Professional Documents
Culture Documents
Strauss
Strauss
w1
■223E »!¥•[• IL-j^Miimi!
SUMMARIES-WHERE to LOOK for HELP
PREREQUISITES ~
Your algebra, tri^m>m(^r)7or)3recalculus textbooks, if you kept them, are good places to look because you are familiar with
the notatioruin^prerentation. If you do not have these books for reference, we have provided the necessary review in Student
MathemStbes Handbook. Also see
Properties of absolute value, see Table 1.1, p. 3
Exact values of trigonometric functions, see Table 1.2, page 9
Directory of curves, see Table 1.3, page 27
Definition of inverse trigonometric functions, see Table 1.4, p. 38
TECHNOLOGY
See Technology Manuals for Maple, Matlab, and Mathematica. We have also found the TI-92 to be of great value in
working many calculus problems. Inexpensive software includes Derive (which is included with the TI-92) and Converge
for graphical representation of many calculus concepts.
INTEGRATION
Definition, p. 274 (indefinite), p. 293 (definite)
Summation formulas, p. 287
Fundamental Theorem of Calculus, p. 303 (first), p. 306 (second)
Integration formulas, see back endsheets
Short integration table, see Appendix D
Integration table, see Student Mathematics Handbook
Volumes of revolution, see Table 6.1, p. 371
Integration strategy, see Table 7.2, pp. 457^)58
Modeling applications involving integration, see Table 10.3, pp. 689-690
INFINITE SERIES
Guidelines for determining convergence of series, Table 8.1, pp. 540-541
Power series for elementary functions, Table 8.2, p. 563
DIFFERENTIAL EQUATIONS
Introduction to differential equations, see Section 5.6, p. 316
Summary of growth models, see Table 7.3, p. 466
First-order differential equations. Section 7.6, p. 461
Summary of strategies for solving first-order differential equations, Table 14.1, p. 942
Second-order linear differential equations, Sections 14.2 and 14.3
MISCELLANEOUS FORMULAS
sin h cos h — 1
SPECIAL TRIGONOMETRIC LIMITS: lim = 1 lim-= 0
0 h h^o h
APPROXIMATION INTEGRATION METHODS: Let A* =- and, for the Arth subinterval, xk = a + AAx; also
n
suppose / is continuous throughout the interval [a, b].
Rectangular Rule: (right endpoints) / /(x) dx ~ [f(a + Ax) + /(a + 2Ax) H-f- f(a + n Ax)] Ax
Ja
Simpson Rule: / /(x) dx ^ \[f (x0) + 4/(xO + 2/(x2) H-b 4/(x„_0 + /(x„)]Ax
-■!
ARC LENGTH: Let / be a function whose derivative /' is continuous on the interval [a, b]. Then the arc length, s, of the
graph of y — f(x) between x = a and x — b is given by the integral
fb _
S = ) yj 1 + [f'(x)]2dx
Ja
If C is a parametrically described curve x = x(t), y = y{t) that is simple for a < t < b, then
-f (SMI)*
SERIES Suppose there is an open interval I containing c throughout which the function / and all its derivatives exist.
FIRST-ORDER LINEAR D.E. The general solution of the first-order linear differential equation
dy
-j- + P(x)y = Q(x)
dx
is given by
y =-i- [ Q(x)I(x)dx + C
/(*) u
where /(x) = g/ pMdx an(i c is an arbitrary constant.
3 3 3
VECTOR-VALUED FUNCTIONS Del operator: V = —i + —j + —k
3x dy dz
r . • ^2f 1 d2f , , , , ,
Laplacian: V / = ~ *xx + *yy + *zz
Monty J. Strauss
Texas Tech University
Gerald L. Bradley
Claremont McKenna College
Karl J. Smith
Santa Rosa Junior College
Prentice
Mall
©2002, 1999, 1995 by Prentice-Hall, Inc.
Upper Saddle River, New Jersey 07458
10 98765432
ISBN D-13-mifl71-7
Preface ix
Input value x
1.1 Preliminaries 2
1.2 Lines in the Plane 13
1.3 Functions and Graphs 19
1.4 Inverse Functions; Inverse Trigonometric Functions 33
Multiply Chapter 1 Review 42
by 5
Guest Essay: Calculus Was Inevitable, John Troutman 46
Mathematical Essays 48
Output value
5x2 + 2
III
iv Contents
3 Differentiation
3.1 An Introduction to the Derivative: Tangents 98
3.2 Techniques of Differentiation 110
3.3 Derivatives of Trigonometric, Exponential, and Logarithmic Functions
3.4 Rates of Change: Modeling Rectilinear Motion 125
3.5 The Chain Rule 138
3.6 Implicit Differentiation 146
3.7 Related Rates and Applications 157
3.8 Linear Approximation and Differentials 165
Chapter 3 Review 177
Group Research Project: Chaos 181
&xsa
Integration
5.1 Antidifferentiation 272
5.2 Area as the Limit of a Sum 282
5.3 Riemann Sums and the Definite Integral 290
5.4 The Fundamental Theorems of Calculus 302
5.5 Integration by Substitution 309
5.6 Introduction to Differential Equations 316
5.7 The Mean Value Theorem for Integrals; Average Value 328
5.8 Numerical Integration: The Trapezoidal Rule and Simpson's Rule
5.9 An Alternative Approach: The Logarithm as an Integral 342
Chapter 5 Review 346
Guest Essay: Kinematics of Jogging, Ralph Boas 351
Mathematical Essays 352
Cumulative Review: Chapters 1-5 353
Contents v
y ~TT TT~T~r TT l 1
l 1 1 ) \ \ TT 1 1 l 1
7.1 Review of Substitution and Integration by Table 426
\ i »\ i i i > \ i i \ 7.2 Integration by Parts 435
-3-
\ \ \ \ i » \ \ \ \
-2i 7.3 Trigonometric Methods 441
\\ \ \ \ \ \ c \ V \ \ \ \ 7.4 Method of Partial Fractions 448
'•Vh y\ \ \ \ ^ \ > \ \ \ \
i / i i / / / / 11 11
7.7 Improper Integrals 472
i / i i 1 1 / i 11 11 7.8 Hyperbolic and Inverse Hyperbolic Functions 481
1 1 1 1 1 1 1 1 1 1 1 1
.U JLLJLJLj u L±j Chapter 7 Review 487
Group Research Project: Buoy Design 491
Appendices
Introduction to the Theory of Limits A-l
B Selected Proofs A-8
Significant Digits A-17
Short Table of Integrals A-21
Trigonometric Formulas A-31
Answers to Selected Problems A-34
G Credits A-80
Index
Preface
his text was developed to blend the best aspects of calculus reform with the rea¬
sonable goals and methodology of traditional calculus. It achieves this middle ground
by providing sound development, stimulating problems, and well-developed pedagogy
within a framework of a traditional structure. “Think, then do,” is a fair summary of
our approach.
Organization
• In this edition we introduce ex and In * in Chapter 2 after we have defined the notion
of a limit. This is beneficial because it allows the number e to be properly defined
using limits. We also assume a knowledge of the conic sections and their graphs. A
free Student Mathematics Handbook is available that provides review and reference
material on these transcendental functions.
• I’Hopital’s Rule is now covered earlier in Chapter 4. This placement allows instruc¬
tors to explore more interesting applications like curve sketching.
• A new section covering applications to business, economics, and the life sciences
has been added to Chapter 6 on Applications of the Integral. This new material is
designed to help students see how calculus relates to and is used in other disciplines.
• The chapter on polar coordinates and parametric forms has been distributed to other
chapters in the book. The polar coordinate system and graphing in polar forms is in
Chapter 6 in the context of the integration topic of finding areas. Parametric repre¬
sentation of curves now appears in the book where it is first needed, in Chapter 9.
• Modeling continues as a major theme in this edition. Modeling is now introduced in
Section 3.4, and then appears in almost every section of the book. These applications
are designated MODELING PROBLEMS. Some authors use the words “Modeling
Problem” to refer to any applied problem. In the third edition of Calculus, we make
a distinction between modeling problems and application problems by defining a
modeling problem as follows. A modeling problem is a problem that requires that
the reader make some assumptions about the real world in order to derive or come up
IX
with the necessary mathematical formula or mathematical information to answer the
question. These problems also include real-world examples of modeling by citing
the source of the book or journal that shows the modeling process.
Problem Sets
Supplements
• Interactive CD (free with every new copy). The new CD-ROM is designed to en¬
hance students’ computational and conceptual understanding of calculus. This CD-
ROM is not an add-on of extra material to the text but rather an incredibly useful
expansion of the text. See the media supplement portion of the Walk-Through for a
complete description of each CD.
• TestGen-EQ. This easy to use test generator contains all of the questions from the
printed Test Item File.
• Prentice Hall Online Homework Grader. For more details, see the PH Homework
Grader section in the Walk-Through.
Hallmark Features
Some of the distinguishing characteristics of the earlier editions are continued with this
edition:
• It is possible to begin the course with either Chapter 1 or Chapter 2 (where the
calculus topics begin).
• We believe that students learn mathematics by doing mathematics. Therefore, the
problems and applications are perhaps the most important feature of any calculus
book. You will find that the problems in this book extend from routine practice to
challenging. The problem sets are divided into A Problems (routine), B Problems
(requiring independent thought), and C Problems (theory problems). You will find
the scope and depth of the problems in this book to be extraordinary while engineer¬
ing and physics examples and problems play a prominent role, we include applica¬
tions from a wide variety of fields, such as biology, economics, ecology, psychology,
and sociology. In addition, the chapter summaries provide not only topical review,
but also many miscellaneous exercises. Although the chapter reviews are typical
of examinations, the miscellaneous problems are not presented as graded problems,
but rather as a random list of problems loosely tied to the ideas of that chapter. In
addition, there are cumulative reviews located at natural subdivision points in the
text: Chapters 1-5, Chapters 6-8, Chapters 1-10, and Chapters 11-13. For a full
description of each type of problem available, see pages xviii through xxi of Walk-
Through.
• We understand that students often struggle with prerequisite material. Further, it is
often frustrating for instructors to have to reteach material from previous courses. As
Preface xi
Because of careful checking and proofing by several people besides each of the three
authors, the authors and publisher believe this book to be substantially error free.
For any errors remaining, the authors would be grateful if they were sent to Monty
J. Strauss at M.Strauss@ttu.edu or Monty J. Strauss, Department of Mathematics
and Statistics, Texas Tech University, Lubbock, TX 79409-1042.
Acknowledgments
The writing and publishing of a calculus book is a tremendous undertaking. We take
this responsibility very seriously because a calculus book is instrumental in transmit¬
ting knowledge from one generation to the next. We would like to thank the many
people who helped us in the preparation of this book. First, we thank our editor George
Lobell, who led us masterfully through the development and publication of this book.
We sincerely appreciate Henri Feiner, who not only worked all of the problems but also
read and critiqued each word of the manuscript, and Carol Williams who transcribed
our manuscript into TgX. We would like to thank Dennis Kletzing for his work in
transforming the manuscript into finished pages. Finally, we would like to thank Lynn
Savino Wendel, who led us through the production process.
Of primary concern is the accuracy of the book. We had the assistance of many:
Henri Feiner, who read the entire manuscript and offered us many valuable sugges¬
tions, and Nancy and Mary Toscano, who were meticulous in their checking of our
manuscript. Thanks also to the accuracy checkers of the previous editions, Jerry
Alexanderson, Mike Ecker, Ken Sydel, Diana Gerardi, Kurt Norlin, Terri Bittner,
Nancy Marsh, and Mary Toscano. We would also like to thank the following read¬
ers of the text for the many suggestions for improvement:
Monty J. Strauss
Gerald L. Bradley
Karl J. Smith
)
CALCULUS
HALLMARK FEATURES
Solution
The general situation: (See Student Mathematics Hundltook. Problem 37. Problem
Ste|i l Let x denote the length (in feet) of the person's shadow, and y, the distance
between the person and tlte street light, as shown in Figure 3 46 Let / denote
the time (in seconds).
Step 2. Since AABC and A DEC are similar, we have
(Included on the Student CD-ROM as well as in an optional print supplement)
7 dx
'id!
Step 4. List the known quantities. We know that dy/dt = 7. Our goal is to find
integral tables. Students are guided from the text to this handbook o A person \*rsilking
m n tfrmlump
dxfdt Substitute and then solve for die unknown value:
dv _ 7 dx
for other features. The length of the person’s shadow is increasing at the rate of 3 ft/*. ■
Solution
The general situation: l et .r and v be the distance* from the base of tlte wall to
the foot and top of tlte ladder, respectively, as shown in Figure 3.47. (See Student
Mathematics Handbook. Problem 38, Problem Set 1.)
Notice that ATOB is a right triangle, so a relevant formula is (Ik Pythagorean
theorem;
x14- jr = 25
Differentiate both sides of this equation with respect to t:
, dx dv
2x— + 2y-f- = 0
dt dt
Tile specific situation: At the particular instant in question, x — 4 and
1 dv
We also know that - 2. ami the goal is to find ~ at
dt dt
this instant. We have
2(4)(2) -f 2(3) — =0
This tells us that, at the instant in question, the hag is descending (since dy/dt is
negative) at the rale of 8/3 =« 2.7 m/sec. H
page 159
Solution
We are seeking a faintly of curses Bach curve m that family intersects each curve in
jhc family x v = C ai right angles, as shown in Figure 5,2fi. Assume that a typical
point on a given curve in tlie family xy = C has coordinates lx. yt anil that a typical
point on the orthogonal irtijcctory curve has coordinates (X.
xymC
dy
X — *t* y = 0 Product rule
dx =T
' We Use upfmraw teller- tor une kind ill curse wxl InvxnEisc fix Ihc oilier In nuke it easier In Cell nhk'il
curve it hems nicniiimcil at each stage at Ihc lottouinjr ilucuiunn.
page 322
124 Orapltr 3 Diffcranfahoa
‘i (1+s)]
ttnvor rule for logarithm*
Since In i I* eontinuou*,
um th* composition limit rule.
Definition of <
Differentiate /lx) = —
-I
gives students insight into how technological advances can be used instetidol • cstrx;
•-A{Trs&) (tln(>
to help understand calculus. Instructors can also encourage students ■n—a—ai
E-mi at an —
nslead of see x tan x: and
to use technology to complete exercise sets. Laboratory manuals are - (esc xi = - Itjwr 12! .Sample m
lliai the fotm of the an
in the lest.
available at a discount when packaged with the text. See supple¬ instead of - esc xcofx. A solution to fixample 9.
using a Tl-92, is show n m Figure 3.21.
Although the fomt of this calculator solution is different from tlie form shown ill tltc example, you should notice that Use)
ments section of a complete list of available technology. ore equivalent by using ulgcbra (and recalling some of the fundamental identities from trigonometry i.
A good test for your calculator is to try to find the derivative ol |x| at » - 0, which we know docs not exist Many
calculators will give the incotrcct answer of 0.
page 124
xvn
PROBLEM SETS
page 227
Interpretation Problems ►
These problems require exposi¬
58. Interpretation Problem Evaluate lim x1-—*_
tion that requires a line of thinking •r—[_ 1,000.000,000 '
Explain why a calculator solution may lead to an incorrect
that is not directly covered in the conclusion about the limit.
textbook. Encourages students to
think critically about calculus
page 69
concepts.
page 69
xviii
''4r
20. Modeling Problem Two towns A and B are 12.0 mi apart and 4 Modeling Problems
are located 5.0 and 3.0 mi, respectively, from a long, straight
highway, as shown in Figure 4.62. Designed to help students see the applications of calculus
to multiple disciplines and real-world situations.
page 246
67. Counterexample Problem When the line segment y — x — 2
between x — 1 and .v = 3 is revolved about the .v-axis, it gen¬
Counterexample Problems ► erates part of a cone. The formula obtained in Section 6.4 says
that the surface area of the cone is
These problems help students develop the ability to con¬
struct counterexamples. Students are asked to formulate
2n{x — 2) V2 dx
an example that satisfies certain conditions.
page 421
page 333
XIX
PROBLEM SETS
Spy Problems ►
54. Spy problem Just as the Spy is about to catch up with Sccldrat
A student favorite, these problems which run throughout
(Problem 70 ol Chapter 10 Supplementary Problems), the
the text like a movie serial, trace the heroic events of an snow gives way and he lalls into a cavern. He slaggers to his
international spy. His survival requires the students' help feet and removes his skis. Why is it so warm? Good grief—
the cave is a large roasting oven! Fortunately, he is wearing
by successfully answering calculus questions. his heat-detector ring, which indicates the direction of great¬
est temperature decrease. Suppose the bunker is coordinalized
so that the temperature at each point (jc, y) on the floor of the
bunker is given by
HI. HISTORICAL'QUEST
Romcmujon was that rarest of T(x,y) = 3(x — 6)2 + 1.5(y - l)2 +41
mathematicians, an instinctive
genius witii virtually no for¬ degrees Fahrenheit, where x and y are in feet. The Spy begins
mal training. Like his Hindu at the point (1,5) and stumbles across the room at the rate of
predecessor, Bhaskara (see the 4 It/inin, always moving in the direction of maximum temper¬
ature decrease. But he can last no more than 2 minules under
Historical .uest in Problem 55
these conditions! Assuming that there is an escape hole at the
of Section 4 4), Ramanujan had
point where the temperature is minimal, does he make it or is
an uncanny instinct for numeri¬ (he Spy toast at last?
cal "truth" and conceived of his
results in much the same way
that a sculptor "sees" a statue
in a raw block of stone. In his short life, he initiated new ways page 748
of thinking about number' theory and mode conjectures that
are the subject of mathematical inquiry to this day.
A story related by his friend and mentor, the eminent British
number theor ist G. H. Hardy (1877-1947), serves to illustrate
the resonance between Ramanu jan's mind and the concept of
number. Ramanujan was ill and Hardy came to visit Him in the
hospital. At a loss for how to begin the conversation. Hardy
idly remarked that he had arrived in a taxi with the "dull"
4 Historical Quest Problems
number 1729. Ramanujan immediately grew excited and ex¬
These problems help students to place calculus in a historical
claimed, "No, Hardy, it is a very interesting number, for it is
the smallest integer that can be expressed as the sum of cubes context, and further encourages them to take part in the math¬
in two different ways." (1729 = U - 12* — 9l -f- (()'.) ematical discovery process.
In a letter from Ramanujan to G. H. Hardy at Cambridge
University. Ramanujan stated
. /1 \ / in / 1 ■ f 5\ * >
+ 9 ( -—- 1 Iff • , . . _ JI
V 2/ \ 2•4/ \ 2 • 4 • 6/ j7
Hardy spent a great deal of time wondering how this sum could
be equal to 2 /t ■
V
page 568
page 719
XX
END-OF-CHAPTER PROBLEMS
■4 Concept Problems
Concept problems are designed to develop
students’ conceptual understanding of calculus
by learning how to think and draw conclu¬
sions about calculus topics.
page 566
Supplementary Problems
Describe the domain of each function given in Problems 1-4.
2 2
1. fix, v) = -/I6 - x2 - y2 2. f(x.v) = -—-A-
x -y
T Putnam Examination Problems 3. /(x, y) = sin-1 x + cos-1 y 4. fix. y) = ex+y tan"1 j
These challenging problems are provided to give insight Find the partial derivatives fx and /,. for the functions defined in
into the types of problems that are asked in mathemati¬ Problems 5-10.
97. Putnam Examination Problem Let / be a real-valued function 9. fix, y) = 2x}y -1- 3xy2 + - 10. fix. v) = x\e?y
X
having partial derivatives defined for x2 + y2 < 1 that sat¬ For each function given in Problems 11-15, describe the level
isfies |/(.v, y)| < 1. Show that there exists a point (r0, .Vo) in cune or level surface f = cfor the given values of the constant c.
the interior of the unit circle such that
11. fix. y) = x2 - y; c = 2, c = -2
1/rOo, yo)]2 + Uy(x0, v0)]2 < 16.
12. fix. y) = 6x 4- 2y; c = 0, c = 1, c = 2
98. Putnom Examination Problem Find the smallest volume bounded
by the coordinate planes and a tangent plane to the ellipsoid ix ft
13. z = fix.y)i = \A2 + .V2 ifz>0_
|y| if z < 0
7 2 2
c = 0, c=l,c=-l
14. fix, y. z) = x2 + y2 + z2; c = 16, c — 0, c = —25
■, v2 z2
99. Putnam Examination Problem Find the shortest distance between 15. fix, y, z) = .v2 + ; c = 1, c = 2
the plane Ax + By + Cz + 1 = 0 and the ellipsoid
Evaluate the limits in Problems 16 and 17, assuming they exist.
A Supplemental Problems
Extensive sets of problems (usually about 100) occur at
the end of each chapter. These problems are based on
randomly selected topics from the complete chapter.
Also helps students to check their comprehension of the
entire chapter.
xxi
MEDIA SUPPLEMENTS
Live Examples
The heart of this CD is the Live Examples powered by
Live Math. Nearly every geometric text example is ani¬
mated. The traditional text environment is enhanced and
brought to life with animations and questions exploring
what-if scenarios. Algebraic solutions are paired with
geometric solutions. There are hundreds of Live Examples
to accompany the complete text.
XXII
MEDIA SUPPLEMENTS
Companion Website ▼
www.prenhall.com/strauss
This website contains much of what is on the CD-ROM along with other activities better suited to the
Internet and includes:
Live Examples
The heart of this website lies in the Live Examples powered by Live Math. Nearly every geometric text example
is animated. The traditional text environment is now enhanced and brought to life with animations and questions
exploring what-if scenarios. Algebraic solutions are paired with geometric solutions. There are hundreds of Live
Examples.
Homework Starters
Each text section includes teaching hints for 3-5 homework problems. These hints often include geometric ani¬
mations. There are more than 150 Homework Starters.
xxiv
OTHER AVAILABLE SUPPLEMENTS
STUDENT SUPPLEMENTS
Student Survival and Solutions Manual (0-13-067245-9)
Contains carefully worked out solutions to all odd-numbered exercises in the text. Contains helpful hints to
bridge the gap between the textbook and a working knowledge of calculus. These hints anticipate the types of
errors and difficulties a student may have while taking the course.
Student CD-ROM
Comes free packaged with each new text. Contains animations of nearly every text example and 10 True/False
questions per section. Also includes Homework Starters, animations, video clips and the complete contents of the
Student Math Handbook.
Companion Website
www.prenhall.com/strauss
Provides additional applied examples and problems, animations that explore “what if’ scenarios, exciting links
to other math sites on the web, True/False quizzes to emphasize key concepts, and a syllabus manager for pro¬
fessors. Contains all the materials from the Student CD-ROM along with a Syllabus Manager and links to other
appropriate and helpful math sites on the web.
TECHNOLOGY MANUALS
Available at a discount when packaged with the text.
XXV
OTHER AVAILABLE SUPPLEMENTS
INSTRUCTOR SUPPLEMENTS
Instructor's Edition (o-i3-0920io-x)
The Instructor’s Edition includes the answers to all odd-numbered exercises and detailed Teaching Notes.
TestGen-EQ (0-13-092018-5)
This easy to use test generator contains all of the questions from the Printed Test Bank.
Companion Website
www.prenhall.com/strauss
Provides additional applied examples and problems, animations that explore “what if’ scenarios, exciting links to
other math sites on the web, True/False quizzes to emphasize key concepts, and a syllabus manager for profes¬
sors. Contains all the materials from the Student CD-ROM along with a Syllabus Manager and links to other
appropriate and helpful math sites on the web.
TECHNOLOGY MANUALS
Available at a discount when packaged with the text.
XXVI
1
TEXT OPTIONS
Single Variable Calculus, 3e
Includes Chapters 1-10 of the three semester full text. This version is intended for a two-semester course in
single variable calculus.
3 Differentiation
Contents 54
3.5
tondOae
bUanlUi 131
U fettaottalham 1*3
U iteas Nl
U '»5lak«<tiii«n>;nMqriMT 713
It M&d-.tm 77*
5 Integration
Please see Supplement section of this Guide for more complete descriptions.
xxvn
TEXT OPTIONS
Multivariable Calculus, 3e
Includes Chapters 9-14 of the three-semester full text. This version is intended for a one-semester course in
multivariable calculus.
Please see Supplement section of this Guide for more complete descriptions.
XXVlll
CONTENTS
1.1 Preliminaries
Distance on a number line
Functions and Graphs
Absolute value
Distance in the plane
Trigonometry
Solving trigonometric equations
1
2 Chapter 1 Functions and Graphs
1.1 Preliminaries
IN THIS SECTION distance on a number line, absolute value, distance in the plane, trigonometry, solving
trigonometric equations
This section provides a quick review of some fundamental concepts and techniques
Every mathematical book that is worth
from precalculus mathematics. If you have recently had a precalculus course, you may
reading must be read “backwards and for¬
skip over this material.
wards,” if I may use the expression. I
would modify Lagrange’s advice a lit¬ Algebra, geometry, and trigonometry are important ingredients of calculus. Even
tle and say, “Go on, but often return to though we will review many ideas from algebra, geometry, and trigonometry, we will
strengthen your faith.” When you come not be able to develop every idea from these courses before we use it in calculus. For
on a hard and dreary passage, pass it over; example, the law of cosines from trigonometry may be needed to solve a problem in
come back to it after you have seen its im¬ a section that never mentions trigonometry in the exposition. For this reason, we have
portance or found its importance or found made available a separate reference manual, Student Mathematics Handbook, which
the need for it further on. includes the background material you will need for this course. We suggest that you
-George Chrystal,
keep it close at hand. References to this handbook are indicated by the logo
Algebra, Part 2 (Edinburgh, 1889)
You are probably familiar with the set of real numbers and several of its subsets,
including the counting or natural numbers, the integers, the rational numbers, and the
(jsmfP) The real numbers, along with irrational numbers.
The real numbers can be visualized most easily by using a one-dimensional co¬
these subsets, are found in Chapter 2
ordinate system called a real number line, as shown in Figure 1.1.
of the Student Mathematics
Handbook.
-2-1012
Notice that a number a is less than a number b if it is to the left of b on a real number
line, as shown in Figure 1.2. Similar definitions can be given for a > b, a < b, and
a > b.
a < b:
-2 -1 0
distance from x to the origin is \x\
The location of the number 0 is chosen arbitrarily, and a unit distance is picked
(meters, feet, inches, ...). Numbers are ordered on the real number line according to
the following order properties.
Order Properties
For all real numbers a, b, and c
ABSOLUTE VALUE
a if a > 0
<2 —
—a if a < 0
The number x is located |x | units away from 0 — to the right if x > 0 and to the left if
x < 0.
Absolute value is used to describe the distance between points on a number line.
Distance Between Two Points The distance between the numbers xi and x2 on a number line is
on a Number Line
1*2 -*ll
WARNING . , ,
Note that
|x2 -x,| = |x, x21 For example, the distance between 2 and —3 is \2 — (—3)| = 5 units, and between —2
and —3 is |— 2 — (—3)| = 1 unit.
Several properties of absolute value that you will need in this course are summa¬
rized in Table 1.1.
Property Comment
7. Let b > 0; \a\ = b if and only 7. This property is useful in solving absolute
WARNING >jfandon|yif^isusec|
if a = ±b value equations. See Example 1.
to mean that both a statement and its
8. Let b > 0; |a| < b if and only 8. and 9. These are the main properties used
converse are true. That is,
if — b < a < b in solving absolute value inequalities. See
if p, then q, and Example 2.
if q, then p.
9. Let b > 0; |a| > b if and only
For example, property 8 has two parts: if a > b or a < — b
(i) If |a| < b, then — b < a < b; 10. \a + b\ < \a\ + \b\ 10. This property is called the triangle in¬
(ii) If — b < a < b, then \a\ < b. equality. It is used in both theory and nu¬
merical computations involving inequalities.
Chapter 1 Functions and Graphs
and
a\ > b if and only if a > b or a < — b
x < b (-oo.fe]
b
x < b (-oo, b)
b
Solution
If 2x — 6 > 0, then \2x — 6| = 2x — 6 so that we solve
2x — 6 — x or x =6
1.1 Preliminaries 5
— (2x — 6) = x
—3x = —6
x = 2
\x — a\ = b
is satisfied by two values of x that are a given distance b from a when represented on
a number line. For example, |x — 5| = 3 states that x is 3 units from 5 on a number
line. Thus, x is either 2 or 8.
3 units 3 units
--♦ » *- X — 5| = 3
2 5 8
X - 5 = ±3
1 X = 5±3
two values that are 3 units from 5
x = 8 or 2
less than 3 units less than 3 units greater than 3 units greater than 3 units
*-" $- "-4- - - --.- ■: ►
2 5 8 2 5 8
t
midpoint midpoint
Solution
-4 + 3<2x-3 + 3<4 + 3
-1 < 2x < 7
1 2x 7
— < — < -
2 ” 2 “ 2
1 7
— < x < -
2 “ ~ 2
Solution
Let w = weight of the bag of cement in pounds. Then
90 — 2<u;<90 + 2
—2 < w — 90 < 2
Absolute value is used to find the distance between two points on a number line. To
find the distance between two points in a coordinate plane, we use the distance formula,
which is derived by using the Pythagorean theorem.
where Ax (read “delta x”) is the horizontal change X2 — xi (sometimes called run)
and Ay (read “delta y”) is the vertical change y2 — yi (sometimes called rise).
Proof Using the two points, form a right triangle by drawing lines through the given
points parallel to the coordinate axes, as shown in Figure 1.4.
The length of the horizontal side of the triangle is |x2 — xj | = | Ax|, and the length
of the vertical side is |y2 — yi | = | Ay|. Then
Midpoint formula Related to the formula for the distance between two points is the
formula for finding the midpoint of a line segment, as shown in Figure 1.4.
Midpoint Formula The midpoint, M, of the segment with endpoints P|(xi, yi) and P2(x2, yf) has
coordinates
M(xj±*t.y±±yi\
1.1 Preliminaries 7
Notice that the coordinates of the midpoint of a segment are found by averaging the
first and second components of the coordinates of the endpoints, respectively. You are
asked to derive this formula in Problem 74.
Graph of an Equation The graph of an equation in two variables x and y is the collection of all points
P(x, y) whose coordinates (x, y) satisfy the equation.
In Example 4, we use the distance formula to derive the equation of a circle. This
means that if x and y are numbers that satisfy the equation, then the point (x, y) must
lie on the circle. Conversely, the coordinates of any point on the circle will satisfy the
equation.
Solution
Let (x, y) be any point on a circle. Recall that a circle is the set of all points in the
plane a given distance from a given point. The given point (h, k) is the center and the
given distance is the radius r.
r - DISTANCE FROM (h, k) TO (x, y)
A unit circle is a circle of radius 1 and center at the origin, so its equation is
x2 + y2 = 1.
Solution
See Figure 1.5. The radius is the distance from the center to the given point:
= 74+169
= 7173
4x2 + 4 y2 — 4x + 8y — 5 — 0
Solution
We need to convert this equation into standard form. To do this we use a process called
completing the square (see Section 2.6 of the Student Mathematics Handbook).
4x2 + 4y2 - 4x + 8y - 5 = 0
(x - |)2 + (y + l)2 = f = f
TRIGONOMETRY
One of the prerequisites for this book is trigonometry. If you need some review, consult
a text on trigonometry or Chapter 4 of the Student Mathematics Handbook that accom¬
panies this book. A summary of useful trigonometric formulas is found in Appendix E.
WARNING , . . . Angles are commonly measured in degrees and radians. A degree is defined to be
In calculus, angles are
usually measured in terms of radians 3^g revolution and a radian ^ revolution. Thus, to convert between degree and radian
rather than degrees since radian measure use the following formula:
measure often leads to simpler
formulas. You may assume that when 9 measured in degrees 0 measured in radians
we write expressions such as sinx,
360 2it
cosx, and tanx, the angle x is in
radians unless otherwise specified by a
degree symbol.
EXAMPLE 7 Converting degree measure to radian measure
Convert 255° to radian measure.
255 e
Solution
360 2n
e
Solution
360 2tx
« 57.29577951
1.1 Preliminaries 9
There will be many times in calculus when you will need to solve a trigonometric
equation. As you may remember from trigonometry, solving a trigonometric equation
is equivalent to evaluating an inverse trigonometric relation. Inverse functions will
be introduced in Section 1.4; for now, we will solve trigonometric equations whose
solutions involve the values in Table 1.2 (called a table of exact values). These exact
values from trigonometry are reviewed in the Student Mathematics Handbook.
7t 7T 7T 7T
Function \ Angle 0
6 4 3 2
V3 V2 1
cos 6 0
2 2 2
1 V2 V3
sin# 1
2 2 2
V3
tan 9 a
1 V3 undefined
It is customary to use the values from Table 1.2 whenever possible. The approximate
calculator values will be given only when necessary.
Solution
If you use a calculator, make certain it is in the proper mode (radian or degree).
Solution
For an algebraic solution,
V3cosx = 1 — sinx
3 cos x = 1 — 2 sinx + sin“x Square both sides.
2 + 2 sin x — 4 sin2 x = 0
2 sin2 x — sin x — 1 = 0
(2sinx + l)(sinx — 1) = 0
sinx = — ^ sinx = 1
_ ln_ 11 n _ 7r
x ~ 6 ’ 6 1 ~ 2
Flowever, since we squared both sides, we need to check for extraneous roots by
substituting into the original equation, since squaring sometimes introduces extraneous
solutions (that is, solutions that do not satisfy the given equation). Checking, we see
that x = ^ is extraneous, and the solution is x = f
It is often worthwhile to check by finding a geometric solution. We graph the left
and right sides of the equation on the same axes:
y\ = sinx + x/3cosx
yi = 1
The graphs are shown in Figure 1.7, and for a solution we look to the intersection
points.
b. [3,5] b. [f,V2]
c- [-2,1) x > —3 c.
3. Represent each of the following on a number line, 48. WHAT DOES THIS SAY? Describe a process for solving ab¬
a. — 1 < x < 4 b. — 1 < x < 3 solute value equations.
c. (0, 2) d. [-2, 1) 49. WHAT DOES THIS SAY? Describe a process for solving ab¬
solute value inequalities.
4. Represent each of the following on a number line,
50. WHAT DOES THIS SAY? Describe a process for solving
a. x > 2 b. x < 4
trigonometric equations.
c. (—oo, 2) U (2, oo) d. — 2 < x < 3 or x > 5
Specify the period for each graph in Problems 51-56.
In Problems 5-6, plot the given points P and Q on a Cartesian Also graph each curve.
plane, find the distance between them, and find the coordinates of
51. a. y = sinx b. y = cos* c. y = tanx
the midpoint of the line segment PQ.
52. a. y = 2sin2;rx b. y = 3cos37rx c. y = 4tan(^)
5. a. P(2, 3), Q(—2, 5) b. P(-2, 3), 0(4. 1)
53. y = tan(2x — |) 54. y = 2cos(3x + 2n) — 2
6. a. P(—5, 3), Q(—5, —7) b. P(-4, 3), <2(3, -4)
55. y = 4 sin(^x 4- 2) — 1 56. y = tan(^x + |)
Solve each equation in Problems 7-24. Assume that a, 57. The current I (in amperes) in a certain circuit (for some con¬
b, and c are known constants. venient unit of time) generates the following set of data points:
7. x2 — x = 0 8. 2y 2 + y - 3 = o
9. y2 -5y + 3 = 17 10. X2 ~h 5x + a = 0 Time Current Time Current
11. 3x2 — bx = c 12. 4x 2 + 2(k + 25 = 0 0 -60.00000 10 30.00000
13. |2x +4| = 16 14. |5y + 2| = 12 1 -58.68886 11 40.14784
15. |3 -2w\ = 7 16. 15 - 3f| = 14 2 -54.81273 12 48.54102
17. |3x + l| = -4 18. 11
NJ
ca
3 -48.54102 13 54.81273
II
1
*Many problems in this book are labeled What Does This Say? Follow¬ y — A — B cos C(x — D)
ing the question will be a question for you to answer in your own words,
or a statement for you to rephrase in your own words. These problems are Determine a possible equation of a curve that is generated by
intended to be similar to the “What This Says" boxes. this water wheel.
12 Chapter 1 Functions and Graphs
59. The sun and moon tide curves are shown here.* During a new
moon, the sun and moon tidal bulges are centered at the same
longitude, so their effects are added to produce maximum high
tides and minimum low tides. This produces maximum tidal
range (the distance between the low and high tides).
3d cos 0
\/7 + 9 cos2 6
17 -— O
Quadrature
In Mathematical Circles, Boston: Prindle, Weber & Schmidt,
1 969.) One possible explanation is put forth by Otto Neuge-
bauer, scholar and authority on early Babylonian mathematics
position and astronomy. In early Sumerian times, there existed a Baby¬
lonian mile, equal to about seven of our miles. Sometime in the
Sun curve -
Moon curve - first millennium B.C., when Babylonian astronomy reached the
Combined curve- stage in which systematic records of the stars were kept, the
Babylonian mile was adapted for measuring spans of time.
Since a complete day was found to be equal to 1 2 time-miles,
and one complete day is equivalent to one revolution of the
sky, a complete circuit was divided into 1 2 equal parts. Then,
for convenience, the Babylonian mile was subdivided into 30
equal parts. One complete circuit therefore has (12)(30) = 360
equal parts. ■
64. HISTORICAL QUEST The numeration system we use 67. Exploration Problem If sin ax — b, w hat effect does changing
(base 10) evolved over a long period of time. It is often called a have on the solution (a A 0; keep b fixed)?
the Hindu-Arabic system because its origins can be traced 68. If c > 0, show that |x| < c if and only if — c < x < c.
back to the Hindus in Bactria (now Afghanistan). Later, in A.D. 69. Show that — |x| < x < |x| for any number x.
700, India was invaded by the Arabs who used and modified
70. Prove that |a| = \b\ if and only if a = b or a — —b.
the Hindu numeration system, and, in turn, introduced it to
Western civilization. The Hindu Brahmagupta stated the rules 71. Prove that if |a| < b and b > 0, then —b<a<b.
for operations with positive and negative numbers in the sev¬ 72. Prove the triangle inequality:
enth century A.D. There are some indications that the Chinese
had some knowledge of negative numbers as early as 200 B.C. \x + y\ < |x| + |y|
On the other hand, the Western mathematician Girolamo Car¬
73. Show that ||x| — |y|| < |* — y| for all x and y.
dan (1501-1576) was calling numbers such as (-1) absurd as
late as 1545. ■ 74. Derive the midpoint formula
Write a paper on the history of the real number system.
*i + *2 yi + y2 \
Q 65. Exploration Problem If ax + b — 0, what effect does changing
2 ’ 2 )
SLOPE OF A LINE
A distinguishing feature of a line is the fact that its inclination with respect to the
horizontal is constant. It is common practice to specify inclination by means of a
concept called slope. A carpenter might describe a roof line that rises 1 ft for every 3
ft of horizontal “run” as having a slope or pitch of 1 to 3.
Let Ax and Ay represent, as before, the amount of change in the variables x and
y, respectively. Then a nonvertical line L that rises (or falls) Ay units (measured from
bottom to top) for every Ax units of run (measured from left to right) is said to have a
slope of m = Ay/Ax. (If Ay is negative, then the “rise” is actually a fall; and if Ax is
negative, then the run is actually right to left.) In particular, if P(x\, yi) and Q(x2, y2)
are two distinct points on L, then the changes in the variables x and y are given by
Ax = x2 - xi and Ay = y2 — yi, and the slope of L is
Ay y2 - y i
m for Ax ^ 0 See Figure 1.9.
Ax x2 - Xi
L/
Q(X 2d^2);
yi r y .
ay
Ajc
P(
x2 -X .
[ ’■’I
k2-7i
b. The slope is given by m =
x2~x\
Slope of a Line A nonvertical line that contains the points P(x\, yO and Q(xi, yf) has slope
Ay _ y2 - y i
Ax *2 — X\
We say that a line with slope m is rising (when viewed from left to right) if m > 0,
falling if m < 0, and horizontal if m = 0.
There is a useful trigonometric formulation of slope. The angle of inclination of
a line L is defined to be the nonnegative angle 0 (0 < 0 < n) formed between L and
the positive x-axis.
Angle of Inclination The angle of inclination of a line L is the angle 0 (0 < 0 < tc) between L and
the positive x-axis. Then the slope of L with inclination 0 is
WARNING .
Sometimes we say a m — tan 0
vertical line has infinite slope.
We see that the line L is rising if 0 < 0 < j and is falling if | < 0 < n. The
line is horizontal if 0 = 0 and is vertical if 0 = |. Notice that if 0 = j, tan0 is not
defined; therefore the slope, m, is undefined for a vertical line.
To derive the trigonometric representation for slope we need to find the slope of
the line through P{x\, yi) and <2(x2, yf), where 0 is the angle of inclination. From the
definition of the tangent, we have
V2 — Vi Ay
tan0 = 1-— = —— = m See Figure 1.10.
X2 — xi Ax
yt \ ~ i t y* >
-l c
'
X X
(H 0
r
1=, 0
(2, -3)
[-5, -10)
> r
Zero slope (Ay = 0); Slope is undefined;
line is horizontal. (Ax = 0) line is vertical.
In algebra you studied several forms of the equation of a line. The derivations of some
of these are reviewed in the problems. Here is a summary of the forms most frequently
used in calculus.
1.2 Lines in the Plane 15
Solution
The slope of the equation passing through the given points is
b- 0 _ b
0—a a
Use the point-slope form with h = 0, k = b. (You can use either of the given points.)
b
y — b =-(x — 0)
a
ay — ab — —bx
bx + ay = ab
x y
—I-= 1 Divide both sides by ab.
a b
Solution
Because d is linearly related to w, we know that points (w, d) lie on a line, and the
given information tells us that two such points are (3, 4) and (6, 6) as shown in Fig¬
ure 1.12. We first find the slope of the line and then use the point-slope form to derive
its equation.
6-4 2
5 = |w + 2
It is often useful to know whether two given lines are either parallel or perpendicular.
A vertical line can be parallel only to other vertical lines and perpendicular only to
horizontal lines. Cases involving nonvertical lines may be handled by the criteria given
in the following box.
The key ideas behind these two slope criteria are displayed in Figure 1.13.
1-
T N
K
T I - 42 1
2
i_
Vi
✓
X
_
a. Parallel lines have b. Perpendicular lines have negative
equal slope: mj = m2. reciprocal slopes: m-im2 = -1.
Figure 1.13
a. Find an equation of the line that is parallel to L and passes through P(4, 7).
b. Find an equation of the line that is perpendicular to L and passes through P(4, 7).
1.2 Lines in the Plane 17
Solution
By rewriting the equation of L as y = — \x + §, we see that the slope of L is m — —
a. Any line that is parallel to L must also have slope mx — The required line
—
contains the point P(4, 7). Use the point-slope form to find the equation and write
your answer in standard form:
y- 7 =-§(*—4)
2y -14 = -3x + 12
3x + 2y — 26 = 0
y-7 = f(x-4)
3y — 21 = 2x — 8
Figure 1.14 Graph of lines parallel 2x — 3y + 13 = 0
and perpendicular to the given line
3x + 2y — 5 These lines are shown in Figure 1.14. ■
5. passing through the point (1, |) with slope 0 32. Find the equation of the perpendicular line passing through the
6. slope 2 and y-intercept (0, 5) midpoint of the line segment connecting (-3, 7) and (4, -1).
7. vertical line through (—2, —5) 33. Three vertices of a parallelogram are 4(1,3), C(4, 11), and
B(3, —2). If A and B lie on the same side, what is the fourth
8. slope —3, x-intercept (5, 0)
vertex?
9. x-intercept (7, 0) and y-intercept (0, -8)
Use a graphing calculator in Problems 34-45 to solve the systems
10. x-intercept (4.5, 0) and y-intercept (0, -5.4)
of equations graphically, and approximate the coordinates of the
11. passing through (—1,8) parallel to 3x + y = 7 intersection point. Some calculators have an INTERSECT feature.
12. passing through (4, 5) parallel to the line passing through Check your owner’s manual. Then solve the system algebraically
(2, 1) and (5, 9) and compare the results.
13. passing through (3, —2) perpendicular to 4x — 3y + 2 = 0
2x — 3y = —8 12x — 3y = —8
14. passing through (—1,6) perpendicular to the line through the 34. 35.
x 4- y = 6 14x — 6y = 0
origin with slope 0.5
15. perpendicular to the line whose equation is x — 4y + 5 = 0 2x — 3y = —8 13x — 4y = 16
where it intersects the line whose equation is 2x 4- 3y — 1 = 0 36. 37.
y = fx 4-1 | -x 4- 2y = -6
In Problems 16-29, find, if possible, the slope, the y-intercept,
y = 3x 4- 1 |2x 4- 3y = 12
and the x-intercept of the line whose equation is given. Sketch 38. 39.
the graph of each equation. x — 2y = 8 | —4x 4- 6y = 18
44 {-v2 + y2 = 10 \x2 + y2 = l5 a. There are three possible choices for P4. One is A(3, 1).
• |U_3)2 + y2 = 9 ^ \{x+4)2 + y2= 16 What are the others, which we will call B and Cl
46. A life insurance table indicates that a woman who is now A b. The centroid of a triangle is the point where its three medi¬
years old can expect to live E years longer. Suppose that A ans intersect. Find the centroid of A ABC and of A P\ P1P3.
and E are linearly related and that E — 50 when A = 24 and Do you notice anything interesting?
E — 20 when A = 60. 54. Exploration Problem Ethyl alcohol is metabolized by the human
body at a constant rate (independent of concentration). Sup¬
a. At what age may a woman expect to live 30 years longer?
pose the rate is 10 mL per hour.
b. What is the life expectancy of a newborn female child?
a. Express the time t (in hours) required to metabolize the ef¬
c. At what age is the life expectancy zero?
fects of drinking ethyl alcohol in terms of the amount A of
47. On the Fahrenheit temperature scale, water freezes at 32° and
ethyl alcohol consumed.
boils at 212°; the corresponding temperatures on the Celsius
scale are 0° and 100°. Given that the Fahrenheit and Celsius b. How much time is required to eliminate the effects of a liter
temperatures are linearly related, first find numbers r and 5 so of beer containing 3% ethyl alcohol?
that E — rC + s, and then answer these questions. c. Discuss how the function in part a can be used to determine
a. Mercury freezes at — 39°C. What is the corresponding a reasonable “cutoff” value for the amount of ethyl alcohol
Fahrenheit temperature? A that each individual may be served at a party.
b. For what value of C is F = 0? 55. Since the beginning of the month, a local reservoir has been
c. What temperature is the same in both scales? losing water at a constant rate (that is, the amount of water in
48. The average SAT mathematics scores of incoming students at the reservoir is a linear function of time). On the 12th of the
an eastern liberal arts college have been declining in recent month, the reservoir held 200 million gallons of water; on the
years. In 1996, the average SAT score was 575; in 2001, it 21st, it held only 164 million gallons. How much water was
was 545. Assuming the average SAT score varies linearly with in the reservoir on the 8th of the month?
time, answer these questions. 56. To encourage motorists to form car pools, the transit authority
a. Express the average SAT score in terms of time measured in a certain metropolitan area has been offering a special re¬
from 1996. duced rate at toll bridges for vehicles containing four or more
b. If the trend continues, what will the average SAT score of persons. When the program began 30 days ago, 157 vehicles
incoming students be in 2014? qualified for the reduced rate during the morning rush hour.
c. When will the average SAT score be 455? Since then, the number of vehicles qualifying has increased
at a constant rate (that is, the number is a linear function of
*49. Spy Problem An internationally famous spy has escaped from
time), and 247 vehicles qualified today. If the trend continues,
the headquarters of a diamond smuggling ring in the tiny
how many vehicles will qualify during the morning rush hour
Mediterranean country of Azusa. Our hero, driving a stolen
14 days from now?
milk truck at 72 km/hr, has a 40-minute head start on his pur¬
suers, who are chasing him in a Ferrari going 168 km/hr. The 57. Exploration Problem The value of a certain rare book doubles
distance from the smugglers’ headquarters in Azusa to free¬ every 10 years. The book was originally worth $3.
dom in the neighboring country of Duarte is 83.8 km. Is this a. How much is the book worth when it is 30 years old? When
the end of the Spy, or does he live to return in Chapter 2? it is 40 years old?
50. A certain car rental agency charges $40 per day with 100 free
b. Is the relationship between the value of the book and its
miles plus 34^ per mile after the first 100 miles. First express age linear? Explain.
the cost of renting a car from this agency for one day in terms
of the number of miles driven. Then draw the graph and use it 58. HISTORICAL QUEST The region between the Tigris and
to check your answers to these questions. Euphrates Rivers (present day Iraq) is rightly known as the
Cradle of Civilization. During the so-called Babylonian pe¬
a. How much does it cost to rent a car for a 1-day trip of 50
riod (roughly 2000-600 B.C.),+ important mathematical ideas
mi?
began to germinate in the region, including positional nota¬
b. .How many miles were driven if the daily rental cost was
tion for numeration. Unlike their Egyptian contemporaries who
$92.36?
usually wrote on fragile papyrus, Babylonian mathematicians
51. A manufacturer buys $200,000 worth of machinery that de¬ recorded their ideas on clay tablets. One of these tablets, in
preciates linearly so that its trade-in value after 10 years will the Yale Collection, shows a system equivalent to
be $10,000. Express the value of the machinery as a func¬
tion ol its age and draw the graph. What is the value of the
ixy = 600
machinery after 4 years?
) (* + y)2 - 150U - y)2 = 100 .
52. Show that if the point P(x,y) is equidistant from A(l,3)
and B( —1,2), its coordinates must satisfy the equation
Ax + 2y — 5 = 0. Sketch the graph of this equation. Find a positive solution (x > 0, y > 0) for this system correct
53. Let P| (2. 6), P’2(—1, 3), FR0. —2), and P4(a, b) be points in to the nearest tenth.
the plane that are located so that P\ P2P3P4 is a parallelogram.
' For an interesting discussion of Mesopotamian mathematics, see A His¬
Take note: The Spy will accompany us on our journey through this book, tory of Mathematics, 2nd ed., by Carl B. Boyer, revised by Uta C.
if he can survive (which, of course, will require your help). Merzbach, John Wiley and Sons, Inc., New York, 1968, pp. 26-47.
1.3 Functions and Graphs 19
y — k = m(x — h)
61. If A(xi, yi) and B(x2, y2) with jci7^ x2 are two points on the
Figure 1.15 Angle (p between
graph of the line y — mx + b, show that
given lines
m
yi - yi
= - 64. Journal Problem Ontario Secondary School Mathematics
X2 - X]
Bulletin* ■ Show that there is just one line in the family
Use this fact to show that the graph of y = mx + b is a line y — 8 = m(x + 1) that is five units from the point (2,4).
with slope m and then show that the line has y-intercept (0, b).
62. Show that the distance s from the point (x0, yo) to the line 65. The centroid of a triangle with vertices A(xi, yi), B(x2, y2),
C(x3, y3) is the point where its three medians intersect. It can
Ax + By + C = 0 is given by the formula
be shown that the centroid P(x0, yo) is located 2/3 the dis¬
I Axo + Byo + C | tance from each vertex to the midpoint of the opposite side.
Use this fact to show that
^ ~~ VA2 + B2
X1+X2 + X3 yi + y2 + 3b
x0 =--- and y0 =---
63. Let L1 and L2 have slopes m\ and m2, respectively, and let
cp be the angle between L\ and L2, as shown in Figure 1.15.
m2 — m\
Show that tan (p —-. ^Volume 18 (1982), issue 2, p. 7.
1 + m\m2
DEFINITION OF A FUNCTION
Function A function / is a rule that assigns to each element x of a set X a unique element
y of a set Y. The element y is called the image of x under / and is denoted by
f(x) (read as “/ of x”). The set X is called the domain of /, and the set of all
images of elements of X is called the range of the function.
A function whose name is / can be thought of as the set of ordered pairs (x, y) for
which each member x of the domain is associated with exactly one member y = f(x)
of the range. The function can also be regarded as a rule that assigns a unique “output”
Chapter 1 Functions and Graphs
Input value x in the set 7 to each “input” from the set X, just as the function “machine” shown in
Figure 1.16 generates an output of 5x2 + 2 for each input x.
+ What This Says To be called a function, the rule must have the property
that it assigns to each legitimate input one and only one output. For instance, —2
is not a legitimate input for the function that computes the square root because
we are working in the set of real numbers.
Output value
5x2 + 2
Domain
Note that it is quite possible for two different elements in the domain X to map into
the same element in the range, and that it is possible for Y to include elements not in
the range of /. If, however, the range of / does consist of all of 7, then / is said to
map X onto 7. Furthermore, if each element in the range is the image of one and only
one element in the domain, then / is said to be a one-to-one function. A function /
is said to be bounded on [a, b] if there exists a number B so that |/(jc)| < B for all x
in [a, b].
Most of our work will be with real-valued functions of a real variable; that is,
functions whose domain and range are both sets of real numbers.*
FUNCTIONAL NOTATION
Functions can be represented in various ways, but usually they are specified by using a
mathematical formula. It is traditional to let x denote the input and y the corresponding
output, and to write an equation relating x and y. The letters x and y that appear in such
an equation are called variables. Because the value of the variable y is determined
by that of the variable x, we call y the dependent variable and x the independent
variable.
In this book, when we define a function by an expression such as
/(*) = x2 + 4x + 5, we mean the function / is the set of all ordered pairs (x, y)
satisfying the equation y = x2 + 4x + 5. To evaluate a function / means to find
the value of / for some particular value in the domain. For example, to evaluate / at
x = 2 is to find /(2).
' Most functions that appear in this book belong to a very special class of elementary functions, defined
by Joseph Liouville (1809-1882). Certain functions that appear in physics and higher mathematics are not
elementary functions.
1.3 Functions and Graphs 21
Solution
In this case, the defined function / tells us to subtract the independent variable x from
twice its square. Thus, we have
/(—1) = 2(— 1 )2 — (- 1) = 3
To find f(x + h) we begin by writing the formula for / in more neutral terms, say as
/(□) = 2O2 - O)
Then we insert the expression x + h inside each box, obtaining
Finally, if h 0,
Certain functions are defined differently on different parts of their domain and are
thus more naturally expressed in terms of more than one formula. We refer to such
functions as piecewise-defined functions.
Solution
To find /(—0.5), we use the first line of the formula because -0.5 < 2:
/(—0.5) = —0.5 sin(—0.5) = 0.5 sin 0.5 This is the exact value.
0.2397 This is the approximate value.
To find /(f), we use the first line of the formula because f « 1.57 < 2:
Finally, because 2 > 2, we use the second line of the formula to find /(2).
/(2) = 3(2)2 + 1 = 13
a. How far will the object fall in the first second? In the next 2 seconds?
b. How far will it fall during the time interval t = 1 to t = 1 + h seconds?
c. What is the average rate of change of distance (in feet per second) during the time
t — 1 sec to t — 3 sec?
d. What is the average rate of change of distance during the time t = x seconds to
t = x + h seconds?
Solution
a. 5(1) = 16(1 )2 = 16. In the first second the object will fall 16 ft. In the next two
seconds the object will fall
DOMAIN OF A FUNCTION
In this book, unless otherwise specified, the domain of a function is the set of real
numbers for which the function is defined. We call this the domain convention and
it will be used throughout this text. If a function / is undefined at x, it means that
x is not in the domain of /. The most frequent exclusions from the domain are those
values that cause division by 0 or negative values under a square root. In applications,
the domain is often specified by the context. For example, if x is the number of people
on an elevator, the context requires that negative numbers and nonintegers be excluded
from the domain; therefore, x must be an integer such that 0 < x < c, where c is the
maximum capacity of the elevator.
a. /(x) = 2x — 1 b. g(x) = 2x — 1, x —3
(2x — l)(x + 3)
c. h(x) = d. F(x) = y/x + 2 e. G(x) =---
x +3 5 — cosx
Solution
a. All real numbers; (—00, 00)
b. All real numbers except -3. This can be written as (-00, -3) U (-3, 00).
c. Because the expression is meaningful for all x ^ -3, the domain is all real numbers
except —3.
1.3 Functions and Graphs 23
d. F has meaning if and only if x + 2 is nonnegative; therefore, the domain is * > —2,
or [—2, oo).
e. G is defined whenever 5 — cos* ^ 0. This imposes no restriction on * since
|cos * | < 1. Thus, the domain of G is the set of all real numbers; this can be written
(—oo, oo). ■
Equality of Functions Two functions / and g are equal if and only if both
(2x — l)(x + 3)
/(x) = 2x — 1 g(x) = 2x — 1 for * 7^ —3 h(x) =
* T" 3
Solution
A common mistake is to “reduce” the function h to obtain the function /:
This is wrong because the functions / and h have different domains. In particular, -3
is a valid input for /, but it is not a valid input for h. We say that the function h has a
hole at x = -3. In other words, this WRONG calculation is valid only if * ^ -3, so
the correct answer is
COMPOSITION OF FUNCTIONS
There are many situations in which a quantity is given as a function of one variable
that, in turn, can be written as a function of a second variable. Suppose, for example,
that your job is to ship * packages of a product via Federal Express to a variety of
addresses. Let * be the number of packages to ship, and let /(*) be the weight of the
x objects and g(ie) be the cost of shipping a package of weight w. Then
(f °g)(x) = f[g(x)]
for each x in the domain of g for which g(;t) is in the domain of /.
Solution
The function / o g is defined by /[gQt)].
L(p) = 0J0y/p2 + 3
parts per million (ppm). A second study predicts that t years from now, the population
will be p(t) = 1 + 0.02?3 hundred thousand people. Assuming these formulas are
correct, what level of air pollution should be expected in 4 years?
Solution
The level of pollution is L(p) = 0.70^p2 + 3, where p(t) = 1 + 0.02r3. Thus, the
pollution level at time t is given by the composite function-
Solution
There are often many ways to express fix) as a composite g[«(x)]. Perhaps the most
natural is to choose u to represent the “inner” portion of / and g as the “outer" portion.
Such choices are indicated in the following table.
GRAPH OF A FUNCTION
Graphs have visual impact. They also reveal information that may not be evident
from verbal or algebraic descriptions. Two graphs depicting practical relationships
are shown in Figure 1.19.
Graph of a Function The graph of a function / consists of points whose coordinates (x, y) satisfy
y = f(x), for all x in the domain of /.
In Chapter 4, we will discuss efficient techniques involving calculus that you can
use to draw accurate graphs of functions. In algebra you began sketching lines by plot¬
ting points, but you quickly discovered that this is not a very efficient way to draw more
complicated graphs, especially without the aid of a graphing calculator or computer.
Table 1.3 includes a few common graphs you have probably encountered in previous
courses. We will assume that you are familiar with their general shape and know how
to sketch each of them either by hand or with the assistance of your graphing calcu¬
lator. There are, however, certain functions, such as f(x) — sin that are hard to
graph even with a calculator or computer. That said, be assured that most functions
that appear in this text or in practical applications can be graphed.
Vertical Line Test By definition of a function, for a given x in the domain there is
only one number y in the range. Geometrically, this means that any vertical line x = a
crosses the graph of a function at most once. This observation leads to the following
useful criterion.
y. ♦
1
1
! ^
J.y X
i
i
i
Function Function Not a function Not a function
a. The graph of a function: No vertical line b. Not the graph of a function: The curve inter¬
intersects the curve more than once. sects at least one vertical line more than once.
Intercepts The points where a graph intersects the coordinate axes are called inter¬
cepts. Here is a definition.
Intercepts If the number zero is in the domain of / and /(0) = b, then the point (0, b) is
called the y-intercept of the graph of /. If a is a real number in the domain of
/ such that f(a) — 0, then (a, 0) is an x-intercept of /.
+ What This Says To find the x-intercepts, set y equal to 0 and solve for x.
To find the y-intercept, set x equal to 0 and solve for y.
1.3 Functions and Graphs 27
>’ -X y — xz
y = W y — yfx
Standard Reciprocal Function Standard Reciprocal Squared Function Standard Square Root Reciprocal Function
1 1 1
i r
2
y
-1-
1 X
-i -f - -2 4 8
1
J
Cosine Function Sine Function Tangent Function
Solution
The ^-intercept is (0, /(0)) = (0, 2). To find the x-intercepts, solve the equation
/(x) = 0. Factoring, we find that
—x1 T x T 2 = 0
x2 — x — 2 = 0
(JC + l)(x -2) = 0
x = —lorx = 2
Figure 1.21 Graph of
f f(x) = —x2 + x + 2 Thus, the x-intercepts are (—1,0) and (2, 0), and the y-intercept is (0, 2). The graph
using the intercepts of / is shown in Figure 1.21. ■
Symmetry There are two kinds of symmetry that help in graphing a function, as
shown in Figure 1.22 and defined in the following box.
Symmetry The graph of y = f(x) is symmetric with respect to the y-axis if whenever
P(x, y) is a point on the graph, so is the point (—x, y) that is the mirror image of
P about the y-axis. Thus, y-axis symmetry occurs if and only if /(—x) = /(x)
for all x in the domain of /. A function with this property is called an even
function.
The graph of y = /(x) is symmetric with respect to the origin if whenever
P(x, y) is on the graph, so is (—x, —y), the mirror image of P about the origin.
Symmetry with respect to the origin occurs when /(—x) = -/(x) for all x. A
function that satisfies this condition is called an odd function.
WARNING _. . r
1 he graph or a nonzero Figure 1.22 Graphs of even and odd functions
function cannot be symmetric with
respect to the x-axis. In other words, if
the curve is symmetric with respect to There are many functions that are neither odd nor even. You may wonder why
the x-axis, such as the circle, then it will we have said nothing about symmetry with respect to the x-axis, but such symmetry
not be a function.
would require /(x) = —/(x), which is precluded by the vertical line test (do you see
why?).
1.3 Functions and Graphs 29
Solution
a. /(a) = a2 is even, because
The graphs of these three functions are shown in Figure 1.23. Note the symmetry in
the first two graphs and the lack of symmetry with respect to the axis or the origin in
the third.
y■ L
1
\ ■J
/
V 7
\ /
-3
<
-1
/
x
CLASSIFICATION OF FUNCTIONS
We will now describe some of the common types of functions used in this text.
Rational Function A rational function is the quotient of two polynomial functions, p(x) and d(x),
which can be written in the form
= d(x)^ 0
d(x)
When we write d(x) ^ 0, we mean that all values c for which die) = 0 are excluded
from the domain of d.
If r is any nonzero real number, the function fix) — xr is called a power function
with exponent r. You should be familiar with the following cases:
Power functions can also have irrational exponents (such as \Jl or n), but such func¬
tions must be defined in a special way (see Section 2.4).
A function is called algebraic if it can be constructed using a finite number of al¬
gebraic operations (such as adding, subtracting, multiplying, dividing, or taking roots)
starting with polynomials. Functions that are not algebraic are called transcendental.
The following are transcendental functions:
Trigonometric functions are the functions sine, cosine, tangent, secant, cosecant,
and cotangent. The basic forms of these functions are reviewed in Chapter 4 of the
Student Mathematics Handbook. You can also review these functions by consult¬
ing a trigonometry or precalculus textbook.
Exponential functions are functions of the form fix) = bx, where b is a positive
constant, b / 1. We will introduce these functions in Section 2.4.
Logarithmic functions are functions of the form fix) = logfcx, where b is a
positive constant, b 7M. We will also study these functions in Section 2.4.
1.3 Functions and Graphs 31
19. fix) = -
x
x+ \
20. fix)
x - 1 Problem 52
State whether the functions f and g in Problems 21-26 are equal.
2x2 + x Figure 1.24
21. fix) = ; gix) = 2x + 1
52. If point B in Figure 1.24 has coordinates (3, g(3)), what are
2x2 + x
22. fix) = ; gix) — 2x + 1, x / 0 the coordinates of R and 5?
Find the x-intercepts, if any, for the functions given in Problems
2x2 — 6
23. fix) = ; gix) = 2x + 3, x 7^ 2 53-60.
(3x + l)(x ■ 6) 1
g(x) = ,x # 2 59. fix) =
x —6 x2 + 2
Classify the functions defined in Problems 27-34 as even, odd, or x(x2 - 3)
neither.
60. fix)
x2 + 5
32 Chapter 1 Functions and Graphs
Q 61. Suppose the total cost (in dollars) of manufacturing q units of 65. At a certain factory, the total cost of manufacturing q units
a certain commodity is given by during the daily production run is C(q) = q2 + q + 900 dol¬
lars. On a typical workday, the number of units manufactured
C(q) = q2 - 3(V + 400? + 500 during the first t hours of a production run can be modeled by
the function q(t) = 25/.
a. Compute the cost of manufacturing 20 units. a. Express the total manufacturing cost as a function of t.
b. Compute the cost of manufacturing the 20th unit. b. How much will have been spent on production by the end
62. An efficiency study of the morning shift at a certain factory in¬ of the third hour?
dicates that an average worker who arrives on the job at 8:00 c. When will the total manufacturing cost reach $11,000?
A.M. will have assembled 66. A ball is thrown directly upward from the edge of a cliff in
such a way that t seconds later, it is
/(a) — —a3 + 6a + 15a2
s — — 16t2 + 961 + 144
CD players a hours later (0 < a < 8).
feet above the ground at the base of the cliff. Sketch the graph
a. How many players will such a worker have assembled by of this equation (making the f-axis the horizontal axis) and
10:00 A.M.? then answer these questions:
b. How many players will such a worker assemble between a. How high is the cliff?
9:00 A.M. and 10:00 a.m.? b. When (to the nearest tenth of a second) does the ball hit the
63. In physics, a light source of luminous intensity K candles is ground at the base of the cliff?
said to have illuminance I = K/s2 on a flat surface 5 ft away. c. Estimate the time it takes for the ball to reach its maximum
Suppose a small, unshaded lamp of luminous intensity 30 can¬ height. What is the maximum height?
dles is connected to a rope that allows it to be raised and low¬ 67. Charles’s law for gases states that if the pressure remains con¬
ered between the floor and the top of a 10-ft-high ceiling. As¬ stant, then
sume that the lamp is being raised and lowered in such a way
that at time t (in min) it is s — 6t — t2 ft above the floor. v(n = v„.(i + |5)
a. Express the illuminance on the floor as a composite func¬ where V is the volume (in.3), Vo is the initial volume (in.3),
tion of t for 0 < t <6. and T is the temperature (in degrees Celsius).
b. What is the illuminance when t = 1? When t = 4? a. Sketch the graph of V(T) for Vo = 100 and T > —273.
64. Biologists have found that the speed of blood in an artery is a b. What is the temperature needed for the volume to double?
function of the distance of the blood from the artery’s central 68. To study the rate at which animals learn, a psychology student
axis. According to Poiseuille’s law, the speed (cnr3/sec) of performed an experiment in which a rat was sent repeatedly
blood that is r cm from the central axis of an artery is given through a laboratory maze. Suppose that the time (in minutes)
by the function required for the rat to traverse the maze on the nth trial was
approximately
S(r) = C • (R2 -r2) 12
f(n) = 3-1-
n
where C is a constant and R is the radius of the artery.* Sup¬ 12
pose that for a certain artery, C = 1.76 x 105 cm/sec and a. What is the domain of the function /(a) = 3 + — ?
A
R = 1.2 x 10-2 cm. (See Figure 1.25.) b. For what values of n does /(n) have meaning in the con¬
text of the psychology experiment?
c. How long did it take the rat to traverse the maze on the
third trial?
d. On which trial did the rat first traverse the maze in 4 min¬
utes or less?
e. According to the function /, what will happen to the time
required for the rat to traverse the maze as the number of
trials increases? Will the rat ever be able to traverse the
maze in less than 3 minutes?
69. The trajectory of a cannonball shot from the origin with ini¬
tial velocity v and angle of inclination a (measured from level
ground), is given by the equation
Figure 1.25 Artery
y = mx — 16u-2(1 + m2) x2
a. Compute the speed of the blood at the central axis of this
where m = tana. For this problem assume the initial velocity
artery.
is v = 200 ft/s and that the angle of inclination is 42°.
b. Compute the speed of the blood midway between the a. Using a graphing utility of a calculator, determine the max¬
artery’s wall and central axis.
imum height reached by the cannonball.
*The law and the unit poise, a unit of viscosity, are both named for the b. Estimate the point where the cannonball will hit the
French physician Jean Louis Poiseuille (1799-1869). ground.
1.4 Inverse Functions; Inverse Trigonometric Functions 33
c. Which curve in Table 1.3 shows the basic shape of this 73. I CA L < One of the
graph? best known mathematical theorems is
70. Draw the path of a cannonball (see Problem 69) for an angle of the Pythagorean theorem, named
inclination of 47°. Determine the angle a that will maximize after the Greek philosopher Pythago¬
the distance the cannonball will travel. ras. Very little is known about the
71. It is estimated that t years from now, the population of a cer¬ life of Pythagoras, but we do know
tain suburban community will be he was born on the island of Samos.
He founded a secret brotherhood called
P(t) =20- — the Fythagoreans that continued for at
t + 1 least 100 years after Pythagoras was ca. 500 B.c.
thousand people. murdered for political reasons. Even
a. What will the population of the community be nine years though the cult was called a "brotherhood," it did admit
from now? women. According to Lynn Osen in Women in Mathematics,f
b. By how much will the population increase during the ninth the order was carried on by his wife and daughters after his
year? death. In fact, women were probably more welcome in the
centers of learning in ancient Greece than in any other age
c. What will happen to the size of the population in the “long
from that time until now. ■
run”?
72. Journal problem: The Mathematics Student Journal.* ■ Given State and prove the Pythagorean theorem.
that /(11) = 11 and
fix) - 1
fix + 3)
fix) + 1
for all x, find /(2000).
*VoIume 28 (1980), issue 3, p. 2. Note that the journal problem requests ' MIT Press, Cambridge MA, 1975.
/(1979), which, no doubt, was related to the publication date. We have
taken the liberty of updating the requested value.
-•
INVERSE FUNCTIONS
WARNING For a given function /, we write >'o = fix o) to indicate that / maps the number xo in
The symbol / 1 means the
its domain into the corresponding number yo in its range. If / has an inverse f~x, it is
inverse of / and does not mean
the function that reverses the effect of / in the sense that
reciprocal, 1// •
f 1 (To) = *o
For example, if
In the case where the inverse of a function is itself a function, we have the follow¬
ing definition.
34 Chapter 1 Functions and Graphs
Inverse Function Let / be a function with domain D and range R. Then the function / 1 with
domain R and range D is the inverse of / if
and
— y for all y fo ^
The inverse function /_1 reverses the effect of /. This relationship can be illus¬
trated by function “machines,” as shown in Figure 1.27. Note that the value input into
the first “/ machine” is x and the output /(x) from this machine is the input into a
second “/_1 machine” whose output is the original input value, x.
X
The language of this definition suggests that there is only one inverse function of
/. Indeed, it can be shown (see Problem 54) that if / has an inverse, then the inverse
is unique.
Solution
The inverse simply reverses the ordered pairs:
Solution
To find f~x, let y = fix), then interchange the x and y variables, and finally solve
for y.
Inverse: y = f(jc + 3)
Thus, we represent the inverse function as f~\x) = ^(jc + 3). To verify that these
functions are inverses of each other, we note that
and
rl[f(x)] = r\2x - 3) = \[{2x - 3) + 3] = \(2x) = x
for all x. ■
The inverse of a function may not exist. For example, both the functions
do not have inverses because if we attempt to find the inverses, we obtain relations that
are not functions. In the first case, we find
Possible inverse of /: {(0, 0), (1, 1), (1, —1), (4, 2), (4, —2)}
This is not a function because not every member of the domain is associated with a
single member in the range: (1,1) and (1, —1), for example.
In the second case, if we interchange the x and y in the equation for the function
g where y — x2 and then solve for y, we find
x — y2 or y = ±s/x forx>0
But this is not a function of jc, because for any positive value of x, there are two
corresponding values of y, namely, y/x and — yfx.
A function / will have an inverse /_1 on the interval I when there is exactly one
number in the domain associated with each number in the range. That is, /~' exists if
f(xi) and f(x2) are equal only when x\ = X2. A function with this property is said to
be one-to-one. This is equivalent to the graphical criterion, called the horizontal line
test, shown in Figure 1.28.
so there is no way to have f(x\) — f(x2) unless x\ = x2. This observation is formal¬
ized in the following theorem.
Proof We have already commented on why exists. To show that /~' is strictly
increasing whenever / is increasing, let yi and y2 be numbers in the range of /, with
y2 > yi. We will show that f~l(y2) > /_1(ji)- Because yj, y2 are in the range
of /, there exist numbers x\, x2 in the domain I such that y\ — f(x\) and y2 —
f(x2). Because y2 > y\, it follows that f(x2) > f {x 1), and because / is strictly
increasing, we must have x2 > x\. Thus, f~x{y2) > f~'(y 1), and f~x is strictly
increasing. Similarly, if / is strictly decreasing, then so is f~x. (The details are left
for the reader.) J
GRAPH OF/-1
The graphs of / and its inverse /“' are closely related. In particular, if (a, b) is a point
on the graph of /, then b = f(a) and a = f~x{b), so (b, a) is on the graph of f~x.
It can be shown that (a, b) and (b, a) are reflections of one another in the line y = x.
Figure 1.29 The graphs of / and / 1 (See Figure 1.29.) These observations yield the following procedure for sketching the
are reflections in the line y — x graph of an inverse function.
The trigonometric functions are not one-to-one, so their inverses do not exist. However,
if we restrict the domains of the trigonometric functions, then the inverses exist on
those domains. In trigonometry you probably distinguished between the sine curve
with unrestricted domain and the sine curve with restricted domain by writing y =
sinx and y = Sinx, respectively. In calculus, it is not customary to make such a
distinction by using a capital letter.
Let us consider the sine function first. We know that the sine function is strictly
increasing on the closed interval [— j, f ], and if we restrict sinx to this interval, it
does have an inverse, as shown in Figure 1.30.
1.4 Inverse Functions; Inverse Trigonometric Functions 37
7t TC
Inverse Sine Function y — sin 1 x if and only if x = sin y and-< y <
2~ y ~ 2
The function sin 1 x is sometimes written arcsinx.
WARNING
The function sin 1 x is Inverses of the other five trigonometric functions may be constructed in a similar
NOT the reciprocal of sinx. To denote manner. For example, by restricting tanx to the open interval (—f, f) where it is
the reciprocal, write (sinx)-1. In other one-to-one, we can define the inverse tangent function as follows.
words, sin-1 x ^ ——, whereas
sinx
I
(sinx)' 7T 7T
sinx
y = tan 1 x if and only if x = tan y and -< V < —
2 ' 2
Inverse Tangent Function The function tan-1 x is sometimes written arctanx.
Quadrants I and II
y, k y k
jL
1
1
1
1
1
1
i
i
1
1
/
f 1 l»
-3 -1 1 i -3 -1 1
_K _K_ _ ~
1 1 ~
L
b. y = cot 'x c. y = sec 1x d. y = esc be
all x M>i hl>i
0<y<n 0 <y < n 0<y<7t,_y*| -f <y<ly*0
Solution
a. sin-1(^) = -f Think: x = is negative, so y is in Quadrant IV;
the reference angle is the angle whose sine is it
is f, so in Quadrant IV the angle is — |.
b. sin-1 0.21 ^ 0.2115750 By calculator; be sure to use radian mode and inverse
sine (not reciprocal).
The definition of inverse functions yields four formulas, which we call the inversion
formulas for sine and tangent.
WARNING Tl . . , , ,
The inversion Formulas tor Inversion Formulas
sin-1 and tan-1 are valid only on the
specified domains. sin(sin_1 x) = x for — 1 < x < 1
Solution
a. sin(sin_1 0.5) = 0.5, because —1 < 0.5 < 1.
b. sin(sin_1 2) does not exist, because 2 is not between —1 and 1.
c. sin-1 (sin 0.5) = 0.5, because — | < 0.5 < j.
d. sin-1 (sin 2) = 1.1415927, by calculator.
For exact values, notice that
It is true that
sin_1(—x) = — sin_1(x)
but in general,
cos-1 (—x) 7^ cos-1 x
Solution
a. Let y = sin-1 (—x)
sin y = — x Definition of inverse sine
— sin y — x
'"R
ii
y — — sin-1 x
b. Let a = sin-1 x, so sin a = x, where — | < a < Here we consider the case
where 0 < a < and the other case where — j < a < 0 is handled similarly.
Construct a right triangle with an acute angle a. and hypotenuse 1, as shown in
Figure 1.33. We call this triangle a reference triangle.
The side opposite a is x (since sin a = x), and by the Pythagorean theorem,
the adjacent side is Vl — x2. Thus, we have
cos(sin_1 x) = cos cr
1 — x~
Figure 1.33 A reference triangle
~ T
Substitute cos cr = x and choose the
= yj\ — X2 positive value for the radical because
a is in Quadrant I. ■
Reference triangles, such as the one shown in Figure 1.33, are extremely use¬
ful devices for obtaining inverse trigonometric identities. For instance, let a and f
be angles of a right triangle with hypotenuse 1. If the side opposite a is jc (so that
since = x), then a = sin-1 x and f> = cos-1 x so that
since the acute angles of a right triangle must sum to tt/2. The same reasoning can
also be used to show that
7T
tan 1 x + cot 1 x =
2
and
71
sec-1 x + esc-1 x =
2
Identities involving inverse trigonometric functions have a variety of uses. For
instance, most calculators have keys for evaluating sin-1 x, cos-1 x, and tan-1 x, but
what about the other three inverse trigonometric functions? The answer is given by
the following theorem, which allows us to compute sec-1 x, esc-1 x, and cot-1 x using
reciprocal identities involving the three inverse trigonometric functions the calculator
does have.
Proof We will prove that sec 1 x = cos 1 — and leave the other two parts as an
x
exercise. Let y = sec-1 x. Then secy = x and by using the reference triangle in
-l _i 1
y = sec x — cos —
x
Figure 1.34 Reference triangle Note that the reference triangle is valid only if |x| > 1. □
1.4 Inverse Functions; Inverse Trigonometric Functions 41
Solution
a. sec_1(—3) = cos_1(—2) ~ 1.910633236
b. esc”1 7.5 = sin-1 (73) ~ 0.1337315894
c. cot"1 2.4747 = § - tan"1 2.4747 « 0.3840267299 ■
6. fix) = -,r^0; gix) = -,r/0 Q Sketch the graph of f in Problems 33-38 and then use the horizon¬
x x
tal line test to determine whether f has an inverse. If f~x exists,
7. fix) — x2, x < 0; gix) = yfx, x > 0
sketch its graph.
8. fix) = x2, x > 0; gix) y/x, x > 0
33. fix) — x2, for all x 34. fix) = x2, x < 0
Find the inverse (if it exists) of each function given in Problems
35. fix) = Vl — x2, on (—1, 1)
9-18.
36. fix) = x(x — 1 )(x — 2), on [1, 2]
9. / = {(4,5), (6, 3), (7,1), (2. 4)}
37. fix) = cosx, on [0, n]
10. g = {(3,9), (-3, 9), (4, 16), (-4, 16)}
38. fix) = tanx, on (—f, f)
11. fix) = 2x + 3 12. g{x) = -3x + 2
Simplify each expression in Problems 39-44.
13. fix) = x2 — 5, x > 0 14. gix) — x2 — 5, x <0
39. sin(2tan_1x) 40. tan(2tan“'x)
15. Fix) = y/x + 5 16. Gix) = 10 - yfx
41. tan(cos_1 x) 42. cos(2sin“'x)
2x — 6 2x + 1
17. hix) = 18. hix) = 43. sin(sin_l x + cos-1 x) 44. cos2(sin-'x + cos-'x)
3x H- 3 x
45. Use reference triangles, if necessary, to justify each of the fol¬
Give the exact values for functions in Problems 19-30.
lowing identities,
19. a. cos-1 j b. sin-1(—^) a. cor1 x = | — tan-1 x for all x
20. a. sin-1 (—2) b. cos_I(—|)
b. esc 1 x = sin for all |x| > 1
21. a. tan_1(—1) b. cot_1(—a/3) \x /
22. a. sec_l(—\/2) b. esc *(—s/2) 46. Use the identities in Theorem 1.3 to evaluate each function
rounded to four decimal places,
23. a. sin_1(-^) b. sec~'(—1)
a. cot-1 0.67 b. sec-1 1.34
24. a. sec_,(^) b. cot-1 (— 1)
c. esc-1 2.59 d. cot-1 (—1.54)
25. cos(sin_l 2) 26. sin (cos -1 _L
s/2
) 47. Use the identities in Theorem 1.3 to evaluate each function
28. tan(sin 1 2) rounded to four decimal places.
27. cot(tan-1 2)
a. cot-1 1.5 b. cot-1 (—1.5)
29. cos(sin_l 2 + 2cos_1 2)
c. sec_1(—1.7) d. csc—l (— 1.84)
(Imh) Hint: Use the addition law for cos(a + f). 48. A painting 3 ft high is hung on a wall in such a way that its
lower edge is 7 ft above the floor. An observer whose eyes are
30. sin(sin 1 2 + cos 1 2)
5 ft above the floor stands x ft away from the wall. Express
(jmIT) Hint: Use the addition law for sin(a + ft). the angle 0 subtended by the painting as a function of x.
42 Chapter 1 Review
49. To determine the height of a building (see Figure 1.35), select 52. We proved the identity
a point P and find the angle of elevation to be a. Then move
out a distance of x units (on a level plane) to point Q and find esc-1 x = sin-1 for all |x| > 1
that the angle of elevation is now /3. Find the height h of the
building, as a function of x.
By examining the graph of y = esc-1 x, conjecture an identity
B
of the form
@ 50. a. Find tan-1 1 + tan-1 2 + tan-1 3 using a calculator. Make The graph of y = sec 1 x suggests that
a conjecture about the exact value. Prove your conjecture
using reference triangles. sec-1 (—x) =7r + sec-1 x for x > 1
b. Prove the conjecture from part a using trigonometric iden¬
Either prove this identity or find a counterexample.
tities. Hint: Find tan(tan_1 1 + tan-1 2 + tan-1 3).
54. Show that if /-1 exists, it is unique. Hint: If g\ and g2 both
51. Prove the conjecture from Problem 50a using right triangles
satisfy
as follows. You may use the figure shown in Figure 1.36; as¬
sume that A ABC, AABD, and A DEF are all right triangles
gi [/(*)] = x = /[gi(x)j
with lengths of sides as shown in the figure.
g2[/0)] — x = f[g2(x)]
B
then g\ (x) = g2(x).
55. Suppose that A ABC is not a right triangle but has an obtuse
angle y. Draw BD perpendicular to AC, forming right trian¬
gles AABD and ABDC (with right angles at D). Show that
sin a sin y
Figure 1.36 6\ + 62 + #3 = n a c
-•
o. inverse sine function 6. the circle with center on the y-axis that passes through the ori¬
p. inverse tangent function gin and is tangent to the line 3x 4- 4y — 40 = 0
q. inverse secant function 7. the line through the two points where the circles
r. inverse cosecant function x2 + y2 — 5x 4- 7y = 3 and x2 + y2 + 4y = 0 intersect
/ jt \ A + tan x
s. inverse cotangent function 8. Find constants A and B so that tan x H-= -.
V 3/ 1 4- B tan x
12. What is a polynomial function?
9. Find constants A and B so that sin’ x = A sin 3x + B sinx.
13. What is a rational function?
10. In a triangle, the perpendicular segment drawn from a given
14. a. Define an inverse function.
vertex to the opposite side is called the altitude on that
b. What is the procedure for graphing the inverse of a given side. Consider the triangle with vertices (—2, 1), (5, 6), and
function? (3, —2). Find an equation for each line containing an altitude
15. What is the horizontal line test? of this triangle.
16. State the inversion formulas for sine and tangent. 11. Show that the three lines found in Problem 10 intersect at the
17. State the reciprocal inverse trigonometric identities. same point. Find the coordinates of this point. Is this the same
point where the three medians meet?
PRACTICE PROBLEMS 12. Let fix) = x2 + 5x — 9. For what values of x is it true that
18. Find an equation for the lines satisfying the given conditions: /(2x) - /(3x)?
a. through (—|, 5) with slope m — — | 13. If an object is shot upward from the ground with an initial ve¬
b. through (—3,5) and (7, 2) locity of 256 ft/sec, its distance in feet above the ground at the
c. with x-intercept (4, 0) and y-intercept (0, — |) end of t seconds is given by sf) = 2561 — 16?2 (neglecting
air resistance). What is the highest point for this projectile?
d. through (— 5) and parallel to the line 2x 4- 5y — 11 = 0
14. It is estimated that t years from now, the population of a cer¬
e. the perpendicular bisector of the line segment joining tain suburban community will be
Pi—3, 7) and Q(5, 1)
Sketch the graph of each of the equations in Problems 19-26. 111 + 12
Pit) =
2/4-3
19. 3x + 2y — 12 = 0 20. y - 3 = \x + 11
21. y —3 = -2(x- l)2 22. y = x2 -Ax - 10 thousand people. What is the current population of the com¬
23. y = 2cos(x — 1) 24. y 4- 1 = tan(2x + 3) munity? What will the population be in 6 years? When will
there be 5,000 people in the community?
25. y = sin~'(2x) 26. y = tan-1 x2
15. Evaluate each of the given numbers (calculator approxima¬
27. If f(x) = —-—, what value(s) of x satisfy tions).
x + 1
1 \ / 2x + 1 a. tan 2 b. cot 1 2 c. sec '(—3.1)
f = f 16. Evaluate each of the given numbers (exact values),
x + 1 2x +4
a. sin(cos-1 ^)
4
b. sin(2tan-13)
28. If f(x) - —and g(x) = —=J=, does f = g? Why -1 5
V x — 1 V* - 1 c. sin (cos-1 | + sin-1 ^|)
or why not?
17. Solve yfx = cos '0.317 4-sin '0.317
29. If fix) = sinx and g(x) = Vl - x2, find the composite
In Problems 18-21, find /-1 if it exists.
functions fog and go/.
30. An open box with a square base is to be built for $96. The 18. fix) = 2x3 - 7 19. fix) = 3/2x + 1, x >
sides of the box will cost $3/ft2 and the base will cost $8/ft2. x +5
20. fix) = Vsinx, 0 < x < f 21. fix) —--,
Express the volume of the box as a function of the length of x -7
its base. x +a
22. Show that for any constant a ^ 1, the function fix) =
x - 1
is its own inverse.
Supplementary Problems' ax + b
In Problems 1-3, find the perimeter and area of the given figure. 23. Let fix) , for constants a, b, c, and d. Find / 1 (x)
cx + d
1. the right triangle with vertices ( — 1. 3), (—1, 8), and (11,8) in terms of a, b, c, and d.
X 4- 1
2. the triangle bounded by the lines y = 5, 3y — 4x = 11, and 24. Find /-1(x) if fix) =-. What is the domain of /“'?
12x + 5y = 25 x — 1
3. the trapezoid with vertices A(—3.0), #(5,0), C(2, 8), and x +y
25. First show that tan-1 x 4- tan-1 y = tan” for
D(0, 8) 1 -xy
TC
In Problems 4-7, find an equation for the indicated line or circle. xy 1 whenever — — c tan -■ (1+1.) < 1. Then es-
Vl-W 2
4. the vertical line through the point where the line y — 2x —1 tablish the following equations.
intersects the parabola y = x2 4- 6x — 3
a. tan i1 \1 + tan 1 |i — |
tL 5. 2tan-1 | 4-tan-1 |
5. the circle that is tangent to the x-axis and is centered at (5, 4)
c. 4 tan-1 j - tan-1 ^ |
*The supplementary problems are presented in a somewhat random order, Note: The identity in part c will be used in Chapter 8 to esti¬
not necessarily in order of difficulty. mate the value of n.
44 Chapter 1 Review
26. Counterexample Problem Each of the following equations may 31. Find the volume and the surface area of each of the following
be either true or false. In each case, either show that the equa¬ solid figures:
tion is generally true or find a counterexample.
a. a sphere with radius 4
sin-1 x 1
a. tan x =-;— b. tan x =- b. a rectangular parallelepiped (box) with sides of length 2, 3,
cos_1x tanx
and 5
c. cot~' x = | — tan-1 x d. cosfsin-' x) = Vl — x2
c. a right circular cylinder (including top and bottom) with
e. sec~1 (7) = cos-1 x
height 4 and radius 2
27. Many materials, such as brick, steel, aluminum, and concrete,
d. an inverted cone with height 5 and top radius 3 (lateral sur¬
expand with increases in temperature. This is why spaces are
face only)
placed between the cement slabs in sidewalks. Suppose you
have a 100-ft length of material securely fastened at both ends, 32. Consider the triangle with vertices A(—1,4), B(3,2), and
and assume that the buckle is linear. (It is not, but this assump¬ C(3, —6). Determine the midpoints M, and M2 of sides AB
tion will serve as a worthwhile approximation.) If the height and AC, respectively, and show that the line segment M\M2
of the buckle is x ft and the percentage of swelling is y, then is parallel to side BC with half its length.
x and y are related as shown in Figure 1.37. Find the amount
of buckling (to the nearest inch) for the following materials: 33. Generalize the procedure of Problem 32 to show that the line
a. brick; y = 0.03 [This means (0.03%)(100 ft) = 0.03 ft, segment joining the midpoints of any two sides of a given tri¬
which is y in Figure 1.37.] angle is parallel to the third side and has half its length.
b. steel; y = 0.06 34. Let ABCD be a quadrilateral in the plane, and let P, Q, R,
c. aluminum; y = 0.12 d. concrete; y = 0.05 and S be the midpoints of sides AB, BC, CD, and DA, as
shown in Figure 1.38. Show that PQRS is a parallelogram.
28. A ball has been dropped from the top of a building. Its height
(in feet) after t seconds is given by the function
600x will have a v-intercept of (0, —5). What are the x-intercepts
fix) =
300 - x of the graph?
a. What is the domain of the function /? 36. A bus charter company offers a travel club the following ar¬
b. For what values of x does /(x) have a practical interpreta¬ rangements: If no more than 100 people go on a certain tour,
tion in this context? the cost will be $500 per person, but the cost per person will
c. How many worker-hours were required to distribute new be reduced by $4 for each person in excess of 100 who takes
telephone books to the first 50% of the households? the tour.
d. How many worker-hours were required to distribute new a. Express the total revenue R obtained by the charter com¬
telephone books to the entire community? pany as a function of the number of people who go on the
e. What percentage of the households in the community had tour.
received new telephone books by the time 150 worker- b. Sketch the graph of R. Estimate the number of people that
hours had been expended? results in the greatest total revenue for the charter company.
30. Find the area of each of the following plane figures:
37. A mural 7 feet high is hung on a wall in such a way that its
a. the circle with Pi0. 0) and (2(2, 3) endpoints of a diameter
lower edge is 5 feet higher than the eye of an observer stand¬
b. the trapezoid with vertices A(0, 0), BiA, 0), C(l, 3), and ing 12 feet from the wall. (See Figure 1.39.) Find the angle 9
Di2, 3) subtended by the mural at the observer’s eye.
Chapter 1 Review 45
46
Guest Essay 47
together with examples of each, such as the construction of tangent lines to a circle
and the area under a parabolic curve. The Hindu and Arabic mathematicians had ex¬
tended the number system and the formal language of algebra with certain of its laws
by 1300 A.D., but it was not until Newton’s own era that the methods of algebra and
geometry were combined satisfactorily in the analytic geometry of Rene Descartes
(1596-1650) to produce the recognition that a geometrical curve could be regarded as
the locus of points whose coordinates satisfied an algebraic equation. This provided
potential numerical exactitude to geometric constructions as well as the possibility of
giving geometric proofs for limiting algebraic arguments. Since Euclidian geometry
was then generally regarded as the only reliable mathematics, it was the latter direction
that was most frequently taken. And it was this direction that was taken in the seven¬
teenth century by de Roberval, Fermat, Cavalieri, Huygens, Wallis, and others, not the
least of whom was Newton’s own teacher at Cambridge, Isaac Barrow (1630-1677).
These giants, as Newton called them, obtained equations for tangent lines to, and the
(correct) areas under, polynomial curves with equations y = x" for n — 1,2, 3,..., 9,
and certain other geometrically defined curves such as the spiral and the cycloid.
In his analysis of tangency, Barrow incorporated the approximating infinitesimal
triangle with sides Ax, Ay, As that is now standard in expositions of calculus, and
Cavalieri attempted to “count” an indefinite number of parallel equidistant lines to
obtain areas. It was known that in specific cases these problems were related, and
that they were equivalent, respectively, to the kinematic problems of characterizing
velocity and distance traveled during a motion, problems which directly confronted
the paradoxes of Zeno.
All that remained was for some mathematicians to sense the generality underlying
these specific constructions and to devise a usable notation for presenting the results.
This was accomplished essentially independently by Newton (who, justly mistrusting
the required limiting arguments, suppressed his own contributions until he could val¬
idate them geometrically) and by the only slightly less cautious Leibniz. However,
as we have argued, by this time (about 1670), it was almost inevitable that someone
should do so.
What calculus has provided is a mathematical language that, by means of the
derivative, can describe the rates of change used to characterize various physical pro¬
cesses (such as velocity) and, by means of the integral, can show how macroscopic
entities (such as area or distance) can emerge from properly assembled microscopic el¬
ements. Moreover, the fundamental theorem, which states that these are inverse opera¬
tions, supplies an exact method for passing between these types of description. Finally,
the ability to relate the results of limiting arguments by simple algebraic formulas per¬
mits the correct use of calculus while retaining skepticism regarding its foundations.
This has enabled applications to go forward while mathematicians have sought an ap¬
propriate axiomatic basis.
Our present technological age attests to the success of this endeavor, and to the
value of calculus.
48 Chapt er 1 Mathematical Essays
ArcFiilles fires an arrow, but it can never reach) its target. WFiy? Because if the arrow
takes T seconds to travel half the distance to the target, it will take T/2 seconds to
finish) half the rest, and so on for each) half. No matter Flow close the arrow is to the
target, it will take finite time to travel half the distance that remains, so the arrow will
never reach its destination.
Common sense tells that the arrow will strike home in 2T seconds, so where is the
error in reasoning? If we measure the total time of the arrow’s flight by the “half,
then half of that, and so on,” approach of the paradox, it will take
T T
T + - + - + •■•
2 4
seconds for the arrow to strike the target; that is, an infinite sum of finite time inter¬
vals. The Greeks of Zeno’s time assumed that such a sum would have to be infinite.
Flowever, the limit process can be used to show that this particular sum is not only
finite, but equals 2T, as common sense would suggest. We will study infinite sums
in Chapter 8, but first it is necessary to introduce and explore a different kind of
limit, the limit of a function, and that is the goal of this chapter.
In a very real sense, the concept of limit is the threshold to modern mathemat¬
ics. You are about to cross that threshold, and beyond lies the fascinating world of
calculus.
49
50 Chapter 2 Limits and Continuity
The limit of a function / is a tool for investigating the behavior of f(x) as x gets closer
and closer to a particular number c. That is, the limit concerns the tendency of f(x)
for x near c rather than the value of / at x = c. The following example illustrates how
such a tendency can be used to compute velocity.
Solution
What we want is the instantaneous velocity after 2 seconds, which can be thought
of as the “speedometer reading” of the falling body at t = 2. To approximate the
instantaneous velocity, we compute the average velocity of the body over smaller and
smaller time intervals, either ending with or beginning with t = 2. For instance, over
the time interval 1.9 < t < 2, the average velocity, v{t), is
2-1.9
16(2)2 - 16(1.9)2
~ Ol
= 62.4 ft/s
Similar computations of average velocity over short time intervals ending or beginning
with time t = 2 are contained in the following table:
Time interval <t< 2 1.9 < t < 2 1.99 < r < 2 1.999 <t< 2 2 < t < 2.0001 2 < t < 2.001 2 < t < 2.01
Interval length (sec) 0.2 0.1 0.01 0.001 0.0001 0.001 0.01
Average velocity (ft/s) 60.8 62.4 63.84 63.98 64.0016 64.016 64.16
Examining the bottom row of this table, we see that the average velocity seems to
be approaching the value 64 as the time intervals become smaller and smaller. Thus, it
is reasonable to expect the velocity at the instant when t — 2 to be 64 ft/s.
In symbols, the average velocity of the falling body over the time interval
2 < t < 2 + h is given by
We say that the average velocity has a limiting value of 64 as the length h of the time
interval tends to zero (gets smaller and smaller), and we denote this tendency by writing
16(2 + h)2 - 16(2)'
lim = 64
h->0 h
is read “the limit of f(x) as x approaches c is L” and means that the functional
values / (x) can be made arbitrarily close to a unique number L by choosing x
sufficiently close to c (but not equal to c).
The numbers on the bottom row of this table suggest that fix) -3 as x -> -2;
that is,
jc2 + jc — 2
L = lim --— = —3
Figure 2.1 Graph of x—*—2 X + 2
jc2 + x — 2
The graph of / is shown in Figure 2.1. Notice that the graph is a line with a “hole” at
/(*) = x + 2
the point (—2, —3).
EXAMPLE 3 Finding limits of trigonometric functions
Evaluate lim sin* and lim cosx.
x—>0 x—>0
Solution
We can evaluate these limits by table.
2s/x + 1 - x - 2
L = lim---
x-yO Xz
Solution
Form a table of values of
2 V* + 1 — x — 2
fix) = -r-
X1
for x near 0:
The numbers on the bottom line suggest that the limit is L — —0.25.
In Example 4, note that if you decide to try x = 0.0000001 just to make sure you
have taken numbers close enough to 0, you may find that the calculator gives the value
0. Does this mean the limit is actually 0 instead of -0.25? No, the problem is that the
numerator 2s/x + 1 - x - 2 is so close to 0 when x = 0.0000001 that the calculator
gives a false value for /.
The point is this: When you use your calculator to estimate limits, always be aware
that this method can introduce errors. In the next section, we will develop an algebraic
approach for computing limits that is more reliable. The three basic approaches (ex¬
amining appropriate data by using a calculator and/or a table, drawing a graph to find a
limit, or using algebraic rules) each has its virtues and faults. As you study this section
and the next, you need to learn all three approaches, along with their virtues and faults,
and this knowledge will serve you well for the rest of your mathematical course work.
ONE-SIDED LIMITS
Sometimes we will be interested in the limiting behavior of a function from only one
side; that is, the limit as x approaches a number c from the left or the analogous limit
from the right.
2.1 The Limit of a Function 53
One-Sided Limits Right-hand limit We write lim f(x) = L if we can make the number f(x)
X—>c+
Left-hand limit We write lim f(x) = L if we can make the number /(x) as
X—*C~
For instance, note that the graph in Figure 2.2 is “broken” at x = 3. The limit as x
approaches 3 from the left is 1, while the limit from the right at the same point is -1.
Observe that the “two-sided” limit, lim h(x), defined earlier in this section, does
x—>3
not exist since h(x) does not tend toward a single limiting value L as x approaches
c = 3 from either side. It should be clear that, in general, a two-sided limit cannot
exist if the corresponding pair of one-sided limits are different. Conversely, it can be
shown that if the two one-sided limits of a given function / as x -> c~ and x -* c+
both exist and are equal, then the two-sided limit lim /(x) must also exist. These
x—>c
observations are so important that we state them in the form of a theorem.
Figure 2.3 We say that lim /(x) — L if and only if lim /(x) = lim /(x) — L.
x—>c x—>c~ x—*c+
Solution
a. Take a good look at the given graph; notice the open circles on the graph at x — 0
and x — -2 and also notice that /(0) = 5. To find lim /(x) we need to look at
x—>0
both the left-hand and right-hand limits. Look at Figure 2.4a to find
so lim /(x) exists and lim /(x) = 1. Notice here that the value of the limit as
x y0 x —y 0
x -»• 0 is not the same as the value of the function at x =0.
b. Look at Figure 2.4b to find
Solution
sinx
/(*) = -is an even function because
x
sin(—x) — sinx sinx
f(~x) = /(*)
—x —x x
This means that we need to find only the right-hand limit at 0 because the limiting
behavior from the left will be the same as that from the right. Consider the following
table.
smx , sinx
The table suggests that lim -= 1 and hence that lim-= 1. We will revisit
x->0+ X x-*0 X
cin X sinx
Figure 2.5 Graph of f(x) = -- this limit in Section 2.2. The graph of /(x) = is shown in Figure 2.5. ■
x X
2.1 The Limit of a Function 55
It may happen that a function / does not have a (finite) limit as x -* c. When lim /(x)
fails to exist, the functional values f(x) are said to diverge as x —> c.
Solution
x approaches 0 from the left; x—>0 . ->■ x approaches 0 from the right; ,r->0+.
o
1 x 104
X
100 400 1 x 106 undefined lxlO6
II
—► 0 <—
Infinite Limits A function / that increases or decreases without bound as .v approaches c is said
to tend to infinity (oo) at c. We indicate this behavior by writing
Solution
sinx 1
Note this is not the same as lim The values of /(x) sin — oscillate in-
*-►0 x x
finitely often between 1 and -1 as i approaches 0. For example, /(x) = 1 for
r _ 2 _2_ 2_ _2_
- _2_
A tt ’ W,’ Q-rr
9ff ’.
• • • , and /(*) = -1 for X =
5Stt * 3tr ' In ifc. •••• The graph of /(x)
is shown in Figure 2.7.
Because the values of /(x) do not approach a unique number L as x —» 0, the limit
does not exist. This kind of limiting behavior is called divergence by oscillation. ■
In the next section, we will introduce some properties of limits that will help us
evaluate limits efficiently. In the following problem set remember that the emphasis is
on an intuitive understanding of limits, including their evaluation by graphing and by
table.
Our informal definition of the limit provides valuable intuition and allows you to de¬
velop a working knowledge of this fundamental concept. For theoretical work, how¬
ever, the intuitive definition will not suffice, because it gives no precise, quantifiable
meaning to the terms “arbitrarily close to L” and “sufficiently close to c.” In the
nineteenth century, leading mathematicians, including Augustin-Louis Cauchy (1789—
1857) and Karl Weierstrass (1815-1897), sought to put calculus on a sound logical
foundation by giving precise definitions for the foundational ideas of calculus. The
following definition, derived from the work of Cauchy and Weierstrass, gives preci¬
sion to the limit notion.
means that for each number € > 0, there corresponds a number 8 > 0 with the
property that
For each e > 0 there is a 8 > 0 such that if 0 < |x-c| < 5, then [/'(jc) -L\ < e.
This forms an interval This forms an interval This says that if x is in the ... then f(x) is in
around L on the y-axis. around c on the x-axis. 5-interval on the x-axis... the e-interval on the y-axis.
W-
L1
►
L. _ IT 5 _ i
Because the Greek letters e (epsilon) and <5 (delta) are traditionally used in this con¬
text, the formal definition of limit is sometimes called the epsilon-delta definition of
the limit. The goal of this section is to show how this formal definition embodies
our intuitive understanding of the limit process and how it can be used rigorously to
establish a variety of results.
Do not be discouraged if this material seems difficult—it is. Probably your best
course of action is to read this section carefully and examine the details of a few ex¬
amples closely. Then, using the examples as models, try some of the exercises. This
material often takes several attempts, but if you persevere, you should come away with
an appreciation of the epsilon-delta process—and with it, a better understanding of
calculus.
Behind the formal language is a fairly straightforward idea. In particular, to estab¬
lish a specific limit, say lim /(x) = L, a number e > 0 is chosen first to establish a
X—>C
desired degree of proximity to L, and then a number 8 > 0 is found that determines
how close x must be to c to ensure that /(x) is within € units of L.
The situation is summarized in Figure 2.9, which shows a function that satisfies
the conditions of the definition. Notice that whenever x is within 8 units of c (but not
equal to c), the point (x, /(x)) on the graph of / must lie in the rectangle (shaded
Figure 2.9 The epsilon-delta definition
region) formed by the intersection of the horizontal band of width 2e centered at L
of limit
and the vertical band of width 28 centered at c. The smaller the e-interval around
the proposed limit L, generally the smaller the <5-interval will need to be for /(x) to
lie in the e-interval. If such a 8 can be found no matter how small e is, then /(x)
and L are arbitrarily close, so L must be the limit. The following examples illustrate
epsilon-delta proofs, one in which the limit exists and one in which it does not.
Solution
From the graph of /(x) = 4x - 3 (see Figure 2.10), we guess that the limit as x 2
is 5. The object of this example is to prove that the limit is 5. We have
|/(x) — L| = |4x — 3 — 5|
= |4x - 8|
= 4 lx -21
Solution
1
Let f[x) = - and L be any number. Suppose that lim /(x) = L. Look at the graph of
x x-*0
f, as shown in Figure 2.11. It would seem that no matter what value of € is chosen, it
would be impossible to find a corresponding 8. Consider the absolute value expression
required by the definition of limit: If
1
\f(x) — L\ < e, or, for this example, L < €
x
Figure 2.11 lim- then
jc-*0 X
1 Property of absolute value
—e <-L < €
x (Table 1.1, p. 3)
and
1
L —€ < - < L +€
X
1
< \L\ + 1
x
1
x\ >
L\ + 1
In general, given any € > 0, then no matter how <5 > 0 is chosen, there will always be
numbers x in the interval 0 < |x — 0| < 8 such that — > L + e. Since L was chosen
|x|
arbitrarily, it follows that the limit does not exist. ■
Appendix B gives many of the important proofs in calculus, and if you look there,
you will see many of them are given in €-8 form. In the following problem set, re¬
member that the emphasis is on an intuitive understanding of limits, including their
evaluation by graphing and by table.
2.1 The Limit of a Function 59
/(*) 15.33
fix) -* ?
fir 1
Graph of t Describe each illustration in Problems 11-16 using a limit state¬
\
ment.
Figure 2.12 Graphs of the functions /, g, and t
a. lim fix) b. lim f(x) C. lim fix)
x-+3 x—>2 .v—>-0
fix) 3
fix) -> ?
x3 - 8
8. lim g(x), where g(x) =
x—*2~ x2 + 2x + 4
x -* 2"
x 1 1.5 1.9 1.99 1.999 1.9999
cosx
26. a. lim 46. Tom and Sue are driving along a straight, level road in a car
*->0
x whose speedometer needle is broken but that has a trip odome¬
cosx
b. lim- ter that can measure the distance traveled from an arbitrary
*->* X
starting point in tenths of a mile. At 2:50 P.M., Tom says he
would like to know how fast they are traveling at 3:00 P.M.,
so Sue takes down the odometer readings listed in the table,
makes a few calculations, and announces the desired velocity.
What is her result?
1 — COS X
27. a. lim
*—>o X time t 2:50 2:55 2:59 3:00 3:01 3:03 3:06
1 — COS X
b. lim odometer
X—>Tl 33.9 38.2 41.5 42.4 43.2 44.9 47.4
reading
s(t) = -16r + 40r + 24 Q In Problems 50-55, use the format definition of the limit to prove
or disprove the given limit statement.
a. Compute the limit 50. lim(x + 3) = 5 51. lim(3t — 1) = 0
*—>2 r-> 0
a. Construct a table showing the values of /(x) for f(x)' forx = A tt-, :«c", sr-, Now what would
J ' 2tt ’ 11 n 20n ’ 50tt ’ 307r ’ 5n
^ 'y n n O'}
x — -.r.Tr.rr. r.T' Based on this table,’ what
7T ’ 9n ’ 137T ’ \9n ’ In ’ 3tt you say about lim /(x)?
*—►0
would you say about lim f (x)?
c. Based on the results in parts a and b, what do you conclude
b. Construct a second table, this time showing the values of about lim sin -?
x—>0 x
i
! Difference rule lim [f{x) - g(x)] = lim fix) - lim gix)
i x-+c jc—>c x-*c
i
i
i
1 , The limit of a difference is the difference of the limits.
1 X
* — > C i— K
b. Limit of x: lim x = c It is fairly easy to justify graphically the rules for the limit of a constant and the
x—*c
limit of x, as shown in Figure 2.13. All of these properties of limits (including the two
Figure 2.13 Two basic limits basic limits shown in Figure 2.13) can be proved using the formal definition of limit.
62 Chapter 2 Limits and Continuity
Solution
lim(2x5 - 9x3 + 3x2 — 11) = lim(2x5) — lim(9x3) + lim(3x2) — lim (11)
,v->2 x—>2 x—*2 x—>2 x—>2
- 5 3 -
Power rule
Solution
lim (z3 - 3z + 7)
z3 - 3z + 7 z->—1
lim Quotient rule
z-* —1 5z2 + 9z + 6 lim (5z2 + 9z + 6)
WARNING w z——l
You must be carerul as to
when you write the word "limit" and (— 1 )3 — 3(—1) + 7
when you do not; pay particular 5(—l)2 + 9(—1) + 6
attention to this when looking at the
examples in this section. 9 Both numerator and denominator
2 are polynomial functions. ■
Notice that if the denominator of the rational function is not zero, the limit can be found
by substitution. If the limit of the denominator is zero, we find ourselves dividing by
zero, an illegal operation.
Once again, for values of the function for which /(c) is defined, the limit can be found
by substitution.
In the previous section we used a table to find that lim sinx = 0 and lim cosx = 1.
x—>0 x—>0
In the following example we use this information, along with the properties of limits,
to find other trigonometric limits.
Solution
2
a. lim sin x — lim sinx Power rule
x—>0 x—0
lim sinx = 0
x—>0
The following theorem states that we can find limits of trigonometric functions by
direct substitution, as long as the number that x is approaching is in the domain of the
given function. Proofs of several parts are outlined in the problem set, and in fact an
examination of the graphs of the functions indicates why they are true.
Proof We will show that lim sinx = sine. The other limit formulas may be proved
X—>C
- sine
Note that sin c and cos c do not change as h —>• 0 because these are constants with
respect to h. □
Solution
- 2 -
a. lim(* cos ttx) = lim x lim cos jtx l2 COS 71 = —1
x^\ A'—> 1 x->\
lim *
x->0 0
b. lim = 0
jr^-ocos* lim cos* 1
x—>0
Solution
If you try substitution on this limit, you will obtain
If* = 2, then *2 + * — 6 = 0.
I
*2 + * - 6 0
lim- -
x-+2 * — 2 0
t
If* = 2, then * — 2 = 0.
The form is called an indeterminate form because the value of the limit cannot be
determined without further analysis.
If the expression is a rational expression, the next step is to try to simplify the
function by factoring and checking to see if the reduced form is a polynomial.
*2 + * — 6 (* + 3)(* - 2)
lim---- — lim = lim(* + 3)
x-+2 * — 2 x-*2 X — 2 x-*2
This simplification is valid only if * 7^ 2. Now complete the evaluation of the reduced
function by direct substitution. This is not a problem, because the limit is concerned
with values as x approaches 2, not the value where * = 2. In the preceding expression
x~ + x — 6
lim — lim(* + 3) = 5
x—>2 * — 2 x-*2
2.2 Algebraic Computation of Limits 65
Another algebraic technique for finding limits is to rationalize either the numerator
or the denominator to obtain an algebraic form that is not indeterminate.
Solution
Once again, notice that both the numerator and denominator of this rational expression
are 0 when x = 4, so we cannot evaluate the limit by direct substitution. Instead,
rationalize the numerator:
lim —
X-*\J<Xr~^47(V^ + 2)
This method will work only 1
if the resulting numerator allows the lim ——-
y/x + 2
fraction to be simplified. Pay close
1 1
attention to this example because it
illustrates a procedure we will use over \/4 + 2 4
and over.
x +5 if x > 0
Find lim /(x), where /(x) =
x-*-0 x if x < 0
Solution
Notice that /(0) is not defined, and that it is necessary to consider left- and right-hand
limits because /(x) is defined differently on each side of x = 0. We have
= 0
= 5
Because the left- and right-hand limits are not the same, we conclude that lim /(x)
x->0
does not exist. If we look at the graph, as shown in Figure 2.14, it is easy to see that
Figure 2.14 Graph of / the left- and right-hand limits are not the same. ■
(x + 1 if x > 0
Find lim g(x), where g(x) = \ 2 „•
x—>0 Xz + 1 if X < 0
/
66 Chapter 2 Limits and Continuity
Because the left- and right-hand limits are equal, we conclude that lim g(x) = 1. This
x—>0
time, by looking at the graph in Figure 2.15, we see that the left- and right-hand limits
are the same (even though the function is not defined at x = 0). Compare this graph
with the graph in Example 8. ■
The following theorem contains two additional limits that play an important role in
calculus. We evaluated the first of these limits by table and graphically in Example 6
Figure 2.15 Graph of g of Section 2.1.
Proof A geometric argument justifying these rules is given at the end of this section.
□
For now, let us concentrate on illustrating how these rules can be used.
Solution
a. We prepare the limit for evaluation by Theorem 2.3 by writing
sin 1 x
lim lim = 1
x^-o x M->osm»
c. We begin by writing
Thus, we have
cos h — 1 — sirr h
lim---= lim —-
h-*o h2 h-*o h2(cosh + 1)
sin2 h 1
-i 2 r
sin h .. 1
lim lim-
/i-»0 /i—>o cos h + 1
= -a r 1 + 1
i
“ ~2
Squeeze Rule*
If #00 ^ f(x) 5: h(x) on an open interval containing c, and if
With this squeeze rule in our toolkit, we are now ready to justify Theorem 2.3.
sin/i
Proof Proof that lim- = 1.
/2-+0 h
To prove the sine limit theorem requires some principles that are not entirely ob¬
vious. However, we can demonstrate its plausibility by considering Figure 2.16, in
which AOC is a sector of a circle of radius 1 with angle h measured in radians. The
line segments AD and BC are drawn perpendicular to segment OC.
sin A
Figure 2.16 Trigonometric relationships in the proof that lim = 1
Now, compare the area of the sector AOC with those of AAOD and ABOC. In
particular, since the area of the circular sector of radius r and central angle 0 is \r20
SMH (see the Student Mathematics Handbook), sector AOC must have area
5d )2h = \h
l sin h
cos h sin h < \h <
2 cos h
h 1
cos h < < Divide by | sin h.
sin h cosh
1 sin h
> > cosh Take reciprocals
cos h h
(reverse order).
sin h 1
cos h < Interchange left
~~h~ cos h
and right sides
(reverse order
again).
This same inequality holds in the interval (—-, 0), which is Quadrant IV for the angle
h. This can be shown by using the trigonometric identities cos(—h) = cosh and
sin(—h) = — sin h. Finally, we take the limit of all parts as h —» 0 to find
sin/i 1
lim cos h < lim- < lim-
/)—>0 h-* 0 h h^-ocosh
sin h 1
1 < lim- < -
h->0 h 1
sin h
From the squeeze rule, we conclude that lim = 1.
h-+ 0 h
You are asked to prove the second part of this theorem in Problem 59. □
z2 + z — 3 c2 + 3x - 10 4 — M2 x2 -4x +4
5. lim 6. lim 11. lim 12. lim
z—i z+ 1 *-3 3a:2 + 5jc - 7 u—> -2 2 + u x-*2 x2 — x — 2
2.2 Algebraic Computation of Limits 69
-L - 1 (* + l)2 x2 + 4x + 3
13. lim -- 14. lim 46. lim 47. lim
x—*l X — 1 x->0 r->2 t2 - 4t + 4 *-+3
WHAT DOES THIS SAY? In Problems 44-51, explain why the 63. Show that lim cosx = cosxo- Hint: You will need to use the
X-*-XQ
given limit does not exist. trigonometric identity coslA + B) = cos A cos B — sin A sin B.
1 .. 1
44. lim- 45. lim —p-
x—> 1 X — x—>2+ \f x 64. Show that lim tanx = tanx0 whenever cosx0 ^ 0.
X—*XQ
70 Chopter 2 Limits and Continuity
2.3 Continuity
IN THIS SECTION intuitive notion of continuity, definition of continuity, continuity theorems, continuity on
an interval, the intermediate value theorem
lim/(x) exists and is equal to/(c). The idea of continuity may be thought of informally as the quality of having parts
X—>C
that are in immediate connection with one another, as shown in Figure 2.17. The idea
evolved from the vague or intuitive notion of a curve “without breaks or jumps” to a
rigorous definition first given toward the end of the nineteenth century (see Historical
Quest Problem 45).
We begin with a discussion of continuity at a point. It may seem strange to talk
about continuity at a point, but it should seem natural to talk about a curve being
“discontinuous at a point.” A few such discontinuities are illustrated by Figure 2.18.
a. HOLE: /(c) is not defined b. HOLE: /(c) is defined c. JUMP: lim /(c) is not
X—»C“
lim/(jt) exists lim f(x) exists and is not equal the same as lim/fx)
x->c+
to/(c)
DEFINITION OF CONTINUITY
Let us consider the conditions that must be satisfied for a function / to be continuous
at a point c. First, /(c) must be defined or we have a “hole” in the graph, as shown
in Figure 2.18a. (An open dot indicates an excluded point.) In Figure 2.18b, there is a
“hole” even though /(c) is defined. If lim f(x) has one value as x —> c~ and another
X—>C
as x —»■ c+, then lim f{x) does not exist and there will be a “jump” in the graph of /,
as shown in Figure 2.18c. Finally, if one or both of the one-sided limits at c are infinite
(+oo or — oo), there will be a “pole” at x = c, as shown in Figures 2.18d and ef.
2.3 Continuity 71
1. /(c) is defined;
2. lim f(x) exists;
A' —> C
^ What This Says The third condition lim f(x) = /(c) summarizes the
X—>C
idea behind continuity. It says that if x is close to c, then fix) must be close
to /(c).
-2 + 2x - 3
a. fix) -
Figure 2.19 The geometric 1
interpretation of continuity x2 + 2x - 3
b. g(x) = -it x 7^ 1 and g(x) = 6 if x = 1
x — 1
x2 + 2x - 3
WARNING _ . , . c. hix) = - it X 1 and hix) — 4 if x - 1
Do not rely on graphing
calculators to test for continuity. For x T3
A” 2 d. Fix) — if x 7^ 1 and Fix) = 4 if x = 1
example, the graph of y = --- looks x — 1
(x - 1) (x + 3)
= lim---
*-*i x — 1
= lim(x -|- 3) = 4
*-►1
CONTINUITY THEOREMS
Proof The proof of the continuity theorem is based on the limit properties stated in
the previous section. For instance, a polynomial is a function of the form
where a0, a\,..., an are constants. We know that lim do — do and that lim xm = cm
x^c x-*-c
for m = 1,2,,n. This is precisely the statement that the function g(x) = dxm
is continuous at any number x — c, or simply continuous. Because P is a sum of
functions of this form and a constant function, it follows from the limit properties that
P is continuous.
The proofs of the other parts follow similarly. □
The limit properties of the previous section can also be used to prove a second
continuity theorem. This theorem tells us that continuous functions may be combined
in various ways without credting d discontinuity.
2.3 Continuity 73
Product fg
f
Quotient — provided g(c) ^ 0
8
Composition f °8 provided g is continuous at c
and / is continuous at g(c)
Proof The first four properties in this theorem follow directly from the basic limit
rules given in Section 2.2. For instance, to prove the product property, note that since
/ and g are given to be continuous at x = c, we have
= f(c)g(c) = Pic)
That is,
lim f[g(x)] = f{L) = / (lim g(x))
x^>c \x -*c /
This property applies in the same way to other kinds of limits, in particular to one-sided
limits.
Now we can prove the continuous composition property of Theorem 2.5. Let
h(x) — if o g)ix). Then we have
= f[gic) ] g is continuous at x = c.
^ What This Says The limit of a continuous function is the function of the
limiting value. The continuous composition property says that a continuous func¬
tion of a continuous function is continuous.
74 Chapter 2 Limits and Continuity
One-Sided Continuity The function / is continuous from the right at a if and only if
CONTINUITY ON AN INTERVAL
d. faix) - sin —
jc
e. /5CO =
(x sin 4
0 '
if x 7^ 0
if jc = 0
Solution
a. Function f\ is not defined when x2 - 4 = 0; that is, when x = 2 or x = -2. The
function is continuous on (—00, —2) U (—2, 2) U (2, 00).
b. Function fa is continuous on (—00, 00).
c. The cosecant function is not defined at x = tin, n an integer. At all other points it
is continuous.
2.3 Continuity 75
Usually there are only a few points in the domain of a given function / where a
discontinuity can occur. We use the term suspicious point for a number c where either
x2 - 1
a. —-has suspicious points for division by zero when x = 2 and x = —2.
x2 — 4
b. |x2 — 4| = x2 — 4 when x2 — 4 > 0 and |x2 — 4| = 4 — x2 when x2 — 4 < 0. This
means the definition of the function changes when x2 — 4 = 0, namely, when x = 2
and x = —2.
c. There are no suspicious points; we know the function cannot be continuous at places
where the function is not defined.
d. sin | has a suspicious point when x = 0 (division by 0).
e. x sin i has a suspicious point when x = 0 (division by 0).
Solution
We find that the domain for both functions is [—5, 5); the continuity theorem tells
us both functions are continuous everywhere on that interval except possibly at the
suspicious points. Consider the graphs of / and g as shown in Figure 2.20. Examining
/, we see
lim fix) = lim (3 — x) = 1 and lim fix) - lim (x - 2) = 0
x-*2~ x->2~ x-»2+ x-+2+
so lim fix) does not exist and / is discontinuous at x = 2. Thus, / is continuous for
x—>2
—5 < x < 2 and for 2 < x <5.
For g, the domain is also —5 < x < 5, and again, the only suspicious point is
x = 2. We have g(2) =0 and
This is the graph of g. Does the graph lim gix) — lim (2 - x) = 0 and lim g(x) = lim (x - 2) = 0
x->-2~ x-+2~ x—>-2+ >2+
reinforce our conclusions?
Therefore, lim gix) = 0 = gi2), and g is continuous at x = 2. Hence, g is continuous
x—*2
Proof This theorem is intuitively obvious, but it is not at all easy to prove. A proof
may be found in most advanced calculus textbooks. □
The intermediate value theorem can be used to estimate roots of the equation
fix) — 0. Suppose fia) > 0 and fib) < 0, so the graph of / is above the
x-axis at x = a and below for x — b. Then, if / is continuous on the closed in¬
terval [a, b], there must be a point x = c between a and b where the graph crosses the
Figure 2.22 Because f(a) > 0 and x-axis, that is, where / (c) = 0. The same conclusion would be drawn if fia) < 0 and
/ (b) < 0, then /(c) = 0 for some / ib) > 0. The key is that / (x) changes sign between x = a and x = b. This is shown
c between a and b. in Figure 2.22 and is summarized in the following theorem.
Proof This follows directly from the intermediate value theorem (see Figure 2.22).
The details of this proof are left as a problem.
Solution
We can use a graphing utility to estimate the point of intersection, as shown in Fig¬
ure 2.23.
We will now use the root location theorem to confirm the intersection point shown
in Figure 2.23. By calculator, we estimate c « 1.16.
Notice that the function fix) — cosjc — x3 + x is continuous on [j, y]. We find
that
Therefore, by the root location theorem there is at least one number c on (|, for
which /(c) = 0, and it follows that cos c = c3 — c. ■
Since the definition of continuity involves the limit, formal proofs of continuity
theorems can be given in terms of the €-8 definition of limit. Here is an example.
Solution
The continuity of / and g at x = xo says that
This means that if e > 0 is given, there exist numbers <5i and <52 such that
and
|g(x) - g(*o)l < - whenever 0 < |x - x0| < <$2
Let 8 be the smaller of <5i and <S2; that is, 8 — min(<5i, <52). Then
must both be true whenever 0 < |x - x0| < <5 so that (by the triangle inequality)
Thus, |(/ + g)(x) - (/ + g)(x0)| < e whenever 0 < |x - x0| < <5, and it follows
from the definition of limit that
3. the charges for a taxi ride across town considered as a function Show that / + g is continuous at x = 0 even though / and g
of mileage are both discontinuous there.
4. the charges to mail a package as a function of its weight In Problems 27-32, show that the given equation has at least one
Identify all suspicious points and determine all points of disconti¬ solution on the indicated interval.
nuity' in Problems 5-14. 27. f/x — x2 + 2x — 1 on (0, 1)
_ „ 3x + 5 1
5. fix) = x3 — lx + 3 6- fix) = --- 28. =x —x
2x — 1 1 on (1,2)
x+ 1
3x
7. fix) = 8. fit) = 3 - (5 + 2t)3 29. v'x — 8 + 9x2/3 = 29 on (0, 8)
\, i LV I
fix) = 3 if x = 1
-2 rc >S\ 7C / 2.JC
V l x2 — 4x + b + 3 if x < 1
i i
i i
i
ft i
O i
l > i 35. Interpretation Problem Use the intermediate value theorem to
In Problems 15-20, the given function is defined for all x > 0, ex¬ explain why the hands of a clock coincide at least once every
cept at x = 2. In each case, find the value that should be assigned hour.
to f (2), if any, to guarantee that f will be continuous at 2.
36. Interpretation Problem The graph shown in Figure 2.24 models
x2 — x — 2 how the growth rate of a bacterial colony changes with tem¬
15. fix) = 16. fix) = perature.*
x-2
sininx) cos -
17. fix) = 18. fix) = _X_
yi
x — 2
C 1, J
if x > 2 O
. a... y-- R :0/
19. fix) = \ 2x + 5
nera
x —2 —
What happens when the temperature reaches 45°C? Does it 45. [ RJCAL ( The first
make sense to compute lim R)t)? Write a paragraph describ- modern formulation of the notion of
r^50
continuity appeared in a pamphlet
ing how temperature affects the growth rate of a species.
published by Bernard Bolzano, a
Czechoslovakian priest whose mathe¬
37. A fish swims upstream at a constant speed v relative to the matical work was, for the most part,
water, which in turn flows at a constant speed vw (vw < v) overlooked by his contemporaries. In
relative to the ground. The energy expended by the fish in explaining the concept of continuity, BERNARD BOLZANO
traveling to a point upstream is given by Bolzano said that one must understand
1781-1848
b. What happens to E(v) as u +oo? Interpret your result as small as one wishes."
in words. In his book, Cours d1 Analyse, Cauchy (see Historical Quest
Problem 49, Section 2.1) introduces the concept of continuity
38. The population (in thousands) of a colony of bacteria t min¬
for a function defined on an interval in essentially the same
utes after the introduction of a toxin is given by the function
way as Bolzano. In this book Cauchy points out that the con¬
tinuity of many functions is easily verified. As an example,
t2 + 1 if 0 < f < 5 he argues that sinx is continuous on every interval because
Pit) = "the numerical value of sin \a, and consequently that of the
—St + 66 if t > 5
difference
8
H
39. f(x) = ondary school teacher who did his research at night. The histo¬
II
5
H
II
*E. Batschelet, Introduction to Mathematics for Life Scientists, 2nd ed.. + A. L. Cauchy, Cours d'Analyse de I'Ecole Royale Polytechnique, Oeu¬
vres, Ser. 2, Vol. 3. Paris: Gauthier-Villars, 1897, p. 44.
New York: Springer-Verlag, 1976, p. 280.
80 Chapter 2 Limits and Continuity
48. Counterexample Problem Find functions / and g such that / is uous from the right and continuous from the left at c.
discontinuous at x — 1 but fg is continuous there. 52. Show that if / and g are both continuous at x = c and
49. Counterexample Problem Give an example of a function defined g(c) ^ 0, then f/g must also be continuous there.
for all real numbers that is continuous at only one point. 0 if x 7^ 0
53. Let u(x) = x and f(x)
50. Prove the root location theorem, assuming the intermediate 1 if jc = 0
value theorem.
51. Show that / is continuous at c if and only if it is both contin¬ Show that lim f[u(x)\ fz / lim u(x)
*—>•0 x^-0
EXPONENTIAL FUNCTIONS
bn = b ■ b ■ b.b
• n factors
••
_ I
« If b ^ 0, then b° — 1, b n = —, and furthermore, if b > 0, then bl^n = \fb. Also, if
••• bn
• m and n are any integers, and m /n is a reduced fraction, then
••f
• bm!n = (bl!n)m
•
• This definition tells us what bx means for rational values of x. However, we now wish
• to enlarge the domain of x to include all real numbers. To get a feeling for what is
involved in this problem, let us examine the special case where b — 2. In Figure 2.25,
_ we have plotted several points with coordinates (r, 2r), where r is a rational number.
We must now attach meaning to bx if x is not a rational number by filling those holes
Figure 2.25 Graph of y = 2V for or spaces to make the graph continuous for all x. We use the following fundamental
rational exponents fact about the real number system, called the completeness property.
For any real number x, there exist rational numbers r„ such that
x = lim rn
/?-> OO
which means that for any number e > 0, there exists a number N such that
|x — r„| < e whenever n > N.
Another way of expressing the completeness property is to say that each real num¬
ber x can be approximated to any degree of accuracy by a rational number. Using the
completeness property, we can now define an exponential function.
2.4 Exponential and Logarithmic Functions 81
Exponential Function Let x be a real number, and let rn be a sequence of rational numbers such that
x = lim r„. Then the exponential function with base b > 0 (b y 1) is given
n-+oo
by
bx = lim br"
n—xx)
V3 ^ 1.732050808-
which means that V?> can be approximated to any desired degree of accuracy by mem¬
bers of the sequence of rational numbers
n = 1, r2 = 1.7, r3 = 1.73, r4 - 1.732, r5 = 1.7320, r6 = 1.73205, ...
Therefore, the exponential number 2^ is given by the limit of the sequence of numbers
2^ 21-7 21'73 21'732 21'7320 21'73205
In this fashion, we can fill the holes and gaps shown in Figure 2.25 to obtain the con¬
tinuous graph of y — 2X, as shown in Figure 2.26.
The shape of the graph of y — bx for any b > 1 is essentially the same as that of
y = 2X. The graph of y = bx for a typical base b > 1 is shown in Figure 2.27a and
is rising (when viewed from left to right). For a base b where 0 < b < 1 the graph is
Figure 2.26 Graph of y = 2 _ ... , . ^ ~,
falling, as shown in Figure 2.27b.
Quotient rule
by
/a\x ax
Power rules (bxy =bxy-Aab)x =ax= —
82 Chapter 2 Limits and Continuity
Graphical Properties
The function y — bx is continuous for all x. Furthermore, its graph is always
above the x-axis (bx > 0), and is always rising for b > 1 and always falling for
0 < b < 1.
, — = 3x — 2
2
c2 - 6x + 4 = 0
Equality rule
6±V36-4(1)(4)
x = = 3± V5
LOGARITHMIC FUNCTIONS
Logarithmic Function If b > 0 and b ^ 1, the logarithm of x to the base b is the function y — logfc x
that satisfies by = x; that is,
y - log6 x means by = x
y - logb x
y is the logarithm to the base b of x.
y is the exponent to the base b that gives x.
by — x
MMuxumiMmMWMmmmmmmmmmiBBmmmmmHBBmKm'
2.4 Exponential and Logarithmic Functions 83
Notice that y — log*x is defined only for x >0 because by > 0 for all y. We have
sketched the graph of y — log* x for b > 1 in Figure 2.28 by reflecting the graph of
y = bx in the line y = x.
Proof Each part of this theorem can be derived by using the definition of the log¬
arithm in conjunction with a suitable property of exponentials. (See Problems 73-
75.) a
Solution
Using Theorem 2.9, we find that
log2 (|) + log2 128 = log2(2)"3 + log2 27
= log2[2_3(27)] = log2 24 = 41og2 2 = 4(1) =4
84 Chapter 2 Limits and Continuity
What we have shown here is longer than what you would do in your work, which would
probably be shortened as follows:
Solution
Use Theorem 2.9, and remember that both 2x + 1 > 0 and x — 3 > 0.
EXP NOTATION
COMMON LOGARITHM
NATURAL LOGARITHM
The “exp” notation for the natural exponential function is especially useful for
representing composite functions. For instance, the awkward expression
exp(3*2 — 2 sin* + 8)
The following theorem summarizes the key properties of natural exponential and log¬
arithmic functions.
a. In 1 = 0 b. In e = 1
c. elnx — x for all * > 0 d. In ey = y for all y
e. bx = exXnb for any b > 0 {b 1)
Proof Parts a and b follow immediately from the definitions of In* and ex. Parts c
and d are just the inversion rules for base e.
We show the proof of part e.
WARNING ., r. On calculators, there are usually two logarithm keys, [LOG]and Qj\[|. For other log¬
It you are using a software
arithms, you will need the useful conversion formula contained in the following theo¬
package such as Mathematica, Derive,
or Maple, sometimes referred to as rem.
CAS (Computer Algebra System), be
careful about the notation. Some Change-of-base theorem
THEOREM 2.11
versions do not distinguish between
In*
log* (common logarithm) and In* for any b > 0 (b 1)
log/, x =
(natural logarithm). All logarithms on h^
these versions are assumed to be
natural logarithms, so to evaluate a
Proof Let y = logfe *. Then
common logarithm requires using the
change-of-base theorem. by = * Definition of logarithm
In*
Divide both sides by In b. □
'V = h7b
EXAMPLE 4 Solving an exponential equation using the change-of-base
theorem
Solve 6X = 200.
Solution 6X — 200
x — log6 200 Definition of exponent
In 200
Change-of-base theorem
“ In 6
« 2.957047225 Approximate with a calculator. ■
Solution
We want to solve the equation
10 log51 + 3 log2 t — 20
In t In t
10-—- + 3-—— = 20 Change-of-base formula
In 5 In 2
/ 10 3 \
I -—- -)-) In t = 20 Factor.
\ In 5 In 2 /
/ 10 3 V1
In r = 20 ( -— +- « 1.89727
\ln5 ln2/
t ^ el-89727 ^ 6 6677
Thus, it takes about 6.67 seconds for the velocity to reach 20 ft/s. ■
Solution
We wish to find N for
\02x = eN
In 102x = \neN Log of both sides
2x In 10 — Nine Power rule of logarithms
2x In 10 = N lne = 1
Thus, I02x = e2xlDl°.
Solution
We have
1 _ 0.0004251
2
Remember that —0.000425t is the
—0.000425f = In 0.5
exponent on base e that equals \ = 0.5
In 0.5
1630.934542 , ■
1 ~ -0.000425
2.4 Exponential and Logarithmic Functions 87
P(t) = P0ekt
where P0 is the initial population and k is a positive constant. Suppose the colony
begins with 5,000 individuals and contains a population of 7,000 after 20 min. Find
k and determine the population (rounded to the nearest hundred individuals) after 30
min.
Solution
Because Pq = 5,000, the population after t minutes will be
P(t) = 5,000/'
P(20) = 5,000/(20)
7,000 - 5,000c20*
7 _ „20k
20 k = ln(|)
k ~ 20 )
% 0.0168236
Finally, to determine the population after 30 min, substitute this value for k to find
P(30) = 5,000<?30*
% 8,282.5117
One reason e is called the “natural” exponential base is that many natural growth phe¬
nomena can be described in terms of ex. As an illustration of this fact, we close this
section by showing how ex can be used to describe the accounting procedure called
continuous compounding of interest.
Suppose a sum of money is invested and the interest is compounded once during
a particular period. If P is the initial investment (the present value or principal) and
i is the interest rate during the period, then the future value (in dollars). A, after the
interest is added will be
A= P + Pi = P(l+i)
t t
principal interest
Thus, to compute the balance at the end of an interest period, we multiply the balance
at the beginning of the period by the expression 1 + /.
Interest is usually compounded more than once a year. The interest added to the
account during one period will itself earn interest during subsequent periods. If the
annual interest rate is r and the interest is compounded n times per year, then the year
is divided into n equal interest periods and the interest rate during each such period is
j = -. To compute the balance at the end of any period, you multiply the balance at
the beginning of that period by the expression 1 + so the balance at the end of the
first period is
and so on. At the end of t years, the interest has been compounded nt times, and the
future value is
/ f \nt
Am = P( l + n)
The first two graphs in Figure 2.29 show how an amount of money in an account
over a one-year period of time grows, first with quarterly compounding and then with
monthly compounding. Notice that these are “step” graphs, with jumps occurring at
the end of each compounding period.
Figure 2.29 The growth of an account over a one-year period with different compounding
frequencies
A(t) = lim P (\ +
f \ nt
7,
Let k — r so that krt — nt and -nk
= f;
«-»• oo \ n/ also k —» oo as n —> oo.
krt
r
= lim P 1 +
A—>oo
-I rt
V
lim P
k-* oo l + k.
n rt
1
= P lim 1 + - Scalar rule for limits
A:—► oo
= Pert Definition of e
Future Value
If P dollars are compounded n times per year at an annual rate r, then the future
value after t years is given by
/ r\nt
A(,) = />(l + -)
and if the compounding is continuous, the future value is
A(t) = Pert
a. monthly. b. continuously.
c. If the interest is compounded continuously, how long will it take for the money to
double?
Solution
P = $12,000; t — 5; and r = 18% = 0.18 are given.
12,000<?°18' = 24,000
e0-18' = 2
0.18r = In 2 Definition of logarithm
In 2
3.851
15. 5 log3 9 — 2 log2 16 16. (log21) (log3 27) 39. log3 x + log3 (2x + 1) =
17. (3,og71 )(log5 0.04) 18. e5ln2 2x
40. In ^ p^ = lnx + In
19. log3 34 — In e0 5 20. ln(log 10e) 1 +x
90 Chapter 2 Limits and Continuity
In Problems 41-46, evaluate the given limits. 55. According to the Bouguer-Lambert law, a beam of light that
strikes the surface of a body of water with intensity 70 will
41. a. lim x2e x b. limx2e x
x-*0+ JC—»■ 1 have intensity / at a depth of x meters, where
42. a. lim e~xl b. lim e~x* / = I0ekx, k> 0
.v—>0 X—► 1
The constant k, called the absorption coefficient, depends on
43. a. lim (1 + x)'/x b. lim(I +x)
.t—>0+ x->\ such things as the wavelength of the beam of light and the pu¬
44. a. lim exjx b. lim lnx rity of the water. Suppose a given beam of light is only 5% as
*-►0- *-►0+ intense at a depth of 2 meters as it is at the surface. Find k and
In sfx determine at what depth (to the nearest meter) the intensity is
45. lim x In x2 46. lim
x—>e x 1% of the intensity at the surface. (This explains why plant
life exists only in the top 10 m of a lake or sea.)
047. If log* 1,296 = 4, what is (§Z>)3/2?
56. A certain bank pays 6% interest compounded continuously.
48. If log^ 106 = 2, what is -Jb — 25? How long will it take for $835 to double?
49: Solve log, x + log5(2x + 1) = lnx correct to the nearest tenth. 57. If an amount of money is compounded continuously at an an¬
50. Solve logt 2 = log3 x correct to the nearest tenth. nual rate r, how long will it take for that amount of money to
Data points with a curve fit to those points are shown in Prob¬ double?
lems 51-54. Decide whether the data are better modeled by an 58. First National Bank pays 7% interest compounded monthly,
exponential or a logarithmic function. and World Savings pays 6.95% interest compounded continu¬
ously. Which bank offers a better deal?
y
59. A person invests $8,500 in a bank that compounds interest
• % •• continuously. If the investment doubles in 10 years, what is
• the (annual) rate of interest (correct to the nearest hundredth
••
percent)?
3
• •
• :*
•• 60. In 1626, Peter Minuit traded trinkets worth $24 for land on
• Manhattan Island. Assume that in 2000 the same land was
• worth $43.9 billion. Find the annual rate of interest com¬
pounded continuously at which the $24 would have had to be
1 invested during this time to yield the same amount.
1
61. Biologists estimate that the population of a bacterial colony is
1 X
Pit) = P0 2kl
•
• « •
• •• at time t (in minutes). Suppose the population is found to be
• :* 1,000 after 20 minutes and that it doubles every hour.
•• » • i
a. Find P0 and k.
•y
• b. When (to the nearest minute) will the population be 5,000?
/ 62. Spy Problem Having escaped from the smugglers (Problem
/ _ 49, Section 1.2), the Spy receives an e-mail message that his
2 * best friend, Sigmund (“Siggy”) Leiter, has been found mur¬
dered, the body stuffed unceremoniously in a freezer. Fight¬
ing back tears of grief, he remembers that t hours after death,
a body has temperature
T = A + (B — A)e~003'
„ log £ — 11.4
M —
1.5
2.4 Exponential and Logarithmic Functions 91
c. What percent can be expected to fail between the 40th and a. Show that if Lj =
nog Aj and L2 — nog N2, then
/A,A2\
50th months? nog A, + nog N2 = n°g I ~)'
67. If $3,600 is invested at 15% compounded daily, how much
money will there be in 7 years? b. State and prove similar results for a quotient rule and a
power rule for Napier logarithms.
a. Use a 365-day year; this is known as exact interest.
c. What is nog 107?
b. Use a 360-day year; this is known as ordinary interest.
d. If nog A = 107, then A er for some rational number r.
68. If P dollars are borrowed for a total of A months compounded
What is r?
monthly at an annual interest rate of r, then the monthly pay¬
e. Fortunately, Napier’s 1614 paper on logarithms was read
ment is found by the formula
by a true mathematician, Henry Briggs (1561-1630), and
n fj) together they decided that base 10 made a lot more sense.
m l-a + fj)-* In the year Napier died, Briggs published a table of com¬
mon logarithms (base 10) that at the time was a major ac¬
a. Use this formula to determine the monthly car payment for complishment. In this paper he used the words “mantissa”
a new car costing $ 17,487 with a down payment of $7,487. and “characteristic.” What is the meaning that Briggs gave
The car is financed for 4 years at 12%. to these words?
b. A home loan is made for $210,000 at 8% interest for 30 f. Who was the first person to publish a table of natural loga¬
years. What is the monthly payment with a 20% down rithms (base e)l
payment? Q 71. For b > 0 and all positive integers m and n, show that
69. A cool drink is removed from a refrigerator and is placed in a bm
a. bmbn = bm+n b. — = bm~n
room where the temperature is 70°F. According to a result in bn
physics known as Newton’s law of cooling, the temperature of 72. For b > 0 and all positive integers m and n, show that
the drink in t minutes will be a. (A")" = bmn b. (Vb)m = s/b™
F(t) = 70 - Ae~k‘ 73. Prove:
a. logfe x + log6 y = logfcUy)
where A and k are positive constants. Suppose the temperature x
of the drink was 35° F when it left the refrigerator, and 30 min¬
b. \ogb x - log* y = logfc -
utes later, it was 50°F [that is, E(0) = 35 and F(30) = 50], 74. Let b be any positive number other than 1. Show that
a. Find A and e~30k.
Xx = bx log''x
b. What will the temperature of the drink be after one hour
(to the nearest degree)? 75. Let a and b be any positive numbers other than 1. Show that
c. What would you expect to happen to the temperature as
a. log x = b. (loga b)(\og,,a) = 1
t —► + oo? log ha
92 Chapter 2 Review
sin9x |2x
Proficiency Examination 23. lim 24. lim
sin 5x i - 2x - 1
CONCEPT PROBLEMS
25. Decide whether each of the curves is exponential or logarith¬
1. State the informal definition of a limit of a function. Discuss mic.
this informal definition. -
a. i ] n
y. "a._. b. y r-
2. State the formal definition of a limit.
3. State the following basic rules for limits:
a. limit of a constant b. multiple rule
1
1
c. sum rule d. difference rule -t -2
X
-t -A -z \ jd
0 si?
e. product rule f. quotient rule
g. power rule
h. limit of a polynomial function
i. limit of a rational function c. > y> - i
Evaluate the limits in Problems 19-24. 6. log2x + log2(x — 15) = 4 7. 32*-1 = 6X3X~X
4x yfx — 2 8. ln(x - 1) + ln(x + 1) = 2 In 712
19. lim 20. lim
3 X2 +X - Jt-*4 x — 4
2 -5x *The supplementary problems are presented in a somewhat random order,
21. lim 22. lim .Lt.cos*
x-+2 x2 -4 *-►0 2 tan x not necessarily in order of difficulty.
Chapter 2 Review 93
1 if x is rational 52. How much should you invest now at an annual interest rate of
41. rix) = 6.25% so that your balance 10 years from now will be $2,000
—1 if x is irrational
if interest is compounded
42. /(x) = |x — 2| on [—5, 5] a. quarterly? b. continuously?
53. A decibel (named for Alexander Graham Bell) is the smallest
43. a. fix) = --on t0- 5]
x +2 increase of the loudness of a sound that is detectable by the
x2 — x — 6 human ear. In physics, it is shown that when two sounds of in¬
b. fix) = on [—5, 5]
x+2 tensity I\ and f (watts/cm3) occur, the difference in loudness
is D decibels, where
x2 — x — 6
on [—5, 5], x 7^-2
c. fix) - x +2
-4 for x — —2 °= 10log(|)
x2 — x — 6 When sound is rated in relation to the threshold of human
on [—5, 5], x #-2
d. fix) = x+2 hearing (/o = 10-16), the level of normal conversation is 50
-5 for x = —2 decibels, whereas that of a rock concert is 110 decibels. Show
:2 — 3x — 10 that a rock concert is 60 times as loud as normal conversation
44. a. g(x) = on [0, 5] but a million times as intense.
x +2
94 Chapter 2 Review
54. If a function / is not continuous at x = c, but can be made W. B. Gearhart and H. S. Schultz, “The Function sinx/x,”
continuous at c by being given a new value at the point, it is College Mathematics Journal (1990): 90-99.
said to have a removable discontinuity at x = c. Which of the L. Gillman, “n and the Limit of (sina)/a,” American Mathe¬
following functions has a removable discontinuity at x = cl matical Monthly (1991): 345-348.
2x2 4- x — 15 F. Richman, “A Circular Argument,” College Mathematics
a. f{x) =-r-at c
x+3 Journal(1993): 160-162.
x -2 D. A. Rose, “The Differentiability of sinx,” College Mathe¬
b. fix) - at c = 2
x 2| matics Journal (1991): 139-142.
2 -
P. Ungar, “Reviews,” American Mathematical Monthly
c. fix) at c (1986): 221-230.
4—x
Some of these articles argue that the demonstration shown in
55. If fix) = x3 — x2 + x + 1, show that there is a number c such
the text is circular, since we use the fact that the area of a circle
that / (c) = 0.
is 7rr2. How do we know the area of a circle? The answer, of
56. Prove that y/x + 3 = ex has at least one real root, and then use
course, is that we learned it in elementary school, but that does
a graphing utility to find the root correct to the nearest tenth.
not constitute a proof. On the other hand, some of these arti¬
57. Show that the tangent line to the circle x2 + y2 = r2 at the cles argue that the reasoning is not necessarily circular. Read
point (x0, yo) has the equation y0y + x0x = r2. Hint: Recall one or more of these journal articles and write a report.
that the tangent line at a point P on a circle with center C is
62. a. Show that sinx < x if 0 < x < |. Refer to Figure 2.30.
the line that passes through P and is perpendicular to the line
Hint: Compare the area of an appropriate triangle and sec¬
segment C P.
tor.
58. Counterexample Problem It is not necessarily true that lim fix)
D
X—>C
fix)
and lim gix) exist whenever lim exists. Find functions
gix)
fix)
/ and g such that lim exists, but neither lim fix) nor
0 gix)
lim gix) exists.
x—*0
AT i°C) 12.5 7.5 5.8 5.0 4.5 3.5 2.9 2.8 sin(a + f3) — sin a cos f + cos a sin fi
P(t) = PQen
1.67-55 if u > 45
97
98 Chapter 3 Differentiation
TANGENT LINES
Since classical times, mathematicians have known that the tangent line at a given point
on a circle has the property that it is the only line that intersects the circle exactly
once, but the same principle cannot be applied to finding tangent lines to more general
curves. Indeed, as shown in Figure 3.1, the line we may intuitively think of as being
tangent to the graph of y = fix) at point P can intersect the curve at other points,
and for that matter, there may be many lines that intersect the curve only at P, but are
clearly not tangent lines.
y = f(x) only at P
We begin this section by computing the slope of a tangent line as a limit, using a
procedure originally developed by the French mathematician Pierre de Fermat (1601-
1665). Fermat’s use of the “dynamic” limit process rather than the “static” procedures
of classical geometry and algebra was a breakthrough in thinking that eventually led to
the development of differential calculus by Isaac Newton (1642-1727) and Gottfried
Leibniz (1646-1716) in the latter half of the seventeenth century. Following in the
footsteps of these giants, we, too, will use what we discover about tangent lines as a
springboard for introducing the derivative and studying its basic properties.
Suppose we wish to find the slope of the tangent line to y = fix) at the point
P(xo, f{xo)). The strategy is to approximate the tangent line by other lines whose
slopes can be computed directly. In particular, consider the line joining the given point
WARNING ^ , P to the neighboring point Q on the graph of /, as shown in Figure 3.2. This line is
Do not contuse secant line
(a line that intersects a curve in two or
called a secant line. Compare the secant lines shown in Figure 3.2.
more points) with the secant function of
trigonometry.
Notice that a secant line is a good approximation to the tangent line at point,/5 as long
as Q is close to P.
3.1 An Introduction to the Derivative: Tangents 99
WARNING A . . , . . . To compute the slope of a secant line, first label the coordinates of the neigh¬
Ax is a single symbol and
does not mean delta times x. Do not boring point Q, as indicated in Figure 3.2. In particular, let Ax denote the change
forget that as Ax —> 0, Ax is getting in the x-coordinate between the given point P(x0, /(x0)) and the neighboring point
close to 0, but is not equal to 0. Q(x0 + Ax, f(xo + Ax)).
The slope of this secant line, msec, is easy to calculate:
To bring the secant line closer to the tangent line, let Q approach P on the graph of f
by letting Ax approach 0. As this happens, the slope of the secant line should approach
the slope of the tangent line at P. We denote the slope of the tangent line by mtan to
distinguish it from the slope of a secant line. These observations suggest the following
definition.
Slope of a Line Tangent to a At the point P(x o, fix o)), the tangent line to the graph of / has slope given by
Graph at a Point the formula
fix0 + Ax) - /(x0)
mtan = lint ---
>0
A.v— Ax
provided this limit exists.
Solution
Figure 3.3 shows the tangent line to / at x = — 1. The slope of the tangent line is given
by
Solution
Figure 3.4 shows a tangent line at an arbitrary point P(x, x2) on the curve. From the
definition of slope of the tangent line,
The result of Example 2 gives a general formula for the slope of a line tangent to
the graph of /(x) = x2, namely, mtan = 2x. The answer from Example 1 can now be
verified using this formula; if x = —1, then mtan = 2(— 1) = —2.
THE DERIVATIVE
The expression
f(x + Ax) - fjx)
Ax
which gives a formula for the slope of a secant line to the graph of a function /, is
called the difference quotient of /. The limit of the difference quotient
fix + Ax) - fix)
lim
Ax—>0 Ax
which gives a formula for the slope of the tangent line to the graph of / at the point
(x, fix)), is called the derivative of / and is frequently denoted by the symbol fix)
(read “eff prime of x”). To differentiate a function f-at x means to find its derivative
at the point (x, fix)).
1. If the limit for the difference quotient exists, then we say that the function / is
differentiable at x.
3.1 An Introduction to the Derivative: Tangents 101
2. The value of a derivative depends only on the limit process and not on the symbols
used in that process. In Example 2, we found that if /(x) — x2, then fix) — 2x.
This means that we also know
Finding the slope of a tangent line is just one of several applications of the deriva¬
tive that we will discuss in this chapter. In Section 3.4, we will examine rectilinear mo¬
tion and other rates of change, and in Section 3.8, marginal analysis from economics.
In Chapter 4 we will examine more complex applications such as curve sketching and
optimization.
f t -\- At — ft ( f1 4- At T ft \
= lim - I . :-— 1 Rationalize numerator.
Ar->0 At \ft + At + ft)
jt + At)-t
- lim
A/->0 At ift T" At -)- sft)
At
= lim
A r->0 Ati\/t -(- At T- \/t)
1
= lim ■-— Reduce fraction.
Ar->0 v t T At T t
1
For t > 0 ■
2 ft
Difference
quotient
n Av EXAMPLE 4 Estimating a derivative using a table
Ax — + Ax
6 Ax Estimate the derivative of fix) = cos* atx = | by evaluating the difference quotient
1 1.523598776 -0.81885
Ay _ fix + Ax) - fjx)
0.5 1.023598776 -0.69146
0.648598776 -0.55276 Ax Ax
0.125
PREVIEW
An encouraging word is necessary after reading this first example of finding a
derivative. By now you are aware of the fact that the derivative concept is one
of the main ideas of calculus, and the previous example indicates that finding a
derivative can be long and tedious. In the next section, we will begin to simplify
the process of finding derivatives, so that you can quickly and efficiently find the
derivatives of many functions (without using this definition directly). For now,
however, we focus on the derivative concept and definition.
Proof To find the equation of the tangent line to the curve y = fix) at the point
P(xo, yo), we use the fact that the slope of the tangent line is the derivative f'{xo) and
apply the point-slope formula for the equation of a line:
Solution
The graph of the function y = j, the point where x = 2, and the tangent line at the
point are shown in Figure 3.5. First, find fix):
Solution
In Example 5, we found that the slope of the tangent line is /'(2) = — | and that the
point of tangency is (2, j). In Section 1.2, we saw that two lines are perpendicular if
and only if their slopes are negative reciprocals of each other. Thus, the perpendicular
line we seek has slope 4 (the negative reciprocal of m = — |). The desired equation is
Normal Line to a Graph The normal line to the graph of / at the point P is the line that is perpendicular
to the tangent line to the graph at P.
Figure 3.6 shows the graph of the function from Example 6, along with the graph
y No rmal line
1 1
of its normal line.
y~X
/ y = 4 (x -21 + 2
RELATIONSHIP BETWEEN THE GRAPHS OF / AND /'
It is important to take some time to study the relationship between the graph of a
function and its derivative. Since slope is measured by /', it follows that at points
x where f'(x) > 0, the tangent line must be tilted upward, and the graph is rising.
Similarly, where /'(x) < 0, the tangent line is tilted downward, and the graph is
(z. ly falling. If f'(x) — 0, the tangent line is horizontal at x, so the graph “flattens.” These
■ £
■
3 i . i0 x observations are illustrated in Figure 3.7.
/'<
gr
y=f(x)
Solution
Notice that the graph of / is rising for x < — 1 and for x > 3; it is falling for — 1 <
x < 3; and it has horizontal tangent lines at x = — 1 and x = 3. Thus, the graph of f
Figure 3.9 A possible graph for is above the x-axis (f'(x) > 0) for x < — 1 and x > 3, below the axis for — 1 < x < 3
the derivative of the function (/'(x) < 0), and crosses the axis at x = —1 and at x = 3. One possible graph with
whose graph is given in Figure 3.8 these features is shown in Figure 3.9. ■
EXISTENCE OF DERIVATIVES
We observed that a function is differentiable only if the limit in the definition of deriva¬
tive exists. At points where a function / is not differentiable, we say that the deriva¬
tive of f does not exist. Three common ways for a derivative to fail to exist at a point
(c, /(c)) in the domain of / are shown in Figure 3.10.
1 -3
EXAMPLE 8 A function that does not have a derivative because of a corner
Show that the absolute value function /(x) = |x| is not differentiable at x = 0.
Solution
Figure 3.11 f(x) = |x| is not
differentiable at x = 0 because the The graph of /(x) = |x| is shown in Figure 3.11. Note that because the slope “from
slope from the left does not equal the left” at x = 0 is — 1 while the slope “from the right” is +1, the graph has a corner
the slope from the right at the origin, which prevents a unique tangent line from being drawn there.
3.1 An Introduction to the Derivative: Tangents 105
f (0 + Ax) — f (0)
/'(0)= lim
Ajc-+0 Ax
/(Ax) - /(0)
= lim
A^->0 Ax
I Ax I
= lim
Ajt—>0 Ax
| Ax | — Ax
lim - lim = -1 Derivative from the left
Ax^O- Ax Ajt—>0~ Ax
|Ax| Ax
lim - lim — = 1 Derivative from the right
AJC-+0+ Ax ajc-*o+ Ax
The left- and right-hand limits are not the same; therefore, the limit does not exist. This
means that the derivative does not exist at x = 0. Hi
If the graph of a function has a tangent line at a point, we would expect to be able to
draw the graph continuously (without the pencil leaving the paper). In other words, we
expect the following theorem to be true.
Proof Recall that for / to be continuous at x = c: (1) /(c) must be defined; (2)
lim /(x) exists; and (3) lim /(x) = /(c). Thus, continuity can be established by
X-+C X—>C
= f\c) -0 = 0
Thus, lim /(x) = /(c), and we see that the conditions for continuity are satisfied. □
106 Chapter 3 Differentiation
WARNING
Be sure you understand what we have just shown with Example 8 and
Theorem 3.2: If a function is differentiable at x — c, then it must be continuous at that point.
The converse is not true: If a function is continuous at x — c, then it may or may not be
differentiable at that point. Finally, if a function is discontinuous at x = c, then it cannot
possibly have a derivative at that point. (See Figure 3.13c.)
DERIVATIVE NOTATION
dy
dy 7
For instance, we would evaluate — = 4x at x = 3 by writing
dx
dy_
= 4x2 lx=3
= 4(3)2 = 36
dx x=3
d -i
— (x2) - 2x
dx
WARNING
Despite its appearance, ^4 is a single symbol and is not a fraction. In
Section 3.8, we introduce a concept called a differential that will provide independent meaning
to symbols like dy and dx, but for now, these symbols have meaning only in connection with
the Leibniz derivative symbol ^.
dy
At x = — 1
nDer i'jOr-37X? -1) dx
3.000001
If you are using technology to find derivatives, most formats require that you input
not only the function, but the variable in the expression and the value at which you
wish to find the derivative. The usual format is nDeri^{expression, variable, value).
The symbol “nDeriv” depends on the calculator or software, and sometimes is “nDer”
Figure 3.14 Typical calculator
(TI-85/86), “d” (TI-92), “diff” (Maple V), or “Dif” (Derive). This notation is used in
output for gL=_r Figure 3.14. Some software and calculators will find the exact value of a derivative,
where y — x3 whereas some will do a numerical evaluation (as shown in Figure 3.14).
In each of Problems 11-16, a function f is given along with a 44. Suppose fix) = x3.
number c in its domain. a. Compute the slope of the secant line joining the points on
a. Find the difference quotient of f. the graph of / whose x-coordinates are 1 and 1.1.
b. Find /'(c) by computing the limit of the difference quotient. b. Use calculus to compute the slope of the line that is tangent
11. f{x) = 3 at c = — 5 12. f{x) = x at c = 2 to the graph when x = 1 and compare this slope to your
answer from part a.
13. fix) = 2x at c — 1 14. fix) = 2x2 at c = 1
45. Sketch the graph of the function y = x2 — x. Determine the
15. fix) = 2 — x2 at c = 0 16. /(x) - — x2 at c = 2
value(s) of x for which the derivative is 0. What happens to
(/.sc f/zc definition to differentiate the functions given in Prob¬ the graph at the corresponding point(s)?
lems 17-28, and then describe the set of all numbers for which 46. a. Find the derivative of fix) = x2 — 3x.
the function is differentiable.
b. Show that the parabola whose equation is y = x2 — 3x has
17. fix) = 5 18. g(x) = 3x one horizontal tangent line. Find the equation of this line.
19. fix) = 3x — 1 20. gix) = 4 — 5x c. Find a point on the graph of / where the tangent line is
21. g(x) = 3x2 22. hix) = 2x2 + 3 parallel to the line 3x + y = 11.
23. fix) = x2 24. git) =4-t2 d. Sketch the graph of the parabola whose equation is
y — x2 — 3x. Display the horizontal tangent line and the
25. fis) = {s- l)2 26. fix) = -1-
2x tangent line found in part c.
27. fix) = V5x 28. fix) — ffx + 1 47. a. Find the derivative of fix) = 4 — 2x2.
Find an equation for the tangent line to the graph of the function b. The graph of / has one horizontal tangent line. What is its
at the specified point in Problems 29-34. equation?
c. At what point on the graph of / is the tangent line parallel
29. fix) = 3x - 7 at (3, 2) 30. g(x) = 3x2 at (-2, 12)
to the line 8x + 3y = 4?
31. fis) = r at i = 32. git) = 4 — f2 at t = 0
48. Show that the function fix) — |x — 2| is not differentiable at
1
33. /(x) = at x = 2 34. g(x) — ffx — 5 at x x = 2.
x +3
49. Is the function fix) = 2 |x + 11 differentiable at x = 1 ‘
Find an equation of the normal line to the graph of the function at
the specified point in Problems 35-38. | —x2 if x < 0
50. Let f(x)
if x > 0
35. fix) = 3x — 7 at (3, 2) 36. gix) = 4 — 5x at (0, 4)
Does /'(0) exist? Hint: Find the difference quotient and take
37. /(x) = —— at x = 3 38. fix) — y/5x at x = 5 the limit as Ax —»• 0 from the left and from the right.
x +3
dy if x < 1
Find for the functions and values of c given in Prob- 51. Let fix) —
dx X—C
if x > 1
lems 39—42. a. Sketch the graph of /.
53. fix) = (2x — 1 )2 for c — 1 54. f(x) —-for c — 2 66. HISTORICAL QUEST In the guest
x + 1 essay at the end of Chapter 1, it was
55. fix) = sin x for c — | 56. /(x) = cos x for c = ^ noted that Isaac Newton, who invented
calculus at the same time as Leibniz,
57. /(x) = s/x for c = 4 58. /(x) = ^/x for c = 8
considered Fermat as "one of the gi¬
59. Show that the tangent line to the parabola y = Ax2 (for
ants" on whose shoulders he stood. Fer¬
A ^ 0) at the point where x — c will intersect the x-axis mat was a lawyer by profession, but
at the point (|, 0). Where does it intersect the y-axis? he liked to do mathematics in his spare
60. Find the point(s) on the graph of /(x) = —x2 such that the time. He wrote well over 3,000 math¬ PIERRE DE FERMAT
ematical papers and notes. Fermat 1601-1665
tangent line at that point passes through the point (0, —9).
developed a general procedure for finding tangent lines that
61. Interpretation Problem
is a precursor to the methods of Newton and Leibniz. ■
a. Find the derivatives of the functions y = x2 and y = x2 — 3
We will explore this procedure by finding a tangent line to
and account geometrically for their similarity.
the curve
b. Without further computation, find the derivative of x3 + y3 - 2xy = 0
y = x2 + 5.
A graph is shown in Figure 3.15.
62. Interpretation Problem
a. Find the derivative of /(x) = x2 + 3x.
b. Find the derivatives of the functions g(x) = x2 and
h(x) = 3x separately. How are these derivatives related
to the derivative in part a?
c. In general, if/(x) = g(x) + h(x), what would you guess is
the relationship between the derivative of / and the deriva¬
tives of g and hi
63. Consider a graph of the function defined by /(x) = x2^3.
/(x) = (x - 2)2/3 + 2x3 c. With the answer from part b, you can follow the steps of
Fermat by dividing both sides of the equation by E. Fer¬
which gives trouble in seeking the tangent line at (2, /(2)). mat reasoned that since E was close to 0, this step should
Attempt to compute /'(2), either “by hand” or using a calcu¬ be permitted. Now, after doing this, Fermat further rea¬
lator. Describe what happens; in particular, do you see why soned that since E was close to 0, he could now set E = 0
the tangent line there is meaningless? and solve for A. Carry out these steps to write A in terms
of Xo and yo-
.
65 Compute the difference quotient for the function defined by
d. Fermat then constructed the tangent line at P by joining P
to S. Draw the tangent line for the given curve at the point
sinx
if x 7^0 (0.5, 0.93) by plotting the point corresponding to the cal¬
fix) = x
culated value of A. Find an equation for the tangent line to
1 if x = 0
the curve
x3 + y3 — 2xy = 0
Do you think fix) is differentiable at x = 0? If so, what is
the equation of the tangent line at x = 0? at the point P(0.5, 0.93).
110 Chapter 3 Differentiation
67. HISTORICAL QUEST The Descartes' next step was to use the equation of the given curve
groundwork for much of the mathe¬ y = f{x) (actually any equation involving two variables) to
matics we do today, and certainly a eliminate one of the variables (usually y) from the equation of
necessity for calculus, is the develop¬ the circle. Descartes reasoned that the circle will cut the given
ment of analytic geometry by Descartes curve in two places except when P Q is normal, in which case
and Pierre de Fermat (see Problem 66). the two intersection points will coincide and the circle will be
Descartes' ideas for analytic geome¬ tangent line to the curve at P. That is, Descartes imposed
try were published in 1637 as one of the condition that the resulting equation (after substitution) has
three appendices to his Discourse on the RENE DESCARTES only one root to solve for Xi. The point Q is thus found and the
Method (of Reasoning Well and Seek¬ 1596-1650 normal line was then known to Descartes. The tangent line can
ing Truth in the Sciences). In that same be taken as the perpendicular through P to the normal line. ■
year, Fermat sent an essay entitled Introduction to Plane and Carry out Descartes circle method for the parabola
Solid Loci to Paris, and in this essay he laid the foundation y2 = 4x at the point (4, 4).
for analytic geometry. Fermat's paper was more complete
68. i IS I OR]CAL QUEST Pierre de Fermat (see Problem 66)
and systematic, but Descartes' was published first. Descartes
also obtained the first method for differentiating polynomials,
is generally credited with the discovery of analytic geometry,
but his real love was number theory. His most famous prob¬
and we speak today of the Cartesian coordinate system and
lem has come to be known as Fermat's last theorem. He wrote
Cartesian geometry to honor Descartes' discovery. Today we
in the margin of a text: "To divide a cube into two cubes, a
describe analytic geometry from two viewpoints: (1) Given a
fourth power; or in general, any power whatever above the
curve, describe it by an equation (Descartes' viewpoint); and
second, into powers of the same denominations, is impossible,
(2) given an equation, describe it by a curve (Fermat's view¬
and I have assuredly found an admirable proof of this, but the
point).
margin is too narrow to contain it." ■
In this Historical Quest we will describe Descartes' circle
method for finding a tangent line to a given curve. This method In 1993 an imaginative, but lengthy, proof was con¬
uses algebra and geometry rather than limits. Descartes' structed by Andrew Wiles, and in 1997 Wiles was awarded the
method for finding the tangent line to the curve y — fix) at Wolfskehl Prize for proving this theorem. An engaging book,
the point P(x0, f (x0)) involved first finding the point Q{xx, 0), Fermat’s Enigma, shows how a great mathematical puzzle of
which is the point of intersection of the normal line to the curve our age was solved: Lock yourself in a room and emerge seven
at P with the x-axis, as shown in Figure 3.16. years later. Write a paper on the history of Fermat’s last the¬
orem, including some discussion of the current status of its
proof. [You might begin with “Fermat’s Last Theorem, The
Four Color Conjecture, and Bill Clinton for April Fools’ Day,”
by Edward B. Burger and Frank Morgan, American Mathe¬
matical Monthly, March 1997, pp. 246-250. Next, see the
NOVA film, The Proof, which was shown in 1997 on many
PBS stations (check with your library or PBS Online). Fi¬
nally, see “Fermat’s Last Stand” by Simon Singh and Kenneth
A. Ribet, Scientific American, November 1997, pp. 68-73.]
©69. If /'(c) 7^ 0, what is the equation of the normal line to
y = fix) at the point Pic, /(c))? What is the equation if
f'ic) = 0?
Figure 3.16 Descartes’ circle method 70. Suppose a parabola is given in the plane along with its axis
for finding tangent lines of symmetry. Explain how you could construct the tangent
line at a given point P on the parabola using only compass
Descartes then wrote the equation of the circle with center Q and straightedge methods. Hint: You may assume that the
passing through P: parabola has an equation of the form y = Ax2 in which the
y-axis is the axis of symmetry and the vertex of the parabola
(x - x,)2 + y2 = (x0 - x,)2 + y2 is at the origin. Then use the result of Problem 59.
If we had to compute the limit of a difference quotient every time we wished to find a
derivative, differentiation would not be the valuable tool that it is. Fortunately, there is
a better way, and in this section, we will derive several formulas that will enable us to
compute derivatives indirectly, without evaluating any limits.
3.2 Techniques of Differentiation
We begin by proving that the derivative of any constant function is zero. Notice that
this is plausible because the graph of the constant function fix) = k is a horizontal
line, and its slope is zero. Thus, for example, if /(x) = 5, then f'(x) = 0.
Proof Note that if /(x) = k, then /(x 4- Ax) = k for all Ax. Therefore, the
difference quotient is
f{x + Ax) - f(x) _ k-k _ Q
Ax Ax
and
fix + Ax) - fjx)
WARNING
Remember Ax f 0 even
/'(*) lim
A*—>0 Ax
lim 0 = 0
Ax—>0
though Ax 0. as claimed. □
Recall that a power function is a function of the form fix) = x", where n is a real
number. For example, fix) — x2, gix) = x-3, /i(x) = x1/2 are all power functions.
So are
—(x") = nx"_1
dx
Proof If the exponent n is a positive integer, we can prove the power rule by using the
binomial theorem with the definition of derivative. Begin with the difference quotient:
Ax
Ax
= nx"-1 + /7^2~~x"~2Ax -^
Note that Ax is a factor of every term in this expression except the first. Hence, as
Ax —► 0, we have
fix + Ax) - fix)
fix) — lim
Aat-^0 Ax
= nx
112 Chapter 3 Differentiation
For the following examples, and the problems at the end of this section, you may
assume that the power rule is valid when the exponent n is any real number. □
Solution
a. Applying the power rule with n = 8, we find that
*1/3
c. For this part you need to recognize that h(x) = —:— = jc-5/3. Applying the power
X2
rule with n = — we find that
The next theorem expands the class of functions that we can differentiate easily by
giving rules for differentiating certain combinations of functions, such as sums, differ¬
ences, products, and quotients. We will see that the derivative of a sum (difference) is
the sum (difference) of derivatives, but the derivative of a product (or quotient) does
not have such a simple form. For example, to convince yourself that the derivative of a
product is not the product of the separate derivatives, consider the power functions
df
Constant multiple [c/(x)]' = c/'(x) c
dx
df dg
Sum rule [f(x) + g(x)]' = fix) + g'(x) + g)
dx dx dx
The constant multiple, sum, and difference rules can be combined into a single rule,
which is called the linearity rule.
d
Linearity rule [af(x) + bg{x)]' = a fix) + bg'(x) (af + bg) + b**
dx dx
d , fdg df
Product rule [/(x)g(x)]' = /(x)g'(x) + f'(x)g{x) ~rifg) = f~r +g
dx dx dx
df .dg
'fix)T = gix) fix) - fjx)g'jx) gi
dx— /y~
dx
Quotient rule
g(*)J fgU)]2 g2
WARNING
Note that the order of terms in the quotient rule matters because
gf ~ fg' # fg' ~ gf
Proof We will prove the product rule in detail, leaving the other rules as problems.
Let f{x) and g(x) be differentiable functions of x and let p(x) — f(x)g(x).
We will add and subtract the term f(x + Ax)g(x) to the numerator of the difference
quotient for p(x) to create difference quotients for fix) and g(x). Thus,
Note that in each part of Theorem 3.5, we prove the differentiability of the appro¬
priate functional combination at the same time we are establishing the differentiation
formula.
EXAMPLE 2 Using the basic rules to find derivatives
Differentiate each of the following functions.
Solution
a. Apply the linearity rule (constant multiple, sum, and difference) and power rules:
b. Apply the product rule; then apply the linearity and power rules:
= 6x(5x3 — x + 7)
c. Apply the quotient rule, then the linearity and power rules:
F(x) — \x 2 — |x + | + 1 -f" x 1
In applying the power rule term by term in Example 2e, we really used the follow¬
ing generalization of the linearity rule.
Example 3 illustrates how the extended linearity rule can be used to differentiate a
polynomial.
Solution
dx
dr d d d
= 2—(x5) - 3 —(x2) + 8 —(x) - —(5) Extended linearity rule
dx dx
= 2(5x4) -3(2x) + 8(1) -0 Power rule; constant rule
— 10x4 — 6x + 8
Solution
We could expand the product function p(x) as a polynomial and proceed as in Exam¬
ple 3, but it is easier to use the product rule.
p'(x) = (x3 — 4x + 7)(3x5 — x2 + 6x)' + (x3 — 4x + 7)'(3x5 — x2 + 6x)
Solution
Evaluating /(x) atx = —1, we find that /(—1) = -4 (verify); therefore, the point of
tangency is (—1, -4). The slope of the tangent line at (—1, —4) is /'(— 1). Find f'(x)
by applying the quotient rule;
, (2x2 + x - 3)(3x2 + 5)' - (3x2 + 5)(2x2 + x - 3)'
f
y.
/
/
7
/
f{x)~ (2x2 + x — 3)2
(2x2 + x — 3)(6x) — (3x2 + 5)(4x + 1)
r
J (2x2 + x — 3)2
// * The slope of the tangent line is
—
/
/
1
1
x\
(2 — 1 — 3)(—6) — (3 + 5)(—4 + 1) (—2)(—6) — (8)(—3)
tat igeht line
at +1 — ) ^ ^ From the formula (Theorem 3.1) y = /'(x0)(x -x0) + f(x 0), we find that an equation
for the tangent line at (—1, —4) is
Solution
The tangent line will be horizontal when dy/dx — 0, because the derivative dy/dx
measures the slope and a horizontal line has slope 0 (see Figure 3.18). Applying the
product rule, we find
= 2x2 - 8 + x2 + 4x - 1
Thus, -j- = 0 when x = | or x = —3. The corresponding points (f, — §7) and
(—3, 50) are the points on the curve at which the tangent line is horizontal. ■
In the following example, we use the quotient rule to extend the proof of the power
rule to the case in which the exponent n is a negative integer.
Solution
We have /(x) = xn — x~m = l/xm, so apply the quotient rule:
HIGHER-ORDER DERIVATIVES
Leibniz notation
dj
First derivative: y fix)
dx
or A/to
dx
d2y d2
Second derivative: y" fix) or
dx \dx ) dx2
d3y d3 '
Third derivative: y"> fix) or
dx \dx2 ) dx3
3.2 Techniques of Differentiation 117
Leibniz notation
d4y d4
Fourth derivative: f(4\x) or ——T fix)
dx4 dx4
d" y d"
nth derivative f(n)(x) or - fix)
dxn dx"
You should note that all higher-order derivatives of a polynomial p(x) will also be
polynomials, and if p has degree n, then p{k\x) = 0 for k > n + 1, as illustrated in
the following example.
Solution
27. fix) = x2 — 3x — 5, where x = —2 53. HISTORICAL QUEST When working with rational ex¬
28. fix) = x5 — 3x3 — 5x + 2, where x = 1 pressions, we need to be careful about division by zero. One
29. fix) = (x2 + 1)(1 — x3), where x — 1
of the earliest recorded treatments of division by zero is at¬
tributed to the Hindu mathematician Aryabhata (476-550), in
x + 1
30. fix) , where x = 0 whose honor the first Indian satellite was named. He also gave
x - 1
rules for approximations of square roots and sums of arithmetic
x2 + 5
31. fix) , where x 1 progressions as well as rules for basic algebraic manipula¬
x +5 tions. One example of his work is the following calculation for
1 2 it: "Add four to one hundred, multiply by eight and add again
32. fix) 1-1-—, where x = 4
x V* sixty-two thousand; the result is the approximate value of the
Fine/ the coordinates of each point on the graph of the given func¬ circumference of a circle whose diameter is twenty-thousand."
tion where the tangent line is horizontal in Problems 33-39. Follow the steps of Aryabhata’s approximation for n. After
33. f(x) = 2x3 - lx2 + 8x - 3 you have completed this demonstration, discuss the procedure
34. f{t) = r4 + 4t3 - St2 + 3 35. g{x) = (3x - 5)(x - 8) and technology you used and contrast it with the tools that
Aryabhata must have had available.
36. fit) = — - — 37. fix) = y/xix - 3) Q 54. What is the relationship between the degree of a polynomial
function P and the value of k for which P(k\x) is first equal
4x2 + 12x + 9
38. hiu) = —— iu + 9) 39. hix) = to 0?
■s/u 2x + 3
55. Prove the constant multiple rule (c/)' = cf.
Q 40. a. Differentiate the function fix) = 2x2 — 5x — 3.
56. Prove the sum rule (/ + g)' — f + g'.
b. Factor the function in part a and differentiate by using the
product rule. Show that the two answers are the same. 57. Use the definition of the derivative to find the derivative of f2,
given that / is a differentiable function.
2x - 3
41. a. Use the quotient rule to differentiate fix) —--—. 58. Prove the product rule by using the result of Problem 57 and
x3
the identity
b. Rewrite the function in part a as fix) = x_3(2x — 3) and
differentiate by using the product rule.
fg = |[(/ + g)2 - f2 ~ g2]
c. Rewrite the function in part a as fix) = 2x~2 — 3x-3 and
differentiate. 59. Prove the quotient rule
d. Show that the answers to parts a, b, and c are all the same.
42. Find numbers a, b, and c that guarantee that the graph of the /V _gf'-fg'
function fix) = ax2 + bx + c will have x-intercepts at (0, 0) s) g2
and (5, 0) and a tangent line with slope 1 where x = 2.
where g(x) 0. Hint: First show that the difference quotient
43. Find the equation for the tangent line to the curve with equa¬
for f/g can be expressed as
tion y = x4 — 2x + 1 that is parallel to the line 2x — y — 3 = 0.
44. Find equations for two tangent lines to the graph of / /
3x T 5 -(x + Ax) - -(x) f{x + Ax)g{x) _ /(*)g(x + Ax)
fix) =-that are perpendicular to the line 2x — y = 1.
1 +x Ax (Ax)g(x + Ax)g(x)
45. Let fix) = (x3 — 2x2)(x + 2).
and then subtract and add the term g(x) f(x) in the numerator.
a. Find an equation for the tangent line to the graph of / at
60. Show that the reciprocal function r(x) = l//(x) has the
the point where x = 1.
derivative r'(x) = —f'ix)/[ Ax)]2 at each point x where f is
b. Find an equation for the normal line to the graph of / at
differentiable and fix) jz 0.
the point where x = 0.
61. If /, g, and h are differentiable functions, show that the prod¬
46. Find an equation for a normal line to the graph of
uct fgh is also differentiable and
fix) — (x3 — 2x2)(x + 2) that is parallel to the line
x - 16y + 17 = 0. ifgh)' = fgh' + fgh + fgh
47. Find all points (x, y) on the graph of y = 4x2 with the prop¬
erty that the tangent line at (x,y) passes through the point 62. Let / be a function that is differentiable at x.
(2, 0). a. If gix) = [fix)]3, show that g\x) = 3[/(x)]2/'(x).
48. Find the equations of all the tangent lines to the graph of the Hint: Write g(x) — [/(x)]2/(x) and use the product rule.
function fix) = x2 — 4x + 25 that pass through the origin. b. Show that p(x) = [/(x)]4 has the derivative
Determine which (if any) of the functions y = fix) given in Prob¬
lems 49-52 satisfy the equation p'(x) = 4 [fix)]3 fix).
y'" + y" + y' = x + 1 63. Find constants A, B, and C so that y = Ax3 + fix + C satisfies
the equation
49. fix) 2 + 2x - 3 50. f(x) - XJ
-3 -p X2 + X
/// i /-> h o r ,
51. fix) = \x22 + 3 .2
52. f(x) — 2x2 +x y +2y -3y+y = x
3.3 Derivatives of Trigonometric, Exponential, and Logarithmic Functions 119
Figure 3.19 Calculator graph From the graph of this difference quotient, it appears that the derivative of /(x) = sin x
of the difference quotient is /'(x) = cosx. We now verify this with the following theorem, which uses the
for y — sinx following limits established earlier in Theorem 2.3:
sin/? cos h — 1
lim-= 1 lim---= 0
h-s-0 h h—>0 h
Proof The proofs of these two formulas are similar. We will prove the first using the
trigonometric identity
sin(a + f) = sin a cos f + cos a sin f
and leave the proof of the second formula as a problem. From the definition of the
derivative
d . sin(x + Ax) — sinx
— sinx = lim -
dx Ax—o Ax
sin x cos Ax + cos x sin Ax — sin x
= lim
Ax—0 Ax
cos Ax sin Ax sinx
= lim sin x + cosx
Ax—0 Ax Ax
cos Ax — 1 sin Ax
(sinx) lim 4* (cosx) lim
Ax—o Ax Ax->0 Ax
= (sinx)(0) + (cosx)(l)
= cosx □
Solution
f'(x) — —-(x2 sinx)
dx
, d d j
x — (sinx) + sinx— (x ) Product rule
dx dx
x2 cosx + 2x sinx Power rule and derivative of sine ■
^t-1/2(cost + 21 sint)
Common factor ^t
cos2 t
cos t + 21 sin t
2-s/t cos21
You will need to be able to differentiate not only the sine and cosine functions, but also
the other trigonometric functions. To find the derivatives of these functions you will
need the following identities, which are given in the Student Mathematics Handbook
and Appendix E.
sinx cosx
tanx = cotx =
cosx sinx
1 1
secx = cscx =
cosx sinx
You will also need the following identities:
cos2 x + sin2 x = 1, 1 + tan2 x = sec2 x, cot2 x + 1 = esc2 x
3.3 Derivatives of Trigonometric, Exponential, and Logarithmic Functions 121
d d
— sec x - sec x tan x — CSC X = — CSC X cot X
dx dx
Proof The derivatives for the sine and cosine functions were given in Theorem 3.6.
All the other derivatives in this theorem are proved by using the quotient rule along
with formulas for the derivatives of the sine and cosine functions. We will obtain the
derivative of the tangent function and leave the rest as problems (see Problems 56-59).
d d sinx
— tanx =- Trigonometric identity
dx dx cosx
d d
cosx— sinx — sinx— cosx
dx_dx_ Quotient rule
cos2x
cos2 x + sin2 x
cos2 x
1
cos2 x + sin2 x = 1
COS2 X
, d
Solution f {0) — — (30 sec 6)
= 30— sect? + sec# — (30) Product rule
dO dO
= 30 sec 0 tan 0 + 3 sec 0 ■
, d
Solution f (x) = —(secx tanx)
dx
d d „ ,
= secx—tan x +tanx—secx Product rule
dx dx
— secx(sec2 x) + tanx (secx tanx)
Solution
When x = we have
2tv 2n -V3 / 2a/3\ -5^3
3 3 3 \ 3 ) 3
so the point of tangency is P(-f). To find the slope of the tangent line at P, we
first compute the derivative ^:
dy d
— = —(cotx — 2cscx)
dx dx
V
-8
T
so the tangent line at (y-, —^) with slope — | is
5 a/3 8/ 2tt\
•v +1-~ _3 vx _ tJ
24x + 9y + 15 >/3 — 16tt = 0
The next theorem, which is easy to prove and remember, is one of the most important
results in all of differential calculus.
{ex) = e*
dx
lim ( 1 H— I = e
n-> °o V n )
Let n = so that lim (1 + Ax)1/Ax = e. This means that for Ax very small.
Aa-»0
,Aa
e ^ (1 + Ax)1/4* or eAx ^ 1 + Ax, so that eAx — 1 Ax. Thus, lim 1.
aa—>o Ax
Finally, using the limit in the definition of derivative for ex, we obtain
^(a+Aa) _ ex
d y
— (e ) — lim
dx aa—>0 Ax
ex(eAx - 1)
= lim
Aa->0 Ax
?Ajr - 1
= ex lim
aa—>o Ax
= ex{\) = ex
3.3 Derivatives of Trigonometric, Exponential, and Logarithmic Functions 123
y.
y. xX 4
A Note: An easier proof of the derivative formula in Theorem 3.8 is given in Example 6
y
y
of Section 3.6 using methods developed in that section.
_ y
y
y
/ -'
The fact that — {ex) = ex means that the slope of the graph of y = ex at any point
ea kk Ipj >e c f li ie {a, ea) is m = ea, the y-coordinate of the point, as shown in Figure 3.20. This is one
i >> m ,a
,a
y
y
of the features of the exponential function y = ex that makes it “natural.”
y X
✓
y
S
✓
✓
y
* EXAMPLE 7 A derivative involving ex
//
y Differentiate fix) — x2ex. For what values of x does fix) = 0?
Solution
Figure 3.20 The slope of y = e*
ex at Using the product rule, we find
each point (a, ea) is m — ea.
_dx J _dx
= x2ex + 2xex Exponential and power rules
Solution
Apply the product rule twice:
x x
Let h = -, so Ax = —.
Ax h
124 Chapter 3 Differentiation
h ( r
lim — In I 1 + — h —+oo as Ax -> 0
h-++oo X V h
1
lim h In ( 1 + —
X h->+oo \ h
1
lim In ( 1 + - Power rule for logarithms
X h-*+oo \ h
( 1\
= - In lim 1 + — ) Since lnx is continuous,
x h—>+oo V h) use the composition limit rule.
1
= — In e Definition of e
x
_ 1
\ne — 1 □
x
TECHNOLOGY NOTE
When you are using calculators or computer programs such as Mathematica, Derive, or Maple, the form of the derivative may
vary. For example, you might obtain
d f,
— (tanx) = tan-x + 1
dx Fi7*r F2t f F3t FAT r fs i Fb )
t i— R1 gebra Ca 1 c Other PrgnlO Clear a-z...
instead of sec2 x as found in this section;
d ^
— (cotx) = — cot x — 1
dx
instead of — esc2 x;
d \ ln(x) T ln(x)•cos(x) 1
d sinx dx{.sin(x)J x ■ sin(x)
— (secx) = —— (sin(x))2
dx coszx ti!<ln<x>/sin<x>,x>
MAIN RAD AUTO FUNC 1/30
instead of sec x tan x; and
d cosx Figure 3.21 Sample output for Example 9. Notice
— (esex) = —^r—
dx sin x that the form of the answer differs from that shown
instead of — escx cotx. A solution to Example 9, text-
using a TI-92, is shown in Figure 3.21.
Although the form of this calculator solution is different from the form shown in the example, you should notice that they
are equivalent by using algebra (and recalling some of the fundamental identities from trigonometry).
A good ^est for your calculator is to try to find the derivative of |x| at x = 0, which we know does not exist. Many
calculators will give the incorrect answer of 0.
3.4 Rates of Change: Modeling Rectilinear Motion 125
The speed of a car or airplane; interest rates; the growth rate of a population; the drip
rate of an intravenous injection—these are examples of the many situations in which
rates of change are an important consideration in practical problems. In this section, we
begin by showing how rates of change can be computed using differentiation. Then we
126 Chapter 3 Differentiation
y,
H
rr
Ay
)
de :>er ds on soints chc)se 1.
A / i
i
i
i
i
/ Ay i
l
,y)
AX
__
0.5 i5 25 35 ■ 45 *
a. Linear function: rate of b. Nonlinear function: rate of change
change Ay/Ax is constant. Ay/Ax depends on chosen points.
Because the slope of the tangent line is given by the derivative of the function,
the preceding geometric observations suggest that the rate of change of a function is
measured by its derivative. This connection will be made more precise in the following
discussion.
Suppose y is a function of x, say y = /(x). Corresponding to a change from
x to x + Ax the variable y changes from /(x) to /(x + Ax). The change in y is
Ay = /(x + Ax) — /(x), and the average rate of change of y with respect to x is
WARNING
This formula for the change in y is important. Find Ay in Figure 3.23.
As the interval over which we are averaging becomes shorter (that is, as Ax -> 0),
the average rate of change approaches what we would intuitively call the instanta¬
neous rate of change of y with respect to x, and the difference quotient approaches
Figure 3.23 A change in Ay dy
corresponding to a change Ax the derivative —. Thus, we have
dx
To summarize,
3.4 Rates of Change: Modeling Rectilinear Motion 127
Instantaneous Rate of Change Suppose f(x) is differentiable at * = xo. Then the instantaneous rate of
change of y = f(x) with respect to x at x0 is the value of the derivative of
/ at xq. That is,
Solution
For any x, the instantaneous rate of change is the derivative,
dy n 2
— = 2x sin x + x cos x
ax
Thus, the rate when x = n is
dy
= lit sin;r +7r2cos7r = 2jt(0) + 7r2(—1) = —tt2
dx
The negative sign indicates that when x = tc, the function is decreasing at the rate of
7i2 ~ 9.9 units of y for each one-unit increase in x. ■
Let us consider an example comparing the average rate of change with the instan¬
taneous rate of change.
Solution
a. The (average) rate of change from x = 3 to x = 5 is found by dividing the change
in / by the change in x. The change in / from x — 3 to x = 5 is
/(5) - /(3) = 8
5-3 2
The slope of the secant line is 4, as shown in Figure 3.24.
b. The derivative of the function is
/'(3) = 2(3) -4 = 2
For instance, if you are earning $15,000/yr and receive a $3,000 raise, you would
probably be very pleased. However, that $3,000 raise might be much less meaningful
if you were earning $100,000/yr. In both cases, the change is the quantity $3,000, but
in the first case the relative rate of change is 3,000/15,000 = 20%, whereas in the
second, it is a much smaller 3,000/100,000 = 3%.
In terms of the derivative, the relative rate of change may be defined as follows:
Relative Rate of Change The relative rate of change of y = f(x) at x — xq is given by the ratio
fix o)
RELATIVE RATE OF CHANGE =
fix o)
lnx — 2
A Ot) = 110
x
where x is the person’s age. What is the relative rate of change of the aerobic rating of
a 20-year-old person? A 50-year-old person?
Solution
Applying the quotient rule, we obtain the derivative of AQc):
'3 — lnx"
o
X2 3 — lnx
RELATIVE RATE =
1
A'(20) 3 - In 20
0.0002143
A (20) (In 20-2)20
In other words, at age 20, the person’s aerobic rating is increasing at the rate of
0.02%/yr. However, at age 50, the relative rate is
A'(50) 3-In50
-0.0095399
A(50) “ (In50-2)50
and the person’s aerobic rating is decreasing (because of the negative sign) at the rela¬
tive rate of 0.95%/yr. ' ■
3.4 Rates of Change: Modeling Rectilinear Motion 129
FsUA
(FORCE = MASS ■ ACCELERATONI
Q « AT. A V
(HEATING OF AIR PROOUCES
change in temperature
AND VOLUME)
a. _ b. _ c. _ d. _ e.
X J X y X y X X y
-4 2.00 -4 - 20.7 0.0 0.0 -4 0.27 3 0.00
-3 1.28 -3 - 16.5 1.0 1.7 -3 0.30 4 0.69
-2 0.72 -2 -12.3 1.5 2.0 -2 0.33 5 1.10
-1 0.32 -1 - 8.2 2.5 1.2 -1 0.36 6 1.39
0 0.08 0 - 3.8 3.5 - 0.7 0 0.40 7 1.61
1 0.00 1 0.4 4.5 -1.9 1 0.44 8 1.79
2 0.08 2 4.7 5.5 - 1.4 2 0.49 9 1.95
3 0.32 3 8.9 6.0 - 0.6 ' 3 0.52 10 2.08
4 0.72 4 13.12 6.5 0.4 4 0.60 11 2.20
12 2.30
Solution
Sketch each graph and observe the pattern (if any) of the graphs.
-t - -2 i5. , 3 V
*10
!15
d. The data points look almost linear (but not e. The data points seem to indicate a
quite), so we hypothesize that the data logarithmic model.
points may indicate an exponential model.
y T
:
•
1.0-
_ 1 Lw
7 1i 0 1 1 l'2 13:v
..LI L.
3.4 Rates of Change: Modeling Rectilinear Motion 131
Mathematical models will appear in both examples and exercises throughout this
text. In the next subsection, we begin our work with modeling by studying rectilinear
motion, an important topic from physics.
Rectilinear motion is motion along a straight line. For example, the up and down
motion of a yo-yo may be regarded as rectilinear, as may the motion of a rocket early
in its flight.
When studying rectilinear motion, we may assume that the object is moving along
a coordinate line. The position of the object on the line is a function of time t and is
often expressed as s(t). The rate of change of s(t) with respect to time is the object’s
velocity v(t), and the rate of change of the velocity with respect to t is its acceleration
a(t). The absolute value of the velocity is called speed. Thus,
ds
Velocity is v(t)
It
ds
Speed is |u(?)|
dt
dv d2s
Acceleration is a(t) = — =
dt dt2
If v(t) > 0, we say that the object is advancing, and if v(t) < 0, it is retreating. If
v(t) = 0, the object is neither advancing nor retreating, and we say it is stationary.
The object is accelerating when a(t) > 0 and is decelerating when a(t) < 0. The
significance of the acceleration is that it gives the rate at which the velocity is chang¬
ing. When you are riding in a car, you do not feel the velocity, but you do feel the
acceleration. That is, you feel changes in the velocity.
These ideas are summarized in the following box.
Rectilinear Motion
An object that moves along a straight line with position s(t) has velocity
ds _ dv d2s
v(t) = — and acceleration a(t) = —— = —— when these derivatives exist. The
dt dt dt2
speed of the object is )v(i)|.
for t on [0, 8]. Find the initial position, velocity, and acceleration for the object and
discuss the motion. Compute the total distance traveled.
132 Chapter 3 Differentiation
Solution
The position at time t is given by the function s. The initial position occurs at time
t — 0, so
s(0) = 0 The object starts at the origin.
The velocity is determined by finding the derivative of the position function.
v(t) = s'(t) = 9f2-81f + 162
4 192 -18 -9
—
4.5 182.25 -20.25 0 r=3 f=0
a = -27 a = -81
5 172.5 -18 9
6 162 0 27
7 178.5 36 45
-20.25 0 90 162 v
8 240 90 63
182.25
a. Position of the object: when v > 0, the b. Velocity of the object: when a > 0, the
object advances, and when v < 0, the velocity increases, and when a < 0, the
object retreats. velocity decreases.
Recall that the speed of the object is the absolute value of its velocity. The speed
decreases from 162 to 0 between / = 0 and t = 3 and increases from 0 to 20.25 as
the velocity becomes negative between t = 3 and t — 4.5. Then for 4.5 < t < 6, the
object slows down again, from 20.25 to 0, after which it speeds up.
The total distance traveled is
|s(3) - 5(0)| + |s(6) - 5(3)| + k(8) - 5(6)| = 202.5 + 40.5 + 78 = 321 ■
3.4 Rates of Change: Modeling Rectilinear Motion 133
The speed increases (particle speeds up) when the velocity and acceleration have
the same signs.
The speed decreases (particle slows down) when the velocity and acceleration have
opposite signs.
where so and i>o are the projectile’s initial height and velocity, respectively. You are
asked to derive this formula for earth’s gravity and a particular initial velocity and a
particular initial position in a problem in Section 5.1.
r=3 Suppose a person standing at the top of the Tower of Pisa (176 ft high) throws a ball
h = 320 directly upward with an initial speed of 96 ft/s.
v=0
a. Find the ball’s height, its velocity, and acceleration at time /.
b. When does the ball hit the ground, and what is its impact velocity?
c. How far does the ball travel during its flight?
Solution
First, draw a picture such as the one shown in Figure 3.27 to help you understand the
problem.
a. Substitute the known values into the formula for the height of an object:
d2h
= -32
dt2
This means that the acceleration of the ball is always decreasing at the rate of 32
ft/s2.
b. The ball hits the ground when h(t) = 0. Solve the equation
to find that this occurs when t & —1.47 and t & 7.47. Disregarding the negative
value, we see that impact occurs at t & 7.47 sec. The impact velocity is
The negative sign here means the ball is coming down at the moment of impact,
c. The ball travels upward for some time and then falls downward to the ground, as
shown in Figure 3.27. We need to find the distance it travels upward plus the dis¬
tance it falls to the ground. The turning point at the top (the highest point) occurs
when the velocity is zero. Solve the equation
-321 + 96 = 0
to find that this occurs when t = 3. The ball starts at h(0) = 176 ft and rises to a
maximum height when t — 3. Thus, the maximum height is
c. Describe the motion of the object; that is, tell when it is ad- 29. 30.
vancing and when it is retreating. Compute the total distance X y X y
traveled by the object during the indicated time interval.
-4 238 -4 4.84
d. Tell when the object is accelerating and when it is decelerat¬ -3 173 -3 2.46
ing. -2 118 -2 -1.41
For a review of solving equations, see Sections 2.5 and -1 73 -1 -4.42
4.7 of the Student Mathematics Handbook.
0 38 0 -4.76
1 13 1 -2.20
17. s(t) = r2 — 2f + 6 on [0. 2] 2 -2 2 1.68
3 -7 3 4.55
18. s(t) = 3t2 + 2t — 5 on [0, 1]
4 -2 4 4.66
37. An object moving on the x-axis has position 46. An astronaut standing at the edge of a cliff on the moon throws
a rock directly upward and observes that it passes her on the
x (0 = ?3 - 9r + 24r + 20 way down exactly 4 seconds later. Three seconds after that,
after ? seconds. What is the total distance traveled by the ob¬ the rock hits the ground at the base of the cliff. Use this in¬
ject during the first 8 seconds? formation to determine the initial velocity v0 and the height of
the cliff. Note: g = 5.5 ft/s2 on the moon.
38. A car has position
47. Answer the question in Problem 46 assuming the astronaut is
s(t) = 50? In? + 80 on Mars, where g = 12 ft/s2.
48. A car is traveling at 88 ft/s (60 mi/h) when the driver applies
ft after ? seconds of motion. What is its acceleration (to the
the brakes to avoid hitting a child. After t seconds, the car is
nearest hundredth ft/s2) after 10 seconds?
s(t) = 88? — 8?2 feet from the point where the brakes were
39. An object moves along a straight line so that after ? minutes,
first applied. How long does it take for the car to come to a
its position relative to its starting point (in meters) is
stop, and how far does it travel before stopping?
49. It is estimated that t years from now, the circulation of a local
s(t) = 10? H-
•* e' newspaper can be modeled by the formula
a. At what speed (to the nearest thousandth m/min) is the ob¬
C(t) = 100?2 + 400? + 50? In t
ject moving at the end of 4 min?
b. How far (to the nearest thousandth m) does the object ac¬ a. Find an expression for the rate at which the circulation will
tually travel during the fifth minute? be changing with respect to time ? years from now.
40. A bucket containing 5 gal of water has a leak. After ? seconds, b. At what rate will the circulation be changing with respect
there are to time 5 years from now?
c. By how much will the circulation actually change during
e«) = 5(1 - i)2 the sixth year?
50. An efficiency study of the morning shift at a certain factory
gallons of water in the bucket. indicates that an average worker who arrives on the job at
a. At what rate (to the nearest hundredth gal) is water leaking 8:00 A.M. will have assembled
from the bucket after 2 seconds?
f(x) = — |x3 + |x2 + 50x
b. How long does it take for all the water to leak out of the
bucket?
units x hours later.
c. At what rate is the waterjeaking when the last drop leaks
a. Find a formula for the rate at which the worker will be as¬
out?
sembling the units after x hours.
Modeling Problems In Problems 41-48, set up and solve an ap¬ b. At what rate will the worker be assembling units at
propriate model to answer the given question. Be sure to state your
9:00 A.M.?
assumptions.
c. How many units will the worker actually assemble between
41. A person standing at the edge of a cliff throws a rock directly 9:00 A.M. and 10:00 A.M.?
upward. It is observed that 2 seconds later the rock is at its
51. An environmental study of a certain suburban community
maximum height (in ft) and that 5 seconds after that, it hits
suggests that ? years from now, the average level of carbon
the ground at the base of the cliff.
monoxide in the air can be modeled by the formula
a. What is the initial velocity of the rock?
b. How high is the cliff? q{t) = 0.05?2 + 0.1? + 3.4
c. What is the velocity of the rock at time ??
parts per million.
d. With what velocity does the rock hit the ground?
a. At what rate will the carbon monoxide level be changing
42. A projectile is shot upward from the earth with an initial ve¬
with respect to time one year from now?
locity of 320 ft/s.
b. By how much will the carbon monoxide level change in the
a. What is its velocity after 5 seconds?
first year?
b. What is its acceleration after 3 seconds?
c. By how much will the carbon monoxide level change over
43. A rock is dropped from a height of 90 ft. One second later the next (second) year?
another rock is dropped from height H. What is H (to the
52. According to Newton’s law of universal gravitation, if an ob¬
nearest foot) if the two rocks hit the ground at the same time?
ject of mass M is separated by a distance r from a second
44. A ball is thrown vertically upward from the ground with an
object of mass m, then the two objects are attracted to one an¬
initial velocity of 160 ft/s.
other by a force that acts along the line joining them and has
a. When will the ball hit the ground? magnitude
b. With what velocity will the ball hit the ground?
GmM
c. When will the ball reach its maximum height?
F = —
45. An object is dropped (initial velocity v0 = 0) from the top
of a building and falls 3 seconds before hitting the pavement where G is a positive constant. Show that the rate of change
below. Determine the height of the building in ft. of F with respect to r is inversely proportional to r3.
3.4 Rates of Change: Modeling Rectilinear Motion 137
53. The population of a bacterial colony is approximately 61. An object attached to a helical spring is pulled down from
its equilibrium position and then released, as shown in Fig¬
P(t) = P0 + 61r + 3/2 ure 3.28.
P(x) = 2x + 4x3/2 + 5,000 Suppose that t seconds later, its position (in centimeters, mea¬
sured relative to the equilibrium position) is modeled by
a. At what rate will the population be changing with respect
to time 9 months from now? s(t) = 7 cos t
b. At what percentage rate will the population be changing a. Find the velocity and acceleration of the object at time t.
with respect to time 9 months from now?
b. Find the length of time required for one complete oscilla¬
56. Assume that your starting salary is $30,000 and you get a raise tion. This is called the period of the motion.
of $3,000 each year.
c. What is the distance between the highest point reached by
a. Express the percentage rate of change of your salary as a the object and the lowest point? Half of this distance is
function of time. called the amplitude of the motion.
b. At what percentage rate will your salary be increasing after 62. Two cars leave a town at the same time and travel at constant
one year? speeds along straight roads that meet at an angle of 60° in the
c. What will happen to the percentage rate of change of your town. If one car travels twice as fast as the other and the dis¬
salary in the long run? tance between them increases at the rate of 45 mi/h, how fast
57. The GDP of a certain country is growing at a constant rate. In is the slower car traveling?
1999 the GDP was 125 billion dollars, and in 2001 it was 155 63. Spy Problem In an attempt to divert the Spy from seeking re¬
billion dollars. At what percentage rate did the GDP grow in venge for the death of his friend Siggy (Problem 62, Sec¬
2000? tion 2.4), his superior, Lord Newton Fleming (affectionately
58. If y is a linear function of x, what will happen to the percent¬ known as the “Flamer”), assigns him to a mission in space.
age rate of change of y with respect to x as x increases without However, an encounter with an enemy agent leaves him with
bound? a mild concussion that causes him to forget where he is. For¬
59. According to Debye’s formula in physical chemistry, the ori¬ tunately, he remembers the values of g for various heavenly
entation polarization P of a gas satisfies bodies (32 ft/s2 on earth, 5.5 ft/s2 on the moon, 12 ft/s2 on
Mars, and 28 ft/s2 on Venus). To deduce his whereabouts, he
throws a rock directly upward (from ground level) and notes
that it reaches a maximum height of 37.5 ft and hits the ground
5 s after leaving his hand. Where is he?
where p,, k, and N are constants and T is the temperature of Q 64. Find the rate of change of the volume of a cube with respect
the gas. Find the rate of change of P with respect to T. to the length of one of its edges. How is this rate related to the
60. A disease is spreading in such a way that after t weeks, for surface area of the cube?
0 < t < 6, it has affected 65. Show that the rate of change of the volume of a sphere with
respect to its radius is equal to its surface area.
N(t) = 5- t2(t - 6)
66. Van der Waal’s equation states that a gas that occupies a vol¬
hundred people. Health officials declare that this disease will ume V at temperature T (Kelvin) exerts pressure P, where
reach epidemic proportions when the percentage rate of in¬
crease of N(t) at the start of a particular week is at least 30% (p +-^J(V - B) = kT
per week. The epidemic designation level is dropped when the
percentage rate falls below this level.
and A, B, and k are physical constants. Find the rate of change
a. Find the percentage rate of change of N(t) at time t.
of pressure with respect to volume, assuming fixed tempera¬
b. Between what weeks is the disease at the epidemic level? ture.
138 Chapter 3 Differentiation
Suppose it is known that the carbon monoxide pollution in the air is changing at the
rate of 0.02 ppm (parts per million) for each person in a town whose population is
growing at the rate of 1,000 people per year. To find the rate at which the level of
pollution is increasing with respect to time, we form the product
Proof A rigorous proof involves a few details that make it inappropriate to include
at this point in the text. A justification (partial proof) is given at the end of this section,
and a full proof of the chain rule is included in Appendix B.
WARNING n .. r . „ , . ,. ,. ,dv . . ,
Kecall our earlier warning in Section 3.1 against thinking ot ^ as a traction. That
said, there are certain times when this incorrect reasoning can be used as a mnemonic device.
For instance, "canceling du," as indicated below, makes it easy to remember the chain rule:
dy dy-dir
dx -drtrdx
Solution
dy 9 du
Because — = 3rr — 6u and — = 2x, it follows from the chain rule that
du dx
dy dy du
~ = —— — (3u2 - 6u)(2x)
dx du dx
Notice that this derivative is expressed in terms of the variables jc and u. To express
dy/dx in terms of x alone, we substitute u — x2 + 2 as follows:
The chain rule is actually a rule for differentiating composite functions. In partic¬
ular, if y = f(u) and u = h(x), then y is the composite function y = (/ o u)(x) =
f[u(x)], and the chain rule can be rewritten as follows:
□
EXAMPLE 2 The chain rule applied to a power
Differentiate y = (3x4 — 7x + 5)3.
Solution
Here, the “inner” function is u{x) — 3x4 — lx + 5 and the “outer” function is u3, so
we have
y' — (u3)'[u(x)]'
= (3m2)(12x3 — 7)
With a lot of work, you could have found the derivative in Example 2 without
using the chain rule either by expanding the polynomial or by using the product rule.
The answer would be the same but would involve much more algebra. The chain rule
allows us to find derivatives that would otherwise be very difficult to handle.
Differentiate g(x) =
1 — 3x
Solution
1/4
and we have
-3/4
1
g\x) -
4 V 1 - 3x 1 - 3x
_3/4 f (1 -3x)(l)-x(-3)
Quotient rule
4 V 1 - 3x (1 -3x)2
140 Chapter 3 Differentiation
i/ , y3/- 1
4 V 1 - 3x ) .(1 — 3x)2_
1
4x3/4(l - 3x)5/4
dx dx
Extended Trigonometric Rules
d . du d du
— sin u — cos u — — cos u — — sin u —
dx dx dx dx
d dU
r\ d 2 du
— tan u — sec' u — — cot u = — csc“u—
dx dx dx dx
d du d du
— sec u = sec u tan u — — esc u = — CSC u cot u —
dx dx dx dx
Extended Exponential and Logarithmic Rules
d „ ,du d l du
—eu — eu — —In u =-
dx dx dx u dx
Solution
Think of this as /(u) = sin u, where u — 3x2 + 5x — 7, and apply the chain rule:
= (6x + 5) cos(3x2 + 5x — 7) ■
dg d 9 d ,
Solution — = — cosx^ -f 5 —(3x 1 + 4)6
dx dx dx
. 9 d 9 d
= — sinx2 —(x“) + 5 6(3x_1 + 4)5-^-(3x_1 +4)
dx dx
= (-sinx2)(2x) + 30(3x 1+4)5(-3x 2)
= —2x sinx2 — 90x-2(3x~' + 4)5
Solution
dy , , d -
— = 4cos3(3x + 1)'— cos(3x + 1)" Extended power rule
dx dx
Solution
(1 — 4x)f
r ——d tan lx — tan lx
r —d (1 — 4x)5f
dp dx dx
dx [(1 — 4x)5]2
Solution
Use the product rule:
d e-lx ■ d
[e 3r sinx] = e — (sinx) + sinx Product rule
dx dx dx
-3jc i
= e “"(cosx) + [e 3x(—3)] sinx
= e~3'v(cosx — 3 sinx)
Solution
Because horizontal tangent lines have zero slope, we need to solve the equation
f'(x) = 0. We find that
= 10x(4x + 5)2(2x + 1)
From the factored form of the derivative, we see that f'(x) = 0 when x = 0, x — — |,
and x so these are the x-coordinates of the points on the graph of / where
horizontal tangent lines occur. ■
C(p) = j0.5p2+ 17
parts per million when the population is p thousand. It is estimated that t years from
now, the population of the community will be
thousand. At what rate will the carbon monoxide level be changing with respect to
time 3 years from now?
Solution
dC dC dp
^(0.5p2 + 17)~1/2 (0.5)(2 p) [0.2f]
dt dp dt
dC
^(0.5-42 + I7)-1/2(4) [0.2(3)] = 0.24
dt t=3
2
The carbon monoxide level will be changing at the rate of 0.24 parts per million. It
will be increasing because the sign of dC/dt is positive. ■
To get a better feel for why the chain rule is true, suppose x is changed by a small
amount Ax. This will cause u to change by an amount An, which, in turn, will cause
y to change by an amount Ay. If An is not zero, we can write
Ay Ay An
Ax An Ax
By letting Ax -> 0, we force An to approach zero as well, since
An =
(£)*■ so lim An = (—
Ax-+Q \dx
(0)=0
It follows that
Historical Note
Mm «_ lim lim —)
A calculus book written by A*—>0 Ax vAw—>o An / \Ax-*o Ax /
the famous mathematician or, equivalently,
G. H. Hardy (1877-1947) con¬ dy dy du
tained essentially this “incorrect”
proof rather than the one given dx du dx
in Appendix B. It is even more Unfortunately, there is a flaw in this “proof” of the chain rule. At the beginning
remarkable that the error was not we assumed that An 0. However, it is theoretically possible for a small change in
noticed until the fourth edition. x to produce no change in n so that An = 0. This is the case that we consider in the
proof in Appendix B.
3.5 The Chain Rule 143
3. y = ir + 1; u = 3x — 2
4. y = 2u2 — u + 5\ u — l — x2
5. y = —; u -9
6. y = cosn; u — x2 + 7
6
7. y — u tanw; u — 3x 4— Figure 3.29 Find the slope of y = flgix)].
x
8. y = u2; u =\nx a. Find the approximate value of u at x = 2. What is the
slope of the tangent line at that point?
Differentiate each function in Problems 9-12 with respect to the
b. Find the approximate value of y at x = 5. What is the
given variable of the function.
slope of the tangent line at that point?
9. a. giu) = w3 b. uix) = x2 + 1
c. Find the slope of y = f[gix)] at x = 2.
c. f{x) = (x2 + l)3
48. The graphs of y = fix) and y = gix) are shown in Fig¬
10. a. g(u) = u5 b. uix) = 3x — 1 ure 3.30.
c. f(x) = (3x - l)5
11. a. giu) = u1 b. uix) = 5 — 8x — 12.x2
c. fix) = (5 — 8x — 12x2)7
12. a. giu) = u 15 b. uix) — 3x2 + 5x — 7
c. fix) = (3x2 + 5x - 7)15
In Problems 13-40, find the derivative of the given function.
'5
13. fix) = (5x — 2y 14. /(*) = (X4 _ 7jc)15
15. fix) = (3x2 -2x4 l)4 16. /(x) = (3 -x2 -x4)1 a-y =/(x) b. y = gix)
17. siO) = sin 146 + 2) 18. ci6) — cos(5 — 36)
Figure 3.30 Problem 48
19. fix) = e~x2+3x 20. y = e*3-"
Let hix) = f[gix)].
21. y = esecx 22. gix) = esinx
a. Estimate h\—l) b. h\ 1) c. h'i3)
23. fit) = exp(t2 + t + 5) 24. g(r) = + (In r)2
49. Repeat Problem 48 for hix) = g[ fix)].
tan 3x
25. gix) = x sin 5.x 26. hix) = —-— 50. If gix) = f[f (x)], use the table to find the value of g'i2).
38. gix) - ln(lnx) gix) 0.8 2.0 2.5 1.8 0.9 0.4
37. /(x) = j/x + V2x
39. fix) = ln(sinx + cosx) 40. T(x) = ln(secx +tanx)
52. Assume that a spherical snowball melts in such a way that its
Find the x-coordinate of each point in Problems 41-46 where the radius decreases at a constant rate (that is, the radius is a linear
graph of the given function has a horizontal tangent line. function of time). Suppose it begins as a sphere with radius 10
42. g(x) = x2(2x + 3)2 cm and takes 2 hours to disappear.
41. fix) = xff\ — 3x
a. What is the rate of change of its volume after 1 hour? (Re¬
(x - l)2
43. qix) = 44. fix) = (2x2 - 7)3 call V — \nr3.)
(x + 2)3
b. At what rate is its surface area changing after 1 hour? (Re¬
In Jx
46. V (x) = —5- call S = 47TA-2.)
45. T(x) =x2e'-3x
144 Chapter 3 Differentiation
53. It is estimated that t years from now, the population of a cer¬ a. How fast is the illuminance on the wall increasing when
tain suburban community is modeled by the formula the person is 19 m from the wall?
b. How far is the person from the wall when the illuminance
pit) -20- is changing at the rate of 1 lux/s?
t+ 1
where p(t) is in thousands of people. A separate environ¬ 58. A lamp of luminous intensity 40 candles is 20 m above the
mental study indicates that the average daily level of carbon floor of a room and is being lowered at the constant rate of
monoxide in the air will be 2 m/s. At what rate will the illuminance at the point on the
floor directly under the lamp be increasing when the lamp is
Lip) = 0.57 p2 + p + 58 15 m above the floor? Your answer should be in lux/s rounded
to the nearest hundredth.
ppm when the population is p thousand. Find the rate at which
the level of carbon monoxide will be changing with respect to 59. To form a simple pendulum, a weight is attached to a rod that
time two years from now. is then suspended by one end in such a way that it can swing
54. At a certain factory, the total cost of manufacturing q units freely in a vertical plane. Let 6 be the angular displacement of
during the daily production run is the rod from the vertical, as shown in Figure 3.31.
q(t) = t2 + lOOt
dollars per pound. At what rate will the weekly demand for 60. A lighthouse is located 2 km directly across the sea from a
the coffee be changing with respect to time 10 weeks from point O on the shoreline, as shown in Figure 3.32.
now? Will the demand be increasing or decreasing?
56. When electric blenders are sold for p dollars apiece, local con¬
sumers will buy
P
blenders per month. It is estimated that t months from now,
the price of the blenders will be
p(t) = 0.04t3/2 + 15
61. The average Fahrenheit temperature t hours after midnight in where /r is a constant (the coefficient of friction). For the case
a certain city is modeled by the formula dF
where W = 100 lb and ji — 0.25, find —- when 9 — 21°.
(nt 5\
T(t) = 58 -t- 17 sin -j
d . du
— cos u = — sin u —
dx dx
-f vT^ = -fa
ax dx
- *2>1/2 =
2
a - xY'f\-2x) \ —x
Vl — x2
To obtain the derivative of the same function in its implicit form, we simply differenti¬
ate across the equation x2 + y2 = 1, remembering that y is a function of x:
d
(x2 + y2) Derivative of both sides
dx
dy
2x + 2 y— 0 Chain rule for the derivative of y
dx
dy dy
Solve for —.
dx dx
The procedure we have just illustrated is called implicit differentiation. Our il¬
lustrative example was simple, but consider a differentiable function y = /(x) defined
by the equation
x2y3 — 6 = 5y3 + x
Implicit differentiation tells us that
9 d 'l 'id*} d d 'i d
x2—(y3) + y3 (x2) - —(6)
— = 5 —(y3) + — (X)
dx dx dx dx dx
, dy
3y + y3(2x)-0 = 5(V|^ + l
dx
dy 1 — 2xy3
dx 3x2y2 — 15y2
You may think that we are not finished since the derivative involves both x and y,
but for many applications, that is enough as we may know both x and y simultaneously
at a point. In this example, we could have found y as an explicit function of x, namely
1/3
y =
(x+6
\x2 - 5
and then found dy /dx by the chain rule.
Now consider a differentiable function y = /(x) defined by the equation
x2y + 2y3 = 3x + 2y
Finding y as an explicit function of x is very difficult in this case, and it is not at all hard
to imagine similar situations where solving for y in terms of x would be impossible or
at least not worth the effort. We begin our work with implicit differentiation by using
it to find the slope of a tangent line at a point on a circle.
3.6 Implicit Differentiation 147
Solution
We recognize the graph of jc2 + y2 10 as not being the graph of a function. However,
—
if we look at a small neighborhood around the point ( — 1,3), as shown in Figure 3.36,
we see that this part of the graph does pass the vertical line test for functions. Thus,
the required slope can be found by evaluating the derivative dy/dx at (— 1,3). Instead
of solving for y and finding the derivative, we take the derivative of both sides of the
equation:
x2 + y2 10 Given equation
d
Figure 3.36 Graph of x2 + y2 = 10
(x2 + y2) (10) Derivative of both sides
dx dx
showing a neighborhood about dy
/»(-!, 3) 2x + 2y —— = 0 Do not forget that y is a function of x.
dx
dy_ _ _x_ dy
Solve the equation for
dx y dx
x 1
y (-1,3) 3
x2y + 2y3 = 3x + 2y
dy
find —.
dx
Solution
The process is to differentiate both sides of the given equation with respect to x. To
help remember that y is a function of x, replace y by the symbol y(x):
d
— {x2y(x) + 2[y(x)f} = — {3x + 2y(x)}
dx
2dy 7 dy „dy
x2—— + 6y2— - 2— 3 — 2 xy
dx ' dx dx
dy 3 — 2xy
dx x2 + 6y2 — 2
Notice that the formula for dy/dx contains both the independent variable x and the
dependent variable y. This is usual when derivatives are computed implicitly. ■
sin(x2 + y) = y2(3x + 1)
Solution
There is no obvious way to solve the given equation explicitly for y. Differentiate
implicitly to obtain
3.6 Implicit Differentiation 149
dy
cos(x2 + y) ^2x + = y2(3) + (3x + 1) ^2y
dx
dy
Finally, solve for —:
dx
2 , ^ dy _ 2 , „ , „dy
2x cos(x2 + y) + cos(x + y)— = 3y + 2y(3x + 1)+-
dx dx
dy 3y2 — 2x cos(x2 + y)
dx cos(x2 + y) — 2y(3x + 1)
Solution
■ dy
The slope of a curve y = f(x) is —, which we find implicitly.
dx
x2 + y2 — 5x + 4y
d , , d
— U + y ) = — (5x + 4y)
dx dx
dy A dy
2x + 2 y— = 5+4—
dx dx
4^=5-2,
(2y-4)^ =5-2*
dx
dy 5 — 2x
dx 2y — 4
dy_ 5 — 2x 5 - 2(5) 5
dx (5.4) 2y -4 (5,4) 2(4) - 4 4
Note that the expression is undefined at y = 2; this makes sense when you see that
Figure 3.37 Graph of circle and
the tangent line is vertical there. Look at the graph in Figure 3.37 and see whether you
tangent line
should exclude any other values.
Quotient rule
Note that the expression for the second derivative contains the first derivative dy/dx.
To simplify the answer, substitute the algebraic expression previously found for dy/dx:
dy_
-y +x
dx ->+<T. dy x
Substitute — =-,
y ' y2 dx y
-y2 -x2
y3
— (x2 + y2)
-10
Substitute x2 + y2 = 10.
y-
d2y
Thus,
dx2
In Section 3.3, we found the derivative of f(x) = ex (Theorem 3.8). This deriva¬
tive is more easily found using implicit differentiation, as shown in the following ex¬
ample.
~(ex) - ex
dx
Solution
Let v = ex, so that x = In v. Then
d d
— (x) = —— (In v) Derivative of both sides of x = In v
dx dx
1 dv
1 = —— Implicit differentiation
v dx
dv
dx
x dv d
e = — = —(e ) Because v = ex ■
dx dx
Prove that — (xr) = rxr~x holds for all real numbers r if x > 0.
dx
3.6 Implicit Differentiation 151
Solution
In y = r In x Definition of exponent
Implicit differentiation
y dx \x)
dy
Solve for —.
dx
r
— x Substitute y = xr.
r— I
= rx Property of exponents
Next, we will use implicit differentiation to obtain differentiation formulas for the six
inverse trigonometric functions. Note that these derivatives are not inverse trigonomet¬
ric functions or even trigonometric functions, but are instead rational functions or roots
of rational functions. In fact, the usefulness of inverse trigonometric functions in many
areas is related to the simplicity of their derivatives.
1 du —1 du
— (tan 1 u) — (cot 1 u) =
dx 1 + u2 dx dx 1 + u2 dx
1 du —1 du
—(sec 1 u) = —(esc 1 u) —
dx u\ \Ju2 — 1 dx dx u\ \Ju~ — 1 dx
WARNING
Note that the derivative of each inverse trigonometric function y = cos 1 x,
y — cot 1 x, and y CSC' x is the opposite of the derivative of the corresponding inverse
cofunction y = sin-1 .r, y = tan-1 x, and y = sec-1 x.
Proof We will prove the first formula and leave the others as problems. Let a =
sin-1 x, so x = since. Because the sine function is one-to-one and differentiable on
[—j[/2, n/2], the inverse sine function is also differentiable. To find its derivative, we
proceed implicitly:
152 Chapter 3 Differentiation
Since — 2
< a < f, cos a > 0, so
da 1 1 1 Note the reference
dx cos a triangle in Figure 3.36.
sin2 a Vl — x2
If u is a differentiable function of x, then the chain rule gives
d _, d . i du 1 du
— (sin u) = —(sin u) — = —-
dx dx dx vl — u2 dx
Solution
d
a. Let u — *Jx in the formula for —(tan u):
dx
1 -i -i
(1-0 =
yi - (i -t)2dt y/l ~ (1 -02 V21 ~ t2
c. Let u — e2x:
1
h\x) = — [sec 1 e2x] —- -— d («2y>
dx je2x | y(e2*)2 _ i dx
1
(2e20 =
e2xs/eAx - 1 Ve4x - 1
The derivatives of bu and \ogb u for a differentiable function u = u(x) and a base
b other than e can be obtained using the chain rule and the change-of-base formulas
from Chapter 2. We summarize the results in the following theorem.
WARNING
-£-*“ = (to w-
dx dx dx dx
b“ is a constant to a variable power, d 11 du
not a variable to a constant power. — (log,, u) =---
dx &b In b udx
What This Says The derivatives of bx and logfex are the same as the
derivatives of ex and lnx, respectively, except for a factor of In b that appears
as a multiplier in the formula
= (In b)bx
dx
and as a divisor in the formula
^0og‘X)=(Ink
Solution
Apply the product rule:
l-x
/'(*) = -^-(x2l-x) = x^-(2l~x) + 2
dx dx
= x(ln2)(21_*)(-l) +2I_*(1) = 21~*(1 -x\n2)
then
d 1 du
— In \u\ = ——
dx u dx
lnx if x > 0
fix) =
ln(—x) if x < 0
so
if x > 0
fix) =
—x x
LOGARITHMIC DIFFERENTIATION
Solution
The procedure called logarithmic differentiation requires that we first take the loga¬
rithm of both sides and then apply properties of logarithms before attempting to take
the derivative.
eIX(2x - If
y
(x3 + 5)2(4 — lx)
e2x(2x - If
In y = In
(x3 + 5)2(4-7x).
dy
Next, differentiate both sides with respect to x and then solve for
dx
1 dy 1 " 1 1
=2 + 6 ——-(2) -2 (3x2) (-7)
y dx [2x - 1 J x3 + 5 -
A — lx
dy_ 12 6x2
=y 2+ +
dx 2x - 1 3+5 4 - 7x
This is the derivative in terms of x and y. If we want the derivative in terms of x alone,
we can substitute the expression for y: .
dy elx(2x — if 12 6x2 7
2+ +
dx (x3 + 5)2(4 — 7x) 2x — 1 x3 + 5 4 — 7x
EXAMPLE 11 Derivative with variables in both the base and the exponent
dy
Find —where y = (x + 1)2+
dx
\2x
Solution y = (x + if
1 dy id } d
(2x) ln(x + 1) Product rule
~dx=1X \Tx [ln(*+1)] + dx
1
= 2x (1) + 2 ln(x + 1)
x + 1
2x
+ 2 ln(x +1)
x + 1
dy 2x 2x
+ 21n(x + 1) (x + 1)
dx x+ 1
3.6 Implicit Differentiation 155
7. - + - l 8. (2x + 3y)2 = 10
y x
x
9. sin(x + y) = x — y 10. tan — = y
y
11. cosxy = 1 — x2 12. e*y + i = x2
13. ln(xy) = e2* 14. exy + In y2 = x
dy
In Problems 15-18, find — in two ways:
dx
a. By implicit differentiation of the equation 43. Find an equation of the normal line to the curve x2+2x v = y2
at (1,-1).
b. By differentiating an explicit formula for y
44. Find an equation of the normal line to the curve
15. x2 + y2 = 12 16. xy + 2y = x2
x2y/y — 2 = y2 — 3x — 5 at (1,3).
17. x 4- 18. xy — x = y + 2 Use implicit differentiation to find the second derivative y" of the
y
functions given in Problems 45-46.
dy .
Find the derivative — in Problems 19-32. 45. lx + 5y2 = l 46. x2 + 2>>3 = 4
dx
19. y = sin-1(2x + 1) 20. y = cos-1(4x + 3) Q 47. Interpretation Problem Compare and contrast the derivatives of
the following functions:
21. y = tan-1 ffx2 + 1 22. y = cot-1 x2
a. y — x2 b. y = 2X c. y - ex d. y = xe
23. y = (sin-1 2x)3 24. y = (tan-1 x2)4
26. y = In |sin-1 x| 48. Interpretation Problem Compare and contrast the derivatives of
25. y = sec-1(e-jr)
the following functions:
27. y = tan-1 28. y = cos-1 (sin x), sinx > 0 a. y = log x b. y = lnx
49. Interpretation Problem Discuss logarithmic differentiation.
29. y = sin-1(cosx), cos* > 0 30. y = ln[sin-l(ev)]
Use logarithmic differentiation in Problems 50-55 to find dy /dx.
31. x sin 1 y + y tan 1 x = x 32. sin 1 y + y = 2xy
You may express your answer in terms of both x and y, and you do
In Problems 33-38, find an equation of the tangent line to the not need to simplify the resulting rational expressions.
graph of each equation at the prescribed point.
50. y = ^/(x10 + l)3(x7 3)8
33. x2 + ;y2 = 13 at (-2, 3)
(2x — l)5
34. x3 + y3 = ;y + 21 at (3, —2) 51. y =
9(x + 3)2
35. sin(x — y) — xy at (0, n)
36. 3X + log2(x^) = 10 at (2, 1) 52. y =
(x2 — 3)2 In s/x^—
V
37. x tan-1 y = x2 + y at (0, 0)
38. sin-1(xy) + | = cos-1 y at (1, 0) e2xl
53. y =
Find the slope of the tangent line to the graph at the points indi¬ (x3 + l)2(4x -7)-
In y/x
cated in Problems 39-42. 54. y=x" 55. y — x
39. bifolium: 40. lemniscate of Bernoulli:
(x2 + y2)2 = ¥ (x2 - y2) 56. Let — H—- = 1, where a and b are nonzero constants. Find
(x2 + y2)2 = 4x2y a2 b2
du dv
a. b.
dv du
57. Show that the tangent line at the point (a, b) on the curve
whose equation is 2x2 + 3xy + y2 = — 2 is horizontal if
4a + 3b = 0. Find two such points on the curve.
58. Find two points on the curve whose equation is
x2 — 3xy + 2y2 = —2, where the tangent line is vertical.
59. Let g be a differentiable function of x that satisfies g{x) < 0
and x2 + g2(x) = 10 for all x.
dg —x
a. Use implicit differentiation to show that — = ——.
dx g(x)
156 Chapter 3 Differentiation
b. Show that g(x) = — VlO — x2 satisfies the given require¬ 64. The tangent line to the curve x2/3 + y2/3 = 8 at the point
ments. Then use the chain rule to verify that (8, 8) and the coordinate axes form a triangle, as shown in
Figure 3.41. What is the area of this triangle?
dg_ _ -x
dx g(x)
60. Find the equation of the tangent line and the normal line to the
curve
x3 + y3 = 2Axy
Give the explicit form of each function, and sketch its graph.
Figure 3.39 Lemniscate (x2 + y2)2 = 4(x2 — y2) 67. Show that the tangent line to the ellipse
68. Find an equation of the tangent line to the hyperbola 71. If 6 is the angle between curve C\ and curve C2 at P and the
2
y2 tangent lines to Ci and C2 at P have slopes m\ and m2, re¬
—r — — = 1 at the point (x0, yo). (See Figure 3.44.)
h- spectively, show that
\m2 - m 11
tan# —
1 + m\m2
72. Find the angle between the circle x2 4- y2 = 1 and the circle
x2 T (y — 1 )2 — 1 at each of the two points of intersection.
73. Journal Problem The Pi Mu Epsilon Journal by Bruce W.
King* : ■ When a professor asked the calculus class to
find the derivative of y2 with respect to x2 for the function
y = x2 — x, one student found
dy y
dx x
x y
Figure 3.44 Hyperbola — — — = 1
Was this answer correct? Suppose
a2 b2
69. Use implicit differentiation to find the second derivative
y", where y is a differentiable function of x that satisfies
x
ax2 + by2 = c (a, b, and c are constants).
70. Show that the sum of the .v-intercept and the y-intercept of any What is the derivative of y2 with respect to x2 for this func¬
tangent line to the curve +/x + ^Jy = C is equal to C2. tion?
The angle between curves Cj and C2 at the point of intersection P 74. Use implicit differentiation to obtain the differentiation for¬
is defined as the angle 0 < 6 < f between the tangent lines at P. mulas for y = tan-1 u and y = sec-1 u.
Specifically, the angle between C i and C2 is the angle between the
75. Use the identity sin-1 u + cos-' u — | to obtain the differ¬
tangent line to C\ at P and the tangent line to C2 at P as shown in
entiation formula for y = cos-1 u. Similar identities can be
Figure 3.45.
used to obtain differentiation formulas for y = cot-1 u and
y = esc-1 u.
Solution
The general situation: Let V denote the volume and r the radius, both of which are
functions of time t (minutes). Since the container is a sphere, its volume is given by
•* A = L (LA
dt dt \3 /
dr
= 47rr2 Do not forget to use the chain rule
dt because r is also a function of time.
dV dr 1
The specific situation: Our goal is to find — at the time when r = 2 ft and —
F dt dt 6 ‘
dV 87T
= 4?r(2)2
dt r=2
T
This means that the volume of the container is increasing at about 8.38 ft3/min when
the radius is 2 ft. ■
Although each related rate problem has its own “personality,” many can be handled
by the following summary:
Step 4. Substitute specific numerical values and solve algebraically for any
required rate. List the known quantities; list as unknown the quantity
you wish to find. Substitute all values into the formula. The only
remaining variable should be the unknown, which may be a variable
or a rate. Solve for the unknown.
Solution
The general situation: (See Student Mathematics Handbook, Problem 37, Problem
Set 1.)
Step 1. Let jc denote the length (in feet) of the person’s shadow, and y, the distance
between the person and the street light, as shown in Figure 3.46. Let t denote
the time (in seconds).
Step 2. Since A ABC and A DEC are similar, we have
x 4- y _ x
20 ~ 6
Step 3. Write this equation as x + y = =§x, or y — and differentiate both sides
with respect to t.
dy 7 dx
dt 3 dt
The specific situation:
Step 4. List the known quantities. We know that dy/dt = 7. Our goal is to find
Figure 3.46 A person walking dx/dt. Substitute and then solve for the unknown value:
away from a streetlamp dy 1 dx
~dt = 3~dt
1 dx
7 =-Substitute.
3 dt
dx 3
3 = — Multiply both sides by y.
Solution
The general situation: Let a: and y be the distances from the base of the wall to
the foot and top of the ladder, respectively, as shown in Figure 3.47. (See Student
Mathematics Handbook, Problem 38, Problem Set 1.)
Notice that ATOB is a right triangle, so a relevant formula is the Pythagorean
theorem:
x1 y2 — 25
Differentiate both sides of this equation with respect to t:
„ dx ^ dy
2x—+2y—=0
dt dt
Figure 3.47 A ladder sliding
down a wall The specific situation: At the particular instant in question, x = 4 and
dx dy
y = V25 - 42 3. We also know that — —2, and the goal is to find — at
dt dt
this instant. We have
2(4)(2) + 2{3)-j- = 0
dt
dy _ 8
~dt ~ ~3
This tells us that, at the instant in question, the bag is descending (since dy/dt is
negative) at the rate of 8/3 » 2.7 m/sec. ■
160 Chapter 3 Differentiation
Solution
The general situation: The required equation was given, so we begin by differentiat¬
ing both sides with respect to t. Remember, because C is a constant, its derivative with
respect to t is zero.
nAdV \AdP
\.4PV0A-b V - = 0 Product rule
dt dt
The specific situation: At the instant in question, P — 20, V — 280, and
dV/dt — —5 (negative because the volume is decreasing). The goal is to find dP/dt.
First substitute to obtain
dP
(20) (1.4) (280)0'4 (—5) + (280)1 4 —;— = 0
dt
Solution
The general situation: Consider a conical tank with height 20 ft and circular base of
radius 5 ft, as shown in Figure 3.48. Suppose that the water level is h ft and that the
radius of the surface of the water is r.
Let V denote the volume of water in the tank after t minutes. We know that
V = \nr2h
3.7 Related Rates and Applications 161
(See Student Mathematics Handbook, Problem 39, Problem Set 1.) Once again, we
use similar triangles (see Figure 3.48) to write ^ = £, or r — We substitute this
into the formula to obtain
Solution
The general situation: Let x denote the horizontal distance between the plane and the
observer, as shown in Figure 3.49. The height of the observer is insignificant when
compared to the height of the plane.
The specific situation: At the instant when x = 3, we are given that — = —400
(negative because the distance is decreasing). Thus, the angle of elevation is changing
at the rate of
d9
--(—■400) — 64 rad/hr
dt 16 + 32 '
The angle of elevation is changing at the rate of 64 radians per hour or, equivalently,
'360deg\/ 1 hr \
(64 rad/hr) 1.02 deg/sec
2tt rad / V 3,600 s
38. A weather balloon is rising vertically at the rate of 10 ft/s. An b. If water enters the trough at the rate of 10 ft3/min, how fast
observer is standing on the ground 500 ft horizontally from is the water level rising (to the nearest ~ in./min) when the
the point where the balloon was released. At what rate is the water is 1 ft deep?
distance between the observer and the balloon changing when 43. At noon, a ship sails due north from a point P at 8 knots (nau¬
the balloon is 400 ft high? tical miles per hour). Another ship, sailing at 12 knots, leaves
39. An observer watches a plane approach at a speed of 500 mi/h the same point 1 h later on a course 60° east of north. How
and an altitude of 3 mi. At what rate is the angle of eleva¬ fast is the distance between the ships increasing at 2 P.M.? At
tion of the observer’s line of sight changing with respect to 5 P.M.? Hint: Use the law of cosines.
time when the horizontal distance between the plane and the
44. A swimming pool is 60 ft long and 25 ft wide. Its depth varies
observer is 4 mi? Give your answer in radians per minute.
uniformly from 3 ft at the shallow end to 15 ft at the deep end,
40. A person 6 ft tall is watching a streetlight 18 ft high while
as shown in Figure 3.56.
walking toward it at a speed of 5 ft/s, as shown in Figure 3.53.
At what rate is the angle of elevation of the person’s line of 60
sight changing with respect to time when the person is 9 ft
* from the base of the light?
Figure 3.54 Problem 41 If water is leaking from the bottom of the bucket at the rate
of 8 in.3/min, at what rate is the water level falling when the
When the beachcomber is 1 mi from the point on the shore
depth of water in the bucket is 6 in.? Hint: The volume of
that is closest to the lighthouse, the searchlight is turning at
the frustum of a cone with height h and base radii r and R is
the rate of 0.25 rev/h. How fast is the beachcomber walking 7th 0
at that moment? Hint: Note that 0.25 rev/h is the same as V = —(R2 + rR + r2).
3
| rad/h.
46. A lighthouse is located 2 km directly across the sea from a
42. A water trough is 2 ft deep and 10 ft long. It has a trapezoidal
point S on the shoreline. A beacon in the lighthouse makes
cross section with base lengths 2 ft and 5 ft, as shown in Fig¬
3 complete revolutions (6n radians) each minute, and during
ure 3.55.
part of each revolution, the light sweeps across the face of a
row of cliffs lining the shore.
a. Show that t minutes after it passes point S, the beam of
light is at a point P located s(t) = 2 tan 6nt km from S.
b. How fast is the beam of light moving at the time it passes
a point on the cliff located 4 km from the lighthouse?
47. Modeling Problem A car is traveling at the rate of 40 ft/s along
a straight, level road that parallels the seashore. A rock with a
Figure 3.55 Water trough family of seals is located 50 yd offshore.
a. Find a relationship between the volume of water in the a. Model the angle 0 between the road and the driver’s line of
trough at any given time and the depth of the water at that sight as a function of the distance x to the point P directly
time. opposite the rock.
3.8 Linear Approximation and Differentials 165
b. As the distance x in Figure 3.58 approaches 0, what hap¬ c. Suppose the car is traveling at v ft/s. Now what happens
pens to d6/dtl to dd/dt as x —»• 0? What effect does this have on a pas¬
senger looking at the seals if the car is traveling at a high
rate of speed?
If f(x) is differentiable at x = a, the tangent line at a point P(a, f(a)) on the graph
of y = f(x) has slope m — f'(a) and equation
4.02
4.01
3.99
3.98
3.97
3.96
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0.1 0.92 0.' 1 1.04
Our observation about tangent lines suggests that if x\ is near a, then f(x\) must
be close to the point on the tangent line to y — f(x) at x = x\. That is,
fix i) « f{a) + f\a)(x i - a)
We refer to this as a linear approximation of /(x) at x = a, and the function
L(x) = f(a) + f'(a)(x - a)
Recall from Section 1.1, that for a line, we used the notation A* for the horizon¬
tal change and Ay for the vertical change. Then, the linear approximation formula
measures the vertical change from point R to point Q even though we really want the
vertical change from the point R to the point P. If the distance A* is small, then these
two vertical distances should be approximately equal. This version of linear approxi¬
mation is sometimes called the incremental approximation formula.
Solution
The approximation formula A/ = /(*o + A*) — /(*0) ~ f'(xo)Ax implies that
A/ _ fixo + A*) - /(*q)
fix o)
A* A*
Because /(*) = sin*, fix) = cos*, and
sin(* -f A*) — sin*
-~ cos*
A*
Figure 3.61 shows the graphs for three different choices of A*. Notice as A* becomes
smaller, it is more difficult to see the difference between / and g; in fact, for very small
A*, the graphs are virtually indistinguishable.
THE DIFFERENTIAL
We have already observed that writing the derivative of /(x) in the Leibniz notation
clf/dx suggests that the derivative may be incorrectly regarded as a quotient of “df”
by “dx.” It is a tribute to the genius of Leibniz that this erroneous interpretation of his
notation often turns out to make good sense.
To give dx and dy meaning as separate quantities, let x be fixed and define dx to
be an independent variable equal to Ax, the change in x. That is, define dx, called the
differential of x, to be an independent variable equal to Ax, the change in x. Then, if
/ is differentiable at x, we define dy, called the differential of y, by the formula
If we relate differentials to Figure 3.62, we see that dx = Ax and that Ay is the rise of
/ that occurs for a change of Ax, whereas dy is the rise of a tangent line relative to the
same change in x (Ay and dy are not the same thing). This is shown in Figure 3.62.
We now restate the standard rules and formulas for differentiation in terms of dif¬
ferentials. Remember that a and b are constants, while / and g are functions.
Differential rules
Trigonometric rules
d(ex) = ex dx d(lnx) = - dx
Differential rules (Continued)
dx —dx
d(sec~] x) = d(csc-1 x)
1*1 Vx2 - 1 1*1 V*2 — l
Note that each of the above differential formulas resembles the earlier derivative
formula, and they can be remembered by “multiplying through” the corresponding
derivative formula by “dx.”
Solution
You can work with the differential, as follows:
d(x2 sinx) = x2 d(sinx) + sinx d(x2)
= x (cosx dx) + sin x (2x dx)
= (x2 cosx + 2x sinx) dx
Or, you can work with the derivative, as shown:
W n
Solution
Let /(x) = i = x_1, so f'(x) — —x-2. We know
ERROR PROPAGATION
In the next example, the approximation formula is used to study propagation of er¬
ror, which is the term used to describe error that accumulates from other errors in an
approximation. In particular, in the next example, the derivative is used to estimate the
maximum error in a calculation that is based on figures obtained by imperfect mea¬
surement.
Solution
The volume of the cube is FQc) = a;3, where a: is the length of a side. If you take
the length of a side to be 10 when it is really 10 ± Ax, your error is A a:; and your
corresponding error when computing the volume will be A V, given by
AV = V(10± A*) - V(10) « V'(10)Ajc
Now, V\x) — 3jc2, so V'(10) = 300. Also, your measurement of the side can be
off by as much as 2%—that is, by as much as 0.02(10) = 0.2 cm in either direction.
Substituting Aa; = ±0.2 in the incremental approximation formula for A V, we get
AV « 3(10)2(±0.2) = ±60
Thus, the propagated error in computing the volume is approximately ±60 cm3. Hence
the maximum error in your measurement of the side is | Ax\ —0.2 and the correspond¬
ing maximum error in your calculation for the volume is
|AV| % V'(10) \Ax\ = 300(0.2) = 60
This says that, at worst, your calculation of the volume as 1,000 cm3 is off by 60 cm3,
or 6% of the calculated volume, when your maximum error in measuring the side is
2%. ■
Error Propagation If A' represents the measured value of a variable and x + Ax represents the exact
value, then Ax is the error in measurement. The difference between f(x + Ax)
and / (a) is called the propagated error at x and is defined by
Relative Error
Af df
The relative error is —— ^
f '
Percentage Error
Solution
Figure 3.63 shows the container. We have
Height
Lateral side
S=6nr2
where r is the radius of the cylinder’s base. Then the approximate propagated error is
Figure 3.63 Surface area of AS « S'(r)Ar = \2nrAr = 12tt(17.3)(±0.02) « ±13.0438927
a right circular cylinder
Thus, the maximum error in the measurement of the surface area is about 13.04 cm2.
Is this a large or a small error? The relative error is found by computing the ratio
AS \2nrAr
= 2r~lAr 2(17.3)-'(±0.02) « ±0.0023121
~Y 6nr2
This tells us that the maximum error of approximately 13.04 is fairly small relative to
the surface area S. The corresponding percentage error is found by
and for this reason, we will compute marginal cost by the derivative C'(x).
Similarly, if R(x) is the revenue obtained from producing x units of a commodity,
then the marginal revenue, MR(x), is the additional revenue obtained from producing
one more unit, and we compute MR by the derivative R'(x). To summarize:
where x is the number of items produced (0 < x < 50). The function p is called the
demand function which is the market price per unit when x units are produced.
b. Use marginal cost to estimate the cost of producing the 9th unit. What is the actual
cost of producing the 9th unit?
c. Use marginal revenue to estimate the revenue derived from producing the 9th unit.
What is the actual revenue derived from producing the 9th unit?
Solution
You can compare the values on [0, 50] using a calculator in order to get a sense of the
No. Total cost; Avg. cost Price per item
meaning of the problem.
0 $98.00 $25.00
5 $116.13; $23.23 $23.33
a. The marginal cost is
10 $140.50; $14.05 $21.67
C'(x) = +3
15 $171.13; $11.41 $20.00
20 $208.00; $10.40 $18.33 To find the marginal revenue, we must first find the revenue function:
25 $251.13; $10.05 $16.67
R(x) = xp(x) = * (|) (75 — x) — -|x2 + 25x
30 $300.50; $10.02 $15.00
35 $356.13; $10.18 $13.33 Thus, the marginal revenue is
40 $418.00; $10.45 $11.67
R'(x) = — jx + 25
45 $486.13; $10.80 $10.00
50 $560.50; $11.21 $8.33 b. The cost of producing the 9th unit is the change in cost as x increases from 8 to 9
and is estimated by
C'(8) = |(8) + 3 = 5
We estimate the cost of producing the 9th unit to be $5. The actual cost is
WARNING Remember that the
marginal revenue from the sale of the AC = C(9) - C(8) = [|(9)2 + 3(9) + 98] - [|(8)2 + 3(8) + 98]
9th item is not the revenue derived from
selling 9 items. Rather, it is the
— 5^ = 5.125 (that is, $5.13)
additional revenue the company has c. The revenue (to the nearest cent) obtained from the sale of the 9th unit is approxi¬
earned by selling the 9th item—that is,
mated by the marginal revenue:
the total revenue of 9 items minus the
total revenue of 8 items. R'(8) = -|(8) + 25 = f « 19.67 (that is, $19.67)
The actual revenue (to the nearest cent) obtained from the sale of the 9th unit is
y.
y-- --mf
2
p *0>
—
_ 1
[*t> nxi))
r 1/
1
_
*1 xq *!
7
/*(^/i) /*/ / \ /
Xn+l = xn "777 r / (%n) f 0
f (*n)
Either this sequence of approximations will approach a limit that is a root of the equa¬
tion or else the sequence does not have a limit.
Proof Rather than present a formal proof, we will present a geometric description
of the procedure to help you understand what is happening. Let xo be an initial ap¬
proximation such that f'(xo) 7^ 0. To find a formula for the improved approximation
Choose a tolerance level X\, recall that the slope of the tangent line through (xo, /(To)) is the derivative f'(x 0).
e>0 Therefore (see Figure 3.64b),
*
Compute/'(x).
\ f(x0) - 0
/ (x0) = —~ = ——--
, ^ , Ax xo — jci
I
Slope of the tangent line through (x0, /(x0))
Choose an initial estimate
x0 with or, equivalently (by solving the equation for x0,
fV o)*0
fix 0)
I Xi =x0 - fix 0) 7^ 0
fix 0)
Compute the new
estimate If this procedure is repeated using x\ as the initial approximation, an even better ap¬
proximation may often be obtained (see Figure 3.64b). This approximation, X2, is
related to xj as xj was related to xq. That is,
fix 1)
x2 — Xj fix t) ^ 0
fix 1)
If this process produces a limit, it can be continued until the desired degree of accuracy
is obtained. In general, the nth approximation xn is related to the (n — l)st by the
formula
x„+1 is an estimate
□
with the desired accuracy. fjXn-1)
xn — Xn—\ fixn-1) 7^ 0
f'iXn-1)
Figure 3.65 Flowchart for Here is a step-by-step procedure for applying the Newton-Raphson method. A
the Newton-Raphson method flowchart for the method appears in Figure 3.65.
1. Choose a number e > 0 that determines the allowable tolerance for estimated
solutions.
2. Compute fix) and choose a number xo (with f{x 0) f 0) “close” to a solu¬
tion of /(x) = 0 as an initial estimate.
3. Compute a new approximation with the formula
x„+, = xn - f(xn) f 0
f ixn)
4. Repeat step 3 until |x„+i — x„| < e. The estimate x = xn+\ then has the
required accuracy.
Solution
Let /(*) = x3 + x + 1. Our goal is to find a root of the equation fix) — 0. The
derivative of / is fix) = 3* 2 + 1, and so
fix) X'3 + * + 1 2x — 1
A ~~ fix) ~X 3*2 + 1 ~~ 3x2 + 1
fixn) _ 2x] ~ 1
f'ixn) ~ 3*2 + 1
2x3 — 1
X) = —3- = —0.75 You will need a calculator or a spreadsheet
3*o + 1 to help you with these calculations.
?r3 — 1
*■> = -4-~ -0.6860465
3*2 + 1
*3 = 2x\ ~ 1 « -0.6823396
3*1+1
So, to two decimal places, the root seems to be approximately *„ ~ -0.68.
3. di2+x) 4. d(*5 + V*2 + 5) 23. You measure the radius of a circle to be 12 cm and use the
5. d(xcosx) 6. d(* sin 2*) formula A — nr2 to calculate the area. If your measurement
(tan 3* of the radius is accurate to within 3%, approximately how ac¬
8. dixe~2x) curate (to the nearest percent) is your calculation of the area?
7-+T7-
9. d(ln|sin*|) .
10 d(*tan-1*) 24. Exploration Problem Suppose a 12-oz can of Coke® has a
sin height of 4.5 in. If your measurement of the radius has an
.
11 d\ex In*) 12. d accuracy to within 1%, how accurate is your measurement for
volume? Check your answer by examining a Coke can.
x sec*
13. d 14. d(*V*2 — 1) 25. You measure the radius of a sphere to be 6 in. Use the formula
* —3 V = \nr2 to calculate the volume. If your measurement of
* —5 /In yfx" the radius is accurate to within 1%, approximately how accu¬
15. d 16. d I
+4, V * rate (to the nearest percent) is your calculation of the volume?
17. WHAT DOES THIS SAY? What is a differential?
26. It is projected that t years from now the circulation of a local
18. Interpretation Problem Discuss error propagation, including newspaper will be
relative error, and percentage error.
Q Use differentials to approximate the requested values in Prob¬ Cit) = 100r + 400f + 5,000
lems 19-22 and then determine the error as compared to the cal¬
culator value. Estimate the amount by which the circulation will increase
during the next 6 months.
19. VO.99 20- cos(f+ 0.01)
174 Chapter 3 Differentiation
27. An environmental study suggests that t years from now, the 35. Modeling Problem According to Poiseuille’s law, the speed of
average level of carbon monoxide in the air will be blood flowing along the central axis of an artery of radius R is
modeled by the formula S(R) = cR2, where c is a constant?
0(0 = 0.05t1 2 + 0.1/ + 3.4 What percentage error (rounded to the nearest percent) will
you make in the calculation of S(R) from this formula if you
parts per million (ppm). By approximately how much will the
make a 1 % error in the measurement of R?
carbon monoxide level change during the next 6 months?
28. A manufacturer’s total cost (in dollars) is 36. Modeling Problem One of the laws attributed to Poiseuille mod¬
els the volume of a fluid flowing through a small tube in unit
C{q) = O.I43 - 0.5q2 + 500$ + 200 time under fixed pressure by the formula V = kR4, where
k is a positive constant and R is the radius of the tube. This
when the level of production is q units. The current level of formula is used in medicine to determine how wide a clogged
production is 4 units, and the manufacturer is planning to de¬ artery must be opened to restore a healthy flow of blood. Sup¬
crease this to 3.9 units. Estimate how the total cost will change pose the radius of a certain artery is increased by 5%. Approx¬
as a result. imately what effect does this have on the volume of the blood
29*. At a certain factory, the daily output is flowing through the artery?*
E(T) = 4.6e17'3r/(r+237>
Time 1: Stream is focused. 46. Approximate s/2 to four decimal places by using the Newton-
Raphson method.
Time 2: Stream hits foil.
47. Approximate -\/2 to four decimal places by using the
_i I k
Newton-Raphson method.
48. Use the Newton-Raphson method to estimate a root of the
equation cosx = x. You may start with x0 = 1.
Time 3: Stream is scattered after colliding
49. Use the Newton-Raphson method to estimate a root of the
with foil and disperses in the shape of a cone.
equation e~x = x. You may start with x0 = 1.
50. Use the Newton-Raphson method to estimate a root of the
equation
x6-x5+x3 = 3
54. Suppose the plane described in Problem 53 is located so that 60. Can you solve the equation ^[x. = 0 using the Newton-
it divides the sphere in the ratio 1:3. Find an equation for xc, Raphson method with the initial estimate xo = 0.05? Does
and estimate the value of xc by the Newton-Raphson method. it make any difference if we choose another initial estimate
Hint: You may need the result of Problem 52. (other than xo = 0)?
Q 55. Show that if h is sufficiently small, then 61. Interpretation Problem Suppose that when we try to use
a. y/1 + h is approximately equal to 1 + |. the Newton-Raphson method to approximate a solution of
/(x) = 0, we find that f(xn) = 0 but f'{xn) ^ 0 for some
b. is approximately equal to 1 — h. xn. What does this imply about xn+l, xn+2,... ? Explain.
1 +h
56. If h is sufficiently small, find an approximate value for 62. HISTORICAL QUEST Among the peoples of the region
y/A" + h for constant A. between the Tigris and Euphrates rivers (at different times) dur¬
57. Tangent line approximations are useful only if Ax is small. ing the period 2000-600 B.C. were Sumerians, Akkadians,
Illustrate this fact by trying to approximate y/97 by regarding Chaldeans, and Assyrians. Since the middle half of the nine¬
97 as being near 81 (instead of 100). teenth century, archeologists have found well over 50,000 clay
58. Let f(x) = —x4 + x2 + A for constant A. What value of tablets describing these great civilizations. Records show that
A should be chosen that guarantees that if x0 = | is chosen they had highly developed religion, history, science (includ¬
as the initial estimate, the Newton-Raphson method produces ing alchemy, astronomy, botany, chemistry, mathematics, and
*i = -x0, x2 = x0, x3 = -x0,... ? zoology). ■
59. Suppose that when x units of a certain commodity are pro¬ Mesopotamian culture had iterative formulas for comput¬
duced, the total cost is C(x) and the total revenue is R(x). Let ing algebraic quantities such as roots. In particular, they ap¬
P(x) = R(x) — C(x) denote the total profit, and let proximated \/lV by repeatedly applying the formula
C(x)
A(x) = 1 / N\
x Xn+1 — ~ ( Xfi T I for n = 0, 1, 2, 3, ...
be the average cost.
a. Show that P'{x) = 0 when marginal revenue equals a. Apply the Newton-Raphson method to f(x) = x2 — N to
marginal cost. justify this formula.
b. Show that A'{x) = 0 when average cost equals marginal b. Apply the formula to estimate V 1.265 correct to five deci¬
cost. mal places.
Chapter 3 Review 177
Proficiency Examination 32. Use the definition of the derivative to find ^(x — 3x2).
33. Find the equation of the tangent line to the graph of
CONCEPT PROBLEMS
1. What is the slope of a tangent line? How does this compare to y = (x2 + 3x — 2) (7 — 3x)
the slope of a secant line?
at the point where x = 1.
2. Define the derivative of a function.
3. What is a normal line to a graph? 34. Let /(x) = sin2 Find equations of the tangent line and
4. What is the relationship between continuity and differentiabil¬ the normal line to the graph of / at x = 1.
ity? 35. A rock tossed into a stream causes a circular ripple of water
5. List and explain some of the notations for derivative. whose radius increases at a constant rate of 0.5 ft/s. How fast
6. State the following procedural rules for finding derivatives: is the area contained inside of the ripple changing when the
radius is 2 ft?
a. constant multiple b. sum rule
c. difference rule d. linearity rule
Supplementary Problems
e. product rule f. quotient rule
Find dy I dx in Problems 1-36.
g. State each of these rules again, this time in differential
form. 1. y = x4 + 3x2 - lx + 5 2. y = x5 + 3x3 - 11
7. State the following derivative rules:
a. constant rule b. power rule 3. y = ./ —-- 4. y -
yx2-5 x + sin x
c. trigonometric rules d. exponential rule
5. 2x2 — xy + 2y = 5 6. y = (x2 + 3x — 5)7
e. logarithmic rule f. inverse trigonometric rules
7. y = (x3 +x)10 8. y = V^(x2 + 5)10
8. What is a higher-order derivative? List some of the different
9. y = ^x(x3 + l)5 10. y = (x2 + 3)5(x3 — 5)
notations for higher-order derivatives.
11. y = (x4- l)10(2x4 + 3)7 12. y = Vsin 5x
9. What is meant by rate of change? Distinguish between aver¬
age and instantaneous rate of change. 13. v = yjcos sfx 14. y = (sinx -f cosx)3
10. What is relative rate of change? lx3 -x
11. How do you find the velocity and the acceleration for an object 15. y = ifx + ^/x)5
16-W 4-,:
with position 5(7)? What is speed?
17. y = exp(2x2 + 5x — 3) 18. y = ln(x2 - 1)
12. State the chain rule.
19. y = x32~x 20. y = log3(x2 - 1)
13. Outline a procedure for logarithmic differentiation.
14. Outline a procedure for implicit differentiation. 21. <?AV + 2 = In - 22. y = V^siiT'Ox + 2)
x
15. Outline a procedure for solving related rate problems. 23. y = esin* 24. y = 2X log2 x
16. What is meant by tangent line approximation? 25. y = e~x log5 3x 26. x2-v + y2x = 3
17. Define the differential of x and the differential of y for a func¬
27. ln(x + y2) = x2 + 2y 28. y = e~xs/ln 2x
tion y - f{x). Draw a sketch showing Ax, Ay, dx, and dy.
29. y = sin(sinx) 30. y = cos(sinx)
18. Define the terms propagated error, relative error, and percent¬
31. X1/2 + yO2 = x 32. 4x2 — 16y2 = 64
age error.
19. What is meant by marginal analysis? 33. sin xy = y + x
56. Find equations for the tangent and normal lines to the graph 75. A car and a truck leave an intersection at the same time.
The car travels north at 60 mi/h and the truck travels east at
of v = ( 2x 4— j at the point (1,27). 45 mi/h. How fast is the distance between them changing af¬
ter 45 minutes?
dy
57. Use the chain rule to find — when v = x3 — lx and 76. A spherical balloon is being filled with air in such a way that
dt
x = t sin t. its radius is increasing at a constant rate of 2 cm/s. At what
58. Find /"(x) if /(x) = x2 sinx2. rate is the volume of the balloon increasing at the instant when
its surface has area An cm2?
59. Find /(4,(x) if /(x) = x4 77. Suppose the total cost of producing x units of a particular com¬
4 '
83. A processor who sells a certain raw material has analyzed the d2s
89. The equation + ks — 0 is called a differential equation
market and determined that the unit price should be modeled d?
of simple harmonic motion. Let A be any number. Show that
by the formula
the function sit) — A sin 21 satisfies the equation
p(x) = 60 — x2 d2s
+ 4s = 0
d?
(thousand dollars) for x tons (0 < x < 7) produced. Estimate
the change in the unit price that accompanies each change in 90. Suppose L(x) is a function with the property that
sales: L'(x) = x_1. Use the chain rule to find the derivatives of
the following functions.
a. from 2 tons to 2.05 tons
b. from 1 ton to 1.1 tons a. f(x) = L(x2) b. fix) = L
c. from 3 tons to 2.95 tons
2x + 1
84. A company sends out a truck to deliver its products. To esti¬ c. fix) = L d. fix) = L
Sy/X 1 -x
mate costs, the manager models gas consumption by the for¬
91. Let / and g be differentiable functions such that /(g(x)] = x
mula and g[/(x)] = x for all x (that is, / and g are inverses). Show
that
CW = 300
The rod AB is attached to O A at A, and the end B slides along Show that |/'(x)| < 4 for |x| < 1.
the x-axis. Suppose AB is 5 meters long. What are the veloc¬ 96. Putnam Examination Problem A point P is taken on the curve
ity and the acceleration of the motion of the point B along the y = x3. The tangent line at P meets the curve again at Q.
x-axis? Prove that the slope of the curve at Q is four times the slope
94. The ideal gas law states that for an ideal gas, pressure P is at P.
given by the formula P — kT/V, where V is the volume, T is 97. Putnam Examination Problem A particle of unit mass moves on a
the temperature, and k is a constant. Suppose the temperature straight line under the action of a force that is a function f(v)
is kept fixed at 100°C, and the pressure decreases at the rate of the velocity v of the particle, but the form of this func¬
of 7 lb/in.2 per min. At what rate is the volume changing at tion is not known. A motion is observed, and the distance x
the instant the pressure is 25 lb/in.2 and the volume is 30 in.3? covered in time t is found to be related to t by the formula
x = at + bt2 + ct3, where a, b, and c have numerical values
95. Putnam Examination Problem Suppose /(x) = ax2 + bx + c, determined by observation of the motion. Find the function
where a, b, and c are real numbers and |/(x)| < 1 for x < 1. f(v) for the range of v covered by this experiment.
Group Research Project
Chaos
A n exciting new topic in mathematics attempts to bring order to the universe by con¬
sidering disorder. This topic is called chaos theory, and it shows how structures
of incredible complexity and disorder really exhibit beauty and order.
Water flowing through a pipe offers one of the simplest physical models of chaos.
Pressure is applied to the end of the pipe and the water flows in straight lines. More
pressure increases the speed of the laminar flow until the pressure reaches a critical
value, and a radically new situation evolves—turbulence. A simple laminar flow sud¬
denly changes to a flow of beautiful complexity consisting of swirls within swirls. Be¬
fore turbulence, the path of any particle was quite predictable. After a minute change
in the pressure, turbulence occurs and predictability is lost. Chaos is concerned with
systems in which minute changes suddenly transform predictability into unpredictabil¬
ity.*
For this research project begin with
Use the Newton-Raphson method to investigate what happens as you change the start¬
ing value, x0, to solve /(x) = 0. What would you select as a starting value? Two
This fractal image is an important x-values in our study are
example of what is called
mathematical chaos. and
One would not want to pick xq as either ±53. Why not? Also, what happens if one
picks xq = y5? That is, what are x\, x2,... ?
Generate a good plot of /(x) on [-2, 2], Explain from the plots why you would
Mathematics is one of the oldest of expect that an initial value xo > s3 would lead to lim xn — 1. (Also, by symmetry,
the sciences; it is also one of the most ac¬ , , , , n-*+oo
xo < —S3 leads to lim xn = — 1.)
tive, for its strength is the vigor of perpet¬ n-++oo
ual youth. Explain why if |x0| < s5, you would expect x„ -» 0. Numerically verify your
-A. R. Forsythe assertions. Now to see the “chaos,” use the following x0 values and take 6 to 10 it¬
Nature 84 (1910): 285. erations, until convergence occurs, and report what happens: 0.448955; 0.447503;
0.447262; 0.447222; 0.4472215; 0.4472213. Write a paper telling what you see and
why.
Extended paper for further study Attempt to define chaos. Your paper should in¬
The whole of mathematics clude, but not be limited to, answers to the questions on this page. Some references (to
consists in the organization of a series of
get you started) are listed:
aids to the imagination in the process of
Chaos: Making a New Science, James Gleick, Penguin Books, 1988.
reasoning.
-A. N. Whitehead Chaos and Fractals: New Frontiers of Science, H. O. Peitgen, H. Jurgens, and
Universal Algebra D. Saupe, Springer-Verlag, 1992.
(Cambridge, 1898), p. 12.
Newton’s Method and Fractal Patterns, Phillip D. Straffin, Jr., UMAP Module 716,
COMAP, 1991.
'
CONTENTS
4.1 Extreme Values of a
Continuous Function
Extreme value theorem
Relative extrema
Additional Applications
Absolute extrema
4.2
Optimization
Chapter 4 Review
Group Research Project: Wine Barrel *Edward Batschelet, Introduction to Mathematics for Life Scientists, 2nd ed. (New York: Springer-
Capacity Verlag, 1979), pp. 276-277.
+See Problem 32, Section 4.7.
183
184 Chapter 4 Additional Applications of the Derivative
One of the principal goals of calculus is to investigate the behavior of various functions.
As part of this investigation, we will be laying the groundwork for solving a large class
of problems that involve finding the maximum or minimum value of a function, if one
exists. Such problems are called optimization problems. We begin by introducing
some useful terminology.
Absolute Maximum Let / be a function defined on an interval I that contains the number c. Then
and Minimum
/(c) is an absolute maximum of / on / if f(c) > f(x) for all x in /
/(c) is an absolute minimum of / on 7 if /(c) < f(x) for all x in /
Sometimes we just use the terms maximum and/or minimum if the context is clear.
Together, the absolute maximum and minimum of / on the interval I are called the
extreme values, or the absolute extrema, of / on I. A function does not neces¬
sarily have extreme values on a given interval. For instance, the continuous function
g(x) = t has neither a maximum nor a minimum on the open interval (0, 1), as shown
in Figure 4.1a.
The discontinuous function defined by
for * / 0
h(x) =
for x — 0
has a maximum which is on the closed interval [—1, 1], but no minimum, as shown
in Figure 4.1b. Incidentally, this graph also illustrates the fact that a function may
assume an absolute extremum at more than one point. In this case, the maximum
occurs at the points (—1, 1), (0, 1), and (1, 1). If a function / is continuous and the
interval / is closed and bounded, it can be shown that both an absolute maximum and
an absolute minimum must occur. This result, called the extreme value theorem, plays
an important role in our work.
Proof Even though this result may seem quite reasonable (see Figure 4.2), its proof
requires concepts beyond the scope of this text and will be omitted.
If / is discontinuous or the interval is not both closed and bounded, you cannot
conclude that / has both a largest and smallest value. You will be asked for appropriate
counterexamples in the problem set. Sometimes there are extreme values even when
the conditions of the theorem are not satisfied, but if the conditions hold, the extreme
values are guaranteed.
Note that the maximum of a function occurs at the highest point on its graph and
the minimum occurs at the lowest point. These properties are illustrated in Example 1.
Solution
The highest point on the graph occurs at the right endpoint F, and the lowest point
occurs at C. Thus, the absolute maximum is /(b), and the absolute minimum is /(c2).
186 Chapter 4 Additional Applications of the Derivative
2x if 0 < jc < 1
a. f(x) = b. g(x) = x2 on 0 < x < 2
1 if 1 < x < 2
Does not have a maximum value. g does not have a minimum value
(but it does have a maximum value).
Solution
a. The function / has no maximum. It takes on all values less than, but arbitrarily
close to, 2. However, it never reaches the value 2. This function does not contradict
the extreme value theorem because / is not continuous on [0, 2].
b. Although the functional values of g(x) become arbitrarily small as x approaches 0,
g{x) never reaches the value 0 so g has no minimum. The function g is continuous
on the interval (0, 2], but the extreme value theorem is not contradicted because the
interval is not closed. ■
RELATIVE EXTREMA
Typically, the extrema of a continuous function occur either at endpoints of the interval
or at points where the graph has a “peak” or a “valley” (points where the graph is higher
or lower than all nearby points). For example, the function / in Figure 4.3 has “peaks”
at B, D, and “valleys” at C, E. Peaks and valleys are what we call relative extrema.
Relative Maximum The function / is said to have a relative maximum at the point c if
and Relative Minimum /(c) > /(x) for all x in an open interval containing c. Likewise, / is said
to have a relative minimum at d if f(d) < /(x) for all x in an open inter¬
val containing d. Collectively, relative maxima and relative minima are called
WARNING , relative extrema.
Note that an extremum at
an endpoint is, by definition, not a
relative extremum, since there is not an
open interval around the endpoint
entirely contained in the domain of the Next we will formulate a procedure for finding relative extrema. By looking at
function. Figure 4.3, we see that there are horizontal tangents at B, C, and E, while no tangent
can be defined at the point D. This suggests that the relative extrema of / occur either
where the derivative is zero (horizontal tangent) or where the derivative does not exist
(no tangent). This notion leads us to the following definition.
4.1 Extreme Values of a Continuous Function 187
Critical Numbers and Suppose / is defined at c and either /'(c) = 0 or /'(c) does not exist. Then
Critical Points the number c is called a critical number of /, and the point P(c, /(c)) on the
graph of / is called a critical point.
WARNING . . . ......
Note that if / (c) is not
defined, then c cannot be a critical
number.
EXAMPLE 3 Finding critical numbers
Find the critical numbers for the given functions.
Solution
a. fix) — \2x2 — lOx — 8 is defined for all values of x. Solve
12x2 - lOx - 8 = 0
2(3x — 4)(2x + 1) = 0
4 1
x — These are the critical numbers.
3’ 2
c. Write fix) — 12x1/2 - 2x3/2 so fix) = 6x~1/2 - 3x1/2. The derivative is not
defined at x = 0. We have /(0) = 12(0)1/2 - 2(0)3/2 = 0, so we see that / is
defined at x = 0, which means that x = 0 is a critical number. For other critical
numbers, solve fix) = 0:
bx-1/2- 3x^ = 0
3x-1/2(2 — x) = 0
x = 2
/ (x) = (x — l)2(x + 2)
Solution
Because the function / is a polynomial, we know that it is continuous and that its
derivative exists for all x. Thus, we find the critical numbers by using the product rule
and extended power rule to solve the equation fix) — 0:
The critical numbers are x = ±1. To find the critical points, we need to find the
^-coordinate for each critical number.
Thus, the critical points are (1, 0) and (—1,4). The graph of /(x) = (x — l)2(x + 2)
is shown in Figure 4.4. ■
Note how the relative extrema occur at the critical points. Our observation that
the relative extrema occur only at points on a graph where there is either a horizontal
Figure 4.4 The graph of tangent line or no tangent at all is equivalent to the following result.
/(■*) = (x - l)2(x +2)
Proof Since / has a relative extremum at c, f(c) is defined. If /'(c) does not exist,
then c is a critical number by definition. We will show that if /'(c) exists and a relative
maximum occurs at c, then /'(c) = 0. Our approach will be to examine the difference
quotient. (The case where /'(c) exists and a relative minimum occurs at c is handled
similarly in Problem 61.)
Because a relative maximum occurs at c, we have /(c) > /(x) for every number
x in an open interval (a, b) containing c. Therefore, if Ax not equal to zero is small
enough so c + Ax is in (a, b), then
For the next step we want to divide both sides by Ax (to write the left side as a differ¬
ence quotient). However, as this is an inequality, we need to consider two possibilities:
Now we take the limit of both sides as Ax approaches 0 from the right (because Ax
is positive).
f\c) 0
f\c) < 0
4.1 Extreme Values of a Continuous Function 189
This time we take the limit of both sides as Ax approaches 0 from the left (because
Ax is negative).
f\c) > 0
Because we have shown that /'(c) < 0 and /'(c) > 0, it follows that /'(c) =0. □
Figure 4.5 A relative extremum may not occur at each critical number.
Solution
Figure 4.6 Graph of
/(x) = |x + 1| on [—5, 5] Ifx > — 1, then /(x) =x+l and /'(x) = 1. However, ifx < -l,then/(x) = -x-1
and /'(x) = — 1. We need to check what happens at x = — 1.
/(—1 + Ax + 1) — /(—1 + 1)
/'(-i)= lim
Ax-*-0 Ax
|A*1-|0|
= lim
Ax->0 Ax
I Ax |
= lim
Ax—>-0 Ax
190 Chapter 4 Additional Applications of the Derivative
and
| Ax — Ax
lim - lim - = -1
Ax—>0“ Ax Ax—Ax
Because these are not the same, this limit does not exist and thus the function /(x) =
|x + 11 is not differentiable at x = —1. Because /(—1) is defined, it follows that — 1
is the only critical number. ■
ABSOLUTE EXTREMA
Suppose we are looking for the absolute extrema of a continuous function / on the
closed, bounded interval [a,b]. Theorem 4.1 tells us that these extrema exist and
Theorem 4.2 enables us to narrow the list of “candidates” for points where extrema
can occur from the entire interval [a, b] to just the endpoints x — a, x = b, and the
critical numbers between a and b. This suggests the following procedure.
;
Step 1. Compute f'(x) and find all critical numbers of / on [a, b].
Step 2. Evaluate / at the endpoints a and b and at each critical number c.
Step 3. Compare the values in step 2.
The largest value of / is the absolute maximum of f on [a, b].
The smallest value of / is the absolute minimum of f on [a, b].
Figure 4.7 shows some of the possibilities in the application of this procedure.
Solution
Because / is a polynomial function, it is continuous on the closed interval [—1, 2].
Theorem 4.1 tells us that there must be an absolute maximum and an absolute minimum
on the interval.
4.1 Extreme Values of a Continuous Function 191
Solution
Step I. To find the derivative, rewrite the given function as /(a) = 5a2/3 — 2a5/3.
Then
/'(A) = - $ A2/3 = f A-‘/3(l - A)
Critical numbers are found by solving /'(a) = 0 and by locating the places
where the derivative does not exist. First,
/'(a) = 0 when a = 1
Even though /(0) exists, we note that /'(a) does not exist at a = 0 (notice
the division by zero when a = 0). Thus, the critical numbers are a = 0 and
a = 1.
Solution
Step 1. To find where T'{x) — 0, we begin by finding T'{a):
Since the factor (sin a 4-2) is never zero, it follows that T'{x) = 0 only when
2 cos a - 1 = 0; that is, when x = j. This is the only critical number in
[0, f ].
192 Chapter 4 Additional Applications of the Derivative
= i (I + 5) + 2 (#) - f = l + 1-309853256
OPTIMIZATION
In our next two examples, we examine applications involving optimization. Such prob¬
lems are investigated in more depth in Sections 4.6 and 4.7. 1
Solution
We let b be the length of one side of the base and h be the height of the box, as shown
in Figure 4.10. The volume, V, is
V = b2h
Because our methods apply only to functions of one variable, it may seem that we
cannot deal with V as a function of two variables. However, we know that b + h = 10;
therefore, h = 10 — b, and we can now write V as a function of b alone:
V(b) = b2(\0 - b)
The domain is not stated, but for physical reasons we must have b > 0 and
Figure 4.10 Volume of a box 10 — b — h > 0, so that 0 < b < 10. Figure 4.10 shows the volume for various
choices of b.
Now, find the value of b that produces the maximum volume. First, we find the
critical numbers. Note that V is a polynomial function, so the derivative exists every¬
where in the domain. Write V (b) — 10b2 — b3 and find V'ib) — 20b — 3b2. Then
V\b) = 0
20b -3b2 = 0
b{20 - 3b) = 0
b = 0, f
V(0) =0
Solution
The velocity is
v(t) = s\t) = 4r3 - 24r2 + 361 + 60
(Note: v is a polynomial function, so it has a derivative for all t.) To find the largest
value of v{t), we compute the derivative of v:
v\t) = 12r2 -48? + 36
= 12(r — l)(f — 3)
Setting v'{t) = 0, we find that the critical numbers of v(t) are t — 1,3. Now we
evaluate v at the critical numbers and endpoints.
t = 1 is both a critical number and an endpoint:
u(l) = 4(1)3 - 24(1)2 + 36(1) + 60 = 76
t — 5 is an endpoint:
i,(5) = 4(5)3 - 24(5)2 + 36(5) + 60 = 140
t — 3 is a critical number:
u(3) = 4(3)3 - 24(3)2 + 36(3) + 60 = 60
The largest value of the velocity is 140 at the endpoint where t — 5, and the smallest
value is 60 when t = 3. M
w
5. fix) — x3 on l]
6. g(x) = x3 — 3x on [—2, 2]
7. fix) = x5 — x4 on [—1, 1]
8. git) = 3t5 — 20f3 on [—1,2]
9. hit) — te~' on [0, 2]
In,/* ViBXA<2/3X5-2X)
10. six) = -on [1,3] Xro i n= "10 Yhi i n= "10
x
11. fix) = |x | on [-1, 1]
Xmax=10 Ymax=10
Xscl=l Ysc1=1
12. fix) = |x - 31 on [-4, 4]
13. /(m) = sin2 u + cos u on [0, 2] This graph is not correct. Can you explain the discrepancy?
14. giu) = sinw — cosm on [0, n] It is not our intent to “make up” problems to use with a cal¬
15. WHAT DOES THIS SAY? Outline a procedure for finding culator, but whenever you use a calculator or computer to as¬
the absolute extrema of a continuous function on a closed, sist you with calculus, you must understand the nature of the
bounded interval. Include in your outline a discussion of what functions with which you are working, and not rely only on
the calculator or computer output.
is meant by critical numbers.
194 Chapter 4 Additional Applications of the Derivative
In Problems 18-29, find the absolute maximum and absolute mini¬ d. neither a maximum nor a minimum
mum (largest and smallest values, respectively) of each continuous Note that in each problem, there may be a different solution func¬
function on the closed, bounded interx’al. If the function is not con¬ tion for each side condition.
tinuous on the interval, then the extreme value theorem does not
44. Counterexample Problem For each of the four given conditions,
apply; if this is the case, so state.
find a function that is discontinuous and defined on an open
18. /(«)= 1 -n2/3on[-l, 1] interval.
19. g(t) = (50 + t)2/3 on [-50, 14]
45. Counterexample Problem For each of the four given conditions,
20. g(x) = 2x3 — 3x2 — 36x + 4 on [—4, 4] find a function that is discontinuous and defined on a closed
21. g(x) — x3 + 3x2 - 24x - 4 on [-4, 4] interval.
22. /(x) = |x3 - 5x2 + 8x - 5 on [-4, 4] 46. Counterexample Problem For each of the four given conditions,
find a function that is continuous and defined on an open in¬
23. /(x) = |(x3 — 6x2 + 9x + 1) on [0, 2]
terval.
24. h(x) = tanx 4- secx on [0, 2n]
* 47. Counterexample Problem For each of the four given conditions,
25. s(t) — t cos t — sin / on [0, 27t] find a function that is continuous and defined on a closed in¬
26. /(x) = e~x sinx on [0, 2n\ terval.
31. the smallest value of gix) = —h x — 3 on [1, 9] 51. An object moves along the 5-axis with position
x
x2- 1 5(0 = t4 - 2t3 - I2t2 4- 60r — 10
32. the smallest value of gix) on [-1,1]
x2 + 1
Find the largest value of its velocity on [0. 3].
—t2 — t + 2 if t < 1
33. the largest value of fit) = 52. Find two nonnegative numbers whose sum is 8 and the product
3 —t if t > 1
of whose squares is as large as possible.
on [-2, 3]
53. Find two nonnegative numbers such that the sum of one and
34. the smallest value of fix) ex 4- e~x — x on [0, 2]
twice the other is 12 if it is required that their product be as
35. the largest value of gix) = -on [2, 3] correct to the near- large as possible.
cos x
est tenth. 54. Under the condition that 3x 4- y = 80, maximize xy3 when
Q In Problems 36—43, find the extrema (that is, the absolute maxima x > 0, y > 0.
and minima). 55. Under the condition that 3x 4- y — 126, maximize xy when
36. f!0) — cos3 9 — 4 cos2 9 on [—0.1,7r 4-0.1] x > 0 and y > 0.
37. g!9) = 9 sin 9 on [—2, 2] 56. Under the condition that 2x — 5y = 18, minimize x2y when
38. fix) — 20sin(378;rx) on [— 1, 1] x > 0 and y < 0.
39. giu) = 98n3 - Au2 4- 72m on [0, 4] 57. Show that if a rectangle with fixed perimeter P is to enclose
the greatest area, it must be a square.
40. fiw) = yfwiw - 5)1/3 on [0,4]
58. Find all points on the circle x2 4-y2 = a2 (a > 0) such that the
41. hix) = ffx^/ix - 3)2 on [-1, 4]
product of the x-coordinate and the y-coordinate is as large as
42. hix) — cos-1 x tan-1 x on [0, 1] possible.
43. fix) — e~xicosx 4- sinx) on [0, 27t] Q 59. a. Show that | is the number that is greater than or equal to
In Problems 44—47, find functions that satisfy the stated conditions its own square by the greatest amount.
and each of the following side conditions, if possible: b. Which nonnegative number is greater than or equal to its
a. a minimum but no maximum own cube by the greatest amount?
b. a maximum but no minimum c. Which nonnegative number is greater than or equal to its
c. both a maximum and a minimum nth power (n > 0) by the greatest amount?
4.2 The Mean Value Theorem 195
60. Given the constants a.\, aj,..., an, find the value of x that relative minimum occurs at c, then f'(x) = 0.
guarantees that the sum
62. Explain why the function
S(x) = (a, - x)2 + (a2 - x)2 H-b (an - x)2
8 27
fid) = +
will be as small as possible. sin 9 cos 8
61. Without using Theorem 4.2, show that if /'(c) exists and a must attain a minimum in the open interval (0, |).
s(b) — s(a)
sV)
b —a
Instantaneous velocity
Average velocity
This example illustrates a result called the mean value theorem for derivatives
(abbreviated MVT) that is fundamental in the study of calculus. Our proof of the MVT
is based on the following special case, which is named for the French mathematician,
Michel Rolle (1652-1719), who gave a version of the result in an algebra text published
in 1690.*
ROLLE’S THEOREM
The key to the proof of the mean value theorem is the following result, which is really
just the MVT in the special case where f(a) = f(b). In terms of our car example,
Rolle’s theorem says that if a moving car begins and ends at the same place, then
somewhere during its journey, it must reverse direction, since --^ ^ = 0 for
b —a
f(a) = f(b).
Proof To construct a formal proof, note that since / is continuous on the closed
interval [a, b], it follows from the extreme value theorem (Theorem 4.1) that / must
have both a largest value and a smallest value on [a, b]. The case where / is constant
on [a, b] is easy since then f'{x) = 0 for all x in the interval (see Figure 4.1 la).
‘Incidentally. Rolle was a number theorist at heart and thoroughly distrusted the methods of calculus, which
he regarded as a “collection of ingenious fallacies.”
Chapter 4 Additional Applications of the Derivative
In the case where / is not constant throughout [a,b], the largest and smallest
values cannot be the same, and since /(a) = fib), we can conclude that at least one
extreme value occurs at a number c that is not an endpoint; that is, a < c < b. Finally,
since c is in the open interval (a, b) throughout which f'(x) exists, it follows from the
critical number theorem (Theorem 4.2) that /'(c) = 0. A graph illustrating this case is
shown in Figure 4.1 lb.
y i
fix) = =0
X
/
/ X
/ \
/ \
\ ' /1
/ \x) = (
*
*
_
a. If/is constant on b. If/is not constant on (a, b),
[a, b], then f\x)=0 the graph jnust have at least
throughout (a,b). one horizontal tangent.
Note: Rolle’s theorem asserts only that at least one number c exists between a and b
such that /'(c) = 0. As illustrated in Figure 4.1 lb, there may be more than one such
number.
Rolle’s theorem can be interpreted as saying that there is at least one number c between
a and b such that the tangent line to the graph of / at the point (c, /(c)) is parallel
to the horizontal line through the points (a, f(a)) and (b, fib)). It is reasonable to
expect a similar result to hold if the endpoints of the graph are not necessarily at the
same level of the graph; that is, if the graph in Figure 4.12a is tilted as shown in Figure
4.12b. This “tilted” version can be interpreted as showing that the line segment joining
points (a, f(a)) and (b, fib)) on the graph of a differentiable function y = fix) has
the same slope as the tangent line to the graph at some point c between a and b. This
observation leads to the following statement of the mean value theorem.
Figure 4.12 A geometrical comparison of Rolle’s theorem with the mean value theorem
4.2 The Mean Value Theorem 197
fib) - fja)
gix) = (x - a) + f (a) - fix)
b —a
for a < x < b. Because / satisfies the hypotheses of the MVT, the function g is also
continuous on the closed interval [a, b] and differentiable on the open interval (a, b).
In addition, we find that
'fib) ~ fja)'
g(a) = (a — a) + f (a) - f(a) = 0
b —a
Thus, g satisfies the hypotheses of Rolle’s theorem, so there exists at least one number
c between a and b for which g'(c) = 0. Differentiating the function g, we find that
fib) - fja)
g\x) fix)
b —a
Because g'(c) = 0, we have
f\ „ , fib)-f(a)
o = g (c) = —-—:-/ (c)
a
This means that
fib) ~ fja)
fie)
b —a
as required. □
The MVT asserts the existence of a number, c, but does not tell us how to find it.
The following example illustrates one method for finding such a number c.
fix) = 3x2 + 2x
for all .*. Therefore, we have /'(c) = 3c2 + 2c, and the MVT equation
N /(2)-/(l)
2 - 1
Chapter 4 Additional Applications of the Derivative
is satisfied when
■1 ±V3l
c =
The negative value is not in the open interval (1,2), but the positive value
-1 + V3l
c =--— « 1.522588121
Solution
The inequality is true if x = y. Suppose r/ j; then f(9) = sin0 is differentiable
and hence continuous for all 6, with f'(6) = cos 6. By applying the MVT to / on the
closed interval with endpoints x and y, we see that
/(*) - f(y)
/'(c)
x-y
it follows that
sinx — siny
-= cos c
x-y
Finally, we take the absolute value of the expression on each side, remembering that
|cosc| < 1 for any number c:
sin x — sin y
Icoscl < 1
x-y
The primary use of the MVT is as a tool for proving certain key theoretical results of
calculus. For example, we know that the derivative of a constant is 0, and in Theorem
4.5, we prove a partial converse of this result by showing that a function whose deriva¬
tive is 0 throughout the interval must be constant on that interval. This apparently
simple result and its corollary (Theorem 4.6) turn out to be crucial to our development
of integration in Chapter 5.
Proof Let x\ and x2 be two distinct numbers (xi x2) chosen arbitrarily from the
closed interval [a, b]. The function / satisfies the requirements of the MVT on the
interval with endpoints xi and x2, which means that there exists a number c between
x\ and X2 such that
fix2) ~ fixi)
f'ic)
X2 ~ X\
By hypothesis, fix) = 0 throughout the open interval (a, b), and because c lies within
this interval, we have /'(c) = 0. Thus, by substitution, we have
fix 2) - fix i) _ Q
X2 ~ X\
fixi) = fix l)
Because x\ and x2 were chosen arbitrarily from [a, b], we conclude that fix) = k, a
constant, for all x, as required. -I
fix) = gix) + C
Thus, by Theorem 4.5, hix) = C for some constant C and all x on [a, b], and because
hix) = fix) - gix), it follows that
fix) - gix) = c
fix) = g(x) + C Add g(x) to both sides. -I
^ What This Says Two functions with equal derivatives on an open interval
differ by a constant on that interval.
13. fix) = cosx on [0, |] 37. Let g be defined as shown in Figure 4.14.
14. fix) = sinx + cos x on [0, 27t]
15. f(x) = ex on [0, 1]
16. f(x) = \(ex + e~x) on [0. 1]
In Jx
17. fix) — lnx on [^, 2] 18. f(x) = —— on [1,3]
24. f{x) = |jc|-2 on [0,4] Prove this alternative form of the MVT.
39. Let u and v be any two numbers between — | and |. Use the
25. f(x) = f/x — 1 on [—8, 8] 26. fix) = —-— on [—1, 1] MVT to show that
1 |tanw — tanu| > |u
27. fix) = on [1,2]
x —2 1
40. If fix) — - on [— 1, 1], does the mean value theorem apply?
28. fix) = 3x + secx on [—n, tt]
Why or why not?
29. fix) — sin2x on [—|]
41. If gix) = |x| on [—2, 2], does the mean value theorem apply?
30. fix) = Vln x on [1, 2] Why or why not?
31. Let gix) — 8x3—6x+8. Find a function / with fix) = gfx) 42. Counterexample Problem Is it true that
and /(1) = 12.
|cosx — cosyl < |x — y|
32. Let g(x) = Vx2 + 5. Find a function / with fix) = g\x)
and / (2) = 1. for all x and y? Either prove that the inequality is always valid
x ~\~ A —9 or find a counterexample.
33. Show that fix) = —- and g(x) =- differ by a con-
5 —x x —5 43. Counterexample Problem Consider
stant. Are the conditions of the constant difference theorem
satisfied? Does fix) = g'(x)? 1 if x > 0
fix) =
34. Let fix) = (x — 2)3 and g(x) = (x2 + 12)(x - 6). Use / and -1 if x < 0
g to demonstrate the constant difference theorem.
fix) = 0 for all x in the domain, but / is not a constant. Does
35. Let fix) — (x — l)3 and gix) — (x2 + 3)(x — 3). Use / and this example contradict the zero-derivative theorem? Why or
g to demonstrate the constant difference theorem. why not?
36. Let / be defined as shown in Figure 4.13. 44. a. Let n be a positive integer. Show that there is a number c
between 0 and x for which
(1 +x)n - 1 ft —1
= ni 1 + c)
46. Use the MVT to evaluate 53. Use the MVT to show that
cosx + 1
lim - Vl + x — 4 < |(x — 15)
x -»7r + X — 7T
47. Let fix) = 14—. If a and b are constants such that a < 0 if x > 15. Hint: Let fix) = + x.
x
and b > 0, show that there is no number w between a and b 54. Use the MVT to show that
for which
1 1 2
- > —|—-(2 — x)
fib) - f (a) = f'(w)(b - a) 2x + 1 5 25
48. Show that for any x > 4, there is a number w between 4 and if 0 < x < 2.
x such that
Q 55. Let fix) = tanx. Note that
sfx - 2 _ 1
x — 4 2 y/w fin) = fi 0) = 0
Our next goal is to see how information about the derivative f and the second deriva¬
tive f" can be used to determine the shape of the graph of /. We begin by showing
how the sign of /' is related to whether the graph of / is rising or falling.
202 Chapter 4 Additional Applications of the Derivative
y. 1
2)
A =/ (x)
z
Figure 4.15 A population curve /(. 1)
1
I
1 1 X
T x?
WARNING .
A note on usage: We say
that functions are increasing and that
graphs are rising. For example, if
fix) = x~, we say that the function /
is increasing for x > 0 and that the
graph is rising for x > 0.
Figure 4.17 The graph is rising where /' > 0 and falling where /' < 0.
Notice that the small flags indicate the slope at various points on the graph.
4.3 Using Derivatives to Sketch the Graph of a Function 203
Proof We will prove that / is strictly increasing on (a, b) if fix) > 0 throughout
the interval. The strictly decreasing case is similar and is left as an exercise for the
reader.
Suppose fix) > 0 throughout the interval (a, b), and let x\ and *2 be two num¬
bers chosen arbitrarily from this interval, with x\ < x2. The MVT tells us that
for some number c between x\ and x2. Because both /'(c) > 0 and x2 — x\ > 0, it
follows that /'(c)(x2 — x\) > 0, and therefore
That is, if Xi and x2 are any two numbers in ia,b) such that x\ < x2, then
fix 1) < / (x 2), which means that / is strictly increasing on (a, b). □
Solution
First, we find the derivative:
Next, we determine the critical numbers: fix) exists for all x and fix) = 0 at
x = - \ and x = 3. These critical numbers divide the x-axis into three parts, as shown
in Figure 4.18a, and we select any arbitrary number from each of these intervals. For
example, we select -2, 0, and 4, evaluate the derivative at these numbers, and mark
each interval as increasing (/) or decreasing 4), according to whether the derivative
is positive or negative, respectively. This is shown in Figure 4.18b. The function / is
increasing for x < -1 and for x > 3; / is decreasing for -1 < x < 3. ■
204 Chapter 4 Additional Applications of the Derivative
a. Intervals where/is increasing and b. The graph of/along with the sign graphs
where it is decreasing for/and/1
Solution
The graphs of / and f are shown in Figure 4.19.
a. When /'(x) > 0, the graph of / is rising. This occurs when x < — 1 and when
* > 3.
b. When the graph of / is falling (for —1 < x < 3), we have /'(x) < 0, so the graph
of f is below the x-axis.
c. The critical numbers of / are where f'(x) = 0; that is, at x = —1 and x = 3, so
they are the x-intercepts of the graph of /'. ■
Every relative extremum is a critical point. However, as we saw in Section 4.1, not
every critical point of a continuous function is necessarily a relative extremum. If the
Figure 4.19 Graphs of / and f derivative is positive to the immediate left of a critical number and negative to its im¬
mediate right, the graph changes from increasing to decreasing and the critical point
must be a relative maximum, as shown in Figure 4.20a. If the derivative is negative to
the immediate left of a critical number and positive to its immediate right, the graph
changes from decreasing to increasing and the critical point is a relative minimum
(Figure 4.20b). However, if the sign of the derivative is the same on both immediate
sides of the critical number, then it is neither a relative maximum nor a relative min¬
imum (Figure 4.20c). These observations are summarized in a procedure called the
first-derivative test for relative extrema.
Step 1. Find all critical numbers of a continuous function /. That is, find all
numbers c such that /(c) is defined and either /'(c) = 0 or /'(c) does
not exist.
Step 2. Classify each critical point (c, /(c)) as follows:
c. The point (c, /(c)) is not an extremum if the derivative fix) has
the same sign for all x in open intervals (a, c) and (c, b) on each
side of c.
Solution Because g'(t) = 1 - 2 cos t exists for all t, the only critical numbers occur
when g'(t) = 0; that is, when cos t = Solving, we find that the critical numbers for
git) on the interval [0, 27r] are j and
Next, we examine the sign of g'(t). Thanks to the intermediate value theorem, be¬
cause g'(t) is continuous, it is enough to check the sign of g\t) at convenient numbers
on each side of the critical numbers, as shown in Figure 4.21a. Notice that the arrows
show the increasing and decreasing pattern for g. According to the first-derivative test,
there is a relative minimum at | and a relative maximum at ^f. The graph of g is
shown in Figure 4.21b. ■
Knowing where a given graph is rising and falling gives only a partial picture of the
graph. For example, suppose we wish to sketch the graph of /(x) = x3 + 3x + 1. The
derivative f'(x) = 3x2 + 3 is positive for all x so the graph is always rising. But in
what way is it rising? Each of the graphs in Figure 4.22 is a possible graph of /, but
they are quite different from one another.
Figure 4.22 shows three possibilities for the graph of /. All three are always rising, but
they differ in the way they “bend” as they rise. The “bending” of a curve is measured
by its concavity. If a curve lies above its tangent lines on some interval (a, b), then
we say the curve is concave up, and if the curve lies below its tangent lines, we say
it is concave down on (a, b). We will examine concavity, and then return in Example
4 to determine which of the three candidates in Figure 4.22 is indeed the graph of
f(x) = x3 + 3x + 1.
Concavity If the graph of a function / lies above all of its tangents on an interval /, then it
is said to be concave up on 7. If the graph of / lies below all of its tangents on
I, it is said to be concave down.
To be specific, a portion of a graph that is cupped upward is called concave up, and
a portion that is cupped downward is concave down. Figure 4.23 shows a graph that is
concave up between A and C and concave down between C and E. At various points
on the graph, the slope is indicated by “flags,” and we observe that the slope increases
from A to C and decreases from C to £. This is no accident! The slope of a graph
increases on an interval where the graph is concave up and decreases where the graph
is concave down.
Conversely, a graph will be concave up on any interval where the slope is increas¬
Figure 4.23 The slope of a graph ing and concave down where the slope is decreasing. Because the slope is found by
increases or decreases,
computing the derivative, it is reasonable to expect the graph of a given function / to
depending on its concavity.
be concave up where the derivative /' is strictly increasing. According to the mono¬
tone function theorem (Theorem 4.7), this occurs when (/')' > 0, which means that
the graph of / is concave up where the second derivative f" satisfies f" > 0. Simi¬
larly, the graph is concave down where j " < 0. We use this observation to characterize
concavity.
4.3 Using Derivatives to Sketch the Graph of a Function 207
Solution
We find that /'(a) = 3a2 + 3 and /"(a) = 6a. Therefore, /"(a) < 0 it a <0 and
/"(a) > 0 if a > 0, so the graph of / is concave down for a <0 and concave up for
Figure 4.24 Intervals of concavity a > 0, as indicated in Figure 4.24. Returning to Figure 4.22c, the graph is concave
down to the left of a = 0 and concave up to the right. Hence, this is the graph of /(a).
In Figure 4.23, notice that the graph changes from concave up to concave down at
the point C. It will be convenient to have a special name for such transition points.
Inflection Point A point P(c, /(c)) on a curve is called an inflection point if the graph is concave
up on one side of P and concave down on the other side.
WARNING . . - . . , .,
An inflection point must be
on the graph, meaning /(c) must be
defined if there is an inflection point at
x — c. Returning to Example 4, notice that the graph of /(a) = a~ + 3a + 1 has exactly
one inflection point, at (0, 1), where the concavity changes from down to up.
Various kinds of graphical behavior are illustrated in Figure 4.25. Note that the
graph is rising on the interval [a, cj], falling on [c\, C2], rising on [C2, C3], falling on
[c3, c4], rising on [c4, c5], and falling on [c5, b]. The concavity is up on [pu P2I and
down otherwise. In this figure, and elsewhere, when the graph of / is displayed, we
will indicate the sign of /" on an interval by a string of + signs if /" > 0 on the
interval, and a string of — signs if /" < 0 there.
The graph has relative maxima at cu c3, and c5, and relative minima at c2 and c4.
There are horizontal tangents (/'(a) = 0) at all of these points except at ci and c3,
where there are sharp points, called comers (/'(c 1) and /'(c3) do not exist). There
is a horizontal tangent at pr, that is, f\px) = 0, but no relative extremum appears
there. Instead, we have points of inflection at p\ and p2, because the concavity changes
direction at each of these points.
In general, the concavity of the graph of / will change only at points where
f”(x) = 0 or /"(a) does not exist—that is, at critical numbers of the derivative /'(a).
We will call the number c a second-order critical number if /"(c) = 0 or /"(c)
does not exist, and in this context an “ordinary” critical number (where /'(c) = 0 or
/'(c) does not exist) will be referred to as a first-order critical number. If we do not
specify otherwise, a critical number is always a first-order critical number. Inflection
points correspond to second-order critical numbers and must actually be on the graph
208 Chapter 4 Additional Applications of the Derivative
Figure 4.25 A graph of a function showing critical points and inflection points
of /. Specifically, a number c such that /"(c) is not defined and the concavity of /
changes at c will correspond to an inflection point if and only if /(c) is defined.
Here is an example of one way inflection points may appear in applications.
WARNING A ,
A continuous runction / Solution
need not have an inflection point at We assume that the morning shift runs from 8:00 A.M. until noon and that worker
every number c where /"(c) = 0. For efficiency is maximized when the rate of production
instance, if f(x) = x4, we have
,/ "(0) = 0, but the graph of / is
R(t) = Q'(t) = -30+ 18r + 12
always concave up (see Figure 4.26).
which is zero when t = 3; this is the critical number. Using the optimization criterion
of Section 4.1, we know that the extrema of R(t) on the closed interval [0, 4] must
occur at either the interior critical number 3 or at one (or both) of the endpoints (which
are 0 and 4). We find that
/?(0) = 12 R( 3) = 39 R( 4) = 36
so the rate of production R(t) is greatest and the worker is performing most efficiently
when t — 3; that is, at 11:00 A.M. The graphs of the production function Q and its
derivative, the rate-of-production function R, are shown in Figure 4.27. Notice that the
production curve is steepest and the rate of production is greatest when t = 3.
4.3 Using Derivatives to Sketch the Graph of a Function 209
y.
R(t ) = -3l 2 + 18r + 12^
~ I.
/
12,
0 Ci = 3 t
_
a. A production curve b. Rate of production
Figure 4.27 Graph of a production curve showing the point of diminishing returns
In Example 5, note how the rate of production, as measured by the slope of the
graph of the average worker’s output, increases from 0 to the inflection point I and
then decreases from / to E, as shown in Figure 4.27a. Because the point / marks the
point where the rate of production “peaks out,” it is natural to refer to / as a point of
diminishing returns. It is also an inflection point on the graph of Q. Knowing that
this point occurs at 11:00 A.M., the manager of the factory might be able to increase
the overall output of the labor force by scheduling a break near this time, since the
production rate changes from increasing to decreasing at this time.
It is often possible to classify a critical point P(c, f (c)) on the graph of / by examining
the sign of /"(c). Specifically, suppose /'(c) = 0 and /"(c) > 0. Then there is a
horizontal tangent line at P and the graph of / is concave up in the neighborhood of
P. This means that the graph of / is cupped upward from the horizontal tangent at P,
and it is reasonable to expect P to be a relative minimum, as shown in Figure 4.28a.
Similarly, we expect P to be a relative maximum if /'(c) = 0 and /"(c) < 0, because
the graph is cupped down beneath the critical point P, as shown in Figure 4.28b.
Fet / be a function such that /'(c) = 0 and the second derivative exists on an
open interval containing c.
Solution
Once again, we begin by finding the first and second derivatives:
We see f'(x) = 0 at x = —1,0, and 1. We can plot these points on a number line,
shown in Figure 4.29, and then evaluate f'(x) at test numbers just to the left and just
to the right of each critical number.
Direction:
A
\ l
\
\ Figure 4.29 Sign graph for the derivative of /(x) = 3x5 — 5x3 + 2
15U(a:-1)(j:+1); JS eg£ .tiv e ( ) t'Jeg ati\ rc
We show a number line like this as part of the graph of many of our examples (as
Figure 4.30 The graph of we did in Figure 4.25), but most often we show the derivative to the right and left of
/(x) = 3x5 — 5x3 + 2 is falling on a critical number. For Example 6, the derivative is negative both to the immediate left
both sides of x = 0. and right of 0, which we illustrate as shown in Figure 4.30. Neither kind of extremum
occurs at x = 0 | pattern).
Solution
We begin with the second derivative (from Example 6):
/"(*) = 30x(2x2 - 1)
To find the inflection points, we look at the second-order critical numbers, namely,
where f"(x) = 0; that is, at x = 0 and x = ±y/^ = We can show these on a
sign graph of the second derivative, as in Figure 4.31. Notice that the inflection points
occur where the sign of the second derivative changes.
The graph of / is shown in Figure 4.32, along with the sign graphs for both the
first and second derivatives. Spend some time studying this figure to see how the signs
of the first derivative indicate where the graph is rising and where it is falling, as well
4.3 Using Derivatives to Sketch the Graph of a Function 211
concave concave
concave down up down concave up
+ + + + ++++++++
/"< lilt I I <t> I I I 1 i +->
-1.5 -1.0 -0.5 0.5 1.0 1.5
Figure 4.31 Sign graph for the second derivative of f(x) = 3x5 — 5x3 + 2
as how the signs of the second derivative show the concavity. For completeness, the
figure also shows the sign graph for the function itself: The graph is above the x-axis
where / is positive, below the x-axis where / is negative, and crosses the x-axis where
/(x) - 0.
Example 7 demonstrates both the strength and the weakness of the second-derivative
test. In particular, when it is relatively easy to find the second derivative (as with a poly¬
nomial) and if the zeros of this function are easy to find, then the second-derivative test
provides a quick means for classifying the critical points. Flowever, if it is difficult
to compute f"{c) or if /"(c) = 0, it may be easier, or even necessary, to apply the
first-derivative test.
Now, we are ready to develop a procedure for curve sketching. The key ideas appear
in Example 8.
Solution
The first derivative,
f\x) = 4x3 - 12x2 = 4x2(x - 3)
is zero when x = 0 and when x = 3. Because 4x2 > 0 for x / 0, we have /'(x) <0
for x < 3 (except for x = 0) and /'(x) > 0 for x > 3. The pattern showing where /
is increasing and where it is decreasing is displayed in Figure 4.33a.
Next, to determine the concavity of the graph we compute
f"{x) = 12x2 - 24x = 12x(x - 2)
If x <0 or x > 2, then /"(x) > 0 and the graph is concave up. It is concave down
when 0 < x < 2, because /"(x) < 0 on this interval. The concavity of the graph of /
is shown in Figure 4.33b.
212 Chapter 4 Additional Applications of the Derivative
"
nT ! \ / !
4.v2(ar-3): Negative 0 hleg ative 3 Posi tive 12x(x-2): Positive 0 hlegative P’ositive x
Figure 4.33 Intervals of increase and decrease and concavity for f(x) = x4 — 4x3 + 10
The two diagrams in Figure 4.33 tell us that there is a relative minimum at x = 3
and inflection points at x = 0 and x = 1 (because the second derivative changes sign
at these points).
To find the y-values of the critical points and the inflection points, evaluate / at
* = 0, 2, and 3:
’ ’ »
Finally, to sketch the graph of /, we first place a “cup” (U) at the minimum point
(3, —17) and note that (0, 10) and (2, —6) are inflection points. Remember there is
also a horizontal tangent at (0, 10). The preliminary graph is shown in Figure 4.34a.
Complete the sketch by passing a smooth curve through these points, using the two
diagrams in Figure 4.33 as a guide for determining where the graph is rising and falling
and where it is concave up and down. The completed graph is shown in Figure 4.34b.
yt
f
_
'k
X
/i
i ' /1
\ / i
| \ i
\
; i
i
i
+ + + + +
1 i
i
i
i
—l
f
J
_1 1
1
+ +
0 3
X
— — -1 + V +
f' .4-
J V X
0 ■ 3
+ + _ - - + + + +
Jf"
_ ~i +
+
2 4
X
/2
/(*) =
is increasing, decreasing, concave up, and concave down. Find the relative extrema
and inflection points and sketch the graph. This function plays an important role in
statistics, where it is called the standard normal density function.
4.3 Using Derivatives to Sketch the Graph of a Function 213
Solution
The first derivative is
—x ,~x2/2
/ \ /'(*)
f2n
Because e x~ ^ is always positive, fix) = 0 if and only if x = 0. Hence, the corre¬
X
sponding point
a. Intervals of increase and
decrease for j\x) 0, (0,0.4)
fliz )
1-1- is the only critical point. Checking the sign of /' on each side of 0, we find that /
is increasing for x < 0 and decreasing for x > 0, so there is a relative maximum at
x:J |m ]W
,i , ;
Solution
You probably graphed this function in trigonometry by adding ordinates. However,
with this example we wish to illustrate the power of calculus to draw the graph. Thus,
we begin by finding the first and second derivatives.
T\x) = cosx - sinx T"(x) = - sinx - cosx
We find the critical numbers (both V and T" are defined for all values of x):
The intervals of increase and decrease as well as the concavity pattern are shown in
Figure 4.37.
214 Chapter 4 Additional Applications of the Derivative
Shape: ( h V )
-sin x - cos x: -F —
rL
o
3:t 7ji 3jT
4 4
Find the critical points, the points of inflection, and the endpoints:
Finally, pass a smooth curve through these key points as indicated in the preliminary
sketch in Figure 4.38a to obtain the completed graph shown in Figure 4.38b.
y y>
jt
( 4’ Vz
) p ela try. e M axi mu m 1H 1
/
i—
Is X
(C\ 1 ^ l 1 \ T 1 2 71
] nfl ection Po nt i 1
1 1
1
1
y /
9 1
A
y
1
1
1
1
1
1
1
1
1
1
Pi P2 VI ? 2 n' + 1 + + 1
1 - - { +
f 4
1 1 X
i
4 2 1 i 1
i j
r- y fT\ + 1
- - i + +
4 4
A4 f
n i —r—
1 X
i 571
i
Re lati ve min . 4 - - i
+
i
f-
— 4
+ 4+ + +
I
X
— 371 7k
4 4
Draw a curve that represents the derivative of the function defined e~x
37. f{x) = -—— at x = 1, * = \
by the curves shown in Problems 8—11. 3 — 2x
38. f(x) = (x2 -4)4(x2 - l)3 atx = l,x = 2
39. fix) = fx2 — 48x at x = 4
In Problems 40—43, use the second-derivative test to classify each
of the given critical numbers as a relative minimum, a relative
maximum, or neither.
40. fix) = 2x3 + 3x2 — 12x + 11 atx = 1, x = — 2
o
when x < 2 and when
V
><
In Problems 20-35, determine the intervals of increase and de¬ fix) < 0 when x > 1
crease and concavity for the given function, and then use those fix) > 0 when x < 1
intervals to help you sketch its graph.
fix) > 0 when x > 1
20. fix) = (x - 12)4 - 2(x - 12)3
21. fix) = 1 +2x + 18/x What can you say about the derivative of / when x = 1 ?
22. fiu) = 3m4 - 2u3 - 12u2 + 18m - 5 47. Sketch the graph of a function with the following properties:
There are relative extrema at (—1, 7) and (3, 2). There is an
23. giu) = m4 + 6m3 — 24i/“ + 26
inflection point at (1,4). The graph is concave down only
24. fix) = fx2 + 1 25. git) = it3 + t)2
when x < 1. The x-intercept is (-4, 0) and the y-intercept is
26. fit) = it3 + 3r2)3 27. fix) = (0, 5).
x2 + 1
48. Sketch a graph of a function / that satisfies the following con¬
28. fit) - t - Inf 29. fit) = t2e~3'
ditions:
30. fix) =ex + e~x 31. fix) = (lnx)2
(i) fix) > 0 when x < —5 and when x > 1.
32. tiO) = sin0 - 2cos0 forO < 0 < 2n
(ii) fix) < 0 when — 5 < x < 1.
33. tiO) = 9 + cos20 forO < 0 < n
(iii) /(—5) = 4 and /(l) = — 1.
34. fix) =2x- sin"1 x for -1 < x < 1
49. Sketch a graph of a function / that satisfies the following con¬
35. fix) = x3 + sinx on [-f, f]
ditions:
In Problems 36—39, use the first-derivative test to classify each of
(i) fix) < 0 when x < — 1.
the given critical numbers as a relative minimum, a relative maxi¬
(ii) fix) > 0 when -1 < x < 3 and when x > 3.
mum, or neither.
(iii) /'(-l) = 0and/'(3) = 0.
36. fix) = (x3 - 3x + l)7 atx = l,x = -1
216 Chapter 4 Additional Applications of the Derivative
50. In physics, the formula with the bird’s size and shape.* What speed will minimize
the power? You may assume that v, w, p, S, and A are all
2
positive.
I(0) = Io where 1(0) = lim 1(6)
0->O 55. The deflection of a hardwood beam of length is given by
b. The critical volume Vc is the volume that satisfies will have a relative extremum at (2, 11) and an inflection point
P'(VC) = 0 and P"(VC) = 0. Find Vc. at (1,5). Sketch the graph of /.
c. Find 7c, the point where P"(V) — 0, using the Vc from 62. Find constants a, £>, and c that guarantee that the graph of
part b to write it in terms of a,b,n, and R. Finally, find the
critical pressure Pc = P(VC) in terms of a, b, n, and R. /(x) = x3 + ax2 + bx + c
d. Sketch P as a function of V. will have a relative maximum at (—3, 18) and a relative mini¬
59. Prove or disprove that if the graphs of the functions / and g mum at (1, -14).
are both concave up on an interval, then the graph of their sum
63. Exploration Problem Consider the graph of y = x3 +bx2+cx +
f + g is also concave up on that interval.
d for constants b, c, and d. What happens to the graph as b
60. Use calculus to prove that for constants a, b, and c, the vertex changes?
(relative extremum) of the quadratic function
64. Find constants A, B, C, and D that guarantee that the graph
of
y = ax2 + bx + c (a 0)
/ (x) = 3x4 + Ax3 + Bx2 + Cx + D
occurs atx = —b/(2a). will have horizontal tangents at (2, —3) and (0, 7). There is a
61. Find constants A, B, and C that guarantee that the function third point that has a horizontal tangent. Find this point. Then,
for all three points, determine whether each corresponds to a
f lx) = Ax3 + Bx~ T C relative maximum, a relative minimum, or neither.
-#
LIMITS TO INFINITY
lim f(x) = L
x->+oo
to indicate that /(x) approaches the number L as x increases without bound. Similarly,
we write
lim /(x) = M
*—►—00
to indicate that /(x) approaches the number M as x decreases without bound. Here
are the formal definitions of these limits to infinity.
Limits to Infinity The limit statement lim f{x) — L means that for any number e > 0, there
*->•+00
lin>/(x) = M
X-*-00
With this formal definition, we can show that all the rules for limits established in
Chapter 2 also apply to lim f(x) and lim f(x).
X—»+00 X—>—oo
Limit Rules
If lim f(x) and lim g(x) exist, then for constants a and b\
X—»+00 X—>+00
f(x) /(*)
Quotient rule: lim -= °°-if lim g(x) 7^ 0
*-*•+00 g(jc) lim g(x) x-*-+oo
x—>-+00
Analogous results hold for lim f(x), if it exists.
X—> —OO
The following theorem will allow us to evaluate certain limits to infinity with ease.
1
Proof We begin by proving that lim — — 0. For e > 0, let N = Then for
A' — +OO X
x > N we have
1
x > N = - so that — < €
€ x
4.4 Curve Sketching with Asymptotes: Limits Involving Infinity 219
This means that --0 < e, so that, from the definition of limit we have lim - = 0.
x .x-*+oo x
= A a/ = A a/O = A-0 = 0
V A lim ~
>+0O X L -1
3x3 — 5x + 9
Evaluate lim
a >+oo 5x3 + 2xL
Solution
We may assume that x 7^ 0, because we are interested only in very large values of
x. Dividing both the numerator and denominator of the given expressions by x3, the
highest power of x appearing in the fraction, we find
1
3-- H—j
3x3 — 5x + 9 3x3 — 5x + 9 xz xi
5x3 + 2x2 - 7 5x3 + 2x2 - 7 1 2 7
5 4-r
Thus,
3 -
3x3 — 5x + 9 ~2
X1
H-3
XJ
lim , „ , — lim
a >+oo 5xi + 2xz 7 A—>+00
5 4-t
lim 3- 4-
A—>+00
lim
x—>+oo ,5+x“+.
3-0 + 0 3
5 + 0-0 5
Figure 4.40 Graph of
The graph of the given rational function is shown in Figure 4.40. Notice how the curve
3x3 — 5x + 9
seems to approach y — | as x —> +oo and as x -+ — oo. ■
y = 5x3 + 2x2 - 7
Example 2 illustrates how Theorem 4.8 can be used along with the other limit
properties to evaluate limits to infinity.
Evaluate
3
3x -5 '3x -5'
lim and lim
A —> +00 x -2 x—>—00 x — 2
220 Chapter 4 Additional Applications of the Derivative
Solution
Notice that for x 0,
5' 5
x 3 3- -
3x - 5 x _X_
2 2
x 1 1
x *
Also, according to Theorem 4.8, we know that
5 2
lim — = 0 and lim — = 0
x-++oo X *-*+00 X
We now find the limits using the quotient rule, the power rule, and Theorem 4.8:
l3x -5 / 3x — 5 \ 1/2 /- 3x — 5 \ 1/2
lim ./- = lim - = lim
— - 1
JT-^+OO V x —2 X-++OO V x — 2 ) \*->-+oo x — 2 !
1/2
lim (3 1/2
X—>-+00 v '3-0'
= V3
lim (1 - 2) r^o.
,x^+ooV xJ
Similarly,
'3x — 5'
lim = 33 = 27
X—>— OO x —2
95x3 + 51 x + 30
Evaluate lim
*-+-00 x5 - 1,000
Solution
Dividing the numerator and the denominator by the highest power, x5, we find that
1
95x3 + 51 x + 30 95x3 + 51x + 30
lim = lim
x-^-00 x5 1,000 — x—>—OO x5-1,000
95 57 30
X^ + X4 + X5
lim
x—>—OO 1,000
1 -
0+0+0
1 -0
Figure 4.41 Graph of
= 0
95x3 + 57x + 30
The graph of the rational function is shown in Figure 4.41. ■
y ~ x5 - 1,000
Solution
We cannot use the product rule for limits since lim cosx does not exist (it diverges
x—>+00
by oscillation—do you see why?). Note, however, that the magnitude of
cosx
cosx =
must become smaller and smaller as x —> +00 since the numerator cosx is bounded
between —1 and 1, while the denominator ex grows relentlessly larger with x. Thus by
the squeeze rule,
lim e~x cosx = 0
.*—>•+00
Figure 4.42 Graph of y = e x cos x The graph of y = e~x cosx is shown in Figure 4.42. ■
4.4 Curve Sketching with Asymptotes: Limits Involving Infinity 221
WARNING
Try drawing the graph of y = e~x cos* on your calculator, or other graphing
software, and you will see that what you get does not properly illustrate the oscillatory behavior
of this graph. This demonstrates the importance of not relying on technology for graphing. In
fact, the accompanying graph (without scale) shows more clearly the behavior of the graph of
the function y = e~x cos* for large, positive values of*.
INFINITE LIMITS
In mathematics, the symbol oo is not a number, but is used to describe either the process
of unrestricted growth or the result of such growth. Thus, a limit statement such as
means that the function / increases without bound as * approaches c from either side,
while
lim g(*) = —oo
*—
means that g decreases without bound as * approaches c. Such limits may be defined
formally as follows.
Infinite Limits We write lim /(*) = +oo if for any number N > 0 (no matter how large), it is
X—>C
possible to find a number 8 > 0 such that /(*) > N whenever 0 < |* — c| < 8.
Similarly, lim /(*) = —oo if for any N > 0, it is possible to find a number
X -*C
oscillation), so saying that lim /(*) = +oo or lim /(*) = —oo conveys more
*—>C *-*c
information than simply observing that the limit does not exist.
8 — WBTSZEm
Solution
1 1
Notice that-increases without bound as * approaches 2 from the right and-
* - 2 * - 2
decreases without bound as * approaches 2 from the left. That is,
3* -5
lim -= +oo and lim - = —OO
*—2+ * — 2 * 2
—> _ * — 2
222 Chapter 4 Additional Applications of the Derivative
Figure 4.43 shows a graph that approaches the horizontal line y = 2 as x —>• +oo and
y = — 1 as x -+ — oo, and the vertical line x = 3 as x approaches 3 from either side.
Vertical and Horizontal The line x = c is a vertical asymptote of the graph of / if either of the one-sided
Asymptotes limits
Solution
Vertical Asymptotes First, make sure the rational function is written in simplified
3* - 5
(reduced) form. Because vertical asymptotes for f(x) =-— occur at values of
c for which lim f(x) or lim f(x) is infinite, we look for values that cause the de¬
nominator to be zero (and the numerator not to be zero); that is, we solve d(c) = 0,
where d(x) is the denominator of f(x), and then evaluate lim f(x) and lim f{x) to
*—>c~ *—>c+
ascertain the behavior of the function at x — c. For this example, x = 2 is a value that
causes division by zero, so we find
3* -5 3x — 5
lim -= Too and lim -= —oo
*—>2+ X —2 *—>2~ x — 2
(We found these limits in Example 5.) This means that x — 2 is a vertical asymptote
and that the graph is moving downward as a: -h> 2 from the left and upward as * —> 2
from the right. This information is recorded on the preliminary graph shown in Figure
4.44a by a dashed vertical line with upward (f) and downward (|) arrows.
3x — 5 3x — 5 i 3 — ^ 3-0
lim -= lim - ^ = lim -f- = 3
*—>+oo X — 2 *->+oo X — 2 1 A-T-f OO 1 — - 1 -0
X X
4.4 Curve Sketching with Asymptotes: Limits Involving Infinity 223
and
y*
IT
1
Zonca it
1 up
Falii
Pg
i
i v
i -3
i y
i
1
lor icaye 1 X
i
i
d 3W1 1 i
N i
i
i
1 ' ~
r
i
-v F 2
> L
a. Preliminary sketch b. Completed sketch
3x — 5
Figure 4.44 Graph of f(x)
x — 2
The preliminary sketch gives us some valuable information about the graph, but it
does not present the entire picture. Next, we use calculus to find where the function
is increasing and decreasing (first derivative) and where it is concave up and concave
down (second derivative):
-1
fix) and f"(x) =--
(X - 2)2 J (x — 2)3
Neither derivative is ever zero, and both are undefined at x = 2. Checking the signs
of the first and second derivatives, we find that the graph is always falling and that it is
concave up for x > 2 and concave down for x <2. However, it does not have a point
of inflection (the function is not defined at x = 2). This information is added to the
preliminary sketch shown in Figure 4.44a. The completed graph is shown in Figure
4.44b, which also shows the x- and y-intercepts at (|, 0) and (0, |). ■
Solving f(x) = 0, we see that the critical numbers are x = 1 and x = 5. Testing on
each side of the critical numbers and x = 3, where /(x) is not defined, we obtain the
intervals of increase and decrease and concavity shown in Figure 4.45.
Note that there is a relative maximum at x = 1 and a relative minimum at x = 5.
The concavity changes (from | to f) at x = 3, but this does not correspond to an
inflection point since /(3) is not defined. We look for vertical asymptotes where /(x)
(in reduced form) is not defined; in this case, at x = 3. Testing with limits, we find that
_j._2 x~ — x — 2
lim -— —oo and lim -= +oo
x^>3~ X — 3 jc->3+ x — 3
224 Chapter 4 Additional Applications of the Derivative
—
t \ f— /
} \ J
7 —•,
/ u
x2 — x — 2
Figure 4.45 Intervals of increase and decrease and concavity for f(x) =--—
x2 - x - 2 x2 — x — 2
lim -= +oo and ltm --— = —oo
*->+oo x — 3 x-*—oo X — J
Since neither limit of fix) at infinity is finite, there are no horizontal asymptotes.
We plot some points: the relative maximum (1, 1); the relative minimum (5, 9);
the x-intercepts (2, 0), (-1,0); and the y-intercept (0, §). We also show the vertical
Figure 4.46 Graph of
asymptote x = 3, and, using the intervals of increase and decrease and concavity
x — 2
/(*) - indicated in Figure 4.45, we obtain the graph shown in Figure 4.46. ■
Suppose the function / is continuous at the point P where x = c and that fix)
becomes infinite as x approaches c. Then the graph of / has a vertical tangent at P
if the graph turns smoothly through P and a cusp at P if the graph changes direction
abruptly there. These graphical features can be defined in terms of limits, as follows:
Vertical Tangents and Cusps Suppose the function / is continuous at the point Pic, f (c)). Then the graph of
/ has
a vertical tangent at P if lim fix) and lim fix) are either both +oo or
x-+c~ >c+
both — oo;
a cusp at P if lim fix) and lim fix) are both infinite with opposite
x-*c~ X-+C+
c c c c
Solution
a. Writing f(x) — 2x5^3 + 5x2/3, we find that
and
it follows that there is a cusp on the graph of / at (0, 0). The graph is shown in
Figure 4.48a.
b. The derivative
g'(x) = \x~2/i{x + 1)
and
Since the limit approaches +oo as jc approaches zero from both the left and the
right, we conclude that a vertical tangent occurs at the origin (0, 0). The graph is
shown in Figure 4.48b.
Figure 4.48 A graph with a cusp and another with a vertical tangent
Step Procedure
Simplify. If possible, simplify algebraically the function you wish to graph. For example, if
/ ^ ^W ^ I ^\
f(x) = ——-, you must write this as /(x) - x + 3, x # 2 before beginning the
x — 2
procedure listed here.
Find derivatives Compute the first and second derivatives; set each equal to zero and solve (by factoring, if
and critical numbers. possible), and find the first- and second-order critical numbers.
Apply the second-derivative test. Use the second-derivative test to find the relative maxima or minima: Substitute the first-order
critical numbers, C], with the following test:
Determine concavity and points Find the points of inflection. These are located where the concavity changes from up to down or
of inflection. down to up, and are found by checking the intervals on either side of the second-order critical
numbers.
Apply the first-derivative test. Use the first-derivative test if the second-derivative test fails or is too complicated.
fix) > 0 for all x in an open interval (a, Ci) to the left of c\
/ \
Maximum
fix) < 0 for all x in an open interval (ci, b) to the right of cj.
Find asymptotes, vertical 1. Vertical asymptotes—Vertical asymptotes, if any exist, will occur at values x = c for which /
tangents, and cusps. is not defined. Use the limits lim /(x) and lim /(x) to determine the behavior of the graph
X ->c+
if lim /'(x) and lim f(x) are both infinite with opposite signs.
A—>C“ X—
Plot points. 1. If /(c) is a relative maximum, relative minimum, or inflection point, plot (c, /(c)). Show the
relative maximum points by using a “cap” (Pi) and relative minimum points by using a "cup
(U).
2. x-intercepts: Set y = 0 and if x = a is a solution, plot (a, 0).
y-intercepts: Set x = 0 and if y = b is a solution, plot (0, b)\ a function will have, at most,
one y-intercept.
50. In an experiment, a biologist introduces a toxin into a bacterial and creation. Bhaskaracharya gave a solution for the so-
colony and then measures the effect on the population of the called Pell equation
colony. Suppose that at time t (in minutes) the population is
x2 = 1 + py2 x, y both nonnegative
Pit) = 5 + e~0 04' (t + 1)
which is related to the problem of cutting a given sphere so the
thousand. At what time will the population be the largest? volumes of the two parts have a specified ratio.
What happens to the population in the long run (as t —> +oo)? Solve this equation for p = 2/3 and write this Pell equation
Find where the graph of P has an inflection point, and inter¬ as y = /(x).f Find lim f(x).
*-►+00
pret this point in terms of the population. Sketch the graph
Counterexample Problems In Problems 56—59, either show
of P.
that the statement is generally true or find a counterexample.
51. Let P{x) =a„xn+an-Xxn-]+ --- + cnx + a0 be a polynomial
with an ± 0 and let L = lim P(x) and M = lim P(x). 56. If / is concave up and g is concave down on an interval I,
*-►-00 *->+oo then fg is neither concave up nor concave down on /.
Fill in the missing entries in the following table:
57. If / and g are concave up on the interval /, then so is / + g.
58. If fix) > 0 and g(x) > 0 for all x on / and if / and g are
Sign of a„ n L M
concave up on 1, then fg is also concave up on I.
+ even a. +00 59. If f(x) < 0 and fix) > 0 for all x on 7, then the function
g = f2 is concave up on /.
+ odd —00 b.
60. Exploration Problem State what you think should be the formal
— even c. d. definition of each of the following limit statements:
b. Sketch the graph of each of the following functions: 62. Prove the following limit rule. If lim fix) = +00 and
X—>C
53. Find constants a and b that guarantee that the graph of the
Hint: Notice that because limg(x) = A, the function
function defined by X—>C
ate a limit of the form lim , where lim f(x) and lim g(x) are either both 0 or both
-v->c g(x) x-+c X->c
oo. Such limits are called 0/0 indeterminate forms and oo/oo indeterminate forms, re¬
spectively, because their value cannot be determined without further analysis. There is
a rule to evaluate such indeterminate forms, known as I’Hopital’s rule, which relates
f'(x)
the evaluation to a computation of lim :-, the limit of the ratio of the derivatives of
x^c g (x)
f and g. Here is a precise statement of the rule.
g'(x)
..
lim- -
/(*) L
g(x)
The theorem also applies to one-sided limits and to limits at infinity (where x -» +oo
and x —> —oo).
we can write
so
,. fix) f'{a)
lim-= □
x^a g(x) g'(a)
Solution
Note that this is an indeterminate form because sin x and x both approach 0 as x ^ 0.
This means that I’Hopital’s rule applies:
sinx cosx
lim-= lim-= 1 ■
x-^-0 X 1
Solution
For this example, /(x) = x7 - 128 and g(x) = x3 - 8, and the form is 0/0.
x1 - 128 . 7x6
lim lim I'Hopital's rule
x-t-2 x3 - 8 x->2 3x2
Simplify.
7(2)4 112
Limit of a quotient
~T~ ~ T~
Solution
You must always remember to check that you have an indeterminate form before ap¬
plying I’Hopital’s rule. The limit is
lim(l — cosx)
1 — cosx x-+0
lim-
jc-+o secx lim sec x
x->0
WARNING
If you blindly apply I'Hopital's rule in Example 3, you obtain the WRONG answer:
1 — cos x cm y
lim- = lim- This is NOT correct.
x-+o secx *-►() secx tanx
cosx
= lim-
*-►0 secx
This answer is blatantly WRONG, as you can see by looking at the Technology Note.
4.5 I’Hopital's Rule 231
TECHNOLOGY NOTE
V V
instance, the limit in Example 3 can easily
be checked by looking at the graph. If you
have a graphing calculator, you can see that
as x -* 0 from either the left or the right,
the limit looks the same, namely, 0.
ViBC1-cos X>cos
X
Xpi i n= "6. 283185...
Xniax=6.2831853...
Xsc 1 = 1.5707963...
Ypiin= _3
Vmax=1
Yscl=l
Solution
This is a 0/0 indeterminate form, and we find that
x — sin x 1 — cos*
lim-r-= lim ——
x—>0 X3 x->0 3x2
This is still the indeterminate form 0/0, so I’Hopital’s rule can be applied once again:
Solution
Using the methods of Section 2.2, we could compute this limit by multiplying by
(l/x2)/(l/*2). Instead, we note that this is of the form 00/00 and apply I’Hopital’s
rule:
2*2 - 3* + 1 4* - 3
lim -= hm --- Apply I'Hopital's rule again.
*-*•+00 3*2 + 5* — 2 ^->+00 6* + 5
4 2
= lim —
a:—>+00 6 3
It may happen that even when I’Hopital’s rule applies to a limit, it is not the best
way to proceed, as illustrated by the following example.
Solution
This limit has the form 0/0, but direct application of l’Hopital’s rule leads to a real
mess (try it!). Instead, we compute the given limit by using the product rule for limits
first, followed by two simple applications of THopital’s rule. Specifically, using the
product rule for limits (assuming the limits exist), we have
o
*-►0
H
Hf
t
sinx 4cos4x 1
lim- lim- lim-
.x—>o 2x *—>0 1 *-►0 cos*
(4) (1) = 2
Solution
This limit has the indeterminate form oo/oo. If you try to apply THopitaTs rule, you
find
x 4- sinx . 1 + cosx
lim -= lim -
*—►+00 x —> cosx x->4-oo 1 + sinx
The limit on the right does not exist, because both sinx and cosx oscillate between
f\x)
— 1 and 1 as x —> +oo. Recall that I’Hopital’s rule applies only if lim — L or is
g'(x)
±oo. This does not mean that the limit of the original expression does not exist or that
we cannot find it; it simply means that we cannot apply THopital’s rule. To find this
limit, factor out an x from the numerator and denominator and proceed as follows:
sinx
x + sinx x(l + 2M) 1 + Jt 1+0
lim -
*->4-oo x — COSX Ja. ,(i - +> lim
*->4-oo l - 1 -0
= 1 ■
WARNING ..... . „ .... Other indeterminate forms, such as l00, 0°, oo°, oo - oo, and 0 • oo, can often be ma¬
I Hopital s rule itself
applies only to the indeterminate form nipulated algebraically into one of the standard forms 0/0 or oo/oo and then evaluated
0/0 and oo/oo. using THopital’s rule. The following examples illustrate such procedures.
We begin by deriving a formula given without proof in Section 2.4.
Solution
Note that this limit is indeed of the indeterminate form l00. Let
x
L = lim
*->4-oo X
4.5 I'Hopital's Rule 233
In L = In lim 1 + -
*-►+00 V X .
In 1 +
0
— lim Form -
*-»+oo 1 0
X
1 + 1/x \
= lim I'Hopital's rule
*—►+00
= lim -— Simplify
*-►+00 „ 1
1 + -
JC
1
r+o
Thus, In L = 1 and L = ex = e.
Solution
This limit has the form 0 • 00, because
1
Write tanx = to obtain
cotx
X - -
x 2
lim (x — — ) tanx = lim Form
*—►7T/2" V 2/ x-*jt/2~ cotx 0
1
= lim I'Hopital's rule
x—nz/2 csc2x
= lim (—sin“x) =
*—*n/2~
Look at the graph as x—> | from both
the left and the right.
Solution
This is a 0° indeterminate form. From the graph shown in Figure 4.49, it looks as
though the desired limit is 1. We can verify this conjecture analytically.
234 Chapter 4 Additional Applications of the Derivative
L = lim xsinr
*->o+
In L = In lim xsinjr
*-> 0+
Thus, L = e° = 1.
Solution
This is a limit of the indeterminate form oo°.
If L = lim x*^,then
*—>+00
lnL = ln lim xx/x
*—>+00
= lim lnx1/x The limit of a log is the log of the limit.
*—>+00
1
= lim - In x
* >+oo X
lnx oo
= lim - This is — form.
* >+oo X oo
1
= lim — I'Hdpital's rule
*—>+oo 1
= 0
Solution
To test for horizontal asymptotes, we compute
1 1
showing the x-axis as a horizontal Evaluate lim I-
*-*0+ \x sinx
asymptote
Solution
1
As it stands, this has the form 00 — 00, because-> +00 and +00 as x —► 0
x sinx
from the right. However, using a little algebra, we find
(1-=
1 \
lim -
sinx — x
x sinx/ x-±o+ xsinx
This limit is now of the form 0/0, so the hypotheses of I’Hopital’s rule are satisfied.
Thus,
sinx — x cosx — 1 0
lim -= lim - Again, the form
*->0+ xsinx *->-o+ sinx + x cosx 0
— sin x
= lim -;-
*-»o+ cosx + x (— sinx) + cosx
WARNING
Not all limits that appear indeterminate actually are indeterminate. For example,
Other such "false indeterminate forms" include +00 + (+00), OO/0, and OO • OO, which are
all actually infinite.
We close this section with a theorem that summarizes the behavior of certain important
special functions involving ex and lnx near 0 and at ±00.
lnx lnx
lim -= —00 lim - = 0
*-+0+ x” x—►-Too x”
ekx
lim — = +00 lim xne~kx = 0
jt-*-+oo x” jc-*+oo
236 Chapter 4 Additional Applications of the Derivative
Proof These can all be verified directly or by applying l’Hopital’s rule. For example,
lnx 1
lim - lim —1— lim - = 0
*->■+00 xn *->+00 nxn x-++oo nx"
ekx lnx
lim — = +00 and lim - = 0
x ►-Too xn X-++00 xn
are especially important. They tell us that, in the long run, any exponential ekx
dominates any power xn, for k and n positive, which, in turn, dominates any
logarithm.
sinx cosx
b. lim lim-= 0 27. lim ( —--— 28. lim x-5lnx
x -+jt/2 x x-^n/2 1 -t-»oysin2x 2x X—>-+00
2. Exploration Problem Sometimes l’Hopital’s rule leads to faulty 29. lim x~5lnx 30. lim (sinx) Inx
X-+0+
computations. For example, observe what happens when the
ln(lnx) ,2x
rule is applied to
31. lim 32. lim ( 1 — —
*->■+00 x *->—oo \ x
lim 3*
*^+00 Jx2 _ l
33. lim ( 1 H- 34. lim (lnx)1/;i
*->•+00 \ 2x x—>+00
Use any method you wish to evaluate this limit.
35. lim (ex + x) \/x 36. lim (sinx)1/lnv/*
Find each of the limits in Problems 3-36. *->0+ *-»0+
, *3~1 ... x3-27 Find each of the limits in Problems 37-51.
3. lim - 4. lim
*-*•1 x1 1 x-*2 X 1
.10
1 .10 37. lim I --(- tan x
- 1 x—>(.n/2)~ \n — 2x
5. lim 6. lim
*-»i x — 1 x-*—\ x + 1
38. lim (\Jx2 — x — x')
1 — COS2 X „ 1 — COS2X ■x—>+00 V /
7. lim 8. lim-
sinJ x *—o 3 sinx 39. lim [x — ln(x3 — 1)] Hint: lne* = x
X—>-+00
cos * 1 — cosx
9. lim-— 10. lim--- 40. lim [x — \n{ex + e~x)] Hint: \nex = x
x->n n — X x—*0 X2 X—t+OO
„„ sin ax tan 3x
11. lim-, ab ± 0 12. lim-
*-*■0 cos bx *-*•0 sin5x 41. lim ( —— lny^l 42. \im{ex - 1 - x)x
*-►0+ \x2 / x—>0
1 — cos2 x
13. lim SinJC 14. lim 43. lim (lnx)(cotx) 44. lim (ex - l)1/lnjr
*->o tanx — x *-*•0 x tan x *-*•0+ jr->0+
tan *(3x) — 3 tan 1 x distribution. He did a great deal of work with differential equa¬
50. lim
X—.0 tions. Johann was jealous and cantankerous; he tossed a son
1 1 (Daniel) out of the house for winning an award he had expected
51. lim jc5 sin to win himself. ■
X-»+00 x 6x3
Write a report on the Bernoulli family.
* + * V _ c
52. Find A so that lim Q Find the limits in Problems 60-62 using the following methods,
x-»+oo JC -2 Af
a. graphically b. analytically c. numerically
In Problems 53-56, use I'Hdpital's rule to determine all horizon¬
Compare, contrast, and reconcile the three methods.
tal asymptotes to the graph of the given function. You are NOT
X
required to sketch the graph. 60. lim x 61. lim ixxY
x-»0+ x—>0+
-0.0 lx
lnx5 .7
53. f(x) = x3e 54. fix) = 1
JC002 62. lim x sin-1
X —> +00 JC
(x + 3\2x
55. f{x) = (In «fx)2lx 56. fix) = 63. Find constants
\x + 2)
57. Exploration Problem Write a paper on using technology to eval¬ sin 2x a
uate limits. lim + ^+b
x—*0
58. HISTORICAL QUEST The French mathematician Guil¬
laume de I'Hopital is best known today for the rule that bears
his name, but the rule was discovered by I'Hopital's teacher, 64. Find all values of A and B so that
Johann Bernoulli. Not only did I'Hopital neglect to cite his sin Ax + Bx
sources in his book, but there is also evidence that he paid lim---= 36
x—>0 Xi
Bernoulli for his results and for keeping their arrangements
for payment confidential. In a letter dated March 17, 1694,
he asked Bernoulli "to communicate to me your discoveries 65. For a certain value of C, the limit
... "—with the request not to mention them to others—"... it
would not please me if they were made public." L'Hopital's lim (x4 + 5jc3 + 3)c — jc
X —>-+00
fia + dx) _ fja) + fia)dx 66. For which values of constants D and E is it true that
OPTIMIZATION PROCEDURE
When light travels from one medium to another—say, from air into water—which path
does it follow? In a mechanical system involving pulleys, which arrangement results in
NOTHING TAKES PLACE in the world
whose meaning is not that of some maxi¬
minimal potential energy? At what temperature does a given mass of water occupy the
mum or minimum. least volume (the answer is not 0°C)? These are typical optimization problems from the
-Leonhard Euler physical sciences and engineering and will be examined in the examples and exercises
of this section. The process of finding the maximum or minimum of a function is called
optimization. Then, in the next section, we will explore optimization involving appli¬
cations to business, economics, and the social and life sciences. We begin by outlining
a general optimization procedure based on ideas originally suggested by George Polya
(1887-1985), one of the great problem solvers of twentieth-century mathematics (see
the Historical Quest in Problem 61 of Section 7.4).
Optimization Procedure
Step 1 . Draw a figure (if appropriate) and label all quantities relevant to the
problem.
Step Focus on the quantity to be optimized. Name it. Find a formula for the
quantity to be maximized or minimized.
Step Use conditions in the problem to eliminate variables in order to ex¬
press the quantity to be maximized or minimized in terms of a single
variable.
Step Find the practical domain for the variables in Step 3; that is, the inter¬
val of possible values determined from the physical restrictions in the
problem.
Step If possible, use the methods of calculus to obtain the required optimum
value.
Solution
A picture of the play zone is shown in Figure 4.51. Let x and y denote the length and
width of the inscribed rectangle. The appropriate formula for the area is
A = (length) (width) = xy
CsmhT>
See Section 1.5 of the Student
y 3X
Mathematics Handbook for similar triangles.
4.6 Optimization in the Physical Sciences and Engineering 239
The domain for A is 0 < x < 12. The critical numbers for A are values such that
A'(x) = 0 (since A'(a) exists for all a). Since
A'(x) = 4 — |x
the only critical number is a = 6. Evaluate A (a) at the endpoints and the critical
number:
A(6) = 4(6) — |(6)2 = 12; A(0) = 0; A(12) = 0
y = 4-i(6) = 2
Thus, the largest rectangular play zone that can be built in the triangular plot is a
rectangle 6 m long and 2 m wide. ■
For what dimensions does the box have the greatest possible volume?
Solution
If each square comer cut out has side a, the box will be a inches deep, 45 — 2a inches
long, and 24 - 2a inches wide. The volume of the box shown in Figure 4.52 is
and this is the quantity to be maximized. To find the domain, we note that the dimen¬
sions must all be nonnegative; therefore, a > 0, 45 — 2x > 0 (or a < 22.5), and
24 — 2a > 0 (or a < 12). This implies that the domain is [0, 12].
To find the critical numbers (the derivative is defined everywhere in the domain),
we find values for which the derivative is 0:
The critical numbers are a = 5 and a = 18, but a = 18 is not in the domain, so the
only relevant critical number is a = 5. Evaluating V(x) at the critical number a = 5
and the endpoints a = 0, a = 12, we find
Thus, the box with the largest volume is found when a = 5. Such a box has dimensions
5 in. x 14 in. x 35 in.
240 Chapter 4 Additional Applications of the Derivative
You will note in Example 2 we did not have to test whether the critical point was
a maximum or a minimum, but knew that it was a maximum because the continuous
function V is nonnegative on the interval [0, 12] and is indeed zero at the endpoints.
Since there is only one critical number x — 5, that must yield the maximum. Similar
reasoning can be used in many examples, allowing us to forgo having to test for a
maximum or minimum with the first- or second-derivative test. We summarize this
procedure with the following definition.
EVT Convention If the hypotheses of the extreme value theorem apply to a particular problem, we
can often avoid testing a candidate to see whether it is a maximum or a minimum.
Suppose we have three candidates and two are obviously minima—the third must
be a maximum. We will refer to this principle of not testing a candidate as the
EVT convention because it is shorthand for using the extreme value theorem.
Solution
We note that
kcosO
1(0) =
Figure 4.53 Illumination on a
kitchen table where k is a (positive) constant of proportionality. Moreover, from the right triangle
shown in Figure 4.53,
.4 4
sin 9 = - or d —-
d sin#
Hence,
Evaluating 1(9) at the endpoints, 6 = 0 and 9 = y, and at the interior critical number
tan-1 yfl ~ 0.9553, we find that
k ,
/ (0) = —(cosO)(sin- 0) = 0
16
Finally, to find the height h that maximizes the illumination / , observe from Figure
4.53 that
4
tan# = -
h
4 4
h = 2.83
tan 9 s/2
To maximize the illumination, the lamp should be placed about 2.8 ft (2 ft 10 in.) above
the table. ■
Solution
Suppose the driver heads for a point C located * km down the road from B toward her
destination, as shown in Figure 4.54. We want to minimize the time. We will need to
d
remember the formula d = rt, or in terms of time, t = -.
buggy r
The domain of T is [0, 50]. Next, find the derivative of time T with respect to x:
rw - h ~(x2 + 1,600)“1/2(2jc)
45y*2 + l,600 75
5* — 3-y/x2 + 1,600
2257*2 + l,600
5* — 37^+600 = 0
242 Chapter 4 Additional Applications of the Derivative
Solving this equation, we find x = 30 (-30 is extraneous). Evaluating 7(x) here and
at the endpoints, we find that
/302 + 1,600 50 - 30
7(30) ~ 1.3778 hr % 83 min
45 + 75
JO2 + 1,600 50 -0
7(0) - ^ 1.5556 hr % 93 min
45 75
v/502 + 1,600 50 - 50
7(50) ~ 1.4229 hr % 85 min
The driver can minimize the total driving time by heading for a point that is 30 miles
from the point B and then traveling on the road to point D. She wins the prize because
this minimal route requires only 83 minutes. ■
Solution
To understand the problem we draw a sketch, as shown in Figure 4.55, and label the
radius of the circle r and the side of the square s.
\Wire, length L
Figure 4.55 Forming a circle and a square from a wire with length L
We need to write the radius r and the side s in terms of the length of wire, L.
= 2jzr + 4s
so 5 = |(L — 2nr). (We could just as easily have solved for r.) By substitution,
To find the domain, we note that r > 0 and that L - 2nr > 0, so the domain is
L 7
0 < r < —. Note when r = 0 there is no circle, and when r = — there is no square.
The derivative of A (r) is
7
Solve A'(r) — 0 to find r =
2(7T + 4) ’
4.6 Optimization in the Physical Sciences and Engineering 243
Thus, the extreme values of the area function on 0 must occur either at the
' 2n
\1 H-1 r L - 2tt ( —l w L2
£
= 71
i / 16 \2xJ_ 4n
1
A - TC 27T
27r + 8 2jc T 8 + 16 2tt T 8 4(7r + 4)
L
Comparing these values, we see that the smallest area occurs at r - and the
27r + 8
largest area occurs at x = L/{2ix). To summarize,
1. To maximize the sum of the areas, do not cut the wire at all. Bend the wire to form
L
a circle of radius r = —.
2tz
2ttL
2. To minimize the sum of the areas, cut the wire at the point located 2irr =
2tt T 8
ttL
units from one end, and form the circular part with resulting radius
71 + 4
L
27r + 8
Figure 4.56 Graph of If you use a graphing calculator or a computer, you can verify this result (for
A{r) = nr2 + -^(1 — 2nr)2 L = 1) by looking at the graph of the area function, as shown in Figure 4.56. ■
Solution
In Figure 4.57, 6 is the angle whose vertex occurs at the observer’s eyes at a point
O located jc feet from the wall. Note that a is the angle between the line O A drawn
horizontally from the observer’s eyes to the wall and the line OB from the eyes to
the bottom edge of the painting. In AOAB, the angle at O is a, with cot a = In
AO AC, the angle at O is (a + 9), and cotter + 6) = f. By using the definition of
inverse cotangent, we have
Figure 4.57 The angle 9 is subtended
by the observer’s eye 9 = (a + 9) - a = cot-1 - cot-1 (^j
To maximize 9, we first compute the derivative:
-5 2
25 +x2 + 4 + x2
-3x2 + 30 = 0
x = ±>/io
Because distance must be nonnegative, we reject the negative value. We apply the first-
derivative test to show that the positive critical number </l0 corresponds to a relative
maximum (verify). Thus, the angle 9 is maximized when the observer stands vld) ft
away from the wall. ®
244 Chapter 4 Additional Applications of the Derivative
Light travels at different rates in different media; the more optically dense the medium,
the slower the speed of transit. Consider the situation shown in Figure 4.58, in which
a beam of light originates at a point A in one mediuni, then strikes the upper surface
of a second, denser medium at a point P, and is refracted to a point B in the second
medium.
Figure 4.58 The path of a light beam through two media of different density
Suppose light travels with speed v\ in the first medium and speed v2 in the second.
What can be said about the path followed by the beam of light?
Our method of investigating this question is based on the following optical prop¬
erty.
This problem is very similar to the dune buggy problem of Example 4: Light
minimizes the time, and determining minimum time was the goal of that example. In
Figure 4.58, let a — |AC|, b = |CD|, and c = |DB|, and let x denote the distance
from C to P. Because A and B are fixed points, the path APB is determined by the
location of P, which, in turn, is determined by x. Because iq is a given constant, the
time required for the light to travel from A to P is given by
AP \Ja1 + x2
Vi vl
and the time required for the light to go from P to B is
PB s/b2 + (c — x)2
V2 l>2
for another given constant v2. Therefore, the total time of transit is
Note from Figure 4.58 that if a is the angle of incidence of the beam of light and /) is
the angle of refraction, then (from the definition of sine)
x c — X
sin cz = , and sin B = ■
Va2 + *2 V^2 + (c - x)2
Therefore (by substitution), we see the derivative of T can be expressed as
dT sin a sin/)
dx V\ l>2
11. Each edge of a square has length L, as shown in Figure 4.59. the camera be placed to photograph the mural with the largest
possible angle?
8 ft
13 ft
Figure 4.59 Square in a square
14 cm
23. Spy Problem It is noon. The Spy has returned from space
(Problem 63 of Section 3.4) and is driving a jeep through the
sandy desert in the tiny principality of Alta Loma. He is 32
km from the nearest point on a straight, paved road. Down the
road 16 km is a power plant in which a band of international
terrorists has placed a time bomb set to explode at 12:50 P.M. Figure 4.65 Maximizing the amount of light
The jeep can travel at 48 km/h on the sand and at 80 km/h on
the paved road. If he arrives at the power plant in the short¬ The rectangular part of the window is of clear glass and trans¬
est possible time, how long will our hero have to defuse the mits twice as much light per square foot as the triangular part,
bomb? which is made of stained glass. If the entire window has a
perimeter of 20 ft, find the dimensions (to the nearest ft) of
the window that will admit the most light.
The distance / from the center O of the lens to the point la¬
beled F is called the focal length of the lens.
a. Using similar triangles, show that
1 1 _ 1
p q f
Figure 4.64 Fabricating a cone from a sheet of tin is the largest value of / for which this condition can be
satisfied?
K(l + c2x3)
ECx) =
(1 +V)3
32. Modeling Problem In crystallography, a fundamental problem miles above the center of the earth t minutes after achieving
is the determination of the packing fraction of a crystal lattice, orbit.
which is defined as the fraction of space occupied by the atoms
a. Sketch the graph of r(t).
in the lattice, assuming the atoms are hard spheres. When the
b. What are the lowest and highest points on the satellite’s
orbit? These are called the perigee and apogee positions,
respectively.
36. The theory of relativity models the Doppler shift 5 of an object 39. It is known that water expands and contracts according to its
traveling at a velocity v: temperature. Physical experiments suggest that an amount of
water that occupies 1 liter at 0°C will occupy
cTv
s - 1
V(T) = 1-6.42 x 10“5T
cut
20-x
cos 26 = 2 cos2 6 — 1
Figure 4.72 Constructing a pyramid
to eliminate 6 and express L in terms of x alone. What is the largest volume of the pyramid that can be formed
38. Find the length of the longest pipe that can be carried hori¬ in this manner?
zontally around a comer joining two corridors that are 2y/2 ft 42. Modeling Problem A universal joint is a coupling used in cars
wide. Hint: Show that the length L can be written as and other mechanical systems to join rotating shafts together
at an angle. In a model designed to explore the mechanics of
2V2 2V2 the universal joint mechanism, the angular velocity /3(f) of the
L(6) = —- T--
sin 6 cos 6 output (driven) shaft is modeled by the formula
where a (in rad/s) is the angular velocity of the input (driving) a. At the interface points, A, B, and C, the light beam is de¬
shaft, and y is the angle between the two shafts, as shown in flected (from a straight line path) and loses intensity. For
Figure 4.73.* instance, at A, the incoming beam is deflected through a
clockwise rotation of (a — ft) radians. Show that the total
a rad/sec deflection (at all three interfaces) is
|3 rad/sec
D = n + 2a — 4/3
Rain drop
Incoming light a
ECONOMICS
Maximizing Profit
Solution
Let x denote the number of $1 price increases, and let P(x) represent the corresponding
profit.
Recall that 4,000 pairs of earrings are sold each month when the price is $5 per pair
and 400 fewer pairs will be sold each month for each added dollar in the price. Thus,
Knowing that the price per pair is 5 + a, we can now write the profit as a function of a:
= (4,000 - 400a)[(5 + a) - 3]
= 400(10 -a)(2 + a)
To find the domain, we note that a > 0. And 400(10 — a), the number of pairs sold,
should be nonnegative, so a < 10. Thus, the domain is [0, 10].
The critical numbers are found when the derivative is 0 {P is a polynomial func¬
tion, so there are no values for which the derivative is not defined):
The critical number is a = 4 and the endpoints are a = 0 and a = 10. Checking for
the maximum profit,
The maximum possible profit is $14,400, which will be generated if the earrings are
Figure 4.76 The profit function P(x) sold for $5.00 + $4.00 = $9.00 per pair. The graph of the profit function is shown in
Figure 4.76. ■
Sometimes the function to be optimized has practical meaning only when its in¬
dependent variable is a positive integer. Such functions are said to have a discrete
domain and are called discrete functions. Technically, the methods of calculus cannot
be applied to discrete functions because the theorems we have developed apply only
to continuous functions. However, useful information about a discrete function f (n)
can often be obtained by analyzing the related continuous function /(a) obtained by
replacing the discrete variable n by a real variable a. Here is an example that illustrates
the general procedure.
252 Chapter 4 Additional Applications of the Derivative
/ • Solution
(• f
• The domain of P{n) is the set of integers n between 10 and 40. To optimize this
—7 I
•• function, we apply the methods developed in this chapter to the related continuous
function
P(x) = -x3 + 27.6x2 + 970.2x - 4,235
i0 20 30 40 *
where x is a real variable defined on the interval 10 < x < 40. To determine the
critical numbers of P(x), we solve
and obtain x = 29.4 and x = —11. Only x = 29.4 is in the domain, and by testing the
sign of P'(x) on each side of this critical number, we find that the continuous function
P(x) is increasing for 0 < x < 29.4 and decreasing for 29.4 < x < 40. Therefore,
the original discrete function P(n) is maximized either at n = 29 or at n — 30. Since
we conclude that the maximum profit is $22,723.40, and occurs when 29 people book
Figure 4.77 Discrete and continuous the tour. The graphs of P(n) and the related continuous function P{x) are shown in
graphs of the profit function P Figure 4.77. ■
In Section 3.8, we worked with marginal quantities and their role as rates of
change, namely, C'(x), the marginal cost, and R'(x), the marginal revenue.*
We will now consider these functions in optimization problems. For example, the
manufacturer certainly would like to know what level of production results in maxi¬
mum profit. To solve this problem, we want to maximize the profit function
Thus, P\x) — 0 when R {x) = C'(x), and by using economic arguments we can show
that a maximum occurs at the corresponding critical point.
Maximum Profit
Profit is maximized when marginal revenue equals marginal cost.
and all units can be sold at a price of p(x) = 49 - jc dollars per unit. Determine the
price that corresponds to the maximum profit.
Solution
The marginal cost is C\x) = |x + 4. The revenue is
R(x) = xp(x)
= x(49-x)
— 49x — x2
R\x) = C\x)
49 — 2x = +4
x = 20
‘Recall (from Section 3.8) that marginal cost is the instantaneous rate of change of production cost C with
respect to the number of units x0 produced and the marginal revenue is the instantaneous rate of change of
revenue with respect to the number of units x0 produced.
Chapter 4 Additional Applications of the Derivative
To justify this second business principle, find the derivative of the average cost
function:
C(x)
C'(x) C'(x) — A(x)
xC'(x) — C(x) X
A'(x) =
X X
Thus, A'(x) — 0 when C'(x) = A(x). Once again, economic theory can be used to
justify this result as a minimum.
and they can all be sold at a price of p(x) = 49 — x dollars per unit. Determine the
level of production at which the average cost is minimized.
Solution
C(x)
The average cost is A(x) = ±4±200x“‘ and A(x) is minimized when
x
C'(x) = A(x). Thus,
^x T 4 — ^x T 4 T 200x
If we disregard the negative solution, it follows that the minimal average cost occurs
when x = 40 units.
You might have noticed that we did not need the information p(x) = 49 — x in
arriving at the solution to this example. When doing real-world modeling, you must
often make a choice about which parts of the available information are necessary to
solve the problem. ■
BUSINESS MANAGEMENT
An Inventory Model For each shipment of raw materials, a manufacturer must pay
an ordering fee to cover handling and transportation. When the raw materials arrive,
they must be stored until needed and storage costs result. If each shipment of raw ma¬
terials is large, few shipments will be needed and ordering costs will be low, but storage
costs will be high. If each shipment is small, ordering costs will be high because many
shipments will be needed, but storage cost will be low. Managers want to determine
the shipment size that will minimize the total cost. Here is an example of how such a
problem may be solved using calculus.
Solution
We begin by writing the cost function:
Iml^ent We need to find an expression for each of these unknowns. Assume that the same
Cdnstantiinvent aryjof | batteries
number of batteries must be ordered each time an order is placed; denote this number
"Y
by x so that C(x) is the corresponding total cost. The average number of batteries in
storage during the year is half of a given order (that is, x/2), and we assume that the
total yearly storage cost is the same as if the x/2 batteries were kept in storage for the
entire year. This situation is shown in Figure 4.79.
Figure 4.79 Inventory graph
/ AVERAGE NUMBER /COST OF STORING 1
STORAGE COST =
yIN STORAGE PER YR \ BATTERY FOR 1 YR
= 0.48x
To find the total ordering cost, we can multiply the ordering cost per shipment by the
number of shipments. We also note that because 6,000 batteries are ordered during
the year and because each shipment contains x batteries, the number of shipments is
6,000/x.
120,000
= (20)
x
= 6,000(0.25) = 1,500
120,000
C(x) = 0.48x +---+ 1,500
x
The goal is to minimize C(x) on (0, 6000]. To obtain the critical numbers, we find
the derivative
C\x) = 0.48 - 120,000x“2
0.48 - 120,000x~2 = 0
0.48jc2 = 120,000
x = ±500
The root x = —500 does not lie in the interval (0, 6000], It is easy to check that C is
Figure 4.80 The total cost function decreasing on (0, 500) and increasing on (500, 6000], as shown in Figure 4.80. Thus,
C(x) = 0.48.r + 120.000.x-1 the absolute minimum of C on the interval (0, 6000] occurs when x = 500, and we
+ 1,500 conclude that to minimize cost, the manufacturer should order the batteries in lots of
500. ■
Chapter 4 Additional Applications of the Derivative
Optimal Holding Time Even with an asset that increases in value, there often comes
a time when continuing to hold the asset is less advantageous than selling it and invest¬
ing the proceeds of the sale. How should the investor decide when to sell? One way is
to hold the asset until the time its present value at the current prevailing rate of interest
is maximized. In other words, hold until today’s dollar equivalent of the selling price
is as large as possible, and then sell. Here is an example that illustrated this strategy.
Solution
The present value of the asset in t years is modeled by the function P{t) = V(t)e~rt,
where r is the annual interest rate and t is the time in years. Thus,
Pit) = lO.OOOe^e-008'
PHYSIOLOGY
Calculus can be used to model a variety of situations in the biological and life sciences.
We will consider two examples from physiology. In the first, we use a model for the
concentration of a drug in the bloodstream of a patient to determine when the maximum
concentration occurs.
b —a
where a, b (with b > a), and k are positive constants that depend on the drug.* At
what time does the largest concentration occur? What happens to the concentration as
t +oo?
Solution
To locate the extrema, we solve C'(t) = 0.
k
b—a b —a
*E. Heinz, “Probleme bei der Diffusion kleiner Substanzmengen innerhalb des menschlichen Korpers,”
Biochem., Vol. 319 (1949), pp. 482-492.
4.7 Optimization in Business, Economics, and the Life Sciences 257
—at
be~bt ae
b
eb,e-at = ebt~at
a
bt — at In — Definition of logarithm
a
t
b — a a
The second-derivative test can be used to show that the largest value of C(t) occurs at
this t (see Problem 49).
To see what happens to the concentration as t -> +oo, we compute the limit
k
lim C(t) = lim -[e~at — e~bl]
«->+oo r-»-+oo b —a
k 1
lim — lim —
b — a t—»+oo eat ?-»+oo e°t
k
[0-0]
b — a
radius Optimal Angle for Vascular Branching The blood vascular system operates in such
r a way that the circulation of blood—from the heart, through the organs of the body,
radius and back to the heart—is accomplished with as little expenditure of energy as possible.
Thus, it is reasonable to expect that when an artery branches, the angle between the
“parent” artery and its “daughter” should minimize the total resistance to the flow of
blood. Figure 4.82 shows a small artery of radius r branching from a larger artery of
radius R. Blood flows in the direction of the arrows from point A to the branch at B
and then to points C and D. For simplicity, we assume that C and D are located in
Figure 4.82 Vascular branching
such a way that CD is perpendicular to the main line through A, B, and D. We wish
to find the value of the branching angle 6 that minimizes the total resistance to the
flow of blood as it moves from A to B and then to point C, which is located a fixed
perpendicular distance h from the line through A and B.* (See Figure 4.82.)
*The key to solving this problem is a result due to work of the nineteenth-century French physiologist
Jean Louis Poiseuille (1789-1869). Our discussion of vascular branching is adapted from Introduction
to Mathematics for Life Scientists, 2nd edition, by Edward Batschelet (New York: Springer-Verlag, 1976,
pp. 278-280). In this excellent little book, Batschelet develops a number of interesting applications of
calculus, several of which appear in the problem set.
258 Chapter 4 Additional Applications of the Derivative
where r is the radius of the smaller artery, k is a viscosity constant, s\ = |Aft|, and
s2 = |ftC'|. Thus, the total resistance to flow may be modeled by the sum
ks\ ks2 , / Si So \
f=h + h = ^4 + ~pr =k (** + ~i)
The next task is to write / as a function of 6. To do this, reconsider Figure 4.82 by
labeling s\, s2, h, and t as shown in Figure 4.83.
We want to find equations for s\ and s2 in terms of h and 6; to this end we notice
that
C h
sin 0 = — tan0 =
$2 l-s\
h
S2 — l-s{ =
sin 9 tan#
= h esc 6 s, = € -
tan#
= l — h cot #
Figure 4.83 Minimizing the
resistance to blood flow We can now write f as a function of #:
h cot # h esc #'
+ A
R4
To minimize /, we need to find the critical numbers (remember that /? and k are positive
constants):
CSC # cot #
= 0
R4 A
CSC # cot #
R4 ~ ~r^
1 cos#
/?4sin# r4sin#
= cos #
R4
By finding the second derivative and noting that both r and R are positive, we can show
that any value #,„ that satisfies this equation yields a value /(#„,) that is a minimum.
Thus, the required optimal angle is #,„ = cos-1 (r4/ft4).
4.7 Optimization in Business, Economics, and the Life Sciences 259
6. A toy manufacturer produces an inexpensive doll (Flopsy) and 12. Some psychologists model a child’s ability to memorize by a
function of the form
an expensive doll (Mopsy) in units of x hundreds and y hun¬
dreds, respectively. Suppose it is possible to produce the dolls f/lnZ + 1 if 0 < z < 4
in such a way that s(')={, if,=o"
82 - lOx where t is time, measured in years. Determine when the
0 < x < 8
10 — x largest and smallest values of g occur.
13. It is estimated that t years from now the population of a certain
and that the company receives twice as much for selling a country will be
Mopsy doll as for selling a Flopsy doll. Find the level of pro¬ 160
duction for both x and y for which the total revenue derived Pit) =
1 + 8<?-°-01'
from selling these dolls is maximized. What vital assumption million. When will the population be growing most rapidly?
must be made about sales in this model? Qu. Modeling Problem The owner of the Pill Boxx drugstore ex¬
7. A business manager estimates that when p dollars are charged pects to sell 600 bottles of hair spray each year. Each bottle
for every unit of a product, the sales will be x = 380 — 20p costs $4, and the ordering fee is $30 per shipment. In addi¬
units. At this level of production, the average cost is modeled tion, it costs 90j( per year to store each bottle. Assuming that
by the hair spray sells at a uniform rate throughout the year and
that each shipment arrives just as the last bottle from the previ¬
*«-5+s5 ous shipment is sold, how frequently should shipments of hair
a. Find the total revenue and total cost functions, and express spray be ordered to minimize the total cost?
the profit as a function of x. 15. Modeling Problem An electronics firm uses 18,000 cases of
connectors each year. The cost of storing one case for a year is
b. What price should the manufacturer charge to maximize
$4.50, and the ordering fee is $20 per shipment. Assume that
profit? What is the maximum profit?
the connectors are used at a constant rate throughout the year
8. Suppose a manufacturer estimates that, when the market price and that each shipment arrives just as the preceding shipment
of a certain product is p, the number of units sold will be has been used up. How many cases should the firm order each
time to keep total cost to a minimum?
x = 16. Suppose the total cost (in dollars) of manufacturing x units of
a certain commodity is
It is also estimated that the cost of producing these x units will C(x) = 3x2 + 5x + 75
be
a. At what level of production is the average cost per unit the
C(x) = 4xe~*/6 + 30
smallest?
a. Find the average cost, the marginal cost, and the marginal b. At what level of production is the average cost per unit
revenue for this production process. equal to the marginal cost?
b. What level of production x corresponds to maximum c. Graph the average cost and the marginal cost functions on
profit? the same set of axes, for x > 0.
260 Chapter 4 Additional Applications of the Derivative
17. Suppose the total revenue (in dollars) from the sale of x units 23. A tour agency is booking a tour and has 100 people signed
of a certain commodity is up. The price of a ticket is $2,000 per person. The agency
has booked a plane seating 150 people at a cost of $125,000.
R(x) = — 2x2 + 68x — 128 Additional costs to the agency are incidental fees of $500 per
person. For each $10 that the price is lowered, a new person
a. At what level of sales is the average revenue per unit equal
will sign up. How much should the price be lowered for all
to the marginal revenue?
participants to maximize the profit to the tour agency?
b. Verify that the average revenue is increasing if the level of
24. A bookstore can obtain the best-seller 20,000 Leagues Under
sales is less than the level in part a and decreasing if the
the Majors from the publisher at a cost of $6 per book. The
level of sales is greater than the level in part a.
store has been offering the book at a price of $30 per copy and
c. On the same set of axes, graph the relevant portions of the
has been selling 200 copies per month at this price. The book¬
average and marginal revenue functions.
store is planning to lower its price to stimulate sales and esti¬
18. A manufacturer finds that the demand function for a certain
mates that for each $2 reduction in the price, 20 more books
product is will be sold per month. At what price should the bookstore
73
x(p) -- — sell the book to generate the greatest possible profit?
sfP
25. A Florida citrus grower estimates that if 60 orange trees are
Should the price p be raised or lowered to increase consumer planted, the average yield per tree will be 400 oranges. The
expenditure? Explain your answer. average yield will decrease by 4 oranges for each additional
19. Suppose you own a rare book whose value t years from now is tree planted on the same acreage. How many trees should the
modeled as 300c^. If the prevailing rate of interest remains grower plant to maximize the total yield?
constant at 8% compounded continuously, when will be the 26. Farmers can get $2 per bushel for their potatoes on July 1, and
most advantageous time to sell? after that the price drops 2<f. per bushel per day. On July 1,
20. Suppose you own a parcel of land whose value t years from a farmer has 80 bushels of potatoes in the field and estimates
now is modeled as that the crop is increasing at the rate of 1 bushel per day. When
should the farmer harvest the potatoes to maximize revenue?
P(t) - 200 In y/lt
27. A viticulturist estimates that if 50 grapevines are planted per
thousand dollars. If the prevailing rate of interest is 10% com¬ acre, each grapevine will produce 150 lb of grapes. Each ad¬
pounded continuously, when will be the most advantageous ditional grapevine planted per acre (up to 20) reduces the av¬
time to sell? erage yield per vine by 2 lb. How many grapevines should be
21. A store has been selling skateboards at the price of $40 per planted to maximize the yield per acre?
board, and at this price skaters have been buying 45 boards a 28. A commuter train carries 600 passengers each day from a sub¬
month. The owner of the store wishes to raise the price and es¬ urb to a city. It now costs $5 per person to ride the train. A
timates that for each $1 increase in price, 3 fewer boards will study shows that 50 additional people will ride the train for
be sold each month. If each board costs the store $29, at which each 25j? reduction in fare. What fare should be charged to
price should the store sell the boards to maximize profit? maximize total revenue?
22. Modeling Problem As more and more industrial areas are con¬ 29. Modeling Problem To raise money, a service club has been col¬
structed, there is a growing need for standards ensuring con¬ lecting used bottles, which it plans to deliver to a local glass
trol of the pollutants released into the air. Suppose that the company for recycling. Since the project began 80 days ago,
pollution at a particular location is based on the distance from the club has collected 24,000 pounds of glass, for which the
the source of the pollution according to the principle that for glass company offers 1 <f. per pound. However, because bottles
distances greater than or equal to 1 mi, the concentration of are accumulating faster than they can be recycled, the com¬
particulate matter (in parts per million, ppm) decreases as the pany plans to reduce the price it will pay by 1 per 100 pounds
reciprocal of the distance from the source. This means that of used glass.
if you live 3 mi from a plant emitting 60 ppm, the pollu¬
a. What is the most advantageous time for the club to con¬
tion at your home is y = 20 ppm. On the other hand, if
clude its project and deliver all the bottles?
you live 10 mi from the plant, the pollution at your home is
yjj = 6 ppm. Suppose that two plants 10 mi apart are releas¬ b. What assumptions must be made in part a to solve the prob¬
ing 60 and 240 ppm, respectively. At what point between the lem with the given information?
plants is the pollution a minimum? Where is it a maximum? 30. Suppose that the demand function for a certain commodity is
expressed as
/120 — x
60 ppm 240 ppm p{x) - J ——— for 0 < x < 120
|« x mi *{« 10-x mi b. Graph the relevant portions of the demand and revenue
functions.
4.7 Optimization in Business, Economics, and the Life Sciences 261
31. Suppose the demand function for a certain commodity is lin¬ was developed by A. Lasota that involved the cell production
ear, that is, function
P(x) = Axse~sx,r
b —x
p(x) —- for 0 < x < b where A, s, and r are positive constants and x is the number
a
of granulocytes (a type of white blood cell) present.*
where a and b are positive constants. a. Find the granulocyte level x that maximizes the production
function P. How do you know it is a maximum?
a. Find the total revenue function explicitly and use its first
derivative to determine its intervals of increase and de¬ b. If 5 > 1, show that the graph of P has two inflection points.
crease. Sketch the graph of P and give a brief interpretation of the
inflection points.
b. Graph the relevant portions of the demand and revenue
functions. c. Sketch the graph for the case where 0 < s < 1. What is
different about this case?
32. Modeling Problem Homing pigeons will rarely fly over large
36. Modeling Problem During a cough, the diameter of the trachea
bodies of water unless forced to do so, presumably because
decreases. The velocity v of air in the trachea during a cough
it requires more energy to maintain altitude in flight over the
may be modeled by the formula
cool water. Suppose a pigeon is released from a boat floating
on a lake 3 mi from a point A on the shore and 10 mi away v — Ar2(ro — r)
from the pigeon’s loft, as shown in Figure 4.84.
where A is a constant, r is the radius of the trachea during
the cough, and r0 is the radius of the trachea in a relaxed state.
Find the radius of the trachea when the velocity is greatest, and
find the maximum velocity of air. (Notice that 0 < r < ro.)
37. Modeling Problem The work of V. A. Tucker and K. Schmidt-
Koenig+ models the energy expended in flight by a certain kind
of bird by the function
1 9
10 E = -[a(v — b)2 + c]
v
where a, b, and c are positive constants, v is the velocity of the
Figure 4.84 Flight path for a pigeon bird, and the domain of v is [16, 60]. Which value of v will
minimize the energy expenditure in the case where a = 0.04,
b = 36, and c = 9?
Assuming the pigeon requires twice as much energy to fly over 38. Modeling Problem A plastics firm has received an order from
water as over land and it follows a path that minimizes total the city recreation department to manufacture 8, 000 special
energy, find the angle 9 of its heading as it leaves the boat. styrofoam kickboards for its summer swimming program. The
(This is the situation in the Perspective section of the chapter firm owns 10 machines, each of which can produce 50 kick-
introduction.) boards per hour. The cost of setting up the machines to pro¬
duce the kickboards is $800 per machine. Once the machines
33. Modeling Problem In an experiment, a fish swims ^ meters up¬
have been set up, the operation is fully automated and can be
stream at a constant velocity v m/sec relative to the water,
overseen by a single production supervisor earning $35 per
which itself has velocity Vi relative to the ground. The results
hour.
of the experiment suggest that if the fish takes t seconds to
a. How many machines should be used to minimize the cost
reach its goal (that is, to swim s meters), the energy it expends
of production?
may be modeled by
b. How much will the supervisor earn during the production
E = cvkt run if the optimal number of machines is used?
39. Modeling Problem After hatching, the larva of the codling moth
where c > 0 and k > 2 are physical constants. Assuming iq goes looking for food. The period between hatching and find¬
is known, what velocity v minimizes the energy? ing food is called the searching period. According to a model
34. In a learning model, two responses (A and B) are possible for developed by P. L. Shaffer and H. J. Gold1 the length (in days)
each of a series of observations. II there is a probability p ot of the searching period is given by
getting response A in any observation, the probability of get¬
S(T) = (-0.03T2 + 1.67T - 13.67)-1
ting a response A exactly n times in a series of M observations
is
*See “A Blood Cell Population Model, Dynamical Diseases, and Chaos,”
by W. B. Gearhart and M. Martelli, UMAP Modules 1990: Tools for Teach¬
F{p) = pn(l-p)M-"
ing. Arlington, MA: Consortium for Mathematics and Its Applications
(CUPM) Inc., 1991.
The maximum likelihood estimate is the value of p that max¬
imizes F(p) on [0, 1]. For what value of p does this occur? ty A. Tucker and K. Schmidt-Koenig, “Flight of Birds in Relation to En¬
ergetics and Wind Directions,” The Auk. Vol. 88 (1971). pp. 97-107.
35. Modeling Problem The production of blood cells plays an im¬ tp. L. Shaffer and H. J. Gold, “A Simulation Model of Population Dynam¬
portant role in medical research involving leukemia and other ics of the Codling Moth Cydia Pomonella," Ecological Modeling, Vol. 30
so-called dynamical diseases. In 1977, a mathematical model (1985) pp. 247-274.
262 Chapter 4 Additional Applications of the Derivative
where T (°C) is the air temperature (20 < T < 30), and the 45. Modeling Problem A store owner expects to sell Q units of a
percentage of larvae that survive the searching period is certain commodity each year. It costs S dollars to order each
new shipment of x units; and it costs t dollars to store each
N(T) = -0.8572 + 45.457 - 547 unit for a year. Assuming that the comrrtodity is used at a con¬
stant rate throughout the year and that each shipment arrives
a. Sketch the graph of N and find the largest and smallest sur¬ just as the preceding shipment has been used up, show that
vival percentages for the allowable range of temperatures, the total cost of maintaining inventory is minimized when the
20 < T < 30. ordering cost equals the storage cost.
b. Find S'(T) and solve S'(T) = 0. Sketch the graph of 5 for 46. Modeling Problem Sometimes investment managers determine
20 < T < 30. the optimal holding time of an asset worth V (t) dollars by
c. The percentage of codling moth eggs that hatch at a given finding when the relative rate of change V'(t)/V{t) equals
temperature T is given by the prevailing rate of interest r. Show that the optimal time
determined by this criterion is the same as that found by max¬
H{T) = -0.53T2 + 25T - 209 imizing the present value of V (r).
47. An important quantity in economic analysis is elasticity of de¬
for 20 < T < 30. Sketch the graph of H and determine
mand, defined by
the temperatures at which the largest and smallest hatching
p dx
percentages occur. E(x) = ^ —
40. According to a certain logistic model, the world’s population x dp
(in billions) t years after 1960 is modeled by the function where x is the number of units of a commodity demanded
when the price is p dollars per unit.* Show that
1 + 12e-°'08'
dR r r i
dx x _ E(x)
a. If this model is correct, at what rate will the world’s popu¬
lation be increasing with respect to time in the year 2010? That is, marginal revenue is [1 + \/E{x)] times average rev¬
At what percentage rate will it be increasing at this time? enue.
b. Sketch the graph of P. What feature on the graph corre¬ 48. Modeling Problem There are alternatives to the inventory model
sponds to the time when the population is growing most analyzed in Example 5. Suppose a company must supply N
rapidly? What happens to P(t) as t —> +oo (that is, “in units/month at a uniform rate. Assume the storage cost/unit is
the long run”)? Si dollars/month and that the setup cost is S2 dollars. Further
assume that production is at a uniform rate of m units/month
41. Modeling Problem Generalize Problem 38. Specifically, a man¬
(with no units left over in inventory at the end of the month).
ufacturing firm received an order for Q units of a certain com¬
Let x be the number of items produced in each run.
modity. Each of the firm’s machines can produce n units per
hour. The setup cost is S dollars per machine, and the operat¬ a. Explain why the total average cost per month may be mod¬
ing cost is p dollars per hour. eled by
Six N \ | s2n
a. Derive a formula for the number of machines that should C(x) =
~1 m) x
be used to minimize the total cost of filling the order.
b. Show that when the total cost is minimal, the setup cost is b. Find an expression for the number of items that should be
equal to the cost of operating the machines. produced in each run in order to minimize the total average
cost C.
42. Modeling Problem (Continuation of Problem 39) Suppose you
have 10,000 codling moth eggs. Find a function F for the Economists refer to the optimum found in this inventory
number of moths that hatch and survive to have their first model as the economic production quantity (EPQ), whereas
meal when the temperature is T. For what temperature T on the optimum found in the model analyzed in Example 5 is
[20, 30] is the number of dining moths the greatest? How large called the economic order quantity (EOQ).
is the optimum dining party and how long did it take to find 49. In this section we showed that the Heinz concentration func¬
its meal? tion,
43. An epidemic spreads through a community in such a way that C(t) = -~~(e~at - e-b‘)
t weeks after its outbreak, the number of residents who have b —a
been infected is modeled by a function of the form for b > a has exactly one critical number, namely,
A In
tc —
m=
1 + Ce~kt
where A is the total number of susceptible residents. Show Find C"(t) and use the second-derivative test to show that a
that the epidemic is spreading most rapidly when half the sus¬ relative maximum for the concentration C(t) occurs at time
ceptible residents have been infected. t = tc.
50. Let f(9) be the resistance to blood flow obtained on page 257, minimizes the surface area of the cell (assuming that 5 and Ir
namely are fixed)?
h cot 6 h esc 6 \
f(8) = k + s
RA
We found that 9m — cos~'(r4//?4) is a critical number of /.
Find f"{6) and use the second-derivative test to verify that a
relative minimum occurs at 6 = 9m. Explain why this must be
an absolute minimum.
51. Beehives are formed by packing together cells that may be
modeled as regular hexagonal prisms open at one end, as
shown in Figure 4.85. It can be shown that a cell with hexag¬
onal side of length 5 and prism height h has surface area
for 0 < 9 < |. What is the angle 9 (to the nearest degree) that Figure 4.85 Beehive (left) and beehive cell (right)
Proficiency Examination Sketch the graph of each function in Problems 22-27. Include a1!
key features, such as relative extrema, inflection points, asymp¬
CONCEPT PROBLEMS totes, intercepts, cusps, and vertical tangents.
1. What is the difference between absolute and relative extrema 22. f{x) = x3 + 3x2 - 9.x + 2
of a function? X‘~ — 1
2. State the extreme value theorem. 23. fix) = .x1/3 (27 - x) 24. fix) = ——-
Xz — 4
3. What are the critical numbers of a function? What is the dif¬
25. fix) = ix2 - 3)e~x 26. fix) = x + tan1 x
ference between critical numbers and critical points?
27. f(x) = sin2x — 2cosx on [0. 2n]
4. Outline a procedure for finding the absolute extrema of a con¬
28. Determine the largest and smallest values of
tinuous function on a closed interval [a, b],
5. State both Rolle’s theorem and the mean value theorem, and
fix) — x4 — 2x? + 5
discuss the relationship between them.
6. What is the zero-derivative theorem? on the closed interval [0, 1].
7. State the first-derivative and second-derivative tests, and dis¬ 29. A box is to have a square base, an open top, and volume of 2
cuss when and in which order you would use them. ft3. Find the dimensions of the box (to the nearest inch) that
8. What is a vertical asymptote? A horizontal asymptote? uses the least amount of material.
9. How would you identify a cusp? A vertical tangent? 30. The personnel manager of a department store estimates that
10. State l'Hopital’s rule. if N temporary salespersons are hired for the holiday season,
the total net revenue derived (in hundreds of dollars) from their
11. Define lim fix) = L and lim fix) = -Foo.
*-»+oo x >c efforts may be modeled by the function
12. Outline a graphing strategy for a function defined by
R(N) = —3/V4 + 501V3 - 261/V2 + 540N
y = /(*)•
13. What do we mean by optimization? Outline an optimization
for 0 < N < 9. How many salespersons should be hired to
procedure.
maximize total net revenue?
14. What is Fermat’s principle of optics?
15. What is Snell’s law of refraction?
16. When is the profit maximized in terms of marginal revenue
Supplementary Problems
and marginal cost? When is the average cost minimized?
Sketch the graph of each function in Problems 1-20. Use as many
17. What is Poiseuille’s law of resistance to flow?
as possible of the key features such as domain, relative extrema,
inflection points, concavity, asymptotes, intercepts, cusps, and ver¬
PRACTICE PROBLEMS
tical tangents.
Evaluate the limits in Problems 18-21.
1. fix) = x3 + 6x2 + 9x - 1 2. fix) = x4 4- 4x3 + 4x2 + 1
sin 2x 1 — a/x
18. lim 19. lim 3. fix) = 3x4 - 4x3 + 1 4. fix) = 3x4 - 4x2 + 1
JC—vjr/2 COS* *->i .x - 1
3* 5. fix) = 6x5 - 15x4 + 10x3
20. lim I - 21. lim 6. fix) = 3x5 - 10x3 + 15x + 1
*-►+00
x-A+oo \X
264 Chapter 4 Review
15. fix) = x2 In y/x 16. fix) — sin-1 x + cos-1 x lim is/x2 — x — x) 32.
*—►-+-00 *—►-+-00
(x — 1
17. fix) = xie~2x + e~x) 18. fix) = In x sinx x sinx
vx + 1 lim-r— 34. n-r—
In Problems 21-24, the graph of the given function fix) for x > 0 sin2x
hm -—- 38.
is one of the six curves shown in Figure 4.86. In each case, match *-►0 sinx7 j—^0 \ X1,2 x2secx
the function to a graph. sec2 x
lim —-— 40.
In y/x * —>7t ji~ sec2 3x -V-.JT/2-
21. fix) = x2~x 22. fix)
lim (sec x —tan x)
ex *->7T/2_
23. fix) = - 24. fix) — e x sinx
x lim (\Jx2 + 4 — \/x2
a. b. *-►+00V -4)
y, 1
/ lim (1 +x)4/x
x^0+
44.
1 —X
// lim xtanx
*->o+
46.
x—>0
-1
X
*
lim-
*-*■0
ln(x2 + 1)
X
48. Um \x
fi'1
JT->+00
ex cos x — 1
lim
jt-»+oo
(4 — 50. lim -
*-►+00 x
51. Journal Problem Mathematics Teacher* ■ Which of the
c. graphs in Figure 4.87 is the derivative and which is the func¬
tion?
e.
55. Find numbers A, B, C, and D that guarantee that the function 63. Journal Problem AMATYC Review* m You are to build a rect¬
angular enclosure with front side made of material costing $10
f(x) = Ax} + Bx2 + Cx + D
per foot and with the other three sides (back, left, right) of ma¬
will have a relative maximum at (—1, 1) and a relative mini¬ terial costing $5 per foot. If the enclosure is to contain exactly
mum at (1, —1). 600 square feet and to be built at minimum total cost, how
long should the side be and what is the total cost?
56. A Norman window consists of a rectangle with a semicircle
surmounted on the top. What are the dimensions of the Nor¬ Answer: $600.
man window of largest area with a fixed perimeter of PQ me¬ Here is the real question: By coincidence, the answer matches
ters? the constrained area (both 600, aside from units, of course).
57. An apartment complex has 200 units. When the monthly rent Question: Characterize when this happens more generally, and
for each unit is $600, all units are occupied. Experience indi¬ thus describe the relationship that exists in such problems in
cates that for each $20-per-month increase in rent, 5 units will which the resultant minimum cost matches the given area (all
become vacant. Each rented apartment costs the owners of in appropriate units).
the complex $80 per month to maintain. What monthly rent 64. Journal Problem Quantum' ■ Two students are pondering the
should be charged to maximize the owner’s profit? What is following problem: A tin can takes the form of a right circular
maximum profit? How many units are rented when profit is a cylinder with radius R and height H. An ant is sitting on the
maximum? border circle of one of its bases (point A in Figure 4.88). It
58. A fanner wishes to enclose a rectangular pasture with 320 ft wants to crawl to the most distant point B at the border circle
of fence. Find the dimensions that give the maximum area in of the other base (symmetric to A with respect to the center of
these situations: the tin). Find the shortest path for the ant.
66. Modeling Problem A manufacturer receives an order for 5,000 6,200 1990 39,000
1985
items. There are 12 machines available, each of which can
produce 25 items per hour. The cost of setting up a machine 1986 10,660 1991 45,000
for a production run is $50. Once the machines are in opera¬
tion, the procedure is fully automated and can be supervised
by a single worker earning $20 per hour. Set up and analyze
a. To model the number of annual AIDS deaths, a data anal¬
a model to determine the number of machines that should be
ysis program produced the cubic polynomial
used to minimize the total cost of filling the order. State any
assumptions that must be made.
N{x) = -8.58197x3 + 732.727x2 - 3,189.9x + 4,375.09
67. Show that the graph of a polynomial of degree n, with n > 2,
has at most n — 2 inflection points.
Sketch the graph of N and determine the time when its
68. Show that the graph of the function /(x) = x" with n > 1 has
highest point occurs for 0 < x < 20, where x = 0 repre¬
either one or no inflection points, depending on whether n is
sents 1982.
odd or even.
69. Each tangent line to the circle x2 + y1 = 1 at a point in the first b. The same report gives the number of cases of reported
quadrant will intersect the coordinate axes at points {x\, 0) and AIDS for the period 1984-1991:
(0, >i). Determine the line for which xj + yq is a minimum.
70. An accelerated particle moving at speed close to the speed of
Year Deaths Year Deaths
light emits power P in the direction 9 given by
f. Based on the information you obtained in part e, find new 84. H1STORICA L Q UE> f Leonhard
models for the information in parts a and b using all the Euler is one of the giants in the history
data you have available. of mathematics. His name is attached
to almost every branch of mathemat¬
Using the graphing and differentiation programs of your computer
ics. He was the most prolific writer
or calculator, graph f(x), fix), f"(x) for each function in Prob¬
on the subject of mathematics, and his
lems 76-79. Print out a copy, if possible. On each graph, indicate
mathematical textbooks were master¬
where
fully written. His writing was not at
a. /'(x) >0 b. f'(x) <0 c. f"(x) > 0 all slowed down by his total blindness LEONHARD EULER
d. f"(x) < 0 e. f'(x) — 0 /. /'(x) does not exist for the last 17 years of his life. He 1707-1783
80. Consider a string 60 in. long that is formed into a rectangle. lim (l + iY
n-*+oo y n )
Using a graphing calculator or a graphing program, graph the
area A(x) enclosed by the string as a function of the length x Instead, Euler used series in his work (which we will study in
of a given side (0 < x < 30). From the graph, deduce that Chapter 8). ■
the maximum area is enclosed when the rectangle is a square. In this Historical Quest problem we will lay the ground¬
What is the minimum area enclosed? Compare this problem work for a Historical Quest in Chapter 8. Euler introduced
to the exact solution. What conclusions do you draw about loga x (which he wrote as lx) as that exponent y such that
the desirability of analytical solutions and the role of the com¬ ay = x. This was done in 1748, which makes it the first ap¬
puter? pearance of a logarithm interpreted explicitly as an exponent.
He does not define a0 = 1, but instead writes
81. Modeling Problem The beach of a lake follows contours that
are approximated by the curve 4x2 + y2 = 1, and a nearby a€ = 1 + ke
road lies along the curve y = 1/x for x > 0. Using your
for an infinitely small number e. In other words,
graphing calculator or computer software, determine the clos¬
est approach of the road to the lake in the north-south direc¬ a* - 1
k = lim-
tion. Take the positive y-axis as pointing north. e—o e
Explain why k — In a.
82. Using your graphing and differentiation programs, locate and When Euler was 13, he registered at the University of
identify all the relative extrema for the function defined by Basel and was introduced to another famous mathematician,
Johann Bernoulli, who was an instructor there at the time (see
fix) = ix5 - fx4 + 2x2
Historical Quest, Section 4.5, Problems 58-59). If Bernoulli
thought that a student was promising, he would provide, some¬
This may require judicious choices of the window settings.
times gratis, private instruction. Here is Euler's own account of
83. Modeling Problem Suppose you are a manager of a fleet of de¬ this first encounter with Bernoulli
livery trucks. Each truck is driven at a constant speed of x / soon found an opportunity to gain introduction to the fa¬
mi/h (15 < x < 55), and gas consumption (gal/mi) is mod¬ mous professor Johann Bernoulli, whose good pleasure it was
eled by the function to advance me further in the mathematical sciences. True,
because of his business he flatly refused me private lessons,
but he gave me much wiser advice, namely, to get some more
difficult mathematical books and work through them with all
industry, and wherever l should find some check or difficul¬
Using a graphing and differentiation program (and/or root¬ ties, he gave me free access to him every Saturday afternoon
solving program on your computer or calculator), answer the and was so kind as to elucidate all difficulties, which hap¬
following questions: pened with such greatly desired advantage that whenever he
a. If gas costs $1.70/gal, estimate the steady speed that will had obviated one check for me, because of that ten others dis¬
minimize the cost of fuel for a 500-mi trip. appeared right away, which is certainly the way to make a
happy advance in the mathematical sciences.*
b. Estimate the steady speed that minimizes the cost if the
driver is paid $28 per hour and the price of gasoline re¬ *From Elements of Algebra by Leonhard Euler, 1840. London: Longman,
mains constant at $1.70/gal. Orme, and Co. Reprinted by Springer-Verlag.
268 Chapter 4 Review
has only negative values for 0 < x < a. Hint: Show that if
x = g(l — y), the polynomial becomes a6y2g(y), where
A wine barrel has a hole in the middle of its side called a bung hole.
To determine the volume of wine in the barrel, a bung rod is inserted
in the hole until it hits the lower seam. Determine how to calibrate
such a rod so that it will measure the volume of the wine in the barrel. You
should make the following assumptions:
Your paper is not limited to the following questions, but it should include these con¬
cerns: You should show that the volume of the cylindrical barrel is
V =27rA3r(4 + r2r3/2
and you should find the approximate ideal value for t. Johannes Kepler was the first
Johannes Kepler (1571-1630) is person to show mathematically why coopers were guided in their construction of wine
usually remembered for his work in barrels by one rule: make the staves (the boards that make up the sides of the barrel)
astronomy, in particular for his three one and one-halftimes as long as the diameter. (This is the approximate f-value.) You
laws of planetary motion. Tycho should provide dimensions for the barrel as well as for the bung rod.
Brahe (1546-1601) was working for
Rodolf II, Holy Roman Emperor in
Prague in 1599, and he asked Kepler
to work with him. The historian
Burton describes this as a fortunate
alliance. “Tycho was a splendid
observer but a poor mathematician,
while Kepler was a splendid
mathematician but a poor observer."
Because Kepler was a Protestant
during a time when most
intellectuals were required to be
Catholic, Kepler had trouble
supporting himself, and
consequently worked for many
benefactors. While serving the
Austrian emperor Matthew I, Kepler
observed with admiration the ability
of a young vintner to declare
quickly and easily the capacities of a
number of different wine casks. He
describes how this can be done in
his book The New Stereometry of
Wine Barrels. Mostly Austrian.
*The idea for this group research project comes from research done at Iowa State University as part of a
National Science Foundation grant. Our thanks to Elgin Johnston of Iowa State University.
*
.I—
CONTENTS
5.1 Antidifferentiation
Reversing differentiation
Antiderivative notation
Antidifferentiation formulas
Applications
Area as an antiderivative
5.2 Area as the Limit of a Sum
Area as the limit of a sum
The general approximation scheme
Summation notation
Area using summation formulas
5.3 Riemann Sums and the Definite Integral PREVIEW
Riemann sums
A traveler, driving a car down a straight, level road at 40 rni/h, applies the brakes,
The definite integral
Area as an integral which decelerate the car at the rate of 21 ft/s2. How far does the car travel before
Properties of the definite integral coming to a full stop? An ecologist measures the width of an oil spill at intervals of
Distance as an integral 5 ft along its length. What is the approximate area of the spill? On Christmas day,
5.4 The Fundamental Theorems of Calculus it starts snowing at a steady rate. A snowplow starts out at noon, going 2 mi during
The first fundamental theorem of the first hour and 1 mi during the second hour. When did it start to snow?
calculus
We will examine questions such as these in the examples and problems of this
The second fundamental theorem
of calculus chapter, as we study integral calculus, the companion to the differential calculus
5.5 Integration by Substitution developed in Chapters 3 and 4.
Substitution with indefinite
integration
Substitution with definite
integration PERSPECTIVE
5.6 Introduction to Differential Equations
The key concept in integral calculus is integration, a procedure that involves com¬
Introduction and terminology
Direction fields puting a special kind of limit of sums called the definite integral We will find that
Separable differential equations such limits can often be computed by reversing the process of differentiation; that is,
Modeling exponential growth and given a function /, we find a function F such that F' — f. This is called indefinite
decay integration, and the equation F' — f is an example of a differential equation.
Orthogonal trajectories
Modeling fluid flow through an
Finding integrals and solving differential equations are extremely important
orifice processes in calculus. We begin our study of these topics by defining definite and
Modeling the motion of a indefinite integration and showing how they are connected by a remarkable result
projectile: escape velocity called \ht fundamental theorem of calculus. Then we examine several techniques of
5.7 The Mean Value Theorem for Integrals; integration and show how area, average value, and other quantities can be set up and
Average Value analyzed by integration. Our study of differential equations begins in this chapter
Mean value theorem for integrals and will continue in appropriate sections throughout this text. We also establish a
Modeling average value of a
mean value theorem for integrals and develop numerical procedures for estimating
function
5.8 Numerical Integration: The Trapezoidal the value of a definite integral.
Rule and Simpson's Rule
Approximation by rectangles
Trapezoidal rule
Simpson’s rule
Error estimation
5.9 An Alternative Approach: The Logarithm
as an Integral
Natural logarithm as an integral
Geometric interpretation
The natural exponential function
Chapter 5 Review
Guest Essay: Kinematics of Jogging
Ralph Boas
_Cumulative Review, Chapters 1-5
271
272 Chapter 5 Integration
5.1 Antidifferentiation
IN THIS SECTION reversing differentiation, antiderivative notation, antidifferentiation formulas, applica¬
tions, area as an antiderivative
REVERSING DIFFERENTIATION
A physicist who knows the acceleration of a particle may want to determine its velocity
or its position at a particular time. An ecologist who knows the rate at which a certain
pollutant is being absorbed by a particular species of fish might want to know the
actual amount of pollutant in the fish’s system at a given time. In each of these cases, a
derivative f is given and the problem is that of finding the corresponding function /.
Toward this end, we make the following definition.
F\x) = f{x)
for all x in 1.
Suppose we know /(x) = 3x2. We wish to find a function F(x) such that
F'(x) — 3x2. It is not difficult to use the power rule in reverse to discover that
F(x) = x3 is such a function. However, that is not the only possibility:
G(x) = F(x) + C
^ What This Says Two antiderivatives of the same function differ by a con¬
stant.
Proof If F and G are both antiderivatives of /, then F' — f and G' = / and
Theorem 4.6 (the constant difference theorem) tells us that
G(x) - F(x) = C
so G(x) = F(x) + C. □
Solution
a. If F(x) = x6, then F'(x) = 6x5, so we see that a particular antiderivative of
Recall that the slope of a function y = /(x) at any point (x, y) on its graph is
given by the derivative f'(x). We can exploit this fact to obtain a “picture” of the
graph of /. Reconsider Example lc where / = 1/x. There is an antiderivative F(x)
of 1/x such that the slope of F at each point (x, F(x)) is 1/x for each nonzero value
of x. Let us draw a graph of these slopes:
If we continue to plot these slope points for different values of x, we obtain many little
slope lines. The resulting graph shown in Figure 5.1c is known as a slope field for
the equation / = 1/x. Beginning in Section 5.6 slope fields are also called direction
fields.
Finally, notice the relationship between the slope field for y' = 1/x and its an¬
tiderivative y = In |x| + C (found in Example lc). If we choose particular values for
C, say C = 0, C = - In 2, or C = 2, and draw these particular antiderivatives in
Figure 5.1c, we notice that these particular solutions are anticipated by the slope field
drawn in part b. That is, the slope field shows the entire family of antiderivatives of the
original equation.
/ C=0
c = -In 2
—C ■ : y
UJ
J
' i
Figure 5.2 Slope field for / = ex~ graph of the antiderivative of ex~ that passes through the point (0. 0).
274 Chapter 5 Integration
Solution
2
Each little segment represents the slope of ex for a particular x value. For example, if
x = 0, the slope is 1, if x = 1, the slope is e, and if x = 2, the slope is e4 (quite steep).
However, to draw the antiderivative, step back and take a large view of the graph, and
“go with the flow.” We sketch the apparent graph that passes through (0, 0), as shown
in Figure 5.3. ■
ANTIDERIVATIVE NOTATION
/ /(x) dx = F(x) + C
^ What This Says This is nothing more than the definition of the antideriva¬
tive, along with a convenient notation. We also agree that in the context of anti¬
differentiation, C is an arbitrary constant.
Figure 5.4 Several members of the The graph of F(x) + C for different values of C is called a family of functions
family of curves y = F(x) + C (see Figure 5.4).
WARNING
It is important to remember that
just a single function. /
f f(x) dx represents a family of functions, not
Because each member of the family y = F(x) + C has the same derivative at x,
the slope of the graph at x is the same. This means that the graph of all functions of
the form y = F{x) + C is a collection of parallel curves, as shown in Figure 5.4.
The process of finding indefinite integrals is called indefinite integration. Notice
that this process amounts to finding an antiderivative of / and adding an arbitrary
constant C, which is called the constant of integration.
EXAMPLE 3 Antidifferentiation
Find each of the following indefinite integrals.
/ 5x3 dx b. J sec2xdx c
:7 ex dx
Solution
d A -i d ( 5x4\ o
7 f,
b. Because —-(tanx) = sec x, we have / sec x dx = tanx + C.
dx J
(I c
c. Since — (ex) = ex, we have / ex dx = ex + C.
dx J
5.1 Antidifferentiation 275
ANTIDIFFERENTIATION FORMULAS
PROCEDURAL RULES
d df
Constant multiple: J cf(u)du=c J f(u)du
Tu(cf) = cTu
d£ + dg_
Sum rule: -l~(f + g) J[f(u) + g(u)]du = J f(u)du + J g(u) du
du du du
d df dg
Difference rule: J[f(u) - g(u)]du = J f (u) du — J g(u) du
Tff-g) = d-u~Tu
Linearity rule: f(af + bg)=aj- + b‘f- J[af(u) + bg(u)] du — a J f(u) du + b J g(u) du
du du du
BASIC FORMULAS
Constant rule:
d ,
— (c) = 0
du / 0 du — 0 + C
Exponential rule:
du
eu
/ eu du — eu + C
in+l
+ C; n / -1
Power rule:
du
nun~
i
/ un du = TT
n- n = —1
d 1
Logarithmic rule:* — (In |m|)
du u
Trigonometric rules:
d
— (cos u)
du
- — sin u
/ sin udu = — cos u + C
N
d .
— (sin u)
du
= cos u
/ cos u du — sin u + C
d
— (tan u)
du
— sec2 u
/ sec2 u du — tan u + C
d
—(sec u)
du
= sec u tan u
/ sec u tan u du = sec u + C
d ,
—(esc u)
du
— — CSC ucot u
/ esc u cot u du = — esc u + C
d
— (cotn)
du
= — CSC2 u
/ esc2 udu — — cot u + C
Note the absolute value. This is the result of reversing Theorem 3.13, Section 3.6.
276 Chapter 5 Integration
r du
Inverse trigonometric rules: —(sin w) = -. sin 1 u + C
du VI - u2 J Vl — u2
d _i _ i
The other three inverse trigonometric ^(tan w) — = tan 1 u + C
1 + u1 M
rules are not needed since, for exam-
71 i d , 1 f du
pie, sin u + cos u = —, so that —(sec-1 u) — . sec ' u + C
2 du |W|V^TT J \u\ \Ju1 — 1
7T
/ yr cos 1 u + Ci
— — cos 1 u + C
Proof Each of these parts can be derived by reversing the accompanying derivative
formula. For example, to obtain the power rule, note that if n is any number other than
— 1, then
d 1 n+1 1
~ [(t7 + 1 )un] — Un
du 77 + 1 n +
/ un du =
77 + 1
un+l + C for 77 ^ — 1
Solution
The first two steps are usually done mentally:
5+1 j^-2+1
3 lx + C Power rule
5+1 2+1
= 2x6 — x3 — lx + C
Solution
v-3/2
+ 4(— cosx) + C Power and trig rules
= fx3/2 -4cosx + C ■
APPLICATIONS
In Chapter 3, we used differentiation to compute the slope at each point on the graph
of a function. Example 6 shows how this procedure can be reversed.
EXAMPLE 6 Finding the function with a given slope through a particular point
The graph of a certain function F has slope 4a-3 - 5 at each point (a, y) and contains
the point (1,2). Find the function F.
Solution
We will work this problem twice; first we find an analytic solution and the second time
we approximate the solution using technology, which is often sufficient.
Analytic solution: Because the slope of the tangent at each point (a, y) is given by
F'{a), we have
F\a) = 4a3 - 5
J F\a) dx — J (4a3 — 5) dx
F(x) — 4 4! Vs x +c
= x4 — 5a + C
The family of curves is y = x4 - 5x + C. To find the one that passes through (1,2),
substitute:
2 = l4 — 5(1) + C
6 = C
The curve is y = a4 — 5a + 6.
Technology solution: Begin by drawing the slope field, as shown in Figure 5.5a. We
are interested in drawing the particular solution passing through (1,2). Remember to
“go with the flow” as shown in Figure 5.5b. If we compare the analytic solution and
the graphical solution, we see that the graph of the equation in the analytic solution is
the same as the one found by technology.
<t t Tj i t i
l i \ \ * ,•
L±1 if li
T v \
y / 1!
1 V 50 “ ✓/ 1 1 ■ > l
v t \ \ \ \- // ! 1 1
\ t f /
\ i t i : f l
_I
< J_.
j-s< - y?
j ; *
t i y/
t l
1* 1
*20 \ r\i ( !
\ \
; < v
y/
-41) .
ss
t'
v{t) = J a(t) dt
Solution
Because a{t) = v'(t), it follows that
Solution
Let a(t), v(t), and s(t) denote the acceleration, velocity, and position of the car t
seconds after the brakes are applied. We will assume that 5 is measured from the point
where the brakes are applied, so that 5(0) = 0.
<3
II
= —221 + 88
Similarly,
a.
Co
II
= J (—22t + 88) dt
Finally, the car comes to rest when its velocity is 0, so we need to solve v(t) = 0 for t:
-221 + 88 = 0
t = 4
This means that the car decelerates for 4 sec before coming to rest, and in that time it
travels
5(4) = —11 (4)2 + 88(4) = 176 ft ■
Indefinite integration also has applications in business and economics. Recall from
Section 4.7 that the demand function for a particular commodity is the function p(x),
which gives the price p that consumers will pay for each unit of the commodity when a
units are brought to market. Then the total revenue is R(x) = xp(x), and the marginal
revenue is R'(x). The next example shows how the demand function can be determined
from the marginal revenue.
Solution
R{x) = J R\x)dx
= J (240 + 0.1a)</a
= 240a + 0.05a2 + C
Because R(x) — xp(x), where p(x) is the demand function, we must have R(0) = 0
so that
240(0) + 0.05 (0)2 + C = 0 or C = 0
AREA AS AN ANTIDERIVATIVE
In the next section, we will consider area as the limit of a sum, and we conclude this
section by showing how area can be computed by antidifferentiation. The connection
between area as a limit and area as an antiderivative is then made by a result called the
fundamental theorem of calculus (see Section 5.4).
Proof Define an area function, A{t), as the area of the region bounded by the curve
y — /(a), the A-axis, and the vertical lines x = a, x — t for a < t < b, as shown in
Figure 5.6 Area function A(t) Figure 5.6. We need to show that A(t) is an antiderivative of / on the interval [r/, b];
that is, we need to show that A'{t) = f(t).
Let h > 0 be small enough so that t + h < b and consider the numerator of the
difference quotient for A(t), namely, the difference A(t + /?) - A(t). Geometrically
this difference is the area under the curve y = /(a) between a = t and a = t + h, as
shown in Figure 5.7a.
280 Chapter 5 Integration
y=f(x)
If h is small enough, this area is approximately the same as the area of a rectangle with
base h and height /(c), where c is the midpoint of the interval [t, t + h], as shown in
Figure 5.7b. Thus, we have
A(t + h) - A{t)
h
m
Finally, by taking the limit as h -» 0+, we find the derivative of the area function A(t)
satisfies
v A(t + h)-A(t)
lim - lim /(c)
h-* 0+ h 0+
A\t) = f{t)
The limit on the left is the definition of derivative, and on the right we see that since
/ is continuous and c is the midpoint of the interval \t, t + h], c must approach t as
h -> 0+. A similar argument works as h -» 0~. Thus, A(t) is an antiderivative of
fit). □
Solution
Let Ait) be the area function for this example—namely, the area under y = x2 on
[0, t]. Since / is continuous and fix) > 0 on [0, 1], Theorem 5.3 tells us that Ait) is
an antiderivative of fit) — t2 on [0, 1]. That is,
Ait) = J t2dt = + C
A(l) = I(l)3 +0 = 1
5.1 Antidifferentiation 281
2 dx -4Jx
'■! ■h‘
7- !tx
f dx
i. J 14e* r/x
,/ sec2 6 dO 12
7
sec# tan# dO
s*
cos#
35. slope
x+1
36. slope
C\A2 — 1
13. / 2 sin 0 dO 14
•/ •/
/■
>5' f F-y1
5
dy
’■ J 10(1 + x2)
J (m3^2 - n,l/2 + 10 ) du
!. J (x3 — 3x + — 5) dx
x2- 1
x(x + \/x) dx 20. y(y2 - 3y)dy 37. slope x + e* 38. slope
’•/ x2+ 1
with point (0, 2)
21. 22.
nH1* cva vy;
vTs \ \ ■
i \ \ \ \ ’ tt
71
\ \ -
<CT;
f.
t /
/ /
rrr TT~ rrr
i i i 1
i t
\ !
/ 1
(
t
1
1
m
t »
i t
i i
i i
with point (0, 0)
r~T~7 "7-7™,
< / / /y'. / s
// / // fs
7// // + /s /• ^ •
<// //
>’>
• 1N
N -
x'/' -
** // -
r-jr-T V 7?
/ / PS,
y / / /y
///
' s/
“•/HX1-?
1. V \ \ N ' l ( i i 1 '. \ \ : 1 t // / s _
> i f /■ /brn ' / * v/ • //
x2 + x - 2 ^ v. V
t \ y v \ ’ f t t i / i { f /s// ' s/ • k-\ ' //
28. dx 3 ^4
2 dX ^ \ V
k ^. —s
\\> f l i i
/ ; i
( l I
! t ; i
'/ / / / - ✓ /
rs /
- v \ \ •>-
A-
- ^ X1 //
' //
' /'/'
\ • it'
r " ' \ \ V V V V f 1 SYs /Y ' / > • •" X1 /✓
^ \ \ S \ \N' — / r 1 i i 1 1 t i i / / s //* ' / / - *•- N // ' /X
\.AJ A. V LJ-. -L-L. i l _J—L i—i—t
s/1 -X2 - 1
'•/
-dx
BS along with one point (xo, Vo) on //ic graph. Use this information b. Sketch the graphs of y = F(x), y = F(x) + 3, and
to find F both graphically and analytically. y = F(x)-l.
c. Find a constant C0 so that the largest value of
31. F'(x) = x2 + 3x 32. F'(x) = (2x - l)2
G(x) = F(x) + Co is 0.
with point (0, 0) with point (1,3)
\ | i 1 1 1 7X %
1 1
1 / 1 I t I 1 1 y. / / / f i f
/ / / { uU
40. A ball is thrown directly upward from ground level with an
1 i '.o' / i Jt }
1 1 ! I ! t t ! i •/ / 1 1
1 f
i <
; |
initial velocity of 96 ft/s. Assuming that the ball’s only ac¬
i 1 1 1 l 1 \ \ ( l * ' I f
/ / ■- ■ J I 1 l
! 1 f 1 1 1 Milt
i (
I 1
1 l
i i
J-L LL. Li-J. i L
celeration is that due to gravity (that is, alt) = -32 ft/s2),
1 1 1 1 l i .LILLI
5 1 l ) ! ! / \ i i i l
1 i
i l t i
L L L1
determine the maximum height reached by the ball and the
1 ! i 1 I ! ( \ 1 ! i i
1 w / i l i
1 l i 1 ! i
_ _ _ —►i- time it takes to return to ground level.
_
—
—L *
$ 4 *11 - -i
/ ( 41. The marginal cost of a certain commodity is
1 < i i I | , 1 ■J-L 1 i
1 l
lit
? C'(x) = 6x2 - 2x + 5, where x is the level of production. If it
J! 1 / costs $5 to produce 1 unit, what is the total cost of producing
J L
8- 5 units?
282 Chapter 5 Integration
42. The marginal revenue of a certain commodity is the car. After a reaction time of 00.7 seconds, he steps on the
R’(x) = —3x2 + Ax + 32, where x is the level of production brakes. Will he stop before hitting the camel?
(in thousands). Assume R(0) = 0. 51. A particle moves along the x-axis in such a way that at time
a. Find the demand function p(x). t > 0, its velocity (in ft/s) is
b. Find the level of production that results in maximum rev¬
v(t) — t~l + t
enue. What is the market price per unit at this level of
production? How far does it move between times t = 1 and t = e2l
43. It is estimated that t months from now, the population of a 52. A manufacturer estimates that the marginal cost in a certain
certain town will be changing at the rate of 4 + 5f2/3 people production process is
per month. If the current population is 10,000, what will the
population be 8 months from now? C'(x) = 0.1e* + 2lVx
44. A particle travels along the x-axis in such a way that its accel¬
eration at time / is a(t) = *Jt + t2. If it starts at the origin
when x units are produced. If the cost of producing 1 unit is
with an initial velocity of 2 (that is, 5(0) = 0 and u(0) = 2),
$100, what does it cost (to the nearest cent) to produce 4 units?
determine its position and velocity when t = 4.
45. An automobile starts from rest (that is, v(0) = 0) and travels In Problems 53-58, find the area under the curve defined by the
with constant acceleration a(t) — k in such a way that 6 sec given equation, above the x-axis, and over the given interval.
after it begins to move, it has traveled 360 ft from its starting 53. y = x2 over [1, 4] 54. y = Vx over [1,4]
point. What is k? x+ 1
55. y — ex — x over [0, 2] 56. y =-over [1,2]
46. The price of bacon is currently $ 1.80/lb in Styxville. A x
consumer service has conducted a study predicting that t 57. y = cos x over [0, |] 58. y — (1 — x2)~1/2 over [0, |]
months from now, the price will be changing at the rate of
dy
0.084 + 0.012 V? cents per month. How much will a pound of
bacon cost 4 months from now?
47. An airplane has a constant acceleration while moving down
59. Evaluate
/ yVy2 -
2 1
using the indicated methods.
at a point 121 ft from the point where the brakes were applied. = a J f(x)dx+bj g(x)dx+c J h{x)dx
What is A:?
50. Spy Problem The Spy, having defused the bomb in Problem 61. Use the area as an antiderivative theorem (Theorem 5.3) to
23, Section 4.6, is driving the sports car in Problem 48 at a find the area under the line y — mx + b over the interval
speed of 60 mi/h on Highway 1 in the remote republic of San [c, d], where m > 0 and me + b > 0. Check your result by
Dimas. Suddenly he sees a camel in the road 199 ft in front of using geometry to find the area of a trapezoid.
Computing area has been a problem of both theoretical and practical interest since
ancient times, but except for a few special cases, the problem is not easy. For example,
you may know the formulas for computing the area of a rectangle, square, triangle,
circle, and even a trapezoid. You have probably found the areas of regions that were
more complicated but could be broken up into parts using these formulas. In Example
10 of the previous section we found the area between the parabola y = x2 and the
5.2 Area as the Limit of a Sum 283
jt-axis on the interval [0, 1] (see Figure 5.9a). We revisit this example to demonstrate
a general procedure for computing area.
Solution
In the previous section we found this area using the area function. In this example,
we will estimate the area by adding the areas of approximating rectangles constructed
on subintervals of [0, 1], as shown in Figure 5.9b. To simplify computations, we will
require all approximating rectangles to have the same width and will take the height of
each rectangle to be the y-coordinate of the parabola above the right endpoint of the
subinterval on which it is based.*
For the first estimate,we divide the interval [0, 1 ] into 5 subintervals, as shown in
Figure 5.10a. Because the approximating rectangles all have the same width, the right
endpoints are x\ = 0.2, x2 = 0.4, x3 = 0.6, x4 = 0.8, and x5 = 1. This subdivision is
called a partition of the interval. The width of each subdivision is denoted by A* and
is found by dividing the length of the interval by the number of subintervals:
a. Partition into 5 subdivisions
b. Detail showing one rectangle Even though S5 = 0.44 serves as a reasonable approximation of the area, we see
from Figure 5.10b that this approximation seems too large. Let us rework Example
Figure 5.10 Partitioning of
Figure 5.8a into 5 subdivisions * Actually, there is nothing special about right endpoints, and we could just as easily have used any other
point in the base subinterval—say, the left endpoint or the midpoint.
284 Chapter 5 Integration
1 using a general scheme rather than a specified number of rectangles. Partition the
interval [0, 1] into n equal parts, each with width
1-0 1
Ax =-= -
n n
1 A
For A = 1, 2, 3, ..., n, the Ath subinterval is , and on this subinterval we
n n
1 (k\2
then construct an approximating rectangle with width Ax = - and height — ) , since
n \n)
y = x2. The total area bounded by all n rectangles is
'1
Sn
n
+ ••• +
O n
n=4 n = 16 n = 64
We now compute the area under any curve y = /(x) on an interval [a, b], where /
is a nonnegative continuous function. We first partition the interval [a, b] into n equal
subintervals, each of width
n
For k — 1,2, 3,..., n the Ath subinterval is [a + (A — 1) Ax, a + AAx], and the Ath
approximating rectangle is constructed with width Ax and height f{a 4-AAx) equal to
the height of the curve y = /(x) above the right endpoint of the subinterval. Adding
the areas of these n rectangles, we obtain
Area of first rectangle Area of second rectangle Area of nth rectangle
Y = X2 Y = X2
(Network points are left endpoints.)|
WANT MORE?
y
NBR. RIEMANN SUM OVER 28
TERMS 11, 5] A
<1 30
8 35.5
14 38.375
32 39.84375
64 40.58594
128 40.95898
' ' 1 ''I
as an estimate of the area under the curve. In advanced calculus, it is shown that the
continuity of / guarantees the existence of lim Sn, and we use this limit to define the
Ax—>0
required area, as shown in the following box.
The definition of area as a limit of a sum is consistent with the area concept in¬
troduced in plane geometry, and with the area function defined in Section 5.1. For
example, it would not be difficult to use this formula to show that a rectangle has area
A = iw or that a triangle has area A = \bh. You will also note that we maintained
everyday usage in saying that the formula for the area of a rectangle is A — Iw or the
area of the region under the curve y = f(x) is a limit, but what we are really doing is
defining area as a limit.
The problem we now face is how to implement this definition of area as the limit of
a sum. The immediate answer (discussed in this section) is to use summation formulas
and technology. The long-range goal is to develop integral calculus, which is discussed
in the next section.
SUMMATION NOTATION
The expanded form of the sum for the definition of area makes it awkward to use.
Therefore, we will digress to introduce a more compact notation for sums, and that
notation will motivate integral notation. Using this summation notation, we express
the sum a\ + + • • • + an as follows:
n
a\ + Cl2 "T • • • + = y ak
k= 1
The summation notation is sometimes called the sigma notation because the uppercase
Greek letter sigma (E) is used to denote the summation process. The index k is called
the index of summation (or running index). The terminology used in connection with
the summation notation is shown:
Upper limit of summation
General term
t I
k= 1 f
Index of summation
Note that in the summation process, the choice of summation index is immaterial.
For example, the following sums are all exactly the same:
Proof These properties can all be established by applying well-known algebraic rules
(see Problem 50). □
Using summation notation, we can streamline the symbolism in the formula for the
area under the curve y = /(x), /(x) > 0 on the interval [a, b]. In particular, note that
the approximating sum Sn is
where Ax = ——. Thus, the formula for the definition of area is shown:
n
From algebra we recall certain summation formulas (which can be proved using
mathematical induction) that we will need in order to find areas using the limit defini¬
tion.
Summation Formulas
k=1
n
n(n + 1)
^ k — 1+2 + 3 +-\- n —
k=i
n
n{n + 1) (2/2 + 1)
W = l2 + 22 + 32 + ...+n2
h 6
Solution
= .J5.fi©©
&=1
n
k2 Note that 1 /n is independent of
= lim
n—t+OO
y n3 the index of summation, k.
k= 1
n
Scalar multiple rule
n-*+oo n3 Ek2
k=\
1 ~n(n + l)(2n + 1)
lim — Summation formula for squares
n->+oo ni 6
1 3 1 1
lim — 2 H-1-r
/I->+0O 6 n /r 3
This is the same as the answer found by antidifferentiation in Example 10 of Section
5.1 and is consistent with the table on page 284. ■
Solution
K —0 K
We see a = 0 and b = —, so Ax = -= —. The right endpoints are
n 2n
/ tc \ 7t 2k 7x „
3k
a T Ax = 0 4“ ( —— ) — —; a + 2Ax = — = —; a + 3Ax = —;..
\2 nt In 2n n 2n
UK K
a + n Ax = — = — = b
2n 2
500 1.00157 Thus, we expect the actual area under the curve to be 1 square unit. ■
Use the properties of summation notation in Problems 9-12 to 33. Recall that a function / is said to be even if /(—x) = fix)
evaluate the given limits. for all x. If / is even and fix) > 0 throughout the inter¬
n L. n ,3 val [—a, a], then the area under the curve y = fix) on this
9. lim V— 10. lim V — interval is twice the area under y = fix) on [0, a],
n—► +00 f—J Yll n-*+oo / yc
k=\ k—1 34. Show that the region under the curve y = x3 on the interval
[0, 1 ] has area £ square units.
35. Use the definition of area to show that the area of a rectangle
equals the product of its length l and its width w.
First sketch the region under the graph of y = fix) on the inter¬
val [a, b] in Problems 13-21. Then approximate the area of each 36. Show that the triangle with vertices (0, 0), (0, h), and {b, 0)
region by using right endpoints and the formula has area A = \bh using the area as the limit of a sum.
37. a. Compute the area under the parabola y = 2x2 on the inter¬
val fl, 2] as the limit of a sum.
S„ = ^ f{a + kAx)Ax
k= 1
b. Let fix) = 2x2 and note that g(x) = |x3 defines
b—a a function that satisfies g\x) — fix) on the interval
for Ax — -and the indicated values of n. [1,2]. Verify that the area computed in part a satisfies
n
A = g(2) - g(l).
13. f(x) — 4x + 1 on [0, 1] for
c. The function defined by
a. n — 4 b. n = 8
14. fix) = 3-2x on [0, 1] for hix) — |x3 + C
c. Combine parts a and b to show that Compare this with the procedure outlined in Example I. Note
that when the interval [0, 1] is subdivided into n equal parts,
^ , n(n + 1) the A th subinterval is
2s ~ 2
*=i ~k — 1 k~
n n_
47. a. First find constants a, b, c, and d such that
49. Develop a formula for area based on approximating rectangles
k3 = a[kA -(k- 1 )A] + bk7 +ck+d
with heights taken at the midpoints of subintervals.
50. Use the properties of real numbers to establish the summation
b. Modify the approach outlined in Problem 46 to establish
formulas in Theorem 5.4. For example, to prove the linearity
the formula
rule, use the associative, commutative, and distributive prop¬
n(n + \)(2n + 1) erties of real numbers to note that
E*! 6 n
y^(cak +dbk)
48. The purpose of this problem is to verify the results shown in k=1
Figure 5.11. Specifically, we will find the area A under the = (ca \ + db\) + {ca 2 + dbf) + ■ • • + (can + db,,)
parabola y = x2 on the interval [0, 1 ] using approximating = (ca) + ca.2 + • • • + can) + (db\ + dbj + • • • + dbn)
rectangles with heights taken at the left endpoints. = c(a\ + + • • • + a,,) + d{b\ + bj + • • • + bn)
Verify that n n
= c^ak + d^bk
k=1 k=1
We will soon discover that not just area, but other useful quantities such as distance,
volume, mass, and work, can be first approximated by sums and then obtained ex¬
actly by taking a limit involving the approximating sums. The special kind of limit
of a sum that appears in this context is called the definite integral, and the process of
finding integrals is called definite integration or Riemann integration in honor of the
German mathematician Georg Bernhard Riemann (1826-1866), who pioneered the
modern approach to integration theory. We begin by introducing some special notation
and terminology.
RIEMANN SUMS
Recall from Section 5.2 that to find the area under the graph of the function y = /(x)
on the closed interval [a, b] using sums where / is continuous and f(x) > 0, we
proceed as follows:
Sn = ^ f(a + k Ax)Ax.
*=1
4. Because we expect the estimates S„ to improve as Ax decreases, we define the area
A under the curve, above the x-axis, and bounded by the lines x = a and x = b, to
be the limit of S„ as Ax -> 0. Thus, we write
n
A = lim
/!-»+0O E f(a + k Ax)Ax
5.3 Riemann Sums and the Definite Integral 291
if this limit exists. This means that A can be estimated to any desired degree of
accuracy by approximating the sum Sn with Ax sufficiently small (or, equivalently,
n sufficiently large).
This approach to the area problem contains the essentials of integration, but there
is no compelling reason for the partition points to be evenly spaced or to insist on
evaluating / at right endpoints. These conventions are for convenience of computation,
but to accommodate easily as many applications as possible, it is useful to consider a
more general type of approximating sum and to specify what is meant by the limit
of such sums. The approximating sums that occur in integration problems are called
Riemann sums, and the following definition contains a step-by-step description of
how such sums are formed.
Riemann Sum Suppose a bounded function / is given along with a closed interval [a, b] on
which / is defined. Then
This is the Riemann sum associated with /, the given partition P, and
the chosen subinterval representatives x*, x|,..., x*.
What This Says We will express quantities from geometry, physics, eco¬
nomics, and other applications in terms of a Riemann sum
n
£/(***) Ax*
k=i
Riemann sums are generally used to model a quantity for a particular application.
Note that the Riemann sum does not require that the function / be nonnegative,
nor does it require that all the intervals must be the same length. In addition, x£
is any point in the kth subinterval and does not need to be something “nice” like
the left or right endpoint, or the midpoint.
jc6 = 1 = b. Find the norm of this partition P and the Riemann sum associated
with the function /(x) = 2x, the given partition, and the subinterval representatives
x* = -1.81, x* = -1.12, x* = —0.55, x% = -0.17, x* = 0.43, and x* = 0.94.
Solution
Before we can find the norm of the partition or the required Riemann sum, we must
compute the subinterval width Axk and evaluate / at each subinterval representative
x*. These values and computations are shown in Figure 5.13.
*0 X1 x2 *3 *4
X5 A
6
•- 11 •- 11 11 • U 1_1 a | A
*7
*
O -
7
Y *-
*
l *1* *2-1 x* •*6
Given Given
i
\
1
[I
X /(x*) = 2(x*)
<
k xk
H
H
II
1
1 -1.6- (-2) = 0.40 -1.81 /(—1.81) = -3.62
From the table, we see that the largest subinterval width is AX3 = 0.72, so the partition
WARNING .. . , „ , . has norm \\P\\ = 0.72. Finally, by using the definition, we compute the Riemann sum:
Notice rrom Example I
that the Riemann sum does not
necessarily represent an area. The sum R6 = (—3.62) (0.40) + (—2.24) (0.67) + (-1.10)(0.72) + (-0.34)(0.56)
found is negative (and area must be
+ (0.86X0.47) + (1.88X0.18)
nonnegative).
^-3.1886 ^-3.19 ■
By comparing the formula for the Riemann sum with that of area in the previous sec¬
tion, we recognize that the sum S„ used to approximate area is actually a special kind
of Riemann sum that has
A xk = Ax =- and x£ = a + k Ax
n
for k = 1,2,..., n. Because the subintervals in the partition P associated with Sn are
equally spaced, it is called a regular partition. When we express the area under the
curve y — f(x) as A = lim Sn, we are actually saying that A can be estimated to any
Ax—>0
desired accuracy by finding a Riemann sum of the form Sn with norm
b — a
mi = n
sufficiently small. We use this interpretation as a model for the following definition.
5.3 Riemann Sums and the Definite Integral 293
Definite Integral If / is defined on the closed interval [a. b), we say / is integrable on [a, b] if
n
I f (x) dx
- Ja
^ What This Says To say that / is integrable with definite integral I means
the number I can be approximated to any prescribed degree of accuracy by any
Riemann sum of / with norm sufficiently small. As long as the conditions of
this definition are satisfied (that is, / is defined on [a, b) and the Riemann sum
exists), we can write
Formally, I is the definite integral of / on [a, b] if for each number e > 0, there
exists a number 8 > 0 such that if
n
y f(x*k)Axk
k=1
is any Riemann sum of / whose norm satisfies || P < 8, then
n
i - y f(xk)Axk < 6
k=1
In advanced calculus, it is shown that when this limit exists, it is unique. Moreover,
its value is independent of the particular way in which the partitions of [a, b] and the
subinterval representatives xk are chosen.
Take a few minutes to make sure you understand the integral notation and
terminology.
The function / that is being integrated is called the integrand; the interval [a, b]
is the interval of integration; and the endpoints a and b are called, respectively, the
lower and upper limits of integration.
In the special case where a = b, the interval of integration [a, b] is really just a
point, and the integral of any function on this “interval” is defined to be 0; that is,
Also, at times, we will consider integrals in which the lower limit of integration is a
larger number than the upper limit. To handle this case, we specify that the integral
from b to a is the negative of the integral from a to b:
[ f(x)dx = -[ f{x)dx
Jb Ja
To summarize,
294 Chapter 5 Integration
f(x)dx = - f (a) dx
At first, the definition of the definite integral may seem rather imposing. How
are we to tell whether a given function / is integrable on an interval [a, b]l If / is
integrable, how are we supposed to actually compute the definite integral? Answering
these questions is not easy, but in advanced calculus, it is shown that / is integrable
on a closed interval [a, b] if it is continuous on the interval except at a finite number
of points and if it is bounded on the interval (that is, there is a number A > 0 such that
|/(a)| < A for all a in the interval). We will state a special case of this result as a
theorem.
Proof The proof requires the methods of advanced calculus and is omitted here. □
Our next example illustrates how to use the definition to find a definite integral.
Evaluate J 4xdx.
Solution
The integral exists because /(a) = 4a is continuous on [—2, 1]. Because the inte¬
gral can be computed by any partition whose norm approaches 0 (that is, the integral
is independent of the sequence of partitions and the subinterval representatives), we
will simplify matters by choosing a partition in which the points are evenly spaced.
Specifically, we divide the interval [—2, 1] into n subintervals, each of width
A 1 - (-2) 3
n n
For each k, we choose the kth subinterval representative to be the right endpoint of the
kth subinterval; that is,
Xk — —2 + kAa = —2 + k ( —
12 / n(n + 1)
= lim — ( (—2n)n + 3 Summation formulas
rt->+oo ft- V
12 / —An1 + 3n2 + 3n
lim — ---
n->+oc nz \ 2
—6n2 + 18/7
lim --- - —6
n—>+oo
AREA AS AN INTEGRAL
Because we have used the development of area in Section 5.2 as the model for our
definition of the definite integral, it is no surprise to discover that the area under a
curve can be expressed as a definite integral. However, integrals can be positive, zero,
or negative (as in Example 2), and we certainly would not expect the area under a curve
to be a negative number! The actual relationship between integrals and area under a
curve is contained in the following observation, which follows from the definition of
area as the limit of a sum, along with Theorem 5.5.
Area as an Integral
WARNING ... ... , ,
We will find areas using a
Suppose / is continuous and f(x) >0 on the closed interval [a, b]. Then the
area under the curve y — fix) on [a, b] is given by the definite integral of / on
definite integral, but not every definite
integral can be interpreted as an area. [a, b]. That is,
AREA = / f(x)dx
Ja
Usually we find area by evaluating a definite integral, but sometimes area can be
used to help us evaluate the integral. At this stage of our study, it is not easy to evaluate
Riemann sums, so if you happen to recognize that the integral represents the area of
some common geometric figure, you can use the known formula instead of the definite
integral, as shown in Example 3.
Evaluate — x2 dx.
Solution
Let fix) — V9 — x2. The curve y — V9 - x2 is a semicircle centered at the origin
of radius 3, as shown in Figure 5.14. The given integral can be interpreted as the area
under the semicircle on the interval [-3, 3]. From geometry, we know the area of the
circle is A = nr2 = 7r(3)2 = 97r. Thus, the area of the semicircle is and we
conclude that
Figure 5.14 The curve
9tc
V = a/9 — x2 is a semicircle of — x2 dx
radius 3. ~2
296 Chapter 5 Integration
If / is continuous and /(x) < 0 on the closed interval [a, b], then the area between
the curve y = f{x) and the x-axis on [a, b] is the opposite of the definite integral of /
on [a, b], since —/(x) > 0. That is,
fb e
AREA = — / f(x)dx
Ja
More generally, as illustrated in Figure 5.15, if a continuous function / is sometimes
positive and sometimes negative, then
b
f(x) dx — A\ — A2
where
A | is the sum of all areas of the regions above the x-axis and below the graph of
/ (that is, where /(x) > 0).
A2 is the sum of all areas of the regions below the x-axis and above the graph of
/ (that is, where /(x) < 0).
Figure 5.15 Aj is the sum of the areas marked “+” and A2 is the sum of the areas marked
We just worked Example 2 (see Figure 5.16) using a Riemann sum, and it is easy to see
from our knowledge of areas of triangles that the area of Triangle I (above the x-axis)
is
A] = jbh = |(4)(1) = 2
Dominance rule If / and g are integrable on [a, b] and fix) < g(x)
throughout this interval, then
pb pb
/ f(x)dx < / g(x)dx
Ja Ja
Proof Each of these rules can be established by using a familiar property of sums or
limits with the definition of the definite integral. For example, to derive the linearity
rule, we note that any Riemann sum of the function rf + sg can be expressed as
The dominance rule and the subdivision rule are interpreted geometrically for non¬
negative functions in Figure 5.17.
Notice that if g(x) > f(x) > 0, the curve y = g(x) is always above (or touching) the
curve y = fix), as shown in Figure 5.17a. The dominance rule expresses the fact that
the area under the upper curve y — g(x) cannot be less than the area under y — fix).
The subdivision rule says that the area under y = fix) above [a, b] is the sum of the
area on [a, c] and the area on [c, b], as shown in Figure 5.17b. The following example
illustrates one way of using the subdivision rule.
Solution
According to the subdivision rule, we have
Therefore,
DISTANCE AS AN INTEGRAL
Many quantities other than area can be computed as the limit of a sum. For example,
suppose an object moving along a line is known to have continuous velocity v(t) for
each time t between t = a and t = b, and we wish to compute the total distance
traveled by the object during this time period.
Let the interval [a, b] be partitioned into n equal subintervals, each of length
b —a
At =-as shown in Figure 5.18.
n
a + {k-\)At a + kAt
1 1
1 1
a Ar = ^7T b
Figure 5.18 The distance traveled during the /cth time subinterval
The kth subinterval is[a + (k—l)At,a + kAt], and if At is small enough, the velocity
u(t) will not change much over the subinterval so it is reasonable to approximate v(t)
by the constant velocity of v[a + (k — l)At] throughout the entire subinterval.
The corresponding change in the object’s position will be approximated by the
product
v[a + (k — l)Ar]Af
and will be positive if v[a + (k — l)At] is positive and negative if v[a + (k — 1) Ar] is
negative. Both cases may be summarized by the formula
|v[a + (k — l)Af]| At
and the total distance traveled by the object as t varies from t = a to t = b is given by
the sum
Sn — |v[a + (k — l)At]| At
k= 1
which we recognize as a Riemann sum. We can make the approximation more precise
by taking more refined partitions (that is, shorter and shorter time intervals At). There¬
fore, it is reasonable to define the exact distance S traveled as the limit of the sum Sn
as At —► 0 or, equivalently, as n —> +oo, so that
n pb
S — lim Y \v[a + (k — 1)Ar]| At = / \v(t)\ dt
«->•+ 00 ^—‘
Ja
The reason that we use |u| is that if we travel 300 miles east and then 300 miles west,
for example, the net distance traveled is 300 — 300 = 0 miles, but we are usually
interested in the total distance traveled, 300 4- 300 = 600 miles.
Distance The total distance traveled by an object with continuous velocity v(t) along a
straight line from time / = a to / = b is
S=( |u(/)| dt
5.3 Riemann Sums and the Definite Integral 299
Note: There is a difference between the total distance traveled by an object over a
given time interval [a, b], and the net distance or displacement of the object over the
same interval, which is defined as the difference between the object’s final and initial
positions. It is easy to see that displacement is given by
D — f v(t) dt
Solution
2-1 1
We have a — 1, b = 2, and At =-= therefore, the required distance is
n n
= J |u(/)| dt
T
= lim
rt—►+ oo
T '
l + (*-l)l-
n
k=1
n
'n + k — 1 T
lim Y'
n—>+oo
->4-oo Z—' n
k=\
= ,im v("+yi)2 a
n—>-+oo * \ n
k= 1
l ^
= lim — \ ^[(n- — 2n + 1) + k" + 2(n — 1)^]
n—►-f-oo ni ^'
k=1
n(n + 1) (2 n + 1) n(n + 1)
= lim — (n — 2n + l)n + + 2(n - 1)
n->+oo n5
14n3 — 9 n2 + n 14 7
lim
n->+oo 6n3 6 3
Thus, we expect the object to travel | units during the time interval [1,2].
By considering the distance as an integral, we see again that zero or negative values
of a definite integral can also be interpreted geometrically. When v(t) > 0 on a time
interval [a, b], then the total distance S traveled by the object between times t — a and
t = b is the same as the area under the graph of v(t) on [a, b].
When v(t) > 0, the object moves forward (to the right), but when v(t) < 0, it
reverses direction and moves backward (to the left). In the general case, where v(t)
changes sign on the time interval [a, b], the integral
b
v(t) dt
measures the net distance or displacement of the object, taking into account both for¬
ward (v > 0) and backward (v < 0) motion. For instance, an object that moves for¬
ward 2 units and back 3 on a given time interval has moved a total distance of 5 units,
but its displacement is -1 because it ends up 1 unit to the left of its initial position.
For example, for the velocity function v(t) graphed in Figure 5.19a, the net dis¬
placement is 0 because the area above the t-axis is the same as the area below, but in
Figure 5.19b, there is more area below the r-axis, which means the net displacement is
negative, and the object ends up “behind” (to the left of) its starting position.
PREVIEW
Students often become discouraged at this point, thinking they will have to com¬
pute all definite integrals using the limit of a Riemann sum definition, so we
offer an encouraging word. Recall the definition of derivative in Chapter 3. It
was difficult to find derivatives using the definition, but we soon proved some
theorems to make it easier to find and evaluate derivatives. The same is true of
integration. It is difficult to apply the definition of a definite integral, but you will
soon discover that computing most definite integrals is no harder than finding an
antiderivative.
As a preview of this result, consider an object moving along a straight line.
We assume that its position is given by s(t) and that its velocity v{t) — s'{t)
is positive (v{t) > 0) so that it is always moving forward. The total dis¬
tance traveled by such an object between times t = a and t — b is clearly
S = s(b) — s(a), but earlier in this section, we showed that this distance is also
given by the definite integral of v{t) over the interval [a, b]. Thus we have
TOTAL DISTANCE = f
Ja
v(t) dt — s(b) - s(a)
ri , i ,1 7
S = f t2dt = 5(2) -5(1) = -(2)3--(l)3
'l 3 3 3
which coincides with the result found numerically in Example 5.
5.3 Riemann Sums and the Definite Integral 301
5. j (1-3x)dx 6.
f3
7.
7-i
/
d —7Tr/2
cosxdx 8.
Jo
/
C2 dx
(x + sinx)dx
J /(x) dx = 3; Ji f(x)dx — —2; /(x)dx = 5
ex dx 10
Jl * 2 for -1 < x < 1
In Problems 11-16, v(t) is the velocity of an object moving along 28. Let /(x) = 3 —x forl<x<4
a straight line. Use the formula 2x — 9 for 4 < x < 5
n Sketch the graph of / on the interval [—1,5] and show that
Sn — ^ |u(a + ^ Af)| At / is continuous on this interval. Then use Theorem 5.6 to
*=i
evaluate
b—a
where At — - to estimate (using right endpoints) the total
n
J fix) dx
distance traveled by the object during the time interval [a, b\. Let
5 for —3 < x < — 1
‘ n = 4.
29. Let /(x) = ■ 4 — x for — 1 < x < 2
11. v(t) = 3M- 1 on [1,4] 12. v(t) = 1 + 2r on [1, 2]
2x — 2 for 2 < x < 5
13. u(T) = sin t on [0. n] 14. v(t) = cos t on [0, f ]
Sketch the graph of / on the interval [—3, 5] and show that
15 . v(t) = e~l on [0, 1] 16 . v(t) = on [0, 1] / is continuous on this interval. Then use Theorem 5.6 to
evaluate
Evaluate each of the integrals in Problems 17-22 by using the
following information together with the linearity and subdivision J fix)dx
properties:
30. HISTORICAL QUEST
r2
/
__
x2 dx =2>j / x" dx =
, 1
x dx = —;
i
/
,2 3 rl
During the eighteenth century,
integrals were considered
J-
1 J-1 3 J -l 2 Jo
simply as antiderivatives. That
/» —1 r2 o is, there were no underpin¬
17 / . x2 dx 18 / (x2+x)c?x . nings for the concept of an in¬
Jo J-1
tegral until Cauchy formulated
r2 r2 o
.
19 1 (2x2 — 3 x)dx 20. / x2 dx the definition of the integral in
J-1 Jo 1823 (see Historical Quest GEORG FRIEDRICH RIEMANN
r0 1826-1866
r° o 49, Problem Set 2.1). This
21 .
/ x dx 22. / (3x2 — 5x)dx
formulation was later com¬
pleted by Georg Friedrich Riemann. In this section, we see that
Use the dominance property of integrals to establish the given in¬
history honored Riemann by naming the process after him. In
equality in Problems 23-24.
his personal life he was frail, bashful, and timid, but in his
23. / x3 dx < ^ Hint: Note that x3 < x on [0, 1]. professional life he was one of the giants in mathematical his¬
’• /' tory. Riemann used what are called topological methods in his
1. / sinx
24.
Jo
f
sir dx < n Hint: sinx < 1 for all x.
theory of functions and in his work with surfaces and spaces.
Riemann is remembered for his work in geometry (Riemann
,4 '4 surfaces) and analysis. In his book Space Through the Ages,
025. Given J [5 fix) 4- 2g(x)] dx = 7 and ^[3/(x) + g(x)] dx = 4, Cornelius Lanczos wrote, "Although Riemann's collected pa¬
pers fill only one single volume of 538 pages, this volume
find
weighs tons if measured intellectually. Every one of his many
J
fix) dx and gix)dx J discoveries was destined to change the course of mathematical
science." ■
302 Chapter 5 Integration
For this Historical Quest, investigate the Konigsberg 37. If the numbers ak and bk satisfy a* < bk for k = 1, 2,..., n,
n n
bridge problem, and its solution.
This famous problem was formulated by Leonhard Euler
then E ak < E bk. Use this dominance property of sums to
k=1 k=l
(1707-1783). The branch of mathematics known today as establish the dominance property of integrals.
topology began with Euler's work on the bridge problem and
38. Counterexample Problem Either prove that the following result
other related questions and was extended in the nineteenth
is generally true or find a counterexample: If / and g are con¬
century by Riemann and others.
tinuous on [a, b], then
31. Generalize the subdivision property by showing that for
a < c < d < b
f
Ja
f (x)g(x) dx = f (x) dx g(x) dx
whenever all these integrals exist. 39. Counterexample Problem Either prove that the following re¬
sult is generally true or find a counterexample: There ex¬
32. If Cx + D > 0 for a < x < b, show that
ists a nonzero continuous function / defined on [a, b] with
f
Ja
(Cx + D) dx = (b — a) y(& + a) + D
In the previous section we observed that if v(t) is the velocity of an object at time t as
it moves along a straight line, then
b
v(t) dt — s(b) — s(a)
where s(t) is the displacement of the object and satisfies s'(t) = v(t). This result is
an application of the following general theorem which was discovered by the English
mathematician Isaac Barrow (1630-1677), Newton’s mentor at Cambridge.
5.4 The Fundamental Theorems of Calculus 303
b
f(x)dx = Fib) - F(a)
Proof Let P = {x0, x\, x2,..., xn) be a regular partition of the interval, with subin¬
terval widths Ax = -——. Note that F satisfies the hypotheses of the mean value the¬
ft
orem (Theorem 4.4, Section 4.2) on each of the closed subintervals [x*_i, **]. Thus,
the MVT tells us that there is a point x*k in each open subinterval (**_i, **) for which
F(xk) - F{xk- ) 1
-= F (xk)
Xk ~ *k-1
F(xk) - F(xk-i) = F'(xk)(xk - xk-\)
b—a
Because F'(jc?) = f{xt) and xk -^_, = Ax =-, we can wnte
K n
F{xk) - F{xk-\) = f{xf)Ax, so that
Finally, we take the limit of the left side as || P || —>■ 0 (i.e., n —> +oo), and because
Fib) — Fia) is a constant, we have
[ fix)dx — Fib)-Fia) -l
To give you some insight into why this theorem is important enough to be named
the fundamental theorem of calculus, we repeat Example 2 from Section 5.3 on the left
and then work the same example using the fundamental theorem, on the right.
Evaluate J 4x dx using the definition of the definite integral and also using the fun¬
damental theorem.
Solution
Solution from Section 5.3 using Riemann sums: Solution using the fundamental theorem of calculus:
12 ”
— lim — Y (—2 n + 3k) = 2(1)2 - 2(—2)2 = -6
n-*+oo
k= 1
Note that if we choose a different antiderivative,
say G(x) — 2x2 + 3, then
\*=1 k=1 G(l) — G(—2) = —6, as well.
r 12 (( o n
= lim — I (—2n)n +, 3
a + 1)
rt-»+oo nl \
12 /—4n2 + 3n2 + 3n'
= lim — I ---
n-*+ oo flA \ 2
—6 n2 + 18 n
= lim
n—»+oo
= -6
The variable used in a definite integral is a dummy variable in the sense that it
can be replaced by any other variable with no effect on the value of the integral. For
instance, we have just found that
/: 4x dx = —6
J 4tdt = -6 4u du = —6 4 NdN = -6
2 /_
J-2
Sometimes we also write [F(x)]£, where F'(x) = fix) on [a, b]. This notation is
illustrated in Example 2.
EXAMPLE 2 Finding the area under a curve using the fundamental theorem
of calculus
Find the area under the curve v = cosx on [— j, f ].
5.4 The Fundamental Theorems of Calculus 305
Solution
Note that /(;c) = cos* is continuous and /(*) > 0 on [— y, y], as shown in Figure
5.20. Because the derivative of sin.v is cos*, it follows (from the definition of the
antiderivative) that sin* is an antiderivative of cos*. Thus, the required area is given
by the integral
/ 7T/2
WARNING |t j$ jmportant f0 remember that, for fixed numerical values of a and b, the
on jj- ~, 5 , you can see that an area family of functions. The relationship between the indefinite and definite integral is given by
of 2 seems reasonable.
dx = + x] dx = (7) +
= (6+a)-(4+^)=34i
Ja
r nt) dt
where the upper limit of integration is a variable instead of a constant. As * varies, so
does the value of the integral, and
Fix)
-r fit) dt
306 Chapter 5 Integration
is a function of the variable x. (Note that the integrand does not contain x.) For
instance,
The first fundamental theorem of calculus tells us that if / is continuous on [a, b], then
-b
G(x)
-f f if) dt
[f
= lim —
Ax->0 Ax I_Ja
fit)dt
f
J LI
f i{) dt
= lim
Aat^-0 Ul
fit)dt+
r
JX
fit)dt
Since / is continuous on [a, b], it is continuous on [x, x + Ax] for any x in [a, b]. Let
mix) and M(x) be the smallest and largest values, respectively, for f(t) on [x, x+Ax],
iNote: In general, m and M depend on x and Ax but are constants as far as t-integration
/
is concerned.) Since m(x) < fit) < Mix), we have (by the dominance rule)
x+Ax
m(x)dt < /
rx+Ax
fit)dt < /
px+Ax
Mix)dt
so
/.x+Ax
or
j rx+Ax
mix) < fit)dt < Mix)
Ax Jx
Since
mix) < fix) < M{x)
on [x,x + Ax], it follows that m(x) and M(x) are both “squeezed” toward fix) as
Ax —> 0 and we have
‘X+Ax
G\x) = lim ~ [
a*->o Ax Jx
fit)dt = fix) □
tX
5.4 The Fundamental Theorems of Calculus 307
G(x) =
Ja
f f{t)dt
found in Theorem 5.8 satisfies G(x) = F(x) + C for some constant C and all x on the
interval [a, b]. In particular, when x = a, we have
b
f(t)dt = Gib) = F(b) + C = F(b) + [-F(a)] = F(b) - F(a)
Solution
From the second fundamental theorem, we can obtain F\x) by simply replacing / with
x in the integrand fit) — 2t — 3. Thus,
d Cx
F,ix) = — / i2t-3)dt
r — 2x — 3
dx \_J~i
f5 sinn ,
Giz) = / -du
Jz u
reverse the order of integration and apply the second fundamental theorem of calculus
as before:
d
dz Jz
f5 sinn ,
/
u
Cl n d
dz .
r
J5
smudu]
u
d
dz J 5
fz sinu
u
sinz
z
Jo
Solution
2
F\x) Chain rule; u — x
(*2)2 i2x)
2x5
308 Chapter 5 Integration
J-10
2
45
/
correct.
, u sin 2 au
cos au du — — H-:-- + C
3. J (2x+a)dx 4. JI (b
(b-x)dx
— 2 4a
2
46. / a cos2 audu =-1- + + C
5. J ax2, dx . J (x3 + bx2)dx J 4 4a a2
/
6 8
u du 1
47 C
7. f^dx 8 . r'f-dx (a2 - a2 )372 Va^u2
Ji x3 y_2 x2
f du 1 a —a
r9 _ r27 48 = — In - + C
9, / 7x dx 10 . / Tx dx ' J u2^ a*- 2a u +a
Jo Jo u du
11. I
Jo
(5uljr7t2)du 12 . f
Jo
(7x*-\-x/n) dx
49
•/ a' — a
— — s/a1 — 2 + C
50
13' /' x2u dx, a xfi—\ 14. (2x)n dx •/ (In |a |)2 du = a (In |a|)2 — 2a In |a| + 2a + C
17. f
Jo
Vx(x + l)dx 18. f
Jo
\ft(t — \ft)dt
Figure 5.21a is 8 times the area of the shaded region in Figure
5.21b. What is c in terms of a?
f2 x3 +1 „A r4 x:
20
• l
'1 —,ta
X
,2 20-1 ~ yx
/*rV3
v3
6a 05 ddx
21. / -- dx 22
•/
'•/
/ 3
T + x"
(sin2 x + cos2 x) dx 24. J/»7T /4
yr
(sec2 x — tan2 x) dx
27. f
Jo
—dx
x —2
28. f •'x2-l
/ —-dx
Jo x2 + 1
b.
1 +V
In Problems 39-44, find the derivative of the given function.
55. Evaluate
rx _ j
/2 rx
39. F{x) = /I _dt 40 Fix) — / {t+\)fftdt
40.
Jo vM" 1 J-2 | cosx if 0 < x < 7t/2
/ (x) dx where / (x) =
f sinx C2 e f
Jo IX if 7T/2 < X < TC
41. F{t) = / —-dx 42. F(0 = / — dx
7i * J, x
dr [x ,
43. Fix) 44. F(x) = / sec Mantdf
Jx 71 + 3/2 J n! 3 * February 1995, p. 15.
5.5 Integration by Substitution 309
56. a. If Fix) — J ^— 4 j dx, find F so that E(l) = 0. b. Where does g have a relative maximum value on [0, 5]?
c. If g(l) = 1, g(2.5) = -2.5, and g(4) = 4, sketch a rough
b. Sketch the graphs of y = Fix), y = Fix) + 3, and graph of g.
y = F{x) - 1. Q 59. Counterexample Problem The purpose of this problem is to pro¬
c. Find a constant K such that the largest value of vide a counterexample showing that the integral of a product
G(x) — Fix) 4 K is 0. (or quotient) is not necessarily equal to the product (quotient)
of the respective integrals,
57. Let g(x) = / f(t)dt, where / is the function defined by
Jo a. Show that fxsfxdx + if x dx)(f fxdx).
the following graph. Note that / crosses the x-axis at 3 points
x , , f Jxdx
on [0, 2]; label these (from left to right), x = a, x — b, and
x = c.
b. Show that
/ x ^x ^ 3) '
J? x dx
60. Suppose / is a function with the property that fix) = fix)
for all x.
a. Show that /
I fix)dx = fib) — fia).
Fix) dt
v(x)
Recall that according to the chain rule, the derivative of (x2 + 3x + 5)4 is
J fix)dx = J gO)—
du f r —d Giu) i dx - Giu) + C
dx _ dx = ) dx
as required.
Now complete the integration of giu) and, when you are finished, back-substitute to
express u in terms of x:
,9 \
/ 9 «8 du —9 + C = ix2 + 3* + 5)9 + C
Let u — 2x + 7, so du = 2 dx.
ite function f(x) = g(u(x)), it is usually a good idea to pick the “inner” function
u — u(x). In Example 2, we chose u = 2x + 7, NOT the entire power (2x + 7)5.
When making a substitution u — u(x), remember that du = u'(x) dx, so you can¬
not simply replace dx by du in the transformed integral. For example, when you use
the substitution u = sinx, you have du = cos x dx and then replace the entire expres¬
sion cos x dx by du in the transformed integral, as shown in the following example.
Solution
du
Fet u = 4 - 2 cos 6, so — = —2(— sin 6) = 2 sin 0 and dO — -; substitute:
2 sin#
f t, du 1 f 3
j (4 — 2cos#)3 sin9 dO / u sin#-= - / u~ du
J 2sin# 2J
- ( — ^ + C — -(4 — 2cos#)4 + C
2V4J 8
Note: We will use a shorthand notation for substitution by placing a “box” under the
integrand as follows:
d6
j{4 — 2 cos #)3 sin # d9
I m3 sin #
/
Evaluate / xe4 x dx.
u — 4 — x2\ du = — 2x dx
= -\eu + C
= -\e4~x2 + C
WARNING Y^en using the method of substitution, all terms involving x in the original
integral f f(x) dx must be converted into terms involving u in the transformed integral
f g(u) du. That is, after you have made your substitution and simplified, there should be no
"leftover" x-values in the integrand.
Solution f , , f ,/du\ 1 f ,
J x(4x — 5) ax = J xu J = - J xu du
t
u = 4x — 5; du = Adx There is a leftover a-value.
We are not ready to integrate until the leftover x-term has been eliminated. Because
u = 4x — 5, we can solve for x:
m T5
x =
4
so
1
4 / xa3 + 5u )
1
+ C = -L(4x - 5)5 4- |-(4x - 5)4 + C U
Note: Always introduce the constant 16 ol) o4
of integration immediately after the
last integration.
The next example derives formulas that will be used many times throughout the
text, often enough, in fact, that you should remember them or mark them for future
reference.
1
-/( cosx
(sinxdx) Let u — cosx, du = — sinx dx.
du
u
= — In \u\ -f C
= — In |cosx| + C
/
This can also be written as In |(cosx) 11 + C = In |secx| + C. Similarly,
[ cos
cosxx
cotx dx — I - dx = In |sinx| + C.
J sinx
Here is a problem in which the rate of change of a quantity is known and we use
the method of substitution to find an expression for the quantity itself.
Solution
Because the rate at which the volume V is changing is dV/dt,
u = 3t + 1; du — 3 dr
V = i J ul/2du
V = |(3r + l)3/2 + C
Notice that the definite integral eliminates the need for finding C. Furthermore, the fol¬
lowing theorem eliminates the need for returning to the original variable (even though
sometimes it is more convenient to do so).
Proof Let F(u) = f f(u) du be an antiderivative of /(«)• Then the first fundamen¬
tal theorem of calculus tells us that
b
f[u(x)]u\x)dx = F[u(x)]\xxZba = F[u(b)] - F[u(a)]
U(b)
f{u)du = F{u)= F[u(b)] - F[u{a)]
Thus,
□
314 Chapter 5 Integration
We now have two valid methods for evaluating a definite integral of the form
b
f[u(x)]u'(x) dx:
1. Use substitution to evaluate the indefinite integral and then evaluate it between the
limits of integration x — a and x = b.
2. Change the limits of integration to conform with the change of variable u = u(x)
and evaluate the transformed definite integral between the limits of integration
u = u(a) and u = u{b).
These two methods are illustrated in Example 8.
Solution
Method I: We first use a substitution with an indefinite integral.
Let u = 4x — 5; du =4dx
Now evaluate the definite integral of fix) between x — l and x = 2 by using the
fundamental theorem of calculus:
= £(81) - £<!)
=5
Method II: This is similar to method I, except that as part of the integration process,
we change the limits of integration when we are doing the substitution.
Let u — 4x — 5; du — 4dx.
If .v = 2, then u — 4(2) — 5 = 3.
4 4
WARNING w , ...
— 5)3 dx You cannot change variables
and keep the orginal limits of integration
t t because the limits of integration are
If JC - 1, then u = 4(1) — 5 = — 1. expressed in terms of the variable of
integration.
1
4
= £(81-1) = 5
5.5 Integration by Substitution 315
r
f4
1. a. / (2r + 4) dt b. / (2r + 4)~1/2 dt
Jo Jo 41. tan 2x dx 42. x2(x3 + 9)1/2 dx
px/2 2 [
2. a. / sin Odd b. / sin 20 dO 0.58 ,12 5,000
r
Jo Jo 43.
j: 1 + e -0.2x
dx 44.
fi + 10e-'/5
3. a. / cost dt b. / tcosrdt 45. HISTORICAL QUEST Johann Pe¬
Jo Jo ter Gustav Lejeune Dirichlet was a pro¬
fo
/*4 fessor of mathematics at the Univer¬
4. a. / y/x dx b. 1 y/—x dx
Jo J- 4 sity of Berlin and is known for his role
/> 16 r° in formulating a rigorous foundation
5. a. / ffx dx b. / f/—x dx for calculus. He was not known as a
Jo J-16
good teacher. His nephew wrote that
6. a. J x(3x2 — 5) dx b. J x(3x2 -5)50dx the mathematics instruction he received
LEJEUNE DIRICHLET
from Dirichlet was the most dreadful ex¬
7. a.
J
f x2V2x3 dx b. f x2\/2xi — 5 dx
perience of his life. Howard Eves tells
of the time Dirichlet was to deliver a lec¬
1805-1859
*7 (2x + 3 y dx
107 V31 -5dt
BaccluAc itineAA, 9
11. j [tan(x2 + 5x + 3)(2x + 5)] dx 12. J (11 — 2x) cannot tcctuAc today,
f xdx f x2 dx
B&cauAc o$ ifhveAA. 9
21.
1 2x2 + 3 22-J x3 + 1 rOi (/ ErcLev)
f 4x dx cannot tedtuAc txpebay.
23. 1 x\j2x2 + 1 dx EcuHr
2x + 1
QddcAtet
j" e^ dx
25. 1 y[xex'J* dx 26. j
X2?3
27. 1 x(x2 + 4)1/: dx 28. J( x3(x2 + 4)l/2 dx Michaelmas and Easter were school holidays, and 1
Frdor (Friedrichsd’or) was the customary honorarium for a
(i«x+odx semester's worth of lectures. ■
30. j
29.
/¥* f
x+ 1
dx
a. What is the answer when you integrate the student-
r dx doctored note?
31. 32. J
f ■s/x(dx + 7) x2/3(^ + l) b. The so-called Dirichlet function is often used for coun¬
C e' dt r e^ dt terexamples in calculus. Look up the definition of this
33. 34. j
1 e< + 1 s/t(e4> + 1) function. What special property does it have?
Evaluate the definite integrals given in Problems 35-44. Approxi¬ Q Find the area of the region under the curves given in Problems
mate the answers to Problems 43 and 44 to two significant digits. 46-49.
35.
Jo
f 5x2 dx
2x3 + 1 -r sfx
dx
46. y = tdt1 + 9 on [0, 4] 47. y =
tz
1
- on
t
316 Chapter 5 Integration
48. y = x(x — l)l/3 on [2, 9] 49. y = ■ - — on [1, 3] 58. A particle moves along the r-axis in such a way that at time t,
a/x2 4- 1 its velocity is v(t) = /2(/3 — 8)1/3.
50. a. In Section 5.3 (Problem 40), you were asked to prove that
a. At what time does the particle turn around?
if / is continuous and odd [that is, f(—x) = —/(x)] on
b. If the particle starts at a position which we denote as 1,
the interval [—a, a], then
where does it turn around?
59. A rectangular storage tank has a square base 10 ft on a side.
/(x) dx = 0
Water is flowing into the tank at the rate modeled by the func¬
tion
b. In Section 5.2 (Problem 33), you were asked to find a coun¬
terexample or informally show that if / is continuous and R(t) = /(3/2 + 1)~1/2 ft3/s
even [/(—x) — /(x)] on the interval [—a, a], then
at time t seconds. If the tank is empty at time t = 0, how
much water does it contain 4 sec later? What is the depth of
the water (to the nearest quarter inch) at that time?
60. Journal Problem College Mathematics Journal.* ■ Evaluate
For this problem, use the properties of integrals to prove
each of these statements.
/[(x2-i)(x + i)r2/3</x
Use the results of Problem 50 to evaluate the integrals given in
Problems 51-54.
/
Jo
7T/2
Jo
cosx dx = I
/»U
si
sinx dx
between 7:00 A.M. and 7:00 P.M. when the highest level
of ozone occurs. What is the highest level?
c. Use your graphing utility or another utility of your calcu¬
-jt/2 J—71
lator to determine a second time during the day when the
56. The slope at each point (x, y) on the graph of y = F(x) is
ozone level is the same as it is at 11:00 A.M.
given by x(x2 — 1)1/3, and the graph passes through the point g
(3, 1). Use this information to find F. Sketch the graph of F. O62- A logistic function is one of the form Q{t) =-.
1 + Ae~"
57. The slope at each point (x, y) on the graph of y = F(x) is
2x Evaluate / Q(t)dt.
given by i -o -0. What is F(x) if the graph passes through
1 - 3x2'
(0, 5)) *Vol. 20, No. 4, Sept. 1989. p. 343. Problem by Murray Klamkin.
because
dy d d 9
— = — (y) = —(dr) — 2x
dx dx ' dx
Moreover, because any solution of this equation must be an indefinite integral of 2x, it
follows that
, = /2x<fa=,» + C
Solution
Ax — 3 y~ 10
d
— (4x-3y2) — (1°) Take the derivative of both sides.
dx dx
dy
4 — 6y 0 Don’t forget the chain rule,
dx since y is a function of x.
dy A _ 2 dy
Solve for —.
dx 6y 3y dx
Solution
Let R(t) denote the total revenue up to time t. Then the rate of change of revenue is
d ft
—, the number of dollars received per barrel is p(t) = 30 + 0.3y/t, and the number
dt
of barrels sold per day is 300. Thus, we have
dR dR
This is actually a statement of the chain rule: —, where B denotes the
dt 1b dt
number of barrels extracted and R denotes the revenue. It is often helpful to use the
chain rule in this way when setting up differential equations.
We solve this differential equation by integration:
Since /?(0) = 0, it follows that C — 0. We also are given that the well will run dry in
3 years or 1,095 days, so that the total revenue obtained during the life of the well is
DIRECTION FIELDS
We can use slope fields to help with a visualization of a differential equation. In Section
5.1 we looked at small segments with slope defined by the derivative of some function.
The collection of all such line segments is called the slope field or direction field of
the differential equation. Today, with the assistance of computers, we can sometimes
draw the direction field in order to obtain a particular solution (or a family of solutions)
as shown in the following example.
Solution
a. The initial value y(2) = 1 means the solution passes through (2, 1) and is shown in
Figure 5.23a.
b. Since y(0) = 1, the solution passes through (0, 1), as shown in Figure 5.23b.
5.6 Introduction to Differential Equations 319
dy_ _ g(x)
dx f(y)
To solve such an equation, first separate the variables into the differential form
f(y)dy = g(x)dx
J f(y)dy = J g(x) dx
A differential equation expressible in the form dy/dx = g(x)/f(y) is said to be sepa¬
rable. This procedure is illustrated in Example 4.
dy_ X
Solution
dxy
ydy = x dx
Jydy
-h
l\y2 + Ci ‘x2
- 2x
—
A process is said to undergo exponential change if the relative rate of change of the
process is modeled by a constant; in other words,
Q'(t) dQ_
or = *(2(0
0(0 dt
If the constant k is positive, the exponential change is called growth, and if k is neg¬
ative, it is called decay. Exponential growth occurs in certain populations, and expo¬
nential decay occurs in the disintegration of radioactive substances.
To solve the growth/decay equation, separate the variables and integrate both sides.
dQ
— kQ
dt
n-h*
In | <21 = kt + C\
ekteCl = 0
Thus, 0 = Cekt, where C = ecK Finally, if we let 0O be the initial amount (that is,
when t — 0), we see that
Qo = Ce° or C = f2o, so 0(0 = Qoekt
Growth/Decay Equation
The growth/decay equation of a substance is
0(0 = Qoekt
where 0(0 is the amount of the substance present at time t, Qo is the initial
amount of the substance, and k is a constant that depends on the substance. If
k > 0, it is a growth equation; if A: < 0, it is a decay equation.
th (half life)
The graph of 0 = Qoekt for k < 0 is shown in Figure 5.24. In Figure 5.24 we
have also indicated the time th required for half of a given substance to disintegrate.
Figure 5.24 The decay curve for a The time th is called the half-life of the substance, and it provides a measure of the
radioactive substance substance’s rate of disintegration.
Solution
From the decay equation,
0(0 Qoekt
0(0
00
1
ek(600) Half-life is 600 years.
2
A(600) Ini Definition of logarithm
In i
* -0.0011552453
600
5.6 Introduction to Differential Equations 321
Next, to see how much of a 50-g sample will remain after 125 yr, substitute Q0 = 50,
k = —0.0011552453, and t — 125 into the decay equation:
R = R0ekt
ln(1/2)
where k — and R0 is the ratio of 14C to l2C in the atmosphere. By comparing
5,730
R(t) with Rq, scientists can estimate the age of the object. Here is an example.
Solution
The age of the specimen is the value of t for which R{t) = 0.20Ro:
0.20R0 R0ekt
0.20 ekt
kt In 0.20
1 ln(l/2)
t -In 0.2 k —
k 5,730
13,304.64798
ORTHOGONAL TRAJECTORIES
A curve that intersects each member of a given family of curves at right angles is called
an orthogonal trajectory of that family and is shown in Figure 5.25. Orthogonal fami¬
lies arise in many applications. For example, in thermodynamics, the heat flow across a
planar surface is orthogonal to the curves of constant temperature, called isotherms. In
the theory of fluid flow, the flow lines are orthogonal trajectories of velocity potential
curves. The basic procedure for finding orthogonal trajectories involves differential
equations and is demonstrated in Example 7.
*Carbon dating is used primarily for estimating the age of relatively “recent” specimens. For example, it
was used (along with other methods) to determine that the Dead Sea Scrolls were written and deposited
in the Caves of Qumran approximately 2,000 years ago. For dating older specimens, it is better to use
techniques based on radioactive substances with longer half-lives. In particular, potassium 40 is often used
as a “clock” for events that occurred between 5 and 15 million years ago. Paleoanthropologists find this
substance especially valuable because it often occurs in volcanic deposits and can be used to date specimens
trapped in such deposits. See Mathematical Essay 5 on page 352.
Orthogonal
trajectory
curve
Family of
curves
Solution
We are seeking a family of curves. Each curve in that family intersects each curve in
the family xy = C at right angles, as shown in Figure 5.26. Assume that a typical
point on a given curve in the family xy — C has coordinates (x, y) and that a typical
point on the orthogonal trajectory curve has coordinates (X, Y).*
xy = C
dy
x +y = 0 Product rule
dx
dy _ —y
dx x
* We use uppercase letters for one kind of curve and lowercase for the other to make it easier to tell which
curve is being mentioned at each stage of the following discussion.
5.6 Introduction to Differential Equations 323
dY i i X x
~dX ~ dy -y y Y
dx x
According to this equation, the coordinates (X, Y) of the orthogonal trajectory curve
satisfy the separable differential equation
dY _ X
dX ~ 7
discussed in Example 4. Using the result of that example, we see that the orthogonal
trajectories of the family xy — C are the curves in the family
X2 — Y2 = K
where K is a constant. The given family of curves xy — C and the family of orthogonal
trajectory curves X2 — Y2 — K are shown in Figure 5.26. ■
Consider a tank that is filled with a fluid being slowly drained through a small, sharp-
edged hole in its base, as shown in Figure 5.27. By using a principle of physics known
as Torricelli’s law,1' we can show that the rate of discharge dV/dt at time t is propor¬
tional to the square root of the depth h at that time. Specifically, if all dimensions are
given in terms of feet, the drain hole has area Ao, and the height above the hole is h at
time t (seconds), then
is the rate of flow of water in cubic feet per second. This formula is used in Example 8.
Figure 5.27 The flow of a
fluid through an orifice
EXAMPLE 8 Fluid flow through an orifice
A cylindrical tank (with a circular base) is filled with a liquid that is draining through
a small circular hole in its base. If the tank is 9 ft high with radius 4 ft and the drain
hole has radius 1 in., how long does it take for the tank to empty?
Solution
Because the drain hole is a circle of radius ~ ft (= 1 in.), its area is nr2 = tv (^) ,
and the rate of flow is
dV
dt
Because the tank is cylindrical, the amount of fluid in the tank at any particular
time will form a cylinder of radius 4 ft and height h. Also, because we are using
g — 32 ft/s2, it is implied that the time, t, is measured in seconds. The volume of such
a liquid cylinder is
fTorricelli’s law says that the stream of liquid through the orifice has velocity -J2gh, where g = 32 ft/s2 is
the acceleration due to gravity and h is the height of the liquid above the orifice. The factor 4.8 that appears
in the rate of flow equation is required to compensate for the effect of friction.
324 Chapter 5 Integration
2 hx/1 + Ci = -0.002 k + C2
2 sfh = -0.002k+ C
2a/9 = —0.0021(0) + C, or C = 6
2 Vh = -0.002k + 6
Consider a projectile that is launched with initial velocity v0 from a planet’s surface
along a direct line through the center of the planet, as shown in Figure 5.28. We will
find a general formula for the velocity of the projectile and the minimal value of u0
required to guarantee that the projectile will escape the planet’s gravitational attraction.
We assume that the only force acting on the projectile is that due to gravity, al¬
though in practice, factors such as air resistance must also be considered. With this
assumption, Newton’s law of gravitation can be used to show that when the projectile
is at a distance s from the center of the planet, its acceleration is given by the formula
Figure 5.28 A projectile launched
from the surface of a planet -gR2
a = —2
sz
where R is the radius of the planet and g is the acceleration due to gravity at the planet’s
surface (see Problem 69).*
Our first goal is to express the velocity v of the projectile in terms of the height 5.
Because the projectile travels along a straight line, we know that
dv ds
a = — and v = —
dt dt
and by applying the chain rule, we see that
dv dv ds dv
dt ds dt ds
* According to Newton’s law of gravitation, the force of gravity acting on a projectile of mass m has magni¬
tude F — mk/s2, where k is constant. If this is the only force acting on the projectile, then F = ma, and we
have ma = F = mk/s2. By canceling the m’s on each side of the equation, we obtain a = k/s:.
5.6 Introduction to Differential Equations 325
v dv = —gR~s ~ ds
vdv = / —gR2s~2ds
jv2 + Ci = gR~s 1 + C2
v2 = 2gR2s~l + C
To evaluate the constant C, recall that the projectile was fired from the planet’s surface
with initial velocity uq. Thus, v = vq when .v = R, and by substitution
so
v2 = 2gR2s 1 + i»o - 2gR
Because the projectile is launched in a direction away from the center of the planet,
we would expect it to keep moving in that direction until it stops. In other words, the
projectile will keep moving away from the planet until it reaches a point where v — 0.
Because 2gR2s~l > 0 for all 5 > 0, v2 will always be positive if Vq - 2gR > 0. On
the other hand, if — 2gR < 0, then sooner or later v will become 0 and the projectile
will eventually fall back to the surface of the planet.
Therefore, we conclude that the projectile will escape from the planet’s gravita¬
tional attraction if v$ > 2gR\ that is, v0 > J2gR. For this reason, the minimum speed
for which this can occur, namely,
u0 = y/2 gR
is called the escape velocity of the planet. In particular, for the earth, R = 3, 956 mi
and g = 32 ft/s2 = 0.00606 mi/s2, and the escape velocity is
„ , , „ , dy x
1. If x2 + y2 — 7, then —- = —•
dx y
Find the particular solution of the first-order linear differential 17. (0,0) 18. (3,3)
equations in Problems 9-14. A graphical solution within its di¬ 1 1 1 1 i rrnr 1 1 1 11 1 1 1 i i
f y ^.
•
1 1 1 1 1 'on! 1 1 1 11 1 1 1 i i - ^
<* / . V \ ' \
rection field is shown. 1 1 i i i i.i 1 1 1 11 1 1 1 i i
• / / ^ " ■ \ \ > \
i i 1 I / / / /1 1 1 1
/ /1 1 1 1 i i r / / ✓< ^ \ \ '
i / 1 1 ! I1.-5/ / /
1 1 i / 1 1 L / / /1 1 1 1 i i _ \ \ \ '
i i 11 1 / / /V0/ // / // / 1 1 i i ' / / / • \ \ \ \ 3 \
< f / / / . \ V \ \ 1 \
i i i / / / / A-.- ss / // / 1 1 i i
y f / 1 \ '
\ \!
\ n!
dx T l i / / / / 1 1 11
' /■ \ ’ J ;
1 l ! / / A - LLaI'LllLL
passing through (2, 2) passing through (—2, —9) /ii in
— t i 0 1.5 2.0 \x {-4
r2 a U q'-Xf ^ / >
1 1 1 I l 1 ■ i l)V t 11 1 1 1 1 1 1 7 1 i / 1 / / / ,-5' // / / 1 1 1 1 \ \ V V ' / / f
1 1 1 1 i l i i 11 1 1 1 1 1 1 1 1 i / / / / /\ 1 1 1 1 l 1 \ \ v / / / /
1 / \ \
1 1 1 1 i i i i 4>{ 1 1 11 i i 1 1
1 1 1 1 i i i i l l 1 1 1 1 i i 1 1 T/ 1 1 1 TlP/ 7 / / 1 1 1 Mil \ k f / / V
11 1 1 1 1 II- O / / / / 1 1 ii a i \ , \ \ / r / / / /
1 1 1 1 i i i i i 1 1 1 i i 1 1 T I 1 1 11 i rLAf / / / 1 1 1 MM \ s \ N / / /
1 1 1 1 1 1 1 ! I 1 1 1 1 i | 1 1 1 1 1 11 1 1 1 1 1 II II \ A *s / y
1 1 1 1 i i 1 1
ih1
ID j 1 1 1 1 i i 1 1 1 LVft1- 1 V \ \ N - -4“ " s X >
1 1 i i 1 1 1 1 l 1 i i 1 1 1 1 Tl tt4ui 1 1 ri I 1 T 1 1 1
1 1 • /
1 1 1 1 1 1 Mill 1 1 i i 1 1 LL 1 1 \ 'V \ A ' s'
1 1 1 1 1 I ! 1
i i 1 1 i i 1 1
1 1 1 1 1 | 1 1
1 1 I i 1 1
l\ I -1 5/-1 (5 K
/ ) ''
i it ' W I1 5; i Oi 7
19. (1,0) 20. (0,0)
•Jjl
1
5 i \ \ \ \ ‘ ! / ' f 1 1 1 r y* i \ 1 1 ! 1 ! i
/ / /////j// 1 / / / / / / / /
//x//\// w/y 1 '/S // 1 \ \ \ \ \ 1 s / Y / / / / / t 1 5 1 I i i / \ 1 1 1 I t
XXXXXjXV sHN W XXs X X
1
\ \ \ \ \ \ \!\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ ' / / / / f 1 1 t ) 1 ; i i i / - * \ \ 1 1 1 1 t
\ \ \ \ \ \ \ \ V \ \ \ \ \ \ / / ■ 1 t 5 * i i > ^ V \ i t t (
\ \
1
\ \ t \ \ \ 1 1 1 1 l l \ , \ \ \ 1 ' / / / / ’ / f 1 t 1 t S \ 1 \ 1 l 1 1
J 111 i V
1 \ » ' l 1 1 1 \ \ 1 \ t \ i i V V ’. \ \ \ X ". X X * / / / / f / / 5 t f 1 i i i t / / . ■M \ \ t \ l
\ \ \ \ ( t ?! l 1 \ t 1 1 \ 1 i t [■-V "V
\ v . \ \ \ x * ' y s z' / f / / f t 1 t / i i t i f / y
\ \ :: ' t L / / ,
dy_ 5 -1 o 2..
12.
i 0 1 5 lv \ \~ . . i c ,
11.
A = gx+y \ ~1 - /, \ *
dx dx / X ' / X s' ■ — XX »x X \ \ ’ , \ \ \ Ns • - / / 1 t » l
' / / y s > X X
passing through (n, 1) passing through (0, 0) / / \ X 1 t V o rmr f i I 1
/ / ' / / s > S X X1 \ \ . \ \ 5 \ \ \ v \ ' 1 1 .
y , $ 5 5 1 / / ’ / / / /* - ’ s' -- «. X X \ \ \ \ i ! \ l \ \ ■' ~ / i 1 1 t
. . \ \ 5 \
8- i i \ i f / ! / / / s X s X \ \ \ . \ \ 1 1 ! 1 1 \ \ / //til > i 1 1
/ / / / / / • / -- X. , \ N
I i 1 J 1 \ \ / t
* 1 l 5 l 1 1 1
! i
J * Find the general solution of the separable differential equations
—
l, * / / given in Problems 21-30. Note that for some problems, the gen¬
,
S7 X eral solution will be a relationship between x and y, rather than
-8 /
1 ! y = fix).
dy_ dy_
21. 3xy 22.
II
dx dx
dy_ dy_
23. — *VT^T2 24. = CV “ 4)2
dy j r- dx y dx
— =y ffx
dx dy
25. xy dx + ffxy dy = 0 26
y
passing through (9, — yg)
dx Vl - 2x2
y. 1 1 11 1 1 1
!! !! 1 1 1 1 1 1 1 sinx
1 1 I 1 1 1 1 1 1 1 1
! 1 1 1 1 1 1 i 28. x2 dy + sec y dx
i i 1 1 1 1 1 1 1 l l cos y
i i 1 l 1 1 / 1 1 1 1
i i ii 1 1 1 111 1
11 l l 1 1 11 l l l lnx dy_
n u///> // t// 29. 30. = ey~x
_ __
A A -4 •
" cmill:
f' '//V/: i//i
7^7
/ / 11 i i 1 / /
4- 1 11 11 11 11 11 Tl
11 f f 1 In Problems 31-34, find the general solution of the given differen¬
1 1 1
1 1 1 1 1 1 1 1 1 1 1 tial equation by using either the product or the quotient rule.
1 1 II 1 m
1 1 1 1 1 1 1 1 111
I 1 1 1 1 1 1 1 111 x dy — y dx
1 1 1 1 1 1 1 1 111 31. x dy + y dx = 0 32. = 0
In each of Problems 15-20, sketch the particular solution passing 33. y dx — x dy, x > 0, y > 0 34. x2y dy + xy2 dx —0
through the given point for the differential equation whose direc¬
tion field is given. Q Find the orthogonal trajectories of the family of curves given in
Problems 35—42. In each case, sketch several members of the given
15. (0, 1) 16. family of curx’es and several members of the family of orthogonal
1111111 trajectories on the same coordinate axes.
/ i 11111
\i)Ui 111
I I //l// 35. the lines 2x — 3y = C 36. the lines y = x + C
•t t # t i f f
/ ///l//
////////
/ J./if t
//
37. the cubic curves y = x3 + C
38. the curves y = x4 + C
Modeling Problems Write a differential equation to model the Swiss Federal Institute of Technology were permitted to test it.
situation given in each of Problems 43—48. Be sure to define your It was found that the cloth contained 92.3% of the original l4C.
■
unknown function first. Do not solve.
According to this information, how old is the Shroud?
43. The number of bacteria in a culture grows at a rate that is pro¬
54. A cylindrical tank of radius 3 ft is filled with water to a depth
portional to the number present.
of 5 ft. Determine how long (to the nearest minute) it takes
44. A sample of radium decays at a rate that is proportional to the to drain the tank through a sharp-edged circular hole in the
amount of radium present in the sample. bottom with radius 2 in.
45. The rate at which the temperature of an object changes is pro¬ 55. Rework Problem 54 for a tank with a sharp-edged drain hole
portional to the difference between its own temperature and that is square with side of length 1.5 in.
the temperature of the surrounding medium. 56. Experiment A rectangular tank has a square base 2 ft on a side
46. When a person is asked to recall a set of facts, the rate at which that is filled with water to a depth of 4 ft. It is being drained
the facts are recalled is proportional to the number of relevant from the bottom of the tank through a sharp-edged square hole
facts in the person’s memory that have not yet been recalled. that is 2 in. on a side.
Hint: Let Q denote the number of facts recalled and N the a. Show that at time f, the depth h satisfies the differential
total number of relevant facts in the person’s memory. equation
47. The rate at which an epidemic spreads through a community
— = --Vh
of P susceptible people is proportional to the product of the dt 30
number of people who have caught the disease and the number
b. How long will it take to empty the tank?
who have not.
c. Construct this tank and then drain it out of the 4 in.2 hole.
48. The rate at which people are implicated in a government scan¬ Is the time that it takes consistent with your answer to part
dal is proportional to the product of the number of people al¬
b?
ready implicated and the number of people involved who have
57. A toy rocket is launched from the surface of the earth with ini¬
not yet been implicated.
tial velocity v0 — 150 ft/s. (The radius of the earth is roughly
49. What do you think the orthogonal trajectories of the family of 3,956 mi, and g = 32 ft/s2.)
curves x2 — y2 = C will be? Verify your conjecture. Hint:
a. Determine the velocity (to the nearest ft/s) of the rocket
Take another look at Example 7.
when it is first 200 feet above the ground. (Remember, this
50. The following is a list of six families of curves. Sketch several is not the same as 200 ft from the center of the earth.)
members of each family and then determine which pairs are
b. What is 5 when v = 0? Determine the maximum height
orthogonal trajectories of one another.
above the ground that is attained by the rocket.
a. the circles x2 + y2 - A b. the ellipses 2x2 + y2 = B2
58. Determine the escape velocity of each of the following heav¬
c. the ellipses x2 + 2y2 — C d. the lines y = Cx enly bodies:
e. the parabolas y2 - Cx f. the parabolas y — Cx2 a. moon (R = 1,080 mi; g = 5.5 ft/s2)
51. The Dead Sea Scrolls were written on parchment in about b. Mars (R — 2,050 mi; g = 12 ft/s2)
100 B.C. What percentage of 14C originally contained in c. Venus (R = 3,800 mi; g = 28 ft/s2)
the parchment remained when the scrolls were discovered in
59. Population statistics indicate that t years after 1990, a certain
1947?
city was growing at a rate of approximately 1,500/_I/2 people
52. Tests of an artifact discovered at the Debert site in Nova Scotia per year. In 1994, the population of the city was 39.000.
show that 28% of the original 14C is still present. What is the
a. What was the population in 1990?
probable age of the artifact?
b. If this pattern continued, how many people were living in
53. HISTORICAL QUEST The
the city in the year 1999?
Shroud of Turin is a rectangular
linen cloth kept in the Chapel of
60. The radius of planet X is one-fourth that of planet Y, and the
acceleration due to gravity at the surface of X is eight-ninths
the Holy Shroud in the cathedral
that at the surface of Y. If the escape velocity of planet X is
of St. John the Baptist in Turin,
6 ft/s, what is the escape velocity of planet T?
Italy. It shows the image of a
man whose wounds correspond 61. A survey indicates that the population of a certain town is
with the biblical accounts of the growing in such a way that the rate of growth at time t is pro¬
crucifixion. portional to the square root of the population P at the time. If
In 1389, Pierre d'Arcis, the the population was 4,000 ten years ago and is observed to be
Bishop of Troyes, wrote a memo 9,000 now, how long will it take before 16,000 people live in
to the Pope, accusing a col¬ the town?
league of passing off "a cer¬ 62. A scientist has discovered a radioactive substance that disinte¬
tain cloth, cunningly painted" grates in such a way that at time t, the rate of disintegration is
as the burial shroud of Jesus proportional to the square of the amount present.
Christ. Despite this early testimony of forgery, this so-called a. If a 100-g sample of the substance dwindles to only 80 g in
Shroud of Turin has survived as a famous relic. In 1988, 1 day, how much will be left after 6 days?
a small sample of the Shroud of Turin was taken and scien¬
b. When will only 10 g be left?
tists from Oxford University, the University of Arizona, and the
328 Chapter 5 Integration
63. The shape of a tank is such that when it is filled to a depth of is g. According to Newton’s law of gravitation, the force of
h feet, it contains V = 9jt/73 ft3 of water. The tank is being gravity acting on a projectile of mass m has magnitude
drained through a sharp-edged circular hole of radius 1 in. If
the tank is originally filled to a depth of 4 ft, how long does it
take for the tank to empty?
64. A rectangular tank has a square base 4 ft on a side and is 10 ft where k is a constant. If this is the only force acting on the
high. Originally, it was filled with water to a depth of 6 feet, projectile, then F = ma, where a is the acceleration of the
but now is being drained from the bottom of the tank through projectile. Show that
a sharp-edged square hole 1 in. on a side.
a. Find an equation involving the rate dh/dt. -gR2
2
b. How long will it take to drain the tank? a = - ~
Sz
In Section 4.2 we established a very useful theoretical tool called the mean value theo¬
rem, which said that under reasonable conditions, there is at least one number c in the
interval (a, b) such that
fib) - f(a)
= f\c)
b—a
The mean value theorem for integrals is similar, and in the special case where
fix) >0 it has a geometric interpretation that makes the theorem easy to understand.
5.7 The Mean Value Theorem for Integrals; Average Value 329
In particular, the theorem says that it is possible to find at least one number c on the
interval (a, b) such that the area of the rectangle with height f(c) and width (b — a)
has exactly the same area as the region under the curve y = /(x) on [a, b]. This is
illustrated in Figure 5.30.
Proof Suppose M and m are the absolute maximum and the absolute minimum of
The shaded rectangle has the same area
as the region under the curve y = f(x) on /, respectively, on [a, b]. This means that
[a. b]. m < /(x) < M when a < x < b
Ja
/(x) dx < I
fa
M dx Dominance rule
integrals
m(b — a) < f
Ja
f{x)dx < M(b — a)
—
m < f
b - a Ja
f(x) dx < M
Because / is continuous on the closed interval [a, b] and because the number
1 fb
I = —— / f{x)dx
b - a Ja
lies between m and M, the intermediate value theorem (Theorem 2.6 of Section 2.3)
says that there exists a number c between a and b for which /(c) = /; that is,
7~ [ f(x)dx = f(c)
b -a Ja
f(x)dx = f (c)(b — a) □
Ja
The mean value theorem for integrals does not specify how to determine c. It
simply guarantees the existence of at least one number c in the interval. However,
Example 1 shows how to find a value of c guaranteed by this theorem for a particular
function and interval.
Solution
There are many practical situations in which one is interested in the average value of
a continuous function on an interval, such as the average level of air pollution over
a 24-hour period, the average speed of a truck during a 3-hour trip, or the average
productivity of a worker during a production run.
You probably know that the average value of n numbers xi, X2, ■ ■ ., xn is
*1 + X2 + • • • + xn
n
but what if there are infinitely many numbers? Specifically, what is the average value
of f(x) on the interval a < x < bl To see how the definition of finite average value
can be used, imagine that the interval [a, b] is divided into n equal subintervals, each
of width
b —a
Ax —-
n
For k = 1, 2, ..., n, let xk be a number chosen arbitrarily from the &th subinterval.
Then the average value AV of f on [a, b) is estimated by the sum
n Xk
b —a 1 1
Because Ax =-, we know that Ax and
n b —a
-| n 1 n
1 ^ 1
s- = -E
k= 1
/«> b —a
Ax
- k= 1 D a k= 1
f(x*k)Ax
b —a
The sum on the right is a Riemann sum with norm ||P|| =-. It is reasonable to
n
expect the estimating average Sn to approach the “true” average value AV of f(x) on
[a, b] as n -> +00. Thus, we model average value by
AV= n^+00
lim r-—y2f(xk)Ax
b—a = Z~~f
b — a Ja fWdx
Average Value If / is continuous on the interval [a, b], the average value (AU) of / on this
interval is given by the integral
1 rb
AV =- / f(x)dx
b — a Ja
kilometers per hour, where t is the number of hours past noon. Compute the average
speed of the traffic between the hours of 1.00 P.M. and 4:00 P.M.
5.7 The Mean Value Theorem for Integrals; Average Value 331
Solution
Our goal is to find the average value of S(t) on the interval [1,4]. The average speed
is
4
(2t3 -2lt2 + 60t + 20)dt -t4 - It3 + 3012 + 20r
2 1
1
Figure 5.32 Graph of S and the f(x)dx = f (c)
average value b —a
which says that the average value of a continuous function f on [a, b] equals the value
of f for at least one number c between a and b. This is quite reasonable since the inter¬
mediate value theorem for continuous functions assures us that a continuous function
/ assumes every value between its maximum M and minimum m, and we would ex¬
pect the average value to be between these two extremes. Example 3 illustrates one
way of using these ideas.
A graph of this formula is shown in Figure 5.33. Find the average temperature over the
time period from 2:00 A.M. to 2:00 P.M., and find a time when the average temperature
actually occurs.
Solution
We wish to find the average temperature T on the interval [2, 14] (because 2 p.m. is 14
hours after midnight). The average value is
14-2 AL 7 J 12 7 3 J
1 I” 587 1,415
-3.2857143
12 [H ~ 21
Thus, the average temperature on the given time period is approximately 3.3°C below
zero. To determine when this temperature actually occurs, solve the equation
37 = (jc — 13)2
The first value is to be rejected because it is not in the interval [2, 14], so we find that the
average temperature occurs 6.917 hr after midnight, or at approximately 6:55 A.M.
332 Chapter 5 Integration
was developed by the United States Bureau of the Census to represent the population
of the United States (in millions) during the period 1790-1990. Time t in the formula
is the number of decades after 1790. Thus, t = 0 for 1790, t = 20 for 1990. Use
this formula to compute the average population of the United States between 1790 and
1990. When did the average population actually occur?
Solution
The average population is given by the integral
1 f-u
’20 202.31 dt -a314r
Multiply by
“ 20 - 0 Jo0 Tr +£ 3.938—0.314r ,0.314r'
20 202.3le0-314'
dt Let u = e0,314r -\- g3.938.
= To[ + e 3.938
du = 0.314e0 314r dt.
1 '202.31 r!=20 du
20 ^ 0.314 Jt= 0 M
1 '202.31
In |^0314' 4- ^3'938|[q° 77.7827445296
20 0.314
To find when the average population of 77.7827 million actually occurred, solve
P(t) = 77.7827:
202.31
77.7827
1 g3.938—0.314/
„3.938—0.314/ 202.31
1
77.7827
202.31
3.938 -0.314? = In - 1
11.1827
t » 11.043
The average population occurred approximately 11 decades after 1790, around the year
1900. ■
f f(x)dx = f{c)(b - a)
c on [a, b] so that the area of the rectangle is equal to the area of
the given region.
34. Suppose a study indicates that t years from now, the level of
13. y = x2 + 2x + 3 on [0, 2] 14. y = — on [0.5, 2]
x1 carbon dioxide in the air of a certain city will be
Lit) = te-°ml
2, OOO
Q(t) ~ 1 + 0.3e-°-276'
s — — + hh + + jvoih + to) a. Show that for each fixed t, fit) can be thought of as the
average value of a function of the form
32. What is the average velocity for the object described in Prob¬
lem 31 during the same time period?
g(A) = (At2 + Bt + C)e~k'
33. Records indicate that t hours past midnight, the temperature
at the local airport was
over the interval a < X < b. Find A, B, and C.
fit) = —O.lr +1 + 50 b. Exploration Problem So what? In particular, can you see any
degrees Fahrenheit. What was the average temperature at the value in the interpretation of the Heinz concentration func¬
airport between 9:00 A.M. and noon? tion as an average value? Explain.
334 Chapter 5 Integration
APPROXIMATION BY RECTANGLES
If /(*) > 0 on the interval [a, b], the definite integral /j’ fix) dx is equal to the area
under the graph of / on [a, b]. As we saw in Section 5.2, one way to approximate this
area is to use n rectangles, as shown in Figure 5.34. In particular, divide the interval
[a, b] into n subintervals, each of width Ax =--, and let x£ denote the right end¬
Figure 5.34 Approximation by point of the kth subinterval. The base of the kth rectangle is the kth subinterval, and
rectangles its height is f(x£). Hence, the area of the kth rectangle is f(x£) Ax. The sum of the
areas of all n rectangles is an approximation for the area under the curve and hence an
approximation for the corresponding definite integral. Thus,
TRAPEZOIDAL RULE
Figure 5.35 Approximation by
trapezoids The accuracy of the approximation generally improves significantly if trapezoids are
used instead of rectangles. Figure 5.35 shows the area from Figure 5.34 approximated
by n trapezoids instead of rectangles. Even from these rough illustrations you can see
how much better the approximation is in this case.
Suppose the interval [a, b] is partitioned into n equal parts by the subdivision
points xo, xi, ..., x„, where xq = a and xn — b. The kth trapezoid is shown in greater
detail in Figure 5.36. Recall the formula for the area of a trapezoid:
A = |(Zq + b2)h
Figure 5.36 The kth trapezoid has The sum Tn estimates the total area under the curve y = f{x) on the interval [a, b]
area ^ [ / (x* _,) + / (x*) ] Ax. and hence also estimates the integral
b
f (x) dx
This approximation formula is known as the trapezoidal rule and applies as a means
of approximating the integral, even if the function / is not positive.
5.8 Numerical Integration: The Trapezoidal Rule and Simpson's Rule 335
Trapezoidal Rule
Let / be continuous on [a, b). The trapezoidal rule is
b
f (x)dx |[/(xq) + 2/(xi) + 2/(X2) + • • • + 2/(x„_i) + /(x„)]Ax
where Ax = ——— and, for the Ath subinterval, = a + A Ax. Moreover, the
Our first example uses the trapezoidal rule to estimate the value of an integral that
we can compute exactly by using the fundamental theorem.
Solution
The interval is [a, b] = [—1,2], so a = —1 and b = 2. Then
. 2 — (—1) 3
Ax =-= - = 0.75. Thus,
4 4
X4 = a+ 4-Ax = b — 2 f(M) = 22 = 4
T, = 1[1 +0.125 + 0.5+ 3.125 +4](0.75) = 3.28125
Trapezoid M 3 • 2A125
The negative sign indicates that the trapezoidal formula overestimated the true value
Trapezoid 10 3.QMS
of the integral in Example 1.
Trapezoid 100 3-ODOMS
Software for numerical evaluation of integrals is common (see the Technology
Trapezoid 1DDD 3-000005
Manual). The output for Example 1 is shown in the margin.
SIMPSON’S RULE
Roughly speaking, the accuracy of a procedure for estimating the area under a curve de¬
pends on how well the upper boundary of each approximating area fits the shape of the
given curve. Trapezoidal strips often result in a better approximation than rectangular
strips, and it is reasonable to expect even greater accuracy to occur if the approximating
regions have curved upper boundaries, as shown in Figure 5.37.
336 Chapter 5 Integration
The name given to the procedure in which the approximating strip has a parabolic arc
for its upper boundary is called Simpson’s rule.*
As with the trapezoidal rule, we will derive Simpson’s rule as a means for approxi¬
mating the area under the curve y — f(x) on the interval [a, b], where / is continuous
and satisfies f(x) > 0. First, we partition the given interval into a number of equal
subintervals, but this time, we require the number of subdivisions to be an even number
(because this requirement will simplify the formula associated with the final result).
If xo, xi.x„ are the subdivision points in our partition (with xq = a and
xn = b), we pass a parabolic arc through the points, three at a time (the points with
x-coordinates *o, *i, x2, then those with x2, *3, jc4, and so on). It can be shown
(see Problem 42) that the region under the parabolic curve y — f(x) on the interval
[x2k-2, x2k] has area
By adding the area of the approximating parabolic strips and combining terms, we
obtain the sum Sn of n parabolic regions:
*This rule is named for Thomas Simpson (1710-1761), an English mathematician who, curiously, neither
discovered nor made any special use of the formula that bears his name.
5.8 Numerical Integration: The Trapezoidal Rule and Simpson's Rule 337
Simpson's Rule
Let / be continuous on [a, b]. Simpson’s rule is
b
+ 4/(x„_i) + / (x„)] Ax
Solution
2 - 1
We have Ax = — =0.1, and xo = a = 1, xi = 1.1, x-i = 1.2, ...» X9 = 1.9,
xio = b = 2.
dx & S10
Jf-
1 x
1/1 4 2 4 2 4 , 2 , 4 , 2 , 4 , l\,ft n
— — [ — -4- - -4- - -4- - -4“ - -4- - ~b - ~b - “I” - “b - d- — I (H* 1)
3 \1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2)
Comparison of Simpson's and trapezoidal rules
~ 0.6931502
Type of * of sub- Estimate over
estimate intervals Cli 2]
This estimate compares well with the value found directly by applying the fundamental
Trapezoid
Simpson
ID
10
.^3771403175
.^3150230^
theorem of calculus:
Trapezoid 100 .1^3153430462
Simpson 100 • 1^3147160672 r dx
Trapezoid 1000 .b131472430b0 In 2 — In 1 % 0.6931472
Simpson 1000 • be131471005b0
J1 X
Comparison of Simpson’s and An example of the output of a computer program for this type of problem is shown in
trapezoidal rules the margin.
ERROR ESTIMATION
The difference between the value of an integral and its estimated value is called its
error. Since this error is a function of n we denote it by En.
Trapezoidal error: |En | < ———M, where M is the maximum value of |/"(x)| on [a, b].
12 n2
Moreover, if / has a continuous fourth derivative on [a, b], then the error En in even)
Simpson’s error: \E„\ < ——^—K, where K is the maximum value of |/l4)(x)| on [a, b].
1 ” 180n4
Proof The proofs of these error estimates are beyond the scope of this book and
can be found in many advanced calculus textbooks and in most numerical analysis
books.
338 Chapter 5 Integration
Solution
If /(x) = —, we find that /(4)(x) = 24x-5. The maximum value of this function
x
will occur at a critical number (there is none on [ 1,2]) or at an endpoint. Thus, the
largest value of |/(4)(x)| on [1,2] is |/(4>(1)| — 24. Now, apply the error formula
with K — 24, a = 1, b = 2, and n = 10 to obtain
^ k(b — a)5 24(2 - l)5
E io 0.0000133
~ 180/i4 180(10)4
That is, the error in the approximation in Example 2 is guaranteed to be no greater than
0.0000133. " " ■
With the aid of the error estimates, we can decide in advance how many subinter¬
vals to use to achieve a desired degree of accuracy.
Solution
To be correct to four decimal places means that \E„\ < 0.00005. Because /(x) = x_1,
we have /"(x) = 2x~3. On [1, 2] the largest value of |/"(x)| is |/"(1)| = 2, so
M — 2, a = \, b — 2, and
,^ , 2(2- l)3 1
\En\ < - - -
12n2 6 n2
The goal is to find the smallest positive integer n for which
1
— < 0.00005
6 n1
10,000 < 3n2 Multiply by the positive number 60,000/r.
1 = [ f(x)dx
5.8 Numerical Integration: The Trapezoidal Rule and Simpson's Rule 339
we have focused attention on the “error term,” but this only measures the error that
comes from estimating I by the trapezoidal or Simpson approximation sum. There are
other kinds of errors that must be considered in this or any other method of approxima¬
tion. In particular, each time we cut off digits from a decimal, we incur what is known
as a round-off error. For example, a round-off error occurs when we use 0.66666667 in
place of | or 3.1415927 for the number jt . Round-off errors occur even in large com¬
puters and can accumulate to cause real problems. Specialized methods for dealing
with these and other errors are studied in numerical analysis.
The examples of numerical integration examined in this section are intended as
illustrations and thus involve relatively simple computations, whereas in practice, such
computations often involve hundreds of terms and can be quite tedious. Fortunately,
these computations are extremely well suited to automatic computing, and the reader
who is interested in pursuing computer methods in numerical integration will find an
introduction to this topic in the Technology Manual.
17. / tan 1 x dx with error less than 0.01. Use Simpson’s rule.
Approximate the integrals given in Problems 3-8 with the specified
Jo
number of subinterx’als using
a. the trapezoidal rule
dx
b. Simpson’s rule 18.
f/ e x sin x dx to three decimal places. Use Simpson’s rule.
Jo
with n = 4 In Problems 19-23, determine how many subintervals are required
3*
/
L
o
TT^2
_
+ x2 dx with n = 4
to guarantee accuracy to within 0.00005 using
a. the trapezoidal rule b. Simpson’s rule
2 f4 dx
21. 22. / cosx dx
6. I x cos x dx with n = 6 J1 yfx Jo
’■L
7. / xe x dx with n = 6 23. J In y/xdx 24. f e-lx dx
L Jo
prescribed accuracy.
a. Estimate n correct to one decimal place by applying the
dx trapezoidal rule to this integral.
11. [ — with error less than 0.05. Use the trapezoidal rule.
Jo x2 +\ b. Estimate n correct to one decimal place by applying Simp¬
son’s rule to this integral.
+ x2 dx with error less than 0.05. Use Simpson’s rule.
,27> 26. Find the smallest value of n for which the trapezoidal rule es¬
timates the value of the integral
13. / cos 2x dx accurate to three decimal places. Use Simpson’s
Jo
rule.
w7
x 1 dx
f
x 1 dx accurate to three decimal places. Use Simpson’s
Minutes 25 30 35 40 45 50 55 60
/ = / sin x dx = — cos x = 2
Jo
Speedometer 60 49 53 47 39 42 48 53
a. Use the trapezoidal rule to estimate / for n — 10, 20, 40,
Use the trapezoidal rule to estimate the total distance traveled and 80. Compute the error En in each case and compute
by the couple from noon to 1:00 P.M. En ■ nk for k = 1,2, 3, 4. Based on your results, you
29. An industrial plant spills pollutant into a lake. The pollutant should be able to conclude that the order of convergence of
spread out to form the pattern shown in Figure 5.40. All dis¬ the trapezoidal approximation is n2.
tances are in feet. b. Repeat part a using Simpson’s rule. Based on your results,
what is the order of convergence for Simpson’s rule?
c. Repeat part a using a rectangular approximation with right
endpoints. What is the order of convergence for this
method?
method introduced in this section. This method, known as the between the first and fourth; then ... the area between the first
yenri (circle principle), found the area of a circle by dividing and fourth ordinates is approximated by |(A + 3B)R. This is
the circle into small rectangles, as shown in Figure 5.41. known today as the "Newton-Cotes three-eighths rule," which
can be expressed in the form
/ tan 1 x dx
Jo
b. If v = p(x) is the quadratic function found in part a, show c. Let <2i(*i> /(*i)), Qi(x2, fix 2)), Q3ix3, fix 3)) be points
that with X2 = x\ + h, and X3 = Xi + 2h. Explain why
fx3
p(x) dx = ~[p(-h) + 4p(0) + p{h)] p(x)dx = ~[p(Xi) + 4p(x2) + p(x3)]
Jxi
You may have noticed in Section 2.4 that we did not prove the properties of expo¬
“When I use a word,” Humpty Dumpty nential functions for all real number exponents. To treat exponentials and logarithms
said, “it means just what I choose it to rigorously, we use the alternative approach provided in this section. Specifically, we
mean—nothing more or less.” use a definite integral to introduce the natural logarithmic function and then use this
-Lewis Carroll (1832-1898) function to define the natural exponential function.
x > 0
At first glance, it appears there is nothing “natural” about this definition, but if this in¬
tegral function has the properties of a logarithm, why should we not call it a logarithm?
We begin with a theorem that shows that lnx does indeed have the properties we would
expect of a logarithm.
J px dt
—
1 t
Let x > 0 and y > 0 be positive numbers. Then
a. Let x = 1. Then
lf1 dt
7 =0'
b. For fixed positive numbers x and y, we use the additive property of integrals as
follows:
Let u = —, so t = ux;
x
dt = x du.
y xdu If t = x, then u = 1;
= lnx +
l ux if t = xy, then u = y.
n
'y di
du
= lnx +
J1 U
= In x + In y
c. and d. The proofs are outlined in the problem set. □
5.9 An Alternative Approach: The Logarithm as an Integral 343
GEOMETRIC INTERPRETATION
An advantage of defining the logarithm by the integral formula is that calculus can be
used to study the properties of ln.v from the beginning. For example, note that if x > 1,
the integral
fx dt
In* = / —
J i t
tion 5.4 that according to the second fundamental theorem of calculus, if / is continu¬
ous on [a, b], then
F(x) = [ f (t)dt
Ja
dF
is a differentiable function of x with derivative —— = /(x) on any interval [a,x].
dx
1 . dt . fx
Therefore, because - is continuous for all t > 0, it follows that lnx = / — is
t Ji t
differentiable for all x > 0 with derivative —-(lnx) = -. By applying the chain rule,
dx x
we also find that
d 1 du
— (Inn) = - —
dx u dx
for any differentiable function u of x with u > 0.
To analyze the graph of /(x) = lnx, we use the curve-sketching methods of
Chapter 4:
a. lnx is continuous for all x > 0 (because it is differentiable), so its graph is “unbro¬
ken.”
b. The graph of In x is always rising, because the derivative
d
— (lnx) = —
„ 1
dx x
is positive for x > 0. (Recall that the natural logarithm is defined only for x > 0.)
c. The graph of lnx is concave down, because the second derivative
d (\ 1
(lnx) = — ( -
dx2 dx \x
is negative for all x > 0.
d. Note that In 2 > 0 (because f~ y > 0) and because In 2p = p In 2, it follows
that lim In2P = +oo. But the graph of /(x) = lnx is always rising, and thus
p->-+oo
lim lnx = Too. Similarly, it can be shown that lim lnx = -oo.
x—>+oo x—>0+
e. If b is any positive number, there is exactly one number a such that In a = b (be¬
cause the graph of lnx is always rising for x > 0). In particular, we define x = e
as the unique number that satisfies lnx = 1.
Figure 5.43 The graph of the
natural logarithm function, In.v These features are shown in Figure 5.43.
344 Chapter 5 Integration
Originally, we introduced the natural exponential function ex and then defined the natu¬
ral logarithm In* as the inverse of ex. In this alternative approach, we note that because
the natural logarithm is an increasing function, it must be one-to-one. Therefore, it has
an inverse function, which we denote by E(x).
Because In* and E(x) are inverses, we have
Proof
=x +y Inversion formula
= In E(x + y) Inversion formula
E(x)E(y) = E(x + y)
c. and d. The proofs are similar and are left as exercises. □
Proof It can be shown that the inverse of any differentiable function is also differ¬
entiable so long as the derivative is never zero (see Appendix B). Since In* is dif¬
ferentiable, it follows that its inverse £(*) is also differentiable, and using implicit
differentiation, we find that
In y — ln[£(x)] y = E(x)
In y = x Inversion formula
1 dy
1 Differentiate implicitly.
y dx
dy_ _
y
dx
dE dy dE
E(x) Because — = —— and E(x) = y □
dx dx dx
ln(er) — r\ne — r
so that
E(r) = er
This means that E(x) = ex when x is a rational number, and we define E{x) to be
ex for irrational x as well. In particular, we now have an alternative definition for e:
£(1) = e1 = e.
In M — In N
to show that In* —> —oo as * —> 0+. Hint: What happens to
111(2“^) as N grows large without bound? if and only if M = N. Hint: Show that if M 7^ A, Rolle’s
theorem implies that
2. Use Simpson’s rule with n = 8 subintervals to estimate
— (In*) = 0
dx
for some number c between M and N. Why is this impossi¬
Compare your estimate with the value of In 3 obtained from ble?
your calculator. 7. The product rule for logarithms states
3. Use the error estimate for Simpson’s rule to determine the ac¬
curacy of the estimate in Problem 2. How many subintervals logb(MN) = \ogb M + logh N
should be used in Simpson’s rule to estimate In 3 with an error
For this reason, we want the natural logarithm function In* to
not greater than 0.00005?
satisfy the functional equation
4. Prove the quotient rule for logarithms.
f(xy) = f(x) + f(y)
In = In* — In y
Suppose /(*) is a function that satisfies this equation through¬
out its domain D.
for all * > 0, y > 0. Hint: Use the product rule for loga¬ a. Show that if /(1) is defined, then /(l) = 0.
rithms.
b. Show that if /(—1) is defined, then /(—1) = 0.
5. Show that ln(*p) = pin* for* > 0 and all rational exponents
c. Show that /(-*) = /(*) for all * in D.
p by completing the following steps.
d. If /'(*) is defined for each * 7 ^ 0, show that
a. Let F(x) = ln(*p) and G{x) = p In*. Show that
F'(x) — G'(*) for* > 0. Conclude that
F(x) = G(x) + C. *
346 Chapter 5 Review
Then show that 8. Show that for each number A, there is only one number x for
which lnx = A. Hint. If not, then lnx = lny = A for x y.
fix) for x > 0
Why is this impossible?
e. Conclude that any solution of 9. a. Use an area argument to show that In 2 < 1.
f{xy) = fix) + fiy) b. Show that In 3 > 1, and then explain why 2 < e < 3.
that is not identically 0 and has a derivative for all x 0
6. What is an area function? What do we mean by area as an 23. Find F\x) if Fix) = J r5v/cos(2r + 1 )dt.
antiderivative? What conditions are necessary for an integral
to represent an area? Evaluate the integrals in Problems 24-29.
7. What is the formula for area as the limit of a sum?
8. What is a Riemann sum?
24.
tpi?
' ix
l dx
14. Describe in your own words the process of integration by sub¬ 32. Find the average value of y = cos 2x on the interval [0, -].
stitution. 33. Modeling Problem In 1995, a team of archaeologists led by
15. What is a differential equation? Michel Brunet of the University of Poitiers announced the
identification in Chad of Australopithecus specimens believed
16. What is a separable differential equation? How do you solve
to be about 3.5 Myr (million years) old.f Using the expo¬
such an equation?
nential decay equation, explain why the archaeologists were
17. What is the growth/decay equation? Describe carbon dating. reluctant to use carbon-14 to date their find.
18. What is an orthogonal trajectory?
19. State the mean value theorem for integrals. *There is an interesting discussion of this integral in the May 1999 issue
of The American Mathematical Monthly in an article entitled, "What Is a
20. What is the formula for the average value of a continuous
Closed-Form Number?” The function / ex~ is an example of a function that
function? is not an elementary function (that we mentioned in Section 1.3).
21. State the following approximation rules: 4*
“Early Hominid Fossils from Africa,” by Meave Leakey and Alan Walker.
a. rectangular approximation Scientific American, June 1997, p. 79.
Chapter 5 Review
34. A slope field for the differential equation 14. fx + ldx 15.
f 3-a ^
I 2a 1 Vl-A2
/ =x+y
16. j (e~x + 1 )ex dx 17. /
1
f sin a — cos a ,
-dx
sinA + cosA
is given in Figure 5.45.
18. I V^(A2 + y/x + 1) dx 19. 1 (a — l)2 dx
24.
f
1 y/x(y/x + l)2
dx
25. j x\/1 — 5x2 dx
26. j Vsin x — cosx(sinx + cosx) dx
1 l
l
t
i 1
i 1
) 1
\ 1 n
i l rr
H
1 i ) i t
i l
till!
++L++
1 / r / /
t i
l i
i i
1 t
f » i
1 / / / i
t 1 1 / /
/ / t
t / / / / / t f / / / / ( / / ' t / L
/ / / / 1 //
J J
' / /
{ \ s \ nmu i \ 1 \ l
\ 1 ii uh'ii » t \
(3a2 + 2x — 1) dx. 1 1 t t ; \ \ i ih \
X \ )i \ t| 1 \
t \ t
1 t t \ \
11
/
3. Use the definition of the definite integral to find 34. y(0) — 2
11 i T :: 71
(4a3 + 6a2 + 3) dx. / /
! 1 i
Jo » v *
Find the definite and indefinite integrals in Problems 4-31. If you j 1:1! 1; 51
i; //i
do not have a technique for finding a closed (exact) answer, ap¬ 1-
1 11 !
proximate the integral using numerical integration. //
sA u~; C-i
-2
4. f
Jo
(5a4- 8a3 4-1 )dx 5. j 30(5* - 2Y dx
t t|
\ (\ \ \\
>'
!i:
Jof
"2 x2dx 1 \ :
6. (xs/x+2)2dx ; 11
11
VT’TT
0 dx 36. Find F'(a), where F(x) r cos41 dt
. / (a + sin a)3 dx 9.
8
Ji 1 y/\ -2x 37. Find the area under /(a) = a-1 on [ 1,4].
0 dx 38. Find the area under /(a) = 2 4- x — x2 on [— 1, 1 ].
10 .
Jif A- s/3x - 1
11 .
1 39. Find the area under /(a) = e4x on [0, 2].
40. Find the area bounded by the curve y = a Va2 4- 5, the .v-axis,
. f
5 a-2 — 2x + 1
13. dx
12
J A3 \dx and the vertical lines a = — 1 and a = 2.
348 Chapter 5 Review
41. Find f (t) if /"(f) = sin4f — cos 21 and / (|) = /' (|) = 1. A prime number, p, is a counting number that has ex¬
42. Find /(x) if f"\x) = 2x3 + x2, given that f"(l) = 2, actly two divisors. The prime number theorem, conjec¬
/'(l) = 1, and /(l) = 0. tured by Gauss in 1793, says that the number of primes tt(x)
less than a real number x is a function that behaves like x/ lnx
Solve the differential equations in Problems 43-50.
as x —>• oo; that is,
„ dy , dy cos4x
43. —- = (1 — y)2 7T(X)
dx lim = 1
c-^+oo x/ lnx
dy_ =
45. ^
dx \ sin x /
47. ~ = y(x2 + 1)
dx
49 dy_ _ cos2y
1
c/x cotx
sinx
51. Find the average value of /(x) = on the interval 1
cos2x
[0- ]• f
52. Find the average value of f(x) = sinx
a. on [0,7r] b. on [0, 2n] __
0 10 20 30 40 50 60 70 80 90 100
53. Use the trapezoidal rule with n = 6 to approximate a. n(n) for 0 s n s 100
/ sin x dx. Compare your result with the exact value of this
Jo
integral.
55. Estimate
f V71 + X-
using the trapezoidal rule with n = 8.
58. Use the trapezoidal rule to estimate to within 0.00005 the b. jt(n) for 0 s n <. 1,000
value of the integral
Figure 5.46 Graphs of n(n)
60. HISTORICAL There are 68. A manufacturer estimates that the marginal revenue of a cer¬
many ways to estimate the prime distri¬ tain commodity is
bution function n(x) defined in Problem
59 other than by the integral function R'(x) = ^(x3/2 + ir1/2
G(x) used by Gauss. Adrien-Marie
dollars per unit when x units are produced. Assuming no rev¬
Legendre is best known for his work
enue is obtained when x — 0, how much revenue is obtained
with elliptic integrals and mathematical
from producing x = 4 units?
physics. In 1794, he proved that tt was
an irrational number and formulated a ADRIEN-MARIE 69. Find a function whose tangent has slope x\/x2 + 5 for each
conjecture that is equivalent to the prime LEGENDRE value of x and whose graph passes through the point (2, 10).
number theorem. He made his estimate 1752-1833 70. A particle moves along the x-axis in such a way that after t
using the function seconds its acceleration is a(t) = 12(2/ + 1)_3/2. If it starts at
X
rest at x = 3, where will it be 4 seconds later?
n(x) % L(x) - 71. An environmental study of a certain community suggests that
hjx^T08366
/ years from now, the level of carbon monoxide in the air will
be changing at the rate of 0.1/ + 0.2 parts per million per year.
For this Quest, use L(x) to estimate n(x) for x — 100, 1,000,
If the current level of carbon monoxide in the air is 3.4 parts
and 10,000, and then compare with the results found using
per million, what will the level be 3 years from now?
G(x) in Problem 59.
72. A woman, driving on a straight, level road at the constant
61. Approximate the average value of the function defined by
speed v0, is forced to apply her brakes to avoid hitting a cow.
cos* The car comes to a stop 3 seconds later, j0 ft from the point
/(*) =
x2 where the brakes were applied. Continuing on her way, she
1- increases her speed by 20 ft/s to make up time but is again
2
forced to hit the brakes, and this time it takes her 5 seconds
on [0, 1 ]. Hint: Use the trapezoidal rule with n = 6.
and .s'! feet to come to a full stop. Assuming that her brakes
62. The brakes of a certain automobile produce a constant decel¬
supplied a constant deceleration k ft/s2 each time they were
eration of k m/s2. The car is traveling at 25 m/s when the
used, find k and determine v0, 5o, and si.
driver is forced to hit the brakes. If it comes to rest at a point
73. A study indicates that x months from now the population of a
50 m from the point where the brakes are applied, what is kl
certain town will be increasing at the rate of 10 -f 2-Jx peo¬
63. A particle moves along the /-axis in such a way that
ple per month. By how much will the population of the town
a{t) — —4s(t), where s(t) and a(t) are its position and ac¬
increase over the next 9 months?
celeration, respectively, at time t. The particle starts from rest
74. It is estimated that / days from now a farmer’s crop will be
at s = 5.
increasing at the rate of 0.3/2 4- 0.6/ + 1 bushels per day. By
a. Show that v2 + 4s2 = 100, where v(t) is the velocity of
how much will the value of the crop increase during the next
the particle. Hint: First use the chain rule to show that
6 days if the market price remains fixed at $2 per bushel?
a{t) — v{dv/ds).
75. Records indicate that / months after the beginning of the year,
b. What is the velocity when the particle first reaches 5 = 3?
the price of turkey in local supermarkets was
Note: At the time in question, the sign of v is determined
by the direction the particle is moving. P(t) = 0.06/2 - 0.2/ + 1.2
64. An object experiences linear acceleration given by
dollars per pound. What was the average price of turkey dur¬
a{t) = 2t + 1 ft/s2 ing the first 6 months of the year?
Find the velocity and position of the object, given that it starts 76. Modeling Problem V. A. Tucker and K. Schmidt-Koenig have
its motion (at t = 0) at 5 = 4 with initial velocity 2 ft/s. investigated the relationship between the velocity v (km/h) of
65. When it is x years old, a certain industrial machine generates a bird in flight and the energy E(v) expended by the bird.*
revenue at the rate of R'(x) — 1,575 — 5x2 thousand dollars Their study showed that for a certain kind of parakeet, the rate
per year. Find a function that measures the amount of revenue, of change of the energy expended with respect to velocity is
and find the revenue for the first five years. modeled (for v > 0) by
77. Modeling Problem A toxin is introduced into a bacterial cul¬ 83. Scientists are observing a species of insect in a certain swamp
ture, and t hours later, the population P(t) of the culture, in region. The insect population is estimated to be 10 million and
millions, is found to be changing at the rate is expected to grow at the rate of 2% per year. Assuming that
the growth is exponential and stays that way for a period of
— = —(In 2)25-' years, what will the insect population be in 10 yr? How long
dt will it take to double?
If there were 1 million bacteria in the culture when the toxin
84. Solve the system of differential equations
was introduced, when will the culture die out?
78. Modeling Problem The Snowplow Problem of R. P. Agnew* ■
dx d\
(This problem was mentioned in the preview for this chapter.) 2—+5 -f-=t
dt dt
One day it starts snowing at a steady rate sometime before
noon. At noon, a snowplow starts to clear a straight, level sec¬ dx dy
-b 3— =7 cos t
tion of road. If the plow clears 1 mile of road during the first dt dt
hour but requires 2 hr to clear the second mile, at what time
, did it start snowing? Answer this question by completing the dx dy
Hint: Solve for — and — algebraically, and then integrate.
following steps: dt dt
a. Let t be the time (in hours) from noon. Let h be the depth 85. Spy Problem While lunching on cassoulet de chameau at his
of the snow at time t, and let 5 be the distance moved by favorite restaurant in San Dimas (recall Problem 50, Section
the plow. If the plow has width w and clears snow at the 5.1), the Spy finds a message from his esteemed superior, Lord
ds
constant rate p, explain why wh— — p. Newton Fleming, spelled out in his alphabet soup, “The aver¬
dt age of the temperature F{t) — at3 + bt1 + cr + d over the
b. Suppose it started snowing f0 hours before noon. Let r de¬
time period from 9a.m. to 3 p.M. is the average temperature
note the (constant) rate of snowfall. Explain why
at two fixed times f, and t2. Mother waits at the well.”
h{t) = r{t + t0) The Spy knows that “Mother” is the codename of Lord
Newton himself, and “the well” is a sleazy bar at the edge
By combining this equation with the differential equation
of town. He decides that time t is measured from noon be¬
in part a, note that
cause the soup was served at that time. The cryptic nature of
ds the message suggests that t\ and t2 are independent of a, b,
wr(t + to)~ = P
dt c, and d and that one of them is the rendezvous time. When
Solve this differential equation (with appropriate condi¬ should he arrive for the meeting? (Remember, it’s not wise to
tions) to obtain to and answer the question posed in the go to the well too often.)
problem. 86. Putnam Examination Problem Where on the parabola 4ay = x2
79. a. If / is continuous on [a, b], show that (a > 0) should a chord be drawn so that it will be normal to
the parabola and cut off a parabolic sector of minimum area?
\[b f(x) dx <
I Ja
f
Ja
1/001 dx That is, find P so that the shaded area in Figure 5.47 is as
small as possible.
b. Show that
or stnx
dx
Kinematics of Jogging
Ralph Boas (1912-1992), who was a professor of mathematics at Northwestern
University, wrote this guest essay. Professor Boas was well known for his many
papers and professional activities. In addition to his work in real and complex
analysis, he wrote many expository articles, such as this guest essay, about
teaching or using mathematics.
/(* +1)-/(*)# 8
Because
/(*+l)-/(*)-8
is continuous and never 0, it must either always be positive, or else always negative;
let us suppose the former. Write the corresponding facts for x =0, 1,... ,n:
/(n)-/(n — 1) > 8
But we started with the assumption that f(n) = 8/? and /(0) = 0, so assuming that
f(x + 1) — f (x) is never 8 leads to a contradiction.
352 Chapter 5 Mathematical Essays
It is somewhat harder to show that only integral values of n will work. Suppose
you jog so that your time to cover x miles is
2 71X
J(x) = k sin"-b 8x
n
where n is not an integer and A: is a small number. This is a legitimate assumption,
because J is an increasing function (as a time has to be), if k is small enough. To be
sure of this, we calculate J'(x)—and here we actually have to use some calculus (or
have a calculator that will do it for us). We find
kn . 2.71 x
J\x) — sin-b 8
n n
If k is small enough (k < f), then J\x) > 0. This shows not only that J increases
but also that
J(x + 1) — /(x)
cannot be eight. Because J(x + 1) - J{x) is never negative, if you jog so that your
time is 7(x), you will never cover a whole mile in less than 8 min.
Use the Internet, a library, or references other than this textbook J(x) = sin2 — + 8x
to research the information necessary to answer questions in Prob¬
lems 1-11 Graph this function on [0, 8].
1. HISTORICAL QUEST The derivative is one of the great 8. Use calculus to find how small k (from the guest essay) needs
ideas of calculus. Write an essay of at least 500 words about to be in the expression
some application of the derivative that is not discussed in this 7 trx
text. J(x) = k sin"-b 8*
n
2. HISTORICAL QUEST The concept of the integral is one so that Jfx) > 0.
of the great ideas of calculus. Write an essay of at least 500 9. a. Find a number x, x ^ 0, that, when divided by 2, gives a
words about the relationship of integration and differentiation display of 0 on your calculator. Write a paper describing
as it relates to the history of calculus. your work as well as the processes on your calculator,
3. HISTQRICAL QUEST Write a report on Georg Riemann. b. Calculate \fl using a calculator. Next, repeatedly subtract
(See Historical Quest in Section 5.3.) Include the 1984 devel¬ the integer part of the displayed number and multiply the
opment in the solution of the Riemann hypothesis. result by 10. Describe the outcome, and then devise a
method for finding a/2 using calculus. Write a paper com¬
4. HISTORICAL QUEST As we saw in a Historical Quest in
paring these answers.
Section 5.8, the mathematician Seki Kowa was doing a form
of integration at about the same time that Newton and Leibniz 10. HfflTHTfll Eli Maor, a native of Israel, has a long-standing
were inventing the calculus. Write a paper on the history of interest in the relations between mathematics and the arts.
calculus from the Eastern viewpoint. Read the fascinating book To Infinity and Beyond, A Cultural
History of the Infinite (Boston: Birkhauser, 1987), and prepare
5. “Lucy,” the famous prehuman whose skeleton was discovered a book report.
in East Africa, has been found to be approximately 3.8 million
11. In Section 3.1, we told the story of Fermat’s last
years old. About what percentage of l4C would you expect
theorem. The book Fermat’s Enigma by Simon Singh (New
to find if you tried to “date” Lucy by the usual carbon dat¬
York: Walker & Company, 1997) tells the story of how An¬
ing procedure? The answer you get to this question illustrates
drew Wiles solved the greatest mathematical puzzle of our
why it is reasonable to use carbon dating only on more re¬
age: He locked himself in a room and emerged seven years
cent artifacts, usually less than 50,000 years old (roughly the
later. Read this book and prepare a book report.
time since the last major ice age). Read an article on alterna¬
12. Make up a word problem involving an application of the inte¬
tive dating procedures such as potassium-argon and rubidium-
gral. Send your problem and solution to
strontium dating. Write a paper comparing and contrasting
such methods. Strauss, Bradley, and Smith
Prentice Hall Publishing Company
6. In the guest essay it was assumed that the time to cover x miles
1 Lake Street
is Upper Saddle River, NJ 07458
. -> 7XX
J(x) = k sin"-K 8*
n The best one submitted will appear in the next edition (along
Suppose that n = 5. What choices for k seem reasonable? with credit to the problem poser).
Chapters 1-5 Cumulative Review 353
x sin x sin 3x
9. Iim 10. Iim
x~>° x + sin‘x Jt—>0 x
2/x
tan x
11. Iim (1 + x) 12. Iim
X —► -(-oo *->■0
a. Find F(l).
Find the derivatives in Problems 14-22.
b. Find all values on the interval (—3, 12) at which F has a
14. y = 6x3 — 4x + 2 15. relative maximum.
x2 -4 c. Write an equation for the line tangent to the graph of F at
16. 17.
“ 3x + 1 x + cosx x = 7.
18. x2 + 3xy + y2 = 0 19. d. Find the x-coordinate of each point of inflection of the
20. v = e5*“4 21. graph of F on the interval (—3, 12).
38. An electric charge <2o is distributed uniformly over a ring of
22. v = tan 1(x2 — 3)
radius R. The electric field intensity at any point x along the
Find the integrals in Problems 23-28. axis of the ring is given by
*9
23.
f
/4
dO 24
• L 50(2x - 5)3 dx
E(x) =
(x2
QoX
+ R2yp
25.
i:0
V9 + x
xdx
dx
26
•/
esc 36 cot 30 dO
x3 + 2x — 5
At what point on the axis is £(x) maximized?
39. A rocket is launched vertically from a point on the ground
5,000 ft from a TV camera. If the rocket is rising vertically at
27. 28 dx
+2 •/ the rate of 850 ft/s at the instant the rocket is 4,000 ft above
29. Approximate
f dx
V1 + -
using Simpson’s rule with n = 6.
the ground, how fast must the TV camera change the angle of
elevation to keep the rocket in the picture?
40. A particle moves along the x-axis so that its velocity at any
30. Sketch the graph of y = x3 — 5x2 + 2x + 8.
time t > 0 is given by v(t) — 612 — 2t — 4. It is known that
4 -(- x2 the particle is at position x = 6 for / = 2.
31. Sketch the graph of v =-
4 — x2 a. Find the position of the particle at any time t > 0.
32. Find the largest value of /(x) = ^x3 — 2x2 + 3x — 10 on b. For what value(s) of t, 0 < t < 3, is the particle’s instan¬
[0, 6], taneous velocity the same as its average velocity over the
Solve the differential equations in Problems 33-34. interval [0, 3].
c. Find the total distance traveled by the particle from time
dy = x2y2
33. ~ 34. (1 4- x2) dy — (x + l)y dx t — 0 to t = 3.
dx
Find the particular solution of the differential equations in Prob¬
41. Let / be the function defined by /(x) = 2cosx. Let P(0, 2)
and Q (|, 0) be the points where / crosses the y-axis and
lems 35-36.
x-axis, respectively.
35. — = 2(5 — y); y = 3 when x = 0 a. Write an equation for the secant line passing through points
dx
P and Q.
dy
36. — = ey sinx; y = 5 when x = 0 b. Write an equation for the tangent line of / at point Q.
dx
354 Chapters 1-5 Cumulative Review
c. Find the x-coordinate of the point on the graph of /, be¬ 44. An object is thrown directly downward with velocity 5 ft/s
tween point P and Q, at which the line tangent to the graph from an airplane, and t seconds later is falling with accelera¬
of / is parallel to PQ. Cite a theorem or result that assures tion
the existence of such a tangent. a(t) = 32-0.08u
42. Let / be the function defined by f(x) — y/x — 2.
a. Sketch / and shade the region R enclosed by the graph of where v(t) is the object’s velocity at time t.
/, the x-axis, and the vertical line x = 5. a. Find an expression for u in terms of t.
b. Find the area of R. b. If the terminal velocity is defined as lim v(t), find the
r-*+oo
43. Let / be the function defined by /(x) = x3 — 6x2 + k for k
terminal velocity of the object (to the nearest ft/s).
an arbitrary constant.
a. Find the relative maximum and minimum of / in terms of c. The object can land safely if the impact velocity is no
k. greater than 60 ft/s. What is the maximum altitude the
b. For which values of k does / have three distinct roots? airplane can fly to guarantee a safe landing for the object?
355
356 Chapter 6 Additional Applications of the Integral
In some practical problems, you may have to compute the area between two curves.
Suppose / and g are functions such that /(x) > g(x) on the interval [a, b], as shown
in Figure 6.1. We do not insist that both / and g be nonnegative functions, but we
begin by showing that case in Figure 6.1.
To find the area of the region R between the curves defined by y = /(x) and
y = g(x) from x = a to x — b, we subtract the area between the lower curve y — g(x)
and the x-axis from the area between the upper curve y — /(x) and the x-axis; that is,
rb />b nb
AREA OF R — / /(x) dx — / g(x) dx = / [/(x) - g(x)] dx
Ja J ci Ja
This formula seems obvious in the situation where both /(x) > 0 and g(x) > 0, as
shown in Figure 6.1. However, the following derivation requires only that / and g be
continuous and satisfy /(x) > g(x) on the interval [a, b]. We wish to find the area
between the curves y = /(x) and y — g(x) on this interval. Choose a partition
of the interval [a, b] and a representative number x^ from each subinterval [x*_i, x*].
Next, for each index k, with k= 1,2,,n, construct a rectangle of width
Ax* = xk - Xk-i
and height
/(Xfe) - g(x$
equal to the vertical distance between the two curves at x = x^. A typical approximat¬
ing rectangle is shown in Figure 6.2b. We refer to this approximating rectangle as a
vertical strip.
Figure 6.2 Using Riemann sums to find the area between two curves
6.1 Area Between Two Curves 357
and the total area between the curves y — /(x) and y = g(x) can be estimated by the
Riemann sum
n
A= - g(.4n&xk
k= 1
which we recognize as the integral of the function /(x) — g(x) over the interval [a, b].
These observations may be used to define the area between the curves.
^ What This Says To find the area between two curves on a given closed
interval [a, b] use the formula
rb
A— [TOP CURVE - BOTTOM CURVE] dx
Ja
Solution
The region is shown in Figure 6.3. We need to know which curve is the top curve on
[0, 1], Solve
x3 = x2 — x or x{x2 — x T 1) = 0
to find the points of intersection. The only real root is x = 0 (x2 — x + 1 =0 has no real
roots). Thus, by the intermediate value theorem, the same curve is on top throughout
the interval [0, 1]. To see which curve is on top, take some representative value in
[0, 1], such as x = 0.5, and note that because (0.5)3 > 0.52 — 0.5, the curve y — x3
must be above y — x2 — x. Thus, the required area is given by
Figure 6.3 The area between the A = [ [ x3 - (x2 x)]dx = (|x4 - |x3 + ^x2)|‘ =
curves y — x} and y = x2 — x
Jo ---'
Top curve Bottom curve
on [0. 1]
358 Chapter 6 Additional Applications of the Integral
In Section 5.1, we defined area for a continuous function / with the restriction that
f(x) > 0. Example 1 shows us that when considering the area between two curves, we
need to be concerned no longer with the nonnegative restriction but only with whether
f(x) > g(x). We now use this idea to find the area for a function that is negative.
Solution
We find the points of intersection of the curve and the x-axis:
The graph of f(x) = e2x — 3ex + 2 is shown in Figure 6.4. We see that on the interval
[0, In 2], /(x) < 0, but we can find the area of the given region by considering the area
between the curves defined by equations y = 0 and y = e2x — 3ex + 2.
In 2 02x In 2
-1 [ 0
Top curve
- (e2x - 3ex + 2)] dx
Bottom curve
+ 3ex — 2x
0
3
Figure 6.4 Area of region 2 In 2 ~ 0.114
2
Although the only mathematically correct way to establish a formula involving inte¬
grals is to form Riemann sums and take the limit according to the definition of the
definite integral, we can simulate this procedure with approximating strips. This sim¬
plification is especially useful for finding the area of a complicated region formed when
two curves intersect one or more times, as shown in Figure 6.5. Note that the vertical
strip has height /(x) — g(x) if y — f(x) is above y = g(x), and height g(x) — f(x)
if y = g(x) is above y = f(x). In either case, the height can be represented by
| f(x) — g(x)|, and the area of the vertical strip is
^ = f | f(x) - g(x)| dx
WARNING w .
You cannot use the
formula A = /[/ — g]dx directly
here because the hypothesis / > g is
not satisfied. In order to use
A =f\f~g\ dx over the entire
interval, you must remember that
1/ — g| might be / — g over part of
the interval and g — / over another
part (see Figure 6.5b). Make sure you
check to see which curve is on top.
Because the curves cross in Example 3,
the interval must be subdivided
accordingly. a. A typical vertical strip b. Approximation by strips
Solution
The region between the curve and the line is the shaded portion of Figure 6.6. Part of
the process of graphing these curves is to find which is the top curve and which is the
bottom. To do this we need to find where the curves intersect:
The points of intersection occur at x = —3, 0, and 1. In the subinterval [—3, 0], labeled
A | in Figure 6.6, the curve y = x3+2x2 is on top (test a typical point in the subinterval,
such as x — —1), and on [0, 1], the region labeled A2, curve y — 3x is on top. The
representative vertical strips are shown in Figure 6.6, and the area between the curve
and the line is given by the sum
Solution
We sketch the region as shown in Figure 6.7. The area we seek is found by
r0 rn/l
/ (0 — sinx)dx+ / (sinx — 0) dx — cos xl0^ + (— c°sx)|q
J-jt/2 Jo
= (1 -0) + (0+l) = 2
There is an easier procedure using symmetry of the curve y — sin x with respect to
the origin. Symmetry uses the fact that the area bounded by the lines x = —7r/2 and
x = 0 is the same as the area bounded by the lines x = 0 and x = it 12. As a result,
Figure 6.7 Area bounded by y — sinx,
y — 0, x = —7r/2 and x = 7r/2 we often work this type of problem as shown here:
7T/2
= 2/
2 /
0
(sinx — 0)dx By symmetry
2(— cosx)!^"
= 2(0+1)
= 2
WARNING
/
A common mistake is to work this problem by evaluating the integral
’?r/2
7T/Z
sinx dx = (— cosx)|*£,2 = —0 + 0 = 0
„/2:
forgetting that the integral does not give the area unless the function is nonnegative.
360 Chapter 6 Additional Applications of the Integral
For many regions, it is easier to form horizontal strips rather than vertical strips. The
procedure for horizontal strips duplicates the procedure for vertical strips. If we want
to find the area between two curves of the form x = F(y) and * = G(y) on the
interval [c, d] on the y-axis, we form horizontal strips. In Figure 6.8, such a region and
its typical horizontal approximating rectangles, or horizontal strips, are illustrated.
Note that the width of the horizontal strip is denoted by Ay.
Regardless of which curve is “ahead” or “behind,” the horizontal strip has length
|F(y) — G(y)| and area
AA = |F(y) - G(y)| Ay
However, in practice, you must make sure to find the points of intersection of the
curves and divide the integrals so that in each region one curve is the leading curve
(“right curve”) and the other is the trailing curve (“left curve”). Suppose the curves
intersect where y = b for b on the interval [c, d], as shown in Figure 6.8. Then
A = f
Jc
[G(y) - F(y)]dy + [ [F(y) - G(y)]dy
K-v-' Jb '---'
G leading F F leading G
Solution
Figure 6.9 shows the region between the parabola and the line, together with a typical
horizontal strip. To find where the line and the parabola intersect, solve
4y — y2 = 2y — 3 to obtain y = — 1 and y = 3
Throughout the interval [—1,3] the parabola is to the right of the line (test a typical
point between —1 and 3, such as y = 0). Thus, the horizontal strip has area
and the area between the parabola and the line is given by
Figure 6.9 The area between
the curves x = Ay — y2 and
x = 2y — 3 using horizontal strips
A — J [(4y — y2) — (2y — 3)] dy = J (3 + 2y-y2)dy
In Example 5, the area can also be found by using vertical strips, but the procedure
is more complicated. Note in Figure 6.10 that on the interval [—5, 3], a representative
vertical strip would extend from the bottom half of the parabola y2 — Ay + x = 0 to the
line y — j(x + 3), whereas on the interval [3, 4], a typical vertical strip would extend
from the bottom half of the parabola y2 - 4y + x = 0 to the top half. Thus, the area
is given by the sum of two integrals, the second of which requires solving the equation
y2 — 4y + x = 0 for y using the quadratic formula. With some effort, it can be shown
that the computation of area by vertical strips gives the same result as that found by
horizontal strips in Example 5.
Q 26. WHAT DOES THIS SAY? When finding the area between
5. x — y2 5y, 6. x — y2 — 6y, two curves, discuss criteria for deciding between vertical and
x =0 x— y horizontal strips.
27. Find the number k (correct to two decimal places) so that the
line y = k bisects the area under the curve y = sin-1 x for
0 < x < 1.
28. Find the area of the region that contains the origin and is
bounded by the lines 2y = 11 — x and y = lx + 13 and
the curve y = x2 — 5.
29. Show that the region defined by the inequalities x2 + y2 < 8,
Sketch the region bounded between the given curves and then find x > y, and y > 0 has area n.
the area of each region in Problems 7-25. 30. Find the area of the region bounded by the curve \/x + yy = 1
7. y = x2, y — x, x = — 1, x = 1 and the coordinate axes.
362 Chapter 6 Additional Applications of the Integral
Q 31. Modeling Problem Imagine a cylindrical fuel tank of length L may be modeled by
lying on its side; the ends are circular with radius b. Determine
the amount of fuel in the tank for a given level by completing
V(h) = 2 L y1 dy
these steps:
a. Explain why the volume of the tank may be modeled by
c. Finally, for b = 4 and L = 20, numerically compute V(h)
for h = —3, —2,..., 4. Note: V(0) and V(4) will serve as
a check on your work.
32. The curve with the equation (x2 + y2)3 = x2 has a loop for
0 < x < 1. Find the area of the loop. Hint: Use the substitu¬
b. Explain why the volume of fuel at level h (—b < h < b) tion u — x1/3 in the area integral.
6.2 Volume
IN THIS SECTION method of cross sections, method of disks and washers, method of cylindrical shells
The methods used in Section 6.1 to compute area by integration can be modified to
compute the volume of a solid region. We will develop methods for computing volume
of several kinds of solids in this section, beginning with the case where the solid has a
known cross section.
Let S' be a solid and suppose that for a < x < b, the cross section of S that is per¬
pendicular to the x-axis at x has area A(x). Think of cutting the solid with a knife at
x = xl and removing a very thin slab whose face has area A(xk) and whose thickness
is Ax*, as shown in Figure 6.11.
Vn = J2A(x*k)Axk
6.2 Volume 363
V = lim A(xr)Axk
'IPII-vO
k= 1
Volume of a Solid with Known A solid S with cross-sectional area A (jc) perpendicular to the x-axis at each point
Cross-Sectional Area on the interval [a, b] has volume
V = f A(x) dx
Solution
The solid resembles a tapered brick, and it may be constructed by “gluing” together a
number of thin slabs with square cross sections, like the one shown in Figure 6.12c.
We begin by drawing the base in two dimensions and then find the volume of the kth
slice.
Ax
To model this construction mathematically, we subdivide the interval [0, 3], form
a vertical approximating rectangle on each resulting subinterval, and then construct
a slab with square cross section on each approximating rectangle. If we choose the
thickness of a typical slab to be Ax and the height and width to be L, the slab will have
volume A V, where
AV - L2Ax
= L2(x) Ax
= [(2x + 5) - (1 — x)]2Ax
= (3x 4)2Ax
364 Chapter 6 Additional Applications of the Integral
The volume of the entire solid is obtained by integrating to “add up” all the volumes
A V, and we find that the volume of the solid is
V = I (3x + 4)“ dx
I
This pyramid can be constructed by stacking a number of thin square slabs. Suppose
that a representative slab has side i and thickness Ah, and that it is located /? units
above the base of the pyramid as shown in Figure 6.13. By creating a proportion from
corresponding parts of similar triangles, we see that
l H-h ( h \
L H \ H)
AV = l2 Ah
= V - 77/ L Ah
To compute the volume V of the entire pyramid, we integrate with respect to h from
the base of the pyramid (h = 0) to the apex (h = H). Thus,
H H
h
L2dh = L2 f 1-h + dh
-L H' JO H H2
,2 l3 ' H
h H2 H3 \ 1
— L1 \ h + = -HL1
H 3 H2 o 3 H2)
Other volume formulas can be found in a similar fashion (see Problems 61 and 62).
6.2 Volume 365
AVk(x) = jr[f(xk)]2Axk
Area of circular cross section
The disk method is used to find a volume generated when a region R is revolved about
an axis L that is perpendicular to a typical approximating strip in R. Suppose R is the
region bounded by the curve y = /(x), the x-axis, and the vertical lines x = a and
x = b. Then if R is revolved about the x-axis, it generates a solid with volume
V - 7r[/(x)]2 dx
+ What This Says The following diagram may help you remember the key ideas
behind the disk method.
Representative
Element
Solution
(a2 + 1 y dx
=* f (a4 + 2a2 + 1) dx
^5,^3 206
a. Theregion under
y = x1 + 1 on [0,2]
= 71 -x H-X +A
nr-jc ^43.14453911
With a small modification of the disk method, we can find the volume of a solid
figure generated by revolving about the A-axis the region between two curves y = f(a)
and y = g{a), where /(a) > g(A) for a < x < b. When a typical vertical strip is
revolved about the A-axis, a “washer” with cross-sectional area 7t([/(a)]2 — [g(A)]2)
is formed, as shown in Figure 6.16. This can be remembered by 7xR2 — nr2, where R
is the outer radius and r is the inner radius.
The volume of the solid of revolution is found by the formula given in the follow¬
ing box.
The washer method is used to find volume when a region between two
curves is revolved about an external axis perpendicular to the approximating
strip. In particular, suppose / and g are continuous functions on [a,b] with
f(x) > #(a) >0. If R is the radius of the (outer) curve y — /(a) and r is
the radius of the (inner) curve y = g(a), and the region bounded by y — /(a),
y — g(x), x — a, and a = b is revolved about the A-axis, then the volume thus
formed is
b
WARNING
f2-82^(f~g)2 n{[f(x)]2 ~ [gU)]2) dx
Outer radius Inner radius
The disk method and the washer method also apply when the axis of revolution is
a line other than the A-axis. In Example 4, we consider what happens when a particular
region R is revolved not only about the A-axis, but also about other axes.
6.2 Volume 367
Solution
The region D to be rotated is shown in Figure 6.17. Note that since x = x2 at x — 0
and at x — 1, the line and parabola intersect at the origin and at (1, 1).
a. The line is always above the parabola on the interval [0, 1], so when we form a
washer to approximate the volume of revolution, the outer radius is R = x and the
inner radius is r = x2, as shown in Figure 6.17a. Thus, the required volume is
2n
= 7r ft; ~15
b. Because we are revolving R about the y-axis, we use horizontal strips to approx¬
For the disk method or the
washer method, make sure that the
imate the solid of revolution, as shown in Figure 6.17b. Note that the parabola
approximating strips are perpendicular x = y/y is to the right of the line x = y on the interval [0, 1], so the approximating
to the axis of revolution. washer has outer radius R — Jy and inner radius r = y; thus.
V =n [ [(Vy)2 ~ (yf]dy
Jo
= n f\y- y2)dy
Jo
yr
rz!
1.2 ” 3 . o 6
c. Since the axis of rotation is the line y — 2, the outer radius of a typical approximat¬
ing washer is R = 2 — x2, and the inner radius is r = 2 — x, as shown in Figure
6.17c. Thus, the volume of revolution is
Radius Circumfereii'
a. A vertical strip in the region b. When the strip is revolved c. The unwrapped
R under the curve y = f(x) about the y-axis, a shell is “flattened” shell has
on the interval [a, b) generated. volume XV = 2xxf(x)Ax.
Because the strip is x units from the axis of rotation and is assumed to be very thin,
the cross section of the shell (perpendicular to the y-axis) will be a circle of radius x
and circumference 2jtx. If we imagine the shell to be cut and flattened out, it is seen
to be a rectangular slab of volume
Thickness
AV = 2irxf(x) Ax
Area of the rectangular slab
6.2 Volume 369
(See Figure 6.18c.) Thus, the total volume of the solid is given by the integral
V
-f Ja
2nxf(x) dx
The approximating shells are shown in Figure 6.19 and the formula is repeated in the
following box.
The shell method is used to find a volume generated when a region R is revolved
about an axis L parallel to a typical approximating strip in R. In particular, if
R is the region, as shown in Figure 6.20, bounded by the curve y — /(x), the
x-axis, and the vertical lines x — a and x — b, where 0 < a < b, then the solid
generated by revolving R about the y-axis has volume
-r
V — I
*b
2:rxf(x)dx
Solution
The region to be rotated is shown in Figure 6.21, along with a typical vertical strip of
height f{x) — x3 + x1 + 1 and width Ax. The volume, by the method of shells, is
Solution
The region R is shown in Figure 6.22, along with a typical vertical strip of height
f(x) = x~2. If we were revolving this strip about the y-axis, the distance from the
axis to the strip would be L = x, but since the axis of rotation is x — —1, the distance
is L = x + 1.
V = dx
+ x 2) dx
= 2n [lnx — jc *] |2
= 2n [in 2 + |]
7.4968
Figure 6.22 Volume by shells
(rotating about x = — 1) Sometimes you have a choice between using shells or washers to compute a vol¬
ume of revolution. The decision as to which method to use often comes down to which
leads to the easier integration. Which method would you choose in the following ex¬
ample?
An 2)3/2
= 4lr[_i(*2—^)3/2]|ff = -(R2
= Y<*2-r2)3/2 ■
Figure 6.25 Volume of solid by
washer method; strip Table 6.1 compares and contrasts the disk, washer, and shell methods for comput¬
perpendicular to axis of rotation ing volume.
6.2 Volume 371
TABLE 6.1 Volumes of revolution when the axis of revolution is either the x-axis or the y-axis.
Summary
a. Disk method: A representative rectangle is perpendicular b. Washer method: A representative rectangle is perpendicular
to the axis of revolution. The axis of revolution is a boundary of to the axis of revolution. The axis of revolution is not part of the
the region. boundary.
Horizontal axis of revolution Vertical axis of revolution Horizontal axis of revolution Vertical axis of revolution
* y-
Vertic al axis af revolution Horizontal a xis of revolution
A.v dy‘ Vertical axis )f revoluti on
y = fO-)
A d-
=g< x) .x ■-= /( ■0
> y
\ A
\ y = g(x )
\\ f
xy = g(yf
L X
c b_
X X
Distance to axis
372 Chapter 6 Additional Applications of the integral
C(3, -2) N,
In Problems 9-12, sketch the region and then find the volume of the l =—
solid where the base is the given region and which has the property \
that each cross section perpendicular to the x-axis is a semicircle. \
’r
9. the region bounded by the y-axis, the parabola y = x2, and
the line 2x + y — 3 = 0
10. the region bounded above by y — cosx, below by y = sinx, .|.
■3.
and on the left by the y-axis ■
11. the region bounded above by the curve y = tanx and below
by the x-axis, on the interval [0, |] '"'N
12. the region bounded by the x-axis and the curve y = ex be¬
1.3
:: \
tween x = 1 and x = 3 \
In Problems 13-20, find the volume of the solid formed when the jc\
;
region described is revolved about the x-axis using washers or
disks.
31. y = e *, 0 < x < 1
13. the region under the curve y = >/x on the interval 0 < x < 1
14. the region under the curve y = fife on the interval 0 < x < 8
15. the region bounded by the lines y = x, y = 2x, and x = 1
/T dy
16. the region bounded by the lines x = 0, x = 1, y = x + 1, and
/ 1 wi
y =x + 2
i0 1.5 *1
17. the region under the curve y — x2 + x3 on the interval
0 < x < 7r; approximate your answer to the nearest unit
18. the region under the curve y = \/2sinx on the interval
0 <x <n
33. y = sin 1 x, 0 < x < 1 34. y = lnx, 1 < x < 2
19. the region between y = sinx and y = cosx on 0 < x < n/4
20. the region bounded by the curves y = ex and y = e~x on
[0. 2]
In Problems 21-26, use shells to find the volume of the solid formed
by revolving the given region about the y-axis.
21. the region bounded by the lines y = 2x, the y-axis, and y = 1
22. the region bounded by the curve y = -Jx, the y-axis, and the
line v = 1
6.2 Volume 373
In Problems 35-40, draw a representative strip and set up an in¬ In Problems 46-49, find the volume of the solid whose base is
tegral for the volume of the solid formed by revolving the given bounded by the circle x2 + y2 = 9 with the indicated cross sections
region: taken perpendicular to the x-axis.
a. about the x-axis b. about the y-axis 46. squares
Set up the integral only; DO NOT EVALUATE.
35. the region bounded by 36. the region bounded by
x = y2 and y = x2 y = \x and x = y2
37. the region bounded by the 38. the region bounded by the
lines y — 1, x = 2, and the curves y — x2 and
curve v = x2 + 1 y - —x2 — 4x
A
2_
y= -X! 4x\ 48. isosceles right triangles
y-- X2
J
kl\-
_> -2 -1 *
-1-
\
\
39. the region bounded by 40. the region bounded by
y = O.lx2 and y = lnx y = ex — 1, y = 2e~x, and 49. semicircles
x =0
In Problems 50-51, find the volume V of the solid with the given
information regarding its cross section.
Q In each of Problems 41-44, find the volume by the 50. The base of the solid is the hyperbola
a. disk/washer method b. shell method
41. Revolve the region bounded by y — x2 and y = x3 about the
x-axis.
42. Revolve the region bounded by y — x2 and y = x3 about the for 2 < x < 5, and the cross sections perpendicular to the
y-axis. x-axis are squares.
43. Revolve the region bounded by y — x, y — 2x, and y = 1
about the x-axis.
44. Revolve the region bounded by y = x, y = 2x, and y = 1
about the y-axis.
45. Derive the formula for the volume of a sphere of radius r by
revolving the region bounded by the semicircle y = -Jr2— x2
about the x-axis.
374 Chapter 6 Additional Applications of the Integral
51. The base of the solid is an equilateral triangle, each side of 59. Cross-sectional areas are measured at 2-meter intervals along
which has length 4. The cross sections perpendicular to a the length of an irregularly shaped object, with the results
given altitude of the triangle are squares. listed in the following table (x in meters and A in nr):
X 0 2 4 6 8 10
x 12 14 16 18 20
54. When viewed from above, a swimming pool has the shape of In this Historical Quest we look at Kepler’s derivation for
the ellipse the volume of a torus, generated by revolving a circle of radius
xz2 yL2 a around a vertical axis at a distance b from its center (b > a).
= 1 He found V = (7ra2){2nb) = 2n2a2b. He derived this for¬
900 + 400
mula by dissecting a torus into infinitely many thin vertical
The cross sections of the pool perpendicular to the ground and
circular slices by considering planes perpendicular to the axis
parallel to the y-axis are squares. If the units are in feet, what
of revolution, as shown in Figure 6.26.
is the volume of the pool?
55. Find the volume of the solid generated when the region
= r-V2 on the interval [ 1,4] is revolved about
y = *
ing the trapezoidal rule. * Vol. 1, James R. Newman (New York: Simon & Schuster, 1956), p. 220.
6.3 Polar Forms and Area 375
62. Use the method of cross sections to show that the volume of a 66. The frustum of a cone is the solid region bounded by a cone
regular tetrahedron of side a is -L \Z2a}. and two parallel planes, as shown in Figure 6.28. Suppose the
63. A student, Frank Kornercutter, conjectures that a tetrahedron planes are h units apart and intersect the cone in plane regions
is nothing more than a solid figure with an equilateral triangu¬ of area A\ and A2, respectively. Use integration to show that
lar base and cross sections perpendicular to that base that are the frustum has volume
also equilateral triangles. Is Frank correct this time? Either
h r-
prove that he is or show that his conjecture must be false. V — — (A | + f A\A2 + A2)
64. Find the volume of the football-shaped solid (called an ellip¬
soid) formed by revolving the ellipse
y.
In the polar coordinate system, points are plotted in relation to a fixed point O, called
the origin or pole and a fixed ray emanating from the origin called the polar axis.
Radial axis/
pA. pUilll WIIH We then associate with each point P in the plane an ordered pair of numbers P(r, 9),
yp(r.9 )
where r is the distance from O to P and 6 is the angle measured from the polar axis
vallue for r T i n-lA point With to the ray OP, as shown in Figure 6.29. The number r is called the radial coordinate
(eraser end) a positive value
1 ' d (/
f for r (pourt end)
\ _l i_If_*
of P, and 0 is the polar angle. The polar angle is regarded as positive if measured
Jr • ' V
-4 4 x counterclockwise up from the polar axis, and negative if measured clockwise. The
' F ole Polar axi s
origin O has radial coordinate 0, and it is convenient to say that O has polar coordinates
C }(r, 6
0 may be (0, 6), for all angles 9.
positive or
negative If the point P has polar coordinates (r, 9), we say that the point Q obtained by
/ _!_1 _ reflecting P in the origin O has coordinates (-r, 9). Thus, if you think of a pencil
lying along the directed line segment OP, with its midpoint at O and tip at P(r, 9),
then the eraser will be at Q(-r, 9), as illustrated in Figure 6.29.
Figure 6.29 Polar-form points
While each point in the plane is associated with exactly one pair of rectangular co¬
ordinates, a given point in polar coordinates has an infinite number of representations.
For example, (5, ^), (-5, §), and (5, -f) all represent the same point in polar co¬
(ImT) For a more complete ordinates. The nonuniqueness of representation in the polar coordinate system causes
discussion of the polar coordinate
some difficulties, but they can be handled by exercising a little care. We will point out
system, see Chapter 5 of the Student situations in our examples where the multiplicity of polar representation must be taken
Mathematics Handbook. into account.
376 Chapter 6 Additional Applications of the Integral
Note that the preceding formula gives r as the distance from the pole to the point.
This does not mean that ordered pairs in polar form need to have positive first compo¬
nents. For example (l, is the same point as (—1, |).
Solution
3y/2
3sf2
3sf2 3y/2.\
The rectangular coordinates are
5a/3 5'
Write polar-form coordinates for the point with rectangular coordinates
2 ’ 2
'5J3' /75 25
Solution r = — + — = 5
+ -2 = 4 4
\
-1
1 \ n ll7r
0 = tan-1 = tan -7= I = —; thus, 9 = —— (Quadrant IV)
V3 y/3j 6 6
2
Polar-form coordinates are (5, -^p). A representation with a negative r is (-5, ^).
6.3 Polar Forms and Area 377
POLAR GRAPHS
The graph of an equation in polar coordinates is the set of all points P whose polar
coordinates (r, 0) satisfy the given equation. Circles, lines through the origin, and rays
emanating from the origin have particularly simple equations in polar coordinates.*
a. r = 6 b. 6 = |-
Solution
a. The graph is the set of all points (r, 6) such that the first component is 6 for any
angle 6. This is a circle with radius 6 centered at the origin, as shown in Figure
6.30a.
b. The graph is the set of all polar points (r, 0) with polar angle 6 — |. This is a line
through the origin (pole) that makes an angle n/6 with the positive x-axis, as shown
in Figure 6.30b. If we further require that r > 0, then the graph is a ray (sometimes
called a half-line). B
x2 + y2 = 4x r2 — x2 + y2 and x — r cos 9
x2 - 4x + y2 = 0
This is a circle with center at (2, 0) and radius 2, as shown in Figure 6.31.
Figure 6.31 Graph of
r = 4 cos 9
In our next example, we sketch a more complicated polar graph by plotting points
in a polar coordinate plane. The graph is called a cardioid because of its heart-like
shape.
*Many books use what is called polar graph paper, but such paper is not really necessary. It also obscures
the fact that polar curves and rectangular curves are both plotted as ordered pairs, only with a different
meaning attached to the ordered pairs. You can estimate the angles as necessary without polar graph paper.
378 Chapter 6 Additional Applications of the Integral
0 0
l 0.9193954
2 2.832294
3 3.979985
4 3.307287
i-iB2 C1-cos 0) Xmin= "5
Xmax=5 5 1.432676
Xscl=l
Vnin= "5 6 0.079659
Yroax=5
Ysc1=1 The points are connected as shown in
Figure 6.32. Figure 6.32 Graph of r — 2(1 — cos 6) ■
r — b ±a cos 9 or r = b db a sin 9
is called a limagon (derived from the Latin word Umax, which means “slug”; in French,
it is the word for snail). The special case where a = b is the cardioid. Figure 6.33
shows four different kinds of lima^ons that can occur. Note how the appearance of the
graph depends on the ratio b/a.
y>
/ r~%
f r *
*x
-i -
1
Summary of Polar-Form Curves There are several polar-form curves which you
can graph by plotting points or by using a graphing calculator (be sure to convert to
polar form). Table 6.2 gives you the names of some of the special types of polar-form
curves you will encounter. There are many others, some of which are represented in the
problems. Others may be found in Chapter 5 of the Student Mathematics Handbook.
6.3 Polar Forms and Area 379
Limatons in which a — b
If n is odd, the rose has n petals (leaves); if n is even, it has 2n petals. Two loops
r = a cos 0 r = a sin 0
standard form ^ rotation,
one (circular) petal
one (circular) petal
To find the points of intersection of graphs in rectangular form, you need only find the
simultaneous solution of the equations that define those graphs. It is not even necessary
to draw the graphs, because there is a one-to-one correspondence between ordered pairs
satisfying an equation and points on its graph. However, in polar form, this one-to-one
property is lost, so that without drawing the graphs you may fail to find all points of
intersection. For example, (— 1, n) and (1,0) both represent the same point, in polar
coordinates. Therefore, our method for finding the intersection of polar-form curves
will include sketching the graphs.
r = §—cosf = §- H) =2
0 = | — cos 0 or cos 6 — |
and a cosine cannot be larger than 1.
Step 3. Now look at the graphs, as shown in Figure 6.34. We see that (—1, =f) may
also be a point of intersection. It satisfies the equation 6 — but what
about r = | — cos0? Check (—1, ^):
— 1 = | — cos (^)
Figure 6.34 Graphs of
r = | — cos 9 and 6 — = i-H)
= 2 Not satisfied
= 1 Satisfied
Be sure to check for points of intersection that you may have missed. ■
POLAR AREA
To find the area of a region bounded by a polar graph, we use Riemann sums in much
the same way as when we developed the integral formula for the area of a region
described in rectangular form. However, instead of using rectangular areas as the basic
units being summed, in polar form, we sum areas of circular sectors. A typical circular
Figure 6.35 Sector of sector is shown in Figure 6.35. Recall the formula for the area of such a sector from
a circle trigonometry.
6.3 Polar Forms and Area 381
Area of a Sector
The area of a circular sector of radius r is given by
A = \r20
Using this formula for the area of a sector, we now state a theorem that gives us a
formula for finding the area enclosed by a polar curve. If r = f(9), then we call r a
polar function of 9.
Proof The region is shown in Figure 6.36. To find the area bounded by the graphs of
the polar functions, partition the region between 9 — a and 9 = f by a collection of
rays, say 9q,9\, ... ,6n. Let a = 9q and f = 9n.
Pick any ray 9 — 9£, with 9^-1 < 9£ < 9*. Then the area Aof the circular
sector is approximately the same as the area of the region bounded by the graph of /
and the lines 9 — 9k-1 and 9 = 9k. Because this circular sector has radius f(9£) and
central angle A9k, its area is
The sum ^A Ak is an approximation to the total area A bounded by the polar curve,
*=t
and by taking the limit as n —> oo, we obtain
M 1
A = lim / ^ A Ak = lim — \f A9k Riemanri sum
«-»oo ' n-+oo 2 L’ ' /J
k=l k=1
1 Cp
= - / [fm2de
^ Ja
382 Chapter 6 Additional Applications of the Integral
You will probably find that the most difficult part of the problem is deciding on the
limits of integration. A sketch of the region will help with this.
Solution
The cardioid is shown in Figure 6.37. Note that the top half of the graph lies between
the rays 0 = 0 and 0 = tv. Hence the required area is given by
=i
2 Jo
r (i +cos eyde =if (1+2 cos 6 + cos2 9) d9
2 Jo
1 0 sin 26
9 -\- 2 sin 6 + — -\--— Integration table, formula 317
2 2 4 o
TV 0 37T
Figure 6.37 Area of the top half tv + 2(0) + — + - — 0 T
of the cardioid r = 1 + cos 6
Solution
The rose curve is shown in Figure 6.38. We will find the area of the top half of the right
loop (shaded portion) to make sure f{9) > 0. We see that this corresponds to angles
with measures from 9 — 0 to 9 — tv/4. If you have a calculator, the easiest way to see
this is to plot the graph and then use the trace feature to find the 0-values for the top
half of the first leaf. By symmetry, the entire area enclosed by the four-leaved rose is 8
times the shaded region. Thus, the required area is given by
A — 8
r -i J/,7r/4 cos2 29 d9
I
Figure 6.38 Area enclosed by the 7T/4
\9 sin 49
four-leaved rose r = cos 26 — 4 — +- Integration table, formula 317
2 8 0
TV
= 4 + 0-0
Li
TV
~ 2
Solution
The circles are shown in Figure 6.39. Solve a cos 9 = a sin 9 to find that the circles
intersect at 9 = 7r/4 and at the pole. To set up the integrals properly, remember to think
in terms of polar coordinates and not in terms of rectangular coordinates. Specifically,
we must scan radially. This means that we need to find the area of intersection by
adding the areas of the regions marked R\ and /?2- Region R\ is bounded by the circle
r = a sin 9 and the rays 9 = 0, 9 — f. Region R2 is bounded by the circle r = a cos 9
Figure 6 1 Area bounded by and 9 — j, 9 = j. Note that the rays 0 = 0 and 0 = ~ are not necessary as geometric
r = a cos 6 and r = a sin 9 boundaries for R\ and R2, but they are necessary to describe which part of each circle
6.3 Polar Forms and Area 383
is being calculated.
A = AREA OF R] + AREA OF R2
] P*/4 1 pn/2
Pn/2
= - / «2 sin2 0 dO H— / 2 cos2 0 dO
2 Jo /
2 7+4 °
a 0 sin 20 +4 a
2 r0 sin 20
+2
Integration table,
2 4 0 +T 2 ^ 4
n/A formulas 340 and 317
o
7T 1 a2 ' n tc 1" a TC 1 "
+ — — +0— — — — — —
8-4-° 2 [_ 4 8 4J 2 4 ~ 2
—g~{tz — 2)
8
Solution
As usual, begin by drawing the graphs, as shown in Figure 6.40. Note that both the
limagon and circle are symmetric with respect to the x-axis, so we can find the area in
the first quadrant and multiply by 2.
Next, we need to find the points of intersection. We see that the curves do not
intersect at the pole. Now solve
5 cos0 = 2 + cos0
To find the area, divide the region into two parts, along the ray 0 = n/3. The right part
(gray region above the x-axis) is bounded by the limagon r = 2 + cos 0 and the rays
0=0 and 0 = 7r/3. The left part (shaded region shown in color above the x-axis)
is bounded by the circle r = 5 cos0 and the rays 0 = jt/3 and 0 = tx/2. Using this
preliminary information, we can now set up and evaluate a sum of integrals for the
required area
1 rn/1
= 2
2
7T/3
L (2 + cos 0)2 dO +
' 2 Jjt/3
(5 cos Oy dO
3 2 6 8 4 6 8
43tt
-V3^ 9.53
1Y
384 Chapter 6 Additional Applications of the Integral
a. r = 2 cos 29 b. r = 4 sin 30° 45. Find the area of the region that is inside the circle r = a and
c. r + 2 = 3 sin 9 d. r + 3 = 3 sin 9 outside the cardioid r = a( 1 — cos0).
e. 9 = 4 f. 9 = tan f 46. Find the area of the region that is inside the circle r = sin 0
g. r = 3 cos 50 h. r cos 9 = 2 and outside the cardioid r = 1 — cos 0.
Graph the polar-form curves given in Problems 6-21. 47. Find the area of the region that is inside the circle r = 6 cos 0
and outside the cardioid r = 2(1+ cos 0).
6. r = 3, 0 < 9 < f 7. 0 = -f,0 < r < 3
8. r = 9 + \,0 <9 <n 9. r = 20, 0 > 0 48. Find the area of the portion of the lemniscate r2 = 8 cos 20
that lies in the region r >2.
10. r = 2 cos 20 11. r = 5 sin 30
12. r2 = 16 cos 20 13. r = 3 cos 3(0 — j) 49. Find the area between the inner and outer loops of the limagon
r = 2 — 4sin0.
14. r = 5 cos 3(0 - f) 15. r = sin(20 + |)
50. Find the area to the right of the line r cos 0 = 1 and inside the
16. r2 = 16 cos 2(0 - f) 17. r = 2 + cos0
lemniscate r2 = 2 cos 20.
18. r = 1 + sin0 19. r cos 0 = 2
51. Find the area to the right of the line r cos 0 = 1 and inside the
20. r = 1 + 3 cos 0 21. r = —2sin0
lemniscate r2 = 2 sin 20.
Find the points of intersection of the curves given in Problems 22-
52. Find the maximum value of the y-coordinate of points on the
29.
lima9on r = 2 + 3 cos 0.
r = 4cos0
22. • 23. 53. Find the maximum value of the x-coordinate of points on the
r = 4sin0
cardioid r = 3 + 3 sin0.
r2 = 9 cos 20 \ r = 2(1 + sin0) G54. a. Show that if the polar curve r = f(9) is rotated about the
25.
[r = 2(1 - sin0) pole through an angle a, the equation for the new curve is
24 l'-3
r = f (9 — a).
[ r2 = sin 20
II
32. /(0) = sec0, -f < 0 < | 33. /(0) = Vsiiifl, £ < 0 < f In particular, show that the graph has a vertical asymptote at
34. f(9) = e0/2,O<9 <2n x = -
X 2 .
6.4 Arc Length and Surface Area 385
57. If / is a differentiable function of 9, then show that the tan¬ Let P be a point on the polar curve r = f (9), and let a be
gent line to the polar curve r = f{9) at the point P(r, 6) has the angle extending from the radial line to the tangent line, as
slope shown in Figure 6.41. Assuming that f'{9) 0, show that
f(9) cos 6 + f'(9) sin 9
m = - f(0)
—f(9) sin 0 + f'{9) cos 9 tana =
no)
U . , dy dy/dd
whenever the denominator is not zero. HmV. — =-. tan (p — tan 9
dx dx/d6 Hint: Use the formula tana — tan(</> — 9) =
58. Because the formula for the slope in polar form is complicated 1 + tan (p tan 9
(see Problem 57), it is more convenient to measure the incli¬ 59. Use the formula in Problem 58 to find tana for each of the
nation of the tangent line to a polar curve in terms of the angle following:
a extended from the radial line to the tangent line, as shown a. the circle r = a cos 9
in Figure 6.41. b. the cardioid r = 2(1 — cos 9)
c. the logarithmic spiral r — 2e3e
60. Journal Problem* School Science and Mathematics ■ Graph
59
the polar curve r = 4 + 2 sin — and find the total area inte¬
rior to this curve. Then find the area of the “star.”
k=1
where ASk is the length of the segment joining to PL By applying the distance
formula and rearranging terms with Ax* = x* - x*_i and Ay* = y* — y*_i, we find
Chapter 6 Additional Applications of the Integral
that
(Ax*)2 + (Ay*)2
Ax*
(Ax*)2
It is reasonable to expect a connection between the ratio Ay*/Ax* and the derivative
dy/dx — f'(x). Indeed, it can be shown (see Problem 42) that
for some number x* between x*_i and x*. Therefore, the arc length of the graph of /
on the interval [a, b] may be estimated by the Riemann sum
£jl + [f'(x*k)?Axk
k= 1
lim Wl +if'(x*k)]2Axk
\p\\^o
k= 1
Notice that if f is continuous on the interval [a, b], then so is y/\ + [f'(x)]2. Thus,
this limit exists and is the integral of sj 1 + [f'(x)]2 with respect to x on the interval
[a, b]. These observations lead us to the following definition.*
*We have used integration to obtain a meaningful definition of arc length for a function f(x) with a contin¬
uous derivative f'(x). It is possible to define arc length for certain curves y = f(x) when /(.v) does not
have a continuous derivative, but we will not pursue this more general topic.
6.4 Arc Length and Surface Area 387
Arc Length Let / be a function whose derivative f is continuous on the interval [a, b] and
differentiable on (a,b). Then the arc length, s, of the graph of y = f(x)
between x = a and x — b is given by the integral
Similarly, for the graph of x = g(y), where g' is continuous on the interval
[c, d], the arc length from y = c to y = d is
Solution
The graph is shown in Figure 6.43. Let /(x) = x3/2; therefore, f'(x) = |x1/2.
3-
;
1
Figure 6.43 What is the length of
this curve?
EXAMPLE 2 Arc length of a curve jc = g(y)
Find the arc length of the curve x = |y3 + |y_1 from y = 1 to y = 3.
Solution
, which is con¬
Solution
The region is shown in Figure 6.44. Because y' — cosx, we have, from the arc length
formula,
I>2jt _
s — \/1 + cos2 a: dx
Jo
We do not have techniques to allow us to evaluate this integral, so we turn to numerical
integration. If we consider four rectangles, we find, by various methods:
Method (» = 4) Approximation
Rectangles
Figure 6.44 Finding the length of
an arc Left endpoints 7.584476
Midpoints 7.695299
In practice, you would choose just one of the methods whose approximate values are
given. You might also note that, for this example, Simpson’s rule performs worse
than the trapezoidal, midpoint, or even the left- and right-endpoint methods. For more
accurate results, you might wish to use a computer program. A simple output is shown
in Figure 6.45.
The actual arc length (rounded to six places) computed as an area is 7.640396. ■
When the arc of a curve is revolved about a line L, it generates a surface called a
surface of revolution, as shown in Figure 6.46a. In particular, if the generating arc is
a line segment, the frustum of a cone is generated (Figure 6.46b).
We will obtain a formula for the lateral area of the frustum of a cone, which will
then be used as part of an approximation scheme for the area of a general surface of
revolution. Accordingly, first note that a cone with slant height s and base radius r has
lateral surface area A = nrs. To see this, cut the cone along a dashed line as shown
in Figure 6.47a and flatten it out as shown in Figure 6.47b. The angle 6 is determined
by noting that the circumference of the circular base of the cone is 2nr, which is the
6.4 Arc Length and Surface Area 389
c
f] A
»
A i L
1 L 1
-&4 1
1 V
/
V
w
b. The line segment of
length t generates the
a. The surface generated
frustum of a cone.
by revolving the curve C about L.
same as the arclength sO after the cone is flattened. Thus, 2rcr = sO and 9 — —
.y
The lateral surface area of the cone is the area of sector COD\ namely.
Arc length
is s0
Figure 6.47 A cone with slant height s and base radius r has area A — rrrs
Finally, the frustum of the cone shown in Figure 6.48 has slant height i, top radius
n and bottom radius /a and can be formed by removing a small cone of base radius r\
and slant height i\ from the larger cone of base radius r2 and slant height £2 = ^1+^-
Thus, the lateral surface area F of the frustum can be found by subtracting the area of
figure 6.48 Frustum of cone the smaller cone from that of the larger:
F = nr2t2 — ixr\i\
— 1 + F) — Tir\i\
= x[(r2 -r\)l\ +r2l]
i\ 11 + i
— —- or r\l = (r2 - r\)t\
r\ r2
so that
Next, assuming that the function /(x) is continuous with f(x) > 0 on the interval
[a, b], we will derive a formula for the area of the surface generated by revolving the
curve y = f(x) about the x-axis. Rather than argue formally with Riemann sums,
we will proceed heuristically. Figure 6.49a shows an approximating line segment with
length As, whose endpoints are, respectively, distances r\ and r2 above the x-axis. Let
/(x) be the average height of the segment above the x-axis; that is, f(x) — \ (ri + r2).
By revolving this segment about the x-axis, we generate the frustum of a cone, with
surface area
r 1 + r2
AS - 2jt As = 2nf(x)y/1 + [/'(x)]2 Ax
The area of the entire surface of revolution S can be obtained by integrating this
expression over the interval [a,b]. These considerations lead us to the following defi¬
nition.
Surface Area Suppose f is continuous on the interval [a, b] and differentiable on {a, b). Then
the surface generated by revolving about the x-axis the arc of the curve y = /(x)
on [a, b] has surface area
S — 2n + [f'(x)]2dx
WARNING , .
An easy-to-remember
form for surface area is
S = 2n I v ds since
ds = yj\ + [f'(x)]2dx.
You may find it instructive to derive this formula by the more rigorous approach
outlined in Problem 43. We close with two examples that illustrate the use of the
surface area formula.
Solution
y
The graph is shown in Figure 6.50. Because f(x) = x3, we have f'(x) = 3.v2, which
is certainly continuous on the interval [0, 1].
(i i)
(0*0)
X
L
Let u — 1 + 9x4; du — 36x3 dx.
j\ If x = 1 , then w = 10.
If x = 0, then u = 1.
to
Solution
y. ■ We can generate the surface of the sphere by revolving the semicircle y — Jr2 — x2
about the x-axis (see Figure 6.51). We find that
Because the semicircle intersects the x-axis at jc — r and x = —r, the interval of
$ integration will be [—r, r]. Finally, by applying the formula for the surface area, we
A
-t r t
find
To generalize the formula for surface area to apply to any vertical or horizontal axis
of revolution, suppose that an axis of revolution is R(x) units from a typical element
of arc on the graph of y — f(x), as shown in Figure 6.52.
i
i
i
i i ]
iS i/i
fit y\
i
1
i i
i i
! 1
1 i
| 1
a b x
Then 2tc R(x) is the circumference of a circle of radius R(x), and it can be shown that
an element of surface area is
= 2 Tt f Xy/l + [f'(x)]2dx
Ja
Sometimes we wish to find arc length and surface area in polar form. Figure 6.53
shows the portion of a polar curve subtended by a central angle AO. If AO is small, the
corresponding arc length As may be thought of as the hypotenuse of a right triangle
with sides Ar and r AO, the length of the circular arc AB with radius r. We have,
By formalizing these observations, we are led to compute polar arc length by the fol¬
lowing formula.
Figure 6.53 Polar arc length
Solution
This circle is shown in Figure 6.54, and since this is a unit circle, we know the circum¬
ference to be 2n. We will verify this conclusion using the arc length formula. Note
that 0 < 0 < 7r.
= 2jt
6.4 Arc Length and Surface Area 393
We conclude this section with an example showing the use of this formula.
Solution
The cardioid is shown in Figure 6.55. Since r — 1 + cos# and r' — — sin#, the surface
area is given by
In Problems 27 and 28, use numerical integration to find the arc 40. Show that a cone of radius r and height h has lateral surface
length. area
27. f(x) — sinx on [0,7r] 28. f{x) = tanx on [0, 1]
S — nrfir2 + h2
29. Find the area of the surface generated when the arc of the
curve Hint: Revolve part of the line hy — rx about the x-axis.
y — |x3 + (4x)_1
Q 41. Show that when the arc of the graph of /(x) between x = a
between x = 1 and x — 3 is revolved about: and x — b is revolved about the y-axis, the surface generated
a. the x-axis b. the y-axis has area
30. Find the area of the surface generated when the arc of the
curve S = 2jt + [f'(x)]2dx
x = b5/3 |y/3
-
35. The graph of the equation x2/3 + y2/3 = 1 is an astroid. 43. Verify the surface area formula by completing the following
steps (see Figure 6.56).
a. Let P = {xo, Xi,..., x„} be a partition of the interval
[a,b], and for k = 0,1,_n, let Pk denote the point
(x*, yk), where yk = f{xk). Show that when the line seg¬
ment between Pk-\ and Pk is revolved about the x-axis, it
generates a frustum of a cone with surface area
7r(y*-i + yk)h
44. The center of a circle of radius r is located at (R, 0), where from x = 2 to x — 3. Find a general formula for the length of
R > r. When the circle is revolved about the y-axis, it gen¬ the curve
erates a torus (a doughnut-shaped solid). What is the surface
area of the torus? y = Cxn + Dx2-"
45. Let n be a positive integer and C, D be constants such that
from x — a to x = b for
1
«(« — !) = 1
16CD n(n — 2) =
4 CD
Find the length of the curve
47. Suppose it is known that the arc length of the curve y — f(t)
y = Cx2n + Dx20~n) on the interval 0 < t < x is
WORK
In physics, “force” is an influence that tends to cause motion in a body. When a con¬
stant force of magnitude F is applied to an object through a distance d, it performs
work measured by the product W = Fd.
Work Done by a Constant Force If a body moves a distance d in the direction of an applied constant force F, the
work W done is
TABLE 6.3 Common Units of Work W = Fd
and Force
kg m newton (N) joule For example, the work done in lifting a 90-lb bag of concrete 3 ft is
g cm dyne (dyn) erg W = Fd = (90 lb)(3 ft) = 270 ft-lb. Notice that this definition does not conform
slug ft pound ft-lb to everyday use of the word work. If you work at lifting the concrete all day, but you
are not able to move the sack of concrete, then no work has been done. Table 6.3 shows
some common units of work and force.
If there is no movement, there is no work!
To find the work done in moving an object against a variable force, calculus is
required. Suppose F is a continuous function and F(x) is a variable force that acts on
an object moving along the x-axis from x = a to x — b. We will heuristically define
the work done by this force. Partition the interval [a, b] into n subintervals, and let Ax*
be the length of the kth subinterval /*. If Ax* is sufficiently small, we can expect the
force to be essentially constant on /*, equal, for instance, to F(x£), where x^ is a point
chosen arbitrarily from the interval /*. It is reasonable to expect that the work AW*
required to move the object on the interval /* is approximately AW* = F(x£)Ax*, and
the total work is estimated by the sum
^ ^(x*)Ax*
k= 1
m)
H-h-1-1-1--1-h
a xk_! x*k xk b
Axk
By taking the limit as the norm of the partition approaches 0, we find that the total
work may be modeled by
W = f F(x) dx
Solution
Hooke’s law, named for the English physicist Robert Hooke (1635-1703), states
that when a spring is pulled x units past its equilibrium (rest) position, there is a
restoring force F(x) — kx that pulls the spring back toward equilibrium.* (See Figure
6.58.) The constant k in this formula is called the spring constant.
* Actually, Hooke’s law applies only in ideal circumstances, and a spring for which the law applies is some¬
times called an ideal (or linear) spring.
6.5 Physical Applications: Work, Liquid Force, and Centroids 397
T n
5
The natural length of a certain spring is 10 cm. If it requires 2 ergs of work to stretch
the spring to a total length of 18 cm, how much work will be performed in stretching
the spring to a total length of 20 cm? (See Figure 6.59.)
Solution
Assume that the point of equilibrium is at 0 on a number line, and let a- be the length the
10 cm —
free end of the spring is extended past equilibrium. Because the stretching force of the
spring is F(x) = kx, the work done in stretching the spring b cm beyond equilibrium
is
b rb
In particular, when the total length of the spring is 20 cm, it is extended b — 10 cm,
and the required work is
Solution
Set up a coordinate system with the origin at the vertex of the cone and the y-axis as
the axis of symmetry (Figure 6.60b). Partition the interval 0 < y < 6 (remember, the
water only comes up 6 ft). Choose a representative point y£ in the kth subinterval, and
construct a thin disklike slab of water y£ units above the vertex of the cone. Our plan
is to think of the water in the tank as a collection of these slabs of water piled on top of
each other. We will find the work done to raise a typical water disk and then compute
the total work by using integration to add the contributions of all such slabs.
398 Chapter 6 Additional Applications of the Integral
Note that the force required to lift the slab is equal to the weight of the slab, which
equals its volume multiplied by the weight per cubic foot of water. Let be the radius
of the kth slab. By similar triangles (Figure 6.60b), we have
xt 3 y*
— = —, which implies xt = —
yk 12 4
2 Thickness
AV = nxk2Ayk = A yk
Radius
62.47r Ayu. lb
From Figure 6.60b, we see that the slab of water must be raised 12 — feet, which
means that the work A W required to raise this kth slab is
v ' Distance
Weight (force)
Finally, to compute the total work, we add the work required to lift each thin slab
and take the limit of the sum as the norm of the partition approaches 0.
n
W = lim 62.4tt
n-> oo E (12 - y*)Ayi
If the water is pumped to a height of 3 ft above ground level, all that changes is the
distance moved by the kth slab of water. It becomes 12 + 3 - yk = 15 - y£, and the
work is given by
Anyone who has dived into water has probably noticed that the pressure (that is, the
force per unit area) due to the water’s weight increases with depth. Careful observa¬
tions show that the water pressure at any given point is directly proportional to the
depth at that point. The same principle applies to other fluids.
In physics, Pascal’s principle (named for Blaise Pascal, 1623-1662, a French
Figure 6.61 Pascal’s principle: mathematician and scientist) states that fluid pressure is the same in all directions (see
In an equilibrium state, the Figure 6.61). This means that the pressure must be the same at all points on a surface
fluid pressure is the same in submerged horizontally, and in this case, the fluid force on the surface is given by the
all directions. formula in the following box.
6.5 Physical Applications: Work, Liquid Force, and Centroids 399
Fluid Force If a surface of area A is submerged horizontally at a depth h in a fluid, the weight
of the fluid exerts a force of
on the surface, where p is the fluid density (weight per unit volume). This is
called fluid force or hydrostatic force.
Table 6.4 shows the fluid density for some common fluids. It turns out that this
TABLE 6.4 Weight-density, p (lb/ft3)
fluid force does not depend on the shape of the container or its size. For instance, each
Water 62.4 of the containers in Figure 6.62 has the same pressure on its base because the fluid
64.0 depth h and the base area A are the same in each case.
Seawater
Gasoline 42.0
Kerosene 51.2
Milk 64.5
Mercury 849.0
Figure 6.62 The fluid force F - ph A does not depend on the shape or size of the container
When the surface is submerged vertically or at an angle, however, this simple formula
does not apply because different parts of the surface are at different depths, as shown in
Figure 6.63. We will use integration to compute the fluid force in such cases. Consider
a plate that is submerged vertically in a fluid of density p as shown in Figure 6.64a.
We set up a coordinate system with the horizontal axis on the surface of the fluid and
the positive h-axis (depth) pointing down. Thus, greater depths correspond to larger
values of h (see Figure 6.64b). For simplicity, we assume the plate is oriented so its
top and bottom are located a units and b units below the surface, respectively.
Pressure on a
Figure 6.63
submerged object
Our strategy will be to think of the plate as a pile of subplates or slabs, each so thin
that we may regard it as being at a constant depth below the surface. A typical slab
(a horizontal strip) is shown in Figure 6.64b. Note that it has thickness Ah and length
L — L(h), so that its area is A A = LAh. Because the slab is at a constant depth h
below the surface, the fluid force on its surface is
and by integrating as h varies from a to b, we can model the total force on the plate.
400 Chapter 6 Additional Applications of the Integral
F — f phL(h) dh
Ja
where h is the depth and L(h) is the corresponding length of a typical horizontal
approximating strip.
Solution
First, we set up a coordinate system in which the horizontal axis lies at the fluid surface
and the positive vertical axis (the h-axis) points down (see Figure 6.65b). Next, we find
expressions for the length and depth of a typical thin slab (horizontal strip) in terms
of the variables. We assume the slab has thickness Ah and length L. Then by similar
triangles (see Figure 6.65c) we have
L 3-h 2
— = —-— so that L = -(3 — h)
4 6 3
We see that the approximating slab has area A A = LAh = |(3 - h)Ah. Finally, we
multiply the product of the fluid’s weight-density at a given depth (ph) and the area of
the approximating slab (AA) and integrate on the interval of depths occupied by the
vertical plate.
— | (62.4) f
Jo
(3/j — h2) dh p = 62.4 lb/ft3 for water
Solution
Instead of putting the x-axis on the surface of the water (at the top of the dam), we place
the origin at the vertex of the parabola and the positive y-axis along the parabola’s axis
of symmetry (Figure 6.66b). The advantage of this choice of axes is that the parabola
can be represented by an equation of the form y = cx2. Because we know that y = 40
when x = 10, it follows that
40 = c(10)2 so that c = \ and thus y — \x2
The typical horizontal strip on the face of the dam is located y feet above the x-axis,
which means it is h = 40 - y feet below the surface of the water. The strip is Ay feet
wide and L = 2x feet long. Write L as a function of y:
y = |;c2 so that x = y
and
L(y) = 2x = 2 (/§7) = /lOy
Therefore,
A A = LAy = yiOy Ay
and the total force may be modeled by the integral
Depth below water: h = 40 — y
r40
-40 i
y rw
r-40 -
40
= 62.4A0 (y A2 - |r5/2)
0
40
= 62.4,/lb (A2) (f -
= 532,480 lb
402 Chapter 6 Additional Applications of the Integral
In mechanics, its often important to determine the point where an irregularly shaped
Centroid
plate will balance. The moment of a force measures its tendency to produce rotation
V
in an object and depends on the magnitude of the force and the point on the object
where it is applied. Since the time of Archimedes (287-212 B.C.), it has been known
that the balance point of an object occurs where all its moments cancel out (so there is
no rotation).
The mass of an object is a measure of its inertia; that is, its propensity to maintain
a state of rest or uniform motion. A thin plate whose material is distributed uniformly,
so that its density p (mass per unit area) is constant, is called a homogeneous lamina.
The balance point of such a lamina is called its centroid and may be thought of as its
geometrical center. We will see how centroids can be computed by integration in this
section, and will examine the topic even further in Section 12.6.
Consider a homogeneous lamina that covers a region R bounded by the curves
y = fix) and y = g(x) on the interval [a, b], and consider a thin, vertical approxi¬
mating strip within R, as shown in Figure 6.67. The mass of the strip is given by
Am = p[f(x) - g(x)]Ajc
Theoretically, the lamina should balance
t'--
on a point placed at its centroid. _ 1 . Area of strip
Density
m = pj [f(x)-g(x)]dx
Ja
Note that m = AREA if p = 1.
The centroid (geometrical center) of the approximating strip is (x, y), where x = x
and y = ~[f (jc) + g(x)]. The moment of the approximating strip about the y-axis
is defined to be the product
A My = x Am = x{p[f(x) — g(x)]Ax}
Mass of strip
Distance of the
strip from y-axis
Figure 6.67 Homogeneous lamina This product provides a measure of the tendency of the strip to rotate about the y-axis.
with a vertical approximating strip
Similarly, the moment of the strip about the x-axis is defined to be the product
= iP{[/U)]2-[g(*)]2} A*
This product provides a measure of the tendency of the strip to rotate about the x-axis.
Integrating on [a, b], we find the moments of the entire lamina R about the y-axis
and x-axis may be modeled by
If the entire mass m of the lamina R were located at the point (Jc, y), then its
moments about the x-axis and y-axis would be my and mx, respectively. Thus, we
have mx = My and my = Mx, so that
rb
x[fix) - gix)]dx M jP {[fix)]2-[gix)]2}dx
and y = —- =-^—r-
m m
[fix) - gix)]dx fix) - gix)]dx
A
J a
Let / and g be continuous and satisfy /(a) > g(x) on the interval b], and
consider a thin plate (lamina) of uniform density p that covers the region R
between the graphs of y = /(a) and y = g(x) on the interval [a, b]. Then
[ x[f(x)-g(x)]dx
x — —- — a
[
m
[/(*) - g(x)]dx
{[f(x)]2~[g(x)]2}dx
Mx
and y = m
[f(x) - g(x)]dx
Ja
Solution
We see that the line and the parabola intersect at the origin and at the point (1, 1), as
shown in Figure 6.68. Because p — 1, the mass is the same as the area of the region
R. That is.
1
m (a — A ) dx
6
and we find that the region has moments My and Mx about the y-axis and a-axis,
respectively, where
Figure 6.68 The centroid of a
planar region l 1
My a(a — A ) dx =
12
and
1
Mx = (a + a2)(a — a2) dx — a4) dx
2
1
2
X 2
-_ _ !L _ A - _ Mx_ 15
i
m \D 2 5
6
There is a remarkable result attributed to Pappus of Alexandria (ca. 300 A.D.), whom
many regard as the last great Greek geometer, which gives a connection between cen¬
troids and volumes of revolution.
404 Chapter 6 Additional Applications of the Integral
y‘ ‘ Proof First, choose a coordinate system in which the y-axis coincides with the axis of
-The approximating solid
revolution. Figure 6.69 shows a typical region R together with a vertical approximating
of revolution has yoltime
AV' = \kix(AA). rectangle. We will assume that this rectangle has area A A and that it is located x units
from the y-axis.
Notice that when the rectangle is revolved about the y-axis, it generates a shell of
volume AV = 2nxA A. Thus, by partitioning the region R into a number of rectangles
and taking the limit of the sum of the volumes of all the related approximating shells
as the norm of the partition approaches 0, we find that
V —
n
lim Y InxkA Ak = / 2rcxdA
n - r 1 A / x dA
-
f
= 2nj
2n / xdA
— 2ttx / dA Because x — ~-
_ SdA
Figure 6.69 The volume theorem
of Pappus = 2jtxA Because f dA = A
We close this section with an example that illustrates the use of the volume theorem
of Pappus.
Solution
The circle has area A — nr2 and its center (which is its centroid) travels 2n R units.
Comparing this example to the volume theorem of Pappus, we see a = 2nR. There¬
fore, the torus has volume
Incidentally, the volume found in Example 7 using the theorem of Pappus was worked
in Problem 60 of Section 6.2 by using washers.
6.5 Physical Applications: Work, Liquid Force, and Centroids 405
35. A hemispherical bowl with radius 10 ft is filled with water to b. Set up and analyze a model to determine the amount of
a level of 6 ft. Find the work done to the nearest ft-lb to pump work performed in moving one particle from an “infinite”
all the water to the top of the bowl. distance to a distance of 8 cm from the other. State any
36. A holding tank has the shape of a rectangular parallelepiped assumptions that you must make.
20 ft by 30 ft by 10 ft. 44. Modeling Problem Figure 6.71 shows a dam whose face against
a. How much work is done in pumping all the water to the the water is an inclined rectangle.
top of the tank? a. Set up and analyze a model for determining the fluid force
b. How much work is done in pumping all the water out of against the face of the dam when the water is level with the
the tank to a height of 2 ft above the top of the tank? top.
b. What would the fluid force be if the face of the dam were
a vertical rectangle (not inclined)?
37. A cylindrical tank of radius 3 ft and height 10 ft is filled to a 45. Modeling Problem Figure 6.72 shows a swimming pool, part of
depth of 2 ft with a liquid of density p — 40 lb/ft3. Find the whose bottom is an inclined plane. Set up and analyze a model
work done in pumping all the liquid to a height of 2 ft above for determining the fluid force on the bottom when the pool is
the top of the tank. filled to the top with water.
18 ft
41. An object weighing 800 lb on the surface of the earth is pro¬ 1.2 3.3
pelled to a height of 200 mi above the earth. How much work 1.4 3.6
is done against gravity? Hint: Assume the radius of the earth 1.6 3.7
is 4,000 mi, and use Newton’s law, F — —k/x2, for the force
1.8 3.7
on an object x miles from the center of the earth.
2.0 3.6
42. A dam is in the shape of an isosceles trapezoid 200 ft at the
top, 100 ft at the bottom, and 75 ft high. When water is up to 2.2 3.4
the top of the dam, what is the force on its face? 2.4 2.9
43. Modeling Problem According to Coulomb’s law in physics, two 2.6 0.0
similarly charged particles repel each other with a force in¬
versely proportional to the square of the distance between A region R with uniform density is said to have moments of inertia
them. Suppose the force is 12 dynes when they are 5 cm apart, Ix about the x-axis and Iv about the y-axis, where
The dA represents the area of an approximating strip. In Problems 52. Let R be the triangular region bounded by the line v = x, the
47-48, find Ix and ly for the given region. x-axis, and the vertical line x = r. When R is rotated about
the x-axis, it generates a cone of volume V = \nr3. Use Pap¬
47. the region bounded by the parabola y — y/x, the x-axis, and
pus’ theorem to determine y, the y-coordinate of the centroid
the line x = 4
of R. Then use similar reasoning to find x.
48. the region bounded by the curve y — vT — x3 and the coor¬
dinate axes 53. Find the volume of the solid figure generated when a square
49. Modeling Problem If a region has area A and moment of inertia of side L is revolved about a line that is outside the square,
Iy about the y-axis, then it is said to have a radius of gyration parallel to two of its sides, and located s units from the closer
side.
In Section 6.5, we saw how integration can be applied to the physical sciences. The
goal of this section is to explore applications to a variety of other areas, especially
business, economics, and the life sciences.
Recall from Chapter 2 that when P0 dollars are invested at the continuous compounding
annual rate r, then after t years, the account is worth A = Poen dollars. The amount
A is called the future value of the initial value Po dollars, but what is the future value if
the money were not deposited just once, in a lump sum Po, but continuously over the
entire time period? The difficulty is that at any given time, the “old" money deposited
earlier in the time period has been accumulating interest longer than “new” money
deposited more recently. To compute future value, in this case, our strategy will be
to approximate the continuous income flow by a sequence of discrete deposits called
an annuity. Then, the amount of the approximating annuity is a sum whose limit is a
definite integral that gives the future value of the income flow. The general procedure
is illustrated in Example 1.
Solution
To approximate the future value of the income flow, divide the 5-yr time period
0 < t < 5 into n intervals, each of length Ant = (5 — 0)/n, and for k — 1,2,.... n,
let tk-1 denote the beginning of the Ath subinterval [tk-\, tk]. The amount of money
Chapter 6 Additional Applications of the Integral
5-f/c i years
1500 a nt 1500 e07(5-'*-i)A„r
5
4 4-i 4 t
Figure 6.73 Approximate future value of money deposited during the jfcth subinterval
The future value of the entire income stream can be approximated by the sum of
the future values of the money deposited during all n time subintervals. Hence, we
have
n
[FUTURE VALUE OF INCOME FLOW] % ^ l,500ea07(5_f*-l)
*=1
As n increases without bound, the length of each subinterval tends toward zero, and
there is less and less error in our assumption that all the money in the jfcth subinterval
is deposited at time t = 4-i • Therefore, it is reasonable to compute the total future
value of the income flow by taking the limit of the approximating sum as n —> +oo.
We recognize this sum as a Riemann sum, and the limit as the definite integral. That is
n
[FUTURE VALUE OF INCOME FLOW] = lim 1 500e>°07(5_4-i) ^ t
n->+oo^ ' n
k=l
=f l,500e°-07(5-f) dt
Jo
= l,500ea35 I* e-°07,dt
Jo
= l,500e°'35
= 8,980.02
Recall from Chapter 2, that the present value, PV, of A dollars invested at the
annual interest rate r compounded continuously for a term T years is the amount of
money PV = Ae " that must be deposited today at the prevailing interest rate to
generate A dollars at the end of the term. Likewise, the present value of an income
stream /(t) generated continuously at a certain rate over a specified term is the amount
of money that would have to be deposited at the beginning of the term at the prevailing
interest rate to generate the income stream over the entire term. As with future value,
the strategy is to approximate the continuous income stream by a sequence of discrete
payments; that is, by an annuity. The present value of the approximating annuity is a
sum, and by taking the limit of that sum, we obtain an integral
[ f(t)e-rldt
Jo
that represents the present value of the entire income flow. The details of this approx¬
imation scheme are outlined in the problem set. Here is a summary of the formulas to
be used for computing future and present value of an income flow.
6.6 Applications to Business, Economics, and Life Sciences 409
Future Value and Present Value Let fit) be the amount of money deposited at time t over the time period [0, T]
of an Income Flow in an account that earns interest at the annual rate r compounded continuously.
Then the future value of the income flow over the time period is given by
FV=f f{t)er,dt
Jo
and the present value of the same income flow over the time period is
In certain applications, we are given the rate of change Q'(x) of a quantity Q(x) and
are required to find the net change Q(b) — Q(a) in Q(x) as jc varies from x = a to
x = b. But because Q(x) is an antiderivative of Q'{x), the fundamental theorem of
calculus tells us that the net change is given by the definite integral
Qib)-Q(a)=[ Q\x)dx
Since the definite integral on the right “adds” all the instantaneous changes in Q{x)
over the time period [a, b], the integral can also be thought of as cumulative change.
For example, if N(t) is the population of a certain species that is growing at the rate
N\t) at time t, then the cumulative change in population over the time period [ti, tf\
is given by the integral
y
A business application involves the net earnings of an industrial process. Suppose
dx) represents the total cost of producing x units and Rix) represents the revenue
generated by the sale of those units. Then Eix) — Rix) - Cix) represents earnings,
/) =
and if the marginal cost C'ix) and marginal revenue R'ix) are known, the net earnings
for the production range a < x < b are given by the definite integral
Net earnings can be interpreted geometrically as an area between the marginal rev¬
Figure 6.74 Net earnings as an area enue and marginal cost curves as shown in Figure 6.74. These ideas are illustrated in
Example 2.
respectively. What are the net earnings of the process as * ranges from x = 0 to
x = xm, where xm is the level of production at which marginal cost equals marginal
revenue?
410 Chapter 6 Additional Applications of the Integral
Solution
First, note that marginal cost equals marginal revenue when
O.lx2 + 4x + 10 = 70 — x
O.lx2+ 5x -60 = 0
x2 + 5Ox - 600 = 0
(x + 60)(x — 10) = 0
x = 10, -60
Reject the negative value, giving the solution xm = 10, so that the net earnings as x
ranges from x = 0 to x = 10 are given by
n 1U
£(10)-£(0) = / [R\x)-C\x)]dx
Jo
/•to
= / [70 — x — (0. lx2 + 4x + 10)] dx
Jo
950
= / [—O.lx2 — 5x + 60] dx = —= 316.6
Jo
Figure 6.75 Net earnings for the That is, the net earnings are $316,667, and are shown as the area between the marginal
marginal cost and marginal revenue cost and marginal revenue curves in Figure 6.75. ■
In a competitive economy, the total amount that consumers actually spend on a com¬
modity is usually less than the total amount they would have been willing to spend.
The difference between the two amounts can be thought of as a savings realized by
consumers and is known in economics as the consumer’s surplus.
To get a better feel for the concept of consumer’s surplus, consider an example
of a couple who are willing to spend $500 for their first TV set, $300 for a second
set, and only $50 for a third set. If the market price is $300, then the couple would
buy only two sets and would spend a total of 2 x $300 = $600. This is less than
the $500 + $300 = $800 that the couple would have been willing to spend to get
the two sets. The savings of $800 - $600 = $200 is the couple’s consumer’s surplus.
Consumer’s surplus has a simple geometric interpretation, which is illustrated in Figure
6.76.
Note that p and q denote the market price and the corresponding demand, respec¬
tively. Figure 6.76a shows the region under the demand curve p = D(q) from q = 0
to q = qo, and the area of this region represents the total amount that consumers are
willing to spend to get q0 units of the commodity. The rectangle in Figure 6.76b has
an area of poq0 and represents the actual consumer expenditure for q0 units at p0 dol¬
lars per unit. The difference between these two areas (Figure 6.76c) represents the
6.6 Applications to Business, Economics, and Life Sciences 411
consumer’s surplus. That is, consumer’s surplus is the area of the region between the
demand curve p = D(q) and the horizontal line p = p0 so that
M0 MO MO MO
/ [D(q)-p0]dq= / D(q) dq — I p0dq = / D(q) dq - p0q0
Jo Jo Jo Jo
Consumer's Surplus
If qo units of a commodity are sold at a price of po dollars per unit, and if
p = D(q) is the consumer’s demand function for the commodity, then
MO
CS - / D(q)dq-p0qo
'o
Producer’s surplus is the other side of the coin from consumer’s surplus. In par¬
ticular, a supply function p = S(q) gives the price per unit that producers are willing
to accept in order to supply q units to the marketplace. However, any producer who is
willing to accept less than po = S(qo) dollars for qo units gains from the fact that the
price is actually po- Then producer’s surplus is the difference between what produc¬
ers would be willing to accept for supplying q units and the price they actually receive.
Reasoning as we did with consumer’s surplus, we obtain the following formula for
producer’s surplus.
Producer's Surplus
If qo units of a commodity are sold at a price of po dollars per unit, and if
p = S(q) is the producer’s supply function for the commodity, then
TOTAL AMOUNT
ACTUAL CONSUMER
PRODUCER’S PRODUCERS RECEIVE
EXPENDITURE FOR
SURPLUS WHEN qo UNITS
q0 UNITS
ARE SUPPLIED
PS = poqo - S{q) dq
p = D{q) = -O.I42 + 90
dollars/tire, and the same number of tires will be supplied when the price is
p — S(q) — 0.2 q2 + q + 50
dollars/tire.
a. Find the equilibrium price (that is, where supply equals demand) and the quantity
supplied and demanded at that price.
b. Determine the consumer’s and producer’s surplus at the equilibrium price.
412 Chapter 6 Additional Applications of the Integral
Solution
a. The supply and demand curves are shown in Figure 6.77. Supply equals demand
when
0.3q2 + q -40 = 0
q = 10 (reject q = — y)
1,850
—
= (80)(10)
- 183.3
Since q is in thousands, the consumer’s surplus is about $66,667, and the producer’s
surplus is $183,333. The consumer’s surplus and producer’s surplus are the regions
labeled CS and PS, respectively, in Figure 6.77. ■
There is an important category of problems in which a survival function gives the frac¬
tion of individuals in a group of population that can be expected to remain in the group
and a renewal function gives the rate at which new members can be expected to join
the group. Such survival and renewal problems arise in many fields, including soci¬
ology, ecology, demography, and even finance, where the “population” is the number
of dollars in an investment account and “survival and renewal” refers to results of an
investment strategy. The following example illustrates the general procedure for mod¬
eling with survival and renewal functions.
Solution
Since /(15) is the fraction of patients whose treatment continues at least 15 months, it
follows that ot the current 300 patients, only 300/(15) will still be receiving treatment
15 months from now.
To approximate the number of new patients who will be receiving treatment 15
months from now, divide the 15-month time interval 0 < t < 15 into n equal subinter¬
vals of length At months and let /■_ j denote the beginning of the jth subinterval. Since
6.6 Applications to Business, Economics, and Life Sciences 413
new patients are accepted at the rate of 10/mo, the number of new patients accepted
during the y'th subinterval is lOAf. Fifteen months from now, approximately 15 - /_i
months will have elapsed since these 10At new patients had their initial visits, and so
approximately (lOAt) /(15-/_i) of them will still be receiving treatment at that time
(see Figure 6.78).
15 -1:_i months
10 At ->—- 10Ac/(15 - tj)
0 At 15
+ t
to 7-t
It follows that the total number of new patients still receiving treatment 15 months
from now can be approximated by the sum
n
10/(15 -tj-OAt
j=1
Adding this to the number of current patients who will still be receiving treatment in
15 months, we obtain
n
P % 300/(15) + 10/(15 - tj-i)At
y'=i
where P is the total number of patients who will be receiving treatment 15 months
from now. As n increases without bound, the approximation improves and approaches
the true value of P. That is,
= 300/(15)+ [ 10/(15 — t) dt
Jo
Since f(t) — e~^20, we have /(15) = e~3^4 and /(15 — f) = e-O5-r)/20 _ e-3/4gc/20
Thus,
That is, 15 months from now, the clinic will be treating approximately 247 patients. ■
In biology, Poiseuille’s law of flow* asserts that the speed of blood in an artery is a
function of the distance of the blood from the artery’s central axis. That is, the speed
(in cm/s) of blood that is r centimeters from the central axis of the artery may be
modeled by
S(r) = k(R2 — r2)
where R is the radius of the artery and k is a constant. In the following example, you
will see how to use this information to compute the rate (in cm3/s) at which blood
flows through the artery.
*The law of flow is named for the same French physician, Jean Louis Poiseuille (1799-1869), responsible
for the resistance to flow law cited in Section 4.7.
414 Chapter 6 Additional Applications of the Integral
Solution
To find the approximate volume of blood that flows through a cross section of the artery
each second, divide the interval 0 < r < R into n equal subintervals of width Ar cm
and let rj denote the beginning of the y'th subinterval. These subintervals determine n
concentric rings, as illustrated in Figure 6.79.
If Ar is small, the area of the yth ring is approximately the area of a rectangle
whose length is the circumference of the (inner) boundary of the ring and whose width
is Ar. That is,
Figure 6.79 Subdividing a
cross section of an artery AREA OF THE y TH RING « 2nrjAr
= 27tk(R2rj — r])Ar
J J
The volume rate of flow of blood through the entire cross section is the sum of n such
terms, one for each of the n concentric rings. That is,
27Tk(R2r — r3) dr
o
nkR4 ,
—-— in cm" /s
In Problems 8-11, p = D(q) is the price (in dollars/unit) at which 23. Suppose an industrial machine that is x years old generates
q units of a particular commodity will be demanded by the market revenue at the rate of
(that is, all q units will be sold at this price), and q0 is a speci¬
fied level of production. In each case, find the price p0 = D(q0) R'(x) = 6,025 - lO.v2
at which q0 units will be demanded and compute the correspond¬ dollars per year and costs accumulate at the rate of
ing consumer’s surplus. Sketch the demand curi’e y = D(q) and
shade the region whose area represents the consumer’s surplus. C'(jc) = 4,000+ 15x2
8. D(q) = 2(64 — q2)\ q0 = 3 units dollars per year for operation and maintenance.
9. D(q) = 150 — 2q — 3q2\ q0 = 6 units a. The machine will be sold when earnings
E(x) = R(x) — C(x) are maximized. When does this oc¬
10. D(q) = 40f'~ao5‘?; q0 = 5 units
cur?
11. D(q) = 75e~OMq\ q0 = 3 units
b. What are the net earnings generated by the machine be¬
In Problems 12-15, p = S{q) is the price (in dollars/unit) at which tween now and when it is sold? Interpret this amount as
q units of a particular commodity will be supplied to the market by the area between two curves.
producers, and q0 is a specified level of production. In each case, 24. After t hours on the job, Joan is producing units at the rate of
find the price po = S(qo) at which q0 units will be supplied and
compute the corresponding producer’s surplus. Sketch the supply Q\(t) = 60 — 2(/ — l)2
curx’e y = S(q) and shade the region whose area represents the units per hour, and Jill is producing at the rate of
producer’s surplus.
Q’2(t) = 50-51
12. S(q) = 0.3q2 + 30; <70 = 4 units
13. S(q) — 0.5q + 15; qo = 5 units units per hour. If both arrive on the job at 8:00 A.M., how
many more units (to the nearest unit) will Joan have produced
14. S(q) — 10 + 15e0 039; qo = 3 units
by noon than Jill? Interpret your answer as the area between
15. S(q) = 17 + lie00'9; qo = 1 units two curves.
^ In Problems 16-19, find the consumer’s surplus at the point of 25. Suppose that x years from now, one investment plan will be
market equilibrium (the level of production where supply S(q) generating profit at the rate of
equals demand D(q)).
P[(x) = 100 + x2
Demand function Supply function
dollars per year, whereas a second plan will be generating
16. D(q) = 14 -q2 S(q) = 2 q2 + 2
profit at the rate of
17. D(q) = 25 — q2 S(<7) = 5<72+1
P2(x) = 190 + 2x
18. D(q) = 32 - 2q2 S(q) = \q2 + 2q +5
dollars per year. Neither plan generates a profit in the begin¬
19. D(q) = 27 - q2 S(q) = W + \<l + 5
ning; that is, FilO) = /MO) = 0.
20. A manufacturer has determined that when q units of a partic¬
a. For how many years will the second plan be more prof¬
ular commodity are produced, the price at which all the units
itable?
can be sold is p = D(q) dollars per unit, where D is the de¬
b. How much more total profit will you earn if you invest in
mand function given by
the second plan instead of the first for the period of time in
300 part a? Interpret the extra profit as the area between two
D(q) = curves.
(0.1c/ + I)2
26. Parts for a piece of heavy machinery are sold in units of 1,000.
The demand for the parts (in dollars) is given by
a. How many units can the manufacturer expect to sell if the p(x) = 110 — x. The total cost is given by
commodity is priced at po — $ 12/unit? C(x) = x2 — 25x2 + 2x + 30 (dollars).
b. Find the consumer’s surplus that corresponds to the level a. For what value of x is the profit
of production qo found in part a.
P(x) — xp(x) — C(x)
21. Money is transferred continuously into an account at the con¬
stant rate of $3,000/yr. The account earns interest at the an¬ maximized?
nual rate of 8% compounded continuously. b. Find the consumer’s surplus with respect to the price that
a. What is the future value of the income flow in three years? corresponds to maximum profit.
27. Repeat Problem 26 with p(x) = 124 — 2x and
b. What is the present value of the income flow over a three-
C(x) = 2x3 — 59x2 + 4x + 76.
year term?
28. Suppose when q units of a commodity are produced, the de¬
22. The management of a small firm expects to generate income
mand is p(q) = 45 — q2 dollars per unit, and the marginal cost
at the rate of f(t) — 2,000e°03' dollars/yr. The prevailing rate
is
of interest is 7% compounded continuously. dC_
= 6 +W
a. What is the future value for the firm over a five-year term? dq
b. What is the present value for the firm over the same five- Assume there is no overhead (so C(0) = 0).
year term? a. Find the total revenue and the marginal revenue.
416 Chapter 6 Additional Applications of the Integral
b. Find the value of q (to the nearest unit) that maximizes b. Compute the net excess profit assuming that you invest in
profit. the second plan for the time period determined in part a.
c. Find the consumer’s surplus at the value of q where profit c. Sketch the rate of profitability curves y — P[{t) and
is maximized. (Use the exact value of q.) y — Pn(t) and shade the region whose area represents the
29. Repeat Problem 28 with net excess profit computed in part b.
40. Answer the questions in Problem 39 for two investments with
p(q) = \{\0 - q)1 and ~ = \q2 + 5 respective rates of profitability P[(t) — 130 +12 hundred dol¬
dq 4 lars per year and P2(t) = 306 + 5/ hundred dollars per year.
41. Answer the questions in Problem 39 for two investments with
30. A company plans to hire additional personnel. Suppose it
respective rates of profitability P,'(r) = 60<?012' thousand dol¬
is estimated that as x new people are hired, it will cost
lars per year and P2(t) = 160<?0 08' thousand dollars per year.
C(x) = 0.2x thousand dollars and that these x people will
42. Answer the questions in Problem 39 for two investments with
bring in R(x) — -fix thousand dollars in additional revenue.
respective rates of profitability P[{t) = 90<?° lr thousand dol¬
How many new people would be hired to maximize the profit?
lars per year and P2{t) = 140c0 07' thousand dollars per year.
How much net revenue would the company gain by hiring
these people? 43. Suppose that when it is / years old, a particular industrial ma¬
chine generates revenue at the rate R'(t) = 6,025 — 812 dol¬
31. Suppose that the demand function for a certain commodity is
lars/yr and that operating and servicing costs accumulate at
D(q) = 20—4<72, and that the marginal cost is C'(q) = 2q+6,
the rate of C'(t) = 4,681 + 13/2 dollars/yr.
where q is the number of units produced. Find the consumer’s
surplus at the sales level q0 where profit is maximized. a. How many years pass before the profitability of the ma¬
chine begins to decline?
32. The resale value of a certain industrial machine decreases over
a ten-year period at a rate that changes with time. When the b. Compute the net earnings generated by the machine over
machine is x years old, its value is decreasing at the rate of the time period determined in part a.
220(x - 10) dollars/yr. By how much does the machine de¬ c. Sketch the revenue rate curve y = R'(t) and the cost rate
preciate during the second year? curve y = C'(t) and shade the region whose area repre¬
sents the net earnings computed in part b.
33. The promoters of a country fair estimate that / hours after the
gates open at 9:00 A.M. visitors will be entering the fair at the 44. Answer the questions in Problem 43 for a machine that gen¬
rate of —4(/ + 2)3 + 54(/ + 2)2 people/hr. How many people erates revenue at the rate /?'(/) = 7,250 — 18/2 dollars/yr and
will enter the fair between 10:00 A.M. and noon? for which costs accumulate at the rate C'(t) — 3,620 + 12/2
dollars/yr.
34. At a certain factory, the marginal cost is 6(q — 5)2 dollars/unit
when the level of production is q units. By how much will the 45. After / hours on the job, one factory worker is producing
total manufacturing cost increase if the level of production is Q\(t) = 60 — 2(/ — l)2 units/hr, while a second worker
raised from 10 to 13 units? is producing Q2(t) = 50-5/ units/hr.
35. A certain oil well that yields 400 barrels of crude oil a month a. If both arrive on the job at 8:00 A.M., how many more units
will run dry in two years. The price of crude oil is currently will the first worker have produced by noon than the sec¬
$ 18/barrel and is expected to rise at a constant rate of $0.03 ond worker?
per barrel per month. If the oil is sold as soon as it is extracted b. Interpret the answer in part a as the area between two
from the ground, what will be the total future revenue from curves.
the well? 46. A national consumers’ association has compiled statistics sug¬
36. It is estimated that / days from now a farmer’s crop (in gesting that the fraction of its members who are still active /
bushels) will be increasing at the rate of 0.3/2 + 0.6/ + 1 months after joining is given by the function /(/) = e~0-2'. A
bu/day. By how much will the value of the crop increase new local chapter has 200 charter members and expects to at¬
during the next five days if the market price remains fixed at tract new members at the rate of 10/mo. How many members
$3/bu? can the chapter expect to have at the end of 8 months?
37. It is estimated that the demand for a manufacturer’s product 47. The population density r miles from the center of a certain
is increased exponentially at the rate of 2%/yr. If the cur¬ city is D(r) = 25,OOOe~005r‘ people/mi2. How many people
rent demand is 5,000 units/yr and if the price remains fixed at live between 1 and 2 miles from the city center?
$400/unit, how much revenue will the manufacturer receive 48. The population density r miles from the center of a certain
from the sale of the product over the next two years? city is D(r) = 5,000c“°lr‘ people/mi2. How many people
38. After / hours on the job, a factory worker can produce live within 3 miles from the city center?
100(2 + sin/) units/hr. How many units does a worker who 49. Calculate the volume rate (in cm3/s) at which blood flows
arrives on the job at 8:00 A.M. produce between 10:00 A.M. through an artery of radius 0.1 cm if the speed of the blood
and noon? r cm from the central axis is 8(1 — 100r2) cm/s.
39. Suppose that / years from now, one investment plan will be 50. Modeling Problem Blood flows through an artery of radius R.
generating profit at the rate of /»,'(/) = 100 + /2 hundred dol¬ At a distance r cm from the central axis of the artery, the speed
lars per year, while a second investment will be generating of the blood is given by S(r) = k(R2 - r2). Show that the av¬
profit at the rate of P[{t) = 220 + 2/ hundred dollars per year. erage speed of the blood is one-half the maximum speed.
a. For how many years does the rate of profitability of the 51. A study indicates that x months from now the population of
second investment exceed that of the first? a certain town will be increasing at the rate of 5 4- 3x2/3
Chapter 6 Review 417
people/mo. By how much will the population of the town members will the service have ten months from now? Hint:
increase over the next 8 months? Divide the interval 0 < t < 10 into n equal subintervals and
52. An object is moving so that its speed after t minutes is form a Riemann sum.
5 + 21 + 312 m/min. How far does the object travel during 58. Modeling Problem A certain nuclear power plant produces ra¬
the second minute? dioactive waste in the form of strontium-90 at the constant
53. It is estimated that the demand for oil is increasing exponen¬ rate of 500 pounds/yr. The waste decays exponentially with a
tially at the rate of 10%/yr. If the demand for oil is currently half-life of 28 years. How much of the radioactive waste from
30 billion barrels per year, how much oil will be consumed the nuclear plant will be present after 140 years. Hint: Think
during the next 10 years? of this as a survival and renewal problem and form a Riemann
54. It is estimated that t weeks from now, contributions in re¬ sum.
sponse to a fund-raising campaign will be coming in at the rate 59. Modeling Problem A certain investment generates income con¬
of R {t) = 5,OOOe-0 2' dollars/wk, while campaign expenses tinuously over a period of N years. After / years, the invest¬
are expected to accumulate at the constant rate of $676/wk. ment will be generating income at the rate of f(t) dollars/yr.
a. For how many weeks does the rate of revenue exceed the The prevailing annual rate of interest, compounded continu¬
rate of cost? ously, is r. Derive the formula
b. What net earnings will be generated by the campaign dur¬
ing the period of time determined in part a.
c. Interpret the net earnings in part b as an area between two
curves.
for the present value of the income flow by completing these
55. Spy Problem The Spy arrives at the sleazy bar at the appointed
steps.
time for his meeting with his superior (recall Supplementary
Problem 85, Chapter 5), but, as usual, his peerless leader is a. Subdivide the time interval [0, N] into n equal subinter¬
fashionably late. While waiting, the Spy reflects on the dan¬ vals, the &th of which is [4_i, tk] for k = 1,2
gerous life he leads. Ten years ago he took his spying oath, fol¬ Assuming that the income generated during the entire kth
lowed on the same day by 6 friends from college. On average, subinterval is /(4), what is the present value of this in¬
30 new people join each year, and he estimates that the frac¬
come?
tion of spies still alive after t years of service is F{t) = e“,/20. b. Approximate the present value of the total income flow by
Approximately how many people who joined the service after an appropriate sum.
the Spy are still alive? c. Note that the sum in part b is a Riemann sum. Make the ap¬
56. The administrators of a town estimate that the fraction of peo¬ proximation precise by taking a limit to obtain the required
ple who will still be residing in the town t years after they integral formula for present value.
arrive is given by the function fit) = e~0 04'. If the current 60. Modeling Problem A manufacturer receives N units of a certain
population is 20,000 people and new townspeople arrive at raw material that are initially placed in storage and then with¬
the rate of 500/yr, what will the population be ten years from drawn and used at a constant rate until the supply is exhausted
now? Hint: Divide the interval 0 < t < 10 into n equal subin¬ one year later. Suppose storage costs remain fixed at p dollars
tervals and form a Riemann sum. per unit per year. Use definite integration to find an expression
57. The operators of a new computer dating service estimate that for the total storage costs the manufacturer will pay during the
the fraction of people who will retain their membership in year. Hint: Let Q(t) denote the number of units in storage
the service for at least t months is given by the function after t years and find an expression for Q{t). Then subdivide
fit) = There are 8,000 charter members, and the op¬ the interval 0 < t < 1 into n equal subintervals and express
erators expect to attract 200 new members/mo. How many the total storage cost as the limit of a sum.
12. How do you find the force exerted by a liquid on an object 26. Find the arc length of the curve y — 2 — x3/2 from (0, 2) to
submerged vertically? (1, 1). Give the exact value.
13. What are the formulas for the mass and centroid of a homoge¬ 27. Find the surface area obtained by revolving the parabolic arc
neous lamina? y2 — x from (0, 0) to (1, 1) about the x-axis. Give the answer
14. State the volume theorem of Pappus. to the nearest hundredth unit.
15. What formulas are used to compute future value and present 28. Find the centroid of the homogeneous region between
value of an income flow? y — x2 — x and y = x — x3 on [0, 1],
16. How do you find the net change of a function / whose deriva¬ 29. An observation porthole on a ship is circular with diameter of
tive /' is given? 2 feet and is located so its center is 3 feet below the waterline.
17. Describe the process for finding consumer’s and producer’s Find the fluid force on the porthole, assuming that the boat is
surplus. in seawater (density is 64 lb/ft3). Set up the integral only.
18. Describe a typical survival and renewal problem. 30. Money is transferred continuously into an account at the con¬
stant rate of $l,200/yr. The account earns interest at the an¬
19. What formula is used for computing the flow of blood through
nual rate of 8% compounded continuously. How much money
an artery?
will be in the account at the end of 5 years?
PRACTICE PROBLEMS
20. Journal Problem Mathematics Teacher* ■ Use the integrals Supplementary Problems
A-G to answer the given questions. Name and sketch the curves whose equations are given in Prob¬
/f*ob rd
pa
lems 1-8.
[f(x)]2dx B. n j^ [f(y)]2dy
1. r — 2 sin# — 3 cos# 2. r = -29, 9 > 0
3. r — — esc 9 4. r — sin 3#
A(x) dx My) dy
c-/ *■[
"5
5.
o
o
6. r2 = 3 cos 29
cr>
II
i
E. 7T f [ fix) - g(x)]dx
4
7. r = -
5
8. r — -
Ja
Find the points of intersection of the polar-form curves in Prob¬
F. Tt Jf
a
{[f(x)]2~[g(x)]2}dx lems 9-12.
\r = 2 cos 6 r = 1 + sin 6
{[f(y)?-[g{y)]2}dy 9. 10 .
°-n[ I r — 1 + cos 6 r — 1 + cos#
a. Which formula does not seem to belong to this list?
r cos 6 + 2r sin 0 = 4 r — 2 sin 2d
b. Which expressions represent volumes of solids of revolu¬ 11. r — 2 sec 9
12.
r = 1 for 0 < 6 < |
tion?
c. Which expressions represent volumes of solids containing Find the area of the regions bounded by the curves and lines in
holes? Problems 13-21.
d. Which expressions represent volumes of a solid with cross 13. 4y2 = x and 2y = x — 2 14. y = x3, y = x4
sections that are perpendicular to an axis? 15. x = y2/3, x — y2
e. Which expressions represent volumes of solids of revolu¬ 16. y = \/3sinx, y = cosx, and the vertical lines x = 0 and
tion revolved about the x-axis? X = Tt/2
f. Which expressions represent volumes of solids of revolu¬ 17. the region bounded by the polar curve r — 4 cos 29
tion revolved about the y-axis? 18. the region inside both the circles r — 1 and r — 2 sin 9
21. Find the area above the curve /(x) = 3x2 - 6 and below the
19. the region inside one loop of the lemniscate r2 = cos 29
x-axis.
20. the region inside one petal of the rose curve r = sin 29
22. Find the area between the curves y = x2 and y3 — x.
21. the region inside the circle r = 2a sin 9 and outside the circle
23. The base of a solid in the xy-plane is the circular region given
r —a
by x2 + y2 — 4. Every cross section perpendicular to the
x-axis is a rectangle whose height is twice the length of the
Find the volume of a solid of revolution obtained by revolving the
side that lies in the xy-plane. Find the volume of this solid.
region R about the indicated axis in Problems 22-25.
24. Let R be a region bounded by the graphs of the equations 22. R is bounded by y = y/x, x = 9, y = 0; about the x-axis
y = (x — 2)2 and y = 4. 23. R is bounded by y = x2, y = 9 - x2, and x = 0; about the
a. Sketch the graph of R and find its area. y-axis
b. Find the volume of a solid of revolution of R about the 24. R is bounded by y = x2 and y = x4; about the x-axis
x-axis. 25. R is bounded by y = y/cosx, x = x = and y = 0;
c. Find the volume of a solid of revolution of R about the about the x-axis.
y-axis. 26. The base of a solid is the region in the xy-plane that is bounded
25. Find the area of the intersection of the circles r = 2a cos 6 by the curve y = 4 — x2 and the x-axis. Every cross section
and r = 2a sin #, where a > 0 is constant. of the solid perpendicular to the y-axis is a rectangle whose
height is twice the length of the side that lies in the xy-plane.
*Vol. 83, No. 9, December 1990, p. 695. Find the volume of the solid.
Chapter 6 Review 419
29. A force of 100 lb is required to compress a spring from its nat¬ 39. A container has the shape of the solid formed by rotating the
ural length of 10 in. to a length of 8 in. Find the work required region bounded by the curve y = x3 and the lines x = 0 and
to compress it to a length of 7 in. from its natural length. y = 1 about the y-axis, where x and y are measured in feet. If
30. A bowl is formed by revolving the region bounded by the the container is filled with water, how much work is required
curve y — lx2 and the lines x = 0 and y = 1 about the y-axis to pump all the water out of the container?
(units are in ft). The bowl is filled with water to a height of 9 40. A gate in a water main has the shape of a circle with diameter
in. 10 ft. If water stands 2 ft high in the main, what is the force
exerted by the water on the gate?
a. How much work is done in pumping the water to the top of
the bowl? Suppose a particle of mass m moves along a number line under
the influence of a variable force F which always acts in a direc¬
b. How much work is done in pumping the water to a point
tion parallel to the line of motion. At a given time the particle is
0.5 ft above the top of the bowl?
said to have kinetic energy
31. Two solids S\ and S2 have as their base the region in the xy-
plane bounded by the parabola y = x2 and the line y = 1. K — \mv2
For solid Si, each cross section perpendicular to the x-axis is
a square, and for S2, each cross section perpendicular to the where v is the particle’s velocity, and it has potential energy
y-axis is a square. Which solid has the greater volume?
P — — f F dx
32. Use calculus to find the lateral surface area of a right circular Js0
cylinder with radius r and height h.
relative to the point s on the line of motion. Use these definitions
33. Use calculus to find the surface area of a right circular cone
in Problems 41^42.
with (top) radius r and height h.
41. A spring at rest has potential energy P = 0. If the spring con¬
34. A 20-lb bucket in a well is attached to the end of a 60-ft chain stant is 30 lb/ft, how far must the spring be stretched so that
that weighs 16 lb. If the bucket is filled with 50 lb of water, its potential energy is 20 ft-lb?
how much work is done in lifting the bucket and chain to the
42. A particle with mass m falls freely near the earth’s surface. At
top of the well?
time t — 0, the particle is s0 units above the ground and is
35. A swimming pool is 3 ft deep at one end and 8 ft deep at the falling with velocity i>o.
other, as shown in Figure 6.80. The pool is 30 ft long and 25 ft a. Find the potential energy P of the particle in terms of po¬
wide with vertical sides. Set up and analyze a model to find sition, s, if the potential energy is 0 when s = s0.
the fluid force against one of the 30-ft sides.
b. Find the potential energy P and the kinetic energy K of the
particle in terms of time t.
30 ft
43. An object weighs w — 1,000r"2 grams, where r is the object’s
distance from the center of the earth. What work is required
to lift the object from the surface of the earth (r — 6,400 km)
to a point 3,600 km above the surface of the earth?
44. Modeling Problem A certain species has population P(t) at
time t whose growth rate is affected by seasonal variations
Figure 6.80 Cross section of a swimming pool in the food supply. Suppose the population may be modeled
by the differential equation
36. The ends of a container filled with water have the shape of the dP ,
region R bounded by the curve y = x4 and the line y = 16. —— = k(P — A) cos t
dt
Find the fluid force (to two decimal places) exerted by the wa¬
ter on one end of the container. where k and A are physical constants. Solve the differential
equation and express P{t) in terms of k, A, and PQ, the initial
37. A conical tank whose vertex points down is 12 ft high with
population.
(top) diameter 6 ft. The tank is filled with a fluid of density
p = 22 lb/ft3. If the tank is filled to a depth of 6 ft, find the 45. Find the center of mass of a thin plate of constant density if
work required to pump all the fluid to a height of 2 ft above the plate occupies the region bounded by the lines 2y = x,
x + 3y = 5, and the x-axis.
the top of the tank.
46. The portion of the ellipse (Figure 6.82)
38. Each end of a container filled with water has the shape of an
isosceles triangle, as shown in Figure 6.81. Find the force ex¬
erted by the water on an end of the container.
420 Chapter 6 Review
that lies in the first quadrant between x — r and a = a, for 54. A bag of sand is being lifted vertically upward at the rate of
0 < /• < a, is revolved about the A-axis to form a solid S. 31 ft/min. If the bag originally weighed 40 lb and leaks at the
Find the volume of S. rate of 0.2 lb/s, how much work is done in raising it until it is
empty?
56. The radius of a circular water main is 2 ft. Assuming the main
is half full of water, find the total force exerted by the water
on a gate that crosses the main at one end (see Figure 6.83).
■*
that lies in the first quadrant. Use the theorem of Pappus Figure 6.84 Problem 59
to find the coordinates of the centroid of R. You may use
the facts that R has area A = |nab, and the semiellipsoid If the tank is filled with a fluid of density p — 24 lb/ft3, how
it generates when it is revolved about the A-axis has volume much work is done in bringing all the fluid to the top of the
V = \nab2. tank?
53. Modeling Problem A container of a certain fluid is raised from 60. Modeling Problem A piston in a cylinder causes a gas either to
the ground on a pulley. While resting on the ground, the con¬ expand or to contract. Assume the pressure P and the volume
tainer and fluid together weigh 200 lb. However, as the con¬ of gas V in the cylinder satisfy the adiabatic law (no exchange
tainer is raised, fluid leaks out in such a way that at height a of heat)
feet above the ground, the total weight of the container and
P(VlA) = C
fluid is 200 — 0.5a lb. Set up and analyze a model to find the
work done in raising the container 100 ft. where C is a constant.
Chapter 6 Review 421
A
a. If the gas goes from volume V\ to volume V2, show that the
D
work done may be modeled by
I
, B
y -- Ax3 H-
X
69. Modeling Problem An object below the surface of the earth is 74. Find the centroid of a right triangle whose legs have lengths a
attracted by a gravitational force that is directly proportional and b. Hint'. Use the volume theorem of Pappus.
to the square of the distance from the object to the center of
75. The largest of the pyramids at Giza in Egypt is roughly 480 ft
the earth. Set up and analyze a model to find the work done
high and has a square base 750 ft on a side. How much work
in lifting an object weighing P lb from a depth of s feet (be¬
was required to lift the stones to build this pyramid if we as¬
low the surface) to the surface. (Assume the earth’s radius is
sume that the rock in the pyramid had an average density of
4,000 mi.)
160 lb/ft3?
In Problems 70-71, consider a continuous curve y = / (x) that
passes through (-1,4), (0, 3), (1,4), (2, 6), (3, 5), (4, 6), (5, 7), 76. Find the length of the curve y = ex on the interval [0, 1],
(6, 6), (7, 4), (8, 3) and (9, 4). 77. Find the volume of the solid generated when the region under
70. Use Simpson’s rule to estimate the volume of the solid formed the curve
by revolving this curve about the x-axis for [—1, 9], Assume
the curve passes through (9, 4).
71. Use the trapezoidal rule to estimate the volume of the solid 'V “ V9^2
' formed by revolving this curve about the line y = —1 on
[-1,9]. on the interval [0, 2] is revolved about the y-axis.
72. Modeling Problem In this computational problem we revisit a 78. Find the volume of the solid formed by revolving about the
problem similar to the group research project at the end of x-axis the region bounded by the curve
Chapter 4, but add a couple of complications. Imagine a cylin¬
drical fuel tank of length L lying on its side; its ends are ellip- 2
tically shaped and are defined by the equation -V “ v/UT^2
79. Find the area between the curves y = tan2 x and y = sec2 x
We are concerned with the amount of fuel in the tank for a
on the interval [0, |],
given level.
a. Explain why the area of an end of the tank can be modeled 80. Find the surface area of the solid generated by revolving the
by region bounded by the x-axis and the curve
A=2aLf~Wdy y = ex + \e x
V(h) = 2La
"Houdini's Escape"
This project is to be done in groups of three or four students. Each group will submit a
single written report.
H arry Houdini was a famous escape artist. In this project we relive a trick of his
that challenged his mathematical prowess, as well as his skill and bravery. It may
challenge these qualities in you as well.
Houdini had his feet shackled to the top of a concrete block which was placed on
the bottom of a giant laboratory flask. The cross-sectional radius of the flask, mea¬
sured in feet, was given as a function of height z from the ground by the formula
r{z) = 10z-1/2, with the bottom of the flask at z — 1 ft. The flask was then filled with
water at a steady rate of Tin ft3/min. Houdini’s job was to escape the shackles before
he was drowned by the rising water in the flask.
Now Houdini knew it would take him exactly 10 minutes to escape the shackles.
For dramatic impact, he wanted to time his escape so it was completed precisely at the
moment the water level reached the top of his head. Houdini was exactly 6 ft tall. In
the design of the apparatus, he was allowed to specify only one thing: the height of the
concrete block he stood on.
Your paper is not limited to the following questions but should include these con¬
cerns: How high should the block be? (You can neglect Houdini’s volume and the
Harry Houdini (born Ehrich Weiss) volume of the block.) How fast is the water level changing when the flask first starts
(1874-1926) to fill? How fast is the water level changing at the instant when the water reaches the
top of his head? You might also help Houdini with any size flask by generalizing the
derivation: Consider a flask with cross-sectional radius r(z), an arbitrary function of z
with a constant inflow rate of dV/dt = A. Can you find dh/dt as a function of h{t)l
MATHEMATICS is the gate and key of
the sciences. ... Neglect of mathematics
works injury to all knowledge, since he
who is ignorant of it cannot know the other
sciences or the things of this world. And
what is worse, those who are thus ignorant
are unable to perceive their own ignorance
and so do not seek a remedy.
-Roger Bacon
Opus Majus, Part 4,
Distinctia Prima, cap. 1.
Chapter 7 Review
Group Research Project: Buoy Design
425
426 Chapter 7 Methods of Integration
REVIEW OF SUBSTITUTION
Remember that when doing substitution you must choose u, calculate du, and then
substitute to make the form you are integrating look exactly like one of the known
integration formulas. We will review substitution by looking at different situations and
special substitutions that prove useful.
Solution
Let u be the value in parentheses; that is, let u — x3 - 2. Then du = 3x2 dx, so by
substitution,
You can check your work by finding the derivative of your answer to see if you obtain
the integrand. For this example,
ix’ - 2)5
Solution
We notice the similarity between this and the formula for inverse sine if we let a
and u = t2. Then du = 21 dt and
du 2 du
If u = t , then du — 21 dt, t dt = —, and
Formula 22, Appendix D (inverse
sine): Hrh-S V'1 ~“2 Vi - r4 = yi - (r2)2 = yr ir
2
du
/ ■-
Va2 — u2
.-l
- = sin -
a
-II /1 s
du
— U2
Don't forget the factor
- j sin u + C = sin t2 + C
The method of substitution (Section 5.5) is very important, because many of the
techniques developed in this chapter will be used in conjunction with substitution.
Examples 3 and 4 illustrate additional ways substitution can be used in integration
problems.
Solution
Multiply the integrand sec* by secjc + tan* and divide by the same quantity:
You may wonder why anyone would think to multiply and divide the integrand
sec a: in Example 3 by sec a: + tana:. To say that we do it “because it works” is prob¬
ably not a very satisfying answer. However, techniques like this are passed on from
generation to generation, and it should be noted that multiplication by 1 is a common
method in mathematics for changing the form of an expression to one that can be ma¬
nipulated more easily.
+ ex
Solution
The straightforward substitution u— \ + ex does not work:
du
f dx _ f 7 T _ f du
This is not an appropriate form
J 1 + ex J u J ex u because x has not been eliminated.
u = I + ex\ du — ex dx
428 Chapter 7 Methods of Integration
1 e~x ( 1 \
then make the substitution
1 + ex 1 T ex e~x T 1
T 1; du = —e x dx
f dx f e x dx f —du , , ,
/ :—~= / . = / - = — In \u\ T C = — ln(e T 1) T C
J l + ex J e x + 1 J u
When the integrand involves terms with fractional exponents, it is usually a good
idea to choose the substitution x = un, where n is the smallest integer that is divisible
by all the denominators of the exponents (that is, the least common multiple of the
denominators). For example, if the integrand involves terms such as x1/4, x1^, and
;c1//6, then the substitution x = un is suggested, because 12 is the smallest integer
divisible by the exponential denominators 4, 3, and 6. The advantage of this policy
is that it guarantees that each fractional power of x becomes an integral power of u.
Thus,
Find J dx
■M 3+xl/2'
Solution
Because 6 is the smallest integer divisible by the denominators 2 and 3, we set
- xvl/6
' , so that and 6u5 du = dx. We now use substitution
dx 6u5 du ,
/ x!/3+x1/2
// (i/6)1/3 + (h6)1/2
6u*du
6u5du fr A"5
6m5du
Let x = u .
f 6u^du
u2 + u3 J «2(1 + u) J 1 +u
Substitution does not guarantee an obvious integrable form. When the degree of the
numerator is greater than or equal to the degree of the denominator, division is often
helpful. By long division
6 w3 -6
= 6 u" — 6u + 6 +
1 4~ u It u
f 6«3 du f /
J IT u J\ 6«2 — 6m T 6 T
1 Tu
= 2m3 — 3m2 T 6u 6 In 11 T u | T C
Example 5 (especially the part involving long division) seems particularly lengthy.
When faced with the necessity of integrating a function such as
f 6«3du
J IT u
7.1 Review of Substitution and Integration by Table 429
most people would turn to a computer, calculator, or a table of integrals. If you look,
for example, at the short integration table (Appendix D), you will find (Formula 37)
When using integration tables, we note that the algebraic form does not always match
the form we obtain by direct integration. You might wish to show algebraically that
the form we have just found and the form
2u3 — 3w2 -f 6w — 6 In 11 + m | + C
Take a few moments to look at the integration table in the Student Mathemat-
✓'jjjjjjjps ics Handbook (or Appendix D). Notice that the table has two basic types of integration
^-J formulas. The first gives a formula that is the antiderivative, whereas the second, called
a reduction formula, simply rewrites the integral in another form. The following illus¬
tration demonstrates each of these forms.
Find/
Solution
Even though this was not too difficult, it is a bit tedious, so we might think to evaluate
this integral by using a table of integrals. This is an integral involving an expression of
the form au + b\ we find that this is Formula 32, where u = a:, a — -1, b = 3, and
n = 5.
One of the difficult considerations when using tables or computer software to carry
out integration is recognizing the variety of different forms for acceptable answers.
Note that it is not easy to show that these forms are algebraically equivalent. If we
use computer software for this evaluation, we find yet another algebraically equivalent
form:
f 9 , x8 15.x-7 , 405a-4 ,
/ a2(3 - a)5 dx =-+-15a6 + 54a5-+ 81a3
J 8 7 4
Once again, remember that you must add the + C to the computer software form. ■
Solution
The integrand is in logarithmic form; from the table of integrals we see that Formula
501 applies, where u = x and n = 4. Take a close look at Formulas 499-502, and note
that Formula 501 gives another integral as part of the result.
This is called a reduction formula because it enables us to compute the given integral
in terms of an integral of a similar type, only with a lower power in the integral.
/
EXAMPLE 8
.x dx
—-
Using an integral table after substitution
V8 - 5x2
Solution
This is an integral of the form Va2 - u2, but it does not exactly match any of the
SMH
formulas.
Note, however, that except for the coefficient of 5, it is like Formula 225. Let u = V5 x
(so u2 = 5a:2); then du = \/5 dx:
r— ll
If udu
5 J \/8 — w2
u du
, , /--7 = i(-V8-M2) + C = -^8-5x2 + C
\!
•> J V8 — uz J 5
As you can see from Example 8, using an integral table is not a trivial task. In fact,
other methods of integration may be preferable. For Example 8, you can let u = 8-5x2
and integrate by substitution:
du
f xdx f ' \-10x/ 1 f _l/2
J 7T^ = J —7^~ = ~r0Ju du
= -T(2u>/2) + C = -1^8- 5*2 + C
Of course, this answer is the same as the one we obtained in Example 8. The point
of this calculation is to emphasize that you should try simple methods of integration
before turning to the table of integrals.
Find / 5x2\/3x2~+Tdx.
7.1 Review of Substitution and Integration by Table 433
Solution
This is similar to Formula 170, but you must take care of the 5 (constant multiple) and
the 3 (by making a substitution).
u2 = 3x2 so that x2 — —
3
3V3
J u2\[u 2 + 1 du Use Formula 170, where a — 1.
5
[2V3 ;t(3;c2 + 1)3/2 - y/3xy/3x2 + 1 - In s/3x + x/3x2 + l|] + C
24V3
If you use a calculator or computer, you will probably obtain an alternate, but equiva¬
lent, form:
l y/x2 + 5x
dx 2
/¥ dx
Sm.v dx
Integrate the expressions in Problems 13-24 using the short table
of integrals given in Appendix D.
4 cosx e
■rx In* 7cw f dx f dx
t2 dt 13. 14.
x dx ' X2y/a2 — X2
6 1 x2y/x2-a2
•/ 4 + x4 •/ 9 + /6
f 4x
4xi — 4x 15. x In .v dx 16. f \nxdx
8 dx
•/ (1 + cot a:)4 esc2 x dx
' J ^ 2jc2 + 3
r dx
2x + 4
■I ^
xJ — X
dx 10 dx 17. f xeat dx 18.
lx2 + 3)2 •/ x 2 + 4x +• 3 1 a + be2x
434 Chapter 7 Methods of Integration
27
;i x(l + x)3 dx 28. Xy/\ + xdx
the curve
y =
[-3/2
V^T9
29 xe4x dx 30. x In 2x dx
•/ between x = 0 and x = 9 is revolved about the x-axis.
dx
31 32. In3 x dx
•/ 1 +
x3 dx
33
•/ V4x4 + 1
Q Use the Student Mathematics Handbook or other avail-
TSMfT) a^e inte8rafi°n tables to find the integrals given in
Problems 34-39. Cite the formula number or source
you are using.
dx 35 dx
7sec7)
*7 sin6 x dx 37
•/ 9x2 + 6x + 1 54. Find the volume of the solid generated when the curve
y = x(l — x2)1/4 from x = 0 to x = 1 is revolved about
sin2 x
38. /<9-M
u.f sin2 x cos2 x dx Hint: Use the identities shown in 61. Show that J cscx dx = - In |cscx + cotx| + C.
Problems 40 and 41.
q$q x -{“ cot X
45. Exploration Problem Using Problems 42-44 formulate a proce¬ Hint: Multiply the integrand by-4
dure for integrals of the form cscx + cotx
/ sin™ * cos" * dx
a. Let u = cosx.
7.2 Integration by Parts 435
b. Let u — sin x.
c. Write 2 sin x cos x = sin 2x and carry out the integration. Torus generated by
the circle x2 + (y-b)2 = 1
d. Show that the answers you obtained for parts a-c are the
same.
^ 63. Derive the formula
/»7T r*7T
/ x/(sinx) dx = — / f(sinx)dx
Jo 2 J0
64. Find the surface area of the torus generated when the circle
x2 + (y — b)2 = 1, b > 1
Integration by parts is a procedure based on reversing the product rule for differen¬
tiation. We present this section not only as a technique of integration but also as a
procedure that is necessary in a variety of useful applications.
Recall the formula for differentiation of a product. If u and v are differentiable func¬
tions, then
d(uv) = u dv + v du
Integrate both sides of this equation to find the formula for integration by parts:
/ f
d(uv) — j u dv -F Jf vdu
UV — / udv + if vdu
1
If we rewrite this last equation, we obtain the formula summarized in the following
box.
/ udv =uv—
/ v du
To use integration by parts, we must choose u and dv so that the new integral is easier
to integrate than the original.
J x ex dx
u dv
Let u =x and dv = ex dx
du = dx v — f ex dx — ex
!xe*ix I dx = xex — ex + C
du
You can check your work with integration by parts by differentiating, using software, or
using the integration table found in Appendix D or the Student Mathematics Handbook
(Formula 484, with a — 1). ■
Note: You may wonder why an arbitrary constant (call it K) was not included when
performing the integration associated with f dv in integration by parts. The reason is
that when applying integration by parts, we need just one function v whose derivative
is dv, so we take the simplest one—the one with K — 0. You may find it instructive to
see that taking v — ex + K gives the same result.
Integration by parts is often difficult the first time you try to do it because there is
no absolute choice for u and dv. In Example 1, you might have chosen
u = ex and dv — x dx
du = ex dx v =
Then
J xex dx = ex y - J y e^dx
V V
— -x2ex — - f x2ex dx
2 2 J
Note, however, that this choice of u and dv leads to a more complicated form than the
original. In general, when you are integrating by parts, if you make a choice for u and
dv that leads to a more complicated form than when you started, consider going back
and making another choice for u and dv.
Generally, you want to choose dv to be as difficult as possible (and still be some¬
thing you can integrate), with the remainder being left for the w-factor.
Solution
Let u = lnx and dv = dx
v =x
du dx
Sometimes integration by parts must be applied several times to evaluate a given inte¬
gral.
1
du = 2x dx
a
II
1
/ A-*C=X*(-.-)-/(-.-X2xdx)
/
= —xe x + 2 / xe x dx
Let u —x and dv — e x dx
1
du = dx
3
II
1
= —* + 2 Jc(-0 e“*)d;c
-/<-
Do not forget the grouping symbols here.
— —e X(x2 + 2x + 2) 4- C
Find J e2xsinxdx.
Solution For this problem you will see that it will be useful to call the original inte¬
gral I. That is, we let
e2x sinx dx
-I
Let u = e 2x and dv = s'mxdx
du = 2e2x dx V — — cosx
/ = /^sin,^ = ^(-cos,)-/(-cos,)(2^rf,)
= -e2*coSx + 2j e2'cos, dx
2x
= —e cos x + 2 e2x (sinx) sinx(2e2' dx)
2x
— —e cosx + 2elx sinx — 4 2x sin x dx
Check with the integration table (Formula 492, where a = 2 and b = 1), or by
taking the derivative. g
The integration by parts formula can be used for definite integrals, as summarized in
the following box.
You should recognize that this formula for definite integrals is the same as the formula
for indefinite integrals, where the first term after the equal sign has been evaluated at
the appropriate limits of integration. This is illustrated by the following example.
TECHNOLOGY NOTE
l xe2x dx = -xe2x
2 o 1 Jo
e2x dx
Solution
n /4
1
x tan 1 x-ln(l + x2)
2 0
n (
— (tan
L4 \
Jt \
7
i
“21"!v1 +
(
Te)J
7T2\1 1
0 - - In 1
2
= - tan-1 - - - In (1 + —) » 0.2827
4 4 2 V 16/
Check with an integration table (Formula 457, for example, with a = 1). ■
LJ
( xe-2* dx 2- ^ j x sinx dx U 23. WHAT DOES THIS SAY? Describe the process known as in¬
tegration by parts.
3-J
r
( x In x dx 4- I
r
x tan 1 x dx In Problems 24-27, first use an appropriate substitution and then
r r integrate by parts to evaluate the integral. Remember to give your
5. 1f sin-lx4x 6. / x2sinx4x answers in terms of x.
lnx sin(lnx)
7. I e 3x cos 4x dx 8. I e2* sin 3x dx
*7'
24.
/ x
dx 25. /[si
[sin 2x ln(cosx)] dx
by parts.
7
find an approximate answer correct to four decimal places.
17.•
f* 4
jf \[x sfx lnx4x 18.
7
/
*
x3lnx4x
32. Use Problem 30 to evaluate
parts.
/ Vx^TT using integration by
jrw 20. 1
• /;1/3
3(ln3x) 4x 33 . Find J x" lnx dx, where « is any positive real number.
440 Chapter 7 Methods of Integration
34. After t seconds, an object is moving along a line with veloc¬ decibels, where I0 = 10~12 watt/m2 is the threshold of au¬
ity of fe~'/2 meters per second. Express the position of the dibility (the lowest intensity that can be heard). What is the
moving object as a function of time. average value of L as the intensity of a TV show ranges be¬
35. After t hours on the job, a factory worker can produce tween /o and /] = 3 • 10~5 watt/m2?
lOOre-0 5' units per hour. How many units does the worker 49. The displacement from equilibrium of a mass oscillating at the
produce during the first 3 hours? end of a spring hanging from the ceiling is given by
36. After t weeks, contributions in response to a local fund-raising
campaign were coming in at the rate of 2,000re~°-2' dollars per y = 2.3e-0'25' cos5r
week. How much money was raised during the first 5 weeks?
feet. What is the average displacement (rounded to the near¬
37. Find the volume of the solid generated when the region under est hundredth) of the mass between times t — 0 and t = n/5
the curve y = e~x on the interval [0, 2] is revolved about the seconds?
y-axis.
38. Find the volume of the solid generated when the region under
the curve y = sinx + cosx on the interval [0, |] is revolved
about the y-axis.
39. Find the volume of the solid generated when the region under - equilibrium point
and r and k are, respectively, the expulsion speed and the rate 53. Derive the reduction formula
of consumption of the fuel, which are assumed to be con¬
stant. As usual, g — 32 ft/s2 is the constant acceleration
due to gravity. Suppose w = 30,000 lb, r = 8,000 ft/s,
J xnex dx — xnex — n J xn lex dx
L = 10 log —
/ (lnx)" dx = x(lnx)
■-"/
(lnx)" 1 dx
Q 55. Wallis's formula If n is an even positive integer, use reduction 56. Exploration Problem State and prove a result similar to the one
formulas to show in Problem 55 for the case where n is an odd positive integer.
/•tt/2 /<n/2
/ sin" x dx — <cos" x dx
Jo Jo
1 • 3 • 5.(n — 1) 71
2-4-6.n 2
Problems 40-45 of Problem Set 7.1 anticipated the results of this section. We begin by
considering a product of powers of sine and cosine, which we represent in the form
/ sinm x cos" x dx
There are essentially two cases that must be considered, depending on whether the
powers m and n are both even or not. We will state the general strategy for handling
each case and then illustrate the procedure with an example.
Evaluate
/ sin4 x cos3 x dx.
Solution
Since n — 3 is odd, peel off a factor of cos x and use cos2 x = 1 — sin~ x to express the
integral as a polynomial in sinx.
= /Sin4*a-si„M(cos^)
/
= / w4(l — u~)du
I,.5 — zjU
= iU' 1..7 + C
w=sinx; du = cosxdx
5*
= J sin3 x — ^ sin7 x + C
Case II: Neither m nor n is odd
General strategy: Convert this into Case I by using the identities
sin2* = |(1 — cos2*) and cos2* = ^(1 + cos2*). These are sometimes called
half-angle identities.
Evaluate J sin2*cos4*dx.
Solution
TECHNOLOGY NOTE
/ tan* dx = In |sec *| + C
Case I: n is even
General strategy: Peel off a factor of sec2 x from the integrand and use the identity
sec2 x = tan2 x + 1 to express the integrand in powers of tanx except for (sec2 x dx);
substitute u = tanx, du = sec2 x dx, and integrate by using the power rule.
Solution
= | tan5 x + | tan3 x + C ■
Solution
/sec" xdx —
sec” 2xtanx
n — 1
+
n- J
n — 2 C
1
sec
n-2
x dx
EXAMPLE 5 Power of the tangent is even and power of the secant is odd
Solution
/ tan2 x sec3 x dx
-s (sec2 x — 1) sec3 x dx
= J sec3 x dx — J sec3 x dx
' sec3 x tan x 3
+ - J sec3 x dx j — J sec3 x dx
sec3 x tan x 1
/ sec3 x dx
sec3 x tan x
sec3 x tan x
1 sec x tan x
sec x tan x
1
---1— / sec x dx
1
\s
In |secx + tanxl + C
8 8
TRIGONOMETRIC SUBSTITUTIONS
Thus, the substitution u = a sin 9, du — a cos 9 d9 eliminates the square root and may
convert the given integral into one involving only sine and cosine. This substitution
can best be remembered by setting up a reference triangle. This process is illustrated
in the following example.
Find fV^dx.
Solution
First, using a table of integration, we find (Formula 231; a = 2)
f V4^dn = i^Z+2si„->(i)+C
Our goal with this example is to show how we might obtain this formula with a trigono¬
metric substitution. Refer to the triangle shown in Figure 7.1.
Let x = 2sin0, so dx — 2 cos 9 d9. Then,
sin e = | [ 1 + cos 29 ,
= 4 / ---d9 Half-angle identity
The final step is to convert this answer back to terms involving x. Using the refer¬
ence triangle in Figure 7.1, we find that
Similar methods can be used to convert integrals that involve terms of the form
s/a1 + u2 or y/u2 — a2 into trigonometric integrals, as shown in Table 7.1. For this
table, we require 0 <6 <
Solution
Let x — 3 tan#, dx = 3 sec2 0 dd\ then
sec 0 = + x~
= 81 J tan2 9 sec3 0 d6
3
This integration can now be completed by looking at Example 5. In order to express
Figure 7.2 Reference triangle the antiderivative in terms of the original variable x, you can use the reference triangle
with form y/a2 + u2 in Figure 7.2.
Since tan# = x/3, we have sec# = (y/9 + x2)/3, and by substituting into the
solution shown in Example 5, we find
(9 + x2y/2 x
= -(9 + x2)3'2 - —(9 + X2)1'2 - - In + C
4 8 8 3 h 3
446 Chapter 7 Methods of Integration
Solution
Let x = sec 0, dx — sec 9 tan 9 d6\ we use the reference triangle shown in Figure 7.3.
tan" 9 = sec" 9 — 1
5 tan5 0 + | tan3 0 + C
1(x2_1)5/2+ .(;c2_1)3/2 + c
QUADRATIC-FORM INTEGRALS
Solution
Complete the square within the radicand:
16jc — 2x2 — 23 = — 2(x2 — Sx ) — 23
= -202 - 8x + 42) + 2 • 42 - 23
= -2{x - 4)2 + 9
Thus,
Where u = y/l(x — 4)
Where u — 3 sin 9
Half-angle identity
7.3 Trigonometric Methods 447
-1 72
sin + 7r^Vl6x - 2x2 - 23 + C
2^2 T^-4)
This last step requires back-substituting from 0 to u and then from « tox. Details are
left as an exercise. ■
13.
r. , , ,
/ sm~ x cos x dx 14.
C sin xdx and the x-axis is revolved about the y-axis 0 < x < n.
! cos5 X 54. A particle moves along the x-axis in such a way that the ac¬
celeration at time t is a(t) = sin21. What is the total distance
15. 1 tan 20 d9 16.
(sec(7 traveled by the particle over the time interval [0,7r] if its initial
velocity is v(0) — 2 units per second?
17. 1 tan3 x sec4 x dx 18. 1 sec5 x tan xdx
21. j” tan2 u sec u du 22. sec4 x dx In Problems 56-59, use the following identities:
r
35.
’f XALdx
V4 + x2 36-j
j \/9 + x2dx
on the interval [0, |]. Find the arc length of the curve
y — f(x) over this interval.
f dx f dx O f x dx
37.
‘ sjx2-l 38-j ' 5 + 2x2
61. Evaluate / ——-—,.—.
J 9 - x2- V9 22^2
448 Chapter 7 Methods of Integration
You are familiar with the algebraic procedure of adding a string of rational expressions
to form a combined rational function with a common denominator. For example,
2 -3 2Qc + 2) + (—3)Qc + 1) -x + 1
x + 1 x +2 (x + l)(x + 2) x2 + 3x + 2
In partial fraction decomposition, we do just the opposite: We start with the reduced
fraction
—x + 1
x2 + 3x + 2
2 -3
x -f- 1 x 2
2
This procedure has great value for integration because the terms and-are
x + 1 x T" 2
easy to integrate. In particular,
f -X +1 . f 2 J f -3 J
/ z-- dx = / -- ax + / -- dx
J x2 4- 3x + 2 J x 1
J x -\- 2
= 2 In |x + 11 — 3 In |x + 2| + C
P(x)
f(x) =
D(x)
that are reduced in the sense that P(x) and D(x) are polynomials in x which have no
common factors and the degree of P is less than the degree of D. In algebra, it is
shown that if P(x)/D(x) is such an expression, then
P(x)
—— = F\(x) + F2{x) 3-h Fn{x)
D(x)
A Ax + B
- or —--
(x — r)n (x2 + sx + t)m
P(x)
If is not reduced, then we simply divide until a reduced form is obtained,
D(x)
For example (x ^ 1),
^
+ ••• +
x -r (x - r)2 (x — r)n
x2 — 6x + 3 Ai i A2 A3
Solution +
(x - 2)3 “ x -2 + (x - 2)2 (x - 2)3
This implies (by equating the coefficients of the similar terms) that
1 = A1 x2 terms
—6 = —4A\ + A2 x terms
3 = 4Ai — 2A2 — 5 Constants
x2 — 6x + 3 1 —2 —5
(x — 2)3 x — 2 (x — 2)2 (x — 2)3
If there are two or more linear factors in the factorization of D(x), there must be a
separate cascade for each power. In particular, if D(x) can be expressed as the product
of n distinct linear factors, then
_P(x)_
(x — r 1)(x — r2) ■ ■ ■ (x - rn)
8x - 1 8* - 1
Solution First factor the denominator.
x2 — x — 2 (x — 2)(x + 1)
Break up the fraction into parts,
- ^2
each with a linear denominator.
x — 2 x + 1 The task is to find A\ and A2.
A\(x + 1) + A2(x — 2) Obtain the least common
Note that the degree of the
denominator is the same number as the ~~ (x — 2)(x + 1) denominator on the right.
number of arbitrary constants, A\ and
Aj. This provides a quick intermediate Now, multiply both sides of this equation by the least common denominator, which is
check on the correct procedure. (x — 2)(x + 1) for this example.
8x — 1 = Ai(x + 1) + ^2(x — 2)
Substitute, one at a time, the values that cause each of the factors in the least common
denominator to be zero.
8x - 1 5 3
Thus,
x2 - x -2 x —2 x T 1
If there is a mixture of distinct and repeated linear factors, we combine the proce¬
dures illustrated in the preceding examples. For example,
5x2 + 21x + 4 A] A2 A3 A4
--- is decomposed as
(x + l)3(x - 3) F x + 1 (x + l)2 (x + l)3 x —3
Note that the degree of the denominator is 4, so we use four constants.
If the denominator D(x) in the rational expression P(x)/D(x) contains an irre¬
ducible quadratic power, the partial fraction decomposition contains a different kind of
cascading sum.
Decompose
7.4 Method of Partial Fractions 451
Solution
The decomposition gives
Multiply both sides of this equation by (x2 + l)2, and simplify algebraically.
Next, equate the corresponding coefficients on each side of this equation and solve the
resulting system of equations to find A\ — -3, A2 = 2, B, = 0, and B2 = 0. This
means that
—3x3 — x —3x 2x
Many of the examples we encounter will offer a mixture of linear and quadratic
factors. For example,
x2 + 4x - 23 A3
—-z-7—-—r is decomposed as A|* + Bl , ,
(x2+ 4)(x+ 3)2 F x2 + 4 x +3 (X + 3)2
The degree of the denominator is four, and there are four constants.
In algebra, the theory of equations tells us that any polynomial P with real co¬
efficients can be expressed as a product of linear and irreducible quadratic powers,
some of which may be repeated. This fact can be used to justify the following general
procedure for obtaining the partial fraction decomposition of a rational function.
Step 1. If the degree of P is greater than or equal to the degree of D, use long
P(x)
(or synthetic) division to express — ■ - as the sum of a polynomial
D(x)
R(x)
and a fraction in which the degree of the remainder polynomial
D(x)
R(x) is less than the degree of the denominator polynomial D(x).
Step 2. Factor the denominator D(x) into the product of linear and irreducible
quadratic powers.
P(x)
Step 3. Express ——- as a cascading sum of partial fractions of the form
D(x)
At AkX + Bk
and
(x — r)n (x2 + sx + t)r'
Solution
From Example 1, we have
[ x2 — 6x + 3 -2 -5
J / -^— ax
(x- 2)3 / +-rrr +
= In |x — 21 + + + C
-2 2(x — 2)2
Solution
We have P(x) = x4 + 2x3 — 4x2 + x — 3 and D(x) = x2 — x — 2; because the degree
of P is greater than the degree of D, we carry out the long division and write
/ xc2 — x — 2
--dx = I I x‘ + 3x + 1 + -
J
./ \
\ -/( x2
xz—x — 2.
dx
The polynomial part is easy to integrate. The rational expression was decomposed into
partial fractions in Example 2.
-/( x -f- 3x H- 1 H-
x —2 x+ 1
'3 3x2
— T “(“X + 5 In |x — 2| + 3In |x T 1| C
Solution
From Example 3,
—3x- 2x f —3x
/ (x2 + l)2
dx
-I (x2 + l)2
rdx + / —-r dx
X2 + 1
u — x2 T 1; du — 2x dx
= J u~2du-lj u du
= — u-In \u\ + C
2
3
ln(x2+ 1) + C
x2+l
7.4 Method of Partial Fractions 453
Solution
The partial fraction decomposition of the integrand is
5x2 + 21x + 4 = (A2 + A3)x~ + (Ai — 2A2 + 2A3)x + (—3Aj — 3A2 -f- A3)
Equate the coefficients of x2, x, and 1 (the constant term) on each side of the equation:
5 = A2 + A3 x2 terms
21 — A\ — 2At + 2A3 x terms
4 = —3A] — 3A2 + A3 Constants
Solution
The partial fraction decomposition of the integrand has the form
x2 + 4x —23 A\x + B\ A2
-—- -— 1
(x2 + 4)(x + 3) x2 + 4 x +3
Multiply both sides by (x2 + 4)(x + 3) and then combine the terms on the right:
Ai + A2 = 1 x2 terms
- 3Ai + B\ = 4 x terms
3B\ + 4A2 = —23 Constants
Solve this system (the details are not shown) to find A\ = 3, B\ = —5, and A2 = —2.
We now turn to the integration:
x2 + 4x - 23 , 3x -5 —2 dx
/ -r-
(x2 + 4)(x + 3)
dX
/ x2 + 4
dx T
/ * +3
454 Chapter 7 Methods of Integration
C x dx f dx 2 [ dx
J
i
-v-
x2 + 4
-j
5 J x2 + 4 J x +3
T_
u — x2 + 4; du = 2x dx
= 3 - ln(x2 + 4) 2 In lx 4" 31 + C
The German mathematician Karl Weierstrass (1815-1897) noticed that the substitution
x
u — tan — — tv < x < tv
2
will convert any rational function of sin x and cos x into a rational function of x. To
see why this is true, we use the double-angle identities for sine and cosine (see Figure
7.4):
x x
sin x — 2 sin — cos — and cos x = cos- sin
2 2
u
= 2
V Vl + u2, .VT+u* . a/1 + U- ,\/l 4r U2,
2u 1 u
Figure 7.4 Weierstrass 1 + u* 1 + u2
substitution x 2
Finally, because u — tan —, we have x — 2 tan u, so dx =-- du.
2 1 + u1
X
Weierstrass Substitution For —tv < x < tv, let u = tan - so that
2
2u 1 — u2 2
sinx =--, cosx = ---, and dx —-- du
1 +u2 1 +u2 1 + u2
This is called the Weierstrass substitution.
The following example illustrates how this substitution can be used along with
partial fractions to integrate a rational trigonometric function.
x 2 du
Use the Weierstrass substitution—that is, let u — tan—. Remember, dx =-
2 1 + uz
1 — u2 2u
cos a: = -and sinx =
1 + u- 1 + u2
2 du
dx
/ 3 cos x — 4 sin x -/-7TTP
J
1 -f u:
(\ — u2\
31 --r 1-4
vl+w2/
(
\l+u2J
2u
Simplify.
r —2du C —2du
J 3n2-f 8w — 3 J (3« — 1)(m + 3)
7.4 Method of Partial Fractions 455
dx C —2du
f 3 cos a: — 4 sin x l)(w + 3)
j du
-l&d 3 1
m T 3
— — • - In |3m — 11 H— • In |m + 3| +C
5 3 5
1 X 1 X
-In 3 tan-1 + - In tan —h 3 + C
5 2 5 2
Once again, observe that when carrying out integration, you may obtain very dif-
ferent forms for the result. For Example 9, you might use an integration table (Formula
393 in the Student Mathematics Handbook, for example) to find
^au + tan 1 ^^
du
/ p sin mm + q cos mm a^Jp2 + q2
In tan
dx 1 T + tan-'fiTT
/ 3 cosx — 4 sin x
= - In tan
5
+ C
5.
4
2x2 + x
6.
x2 — x + 3
x2(x — 1)
15./
x{x — 3)
dx
,6./ 3x
dx
7.
Ax3 + 4.r2 + x — 1
x2(x + l)2
8.
x2 — 5x — 4
(x2+ l)(x-3) i77 3x2 + 2x
4dx
x (x + 1)
dx
187 2x2 + 5jc — 1
x(x2 — 1)
f x22 -
-xx +3
dx
Find the indicated integrals in Problems 21-36. 59. Find the volume (to four decimal places) of the solid generated
when the region under the curve
2x3 + 9x - 1 x2 + 2
2i7 x2(x2 - 1)
dx
/ x2(x — 1)
dx
1
a:2 + 1
*7
f dx Vx2 + 4x + 3
dx 24
x2 + x - 2 ' J ^8
on the interval [0, 3] is revolved about
f x44 + 1 fx3 + l
25.
JI '——^dx
x4- 1
26
i x3 - 1 dx a. the x-axis b. the y-axis
60. Find both the exact and approximate volume of the solid gen¬
f 2x dx
4—x
27
•/ (x + l)2 28
'J (x - 2)2 erated when the region under the curve y2 = x"
4+x
on
x+
dx
7 the interval [0, 4] is revolved about the x-axis.
2
29 30 dx
•/ x(x + l)(x - 2) x(x — l)2 61. HISTORICAL QUEST George
*7
x ofx
(x + l)(x +
5x + 7
2)2
32,
7 X + 1
x(x2 + 2)
5x dx
dx
Polya was born in Hungary and at¬
tended universities in Budapest, Vienna,
Gottingen, and Paris. He was a pro¬
/ x2 + 2x — 3
r/x 34
7 ^ 6x +9
3x2 + 4x + 1
fessor of mathematics at Stanford Uni¬
versity. Polya's research and win¬
3x2 - 2x + 4
x2 + 4x _
037. WHAT DOES THIS SAY?
4
dx 36
7 x3 + 2x2 + x —
Describe the process of partial
2
dx
ning personality earned him a place
of honor not only among mathemati¬
cians, but among students and teach¬ GEORGE POLYA
ers as well. His discoveries spanned 1887-1985
fraction decomposition.
. an impressive range of mathematics,
Find the indicated integrals in Problems 38-55. including real and complex analysis, probability, combina¬
cos x dx f ex dx torics, number theory, and geometry. Polya's book, How to
38.
sin2 x — sin x — 2 39'j' 2e2x — 5ex — 3 Solve It, has been translated into 20 languages. His books
have a clarity and elegance seldom seen in mathematics, mak¬
ex dx f sin x dx
40. ing them a joy to read. For example, here is his explanation
e2x - 1 4lj1 (1 -j- cosx)2
of why he was a mathematician: "It is a little shortened but
tanx dx f sec2 x dx
7 sec2 x + 4
dx
43-j' tan x + 4
f dx
not quite wrong to say: I thought I am not good enough for
physics and I am too good for philosophy. Mathematics is in
between." ■
44
7 X1/4 - X
dx
45‘j 1
f
X2/3_XM2
dx
A story told by Polya provides our next Quest. “A number
of years ago,” Polya related with his lovable accent, “I delib¬
46. erately put the problem
sin x — cos x 47-j1 3 cos x + 4 sin x
48.
dx f sinx — cosx
49. 1 . WA
/ x dx
5 sin x + 4 ' sinx + cosx ^9
dx f dx
7 4cosx + 5
dx
51-j' sec x — tan x
f dx
as the first problem on a test of techniques of integration, to
give my students a boost as they began the exam. With the
7 3 sinx + 4cosx + 5
dx
53. ^
' 4sinx — 3cosx — 5
f dx
substitution u — x2 — 9, which I expected the students to use,
you can knock the problem off in just a few seconds. Half of
the students did this, and got off to a good start. But a fourth of
7 2 esc x — cot x + 2
55. ^
' x(3 - lnx)(l - lnx)
1
them used the correct but time-consuming procedure of partial
fractions-—and because they spent so much time on the prob¬
56. Find the area under the curve y — between lem, they did poorly on the exam. Half of the rest used the trig
x2 + 5x + 4 substitution x = 3 sin 9—also correct but so time-consuming
x = 0 and x = 3. that they wound up very far behind and bombed the exam. It
57. Find the area of the region bounded by the curve is interesting that the students who used the harder techniques
showed they knew ‘more,’ or at least more difficult mathe¬
y —-— and the lines x = \, x = and y = 0.
6 — 5x + x2 J 4 matics than the ones who used the easy technique. But they
showed that ‘it’s not just what you know; it’s how and when
58. Find the volume (to four decimal places) of the solid generated you use it.’ It’s nice when what you do is right, but it’s much
when the curve better when it’s also appropriateCarry out all three meth¬
ods of solution of the given integral Polya described in this
1
is revolved about
* “Polya, Problem Solving, and Education,” by Alan H. Schoenfeld, Math¬
a. the y-axis b. the x-axis c. the line x = — 1 ematics Magazine, Vol. 60, No. 5, December 1987, p. 290.
7.5 Summary of Integration Techniques 457
62. Spy Problem After waiting hours at the sleazy bar for his supe¬ b. Suppose a — b. What is x(t)l What happens to x(t) as
rior, Lord Newton Fleming (recall Problem 55, Section 6.6), t -*■ +oo in this case?
the Spy is about to leave when a courier arrives with shocking 64. Use partial fractions to derive the integration formula
news—Lord Newton has been kidnapped by Ernst Stavro Blo-
hardt! The Spy hurries to the village outside the thug’s chateau dx 1 a+x
disguised as an old duck plucker. On the day he arrives, his
true identity is known only to the Redselig twins, Hans and
f —-- = — In
a2 — x2 2a a —x
+C
Franz, but the next day, Hans’ girlfriend, Blabba, finds out.
Soon, word begins to circulate among the 60 citizens of the 65. Use partial fractions to derive the integration formula
village as to the Spy’s true identity. He determines that the
dx 1 ,
rate at which the rumor concerning his identity will spread is
jointly proportional to the product of the number N of those
who already know and the number 60 — N of those who do
/ - = - In
x(ax + b) b ax + b
+C
not, so that
66. Exploration Problem Consider the integral J \/l — cos x dx.
dN
— = kN(60 - N)
dt a. If you have access to a CAS, report on your attempts to
for a constant k. He needs a week to get the information to find this integral.
break into the chateau to free Lord Newton, but figures that b. Use a Weierstrass substitution to find this integral.
as soon as 20 or more villagers know his identity, Blohardt is c. Can you think of an even more clever way to perform the
sure to find out, too. Does he complete his mission, or is he integration?
about to be a dead duck plucker?
Q 67. Derive the formula
63. Two substances, A and B, are being converted into a single
compound C. In the laboratory, it is shown that the time rate
of change of the amount x of compound C is proportional to J secxdx = In |secx + tanx| + C
the product of the amount of unconverted substances A and B.
Thus, using a Weierstrass substitution.
dx
— = k(a — x){b — x) 68. Derive the formula
dt
for initial concentrations a and b of A and B, respectively,
a. Solve this differential equation to express x in terms of
J CSCxdx = - In |csc JT + cot jr| + C
■%
Step 1. Simplify. Simplify the integrand, if possible, and use one of the procedural rules (see the inside back cover
or the table of integrals).
Step 2. Use basic formulas. Use the basic integration formulas (1-29 in the integration table in Appendix D and the Student
Mathematics Handbook). These are the fundamental building blocks for integration. Almost
every integration will involve some basic formulas somewhere in the process, which means that
you should memorize these formulas.
Step 3. Substitute. Make any substitution that will transform the integral into one of the basic forms.
Step 4. Classify. Classify the integrand according to form in order to use a table of integrals. You may need to
use substitution to transform the integrand into a form contained in the integration table. Some
special types of substitution are contained in the following list:
458 Chapter 7 Methods of Integration
I. Integration by parts
A. J sinm x cos" x dx
m odd: Peel off a factor of (sinx dx) and let u = cosx.
n odd: Peel off a factor of (cosx dx) and let u = sinx.
m and n both even: Use half-angle identities:
B. f tanm x sec" x dx
n even: Peel off a factor of (sec2 x dx) and let u = tanx.
m odd: Peel off a factor of (sec x tan x dx) and let u = sec x.
m even, n odd: Write using powers of secant and use a reduction formula (Formula
428 in the integration tables).
C. For a rational trigonometric integral, try the Weierstrass substitution.
r x i • 2m 1 — u2 2
Let u = tan so that sinx =-cosx - —--, and dx =-- du.
2 1+m2 1 + u2 1+M2
Solution
Keep in mind that there may be several correct approaches, and the way you proceed
from problem to solution is often a matter of personal preference. However, to give
you some practice, we will present several hints and suggestions in the context of this
example.
/ sin y/x
dx
The integrand is simplified and is not a fundamental type. Substitution will work
well for this problem if you let u = y/x. After you make this substitution you can
integrate using a basic formula.
7 (1 + tan“ 9) dO
The integrand can be simplified using a trigonometric identity (1 +tan2 9 — sec2 9),
so that it can now be integrated using a basic formula.
The integrand is simplified, it is not a basic formula, and there does not seem to
be an easy substitution. Next, try to classify the integrand. We see that it involves
powers of trigonometric functions of the type / sin™ x cos'1 .x dx, where m = 3
(odd) and n — 2 (even), so we make the substitution u = cosx.
7 4;t2 cos 3x dx
The integrand is simplified (you can bring the 4 out in front of the integral, if you
wish), it is not a basic formula, and there is no easy substitution. You might try
integration by parts (several steps) or you can check the table of integrals and find
it to be Formula 313.
7 77 x dx
The integrand is simplified and is not a basic formula. However, we might try the
substitution u = 9 - x2. It looks like this will work because the degree of the
numerator is one less than the degree of the denominator.
x2 dx
7 77^7
The integrand is simplified, it is not a basic formula, and it looks as though an
algebraic substitution will not work because of the degree of the numerator. Since
/
the integrand involves 79 — x2, we use the trigonometric substitution jx = 3 sin0,
jc^ dx
. you would have used the
V*2 — 9
substitution x = 3 sec 9, dx = 3 tan 9 sec 9 d9.
7 e2x sin 2x dx
cos4 x dx
h.
/ 1 — sin2 x
The integration can be simplified by writing 1 — sin2* = cos2x. After doing this
substitution, you will obtain
cos4 x dx cos4 x dx
/ 1 — sin2 x 1 cos2 x I ™2xdx
This form can be integrated by using a half-angle identity or Formula 317 in the
table of integrals. It can also be integrated by parts. ■
f 2X~] dx 2. ^ f 2x+3
41
•/ x
2x + 3
dx
/ yjx2 — 16
dx
\1 (x - X2)3 "
3. j" (x sec 2x2) dx 4.
1 Vx2 + 3x
j (x2 esc2 2x3) dx
43
/ V2x2 - 1
dx
dx 44
•/ 1
f tan *Jxdx
45
■/ Xy/x2 + 1
46 [xjx 2 + 1 dx
5. j(ex cot ex)dx 6.
1 tan(lnx)dx
1 yff
47
/
(2x + 1) dx
\/4x — x2 — 2
48 .js + 4x — 4x2 dx
7. 8. j y/cOtX CSC2 X dx
cos x dx sec2 x dx
49
9-
f (3 H- 2 sin/) ^
10. 1
f 2 + cosx j
dx f yj 1 + sin2 x •/ Vsec 2 x — 2
/ cosr 1 sin*
11.
C e2' dt
12.
f sin 2x dx 51 . J sin x dx
5 52,
Icos‘*ix
13.
1 \+e4<
f x2 + X + l dx
1 x2 + 9
14.
’ 1
f 3X + 2
+ sin4 x
537 tan4 x dx S4.fS'c'xdx
1 y/A-X2 Find the exact value of the definite integrals in Problems 55-62.
f e1_v^dx
15. fl+eXdx 16. x2 dx
I l-ex Z yfx • f V*
Jo • L 7?
r 2t2dt r t3 dt
dt
■L
17. 18.
1 Vi -t6 1 + ts ?' s/1 + e2' dt
■l
28
At2 + At + 5
C dx r dx
19. 20.
■
1 1 + e2* 1 A-e~x dx
21.
r
1 x2 + 2x +
dx
2
22.
C dx
1 x2 + x + 4
/
•/ x4Vx2_+_3
2\/3 _
x3 Vx2 + 4 dx 62. ^
l (3 + x2)3/2
/>\/5
x2y/5 — x2 dx
f dx C dx
23. 24.
I x2 + x + 1 1 X2 — X + 1 Find each integral in Problems 63-72.
(2x + 1) dx
25. 1 tan ~xxdx 26. ^ x3sinx2dx 63. 64.
' y/\ 4-e2* VAx2 + Ax + 2
75. Exploration Problem Integrals of the general form 83. Find the centroid (to the nearest hundredth) of the region
bounded by the curve y = x2e~x and the x-axis, between
78. Find the volume of the solid generated when the region under where m and n are positive integers. (This is Formula 510 in
the curve y = cos x between x = 0 and x = j is revolved the Student Mathematics Handbook.) Use the formula to find
about the x-axis.
79. Find the area of the region bounded by the graphs of
v =
2x
—, y = 0, x = 0, and x = 4.
/ x"(lnx)3 dx
Vx2^ '
87. Derive the reduction formula
80. What is the volume of the solid obtained when the region
bounded by y — «/xe~x2, y — 0, x = 0 , and x = 2 is re¬ — sin" 1 Ax cos Ax n — 1
/sin" Ax dx = -:-b Ax dx
volved about the x-axis? An
81. What is the volume of the solid obtained when the region (This is Formula 352 in the Student Mathematics Handbook.)
bounded by y = xV9 — x2 and y — 0 is revolved about the Use this formula to evaluate
x-axis?
82. Find the length of the curve y = ln(secx) from x = 0 to
/ sin4 4x dx
We introduced separable differential equations in Section 5.6, but not all first-order
differential equations have the separable form
dy_ _ fix)
dx g(y)
The goal of this section is to examine a second class of differential equations, called
first-order linear, that have the general form
dx
+ pWy = GU)
For example, the differential equation
dy y _ -x
dx x
462 Chapter 7 Methods of Integration
x^-(x2 + 2)y = xs
ax
? dy
can be put into this form by dividing by xz, the coefficient of —, to obtain
dx
dy_ x5
dx
( x2 + 2\ x5
(Note that P{x) — — (--—j = — 1 — 2x~2 and Q(x) = —- = x3.)
I Q(x)I(x) dx + C
dy
/ (x)— + I (x)P(x)y = I (x)Q(x)
dx
4- U U)y]
dx
= 1 (x)g(x)
_ Substitute.
/ (x)y I(x)Q(x)dx + C
■/
As usual, the constant of integration
C is added after the last integration.
1
y
/(x) I I(x)Q(x) dx + C □
7.6 First-Order Differential Equations 463
dy
— + P(x)y = Q(x)
dx
by I(x) makes the left side an exact derivative, the function I(x) is called an integrat¬
ing factor of the differential equation.
A first-order initial value problem involves a first-order equation and the value of
y at a particular value x = x0. Here is an example of such a problem.
Solve cll. — e~x -2y,x > 0, subject to the initial condition y — 2 when x = 0.
dx
Solution
A graphical solution (using technology) is found by looking at the direction field as
shown in Figure 7.5.
y \ ITT-rrrrr TT
-4- _\ II 1 i Lin 11 i i
\ \ \ t \ i \ \ i l 1 i i
\ \ \ \ V \ \ \ \ \ \
> \ \ \ \ \ \ 1 \ \ \ \ \ \
v. \ > \ \ k \ N \ \ \ \
^ —1
i r
-2 , - r / / ,2/ /y / / yf //
-1 -
/ / / / 1111/ / / / /
-M y—/—jf- /—/ M
1 / 11 / / | / / / / i i
1 / /MM 1 1 i i
1 r f till 1 1 rt
-LI II M l.U 1 J u
To find an analytic solution, we use Theorem 7.1. The differential equation can be
expressed in first-order linear form by adding 2y to both sides:
We have P(x) = 2 and Q(x) = e~x. Both P(x) and Q(x) are continuous in the
domain x > 0. An integrating factor is given by
We now use the first-order linear differential equation theorem, where / (x) = e2x and
Q(x) = e~x, to find y.
J e2*e~x dx + C
—2x
for x > 0
= T7 [«* + C] =
dy Or
x——|-y — xe , x > 0
ax
Solution
As a first step, divide each term in the given equation by x so that we can use integrating
factors, as in Theorem 7.1.
£
ax
+ x2 = .*
For this equation, P(x) = - and Q(x) — e2x. The integrating factor is
1
-xe2x -~e2x + C See Example 5, Section 7.2.
x 2 4
1 2x ^ C
= -e2x-F -
2 4x x
Modeling Logistic Growth When the population <2(0 of a colony of living organ¬
isms (humans, bacteria, etc.) is small, it is reasonable to expect the relative rate of
change of the population to be constant. In other words,
dQ
dQ
-^-=k or = kQ
Q dt
where k is a constant (the unrestricted growth rate). This is called exponential
growth. As long as the colony has plenty of food and living space, its population will
obey this unrestricted growth rate formula and Q(t) will grow exponentially.
However, in practice, there often comes a time when environmental factors begin
to restrict the further expansion of the colony, and at this point, the growth ceases to be
purely exponential in nature. To construct a population model that takes into account
the effect of diminishing resources and crowding, we assume that the population of the
species has a “cap” B, called the carrying capacity of the species. We further assume
that the relative rate of change in population is jointly proportional to the current popu¬
lation <2(0 and the difference B — Q, which can be thought of as the potential unborn
population of the species. That is.
dQ
dt dQ
= k(B - Q) or = kQ(B - Q)
Q dt
This is called a logistic equation, and it arises not only in connection with population
models, but also in a variety of other situations. The following example illustrates one
way such an equation can arise.
7.6 First-Order Differential Equations 465
Solution
Let <2(0 denote the number of residents who have been infected by time t and B the
total number of susceptible residents. Then the number of susceptible residents who
have not been infected is B - Q, and the differential equation describing the spread of
the epidemic is
d_Q
— kQ(B — Q) k is the constant of proportionality.
dt
dQ
— k dt Use separation of variables.
Q(B-Q)
dQ
( dQ =[ k dt
J Q(B-Q)
Q(B-Q) J
The integral on the right causes no difficulty, but the integral on the left requires partial
fractions:
1 1 1 1
<2(5-Q) 5 lQ + B-Q]
We now substitute the form on the left into the equation and complete the integration:
dQ
+ \l B-Q ~Ikdl
1 Integrate each term.
- - In \B — Q — kt T C
B B
Q Division property of logs
-In = kt + C
B B-Q
Multiply both sides by B
Q ) = Bkt + BC (note that Q > 0, B > Q).
B-Q)
Q _ Bkt+BC Definition of In
B-Q
Because eBC is a constant,
Q = (B- Q)eBkteBC
we can let A\ = eBC.
Q = A\BeBkt - A\QeBkl
Q + Ai<2^' = AxBeBkt
A\BeBkt
Q =
1 + A\eBkt
1
To simplify the equation, we make another substitution; let A — —, so that (after
BeBkt
B
Q = ,Bkt -Bki
1 + Ae
1 +
Figure 7.6 A logistic curve:
The graph of Q(t) is shown in Figure 7.6. Note that the curve has an inflection point
Q(t) = 1 + Ae~Bkl
where
£><0 = y
466 Chapter 7 Methods of Integration
since
d2Q
kQ(-l) + k(B-Q)
~di2
k(B -2Q)
B
0 when Q = —.
2
This corresponds to the fact that the epidemic is spreading most rapidly when half the
susceptible residents have been infected.
Note also in Example 3 that y = Q(t) approaches the line y = B asymptotically
as / —> +oo. Thus, the number of infected people approaches the number of those
susceptible in the long run. ■
)0 is the initial population, and k is a constant of proportionality. If Q(t) has a limiting value, let it be
denoted by B.
dQ
Uninhibited growth — =kQ Exponential growth; interest
(k > 0) dt
compounded continuously;
inflation; price/supply curves
Rate is proportional to the Solution:
amount present.
Q = Qoekl
L-curve
dQ
1 Uninhibited decay — =kQ Radioactive decay; depletion
dt
A
Cto (k < 0) of natural resources;
price/demand curves; electric
Rate is proportional to the Solution:
Q condenser discharge
amount present.
Q = Qoeh
5-curve
dQ
Inhibited (logistic) ~tt — kQ(B — Q) Long-term restricted
growth at
population growth; spread of a
disease in a population; growth
Rate is jointly proportional to Solution:
of a business; spread of a
the amount present and to the
B rumor
difference between the amount
present and the maximum Q ~ 1 + Ae~Bkt
amount possible B (called the where A is an arbitrary
carrying capacity). constant
dQ
Limited growth -j~ = k(B — Q) Learning curve; diffusion of
(k > 0) at
information by mass media;
amount of drug retained after
Rate is proportional to the Solution:
intraveneous injection;
difference between the amount
Newton’s law of cooling; sales
present and a fixed limiting Q = B - Ae~k'
of new products; growth of a
value B. where A is an arbitrary
business
constant.
7.6 First-Order Differential Equations 467
Solution
Let S(t) denote the amount of salt in the tank at the end of / minutes. Because 3 gal of
Inflow rate of salt: 6 lb/min
brine flows into the tank each minute and each gallon contains 2 lb of salt, it follows
that (3)(2) = 6 lb of salt flows into the tank each minute (see Figure 7.7). This is the
inflow rate.
For the outflow, note that at time /, there are 5(/) lb of salt and 50+(3 -2)/ gallons
of solution (because solution flows in at 3 gal/min and out at 2 gal/min). Thus, the
concentration of salt in the solution at time / is
S(t) lb/gal
50 + /
and the outflow rate of salt is
50 +1
5(0 r i 25(0 „ , .
Figure 7.7 Rate of flow equals lb/gal [2 gal/minj - gQ + f lb/min
50 + /
inflow rate minus outflow rate.
Combining these observations, we see that the net rate of change of salt dS/dt is given
by
or
dS 2S
= 6
dt 50 +/
which we recognize as a first-order linear differential equation with
1 — [2(50 + /)3 + C]
(50 +
C
= 2(50 + /) +
(50 + /)2
To evaluate C, we recall that there are 20 lb of salt initially in the solution. This
means 5(0) = 20, so that
C
5(0) = 2(50 + 0) +
(50 + 0)2
20=100+^
—80(502) = C
Thus, the solution to the given differential equation, subject to the initial condition
5(0) = 20, is
80(50)2
5(f) = 2(50 + f) —
(50+ /)2
Specifically, at the end of 1 hour (60 min), the tank contains
80(50)2
5(60) = 2(50 + 60)-—— ~ 203.4710744
(50 + 60)"
The tank contains about 200 lb of salt.
R (ohms)
EMF, Current L (henrys)
m© 7(0 / (amperes)
V (volts)
Inductor
8885 L
The EMF source, which is usually a battery or generator, supplies voltage that causes
a current to flow in the circuit. According to Kirchhoff’s second law, if the circuit is
closed at time t = 0, then the applied electromotive force is equal to the sum of the
voltage drops in the rest of the circuit. It can be shown that this implies that the current
I(t) that flows in the circuit at time t must satisfy the first-order linear differential
equation
dl
L— + RI — E
dt
where L (the inductance) and R (the resistance) are positive constants.* Write
dl R E
—-b —I — —
dt L L
R E
to see that P{t) = — (a constant) and Q(t) = —. In the case where E is a constant,
we have
= e — (R/L)l | J eWL>' dl + C
= e -(R/L)t - • -«<*"■>' + C
L R
- — _l Ce~(R/L)t
R +
*It is common practice in physics and applied mathematics to use I as the symbol for current. Of course,
this has nothing to do with the concept of integrating factor introduced earlier in this section.
7.6 First-Order Differential Equations 469
/ = §(
~ ~R
E
This means that, in the long run, the current I must approach —. The solution of the
differential equation consists of two parts, which are given special names:
E
— is the steady-state current.
E
-e~Rt,L is the transient current.
R
Figure 7.9 The current in an
RL circuit with constant EMF Figure 7.9 shows how the current I(t) varies with time t.
dy „ dy (2x + 1 \ _2x
I ' II il I I I I
h i 1 11 I 1 I I
/1 II I I i l I
,3 l l I Ii lI
5. x-j- + 2y = xe i i i i i Til iwK-XJi
dx 6-z, + {-r)y=‘ •// j I I I I II I I
/1 4 i-M* » u i i
1' \N- l N
i i! \ \ y / ( m ii ii i i
• r' i'
\- v vr; j ! 1r n»1r 1i
\ yrTrrf ■11 xKxijiiiiTTTT
„ dy y -i dy 2y \nx II \ \ y /1; i i i m m i nxk\/1 ii 1111 l
7. — + - = tan Jt 8. + — =- -tl-vhl'FfFHITFF
■
-f- . |. vi \. y U /- FT. 1. 4-1. M I- J-
dx x x n t \ y / i m ji ii i i M \ k\/ I I I I I I I I
dx x l \ v s U / I I i i i ii i • 11 \ K\/ 1111 11 III
dy \ ■ ~ 0 In Problems 15-18, find the orthogonal trajectories of the given
dy
9. — + (tanx)y = sinx 10. — + (secx)y = sin2x
dx dx family of curves. Recall from Section 5.6 that a curve is an or¬
thogonal trajectory of a given family if it intersects each member
In Problems 11-14. solve the initial value problems. A graphical
of that family at right angles.
solution is shown as a check for your work.
15. the family of parabolas y2 = 4kx
dy xy dy 2xy 16. the family of hyperbolas xy = c
11. ——h -- = *(1 +x) 12 — -l-— = sin x
dx 1 +x dx 1+x2 17. the family of circles x2 + y2 = r2
with y = — 1 when x = 0 with y = 1 when x = 0 18. the family of exponential curves y = Ce~x
/ / / / / / Itt '\\ !\A' tl' It'I \\ N\ // >/ \\ |
/ ii /1/ /1/ /. * y 'I ; Modeling Problems. In Problems 19-30, set up and solve an ap¬
y->.- >. a / / / a 2.!/ IV-t-A
■ \ - ' I1 1 1 i / \ \ A I / :/ A P ■ t \ A t It t K / / \ I
A \ Iw 1/ /!/ < V«*_ t \ t i i i k ' k \ I propriate model to answer the given question. Be sure to state your
/ f i 1 1 1 / \ \ / i ii/ yy trwrlrTS
V \ t t tw/t
/ y vi
growth rate from 1992 to 2000 was 4.0% per year. Predict 30. The Motion of a Body Falling through a Resisting Medium A body of
the GDP in 2002. Consult an almanac to see if this prediction mass m is dropped from a great height and falls in a straight
is correct. line. Assume that the only forces acting on the body are the
21. According to the Department of Health and Human Services, earth’s gravitational attraction mg and air resistance kv. (Re¬
the divorce rate in 1990 in the United States was 4.7% and call g = -32 ft/s2.)
there were 1,175,000 divorces that year. How many divorces a. According to Newton’s second law,
will there be in 2005 if the divorce rate is constant?
dv
22. According to the Department of Health and Human Services, m — = mg — kv
the marriage rate in 1990 in the United States was 9.8% and dt
there were 2,448,000 marriages that year. How many mar¬ Solve this equation, assuming the object has velocity
riages will there be in 2005 if the marriage rate is constant? Vo — 0 at time t = 0.
23. A tank contains 10 lb of salt dissolved in 30 gal of water. Sup¬ b. Find the distance 5(f) the body has fallen at time t. Assume
pose 2 gal of brine containing 1 lb of dissolved salt per gallon s = 0 at time t = 0.
runs into the tank every minute and that the mixture (kept uni- c. If the body weighs W = 100 lb and k = 0.35, how long
' form by stirring) runs out at the same rate. does it take for the body to reach the ground from a height
a. Find the amount of salt in the tank at time t. of 10,000 ft?
b. How long does it take (to the nearest second) for the tank Recall from Section 5.6 that a tank filled with water drains at the
to contain 15 lb of salt? rate
24. In Problem 23, suppose the tank has a capacity of 100 gal and
dV r-
that the mixture flows out at the rate of 1 gal/min (instead of — = -4.8A0V6
2 gal/min). dt
a. How long will it take for the tank to fill? where h is the height of water at time t (in seconds) and Ao is the
b. How much salt will be in the tank when it is full? area (in ft2) of the drain hole. This formula, called Torricelli’s law,
25. The rate at which a drug is absorbed into the blood system is is used in Problems 31 and 32.
given by 31. A full tank of water has a drain with area 0.07 ft2. If the tank
has a constant cross-sectional area of A = 5 ft2 and height of
ho — 4 ft, how long does it take to empty?
32. A full tank of water of height 5 ft and constant cross-sectional
where b(t) is the concentration of the drug in the bloodstream area A = 3 ft2 has two drains, both of area 0.02 ft2. One drain
at time t. What does b(t) approach in the long run (that is, as is at the bottom and the other at height 2 ft. How long does it
t -> Too)? At what time is b(t) equal to half this limiting take for the tank to drain?
value? Assume that b(0) = 0. 33. The Euler Beam Model For a rigid beam with uniform loading,
26. An RL circuit has a resistance of R ohms, inductance of L the deflection y(x) is modeled by the differential equation
henries, and EMF of E volts, where R, L, and E are constant. y<4) = —k, where k > 0 is a constant and x measures the
Suppose no current flows in the circuit at time t = 0. If L distance along the beam from one of its ends. Assume that
is doubled and E and R are held constant, what effect does y(0) = y(L) = 0, where L is the length of the beam, and that
this have on the “long run” current in the circuit (that is, the y"(0) = y"(L) = 0.
current as t —>■ Too)? a. Solve the beam equation to find y(x).
27. The 1990 census recorded a population of 16.1 million His- b. Where does the maximum deflection occur? What is the
panics, while in 2000 the figure was 32.8 million. Assuming maximum deflection?
that the rate of population growth is proportional to the popu¬
c. Suppose the beam is cantilevered, so that y(0) = y(L) = 0
lation, predict the Hispanic population in the year 2005.
and y"(0) = y'(L) = 0. Now where does the maximum
28. A population of animals on Catalina Island is limited by the deflection occur? Is the maximum deflection greater or less
amount of food available. Studies show there were 1,800 ani¬ than the case considered in part b? Would you expect the
mals present in 1980 and 2,000 in 1986 and suggest that 5,000 graph of the deflection y — y(x) to be concave up or con¬
animals can be supported by the conditions present on the is¬ cave down on [0, LJ? Prove your conjecture.
land. Use a logistic model to predict the animal population in
34. Modeling Problem A tank initially contains 5 lb of salt in 50 ft3
the year 2005.
of solution. At time t — 0, brine begins to enter the tank at the
29. In 1986 the Chernobyl nuclear disaster in the Soviet Union rate of 2 ft3/h, and the mixed solution drains at the same rate.
contaminated the atmosphere. The buildup of radioactive ma¬ The brine coming into the tank has concentration
terial in the atmosphere satisfies the differential equation
C(t) = 1 - e-°-2' lb/ft3
dM
/
M = 0 when t = 0
~df t hours after the dilution begins.
a. Set up and solve a differential equation for the amount of
where M — mass of radioactive material in the atmosphere
salt S(t) in the solution at time t.
after time t (in years); k is the rate at which the radioactive
material is introduced into the atmosphere; r is the annual de¬ b. How much salt is eventually in the tank (as 1 —> Too)?
cay rate of the radioactive material. Find the solution, M(t), c. At what time is S(t) minimized? What is the minimum
of this differential equation in terms of k and r. amount of salt in the tank?
7.6 First-Order Differential Equations 471
35. Modeling Problem Two 100-gallon tanks initially contain pure a. If / (f) = 0 (the patient intakes only water), the amount of
water. Brine containing 2 lb of salt per gallon enters the first substance Q(t) in the patient’s body at time t satisfies
tank at the rate of 1 gal/min, and the mixed solution drains
dQ
into the second tank at the same rate. There, it is again thor¬ = 0 — R(t)
oughly mixed and drains at the same rate, 1 gal/min. dt
a. Set up and solve a differential equation of the amount of Solve this equation for Q(t).
salt Si(t) in the first tank at time t (minutes). b. Suppose the patient intakes the substance at a constant rate
b. Set up and solve a second differential equation for the for part of the day. Specifically,
amount of salt S2(t) in the second tank at time t.
10.4 g/h for 10 < t < 20
c. Let S(t) — Si — S2. Intuitively, S(t) > 0 for all t. At what I(t) =
time is the excess S(t) maximized? What is the maximum
excess?
1°
otherwise
/ + P(x)y = Q(x)yn
find P(t).
38. Modeling Problem Uranium-234 (half-life 2.48 x 105 yr) decays where n is a real number, n/0,n/l.
to Thorium-230 (half-life 80,000 yr). a. Show that the change of variable u = y'~" transforms such
a. If U(t) and T{t) are the amounts of uranium and thorium an equation into one of the form
at time f, then
u\x) + (1 - n)P(x)u(x) = (1 - n)Q(x)
dU dT
—k\U, —k2T + k\U
It ~dt This transformed equation is a first-order linear differen¬
tial equation in u and can be solved by the methods of this
Solve this system of differential equations to obtain U (?) section, yielding a solution to the given Bernoulli equation,
and T(t).
b. Use the change of variable suggested to solve the Bernoulli
b. If we start with 100 g of pure U-234, how much Th-230 equation
will there be after t = 5,000 yr?
39. Solve the differential equation
dS „ / S\dN pS2
and the rate of intake is I(t) for 0 < / < 24. The patient’s
body contains 200 g of the substance when t = 0.
77 = -pS+(nj1T + '^n
472 Chapter 7 Methods of Integration
In this equation S(t) is the number of people at age t that are 47. An object of mass m is projected upward from ground level
susceptible to smallpox; N(t) is the number of people at age with initial velocity Vo against air resistance proportional to its
t who survive; p is the probability of a susceptible person get¬ velocity v(t). Thus,
ting the disease, and 1/m is the proportion of those who die
from the disease. ■ dv
m— = —mg — kv
Let y — S/N, and solve the resulting equation for y(t). dt
Then write N as a function of S. (Hint: You will need the
a. Solve this equation for v(t), and then find s(t), the height
result of Problem 44.)
of the object above the ground at time t.
46. Consider a curve with the property that when horizontal and
b. When does the object reach its maximum height? What is
vertical lines are drawn from each point P(x, y) on the curve
the maximum height?
to the coordinate axes, the area A, under the curve is twice the
area Ai above the curve, as shown in Figure 7.10. c. Suppose k = For an object weighing 20 lb with ini¬
tial velocity v0 = 150 ft/s, what is the maximum height?
When does the object hit the ground?
d. Suppose the same object as in part c is launched with the
same initial velocity but in a vacuum, where there is no air
resistance (k = 0). Would you expect the object to hit the
ground sooner or later than the object in part c? Prove your
conjecture.
48. A body of mass m falls from a height of s0 ft against air resis¬
tance modeled to be proportional to the square of its velocity
v, so that
dv 2
Figure 7.10 Problem 46 m— = mg — kv
dt
Show that x and y satisfy the differential equation a. Solve this differential equation to find v(t). Then find the
dy^ _ y_ height 5 (t) of the object at time t (in seconds).
dx 2x b. If the object weighs 1,000 lb, 50 = 100 ft, and k — 0.01,
Solve the equation and characterize the family of all curves how long does it take for the object to hit the ground? In¬
that satisfy the given geometric condition. terpret this result.
where the integrand f(x) is bounded. In this section, we extend the concept of integral
to the case where the interval of integration is infinite and also to the case where /
is unbounded at a finite number of points on the interval of integration. Collectively,
these are called improper integrals.
In physics, economics, probability and statistics, and other applied areas, it is useful
to have a concept of an integral that is defined on the entire real line or on half-lines
of the form x > a or x < a. If /(x) > 0, the integral of / on the interval x > a
can be thought of as the area under the curve y = f(x) on this unbounded interval,
as shown in Figure 7.11. A reasonable strategy for finding this area is first to use a
Figure 7.11 Finding the area under a definite integral to compute the area from x = a to some finite number x = N and
curve on an unbounded region then to let N approach infinity in the resulting expression. Here is a definition.
7.7 Improper Integrals 473
pN
Improper Integrals (First Type) Let a be a fixed number and assume / /(x) dx exists for all N > a. Then if
Ja
pN p+oo
p-rOO ply
r \1
11
Solution
, i
y ■
/ i\
\
r x: The region under the curve y = 1/x2 for x > 1 is shown in Figure 7.12. This region is
i > unbounded, so it might seem reasonable to conclude that the area is also infinite. Begin
—i—
i * by computing the integral from 1 to N:
i
j
i rN N
l dx 1
I + 1
i
Jl
~~2
x N
Figure 7.12 Graph of y — Let us consider the situation more carefully, as shown in Figure 7.13.
7 i
!
| T '
1
1 J. i
i
i J i
i i 1 i
1 i
\ A rea is >/3j . ' 1
0.8-
i
_1 Ar ja s 9 io[ U.b-
Area is( 19/1 00. _L„
f\ 1
n
1
1
t 1
1 i l
U.2
i i
i 1 J_l^
IT I1 I1 11
! 1 V,
i 1 i r 1 1 i i i 7
i 3 * 2 10 1 2 1 4 1 ‘.'1 lC 20 60 40 5 0 6( 70 80 90 100
For N = 2, 3, 10, or 100, this integral has values j, and respectively, sug¬
gesting that the region under y = \/x2 for x > 1 actually has finite area approaching
1 as N gets larger and larger.
To find this area analytically, we take the limit as N -+ +oo, as shown in Figure
7.14.
r^=
J\
lim rd4
x2 N^+ooJi X2
i N
= lim [—x !1
At—>+oo L J
= lim \-N~l + ll
N—*+oo L J
+°° dx 'N dx
— = lim
l X N-y+oo i; X
N
= lim In |jc 11,
N-S-+00
= lim In N — +oo
N—y+oo
The limit is not a finite number, which means the improper integral diverges
+°° dx
We have shown that the improper integral — converges and that the im-
Figure 7.15 The area to the right of the
line x — 1 bounded by the curve r+°° dx
y = \/x and the x-axis is infinite.
proper integral / — diverges. In geometric terms, this says that the area to the
J1 x
right of x = 1 under the curve y = 1 /x2 is finite, whereas the corresponding area
under the curve y = 1 /jc is infinite. The reason for the difference is that as x increases,
1/x2 approaches zero more quickly than does l/x, as shown in Figure 7.16.
Solution
For this example, remember that p is a constant and x is the variable. We have seen in
Example 1 that the integral converges for p = 2. We also know from Example 2 that
the integral diverges for p = 1, so we can assume p ^ 1 in the following computation:
N
~P+1 1
[+°° dx f'N dx
— = lim = lim [N l~p
Figure 7.16 An unbounded region may Xp N-y+oo ~P+ 1. j N-^y+oo 1 — p
+°° dx 1 1 1
— = lim —— IN1-? - 1] = ———[—1] = ——r
/1
r XP
Xr iv—y-rw 1
N^y+oo i — p
p L 1 —
P
p P
p — 1
■+0° dx 1
— lim — - 1] = +oo
i; Xp N-y+oo 1
Thus, the improper integral diverges for p < 1 and converges for p > 1.
Evaluate
Solution
rN
lim / x e 2a dx By parts:
N-++00 Jq
u dv Let m = x and dv = e~2x dx
du = dx v =
+ o + i_
I’Hopital's rule
2 n->+oo \ 2e2N
1
4
EXAMPLE 5 Gabriel's horn: a solid with a finite volume but infinite surface
area
Gabriel’s horn is the name given to the solid formed by revolving about the x-axis the
unbounded region under the curve y — j for x > 1. Show that this solid has finite
volume but infinite surface area.
Solution
We will find the volume, V, by using disks, as shown in Figure 7.17
r+oo rN / 1
V = n / (x-1)2 dx — tx lim x 2 dx = tx lim (-1-1 It
Jl N-*+oo /i
N^+ooJl A?->+oo \ N
+0°
du — tx lim
/ N
y/u2 +\
du
=2nL --—
2u- N-> + 00
N
— yju2 + 1
= tx lim + In (u + yju2 + l) Formula 150
A^—>+oo
We can also define an improper integral on an interval that is unbounded to the left
or on the entire real number line.
476 Chapter 7 Methods of Integration
Improper Integrals Let He a fixed number and assume fix) dx exists for all t < b. Then if
(First Type, Extended)
lim / f{x) dx exists, we define the improper integral
r^-oo J,
/b
f(x)dx= lim /
rb
f(x)dx
-00
L f{x)dx and
f
J — OC
converge for some number a, the improper integral of fix) on the entire x-axis
f(x)dx
is defined by
/-(-oo
fix)dx= /
pa
f(x)dx + I
p-\-co
fix)dx
-00 J —OO Ja
Solution
a.
1
b. Since yix) — yi~x), we see that y = is symmetric about the v-axis, and it
1 + x2
is reasonable to conjecture that the integral in part b is twice the integral in part a.
We have
CN dx
+ lim /
N-^+oc J0 1 +x2
7T
-2M V
= lim
M ->+oo
~ 4
If / is unbounded at c and a < c < b, the Riemann integral f(x)dx is not even
defined (only bounded functions are Riemann integrable). However, it may still be
possible to define f(x) dx as an improper integral in certain cases.
Let us examine a specific problem. Suppose
1
fix) = —^ for 0 < x < 1
■s/X
fix) = -7=
y/X
1
Figure 7.19 Graph of y = is continuous on every interval [r, 1] for t > 0, as shown in Figure 7.19. For any such
■s/x
interval [t, 1], we have
1
x~l/2 dx — 2^/x^ = 2 — 2y/t
[X dX r-
lim / —= - lim (2 — 2-Jt) — 2
0+ J, s/x t-*0+
478 Chapter 7 Methods of Integration
This is called a convergent improper integral with value 2, and it seems reasonable
to say
Iipproper Integral If / is unbounded at a and / fix) dx exists for all t such that a < t < b, then
(Second Type)
nb pb
/ f(x)dx= lim / f(x)dx
Ja <^a+ Jt
If the limit exists (as a finite number), we say that the improper integral con¬
verges; otherwise, the improper integral diverges. Similarly, if / is unbounded
at b and / f(x)dx exists for all t such that a <t < b, then
If / is unbounded at c, where a < c < b, and the improper integrals /flc /(*) dx
and fcb fix) dx both converge, then
pb pc rb
/ f{x)dx= I f(x)dx+ / f(x)dx
Ja Ja Jc
We say that the integral on the left diverges if either (or both) of the integrals on
the right diverges.
^ What This Says If a function is continuous on the open interval {a, b) but
is unbounded at one of its endpoints of the interval, then replace that endpoint by
t, evaluate the integral, and take the limit as t approaches that endpoint. On the
other hand, if / is continuous on the interval [a, b], except for some c in (a, b)
where / has an infinite discontinuity, rewrite the integral as
/•b pc pb
I f(x)dx= / f(x)dx+ / f(x)dx
Ja Ja Jc
You will then need limits to evaluate each of the integrals on the right.
Solution
fix) = ix - 1)~2/3 is unbounded at the right endpoint of the interval of integration
and is continuous on [0, /] for any t with 0 < t < 1. We find that
r1 dx r1 dx
= lim [3Qc - l)l/3
Jo = r-iT- 7o (X - 1)2/3
7.7 Improper Integrals 479
That is, the improper integral converges and has the value 3.
Solution
Because sec x is unbounded at the left endpoint | of the interval of integration and is
continuous on [t, re] for any t with | < t < n, we find that
L 71/2
secx dx = lim
t->(7t/2) + f
sec x dx — lim
r->(7r/2)+
In |secx + tanx|
Solution
WARNING . ...
A common mistake is to The integral is improper because the integrand is unbounded at x = 2. If the improper
fail to notice the discontinuity at x = 2. integral converges, we have
It is WRONG to write
f
Jo
(x — 2)—1 dx = f
Jo
(x — 2)-1 dx + f
J2
(x — 2)~l dx
(x — 2) 1 dx = In |x — 2| |n
I, — lim [ (x — 2)—1 dx T lim [ (x — 2)~] dx
= In 1 — In 2 t-*2~ Jo t^2+ Jt
= - In 2 If either of these limits fails to exist, then the original integral diverges. Because
proper integrals.
r+0° dx r+0° dx
9. 10.
In Problems 3—46, either show that the improper integral con¬
verges and find its value, or show that it diverges.
Ji 2x - 1 Ji ^2x - 1
r+oo
C+°° dx
r+°° dx 11. 12. / dx
3. 4- Ji (2x - l)2 Jo
n 11 ^
r+0° dx r+°° dx /»+oo /•+00
/•+00
x2 dx r+oc x2dx 50. Suppose that an oil well produces P(t) thousand barrels of
16.
15+ (x3 + 2)2 i, x3 +2 crude oil per month according to the formula
r»+oo
x2 dx r+°° 2 P(t) = lOOe-0 02' - lOOe
-0.1/
18. / xe_Jr dx
17'/ Vx3 + 2 Jo
where t is the number of months the well has been in produc¬
f+OO
tion. What is the total amount of oil produced by the oil well?
19. [ —— dx 20. / xe~x dx
*Jx Jo
r+0° lnx dx 1
21.
I 5xe10~x dx 22.
J1 x 51. Let /(x) =
—
1
forx > 1
23.
r dx
x In x
24.
26
r+0° dx
x3e * dx
^ex+l forx < — 1
Sketch the graph of / and evaluate
25.
i:
x2e * dr
•1
+0° x dx
f +00
/ (x) dx
27,
L 2a: dx
x2 + 1
28
4
(1+x2)2
dx
52. Find all values of p for which /
r*+00
dx
converges, and
f dx J2 x(lnx)/7
29 30 find the value of the integral when it exists.
-oo "%/2 "a ’•/_'-00 (5 - *)2
f+OO
+0° dx 53. Find all values of p for which f — converges, and find the
31 xe-1*1 dx 32
f
-00 •/_ X2 + 1 Jo xp
value of the integral when it exists.
33. f dx
xW
34
dx
Xsfx. 54. Find all values of p for which
/
,1/2
x(ln x)p
dx
converges, and
+00 3x dx
37. 38 CA dx —
—11 1
/_;^ "Loo (3x2 + 2)3 = —[1 — (—1)] = —2
J-i *2 *
/•+00
39.
r *+00
lnx dx
dx
40
'■I
/■' x dx
a
lnx dx Is the calculation correct? Explain.
56. Journal Problem College Mathematics Journal* ■ Peter Lind-
41.
/ x(lnx)2
42
■l -
strom of North Lake College in Irving, Texas, had a student
who handled an oo/oo form as follows:
43.
/ g-(l/2)lnx
71/3 c'“^2
rn/3 sec2 x dx
44>•/
o
77/2
e* + e-*
sinxdx
/+CO
(x — l)e~x dx — I
/*
+0O 1
x - 1
ex
dx
48. Find the area of the unbounded region between the x-axis and 57. Find / /(x) dx, where
the curve
y =-- for x < 2
J (x - 4)3 for 0 < x < 1
Vx3
49. The total amount of radioactive material present in the atmo¬ fix) =
sphere at time T is modeled by for 1 < x < 2
\/(x - l)3
J* +0O
Pe~r' dt
58. Evaluate the improper polar integral 6e~e dO.
l 1
where P is a constant and t is the number of years. Suppose a 59. Find the total area between the spirals r — e 20 and r = e 50
recent United Nations publication indicates that, at the present for 9 > 0.
time, r = 0.002 and P = 200 millirads. Estimate the total fu¬
ture buildup of radioactive material in the atmosphere if these
values remain constant. *Vol. 24, No. 4, September 1993, p. 343.
7.8 Hyperbolic and Inverse Hyperbolic Functions 481
© The Laplace transform of the function f is defined by the improper a. £ [t3e 2,\ b. £ (e 31 cos 2f}
integral
'+00
63. If F(s) = £{/(f)}, show that jC{ea,f(t)} = F(s — a). y' + 3y = sin 2 f
64. Use the result in Problem 63, along with those of Problem with y(0) — 2.
60, to find the following Laplace transforms and inverse trans¬ c. Solve the initial value problem in part b by using inverse
forms. Laplace transforms to obtain y.
HYPERBOLIC FUNCTIONS
In physics, it is shown that a heavy, flexible cable (for example, a power line) that is
suspended between two points at the same height assumes the shape of a curve called
a catenary (see Figure 7.20), with an equation of the form
y = \ (eX,a + e~x/a)
This is one of several important applications that involve sums of exponential func¬
tions. The goal of this section is to study such sums and their inverses.
In certain ways, the functions we will study are analogous to the trigonometric
functions, and they have essentially the same relationship to the hyperbola that the
trigonometric functions have to the circle. For this reason, these functions are called
hyperbolic functions. Three basic hyperbolic functions are the hyperbolic sine (de¬
Figure 7.20 The hanging cable problem
noted sinh;t and pronounced “cinch”), the hyperbolic cosine (coshx; pronounced
“kosh”), and the hyperbolic tangent (tanhx; pronounced “tansh”). They are defined
as follows.
482 Chapter 7 Methods of Integration
Hyperbolic Functions
Graphs of the three basic hyperbolic functions are shown in Figure 7.21.
The list of properties in the following theorem suggests that the basic hyperbolic
functions are analogous to the trigonometric functions.
cosh2 x — sinh2 x = 1
sinh(-A) = - sinh* (sinh a: is odd)
WARNING . .
A ma|or aitterence cosh(—x) — cosh a (cosh a is even)
between the trigonometric and tanh(—x) = — tanh a (tanha: is odd)
hyperbolic functions is that the
sinh(A + y) = sinh x cosh y + cosh at sinh y
trigonometric functions are periodic,
but the hyperbolic functions are not. cosh(A + y) = cosh a cosh y + sinh a- sinh y
Proof We will verify the first identity and leave the others for the reader.
r\
cosh a — sinh a =
m’-m'
e2x+2 + e~2x e2x - 2 + <r2*
□
TECHNOLOGY NOTE
4 4
Some of the software packages show
these in simplified form as: There are three additional hyperbolic functions: the hyperbolic cotangent (coth a),
the hyperbolic secant (sechA), and the hyperbolic cosecant (csch a). These functions
e2x + 1 are defined as follows:
cothA =
e2-1 - 1
1 ex + e x
COthA = -;-
2ex tanh a ex — e~x
sechA = y2.X
+ 1 2
sechA = —-—
2ex cosh A ex + e~x
csch x = - 1 2
2* - 1
csch A =-
sinh a ex — e x
7.8 Hyperbolic and Inverse Hyperbolic Functions 483
Rules for differentiating the hyperbolic functions are listed in Theorem 7.3.
Proof Each of these rules can be obtained by differentiating the exponential func¬
tions that make up the appropriate hyperbolic function. For example, to differentiate
sinhx, we use the definition of sinhx:
d . d e* — e~
— (sinhx) — —
dx dx
— iex _ ie (— 1) = \{ex + e x) = cosh.
The proofs for the other derivatives can be handled similarly. □
Solution
a. We have
d d
— (cosh Ax) = sinh(Ax) — (Ax) = A sinh Ax
dx dx
b. We find that
Proof The proof of each of these formulas follows directly from the corresponding
derivative formula. -1
7 cosh1 x sinh x dx
b7 x seclr(x ) dx
7 tanhx dx
Solution
r\ r\
f x sech (x ) dx
-! sech2(x2)(x dx)
Let u = x2; du = 2x dx
— J sech2 u du
= i tanh m + C
= i tanhx2 + C
sinhx dx
c. J tanh x dx = J cosh x
Let w = coshx; du = sinhx dx
=f — = In |u| + C = ln(coshx) + C
u
Inverse hyperbolic functions are also of interest, mainly because they enable us to
express certain integrals in simple terms.
Because sinhx is continuous and strictly monotonic (increasing), it is one-to-one
and has an inverse, which is defined by
y = sinh-1 x if and only if x = sinh y
for all x and y. This is called the inverse hyperbolic sine function, and its graph is
obtained by reflecting the graph of y = sinhx in the line y = x, as shown in Figure
7.22. Other inverse hyperbolic functions are defined similarly (see Problem 60).
Because the hyperbolic functions are defined in terms of exponential functions, we
may expect to be able to express the inverse hyperbolic functions in terms of logarith¬
The graph of y = sintrhc mic functions. We summarize these relationships in the following theorem.
is obtained by reflecting
y = sinh x in the line y = x,
THEOREM 7.5 Logarithmic formulas for inverse hyperbolic functions
Figure 7.22 The graph of
y = sinh-1 x
sinh-1 x = ln(x + \/x2 + 1), all x csch 1 x = In ( - H--7— ] , x / 0
1 + y/\ — X2
cosh-1 x = ln(x + s/x1 - 1), x > 1 sech 1 x = In ,0 < x < 1
1 1 + X -i 1 x + 1
tanh x = - In --, |x| < coth x = - In--, |x| >
2 1 —x 2 x — 1
7.8 Hyperbolic and Inverse Hyperbolic Functions 485
Proof We will prove the first formula and leave the next two formulas for you to
verify (see Problem 61). The other formulas are proved similarly. Let y = sinh-1 x;
then its inverse is
x — sinh y
= ^-e-n
ey
ely — 2xey —1=0 Quadratic formula with
a = \,b — —2x, c — — 1
v 2x ± V4x2 + 4
e- =-
2
= x ± \/x2 + 1
Because ey > 0 for all y, the only solution is ey = x + s/x2 + 1, and from the definition
of logarithms,
y = ln(x + \!x2 + 1) □
1 du du
— (sinh 1 u) —
ax Vl T u2 dx
1 du
I W+u2
du
= sinh u fC
—(cosh 1 u)
./A Vu2 - 1 dx
1 du
/ —r:
sfu2^T
du
- COSh U + C
/
.
— (tanh 1 u) — --r — tanh 1 u + C
dx 1 — m2 dx 1 — u1
d -1 du du
—(csch
dx
u) =
|m| Vl + u2
du
dx / --j=.r = — csch
W V 1 + M2
|w I + C
/
-1
—— (sech 1 u) — —. = — sech MmI + C
dx M Vl — M2 dx mV 1 - m2
d 1 du C?M ,
— (coth
dx
u) —
1 — u2 dx / --r
1 — M^
= COth 1 M + C
Proof The derivative of each inverse hyperbolic function can be found either by dif¬
ferentiating the appropriate logarithmic function or by using the definition in terms of
hyperbolic functions. We will show how this is done for y — sinh-1 x and then leave
it for you to apply the chain rule for u, a differentiable function of x. By definition of
inverse, we see
dx
x = sinh v and — - cosh y
dy
Thus,
d . ^_i \ dy if 1 1
sTsi x) - Tx ~ 77 - jst; - /,+sinh= - ytTF
dy y
since cosh2 y — sinh2 y = 1 and sinh v = x. The integration formula that corresponds
to this differentiation formula is
f dx
/ . — sinh 1 x + C
J Vl + x2
The other differentiation and integration formulas follow similarly.
486 Chapter 7 Methods of Integration
1 d
b. [cosh '(secx)] = — (sec x)
dx V sec2 x — 1
sec x tan x
= —= secx ■
V tan2 x
= ln(x + yjX2 + 1)
0
= ln(l + V2)
C seclr(lnx) ufx 53. a. Verify that y = a cosh cx + b sinh ex satisfies the differen¬
32. 33. J coth x dx
J x tial equation
f dx
34. 35. f ^ 0
J V4x2 + 16 J y/912 - 16
r dt r cos xdx b. Use part a to find a solution of the differential equation
36. 37.
J 36- 1612 J y/\ + sin2.
y" — Ay = 0
r xdx C x2 dx
38. 39.
J V1 + x4 J 1-x6 subject to the initial conditions y(0) = 1, y'(0) = 2.
r'/2 dx f3 dx 54. Find the volume V generated when the region bounded by the
40. 41.
curves y = sinhx, y = coshx, the y-axis, and the line x = c
Jo l-*2 A l-x2
(c > 0) is revolved about the x-axis. For what value of c does
r' t5 dt f ex dx
V = 1?
42. 43.
Jo Vl +t'2 J\ y/e2x — 1 x
55. Find the length of the catenary y = a cosh — between x = —a
Mn2 rl a
44. / sinh 3x dx 45. / xsech2x2 and x = a.
Jo Jo 56. Find the volume of the solid formed by revolving the region
46. Show that
bounded by the curve y = tanhx on the interval [0, 1] about
tanh x + tanh y
a. tanhfx + y) =- the x-axis.
1 + tanh x tanh y
57. Find the surface area of the solid generated by revolving the
b. sinh 2x = 2 sinhx coshx
curve y = coshx on the interval [—1. 1] about the x-axis.
c. cosh 2x = cosh2 x + sinh2 x
Q 58. Prove the following formulas
sinh'1 x
47. Evaluate lim a. sech2 x + tanh2 x = 1 b. coth2 x — csch2 x = 1
-t-*+oo cosh -i■ x
59. Derive the differentiation formulas for coshn, tanhn, and
48. Show that sech «, where u is a differentiable function of x.
a. — 1 < tanh x < 1 for all x 60. First give definitions for cosh-1 x, tanh-1 x, and sech-1 x, and
b. lim tanhx = 1 and lim tanhx = —1 then derive the differentiation formulas for these functions by
*-*•4-00 *->—00
using these definitions.
49. Determine where the graph of y = tanhx is rising and falling
61. Prove
and where it is concave up and concave down. Sketch the
graph and compare with Figure 7.21c. a. cosh-1 x = ln(x + Vx2 — 1), x > 1
1 1+x
50. Sketch the graph of y — cothx. Be sure to show key features b. tanh x = - In-, |x| < 1
such as intercepts, high and low points, and points of inflec¬ 2 1 —x
tion. 62. Show that
51. First show that coshx + sinhx = ex, and then use this result to dx . , x „
prove that (sinh x + cosh x)n = cosh nx + sinh nx for positive
integers n.
/ r. = sinh
s/x2 + a2
—b C
a
2
Xz V
2
1
/ yh2
b2 with x > a > 0.
11. State the differentiation and integration formulas for the in¬ dx f dx
10 11.
verse hyperbolic functions. •/ (x2 4- x + l)3/2 I y/x(l 4- f/x)
\j9x2 — 1 dx
PRACTICE PROBLEMS 12 13. 1 x2 tan-1 x dx
12. Evaluate each of the given numbers.
•/ X
dx
14 15. j" ex \J4 — e2* dx
a. tanh-1(0.5) b. sinh(ln3)
Find the integrals in Problem 13-18.
c. coth-,(2)
•/ sin x 4- tan x
n7 v/l +1/*2
X X
2x + 3 16. / cos - sin - dx dx
13
J Vx2 + 1
dx
147 x sin 2x dx •/ 2 3
sinx dx
15. J sinh(l — 2x)dx •/ cos5 x
19 . JvT + sinx dx
17.
, J
f__
(x2 + l)(x - 1) «-!£h
20.
// cosx ln(sinx) dx 217 sin(lnx) dx
_ f sinhx dx
Evaluate the definite integrals in Problems 19-22. 22. J e sechfe ) dx 23. j
2 + coshx
21.
f4
i3
dx
2x - x2
22.
»tt/4
y=
V9
1
/ r
sec5xtm2xdx
dx
39. / ——-——
J
on the interval [0, 2] is revolved about the y-axis.
29. Solve the first-order linear differential equation 40
/ sinxifx
--
sin x + cos x
_
41.
f
/ -
X1/2 +X2/3
cosx + sinx
J 1 + cos x — sin x
dy , xy f dx C e l dx
dx
-I-- = e
x + 1
42
TT 44 J gx/l — gx/t
subject to the condition y = 1 when x = 0. In Problems 44^48, solve the given differential equations,
30. A tank contains 200 gal of saturated brine with 2 lb of salt dy /x2 - 1 dy 1 -y
per gallon. A salt solution containing 1.3 lb of salt per gal¬ 44. 45.
dx y2 + 1 dx 1 +x
lon flows in at the rate of 5 gal/min, and the uniform mixture
flows out at 3 gal/min. Find the amount of salt in the tank dy 1
46. 47. y' + (tanx)y = sec3 x
after 1 hour. dx 2x x'/2 + 1
48. (cos3 x)y' = sin3 x cos y
Supplementary Problems Determine whether the improper integral in Problems 49-52 con¬
dy verges, and if it does, find its value.
Find the derivative — in Problems 1-5.
dx 77/2 cos xdx __ f+00 dx
49. f
1. y — tanh-1 - 2. y — x cosh *(3x 4- 1) Jo Vsinx ■L 00 4 + x2
x +00
_ sinh x
3. y = -
6.
■ S^'xdx
/ cos-* x dx 7.
?7i=
f xne~ax dx
IsHht 97
9. I ~-A dx
x3 - 2x2
54. Find
n sin x dx
V cos 2x
Chapter 7 Review 489
region bounded by the curve y = coshx, the y-axis, and the f cos" x dx
78. Derive a reduction formula for
line y = 2. J sin x
66. The region bounded by the graph of y = sinx and the x-axis
79. Derive a reduction formula for / x"(x2 + a2)~1/2 dx.
between x = 0 and x = n is revolved about the x-axis to
generate a solid. Find the volume of the solid and its surface 80. Derive the reduction formula
area.
dx —\/ax 4- b (2n — 3)a r dx
*L. L. Thurstone, “The Learning Function,” J. of General Psychology 3 It. s/ax + b (n - 1 )bxn~1 (2n - 2)b J x”—1 *J~ayx + b
(1930), pp. 469^193.
490 Chapter 7 Review
1
f (x + l)(x- + X + 1)
if it exists. Hint: Let that is,
dp
u — —.
= k(D - S)
Jt
83. Spectroscopic measurements based on the Doppler effect yield
an observed mass mo for a certain type of binary star. The true Suppose for a certain commodity, demand is linear,
mass m is then estimated by m — mo//, where
D(p) = c — pd
/•tt/2
I sin4 x dx and supply is cyclical
= L '
Determine the number I. S(t) = a sin(bt)
W dv
— — = W - B-kv
f
io
Jo 1 + (tanx)'72
g dt
90. Putnam Examination Problem Evaluate
where g is the constant acceleration due to gravity. Solve
this equation to express v(t) in terms of W, k, B, and g.
b. Integrate the expression for v(t) in part a to find s(t).
l V^pf-lSSS (, + !)'
»+oo
dt
Buoy Design
This project is to be done in groups of three or four students. Each group will submit a
single written report.
Y ou have been hired as a special consultant by the U.S. Coast Guard to evaluate
some proposed new designs for navigational aids (buoys).
The buoys are floating cans that need to be visible from some distance away, with¬
out rising too far out of the water. Each buoy has a circular cross section and will be
fitted with a superstructure that carries equipment such as lights and batteries.
To be acceptable, a fully equipped buoy must float with not less than 1.5 ft nor
more than 3 ft of freeboard. (Freeboard is the distance from the water level to the top
of the device.)
You should make the following assumptions:
a. A floating object will displace a volume of water whose weight equals the weight
of the floating object. (This is Archimedes’ principle.)
b. Your devices will be floating in salt water, which weighs 65.5500 lb/ft3.
c. The buoys will be constructed of ~ in.-thick sheet metal that weighs 490 lb/ft3.
d. You can estimate an additional 20% in weight attributable to welds, bolts, and the
like.
e. Each buoy will be fitted with a superstructure and equipment weighing a total of
2,000 lb.
f. Each design you are given to evaluate will be presented to you in the form of a curve
The BOOK OF NATURE is written in
x = f(y), to be revolved about the y-axis.
the language of mathematics.
-Galileo
This project is adapted from a computer project used at the U.S. Coast Guard Academy.
491
*
CONTENTS
8.1 Sequences and Their Limits
Sequences
Infinite Series
Limits of sequences
Bounded, monotonic sequences
8.2 Introduction to Infinite Series; Geometric
Series
Definition of infinite series
General properties of infinite series
Geometric series PREVIEW
Applications of geometric series Is it possible for the sum of infinitely many nonzero numbers to be finite? This
8.3 The Integral Test; p-Series concept, which may seem paradoxical at first, plays a central role in mathematics
Divergence test and has a variety of important applications. The goal of this chapter is to examine the
Series of nonnegative numbers; the theory and applications of infinite sums, which will be referred to as infinite series.
integral test Geometric series, introduced in Section 8.2, are among the simplest infinite series
p-series
we will encounter and, in some ways, the most important. In Sections 8.3-8.6, we
8.4 Comparison Tests will develop convergence tests, which provide ways of determining quickly whether
Direct comparison test or not certain infinite series have a finite sum. Next, we will turn our attention to
Limit comparison tests
series in which the individual terms are functions instead of numbers. We will be
8.5 The Ratio Test and the Root Test especially interested in the properties of power series, which may be thought of as
Ratio test polynomials of infinite degree, although some of their properties are quite different
Root test
from those of polynomials. We will find that many common functions, such as ex,
8.6 Alternating Series; Absolute and Infix: T 1), sinx, cosx, and tan-1 x, can be represented as power series, and we
Conditional Convergence will discuss some important theoretical and computational aspects of this kind of
Alternating series test
representation.
Error estimates for alternating
series
Absolute and conditional
convergence PERSPECTIVE
Summary of convergence tests for
series Series, or sums, arise in many different ways. For example, suppose it is known
Rearrangement of terms in an that a certain pollutant is released into the atmosphere at weekly intervals and is
absolutely convergent series dissipated at the rate of 2% per week. If m grams of pollutant are released each
8.7 Power Series week, then at the beginning of the first week, there will be Si = m grams in the
Convergence of a power series atmosphere, and at the beginning of the second, there will be 0.98m grams of “old”
Term-by-term differentiation and pollutant left plus m grams of “new” pollutant, to yield a total of S2 = m + 0.98m
integration of power series grams. Continuing, at the beginning of the nth week there will be
8.8 Taylor and Madaurin Series S„ = m + 0.98m + (0.98)2m T • • • T (0.98)B_1m
Taylor and Madaurin polynomials
Taylor’s theorem grams. It is natural to wonder how much pollutant will accumulate in the “long run”
Taylor and Madaurin series (as n -* Too), and the answer is given by the infinite sum
Operations with Taylor and
Madaurin series S+OQ = m T 0.98m T (0.98)2m T • • •
Chapter 8 Review But just exactly what do we mean by such a sum, and if the total is a finite number,
Group Research Project: Elastic Tightrope how can we compute its value? We seek answers to these questions in this chapter.
493
494 Chapter 8 Infinite Series
The making of a motion picture is a complex process, and editing all the film into a
movie requires that all the frames of the action be labeled in chronological order. For
example, R21-435 might signify the 435th frame of the 21st reel. A mathematician
might refer to the movie editor’s labeling procedure by saying the frames are arranged
in a sequence.
SEQUENCES
^ 15 @2 5 • • • ? &n ? • • •
{@n }
The number an is called the general term of the sequence. We will deal only with
infinite sequences, so each term an has a successor an+l and for n > 1, a predecessor
an-1- For example, by associating each positive integer n with its reciprocal -, we
Yl
obtain the sequence denoted by
Solution
It is sometimes useful to start a If n = l,thenai = (|)* = 1. Similarly,
sequence with a0 instead of a^.
The reverse question, that of finding a general term given certain terms of a se¬
quence, is a more difficult task, and even if we find a general term, we have no assur¬
ance that the general term is unique. For example, consider the sequence
2, 4, 6, 8,...
This seems to have a general term a„ — 2n. However, the general term
an = (n -l){n- 2){n - 3)(n - 4) + In
has the same first lour terms, but = 34 (not 10, as we would expect from the
sequence 2, 4, 6, 8).
So far, we have been discussing the concept of a sequence informally, without a
definition. We have observed that a sequence {an} associates the number a„ with the
positive integer n. Hence, a sequence is really a special kind of function, one whose
domain is the set of all positive (or possibly, nonnegative) integers.
8.1 Sequences and Their Limits 495
Sequence A sequence {a,,} is a function whose domain is a set of nonnegative integers and
whose range is a subset of the real numbers. The functional values a\, a2, ai,...
are called the terms of the sequence, and an is called the nth term, or general
term, of the sequence.
LIMITS OF SEQUENCES
It is often desirable to examine the behavior of a given sequence {an} as n gets arbi¬
trarily large. For example, consider the sequence
L = lim an
n-+oo
(Note that for simplicity, the limiting behavior is denoted by n -» 00 instead of the
usual n -> +00.) For instance, in our example we would expect
n
lim -= 1
n->oo n + 1
b. Graphing a sequence in two
dimensions
This limiting behavior is analogous to the continuous case (discussed in Section 4.4),
Figure 8.1 Graphing the sequence and may be defined formally as follows.
n
an —
n + 1
Convergent Sequence The sequence {a,,} converges to the number L, and we write
L — lim an
n->oo
+ What This Says The notation L = lim an means that eventually the terms
n —> 00
of the sequence {an} can be made as close to L as may be desired by taking n
sufficiently large.
496 Chapter 8 Infinite Series
“5 "4 fl3 “2 al A geometric interpretation of this definition is shown in Figure 8.2. Note that the
* » *(-->- numbers an may be practically anywhere at first (that is, for “small” n), but eventually,
L-e L L+e
an the an must “cluster” near the limiting value L.
a. one dimension The theorem on limits of functions carries over to sequences. We have the follow¬
ing useful result.
Root rule for sequences lim ’tfaf — a/l provided Ufa is defined for
n—>oo n
Figure 8.2 Geometric interpretation
all n and \/L exists
of a converging sequence
Proof The proof of these rules is similar to the limit rules that were stated in Section
4.4. □
3 n4 + n — 1
5n4 + 2 n2 + 1
Solution
a. As n grows arbitrarily large, 100/n gets smaller and smaller. Thus,
1 4-
2n2 + 5n - 7 _ 2 5 7
l n3 n n2 n2
(
[) N-
and by using the linearity rule, we find that
(
f) 5-
'
0 4- <
n 7- ( 2n2 + 5n — 1 2 1 1
lim ---= lim - + 5 lim — — 7 lim —
__ n-nx) nJ n-too Yl n->-oo n-+ oo n2
: 7 1 3 "j
=0+0+0
c. Divide the numerator and denominator by ft4, the highest power of n that occurs in
the expression, to obtain
1 1
3ft4 + ft — 1 ^ „3 „4 3
lim —---- = lim -%-= -
n^-oo 5ft4 + 2ft2 + 1 n-K>o 2 1 5
^ ' 9 ' Z
nL nr
Note that this procedure is very similar to the one we used in Section 4.4 when we
evaluated limits of functions to infinity. A graph of this sequence is shown in Figure
8.5. ■
Solution
a. The sequence defined by {(-1)"} is -1, 1,-1, 1-- and this sequence diverges by
oscillation because the ftth term is always either 1 or —1. Thus an cannot approach
one specific number L as n grows large. The graph is shown in Figure 8.6a.
ft5+ft +2 n2 n5
b. lim —:-z—- = lim ----T
n-*oo 7ft4 + ft2 + 3 n-»oo 7 13
I o I c
n nJ nD
The numerator tends toward 1 as n -> oo, and the denominator approaches 0.
Hence the quotient increases without bound, and the sequence must diverge. The
graph is shown in Figure 8.6b.
Y = (X5+X3+2)/(7X4+X2+3)
2.5
X
6 .8774009
7 1.017379 2.0
8 1.158078
9 1.299255
10 1.440766
11 1.58252
12 1.724458
13 1.866536
14 2.008725
15 2.151003
16 2.293352
17 2.435762
18 2.578221 x
19 2.720721 -H—I-1—I- H—V -t-h H
20 2.863257 4 8 12_ 16_
n5 + n3 + 2
b. an
lnA + ft2 + 3
If lim an does not exist because the numbers a„ become arbitrarily large as
n—»oo
n _> oo, we write lim an = oo. We summarize this more precisely in the follow-
n-+oc
ing box.
Limit Notation lim an = oo means that for any real number A, we have an > A for
n-+ oo
all sufficiently large n.
\
lim bn = — oo means that for any real number B, we have bn < B for
n-> oo
all sufficiently large n.
498 Chapter 8 Infinite Series
n5 + n3 + 2
lim = oo
n —> oo In4 + n2 + 3
Y = Vx2 + 3X - X
Determine the convergence or divergence of the sequence j Vn2 + 3n - n j.
x =E3 y
2.5-
Solution
X Y
6 1.348469 Consider
7 1.3666
8 1.380832
9
10
1.392305
1.401754
1.5 lim (y/n2 + 3n — n)
11 1.409674
n—>oo
12 1.416408
* 13
' 14
1.422205 It would not be correct to apply the linearity property for sequences (because neither
1.427249
15 1.431677
16 1.435596 lim \/n2 + 3n nor lim n exists). It is also not correct to use this as a reason to sav
17 1.439089 n—► oo n—>oo J
18 1.442222
19
20
1.445048 4 8 12 16 x that the limit does not exist. You might even try some values of n (shown in Figure
1.447611
8.7) to guess that there is some limit. To find the limit, however, we will rewrite the
general term algebraically as follows:
Figure 8.7 Graph of
an = Vn2 + 3/i — «; convergent
V nl + 3n — n = (y nz + 3n
n) - —--
or divergent?
Vn2 + 3n + n
1
n2 + 3n — n2 „
\/n2 + 3n + n 1
n
y
L. y/n2 + 3 n
+ 1 ' n2 + 3n
+ 1 1 +-+ 1
n
Hence,
3
1.0- lim
n—>oo
(a/n2 + 3n — n) lim
tt—»oo 2
1 + - + 1
n
__ Note the graph of the sequence in Example 4 in Figure 8.7. The graph of a se¬
<
quence consists of a succession of isolated points. This can be compared with the
L
graph of y = Vx1 + 3x — x, x > 1, which is a continuous curve (shown in Figure
Figure 8.8 Graph of 8.8). The only difference between lim an — L and lim f(x) = L is that n is required
/(X) = y/x2 + 3x — X, X > 1 to be an integer. This is stated in the hypothesis of the following theorem.
Proof Let e > 0 be given. Because lim fix) = L, there exists a number N > 0
*-XX)
such that
Be sure you read this theorem correctly. In particular, note that it does not say that
if lim an = L, then lim fix) = L (see Figure 8.9b).
n->oo x-*oo
8.1 Sequences and Their Limits 499
Solution
_
Theorem 8.2 tells us that lim _ is the same as lim fix), provided this latter
n—»oo 1 — en n-+ 00
limit exists. Since the function / is continuous for all * 7^ 0, we use I’Hopital’s rule.
2x
lim - lim 'Hopital's rule
x—nx 1 — ex x —> 00 —e
-0
Proof The proof of this theorem follows the same steps as the proof of the squeeze
rule given in Appendix A.
1
, r Inn In a: ~
= 0
Thus, L = e° = 1.
b. We cannot use THopitaTs rule because xl is not defined as an elementary function
when x is not an integer. Instead, we use the squeeze rule for sequences. Accord¬
n! n (n — 1) (n — 2) ■ • • 1
n'1 n • n • n • • n
n — 1 n — 2
0
<
oo-oc,
Because lim an — 0 and lim cn — 0, it follows from the squeeze rule that
n—>oo n—>oo 1
lim
►oo nn
We introduce, along with a simple example, some terminology associated with se¬
quences {an}.
Strictly increasing Q\ < d2 < ■ ■ < Clk-1 < dk < ■ • ■ 1,2, 3, 4, 5,...
Proof A formal proof of the BMCT is outlined in Problem 53. For the following
informal argument, we will assume that {an} is an increasing sequence. You might
wish to see whether you can give a similar informal argument for the decreasing case.
Because the terms of the sequence satisfy a\ < ci2 < ay < ■ • ■, we know that the
sequence is bounded from below by a\ and that the graph of the corresponding points
(n, a„) will be rising in the plane. Two cases can occur, as shown in Figure 8.11.
y‘ y
M
i
L
i • «
e
■
€ t
« I
— h- — “3“
•
—
h~ a2-
■~ h “
_
j 7 i ... i x
a. If a„
< M for n = 1,2, ....the graph b. If\an) is not bounded from above,
of the points (n, an) will approach a the graph rises indefinitely.
horizontal “barrier” line y = L.
Suppose the sequence {a,,} is also bounded from above by a number M, so that
ax < an < M for n — 1,2,.... Then the graph of the points (n, an) must continually
rise (because the sequence is monotonic) and yet it must stay below the line y = M.
The only way this can happen is for the graph to approach a “barrier” line y = L
(where L < M), and we have lim an — L, as shown in Figure 8.1 la. However, if the
n —> oo
sequence is not bounded from above, the graph will rise indefinitely (Figure 8.1 lb),
and the terms in the sequence {an} cannot approach any finite number L.
Solution
The first few terms of this sequence are
1 1-3 3 1-3-5 _ 5
ai~2 ai ~ 2~4 _ 8 a3~2-4-6~16
13 5
Because - > - > —, it appears that the sequence is decreasing (that is, it is mono-
2 8 16
tonic). We can prove this by showing that an+\ < an for all n > 0, or equivalently.
502 Chapter 8 Infinite Series
ttn +1
< 1. (Note that a„ ^ 0 for all n.)
a„
1 • 3 • 5 • • • [2(n + 1) 1]
&n +1 2 • 4 • 6 • • • [2(n + 1)]
a. 1 • 3 - 5 - - - (2w — 1)
2 • 4 • 6 • ■•(2n)
1 3 • 5 • • ■ (2n + 1) 2 • 4 • 6•••(2n)
2 • 4 • 6 • • (2n + 2) ' 1 • 3 • 5 • • • (2n - 1)
2n + 1
< 1
2n + 2
for any n > 0. Hence n„+i < an for all n, and {an} is a decreasing sequence. Because
an > 0 for all n, it follows that {«„} is bounded below by 0. Applying the BMCT, we
see that {an} converges, but the BMCT tells us nothing about the limit. ’ ■
The BMCT also applies to sequences whose terms are eventually monotonic. That
is, it can be shown that the sequence {an} converges if it is bounded and there exists an
integer N such that [an] is monotonic for all n > N. This modified form of the BMCT
is illustrated in the following example.
Y = (In X/Vx)
Show that the sequence | converges.
y *
1 10000000000
1
Solution
0.8-
20
30
0.6698661
0.6209708
A.
We will apply the BMCT. A few values are shown in Figure 8.12. The succession
40 0.5832631 0.6- * • .
50
60
0.5532436
0.5285776 »
•
• •
of numbers suggests that the sequence increases at first and then gradually begins to
70
30
0.5077923
0.4899255
# •
decrease. To verify this behavior, we let
90 0.4743216 0.4-
100 0.460517
1000 0.2184424
10000 0.0921034
100000
1000000
0.0364071
0.0138155
0.2-
tt \ = ln*
fix)
10000000 0.005097
10000000000 0.0002303
—i—i—i—i—i—i—i—i—i—
123456789x
and find that
Thus, x = e2 is the only critical value, and you can show that f'(x) < 0 for x > e2.
This means that / is a decreasing function for x > e2. Thus, the sequence J J must
be decreasing for n > 8 (because <?2 is between 7 and 8). We see that the sequence
Inn | .
—= 1 is bounded from below because
yfn J
Inn
0 < for all n > 2
yfn
Therefore, the given sequence is bounded from below and is eventually decreasing, so
it must converge. _
8.1 Sequences and Their Limits 503
The BMCT is an extremely valuable theoretical tool. For example, in Section 2.4,
we defined the number e by the limit
„ra(i+iy=,
n->oo y n)
but to do so we assumed that this limit exists. We can now show this assumption is
warranted, because it turns out that the sequence {(1 + £)"} is increasing and bounded
from above by 3 (see Problem 80 in the Supplementary Problems). Thus, the BMCT
assures us that the sequence converges, and this in turn guarantees the existence of the
limit. We end this section with a result that will be useful in our subsequent work. The
proof makes use of the formal definition of BMCT.
Proof The case where r = 0 is trivial. We will prove the theorem for the case
0 < r < 1 and leave the case — 1 < r < 0 as an exercise.
lim r" = L
n —> oo
lim rrn = rL
n—>oo
But lim r"+l = lim rn = L by the definition of limit. Hence L = r L < L, which is
n—>oo n —> oo
impossible, so lim rn =0. -1
n—>oo
5. n
6. Inn sin——
sin
20 .
2n
21.
8n — 500 y/n I
n +1 fn 2n + 800y/n J
3n + 1 n2
7. 8. n- + n 22.
3 yfn
23.
Inn
n T2 n2
5 sfn + Zfn
9. {a,,} where a\ — 256 and a„ = fan _i for n >2
24. {25/n} 25. {n3>n\
10. {«,,} where ci\ = — 1 and an = n + o„_i for n > 2 3
26. 1 27. {(n + 4)1/"}
11. {o,,} where ci\ — 1 and a„ = («„_i)2 + a„_i + 1 for n > 2 + -
n
504 Chapter 8 Infinite Series
' \
3n 4n + 5
38t
n
39.
5 8
&4b 4>
V t * ♦ V
3n — 7 Same pair
40. 41. {^7} (rabbits never die)
2"
53. Proof of the BMCT Suppose there exist a number M and a pos¬ show that if e > 0 is given, there exists a positive integer
itive integer N so that an < an+\ < M for all n > N. N so that
a. It is a fundamental property of the real numbers that a A - e < an < A
sequence with an upper bound must have a least upper
for all n > N.
bound; that is, there must be a number A with the prop¬
erty that an < A for all n but if c is a number such that b. Show that \a„ — A\ < e for all n > N and conclude that
c < A, then am > c for at least one m. Use this property to lim an = A.
n-A-oo
If it is possible to attach a numerical value to the infinite sum, we would expect the
partial sums Sn — a\ + a2 + ■ ■ ■ + an to approach that value as n increases without
bound. These ideas lead us to the following definition.
S,i — 01 + 02 +•••+ On = Ok
k=1
The series is said to converge with sum S if the sequence of partial sums (S„)
converges to 5. In this case, we write
Ecik = lim Sn = S
n-+oc
k=1
ok
k= 1
What This Says An infinite series converges if its sequence of partial sums
converges and diverges otherwise. If it converges, its sum is defined to be the
limit of the sequence of partial sums.
506 Chapter 8 Infinite Series
ak
k=t Kk= 1
and the symbol ^ ak is used to represent both the series and its sum.
k=1
We will also consider certain series in which the starting point is not 1; for exam¬
ple, the series
111 OO j OO ^
Solution
This series has the following partial sums.
1
51 = - Note: 1-- =
2
1 1 3
52 = —I— -
2 4 4
p 1117
2 4 8 8
1 1 1
^"_2 + 4_l-^ 2"
The sequence of partial sums is §, jf, §, g, {§, • • •, and in general (by mathe¬
matical induction)
Sn = 1 - —
2n
Because hm_ ^1 — — ^ = 1, we conclude that the series converges and its sum is
S = 1. ■
Solution
The series can be expanded (written out) as
X^(-l)* = —1 + 1 — 1 + 1 — 1 + 1-
k= 1
and we see that the nth partial sum is
—1 if n is odd
S„ =
0 if n is even
Because the sequence {5,,} has no limit, the given series must diverge.
8.2 Introduction to Infinite Series; Geometric Series 507
Solution
Using partial fractions, we find that
1 1 1 1
= T +
k2 + k k(k +1) k k+ 1
Thus, the nth partial sum of the given series can be represented as follows:
1 1
k=l k= 1 L k k+ l
1 1
- ri-iW fi-TU + ••• +
n n + 1
1 1 1 1 1 1
= 1+ -- + - + -T +» +•••+ —n + n n + 1
= 1 -
n + 1
The limit of the sequence of partial sums is
lim Sn - lim = 1
n-^-oo n—>• oo n + 1
Y ak instead of £ dk
k=1
Proof Compare this with the linearity property in Theorem 5.4. In Chapter 5, the
limit properties are for finite sums, and this theorem is for infinite sums. The proof of
this theorem is similar to the proof of Theorem 5.4, but in this case it follows from the
linearity rule for sequences (Theorem 8.1). The details are left as a problem.
Solution
oo 1 00 |
Because we know from Examples 1 and 3 that ^ —- and ^ — both converge
k=l ^ ^ k= 1 ^
with sum 1, the linearity property allows us to write the given series as
OO i OO i
^k^+k
k=1 1 k=1
We can now conclude that the series converges and that the sum is
OO j OO |
4Ej^-6E^=4<>)-6(l) = -2
*=i k=1
The linearity property also provides useful information about a series of the form
T,(cak +dbk) when exactly one of Hak or Tjbk diverges; that is, one converges and the
other diverges.
Proof Suppose Eak diverges and Y,bk converges. Then if the series T,(ak + bk) also
converges, the linearity property tells us that the series
£[(«* + bk) ~ h] = J2ak
must converge, contrary to hypotheses. It follows that the series H{ak + bk) diverges.
□
oo r
1
This theorem tells us, for example, that ^ + (-ir must diverge be-
k2 + k
k= 1
OO | OO
cause even though ^ ^ converges (Example 3), the series ^(—1)* diverges
*=i k=1
(Example 2).
GEOMETRIC SERIES
We are still very limited in the available techniques for determining whether a given
series is convergent or divergent. In fact, the purpose of a great part of this chapter is
to develop efficient techniques for making this determination. We begin this quest by
considering an important special kind of series.
Geometric Series A geometric series is an infinite series in which the ratio of successive terms in
the series is constant. If this constant ratio is r, then the series has the form
OO
The geometric series ^^ark with a ^ 0 diverges if |r| > 1 and converges if |r| < 1
k=0
with sum
E ar
00
k
— -
a
I — K
k=0
Proof Note that the nth partial sum of the geometric series is
Sn = a + ar + ar2 + • • • + arn~l
Thus,
rSn = ar + ar2 + ar3 + • • • + arn Multiply both sides by r
If |r | > 1, the sequence of partial sums {5,,} has no limit since r" is unbounded in this
case, so the geometric series must diverge. However, if |r| < 1, Theorem 8.5 tells us
that rn -* 0 as n —> oo, so we have
o° / „n i \
o(0- 1)
Eark — lim Sn = lim a ( —
n—KX n—>oo \ r r - 1
a
1 -r
k=0 v
To complete the proof, it must be shown that the geometric series diverges when
|r| = 1, and we leave this final step as a problem. J
1
b- E3 -
k=0 k=2
Solution
a. Because r — \ satisfies |r| > 1, the series diverges.
b. We have r — — ^ so |r| < 1, and the geometric series converges. The first value of
k is 2 (not 0), so the value a (the first value) is a = 3 (—|) = and
r a 1
E3
k =2
— r
25
-(-*) To
Geometric series can be used in many different ways. Our next three examples illus¬
trate several applications involving geometric series.
Recall that a rational number r is one that can be written as r = p/q for inte¬
ger p and nonzero integer q. It can be shown that any such number has a decimal
representation in which a pattern of numbers repeats. For instance,
Solution
The bar over the 23 indicates that this block of digits is to be repeated, that is,
15.423 = 15.423232323- The repeating part of the decimal can be written as a
geometric series as follows.
23 23 23
15.423 = 15 + — +
10 lO3 + Tq5 + 1Q7 +
4 23
= 15 + — +
10 103 1 + 7^ + 75i +
23 1
15 4“ ~ + TTT Sum of a geometric series
10 IQ1 1 - 100 with a — 1 and r = 1/100
23 100 '
= 15 + — +
10 103 ~99
23
= 15 + — +
10 990
15,269
~ 990
A tax rebate that returns a certain amount of money to taxpayers can result in
spending that is many times this amount. This phenomenon is known in economics
as the multiplier effect. It occurs because the portion of the rebate that is spent by
one individual becomes income for one or more others, who, in turn, spend some of it
again, creating income for yet other individuals to spend. If the fraction of income that
is saved remains constant as this process continues indefinitely, the total amount spent
as a result of the rebate is the sum of a geometric series.*
Solution
The amount (in billions) spent by original recipients of the rebate is 40. This becomes
new income, of which 90%, or 0.9(40), is spent. This, in turn, generates additional
spending of 0.9[0.9(40)], and so on. The total amount spent if this process continues
indefinitely is
40
.1 — 0.9,
400 (billion)
*The fraction .v of income that is saved is called the marginal propensity to save and c — 1 — .v is the marginal
propensity to consume. In Example 7, we have .9 =0.1 and c — 0.9.
8.2 Introduction to Infinite Series; Geometric Series 511
Solution
Of the original dose of 10 units, only 10e-1/5 are left in the patient’s body after the first
day (just prior to the second injection). That is,
Si = 10e_1/5
The medication in the patient’s body after 2 days consists of what remains from the
first two doses. Of the original dose, only 10e~2/5 units are left (because 2 days have
elapsed), and of the second dose, lO^-1/5 units remain:
S2 - 10e-1/5 + 10e~2/5
The amount S of medication in the patient’s body in the long run is the limit of Sn as
n —> oo. That is,
S = lim S„
n—>oo
oo
Geometric series with
= \0e-k/5
a = 10e-1/5 and r = e_1/5
k= 1
10e~^
~ \ — g-i/5
45.166556
WARNING
As a final note, remember that a sequence is a succession of terms, whereas a
series is a sum of such terms. Do not confuse the two concepts because they have very different
properties. For example, a sequence of terms may converge, but the series of the same terms
may diverge:
3 5 9 17
is the sequence which converges to
Vx
i \ 3
2
5
4
9
8
n— I '
"•S-H) 28. £
*=1 (2k — l)(2k + 1)
47. A flywheel rotates at 500 rpm and slows in such a way that
during each minute it revolves at two-thirds the rate of the pre¬
2k + 1 ceding minute. How many total revolutions does the flywheel
»• E- k2(k + l)2 make before coming to rest?
^Zv/F+T-Vk ^— r- ,_ 48. A ball is dropped from a height of 10 ft. Each time the ball
30. ) -- —. Hint: Note that y/k2 + k — y/k y/k + 1. bounces, it rises 0.6 the distance it had previously fallen. What
is the total distance traveled by the ball?
k- 1 Ak B(k + 1)
2*+i “ 2 *
2*~ 2*+1 49. Suppose the ball in Problem 48 is dropped from a height of h
feet and on each bounce rises 75% of the distance it had pre¬
k 1 viously fallen. If it travels a total distance of 21 ft, what is
b. Evaluate f
1k+\ as a telescoping series. hi
k= 1 A
2* - 1 50. Suppose that nationwide approximately 92% of all income is
36. Evaluate ^ j Hint: Follow the procedure in
spent and 8% is saved. How much total spending will be gen¬
*=1
Problem 35. erated by a $50 billion tax cut if savings habits do not change?
-n+\ 51. Suppose that a piece of machinery costing $10,000 depreci¬
In
(n + D" ates 20% of its current value each year. That is, the first year
37. Evaluate E Hint: Use the properties of log- $10,000(0.20) = $2,000 is depreciated. The second year’s
„=i »(« + D
depreciation is $8,000(0.20) = $1,600, because the value for
arithms to show that the series telescopes.
the second year is $10,000 — $2,000 = $8,000. If the de¬
00 n preciation is calculated this way indefinitely, what is the total
38. Evaluate E -———. Hint: Express as a telescoping series.
*= 1 depreciation?
8.2 Introduction to Infinite Series; Geometric Series 513
52. Winnie Winner wins $100 in a pie-baking contest run by the 58. His ; a pro¬
Hi-Do Pie Co. The company gives Winnie the $100. How¬ fessor of mathematics at the Univer¬
ever, the tax collector wants 20% of the $100. Winnie pays sity of Warsaw, Waclaw Sierpinski
the tax. But then she realizes that she didn’t really win a $100 was the first person to give a lecture
prize and tells her story to the Hi-Do Co. The friendly Hi-Do course devoted entirely to set theory.
Co. gives Winnie the $20 she paid in taxes. Unfortunately, He became interested in set theory
the tax collector now wants 20% of the $20. She pays the tax in 1907 when he came across a
again and then goes back to the Hi-Do Co. with her story. As¬ theorem which stated that points in
sume that this can go on indefinitely. How much money does the plane could be specified with a
the Hi-Do Co. have to give Winnie so that she will really win single coordinate. He wrote to a WACLAW SIERPINSKI
$100? How much does she pay in taxes? friend and asked how such a result 1882-1969
53. A patient is given an injection of 20 units of a certain drug could be possible, and he received
every 24 hr. The drug is eliminated exponentially, so the part a one word reply: "Cantor." Georg Cantor (1845-1918)
that remains in the patient’s body after t days is /(f) = e~,/2. wrote a series of eight papers that provided the basis for what
If the treatment is to continue indefinitely, approximately how we today call set theory. ■
many units of the drug will eventually be in the patient’s body For this quest, we investigate an interesting geometrical
just before an injection? figure named after Sierpinski, namely the Sierpinski gasket,
54. A certain drug is injected into a body. It is known that the shown in Figure 8.13.
amount of drug in the body at the end of a given hour is ^ the
amount that was present at the end of the previous hour. One
unit is injected initially, and to keep the concentration up, an
additional unit is injected at the end of each subsequent hour.
What is the highest possible amount of the drug that can ever
be present in the body?
55. Each January 1, the administration of a certain private college
adds six new members to its board of trustees. If the fraction of
trustees who remain active for at least t years is /(f) = e~02',
approximately how many active trustees will the college have
in the long run? Assume that there were six members on the
board on January 1 of the present year. Figure 8.13 Sierpinski gasket
56. How much should you invest today at an annual interest rate To form the gasket, begin by drawing an equilateral triangle,
of 15% compounded continuously so that, starting next year, 7, with sides of length 1. Remove the middle triangle formed
you can make annual withdrawals of $2,000 forever? by joining the midpoints of the sides of 7, as shown in Fig¬
ure 8.14a. This leaves three equal triangular regions, and the
57. HISTORICAL QUEST The Greek philosopher Zeno of
next step is to remove the middle triangle from each of these,
Elea (495-435 B.C.) presented a number of paradoxes that
as shown in Figure 8.14b. If you continue this process indef¬
caused a great deal of concern to the mathematicians of his
initely, what remains is the Sierpinski gasket. Show that the
period. These paradoxes were important in the development
Sierpinski gasket has area 0. Hint: What is the sum of the
of the notion of infinitesimals. Perhaps the most famous is the
areas of the regions removed?
racecourse paradox, which can be stated as follows:
A runner can never reach the end of a race: As he runs the
▼
track, he must first run | the length of the track, then ^ the re¬
maining distance, so that at this point he has run | + i = iof
the length of the track, and ^ remains to be run. After running
A
half this distance, he finds he still has | of the track to run, and a. First step b. Second step c. Third step
so on, indefinitely.
Figure 8.14 Construction of a Sierpinski gasket
understatement, as illustrated in the following anecdote, stated b. Use part a to show that if
in the form of a quest problem. ■
Two trains, each traveling at 10 mph approach each other lim ak = A, then W(ak — ak+2) = a\ + a2 — 2 A
k —> oo
on the same straight track. When the trains are 5 miles apart, k= 1
a fly begins flying back and forth from train to train at the rate
of 16 mi/h. Between the time the fly starts and the time the
c. Use part b to evaluate 53 \k]/k — (k + 2)l/<A+2)].
trains collide, how far does the fly fly?
4=1
a. Find the fly’s total distance traveled by determining how
long the fly flies. d. Evaluate
k2~ 1
b. When von Neumann gave the correct answer almost im¬ 4=2
mediately, the poser of the problem was convinced he had 66. Square ABCD has sides of length 1. Square EFGH is
used the "trick" suggested in part a. "What trick?" asked formed by connecting the midpoints of the sides of the first
von Neumann. "I summed the series." Follow von Neu¬ square, as shown in Figure 8.15. Assume that the pattern of
mann’s lead by summing an infinite series whose terms shaded regions in the square is continued indefinitely. What is
are the distances traveled by the fly as it flies between the the total area of the shaded regions?
trains.
n D G C
O60- a. Let ^ ak = Sn be the nth partial sum of an infinite series.
4= 1
Show that S„ — S„_i = a„.
b. Use part a to find the nth term of the infinite series whose
n
nth partial sum is S„ = -.
2n + 3
61. Prove that if Ea^ = A and Ebk = B, then
E (ak + bk) = A + B.
62. Show that if T,ak converges and Y,bk diverges, then E(ak—bk)
diverges. Hint: Note that bk — ak — (ak — bk).
63. Counterexample Problem Either prove or disprove (by finding a
counterexample) each of the following statements:
a. If T,ak and T,bk are both divergent series, then E{ak + bk)
is also divergent. Figure 8.15 Find the area of the shaded regions.
b. If Eak and Ebk are both convergent series, then E(ak + bk)
\2k
is also convergent. 67. Evaluate > —■—
(Ak _
OO
4=1
(4* - 3*)(4*+1 - 3*+1)
64. Show that 53ark diverges if |r| > 1. Hint: Find A and B so that
4=0
n 3 kA _ 3 kB \2k
65. a. Show that - ak+2) = (ax + a2) — (an+i + an+2). 44334 + 44+I _ 34+I ~ (44 _ 34)(44+1 _34+l)
4=1
DIVERGENCE TEST
When investigating the infinite series Ea*, it is easy to confuse the sequence of general
terms [cik] with the sequence of partial sums {5,,}, where
n
Sn = ^ Uk
4=1
8.3 The Integral Test; p-Series 515
Because the sequence {ak} is usually more accessible than {£„}, it would be convenient
if the convergence of {5,,} could be settled by examining
lim
k—> oo
Even though we do not have one simple, definitive test for convergence, the following
theorem tells us that if certain conditions are not met, the series must diverge.
lim Sn-1 — L
n—>• oo
If the series J2 ak converges, then lim ak = 0. Thus, if lim ak 7^ 0, the series Ea*
k->oo k-+oo
must diverge. □
What This Says The divergence test can only tell us that Ea* diverges if
lim ak 7^ 0, but it cannot be used to show convergence of a series. In Example
k-+oo
2, we show that the series E } diverges even though lim - = 0.
K 00 k
Solution
Taking the limit of the Ath term as A —>• 00, we find
A - 300 1
lim -- = -
k->oo 4A + 750 4
Because this limit is not 0, the divergence test tells us that the series must diverge. ■
Series whose terms are all nonnegative numbers play an important role in the general
theory of infinite series and in applications. Our next goal is to develop methods for
investigating the convergence of nonnegative-term series, and we begin by establishing
the following general principle.
Proof Suppose Ea* is a series of nonnegative terms, and let Sn denote its nth partial
sum; that is
n
This convergence criterion is often difficult to apply because it is not easy to deter¬
mine whether the sequence of partial sums {5,,} is bounded from above. Our next goal
in this section and in Sections 8.4-8.6 is to examine various “tests for convergence.”
These are procedures that allow us to determine indirectly whether a given series con¬
verges or diverges without actually having to compute the limit of the partial sums.
We begin with a convergence test that relates the convergence of a series to that of an
improper integral.
WARNING w, , . , Proof We will use a geometric argument to show that the sequence of partial sums
When applying the oo rOO
integral test, the function / need not be {£„} of the series is bounded from above if the improper integral / f(x)dx
decreasing for all x > 1, only for all
x > N for some number N. That is, / converges and that it is unbounded if the integral diverges. Accordingly, Figures 8.16a
must be decreasing "in the long run." and 8.16b show the graph of a continuous decreasing function / that satisfies f(k) —
cik for k — 1, 2, 3,.... Rectangles have been constructed at unit intervals in both
figures, but in Figure 8.16a the /cth rectangle has height f {k + 1) — ak+1, whereas in
Figure 8.16b, the comparable rectangle has height f(k) = Ok. Notice that in Figure
8.16a, the rectangles are inscribed so that
AREA OF THE FIRST n - 1 RECTANGLES < AREA UNDER y = f(x) OVER [1, n]
a. The /cth rectangle has height b. The /cth rectangle has height
f(k + 1) = ak+l. m = ak.
/
AREA UNDER y = f(x) OVER [1, n + 1 ] < AREA OF THE FIRST n RECTANGLES
n+1
/
Let Sn = a 1 + <32 + <33 + • • • + <3ii be the nth partial sum of the series so that
n +1
f(x)dx < Sn < a\ + /
pn
f{x)dx
Now, suppose the improper integral J f{x)dx converges and has the value /;
/oo *
fix) dx — I. Then
/ n
fix) dx < cii + /
r 00
fix) dx = «i + /
It follows that the sequence of partial sums is bounded from above (by 01 + I), and the
OO
convergence criterion for nonnegative term series tells us that the series ^ must
converge.
/
k= 1
OO
lim /
/ n-
fix) dx — 00
n —> 00 i
'ii+1
It follows that lim Sn = 00 also because / fix)dx < Sn. This means that the
n^oo Jj
series must diverge. Thus, the series and the improper integral either both converge or
both diverge, as claimed. J
^1 111
Er1
k= 1
+ i + t + 4 + "'
This is called the harmonic series, and in the next example, we use the integral test to
show that this series diverges.
Solution
1
Because fix) = - is positive, continuous, and decreasing for x > 1, the conditions of
x
the integral test are satisfied.
r°0 1 rb 1
/ - dx = lim / — dx — lim [In ft — In 1] = 00
Jl X b-+ooj\ X b^-oo
Solution
The function f(x) = ,x/5
— xe x/l5 is positive and continuous for all x > 0. We find
that
,-x/5
f'(x) = x + e~"s = (l - i)
(l - i) e~"* = 0
1 — — = 0 Since e~x 3 7^ 0
x =5
We see that f'(x) < 0 for x > 5, so it follows that / is decreasing for x > 5. Thus,
the conditions for the integral test have been established, and the given series and
the improper integral either both converge or both diverge. Computing the improper
integral, we find
/■oo pb
/ xe~x/5 dx — lim / xe~x/5 dx
J5 b^°° J 5
Let u — x dv e x/5 dx
du = dx v -5e~x/5
Y—
pk/5 = 50<f lim = 0
k= 1 K b—>00
Thus, the improper integral converges, which in turn assures the convergence of the
converges to the same number 50e_1 given series. ■
p-SERIES
The harmonic series El/k is a special case of a more general series form called a
p-series.
V— - — 1 1
k= 1 kP ~ IP + 2p + 3r "*
The following theorem shows how the convergence of a p-series depends on the
value of p.
Proof We leave it for the reader to verify that f(x) = — is continuous, positive,
xP
and decreasing for x > 1 and p > 0. We know the harmonic series (p — 1) diverges.
For p > 0, p 1, we have
00 dx bx~p — i [- if p > 1
/ — -
Xp
lim
b—► oo /
.~P dx = lim -= < p — 1
b—>00 1 O
1
I
^ oo
■r |
if 0 < p < 1
That is, this improper integral converges if p > 1 and diverges if 0 < p < 1. For
p = 0, the series becomes
OO
1 1 1
E¥ + - + - +
and if p < 0, we have lim — = oo, so the series diverges by the divergence test
k->oo kP
(Theorem 8.9). Thus, a p-series converges only when p > 1. □
What This Says It should be clear that all functions of the form y = \/xp
with p > 0 are decreasing, but why should the series £ ^ converge only for
p > 1 ? The answer lies in the rate of decrease for y — \/xp. If p > 1, the curve
y = \/xp falls fast enough to guarantee that the area under the curve for p > 1
is finite (see Figure 8.17a), while if p < 1, the curve \/xp falls more gradually,
and the area under the curve is infinite (Figure 8.17b).
Solution
a. Here VT3 = A3/2, so p = | > 1, and the series converges.
We note that
1
converges, because it is geometric with |r | = — < 1
e
and
OO
1
E Vk diverges, because it is a /7-series with /? = -<!
Because one series in the difference converges and the other diverges, the given
series must diverge (Theorem 8.7). ■
k=\ K
oo , EO
*=l '
+r
s' IfTk
6-e
nr 2a^a
V v-ti 24-
00
5
E4=
k=1 nr
Show that the function f determined by the nth term of each se¬
OO oo
ries given in Problems 7-12 satisfies the hypotheses of the integral
25. J2k~V* 26. f>4'3
test. Then use the integral test to determine whether the series
k=1 k=\
*=1
converges or diverges. oo oo
oo k2
k
oo 1 oo 27. Y , 28. £
7. y 1 , 8. J](2 + A)-3/2 Y VFT2
tt (2 + 3A)2 *=i OO oo ,
k=2 K OO
31# yln^
00 32. E '
jt=2 v/* ^2 *Vln A
n. i2. Yke~k2
t! ' + *2 k=] ,, ^ A _ ^A2
In Problems 13-18, either use the divergence test to show that the
33, E
*=1
ek 34- E
*=1 e
given series diverges or show that the divergence test does not ap¬ « tan"1* _ ^ ,
ply. 35. e j , k2 36’ Ecot k
k=1 1 ^ * *=1
00 /
OO OO 1
i3. y k
t<k+' *SHr 37. £ tan 1 A;
*=1
38. £
*=1 e* +e
00
£ _1_
00 k ^ jfc — sin A
is. E fi—
n? va2 -1
10. 39. > -
“ 3A + 2 sin k
40. (V*2 + k - A)
*=1 *=i v
8.4 Comparison Tests 521
i 1 1
4i. y>i sin 59. E 60. £
4=1 4=4 —t .L‘“v.'J
4=3 /:ln£[ln(ln&)]p —t k(\nk)p
4=2
Use the divergence test and the p-series test to determine which 61. Counterexample Problem If T,ak converges, E6k diverges, and
of the series in Problems 43-54 converge. Write “inconclusive" if lim bk = 0, then Eakbk must diverge. Either prove this state-
there is not yet enough information to settle convergence. k->-oc
ment or provide a counterexample to show it is not generally
°° k- 1 ~ £2+1
true.
«-E
4=1 k + 1
44. ^
4=1 * + 5 62. Counterexample Problem If lim ak = 0, lim bk 0, and Ea*
k —> oo k—>oo
00 k2 +1 ^ ^2k4 + 3
45. •£ 46. ^ and Ebk both diverge, then Eakbk must also diverge. Either
k=1 k3 k= 1
k5 prove this statement or provide a counterexample to show it is
CXJ
oo , OO not generally true.
-k5 +k2 + l
48. £
*=1 4=1 fc5 +2 Q 63. Let f{x) be a positive, continuous, decreasing function for
00
x > 1, and let ak — f{k) for k = 1,2.If E ak converges,
rCV5 + 1
show that
«-E^
*=1
50. £
fc=l
£2.236
pOO 00 poo
00 00
l jt+i / f{x)dx < < <3| + / f(x)dx
51. J2 52. E Ji ■'i
k=1 it + k + 2 4=1 21 ' l 2 4=1
00 OO
1 2>t + 3 1
53. x; 64. Getting a feeling for a series by examining its graph is a good
2* + 3£ + 4
54. E 2^
*=i 4=1 idea, but you should be careful because you cannot always
Q Find all values of p for which each series given in Problems 55-60 obtain a graph. Decide whether
converges.
iooy
55. V-t—
k _ ^
56. £
k2
E >
k=2 (k2 - 1 )p jfcf (k2 + 4)P
00
1 converges or diverges, and then see if you can verify your re¬
57. £ 58. £ln*
sult graphically.
4=3
In 4: k=2 kP
Proof Suppose Ea* is a given nonnegative term series that is dominated by a con¬
vergent series Sq in the sense that for sufficiently large k,0 < ak < ck for all k. Then,
because the series Ec* converges, its sequence of partial sums is bounded from above
(say, by M), and we have
n n
^ What This Says Let Ea^, Ec* and Edk be series with positive terms. The
series Ert^ converges if it is “smaller” than (dominated by) a known convergent
series Ec* and diverges if it is “larger” than (dominates) a known divergent series
Edk. That is, “smaller than convergent is convergent,” and “bigger than divergent
is divergent.”
Solution
We will compare the given series with a geometric series. Specifically, note that if
k >2, then 1/A:! > 0 and
A! = A(A - 1)(A — 2) • • • 3 • 2 • 1 > 2 • 2 • 2 ■ • ■ 2 • 2 • 1 = 2k~{
k — 1 terms
Therefore, we have 0 < — < ——and because the given series is dominated bv the
A! 2k~'
OO
It is not always easy or even possible to make a suitable direct comparison between
two similar series. For example, we would expect the series E 1/(2* — 5) to converge
because it is so much like the convergent geometric series El/2*. To compare the
series, we must first note that ——- is negative for k — 1 and k = 2 and positive for
k > 3. Thus, if k > 3,
1 1
0 < —- <
” 2* 2* - 5
That is, El/(2* — 5) dominates the convergent series El/2*, which means the com¬
parison test cannot be used to determine the convergence of E1 /(2* — 5) by comparing
with E1 /2*. In such situations, the following test is often useful.
Proof The limit comparison test is proved by using the direct comparison test. De¬
tails of the proof are given in Appendix B. -I
^ What This Says We have the following procedure for testing the conver¬
gence of Ea*.
Step 1. Find a series T,bk whose convergence properties are known and whose
general term bk is “essentially the same” as a*.
Ok
Step 2. Verify that lim — exists and is positive.
k-*oo bk
Step 3. Determine whether E^ converges or diverges. Then the limit com¬
parison test tells us that Eak does the same.
Solution
Because the given series has the same general appearance as the convergent geometric
series E1/2*, compute the limit
1
2* -5 2*
lim lim = 1
k—*00 1 k-*oo 2* -5
2*
This limit is finite and positive, so the limit comparison test tells us that the given
series will have the same convergence properties as E1 /2*. Thus, the given series
converges.
EXAMPLE 5 Divergence using the limit comparison test
OO
3k+ 2
Test the series for convergence.
k= l Vk(3k - 5)
Solution
For large values of k, the general term of the given series
3k+ 2
ak = ~7=-
Vk(3k - 5)
seems to be similar to
3k _ 1
y/k(3k) \[k
3k+ 2
Vk(3k - 5) 3k + 2
lim — — lim lim - = 1
k—^ oo bk k—yoo 1 k >oo 3k — 5
\fk
Because the limit is finite and positive, it follows that the given series behaves like the
series El/>/&, which is a divergent p-series (p = |). We conclude that the given
series diverges. ■
EXAMPLE 6
OO
Ik + 100
Test the series
E
k= 1
for convergence.
Solution
For large k.
Ik + 100 k
ak = -r - -7 seems to behave like bk
_ 70 ^kj5
Ik + 100
,. e*/s - 70 ek^5(lk + 100) „
lim --= lim -—-= 7
k-+oo k k^>oo k(ekl^ — 70)
^kj5
In Example 3, Section 8.3, we showed that the series Y,k/ek^5 converges, and therefore
the limit comparison test tells us that the given series also converges. ■
Occasionally, two series Ea* and Ebk appear to have similar convergence proper¬
ties, but it turns out that
fc->oo bk
is either 0 or oo, and the limit comparison test therefore does not apply. In such cases,
it is often useful to have the following generalization of the limit comparison test.
8.4 Comparison Tests 525
Proof The first part of the proof is outlined in Problem 61; the second part follows
similarly.
Solution
We will carry out this proof in three parts:
Case I: {q > 1)
Let p be a number that satisfies 1 < p < q, and let
In* 1
Ok — and bk = —
*<? kP
Then
\nk
kf
lim — — lim
k—► oo bk k-+oo 1
kP
In*
= lim This is the indeterminate form
k-+oo k^-P
oo/oo because q — p > 0.
Because El/kp converges (p-series with p > 1), the series converges by the zero-
infinity limit comparison test.
Because E»* = £ — diverges (we know p < 1), it follows from the zero-infinity
*=i k=\ ”
^ln* .
comparison test that 2_, — diverges when q < 1.
k=l
526 Chapter 8 Infinite Series
Case III: (q = 1)
In A:
Here, the series is ^ which diverges by the integral test:
k= 1
lnx ln"x
dx — lim = oo
b—>oo
Let u — \nx\ du = — dx
x
1. geometric series: ^ rk
k=o
2. p-series: ^ —
k=l kP
oo
31. T-f-
k=i 'Jk2k
1
32. f ;uwo
k=1 Vk 3*
Each series in Problems 3-12 can be compared to the geometric OOOO |
series or p-series given in Problems 1 and 2. State which, and then 33. V |sin(^!)l
^
*=i A:2 M' § VJ In*
determine whether it converges or diverges.
2k3 + A: + 1
e
OO
i3. y -4
1
- ■ Y
14. —
OO i
1 45. f;-i-
A:33 + 4
E-J lr
4
46. L_J k _ l
k= 1 k=2
k=1 A:2 + A: fak2 + 3k + 2
oo
oo ln(A: + 1) 00 In A:
1
15- E4= 16. T-E 47- E
kJ (*+D3
<8-E^
*=i
ks/k
oo
oo 1
1 1
i 4». e 50. E
17. Y _ is. e S (A: + 3) (In A:) 1.1
k=2 S (A: + 3)(lnA:)09
fbr v^FT3 *fk(k + 1) k=2
OO OO
oo 1
1 si. e*"- 52. Yk(l+k)/k
19- E^= 20. £ k=\ *= 1
fcr vf+2 rT n/FTT
*=i 53. Show that the series
°° 94.2 OO
A: + 1
2i. x:2*
k=1 *4-4
22- E
k=\ A:2 + 1
k2 14 9
(£ + 3)! _ 4! + 5! + 6! + '''
(* + 2)(* + 3) + l)3
is. e kV2 M- E
(k
k9'2 converges by using the limit comparison test.
*=1 *=i
OO
2k+ 3
54. Show that the series
3k2 + 2
25. E
frf A:2 + 3* + 2
2«. e
k=1 A:2 + 3A: + 2 i , 1 , 1
1 -|- --"h -- 4- ...
1 2kk\
OO /
00 5 1-3 1-3-5 1-3-5-7 (2* + l)!
27- E 28. ^
-T *=i (* + 2)2* >t=i 4* + 3 converges. Hint: Compare with the convergent series ]T 1/A:!.
8.5 The Ratio Test and the Root Test 527
55. Use a comparison test to show that 58. Suppose 0 < ak < A for some number A and bk > k2. Show
ak
that E — converges.
°° i
y_1 —
“ (ln&)ln*
bk
59. Suppose ak > 0 for all k and that Y,ak converges. Show that
1
E— diverges.
converges. Hint: Use the fact that In A: > e2 for sufficiently ak
large k. 60. Counterexample Problem Suppose that 0 < ak < I for every k,
56. a. Let {a,,} be a positive sequence such that and that Eak converges. Use a comparison test to show that
Eo2 converges. Find a counterexample where Eal converges,
lim kpak but Eak diverges.
k—>oo
61. Show that if ak > 0, bk > 0, and
exists. Use the limit comparison test to show that Y.ak con¬
ak
verges if p > 1. lim -— = 0
k->-OQ bk
b. Show that Y,e~k' converges. Hint: Show that lim k2e~k~
k-+oc
then Eak converges whenever Ebk converges. Hint: Show
exists and use part a.
that there is an integer N such that ak < bk if k > N. Then
57. Let Y,ak be a series of positive terms and let {bk} be a sequence
use a comparison test.
of positive numbers that converge to a positive number. Show
that Ea* converges if and only if Y.akbk converges. Hint: Use 00 1
62. Show that y-converges if and only if q > 1.
the limit comparison test. f^kQnty
RATIO TEST
Intuitively, a series of positive terms 'Eak converges if and only if the sequence {a*}
decreases rapidly toward 0. One way to measure the rate at which the sequence {a*} is
decreasing is to examine the ratio ak+\/ak as k grows large. This approach leads to the
following result.
Proof In a sense, the ratio test is a limit comparison test in which Yjcik is compared
to itself. We will use the direct comparison test to show that T,ak converges if L < 1.
Choose R such that 0 < L < R < 1. Then there exists some N > 0 such that
°k+l < /? for all A: > N; Since the ratios ak+\/ak
ak are eventually very close to L
Therefore,
on+i < OnR
2
fliV+2 < aN+] R <
2 3
(Ift-(-3 < O/v+2^ < aN+1 7? < OnR
528 Chapter 8 Infinite Series
Ep
k=i K
converges with lim gw = Um (k+ll - ,
£->oo a* £->oo 1
k2
it!
Thus L < 1, and the ratio test tells us that the given series converges.
Solution
kk
Let a* = — and note that
A:!
(k + 1)*+1
k\(k + 1)*+1
£ = lim ^±i = lim (k + 1)1 — lim
£->00 a* £->00 kk £->oo kk(k + 1)!
T\
(k + 1)* 1
= lim = lim 1 + - = e >
kk £->oo
£->oo V k)
Because L > 1, the given series diverges.
8.5 The Ratio Test and the Root Test 529
TECHNOLOGY NOTE
It is instructive to compare and contrast two different ways you can use technology to help you with problems such as Example
2. First, as you have seen, you can use a calculator or a computer to calculate the partial sums directly. This table is shown
below at the left.
A second procedure is to use the ratio test for Example 2. In this case, we are interested in
(k + 1)*+1
ak+1 (k + 1)1
lim - lim = lim 1 + 1
k-*oo Clk k—>oo kk k-*oo
k\
We can look at terms of the sequence a* to see whether this limit is greater than, less than, or equal to 1. This is
By table: By graph:
OO
kk
E k\ lim
k—*oo
•
•
•
_1_1_
•
3 7.5
•
2.5 -
•
•
18.16667 X Y
•
4
•
5 44.20833 2.0 -
1 2
6 109.0083 2 2.25
3 2.37037 1.5 -
7 272.4097 2.44141
4
8 688.5113 5 2.4883
6 2.521626 1.0 -
9 1756.138 2.546500
7
10 4511.87 8 2.565784
9 2.581175 0.5 -
11 11659.53 2.593742
10
12 30273.46 100 2.704814 1 1 | | | 1 1 1 1
1000 2.716924 1 1 1 1 1 1 1 1 1
13 78912.3 10000 2.718146 123456789
14 206375.3 100000 2.718268
15 541239.9
The partial sums seem to diverge. We see that the ratio test gives a limit greater than 1.
Finally, you might test this using a calculator to verify that it diverges.
Solution
1
Let ak — and find
2k -3
1
,. 2(k + 1) — 3 ,. 2k — 3
L = lim —-t1-= lim —-- = 1
k—> oo 1 k-*-oo 2k — 1
2k-3
The ratio test is inconclusive. We can use the integral test or the limit comparison test
to determine convergence. Note that
1
y- is similar to the known divergent series 2J
2k
k=2 k=2
J2 k3xk =x + 2 V + 3 V + • ■ •
k=l
converges. This is called a power series, and it will be discussed in more detail in
Sections 8.7 and 8.8.
Solution
, (k+l)3xk+1 'k + 1
L = lim -——--— hm = jc
&-»oo kiXk k-+oo
Thus, the series converges if L — x < 1 and diverges if x > 1. When x — 1, the series
OO
ROOT TEST
Of all the tests we have developed, the divergence test is perhaps the easiest to apply,
because it involves simply computing lim ak and observing whether that limit is zero.
fc—>0o
Unfortunately, most “interesting” series have lim ak = 0, so the divergence test cannot
k->oo
be used to determine whether they converge or diverge. However, the following result
shows that it may be possible to say more about the convergence of Yiak by examining
what happens to as k —» oo. This test will prove particularly useful with a series
involving a *th power.
Proof The proof of this theorem is similar to that of the ratio test and is outlined in
Problem 57. □
Solution
T 1
Let ak = ——— and note that
(In*)*
Because L < 1, the root test tells us that the given series converges. ■
8.5 The Ratio Test and the Root Test 531
Solution
*2-1 V*
= k—>
limoo (l
y
+ j)
kJ
=e > 1
kP
*/T
lim lim = rp = l Inconclusive root test
*-> oo V kP " *—>00 where I'Hopital's rule
is used to evaluate the limit.
See Example 11 of Section 4.5.
k\ 1! 2! 3!
1 + + + + •
£7 4 • 7 ■ • • (3k + 1) 1-4 1-4-7 1-4-7-10
*=0
Solution
k\
Let ak = Because ak involves k\, we try the ratio test.
1 - 4-7---(3it+ 1)
(* + !)!
fl*+i 1 • 4 • 7 • • • [3(it + 1) + 1] _(k + 1)! • [1 • 4 • 7 • • • (3k + 1)]
Solution
k\
The general term ak = — involves a Ath power, which suggests using the root test, but
the presence of the factorial k\ makes using the ratio test much more reasonable.
(k + 1)!
r (k + i)*+i (k + iy.kk (k + \)kk
lim - -= lim -;—- = lim --—-
—
3- E k\
4=1
4.±k'
4=1
2* 29. Y5k + 2 30. £
(A!)2
A2* kk
4=1 4=1
OO ; i oo
5 T—
93k 23*
6. V- oo oo
3A + 5
k=1
31- Ett 32. E A3*
oo nk 4=1 4=1
00 *
9- E 10. J]Ae
35
oo
v EE1
oo
k=1 4=1
* Z-4 £4+°.;
t4+0.5
36. V —
4=1 ^ A*
4=1
"•EM
*=1 -S‘(D‘
oo
37. £
k\ oo 91,0004
4=1 0 In Problems 45-52, assume x > 0 and find all x for which the
given series converges.
21. Y —L 22 . £ [(2A)!12
W>2
(2*)! 4=1 45. £tV 46. 'Yl kxk
OO 4=1 4=1
23. X>22-* 24. ;r*43- f(,+0.5)‘
00
Ox - 0.4)*
4=1 *=1 48. £
47-S^ 4=1
A2
00
oo oo ^24
x-tn~2 26. £
4= 1 4=1
2k + 1 49. £
4=1
A! 50- E V 4=1
8.6 Alternating Series; Absolute and Conditional Convergence 533
OO OO
57. Prove the root test by completing the steps in the following
51. 'Y2(ax)k for a > 0 52. Yt kxJ outline:
k=I k=1
53. Use the root test to show that Y,kpe~k converges for any fixed a. Let lim ^fcT„ — R. If R < 1, choose r so R < x < 1.
n—>oo
positive real number p. What does this imply about the fol¬ Explain why there is an N such that
lowing improper integral?
0 < !fan < x
xpe xdx
r for all n > N. Then use the direct comparison test to show
OO
that Ea„ converges.
54. Consider the series ^2-*+<-1) . What can you conclude b. If R > 1, explain why an > 1 for all but a finite number of
k=\ n. Use the divergence test to show diverges.
about the convergence of this series if you use each of the
c. Find two series Ea„ and Y,b„ with R — 1, so that
following?
converges but Y.b„ diverges.
a. the ratio test b. the root test
58. Exploration ProblemRecall from Problem 48, Section 8.1, that
55. Let {ak} be a sequence of positive numbers and suppose that
the Fibonacci sequence {an} is
lim = p where 0 < L < 1.
k-* oo ak
1, 1,2,3,5,8, 13,21,...
a. Show that lim ak = 0. Hint: Note that Y.ak converges.
k—> oo
111111 1 1 1 1 1
1 + - + - + - + T + - + - + 77 + '
2 2 4 4 8 8 16 E ak =,+,+_+_+_+
There are two classes of series for which the successive terms alternate in sign, and
each of these series is appropriately called an alternating series:
OO
In general, just knowing that lim ak = 0 tells us very little about the convergence
k-^-oo
properties of the series Y,ak, but it turns out that an alternating series must converge il
the absolute value of its terms decreases monotonically toward zero. This result, first
established by Leibniz in the 17th century, may be stated as shown in the following
theorem.
534 Chapter 8 Infinite Series
E(-Dt+1^
£=i
«i satisfies the two required properties, it converges. The steps for the other type of alter¬
nating series are similar. For this proof we are given that
We need to prove that the sequence of partial sums {£„} converges, where
o Si n
(a)
Sn — ^ ^(~l) + ak — a\ — + #3 — a4 + - 1 • + (—1 )n+lan
k= 1
a\
a2 Our strategy will be to show first that the sequence of even-indexed partial sums {S2,i}
is increasing and converges to a certain limit L. Then we will show that the sequence
of odd-indexed partial sums {S^-if also converges to L.
To understand why we would think to break up Sn into even- and odd-indexed
0 s2 Sj partial sums, consider Figure 8.18. Start at the origin. Because ax > 0, S\ will be
(b) somewhere to the right of 0 as shown in Figure 8.18a. We know that a2 < a{ (decreas¬
ing sequence) so S2 is to the left of Sx but to the right of 0, as shown in Figure 8.18b.
As you continue this process you can see that the partial sums oscillate back and forth.
Because an -> 0, the successive steps are getting smaller and smaller, as shown in
Figure 8.18c.
Note that the even partial sums of the alternating series satisfy
S2 = a! - a2
Sequence of even Sequence of odd
S4 = (ax — a2) + (#3 — af)
partial sums {S2n) partial sums {S2„-i}
is increasing is decreasing
(c)
Sin — (tfi — a2) + (a3 — a4) +-1- (a2n_\ — a2n)
Figure 8.18 Alternating series test
Because {ak} is a decreasing sequence, each of the quantities in parentheses (a2j
a2j) is nonnegative, and it follows that {S2,,} is an increasing sequence:
S2<S4<S6<---
Moreover, because
S2 = ax — a2 < ax
S4 = ax — (a2 — af) — 04 < ax
lim S2n = L
8.6 Alternoting Series; Absolute and Conditional Convergence 535
Since the sequence of even partial sums and the sequence of odd partial sums both
converge to L, it follows that {S„} converges to L, and the alternating series must
converge (in fact, to L). □
REMARK: When you are testing the alternating series I,(-\)k+lak for conver¬
gence, it is wise to begin by computing lim ak. If lim ak 7^ 0, you know immediately
fc->oo k-t 00
that the series diverges (by the divergence test), and no further computation is neces¬
sary. However, if this limit is 0, you can show that the alternating series converges by
verifying that {ais a decreasing sequence.
Solution
The series can be expressed as E(—1 )kak, where a* -. We have lim - = 0 and
WARNING Remember that the Ctk k k-Kso k
used in the alternating series test does because
NOT include the (—1)*+1. 1 1
k + 1 k
for all k > 0, the sequence {a^} is decreasing. Thus, the alternating series test tells us
that the given series must converge. ■
Solution
In h-
Express the series in the form E(— l)*+la*, where a* = —j— (note that a* > 0 for
k > 1). The graph of y = {\nx)/x is shown in Figure 8.19. By applying I’Hopital’s
rule, we find that
1
In* .. r
lim -= lim = 0
JC->00 X X^OO 1
It remains to show that the sequence {a*} is decreasing. We can do this by setting
lnx , , , , .
f(x) —-and noting that the derivative
satisfies f{x) < 0 for all x > e. Thus, the sequence {ak} is (eventually) decreasing
for all k > 3 > e. The conditions of the alternating series test are satisfied, and the
given alternating series must converge.
536 Chapter 8 Infinite Series
Solution
Since
lim ex,k = exp (lim - ) = e° = 1 ^ 0
k-*oo yk^-oo k )
00
(-1)*+1
Alternating p-Series The series is called the alternating p-series.
kP
k=1
The following example shows that the alternating p-series converges for all p > 0.
Solution
If dk = —, then lim ak = 0 for p > 0. To show that the sequence {<3^1 is decreasing,
kP k-*oo
we note that
1
flfe+i _ (k + 1 )p _ kp
ak J_ “ (k + \)p <
kP
so ak > ak+\. Thus, the alternating p-series converges for all p > 0. ■
For any convergent series with sum L, the nth partial sum is an approximation to L that
we expect to improve as n increases. In general, it is difficult to laiow how large an n to
pick in order to ensure that the approximation will have a desired degree of accuracy.
However, if the series in question satisfies the conditions of the alternating series test,
the following theorem provides a convenient criterion for making this determination.
or ^(-l)*+1a*
k=1 A=1
15 ~ Sn | < an+1
Proof We will prove the result for the second form of the alternating series and leave
the first form for the reader. That is, let
oo n
= X] (-Dt+1«*
Jt=n+1
Because the sequence {a,,} is decreasing, we have ak—ak+\ > 0 for all k, and it follows
that
1*5 Sn | = I1 (<3/i+2 <3/54-3) (<3/i+4 <3/?+5) I — <3/i+l
^ What This Says If an alternating series satisfies the conditions of the al¬
ternating series test, you can approximate the sum of the series by using the nth
partial sum (S„), and your error will have an absolute value no greater than the
first term left off (namely, an+1).
. ^ (-1)*+1
Consider the convergent alternating series 2_^ —n— •
k=\ k
a. Estimate the sum of the series by taking the sum of the first four terms. How accu¬
rate is this estimate?
b. How many terms of the series are necessary to estimate its sum with three-decimal-
place accuracy? What is this estimate?
Solution
1
a. Let ak — — and let S denote the actual sum of the series. The error estimate tells
k4
us that
IS — S4I = as
where S4 is the sum of the first four terms of the series. Using a calculator, we find
S4 = 1 1 J_L 0.9459394
F ~ 2* + 3^ ~ 44
^2,000 < n + 1
6.687403 - 1 <
Thus, n > 5.687403, so we need at least 6 terms to achieve the required accuracy.
We find that
n 111111
The convergence tests we have developed cannot be applied to a series that has mixed
terms or does not strictly alternate. In such cases, it is often useful to apply the follow¬
ing result.
Proof Assume £ |a*| converges and let bk = ak + |a*| for all k. Note that
2 \ak\ if ak > 0
bk
0 if ak < 0
Thus, we have 0 < bk < 2 |a*| for all k. Because the series £ |a*| converges and both
Y,bk and £ |a*| are series of nonnegative terms, the direct comparison test tells us that
the dominated series 'Zbk also converges. Finally, because
Qk — bk | Qk |
and both £fi* and £ |a*| converge, it follows that Y,ak must also converge. □
Solution
This is not a series of positive terms, nor is it strictly alternating. However, we find that
the corresponding series of absolute values is the convergent p-series
, 111 ^1
' + i + 9 + !6+'" = £^
Because the absolute value series converges, the absolute convergence test assures us
that the given series also converges. ■
8.6 Alternating Series; Absolute and Conditional Convergence 539
sink
~lF~
which is dominated by the convergent geometric series £ ^ because |sink| < 1 for all
k; that is,
sink
0 < for all k
~2r
Therefore, the absolute convergence test assures us that the given series converges. The
sinx
Figure 8.20 Graph of y = graph of the related continuous function y = (sinx)/(2'r) is shown in Figure 8.20. ■
IT
If Y>ak converges, then £ \ak\ may either converge or diverge. The two cases that
can occur are given the following special names.
Absolutely And Conditionally The series Y,ak is absolutely convergent (or converges absolutely) if the related
Convergent Series series £ \ak\ converges. The series Y,ak is conditionally convergent (or con¬
verges conditionally) if it converges but £ \ak \ diverges.
WARNING .. ,
Note that a convergent
series is either absolutely convergent or
conditionally convergent, but it cannot
be both. For example, the series in Examples 6 and 7 are both absolutely convergent, but the
alternating harmonic series
OO
(-Dt+|
E k
is conditionally convergent because it converges (see Example 1), whereas the related
series of absolute value terms £ 1/k (the harmonic series) diverges.
The ratio test and the root test for positive-term series can be generalized to apply
to arbitrary series. The following is a statement and proof of the generalized ratio test.
lim
k-t-oc <Ik
Proof Assume L < 1; then the positive series E \a^ \ converges by the ratio test, and
hence Ea^ converges absolutely. Assume L > 1; then the sequence {|^|} is eventually
increasing, which means that {ak) cannot converge to 0 as k —>• oo, so Ec^ diverges
by the divergence test. Assume L = 1; see Problem 61 for this part of the proof. □
Solution
Let ak = kxk.
&k+1
lim
k-+oo Clk
(.k + l)xk+l
lim
k—>oo
lim
k—too
lim
k—>oo
Thus, by the generalized ratio test, since L = \x\ the series converges for |jc| < 1
and diverges if |x| > 1. If x — 1, the series is T,k(l)k, which clearly diverges by the
divergence test. Similarly, if x = — 1, E£(— 1)* diverges by the divergence test. Thus,
the series converges for \x\ < 1 and diverges for |x| > 1. ■
The statement and proof of the generalized root test is left to the problems (see
Problem 60).
This concludes our basic study of convergence tests. There are no firm rules for decid¬
ing how to test the convergence of a given series Ea*, but we can offer a few guidelines
(Table 8.1).
OO
Geometric series, E ork diverges if |r| > 1, and converges if |r| < 1 with sum S —
k=o
|
i V" 1°
4-log senes ^ —
k
converges if q > 1 and diverges if q < 1
*=i 'c
8.6 Alternating Series; Absolute and Conditional Convergence 541
Divergence test Compute lim ak. If this limit is anything but 0, the series diverges.
k^-oo
Tests for Series Yak with ak > 0
Limit comparison test Check to see whether Yak is similar in appearance to a series Ybk
whose convergence properties are known, and apply the limit com¬
parison test. If lim ak/bk = L, where L is finite and positive, then
£—►00
Yak and Ybk either both converge or both diverge.
, (Ifc-L- 1
Ratio test If ak involves k\, kp, or c , try the ratio test, lim - = L. Con-
k-* oo ak
verges if L < 1, diverges if L > 1, test is inconclusive if L = 1.
Root test If it is easy to find lim ty~a~k = L, try the root test. Converges if
k—>oo
L < 1, diverges if L > 1, test is inconclusive if L = 1.
all A:, then ak and the improper integral / f(x) dx either both
Absolute convergence test Take the absolute value of each term and proceed to test for conver¬
gence of positive series, as summarized previously.
Alternating series test (conditional convergence) The alternating series E( —1 )kak or E(—1 )k+xak with ak > 0 con¬
verges if <3*+! < ak for k — 1,2,... and lim ak — 0.
k—>oo
You may be surprised to learn that if the terms in a conditionally convergent series are
rearranged (that is, the order of the summands is changed), the new series may not
converge or it may converge to a different sum from that of the original series! For
example, we know that the alternating harmonic series
(-1)*+1
In 2
542 Chapter 8 Infinite Series
If we rearrange this series by placing two successive subtracted terms after each added
term, we obtain
i + i- i 1 + i -
4 ' 3 6 8 ' 5
= i1 i1,1
4- I
o I
,
2 4 ' 6
= iln2
1 1 1 i
b. Show that lim S„ = - In 2. Hint: You many find it helpful
n-> oo 2
55_1_22 + 32~42 + 52
to use
46. Find an upper bound for the error if the convergent alternating
60. Prove the generalized root test, which may be stated as fol¬ fies the same condition.
lows: Suppose that lim y|a*| = L. Then 62. Let E(— l)*+1a/t be an alternating series such that {ak} is de¬
k —*■ OO
creasing and
If L < 1, the series Y,ak converges absolutely.
If L > 1, the series 'Eak diverges. lim ak — 0
k-+OQ
If L = 1, the test is inconclusive.
Show that the sequence of odd partial sums {S2n-il is decreas¬
61. Show that if
ing.
Qk+1 Construct a divergent alternating series
L = lim 63. Exploration Problem
k—>oo OO
<*k
Xy-i/+1 ak such that lim ak = 0, but {ak} is not decreas-
OO
the series Y,ak can either converge or diverge. Hint: Find a
convergent series with L = 1 and a divergent series that satis¬ ing.
is called a power series in (x — c). The numbers ao, a\, a2,... are the coefficients of
the power series. If c = 0, the series has the form
OO
For what numbers x does a given power series converge? The following theorem
answers this question for the case where c — 0.
The set of values for which a series converges is called the convergence set for the
series. The following three examples show each of these possibilities.
Solution
If x = 0, then the series is trivial and converges. For x ^ 0, we use the generalized
ratio test to find
Xk+1
(* + !)! xk+lkl
L — lim = lim lim = 0
*->oo k—>oo (k + 1)!jc* *->■00 k + 1
k\
Because L = 0 and thus L < 1, the series converges for all x.
Solution
We use the generalized ratio test to find
(it + l)!x*+1
lim = lim (k + 1) \x\
k—*oo k\xk *—>00
For any x other than 0, we have L = oo. Hence, the power series converges only when
jc = 0. ■
k=1
Solution
Using the generalized ratio test, we find
xk+x
y/k + 1 y/k
L = lim lim X
*—>00 *—>00 y/k -f- 1
The power series converges absolutely if |jc| < 1, and diverges if |jc| > 1. We must
also check the endpoints of the interval \x\ < 1, namely x — 1 and x = 1: —
(-1)*
x = —1 : > —— converges by the alternating series test
*=l vk
( ‘ and lim —— = 0
\ y/k + 1 *->00 y/k
According to Theorem 8.22, the set of numbers for which the power series 'Za/(xk
converges is an interval centered at x = 0. We call this the interval of convergence
of the power series. If this interval has length 2R, then R is called the radius of
convergence of the series, as shown in Figure 8.21. If the series converges only for
x = 0, the series has radius of convergence R — 0, and if it converges for all x, we say
that R — oo.
546 Chapter 8 Infinite Series
R 0 R
Solution
We apply the generalized ratio test to find
2^+ij^+i
k+ 1 2k
L = lim = lim
k—roo k „k k—rOO k+ 1
2kX
Thus, the series converges absolutely for 2|x| < 1; that is, for — ^ < x < The
WARNING radius of convergence is R — Checking the endpoints, we find
Do not forget to check the
endpoints.
1 ~ 2* (\\k ^ 1
Harmonic series
2: £7(2) =£;dlverges
1 ^2k r ("1)A
~ 9 '
k=1
l =E k=1
converges Alternating harmonic series
*=i
Solution
Applying the generalized ratio test, we find
2k+lxk+l
(k + 1)!
L = lim = lim Ul = 0
kvk k+
<:->oo 2kx k—KX)
Id
Thus, the power series converges absolutely for x, and the radius of convergence is
infinite. ■
8.7 Power Series 547
k+\
Find the radius of convergence of the power series
£=1
Solution
Using the root test to examine absolute convergence, we find that
*2 1.2 1 /k
'k + 1 'k + 1
L - lim = lim
k—>oo k—>oo
N
1
= lim 1 + \x\ = e\x\
£-►00
Thus, the power series converges absolutely for ^ |jc | < 1; that is, for —e~x < x < e~l.
It follows that the radius of convergence is R — e~x. We do not need to test the
endpoints because we are not finding the interval of convergence. ■
(x + 1)*
Solution
Applying the generalized ratio rest, we obtain
(x + 1) £+1
3£+l
L = lim = lim \x + 1| = lim - \x 4- 1|
£-►00 (x +1 r £—►00
3£+l k^-oo 3
3k
Thus, the power series converges absolutely for
^ \x + 11 < 1
\x + 1| <3
—3 < x + 1 < 3
-4 < x <2
00 (_4 00
x — —4 : ^--- = l)k diverges (by oscillation)
£=0 ^ £=0
00 (2 -j- l )* 00
*-2= E^ = E 1* diverges (by divergence test)
£=o £=0
specifically, if f(x) = ^~^akxk for \x\ < R, then for |x| < R we have
k=0
OO
and
^ What This Says In many ways, a function defined by a power series be¬
haves like a polynomial. It is differentiable, hence continuous, inside its interval
of absolute convergence, and its derivative and integral can be determined by
differentiating and integrating term by term, respectively.
00 xk
f(x) = X] 17 for all X
k=o k-
Show that f'(x) = f(x) for all x, and deduce that f(x) = ex.
Solution
The given power series converges for all x (by the generalized ratio test—see Example
1), and Theorem 8.23 tells us that it is differentiable for all x. Differentiating term by
8.7 Power Series 549
term, we find
d „ X“2 XJ3 X4
/'(*) = 1 —|— X —(— - 4" -(- 4"
dx 2! 3! 4!
2x 3x2 4x3
= 04-14-1-1-b
2! 3! 4!
x2 x3
= 1 4 X 4 — 4~ — + • • •
2! 3!
- fix)
In Chapter 5, we found that the differential equation fix) — fix) has the general
solution fix) = Cex. Substituting x = 0 into the power series for fix), we obtain
02 03
/(0) = 14-04- — 4- — 4-=1
and by solving the equation 1 = Ce° for C, we find C = 1; therefore, fix) = ex. ■
A power series can be differentiated term by term—not just once, but infinitely
often in its interval of absolute convergence. The key to this fact lies in showing that if
/ satisfies
00
fix) = ^2akXk
k=0
and R is the radius of convergence of the power series on the right, then the derivative
series
fix) = y, kakxk 1
k=l
also has the radius of convergence R. (You are asked to show this in Problem 44.)
Therefore, Theorem 8.23 can be applied to the derivative series to obtain the second
derivative
OO
•TOO = Z!*(*-1)fl*x*-2
k=2
for |x| < R. Continuing in this fashion, we can apply Theorem 8.23 to f"(x) , /(4),
and all other higher derivatives of /.
For example, we know that the geometric series
00
d
4" 2x 4- 3x_ 4-
dx k=1
converges to fix) — for |x| < 1, and the term-by-term second derivative
(1 -x)2
d2 k-1 k-2
dx2 E
*=o
X
dx
J2kx
*=1
— ^jfkik — l)x
k=2
converges to f"ix) — , and so on. These ideas are illustrated in our next
(1 ~xf
example.
550 Chapter 8 Infinite Series
^ (-l)*x2*
WARNING
Do not forget, 0! = 1. /w = E k=0
(2*)!
Solution
First, we use the ratio test to verify that the given power series converges absolutely
for all x.
k-\-1 v2(£+l)
(-i r+ix
r2k+2
[2(k + 1)]! (2k)'.
L = lim = lim — lim = 0
kv2k
k-+oo (-1 Yx fc-»oo[2(Jk + l)]! x2k k^oo (2k + 2)(2k + 1)
(2 k)\
Because L < 1, the series converges for all x. Next, differentiate the series term by
term:
d X2 X4 X6 2x 4x3 6x5
fix) = —— + — -— +
dx 2! +4! “6! + 2! 4! 6!
3 5
X X X
= “IT + 3! ~ 5! + ‘ "
Finally, by differentiating term by term again, we obtain
3 5
// d X X X 3x2 5x4
fix) = -r
dx _l! + 3\ ~ 5? + 1 + ir“^r +
x2 x4
1-+- = -/(*)
2! 4!
-55- xk+x
for — 1 < x < 1
E^m =-'n(1-j:)
k=o K + 1
Solution
OO ]
Integrating the geometric series ) uk —- term by term in the interval
f—^
k=o
1 — w
— 1 < u < 1, we obtain
Jo
r^-du=r
l —U Jo J2uk du
r
[1 u -\- -{-vc* H- • • • ] du
k=0
X2 X3
=x + - + t + ... = e
A A 5°- x*+1
v—>
k= 0
k+ 1
for — 1 < x < 1
We also know
rx du
ln(l — x) for —1 < x < 1
Jo 1 - U
Thus,
r du ^ **+*
ln( 1 — x) = / -=> -- for — 1 < x < 1
Jo
'° u
1* - ” Tk=50(k+ 1
If we check endpoints, we find the series actually converges for — 1 < x < 1.
8.7 Power Series 551
Solution
fx dt
We will use the fact that tan-1
series:
jc
~Jo TT72'
Consider the integrand as a geometric
4 Hi) +i(J) —■
i
239 ~ 5 (259) + 3 (2T9)
552 Chapter 8 Infinite Series
By comparing the spreadsheet in the margin with the one in Figure 8.22, we see that
Machin’s formula for n converges much more rapidly than the Leibniz formula. In¬
deed, the sum of the first seven terms M7 = 3.141592654 already coincides with n for
the first eight decimal places.
k=0
k= 0
4* 8- E k=0
3* In Problems 39^12, find / f(u)du by integrating term by term.
oo
OO i.w.. i \k oo Jo
9- E
ik!(jc - 1)*
5k
io. £ (x - 15)* o° £ oo k
k=0 *=o ln(/: + 1} 39. /(*) = £ (-) 40. /(*) = £ -
*=0 z *=i K
°° k2 2* (x - 3)*
«• Eji'-1)
*=l 2k
22-E *(*+1)
oo
»-E k= 1
*=1
kk
18. E In (lk + 2)
*=i converges for all x. Differentiate term by term to obtain the
oo
00 /(-l)kxk
13*^ oo /"2„\&
(3x)* series
»-E
tzk^2
2«- E 2,+,
&=0 ^
^4 k! cos(fc!x)
oo /■!„ ;,\ t
(ln&)x* lim = 4
” jr* k-fOO A
»-E
*=i
28. £
*=1
Show that the power series
Find the radius of convergence R in Problems 29-34.
PC 00
00
29. £]*2(x + 1)“+1 30. J]2'/*(x - 1)* k-1
Y^kakx
*=i *=i k= 1
m. y:
*=l
k 'xk
kk
32. E
00
-1
*=i
m2xk
(2k)! has radius of convergence R.
8.8 Taylor and Madaurin Series 553
45. Show that if /(x) = ^^a*xA has radius of convergence For what values of x does this power series converge?
*=i
R > 0, then the series (k + 3)!x*
50. Find the radius of convergence for ^
t{ k'-(k + 4)!'
E
*=i
akx kp
^ i - 2 - 3 • • • ki~x)2k~l
51. Find the radius of convergence for > -.
where p is a positive integer, has radius of convergence Rl/p. fe? 1 • 3 - 5 - - - (2Jk - 1)
OO
EX
- converge?
k=\ k + x
52. Let / be the function defined by the power series
Note: This is not a power series. 00 ('_n*x2*+i
” lc fix) = ^2 ntr ■ n, for a11 Show that /"(*) = -fW
47. For what values of x does the series / J —k converge?
i y for all x.
k= 1 A
Consider a function / that can be differentiated n times on some interval /. Our goal
is to find a polynomial function that approximates / at a number c in its domain.
For simplicity, we begin by considering an important special case where c = 0. For
example, consider the function fix) = ex at the point x = 0, as shown in Figure 8.23.
To approximate / by a polynomial function Mix), we begin by making sure that both
the polynomial function and / pass through the same point. That is, M(0) = /(0).
We say that the polynomial is expanded about c = 0 or that it is centered at 0.
M2(x) = 1 + X + y
There are many polynomial functions that we could choose to approximate / at
x = 0, and we proceed by making sure that both / and M have the same slope at
x = 0. That is,
M\0) = /'(0)
The graph in Figure 8.23 shows that we have fix) = ex, so fix) = ex and
/'(0) = /'(0) = 1. To find M, we let
Figure 8.23 Graph of /(x) = ex We see from Figure 8.23 that close to x = c the approximation of / by M\ is good,
and approximating polynomials but as we move away from (0, 1), M\ no longer serves as a good approximation. To
Mi (x) and M2ix) improve the approximation, we impose the requirement that the values of the second
derivatives of M and / agree at x = 0. Using this criterion, we find
so that
M'2 (x) = a\ + 2a2x and M2(x) = 2a2
2a2 =1 or a2 = \
Thus,
Mi_(x) — 1 T x T ^x“
To improve the approximation even further, we can require that the values of the ap¬
proximating polynomials Af3, Af4, • • • , Mn at x = 0 have derivatives that match those
of / at x = 0. We find that
TAYLOR S THEOREM
Instead of stopping at the nth term, we will now approximate a function / by an infinite
series. A function / is said to be represented by the power series
OO
^ak{x - c)k
k=0
OO
functions are extremely useful, but before we can deal effectively with such represen¬
tations, we must answer two questions.
1. Existence: Under what conditions does a given function have a power series repre¬
sentation?
2. Uniqueness: When / can be represented by a power series, is there only one such
series, and, if so, what is it?
for — R < x—c < R. Then there is exactly one such representation, and the coefficients
ak must satisfy
f(k)(c)
ak = for k = 0, 1,2,...
k\
8.8 Taylor and Madaurin Series 555
Proof The uniqueness theorem may be established by differentiating the given power
series term by term and evaluating successive derivatives at c. We start with
OO
and thus, f(c) — a\ + 2a2(Q>) + 3a3(0)2 + • • • = a\. Differentiating once again and
substituting x = c, we find
f"(c)
f"(c) — 2a2 so that a2 = —-—
f(k)(c)
fk)ic) — k\cik so that a * =- □
k\
The question of existence—that is, under what conditions a given function has
a power series representation—is answered by considering what is called the Taylor
remainder function:
This relationship among /, its Taylor polynomial T„(x), and the Taylor remainder
function R„ix) is summarized in the following theorem.
fn+l)iZn)
Rnix) (x - c)"+1
(« + !)!
The formula for R„(x) is called the Lagrange form of the Taylor remainder func¬
tion, after the French/Italian mathematician Joseph Lagrange (1736-1813). There are
other forms for the remainder, principally one developed by Cauchy (see the HISTOR¬
ICAL QUEST in Problem 49 of Section 2.1), but we will consider only the Lagrange
form in this text.
When applying Taylor’s theorem, we do not expect to be able to find the exact
value of zn- However, we can often determine an upper bound for |/?„(x)| in an open
interval. We will illustrate this situation later in this section.
556 Chapter 8 Infinite Series
The uniqueness theorem tells us that if / has a power series representation at c, it must
be the series
/'(c) , , /"(c), ,2 , /w(c),
f(c) + u(x c) H-—- (x-c) H-- (x c)3 +
2! 3!
and Taylor’s theorem says that /(x) is represented by such a series precisely where
the remainder term R„(x) -> 0 as n -» oo. It is convenient to have the following
terminology.
Taylor And Maclaurin Series Suppose there is an open interval I containing c throughout which the function
/ and all its derivatives exist. Then the power series
WARNING tL ,
The uniqueness theorem /'(c) /"(c) , /"'(c)
c) + J (x - c)~ H-(x c)3 +
may be summarized by saying that the /(c) + IT '(X 2! 3!
Taylor series of / at c is the only power
is called the Taylor series of / at c. The special case where c = 0 is called the
series of the form (x — c)k that
Maclaurin series of /:
can possibly represent / on 7. But it
does not say that / is equal to the
power series on I. All we really know nv) 1! 2!
is that if / has a power series
representation at c, then that
representation must be its Taylor series.
Solution
First, note that / is infinitely differentiable at x = 0. We find
You might ask about the relationship between a Maclaurin series and a Maclaurin
polynomial. Figure 8.24 shows the function f(x) = cosx from Example 1 along with
some successive Maclaurin polynomials. We use the notation Mn(x) to represent the
first n + 1 terms of the corresponding Maclaurin series. Note that the polynomials are
quite close to the function near x = 0, but that they become very different as x moves
away from the origin.
Solution
First find the Maclaurin series for f(x)=ex.
f(x) = ex /(0) = 1
f\x) = ex /'( 0) = 1
A comparison of / and M5(a) is shown in Figure 8.25. Notice from the graphs of
these functions that the error seems to increase as a moves away from the origin.
To determine the accuracy, we use Taylor’s theorem
*>=1 + 1 + - + - + — + — + 7?5(1)
2 6 24 120
= 2.716 + R5(l)
The remainder term is given by
e^5
5+1
Rs(1) = ;(1)
(5 + 1)!
for some z5 between 0 and 1. Because 0 < z5 < 1, we have ez$ < e, and since e < 3,
b. Zoom showing the
error at x = 1 it follows that
Solution
Note that / is infinitely differentiable at x — 1. We find
/(*) = lnx /(i) = o
1
fix) = /'(l) = 1
X
-1
/"(*) - 9 /"(!) = -!
X1
2
/"'(*) - /'"(I) = 2
X3
—6
II
/(4)d) - -6
I*
f(k\x) = ( 1} +
xk
—— f f(k\ 1) = (-1 )k+1(k - 1)!
f (-l)w(^ - D* a
k=i ^
EXAMPLE 4 A function defined at points where its Taylor series does not
converge
Show that the function In * is defined at points for which its Taylor series at c = 1 does
not converge.
Solution
We know that In x is defined for all x > 0. From the previous example, we know the
Taylor series for lnx at c = 1 is
k= 1 ^
We find the interval of convergence for this series centered at c = 1 by computing
(-l)*(x - 1)*+1 (-1)*
*+l k T 1
L = lim = lim
k-+oo (-l)*-1^ ~ 1)* k-+oo (-I)*"1
k k
= lim (Mu n- |X — 11
k—nx U+iJ
8.8 Taylor and Maclaurin Series 559
|x - 1| < 1
0 < x < 2
(-1)*-1(0- 1/ -1
0 : E
k=1 k=\
— diverges Opposite of
harmonic series
k-\
(—1)*—1 (2 — 1)^ (-D
= ’=E
k=\
=E
k=1
converges Alternating
harmonic series
Figure 8.26 The graphs of f(x) = lnx Thus, the series converges only on (0, 2], but the function Inx is defined for all
and the Taylor polynomial x > 0. The function is compared with the sixth degree Taylor approximation in Figure
6 (—n*_1 8.26. ■
t^x) = Y,—i
k=i K
—(x_ 1}
The next example introduces a function whose Maclaurin series does not represent
are very close near 1. that function on any interval.
if x = 0
fix) -
if x ^0
Solution
It can be shown that / is infinitely differentiable at 0 and that fa)(0) = 0 for all k.
Therefore, / has the Maclaurin series
0 0 2 0 3
0+-J + -X- + -X- +
which converges to the zero function for all x and thus represents /(x) only at x 0 .
Figure 8.27 Graph of / and its
Maclaurin series y = 0 The graph of / and the Maclaurin series are shown in Figure 8.27.
According to the uniqueness theorem, the Taylor series for / at c is the only power se-
OO
ries in x — c that can satisfy/(x) = ak (x - c)k for all x in some interval containing
k=0
c. The coefficients in this series can always be found by substituting into the formula
f(k)( )
ak = i—_5 but occasionally they can also be found by algebraic manipulation. In
kt . .
several of the examples that follow, we will use the geometric series equation
1 00
-= V uk for \u\ < 1
1 — u ‘ J
1 M *=o
to obtain Maclaurin series for rational functions. The general procedure is illustrated
in Example 6.
Solution
00 |
We want a series of the form akUk that represents-- on an interval containing
to l+x2
0. We could proceed directly using the definition of a Maclaurin series, but instead we
will modify a known series. We know that if |m| < 1 we can write
1 ?
— 1 -f- u -)- u -(-•••
— u 1
so by substitution (u = — x2) we have
1 1
= l- x2+xA x6 +
1 + X2 1 — (—x2)
provided |— x2\ < 1; that is, —1 < x < 1. Hence, by the uniqueness theorem, the
desired representation is
j OO
Solution
1
Our strategy is to rewrite /(*) in the form of a geometric series:
1 — u
5 -2x 8
= -l + Long division
3 -f- 2x 3 + 2x
8
3 1
= -l + -7—r— Writing in the form -
i-(-§*) i-«
OO Q -ik
k=0 3X
, 8 16 32 , 64 ,
= — 1 + --X -(-x2-JT + ■
39 27 81
OO / 2 \*
The convergence set for f(x) is the same as for the series I —x) ; namely,
k=\ k ^ /
\~lx\ < 1
~\ <x < \
a. cosx2 b. cos2*
Solution
a. In Example 1 we found that
u2 u4 u6
cos « = 1-1- + for all u
2! 4! 6!
Therefore, by substituting u = x2, we obtain
2^
2 , U2)2 , (X2)4 (x2)
cos* = 1-—-(- +
2! 4! 6!
*8 *12
for all x
" 2! + 4! 6! +
8.8 Taylor and Maclaurin Series 561
b. For cos2 x we could use the definition of a Maclaurin series, but instead we will use
a double-angle trigonometric identity.
2 1 1
cos x — —I— cos 2x
2 2
1 1 (2x)2 (2x)4 (2x)6
1 - + Let u = 2x.
~ 2 + 2 2! 4! 6!
5 ,.6
11 2x2 23x4 25x
+
= 2 + 2 2T + ~4r " ~6\
_ i _ v2 + -jc4 ——x6 + for all x
“ * + 3 45 +
Solution
For this function, we first use a property of logarithms:
l+x
fix) - In = ln(l + x) — ln(l — x)
1 — X
so that
ln(l + x) = [1 +x - 1] - i[l +x - l]2 + |[1 + * - l]3-
2 3 4
XA JC X
=*~t+t ~7 +"'
2 3
X XJ X4
= _x ______
1 . (i+ (ill, 2
ln2 = 2
3 + 3 5
=
+(+
According to Taylor’s theorem, we find
2 2 /1 \3 2/1
+ ••• +
ln2=3 + 3 3 +5 3 2n + 1 V 3
562 Chapter 8 Infinite Series
where R„ (|) is the remainder term. To estimate the remainder, we note that Rn (|) is
the tail of the infinite series for In 2. Thus,
2n+3 2n+5 2n+l
+ +
2«+3
<
2n+3 r
2n + 3
Some of the more common power series that we have derived are given in Table
8.2. For a more complete list, see the Student Mathematics Handbook. Before we
present this table, there is one last result we should consider. It is a generalization of
the binomial theorem that was discovered by Isaac Newton while he was still a student
at Cambridge University.
»+xy=i+PX++ p(p - m - + +
2! =E 3!
3
k=0
P'-
where for k = 1,2,....
k\(p — k)\
Proof We begin by showing how the Maclaurin series is found. Let /(jc) = (1 +jc)+
/(*) = ( \+x)p /(0) = 1
fix) = pi \+x)p~l f'(0) = p
s=s + s0'-c) + Z^
2!
+ ^^
3!
+ --- + ?K-+
k\
(—00, 00)
u2 u4 u6 (-1 )ku2k
Cosine series C0S„= 1 (—00, 00)
(x — c)2 (x — c)3 .
cos x — cos c — {x — c) sin c-—— cos c H-—— sin c + (—00, 00)
2! 3!
u3 u5 u7 (-1 )ku2k+l
Sine series si„„ = (—00, 00)
(x — c)2 . (x — c)3
sin x = sin c + (* — c) cos c-—— sin c-—— cos c + (—00, 00)
2! 3!
1 , 1 , 1 4 (-l)nxn+1 ,
n 15
n u7 (-1 )ku2k+l
Inverse tangent tan- “ = “--3+7-7 + -+ 2k+l + [-1,1]
series
1 ■ 3 • 5u7
2*-l
n3 1 • 3n5 . 1 • 3 • 5 • • • (2k - 3)«
Inverse sine sin 1 u =u+ + + + ••• + + [-1,1]
2-3 ' 2-4-5 ' 2 • 4 • 6 • 7 ' ' 2•4• 6 ■ • • (2ik - 2){2k - 1)
series
p(p~ 1) 2 , P(P~ 1)(P~2)..3
-nJ + • • • + I In9 +
Binomial series (1 + u)p = 1 + pn + -n +
2! 3!
which produces the Maclaurin series. We could use the ratio test to show that this series
converges on the interval (-1, 1), but to show that it converges to (1 + x)p requires
advanced calculus. -1
The binomial series converges on the open interval -1 < * < 1, but convergence
at the endpoints x = -1 and jc = 1 of this interval depends on the exponent p. In
particular, it can be shown that
Moreover, if p is a nonnegative integer, the series terminates after only a finite num¬
ber of terms (since (^P^j =0 for k > p) and thus reduces to the ordinary binomial
expansion, which, of course, converges for all x. The next example illustrates how the
binomial series theorem can be used.
564 Chapter 8 Infinite Series
Solution
V^=3(l + if
1 + - 0 fx\2 , 5 (i - 0 (i - 2)
1 + 2 V9 +
' 1 1 2 1
x3-
18' 648' + 11,664
1 1 2 1
— 3 + —x X2 + -x3-
6 216 3,888'
Since the exponent p — \ satisfies p > 0, p not an integer, it follows from Theorem
8.26 that the series converges for < 1, that is, for |x| <9.
19. ex + sin x 20. sinx + cosx 0 49. Use term-by-term integration to show that
21. —1— 22. 1 ,a £ 0 00
1 +4x 1 — ax
ln(l +x) = ^-7- f°r 1^1 < 1
23. —— ,a ^0
a +x
24.
a1 + x2
1
,fl^0 k=\ *
52. Use the Maclaurin series expansion of ex to estimate y/e with In this Quest we seek to duplicate Bernoulli's work. He
three-decimal-place accuracy? How many terms are needed? started by writing
53. Use the Maclaurin series expansion of ex to estimate ffe with
three-decimal-place accuracy. How many terms are needed? / dn \
n dx — ndx + x dn — x— dx \
Find the Maclaurin series for the functions given in Problems 54- V dx J
56. Hint: Begin by expressing each function as a sum of partial ( dn \ (x2 d2n x2 d2n \
— ndx + [xdn-x-dxj
fractions. \v.J?dx - v.J7-dx)
2x 1 x2 d2n
55. ( dn \
54. f{x) = —— /(*) = x2 - 3x + 2 = ndx + I x dn — x— dx dx — dx
V dx J \ 2! dx2 2! dx2
71. The functions y0(x) and -A fix) are defined as the following
power series: and
1 +2x
and JiW=S«or+i)B»« (1 -x)(l — 2x)(l — 4x) = E^"
a. Show that AoU) and J\ (x) both converge for all x. show that bn = a2 for all n. Hint: Use a partial fraction de¬
b. Show that J${x) = — J\{x). composition.
These functions are called Bessel functions of the first kind. 75. Journal ProblemThe Pi Mu Epsilon Journal by Robert C. Geb-
* Bessel functions were first used in analyzing Kepler’s laws of 00 x”
planetary motion. These functions play an important role in hardt* ■ For what x does -converge, and what is the
h (2«)!
physics and engineering. sum?
72. Show that J0, the Bessel function of the first kind defined in
Problem 71, satisfies the differential equation *Volume 8, 1984, Problem 586.
x2Jq(x) +xJq(x) + x2/0(x) — 0.
73. For x in the open interval (—1, 1), let
f(x) = x +
Proficiency Examination 20. How do you make an error estimate for an alternating series?
14. State the direct comparison test. b. Test the series > — for convergence.
L—'
k=1 k\
15. State the limit comparison test.
16. State the zero-infinity limit comparison test. c. Discuss the similarities and/or differences of parts a and b.
17. State the ratio test. In Problems 33-37, test the given series for convergence.
18. State the root test. 1
33. £
19. What is the alternating series test?
k=2 k In k 34-£if
k=1 K
Chapter 8 Review 567
oo
1 _ f^3k2-k+\
35. •£ 32. £>'* 33. V -
k=2
(In k)x/k 36' (1-2*)* 11=0
11111
1 1 (-1)',+1 OO nk OO
38. 39. y
40. Find the Taylor series for f(x) = — at c = 4. A:(ln A:)11 f^_k(\nk)2
3 2
OO OO j
40.
Supplementary Problems
Determine whether each sec/uence in Problems 1-16 converges or OO 00
y, * - 1
diverges. If it converges, find its limit. 42. 43. y_-—
(-D*
123,456,789
54.
,s- £4G -EG)
*=1
'• Et-'J'-iaVr)
K=1 N
oo
1 oo ^_jjika+l)/2
*=2 K 1
21' E
t=i
4*2 _ , E—«—
55- E—*—
- z1=1
oo °o ek + 3k~l
Find the interval of convergence for each power series in Problems
22~ E
A=0 ¥ Hi »E 6i+1 56- 67.
00 OO / k OO
1 1
56. V 57. y k(x
24 E(-1)t+‘G+rrr)
*=i “-E ln(fc + ) In A: k—2 L
1
k= 1 k= 1
l)*
00 oo oo PC
k InJfcU2)*
26. y;
GHt)4*
k= o
21 E
Ef (Jt + l)(* + 2)(* + 3) ■
58. £
A=1 *(*+1)
5». £
fc=l ~7T
7c.s7 the series in Problems 28-45for convergence. oo OO
(~\)kxk
6o.£^±iE 6i. jr
^5kk\ - ^ 1 *=i kk
29. Y] - k=\
28- Eir h vf+4
A=1 K oo
,, J2, (-1)‘(2» - 1)‘ Ivx 93*
OO 1
^ *! 61 E-f5-
30- E v *=o
34 E sG,
k=\
568 Chapter 8 Review
/ l\”
64. 65. £ 22!,- b. Use part a to show that f 1 H— 1 < 3 for all n.
^
k=1 ft + 3)! >1=1
Yl\
66.
oo
£ (-9 1)"*"
- 1
n2
c. Use part a to show that j ^1 + | is an increasing se¬
rt=l quence. Then use the BMCT to show that this sequence
3 5 7
* XD x‘ (-l)i+1x2*-' converges.
67. x — — -F — — — + + +
3! 5! 7! (2k — 1)! 81. HISTORICAL QUEST
Find the Maclaurin series for each function given in Problems Ramanujan was that rarest of
68-73. mathematicians, an instinctive
genius with virtually no for¬
68. f(x) = x2e 3jr 69. f(x) = x3 sinx
mal training. Like his Hindu
70. f(x) = 3X 71. f(x) = x2 + tan
predecessor, Bhaskara (see the
5 + 7x llx - 1 Historical Quest in Problem 55
72. fix) = 73. f(x) =
1 + 2x - 3x2 2 + x — 3x2 of Section 4.4), Ramanujan had
74. Compute the sum of the series an uncanny instinct for numeri¬
cal "truth" and conceived of his
1 1 1_ SRINIVASA RAMANUJAN
results in much the same way
1887-1920
1 ~~ 2 + 2 ! 3! 2 + that a sculptor "sees" a statue
in a raw block of stone. In his short life, he initiated new ways
correct to three decimal places. of thinking about number theory and made conjectures that
75. Find a bound on the error incurred by approximating the sum are the subject of mathematical inquiry to this day.
of the series A story related by his friend and mentor, the eminent British
number theorist G. H. Hardy (1877-1947), serves to illustrate
1 2 3
1 — — -{- — — — -}-•-• the resonance between Ramanujan's mind and the concept of
2! 3! 4! number. Ramanujan was ill and Hardy came to visit him in the
by the sum of the first eight terms. hospital. At a loss for how to begin the conversation, Hardy
idly remarked that he had arrived in a taxi with the "dull"
76. Use geometric series to write 12.342132 as a rational number.
number 1729. Ramanujan immediately grew excited and ex¬
77. A ball is dropped from a height of A feet. Each time it drops claimed, "No, Hardy, it is a very interesting number, for it is
h feet, it rebounds 0.8h feet. If it travels a total distance of 20 the smallest integer that can be expressed as the sum of cubes
feet before coming to a stop, what is A? in two different ways." (1729 = l3 + 123 = 93 + 103.)
78. Find the first three nonzero terms of the Maclaurin series for In a letter from Ramanujan to G. H. Hardy at Cambridge
tanx and then use term-by-term integration to find the first University,* Ramanujan stated
three nonzero terms of the Maclaurin series for In |cosx|.
79. Spy Problem The Spy, who is smarter than he looks, aborts 1 -3 1- 3-5
+9 - 13
his mission to the village after six days of duck plucking (re¬ 5,I 2L4 2- 4-6 n
call Problem 62, Section 7.4). As he races away, he is con¬
fronted by four of Blohardt’s thugs and soon finds himself in a Hardy spent a great deal of time wondering how this sum could
gun battle using a revolver he managed to steal from Blabba. be equal to 2/n. m
While dodging bullets, he recalls from his math class at Spy OO
Academy that if the probability of a bullet being “good” is p, a. Find ak so this sum can be expressed as ^ ak.
then on average, the number of bullets fired before a bad one i=i
OO
+ 3! ('
+ 7 ('\
* Quantum, March/April 1998.
Chapter 8 Review 569
ax = aNe = (ae)N 90. Use series to find In 1.05 correct to five decimal places.
= (1 +ke)N 91. Use a Maclaurin series to find a cubic polynomial that approx¬
imates e2* for |jc| < 0.001. Find an upper bound for the error
kx in this approximation.
= 1 +
~N x3
92. For what values of x can sinx be replaced by x ——if the
kx\ N(N - 1) (kx
= 1 + N allowable error is 0.0005?
N 2!
x2
N(N — \)(N — 2) (kx 93. For what values of x can cos x be replaced by 1 ——if the
+ allowable error is 0.00005?
3! ¥ 1 +'
2 tan x
94. Find the Maclaurin series for f(x) — --r— .
= 1 + kx + 1 NiN~l)k2x2 J 1 + tan2 x
2! N2
1 N(N — l)(N — 2) 95. Modeling Problem When we studied carbon dating in Section
+ k3x3 + • • 5.6, it was pointed out that radiocarbon methods apply only
3! N3
to objects that are not too “old” (roughly 40,000 years). To
Because N is infinitely large, Euler assumed that date older objects, radioactive isotopes other than 14C are of¬
ten employed. Regardless of the radioactive substance used,
N-\ N-2 it can be shown that the ratio of stable isotope to radioactive
1 =
N N isotope at any given time may be modeled by
Substitute these values into the given derivation and also sub¬
S(0) j _ e(in2)tjk
stitute e for a to obtain
R(t)
1 1 1
where R{t) is the number of atoms of radioactive material at
e = 1 + TT + 2T + 3!H time t, S(t) is the number of atoms of the stable product of
Euler calculated e to 26 places: radioactive decay, 5(0) is the number of atoms of stable prod¬
uct initially present (at t = 0), and X is the half-life of the
e « 2.71828182845904523536028 radioactive isotope.*
IV a. Use the first two terms of the Maclaurin series for ex to ap¬
Show that Euler’s derivation is equivalent to lim I 1 + - . proximate g(ln2)'A> and then solve the modeling equation
n->+oo
writing the first four nonzero terms. Hint: Use the binomial as a rational number.
theorem and then integrate term by term. 97. Putnam Examination Problem Find the sum of the convergent al-
86. Estimate the value of the integral (-1)"+1
temating series 'Y
3« — 2
f0A dx n=0
98. Putnam Examination Problem For positive real x, let
Jo y/1 + X3
Ona)* k
for all x V-* Rk(2)
■* = £ k\
k=0 (k log2 k)2
1 *This model is discussed in detail in the article, “How Old Is the Earth?”
I = - - -I(jc - c) + ~(x - c)2 - -(x - cY + by Paul J. Campbell, UMAP Modules 1992: Tools for Teaching, Lexing¬
X c c2 c3 c*
ton, MA: Consortium for Mathematics and Its Applications, Inc., 1993,
89. Use series to find sin 0.2 correct to five decimal places. pp. 105-137.
Group Research Project*
Elastic Tightrope
This project is to be done in groups of three or four students. Each group will submit a
single written report.
S uppose a cockroach starts at one end of a 100-m tightrope and runs toward the
other end at a speed of 1 m/s. At the end of every minute, the tightrope stretches
uniformly and instantaneously, increasing its length by 100 meters each time.
Thus, after the first minute, the roach has traveled 60 meters and the length of the
tightrope is 200 m. After the second minute, the roach has traveled another 60 meters
and the tightrope has become 300 m long, and so on.
Does the roach ever reach the other end? If so, how long does it take?
Your paper is not limited to the following questions but should include these con¬
cerns. Suppose it is you standing on the elastic rope b meters from one end and c
meters from the other end. Then suppose the entire rope stretches uniformly, increas¬
ing its length by d meters. Only the tightrope stretches; units of length and time remain
constant. How far are you from each end? You might also try to generalize the roach
problem.
-Augustus De Morgan
Graves’ Life ofW.R. Hamilton
(New York: 1882-1889), p. 249
This group project is courtesy of David Pengelley, New Mexico State University, Las Cruces.
570
Chapters 6-8 Cumulative Review 571
5. J d In 6.
/ sin2 x cos3 x dx
7.
/ 1 4-x2
dx 8.
/ sec3/2 x tan x dx
9. x2 dx
x2 + 17x - 8
10 .
/ l-x2dx
1 +x2 Figure 8.28 Submerged plate
11. Let /(x) 25. Find the arc length of the curve y = |x2/3 on the interval
(2x + 3)(x - l)2
a. Express /(x) as a partial fraction sum. [*.»]-
26. Let R be the region bounded by the x-axis, the curve
b. Find / f{x)dx.
y = ■ —, and the lines x = 3 and x = 8. Find the volume
POO
>/x + 1
c. Either evaluate the improper integral / fix) dx or show of the solid generated when R is revolved about
12. In each of the following cases, either evaluate the given im¬ 27. Solve each of the given first-order linear differential equa¬
proper integral or show that it diverges. tions.
ctt/2 a. y
xe■'~2x dx b.
»•// sec x dx x 1 + x2
-/ b. y’ 4- (tanx)y = tanx, where y = 3 when x = 0
In Problems 13-18, test the given series for convergence. 28. A ball is dropped from a height of 20 ft. Each time it bounces,
^ k23k+l 14.
^ k + 3k3/2 it rises to | of its previous height. What is the total distance
2—j Ak f-f 5k2 - k + 3 traveled by the ball?
k= 0 k=\
y, (-1 )k+'k
16.
^ f2k + 5\k 29. Find the Maclaurin series for each of the following functions.
h U-2J
a. f(x) = sinx2
h
oo
*+i 00 12
b. f{x) — s[e*
3x -2
Eke~1’
k= 1
18. yL
hk'
c. fix) =
5 + 2x
In Problems 19 and 20, find the sum of the given convergent series. 30. A tank contains 15 lb of salt dissolved in 70 gallons of water.
oo / /} \ 2k—1 oo 9
Suppose four gallons of brine containing 3 lb/gal of salt run
into the tank every minute and the mixture (kept uniform by
w'E5(-r)
*=1 \ - /
20-E
k=\ (*+!)(* +3) stirring) runs out of the tank at the rate of 2 gal/min.
In Problems 21 and 22, find the convergence set of the given power a. Set up and solve a differential equation for the amount of
series. salt Sit) in the tank at time t.
00 rk (—r)k b. Suppose the capacity of the tank is 100 gal. How much salt
21. 22- E } is in the tank at the time it just begins to overflow?
k+ 1
*=i k3 *=i
.
CONTENTS
9.1 Vectors in R2
Introduction to vectors
Vectors in the Plane and in Space
Vectors in component form
Standard representation of vectors
in the plane
9.2 Coordinates and Vectors in R3
Coordinates in R3
Graphs in R3: spheres and PREVIEW
cylinders
In this chapter, we focus on various algebraic and geometric aspects of vector repre¬
Vectors in R3
sentations. Then in Chapter 10, we see how vectors can be combined with calculus
9.3 The Dot Product to study motion in space and other applications.
Definition and basic properties of
the dot product
Angle between vectors
Direction cosines PERSPECTIVE
Projections Suppose a person is pulling a block of ice across the floor. The effect of this effort
Work as a dot product on the block of ice is dependent on the direction it is pulled, as well as the magnitude
9.4 The Cross Product of the force that is applied when pulling. In this chapter, we consider the concept
Definition and basic properties of of a vector, which is defined as a quantity with both direction and magnitude. In
the cross product Problem 43 in Section 9.3 we consider the amount of work done by dragging a large
Geometric interpretation of the block of ice, which we compute using vectors and vector operations.
cross product In general, a scalar quantity is one that can be described in terms of magnitude
Applications of the cross product: alone, whereas a vector quantity requires both magnitude and direction. Force,
area and torque velocity, and acceleration are common vector quantities, and an important goal of
The scalar triple product this chapter is to develop methods for dealing with such quantities.
9.6 Planes in R3
Forms for the equation of a plane
in R3
Vector methods for measuring
distances in R3 Finding the amount of work to pull a box across
the floor can be formulated as a vector calculation.
9.7 Quadric Surfaces
Methods for graphing quadric
surfaces
A catalog of quadric surfaces
Chapter 9 Review
Group Research Project: Star Trek
573
574 Chapter 9 Vectors in the Plane and in Space
9.1 Vectors in R2
IN THIS SECTION introduction to vectors, vectors in component form, standard representation of vectors
in the plane
INTRODUCTION TO VECTORS
A vector is a quantity (such as velocity or force) that has both magnitude and direction.
We sometimes represent a vector as a directed line segment, an “arrow” with initial
point P and terminal point Q. The direction of the vector is that of the arrow, and
its magnitude is represented by the arrow’s length, as shown in Figure 9.1. We will
indicate such a vector by writing PQ in boldface type, but in your work you may write
an arrow over the designated points: The order of letters you write down is
important: PQ means that the vector is from P to Q, but QP means that the vector is
from Q to P. The first letter is the initial point and the second letter is the terminal
point.
Q (terminal point)
i P (initial point)
a. The triangle rule for b. The parallelogram rule c. The difference rule
adding vectors u and v for vector addition
of vector addition. In particular, to add the vector v to the vector u, we place the initial
point of v at the tip (terminal point) of u and define the sum, also called the resultant
u + v, to be the vector that extends from the initial point of u to the terminal point of v.
Equivalently, u + v is the diagonal of the parallelogram formed with sides u and
v, as shown in Figure 9.3b. The difference u — v is just the vector w that satisfies
v + w = u, and it may be found by placing the initial points of u and v together and
extending a vector from the terminal point of v to the terminal point of u (see Figure
9.3c). Note that u + v = v + u. The vector v — u may be found similarly and has the
same magnitude, but points in the opposite direction to u — v.
Using the geometric relationships shown in Figure 9.5, the vector PQ with initial
point P(a, b) and terminal point Q(c, d) equals the vector OR, where (9(0, 0) is the
origin and R is the point with coordinates (c — a, d — b).
Q(c,d)
d-b
Figure 9.5 Given P(a, b) and Q(c, d), the vector PQ is (c - a,d - b).
Vector properties are easily represented when vectors are given in component
form. In particular, we have
These formulas may be verified geometrically. For instance, the rule for multiplication
by a scalar can be obtained by using the relationships in Figure 9.6a, and Figure 9.6b
illustrates the rule for vector addition.
a. u + v b. |u c. 3u —
Solution
a. u + v = (2, -3) + (-1, 7) = (2 + (-1), -3 + 7) = (1, 4)
b. |u = I (2, -3) = (2(2), |(-3» = (§, =*)
c. 3u — = 3 (2, —3) — \ (—1, 7)
= (6, —9) + (|, —Scalar multiplication
= (6 + j, —9 — Vector addition
= (¥,-¥) Simplify ■
s(u + v) = 5U 4- 5V
You are asked to prove other of these properties in the problem set. □
Many results from geometry can be proved easily by using vector notation and
vector properties. An illustration of this procedure is proved in Example 2.
Solution
Consider A ABC, and let P and Q be the midpoints of sides AC and BC, respectively,
as shown in Figure 9.7.
C
AB = AP + PQ + QB
= ±AC + PQ - BQ AP = iAC and QB = -BQ
= i(AB + BC) + PQ - |BC AC = (AB + BC) and BQ = ^BC
= ^AB+ iBC + PQ- ^BC
= iAB + PQ
iAB = PQ Subtract ^AB from both sides. ■
When a vector u is represented in component form u = (mi, m2), its length is given
by the formula
I|U|| = yju\+u\
578 Chapter 9 Vectors in the Plane and in Space
This is a simple application of the Pythagorean theorem as shown in Figure 9.8a. An¬
other important relationship involving the length of vectors is the triangle inequality
||u +v|| < ||u|| + IIv||
for any vectors u and v. Equality will occur precisely when u and v are nonnegative
multiples of one another (that is, when u and v have the same direction). To establish
the inequality, we observe that if u and v are positioned as two sides of a triangle in
the plane, then the third side, u + v, has length ||u + v|| and is “shorter” than the sum
|| u || + || v|| of the lengths of the other two sides, as shown in Figure 9.8b.
Wj
Solution
Begin by drawing a diagram, like the one shown in Figure 9.9.
IIVII = 12
Let B be the velocity vector of the boat in the direction of the angle 9. If the river’s
current has velocity C, the given information tells us that ||C|| = 5 mi/h and that C
points directly south. Moreover, because the boat is to cross the river from east to west
in 20 min (that is, \ hr), its effective velocity after compensating for the current is a
vector V that points west and has magnitude
WIDTH OF THE RIVER 4 mi
= 12 mi/h Since d — rt
TIME OF CROSSING I h
3
The effective velocity V is the resultant of B and C; that is, V = B + C. Because
V and C act in perpendicular directions, we can determine B by referring to the right
triangle with sides ||v|| = 12 and ||C|| =5 and hypotenuse ||B||. We find that
Thus, the boat should travel at 13 mi/h in a direction of approximately 0.3948 radian. In
navigation, it is common to specify direction in degrees rather than radians (0.3948 ~
22.6°) and to measure the direction as an angle from either north or south. In this case,
it is 90° - 22.6° = 57.4°, or N57.4°W (that is, 57.4° west of north).
A unit vector is a vector with length 1, and a direction vector for a given nonzero
vector v is a unit vector u that points in the same direction as v. Such a vector can be
found by dividing v by its length ||v||; that is,
v
u =-
IIV ||
Solution
The vector v has length (magnitude) ||v|| = y/22 + (—3)2 = VJ3. Thus, the required
direction vector is the unit vector
v (2, -3) 1 / 2 -3 \
The unit vectors i = (1,0) and j = (0, 1) point in the directions of the positive x- and
y-axes, respectively, and are called standard basis vectors. Any vector v = (t>i, V2)
in the plane can be expressed as a linear combination of the vectors i and j, because
Solution
Using Theorem 9.1, we find that
2u + 5v — w = 2(3i + 2j) + 5(-2i + 5j) - (i - 4j)
= [2(3) + 5(—2) - l]i + [2(2) + 5(5) - (—4)]j
= —5i + 33j ■
580 Chapter 9 Vectors in the Plane and in Space
Solution
The component form of PQ is
fj
Find the magnitude and direction of the additional force F3 that must be applied to
keep the body at rest.
Solution
According to the given information, we have
(—3 + | + «) i + (— | + b'j j = Oi + Oj
Because the standard representation is unique, we must have
—3 T ^6 T1 a- -
— n
0 and ~l+b=0
a = b=l
The required force is F3 = f i + f j. This is a force of magnitude
5. P(3, -1), 0(7.2) 6. P(5, -2), 0(5, 8) 13. i + j 14. + |j 15. 3i — 4j 16. —4i + 7j
9.1 Vectors in M2 581
Let u = (—3,4) and v = (1, — 1). Find scalars s and t so that the 47. Two forces F| = 3i + 4j and F2 = 3i — 7j act on an object.
given equation in Problems 17-20 is satisfied. What additional force should be applied to keep the body at
17. xu + t\ = (6, 0) 18. ^(0, —3) + fu = v rest?
19. s\ + t(—2, 1) = u 20. 5U + (8, 11) = t\ 48. Three forces Fi = i — 2j, F2 = 3i - 7j, and F3 = i + j act on
an object. What additional force F4 should be applied to keep
Suppose u = 3i — 4j, v = 4i — 3j, and w = i + j. Represent each
the body at rest?
of the expressions in Problems 21-24 in standard form.
49. A river 2.1 mi wide flows south with a current of 3.1 mi/h.
21. 2u + 3v - w 22. ^(u + v) — |w
What speed and heading should a motorboat assume to travel
23. ||v||u + ||u||v 24. 11 u||||v||w across the river from east to west in 30 min?
Find all real numbers x and y that satisfy the vector equations 50. Four forces act on an object: F, has magnitude 10 lb and acts
given in Problems 25-28. at an angle of | measured counterclockwise from the positive
25. (x - y - l)i + (2x + 3y - 12)j = 0 x-axis; F2 has magnitude 8 lb and acts in the direction of the
26. xi - 4y2j = (5 - 3y)i + (10 - 7x.)j vector j; F3 has magnitude 5 lb and acts at an angle of 4tt/3
measured counterclockwise from the positive x-axis. What
27. (.x2 + y2)i + yj = 20i + (x + 2)j
must the fourth force F4 be to keep the object at rest?
28. (y - l)i + yj = (logx)i + [log 2 + log(x + 4)]j
Q 51. Use vector methods to show that the diagonals of a parallelo¬
In Problems 29-32, find a unit vector u with the given character¬ gram bisect each other.
istics.
52. In a triangle, let u, v, and w be the vectors from each vertex
29. u makes an angle of 30° measured from the positive x-axis. to the midpoint of the opposite side. Use vector methods to
30. u has the same direction as the vector 2i — 3j. show that u + v + w = 0.
31. u has the opposite direction of the vector —4i + j. 53. Prove that the medians of a triangle (see Figure 9.12) intersect
32. u has the direction of the vector from P(— 1,5) to Q(7, —3). at a single point by completing the following argument.
C
55. Let A, B, C, and D be any four points in the plane. If M and
N are midpoints of AC and BD, show that
MN = |(AB + AD + CB + CD)
COORDINATES IN R3
We have already considered ordered pairs in R2, and because we exist in a three-
dimensional world, it is also important to consider a three-dimensional system. We
call this three-space and denote it by R3. We introduce a coordinate system to three-
space by choosing three mutually perpendicular axes to serve as a frame of reference.
The orientation of our reference system will be right-handed in the sense that if you
stand at the origin with your right arm along the positive x-axis and your left arm along
the positive y-axis, as shown in Figure 9.14, your head will then point in the direction
of the positive z-axis.
To orient yourself to a three-dimensional coordinate system, think of the x-axis
and y-axis as lying in the plane of the floor and the z-axis as a line perpendicular to the
floor. All the graphs that we have drawn in the first eight chapters of this book would
now be drawn on the floor. If you orient yourself in a room as shown in Figure 9.15,
you may notice some important planes. Assume that the room is 25 ft x 30 ft x 8 ft
Figure 9.14 A “right-handed” and fix the origin at a front corner (where the board hangs).
rectangular coordinate system
for M3
yz-plane
.rz-plane
Figure 9.15 A typical classroom; assume the dimensions are 25 ft by 30 ft with an 8-foot ceiling
The xy-, xz-, and vz-planes are called the coordinate planes. We will obtain
equations for planes in space after we discuss vectors in R3. However, in beginning to
visualize objects in R3, we do not want to ignore planes, because they are so common
(for example, the walls, ceiling, and floor in Figure 9.15). We will show that the graph
of Ax + By + Cz = D is a plane if A, B, C, and D are real numbers (not all zero).
Points in R3 are located by their position in relation to the three coordinate planes and
are given appropriate coordinates. Specifically, the point P is assigned coordinates
(a, b, c) to indicate that it is a, b, and c units, respectively, from the yz-, xz-, and
xy-planes.
Solution
The points are plotted as shown in Figure 9.16.
In R2, the distance from the origin (0, 0) to the point (a, b) is d — Va2 + b2, and
in R3, the distance from (0, 0, 0) to (a, b, c) is d = -Ja2 + b2 + c2, as shown in Figure
9.17. We can now summarize.
Distance Formula
The distance \P\P2 \ between P\(x\, yi, zi) and P2U2, yi, Z2) is
= aJ a2 + b2 + c2 = 6/l9 ■
In R3, the collection of all points Qc, y, z) whose coordinates satisfy a given equation
is called the graph of an equation. Geometrically, the graph of an equation in .v, y, z
is a surface in R3. In this section, we will examine two kinds of surfaces, spheres and
cylinders. Later in the chapter, we will examine a variety of other surfaces.
584 Chapter 9 Vectors in the Plane and in Space
Spheres A sphere (see Figure 9.18) is defined as the collection of all points located
a fixed distance (the radius) from a fixed point (the center).
In particular, if P(x,y,z) is a point on the sphere with radius r and center
C(a, b, c), then the distance from C to P is r. Thus,
If you square both sides of this equation, you can see that it is equivalent to the equation
of a sphere displayed in the following box. Conversely, if the point (x,y,z) satisfies
an equation of this form, it must lie on a sphere with center (a, b, c) and radius r.
Figure 9.18 Graph of a sphere
with center (a, b, c), radius r
is a sphere with center (a, b,c) and radius r, and any sphere has an equation of
this form. This is called the standard form of the equation of a sphere (or
simply standard-form, sphere).
Solution
By completing the square in both variables x and y, we have
(x2 + 4x) + (y2 - 6y) + z2 = 3
(x + 2)2 + (y — 3)2 + z2 = 16
Comparing this equation with the standard form, we see that it is the equation of a
Figure 9.19 Graph of a sphere sphere with center (-2, 3, 0) and radius 4, as shown in Figure 9.19.
Generatrix Generatrix
L
We will deal primarily with cylinders in which the directrix is a conic section and
the generatrix L is one of the coordinate axes. Such a cylinder is often named for the
type of conic section in its principal cross sections and is described by an equation
involving only two of the variables x, y, z. In this case, the generating line L is parallel
to the coordinate axis of the missing variable. Thus,
+ y2 = 5 is a circular cylinder with L parallel to the z-axis
Elliptic
cross section
VECTORS IN R3
A vector in R3 is a directed line segment (an “arrow”) in space. The vector P1P2 with
initial point Pi(xi, yi, zi) and terminal point P2&2, yi, Z2) has the component form
analogous to the R2 case
a. The standard basis vectors in R3 b. The vector from O to Q{d\, a2, a2)
is OQ = a,i + a2 j + 03k.
Solution
We have
As in R2, if v is a given nonzero vector in R3, then the unit vector u that points in
the same direction as v is
Thus,
As in R2, two vectors in R3 are parallel if they are scalar multiples of one another.
Similarly, nonzero vectors u and v are parallel if and only if u = sv for some nonzero
scalar s.
9.2 Coordinates and Vectors in R3 587
Solution
Let Q have coordinates (a\,a2, a3). Then
a\ — 1 = 2s a2 = —3.v a3 + 3 = 6s
a\ = 2s + 1 a3 = 6s — 3
There are two points that satisfy the conditions for the required terminal point Q:
(y, — |, — f) and (f, f, — y). ■
33. Find an equation for a sphere, given that the endpoints of a In Problems 51 and 52, determine whether the given points are
diameter of the sphere are (1,2, —3) and (—2, 3, 3). col linear (that is, lie on the same line). Note: For three points A,
B, and C to be collinear, AC must be a multiple of AB.
Evaluate the expressions given in Problems 34-37.
51. (2, 3, 2), (-1,4,0), (-4, 5,-2)
34. ||i + j + k|| 35. Hi — j + k||
52. (3,0, 3), (2,-3, 5), (4, 3, 1)
36. ||2i + j — 3k||2
53. A 150-lb tennis official is sitting on a three-legged judge’s
37. ||2(i — j + k) — 3(2i + j — k)||2
stool. Suppose a coordinate system is chosen with the of¬
Let v = i — 2j 4- 2k and w — 2i + 4j — k; find the vector or scalar ficial at point P(0, 0,6) and the feet of the stool’s legs at
requested in Problems 38—41. A(0, —2, 0), B( — \f3, 1,0), and C(V3, 1,0), as shown in Fig¬
38. 2||v|| — 31|w|| 39. ||v||w ure 9.23. Find the force exerted on each of the three legs.
40. ||2v — 3w|| 41. ||v-w||(v + w)
Determine whether each vector in Problems 42-45 is parallel to
u = 2i - 3j + 5k.
42. v = (4, -6, 10) 43. v= (-2,6,-10)
44. v= (1, —|, 2) 45. v= (-1, |,-f)
Q The vertices A, B, and C of a triangle in R3 are given in Problems
46-49. Find the lengths of the sides of the triangle and determine
whether it is a right triangle, an isosceles triangle, both, or neither.
46. A(3, —1, 0), 5(7, 1, 4), C(l, 3,4) Figure 9.23 Tennis judge
47. A( 1, 1, 1), B(3, 3, 2), C(3, -3, 5)
Q 54. Let P(3, 2, —1), Q(—2, 1, c), and R(c, 1,0) be points in R3.
48. A(l, 2, 3), B{—3, 2, 4), C(l, —4, 3) For what values of c (if any) is PQR a right triangle?
49. A(2, 4, 3), B(-3, 2, -4), C(-6, 8, -10) 55. Find the point P that lies | of the distance from the point
50. Let u = (—1, 1,2), v = <0, 2, —3), and w = (5,-1,0). Find A(—1, 3, 9) to the midpoint of the line segment joining points
a vector q so that 2u — v + 5q = 3w. B(—2, 3, 7) and C(4, 1, -3).
In the first two sections of this chapter, you have learned how to add vectors and mul¬
tiply a vector by a scalar. We now turn our attention to the product of two vectors. The
dot (scalar) product and the cross (vector) product are two important vector opera¬
tions. We will examine the dot product in this section and the cross product in the next
section. The dot product is also known as a scalar product because it is a product of
vectors that gives a scalar (that is, real number) as a result. Sometimes the dot product
is called the inner product.
Dot Product The dot product of vectors v = ati + a^ + a^k and w = bfi + b2j + b2k is the
scalar denoted by v • w and defined by
The dot product of two vectors v = a\i + «2j and w = b\ \ + b^j in a plane is given by
a similar formula with a3 = b2 = 0, namely,
v • w = a\b\ + a2b2
9.3 The Dot Product 589
Before we can apply the dot product to geometric and physical problems, we need
to know how it behaves algebraically. A number of important general properties of the
dot product are listed in the following theorem.
Zero product 0 • v = 0
Commutativity v • w = W • V
Distributivity u-(v + w) = u- v + u- w
Magnitude of a vector
yja] + aj + a^j = a\ + aj + a] = v • v
Distributivity
U • (V + W)
and
Thus, u • (v 4- w) = u ■ v + u • w.
The angle between two nonzero vectors v and w in R2 or R3 is defined to be the angle 0
with 0 < 0 < n that is formed when the vectors are in standard position (initial points
at the origin), as shown in Figure 9.24.
The angle between two vectors plays an important role in certain applications and
Figure 9.24 The angle
between two vectors may be computed by using the following formula involving the dot product.
590 Chapter 9 Vectors in the Plane and in Space
B{bi, b2, b3) Proof Suppose v = a\\ + a2j + a2k and w = byi + + b2k, and consider AAOB
with vertices at the origin O and the points A(cq, a2, a2) and B(b\, b2, b2), as shown
in Figure 9.25.
Note that sides OA and OB have lengths
|v||2 + ||w||2 V — w
cost? = Solve for cos 0.
2||v|| || w|
Solution
Draw A ABC and label the angle formed at A as a, as shown in Figure 9.26. The angle
a is the angle between vectors AB and AC, where
The formula for the angle between vectors is often used in conjunction with the dot
product. If we multiply both sides of the formula by ||v|| ||w||, we obtain the following
alternate form for the dot product formula.
where 0 is the angle (0 < 0 < re) between the vectors v and w.
9.3 The Dot Product 591
v•w = 0
Proof If the vectors are orthogonal, then the angle between them is therefore,
^ What This Says The orthogonal vector theorem allows you to show that
two vectors are orthogonal by finding the dot product of the vectors and showing
this dot product is 0. On the other hand, if you know the vectors are orthogonal,
then it follows that the dot product of the vectors is 0. In particular, i j = j • k
= k • i = 0.
u = 3i + 7j - 2k v = 5i - 3j - 3k w =j -k
Solution
u v = 3(5) + 7(—3) + (—2)(—3) = 0; u and v are orthogonal.
u w = 3(0) + 7(1) + (—2)(—1) = 9; u and w are not orthogonal.
v w = 5(0) + (—3)(1) + (—3)(-l) = 0; v and w are orthogonal. ■
DIRECTION COSINES
The direction of a vector in R2 positioned with its initial point at the origin may be
specified by determining its angle of inclination; that is, the angle (p is the smallest
positive angle the vector makes with the positive x-axis, as shown in Figure 9.27a. In
R3, it is common practice to measure the direction of a nonzero vector v in terms of
angles cc, f, and y between v and the positive coordinate axes as shown in Figure 9.27b.
These angles are called the direction angles of v, and their cosines are the direction
cosines of v.
The direction cosines of v = (vi, i>2, V3) can be computed by taking the dot
product of v with the three standard basis vectors, i = (1,0, 0), j = (0, 1,0), and
k = (0, 0, 1). For example, to find cos a, note that
V\
v = v • i = IIv|| llill cosa, so that coscr =-
IMI
and, similarly,
a v2 , v3
cos p —- and cos y — —
IMI l|v|
It can be shown (Problem 53) that
This identity can be used to check whether you have computed the direction cosines
correctly.
Solution
vi -2
__
coso: = -0.3244428 a ~ cos-1 (-0.324428) % 109°
IMI ~~ a/38
l>2 3
cos P — 0.4866642 j3 « cos-1 (0.4866642) % 61°
IMI V38
v3 5
cos y - 0.8111071 y « cos_1(0.8111071) % 36°
x/38
and
Figure 9.28 The vector
v = —2i + 3 j + 5k cos2 a + cos2 /3 + cos2 y —
forms the angles 109°, 61°, and 36° 3 ) +(tW +(^l) =1
with the coordinate axes These direction angles are shown in Figure 9.28.
PROJECTIONS
Let v and w be two vectors in R3 drawn so that they have a common initial point,
as shown in Figure 9.29. If we drop a perpendicular from the tip of v to the line
determined by w, we determine a vector called the vector projection of v onto w,
which we have labeled u in Figure 9.29. To find a formula for the vector projection,
note that u = fw for some scalar t and that \ — tw is orthogonal to w. Thus,
(v — tw) • w = 0
v • w = t(w ■ w)
V w
t —-
w w
Figure 9.29 The vector u is the and the vector projection is
projection of v onto w.
u = w
V• w\
ProJw v = w
Nliv
9.3 The Dot Product 593
projw v
w v • w
where is the direction vector for w. The scalar is called the scalar
WARNING , /v'w\ . ||w|| V ||w|| /
Note that (-) w is projection of v onto w, or the component of v in the direction of w, and is denoted by
Vw • w/
compw v. Thus,
not the same as ( — ) w; you cannot
V\v/ V • w
"cancel" the vector w. Remember, comp v = —— - ||v|| cos 9 Since v w = ||v|| ||w|| cos#
W
v • w and w • w are numbers, whereas
v/w is not defined. so compw v > 0 when 0 < 9 < y and compw v < 0 when j <9 < n.
To summarize
proJWV = u^)w
the scalar projection of v onto w is the number:
v• w
COmp”V= Ml
It is important to note that the i, j, and k components of any vector v are the scalar
projections of v onto the appropriate basis vectors, as shown in Figure 9.30.
Solution
The vector projection of v onto w is
2(2) + (3) (—2) + 5(— 1)
(2i - 2j - k)
proj»v = (^)w = ( 22 H- (—2)2 H- (-1)2
= -|(2i - 2j - k) = —yi + + ?k
In Section 6.5, we noted that when a constant force F acts in the direction of motion of
an object moving from P to Q along a straight line, it produces work, W, given by
More generally, if the force F remains constant but points in a direction that makes
an angle 9 with the direction of motion (see Figure 9.31b), then the work W done
by F in moving the object along the line from P to Q is given by the product of the
component of F on the displacement vector PQ times the distance moved; that is,
W = F PQ
Note that this formula also gives the work in the case where 0 = 0.
Solution
The wind force is ||F|| = 500 lb, acting in a direction 0 — 30°, as shown in Figure 9.32.
The displacement vector is PQ = lOOj, so ||PQ|| = 100 ft. Thus,
F = 500 cos 60° i + 500 sin 60° j = 250i + 250\/3 j. The work performed is
Side View
W = F PQ = 100(250\/3) = 25,000^3
Figure 9.32 Work performed on a sail Therefore, the work is approximately 43,300 ft-lb.
9.3 The Dot Product 595
43. A block of ice is dragged 20 ft across a floor, using a force and discontinuous, can be represented by trigonometric series
of 50 lb, as shown in Figure 9.35. Find the work done if the of the general form
direction of the force is inclined 6 to the horizontal, where
i<30 + a\ cos x + a2 cos 2x + a2 cos 3x 4- • ■ •
a. 9 = j b. 6 = f
+ b\ sinx + b2 sin 2x + b2 sin 3x 4- • • •
f
Ja
f(x)g(x)dx = 0
b. Use the Fourier series in part a to evaluate the convergent
alternating series
a. Show that the two functions x2 and x3 — 5x are orthogonal (-1)*+1 1111
on [—b, b] for any positive b. 2k - 1 3 5 7 9
b. For distinct positive integers k and n, show that sinkx and
Q 47. Let u0 = ai + bj + ck and let u = xi + yj + zk. Describe the
sinnx are orthogonal on the interval [—tv, tv]. That is, the
set of points in R3 defined by
family of functions sinx, sin 2x, • ■ • are mutually orthogo¬
nal on [—tv, tv]. You may need the product-to-sum identity lluo - u|| < r
(trigonometric identity 28 from the Student Mathematics
Handbook): where r > 0 and a, b, c are constants.
48. Find the angle (to the nearest degree) between the diagonal of
2 sin a sin f$ = cos(o? — f) — cos(a + f) a cube and a diagonal of one of its faces. Assume both diago¬
nals have the same initial point.
46. HISTORICAL QUEST Jean Bap¬
49. Show that the vector B = ||v||u 4- ||u||v bisects the angle 6
tiste Joseph Fourier began his stud¬
between the two nonzero vectors u and v.
ies by training for the priesthood. In
50. Use vector methods to prove that an angle inscribed in a semi¬
spite of this first calling, his interest
circle must be a right angle.
in mathematics remained intense, and
in 1789 he wrote, "Yesterday was 51. a. Show that (v 4- w) • (v 4- w) = ||v||2 + ||w||2 + 2(v • w).
my 21st birthday, and at that age b. Use part a to prove the triangle inequality:
Newton and Pascal had already ac¬
JEAN FOURIER I|V +w|| < ||v|| 4- ||w||
quired many claims to immortality."
1768-1830
He did not take his religious vows Hint: Note that
but continued his mathematical research while teaching.
Fourier was renowned as an outstanding teacher, but it was ||v 4- w||2 = (v 4- w) • (v 4 w)
not until around 1804-1807 that he did his important mathe¬
52. The Cauchy-Schwarz inequality in R3 states that for any vec¬
matical work on the theory of heat. ■
tors v and w
During that time, he expanded on an idea of Daniel
Bernoulli by showing how many functions, both continuous |V • w| < IIv|| ||w||
9.4 The Cross Product 597
The dot product is not the only way to multiply vectors, In this section, we will study
a second product involving vectors, called the cross product, vector product, or some¬
times the outer product. Unlike the dot product, in which two vectors are multiplied to
produce a scalar, the cross product produces a vector. Here is the definition.
Cross Product If v = aii + ci2j +«3k and w = bx\ + b2] + b3k, the cross product of the vector
v with the vector w, written v x w, is the vector
The form of the cross product may seem odd to you, but you will soon discover
^ -9 Determinants are reviewed in that the cross product has a variety of useful properties.
Section 2.8 of the Student A convenient way of remembering the cross product is the following determinant
Mathematics Handbook.
form:
i j k
V X w = ax a2 a3
bi b2 b3
To verify the determinant formula for the cross product, we expand about the first
row.
k
i j
a2 ai fli 03 ai a2
V X w = a1 ai a3 —
b2
i — j +
b\
b3 b\ b3 b2
b\ b2 b3
Solution
k
V X w 3
-4
See Example 2.31 and Problem 66 of Problem Set 2 of the Student Mathematics
Handbook.
TECHNOLOGY NOTE
i j k i j k
V X w = a\ a2 = — b\ b2 b2 = — (w X v)
b\ b2 bj a\ a2 «3
Zero product v x 0 = 0 x v = fl
Proof Let u = ci\i + 02} + 03k, v = b\\ + bij + b^k, and w = c\\ + C2J + 03k.
Scalar distributivity and vector distributivity are proved by using the definition of cross
product and the corresponding properties of real numbers.
Anticommutativity was proved in the paragraph preceding this theorem.
Product of a multiple
i j k i j k
v X s\ = b\ b2 h = s b\ b2 63 = 0 A determinant with two
sb 1 sb2 sb?, b\ bi bi identical rows is 0.
Lagrange's identity
— ||v||2||w||2 — (v ■ w)2
You are accustomed to dealing with products in which the order does not matter,
but the order of the terms in the cross product V X w matters very much, since
w x v = —(v x w).
Cross products involving the basic vectors i, j, and k occur so frequently in com¬
putations that it is useful to have the following list:
i x i = 0 » xj = k i x k = —j
j x i = -k j xj = 0 j x k = i
k x i =j k x j = —i k x k = 0
These nine rules can be established by using the definition of cross product, or by using
the determinant form. For example,
i j k
j x i = 0 1 0 -k
1 0 0
The cross product is frequently used for geometric purposes, such as determining a
vector that is orthogonal to two given vectors. The following theorem lists one of
the most important geometric properties of the cross product. It says that for any
Figure 9.37 vxwis orthogonal to two nonzero vectors v and w, the cross product is a vector orthogonal to the plane
both v and w determined by the vectors v and w, as shown in Figure 9.37.
Proof The theorem is illustrated in Figure 9.37. We will show that (v x w) is or¬
thogonal to v and leave the proof that vxwis orthogonal to w as an exercise. Let
v = fiqi + «2j + 03k and w = b\\ + bfi + ^k. Then
J k
V X w = a1 ai «3 — («2^3 — «3^2)i ~ i^3 — aib\)i -L {a\b2 - 02^1 )k
bi b2 h
600 Chapter 9 Vectors in the Plane and in Space
so that
Solution
1 J k
V X w -2 3 -7 = (27 + 0)i - (-18 + 35)j + (0 - 15)k
5 0 9
In Section 9.3, we showed that the dot product satisfies v w = ||v||||w|| cos0,
where 0 is the angle between v and w. The following theorem establishes a similar
result for the cross product.
|| v X w||2 = || v||21| w||2 - [|| v|| || w|| cos 0]2 Lagrange's identity
— IMI2||w||2(1 - cos2 0)
= ||v||2||w||2 sin2 6
Thus,
From Theorem 9.7, we know the magnitude of the cross product v x w, but what
is its direction? It turns out that this direction can be described by the following rule:
9.4 The Cross Product 601
Figure 9.38 The right-hand rule: If you place the little finger
of your right hand along v and curl your fingers toward w
to cover the smaller angle between v and w, then your thumb
points in the direction of v x w.
The cross product can be used in a variety of ways in geometry and physics. For
example, using Theorem 9.7, it can be shown that the parallelogram having v and w as
adjacent sides has area ||v x w|| (see Figure 9.39). To see this, note that the height of
the parallelogram is h = ||w|| sin 9, so that the area is given by
From this formula, it follows immediately that the area of a triangle in R1 is half the
v
area of the corresponding parallelogram. For future reference, we let v be defined by
the points A and B, and w by the points A and C, and summarize these formulas for
Figure 9.39 A geometric
area in the following box.
interpretation of ||v x w||
AREA = IIAB X AC II
Let A, B, and C be points in K3. Then the triangle with vertices A, B, and C has
AREA = ^ II AB X AC II
602 Chapter 9 Vectors in the Plane and in Space
Solution
Draw this triangle as shown in Figure 9.40. Then APQR has half the area of the
parallelogram determined by the vectors PQ and PR; that is, the triangle has area
A = ^ IIPQ x PR||
First find
PQ = (0 + 2)i + (7 - 4)j + (-4 - 5)k = 2i + 3j - 9k
PR = (-1 + 2)i + (5 - 4)j + (0 - 5)k = i +j - 5k
and compute the cross product:
1 j k
PQ x PR = 2 3 -9
* Figure 9.40 Area of a triangle
by cross product
1 1 -5
Solution
We represent the force by F = —30i (see Figure 9.42). Because the door is half open,
it makes an angle of j with the positive x-axis, and we can represent the “arm” PQ by
the vector
T = PQ x F = 3 72 3y/2
0 = 45\/2k
2
-30 0 0
The magnitude of the torque (45 v/2 lb-ft) is a measure of the tendency of the door to
rotate about its hinges. ■
9.4 The Cross Product 603
The cross product can also be used to compute the volume of a parallelepiped in R3.
Consider the parallelepiped determined by three nonzero vectors u. v, and w that do
not all lie in the same plane, as shown in Figure 9.43.
U X V
It is known from solid geometry that this parallelepiped has volume V = Ah,
where A is the area of the face determined by u and v and h is the altitude from the tip
of w to this face (see Figure 9.43b).
The face determined by u and v is a parallelogram with area A = ||u x v||, and we
know that the cross-product vector u x v is perpendicular to both u and v and hence to
the face determined by u and v. From the geometric formula for the dot product, we
have
where 0 is the angle between u x v and w. Since the parallelepiped has altitude
(u x v) • w
h = 11| w || cos 0| —
U X V
(u X v) • w
V = Ah = || u x v |(u X v) • w
llu X v||
V = |(u x v) • w|
u = i - 2j + 3k v = —4i + 7j - Ilk w = 5i + 9j - k
604 Chapter 9 Vectors in the Plane and in Space
Solution
We first find the cross product
1 J k
11 X V — 1 -2 3
-4 7 -11
= (22 —21)i—(—11 + 12)j + (7 — 8)k
= i-j -k
Thus,
a1 a2 a2 components of u
(u X v)•w = b\ b2 b3 components of v
c1 ci c3 components of w
Proof The proof follows by expanding the determinant (see Problem 49). □
We can use Theorem 9.8 to rework Example 5:
1 -2 3
(u x v) • w = -4 7 -11 -3
5 9-1
In the problem set, we have included several exercises involving the scalar triple
product and the triple vector product u X V X W. In case you wonder why we neglect the
product (u • v) x w, notice that such a product makes no sense, because u • v is a scalar and
the cross product is an operation involving only vectors. Thus, the product u • v x w must mean
u • (v x w).
5. v = 3i — 2j + 4k; w = i + 4j - 7k 15. v = 2i + k; w = i - j - k
16. v = j - 3k; w = -i + j + k
6. v = 5i — j + 2k: w = 2i + j — 3k
17. v = i + j + k; w = 3i + 12j — 4k
7. v = 3i — j + 2k; w = 2i + 3j — 4k
18. v = 2i - 2j + k; w = 4i + 2j - 3k
8. v = — j + 4k; w = 5i + 6k
Find the area of the parallelogram determined by the vectors in
9. v = i — 6j + 1 Ok; w = —i + 5j — 6k Problems 19-22.
10. v = cos 61 + sin 9y, w = — sin 64 + cos 0j for any angle 0
19. v = 3i + 4j and w = i + j — k
Use the cross product to find sin 9 where 9 is the angle between v 20. v = 2i — j + 2k and w = 4i — 3j
and w in Problems 11-14. 21. v = 4i — j + k and w = 2i + 3j — k
11. v = i + k; w = i + j 12. v = i + j; w = i + j + k 22. v = 2i + 3k and w = 2j — 3k
9.4 The Cross Product 605
53. Show that Establish the validity of the equations in Problems 56-59 for ar¬
bitrary vectors u, v, w, and z in R3. You may use the result of
v• w Problem 55.
where 9 (0 < 6 < |) is the angle between v and w. 56. (u x v) x (w x z) = (u • w x z)v — (v • w x z)u
54. Let u, v, and w be nonzero vectors in R3 that do not all lie in
57. u x (v x w) + v x (w x u) + w x (u x v) = 0
the same plane. Show that
58. u x v = (u • v x i)i + (u ■ v x j)j + (u • v x k)k
|u • (v x w)| = |v ■ (u X w)|
What other scalar triple products involving u, v, and w have 59. u x [u x (u x v)] • w = — ||u||2u • v x w
the same absolute values? 60. Show that if vectors OP, OQ, OR, and OS lie in the same
55. Let a, b, and c be vectors in space. Prove the “cab - bac” plane, then
formula.
a x (b x c) = (c ■ a)b — (b ■ a)c (OP x OQ) x (OR x OS) = 0
PARAMETRIC EQUATIONS
Until now, most of the curves we have considered have been graphs of functions of
the form y = f(x). This form of representation is limited by the requirements of the
vertical line test, which precludes closed curves such as circles or any curve with a
self-intersection. However, such restrictions are often not necessary for curves where
the coordinates x and y of each point P(x, y) on the curve are themselves functions of
a third variable, called a parameter. We will begin by examining parametric represen¬
tations in R2, and then extend our results to R3.
Parametric Representation Let / and g be continuous functions of t on an interval I; then the equations
for a Curve in R2
x = /(f) and y = g{t)
are called parametric equations with parameter t. As t varies over the para¬
metric set /, the points (x, y) — (f{t), g(t)) trace out a parametric curve.
Note: The letter “f” used for the parameter does not necessarily denote time, al¬
though in many applications, time is a suitable parameter. Indeed, any letter or symbol
may be used to denote a parameter.
Solution
Values of x and y corresponding to various choices of the parameter t are shown in the
following table:
9.5 Parametric Representation of Curves; Lines in K3 607
t X y
-3 0 -l (Starting or initial point)
2
-2 -5 ~3
-1 -8 3
0 -9 0
1
1 -8 3
2
2 -5 3
(Ending or terminal point)
x = (3y)2 - 9 = 9y2 - 9
which is the Cartesian equation for a parabola that opens to the right. Because of the
domain of the parameter t, we see that the parametric curve in Figure 9.47 is a subset
of the set of points that satisfy the equation x = 9y2 — 9.
Parameterizations are not unique. For example, the curve with parametric equa¬
tions
Solution
Using a double angle identity, we find
so that
y = 1 — 2jc2
We recognize this as a Cartesian equation for a parabola. Because y' = —Ax, we can
find the critical number x = 0, which locates the vertex of the parabola at (0, 1). The
parabola is the curve shown in color as the dashed curve in Figure 9.48.
Because t is restricted to the interval 0 < t < 0.5, the parametric representation
Figure 9.48 The parabolic arc involves only part of the right side of the parabola y = 1 — 2.t2. The curve is oriented
x = sin nt, y = cos 2nt from the point (0, 1), where t = 0, to the point (1, —1), where t = 0.5, and is the
forO < t < 0.5 portion of the parabola shown in black in Figure 9.48.
Solution
We have no convenient way of eliminating the parameter so we write out a table of
values (x, y) that correspond to various values of t. The curve is obtained by plotting
these points in a Cartesian plane and passing a smooth curve through the plotted points,
as shown in Figure 9.49.
t X y
0 1 0
n
4
0.32 0.32
l 0.20 0.31
7T
2
0 0.21
2 -0.06 0.12
71 -0.04 0
3n
2
0 - 0.01
2tt 0.00 0 Figure 9.49 Graph of x = e 'cos t,
y = e~‘ sin t for t > 0
Note that for each value of t, the distance from P(x, y) on the curve to the origin is
PARAMETERIZING A CURVE
So far, our examples have dealt with sketching a parametric curve given the parametric
equations. In general, this process may be tedious, but generally can be done easily
using technology. However, the reverse process, finding a suitable set of parametric
equations for a given curve, is an art for which there is no simple procedure. Indeed, a
given curve can have many different parameterizations and there are curves for which
no simple parameterization can be given. The following two examples illustrate vari¬
ous methods for parameterizing a given curve.
Solution
a. The usual parameterization for a parabola is to let the parameter t be the variable
that is squared: x = t, y — 912. However, another parameterization is to let t = 3x
so that x = jt and y = t2.
b. In polar coordinates we have x — r cos#, y = r sin#, so we can parameterize x
and y in terms of the parameter #:
x — r cos # y = r sin 6
— 5 cos4 0
9.5 Parametric Representation of Curves; Lines in R3 609
Solution
Assume that the wheel rolls along the x-axis and that the center C of the wheel begins
at (0, a) on the y-axis. Further assume that P also starts on the y-axis, d units below
C. Figure 9.51 shows the initial position of the wheel and its position after turning
through an angle 9 (radians).
We begin by labeling some points: The point A is on the x-axis directly beneath
C, whereas B is the point where the horizontal line through C meets the rim of the
wheel. Finally, Q is the point on 5C directly beneath P, and S is the point where the
line through C and P intersects the rim. Let P have coordinates (x, y).
We need to find representations (in terms of a, d, and 9) for x and y.
x = | OA | T |CQ|
= a9 + | C Q | Because the wheel rolls along the x-axis
without slipping, | O A | is the same as
the arc length from A to S, so \ 0A\ — ciB.
y = \ac\ + [QP\
= a + \QP\
Chapter 9 Vectors in the Plane and in Space
To complete our evaluation of x and y, we need to compute \CQ\ and |(9P|. These
are sides of A PC Q. Note that LPCQ — f — 0; therefore, by the definitions of cosine
and sine, we have
x = aO + | C Q | = a0 — d sin 6
y = a + I QP I = a — d cos 0 ■
The special case where P is on the rim of the wheel in Example 5 (when d = a) is
a curve called a cycloid. There are several problems involving these and similar curves
in the problem set.
If /, g, and h are continuous functions of a variable t on an interval /, then
are parametric equations with parameter t, and as t varies over /, the points
(jc, y, z) = {fit), g(t), h{t)) trace out a parametric curve in R3. The rest of this sec¬
tion is devoted to examining lines in R3, which we will represent in parametric form.
More general curves in R3 will be studied in Chapter 10.
LINES IN R3
As in the plane, a line in R3 is completely determined once we know one of its points
and its direction. We used the concept of slope to measure the direction of a line in the
plane, but in space, it is more convenient to specify direction with vectors.
Suppose Lisa line in R3 that is parallel to the vector v = Ai + Bj + Ck and also
contains Q{xo, yo, zo), as shown in Figure 9.52. We say that the line has direction
numbers A, B, and C and denote these direction numbers by [A, B, C]. The vector
v is called a direction vector of the line L. If P{x, y, z) is any point on L, then the
vector QP is parallel to v and must satisfy the vector equation QP = t\ for some
Figure 9.52 If L is parallel to
number t. If we introduce coordinates and use the standard representation, we can
v and contains Q, then P
rewrite this vector equation as
is on L whenever QP = tv.
(x - *0)i + (y - y0)j + iz- Zo)k = t(Ai + Bj + Ck)
By equating components on both sides of this equation, we find that the coordinates of
P must satisfy the linear system
X — Xo = tA y-y0 = tB Z-Zo — tC
where t is a real number. Note that each point on L corresponds to exactly one value
of the parameter t.
X=Xo + tA y = y0 + tB z = Zo + tC
Conversely, the set of all points (jc, y, z) that satisfy such a set of equations is
a line that passes through the point (*o, yo, Zo) and is parallel to a vector with
direction numbers [A, B, CJ.
9.5 Parametric Representation of Curves; Lines in M3 611
Solution
The direction numbers are [-1, 1, -2] and x0 = 3, y0 = 1, z0 = 4, so the line has the
parametric form
x = 3 - t y = 1 +/ z = 4-2/
0 = 4 — 2/ implies t — 2
Solution
The point of intersection of the line with the xy-plane is (y, y,0). Similarly, the
other points of intersections are
has direction numbers [2, 0,-1] and lies in the plane y — 5, parallel to the xz-plane.
The symmetric form for this line is
x - 3 _ z - A
Solution
L\ has direction numbers [2, 1,4] and Lj has direction numbers [4, —3, —1]. Because
there is no t for which [2, 1,4] = /[4, -3, -1], the lines are not parallel. Express the
lines in parametric form:
L i : x = 1 4- 2s y — —1 +£ z = 2 + As
Li : x — — 2 + At y = —31 z = t
s + 3t = 1
+
or
II
1
3
2 + 4 s = \-t or As + t = 2
9.5 Parametric Representation of Curves; Lines in R3 613
4(-i) +
If the solution did not satisfy the third equation, the lines would be skew. To find the
coordinates of the point of intersection, substitute s = —\ into the parametric-form
WARNING K1 . , , equations for L\ (or substitute t — \ into Li) to obtain
Notice that we used two
parameters (s and t) for the lines in
Example 8. This is because it is quite
*0= l+2(—±) = 0
x — 1 y+1 Z 2 , t x +2 y z+ 1
and L2
1
Solution
Note that L\ has direction numbers [2, 1,4]; that is, L\ is parallel to (2i + j + 4k) and
L2 has direction numbers [4,-3, 1], If we try to solve
for t, we find no possible solution for any value of t. This implies that the lines are not
parallel.
Next we determine whether the lines intersect or are skew. Note that 5(1, —1,2)
lies on L\ and T(—2, 0, —1) lies on L2. The lines intersect if and only if there is a
point P that lies on both lines. To determine this, we write the equations of the lines in
parametric form:
L1 : x — 1 + 2s y — ~ 1+J z = 2 + 4s
L,2 '• x = —2 + 41 y — —31 z = —1 +t
* = 1 + 2s = — 2 + At or
>>
II
1
1
u>
+
or 5 + 3? —
V
1
II
II
1
1
CO
z = 2 + 4x = -1 +t or
II
1
'is -At = -3
• s + 3t = 1
4.v — t = —3
Any solution of this system must correspond to a point of intersection of L\ and Zo,
and if no solution exists, then L\ and L2 are skew. Because this is a system of three
equations with two unknowns, we first solve the first two equations simultaneously to
find s = -5, t = 2. However, (5, f) = (—5. 5) does not satisfy the third equation, so
it follows that L\ and L2 do not intersect and they must be skew.
614 Chapter 9 Vectors in the Plane and in Space
Q 52. What can be said about the lines Show that this epicycloid may be modeled by the parametric
equations
x -x0 _ y - y0 _ z- Z o
ai bx ci a+ R
x = (a + R) cos t — R cos t,
R
and
a+ R
x - xp _ y - yo _ z - z0 y = (a + R) sin t — R sin t
R
CI2 bj C2
-•
Planes in R?
IN THIS SECTION forms for the equation of a plane in R3, vector methods for measuring distances in R3
Planes in space can also be characterized by vector methods. In particular, any plane
is completely determined once we know one of its points and its orientation; that is,
the “direction” it faces. A common way to specify the direction of a plane is by means
of a vector N that is orthogonal to every vector in the plane, as shown in Figure 9.56.
Such a vector is called a normal to the plane.
Solution
The normal vector N is orthogonal to every vector in the plane. In particular, if
P(x, y, z) is any point in the plane, then N must be orthogonal to the vector
Because the dot (or scalar) product of two orthogonal vectors is 0, we have
a. 5x + 7y — 3z = 0 b. 3x — Iz = 10
Solution
a. A normal to the plane 5x + 7y — 3z = 0 is N = 5i + 7j — 3k.
b. For 3x — lz = 10 a normal is N = 3i — 7k. ■
Forms for the Equation of a An equation for the plane with normal N = Ai -f Bj + Ck that contains the point
Plane (xo, To> Zo) has the following forms:
N = Ai + B) + Ck
nN
Find the standard form equation of a plane containing P(—1, 2, 1), <2(0, —3, 2), and
/?(!, 1, -4).
26i + 7j + 9k
Solution
Because a normal N to the required plane is orthogonal to the vectors PQ and PR, we
find N by computing the cross product N = PR x PQ. The situation is illustrated in
Figure 9.58.
PQ (0 + l)i + (-3 - 2)j + (2 - l)k = i - 5j + k
Figure 9.58 A plane passing through
three given points PR (1 + l)i + (1 - 2)j + (-4 - l)k = 2i - j - 5k
i j k
N PQ x PR = 1 -5 1
2 -1 -5
= 26i + 7j + 9k
9.6 Planes in R3 617
We can now find the equation of the plane using this normal vector and any point in
the plane. We will use the point P:
26x + 26 + 7y — 14 + 9z — 9 = 0
26x + 7y + 9z + 3 = 0 ■
Solution
Figure 9.59 Intersection of a line Then, the point of intersection is found by substituting t — — 1 for x, y, and z:
and a plane
x = 2 + 3(-l) = -1
y = -1 — 7(—1) = 6
z = 3 + 5(—1) = -2
Solution
By inspection, we see that normals to the given planes are Ni = 3i — 2j + k and
N2 = i + 2j + 3k. The desired line is perpendicular to both of these normals, so a
vector parallel to the line is found by computing the cross product:
i j k
Ni x N2 = 3 -2 1 (—6 — 2)i — (9 — l)j + (6 + 2)k
1 2 3
= —8i - 8j + 8k
= —8(i + j — k)
Thus, the required line passes through (—1,2, 3) and is parallel to the vector (1, 1,-1)
so it has parametric form
x = — 1 +1, y = 2 + t, z = 3 - t
618 Chapter 9 Vectors in the Plane and in Space
An equation for the line of intersection of two planes can be found by determining
a point P in the intersection, and then proceeding as in Example 5 (using P instead of
the point (—1,2, 3)). In Example 6, we turn things around by finding the equation of a
plane containing two given intersecting lines.
= -6i - 7j + k
The points (2, —5, —1) and (—1,0, 16) are both in the required plane. We use
(2, —5, —1) to obtain
—6(x — 2) — 7(y + 5) + l(z + 1) = 0
6x 4- 7y — z + 22 = 0
as the equation for the plane.
Vector methods can also be used for measuring distances in R3, between points, lines,
and planes. We begin by deriving the following formula for the distance from a point
to a plane.
Plane Ax + By + Cz + D = 0 Proof The required distance is found by projecting the vector QP onto the normal
N. Thus, the distance from the point to the plane is given by
J linnil, ||QP||||N|||cos0| |QP-N|
d = QP cos# =-=-
l|N|| IIN ||
where 0 is the angle between QP and N (see Figure 9.60). Suppose Q(x\, y\,z\) is
any particular point in the plane, so that
Q
QP = Uo - *i)i + (y0 - yi)j + (z0 - zt)k
The dot product of QP with N = Ai + B\ + Ck is given by
Figure 9.60 The distance from a point QP • N = (*0 - x\)A + (y0 - y\)B + (z0 - Zi)C
to a plane in K3
= (A*o + Byo + Czo) - (Axi + By\ + CzQ
= Axo + Byo + Czo — (~D) Since Ax\ + By\ + Cz\ + D = 0
Because the normal vector has length ||N|| = \JA2 + B2 + C2, we can substitute into
the formula for d to obtain
j _ IQP • N| |Axo + Byo + Czo + D\
^
IIN || _ \/ A2 + B2 + C*~
9.6 Planes in M3 619
Solution
Recall from Section 9.5 that skew lines do not intersect and are not parallel, but do lie
in parallel planes. Thus, the distance d between the lines L\ and L2 is the same as
the distance between the parallel planes p\ and P2 that contain them (see Figure 9.62).
Our strategy for finding the distance d is to find the distance from a point Q on p\ to
the plane P2.
The distance d between the lines L\ and Lj is the same as the distance from the point
Q( 1, — 1,2) to the plane pi. Thus
In the final example of this section, we find the distance from a point P to a line
L in R3. Let Q be a point on L and let v be a vector parallel to L. Then, as shown in
Figure 9.63, the distance from P to L is given by
d = ||QP|| |sin#|
where # is the angle between v and the vector QP. This reminds us of the cross product
vxQP, and because
Figure 9.63 The distance from
a point to a line in R3
|v x QP|| = ||v||||QP|| sin#
we have
v x QP
d = ||QP|| sin#
IMI
These observations are summarized in the following theorem.
||v x QP||
d =-
|| v||
x — 3 y+7 z+2
3 ~ -1 ~ 5
Solution
We need to find a point Q on the line. We see that <2(3, —7, —2) is on the line and that
QP = (0, —1,3). A vector parallel to L is v = (3,-1, 5), so that
j k
v x QP = 3 -1 5 = 2i — 9j — 3k
0 -1 3
Thus,
42. Find an equation for the line that passes through the point
P(0, 1,-1) and is parallel to the line of intersection of the
31. P(l,-2.2); J = y =
planes 2x 4- y — 2z — 5 and 3x — 6y — 2z — 1 ■
622 Chapter 9 Vectors in the Plane and in Space
43. Find a vector that is parallel to the line of intersection of the O 56. The planes Ax + By + Cz = D, and Ax + By + Cz — D2
planes 2x + 3>y — 0 and 3jc — y + z — 1. are parallel.
44. Find an equation for the line of intersection of the planes a. Is the distance between the planes |Z)| — D2|? If not, what
2x - y + z — 8 and x + y — z — 5. is the correct formula?
45. Find the direction cosines of a vector determined by the line of
b. Find the distance between the parallel planes x+y+2z = 2
intersection of the planes x + y + z — 3 and 2x + 3y — z = 4. and x + y + 2z = 4.
46. Find an equation for the plane that passes through P(1, — 1, 2)
57. Show that the angle between the planes
and is normal to PQ where Q is Q(2, 1,3).
47. Find an equation for the line that passes through the point
A\X + B\y + C\Z + D\ = 0
(1, —5, 3) and is orthogonal to the plane 2x — 3y + z — 1.
48. Find two unit vectors that are parallel to the line of intersection and
of the planes x + y = 1 and x — 2z — 3.
49. Find two unit vectors that are parallel to the line of intersection A2x + B2y + C2z + D2 — 0
•* of the planes x + y + z = 3 and x — y + z — 1.
50. Find an equation for the plane that contains the point is 7r/2 if and only if
(2, 1, —1) and is orthogonal to the line
A] A2 + B\B2 + C\C2 = 0
x -3 _ y + 1 z
3 “ 5 ~ 2
58. Show that a plane with x-intercept a, y-intercept b, and
51. Find a plane that passes through the point (1,2, —1) and is z-intercept c has the equation
parallel to the plane 2x — y + 3z = 1.
52. Show that the line
-
a + T
b + —
c 1
x - 1 _ y+ 1 _ Z-2
2 ~ 3 ~ 4 assuming a, b, and c are all nonzero.
is parallel to the plane x — 2y + z = 6. 59. Suppose planes px and p2 intersect. If Vi and Wj are nonzero
53. Find the point where the line vectors on p{ and v2 and w2 are nonzero vectors on plane p2,
then show that
x-1 _ y+1 z
2 ~ -1 “ 3 (vt x Wi) x (v2 x w2)
intersects the plane 3x + 2y — z = 5.
54. The angle between two planes is defined to be the acute an¬
is parallel to the line of intersection of the planes.
gle between their normal vectors. Find the angle between the 60. Let L| and L2 be skew lines that contain the points P\ and P2,
planes 2x + y — 4z = 3 and x — y + z — 2, rounded to the respectively, and are parallel to the vectors Vi and v2. Show
nearest degree. that the distance between the lines is given by
55. Find an equation for the line that passes through the point
P(2, 3, 1) and is parallel to the line of intersection of the (Vi X V2) • PiP2
d =
planes x + 2y - 3z = 4 and x - 2y + z = 0. IIVi X v2||
Quadric Surfaces
methods for graphing quadric surfaces; a catalog of quadric surfaces:
You might want to review graphs whose equations are of the form
Ax2 + Bxy + Cy2 + Dx + Ey + F = 0
which are called conic sections (see Chapters 6 and 7 of the Student Mathematics
SMH ) Handbook). Conic sections are graphs in M2 that can be represented by a second-
degree equation in two variables, and quadric surfaces are graphs in JR3 which can be
represented by a second-degree equation in three variables.
Spheres, and elliptic, parabolic, and hyperbolic cylinders are examples of quadric
surfaces. In general, such a surface is the graph of an equation of the form
x2 k2 z2 , x2 z2 , k2
(2,0.0)
■4 + y + ~g - 1 or-1-= 1 - —
4 9 5
which is an ellipse for k2 < 5, as is the trace in the plane x — k (for k2 < 4):
k2 y2 z2 . y2 z2 . k2
Figure 9.65 Graph of the ellipsoid 7 + 7+9=1 or 7+9='-7
2 2 2
— + — + —• = i Sketch these traces in R3 as shown in Figure 9.65. Since the traces in all three principal
4 5 9 directions are ellipses, we call this surface an ellipsoid. ■
Solution
We begin by setting z = k, to obtain
9x2 + 36y2 = 4k2 - 36
and for y = k
4Z2 - 9x2 = 36 + 36k2
z To summarize, the surface has hyperbolic traces for planes parallel to the yz-plane
and the xz-plane and elliptical traces for planes parallel to the xy-plane, except for a
“gap” corresponding to the planes of the form z = k with |&| < 3. This surface is
called a hyperboloid of two sheets. The graph is shown in Figure 9.66.
Solution
For z = 0, we have
9 16
and the corresponding trace is a hyperbola (along the y-axis if k > 0, and along the
x-axis if k < 0). These traces are also shown in Figure 9.67a.
If x = k, we have
z= r e
9 16
and the trace is a parabola that opens up (Figure 9.67b), while if y = k, the corre¬
sponding equation
z =
9 16
A catalog of quadric surfaces is shown in Table 9.1, where x, y, and z are variables
and a, b, and c are constants.
Hyperboloid of one sheet The trace in the Hyperboloid of two sheets There is no trace in the
xy-plane is an ellipse; xy-plane. In planes
in the xz- and parallel to the xy-plane
yz-planes, the traces that intersect the
are hyperbolas: surface, the traces are
ellipses. Traces in the
xz- and yz-planes are
a2 b2 c2 hyperbolas:
Ellipsoid The traces in the Hyperbolic paraboloid, also called a The trace in the
coordinate planes are saddle surface xy-plane is two
ellipses: Z, intersecting lines; in
planes parallel to the
xy-plane, the traces are
+ = 1
hyperbolas. Traces in
the xz- and yz-planes
are parabolas:
y2 x2
Z~b2~a2
Solution
a. We rewrite the equation in one of the two standard forms:
z2 = t
4 + t
7
We see this is an elliptic cone,
b. Rewrite as
x2 yL2
Z=J+t
We see that the surface is an elliptic paraboloid.
3. x2 = z2 + y2
7. x2 + y2 + z2 = 9
2 2
y_ r_
9. x
25 + 16
9.7 Quadric Surfaces 627
2 2 2
xL y z_
a2 b2 c2
aemm»
Proficiency Examination 31. What is the formula for the distance from a point to a line?
32. What is a quadric surface?
CONCEPT PROBLEMS
1. What is a vector and what is a scalar? PRACTICE PROBLEMS
2. Give both an algebraic and a geometric interpretation of mul¬ 33. Given v = 2i —3j + k, w = 3i —2j. Find each of the following
tiplication of a vector by a scalar. vectors or scalars.
3. What is the parallelogram rule for vector sums? a. 2v + 3w b. 11v||2 — ||w||2
4. State each of the following properties of vector operations: c. v • w d. v x w
a. commutativity of vector addition e. vector projection of v onto w
b. associativity of vector addition
f. scalar projection of w onto v
, c. identity for vector addition
34. Given u = 2i - 3j + k, v = i + j - 2k, and w = 3i + 5k.
d. inverse property for vector addition In each of the following cases, either perform the indicated
e. magnitude of a vector for dot product computation or explain why it is not defined.
f. commutativity for dot product a. (u x v) ■ w b. (u • v) x w
g. multiple of a dot product c. (u x v) x w d. (u v) w
h. distributivity for dot product over addition Find the equations for the lines and planes in Problems 35-38.
i. cross product of a zero vector 35. the line through the points P{— 1, 4, —3) and Q(0, —2, 1)
j. anticommutativity for cross product 36. the plane that contains the point P(l, 1,3) and is normal to
k. distributivity for cross product over addition the vector v = 2i + 3k
5. If u and v are parallel vectors, what can be said about u x v? 37. the line of intersection of the planes 2x + 3y + z = 2 and
6. How do you find the length of a vector? What is a unit vector? y — 3z — 5
7. State the triangle inequality. 38. the plane that contains the points P(0, 2, —1), <2(1, —3, 5),
8. What are the standard basis vectors in R3? and R(3, 0, —2)
9. What is the standard-form equation of a sphere? 39. Find the direction cosines and the direction angles of the vec¬
10. What is a cylinder? tor u = —2i + 3j + k. Round to the nearest degree.
11. What is the distance between two points in R3? 40. In each case, determine whether the lines intersect, are paral¬
lel, or are skew. If they intersect, find the point of intersection.
12. How do you find a unit vector u in the direction of a given
vector v? a. x = 2t — 3, y = 4 — t, z = 2t and ~ ^; z = 3
13. What is Lagrange’s identity?
x—1 y-6 z-8 x -8 y-6 z-9
14. Define dot product. b. —-— = 1-=-and-=-=-
5 4 5 6 4 6
15. What is the formula for the angle between two vectors? 41. Let u = 2i + j, v = i - j — k, and w = 3i + k.
16. What is meant by orthogonal vectors? What is the algebraic a. Find the volume of the parallelepiped determined by these
condition for orthogonality?
vectors.
17. What is a vector projection? Give a formula for finding the
b. Find a positive number A that guarantees that the tetrahe¬
vector projection of v onto w.
dron determined by Au, A\, and w has volume that is twice
18. What is a scalar projection? Give a formula for finding the the volume of the original tetrahedron. Hint: See Problem
scalar projection of v onto w. 50, Section 9.4.
19. What is the vector formula for work? 42. Find the distance from P(—1, 1,4) to 2x + 5y — z — 3.
20. Define cross product. 43. Find the distance between the skew lines x = t, y = 2t,
21. What is the right-hand rule for a coordinate system? Z = 3f — 1 and x = 1 — t, y = t + 2, z = t.
22. a. Give a geometric interpretation of cross product. 44. Find the distance from the point P(4, 5, 0) to the line
b. What is the formula for the magnitude of a cross product?
x -2 _ y _ z+ l
23. What is the determinant form for the scalar triple product?
3 “ 5 ~ -1
24. How do you find the volume of a parallelepiped?
25. a. What is the parametric form of a line in R3? 45. An airplane flies at 200 mi/h parallel to the ground at an alti¬
b. What is the symmetric form of a line in R3? tude of 10,000 ft. If the plane flies due south and the wind is
26. What are the direction cosines of a vector in R3? blowing toward the northeast at 50 mi/h, what is the ground
speed of the plane (that is, effective speed)?
27. What is the normal vector of a plane in R3?
46. A girl pulls a sled 50 ft on level ground with a rope inclined at
28. What is the point-normal form of a plane in R3?
an angle of 30° with the horizontal (the ground). If she applies
29. What is the standard form of a plane in R3?
3 lb of tension to the rope, how much work is performed on
30. What is the formula for the distance from a point to a plane? the sled?
Chapter 9 Review 629
47. Sketch the graph of x = 2', y — 2'+1 for t > 0 by 14. Show that three planes whose normals u, v, w satisfy
a. using the parametric form b. eliminating the parameter uxvw^O intersect in exactly one point.
x y 15. Show that u x (v x w) = (u x v) x w if and only if
48. Graph the line — = ——
F 4 9 v x (w x u) = 0.
a. using the rectangular form b. by parameterizing 16. Given the vectors v = 3i — 2j + k and w = 4i + j — 3k, find
x y z
49. Graph the line - = —= - ||v||, v — w, and 2v + 3w.
a. using the symmetric form b. by parameterizing 17. Given u = i - j + k, v = 3i — 2j + 5k, and w = i + j — k,
find (u — v) • w and (2u + v) x (u — w).
50. Identify the graph of each given equation.
b. x2 = z + y1 18. Given the vectors v = 4i + 2j + k, w = 2i + j - 5k. Find
a. x2 — z2 + y
a. 5v — 3w b. ||2v — w||
x2 y2 x2 y2
c.- 1 c. vector projection of v onto w
2 4 "■ T + T
d. scalar projection of w onto v
e.
x _y _ = 1
y _ £
2 4 9 4 9 19. Find the direction cosines for v = (2i + j) x (i + j — 3k).
g. x2 + y2 + z2 = 9 h. y x2 + z2 20. Find two unit vectors that are parallel to the line
i. y2 + z2 - x2 = -1 j. xz2 _ „2 _L ,2
- yz +z _
x _ y _ z— 1
6 “ 2 ~ 6
Supplementary Problems
21. Find the (acute) angle, rounded to the nearest degree, between
1. A triangle in M3 has vertices A(0, 2, — 1), 5(1, 1,3), and
the intersecting lines
C(1,0, -4).
a. Find the perimeter of the triangle. y - 3 _ z+5 x — 1 y — 3 z +5
and
b. Find the area of the triangle. -1 ~ 2 2 “ -1 -2
c. Find the three vertex angles of the triangle. (Round to the
nearest degree.) 22. Find the area of the parallelogram determined by 3i — 4j and
d. Find a number p such that the points A, 5, C, and -i-j + k
D(p, p, 0) form a tetrahedron of volume v = 100 cubic 23. Find the center and the radius of the sphere
units. Hint: See Problem 50, Section 9.4.
4x2 + 4y2 + 4z2 + 12y — 4z + 1 = 0
Find equations, in both parametric and symmetric forms, of the
lines described in Problems 2—3. Find two additional points on
24. Find the points of intersection of the line x — 6 + 31,
each line.
y — 10 — 2r, z = 5r with each of the coordinate planes.
2. passing through A (1, —2, 3), 5(4, — 1, 2)
25. Find the point of intersection of the planes 3x — y + 4z = 15,
3. passing through 5(1,4, 0) with direction numbers [2, 0, 1 ]
2x + y — 3z = 1, and x + 3y + 5z = 2
4. Find an equation for the line passing through 5(3,4,—1)
26. Find the equation of the plane determined by the intersecting
and parallel to the line of intersection of the planes
lines
x + 2y + 2z + 5 = 0 and 2x + y — 3z - 6 = 0.
Find an equation for the plane satisfying the conditions given in x+3 y z -7 x + 6_y + 5_z-l
Problems 5-12. 3 = ^2 = 6 an 1 “ -3 “ 2
5. the xy-plane
27. Find an equation for the set of all points that are equidistant
6. the plane parallel to the xz-plane passing through (4, 3, 7)
from the planes 3x — 4y + 12z = 6 and 4x + 3z = 7.
7. the plane through (1, -3, 4) with attitude numbers [3, 4,-1]
28. Graph the planes defined by the given equations.
8. the plane through (—1,4,5) and orthogonal to a line with
a. x + 3y 4- 2z = 6 b. y + z = 5 c. x = 4
direction numbers [4, 4, —3]
29. Graph the curve defined by the functions x = 2 cos t,
9. the plane through (4, —3, 2) and parallel to the plane
y = 2 sin t for 0 < t < 2n.
5x — 2y + 3z — 10 = 0
30. Vertices 5 and C of A ABC lie along the line
10. the plane containing
x-3 z— 1 „ , x-3 y +2 z- 1 x + 2_ y — 1 _ z
— = — iy = -2l- and — = — = — 2 “ 1 ~ 4
11. the plane passing through 5(4, 1,3), Q{—4, 2, 1), and Find the area of the triangle given that A has coordinates
5(1, 0,2) (1, —1, 2) and line segment BC has length 5.
12. the plane passing through (4, —1,2) and parallel to the lines 31. Find the work done by the constant force F = 5i + 4j + k
in moving a particle along the line from 5(2, 1,-1) to
x +2 _ y —2 _ z+ 1 ^ x —2 _ y — 3 _ z —4
(2(4, 1,2).
3 “ -1 ~~ 2 an 1 _ 2 3 32. How much work does it take to move a container 25 m along a
13. If u and v are orthogonal unit vectors, show that horizontal loading platform onto a truck using a constant force
(u x v) x u = v. What is (u x v) x v? of 100 newtons at an angle of | from the horizontal?
630 Chapter 9 Review
33. Find a formula for the surface area of the tetrahedron deter¬ 52. Figure 9.69 shows a parallelogram ABCD. If M is the mid¬
mined by vectors u, v, and w. Assume the vectors do not all point of side A6, show that line CM intersects diagonal 5D
lie in the same plane. at a point P located one-third of the distance from B to D by
34. Suppose v and w are nonzero vectors. Show that || v||w+ ||w|| v completing the following steps.
and ||v||w — ||w||v are orthogonal vectors.
35. Let v = cos 6 i + sin 6 j and w = cos <p i + sin 0 j. Find v x w.
Interpret this cross product geometrically, and use it to derive
a well-known trigonometric identity.
36. Find the three vertex angles (rounded to the nearest degree)
of the triangle whose vertices are (1, -2, 3), (-1,2, -3), and
(2, 1,-3).
37. Find a relationship between the numbers au b,, and ci so
that the angle between the vectors v = ati + b\j + cjk
, and w, = i — 2j is the same as the angle between v and
Figure 9 Parallelogram ABCD
w2 = 2i + k.
38. Find an equation of the plane that passes through (a, 0, 0), a. Let a and b be scalars such that MP = nMC and
(0, a, 0), and (0, 0, a). BP = &BD. Show that
39. Find the area of the triangle with vertices A(0, —1,2),
5(1,2,-1), and C(3,-1,2). 2AB + Z?(AD - AB) = a (AD + ^AB)
44. Find a number A such that the angle between the planes w = cj + cij + c3k are linearly dependent (see Problem 51),
if and only if
2Ax + 3y + z = 1 and x - Ay + 3z = 5 is n/2.
45. The lines L\ and L2 are parallel to the vectors Vj = i — j and a2
a1
V2 = i - j + 2k, respectively. Find an equation for the line b\ b3
b2
L that passes through the point (-1,2, 0) and is orthogonal to C\ c2 c3
both Lx and L2.
46. A parallelepiped is determined by the vectors u = i - j + k, 54. Show that
v = > + 2j — k, and w = 2i + j + k. Find the altitude from the
tip of w to the side determined by u and v. ui • v, U, V2
= (u, X U2) • (Vi X v2)
U2 • V, u2 • v2
47. In Chapter 10, we will show that T = i + 2xj is a vector in
the direction of the tangent line at each point 6(x, x2) on the
55. Let A(—2, 3, 7), 6(1,5, —3), and C(2, 8, —1) be the vertices
parabola y — x2. Find a unit vector normal to the parabola at
the point (3, 9). of a triangle in K3. What are the coordinates of the point M
where the medians of the triangle meet (the centroid)?
48. For nonzero vectors u, v, and w, show that the vector
(u x v) x (u x w) is parallel to u. 56. The medians of a triangle meet at a point (the centroid) located
two-thirds of the distance from each vertex to the midpoint of
49. Let v = ai + bj + ck and w = Ai + 6j + Ck, where a, b, c,
the opposite side. Generalize this result by showing that the
A, 5, and C are constants. Describe the set of vectors v + / w,
four lines that join each vertex of a tetrahedron to the centroid
where t is any scalar.
of the opposite face meet at a point located three-fourths of
50. Lise vectors to show that the sum of the squares of the lengths the distance from the vertex to the centroid.
of the sides of a parallelogram equals the sum of the squares
57. Spy Problem The Spy forces Blohardt’s helicopter to land (see
of the lengths of the diagonals.
Problem 53, Section 9.5), but when he and his men approach,
51. The vectors u, v, and w are said to be linearly independent in they discover no occupants—the helicopter is remote con¬
IR3 if the only solution to the equation au + by + cw = 0 trolled! There is, however, a note attached to the control panel.
is a = b = c = 0. Otherwise, the vectors are linearly There are map coordinates, a crude drawing, and these words:
dependent. The simplest example of a set of linearly inde¬ “From the snowman, pace off the distance to the woodpile,
pendent vectors is i, j, and k. Determine whether the vectors turn left, pace off an equal distance and drive a stake. From
u = “• + 2k, v = 2i — j + 3k, w = i + 3 j — 2k are linearly the snowman again, pace off the distance to the flagpole,
independent or dependent. turn right, pace off an equal distance and drive a second
Chapter 9 Review 631
stake. Dig halfway between the stakes.” A bloody letter a. Show that u, a, fi are mutually orthogonal (any pair is or¬
“F” marks the spot on the map between the stakes. thogonal).
b. If y is any vector in R3, show that
“Flamer!” gasps the Spy. Gripped by trepidation, he
arrives at the location indicated by the map coordinates. y a
y u ~y_P'
There is a woodpile and a flagpole, but the snowman has y = u+ of + P
Ju||2_ JPW2.
melted. How can the Spy know where to dig?
58. A triangle in R3 is determined by the vectors v and w as shown 61. Exploration Problem
in Figure 9.70. a. In Figure 9.72, A BCD is a rectangle, with M the midpoint
1
AR<..+i — AB
2£ + 3
A = ^IMPIMI2 - (v-w)2
Lx : x = 1, y = o
Figure 9.71 Problem 59
L2 : y = i. z = 0
It can be shown that APQR is also equilateral, and x = 0
L3 : z = L
||PM|| = IIQNH = ||RO|| and ||AP|| = ||BQ|| = ||CR||.
L4: x — y = —6z
a. Show that PQ = |AN, then show that ||AN||2 = 11|AB||2.
Hint: Use the law of cosines.
63. Putnam Examination Problem Find the equation of the smallest
b. Show that APQR has area | that of A ABC. sphere that is tangent to both the lines
c. Do you think the same result would hold if A ABC were
not equilateral? Investigate your conjecture. L\ : x = t + 1, y = 2t +4, z — -31 + 5 and
60. Gram-Schmidt orthogonalization process Let u, v, and w be L2 : jc = 4r - 12, y = t + 8, z — t + \l
nonzero vectors in R3 that do not lie on the same plane. Define
vectors a and /J as follows: 64. Putnam Examination Problem The hands of an accurate clock
have lengths 3 cm and 4 cm. Find the distance between the
v•u w•u W Of
tips of the hands when the distance is increasing most rapidly.
a =v— u and ft = w u — Of
U Of
Group Research Project*
Star Trek
This project is to be done in groups of three or four students. Each group will submit a
single written report.
T he starship Enterprise has been captured by the evil Romulans and is being held in
orbit by a Romulan tractor beam. The orbit is elliptical with the planet Romulus
at one focus of the ellipse. Repeated efforts to escape have been futile and have
almost exhausted the fuel supplies. Morale is low and food reserves are dwindling.
In searching the ship’s log, Lieutenant Commander Data discovers that the Enter¬
prise had been captured long ago by a Romulan tractor beam and had escaped. The key
to that escape was to fire the ship’s thrusters at exactly the right position in the orbit.
Captain Picard gives the command to feed the required information into the computer
to find that position. But, alas, a Romulan virus has rendered the computer useless for
this task. Everyone turns to you and asks for your help in solving the problem.
Here is what Data discovered. If F represents the focus of the ellipse and P is
TIME IS SAID to have only one dimen¬
the position of the ship on the ellipse, then the vector FP can be written as a sum
sion, and space to have three dimensions.
... The mathematical quaternion par¬ T + N where T is tangent to the ellipse and N is normal to the ellipse (not necessarily
takes of both of these elements; in tech¬ unit vectors). The thrusters must be fired when the ratio ||T||/||N|| is equal to the
nical language it may be said to be “time eccentricity of the ellipse.
plus space,” or “space plus time”; and in Your mission is to save the starship from the evil Romulans.
the sense it has, or at least involves a ref¬
erence to, four dimensions.
-W. R. Hamilton
Graves’ Life of Hamilton
(New York, 1882-1889)
Vol. 3, p. 635
' MAA Notes: 17 (1991) “Priming the Calculus Pump: Innovations and Resources,” by Marcus S. Cohen,
Edward D. Gaughan, R. Arthur Knoebel, Douglas S. Kurtz, and David J. Pengelley.
CONTENTS
10.1 Introduction to Vector Functions
Vector-valued functions
Vector-Valued Functions
Operations with vector functions
Limits and continuity of vector
functions
10.2 Differentiation and Integration of Vector
Functions
Vector derivatives PREVIEW
Tangent vectors
The marriage of calculus and vector methods forms what is called vector calculus.
Properties of vector derivatives
The key to using vector calculus is the concept of a vector-valued function. In this
Modeling the motion of an object
chapter, we introduce such functions and examine some of their properties. We
in R3
will see that vector-valued functions behave much like the scalar-valued functions
Vector integrals
studied earlier in this text.
10.3 Modeling Ballistics and Planetary
Motion
Modeling the motion of a
projectile in a vacuum PERSPECTIVE
Kepler’s second law A car travels down a curved road at a constant speed of 55 mi/h. What additional
10.4 Unit Tangent and Principal Unit Normal information do we need about the car to determine whether it will stay on the road
Vectors; Curvature or skid off as it rounds a particular curve? Can we modify the road (say, by banking)
Unit tangent and principal unit so an average-sized car can travel at moderate speeds without skidding? A soldier
normal vectors fires a howitzer whose muzzle speed and angle of elevation are known. If the shell
Arc length as a parameter overshoots its target by 40 yd, how should the angle of elevation be changed to
Curvature ensure a hit on the next shot? If a satellite is in orbit above the earth, how fast must
10.5 Tangential and Normal Components of it travel to remain stationary above a particular point on the equator? These and
Acceleration other similar questions can be answered using vector calculus.
Components of acceleration
Applications
Chapter 10 Review
Guest Essay: The Stimulation of Science
Howard Eves
Cumulative Review, Chapters 1-10
633
634 Chapter 10 Vector-Valued Functions
VECTOR-VALUED FUNCTIONS
where / and g are functions of t on some interval. We extend this definition to three
dimensions. A curve in R3 is the set of all ordered triples if ft), fiit), hit)) satisfy¬
ing the parametric equations x = f\it), y = fft), z = fft), where fu f2, and /3 are
functions of t on some domain D, which lies in R.
The concept of a vector-valued function is fundamental to the ideas we plan to
explore. Here is a definition.
Vector-Valued Function A vector-valued function (or, simply, a vector function) F of a real variable
with domain D assigns to each number t in the set D a unique vector F(f). The
set of all vectors v of the form v = F(f) for t in D is the range of F. That is,
where fu f2, and /3 are real-valued (scalar-valued) functions of the real number
t defined on the domain set D. In this context, /), f2, and /3 are called the com¬
ponents of F. A vector function may also be denoted by F(f) = (/iit), fiiO)
orF(0 = (/i(0,/2(0,/3(0).
Figure 10.1 The graph of the Let F be a vector function, and suppose the initial point of the vector F(f) is at the
vector function F(r) is traced origin. The graph of F is the curve traced out by the terminal point of the vector F(f) as
out by the terminal point of F(t) t varies over the domain set D, as shown in Figure 10.1. In this context, F(f) is called
as t varies over the domain D. the position vector at t for the point P{fx (t), f2{t), /3(/)) on C.
for all t.
Solution
The graph is the collection of all points (x, y, z) with
x = 3 -t y — 2t z = -4 + 3f
for all t. We recognize these as the parametric equations for the line in R3 that contains
the point Pq(3, 0, —4) and is parallel to the vector
v = -i + 2j + 3k
Figure 10.2 The graph of
F(r) = (3 - r)i + (2f)j + (3f - 4)k as shown in Figure 10.2. ■
Solution
The graph of F is the collection of all points (x, y, z) in R3 whose coordinates satisfy
x = 2sin t y = ~2cosr z — 3t for all t
The first two components satisfy
x2 + y2 = (2 sin t)2 + (—2 cos t)2 — 4(sin21 + cos21) = 4
which means that the graph lies on the surface of the circular cylinder with radius 2 as
shown in Figure 10.3.
Next notice the axis of symmetry for the graph in Figure 10.3 is the z-axis. We
also know that as t increases, the z-coordinate of the point P(x,y, z) on the graph of F
increases according to the formula z — 31, which means that the point (*, y, z) on the
graph rises in a spiral on the surface of the cylinder x2+y2 — 4. The point on the graph
of F that corresponds to t = 0 is (0, —2, 0), and the points that correspond to t = f
and t = tv are (2, 0, and (0, 2, 3;r), respectively. Thus, the graph spirals upward
5 33' 5'
.-MESSENGER RNA
counterclockwise (as viewed from above). The graph, known as a right circular helix,
PROTEIN
is shown in Figure 10.3. ■
A
z,. zj,
SERINE
Circular cylinder
x2 + y2 = 4
ISOLEUCINE
GLYCINE
ALANINE
ALANINE
a. DNA molecule b. A computer-generated image of the helix in c. A computer-generated image of the helix:
Example 2: F(f) = (2 sin f)i - (2 cos f)j + 3fk F(f) = (cos 2f)i + (sin 2f)j + 0.2rk
Hemisphere Find a vector function F whose graph is the curve of intersection of the hemisphere
. I . O
Curve of"
intersection C z = a/4 — x2 — y2 and the parabolic cylinder y — x2, as shown in Figure 10.5.
Solution
Finding a parametric representation is sometimes called parameterizing the curve.
There are several ways this can be done, but one choice is to let x = t. Then y = t2
Cylinder y = x2 (from the equation of the parabolic cylinder), and by substituting into the equation for
the hemisphere, we find
It follows from the definition of vector operations that vector functions can be added,
subtracted, multiplied by a scalar function, and multiplied together. We summarize
these operations in the following box.
Vector Function Operations Let F and G be vector functions of the real variable t, and let f(t) be a scalar
function. Then, F + G, F — G, /F, and F x G are vector functions, and F G is
a scalar function. These operations are defined as follows:
Vector functions:
Solution
a. (F + G)(t) — F(t) + G(t)
1
= l42i + t) - (sin r)k] + t\ + -j + 5k
10.1 Introduction to Vector Functions 637
= [ri + tj — (sinf)k] x ti + yj + 5k
t t — suit
1
t - 5
t
sin t
51 + i — [5f2 + t sin t Jj + [t — f2]k
d. (F-G)(r) = F(r)-G(f)
= [?2i + t] - (sinf)k] • t\ + -j + 5k
= t3 + 1 — 5 sin t
TECHNOLOGY NOTE
Representing a vector function using technology is generally the same whether using a calculator or a software package.
Square brackets are often used to denote vector functions. For Example 4, we denote F as the function y 1 (jc), and G as y2(x):
A sample of these input values is shown in Figure 10.6a. Next, we work through each of the parts of Example 4, as shown in
Figure 10.6b. Notice that if the answer is a vector, it is shown in brackets (as in parts a, b, and c). If the answer is a scalar (as
in part d), the answer is not shown in brackets.
Limit of a Vector Function Suppose the components f\, f2, f3 of the vector function
Solution
= i+ e2j m
Vector limits behave like scalar limits. The following theorem contains some use¬
ful general properties of such limits.
These limit formulas are also valid as t -» +oo or as t —► — oo, assuming all expres¬
sions have finite limits.
Proof We will establish the formula for the limit of a dot product and leave the rest
of the proof as an exercise. Let
Apply the limit of a vector function along with the sum rule and product rule for scalar
limits to write
lim f (t) lim g!(0 + lim f2(t) lim g2(t) + lim f3(t) lim g3(t)
t->to _*-»«0 j-yto J-yto j-yto
lim fi{t) i+ lim f2(t) j + lim /3(0 lim gi(t) i+ lim g2(t) j + lim g3(t)
t-yto t-yto t-y to t-yto t-*t0 t-*t0
Solution
j k
F(t) x G(r) -t t2
0 —(3 + e‘)
Thus,
i j k
limF(r) X lim G(t) 1 0 1
t-+1 t-* 1
e 0 —3 — e
What This Says This is the same as requiring each component of F(t) to
be continuous at to- That is,
The rules for vector limits listed in Theorem 10.1 can be used to derive general
properties of vector function continuity.
Solution
The vector function F is continuous where its component functions
/i(t) = sint /2(t) = (1 - t)_1 /3(f) = In t
are continuous. The function f\ is continuous for all t; /2 is continuous whenever
1 — t 7^ 0 (that is, when t ^ 1); /3 is continuous for t > 0. Thus, F is continuous
when t is a positive number other than 1; that is, t > 0, t ^ 1.
640 Chapter 10 Vector-Valued Functions
1 -t 1 -t7
R(0 = ti + j+
43. Which of the following curves lies on the surface z = x2 + y2? c. lim[F(r) x G(0] = limF(0 lim G{t)
t-+ l t-+i
a. x = sin t, y = cos t, z = t Q 62. If H(0 is a vector function, we define the difference operator
b. x = e~' cos t, y = e~' sin t, z — e~2' AH by the formula
, 1 1 2
c. x = 1 + y = 1-,z = 2 + — AH = H(t + At) - H(0
t t t2
44. Find parametric equations for the curve of intersection of the
where At is a change in the parameter t. {Note: Usually, | At|
elliptical cylinder x2 + 3y2 — 1 and the parabolic cylinder
is a small number.) If F(r) and G(0 are vector functions, show
z = 2x2 — 1.
that A(F x G)(0 = F(r +At) x AG(r) +AF(0 x G(0- Hint:
Find each limit indicated in Problems 45-52. Observe that
45. lim f2/i — 3j + e'kl
r-^1 L J A(F x G)(0 = F(t + At) x G{t + At)
46. lim[3ti + e"'j + (sin7rt)k]
t-*i - F(t + At) x G(0 + F{t + At) x G(0
(sint)i — tk - F(0 x G(0
47. lim
o t2 + t - 1
r,3 1- ^ — 3f + 2 # j . 63. Prove the limit of a sum rule of Theorem 10.1. (The difference
48. lim 1+ -r--—J + (f2 + iy k rule is proved similarly.)
i t - 1 t2 + t -2
1 T 64. a. Prove the limit of a scalar multiple rule of Theorem 10.1.
te'
49. lim i+ b. Prove the limit of a cross product rule of Theorem 10.1.
/-»o 1 — e' cos t
65. Counterexample Problem Let F(r) and G(r) be vector functions
sin t 1 — cos t .
50. lim + -j + e1 'k that are continuous at t0, and let h{t) be a scalar function con¬
r->0 t t tinuous at t0. In each of the following cases, either show that
sin 31 ln(sint). the given function is continuous at /0 or provide a counterex¬
51. lim ■ 1 i 71—77J + (?lnGk
f->0+ sin2t ln(tant) ample to show that it is not continuous.
52. lint [(2i — tj + e'k) x (r2i + 4 sin fj)] a. 3F(r) -f- 5G(r) b. F(?) ■ G(t)
c. (/t(r))_lF(r) d. F(r) x G(r)
Determine all values of t for which the vector function given in
Problems 53-58 is continuous. 66. Counterexample Problem Find a vector function
F(0 = ifiit), fjit), /3(G) such that ||F(r)|| is continuous at
53. F(r) = fi + 3j - (1 - t)k 54. G(r) = ti - yk t = 0 but F(r) is not continuous at t — 0.
642 Chapter 10 Vector-Valued Functions
VECTOR DERIVATIVES
Derivative of a Vector Function The derivative of the vector function F is the vector function F' determined by
the limit
v AF F(f + At) — F(f)
F it) = lim -= lim -
A?—>0 At Ac—>-0 At
wherever this limit exists. In the Leibniz notation, the derivative of F is denoted
The following theorem establishes a convenient method for computing the deriva¬
tive of a vector function.
Proof We use the definition of the derivative, along with rules of vector limits (The¬
orem 10.1), and the fact that the scalar derivatives flit), flit), and /3'(f) all exist.
Fit + At)-Fit)
F'(f) = lim -
A/—>0 At
Solution
The component functions cost and t — 5 are differentiable for all t, but |/| is not
differentiable at t — 0. Thus, the vector function G is differentiable for all / / 0. ■
10.2 Differentiation and Integration of Vector Functions 643
Solution
Differentiating each component separately, we find that
TANGENT VECTORS
Recall that the scalar derivative f'(x0) gives the slope of the tangent line to the graph
of / at the point where x — xo and thus provides a measure of the graph’s direction
at that point. Our first goal is to extend this interpretation by showing how the vector
derivative can be used to find tangent vectors to curves in space. Let t be a number
in the domain of the vector function F(0, and let P0 be the point on the graph of F
that corresponds to to, as shown in Figure 10.7. Then for any positive number At, the
difference quotient
Direction of P0 AF
hrn - ^ 0
Ar->0 At
Then, as At —> 0, the direction of the secant vector, PoQ, and hence that of the
difference quotient AF/Ar, will approach the direction of the tangent vector at Pq,
as shown in Figure 10.8. Thus, we expect the tangent vector at Po to be the limit vector
AF
lim -
Ar->0 At
Figure 10.8 As At —► 0, the vector P0Q
and hence the difference quotient which we recognize as the vector derivative F'(fo). These observations lead to the
AF/Ar approaches the tangent vector following interpretation.
at P0.
Tangent Vector Suppose F(f) is differentiable at to and that F'(fo) 7^ 0. Then F'(fo) is a tangent
vector to the graph of F(/) at the point where t = to and points in the direction
of increasing t.
Solution
The derivative of F(t) is
F'(0 = 2e2,i + (2t — l)j + r'k
If F'(ro) ^ 0 and we also require the derivative F' to be continuous at to, then the
tangent vector at each point of the graph of F near Pq will be close to the tangent vector
at Pq. In this case, the graph is said to be smooth at Pq.
Smooth Curve The graph of the vector function defined by F(?) is smooth on any interval of
t where F' is continuous and F\t) ^ 0. The graph is piecewise smooth on an
interval that can be subdivided into a finite number of subintervals on which F is
smooth.
Solution
The derivative
F'(t) = (21, — sin t, e' — e~l)
is continuous for all t, but F'(0) = (0, 0, 1 - 1) = 0. Thus, the graph is not smooth for
all t, but it is smooth on any interval not containing t = 0.
A graph will not be smooth on any interval containing a point where there is an
X
abrupt change in direction. For example, the graph in Figure 10.9 is not smooth on
any interval containing the point corresponding to the sharp corner P0. In Example 4,
e 10.9 A curve that i s not such a point occurs when t — 0, namely P0(\, 1,2). However, this curve is piecewise
smooth smooth because it consists of a finite number of smooth pieces.
d7, F
third derivative, by
~d?‘
10.2 Differentiation and Integration of Vector Functions 645
Solution
F'(t) = 2e2ti + (—2f)j + (-2 sin 2t)k
F"(t) = 4e2li — 2j — (4cos2r)k
¥"'{t) = 8e2li + (8 sin 2r)k ■
Several rules for computing derivatives of vector functions are listed in the follow¬
ing theorem.
Solution
First find the derivative of the cross product:
i j k
(F' x G)(0 0 1 21 (3 - 2te')i - (—2/2)j + (-r)k
t e* 3
i j
(F x G')(0 1 t = (-tV)i - (-/2)j + (er - Ok
1 e’
(F' x G)(0 + (Fx G')(0 = [(3 - 2 tel) - /V] i + [212 + t2] j + [-/ + e‘ - f]k
= (3 — 2 teT — t2e')i + (3t2)j + (e‘ — 2t)k
Solution
d¥ dG
Begin by finding —- and -—
dt dt
dF t dG
— = e j + 2tk and — = 6ti — e j — 2k
dt J dt
d , dF -,d G d ,
a. —[2F(0 + t G(0] — 2——f ^- + -(,3)G,0
dt dt
2(Oj + 2tk) + t\6ti - e~!j - 2k) + 3r2(3r2i + e~‘j - 2tk)
d dF ^ dG
b. —[F(0 • G O] = -T- ' G O + F 0 • —
dt dt dt
= (Oj + 2tk) ■ (3/2i + e~'j - 2rk) 4- (i + Oj + r2k) • (6fi - e~'\ - 2k)
= —612 + 6/ ■
We will use Theorem 10.3 in both applied and theoretical problems. In the follow¬
ing theorem, we establish an important geometric property of vector functions.
Proof We are given that ||F(/)|| = r for some constant r and all t. To prove that F
and F' are orthogonal, we will show that F(t) • F'(/) = 0 for all t. First, note that
r2 = ||F(f)||2 = F(0 - F(0
for all t. We take the derivative of both sides (remember that the derivative of r2 is zero
because it is a constant).
[r2]' = [F(0 • F(/)]'
0 = F'(f) • F(0 + F(r) • F'(r)
0 = 2F(r) • F\t)
Tangent 0 = F(f) • F'(f)
Thus, F and F' are orthogonal. □
Recall from Chapter 3 that the derivative of an object’s position with respect to time is
the velocity, and the derivative of the velocity is the acceleration. We frequently know
the acceleration and can use integration to find the velocity and the position. We will
now express these concepts in terms of vector functions.
Vector Motion An object that moves in such a way that its position at time t is given by the
vector function R(t). Then
dR
Velocity vector V is —
d\ _ d2R
dt dt2
The graph of the position vector R(f) is called the trajectory of the moving object.
According to the results obtained earlier in this section, the velocity V(/) = R'(r) is a
Figure 10.11 The velocity vector is tangent vector to the trajectory at any point where \(t) exists and V(/) ^4 0 (see Figure
tangent to the trajectory of the moving 10.11). When \(t) = 0, the object is stationary. The direction of motion is given by
object the unit vector V/||V||, so the velocity satisfies
= (SPEED) (DIRECTION)
648 Chapter 10 Vector-Valued Functions
dR
Solution V (— sin t) i + (cos t)j + 3rk
dt
d\
A (— cos t) i — (sin t )j + 6fk
dt
The velocity at time t — 2 is V(2):
(- sin 2)i + (cos 2)j + 3(2)2k « -0.91i - 0.42j + 12k
The acceleration at t — 2 is
(— cos 2)i — (sin2)j + 6(2)k w 0.42i — 0.9lj + 12k
The speed is
= y71 + 9t4
At time t = 2, the speed is ||V(2)|| = y/\ + 144 ~ 12.04. The direction of motion is
1
[(— sin t )i + (cos r)j + 3r k]
l|V|| 7145
At time t = 2, the direction is V(2)/||V(2)||:
1
[(— sin 2)i -f (cos 2)j + 12k] % —0.08i — 0.03j + k
Figure 10.12 Graph of R and V 7145
at t = 2 The geometry of this example is shown in Figure 10.12.
VECTOR INTEGRALS
Vector Integrals Let F(f) = f\(t)i + {t)j + /3(f)k, where f\, fi, and /3 are continuous on the
closed interval a < t < b. Then the indefinite integral of F(t) is the vector
function
where C = Cii + C2j + C3k is an arbitrary constant vector. The definite integral
of F(t) on a < t < b is the vector
1
f f
" .
m
(sin t)k] dt = tdt i + j - / sin tdt
Jo .Jo Jo
i + (3f)lo j - (-cosr)lo k
= \tX2{
2
+ 371} + (cos Tt — cosO)]k
= j1 7^2
z i + 3tt} — 2k
Solution
We need to solve the initial value problem that consists of
dK , 9
The differential equation \(t) = — = e i + rj + (cos 2t)k
dt
so
C\ = 1, C2 = 1, C3 = -l
and
i+j-k = C
Find the tangent vector to the graph of the given vector function F
46. (3F - 2G)'fr) = 3F'(r) - 2G'(r)
at the points indicated in Problems 19-22.
47. (F • G)'(r) = (F • G)(r) + (F • G')(r)
19. F(r) = r2i + 2rj + (t3 +12)k; t = 0, / = 1, t
In Problems 48-49, find a value of a that satisfies the given equa¬
ri 4- r2j + r3k
20. F it) = ; t = 0, r = 2 tion.
1 +2r
21. F(r) = r2i 4- (cos r)j 4- (r2 cos r)k; r = 0 t = , | 48. [[{tVTLr) i+( 1
1 +r2
1 j dt 42V2- l)i+fj
Find the position vector R(r), given the velocity V(r) and the ini¬
55. Let F(r) be a differentiable vector function and let h(t) be a
tial position R(0) in Problems 31-34.
differentiable scalar function of r. Show that
In general, it can be quite difficult to analyze the motion of a projectile, but the problem
becomes manageable if we assume that the acceleration due to gravity is constant and
that the projectile travels in a vacuum, so we will make this assumption. A model of the
motion in the real world (with air resistance, for example) based on these assumptions
is fairly realistic as long as the projectile is reasonably heavy, travels at relatively low
speed, and stays close to the surface of the earth.
Figure 10.13 shows a projectile that is fired from a point P and travels in a vertical
plane coordinatized so that P is directly above the origin O and the impact point / is
on the x-axis at ground level. We let 50 be the height of P above O, uq be the initial
speed of the projectile (the muzzle speed), and a be the angle of elevation (the angle
at which the projectile is launched).
Because the projectile is to travel in a vacuum, we will assume that the only force
acting on the projectile at any given time is the force due to gravity. This force F
is directed downward. The magnitude of this force is the projectile’s weight. If m
is the mass of the projectile and g is the “free-fall” acceleration due to gravity (g is
Figure 10.13 The motion of a projectile approximately 32 ft/s2 or 9.8 m/s2), then the magnitude of F is mg. Also, according
to Newton’s second law of motion, the sum of all the forces acting on the projectile
must equal mA{t), where A(t) is the acceleration of the projectile at time t. Because
F = — mg’] is the only force acting on the projectile, it follows that
—mgj — F = mA(t)
v0 sin a A(t) — —gj for all t > 0
The velocity V(r) of the projectile can now be obtained by integrating A(f). Specif¬
v0 cos a
ically, we have
Next, by integrating V with respect to t, we find that the position vector of the
projectile is
where C2 = R(0) is the position at time t = 0. Because the projectile begins its flight
at P, the initial position is R(0) = soj, and by substituting this expression for C2, we
find that the position at time t is
Specifically, R(t) gives the position of the projectile at any time t after it is fired and
before it hits the ground. If we let x(t) and y(t) denote the horizontal and vertical
components of R(t), respectively, we can make the following statement.
The parametric equations for the motion of a projectile in a vacuum provide useful
general information about a projectile’s motion. For instance, note that if a 7^ 90°, we
can eliminate the parameter t by solving the first equation, obtaining
t
r cos a
By substituting into the second equation, we find
-12
1
y = :8 + (v0 sin or) + £0
Vqcos a Vqcos a
~g
x + (tanar)x + jq
2(vq cos a)2
This is the Cartesian equation for the trajectory of the projectile, and because the equa¬
tion has the general form y = ax2 + bx + c, with a < 0, the trajectory must be part
Figure 10.15 The motion of a of a downward-opening parabola. That is, if a projectile is not fired vertically (so that
projectile. The trajectory is part of tana is defined), its trajectory will be part of a downward-opening parabolic arc, as
a parabolic arc. shown in Figure 10.15.
The time of flight Tf of a projectile is the elapsed time between launch and impact,
and the range is the total horizontal distance R / traveled by the projectile during its
flight. Because y = 0 at impact, it follows that Tf must satisfy the quadratic equation
Then, because the flight begins at x — 0 when t — 0 and ends at x = Rf when t = Tf,
the range R f can be computed by evaluating x(t) at t = Tf. That is.
Rf = (v 0 cosajTy
10.3 Modeling Ballistics and Planetary Motion 653
a. What are the time of flight of the ball and its range?
b. What are the velocity of the ball and its speed at impact?
c. What is the highest point reached by the ball during its flight?
Solution
We know that g = 32 ft/s2, and we are given so = 48 ft, vq = 64 ft/s, and a — 30°,
so the ball’s trajectory has parametric equations
= -16r2 4- 32? 4- 48
a. The ball hits the ground when y = 0, and by solving the equation
— 16r2 4 32f 4- 48
- 0 for t > 0, we find that t — 3, so the time of flight is
= Tf — 3
seconds. The range is
That is, the ball hits the ground about 166 ft from the base of the cliff.
b. We find that x'(t) = 32^ and y'(t) — -321 4- 32, so the velocity at time t is
Thus, at impact (when t = 3), the velocity is V(3) = 32^3 i - 64j, and its speed is
and the highest point reached by the ball has coordinates (rounded to the nearest
foot) of (55, 64). ■
In general, if the projectile is fired from ground level (so so = 0), the time of flight
Tf satisfies
/ 2uo sin or \ Vn . N Un . _
Rf = x{Tf) = (u0 cos a) I - I = —(2 sin a cos a) = — sin 2a
\ $ $
Notice that for a given initial speed vq, the range assumes its largest value when
sin 2a = 1; that is, when or = 45° = f. To summarize.
654 Chapter 10 Vector-Valued Functions
2 vl;
Tf = -vosina and Rf = — sin2o'
8 8
v2
The maximal range is Rm = —, and it occurs when a = |.
Solution
With g — 32 ft/s2, we see that the flight time Tf is
2 2
Tf — -u0 sin a = — (110)(sin40°) 4.4191648
8
The range Rf is
v2 ■ (HO)2
Rf = — sm 2c* = 2__L(sin 80°) ^ 372.38043
That is, the projectile travels about 372 ft horizontally before it hits the ground, and the
flight takes a little more than 4 seconds. ■
In the seventeenth century, the German astronomer Johannes Kepler (1571-1630) for¬
mulated three useful laws for describing planetary motion. The guest essay at the end
of this chapter discusses the discovery of these laws.
Kepler's Laws
1. The planets move about the sun in elliptical orbits, with the sun at one focus.
2. The radius vector joining a planet to the sun sweeps out equal areas in equal
intervals of time.
3. The square of the time of one complete revolution of a planet about its orbit
is proportional to the cube of the length of the semimajor axis of its orbit.
Kepler’s approach was empirical, but his three laws were proved eighty years later
by Isaac Newton, in the Principia Mathematica (1687). We will prove the second law
using vector methods. The other two laws can be established similarly.
We begin by introducing some useful notation involving polar coordinates. Let ur
Figure 10.16 Description of the and u# denote unit vectors along the radial axis and orthogonal to that axis, respec¬
motion of a particle along a curve tively, as shown in Figure 10.16. Then, in terms of the unit Cartesian vectors i and j,
we have
dur
The derivatives I clyXl- satisfy
~cW dQ
dllr
= (— sin0)i + (cos0)j =
Ho
due
(— cos 0)i + (— sin 0) j = — U,
~dO
Now, suppose the sun S is at the origin (pole) of a polar coordinate system, and
consider the motion of a body B (planet, comet, artificial satellite) about S. The radial
vector R = SB can be expressed as
R = ru,- = (r cos 0)\ + (r sin 0)j
where r = ||R||, and the velocity V satisfies
dR dr du,
V =-= —u,- + r- Derivative of a scalar multiple
dt dt dt
dr dur dO
— —u,- + r-Chain rule
dt d6 dt
drd6
= + r—ue
—Ur
dt dt
A similar formula for acceleration can be derived (see Problem 41). We summarize
these formulas in the following box.
dr dO
V(f) = — W +r— ue
dt dt
K)|
to
0
II
II
II
U,. + u
to
Si.
EC
Solution
We note that on the trajectory, x = 2t, y — —t2, so R has length
and because the polar angle satisfies 0 = tan 1 for all points (x, y) on the curve
(except where x = 0), we have
dO 1 / 1\_ -2
dl~ l+(-|)2\ 2/ _ t2 + 4
656 Chapter 10 Vector-Valued Functions
Thus,
Time T Proof The situation described by Kepler’s second law is illustrated in Figure 10.17.
We will assume that the only force acting on a planet is the gravitational attraction of
the sun. According to the universal law of gravitation, the force of attraction is given
by
mM
F = -G u,
Figure 10.17 The radial line sweeps out
where G is a physical constant, and m and M are the masses of the planet and the sun,
equal area in equal time
respectively. Because this is the only force acting on the planet, Newton’s second law
of motion tells us that F = mA, where A is the acceleration of the planet in its orbit.
By equating these two expressions for the force F we find
mM
m A = — G—-—ur
— GM
A = —Tu,
rl
This says that the acceleration of a planet in its orbit has only a radial component.
Note how this conclusion depends on our assumption that the only force on the planet
is the sun’s gravitational attraction, which acts along radial lines from the sun to the
planet.
This means that the u<9 component of the planet’s acceleration is 0, and by exam¬
ining the polar formula for acceleration, we see that this condition is equivalent to the
differential equation
d20 dr dO
r "TT 2— — = 0
dt2 dt dt
« dO . d20 dw
It we set w = —, we obtain —- ——, so that
dt dt2 dt
dw dr
r- + 2w — 0 Separation of variables
dt dt
-l
dw dr
w —2 r~l
dt dt
j w 1 dw = J (—2r J) dr
In | id | + C\ = —2 In | r | -f- C2
Planet
Now, let [fi, r2] and fo, £*] be two time intervals of equal length, so f2 —fi — U — t^.
According to Theorem 6.1, the area swept out in the time period [t\, t{] (see Figure
10.18) is
ti i
,de *2 1 2 d0 „
dt Cdt Since r — — C
s'=[\rU9=L 2 dt -L dt
Thus, Si = \C(t2 - t\). Similarly, the area swept out in time period |>3,/4] is
S2 = \C(t4, — tf), and we have
Figure 10.18 The radial line sweeps An object is said to move in a central force field if it is subject to a single force
fSllf- oppo of tnp
dA _
1 2dd
_ __ _
that is always directed toward a particular point. Our derivation of Kepler’s second
dt 2 dt' law is based on the assumption that planetary motion occurs in a central force field. In
particular, the formula
-G M
A = —— u,
rL
shows that a planet’s acceleration has only a radial component, and this turns out to be
true for an object moving in a plane in any central force field.
Much of what we know about the motion of the planets is derived from Kepler’s
laws. Some of these data are listed in Table 10.1.
15. A shell fired from ground level at an angle of 45° hits the
ground 2000 m away. What is the muzzle speed of the shell?
16. A gun at ground level has muzzle speed of 300 ft/s. What
angle of elevation (to the nearest tenth of a degree) should be
used to hit an object 1,500 ft away?
17. At what angle (to the nearest tenth of a degree) should a pro¬
jectile be fired from ground level if its muzzle speed is 167.1
ft/s and the desired range is 600 ft? Figure 10.21 Evel Knievel’s jump
18. A shell is fired at ground level with a muzzle speed of 280 ft/s Assume that the opposite edges of the canyon are at the same
and at an elevation of 45° from ground level. height and neglect wind resistance.
a. Find the maximum height attained by the shell. 24. Payton Manning, a quarterback for the Indianapolis Colts,
throws a pass at a 45° angle from a height of 6.5 ft with a
b. Find the time of flight and the range of the shell.
speed of 50 ft/s (see Figure 10.22). A receiver races straight
c. Find the velocity and speed of the shell at impact. downfield on a fly pattern at a constant speed of 32 ft/s and
catches the ball at a height of 6 ft. If Manning and the receiver
19. Derek Jeter hits a baseball at a 30° angle with a speed of 144
both line up on the 50 yard line at the start of the play, how far
ft/s. If the ball is 4 ft above ground level when it is hit, what
does Manning fade back (retreat) from the line of scrimmage
is the maximum height reached by the ball? How far will it
before he releases the ball?
travel from home plate before it lands? If it just barely clears
a 5-foot wall in the outfield (see Figure 10.19), how far (to the
nearest foot) is the wall from home plate?
23. Modeling Problem In 1974, Evel Knievel attempted a skycycle 27. Spy Problem The Spy figures out where to dig (Supplemen¬
ride across the Snake River (see Figure 10.21), and at the time tary Problem 57 of Chapter 9) and unearths a human-size box.
there was a great deal of hype about “will he make it?” If With shaking hands, he opens the lid, and to his relief, finds
the angle of the launching ramp was 45° and if the horizontal not the body of Lord Fleming, but his favorite pipe and a ring¬
distance the skycycle needed to travel was 4.700 ft, at what ing cell phone. He picks up the phone and hears the cold, cal¬
speed did Evel have to leave the ramp to make it across the culating voice of Ernst Stavro Blohardt. “Do what you’re told
Snake River Canyon? or the next hox will be heavier!” he threatens. “You’re going
10.3 Modeling Ballistics and Planetary Motion 659
to help me eliminate my competition, starting with Scelerat.” In Problems 31-34, consider a particle moving on a circular path
The Spy is still peeved with Scelerat for icing his friend, Siggy of radius a described by the equation
Leiter (Problem 62, Section 2.4), so he agrees. He’s told that d6
R(f) = (acosa)t)i + (a sin cur)j, where a> = — is the constant
the Frenchman and his gang are holed up in a bunker on the dt
side of a hill inclined at an angle of 15° to the horizontal, as angular velocity.
shown in Figure 10.24. The Spy returns to his helicopter and 31. Find the velocity vector and show that it is orthogonal to R(r).
flies toward the bunker traveling at 200 ft/s at an altitude of 32. Find the speed of the particle.
10,000 ft. Just as the helicopter flies over the base of the hill,
33. Find the acceleration vector and show that its direction is al¬
the Spy sights a ventilation hole in the top of the bunker at an
ways toward the center of the circle.
angle of 20°. He decides to drop a canister of knockout gas
into the bunker through the hole. How long should he wait 34. Find the magnitude of the acceleration vector.
before releasing the canister? 35. A mortar is placed at the top of a hill sloping downward at a
20° angle, as shown in Figure 10.27.
where Rm tan am = —.
vo
8
c. Show that
Figure 10.26 Path of an object thrown from a building
30. A particle moves along the polar path (r, 9), where vo
R max — Jvl + 2gSo and am = tan
r(t) = 3 + 2sin/, 6{t) = f3. Find V(r) and A(r) in terms 8 v
of u, and us.
660 Chapter 10 Vector-Valued Functions
41. Use the formula 43. Modeling Problem A cannon is placed at the top of a
. dr d6 downward-sloping hill as in Problem 35. A shell is fired with
\{t) = — u, +r—u0 muzzle speed u0 at an angle of elevation a with respect to the
dt dt
horizontal. Assume the hill is sloped at an angle with the
to derive the formula for polar acceleration: horizontal, as shown in Figure 10.28.
MO =
dt2 r-Vl
{dt) _
U, +
' d20
dt2
dr dO 1
r—— + 2--
dt dt
ue
dur dO du„ dO
— = lie— and —— = -ur —
dt dt dt dt
42. Use the results of Problem 41 to show that the force F(?) act-
« ing on an object moving with acceleration A(t) acts only in a
radial direction (parallel to ur) if and only if Figure 10.28 Projectile fired on a sloping hill
7d0 a. Derive a formula for the time of flight Tf of the shell and
mr~— = C
dt the downhill range Rd (measured along the slope).
where C is a constant. Under what conditions will F(f) act b. What should a be in order to maximize the downhill range
only in a direction orthogonal to ur? Rdl
In Section 10.2, we observed that if R(?) is a vector function whose graph is a smooth
curve C, then the nonzero derivative R'(t) is tangent to C at the point P that corre¬
sponds to t. Since the curve C is smooth, we have R'(f) ^ 0 and a unit tangent
vector T to C can be obtained by dividing R'(t) by its length, that is,
R V)
T(0 =
IIR'(Oil
Figure 10.29 shows a smooth curve C and the unit tangent T at a particular point P.
orthogonal to T
According to Theorem 10.4 (orthogonality of a function of constant length and its
Figure 10.29 Unit tangent vector T and
derivative), since T(0 has constant magnitude (namely ||T(OH = 1), it follows that
vectors orthogonal to T on a given its derivative T'(0 is orthogonal to T(0 and hence is normal to the curve C at P. As
curve indicated in Figure 10.29, there are infinitely many vectors that are orthogonal to T
and thus deserve to be called normal vectors. To distinguish T'(f) from the others, we
normalize it by dividing by its length and refer to it as the principal unit normal N(f),
so
T'(r)
N(f) =
T'(f)ll
To summarize.
10.4 Unit Tangent and Principal Unit Normal Vectors; Curvature 661
Unit Tangent and Principal Unit If R(r) is a vector function that defines a smooth graph, then at each point a unit
Normal Vectors tangent vector is
R'(f)
T (t)
|R'(OH
and the principal unit normal vector is
T(t)
N(f) =
|T'(OII
Note these are pairs of unit vectors that change direction depending on where you are
on the curve.
Solution
We have R'(f) = (3 cost, 4, -3 sinr), so
= V9 cos2 r + 16 + 9 sin21
R' (j\
T(r) =-= \ (3 cos t, 4, —3 sin /)
IIR'(Oil 5
= cost, | sinr)
To find the principal unit normal N, we first find T'(r) and then its magnitude:
= Vitsin2 r + cos2 r)
_ 3
— 5
Suppose an object moves along the graph C of a vector function R(r) in the direc¬
tion of increasing t. Then at each point P on the trajectory C, the unit tangent T(f)
points in the direction of motion, while the principal unit normal N(f) points in the
direction the object is turning, as shown in Figure 10.30a. For a trajectory in a plane,
this means that N(/) points toward the concave side of the trajectory (Figure 10.30b).
Figure 10.30 Unit tangent T and principal unit normal N at each point on the
trajectory of a moving object
Recall from Section 6.4 that the arc length of the graph of y = F(x) on the interval
[a, b] is given by the integral
5 =
If the graph is described by parametric equations x = x{t) and y — y(t), then by the
, . , dy dy dx
chain rule —■ =- so we have
dt dx dt
dy_
dy_ _ dt „ dx
for _ # 0.
dx dx
dt
Thus, if the graph is traced out exactly once over the parameter interval t\ < t < r2, its
arc length is given by
dt
10.4 Unit Tangent and Principal Unit Normal Vectors; Curvature 663
Often, however, we want a formula for arc length that does not depend on / or any
other particular parameter used in its computation. It can be shown that the following
definition of the arc length function is, indeed, independent of parameterization.
Arc Length Function Let C be a piecewise-smooth curve that is the graph of the vector described pa-
rameterically by R(r) = jc(f)i + y(/)j + z(f)k. and let PQ = P(t0) be a particular
point on C (called the base point). Then the length of C from the base point PQ
to the variable point P(t) is given by the arc length function s(t) defined by
s(t) - (^)\(^)2+(d-i)2du
\ du ) \du ) \du )
4 What This Says The arc length function s(t) measures the distance along
the curve C from P(t0) to P(t) if t > t0 and the opposite of that distance if t < t0
(see Figure 10.31). Since the length along C is a geometric property, it does not
depend on the fact that we have used t as a parameter for describing R(t).
Figure 10.31 s(t) is the length of the This definition allows us to use arc length s(t) as a parameter for describing a
curve C from the point P(to) to curve. While time is a natural parameter for studying motion in space, it is usually more
P(t) if r > t0. fruitful to use the arc length parameter for describing a curve’s geometric features.
Solution
We find that
dx , „ dy dz _
— = 12, — = -5 sin /, — = —5 cos /
dt dt dt
Thus,
= 26
664 Chapter 10 Vector-Valued Functions
The formula for arc length can be used to obtain the following useful formula for
the speed of an object moving in space.
Proof Apply the second fundamental theorem of calculus to the arc length formula
(where we use u as the variable of integration to distinguish it from t):
= HR'(Oil □
Solution
By differentiating R(t) with respect to t, we find the velocity:
The distance traveled by the object between times t — 0 and t = 1 is the arc length
and is given by
To change a given parameterization from a given form, say R(t), to a form involv¬
ing the parameter 5, we first find s(t) using the arc length formula (as in the proof of
Theorem 10.6), and then solve for t as a function of s. The graph of R(t) can then
be parameterized in terms of s by substituting t = t(s). The general procedure is
illustrated in the following example.
Solution
The point P$ corresponds to t = 0, so arc length s(t) is given by the integral
it
s(t) =
l du ) \du) + ©du
— / y/~5 du
Jo
= V5i
1
Solving 5 = \/5 t for t, we obtain t = —s, and the required reparameterization is
V5
/ s \ 2s
x — sin
.VS. y = a*V7s. V5
Expressing a given position vector R in terms of the arc length parameter 5 often
leads to complicated computation. However, once we have such a parameterization,
the corresponding unit tangent vector T and principal unit normal vector N can be
expressed in the relatively simple forms shown in the following theorem.
dR dRds , . ds . .
Proof By the chain rule, we have —— = ———, and since — = ||R (t)||, it follows
J J dt ds dt dt
that
dR
dR dt R'(0
= T
ds ds |R'(0II
dt
Next, to obtain the formula for the principal unit normal vector
T'(0
N =
|T'(f)ll
in terms of the arc length parameter 5, we first note that
, dT dT ds
Chain rule
w dt ds dt
ds dT dT . . L
Because — > 0, it follows that the vector derivatives — and — point in the same
dt dt ds
direction, and N can be computed by finding — and dividing by its length; that is,
ds
dT
ds 1 dl
N =
dT k ds
ds
as claimed.
666 Chapter 10 Vector-Valued Functions
CURVATURE
Let C be a smooth curve that is the graph of a vector function R(s) parameterized in
terms of arc length 5. Since the unit tangent T(s) is a unit vector, only its direction
changes with s and the rate of change of that direction may be measured by the mag-
dT
nitude of the derivative —. As shown in Figure 10.32a, if the curve is a line, then the
ds
. dT
direction remains constant so that — = 0, but if the curve is bent sharply, the direc-
ds
dT
tion of T changes rapidly so — is relatively large in magnitude (Figure 10.32b). The
ds
dT
magnitude of — is called the curvature of C.
ds
C^/T
Figure 10.32 The sharpness of a curve C is related to the magnitude of —.
ds
Curvature Suppose the smooth curve C is the graph of the vector function R(s), parameter¬
ized in terms of the arc length 5. Then the curvature of C is the function
dT
k(s) =
ds
where T(j) is the unit tangent vector.
First, notice that k in Theorem 10.7 is the curvature of Rfs). Since a vector func¬
tion R is often represented in terms of a parameter other than arc length s, it may not
jm
be especially convenient to compute the curvature using the formula k — —r~ . For
ds
instance, if the parameter is t, then according to the chain rule, we have
dT dT ds
~dt ds dt
so that
dT
dT dt
ds ds
dt
o . ds
(Remember, s increases in the same direction as /, so — > 0.) Since — = IIR'(0II,
dt dt
it follows that
moii
IIR'COII
10.4 Unit Tangent and Principal Unit Normal Vectors; Curvature 667
which we will refer to as the two derivatives form of the curvature. This form is
often used in practice to compute k because it involves only the magnitudes of R'(f)
dT
and T'(0, both of which are usually easier to compute than . In Example 5, we
ds
illustrate the use of the two derivatives form by computing the curvature of a circle.
Solution
Since R'(f) = (—r sin t, r cos t), we have
Then
and
Since
the curvature is
K
mo ii
- - = -
i _
■
IIR'(Oil r
||T'(OII
The definition of curvature, k =-, can lead to difficult computations when
IIR'COII
||R'(OH is not a constant. The formula for curvature derived in the following theorem
may then be easier to apply.
* “ IIR'II3
/ ds
Proof Since ||R || = —, we have
Chapter 10 Vector-Valued Functions
Therefore, we have
/ds \ /d2s^ ds .
R x R = I —T ) x ( —T + —T
\dt2 dt
( ds\
= ( — J (T x T') Since T x T = 0
Since ||T|| = 1, it follows from the orthogonality of a function of constant length and
its derivative (Theorem 10.4) that T • T' = 0, so the angle between T and T' is 0 —
Thus,
|R/||2||T,II
|T'||
K
jl^jf
|| R' x R"|
IIRHI2
IIR'II
IR' x R" ||
□
IR'II3
Solution
i j k
R x R = —a sin? a cost b = (ab sin ?)i — (ab cos ?)j + cr k
—a cos t —a sin t 0
We now need to find the magnitude of this vector, as well as that of R':
= Ja2b2(sin-1 + cos2 t) + aA
= yj u2b2 + a4
and
Then we have
|R x R i\J a2 + b2 a\! a1 + b2
k =
IR'II3 (■s/a2 + b2)3 {a2+b2)Va2Tb2 «2 + b2
In the special case where C is a curve in the plane with the equation y = f(x), the
curvature has the form given in the following theorem.
_\f
" (i + [/'«]2)3/2
wherever fix) and its derivatives fix) and fix) all exist.
Proof Using jc as a parameter, we can regard the curve C as the graph of the vector
function R(x) = xi + /(;c)j. Then, R' = i + fix)} and R" = fix)}, so
i j k
R x R' 1 fix) 0 = fix) k
0 fix) 0
HR' x R"||
(v/r+Trciyp)3
\f"M\
(1 + 1/'U)12)3'2
Solution
For f(x) = x~\ we have fix) = —x~2 and fix) = 2x-3, so the curvature is
k =
|ru)|
(i+[/'(*)p)3/2
2x-3
(1 + [--v-2]2)V2
2x3
(x4 + l)3/2
To find the largest value of k, we compute the derivative of k (x) and solve the equation
k'(x) = 0:
6x2(1 — x4)
k'(x)
(x4 + l)5/2
6x2(1 — x4) = 0
Since k\x) > 0 for x < 1 and k'(x) < 0 for x > 1, it follows that the largest value of
the curvature must occur at x = 1. ■
Suppose a curve C in the plane has curvature k ^ 0 at a point P. Then the circle
of radius p — — that is centered on the concave side of the curve and shares a common
K
tangent with C at P is called the osculating (kissing) circle at P. The radius p of the
osculating circle is called the radius of curvature and the center of the circle is called
the center of curvature. The osculating circle and the curve C are tangent and have
the same curvature at P, and in this sense, the osculating circle is the circle that best
approximates the curve C at P.
Figure 10.34 Osculating circles to Figure 10.34 shows the curve y = x-1 for x > 0 whose curvature we found in
the curve y = x-1 at the points Example 7, along with the osculating circle at the point P(l, 1) that corresponds to
(1, 1) and (3, |) x = 1. Note that the curvature at P is
2(1)3 1
*(!) =
(14+ 1)3/2 T2
P(1) = -E = v/2
k{ 1)
In Problem 56, you are asked to show that the osculating circle at P has the equation
(x - 2)2 + (y - 2)2 = 2
dT
Arc length parameter R(s) Definition
ds
37. y — x2\ functional form b. Find an equation for the osculating circle at the point
P(32, 12) on the curve C defined by x = 32/,
38. y — jv3; functional form
y = 16/2 — 4.
39. y = x~2', functional form
51. The Cornu spiral is a parametric curve described by the equa¬
40. y — sin.x; functional fonn
tions
41. the spiral r = e6\ polar form
42. the cardioid r — 1 + cos 0, for 0 < 0 <2n\ polar form
43. A pestus houseflyus is observed to zip around a room in such
a way that at time / its position with respect to the nose of an
It is used in optics to study amplitudes of light waves, and in
observer is given by the vector function
the design of highway off ramps. It also has interesting math¬
ematical properties.
R(/) — /u + /2v + 2 (|r)3/2 (u x v)
a. Show that the arc length of the spiral measured from the
where u and v are unit vectors separated by an angle of 60°. origin (0, 0) satisfies s = t.
* Compute the fly’s speed and find how long it takes to move a b. Show that the curvature of the spiral satisfies k(s) — n |s|.
distance of 20 units along its path (starting from the nose), as Thus, for s > 0, the curvature increases at a constant rate.
shown in Figure 10.35. c. What happens to the curvature, k, as s —> +oo? That is,
evaluate lim k(s).
J-J-+00
Q 52. If T and N are the unit tangent and normal vectors, respec¬
tively, on the trajectory of a moving body, then the cross prod¬
uct vector B = T x N is called the unit binormal of the tra¬
jectory. Three planes determined by T, N, and B are shown in
Figure 10.36.
Unit Normal
Figure 10.35 Path of a pestus houseflyus
binormal plane
44. WHAT DOES THIS SAY?
arc length as a parameter.
Describe what is meant by using
plane A Curve
45. WHAT DOES THIS SAY? Discuss the various curvature for¬ Unit principal
mulas given in Table 10.2. Unit
normal
46. Let C be a smooth curve in M2 described by the parametric
equations x — x(t) and y = y(/).
Osculating
a. Show that the curvature of C is given by plane
|x'y" - y'x" |
Figure 10.36 Three planes determined by T, N, and B
[Or')2 + (y')2]3/2
dB
b. Find the curvature of the curve described by x = 1 — cos /, a. Show that T is orthogonal to —. Hint: Differentiate B • T.
ds
y = 2 + sin/.
dB
47. Let / be a twice differentiable function in M2 described by the b. Show that B is orthogonal to —. Hint: Differentiate B • B.
ds
polar function r = /(0). Show that the curvature is
dB
c. Show that — = —rN for some constant r. Note: z is
\r2+ 2ra-rr"\ ds
called the torsion of the trajectory.
K^ = (r2 + r/2)3/2
53. Prove the Frenet-Serret formulas:
Hint: Use the formula in Problem 46 with x = f(6) cos (9,
dT dN dB
y = /(0)sin0. = KN = — /cT + rB = -rN
ds ds ds
48. Use the formula in Problem 47 to find the curvature of the
graphs with the given equations. where k is the curvature and r = r(s) is a scalar function
a. the cardioid r = 1 + cos 0 called the torsion, which provides a measure of the amount of
twisting at each point on the trajectory.
b. the spiral r = 0
54. Show that the torsion may be computed by the formula
c. the circle r = sin 0
49. Show that the formula for polar curvature in Problem 47 re- [R'(/) x R"(/)] • R'"(/)
2
duces to k = — at the pole (where r = 0). IIR'(/) x R"(/)||2
\r'\
50. Let P(a,b) be a point on the graph C of the vector function 55. A highway has an exit ramp that begins at the origin and fol¬
R(r). lows the curve y — y^x5/2 to the point (4, 1). Then it follows
a. Describe a general procedure for finding an equation for the shape of the osculating circle at (4, 1) until the point where
the osculating circle to C at P. y = 3. What is the total length of the exit ramp?
10.5 Tangential and Normal Components of Acceleration 673
56. Show that the osculating circle at the point (1, 1) on the curve What are the largest and smallest values of the curvature K{t)l
y — x~x has the equation
(X - 2)2 + (y - 2)2 = 2 58. Counterexample Problem Let / be a function that is twice dif¬
ferentiable. Either prove that the largest curvature of the curve
57. Find the curvature of the cycloid with parametric equations
y — f (x) occurs at a relative extremum of / or find a coun¬
x = t — sin t, y = 1 — cos t terexample.
COMPONENTS OF ACCELERATION
When a body is caused to accelerate or brakes are applied, it is of interest to know how
much of the acceleration acts in the direction of the body’s motion, as indicated by the
unit tangent vector T and how much in the direction of the principal unit normal N.
This question is answered by the following theorem.
Proof The formula for V follows from the fact that the velocity vector has magnitude
ds
|| V || = — from Theorem 10.6 and points in the direction of the unit tangent. To
dt
establish the formula for A, first note that
dT dT
N = /cN
ds ds
since the derivative of T(^) points in a direction orthogonal to T. Using this formula,
we find
dV
A = Definition of A
dt
d r
— T—
dt dt
d2s ds dT
—T H-— Product rule
dt1 dt dt
d2s ds ~dTds
Chain rule
dt2 dt ds dt
674 Chapter 10 Vector-Valued Functions
What This Says At each point on the trajectory of a moving object, the
velocity V points in the direction of the unit tangent T, but the acceleration A
may have both a tangential and normal component. The trajectory may twist and
turn, but the acceleration is always in the plane determined by T and the unit
normal N.
where
d~s
At — is the tangential component.
dt2
/ ds \
II
3:
R R \R' x R
At — and An —
IR' HR'II
Proof Let 6 be the angle between R' and R". Then, since the acceleration is given
by A = R", we have
|R'||||A|| cos# R R
At = IIA II cos 6 =
IR IR
and
R ||A|| sin# R' x R"||
An — ||A|| sin# □
IR IIR'II
Solution
Since R'(t) = \{t) = (312, 21, 1) and R"(r) = A(t) = (6r, 2, 0), we have
and
j k
R x R" = V x A = 312 2 t 1 —2i + 6/j — 6rk
61 2 0
R R V A 18t3+41 _ 18r3 + At
llR'll l|V|| “ 7(312)2 + (2t)2 + 1 _ 79r4 + 4r2 + 1
Note in Figure 10.37 that || A|| is the hypotenuse of a right triangle whose other two
sides have lengths ||A7-T|| = lAyl and ||AyvN|| = An. Therefore, by the Pythagorean
theorem, we have
An = y||A||2 - A2
Solution
The trajectory is a helix, and we find
dR d\
V = —- = (— sint, cost, 1) and A = — = (— cost, — sint, 0)
at dt
ds 1/— - -- ■ d*~ s
-t— = ||V|| = v sin21 + cos2 t + 1 = 72 and AT = —-z = 0
dt dt2
The tangential and normal components of acceleration are 0 and 1, respectively. This
means that the acceleration satisfies
That is, the acceleration vector is the principal unit normal N, and the acceleration is
always normal to the helical trajectory. ■
APPLICATIONS
Now that we know how to compute the tangential and normal components of acceler¬
ation, A7- and An, we will examine some applications. First, according to Newton’s
second law of motion, the total force acting on a moving object of mass m satisfies
F = znA, where A is the acceleration of the object. Because A = A7-T + A^N, we
have
where
676 Chapter 10 Vector-Valued Functions
For instance, experience leads us to expect a car to skid if it makes a sharp turn at
moderate speed or even a gradual turn at high speed. Mathematically, a “sharp turn"
occurs when the radius of curvature p — \/k is small (that is, when k is large), and
“high speed” means that ds/dt is large. In either case,
will be relatively large, and the car will stay on the road (its trajectory) only if (assum¬
ing no banking) there is a correspondingly large frictional force between the tires of
Figure 10.38 Tendency to skid the car and the surface of the road (see Figure 10.38).
When forces, masses, and accelerations are involved, it is customary to express the
mass m in slugs of an object whose weight W has been given in pounds.* From the
definition of weight,
W
m = —
8
where g is the acceleration of gravity (g ~ 32 ft/s2).
Solution
' ds
The force tending to push the car off the road is FN — iuk ^ — J , where m is the mass
of the car and k is the curvature of the road. This is the force that must be compensated
ds
for by friction if the car is not to skid. We know that — = 56 ft/s, and because the car
dt
W 2,700
weighs W = 2,700 lb, its mass is m = 84.38 slugs. Because the turn
8 ~32~
radius is 70 ft, we have k = so that
lb - s2 1 ft2
3,780 3,7801b ■
ft ft s2
There are certain important applications in which an object moves along its tra¬
jectory with constant speed ds/dt, and when this occurs, the acceleration A can have
only a normal component, because drs/dt2 = 0.
Proof Notice that we really do not need to know anything about components of ac¬
celeration to prove this result. Saying that the object has constant speed means that
||R'(f)|| is constant, and by Theorem 10.4, we conclude that R\t) is orthogonal to its
derivative R"(f) = A(t). But R'(t) points in the direction of the object’s motion along
its trajectory, which means that the acceleration A is orthogonal to the direction of
motion. □
*A slug is a unit of measurement defined as the unit of mass that receives an acceleration of 1 ft/s2 when a
1 lb lb s2
force of 1 lb is applied to it. That is, 1 slug = ^ = ——
10.5 Tangential and Normal Components of Acceleration 677
As an application of this result, note that when an object moves with constant
speed Do along a circular path of radius R (so that k = \/R), its acceleration is directed
toward the center of the path and has magnitude
. 2
An = «
R
(See Figure 10.39.)
Solution
Let m denote the satellite’s mass and v denote its speed. We will assume that the earth
is a sphere of radius 6,440 km, so the curvature of the path is k = 1 / R, where
Height
Use the formulas in Theorem 10.11 to find Ar and A N for the given show that
position vector R(?) in Problems 28-31.
„ f\x)f'\x) A \f"(x)\
28. R(r) = r3i + r2j + rk 29. R(r) = ri + 2tj + r2k A-j- — — ; An —
/TTTTW v7! + [ fix)]2
30. R(r) = (cos t)i + (sin r)j + k
31. R(r) = {e' cos f)i + (e* sin r)j + e'k 36. If the graph of the function / has an inflection point at x = a
and f"(a) exists, show that the graph of / has curvature 0 at
32. Modeling Problem An amusement park ride consists of a large
(a, f (a)).
(25-ft radius), flat, horizontal wheel. Customers board the
wheel while it is stationary and try to stay on as long as pos¬ 37. A projectile is fired from ground level with an angle of eleva¬
sible as it begins to rotate. The purpose of this problem is to tion a and muzzle speed u0- Find formulas for the tangential
discover how fast the wheel can rotate without losing its pas¬ and normal components of the projectile’s acceleration at time
sengers. t. What are Ar and AN at the time the projectile is at its max¬
imum height?
a. Suppose the wheel rotates at to revolutions per minute and
that a volunteer weighing W lb sits 15 feet from the center 38. An object connected to a string of length r is spun counter¬
of the wheel, as shown Figure 10.44. Find FN = mAN, clockwise in a circular path in a horizontal plane. Let co be the
where m — W/g is the volunteer’s mass. This is the force constant angular velocity of the object.
tending to push the volunteer off the wheel. a. Show that the position vector of the object is
R(r) = (r cos cot)i + (r sin cot)j.
b. Find the normal component of the object’s acceleration.
c. If the angular velocity co is doubled, what is the effect on
the normal component of acceleration? What is the effect
on A if co is unchanged but the length of the string is dou¬
bled? What happens to AN if co is doubled and r is halved?
An artificial satellite travels at constant speed in a stable cir¬
cular orbit R km above the earth’s surface (as in Example 4).
a. Show that the satellite’s period is given by the formula
-•
(e fi + rj + 3k) dt
24. Sketch the graph of R(t) — (3 cos f)i + (3 sin ?)j + ?k, and find •/_;
the length of this curve from t = 0 to t = 2n.
25. If F(r) = y-^i + ^ j + (cos Ok, find F(f) and F"(0- •r [(1 -t)i-rlj + e‘k\dt
64. Basketball player Kobe Bryant attempts a shot while standing 69. The path of a particle P is an Archimedean spiral described
20 ft from the basket, as shown in Figure 10.47. If Bryant in polar coordinates by r — lOr and 9 = 2nt, where r is ex¬
shoots from a height of 7 ft and the ball reaches a maximum pressed in inches and t is in seconds. Determine the velocity
height of 12 ft before passing through the 10-ft-high basket, of the particle (in terms of ur and ue) when
what is the initial speed of the ball? a. t = 0 b. t = 0.25 sec
70. Spy Problem After dropping the knockout gas into Scelerat’s
bunker (Problem 27, Section 10.3), the Spy lands and, as
expected, enters a bunker full of sleeping thugs—but no
Scelerat! He races through the back door of the bunker and
finds himself in a ski resort. By the time he locates Scelerat,
the French fiend is skiing down the mountain. The Spy grabs
a pair of skis and runs over to a conveniently located ski jump.
His keen mind quickly determines that the slope of the moun¬
tain is 17° and the angle at the lip of the jump is 10°, as shown
in Figure 10.49.
20 ft
a. What is the smallest value of v0 that will allow water from 76. differential equation: —— = —32j;
the hose to reach a window 11 m above the ground? dt2
dF
b. Suppose no = 50 m/s. At what angle should the hose be initial conditions: F = 50j and — = 5i + 5k at t = 0
dt
held so the water can be directed onto a fire on the roof of
77. Putnam Examination Problem A shell strikes an airplane flying at
the building (at a height of 15 m)?
a height h above the ground. It is known that the shell was fired
67. A DNA molecule has the shape of a double helix (see Figure
from a gun on the ground with muzzle speed t>0, but the po¬
10.4a). The radius of each helix is about 10-8 jum. Each helix
sition of the gun and its angle of elevation are both unknown.
rises about 3 x ICE8 jxm during each complete turn and there
Deduce that the gun is situated within a circle whose center
are about 3 x 10s complete turns. Estimate the length of each
lies directly below the airplane and whose radius is
helix.
68 . Show that the tangential component of acceleration of a mov¬
ing object is 0 if the object has constant speed. Is the converse -y/va-lgh
$
statement also true? That is, if Aj = 0, can we conclude that
the speed is constant? Neglect resistance of the atmosphere.
Chapter 10 Review 683
78. Putnam Examination Problem A coast artillery gun can fire at any again at a point Q, without coming to rest at any intermediate
angle of elevation between 0° and 90° in a fixed vertical plane. point. Prove that the acceleration vector of the particle does
If air resistance is neglected and the muzzle speed is constant not vanish at any point between P and Q, and that, at some
(v = vo), determine the set H of points in the plane that can point R between P and Q, the acceleration vector points in
be hit. Consider only those points above the horizontal. along the radius RO.
79. Putnam Examination Problem A particle moves on a circle with
center O, starting from rest at a point P and coming to rest
Guest Essay
T he mighty Antaeus was the giant son of Neptune (god of the sea) and Ge (goddess
of the earth), and his strength was invincible so long as he remained in contact
with his Mother Earth. Strangers who came to his country were forced to wrestle
to the death with him, and so it chanced one day that Hercules and Antaeus came to
grips with one another. But Hercules, aware of the source of Antaeus’ great strength,
lifted and held the giant from the earth and crushed him in the air.
There is a parable here for mathematicians. For just as Antaeus was bom of and
THE DIE IS CAST; I have written my
nurtured by his Mother Earth, history has shown us that all significant and lasting
book; it will be read either in the present
mathematics is born of and nurtured by the real world. As in the case of Antaeus, so
age or by posterity, it matters not which;
long as mathematics maintains its contact with the real world, it will remain powerful.
it may well await a reader, since God has
waited six thousand years for an interpreter
But should it be lifted too long from the solid ground of its birth into the filmy air
of his words. of pure abstraction, it runs the risk of weakening. It must of necessity return, at least
-Johannes Kepler occasionally, to the real world for renewed strength.
From James R. Newman, The World of Such a rejuvenation of mathematics occurred in the seventeenth century, following
Mathematics, Volume /, (New York: Simon discoveries made by two eminent mathematicians-scientists—Galileo Galilei (1564-
and Schuster, 1956), p. 220. 1642) and Johannes Kepler (1571-1630). Galileo, through a sequence of experiments
started before his 25th birthday, discovered a number of basic facts concerning the
motion of bodies in the earth’s gravitational field, and Kepler, by 1619, had deduced
all three of his famous laws of planetary motion. These achievements proved to be so
influential on the development of so much of subsequent mathematics that they must
be ranked as two of the GREAT MOMENTS IN MATHEMATICS. Galileo’s discoveries
led to the creation of the modern science of dynamics and Kepler’s to the creation of
modern celestial mechanics; and each of these studies, in turn, required, for their de¬
velopment, the creation of a new mathematical tool—the calculus—capable of dealing
with change, flux, and motion.
Galileo was born in Pisa in 1564 as the son of an impoverished Florentine noble¬
man. After a disinterested start as a medical student, Galileo obtained parental permis¬
sion to change his studies to science and mathematics, fields in which he possessed a
strong natural talent. While still a medical student at the University of Pisa, he made
his historically famous observation that the great pendulous lamp in the cathedral there
oscillated to and fro with a period independent of the size of the arc of oscillation.'
Later, he showed that the period of a pendulum is also independent of the weight
of the pendulum’s bob. When he was 25, he accepted an appointment as professor
of mathematics at the University of Pisa. It was during this appointment that he is
alleged to have performed experiments from the leaning tower of Pisa, showing that,
contrary to the teaching of Aristotle, heavy bodies do not fall faster than light ones. By
rolling balls down inclined planes, he arrived at the law that the distance a body falls
is proportional to the square of the time of falling, in accordance with the now familiar
formula 5 = \gt2.
Unpleasant local controversies caused Galileo to resign his chair at Pisa in 1591,
and the following year he accepted a professorship in mathematics at the University of
Padua, where there reigned an atmosphere more friendly to scientific pursuits. Here at
Padua, for nearly 18 years, Galileo continued his experiments and his teaching, achiev¬
ing a widespread fame. While at Padua, he heard of the discovery, in about 1607, of
the telescope by the Dutch lens-grinder Johann Lipersheim, and he set about making
instruments of his own, producing a telescope with a magnifying power of more than
30 diameters. With this telescope he observed sunspots (contradicting Aristotle’s as¬
sertion that the sun is without blemish), saw mountains on the moon, and noticed the
phases of Venus, Saturn’s rings, and the four bright satellites of Jupiter (all three of
these lending credence to the Copernican theory of the solar system). Galileo’s dis¬
coveries roused the opposition of the Church, and finally, in the year 1633, he was
summoned to appear before the Inquisition, and there forced to recant his scientific
findings. Not many years later the great scientist became blind. He died, a prisoner in
his own home, in 1642, the year Isaac Newton was born.
Johannes Kepler was born near Stuttgart, Germany in 1571 and commenced his
studies at the University of Tubingen with the intention of becoming a Lutheran min¬
ister. Like Galileo, he found his first choice of an occupation far less congenial than
his deep interest in science, particularly astronomy, and he accordingly changed his
plans. In 1594, when in his early twenties, he accepted a lectureship at the University
of Graz in Austria. Five years later he became assistant to the famous Danish-Swedish
astronomer Tycho Brahe, who had moved to Prague to serve as the court astronomer
to Kaiser Rudolph II. Shortly after, in 1601, Brahe suddenly died, and Kepler inherited
both his master’s position and his vast collection of very accurate data on the positions
of the planets as they moved about the sky. With amazing perseverance, Kepler set out
to find, from Brahe’s enormous mass of observational data, just how the planets move
in space.
It has often been remarked that almost any problem can be solved if one but con¬
tinuously worries over it and works at it a sufficiently long time. As Thomas Edison
said, genius is 1% inspiration and 99% perspiration. Perhaps nowhere in the history
of science is this more clearly demonstrated than in Kepler’s incredible pertinacity in
solving the problem of the motion of the planets about the sun. Thoroughly convinced
of the^Copernican theory that the planets revolve in orbits about the central sun, Ke¬
pler strenuously sought to determine the nature and position of those orbits and the
manner in which the planets travel in their orbits. With Brahe’s great set of observa¬
tional recordings at hand, the problem became this: to obtain a pattern of motion of
the planets that would exactly agree with Brahe’s observations. So dependable were
Brahe’s recordings that any solution that should differ from Brahe s observed positions
by even as little as a quarter of the moon’s apparent diameter must be discarded as in¬
correct. Kepler had, then, first to guess with his imagination some plausible solution
and then, with painful perseverance, to endure the mountain ot tedious calculations
needed to confirm or reject his guess. He made hundreds of fiuitless attempts and pei
formed reams and reams of calculations, laboring with undiminished zeal and patience
for many years. Finally he solved his problem, in the form of his three famous laws of
+This is only approximately true, the approximation being very close in the case of small amplitudes of
oscillation.
686 Chapter 10 Vector-Valued Functions
planetary motion, the first two around 1609 and the third one 10 years later in 1619:
I. The planets move about the sun in elliptical orbits with the sun at one focus.
II. The radius vector joining a planet to the sun sweeps over equal areas in equal
intervals of time.
III. The square of the time of one complete revolution of a planet about its orbit is
proportional to the cube of the orbit’s semimajor axis.
The empirical discovery of these laws from Brahe’s mass of data constitutes one
of the most remarkable inductions ever made in science. With justifiable pride, Kepler
prefaced his Harmony of the Worlds of 1619 with the following outburst:
I am writing a book for my contemporaries or—it does not matter—for posterity. It may be
that my book will wait for a hundred years for a reader. Has not God waited 6000 years
for an observer?
Kepler’s laws of planetary motion are landmarks in the history of astronomy and
mathematics, for in the effort to justify them, Isaac Newton was led to create modem
celestial mechanics. It is very interesting that 1800 years after the Greeks had de¬
veloped the properties of the conic sections there should occur such an illuminating
practical application of them. One never knows when a piece of pure mathematics may
receive an unexpected application.
In order to compute the areas involved in his second law, Kepler had to resort to a
crude form of the integral calculus, making him one of the precursors of that calculus.
Also, in his Stereometria doliorum vinorum (Solid Geometry of Wine Barrels, 1615),
he applied crude integration procedures to the finding of the volumes of 93 different
solids obtained by revolving arcs of conic sections about axes in their planes. Among
these solids were the torus and two that he called the apple and the lemon, these latter
being obtained by revolving a major and a minor arc, respectively, of a circle about
the arc’s chord as an axis. Kepler’s interest in these matters arose when he observed
some of the poor methods in use by the wine gaugers of the time. (See Group Research
Project, page 269.)
It was Kepler who introduced the word focus (Latin for “hearth”) into the geometry
of conic sections. He approximated the perimeter of an ellipse of semiaxes a and b by
the formula 7v(a + b). He also laid down a so-called principle of continuity, which
postulates the existence in a plane of certain ideal points and an ideal line having many
of the properties of ordinary points and lines, lying at infinity. Thus he explained
that a straight line can be considered as closed at infinity and that a parabola may be
regarded as the limiting case of either an ellipse or a hyperbola in which one of the foci
has retreated to infinity. The ideas were extended by later geometers.
Kepler was a confirmed Pythagorean, with the result that his work is often a blend
of the fancifully mystical and the carefully scientific. It is sad that his personal life was
made almost unendurable by a multiplicity of worldly misfortunes. An infection from
smallpox when he was four years old left his eyesight much impaired. In addition to
his general lifelong weakness, he spent a joyless youth; his marriage was a constant
source of unhappiness; his favorite child died of smallpox; his wife went mad and
died; he was expelled from his lectureship at the University of Graz when the city
fell to the Catholics; his mother was charged and imprisoned for witchcraft, and for
almost a year he desperately tried to save her from the torture chamber; he himself
very narrowly escaped condemnation for heterodoxy; and his stipend was always in
arrears. One report says that his second marriage was even less fortunate than his
first, although he took the precaution to analyze carefully the merits and demerits of
11 women before choosing the wrong one. He was forced to augment his income by
casting horoscopes, and he died of a fever in 1630 at the age of 59 while on a journey
to try to collect some of his long overdue salary.
Chapter 10 Mathematical Essays 687
lim fix) = L
X-+C
means that for each e > 0, there corresponds a number S > 0 with the property that
The notion of a limit is fundamental for many of the ideas of calculus, primarily
in the definitions of both the derivative and integral.
Definition of integral If / is defined on the closed interval [a, b], we say / is inte¬
grate on [a, b] if
n
1 = lim fixt)Axk
exists, where ||P|| = max A Xk, k = 1,2,... ,n. This limit is called the definite inte¬
gral of / from a to b. The definite integral is denoted by
1- fJ Cl
f(x)dx
Table 10.3 summarizes this relationship between modeling and the Riemann sum.
In the last two chapters we considered vectors in R2 and R3, as well as vector¬
valued functions. We summarize these ideas in Table 10.4.
Chapters 1-10 Cumulative Review 689
Area Find the area under a curve Find the sum of n rectangles. Area = / f(x)dx
5.2 Ja
defined by y = f(x), where f{x) > 0,
J^fix^Ax,
above the x-axis and bounded by the
lines x — a and x = b. Area of Ath rectangle
Area between curves Find the area £[/(***)-*(*;)] Axk Area [/(*) -g(x)]dx
6.1 -f
between the curves y = f(x) and k=i'-v '
n
6.5 Mass Find the mass of a thin plate y^plfixf) - g(x*k)]Axk m = p [fix) - gix)]dx
with constant density p, bounded by k— 1 Ja
y = fix) and y = g(x).
6.5 First moment Find the moment of a First moment about the y-axis Mv = p x[f (x) — g(x)] dx
n Ja
thin plate with constant density p
bounded by y — f(x) and y = g(x) Pxklf(xk) - g(x*k)\Axk
k=2
(about the y-axis).
First moment about the x-axis = {Uix)?~[gix)]2} dx
n
-# EvU/^)]2-ku;)]2} ax,
*=iz
Centroid Find the centroid of a (x,y) = (—, —^
\ m m /
homogeneous lamina, (p constant.)
6.5 Work Find the work done when an J2F(x*k)Axk Work = / Fix) dx
object moves along the x-axis from a k= 1 Ja
d2 R dX
Acceleration vector
M,) ~ ~ 71 ~ AtT + AnN
where
d2s R R'
At — is the tangential component, and
dt2 IR
ds |R x R
A\i — K = IIA||2 — A\ is the normal component.
~dt IR
dT |R x R !/"(*)!
Curvature K = if y — f (x) then k =
113 3/2
ds IR '
[1 +t fix)]2]
dR
Torsion r —
~d7
Chapters 1-10 Cumulative Review 691
2. WHAT DOES THIS SAY? In your own words, outline a pro¬ 27. Find the volume of the solid formed by revolving about the
cedure for deciding whether a given series converges or di¬ x-axis the region bounded by the curve
verges. 2
3. WHAT DOES THIS SAY? What is a vector? What is a vector -V “ V3x -2
function? In your own words, discuss what is meant by vector
calculus. 28. Find the equation of the line through (-1,2, 5) and perpen¬
dicular to the plane 2x - 3y + z = 11 in symmetric form.
Evaluate the limits in Problems 4-9, or explain why the limit does
29. Find the equation of the plane satisfying the given conditions.
not exist.
a. passing through (5, 1, 2), (3, 1, —2), and (3, 2, 5)
4. lim — ---- -
lx3 — 5x — 2
x->\ 3x2 + 5x — 8
b. passing through (-2, -1,4) and perpendicular to the line
x y + 1 _ 3 — z
(2x2 -3x + 1)(1 -5x)
5. lim 2 = 5 ~ 2
k^+oo 5x3 + 4x — 9
3a;
Test the series in Problems 30-35for convergence.
sin 3x 0° ~
6. lim 7. lim 1 - - ,3 1
cosx — 1 k=0 k= 1
/2x + 3
8.
i. lim (:
I-esc x 9. lim 00 3k2 — Ik + 2 00 it!
x->0+ \ x / *->-o ex — x — 1
33. £
32. £
% (2k — l)(k + 3) 2k -k
Find the derivatives in Problems 10-15. k=1
41. Find T and N for R(f) = 2(sin 2t)\ + (2 + 2 cos 2r)j + 6rk.
42. Show that the alternating series
OO / 1 \t+l
s = E( Vk
k=1
23. Let f be a continuous differentiable function with f(0) = 0 approximates the entire sum S with four-decimal-place accu¬
and -2 < /'(x) < 4 for -3 < x < 3. racy.
a. Can /(2) be negative? Give a reason for your answer. 43. a. Find the Maclaurin series representation for
b. Can /(—2) be negative? Give a reason for your answer. fix) = x2e~x\
c. Must / have a critical point between —3 and 3 ? b. Use the representation in a to approximate
24. Find the area bounded by the curve y — xVx2 + 8 and the
x2e x‘ dx
x-axis on [—2, 1].
f
Jo
25. Find the area bounded by the curves y - sin x and y = cos x
with three-decimal-place accuracy.
and the lines x = 0 and x = 1.
692 Chapters 1-10 Cumulative Review
44. Let P{x) = 7 - 3(x - 4) + 5(x - 4)2 - 2(x - 4)3 + 6(x - 4)4 e. What is the distance traveled by the particle from t = 1.25
be the fourth-degree Taylor polynomial for the function / to t — 1.75.
about 4.
47. Find an equation in terms of x and y for the line tangent to the
a. Find /(4) and /"'(4). curve given by
b. Use the third-degree Taylor polynomial for f about x = 4
to approximate /'(4.2). x = r2-2r-l, y = tA —4t2 + 2
* subject to the condition y — 2 when x = 0. b. Find a Cartesian equation for the particle’s trajectory.
Chapter 11 Review
Group Research Project: Desertification
Mount St. Helens map courtesy of Bill Lennox, Humboldt State University.
693
694 Chapter 11 Partial Differentiation
BASIC CONCEPTS
Physical quantities often depend on two or more variables (see the Perspective at the
beginning of this chapter). For example, we might be concerned with the temperature,
T, at various points (x, y) on a metal plate. In this situation, T might be regarded as a
function of the two location variables, x and y. Extending the notation for a function
of a single variable, we can denote this relationship by T(x, y).
Function of Two Variables A function of two variables is a rule / that assigns to each ordered pair (x, y) in
a set D a unique number f(x, y). The set D is called the domain of the function,
and the corresponding values of /(x, y) constitute the range of /.
Functions of three or more variables can be defined in a similar fashion. For ex¬
ample, the temperature in our introductory example may vary not only on the plate,
but also with time t, in which case it would be denoted by T(x, y, t). Occasionally,
we will examine functions of four or more variables, but for simplicity, we will focus
most of our attention on functions of two or three variables.
When dealing with a function / of two variables, we may write z = /(x, y) and
refer to x and y as the independent variables and to z as the dependent variable.
Often, the functional “rule” will be given as a formula, and unless otherwise stated, we
will assume that the domain is the largest set of points in the plane (or in M?)for which
the functional formula is defined and real-valued. These definitions and conventions
are illustrated in Example 1 for a function of two variables.
Solution
a. /(2, l) = v/9 — 22— 4(1)2 = 1
Functions of several variables can be combined in much the same way as functions
of a single variable, as noted in the following definition.
Operations with Functions of If /(x, y) and g(x, v) are functions of two variables with domain D, then
Two Variables
Sum if + 8)(x,y) = f(x, y) + g(x, y)
Cxmyn
3x5y3 — lx2y + 2x — 3y + 11
By analogy with the single-variable case, we define the graph of the function f (x, y)
to be the collection of all 3-tuples (ordered triples) (x, y, z) such that (x, y) is in the
domain of / and z = /(x, y). The graph of /(x, y) is a surface in K3 whose projection
onto the xy-plane is the domain D.
It is usually not easy to sketch the graph of a function of two variables without the
assistance of technology. One way to proceed is illustrated in Figure 11.2.*
z*
Z«ll
z=8
Z=5
Notice that when the plane z = C intersects the surface z = fix, y), the result is the
curve with the equation fix, y) — C. Such an intersection is called the trace of the
graph of / in the plane z = C. The set of points (x, y) in the xy-plane that satisfy
f (x, y) = C is called the level curve of / at C, and an entire family of level curves is
generated as C varies over the range of /. We can think of a trace as a “slice” of the
surface at a particular location and a level curve as its projection onto the xy-plane. By
sketching members of this family on the xy-plane, we obtain a useful topographical
map of the surface z = fix, y). Because level curves are used to show the shape of a
surface (a mountain, for example), they are sometimes called contour curves.
For instance, imagine that the surface z = fix, y) is a “mountain” and that we
wish to draw a two-dimensional “profile” of its shape. To draw such a profile, we
indicate the paths of constant elevation by sketching the family of level curves in the
plane and pinning a “flag” to each curve to show the elevation to which it corresponds.
*See the Technology Manual accompanying this book for some suggestions about using technology to help
you graph functions of two variables.
696 Chapter 11 Partial Differentiation
as shown in Figure 11.3b. Notice that regions in the map where paths are crowded
together correspond to the steeper portions of the mountain. An actual topographical
map of Mount Rainier is shown in Figure 11.3c.
z = 1500
z = 100
a. The surface z = /(*, y) as a mountain
You probably have seen level curves on the weather report in the newspaper or
television news shows, where level curves of equal temperature are called isotherms
(see Figure 11.4). Other common uses of level curves involve curves of equal pressure
(called isobars) and curves of electric potential (called equipotential lines).
a. Computer generated graph of surface b. Simplified graph showing traces c. Level curves
The level curves of a function /(x, y) provide information about the cross sections of
the surface z — /(x, y) perpendicular to the z-axis. However, a more complete picture
of the surface can often be obtained by examining cross sections in other directions as
well. This procedure is used to graph a function in Example 3.
Solution
The level curve x2 + y2 = 0 (that is, C = 0) is the point (0, 0), and for C > 0, the
level curve x2 + y2 - C is the circle with center (0, 0) and radius y/C (Figure 11.6a).
There are no points (jc, y) that satisfy x2 + y2 — C for C < 0.
We can gain additional information about the appearance of the surface by exam¬
ining cross sections perpendicular to the other two principal directions; that is, to the
x-axis and the y-axis. Cross-sectional planes perpendicular to the x-axis have the form
x = A and intersect the surface z = x2 + y2 in parabolas of the form z = A2 + y2-
For example, the cross section of the plane x = 5 intersects the surface in the parabola
z = 25 + y2. That is, it is the set of point (5, y, 25 + y2) as y varies. Similarly,
cross-sectional planes perpendicular to the y-axis have the form y = B and intersect
the surface in parabolas of the form z = x2 + B2.
To summarize, the surface z = x2 + y2 has cross sections that are circles in planes
perpendicular to the z-axis and parabolas in the other two principal directions. Since
the surface is formed by revolving a parabola about its axis, it is called a circular
paraboloid or a paraboloid of revolution. The graph of the surface is shown in Figure
11,6b.
698 Chapter 11 Partial Differentiation
The concept of level curve can be generalized to apply to functions of more than
two variables. In particular, if / is a function of 3 variables x, y, and z, then the graph
of the equation /(x, y, z) = C is a surface of R3 called a level surface of / at C.
Solution
The isothermal surfaces are given by T(x,y,z) = k for constant k\ that is,
x2 + y2 + z2 — 100 — k. If 100 — k > 0, the graph of x2 + y2 -f z2 = 100 — k
is a sphere of radius V100 - k and center (0, 0, 0). When k = 100, the graph is a
single point (the origin), and T{0, 0, 0) = 100°C. As the temperature drops, the con¬
stant k gets smaller, and the radius V100 - k of the sphere gets larger. Hence, the
isothermal surfaces are spheres, and the larger the radius, the cooler the surface. This
situation is illustrated in Figure 11.7.
r= VlOO- Cj
As you can see by the examples in this section, you will need to recall the graphs
of quadric surfaces (see Section 9.7). It will also help if you can recognize the surface
by looking at its equation. What you will need to remember is summarized in Table
9.1, page 625.
11.1 Functions of Severol Variables 699
TECHNOLOGY NOTE
Computer software is now available for graphing functions of two variables. Software packages such as Mathematical Maple,
Matlab, and Derive will sketch very sophisticated graphs in three dimensions. Software usually allows a given graph to be
visualized from different viewpoints by rotating the surface until a desired viewpoint is found. A few examples of computer
generated surfaces are shown in Figure 11.8.
a, /(0,0,0) b. /(I, —1,1) c. /(-1,1, —1) Sketch some level curves f ix, y) — C for C > 0 for each function
given in Problems 15—20.
d. -^-fix,x,x) e. — /(l,y, 1) f- —/(1.1.Z2) 16. fix, y) =x2 -y2
dx dy dz 15. fix, y) — 2x — 3 y
3. WHAT DOES THIS SAY? Discuss what is meant by a func¬ 17. fix, y)=x2-y 18. gix, y) — x2 — y
tion of two variables. Your discussion should include exam¬ 2 y2
19. hix, y) = x- + — 20. fix, y) = -
ples of a function of two variables not discussed in this section.
y
4. WHAT DOES THIS SAY? Describe how level curves can be In Problems 21-26, sketch the level surface fix, y, z) = C for the
used to sketch the graph of a function fix, y). Include some given value of C.
examples in your discussion (different from those in the text).
21. /ix, y, z) = y2 + z2 for C = 1
Find the domain and range for each function given in Problems 22. fix, y, z) = x2 + z2 for C = 1
5-14.
1
23. fix, y,z)—x + y — z for C = 1
5. fix, y) = Jx -y 6. fix,y) = _— 24. fix, y, z) = x d- y — Z for C = 0
25. fix, y, z) = ix + l)2 + (y — 2)2 + (z - 3)2 for C = 4
7. fiu,v) — y/Tiv 8. fix,y) = J^ 26. fix, y, z) = 2x2 + 2y2 - z for C = 1
700 Chapter 11 Partial Differentiation
x2 y2 x
29. — + — :2 = 1 30. y2-z2=l
4 9 ~9
Eix, y) -
V3 + x2 + 2y2
Sketch the equipotential curves that correspond to E = 1,
E = 2, and E = 3.
54. According to the ideal gas law, PV = kT, where P is pres¬
sure, V is volume, T is temperature, and A: is a constant. Sup¬
pose a tank contains 3,500 in.3 of a particular gas at a pressure
of 24 lb/in.2 when the temperature is 270° K (degrees Kelvin).
_L 1 _ 1
d0 di L
56. The EZGRO agricultural company estimates that when lOO.v salespeople will sell approximately
worker-hours of labor are employed on y acres of land, the
i
57. Suppose that when x machines and y worker-hours are used 61. Modeling Problem A manufacturer with exclusive rights to a
each day, a certain factory will produce Q(x, y) — lOxy mo¬ sophisticated new industrial machine is planning to sell a lim¬
bile phones. Describe the relationship between the “inputs” ited number of the machines to both foreign and domestic
x and y that result in an “output” of 1,000 phones each day. firms. The price that the manufacturer can expect to receive
Note: You are finding a level curve of the production func¬ for the machines will depend on the number of machines made
tion Q. available. It is estimated that if the manufacturer supplies x
machines to the domestic market and y machines to the for¬
58. Modeling Problem At a certain factory, the daily output is mod¬ eign market, the machines will sell for
eled by Q = CKrLl~r units, where K denotes capital in¬
vestment, L is the size of the labor force, and C and r are 2^
constants, with 0 < r < 1 (this is called a Cobb-Douglas pro¬ 20
duction function). What happens to Q if K and L are both thousand dollars apiece at home and
doubled? What if both are tripled?
50 - — + —
59. Sketch the level curves for the graph of f(x, y) = xy in the 10 20
first quadrant.
thousand dollars apiece abroad. Express the revenue R as a
60. A publishing house has found that in a certain city each of its function of x and y.
Most commonly considered functions of a single variable have domains that can be
described in terms of intervals. However, dealing with functions of two or more vari¬
ables requires special terminology and notation that we will introduce in this section
and then use to discuss limits and continuity for functions of two variables.
An open disk centered at the point C(a, b) in R2 is the set of all points (x, y) such that
a. Open disk
for r > 0 (see Figure 11.10a). If the boundary of the disk is included (that is, if
y/(x — a)2 + (y — b)2 < r), the disk is said to be closed (see Figure 11.10b). Open
and closed disks are analogous to open and closed intervals on a coordinate line.
A point PoUo* yo) is said to be an interior point of a set S in R2 if some open disk
centered at P0 is contained entirely within 5, as shown in Figure 11.11a. If S is the
empty set, or if every point of S is an interior point, then S is called an open set (Figure
11.11b). The point P0 is called a boundary point of S if every open disk centered at
PQ contains both points that belong to 5 and points that do not. The collection of all
b. Closed disk
boundary points of S is called the boundary of S, and S is said to be closed if it
Figure 11.10 Open and closed disks contains its boundary (Figure 11.11c). The empty set and R2 are both open and closed.
702 Chapter 11 Partial Differentiation
Boundary
Similarly, an open ball centered at C(a, b, c) is the set of all points P(x, y, z)
such that
for r > 0. A point Po is an interior point of a set S in R3 if there exists an open ball
centered at Po that is contained entirely within S, and the nonempty set S is an open
set if all its points are interior points. The point Po is a boundary point of S if every
open ball centered at Po contains both points that belong to S and points that do not,
and S is closed if it contains all its boundary points. As with R2, the set R3 is both
open and closed.
In Chapter 2, we informally defined the limit statement lim f(x) = L to mean that
X—>-C
f(x) can be made arbitrarily close to L by choosing x sufficiently close (but not equal)
to c. For a function of two variables we write
lim f(x, y) = L
(x,y)-*(x0,y0)
to mean that the functional values f(x,y) can be made arbitrarily close to the number
L by choosing the point (*, y) sufficiently close to the point (xo, yo). The following is
a more formal definition of this limiting behavior.
means that for each given number € > 0, there exists a number 8 > 0 so that
whenever (x, y) is a point in the domain D of / such that
^ What This Says If lim fix, y) = L, then given e > 0, the func-
When considering lim f(x) we need to examine the approach of x to c from two
X —>C
directions (the left- and right-hand limits). However, for a function of two variables, we
write (x, y) -> (xo, Vo) to mean that the point (x, y) is allowed to approach (xo, yo)
along any curve in the domain of / that passes through (xo, yo)- For example, no
matter how you get to the 15th step of Denver’s state capitol, you will still be at 5,280
ft when you get there. If the
lim f(x, y)
(*.y)-K-*o>:yo)
is not the same for all approaches, or paths within the domain of /, then the limit does
not exist.
Solution
First, note that for x ^ y.
+ x — xy — y (x + l)(x — y)
fix, y) = =x+ l
x — y x -y
Therefore, since fix, y) is defined only for x / y, we have
x2 + x — xy — y
lim = lim (x -F 1)
(jc,y)->(0.0) x - y (*,y)-*(0,0)
Solution
First note that the denominator is zero at (0, 0), so /(0, 0) is not defined. If we ap¬
proach the origin along the x-axis (where y = 0), we find that
2x(0)
/(*. 0) = 0
x2 + 0
so /(x, y) —> 0 as (x, y) —> (0, 0) along y = 0 (and x ^ 0). We find a similar result
if we approach the origin along the y-axis (where x = 0); see Figure 11.13.
However, along the line y = x, the functional values are
line y = x
2x2
Figure 11.13 Graph of f(x, y) = f(x,x) = = 1 for x / 0
X2 + X2
2xy
z = ———r and limits as
x2 + y2
(x, y) -> (0, 0) so that /(x, y) -> 1 as (x, y) —>• (0, 0) along y = x. Because /(x, y) tends toward
different numbers as (x, y) —»■ (0, 0) along different curves, it follows that / has no
limit at the origin. H
(*,?)->(0,0) x4 + y2
Solution
If we approach this problem as we did the one in Example 2, we would take limits as
(x, y) —>■ (0, 0) along the x-axis, the y-axis, and the line y = x. All these approach
lines have the general form y = mx. We find
along y = mx:
2
* y x2(mx)
hm —-- lim
(jc,;y)->(0.0) X4 + y2 (x.y)-*>(0,0) x4 + (mx)2
mx
lim
(jc,y)-»-(0,0) x2 + m2
WARNING , . , ... . 2
It is otten possible to snow x y x2(x2)
lim lim
that a limit does not exist by the (x,y)->(0,0) x4 + y2 0c,y)—>(0,0) X4 + (x2)2
methods illustrated in Examples 2 and
x4 _ 1
3. However, it is impossible to try to lim
prove that lim /(x, y) exists by (x.y)—>(0,0) 2X4 “ 2
Or, v)—>(0,0)
showing the limiting value of /(x, y)
is the same along every curve that Therefore, since approaching (0, 0) along the lines y = mx gives a different limiting
passes through (xo, yo) because there value from approaching along the parabola y = x2, we conclude that the given limit
are too many such curves. does not exist. ■
11.2 Limits and Continuity 705
We have just observed that it is difficult to show that a given limit exists. However,
limits of functions of two variables that are known to exist can be manipulated in much
the same way as limits involving functions of a single variable. Here is a list of the
basic rules for manipulating such limits.
Basic Formulas and Rules for Suppose lim /(x, y) = L and lim g(x, y) = M. Then
Limits of a Function of Two (*,}>)-> U0..yo) (*,>>)->■ Uo.yo)
Variables Scalar multiple rule lim [af](x, y) — aL for constant a
U,y)^-(x0,yo)
7
Quotient rule lim (x,y) = — ifM^O
ix,y)-*-(xo,yo) 8J M
Solution
a. lim (x2 + xy + y2) = (3)2 + (3)(—4) + (—4)2 = 13. A graph is shown in
(*,y)-+(3,-4)
Figure 11.14a.
_ lim 2xy
. .. 2 xy (*.y)-+-(1.2) '
b. lim —-- =-^-x-
(x,y)->(i,2) X2 + y2 lim (x + y )
C*.y)—>-(1,2)
Figure 11.14 Graphs showing limits The graph is shown in Figure 11.14b.
CONTINUITY
Using the definition of the limit of a function of two variables, we can now define the
continuity of a function of two variables analogously.
706 Chapter 11 Partial Differentiation
Continuity of a Function of The function /(x, y) is continuous at the point (xo, To) if
Two Variables
1. /(xo, yo) is defined;
2. lim /(x, y) exists;
(x:y)^(x o,y0)
+What This Says The function / is continuous at (xo, jo) if the functional
value of /(x, y) is close to /(xo, yo) whenever (x, y) in the domain of / is
* sufficiently close to (xo, Jo)- Geometrically, this means that / is continuous if
the surface z = /(x, y) has no “gaps” or “holes” (see Figure 11.15).
a.f{x,y) = b.f(x,y)=
x2 + v2 y-
is discontinuous at (0,0); a hole is discontinuous on the parabola y =x2\ a gap
The basic properties of limits can be used to show that if / and g are both con¬
tinuous on the set S, then so are the sum f + g, the multiple af, the product fg, the
quotient f/g (whenever g ^ 0), and the root iff wherever it is defined. Also, if F
is a function of two variables that is continuous at (xo, yo) and G is a function of one
variable that is continuous at F(xo, yo)> it can be shown that the composite function
G o F is continuous at (xo, yo)-
Many common functions of two variables are continuous wherever they are de¬
fined. For instance, a polynomial in two variables, such as x3y2 + 3xy3 — lx + 2, is
continuous throughout the plane, and a rational function of two variables is continuous
wherever the denominator polynomial is not zero. Unless otherwise stated, the func¬
tions of two or more variables considered in this text will be continuous wherever they
are defined.
Solution
a. The function / is a rational function of x and y (since x — y and x2 + y2 are both
polynomials), so it is discontinuous only where it is undefined; namely at (0, 0).
b. Again, this is a rational function and is discontinuous only where it is undefined;
that is, where
y — x2 — 0
Thus, the function is continuous at all points except those lying on the parabola
y = x2. U
. 1
y sin — x ^ 0
/(*, y) x
0 x — 0
Solution
Front view The graph is shown in Figure 11.16. To prove continuity at (0, 0), we must show that
for any € > 0, there exists a 8 > 0 such that
(We use x2 + y2 and S2 here instead of jx2 + y2 and 8 for convenience.) Note that
y sin - < ly | for all jc 7^ 0, because sin — < 1 for x ^ 0. If (x, y) lies in the disk
Side view x x
x2 + y2 < 82, then the points (0, y) that satisfy y2 < 82 lie in the same disk (let x = 0
Figure 11.16 Graph of / in Example 6 in x2 + y2 < s2). In other words, points satisfying |y| < <5 lie in the disk, and if we let
8 — € it follows that
The concepts we have just introduced for functions of two variables in R2 extend nat¬
urally to functions of three variables in R3. In particular, the limit statement
lim f(x,y,z) = L
(x,y,z)^-(x0.yO'Zo)
means that for each number e > 0, there exists a number <5 > 0 such that
Solution
The function f(x,y,z) is continuous except where it is not defined; that is, for
x2 + y2 — 2z < 0. Thus, f(x, y, z) is continuous at any point not inside or on the
paraboloid z = \(x2 + y2). ■
3. lim (xy2+x3y + 5)
sin(x + y)
15. lim 16. lim (sin x +cosy)
(x,y)-+(0,0) x+y (jc.y)->(0,0)
X*4 — 4
x 4 — y4 sin(x2 + y2)
17. lim 18. lim 27. lim (sin x —cosy) 28. lim
(,t,y)->(5,5) X2 — y2 (x,y)^-(a,a) X2 — y2 ’ (*,y)->(0.0) (*,y)-K0,0) x2 + y2
a is a constant.
x2 - 4y2 x2 + y2
19. lim 20. lim
U,>)->(2.i) x — 2y (,t,.v)—>(0,0) _(_ _y2 _|_ 4 _ 2
x3 + y3
for (x,y) # (0.0)
f(x,y) = x2 + y2
A for (x, y) = (0, 0)
3x3 — 3y3
for x2 ^ y2
fix, y) = x2 — y2
B otherwise
sin(x2 + y2)
42. Let fix, y) — for (x, y) (0, 0). For what
x2 + y2
value of /(0, 0) is fix, y) continuous at (0,0)? Hint: Use
polar coordinates.
43. Consider the function
2x2 — x2y2 4- 2y
for (x, y) 7^ (0, 0)
fix, y) x2 + v2
for ix,y) — (0, 0)
49. If fix, y) is continuous for all x ^0 and y 7^ 0, and 57. Prove the sum rule:
f (0, 0) = 0, then lim fix, y) — 0.
(x,y)-+(0,0)
50. If f(x) and g(y) are continuous functions of x and y, respec¬ lim [f + g](x,y) = L +M
(*.y)-n*o.yo)
tively, then
x2 — V2 x2 — v2
53. lim -— — 0 54. lim -— = 2 -2 2 for (x,y) ^ (0, 0)
(X.JO-KO.O) X + y (*,y)->(l,-l) x + y fix, y) x2 + y2
55. Prove that if / is continuous and f(a,b) > 0, then there ex¬
0 for (x, y) = (0, 0)
ists a 5-neighborhood about (a, b) such that /(x, y) > 0 for
every point (x, y) in the neighborhood.
56. Prove the scalar multiple rule: Let g(x) = f(x, 0) and h(y) = f (0, y). Show that both g(x)
and h(y) are continuous at 0 but that fix, y) is not continuous
lim [affx,y)—a lim fix, y) at (0, 0).
(jr.y)—j-(jcq.Xo) ' ix,y)^y(x0,yo)
PARTIAL DIFFERENTIATION
It is often important to know how a function of two variables changes with respect to
one of the variables. For example, according to the ideal gas law, the pressure of a gas
kT
is related to its temperature and volume by the formula P = —, where A: is a constant.
If the temperature is kept constant while the volume is allowed to vary, we might want
to know the effect on the rate of change of pressure. Similarly, if the volume is kept
constant while the temperature is allowed to vary, we might want to know the effect on
the rate of change of pressure.
The process of differentiating a function of several variables with respect to one of
its variables while keeping the other variable(s) fixed is called partial differentiation,
and the resulting derivative is a partial derivative of the function.
Recall that the derivative of a function of a single variable / is defined to be the
limit of a difference quotient, namely.
Partial Derivatives of a If z = /(x, y), then the partial derivatives of / with respect to x and y are the
Function of Two Variables functions fx and fy, respectively, defined by
What This Says For the partial differentiation of a function of two vari¬
ables, z = /(x, y), we find the partial derivative with respect to x by regarding
y as constant while differentiating the function with respect to x. Similarly, the
partial derivative with respect to y is found by regarding x as constant while
differentiating with respect to y.
Solution
a. For fx, hold y constant and find the derivative with respect to a:
b. For fy, hold a constant and find the derivative with respect to y:
fyix, y) = a3 + 2x2y ■
Several different symbols are used to denote partial derivatives, as indicated in the
following box.
3/ 3/
= fx(a,b) and = fy(a,b)
3a (a.b) 3y (a.b)
dz
a. Evaluate — . b. Evaluate at (1, 1).
3* (tt/3,0)
712 Chapter 11 Partial Differentiation
Solution
Thus,
TECHNOLOGY NOTE
Finding partial derivatives using technology is a natural extension of the way you have been finding other derivatives. The
general format for most calculators and computer programs is the same: derivative operator, function, variable of differentia¬
tion. Evaluating the partial derivative is then accomplished by using the evaluate feature. For example, Figure 11.17a displays
the computation of the partial derivative of f(x, y) — x3y + x2y2 with respect to x from Example 1, while Figure 11.17b
displays the evaluation of the partial derivative of z — x2 sin(3x + y3) with respect to x at the point (j, 0) found in Example 2.
■-^(x3y+ x2y2) 3 • x2 • y + 2 • x ■ y2
c7CxA3y+xA2y^2J.x>
MAIN_FiAD AUTO_FUNC 1/30
a. The partial derivative +x2y2) b. The partial derivative sin (3* + y3)) evaluated at (y, 0)
Solution
a. For fx, think of / as a function of x alone with y and z treated as constants:
fx (x, y, z) = 2x + 2y2
WARNING , x . b. fy(x, y, z) =4xy + z3
In f (x, y, z), Z is an
independent variable. c. fz(x, y, z) = 3yz2 ■
x2z 4- yz3 = x
Solution
Differentiate implicitly with respect to x, treating y as a constant:
2dz . 2 dz
2xi
2 xz + x2— + 3 yz2 — = 1
ox dx
dz
Then solve this equation for
dx
dz 1 — 2 xz
dx x2 + 3 yz2
Similarly, holding x constant and differentiating implicitly with respect to y, we find
*2^+z3 + 3^fi=0
dy dy
so that
dz —z
dy x2 + 3yz2
line
line Curve C
z=f(x,yo)
Curve D
z=f(x0,y)
Surface Surface
l =/(*,y) z=f(x,y)
Plane
Plane
y = yo
x = x0
Solution
If f(x, y) = x^/x + y = x(x + y)1,/2, then the required slope is fx{ 1, 3).
Thus, ml 3) = —_ + yrr3 = -. ■
2\/1 +3 4
The derivative of a function of one variable can be interpreted as a rate of change, and
the analogous interpretation of partial derivative may be described as follows.
Solution
Since I = ER~l, we have
3/ 3/
— = R and = -ER
3E 3~R
and thus, when E = 120 and R = 15, we find that
3/ _ . 3/
15 0.0667 and = — (120) (15) -2 -0.5333
3E 3R
This means that if the resistance is fixed at 15 ohms, the current is increasing (because
the derivative is positive) with respect to voltage at the rate of 0.0667 ampere per
volt when the EMF is 120 volts. Likewise, with the same fixed EMF, the current is
decreasing (because the derivative is negative) with respect to resistance at the rate of
0.5333 ampere per ohm when the resistance is 15 ohms. ■
9’/ JL(Bf
dy2 ^ hr; = (/>)> = f,
'yy
9y \9y
92f 9 (df
= (fy)x = fyx
dxd y dx ,9y.
92f 3 'df
= (fx)y = fxy
dydx dy dx
The notation fxy means that we differentiate first with respect to x and then with
respect to y, while -—— means just the opposite (differentiate with respect to y first and then
dxdy
with respect to x).
Solution
a. First differentiate with respect to y; then differentiate with respect to x.
dz 9
-~ = -2x + 9y2
9y
92z d dz
■(-2x + 9 y2) = -2
3x3y 3x
dx \3ydy,J dx
b. Differentiate first with respect to x and then with respect to y:
df
lOx — 2 y
3x
d2f _ 3 (df
= - 2y) = -2
dydxdy \dx dy
c. Differentiate with respect to x twice:
d2z 3 ^ dz ^
= —(10x-2y) = 10
3x2 3x
dx \3x^ 3x
d. Evaluate the mixed partial found in part b at the point (3.2):
fxy 0,2) = -2
92z 9 2z
Notice from parts a and b of Example 7 that . This equality of
dxdy dydx
mixed partials does not hold for all functions, but for most functions we will encounter,
it will be true. The following theorem provides sufficient conditions for this equality
to occur.
716 Chapter 11 Partial Differentiation
Solution
We have the partial derivatives
The mixed partial derivatives (which must be the same by the previous theorem) are
3T _ 2d2T
31 dx2
where T(x,t) is the temperature in a thin rod at position * and time t. The constant c
is called the diffusivity of the material in the rod. In the following example, we verify
that a certain function satisfies this heat equation.
Solution
3T -t ^
-e cos -
~dt
and
3 2T
— I —e sin -
dx2 3r \ c c
= —-e cos
3T 3 2T
Thus, T satisfies the heat equation
~dt dx2
Analogous definitions can be made for functions of more than two variables. For
example,
fzzz —
33/
3 z2
3
3z
ra Uz/J
dz
or
3V
dzdydx
3 r9 (df) 1
dz _3y
11.3 Partial Derivatives 717
Solution
First, compute the partials:
fx(x, y, z) = yz + 2xy3z4
Determine fx, fy, fxx> and fyx in Problems 3-8. 24. 3x2 + 4y2 + 2z2 = 5
3. f(x, y) — x3 + x2y + xy2 + y3
4. /(x, y) = (x+xy + y)3 25. 3x2y + y3z — z2 = 1 26. x3 — xy2 + yz2 — z3 = 4
Determine fx and fy in Problems 9-16. In Problems 29—32, compute the slope of the tangent line to the
graph of f at the given point Pq in the direction parallel to
9. a. /(x, y) = (sinx2) cosy b. /(x, y) = sin(x2 cosy)
a. the xz-plane b. the yz-plane
10. a. /(x, y) = (sin y/x) In y2 b. /(x, y) = sint^/x In y2)
29. /(x, y) =xy3+x3y; P0(l,-1. ~2)
11. /(x, y) = y/3x2 + y4 12. /(x, y) = xy2 ln(x + y)
2 2
13. /(x, y) = x2ex+y cos y 14. /(x, y) = xy3 tan-1 y
30. fix, y) = Y +>’ ; P0(l. -1,-2)
15. /(x, y) = sin~‘(xy) 16. /(x, y) = cos~'(xy) xy
Determine fx, fy, and fz in Problems 17-22. 31. fix, y) = x2 sin(x + y); P0 (f, f, 0)
17. f(x, y, z) = xy2 + yz3 + xyz 32. f(x, y) = x ln(x + y2); P0ie, 0, e)
18. f(x, y, z) = *ye:
Q 33. Determine fx and fy for
x + y2
19. /(x, y, z) =
z
xy -I- yz
20. /(x,y,z) = f(x,y) it2 + 2t+ \)dt
xz
21. /(x, y, z) = ln(x + y2 + z3)
22. /(x, y, z) = sin(xy + z) Hint: Review the second fundamental theorem of calculus.
718 Chapter 11 Partial Differentiation
34. Determine fx and fy for 45. Modeling Problem Biologists have studied the oxygen con¬
sumption of certain furry mammals. They have found that if
the mammal’s body temperature is T degrees Celsius, fur tem¬
f(x,y) (<?' +3 t)dt
perature is t degrees Celsius, and the mammal does not sweat,
then its relative oxygen consumption can be modeled by
Hint: Review the second fundamental theorem of calculus.
C(m, t, T) = oiT — t)m~067
A function f(x, y) is said to be harmonic on the open set S if fxx
and fyy are continuous and (kg/h), where m is the mammal’s mass (in kg) and a is a
physical constant. Compute the rate (rounded to two deci¬
fxx + fyy = 0 mal places) at which the oxygen consumption changes with
respect to
throughout S. Show that each function in Problems 35-38 is har¬ a. the mass m
monic on the given set. b. the body temperature T
35.. f(x, y) = 3x2y — y3; S is the entire plane. c. the fur temperature t
46. Modeling Problem A gas that gathers on a surface in a con¬
36. fix, y) = ln(x2 + y2); S is the plane with the point (0, 0)
removed. densed layer is said to be adsorbed on the surface, and the
surface is called an adsorbing surface. The amount of gas ad¬
37. f{x, y) = ex sin y; S is the entire plane.
sorbed per unit area on an adsorbing surface can be modeled
38. /(x, y) = sin x cosh y; S is the entire plane. by
39. For f(x, y) = cos xy2, show fxy = fyx. Sip, T, h) = apeh/{bT)
40. For fix, y) = (sin2x)(siny), show fxy = fyx. where p is the gas pressure, T is the temperature of the gas, h
is the heat of the adsorbed layer of gas, and a and b are phys¬
41. Find fxzy - fyzz, where fix, y, z) = x2 + y2 - 2 xy cos z.
ical constants. Compute the rate of change of S with respect
42. Two commodities Q\ and Q2 are said to be substitute com¬ to
modities if an increase in the demand for either results in a. p b. h c. T
a decrease in the demand of the other. Let D\(pi, p2) and
47. The ideal gas law says that PV = kT, where P is the pres¬
DiiP\, Pi) be the demand functions for Qx and Q2, respec¬
sure of a confined gas, V is the volume, T is the temperature,
tively, where pi and p2 are the respective unit prices for the and & is a physical constant.
commodities.
„ , dV 3p
^ , . , dD 1 3D, a. Calculate —. b. Calculate —-.
a. Explain why —— < 0 and- < 0. 3T dV
dp\ 3 p2 P1 , 3 P dV 3 T
u A 3D, j 3D2 c. Show that---- — = — 1.
d. Are —— and —— positive or negative? Explain. dV 3 T dP
dp2 3pi 48. At a certain factory, the output is given by the production func¬
c. Give examples of substitute commodities. tion Q — 120K2/3L2/5, where K denotes the capital invest¬
ment (in units of $1,000) and L measures the size of the labor
43. Two commodities Q\ and Q2 are said to be complementary
force (in worker-hours).
commodities if a decrease in the demand for either results in
a decrease in the demand of the other. Let Dfpu p2) and a. Determine the marginal productivity of capital, dQ/dK,
and the marginal productivity of labor, dQ/dL.
33lip\. Pi) be the demand functions for Qx and Q2, respec¬
tively, where p\ and p2 are the respective unit prices for the b. Determine the signs of the second-order partial derivatives
commodities. d2Q/dL2 and d2Q/dK2, and give an economic interpreta¬
T • , 3D, 3Di tion.
a. Is it true that -— < 0 and -— < 0? Explain.
dp 49. The temperature at a point (x, y) on a given metal plate in the
3 Pi
dD 3 D2 xy-plane is determined according to the formula
b. Determine whether —— and-are positive or negative? Tlx, y) — x3 + 2xy2 + y degrees. Compute the rate at which
dpi dp\
Explain. the temperature changes with distance if we start at (2, 1) and
move
c. Give examples of complementary commodities.
a. parallel to the vector j
44. Modeling Problem The flow (in cm3/s) of blood from an artery b. parallel to the vector i
into a small capillary can be modeled by
50. In physics, the wave equation is
d2Z _ 2 d2Z
Fix, y, z) = —— y/y - z
4 312 dx2
and the heat equation is
lor constant c > 0, where x is the diameter of the capillary, y
is the pressure in the artery, and z is the pressure in the cap¬ 3z 2 d2z
illary. Compute the rate of change of the flow of blood with ~dt ~ C dx2
respect to
In each of the following cases, determine whether z satisfies
a. the diameter of the capillary the wave equation, the heat equation, or neither.
b. the arterial pressure a- Z = e~' (sin—f cos — ^ b. z = sin 3cr sin 3x
c. the capillary pressure
c. z = sin 5ct cos 5x
11.3 Partial Derivatives 719
51. The Cauchy-Riemann equations are 58. Modeling Problem Suppose a substance is injected into a tube
containing a liquid solvent and that the tube is placed so that
3u dv 3u dv
its axis is parallel to the x-axis, as shown in Figure 11.19.
dx 3y dy dx
-
b. u = x2 + y2, v = 2.vy x=0
c. u = ln(x2 + y2), v = 2tan~' ^ —^
Figure 11.19 Problem 58
52. When two resistors R\ and R2 are connected in parallel, their
combined resistance R satisfies
Assume that the concentration of the substance varies only
1 _ 1 1 in the x-direction, and let C(x, t) denote the concentration at
~R ~ ^ + position x and time t. Because the number of molecules in
this substance is very large, it is reasonable to assume that C
32R 32R i
What is is a continuous function whose partial derivatives exist. One
Jr2 3r| model for the flow yields the diffusion equation in one dimen¬
53. Show that the production function P(L, K) = LaKp, where sion., namely,
a and f> are constants, satisfies
3P 3P 3C 32C
L-b K — (a + fi)P — =<5—v
3L 3K 31 dx2
54. The kinetic energy of a body of mass m and velocity v is where 8 is the diffusion constant.
K - \mv2. Show that
a. What must <5 be for a function of the form
3K 32K
C(x,t) = eax+b,
3m dv2
55. A study of ground penetration near a toxic waste dump mod¬ (,a and b are constants) to satisfy the diffusion equation?
els the amount of pollution P at a depth x (in feet) and time t
b. Verify that
(in years) by the function
TANGENT PLANES
Suppose S is a surface with the equation z — f(x, y), where / has continuous first
partial derivatives fx and fy. Let P0(-^o. ,yo> Zo) be a point on S, and let C\ be the
curve of intersection of S with the plane x = x0 and C2, the intersection of 5 with the
plane y = yo, as shown in Figure 11.20a. The tangent lines T\ and T2 to C\ and C2,
respectively, at P0 determine a unique plane, and in Section 11.6 we will find that this
plane actually contains the tangent to every smooth curve on S that passes through Pq.
We call this plane the tangent plane to S at Pq (Figure 11.20b).
*
Surface S
Surface S
Tangent
plane
Plane x = x0
Plane y = y0
a. Cj and C2 are the curves of b. The tangent plane to S at P0
intersection of the surface S contains the tangent lines Tj and T2
with the planes x = x0 and y = y0. to Cj and C2, respectively.
To find an equation for the tangent plane, recall that the equation of a plane with
normal N = Ai + B\ + Ck is
A(x - jr0) + B(y - y0) + C(z - zo) = 0
If C ^ 0, divide both sides by C and let a = —A/C and b = — B/C to obtain
Z ~ Zo = a(x - xo) + b(y - y0)
The intersection of this plane and the plane x = *0 is the tangent line T\, which
we know has slope fy(x0, yo) from the geometric interpretation of partial derivatives.
Setting x = xq in the equation for the tangent plane, we find that T\ has the point-slope
form
z - ZO = b(y - yo)
dz
so we must have b = fy(x0, yo) . Similarly, setting y = y0, we obtain
dy
Uo.yo)
z - zo = a(x - *0)
which represents the tangent line T2, with slope a = fx(x0, yo). To summarize:
Suppose S is a surface with the equation z = f(x,y) and let Po(*o. >o» Zo) be
a point on S at which a tangent plane exists. Then an equation for the tangent
plane to S at Pq is
Ax + By -f- Cz -f- D — 0
Solution
-2
-yx -y -V3 -V3
fxix, y) = /,( l,>/3) =
yx2 x2 + y2 ' 1 +3 ~ 4
X-1 1
fy(x, y) =
2 x2 -f- y2 ’ 4
7T
Z~3 =
INCREMENTAL APPROXIMATIONS
In Chapter 3, we observed that the tangent line to the curve y = /(x) at the point
P(x0, Jo) is the line that best fits the shape of the curve in the immediate vicinity of P.
That is, if / is differentiable at x = xo and the increment Ax is sufficiently small, then
or, equivalently,
Similarly, the tangent plane at P(xo, yo, Zo) is the plane that best fits the shape of
the surface z = fix, y) near P, and the analogous incremental (or linear) approxi¬
mation formula is as follows.
so that
or
Tangent plane
z=f(x,y)
As long as we are near (xo, yo), the height of the tangent plane is approximately
the same as the height of the surface. Thus, if |Ax| and |Ay| are small, the point
(xo + Ax, yo + Ay) will be near (xo, yo) and we have
/(x0 + Ax, y0 + Ay) ~ /(x0, yo) + fx(xo, yo)Ax + fy(x0, y0)Ay
Solution
An open (no top) box with length x, width y, and height z has surface area
S — xy + 2xz + 2yz
Bottom Four side faces
Because the sides cost $2/ft2 and the bottom $3/ft2, the total cost is
C(x, y, z) — 3xy + 2(2xz + 2yz)
The partial derivatives of C are
Figure 11.22 Construction
of a box Cx — 3y + 4z Cy = 3x + 4z C, = 4x + 4y
and the dimensions of the box change by
Solution
We are given that
AR AH
< 0.03 and
R H
AV « tRH) AR + (ijrR2) AH
df = dx + dy = /v(x, y) dx + fy(x, y) dy
dx dy
where dx and dy are independent variables. Similarly, for a function of three
variables w — f(x,y,z) the total differential is
df 3/ df
df = dx + —- dy + dz
dx dy dz
Solution
a. df = — dx + — dy + — dz — 6x2 dx + 20y3 dy -6 dz
dx dy dz
724 Chapter 11 Partial Differentiation
. 3/ . 3/ .
b. df = —dx + — dy
ox dy
1
_
SO
-—1
-2
2x ln(3y2 — 2x) + x2 dx + dy
H
m
(N
3 y2 — 2x
H
1
.
2x2 6x2y
2x ln(3y2 — 2x)
3y2 — 2x _ dx + W^dy
Solution
The change in output is estimated by the total differential dQ. We have K = 900,
L = 1,000, dK = AK = 1, and dL — AL = —2. The total differential of Q(x, y) is
3Q , 3Q
dQ = — dK + —dL
3K 3L
= 60 (\)K~l/2Ll/3 dK + 60 (\)KXI2L~213 dL
= 30K~l/1L1/3 dK + 20K{/2L~213 dL
Substituting for K, L,dK, and dL,
I - J_ 1
R~Y{+Y2
If is measured as 300 ohms with a maximum error of 2% and R2 is measured as
500 ohms with a maximum error of 3%, what is the maximum percentage error in /??
Solution
We are given that
dRx d R2
< 0.02 and < 0.03
Ri r2
dR 7?i Ri
and we wish to find the maximum value of . Because R (solve the
~R~ K\ + R2
given equation for R), we have
3R R2 3R
and Quotient rule
dR\ (Ri + R2)2 dR2 (/?, + R2)2
it follows that the total differential of R is
dR dR
dR — -— dR\ T ^ dR2
dR dR2
dR\ + dR2
(/?! + Rl)2 (R\ + R2)‘
11.4 Tangent Planes, Approximations, and Differentiability 725
dR R\ + Ri
We now find — by dividing both sides by R; however, since — — it follows
R R Ri R2
that
R R i + A*2
dR- - = dR\ + dR2
R (Rx + Ri)2 (R\ + R2)2 R1R2
dR R2 dR\ Ri d R2
+
~R Ri +R2 Ri Ri + /?2 ^2
dR Ri dR\ Ri dR2
< +
R R\ + R2 Ri Ri + R2 r2
500 300
< (0.02) + (0.03) = 0.02375
300 + 500' ' ' 300 + 500
The maximum percentage is approximately 2.4%.
DIFFERENTIABILITY
Definition of Differentiability The function /(x, y) is differentiable at (x0, yo) if the increment of f can be
expressed as
A/ = fx(x0, yo)Ax + fy(x0, yo)Ay + e\Ax + e2Ay
We wish to show that /(x, y) -* /(x0, y0) as (x, y) -* (x0, y0) or, equiva¬
lently, that
WARNING D , , . .
Be careful about how you use the word differentiable. In the single-variable
case, a function is differentiable at a point if its derivative exists there. However, the word is
used differently for a function of two variables. In particular, the existence of the partial
derivatives fx and fy does not guarantee that the function is differentiable, as illustrated in the
following example.
That is, the function / has the value 1 when (x, y) is in the first quadrant and is 0
elsewhere. Show that the partial derivatives fx and fy exist at the origin, but / is not
differentiable there.
Solution
Since /(0, 0) = 0, we have
/(0 + Ax, 0) - /(0, 0)
A (0,0) lim = 0
Ajt-*0 Ax
and similarly, fy(0, 0) = 0. Thus, the partial derivatives both exist at the origin.
If fix, y) were differentiable at the origin, it would have to be continuous there
( Theorem 11.2). Thus, we can show f is not differentiable by showing that it is not
continuous at (0, 0). Toward this end, note that lim fix, y) is 1 along the line
C*,y)-*(0,0)
y = x in the first quadrant but is 0 if the approach is along the x-axis. This means that
the limit does not exist. Thus, f is not continuous at (0, 0) and consequently is also
not differentiable there.
11.4 Tangent Planes, Approximations, and Differentiability 727
Although the existence of partial derivatives at P(xo, yo) is not enough to guaran¬
tee that f{x, y) is differentiable at P, we do have the following sufficient condition for
differentiability.
Note that the function in If / is a function of x and y, and /, fx, and fy are continuous in a disk D centered at
Example 7 does not contradict Theorem (x0, yo). then / is differentiable at (xo, yo)-
11.3. because there is no disk centered
at (0, 0) on which / is continuous. Proof The proof is found in advanced calculus texts. -i
Show that fix, y) = x2y + xy3 is differentiable for all (x, y).
Solution
Compute the partial derivatives
/yO> y) = 3y
+ *j3) = x2 +3^2
Because /, fx, and fy are all polynomials in x and y, they are continuous throughout
the plane. Therefore, the sufficient condition for differentiability theorem assures us
that / must be differentiable for all x and y. ■
5. z = tan 1 - at P0 (2, 2, 33. A closed box is found to have length 2 ft, width 4 ft, and height
3 ft, where the measurement of each dimension is made with
6. z = In |x + >,2| at Po(-3, —2, 0)
a maximum possible error of ±0.02 ft. The top of the box is
Determine the total differential of the functions given in Problems made from material that costs $2/ft2; and the material for the
7-18. sides and bottom costs only $1.50/ft2. What is the maximum
U>
error involved in the computation of the cost of the box?
00
7. fix, y) — 5 x2y3 8. fix, y)
*
ii
9. fix, y) — sinxy 34. A cylindrical tank is 4 ft high and has an outer diameter of
10. fix, y) = cosx
2 ft. The walls of the tank are 0.2 in. thick. Approximate the
y
11. fix, y) —
12. fix, y) _ *2 volume of the interior of the tank assuming the tank has a top
X y and a bottom that are both also 0.2 in. thick.
13. fix, y) — yex 14. fix. y) = exl+y 35. The Higrade Company sells two brands, X and Y, of a com¬
15. fix, y, z) = 3x3 - 2y4 ± 5z mercial soap, in thousand-pound units. If x units of brand X
16. fix, y, z) = sinx + sin y + cos z and y units of brand Y are sold, the unit price for brand X is
40. Modeling Problem A business analyst models the sales of a new the specific gravity of the object is
product by the function
P(x,y) = (x — 30) (70 — 5x + 4y) + (y — 40) (80 + 6x — 7y) has volume V — |nabc.
Q 47. Show that the following function is not differentiable at (0, 0):
dollars, where x cents is the price per can of the first brand and
y cents is the price per can of the second. Currently the first
brand sells for 50/ per can and the second for 52/ per can. 2 if(x,y) ^ (0,0)
fix, y) x2 + y2
Use the total differential to estimate the change in the weekly
0 if (x, y) = (0, 0)
profit that will result if the grocer raises the price of the first
brand by 1 / per can and lowers the price of the second brand
48. Compute the total differentials
by 2/ per can.
43. A juice can is 12 cm tall and has a radius of 3 cm. A manu¬
facturer is planning to reduce the height of the can by 0.2 cm
and the radius by 0.3 cm. Use a total differential to estimate
the percentage decrease in volume that occurs when the new Why are these differentials equal?
cans are introduced. (Round to the nearest percent.) 49. Let A be the area of a triangle with sides a and b separated by
44. Modeling Problem It is known that the period T of a simple an angle 9, as shown in Figure 11.24.
pendulum with small oscillations is modeled by
where L is the length of the pendulum and g is the accelera¬ Figure 11.24 Problem 49
tion due to gravity. For a certain pendulum, it is known that
L — 4.03 ft. It is also known that g = 32.2 ft/s2. What is Suppose 6 = |, and a is increased by 4% while b is decreased
the approximate error in calculating T by using L — 4 and by 3%. Use differentials to estimate the percentage change in
A.
g = 32?
45. If the weight of an object that does not float in water is x 50. In Problem 49, suppose that 6 also changes by no more than
pounds in the air and its weight in water is y pounds, then 2%. What is the maximum percentage change in A?
We begin with a differentiable function of two variables fix, y). If x = x(f) and
y = y{t) are, in turn, functions of a single parameter t, then z = fix it), yit)) is
730 Chapter 11 Partial Differentiation
a composite function of a parameter t. In this case, the chain rule for finding the
derivative with respect to one parameter can now be stated.
dz dz dx dz dy
dt dx dt dy dt
dt —— __dt
t t
*Tree diagram illustrating the chain rule What This Says The tree diagram shown in the margin is a device for re¬
membering the chain rule. The diagram begins at the top with the dependent
variable z and cascades downward in two branches, first to the independent vari¬
ables x and y, and then to the parameter t on which they each depend. Each
branch segment is labeled with a derivative, and the chain rule is obtained by
first multiplying the derivatives on the two segments of each branch and then
adding to obtain
dz dz dx dz dy
dt dx dt dy dt
left branch right branch
Proof Recall that because z = f(x, y) is differentiable, we can write the increment
Az in the following form:
Az = — Ax + — Ay + <?iAx + €2Ay
dx dy
Az dz Ax
At 3x At
We know that x and y both vary continuously with t (remember, they are differen¬
tiable), and it follows that Ax -► 0 and Ay -»• 0 as At -► 0, so that e, ->■ 0 and
£2 ^ 0 as At —> 0. Therefore, we have
dz dz Ax dz Ay Ax Ay
lim — = lim
dt Ar->0 At At-^0 dx At 3y At ^' At + At
dz dx dz dy dx dy
□
3x dt ^ dy dt dt ^ ^ dt
Solution
dz
Thus, 2r3 + 4U.
dt
b. Because z = a2 + y2 and a = t 1, y = f2,
dz dz _ dx _ 2 dy n
— =2t
3a “ dy ^V’ dt dt
Use the chain rule for one independent parameter:
dz dz dx dz dy
dt 3a dt dy dt
— (2a ){-t~2) + 2y(2t)
= -2 r3 + At2
Solution
dz
= Ux2 + 2xy)-l/2(2x + 2 y) and — = Ux2 + 2a.v)-1/2(2a)
dx z dy
dx dy
= — sin 6 and — = cos 9. Use the chain rule for one independent parameter
to find
dz dz dx dz dy
d6 dx dO 3y d6
= i(A2 + 2av)~1/2(2a + 2y)(— sin6>) + |(a2 + 2Ay)“1/2(2A)(cos0)
Solution
dr dh
The volume of the cylinder is V = 7ir2h, and we are given 3 and — = -5. We
Tt dt
find that
dV dV 9
— = tt(2 r)h and — = nr2(\)
dr dh
Figure 11.25 Right circular By the chain rule for one parameter,
cylinder dV dV dr dV dh n dr ?dh
— --1-= 2nrh— + n r~ —
dt dh dt dr dt dt dt
Thus, at the instant when r = 10 and h = 8, we have
dV 2
-= 2tt(10)(8)(3) + tt(10)2(—5) = -20tt
dt
The volume is decreasing at the rate of about 62.8 in.3/min.
732 Chapter 11 Partial Differentiation
3 F dx dF dy 3F dF dy
dx dx dy dx dx dy dx
so
dF
dy 3v Fx
— = -g^r- = provided Fy £ 0
3y
This formula provides a useful alternative to the procedure for implicit differentiation
introduced in Section 3.6. This alternative procedure is illustrated in the following
•*
example.
sin(x + y) + cos(x — y) = y
find —
dx
Solution
Let F(x, y) = sin(x + y) + cos(x - y) - y, so that F(x, y) = 0. Then
Fx = cos(x + y) - sin(x - y)
Fy — cos(x + y) — sin(x — y)(—1) — 1
so
dy _ Fx _ —[cos(x + y) — sin(x - y)]
dx Fy cos(x + y) + sin(x — y) — 1
Solution
By using the chain rule, we have
dz 3z dx dz dy
dt 3x dt ^ 3y dt
and
3£ r d_ 'dx\ dx dz d_(dy_ dy
dx dt sdi) dt 3y dt dt
dz d2x dx d2z dx d2z dy dz d2y (£^dx+yzdy\
dx dt2
+ — + +
dx2 dt dxdy dt dy dt2 \ 3y3x dt dy2 dt J
11.5 Chain Rules 733
_ , . . dx dy
Substituting — = a and — = b,we obtain
dt dt
dh dz / d2z d2z dz / 92 d2z
—— (0) -\- ci \ ——a -f- b + —(0) + b
dt2 dx \dx2 dxdy dv 8 y3d+S?”
d2Z _ . d2Z ,id2z d2Z d2Z
— Cl + 2 ab- +b Note
dx2 ' dxdy ' dy2 dxdy dydx
Next, we will consider the kind of composite function that occurs when x and y are
both functions of two parameters. Specifically, let z = F(x, y), where x = x(u, v)
and y — y(u, v) are both functions of two independent parameters u and v. Then
Z = F[x(u, v), y(u, u)] is a composite function of u and v, and with suitable assump¬
tions regarding differentiability, we can find the partial derivatives dz/du and dz/dv
by applying the chain rule obtained in the following theorem.
dz dz dx dz dy dz dz dx dz dy
du dx du dy du dv dx dv dy dv
Proof This version of the chain rule follows immediately from the chain rule for one
independent parameter. For example, if v is fixed, the composite function
z — f[x(u, v), y(u, u)] depends on u alone, and we have the situation described in
the chain rule of one independent variable. We apply this chain rule with a partial
derivative (because x and y are functions of more than one variable):
dz dz dx dz dy
du dx du dy du
dz
The formula for — can be established in a similar fashion. □
dv
Solution
First find the partial derivatives:
= JL(4X - yl) = 4
dx dx dy dy
dy d , 2
_ = —(uv2) = V2 — — —(u3v) — 3 u2v
du du du du
and
dx d 2 dy 9 , 3
— = —(uv ) = 2uv — = —(uv) — u
dv dv dv dv
Therefore, the chain rule for two independent parameters gives
dz dz dx dz dy
du dx du dy du
= (4)(u2) + (—2y)(3u2v) = 4v2 — 2(u3v)(3u2v) = 4u2 — 6 u5v2
734 Chapter 11 Partial Differentiation
and
3z 3z dx 3z dy
dv dx dv + dy dv
= (4)(2wu) + (—2 y)(w3) = 8«t> — 2(m3u)m3 = 8wu — 2u6v ■
Solution
dw dx dw dy dw dz
dx ds dy ds ^ dz ds
~(4* + y2 + z3)
_3*
~-(ers2)
[3^
+ ^-(4* + y2 + z3) LL9^8 Ur+S)]
V /.t
/ \
= 4[ers\2rs)j+2y -j+3 z2irt2)
r +s
\~ /
2y 2„2
= 8 rsers + + 3 rt
r + s
11.5 Chain Rules 735
3w 2 'l r _1_ c
= 8 rse +-In — + 3 r W
ds r 4- s t
1^
1^
3x3 y b' dx2
In Problems 8—11, the function Fix, y) depends on x and y. Let
28. x3 + y2 +- z2 = 5 29. xyz = 2
x — x(u, d) and y = y(u, v) be given functions of u and v. Let
Vto
+
30. ln(x + y) 31. x-' +y"‘ + z-1 =3
II
Z — F[x(u, v), y(u, u)] and find the partial derivatives dz/du and
dz/dv in these two ways: 32. x cos y = y + z 33. z2 + sinx = tan y
a. Express z explicitly in terms ofu and v. 34. Let fix. y) be a differentiable function of x and y, and
x — r cos 6, y — r sin 8 for r > 0 and 0 <9 <2n.
b. Apply the chain rule for two independent parameters.
8. F(x, y) — x y2, where x — u + v and y — u — v a. If z = /[x(r, 9), y(r, 9)], find ^ and
dr dO
9. F(,x, y) = x2 + y2, where x — u sin v and y = u — 2v b. Show that
10. Fix, y) — exy, where x — u — v and y = u + v dz
2
11. Fix, y) — lnxy, where x — euv , y = euv dy
Write out the chain rule for the functions given in Problems 12-15.
35. Let z = fix, y), where x — au and y = bv, with a, b con¬
12. z = fix, y), where x = xls, t), y = yls, t) stants. Express d2z/du2 and d2z/dv2 in terms of the partial
13. w = f lx, y, z), where x = x(s, t), y — yls, t), z = zls, t) derivatives of z with respect to x and y. Assume the existence
and continuity of all necessary first and second partial deriva¬
14. t = flu, v), where u = w(x, y, z, w), v = u(x, y, z, w)
tives.
15. w — flx,y,z), where x = xls,t,u), y — yls,t,u), 36. Let (x, y, z) lie on the ellipsoid
z = zls, t, u)
Find the indicated derivatives or partial derivatives in Problems
16-21. Leave your answers in mixed form (x, y, z, t).
Without solving for z explicitly in terms of x and y, compute
16. Find —, where w = ln(x + 2y — z2) and x = 2t — \,y — ~, the higher-order partial derivatives
dt t
z = J~t- d2Z d2Z
—- and -
dw . , , _ i_, 3x2 3x3y
17. Find —, where w = stnxyz and x = 1 - 31, y = e ,
dt
Z = 41. 37. The dimensions of a rectangular box are linear functions of
dw r„2 time, lit), wit), and hit). If the length and width are increas¬
18. Find —, where w — ze • and x .= sinr, y = cost,
dt ing at 2 in./sec and the height is decreasing at 3 in./sec, find
z = tan 21. the rates at which the volume V and the surface area 5 are
i 9
changing with respect to time. If f(0) = 10, w(0) = 8, and
19. Find —, where w = ex +yz and x = -, y = ln(2/ — 3), h(0) = 20, is V increasing or decreasing when t = 5 sec?
dt t '
.?
What about S when t = 5?
736 Chapter 11 Partial Differentiation
38. Van der Waal’s equation in physical chemistry states that a gas 43. The combined resistance R produced by three variable resis¬
occupying volume V at temperature T (Kelvin) exerts pres¬ tances R\, R2, and R2 connected in parallel is modeled by the
sure P, where formula
1 _ i J_ J_
R ~ Yi+ Y2 + T3
(P + 72) (y " fl) = kT Suppose at a certain instant, R\ = 100 ohms, R2 = 200 ohms,
R3 — 300 ohms, and that R] and R2 are decreasing at the rate
for physical constants A, B, and k. Compute the following of 1.5 ohm/s while R2 is increasing at the rate of 2 ohms/s.
rates: How fast is R changing with respect to time at this instant? Is
a. the rate of change of volume with respect to temperature it increasing or decreasing?
b. the rate of change of pressure with respect to volume In Problems 44^18, assume that all functions have whatever
39. The concentration of a drug in the blood of a patient t hours derivatives or partial derivatives are necessary for the problem
after the drug is injected into the body intramuscularly is mod- to be meaningful.
• eled by the Heinz function 44. If z — f(uv2), show that
3z 3z
C = —-—(e~al — e~bt) b > a 3u 3
b—a
45. If z = f(u — v, v — u), show that
where a and b are parameters that depend on the patient’s
metabolism and the particular kind of drug being used. 3z 3z
3C 3C , 3C 3u dv
a. Compute the rates —, —, and —.
da db at 46. If z = u + f(uv), show that
b. Explore the assumption that a = (In b)/t, b constant for
dz dz
t > (In b)/b. In particular, what is dC/dtl u-v — - u
3u dv
40. A paint store carries two brands of latex paint. An analysis of
sales figures indicates that the demand Q for the first brand is
47. If w = f , show that
modeled by
3w dw
Q(x, y) = 210 — \2x2 + 18y r-\- s— = 0
dr 3s
gallons/month, where x, y are the prices of the first and second 48. If z = xy + f(x2 + y2), show that
brands, respectively. A separate study indicates that t months
from now, the first brand will cost x — 4 + 0.18? dollars/gal dz dz 2 2
y-x— = y — x
and the second brand will cost y = 5 + 0.3dollars/gal. At dx dy
what rate will the demand Q be changing with respect to time
49. Let w = /(f) be a differentiable function of t, where
9 months from now?
r = (x2 + y2 + z2)I//2. Show that
41. To model the demand for the sale of bicycles, it is assumed
\ 2
that if 24-speed bicycles are sold for x dollars apiece and the dw\2 dw (/ dw
o
\ dw
price of gasoline is y cents per gallon, then ~dt ) dx ) + V 3y ) + V 3z
where E is the number of employees and T (°C) is the aver¬ d2u d2u
age temperature during the workday. Currently, there are 142 for a constant c. Show that = c~ -. Hint: Let
dt2 dx2
employees and the average temperature is 18°C. If the aver¬ r = x + ct\ s = x — ct.
age daily temperature is falling at the rate of 0.23°/day and 53. Suppose z = /(x,y) has continuous second-order partial
the number of employees is increasing at the rate of 3/month, derivatives. If x = er cos6 and y = er sin#, show that
what is the corresponding effect on the rate of pollution? Ex¬
press your answer in units/day. For this model, assume there 32Z d2z _ _2r d2z d2z
are 22 workdays/month. dx2 + dy2~(' [dr2 + S#2
11.6 Directional Derivatives and the Gradient 737
In Chapter 3 we defined the slope of a cur\>e at a point to be the ratio of the change
in the dependent variable to the change in the independent variable at the given point.
To determine the slope of the tangent line at a point Po(*o. yo) on a surface defined by
z = f(x, y), we need to specify the direction in which we wish to measure. We do
this by using vectors. In Section 11.3 we found the slope parallel to the xz-plane to
be the partial derivative fx(xo, yo). We could have specified this direction in terms of
738 Chapter 11 Partial Differentiation
the unit vector i (jc-direction), while fy(x, y) could have been specified in terms of the
unit vector j. Finally, to measure the slope of the tangent line in an arbitrary' direction,
we use a unit vector u = u \ i + utJ in that direction.
To find the desired slope, we look at the intersection of the surface with the vertical
plane passing through the point P0 parallel to the vector u, as shown in Figure 11.26.
This vertical plane intersects the surface to form a curve C, and we define the slope of
the surface at Pq in the direction of u to be the slope of the tangent line to the curve C
defined by u at that point.
We summarize this idea of slope in a particular direction with the following definition.
Directional Derivative Let / be a function of two variables, and let u = u\i + u2j be a unit vector. The
directional derivative of / at Po(*o? Jo) in the direction of u is given by
WARNING
Remember, u must be a fix o + hui,y0 + hu2) - f(x o, yo)
unit vector. Du fix o, yo) = lim
/i—>0 h
provided the limit exists.
At a particular point Pq(xq, yo), there are infinitely many directional derivatives
to the graph of z = f(x, y), one for each direction radiating from Pq. Two of these
are the partial derivatives fx(xo, yo) and fy(xo, yo)- To see this, note that if u = i (so
u\ = 1 and u2 = 0), then
fix 0 + h, y0) - fjx o, yo)
Difix o, yo) = fim fxix 0, yo)
/!-» 0 h
and if u = j (so m i =0 and u2 = 1),
Solution
First, find the partial derivatives fx(x,y) = -4x and fy(x,y) — 3y2. Then since
u\ = j and u2 — — we have
Go (1.2,9)
The directional derivative of f(x, y) at the point PoUo. To) in the direction of the
unit vector u = (u\,u2) can be interpreted as both a rate of change and a slope. For
instance, in Example 1, the intersection of the surface z — 3 — 2x2 + y3 with the
vertical plane through the point P( 1,2) parallel to the unit vector u = is a
Ml. 2) curve C, and the directional derivative Duf( 1, 2) = —12.3 is the slope of C at the
point <2(1, 2, 9) on the surface above P, as shown in Figure 11.27. The directional
Figure 11.27 The graph of the surface derivative also gives the rate at which the function f(x, y) = 3 — 2x2 + y3 changes as
z = 3-2x2 + y3 a point (x, y) moves from P in the direction of u.
THE GRADIENT
Gradient Let / be a differentiable function at (x, y) and let f{x,y) have partial derivatives
fx(x, y) and fy(x, y). Then the gradient of /, denoted by V/ (pronounced “del
eff”), is a vector given by
Solution
Begin with the partial derivatives:
3 3
fx(x,y) =—(x2y+ y3) = 2xy and fy(x, y) =—(x2y + y3) = x2 + 3y2
dx 3y
Then
V/(x,y) = 2xyi + (x2 + 3y2)j U
The following theorem shows how the directional derivative can be expressed in
terms of the gradient.
Proof Because V/0 = fx(xo, yo)i + /yUo, yo)j and u = u\i + u2j, we have
Daf(x0, y0) = V/0 • u = fx(xo, y0)«i + fy(x0, y0)u2 □
Solution
2x 2
fx(x, y) = so fx( 1,-3) =
x2 + y3 ’ 26
3y2 27
fy(x, y) = SO fy ( 1 , -3) =
x2 + y3 ’ 26
V/0 = V/(l,-3) = -|i-gj
Du(x, y) = V/ • u =
77^13
my,)'
338
Although a differentiable function of one variable f(x) has exactly one derivative
f'(x), a differentiable function of two variables F(x, y) has two partial derivatives and
an infinite number of directional derivatives. Is there any single mathematical concept
for functions of several variables that is the analogue of the derivative of a function
of a single variable? The properties listed in the following theorem suggest that the
gradient plays this role.
11.6 Directional Derivatives and the Gradient 741
V(o/ + bg) = (af + &g)*i + (af + 6g)vj = (afx + 6g.,)i + (afy + bgy) j
= afx i + ^g.vi + afy j + /?gvj = a(/*i + />j) + + g.yj)
= aVf + bVg
Power rule
The other rules are left for the problem set (Problem 57). □
In applications, it is often useful to compute the greatest rate of increase (or decrease)
of a given function at a specified point. The direction in which this occurs is called the
direction of steepest ascent (or steepest descent). For example, suppose the function
z = f(x, y) gives the altitude of a skier coming down a slope, and we want to state a
theorem that will give the skier the compass direction of the path of steepest descent
(see Figure 11,28b). We emphasize the words “compass direction” because the gradi¬
ent gives direction in the xy-plane and does not itself point up or down the mountain.
The following theorem shows how the direction of maximum change is determined by
the gradient (see Figure 11.28).
What This Says The theorem states that at Pq the function / increases
most rapidly in the direction of the gradient V/0 and decreases most rapidly in
the opposite direction (see Figure 11.29).
Solution
We begin by finding the gradient of /:
V/ = h i + fy) = [e2y~x + xe2y~x + [xe2y~x{2)]\
= e2y~x[(l-x)i + 2xj]
At (2, 1), V/o = e2(1) 2[(1 — 2)i + 2(2)j] = — i + 4j. The most rapid rate of increase
is IIV/o|| = -y/(— l)2 + (4)2 = \f\l and it occurs in the direction of —i + 4j. The most
rapid rate of decrease occurs in the direction of — V/0 = i — 4j. ■
The directional derivative and gradient concepts can easily be extended to functions of
three or more variables. For a function of three variables, f(x, y, z), the gradient V/
is defined by
V/ = fXi+ fy} + fz k
and the directional derivative Duf of f(x, y, z) at Po(x0, yo> Zo) in the direction of the
unit vector u is given by
Duf = V/o • u
where, as before, V/0 is the gradient V/ evaluated at Po. The basic properties of the
gradient of /(x, y) (Theorem 11.8) are still valid, as is the maximal direction property
of Theorem 11.9. Similar definitions and properties are valid for functions of more
than three variables.
11.6 Directional Derivatives and the Gradient 743
Solution
Begin with the partial derivatives:
fx — y sin(xz) + xy(z cos(xz))', fx( 1, —2, 7i) — —2sin7r — 2tx costt = 2tx
fy = x sin(xz); fy( 1, —2, tt) = 1 sin7r = 0
fz — xy(x cos(xz)); fz( 1, —2, tt) — (1)(—2)(l)cos7r = 2
Thus, the gradient of / at Pq is
V/0 = 27ri + 2k
To find Duf we need u, the unit vector in the direction of v:
v —2i + 3j - 5k 1
u— (—21 + 3j - 5k)
Ml V(-2)2 + (3)2 + (-5)2 V38
Finally,
I
Duf (1, -2, tt) - V/0 • u -=(-4tt - 10) w -3.66 ■
V38
Normal
Proof Let C be any smooth curve on the level surface f(x, y,z) — K that passes
through P0(x0, y0, Zo)> and describe the curve C by the vector function
R(f) = x(t)i + y(0j + z(t)k for all t in some interval /. We will show that the
gradient V/0 is orthogonal to the tangent vector dR/dt at P0.
Because C lies on the level surface, any point P(x(t), y(t), z(t)) on C must satisfy
f[x(t), y(t), z(t)] = K, and by applying the chain rule, we obtain
d dx dy , dz
~[f(x(t), y(t), z(0)] = fx(x, y, z)— + fy(x, y, z)— + ft(x, y, z)—
744 Chapter 11 Partial Differentiation
d
y(t), z(f))]
dt r=r0
dx dy dz
= fx(x(t0), y(to), z(to))— + fy(x(t0), y(t0), z(t0))-j- + fz(x(t0), y(t0), z(t0)) —
dt dt dt
dR
= V/„.~
dR dx. dy. dz
since — = —i + — j + —k. We also know that f(x(t), y(t), z(t)) = K for all t
dt dt dt dt
in / (because the curve C lies on the level surface f(x, y, z) = K). Thus, we have
^;{f[x(t),y(t),zm
dt = 4-(K)
dt =0
dR
and it follows that V/0 • — = 0. We are given that V/o ^ 0, and dR/dt ^ 0 because
dt
the curve C is smooth. Therefore, V/o is orthogonal to dR/dt, as required. □
^ What This Says The gradient V/0 at each point Po on the surface
JR
Solution
Since the gradient vector at Pq is perpendicular to the level surface, we have
Solution
The level curve for C = 1 is a hyperbola given by x2 — y2 — 1, as shown in Figure
11.31. The gradient vector is perpendicular to the level curve. We have
Figure 11.31 The level curve so at the point (2, \/3), V/o = 4i — 2>/3j is the required normal. This normal vector
x2 - y2 = 1 and a few others are shown in Figure 11.31. ■
11.6 Directional Derivatives and the Gradient 745
Solution
The flow of heat in the region is given by a vector function H(x, y), whose value at
each point (x, y) depends on x and y. From physics it is known that H(x, y) will be
perpendicular to the isothermal curves T(x, y) = C for C constant. The gradient VF
and all its multiples point in such a direction. Therefore, we can express the heat flow
as H = —kVT, where k is a positive constant (called the thermal conductivity). The
negative sign is introduced to account for the fact that heat flows “downhill” (that is, in
the direction of decreasing temperature).
Figure 11.32 Isotherms of Because T(3, 4) = 25, the point Po(3, 4) lies on the isotherm F(x, y) = 25, which
T(x, y) — x2 + y2. Heat flow at P0 is part of the circle x2 + y2 = 25, as shown in Figure 11.32. We know that the heat
is in the direction of
flow Ho at P0 will satisfy H0 = —kVTo, where V7o is the gradient at Po. Because
-VT0 = — 6i - 8j.
VF = 2xi + 2vj, we see that VT0 = 6i + 8j. Thus, the heat flow at Po satisfies
H0 = -kVT0 = -*(6i + 8j)
Normal line
Because the thermal conductivity k is positive, we can say that heat flows from Po in
the direction of the unit vector u given by
Tangent
Normal vector — (6i + 8j) 3. 4.
u= , = —1-1
v/(-6)2 + (-8)2 5 5
Tangent planes and normal lines to a surface are the natural extensions to M2 of the
tangent and normal lines we examined in IR2. Suppose S is a surface and N is a vector
Surface S normal to S at the point P0. We would intuitively expect the normal line and the tangent
plane to S at P0 to be, respectively, the line through Po with the direction of N and the
plane through Po with normal N (see Figure 11.33). These observations lead us to the
Figure 11.33 Tangent plane and
following definition.
normal line
Tangent Plane and Normal Line Suppose the surface S has a nonzero normal vector N at the point P0. Then the
to a Surface line through P0 parallel to N is called the normal line to S at P0, and the plane
through P0 with normal vector N is the tangent plane to S at P0.
EXAMPLE 9 Finding the tangent plane and normal line to a given surface
Find equations for the tangent plane and the normal line at the point P0(l, —1,2) on
the surface S given by x2y + y2z + z2x = 5.
Solution
We need to rewrite this problem so that the normal property of the gradient theorem
applies. Let F(x, y, z) = x2y + y2z + z2x, and consider S to be the level surface
F(x, y, z) = 5. The gradient VF is normal to 5 at P0. We find that
VF(x, y, z) = (2xy + z2)i + (x2 + 2yz)j + (y2 + 2xz)k
746 Chapter 11 Partial Differentiation
so a normal vector at Pq is
x = l+21, y = —\ — 3t, z = 2 + 5t ■
Fx(xo, yo, Zo)(* - *o) + Fy(xo, To, zo)(y - To) + Fz(x0, yQ, zo)(z - z0) = 0
and the normal line to S at Po has parametric equations
Solution
If Po(x0, To, zo) is the point of tangency and x0 — 3, yo = 4, and z0 > 0, then
Zo = ^ + To2 = V9 + 16 = 5
If we consider F(x, y, z) = x2 + y2 — z2, then the cone can be regarded as the level
surface F(x, y, z) = 0. The partial derivatives of F are
Fx =2x Fy = 2 y Fz = -2 z
so at P0(3, 4, 5),
11
->+j
y
13. fix, y) == ln(x2 + 3y) (1, 1) i +j 40. Find the directional derivative of /(x, y) — x2 -\- xy + y2 at
Pod, —1) in the direction toward the origin.
14. /(*. y) == ln(3x + y2) (0, 1) >- J 2 2
15. fix,y) == sec(xy — y3) (2,0) —i-3j 41. Find the directional derivative of f(x, y) = ex r at P0( 1, — 1)
in the direction toward Q(2, 3).
16. fix, y) == sinxy i^/n, yfn) 3;ri -
42. Let f(x, y, z) = 2x2 — y2 + 3z2 — 8x — 4y + 201, and let P0
Find a unit vector that is normal to each surface given in Prob¬ be the point (2, — |, 4)
lems 17-24 at the prescribed point, and the standard form of the
a. Find V/0.
equation of the tangent plane at that point.
b. Find cos 6, where 6 is the angle between V/0 and the vec¬
17. x2 + y2 + z2 = 3 at (1, — 1, 1) tor toward the origin from P0.
18. x4 + / + z4 = 3at(l,-l - 1) 0^3. Let fix, y, z) = xyz, and let u be a unit vector perpendicular
19. cosz = sin(x + y) at (|, f) to both v = i — 2j -p 3k and w = 2i + j — k. Find the directional
derivative of / at P0(l, — 1,2) in the direction of u.
20. sin(x 4- >0 + tan(y + z) = 1 at (f, f, -f)
44. Let fix, y, z) = yex+z + zey~x. At the point P(2, 2, —2), find
21. In f^ = Oat (2, 5,3) the unit vector pointing in the direction of most rapid increase
of /.
22. In ( —^ ) = x — z at (1, 0, 1) 45. Modeling Problem Suppose a box in space given by 0 < x < 2,
\y + zj 0<y<2, 0<z<2is temperature-controlled so that
23. zex+2y = 3 at (2,-1,3) the temperature at a point Pix, y, z) in the box is modeled by
T(x, y, z) = xy + yz + xz. A heat-seeking missile is located
24. ze*2-? =3 at (1,1, 3)
at P0(l, 1, 1). In what direction will the missile move for the
Find the direction from P0 in which the given function f increases
temperature to increase as quickly as possible? What is the
most rapidly and compute the magnitude of the greatest rate of
maximum rate of change of the temperature at the point P(l?
increase in Problems 25-34.
Modeling Problem A metal plate covering the rectangular re¬
25. fix, y) — 3x + 2y - 1; P0(l, -1) gion 0 < x < 6, 0 < y < 5 is charged electrically in such
26. fix, y) — 1 -x2-y2; P0(l, 2) a way that the potential at each point (x, y) is inversely pro¬
27. fix, y) — x3 + y3; P0(3, -3) portional to the square of its distance from the origin. If an
object is at the point (3, 4), in which direction should it move
28. fix,y) = ax + by + c; Po(a, b)
to increase the potential most rapidly?
29. fix, y, z) — ax2 + by2 + cz2; P0(a, b, c);
A hiker is walking on a mountain path when it begins to rain.
assume a4 + b4 + c4 ^ 0.
If the surface of the mountain is modeled by z = 1 — 3x2 — pv2
30. fix, y) — ax3 + by3-, P0ia, b); assume a6 + b6 0. (where x, y, and z are in miles) and the rain begins when the
31. fix, y) — In^x2 + y2; Pod, 2) hiker is at the point P0 (|, — 5, j|), in what direction should
she head to descend the mountainside most rapidly?
32. fix,y) = sinxy; P0
Let / have continuous partial derivatives, and assume the
33. fix,y, z) = (x + y)2 + (y + z): + ix + z)2; Po(2, —1,2) maximal directional derivative of / at (0, 0) is equal to 100
and is attained in the direction toward (3, —4). Find the gradi¬
34. fix, y. z) = zln P0(l.*, -1)
ent V/ at (0, 0).
In Problems 35-38, find a unit vector that is normal to the given 49 Suppose at the point P0(— 1,2), a certain function fix, y) has
graph at the point P0(x0, y0) on the graph. Assume that a, b, and directional derivative 8 in the direction of V] = 3i — 4j and
c are constants. 1 in the direction of v2 = 12i + 5j. What is the directional
35. the line ax + by — c 36. the circle x2 + y~ = a~ derivative of / at P0 in the direction of v = 3i — 5j?
748 Chapter 11 Partial Differentiation
u= (cosf)i+(sin|)j
T(x, y) = 3(x — 6)2 + L5(y — l)2 + 41
60. Suppose that u and v are unit vectors and that / has continu¬
degrees Fahrenheit, where x and y are in feet. The Spy begins ous partial derivatives. Show that
at the point (1,5) and stumbles across the room at the rate of
4 ft/min, always moving in the direction of maximum temper¬ D-+v/ = — j- (Duf + Dyf)
ature decrease. But he can last no more than 2 minutes under l|u +v||
these conditions! Assuming that there is an escape hole at the 61. Counterexample Problem If /, fx, and fy are continuous and
point where the temperature is minimal, does he make it or is V f(x, y) — 0 inside the disk x2 + y2 < 1, then either show
the Spy toast at last? that f(x,y) is a constant function inside the disk or find a
55. Recall that an ellipse is the set of all points P(x, y) such that counterexample.
the sum of the distances from P to two fixed points (the foci) 62. Counterexomple Problem If / is differentiable at P0(x0, y0)
is constant. Let P(x, y) be a point on the ellipse, and let r\ and D„f(x0, y0) = 0 for unit vectors U| and u2, where
and r2 denote the respective distances from P to the two foci, U| x u2 7^ 0, then show that Duf(xo, yo) = 0 for every unit
F\ and F2 as shown in Figure 11.36. vector u. or find a counterexample.
11.7 Extrema of Functions of Two Variables 749
63. Let R = xi + vj + ek, and let r = ||R|| = fx2 + y2 + z2. 64. Suppose the surfaces F(x, y, z) = 0 and G(x, y, z) = 0 both
pass through the point Poixo, yo, Zo) and that the gradients
a. Show that Vr is a unit vector in the direction of R.
VF0 and VGo both exist. Show that the two surfaces are tan¬
b. Show that V(r") = nrn~2R, for any positive integer n. gent at P0 if and only if VF0 x VGo = 0.
There are many practical situations in which it is necessary or useful to know the
largest and smallest values of a function of two variables. For example, if T(x, y) is
the temperature at a point (x, y) in a plate, where are the hottest and coldest points
in the plate and what are these extreme temperatures? If a hazardous waste dump is
bounded by the curve F(x,y) — 0, what are the largest and smallest distances to the
boundary from a given interior point PqI We begin our study of extrema with some
terminology.
Absolute Extrema The function f(x,y) is said to have an absolute maximum at (x0, yo) if
/(To, .Vo) > fix, y) for all (x, y) in the domain D of /. Similarly, / has an
absolute minimum at (x0, yo) if fix 0, yo) < fix, v) for all (x, y) in D. Col¬
lectively, absolute maxima and minima are called absolute extrema.
Relative Extrema Let / be a function defined on a region containing (x0, yo)- Then
f(x o, yo) is a relative maximum if fix, y) < /(x0, yo) for all (x, y) in an
open disk containing (x0, yo)-
/ (To, Vo)is a relative minimum if fix, y) > f (x0, yo) for all (x, y) in an
open disk containing (x0, yo).
RELATIVE EXTREMA
Relative
maxima
(peaks)
Relative
minimum
(valley)
Proof Let F(x) = f(x, yo). Then F(x) must have a relative extremum at x — xo,
so F'{x0) = 0, which means that fx(x0, y0) = 0. Similarly, G(y) = /(x0, y) has a
relative extremum at y = yo, so G'(yo) = 0 and fy(xo, yo) = 0- Thus, we must have
both fx(xo, yo) = 0 and fy(xo, yo) = 0, as claimed. □
WARNING
There is a horizontal tangent plane at each extreme point where the first partial
derivatives exist. However, this does not say that whenever a horizontal tangent plane occurs
at a point P, there must be an extremum there. All that can be said is that such a point P is a
possible location for a relative extremum.
In single-variable calculus, we referred to a number xo, where f'{xo) does not exist
°r /'(*o) = 0 as a critical number. This terminology is extended to functions of two
variables as follows.
Critical Points A critical point of a function / defined on an open set D is a point (x0, yo) in D
where either one of the following is true:
a. z = x2 + y2 b. z = 1 - x2 - y2 c. z = y2 - x2
Solution
The graphs of these quadric surfaces are shown in Figure 11.38. Let /(x, y) = x2+y2,
g(x, y) = 1 — x2 — y2, and h{x, y) = y2 — x2. We find the critical points:
a* /*(*, y) = 2x, fy (x, y) = 2y; critical point (0, 0). The function f has a relative
minimum at (0, 0) because x2 and y2 are both nonnegative, yielding x2 + y2 > 0
for all nonzero x and y.
11.7 Extrema of Functions of Two Variables 751
Minimum in yz-plane z
Maximum in xz-plane
A critical point P0(xQ, Jo) is called a saddle point of /(x, y) if every open disk
centered at P0 contains points in the domain of / that satisfy /(x, y) > f (x0, yo) as
well as points in the domain of / that satisfy f(x, y) < f(x0, yo)- An example of a
saddle point is (0, 0) on the hyperbolic paraboloid z = y2 ~ x2, as shown in Figure
11.38c.
D — fxxfyy ~~ fXy
752 Chapter 11 Partial Differentiation
Then
A saddle point occurs at P0 if D(x0, y0) < 0. z =y2 -x2 (See Ex lc)
Zk
+ Rel. max.
+ + Rel. min.
NA Saddle point If D(x0, y0) = 0, then the test is inconclusive. Nothing can be said about the nature of
0 NA Inconclusive the surface at (x0, yo) without further analysis.
Proof The second partials test can be proved by using an extension of the Taylor se¬
ries expansion (Section 8.8) that applies to functions of two variables f(x, y). Details
can be found in most advanced calculus texts (also see Problem 60). □
The discriminant D = fxx fyy — f*y may be easier to remember in the equivalent
SMH determinant form
The definition of a 2 x 2
determinant is presented in Section r) _ fxx fxy
2.8 of the Student Mathematics fxy fyy
Handbook. Note that if D > 0 at the critical point Pq(xq, Jo), then fxx and fyy must have the same
sign. This is the reason that when D > 0, either fxx > 0 or fyy >0 is enough to
guarantee that a relative minimum occurs at P0 (or a relative maximum if fxx < 0 or
fyy < 0).
Geometrically, if D > 0 and fxx > 0 and fyy > 0 at P0, then the surface
z — f(x,y) curves upward in all directions from the point £?(x0, jo, Zo). so there
is a relative minimum at Q. Likewise, if D > 0 and fxx < 0 and fyy < 0 at P0,
then the surface curves downward in all directions from P0, which must therefore be
a relative maximum. However, if D < 0 at PQ, the surface curves up from Q in some
directions, and down in others, so Q must be a saddle point.
WARNING
The second partials test says nothing about the geometry of the surface at Q if
D — 0 at Pq. Example 4 and Problem 54 show that a relative minimum, a relative maximum,
a saddle point, or something entirely different may occur if D = 0.
11.7 Extrema of Functions of Two Variables 753
Solution
First, find the critical points:
fx — Ax + 2y — 2 fy = 2x+2y-2
4x + 2y — 2 = 0
2x + 2y — 2 = 0
and solve to obtain x = 0, y = 1. Thus, (0, 1) is the only critical point. To apply the
second partials test, we obtain
For the critical point (0, 1) we have D = 4 > 0 and fxx = 4 > 0, so there is a relative
minimum at (0, 1).
EXAMPLE 3 Second partials test with a relative minimum and a saddle point
Find all critical points on the graph of /(x, y) = 8x3 — 24xy + y3, and use the second
partials test to classify each point as a relative extremum or a saddle point.
Solution
fx O, y) = 24x2 - 24y, fy(x, y) = -24x + 3y2
24x2 - 24y = 0
—24x + 3y2 = 0
From the first equation, y — x2; substitute this into the second equation to find
—24x + 3(x2)2 = 0
x(x3 — 8) = 0
To view the situation graphically, we calculate the coordinates of the saddle point
(0, 0, 0), and the relative minimum (2, 4, -64), as shown in Figure 11.39.
754 Chapter 11 Partial Differentiation
Saddle Point
Relative
minimum
(2,4,-64)
». View near the origin (showing b. View away from the origin (showing c. Level curves
the saddle point) the relative minimum point)
f(x, y) = x2y4
The graph is shown in Figure 11.40.
Solution
Since fx(x, y) = 2xy4, fy(x, y) — 4x2y3, we see that the critical points occur only
when x = 0 or y — 0; that is, every point on the x-axis or y-axis is a critical point.
Because
the discriminant is
12x2y2
H
fxy fyy
Since D = 0 for any critical point (x0, 0) or (0, y0), the second partials test fails.
However, f(x,y) =0 for every critical point (because either x = 0 or y — 0, or both),
and because / (x, y) = x2y4 > 0 when x / 0 and y ^ 0, it follows that each critical
point must be a relative minimum. ■
The extreme value theorem (Theorem 4.1) says that a function of a single variable
/ must attain both an absolute maximum and an absolute minimum on any closed,
bounded interval [a, b] on which it is continuous. In R2, a nonempty set S is closed if
it contains its boundary (see the introduction to Section 11.2) and is bounded if it is
contained in a disk. The extreme value theorem can be extended to functions of two
variables in the following form.
Solution
Step 1. fx(x, y) = 2xexl~yl and fy(x, y) = -2ye*2~y2. These partial derivatives
are defined for all (x, y). Because fx(x, y) = /v(x, y) = 0 only when x = 0
and y = 0, it follows that (0, 0) is the only critical point of / and it is inside
the disk.
Step 2. Examine the values of / on the boundary curve x2 + y2 = 1. Because
y2 = 1 — x2 on the boundary of the disk, we find that
Solution
If Q(x, y, z) is a point on the plane x + 2y + z = 5, then z = 5 - x - 2y and the
distance from P to Q is
= y/ x1 + (y - 3)2 + (5 - x - 2y - 4)2
since the minimum of d will occur at the same points where d2 is also minimized.
To minimize f(x,y), we first determine the critical points of / by solving the
system
fx — 2x - 2(1 - x - 2y) = Ax + 4y - 2 = 0
fy = 2(y - 3) - 4(1 - x - 2y) = 4x + lOy -10 = 0
Check: You might want to check your work by using the formula for the distance
from a point to a plane in 1R3 (Theorem 9.9):
In the following example, calculus is applied to justify a formula used in statistics and
in many applications in the social and physical sciences.
Solution
We wish to find values of m and b that minimize the sum of the squares of the differ¬
ences between the y-values and the line y = mx -f b. The line that we seek is called
the regression line. Suppose that the point P( has components (x*, y*). At this point
the value on the regression line is y = mxk + b and the value of the data point is yk.
The “error” caused by using the point on the regression line rather than the actual data
point can be measured by the difference
y* - (mxk + b)
11.7 Extrema of Functions of Two Variables 757
The data points may be above the regression line for some values of k and below the
regression line for other values of k. We see that we need to minimize the function that
represents the sum of the squares of all these differences:
n
= 2m yt x\ + 2b \xk - 2 xkyk
k=l k= 1 *=1
Figure 11.42 Least squares
n
approximation of data
Fb(m, b) = J^2[y* - (mxk + 6)](-l)
*=i
n n »
= 2m ^ ** + 2b ^ 1 — 2 ^ y*
*=i *=i *=i
n n
= 2m ^ xk + 2bn - 2 ^ yjt
k=l k= 1
Set each of these partial derivatives equal to 0 to find the critical values (see Problem
61).
n n
"E
*=i
x*yk - (^
vk= 1
^) (51) E4E»~
k=l k=1
EE«‘
a=i a=i
m = and b -
n E4-E*»
k=1 a=i *=i \*=i
It can be shown that these values of m and b yield an absolute minimum for F(m, b).
Most applications of the least squares formula stated in Example 7 involve using
a calculator or computer software. The following technology note provides an exam¬
ple.
TECHNOLOGY NOTE
Many calculators will carry out the calculations required by the least
squares approximation procedure. Look at your owner s manual tor
specifics. Most calculators allow you to input data with keys labeled |STAT]
and |DATA!. After the data are input, the m and b values are given by pressing
the |LinReg| choice. For example, ten people are given a standard IQ test.
Their scores are then compared with their high school grades:
Find the critical points in Problems 4-23, and classify each point
as a relative maximum, a relative minimum, or a saddle point.
9x
8- f(x,y)
x2 + y2 + 1
20. /(x, y) = x2 + y2 + —
xy
Find the least squares regression line for the data points given in York water for y dollars per bottle, then consumers will buy
Problems 31-34. approximately 40 — 50x + 40y bottles of California water and
20 + 60x — 70y bottles of the New York water each day. How
31. (-2.-3), (-1,-1), (0,1), (1,3), (3,5)
should the owner price the bottled water to generate the largest
32. (0, 1), (1, 1.6), (2.2,3), (3.1.3.9), (4,5)
possible profit?
33. (3,5.72), (4,5.31), (6.2,5.12), (7.52,5.32), (8.03,5.67)
46. Modeling Problem A telephone company is planning to intro¬
34. (-4,2), (-3,1), (0,0), (1,-3), (2,-1), (3,-2) duce two new types of executive communications systems that
35. Find all points on the surface y2 = 4 + xz that are closest to it hopes to sell to its largest commercial customers. To create
the origin. a model to determine the maximum profit, it is assumed that
36. Find all points in the plane x + 2y + 3z = 4 in the first octant if the first type of system is priced at x hundred dollars per
where /(x, y, z) = x2yz3 has a maximum value. system and the second type at y hundred dollars per system,
approximately 40 - 8x + 5y consumers will buy the first type
37. A rectangular box with no top is to have a fixed volume. What
and 50 + 9x - 7y will buy the second type. If the cost of
should its dimensions be if we want to use the least amount of
manufacturing the first type is $1,000 per system and the cost
material in its construction?
of manufacturing the second type is $3,000 per system, how
38. A wire of length L is cut into three pieces that are bent to form
should the telephone company price the systems to generate
a circle, a square, and an equilateral triangle. How should the
maximum profit?
cuts be made to minimize the sum of the total area?
47. Modeling Problem A manufacturer with exclusive rights to a
39. Find three positive numbers whose sum is 54 and whose prod¬
sophisticated new industrial machine is planning to sell a lim¬
uct is as large as possible. ited number of the machines to both foreign and domestic
40. A dairy produces whole milk and skim milk in quantities x firms. The price the manufacturer can expect to receive for the
and y pints, respectively. Suppose the price (in cents) of machines will depend on the number of machines made avail¬
whole milk is p(x) = 100 - x and that of skim milk is able. For example, if only a few of the machines are placed
q(y) = 100 —y, and also assume that C(x, y) = x2 +xy + y2 on the market, competitive bidding among prospective pur¬
is the joint-cost function of the commodities. Maximize the chasers will tend to drive the price up. It is estimated that if
profit the manufacturer supplies x machines to the domestic market
P(x, y) = px + qy - C(x, y) and y machines to the foreign market, the machines will sell
for 60 - 0.2x + 0.05y thousand dollars apiece at home and
41. Let R be the triangular region in the xy-plane with vertices
50 — 0.1 y + 0.05x thousand dollars apiece abroad. If the man¬
(-1, -2), (-1,2), and (3, 2). A plate in the shape of R is
ufacturer can produce the machines at a total cost of $10,000
heated so that the temperature at (x, y) is
apiece, how many should be supplied to each market to gen¬
T(x, y) = 2x2 - xy + y2 - 2y + 1 erate the largest possible profit?
48. A college admissions officer. Dr. Westfall, has compiled
(in degrees Celsius). At what point in R or on its boundary is the following data relating students’ high-school and college
T maximized? Where is T minimized? What are the extreme GPAs:
temperatures?
HS GPA 2.0 2.5 3.0 3.0 3.5 3.5 4.0 4.0
42. A particle of mass m in a rectangular box with dimensions x,
y, z has ground state energy College GPA 1.5 2.0 2.5 3.5 2.5 3.0 3.0 3.5
k2 Plot the data points on a graph and find the equation of the
E(x, y, z) = — regression line for these data. Then use the regression line to
8m
predict the college GPA of a student whose high school GPA
where k is a physical constant. If the volume of the box is fixed is 3.75.
(say Vo = xyz), find the values of x, y, and z that minimize 49. It is known that if an ideal spring is displaced a distance y from
the ground state energy. its natural length by a force (weight) x, then v = kx, where k
43. A manufacturer produces two different kinds of graphing is the so-called spring constant. To compute this constant for
calculators, A and B, in quantities x and y (units of a particular spring, a scientist obtains the following data:
1,000), respectively. If the revenue function (in dollars) is
x (lb) 5.2 7.3 8.4 10.12 _12.37
R(x v) = -*2 _ 2y2 + 2xy + 8x + 5y, find the quantities of
A and B that should be produced to maximize revenue. y (in.) 1L32 LL56 17.44 21.96 26.17
44. Suppose we wish to construct a closed rectangular box with
Based on these data, what is the “best” choice for kl
volume 32 ft3. Three different materials will be used in the
construction. The material for the sides costs $ 1 per square
50. Exploration Problem The following table gives the approximate
foot, the material for the bottom costs $3 per square foot, and U.S. census figures (in millions):
the material for the top costs $5 per square foot. What are the 1920 1930 1940
Year: 1900 1910
dimensions of the least expensive such box?
Population: 76.2 92.2 106.0 123.2 132.1
45. Modeling Problem A store carries two competing brands of bot¬
tled water, one from California and the other from upstate New
Year: 1950 1960 1970 1980 1990
York. To model this situation, assume the owner of the store
Population: 151.3 179.3 203.3 226.5 248.7
can obtain both at a cost of $2/bottle. Also assume that if the
California water is sold for x dollars per bottle and the New
760 Chapter 11 Partial Differentiation
a. Find the least squares regression line for the given data and b. Show that g(x, y) = x2y2 has a relative minimum at (0, 0).
use this line to “predict” the population in 1997. (The ac¬
c. Show that h(x, y) — x3 + y3 has no extremum or saddle
tual population was about 266.5 million.)
point at (0, 0).
b. Use the least squares linear approximation to estimate the
55. Counterexample Problem If / is a continuous function of one
population at the present time. Check your answer by look¬
variable with two relative maxima on a given interval, there
ing up the population using the Internet. Comment on the
must be a relative minimum between the maxima. By consid¬
accuracy (or inaccuracy) of your prediction.
ering the function*
51. Exploration Problem The following table gives the Dow Jones
Industrial Average (DJIA) Stock Index along with the per fix, y) = 4x V - 2a4 - e4-v
capita consumption of wine (in gallons) for those years.
show that it is possible for a continuous function of two vari¬
Year: 1965 1970 1975 1980 1985 1990 1995
ables to have only two relative maxima and no relative min¬
DJIA Index: 911 753 802 891 1,328 2,796 3,838 ima.
Consumption: 0.98 1.31 1.71 2.11 2.43 2.05 1.79
56. Counterexample Problem If a continuous function of one vari¬
a. Plot these data on a graph, with the DJIA Index on the x- able has only one critical number on an interval, then a rela¬
axis and consumption on the y-axis. tive extremum must also be an absolute extremum. Show that
b. Find the equation of the least squares line. this result does not extend to functions of two variables by
considering the functionf
c. Determine whether the consumption figures predicted by
the least squares line in part b are approximately correct by
fix, y) = 3xe-v — a3 - eiy
using the most recent figures available at the time of this
writing: The DJIA opened 1997 at 6,442 and U.S. wine In particular, show that it has exactly one critical point, which
consumption was 1.95 gallons. Interpret your findings. corresponds to a relative maximum. Is this also an absolute
d. In 2001 the stock market began the year at 10,790. Use the maximum? Explain.
least squares line to predict per capita wine consumption 57. Tom, Dick, and Mary are participating in a cross-country relay
that corresponds to this stock value. race. Tom will trudge as fast as he can through thick woods
52. Linearizing nonlinear data In this problem, we turn to data that to the edge of a river, and then Dick will take over and row to
do not tend to change linearly. Often we can “linearize” the the opposite shore. Finally, Mary will take the baton and run
data by taking the logarithm or exponential of the data and along the river road to the finish line. The course is shown in
then doing a linear fit as described in this section. Figure 11.44.
a. Suppose we have, or suspect, a relationship y = kxm.
Show that by taking the natural logarithm of this equation Road |«—.- 4.3 mi-►)
we obtain a linear relationship: Y = K +mX. Explain the ■4j(0,3.7) _ Mary ? F (4.3,3.7)
new variables and constant K. River ' A
58. Exploration Problem Consider the function 60. Prove the second partials test. Hint: Compute the second
f(x, y) = (y — x2)(y — 2x2). Discuss the behavior of this directional derivative of / in the direction of the unit vector
function at (0. 0). u = hi + £j and complete the square.
59. Exploration Problem Sometimes the critical points of a func¬ 61. Verify the formulas for m and b associated with the least
tion can be classified by looking at the level curves. In each squares approximation.
case shown in Figure 11.45, determine the nature of the criti¬ 62. This problem involves a generalization of the least squares
cal point(s) of z — f(x, y) at (0, 0). procedure, in which a “least squares plane” is found to pro¬
duce the best fit for a given set of data. A researcher knows
z = -l that the quantity z is related to x and y by a formula of the
form z = k\x + k2y, where k\ and k2 are physical constants.
To determine these constants, she conducts a series of experi¬
ments, the results of which are tabulated as follows:
Proof Denote the constraint curve g(x, y) = c by C, and note that C is smooth. We
represent this curve by the vector function
R (0 = x(r)i + y(0j
for all t in an open interval /, including t0 corresponding to P0, where x'{t) and y'(t)
exist and are continuous. Let F(t) = f(x(t), y(t)) for all t in I, and apply the chain
rule to obtain
Because f(x, y) has an extremum at P0, we know that F(t) has an extremum at /0, the
value of t that corresponds to P0 (that is, P0 is the point on C where t = t0). Therefore,
we have F'(t0) = 0 and
The general procedure for the method of Lagrange multipliers may be described as
follows.
Suppose / and g satisfy the hypotheses of Lagrange’s theorem, and that f(x, y)
has an extremum subject to the constraint g(x, y) = c. Then to find the extreme
value, proceed as follows:
2. Evaluate / at all points found in step 1 and all points on the boundary of the
constraint. The extremum we seek must be among these values.
Solution
x (L i) = i (l)2-(l)2 = l
The endpoints of the line segment
x+y = 1 for x > 0, y > 0
z = 1 - x2 - y2
are at (1,0) and (0, 1), and we find that
/(1,0) = 1 - l2 -02 = 0
Figure 11.46 The maximum is the
high point of the curve of intersection /(0, 1) = 1 — 02 — l2 = 0
of the surface and the plane. Therefore, the maximum value is ^ at (j, |), and the minimum value is 0 at (1,0) and
(0, 1). (See Figure 11.46.) ~ g
11.8 Lagrange Multipliers 763
Plane Solution
The cube and plate are shown in Figure 11.47. We will use Lagrange multipliers to
find all critical points in the interior of the plate, and then we will examine the plate’s
boundary. To apply the method of Lagrange multipliers, we must solve VT = AVg,
where g(x, y, z) — x 4- y + z. We obtain the partial derivatives.
ro
&0
Co
Tx = ~4x Ty = —2y
II
II
II
II
1
to
We must solve the system
II
II
i
SMH)
See Problem 29 of
<N
r<
Ctq
II
II
1
V;
Figure 11.47 Find the hottest and
Problem Set 3 of the Student
E4T
to
^O
1
coldest points on the plate inside
II
II
1
to
Mathematics Handbook.
the cube. x +y + z = 1 4 - g(x, y, z) = 1
The solution of this system is (±, §, §). The boundary of the plate is a triangle with
vertices A(l, 0, 0), 5(0, 1,0), and C(0, 0, 1). The temperature along the edges of this
triangle may be found as follows:
Edge AC: x + z — 1, y = 0 7j(x) = 4-2x2- (0)2 - (1 - x)2 = 3 - 3x2 + 2x, 0 < x < 1
Edge AB: x + y = 1, z = 0 T2(x) = 4 - 2x2 - (1 - jc)2 - (0)2 = 3 - 3a:2 + 2x, 0 < x < 1
The last step is to evaluate T at the critical points and the endpoints:
t(L 2 2\ 33.
1 V5’ 5’ 5/ J5’
T G’°> I) = 35;
T(i,f,0) = 3i; 7 (0, i) = 3^;
Comparing these values (remember that the temperature is in hundreds of degrees Cel¬
sius), we see that the highest temperature is 360°C at (2, |, |) and the lowest temper¬
ature is 200°C at (1,0, 0).
764 Chapter 11 Partial Differentiation
Notice that the multiplier is used only as an intermediary device for finding the
critical points and plays no role in the final determination of the constrained extrema.
However, the value of A is more important in certain problems, thanks to the interpre¬
tation given in the following theorem.
The coordinates of the optimal ordered pair (x, y) depend on c (because different con¬
straint levels will generally lead to different optimal combinations of x and y). Thus,
dE dE dx dE dy
dc 3x dc dy dc
„ dx dy
fx~jdc b fy ~r~
dc
Because E — f(x, y)
dx dy
^gx — + kgy — Because fx = Agx and fy = Agy
dc dc
( dx dy'
A* Chain rule
dc
dg
A Because —— = 1 (remember g = c) □
This theorem can be interpreted as saying that the multiplier estimates the change
in the extreme value E that results when the constraint c is increased by 1 unit. This
interpretation is illustrated in the following example.
Q(x, y) = 50x2/5y3/5
units. If $150,000 is available, how should this capital be allocated between labor and
equipment to generate the largest possible output? How does the maximum output
change if the money available for labor and equipment is increased by $1,000? In
economics, an output function of the general form Q(x, y) = x“v^-a is known as a
Cobb-Douglas production function.
Solution
Because x and y are given in units of $1,000, the constraint equation is x + y = 150.
If we set g(x, y) = x + y, we wish to maximize Q subject to g(x, y) = 150. To apply
the method of Lagrange multipliers, we first find
Qx = 20x~3/5y3/5 Qy = 30x2/5y-2/5 gx = 1 gy = 1
11.8 Lagrange Multipliers 765
30 x2/5y“2/5 = A(l)
x + y = 150
20*-3V/5 = 30 *2/5y-2/5
20y = 3 Ox
y - 1,5x
Substitute y = 1.5x into the equation x + y — 150 to find x = 60. This leads to the
solution y — 90, so that the maximum output is
1 = 20(60)~3/5(90)3/5 « 25.50849001
Thus, the maximum output is about 3,826 units and occurs when $60,000 is allocated
to labor and $90,000 to equipment. We also note that an increase of $1,000 (1 unit)
in the available funds will increase the maximum output by approximately A ~ 25.51
units (from 3,826.27 to 3,851.78 units). ■
The method of Lagrange multipliers can also be applied in situations with more than
one constraint equation. Suppose we wish to locate an extremum of a function defined
by /(x, y, z) subject to two constraints, g(x, y, z) = c\ and h(x, y, z) — C2, where
g and h are also differentiable and Vg and V/7 are not parallel. By generalizing La¬
grange’s theorem, it can be shown that if (xo, yo, Zo) is the desired extremum, then
there are numbers A and /x such that g(xo, yo, Zo) = c\, h(x0, yo, zo) = Q, and
As in the case of one constraint, we proceed by first solving this system of equations
simultaneously to find A, /x, xo, yo, Zo and then evaluating /(x, y, z) at each solution
and comparing to find the required extremum. This approach is illustrated in our final
example of this section.
Solution
The distance from a point (x, y, z) to the origin is s = y^x2 + y2 + z2, but instead of
minimizing this quantity, it is easier to minimize its square. That is, we will minimize
Find the point on th /(x, y, z) =x2 4-y2 + z2
intersection curve that x + 2y + z = 10
is closest to the origin. subject to the joint constraints
fx = 2x fy = 2 y fz = 2z
gx = 1 gy = 2 gz = 1
hx ■ 2x hy = 2 y h: = — 1
To apply the method of Lagrange multipliers, we use the formula
V/(xo, yo, zo) = AVg(xo, y0, z0) + +V/?(x0, y0, Zo)
which leads to the following system of equations:
2x = A(l) + /x(2x)
2 y = A (2) + /x(2y)
■ 2 z = A(l) + /x(— 1)
x + 2y + z — 10
z = x2 + y2
This is not a linear system, so solving it requires ingenuity.
Multiply the first equation by 2 and subtract the second equation to obtain
4x — 2 y — (4x — 2 y)/x
(4x — 2 y) — (4x — 2y)/x = 0
(4x — 2y)(l — /x) = 0
4x — 2y = 0 or 1 — /x = 0
CASE I: If 4x — 2y = 0, then y = 2x. Substitute this into the two constraint equa¬
tions:
x -1- 2y + z 10 x2 + y2 - z = 0
—
x + 2(2x) + 2 = 10 x2 + (2x)2 -2 = 0
L/i
2 = 5x2
iN
*
II
1
By substitution we have 5x2 = 10 — 5x, which has solutions x = 1 and
x = —2. This implies
x — —2
II
y = 2x = 2(1) = 2 y = 2x = 2(—2) = -4
2 = 5x2 = 5(1)2 = 5 2 = 5x2 = 5(-2)2 = 20
Thus, the points (1,2, 5) and (—2, —4, 20) are candidates for the minimal
distance.
CASE II: If 1 - ft = 0, then /I — 1, and we look at the system of equations involving
x, y, z, A., and /x.
2x = A(l) + /x(2x)
2 y = A (2) + /x(2y)
■ 2z = A(l) + /x(— 1)
x + 2v + z — 10
z — x2 + y2
The top equation becomes 2x = A + 2x, so that X = 0. We now find z from
the third equation:
2z = — 1 or z = —\
Next, turn to the constraint equations:
x + 2y + 2 = 10 x2 + y2 - 2 = 0
X + 2 y — 4 — 10 x2 + y2 + \ = 0
x + 2y = 10 + \ x2 + y2 =
There is no solution because x2 + y2 cannot equal a negative number.
We check the candidates for the minimal distance:
/(x, y, z) = x2 + y2 + z2 so that
/(1,2,5) = l2 + 22 + 52 = 30
/(—2, -4, 20) = (—2)2 + (—4)2 + 202 = 420
Because /(x, y, z) represents the square of the distance, the minimal distance is \/30
and the point on the intersection of the two surfaces nearest to the origin is (1,2, 5). ■
11.8 Lagrange Multipliers 767
A GEOMETRIC INTERPRETATION
~gx
Slope of constraint curve g(x, y) = c is
gy '
of f{x, y)
Slope of each level curve is zA
Figure 11.49 Increasing level curves fy
and the constraint curve The condition that the slopes are equal can be expressed by
~fx ~gx . ,
or, equivalently,
.. fx A
fy 8y gx gy
Let X equal this common ratio,
fx fy
X = and
gx gy
so that
fx = Zgx and fy = A.gy
and
v/ = /x i + fyj = X(gxi + gyj) = XVg
Because the point in question must lie on the constraint curve, we also have
g(x, y) — c. If these equations are satisfied at a certain point (a, b), then / will
reach its constrained maximum at (a, b) if the highest level curve that intersects the
constraint curve does so at this point. On the other hand, if the lowest level curve
that intersects the constraint curve does so at (a, b), then / achieves its constrained
minimum at this point.
20. Find the maximum and minimum distance from the origin to 34. Find the radius of the largest cylinder of height 6 in. that can
the ellipse 5x2 — 6xy + 5y2 — 4. be inscribed in an inverted cone of height H, radius R, and
21. Find the point on the plane 2x + y + z = 1 that is nearest to lateral surface area 250 in.2.
the origin. 35. In Problem 42 of Problem Set 11.7, you were asked to mini¬
22. Find the largest product of positive numbers x, y, and z such mize the ground state energy
that their sum is 24.
23. Write the number 12 as the sum of three positive numbers x, k2 ( 1 1 1 \
y, z in such a way that the product xy2z is a maximum.
24. A rectangular box with no top is to be constructed from 96 ft2
£fc,-!l=sb+?+?)
of material. What should be the dimensions of the box if it is subject to the volume constraint V = xyz = C. Solve the
to enclose maximum volume? problem using Lagrange multipliers.
25. The temperature T at point (x, y, z) in a region of space is
36. Modeling Problem A university extension agricultural service
given by the formula T — 100 — xy — xz — yz. Find the
concludes that, on a particular farm, the yield of wheat per
lowest temperature on the plane x + y + z = 10.
acre is a function of water and fertilizer. Let x be the number
26. A farmer wishes to fence off a rectangular pasture along the
of acre-feet of water applied, and y the number of pounds of
bank of a river. The area of the pasture is to be 3,200 yd2, and
fertilizer applied during the growing season. The agricultural
no fencing is needed along the river bank. Find the dimensions
service then concludes that the yield B (measured in bushels),
of the pasture that will require the least amount of fencing.
can be modeled by the formula B(x, y) — 500 + x2 -t- 2y2.
27. There are 320 yd of fencing available to enclose a rectangu¬ Suppose that water costs $20 per acre-foot, fertilizer costs $12
lar field. How should the fencing be used so that the enclosed per pound, and the farmer will invest $236 per acre for wa¬
area is as large as possible? ter and fertilizer. How much water and fertilizer should the
28. Use the fact that 12 oz is approximately 6.89tt in.3 to find the farmer buy to maximize the yield?
dimensions of the 12-oz Coke can that can be constructed us¬
37. How would the farmer of Problem 36 maximize the yield if
ing the least amount of metal. Compare your answer with an
the amount spent is $100 instead of $236?
actual can of Coke. Explain what might cause the discrepancy.
29. A cylindrical can is to hold 4jt in.3 of orange juice. The cost 38. Present post office regulations specify that a box (that is, a
per square inch of constructing the metal top and bottom is package in the form of a rectangular parallelepiped) can be
twice the cost per square inch of constructing the cardboard mailed parcel post only if the sum of its length and girth does
side. What are the dimensions of the least expensive can? not exceed 108 inches, as shown in Figure 11.50. Find the
30. Find the volume of the largest rectangular parallelepiped that maximum volume of such a package. (Compare your solution
can be inscribed in the ellipsoid here with the one you might have given to Problem 17, Section
4.6, page 246.)
42. Find the maximum of /(x, y, z) = xyz subject to x1 + y2 = 3 where p, q, and r are the proportions of blood types A, B, and
and y = 2z. O in the population. Given that p + q + r — 1, what is the
43. Maximize /(x, y, z) = xy + xz subject to 2x + 3z = 5 and largest value of PI
xy = 4.
© 48. A farmer wants to build a metal silo in the shape of a right cir¬
44. Minimize /(x, y, z) - 2x2+3y2+4z2 subject to x+y+z = 4 cular cylinder with a right circular cone on the top (the bottom
and x — 2y + 5z = 3. of the silo will be a concrete slab). What is the least amount
45. Modeling Problem A manufacturer is planning to sell a new of metal that can be used if the silo is to have a fixed volume
product at the price of $150 per unit and estimates that if x Vo?
thousand dollars is spent on development and y thousand dol¬ 49. Find the volume of the largest rectangular parallelepiped (box)
lars on promotion, then approximately that can be inscribed in the ellipsoid
320y 160x
-r H-t
y+2 x +4
a. If the manufacturer has a total of $8,000 to spend on the 50. Exploration Problem The method of Lagrange multipliers gives
development and promotion, how should this money be al¬ a constrained extremum only if one exists. If the method is
located to generate the largest possible profit? applied to optimizing /(x, y) = x + y subject to xy = 1,
b. Suppose the manufacturer decides to spend $8,100 instead the method yields two candidates for an extremum. Is one a
of $8,000 on the development and promotion of the new maximum and the other a minimum? Explain.
product. Estimate how this change will affect the maxi¬ In Problems 51-54, let Q = f(x, y) be a production function in
mum possible profit. which x and y represent units of labor and capital, respectively.
c. If unlimited funds are available, how much should the man¬ If p and q represent unit costs of labor and capital, respectively,
ufacturer spend on development and promotion to maxi¬ then C(x,y) — px -\- qy represents the total cost of production.
mize profit?
51. Use Lagrange multipliers to show that subject to a fixed pro¬
d. What is the Lagrange multiplier in part c? Your answer duction level Qo, the total cost is smallest when
should suggest another method for solving the problem in
part c. Solve the problem using this alternative approach.
— = — and Q = /(x, y) = Q0
46. Modeling Problem A jewelry box with a square base has an in¬
P 4
terior partition and is required to have volume 800 cm3 (see
Figure 11.51). (provided V/ ^ 0, and p 0, q 0). This is often referred
to as the minimum cost problem, and its solution is called the
least cost combination of inputs.
52. Show that the inputs x, y that maximize the production level
Q — /(x, y) subject to a fixed cost k satisfy
fx fy ... , .
— = — with px + qy = k
P 4
a. The material in the top costs twice as much as that in the a. Show that such a function is maximized with respect to the
sides and bottom, which in turn, costs twice as much as fixed cost px + qy = k when x = ak/p and y = fik/q.
the material in the partitions. Find the dimensions of the b. Where does the maximum occur if we drop the condition
box that minimize the total cost of construction. Does it a + f — 1? How does the maximum output change if k is
matter that you have not been told where the partitions are increased by 1 unit?
located?
54. Show that the cost function
b. Suppose the volume constraint changes from 800 cm3 to
801 cm3. Estimate the appropriate effect on the minimal C(x, y) = px + qy
cost.
47. Three alleles, A, B, and O, determine the four blood types A, is minimized subject to the fixed production level Axayt> = k,
B, O, and AB. The Hardy-Weinberg law says that the propor¬ with a + f = 1 when
tions of individuals in a population who carry two different
alleles is k /aqV k /fipY
P = 2 pq -p 2 pr + 2rq x = -a\Tp) , =
770 Chapter 11 Review
55. TORICAL QUEST Joseph La¬ For this Quest, we consider Lagrange’s work with solv¬
grange is generally acknowledged as ing algebraic equations. You are familiar with the quadratic
one of the two greatest mathemati¬ formula, which provides a general solution for any second-
cians of the eighteenth century, along degree equation ax2 + bx + c — 0, a ^ 0. Lagrange made
with Leonhard Euler (see HISTORI¬ an exhaustive study of the general solution for the first four
CAL QUEST #84 of the Supplemen¬ degrees. Here is what he did. Suppose you are given a general
tary Problems of Chapter 4). There algebraic expression involving letters a, b, c,...; how many
is a distinct difference in style between different expressions can be derived from the given one if the
JOSEPH LAGRANGE letters are interchanged in all possible ways? For example,
Lagrange and Euler. Lagrange has
1736-1813
been characterized as the first true an¬ from ab + cd we obtain ad + cb by interchanging b and d.
alyst in the sense that he attempted to This problem suggests another closely related problem,
write concisely and with rigor. On the other hand, Euler wrote also part of Lagrange’s approach. Lagrange solved general
using intuition and with an abundance of detail. Lagrange algebraic equations of degrees 2, 3, and 4. It was proved later
was described by Napoleon Bonaparte as "the lofty pyramid (not by Lagrange, but by Galois and Abel) that no general so¬
of the mathematical sciences," and followed Euler as the court lution for equations of degree greater than 4 can be found. Do
mathematician for Frederick the Great. He was the first to use some research and find the general solution for equations of
the notation fix) and f"(x) for derivatives. In this section we degrees 1, 2, 3, and 4.
were introduced to the method of Lagrange multipliers, which
provide a procedure for constrained optimization. This method
was contained in a paper on mechanics that Lagrange wrote *From Men of Mathematics by E. T. Bell, Simon & Schuster, New York,
when he was only 1 9 years old.* ■ 1937, p. 165.
Proficiency Examination 18. State the following basic properties of the gradient,
a. constant rule b. linearity rule
CONCEPT PROBLEMS
c. product rule d. quotient rule
1. What is a function of two variables? e. power rule
2. What are the domain and range of a function of two variables? 19. Express the directional derivative in terms of the gradient.
3. What is a level curve of the function defined by f(x, y)? 20. State the optimal direction property of the gradient (that is, the
4. What do we mean by the limit of a function of two variables? steepest ascent and steepest descent).
5. State the following properties of a limit of functions of two 21. State the normal property of the gradient.
variables. 22. Define the normal line and tangent plane to a surface S at a
point P0.
a. scalar rule b. sum rule
23. Define the absolute extrema of a function of two variables.
c. product rule d. quotient rule
24. Define the relative extrema of a function of two variables.
6. Define the continuity of a function defined by fix, y) at a
25. What is a critical point of a function of two variables?
point (*0, yo) in its domain and continuity on a set S.
26. State the second partials test.
7. If z = /(x, y), define the first partial derivatives of / with
27. State the extreme value theorem for a function of two vari¬
respect to x and y.
ables.
8. What is the slope of a tangent line to the surface defined by
28. What is the least squares approximation of data, and what is a
z — f(x, y) that is parallel to the xy-plane at a point P0 on /?
regression line?
9. If z — fix, y), represent the second partial derivatives. 29. State Lagrange’s theorem.
10. If z = /(x, y), what are the increments of x, y, and z? 30. State the procedure for the method of Lagrange multipliers.
11. What does it mean for a function of two variables to be differ¬
entiable at ix0, yo)? PRACTICE PROBLEMS
12. State the incremental approximation of fix, y). 31. If fix, y) = sin-1 xy, verify that fxy = fyx.
13. Define the total differential of z = fix, y). 32. Let w = x2y + y2z, where x — t sin t, y — t cos t, and z = 2t.
dw
14. State the chain rule for a function of one parameter. Use the chain rule to find —, where t = n.
dt
15. State the chain rule for a function of two independent param¬ 33. Let fix,y,z) — xy + yz + xz, and let P0 denote the point
eters. (1,2, —1).
16. Define the d rectional derivative of a function defined by a. Find the gradient of / at Pq.
Z = fix, y). b. Find the directional derivative of / in the direction from
17. Define the gradient V/(x, y). P0 toward the point Q(— 1, 1, —1).
Chapter 11 Review 771
c. Find the direction from P0 in which the directional deriva¬ Show that each limit in Problems 18 and 19 does not exist.
tive has its largest value. What is the magnitude of the 3
y xJy-
largest directional derivative at P01 18. lim 19. lim
(j:,y)->(0,0) x- + yJ (j;.y)->(0,0) Xft + y4
34. Show that the function defined by
Find the derivatives in Problems 20-23 using the chain rule. You
may leave your answers in terms of x, y, t, u, and v.
if (x,y) # (0. 0) dz
fix, y) x3 + y3 20. Find —, where z = -xy + v3, and x = -'it2, y = 1 + r3.
0 dt ’ J
if (x, y) = (0,0)
21. Find —, where z — xy + v2, and x = e'r1, y = tan/.
is not continuous at (0, 0). dt
dZ Z
35. If fix, y) = In find fx, fy, fyy, and fxy. 22. Find — and —, where z = x2 - y2, and x = u + 2v,
du dv
36. Show that if fix, y, z) = x2y + y2z + z2x, then y = u — 2v.
39. Use the method of Lagrange multipliers to find the maximum 28. fix, y) = tan-1 xy 29. fix, y) = sin-1 xy
and minimum values of the function 30. fix, y) = x2 4- y3 — 2xy2 31. fix, y) — e*2+y2
f(x, y) — x2 + 2_v2 4- 2x + 3
32. /(x,y)=xlny 33. /(x, y) = / sin(cos/)uf/
subject to the constraint x2 + y2 = 4.
Find equations for the tangent plane and normal line to the sur¬
40. Find the largest and smallest values of the function
faces given in Problems 34-36 at the prescribed point.
fix, y) — x2 — 4 y2 + 3x + 6y
34. x2y3z = 8 at /50(2, —1, —2)
on the region defined by — 2 < x < 2, 0 < y < 1.
35. x3 4-2xy2 - 7x3 + 3y 4- 1 = 0 at P0i\, 1, 1)
-4
36. z at Pq(1, 1,-1)
Supplementary Problems 2 + x2 + y2
Describe the domain of each function given in Problems 1-4. Find all critical points of fix, y) in Problems 37-42 and classify
each as a relative maximum, a relative minimum, ora saddle point.
1. f{x, y) = v716 -x2 - y2 2. fix, y) - 37. fix, y) = x2 — 6x + 2y2 + 4y — 2
38. fix, y) = x3 + y3 - 6xy
3. /(x,y) = sin 1 x + cos 1 y 4. fix,y) tan
39. fix, y) = (x - 1) (y - l)(x + y - 1)
Find the partial derivatives fx and fy for the functions defined in
40. fix, y) = x2 + y3 4 6xy - lx - 6y
Problems 5-10.
41. fix, y) = x3 + y3 4- 3x2 - 18y2 + 81y + 5
x 2 — yz2
5. fix, y) = 6. fix, y) = x3e3yl(2x) 42. fix, y) = sin(x+y)+sinx4-siny for 0 < x < n, 0 < y < n
X +y
(See Figure 11.52.)
xy
7. fix, y) = x2y 4- sin - 8. fix, y) = In
x + 2y
y/x2 4- y2 if x > 0
13. z = fix, y) -
— |y| if x < 0
c = 0, c = l, c = — 1
14. fix, y, z) = x2 4- y2 + z2; c — 16, c = 0, c = -25 Figure 11.52 Graph of fix, y) = sin(x 4- y) + sinx 4- sin y
2 2
15. fix, y,z)=x2 + y+ ^-;c=l,c = 2
In Problems 43-46, find the largest and smallest values of the func¬
Evaluate the limits in Problems 16 and 17, assuming they exist. tion f on the specified closed, bounded set S.
44. f(x,y) — x2 -F 2 v2 — x — 2y; S is the triangular region with 66. Modeling Problem A plate is heated in such a way that its
vertices (0, 0), (2, 0), (2, 2) temperature at a point (x, y) measured in centimeters on the
plate is given in degrees Celsius by
45. /(x, y) = x2 + y2 — 3y; S is the disk x2 + y2 < 4
64
46. /(x, y) = 6x — x2 + 2xy — _y4; S is the square 0 < x < 3, T(x, y)
0 < y < 3 x2 + y2 + 4
cl Z a. Find the rate of change in temperature at the point (3, 4) in
47. Use the chain rule to find — if z = x2 —3xy2; x = 2t,y = t2.
the direction 2i + j.
b. Find the direction and the magnitude of the greatest rate of
dz
48. Use the chain rule to find — if z — x In v; x = 2t, y = e . change of the temperature at the point (3,4).
dt
67. Modeling Problem The beautiful patterns on the wings of but¬
49. Let z — we"2'"2, where u — 2x2 + 3y2 and v — 3x2 — 2y2. terflies have long been a subject of curiosity and scientific
dz dz study. Mathematical models used to study these patterns of¬
Use the chain rule to find — and —.
3x dy ten focus on determining the level of morphogen (a chemical
that affects change). In a model dealing with eyespot patterns,
50. Use implicit differentiation to find — and —, where x, y, a quantity of morphogen is released from an eyespot and the
dx dy
and z are related by the equation x3 + 2xz — yz2 — z3 — 1. morphogen concentration t days later is modeled by
52. Find the slope of the level curve of xey = 2, where x = 2. where r measures the radius of the region on the wing affected
by the morphogen, and k and y are positive constants.*
53. Find the equations for the tangent plane and normal line to the
surface z — sinx + exy + 2y at the point 7*0(0, 1,3). a. Find tm so that dS/dt = 0. Show that the function Sm(t)
formed from S(r, t) by fixing r has a relative maximum
54. The electric potential at each point (x, y) in the disk
at tm. Is this the same as saying that the function of two
x2 + y2 < 4 is V = 2(4 — x2 — y2)~1/2 volts. Draw the
variables S(r, t) has a relative maximum?
equipotential curves V = c for c = \/2, 2/V3, and 8.
b. Let M(r) denote the maximum found in part a; that is,
55. Let f(x, y, z) x3y + y3z + z3x. Find a function g(x, y, z) M(r) = S(r, tm). Find an expression for M in terms of
df df df
such that-1-1- x3 + y3 + z3 + 3g(x, y, z). z = 0 +4 ykr2)1!2.
3x dy dz dM
c. Show that- < 0 and interpret this result.
x y du du dr
56. Let m = sin —F In Show that y-F x— =0.
y x dy dx 68. Modeling Problem Certain malignant tumors that do not
respond to conventional methods of treatment (surgery,
57. Let w = ln(l + x2 + y2) 2 tan 1 y, where x = ln(l + t2)
dw chemotherapy, etc.) may be treated by hyperthermia, a pro¬
and y e . Use the chain rule to find —. cess involving the application of extreme heat using mi¬
dt
crowave transmission (see Figure 11.53). For one particular
58. Let /(x,y) = tan'! —. Find the directional derivative of f kind of microwave application used in such therapy, the tem¬
x
at (1,2) in the direction that makes an angle of | with the perature at each point located r units from the central axis of
positive x-axis. the tumor and It units inside it is modeled by the formula
59. Let /(x,y) = yx. Find the directional derivative of / at T(r,h) = Ke-pr\e~qh - e~sh]
7*o(3, 2) in the direction toward the point <2(1, 1).
where K, p, q, and s are positive constants that depend on the
60. According to postal regulations, the largest cylindrical can that properties of the patient’s blood and the heating application.1
can be sent has a girth (27rr) plus length l of 108 inches. What
is the largest volume cylindrical can that can be mailed?
61. Let f(x, y, z) = z(x - y)5 + xy2z3.
a. Find the directional derivative of f at (2, 1, —1) in the
direction of the outward normal to the sphere
x2 + y2 -F z2 = 6.
b. In what direction is the directional derivative at (2, 1,-1)
largest?
62. Find positive numbers x and y for which xyz is a maximum,
given thatx+y+z = 1. Assume that the extreme value exists.
63. Maximize /(x, y, z) = x2yz given that x, y, and z are all pos¬
itive numbers and x -F y -F z = 12. Assume that the extreme Figure 11.53 Hyperthermia treatment
value exists.
*J. D. Murray, Mathematical Biology, 2nd edition. Springer-Verlag. New
,, ,*• 7, , ^ dz York, 1993, p. 464.
64. If z = f(xx — yU, evaluate y-Fx —.
dx dy
! “Heat Therapy for Tumors,” by Leah Edelstein-Keshet. UMAP Modules
65. Find the shortest distance from the origin to the surface 1991: Tools for Teaching, Consortium for Mathematics and Its Applica¬
z2 = 3 -F xy. Assume the extreme value exists. tions. Inc., Lexington, MA. 1992, pp. 73-101.
Chapter 11 Review 773
a. At what depth inside the tumor does the maximum temper¬ 83. The geometric mean of three positive numbers x, y, z is
ature occur? Express your answers in terms of K, p, q, G = (xyz)1/3 and the arithmetic mean is A = |(x + y + z).
and s. Use the method of Lagrange multipliers to show that
b. The article on which this problem is based discusses the G(x, y, z) < A(x, y, z) for all x, y, z.
physiology of hyperthermia in addition to raising several 84. Liquid flows through a tube with length L centimeters and in¬
other interesting mathematical issues. Read this article and ternal radius r centimeters. The total volume V of fluid that
discuss assumptions made in the model. flows each second is related to the pressure P and the viscosity
69. Modeling Problem The marketing manager for a certain com¬ 7r Pr4
a of the fluid by the formula V = What is the maxi¬
pany has compiled the following data relating monthly adver¬ 8 aL
tising expenditure and monthly sales (in units of $1,000). mum error that can occur in using this formula to compute the
viscosity a, if errors of ±1% can be made in measuring r and
Advertising 3 4 7 9 10 L, ±2% in measuring V, and ±3% in measuring PI
Sales 78 86 138 145 156 85. Suppose the functions / and g have continuous partial deriva¬
tives and satisfy
a. Plot the data on a graph and find the least squares line.
9f = dg 9/ =_9g
b. Use the least squares line to predict monthly sales if the and
3x 3y 3y 3x
monthly advertising expenditure is $5,000.
70. Find fxx - fxy + fyy, where fix, y) = x2y3 + x3y2. These are called the Cauchy-Riemann equations.
a. Show that level curves of / and g intersect at right angles
71. Find fxyz, where /(x, y, z) = cos(x2 + y3 + z4).
provided V/ ^ 0 and Vg ^ 0.
72. Let z — fix, y), where x — t + cos t and y = e'.
b. Assuming that the second partials of / and g are continu¬
a. Suppose fx( 1, 1) = 4 and fy(\, 1) = —3. Find — when ous, show that / and g satisfy Laplace's equations
dt
t = 0. fxx + /vy = 0 and gxx + gyy = 0
3z
b. Suppose fx(0,2) = —1 and /v(0,2) = 3. Find — and 86. Show that if z = fir, 6), where r and 9 are defined
d?'
as functions of x and y by the equations x = rcos#,
— at the point where r = 2, 0 — f, and x = r cos 9,
30 F 2 32z 32z
y = rsin#, then the equation —- + -—z=0 becomes
y = r sin#. 3x2 3y2
73. Suppose / has continuous partial derivatives in some region 3 2z 1 3 2z 1 3z
—- -|—-—- -|-=0. This is Laplace s equation in po-
D in the plane, and suppose f(x, y) = 0 for all (x, y) in D. If 3r2 r2 392 r dr
lar coordinates.
(1, 2) is in D and fx( 1,2) — 4 and fy( 1,2) = 6, find dy/dx
when x = 1 and y - 2. 87. A circular sector of radius r and central angle 9 has area
A = \rl9. Find 9 for the sector of fixed area A0 for which
74. Suppose V/(x, y, z) is parallel to the vector xi + yj + zk for
the perimeter of the sector is minimized. Use the method of
all (x, y, z). Show that /(0, 0, a) = /(0, 0, —a) for any a.
Lagrange multipliers and assume that the minimum exists.
75. Find two unit vectors that are normal to the surface given by
88. For the production function given by Qix, y) = xayb, where
z = /(x, y) at (0, 1), where /(x, y) = sinx + exy + 2y.
a > 0 and b > 0, show that
76. Let /(x, y) = 3(x - 2)2 - 5(y + l)2. Find all points on
90
the graph of / where the tangent plane is parallel to the plane
+ y^r
3y = (a + b)Q
2x + 2 y-z = 0.
77. Let z be defined implicitly as a function of x and y by the In particular, if b = 1 — a with 0 < a < 1,
d2z
equation cos(x + y) + cos(x + z) = 1. Find -—— in terms of 9Q
3y3x x
3x
x, y, and z.
78. Suppose F and F' are continuous functions of t and that 89. The diameter of the base and the height of a closed right cir¬
F'(t) = C. Define / by /(x, y) = F(x2 + y2). Show that the cular cylinder are measured, and the measurements are known
direction of V/(u, b) is the same as the direction of the line to have errors of at most 0.5 cm. If the diameter and height
joining (a, b) to (0. 0). are taken to be 4 cm and 8 cm, respectively, find bounds for
the propagated error in
79. Let f(x, y) = 12x_l + 18y_1 + xy, where x > 0, y > 0.
a. the volume V of the cylinder
How do you know that / must necessarily have a minimum in
the region x > 0, y > 0? Find the minimum. b. the surface area S of the cylinder
90. A right circular cone is measured and is found to have base
80. Let fix, y) = 3x4 - 4x2y + y2. Show that / has a minimum
radius r = 40 cm and altitude h — 20 cm. If it is known that
at (0, 0) on every line y = mx that passes through the origin.
each measurement is accurate to within 2%, what is the maxi¬
Then show that / has no relative minimum at (0, 0).
mum percentage error in the measurement of the volume?
In Problems 81-83, you may assume the required extremum exists.
91. An elastic cylindrical container is filled with air so that the ra¬
81. Find the minimum of x2 + y2 + z2 subject to the constraint dius of the base is 2.02 cm and the height is 6.04 cm. If the
ax + by + cz — 1 (with a ^ 0, b £ 0, c ^ 0). container is deflated so that the radius of the base reduces to
82. Suppose 0 < a < 1 and x > 0, y > 0. Find the maximum of 2 cm and the height to 6 cm, approximately how much air has
x“y'~a subject to the constraint ax + (1 — a)y = 1. been removed? (Ignore the thickness of the container.)
774 Chapter 11 Review
92. Suppose f is a differentiable function of two variables with the partial derivatives Dy and D- to 0 gives the critical point
fx and fy also differentiable, and assume that fxx, fyy and fxy y = 0, z = 0.5. Solving for x on the paraboloid gives the
are continuous. The second directional derivative of / at the points Qi(VT5, 0, 0.5) and Q2(—\/T5, 0, 0.5). These points
point (x, v) in the direction of the unit vector u = fli + £>j is are not minimal because (0, 1,0) is closer. Explain what is go¬
defined by D2/(x, y) = Du[Duf(x,y)]. Show that ing on here. Our thanks to Herbert R. Bailey, who presented
this problem in The College Mathematics Journal (“ ‘Hidden’
Dif(x, y) = a2fxx{x, y) + 2 abfxy(x, y) + b2fyy(x, y) Boundaries in Constrained Max-Min Problems,” May 1991,
Vol. 22, p. 227).
93. A capsule is a cylinder of radius r and length l, capped on each
end by a hemisphere. Assume that the capsule dissolves in the 96. It can be shown that if z = /(x, y) has the least surface area
stomach at a rate proportional to the ratio R — S/V, where V of all surfaces with a given boundary, then
is the volume and S is the surface area of the capsule. Show
that
(1 + Zy)Zxx ~ ZZxy + (1 + Z2x)Zyy = 0
Z.
Figure 11.54 7
X
This project is to be done in groups of three or four students. Each group will submit a
A
single written report.
friend of yours named Maria is studying the causes of the continuing expansion of
deserts (a process known as desertification). She is working on a biological index
of vegetation disturbance, which she has defined. By seeing how this index and
other factors change through time, she hopes to discover the role played in desertifica¬
tion. She is studying a huge tract of land bounded by a rectangle. This piece of land
surrounds a major city but does not include the city. She needs to find an economi¬
cal way to calculate for this piece of land the important vegetation disturbance index
J(x, y).
Maria has embarked upon an ingenious approach of combining the results of pho¬
tographic and radar images taken during flights over the area to calculate the index J.
She is assuming that J is a smooth function. Although the flight data do not directly
reveal the values of the function J, they give the rate at which the values of J change as
the flights sweep over the landscape surrounding the city. Maria’s staff has conducted
numerous flights, and from the data she believes she has been able to find actual for¬
mulas for the rates at which J changes in the east-west and north-south directions. She
has given these functions the names M and N. Thus, M(x, y) is the rate at which J
changes as one sweeps in the positive x-direction and N(x, y) is the corresponding
rate in the y-direction. Maria shows you these formulas:
Mathematics in its pure form, as
arithmetic, algebra, geometry, and the M(x,y) = 3.4ex{y — 7.8)2 and N(x, y) = 22 sin(75 — 2*y)
applications of the analytic method, as
Convince her that these two formulas cannot possibly be correct. Do this by show¬
well as mathematics applied to matter and
ing her that there is a condition that the two functions M and N must satisfy if they
force, or statics and dynamics, furnishes
the peculiar study that gives to us, whether are to be the east-west and north-south rates of change of the function J and that her
as children or as men, the command of na¬ formulas for M and N do not meet this condition. However, show Maria that if she can
ture in this its quantitative aspect; math¬ find formulas for M and N that satisfy the condition that you showed her, it is possible
ematics furnishes the instrument, the tool to find a formula for the function J from the formulas for M and N.
of thought, which we wield in this realm.
-W. T. Harris
Psychological Foundations of Education
(New York, 1898), p.325.
* Marcus S. Cohen, Edward D. Gaughan. R. Arthur Knoebel, Douglas S. Kurtz, and David J. Pengelley,
"Priming the Calculus Pump: Innovations and Resources," MAA Notes 17(1991).
775
■*
CONTENTS
12.1 Double Integration over Rectangular
Regions
Multiple Integration
Definition of the double integral
Properties of double integrals
Volume interpretation
Iterated integration
An informal argument for Fubini’s
theorem
12.2 Double Integration over Nonrectangular PREVIEW
Regions The single integral f* f(x) dx introduced in Chapter 5 has many uses, as we have
Double Integrals over Type I and
seen. In this chapter, we will generalize the single integral to define multiple in¬
Type II regions
tegrals in which the integrand is a function of several variables. We will find that
More on area and volume
Choosing the Order of Integration
multiple integration is used in much the same way as single integration, by “adding”
in a Double Integral small quantities to define and compute area, volume, surface area, moments, cen¬
troids, and probability.
12.3 Double Integrals in Polar Coordinates
Change of variables to polar form
Area and volume in polar form
12.4 Surface Area PERSPECTIVE
Definition of surface area
Surface area projections
What is the volume of a doughnut (torus)? Given the joint probability density func¬
Area of a surface defined tion for the amount of time a typical shopper spends shopping at a particular store
parametrically and the time spent in the checkout line, how likely is it that a shopper will spend
no more than 30 minutes altogether in the store? If the temperature in a solid body
12.5 Triple Integrals
Definition of the triple integral is given at each point (x,y,z) and time t, what is the average temperature of the
Iterated integration body over a particular time period? Where should a security watch tower be placed
Volume by triple integrals in a parking lot to ensure the most comprehensive visual coverage? We will answer
12.6 Mass, Moments, and Probability Density these and other similar questions in this chapter using multiple integration.
Functions
Mass and center of mass
Moments of inertia
Joint probability density functions
12.7 Cylindrical and Spherical Coordinates
Cylindrical coordinates
Integration with cylindrical
coordinates
Spherical coordinates
Integration with spherical
coordinates
12.8 Jacobians: Change of Variables
Change of variables in a double
integral
Change of variables in a triple
integral
Chapter 12 Review
Group Research Project: Space-Capsule
Design
777
778 Chapter 12 Multiple Integration
Recall that in Chapter 5, we defined the definite integral of a single variable f{x)dx
n
Y f(x%)Axk, where a = xo < x\ < X2 < ■ ■ ■ < xn = b
k=t
are points in a partition of the interval [a, b] and x% is a representative point in the
subinterval [x*_i, .**]. We now apply the same ideas to define a definite integral of two
Step 1. Partition the interval a < x < b into m subintervals and the interval
c < y < d into n subintervals. Using these subdivisions, partition the rect¬
angle R into N = mn cells (subrectangles), as shown in Figure 12.1. Call
this partition P.
Step 2. Choose a representative point (x£, y%) from each cell in the partition of the
rectangle. Form the sum
When this process is applied to the Riemann sum and the norm decreases indefi¬
nitely to zero, we write
N
lim
IIUKO E/<***•
k=1
If this limit exits, its value is called the double integral of / over the rectangle R.
Double Integral If / is defined on a closed, bounded rectangular region R in the xy-plane, then
the double integral of / over R is defined by
JJ
R
f(x,y)dA lim
IIUI-O
£
N
k=\
/at, rf)AA,
provided this limit exists, in which case, / is said to be integrable over R.
12.1 Double Integration over Rectangular Regions 779
means that for any e > 0, there exists a 8 > 0 such that
N
/ - J2 < €
k=1
N
whenever ^ /(x£, yl)AAk is a Riemann sum whose norm satisfies | P|| < 8. In this
;t=i
process, the number of cells N depends on the partition ft, and as || P —► 0, it follows
that N ^ oo.
It can be shown that if f(x, y) is continuous on a rectangle ft, then it must be inte¬
grate on R, although it is also true that certain discontinuous functions are integrate
as well. Moreover, it can also be shown that if the limit that defines the definite integral
exists, then it is unique in the sense that the same limiting value results no matter how
the partitions and subinterval representatives are chosen. We will extend the definition
of the definite integral to nonrectangular regions in Section 12.2. However, issues in¬
volving the existence and uniqueness of the definite integral are generally dealt with in
a course in advanced calculus.
Double integrals have many of the same properties as single integrals. Three of these
properties are contained in the following theorem.
Dominance rule: If f(x, y) > g(x, y) throughout a rectangular region ft, then
VOLUME INTERPRETATION
t f(xtyt)
If g(x) > 0 on the interval [a, b], the single integral f g{x)dx can be interpreted
Area of
Rk is AAk. as the area under the curve y = g(x) over [a,b]. The double integral JJ f(x,y)dA
y%) in Rk R
has a similar interpretation in terms of volume. To see this, note that if f(x, y) > 0
on the rectangular region ft and we partition ft, then the product /(xf yk)AAk is the
Figure 12.2 The approximating
volume of a parallelepiped (a box) with height f(xf yk ) and base area A Ak, as shown
parallelepiped has volume
A14 = f(x*k,y*k)AAk in Figure 12.2.
780 Chapter 12 Multiple Integration
E /u;,y;)A At
provides an estimate of the total volume under the surface z = fix, y) over R, and
if / is continuous, we expect the approximation to improve by using more refined
partitions (that is, more rectangles with smaller norm). Thus, it is natural to define the
total volume under the surface as the limit of Riemann sums as the norm tends to 0.
That is, the volume under z — f(x, y) over the domain R is given by
y
V = ]P\lfi0J2^x^y^AAk
k=\
ff fix, y) dA
^J
=
Evaluate (2 — y) dA, where R is the rectangle in the xy-plane with vertices (0, 0),
Solution
Because z = 2 — y satisfies z > 0 for all points in R, the value of the double integral is
the same as the volume of the solid bounded above by the plane z = 2 — y and below
by the rectangle R. The solid is shown in Figure 12.4. Looking at it endwise, it has a
triangular cross section of area B and has length 3. We use the formula V — Bh.
J[(2-y)dA=6
12.1 Double Integration over Rectangular Regions 781
ITERATED INTEGRATION
As with single integrals, it is often not practical to evaluate a double integral even over
a simple rectangular region by using the definition. Instead, we will compute double
integrals by a process called iterated integration, which is like partial differentiation
in reverse.
Suppose fix, y) is continuous over the rectangle R: a < x < b, c < y < d.
Cd
Then we write / f(x, y) dy to denote the integral obtained by integrating /(x, y)
with respect to y over the interval [c, d] with x held constant. The integral obtained
, fd
by this partial integration is a function of x alone, G(x) — I f(x,y)dy, which we
Similarly, if we integrate f(x,y) first with respect to x over [a, b] holding y constant,
and then with respect to y over [c, d], we obtain the iterated integral
fix, y)dA
can be evaluated in terms of iterated integrals, was proved in a more general form by
the Italian mathematician Guido Fubini (1879-1943) in 1907.
JI fix, y)dA
fix, y) dx dy
“ 11m y) dy dx
i-cl
rb
/ fix, y) dy dx or / fix,y)dx dy
JC l Ja
t _ t
Limits of y (variable inside brackets) Limits of x (variable inside brackets)
Note that Fubini’s theorem applies only when a, b, c, and d are constants.
782 Chapter 12 Multiple Integration
Proof We will provide an informal, geometric argument at the end of this section.
The formal proof may be found in most advanced calculus textbooks. □
Let us see how Fubini’s theorem can be used to evaluate double integrals. We
begin by taking another look at Example 1.
the rectangle with vertices (0, 0), (3, 0), (3, 2), and (0, 2).
Solution
The region of integration is the rectangle 0<x<3,0<y<2 (see Example 1 and
Figure 12.4). Thus, by Fubini’s theorem, the double integral can be evaluated by the
iterated integral:
r 21 2
-L
’3 r 4
-[[ 4 - - - (0)
= / 2dx = 2x\l = 6
Jo
which is the same as the result obtained geometrically in Example 1.
Evaluate JJ x2y5 dA, where R is the rectangle 1 < x < 2, 0 < y < 1, using an
R
iterated integral with
Solution
a. The graph of the surface z = x2y5 over the rectangle is shown in Figure 12.5a, and
the rectangular region of integration in Figure 12.5b.
TECHNOLOGY NOTE
Using technology for multiple integrals offers no special difficulties. You simply integrate with respect to one variable and
then with respect to the other variable. You must be careful, however, to input properly the correct limits ot integration
for each integral. Notice that for / / x2y2 dy dx, most calculators and software programs require the following syntax:
Ji Jo .
integration operator, function, variable of integration, lower limit of integration, upper limit of integration.
/((/(i2> ? ^ )
function
inner variable outer variable
of integration of integration
The result of this operation is shown in Figure 12.6 (compare with Example 3).
7/18
1 [^ | 7/Y4 1 jo [i (x2y5)dX(iy
J <I<x^2yA5 ,y ta <xA2yA5,xJ.lJ.2>J.y
HrtIN
tour
FUNC 1/20
MftIN F;HD rtUTO FUNi l/SO
f
Jo LJo
x cosxy dx dy
The inner integral requires integration by parts. However, integrating with respect to y
first is much simpler:
r jf/2 ~x sinxyl
•w/2
r /r1 xcos xv dy dx - / -- dx
Jo X
Jo
en/2 n/2
(sin x — sin 0) dx = — cos x = 1
=L '
784 Chapter 12 Multiple Integration
We can make Fubini’s theorem plausible with a geometric argument in the case where
f(x,y) > 0 on R. If// f(x,y)dA is defined on a rectangle R: a < x < b,
R
c < y < d, it represents the volume of the solid D bounded above by the surface
^ = fix, y) and below by the rectangle R. If A(yJ) is the cross-sectional area perpen¬
dicular to the y-axis at the point y£, then A(y£)Ayk represents the volume of a “slab”
that approximates the volume of part of the solid D, as shown in Figure 12.7.
Solid D
area A(y*k)
By using a limit to “add up” all such approximating volumes, we obtain the vol¬
ume, V, of the entire solid D; that is,
JJ f{x,y)dA=V = \imJ^A{y*k)Ayk
R k=1
The limit on the right is just the integral of A(y) over the interval c < y < d, where
A{y) is the area of a cross section with fixed y. In Chapter 5, we found that the area
A(y) can be computed by the integral
fb
A(y) = I f(x, y) dx Integration with respect to x (y is fixed)
Ja
We can now make this substitution for A{y) to obtain
JJ f(x,y)dA = V = hmJ2A(y*k)^yk
= Jc A(y)dy
■rtf fix, y) dx
a O’)
dy Substitution
JJ f(x,y)dA=J J fix, y) dy dx
can be justified in a similar fashion (you are asked to do this in Problem 46). Thus, we
have
rd r nb
/
Jc
/ fix,y)dx
L Ja dyssfff(x,y)dA = Ja \_Jc fix^y)dy dx
12.1 Double Integration over Rectangular Regions 785
Use an appropriate volume formula to evaluate the double integral 19 . Jj sin(x + y)dA; R: 0 < x < f, 0 < y < \
given in Problems 7-12.
■If
R
R:0 <x <2,0 <y <4 R:2 <x <5,1 <y <3
Find the volume of the solid bounded below by the rectangle R in
the xy-plane and above by the graph of z = f{x,y) in Problems
21-31.
21. f{x, y) = 2x + 3y; R: 0 < x < 1, 0 < y < 2
22. fix, y) = 5x + 2y\ R: 0 < x < 1, 0 < y < 2
’•//«- y)dA\
10-//<4-
2y) dA;
y
xy
27. fix, y) = - = ; R: 0 < x < 1,0 < y < 1
^/x2 + y2 + 1
28. fix, y) = xexy; R: 0 < x < 1, 0 < y < ln3
29. fix, y) = ix + y)5; R: 0 < x < 1,0 < y < 1
30. fix, y) = yfT+y\ /?: 0 <x < 1,0 < y < 1
31. fix,y) = x cosy + y sinx; /?:0<x<|,0<><|
Q 32. WHAT DOES THIS SAY? Discuss the definition of the dou¬
ble integral.
33. WHAT DOES THIS SAY? Explain how Fubini’s theorem is
used to evaluate double integrals.
34. Modeling Problem Suppose R is a rectangular region within the
boundary of a certain national forest that contains 600,000
trees per square mile. Model the total number of trees, T,
per square mile in the forest as a double integral. Assume that
x and y are measured in miles.
35. Modeling Problem Suppose mass is distributed on a rectangu¬
lar region R in the xy-plane so that the density (mass per unit
area) at the point (x, y) is Six, y). Model the total mass as a
double integral.
Use iterated integration to compute the double integrals in Prob¬ 36. Modeling Problem Let R be a rectangular region within the
lems 13-20 over the specified rectangle. boundary of a certain city defined by R: -3 < x < 2,
_1 < y < 1, where units are in miles and (0, 0) is the city
13. JI x2y dA\ R\l<x<2,0<y<l center. Assume the population density is 12e~0 07' thousand
R people per square mile, where r = yjx2 + y2. Model the total
population of the region of the city as a double integral, but do
14. JI dA, R: 0 < x < 1, 0 < y < 1 not evaluate the integral.
786 Chapter 12 Multiple Integration
In Problems 37-40, evaluate the integral. Note that one order of For this Historical Quest, write several paragraphs about
integration may be considerably easier than the other. the nature of differential projective geometry.
40. Evaluate Jj xexy dA, where R is the rectangle 0 < x < 1, JI f(x,y)dA fix, y) dy dx
R R
1 < y < 2.
Hint: Modify the argument given in the text by taking cross-
41. Explain why //<(4 — x2 — y2)dA > 2, where R is the rec¬ sectional areas perpendicular to the x-axis.
47. Counterexample Problem Show that the iterated integrals
tangular domain in the plane given by 0 < x < 1, 0 < y < 1.
42. Use a grid with 16 cells to approximate the volume of the solid nl
lying between the surface f{x, y) — 4-x2-y2 and the square ——~~x dx dy
(x + y)3
region R given by0<x<l,0<y<l.
43. Use a grid with 16 cells to approximate the volume of the solid have different values. Why does this not contradict Fubini’s
lying between the surface fix, y) = x2+y2 +1 and the square theorem?
region R given by 0 < x < 1, 0 < y < 1. 48. Exploration Problem You want to evaluate
44. I HISTORICAL QUEST Guido ' Fu-
bini taught at the Institute for Advanced y sin(xy) sin2(7ry)
Study in Princeton. He was nicknamed
the "Little Giant," because of his small
body but large mind. Even though over the rectangle D: 0 < x < 7r,0 < y < Which order of
the conclusion of Fubini's theorem was integration is easier?
known for a long time and success¬
fully applied in various instances, it was
not satisfactorily proved in a general GUIDO FUBINI
setting until 1907. His most impor¬ 1879-1943
tant work was in differential projective
geometry. In 1938 he was forced to leave Italy because of the
Fascist government, and he immigrated to the United States. ■
Let f{x, y) be a function that is continuous on the region D that can be contained in a
rectangle R. (See Figure 12.8.)
Define the function F(x, y) on R as /(x, y) if (x, y) is in D and 0 otherwise. That
is,
The function F(x, y) may have discontinuities on the boundary of D, but if f(x, y)
is continuous on D and the boundary of D is fairly “well behaved,” then it can be
shown that j'f F(x, y)dA exists and hence that f(x, y)dA exists. This proce-
R
D
dure is valid for the type I and type II regions we discuss next, although a general
notion of what constitutes a “well-behaved” boundary is a topic for advanced calculus.
A type I, or vertically simple region, D, in the plane is a region that can be described
by the inequalities
where h \ (y) and h2(y) are continuous functions of y on [c, d]. Vertically and horizon¬
tally simple regions are illustrated in the following box.
A type I region is the set of all points A type II region is the set of all points
(x, y) such that for each fixed x be¬ (*, y) such that for each fixed y be¬
tween x = a and x = b, the vertical tween y — c and y = d, the horizontal
line segment gi(x) < y < g2(x) lies line segment h\(y) < x < h2(y) lies
in the region. in the region.
we have
because for each fixed x in the interval [a, b], F(x, y) — 0 if y < ^(x) and also if
y > 82CO and F(x, y) = f(x, y) for gi(x) < y < g2(x). Therefore,
b r rd
II f(x,y)dA = II F(x,y)dA = £ Fix, y) dy dx
D] R
nglM
fix, y) dy dx
'a J g\(x)
■
iSnx'y)dA=[ Lf
d2
F(x, y) dx dy
d rh2(y)
-n 'c Jh\iy)
fix, y) dx dy
Evaluate f f 160xy3dydx.
JO Jx2
Solution
[lOx4 — 4xI0]|('
= 6
12.2 Double Integration over Nonrectangular Regions 789
When using Fubini’s theorem for nonrectangular regions, it helps to sketch the
region of integration D and to find equations for all boundary curves of D. Such a
sketch often provides the information needed to determine whether D is a type 1 or
type II region (or neither, or both ) and to set up the limits of integration of an iterated
integral.
(x + y)d A
Solution
a. To set up the limits of integration in the iterated integral, we draw the graph as
shown in Figure 12.9 and note that for fixed x, the variable y varies from y = 0 (the
x-axis) to the line y = 2x. These are the limits of integration for the inner integral
(with respect to y first). The outer limits of integration are the numerical limits of
integration for x; that is, x varies between x = 0 and x = 1.
(x + y)dA (x + y) dy dx
= f 4x2dx = f x3|^Io = |
Jo
b. Reversing the order of integration, we see from Figure 12.10 that for each fixed v,
the variable x varies (left to right) from the line x = y/2 to the vertical line x = 1.
The outer limits of integration are for y as y varies from y = 0 to y = 2.
Even though we can find the area between curves with single integrals, it is often
easier to compute area using a double integral. It ./(x, y) > 0 ovei a legion D in
the xy-plane, then ff fix, y) dA gives the volume of the solid bounded above by the
surface z = fix, y) and below by the region D. In the special case where f (x, y) = 1,
we have / / 1 dA = AREA OF D.
o
790 Chapter 12 Multiple Integration
A = dA
If / is continuous and /(x, y) > 0 on the region D, the volume of the solid
under the surface z = fix, y) above the region D is given by
V = JI fix, y) dA
Solution
a. The graph is shown in Figure 12.11. We find that
nJt/4
r r /»7r/4 pcosx /»
y=c osx
b. A = / / dA = I I 1 dy dx = / [V] dx
JJ do J sin x Jo
D
r”/4
= / [cosx — sinx] dx = \fl — 1
Jo
Figure 12.11 The area of the region
between y = cosx and y = sinx The area is \Jl — 1 % 0.41 square unit.
In comparing the single and double integral solutions for area in Example 3, you
might ask, “Why bother with the double integral, because it reduces to the single inte¬
gral case after one step?” The answer is that it is often easier to begin with the double
integral
Solution
The projection in the xy-plane is shown in Figure 12.12. We can regard D as either
a type I or type II region, and because we worked with a type I (vertical) region in
Figure 12.12 The quarter disk Example 3, we will use a type II (horizontal) region for this example. Accordingly,
x2 + y2 < 1, x > 0, y > 0 note that for each fixed number y between 0 and 1, x varies between x = 0 on the left
12.2 Double Integration over Nonrectangular Regions 791
v = JJ /(*. y)dA
D
dy
-L x=0
= fyA
JO
y2 dy let u — 1 — yL to integrate.
1 y=i 1
= |-3<.-^
3
y=o
Often a region D is both vertically and horizontally simple, and to evaluate an integral
fDf(x,y)dA you have a choice between performing x-integration before
y-integration, or vice versa. In the following example, you are given one order of
integration and are asked to reverse the order of integration.
f(x, y) dy dx
Solution
Draw the region D by looking at the limits of integration for both x and y in the double
integral. For this example, we see that the y-integration is done first, so D is a type I
region. The inner limits are
a. D as a type I region; y
varies from 1 to ex y — ex (top curve) and y = 1 (bottom curve)
These are shown in Figure 12.13a. Next, draw the appropriate limits of integration
for x:
x = 0 (left point) and x-2 (right point)
Figure 12.13 The region of integration The two different ways of representing the integral in this example may be summarized
for Example 5 as follows:
792 Chapter 12 Multiple Integration
Solution
The parabola and the line intersect where
x2 — 2 = x
x2 — x — 2 = 0
(x - 2)(x + 1) = 0
Thus, they intersect at (2, 2) and (—1, —1). The graph of D is shown in Figure 12.14.
As a type I region, D can be described as the set of all points (x, y) such that for
each fixed x in [-1, 2], y varies from x2 - 2 to x (see Figure 12.14). The area, A, of
D is given by
= ff
J-1 Jx 2-2
dy dx
=J [X - (x2 - 2 )]dx
a. Type I description
The set of all (x, y) such that for each fixed y between —1 and 2, x varies from
the line x = y to the right branch of the parabola x = Jy + 2;
2)
and
The set of all (x, y) such that for each fixed y between -2 and -1, x varies from
the left branch of the parabola x = ~Vy + 2 to the right branch x = Vy + 2.
Evaluate dy dx
Solution
We cannot evaluate the integral in the given order (y-integration first) because the inte¬
grand ey2 has no elementary antiderivative. We will evaluate the integral by reversing
the order of integration. The region of integration is sketched in Figure 12.15a. Note
that for any fixed x between 0 and 1, y varies from a: to 1.
To reverse the order of integration, observe that for each fixed y between 0 and 1,
x varies from 0 to y, as shown in Figure 12.15b.
n
Jn/3 .
In/'S JO
ir/4
in-1 y
i cos - dx dy
y y
sec x dx dy
Sketch the region of integration in Problems 3-16, and compute
the double integral (either in the order of integration given or with 20. f f y 1 + cos2 x cos x dx dy
the order reversed). Jo J sin-1 y
/•4 r4x Evaluate the double integral given in Problems 21-32 for the spec¬
1
JJ
pe p]nx
21. JJ (x + y)dA\ D is the triangle with vertices (0, 0), (0, 1),
5.
LX dy dx
p2\f2 p -d 12-y2
6.
c\ r-Sy+2
xy dy dx
D
(1,1).
7.
Jo Jy2 /4
dx dy 8. / 1
J-2 Jy2+Ay
dx dy
22. JI (2y - x)dA; D is the region bounded by y = x2 and
nr
D
9. (x + y) dy dx 10. J j^ (x + y2)dy dx y = 2x.
ff dA
26. 11 —--; D is the triangle bounded by x = 2y, y = —x, 52. The solid bounded above by the paraboloid z — 6 — 2x2 — 3y2
and below by the plane z = 0
D
and y — 2.
277/ D
\2x2ey dA; D is the region in the first quadrant bounded
53. The solid that lies inside both the cylinder x2 + y2 = 3 and the
sphere x2 + y2 + z2 = 7
54. The solid that lies inside both the sphere x2 + y2 + z2 = 3 and
by y — x3 and y = x.
the paraboloid 2z = x2 + y2
28. jJ cose* dA; D is the region bounded by y — ex, y — —ex, x y z
2 2 2
D
55. The solid bounded by the ellipsoid-1-1-= 1
a1 b2 c2
x = 0, and x = In 2.
f f sin x 56. The solid bounded above by the plane z = 2 — 3x — 5y and
29. / / -dx dy; D is the triangle bounded by the lines x = 0, below by the region shown in Figure 12.16.
D
» y = n, and y — x.
317/ D
x dy dx; D is the region between the parabola y — x2 and
the line y = 2x + 3.
ff 4-2x
32. J J --—dy dx; D is the triangle with vertices (0, —2),
xy-plane with vertices (0, 0), (1,0), and (4, 1). x2 + y2 < 9 in the xy-plane.
*4
where D is the triangle with vertices (—1,0), (2,0), and
(0, 1).
74. The region D shown in Figure 12.17 is the union of a type I
and a type II region. Evaluate
xy dA
66. Find the volume under the plane z = x + 2y + 4 and above the
region D bounded by the lines y — 2x, y = 3 — x, and y = 0.
67. Find the volume under the plane 3x + y — z = 0 and above the
elliptic region bounded by 4x2 + 9y2 < 36, with x > 0 and
y > 0.
68. Find the volume under the surface z = x2 + y2 and above the
square region bounded by |x| < 1 and |y| < 1.
Evaluate each integral in Problems 69—71 by relating it to the vol¬
ume of a simple solid. Figure 12.17 Union of type I and type II regions
____•
Polar coordinates are used in double integrals primarily when the integrand or the
region of integration (or both) have relatively simple polar descriptions. As a preview
of the ideas we plan to explore, consider Figure 12.18 and let us examine the double
integral
(x2 + y2 + 1 )dA
R
I
where R is the region (disk) in the xy-plane bounded by the circle x2 + y2 = 4.
Figure 12.18 Graph of the surface Interpreting R as a vertically simple region (type I), we see that for each fixed
fix, y) — x2 + y2 + 1 along with x between —2 and 2, y varies from the lower boundary semicircle with equation
the' region R shown in the xv-plane y = —x/4 - x2 to the upper semicircle y = V4 - x2, as shown in Figure 12.19a.
Using the type I description, we have
(x2 + y2 + 1) dA (x2 + y2 + 1) dy dx
R
The iterated integral on the right is difficult to evaluate, but both the integrand and the
domain of integration can be represented quite simply in terms of polar coordinates.
Specifically, using the polar conversion formulas
and the region of integration R is just the interior of the circle r — 2. Thus, R can be
described as the set of all points (r, 0) so that for each fixed angle 6 between 0 and 2tt,
r varies from the origin (r = 0) to the circle r = 2, as shown in Figure 12.19b.
II fir, 6) d A fir, 9) r dr dO
D
^ What This Says The procedure for changing from a Cartesian integral
fix, y)dA
Proof A polar region described by r\(9) < r < r2(9), a <9 < f> can be subdivided
b. A typical polar rectangle into polar rectangles. A typical polar rectangle is shown in Figure 12.20.
in the partition
We begin by subdividing the region of integration into polar rectangles. Next we
pick an arbitrary polar-form point irk, 0%) in each polar rectangle in the partition and
Figure 12.20 A polar rectangle then take the limit of an appropriate Riemann sum
E/<r*’A‘) AA‘
k= 1
See Section 1.2 of the In Chapter 6 we used the fact that a circular section of radius r and central angle 9 has
Student Mathematics Handbook. area \r29 (see the Student Mathematics Handbook for details). Thus, a typical polar
rectangle has area
Finally, we compute the given double integral in polar form by taking the limit
N N
/(r, 9) r dr d9 □
798 Chapter 12 Multiple Integration
If you carefully compare the result in the preview box with the result of Theorem
12.4, you will see that they are not exactly the same. The region D in Theorem 12.4
already is in polar coordinates. The more common situation is that we are given / and
a region R in rectangular coordinates that we need to change to polar coordinates.
You will need to be familiar with the graphs of many polar-form curves. These can be
found in Table 6.2 on page 379, and also in the Student Mathematics Handbook. We
now present the example we promised in the introduction.
Chapter 5 of the Student
Mathematics Handbook.
EXAMPLE 1 Double integral in polar form
Solution
In this example, the region R is given in rectangular form. We will describe this as a
polar region D. Earlier, we observed that in D, for each fixed angle 6 between 0 and
2jt, r varies from r = 0 (the pole) to r = 2 (the circle). Thus,
It f(x, y) f(rcosO,rsin0) ‘I A
p Z7T pn2a 2n r ^4 r2 i 2 p2 7T
(r -Hr) dr dO do 6 d6 = 60|nT — 127r
Jo Jo =L 4 +J o Jo
12.3 Double Integrals in Polar Coordinates 799
Solution
The circle *2 + y2 — 2y = 0 and the line y = x are shown in Figure 12.21. The polar
form for the line y — x is 9 = n/A and for the circle is
*2 + y2 = 2y
Fixed 0
r2 = 2r sin 6
r = 2 sin 9
Thus, the region D is determined as r varies from 0 (the pole) to 2 sin 0 (the circle) for
0 between 0 and 7t/4. The area is given by the integral
r r ■'jt/4
/»7T/4 |'2sin 6
/•ZSintf
-L *7r/4
-r
r=0
de
(4 sin2 0) d9
I.,
n/A
1 — cos 29
= 2fJO
dO
7l/4
sin 29 TC —2
9 -
Solution
We will compute the required volume by doubling the volume of the solid hemisphere
x2 + y2 + z2 < a2, with z > 0. This hemisphere (see Figure 12.22) may be regarded
as a solid bounded below by the circular disk x2 + y2 < a2 and above by the spherical
surface.
We need to change the equation of the hemisphere to polar form:
Figure 12.22 The solid hemisphere
z < \]a1 — x2 — y2 is bounded In rectangular form: z = \Ja2 — x2 — y2
above by the sphere and below by In polar form: z — Va2 — r2 Because r~ — x2 + y2
the disk x2 + y2 < a2.
Describing the disk D in polar terms, we see that for each fixed 9 between 0 and 27r, r
varies from the origin to the circle x2 + y2 — a2, which has the polar equation r = a,
as shown in Figure 12.23. Thus, the volume is given by the integral
Rectangular form
Evaluate jj — dA, where D is the region that lies inside the circle r — 3cos# and
D
outside the cardioid r = 1 + cos#.
Solution
Begin by sketching the given curves, as shown in Figure 12.24. Next, find the points
of intersection:
3 cos# = 1 + cos#
2cos# - 1
cos# = i
n _ _7r 7T
U — 3 ’ 3
Notice that D is the region such that, for each fixed angle # between — k/3 and k/3, r
varies from 1 + cos # (the cardioid) to 3 cos # (the circle). This gives us the limits of
integration, and in polar form, the integrand becomes
1 1
x r cos#
Figure 12.24 Region D
Rectangular form Polar form
Thus,
1
Ill'll r cos#
r dr dO
D
/ D
7r/3
K/3 />3cos0 j
dr dO
k/3 J
-k/3 Jl+cos0
1 COS#
r r r=3cos0
dO By symmetry
-n cos#- r=l+cos0
k/3 j
[3cos# — (1 Tcos#)] dO
=2i/ o cos#
[•k/3
= 21 (2 — sec#) dO
Jo
k/3
= 2 [2# — In |sec# + tan #|]|
= Y -21n(2 + x^)
Solution
The region of integration D is the set of all points (x, y) such that for each point x in
the interval [0, 2], y varies from y = 0 to y = V2x — x2. Note that
y = sj'lx — X2
y2 = 2x — x2
x2 — 2x + y2 = 0
Figure 12.25 Region of integration in
Example 5 (x- l)2 + y2= 1
12.3 Double Integrals in Polar Coordinates 801
y = y/2x — x2
y2 = 2x — x2
X2 T y2 — 2x
which has the polar form
r = 2r cos #
r — 2 cos#
In polar coordinates, D is the region in which r varies from 0 to 2 cos # for # between
n
0 and j. Thus, the integral is
y/2x—x2
y\Jx2 + y2 dy dx = x2 + y2 dA
n 2cos 9
(r sin#)r (r dr dO) = /
D
Jo
rn/2
sin#
r a-i 2 cos 9
r
d6
nil
i r/2
= 4Jo
16 cos4 # sin # dO
-H cos5 #
1. Jr'2 J f3 2
re~r dr d6 2. / J y/A — r2 r dr d9
1J-
32.
// (x2 + y2) dA, where D is the region bounded by the
39. f xdydx
Jo Jo
"2 J
/r ^ 8 —V2 1
40. I I —-—: dx dy
"IIy v/l +x2 + y2
2 rJfUfi
„*2+v2
dx dy
23. Use a double integral to find the area inside the inner loop of ■n
the lima?on r = 1 + 2 cos 9. ’ 9—x2
24. Use a double integral to find the area bounded by the parabola
42' ff »4
cos(x2 + y2) dy dx
n
r = 43 . f r dx dy
1 + COS0
Jo Jo yjx2 + y2
'9-x2
60. Use polar coordinates to evaluate the double integral
46. ln(x2 + y2 + 9) dy dx
ff (x2 + y2)dA over the shaded region.
t9-x2
47.
n 2 r \Jlx— X
x -y
X2 + y2
dy dx
Pierre-Simon Laplace has been called the Newton of where the integral on the right is over the entire xy-plane, de¬
France. He taught Napoleon Bonaparte, was appointed for scribed in polar terms by 0 < r < oo and 0 < 6 < 2n.
a time as Minister of Interior, and was at times granted favors 63. For constant a where 0 < a < R, the plane z = R — a cuts
from his powerful friend. Today, Laplace is best known as the off a “cap” from the hemisphere
major contributor to probability, taking it from gambling to a
true branch of mathematics. He was one of the earliest to eval¬ Z = y/R2-x2-y2
uate the improper integral
Use a double integral in polar coordinates to find the volume
of the cap. For what value of a is the volume of the cap half
the volume of the hemisphere?
which plays an important role in the theory of probability. ■
Show that / = s/n. Hint: Note that
oo n oo n oo /»oo
dx ■ / dy = / e~(x2+y2) dx dy
-oo J —oo J— oo J — 00
L = + [/'(*)]2 dx
b. The surface area above Rm is AB = Axmi + [fx(x*, y*) Axm ]k and AC = Ay„,j + [fy(x*, y*) Aym]k
approximated by the area of the
parallelogram on the tangent plane.
In Chapter 9, we showed that such a parallelogram with sides AB and AC has area
Figure 12.26 Surface area
||AB x AC||
Km = || AB x AC ||
i j k
AB x AC = Axm 0 Mx*,y*)Axm
0 A ym fy(x*m,y*m)Aym
Km = || AB x AC ||
Finally, summing over the entire partition, we see that the surface area over R may
be approximated by the sum
where AAm — AxmAym. This is a Riemann sum, and by taking an appropriate limit
(as the partitions become more and more refined), we find that surface area, S, satisfies
S = lim
n-+oO
V J[fx(Xm, .V*)]2 +
^' V
y*)]2 + 1 AA„
m=1
S = JJ J[Mx,y)]2 + [fy(x,y)¥+ldA
/ dx /pb
Ja
pb
ds = j yir (x)]2 + 1 dx
Ja
ffdA
R
jj dS = Jj y/[fx(x, y)]2 + [fy(x, y)]2 + 1 dA
R R
Solution
The plane x + y + z = 1 and the triangular region T that lies beneath it in the
xy-plane are shown in Figure 12.28. We begin by letting /(x, y) = 1 — x — y. Then
fx(x, y) = -1 and fy(x, y) = — 1, and
pt rn———i—i—n—'
i-i-i— J{-\)2 + (-l)2+\dA
T
= // l r\-x
y/3 dy dx
Look at Figure 12.26 to see the region T
and the limits for both x and y.
Fixed x_1
Solution
To determine the limits for the region D note that the paraboloid intersects the plane
Z = 1 in the circle x2 + y2 — 4. Thus, the part of the paraboloid whose surface area
we seek projects onto the disk x2 + y2 < 4 in the jcy-plane, as shown in Figure 12.29
(note the shadow).
It is easier if we convert to polar coordinates by letting x = r cosO, y — r sin#.
Because the region is bounded by 0 < r < 2 and 0 < 9 < 27r, we have
D
Projected region D first in rectangular
coordinates and then f*2rc p2
switch to polar form, r2 + \ r dr d9 Let u — Ar2 + 1, so
if appropriate. Jo Jo du — 8r dr.
r2lt rxl du If r = 2, then u = 17
= / / ux'2 — dO and if r = 0, u = 1.
Jo J1 8
»2tt
1 2
T — • — 3/2 |17 dQ
Jo 8 3
n 2.7T
/ (173/2 I )d0
Disk x2 + y2 < 4 in polar 12 Jo
coordinates
We have been projecting the given surface onto the jcy-plane, but we can easily mod¬
ify our procedure to handle other cases. For instance, if the surface y = f(x, z) is
projected onto the region Q in the xz-plane, the formula for surface area becomes
Q
An analogous formula for projection onto the region T in the yz-plane would be
Solution
The cylinder is shown in Figure 12.30a. We can find the lateral surface area using the
formula for a rectangle by imagining the cylinder “opened up.” The length of one side
is the height of the cylinder (3 units), and the length of the other side of the rectangle
b. The projection of the is the circumference of the bottom, namely, 27t(2) = 47r. Thus, the lateral surface area
cylinder on the xz plane is 12k.
We will illustrate our integral formula by using it to recompute this surface area.
Figure 12.30 Surface area of a cylinder Specifically, we will compute the surface area of the right half of the cylinder and then
808 Chapter 12 Multiple Integration
double it to obtain the required area. If we project the half-cylinder onto the xy-plane,
we will not be accounting for the height of the cylinder. Instead, we will project the
surface onto the xz-plane to obtain the rectangle Q bounded by x = 2, x = —2, z = 0,
and z = 3, as shown in Figure 12.30b. Because we are projecting onto the xz-plane,
we solve for y to express the surface as y = /(x, z). Since
y = f(x, z) = \J4 — x2
—x
we have fx{x, z) = —7= and /z(x, z) =0 so that
V4 x2 -
s = //vS<iA = £f Vr^dzdx
Q
,
= 6 sin
• -1 X 2
- —bn
r
2 -2
The lateral surface area of the whole cylinder is 2(67r) = 127r. ■
Suppose a surface S (see Figure 12.31) is defined parametrically by the vector function
R(u, v) = x{u, u)i + y(u, u)j + z(u, u)k
for parameters u and v.
5 = ||RU x R„|| du dv
Solution
We find that
i j k
R;/ X R„ = sin v cos v 2u (2m2 sin u)i + (2 m- cos u)j — uk
u cos v —u sinu 0
We find that
Notice that in the special case where the surface under consideration has the ex¬
plicit representation z = fix, y), it can also be represented in the vector form
where x and y are used as parameters ix — u and y = v). With this vector representa¬
tion, we find that
i j k
Rv X Ry = 1 0 fx = -fxi - fyj + k
0 1 /v
810 Chapter 12 Multiple Integration
Therefore, in the case where z — fix, y), the surface area over the region D is given
by
= JJ y/ fx + fy + 1 dx dy
D
25. Find the surface area of that portion of the plane Compute the magnitude of the fundamental cross product for the
suiface defined parametrically in Problems 37-40.
x +y+z —a
37. R(u, v) — (2u sin u)i + (2u cos u)j + rrk
38. R(m, v) = (4sin u cos u)i + (4 sin u sin t>)j + (5 cos u)k
that lies between the concentric cylinders x2 + y2 = —■ and
4 39. R(u, u) = ;ri + irj + «3k
x2 + y2 = a2 (a > 0).
40. R(u, v) = (2u sin u)i + (2u cos u)j + (w2 sin 2u)k
26. Find the surface area of the
41. Find the area of the surface given parametrically by the equa¬
portion of the cylinder
tion
x2 + z2 — 9 that lies inside the
R(u, v) = uvi + (u — u)j + (u + u)k
cylinder y2 + z2 — 9.
for u2 + ir < 1. Hint'. Use polar coordinates with u = r cos 9
and v = r sin#.
42. A spiral ramp has the vector parametric equation
27. Find a formula for the area of the conical surface R(w, v) = (u cos u)i + (u sin u)j + uk
z = s/x2 + y2 between the planes z = 0 and z — h.
28. Find a formula for the surface area of the frustum of the cone for 0 < m < 1, 0 < u < 7T. Find the surface area of this ramp.
z — 4,/x2 + y2 between the planes z — h\ and z = h2, where
h i > /z 2 • Q 43. The surface given parametrically by
Figure 12.33 The box B contains D and is subdivided into smaller boxes.
k= 1
If we repeat the process with more subdivisions, so that the norm approaches zero, we
are led to the following definition.
Triple Integral If / is a function defined over a closed bounded solid region D, then the triple
integral of / over D is defined to be the limit
Subdivision rule If the solid region of integration D can be subdivided into two
Figure 12.34 Dividing the solid D solid subregions D\ and Di (see Figure 12.34), then
into two solid subregions
ITERATED INTEGRATION
The iterated integration can be performed in any order, with appropriate adjustments
to the limits of integration:
Proof The proof, which is similar to the two-dimensional case, can be found in an
advanced calculus course.
Solution
z2yex dx dy dz
///
B
^X’ ^ Z^dV u:l
Treat y arid z as constants.
• 1 />2
Treat z as a constant.
>1^
y=2 ’22 12‘
to
I?'
II
dz = (e - 1) / z2
1
-« J-i Li J y=1 J-1 .2 2 _
Z=1
3 T1 3
z2dz = ^(e - 1) ^
2 z=-
(e_l) = 6—1
As an exercise, verify that the same result is obtained by using any other order of
integration—for example, dz dy dx. ■
Next, we will see how triple integrals can be evaluated over solid regions that are
not rectangular boxes. We will assume the solid region of integration D is z-simple
Upper boundary
surface z = v (x, y)
in the sense that it has a lower bounding surface z = u(x, y) and an upper bounding
surface z = v(x, y), and that it projects onto a region A in the xy-plane that is of either
type I or type II. Such a solid is shown in Figure 12.36.
The region D can be described as the set of all points (x,y,z) such that
u(x, y) < z < v(x, y) for all (x, y) in A. Then the triple integral of fix, y, z) over the
Lower boundary solid region D can be expressed as an iterated integral with inner limits of integration
surface z = u (x, y) (with respect to z) m(x, y) and v(x,y). We summarize in the following theorem.
Proof Even though a proof is beyond the scope of this text, we note that if A is
vertically simple (type I), then for each fixed x in an interval [a, b], y varies from
<?i (x) to g2(x), and the triple integral of f(x, y, z) over D can be expressed as
ng2(x) />v(x,y)
/(x, y, z)dz dydx
/// ^K' y' Z^dV .lW Ju(x.y)
D
Likewise, if A is horizontally simple (type II), then for each fixed y in an interval [c, d],
x varies from h i (y) to h2(y), and
rd rh2(y) rv(x.y)
IJI fU.y.z)dv fix, y, z)dzdx dy □
Jc Jh\(y) Ju(x.y)
D
Evaluate
-///
D
x d V, where D is the solid in the first octant bounded by the cylinder
Solution
The solid is shown in Figure 12.37. The upper boundary surface of D is the plane
z = 4 — 2y, and the lower boundary surface is the xy-plane, z = 0. The projection A
of the solid on the xy-plane is the quarter disk x2 + y2 <4 with x > 0, y > 0 (because
D lies in the first octant). This projection may be described in type I form as the set
of all (x, y) such that for each fixed x between 0 and 2, y varies from 0 to y/4 — x2.
Thus, we have
C r r4—2y p2 p \/ 4—x2 p4—2y
/// D
xdV = // /
J J Jo
xdzdA = / /
Jo Jo
/
Jo
xdzdydx
r 32 1 20
-|(4 - x2)3/2 - 2x2 + ix4 = 0-8 + 4+y +0-0 = y
Just as a double integral can be interpreted as the area of the region of integration, a
triple integral may be interpreted as the volume of a solid. That is, if V is the volume
of the solid region D. then
Solution
The tetrahedron T is shown in Figure 12.38a. The upper surface of T is the plane
z = 1(6 - 2x - y) and its lower surface is z = 0. Note that T projects onto a triangle,
A, in the xy-plane, as shown in Figure 12.38b. Described in type I form, the triangle
A is the set of all (x, y) such that for each fixed x between 0 and 3, y varies from 0 to
6 — 2x. Thus,
T A
,5
-3 po-zx
r6-2x p (>-2x-y)
= / dz dy dx
Jo Jo Jo
/»3 r*(j—2x
= / [5(6 - 2x - y) - 0] dy dx
Jo Jo
= f'[2.v-^-iy]|;:r *
Jo
Chapter 12 Multiple Integration
z,
l
= / iK36 24x + 4x2] dx = 6
III
D
^ Z^dV f(x, y, z) dx dA
r r r
f(x,y,z) dx dA f(x,y,z) dy dA
W y
a. A solid D with “front” surface b. A solid D with “side” surfaces
x = x2{y, z) and “back” surface y=y2 (*, z) and y = yx (x, z)
x = xx (y,z)
On the other hand, if the solid region of integration D is bounded by the surfaces
y = (*> z) and y = y2(x, z), and the boundary surfaces project onto a region A in
the *z-plane as shown in Figure 12.39b, then
f(x, y, z) dy dA
D A
12.5 Triple Integrals 817
Solution
Note that T is bounded by the yz-plane and the plane 2x + y + 3z = 6, which we
express as x = ±(6 - y - 3z). (See Figure 12.40a.) The volume is given by
V = dx dA
where B is the projection in the yz-plane. This projection is the triangle bounded by
the lines z = 0, y = 0, and z = 5(6 - y), as shown in Figure 12.40b.
z,
Thus, for each fixed y between 0 and 6, z varies from 0 to \ (6 - y), and we have
n i(6-y) A(6-y-3z)
/
J0
dxdzdy = /
*6
Z*0
/
Jo Jo
rU6~y)
[;
5(6 — y — 3 z)dz dy
= f [3z-',yzdy
Jo
This is the same result we obtained in Example 3 by projecting onto the xy-plane.
Projection on
xy-plane
Solution
First, note that the intersection of the plane and the sphere occurs above the xy-plane
(where z > 0), so the sphere can be represented by the equation
Z = \/4 — x2 — y2
(the upper hemisphere). To find the limits of integration for x and y we consider
the projection of D onto the xy-plane. To this end, consider the intersection of the
hemisphere and the plane z — 2 — y\
sjA -x2 - y2 = 2 - y
4 — x2 — y2 = 4 — 4y + y2
x2 + 2y2 - 4y = 0
x2 + 2(y — l)2 = 2
Although this intersection occurs in R3, its equation does not contain z. Therefore,
the equation serves as a projection on the xy-plane, where z = 0. This is an ellipse
centered at (0, 1), as shown in Figure 12.42.
We consider this as a type II region, which means we will integrate first with re¬
spect to x, then with respect to y. Because x2 + 2(y — l)2 = 2 we see that for fixed
y between 0 and 2, x varies from —y/2 — 2(y — l)2 = — yjAy — 2y2 to y/Ay — 2y2.
Figure 12.42 The projection of D However, using symmetry, we see the required volume V is twice the integral as x
onto the xy-plane varies from 0 to y/Ay — 2y2. This leads us to evaluate V by the following triple inte¬
gral.
n y/4y—2y2
/
Jl-y
p y/4—x2—y2
dzdxdy ■
Solution
The graph of the solid D is shown in Figure 12.43a. The projection of the solid onto
the x v-plane is the graph of the equation
2 2
x + y = 3 — 2x Substitute the second equation into the first.
2 2
(x + 1) -|-y =4 Complete the square for x.
12.5 Triple Integrals 819
Figure 12.43 The solid D bounded by z = x2 + y2 and 2x + z — 3 and projections on two coordinate planes.
This is a circle of radius 2 centered at (—1, 0), as shown in Figure 12.43b. Proceeding
as in Example 5, we find that the volume is given by the integral
/ I n y/3—2x—X2
/ /
i>3— 2x
dzdydx
-3 Jo Jx2+y2
A : set of all (x, z) such that for each fixed .v in the interval
—3 < x < 1, z varies from z = t2 to z = 3 — 2x
= 2 J* ^ Vz-x2dzdx
/■1 2
= 2 J-3 3(Z _ *
= 5
— - f [4 — (x + 1 )2}2’/1 dx Complete the square.
3 J-3 -——7 ^-7-TT
Let x + 1 = 2 sin 9\ dx = 2 cos Odd
4 r11 x
= - 8 cos3 9(2 cos 0 dO)
3 J-n/2
128 r/2
f
3 Jo
cos4 0 dO Symmetry
820 Chapter 12 Multiple Integration
128 /3tt\
Formula 320
1“ VT6 7
= 87T
■ns:
J-\ JO J-2 ties 0<jc< l,0<y<x, and 0 < z < x + y
8x2yz3 dx dy dz 22- SIS yzdV, where D is the solid in the first octant bounded
■n
D
ms: x+y
x2y2z2 dx dy dz
xyzdzdy dx
by the hemisphere x = y/9 - y2 - z2 and the coordinate
planes
Find the volume V of the solids bounded by the graphs of the equa¬
tions given in Problems 23-32 by using triple integration.
pi r-s/l+x pxy 23. x + y + z = 1 and the coordinate planes
8• / y~'zdzdy dx
Jo Jj;
IJx Jo
/»2Z /»7T
/» 71 /»4
/*4
9 / / yz cos xy dzdx dy
\ Jo J\
rTT n1 /* 1
10 ,
m/
Jo Jo Jo
/ / x2y cos xyzdzdy dx
\ny
n 2z
/
piny
Jo
ez+2x dzdxdy
ye~x dx dy dz
4 r2z r-fhx
n- s /_, l xz + y
24. y — 9 — x2, z = 0, z = y
nx2 py
14 / (x + y)dzdydx
— 1 J —X
Evaluate the triple integrals in Problems 15-22.
SSI
D
(xy + 2yz)dV, where D is the box 2 < x < 4,
25. (x - l)2 + (y - 2)2 + (z - 3)2 = 1 For each given iterated integral there are five other equivalent iter¬
ated integrals. Find the one with the requested order in Problems
33-38.
33.
fff fix, y, z) dzdy dx;
34. fi I f
J Jo Jo
f(x,y,z)dydxdz\
n
change the order to dx dy dz.
nl V4-^2
r\f4—x2
1
Jo
fV 4-x2
f \!4-x2
f(x,y,z)dzdydx;
37 . f rf
Jo Jo Jo
f(x, y, z)dx dzdy;
28. x2 + y2 + z3 = 9, z = 0
■H
: >
R
44. Find the volume of the solid region in the first octant that is 51. Evaluate the triple integral
bounded by the planes z = 8 + 2x + y and y — 3 - 2x.
Q 45. Use triple integration to find the volume of a sphere.
46. Use triple integration to find the volume of a right pyramid
with height H and a square base of side S.
Iff sin(;r — z)2 dzdydx
=U. /(H[fs<H [/
h(z) dz
Also, show that where H is the four-dimensional region bounded by the hy¬
A (planar) lamina is a flat plate that occupies a region R in the plane and is so thin it
can be regarded as two dimensional. If m is the lamina’s mass and A is the area of the
region R, then 8 = j is the density of the lamina. The lamina is homogeneous if its
density <5(x, y) is constant over R and nonhomogeneous if <$(jt, y) varies from point
to point. We considered homogeneous lamina in Section 6.5, and in this section, we
examine nonhomogeneous lamina.
Figure 12.45 shows a lamina covering a region R that has been partitioned into a
number of rectangles. Choose a representative point (x£, y*) in each rectangle of the
partition, and note that the mass of the part of the lamina that lies in the &th subrectangle
is approximated by
m ^^8{xl,yl)AAk
*=l
12.6 Mass, Moments, and Probability Density Functions 823
so that
m — lim V8(x*y*k)AAk
// 8(x, y) dA
m = JI 8(x, y)dA
R
' / f *2dydx
=J x2(2 — x2 — x) dx = J ^ ^ v3>
(2x2 — x4 — -O dx
2jc3 Xs JT 63
= — = 3.15
"T -1 ~ T -2 20
824 Chapter 12 Multiple Integration
The moment of an object about an axis measures the tendency of the object to
rotate about the that axis. It is defined as the product of the object’s mass and the
signed distance from the axis (see Figure 12.47). Thus, by partitioning the region R
as before (with mass), we see that the moments Mx and My of the lamina about the
x-axis and y-axis, respectively, are approximated by the Riemann sums
n
mx = My and my = Mx
The center of mass (x, y) may also be thought of as the point from which the lamina
may be suspended without movement. For future reference, these observations are
summarized in the following box.
Furthermore, if m is the mass of the lamina, the center of mass is (x, y), where
Mv MY
x = and y
m m
If the density 8 is constant, the point (x, y) is called the centroid of the region.
Solution
In Example 1 we found that for each fixed x between -2 and 1, y varies from x to
2 - x2 (see Figure 12.47). Thus, we have
-II
Mx= 11 y(xz) dA
R
-II
My = 11 x(xz) dA
R
2—x
= hi
2—x2
yx‘dydx
-h X'3 dy dx
ll "2x4 x6 x5"
= HK-^5 + H1‘1-2 . 4 6 5 .
11
5
of mass for a lamina The center of mass for Example 1 is shown in Figure 12.48.
In a completely analogous way, we can use the triple integral to find the mass and
center of mass of a solid in R3 with density <5(x, y, z). The mass m, moments Myz,
Mxz, Mxy about the yz-, xz-, and xy-planes, respectively, and coordinates x, y, z, of
the center of mass (see Figure 12.49) are given by:
Moments M
:fil
r
x<5(x, y,z)dV
t
Distance to the yz-plane
Mxz
-ill
R
y8(x, y, z)dV
As before, if the density is constant, the center of mass is called the centroid. Example
3 illustrates how this point can be found by multiple integration.
Solution
a. A tetrahedron
The tetrahedron may be described as the region in the first octant that lies beneath the
plane x + y + z = 1, as shown in Figure 12.50a. The boundary of the projection
of the top face of the tetrahedron in the xy-plane is found by solving the equations
JC + }; + z= 1 and z = 0 simultaneously. We find that the projection is the region
bounded by the coordinate axes and the line x + y = 1, as shown in Figure 12.50b.
This means that for each fixed x between 0 and 1, y varies from 0 to 1 — x.
m
- Iffsdv-I'll ‘ C ’ w*
R
l-x r\-x-y I
Mxy
=ui zdv=j» r l
6
bz dzdydx = -
MOMENTS OF INERTIA
In general, a lamina of density 8 (*, y) covering the region R in the first quadrant of the
plane has (first) moment about a line L given by the integral
where dm = 8(x,y)dA and s = s(x, y) is the distance from a typical point P(x, y)
in R to L. Similarly, the second moment, or moment of inertia, of R about L is defined
by
In physics, the moment of inertia measures the tendency of the lamina to resist a
change in rotational motion about the axis L. Moments of inertia about the coordinate
axes are given by the formulas listed in the following box.
Moments of Inertia
Solution
The graph of R is shown in Figure 12.51. We see that for each fixed x between 1 and
2, y varies from 1 to x2.
Ix JJ v2 dm = JJ y2S(x, y) dA — JJ y2(x2y) cl A
=
i I '' ~~/
, o 1,516
x2(x8 - 1) dx = —-« 45.94
33
-a; 2 A A 1, 138
4(x4 - 1 )dx = -4— % 25.29
<"/// R
(y2 +z2)S(x,y,z)dV
t Increment of mass
Square of the distance to the .v-axis
-ill
R
(.x2 + z‘)5(jc, y, z)dV
t Increment of mass
Square of the distance to the y-axis
I-
-II!
R
(x2 + y2)S(x, y, z)dv
t Increment of mass
Square of the distance to the z-axis
Solution
In Example 3, we observed that the solid S can be described as the set of all (*, y, z)
such that for each fixed x between 0 and 1, y lies between 0 and 1 - x, and
0<z<l-*-y. Thus,
I- x(x2 + y2)dzdy dx
=!!I(x2+y2m’y’z)dv=l l I
s
"x3(l — x)2 *(1 — x)4
= [ f x(x2 + y2)(l - x - y)dy dx = f dx — —
~2 \2~ 90
Jo Jo Jo +"
828 Chapter 12 Multiple Integration
1 9 9
K rot jco r dm r~ dm
Figure 12.53 Kinetic energy of a
rotating disk
Since r2 = x2 + y2, we see that the integral in this formula is just the moment of inertia
of R about the z-axis, so the rotational kinetic energy can be expressed as
^rot — 2 h<*>
Comparing this formula to the kinetic energy formula K\in = \mv2, we see that the
moment of inertia may be thought of as the rotational analogue of mass or, equivalently,
as rotational inertia.
In general, f(x) >0 for all x, and since the value of X is always some real number, it
follows that P(—00 < X < 00) = 1, so
f{x)dx = 1
In geometric terms, the probability P(a < X < b) is the area under the graph of /
over the interval a < x < b. (See Figure 12.54.)
Figure 12.54 Probability as the If X and Y are both continuous random variables, then the joint probability den¬
area under the graph of the sity function for two random variables X and Y is a function of two variables fix, y)
probability density function such that f(x, y) > 0 for all (x, y) and
/
where P[(X, Y) in D] denotes the probability that (X, F) is in the region D. Note that
OO
/
POO
f(x,y)dxdy— 1
-OO J —OO
Geometrically, P[(X, Y) in D] may be thought of as the volume under the surface
Z = fix, y) above the region D.
The techniques for constructing joint probability density functions from experi¬
mental data are outside the scope of this text and are discussed in many texts on proba¬
bility and statistics. The use of a double integral to compute a probability with a given
joint density function is illustrated in the next example.
12.6 Mass, Moments, and Probability Density Functions 829
Find the probability that the customer’s total time in the store will be no greater than
30 min.
Solution
The goal is to find the probability that X + Y <30. Stated geometrically, we wish to
Figure 12.55 Triangle comprised find the probability that a randomly selected point (.x, y) lies in the region R in the first
of all points (X, Y) such that quadrant that is bounded by the coordinate axes and the line x + y — 30 (see Figure
X + Y < 30, X > 0, Y > 0 12.55). This probability is given by the double integral
_ j_
to .
e~3 - 2e~3/2 + 1
0.6035
Thus, it is about 60% likely that the shopper will spend no more than 30 minutes in the
store.
P(E) + P(E) = 1
where E and ~E together constitute the set of all possible outcomes (that is, they are
complementary events). For Example 6, since the events E — {the shopper will spend
no more than 30 minutes in the store} and E = {the shopper will spend more than 30
minutes in the store} are complementary, we can find the probability that the shopper
will spend more than 30 minutes in the store by using this complementary property:
1 - 0.6035 = 0.3965
Find the centroid for the regions described in Problems 5-12. 26. Find the centroid (correct to the nearest hundredth) of the part
5. A lamina with 8 — 5 over the rectangle with vertices (0, 0), of the large loop of the limagon r = 1 + 2 cos 9 that does not
(3.0), (3,4), (0, 4) include the small loop.
27. Find the center of mass of the lamina that covers the triangular
6. A lamina with 8 = 4 over the region bounded by the curve
region with vertices (0, 0), (a. 0), (a. b), if a and b are both
y = sfx and the line x = 4 in the first quadrant
positive and the density at P(x, y) is directly proportional to
7. A lamina with 8 = 2 over the region between the line y = 2x the distance of P from the y-axis.
and the parabola y = x2
28. A rectangular lamina has vertices (0, 0), (a, 0), (a, b), (0, b),
8. A lamina with 8 = 4 over the region bounded by y = sin |x, and its density at any point (x, y) is the product <5(x, y) = xy.
x = 0, y = 0, * = \ Find the center of mass of the plate.
9. A thin homogeneous plate of density 8 = 1 with the shape of 29. A homogeneous lamina covers the circular disk with bound¬
the region bounded above by the parabola y = 2 — 3x2 and ary .v2 + y2 = ax. Find the moment of inertia of this circular
below by the line 3x + 2y = 1 plate about the vertical line passing through the center of the
10-» The part of the spherical solid with density 8 = 2 described lamina. Hint: Use polar coordinates.
by x2 + y2 + z2 < 9, x > 0, y > 0, z > 0 30. Show that a homogeneous lamina of mass m that covers the
11. The solid tetrahedron of density <5 = 4 bounded by the plane circular region x2 + y2 = a2, will have moment of inertia
x + y + z = 4 in the first octant ma2/4 with respect to both the x- and y-axes. What is the
moment of inertia of the lamina with respect to the z-axis?
12. The solid bounded by the surface z = sin*, x = 0, x = n,
31. Show that a homogeneous lamina of density <5 and mass m that
y = 0, z = 0, and y + z — 1, where the density is <5 = 1
covers the ellipse
Use double integration in Problems 13-18 to find the center of
mass of a lamina covering the given region in the plane and hav¬
— + -Tj <\
ing the specified density 8. a1 bl
13. 8(x, y) = x2 + y2 over x2 + y2 < 9, y > 0 has moment of inertia about the x-axis equal to
14. <5(x, y) = k(x2 + y2) over x2 + y2 < a2, y > 0
Ix — jab28 = \mb2
15. <5(x, y) = lx over the triangle with vertices (0, 0), (6, 5), and
(12, 0) where m = 8 (nab).
16. <5(x, y) = 3x over the region bounded by y = 0, y = x2, and
Area of ellipse
x = 6
32. Find the center of mass of the tetrahedron in the first octant
17. <5(x, y) = x~' over the region bounded by y = lnx, y = 0, bounded by the plane
x = 2
18. <5(x, y) = y over the region bounded by y = e~x, x = 0,
x = 2, y = 0
19. A lamina in the xy-plane has the shape of the semicircular re¬ where o, b, and c are all positive constants. Assume the den¬
gion x2 + y2 < a2, y > 0. Find the center of mass if the sity is 5 = x.
density at any point in the lamina is: 33. A solid has the shape of the homogeneous sphere
a. directly proportional to the distance of the point from the x2 + y2 + z2 < a2. Find the centroid of the part of the solid in
origin the first octant (x > 0, y > 0, z > 0).
b. directly proportional to the polar angle 34. Suppose the joint probability density function for the random
variables X and Y is
20. A lamina has the shape of a semicircular region x2 + y2 < a2,
y > 0. Find the center of mass of the lamina if the density at 2e ^e y if x > 0, y > 0
each point is directly proportional to the square of the distance n*,y) 0 otherwise
from the point to the origin.
21. Find the centroid of a homogeneous lamina that covers the re¬ Find the probability that X -f Y < 1.
gion bounded by the curve y = lnx and the lines x = e2, and 35. Suppose the joint probability density function for the random
y = 0. variables X and Y is
22. Find Ix, the moment of inertia about the x-axis, of the lamina
that covers the region bounded by the graph of y = 1 — x2 and xe xe y if x > 0, y > 0
fix, y) =
the x-axis, if the density is 8(x, y) = x2. 0 otherwise
23. Find /., the moment of inertia about the z-axis, of the
Find the probability that X + Y < 1.
lamina that covers the square in the plane with vertices
36. Modeling Problem Suppose X measures the length of time
(-1,-1), (1,-1), (1,1), and (—1,1), if the density is
<5(x, y) = x2y2. (in days) that a person stays in the hospital after abdominal
surgery, and Y measures the length of time (in days) that a
24. Find the center of mass of the cardioid r = 1 + sin 0 if the
person stays in the hospital after orthopedic surgery. On Mon¬
density at each point (r. 9) is S(r, 9) = r.
day. the patient in bed 107A undergoes an emergency appen¬
25. Find the centroid (correct to the nearest hundredth) of the loop dectomy (abdominal surgery), while the patient’s roommate in
ot the lemniscate r2 = 2 sin 29 that lies in the first quadrant. bed I07B undergoes orthopedic surgery for the repair of torn
12.6 Mass, Moments, and Probability Density Functions 831
knee cartilage. Suppose the joint probability density function The average value of the continuous function f over R is given by
for X and Y is
1
,, x \\e~xl2e-yl* if x > 0, y > 0 One variable : [ f (x) dx
f(x, y) = { " length of segment R Jr
10 otherwise
1
Find the probability (to the nearest hundredth) that both pa¬
Two variables : JI fix, y)dA
area of region R
tients will be discharged from the hospital within 3 days.
37. Modeling Problem Suppose X measures the time (in min¬
1
utes) that a person stands in line at a certain bank and Y, the
duration (in minutes) of a routine transaction at the teller’s
Three variables
volume of solid R Iff fix,y,z)dV
x \\e-x'2e-yl* if x > 0, y > 0 40. Find the average value of /(x, y) = ex\ where R is the region
fix, y) = { 8 in the first quadrant bounded by y = x2, y — 0, and x = 1.
10 otherwise
41. Find the average value of f(x, y) = exy~x/2 where R is the
Find the probability that you will complete your business at region in the first quadrant bounded by y = x2, x = 0, and
the bank within 8 min. y = 1.
38. Modeling Problem Suppose X measures the time (in min¬ 42. Find the average value of the function f(x, y, z) — x+2y+3z
utes) that a person spends with an insurance agent choosing over the solid region S bounded by the tetrahedron with ver¬
a life insurance policy and Y, the time (in minutes) that the tices (0, 0, 0), (1, 0, 0), (0, 1,0), and (0, 0, 1).
agent spends doing the paperwork once the client has decided.
If the joint probability density function for X and Y is 43. Find the average value of the function f(x,y,z) = xyz over
the solid sphere x2 + y2 + z2 < 1.
\ ^e~x/3°e-y/w if x > 0, y > 0 44. In a psychological experiment, x units of stimulus A and y
/(JC’J,)= jo otherwise units of stimulus B are applied to a subject, whose perfor¬
mance on a certain task is modeled by the function
Find the probability that the entire transaction will take less
than half an hour. f(x, y) = 10 + xyex~x2~y2
39. Modeling Problem Racing yachts, such as those in the Amer¬
ica’s Cup competition, benefit from sophisticated, computer- Suppose the stimuli are controlled in such a way that the
enhanced construction techniques.* For example, define the subject is exposed to every possible combination (x, y) with
center of pressure on a boat’s sail as the point (x, y) where all x > 0, y > 0, and x + y < 1. What is the subject’s average
aerodynamic forces appear to act. Suppose a sail occupies the response to stimuli?
triangular region R in the plane, as illustrated in Figure 12.56,
and that x and y are modeled by the formulas The radius of gyration for revolving a region R with mass m about
an axis of rotation L, with moment of inertia I, is
// xydA ff y2dA
ffydA y ~ JfydA
R R
Calculate the center of pressure on this sail. Note that if the entire mass m of R is located at a distance d from
the axis of rotation L, then R would have the same moment of in¬
ertia. Use this definition in Problems 45^17.
48. EXPLORATION PROBLEM A solid has the shape of the rectan¬ b. Explain why the integral in part a simplifies to the formula
gular parallelepiped given by — a < x < a, — b < y < b,
Work — 8Vh — Force x distance
—c<z<c, and its density is 6(x, y, z) = x2y2z2.
a. Guess the location of the center of mass and value of the where h is the distance between the center of mass and the
moment of inertia about the z-axis. level to which the material is lifted,
b. Check your response to part a by direct calculation. c. Picture a cylindrical tank of radius 6 ft and height 10 ft po¬
sitioned so its center of mass is at the origin. Compute the
49. Modeling Problem An industrial plant is located on a narrow
work required to lift (that is, pump) the contents (of den¬
river. Suppose Co units of pollutant are released into the river
sity 8) to a level of 15 ft above the center of mass. Compute
at time t — 0 and that the concentration of pollutant t hours
this two ways: by the integral in part a (suggestion—do the
later at a point x miles downstream from the plant is modeled
dx dy integral by inspection), and also by simply lifting the
by the diffusion function
center of mass to the required height. These results should
agree.
Co c-xi/(ut)
C(x, t) 52. Exploration Problem Newton’s inverse square law ■ Recall
y/knt
that if two masses m and M are each concentrated at (or nearly
where k is a physical constant. at) a point and are separated by a distance p, then the attrac¬
a. At what time rm(x0) does the maximum pollution occur at tive force is expressed by F = GmM/p2. What physicists
point x = x0 miles from the plant? What is the maximum and others prefer to do in the case of real-life masses (which
concentration Cm(xo)? occupy some volume) is to use p, the distance between the
two centers of mass. The question you are to explore here (and
b. Define the danger zone to be the portion of the riverbank
again in Section 12.7) is whether, and when, this simplifying
such that 0 < x < xm where xm is the largest value of x
way of computing attractive forces is justified,
such that Cm(xm) > 0.25Co. Find xm.
a. Suppose one point mass, m, is located at (0, 0, h) and the
c. Set up a double integral for the average concentration of
second, M, at (0, 0. 0). We are interested in the total re¬
pollutant over the set of all (x, t) such that 0 < t < tm(x)
sultant force that M exerts on m, and we assume this to be
for each fixed x between x = 0 and x = xm.
in the z-direction. Hence we will sum the vertical compo¬
d. How would you define the “dangerous period” for the pol¬ nents of the forces acting on m. Consider, in M, a small
lution spill? element of volume AV (at x, y, z) and argue that the mag¬
50. Modeling Problem The stiffness of a horizontal beam is mod¬ nitude of the force exerted on m and its vertical component
eled to be proportional to the moment of inertia of its cross are
section with respect to a horizontal line L through its centroid, Gm AV
as illustrated for three shapes shown in Figure 12.57. AF'mag —
GmAV(h — z)2
AF =
CYLINDRICAL COORDINATES
Solution
We use the conversion formulas x = r cos 6 and y = r sin 9.
= r2(cos2 9 + 3 sin2 9)
= r2(l + 2 sin2#) ■
D = {(x,y, z) such that u(x, y) < z < v(x, y) for all (x, y) in A]
Then
,Z
X
A = {(r,9) such that g\(9) < r < g2(9) for a < 9 < fi)
Figure 12.61 Volume element in Transforming the given integral to cylindrical coordinates yields the form shown in the
cylindrical coordinates following box.
12.7 Cylindrical and Spherical Coordinates 835
Triple Integral in Cylindrical Let R be a solid with upper surface z = v{r, 9) and lower surface z = u(r, 9),
Coordinates and let D be the projection of the solid onto the xy-plane expressed in polar
coordinates. Then, if f(x,y,z) is continuous on R, the triple integral of /
over S is given by
rrr rP rgi(e) rv(r,6)
/// f(x,y,z)dV = / / / f(rcos6,rs'm9,z)rdzdrdd
JJJ Ja Jg\(6) Ju(r,6)
Solution
Let S be the region occupied by the solid as shown in Figure 12.63. This surface is
most easily described in cylindrical coordinates.
Cylinder: Cone:
x2 + y2 = 2y z — y/x2 + y2
Figure 12.63 The solid bounded by
r2 = 2r sin# z — r
x2 + y2 — 2y, z = \Jx2 + y2, and
r — 2sin0
the xy-plane
Since the region S lies in the first octant, we have 0 < 9 < j, so S may be described
by
0<z<r 0<r<2sin(9 0 <9 < —
—
rn/2
/
r2s\n9
r2 dr d9 =
r•tt/2 r3 r=2 sin 6
d9 = -
8 r/2 ,
sin3 9 d9
Jo Jo Jo 3 Jo
= -fJor'1
8
3
(1 — cos“ 9) sin 9 d9
jt/2
8 cos3 9 16
— cos 9 +
3 3
836 Chapter 12 Multiple Integration
Solution
The solid S is shown in Figure 12.64. Because the solid is bounded by a cylinder, we
will carry out the integration in cylindrical coordinates.
Cylinder: Paraboloid:
Figure 12.64 The solid bounded by
x2 + y2 = a2 and z = x2 + y2 x2 T" y2 = a2 z = x2 + y2
r2 = a2
r — a (a > 0)
Let (x, y, z) denote the centroid. Using symmetry and the fact that the density function
is constant, we have x — y — 0. Let m denote the mass of S. Since the projected region
is r = a for 0 < 9 < 27T, we find that
z —
Iff zr dz dr dO
pin
Jo
pa prz
zrdzdrdO
Jo Jo _= Ia _ a
K 6 7
Iff 3
m pin pa prz
r dz dr d6
r dz dr dO
Jo Jo Jo
SPHERICAL COORDINATES
Figure 12.65 The spherical coordinate In spherical coordinates we label a point P by a triple (p, 6, 0), where p, 9, and 0
system are numbers determined as follows (refer to Figure 12.65):
Dallas-Fort Worth 9 — the polar angle (as in polar coordinates); we require 0 <9 <2tt.
33°N latitude,
97° W longitude;
0 = the angle measured down from the positive z-axis to the ray from
spherical the origin through P; we require 0 < 0 < it.
coordinates
You might recognize that spherical coordinates are related to the longitude and
latitude coordinates used in navigation. To be more specific, consider a rectangular
coordinate system with the origin at the center of the earth, with the positive z-axis
passing through the north pole and the xz-plane passing through the prime meridian.
Then, a particular location on the surface is denoted by (p, 9, 0), where p is the dis¬
tance from the center of the earth, 9 is the longitude, and y — 0 is the latitude (since
latitude is the angle up from the equator). For example, Dallas-Fort Worth has coordi¬
nates (p, 9, 0) = (3950, 263°, 57°), as shown in Figure 12.66.
Spherical coordinates are desirable when representing spheres, cones, or certain
planes. Some examples are shown in Figure 12.67.
We use the relationships in Figure 12.67 to obtain the remaining conversion for¬
Figure 12.66 Spherical coordinates mulas. All of these formulas can be derived by considering the relationships among
on the earth’s surface the variables as shown in Figure 12.68.
12.7 Cylindrical and Spherical Coordinates 837
to
+
Cylindrical to spherical:
II II
Wr2 4- z2/
Solution
a. Because p = y/x2 + y2 + z2, we see x2 + y2 + z2 — p2, so we can write
2 2
p = a
p = a Because p > 0
2 i 2
b. z = x +y
p cos 0 = (p sin 0 cos 9)2 -f (p sin 0 sin 0)2
Z,
Using this formula, we can form partitions and take a limit of a Riemann sum as the
partitions are refined to obtain the integral form shown in the following box.
Triple Integral in Spherical If / is continuous on the bounded solid region S, then the triple integral of /
Coordinates over S is given by
, y, z)dV
Solution
It seems clear that we should work in spherical coordinates with the origin of the
coordinate system at the center of the sphere, because the equation of the sphere is
p = R for 0 < 9 < 2n and 0 < 0 < n.
2n
V =
III"-L
S
sin (f)dp d(p d6
-v-'
dV
12.7 Cylindrical and Spherical Coordinates 839
Axis of symmetry A toy top of constant density <$0 is constructed from a portion of a solid hemisphere
with a conical base, as shown in Figure 12.70. The center of the spherical cap is at the
point where the top spins, and the height of the conical base is equal to its radius. Find
the moment of inertia of the top about its axis of symmetry.
Solution
We use a Cartesian coordinate system in which the z-axis is the axis of symmetry of the
top. Suppose the cap is part of the hemisphere z = yjR2 — x2 — y2. Since the height
of the conical base is equal to its radius, the cone makes an angle of jz/4 radians (45°)
with the z-axis. Let S denote the solid region occupied by the top. Then the moment
Figure 12.70 Moment of inertia for of inertia Iz with respect to the z-axis is given by
a spinning top
The shape of the top suggests that we convert to spherical coordinates, and we find that
S can be described as
Also,
I? =
ill (x2 + y2)So dV
rn/4 p2n pR
jt/4
f
Jo
{2ti — 0) sin3 0 dcj)
2 R5S0k 1 2R^8qji
COS 0 H-COS 0 l-V2+-^V2+\
2 12
3
- 5>/2) w 0.0912S80R5
It was clear in Example 5 that we should use the spherical coordinate system to
find the volume of a sphere. However, sometimes it is not obvious which coordinate
system to use. We address this question with the following example.
r1 r \J 1 — x2 i'y/2—x2—y2
1=1 / ,_ / zdzdydx
7-1 J-y/T^d J.x2+y2
Solution
The region of integration is D, where
z = x2 + y2 and x2 + y2 + z2 = 2
so
z + z2 = 2 Substitute the first into the second.
(z-l)(z + 2) = 0
z = 1 Reject z = —2, since z > 0.
It follows that the projected region in the xy-plane is the disk x2 + y2 < 1 (see Figure
12.71).
It may seem that transforming to spherical coordinates may result in a less com¬
plicated integration, since the upper boundary of D has the simple form p = V2, but
Figure 12.71 The solid D and the
the lower boundary z = x2 + y2 has the form
projected region in the xy-plane
9 9 9
z =* +y =r
'y
p cos 0 = (p sin 0)
cos 0 , , ,
p = -»— = cot 0 CSC 0
sin 0
Moreover, for 0 < 0 < we have 0 < p < y/2, but 0 < p < cot 0 esc 0 for
f < 0 < y, which means we need two integrals to evaluate the integral in spherical
coordinates:
However, if we use cylindrical coordinates, the integral has the relatively simple form
= / / / zrdzdrdO— / r [jZ2]| 2 r dr dO
Jo Jo Jr2 Jo Jo r
p2n p1 p2n
= / i {r[a-r’1)-ri\drde = {i [r - Jr3 - lr‘] |„
JO JO JO
f
-1 (A) JO = ^(2tt) =
In Problems 7-10 convert from cylindrical coordinates to where R is the cylindrical solid x2 + y2 < 1 with 0 < z < 1.
a. rectangular b. spherical
33. dzdr dO
Jo Jo Jo zr
psind 4 cos 6
41. Find the centroid of the solid bounded by the surface
34. r dz dr dO
J—tt/4 Jo Jo z — Jx2 4- y2 and the plane z = 9.
l'n/2 pcos9 r 1—r2
z,,
35. r sin 0 dz dr dO 9
Jo Jo Jo
l'n/2 pcosO r <t>
44. Find the moment of inertia about the z-axis of the portion of cave has a beehive shape, something like the part of the sur¬
the homogeneous hemisphere z = a/4 — x2 — y2 that lies be¬ face p = 4(1 + cos0) that lies above the xy-plane (where p
tween the cones z — Jx2 + y2 and 2z = Jx2 + y2. Assume is measured in ft). If the Spy needs at least 10 minutes to free
5=1. himself, and if water is entering at the rate of 25 ft3/min, will
he drown or survive?
Recall (from Problems 40^43 of Section 12.6) that the average
Q 58. In Problem 52 of Section 12.6, we noted that it is not always
value of a function f(x, y, z) over a solid region D is given by
accurate to compute the attractive force between two masses
1 by simply using the distance between their centers of mass.
AVERAGE VALUE = z)dV This approximation was applied first in astronomy, where the
VOLUME OF D
bodies are typically spherical in shape. As we will see, in this
case the simple calculation is very appropriate. Consider a
Use this definition in Problems 45—46. sphere of radius a with density S centered at the origin and a
45. Find the average value of the function f(x, y, z) = x + y + z point mass m located at (x, y, z) = (0, 0, R). In this prob¬
over the sphere x2 + y2 + z2 = 4. lem you are to set up the integral to compute the attractive
46. Exploration Problem What do you think the average values of 0 force between the masses; and in the next problem you will
and 0 are over the solid sphere p < a for a > 0? Prove your do some computing.
conjecture. a. Show that the distance p between point
47. Find the mass of the torus p = 2 sin 0 if the density is <5 = p.
(x, y, z) = (0, 0, R) and a point in the sphere, (p, 0, 0),
can be expressed p2 = R2 + p2 — 2Rpcos0. Hint: Use
48. Evaluate JJJ f'x1 + y2 + z2 dx dy dz where R is defined by the law of cosines.
b. In the sphere, consider a small element of volume AV at
x2 + y2 + z2 < 2. point (p, 9, 0). Argue that the magnitude of the force ex¬
erted on m, and the vertical component, are
49. Evaluate JJJ (x2 + y2 + z2) dx dy dz. where R is defined by
A Fmag = Gm(8AV)/p2
Note that the term (R — p cos 4>)/p projects the force vec¬
50. Evaluate JJJ z2 dx dy dz where D is the solid hemisphere tor onto the z-axis, giving the vertical component. Do you
see why?
x2 + y2 + z2 < 1, z > 0. c. Form the Riemann sum involving A F and get the follow¬
dx dy dz ing for the force acting between point mass m and the
51. Evaluate
/// Jx2 + y2 + z2
where D is the solid sphere sphere of radius a:
Gm8(R — p cos fi)dV
x2 + y2 + z2 < 3.
Find the volume of the solid D given in Problems 52-56 by using
integration in any convenient system of coordinates. 2 {R — pcos0) sin0
= 2nGmS
52. D is bounded by the paraboloid z = 1 — 4(x2 + y2) and the P (R2 + p2- 2Rp cos0)3/2 d(t> P
xy-plane.
Note: The integrand is independent of 0.
53. D is bounded above by the paraboloid z = 4—(x2+v2), below 59. Exploration Problem
by the plane z = 0, and laterally by the cylinder x2 + y2 = 1.
a. For an unspecified sphere of radius a and R > a, com¬
54. D is the intersection of the solid sphere x2 4- y2 + z2 < 9 and pute the attractive force between the masses in the previous
the solid cylinder x2 + y2 < 1. problem using the center of mass calculation.
55. D is the region bounded laterally by the cylinder r — 2sin#, b. Set a and R to numerical values and compute the attractive
below by the plane z = 0, and above by the paraboloid force by the integral in part c of the previous problem. Try
Z = 4 - r2. two or three different values of R until you are satisfied that
56. D is the region that remains in the spherical solid p < 4 after the center of mass “approximation” is exact in this case.
the solid cone </> < f has been removed. c. Compare this result with the corresponding calculations of
57. Spy Problem The Spy figures out Coldfinger’s puzzle (Supple¬ Section 12.6 and conjecture why the center of mass argu¬
mentary Problem 94 of Chapter 11), but just as his bonds come ment is sound in the case of the sphere and not for the rec¬
free, the door bursts open and a group of thugs rush in, led by tangular region.
... the Flamer! “Why?” pleads the Spy as he is tied wrist 60. Let D be a homogeneous solid (with density 1) that has the
to ankle, even more securely than before. “Don’t you know?” shape of a right circular cylinder with height h and radius r.
gloats the former Chief of Spies. “1 thought you knew every¬ Use cylindrical coordinates to find the moment of inertia of S
thing! You with your fancy suits and fancy cars and fancy about its axis of symmetry.
... other things—let’s see how fancily you die!” He pulls a 61. Find the sum Ix + Iy + /- of the moments of inertia of the solid
lever by the door, and with a mocking salute to the Spy, leaves sphere
with his men. The cave begins to fill with water. The Spy is
x2 + y2 + < 1
able to pull himself into a sitting position, with his nose 3 feet
above the floor. He looks around desperately and sees that the about the coordinate axes. Assume the density is 1.
12.8 Jacobians: Change of Variables 843
62. How much volume remains from a spherical ball of radius a a. by direct evaluation (exact value)
when a cylindrical hole of radius b (0 < b < a) is bored out
b. by calculator or computer (approximate value)
of its center? (See Figure 12.72.)
JJ f(x, y) dy dx = JJ f[g(u,v),h(u,v)]\J(u,v)\dudv
D D*
where
dx dx
du dv dx dy dy dx
J(u, v) =
dy dy du dv du dv
du dv
is nonzero and does not change sign on D*. The mapping factor J(u,v) is called the
d(x, y)
Jacobian and is also denoted by-.
d(u, v)
Proof A formal proof is a matter for advanced calculus, but a geometric argument
for this theorem is presented in Appendix B. □
Solution
The Jacobian of the change of variables is
dx dx
d(x, y) dr d9 cos 9 —r sin@
r cos2 6 + r sin2 9 = r ■
d (r,0) dy dy sin# r cos 9
dr d9
where D* is the region in the (r, 9) plane that is mapped into the region D in the (A, y)
plane by the polar transformation x = rcos9, y — rsin9, as illustrated in Figure
12.75.
12.8 Jacobians: Change of Variables 845
pute the Jacobian —L The Jacobian ^ needed for converting the integral
3(*, y) d(u,v)
ff f(x,y)dydx can then be computed by the formula
D
3U, y) 3(«, v)
= 1
3(u, v) d(x, y)
Solution
It is not easy to express x and y in terms of u and v. However,
3u du
3(w, v) dx dy y x
-2y2 — 2x2 = -2 (x2 + y2)
d(x, y) dv dv 2x —2 y
dx dv
Since
3(s, y) d(u, v)
Therefore, since = 1, we have
3(u, v) 3(x, y)
d(x, y)
3(u,v) 3(u, v) -2(x2 + y2) 2y/Au2 + v2
d(x, y)
x —y
Compute
//
D
x + y.
dy dx, where D is the triangular region bounded by the line
Solution
This is a rather difficult computation if no substitution is made. The form of the integral
suggests we make the substitution
u — x — y, v = x +y
d(x, y) 1
3 (w, v) 2
To find the image D* of D in the wu-plane, note that the boundary lines x = 0 and
y = 0 for D map into the lines u — — v and u — v, respectively, while x + y = 1
maps into v = 1. Therefore, the transformed region of integration D* is the triangular
region shown in Figure 12.76b, with vertices (0, 0), (1, 1), and (-1, 1).
Vi
u = : -V u -V
V =1 /
In Example 3, the change of variables was chosen to simplify the integrand, but
sometimes it is useful to introduce a change of variables that simplifies the region of
integration.
■// dy dx
circular disk.
we consider the substitution u = - and v = which will map the elliptical region E
a b
onto the circular disk
C:u2 + v2 < 1
JJ dydx = JJ abdudv
e c
du dv
= abfJ
=abLf dA
Solution
If ,4 is the area of D*, the centroid is (x, y), where
= -JJ D*
xdA and y
= 'JJ D*
y dA
We wish to find a transformation that simplifies the boundary curves of the region of
integration, so we let u = - and v = xy and the boundaries of the transformed region
D become
u — \ u = 1 V — 1 u = 5
1 5
6— 4- ,_ «=-
V,
0 k ~2
V == 5
V 1
_1
' 1 0 2 0 ' ' 30 4.0 u\
Solving the equations u = y/x and v — xy for x and y, we obtain x = y/v/u and
y = yfuv, so the Jacobian is
dx dx
9(*. y) 3u dv 1 -1
3 (u, v) 9y 9y 4u 4u 2u
3u dv
The change of variables formula for triple integrals is similar to the one given for
double integrals. Let T be a change of variables that maps a region R* in uvw-space
onto a region R in xyz-space, where
d(x, y, z)
= JJJ f[x(u,v,w),y(u,v,w),z(u,v,w)] dudvdw
3(u, v, w)
R*
Solution
The conversion formulas from rectangular coordinates to spherical coordinates are
cos 0 0 —psin0
JJJ f(x,y,z)dzdx dy
R
277/ x-y
x +y
dy dx
28' // (
•* - y
X +y
dy dx
15. u = ye~x, v = ex the square D in the xy-plane with vertices (0, 0), (1, —2), (3, —1),
y (2, 1) is mapped onto a square in the uv-plane. Use this informa¬
16. u = xy, v = — for x > 0, y > 0
x tion to find the integrals in Problems 33-38.
17. u — x2 y2, v = x2 ■r
■//(-?
2x + y
y dy dx
18. u = r. v 2y+ 5
x2 + y2’ ~ x2 + y2 D
35.
If D
(2x + y)2(x — 2y) dy dx
36 JJ D
s/(2x + y)(x — 2y) dy dx
o'
,C
X
JJ
37.
D
(2x + y) tan_1(x — 2y) dy dx
u = x + y, v = x- y
387/ cos(2x + y) sin(x — 2y) dy dx
39. Evaluate
,{2y-x)/{y+2x)
dA
//
where R is the trapezoid with vertices (0, 2), (1, 0), (4, 0),
and (0, 8).
40. Let R be the region in the xy-plane that is bounded by the
parallelogram with vertices (0,0), (1,1), (2, 1) and (1,0).
23. Suppose the wu-plane is mapped onto the xy-plane by the Use the linear transformation x = « + u, y = uto compute
equations x = u( 1 — v), y = uv. Express dx dy in terms ff(2x — y)dy dx.
of dudv. R
42. Under the change of variables x = s2 — t2, y = 2st, the quar¬ 52. A rotation of the xy-plane through the fixed angle 9 is given
ter circular region in the .sv-plane given by s2 4-12 < 1, s > 0, by
t > 0 is mapped onto a certain region D* of the xy-plane.
x — u cos 9 — v sin 0
Evaluate
y = u sin 9 -(- v cos 9
dy dx
(See Figure 12.79.)
7^+T2
fftxp{-x^-y*)dydx
x y
— H—- = 1, in the first quadrant
a2 b2
Figure 12.79 Rotational transformation
44. Evaluate
2x — 5y 3(x, y)
a. Compute the Jacobian ———.
3x + y d(u, v)
D* b. Let E denote the ellipse
where D* is the region bounded by the lines 2x — 5y = 1,
x2 4- xy + y2 = 3
2x — 5y = 3, 3x + y = 1, 3x + y = 4.
45. Evaluate Use a rotation of | to obtain an integral that is equivalent
to // y dy dx. Evaluate the transformed integral.
e (4*2+5^ dA
E
53. Find the area of the rotated ellipse
2 2
x y 5x2 - 4xy + 2y2 = 1
where D* is the elliptical disk — + — < 1.
Note that 5x2 - 4xy + 2y2 = Au2 + Bv2, where A and B are
46. Evaluate
constants and u — x 4- 2y, v — 2x — y.
54. Find the Jacobian of the cylindrical coordinate transformation
y3(2x — y) cos(2x — y) dy dx
x = rcos0, y = rsin0, z—z
where D* is the region bounded by the parallelogram with 55. Exploration Problem Let R be the region bounded by the par¬
vertices (0, 0), (2,0), (3, 2), and (1, 2). allelogram with vertices (0,0), (2,0), (1, 1), and (—1, 1).
Show that under a suitable change of variables of the form
47. Find the centroid of the region D* bounded by the curves
x = au + bv, y — cu + dv, we have
y = x, y — 2x, xy3 = 3, and xy3 = 6.
48. Evaluate
? VV dA Jlf(x + y)dy dx f(t) dt
a1 + V1+72 ~ x = 5.
51. Evaluate
x -y
// In
x 4- y
dy dx
57. IIS I'ORICAL QUEST Karl G. For this Quest, set up an integral for the circumference of
Jacobi was a gifted teacher and one the ellipse
of Germany's most distinguished math¬ jc2 y2
ematicians during the first half of the
a1 b2
nineteenth century. He made major
contributions to the theory of ellip¬ This is an example of an elliptic integral.
tic functions, but his work with func¬ 58. Let T: x — x(u, v), y = y(u, v) be a one-to-one transfor¬
tional determinants is what secured ml *W7.> mation on a set D so that T-1 has the form u = u(x,y),
his place in history. In 1841 he pub¬ KARL G. JACOBI v = v(x, y). Use the multiplicative property of determinants,
lished a long memoir called "De de- 1804-1851 along with the chain rule, to show that
terminantibus functionalibus," devoted
to what we today call the Jacobian and pointing out that this d(x, y) d(u, v)
determinant is in many ways the multivariable analogue to the 3(u, v) d(x, y)
differential of a single variable. The memoir was published in
«what is usually known as Crelle's Journal, one of the first jour¬ You may assume that all the necessary partial derivatives exist.
nals devoted to serious mathematics. It was begun in 1826 by
August Crelle (1780-1855) with the title Journal fur die reine Properties of determinants are found in Section
und angewandte Mathematik. ■ 2.6 of the 5tudent Mathematics Handbook.
Proficiency Examination 18. Complete the following table for the conversion of coordi¬
nates.
CONCEPT PROBLEMS
1. Define a double integral.
(x, y, z) (r, 0, z) (p,0,<p)
To Rectangular Cylindrical Spherical
2. State Fubini’s theorem over a rectangular region. From
3. What is a type I region? State Fubini’s theorem for a type 1 (x, y, z) X =X r = ? p= 7
region.
Rectangular y = y tanfl = 7 tan0 = 7
z —z z= 7. c/>= 7
4. What is a type II region? State Fubini’s theorem for a type II 0r, e, z) r —r
x — '■ P = ■
region. Cylindrical y= ? o=e 9= 7
5. State the formula for finding an area using a double integral. z— 7 z — z 7
(p.0.<P) X= 7 r = 7 P =P
6. State the formula for finding a volume using a double integral. Spherical 9= 7 9=6
y= 7
7. State the following properties of a double integral. z — 7. z= .7 .. 4>-d
a. linearity rule
19. a. State the formula for a triple integral in cylindrical coordi¬
b. dominance rule nates.
c. subdivision rule b. State the formula for a triple integral in spherical coordi¬
8. What is the formula for a double integral in polar coordinates? nates.
9. State the double integral formula for surface area. 20. Define the Jacobian for a mapping x — x(u, u), y = y(u, v)
10. State the formula for the area of a surface defined parametri¬ from u, v variables to x, y variables.
cally. 21. State the formula for the change of variables in a double inte¬
11. State Fubini’s theorem over a parallelepiped in space. gral.
12. Give a triple integral formula for volume.
PRACTICE PROBLEMS
13. Explain how to use a double integral to find the mass of a pla¬
r*n/3
rn/i psiny
/>s
nar lamina of variable density.
22. Evaluate e * cos ydxdy.
14. How do you find the moment of a lamina with respect to the Jo Jo
x-axis?
15. What are the formulas for the centroid of a region?
23. Evaluate (x + y — z)dxdy dz.
m
16. Define moment of inertia.
24. Use a double integral to compute the area of the region R that
17. What is a joint probability density function? is bounded by the x-axis and the parabola y = 9 — x2.
Chapter 12 Review 853
«jt/2
ex/* (*n/2
r*
25. Use polar coordinates to evaluate dx dy cos(x + y) dx dy
/ 1-x2
'■a: X2 + y2 Jo Jo
26. Use a triple integral to find the volume of the tetrahedron that 3 r9-x2 xey/3
is bounded by the coordinate planes and the plane dy dx 8 x3 dx dy
'■n 9 — y 77'^
Jo J^/y
components will be usable as long as either one of its com¬ 10. Ji-x*-? dy dx
ponents is still operating. The appliance carries a warranty
from the manufacturer guaranteeing replacement if the appli¬ Evaluate the integrals given in Problems 11-34.
ance becomes unusable within one year of the date of pur¬
chase. To model this situation, let the random variables X 11. [ [ eyZ dy dx
and Y measure the life span (in years) of the first and second Jo J J:
components, respectively, and assume that the joint probabil¬ 2 t-2
ity density function for X and Y is
m
zdz dy dx
fix■y) = jo otherwise
\J 1— X
Suppose the quality assurance department selects one of these
dy dx dz
appliances at random. What is the probability that the appli¬
e
ance will fail during the warranty period?
m
28. Set up and evaluate a triple integral for the volume of the solid 15. / / / r2 sin f dr d6 df
region in the first octant that is bounded above by the plane Jo Jo Jo
z = 4x and below by the paraboloid z — x2 + 2y2. See Figure y
12.81.
n .x2
/
Jo
x2y dzdy dx
/>!n at
xez dz dy dx
p\ pl+x pxy
18. / / / xzdzdydx
Jo JI —x Jo
/ • V3
v/3
/
rj3-y2
r \/s—y r9
/ y2dzdxdy
-V3 J-^/l-y2 J(x2+y2)2
Hint: Change to cylindrical coordinates.
where R is the triangular region with vertices (0, 0), (2. 0), D
n > 0
(1, 1).
Supplementary Problems
M-If — dA, where D is the region above the line x + y = 1
D
In Problems 1-10, sketch the region of integration, exchange the and inside the circle x2 + y2 = 1
order, and evaluate the integral using either order of integration.
1. x2y2 dy dx
25. If x3v/4 - y2 dA, where D is the circular disk x2 + y2 < 4
854 Chapter 12 Review
267/ x3 dA, where D is the shaded portion shown in Figure 35. Set up (but do not evaluate) an integral for the mass of a lam¬
D
ina with density 8 that covers the region outside the circle
12.82 r — s/2 a and inside the lemniscate r2 = 4a2 sin 20 in the
first quadrant.
y/xy
7t/2
/•JT/Z ff 2a cos 0
r sin 20 dr d9
Jo Jo
277/ D
xy2 dA, where D is the shaded portion shown in Figure
a: ex/y2 dx dy
12.83
and evaluate using either order.
n
39. Express the integral
fix, y)dxdy + / /
J i Jo
/>4 o(4-y)/3
fix, y) dxdy
f-Jy rl r2-y
/ / fix, y) dxdy + / / fix, y) dxdy
Jo J-yy Ji 7-i
Figure 12.83 Problem 27
as a double integral with the order of integration reversed.
y-x
28./ / exp 41. Use a double integral to find the area inside the circle
’•// D
y+x
dydx, where D is the triangular region
r = cos9 and outside the cardioid r = 1 - cos#.
with vertices (0, 0), (2, 0), (0, 2) Hint: Make a suitable change 42. Use a double integral to find the area outside the cardioid
of variables. r = 1 + cos# and inside the cardioid r — 1 + sin#.
d2f
29-if dxdy
dx dy, where R is the rectangle a < x < b.
43. Use a double integral to find the area outside the small loop of
the lima^on r = 1 + 2 sin # and inside the large loop.
44. Use a double integral to compute the volume of the tetrahe¬
3 2f .
c < y <d and is continuous over R with f (a, c) = 4, dral region in the first octant that is bounded by the coordinate
dxdy
planes and the plane dx + y + 2z — 6.
f(a, d) = -3, f(b, c) = 1, f{b, d) = 5
45. Use a double integral to find the volume of the solid that is
30. IJJ -y/x2 + y2 + z2 dV, where D is the portion of the solid bounded above by the paraboloid z = x2 + v2 and below by
D the square region 0<x < l,0<y< 1,^ = 0.
sphere x2 + y2 + z2 < 1 that lies in the first octant 46. Find the volume of the solid that is bounded above by the sur¬
face z — xy, below by the xy-plane, and on the sides by the
- /// z2 dV, where H is the solid hemisphere x2 + y2+z2 < 1 circular disk x2 + y2 < a2 ia > 0) that lies in the first quad¬
rant.
with z > 0
47. Find the volume of the solid that is bounded above by the
dV
32
'■ SIS yxTTT?
, where H is the solid hemisphere paraboloid z = 4 — x2 — y2 and below by the plane z = 4 —2x.
48. Find the surface area of that portion of the cone z = yjx2 + y2
x2 + y2 + z2 < 1, with z > 0 that is contained in the cylinder x2 + y2 = 1.
31
■SSh
''/ 5
n
y2 _|_ z2 ’ D is the solid region bounded be-
49. Find the surface area of the portion of the paraboloid
z — x2 + y2 that lies below the plane z — 9.
ow by the paraboloid 2z — x2 + y2 and above by the sphere 50. Find the surface area of the portion of the cylinder x2 +z2 — 4
t2 + y1 + z2 = 8 that is bounded by the planes x = 0, x = 1, y = 0. and y = 2.
51. Find the volume of the region bounded by the paraboloid
34. ///«(, “) 1 " dV, where D is the solid bounded by the y2 + z2 = 2x and the plane x + y = 1.
52. Find the volume of the region bounded above by the
surtac = x 2 + y2 and the plane z = 2 paraboloid z = 4 —x2 — y2 and below by the plane z + 2y = 4.
Chapter 12 Review 855
53. Find the volume of the region bounded by the ellipsoid 68. Use a change of variables after a transformation of the form
z2 = 4 — 4r2 and the cylinder r — cos 0. x = Au, y = Bv, z = Cw to find the volume of the re¬
54. Find the volume of the region bounded above by the hemi¬ gion bounded by the surface ~Jx + *j2y + V3z = 1 and the
coordinate planes.
sphere z = -y/9 — x2 — y2, below by the xy-plane, and on the
sides by the cylinder x2 + y2 — 1. 69. Find the centroid of a homogeneous lamina that covers the
part of the plane Ax + By + Cz = 1,A > 0, B > 0, C > 0
55. Show that the solid bounded below by the cone z = Jx2 + y2 that lies in the first quadrant.
and above by the sphere x2 + y2 + v = 2az for a > 0 has
70. A homogeneous plate has the shape of the region in the
volume V = na2.
first quadrant of the xy-plane that is bounded by the circle
56. Find the volume of the solid region bounded above by the x2 + y2 — 1 and the lines y = x and x = 0. Sketch the region
sphere given in spherical coordinates by p = a {a > 0) and and find x, the x-coordinate of the centroid.
below by the cone <p = (p0 where 0 < (p0 < |. 71. Let R be a lamina covering the region in the xy-plane that is
57. A homogeneous solid S is bounded above by the sphere p — a bounded by the parabola y — 1 — x: and the positive coordi¬
(,a > 0) and below by the cone <p = <p0 where 0 < (p0 < \. nate axes. Assume the lamina has density 6(x, y) = xy.
Find the moment of inertia of S about the z-axis. Assume a. Find the center of mass of the lamina.
6 = 1. b. Find the moment of inertia of the plate about the z-axis.
58. The region between the circles x2 + y2 — 1 and x2 + y2 = 4 72. A solid has the shape of the cylinder x2 + y2 < a, with
with 0 < z < 5 forms a “washer.” Find the moment of inertia 0 < z <b. Assume the solid has density 6(x, y, z) = x2 + y2.
of the washer about the z-axis if the density 6 is given by
a. Find the mass of the solid.
a. 8 — k (k a constant) b. 6(x, y) = x2 + y2
b. Find the center of mass of the solid.
c. 6(x,y)=x2y2 73. Use cylindrical coordinates to find the volume of the solid
59. Find x, the x-coordinate of the center of mass, of a triangular bounded by the circular cylinder r = 2a cos 6 (a > 0), the
lamina with vertices (0, 0), (1, 0), and (0, 1) if the density is cone z = r, and the xy-plane.
6(x, y) = x2 + y2. 74. Let u be everywhere continuous in the plane, and define the
60. Find the mass of a cone of top radius R and height H if the functions / and g by
density at each point P is proportional to the distance from P
to the tip of the cone. /(*) = f «(x, y) dy g(y) - ( u(x, y) dx
61. Find z, the z-coordinate of the centroid of a homogeneous
right circular cone with height H and base radius R. As¬ Show that
sume the cone is positioned with its vertex at the origin and the
pb pb
z-axis is its axis of symmetry. / / (x) dx = / g(y)dy
Ja Ja
62. Find the Jacobian V' U ) of the change of variables
3(x,y, z)
u = 2x — 3y + z, v = 2y — z, w — 2z. Hint: Show that both integrals equal Jj u(x. y)dA for a cer¬
3(u, v,w)ri , ,. ...
63. Find the Jacobian -—- of the change of variables tain region D.
3(x, y, z)
u = x2 + y2 + z2, v = 2y2 + z2, w = 2z2. 75. Find the volume of the solid region bounded by the surface
64. Let u = 2x - y and v = x + 2y. Find the image of the unit ^/3 + y2/3+z2/3=fl2/3
78. Show that if z — f{r, 9) is the equation of a surface S in polar Suppose the origin is at the center of the hemispherical dome.
coordinates, the surface area of S is given by If the entire solid is made of the same material with density
5 = 1, where is the centroid?
/(x, y) = 1 + f f
Jo Jo
f(u,v)dudv
Space-Capsule Design
This project is to be done in groups of three or four students. Each group will submit a
S
single written report.
uppose you were part of a team of engineers designing the Apollo space capsule.
The capsule is composed of two parts:
*MAA Notes 17 (1991), “Priming the Calculus Pump: Innovations and Resources,” by Marcus S. Cohen,
Edward D. Gaughan, R. Arthur Knoebel, Douglas S. Kurtz, and David J. Pengelley.
CONTENTS
13.1 Properties of a Vector Field: Divergence
and Curl
Vector Analysis
Definition of a vector field
Divergence
Curl
13.2 Line Integrals
Definition of a line integral
Line integrals with respect to .r, y,
PREVIEW
and z
Line integrals of vector fields In this chapter, we combine what we have learned about differentiation, integration,
Applications of line integrals: and vectors to study the calculus of vector functions defined on a set of points in R:
mass and work or M3. We introduce line integrals and surface integrals to study such things as
13.3 The Fundamental Theorem and Path fluid flow and then obtain a result called Green’s theorem that enables line integrals
Independence to be computed in terms of ordinary double integrals. This result is extended into
Fundamental theorem for line M3 to obtain Stokes’ theorem and the divergence theorem, which have extensive
integrals
applications in areas such as fluid dynamics and electromagnetic theory.
Conservative vector fields
Independence of path
13.4 Green's Theorem
Green’s theorem PERSPECTIVE
Area as a line integral
How much work is done by a variable force acting along a given curve in space?
Green’s theorem for
How can the amount of heat flowing across a particular surface in unit time be mea¬
multiply-connected regions
Alternative forms of Green’s
sured, and is the measurement similar to measuring the flow of water or electricity?
theorem We will use line integrals and surface integrals to answer these and other questions
Normal derivatives from physics and engineering mathematics.
13.5 Surface Integrals
Surface integration
Flux integrals
Integrals over parametrically
defined surfaces
13.6 Stokes'Theorem
Stokes’ theorem
Theoretical applications of Stokes'
theorem
Physical interpretation of Stokes’
theorem
13.7 The Divergence Theorem
The divergence theorem
Applications of the divergence
theorem
Physical interpretation of
divergence
Chapter 13 Review
Guest Essay: Continuous versus Discrete
Mathematics
William F. Lucas
Cumulative Review, Chapters 11-13
859
860 Chapter 13 Vector Analysis
The satellite photograph in Figure 13.1 shows wind measurements over the Atlantic
Ocean. Wind direction is indicated by directed line segments. This is an example of a
vector field, in which every point in a given region of the plane or space is assigned a
vector. Here is the definition of a vector field in R3.
Vector Field A vector field in R3 is a function F that assigns a vector to each point in its
domain. A vector field with domain D in R3 has the form
For example,
9 X
is a vector field with i-component 2xzy, j-component eyz, and k-component tan —.
Figure 13.1 A wind-velocity map of
the Atlantic Ocean A vector field in R2 can be thought of as a special case where there are no
Z-coordinates and no k-components. That is, a vector field in R2 has the form
To visualize a particular vector field F(x, y, z), it often helps to select a number of
points in the domain of F and then draw an arrow emanating from each point P(a, b, c)
with the direction of F(a, b, c) and length representing the magnitude ||F(a, b, c)||. We
will refer to such a representation as the graph of F. Here is an example involving the
graph of a vector field in R2.
Solution
We will evaluate F at various points. For example,
The graph of a vector field often yields useful information about the properties of
the field. For instance, suppose F(x, y) represents the velocity of a compressible fluid
(like a gas) at a point (x, y) in the plane. Then F assigns a velocity vector to each point
in the plane, and the graph of F provides a picture of the fluid flow. Thus, the flow in
Figure 13.4a is a constant, whereas Figure 13.4b suggests a circular flow.
Gravitational, electrical, and magnetic vector fields play an important role in phys¬
ical applications. We will discuss gravitational fields now, and electrical and magnetic
fields later in this section. Accordingly, we begin with Newton’s law of gravitation.
13.1 Properties of a Vector Field: Divergence and Curl 861
TECHNOLOGY NOTE
2
//// / / \ \ \ \ \W
//// / / / l \ \ WW
"/// / / \ \ \\\W
1 sss/ / / / \ \
/ / \ \ \
"——^ / / t \ \ % "x 'x-.-
0
V\ ( f / s y- - -
'x'xW \ \ I / / / S’ S’*'*
-1 \\W \ \ I///S/S
\\\\\\ t 1////s b. A circular flow
\\\\ \ \ I If////
-2 WWW !t /////
Figure 13.4 Flow diagrams
-2-10 1 2
which says that a point mass (particle) m at the origin exerts on a unit point mass
located at the point P(x, y, z) a force F(x, y, z) given by
Gm
F(x, y, z) = -y-— uU, y, z)
xz + yz + zz
where G is a constant (the universal gravitational constant) and u is the unit vector
extending from the point P toward the origin. The vector field F(x, y, z.) is called the
gravitational field of the point mass m. Because
Note that the gravitational field F always points toward the origin and has the same
magnitude for any point m located r = y/x2 + y2 + z2 units from the origin. Such a
vector field is called a central force field. This force field is shown in Figure 13.5a.
Other physical vector fields are shown in Figures 13.5b and 13.5c.
DIVERGENCE
Divergence and curl are two operations on vector fields that originated in connection
with the study of fluid flow. Divergence may be defined as follows.
Solution
d d
a. divF = —(x2y) H-(xy3) — 2xy + 3xy2
Bx ~ By
represents the velocity of a fluid with density 8(x, y, z) at a point (x, y, z) in a certain
div D > 0
region R in R3. Then the vector field 3 V is called the flux density and is denoted by
D. We can think of D = 3 V as measuring the “mass flow” of the liquid.
Assuming there are no external processes acting on the fluid that would tend to
create or destroy fluid, it can be shown that div D gives the negative of the time rate
Fluid flows from a source point. change of the density, that is,
div D < 0
div D =-
31
'Jv'
This is often referred to as the continuity equation of fluid dynamics. (A derivation is
given in Section 13.7.) When divD = 0, D is said to be incompressible. If divD > 0
Fluid flows toward a sink point. at a point (x0, yo, Zo), the point is called a source; if divD < 0, the point is called a
sink (see Figure 13.6). The terms sink, source, and incompressible apply to any vector
div D = 0 field F and are not reserved only for fluid applications.
A useful way to think of the divergence divV is in terms of the del operator
defined by
3 . 3 . 3
Fluid is incompressible. V = —i H-j H-k
3x 3yJ Bz
Figure 13.6 Flow of a fluid across Recall (from Section 11.6) that applying the del operator to the differentiable function
a plane region, D f{x,y,z) produces the gradient field
13.1 Properties of a Vector Field: Divergence and Curl 863
Similarly, by taking the dot product of the operator V with the vector field
V = u(x, y, z)i + y, z)j + w(x, y, z)k, we obtain the divergence
/ 9 9 9 \
v • V = —i + —j + —k j • («i + uj + ink)
9 9 9
- — (u) + — (u) + —(w)
dx dy dz
du dv dw
dx dy dz
= div V
CURL
The del operator may also be used to describe another derivative operation for vector
fields, called the curl.
dw\ / dv
curl V = k
dx / \9x
Note that
dv\ . /dw du \ . /dv
curl V = k
~dz)X~
9 9 9
V
9jc dy dz
u v w 4- V
= V x V
The determinant form of curl V is a convenient device for remembering the definition
and is helpful in organizing computations.
Solution
i J k
3 3 3
curlF =
dx dy 3z
x^yz xy^z xyz
3 2 3 2 V / 3 2 3 2 \ .
,z)j
+ xy2z * yz k
3x 3y
•* = (xz2 - xy2)i + (x2y - yz2)j + (y2z - x2z)k
i j k
3 3 3
curl G =
3x 3y 3z
x cos y xy2 0
9 2‘
i —
3 3 2 3 r
0 - —(xcosy) j + T~xy - — (xcosy)
dz dz _3x dy J
r\
— (y+ x siny)k
Solution
TECHNOLOGY NOTE
Derive, Maple, Matlab, and Mathematica will carry out most vector operations. For
Example 2, we define G(x, y, z) = xi + y3z2j + xz3k, and use one of these programs:
You may need to enter a zero, as with F(x, y) = x2yi + xy3j, as shown here:
Finally, consider the vector field F from Example 3, F = x2yzi + xy2zj + xyz2k. We
can find the curl:
The Laplacian Operator Let f(x,y,z) define a function with continuous first and second partial deriva¬
tives. Then the Laplacian of / is
Thus, / is harmonic.
TECHNOLOGY NOTE
Derive, Maple, M ATLAB, and Mathematica will find the Laplacian of a given function.
For Example 5, we obtain
LAPLACIAN0* cos(y), [x, y]) which gives 0
In many ways, the study of electricity and magnetism is analogous to that of fluid
dynamics, and the curl and divergence play an important role in this study. In electro¬
magnetic theory, it is often convenient to regard interaction between electrical charges
as forces somewhat like the gravitational force between masses and then to seek quan¬
titative measure of these forces.
One of the great scientific achievements of the nineteenth century was the discov¬
ery of the laws of electromagnetism by the English scientist James Clerk Maxwell (see
Historical Quest, Section 13.7, Problem 32). These laws have an elegant expression
in terms of the divergence and curl. It is known empirically that the force acting on a
866 Chapter 13 Vector Analysis
charge due to an electromagnetic field depends on the position, velocity, and amount
of the particular charge, and not on the number of other charges that may be present or
how those other charges are moving. Suppose a charge is located at the point (x, y, z)
at time t, and consider the electric intensity field E(x, y, z, t) and the magnetic in¬
tensity field H(x, y, z, t). Then the behavior of the resulting electromagnetic field is
determined by
Q
divE = — div(|iH) = 0
€
9 , 9E J
curlE =-(/xH) c“(curlB) =-1—
dt dt e
where Q is the electric charge density (charge per unit volume), J is the electric cur¬
rent density (rate at which the charge flows through a unit area per second), B is the
magnetic flux density, c is the speed of light, and p. and c are constants called the
permeability and permittivity, respectively. Working with these equations and terms is
beyond the scope of this course, but if you are interested there are many references you
can consult. One of the best (despite being almost 40 years old) is the classic Feyn¬
man Lectures in Physics (Reading, Mass.: Addison-Wesley, 1963), by Nobel laureate
Richard Feynman, Robert Leighton, and Matthew Sands.
13. F = (sinx)i + (cosy)j 14. F = (— cosx)i + (siny)j 37. If F(x, y, z) = xyi + yzj + z2k and G(x, y, z) = xi + yj -zk,
find curl(F x G).
15. F = xi — yj 16. F = — xi + yj
38. If F(x, y, z) = 2i + 2xj + 3yk and G(x, y, z) = xi — yj + zk,
17. F = ---■ i + -—---j
find div(F x G).
y/x2 + y2 Jx2 + y2
18. F = x2i — y2j 39. If F(x, y, z) = xyi + yzj + z2k and G(x, y, z) = xi + yj - zk,
find div(F x G).
19. F = axi + by] + ck for constants a, b, and c
20. F = (ex sin y)i + (ex cos y)j + k O40. Let A be a constant vector and let R = xi + yj + zk. Show
that div(A x R) = 0.
21. F = x2i + y2j + z2k
41. Let A be a constant vector and let R = xi + yj + zk. Show
22. F = yi + z] + xk
that curl(A xR) = 2A.
23. F = xyi + yzj + xzk
42. If F(x, y) = u(x, y)i + u(x, y)j, show that curlF = 0 if and
24. F = 2xzi + 2yz2j - k
9i>
25. F = xyzi + x2y2z2j + v2z3k only if —
dy 3x
13.2 Line Integrals 867
43. Consider a rigid body that is rotating about the z-axis (coun¬ 50. curl(/F) = / curl F + (V/ x F)
terclockwise from above) with constant angular velocity 51. div(/F) = / div F + (V/ • F)
<o = ai + bj -f ok. If P is a point in the body located at
52. curl(V/ + curlF) = curl(V/) + curl(curlF)
R = xi + yj + zk, the velocity at P is given by the vector field
V = (o x R. 53. div(/Vg) = / div Vg + V/ - Vg
a. Express V in terms of the vectors i, j, and k. 54. The curl of the gradient of a function is always 0. That is,
b. Find div Y and curl V. V x (V/) = 0.
44. Exploration Problem If F = {/, g, h) is an arbitrary vector field 55. The divergence of the curl of a vector field is 0. That is,
whose components are twice differentiable, what can be said div(curl F) = 0.
about curl(curl F)?
Q In Problems 56-60, R = (x, y, z), and
45. Which (if any) of the following is the same as div(F x G) for
r = ||R|| = y/x2 + y2 -f Z2- In each case, verify the given identity
all vector fields F and G?
or answer the question.
I. (div F) (div G) 56. curl R = 0; what is div R?
II. (curl F) G - F (curl G)
III. F(divG) + (divF)G 57. div^R^=0 58. curl ^Rj = 0
IV. (curl F) • G + F • (curl G)
2
In Problems 46-55, prove the given property for the vector fields 59. div (rR) = 4r 60. div(Vr) = -
r
F and G, scalar c, and scalar functions f and g. Assume that all 61. Exploration Problem State and prove an identity for
required partial derivatives exist and are continuous. div(V(/g)), where / and g are differentiable scalar functions
46. div(cF) = c div F of x, y, and z.
47. div(F + G) = div F + div G 62. Counterexample Problem Let F = (x2y, yz2, zy2). Either find a
48. curl(F + G) = curl F + curl G vector field G such that F = curl G, or show that no such G
49. curl(cF) = c curl F exists.
E /W*. V 4) A*
k= 1
Figure 13.7 The curve C partitioned
into subarcs and let || As|| denote the largest subarc length in the partition. Then, if the limit
lim
IIAsIKO E /(4. yl
k= 1
exists, we call this limit the line integral of / over C and denote f(x,y,z)ds.
868 Chapter 13 Vector Analysis
Line Integral over a Smooth If /(jc, y, z) is defined on the smooth curve C with parametric equations
Curve x = x(t), y — y(t), z = z(t), then the line integral of / over C is given by
provided that this limit exists. If C is a closed curve, we indicate the line integral
of / around C by <j) f ds.
It can be shown that the limit that defines the line integral J f ds always exists if
/ is continuous at each point of C. Also, since the curve C is smooth, the component
functions x(t), y(t), and z(t) will all be continuously differentiable. Thus, we have
ds = Vix'(t)]2 + [y'(t)]2 + [z'(t)]2 dt
so the line integral can be written entirely in terms of t:
Similarly, if fix, y) is a function of only two variables and C is a curve in the plane,
then
J f ix,y)ds = j" f ixit), yit))V[x'it)]2 + [y'O)]2 dt
Evaluate the line integral J x2z ds, where C is the helix x = cos t, y = 2t, z — sin t,
for 0 < t < n.
Solution
Since x\t) = — sin t, yft) = 2, z!it) = cos t, the line integral is
= / V5 cos2 t sin t dt
Jo
-V5
cos31
3
2^5
The definition of a line integral can be extended to curves that are piecewise smooth
in the sense that they are the union of a finite number of smooth curves with only
Figure 13.8 Piecewise smooth curve endpoints in common, as shown in Figure 13.8.
In particular, if C is comprised of a number of smooth subarcs C\, Cj,.. •, C„,
then
Evaluate the line integral J xyds, where C consists of the line segment Cj from
(—3, 3) to (0, 0), followed by the portion of the curve Ci: 16v = x4 between (0, 0)
and (2, 1).
Solution
The curve C is shown in Figure 13.9. The segment C\ is part of the line y = —x and
can be parameterized by the equations x = t, y — — t over the interval —3 < / < 0.
On this curve, we have
f3)
U) x\t) = 1, y'(t) = —1, so ds — y/ (l)2 + (— l)2 dt = \Pldt
and the line integral of /(x, y) = xy over C\ is
Cl
1 6y = T 4
9 = -X
'c2
£2
X
- - (0 ,0)
The curve C2, 16y = x4, can be parameterized by the equations x = 21,
Figure 13.9 The curve C in Example 2 y — -j^x4 = |L(21)4 = t4 for 0 < t < 1. We find that
x\t) = 2, y\t) — 413, so ds = y/(2)2 + (4r3)2 c/r
and the line integral over C2 is
- -(1 +4t6)3/2
9 0
= ^(5^5-1)
7
Proof The proof follows directly from the properties of limits and the definition of a
line integral. -J
870 Chapter 13 Vector Analysis
J f(x,y,z)dx = J f[x(t),y(t),z(t)]x'(t)dt
/8dy = f
J hdz — J h[x(t),y(t),z(t)]z'(t)dt
By combining the line integrals with respect to the coordinate variables x, y, and z, we
obtain a line integral of the form
J [y dx — zdy + x dz]
where C is the curve with parametric equations x = f , y = e z = for 0 < £ < 1
Solution
Since x'(t) — 2t, y'(t) = — e~‘, and z'(t) — e‘, we have
= I [Ite-1 + 1 +t2e*]dt
Jo
— [—2e 1 (t + 1) + t + e1 (t2 — 2t + 2)]|0
We will now discuss what it means to compute the line integral of a vector field.
Line Integral of a Vector Field Let F(x, y, z) = u(x, y, z)i + u(x, y, z)j + tu(x, y, z)k be a vector field, and let
C be a piecewise smooth orientable curve with parametric representation
Evaluate J F • c/R, where F = (y2 — z2)i + (2yz)j — ;c2k and C is the curve defined
parametrically by x = t2, y = 2t, and z = t for 0 < t < 1.
Solution
Rewrite F using the parameter t:
F = [(2/)2 - (f)2]i + [2(20(01 j - [02)2]k = 3t2i + 4rj - Ok
Because R(0 = t2i + 2/j + tk, we have dR = {It dt)i 4- (2dt)j + dtk, so
F dR = (3t2)(2t dt) + (4r2)(2 dt) + (~tA)(dt)
= (60 + 80 - t4)dt
Thus,
f
Jc
F dR = [ (60 + 8/2 — 0) dt
Jo
3 4 8 3
-0 + -0
119
"30"
A line integral over a curve C does not depend on the parameterization used for
C. This property of line integrals is illustrated in the following example.
Solution
a. With the parameterization * = 2 cos 0, y = 2 sin 0, we find that x'{6) — —2 sin 0,
Figure 13.10 Graph of the y'{6) = 2 cos 0
semicircle C
LFdR=L [y dx + x dy]
T
= 0
L 4 cos 26 d6
2 -4 + 2/2
dt
-L -2 \/4 — t2
—tyj4 "
-2
=0
This is the same result that was obtained using the parameterization in part a.
872 Chapter 13 Vector Analysis
Solution
The two paths we are considering are shown in Figure 13.11.
a. The line joining the given points has equation y = 2x, which may be parameterized
by setting x — t, y = 2t for 0 < t < 2. Thus,
[ F dR = [ 8t2dt = [2t4]|o = 32
Jc Jo
b. The parabola y = ;c2 can be parameterized by setting x = t, y — t2 for 0 < t <2.
Thus,
In terms of t,
and
In Example 6 we see that the value of the line integral is the same for both paths.
Indeed, it can be shown that for the vector field F = xy2i + Jc2yj, the line integral
Note: The "line" integral should be lc¥dR along any path C joining (0, 0) to (2, 4) has the same value. This is not
thought of as the "curve" integral. The
equation(s) of a curve are needed so true for every F (see Problem 58) but when it is true, the line integral is said to be
that the integral can be expressed in independent of path or path independent. Path independence is an important feature
terms of a single variable. of certain vector fields and will be discussed in detail in the next section.
Line integrals were developed in the nineteenth century, primarily to deal with prob¬
lems in physics involving force, fluid flow, electricity, and magnetism. We will show
how line integration can be used to compute the mass of a thin wire and the work per¬
formed by an object moving along a curve in a force field. Additional applications are
examined in the problem set and in Sections 13.3 and 13.4.
Consider a thin wire with the shape of a curve C and let 8(x, y, z) be the density
(mass per unit length) at each point P(x, y, z) on the wire. Suppose the curve is de¬
scribed by parametric equations x = x{t), y = y(t), z = z(t) for a < t < b, and
subdivide the parameter interval [a, b] into n equal parts. This induces a partition of
the portion of the curve covered by the wire into n subarcs. Let ASk be the length of the
&th subarc, and let Pk(x£, y%, z*k) be a point chosen arbitrarily from this subarc. Then,
the mass of the subarc is approximately
and the sum ^ Amk approximates the total mass of the wire. We improve the ap-
k=1
proximation by taking more and more subdivision points, and the actual mass of the
wire is given by the limiting value
If the wire has density 8(x, y, z) = xyz at each point (x, y, z), what is its mass?
Solution
We note that x\t) = \/2 cos t, y'(t) = z\t) = - sin t. The mass is given by the line
integral
Zi,
■L
m y, z)dx
X
cos2 t sin t dt
a. An object moving along a curve
C in a force field F 4
3
In Problem 49, you are asked to determine the center of mass of this wire. Note that
the center of mass may not lie on the curve, or wire, itself. ■
Wk^¥k- TkAsk
874 Chapter 13 Vector Analysis
By adding the contributions along all n subarcs, we obtain the sum ^ • T* As* as
k=l
an approximation to the total work performed as the object moves along C in the force
field F. As the length of the largest subarc ||As|| tends to 0, this approximating sum
W = J F • T ds
where T is the unit tangent at each point on C.
dR
Recall (from Section 10.4) that T = — where R is the position vector of the
ds
object moving on C. Thus, the work can also be given by the line integral
f dR f
W= / F-ds= / F dR
Jc ds Jc
This form of the line integral for work is used in the following example.
Solution
If C is the trajectory of the moving object, the work performed is W = J) F • dR. Let
Ci be the top (elliptical) part of C, and let Cj be the bottom (linear) part.
The curve Ci can be parameterized by the equations
W\
L F • dRi F[R,(0] • R\(t)dt
= (6 cos-1
2 + 2 cos t — 2) sin t dt Let u = cos t;
Jo
du = — sinf r/f.
If t = 0, u = 1,
-i: (6u~ + 2u — 2) du and if t = u = 0.
= 1
For the line segment C2, a parameterization is
The position vector is R2 = <2r, 1 — t), so IT, = (2,-1) and the work performed as
the object moves along C2 is
= [ F • dR2 — f F[R2(0]-R;(0^
Jc2 Jo
=f [61~ — 61] dt
= -1
Thus, the total work performed as the object moves along the curve C = Cj U C2 is
It can be shown that the work performed as an object moves around any closed
path in the force in Example 8 will always be 0. When this occurs, the force is said to
be conservative. We will discuss conservative force fields in Sections 13.3 and 13.4.
6. (x2 + y2) ds for C: x = e~‘ cos t, y = e~' sin t, 0 < t < | 14. Evaluate J [x2y dx + (x2 — y2)dy] for these choices of the
curve C:
7. J (—ydx +x dy) for C: y = 4x2 from (— 1, 4) to (0, 0) a. C is the arc of the parabola y = x2 from (0, 0) to (2, 4).
16. Evaluate J (—xy2 dx + x2 dy), where C is the path shown in 26. Jr (—y dx + x dy + xz dz), where C is the following path:
Figure 13.14. a. the helix defined by x = cos t, y = sinr, z = t for
0 < t < 2jt
b. the unit circle x2 + y2 = 1, z = 0, traversed once counter¬
clockwise as viewed from above
Is
17. Evaluate j (—y2 dx + x2 dy), where C is the path shown in
2
Figure 13.15.
29.
l
F • dR, where F = xi+xyj+x2yzk, and C is the elliptical
\\
\
C
y — i fix -'
circle of radius 1 centered on the z-axis in the plane z — 2,
traversed once counterclockwise as viewed from above
0)
31.
L 2 xy2z ds for C: x = t, y = t2, z = ft3 for 0 < t < 1
-
X
32. J yexz ds, where C is the line segment from (0, 0, 0) to
(2, 1,3)
Figure 13.15 Path of C for Problem 17
43. Find the work done by the force field F = (x2 + y2)i + (.v + y)j
as an object moves counterclockwise along the circle
x2 + y2 — 1 from (1,0) to (—1,0), and then back to (1,0)
along the x-axis.
53. A 180-lb laborer carries a bag of sand weighing 40 lb up a cir¬ Show that
cular helical staircase (Figure 13.16) on the outside of a tower
50 ft high and 20 ft in diameter. How much work is done as F • dR ^ F dR
the laborer climbs to the top in exactly five revolutions?
In general, the value of the line integral jc F • dR depends on the path of integration C,
but in certain cases, the integral will be the same for all paths in a given region D with
the same initial point P and terminal point Q. In this case, we say the line integral
is Independent of path in D (see Figure 13.17). In this section, we will study path
independence and characterize the kind of vector field F for which it occurs.
878 Chapter 13 Vector Analysis
The fundamental theorem of calculus (Section 5.4) says that if the function / is con¬
tinuous on [a, b], then
b
f(x)dx = F(b) - F(a)
where F is any antiderivative of /; that is, F'(x) = f(x). For a function of two or
three variables, the analogue of the derivative is the gradient, and the corresponding
analogue of the fundamental theorem is the following theorem.
Figure 13.17 A line integral THEOREM 13.2 Fundamental theorem for line integrals
fc F • dR is independent of path Let C be a piecewise smooth curve that is parameterized by the vector function R(r)
in D if its value is the same for all for a < t < b, and let F be a vector field that is continuous on C. If / is a scalar
curves joining any two points in D. function such that F = V/, then
J F-dR = f(Q)-f(P)
where Q — R(b) and P = R(a) are the endpoints of C.
Proof We will prove this theorem for the case where f(x, y, z) is a function of three
variables, such that F = V f(x, y, z). Suppose R(f) = {x(f), y(t), z(t)) and let G be
the composite function G(t) = f[x(t), y(t), z(t)J■ Then, according to the chain rule
dG_ _ 3/ dx df dy df dz
dt dx dt dy dt dz dt
and we have
/cF'rfR = /c Wf-dR
df 3 , 3
-dx + ^-dy + ^dz
/ /
-L dx dy dz
/ dt
Substitution
= f(Q) - f(P)
and C is the path described by R(f) = [r3 sin |f] i- [| cos (ft + f)] j for 0 < t < 1.
13.3 The Fundamental Theorem and Path Independence 879
Solution
First, note that the hypotheses of the fundamental theorem for line integrals are satisfied
since
f(x, y) = ex sin y — xy — 2y
/(0,0) = e° sin 0 — 0 — 0 = 0
A key requirement for evaluating the line integral fc F dR by the fundamental theorem
of line integrals is that F be the gradient of some scalar function /; that is, F = V/.
Our next goal is to learn how to determine when a given vector field F can be expressed
as a gradient field, and we begin by introducing some terminology.
Conservative Vector Field A vector field F is said to be conservative in a region D if F = V/ for some
scalar function / in D. The function / is called a scalar potential of F in D.
That is,
F = V/ for (x, y) on D
t t
Conservative vector field Scalar potential
The term conservative comes from physics and is related to the law of conserva¬
tion of energy (see Problem 54).
Solution
V/ = 2xyi + x2j and this is the same as F, so F is conservative. ■
A region D with property (1) is connected, and a connected region with property (2)
is simply connected. Roughly speaking, a simply connected region is one with no
“holes,” as shown in Figure 13.18.
THEOREM 13.3 Cross-partials test for a conservative vector field in the plane
Consider the vector field F(x, y) = u(x, y)i + u(x, y)j, where u and v have continuous
first partials in the open, simply connected region D in the plane. Then F(x,y) is
a. A region that is simply conservative in D if and only if
connected (no holes)
3u dv
throughout D
3y dx
Proof We outline the proof in Problem Set 13.4 after our discussion of Green’s the¬
orem. □
Solution
First note that the component functions u(x, y) — ex sin y — y and
v(x, y) = ex cosy — x — 2 have continuous partial derivatives. Then
3u dv
— = e cos y — 1 and — = e cos y — 1
3y dx ^
= ex sin y — xy + k(y)
Note that k(y) is a function of y alone—a
“constant” as far as x-integration is concerned.
Because / must also satisfy fy(x, y) = u(x, y), we compute the partial derivative of
this / with respect to y:
dk
fy(x, y) = —[ex siny - xy + k(y)] = e^cosy - x +
3y dy
dk
ex cos y x + — = e cos y — x — 2
dy
dk
= -2
dy
k{y) = -2 y + C
fix, y) = ex sin y - xy - 2y
13.3 The Fundamental Theorem and Path Independence 881
In Example 3, we began by using the fact that u = fx. In general, the issue of
whether to start with u = fx or v = fy is often determined by which equation leads to
the simpler integration.
Solution
We have u(x,y) = yexy and v(x, y) — xexy +
du rv rv xv xv
— = Xyexy + exy — = xyexy + exy + 1
3y ' dx
du dv
so — ± —, and F is not conservative. ■
dy dx
Note that a vector field F = (u(x, y), v(x, y)) in R2 can be regarded as the vector
field G = (u(x, y, 0), v(x, y, 0), 0) in R3. Since
i j k
3 3 3
curl G —
dx dy dz
u(x,y,0) v(x,y,0) 0
f dv du\
= °' + 0j + U-T)k
dv du
we have curl G = 0 if and only if — = —. Thus, Theorem 13.4 becomes the cross-
3x dy
partials test if F is in R2.
Solution
To show that F is conservative in R3, note that the components of F are continuous
with continuous partial derivatives and that
i j k
curlF = A A JL
dx dy dz
20x3z + 2 y2 4 xy 5*4 + 3z2
= (0 - 0)i - (20x3 - 20x3)j + (Ay - 4y)k
= 0
882 Chapter 13 Vector Analysis
where g(y, z) is a constant with respect to x-integration. Taking the partial derivative
of this expression with respect to y, and then comparing the result to the required
df
equation — = 4xy, we see that
dy
dg
4xy H—- = 4xy
dy
Thus,
dg
— =0 and g(y,z) = h(z)
dy
where h(z) is a constant with respect to x- and y-integration, so
h'(z) = 3z2
h(z) = z3 + C
Therefore, any function of the form
INDEPENDENCE OF PATH
We now have the tools to discuss path independence for a line integral. Here are the
definition and notation we will use.
Independence of Path The line integral fc F • dR is independent of path in a region D if for any two
points P and Q in D the line integral along every piecewise smooth curve in D
from P to Q has the same value.
Proof To prove this theorem, it is enough to prove that (i) implies (ii), that (ii) implies
(iii), and that (iii) implies (i). Then, if any one is true, all are true.
(i) implies (ii) Assume that F is conservative, and let C be a closed curve in D, as
shown in Figure 13.19a. Then any point P on C can serve as both the initial point
and the terminal point of the curve, and according to the fundamental theorem of line
integrals, we have
(ii) implies (iii)Let C\ and C2 be two curves in D with the same initial point P and
terminal point Q. Then the curve C formed by C\ followed by — C2 (the reverse of
C2) is a closed curve beginning and ending at P. According to condition (ii), the line
integral around this closed curve must be 0, so we have
The closed
0 = (j> F ■ dR = f F-dR+ f F dR
curve C = Ci u - C2
JC J C\ J —C2
b. (ii) implies (iii) V
Condition (ii) Additivity of line integrals
Figure 13.19 Equivalent conditions and it follows that (see Figure 13.19b)
for path independence
f
JCi
F dR — — f
J-c2
F dR = I
J c2
F dR
Since C\ and C2 were chosen as any two curves in D with the same endpoints, it
in D and construct a scalar function / such that F = V / in order to show that F must
be conservative. Details of this implication are outlined in Problem 55. □
Solution
In such a force field F, we have V/ = F, where / is a scalar potential of F, and
because the path of motion is closed, it begins and ends at the same point P. Thus, the
work is given by
We now have several ways for evaluating a given line integral Jc F • dR. We can
1. Parameterize C and use the parameterization to convert the line integral into an
“ordinary” integral in t over an interval a < t < b.
2. Check to see whether F is conservative. If it is, find a scalar potential function f
and then use the fundamental theorem of line integrals.
3. If F is conservative, find a convenient path C\ with the same endpoints as C and use
the fact that fC]F ■ dR = J'c F • dR since the line integral is independent of path.
Solution
First, we check to see whether F is conservative using the cross-partials test:
d 1
x*e~xy + (+3y — 3x2)e xy
dx y2 + 1 y2 + 1
i F dR = 0.
b. The curve C2 has initial point P(0, 0), where t = 0, and terminal point Q(- 1, 0),
where t — 1. Since F is conservative, the line integral fc F • dR is independent of
path, so the given line integral has the same value as fc F • JR, where C3 is the
line segment from P to Q. A parameterization for C3 is x — —t, y = 0. Thus, if
R = (—t, 0), we have R' = (—1,0) and
■d R =
L F JR = F[R(0] • R'(0 dt
-f
= 1
21 dt
In the next section, we will develop another criterion for path independence as part
of our study of an important result known as Green’s theorem.
Evaluate the line integrals in Problems 10-13 for each of the given 13.
paths. L [2xy dx + x2 dy]
10. [(3x + 2y) dx + (2x + 3y) dy]
Verify that each line integral in Problems 31-36 is independent of 45. The gravitational force field F between two particles of
path and then find its value. masses M and m separated by a distance r is modeled by
b. Due to the effect of the harbor you are entering, suppose 52. Let / and g be differentiable functions of one variable. Show
the wind force is {—a, —ae~y). Again, compute the work that the vector field F = [f(x) + v]i + [g(y) + -v]j is conser¬
along two paths as in part a. Does the path matter here? vative, and find the corresponding potential function.
Why or why not? 53. Let R = xi + yj + zk and r = ||R||. Show that the work done
c. Repeat part b for {—a, —ae~y+x/9). in moving an object from a distance r, to a distance r2 in the
50. Exploration Problem An important type of problem in several central force field F = R/r1 is given by
fields of application is, “Can we find the optimal path to min¬
imize the work?” You are to explore this issue in regard to the
wind force in Problem 49c, ri r2
F = (-a, —ae~v+v/9) 54. An object of mass m moves along a trajectory R(r) with ve¬
locity V(r) in a conservative force field F(q). Let R(to) = Qo
a. You should have the work computed for two different paths and R(q) = Qi be the initial and terminal points on the tra¬
from the previous problem. By looking at these numbers jectory.
and carefully studying F, you should see that we will be a. Show that the work done on the object is
rewarded (or punished) by changing the path slightly from W = K{ti) - K{t0), where K(t) = ||V(f)||2 is the
the straight-line path. What do your observations suggest object’s kinetic energy.
regarding trying to minimize the work?
b. Let / be a scalar potential for F. Then P(t) — —fit) is
b. Attempt to find an (approximate) optimal path, starting the potential energy of the object. Prove the law of conser¬
with your observations in part a and common sense. For vation of energy—namely,
example, the path could be described by a parabola (or
higher-degree polynomial) or by some trigonometric func¬ Pito) + AT (?o) = Pih) + K(h)
tion. Find a “good” path for this purpose.
c. Explore the following question: “Is there a realistic opti¬ 55. Complete the proof of Theorem 13.5 by showing that if
mal path from A to BT Consider two conditions: First, Jc F ■ dR is independent of path in the open connected set D,
there is no restriction on the path; second, suppose there then F is conservative. Do this by completing the following
is a shoreline at y = 1 so that one’s path cannot exceed steps:
this limit. Hint-. One approach might be to consider all a. Let Pia, b) be a fixed point in D, and let Qix, y) be any
parabolic paths of the form y = x(b — ax), where a and point in D. Define the function fix, y) by
b are nonnegative, and y(2) = 1. In this case, you can
eliminate b and do a one-parameter study.
d. Assume the path in part c cannot go above y — 1. You may
fix, y) = J F
have discovered the optimal path. What is it? If you did the
parabolic study suggested in part c, there is an optimal path
Let <2i (x, y) be a point in D other than Q that lies on the
same horizontal line as Q. Let C\ be any curve in D from
in this case. What is it?
P to and let C2 be the horizontal line segment joining
Q 51. Show that if the vector field
Qi to Q. Show that if F = Mix, y)i + Nix, y)j, then
3/
F(jc, y, z) = M{x, y, z)i + Nix, y, z)j + P(x, y, z)k -f = Mix,y).
ox
is conservative, then b. Using a similar argument (only with vertical line seg-
9f
3P 3N 3M _ 3 P 3N _ 3M ments), show that — = Nix, y). Conclude that since
3y
3y 3z 3z dx dx 3y F = V/, F must be conservative in D.
dF
The fundamental theorem of calculus dx — Fib) — Fia), can be described
■f
lx
dF
as saying that when the derivative — is integrated over the closed interval a < x <b,
dx
the result is the same as that obtained by evaluating Fix) at the “boundary points” a
and b and forming the difference Fib) - Fia). We obtained the analogous result
L V/-dR=/(G)-/(P)
888 Chapter 13 Vector Analysis
in Section 13.3, and our next goal is to obtain a different kind of analogue to the funda¬
mental theorem of calculus called Green’s theorem after the English mathematician
George Green (Historical Quest in Problem 39).
GREEN’S THEOREM
Green’s theorem relates a line integral around a closed curve to a double integral over
the region contained by the curve. We begin with some terminology. A Jordan curve,
named for the French mathematician Camille Jordan (1838-1922), is a closed curve C
that does not intersect itself (see Figure 13.20). A simply connected region D in the
plane has the property that it is connected and the interior of every Jordan curve C in
D also lies in D, as shown in Figure 13.20.
Not a Jordan curve Not a simply connected region; A positively oriented boundary
a Jordan curve in D whose interior curve. The region is “on the left'
contains points not in the region of the curve.
Picture yourself as a point moving along a curve. If the region D stays on your
left as you, the point, move along the curve C with increasing t, then C is said to be
positively oriented (see Figure 13.20). Now we are ready to state Green’s theorem.
fwdx + Ndy)=jj dA
D
Graffiti on the wall of a high school What This Says This theorem expresses an important relationship be¬
playground in Ramat Gan (a suburb tween a line integral around a simple closed curve (a curve is simple if there
of Tel Aviv). Courtesy of Eli Maor.
are no self-intersections and closed if there are no endpoints) in the plane and
a double integral over the region bounded by the curve. It is one of the most
important and elegant theorems in calculus. Take special note that D is required
to be simply connected with a positively oriented boundary C.
Proof A standard region is one in which no vertical or horizontal line can inter¬
sect the boundary curve more than twice (see Figure 13.21). We will prove Green’s
theorem for the special case where D is a standard region, and then we will indicate
13.4 Green's Theorem 889
how to extend the proof to more general regions. Suppose D is a standard region with
boundary curve C. We begin by showing that
3M
dx dy M dx
3y
r f bm rr am , r r rf2(x) 3m ,
Figure 13.21 Standard region: No
vertical or horizontal line intersects
the boundary more than twice.
JJicdxdy=JJi»dydx=L L wdy
D D
dx
/
'Cu
J
M dx + / M dx
JcL
= — (j) M dx
3N 3 M'
D
dx dy J
dA
ffj;dxdy-ffjidxdy
JJ
D
dx ' JJ
D
dy
This completes the proof for a standard region. If D is not a standard region, it can
be decomposed into a number of standard subregions by using horizontal and vertical
“cuts,” as shown in Figure 13.22. The proof for the standard region is then applied to
Figure 13.22 General case: The region each of these subregions, and the results are added. The line integrals along the cuts
is decomposed into a finite number cancel in pairs, and after cancellation, the only remaining line integral is the one along
of standard regions by cuts. the outer boundary C. Thus,
(M dx + N dy) dA
Show that Green’s theorem is true for the line integral (—y dx + x dy), where C is
Solution
Figure 13.23 Path C First, evaluate the line integral directly. The curve C consists of the line segment Ci
from (—1, 0) to (1, 0), followed by the semicircular arc C2 from (1,0) back to (-1,0).
890 Chapter 13 Vector Analysis
Ci : x = t, y = 0 -1 < t < 1
dx = dt, dy = 0
Cj '■ x — cos u, y = sin w 0 < u < n
dx = — sin u du, dy = cos udu
= / (sin2 u cos2 u) du = \ du — rt
Jo Jo
Next, we use Green’s theorem to evaluate this integral. Note that the boundary C
is a Jordan curve and M — —y, N — x, so that F(jc, y) = —yi + xj is continuously
differentiable. We now apply Green’s theorem:
(—y dx + x dy)
If (fx(X) - Fy(-y)) dA ll2dA
— 2(area of semicircle) = 2 [^(l)2] = jt ■
Solution
The work done, W, is given by the line integral <j) F • dR. Note that F is continuously
3 d 1
W — (2x2y - 2y2 sin y) - — (x + xy2) dA
= /cF'rfR = // dx dy
D
=
J0
f (X-X5)dx= [ix2-|x6]|o = i
A line integral can be used to compute an area of a region in the plane by applying the
following theorem.
= \i[xdy ydx]
and leave the other two area formulas for the reader (see Problem 44).
Let F(x, y) = —yi+xj. Then since F is continuously differentiable on D, Green’s
WARNING Tl . . ,
I his gives us yet another theorem applies. We have
technique for finding the area of a
9
region, especially when its boundary is
specified in parametric form. In finding
area, the function F = —yi + xj does
(—y dx + x dy)
IID
dx
(x)
9y
(~y) dA
-II D
2 dA = 2 A
A = j (—y dx + x dy)
2n
[—{b sin 9)(—a sin 0 d6) + {a cos 6){b cos 6 dd)]
2n
ab{sin2 0 + cos20) d6
2tt
abdO
— \ab{2n — 0)
= nab ®
In the statement of Green’s theorem, we require the region R inside the boundary curve
C to be simply connected, but the theorem can be extended to multiply-connected
regions, that is, regions with one or more “holes.” A region with a single hole is shown
in Figure 13.26a. The boundary of this region consists of an “outer” curve Ci and an
“inner” curve C2, oriented so that the region R is always on the left as we travel around
the boundary, which means that C1 is oriented counterclockwise and C2 clockwise.
a. A doubly-connected region We now make cuts AB and CD through the region to the hole, as indicated in
with oriented boundary Figure 13.26b. Let R\ be the simply connected region contained by the closed curve
curves C{ and C2.
C3 that begins at A, extends along the cut to B, and then clockwise along the bottom
of the curve C2 to C, along the cut to D, and counterclockwise along the bottom of
Ci back to A. Similarly, let R2 be the region contained by the curve C4 that begins
at D and extends to C along the cut, to B along the top of C2, to A along the second
cut, and back to D along the top part of Ci. Then, if the vector field F = Mi + Nj is
continuously differentiable on R. we can apply Green’s theorem to show
But the line integrals from A to B and C to D cancel those from B to A and D to C,
leaving only the line integrals along the original boundary curves C\ and Ci. Thus, we
have
(M dx + N dy) + ® (M dx + N dy)
l:
dA = (p (M dx + N dy) + (p (M dx + N dy)
JC\ Jc2
Solution
Let M(x, y) —
-y and N — -. Then
xz + y" x2 + y2
dN y 2 - Jr2 9M
dx (x2 + y2)2 dy
at any point (x, y) other than the origin. Next, let C\ be a circle centered at the origin
with radius r so small that the entire circle is contained in C, and let R be the region
Figure 13.27 The region R between the curve C and the circle Cj, as shown in Figure 13.27.
for doubly-connected regions dN dM
We know that —— = —— throughout R (since R does not contain the origin) and
dx dy
Green’s theorem for doubly-connected regions tells us that
—y dx + x dy -y dx + x dy 'dN dM'
i x2 + y2
+
/ x2 + y2 If dx dy,
dA = 0
so that
—y dx + x dy -y dx + x dy -y dx + x dy
i
,c xz + yz --L JCl xz + yz -l /_ Cl
where — Ci is the circle Ci traversed counterclockwise instead of clockwise. In other
xz + y2
words, we can find the value of the given line integral about the curve C by finding the
line integral about the circle —Cj. To do this, we parameterize — C\ by
x = r cos 9, y — r sin 0 for 0 < 0 < 2tt
and find that
—y dx + x dy r2n —r sin 6(—r sin 0 dO) + r cos 6(r cos 9 d6)
£-Cl
73 ,
x2 + y2 =f /
r2 cos2 6 + r2 sin2 9
r2n
2n r2(sin2 9 + cos2 9) /•2n
d9 1 d9 = 27T
Jo r 2(sin2 9 -I- cos2 9) Jo
13.4 Green's Theorem 893
Thus,
-y dx + x dy — y dx + x dy
i x2 + y2 =1 x2 + y2
2:t
Note that the line integral in Example 4 is not independent of path in R because
there are closed paths (like Ci) along which the line integral is not 0.
Green’s theorem can be expressed in two forms that generalize nicely to M3. For the
first, note that the curl of the vector field F(jc, y) = M(x, y)i + N(x, y)j is given by
i J k
3 3 3
curlF =
dx 3y 3z
M(x,y) N(x,y) 0
“3 N 3 AT
= 0i + Oj +
dx dy
so
/ dN 3 M\
If
D
V ~dx 3y )
dA
= //<curlF'
D
k) dA
On the other hand, the line integral in Green’s theorem can be expressed as
(j) F • dR = <j) dR
F • —j— ds — (j) F • T ds
3jV 3 AT
lFdR = IJ D
-) dA
,dx dy J
Green's theorem
<j) F • T ds — JJ (curl F • k) dA
D
Solution
Parameterize C with the arc length parameter s, so that R(s) = ;c(s)i + y(s)j
is the position vector on C. A unit tangent vector T to the curve C is
T = R'(s) = jc'(s)i + y'(s)j, which means that an outward normal vector is
N = /(s)i — ;L(s)j (see Figure 13.28).
We now apply Green’s theorem to find a representation using divF.
J
Figure 13.28 The outward unit
i F • Nr/s = (Mi + ZVj) • [;/(s)i — x'(s)j] ds
-II*
D
div F d A
//
'dN
,dy ,
dM'
dA JJ (curl F • k) d A
D D
^ D
//
D
divF
dA
Normal component of F
NORMAL DERIVATIVES
In physics, some important applications of Green’s theorem involve the normal deriva¬
tive of a scalar function /, which is defined as the directional derivative of / in the
direction of the outward normal vector N to some curve or surface.
Normal Derivative The normal derivative of /, denoted by df/dn, is the directional derivative of
/ in the direction of the normal vector pointing to the exterior of the domain of
/. In other words,
3/
= V/-N
dn
where N is the outward unit normal vector.
13.4 Green's Theorem 895
The following example illustrates how Green’s theorem can be used in connection
with the normal derivative. Additional examples are found in the problem set.
Jj V2fdxdy = j) 9/
dn
ds
D
9/
where V"/ = fxx + fyy is the Laplacian of / and — = V f ■ N is the normal deriva-
dn
tive vector.
Solution
9/ 9/
Let u =-and v = —. Then we have
3y dx
dv 3u
V2/ = fXX + fyy =
dx dy
3u
II V2/ dx dy
D
m-
D
dy
dx dy
( dx dy\
u-h v— I ds
c V ds ds J
+ ds
c V 3y ds dx ds J
f ( dy dx\
v/.f^i-LA ds
c \ ds dsJJ
dy. dx.
= I V / -N ds Where N — —l-j is the
ds ds
outward unit normal vector to C
= 1^
Jc 9n
896 Chapter 13 Vector Analysis
1. 2. (y3 dx — x3 dy)
£ {y2 dx + x2 dy)
y.
£
L
,(° 1) (i
c
10. [(x + y)dx - (3x — 2y) dy]
,(i ,0
(0 0) X y ,(2 .4)
(0. 4)J
\
c
yf
6 . — 3x dy) .
(0 0)
L ,(2
,0
a:
.1
(2, 0)
In Problems 7-14, use Green’s theorem to evaluate the given line
(OjO) a:
integral around the indicated closed curve C.
.1. AD
x2y2
7. (2y dx — x dy) 8 . (eJt dx — sinx dy) 14.
£
2x tan 1 y dx dy
£ £ l+y2
y,
(0, 2) .(2. 2)
_ (2
1 a:
(0, 0)
13.4 Green's Theorem 897
15. Use Green’s theorem to find the work done by the force field -y
27. Evaluate + - 1 dx + - dy , where C is any Jor¬
F(x, y) = (3y - 4x)i + (4x - y)j
£
dan curve that does not intersect the y-axis, traversed counter¬
clockwise.
when an object moves once counterclockwise around the el¬
f x dx + y dy
lipse 4x2 + y2 = 4. 28. Evaluate ® -—-—^r-1-, where C is any Jordan curve whose
Jc *2 + y2
16. Find the work done when an object moves in the force field interior does not contain the origin, traversed counterclock¬
F(x, y) = y2i + x2j counterclockwise around the circular path wise.
x2 + y2 = 2.
29. Evaluate the line integral
Use Theorem 13.7 to find the area enclosed by the regions de¬
scribed in Problems 17-20, and then check by using an appro¬
x dx + y dy
priate formula.
17. circle x2 + y2 = 4
£ x2 + y2
18. triangle with vertices (0, 0), (1, 1), and (0, 2) where C is any piecewise smooth Jordan curve enclosing the
19. trapezoid with vertices (0, 0), (4, 0), (1, 3), and (0, 3) origin, traversed counterclockwise.
20. semicircle y = V4 — x2 / —y dx + (x — 1) dy
30. Evaluate ® ----—, where C is any Jordan
Jc (x —1 )2 + y2
Q 21. Evaluate the line integral <j) (x2ydx — y2xdy), where C is curve whose interior does not contain the point (1,0), tra¬
the boundary of the region between the x-axis and the semi¬ versed counterclockwise.
circle y — -Jo.1 — x2, traversed counterclockwise (including t- , . I “O' + 2)dx + (x -\)dy
31. Evaluate ® ----—, where C is any Jor-
the x-axis). Jc (x - l)2 + (y+ 2)2 J
22. Evaluate the line integral dan curve whose interior does not contain the point (1, —2),
traversed counterclockwise.
(£ (3y dx — 2x dy) (* 9 z,
32. Evaluate ® — ds, where z(x, y) = 2x2 + 3y2, and C is the
Jc
circular path x2 + y2 = 16, traversed counterclockwise.
where C is the cardioid r — 1 + sin 6, traversed counterclock¬
wise. 3/
33. Evaluate ® — ds, where /(x, y) = x2y — 2xy + y2, and C
23. Show that Jc dn
is the boundary of the unit square 0<x < l,0<y < 1,
traversed counterclockwise.
£ [(5 -xy- y2) dx - (2xy - x2) dy] = 3 x I dx
34. Evaluate ® x—ds, where C is the boundary of the unit
where C is the square 0 < x < 1, 0 < y < 1 traversed coun¬ Jc dn
square 0<x<l,0<y<l, traversed counterclockwise.
terclockwise and x is the x-coordinate of the centroid of the
square. 35. If C is a Jordan curve, show that
24. Find the work done by the force field F(x, y) = (x + 2y2)j
as an object moves once counterclockwise about the circle (j) [(x — 3y) dx + (2x — y2) dy] = 5A
(x -2)2 + y2 = 1.
25. Let D be the region bounded by the Jordan curve C, and let A
where A is the area of the region D enclosed by C.
be the area of D. If (x, y) is the centroid of D, show that
36. Use line integration to find the area of the region bounded by
and the curve C: x = cos31, y = sin31 for 0 < t < 2rc.
£ y2 dx 037. Prove the following theoretical application of Green’s theo¬
where C is traversed counterclockwise. rem: Let F(x, y) = u{x, y)i + u(x, y)j be continuously dif¬
26. Use Theorem 13.7 and the polar transformation formulas ferentiable on the simply connected region D. Then F is con¬
x = r cos 9, y — rsin0 to obtain the area formula in polar servative if and only if
coordinates, namely,
du dv
dy dx
throughout D.
38. Recall that a scalar function / with continuous first and sec¬
ond partial derivatives is said to be harmonic in a region D if
W2f — 0 (that is, if fxx + fyy = 0). If / is such a function and
D is a simply connected region enclosed by the Jordan curve
C, show that
b. Once again start with the line integral and derive what is
known as Green’s second formula.
Figure 13.30 A triply-connected region
-•
SURFACE INTEGRATION
A surface is said to be smooth if there is a nonzero normal vector at each of its points,
and it is piecewise smooth if it is composed of a finite number of smooth pieces. A
13.5 Surface Integrals 899
z
surface integral is a generalization of the line integral in which the integration is over
(**• jM) S:z=f(x,y) a surface in space rather than a curve.
We begin by defining the surface integral of a continuous scalar function g(x, y, z)
over a piecewise smooth surface S, as shown in Figure 13.31a. Partition S into n
subregions, the kth of which has area ASk, and let Pfixk, yk, z*k) be a point chosen
arbitrarily from the kth subregion, for k = 1,2.n (see Figure 13.31b). Form the
Riemann sum
n
g(x, y, z)dS
s
Recall from Section 12.4 that when the surface S projects onto the region R in
the xy-plane and S has the representation z = fix, y), then dS — J f2 + f2 + 1 dA,
where dA is either dxdy or dydx. We can now state the formula for evaluating a
b. Construction of a Riemann sum surface integral.
Figure 13.31 Definition of a surface
integral
Surface Integral Let S be a surface defined by z = fix, y) and R be its projection on the xy-
plane. If /, fx, and fy are continuous in R and g is a continuous function of
three variables on S, then the surface integral of g over S is
s
IIsdS
shown in Figure 13.32, where g(x, y, z) — xz + 2x2 — 3xy and S is that portion of the
plane 2x — 3y + z = 6 that lies over the unit square R: 2 < x < 3, 2 < y < 3.
Solution
First, note that the equation of the plane can be written as z = f(x,y) where
fix, y) — 6 — 2x + 3y. We have fxix, y) = -2 and fyix, y) = 3, so that
— V~\A JJ 6x dy dx
Figure 13.32 The portion of the plane
that lies above the unit square R
6/l4
a: x dy dx = 6\/l4 J x dx = 15VT4
900 Chapter 13 Vector Analysis
If 8(x, y, z) is the density (mass per unit area) at each point (x, y, z) on the lamina,
then the total mass, m, of the lamina is given by the surface integral
Mass of a lamina m
ff 8(x, y, z)dS
and the center of mass of the surface is the point C(x, y, z), where
Center of mass
-iff s
x8 dS, y
-iff s
y8 dS,
-iff s
zSdS
These formulas may be derived by essentially the same approach used in our previous
work with moments and centroids of solid regions. (See, for example, Sections 6.5 and
12.6.)
Solution
We begin by calculating dS.
Zx = |(a2 - x2 - y2) 1/2(—2x) =-x(a2 - x2 - y2) 1/2
The surface projects onto the disk x2 + v2 < a2 in the xy-plane, so the mass of the
B C hemisphere S is given by
A - - D
B D m JJ S(x, y, z) dS = JJ zdS
s s
FLUX INTEGRALS
An important application of surface integrals involves the flow of a fluid through a sur¬
face. To discuss such applications, we define a surface S to be orientable if S has a unit
normal vector field N that varies continuously over S. Most common surfaces, such
as spheres, cones, cylinders, ellipsoids, and paraboloids, are orientable, but it is not
difficult to construct a fairly simple surface that is not. For example, the Mobius strip,
formed by twisting a long rectangular strip before joining the ends, is not orientable
b. Two-sided (orientable) surface: a football. (see Figure 13.35a). If S is an orientable surface, there will be two possibilities for
This is also a closed surface. N, which can be thought of as outward (pointing toward the exterior of S) or inward
(pointing toward the interior), as illustrated in Figure 13.35b.
Figure 13.35 Surfaces in space Consider a fluid of density 8 flowing steadily through a surface S with the contin¬
uous unit normal field N. If V is the velocity of the fluid flow, then the vector field
F = 8\ is the flux density, which measures the volume of fluid crossing the surface
per unit area in unit time. Think of water flowing through a net, as shown in Figure
13.36.
If AS1 is the area of a small patch of the surface S, then the amount of fluid crossing
this patch in unit time in the direction of N may be estimated by the volume of the
cylinder of height F • N, and base area AS (see Figure 13.37); that is,
AV as (F • N)AS
Thus, we can measure the total volume crossing the surface in unit time by computing
the surface integral of F • N, which is called a flux integral.
N
Figure 13.36 Water flowing through
a net
Flux Integral Let F be a vector field whose components have continuous partial derivatives on
the surface S, which is oriented by the unit normal field N. Then the flux of F
across S is given by the surface integral
Flux = F-N dS
VG
N =- where VG = (—fx, —fy, 1)
l| VG || y
Therefore,
and since || VG || = y/i~fx)2 + (—fy)2 + 1, the flux integral of F over S can be written
as
-VG
N =
I VG |
Solution
Let fix, y) — z = 1 — x — y. Then fx = — 1, fy — —1, and the flux integral is
nD
[xy + (1 - x - y) + x + y] dy dx
n [xy + 1 ] dy dx
f
Jo
[\xy2 + y]\l~x dx
jf 1
(x — 2x — x + ) dx 2
13
24
below by the xy-plane. Experiments indicate that the velocity of heat flow is given by
the vector field H = —KVT, where
T(x, y, z) = 2x + y - 3 r
is the temperature at each point P(x, y, z) in the region and K is a constant (the heat
conductivity, which is obtained experimentally for each different substance). Find the
total heat flow J'f H-NdS out of the region (that is, N is the outer unit normal, the one
R
pointing away from the origin).
Solution
The surface is shown in Figure 13.39. Note also that VT = (2, 1, —6z) and that the
surface S is composed of a top surface 5] and a bottom surface S2. The top surface is
S\: z — 9 — x2 — y2 and the bottom surface is the disk S2'. x2 + y2 < 9, which is also
the projection D of S) on the xy-plane.
For Si: Let G = z + x2 + y2 — 9; then, VG = (2x, 2y, 1). The top surface Si has
upward pointing normal, and the flux integral over Si is
-II D
K[Ax + 2y — 6(9 — x2 — y2)] dA Si nee
Z = 9 — x — y on S
2 2
/*2tt /»3
Figure 13.39 The surface 5 of the [4r cos 6 + 2r sin 9 — 6(9 — /-)] r dr d6
region bounded above by —Kl Jo
z = 9 — x2 — y2 and below by
36cos0 + 18 sin# — dO
the xy-plane = -*/
= 2A3n K
For S2: The outer normal N2 —k = (0.0, —1) points downward and the flux
904 Chapter 13 Vector Analysis
integral is
II
= —K IJ 6zdA
D
=—K
IIjj
D
0 dA 5ince z — 0 on S2
= 0
Thus, the total heat flow is
II HNdS = II
S
H'N|<rt+ //« • N2 dS
s2
= 2A3jtK + 0 = 243tzK ■
II
D
|R„ x Ry|| du dv
jf f(x,y,z)ds = jj mm u x Ryll du dv
D
Solution
Thus, ||R„ x Ryll = y52 + (—5)2 + (—5)2 = 5>/3. We now substitute these values
into the surface integral formula:
n
y z)dS
s D
l p2
(Au + 2v)(5V3) du dv = 5V3 / [hi2 + 2uv]\' dv
Jo
= 5^3 f
Jo
(2 + 2v) dv = 5^3 [2v + v2] ^ = 5^3(8) = 40^3
For a flux integral over a surface S parameterized by R(u, v) that projects onto a
region D in the uv-plane, we have
x R,
|R„ x RJ| dA
//F ( K x R„
D
= JJ F • (R„ x RJ dA
D
Solution
First, note from R(m, v) that x = uv, y — u — v, and z = 2u + v, so we have
i j k
< Ry = V l 2
u — 1 1
= 3i — (u — 2u )j + (—v — u)k
■ (R„ x Rv)dA
1 —u
(2u T v, uv, 3m) • (3,2u — v, —(u + v)) dv du
1 -u
[(2 u + u)(3) + (uv)(2u — v) + 3 u(—u — v)]dv du
906 Chopter 13 Vector Analysis
1 /> 1 — u
1)(8m3 u1 16m — 9) du
-L>-
— —
137
120
x2 y 2
4. S: z = 10, — + — < 1
4 1
Let S be the hemisphere x2 + y2 + z2 = 4, with z > 0, in Problems plane z = x +■ 1 that lies inside the cylinder x + y2 = 1. 2
9.
ffzdS
s
10.
ffzldS
11. JJ (* - 2y)dS 12.
s s
,5. //,«
■II s
(4 -z)dS in Problems 22-29. Assume N is the outward directed normal field.
22. F = xi + 2yj + zk, and S is the triangular region bounded by
■II
the intersection of the plane x + y + z = with the positive
2 1
dS
177/ Vl +4 zdS coordinate planes.
VT+4z
s 23. F = xi 4- 2yj — 3zk, and S is that part of the plane
15x — v + 3z = that lies above the unit square < x < ,
If
12 6 0 1
28. F = x2i+y2j+z2k, and S is the part of the cone z = V'x2 + y2 44. The temperature at each point (x, y, z) in a region D is
below the plane z = 2. T — 3x2 + 3z2. If the heat conductivity is K — 5.8, find the
29. F = yi + z2j — 2zk, and S is the part of the surface cut from rate of heat flow outward across the cylinder x2 + z2 = 4 for
the cylinder z = 3 — x2 by y = 0, y = 1, and z — 0. 0 < y < 3.
Q 45. a. A lamina has the shape of the portion of the sphere
Q 30. Evaluate Jj(3x — y + 2z)dS, where S is the surface deter-
x2 + y2 + z2 = a2 that lies within the cone z = y/x2 + y2.
5 Determine the mass of the lamina if <$(x, y, z) = x2y2z.
mined by R(m, v) = ui + u) — uk, 0 < u < 1,1 < v <2.
b. Let 5 be the spherical shell centered at the origin with ra¬
31. Evaluate JJ (x — y2 + z) dS, where S is the surface defined dius a, and let C be the right circular cone whose vertex
s is at the origin and whose axis of symmetry coincides with
by R(m, v) = w2i + uj + «k, 0<m<1,0<u<1. the z-axis. Suppose the vertex angle of the cone is </>0, with
0 < 0o < tt/2. Determine the mass of that portion of
32. Evaluate Jj(tan^1 x + y — z2) dS, where S is the surface de- the sphere that is enclosed in the intersection of S and C.
s Assume 8{x, y, z) = 1.
fined by R (u, v) = ui + irj — uk, 0<m<1,0<u<1. 46. Recall the formula
//
36. Evaluate / / F • Nr/S, where F = xi + yj + z4k and S is the its diameter is |ma2, where m is the mass and a is the radius.
Assume <$(x, y, z) = 1.
48. HISTORICAL QUEST August
parametric surface
Mobius studied under Karl Gauss
x = u cos v y = u sin u z — u (1777-1855), as well as Gauss' own
teacher Johann Pfaff (1765-1825). He
for 0 < u < 2, 0 < v < 2n. was a professor at the University of
In Problems 37-42 find the mass of the homogeneous lamina that Leipzig and is best known for his work in
has the shape of the given surface S. topology, especially for his conception
37. S is the surface z = 4 — x — 2y, with z > 0, x > 0, y > 0; of the Mobius strip (a two-dimensional
8 — x. surface with only one side). ■ AUGUST MOBIUS
1790-1868
38. S is the surface z = 10 — 2x — y, with z > 0, x > 0, y > 0;
An example of a Mobius strip is given parametrically by:
8 = y.
39. S is the surface z = x2 + y2, with z < 1; 8 — z. Hint: Use
x = cos v + u cos - cos v
cylindrical coordinates. 2
40. S is the surface z = 1 — x2 — y2, with z > 0; 8 = x2 + y2 + z2. v .
y = sin v T u cos — sin v
Hint: Use cylindrical coordinates. 2
41. S is the surface r2 + z2 = 5, with z > 1; 8 — 62. z = u sin
42. S is the triangular surface with vertices (1,0, 0), (0, 1,0), and
(0, 0, 1); 8 = x + y. where — 2 < u < 0 < v < 2n. Construct a three-
43. A fluid with constant density p flows with velocity dimensional model of a Mobius strip and show that a Mobius
V = (xy)i + (yz)j + (xz)k. Find the upward rate of fluid flow strip is not orientable. If you have access to a CAS program,
through the paraboloid z = 9 — x2 — y2. sketch the graph of this surface.
908 Chapter 13 Vector Analysis
13.6 Stokes’Theorem
IN THIS SECTION Stokes' theorem, theoretical applications of Stokes' theorem, physical interpretation of
Stokes' theorem
In Section 13.4, we observed that Green’s theorem can be written
Direction along C is dR
-II (curl F • k) dA
where A is the plane region bounded by the Jordan curve C. Stokes’ theorem is a
generalization of this result to surfaces in space and their boundaries.
STOKES’ THEOREM
Before stating Stokes’ theorem, we need to explain what is meant by a compatible ori¬
entation. We say that the orientation of a closed path C on the surface S is compatible
with the orientation on S if the positive direction on C is counterclockwise in relation
to the outward normal vector N of the surface (see Figure 13.40). That is, if a person
walks around C in a positive (counterclockwise) direction with her head pointing in
the direction of N, then the surface will always be on her left.
walking in a counterclockwise
direction around the boundary curve C. Proof A proof assuming F, S, and C are “well behaved” is given in Appendix B. □
Evaluate (j) (iy2 dx + zdy + x dz), where C is the curve of intersection of the plane
Solution
If we set F = \y2\ + -j + xk, the given line integral can be expressed as
F dR
F'rfR7/ s
(curl F • N)dS
There are many surfaces bounded by the curve C. We choose the surface S that is part
of the plane x + z = 1. We find the required parts of the surface integral; namely,
curlF, N, and dS.
' J k
3 3 3
curl F = -i-j-yk
dx 3y 3z
1 2
ky z x
The upward unit normal vector to the plane x + z — 1 is N = —p(i + k), so that
V2
/ 1 1 \
curl F • N = (— 1, — 1, —y) ■ (—=, 0,
,V2’ V21
1
y
n /2 sfl
= -j=(\+y)
x2 + 2y2 + (1 — x)1 1
X2 - x + y2 0
b-i)2 + / = i
Thus, the projection of C on the xy-plane is the disk D bounded by the circle
(x _ i)2 + y'z = I shown in Figure 13.42. Note that this circle has the polar equation
r — cos 0.
We now calculate the desired line integral using Stokes’ theorem.
s
Figure 13.42 The projection D of 1
=-/7 r rcosd
(1 + r sin 6) r dr dO Change to
polar coordinates.
F = -§y2i-2xyj + yzk
Solution
The surface S is the triangle with vertices (1,0, 0), (0, 1,0), (0, 0, 1), whose boundary
C is traversed in that order, as shown in Figure 13.43. We will show that the line
(0,0,1)
integral ® F • dR and the surface integral f/ (curl F • N) dS have the same value.
Jc s
I. Evaluation of <j) F ■ dR
The three edges of the boundary curve C are expressed as
Ei : x + y = 1, z = 0
E2 : y+z = 1, x = 0
[ F • dR — f (\t2-2t + lt2) dt
J E\ Jo
= /V-2() c=i
J0
1
■d R =
L (1 — s)s ds
6
1
f F • dR = f F dR+ f F dR+ f F-dR=- + -+0
JC J E\ J Ei J Et, ^ 6 3
1 J k
9 9 9
curlF = = zi + yk
9x 9y 9z
-\y2 — 2xy yz
13.6 Stokes'Theorem 911
The triangular region S on the surface of the plane x + y + z — 1 has outward unit
normal vector
1
N = —(l + j + k)
v/3
Because z = 1 — x — y, we find
1
curl F • N = (zi + vk) • —=(i + j + k)
V3
1 1
= —t=(1 — x - y + y) = —=(l -x)
n/3 73
and since zx = — 1, zy = — 1, it follows that
Finally, the surface S projects onto the triangular region D in the xy-plane that is
bounded by the lines x = 0, y = 0, and x + y = 1 (see Figure 13.44). Thus, we
have
—x
(1-x)3 * S 1
x)[(l — x) — 0] dx —
3 o 3
Figure 13.44 The projected region
in the xy-plane is a triangle We see that for this example,
bounded by y = 1 — x and the
positive coordinate axes. (curl F • N) dS
s
for any function F whose components have continuous partial derivatives on both 5]
and 1S2.
Surface S Boundary
z= 1 2y curve C
*2 + 2y2 = 1
Bottom EXAMPLE 3 Using Stokes' theorem to evaluate a surface integral
surface
Evaluate Jj (curlF • N) dS, where F = xi + y2j + zexyk and S is that part of the
Solution
By setting z = 0 in the equation of the surface, we find that the boundary curve C for
S is the ellipse x2 + 2y2 = 1, and the upward unit normal vector N on S induces a
counterclockwise orientation on C, as shown in Figure 13.45. Let S* be the elliptical
Figure 13.45 Surface z = l — x2 — 2y2 disk defined by x2 + 2y2 < 1, z = 0. We see that S and S* have the same boundary
and boundary curve C and the same orientation. Since the upward unit normal to S* is k, we have
5 s*
912 Chapter 13 Vector Analysis
and we use this second integral to calculate the required integral. We obtain
i j k
curlF =
a a a
(zxexy)i - (zyexy)j
dx dy 9z
X
s s*
In physics and other applied areas, Stokes’ theorem is often used as a device for estab¬
lishing general properties. For instance, we now use it to prove the curl criterion for
a conservative vector field in M3 stated without proof in Section 13.3 (Theorem 13.4),
and whose proof was promised at this time.
The curl criterion for a conservative vector field in R3 Suppose the vector field
F and curlF are both continuous in the simply connected region D of R3. Then F is
conservative in D if and only if curl F = 0 in D.
so that
/, F dR — f F
Jc2
dR
It follows that the line integral <^> F • dR is independent of path, so F must be conser-
vative. □
N dS = I
£H dR = I and
where H is the magnetic intensity, and J is the electric current density. Use this infor¬
mation to derive Maxwell’s current density equation curl H = J.
Solution
jH JR = / = JJ J-N dS
s
By Stokes’ theorem,
Jj J-N dS = JI (curlH • N) JS
s s
or, equivalently,
JI (J - curl H )-NdS = 0
Because this equation holds for any surface S bounded by C, it can be shown that
J - curl H = 0
J = curlH
If V is the velocity field of a fluid flow, then curl V measures the tendency of the fluid to
rotate, or swirl (see Figure 13.46). If the fluid flows across the surface S, the rotational
tendency usually will vary from point to point on the surface, and the surface integral
f f (curl V • N) dS provides a measure of the cumulative rotational tendency over the
s
entire surface.
Figure 13.46 The tendency of a fluid Stokes’ theorem tells us that this cumulative measure of rotational tendency equals
to swirl across the surface S is
measured by curl V • N. the line integral j) V • JR. To interpret this line integral, recall that it can be written
£ V ■ T ds in terms of the arc length parameter s and the unit tangent T to the curve.
Since the line integral sums the tangential component of the velocity field V, it mea¬
sures the rate of flow of fluid mass around C and for this reason is called the circulation
of V around C. If curl V = 0, the circulation is zero and V is said to be irrotational.
To summarize,
(curl V • N) JS = V-T ds
s
The circulation of
The cumulative tendency a fluid around the
of a fluid to swirl across boundary curve C
the surface S
914 Chapter 13 Vector Analysis
for the prescribed vector fields and surfaces. In each case, use the
upward unit normal for S.
21. F = 4yi + zj + 2yk, and 5 is the hemisphere 34. HISTORICAL QUEST George
z = \J4 - x2 - y2. Stokes was an English mathematical
physicist who made important contribu¬
22. F = (x tan-1 e~x, y ln(l + y3/2), ze~{/z), and S is the part
tions to fluid mechanics, including the
of the sphere x2 + y2 + z2 = 9 that lies inside the cone
Navier-Stokes equations, which are im¬
z = \j2.x2 + 2y2.
portant for modeling fluid flow. Most
23. F = (6x2eyz, 2x3zeyz, 2x3yeyz), and S is the part of the cone of his research was done before 1 850,
z = yjx2 + y2 that lies above the inner loop of the limayon after which he held the Lucasian chair
r = 1 + 2cos0. of mathematics at Cambridge for the GEORGE STOKES
better part of a half-century. William 1819-1903
Q In Problems 24-26, use Stokes ’ theorem to evaluate the line inte¬
gral Thomson (Lord Kelvin) knew the result
now known as Stokes' theorem in 1850 and sent it to Stokes as
[(1 +y)zdx + (1 +z)xdy + (1 +x)ydz]
a challenge. Stokes proved the theorem, and then included it
as an exam question in 1854. One of the students taking this
for the given closed path C.
particular examination was James Clerk Maxwell (Historical
24. C is the elliptic path x = 2 cos 6, y — sin#, z — 1 for Quest #32, Section 13.7), who derived the famous electro¬
0 < 6 < 2jt. magnetic wave equations ten years later. ■
25. C is the boundary of the triangle with vertices (1,0,0), For this Quest, write a history of the Lucasian chair, which
(0, 1,0), (0,0, 1). was deeded in 1663 as a gift of Henry Lucas. The first and
26. C is any closed path in the plane 2x — 3y + z — 1. second appointees were Isaac Barrow (Historical Quest #6,
In Problems 27-30, the vector field V represents the velocity of a page 48) and Isaac Newton (Historical Quest #1, page 48);
fluid flow. In each case, find the circulation and the chair is currently held by Stephen Hawking.
2 2
A 35. Let S be the ellipsoid — + + z2 = L and let F be a vec-
l \ dR 4 9
tor field whose component functions have continuous partial
derivatives on S. Use Stokes’ theorem to show that
around the boundary C, assuming a counterclockwise orientation
as viewed from above. (curl F ■ N) dS = 0
27. V = xi+(z —x)j+yk, and C is the intersection of the cylinder s
x2 + y2 = y and the hemisphere z = y/\ — x2 — y2.
Does it matter that S is an ellipsoid? State and prove a more
28. V = yi + ln(x2 -f y2)j + (x + y)k, and C is the triangle with general result based on what you have discovered in the first
vertices (0, 0, 0), (1,0, 0), and (0, 1, 0). part of this problem.
29. V = (e*2 + z)i + (x + sin y3)j + [y + ln(tan_1 z)]k, and C is 36. Faraday’s law of electromagnetism says that if E is the elec¬
the intersection of the sphere x2 + y2 + z2 — 1 and the cone tric intensity vector in a system, then
Z = \Jx2 + y2.
30. V = y2i + tan^1 zj + (x2 + l)k, and C is the intersection of (f E-dR = -~
the plane z = y and the cylinder x2 + y2 = 2x. Jc 31
31. Let F = y2i + xyj + xzk, and suppose S is the hemisphere around any closed curve C, where t is time and cp is the
x2 + y2 + z2 = 1 with z > 0. Use Stokes’ theorem to express total magnetic flux directed outward through any surface S
bounded by C. Given that
s
(curl F • N) dS
0 = y/B NdS
s
as a line integral, and then evaluate the surface integral by
evaluating this line integral. where B is the magnetic flux density, show that
3B
32. Let F = zi + xj + yk, and suppose S is a smooth surface curl E =-. Hint: It can be shown that
in R3 whose boundary is given by x — 2 cos 0, y = 3 sin0,
z = sin#, 0 < 6 < 2jt. Use Stokes’ theorem to evaluate 1//b.n« = //^.n«.
s s
(curl F • N) dS 37. The current I flowing across a surface S bounded by the
closed curve C is given by
s
<b F • dR —
Jc 1
(f)
Jc2
F dR
B is magnetic flux density and p is a constant, show that
or find a counterexample. l B dR — pi
916 Chapter 13 Vector Analysis
We used Green’s theorem to show that F-N dS = div F dA, where D is a sim-
D
ply connected domain with the closed boundary curve C. The divergence theorem
(also known as Gauss’ theorem) is a generalization of this form of Green’s theorem
that relates an integral over a closed surface to a volume integral.
F-NdS= divFdV
Evaluate F • N dS, where F = x2i -f xyj + x3y3k and S is the surface of the
5
tetrahedron bounded by the plane x + y + z — 1 and the coordinate planes, with
outward unit normal vector N (see Figure 13.47).
Solution
9 9 9
divF = ^-U'2) + t-C*y) + T'U'V) = 2x +x + 0 = 3x
dx dy dz
Figure 13.47 A tetrahedron in R3 This projects onto the triangular region in the xy-plane described by
13.7 The Divergence Theorem 917
(See Figure 13.47b.) Then, by applying the divergence theorem, we find that
JJ F
S
'NdS = JJJ divFdV = fQ fQ
R
3 xdzdydx
Solution
The solid is shown in Figure 13.48. We will show that the surface integral jf F • Nr/5
s
and the triple integral /// div F d V have the same value, where R is the solid bounded
R
by 5.
Figure 13.48 The surface of the
hemisphere z = \/9 — x2 — y2 I. Evaluation of
///* div F dV
d d d
div F = — (2x) + -(-3y) + —(5z) = 2 - 3 + 5 = 4
dx dy dz
Therefore, ///divFdV = fjf 4dV, but /// dV is just the volume of the
R R R
hemisphere z = y/9 — x'2 — y2. A hemisphere of radius 3 has volume
1 r[|tt(3)3]
4 = 187T, so
2
II. Evaluation of
fl* •Nr/5
5 consists of two parts: 5i, the disk on the bottom of the hemisphere, and 52, the
hemisphere. We will consider these separately and then use
The surface 5t: The disk x2 + y2 < 9 with z — 0 has outward (downward) unit
normal N = —k, so
Si Si
— JJ(—5z)dSi—0 Because z = 0 on 5i
Si
—x —y
Zx - and Zv =
79 "
x2 - y2 \/9 — x2 - y"
918 Chapter 13 Vector Analysis
and the projection of S2 onto the Ay-plane is the disk D : x2 + y2 < 9 or r < 3 in
polar form. (See Figure 13.49.) We find that
—A
//F N dS2
-n
D
(2a-, —3y, 5z) ■ ( -
a2 — y2
—y
x2-y2
, 1 ) dA
2xl 3y2
-II
-
+ 5z dA
v/9 - v2
Az -
- '’2
yi
D
2x2 - 3y2
-IID
\/9 — A2 — y
+ 5v/9" a2 - y2 dA Since z = yj9 — a2 — v2 on S
rut»2 n />3
2r2 cos2 0 — 3r2 sin2 9 Changing to polar
+ 5s/9^~ rrtr-,10
coordinates
Jo Jo s/9 — r2
r-27i
//F S
•N dS = jj F ■ Nr/S, + JJ F-NdS2 = 0 + J2tt =12n
Si s2
Thus, we have
F - N dS + N dS = Nr/S = div F d V
where D is the solid region bounded by S. Thus, if we can compute fff div F d V and
D
S* is the closed surface Jj F • Nr/S, we can compute //F • Nr/S by the equation
of the entire cube.
S2 Si
S is the surface of five
faces of the cube.
F- Nr/S = div F dV F-N dS
This equation can also be used as a device for trading the evaluation of a difficult sur¬
face integral for that of an easier volume integral. Here is an example of this procedure.
a cube with a missing face faces of the unit cube 0<a< 1, 0 < y < 1, 0 < z < l,as shown in Figure 13.50.
13.7 The Divergence Theorem 919
Solution
Note that the surface S is not closed, but we can close it by adding the missing face
Sm, thus forming a closed surface S* that satisfies the conditions of the divergence
theorem. The strategy is to evaluate the surface integral of S* and then subtract the
surface integral over the added face Sm.
ff FNds = ///—
5* cube
ft ■1
Jo Jo
r1
(.y-2z)dydz=
Jo
z) dz f G-2
Also, because the outward unit normal vector to the added face Sm is N = —k and
^ = 0 on this face, it follows that
Therefore,
1
F-NdS= divF d V — F-NdS = -\-0 = 2
Like Stokes’ theorem, the divergence theorem is often used for theoretical purposes,
especially as a tool for deriving general properties in mathematical physics. The fol¬
lowing example deals with an important property of fluid dynamics.
Solution
Recall from Section 13.1 that a point is called a source if div <$F > 0, a sink if div 3F <
0, and incompressible if div <5F = 0. Let S be a smooth surface in that encloses a
solid region D. In physics, it is shown that the amount of fluid flowing out of D across
S in unit time is ff <$F ■ N dS, which must equal the net decrease in density of the fluid;
//w.™ = -///£ dV
S D
By the divergence theorem, it follows that
so that
ds
dS
Iff D
div <5F H-
dt
dV = 0
This equation must hold for any region D, no matter how small, which means that the
integrand of the integral must be 0. That is,
3S
div<5F =- ■
dt
It is known that the total heat contained in a body with uniform density 8 and
specific heat o is fff a ST dV, where T is the temperature. Thus, the amount of heat
D
leaving D per unit of time is given by the derivative
9 dT
aSTdV -a8—dV
dt dt
In the following example, we use this result to obtain an important formula from math¬
ematical physics.
dT K ,
— = —v2r
dt a8
where a is the specific heat of the body and 8 is its density.
Solution
Let S be the closed surface that bounds D. Because F is the velocity of heat flow, the
amount of heat leaving D per unit time is ffF-NdS, and the divergence theorem
s
applies:
— KV2T dV
Since this is the amount of heat leaving D per unit of time, it must equal the heat
integral from physics derived just before this example. Thus,
UI-K^TdV.fff-at^y
D D
This equation holds not only for the body as a whole, but for every part of the body, no
matter how small. Thus, we can shrink the body to a single point, and it then can be
shown that when this occurs the integrands are equal; that is.
dT
-KV2T = —o8
~dt
dT K
— = —
dt cr 8
Recall from Section 13.4 that the normal derivative dg/dn of a scalar function g
defined on the closed surface S is the directional derivative of g in the direction of the
outward unit normal vector N to S\ that is,
dg
/ = Vg • N
dn
We will use this equation in the following example, which involves a generalization of
a property we first obtained for R2 in Example 6 of Section 13.4.
13.7 The Divergence Theorem 921
//A[/v2g + V/
D
Vg]dV=JJf9£dS
s
This is called Green’s first identity.
Solution
We will apply the divergence theorem to the vector field F (note that F is continuously
differentiable), but first we need to express div F in a more useful form.
div(/Vg) = V • (/Vg)
a a a
i H-j H-k
dx dyJ 9z dx dy dz
_a_
fdJL 'fdA
a dg
+ f
dx J dx dy dy. + Tz dz
d2g1 Wdg a2gn
dldA + f d2g1 + VdJL + f +
dx dx dx2 dy dy dy2 dz dz dz2
d/ 3g dfdg dfdg' r da2 g , d2g d2g
+ f + +
dx dx dy dy dz dz. _dx2 dy2 9z2_
= (V/) • (Vg) + fV2g
This calculation gives us the first step in the following computation.
— //
JJ <
(/Vg) -NrfS Divergence theorem
s
-II
s
/(Vg-N)d5
ag 3g
-II a f—dS
n
Because Vg • N =
by definition
3/7
div F0 = lim —— [ [ F • N dS
o V(r) J J
Sir)
where div F0 denotes the value of div F at P0, and S(r) is a sphere centered at P0 with
volume V(r) = |nr3.
922 Chapter 13 Vector Analysis
Solution
Applying the divergence theorem to the solid sphere (ball) B(r) with surface S(r), we
obtain
F-NdS= div F d V
S(r) B(r)
where divF* denotes the value of divF at some point P* in the ball B(r). Combining
these results, we find that
N dS =
Iff divF dV = V (r) divF*
or
1
F-N dS = divF*
VV)
Since the point P* is inside the ball B(r) centered at P0, it follows that P* ->■ P0 as
r -> 0, so div F* —> div Fq and we have
lim —-—
'■-o V(r)JJ
ff
F-N dS = limdivF* = divF0
r^O
S(r)
as claimed.
3. F = 2y2j; D is the tetrahedron bounded by the coordinate 11. F = xi+yj+zk; S is the paraboloid z = x2+y2 for 0 < z < 9.
planes and the x + 4y + z = 8. 12. F = curl(yi+xj—zk); S is the hemisphere z = y/4 — x2 — y2
4. F = 3xi + 5yj + 6zk; D is the tetrahedron bounded by the together with the disk x2 + y2 < 4 in the xy-plane.
coordinate planes and the plane 2x + y + z = 4. 13. F = x2i + y2j + z2k; S is the sphere x2 + y2 + z2 = 4.
Use the divergence theorem in Problems 5-19 to evaluate the sur¬ 14. F = xyzi+xyzj+xyzk; S is the surface of the box 0 < x < 1,
face integral ff F • N dS for the given choice of F and closed 0<y<2, 0<z<3.
s
15. F = xi + yj + (z2 — l)k; S is the surface of a solid bounded
boundary surface S. Assume N is the outward unit normal vec¬
tor field. by the cylinder x2 + y2 = 4 and the planes z — 0 and z = 1.
16. F = (x5+ 10xy2z2)i+(y5 + 10yx2z2)j + (z5+ 10zx2y2)k; S is
5. F = xi + yj + zk; S is the cube 0 < x < 1, 0 < y < 1,
the closed hemispherical surface z = y/\ — x2 — y2 together
0 < z < 1.
with the disk x2 + y2 < 1 in the xy-plane.
6. F = xyzj; S is the cylinder x2 + y2 = 9, for 0 < z < 5.
17. F = xy2i + yz2j + x2yk; S is the surface bounded above by
7. F = (cos yz)i + ex:) + 3z2k; S is the hemisphere the sphere p = 2 and below by the cone (j) = | (in spherical
z = \/4 — x2 — y2 together with the disk x2 + y2 < 4 in the coordinates). Note: S is the surface of an “ice cream cone.”
jcy-plane.
18. F = xy2i + yz2j + x2zk; S is the surface bounded above by
8. F = curl[evzi — 4j + (sin.ryz)k]; S is the ellipsoid the sphere p — 2 and below by the cone f ^ (in spherical
2a-2 + 3y2 + lz2 = 1. coordinates). Note: S is the surface of an “ice cream cone.”
13.7 The Divergence Theorem 923
19. F = x3i + y3j + 3a2zk (constant a > 0); S is the surface 27. Show that if g is harmonic in the region D, then
bounded by the cylinder x2 + y2 = a2 and the planes z = 0
and z = 1 ■
— dS = 0
Q 20. Suppose that S is a closed surface that encloses a solid re¬ dn
gion D.
a. Show that the volume of D is given by
where the closed surface S is the boundary of D. (Recall that
g harmonic means V2g = 0.)
V(D)
=\u (xi + yj + zk) • N dS
T
b. Show that
///(/v’s-svV)^// (/|-4)rfs
D S
fffQdV=*
924 Chapter 13 Review
Chapter Checklist j k
a. F is conservative on D; that is, F = V/ for some function Stokes’ theorem: Let S be an oriented surface with unit
/ defined on D. normal vector field N, and assume that S is bounded by a
L ¥ dR = fiQ)- fiP) where N is the outward unit normal field for the surface S.
where Q = R(b) and P = R(a) are the endpoints of C. EVALUATION OF FLUX INTEGRALS: F-N dS
EVALUATION OF LINE INTEGRALS: JF dR Step 1. If the surface S is a closed surface bounding the solid re¬
gion D, use the divergence theorem to write the flux inte¬
Step 1. Check to see whether F is conservative; if it is, then gral as a triple integral.
y, z)dx
f
Ja
f[x(t), y(t), z(t)}~ dt
dt Proficiency Examination
CONCEPT PROBLEMS
J f(x,y, z)ds 1. What is a vector field?
2. What is the divergence of a vector field?
fh
= / f[x(t), y(t), z(t)Ulx'(t)]2 + [y'(0]2 + [z'(t)]2dt 3. What is the curl of a vector field?
Ja
4. What is the del operator?
from (1,0) to (7i, 0), followed by the fine segment from (n, 0)
to {n, n)
23 . J (x dx + x dy — y dz), where C is the curve 39. F(x, y) = (y/x + 3xy)i + (cx2 + 4y)j
24-l (x2 dx — 3y2 dz), where C is the line segment from 41. F(,.,.Z)=^[(f)i+0j+0k]
(0, 1, 1) to (1, 1,2)
42. Let F = (_y2 + x 2yex,y)i + (2xy + z — x 1 )j + yk. Is F
2S'l (x2 dx + y dy), where C is the curve conservative?
43. Find the work done when an object moves in the force field
R(r) = (t sin r)i + (1 — t cosr)j, 0 < t < 2n
F = 2;ri — (* + z)j + (y — *)k along the path given by
26-i (xy dx — x2 dy), where C is the square with vertices (1,0), R(t) = ri + (r - r)j + 3k, 0 < t < 1.
(0. 1), (—1,0), (0,-1) traversed counterclockwise 44. Show that the force field F = yz2i + (xz2 — 1 )j + (2xyz — 1 )k
is conservative, and determine the work done when an object
21-i (y dx + x dy — 2dz), where C is the curve of intersection moves in the force field from the origin to the point (1,0, 1).
45. Find a region R in the plane where the vector field
of the cylinder x2 + y2 = 2x and the plane x = z, traversed
counterclockwise as viewed from above F = -(i + j) is conservative. Then evaluate / F clR,
28 • £[(,y + z)dx + (x + z)dy + (x + y)dz], where C is the
x+y Jc
where C is any path in R from the point Po(a, b) to Fj (c, d).
curve of intersection of the sphere x2 + (y — 3)2 + z2 = 9 46. If u is a scalar function and F is a continuously differentiable
and the plane x + 2y + z — 3, traversed counterclockwise as vector field, show that curl(MF) = u curl F + (Vu x F).
viewed from above 47. Show that div(F x G) = G • curl F — F • curl G, for any con¬
tinuously differentiable vector fields F and G.
29 ■l (—2y dx + 2x dy + dz), where C is the circle x2 + y2 = 1
48. A vector field F is incompressible in a region I) if divF = 0
in the plane z — 3, traversed counterclockwise as viewed from throughout D. If F and G are both conservative vector fields
above in D, show that F x G is incompressible.
307/ (curl yi) • N dS, where S is the hemisphere 49. If F = curl G, show that F is incompressible. (See Problem
48.)
50. Suppose F = fix, y, z)A, where A is a constant vector and /
Z = y/l - X2 - y
is a scalar function. Show that curl F is orthogonal to A and to
31.
ff (2x' i + y j + z k) • N dS, where S is the surface of the
n-3‘' ~3‘' -3
v/.
51. If A is a constant vector and F is a continuously differentiable
ellipsoid 2x2 + y2 + z2 = 1 vector field, show that div(A x F) = —A • curl F.
, . f x dx — y dy
32. Jf<?Ii + y j + yzk) ■ N dS, where S is the surface of the 52. Evaluate the line integral / --———, where C is any path
Jc x2 - y2
in the *y-plane that is interior to the region x > 0, y < x,
tetrahedron bounded by the plane 2x + 3y + z = 1 and the y > —x and connects the point (5, 4) to (2, 0).
coordinate planes, in the first octant
l
—y 1 \
53. Evaluate —— dx 4— dy , where C is the closed path
33.
// V0 • NdS, where <p(x, y, z) = 2x + 3y and S is the por¬ x1 x
(x — 2)2 + y2 = 1, traversed once counterclockwise.
tion of the plane ax + by + cz = 1 (a > 0, b > 0, c > 0) that 54. Counterexample Problem Let u(x, y) and i>(*. y) be functions
lies in the first octant of two variables with continuous partial derivatives every¬
34. F = x2i + y2j + x2k, and S is the surface of the unit cube dy dx
0<jt<l,0<y<l,0<z<l. for all (*, y). Are u and v necessarily harmonic? Either show
35. F = 2yzi + (tan-1 *z)j + exyk, and S is the surface of the that they are or find a counterexample.
sphere x2 + y2 + z2 = 1. 55. a. Find a region in the plane where the vector field
36. F = ,vi — 4j + 3k, and S is the paraboloid y = x2 + z2 with
x2 + z2 < 9. The disk x2 + z2 — 9 is omitted; that is, the 1+77. / y +*2y
paraboloid is open on the right.
37. F = x yzi + xyzj + xyzk, and S is the surface of the five faces is conservative, and find a scalar potential for F.
of the unit cube 0 < x < l,0<y< 1, 0 < z < 1, missing
z = 0. b. Evaluate lFJR , where C is a path from (1, 1) to (3, 4).
38. F = xyi — 2zj, and S is the surface defined parametrically by Are there any limitations on the path C? Explain.
R(u, d) = ui 4- uj + wk for 0 < u < 1, 0 < v < 1. 56. Determine the most general function u(x, y) for which the
Find all real numbers c for which each vector field in Problems vector field F = u{x, y)i + (2yex + y2eix)j will be conser¬
39-41 is conservative. vative.
928 Chapter 13 Review
58. If S is a closed surface in a region R and F is a twice continu¬ 68. Suppose / and g are both harmonic in the region R with
ously differentiable vector field on R, show that boundary surface S. Show that
// (curl F • N) dS = 0
l|V/||2rfV
where N is the outward unit normal vector to S. “//'£"-///
S D
59. A certain closed path C in the plane 2x + 2y + z = 1 is known
69. Show that curl(curlF) = V(divF) — V2F if the components
' to project onto the unit circle x2 + y2 = 1 in the xy-plane. Let
of F have continuous second-order partial derivatives.
c be a constant, and let R = xi + yj + zk. Use Stokes’ theorem
to evaluate 70. Evaluate the surface integral [[ — dS, where S is the sur-
JJ 3 n
l (ck x R) • dR
b. Let w = x — y + 2z, and let S be the surface of the sphere R(«, v) = [(a + b cos v) cos u]i + [{a + b cos v) sin u]j
x2 + y2 + z2 = 9. Evaluate + (b sin u)k
for 0 < b < a and 0<u<2tt,0<v<27t.
s
72. Evaluate JJ F • N dS, where F = xi + yj + zk and S is the
61. A particle moves along a curve C in space that is given para¬ s
metrically by R(f) = x(r)i + y(f)j + z(?)k for a < t < b. A closed cubic surface with a comer block removed, as shown
force field F is applied to the particle in such a way that F is in Figure 13.52.
always perpendicular to the path C. Flow much work is per¬
formed by the force field F when the particle moves from the
point where t = a to the point where t = bl
62. If D is the electric displacement field, then divD = 0, where
0 is the charge density. A region of space is said to be charge-
free if 0 = 0 there. Describe the charge-free regions of the
electric displacement field D = 2x2i + 3y2j — 2z2k.
63. A satellite weighing 10,000 kg travels in a circular orbit 7,000
km from the center of the earth. How much work is done by
gravity on the satellite during half a revolution?
64. If F is a conservative force field, the scalar function / such
that F = —V/ is called the potential energy (see Problem 54, Figure 13.52 Cube with a comer removed
Section 13.3). Suppose an object with mass 10 g moves in the
73. Counterexample Problem Prove or disprove that it does not mat¬
force field in such a way that its speed decreases from 3 cm/s
ter which comer is removed in Problem 72.
to 2.5 cm/s. What is the corresponding change in the potential
74. Show that a lamina that covers a standard region D in the plane
energy of the object?
with density S = 1 has moment of inertia
65. Find the work done by the force field
y I = (-y3 dx + x3 dy)
-1
F = 2xyzi + + y + — tan
zl
with respect to the z-axis, where C is the boundary curve
in moving an object along the circular helix of D.
15. Use vector analysis to find the centroid of the conical surface
R (t) — (sin jr/)i + (cos7rf)j + (2f + l)k
z = Jx1 + y2 between z = 0 and z — 3.
for 0 < t < |. 76. Putnam Examination Problem A force acts on the element ds of
66. Find the work done when an object moves against the force a closed plane curve. The magnitude of this force is r~] ds,
field F = 4y2i + (3x + y)j from (1,0) to (— 1,0) along the top where r is the radius of curvature at the point considered, and
2 y2 the direction of the force is perpendicular to the curve; it points
half of the ellipse x + — 1. Which value of k minimizes to the convex side. Show that the system of such forces acting
k
the work? on all elements of the curve keeps it in equilibrium.
Guest Essay
929
930 Chapter 13 Vector Analysis
b
J f(x,y, z) ds
J fix) dx
Subdivisions on Subdivisions on a
the x-axis curve C in space
Fundamental theorem of calculus (Section 5.4) Fundamental theorem for line integrals (Section 13.3)
f
Ja
f(x)dx = Fib)- F(a) Jc F • JR = /(Q) — f{P)
Green’s theorem (Section 13.4): F(jt, y) — M(x, y)i + N(x, y)j is continuously differentiable on the simply connected
region D with positively oriented piecewise smooth boundary curve C. Then
f rr/dN 8M'
(J) {M dx + N dy) — dA
Jc dx dy
D
// 8(x,y,z)dS =
S
JJ
R
Six, y, z)yjff + .A2 + 1 dA
where z = fix,y)
n- °f jJ 3x dy dx
y+2 x+5
units of the product. If manufacturing costs for this product
• i cz py~z
.
12
Ill »157t
p\5tt pi7ir
p
(x + y + z) dx dy dz
/'sin
ps d>
are $150 per unit, how much should the manufacturer spend
on development and how much on promotion to generate the
largest possible profit, given the following circumstances?
13. I / / p3 sirup dp dO dtp a. Unlimited funds are available
Jo Jo Jo
b. The manufacturer has only $11,000 to spend on develop¬
14. JI ex+y dA\ R : 0 < x < 1; 0 < y < 1 ment and promotion of the new product
R
25. A heat-seeking missile moves in a portion of space where the
temperature (in degrees Celsius) at the point (x, y, z) is given
15.
II yexy dA\ /?:0<x<l;0<y<2 by
17.
II xsinxydA;/?:0<x<;r;0<y< 1
as (x, y, z) moves from the point P0i—2, 9, 1) toward
0(1, —3, 5).
b. If the heat-seeking missile is at P$, in what direction will it
Evaluate the line integrals in Problems 18-21. travel to maximize the rate of heat increase?
c. What is the maximal rate of increase (in degrees Celsius
dx + Ixyzdy + xy2 dz), where C is any path from
i87c(o, iy2z
o.o) tod. i, i)
per kilometer)?
26. Find the volume of the solid bounded above by the surface
z — x2 -F y2 -F 1, below by the xy-plane, and on the sides by
i97 (5xy dx + 10yzdy + zdz), where C is given by x = t2,
the cylinder x2 + y2 = 1.
y = t, z = 2t3 for 0 < t < 1 27. Find the surface area of that portion of the paraboloid
z — x2 + y2 that lies below the plane z — 16.
F • dR, where F = yzi — xk, and C is the boundary of
Ic
28. Find the center of mass of the lamina that covers the region R
the triangle (1, 1, 1), (1,0, 1), (0, 0, 1), traversed once coun¬ inside the circle x2 + _y2 = 4 in the first quadrant, given that
terclockwise as viewed from the origin the density at any point (x, y) is Six, y) = x + y.
934 Chapters 11-13 Cumulative Review
29. A force field F(x, y) = (x — 2y)i + (y — 2x)j acts on an object 30. A particle of weight w is acted on only by the constant gravi¬
moving in the plane. Show that F is conservative, and find a tational force F = —ink. How much work is done in moving
scalar potential for F. How much work is done as the object the weight along the helical
al path given by x = cost, y = sin t,
moves from (1,0) to (0. 1) along any path connecting these z — t for 0 < t < 2n ?
points?
CONTENTS
935
936 Chapter 14 Introduction to Differential Equations
and observed that such an equation can be solved by separating the variables and inte¬
grating each side; that is,
J f(y)dy = J g(x) dx
dy _y .
— = e y sin v
dx
Solution
Separate the variables and integrate:
dy
— e y sinx
dx
ey dy — sin* dx
j ey dy = j sinxdx
ey = — cosx + C Combine constants of integration.
From the standpoint of applications, one of the most important separable differen¬
tial equations is
dy_
— ky
dx
which occurs in the study of exponential growth and decay. You might review Table
7.3 on page 466, which gives the solution for uninhibited growth or decay, logistic
(or inhibited) growth, and limited growth functions. In this section, we consider an
application of separable differential equations from chemistry.
14.1 First-Order Differential Equations 937
Solution
Since each unit of C is formed from one unit of A and one unit of B, it follows that at
time t, a — Q(t) units of A and — 0(0 units of B remain unconverted. The specific
rate condition can be expressed mathematically as
dQ
= k(a - <2)06 - Q)
dt
where ft is a constant (ft > 0 because Q(t) is increasing).
To solve this equation, we separate the variables and integrate.
dQ
l (a - 0)08 - Q) -!
1
kdt
/At?Q + P Q
dQ
-I ft dt Partial fraction
decomposition
BMe° — a
0 = -—77-
Me0 - 1
0 = pM -a
a
M — —
P
p*e(«-P)kt _a
P
apie^-W 1]
Thus, 0(0
— e(a~P)kt _ ] ae(a-P)k‘ — ft
P
HOMOGENEOUS DIFFERENTIAL EQUATIONS
Sometimes a first-order differential equation that is not separable can be put into sepa¬
rable form by a change of variables. A differential equation of the form
dy r (T\
938 Chapter 14 Introduction to Differential Equations
In other words, dy/dx is isolated on one side of the equation and the other side can
be expressed as a function of y/x. We can then solve the differential equation by
substitution.
To see how to solve such an equation, set v = y/x, so that
vx = y
d d
Take the derivative of both sides.
Tdy)
dv Jy
v + x— - Product rule
dx dx
dv dy /y\ n s
v + x— = f(v) Substitution, = /( — )= f(v)
dx dx Vx /
dv
f(v) - V
dx
dv dx
f(v) - v x
The equation can now be solved by integrating both sides; remember to express your
answer in terms of the original variables x and y (use v = y/x).
Solution
dy
First, show that the equation is homogeneous by writing it in the form — =
dx
2xy dx + (x2 4- y2) dy — 0
dy —2xy
dx x2 + y2
y —2v
Let v = - and f(v) - ——
x 1 + v1
dv dx
f(v) ~ v x
dv dx
—2v x
_ r
J
1 +
(i + v2)
v2 + 3v
V2
dv
vl)dv
— v
r
J
, dx
f
3 +
v ' v2 + 3
3
!„ -/= x dx Partial fraction
decomposition
In + In lx31 = C2 Substituting
®2 +3
v = -;C2 — 3C|
x
14.1 First-Order Differential Equations 939
y- + 3jt
In = C2
y3 + 3 x3 = C Where C — eCl
1
y =
I(x) = efpMdx
Solution
Comparing the given first-order linear differential equation to the general first-order
form, we see
1
y =- /(secxXsec^ + C
secx
= cosx anx + C]
= sinx C cosx
y = sinx — 2 cosx ■
Solution
The downward force is the weight mg of the body and the upward force is —kv, where
k is a positive constant (the negative sign indicates that the force is directed upward).
According to Newton’s second law, the sum of the forces acting on a body at any time
equals the product ma, where a is the acceleration of the body, that is
ma - mg kv
dv (I v
m— = mg kv Since a
dt dt
dv
~T7 ~ &-v
dt m
dv k
——|-v 8
dt m
This is a first-order linear differential equation, where p(t) — — and q(t) = g. The
m
integrating factor is
uf
1_
mg
^ .
— g-kt/m
u
+
v = ek,'m(g)dt + C + Ce~kt/m
ckt/m
k/m
0 = ^ + Ce» = ^ + C
k k
mg
Solving, we obtain C -, so
mg -kt/m
V = T + (-t)
ds
Now, to find the position s(t), we use the fact that v(t) =
dt
ds_^mg__ ing_e-kt/m
mg mg e kt^m
+c
J<') = T'~T^
mg mg rllm
= -r-t + -?-e-k,/m + C
k kl
14.1 First-Order Differential Equations 941
Because i'(O) = 0 (the position s is measured from the point where the object is
dropped), we find that
mg m e n m2g
0= -^(0) + -jfe° + C so that = C
It
Thus, the position is
1)
In the problem set, you are asked to show that no matter what the initial velocity
may be, the velocity reached by the object in the long run (as / -» +oo) is mg/k.
M(x, y) dx + N(x, y) dy = 0
df — M(x, y) dx + N(x, y) dy
for some function /. In this case, the given differential equation is appropriately called
exact and since df — 0, its general solution is given by f(x, y) = C.
But how can we tell whether a particular first-order equation is exact, and if it is,
how can we find /? Since
df df
df — — dx H-dy
J dx dy
df — — dx + — dy — M(x, y) dx + N(x, y) dy
dx 3y
In particular, suppose M and N are differentiable, with
and then the function f(x,y) is found by partial integration, exactly as we found the
potential function of a conservative vector field in Section 13.3. The procedure for
identifying and then solving an exact differential equation is illustrated in Example 6.
Solution
Let M(x,y) = 2.v_y3 + 3y and N(x,y) — 3x2y2 + 3x, and apply the cross-derivative
test
3M „ 2 „ J dN , 2 0
- - 6xy2 + 3 and -— 6xyl + 3
dy ' dx
dM dN , . . _ , . ... „ ,
Because-=-. the equation is exact. To obtain a general solution, we must find
3y dx
a function / such that
To find /, we integrate the first partial on the left with respect to jc:
df 3 _ ,
+ 3xy + «(y)] = 3 x2y2 + 3x + u'{y)
so that
This implies that u is a constant. Taking u = 0, we have / = x2y3 + 3xy, and the
general solution to the exact differential equation is
x2y3 + 3xy = C ■
dy
+P(x)y = q (x) Use the integrating factor I(x) — p{x)dx y -
I(x) J I{x)q{x) dx + C
EULER’S METHOD
dy
fa = fix, y) y(x0) = y0
The key idea in Euler’s method is to increment x0 by a small quantity h and then to
estimate y\ — y(xi) for x\ = x0 + h by assuming x and y change by so little over the
* In our discussion of Euler’s method, we assume that the given initial value problem has a unique solution.
It can be shown that such an initial value problem always has a unique solution if / and df/dy are both
continuous in a neighborhood of (xq, yo). The proof of this result is beyond the scope of this text but can be
found in most elementary differential equations texts.
14.1 First-Order Differential Equations 943
interval [xo, xq] that /(x, y) can be replaced by f(xo, yo) for this interval. Solving the
approximating initial value problem
dy
— = fix o, y0) y(x0) = yo
dx
we obtain,
y - yo = fix o, yo)(x -x0)
In other words, we are approximating the solution curve y = y(x) near (xo, yo) by the
tangent line to the curve at this point, as shown in Figure 14.1a. We then repeat this
process with (xi, yj) assuming the role of (x0, yo) to obtain an approximation of the
solution y = y(x) over the interval x\ < x < X2, where X2 = x\ + h and
y2 = yi + hfixuyi)
a. The first Euler approximation Continuing in this fashion, we obtain a sequence of line segments that approximates
the shape of the solution curve as shown in Figure 14.1b. Euler’s method is illustrated
in the following example.
Solution
Before using Euler’s method, we might first look at a graphical solution. The slope
field is shown in Figure 14.2a, and the particular solution through the point (0, 1) is
b. A sequence of Euler approximations shown in Figure 14.2b.
of Euler’s method // // / / // / / // / /
—2 .J / / / / / / / / f / / s
/ /\a A/ / / /
/ /b'7" / 7 7 /
/ / / / / / / / / / / / / / //!// A A / /
// / / / / / / / / / / A / A A / /
"Y'
A-A-U../.. A* A A
2 .0- // a a // ^A A-A A-A //
a a / / a a // A A A A / A A A\A A A A / /
// A JTT'P'V 7 7 77
✓ / ✓ ✓ / / ✓ / A A A A\A A ' +*A A A A
.5
✓ / ✓ / #A AA " AA A A AA
'
-- a a A a A A A A A a \a A A A A A
0-
0 ^ -- AA A X A A A A
aa\aa A A A A
" - *" -*
( .5
— — — -- -- A a\aa A A
A>\a -
_ — —| _ __—
1
1 1
a 2_ Lo 4_ 0,6 (A
a. Slope field of ~ = x + y2 b. Particular solution through (0,1)
yo = y(0) = 1
yi — yo + hf (x0, yo) — 1 +0.1(0+ l2) = 1.1000
y2 = yi +hfixuyi) = 1.1000 + 0.1(0.1 + 1.10002) = 1.2310
y3 = y2 + hf(x2,y2) = 1.2310 + 0.1(0.2+ 1.23102) « 1.4025
y4 = y3 + /7/(x3,y3) = 1.4025 + 0.1(0.3 + 1.40252) % 1.6292
y5 = y4 + hfix4, y4) = 1.6292 + 0.1 (0.4 + 1.62922) % 1.9347
944 Chapter 14 Introduction to Differential Equations
These points can be plotted to approximate the solution, as shown in Figure 14.3.
Notice that we plotted these points by superimposing them on the direction field shown
in Figure 14.2b. ■
Euler’s method has educational value as the simplest numerical method for solv¬
ing ordinary differential equations and can be found in most computer-assisted pro¬
grams. However, as you might guess by looking at Figure 14.3, as you move away
from (xo, yo)> the error may accumulate. The Euler method can be improved in a vari¬
ety of ways, most notably by a collection of procedures known as the Runge-Kutta and
predictor-corrector methods. These methods are studied in more advanced courses.
r '.
+ x2(y + l) - 3x - dy — 0
l//l//l//i// 6VTJ
14.1 First-Order Differential Equations 945
✓
"Vs. s s y y y y y / y y4 v y y y
/ y y\y yiy y y y y y ✓ / y y y /
47. x— - 3y = x3; x = 1, y = 1
dx
*
<
—"
y y\y y
r
/y y y y y y y y
y / / T /r / / dy
x-
A
y\y///, yy / / / / / / f / / / 48. — = 1 + 3y tanx; x = 0, y = 2
•5‘
x- y\y y^/y y y / / y / // / / / / dx
A
- y/\y/V/ / / / / / / / / / / 49. [sin(x2 + y) + 2x2 cos(x2 + y)] dx + [x cos(x2 + y)] dy = 0;
dr TTvrrsrr rr rr rr t-f- 11
i
y y y/V/// / / / / 1 1 1 1 \ l x = 0, y = 0
y y y y y A/ / / / /1 1 l \ v \ \^ \
n
V-V V-V
—0
✓ /[//]» l 1 \ \\ \\ \ \ S v \\ 50. y— — ex+2y sinx; x = 0, y = 0
/ 1 jj \ v \ \'l' ^
y I wkxkskx V Vs dx
1 1 1
02 _M 0.6 0 8
A
51. yex dy — (y2 + 2y + 2)dx\x = 0, y = — 1
52. (2xy + y2 + 2x) dx + (x2 + 2xy — 1) dy = 0; x = 1, y = 3
dy 5x — 3 xy -n dy_ _ V2 + 2x
29. 53. In the chemical conversion analyzed in Example 2, what hap¬
dx 1+ x2 dx 3y2-2xy'
passing through (0, 0) passing through (0, 1) pens to <2(/) as t —*■ +00 in the case where a 7^ /?? What if
for 0 < x < 0.5, h = 0.1 for 0 < x < 0.5, h = 0.1 a =
54. a. Find an equation for the velocity of the falling body in
JJ
T ✓^ y y y y yy //
.3
-2 x/ y y y y y y. r y / Example 5 in the case where the body has initial velocity
^ Ay y\y y // y y y y/ y y / ✓
—2
/ t-A rr\/-A u0 ^ 0.
y y yy y / / /
✓ / y y /V y y y / / / / / / I
b. Show that for any initial velocity (zero or nonzero) the ve¬
1 .5 ~ y ✓/
— ^ y y / / / / / / / 1 locity reached by the object “in the long run” (as t —> +00)
/ / ✓ / / / / / 1 l 1 1 \ \
is always mg/k.
yyyy-y
1 // / / / 1 l 1 1 i 1 i \ \ \
yy\/
"S A/ // / / 1 11 11 1 1 \ \ \ \ 55. A population of foxes grows logistically until authorities de¬
/./ u u tx U U \ \
d>
sX./ 1
cide to allow hunting at the constant rate of h foxes per month.
1
56. A chemical in a solution diffuses from a compartment where Q 60. Almost homogeneous equations Sometimes a differential equa¬
the concentration is C0(t) = le~' across a membrane with tion is not quite homogeneous but becomes homogeneous with
diffusion coefficient k = 1.75. The concentration C{t) in the a linear change of variable. Specifically, consider a differential
second compartment is modeled by equation of the form
dC dy ax + by + c
— = 1.75[C0(0-C(f)] = f
dx rx + sy + t
Solve this equation for C(f), assuming that C(0) = 0. a. Suppose as ^ br. Make the change of variable x = X +A
57. The formula for the solution of a first-order linear equation
and v = Y + B, where A and B satisfy
requires the differential equation to be linear in x. Suppose,
| a A + bB + c = 0
instead, that the equation is linear in y; that is, it can be written
in the form | r A + sB + t = 0
the Spy is once again hot on the trail of Blohardt and the
Flamer. He enters an open field and sees Blohardt pursuing the Hint: There is a solution of the general form y — Ax.
Flamer in the distance. Why? Could his old friend have had 62. Journal Problem The American Mathematical Monthly* ■
a change of heart? His mathematical mind visualizes the field Find all solutions of the Riccati equation
as a coordinate plane (units in yards) with himself at (100, 0),
/ 2 a
the Flamer at (0, 0), and Blohardt at (0, -80). The Spy and u = u 4—u — b
Blohardt both run directly at the Flamer, who is running along x
the positive y-axis at the rate of 12 ft/s. If Blohardt runs at the a, b ^ 0, that are real rational functions of x.
rate of 15 ft/s, how fast must the Spy run in order to reach his
friend before the villain? *Problem E3055, Vol. 91, 1984, p. 515.
14.2 Second-Order Homogeneous Linear Differential Equations 947
LINEAR INDEPENDENCE
Linear Dependence and The functions yi, y2,..., yn are said to be linearly independent if the equation
Independence
Ciyi + C2y2 H-f- Cnyn = 0 for constants C(, C2,..., C„
has only the trivial solution Ci — C2 = ■■■ — Cn = 0 for all x in the interval I.
Otherwise the yk's are linearly dependent.
The functions yi = cosx and y2 — x are linearly independent because the only
way we can have C\ cos x + C2x = 0 for all x is for C\ and C2 both to be 0. However,
yj = 1, y2 = sin2x, and y3 = cos2x are linearly dependent, because
Wronskian The Wronskian VP(yl5 y2,..., yn) of n functions yx, y2,... ,yn having n - 1
derivatives on an interval I is defined to be the determinant function
y\ yi ■■■ yn
yi y'i ■■■ y'n
W(yi,y2,-- - ,yn) =
are all continuous on a closed interval [c, d]. Then solutions y\, y2, ■ ■ ■, yn of this
differential equation are linearly independent if and only if the Wronskian is nonzero;
that is,
W(yi, y2,---,yn)^0
throughout the interval [c,d].
Solution
e x xe x e3x
'h
1
i
W (e~x, xe~x, e3x) = 3e3x
1
e x (x -2)e x 9e3x
= e-x[9e3x(l-x)e~x — 3e3x
- xe~x[-9e3xe~x - 3e~xe3x]
Proof We will prove that if and y2 are linearly independent solutions, then
y = C\y\ + C2y2 is also a solution. However, the proof that all solutions are of
this form is beyond the scope of this book. Suppose y\ and y2 are solutions, so that
Thanks to the characterization theorem, we now know that once we have two linearly
independent solutions yi, y2 of the equation ay" + by' + cy = 0, we have them all
because the general solution can be characterized as y = Ciyi + C2y2- Therefore,
the whole issue of how to represent the solution of a second-order homogeneous lin¬
ear equation with constant coefficients depends on finding two linearly independent
solutions.
Recall that the general solution of the first order equation y' + ay = 0 is
y = Ce~ax. Therefore, it is not unreasonable to expect the second-order equation
ay" -I- by' + cy = 0 to have one or more solutions of the form y — erx. If y = erx,
then y' = rerx and y" = r2erx, and by substituting these derivatives into the equation
ay" + by' + cy = 0, we obtain
ay" + by' + cy = 0
erx{ar2 + br + c) = 0
ar2 + br + c = 0 erx ^ 0
Solution
Begin by solving the characteristic equation:
r2 + 2r - 3 = 0
(r-l)(r + 3) = 0
r = 1, —3
The particular solutions are yx = ex and y2 = e 3jr. Next, determine whether these
equations are linearly independent by looking at the Wronskian.
e-3x
-2x
W(ex,e~3x) ex(—3e~3x) — exe~3x = -4e
-3e~3x
Because W(ex, e~3x) = -4e~2x ^ 0, the functions are linearly independent, and the
characterization theorem tells us that the general solution is
When a characteristic polynomial r2+ar + b does not factor as easily as the one in
Example 2, it may be necessary to find the roots r\ and r2 of the characteristic equation
by applying the quadratic formula to obtain
—b ± \Jb2 — 4ac
The discriminant for this equation, b2 - 4ac, figures prominently in the following
theorem.
950 Chapter 14 Introduction to Differential Equations
—b + \Jb2 — 4 ac —b — \Jb2 — 4a c
r i = --- and r? =-
2a 2a
then the general solution of the differential equation can be expressed in exactly one of
the three following forms, depending on the sign of the discriminant b2 — 4ac:
b1 - 4ac > 0: The roots r, and r2 are real and distinct and the general solution
is
y = Cierix + C2er2X
for arbitrary constants Cj and C2.
b2 — 4ac = 0: The roots ry and r2 are real and equal, ry = r2 =-. The gen¬
eral solution is
ry = cr + and r2 = a - pi,
b 4ac — b2 __
where a = P =---, and i = V-l.
2a 2a
The general solution is
Proof We have just seen that the solutions of the differential equation of the form
y = erx correspond to the solutions of the characteristic equation ar2 + br + c = 0.
The quadratic formula characterizes the three cases according to the discriminant of
this equation, namely, b2 - 4ac. For each case, we must find two linearly independent
solutions.
enx er2X
W(erix,er2X) = r2e{ri+ri)x - rye{r'+r2)x = (r2 - rl)e(r 1+r2)x
ryer'x r2enx
b2 — 4ac = 0 : In this case, the characteristic equation has one repeated root,
y = Cyerx + C2xerx
are linearly independent and satisfy the given differential equation. You are asked
to verify these facts in Problem 51. □
14.2 Second-Order Homogeneous Lineor Differential Equations 951
Solution
Solve the characteristic equation
r2 + 4r + 4 = 0
(r + 2)2 = 0
r — —2 (multiplicity 2)
The roots are rx — r2 = -2. Thus, yi = e~lx and y2 = xe~2x, so the general solution
of the differential equation is
y = C\6 T C2X6 2x ■
Solution
Solve the characteristic equation:
2r2 + 3r + 5 = 0
—3 ± s/9 — 4(2)(5)
r ~ 2(2)
—3 ± V3l 1
= 4
The roots are
-i
4
Thus, the general solution is
V3l ^ . V3l
y = e(-3/4)x Ci cos ——x + C2 sin ——x
Solution
Solve
4r2 + 12r + 9 = 0
(2r + 3)2 = 0
3
n=r2 = --
y = C,/-3/2)* + C2xe^2)x
3 = Cle° + C2(0)e°
3 = Ci
952 Chapter 14 Introduction to Differential Equations
-2 = -§Ci + C2
—2 = — |(3) + C2 Because Cj = 3
i = c2
Thus, the particular solution is
y = 3e^x + \xe^x U
Solution
Solve the characteristic equation:
r4 — 5r3 + 6 r2 + 4r — 8 = 0
Because this is 4th degree, we use synthetic division and the rational root theorem (or
a calculator) to find the roots -1, 2, 2, and 2. The general solution is
Solution
Solve the characteristic equation:
r1 + 8r5 + 16r3 = 0
WARNING T, , , . f
I he general solution of an
r3(r4 + 8r2 + 16) = 0
nth order linear homogeneous
differential equation involves n r3(r2 + 4)2 = 0
arbitrary constants C\, C2, ..., C„.
r — 0 (multiplicity 3), ±2i (multiplicity 2)
Always check to make sure your
solution has the correct number of The roots (showing multiplicity) are 0, 0, 0, 2i, 2i, -2i, -2i. The general solution is
constants.
y = C1 + C2x + C2x~ + C4 cos 2x + C5 sin 2x + C(,x cos 2x + C2x sin 2x ■
14.2 Second-Order Homogeneous Linear Differential Equations 953
The constant k2 is called the damping constant to contrast it from the spring constant
Figure 14.6 A damped
k\. The characteristic equation is
spring system
mr2 + k2r + k\ = 0
—k2 ± y&2 — 4k im
r = 2m
The three cases that can occur correspond to different kinds of motion for the object
on the spring.
overdamping: k\ - <\k\m > 0 In this case, both roots are real and negative, and
the solution is of the form
where
ki 1 r-
r\ —-1-Jk\ — Ak\in and
2m 2m v
2 2 2
a. Overdamping k2 - Akxm >0 b. Critical damping k2 - 4kxm = 0 c. Underdamping k2 - 4kxm < 0
where
REDUCTION OF ORDER
Theorem 14.2 applies even when a, b, and c are functions of jc instead of constants. In
other words, the general solution of the equation
y = Ciyi + C2y2
where yi(x) and y2(x) are any two linearly independent solutions. Sometimes, we
can find one solution yj by observation and then obtain a second linearly independent
solution y2 by assuming that y2 = uyi, where v(x) is a twice differentiable function of
x. This procedure is called reduction of order because it involves solving the given
second-order equation by solving two related first-order equations. The basic ideas of
reduction of order are illustrated in the following example.
Show that y = x1 is a solution of the equation y" — —y' + 4-y = 0 for x > 0, then
x x2
find the general solution.
Solution
If y\ = x2, then y[ = 2x and y" = 2. We have
so yi is a solution.
Next, let y2 = v(x)x2 and compute
,2
(2v + 4a t/ + x2v")-(2xv + aV) + — (tur) = 0
x xl
2 ii
x~v" + xv' — 0
-1
v'
Integrate both sides of this equation:
In I v' I In A"
1
v
X
V In a
l 2
Thus, y2 vx"
C = a2 In a is a second solution. To show that the solutions y\
= A - a2 and
,2
y = a2 In a are linearly independent, we compute the Wronskian:
A2 A2lnA
W(a2, a2 In a) = a2 ^ 0 Since a > 0
2a a + 2a In a
y — Cia2 + C2A2 In a
21. y" - 10/ + 25y - 0; y(0) = 1, y'(0) = -1 39. The weight is pulled down 10 in. and is released with an initial
upward velocity of 6 ft/s.
22. y" + 6y' + 9v = 0; y(0) - 4; y'(0) = -3
40. The weight is raised 8 in. above the equilibrium point and the
23. y" - 12/ + 1 ly = 0; y(0) = 3, y'(0) - 11
compressed spring is then released.
24. y" + 4/ + 5v = 0; y(0) = -2, y'(0) = 1
41. The weight is raised 12 in. above the equilibrium point and
25. /" + 10/' + 25/ = 0; y(0) = 3, y'(0) = 2, y"(0) = -1
the compressed spring is then released.
26. y(4) - y'" = 0; y(0) = 3, y'(0) = 0, y"(0) = 3, y'"(0) - 4 42. The weight is pulled 6 in. below the equilibrium point and
In Problems 27-30, find the Wronskian, W, of the given set of func¬ is then released. The weight is connected to a dashpot that
tions and show that W ^ 0. imposes a damping force of magnitude 0.4 |u| at all times.
956 Chapter 14 Introduction to Differential Equations
43. The weight is pulled 12 in. below the equilibrium point and Mathematics is invented:
is then released. The weight is connected to a dashpot that
imposes a damping force of magnitude 0.08 |u| at all times. “There is an old Armenian saying, ‘He who lacks sense of the
past is condemned to live in the narrow darkness of his own gen¬
44. Characterize the solution of the second-order equation
eration.' Mathematics without history is mathematics stripped of
y" — lay' + (a2 — b)y = 0 b > 0 its greatness: for, like the other arts-and mathematics is one of the
supreme arts of civilization-it derives its grandeur from the fact
in terms of the functions coshkx and sinh kx, where k = -Jb. of being a human creation.”
-G. F. Simmons,
45. Verify that jq = e2*, y2 = xe2x, and y3 = x2e2x are linearly
Differential Equations with Applications and Historical
independent solutions of the third-order differential equation
Notes, Second Edition, McGraw-Hill, Inc., 1991, p. xix.
y'" - 6y" + 12y' - 8.y = 0.
46. Modeling Problem A 100-lb object is projected vertically up¬ Discuss whether the significant ideas in mathematics are dis¬
ward from the surface of the earth with initial velocity covered or invented.
150 ft/s. Q 48. The motion of a pendulum subject to frictional damping pro¬
a. Modeling the object’s motion with negligible air resis¬ portional to its velocity is modeled by the differential equation
tance, how long does it take for the object to return to „ d2e , do
earth? mL-— ffkL-E mg sin# = 0
dt2 dt
b. Change the model to assume air resistance equal to half the where L is the length of the pendulum, m is the mass of the
object’s velocity. Before making any computation, does bob at its end, and 9 is the angle the pendulum arm makes with
your intuition tell you the object takes less or more time the vertical (see Figure 14.8). Assume 0 is small, so sin# is
to return to earth this time than in part a? Now set up approximately equal to #.
and solve a differential equation to actually determine the
round-trip time. Were you right?
50. Consider the motion of an object of mass m on a spring with a. Use Newton’s second law to show that
spring constant k\ and damping constant k2 for the case where
d20
there is critical damping. mL—- + mg sin 9 = 0
dt2
a. Describe the motion of the object assuming that it begins
at rest: x'(0) = 0 at x(0) = x$. How is this different from Assume that air resistance and the mass of the rod are neg¬
the case where the object begins at x(0) = 0 with initial ligible.
velocity x'(0) = i>o? How do initial velocity and initial b. When the displacement is small (6 close to 0), sin 9 may be
displacement affect the motion? replaced by 9. In this case, solve the resulting differential
b. If x(0) = x'(0), what is the maximum displacement? equation
Show that the time at which the maximum displacement d29 g
- + -9 = 0
occurs is independent of the initial displacement x(0). As¬ dt2 L
sume x(0) # 0. How is the motion of the pendulum like the simple har¬
51. Suppose the characteristic equation of the differential equa¬ monic motion discussed in Problem 49?
tion ay" + by' + cy — 0 has complex conjugate roots 53. Path of a projectile with variable mass A rocket starts from rest
r\ — a + fii and r2 = a — pi. Show that yi = eax cos ySx and and moves vertically upward, along a straight line. Assume
y2 = eax sin fix are both solutions and are linearly indepen¬ the rocket and its fuel initially weigh w kilograms and that
dent. This is Theorem 14.3 for the case where b2 — Aac < 0. the fuel initially weighs Wf kilograms. Further assume that
the fuel is consumed at a constant rate of r kilograms per sec¬
52. Pendulum motion Suppose a ball of mass m is suspended at the ond (relative to the rocket). Finally, assume that gravitational
end of a rod of length L and is set in motion swinging back attraction is the only external force acting on the rocket,
and forth like a pendulum, as shown in Figure 14.9. a. If m(t) is the mass of the rocket and fuel at time t and s(t)
is the height above the ground at time t, it can be shown
that
m(t)s"{t) + m'(t)v0 + m{t)g = 0
where u0 is the velocity of the exhaust gas in relation to the
rocket. Express m(t) in terms of w, r, u0> and g, then inte¬
grate this differential equation to obtain the velocity s'(t).
Note that s'(0) = 0 because the rocket starts from rest.
b. Integrate the velocity s'(t) to obtain s(t).
c. At what time is all the fuel consumed?
d. How high is the rocket at the instant the fuel is consumed?
Figure 14.9 Pendulum motion
54. Journal Problem Canadian Mathematical Bulletin* ■ Solve
Let 9 be the angle between the rod and the vertical at time t, so the differential equation
that the displacement of the ball from the equilibrium position
is s = L6 and the acceleration of the ball’s motion is xAy" — (x3 + 2 axy)y' + 4ay2 = 0
d2s d29
dt2 = LJt2 ^Problem 331 by Murray S. Klamkin, Vol. 26, 1983, p. 126.
-•
Next we will see how to solve a nonhomogeneous second-order linear equation of the
general form ay" 4- by' + cy = F(x), where F(x) ^ 0. The key to our results is the
following theorem.
958 Chapter 14 Introduction to Differential Equations
y = yh + yP
a(y - ypf + b(y - yp)' + c(y - yp) = a(y" - y") + b(y' - y'p) + c(y - yp)
^ What This Says We can obtain the general solution of the nonhomoge¬
neous equation ay" + by' + cy = F(x) by finding the general solution y/, of
the related homogeneous equation ay" + by' + cy — 0 and just one particular
solution yp of the given nonhomogeneous equation.
We can use the methods of the preceding section to find the general solution of the
related homogeneous equation. We now develop two methods for finding particular
solutions of the nonhomogeneous equation.
Solution
The right side F(x) = 2x2 — 4x is a quadratic polynomial. Because derivatives of
a polynomial are polynomials of lower degree, it seems reasonable to consider a trial
solution that is also a polynomial of degree 2. That is, we “guess” that this equation
has a particular solution yp of the general form y p = A\x2 + Ai_x + A3. To find the
constants A\, A2, and A3, calculate
y'p — 2A\x + A2 and y"=2A|
Substitute the values for yp, y' and yp into the given equation:
Comment: Notice that even though the constant term is zero in the polynomial func¬
tion F(x), we cannot assume that y = A\x2 + A2X is a suitable trial solution. In
general, all terms of the same or lower degree that could possibly lead to the given
right-side function F(x) must be included in the trial solution.
Solution
Because the trial solution y is to be “like” the right-side function F(x) = sinx, it
seems that we should choose the trial solution to be yp — A1 sin x, but a sine function
can have either a sine or a cosine in its derivatives, depending on how many derivatives
are taken. Thus, to account for the sinx term, it is necessary to have both sinx and
cosx in the trial solution, so we set
y = A1 sin x + A2 cos x
Differentiating, we find
~y'p = A\ cosx — A2sinx and y" = — A\sinx — A2cosx
with the solution A\ = — ~,A2 = — Thus, the particular solution of the non-
homogeneous equation is yp — — -j^ sin jc — -j^cosx, and the general solution is
y = yh + yP = C\ex + C2e~lx - ^ sin x - cosx. ■
is
,-,
Solution for F — 2x2 — 4x
2 13. 1
n — —x + x - sinx- cos x
p 2 10 10
Solution for G = sin x
Solution
At first glance (looking at F{x) — 4e~2x), it may seem that the trial solution should
be y — Ae~2x. However, yi = e~2x is a solution of y" + yj — 2yi = 0, so it cannot
possibly also satisfy the given nonhomogeneous equation.
To deal with this situation, multiply the usual trial solution by x and consider the
trial solution yp — Axe~lx. Differentiating, we find
—3Ae~lx — Ae~2x
y = yh + yP — c,ex + c2e 2x
To solve ay" + by' + cy = F(x) when F(x) is one of the forms in the preceding
list:
ay" + by' + cy — 0
Solution
a. The related homogeneous equation y"—4y/+4y = 0 has the characteristic equation
r2 — 4r + 4 = 0, which has the root 2 of multiplicity two. Thus, the general
homogeneous solution is
yh = C\e2x + C2xe2x
The part of the trial solution for the nonhomogeneous equation that corresponds
to 3x2 is Ao + Aix + A2X2 and the part that corresponds to 4e~2x is Be~2x. Since
neither part includes terms in yh, we apply the principle of superposition to conclude
that
y p = Ao + A\X + A2X2 + Be
yh = C\e2x + C2xe~x
The expected trial solution for 5xe2x would be (Ao + A\x)elx, but this expression
is contained in y/,. If we multiply by x, part of (A0 + A\x)xe2x is still contained in
yh, so we multiply by x again to obtain
yp = (A0 + A\x)x2e2x
962 Chapter 14 Introduction to Differential Equations
Ordinarily, the trial solution for the nonhomogeneous equation would be of the form
e~x[A cos 2 x + B sin 2.x], but part of this expression is in v/,. Therefore, we multiply
by x to obtain the trial solution
VARIATION OF PARAMETERS
The method of undetermined coefficients applies only when the coefficients a, b, and
c are constant in the nonhomogeneous linear equation ay" + by' + cy = F(x) and the
function F(x) has the same general form as a solution of a second-order homogeneous
linear differential equation with constant coefficients. Even though many important
applications are modeled by differential equations of this type, there are other situations
that require a more general procedure.
Our next goal is to examine a method of J. L. Lagrange (see Historical Quest,
Problem 55 in Section 11.8) called variation of parameters, which can be used to
find a particular solution of any nonhomogeneous equation
In practice, if P(x) and Q(x) are not both constants, these may be difficult to find, but
once we have them, we assume there is a solution of the nonhomogeneous equation of
the form
yP = uyi + vy2
«'yi + u'y2 = o
Remember, we need to find only one particular solution yp, and if imposing this side
condition makes it easier to find such a yp, so much the better! With the side condition,
we have
yp = uyl + u y, + vy2 + v y2
Next, we substitute our expressions for y'p and y" into the given differential equation:
u[y'i + P{x)y\ + Q{x)y\] + i>[ y'{ + P{x)y2 + Q(x)y2] + u'y[ + v'y'2 = F(x)
! ! | II I-' / \
m Vj + v y2 = F(x)
u'y i + v'y2 = 0
u'y\ + v'y2 = F(x)
yp = uyx + vy2
we obtain a particular solution of the given differential equation. Here is a summary of
the procedure we have described.
Variation of Parameters
WARNING tL rr. .
the coerhcient To find the general solution of y" + P(x)y' + Q(x)y — F(x),
of y" must be 1 for this method.
Step 1. Find the general solution, y/, = Cxy\ + C2y2 to the related homoge¬
neous equation, y" + P(x)y' + Q(x)y = 0.
Step 2. Set yp — uy\ + vy2 and substitute into the formulas:
, —y2F(x) yiF(x)
u = -7-7 v = -7-7
y\y2-y2yx yiy2-yiyx
Step 3. Integrate u and v' to find u and v.
Step 4. A particular solution is yp = uyx + vy2, and the general solution is
y = yh + yp.
Solution
Notice that this problem cannot be solved by the method of undetermined coefficients,
because the right-side function F(x) = tan 2* is not one of the forms for which the
procedure applies.
To apply variation of parameters, begin by solving the related homogeneous equa¬
tion y" + 4y = 0. The characteristic equation is r2 + 4 = 0 with roots r = ±2i. These
complex roots have a = 0 and — 2 so that the general solution is
y = e°[Ci cos2x + C2 sin2x] = C\ cos2x + C2 sin2x
964 Chapter 14 Introduction to Differential Equations
This means yi(x) = cos2x and y2(x) = sin 2x. Set yp = uy\ + vy2, where
, -yiFix) — sin2xtan2x sin2 2x
u -
yi y'2 — yiy[ 2 cos 2x cos 2x + 2 sin 2x sin 2x 2cos2x
and
y\F{x) cos2xtan2x
v — = i sin 2x
y\y'2 ~ yiy[ 2 cos 2x cos 2x + 2 sin 2x sin 2x
Integrating, we obtain
f sin2 2x C ,
u(x)
= j-^2xdX vU) = j~2sin 2x dx
= ~i Jf~
2
— cos2 2x
cos 2x
= — | cos 2x
sin 2x
h In |sec 2x + tan2x|
y" + 6/ + 9y = Ffx)
b. Use variation of parameters to find a particular solution yp order linear differential equation
of the given nonhomogeneous equation. Then use Theo¬
rem 14.4 to write the general solution. Hint: Divide the y" + P(x)y' + Q(x)y = Ffx)
differential equation by x2.
and let y2 satisfy
53. Repeat Problem 52 for the equation
y" + P(x)y' + Q(x)y = F2(x)
(1 + x2)y" + 2 xy' — 2 y = tan”1 x
Show that y! + y2 satisfies the differential equation
54. The principle of superposition Let yt be a solution to the second- y" + P(x)y' + Q(x)y = Fx (x) + F2(x)
-9
Proficiency Examination Eft) = 50cos 100r in the circuit, what is the current I(t) at
time tl Assume that when t = 0, there is no charge and no
CONCEPT PROBLEMS current flowing.
1. What is a separable differential equation?
R = 25
2. What is a homogeneous differential equation?
3. What is the form of a first-order linear differential equation?
E
4. What is an exact differential equation?
5. Describe Euler’s method.
6. Define what it means for a set of functions to be linearly inde¬ L = 0.1
pendent.
7. What is the Wronskian, and how is it used to test for linear
independence? Supplementary Problems
8. a. What is the characteristic equation of ay" T by' + cy — 0? 1. Consider the differential equation
13. — dx — dy = 0
r y3 a. Use the slope field to sketch the particular solution that
dy 2x + y passes through (1,2).
14. 15. xy dy = (x2 — y2) dx b. Use Euler’s method to find a solution for the initial value
dx 3x
16. x2 dy — (x2 + y2) dx — 0 problem in part a for 1 < x < 2 and h = 0.2.
17. y" 4- 2 y' + 2 y = sin x Find the general solution of the given first-order differential equa¬
tions in Problems 3-20.
18. (3x2e~y 4- y-2 + 2xy-3) dx
„ dy —Ax
+ (—x2e~y — 2 xy-3 — 3x2y”4) dy = 0 3. — = ——
dx y3
19. A spring with spring constant k = 30 lb/ft hangs in a vertical
position with its upper end fixed and an 8-lb object attached to 4. x dy + (3y — xex2) dx = 0
its lower end. The object is pulled down 4 in. from the equi¬
5. y2 dy — (x2 + —) dx = 0
librium position of the spring and is then released. Find the x
displacement x(f) of the object, assuming that air resistance is x (3 y \
6. - dy -f ( —-5 J dx — 0
0.8u, where v(t) is the velocity of the object at time t.
20. An RLC circuit has inductance L = 0.1 henry, resistance
P = 25 ohms, and capacitance C = 200 microfarads (i.e.,
200 x 10”6 farad). If there is a variable voltage source of 8. dy — (3y 4- ex 4- cosx) dx
Chapter 14 Review 967
9. dy — (5y + e5x sinx) dx = 39. Use reduction of order to find the general solution of the Leg¬
10. x2 dy = (x2 — y2) dx endre equation
11. (1 — xev) dy — ey dx (1 -x2)y"-2xy' + 2y = 0
12. xy dx + (1 + x2) dy = 0 given that yj = x is one solution.
13. x dy — (y + Jx1 — y1) dx 40. Find functions x = x{t) and y = y(r) that satisfy the linear
14. 2xyex~ dx — ex~ dy 0 system
15. (y — xy) dx + x3 dy — 0 dx dy
dy
—- =x + y
dt
— =x
dt
y
16. = 2y cot 2x + 3 esc 2x
dx Sketch the solution curve (x(r), y(f)) that contains the point
1 4, ,2 1 dy y'(t)
17. 4x3y3 + — ) dx + ( 3x4y dy — 0 (0, 2). Hint: Note that — = ;-.
dx x'(f)
18. (—y sin xy + 2x) dx + (3y2 — x sin xy) dy = 0 41. Solve the second-order differential equation xy" + 2y' — x by
19. sin y dx + (ex + e-*) sin y dy = 0 setting p — y' to convert it to a first-order equation in p.
dy 42. Find an equation for a curve for which the radius of curvature
20. — + 2y cotx + sin 2x = 0 is proportional to the slope of the tangent line at each point
dx
P(x, y).
In Problems 21-28, find
43. Solve the differential equation
a. the general solution y* of the related homogeneous equation
dx x
b. a particular solution yp of the nonhomogeneous equation --= ye-
dy y
21 y" - 9y = 1 + x 22. y" - 2y' + y = ex
23. y" - 5y + 6y = x V 24. y" + 2y' + y = sinh x 44. A ship weighing 64,000 tons [mass = (64.000)(2,000)/32
slugs] starts from rest and is driven by the constant thrust of its
25. y" - 3y' + 2y = x V
propellers. Suppose the propellers supply 250,000 lb of thrust
26. 14y" + 29y' — 15y = cosx and the resistance of the water is 12,000u lb, where v(t) is the
27. y" + y = tan2 x 28. y" + y = sec x velocity of the ship. Set up and solve a differential equation
d2y for v(t).
29. Solve = e~3;t subject to the initial conditions
dx* 45. A 10-ft uniform chain (see Figure 14.13) of mass m is hang¬
y(0) = y'(0) = 0. ing over a peg so that 3 ft are on one side and 7 ft are on the
30. Use the substitution p — dy/dx to solve the differential equa¬ other. Set up and solve a differential equation to find how long
tion it takes for the chain to slide off the peg. Neglect the friction.
d2y
/2y /dv\3
dy dy
+
dx2 dx dx
dx dy . . _ dz
= —2x — = -3y + 2x — = 3y
d7 dt dt
Suppose the load on the beam is W pounds per foot and one 50. Find the general solution of the Euler equation
end of the beam is at the origin, and the other end of the beam
is at (L, 0). Then the deflection y of the beam at point x is x2y" + Ixy' + 9y = V*
modeled by the differential equation
Hint: Use the result of Problem 49 along with variation of
d2y 1 ,
El —y— = WLx - -Wx2 parameters.
dx2 2
51. In certain biological studies, it is important to analyze
for positive constants E and 7. predator-prey relationships. Suppose x{t) is the prey popu¬
a. Solve this equation to obtain the deflection y(x). Note that lation and y(t) is the predator population at time t. Then these
y(0) = y(L) = 0. populations change at rates
b. What is the maximum deflection in the beam?
48. Consider the Clairaut equation dx dy
- aux - aX2y = a2xx - a22y
dt dt
y = xy' + f(y')
where axx is the natural growth rate of the prey, aX2 is the pre¬
a. Differentiate both sides of this equation to obtain
dation rate, a2x measures the food supply of the predators, and
[x + /'(/)]/ = 0 a22 is the death rate of the predators. Outline a procedure for
solving this system.
Because one of the two factors in the product on the left
must be 0, we have two cases to consider. What is the so¬ 52. Consider the almost homogeneous differential equation
lution if y" — 0? This is the general solution.
dy / ax + by + c
b. What is the solution if
dx \ rx + sy + t
^ + /V) = 0
with as — br (recall Problem 60, Section 14.1).
(This solution is called the singular solution.) Hint: Use
the parameterization y' = t. ax + by rx + sy
a. Let u — --. Show that u —-and
c. Find the general and singular solutions for the Clairaut a
du a du
equation - 1
dx b dx
y = xy' + %/4 + (y')2 Verify that by making the change of variable suggested in
part a, you can rewrite the given differential equation in the
49. A differential equation of the form
separable form
x2y" + Axy' + By = F{x)
du
with x 0 is called an Euler equation. To find a general — dx
solution of the related homogeneous equation i+if(°jt+±
a \ ru T t
x2y" + Axy' + By = 0
c. Use the procedure outlined in parts a and b to solve the
we assume that there are solutions of the form y — xm.
almost homogeneous differential equation
a. Show that m must satisfy
m2 + (A — l)m + B = 0 dy 2x + y — 3
dx 4x + 2y + 5
This is the characteristic equation for the Euler equation.
b. Distinct real roots. Characterize the solutions of the re¬
53. Putnam Examination Problem Find all solutions of the equation
lated homogeneous equation in the case where
c. Repeated real roots. Characterize the solutions of the re¬ that pass through the point x = 1, y = 1.
lated homogeneous equation in the case where
54. Putnam Examination Problem A coast artillery gun can fire at any
(A - l)2 = 4B angle of elevation between 0° and 90° in a fixed vertical plane.
If air resistance is neglected and the muzzle velocity is con¬
d. Complex conjugate roots. Suppose that
stant (u(0) = u0), determine the set 77 of points in the plane
(A - l)2 < 47? and above the horizontal that can be hit.
55. Putnam Examination Problem Show that
and that a ± fli are the roots of the characteristic equa¬
tion. Verify that yx = xa cos(/6 In |x|) is one solution. Use
2 3 2-4 5 2- 4-6 sin 1 x
reduction of order to find a second solution y2, then char¬ *' + ••• =
X+3X+ 3T5 + 3- 5-7
acterize the general solution. 7f=^
Group Research Project*
H
will submit a single written report.
A 50 gal/h 60 gal/h
B 60 gal/h 75 gal/h
Unfortunately, spring B has become contaminated with salt and is now 10% salt,
G
which means that 10% of a gallon of water from spring B is salt. The yellow perch
F
will start to die if the concentration of salt in the pond rises to 1%. Assume that the
E salt will not evaporate but will mix thoroughly with the water in the pond. There was
D no salt in the pond before the contamination of spring B. Your group has been called
C upon by the Happy Valley Bureau of Fisheries to try to save the perch.
B Your paper should include the number of gallons of water in the pond when the
A water level is exactly even with the top of the spillover dam. The overflow goes over
1 2 3 4 5
the dam into Bubbling Brook. The following table gives a series of measurements of
Figure 14.15 Happy Valley the depth of the pond at the indicated points when the water level was exactly even
Pond is fed by two springs with the top of the spillover dam.
A and B
DEPTH OF HAPPY VALLEY POND (Location/depth)
F5/2 ft G3/1 ft
Let 7 = 0 hours correspond to the time when spring B became contaminated. Assume
it is the dry season and that at time t — 0 the water level of the pond was exactly even
with the top of the spillover dam. Write a differential equation for the amount of salt in
the pond after t hours. Draw a graph of the amount of salt in the pond versus time for
AMONG all the mathematical disciplines the next 3 mo. How much salt will there be in the pond in the long run, and do the fish
the theory of differential equations is the die? If so, when do they start to die? It is very difficult to find where the contamination
most important .... It furnishes the ex¬ of spring B originates, so the Happy Valley Bureau of Fisheries proposed to flush the
planation of all those elementary manifes¬ pond by running 100 gal of pure water per hour through the pond. Your report should
tations of nature which involve time .... include an analysis of this plan and any modifications or improvements that could help
-Sophus Lie save the perch.
Leipziger Berichte, 47 (1895).
This group project is courtesy of Diane Schwartz from Ithaca College, New York.
969
Appendices
lim f(x) = L
X-+C
is read “the limit of f(x) as x approaches c is L" and means that the function values
/(x) can be made arbitrarily close to L by choosing * sufficiently close to c but not
equal to c.
This informal definition was valuable because it gave you an intuitive feeling for
the limit of a function and allowed you to develop a working knowledge of this funda¬
mental concept. For theoretical work, however, this definition will not suffice, because
it gives no precise, quantifiable meaning to the terms “arbitrarily close to L” and “suf¬
ficiently close to c.” The following definition, derived from the work of Cauchy and
Weierstrass, gives precision to the limit definition and was also first stated in Section
2.1.
means that for each number e > 0, there corresponds a number 8 > 0 with the
property that
A-l
THE BELIEVER/DOUBTER FORMAT
The limit process can be thought of as a “contest” between a “believer” who claims
that lim /(*) = L and a “doubter” who disputes this claim. The contest begins with
X~*C
the doubter choosing a positive number e and the believer countering with a positive
number 8.
When will the believer win the argument and when will the doubter win? As you
can see from Figure A.l, if the believer has the “correct limit” L, then no matter how
the doubter chooses e, the believer can find a 8 so that the graph of y = /(*) for
c-5<*<c+S will lie inside the heavily shaded rectangle in Figure A. 1 (with the
possible exception of x = c); that is, inside the rectangular region
On the other hand, if the believer tries to defend an incorrect value w (for “wrong”)
as the limit, then it will be possible for the doubter to choose an e so that at least part
of the curve y = fix) will lie outside the heavily shaded rectangle R, no matter what
value of 8 is chosen by the believer (see Figure A.2).
To avoid an endless chain of €-8 challenges, the believer usually tries to settle the
issue by producing a formula relating the choice of 8 to the doubter’s e that will satisfy
the requirement no matter what € is chosen. The believer/doubter format is used in
Example 1 to verify a given limit statement (believer “wins”) and in Example 2 to
show that a certain limit does not exist (doubter “wins”).
Solution
Let f{x) = 2* + 1. To verify the given limit statement, we begin by having the
doubter choose a positive number €. To help the believer, choose 8 > 0, and note the
computation in the following box.
The believer wants \ fix) — L\ < e, so we see that the believer should choose 28 = €,
or 8 = |
With the information shown in this box, we can now make the following argument,
which uses the formal definition of limit:
Appendix A A-3
2x2 — 3x — 2
Determine whether lim---= 6.
Figure A.3 Verifying lim(2x + 1) = 5: x—>2 X — 2
x—>2
p 6
Given e > 0, choose 8 = -. Solution
2*2 - 3* (2x + l)(x — 2)
= 2x + 1, x 2. Once again, the
Let/W x — 2 x-2
doubter will choose a positive number e and the believer must respond with a 8. As
before, the believer does some preliminary work with /(x) — L:
2x2 — 3x — 2 ^ 2x2 — 3x — 2 — 6x + 12
o —
x — 2 x —2
2x2 — 9x ~F 10
x —2
(2x — 5)(x — 2)
x —2
= 2x - 5|
( s\ r / 8\ i
— —6 = \-8 - II > 1
so |/(x) - 6| < e is not satisfied for all 0 < |x - 2| < <5 (in particular, not for
2x2 — 3x — 2
x = 2 - |). Thus, lim-r-= 6 is false.
x —2
The believer/doubter format is a useful device for dramatizing the way certain
choices are made in epsilon-delta arguments, but it is customary to be less “chatty” in
formal mathematical proofs.
Solution
x2 — 2x + 2
Let f(x) - ■. We have
x —4
x2 — 2x + 2
\f (x) — L\ -(-1)
x-4
x2 — x — 2
x — 4
x + 1
\x — 2\
x-4
This must be less than the given e
whenever x is near 2, but not equal to 2.
X I J
Certainly \x — 2| is small if x is near 2, and the factor - is not large (it is close
x -4
to j)- Note that if \x — 2\ is small it is reasonable to assume
x + l 5 5
Let g(x) =-7 = 1 +-g'(x) < 0 on (1,3). Thus g(x) is de-
x —4 x-4 (x — 4)2
creasing on (1, 3) and
. x + 1 2
-4 < -- < 4 Because-<4
x-4 3
and
x T 1
< 4
x-4
x -f 1
and < 4
x-4
then
x T 1
\f(x)-L\ = \x 2| <-(4) = f
x-4
Thus, we have only to take <5 to be the smaller of the two numbers 1 and | in order to
guarantee that
|/(x) - L \<€
That is, given € > 0, choose 8 to be the smaller of the numbers 1 and We write this
c- A c 1- X~ ~2x + 2
Figure A.5 hm-
x-*2 X — 4
as <5 = min (l, |). We can confirm this result by looking at the graph of y = fix) in
Figure A.5). m
Solution
Let f(x) — — and L be any number. Suppose that lim f(x) = L. Look at the graph
x *->-0
of /, as shown in Figure A.6. It would seem that no matter what value of € is chosen, it
would be impossible to find a corresponding 5. Consider the absolute value expression
required by the definition of limit: If
1
I/O) - L\ < e, or, for this example, L < €
x
then
1 and
Fiqure A.6 lim -
.v->0 X
1
L — € < — < L +€
X
1
<01 + 1
x
1
\x\ >
I L\ + 1
1
and thus x is not close to zero which proves (since L was chosen arbitrarily) that lim -
does not exist. In other words, since |x| can be chosen so that 0 < |x| < —, then
I I I 1
Next, we will prove two theoretical results using the formal definition of the limit.
These two theorems are useful tools in the development of calculus. The first states
that the points on a graph that are on or above the x-axis cannot possibly “tend toward"
a point below the axis, as shown in Figure A.7.
WARNING , . . , r
It may seem reasonable to conjecture that if fix) > 0 throughout an open
interval containing c, then lim fix) > 0. This is not necessarily true, and the most that can be
X —> C
said in this situation is that lim fix) > 0, if the limit exists. For example, if
x2 for x 7^0
fix)
1 for x — 0
Useful information about the limit of a given function / can often be obtained
by examining other functions that bound / from above and below. For example, in
Section 2.1 we discovered
sin x
lim-= 1
x-*0 X
by using a table. We justified the limit statement in Section 2.2 by using a geometric
argument to show that
sinx
cosx < - < 1
x
for all x near 0 and then noting that since cos x and 1 both tend toward 1 as x ap¬
proaches 0, the function
sinx
which is “squeezed” between them, must converge to 1 as well. Theorem A.2 provides
the theoretical basis for this method of proof.
Proof Let c > 0 be given. Since lim gix) = L and lim hix) = L, there are positive
x-*c x->c
numbers and S2 such that
|g(x) — L\ < € and |/i(x) - L| < e
whenever
0 < |x — c| < <$i and 0 < |x — c| < S2
respectively. Let S be the smaller of the numbers and <52. Then, if x is a number that
satisfies 0 < |x — c| < 8, we have
-6 < gix) - L < fix) - L < hix) - L <€
and it follows that |/(x) - L\ <€. Thus, lim fix) = L, as claimed. □
Appendix A A-7
The geometric interpretation of the squeeze rule is shown in Figure A.8. Notice
that since g(x) < fix) < h(x), the graph of / is “squeezed” between those of g and
h in the neighborhood of c. Thus, if the bounding graphs converge to a common point
P as x approaches c, then the graph of / must also converge to P as well.
disprove the given limit statement. rem (Theorem A.l) to the function /i(x) = fix) — g(x).
7. lim(x + 3) = 5 8. lim(3? — 1) = 0 Problems 20-23 lead to a proof of the product rule for limits.
x->2 r->0
9. lim (3x + 7) = 1 10. lim(2x — 5) = — 3 20. If lim fix) = L, show that lim |/(x)| = |L|. Hint: Note that
x->-2 *-»-i
11/001 - IZ.II <\fix)-LXCC
1 1
11. lim(x2 + 2) = 6 12. lim — = - 21. If lim fix) = L, L yk 0, show that there exists a 8 > 0 such
i->2 x—>2 X 2 X—>C
1 that
sin — if x ^0
13. Prove that fix) — x
0 if x = 0 \ \L\ < |/(x)| < \\L\
is not continuous at x = 0. Hint: To show that lim /(x) 0,
x-+0 for all x for which 0 < |x — c\ < 8. Hint: Use e = j \L\ in
choose e = 0.5 and note that for any <5 > 0, there exists an x
Problem 20.
2
of the form x = with n a natural number for which 22. If lim fix) — L and L ^ 0, show that lim[/(x)]2 = L2 by
n(2n + 1) x->c x^-c
0 < |x| < 8. completing these steps:
Q In Problems 14-19, construct a formal c-8 proof to show that the a. Use Problem 21 to show that there exists a Si > 0 so that
given limit statement is valid for any number c.
14. If lim /(x) exists and k is a constant, then |/(x) + L\ < f \L\
15. If lim /(x) and lim g(x) both exist, then b. Given e > 0, show that there exists a 82 > 0 such that
lim[/(x) - g(x)] = lim /(x) - lim g(x).
JC—►C X —>c x—*c
|[/(x)]2 - L2\ < f |L|e
16. If lim f{x) and lim g(x) both exist and a and b are constants,
then lim[a/(x) + bg(x)] = a lim fix) + b limg(x). whenever 0 < |x — c\ < S2.
A-8 Appendices
c. Complete the proof that 25. Show that if / is continuous at L and limg(x) = L, then,
X~*C
lim /[g(x)] = f(L) by completing the following steps:
lim[/(x)]2 = L2
X—>C
a. Explain why there exists a ^ > 0 such that
\f(w) — f(L)\ < e whenever \w — L\ < <5i.
23. Prove the product rule for limits: If lim /(x) = L and
X—>C b. Complete the proof by setting w — g(x) and using part a.
lim g(x) = M, then lim /(x)g(x) = LM. Hint: Use the re-
x—>c x—*c 26. For linear functions, the relation between e and S is clear;
suit of Problem 22 along with the identity but for complicated functions, it is not. However, the cor¬
rect graph can illustrate this relationship. For the following
fg = + g)2 ~ (f ~ g)2] function, illustrate not only that / is continuous at x = 3, but
determine graphically how you would pick 8 to accommodate
24. Show that if / is continuous at c and /(c) > 0, then f(x)> 0 a given e. (For example, 8 = A'“1c.)
throughout an open interval containing c. Hint: Note that
lim /(x) = /(c) and use e = 4/(c) in the definition of limit. , x4 — 2x3 + 3x2 — 5x + 2
- x—>c z
-m
Am
lim g{Au) + lim — lim —
Ax->0 Ax->0 du Ax—>0 Ax
Appendix B A-9
df~ du
Since g is continuous at t = 0
0 + 17,\dI
df du
□
du dx
Let / and g be functions that are continuous on the closed interval [a, b] and differen¬
tiable on the open interval ia, b). If gib) ^ gia) and g'(x) ^ 0 on (a, b), then
fib) - fja) _ /'(c)
g((?) - gia) g'(c)
for at least one number c between a and b.
Proof We begin by defining a special function, just as in the proof of the MVT as
presented in Section 4.2. Specifically, let
fib) — fia)
Fix) = fix) - fia)-—-—[g(x) - g(fl)]
gib) - g{a)
for all x in the closed interval \a,b\. In the proof of the MVT in Section 4.2 we show
that F satisfies the hypotheses of Rolle’s theorem, which means that F\c) = 0 for at
least one number c in ia, b). For this number c, we have
fib) - fia) ,, x
0 = Ffc) = fie) - 777-T^g (c)
gib) - gia)
and the result follows from this equation. □
L’HOPITAL’S RULE*(Chapter 4)
For any number a, let / and g be functions that are differentiable on an open interval
fix) .
ia, b), where g'(x) 0. If lim fix) = 0, lim gix) = 0, and lim ^7— exists,
g'ix)
then
,. fix) ,. fix)
lim -= lim -
gix) x^a+ gfx)
Fiw) _ FfO
Because Fia) = Gia) - 0
Giw) ~ G'it)
fiw) _ fit)
Because Fix) = fix) and Gix) = g(x)
giw) g'it)
This is a special case of l’Hopital’s rule. The other cases can be found in most advanced calculus textbooks.
A-10 Appendices
/ O) fit)
lim = lim
w—ya+ g(w) w->a^ g'it)
f(w) fit)
lim = lim because t is “trapped" between a and w
w—>a+ gin>) t-*a+ g'it)
fix) fix)
lim - lim □
x—>a+ gix) x—>a+ g'ix)
Suppose ak > 0 and bk > 0 for all sufficiently large k and that
ak
lim — — L where L is finite and positive (0 < L < oo)
k—^oo bk
Then Hak and Hbk either both converge or both diverge.
Proof Assume that lim ~ = L, where L > 0. Using e = - in the definition of the
k-* oo bk 2
limit of a sequence, we see that there exists a number N so that
Uk
- L < whenever k > N
L ak L
—— < -- L <
2 bk 2
L ak 3L
— < — < —
2 bk 2
L 3L
-bk < ak < —bk bk> 0
This is true for all k > N. Now we can complete the proof by using the direct compar-
ison test. Suppose Tibk converges. Then the series —-bk also converges, and the
inequality
3L
ak < —bk
tells us that the series Ea* must also converge since it is dominated by a convergent
series.
Similarly, if Yibk diverges, the inequality
tells us that Eak dominates the divergent series ^ —bk, and it follows that Hak also
diverges.
Thus T,ak and Ebk either both converge or both diverge. □
If / and all its derivatives exist in an open interval I containing c, then for each x in /
Proof We will prove Taylor’s theorem by showing that if / and its first n + 1 deriva¬
tives are defined in an open interval / containing c, then for each fixed x in /
fix)
/(c)+
fic)
1! (*
c) + —(x - cY Hr +/^_
n\
+ cr f ^-cr+1
(n + 1)!
where z is some number between x and c. In our proof, we will apply Cauchy’s
generalized mean value theorem to the auxiliary functions F and G defined tor all t in
I as follows:
fit) fMit)
Fit) = fix) - fit) - ix-t)- ix - t)n
1! n\
(x - t)n+l
G{t) =
in + 1)!
Note that Fix) = G(x) = 0, and thus Cauchy’s generalized mean value theorem tells
us
F'iz) _ Fjx) - Fjc) _ F(£)
Gfz) ~ G(x) - G(c) G(c)
for some number z between x and c. Rearranging the sides of this equation and finding
the derivatives gives
Fjc) F'iz)
G(c) G'iz)
-f(n+l)iz)
-Z)n
-ix - z)n
n\
= f(n+X)iz)
Fic) = f(n+l\z)G(c)
fic) (x — c)n+l
fix) = fic) + ^ix-c) + + t^-ix-c)n + fn+l)iz)
ni in -(- 1)!
□
^tC*! y) = fx{x\, y) — fxixo, yo) and 62(x, y) =/^(xo, yi) —/^(xo, yo)
Then since xi is between x and xo, and yi is between y and yo, and the partial deriva¬
tives fx and fy are continuous at (x0, yo), we have
flvlS„ei(i-y)=
(X, y)^> U0A0) (x,y)-+(x0,y0)
Uxixx,y)~ fxixo, yo)} = 0
,Mx'y) = ( 1™ [/r(^o,yi)-/y(xo,yo)] = 0
d( x, ,y)
fix, y)dydx figiu, v), hiu, u)] dv du
3(m, v)
absolute value of Jacobian
djx, y)
A = dv du
3 iu, v)
Figure B.2
B(u + Au, v), C(u, v + Av), and D(u + Au,v + Av), the set R will be the interior
of a curvilinear rectangle with vertices
Note that the curved side of R joining A to B may be approximated by the secant
vector
dx 3y
AB = — (a, u) Am i+ — (b, u) Am
6u au
for some numbers a, b between u and u + Au. If Aw is very small, a and b are
approximately the same as u, and we can approximate AB by the vector
dx 3y
a = — Am i + — Am
3m 3m
where the partials are evaluated at the point (u,v). Similarly, the curved side of R
joining A and C may be approximated by the vector
dx "3y .
— Au i+ — Av
3u dv
The area of the curvilinear rectangle R is approximately the same as that of the paral¬
lelogram determined by a and /?; that is,
dx 3y .
— Am — Am
\a x fi\\ = 3m 3m
dx dy
— Au — Av
dv dv
i j k
dx
* o AvAu
3m 3m
dx
^ 0
dv dv
A-14 Appendices
dx dy dy dx
AvAu
du dv du dv
3(*. y)
AvAu
d(u, u)
By adding the contributions of all cells in the partition of D, we can approximate the
area of D as follows:
d(x,y)
AvAu
=£ d(u, v)
Finally, using a limit to “smooth out” the approximation, we find
A = JJ
D
dy dx = lim ^
d(x, y)
d(u, v)
AvAu
= // d(x, y)
d(u, v)
dv du
D*
Let S be an oriented surface with unit normal vector field N, and assume that S is
bounded by a piecewise smooth Jordan curve C whose orientation is compatible with
that of S. If F is a vector field that is continuously differentiable on S, then
lFdR=II (curl F • N) dS
Proof The general proof requires the methods of advanced calculus. However, a
proof for the case where S is a graph and F, S and C are “well behaved” can be given.
Let S be given by z = g(x, y) where (x, y) is in a region D of the xy-plane. Assume
g has continuous second-order partial derivatives. Let Ci be the projection of C in the
xy-plane, as shown in Figure B.3. Also let F(x, y, z) = f(x, y, z)i + g(x, y, z)j +
Figure B.3 h(x,y,z)k where the partial derivatives of /, g, and h are continuous.
We will evaluate each side of Stokes’s theorem separately and show that the results
for each are the same. If x = x(t), y = y(t), and z = z(t) for a < t < b and
R(0 = x(t)i + y(Oj + z(t)k, then
"b T dx dv d7~
L,dn-L'[/i+’‘i+‘ I' dt
dx dy (dz dx dz dy
dt Chain rule
f^+SY,+h\TXli + TyJ,
dt
-a<
■/Jl
3
-//[ * + hS)-ly{f + h^
D
dx
dA Green's theorem
3g dg dz dh dz dh dz dz d2z \
-II D
,dx dz dx dx dy + dz dx dy + dydx)
' df ^ 3/ 3z
.3y dz dy
dh dz
dy dx
dh dz dz
dz dy dx +
d2z \
dydx J dA
dh dz dg dz df dz dh dz dg df
-II D
dy dx dz dx dz dy dx dy + dx dy _
dA
Appendix B A-15
-in-
D
3h dz
3y dx + dz dx
dg dz df dz
dz dy
3h dz
dx dy
dg
dx
df
dy _
dA
□
/cF'rfR = // (curlF-N )dS
F-N dS = divF dV
Proof Let F(jc, y, z) - fix, y, z)i + g(x, y, z) j + h(x, y, z) k. If we state the diver¬
gence theorem using this notation for F, we have
N) dS + jf
c
*(J.N )dS + // Mk-N)^ = ///^dy + ///|dV + ///|dV
5 D D D
ffK,Nm = ffffxdv
S D
//*(k.N)« = /// 3h
dz
dV
D
Since the proof of each of these is virtually identical, we will show the verification
for the last equation; the other two can be done in a similar fashion. We will evaluate
this third integral by separately evaluating the left and right sides to show they are the
same.
A-16 Appendices
We will restrict our proof to a standard region; that is, a solid bounded above by
a surface ST with equation z = u(x,y) and below by a surface SB with equation
z = f(x, y). We assume that both Sj and SB project onto the region R in the
xy-plane. The lateral surface SL of the region is the set of all (x,y,z) such that
iN Top surface ST v(x, y) 5 Z 5 u(x, y) on the boundary of R, as shown in Figure B.4. The upward
Z = u(x,y) flux across the top surface ST is