Professional Documents
Culture Documents
Ec119 Notes
Ec119 Notes
Nicholas Jackson
ii
EC119 Mathematical Analysis CONTENTS
Contents
0 Introduction 1
0.1 Course description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
0.2 Aims and objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
0.3 Prerequisites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
0.4 Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
0.5 Assignments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
0.6 Recommended books . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
0.7 Learning methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
0.8 The Greek alphabet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1 Set theory 5
1.1 Sets and elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Real Numbers 9
2.1 The Natural Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 The Integers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 The Rational Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Irrational and real numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.5 Some further results for real numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.5.1 Classifying decimals as rational or irrational . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.5.2 Absolute value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.5.3 The Triangle Inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3 Complex Numbers 13
3.1 General Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2 Algebra of Complex Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2.1 Addition, negative and subtraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2.2 Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2.3 Equality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.3 Complex conjugate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.3.1 Multiplicative Inverse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.3.2 Division . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.4 The Fundamental Theorem of Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.5 Geometric interpretation and the polar form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.5.1 Complex multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.5.2 Powers: De Moivre’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.5.3 Reciprocal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.5.4 Division . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.5.5 Conjugate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.6 Roots of polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.6.1 Cube roots of unity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.6.2 The nth roots of unity: generalisation of the above . . . . . . . . . . . . . . . . . . . . . 19
3.7 Exponential form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.7.1 A remarkable result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
CONTENTS iii
EC119 Mathematical Analysis CONTENTS
5 Functions 25
5.1 Domain and codomain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.2 Interval Notation for Subsets of R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.3 Images and image sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.4 One-one functions: injections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
5.5 Onto functions: surjections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
5.6 One-one and onto functions: bijections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
5.7 Composition of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.8 Inverse Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.9 Real-valued functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.10 An alternative definition of function (non-examined) . . . . . . . . . . . . . . . . . . . . . . . . 30
6 Counting 31
6.1 Finite Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
6.2 Infinite sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
6.3 Countable Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
7 Limits 35
7.1 Some well-behaved functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
7.2 Some general properties of limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
7.3 Further properties of limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
7.4 An important limit (1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
7.5 An important limit (2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
7.6 Some more properties and examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
7.7 Limits when x → ±∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
8 Continuity 41
8.1 The Intermediate Value Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
8.2 Numerical methods for solving f(x)=0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
8.2.1 The Newton–Raphson method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
8.2.2 The bisection method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
8.2.3 Direct iteration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
8.3 Monotone functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
9 Differentiability 45
9.1 Differentiation rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
9.2 Differentiating composite functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
9.3 Differentiating inverse functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
9.4 Table of derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
9.5 Leibniz’ Theorem (the extended product rule) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
9.6 Rolle’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
9.6.1 An application of Rolle’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
9.7 The Mean Value Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
9.7.1 Alternative forms of the Mean Value Theorem . . . . . . . . . . . . . . . . . . . . . . . 49
9.8 The Cauchy Mean Value Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
10 L’Hôpital’s Rule 51
CONTENTS iv
EC119 Mathematical Analysis CONTENTS
11 Taylor’s Theorem 53
11.1 Taylor’s Theorem – the nth Mean Value Theorem . . . . . . . . . . . . . . . . . . . . . . . . . 53
11.2 Taylor and Maclaurin Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
11.3 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
11.4 Some Maclaurin series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
11.5 Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
11.6 Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
11.7 Convergence of series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
11.8 D’Alembert’s ratio test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
12 Integration 63
12.1 Indefinite Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
12.2 Linearity Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
12.3 Integral types and methods of integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
12.3.1 Integrals of the form g(f (x))f 0 (x) dx . . . . .
R
. . . . . . . . . . . . . . . . . . . . . . . 63
12.3.2 Integrals re-expressed in partial fraction form . . . . . . . . . . . . . . . . . . . . . . . . 64
12.3.3 Integration by parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
12.3.4 Integrals of the form xα ln x dx . . . . . . . .
R
. . . . . . . . . . . . . . . . . . . . . . . 64
12.4 Definite integration and the Riemann integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
12.4.1 The Newton–Leibniz approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
12.4.2 The Riemann approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
12.4.3 Some properties of definite integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
12.4.4 The Mean Value Theorem for definite integrals . . . . . . . . . . . . . . . . . . . . . . . 67
12.4.5 The Fundamental Theorem of Calculus . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
12.5 Improper Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
12.5.1 Infinite Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
12.5.2 Other improper integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
12.6 Functions defined by integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
CONTENTS v
EC119 Mathematical Analysis CONTENTS
14.4 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
CONTENTS vi
EC119 Mathematical Analysis 0 INTRODUCTION
0 Introduction
Lecturer Dr Nicholas Jackson <Nicholas.Jackson@warwick.ac.uk>
Mathematics B2.38, x28336
Class tutor Dr Mark Cummings <M.S.Cummings@warwick.ac.uk>
Credit 15 CATS
Teaching Two lectures and one class per week, in term 1.
Assessment One 1 21 –hour examination in June (80%)
Five fortnightly coursework assignments (20%)
Lectures Mondays 2pm–3pm R0.12, Fridays 11am–12pm L5
Classes Tuesdays 10am–11am H3.57 and 12pm–1pm H0.58 and Wednesdays 11am–12pm H0.43
By the end of this module you should have acquired a deeper understanding of the mathematics introduced in
other modules such as EC123 Mathematical Techniques B and EC120 Quantitative Techniques, and obtained a
sound foundation for later specialisation in the mathematical and statistical aspects of economics. You will have
had much more practice at thinking in a logical and precise manner, and will hopefully find your capabilities
for clear and rigorous reasoning are improved in a way that will be useful in the rest of your studies.
0.3 Prerequisites
Note that Further Mathematics A–level is not a prerequisite for this module. All that is required is an aptitude
for, and an interest in learning some additional mathematical techniques and establishing the theory behind what
you do in EC123 Mathematical Techniques B, as indicated by a good grade at A–level. (EC121 Mathematical
Techniques A is not a suitable corequisite.) Of course, those students who have done Further Mathematics may
well have an easier ride with some of the topics, but the final results in the summer don’t necessarily reflect
previous knowledge!
0.4 Assessment
The total mark for the module will comprise the following:
0.5 Assignments
Assignments will usually be handed out at the Monday lecture in odd-numbered weeks, and should be handed
in to the Undergraduate Office by the deadline stated on the assignment sheet: this will typically be 12pm
1
EC119 Mathematical Analysis 0 INTRODUCTION
(noon) on the Friday of the following week. This means that you will have just under two weeks to do the
assignment, and you are strongly recommended not to leave all of it until the Thursday evening. Deadlines
will be strictly enforced, and extensions or exemptions granted only in exceptional (and suitably-documented)
unforeseen circumstances such as illness, personal crisis, or natural disasters.
The class tutor will mark your work and return it to you in the following week’s class. Each assignment will
receive a total mark out of 25, comprising up to 20 marks for your answers to the questions, and up to five
marks for clarity of exposition.
Assignments will typically comprise three sections. Section A will consist of easier questions, which you should
do as a warm-up exercise, but which will not be marked or assessed. Section B will consist of harder questions,
which should be handed in by the specified deadline, and which will be marked for credit. Section C consists of
non-assessed questions which may be useful for revision purposes.
When working on your assignment, you may discuss your ideas with other students, but you must write up
your final solutions independently. Any instances of copying may be dealt with severely. (Please see the
Undergraduate Handbook’s section on plagiarism.)
S Lipschutz, Set Theory and Related Topics, Schaum Outline series, McGraw–Hill (1998) QA 248.L4
F M Hart, Guide to Analysis, Macmillan (2001) QA 300.H2
R P Burn, Numbers and Functions, Cambridge (2000) QA 300.B8
G H Hardy, A Course of Pure Mathematics, Cambridge (2008) QA 303.H2
Talk to others
Discuss the material with others on your course; indeed, explaining ideas to someone else is an excellent way of
mastering the material yourself. You may find that other students, for example engineers, scientists, as well as
mathematicians, are doing similar topics.
Use the library
If you find the lecture notes inadequate, look in some of the books suggested.
Seek help
If you have done all of the above, and are still having difficulty, do speak to me at the end of the lecture, or
come to see me during my office hours (no appointment is needed). Don’t suffer in silence.
Work consistently
Since the coursework counts 20% towards your final mark for this module, consistent work really pays off. In
previous years, most students have followed this advice and have done very well overall.
Revise thoroughly
As you will already know, the best way to revise maths is to do lots of problems, not by just reading through
lecture notes. Re-work assignment questions, try past papers, etc.
1 Set theory
Almost all of pure mathematics relies on the concept of a set, an amorphous collection of objects. Usually,
some additional structure is imposed on the set. We might, for example, impose an ordering on the elements
of the set, which then enables us to use symbols like <, >, 6 and > with impunity. If we define a notion of
‘distance’ between two elements of the set, we get a structure called a metric space, which is the subject of an
entire 30-lecture second-year module in the Mathematics Department. If we define some sort of ‘multiplication’
or ‘addition’ operation on the elements of the set, then we get a very versatile structure called a group, which
is important in a wide range of mathematical topics, as well as physics and chemistry.
Next term’s module EC133 Linear Algebra is concerned with objects called vector spaces, which again are
really just sets equipped with some extra structure. But for the rest of this section, we’ll just be looking at
ordinary sets with no extra structure.
The book Set Theory and Related Topics by Seymour Lipschutz [4] is a good (and cheap) reference for the
material in this and some later sections.
Definition 1.1 A set is a collection of objects, which are referred to as the members or elements of the set.
Example 1.2
Sets are usually denoted by capital letters, and the elements, if not specified, by lower case letters.
Definition 1.3 (Membership) If A is a set and x is an element of A, we say that x belongs to A (or x is a
member of A) and we write x ∈ A. We denote non-membership by using the symbol ∈, / thus x ∈
/ A means
that x does not belong to the set A.
We can often describe a set by listing all of its elements. The order in which they are written is unimportant.
Example 1.4
Simon Winchester.
5
EC119 Mathematical Analysis 1 SET THEORY
Example 1.5
(i) {x | x2 − 3x + 2 = 0}
(ii) {x : x is even}
(iii) {x : 0 6 x 6 1} – the set of numbers x such that x lies between 0 and 1 inclusive.
(Strictly speaking, these sets are not properly defined, as we haven’t said what sort of number x represents.)
Definition 1.6 (Finite and infinite sets, cardinality) A set is said to be finite if it contains only a finite
number of elements, and infinite if it contains an infinite number of elements. The order or cardinality of a
set A is defined to be the number of elements the set contains, and is written |A| or card(A).
In Example 1.2, the sets (i), (ii), (iii), (v) and (vi) are finite, while (iv) and (vii) are infinite (but differently so –
more on this later).
Definition 1.7 (Equality) Two sets A and B are equal (written A = B) if they contain exactly the same
elements.
Example 1.8
{x : x2 = 1} = {−1, 1}
Definition 1.9 (Subsets) If A and B are two sets, we say that A is a subset of B (written A ⊆ B, or
sometimes B ⊇ A) if all the members of A are also in B:
‘whenever x ∈ A then x ∈ B’
The symbol ⊆ means that A could be equal to B (compare with the symbol 6). If there are additional
elements of B which are not contained in A, then we say that A is a proper subset of B (written A ⊂ B, or
sometimes B ⊃ A).
Clearly A = B if and only if 2 A ⊆ B and B ⊆ A. That is, every element of A is contained in B, and every
element of B is contained in A.
Definition 1.10 (Intersection) The intersection A ∩ B of two sets A and B is defined to be the set of all
elements which belong to both sets.
A ∩ B := {x : x ∈ A and x ∈ B}
Definition 1.12 (The empty set) The empty set is a set which contains no elements; we represent it with
the symbol ∅, not to be confused with the lowercase Greek letter φ (phi). The empty set is a subset of all
other sets, since all of its elements are elements of any other given set.
Note ∅ =
6 {0}.
2 This phrase, which turns up quite often in mathematics, is occasionally abbreviated as ‘iff’.
Example 1.13
(i) The set of all mountains on earth more than 30 000 feet high is empty.
(ii) If A = {1, 5, 7} and B = {2, 4, 11} then A ∩ B = ∅.
Definition 1.14 (Disjoint sets) If the intersection of two sets is empty, then they are said to be disjoint.
The sets A and B in Example 1.13(ii) are disjoint – they have no elements in common.
Definition 1.15 (Union) Given two sets A and B, their union A ∪ B is defined to be the set of all elements
which belong to A or B or both.
A ∪ B := {x : x ∈ A and/or x ∈ B}
Note that
A ∪ A = A, A ∪ ∅ = A, A ∩ A = A, A∩∅=∅
for any set A. Also note that
A ∪ B = B ∪ A, A∩B =B∩A
for any sets A and B (we say ∪ and ∩ are commutative operations).
Definition 1.17 (The universal set) In any question involving sets, all sets under investigation may be
considered to be subsets of some fixed set, the universal set, which we denote by Ω.
Note that, if A ⊆ Ω,
A ∩ Ω = A, A∪Ω=Ω
for all sets A. (We could, in fact, formally define Ω to be the set satisfying these two properties.)
Definition 1.18 (Difference) Given two sets A and B, their difference A \ B is defined to be the set
consisting of all the elements of A which are not also contained in B:
A \ B := {x : x ∈ A and x ∈
/ B}
Example 1.19
Definition 1.20 (Cartesian product) Given two sets A and B, their Cartesian product A×B is defined to
be the set of all ordered pairs (a, b) such that a ∈ A and b ∈ B. That is,
A × B = {(1, a), (1, b), (2, a), (2, b), (3, a), (3, b)}.
A1 × A2 × · · · An := {(a1 , a2 , . . . an ) : ai ∈ Ai , i = 1, 2, . . . n}
2 Real Numbers
The main purpose of this module is to investigate various properties of functions of real variables. In particular,
we will study calculus (differentiation and integration) in careful detail, and examine concepts such as limits,
continuity and differentiability in a properly rigorous manner.
In this section, we will spend a little time studying the various classes of real numbers.
This is the set of ‘counting numbers’ {1, 2, 3, . . .} and is denoted N (for ‘natural’).
We could extend this set slightly so that we include 0 (call it N∗ say, or N0 ). But if we extend the set further
to include negative numbers as well, the resultant set will satisfy even more of the basic properties specified.
Z+ = {positive integers} = N
Z− = {negative integers} = −N
All integers can be classified as being even (divisible by 2), or odd (not divisible by 2).
All even integers can be written in the form 2n, where n ∈ Z, and all odd integers can be written as 2n + 1, or
if more convenient, 2n − 1, again with n ∈ Z. The evens and odds are disjoint sets:
Zodd ∩ Zeven = ∅
Zodd ∪ Zeven = Z
Example 2.1 The sum of two even integers is even, the sum of two odd integers is even, and the sum of an
even and odd integer is odd.
Proof Let a = 2m and b = 2n be two even integers, where m, n ∈ Z. Then a + b = 2m + 2n = 2(m + n). Since
m and n are both integers, so is m + n and hence a + b = 2(m + n) is an even integer.
Now let a = 2m + 1 and b = 2n + 1. Then a + b = 2m + 2n + 1 + 1 = 2(m + n + 1), which is also even.
Finally, let a = 2m and b = 2n + 1. Then a + b = 2m + 2n + 1 = 2(m + n) + 1, which is an odd integer.
Example 2.2 The product of two even integers is even, the product of two odd integers is odd, and the product
of an even and odd integer is even.
A consequence of this result is that the square of an even integer is even, and the square of an odd integer is
odd.
Also the converse of this is true: given that a2 is even, where a ∈ Z, then a must be even, whereas if a2 is an
odd perfect square, then a must be odd.
m
The set of all numbers of the form n, where m and n are integers and n 6= 0, are called rational numbers,
and is denoted Q (for ‘quotient’).
The rational numbers are dense; that is, between any two rational numbers is at least one other rational
number:
9
EC119 Mathematical Analysis 2 REAL NUMBERS
p
Proof Let a = q and b = rs , where p, q, r, s ∈ Z and q, s 6= 0.
Then
1 1 p r (ps + rq)
2 (a + b) = + = ∈Q
2 q s 2qs
since both numerator and denominator are integers and the denominator is non-zero. Now
Hence there is a rational number 12 (a + b) satisfying a < 21 (a + b). Similarly we can show that 12 (a + b) < b,
and the result follows.
So it looks as though we can ‘fill up’ the whole number line with rational numbers. But this is wrong, as we
shall see next.
√
Theorem 2.4 2 is not a rational number.
This result is attributed to the school of Pythagoras of Samos (480–400BC), although a more rigorous proof
is known to have been in existence by 380BC. This result contradicted a central pillar of the Pythagoreans’
worldview – that the rational numbers underlay the entirety of existence – and an apocryphal legend states that
this was so shocking that Pythagoras had the unfortunate responsible mathematician taken out and drowned.
The following proof uses a technique known as ‘proof by contradiction’ or reductio ad absurdum. Essentially
we start by assuming the desired result is not true, and then proceed logically to show that this leads to an
inconsistency. Thus, our initial assumption must have been incorrect.
√ √
Proof Suppose that 2 is rational. In that case, it can be written in the form 2 = pq , where p and q are
integers which have no factors in common – that is, pq is a proper fraction – and q 6= 0.
Squaring both sides:
p2
2 = 2 =⇒ p2 = 2q 2
q
Thus p2 is an even integer, and by Example 2.2 this means that p also must be even. (If p were odd, then p2
would be odd.)
So we can write p = 2r where r is some integer. Therefore
2q 2 = (2r)2 = 4r2
and so q 2 = 2r2 .
Hence q is also even, for the same reason that p is.
p
This means that both p and q are even, which contradicts the hypothesis that√ q was in its simplest form. So
the only option open to us is to conclude that it is not possible to express 2 as a rational number.
This proof uses a subtle but powerful technique called reductio ad absurdum, or proof by contradiction. Broadly
speaking, we start out by assuming the opposite statement to the one we want to prove, and then proceed
logically from that point until we arrive at a statement which is either inconsistent with the original hypothesis
or clearly false. The only possible conclusion then is that the original assumption was false, and therefore that
the thing we wanted to prove must have been true all along.
The mathematician G H Hardy (1877–1947) wrote, in his book A Mathematician’s Apology:
. . . and reductio ad absurdum, which Euclid loved so much, is one of a mathematician’s finest
weapons. It is a far finer gambit than any chess gambit: a chess player may offer the sacrifice of a
pawn or even a piece, but a mathematician offers the game.
√
Since 2 is not rational, we call it an irrational number. There is no specific symbol for the irrationals, but
all the rational and irrational numbers constitute the real numbers which we denote by the symbol R.
The symbols R+ and R− have analogous meanings to the symbols Z+ and Z− . We can define the irrational
numbers to be R \ Q. Obviously there are an infinite number of square
√ roots which are irrational – as an
exercise, see if you can adapt the proof of Theorem 2.4√to show that 3 is irrational. Similarly, an infinite
number of higher-order roots will also be irrational (eg 3 5).
Irrational numbers which are the solution of polynomial equations (eg x2 = 2) are said to be algebraic.
Irrational numbers which aren’t algebraic are called transcendental – well-known examples of this second
class are π and e.
It isn’t always easy to prove that a number is irrational, and indeed there have been many interesting numbers
which mathematicians have been convinced were irrational, but which were later shown to be rational.
The Feigenbaum constant δ = 4.669201 . . ., which turns up in the study of dynamical systems and chaos
theory (eg in models of population growth) is believed to be transcendental, but has not been proved to be so.
Chaitin’s constant (the probability that a random algorithm halts) is known to be not only transcendental, but
uncomputable.
It is sometimes possible to prove existence statements without finding explicit numbers which satisfy the
statement.
Proposition 2.5 There exist irrational real number x and y such that xy is rational.
√ √ √2
Proof We know that 2 is irrational. Consider the number 2 . It is (obviously) either rational or irrational.
√
If it is rational, then the result follows by setting x = y = 2.
If not, then set √ √2 √
√ √
x = 2 and y = 2.
√ 2 √
Then xy = ( 2 ) 2 = ( 2)2 = 2, which is rational.
√
The fact that 2 is irrational implies that between any two rational numbers there is at least one irrational
number.
√
2
Lemma 2.6 If a, b ∈ Q with a < b, then c = a + 2 (b − a) is an irrational number such that a < c < b.
Case (i) could be included within case (ii) because we can always append an infinite number of zeros to the end
of any finite decimal.
14362
Example 2.7 (i) 0.14362 = 0.1436200000 . . . = 100000 .
1000x = 123.123123123 . . . ,
123
and subtracting x we get 999x = 123 and so x = 999 .
Actually in this example we can simplify further by dividing the numerator and denominator by 3 to give
41
x = 333
If a ∈ R, then
a if a > 0,
|a| :=
−a if a < 0.
√
This means that a 6 |a|. Also |a| = a2 , or |a|2 = a2 . Later we will extend this concept to cover complex
numbers as well.
√
Note Nowadays, is always
√ defined to be the positive square root. So while the solutions of the equation
2
x = 4 are ±2, the symbol 4 means just 2, not ±2.
|a| may be interpreted as the distance between ‘a’ and the origin on the real number line.
|a − b| is the distance between a and b, and |a + b| is the distance between a and −b, (or between b and −a).
|a + b| 6 |a| + |b|
Proof Now (|a + b|)2 = (a + b)2 = a2 + 2ab + b2 . Also since a 6 |a| and b 6 |b|, then
Hence (|a + b|)2 6 (|a| + |b|)2 , and since all absolute values are non- negative, this leads to
|a + b| 6 |a| + |b|
3 Complex Numbers
We now complete the standard hierarchy of number systems
N⊂Z⊂Q⊂R⊂C
In the 16th century, there was much interest in solving algebraic equations. For example: find two numbers
whose sum is ten and product forty.
Algebraically, this is the problem
x + y = 10
xy = 40
x(10 − x) = 40
2
so x − 10x + 40 = 0
√ √ √ √
This has solutions x = 10± 2 −60 = 5 ± −15. So the solutions are x = 5 + −15 and y = 5 − −15.
√
Although, at this point, we haven’t yet attached any meaning to −15, if we manipulate these expressions
using the familiar rules of algebra, we can show that x + y = 10 and xy = 40.
Let us consider the equation x2 + 1 = 0. If x is a real number, this equation has no solutions. We know that
in general, a quadratic equation (a polynomial equation of degree 2) has 2, 1 or 0 real solutions. In some
circumstances, it is useful to think of such equations as having two solutions in all cases. If so, we have to
extend our set of real numbers R to accommodate this.
√
The Swiss mathematician Leonard Euler (1707–1783) introduced the symbol i for −1. We will define it via
i2 = −1, for reasons which will emerge later. In this case we solve x2 − i2 = 0 and the equation x2 + 1 = 0 has
solutions x = i and x = −i.
√ √ √ √ √
In Section
√ 2.7 we obtained √ solutions x = 5 + √−15 and y = 5 − −15. We can write −15 as 15i2 = 15i
(or i 15), and so x = 5 + i 15 and y = 5 − i 15
Any number of the form a + ib (or a + bi) where a and b are real numbers, is called a complex number. We
usually call a complex number z or w rather than x or y. We denote the set of all complex numbers C.
Two complex numbers are added or multiplied term by term using the usual rules of algebra, remembering
that i2 = −1.
Given a complex number z = a + ib, we call the real number a the real part of z and the real number b is the
imaginary part of z. This can be written Re(z) = a and Im(z) = b. Note that these terms are historical and
have no particular significance.
Example 3.1 Re(3 − 5i) = 3 and Im(3 − 5i) = −5, (not −5i).
Given the general complex number z = a + ib, if b = 0, then z is real, so a real number a is equivalent to the
complex number a + 0i.
If z = a + ib and a = 0 then we say that z is purely imaginary. If both a and b are zero, then the complex
number 0 + 0i is equivalent to the real number 0. So zero belongs to C and z = 0 means that both Re(z) = 0
and Im(z) = 0.
13
EC119 Mathematical Analysis 3 COMPLEX NUMBERS
If z = a + bi and w = c + di then
z+w = (a + c) + (b + d)i
−w = −c − di
z−w = (a − c) + (b − d)i
3.2.2 Multiplication
If z = a + bi and w = c + di then
3.2.3 Equality
3.3.2 Division
From this theorem it is clear that a cubic must have one or three real roots.
If a pair of complex conjugate roots are a ± ib, then these give rise to a real quadratic factor because
[x − (a + ib)][x − (a − ib)] = x2 − (a + ib + a − ib)x + (a + ib)(a − ib)
= x2 − 2ax + (a2 + b2 )
Note here that the coefficient of x, viz −2a is ‘−the sum of the roots’ and the constant term a2 + b2 is the
‘square of the conjugates’.
Every complex number z = a + bi can be represented by a point (a, b) in the x–y plane. In this context, the
x–y plane is called the complex plane or Argand diagram.
Im(z)
b z=a+bi
r
θ
a Re(z)
In polar coordinates, the cartesian point (a, b) can be represented as hr, θi. Clearly, a = r cos θ and b = r sin θ.
So we can write
z = a + bi = r(cos θ + i sin θ).
If we square and add we get
a2 + b2 = r2 cos2 θ + r2 sin2 θ = r2
√
and so r = a2 + b2 .
Thus r = |z|, the modulus of z, which was defined earlier. Note that r > 0.
The polar angle θ is called the argument of z, written arg z and is defined by
r cos θ = a
r sin θ = b
Since θ ≡ θ + 2kπ for k ∈ Z, we usually define the principal value of the argument to lie in the range
0 6 θ < 2π or possibly, −π < θ 6 π, (0◦ 6 θ < 360◦ or −180◦ < θ 6 180◦ ).
3.5.3 Reciprocal
If z = r(cos θ + i sin θ) = hr, θi then
1 1
=
z r(cos θ + i sin θ)
1 ((cos θ − i sin θ))
= ·
r (cos θ + i sin θ)(cos θ − i sin θ)
1
= · (cos θ − i sin θ).
r
Now cos(−θ) = cos(θ) and sin(−θ) = − sin(θ) so the above could be written as
1 1
= [(cos(−θ) + i sin(−θ)] .
z r
Thus, in polar form,
1 D1 E
= , −θ
z
r−1
or z −1
= r , (−1)θ
It may also be shown that
3.5.4 Division
If z1 = hr1 , θ1 i and z2 = hr2 , θ2 i then
z1 D1 E
= z1 .z2−1 = hr1 , θ1 i , −θ2 , z2 6= 0.
z2 r2
3.5.5 Conjugate
Example 3.5
(i) Mark on an Argand diagram the complex numbers z1 = 3 − 2i, z2 = −i, z3 = i2 , z4 = −2 − 4i, z5 = 3.
Find the modulus and argument of each complex number and hence write in polar form.
z 3 − 1 = (z − 1)(z 2 + z + 1),
√
3
so z 3 − 1 = 0 when z = 1 and when z 2 + z + 1 = 0, ie when z = − 21 ± 2 i.
√ √
So the roots of z 3 − 1 are 1, − 12 + 23 i and − 12 − 23 i. These are the three cube roots of 1, or the ‘cube roots of
unity’.
Method 2 We solve z 3 = 1 and express both sides of the equation in polar form. If z = hr, θi, then z 3 = hr3 , 3θi
and 1 = h1, 0i, so we solve
hr3 , 3θi = h1, 0i,
giving r3 = 1 and 3θ = 0.
This means that r = 1. We may be tempted to say that the second equation has the one solution θ = 0, but
this is wrong. We have to remember that the argument of a complex number isn’t defined uniquely, but we can
always add on integer multiples of 2π.
Thus we write 3θ = 2kπ, k = 0, 1, 2, . . ., i.e.
Dividing by 3 gives
2π 4π 8π
θ = 0, , , 2π, ,......
3 3 3
It is clear that after the first three values, 0, 2π 4π
3 , 3 , the remaining solutions are just repetitions.
2kπ
So we can write the three solutions as z = 3 , for k = 0, 1, 2. Writing in full, the solutions are
Remarks
2π
(ii) The three roots are distributed evenly around the origin with an angular separation of 3 .
z12 = h12 , 2. 2π 4π
3 i = h1, 3 i = z2
z22 = h1, 8π 2π
3 i ≡ h1, 3 i = z1 .
So if we denote either of z1 or z2 by ω then the other one is ω 2 and the three solutions are 1, ω and ω 2 .
P (z) = z 3 + az 2 + bz + c
and matching up with z 3 + az 2 + bz + c we, see in particular that the sum of the roots is α + β + γ = −a
In (iii) our polynomial is z 3 − 1. Since the term in z 2 is missing, or equivalently, its coefficient is zero,
then the sum of the roots is 0, and hence
1 + ω + ω2 = 0
z n + an−1 z n−1 + · · · a2 z 2 + a1 z + a0 = 0,
the sum of the solutions of the equation is equal to ‘the coefficient of z n−1 ’, i.e. the real number −an−1 .
Now when we look at the equation z n − 1 = 0 we see that an−1 = 0, so this tells us that the sum of the
roots of the polynomial is 0.
(ii) In finding the roots of z n − 1, we know that one root z0 = 1. Consider the root z1 = h1, 2π
n i. Then
Example 3.6
Example 3.8
(i) How would you find the roots of z 5 − z 4 + z 3 − z 2 + z − 1 = 0?
(ii) Can you generalise to find the roots of 1 + z + z 2 + · · · + z n ?
eiπ = −1
Direct Proof
Start with ‘What I know’.
Set up a chain of argument.
End up with ‘What I want’.
21
EC119 Mathematical Analysis 4 PROOF AND REASONING
Contradiction
Assume that ‘the required result is false’.
Work to a contradiction.
Exhaustion
Check all possible cases individually.
Demonstration
Use this method when you are asked to ‘Prove there exists . . . ’.
You only need one example.
Counterexample
Use this method when you need to disprove something.
You need to show that it is false for some case.
One example is enough.
Definition 4.1 (Principle of mathematical induction) Suppose we have a variable proposition P (n) which
depends on some natural number n. Then if P (1) is true, and if P (k) ⇒ P (k + 1) for some k ∈ N, then P (n)
is true for all n ∈ N.
n(n+1)
Example 4.2 Let P (n) be the statement “the sum of the first n natural numbers is 2 .”
We can prove this to be true for all n ∈ N.
Sn = 1 + 2 + 3 + · · · + (n − 1) + n
Sn = n + (n − 1) + (n − 2) + · · · + 2 + 1
2Sn = (n + 1) + (n + 1) + (n + 1) + · · · + (n + 1) + (n + 1)
| {z }
n times
Pn n(n+1)
So Sn = i=1 = 2 .
(ii) Proof by Induction
(a) (The first rung)
P (1): the sum of 1 natural number is obviously 1. Also from the formula,
1(2)
S1 = 2 = 1,
k(k+1)
So we assume Sk = 1 + 2 + · · · + k = 2 . Then
Then Sk+1 = Sk + (k + 1)
k(k+1)
= 2 + (k + 1)
(k+1)(k+2)
= 2
(k+1)(k+1+1)
= 2
P (1) is true, and P (1) ⇒ P (2) ⇒ P (3) ⇒ · · · ⇒ P (n) ⇒ P (n) is true for all n ∈ N.
Example 4.3 The sum of the first n odd numbers is n2 , that is,
n
X
Sn = 1 + 3 + 5 + · · · + (2n − 1) = (2i − 1) = n2 .
i=1
5k+1 − 1 = 5.5k − 1
= 5(4m + 1) − 1
= 20m + 4
= 4(5m + 1)
Example 4.6 Prove that if z is a complex number which in polar form is hr, θi, then z n = hrn , nθi for all
n ∈ N.
5 Functions
We can illustrate the theory in this section by means of the following, rather contrived example
Example 5.1
(i) Consider the following scenario: We have a group of customers and a selection of goods that they can
buy. The rules of the game are as follows:
(a) Every customer has to choose one good (one and only one).
(b) Two or more customers are allowed to choose the same good, (there is a large enough supply).
(c) Not all goods have to be chosen, indeed all customers could choose the same good.
To be more precise, let the customers be a, b, c, d and the goods be p, q, r, s, t, u, v.
We know that a chooses q: so we write this as f (a) = q. Similarly, b chooses p, c chooses q and d chooses
u. So we have f (a) = q, f (b) = p, f (c) = q, f (d) = u.
We can represent this information in a diagram:
p
q
a
r
b
s
c
t
d
u
v
Note that the only goods chosen are p, q and u. Goods r, s, t and v have not been chosen by anyone.
We say that the domain of f is the set {a, b, c, d}, the codomain of f is the set {p, q, r, s, t, u, v} and the
image set is the set {p, q, u}. f is a function from the domain to the codomain.
(ii) We change the rules. Now, any good chosen by one customer cannot be chosen by anyone else, ie suppose
f (a) = q, f (b) = p, f (c) = t, f (d) = u. Again, some of the goods (r, s and v) have not been chosen.
p
q
a
r
b
s
c
t
d
u
v
The domain and codomains are as before, but the image set is now {p, q, t, u}. Since no good is chosen by
more than one customer, the function is said to be one-one or injective.
(iii) Yet another rule change: this time there are fewer goods to choose from. More than one customer is
allowed to choose a particular good, as in (i), but now there are none left over. All goods are chosen by
somebody.
Suppose the customers are a, b, c, d as before, the only goods now are p, q and u, and f (a) = q, f (b) = p,
f (c) = q, f (d) = u.
a
p
b q
c
u
d
Here the domain is the same as in (i) and (ii), the codomain is {p, q, u} and so also is the image set. Since
the image set is equal to the codomain, the function is said to be onto or surjective.
25
EC119 Mathematical Analysis 5 FUNCTIONS
This rather convoluted example illustrates the general properties of functions which we shall investigate next.
Note that I didn’t give a mathematical formula for the rule for f , just a verbal description of what it does.
Definition 5.2 Let A and B be sets. A function f : A → B, where f : x 7−→ y is a rule which assigns to
each x ∈ A a single (that is, exactly one) element y ∈ B. This assigned element is denoted f (x).
A is called the domain of the function f , and B is called the codomain of f .
(i) a < x < b is denoted by (a, b), and is called an open interval. (Some books use ]a, b[ ).
(iv), (v) and (vi) are all called semi-infinite intervals. If you prefer, you can use interval notation wherever
appropriate.
Examples 5.3
(i) Let A = B = R then
(a) f (x) = x2 is a function which assigns to each x ∈ R, the unique value x2 .
(b) f (x) = 0, for all x ∈ R is called the zero function.
(c) f (x) = x, for all x ∈ R is called the identity function.
(ii) Let A = {x : x ∈ R and x > 0}, or you could write this as R+ or using interval notation (0, ∞) and
B = R. Then f (x2 ) = x is not a function since both 2 and −2 are elements which satisfy this relation
(counter example).
√ √
If we write f (x) = x, then this is a function from (0, ∞) to R, since is defined to be the positive
square root.
(iii) Let A = the set of all car owners in GB, B = the set of all registration numbers of cars. f is a function
which assigns to each car owner the registration number of his/her car.
(iv) f , as given in Example 5.1, is a function.
Definition 5.4 (Image) In the above definition of function, corresponding to each element x ∈ A is the
unique element f (x) in B. f (x) is the image of x in B.
Definition 5.5 (Image set) The image set or range of the function f : A → B is the set of all values of
f (x) attained in B, with x ∈ A. It is a subset of the codomain B. and we could write the image set as f (A);
clearly f (A) ⊆ B.
In Example 5.1(i), the image set is {p, q, u}.
Definition 5.6 (Injective function) A function f : A → B is one-one, injective, or an injection if f (a) = f (b)
only when a = b.
The function defined in Example 5.3(ii) is one-one.
Definition 5.7 (Surjective function) A function f : A → B is onto, surjective, or a surjection if its image
set exactly coincides with its codomain. That is, for any b ∈ B there exists at least one (and possibly more
than one) a ∈ A such that f (a) = b.
Definition 5.8 (Bijective function) A function which is both injective and surjective is said to be bijective
or a bijection.
Much of the above can often be illustrated best using graphs, although this isn’t appropriate for all functions.
Example 5.9
(i) The function f : R → R given by f (x) = x2 is not onto because in this case the image set is the set of
non-negative real numbers, not the whole set R. We could write the image set as R+ ∪ {0}, or R+ 0 or
much easier, [0, ∞).
(ii) The function f : R → R+ 2
0 , f (x) = x is onto.
(iii) (i) above is not one-one: for example, f (2) = 4 and f (−2) = 4 (counter-example).
(iv) The function f : (0, ∞) → R, where f (x) = x2 is one-one.
Proof (Take note of the method of proof used here, (also see parts (vi) and (vii)).
We have to show: given f (a) = f (b) for all real a, b > 0 then a = b.
Consider f (b) − f (a) = b2 − a2 = (b − a)(b + a) = 0.
Now (b + a) > 0, since both a and b are positive, therefore a = b.
OR f (a) = f (b) ⇒ a2 = b2 ⇒ a = ±b. Since neither a nor b is negative, this implies that a = b.
(v) Let X = {1, 2, 3, 4}, Y = {a, b, c, d}, then f : X → Y given by f (1) = b, f (2) = c, f (3) = a, f (4) = d is a
bijection since it is both one-one and onto.
(vi) The function f : R → R given by f (x) = x + 1 is a bijection
It is onto because the image set f (R) is the whole of the real numbers R.
It is one-one because if for any a, b ∈ R, f (a) = f (b),
then a + 1 = b + 1 ⇒ a = b.
(vii) Let Y = {x ∈ R : x 6= 0} = R \ {0}.
1
The function f : Y → Y given by f (x) = x is a bijection.
It is onto because the image set is Y
It is one-one because for all a, b ∈ Y .
1 1
f (a) = f (b) ⇒ a = b ⇒ a = b(6= 0)
Definition 5.10 (Composite function) Given two functions f : A → B and g : C → D such that f (A) ⊆ C,
we can form the composite function (g ◦ f ) : A → D by defining (g ◦ f )(a) = g(f (a)) for all a ∈ A.
Example 5.11
(b) Similarly, the image set of g is the whole of R which is the same as the domain of f , and f ◦ g can
be formed.
f ◦ g(x) = f (g(x)) = f (5x) = 3(5x)2 + 2 = 75x2 + 2
(ii) Let A be the set of even integers, ie A = {x ∈ Z : x = 2m, for some m ∈ Z}
Let f : Z → A be given by f (x) = 2x and g : A → Z be given by g(x) = x2
(a) The image set of f is A, so g ◦ f can be formed
(b) The image set of g is of the form {x ∈ Z : x = 4m, for m ∈ Z} and so is a subset of Z, the domain
of f , so f ◦ g can be formed.
Definition 5.12 (Inverse function) Let X and Y be sets, and suppose that f is a function from X to Y .
We say that f is invertible if there exists a function f −1 : Y → X such that:
(i) f −1 ◦ f (x) = x for all x ∈ X,
(ii) f ◦ f −1 (y) = y for all y ∈ Y .
Proof (Omitted.)
Examples 5.14
(ii) Refer to Example 5.9(i) and (ii). Neither of these functions has an inverse.
√
Suppose f : [0, ∞) → [0, ∞), where f (x) = x2 . Then writing y = x2 we obtain x = y. So we write
√
f −1 : [0, ∞) → [0, ∞), f −1 (x) = x
(X and Y are the same in this case).
(iii) Refer to Example 5.9(vi): f : R → R, f (x) = x + 1. Write y = x + 1, then x = y − 1 and the inverse
function is
f −1 : R → R, f −1 (x) = x − 1
(Check: f ◦ f −1 (x) = f (f −1 (x)) = f (x − 1) = (x − 1) + 1 = x).
(iv) f : R → R+ , f (x) = ex . Write y = ex , then x = ln y, so
f −1 : R+ → R, f −1 (x) = ln x
(Check: f −1 (f (x)) = f −1 (ex ) = ln(ex ) = x; similarly, f (f −1 (x)) = f (ln x) = eln x = x).
(v) Y = R \ {0} f : Y → Y , where f (x) = x1 . Write y = x1 , then x = 1
y
f −1 : Y → Y, f −1 (x) = x1 .
This function is self inverse.
Much of this is much easier to grasp when we resort to the use of diagrams and graphs.
The graph of f −1 is just the graph of f reflected in the line y = x.
The product and quotient of two even functions or of two odd functions is even.
The product and quotient of an even and an odd function is odd.
(Note that the domain of any quotient function will have to exclude any values for which the denominator
is zero). In order to refer sensibly to a function as being even or odd, we have to assume that its domain is
symmetric about the origin.
(i) for each x ∈ X, (x, y) ∈ S for some y ∈ Y , ‘there is a y corresponding to each x’.
(ii) if (x, y1 ) ∈ S and (x, y2 ) ∈ S where x ∈ X and y1 ∈ Y , y2 ∈ Y , then y1 = y2 , ‘for each x there corresponds
only one y’.
Example 5.16
(i) Consider the function from N to N given by S = {(1, 2), (2, 4), (3, 6), . . .}.
Clearly, S = N × N, that is, S is a subset of the Cartesian product N × N. (Check that S satisfies the
conditions above). This function corresponds to the function f : N → N given by f (x) = 2x under our
previous definition.
(ii) Let X = {a, b, c} and Y = {d, e, f, g}. Then S = {(a, d), (b, d), (c, e)} is a function. The function
corresponds to f (a) = d, f (b) = d, f (c) = e.
(iii) Let X = {1, 2, 3} and Y = {6, 7, 9}. Then S = {(1, 7), (2, 8)} is not a function because it does not satisfy
condition (i), since (3, −) is missing.
(iv) Let X and Y be as in (iii) above. Then S = {(1, 7), (1, 8), (2, 6), (3, 6)} is not a function because it does
not satisfy condition (ii). Both y = 7 and y = 8 correspond to x = 1.
6 Counting
6.1 Finite Sets
If the set is finite, then counting how many elements are in the set is theoretically a straightforward process,
even if not so in practice.
Example 6.1 Consider the finite sets in Examples 1.2 and 1.4 (in Section 1).
(i) A = {−3, 137, 1391} (ii) B = {the people living in Coventry} (iii) C = {x ∈ R : x2 +3x+2 = 0}
(iv) D = {b, c, d, f, g, h, j, k, l, m, n, p, q, r, s, t, v, w, x, y, z} (v) E = {a, e, i, o, u}
(vi) F = {Everest, K2, Kangchenjunga, Lhotse}
In each case, you would probably count by working systematically from left to right. There is a bijection
between A and the set {1, 2, 3}, between C and {1, 2}, between D and the set {1, . . . , 21}, and between E and
{1, 2, 3, 4, 5}.
The population of Coventry is continually and unpredictably fluctuating, so it’s difficult to say exactly what the
cardinality of B is at any given time. It is, however, finite, and in 2006 was estimated to be just under 310 000.
There are four mountains on Earth with heights above 8 500m, so there is a bijection between F and the set
{1, 2, 3, 4}.
Definition 6.2 A set X is said to be finite if either X = ∅, or for some n ∈ N there is a bijection between X
and the set {1, 2, . . . n}. Then X is said to have size or cardinality |X| = n.
Example 6.4 N, Z, Q, R and C are all infinite (why?) but N ⊂ Z ⊂ Q ⊂ R ⊂ C, so is R more infinite than N,
say?
Definition 6.5 A set X is said to be countable if either X is finite or there exists a bijection from N to
X, (that is, there is an exact match between the two sets). A set X is said to be uncountable if it is not
countable.
Example 6.6
f (1) = 0, f (2) = 1, f (3) = −1, f (4) = 2, f (5) = −2 and so on. f is a bijection (why?).
Example 6.8
31
EC119 Mathematical Analysis 6 COUNTING
(iii) The set of all natural numbers greater than 100 is countable.
(iv) Any subset of the natural numbers is countable, and any infinite set of natural numbers has the same
cardinality as N itself.
Definition 6.9 A set X which is infinite but countable is said to have cardinality ℵ0 (aleph null, aleph nought
or aleph zero).
Q+ = {q ∈ Q : q > 0}.
f : N → Q+ , given by f (n) = an
is a bijection, and Q+ is countable. We can then extend the list to the whole of Q.
0, a1 , −a1 , a2 , −a2 , . . .
Remark Although we do not have an explicit formula for the bijection, we have an explicit prescription for
constructing it:-
1 2 3 4 5 6 7 8 9 10 11 ...
↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓
1
0 1 -1 2 − 12 2 -2 3 -3 1
3 - 13 ...
Theorem 6.11 R is not countable, that is, it is uncountable. There is no bijection from N to R
Proof (Omitted.)
Remark Since R is not the same size as N it must have its own cardinality. R, or any other set X for which
there is a bijection from R to X is said to have cardinality ℵ1 (aleph one).
7 Limits
First we present an informal definition:
Definition 7.1 (informal) If we can make f as close as we like to a real number `, by making x sufficiently
close to, but not equal to, the real number a, we say that f (x) approaches or tends to the limit ` as x tends
to a. We write either
f (x) → ` as x → a or lim f (x) = `
x→a
This definition means that the ‘distance’ between f (x) and ` can be as small as any prescribed number provided
that the ‘distance’ between x and a is small enough. Mathematically this is written as follows:
Definition 7.2 (formal) For a real valued function f of a real variable x, f (x) tends to the limit ` when
x → a if for every choice of ε > 0, however small, there is a corresponding number δ > 0 such that
x x 6= 0
(ii) g(x) =
1 x=0
x
(iii) g(x) = |x|
If f and g are real valued functions of a real variable x and if f (x) → ` and g(x) → m as x → a then
(i) k.f (x) → k`, where k is a constant;
35
EC119 Mathematical Analysis 7 LIMITS
that is, f is always sandwiched between g and h. If both g(x) and h(x) tend to the same limit ` as
x → a, then f (x) → ` as x → a.
(ii) If f (x) → ` and g(x) → m as x → a and f (x) 6 g(x), in some neighbourhood of a, then ` 6 m.
We now let x decrease towards zero, or ‘approach zero through positive values’, which is usually written x → 0+ ,
then cos x increases towards 1.
Hence sinx x also increases towards 1, since it is ‘sandwiched’ between cos x and 1.
This result is also true when x approaches zero through negative values, which we write x → 0− , because sinx x
is an even function.
We can therefore say that sinx x → 1 as x → 0 in any manner, or
sin x
lim x = 1.
x→0
This limit is important because it is used in proving that the derivative of sin x is cos x.
0.9
0.8
0.7
0.6
0.5
–2 –1 0 1 2
x
sin x
Figure 7.1: f (x) = x
sin x
Remark In the above example we have said nothing about whether or not x
In fact sinx x is indeterminate when x = 0 as we have stated above. Also we did not need to use any value for
f (0) at any stage of our calculation of the limit. There is no value of x for which sinx x = 1 and we can say that
the limit is unattained.
This means that when x becomes close to 0, sinx x becomes close to 1, and we can make sinx x as close as we
please to 1 for all values of x sufficiently close to 0.
√
1+x−1
Example 7.6 limx→0 x .
sin x
Example 7.8 We know that x → 1 and (x + 2) → 2 as x → 0. Then
5 sin x
(i) x →
sin x
(ii) x +x+2→
sin x(x+2)
(iii) x →
x 1
(iv) sin x → and x+2 →
sin x
(v) x(x+2) →
xm −am
lim = mam−1 , m∈Q
x→a x−a
x−a = − x(x na
n an (x−a) → − a2n = −na−n−1 = mam−1
The proof is trickier when m is not an integer. This limit is important in proving that the derivative of xn is
nxn−1
Remarks
(i) The phrase ‘x tends to a’ implies that x must be allowed to approach a from both sides, ie from below,
x → a− and from above, x → a+ and obtain the same answer from both sides.
0.8
0.6
0.4
0.2
–2 –1 0 1 2
x
–0.2
1
Figure 7.2: f (x) = x sin x
sin x
(ii) f (x) does not necessarily approach ` steadily as x → a as is the case for x when x → 0, (see Figure 7.1
on page 36).
Consider the function f given by f (x) = x sin x1 (see Figure 7.2).
The function is even because it is the product of x and a sine function, which are both odd. We find that
f is positive for some values of x near 0, and negative for others. What is critical is that the numerical
difference between f and ` should get smaller and smaller, that is, it should be as small as we like,
provided we are close enough to 0.
(iii) The difference between f and `, as described above, must be as small as we require for all values of x
sufficiently near a.
Consider the function f (x) = sin x1 (see Figure 7.3).
0.5
–2 –1 1 2
x
–0.5
–1
1
Figure 7.3: f (x) = sin x
(iv) In the definition of limit, no mention is made of f (a), the value of f at x = a, indeed the function may
not even be defined when x = a. So when we say ‘all values of x sufficiently close to a’, we mean all
values of x in some small interval about x = a, but excluding the value a itself. This interval may be
written in the form a − δ < x < a + δ, for x 6= a or, better,
0 < |x − a| < δ.
(|x − a| is the ‘distance’ between ‘x’ and ‘a’). The interval is called a neighbourhood of a.
(v) If f (x) = k, a constant, for all x in some neighbourhood of a, then limx→a = k.
20
y 10
–10
–20
1 1
Figure 7.4: f (x) = x sin x
1
The function sin assumes all values between 1 and −1, however close we get to x = 0.
x
It follows that f (x) = x1 sin x1 can take any value, positive or negative, however large.
1
But f (x) does not remain large for all values of x near x = 0, eg it is zero when x = nπ , for n ∈ Z, and
so does not tend to ±∞ as x → 0.
We say that
(a) sin x1 oscillates finitely between -1 and +1 (see Figure 7.3),
1 1
(b) x sinoscillates infinitely as x → 0 (see Figure 7.4).
x
(iv) (a) The functions x1 , x12 and x1 sin x1 are all unbounded near x = 0, because they tend to infinity
(positive or negative) as x → 0.
(b) sin x1 is said to be bounded near x = 0 because it lies between two fixed numbers for all small
values of x.
The terms ‘bounded’ and ‘unbounded’ can be applied to the behaviour of any function in the neighbourhood
of any number a.
Theorem 7.9 If f (x) → ` as x → a, then f (x) is bounded near x = a. (` is finite and when x is close to a,
f (x) is close to `).
Definition 7.10 (informal) If the values of f (x) can be made as close as we like to the number ` by making
x sufficiently large, we say that f (x) tends to the limit ` as x tends to infinity. We write
x2 +1
Example 7.11 Consider the behaviour of f (x) = x+1 as x → −1 and as x → ±∞.
1
Example 7.12 (i) limx→∞ xα , α>0
x2 −4x+3
(ii) limx→∞ x3 −x2 +x−1
√ √ √
(iii) limx→∞ x x+a− x
8 Continuity
We want a mathematically precise concept of a continuous function. Intuitively, the graph of such a function
f (x) should have no breaks or jumps in it. That is, at any given point, the value of the function should be
equal to the limit of the function (from both sides):
Examples 8.2
We are particularly interested in functions which are continuous for most, if not all, of their domain:
Theorem 8.4 The sum, difference, product and quotient of continuous functions are also continuous functions
(except for the zeros of the denominator in the case of quotients).
Care must be taken with domains:
The domain of sums, differences, products are the intersections of individual domains.
The domain of a quotient is the intersection of individual domains, minus any values for which the denominator
is zero.
(i) f : R → R, f (x) = xn , n ∈ N.
(ii) g : R∗ → R, g(x) = 1
xn , n∈N
(iii) Polynomials:
p : R → R given by p(x) = a0 + a1 x + a2 x2 + · · · + an xn , ai ∈ R, n ∈ N.
p(x)
(iv) Rational functions, that is, functions which are quotients of polynomials = q(x) . The domain excludes
any values of x for which q(x) = 0.
(v) sin(x), continuous for all x ∈ R.
(vi) cos(x), continuous for all x ∈ R.
41
EC119 Mathematical Analysis 8 CONTINUITY
Examples 8.8
(i) y = sin(x), and −1 6 sin(x) 6 1. Bounds are ±1, which are attained.
(ii) φ(x) = x1 , for 0 < x 6 1, and φ(0) = 0. φ is finite in [0, 1], but it is not continuous on this closed interval.
It tends to −∞ as x → 0, and is not bounded.
(iii) f (x) = sin(x) π
x , where 0 < x 6 2 . An upper bound is 1, but
sin(x)
x never takes the value 1 at any point
within the given interval, and it is not continuous since it is not defined at x = 0.
(
1
sin(x) x 6= 0,
x ∈ 0, π2 .
(iv) g(x) = x
1 x = 0.
2 1
Now, the function is defined at x = 0. We see that M = 1 and m = π, and that limx→0 x sin(x) = 1 = g(0),
hence g is continuous.
x2 −1
Examples 8.10 (i) limx→1 sin x−1
Theorem 8.11 (Simple version) If f is continuous on [a, b] and f (a) and f (b) have opposite signs, then
there is at least one point c ∈ (a, b) at which f (c) = 0.
Theorem 8.12 (General version) If f is continuous on [a, b] and f (a) = α and f (b) = β, then for any
number γ ∈ (α, β) there is at least one number c ∈ (a, b) at which f (c) = γ.
Both versions of these theorems together are called the Intermediate Value Theorem (IVT). They are very
useful, and in fact are just common sense.
Example 8.13 Show that the polynomial equation p(x) = x5 − 5x + 2 = 0 has at least 3 real roots.
By experimentation, we find that p(−2) = −20, p(−1) = 6, p(0) = 2, p(1) = −2, p(2) = 24, so there must be a
root in each of the intervals (−2, 1), (0, 1) and (1, 2).
There could be more. How many?
Corollary 8.14 If f is continuous on [a, b] and is such that a 6 f (x) 6 b whenever a 6 x 6 b, then the
equation f (x) = x has a real root in [a, b].
f (xn )
xn+1 = xn − .
f 0 (xn )
Newton’s method does not always work in finding a specific root. It may converge to the ‘wrong’ root if you
choose the ‘wrong’ initial value.
Example 8.15 Find, to an accuracy of one decimal place, a root of the quadratic polynomial p(x) = x5 − 5x + 2.
We first note that p(0) = 2 and p(1) = −2, so by the IVT, there is a solution in the interval (0, 1). Take the
initial approximation to be x0 = 0.5.
(x5 −5xn +2)
The first derivative of p(x) is p0 (x) = 5x4 − 5, so we have to iterate xn+1 = xn − n(5x4 −5) .
n
Setting x0 = 0.5, we find that x1 = 0.4 and x2 = 0.4021. Since x1 and x2 agree to the required accuracy we
can stop, so x = 0.4 to one decimal place. (Carry on for greater accuracy.)
Example 8.16 Use the bisection method to find an approximate (to an accuracy of one decimal place) solution
of the equation x5 − 5x + 2 = 0.
Interval [0, 1], p(0) > 0, p(1) < 0, so take x = 0.5, p(0.5) < 0,
Interval [0, 0.5], p(0) > 0, p(0.5) < 0, p(0.25) > 0
Interval [0.25, 0.5], p(0.25) > 0, p(0.5) < 0, p(0.375) > 0
Interval [0.375, 0.5], p(0.375) > 0, p(0.5) < 0, p(0.4375) < 0
Interval [0.375, 0.4375], p(0.375) > 0, p(0.4375) < 0, p(0.40625) < 0
This is very slow.
xn+1 = f (xn ).
This will converge to the required root c, i.e. c = f (c), if both the initial value x0 and c both lie within an
interval for which
|f 0 (x)| 6 α, where α < 1
Example 8.18 Show that the equation xex = 1 has a solution between 0 and 1. Find a root correct to 2
decimal places.
Definition 8.19 The function f : X → Y is said to be strictly increasing on a set X if, whenever x1 < x2 ,
for all x1 , x2 ∈ X.
Note that f doesn’t have to be continuous in this definition.
A strictly monotone (or strictly monotonic) function is one which is either strictly increasing or strictly
decreasing.
By a (not strictly) monotonic increasing function, we mean that
9 Differentiability
Having considered what it means for a function to be continuous, we now want to examine what it means for a
function to be differentiable. Intuitively, we expect the graph of the function to be smooth, with no jagged
points in it.
Definition 9.1 (Derivative at a point) Suppose f (x) is defined at all points in the neighbourhood of x = a,
and the ratio
f (a + h) − f (a)
h
tends to a (finite) limit as h → 0, then f is said to be differentiable at a. The value of the limit is called the
derivative of f at a, and is denoted by f 0 (a). That is,
f (a + h) − f (a)
f 0 (a) = lim
h→0 h
Note that the limit must exist and be taken from both sides of x = a.
As before, we can extend the concept of differentiability at a single point, to differentiability over an interval:
Definition 9.2 (Derivative in an interval) Suppose f (x) is defined for all x ∈ (a, b). Then f is said to be
differentiable in (a, b) if it is differentiable at all points of the interval as determined by Definition 9.1.
There is an important connection between the concepts of continuity and differentiability:
Proof It is often more convenient to write x in place of (a + h), and in this case, h is replaced by (x − a) and
with this notation, the limit becomes
Cross multiplying,
lim (f (x) − f (a)) = f 0 (a) lim (x − a) = 0,
x→a x→a
Example 9.5 Are the following functions differentiable? If not, why and where not?
(i) f (x) = xn , (n ∈ N)
(ii) g(x) = |x|
sin(x)−sin(a)
Example 9.6 To find the derivative of sin(x) from first principles, we consider the ratio x−a . Using
the trigonometric identity
sin C − sin D = 2 cos C+D C−D
2 sin 2 ,
we have
x+a x−a
sin(x) − sin(a) 2 cos 2 sin 2 x+a
x−a
x−a
= = cos 2 · sin 2 2
x−a x−a
so
sin(x) − sin(a) 2a
lim = cos 2 · 1 = cos(a)
x→a x−a
Thus the derivative of sin(x) is cos(x).
45
EC119 Mathematical Analysis 9 DIFFERENTIABILITY
Rules (i) and (ii) together mean that differentiation is a linear transformation.
Suppose g is differentiable at x = a and that g(a) = b. Also suppose f is differentiable at b. Then the
composite function f ◦ g or f (g(x)) is differentiable at a, and its derivative at a is f 0 (b)g 0 (a). More generally,
if h(x) = f (g(x)) on some interval, then h0 (x) = f 0 (g(x))g 0 (x).
This is the familiar chain rule when written in Leibniz notation:
Write u = g(x), y = f (u), so that
dy dy du
= · .
dx du dx
cos−1 x
− a21−x2
xn nxn−1 cos(ax) −a sin(ax) a
√
tan−1 x a
eax aeax tan(ax) a sec2 (ax) a a2 +x2
1
ln(ax) x
y 00 = u00 v + 2u0 v 0 + uv 00
000
y = u000 + 3u00 v + 3u0 v 0 + uv 000
Clearly this notation is impractical, so instead we write y 0 = y (1) , y 00 = y (2) , etc, then these equations become
d5 2
Example 9.11 Find dx5 (x sin(x)).
Question Why an open interval in the second hypothesis? Why not a closed interval?
Example 9.13 Verify that Rolle’s Theorem is satisfied for the following function on the given interval. Hence
find the value(s) of c.
f (x) = x(1 − x)3 on [0, 1].
Theorem 9.14 If f is continuous on [a, b] and differentiable on (a, b) and f 0 (x) > 0 in (a, b), then f is strictly
increasing throughout [a, b]. If f 0 (x) > 0, then f is increasing (but not strictly so).
(i) There is at least one solution in (−2, −1), in (0, 1) and in (1, 2) – a minimum of three solutions.
(ii) There are at most 5 solutions for x ∈ R (by the Fundamental Theorem of Algebra). We now show that it
has exactly 3 roots and no more.
(iii) p0 (x) = 5x4 − 5, p0 (x) = 0 when x = ±1, so there is at most one solution of p(x) = 0 between -1 and 1, so
using (a) there is exactly on solution in (0, 1).
(iv) For |x| > 1, p0 (x) > 0, so p is a strictly increasing function.
(v) For x > 1, since p(1) = −2 < 0, there can be at most one solution, which is the solution in (1, 2).
(vi) For x < −1, p(−1) = 6 > 0, p0 (x) > 0, so p decreases as we move to the left and so there can be at most
one solution which is the solution in (−2, −1).
Example 9.17 Show that xesin x = cos x has no more than one root in the interval (0, π2 ). First use the
Intermediate Value Theorem to prove that it has at least one root.
f (b) − f (a)
= f 0 (c).
b−a
This theorem says that the gradient of the chord between the points (a, f (a)) and (b, f (b)) is equal to the
gradient of the tangent to the curve at x = c.
If, as indicated in the diagram, b > a, and in addition, if f 0 (c) > 0, then f (b) > f (a) and the function is
increasing.
Theorem 9.20 If f is continuous on [a, b] and differentiable in (a, b) and f 0 (x) = 0 for all x ∈ (a, b), then
f (x) is constant for all x ∈ [a, b].
Corollary 9.21 If both f and g are continuous in [a, b], and are such that f 0 (x) − g 0 (x) = 0 for all x ∈ (a, b)
then f (x) − g(x) is constant for all x ∈ [a, b].
We can interpret this as f (b) is approximately equal to f (a), and the term (b − a)f 0 (c) is the error or
remainder in making this approximation.
(ii) Let b = a + h (h can be positive or negative) and c = a + θh, where 0 < θ < 1, then
f (a + h) = f (a) + hf 0 (a + θh).
Solution Let f (x) = ln x. We know that f (x) is continuous and differentiable for all x > 0 and that f 0 (x) = x1 .
Using the Mean Value Theorem in the form
f (b) − f (a)
= f 0 (c), then
b−a
ln(b) − ln(a) 1
= , where 0 < a < c < b. (∗)
b−a c
Now consider the function g(x) = x1 (that is, the derivative f 0 (x) of f ).
Now g 0 (x) = − x12 < 0, so g is a decreasing function, and since 0 < a < c < b,
1 1 1
< < (†)
b c a
Substituting into (∗) gives
1 ln (b/a) 1
< = .
b b−a a
Now write a = 1, b = 1 + x, (where x > 0), then
1 ln(1 + x)
< < 1, for x > 0
1+x x
x
or < ln(1 + x) < x.
1+x
(More than was asked for!)
Note that result (†) could be written down very easily for this problem, without adopting the above approach,
but the method indicated is useful for more complicated problems.
Theorem 9.23 (The Cauchy Mean Value Theorem) If f and g are continuous on [a, b] and are differentiable
in (a, b), then there is a number c ∈ (a, b) such that
10 L’Hôpital’s Rule
Theorem 10.1 (L’Hôpital’s Rule) If f (a) = 0, g(a) = 0 and f 0 (a), g 0 (a) both exist, and g 0 (a) 6= 0, then
f (x) f 0 (a)
lim = 0
x→a g(x) g (a)
Remarks
Theorem 10.2 If f (a) = 0, g(a) = 0 and f and g are differentiable in some neighbourhood of x = a, but not
necessarily at x = a itself, and g 0 (x) 6= 0 in this neighbourhood, except possibly at x = a, then
f (x) f 0 (x)
lim = lim 0
x→a g(x) x→a g (x)
ex −1
(ii) limx→0 x
(x−1)3
(iii) limx→1 ln x
eax −e−ax
(iv) limx→0 ln(1+x)
When attempting to use L’Hôpital’s rule (version 1), you may get f 0 (a) = 0, g 0 (a) = 0, in which case you need
to apply version 1 or 2 again, provided the limits exist.
x3 +x2 −x−1
(ii) limx→1 x2 +2x−3
51
EC119 Mathematical Analysis 10 L’HÔPITAL’S RULE
ln(sin x)
Example 10.5 limx→0 ln(sin 2x) .
Then f (x) = ln(sin x) and g(x) = ln(sin 2x) → −∞ as x → 0. Also f 0 (x) = cos x
sin x and g 0 (x) = 2 cos 2x
sin 2x .
Neither of these exist at x = 0, but we can solve as follows:
10.2 Limits as x → ±∞
1
These can be solved directly in the same way as in the preceding examples, or we could replace x by t and let
t → 0.
f (x)
= lim f 1t g 1t
lim provided the limits exist.
x→∞ g(x) t→0
x 1
(iii) limx→1 x−1 − ln x . (Type ‘∞ − ∞’.)
ln(1 − x) 1 . 1
lim = lim − 1 = lim − = −1.
x→0 x x→0 1−x x→0 1−x
Thus ln y → −1 as x → 0, or y → e−1 as x → 0.
11 Taylor’s Theorem
The following is a restatement of Theorem 9.18 on page 48, and is proved (as an exercise in Assignment 8) by
applying Rolle’s Theorem to an appropriate continuous and differentiable function.
Theorem 11.1 (First Mean Value Theorem) If f is continuous on [a, b] and differentiable on (a, b) then
there exists c ∈ (a, b) such that
Theorem 11.2 (Second Mean Value Theorem) If f and its first derivative is continuous on [a, b], and f and
is twice differentiable on (a, b) then there exists c ∈ (a, b) such that
(b − a)2 00
f (b) = f (a) + (b − a)f 0 (a) + f (c), where a < c < b
2!
In fact, these theorems also hold when b < a, so we could write
f (a + h) = f (a) + R1 ,
and f (a + h) = f (a) + hf 0 (a) + R2
2
where h could be positive or negative and the remainders are R1 = hf 0 (c), R2 = h2! f 00 (c).
c lies between a and a + h and as before, we could write c = a + θh, where 0 < θ < 1.
Alternatively we could write
f (x) = f (a) + R1
f (x) = f (a) + (x − a)f 0 (a) + R2
2
In this case, R1 = (x − a)f 0 (c) and R2 = (x−a)
2! f 00 (c), where c lies between a and x.
If we ignore the remainder term, we obtain a polynomial approximation T (x) to the function f (x) near x = a.
In the first case, the approximation is just a constant, and in the second, a linear approximation.
Theorem 11.4 (Taylor’s Theorem) Suppose f possesses derivatives of orders 1, 2, . . . , (n−1) on the closed
interval [a, b]. If
(i) f and its derivatives are all continuous on [a, b], and
(ii) f has a derivative of order n on (a, b)
then there is a number c ∈ (a, b), such that
53
EC119 Mathematical Analysis 11 TAYLOR’S THEOREM
Once again, the last term is the remainder. In this form it is called the Lagrange form of the remainder, that is,
we write
(b − a)n (n)
Rn = f (c).
n!
Remarks
(iii) If we ignore the remainder term, we obtain a polynomial approximation to the function f (x) near the
point x = a.
(iv) If a = 0, we obtain a special case of Taylor’s Theorem
h2 00 hn−1 (n−1)
f (a + h) = f (a) + hf 0 (a) + f (a) + · · · + f (a) + Rn
2! (n − 1)!
hn (n)
where Rn = n! f (a + θh), for 0 < θ < 1.
Example 11.5
(i) Find a polynomial approximation to f (x) = ex near x = 0, up to and including the term in x4 .
(ii) Estimate the remainder when x = 0.5, 1, 2.
Solution
(i) (a = 0) We have already obtained the first two approximations in Example 11.3. If f (x) = ex , it has
derivatives of all orders, and all its derivatives are ex .
Thus f (0) = f 0 (0) = f 00 (0) = f 000 (0) = f iv (0) = 1, and the fourth order polynomial approximation to ex is
x2 x3 x4
T4 (x) = 1 + x + + + .
2! 3! 4!
x5
(ii) The remainder is R5 = 5! eθx , for 0 < θ < 1.
Consider any fixed value of x, then eθx is an increasing function of θ on its domain (0,1).
x5 x 5 ex
When θ = 0, eθx = 1 and when θ = 1, eθx = ex so we can say that 5! < R5 < 5! .
Evaluating for different values of x:
x R5
0.5 2.6 × 10 < R5 < 4.3 × 10−4
−4
So we see that the further we are from x = 0, the greater the error.
It could well be that the series is convergent, but possibly only on some restricted interval of the real line, the
interval of convergence.
In addition, in order to establish convergence to f (x), we would need to show that for x and a in the interval
of convergence for Sn , the remainder Rn converges to 0 as n → ∞.
P∞
So first we should find the range of values for which the series n=0 Sn converges, then show that Rn → 0, for
these values of x, as n → ∞. This can get quite complicated and we shall not pursue this in any detail here.
Note that when a = 0, Taylor series are frequently called Maclaurin Series.
x2 x3 xn−1
exp(x) = 1 + x + + + ··· + + Rn
2! 3! (n − 1)!
xn eθx
where Rn = , 0 < θ < 1.
n!
One method for testing series for convergence is called the Ratio Test, (see Section 11.8 on page 61 for further
details). We compare two consecutive terms in the series as follows:
un+1 xn (n − 1)! |x|
un n! xn−1 = n
= .
As n → ∞, this tends to 0 < 1, regardless of the value of x. The test tells us that the series converges.
xn eθx
The remainder Rn = n! → 0 as n → ∞, so the series actually converges to the function exp(x) for all x.
This example illustrates how intervals of convergence may be found. They aren’t always the whole of R.
Example 11.7 Find the Taylor (Maclaurin) series for the following functions centred at x = 0.
x3 x5
sin x = x − + − ···
3! 5!
Once again it may be shown that this series converges for all x ∈ R.
(ii) f (x) = (1 + x)α , for α ∈ R.
11.3 Remarks
One series can be used to obtain another.
Example 11.8
∞
X 1
(−1)n xn = 1 − x + x2 − x3 + · · · = , |x| < 1
n=0
1+x
Example 11.9 Using a standard series, find the series expansion, and interval of convergence for
(2 + 3x)1/2 .
Convergent power series may be differentiated term by term to give a new convergent series whose sum is the
derivative of the sum of the original series.
11.3 Remarks 56
EC119 Mathematical Analysis 11 TAYLOR’S THEOREM
Example 11.10
∞
X 1
xn = 1 + x + x2 + · · · = , |x| < 1
n=0
1−x
∞
X 1
nxn−1 = 1 + 2x + 3x2 + · · · = , |x| < 1 etc.
n=1
(1 − x)2
Note that while the domain of both functions in this example is R \ {1}, the domain of the series is the interval
(−1, 1), so in each case, the series cannot be equal to the function outside (−1, 1).
Similarly, we may integrate a convergent power series, term by term, to give a new convergent series whose
sum is the integral of the sum of the original series. The interval of convergence remains the same.
Example 11.11
∞
X 1
(−1)n x2n = 1 − x2 + x4 − x6 + · · · = , |x| < 1
n=0
1 + x2
Integrating,
∞
X (−1)n x2n+1 x3 x5 x7
=x− + − + · · · = tan−1 x |x| < 1.
n=0
(2n + 1) 3 5 7
Series can be multiplied, and one series can be substituted into another.
(i) ex sin x
(ii) esin x
π
Example 11.13 Find the Taylor series for f (x) = sin x, about x = 2.
11.3 Remarks 57
EC119 Mathematical Analysis 11 TAYLOR’S THEOREM
∞
x2 x3 X xn
ex = 1 + x + + + ··· = , x∈R
2! 3! n=0
n!
∞
x3 x5 x7 X (−1)n x2n+1
sin x = x − + − + ··· = , x∈R
3! 5! 7! n=0
(2n + 1)!
∞
x2 x4 x6 X (−1)n x2n
cos x = 1 − + − + ··· = , x∈R
2! 4! 6! n=0
(2n)!
∞
x2 x3 x4 X (−1)n−1 xn
ln(1 + x) = x − + − + ··· = , x ∈ (−1, 1]
2 3 4 n=1
n
α(α − 1) 2 α(α − 1)(α − 2) 3
(1 + x)α = 1 + α x + x + x + ··· , |x| < 1, α ∈
/N
2! 3!
∞
X
(1 + x)−1 = 1 − x + x2 − x3 + · · · = (−1)n xn , |x| < 1
n=0
X∞
(1 + x)−2 = 1 − 2x + 3x2 − 4x3 + · · · = (−1)n (n + 1)xn , |x| < 1
n=0
X∞
(1 − x)−1 = 1 + x + x2 + x3 + · · · = xn , |x| < 1
n=0
∞
x2 x3 x4 X xn
ln(1 − x) = − x − − − − ··· = − , x ∈ [−1, 1)
2 3 4 n=1
n
∞
x3 x5 x7 X (−1)n x2n+1
tan−1 x = x − + − + ··· = , |x| < 1
3 5 7 n=0
2n + 1
1 x3 1 3 x5 1 3 5 x7
sin−1 x = x + · + · · + · · · + ··· , |x| < 1
2 3 2 4 5 2 4 6 7
1 2 17 7 62 9 π
tan x = x + x3 + x5 + x + x + ··· , |x| <
3 15 315 2835 2
20
15
10
–3 –2 –1 1 2 3
x
y 1
–10 –5 0 5 10
x
–1
–2
Figure 11.2: Graphs of successive polynomial approximations to f (x) = sin(x), Tn for n = 1, 3, 5, 7, 13, 21
10
6
y
4
–1 0 1 2 3 4
–2 x
–4
–6
–8
–10
Figure 11.3: Graphs of successive polynomial approximations to f (x) = ln(1 + x), Tn for n = 1, 2, 3, 4, 7, 10
11.5 Sequences
Definition 11.15 A sequence is a function whose domain is the integers or a subset of the integers (n ∈ A ⊆
Z).
Example 11.16
In general, if un tends to a finite limit as n → ∞, then we say that the infinite sequence converges.
11.6 Series
11.5 Sequences 59
EC119 Mathematical Analysis 11 TAYLOR’S THEOREM
S1 = u1 , S2 = u1 + u2 , S3 = u1 + u2 + u3 , . . . Sn = u1 + u2 + u3 + · · · + un . . .
If this sequence S1 , S2 , . . . , Sn , . . . tends to a finite limit as n → ∞, then we say that the infinite series
converges.
The value of this limit is called the sum of the infinite series, that is,
n
X
lim Sn = un = S.
n→∞
i=1
Example 11.18 (The geometric sequence and series) Consider the familiar geometric sequence
1, x, x2 , x3 , . . . xn−1 , xn , . . .
You have seen that this sequence converges when −1 < x 6 1. For this sequence we can find a formula for Sn ,
the sum of the first n terms, as follows:
Sn = 1 + x + x2 + x3 + · · · + xn−1
multiplying throughout by x: xSn = x + x2 + x3 + · · · + xn−1 + xn
(1 − xn )
Sn = , provided that x 6= 1.
1−x
What happens as n → ∞?
1
(i) If |x| < 1, Sn → 1−x : Convergent.
So the series only converges when |x| < 1, and in this case we can write
∞
X 1
S = 1 + x + x2 + x3 + · · · + xn + · · · = xn = = (1 − x)−1 , |x| < 1.
n=0
1−x
11.6 Series 60
EC119 Mathematical Analysis 11 TAYLOR’S THEOREM
So we see that although the geometric sequence converges when x = 1, (converges to 1), the series certainly
does not converge for this value of x.
The converse is not true, as illustrated in the next example.
Example 11.19
1 1 1
(i) 1 + 2 + 4 + 8 + · · · is convergent. This is a geometric series, with x = 12 .
1 1 1 1
(ii) 1 + 2 + 3 + 4 + · · · is not convergent even though un = n → 0 as n → ∞.
This series is called the harmonic series.
P∞
Given the series un and limn→∞ uun+1 = `, then:
n=1 n
(ii) if ` > 1 or if the ratio → ∞ (that is, no finite limit ` exists) then the series diverges,
(iii) if ` = 1, no conclusion can be drawn, and the test fails.
12 Integration
Contrary to the customary order of presentation: starting with differentiation and later considering integration,
the ideas of the integral calculus were in fact developed first3 . The original ideas were noted by Archimedes in
223 BC, but the majorR initial contributions were made by Leibniz sometime between 1673 and 1676. He was
the first to use the notation. (Of course, Newton was involved as well).
There are many types of integrals, Riemann, Riemann–Stieltjes, Lebesgue, line integrals, double and other
multiple integrals, improper integrals of various kinds, and so forth. but all can be interpreted as a generalisation
of area. Here, we shall develop the theory in the usual way. First we consider indefinite integrals, then definite
integrals, the connection between differentiation and integration, and finally, improper integrals.
d
Since dx (C) = 0, we should always add an arbitrary constant C to the results above.
R R R
(i) f + g = f + g.
R R
(ii) kf = k f , where k ∈ R is a scalar constant.
Example 12.1
(i) (3 − 4x)1/5 dx
R
(ii) x2x+4 dx
R
63
EC119 Mathematical Analysis 12 INTEGRATION
In Leibniz notation:
Z Z
dv du
u dx = uv − v dx.
dx dx
Typically, this method is used on integrals of the following forms:
Z Z Z
xn eax dx xn cos(ax) dx, xn sin(ax) dx n∈N
Z Z
ax
e cos(bx) dx, eax sin(bx) dx
Essentially we are trying to rearrange the integral in such a way as to end up with an easier integral than we
dv
began with. This usually means that we have to choose u and dx carefully. The method is also used to evaluate
Z
xα ln x dx, (α 6= −1),
sin−1 x dx, and also some integrals which can be solved using alternative methods.
R R
including ln x dx and
Example 12.3
(i) x2 e−5x dx
R
R
(ii) x cos(2x) dx
(iii) e−x sin x dx
R
xα ln x dx
R
12.3.4 Integrals of the form
dv du xα+1
Let u = ln x, dx = xα , dx = x1 ,
α 6= −1, and the integral becomes
v= α+1 ,
xα+1 xα+1
Z Z
1
· ln x − dx = xα+1 ln x − xα dx
(α + 1) x(α + 1) (1 + α)
xα+1
1
= xα+1 ln x − + C, for all α 6= −1.
(1 + α) (α + 1)
Now consider the case α = −1: Z
ln x
dx.
x
du
Let u = ln x, then dx = x1 , so we use substitution, not parts, and the integral is
Z
u du = 12 u2 + C = 12 (ln x)2 + C.
R
Surprisingly, integration by parts still works for the case α = 0: ln x dx.
dv
Let u = ln x, dx = 1, du 1
dx = x , v = x, and
Z Z
ln x dx = x ln x − dx = x ln x − x + C.
If u = sin−1 x, dx
dv
= 1, dudx =
√ 1
1−x2
, v = x, and
Z Z
x p
sin−1 x dx = x sin−1 x − √ dx = x sin−1 x + 1 − x2 + C.
1−x 2
and so on.
a b
b−a
Divide the interval [a, b] into n equal subintervals of width h = n .
Sum the areas of the rectangles indicated above:
Pn n(n+1)
Now i=1 i= 2 , (previously proved by mathematical induction), so
(b − a)2 n(n + 1)
Sn = (b − a)a + ·
n2 2
(b−a)2
and as n → ∞, Sn → (b − a)a + 2 = 12 (b − a)(2a + b − a) = 21 (b2 − a2 )
We call the length of the general interval [xi , xi+1 ], ∆xi and choose any number inside the interval ti . We then
form the sum
n−1
X
Sn = f (t0 )∆x0 + f (t1 )∆x1 + · · · + f (tn−1 )∆xn−1 = f (ti )∆xi .
i=0
Sn obviously depends on the function f , on the form of the subdivision, and on the choice of ti .
Assume n → ∞, and simultaneously the largest of the subintervals ∆xi → 0 and that Sn → a limit S, then f
is said to be Riemann-integrable. It can be shown that every continuous function is Riemann-integrable.
So in addition to continuous functions, this definition of integration also allows us to deal with functions which
are not differentiable, and those which are piecewise continuous (functions which have a finite number of finite
continuities).
Example 12.5
R4
(i) f (x) = |x| is not differentiable at x = 0, but the integral −1 |x| dx exists.
−1, x < 0 R1
(ii) f (x) = sgn x = is not continuous at x = 0, but −2 sgn x dx exists.
1, x > 0
R1 1
(iii) f (x) = x1 . 0 x
dx does not exist because f does not exist at x = 0, ie f is not bounded on [0, 1].
Theorem 12.6 If f is continuous on [a, b] then there is a number c ∈ (a, b) such that
Z b
f (x) dx = (b − a)f (c).
a
(ii) Furthermore, if f is continuous on [a, b], then F is differentiable, and F 0 = f . In this case,
Z b
f (x) dx = F (b) − F (a).
a
RConsidering
∞
type (3), if f is Riemann-integrable on (a, b) for all b > a, then we define the infinite integral
a
f (x) dx, to be the following limit, if it exists:
Z ∞ Z b
f (x) dx = lim f (x) dx
a b→∞ a
If the limit exists and is equal to ` say, we say that the infinite integral converges to `.
If the limit does not exist, we say that the integral diverges.
There are various tests, not gone into here, which enable us to establish convergence, without actually finding
the limit; but we shall not adopt this approach, but rather, work out limiting values for some infinite integrals.
Example 12.8
(i) f (x) = λe−λx , where x > 0 and λ > 0. This is the exponential distribution.
Show that the area under the graph of f over [0, ∞) is 1.
Suppose that b > 0.
Z b b
λe−λx dx = −e−λx 0 = 1 − e−λb
0
−λb
As b → ∞, e → 0, so the integral is convergent to 1.
R∞ 1
(ii) 0 1+x2 dx.
R∞
(iii) 1 x1α dx. Consider
Z b −α+1 b
1 x 1
b1−α − 1 ,
I= α
dx = =
1 x 1 − α 1 (1 − α)
where b > 1 and α 6= 1. Now let b → ∞.
1
If α > 1, then b1−α → 0 and the integral converges to α−1 .
If α < 1, then b1−α → ∞ and the integral diverges.
Rb b
If α = 1, the integral is 1 x1 dx = [ln x]1 = ln b − ln 1 = ln b.
This tends to infinity as b → ∞, so the integral diverges.
Rb
Type (4) −∞
f (x) dx is similar to type (5) above. We define
Z b Z b
f (x) dx = lim f (x) dx
−∞ a→−∞ a
for some real number c. So the integral on the left only exists when both the integrals on the right exist. (c is
frequently taken to be 0, but this isn’t necessary).
R∞ 2
Example 12.12 −∞
xe−cx dx, for c > 0.
provided both limits exist. This result can be extended to functions which have a finite number of infinite
discontinuities.
Example 12.13
R1 1
(i) 0 √1−x 2
dx.
R1
(ii) 0
x−α dx.
Example 12.14
and there are many more (such as the elliptic functions F , E and Π).
R∞
It is an astonishing fact that improper integrals 0 f (t) dt can often be calculated where ordinary integrals
Rb
a
f (t) dt cannot. The following example makes use of techniques which are beyond the scope of this course.
Rb 2 R∞ 2
Example 12.15 There is no elementary formula for a e−t dt, but the value of 0 e−t dt can be calculated
precisely. The usual technique involves squaring to get the double integral
Z ∞ 2 Z ∞ Z ∞ Z ∞ Z ∞
−t2 −x2 −y 2 2
+y 2 )
e dt = e e dx dy = e−(x dx dy
0 0 0 0 0
(x − µ)2
1
f (x) = √ exp − for x ∈ (−∞, ∞).
σ 2π 2σ 2
Verify that f does indeed represent a probability density function, and find the mean and variance of the
distribution.
(i) An ordinary differential equation is an equation containing the derivatives of a function of a single
variable.
(ii) A partial differential equation is one where the unknown function is a function of several variables, so
the equation contains partial derivatives.
(iii) The order of a differential equation is the order of the highest order derivative within the equation.
(iv) A differential equation is said to be linear if it does not contain products, quotients, or powers of the
derivatives, or of the unknown function.
We will use a variety of notations. We should choose the one which is most appropriate to the context, or to
the situation being modelled. For example:
dy
dx = f (x, y) or y 0 = f (x, y) : solve for y
dx
dt = g(t, x) or ẋ = g(t, x) : solve for x, etc.
˙ means ‘differentiate with respect to time t’ .
13.1 Qualitative approach: ‘knowing the direction and finding the path’
Most differential equations do not have solutions that can be found in a neat closed form, so instead, numerical
techniques have to be used. Even when analytical solutions can be found, it is sometimes useful to have a
qualitative picture of the behaviour of possible solutions. Such an approach is illustrated in the following
example.
dy
Example 13.1 dx = x + y.
This can be solved analytically (see later), but we can obtain a qualitative picture of the behaviour of y as
follows.
We can obtain a direction diagram or gradient field, because at each point of the x, y plane, we know that the
dy
gradient dx is equal to x + y. For example, at (1, 2), the gradient is 3, at (−1, 0) the gradient is −1, and so on.
We represent the gradient at the point (x, y) by a short line segment drawn at the point. When the picture is
complete, we have obtained our gradient field.
The line segments give us a means of drawing curves that follow the direction of the gradient field, because we
know that they are tangential to the solution curves or integral curves y = y(x). In applied subjects they are
also called trajectories or time paths. Generally, there will be an infinite number of them.
Through each point (x, y) will pass one and only one solution curve, so we can determine the precise solution
of our differential equation provided we know a single point on it. In particular, if we know the value of y
when x = 0, we can find the appropriate integral curve. Such a condition is called an initial condition. The
problem of solving the differential equation with a given initial condition is called an initial value problem.
See Figure 13.1 for solution curves corresponding to two sets of initial conditions.
71
EC119 Mathematical Analysis 13 FIRST-ORDER DIFFERENTIAL EQUATIONS
10 10
5 5
y(x) 0 y(x) 0
–5 –5
–10 –10
–10 –5 0 5 10 –10 –5 0 5 10
x x
dy
Figure 13.1: Two solution curves for the equation dx = x + y, for initial values y = 3.5, x = −5.0 (left) and
y = 4.1, x = −5.0 (right)
10 10
5 5
y(x) 0 y(x) 0
–5 –5
–10 –10
–10 –5 0 5 10 –10 –5 0 5 10
x x
dy
Figure 13.2: Two solution curves for the equation dx = xy, for initial values y = 2, x = −2 (left) and y = −3,
x = −2 (right)
Example 13.3 t2 dy 6 0. Consider the left hand side and recall the product rule for
dt + 2ty = b(t) for t =
differentiation:-
d dv du
dt (uv) = u dt + v dt .
We see that the left hand side of the differential equation may be expressed as
d 2
dt (t y).
(v) Integrate R
u(t)y = u(t)b(t) dx + C,
1
R
so y = u(t) u(t)b(t) dx + C .
13.3.2 Verification
If R
u(t)y = exp a(t) dt ,
then
du
R
dt = a(t) exp a(t) dt = a(t)u(t).
So
d
dt u(t)y = u(t)b(t)
becomes
u(t) dy
dt = a(t)u(t)y = u(t)b(t),
for u(t) 6= 0, or
dy
dt + a(t)y = b(t).
dx
Example 13.4 dt = xt.
Example 13.7 Consider the differential equation in Example 13.1. Rearrange it to the standard linear form
dy
dx − y = x.
The integrating factor is u(x) = exp − dx = e−x , so the differential equation becomes dx d
e−x y = xe−x .
R
Integrating, we get e−x y = xe−x dx = −xe−x − e−x +C = −e−x (1+x)+C, so that y = −(1+x)+Cex . This is
R
Example 13.8 (A special case) As in the previous example, a linear differential equation which is frequently
encountered in applications is
dy
dt + ay = b,
where a and b are constants. This can be solved in several ways as we shall see later, but for now we’ll use the
integrating factor method.
The integrating factor is exp a dt = eat . Thus
R
d at
e y = beat
dt
b
eat y = eat + C
a
b
y = Ce−at +
a
Note that the arbitrary constant need only be added to one side of the equation.
If possible, evaluate both integrals, and again if possible express the result in the explicig form y = F (x). In
this case, F (x) is the general solution of the differential equation.
dy
In addition, we have to consider separately what happens when g(y) = 0. In this case, dx ≡ 0 and there will be
solutions y = constant for each value of y for which g(y) = 0.
If we know any one point on the time path, any arbitrary constant may be determined.
dx
Example 13.9 dt = xt. Rearranging this gives
Z Z
1
dx = t dt.
x
1 2
Assume first that x > 0, then ln x = 12 t2 + C, so that x = exp 21 t2 + C , which can be written x = Ae 2 t for
some A > 0.
1 2
Alternatively, if x < 0, then the solution is ln(−x) = 12 t2 + D. This leads to x = −Be 2 t for some B > 0.
Both cases are covered by the general solution x = A exp 12 t2 ; whether A > 0 or A < 0 depends on the initial
value.
Note We have already solved this equation using the integrating factor method.
1
Assuming that ay > b, we integrate to give a ln(ay − b) = −t + C, that is,
ay − b = e−at+K = Ae−at
Wt = (1 + r)Wt−1 + Yt − Ct
where Wt is the size of the account (wealth), r is the interest rate, Yt the amount deposited, and Ct the amount
withdrawn.
The continuous-time analogue of this is
dW
= rW (t) + Y (t) − C(t),
dt
where r is constant.
d2 y
dy
=F x, y, or ẍ = F (t, x, ẋ).
dx2 dx
originally studied by Adrien-Marie Legendre (1752–1833) gives rise to a family of functions Pn (x), called
Legendre polynomials, the first few of which are:
P0 (x) = 1
P1 (x) = x
P2 (x) = 12 (3x2 − 1)
P3 (x) = 12 (5x3 − 3x)
P4 (x) = 18 (35x4 − 30x2 + 3)
P5 (x) = 18 (63x5 − 70x3 + 15x)
d2 y dy
x2 2
+x + (x2 − α2 )y = 0
dx dx
was originally studied by Daniel Bernoulli (1700–1782) and Friedrich Bessel (1784–1846) and has no solutions
in terms of elementary functions. Instead, we define a new class of functions (Bessel functions) to be the
solutions of this equation.
d2 x
Example 14.3 The equation dt2 = k, where k is some given constant, can be integrated directly.
dx
Integrating once gives dt= kt + C; integrating again gives x = 12 kt2 + Ct + D., where C and D are two
arbitrary constants.
In order to determine them, we need two pieces of information about x. These could be:
(i) Two points through which the solution curve passes. This is called a boundary value problem.
dx
(ii) x and the gradient function dt at the same point. This is called an initial value problem.
(iii) If we know x at one point and the gradient at another, this is called a mixed problem.
d2 x dx
2
= + t, (or ẍ = ẋ + t), x(0) = 1, ẋ(0) = 2.
dt dt
dx
The easiest approach is to replace dt by a new variable, y say, then the equation becomes a first order one
ẏ = y + t, with y(0) = 2.
77
EC119 Mathematical Analysis 14 SECOND-ORDER DIFFERENTIAL EQUATIONS
Adding these gives L(u1 +u2 ) = 0, so (u1 +u2 ) is also a solution of (9).
Similarly, if u is a solution of (9) so that Lu = 0, and k ∈ R then L(ku) = 0 and ku is also a solution.
V is a vector space, consisting of all the vectors (ie functions) which are mapped to the zero function by the
linear map L.
We can combine the two properties above and say that Au1 +Bu2 is a solution of (9) for all choices of the
(real) arbitrary constants A and B.
Is the general solution of (9) always the form Au1 + Bu2 ?
No, only when u1 and u2 span the solution space of (9) and are linearly independent. In other words, they
must form a basis for the solution space of (9).
d2 y dy
a 2
+b + cy = 0. (11)
dt dt
2
If we try the solution y = emt , then dy
dt = me
mt
and ddt2y = m2 emt .
Substitute into the homogeneous differential equation (11):
These are linearly independent functions, so the complementary solution is a linear combination of them, viz.
Aem1 t + Bem2 t ,
u1 = emt .
Clearly the second solution u2 cannot also be emt , because u1 and u2 must be linearly independent.
The second solution is u2 = temt . This can be verified by substitution into the differential equation. The
complementary solution is
Aemt + Btemt or emt (A + Bt),
where A and B are again arbitrary.
Case 3 (m1 , m2 = α ± iβ complex conjugate, ie (b2 < 4ac))
u1 and u2 are linearly independent, but they are complex, and are not really suitable choices for solutions to a
real problem. Instead we choose two real linearly independent solutions from the complex subspace spanned by
u1 and u2 , viz. eαt cos βt and eαt sin βt.
That is,
u1 + u2 u1 − u2
and .
2 2
The general solution is thus
eαt (A cos βt + B sin βt).
Summary
Examples 14.6
(i) ẍ − 4x = 0
(ii) ẍ − 4ẋ + 4x = 0
(iii) ẍ − 6ẋ + 13x = 0
In each of the above cases, determine what happens in the long run.
17
Example 14.8 Solve ẍ + 5ẋ + 6x = 3e2t subject to x(0) = 0 and ẋ(0) = 20 .
14.4 Stability
This is a very important consideration when dealing with dynamical systems.
If small changes in the initial conditions have no effect on the long run behaviour of the solution, the system
is said to be stable. If small changes in initial conditions can lead to significant differences in the long run
behaviour then the system is unstable.
Recall that the solution is of the form
This will be stable if the complementary solution tends to 0 as t → ∞ for all values of A and B; in this case x
tends to the particular solution as t → ∞, which is independent of the initial conditions.
Thus the equation will be stable in the three cases considered above if
14.4 Stability 81
EC119 Mathematical Analysis REFERENCES
References
[1] R P Burn, Numbers and Functions, Cambridge (2000) QA 300.B8
REFERENCES 82