Professional Documents
Culture Documents
VI Sem Maths Real Analysis
VI Sem Maths Real Analysis
STUDY MATERIAL
B.Sc. Mathematics
VI SEMESTER
CORE COURSE
(2011 Admission)
UNIVERSITY OF CALICUT
SCHOOL OF DISTANCE EDUCATION
THENJIPALAM, CALICUT UNIVERSITY P.O., MALAPPURAM, KERALA - 673 635
357
School of Distance Education
UNIVERSITY OF CALICUT
SCHOOL OF DISTANCE EDUCATION
Study Material
B.Sc.Mathematics
VI SEMESTER
CORE COURSE
REAL ANALYSIS
Prepared by :
Sri.Nandakumar,
Assistant Professor,
N.A.M College,
Kallikkandi, Kannur
Scrutinised by :
Dr. Anilkumar V.
Reader,
Department of Mathematics,
University of Calicut
Contents
1
CONTINUOUS FUNCTIONS ON INTERVALS
Definitions A function f : A is said to be bounded on A if there exists a constant M 0
such that
f ( x) M for all x A .
That is, a function f : A is bounded on the set A if its range f ( A) is a bounded set in .
f : A is unbounded on A, if f is not bounded on A.
i.e., f is unbounded on A if given any M 0 , there exists a point xM A such that
f (x M ) M .
Example We now give an example of a continuous function that is not bounded. Consider the
function f defined on the interval A (0, ) by
1
f ( x)
x
1
is not bounded on A because for any M 0 we can take the point xM in A to get
M 1
1
f ( xM ) M 1 M . However, being the quotient two continuous functions 1 and x, f is
xM
continuous on A.
We now review some basic definitions and theorems.
Cluster Point: Let A . A point c is a cluster point of A if for every 0 there exists
at least one point x A, x c such that | x c | .
i.e., A point c is a cluster point of A if for every -neighborhood V (c) (c , c )
of c contains at least one point of A distinct from c.
The cluster point c may or may note be a member of A, but even if it is in A, it is ignored
when deciding whether it is a cluster point of A or not, since we explicitly require that there be
points in V (c) A distinct from c in order for c to be a cluster point of A
Example We now show that A {1, 20, 323}, the set consisting of three elements 1, 20 and 323,
has no cluster point.
1
The point 1 is not a cluster point of A, since choosing gives a neighborhood of 1 that
2
contains no points of A distinct from 1. The same is true for the point 20 and 323. Also, a real
number other than 1, 20 and 323 cannot be a cluster point. For, if c with c 1, c 20, and
c 323, then choose min c 1 , c 20 , c 323 . Then V (c) A . Hence A has no
cluster point.
Theorem A number c is a cluster point of a subset A of if and only if there exists a
sequence (an ) in A such that lim(an ) c and an c for all n .
Example Let A be the open interval A (2, 3). All the points of A are cluster points of A. Also
the points 2, 3 are cluster points of A, though they do not belong to A. Hence every point of
the closed interval [0, 1] is a cluster point of A.
Example We now show that A finite set has no cluster point.
Let F be a finite set. No number c is a cluster point of F as it is not possible to find a
sequence (an ) in F such that lim(an ) c and an c for all n .
Example The infinite set has no cluster point. No number c is a cluster point of as it
is not possible to find a sequence (an ) in such that lim(an ) c and an c for all n .
Example The set B 1n : n has only the point 0 as a cluster point.
Example If I [0, 1], then the set C I consists of all the rational numbers in
I . It follows from Density Theorem (“If x and y are real numbers with x y ,
then there exists a rational number r such that x r y ”) that every point in I is a
cluster point of C.
Definition Let A , and let c be a cluster point of A. For a function f : A , a real
number L is said to be a limit of f at c if, for every number 0 , there exists a
corresponding number 0 such that if x A and 0 | x c | , then | f ( x) L | . In that
we write lim f ( x) L.
x c
(ii) Given any -neighborhood V ( L) of L, there exists a -neighborhood V (c) of c such that
if x c is any point in V (c) A, then f ( x) belongs to V ( L).
Important Limits of Functions
(1) lim k k . (2) lim x c.
x c x c
1 1
(3) lim x 2 c 2 . (4) lim if c 0.
x c x c x c
The following important formulation of limit of a function is in terms of limits of sequences.
Sequential Criterion for Limits: Let f : A and let c be a cluster point of A. Then the
following statements are equivalent.
(i) lim f ( x) L.
x c
(ii) For every sequence ( xn ) in A that converges to c such that xn c for all n , the sequence
( f ( xn )) converges to f (c) .
Divergence Criterion: Let A , let f : A and let c be a cluster point of A.
(a) If L , then f does not have limit L at c if and only if there exists a sequence ( xn ) in A
with xn c for all n such that the sequence ( xn ) converges to c but the sequence
( f ( xn )) does not converge to f (c) .
(b) The function f does not have a limit at c if and only if there exists a sequence ( xn ) in A
with xn c for all n such that the sequence ( xn ) converges to c but the sequence
( f ( xn )) does not converge in .
Remark The assertion (a) in theorem enables us to show that a certain number L is not the
limit of a function at a point, while (b) says that the function does not have a limit at a point.
We list some examples, which works on the basis of Divergence Criteria. The details are
avoided as it is not mentioned in the syllabus.
1 1
Example If f ( x) ( x 0), then lim f ( x) lim doesnot exist in .
x x 0 x 0 x
Properties of Limits
The following rules hold if lim x c f ( x) L and lim x c g ( x) M (L and M real numbers).
1. Sum Rule: lim[ f ( x ) g ( x )] L M
x c
i.e., the limit of the sum of two functions is the sum of their limits.
2. Difference Rule: lim[ f ( x ) g ( x )] L M
x c
i.e., the limit of the difference of two functions is the difference of their limits.
3. Product Rule: lim f ( x) g ( x) L M
x c
i.e., the limit of the product of two functions is the product of their limits.
4. Constant Multiple Rule: lim kf ( x ) kL (any number k)
x c
i.e., the limit of a constant times a function is that constant times the limit of the function.
f ( x) L
5. Quotient Rule: lim , M 0
x c g ( x) M
i.e., the limit of the quotient of two functions is the quotient of their limits, provided the
limit of the denominator is not zero.
6. Power Rule: If m and n are integers, then
m
lim[ f ( x)] n Ln , provided Ln is a real number.
m m
x c
i.e., the limit of any rational power of a function is that power of the limit of the function,
provided the latter is a real number.
Definition Let A , let f : A , and let c A . We say that f is continuous at c if, for
every number 0 there exists 0 such that if x is any point of A satisfying | x c | ,
then | f ( x) f (c) | .
Definition If f fails to be continuous at c , then f is discontinuous at c .
Similar to the definition of limit, the definition of continuity at a point can be formulated
very nicely in terms of neighborhoods. .
Theorem A function f : A is continuous at point c A if and only if given any -
neighborhood V ( f (c)) of f (c) there exists a -neighborhood V (c) of c such that if x is any
point of A V (c), then f ( x) belongs to V ( f (c)), that is
f ( A V (c)) V ( f (c)).
If c A is a cluster point of A, then a comparison of Definitions 2 and 3 show that f is
continuous at c if and only if
f (c) lim f ( x). …(1)
x c
Thus, if c is a cluster point of A, then three conditions must hold for f to be continuous at c :
(i) f must be defined at c (so that f (c) makes sense),
(ii) the limit of f at c must exist in (so that lim f ( x) makes sense), and
x c
If c A is not a cluster point of A, then there exists a neighborhood V (c) of c such that
A V (c) {c}. Thus we conclude that f is continuous at a point c A that is not a cluster
point of A. Such points are often called isolated points of A. They are of little practical
interest to us, since they have no relation to a limiting process. Since continuity is automatic for
such points, we generally test for continuity only at cluster points. Thus we regard condition
(1) as being characteristic for continuity at c .
Similar to Sequential Criterion for Limits (Theorem 4), we have the Sequential Criterion for
Continuity.
Sequential Criterion for Continuity: A function f : A is continuous at the point c A if
and only if for every sequence ( xn ) in A that converges to c, the sequence ( f ( xn )) converges to
f (c ) .
Discontinuity Criterion: Let A , let f : A , and let c A . Then f is discontinuous at
c if and only if there exists a sequence ( xn ) in A such that ( xn ) converges to c, but the sequence
( f ( xn )) does not converge to f (c) .
1
Example f ( x) is not continuous at x 0.
x
The signum function sgn is not continuous at x 0.
Definition Let A , and let f : A . If B is a subset of A, we say that f is continuous
on the set B if f is continuous at every point of B.
exist.
Example The function f ( x) x sin(1/ x) for x 0 is not continuous at 0 as f ( x) is not defined at
0.
Example Given the graph of f(x), shown below, determine if f(x) is continuous at
x 2, x 0, x 3.
Solution
To answer the question for each point we’ll need to get both the limit at that point and the
function value at that point. If they are equal the function is continuous at that point and if
they aren’t equal the function isn’t continuous at that point.
First x 2 .
The function value and the limit aren’t the same and so the function is not continuous at this
point. This kind of discontinuity in a graph is called a jump discontinuity. Jump
discontinuities occur where the graph has a break in it as this graph does.
Now x 0. .
The function is continuous at this point since the function and limit have the same value.
Finally x 3. .
f (3) 1 and lim f ( x ) 0
x 3
The function is not continuous at this point. This kind of discontinuity is called a removable
discontinuity. Removable discontinuities are those where there is a hole in the graph as there
is in this case.
Example The function
x sin(1/ x) for x 0,
F ( x)
0 for x 0,
is continuous at 0.
Combinations of Continuous Functions:
Let A and let f and g be functions that are defined on A to and let c A . If k is a
constant and f and g are continuous at c, then
(i) f g , f g, fg , and kf are continuous at c.
(ii) f / g is continuous at c, provided g ( x) 0 for all x A.
The next result is an immediate consequence of Theorem 10, applied to every point of A .
Theorem Let A and let f and g be continuous on A to . If k is a constant and f and g
are continuous on A , then
(i) f g , f g , fg , and kf are continuous on A .
(ii) f / g is continuous on A , provided g ( x) 0 for all x A.
Examples A polynomial function is continuous on .
Rational function r defined by
p( x)
r ( x) ,
q( x)
as the quotient of two polynomials p( x) and q ( x) , is continuous at every real number
for which it is defined.
The trigonometric functions sine and cosine functions are continuous on .
The trigonometric functions tan, cot, sec, and csc are continuous where they are defined.
For instance, the tangent function is defined by
sin x
tan x
cos x
provided cos x 0 (that is, provided x (2n 1) / 2, n ). Since sine and cosine functions are
continuous on , it follows that the function tan is continuous on its domain.
Theorem Let A , let f : A , and let f be defined by
f ( x) f ( x) for x A.
(a) If f is continuous at point c A, then f is continuous at c.
(a) If f is continuous on A, then f is continuous on A.
Example Let g ( x) x for x . Then g is continuous on .
Continuity of Composites:
If f is continuous at c , and g is continuous at f (c) , then g f is continuous at c
1
Example Let g ( x) sin x for x and f ( x) for x 0. Then, being the composition of the
x
continuous functions g and f , ( g f )( x) sin(1/ x ) is continuous at every point c 0.
Theorem If X ( xn ) is a convergent sequence and if a xn b for all n , then
a lim( xn ) b .
Sequential Criterion for Continuity A function f : A is continuous at the point c A if
and only if for every sequence ( xn ) in A that converges to c, the sequence ( f ( xn )) converges to
f (c ) .
Hence if f is continuous at x, then
lim xn x lim f ( xn ) f ( x )
n n
Equivalently,
lim xn x lim f ( xn ) f (lim xn )
n n n
converges to f ( x) .
Since a convergent sequence of real numbers is bounded, the convergent sequence f (x )
nr
f ( x* ) f ( x) for all x A .
We say that f has an absolute minimum on A if there is a point x* A such that
f ( x ) f ( x ) for all x A
We say that x* is an absolute maximum point for f on A, and that x* is an absolute minimum
point for f on A, if they exist.
Example We now give an example to show that a continuous function on set A does not
necessarily have an absolute maximum or an absolute minimum on the set.
1
The function f ( x) has neither an absolute maximum nor an absolute minimum on the set
x
A (0, ) (Fig. 1). There can be no absolute maximum for f on A since f is not bounded above on
A. f can have no absolute minimum as there is no point at which f attains the value
0 inf{ f ( x) : x A} . The same function has neither an absolute maximum nor an absolute
minimum when it is restricted to the set (0, 1), while it has both an absolute maximum and an
1
absolute minimum when it is restricted to the set [1, 2] . In addition, f ( x)
has an absolute
x
maximum but no absolute minimum when restricted to the set [1, ) , but no absolute maximum
and no absolute minimum when restricted to the set (1, ) .
Example We now give an example to show that if a function has an absolute maximum point,
then this point is not necessarily uniquely determined.
Consider the function g ( x) x 2 defined for x A [1, 1] has two points x 1 giving the
absolute maximum on A, and the single point x 0 yielding its absolute minimum on A.
Example The constant function h( x) 1 for x is such that every point of is both an
absolute maximum and an absolute minimum point for h.
Theorem (Maximum-Minimum Theorem) Let I [a, b] be a closed bounded interval and let
f : I be continuous on I. Then f has an absolute maximum and an absolute minimum on I.
To prove the result, let
f ( I ) { f ( x) : x I }.
i.e., f ( I ) is the nonempty set of values of f on I.
Since f is continuous and I is closed and bounded, using Boundedness Theorem, we have
f ( I ) is a bounded subset of . Then f ( I ) is bounded above and bounded below. Hence by the
completeness property of , supremum and infimum of the set exist.
Let
s* sup f ( I ) and s* inf f ( I ) .
We claim that there exist points x* and x* in I such that s* f ( x* ) and s* f ( x* ) .
We only establish the existence of the point x* , as the proof of the existence of x* is similar.
1
Since s * sup f ( I ) , if n , then the number s* is not an upper bound of the set f ( I ).
n
Consequently there exists a number xn I such that
1
f ( xn ) s* for all n .
s* … (2)
n
By this way we obtain a sequence X ( xn ) with members in I. Also, since I is bounded, the
sequence X ( xn ) is bounded. Therefore, by the Bolzano-Weierstrass Theorem, there is
subsequence X ( xnr ) of X that converges to some number x* . Since the elements of X belong
to I [a, b] , we have a xnr b and hence it follows from Theorem A that
a lim xnr b .
That is,
a x* b
and hence x* I . Therefore f is continuous at x* and, by Sequential Criterion for Continuity, it
follows that
lim f ( xnr ) f ( x* ) .
Since it follows from (2) that
1
s* f ( xnr ) s* for all r ,
nr
we conclude from the Squeeze Theorem that2
lim( f ( xnr )) s* .
Therefore we have
f ( x* ) lim f ( xnr ) s* sup f ( I ) .
We conclude that x* is an absolute maximum point of f on I. This completes the proof.
Remark With the assumptions of the theorem, s* f ( x* ) f ( I ) and s* f ( x* ) f ( I )
Theorem (Location of Roots Theorem) Let I [a, b] and let f : I be continuous on I. If
f (a ) 0 f (b) , or if f (a) 0 f (b) , then there exists a number c (a, b) such that f (c) 0 .
To prove the result, we assume that f (a ) 0 f (b) . We will generate a sequence of intervals by
successive bisections.
Let I1 [a1 , b1 ] ,
where a1 a, b1 b , and let p1 be the midpoint
1
p1 (a1 b1 ) .
2
If f ( p1 ) 0 , we take c p1 and we are done.
If f ( p1 ) 0 , then either f ( p1 ) 0 or f ( p1 ) 0 .
If f ( p1 ) 0 , then we set a2 a1 , b2 p1 ,
while if f ( p1 ) 0, then we set a2 p1 , b2 b1 .
We continue the bisection process. Suppose that the intervals I1 , I 2 , ... , I k have been
obtained by successive bisection in the same manner. Then we have f (ak ) 0 and f (bk ) 0 ,
and we set
ak bk
pk .
2
If f ( pk ) 0 , we take c pk and we are done.
If f ( pk ) 0 , we set
ak 1 ak , bk 1 pk ,
while if f ( pk ) 0 , we set
ak 1 pk , bk 1 bk .
In either case, we let
I k 1 [ak 1 , bk 1 ];
then I k 1 I k and f (ak 1 ) 0, f (bk 1 ) 0 .
If the process terminates by locating a point pn such that f ( pn ) 0 , then we are done. If the
process does not terminate, then we obtain a nested sequence of closed bounded intervals
I n [an , bn ] such that for every n we have
f (an ) 0 and f (bn ) 0
Furthermore, since the intervals are obtained by repeated bisections, the length of I n is equal to
bn an (b a ) / 2n 1 . It follows from the Nested Intervals Property that there exists a point c that
belongs to I n for all n . Now c I n for all n implies an c bn for all n , and hence
we have
(b a )
0 c an bn an ,
2n 1
and 0 bn c bn an (b a ) / 2n 1 .
Hence, it follows, by Squeeze Theorem, that
1
0 lim(c an ) (b a )lim n 1
2
1
and 0 lim(bn c) (b a )lim n 1 .
2
That is,
0 c lim(an ) 0
and 0 lim(bn ) c 0 .
Hence lim( an ) c lim(bn ) .
Since f is continuous at c , the fact that ( an ) c and (bn ) c implies (by Sequential Criterion
for Continuity) that
lim f (an ) f (c) lim f (bn ) .
The fact that f (an ) 0 for all n implies that
f (c) lim f ( an ) 0 .
6. Show that the function f ( x) 21n x x 2 has root in the interval [1, 2] . Use the Bisection
Method and a calculator to find the root with error less than 102 .
7. Show that the equation x cos x has a solution in the interval [0, 2 ] . Use the Bisection
Method and a calculator to find an approximate solution of this equation, with error less
than 103 .
8. The function f ( x) ( x 1)( x 2)( x 3)( x 4)( x 5) has five roots in the interval [0, 7]. If the
Bisection Method is applied on this interval, which of the roots is located?
9. The function for g ( x) ( x 2)( x 3)( x 4)( x 5)( x 6) has five roots on the interval [0, 7]. If
the Bisection Method is applied on this interval, which of the roots is located?
10. If the Bisection Method is used on an interval of length 1 to find pn with error pn c 105 ,
determine the least value of n that will assure this accuracy.
_____________
2
BOLZANO’S INTERMEDIATE VALUE THEOREM
The next result is a generalization of the Location of Roots Theorem. It assures us that a
continuous function on an interval takes on (at least once) any number that lies between two of
its values.
Bolzano’s Intermediate Value Theorem: Let I be an interval and let f : I be continuous on
I. If a, b I and if k satisfies f (a) k f (b) , then there exists a point c I between a and b
such that f (c) k .
To prove the result, suppose that a b and let g ( x) f ( x) k . Then f (a) k f (b) implies
g (a ) 0 g (b) . By the Location of Roots Theorem there exists a point c with a c b such that
0 g (c).
i.e., such that 0 f (c ) k .
Therefore f (c ) k .
If b a , let h( x) k f ( x) so that h(b) 0 h(a ) . Therefore there exists a point c with
b c a such that 0 h(c) k f (c) , and hence f (c) k . This completes the proof.
Corollary Let I [a, b] be a closed, bounded interval and let f : I be continuous on I. If
k is any number satisfying
inf f ( I ) k sup f ( I )
then there exists a number c I such that f (c) k .
It follows from the Maximum-Minimum Theorem that there are points c* and c* in I such that
inf f ( I ) f (c* ) and f (c* ) sup f ( I ).
Hence by the assumption
inf f ( I ) f (c* ) k f (c* ) sup f ( I ). .
Hence by Bolzano’s Intermediate Value Theorem, there exists a point c I between c* and
c such that f (c) k . This completes the proof.
*
Theorem Let I be a closed bounded interval and let f : I be continuous on I. Then the set
f ( I ) { f ( x) : x I } is a closed bounded interval.
To prove this, we let m inf f ( I ) and M sup f ( I ) , then by the Maximum–Minimum Theorem,
m and M belong to f ( I ) . Moreover, we have
f ( I ) [ m, M ] . … (3)
If k is any element of [ m, M ] , then it follows from Corollary to Bolzano’s Intermediate Value
Theorem that there exists a point c I such that k f (c) . Hence, k f ( I ) and we conclude that
[ m, M ] f ( I ) . … (4)
From (3) and (4), we have f ( I ) [m, M ] .Since [ m, M ] is a closed and bounded interval, f ( I ) is
also a closed and bounded interval. (Also, we note that the endpoints of the image interval are
m and M, resp., the absolute minimum and absolute maximum values of the function). This
completes the proof.
The previous theorem states that the image of closed bounded interval under a continuous
function is also a closed bounded interval. The endpoints of the image interval are the absolute
minimum and absolute maximum values of the function, and all values between the absolute
minimum and the absolute maximum values belong to the image of the function.
If I [a, b] is an interval and f I is continuous on I, we have proved that f ( I ) is the
interval [ m, M ] . We have not proved (and it is not always true) that f ( I ) is the interval
[ f (a), f (b)] .
The preceding theorem is a preservation theorem in the sense that it states that the continuous
image of a closed bounded interval is a set of the same type. The next theorem extends this
result to general intervals. However, it should be noted that although the continuous image of
an interval is shown to be an interval, it is not true that the image interval necessarily has the
same form as the domain interval.
Example We now give examples to show that the continuous image of an open interval need
not be an open interval, and the continuous image of an unbounded closed interval need not be
a closed interval.
If
1
f ( x) 2
x 1
for x , then f is continuous on . It is easy to see that if I1 (1,1) , then f ( I1 ) 12 ,1 , which is
not an open interval. Also, if I 2 is the closed interval given by I 2 0, , then f ( I 2 ) 0,1 ,
which is not a closed interval (Fig. 4).
Preservation of Intervals Theorem Let I be an interval and let f : I be continuous on I.
Then the set f ( I ) is an interval.
To prove the result, let , f ( I ) with . Then there exist points a, b I such that
f (a) and f (b) . Further, it follows from Bolzano’s Intermediate Value Theorem that if
k ( , ) then there exists a number c I with k f (c) f ( I ) .
Hence, [ , ] f ( I ) , showing that f ( I ) possesses property (1) of Characterization Theorem
for Intervals . Therefore f ( I ) is an interval and this completes the proof.
Example Let I [a, b] and let f : I and g : I be continuous functions on I. We show
that the set E {x I : f ( x) g ( x)} is with the property that if ( xn ) E and xn x0 , then x0 E .
To prove this, we note that since f and g are continuous, f g : I is also continuous.
Also, since E {x I : ( f g )( x) 0} ,
we have E ( f g ) 1{0} .
As {0} is a closed set and f g is continuous, ( f g ) 1{0} is a closed subset of I. i.e., E is a
closed subset of I.
Hence if we take a sequence ( xn ) in E with limit x0 , then its limit x0 must belongs to the set E.
Solved Examples
Verify the following two important limits:
Answers
x2 , x 2
f ( x)
1, x 2.
Answers Our task is to show that given 0 there exists 0 such that for all x
0 | x 2 | | f ( x) 4 | .
Step 1 : We solve the inequality | f ( x) 4 | to find on open interval about x0 2 on which
the inequality holds for all x x0 .
For x x0 2, we have f ( x) x 2 , and the inequality to solve is | x 2 4 | :
| x2 4 |
x 2 4
4 x2 4
4 | x| 4 here we assume that 4
4 x 4 this is an open interval about x0 2 that solves the
inequality
The inequality f ( x) 4 holds for all x 2 in the open interval 4 , 4 .
Step 2 : We find a value of 0 that places the centered interval (2 , 2 ) inside the
interval ( 4 , 4 ) :
Take to be the distance from x0 2 to the nearer endpoint of 4 , 4 . In other
words, take
min 2 4 , 4 2 ,
the minimum (the smaller) of the two numbers 2 4 and 4 2 . If has this or any
smaller positive value, the inequality 0 | x 2 | will automatically place x between 4
and 4 to make | f ( x) 4 | . For all x ,
0 | x 2| | f ( x) 4 | .
This completes the proof.
somewhere between -1 and 2. In other words, we want to show that there is a number c such
So we have,
x2
Example If possible, determine if f ( x) 20sin( x 3)cos takes the following values in the
2
interval [0,5].
If possible, suppose the function takes on either of the two values above in the interval [0,5].
First, let’s notice that this is a continuous function and so we know that we can use the
Intermediate Value Theorem to do this problem.
Now, for each part we will let M be the given value for that part and then we’ll need to
show that M lives between and . If it does then we can use the Intermediate Value
Theorem to prove that the function will take the given value.
So, since we’ll need the two function evaluations for each part let’s give them here,
So, by the Intermediate Value Theorem there must be a number such that
(b) In this part we’ll define . We now have a problem. In this part M does not live
between and . So, what does this mean for us? Does this mean that in
[0,5]?
Unfortunately for us, this doesn’t mean anything. It is possible that in [0,5], but
is it also possible that in [0,5]. The Intermediate Value Theorem will only tell us
that c’s will exist. The theorem will NOT tell us thatc’s don’t exist.
In this case it is not possible to determine if in [0,5] using the Intermediate Value
Theorem.
Theory: Let’s take a look at the following graph and let’s also assume that the limit does exist.
What the definition is telling us is that for any number that we pick we can go to our
graph and sketch two horizontal lines at and as shown on the graph above. Then
somewhere out there in the world is another number , which we will need to determine,
that will allow us to add in two vertical lines to our graph at and .
Now, if we take any x in the region, i.e. between and , then this x will be closer
to a than either of and . Or,
If we now identify the point on the graph that our choice of x gives then this point on
the graph will lie in the intersection of the regions. This means that this function value f(x) will
be closer to L than either of and . Or,
So, if we take any value of x in the region then the graph for those values of x will lie in the
region.
Notice that there are actually an infinite number of possible δ’s that we can choose. In
fact, if we go back and look at the graph above it looks like we could have taken a slightly
larger δ and still gotten the graph from that region to be completely contained in the region.
Also, notice that as the definition points out we only need to make sure that the function
is defined in some interval around but we don’t really care if it is defined at .
Remember that limits do not care what is happening at the point, they only care what is
happening around the point in question.
Okay, now that we’ve gotten the definition out of the way and made an attempt to
understand it let’s see how it’s actually used in practice.
These are a little tricky sometimes and it can take a lot of practice to get good at these so
don’t feel too bad if you don’t pick up on this stuff right away. We’re going to be looking at a
couple of examples that work out fairly easily.
Example Use the definition of the limit to prove the following limit.
In this case both L and a are zero. So, let be any number. Don’t worry about what the number is,
is just some arbitrary number. Now according to the definition of the limit, if this limit is to be true we
will need to find some other number so that the following will be true.
Often the way to go through these is to start with the left inequality and do a little simplification and see if
that suggests a choice for . We’ll start by bringing the exponent out of the absolute value bars and then
taking the square root of both sides.
Now, the results of this simplification looks an awful lot like with the exception of the “ ”
part. Missing that however isn’t a problem, it is just telling us that we can’t take . So, it looks like
Verification is in fact pretty much the same work that we did to get our guess. First, let’s again let
be any number and then choose . Now, assume that . We need to show that by
choosing x to satisfy this we will get,
To start the verification process we’ll start with and then first strip out the exponent from the absolute
values. Once this is done we’ll use our assumption on x, namely that . Doing all this gives,
Or, upon taking the middle terms out, if we assume that then we will get,
So, just what have we done? We’ve shown that if we choose then we can find a so that we
have,
Assignments
1. Let I [a, b] , let f :I be continuous on I, and assume that f ( a ) 0, f (b) 0 . Let
W {x I : f ( x ) 0}, and let w sup W . Prove that f ( w) 0 . (This provides an alternative proof
of Location of Roots Theorem).
2. Let I [0, / 2] and let f : I be defined by f ( x) sup{x 2 ,cos x} for x I . Show there exists
an absolute minimum point x0 I for f on I. Show that x0 is a solution to the equation cos x x 2 .
3. Suppose that f : be continuous on and that lim f 0 and lim f 0 . Prove that f is
x x
bounded on and attains either a maximum or minimum on . Give an example to show that
both a 4. Let f : be continuous on and let . Show that if x0 is such that
f ( x0 ) , then there exists a -neighborhood U of x0 such that f ( x) for all x U .
5. Examine which open [respectively, closed] intervals are mapped by f ( x) x 2 for x onto
open [respectively, closed] intervals.
6. Examine the mapping of open [respectively, closed] intervals under the functions
g ( x) 1/( x 2 1) and h( x) x3 for x .
7. Give an example of an open interval that is not mapped by f ( x ) x 2 for x onto open
interval.
8. If f [0,1] is continuous and has only rational [respectively, irrational] values, must f be
constant? Prove your assertion.
9. Let J ( a, b) and let g : J be a continuous function with the property that for every x J ,
the function g is bounded on a neighborhood V x ( x) of x. Show by example that g is not
necessarily bounded on J.
10. Let I [a, b] and let f : I be a (not necessarily continuous) function with the property that
for every x I , the function f is bounded on a neighborhood V x ( x) of x (in the sense of the
Definition “ Let A , let f : A , and let c be a cluster point of A. We say that f is
bounded on a neighborhood of c if there exists a neighborhood V (c) of c and a constant
M 0 such that such that f ( x ) M for all x A V (c) ”). Prove that f is bounded on I.
3
UNIFORM CONTINUITY
A function f is uniformly continuous if it is possible to guarantee that f ( x) and f (u ) be
as close to each other as we please by requiring only that x and u are sufficiently close to each
other; unlike ordinary continuity, the maximum distance between x and u cannot depend on x
and u themselves.
There are other selections that can be made for . (For example we could also take
1 ( , u ) inf{13 u, 32 u 2 } ; however, we still have inf{1 ( , u ) : u 0} 0 .) In fact, there is no way of
choosing one value of that will “work” for all u 0 for the function g ( x ) 1/ x , as we shall
see.
Uniform Continuity
Now it often happens that the function f is such that the number can be chosen to be
independent of the point u A and to depend only on . For example, if f ( x) 2 x for all x ,
then
f ( x) f (u ) 2 x u ,
and so we can choose ( , u ) / 2 for all 0, u . Here is independent of u.
Definition Let A and let f : A . We say that f is uniformly continuous on A if for each
0 there is a ( ) 0 such that if x, u A are any numbers satisfying x u ( ) , then
f ( x) f (u ) .
We note that ( ) in the definition above is independent of u. Also, it is clear that if f is
uniformly continuous on A, then it is continuous at every point of A. In general, however, the
converse does not hold, as is shown by the function g ( x ) 1/ x on the set A {x : x 0} .
Example Show that the function f ( x ) x 2 is uniformly continuous on [ 1,1] .
Let x, y be any two points of [ 1,1] . Then
f ( x) f ( y ) x 2 y 2 x y x y
Now we note that
x y x y 2, since x, y [1,1] .
Hence, for x, y [1, 1]
f ( x) f ( y ) 2 x y .
Hence for any 0 ,
f ( x) f ( y )
whenever x y .
2
Hence taking , we have for any 0 and any x, y [ 1,1]
2
f ( x ) f ( y ) whenever x y .
This shows that the given function is uniformly continuous.
Example Show that the function f ( x) 1/ x is not uniformly continuous on
A {x : 0 x 1}.
Since x is continuous on A (0,1] and x 0 for all x A, 1/ x is continuous on A. If
possible, let f ( x) 1/ x is uniformly continuous on A. Then corresponding to 1 , there exists
a 0 such that
1 1
1 whenever x y
x y
By Archimedean Property of real numbers [“If x , then there exists nx such that x nx ”],
1 1
corresponding to this 0 , we can find a natural number m such that m . Then .
m
1 1
Since m , both and are elements of A (0,1] and
m 2m
1 1 1 1
.
m 2m 2m m
But
1 1 1 1
f f m 2m m 1 .
m 2m 1/ m 1/ 2m
This is a contradiction to the inequality. Hence our assumption is wrong. i.e., f is not uniformly
continuous on A.
Example If g ( x ) 1/ x for x B [1/ 2, 1] , then
ux
g ( x) g (u ) … (1)
ux
1
Since x, u [1/ 2, 1], xu 1/ 2 1/ 4 4.
2
xu
xu
So if x u , f ( x ) f (u ) 4 if / 4.
xu
Here doesn’t depend on x and u. It works for any pair of numbers in the interval [1/ 2, 1] with
x u . Hence g ( x ) 1/ x is uniformly continuous on B [1/ 2, 1] .
Example Show that f ( x) cos x 2 , x is not uniformly continuous on .
If possible let f ( x ) cos x 2 , x is uniformly continuous on . Then corresponding to
1 , there exists a 0 , such that for any x, y ,
f ( x) f ( y ) 1 whenever x y
By Archimedean Property, corresponding to this 0 we can find natural number n such
1 1
that 2 n . Hence .
n
Now let x (n 1) and y n .
Then
(n 1) n
x y (n 1) n
(n 1) n
1
n 1 n 2 n n
But f ( x ) f ( y ) cos x 2 cos y 2 cos( n 1) cos n
(1) n 1 (1) n 2 1 .
This is in contradiction with inequality (6). Hence our assumption is wrong i.e., f is not
uniformly continuous on .
It is useful to formulate a condition equivalent to saying that f is not uniformly continuous
on A. We give such criteria in the next result.
Nonuniform Continuity Criteria: Let A and let f : A . Then the following statements
are equivalent:
(i) f is not uniformly continuous on A.
(ii) There exists an 0 0 such that for every 0 there are points x , u in A such that
x u and f ( x ) f (u ) 0 .
(iii) There exists an 0 0 and two sequences ( xn ) and (un ) in A such that lim( xn un ) 0 and
f ( xn ) f (un ) 0 for all n .
Example Show that g ( x ) 1/ x is not uniformly continuous on A {x : x 0} .
We apply Nonuniform Continuity Criterion. For, if xn 1/ n and un 1/(n 1) , then we have
lim( xn un ) 0 , but
1 1 1 1
g ( xn ) g (un ) n (n 1) 1
xn un 1/ n 1/(n 1)
for all n . Now choose 0 1 . By the Nonuniform Continuity Criterion, g is not uniformly
continuous.
We now present an important result that assures that a continuous function on a closed
bounded interval I is uniformly continuous on I.
Uniform Continuity Theorem: Let I be closed bounded interval and let f : I be
continuous on I. Then f is uniformly continuous on I.
To prove the theorem, we apply Contrapositive Method (which works on the logic
p q q p ) for proving the Theorem. If f is not uniformly continuous on I , then, by the
Nonuniform Continuity Criterion, there exists 0 0 and two sequences ( xn ) and (un ) in I such
that
1
xn un
and f ( xn ) f (un ) 0 for all n .
n
Since I is bounded, the sequence ( xn ) is bounded; and hence by the Bolzano-Weierstrass
Theorem there is a subsequence ( xnk ) of ( xn ) that converges to an element, say, z. Since I is
closed, the limit z belongs to I, by Theorem A in chapter 2.
Since
1
unk z unk xnk xnk z xnk z
nk
1
and since 0 and xnk z 0 as nk , we have the subsequence (unk ) also converges to
nk
z.
Now if f is continuous at the point z, then, by the sequential criterion for continuity, as
( xnk ) z and (unk ) z , we have both of the sequences f (x )
nk and f (u )
nk converge to f ( z ) .
But this is not possible since
f ( xn ) f (un ) 0
for all n . Thus the hypothesis that f is not uniformly continuous on the closed bounded
interval I implies that f is not continuous at some point z I . Consequently, if f is continuous
at every point of I, then f is uniformly continuous on I. This completes the proof.
Example Show that g ( x ) 1/ x is uniformly continuous on A {x :1 x 2}.
To show this, we note that since A [1, 2] is closed and bounded and g is continuous on A , by
Uniform Continuity Theorem, g is Uniformly Continuous.
Lipchitz Functions
If a uniformly continuous function is given on a set that is not a closed bounded interval, then it
is sometimes difficult to establish its uniform continuity. However, there is a condition that
frequently occurs that is sufficient to guarantee uniform continuity.
Definition Let A and let f : A . If there exists a constant K 0 such that
f ( x) f (u ) K x u … (**)
for all x, u A , then f is said to be a Lipchitz function (or to satisfy a Lipchitz condition) on A.
Geometric Interpretation
The condition (**) that a function f : I on an interval I is a Lipchitz function can be
interpreted geometrically as follows. If we write the condition as
f ( x) f (u )
K , x, u I , x u ,
xu
then the quantity inside the absolute value is the slope of a line segment joining the points
x, f ( x) and u, f (u) . Thus a function f satisfies a Lipchitz condition if and only if the slopes of
all line segments joining two points on the graph of y f ( x) over I are bounded by some
number K.
Theorem If f : A is a Lipchitz function, then f is uniformly continuous on A.
To prove this we note that if f is a Lipchitz function, then there exists K 0 such that
f ( x) f (u )
K , x, u I , x u ,
xu
i.e., such that
f ( x) f (u ) K x u , x, u I , x u .
If 0 is given, we can take / K . If x, u A satisfy x u , then
f ( x) f (u ) K
K
Therefore f is uniformly continuous on A. This completes the proof.
Example If f ( x) x 2 on A [0, b] , where b 0 , then
f ( x) f (u ) x u x u 2b x u
f ( x) f (u )
for all x, u in [0, b] . Thus 2b for all x, u [0, b], x u .
xu
Thus f satisfies the Lipchitz condition (4) with K 2b on A, and therefore f is uniformly
continuous on A. Of course, since f is continuous and A is a closed bounded interval, this can
also be deducted from the Uniform Continuity Theorem. (Attention! The function f in the
above example f does not satisfy a Lipchitz condition on the interval [0, ) .)
Converse of the Theorem is not true, in general. Not every uniformly continuous function is a
Lipchitz function. The following example shows this.
Example Show that g ( x) x on [0, 2] is not a Lipchitz function, but a uniformly continuous
function.
If possible let g ( x) x , for x [0, 2] is a Lipchitz function. Then there exists a constant
K 0 such that
g ( x ) g (u ) K x u for all x, u [0, 2]
i.e., x u k x u for all x, u [0, 2]
x u
Then, x u x u
x u
x u 1
for all x, u A
x u K
By Archimedean Property, corresponding to the real number K, we can find a natural number
1 1
n, such that n K . Now consider x 2
and u . Clearly, both x, u [0, 2] .
4n 16n 2
But
1 1 3 1 1
x u
, since n K
2n 4n 4n n K
This is a contradiction. Hence our assumption is wrong. Hence g ( x) is not a Lipchitz function.
Since [0, 2] is closed and bounded and since g is continuous, by Uniform Continuity Theorem g
is uniformly continuous.
Alternatively, There is no number K 0 such that g ( x) K x for all x I . The
1
reason is this: If there is such a K then, in particular for 0 x 1, K , which is not possible
x1/ 2
as the set { x1/1 2 : 0 x 1} is unbounded. Therefore, g is not a Lipchitz function on I.
The following example illustrates that Uniform Continuity Theorem and Theorem 3 can
sometimes be combined to establish the uniform continuity of a function on a set.
Example We now verify the uniform continuity of the function g defined by g ( x) x on the
set A [0, ) .
The uniform continuity of the continuous function g on the closed bounded interval
I [0, 2] follows from the Uniform Continuity Theorem. If J [1, ) , then if both x, u are in J,
we have
xu
g ( x) g (u ) x u 12 x u .
x u
Thus g is a Lipchitz function on J with constant K 12 , and hence, by Theorem 3, g is uniformly
continuous on [1, ) . Now A I J , and for a given 0 , uniform continuity of g on I gives a
, denoted by I ( ) and uniform continuity of g on J gives a , denoted by J ( ).
Now take
( ) inf{1, I ( ), J ( )} .
Then for a given 0 there exists 0 such that
g ( x) g (u ) for x, u A and x u .
Hence g is uniformly continuous on A.
(i) lim f ( x) L.
x c
(ii) For every sequence ( xn ) in A that converges to c such that xn c for all n , the
sequence ( f ( xn )) converges to f (c) .
Cauchy Convergence Criterion A sequence of real numbers is convergent if and only if it is
a Cauchy sequence.
Continuous Extension Theorem: A function f is uniformly continuous on the interval (a, b) if
and only if it can be defined at the endpoints a and b such that the extended function is
continuous on [ a, b] .
Proof.
() This direction is trival.
() Suppose f is uniformly continuous on (a, b) . We shall show how to extend f to a; the
argument for b is similar. This is done by showing that lim f ( x) L exists, and this is
xa
accomplished by using the sequential criterion for limits. If ( xn ) is a sequence in (a, b) with
lim( xn ) a , then it is a Cauchy sequence, and by the preceding theorem, the sequence f ( xn )
is also a Cauchy sequence, and so is convergent by Cauchy Convergence Criterion. Thus the
limit lim f ( xn ) L exists. If (un ) is any other sequence in (a, b) that converges to a, then
lim(un xn ) a a 0 , so by the uniform continuity of f we have
0 L L.
Since we get the same value L for every sequence converging to a, we infer from the sequential
criterion for limits that f has limit L at a. If we define f (a ) L , then f is continuous at a. The
same argument applies to b, so we conclude that f has a continuous extension to the interval [a,
b]. This completes the proof.
Example Examine the uniform convergence of the functions
f ( x) sin(1/ x)
and
g ( x) x sin(1/ x)
on (0, b] for all b 0 .
Since the limit of f ( x) sin(1/ x) at 0 does not exist (Fig. B), we infer from the Continuous
Extension Theorem that the function is not uniformly continuous on (0, b] for any b 0 . On the
other hand, since lim x sin(1/ x ) 0 exists, the function g ( x) x sin(1/ x) is uniformly continuous
x0
0, 2 x 1,
1, 1 x 0,
,1
0 x 12 ,
s ( x) 2
3,
1
2
x 1
2, 1 x 3,
2, 3 x 4,
is a step function.
We will now show that a continuous function on a closed bounded interval I can be
approximated arbitrarily closely by step functions.
kn kn x k (1 x)nk
n
Bn ( x) f
k 0
The polynomial function Bn is called the nth Bernstein polynomial for f; it is a polynomial of
degree at most n and its coefficients depend on the values of the function f at the n 1 equally
spaced points 0, 1/ n, 2 / n, ... , k / n, ... ,1 and on the binomial coefficients
Example If f ( x ) x 2 , for x [0,1] , calculate the first few Bernstein polynomials for f. Show that
1 1
Bn ( x ) 1 x 2 x .
n n
Here f ( x ) x 2 for x [0,1]
Bernstein polynomials for f are given by
n
k
Bn ( x ) f C ( n, k ) x k (1 x ) n k , n 1, 2, ...
k 0 n
1
k
Then B1 ( x ) f C (1, k ) x k (1 x )1 k
k 0 1
1
k 2 C (1, k ) x k (1 x)1 k 0 x x .
k 0
2
k
B2 ( x ) f C (2, k ) x k (1 x ) 2 k
k 0 2
2
k2
C (2, k ) x k (1 x ) 2 k
k 0 4
0 14 2 x (1 x) x 2 12 x 2 12 x
3
k
B3 ( x ) f C (3, k ) x k (1 x )3 k
k 0 3
3
k2
C (3, k ) x k (1 x )3 k
k 0 9
0 19 3 x (1 x) 2 94 3 x 2 (1 x) 99 1 x 3
13 [ x 2 x 2 x 3 ] 34 [ x 2 x 3 ] x 3 32 x 2 13 x .
In general for any n 1, 2,...
n
k2
Bn ( x ) 2
C ( n, k ) x k (1 x ) n k
k 0 n
1 n n!
2 [ k ( k 1) k ] x k (1 x ) n k
n k 0 k !(n k )!
1 n n!
2
x k (1 x) n k
n k 2 (k 2)!(n k )!
n
n!
x k (1 x) n k
k 1 ( k 1)!( n k )!
1 n
(n 2)!
n2
n ( n 1) x 2
x k 2 (1 x) n k
k 1 ( k 2)!( n k )!
n
(n 1)!
nx x k 1 (1 x) n k
k 1 ( k 1)!( n k )!
2
n(n 1) x 2 [ x (1 x)]n 2 nx[ x 1(1 x)]n 1
1
n
1
2 n( n 1) x 2 nx 1 x 2 x .
1 1
n n n
Bernstein’s Approximation Theorem
Let f :[0,1] be continuous and let 0 . There exists an n such that if n n , then we
have f ( x) Bn ( x) for all x [0,1] .
The Weierstrass Approximation Theorem can be derived from the Bernstein Approximation
Theorem by a change of variable. Specifically, we replace f :[ a, b] by a function
F :[0,1] , defined by
F (t ) f (a (b a )t ) for t [0,1].
The function F can be approximated by Bernstein polynomials for F on the interval [0, 1], which
can then yield polynomials on [a, b] that approximate f.
Assignments
1
1. Show that the function f ( x) 2 is uniformly continuous on A [1, ) , but that it is not
x
uniformly continuous on B (0, ) .
1
2. Show that the function f ( x) is uniformly continuous on the set A [ a, ) , where a is a
x
positive constant.
3. Use the Nonuniform Continuity Criterion to show that the following functions are not
uniformly continuous on the given sets.
a) f ( x) x 2 , A [0, )
b) g ( x) sin(1/ x), B [0, )
17. Let f :[0.1] is continuous and has only rational values. Show that f must be constant.
18. If f1 ( x) x for x [0,1] , calculate the first few Bernstein polynomials for f1 . Show that they
coincide with f1 .
4
THE RIEMANN INTEGRAL - PART I
The Riemann integral, created by Bernhard Riemann, was the first rigorous definition of
the integral of a function on an interval. For many functions and practical applications, the
Riemann integral can be evaluated by using the fundamental theorem of calculus or
(approximately) by numerical integration. The Riemann integral is unsuitable for many
theoretical purposes. Some of the technical deficiencies in Riemann integration can be remedied
with the Riemann–Stieltjes integral, and most disappear with the Lebesgue integral.
Often we will denote the partition P by the notation P {[ xi 1 , xi ]}in1 . We define the norm (or
mesh) of P to be the number
P max{x1 x0 , x2 x1 ,..., xn xn 1} .
Thus the norm of a partition is merely the length of the largest subinterval into which the
partition divides [ a, b] .
Clearly, many partitions have the same norm, so the partition is not a function of the norm.
If a point ti has been selected from each subinterval I i [ xi 1 , xi ] , for i 1, 2, ... , n, then the
points are called tags of the sub-intervals I i . A set of ordered pairs
P {([ x , x ], t )}n
i 1 i i i 1
of subintervals and corresponding tags is called a tagged partition of I. (The dot over P
indicates that a tag has been chosen for each subinterval.) The tags can be chosen in a wholly
arbitrary fashion; for example, we can choose the tags to be the left endpoints, or the midpoints
of the subintervals, etc. Note that an endpoint of a subinterval can be used as a tag for two
consecutive subintervals. Since each tag can be chosen in infinitely many ways, each partition
can be tagged in infinitely many ways.
The norm of a tagged partition is defined as for an ordinary partition and does not depend on the
choice of tags.
Definition If P is the tagged partition given above, we define the Riemann sum of a function
f :[ a, b] corresponding to P to be the number
n
S ( f ;P ) f (ti )( xi xi 1 ) .
i 1
Notation We will also use this notation when P denotes a subset of a partition, and not the
entire partition.
Definition If the function f is positive on [ a, b] , then the Riemann sum (2) is the sum of the
areas of n rectangles whose bases are the subintervals I i [ xi 1 , xi ] and whose heights are f (ti )
(Fig.1).
Riemann Integral
We now define the Riemann integral of a function f on an interval [ a, b] .
Definition A function f :[ a, b] is said to be Riemann integrable on [ a, b] if there exists a
number L such that for every 0 there exists 0 such that if P is any tagged
partition of [ a, b] with P , then
S ( f ;P ) L .
The set of all Riemann integrable functions on [ a, b] will be denoted by R [ a, b] .
Remark It is sometimes said that integral L is “the limit” of the Riemann sums S ( f ;P ) as the
norm P 0 . However, since S ( f ;P ) is not a function of P , this limit is not of the type that
we have studied before.
First we will show that if f R [ a, b] , then the number L is uniquely determined. It will be
called the Riemann integral of f over [ a, b] . Instead of L, we will usually write
b b
L f or f ( x)dx .
a a
It should be understood that any letter than x can be used in the latter expression, so long as it
does not cause any ambiguity.
S ( f ;P1 ) L / 2 . … (*)
Also there exists / 2 0 such that if P2 is any tagged partition with P2 / 2 , then
Now let min{ / 2 , / 2 } 0 and let P be a tagged partition with P . Since both
If we use only the definition, in order to show that a function f is Riemann integrable we must
(i) know (or guess correctly) the value L of the integral, and (ii) construct a that will suffice
for an arbitrary 0 . The determination of L is sometimes done by calculating Riemann sums
and guessing what L must be. The determination of is likely to be difficult.
In actual practice, we usually show that f R a, b by making use of some of the theorems
that will be discussed after considering some examples based on the Definition.
Example Every constant function on [ a, b] is in R [a, b] .
Let f ( x) k for all x [a, b] . If P {[ xi 1 , xi ], ti )}in1 is any tagged partition of [a, b], then it is
clear that
n n n
S ( f ;P ) f (ti )( xi xi 1 ) k ( xi xi 1 ) k ( xi xi 1 )
i 1 i 1 i 1
k ( xn xn 1 xn 1 xn 1 xn 2 x0 )
k ( xn x0 )
k (b a) .
Hence, for any 0 , we can choose 1 so that if P , then
S ( f ;P ) k (b a ) 0 .
b
Since 0 is arbitrary, we conclude that f R [a, b] and a
f k (b a ) .
Example Let g :[0,3] be defined by g ( x) 2 for 0 x 1 and g ( x) 3 for 1 x 3 . Find
3
0
g.
A preliminary investigation, based on the graph of g, suggests that we might expect that
3
0
g 2 1 2 3 8 .
Let P be a tagged partition of [0,3] with norm ; we will show how to determine in
order to ensure that S ( g ;P ) 8 . Let P1 be the subset of P having its tags in [0,1] where
g ( x) 2 , and the let P2 be the subset of P with its tags in (1,3] where g ( x) 3 . It is obvious
that we have
S ( g ;P ) S ( g ;P1 ) S ( g ;P2 ) .
Since P , if u [0,1 ] and u [ xi 1 , x], then xi 1 1 so that xi xi 1 1 , and hence
the tag ti [0,1] . Therefore, the interval [0,1 ] is contained in the union of all subintervals in
P with tags ti [0,1] . Similarly, this union is contained in [0,1 ] . Since g (ti ) 2 for these
tags, we have
2(1 ) S ( g ;P1 ) 2(1 ) .
A similar argument shows that the union of all subintervals with tags ti (1,3] contains the
interval [1 ,3] of length 2 , and is contained in [1 , 3] of length 2 . Therefore,
3(2 ) S ( g ;P2 ) 3(2 ) .
Adding these inequalities and using equation (3) we have
8 5 S ( g ;P ) S ( g ;P ) S ( g ;P ) 8 5
1 2
xi xi 1 for i 1,..., n . Also let Q have the same partition points, but where we choose the
tag qi to be the midpoint of the interval I i . Since both ti and qi belong to this interval, we have
ti qi . Using the Triangle Inequality, we deduce
n n
S (h;P ) S (h; Q ) t (x x
i 1
i i i 1 ) qi ( xi xi 1 )
i 1
n
ti qi ( xi xi 1 )
i 1
n
( xi xi 1 )
i 1
( xn x0 ) .
Since S (h; Q ) 12 , we infer that if P is any tagged partition with P , then
S (h;P ) 12 .
Therefore we are led to take . If we choose , we can retrace the argument to
1 1
conclude that h R [0,1] and h
0 0
xdx 12 .
Example Let F ( x) 1 for x 15 , 52 , 53 , 54 , and F ( x) 0 elsewhere on [0, 1]. Show that F R [0,1]
1
and F 0 .
0
Here there are four points where F is not 0, each of which can belong to two sub-intervals in
a given tagged partition P . Only these terms will make a non zero contribution to S ( F ;P ) .
Therefore we choose / 8 .
If P , let P0 be the subset of P with tags different from , , , , and let P1 be the
1 2 3 4
5 5 5 5
S ( F ;P ) S ( F ;P0 ) S ( F ;P1 ) S ( F ;P1 ) . Since there are at most 8 terms in S ( F ;P1 ) and each
Given 0 , let E be the (finite) set of points where G ( x) , let n be the number of points in
E , and let /(2n ) . Let P be a tagged partition such that P . Let P0 be the subset of
P with tags outside of E and let P1 be the subset of P with tags in E . Then, as in the
previous example, we have
0 S (G;P ) S (G;P1 ) (2n ) .
1
Since 0 is arbitrary, we conclude that G R[0,1] and G 0 .
0
To prove (b), we note that if P {([ xi 1 , xi ], ti )}in1 is a tagged partition of [a, b] , then
n n
S (kf ;P ) ( kf )(ti )( xi xi 1 ) k f (ti )( xi xi 1 )
i 1 i 1
n
k f (ti )( xi xi 1 ) kS ( f ;P ) .
i 1
Hence
S ( f ;P ) f / 2 and S ( g ;P ) g / 2
b b
… (4)
a a
S ( f g ;P ) a
b
a
f g S ( f ;P ) S ( g ;P ) f g
b b b
a a
S ( f ;P ) f S ( g ;P ) g
b b
a a
/2 /2 .
Since 0 is arbitrary, we conclude that f g R[a, b] and that its integral is the sum of the
integrals of f and g.
To prove (c), we note that S ( f ;P ) S ( g ;P ) and also (4) implies
S ( f ;P ) and S ( g ;P ) g .
b b
a 2
f
a 2
If we use the fact that S ( f ;P ) S ( g ;P ) , we have
b b
a
f g .
a
b b
But since 0 is arbitrary, we conclude that f g . This completes the proof.
a a
Theorem If f1 ,..., f n are in R [a, b] and if k1 ,..., kn , then the linear combination f i 1 ki f i
n
f i 1 ki fi
b b
n
a a
Hint for the Proof. Use mathematical Induction and part (a) and (b) of Theorem 2 above.
b
Theorem If f R [a, b] and f ( x) M for all x [a, b] , then a
f M (b a ) .
Take
g ( x) M for x [a, b]
S ( f ;P ) L 1 ,
which implies that
S ( f ;P ) L 1 . (***)
f (tk )( xk xk 1 ) L 1 f (t )( x x
ik
i i i 1 ).
S ( f ; Q ) f (tk )( xk xk 1 ) f (t )( x x
ik
i i i 1 ) L 1 ,
If b is an irrational number and 0 , then (by the Archimedean Property) there is a natural
1 1
number n0 such that n0 . i.e., such that . There are only a finite number of rationals
n0
with denominator less than n0 in the interval (b 1, b 1) . Hence 0 can be chosen so small
that the neighborhood (b , b ) contains no rational numbers with denominator less than
1
n0 . It then follows that for x b and x (0, ) , we have h( x) h(b) h( x) . Then
n0
h is continuous at the irrational number a.
We will now show that h R [0,1] .
Let 0 . Then the set E {x [0,1] : h( x) 2 } is a finite set. We let n be the number of
elements in E and let /(4n ) . If P is a tagged partition with P , let P1 be the subset
of P having tags in E and P2 be the subset of P having tags elsewhere in [0,1] . We observe
that P1 has at most 2n intervals whose total length is 2n / 2 and that 0 h(ti ) 1 for
every tag in P1 . Also the total lengths of the subintervals in P2 is 1 and h(ti ) / 2 for every
tag in P2 . Therefore we have
S (h;P ) S ( h;P1 ) S ( h;P2 ) 1 2n ( / 2) 1 .
Since 0 is arbitrary, we infer that h R[0,1] with integral 0.
Assignments
1. If I [1, 8] , calculate the norms of the following partitions:
a) P1 (1,2,3,4,5,6,7,8) b) P2 (1,3,4,6,8)
c) P3 (1,3,6,8) d) P4 (1,5,8)
2. If f ( x) x 2 for x [1,8] , calculate the following Riemann sums where Pi has the same
partition points as in Assignment 1, and the tags are selected as indicated.
a) P1 with the tags at the left endpoints of the subintervals.
b) P1 with the tags at the right endpoints of the subintervals.
c) P2 with the tags at the left endpoints of the subintervals.
d) P2 with the tags at the right endpoints of the subintervals.
3. Show that f :[a, b] is Riemann integrable on [a, b] if and only if there exists L such
that for every 0 there exists 0 such that if P is any tagged partition with norm
P , then S ( f ;P ) L .
a) Show that the union U1 of all subintervals in P with tags in [0,1] satisfies [0, 1 P ]
U1 [0,1 P ]
b) Show that the union U 2 of all subintervals in P with tags in [1,2] satisfies
[1 P , 2 P ] U 2 [1 P , 2 P ] .
5. Let P {( I i , ti )}in1 be a tagged partition of [a, b] and let c1 c2 .
a) If u belongs to a subinterval Ii whose tag satisfies c1 ti c2 , show that
c1 P u c2 P .
13. Let 0 a b , let Q( x) x for x [a, b] and let P {[ xi 1 , xi ]}in1 be a partition of [a, b] . For
2
15. Let 0 a b and m , let M ( x) x m for x [a, b] and let P {[ xi 1 , xi ]}in1 be a partition of
[a, b] . For each i, let qi be the positive mth root of
1
( xim xim 1 xi 1 xi xim11 xim1 ) .
m 1
a) Show that qi satisfies 0 xi 1 qi xi
b) Show that M ( qi )( xi xi 1 ) 1
m 1
( xim 1 xim11 )
c) If Q is the tagged partition with the same subintervals as P and the tags qi , show
1
that S ( M ; Q )
(b m 1 a m 1 )
m 1
d) Show that M R [a, b] and
b b 1
a
M x m dx
a m 1
(b m 1 a m 1 ) .
5
THE RIEMANN INTEGRAL - PART II
P and Q , then
S ( f ;P ) S ( f ; Q ) .
Proof.
() If f R [a, b] with integral L , let 0 be such that if P and Q are any tagged
2
partitions of [a, b] with P and Q , then
S ( f ;P ) L / 2 and S ( f ; Q ) L / 2 .
Therefore we have
S ( f ; P ) S ( f ; Q ) S ( f ;P ) L L S ( f ; Q )
S ( f ;P ) L L S ( f ; Q )
/2 /2 .
() For each n , let n 0 be such that if P and Q are tagged partitions with norms n ,
then
S ( f ;P ) S ( f ; Q ) 1/ n .
Evidently we may assume that n n 1 for n ; otherwise, we replace n by
n min{1 , ..., n } .
For each n , let Pn be a tagged partition with Pn n . Clearly, if m n then both Pm and
S ( f ; Q ) A S ( f ; Q ) S ( f ;PK ) S ( f ;Pk ) A
1/ K 1/ K .
Since 0 is arbitrary, then f R [a, b] with integral A. This completes the proof.
Example Let g :[0,3] be defined by g ( x) 2 for 0 x 1 , and g ( x) 3 for 1 x 3 . We
have seen that if P is a tagged partition of [0,3] with norm P , then
8 5 S ( g ;P ) 8 5 .
Hence if Q is another tagged partition with Q , then
8 5 S ( g ; Q ) 8 5 .
If we subtract these two inequalities, we obtain
S ( g ;P ) S ( g ; Q ) 10 .
In order to make this final term , we are led to employ the Cauchy Criterion with / 20 .
The Cauchy Criterion can be used to show that a function f :[a, b] is not Riemann
integrable. To do this we need to show that: There exists 0 0 such that for any 0 there
exists tagged partitions P and Q with P and Q such that
S ( f ;P ) S ( f ; Q ) 0 .
Example Show that the Dirichlet function, defined by f ( x) 1 if x [0,1] is rational and
f ( x) 0 if x [0,1] is irrational is not Riemann integrable.
Here we take 0 12 . If P is any partition all of whose tags are rational numbers then
S ( f ;P ) 1 , while if Q is any tagged partition all of whose tags are irrational numbers then
S ( f ; Q ) 0 . Since we are able to take such tagged partitions with arbitrarily small norms, we
conclude that the Dirichlet function is not Riemann integrable.
The next result will be used to establish the Riemann integrability of some important classes
of functions.
Squeeze Theorem: Let f :[a, b] . Then f R [a, b] if and only if for every 0 there exist
functions and in R[a, b] with
( x) f ( x) ( x) for all x [a, b] , … (**)
and such that
b
a
( ) … (***)
Proof.
() Take f for all 0 .
() Let 0 . Since and belong to R [a, b] , there exists 0 such that if P is any
tagged partition with P then
S ( f ; Q )
b b
a a
a a
b
( ) 2 3 .
a
Since 0 is arbitrary, the Cauchy Criterion implies that f R [a, b] . This completes the proof.
Proof.
For a tagged partition P {([ xi 1 , xi ], ti )}in1 , Riemann sum is giving by
n
S ( ,P ) (ti )( xi xi 1 ).
i 1
b
Hence by the Definition of Riemann integrability, R [a, b] and that (d c) .
a
proof.
a j Jj
j 1
k j d j c j , using Lemma 2
n
j 1
Hence R [a, b] .
We will now use the Squeeze Theorem to show that an arbitrary continuous function is
Riemann integrable.
Theorem If f :[a, b] is continuous on [a, b] , then
f R [a, b] .
Proof. Being continuous function on a closed and bounded interval, it follows that f is
uniformly continuous on [a, b] . Therefore, given 0 there exists 0 such that if
u, v [a, b] and u v , then we have
f (u ) f (v) .
ba
Let P {I i }in1 be a partition such that P , let ui I i be a point where f attains its
minimum value on I i and let vi I i be a point where f attains maximum value on I i .
Let be the step function defined by ( x) f (ui ) for x [ xi 1 , xi ) (i 1,..., n 1) and
( x) f (un ) for x [ xn 1 , xn ] . Let be defined similarly using the points vi instead of the ui .
i.e., ( x) f (vi ) for x [ xi 1 , xi )(i 1,..., n 1) and ( x) f (vn ) for x [ xn 1 , xn ] . Then we
obtain
( x) f ( x) ( x) for all x [a, b] .
Moreover, it is clear that
0 ( ) ( f (vi ) f (u j ) ( xi xi 1 )
n
i 1
n
( xi xi 1 ) .
i 1 b a
Therefore it follows from the Squeeze Theorem that f R[a, b] . This completes the proof.
Monotone functions are not necessarily continuous at every point, but they are also Riemann
integrable. Before proving this result, we recall Characterization Theorem for Intervals.
Characterization Theorem for Integrals: If S is a subset of that contains at least two points
and has the property
if x, y S and x y then [ x, y ] S ,
then S is an interval.
Theorem If f :[a, b] is monotone on [a, b] , then f R [a, b] .
Proof. Suppose that f is increasing on the interval [a, b] , a b . If 0 is given, we let q be
such that
f (b) f (a)
h .
q ba
Let yk f (a ) kh for k 0,1,..., q and consider sets Ak f 1 ([ yk 1 , yk )) for k 1,..., q 1 and
Aq f 1 ([ yq 1 , yq ]) . The sets Ak are pairwise disjoint and have union a, b . The
Characterization Theorem implies that each Ak is either (i) empty, (ii) contains a single point, or
(iii) is a nondegenerate interval (not necessarily closed) in [a, b]. We discard the sets for which
(i) holds and relabel the remaining ones. If we adjoin the endpoints to the remaining intervals
Ak , we obtain closed intervals I k . It can be shown that relabeled intervals Ak qk 1 are pair
wise disjoint, satisfy [a, b] qk 1 Ak and that f ( x) [ yk 1 , yk ] for x Ak .
We now define step functions and on [a, b] by setting
( x) yk 1 and ( x) yk for x Ak for x Ak
It is clear that ( x) f ( x) ( x) for all x [a, b] and that
q
( ) ( yk yk 1 )( xk xk 1 ) .
b
a
k 1
q
h ( xk xk 1 ) h (b a ) .
k 1
Since 0 is arbitrary, the Squeeze Theorem implies that f R [a, b] . This completes the
proof.
Additivity Theorem: Let f :[a, b] and let c (a, b) . Then f R[a, b] if and only if its
restrictions to [a, c] and [c, b] are both Riemann integrable. In this case
b c b
a
f f f
a c
… (*)
Proof. () Suppose that the restriction f1 of f to [a, c] , and the restriction f 2 of f to [c, b] are
Riemann integrable to L1 and L2 , respectively. Then, given 0 there exists 0 such that if
P1 is a tagged partition of [a, c] with P1 , then S ( f1 ;P1 ) L1 / 3 .
Also there exists 0 such that if P2 is a tagged partition of [c, b] with P2 then
S ( f ; Q ) ( L1 L2 ) . … (**)
(i) If c is a partition point of Q , we split Q into a partition Q1 of [a, c] and a partition Q2 of [c, b]
. Since
S ( f ; Q ) S ( f ; Q1 ) S ( f ; Q2 ) ,
and since Q1 has norm and Q2 has norm , the inequality (**) is clear.
(ii) If c is not a partition point in Q {( I k , tk )}km1 , there exists k m such that c ( xk 1 , xk ) . We let
Q be the tagged partition of [a, c] defined by
1
S ( f ; Q1 ) L1 and S ( f ; Q2 ) L2 ,
3 3
from which we obtain (**). Since 0 is arbitrary, we infer that f R [a, b] and that (6) holds.
() We suppose that f R [a, b] and, given 0 , we let 0 satisfy the Cauchy Criterion.
Let f1 be the restriction of f to [a, c] and let P1 , Q1 , be tagged partitions of [a, c] with P1
and Q1 . By adding additional partition points and tags from [c, b] , we can extend P1 and
Q1 to tagged partitions P and Q of [a, b] that satisfy P and Q . If we use the same
additional points and tags in [c, b] for both P and Q , then
S ( f1 ;P1 ) S ( f1; Q1 ) S ( f ;P ) S ( f ; Q ).
Since both P and Q have norm , then S ( f1 ;P1 ) S ( f ; Q1 ) . Therefore the Cauchy
Condition shows that the restriction f1 of f to [a, c] is in R [ a, c ] . In the same way, we see that
the restriction f 2 of f to [c, b] is in R [c, b] .
The equality (*) now follows from the first part of the theorem. This completes the proof.
Corollary If f R [a, b] , and if [c, d ] [a, b] , then the restriction of f to [c, d ] is in R [c, d ] .
Proof. Since f R [a, b] and c [a, b] , it follows from the theorem that its restriction to [c, b] is
in R [c, b] . But if d [c, b] , then another application of the theorem shows that the restriction of
f to [c, d ] is in R [c, d ] . This completes the proof.
Corollary If f R [a, b] and if a c0 c1 cm b , then the restrictions of f to each of the
subintervals [ci 1 , ci ] are Riemann integrable and
m
f
b ci
a
i 1
ci 1
f.
Proof. If any two of the numbers , , are equal, then (***) holds. Thus we may suppose that
all three of these numbers are distinct.
For the sake of symmetry, we introduce the expression
L( , , ) f f .
It is clear that (***) holds if and only if L( , , ) 0 . Therefore, to establish the assertion, we
need to show that L 0 for all six permutations of the arguments , and .
We note that the Additivity Theorem implies that
L( , , ) 0 when .
But it is easily seen that both L( , , ) and L( , , ) equal L( , , ) . Moreover, the numbers
L( , , ) , L( , , ) and L( , , )
are all equal to L( , , ) . Therefore, L vanishes for all possible configurations of these three
points. This completes the proof.
Exercises
1. Consider the function h defined by h( x) x 1 for x [0,1] rational, and h( x) 0 for
x [0,1] irrational. Show that h is not Riemann integrable.
2. Let f :[a, b] . Show that f R [a, b] if and only if there exists 0 0 such that for every
n there exist tagged partitions Pn and Qn with Pn 1/ n and Qn 1/ n such that
S ( f ;Pn ) S ( f ; Qn ) 0 .
3. Let H ( x) k for x 1/ k (k ) and H ( x) 0 elsewhere on [0,1]. Use Exercise 1 to show that
H is not Riemann integrable.
4. If ( x) x and ( x) x and if ( x) f ( x) ( x) for all x [0,1] , does it follow from the
Squeeze Theorem that f R[0,1] ?
5. If J is any subinterval of [a, b] and if J ( x) 1 for x J and J ( x) 0 elsewhere on [a, b] , we
say that J is an elementary step function on [a, b] . Show that every step function is a linear
combination of elementary step functions.
6. If :[a, b] takes on only a finite number of distinct values, is a step function?
7. If S ( f ;P ) is any Riemann sum of f :[a, b] , show that there exists a step function
a
b
8. Suppose that f is continuous on [a, b] , that f ( x) 0 for all x [a, b] and that f 0 . Prove
a
f (c ) g (c ) .
11. If f is bounded by M on [a, b] and if the restriction of f to every interval [c, b] where c ( a, b)
b b
is Riemann integrable, show that f R [ a, b] and that c
f f as c a .
a
12. Show that g ( x) sin(1/ x) for x (0,1] and g (0) 0 belongs to R [0,1] .
13. Give an example of a function f :[ a, b] that is in R [c, b] for every c ( a, b) but which is
not in R [ a, b] .
14. Suppose that f :[ a, b] , that a c0 c1 ... cm b and that the restrictions of f to [ci 1 , ci ]
m
f
b ci
belong to R [ci 1 , ci ] for i 1,..., m . Prove that f R [ a, b] and that the formula a
i 1
ci 1
f
in Corollary holds.
15. If f is bounded and there is a finite set E such that f is continuous at every point of [ a, b] \ E ,
show that f R [a, b] .
16. If f is continuous on [ a, b] , a b , show that there exists c [ a, b] such that we have
b
a
f f (c)(b a ) . This result is sometimes called the Mean Value Theorem for Integrals.
17. If f and g are continuous on [ a, b] and g ( x ) 0 for all x [a, b] , show that there exists
b b
c [a, b] such that a
f g f (c) g . Show that this conclusion fails if we do not have
a
f
b 1/ n
18. Let f be continuous on [ a, b] , let f ( x) 0 for all x [a, b] , and let M n n
. Show that
a
20. Suppose that f :[ a, b] and that n . Let Pn be the partition of [ a, b] into n subintervals
having equal lengths, so that x ; a i (b a) / n for i 0,1,..., n . Let L ( f ) S ( f ;P ) and
i n n ,l
Rn ( f ) S ( f ;Pn,r ) , where Pn,l has its tags at the left endpoints, and Pn , r has its tags at the
right endpoints of the subintervals [ xi 1 , xi ] .
a) If f is increasing on [ a, b] , show that Ln ( f ) Rn ( f ) and that
(b a )
0 Rn ( f ) Ln ( f ) ( f (b) f ( a )
n
b
b) Show that f (a)(b a) Ln ( f ) f Rn ( f ) f (b)(b a ) .
a
/2 /2
21. If f is continuous on [1,1] , show that f (cos x)dx f (sin x)dx 12 f (sin x)dx . [Hint:
0 0 0
Answers
1. If the tags are all rational, then S (h;P ) 1 , while if the tags are all irrational, then S (h;P ) 0 .
3. Let P be the partition of [0,1] into n equal subintervals with t 1/ n and Q be the same
n 1 n
6. Not necessarily.
8. If f (c ) 0 for some c ( a, b) , there exists 0 such that f ( x) 12 f (c) for x c . Then
b c
a
f
c
f (2 ) 12 f (c) 0 . If c is an endpoint, a similar argument applies.
10. Use Bolzano’s Intermediate Value Theorem (stated in chapter 2).
11. Let c ( x) M and c ( x) M for x [a, c) and c ( x) wc ( x) f ( x) for x [c, b] . Apply the
Squeeze Theorem for c sufficiently near a.]
12. Indeed, g ( x) 1 and is continuous on every interval [c,1] where 0 c 1 . The preceding
exercise applies.
13. Let f ( x) 1/ x for x (0,1] and f (0) 0 .
b
16. Let m inf f ( x) and M sup f . Then we have m(b a ) a
f M (b a) . By Bolzano’s
b
2n 2n
22. Note that x f ( x ) is an even continuous function.
2
6
FUNDAMENTAL THEOREMS OF CALCULUS
In this chapter we will explore the connection between the notions of derivative and integral.
The First form of Fundamental Theorem of Calculus: If f is continuous at every point of
[ a, b] and F is any antiderivative of f on [ a, b] , then
b
a f ( x)dx F (b) F (a) .
The First Form of the Fundamental Theorem provides a theoretical basis for the method of
calculating an integral. It asserts that if a function f is the derivative of a function F, and if f
b
belongs to R [ a, b] , then the integral a
f can be calculated by means of the evaluation
F (b) F ( a ) . A function F such that F ( x) f ( x) for all x [a, b] is called an antiderivative or a
primitive of f on [ a, b] . Thus, when f has an antiderivative, it is a very simple matter to calculate
its integral.
In practice, it is convenient to allow some exceptional points c where F (c) does not exist in
, or where it does not equal f (c) . It turns o
ut that we can permit a finite number of such exceptional points.
We recall Mean Value Theorem.
Mean Value Theorem: Suppose that f is continuous on a closed interval I [a, b] , and that f has
a derivative in the open interval (a, b) . Then there exists at least one point c in (a, b) such that
f (b) f (a) f (c)(b a) .
The following is a result needed in the coming section.
Theorem If f : I has a derivative at c I , then f is continuous at c.
Fundamental Theorem of Calculus (First Form): Suppose there is finite set E in [ a, b] and
functions f , F :[a, b] such that:
a) F is continuous on [ a, b] ,
b) F ( x) f ( x) for all x [a, b] \ E ,
Then we have
b
a
f F (b) F (a ) … (1)
Proof. We will prove the theorem in the case where E {a, b} . The general case can be obtained
by breaking the interval into the union of a finite number of intervals.
Let 0 be given. By assumption (c), f R [ a, b] . Then there exists 0 such that if P is
any tagged partition with P , then
S ( f ;P ) f .
b
… (2)
a
Now let Pu {([ xi 1 , xi ], ui )}in1 , then the sum on right equals S ( f ;Pu ) . If we substitute
F (b) F (a) S ( f ;P ) into (2), we conclude that
u
b
F (b) F (a ) f .
a
But, since 0 is arbitrary, we infer that equation (1) holds. This completes the proof.
Remark If the function F is differentiable at every point of [ a, b] , then (by Theorem C)
hypothesis (a) is automatically satisfied. If f is not defined for some points c E , we take
f (c ) 0 . Even if F is differentiable at every points of [ a, b] , condition (c ) is not automatically
satisfied, since there exist function F such that F is not Riemann integrable (This is illustrated
in the following Example 5).
Example If F ( x) 12 x 2 for all x [a, b] , then F ( x) x for all x [a, b] . Further, f F is
continuous so it is in R[ a, b] . Therefore the Fundamental Theorem (with E ) implies that
b
a
x dx F (b) F (a ) 12 (b 2 a 2 ) .
Example If G ( x) Arc tan x for x [a, b] , then G ( x) 1/( x 2 1) for all x [a, b] ; also G is
continuous, so it is in R[ a, b] . Therefore the Fundamental Theorem (with E ) implies that
b 1
a x 2 1dx Arc tan b Arc tan a .
Example If A( x) x for x [10,10] , then A( x) 1 if x [10,0) and A( x) 1 for x (0,10] .
Then we have A( x) sgn( x) for all x [10,10] \ {0} , where
1 for x 0
sgn( x) 0 for x 0 .
1 for x 1
(sgn is called the signum function).
Since the signum function is a step function, it belongs to R [ 10,10] . Therefore the
Fundamental Theorem (with E {0} ) implies that
10
10
sgn( x)dx A(10) A(10) 10 10 0 .
Thus K is continuous and differentiable at every point of [0, 1] . Since the first term in K is
continuous on [0, 1] , it belongs to R [0, 1] . However, the second term in K is not bounded, so it
does not belong to R [0, 1] . Consequently K R [0,1] , and the Fundamental Theorem does not
apply to K .
Additivity Theorem: Let f :[a, b] and let c (a, b) . Then f R[a, b] if and only if its
restrictions to [a, c] and [c, b] are both Riemann integrable. In this case
b c b
a
f f f
a c
Theorem The indefinite integral F defined by
z
F ( z ) f for z [a, b]
a
z
F ( z ) F ( w) w
f M zw .
Now on the interval [c, c h] the function f satisfies inequality (4), so that (by Theorem D) we
have
ch ch ch
c
( f (c ) )
c
f
c
( f (c) ).
ch
Since
c
f F (c h) F (c) , the above implies
( f (c ) ) h F (c h ) F (c ) ( f (c ) ) h
If we divide by h 0 and subtract f (c) , we obtain
F (c h) F (c )
f (c )
h
which implies
F (c h ) F (c )
f (c ) .
h
But, since 0 is arbitrary, we conclude that the right-hand limit is given by
F (c h ) F (c )
lim f (c ) .
h 0 h
It is proved in the same way that the left-hand limit of this difference quotient also equals f (c)
when c (a, b] . This completes the proof.
If f is continuous on all of [ a, b] , we obtain the following result.
Theorem If f is continuous on [ a, b] then the indefinite integral F, defined by
z
F ( z) f for z [ a, b] ,
a
1 x x 1 .
Thus f R [1,1] and has the indefinite integral1 F ( x) x 1 with the base point 1 . However,
since F (0) does not exist, F is not an antiderivative of f on [1,1] .
Example If h denotes Thomae’s function h :[0,1] defined by defined by h( x) 0 if x [0,1]
is irrational, h(0) 1 and by h( x) 1/ n if x [0,1] is the rational number x m / n where
x
m, n have no common integer factors except 1. Then its indefinite integral H ( x) h is
0
identically 0 on [0, 1] . Here, the derivative of this indefinite integral exists at every point and
H ( x) 0 . But H ( x) h( x) whenever x [0,1] , so that H is not an antiderivative of h on [0,1]
.
Substitution Theorem
The next theorem provides the justification for the “change of variable” method that is often
used to evaluate integrals. This theorem is employed (usually implicitly) in the evaluation by
means of procedures that involve the manipulation of “differentials”.
Substitution Theorem: Let J [, ] and let : J have a continuous derivative on J. If
f : I is continuous on an interval I containing ( J ) , then
( )
f ((t )) (t ) dt
( )
f ( x ) dx … (5)
u
Proof. Define F (u ) f ( x ) dx for u I , and H (t ) F ( (t )) ( F ) (t ) for t J . Then
( )
The hypotheses that f and are continuous are restrictive, but are used to ensure the existence
of the Riemann integral on the left side of (5).
Example Evaluate the integral
4 sin t
1 t dt .
Here we substitute (t ) t for t [1, 4] so that (t ) 1/(2 t ) is continuous on [1, 4] . If we let
f ( x) 2sin x , then the integrand has the form ( f ) and the Substitution Theorem implies
2
that the integral equals 2sin x dx 2cos x |12 2(cos1 cos 2) .
1
sin t 4
Example Consider the integral
t
dt . Since (t ) t does not have a continuous
0
derivative on [0, 4] , the Substitution Theorem is not applicable, at least with this substitution.
Exercises
1. Extend the proof of the Fundamental Theorem to the case of an arbitrary finite set E.
2. If n and H n ( x) x n 1 /(n 1) for x [a, b] , show that the Fundamental Theorem implies
b
that x n dx (b n 1 a n 1 ) /(n 1) . What is the set E here?
a
1 2
3. If g ( x) x for x 1 and g ( x) x for x 1 and if G ( x) x 1 , show that
2
3
2
g ( x) dx G (3) G ( 2) 5 / 2 .
1 1 b
4. Let B( x) x 2 for x 0 and B( x) x 2 for x 0 . Show that
2 2 a
x dx B(b) B(a) .
5. Let f :[ a, b] and let C .
(a) If :[ a, b] is an antiderivative of f on [ a, b] , show that
C ( x) ( x) C is also an antiderivative of f on [ a, b] .
(b) If 1 and 2 are antiderivative of f on [ a, b] , show that
1 2 is a constant function on [ a, b] .
z
6. If f R [ a, b] and if c [ a, b] , the function defined by Fc ( z ) f for z [a, b] is called the
c
x [ x] is called the greatest integer function, for example, [7.2] 7 , [ ] 3 ,
[2.8] 3 ).
x
9. Let f R [ a, b] and define F ( x) f for x [a, b] .
a
x
(a) Evaluate G ( x) f in terms of F, where c [ a, b] .
c
b
(b) Evaluate H ( x) f in terms of F.
x
sin x
(c) Evaluate S ( x) f in terms of F.
x
x2 x
a) F ( x ) (1 t 3 ) 1 dt b) F ( x ) 1 t 2 dt
0 x2
.
15. Use the Substitution Theorem to evaluate the following integrals.
1 2
a) t 0
1 t 2 dt b) 0
t 2 (1 t 3 ) 1/ 2 dt
1 t 4 4 cos t
c)
1
t
dt , d)
1
t
dt
16. Sometimes the Substitution Theorem cannot be applied but the following result, called the
Second Substitution Theorem is useful. In addition to the hypotheses of Substitution
Theorem, assume that (t ) 0 for all t J , so the function : ( J ) inverse to exists
and has derivative ( (t )) 1/ (t ) . Then
( )
f ( (t )) dt
( )
f ( x) ( x) dx
so that G(t ) f ( (t )) . Note that K ( x) G ( ( x)) is differentiable on the interval ( J ) and that
K ( x) G( ( x)) ( x) f ( ( x)) ( x) f ( x) ( x) .
Calculate G ( ) K ( ( )) in two ways to obtain the formula.
17. Apply the Second Substitution Theorem to evaluate the following integrals.
9 dt 3 dt
a) b) 1n(3 2 2) 1n3
1
2 t 1
t t 1
tdt 4 dt
1 t (t 4) Arctan(1) Arctan(1/ 2)
4
c)1 1 t d)
18. Explain why Substitution Theorem and/or Second shifting Theorem cannot be applied to
evaluate the following integrals, using the indicated substitution.
4 tdt cos tdt
4
a) 0
1 t
(t ) t b) 0
t
(t ) t
1 dt
d)
1
c)
1
1 2 t dt (t ) t
0
1 t2
(t ) Arcsin t
7
POINTWISE AND UNIFORM CONVERGENCE
Let A be given and suppose that for each n there is a function f n : A ; we say that
( f n ) is a sequence of functions on A to .
Clearly, for each x A , a sequence of functions gives rise to a sequence of real numbers,
namely the sequence
( f n ( x)) , … (1)
obtained by evaluating each of the functions at the point x. For certain values of x A the
sequence (1) may converge, and for other values of x A this sequence may diverge. For each
x A for which the sequence (1) converges, there is a uniquely determined real number
lim( f n ( x )) . In general, the value of this limit, when it exists, will depend on the choice of the
point x A . Thus, there arises in this way a function whose domain consists of all numbers
x A for which the sequence (1) converges. Definition follows:
Let ( f n ) be a sequence of functions on A to , let A0 A , and let f : A0 . We say that
the sequence ( f n ) converges on A0 to f if, for each x A0 , the sequence ( f n ( x)) converges to
f ( x) in . In this case we call f the limit on A0 of the sequence ( f n ) . When such a function f
exists, we say that the sequence ( f n ) is convergent on A0 , or that ( f n ) converges pointwise on
A0 .
Except for a possible modification of the domain A0 , the limit function is uniquely
determined. Ordinarily we choose A0 to be the largest set possible; that is, we take A0 to be the
set of all x A for which the sequence (1) is convergent in .
If the sequence ( f n ) converges on A0 to f, we denote it by
f lim( f n ) on A0 , or f n f on A0 .
Sometimes, when f n and f are given by formulas, we write
f ( x) lim f n ( x) for x A0 , or f n ( x) f ( x) for x A0
We need the following result.
Theorem Let Y ( yn ) be a sequence of real numbers that converge to y and let x . Then the
sequence
xY ( xyn )
converge to xy . i.e.,
lim( xyn ) x lim( yn ).
Example Show that lim( x / n) 0 for x .
For n , let f n ( x) x / n and let f ( x) 0 for x .
For x , it follows that
lim( f n ( x)) lim( x / n)
x lim(1/ n), using Theorem A
x 0, as lim(1/ n) 0
x
Then (hn ( x)) h( x) for all x .
Example Show that lim(( 1n )sin(nx n)) 0 for x (Fig. 4).
Let Fn ( x) ( 1n )sin( nx n) for x , n , and let F ( x) 0 for x . Since sin y 1 for all
y we have
1 1
Fn ( x) F ( x) sin( nx n) … (2)
n n
for all x . Therefore it follows that lim( Fn ( x)) 0 F ( x) for all x .
It should note that, given any 0 , if n is sufficiently large, then Fn ( x) F ( x) for all values
of x is simultaneously!
Getting an intuition from the remark above, we have the following reformulation of the
definition of pointwise convergence
Lemma 1 A sequence ( f n ) of functions on A to converges to a function f : A0 on
A0 if and only if for each 0 and each x A0 there is a natural number K ( , x) such that if
n K ( , x ) , then
f n ( x) f ( x) . … (3)
1
Example Let f n ( x) for x [0, ), n 1, 2, We show that ( f n ) converges point wise to
xn
f where f ( x) 0 for all x [0, ) .
1
For x 0, f n (0) 1.
n
For fixed x (0, ), it follows that
1
implies xn
1
implies x n.
Now if we choose
k
1
If nk k and xk 12
1
, then g nk ( xk ) , so that
1/ k 1 k
2
2
g nk ( xk ) g ( xk ) 1
2
0 12 .
1
Let 0 . Then, by the lemma above, the sequence ( g n ) doesnot converge uniformly on
4
( 1,1] to g.
x 2 nx
Example Let hn ( x) x , n . We have seen in an earlier example that (hn ) h for
n
all x , where h( x) x for all x .
(k )2 k ( k )
If nk k and xk k , then hnk ( xk ) 0 and h( xk ) k so that hnk ( xk ) h( xk ) k .
k
Let 0 12 k . Then, by the Lemma above, the sequence (hn ) doesnot converge uniformly on to
h.
Example Show that the sequence ( f n ( x)) where
n
f n ( x)
, x0
xn
is uniformly convergent in the closed bounded interval [0, m], whatever m may be, but not in
the interval 0 x .
Here for any given x 0 ,
f ( x ) lim f n ( x )
n
n 1
lim lim
xn ( x / n) 1
= 1.
Then for a given 0 ,
f n ( x) f ( x) ,
n
if 1
xn
x
i.e., if
xn
x
i.e., if
xn
1
i.e., if n x 1 .
Let K ( , x) the smallest integer greater than x (1/ 1) . Also, we note that K ( , x) increases
as x increases and tends as x . Hence it is not possible to choose a positive integer K ( )
such that
f n ( x) f ( x) for every n K ( )
and for every value of x in [0, ] . Hence the convergence is nonuniform in [0, ).
If we consider any finite interval [0, m] where m 0 is a fixed number, then the maximum
value of x (1/ 1) on [0, m] is m(1/ 1) so that if we take
K ( ) any integer greater than m(1/ 1) ,
Then
f n ( x) f ( x) for every n K ( ) and for every x in [0, m] .
This shows that the sequence ( f n ( x)) converges uniformly in the interval [0, m] where m is any
fixed positive number.
Example Show that x 0 is a point of nonuniform convergence of the sequence ( f n ( x)) in
0 x 1 , where
n2 x
f n ( x)
1 n4 x2
Solution
Here
n2 x x / n2
f ( x) lim ( f n ( x)) lim lim 2
1 n x (1/ n ) x
4 2 4
= 0 for 0 x 1
For a given , we have
f n ( x) f ( x) ,
n2 x
if
1 n4 x2
i.e., if n 4 x 2 n 2 x 0
x x 2 4 x 2 2
i.e., if n2
2 x 2
1 1 4 2
i.e., if n2 ..(1)
2x
Also (1) shows that if x 0 , n so that it is not possible to choose K ( ) such that
f n ( x) f ( x) for every n K ( ) and for every x [0, 1] .
Hence the sequence is nonuniformly convergent in [0, 1] .
But if we consider the interval [ m, 1] where 0 m 1 , the convergence is uniform since in
that case it is possible to take
1
1 1 4 2 2
K ( ) an integer just greater than .
2x
Hence x 0 is the point of nonuniform convergence of the given sequence in [0, 1] .
Example Test for uniform convergence the sequence e nx for x 0 .
Let f n ( x) e nx .
1 when x 0
Then lim( f n ( x)) lim (e nx )
0 when x 0
1 when x 0
Let f ( x)
0 when x 0
Then for each x [0, ) , lim ( f n ( x)) f ( x) . Hence ( f n ) converges point wise to f in [0, ) .
Now for any 0 and x (0, ) ,
f n ( x) f ( x) e nx
1
if e nx
1
i.e., if nx log
1 1
i.e., if n log .
x
Hence choosing
K ( , x) = the smallest integer greater than (1/ x) log (1/ )
we get, for all x (0, )
sn ( x) S ( x) for all n K ( , x).
Here, obviously, K ( , x) depends on x. We note that K ( , x) increases and tends to as x
tends to 0 i.e., K ( , x) is not a bounded function of x on (0, ) . Hence it is not possible to find
a natural number K ( ) , depending only on , such that
sn ( x) S ( x) whenever n K ( ) and x [0, )
Hence ( f n ) is not uniformly convergent on [0, ) .
If, however, we consider the interval [a, ) where a is any fixed real number greater than
zero, however small, then K ( , x) is a bounded function on [a, ) . The maximum value of
K ( , x) on [a, ) is the integer just greater than (1/ a ) log(1/ ) . Hence if we take
K ( ) the integer just greater than (1/ a ) log(1/ ) ,
then
f n ( x) f ( x) whenever n K ( ) and x [ a, )
Hence the sequence ( f n ) is uniformly convergent in [a, ) where a 0, but nonuniformly
convergent in [0, ) .
x
Example Show that if g n ( x) for x 0 , then ( g n ) converges uniformly on [0, ) .
nx 1
x x/n
For any x 0, lim( g n ( x)) lim lim 0
nx 1 x 1/ n
Hence ( g n ) converges point wise to g where
g ( x ) 0 for x 0 on [0, ) .
Now for any 0 and x (0, ) ,
g n ( x) g ( x)
x
if
nx 1
x
i.e., if nx 1
1 1
i.e., if n .
x
1 1
Take K ( , x) the positive integer just greater than .
x
Then for all x (0, ) ,
g n ( x) g ( x) for all n k ( , x)
1 1 1
Here K ( , x) = the positive integer just greater than is a bounded above by and is a
x
function of x on (0, ) .
The maximum value of K ( , x) on (0, ) is the integer just greater than 1/ . Hence if we take
K ( ) integer just greater than 1/ ,
then
g n ( x) g ( x) for all n K ( ) and x (0, )
But when x 0, g n ( x) 0 for all n , and hence we have
g n ( x) g ( x) for all n K ( ) and x [0, )
Hence the sequence ( g n ) is uniformly convergent on [0, )
Example Let G ( x) x n (1 x) for x A [0,1] . Then the sequence (Gn ( x)) converges to G ( x) 0
for each x A . To calculate the uniform norm Gn G Gn on A , we find the derivative and
solve
Gn ( x) x n 1 (n (n 1) x) 0
n
to obtain the point xn . This is an interior point of [0,1], and it is easily verified by using
n 1
the First derivative Test that Gn attains a maximum on [0,1] at xn . Therefore, we obtain
1
Gn Gn ( xn ) (1 1n ) n
A
n 1
which converges to ( e ) 0 0 . Thus we see that convergence is uniform on A.
1
By making use of the uniform norm, we can obtain a necessary and sufficient condition for
uniform convergence that is often useful.
n K ( 12 ) then f n f A
12 . Hence if both m, n K ( 12 ) , then we conclude that
f m ( x) f n ( x) f m ( x) f ( x) f n ( x) f ( x) 12 12
for all x A . Therefore f m f n A
, for m, n K ( 12 ) H ( ) .
() Conversely, suppose that for 0 there is H ( ) such that if m, n H ( ) , then
fm fn A
. Therefore, for each x A we have
f m ( x) f n ( x) f m f n A
for m, n H ( ) . … (8)
It follows that ( f n ( x)) is a Cauchy sequence in ; therefore, being a Cauchy sequence, it is a
convergent sequence. We define f : A by
f ( x) lim( f n ( x)) for x A .
From (8), we have for each x A
f m ( x) f n ( x) for m, n H ( ) .
Now fix m H ( ) and let n , then we obtain for x A ,
f m ( x) f ( x) .
Hence for each x A, we have
f m ( x) f ( x) for m H ( ).
Therefore the sequence ( f n ) converges uniformly on A to f. This completes the proof.
Exercises
1. Show that lim( x /( x n)) 0 for all x , x 0
8
UNIFORM CONVERGENCE AND CONTINUITY
We first list some examples which shows that, in general, limit of a sequence of continuous
functions need not be continuous. Similarly, limit of a sequence of differentiable (resp.,
Riemann integrable) functions need not be differentiable (resp., Riemann integrable). In this
text we discuss the limit of a sequence of continuous functions only.
We will consider results that reveal the importance of uniform convergence that allows
interchange of limits which makes the limit of sequence of continuous function also continuous.
Let g n ( x) x n for x [0,1] and n . Then, the sequence ( g n ) converges pointwise to the
function
0 for 0 x 1
g ( x)
1 for x 1
Although all of the functions g n are continuous at x 1 , the limit function g is not continuous at
x 1 . Recall that it was shown in Example 9 in the previous chapter that this sequence does not
converge uniformly to g on [0,1] .
Each of the functions g n ( x) x n for x [0,1] and n has a continuous derivative on [0,1] . For
x [0, 1], g n ( x) nx n 1 . However, the limit function g does not have a derivative at x 1 , since
it is not continuous at that point.
Let f n :[0,1] be defined for n 2 by
n 2 x for 0 x 1n
f n ( x) n 2 ( x n2 ) for 1n x n2
0 for n2 x 1
It is clear that each of the functions f n is continuous on [0,1] ; hence it is Riemann integrable.
1
We note that for x [0,1] , f n ( x) 0 as n . Also, for n 2 ,
0
f n ( x)dx 1. This can be
described as below:
1 1/ n 2/ n 1
0
f n ( x)dx
0
n 2 x dx
1/ n
n 2 ( x n2 )dx 0 dx
2/ n
1/ n 2/n
n x2 n ( x2 n2 x)
2 2 2 2
0 1/ n
1 1
1 for n 2
2 2
2
If x [0, 1] and n , we have x 1 so that f n ( x) 0 as n . Hence the limit
n
function f is such that f ( x) 0 for x [0, 1] . Being a constant function, f is continuous on
1
[0, 1] and hence is Riemann integrable on [0, 1] and f ( x) dx 0 . Therefore we have
0
1 1
0
f ( x) dx 0 1 lim f n ( x) dx
0
In this example,
1 1
lim f
0 n ( x) dx lim f n ( x) dx,
0
2nx
For x [0,1] , hn ( x) 2 0 as n . Hence if we take h ( x ) lim( hn ( x )) , we have
enx
h( x) lim(hn ( x)) 0 for all x [0,1] .
Hence
h( x)dx lim h ( x) dx .
1 1
n
0 0
nx
Let f n ( x) for all x [0,1] n 1, 2,3,...
1 n2 x2
Then
nx x/n
lim( f n ( x)) lim 2 2
lim 0 for all x [0, 1].
1 n x 12 x 2
n
Let us define f :[0,1] by
f ( x) 0 for all x [0, 1] .
Then ( f n ) converges pointwise to f on [0, 1]. Here clearly each f n and f are continuous on
[0,1] . But
fn f [0,1]
sup f n ( x) f ( x) : x [0,1]
nx
sup : x [0,1] … (1)
1 n x
2 2
nx nx
To find sup : x [0,1] i.e., the maximum value of on the set [0, 1] , we can use
1 n x
2 2
1 n2 x2
nx
the Differentiation Method. By the method, the function y , attain an extremum when
1 n2 x2
y 0
(1 n 2 x 2 )n nx 2n 2 x
i.e., when 0
(1 n 2 x 2 ) 2
i.e., when n n 3 x 2 2n 3 x 2 0 .
1
i.e., when x 2
n2
1 nx 1
i.e., when x . Hence y attains a maximum in [0,1] at x and the maximum
n 1 n x
2 2
n
1
n
n 1 . Hence from (1), we have f f 1 1
value is n . Hence f n f [0,1] as
1 [0,1]
2 2
1 n2 2 2
n
n .
Since f n f [0,1] does not tend to zero as n tends to infinity, ( f n ) is not uniformly convergent to
f on [0,1] .
Exercises
1. Show that the sequence (( x n /(1 x n )) does not converge uniformly on [0, 2] by showing that
the limit function is not continuous on [0, 2] .
2. Let f n :[0,1] be defined for n 2 by
n 2 x for 0 x 1/ n
2
f n ( x) n 2 ( x ) for 1/ n x 2 / n
n
0 for 2 / n x 1
Prove that the sequence in ( f n ) is an example of a sequence of continuous functions that
converges non-uniformly to continuous limit.
3. Construct a sequence of functions on [0,1] each of which is discontinuous at every point of
[0,1] and which converges uniformly to a function that is continuous at every point.
4. Suppose ( f n ) is a sequence of continuous functions on an interval I that converges uniformly
on I to a function f. If ( xn ) I converges to x0 I , show that lim( f n ( xn )) f ( x0 ) .
5. Let f : be continuous on and let f n ( x) f ( x 1/ n) for x . Show that ( f n )
converges uniformly on to f.
6. Let f n ( x) 1/(1 x) n for x [0,1] . Find the pointwise limit f of the sequence ( f n ) on [0,1] . Does
( f n ) converges uniformly to f on [0,1] ?
7. Suppose the sequence ( f n ) converges uniformly to f on the set A and suppose that each f n is
bounded on A. (That is, for each n there is a constant M n such that f n ( x) M n for all x A .)
Show that the function f is bounded on A.
8. Let f n ( x) nx /(1 nx 2 ) for x A [0, ) . Show that each f n is bounded on A, but the
pointwise limit f of the sequence is not bounded on A. Does ( f n ) converge uniformly to f on
A?
9
SERIES OF FUNCTIONS
If ( f n ) is a sequence of functions defined on a subset D of with values in , the sequence of
partial sums ( sn ) of the infinite series f n is defined for x in D by
s1 ( x) f1 ( x)
s2 ( x) s1 ( x) f 2 ( x)
sn 1 ( x) sn ( x) f n 1 ( x)
In case the sequence ( sn ) of functions converges on D to a function f, we say that the infinite
series of functions f n converges to f on D. Notation We will often write
f n for f
n 1
n
is not uniformly convergent on [0, 1]. Discuss the continuity of the sum function.
x4 N N
x4
Let f n ( x)
(1 x 4 ) n 1
and s N ( x )
n 1
f n ( x ) 4 n 1
n 1 (1 x )
.
Hence
N
x4
sN ( x) 4 n 1
n 1 (1 x )
1
x 4 1 4 N
(1 x ) (1 x 4 ) N 1
1
1 (1 x 4 ) N 1
1 x4
1
1 x4 if x 0 . …(1)
(1 x 4 ) N 1
When x 0 , f n (0) 0 for all n and hence sN ( x) 0 . Hence sN (0) 0 0 as N .
For x 0 , using (1),
1
sN ( x) 1 x 4 N 1 1 x 4 as N .
(1 x 4 )
1 x 4 , x 0
Hence if f ( x)
0 x0
then, ( sN ) converges point wise to f and hence f n converges pointwise to f on [0, 1] .
Each f n ( x) is a continuous function on [0, 1] . But its limit f ( x) is not continuous at x 0 .
Hence (with the aid of Theorem 1) f n does not converge uniformly on [0, 1] .
Example We now show that the series
x
(nx 1) [(n 1) x 1]
is uniformly convergent on any interval [ a, b], 0 a b , but only point wise on [0, b] .
For each n , let
x 1 1
f n ( x) .
(nx 1) [(n 1) x 1] (n 1) x 1 (nx 1)
Then the sequence of partial sums of the series f n ( x) is given by
N
sN ( x) f n ( x)
n 1
N
1 1
n 1 ( n 1) x 1 n x 1
1
1 when x 0 (as other terms, occur in pairs with same
Nx 1
absolute value but with opposite signs and hence, get
cancelled)
When x 0 , f n (0) 0 for n
and hence sN ( x) 0 for all N .
Combining the above, we have
1
1 , if x 0
sN ( x) Nx 1
0 , if x 0
1
If x 0 lim ( sN ( x)) lim 1 1
Nx 1
and lim sN (0) 0.
Therefore ( sN ) and hence f n converges point wise to f where
1, x 0
f ( x)
0, x 0
For each n , f n ( x) is continuous on [0, b] but f ( x) is not continuous on [0, b] . Hence,
applying Theorem 1, f n does not converge uniformly on [0, b] .
Now let J [a, b] , where 0 a b . Then
sN f J
sup {S N ( x) f ( x) : x [a, b]}
1
sup 1 1 : x [ a, b] , since x 0
Nx 1
1
sup : x [ a , b ]
Nx 1
1
0 as N .
Na 1
Hence ( sN ) converges uniformly to f on [ a, b] . Hence the associated series f n also
converges uniformly to f on [ a, b] (where 0 a b.
x
n 1
n x1 x2 xn
converges if and only if for every 0 there exists an M ( ) such that if m n M ( ) , then
xn 1 xn 2 xm .
N
sN ( x) fi ( x) f1 ( x) f N ( x) for all x D .
i 1
m
i.e.,
i n 1
f i ( x) for all m n M ( ) and x D
i.e., f n 1 ( x) f n 2 ( x) f m ( x)
for all m n M ( ) and x D .
xn 1
Then f n ( x) for 1 x 1
n! n!
1 1 1
Noting that
n ! 2n
and the series 2 n
is convergent, we have, with the aid of comparison
1
test of series of real numbers, the series n! is also convergent.
Thus by Weierstrass’s M-test, it follows that the given series is uniformly convergent for all
values of x [1, 1] .
(1) n x 2 n 1 x3 x5
n 0 (2n 1)!
x
3! 5!
is uniformly convergent in every interval in [ a, b] .
Let M be any positive number greater than each of a and b so that for every x [a, b] , we
have
(1) n x 2 n 1 M 2 n 1
.
(2n 1)! (2n 1)!
we have
eM e M M3 M5
M ….. (2)
2 3! 5!
The identity (2) shows that the series (1) is convergent, whatever the positive constant
M may be. Hence by Weierstrass’s M-test, the given series
(1) n x 2 n 1
n 0 (2n 1)!
1. Discuss the convergence and the uniform convergence of the series f n , where f n ( x) is
given by
a) ( x 2 n 2 ) 1
b) (nx) 2 ( x 0)
c) sin( x / n 2 )
d) ( x n 1) 1 ( x 0)
e) x n /( x n 1) ( x 0)
f) (1) n (n x) 1 ( x 0)
10
IMPROPER INTEGRALS OF FIRST KIND - PART I
b
Evaluation of the definite integrals of the type a
f ( x) dx is required in many problems. So far
you may have been seen those definite integrals that have the following two properties:
1. The domain of integration, from a to b, is finite.
2. The range of the integrand is finite (in other words, the integrand is defined at
every point in the domain).
1
ln x 1 dx. Property 1 is violated in the first
Now consider the definite integrals 1 x2 dx and
0 x
definite integral while Property 2 is not satisfied at x 0 by the second integral. In this chapter
we consider the methods for the evaluation of such definite integrals.
An improper integral is a definite integral which does not satisfy Property 1 or 2 given above.
i.e., integrals with infinite limits of integration and integrals of functions that become infinite at
a point within the interval of integration are improper integrals. When the limits involved exist,
we can evaluate such integrals. We discuss this in this chapter in detail.
a
f ( x)dx
does not follow from the discussion on Riemann integration since the interval [a, ) is not
bounded. Such an integral is called an improper integral of first kind. The theory of this type
of integral resembles to a great extent the theory of infinite series.
We define
a
f ( x)dx
as follows:
If f R [a, s ] for every s > a, then
a
f ( x)dx
F (s)
f ( x) dx
a
(a s ).
Integrals of the type discussed in this section are sometimes called improper integrals of the
first kind.
1
Example We now show that the integral 2
dx is convergent.
1 x
s
1
With the usual notation, we have F ( s) 2
dx,
1 x
1
and then F ( s ) 1 and hence lim F ( s ) 1. Thus
s s
1
2
dx 1.
1 x
1
Example We now show that the integral dx diverges.
1 x
The integral
1
dx
1 x
diverges since
s
1
F (s) dx 2( s 1)
1 x
and lim F ( s ) does not exist.
s
ln x
Example We now valuate
1
x2
dx, if it exists.
1n x s
Here F (s) dx
x2 1
u ( x)dv( x) u ( x)v( x) a du ( x) v( x) .
b b
i.e.,
b
a a
ln s 1
1.
s s
Hence
ln s 1 ln s
lim F ( s ) lim 1 lim 0 1
s s
s s s s
s 1
0 1 1.
lnx
Hence 1 x2 dx lim s
F ( s) 1 .
Example We now Investigate the convergence of dx and dx .
1 x 1 x2
s
dx lim
dx lim ln s ln1 .
1 x s
x s 1
Hence the integral dx is divergent.
1 x
dx lim dx lim 1 +1 =1.
s
Now,
1 x
2 s 1 x 2 s s
Hence the integral dx is convergent and its value is 1.
1 x2
RESULT: If
a
f and
a
g
a
( f g)
a
f
a
g.
(2)
a
cf converges and
a
cf c
a
f.
1.
a
f ( x)dx converges if
a
g ( x)dx converges
2.
a
g ( x)dx diverges if a
f ( x)dx diverges.
Example We now nvestigate the convergence of e x dx
2
By definition,
b
e x dx lim e x dx
2 2
1 b 1
We cannot evaluate the latter integral directly because it is non-elementary. But we can show
b
that its limit as b is finite. We know that e x dx is an increasing function of b . Therefore
2
either it becomes infinite as b or it has a finite limit as b . It does not become infinite:
For every value of x 1 we have e x e x , Also,
2
b
e x dx lim e x dx lim e x lim e b e 1 e 1 0.36788.
b
1 b 1 b 1 b
1 b 1
converges to some definite finite value. We do not know exactly what the value is except that it
is something less than 0.37.
Example We no investigate the convergence of sin x dx .
2
1 x2
2
We know that 0 sin2 x 12 on [1, ) and 12 dx converges. Hence by Direct comparison test,
x x 1 x
sin 2 x
1 x 2 dx converges.
1
Example We now Investigate the convergence of dx .
1
x2 0.1
1
1 x 2 0.1dx diverges because x 2 0.1 x on [1, ) and 1 xdx diverges.
1 1 1
Again, since as N the right side of (4) diverges to infinity, we see that
s1
lim dx
s 1 x
1
RESULT:
a xp
dx , where p is a constant and a 0 , converges if p 1 and diverges if p 1 .
Proof.
s
1 s 1 x p 1
a xp
dx
s a x p
lim dx lim
s p 1
, provided p 1
a
1
lim s1 p a1 p , provided p 1
s 1 p
1
lim s1 p a1 p provided p 1
1 p s
0 if p 1,
But lim s1 p
s
if p 1.
1
Hence a xp
dx is convergent when p 1 and divergent when p 1 .
Also, when p 1 ,
1 s1
p
dx lim dx lim ln s ln a .
a x s a x s
1
Therefore, a xp
dx converges if p 1 and diverges if p 1 .
Example Show that a
e tx dx where t is a constant, converges if t 0 and diverges if t 0 .
s
s e tx
dx lim e dx lim
tx tx
e
a s a s
t n
1 1
lim e at e st e at lim e st
s t t s
0 if t 0,
But lim e ts
s
if t 0.
Hence 0
e tx converges if t 0 and diverges if t 0 .
s
When t 0 , a
e 0 x dx dx lim
a s
a
dx lim( s a) .
s
Therefore, converges if t 0 and diverges if t 0 .
tx
e
a
f ( x)
lim L (0 L )
x g ( x)
then a
f ( x)dx and
a
g ( x)dx both converge or both diverge.
dx and the Limit comparison Test, we now discuss the convergence of
Example Using 1 x2
dx . Also compare the values of each integral.
1 1 x2
With f ( x) 1/ x 2 and g ( x) 1/(1 x 2 ) , we have
2
f ( x)
lim lim 1/ x 2
x g ( x) x 1/(1 x )
x x x x
a positive finite limit. Therefore, dx converges because dx converges.
1 1 x2 1 x2
dx 1 .Also,
However, the integrals converge to different values. By an earlier Example,
1 x2
s
dx lim
dx lim tan 1 x s
1 1 x 2 s
1 1 x
2 s 1
1
We know that x dx converges. Now we use Limit Comparison Test with f ( x) 1x and
1 e e
g ( x) x 3 :
e 5
lim e x 5
x x
f ( x)
lim lim 1/xe
x g ( x ) x 3/(e 5) x 3e
= lim 1 5 x 1 0 1 ,
x 3 3e 3 3
a positive finite limit. As
0 x 3 1x ,
e 5 e
3 dx converges.
by Limit Comparison Test, we conclude that
1 e 5
x
dx
Example We now test for convergence the integral
1
x x2 1
1 1
Let f ( x) and g ( x) .
x x 1 2 x2
Then
f ( x) x2 1
1 0 , as x .
g ( x) x x 12
1
1 2
x
f ( x)
Since lim
x g ( x )
1 , a finite non zero number, by Quotient Test, 1
f ( x) dx and
1
g ( x ) dx
1
converge or diverge together. But
1
g ( x) dx
1 x2
dx converges. Hence
dx
1
f ( x) dx
1
x x2 1
converges.
x2
Example We now test for convergence the integral
1
x5 1
dx.
x2 1
Let f ( x) and g ( x) .
x 1
5
x
Then
f ( x) x5 / 2 1
1 as x .
g ( x) x 1
5
1 1/ x5
f ( x)
Since lim
x g ( x)
1 , a finite non zero number, by Quotient Test
1
f ( x) dx and
1
g ( x ) dx
1 1 1
converges or diverge together. But 1
g ( x) dx
1
x
dx 1/ 2 dx is divergent as 1 . Hence
1 x 2
x2
1
f ( x) dx
1
x5 1
is divergent.
Example We now test for convergence the integral e x dx
2
Since 0 is not a point of infinite discontinuity, we have to examine the convergence at only.
1
e x dx e x dx e x dx . …(1)
2 2 2
0 0 1
Since first integral on the right is a proper integral we need only test the convergence of
e x dx .
2
1
Hence e x 2 for all x 1 .
2
x
1
Since dx converges, by Comparison Test, we have e x dx converges. Hence using (1),
2
1 x2 1
being the sum of two convergent integrals, e x dx converges.
2
log x
Example We now test for convergence the integral
1 x2
dx.
log x 1
Let f ( x) and g ( x) .
x2 x
f ( x) x log x log x
Then as x .
g ( x) x 2 1 2/ x
1 log x
Here 1
g ( x) dx
1 x
dx diverges. Hence by Quotient test, 1
f ( x) dx
1 x2
dx diverges.
Definition If f R [a, s ]for every s a and if
a
| f ( x) | dx converges, we say that
a
f ( x) dx
converges absolutely.
RESULT: If
| f ( x) | dx
s
F (s)
a
and if F is bounded (above) on [ a, ), then lim F ( s ) exists and hence
s a
f ( x) dx converges
absolutely.
RESULT: If
a
f ( x) dx converges absolutely, if g [a, s ] for every s a, and if
| g ( x) | | f ( x)| ( a x ),
then
a
g ( x) dx converges absolutely.
Proof.
s s
G(s) | g ( x) | dx f ( x) dx | f ( x) | dx.
a a a
Hence G is bounded above on [a, ) and the absolute convergence of
a
g ( x) dx follows from the
preceding theorem.
RESULT: If
a
| f ( x)| dx converges absolutely, then
a
f ( x) dx converges
Proof.
f ( x) is either f ( x) or f ( x ) . In either case f ( x) f ( x) 0 and 2| f ( x)| f ( x) | f ( x) |.
Hence
0 f ( x) | f ( x) | 2| f ( x)| ( a x ), … (1)
and since (by assumption)
a
2 | f ( x) | dx converges.
s
a
f ( x) dx converges implies lim F ( s ) exists where F ( s ) f ( x) dx . Being a
s a
s
Hence by the theorem above, lim f ( x) f ( x) dx exists and hence f ( x) f ( x) dx
s a a
converges absolutely.
It follows that
a
f ( x) | f ( x) | dx
converges absolutely. Since f ( x) | f ( x) | 0 this means simply that
f ( x) | f ( x) | dx
a
converges. But, since
| f ( x)| dx converges, it follows (by subtraction) that
a
f ( x) | f ( x) | dx | f ( x) | dx f ( x) dx
a a a
cos x
Example Prove that 1 x2 1
dx converges.
cos x cos x 1
We have 2 2 for all x 1 . since cos x 1
x 1 x 1 x
2
dx cos x
Then, since 1 x2
converges, by comparison test, it follows that
1 x2 1
dx converges, i.e.,
cos x
1 x2 1
dx converges absolutely. Since, every absolutely convergent integral converges,
cos x
1 x 2 1 dx converges.
If
a
f ( x) dx converges but does not converge absolutely, we say that
a
f ( x) dx converges
conditionally.
sin x
Example Show that dx is a conditionally convergent improper integral. Also find the
x
sin x
value of dx .
x
sin x
To show that dx converges, we have for any s (using integration by parts)
x
s s
sin x 1 cos s cos x
dx dx. … (2)
x s x2
| cos x | 1
Now 2 ( x ).
x2 x
s
cos x
Since 2
dx converges (absolutely) it follows that dx converges absolutely and hence
x x2
converges. Thus as s all terms on the right of (2) approach limits. Thus
s
sin x
dx lim
x s
sin x
exists, which proves that dx converges.
x
Letting s , (2) gives
sin x cos x 1
dx dx .
x x 2
Note that the second integral is absolutely convergent while the first integral (as we shall now
show) is not.
sin x
Now we show that dx does not converge absolutely. For any N , we have
x
N 1 N 1
N ( n 1) ( n 1)
|sin x | | sin x | 1
dx dx | sin x | dx
x n 1
n x n 1
(n 1) n
1 N 1 1
n 1 0
| sin (u n) | du.
n 1
Now
sin (u n) sin u cos n cos u sin n sin u cos n.
Since cos n 1 this shows that |sin (u n) | | sin u | . Hence if 0 u , then
| sin(u n) | sin u. Thus
N | sin x | 1 N 1 1 2 N 1 1 2 N 1
dx
n 1 n 1 0 n 1
sin u du . … (3)
x k
n 1 k 2
1
Since the series k is divergent, the right side of (3) can be made as large as we please by
k 2
Exercises
In Exercises 1- , discuss the convergence of the following improper integral of the first kind.
x dx 1
1. 2. 2 dx
2
x 1
3 1
x3
x 1
2
1
3. 4 dx 4. 4 dx
0 x 1 1
x3
2
sin x x
5. dx 6. dx.
0 x 2 1 1 x2
log x 43 x 2
7. dx 8. 1 1 2 x 2 12 x 4 dx.
1 x e x
x dx x
9. 10. dx.
1 3x 4 5 x 2 1 2 1 x2
43 x 2 dx
11.
1
x cos x dx. 12. 1 1 2 x 2 12 x 4 dx
x dx 1
13. ` 14. 1
dx.
2 (log x)3 1
(1 x3 ) 2
x n 1 1
15.
0 1 x
dx 16. 1 1
dx.
(1 x3 ) 3
log x 1 cos x
17.
1 xa
dx 18.
0 x2
dx
x2 1 x 1
2
19.
2
x 6 16
dx 20. 4
0 x 1
dx
dx dx
21. 2
x 4
22. 1
x 3x 2
x 2 dx 2 sin x
23. 24. dx
x x 1 x2 1
2 5/ 2
x 1 1
25. Show that 0 (1 x) 3
dx
2 0 (1 x) 2
dx.
sin x cos x
26. Show that 1 x2
dx is convergent and deduce that 1 x
dx is convergent
1
x2 1
27. Show that 1 (1 x) 2
dx .
2 4
sin x
28. Show that the integral 0 xp
dx where p 1 is absolutely convergent.
29. True or false? If f is continuous on [1, ) and if
1
f ( x) dx converges, then lim f ( x) 0.
x
x sin x
30. Show that 0 1 x2
dx is convergent.
31. If 1
f ( x) dx converges and if lim f ( x ) L, prove that L=0.
x
but f (n) diverges.
n 1
sin x
34. Show that
xp 0
dx, 0 p 1 is convergent but not absolutely convergent.
sin bx
35. Discuss absolute convergence of dx .
1 x2
36. If f ( x ) 0 (1 x ), if f is nonincreasing on [1, ), and if
1
f ( x) dx converges,
then show that lim xf ( x ) 0.
x
cos x
37. Show that
1 x2
dx is conditionally convergent.
0
dx
41. Evaluate
x3
1
dx
42. Evaluate 3/ 2
1 x
dx
43. Evaluate
1
x
0
44. Evaluate e x dx .
dx
45. Evaluate
x x2
2
46. Show that 0
e x dx is convergent.
11
f (n) f ( x) dx f (n 1) f (k )
N
… (5)
1
n 1 n 1 k 2
If
1
f ( x) dx converges to A, then, by (5),
N
f (k )
N
f ( x)dx A.
1
k 2
The partial sums of k 2
f (k ) are thus bounded above. i.e., the sequence of Nth partial
sums of the series is bounded and hence the sequence of the Nth partial sum is convergent, and
hence the series k 2
f (k ) converges and hence k 1
f (k ) converges.
f ( x) dx diverges, the divergence of n 1 f (n) may be established in similar fashion,
If 1
n f (n).
1
2
n 1 n 1
1
Since 1
f ( x) dx
1 x2
dx
is convergent, it follows that n 1
f (n) converges.
If g ( x) 1/( x log x), then g is nonnegative and non-decreasing on [3, ) . Since G( x) g ( x) where
G ( x) log log x, then
s
3
g ( x) dx log log s log log 3,
and hence 3
g ( x) diverges. It follows from Integral Test for Series that n 1
[1/(n log n)]
diverges.
1 1 1 ... 1 ... ,
2 3 n
which is known to be divergent,
Hence we conclude that the series converges for p 1 but diverges for p 1.
a
Integrals of the form
f ( x ) dx
a
An integral of the form f ( x) dx may be treated by the same methods as those used on
a
integrals of the form
b
f ( x) dx. Thus, we say that
f ( x) dx converges to A if
a
lim f ( x) dx A.
s s
a
The change of variable x u will change a
problem into a
b
problem.
2 1
Example Examine the convergence or divergence of
1 x
dx.
For any s 2 we have
2 1 2 1 s 1
s 1 x s 1 u
dx ( 1) du 2 1 u du.
1 1 1 1
Since for 2 u , it follows, from the fact
1 u 2 u 1 x
dx is divergent, that
1 s
du lim
1 u s 2
2 1 2 1
does not exist. Hence lim dx does not exist, which proves that dx does not
s s 1 x 1 x
converge. (In this problem the divergence of
2 1
1 x dx
1
was deduced from the divergence of du. )
2 1 u
x3 x 2
Example Test for convergence x6 1 dx
We write
x3 x 2 0 x x
3 2
x x
3 2
Now
0 x3 x 2 0 ( y ) ( y )
3 2
y3 y 2
dy
0 y6 1
1 y3 y 2 y y
3 2
dy 1 y 6 1 dy
0 y6 1
y3 y 2 1
Taking f ( y ) and g ( y ) 3 , we have
y 1
6
y
f ( y) y3 ( y3 y 2 ) 1 1/ y
lim lim lim 1,
x g ( y) x y 1
6 x 1 1/ y 6
y3 y 2 1
a non zero finite number. Hence, by Quotient Test, 1 y6 1
dy and 1 y3
dy converge or
1 y3 y 2
diverge together. Since
1 y3
dy converges,
1 y6 1
dy converges.
1 y3 y 2 x3 x 2
0
Since
0 y6 1
dy is a proper integral, it follows that x6 1 dx converges.
x3 x 2
0
Similarly, we can show that x 6 1
dx converges.
x3 x 2
Hence x 6 1 , (being a sum of two convergent improper integrals) is convergent.
0 dx
Example Test the convergence of .
(1 3 x ) 2
0 dx 0 1
(1 3 x ) 2
lim
a a (1 3 x ) 2
dx
0
1 1 1
lim lim
a 3(1 3 x ) a 3 3(1 3a )
a
1 1 1 1 1 1
lim 0 .
3 3 a 1 3a 3 3 3
1
Hence, the given integral is convergent and has value .
3
0
Example Test the convergence of
cosh x dx .
0
cosh x dx lim cosh x dx lim sinh x
0 b
a a a a
ea e a
lim sinh 0 sinh a lim 0 .
a a
2
Hence, the given integral is divergent.
dx .
Example Evaluate 1 x2
dx 0 1 1
1 x2 1 x2 dx 0 1 x2 dx
0 1 b 1
lim dx lim dx
a a 1 x 2 b 0 1 x 2
0 b
lim tan 1 x lim tan 1 x
a a b 0
tan 1 0 tan 1 0 , since
2 2
lim tan 1 a and lim tan 1 b
a 2 b 2
2 2
Hence, the given integral is convergent and has value .
Example The cross sections of the solid horn in Fig.1 perpendicular to the x-axis are circular
disks with diameters reaching from the x-axis to the curve y e x , x ln 2. Find the
volume of the horn.
Solution
1
For a typical cross section, radius is y and area is
2
2
A( x) (radius) 2 1 y e 2 x .
2 4
ln 2
Now the volume of the horn in the Fig.1 is
A( x ) dx . i.e., the volume of the horn is the limit
as a of the volume Va of the portion from a to ln 2. By the method of slicing, the volume
Va of this portion is
ln 2
Va
ln 2
A( x)dx
ln 2 e 2 x dx e 2 x
a a 4 8
b
(eln 4 e 2 a ) (4 e 2 a ).
8 8
As a , e 0 and Va ( / 8)(4 0) / 2 . Hence
2a
ln 2
A( x ) dx lim Va .
a 2
i.e., the volume of the given horn is / 2 .
ex1
Example Test for convergence x dx .
Let x y . Then,
y y
ex
1 1e e
x dx y ( dy ) 1 y dy .
e y
Now,
y
e y for all y 1 and, by Example above with a 1 and t 1, we have
1
e y dy is
e y
convergent. Hence by comparison test, 1 y
dy converges. Therefore
y
ex
1 e
x dx 1 y dy
also converges.
Exercises
In Exercises 1- , discuss the convergence of the following improper integral of the first kind.
x dx 1
1. 2. 2
dx
2
x 1
3 1
x 3
x 1
2
1
3. 0 x4 1
dx 4. 1 4
dx
x3
sin 2 x
x
5. 0 x 2 dx 6. 1 x2
1
dx.
log x 43 x 2
7. dx 8. dx.
1 x e x 1 1 2 x 2 12 x 4
x dx x
9. 10. dx.
1 3x 5 x 2 1
4 2 1 x2
43 x 2 dx
11. x cos x dx. 12. dx
1 1 1 2 x 2 12 x 4
x dx 1
13. ` 14. 1
dx.
2 (log x)3 1
(1 x )
3 2
x n 1 1
15. 0 1 x
dx 16. 1 1
dx.
(1 x ) 3 3
log x 1 cos x
17. 1 xa
dx 18. x2
0
dx
x2 1 x 1
2
19.
x 6 16
2
dx 20. 4
0 x 1
dx
dx dx
21. 2
x 4
22. 1
x 3x 2
x 2 dx 2 sin x
23. 24. dx
x x 1 x2 1
2 5/ 2
x 1 1
25. Show that 0 (1 x) 3
dx
2 0 (1 x) 2
dx.
sin x cos x
26. Show that 1 x2
dx is convergent and deduce that 1 x
dx is convergent
1
x2 1
27. Show that 1 (1 x) 2
dx .
2 4
sin x
28. Show that the integral 0 xp
dx where p 1 is absolutely convergent.
29. True or false? If f is continuous on [1, ) and if
1
f ( x) dx converges, then lim f ( x) 0.
x
x sin x
30. Show that 0 1 x2
dx is convergent.
31. If 1
f ( x) dx converges and if lim f ( x ) L, prove that L=0.
x
f ( x) 0 (1 x ),
and such that f (n) converges
n 1
but 1
f ( x) dx diverges.
33. Give an example of a continuous function f such that
f ( x) 0 (1 x )
and such that 1
f ( x) dx converges
but f (n) diverges.
n 1
sin x
34. Show that dx, 0 p 1 is convergent but not absolutely convergent.
0 xp
sin bx
35. Discuss absolute convergence of dx .
1 x2
36. If f ( x ) 0 (1 x ), if f is nonincreasing on [1, ), and if
1
f ( x) dx converges,
then show that lim xf ( x ) 0.
x
cos x
37. Show that
1 x2
dx is conditionally convergent.
0
dx dx
41. Evaluate
1 x3 45. Evaluate x x2
2
dx
42. Evaluate 3/ 2
1 x
46. Show that 0
e x dx is convergent.
dx
43. Evaluate
1
x
0
44. Evaluate e x dx .
12
Real Analysis Page 107
School of Distance Education
IMPROPER INTEGRALS OF
SECOND AND THIRD KINDS – PART I
Improper Integrals of Second Kind
The definition of
1 1
x
dx 0
does not follow from the discussion on Riemann integration because the function f defined by
1
f ( x) (0 x 1)
x
is not bounded. Note, however, that f is bounded (and continuous) on [,1] for every 0. This
suggests treating
1 1 1 1
0 x dx as the
lim
0 x
dx
1
x2
1
b
Definition If f [ a , b] for all such that 0 b a, but f [a, b], we define
a
f ( x) dx
1
The integrand is unbounded at x 0 . Hence the given is an improper integral of second
x2
kind.
1
1 1 1 1
1 1
Hence dx lim dx lim lim 1 .
x2 x
0 2
0 x 0 0
dx
1
Hence 2
is divergent.
0 x
7
dx
Example Find the value of , if it converges.
1
3
x 1
7
dx
Hence converges and has value 6.
1
3
x 1
b
dx
Example Show that converges if p 1 and diverges if p 1 .
a ( x a) p
The given integral is a proper integral if p 0 and hence converges. So let p 0 . Then the
1
integrand is unbounded at x a .
( x a) p
b b
1 1
Hence, dx lim dx
a ( x a) p 0 a ( x a)
p
b
( x a ) p 1
lim , if p 1
0
p 1 a
1
lim (b a ) p 1 p 1 ,if p 1
0 1 p
1 1
(b a ) p 1 lim p 1
1 p 1 p 0
But lim p 1 0, if p 1
0
, if p 1 .
b 1
Hence, a ( x a) p
dx is convergent if p 1 and divergent if p 1 .
When p 1 ,
b 1 b 1
a ( x a) p
dx
a xa
dx , improper integral of second kind
b 1
lim dx
0 a xa
lim ln ( x a)a
b
0
lim ln (b a ) ln
0
ln (b a ) lim ln .
0
b 1
Since the limit on the right does not exist a ( x a) p
dx is not convergent at p 1 .
b 1
Remark As in the above example we can show that a (b x) p
dx converges if p 1 and
diverges if p 1 .
Both improper integrals
b dx b dx
a ( x a) p
and a (b x) p
are widely used as comparison integrals in testing convergence of improper integrals.
Convergence Tests
Direct Comparison Test: Let f and g be two positive functions, both are bounded at x a and
such that
f ( x) g ( x) for a x b .
Then
b b
(i) a
f ( x) dx converges if a
g ( x) dx converges.
b b
and (ii) a
g ( x) dx diverges if a
f ( x) dx diverges.
Limit Comparison Test/ Quotient Test: If f ( x) 0, g ( x) 0 for a x b , f ( x) and g ( x) are
f ( x)
unbounded at x a and if lim A , then
xa g ( x)
b
(a) If A 0 or i.e., if A is a non zero finite number, then the two integrals
a
f ( x) dx and
b
a
g ( x) dx converge or diverge together;
b b
(b) If A 0 and
a
g ( x) dx converges, then a
f ( x) dx converges;
and
b b
(c) If A and a
g ( x) dx diverges, then a
f ( x) dx diverges.
5 dx
Example Test for convergence the improper integral 1
x4 1
.
1 1
We have for all x 1 .
x 1
4
x 1
1 5 1 5
Also,
1
x 1
dx
1
x4 1
dx converges.
5 1
Hence by comparison test, dx converges
1
x4 1
6 log x
Example Test for convergence the improper integral 3 ( x 3) 4
dx
log x 1
We have for all x 3 .
( x 3) 4
( x 3) 4
6 1
Also,
3 ( x 3) 4
diverges since 4 1 .
6 log x
Hence, by comparison test, 3 ( x 3) 4
dx also diverges.
3 dx
Example Investigate the convergence of 2 x 2 ( x 3 8) 2 / 3
.
1 1
Let f ( x) and g ( x) .
x ( x 8)
2 32/3
( x 2) 2 / 3
Then,
f ( x) ( x 2) 2 / 3 1
2 3 2 2 , since x3 23 ( x 2) ( x 2 2 x 4)
g ( x) x ( x 2 ) 3 2/3
x ( x 2 x 4) 2 / 3
Hence
f ( x) 1 1
lim lim 2 2 3 ,
x2 g ( x) x 2 x ( x 2 x 4) 2/3
8 18
3 3
a non zero finite number. Hence by Quotient Test 2
f ( x) dx and 2
g ( x ) dx converges or
3 3 1
diverges together. Now, since 2
3
1, 2
g ( x) dx
2 ( x 2) 2 / 3
dx converges. Hence,
3 3 dx
2
f ( x) dx
2 x ( x 8) 2 / 3
2 3
also converges.
sin x
Example Investigate the convergence of 0 x3
dx
sin x
Let .f ( x)
x3
Since sin x 0 in [0, ], f ( x ) is non negative in [0, ] .
1
Take g ( x) .
x2
Then
f ( x) x 2 sin x sin x
.
g ( x) x3 x
f ( x) sin x
Hence lim
x 0 g ( x)
lim
x 0 x
1 , a non zero finite number. Hence by Quotient Test 0
f ( x) dx
1
and 0
g ( x) dx converge or diverge together. Now, 0
g ( x) dx
0 x2
dx diverges since 2 1 .
sin x
Hence, 0 x3
dx diverges.
5 dx
Example Investigate the convergence of 1
(5 x)( x 1)
The given integrand is unbounded at two points x 1 and x 5.
We write
5 dx 3 1 5 1
1
(5 x)( x 1)
1
(5 x)( x 1)
dx
3
(5 x)( x 1)
dx
3 1
We first consider the improper integral
1
(5 x)( x 1)
dx.
1 1
Let f ( x) and g ( x) .
(5 x)( x 1) x 1
Then
f ( x) x 1 1 1
lim lim lim ,
x 1 g ( x) x 1 (5 x)( x 1) x 1 5 x 2
3 3
a finite non zero number, and hence by the Quotient Test 1
f ( x) dx and
1
g ( x ) dx converge
3 3 1
or diverge together. Now, since 1
2
1, we have
1
g ( x) dx
1 ( x 1)1/ 2
dx converges and hence
3 3 1
1
f ( x) dx
1
(5 x)( x 1)
dx also converges.
5 1
Now consider the improper integral 3
(5 x) ( x 1)
dx .
1 1
Let f ( x) and g ( x) .
(5 x)( x 1) (5 x)1/ 2
Then
f ( x) 5 x 1 1
lim lim lim ,
x 5 g ( x) x 5 (5 x)( x 1) x 5 x 1 2
5 5
a finite no zero number. Hence
3
f ( x) dx and
3
g ( x) dx converge or diverge together. Since
1 5 dx
5 5
1
2
1, 3
g ( x) dx
3 (5 x)1/ 2
dx converges, and hence
3
(5 x)( x 1)
converges. Combining
b b b
RESULT: If a
f ( x) dx converges i.e., if
a
f ( x) dx converges absolutely, then a
f ( x) dx
converges.
b
RESULT: If a
f ( x) dx is an absolutely convergent improper integral, and if
b
| g ( x) | | f ( x) | ( a x b), then g ( x) dx converges absolutely.
a
xp
dx ( p 0)
0
sin (1/ x) 1
converges if p 1. Therefore
0 xp
converges absolutely if p 1 .
1 sin x
Example The improper integral dx converges absolutely,
0
x
|sin x | 1 1 1
since (0 x 1) and converges (absolutely).
0
x x x
1
1 1
1 x dx diverges, we have 0 1 u
As the first kind improper integral lim du
does not exist. Hence lim F () does not exist, and the theorem follows.
0
1 x dx.
1
Example Investigate the convergence of 1
0
1 dx ln 1 x 1
0 1 x 0
ln[1 (1 )] 0
ln .
1
Hence lim 1 dx lim ln( ) 0
0 0 1 x 0
.
The above limit is infinite, so the integral
1
1
1 dx lim 1 dx
0 1 x 0 0 1 x
diverges.
dx
1
Example Examine the convergence of .
0 1 x
1
The integrand is unbounded at x 1 . Hence
1 x
dx 1 1 1
1
lim dx lim 2 1 x
0 1 x 0 0 1 x 0 0
lim 2 2 2 .
0
dx
1
Hence is convergent and has value 2.
0 1 x
dx
2
Example Examine the convergence of
0 2x x
2
1
The integrand is unbounded at both the end points x 0 . and x 2 . Hence
2x x 2
dx 1 1
2 1 2
dx dx
0 2x x 0 2x x 1 2x x
2 2 2
1 2 2 1
1
lim dx lim dx
1 2 x x 2x x2
1 0 2 2 0 1
1 2
1 x 1 x 2
lim ln lim ln , where ln is the natural logarithm
1 0 2 2 x 1 2 0 2 2 x 1
function
1 1 2 2
lim ln1 ln 1 lim ln ln1
1 0 2
2 1 2 0 2 2
1 1 2
lim ln 1 lim ln .
2 0 2 1 2 0 2
dx
2
Since the limits on the right do not exist is not convergent.
0 2 x x2
RESULT: The improper integral
1 1
0
1 x2
dx
is convergent.
1
Proof. The integral is improper since is not bounded on [0,1] . We first show that the
1 x2
improper integral
1 1
0
1 x
dx
/2 dx 1 1
3. 0 (cos x)1/ n
,n 1. 4. 0 4
3
dx.
x
n
x 1 dx
1
5. 1 x dx
0
6.
0
( x 1) 1 x 2
.
2 x x2 1
3
7.
0 1
dx. 8. 2 x2 4
dx .
(16 x ) 4 3
a 1 1 cos x
1
9.
a 1 1
dx. 10. 0 x2
dx .
( x a) 3
sin x 1 log(1/ x)
1
11. 0 3
2
dx. 12.
0
x
dx .
x
1
t 2 3 x2
13. 0 1 t
dt 14. 0 (3 x)2
dx .
1 dx
15. Evaluate
1 x2
1 dx
16. Evaluate
1 x2
0
/2
18. Show that 0
log sin x dx is convergent and hence evaluate it.
x s 1
19. Prove that
0 1 x
is convergent if and only if 0 s 1.
/2 2
20. Show that 0
sin x log (sin x) dx converges and that its value is log log 2 1 .
e
1 log x x
2
21. Show that
0
x
dx is convergent, but
1 log x
dx is divergent.
x sin t
22. Let F ( x) 3
dt (0 x ).
0
2
t
Prove that the maximum value of F ( x) is attained when x .
1 1
23. Prove that
x x
sin dx converges conditionally.
0
1 sin x
24. Show that dx converges absolutely.
0
x
Real Analysis Page 117
School of Distance Education
dx
25. Test for convergence
x x2
0 3 4
sin x
26. Test for convergence dx
0
x
27. (Objective Type Questions) Classify the following according to the type of improper
integral
dx dx
(a) ; (b)
0 1 tan x
x x
0 3 4 2
100 x dx x 2 dx
(c)
5 ( x 3)8
(d) 5 x 4 x 2 1
13
IMPROPER INTEGRALS OF
SECOND AND THIRD KINDS – PART II
The Integrand Becomes Infinite at an Interior Point
( x 1)
1 c
dx lim dx lim 3( x 1)1/ 3
c
0 ( x 1)
2/3 2/3
c 1 0 c 1 0
lim[3(
c 1)1/ 3 3(0 1)1/ 3 ] 3.
c 1
On [1, 3] we have
( x 1)
3 3
dx lim dx
1 ( x 1) 2 / 3 c 1 c
2/3
lim[3(3
1)1/ 3 3(c 1)1/ 3 ] 3 3 2.
c 1
Both limits are finite, so the integral of f from 0 to 3 converges and its value is 3 3 3 2. i.e.,
3 dx
0 ( x 1) 2/3
3 3 3 2.
x x
2
x
for 0 x 1 and since
x
dx is convergent] and J 2 is a convergent improper
0
1 1 1
integral of the first kind [ since 2 2 for1 x < and since 2 is convergent].
x x x 1 x
In general if an integral J can be broken up in this way into two or more improper integrals
J1 , ... , J n of the first or second kinds, and if each J k (k 1,..., n) is convergent, we shall say that J is
a convergent improper integral. However, if one or more of the J k is divergent, we shall say
that J is a improper integral.
1
Example 2 dx is a divergent improper integral since
0 x
1 1 1
0 x 2 dx and 1 x 2 dx
are improper integrals (of the second and first kinds, respectively) one of which is divergent.
Example
1 x
1 x 2 dx
is a divergent improper integral since both
1 x 0 1 x
0 1 x 2 dx and 1 x 2 dx
1 x 1 1
are divergent improper integrals of the second kind. (Note for 1 x .) and dx is
1 x 2
x 1 x
divergent.
As we have just observed, the integral
1 x
1 x 2 dx
s 1 x
diverges since 0 1 x 2 dx … (1)
s 1
and lim 2 du
s 0 1 u2
s 1 x
does exist. The sum of (1) and (2) may be also be written dx . Hence, we have shown
s 1 x2
that
s 1 x
lim dx
s s 1 x 2
1 x
exists, even though dx diverges. We call
1 x 2
s 1 x
lim dx
s s 1 x 2
1 x
the Cauchy Principle Value of dx.
1 x 2
Definition The Cauchy principal value (C.P.V.) of
f ( x) dx is defined to be
s
lim f ( x) dx,
s s
Theorem If
f ( x ) dx converges to A, then
C.P.V.
f ( x) dx A .
Proof.
If
f ( x ) dx converges to A, then
0 b
A f ( x) dx lim f ( x) dx lim f ( x) dx … (1)
a a b 0
Also
f ( x) dx lim f ( x) dx f ( x) dx
R 0 R
lim … (2)
R R R
R 0
The last two limits in (2) are the same as the limits on the right in equation (1). Hence left side
of (1) and (2) are the same. i.e.,
R
A f ( x) dx lim f ( x) dx.
R R
Hence
C.P.V.
f ( x) dx A .
The converse is not, in general, true, because Cauchy Principal Value of
f ( x ) dx may exist
even if the integral diverges. The following is an example.
Example
x dx diverges but
s
x2 s
C.P.V. x dx lim x dx lim lim 0 0.
s s s
2 s s
It may happen that the limits on the right of (1) do not exist when 1 and 2 approach zero
independently. In such case it is possible that by choosing 1 2 in (1), i.e., writing
a
b
f ( x) dx lim
0 a
x0
f ( x) dx
b
x0
f ( x) dx … (2)
the limit des exist. If the limit on the right of (2) does exist, we call this limiting value the
b b
Cauchy Principal Value of a
f ( x) dx and is usually denoted by C.P.V. a
f ( x) dx .
Thus
C.P.V. a
b
f ( x) dx lim
0
a
x0
f ( x) dx
b
x0
f ( x) dx
5 dx
Example Determine whether 1 ( x 1)3
converges (a) in the usual sense (b) in the Cauchy
1 1 5
1 1
lim lim
1 0 2( x 1) 2 2 0 2( x 1) 2
1 1 2
1 1 1 1
lim 2 lim 2
1 0 8
21 2 2 2 32
0
Since the limits on right do not exist, the given integral doesnot converge in the usual sense.
5 dx 1 1 5 1
(b) C.P.V. 1 ( x 1) 3
lim
0
1 ( x 1) 3
dx
1 ( x 1) 3
dx
1 1 1 1 3
lim 2 2 .
0 8 2 2 32 32
Hence the given integral converges in the C.P.V. sense.
Exercises
Discuss the convergence of the following improper integrals of the second kind
sin x 1 1
1. p
dx 2. 2 dx.
0 x 0
x3
/2 dx 1 1
3. 1/ n
,n 1. 4. 4 dx.
0 (cos x) 0
x3
n
1 x 1 dx
5. dx 6. .
0 1 x 0
( x 1) 1 x 2
2 x x2 1
3
7.
0 1
dx. 8. 2 x2 4
dx .
(16 x ) 4 3
a 1 1 cos x
1
9.
a 1 1
dx. 10. 0 x2
dx .
( x a) 3
sin x 1 log(1/ x)
1
11. 0 3
2
dx. 12.
0
x
dx .
x
1
t 2 3 x2
13. 0 1 t
dt 14. 0 (3 x)2
dx .
1 dx
15. Evaluate 1 x2
1 dx
16. Evaluate
1 x2
0
/2
18. Show that 0
log sin x dx is convergent and hence evaluate it.
x s 1
19. Prove that 0 1 x
is convergent if and only if 0 s 1.
/2 2
20. Show that 0
sin x log (sin x) dx converges and that its value is log log 2 1 .
e
1 log x x
2
21. Show that
0
x
dx is convergent, but
1 log x
dx is divergent.
x sin t
22. Let 3
dt F ( x)
(0 x ).
0
t2
Prove that the maximum value of F ( x) is attained when x .
1 1
23. Prove that
x x
sin dx converges conditionally.
0
1 sin x
24. Show that dx converges absolutely.
0
x
dx
25. Test for convergence
0 3 4
x x2
sin x
26. Test for convergence dx
0
x
27. (Objective Type Questions) Classify the following according to the type of improper
integral
dx dx
(a) ; (b)
0 1 tan x
x x
0 3 4 2
100 x dx x 2 dx
(c) 5 ( x 3)8
(d) 5 x 4 x 2 1
In Exercises 1-3, which of the following integrals does the Cauchy principle value exist? Do any
of the integrals converge?
28.
sin t dt.
29.
| sin t | dt.
1
30.
1 t2
dt
s dx
31. Prove that 0 4 x
diverges in the usual sense but converges in the Cauchy principal value
sense.
1 1 1
1
32. Show that C.P.V.
1 x
dx 0 and C.P.V
1 x
dx does not exist.
33. If f is continuous on (, ) and if
f ( x) dx converges to A, prove that
C.P.V.
f ( x ) dx A. .
34. If f is continuous on [0,1], prove that
1 f ( x)
1 x2
dx 0
lim
0 a
c
f
b
c
f .
1 1
(a) Show that C.P.V.
x
dx 0. 1
1 1
(b) Show that C.P.V. dx does not exist.
1 | x |
14
BETA AND GAMMA FUNCTIONS
The Beta Function
x 1 x
n 1
If m, n are positive, then the definite integral m 1
dx is called the Beta function, (or
0
The beta integral is some times called Eulerian Integral of the first kind.
Then
1
0
x m 1 (1 x) n 1 I1 I 2
Convergence of I1
1
We take f ( x) x m 1 (1 x) n 1 and g ( x) 1 m
.
x
f ( x)
Then, (1 x) n 1 1 as x 0 .
g ( x)
f ( x)
i.e., lim 1, a non zero finite number.
x 0 g ( x)
1 1
Hence, by Limit Comparison Test (Quotie nt test) 2 f ( x) dx and 2
g ( x) dx converge or
0 0
diverge together.
1 1
1
But 2 g ( x) dx 2 dx converges if and only if 1 m 1 i.e., m0. Hence
0 0 x1 m
1 1
0
2
f ( x) dx 2 x m 1 (1 x) n 1 dx I1 converges if and only if m 0.
0
Convergence of I 2
1
We take f ( x) x m 1 (1 x) n 1 and g ( x) .
(1 x)1 n
f ( x)
Then, lim lim x m 1 1 , a non zero finite number.
x 1 g ( x) x 1
1 1
Hence, by Limit Comparison Test, both the integrals 1
2
f ( x) dx and 1
2
g ( x) dx converges or
2
Example Express 0
(8 x3 ) 1/ 3 dx in terms of a Beta function.
2
Let I (8 x3 ) 1/ 3 dx
0
2 2/3
Put x 3 8 z . Then x 2 z1/ 3 and dx z dz.
3
Also, when x 0 , z 0 ; when x 2, z 1
Hence (1) becomes
1 2 2 1
I (8 8 z )1/ 3 z 2 / 3 dz 81/ 3 (1 z ) 1/ 3 z 2 / 3 dz
0 3 3 0
1 1
(1 x) 1/ 3 x 2 / 3 dx , using the fact that
3 0
b b
a
f ( z )dz f ( x) dx
a
1 1 2 / 3
3 0
x (1 x) 1/ 3 dx
1 2 1 1 1 2
1, 1 , .
3 3 3 3 3 3
b
Example Evaluate ( x a) m 1 (b x) n 1 dx, m 0, n 0 .
a
[(b a ) z ]m 1 [b a (b a ) z ]n 1 (b a ) dz
1
(b a ) m n 1 z m 1 (1 z ) n 1 dz
0
1
(b a) m n 1 0
x m 1 (1 x) n 1 dx (b a) m n 1 (m, n)
Proof . We have
1
(m, n) x m 1 (1 x) n 1 dx
0
1
(1 x) m 1 (1 (1 x)) n 1 dx , obtained by taking f ( x) x m 1 (1 x) n 1 and a 1 in the
0
a a
result f x dx f a x dx
0 0
1
x n 1 (1 x) m 1 dx
0
n, m
1 1
m m
1 x
n 1 x
n 11 x
m
0 0
n2
1 x dx ,
m
by integrating by parts
1
n 1 m
0
m
x 1 x n 2 dx
0
n 1
m 1, n 1
m
By repeated application, we have
n 1 n 2 n 3
m, n m n 1,1
1
. . . . ...
m m 1 m 2 mn2
n 1 n 2 n 3 1 1
. . . . ... . ,
m m 1 m 2 m n 2 m n 1
1
since by Case 1, m,1
m
(n 1)!
... (1)
m(m 1)(m 2) (m n 1)
x a b x dx b a m, n
m 1 n 1 m n 1
… (3)
a
b a m, n
m n 1
b
7. x
m 1
(b x) n 1 dx b m n 1 m, n
0
Proof. Put x sin 2 , so that 1 x cos 2 and dx 2 sin cos d . When x = 0, = 0 and
when x = 1, = /2.
2
m, n sin 2m 1 cos 2n 1 2 sin cos d
0
2
2 sin 2m 1 cos 2n 1 d
0
y n 1
9. m, n
1 y
0
mn
dy .
1
1
expression m, n x m 1 1 x
n 1
Proof. Consider the dx . Put x , so that
0
1 y
1 y 1
1 x 1 and dx . When x = 1, y = 0 and when x 0, . Hence
1 y 1 y 1 y
2
0 n 1
y
m, n
1 1 dy
m 1 1 y 1 y 2
1 y
y n 1
1 y
0
mn
dy
m 1
x
10. m, n dx , where m 0, n 0
1 x
mn
0
m 1
x m 1 z
1
1
1 x
0
mn
dx
0
1 z
(1 z ) m n
(1 z ) 2
dz , since
z 1
1 x 1 .
1 z 1 z
z m 1 (1 z ) m n
1
dz
0
(1 z ) m 1 (1 z ) 2
1
z m 1 (1 z ) n 1 dz m, n .
0
p 1 q 1
In (4), put m , n or p 2m 1, q 2n 1, and we obtain
2 2
p 1 q 1
2
0 sin x cos x dx.
p q
, 2
2 2
The above formula can be used to evaluate integrals. For example,
/2
1 7 1 5 1 1 1 3! 2! 1
sin x cos5 x dx 4, 3
7
, .
0
2 2 2 2 2 6! 120
1 1
z ( m 1) / 2 (1 z ) n dz
0 2
1 1
z ( m 1) / 2 (1 z ) n dz
2 0
1 m 1 m 1
1, n 1 B , n 1
1
2 2 2 2
1 x2
Example Express 0
1 x5
dx in terms of beta function.
1
1 x2 1
We have
1 x5 0 0
dx x 2 (1 x 5 ) 2 dx.
(b a ) m 1 z m 1 b a 1 z b a dz
1 n 1 n 1
0
1
(b a) m n 1 z m 1 (1 z ) n 1 dz
0
(b a ) m n 1 B(m, n).
by integrating by parts
q 1 q
0 x p (1 x) q 1 dx ( p 1, q ). ... (1)
p 0 p
( p, q 1) ( p 1, q )
… (2)
q p
Also from (2), we obtain
( p, q 1) 1 1 1 1 1
( p 1, q) x p (1 x) q 1 dx x p 1 x(1 x) q 1 dx
q p p 0 p 0
x 1 (1 x) (1 x) q 1 dx
1 1 p 1
p 0
1 1 p 1 1 1
p 0
x (1 x) q 1 dx x p 1 (1 x) q dx
p 0
1 1
( p, q) ( p, q 1
p p
( p, q 1) ( p, q 1) ( p, q )
i.e., .
q p p
( p q ) ( p, q 1) ( p, q )
Hence .
pq p
( p, q 1) ( p, q )
Therefore, . … (3)
q pq
From (2) and (3), we get
( p, q 1) ( p 1, q ) ( p, q )
… (4)
q p pq
(ii) From (4), we get
q
( p, q 1) ( p, q,), … (5)
pq
p
and ( p 1, q ) ( p, q,). …(6)
pq
Adding (5) and (6), we have
q q
( p, q 1) ( p 1, q ) ( p, q,) ( p , q )
pq pq
pq
( p , q ) ( p , q )
pq
The Gamma Function
If n 0, the definite integral e x x n 1dx is called the Gamma Function and is denoted by (n).
0
Thus
(n) e x x n 1dx, x 0, n 0.
0
Case1: n 1
When n 1 , I1 is a proper integral and I 2 is an improper integral of the first kind.
Here f ( x) e x x n 1
1
Take g ( x) .
x2
f ( x)
Then lim lim x n 1 e x 0 .
x g ( x) x
1
Since 1
g ( x) dx
1 x2
dx converges, by Limit Comparison Test, we have
1
f ( x ) dx e x x n 1 dx I 2 is convergent.
1
Case 2: n 1
When n 1 , I1 is an improper integral of the second kind and I 2 is an improper integral of the
first kind.
Here f ( x) e x x n 1
1
Take g ( x) 1 n .
x
f ( x)
Then lim lim e x 1 , a non zero finite number.
x 0 g ( x) x 0
1 1 1
Since
0
g ( x) dx
0 x1 n
dx converges if 1 n 1, i.e., if n 0, by Limit Comparison Test, it
1 1
follows that 0
f ( x) dx x n 1e x dx I1 converges if n 0 . As in Case 1 we can show that
0
I 2 e x dx converges for all n 1
x n 1
1
1 0 t 0
lim x n 1 lim ( n 1) x dx ,
t
1
integrating by parts
t
x n 1 t
e x
lim x ( n 1) lim x n 2 dx
t e t 1
0 0
t n 1 n2 e
x
lim
t et
( n 1) 0 x
1
dx
t n 1
0 ( n 1) ( n 1), since, by applying L’Hospital Rule repeatedly, lim 0
t e t
(n 1)(n 1).
Theorem When n is a positive integer ( n) ( n 1)!
We have ( n) ( n 1)( n 1)
When n is a positive integer, by repeated application of the above formula
(n) (n 1)(n 1)
( n 1)(n 2) ( n 2)
( n 1)( n 2)( n 3) ( n 3)
( n 1)( n 2)( n 3)1 (1).
( n 1)( n 2)( n 3)1 (1).
t
But (1) e x x 0 dx e x dx lim e x dx
0 0 x 0
lim e x lim 1 e t 1 0 1.
t
t 0 t
Hence
( n) ( n 1)( n 2)( n 3)1 1 ( n 1)! ,
when n is a positive integer.
( n)
Example Show that e
a y
y n 1dy .
0
an
Putting x ay in n , we have
n
e
0
ay
ay n 1
a dy a n
e
0
ay n 1
y dy
n
Hence
e
0
a y n 1
y dy
an
m n
Theorem m, n .
m n
n
Proof. Using the relation
a n
0
e a x x n 1dx , we obtain (by taking z in place of a )
m z m
e
0
z x m 1
x dx
z
0
m z x m 1
e x dx
That is,
m n
e
0 0
z 1 x m n 1
z dz x m 1dx ,
m n m, n
m n
Hence m, n
m n
The above can be proved as follows also:
(m, n) x m 1 (1 x) n 1 dx 2 2 sin 2 m 1 cos 2 n 1 d … (1)
0 0
by putting x sin 2 .
(m) e x x m 1 dx 2 e t t 2 m 1 dt , by putting x t 2
2
0 0
2 e x x 2 m 1dx , by changing the variable t to x … (2)
2
Similarly, we have
(n) 2 e y y 2 n 1 dy. … (3)
2
Hence
( m) ( n ) 4 e( x y2 )
x 2 m 1 y 2 n 1dx dy. … (4)
2
0 0
In (4) put x r cos , y r sin .Then dxdy r dr d . Since x and y vary from 0 to , x and y may be
taken as the coordinates of any point in the first quadrant. Also by putting x r cos , y r sin ,
we are transforming Cartesian coordinates into polar coordinates. Again in polar coordinates,
the first quadrant can be covered by varying r from 0 to and from 0 to . So
2
( m ) ( n ) 4 e r (r cos ) 2 m 1 (r sin ) 2 n 1 r dr d
2
2
0 0
4 e r r 2 m 2 n sin 2 n 1 cos 2 m 1 dr d
2
2
0 0
2 e r r 2 m 2 n 1dr 2 2 sin 2 n 1 cos 2 m 1 dr d
0
2
0
... (5)
Now
2 e r r 2 m 2 n 1 dr e 1t m n 1 dt , by putting r 2 t
2
0 0
( m n)
Also, we have
2 2 sin 2 n 1 cos 2 m 1 d ( n, m) ( m, n).
0
1 1 1
, 2 sin 0 cos 0 d 2 d .
2 2 2
0 0
1
Hence .
2
Proof. As
x n 1
m, n dx , for m 0, n 0
1 x
mn
0
m n
and m, n for m 0, n 0 ,
m n
m n
x n 1
we have dx .
m n 0 1 x
mn
Putting m n 1, m 1 n, we obtain
1 n n x n 1
1
0
1 x
dx 0 n 1
x n 1
Also
0
1 x
dx
sin x
.
n 1 n
n
Hence , where 0 < n <1.
sin n
1
n
1
Example Evaluate x m log dx.
x
0
t
Put x e , then dx e t dt. Hence
1
e t e dt e
n
1 t m n t m 1 t n
x m log dx t dt
x
0 0 0
m 1
1
n 1.
dy 1
x m log dx
x n m 1 m 1n 1
0 0
e x dx
2
Example Evaluate
0
dt dt
Put x 2 t , then 2 x dx dt or dx . Hence
2x 2 t
e e
x2 t dt
dx
2 t
0 0
1
1 2 1 t
1 1
t e dt .
2 2 2 2
0
Example
2
1 7 1 5 1
sin cos d 2
7 5
2
,
2
0
4 3 1 3! 2! 1
4,3
1
2 2 7 2 6! 120
Example Express the following in terms of Gamma functions:
1
0
x p (1 x q ) n dx , where p 0, q 0; n 0.
1 1
1 q 1
Put x q z or x z q . Then dx z dz. Also when x 0, z 0 and when x 1, z 1 . Hence
q
p 1
1 1 q 1
x p (1 x q ) n dx z q (1 z ) n
1
0 0 q
z dz
p 1
1 1
q 0
z q
(1 z )( n 1) 1 dz
p 1
(n 1)
1 p 1 1 q
, n 1
q q q p 1 n 1
q
Example Express the following in terms of Gamma functions:
a
o
x p 1 (a x) q 1 dx where p 0, q 0
Here put x az so that dx a dz
When x 0, z 0 and when x a, z 1
dx az
a 1 p 1
x p 1 (a x) a 1 (a az ) q 1 a dz
0 0
1
a p 1 z p 1a q 1 (1 z ) q 1 a dz
0
1
a p q 1 z p 1 (1 z ) q 1 dz
0
( p ).( q )
a p q 1 ( p, q ) a p q 1
( p q )
n 1
(n) log
1 1
Example Show that dx.
0 x
1
Put log
y. Then x e y and dx e y dy.
x
Also when x 0 y and when x 1, y 0.
n 1
log dx y n 1 e y dy e y y n 1 dy
1 1 0
0 x 0
e y y n 1 dy (n).
0
(2m) 22 m 1 ( m) m .
1
2
( m) ( n)
We know that ( m, n ) . … (1)
( m n)
m
2
(m, m) .
2m
1
Also (m, m) x m 1 (1 x) m 1 dx.
0
2 m 1
2 sin 2 m 1
cos 2 m 1
d 2 2sin cos d
2 2
0 0 2
2 m 1
sin 2 1
2. 2 d 2 2 sin 2 m 1 2 d
0
2 0 22 m 1
1 1 1
2 2m2 2
0
sin 2 m 1 2 d
2 2m2
0
sin 2 m 1 .
2
d
by putting 2
1
2 2 m 1 2
0
sin 2 m 1 d
1
2 m 1
2 2 sin 2 m 1 d , using the result
2 0
2a a
0
f ( x) dx 2 f ( x) dx if f (2a x) f ( x) .
0
1
22 m 2 2
0
sin 2 m 1 cos 0 d
2m 1 1 0 1
2
.
1 2
22 m 2 2m 1 1 0 1
2
2 2
(m)
1
1
2 m 1 . 2
m
2 1
2
( m)
1
2 m 1
. … (2)
m
2 1
2
From (1) and (2), we conclude that
( m)
2
1 ( m)
(2m) 2 m 1
m
2 1
2
1
Hence (2m) 22 m 1 (m) (m ).
2
Example Prove that e x dx .
2
0 2
Let I e x dx
2
dz 1
Put x 2 z. Then 2x dx dz and hence dx dz.
2x 2 z
Also, when x 0, z 0 ; when x , z . Thus
dz 1 e z
I e z dz
0
2 z 2 0 z
1 21 z
2 0
z e dz
12 ( 12 1) 12 ( 12 ) 1
2
Example Prove that e x dx
2
dx 2 e , using the result “if f ( x) is an even function of x, then
x2 x2
e
0
a a
a
f ( x) dx 2 f ( x) dx ”
0
2 .
2
1
Example Show that xe x dx .
3
0 3
dz
Put x 3 z so that dx . Also, when x 0, z 0 ; when x , z . Hence
3x 2
dz
xe x dx x e z 2
3
0 0 3x
1 1
x 3/ 2 e z dz ( z1/ 3 ) 3/ 2 e z dz
3 0 3 0
1 1 1 1
z 2 e z ( 12 ) , since ( 12 ) .
3 0 3 3
In an integral of the type e f ( x ) dx , we may put f ( x) z . The above example illustrated this.
(m 2, n 2) m(m 1)
Example Show that
(m, n) (n 1)(n 2)
(m 2) ( n 2)
(m 2, n 2) (m 2 n 2)
(m, n) ( m) ( n)
( m n)
(m 2) ( n 2) ( m 1) m ( m) ( n 2)
,
( m) ( n) ( m) ( n 1)( n 2) ( n 2)
since ( n) ( n 1) ( n 1)]
m(m 1)
.
( n 1) ( n 2)
Example Evaluate
1
(i) 0
e 4 x x3/ 2 dx (ii) 0
x 4 (1 x)3 dx
x2 1
2
(iii) 0
2 x
dx (iv)
0
x5 (1 x3 )3 dx
1
(i) Putting 4x y , we have dx dy . Also, when x 0, y 0 ; when x , y . Hence
4
1
0 e x dx 32 0 e y dy
4 x 3/ 2 y 5 / 2 1
1 5 1 3 1 1 3
.
32 2 32 2 2 2 128
1 (5) (4) 4!3! 1
(ii)
0
x 4 (1 x)3 dx (5, 4)
(9)
8! 280
.
(3) ( 12 ) 4 2 2!( 12 ) 64 2
4 2 .
7 5 3 1 1 15
2 2 2 2 2
(iv) Put x 3 y . Proceeding as above, we get
1 1 1! 3! 1
(2, 4)
3 3 5! 60
Example Evaluate
x (1 x ) x (1 x )
8 6 4 5
(i) dx (ii) 0 (1 x)15 dx
0 (1 x ) 24
x8 (1 x 6 ) x8 x14
(i) I dx 0 (1 x)24 0 (1 x)24 dx
dx
0 (1 x) 24
x91 x151
9 15
dx dx
0 (1 x) 0 (1 x )15 9
8. , 2
3
2
9. 0
2
sin10 d
10. 0
2
sin 7 d cos5 d
11. 0
2
sin 4 sin 6 d
12.
0
2
tan d
dx
13. 2
0
cos x
3
sin 8 x
14.
0
2
cos x
dx
1
15.
0
x5 (1 x3 )10 dx.
1
(1 x
1
16. 2
) 2 dx
0
1
17.
0
x 7 (1 x)8 dx
1
18. 0
x 2 (1 x)3 dx
2 3
19. 0
(4 x 2 ) 2 dx
1 (1 x)
20. 0 x
dx
1
21. Show that 8 x3 3 13 , .
a 1 1
3
0 3
22. Prove that a x
a m 1
x n 1dx a m n 1 (m, n).
0
x m 1 x n 1
1
23. Prove that 0 (1 x)m n dx (m, n).
x m 1 x n 1
24. Prove that 0 (1 x)m n
2 (m, n).
x m 1 (1 x) n 1 1
25. Prove that 0 (a bx) mn
dx
( a b) m a n
(m, n).
p qn m 1 m 1
x m p q x q dx
p
0
n
26. Prove that n 1,
q q
27. Show that if p, q are positive, then
p 1, q p, q 1 p, q .
n
28. Evaluate x log dx.
1 1
m
0 x
a
29. Prove that (a x)m1 x n 1 dx a m n1 (m, n) .
0
1
30. Using the property (m, n) (n, m) , evaluate x (1 x)
3 4/3
dx
0
x m 1 (1 x) n 1
1 (m, n)
31. Show that
dx n
0 (a x) m n
a (1 a)n
32. Show that
m 1
x x n 1
(i) dx 2 (m, n), m 0, n 0
0 (1 x ) m n
x m 1 x n 1
(ii) 0 (1 x) m n
dx 0, m 0, n 0 .
1 1
33. Express x (1 x n ) p dx in terms of Beta function and hence evaluate x (1 x )
m 5 3 3
dx
0 0
x m 1 1
34. Prove that 0 (a bx)mn a nbm (m, n)
where m 0, n 0, a 0, b 0
1 1
35. (i) Express x (1 x n ) p dx in terms of the Beta function and hence evaluate x5 (1 x3 )10 dx .
m
0 0
(n)
m
x 1 x
(ii) Show that
1
m 1 2 n 1
dx m, n
1 1 2
2 m n
0 2 2 1
2
( p, q 1) ( p 1, q ) ( p, q )
(iii) Prove that , p 0, q 0
q p pq
(m 1, n) m
(iv) Prove that , m 0, n 0 .
(m, n) mn
(v) Prove that (m, n) 21 2 m (m, 12 ) .
(vi) Show that 1
2
sin 3 cos5 d (2,3) and hence evaluate it.
0 2
In Exercises 36-45, evaluate:
(7)
36.
2(4)(3)
37. 12 23 52
(3)( 32 )
38.
( 92 )
39. 0
x3e x dx 6
xe x dx
2
40.
0
x 4 e x dx
4
41.
0
e x dx
4
42.
0
e x dx
3
43.
0
1
44. x log( ) dx
1 1 3
0 x
x3e x dx
3
45.
0
1
46. Show that xe x dx
3
0 3
5
47. Show that 4 x 4 e x dx
4
0
4
45
48. Show that 0
x 6 e 2 x dx
8
1 1
49. Show that x 2 (1 x)3 dx
0 60
3 dx
50. Show that
0
3x x 2
1
51. Show that
1 dx
3
0
1 x 3 5
3
6
52. Prove that ( n 1) n( n) , where n 0 .
p 1
1 1 ( p )
53. Show that y q 1
log dy , where p 0, q 0 .
0
y qp
2
1
54. Show that 2
0
sin d
0 sin
d .
1
55. Show that the area bounded by the normal curve y e ( x / 2 ) and x-axis is unity.
2 2
2
1
x3 2
4
1 23 2
56. Prove that
0 (1 x 3 )
dx
3 2 13
3
1
14
1 1
2
57. Show that 1 x 4 dx
0
6 2
1 2
x dx 1 dx
58. Show that 4 12
4 12
.
0
(1 x ) 0
(1 x ) 4
x a
(a 1)
59. Show that if a 1, x dx
0 a (log a)a 1
t 2 dt
60. Prove that 0 1 t 4 2 2
1 dx 1n
61. Prove that .
0
1 xn n( 1n 12 )
a 1
62. Show that (a) 2 et t 2 a 1 dt log du.
2 1 1
0 0 u
n 1
63. Prove that n 1, x n e a x dx n 1
2 2 1
.
0 2a 2
1
64. Show that n
1
0
n 1
log y n 1 dy.
n 1
x m 1
1 x
x
65. Show that mn
dx 0. m 0, n 0.
0
1 n 1
66. (i) Prove that x n e a . Hence show that e a dx
2
x2
dx
2
x2
n 1
0 2a 2 a
(ii) Prove that e a x dx
, a0
2 2
0 2a
1
67. (a) Show that
1 dx
n
0
1 x n 1 1
n
n 2
1
.
1 dx 4
(b) Prove that
0
1 x 4 4 3
4
m 1 n 1
1x (1 x) ( m) ( n )
68. Show that dx n
0 (a x) m n a (1 a) m (m n)
n 1
/2 2
69. (i) Show that sin n
d , n 1
0 2 n2
2
/2 /2 d
(ii) Show that sin d
0 0
sin
/2 /2
(iii) Show that sin p x dx sin p 1 x dx .
0 0 2( p 1)
xa (a 1)
70. Show that dx , if a 1 .
0 a x
(log a)n 1
x 1 1
71. Show that dx
1 ( x 2) 6
20
1 3 5 (2n 1)
72. Show that n
1
.
2 2n
n
1
1 3 5 (2n 1)
2
73. Show that
2 4 6 2n ( n 1)
m 1 n 1
1 m 1 n 1 2 2
74. Show that 0
m n
2 2 2
2
sin cos d ,
mn2
2
2
1 (1 x ) 7 1
4 3/ 4
1
75. Show that dx , .
0 (1 x 4 ) 2
4(2 ) 4 4
1
2