M.Thamban Nair
Department of Mathematics
Indian Institute of Technology Madras
Contents
Preface
vi
1.2
1.1.1
1.1.2
Monotonic sequences . . . . . . . . . . . . . . . . . . . . . . .
1.1.3
Subsequences . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.4
Further examples . . . . . . . . . . . . . . . . . . . . . . . . .
11
1.1.5
Cauchy sequence . . . . . . . . . . . . . . . . . . . . . . . . .
16
1.1.6
Additional exercises . . . . . . . . . . . . . . . . . . . . . . .
19
21
1.2.1
21
1.2.2
23
1.2.3
Alternating series . . . . . . . . . . . . . . . . . . . . . . . . .
26
1.2.4
Absolute convergence . . . . . . . . . . . . . . . . . . . . . .
27
1.2.5
Additional exercises . . . . . . . . . . . . . . . . . . . . . . .
30
M.T. Nair
M.T. Nair 32
Limit of a Function . . . . . . . . . . . . . . . . . . . . . . . . . . . .
32
2.1.1
32
2.1.2
33
35
2.1.3
Some properties . . . . . . . . . . . . . . . . . . . . . . . . .
36
2.1.4
Some Examples . . . . . . . . . . . . . . . . . . . . . . . . . .
37
ii
Contents
2.2
2.1.5
. . . . . . . . . . . . . . . . . . . .
39
2.1.6
Limit at and at . . . . . . . . . . . . . . . . . . . . .
40
2.1.7
Additional Exercises . . . . . . . . . . . . . . . . . . . . . . .
42
Continuity of a Function . . . . . . . . . . . . . . . . . . . . . . . . .
43
2.2.1
43
2.2.2
Some examples . . . . . . . . . . . . . . . . . . . . . . . . . .
44
2.2.3
46
2.2.4
51
52
52
Additional Exercises . . . . . . . . . . . . . . . . . . . . . . .
53
Differentiability of functions . . . . . . . . . . . . . . . . . . . . . . .
54
2.3.1
55
2.3.2
Some Examples . . . . . . . . . . . . . . . . . . . . . . . . . .
57
2.3.3
59
2.3.4
61
Rolles theorem . . . . . . . . . . . . . . . . . . . . . . . . . .
61
62
63
LHospitals rules . . . . . . . . . . . . . . . . . . . . . . . . .
64
Taylors formula . . . . . . . . . . . . . . . . . . . . . . . . .
66
2.3.5
70
2.3.6
71
2.3.7
Additional exercises . . . . . . . . . . . . . . . . . . . . . . .
73
2.2.5
2.3
iii
3 Definite Integral
M.T. Nair 74
3.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
74
3.2
75
3.3
76
3.3.1
78
iv
Contents
3.4
83
3.5
Some Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
86
3.6
Some Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
88
3.6.1
88
3.6.2
88
3.6.3
90
3.7
Appendix
M.T. Nair . . . . . . . . . . . . .
4 Improper Integrals
4.1
92
M.T. Nair 93
Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
93
4.1.1
Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
95
Integrability by Comparison . . . . . . . . . . . . . . . . . . . . . . .
96
4.2.1
Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
97
4.2.2
98
4.3
99
4.4
4.2
5.2
M.T. Nair101
5.1.2
5.1.3
. . . . . . . . . . . . . . . . . . . . . . . . . 103
5.2.1
5.2.2
5.2.3
Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.3
5.4
5.5
5.6
Contents
5.6.2
5.7
5.8
5.7.2
5.7.3
112
M.T. Nair116
6.1.2
6.1.3
6.1.4
6.2
6.3
7 Power Series
M.T. Nair128
7.1
7.2
Uniform Convergence
7.3
7.4
8 Fourier Series
8.1
8.2
M.T. Nair133
8.1.2
8.3
. . . . . . . . . . . . . . . . . . . . . . . . . . 130
Preface
This is based on a course that the author gave to B.Tech. students of IIT Madras
many times since 1996 for the use of students and teachers of IIT Madras.
Comments and suggestions from the readers are welcome.
June 2006
M. T. Nair
mtnair@iitm.ac.in
vi
1
Sequence and Series of Real Numbers
1.1
Suppose for each positive integer n, we are given a real number an . Then, the list
of numbers,
a1 , a2 , . . . , an , . . . ,
is called a sequence, and this ordered list is usually written as
(a1 , a2 , . . . , . . .)
or
(an ).
M.T. Nair
1.1.1
n N,
Remark 1.2 (a) Note that, different can result in different N , i.e., the number
N may vary as varies. We shall illustrate this in Example 1.2.
(b) In Definition 1.2, the relation an a < can be replaced by an a < c0
for any c0 > 0 which is independent of n. In other words, a sequence (an ) in R
converges to a R if and only if for any c0 > 0 and for every > 0, there exists
N N such that an a < c0 for all n N .
If (an ) converges to a, then we write
lim an = a
or an a
or simply as
an a.
as
Note that
an a <
n N
if and only if
a < an < a +
n N.
Thus, limn an = a if and only if for every > 0, there exists N N such that
an (a , a + )
n N.
Thus, an a if and only if for every > 0, an belongs to the open interval
(a, a+) for all n after some finite stage, and this finite stage may vary according
as varies.
Remark 1.3 Suppose (an ) is a sequence and a R. The to show that (an ) does
not converge to a, we should be able to find an > 0 such that infinitely may an s
are outside the interval (a , a + ).
Exercise 1.1 Show that a sequence (an ) converges to a if and only if for every open
interval I containing x, there exits N N such that an I for all n N .
J
EXAMPLE 1.2 The sequences (1/n), ((1)n /n), (1 n1 ) are convergent with
limit 0, 0, 1 respectively:
For the sake of illustrating how to use the definition to justify the above statement, let us provide the details of the proofs:
(i) Let an = 1/n for all n N, and let > 0 be given. We have to identify an
N N such that 1/n < for all n N . Note that
1
1
< n > .
n
1
Thus, if we take N =
+ 1, then we have
an 0 =
1
<
n
n N.
M.T. Nair
1
n
n N := 1.
EXAMPLE 1.4 For a given k N, let Let an = 1/n1/k for all n N. Then an 0
as n .
To see this, first let > 0 be given. Note that
1
1
< n > k .
n1/k
Hence,
1
n1/k
<
1
n N := k + 1.
Thus, 1/n1/k 0.
an (a0 , a0 + )
n N2 .
In particular,
an (a , a + ),
an (a0 , a0 + ) n N := max{N1 , N2 }.
for all
n N1 ,
a0 an  < /2 n N2 .
n N := max{N1 , N2 }.
an + bn a + b as n ,
(b)
Exercise 1.6 If an a and there exists b R such that an b for all n N, then
show that a b.
J
Exercise 1.7 If an a and a 6= 0, then show that there exists k N such that
an 6= 0 for all n k.
J
1/n
1
, n N. Then
Exercise 1.8 Consider the sequence (an ) with an = 1 +
n
show that lim an = 1.
n
1
Exercise 1.9 Consider the sequence (an ) with an = k , n N. Then show that
n
for any given k N, lim an = 0.
n
M.T. Nair
Definition 1.4 (i) If (an ) is such that for every M > 0, there exists N N such
that
an > M
n N,
then we say that (an ) diverges to +.
(ii) If (an ) is such that for every M > 0, there exists N N such that an < M
for all n N , then we say that (an ) diverges to .
Definition 1.5 If (an ) is such that an an+1 < 0 for every n N, that is an changes
sign alternately, then we say that (an ) is an alternating sequence .
An alternating sequence converge or diverge. For example, (Verify that) the
sequence ((1)n ) diverges, whereas ((1)n /n) converges to 0.
Definition 1.6 A sequence (an ) is said to be bounded above if there exists a
real number M such that an M for all n N; and the sequence (an ) is said to be
bounded below if there exists a real number M 0 such that an M 0 for all n N.
A sequence which is bound above and bounded below is said to be a bounded
sequence.
Exercise 1.10 Show that a sequence (an ) is bounded if and only if there exists
M > 0 such that an  M for all n N.
J
Exercise 1.11 Prove the following.
1. If (an ) is not bounded above, then there exists a strictly increasing sequence
(kn ) of natural numbers such that akn + as n .
2. If (an ) is not bounded below, then there exists a strictly increasing sequence
(kn ) of natural numbers such that akn as n .
J
Theorem 1.3 Every convergent sequence is bounded. The converse is not true.
Proof. Suppose (an ) converges to x. Then there exists N N such that an x
1 for all n N . Hence
an  = (an a) + a an a + a 1 + a
n N,
so that
an  max{1 + a, a1 , a2 , . . . , aN 1 }
n N.
To see that the converse of the theorem is not true, consider the sequence ((1)n ).
It is a bounded sequence, but not convergent.
The above theorem can be used to show that certain sequence is not convergent,
as in the following example.
EXAMPLE 1.5 For n N, let
an = 1 +
1 1
1
+ + ... + .
2 3
n
1 1
1
+ + ... + n
2
3
2
1 1 1 1
1
1 1
+
+
= 1+ +
+
+ + +
2
3 4
5 6 7 8
1
1
... +
+ ... + n
n
2 1
2
n
1+ .
2
= 1+
Exercise 1.13 If (an ) converges to a and a 6= 0, then show that there exists k N
such that an  a/2 for all n k, and (1/an+k ) converges to 1/a.
J
1.1.2
Monotonic sequences
We can infer the convergence or divergence of a sequence in certain cases by observing the way the terms of the sequence varies.
Definition 1.7 Consider a sequence (an ).
(i) (an ) is said to be monotonically increasing if an an+1 for all n N.
(ii) (an ) is said to be monotonically decreasing if an an+1 for all n N.
If strict inequality occur in (i) and (ii), then we say that the sequence is strictly
increasing and strictly decreasing, respectively.
Also, a monotonically increasing (respectively, monotonically decreasing) sequence is also called an increasing (respectively, a decreasing) sequence.
We shall make use of some important properties of the set R of real numbers.
M.T. Nair
n N.
1.1.3
Subsequences
10
M.T. Nair
For example, given a sequence (an ), the sequences (a2n ), (a2n+1 ), (an2 ), (a2n ) are
some of its subsequences. As concrete examples, (1/2n), and (1/(2n + 1)), (1/2n )
are subsequences of (1/n).
A sequence may not converge, but it can have convergent subsequences. For
example, we know that the sequence ((1)n ) diverges, but the subsequences (an )
and (bn ) defined by an = 1, bn = 1 for all n N are convergent subsequences of
((1)n ).
However, we have the following result.
Theorem 1.6 If a sequence (an ) converges to x, then all its subsequences converge
to the same limit x.
Exercise 1.16 Prove Theorem 1.6.
What about the converse of the above theorem? Obviously, if all subsequences
of a sequence (an ) converge to the same limit x, then (an ) also has to converge to
x, as (an ) is a subsequence of itself.
Suppose every subsequence of (an ) has at least one subsequence which converges
to x. Does the sequence (an ) converges to x? The answer is affirmative, as the
following theorem shows.
Theorem 1.7 If every subsequence of (an ) has at least one subsequence which converges to x, then (an ) also converges to x.
Proof. Proof is left as an exercise.
We have seen in Theorem 1.3 that every convergent sequence is bounded, but a
bounded sequence need not be convergent.
Question: For every bounded sequence, can we have a convergent subsequence?
The answer is in affirmative:
Theorem 1.8 (BolzanoWeirstrass theorem). Every bounded sequence of real
numbers has a convergent subsequence.
Proof: For a first reading this proof can be omitted.
quence in R. For each k N, consider the set
Ek := {an : n k},
and let bk := sup Ek , k N. Clearly, b1 b2 . . ..
1
From the book: Mathematical Analysis: a straight forward approach by K.G. Binmore.
11
(anj )
j=1 . Again, since (an ) is bounded, (anj )j=1 is bounded, so that it converges.
Remark 1.4 We may observe that the proof of Theorem 1.8 can be slightly modified so that we, in fact, have the following: Every sequence in R has a monotonic
subsequence.
Exercise 1.17 Prove the statement in italics in Remark 1.4.
1.1.4
Further examples
an+1 =
2an + 3
,
4
n = 1, 2, . . . .
2an + 3
an 3
=
+ an
4
2
4
3
an .
2
2.
12
M.T. Nair
Since a1 = 2, one may try to show that (an ) is monotonically decreasing and
bounded below.
Note that
1
2
an+1 :=
an +
an
a2n 2, i.e., an 2.
2
an
Clearly a1 2. Now,
2
1
an +
2
a2n 2 2an + 2 0,
an+1 :=
2
an
i.e., if and only if (an 2)2 0. This is true for every n N. Thus, (an ) is monotonically decreasing and bounded below, so that by Theorem 1.5, (an ) converges.
EXAMPLE 1.9 Consider the sequence (an ) with an = (1+1/n)n for all n N. We
show that (an ) is monotonically increasing and bounded above. Hence, by Theorem
1.5, (an ) converges.
Note that
an
1 n
=
1+
n
1 n(n 1) 1
n(n 1) . . . 2.1 1
+ +
= 1 + n. +
2
n
n 2! n
n! n
1
1
1
1
2
= 1+1+
1
+
1
1
+
2!
n
3!
n
n
1
1
n1
+
1
1
n!
n
n
an+1 .
Also
1 n(n 1)
1 n(n 1) . . . 2.1
+ +
2
2!
n
n!
nn
1
1
1
1 + 1 + + + +
2! 3!
n!
1
1
1
1 + 1 + + 2 + + n1
2 2
2
< 3.
an = 1 + 1 +
In the four examples that follow, we shall be making use of Theorem 1.2 without
mentioning it explicitly.
13
1
1
1
1
+ + + +
1! 2! 3!
n!
n N.
Clearly, (bn ) is monotonically increasing, and we have noticed in the last example
that it is bounded above by 3. Hence, by Theorem 1.5, it converges.
Let (an ) and (bn ) be as in Examples 1.9 and 1.10 respectively, and let a and b
their limits. We show that a = b.
We have observed in last example that 2 an bn 3. Hence, taking limits, it
follows that a b. Notice that
1
1
1
1
2
an = 1 + 1 +
1
+
1
1
+
2!
n
3!
n
n
1
1
n1
... +
1
... 1
.
n!
n
n
Hence, for m, n with m n, we have
1
1
1
2
1
1
+
1
1
+
an 1 + 1 +
2!
n
3!
n
n
1
1
m1
+
1
... 1
.
m!
n
n
Taking limit as n , we get (cf. Theorem 1.2 (c))
x1+
1
1
1
1
+ + + +
= bm .
1! 2! 3!
m!
EXAMPLE 1.11 Let a > 0. We show that, if 0 < a < 1, then the sequence (an )
converges to 0, and if a > 1, then (an ) diverges to infinity.
(i) Suppose 0 < a < 1. Then we can write a = 1/(1 + r), r > 0, and we have
an = 1/(1 + r)n = 1/ (1 + nr + + rn ) < 1/(1 + nr) 0
as
n .
Hence, an 0 as n .
An alternate way: Let xn = an . Then (xn ) is monotonically decreasing and bounded
below by 0. Hence (xn ) converges, to say x. Then xn+1 = an+1 = axn ax. Hence,
x = ax. This shows that x = 0.
(ii) Suppose a > 1. Then, since 0 < 1/a < 1, the sequence (1/an ) converges to
0, so that (an ) diverges to infinity. (Why ?)
14
M.T. Nair
n(n 1) 2
rn ,
2
so that rn2 2/(n 1) for all n 2. Since 2/(n 1) 0, by Theorem 1.2(c), that
rn 0, and hence by Theorem 1.2(c), n1/n = 1 + rn 1.
Remark 1.5 The proof of the result in Example 1.12 does not require the knowledge
that n1/n 1, but uses the facts that 2/(n 1) 0 (which can be shown easily)
and Theorem 1.2. But, if one is asked to show that n1/n 1, then we can give
another proof of the same by using only the definition, as follows:
Another proof without using Theorem 1.2. Let > 0 be given. To find n0 N such
that n1/n 1 < for all n n0 . Note that
n1/n 1 < n1/n < 1 + n < (1 + )n = 1 + n +
n(n 1) 2
+ . . . + n .
2
Hence
2
.
2
So, we may take any n0 N which satisfies n0 2(1 + /2 ).
n1/n 1 < if
i.e., if
n>
Now, another proof for the same: Note that 1 an 21/n , and 21/n 1 as n .
15
0 an q nN aN
n N.
Exercise 1.20 Suppose Let (an ) is a sequence of positive terms such that
an+1
lim
= ` > 1. Then (an ) diverges to .
J
n an
EXAMPLE 1.18 Let 0 < a < 1. Then nan 0 as n .
To see this let an := nan for n N. Then we have
(n + 1)a
(n + 1)an+1
an+1
=
=
n
an
na
n
Hence, lim
n N.
an+1
= a < 1. Thus, the result follows from the last example.
an
Exercise 1.21 Obtain the the result in Example 1.18 by using the arguments in
Example 1.11.
J
(k)
(k)
as
and
(2)
(n)
b(1)
n bn bn 1
as
n .
k
,
n2
1/n
b(k)
n =k
k, n N.
16
M.T. Nair
(k)
(n)
(n)
1
2
n
n+1
1
+
+ + 2 =
n2 n2
n
2n
2
as
and
(2)
(n)
1/n 1/n
b(1)
2 n1/n = (n!)1/n 6 1
n bn bn = 1
as
n .
k! 1/n
kk
1/n
Since for any given k N, k kk!k
k as n , it follows that (n!)1/n 6 1. In
fact, there exists n0 N such that
(n!)1/n
k
2
n max{n0 , k}.
Thus, the sequence is (n!)1/n is unbounded.
n+1
1 + 2 + ... + n
=
n.
n
2
2
1 (n!)1/n n1/n .
2
1.1.5
Cauchy sequence
Theorem 1.9 If a real sequence (an ) converges, then for every > 0, there exists
N N such that
an am  < n, m N.
Proof. Suppose an a as n , and let > 0 be given. Then we know that
there exists N N such that an a < /2 for all n N . Hence, we have
an am  an a + a am  < n, m N.
This completes the proof.
17
n N.
k max{N, n0 }.
Thus, an a as n .
In fact, the converse of Theorem 1.9 also holds:
Theorem 1.11 Every Cauchy sequence of real numbers converges.
Proof of Theorem 1.11. Let (an ) be a Cauchy sequence. By Theorem 1.10(i),
(an ) is bounded. Then, by BolzanoWeierstrass theorem (Theorem 1.8), (an ) has a
convergent subsequence (ank ). Let a = lim ank . Now, by Theorem 1.10(ii), an a
k
as n .
18
M.T. Nair
()
From this, we see that the set {xn : > 0} is bounded above, and
x := sup{xn : > 0}
satisfies
xn < x < xn +
n n .
()
n n .
n N, n 2.
n N, n 2.
1
Since m1 0 as m , given > 0, there exists N N such that an am  <
for all n, m N .
Exercise 1.23 Given a, b R and 0 < < 1, let (an ) be a sequence of real numbers
defined by a1 = a, a2 = b and
an+1 = (1 + )an an1
n N, n 2.
19
x, y J,
an+1 := f (an )
n N,
is a Cauchy sequence. Show also that the limit of the sequence (an ) is independent
of the choice of a.
J
1.1.6
Additional exercises
n
n+1
(ii)
(1)n n
n+1
(iii)
2n
,
3n2 + 1
(iv)
2n2 + 3
.
3n2 + 1
( an ) converges to x.
8. Prove the following:
(a) If {xn } is increasing and unbounded, then xn + as n .
(b) If {xn } is decreasing and unbounded, prove that xn as n .
9. Let a1 = 1, an+1 =
find its limit.
10. Let a1 = 1 and an+1 = 41 (2an + 3) for all n N. Show that {an } is monotonically increasing and bounded above. Find its limit.
20
M.T. Nair
an
for all n N. Show that {an } converges. Find
1 + an
12. Prove that if (an ) is a Cauchy sequence having a subsequence which converges
to a, then (an ) itself converges to a.
13. Suppose (an ) is a sequence such that the subsequences (a2n1 ) and (a2n ) converge to the same limit, say a. Show that (an ) also converges to a.
14. Let {xn } be a monotonically increasing sequence such that {x3n } is bounded.
Is {xn } convergent? Justify your answer.
15. Show that, if a sequence (an ) has the property that an+1 an 0 as n ,
then (an ) need not converge.
16. Let a1 = 1 and an+1 = 41 (2an + 3) for all n N. Show that (an ) is monotonically increasing and bounded above. Find its limit.
17. Give an example in support of the statement: If a sequence {an } has the
property that an+1 an 0 as n , then {an } need not converge.
18. Let a > 0 and xn =
(xn ).
an an
,
an +an
19. If 0 < a < b and an = (an + bn )1/n for all n N,then show that (an ) converges
n
to b. [Hint: Note that (an + bn )1/n = b(1 + ab )1/n .]
n
20. If 0 < a < b and an+1 = (an bn )1/2 and bn+1 = an +b
for all n N with a1 = a,
2
b1 = b, then show that (an ) and (bn ) converge to the same limit. [Hint: First
observe that an an+1 bn+1 bn for all n N.]
n bn
21. Let a1 = 1/2 and b = 1, and let an+1 = (an bn )1/2 and bn+1 = a2an +b
for all
n
n N. Show that (an ) and (bn ) converge to the same limit. [Hint: First
observe that bn bn+1 an+1 an for all n N.]
1
). Prove that
22. Let a1 = 1/2, b = 1, an = (an1 bn1 )1/2 and b1n = 12 ( a1n + bn1
an1 < an < bn < bn1 for all n N. Deduce that both the sequences {an }
and {bn } converge to the same limit , 1/2 < < 1.
23. Prove that if {an } is a Cauchy sequence having a subsequence which converges
to a, then {an } itself converges to a.
1.2
21
n=1 an ,
The number an is called the nth term of the series and the sequence sn :=
P
is called the nth partial sum of the series
n=1 an .
n
X
ai
i=1
1.2.1
P
Definition 1.13 A series
n=1 an is said to converge (to s R) if the sequence
{sn } of partial sums of the series converge (to s R).
P
P
If
n=1 an = s.
n=1 an converges to s, then we write
A series which does not converge is called a divergent series.
A necessary condition
P
Theorem 1.12 If
n=1 an converges, then an 0 as n . Converse does not
hold.
Proof. Clearly, if sn is the nth partial sum of the convergent series
then
an = sn sn1 0 as n .
n=1 an ,
n
n=1 n+1
P
1
EXAMPLE 1.21 We have
that the sequence (sn ) with sn = nk=1 k!
conP seen
1
verges. Thus, the
we have seen that the sequence
P series n=1 n! converges.
P Also,
1
(n ) with n = nk=1 k1 diverges. Hence,
diverges.
n=1 n
22
M.T. Nair
P
n1 , where a, q R.
EXAMPLE 1.22 Consider the geometric series series
n=1 aq
n1
Note that sn = a + aq + . . . + aq
for n N. Clearly, if a = 0, then sn = 0 for all
n N. Hence, assume that a 6= 0. Then we have
(
na
if q = 1,
sn =
a(1q n )
if q 6= 1.
1q
Thus, if q = 1, then (sn ) is not bounded; hence not convergent. If q = 1, then we
have
a
if n odd,
sn =
0
if n even.
Thus, (sn ) diverges for q = 1 as well. Now, assume that q =
6 1. In this case, we
have
sn a = a qn .
1 q 1 q
This shows that, if q < 1, then (sn ) converges to
not bounded, hence diverges.
a
1q ,
n
X
i=k+1
ai =
n
X
bi = n
n k.
i=k+1
From this it follows that the sequence (sn ) converges if and only if (n ) converges.
P
P
From the above theorem it follows if
n=1 bn is obtained from
n=1 an by
omitting or adding a finite number of terms, then
an
converges
n=1
In particular,
X
n=1
an
bn
converges.
n=1
converges
an+k
converges
n=1
for any k N.
The proof of the following theorem is left as an exercise.
P
P
Theorem 1.15 Suppose
n=1 an converges to s and
n=1 bn converges to . Then
P
for every , R, n=1 ( an + bn ) converges to s + .
1.2.2
23
Theorem 1.16 (Comparison test) Suppose (an ) and (bn ) are sequences of nonnegative terms, and an bn for all n N. Then,
(i)
(ii)
n=1 bn
converges =
n=1 an
diverges =
converges,
diverges.
n=1 an
n=1 bn
P
Proof.
Suppose
s
and
be
the
nth
partial
sums
of
the
series
n
n
n=1 an and
P
n=1 bn respectively. By the assumption, we get 0 sn n for all n N, and
both (sn ) and (n ) are monotonically increasing.
(i) Since (n ) converges, it is bounded. Let M > 0 be such that n M for all
n N. Then we have sn M for all n N. Since (sn ) are monotonically increasing,
it follows that (sn ) converges.
(ii) Proof of this part follows from (i) (How?).
Corollary 1.17 Suppose (an ) and (bn ) are sequences of positive terms.
(a) Suppose ` := limn abnn exists. Then we have the following:
P
P
(i) If ` > 0, then
n=1 an converges.
n=1 bn converges
P
P
(ii) If ` = 0, then n=1 bn converges n=1 an converges.
P
P
an
(b) Suppose lim
= . Then
n=1 bn converges.
n=1 an converges
n bn
Proof. (a) Suppose limn
an
bn
= `.
(i) Let ` > 0. Then for any > 0 there exists n N such that ` < abnn < ` +
for all n N . Equivalently, (` )bn < an < (` + )bn for all n N . Had we taken
= `/2, we would get 2` bn < an < 3`
2 bn for all n N . Hence, the result follows by
comparison test.
(ii) Suppose ` = 0. Then for > 0, there exists n N such that < abnn <
for all n N . In particular, an < bn for all n N . Hence, we get the result by
using comparison test.
(b) By assumption, there exists N N such that
result follows by comparison test.
an
bn
P
1
EXAMPLE 1.23 We have already seen that the sequence (sn ) with sn = nk=1 k!
1
1
converges. Here
proof for the same fact: Note that n!
2n1
P is another
Pfor all
1
1
n N. Since n=1 2n1 converges, it follows from the above theorem that
n=1 n!
converges.
24
M.T. Nair
P
1
for all n N, and since the series
n=1 n diverges,
P 1
it follows from the above theorem that the series n=1 n also diverges.
EXAMPLE 1.24 Since
1
n
1
n
n N.
In particular,
an+1 < q an < q 2 an1 < . . . < q n a1 , n N.
P
P
n
Since
n=1 an also converges.
n=1 q converges, by comparison test,
(ii) Let 1 < p < `. Then there exists N N such that
an+1
> p > 1 n N.
an
From this it follows that (an ) does not converge to 0. Hence
n=1 an
diverges.
Theorem 1.19 (Cauchys root test) Suppose (an ) is a sequence of positive terms
such that limn an 1/n = ` exists. Then we have the following:
P
(i) If ` < 1, then the series
n=1 an converges.
P
(ii) If ` > 1, then the series
n=1 an diverges.
Proof. (i) Suppose ` < q < 1. Then there exists N N such that
an 1/n < q
n N.
P n
n
Hence,
P an < q for all n N . Since the n=1 q converges, by comparison test,
n=1 an also converges.
(ii) Let 1 < p < `. Then there exists N N such that
an 1/n > p > 1
n N.
n=1 an
Remark 1.7 We remark that both dAlemberts test and Cauchy test are silent
for the case ` = 1. But, for such case, we may be able to infer the convergence or
divergence by some other means.
xn
n! .
xn
n=1 n!
25
converges:
Hence
x
an+1
=
an
n+1
n N.
n
1
< 1.
2n + 1
2
= 1 = limn an
sn =
n
X
k=1
1/n .
1
n(n+1) .
X
1
=
k(k + 1)
k=1
1
1
k k+1
=1
1
.
n+1
1
n=1 n2 .
1
1
2
(n + 1)
n(n + 1)
Since
1
n=1 n(n+1)
1
test, we see that the series n=1 np converges for p 2 and diverges for p 1.
P
1
EXAMPLE 1.30 Consider the series
n=1 np for p > 1. To discuss this example,
p
consider the function f (x) := 1/x , x 1. Then we see that for each k N,
Z k
1
1
dx
1
k 1 x k = p p = p
.
p
k
x
k
k1 x
Hence,
Z n
n
n Z k
X
X
1
dx
dx
n1p 1
1
=
=
.
p
p
kp
x
x
1
p
p
1
k1
1
k=2
k=2
Thus,
sn :=
n
X
1
1
+ 1.
p
k
p1
k=1
Hence, (sn ) is monotonically increasing and bounded above. Therefore, (sn ) converges.
26
M.T. Nair
f (n)
converges (an )
converges.
n=1
Rn
Rk
P
Proof. First we observe that an = 1 f (x) dx = nk=2 k1 f (x) dx. Now, since
f (x) is a decreasing function for x [1, ), we have for each k N,
k 1 x k = f (k) f (x) f (k 1).
Hence, for k = 2, 3, . . .,
k
Z
f (k)
f (x) dx f (k 1)
k1
so that
n
X
f (k)
k=2
Thus,
n
X
Now, let sn :=
f (x) dx
k1
k=2
Z
f (k)
k=2
n
X
n Z
X
f (x) dx
1
n
X
f (k 1).
k=2
n
X
f (k 1).
k=2
k=1
with the fact that (sn ) is a monotonically increasing sequence, it follows that (an )
converges if and only if (sn ) converges.
1.2.3
Alternating series
P
n+1 u where (u ) is a sequence of
Definition 1.14 A series of the form
n
n
n=1 (1)
positive terms is called an alternating series.
Theorem 1.21 (Leibnizs theorem) Suppose (un ) is a sequence of positive terms
such that
n+1 for all n N, and un 0 as n . Then the alternating
P un un+1
series
(1)
un converges.
n=1
Proof. Let sn be the nth partial sum of the alternating series
We observe that
s2n+1 = s2n + u2n+1 n N.
n+1 u .
n
n=1 (1)
27
X
that sn s, where sn is the nth partial sum of
(1)n+1 un .
n=1
Thus,
s2n1 s u2n ,
s s2n u2n+1 .
Consequently,
s sn  un+1
1.2.4
n N.
Absolute convergence
n=1 an
n=1 an 
con
28
M.T. Nair
a

respectively.
Then
it
follows
that
n
n=1
sn + n = 2pn ,
where pn is the sum of all positive terms from {a1 , . . . , an }. Since {n } converges,
it is bounded, and since pn n for all n N, the sequence {pn } is also bounded.
Moreover, {pn } is monotonically increasing. Hence {pn } converge as well. Thus,
both {n }, {pn } converge. Now, since sn = 2pn n for all n N, the sequence
{sn } also converges.
P
P
a
is
said
to
converge
conditionally
if
Definition 1.16
A
series
n
n=1 an
n=1
P
converges, but
a
diverges.
n
n=1
P
(1)n+1
EXAMPLE 1.31 The series
is conditionally convergent.
n=1
n
P
(1)n+1
EXAMPLE 1.32 The series
is absolutely convergent.
n=1
n2
P
(1)n+1
EXAMPLE 1.33 The series
is absolutely convergent.
n=1
n!
P
sin(n)
is absolutely convergent:
EXAMPLE 1.34 For any R, the series
n=1
n2
Note that
sin(n)
1
n2 n2 n N.
P
P sin(n)
1
Since
converges,
by
comparison
test,
2
n=1 n
n=1 n2 also converges.
EXAMPLE 1.35 The series
X
(1)n log n
is conditionally convergent. To see
n log(log n)
n=3
log n
. Since n log n log(log n) we have
this, let un =
n log(log n)
1
log n
1
n
n log(log n)
log(log n)
()
29
P
Theorem 1.23 Suppose
n=1 an is an absolutely convergent
P series and (bn ) is a
sequence obtainedPby rearranging
the
terms
of
(a
).
Then
n
n=1 bn is also absolutely
P
convergent, and
a
=
b
.
n=1 n
n=1 n
P
Theorem 1.24 Suppose
n=1 an is a conditionally convergent series. The for
every R, there exists a P
sequence (bn ) whose terms are obtained by rearranging
the terms of (an ) such that
n=1 bn = .
To illustrate the last theorem consider the conditionally convergent series
Consider the following rearrangement of this series:
1
Thus, if an =
n=1
(1)n+1
.
n
1 1 1 1 1
1
1
1
+ + +
+ .
2 4 3 6 8
2k 1 4k 2 4k
(1)n+1
n
b3k2 =
1
,
2k 1
b3k1 =
1
,
4k 2
n=1 bn ,
b3k =
where
1
4k
n=1 an
and
1 1 1 1 1
1
1
1
3k = 1 + + +
2 4 3 6 8
2k 1 4k 2 4k
1 1
1 1 1
1
1
1
=
1
+
+ +
2 4
3 6 8
2k 1 4k 2 4k
1 1
1 1
1
1
=
+ +
2 4
6 8
4k 2 4k
1
1
1 1
1
1
=
1
+
+ +
2
2
3 4
2k 1 2k
1
=
s2k .
2
Also, we have
3k+1 = 3k +
1
,
2k + 1
3k+2 = 3k +
1
1
.
2k + 1 4k + 2
s
lim 3k+1 = ,
k
2
s
lim 3k+2 = .
k
2
30
1.2.5
Additional exercises
M.T. Nair
(These problems were prepared for the students of MA1010, AugNov, 2010.)
X
n=1
3n 2
=1
n(n + 1)(n + 2)
X
(n!)2
(a)
(2n)!
n=1
(d)
(g)
X
n=1
X
n=1
n
n+1
n2
X
(n!)2 n
(b)
5
(2n)!
n=1
n2
4n
1
+ 2n + 3
(e)
X
n=1
(h)
n2
X
n
(c)
n+1
n=1
1
n+1 n
n
(f )
X
(1)(n1)
n
n=1
1
2
3
n
+ 2 + 3 + ... + n + ...
2 2
2
2
1
1
1
1
(i) + + + ... +
+ ...
10
20
30
10n
1
1
1
1
(k)
+
+
+ ... +
+ ...
3
3
3
3
n+6
7
8
9
4 9
n2
1
2
3
n
(l) +
+
+...+
+...
2
3
4
n+1
(j) 2 +
(m)
3 4
n+1
+ + ... +
+ ...
2 3
n
1 2 3
n
+ + +...+ 2
+...
2 5 10
n +1
+ ... +
+ ...
n1
n+1
21
2+1
31
3+1
Does the above series converge?
Justify your answer.
4. Find out whether (or not) the following series converge absolutely or conditionally:
1
1
1
1
1
+ 2 2 + 2 + ... + (1)n+1
+ ...
2
3
5
7
9
(2n 1)2
1
1
1
1
1
(b)
+ ... + (1)n
+ ...
ln 2 ln 3 ln 4 ln 5
ln n
(a) 1
1
1
1
+
+ ... +
+ ...
1.2.3 2.3.4
n.(n + 1).(n + 2)
p
p
X
X
X
1
1
(a)
(b)
(c)
n4 + 1 n4 1
n!
(n + 1) n
n=1
n=1
n=1
(d)
(f )
1
,
(a + n)p (b + n)p
n=1
p
X
3
n3
+1n
a, b, p, q > 0
(g)
n=1
(i)
n=1
(n + 6)1/3
(j)
n=1
X
nx n
(l)
1+n
(o)
n=1
(1)n+1
(m)
2n!
n!
(log n)n
n=1
tan
n=1
1
n
1
n(n + 1)
n=1
2
X
1 n
(k)
1+
n
(h)
n=1
(1)n1
n=1
n+1
(e)
31
n
(p)
n
n+1
(1)n1
p
n(n + 1)(n + 2)
n=1
X
(2)n
n=1
n=1
(n)
n2
X
X
X
(1)n+1
(1)n+1 n
(1)n
(a)
(b)
(c)
(2n 1)2
3n1
n(log n)2
n=1
(d)
n=1
X
(1)n log n
n=1
n log log n
(e)
n=1
(1)n+1
n=1
1
n
(f )
(1)n+1
n=1
1
n2n
2
Limit, Continuity and Differentiability
of Functions
In this chapter we shall study limit and continuity of real valued functions defined
on certain sets.
2.1
Limit of a Function
2.1.1
32
Limit of a Function
33
2.1.2
x D, 0 < x a < ,
xa
f (x) b as
x a.
Thus, limxa f (x) = b if and only if for every open interval Ib containing b there
exists an open interval Ia containing a such that
x Ia (D \ {a}) = f (x) Ib .
In the following, whenever we talk about limit of a function f as x
approaches a R, we assume that f is defined on a set D R and a is
a limit point of D.
Exercise 2.1 Show that, if limit of a function, if exists, is unique.
Exercise 2.2 Show that, if lim f (x) = b, then there exists a deleted neighbourhood
xa
34
M.T. Nair
If you see the definitiion of existence of a limit, then you see that, the above statement means the following:
For any b R, there exists > 0 such that for any > 0, there is atleast
one x (a , a + ) such that f (x ) 6 (b , b + ).
How do we show, by first principle, i.e., from the definition, that a function f does
not have a limit as x approaches to a particular point x0 ?
Assume that the limxx0 f (x) exists and it is equal to for some R.
Then we know that for every > 0, there exists := such that
0 < x x0  < = f (x)  < .
Then, for a particular choice of , say = 0 , one must be able to find
an x0 such that x0 x0  < 0 but f (x)  0 .
Let us illustrate this by a simple example.
0, 1 x 0,
We show that
1, 0 < x 1.
lim x 0f (x) does not exist. Suppose limx0 f (x) = for some R and let
> 0 be given. Then, we know from the definition of the limit that there exists
:= such that
Let f : [1, 1] R be defined by f (x) =
x 6= 0,
x (, ) = f (x) ( , + ).
We know that, f (x) takes the values only 0 and 1 and the length of the interval
( , + ) is 2. Hence, if we take 0 such that 20 < 1, then f (x) cannot belong
to ( 0 , +0 ) for all x in a neighbourhood of 0. To see this consider the following
cases:
case (i): = 0. In this case, f (x) 6 ( 0 , + 0 ) for any x > 0.
case (ii): = 1. In this case, f (x) 6 ( 0 , + 0 ) for any x < 0.
case (iii): 6= 0, 6= 1. In this case, f (x) 6 ( 0 , + 0 ) for any x0.
Thus, we arrive at a a contradiction to our assumption.
Limit of a Function
35
we have f (xn ) b.
Proof. Suppose lim f (x) = b. Let (xn ) be a sequence in D such that xn a. Let
xa
> 0 be given. We have to show that there exists n0 N such that f (xn ) b <
for all n n0 .
Since lim f (x) = b, we know that there exists > 0 such that
xa
()
Also, since xn a, there exists n0 N such that xn a < for all n n0 . Hence,
from (), we have f (xn ) b < for all n n0 .
What about the converse of the above theorem? The converse is also true, in a
slight restricted sense.
Theorem 2.3 If for every sequence (xn ) in D which is not eventually constant, the
sequence (f (xn )) converges to b, then lim f (x) = b.
xa
Proof. Suppose for every sequence (xn ) in D which is not eventually constant,
the sequence (f (xn )) converges to b. Assume for a moment that f does not have the
limit b as x approaches a. Then, by the definition of the limit, there exists 0 > 0
such that for every > 0, there exists at least one x D such that
x 6= a
and x a < ,
but f (x ) b > 0 .
1
,
n
Thus, (xn ) is not eventually constant such that xn a but f (xn ) 6 b. Thus we
arrive at a contradiction to our hypothesis.
Remark 2.3 The advantage of Theorem 2.2 is that, if we can find a sequence (xn )
in D such that xn a, but (f (xn )) does not converge, then we can assert that
lim f (x) does not exist. Similarly, by Theorem 2.3, we can determine lim f (x) if
n
we are able to show convergence of (f (xn )) to some b for any arbitrary (not for a
specific) noneventually constant sequence (xn ) in D which converges to a.
36
2.1.3
Some properties
M.T. Nair
The following two theorems can be proved using Theorems 2.2 and 2.3, and the
results on convergence of sequences of real numbers.
Theorem 2.4 We have the following.
(i) If lim f (x) = b and lim g(x) = c, then
xa
xa
xa
xa
6= 0 in a deleted neighbourhood of a
1
.
b
Corollary 2.7 If lim f (x) = b, lim g(x) = c and f (x) g(x) for all x in a deleted
xa
xa
neighbourhood of a, then b c.
Theorem 2.8 Suppose lim f (x) = b and lim g(y) = c. Then lim g(f (x)) = c.
xa
yb
xa
Proof. Let > 0 be given. Then there exists 1 > 0 such that
y b < 1 = g(y) c < .
Also, let 2 > 0 be such that
x a < 2 = f (x) b < 1 .
Hence,
x a < 2 = f (x) b < 1 = g(f (x)) c < .
This completes the proof.
Exercise 2.3 Write details of the proof of Theorem 2.4 and Corollary 2.6 and
Corollary 2.7.
J
Limit of a Function
37
and lim f (x) exists, then prove that lim f ((x)) exists and
xx0
xx0
xx0
2.1.4
Some Examples
x2 4
x2 .
(x + 2)(x 2)
= (x + 2).
x2
38
M.T. Nair
EXAMPLE 2.4 Let D = R \ {0} and f (x) = x1 . Then lim f (x) does not exist. To
x0
see this consider the sequence (xn ) with xn = 1/n for n N. Then we have xn 0
but (f (xn ) diverges to infinity. Therefore, by Theorem 2.2, lim f (x) does not exist.
x0
whenever
But,
f (x) > 1 + b x <
1
.
1 + b
whenever
x <
1
.
1 + b
Thus, we have proved that it is not possible to find a > 0 such that f (x) b < 1
for all x with x < .
EXAMPLE 2.5 We show that (i) lim sin(x) = 0 and (ii) lim cos(x) = 1.
x0
x0
Also,
x0
x0
Also, since cos x = 1 2 sin2 (x/2) and lim sin(x/2) = 0, Theorem 2.4(i) implies
x0
lim cos x = 1.
x0
sin x
= 1.
x0 x
It can be seen, using the graph of sin x that
0<x<
0<x<
sin x
= cos x <
< 1.
2
x
Hence,
Since
sin(x)
x
sin x
x
sin x
= cos x <
< 1.
2
x
sin x
= 1.
x0 x
Limit of a Function
39
exists. Prove that lim f (x) = 0 and lim f (x) = f (c) for every c R.
xc
x0
Hint: Use the facts that f (2x) = 2f (x), Theorem 2.8 and f (x) f (c) = f (x c). J
Exercise 2.6 Suppose is a function defined in a neighbourhood I0 of a point x0
such that
x I0 , x x0  < r = (x) x0  < r
r > 0.
If f is also a function defined in a neighbourhood of x0 and lim f (x) exists, then
xx0
prove that lim f ((x)) exists and limxx0 f ((x)) = limxx0 f (x).
xx0
2.1.5
Definition 2.3 Let f be a real valued function defined on a set D R, and let
a R be a limit point of D.
(i) We say that f (x) has the left limit b R as x approaches a from left
if for every > 0, there exists > 0 such that
f (x) b < whenever
x D, a < x < a.
(ii) We say that f (x) has the right limit b R as x approaches a from
right if for every > 0, there exists > 0 such that
f (x) b < whenever
x D, a < x < a + .
xa
f (x) b as
x a ,
and if f (x) has the right limit b R as x approaches a from right we write
lim f (x) = b or
xa+
f (x) b as
x a+ .
xa
xa+
xa
xa
xa+
xa+
40
M.T. Nair
In view of the above theorem, if (i) lim f (x) does not exist or (ii) lim f (x)
xa
xa+
does not exist or (iii) both lim f (x) and lim f (x) exist but lim f (x) 6= lim f (x),
xa
xa
xa+
xa+
2.1.6
Limit at and at
Definition 2.4 Suppose a function f is defined on an interval of the form (a, ) for
some a > 0. Then we say that f has the limit b as x and write lim f (x) = b,
x
if for every > 0, there exits M > a such that
f (x) b < whenever
x > M.
x < M.
x0
x0
x0
x0
J
Definition 2.6
that
Limit of a Function
41
2. lim f (x) = if for every M > 0, there exists > 0 such that
xa
lim f (x) = + if for every M > 0, there exists > 0 such that
x+
lim f (x) = if for every M > 0, there exists > 0 such that
x+
lim f (x) = + if for every M > 0, there exists > 0 such that
lim f (x) = if for every M > 0, there exists > 0 such that
(Of course, one has to show that this limit exists; which is true!).
1 n
exists, and we denoted it by e. Now
EXAMPLE 2.7 Recall that lim 1 +
n
n
we show that
1 x
lim 1 +
= e.
x
x
Let > 0 be given. We have to find an M > 0 N such that
1 x
e< 1+
< e + whenever x > M.
()
x
Now, we can see that, for every n N, if x R is such that n x n + 1, then
1+
1
1
1
1+ 1+
n+1
x
n
42
so that
1+
Thus is is same as
M.T. Nair
1 n
1 x
1 n+1
1+
1+
.
n+1
x
n
1 x
n 1 +
n ,
x
where
n := 1 +
1 n+1
1 1
1+
,
n+1
n+1
1
1 n
n := 1 +
1+
.
n
n
e < n < e + .
Now, take M = n0 and let x > M . Take n n0 such that n + 1 x n. For this
n and x, we have
1 x
e < n 1 +
n < e + .
x
Thus, we obtained an M > 0 such that
1 x
e< 1+
< e + whenever x > M.
x
Thus, we have proved ().
Using the arguments used in the above example, we obtain a more general result.
Theorem 2.10 Suppose (n ) and (n ) are sequences of positive real numbers and
f is a (real valued) function defined on (0, ) having the following property: For
n N, x R,
n < x < n + 1 = n f (x) n .
If (n ) and (n ) converge to the same limit, say b, then lim f (x) = b.
x
2.1.7
Additional Exercises
x
= 1.
x3 4x 9
x,
if x < 1,
2. Show that the function f defined by f (x) =
does not
1 + x, if x 1
have the limit as x 1.
3 x, if x > 1,
1,
if x = 1, Find lim f (x). Is it f (1)?
3. Let f be defined by f (x) =
x1
2x,
if x, 1.
1. Using the definition of limit, show that lim
b. If b 6= 0, then show that there exists > 0 such that f (x) 6= 0 for every
x (x0 , x0 + ) D0 .
Continuity of a Function
5. Let f be defined by f (x) =
43
1, if x Q,
Show that
0, if x 6 Q.
6. Suppose lim f (x) = and lim g(x) = b. Show that lim g(f (x)) = b.
x
7. Let f : (0, ) R be such that lim f (x) = b. Show that lim f (x1 ) = b.
x0
2.2
2.2.1
Continuity of a Function
Definition and some basic results
44
M.T. Nair
2.2.2
Some examples
x2 4
x2
EXAMPLE 2.11 The functions f and g defined by f (x) = sin x and g(x) = cos x
for x R are continuous at 0.
Note that for x, y R,
sin x sin y = 2 sin
x y
2
cos
x + y
2
so that
 sin x sin y x y
x, y R.
sin x
x
Continuity of a Function
45
EXAMPLE 2.13 Let f (x) = x1 for x 6= 0. Then there does not exist a continuous
function g defined on R such that g(x) = f (x) for all x 6= 0.
EXAMPLE 2.14 The function f defined by f (x) = 1/x, x 6= 0 is continuous at
every x0 6= 0:
Note that for x 6= 0, x0 6= 0,
1
1 = x x0  2x x0 
x x0
xx0 
x20 
whenever
x x0 
x0 
2
since x = x0 (x0 x) x0  x0 x. Hence, for any > 0,
2
1
1 < whenever x x0  < := min{ x0 , x0 }.
x x0
2
2
Thus, f is continuous at every x0 6= 0.
EXAMPLE 2.15 Let f be defined by f (x) = 1/x on (0, 1]. Then there does not
exist a continuous function g on [0, 1] such that g(x) = f (x) for all x (0, 1].
Suppose g is any function defined on [0, 1] such that g(x) = f (x) for all x (0, 1].
Then we have 1/n 0 but g(1/n) = f (1/n) = n . Thus, g(1/n) 6 g(0).
( x + x0 ) x x0 , we have
x x0 
x x0 
 x x0  =
.
x0
x + x0
Thus,
 x x0  < whenever
x x0  < := x0 .
x < k .
46
M.T. Nair
so that
1/k
x1/k x0  =
x x0 
x x0 
.
k1
y k1 + y k2 y0 + . . . + yy k2 + y0
y0k1
Thus,
1/k
11/k
x x0  < := y0k1 = x0
.
EXAMPLE 2.18 For a rational number r, let f (x) = xr for x > 0. Then using
the above example together with Theorem 2.13, we see that f is continuous at every
x0 > 0.
We know that given r R, there exists a sequence (rn ) of rational numbers such
that rn r. For n N, let fn (x) = xrn , x > 0. Since each fn is continuous for
x > 0, one may enquire whether the function f defined by f (x) = xr is continuous
for x > 0.
First of all how do we define the xr for x > 0?
We shall discuss this issue in the next subsection, where we shall introduce two
important classes of functions, namely, exponential and logarithm functions. In fact,
our discussion will also include, as special cases, the Examples 2.16  2.18.
2.2.3
We have already come across expression such as ab for a > 0 and b R, though we
have not proved existence of such numbers, and also defined a number denoted by
X
1 1/n
1
e as the limit of the sequence 1 +
or the series
. Now, we formally
n
n!
n=0
define the following functions:
Exponential function: ex , x R.
Natural logarithm function: ln x, x > 0.
Exponential function: ax , x R for a given a > 0.
Logarithm function with base a > 0:
loga x, x > 0.
P
xn
First, we observe that for every x R, the series
n=0 n! converges absolutely.
This can be seen by using the ratio test. This series plays a very significant role in
mathematics.
Continuity of a Function
47
X
xn
n=0
n!
and
exp(1) = e.
In order to derives some of the important properties of the function exp(x), the
student is urged to do the following exercise.
P
P
Exercise 2.9 Suppose that series
n=0 bn are absolutely convergent.
n=0 an and
Then, the series
n
X
X
ak bnk
()
cn with cn :=
n=0
k=0
P
P
is absolutely convergent. Further, show that if =
n=0 bn , then
n=0 an and =
P
c
=
.
n=0 n
P
The
series
Pn=0 cn defined in () is called the Cauchy product of the series
P
J
n=0 cn .
n=0 cn and
Using the conclusion in the above exercise, it can be proved that
exp(x + y) = exp(x) exp(y)
x, y R.
()
1
exp(x)
have
exp(x) > 0
x R.
and
exp(x) = 1 x = 0.
[From the definition, x > 0 implies exp(x) > 1. Next, suppose x 0. If x = 0, then
exp(x) = exp(0) = 1. If x < 1, then taking y = x, we have y > 1, and hence from
the first part, exp(y) > 1, i.e., 1/ exp(x) = exp(x) > 1 so that exp(x) < 1. Hence,
exp(x) > 1 x > 0. From this, we get exp(x) = 1 x = 0.]
x > y exp(x) > exp(y).
[x > y x y > 0 exp(x y) > 1. But, exp(x y) = exp(x)/ exp(y).]
48
k Z.
M.T. Nair
k N, we have
k N.
m, n N. Hence,
exp(r) = er
r Q.
provided the above limit exists. Thus, our next attempt is to show that the function
exp(x), x R, is continuous.
In view of these observations, we use the following
Proposition 2.15 The following results hold.
(i) exp(x) as x .
(ii) exp(x) 0 as x .
Proof. For x 0, we have
exp(x) = 1 + x +
X
xn
n=2
n!
1 + x.
From this (i) follow. To see (ii), let x < 0 and y = x. Then y > 0 so that by (ii),
exp(x) = exp(y) =
1
0
exp(y)
as
y .
Thus, exp(x) 0 as x .
NOTATION: For brevity of expression, we shall use the notation ex for exp(x).
Theorem 2.16 The function exp(x) is continuous on R
Continuity of a Function
49
X
(x x0 )n
n!
n=1
= ex0 (x x0 )
X
(x x0 )n1
n=1
n!
Thus, if x x0  1, then
X
1
= ex0 (e 1)x x0 .
e e  e x x0 
n!
x
x0
x0
n=1
as
ex as
x ,
x ,
x > 0,
50
M.T. Nair
x > 0,
y = ln x ey = x.
y = loga x ay = x.
loga x =
ln x
.
ln a
Theorem 2.19 The functions ln x and loga x for a > 0 are continuous on (0, ).
Proof. Let x, x0 belong to the interval (0, ), and let y = ln x and y0 = ln x0 .
Then we have ey = x and ey0 = x0 . Assume, without loss of generality that x > x0 .
Since ea > 1 if and only if a > 0, we have y > y0 , and hence
x x0 = ey ey0 = ey0 (eyy0 1) = ey0
X
(y y0 )n
n=1
n!
ey0 (y y0 ).
Hence,
y y0  ey0 x x0 .
Thus, for > 0, we have y y0  < whenever x x0  < ey0 , ln x is continuous on
(0, ). Since loga x = ln x/ ln a, the function loga x is also continuous on (0, ).
Theorem 2.20 For r R, the function f defined by f (x) = xr is continuous at
every point x (0, ).
Proof. For r R and x > 0, we have xr = er ln x . Hence, the result follows from
Theorem 2.19 and Theorem 2.13.
Remark 2.5 Often, the notation log x is used for the natural logarithm function
instead of ln x.
Continuity of a Function
2.2.4
51
Recall that a subset S of R is said to be bounded if there exists M > 0 such that
s M for all s S, and set which is not bounded is called an unbounded set.
Recall that if S is a bounded subset of R, then S has infimum and supremum.
Exercise 2.12 Let S R. Prove the following:
(i) Suppose S is bounded, and say := inf S and := sup S. Then there exist
sequences (sn ) and (tn ) in S such that sn and tn .
(ii) S is unbounded if and only if there exists a sequence (sn ) in S which is
unbounded.
(iii) S is unbounded if and only if there exists a sequence (sn ) in S such that
sn  as n .
(iv) If (sn ) is a sequence in S which is unbounded, then there exists a subsequence
(skn ) of (sn ) such that skn  as n .
(v) If (sn ) is a sequence in S such that sn  as n , and if (skn ) is
J
subsequence of (sn ), then skn  as n .
Definition 2.14 A real valued function defined on a set D R is said to be a
bounded function if the set {f (x) : x D} is is bounded. A function is said to
be an unbounded function if it is not bounded.
The following can be easily deduced from the definition:
A function f : D R is bounded if and only if there exists M > 0 such that
f (x) M for all x D.
A function f : D R is not unbounded if and only if there exists a sequence
(xn ) D such that the f (xn ) as n .
Theorem 2.21 Suppose f is a real valued function defined on a closed and bounded
interval [a, b]. Then f is a bounded function.
Proof. Assume for the time being that f : [a, b] R is not a bounded function.
Then for every n N, there exists a sequence (xn ) in [a, b] such that f (xn )
as n . Since (xn ) is a bounded sequence, by BolzanoWeierstrass property of
R, there exists a subsequence (xkn ) of (xn ) such that xkn x for some x [a, b].
Therefore, by continuity of f , f (xkn ) f (x). In particular, (f (xkn ) is a bounded
sequence. This is a contradiction to the fact that f (xn ) as n .
Thus, we have proved that continuous function cannot be unbounded.
52
M.T. Nair
Proof. By the definition of infimum and supremum, there exist sequences (xn )
and (yn ) in [a, b] such that f (xn ) and f (yn ) as n . Since (xn )
and (yn ) are bounded sequences, there exist subsequences (xkn ) and (ykn ) of (xn )
and (yn ), respectively, such that xkn x and ykn y for some x, y in [a, b]. By
continuity of f , f (xkn ) f (x) and f (ykn ) f (y) as n . But, we already have
f (xkn ) and f (ykn ) . Hence, = f (x) and = f (y).
Remark 2.6 By Theorem 2.22, we say that the infimum and supremum of a continuous real valued function f defined on a closed and bounded interval [a, b] are
attained at some points in [a, b], and in that case, we write
inf{f (x) : x [a, b]} = min f (x),
axb
The conclusion in the above theorem need hold if the domain of the function is not
of the form [a, b] or if f is not continuous. For example, f : (0, 1] R defined by
f (x) = 1/x for x (0, 1] is continuous, but does not attain supremum. Same is the
case if g : [0, 1] R is defined by
1
x , x (0, 1],
g(x) =
1, x = 0.
Thus, neither continuity nor the fact that the domain is a closed and bounded
interval cannot be dropped. This does not mean that the conclusion in the theorem
does not hold for all such functions! For example f : [0, 1) R defined by
0, x [0, 1/2),
f (x) =
1, x [1/2, 1).
Then we see that neither f is continuous, nor its domain of the form [a, b]. But, f
attains both its maximum and minimum.
Intermediate value theorem
Suppose f is a continuous real valued function f defined on a closed and bounded
interval [a, b], and
:= min f (x),
:= max f (x).
axb
axb
Clearly,
f (x)
x [a, b].
Continuity of a Function
53
Now, the question is whether every value between and is attained by the function. The answer is in affirmative. In fact we have the following general theorem,
known as Intermediate value theorem1
Theorem 2.23 (Intermediate value theorem) Suppose f is a continuous real
valued function defined on an interval I. Suppose x1 and x2 are in I such that
f (x1 ) < f (x2 ), and c is such that f (x1 ) < c < f (x2 ). Then there exists x0 lying
between x1 and x2 such that f (x0 ) = c.
Proof. Without loss of generality assume that x1 < x2 . Let
S = {x [x1 , x2 ] : f (x) < c}.
Then S is nonempty (since x1 S) and bounded above (since x x2 for all x S).
Let
:= sup S.
Then there exists a sequence (an ) in S such that an . Note that [x1 , x2 ].
Hence, by continuity of f , f (an ) f (). Since f (an ) < c for all n N, we have
f () c. Note that 6= b (since f () c < f (x2 )).
Now, let (bn ) be a sequence in (, x2 ) such that bn . Then, again by
continuity of f , f (bn ) f (). Since bn > , bn 6 S and hence f (bn ) c. Therefore,
f () c. Thus, we have prove that there exists x0 := such that f (x0 ) c f (x0 )
so that f (x0 ) = c.
The following two corollaries are immediate consequences of immediate from the
above theorem.
Theorem 2.24 Let f be a continuous function defined on an interval. Then range
of f is a an interval.
Corollary 2.25 Suppose f is a continuous real valued function defined on an interval I. If a, b I satisfy a < b and f (a)f (b) < 0, then there exists x0 I such
that
a x0 b and f (x0 ) = 0.
2.2.5
Additional Exercises
Proof taken from the book: A Course in Calculus and Real Analysus by S.R. Ghorpade and
B.V. Limaye (IIT Bombay), Springer, 2006.
54
M.T. Nair
3. There does not exist a continuous function f from [0, 1] onto R Why?
4. Find a continuous function f from (0, 1) onto R.
5. Suppose f : [a, b] [a, b] is continuous. Show that there exists c [a, b] such
that f (c) = c.
6. There exists x R such that 17x19 19x17 1 = 0 Why?
7. If p(x) is a polynomial of odd degree, then there exists at least one R such
that p() = 0.
8. Suppose f : R R is continuous such that f (x) 0 as x . Prove that
f attains either a maximum or a minimum.
9. Suppose f : [a, b] R is continuous such that for every x [a, b], there exists
f (x)
a y [a, b] such that f (y)
. Show that there exists [a, b] such
2
that f () = 0.
1
10. Suppose f : [a, b] [a, b] is continuous such that there f (x) f (y) x y
2
for all x, y [a, b]. Show that there exists [a, b] such that f () = .
2.3
Differentiability of functions
xx0
f (x) f (x0 )
x x0
exists, and in that case the value of the limit is called the derivative of f at x0 .
The derivative of f at x0 , if exists, is denoted by
f 0 (x0 )
or
df
(x0 ).
dx
Differentiability of functions
55
0 (x0 ) = 0 (x0 ).
J
2.3.1
f (x0 + h) f (x0 )
h f 0 (x0 ).0 = 0
h
as h 0.
Thus, f is continuous at x0 .
For the following theorem, we may recall that if is continuous at a point x0 and
(x0 ) 6= 0, then there exists an open interval I0 containing x0 such that (x) 6= 0
for all x I0 .
Theorem 2.27 (Products and quotient rules) Suppose f and g defined on I
are differentiable at x0 I. Then the function (x) := f (x)g(x) is differentiable at
x0 , and
0 (x0 ) = f 0 (x0 )g(x0 ) + f (x0 )g 0 (x0 ).
()
If g(x0 ) 6= 0, then the function (x) := f (x)/g(x) is differentiable at x0 , and
0 (x0 ) =
()
f (x0 + h) f (x0 )
g(x0 + h) g(x0 )
g(x0 + h) + f (x0 )
h
h
f 0 (x0 )g(x0 ) + f (x0 )g 0 (x0 ) as h 0.
=
56
M.T. Nair
we have
(x0 + h) (x0 )
h
0)
f (x0 ) g(x0 +h)g(x
h
g(x0 + h)g(x0 )
0
f (x0 )g(x0 ) f (x0 )g 0 (x0 )
as h 0.
[g(x0 )]2
0 (x0 )
.
g 0 (y0 )
0 (x0 )
.
f 0 (x0 )
(g f )(x) (g f )(x0 )
.
x x0
Differentiability of functions
57
(g f )(xn ) g f )(x0 )
xn x0
g(yn ) g(y0 )
xn x0
g(yn ) g(y0 ) f (xn ) f (x0 )
.
yn y0
xn x0
g(yn ) g(y0 )
g 0 (y0 ).
yn y0
(xn )
g(yn )g(y0 )
yn y0
(g f )0 (x0 )
.
g 0 (y0 )
(g f )0 (x0 )
.
g 0 (y0 )
(iii) Proof of this part is analogous to the proof of (ii). Hence, we omit the
details.
Remark 2.7 The part (ii) and (iii) of Theorem 2.28 is not available in standard
books on Calculus. I found it useful while discussing derivative of logarithm function
in next section (Section 2.3.2).
Exercise 2.17 Prove part (iii) of Theorem 2.28.
2.3.2
Some Examples
We shall assume that the students are familiar with the following:
For c R, if f (x) = c, x R, then f 0 (x) = 0
x R.
x R.
58
M.T. Nair
x R.
x R.
x D.
x R.
ex+h ex
ex
ex X hn
ex = (eh 1 h) =
.
h
h
h
n!
n=2
Now, if h 1, then hn h2 for all n {2, 3, . . .}. Thus,
x+h
x
X
e
1
e
x
e ex h
= ex h(e 2).
h 1 =
h
n!
n=2
EXAMPLE 2.20 For a > 0, the function ax is differentiable for every x R and
(ax )0 = ax ln a
x R.
1
,
x
x > 0.
To see this, let f (x) = ln x and g(x) = ex . Then we have g(f (x) = x for every x > 0.
Since g f is differentiable, g is differentiable, and g 0 (y) = ey 6= 0 for every y R,
by Theorem 2.28, f is differentiable for every x > 0 and we have g 0 (f (x))f 0 (x) = 1.
Thus,
1 = eln x (ln x)0 = x(ln x)0
so that (ln x)0 = 1/x.
Differentiability of functions
59
EXAMPLE 2.22 For a > 0, the function loga x is differentiable for every x > 0,
and
1
(loga x)0 =
,
x > 0.
x ln a
We know that
ln x
loga x =
.
ln a
1
Hence, (loga x)0 =
for every x > 0.
x ln a
EXAMPLE 2.23 For r R, let f (x) = xr for x > 0. Then f is differentiable for
every x > 0 and
f 0 (x) = rxr1 ,
x > 0.
By the composition rule in Theorem 2.28,
f 0 (x) = (er ln x )0 = er ln x
r
xr r
=
= rxr1 .
x
x
1
,
x
x 6= 0.
(ii) For a > 0, the function loga x is differentiable for every x R with x 6= 0,
and
1
(loga x)0 =
,
x 6= 0.
x ln a
J
2.3.3
Recall from Theorem 2.22 that if f : [a, b] R is a continuous function, then there
exists x0 , y0 in [a, b] such that
f (x0 ) f (x) f (y0 )
x [a, b].
60
M.T. Nair
x I (x1 , x1 + ),
x I (x2 , x2 + ).
Differentiability of functions
61
Remark 2.9 A function can have more than one maximum and minimum. For
example, consider
f (x) = sin(4x),
[0, ].
We see that f has maximum value 1 at /8 and 5/8, and has minimum value 1
at 3/8 and 7/8.
Remark 2.10 (a) In view of Theorem 2.29, if a function f is differentiable at an
interior point x0 of an interval I and f 0 (x0 ) 6= 0, then f can not have local maximum
or local minimum at x0 .
(b) It is to be observed that in order to have a maximum or minimum at a point
x0 , the function need not be differentiable at x0 . For example
f (x) = 1 x,
x 1,
x 1,
x < 1.
In this example, we have f 0 (0) = 0. Note that f has neither local maximum nor
local minimum at 0.
Definition 2.18 Suppose f is defined n an interval I and x0 is an interior point
of I. If f 0 (x0 ) exists and f 0 (x0 ) = 0 or if f 0 (x0 ) does not exist, then x0 is called a
critical point of f .
In Section 2.3.6 we shall give some sufficient conditions for existence of local
exrema of functions. Now, let us derive some important consequences of Theorem
2.29.
2.3.4
Rolles theorem
Theorem 2.30 (Rolles theorem) Suppose f is a continuous function defined on
a closed and bounded interval [a, b] such that it is differentiable at every x (a, b).
If f (a) = f (b), then there exists c (a, b) such that f 0 (c) = 0.
Proof. Let g(x) = f (x) f (a). Then we have g(a) = 0 = g(b), and g 0 (x) = f 0 (x)
for every x (a, b).
62
M.T. Nair
x [a, b].
If g(x1 ) = g(x2 ), then g is a constant function and hence g 0 (x) = 0 for all x [a, b].
Hence, assume that g(x2 ) < g(x1 ). Then, either g(x1 ) 6= 0 or g(x2 ) 6= 0. Assume
that g(x2 ) 6= 0, so that x2 6= a and x2 6= b. Hence, by Theorem 2.29, g 0 (x2 ) = 0.
Thus, f 0 (x2 ) = 0.
Similarly, if g(x1 ) 6= 0, then we shall arrive at f 0 (x1 ) = 0.
Exercise 2.19 Show that between any two roots of the equation ex cos x 1 = 0,
there is at least one root of the equation ex sin x 1 = 0.
J
Lagranges mean value theorem
As a corollary to Rolles theorem we obtain the following.
Theorem 2.31 (Lagranges mean value theorem) Suppose f is a continuous
function defined on a closed and bounded interval [a, b] such that it is differentiable
at every x (a, b). Then there exists c (a, b) such that
f (b) f (a) = f 0 (c)(b a).
Proof. Let
f (b) f (a)
(x a),
x [a, b].
ba
Note that is continuous on [a, b], differentiable in (a, b), (a) = 0 = (b), and
(x) := f (x) f (a)
f (b) f (a)
,
x (a, b).
ba
By Rolles theorem (Theorem 2.30), there exists c (a, b) such that 0 (c) = 0.
Thus, f (b) f (a) = f 0 (c)(b a).
0 (x) := f 0 (x)
x (a, b).
x [a, b].
Differentiability of functions
63
f (b) f (a)
[g(x) g(a)],
g(b) g(a)
x [a, b].
Note that is continuous on [a, b], differentiable in (a, b), (a) = 0 = (b), and
0 (x) := f 0 (x)
f (b) f (a) 0
g (x),
g(b) g(a)
x (a, b).
By Rolles theorem (Theorem 2.30), there exists c (a, b) such that 0 (c) = 0. This
completes the proof.
Exercise 2.20 Let 0 < a < b. Show that for every n N, a <
[Hint: take f (x) = xn+1 and g(x) = xn .]
n[bn+1 an+1 ]
< b.
(n + 1)[bn an ]
J
64
M.T. Nair
LHospitals rules
Theorem 2.33 (LHospitals rule)2 Suppose f and g are continuous functions
on [a, b] which are differentiable at every point in (a, b), except possibly at x0 [a, b].
f 0 (x)
f (x)
Suppose f (x0 ) = 0 = g(x0 ) and lim 0
exists. Then lim
exists and
xx0 g (x)
xx0 g(x)
lim
xx0
f (x)
f 0 (x)
= lim 0
.
g(x) xx0 g (x)
f 0 (x)
exists, there exists a deleted neighbourhood D0 of x0
xx0 g 0 (x)
such that g 0 (x) 6= 0 for x D0 [a, b]. By Cauchys generalized mean value theorem
(Theorem 2.32), for every x D0 , there exists x between x and x0 such that
Proof. Since lim
f (x) f (x0 )
f 0 (x )
f (x)
=
= 0
.
g(x)
g(x) g(x0 )
g (x )
f 0 (x )
f 0 (x)
exists,
lim
exists and it is equal to
xx0 g 0 (x )
xx0 g 0 (x)
f 0 (x)
f (x)
f (x)
f 0 (x)
lim 0
(See Exercise 2.4). Thus, lim
exists and lim
= lim 0
.
xx0 g (x)
xx0 g(x)
xx0 g(x)
xx0 g (x)
This completes the proof.
Since x x0  < x x0  and lim
The following theorem is proved using the above theorem by using the change
of variable x 7 y := 1/x.
2
LHospital is pronounced as Lopital. The rule is named after the 17thcentury French mathematician Guillaume de lHospital, who published the rule in his book Analyse des Infiniment Petits
pour lIntelligence des Lignes Courbes (i.e., Analysis of the Infinitely Small to Understand Curved
Lines) (1696), the first textbook on differential calculus.
Differentiability of functions
65
Proof. Let f(y) = f (1/y) and g(y) = g(1/y) for 0 < y < 1/a. We note that
lim f (x) = 0 = lim g(x) lim f(y) = 0 = lim g(y).
y0
y0
Also, since
f0 (y) = [f (1/y)]0 = f 0 (1/y)(1/y 2 ),
we have
f 0 (x)
x g 0 (x)
lim
exists
lim
exists.
xx0
f 0 (x)
6= 0. In this case, since
g 0 (x)
xx0
xx0
xx0
the result follows from Theorem 2.35 by interchanging the roles of f and g.
To consider the general case where is not necessarily zero, let , x R such
that x x0  <  x0  and sufficiently close to x0 such that (x) 6 g(). This is
guaranteed because, g 0 (x) 6= 0 for x sufficiently close to x0 . Then there exists x,
lying between and x such that have
h
i
f ()
0
1
f (x)
f (x, )
f (x) f ()
f (x)
h
i
=
=
0
g()
g (x,)
g(x) g()
g(x) 1
g(x)
66
so that
f (x)
=
g(x)
f 0 (x, )
g 0 (x,)
1
h
1
f ()
f (x)
g()
g(x)
M.T. Nair
i
i.
x0
f 0 (x, )
f 0 (x)
=
lim
.
xx0 g 0 (x)
g 0 (x,)
h
1
x, )
h
Hence, using (, ) arguments, it can be shown that lim 0
x0 g (x,)
1
f (x)
exists and
so that lim
xx0 g(x)
i
h
f ()
0
1
f (x)
f (x, )
f (x)
f 0 (x)
i = lim 0
h
lim
= lim 0
.
xx0 g(x)
xx0 g (x)
x0 g (x,)
1 g()
f 0 (
f ()
f (x)
g()
g(x)
i
i exists
g(x)
xx0
can be treated analogously to the cases already discussed in the above theorems.
Taylors formula
Now, we state another important formula in calculus.
Theorem 2.37 (Taylors formula) Suppose f is defined and has derivatives
f (1) (x), f (2) (x), . . . , f (n+1) (x) for x in an open interval I and x0 I. Then, for
every x I, there exists x between x and x0 such that
f (x) = f (x0 ) +
n
X
f (j) (x0 )
j=1
j!
(x x0 )j +
f (n+1) (x )
(x x0 )n+1 .
(n + 1)!
n
X
f (j) (x0 )
j=1
j!
hj +
f (n+1) (x ) n+1
h
.
(n + 1)!
Differentiability of functions
67
n
X
f (j) (x0 )
j!
j=1
(t x0 )j ,
t I.
j {1, . . . , n}.
Now, let
g(t) = f (t) Pn (t) (x)(t x0 )n+1 ,
t I,
where
(x) :=
f (x) Pn (x)
.
(x x0 )n+1
Note that, by this choice of (x), we have g(x0 ) = 0 and g(x) = 0. Also, we have
g (1) (x0 ) = 0,
g (2) (x0 ) = 0,
...,
g (n) (x0 ) = 0.
Hence, by Rolles theorem, there exists x1 between x0 and x such that g 0 (x1 ) = 0.
Again, by Rolles theorem, there exists x2 between x0 and x1 such that g 00 (x2 ) = 0.
Continuing this, there exists x := xn+1 between x0 and xn such that g (n+1) (x ) = 0.
But,
g (n+1) (t) = f (n+1) (t) Pn(n+1) (t) (x)(n + 1)! = f (n+1) (t) (x)(n + 1)!.
Thus, we have
(x) =
f (n+1) (x )
,
(n + 1)!
so that
f (x) = f (x0 ) +
n
X
f (j) (x)
j=1
j!
(x x0 )j +
f (n+1) (x )
(x x0 )n+1 .
(n + 1)!
f (n+1) (x )
(x x0 )n+1
(n + 1)!
This proof adapted from S. Ghorpade & B.V. Limaye: A Course in Calculus and Analysis,
Springer, 2006
68
M.T. Nair
as
X
f (n) (x0 )
j!
n=1
(x x0 )n ,
x I.
()
Definition 2.20 If f can be represented as a series as in (), for all x in a neighbourhood of x0 , then such a series is called the Taylors series of f around the
point x0 .
A natural question that one may ask is:
Doe every infinitely differentiable function in a neighbourhood of x0 has
a Taylors series expansion?
Unfortunately, the answer is negative. For example, if we define
f (x) =
e1/x , x 6= 0,
0,
x = 0,
then it can be seen that f (0) = 0 and f k) (0) = 0 for all k N. Thus, f does not
have the Taylors series expansion around the point 0,
EXAMPLE 2.25 Using Taylors formula, we shall show that
sin x =
X
(1)n x2n+1
(2n + 1)!
n=0
x R.
For this, let f (x) = sin x and x0 = 0. Since f is infinitely differentiable, and
f 2j (0) = 0,
j N,
we have
f (x) = f (x0 ) +
2n+1
X
j=1
= f (x0 ) +
= f (x0 ) +
n
X
f (2j+1) (0)
j=0
n
X
j=0
(2j + 1)!
x2j+1 +
f (2n+2) (x ) 2n+2
x
(2n + 2)!
Differentiability of functions
69
as
n .
Therefore,
n
h
X
(1)j 2j+1 i
x
0
f (x) f (x0 ) +
(2j + 1)!
as
j=0
X
(1)n x2n+1
n=0
(2n + 1)!
x R.
x I,
k N {0}.
X
f (n) (x0 )
n=1
j!
(x x0 )n ,
x I.
J
Exercise 2.23 Using Taylors formula, prove the following:
(i) cos x =
X
(1)n x2n
(2n)!
n=0
(ii) ex =
X
xn
n=0
n!
for all x R.
for all x R.
(iii)
X
1
=
xn for all x with x < 1.
1x
n=0
(iv) tan
x=
X
(1)n x2n+1
n=0
2n + 1
for all x R.
X (1)n
=
.
4
2n + 1
n=0
70
2.3.5
M.T. Nair
x y = f (x) f (y),
x y = f (x) f (y).
Theorem 2.38 Let f be continuous on [a, b] and differentiable on (a, b). Then
(i) f is increasing iff f 0 (x) 0 for all x (a, b).
(ii) f is decreasing iff f 0 (x) 0 for all x (a, b).
(iii) f is strictly increasing if f 0 (x) > 0 for all x (a, b).
(iv) f is strictly decreasing if f 0 (x) < 0 for all x (a, b).
Proof. (i) Suppose f is increasing and x (a, b). Note that
f (x + h) f (x)
f (x + h) f (x)
= lim
.
+
h0
h
h
h0
f 0 (x) = lim
Since,
f (x+h)f (x)
h
To see the converse and (iii), let x1 , x2 [a, b] with x1 < x2 . Then, by mean
value theorem, theorem there exists (x1 , x2 ) such that
f (x2 ) f (x1 ) = f 0 ()(x2 x1 ).
Hence, if f 0 (x) 0 (respectively, f 0 (x) > 0) for every x (a, b), then f (x1 ) f (x2 )
(respectively, f (x1 ) < f (x2 )). Thus, (i) and (iii) are proved. Similar arguments will
lead to the proof of (ii) and (iv).
Differentiability of functions
71
x < 0.
Hence,
f is strictly increasing on (0, ), and
f is strictly decreasing on (0, ).
2.3.6
x I0 , x < x0
and
f 0 (x) < 0
x I0 , x > x0 ,
x I0 , x < x0
and
f 0 (x) > 0
x I0 , x > x0 ,
Hence, in both the cases, we have f (x) < f (x0 ) so that has maximum at x0 . Thus,
(i) is proved.
Similar arguments will lead to the proof of (ii).
EXAMPLE 2.27 Consider
f (x) = x4 ,
g(x) = 1 x4 ,
x < 1.
Then f 0 (x) = 4x3 is negative for x < 0 and positive for x > 0. Hence, by Theorem
2.39, f has local minimum at 0. Also, g 0 (x) = 4x3 is positive for x < 0 and
negative for x > 0. Hence, by Theorem 2.39, g has local maximum at 0.
72
M.T. Nair
Remark 2.12 The conditions given in Theorem 2.39 are cannot be dropped. For
example, consider f (x) = x3 , x R. Then we see that f 0 (x) = 3x2 > 0 for positive
x and negative x. Note that f does not have extremum at 0.
Theorem 2.40 (Another sufficient condition) Suppose f is defined on an interval I and x0 is an interior point of I. Further, suppose that f continuously twice
differentiable in a neighbourhood of x0 and f 0 (x0 ) = 0. Then we have the following:
(i) If f 00 (x0 ) < 0, then f has local maximum at x0 .
(ii) If f 00 (x0 ) > 0, then f has local minimum at x0 .
Proof. By Taylors theorem, there exists an open interval I0 containing x0 such
that for every x I0 , there exists x between x0 and x such that
f (x) f (x0 ) = f 0 (x0 )(x x0 ) +
f 00 (x )
f 00 (x )
(x x0 )2 =
(x x0 )2 .
2
2
()
f 00 (x0 )
.
2
f 00 (x )
(x x0 )2 < 0
2
x I1 .
g(x) = 1 x4 ,
x < 1.
Then f 0 (0) = 0 = g 0 (0), f has local minimum at 0 and g has local maximum at 0.
But, f 00 (0) = 0 = g 00 (0).
Remark 2.14 How to identify critical points and extreme points of a function?
1. Suppose f is defined on an open interval I.
(a) Find those points at which either f is not differentiable or f 0 vanish.
These points are the critical points of f .
Differentiability of functions
73
2.3.7
Additional exercises
3
Definite Integral
3.1
Introduction
In school you must have come across the definition of integral of a function f :
[a, b] R between the limits a and b, i.e.,
Z b
f (x)dx
a
as the number g(b) g(a), where g : [a, b] R is such that its derivative is f .
One immediate question one raises would be the following:
Given any function f : [a, b] R, does there exist a differentiable function
g : [a, b] R such that g 0 (x) = f (x)?
Obviously, it is not necessary to have such a function g (See the exercise below).
1, 1 x 0,
. Show that there
Exercise 3.1 Consider the function f (x) =
1, 0 x 1
does not exist a differentiable function g : [1, 1] R such that g 0 (x) = f (x) for all
x (0, 1).
Another point one recalls from school is that if g : [a, b] R is a differentiable
function, then g 0 (x) has a geometric meaning, namely, it represents the slope of the
tangent to the graph of g at the point x.
Rb
Do we have a geometric meaning to the integral a f (t)dt?
We answer both the above questions affirmatively for a certain class of functions
by giving a geometric definition of the concept of integral.
Suppose f : [a, b] R is a bounded function. Our attempt is to associate a
number to such a function such that, in case f (x) 0 for x [a, b], then is the
area of the region, say Rf , bounded by the graph of f , xaxis, and the ordinates at
a and b. We may not succeed to do this for all bounded functions f .
Suppose, for a moment, f (x) 0 for all x [a, b]. Let us agree that we have
some idea about what the area under the graph of f . Then it is clear that
m(b a) area(Rf ) M (b a),
74
(3.1.1)
75
where m = inf x[a,b] f (x) and M = supx[a,b] f (x). Thus, we get estimates for
area(Rf ). To get better estimates, let us consider a point c such that a < c < b.
Then we have
m1 f (x) M1 x [a, c];
where
m1 = inf f (x),
x[a,c]
m2 = inf f (x),
x[c,b]
M1 = sup f (x),
M2 = sup f (x).
x[a,c]
x[c,b]
(3.1.2)
Since
m(b a) = m(c a) + m(b c) m1 (c a) + m2 (b c),
M (b a) = M (c a) + M (b c) M1 (c a) + M2 (b c),
we can infer that the estimates in (3.1.2) are better than those in (3.1.1). We can
improve these bounds by taking more and more points in [a, b]. This is the basic
idea of Riemann integration.
3.2
Let f : [a, b] R be a bounded function, and let P be a partition of [a, b], i.e., a
finite set P = {x0 , x1 , x2 , . . . , xk } of points in [a, b] such that
a = x0 < x1 < x2 < . . . < xk = b.
We shall denote such partitions by P = {xi }ki=0 also.
Corresponding to the partition P = {xi }ki=0 , and the function f , we associate
two numbers:
L(P, f ) :=
k
X
mi xi ,
U (P, f ) :=
i=1
k
X
Mi xi ,
i=1
Definition 3.1 The quantities L(P, f ) and U (P, f ) are called lower sum and
upper sum, respectively, of the function f associated with the partition P .
76
Definite Integral
M.T. Nair
Note that if f (x) 0 for all x [a, b], then L(P, f ) is the total area of the
rectangles with lengths mi and widths xi , and U (P, f ) is the total area of the
rectangle with lengths Mi and widths xi for i = 1, . . . , k. Thus, it is intuitively
clear that the required area, say , under the graph of f must satisfy the relation:
L(P, f ) U (P, f )
for all partitions P of [a, b].
Throughout this chapter, functions defined on [a, b] are considered to be bounded
functions, and P denotes the set of all partitions of [a, b]. Thus we have
L(P, f ) U (P, f )
P P.
We observe that
m(b a) L(P, f ) U (P, f ) M (b a)
P P,
exist.
Exercise 3.3 Show that L(P, f ) f f U (Q, f ) for all P, Q P.
Hint: Exercise 3.2.
3.3
P P,
then we say that f is Riemann integrable on [a, b], and this is called the
Riemann integral of f , and it is is denoted by
Z b
f (x) dx.
a
77
Remark 3.1 In the due course, Riemann integral will be simply referred to as
integral
The proof of the following theorem is left as an exercise.
Theorem 3.1 A bounded function f : [a, b] R is integrable if and only if f = f .
Definition 3.3 The quantities f and f are known as lower integral and upper
integral, respectively, and they are usually denoted by
Z b
Z b
f (x)dx, respectively.
f (x)dx and
a
Remark 3.2 Not all functions are integrable! For example, consider f : [a, b] R
defined by
0, x Q,
f (x) =
1, x 6 Q.
For this function we have L(P, f ) = 0 and U (P, f ) = b a for any partition P of
[a, b]. Thus, in this case f = 0, f = b a, and hence, f is not integrable.
Theorem 3.2 Suppose f is integrable on [a, b], and m, M are such that
m f (x) M
Then
Z
m(b a)
x [a, b].
f (x) dx M (b a).
a
In particular, if M0 > 0 is such that f (x) M0 for all x [a, b], then
Z b
f (x) dx M0 (b a).
a
78
Definite Integral
M.T. Nair
EXAMPLE 3.1 Let f (x) = c for all x [a, b], for some c R. Then we have
L(P, f ) = c(b a),
U (P, f ) = c(b a)
Hence
c(b a) = L(P, f ) f f U (P, f ) = c(b a).
Rb
Thus, f is integrable and a f (x)dx = c(b b).
(b a)
,
n
i = 0, 1, . . . , n.
Mi = xi ,
xi =
ba
n
for i = 1, . . . , n.
Hence
n
X
n
X
(b a) (b a)
,
L(Pn , f ) =
xi1 xi =
a + (i 1)
n
n
i=1
i=1
n
n
X
X
(b a) (b a)
U (Pn , f ) =
xi xi =
a+i
.
n
n
i=1
i=1
It is see that
(b a)2
n(n + 1)
+ (b a)2
,
n
2n2
(b a)2 n(n + 1)
U (Pn , f ) = a(b a) +
n2
2
L(Pn , f ) = a(b a)
3.3.1
In Example 3.2, what we have showed is that for a sequence (Pn ) of partitions,
the sequences {U (Pn , f )} and {L(Pn , f )} converge to the same point. We shall see
(Corollary 3.4 below) that this is true in general.
Theorem 3.3 A bounded function f : [a, b] R is Riemann integrable if and only
if for every > 0, there exists a partition P of [a, b] such that
U (P, f ) L(P, f ) < .
79
f < U (P 00 , f ) < f + .
as
Z
f (x) dx
and
U (Pn , f )
f (x) dx
a
as n .
Proof. The proof follows from Theorem 3.3. However, we give details:
Suppose f is integrable. Then, by Theorem 3.3, for each n N, there exists a
partition Pn of [a, b] such that
U (Pn , f ) L(Pn , f ) <
1
.
n
n N,
80
Definite Integral
M.T. Nair
Theorem 3.5 Let f and g be integrable over [a, b]. Then f + g is integrable and
Z
f (x) dx.
f (x) dx +
[f (x) + g(x)]dx =
Proof. By Corollary 3.4, there exists sequences of partitions (Pn0 ) and (Pn00 ) of
[a, b] such that
U (Pn0 , f ) L(Pn0 , f ) 0,
U (Pn00 , g) L(Pn00 , g) 0
Z
L(Pn , f )
L(Pn00 , f ) L(Pn , f )
Hence,
U (Pn , f ) L(Pn , f ) 0,
U (Pn , g) L(Pn , g) 0
()
cxd
cxd
cxd
cxd
Hence,
L(Pn , f ) + L(Pn , g) L(Pn , f + g) U (Pn , f + g) U (Pn , f ) + U (Pn , g).
Hence, from (),
U (Pn , f + g) L(Pn , f + g) 0
as
n .
() implies
Z
an
g(x)dx bn ,
f (x)dx +
a
Z
an
[f (x) + g(x)]dx bn .
a
()
81
Theorem 3.6 Suppose f is integrable on [a, c] and [c, b]. Then f is integrable on
[a, b], and
Z b
Z c
Z b
f (x) dx.
f (x) dx +
f (x) dx =
c
Proof. Let f1 = f [a,c] , f2 = f [c,b] . Let > 0 be given. Since f1 and f2 are
integrable, there exist partitions P1 and P2 of [a, c] and [c, b] respectively such that
Hence,
U (P, f ) L(P, f ) = [U (P1 , f1 ) L(P1 , f1 )] + [U (P2 , f2 ) L(P2 , f2 )] < .
Thus f is integrable. Since
Z
L(P, f )
f (x)dx U (P, f ),
a
Z
L(P1 , f1 ) + L(P2 , f2 )
f (x)dx +
a
it follows that
Z
c
Z c
Z b
Z b
f (x) dx +
f (x) dx
f (x) dx < .
a
Z
f (x) dx
g(x) dx.
a
Rb
Proof. Since L(P, f ) a f (x)dx for every partition P on [a, b], and since
Rb
L(P, f ) 0 by the the assumption on f , we obtain a f (x)dx 0. The general
case follows from the above by applying the above result for the function defined
by (x) = g(x) f (x), x [a, b].
Corollary 3.8 Suppose f is integrable on [a, b]. Then
Z b
Z b
f (x)dx
f (x)dx.
a
82
Definite Integral
M.T. Nair
k
X
f (ti )xi
i=1
is called the Riemann sum for f corresponding to the partition P and the set T
of tags on P .
We may observe that for any partition P and for any set T of tags on P ,
L(P, f ) S(P, f, T ) U (P, f ).
This observation together with Corollary 3.4 gives the following result.
Theorem 3.9 If (Pn ) is a sequence of partitions of [a, b] such that
U (Pn , f ) L(Pn , f ) 0
as
n ,
f (x) dx,
a
Tn00 = {t00i,n : i = 1, . . . , k}
()
83
t00i,n Mi
as
n .
Then
S(P, f, Tn0 ) L(P, f ),
as
n .
n N.
3.4
84
Definite Integral
M.T. Nair
1
,
x
0 < x < 1.
Clearly, f is continuous on I := (0, 1). Now, let > 0 be given. For x, y (0, 1),
f (x) f (y) <
x y
< x y < xy.
xy
Thus, for f to be uniformly continuous, there must exist a > 0 such that
xy, which is not possible. In particular, we are not in a position to choose a
independent of the points x, y.
To see this fact, more clearly, consider xn = 1/n and yn = 1/(n + 1). Then we
know that xn yn  0 as n and f (xn ) f (yn ) = 1 for all n N. Hence,
the condition in the definition of uniform continuity is not satisfied if we take < 1.
The above kind of situation will not arise if the interval I is closed and bounded.
More specifically, we have the following theorem.
Theorem 3.12 Every real valued continuous function defined on a closed and
bounded interval is uniformly continuous.
Proof. Suppose f : [a, b] R is continuous. We have to show that for every
> 0, there exists > 0 such that
x y < = f (x) f (y) < .
Suppose this is not true. Then there exists 0 > 0 such that for every > 0, there
are x, y such that
x y < but f (x) f (y) 0 .
()
Taking = 1/n, there exists xn , yn such that
xn yn  < 1/n
k
X
i=1
85
Since f is continuous on the closed interval [a, b], there exists i , i in [xi1 , xi ] such
that Mi = f (i ), mi = f (i ) for i = 1, . . . , k. Hence,
k
X
U (P, f ) L(P, f ) =
[f (i ) f (i )](xi xi1 ).
i=1
Again, since f is uniformly continuous on [a, b], there exists > 0 such that
f (t) f (s) < /(b a)
whenever t s < .
k
X
[f (i ) f (i )](xi xi1 ) < .
i=1
Therefore, by Theorem 3.3, f is integrable, and the proof of (i) is also over.
Next, suppose (Pn ) is a sequence of partitions such that (Pn ) 0 as n .
Let N N be such that (Pn ) < for all n N . Then, it follows by (i) above that
U (Pn , f ) L(Pn , f ) < for all n N , showing that U (Pn , f ) L(Pn , f ) 0 as
n . Now, the concludions in (ii) follows from the observations
Z
L(Pn , f )
f (x)dx U (Pn , f ),
a
f (x)dx.
a
86
Definite Integral
M.T. Nair
EXAMPLE 3.3 Let f (x) = ex for all x [a, b]. Then, f is continuous. Let
(b a)
,
n
hn =
xi = a + ihn ,
ti = xi1 ,
i = 1, . . . , n.
Then with Pn = {xi }ni=1 and T = {ti }ni=1 , we have (Pn ) 0, and
S(Pn , f, Tn ) = hn
n
X
ea+(i1)hn = hn ea
i=1
n
X
n(i1) = hn ea
i=1
nn 1
,
n 1
ehn 1
hn
nn 1
hn
hn
= [eb ea ] hn
.
= ea [eba 1] hn
n 1
e 1
e 1
3.5
Some Properties
For continuous functions f and g, we can give a simpler proof for Theorem 3.5, as
in the following.
Theorem 3.13 Suppose f and g are continuous functions on [a, b]. Then
b
Z
a
[f (x) + g(x)] dx =
f (x) dx +
g(x) dx
a
a
Z b
Z b
c f (x) dx = c
f (x) dx c R.
a
Z
S(Pn , f + g, Tn )
Now, since
S(Pn , f + g, Tn ) = S(Pn , f, Tn ) + S(Pn , g, Tn ),
S(Pn , c f, Tn ) = c S(Pn , f, Tn )
c R,
Some Properties
87
it follows that
Z
a
b
g(x) dx,
f (x) dx +
[f (x) + g(x)] dx =
Z
f (x) dx
c f (x) dx = c
c R.
Since g(x0 ) 6= 0 for some x0 [a, b] and g(x) 0 for all x [a, b], it follows that
Rb
a g(x)dx > 0. Hence,
Rb
f (x)g(x) dx
m aRb
M.
a g(x)dx
Therefore, by the intermediate value theorem, there exists [a, b] such that
Rb
f (x)g(x) dx
f () = a R b
.
g(x)dx
a
Hence the result.
88
Definite Integral
3.6
Some Results
3.6.1
M.T. Nair
Theorem 3.16 Suppose f is continuous on [a, b], and for x [a, b], let
Z x
f (t)dt.
(x) =
a
x (a, b).
Proof. Let x (a, b) and h > 0 be such that x + h [a, b]. Then we have
x+h
Z
(x + h) (x) =
Z
f (t)dt
Z
f (t)dt =
x+h
f (t)dt.
x
(x + h) (x)
= f (x).
h
lim
(x + h) (x)
= f (x).
h
h0+
Similarly, we have
h0
3.6.2
Some Results
89
all x [a, b] and for some constant c R. Hence, in view of the above theorem, we
have
Z b
f (t)dt.
g(b) g(a) = (b) (a) =
a
n
n
X
X
g(b) g(a) =
[g(xi ) g(xi1 )] =
f (i )(xi xi1 ).
j=1
j=1
Pn
Since the Riemann sum j=1 f (i )(xi xi1 ) is constant for any partition P , it
Rb
P
follows that nj=1 f (i )(xi xi1 ) = a f (x)dx. Thus,
Z
90
Definite Integral
M.T. Nair
k
X
f (i )(xi xi1 ) =
i=1
k Z
X
i=1
xi
f (x) dx.
f (x) dx =
xi1
3.6.3
Z
a
f (x)g(x) dx = [f (x)G(x)]ba
f 0 (x)G(x) dx.
[u(x)v(x)]0 dx =
u0 (x)v(x) dx +
Z
=
u (x)v(x) dx +
a
Thus,
Z
u0 (x)v(x) dx.
Now, taking f (x) = u(x) and v(x) = G(x), we obtain the required formula.
Theorem 3.20 (Change of variable formula) Suppose : [, ] R is a
differentiable function such that () = a and () = b. Then, for any continuous
function f : [a, b] R,
Z
f (x) dx =
a
f ((t)) 0 (t)dt.
Some Results
91
Proof. Let F be an antiderivative of f , i.e., such that F 0 (x)f (x). Then taking
G(t) = F ((t)) for t [, ], we have
G0 (t) = F 0 ((t)) 0 (t) = f ((t)) 0 (t),
t [, ].
f ((t)) (t)dt =
Hence,
Z
f (x) dx.
a
xk+1
x dx =
k+1
b
=
a
bk+1 ak+1
.
k+1
e
a
ex
dx =
b
=
a
eb ea
.
k+1
92
Definite Integral
3.7
Appendix
M.T. Nair
Assume that > 0 is so small that (M m) < /6. Then for any partition P3 on
[c , c + ], we have
4
Improper Integrals
Recall that we defined definite integral of a function f for the case when f is a
bounded function defined on a closed interval [a, b]. In case the assumptions on f
are not satisfied, then can we still have a notion of integral? We discuss a few such
cases.
4.1
Definitions
Rt
Definition 4.1 Suppose f is defined on [a, ). If (t) := a f (x) dx exists for
every t > a, and if limt (t) exists, then we define the improper integral of f
over [a, ) as
Z t
Z
f (x) dx = lim
f (x) dx.
t a
Rb
Definition 4.2. Suppose f is defined on (, b]. If (t) := t f (x) dx exists for
every t < b, and if limt (t) exists, then we define the improper integral of
f over (, b] as
Z b
Z b
f (x) dx = lim
f (x) dx.
t t
Rc
Definition 4.3. Suppose f is defined on R := (, ). If f (x) dx and
R
c f (x) dx exists for some c R, then we define the improper integral of f
over (, ) as
Z
Z c
Z
f (x) dx =
f (x) dx +
f (x) dx.
Rt
We may observe that existence of limt t f (x) dx does not, in general, imply
R
existence of f (x) dx. To see this, consider the following example:
Rt
Let f (x) = x for every x R. Then we have t f (x) dx = 0 for every t R,
Rc
R
but the integrals f (x) dx and c f (x) dx do not exist.
93
94
Improper Integrals
M.T. Nair
Next we consider the case when f is defined on an interval J of finite length, but
limxx0 f (x) = , where x0 either belongs to J or it is an end point of J.
Rb
Definition 4.4. Suppose f is defined on (a, b]. If t f (x) dx exists for every
Rb
t (a, b), and if lim0 a+ f (x) dx exists, then we define the improper integral
of f over (a, b] as
Z b
Z b
f (x) dx.
f (x) dx = lim
0 a+
Rt
Definition 4.5. Suppose f is defined on [a, b). If a f (x) dx exists for every
R b
t (a, b), and if lim0 a f (x) dx exists, then we define the improper integral
of f over [a, b) as
Z b
Z b
f (x) dx = lim
f (x) dx.
0 a
Rc
Definition 4.6. Suppose f is defined on [a, c) and (c, b]. If a f (x) dx and
Rb
c f (x) dx exist, then we define the improper integral of f over [a, b] as
Z c
Z b
Z b
f (x) dx =
f (x) dx +
f (x) dx.
a
Now we combine the situations in Definitions 4.1, 4.2 with Definitions 4.4, 4.5,
to consider improper integrals over intervals of the form (a, ) and (, b).
Rt
Definition
4.7.
Suppose
f
is
defined
on
(a,
).
If
the
integrals
a f (x) dx and
R
f
(x)
dx
exist
as
improper
integrals
for
every
t
>
a,
then
we
define
the
improper
t
integral of f over (a, ) as
Z
Z t
Z
f (x) dx =
f (x) dx +
f (x) dx.
a
Rt
Definition. Suppose f is defined on (, b). If the integrals f (x) dx and
Rb
t f (x) dx exist as improper integrals for every t < b, then we define the improper
integral of f over (, b) as
Z b
Z t
Z b
f (x) dx =
f (x) dx +
f (x) dx.
In case an improper integral exists (resp. does not exists), then we also say that
the improper integral converges (resp. diverges).
Definitions
4.1.1
Examples
Z
1
dx. Note that
x
1
dx = [ln x]t1 = ln t as t .
x
Hence,
R
1
1
x
dx diverges.
Z
1
dx Note that
x2
Z
1
Hence,
95
R
1
1
x2
1
1
dx =
2
x
x
t
=1
1
1
1
t
as
t .
dx converges.
Z
case, we have
t
Z
1
1
dx. In this
xp
p+1 t
1
x
tp+1 1
dx
=
=
.
xp
p + 1 1
p + 1
Note that,
p>1
1
tp+1 1
p + 1
p1
and
p<1
as
tp+1 1
as
p + 1
t ,
t ,
Hence,
Z
1
1
dx
xp
we have
Z
1
xp+1
dx
=
xp
p + 1
1
=
1
dx. In this case,
xp
1 p+1
.
p + 1
Note that,
p>1
and
p<1
p+1 1
as
p + 1
p+1 1
1
p + 1
1p
0,
as 0,
96
Improper Integrals
M.T. Nair
Hence,
Z
1
dx
xp
dx
=
(b x)
Rb
dx
a (bx)
ba
bt
converges:
du
.
u
Now,
Z
lim
tb a
dx
exists
(b x)
ba
lim
du
tb bt u
Z ba
lim
exists
du
exists
u
< 1.
Rb
Exercise 4.1 Suppose f 0 and the integral a f (x)dx exists for every x [a, b)
Rb
and limxb (b x) f (x) exists for some < 1. Then a f (x)dx exists.
[Hint: Observe that for any > 0, there exists x0 [a, b) such that the number
+
:= limxb (b x) f (x) satisfies 0 f (x) (bx)
for all x [x0 , b).]
4.2
Integrability by Comparison
We state a result which will be useful in asserting the existence of certain improper
integral by comparing it with certain other improper integral.
Suppose J is either an interval of finite or infinite length. Suppose f defined on
J, except possibly at one point (It can be generalized to the case when f is not be
defined at some finite number of point in J). We denote the improper integral of f
over J by
Z
f (x) dx,
J
ta t
Z
f (x) dx,
lim
tb a
Z
f (x) dx,
lim
tc
Z
f (x) dx + lim
tc+
f (x) dx.
t
Integrability by Comparison
97
4.2.1
R
J
R
J
R
J
f (x) dx
f (x) dx exists.
Examples
cos x
1
p p
x
x
it follows from Example 4.3 and Theorem ??(ii) that the improper integrals
Z
Z
sin x
cos x
dx
and
dx
p
x
xp
1
1
converge for all p > 1.
R x
R
In fact 1 sin
xp dx and 1
example.
cos x
xp
x
see that 1 cos
dx
converges
for
all
p
> 0.
xp
EXAMPLE 4.8 Since
sin x sin x 1
1
=
xp x xp1 xp1 ,
cos x
1
p p
x
x
98
Improper Integrals
M.T. Nair
R1
1
0 xp1
x (0, 1].
0 < p < 2.
4.2.2
Gamma and Beta Functions are certain improper integrals which appear in many
applications.
Gamma function
We show that for x > 0, the improper integral
Z
(x) :=
tx1 et dt
0
tx1 et
R
Also, we observe that
0 as t , and 1 t2 dt converges. Hence, by
2
t
R
Theorem 4.2, 1 tx1 et dt converges. Thus,
Z
Z 1
Z
x1 t
x1 t
(x) :=
t e dt =
t e dt +
tx1 et dt
0
99
converges. The function (x, y) for x > 0, y > 0 is called the beta function.
Clearly, the above integral is proper for x 1, y 1. Hence it is enough to
consider the case of 0 < x < 1, 0 < y < 1. In this case both the points t = 0 and
t = 1 are problematic. hence, we consider the integrals
Z 1
Z 1/2
x1
y1
tx1 (1 t)y1 dt.
t (1 t)
dt,
1/2
We note that if 0 < t 1/2, then (1t)y1 21y so that tx1 (1t)y1 21y tx1 .
R 1/2
R 1/2
Since 0 tx1 dt converges it follows that 0 tx1 (1 t)y1 dt converges. To deal
with the second integral, consider the change of variable u = 1 t. Then
Z 1/2
Z 1
x1
y1
uy1 (1 u)x1 du
t (1 t)
dt =
0
1/2
x > 0, y > 0
4.3
100
Improper Integrals
M.T. Nair
R
Thus, f (x) < g(x) for all x x0 . Hence, convergence of x0 g(x)dx implies the
R
convergence of x0 f (x)dx. From this the result in (ii) follows.
4.4
Additional Exercises
Z
sin
1. Does
1
1
x2
dx converge?
[ Hint: Note that sin x12 x12 .]
Z
cos x
dx converge?
2. Does
x(log x)2
2
cos x
[Hint: Observe x(log
x(log1 x)2 and use the change of variable t = log x.]
x)2
Z
3. Does
0
sin2 x
dx converge?
x2
2
5. Does
f (x)dx 0 as a, b .
sin2 x
,
x2
Ra
Rb
Rb
[Hint: Note that a f (x)dx = a0 f (x)dx a0 f (x)dx 0 as a, b .]
Z
2
6. Does
ex dx converge?
0
2
1
x2
for 1 x .]
[Hint: Use the change of variable t = log x, and the fact that
diverges.]
Z 1
8. Does
ln xdx converge?
0
log 2 sin t dt
5
Geometric and Mechanical
Applications of Integrals
5.1
5.1.1
Computing Area
Using Cartesian Coordinates
y = f (x),
where f : [a, b] R is a continuous function such that f (x) 0 for all x [a, b].
Then, the area under the curve, i.e., the area of the region bounded by the graph
of f , the xaxis, and the ordinates at x = a and x = b, is
Z b
k
X
lim
f (i )xi =
y dx.
(P )0
j=1
y = (t),
, t ,
such that a = (), b = (). Then the area under the curve takes the form
Z
(t)0 (t)dt.
If f takes both positive and negative values, but changes sign only at a finite
number of points, then the area bounded by the curve, the xaxis, and the ordinates
at x = a and x = b, is given by
Z b
f (x) dx.
a
j=1
101
102
5.1.2
M.T. Nair
= (),
lim
(P )0
5.1.3
j=1
[(i )i ](i ) =
k
X
1
lim
(P )0
j=1
[(i )]2 i =
1
2
2 d.
Some Examples
y = x,
y = x2 , x 0 :
Note that the points of intersection of the curves are at x = 0 and x = 1. Also,
"
xx
2
x3/2 x3
dx =
3/2
3
#1
0
1
= .
3
y = b sin t,
0 t 2 :
y dx = 4
0
/2
/2
(1 cos 2t dt = ab.
EXAMPLE 5.3 We find the area bounded by one arch of the cycloid
x = a(t sin t),
One arch of the cycloid is obtained by varying t over the interval [0, 2]. Thus, the
required area is
Z 2a
Z 2
Z 2
0
y dx =
y(t)x (t) dt =
a2 (1 cos t)2 dt = 3a2 .
0
103
2 d = a2 .
= a cos 2.
The required area is
#
" Z
Z /4
1 /4 2
cos 2d = a2 .
d = a2
2
2 /4
/4
5.2
5.2.1
y = f (x),
lim
(P )0
k p
X
(xi xi1 )2 + (yi yi1 )2 ,
j=1
where
yi yi1 = f (xi ) f (xi1 ) = f 0 (i )xi ,
for some i [xi1 , xi ], i = 1, . . . , k. Hence,
A :=
lim
(P )0
lim
(P )0
lim
(P )0
Z
=
k p
X
(xi xi1 )2 + (yi yi1 )2
j=1
k p
X
(xi )2 + [f 0 (i )xi ]2
j=1
k p
X
1 + [f 0 (i )]2 xi
j=1
s
1+
dy
dx
2
dx.
104
M.T. Nair
1+
a
t ,
y = (t),
dy
dx
2
s
dx =
d
dt
2
+
d
dt
2
dt.
Remark 5.1 Curves in parametric form are assumed to be piecewise smooth, i.e.,
having unique tangents except possible at a finite number of points. Note that
if a curve is given in parametric form as x = (t), y = (t) with t b,
then it has unique tangent at (x0 , y0 ) if 0 (t0 ), 0 (t0 ) exists, where t0 is such that
(x0 = (t0 ), y0 = (t0 ), and 0 (t0 )2 +  0 (t0 )2 6= 0.
5.2.2
= (),
y = sin ,
s
A=
dx
d
2
+
dy
d
2
d.
Note that
dx
dy
= 0 cos + ( sin ),
= 0 sin + cos .
d
d
Hence, it follows that
s
Z 2 2
dx
dy
A =
+
d
d
d
Z p
=
2 + 02 d.
5.2.3
Examples
105
Thus,
a
L := 4a
0
dx
= 2a.
x2
a2
0 2.
y = a sin ,
/2
s
L := 4
0
Z
=
dx
d
2
+
dy
d
2
d
/2 p
y = b sin ,
0 2.
/2
s
L := 4
0
Z
=
dx
d
2
+
dy
d
2
d
/2 p
a2 sin2 + b2 cos2 d
Z
=
/2 p
a2 (1 cos2 ) + b2 cos2 d
Z
=
/2 p
a2 (a2 b2 ) cos2 d
Z
=
4a
/2 p
1 2 cos2 d,
106
M.T. Nair
s
dx
d
L := 4
0
+
dy
d
2
d
/2 p
Z
=
2
4
0
/2 p
Z
=
12a
0
2 = a2 (1 + cos )2 ,
we have
L=
Z
2a
1 + cos d = 4a
5.3
Z
0
cos d = 8a.
2
Suppose that a three dimensional object, a solid, lies between two parallel planes
x = a and x = b. Let (x) be the area of the cross section of the solid at the point x,
with cross section being parallel to the yzplane. We assume that the function (x),
x [a, b] is continuous. Now, consider a partition P : a = x0 < x1 < . . . < xk = b
of the interval [a, b]. Then the volume of the solid is given by
lim
(P )0
k
X
Z
(i )xi =
(x) dx.
a
j=1
EXAMPLE 5.11 Let us compute the volume of the solid enclosed by the ellipsoid
x2 y 2 z 2
+ 2 + 2 = 1.
a2
b
c
For a fixed x [a, a], the boundary of the cross section at x is given by the equation
y2 z2
x2
+
=
1
,
b2
c2
a2
i.e.,
y2
z2
+
= 1,
(x)2 (x)2
where
x2
(x) = b 1 2 ,
a
r
(x) = c 1
x2
.
a2
107
Hence,
x2
(x) = (x)(x) = bc 1 2 ,
a
and the required volume is
Z
x2
4
1 2 dx = abc.
a
3
a
Z
(x) dx = bc
V :=
5.4
a x b.
Z
V :=
(x) dx =
a
y 2 dx.
EXAMPLE 5.12 Let us compute the volume of the solid of revolution of the curve
y = x2 about xaxis for a x a. The required volume is
Z
a
2
V :=
y dx =
a
2
x4 dx = a5 .
5
a
EXAMPLE 5.13 We compute the volume of the solid of revolution of the catenary
y=
a x/a
e
+ ex/a
2
V :=
y dx =
0
Z
2
a2 b 2x/a
a2 x/a
x/a
e
+e
dx =
e
+ e2x/a + 2 .
4
4 0
We see that
V :=
b3
a3 2b/a
e
e2b/a +
.
8
8
108
5.5
M.T. Nair
A :=
lim
(P )0
k
X
2f (i )si ,
j=1
where P : a = x0 < x1 < . . . < xk = b is a partition of the interval [a, b], and
p
1 + [f 0 (i )]2 xi ,
si :=
i = 1, . . . , k.
Thus
A=
lim
(P )0
k
X
Z b
p
y
2f (i ) 1 + [f 0 (i )]2 xi = 2
s
1+
j=1
dy
dx
2
dx.
1+
dy
dx
2
dx
r
Z ap
p
= 2
2px 1 +
dx = 2 p
p + 2x dx
2x
0
0
a
i
2 p h
2
3/2 1
= 2 p (2x + p)
=
(2a + p)3/2 p3/2 .
3
2 0
3
Z
5.6
ap
Centre of Gravity
Suppose A1 , A2 , . . . , An are material particles on the plane at coordinates (x1 , y1 ), (x2 , y2 ), . . . , (xn , yn )
and masses m1 , m2 , . . . mn respectively. Then the centre of gravity of the system
of these particles is at the point A = (xC , yC ), where
Pn
xi mi
xC := Pi=1
,
n
i=1 mi
Pn
yi mi
yC := Pi=1
.
n
i=1 mi
Now we attempt to define the centre of gravity of a material line and material
planar region enclosed by certain curves.
Centre of Gravity
5.6.1
109
r0
M (X, r)
.
r
s
(P )0
(P
)0
i
i=1 i
i=1 i si
Assuming that the function (X) := (x, f (x)) is continuous on [a, b], we see that
r
r
2
2
Rb
Rb
dy
dy
x(x,
y)
1
+
y(x,
y)
1
+
dx
dx
dx
dx
a
a
r
r
xC =
,
y
=
.
C
2
2
Rb
Rb
dy
dy
dx
dx
a (x, y) 1 + dx
a (x, y) 1 + dx
2
2
f (x) := a x , so that it follows that
s
2
dy
a
1+
=
.
2
dx
a x2
Hence, since (x, y) is constant,
xC = 0
2
dy
y
1
+
dx
dx
a
2a
r
yC =
=
.
2
Ra
dy
1 + dx dx
a
Ra
110
5.6.2
M.T. Nair
y = g(x),
with
f (x) g(x)
a x b.
Suppose that the density of the material at the point X is (X). This density is
defined as follows: Suppose M (X, r) is the mass of the circular region S(X, r)
with centre at x and radius r > 0, and (X, r) is the area of the same circular
region. Then the density of the material at the point x is defined by
M (X, r)
.
r0 (X, r)
(X) := lim
Now, in order to find the centre of gravity of , we first look at the following special
case: Suppose is a rectangle given by a1 x b1 , a2 y b2 . Then we can
infer that the centre of gravity of such rectangle is located at the point
a1 + b1 a2 + b2
,
.
2
2
Taking the above obervation into account, we consider a partition P : x0 < x1 <
. . . < xk of the interval [a, b], and consider the rectangular strips:
Ri :
xi1 x xi ,
f (i ) y g(i ),
i = 1, . . . , k,
i = 1, . . . , k.
Assuming that the mass of the rectangular strip Ri is concentrated at its midpoint:
f (i ) + g(i )
Xi : i ,
,
2
we consider the centre of gravity of the system of material points at Xi as
Pn f (i )+g(i )
Pn
mi
i mi
i=1
xC,P := Pn
,
yC,P := i=1Pn 2
.
i=1 mi
i=1 mi
Now the centre of gravity of is defined as
xC =
lim xC,P ,
(P )0
yC =
lim yC,P ,
(P )0
i.e.,
xC
Pn
i [g(i ) f (i )]xi
=
lim Pi=1
n
(P )0
i=1 [g(i ) f (i )]xi
Rb
x[g(x) f (x)] dx
= Ra b
a [g(x) f (x)] dx
Moment of Inertia
111
Pn
yC
1
i=1 2 [f (i + g(i )][g(i ) f (i )]xi
Pn
i=1 [g(i ) f (i )]xi
lim
(P )0
1
2
Rb
+ g(x)][g(x) f (x)] dx
Rb
a [g(x) f (x)] dx
a [f (x
f (x) = a x,
g(x) =
a x,
0 x a.
x[g(x) f (x)] dx
5.7
Moment of Inertia
Suppose there are n material points in the plane. Let their masses be m1 , m2 , . . . mn
respectively. Suppose that these points are at distances d1 , . . . , dn from a fixed point
O. Then the moment of inertia of the system of these points with respect to the
point O is defined by the quantity:
IO :=
n
X
d2i mi .
i=1
If O is the origin, and (x1 , y1 ), (x2 , y2 ), . . . , (xn , yn ) are the points, then
IO :=
n
X
i=1
5.7.1
112
M.T. Nair
n
X
(i2 + i2 )mi .
i=1
Thus,
n
X
IO,P :=
(i2 + i2 )i si .
i=1
Here, si is the length of the arcs joining (xi1 , yi1 ) to (xi , yi ), and i is the
density at the point (i , i ). Note that i si is the approximate mass of the arc
joining (xi1 , f (xi1 ) to (xi , f (xi ). Now, assuming that the functions f (x) and
(x) := (x, f (x)) are continuous on [a, b], the moment of inertial of L with respect
to O is
IO =
lim IO,P
(P )0
n
X
lim
(P )0
(i2 + i2 )i si
i=1
b
2
(x + y )(x, y)
1+
5.7.2
dy
dx
2
dx.
lim
(P )0
n
X
d2i mi ,
i=1
n
X
lim
(P )0
a2 i [ai ] = a3
()d.
i=1
()d = ( )a3 .
Additional Exercises
5.7.3
113
Q:
i1 i i .
Assume that the the mass of this region Rij is concentrated at the point (
j , i ),
n
X
mij d2ij
lim
(Q)0
j=1
(Q)0
j=1
n
X
lim
Z
n
X
[
j i j ] 2j
3j j i
j=1
d i
3
a4
i .
4
m
X
a4
i=1
i =
( )a4
.
4
5.8
Additional Exercises
1. Find the area of the portion of the circle x2 + y 2 = 1 which lies inside the
parabola y 2 = 1 x.
114
M.T. Nair
[Hind: Area enclosed by the circle in the second and third quadrant and
the area enclosed by the
parabola in the first and fourth quadrant. The the
R1
required area is 2 + 2 0 1 x dx.
Ans: 2 + 43 . ]
2. Find the area common to the cardioid = a(1 + cos ) and the circle =
3a
2 .
7
4
9 3
8 .
3. For a, b > 0, find the area included betwee the parabolas y 2 = 4a(x + a) and
y 2 = 4b(b x).
[Hind: Points of intersection of the curves is given byh a(x + a) = b(b x), i.e.,
i
Rb p
R ba p
2 a2
= ba; y = 2 ab. The required area is 2 a
4a(x + a) + ba 4b(b x) dx .
x = b a+b
3
where r1 = a cos , r2 = a(1 cos ). Ans: a3 (3 3 ) ]
7. Find the area of the loop of the curve x = a(1 t2 ), y = at(1 t2 ) for
1 t 1.
[Hint: y = 0 for t {1, 0, 1}, and y negative for 1 t 0 and positive for
0 t 1. Also, Ry 2 = x2 (a
R 0x)/a so that the curve is symmetric w.r.t. the
a
xaxis. Area is 2 0 ydx = 2 1 y(t)x0 (t)dt. Ans: 8a2 /15 ]
8. Find the length of an arch of the cycloid x = a(t sin t), y = a(1 cos t).
[Hint: The curve cuts the xaxisR at p
x = a and x = 2a for t = 0 and t = 2
2
respectively. Thus the length is 0
[x0 (t)]2 + [y 0 (t)]2 dt. Ans: 8a. ]
Additional Exercises
115
9. For a > 0, find the length of the loop of the curve 3a y 2 = x(x a)2 .
[Hint: The curve cuts the xaxis at x = a,rand the curve is symmetric w.r.t.
2
Ra
dy
the xaxis. Thus the required area is 2 0 1 + dx
dx. Note that 6ayy 0 =
(x a)(3x a), so that 1 + y 02 =
10. Find the length of the curve r =
[Hind: ` :=
(3x+a)2
12ax .
2
1+cos ,
Ans:
4a
.
3
0 /2.
R /4
R /2 p
r2 + [r0 ]2 d = 2 0 sec3 d. Ans: 2 + ln( 2 + 1). ]
0
11. Find the volume of the solid obtained by revolving the curve y = 4 sin 2x,
0 x /2, about yaxis.
[Hint: writing y = 4 sin 2x, 0 x /4 and y = 4 sin 2u, /4 u /2, the
R4
R /2
R /4
required volume is 0 (u2 x2 )dy = /4 u2 (8 cos 2u)du 0 x2 (8 cos 2x)dx.
Also, note that the curve is symmetric w.r.t. the line x = /4. Hence, the
R /4
required volume is given by 0 [( 4 x)2 x2 ]dy. Ans: 2 2 .]
12. Find the area of the surface obtained by revolving a loop of the curve 9ax2 =
y(3a y)2 about yaxis.
r
2
R 3a
[Hind: x = 0 iff y = 0 or y = 3a. The required area is 2 0 x 1 + dx
dx.
dy
Ans: 3a2 . ]
13. Find the area of the surface obtained by revolving about xaxis, an arc of the
catenary y = c cosh(x/c) between x = a and x = a for a > 0.
Ra
Ra p
[Hind: The area is 2 a y 1 + y 02 dx = 2 c a cosh2 xc dx. Ans: c [2a +
c sinh 2a
c ]. ]
14. The lemniscate 2 = a2 cos 2 revolves about the line = 4 . Find the area of
the surface of the solid generated.
R /4 p
[Hind: The required surface is 22 /4 h 2 + 02 d, where h := sin
2
= a cos 3 so that 2 + 02 = cosa 2 . Ans: 4a2 . ]
,
15. Find the volume of the solid generated by the cardioid = a(1 + cos ) about
the initial line. [Ans: 83 . ]
6
Sequence and Series of Functions
6.1
6.1.1
Sequence of Functions
Pointwise Convergence and Uniform Convergence
Let J be an interval in R and let F(J) be the set of all real valued functions defined
on F .
Definition 6.1 By a sequence of functions on J we mean a map F : N F(J).
If fn := F (n) for n N, then we denote F by (fn ), and call (fn ) as a sequence of
functions.
Definition 6.2 (a) We say that the sequence (fn ) of functions on J converges
pointwise on J if for each x J, the sequence (fn (x)) of real numbers converges.
(b) Suppose (fn ) converges pointwise on J. Let f : J R be defined by
f (x) = limn fn (x), x J. Then we say that (fn ) converges to f pointwise
on J, and f is the pointwise limit of (fn ), and in that case we may also write
fn f
pointwise on
J.
Thus, (fn ) converges to f pointwise on J if and only if for every > 0 and for
each x J, there exists N N (depending, in general on both and x) such that
fn (x) f (x) < for all n N .
Exercise 6.1 Pointwise limit of a sequence of functions is unique.
EXAMPLE 6.1 Consider fn : R R defined by
fn (x) =
sin(nx)
,
n
xR
1
n
116
n N.
Sequence of Functions
117
n N
and x J,
uniformly on
J.
118
M.T. Nair
any > 0, there exists N N such that xn  < for all n N and for all x [0, 1).
In particular, we have
xN  <
x [0, 1).
Hence, taking limit as x 1, we would get 1 < , which is not possible if we had
chosen < 1.
Here is another example of a sequence of functions which converges pointwise
but not uniformly.
EXAMPLE 6.3 For each n N, let
fn (x) =
nx
,
1 + n2 x2
x [0, 1].
x [0, 1].
x [0, 1].
x R.
x J \ {0}.
Letting x 0, we have /2 = fN (0) /2 < which is not possible if we had
chooses < /2.
Now, we give a theorem which would help us to show nonuniform convergence
of certain sequence of functions.
Theorem 6.1 Suppose fn and f are functions defined on an interval J. If there
exists a sequence (xn ) in J and c 6= 0 such that an := fn (xn )f (xn ) c as n ,
then (fn ) does not converge uniformly to f on J.
Sequence of Functions
119
c
2
n N,
x J.
In particular,
fn (xn ) f (xn ) <
c
2
n N.
c
2
n N,
2nx
,
1 + n4 x2
x [0, 1].
1
2nx
0 fn (x)
= .
4
2
n
2 n x
120
M.T. Nair
1
log(1 + n4 x2 ),
n3
Then we have
0 fn (x)
Taking g(t) :=
1
t3
x [0, 1].
1
log(1 + n4 ) =: n
n3
n N.
In particular,
1
log(1 + n4 ) = 0.
n n3
Thus, by Theorem 6.2, (fn ) converges uniformly to f = 0.
lim
We may observe that in Examples 6.2 and 6.4, the limit function f is not continuous, although every fn is continuous. This makes us to ask the following:
Suppose each fn is continuous function on J and (fn ) converges to f pointwise.
Under what condition can we assert that f is continuous?
Also, we may want to know the answers to the following questions:
Suppose each fn is continuous on J and (fn ) converges pointwise to f on J.
Do we have
Z b
Z b
f (x)dx = lim
fn (x)dx
a
n a
0 x 1.
Sequence of Functions
121
n+1
n
x(1 x2 )
as
n .
Since x(1 x2 ) < 1 for x (0, 1), the result follows. But,
Z
fn (x)dx =
0
n
1
2n + 2
2
as
n .
sin(nx)
,
n
x R.
fn0 (0) = n as n .
6.1.2
()
Let > 0 be given. Since (fn ) converges to f uniformly, there exists N N such
that
fn (x) fn (x) < /3 n N, x J.
Since fN is continuous, there exists > 0 such that
fN (x) fN (x0 ) < /3
whenever
x x0  < .
122
6.1.3
M.T. Nair
Theorem 6.4 Suppose (fn ) is a sequence of continuous functions defined on an interval [a, b] which converges uniformly to a function f on [a, b]. Then f is continuous
and
Z
Z
b
lim
n a
f (x)dx.
fn (x)dx =
a
f
(x)dx
fn (x) f (x)dx.
n
a
f
(x)dx
n
a
6.1.4
= f 0 (x)
x J.
Proof. Let g(x) := lim fn0 (x) for x J, and := lim fn (a). Since the convern
Rx
0
Let (x)
a g(t)dt, x J. Then is differentiable and (x) = g(x) for x J.
R x :=
But, a fn0 (t)dt = fn (x) fn (a). Hence, we have
Series of Functions
6.2
123
Series of Functions
X
X
fn or
fn (x),
n=1
n=1
n=1 fn (x)
sn (x) :=
n
X
fi (x),
i=1
n=1 fn .
P
on an interval J, and let
Definition 6.6 Consider a series
n=1 fn (x) of functions P
sn (x) be its nth partial sum. Then we say that the series
n=1 fn (x)
(a) converges at a point x0 J if (sn )) converges at x0 ,
(b) converges pointwise on J if (sn ) converges pointwise on J, and
(c) converges uniformly on J if (sn ) converges uniformly on J.
The proof of the following two theorems are obvious from the statements of
Theorems 6.4 and 6.5 respectively.
P
Theorem 6.6 Suppose (fn ) is a sequence of continuous functions on J. If
n=1 fn (x)
converges uniformly on J, say to f (x), then f is continuous on J, and for [a, b] J,
Z
f (x)dx =
a
Z
X
fn (x)dx.
n=1 a
Theorem
P differentiable functions on
P 6.70 Suppose (fn ) is a sequence of continuously
f
(x)
converges
uniformly
on
J,
and
if
J. If
n=1 n
n=1 fn (x) converges at some
P
point x0 J, then n=1 fn (x) converges to a differentiable function on J, and
!
X
d X
fn (x) =
fn0 (x).
dx
n=1
n=1
124
M.T. Nair
i=m+1
i=m+1
Pn
where n =
n=1 n converges, the sequence (n ) is a Cauchy
k=1 k Since
sequence. Now, let > 0 be given, and let N N be such that
n m  < n, m N.
Hence, from the relation: sn (x) sm (x) n m , we have
sn (x) sm (x) < n, m N, x J.
This, in particular implies that {sn (x)} is also a Cauchy sequence at each x J.
Hence, {sn (x)} converges for each x J. Let f (x) = limn sn (x), x J. Then,
we have
f (x) sm (x) = lim sn (x) sm (x) < m N, x J.
n
n=1 fn
converges uniformly to f on J.
n=1
cos nx
1
2,
2
n
n
and
1
n=1 n2
is convergent.
cos nx
n2
and
n=1
sin nx
n2
sin nx
1
n2 n2
n N
P
EXAMPLE 6.10 The series
xn is a dominated series on [, ] for 0 < < 1,
n=0
P
n
n
since x  for all n N and n=0 n is convergent. Thus, the given series is a
dominated series, and ehnce, it is uniformly convergent.
P
x
EXAMPLE 6.11 Consider the series n=1 n(1+nx
2 ) on R. Note that
x
1
1
,
n(1 + nx2 )
n 2 n
X
1
converges. Thus, the given series is dominated series, and hence it
3/2
n
n=1
converges uniformly on R.
and
Series of Functions
125
P
x
EXAMPLE 6.12 Consider the series
n=1 1+n2 x2 for x [c, ), c > 0. Note
that
x
x
1
1
2 2 = 2 2
2
2
1+n x
n x
n x
n c
X
1
and
converges. Thus, the given series is dominated series, and hence it
n2
n=1
converges uniformly on [c, ).
n
P
x
EXAMPLE 6.13 The series
is dominated on [0, ): To see this,
n=1 xe
note that
n
xn
xn
n!
xex = nx
= n
n
e
(nx) /n!
n
P n!
and the series n=1 nn converges.
It can also be seen that xex  1/2 for all x [0, ).
P
n1 is not uniformly convergent on (0, 1); in
EXAMPLE 6.14 The series
n=1 x
particular, not dominated on (0, 1). This is seen as follows: Note that
sn (x) :=
n
X
xk1 =
k=1
1 xn
1
f (x) :=
1x
1x
as n .
n
x
1 x < .
Hence, if there exists N N such that f (x) sn (x) < for all n N for all
x (0, 1), then we would get
xN
< x (0, 1).
1 x
This is not possible, as xN /1 x as x 1.
However, the above series is dominated on [a, a], where 0 < a < 1.
P
EXAMPLE 6.15 The series n=1 (1x)xn1 is not uniformly convergent on [0, 1];
in particular, not dominated on [0, 1]. This is seen as follows: Note that
n
X
1 xn
if x 6= 1
k1
sn (x) :=
(1 x)x
=
0
if x = 1.
k=1
as
n .
126
M.T. Nair
n1
converges.
uniformly for x for any (0, 1). This follows since n=1 n
Hence,
X
1
d 1
=
=
nxn1 for x .
(1 x)2
dx 1 x
n=1
The above relation is true for x in any open interval J (1, 1); because we can
choose sufficiently close to 1 such that J [, ]. Hence, we have
X
1
=
nxn1
(1 x)2
n=1
We know that the given series is not uniformly convergent (see, Example 6.14).
6.3
Additional Exercises
X
x2
for
x
0.
Show
that
the
series
fn (x) does not
(1 + x2 )n
n=1
converge uniformly.
x
, x R. Show that (fn ) converge uniformly, whereas (fn0 )
2. Let fn (x) =
1 + nx2
0
does not converge uniformly. Is the relation lim fn0 (x) = lim fn (x) true
n
n
for all x R?
1. Let fn (x) =
log(1 + n3 x2 )
2nx
, and gn (x) =
for x [0, 1]. Show that
2
n
1 + n3 x2
the sequence (gn ) converges uniformly to g where g(x) = 0 for all x [0, 1].
Using this fact, show that (fn ) also converges uniformly to the zero function
on [0, 1].
3. Let fn (x) =
Additional Exercises
2
n x,
n2 x + 2n,
4. Let fn (x) =
0,
127
0 x 1/n,
1/n x 2/n,
2/n x 1.
X
an x2n
a
is
absolutely
convergent.
Show
that
n
n=1
1 + x2n
n=1
is a dominated series on R.
6. Show that for each p > 1, the series
X
xn
n=1
np
n=1
X
n=1
1
1
1X
0 n=1
X
1
1
dx =
1 + (k + 1)x 1 + kx
n=1 0
Z
9. Show that
1X
0 n=1
x
1
dx = .
(n + x2 )2
2
1
1
1
dx.
1 + (k + 1)x 1 + kx
7
Power Series
7.1
n=0 an x
P
n
Note that a power series
n=0 an x converges at the point x = 0. What can we
say about its domain of convergence?
P
n
Theorem 7.1 (Abels theorem) Consider the series
n=0 an x .
(i) If the above series converges at a point x0 , then it converges absolutely at
every x with x < x0 .
(ii) If the above series diverges at a point y0 , then it diverges at every x with
x > y0 .
P
n
Proof. (i) Suppose
n=0 an x converges at a point x0 6= 0. Let x be such that
x < x0 . Then, since we have
n
n
n x
n.
an x  = an x0 
x0
Since an xn0  0 as n , there exists M > 0 such that an xn0  M for all n, so
that we have
n
x
n
n.
an x  M
x0
P
n
Now, since xx0 < 1, it follows, by comparison test that
n=0 an x  converges.
P
P
n
n
(ii) Suppose
n=0 an x diverges at a point y0 6= 0. If
n=0 an x converges at a
point x with x > y0 , then by part (i), it converges at y0 as well, which contradicts
our assumption. Hence, the conclusion holds.
P
n
Theorem 7.1 shows that if the series
n=0 an x converges at some nonzero point
x0 , then there it converges for all x in an interval of the form (a, a), where a x0 .
128
129
Definition
7.2 The domain (or interval) of convergence of a power series
P
n is the set
a
x
n
n=0
D := {x R :
an xn converges at x}
n=0
and
R := sup{x : x D}
P
n
is called the radius of convergence of
n=0 an x .
P
n
Thus, a number
R
with
0
<
R
<
is
the
radius
of
convergence
of
n=0 an x if
P
and only if n=0 an xn converges for all x with x < R, and diverges at all x with
x > R.
P
n
If the power series
n=0 an x converges only at the point 0, then the radius of
convergence is 0, and if it converges at all points in R, then sup{x : x D} does
not exists, and in that case we say that the radius of convergence is , i.e., we write
R = .
P
n
EXAMPLE 7.1 Consider the power series
n=0 x . In this case, we know that
the series converges for x with x < 1, and diverges for x with x > 1. Also, the
series diverges at x {1, 1}. Hence, its radius of convergence is 1, and its domain
(interval) of convergence is D = {x : 1 < x < 1} = (1, 1).
P xn
EXAMPLE 7.2 Consider the power series n=0 n . In this case, we know that the
series converges at x = 1 and diverges at x = 1. Hence, its radius of convergence
is 1, and its domain (interval) of convergence is [1, 1).
P xn
EXAMPLE 7.3 Consider the power series n=0 n2 . We know that this series
converges at x = 1 and x = 1. Since
xn+1 /(n + 1)2 
n
= x
x as n ,
n
2
x /n 
n+1
xn
P
by ratio test the series
n=0 n2 converges for x with x < 1 and diverges for x
with x > 1. Therefore, the radius of convergence is 1, and the domain (interval)
of convergence is [1, 1].
P xn
EXAMPLE 7.4 Consider the power series n=0 n! . Since
xn+1 /(n + 1)!
1
= x
0
n
x /n!
n+1
as
n ,
by ratio test the series converges at every x R. Hence, the radius of convergence
is , and the domain (interval) of convergence is R.
For finding the radius of convergence and domain of convergence, the following
theorem will be useful.
130
Power Series
M.T. Nair
7.2
Uniform Convergence
Theorem
7.3 Suppose R > 0 is the radius of convergence of a power series
P
n
n=0 an x . Then
P
n
(i)
n=0 an x converges uniformly on [, ] for any with 0 < < R, and
P
n
(ii) the function f defined by f (x) :=
n=0 an x , x (R, R), is continuous
on (R, R).
Proof. Let 0 < < R, and r such that < r < R. Then for every x with x ,
we have
x n
n
an xn  an rn 
 an rn 
.
r
r
P
n
n
bounded, say
Since the series
n=0 an r is convergent, the sequence (an r ) is P
n
n
an r  M for all n N, for some M >P0. Also, since r < 1,
n=0 an x is
n
a dominated on [, ]. Hence the series n=0 an x Pis uniformly convergent on
n
[, ]. Hence, the function f defined by f (x) :=
n=0 an x is continuous on
[, ]. Since, this is true for any < R, the function f is continuous at every
x (R, R). [Given any x0 (R, R), we may take such that x0  < < R.]
7.3
Theorem
7.4 Suppose R > 0 is the radius of
convergence of a power series
P
P
n . Then the radius of convergence of
a
x
an xn1 is R. In particun=0 n
P n=1 nn1
lar, for any with 0 < < R, the series n=1 n an x
converge uniformly on
[, ] and it represents a continuous function on (R, R).
131
Proof. Let x0 (R, R), and let such that x0  < < R. Then, we have
nan xn1

0
n1
= nan

x0 
n1
n N.
P
n
n1  0 so that it is bounded, say a n1  M
Since
n
n=0 an converges, an
for all n N. Thus,
nan xn1

0
Since
P
n=0 n
n=0 an
xn
x0 
n1
Mn
x0 
n1
n N.
P
n
n=0 an x0
P
n1 . Suppose
It
remains
to
show
that
R
is
the
radius
of
convergence
of
n=1 nan x
P
n1 converges at some point y with y  > R. Then taking r with
0
0
n=1 nan x
P
n1 converges (by Abels theorem applied
na
r
y0  > r > R, we see that
n
n=1
P
n
to this series). But, nan rn1  an rn /r so that by comparison test
n=1 an r
converges.
This is not possible since r > R. Thus, the radius of convergence of
P
n1 is R.
na
x
n
n=1
Theorem
7.5 Suppose R P
> 0 is the radius of convergence of a power series
P
n
n
n=0 an x for x < R. Then the following hold.
n=0 an x , and let f (x) :=
(i) f (x) is a continuous function for x < R, and for [a, b] (R, R),
Z
f (x)dx =
a
X
an
[bn+1 an+1 ].
n+1
n=0
X
d
f (x) =
nan xn1 .
dx
n=1
Proof. The result in (i) follows from Theorems 7.3 and 6.6. Also, by Theorems
7.4 and 6.7 it follows that
X
d
f (x) =
nan xn1
dx
n=1
for every x [, ], where 0 < < R. Now, if x (R, R), then we may take
such that x [, ], and the result is valid for such x as well.
132
Power Series
7.3.1
M.T. Nair
P
n
Some of the series may not be in the standard form
n=0 an x , but can be converted
into this form after some change of variable. For example, consider the series
(i)
an (x x0 )n , (ii)
n=0
an x3n , (iii)
n=0
an
X
1
,
(iv)
an sinn x.
xn
n=0
n=0
7.4
Additional Exercises
X
(1)n
n=1
(iii)
n=1
nxn
(ii)
X
(1)n 3n
.
(4n 1)xn
n=1
5)n
n(x +
,
(2n + 1)3
(iv)
X
2n sinn x
n2
n=1
X
(n 1)!
n=0
nn
xn .
Ans: e
X
X
X
n n
n2 n
(i)
x for > 0;
(ii)
x for > 0;
(iii)
xn! .
n=0
(iv)
n=0
(1)n
n=0
xn(n+1) ;
(v)
X
sin(n/6)
n=0
2n
(x 1)n ;
(vi)
n=0
X
n=0
(i)n 2n
x .
4n n
4. Show that
(i) log(1 + x) =
(1)n+1
n=1
(ii) tan1 x =
(1)n
n=0
x2n+1
for 1 < x 1,
2n + 1
X
xn
n=1
dominated on R?
xn
for 1 < x 1,
n
n!
8
Fourier Series
While studying the heat conduction problem in the year 1804, Fourier found it
necessary to use a special type of function series associated with certain functions f ,
later known as Fourier series of f . In this chapter we study such series of functions.
8.1
8.1.1
Definition 8.1 Let (an ) and (bn ) be sequences of real numbers. Then a series of
the form
X
c0 +
(an cos nx + bn sin nx)
n=1
PWe observe that the functions cos nx and sin nx are 2periodic. Hence, if c0 +
n=1 (an cos nx + bn sin nx) converges to a function f (x), the f (x) also has to be
2periodic. Thus, only a 2periodic function is expected to have a trigonometric
series expansion.
Definition 8.2 A function f : R R is said to be T periodic for some T > 0 if
f (x + T ) = f (x for all x R.
Now, suppose that f is a 2periodic function. We would like to know whether
f can be represented as a Fourier series. Suppose, for a moment, that we can write
f (x) = c0 +
x R.
n=1
Then what should be an , bn ? To answer this question, let us further assume that
the series can be termwise integrated. For instance if the above series is uniformly
convergent to f , then termwise integration is possible. Recall that if (an ) and (bn )
X
are such that
(an  + bn ) converges, then the above series is dominated and
n=0
133
134
Fourier Series
M.T. Nair
Z
0,
sin nx sin mxdx =
,
Z
if n = m
if n 6= m,
if n = m
if n 6= m,
Therefore, under the assumption that the series can be integrated termwise, we get
Z
f (x)dx = 2c0 ,
Z
f (x) cos nxdx = an n N,
Z
f (x) sin nxdx = bn n N.
Thus, c0 =
1
2
an =
bn =
a0 X
+
(an cos nx + bn sin nx) ,
2
n=1
where
1
an =
1
bn =
a0 X
f (x)
+
(an cos nx + bn sin nx) .
2
n=1
The following two theorems give sufficient conditions for the convergence of the
Fourier series of a function f to the function f at certain points x R.
Theorem 8.1 Suppose f is a bounded monotonic function on [, ). Then the
Fourier series of f converges to a function s(x), where
f (x)
if f continuous at x,
s(x) =
1
[f
(x)
+
f
(x+)]
if f not continuous at x.
2
135
a0 X
s(x) =
+
an cos nx
2
n=1
with
2
an =
In particular,
s(0) =
an ,
s() = a0 +
n=0
(1)n an .
n=1
X
2
bn sin nx with bn =
f (x) sin nxdx.
()
0
n=1
In particular,
s(/2) =
(1)n b2n+1 .
n=0
8.1.2
f (x)
f (x)
if 0 x < ,
,
if x < 0,
136
Fourier Series
M.T. Nair
feven (x) =
f (x)
f (x)
if 0 x < ,
if x < 0,
so that fodd and feven are the odd extension and even extension of f respectively.
Therefore,
a0 X
f (x)
+
an cos nx, x [0, ),
()
2
n=1
and
f (x)
bn sin nx,
x [0, ),
()
n=1
with
Z
Z
2
2
f (x) cos nxdx, bn =
f (x) sin nxdx.
an =
0
0
The expansions () and () are called, respectively, the even and odd expansions
of f on [0, ).
EXAMPLE 8.1 Consider the 2periodic function f with f (x) = x for x [, ].
Note that the f is an odd function. Hence, an = 0 for n = 0, 1, 2, . . ., and the Fourier
series is
X
bn sin nx, x [0, ]
n=1
with
Z
Z
2
2 h cos nx i
cos nx
bn =
x sin nxdx =
x
dx
+
0
n
n
0
0
2 n cos n o (1)n+1 2
=
=
.
n
n
Thus the Fourier series is
X
(1)n+1
2
sin nx.
n
n=1
n=1
n=0
X (1)n+1
n X (1)n+1
=
sin
=
.
4
n
2
2n + 1
EXAMPLE 8.2 Consider the 2periodic function f with f (x) = x for x
[, ]. Note that the f is an even function. Hence, bn = 0 for n = 1, 2, . . ., and
the Fourier series is
Z
a0 X
2
+
an cos nx, x [0, ], an =
x cos nxdx.
2
0
n=1
137
4
,
(2n + 1)2
a2n+1 =
n = 0, 1, 2, . . . .
4 X cos(2n + 1)x
,
2
(2n + 1)2
x [0, ].
n=0
2 X
1
=
.
8
(2n + 1)2
n=0
EXAMPLE 8.3 Consider the 2periodic function f with
1, x < 0,
f (x) =
1,
0 x .
Note that the f is an odd function. Hence, an = 0 for n = 0, 1, 2, . . ., and the Fourier
series is
X
bn sin nx, x [0, ]
n=1
with
2
bn =
2
f (x) sin nxdx =
Thus
sin nxdx =
0
2
(1 cos n).
4 X sin(2n + 1)x
.
f (x)
2n + 1
n=0
X (1)n
=
.
4
2n + 1
n=0
EXAMPLE 8.4 Consider the 2periodic function f with f (x) = x2 for x
[, ]. Note that the f is an even function. Hence, bn = 0 for n = 1, 2, . . ., and
the Fourier series is
Z
a0 X
2
+
an cos nx, x [0, ], an =
x cos nxdx.
2
0
n=1
X (1)n cos nx
2
x
+4
,
3
n2
2
n=1
x.
138
Fourier Series
M.T. Nair
2 X (1)n+1
,
=
12
n2
2 X 1
=
6
n2
n=1
n=1
respectively.
8.2
a0 X
+
(an cos nt + bn sin nt)
2
n=1
where
`t
f
cos ntdt =
Z
1
`t
bn =
f
sin ntdt =
1
an =
1
`
1
`
f (x) cos
nx
dx,
`
f (x) sin
nx
dx.
`
`
`
In particular,
if f is even, then bn = 0 for all n and
2
an =
`
f (x) cos
0
nx
dx,
`
8.2.1
f (x) sin
0
nx
dx,
`
Suppose a function f is defined in an interval [a, b). We can obtain Fourier expansion
of it on [a, b) as follows:
Method 1: Let us consider a change of variable as y = x a+b
2 . Let (y) := f (x) =
a+b
f (y + 2 ) where ` y ` with ` = (b a)/2. We can extend as a 2`periodic
function and obtain its Fourier series as
a0 X
n
n
(y)
+
an cos
y + bn sin
y
2
`
`
n=1
Additional Exercises
where
an =
bn =
1
`
1
`
(y) cos
nx
ydy,
`
(y) sin
nx
ydy.
`
`
`
139
a0 X
n
(y)
+
y
an cos
2
`
n=1
(y) cos
`
nx
ydy.
`
(b) For y [`, 0], define fo (y) = f(y). Thus fo on [`, `] is an odd function.
In this case,
X
n
(y)
bn sin
y
`
n=1
where
1
bn =
`
(y) sin
`
nx
ydy.
`
8.3
Additional Exercises
1,
2 x< 2
(a) f (x) =
3
0, 2 < x < 2 .
x,
2 x< 2
(b) f (x) =
3
x, 2 < x < 2 .
1 + 2x
, x 0
(c) f (x) =
2x
1 , 0 x .
(d) f (x) =
x2
4 ,
x .
140
Fourier Series
M.T. Nair
2. Using the Fourier series in Exercise 1, find the sum of the following series:
1 1
1
1 1 1
(b) 1 + + +
+ . . ..
(a) 1 + + . . .,
3 5 7
4 9 16
1 1
1
1
1
1
(c) 1 +
+ . . .,
(d) 1 + 2 + 2 + 2 + . . ..
4 9 16
3
5
7
sin x, 0 x 4
3. If f (x) =
, then show that
cos x, 4 x < 2
8
sin x sin 3x sin 10x
f (x) cos
+
+
+ ... .
4 1.3
5.7
9.11
4. Show that for 0 < x < 1,
8 sin x sin 3x sin 5x
+
+
+ ... .
xx = 2
13
33
53
2
sin 3x sin 5x
+
+ ... = .
3
5
4
cos 4x +
+ ....
1.3 3.5
5.7 7.9
7. Show that for 0 x ,
cos 2x cos 4x cos 6x
2
+
+
+ ... ,
x( x) =
6
12
22
32
8 sin x sin 3x sin 5x
x( x) =
+
+
+ ... .
13
33
53
8. Assuming that the Fourier series of f converges uniformly on [, ), show
that
Z
1
a2 X 2
[f (x)]2 dx = 0 +
(an + b2n ).
2
n=1
X
X
1
4
(1)n1
2
(a)
=
,
(b)
=
n4
90
n2
12
n=1
X
1
6
(c)
=
n6
945
n=1
(d)
n=1
n=1
(1)n1
3
=
(2n 1)3
32
10. Write down the Fourier series of f (x) = x for x [1, 2) so that it converges to
1/2 at x = 1.