You are on page 1of 65

ii Dr. V. V.

Acharya

FYBSc Calculus
Textbook written according to the latest
syllabus of Fergusson Collge, Pune
(Autonomous)

Dr. V. V. Acharya
Head, Department of Mathematics,
Fergusson College, Pune-411 004.
Dr. M. R. Modak
Former Head, Department of Mathematics,
S. P. College, Pune-411 030.

November 20, 2017


iv Dr. V. V. Acharya

1
Z
4.11.3 2 + a2 )n
dx, n a positive integer . . . . . . . . . . . 117
(x
4.11.4 R(x + a2 )n/2 dx, n a positive integer . . . . . . . . . . . 117
R 2
Contents 4.11.5 xn eax dx. . . . . . . . . . . . . . . . . . . . . . . . . 118
1 Continuity 1
1.1 Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 General Theorems . . . . . . . . . . . . . . . . . . . . . 10
1.2 Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3 Continuous Functions on an Interval . . . . . . . . . . . . . . . . 25
1.4 Applications of Continuity . . . . . . . . . . . . . . . . . . . . . 28
1.5 Uniform Continuity . . . . . . . . . . . . . . . . . . . . . . . . . 32

2 Derivative 43
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.2 Mean Value Theorems . . . . . . . . . . . . . . . . . . . . . . . 46
2.3 Indeterminate Forms . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4 Successive Differentiation . . . . . . . . . . . . . . . . . . . . . 63

3 Taylor’s theorem 69
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.2 Taylor’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.3 Maxima and Minima . . . . . . . . . . . . . . . . . . . . . . . . 76

4 Integration 79
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.2 Integration by substitution . . . . . . . . . . . . . . . . . . . . . 80
4.2.1 Some standard forms . . . . . . . . . . . . . . . . . . . . 81
4.3 Trigonometric functions . . . . . . . . . . . . . . . . . . . . . . 83
4.4 Some standard substitutions . . . . . . . . . . . . . . . . . . . . 86
4.5 The integrals involving
Z quadratic polynomials . . . . . . . . . . . 89
dx
4.6 Integrals of the type . . . . . . . . . . . 94
a + b cos x + c sin x
4.7 Integration by partial fractions . . . . . . . . . . . . . . . . . . . 95
4.8 Integration by parts . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.9 Standard Integrals . . R. . . . . . . . . . . . . . . . . . . . . . . . 107
4.10 Integrals of the form ex [f (x) + f ′ (x)] dx . . . . . . . . . . . 112
4.11 Reduction Formulae
Z . . . . . . . . . . . . . . . . . . . . . . . . . 114
n
4.11.1 I = sin xdx . . . . . . . . . . . . . . . . . . . . . . 114
sinn x cosm xdx . . . . . . . . . .
R
4.11.2 . . . . . . . . . . . 116

iii
vi Dr. V. V. Acharya

Preface
This book has been developed out of the notes written while teaching this course.
Many problems are solved. Exhaustive exercise sets are given at the end of each
chapter. We believe that this will make book student friendly. We have taken at
most care so as to avoid the errors (both mathematical and typographical). How-
ever, if you find any errors please send the errors to me at vvacharya@gmail.com.
Suggestions for improvement of the notes are most welcome.

Dr. V. V. Acharya
November 20, 2017

v
2 Dr. V. V. Acharya

(iii) 0 < |x − 2| < k implies |f (x) − 8| < 0.006 ?


Indeed, we can. For this, we start with the right-hand side of (iii) and argue as
Chapter 1 follows :

We will have |f (x) − 8| < 0.0006


Continuity if 3|x − 2| < 0.006,
i.e. if, on dividing by 3,
1.1 Limits 1
|x − 2| < (0.006) = 0.002.
3
The concept of limit of a function at a point forms the foundation of Calculus.
Therefore we must study the definition of limit carefully and must become familiar
with its meaning by applying it in many particular cases. 9
Consider the function given by 8+ǫ
b

f (x) = 3x + 2, if x 6= 2. 8 8−ǫ
b

b b

We want to examine the behaviour of the values of f when x is near 2 and is


7
different from 2. In other words, we want to look at the values of f when x
varies over a deleted neighbourhood of 2. So we construct the following table:
6
x 1.9 1.98 1.998 2.001 2.01
f (x) 7.7 7.94 7.994 8.003 8.03
5
|x − 2| 0.1 0.02 0.002 0.001 0.01
|f (x) − 8| 0.3 0.06 0.006 0.003 0.03
4
The table shows that
(i) When x is chosen close to 2 (where either x < 2 or x > 2, but x 6= 2) then 3
the value f (x) of f at x is close to 8. For example,

f (1.998) = 7.994, and f (2.001) = 8.003. 2

Hence the number 8 has the property that


1
(ii) When |x − 2| is positive (since x 6= 2) and small, the |f (x) − 8| is also small.
Thus since |f (x) − 8| = |3x + 2 − 8| = 3|x − 2|, we see that 2−δ
b b
2+δ

if 0 < |x − 2| < 0.001, then −1 1 2

|f (x) − 8| = 3|x − 2| < 3(0.001) = 0.003. f

Suppose we want to make |f (x) − 8| still smaller; e.g. suppose we want to make
|f (x) − 8| < 0.006. How small |x − 2| should be to achieve this ? That is, can we
find a number k such that Figure 1.1: Limit of a function

1
Calculus 3 4 Dr. V. V. Acharya

Hence we may take k = 0.002 so that (iii) holds. Of course, there is nothing (ii) For every ǫ-neighbourhood N (l, ǫ) of l, we can find at least one deleted
special about the number 0.006 in (iii). In fact, we can make the difference |f (x)− neighbourhood of l, say N (a, δ) such that
8| as small as we like i.e. we can make |f (x) − 8| less than any give positive
number ǫ, by taking x such that |x − 2| is sufficiently small. x ∈ N (a, δ) ∩ D ⇒ f (x) ∈ N (l, ǫ).

We have |f (x) − 8| < ǫ if 3|x − 2| < ǫ, The above definition gives us a condition to decide whether a given number l is
ǫ the limit of f as x tends to a. Note carefully the order in which the numbers ǫ and
i.e. if, on dividing by 3, |x − 2| < .
3 δ appear in the definition : first the positive number ǫ is given to us and then we
are required to discover a positive number δ for which (2) holds.
Hence it is sufficient to take x such that |x − 2| < ǫ/3, in order that the
In the Theorem below we show that the limit of f as x tends to a, is unique, if
condition |f (x) − 8| < ǫ, is satisfied. So, corresponding to the given positive
it exists. That is, if we find a number l which satisfies the condition stated in the
number ǫ, we have found a positive number δ (namely, δ = ǫ/3) such that
definition, then l is the limit of f as x tends to a.
(iv) 0 < |x − 2| < δ ⇒ |f (x) − 8| = 3|x − 2| < 3δ = ǫ
Theorem 1.1 The limit of a function f as x tends to a, is unique, if it exists.
⇒ |f (x) − 8| < ǫ.
Proof: Suppose as x → a, f has two limits l and m and let, if possible l 6= m.
This statement is expressed by writing
(v) lim f (x) = lim (3x + 2) = 8.
x→2 x→2
Geometrically, (v) means the following: given any neighbourhood (8−ǫ, 8+ǫ) l − ǫl
b b b
l+ǫ
b
m−
mǫ m + ǫ
b b

of 8 on y-axis, we are able to find a deleted neighbourhood of 2, namely N =


(2 − δ, 2 + δ) − {2}, on x-axis such that for every x in N, the value f (x) lies in
the given neighbourhood (8 − ǫ, 8 + ǫ) of 8.See Fig. 1. Figure 1.2: Uniqueness of the limit
Note that (v) remains true if we leave f (2) undefined or define f (2) to be 8 or
any other number. |l − m|
Then taking ǫ = , we see that ǫ > 0 and the ǫ neighbourhoods of l and
4
Definition 1.1 Let D be a subset of R. We say that a is a cluster point of D if for m are disjoint. But by the definition of limit, corresponding to this ǫ, there exist
every deleted neighbourhood of a contains at least one point of D. numbers δ1 > 0 and δ2 > 0 such that

Definition 1.2 Let D be a subset of R and f be a real valued function defined on x ∈ N ′ (a, δ1 )∩D ⇒ f (x) ∈ N (l, ǫ) and x ∈ N ′ (a, δ2 )∩D ⇒ f (x) ∈ N (m, ǫ).
D. Suppose a is a cluster point of D and l is a given number. If for every number
ǫ > 0, there exists (i.e. we can find) a number δ > 0 such that Hence if δ = min{δ1 , δ2 }, then

for every x ∈ D satisfying 0 < |x − a| < δ, we have |f (x) − l| < ǫ, x ∈ N ′ (a, δ) ∩ D ⇒ x ∈ N ′ (a, δ1 ) ∩ D and x ∈ N ′ (a, δ2 ) ∩ D
⇒ f (x) ∈ N (l, ǫ) and f (x) ∈ N (m, ǫ).
then we say that f has limit l as x tends to a and write

f (x) → l as x → a or lim f (x) = l. (1) This is contradicts the fact that the neighbourhoods N (l, ǫ) and N (m, ǫ) are dis-
x→a joint. Hence l = m, i.e. the limit of f is unique.
Two equivalent definitions of the statement (1) are as follows:
Example 1.1 Limit of a constant function: Let f (x) = c = a constant for all x.
(i) Given ǫ > 0, there is δ > 0 such that
Then
a − δ < x < a + δ and x 6= a, x ∈ D ⇒ l − ǫ < f (x) < l + ǫ. (2) lim f (x) = lim c = c.
x→a x→a
Calculus 5 6 Dr. V. V. Acharya

To prove this, let ǫ > 0 be given. Then it is enough to take δ = ǫ [ actually, in this 3
Example 1.4 Evaluate lim .
simple case, any positive number δ will do], because then x→1 x+2
3
Here we are concerned with the behaviour of the function f (x) = when
0 < |x − a| < δ ⇒ |f (x) − c| = |c − c| = 0 < ǫ, x+2

x is near 1. So we take the domain of f to be the deleted 1-nhd S = N (1, 1) =
as required. (0, 2) − {1}.
3
Example 1.2 Limit of an identity function: Let f (x) = x for all x near a. Then Since, when x is near 1, the expression x + 2 is near 3 so is near 1, we
x+2
lim f (x) = lim x = a. 3
x→a x→a expect that lim = 1. To prove this, let ǫ > 0 be given. Now
x→1 x + 2
To prove this, let ǫ > 0 be given. Take δ = ǫ. Then 3 |3 − x − 2| |x − 1|
|f (x) − 1| = − 1 = = .

0 < |x − a| < δ ⇒ |f (x) − a| = |x − a| < δ = ǫ, x+2 |x + 2| |x + 2|
as required. Also,
2 1 1
Example 1.3 Prove that lim x = 4.
x→2 x∈S ⇒ x>0 ⇒ x+2>2⇒ <
2
Here f (x) = x . Since we wish to find the limit of f at 2, it is enough to x+2 2
consider the values of f in some deleted neighbourhood of 2. So let us choose the |x − 1| 1
⇒ |f (x) − 1| = < |x − 1|,
deleted neighbourhood x+2 2

S = N ′ (2, 1) = (1, 3) − {2}, on multiplying by the positive number |x − 1|. (|x − 1| > 0 since x 6= 1.) Hence
it is enough to take δ = min{1, 2ǫ}. Then
as the domain of f. Let ǫ > 0 be given. We now have to decode how small the
difference |x − 2| should be, in order that the inequality |f (x) − 4| < ǫ holds. 0 < |x − 1| < δ ⇒ |f (x) − 1| < ǫ.
Consider the difference Thus f (x) → 1 as x → 1.
d = |f (x) − 4| = |x2 − 4| = |x + 2| · |x − 2|.
Note: We assume, without proof, the following limits:
We wish to find a condition on the factor |x − 2| under which d < ǫ will be true. sin x
For the other factor, |x + 2|, note that if x ∈ S, then |x| = x < 3, so lim sin x = 0, lim cos x = 1, lim = 1,
x→0 x→0 x→0 x
|x + 2| ≤ |x| + |2| < 3 + 2 = 5. lim (1 + x)1/x = e,
x→0
Hence for all x in S, we have ax − 1
lim = loge a, where a is a positive constant,
x→0 x
|f (x) − 4| = |x + 2| · |x − 2| < 5|x − 2|.
loge (1 + x)
and lim = 1.
Therefore, if x is such that 5|x − 2| < ǫ, i.e. such that |x − 2| < ǫ/5, then x→0 x
certainly, it will be true that |f (x) − 4| < ǫ. Hence we take δ = min{1, ǫ/5}.
Then In the examples above, we have directly applied the definition to prove that the
given function has the given number as its limit. This process can be often be
0 < |x − 2| < δ ⇒ x ∈ S since δ ≤ 1, and simplified by using the general properties of limits, given below.
|f (x) − 4| < 5|x − 2| < 5δ ≤ ǫ since δ ≤ ǫ/5.
Here let us note the following points about the above definition of the limit of a
Hence f (x) → 4 as x → 2. function:
Calculus 7 8 Dr. V. V. Acharya

1. Suppose corresponding to ǫ > 0, we have found a number δ > 0 for which Hence (4) follows.
the following condition holds: Actually, there is no number l, such that
x ∈ N ′ (a, δ) ⇒ |f (x) − l| < ǫ. (3) 1
lim cos = l, (5)
x→0 x
Then for every δ1 such that 0 < δ1 < δ, condition (3) also holds because 1
N ′ (a, δ1 ) ⊆ N ′ (a, δ). and this fact is expressed by saying that lim cos does not exist. To prove this,
x
x→0
2. To show that lim f (x) = l as x → a, it is enough to show that for every suppose (5) holds for some l in R. Then corresponding to ǫ = 1/2, there would
ǫ1 > 0, there is δ1 > 0 such that exist δ > 0 such that
1 1
x ∈ N ′ (a, δ1 ) ⇒ |f (x) − l| < kǫ1 , x ∈ N ′ (a, δ) ⇒ cos − l < . (6)

(i) x0 2
where k is a positive constant. To see why this is enough, suppose that we Choose a positive integer n so large that
have proved (i). Then to verify that the definition of limit is satisfied, let 1 1
ǫ > 0 be given. Then by (i), corresponding to ǫ1 = ǫ/k, there is δ1 > 0 x0 = < δ and x1 = < δ.
2nπ (2n + 1)π
such that
x ∈ N ′ (a, δ1 ) ⇒ |f (x) − l| < kǫ1 = ǫ, Then x0 , x1 are in N ′ (0, δ) and so by (6),
as required. 1 1
| cos − l| = |1 − l| < ,
x0 2
It is important to know the negation of the statement lim f (x) = l. Thus the limit so that l > 0; and again by (6),
x→a
of f (x) as x → a is not equal to l means that 1 1
| cos − l| = |1 − l| < ,
there is some ǫ′ > 0 for which we can find no number δ > 0 for which the x1 2
following condition holds: so that l < 0. This contradiction proves that there is no real number l satisfying
for every x in N ′ (a, δ), |f (x) − l| < ǫ′ (ii) (5).
that is, such that for every δ > 0, condition (ii) fails
that is, such that for every δ > 0, there is at least one number x0 in N ′ (a, δ), One-sided limits: If f (x) → l as x → a, then l is called the two-sided limit of f
such that at a. We also need the one-sided limits of f at a, which are defined as follows:
|f (x0 ) − l| ≥ ǫ′ . Definition 1.3 If for every ǫ > 0, there is δ > 0 such that
For example, let us show that a < x < a + δ ⇒ |f (x) − l1 | < ǫ,
1
lim cos 6= 1. (4) we say that l1 is the right limit of f at a and write
x→0 x
lim f (x) = l1 or f (a+) = l1 .
For this, let ǫ′ = 1. Let δ > 0 be any given number. Then we choose a positive x→a+
integer n so large that (ii) If for every ǫ > 0, there is δ > 0 such that
1
x0 = < δ.
(2n + 1)π/2 a − δ < x < a ⇒ |f (x) − l2 | < ǫ,

Then x0 is in (−δ, δ) − {0} i.e. x0 is in N (0, δ), and we say that l2 is the left limit of f at a and write
1 π lim f (x) = l2 or f (a−) = l2 .
cos − 1 = | cos(2n + 1) − 1| = |0 − 1| = 1 = ǫ′ .

x→a−
x0 2
Calculus 9 10 Dr. V. V. Acharya

Example 1.5 Since [x] = 2 if x ∈ (2, 3) and [x] = 3 if x ∈ (3, 4), we get For example, let g(x) = x/|x| for x 6= 0. Then, as seen in Example 2) above,

lim [x] = 2 and lim [x] = 3. lim g(x) = 1 while lim g(x) = −1.
x→3− x→3+ x→0+ x→0−

See Fig. ** Similarly, the reader should check that So the one-sided limits of g at 0 are not equal. Therefore, by the above theorem,
limx→0 g(x) does not exist.
lim (x − [x]) = 1 and lim (x − [x]) = 0.
x→3− x→3+
Finally, we define infinite limits and limits at infinity.
See Fig. *
(i) lim f (x) = ∞ means that for every ∆ > 0, there is δ > 0 such that
x→a
Example 1.6 Let g(x) = x/|x| for x 6= 0. To find lim g(x), we consider only 0 < |x − a| < δ ⇒ f (x) > ∆.
x→0+
the positive values of x near zero. For these values of x, |x| = x, hence g(x) = 1. (ii) lim f (x) = −∞ means that for every ∆ > 0, there is δ > 0 such that
x→a
Hence 0 < |x − a| < δ ⇒ f (x) < −∆.
lim g(x) = lim 1 = 1.
x→0+ x→0+
(iii) lim f (x) = l means that for every ǫ > 0, there is X > 0 such that
x→∞
Similarly, when x is negative and near zero, |x| = −x and g(x) = −1. So x > X ⇒ |f (x) − l| < ǫ.
lim g(x) = lim (−1) = −1. (iv) lim f (x) = l means that for every ǫ > 0, there is X > 0 such that
x→0− x→0− x→−∞
x < −X ⇒ |f (x) − l| < ǫ.
Theorem 1.2 The two-sided limit of f at a exists (and equals l) if and only if
both the one-sided limits of f at a exist and are equal (and their common value is Definitions of statements such as lim f (x) = −∞ are analogous and are left
x→a+
l). to the reader.
Proof: First suppose that f (x) → l as x → a. Then given ǫ > 0 there is δ > 0
such that 1.1.1 General Theorems
x ∈ (a − δ, a + δ) − {a} ⇒ |f (x) − l| < ǫ.
Theorem 1.3 If the limit of a function f as x tends to a, exists, then f is bounded
Hence both the statements in some deleted neighbourhood of a, i.e. the are numbers δ > 0 and k > 0 such
that
a − δ < x < a ⇒ |f (x) − l| < ǫ and a < x < a + δ ⇒ |f (x) − l| < ǫ
|f (x)| ≤ k, for all x ∈ N ′ (a, δ).
are true. So we get
lim f (x) = l = lim f (x). (7) Proof: Suppose lim f (x) = l. Take ǫ = 1. Then there exists δ > 0 such that
x→a− x→a+ x→a

Conversely, suppose that (8) is satisfied. Then given ǫ > 0, there exist δ1 > 0 and
x ∈ N ′ (a, δ) ⇒ |f (x) − l| < 1
δ2 > 0 such that
⇒ |f (x)| = |f (x) − l + l| ≤ |f (x) − l| + |l| < 1 + |l|.
a − δ1 < x < a ⇒ |f (x) − l| < ǫ and a − δ2 < x < a ⇒ |f (x) − l| < ǫ
Hence taking k = 1 + |l|, we see that f is bounded on N ′ (a, δ).
Hence taking δ = min{δ1 , δ2 }, we see that |f (x) − l| < ǫ holds when a − δ <
x < a and also when a < x < a + δ i.e. when x ∈ (a − δ, a + δ) − {a}. Hence Theorem 1.4 Suppose, as x → a, f (x) → l where l is a non-zero number. Then
limx→a f (x) = l. (i) f (x) and l have the same sign in some deleted neighbourhood of a and
Calculus 11 12 Dr. V. V. Acharya

(ii) there exists δ > 0 such that Theorem 1.5 The following two statements are equivalent:
1 (i) lim f (x) = l,
|f (x)| > |l| > 0, for all x ∈ N ′ (a, δ). x→a
2
(ii) lim f (xn ) = l for every sequence {xn } which is properly convergent to
n→∞
Proof: Since l 6= 0, we have |l| > 0, hence for ǫ = 12 |l|, there is δ > 0 such that a.
1 Proof: Suppose (i) holds. Then given ǫ > 0, there is δ > 0 such that
x ∈ N ′ (a, δ) ⇒ |f (x) − l| < |l|
2
1 1 0 < |x − a| < δ ⇒ |f (x) − l| < ǫ. (1)
⇒ l − |l| < f (x) < l + |l|.
2 2
(i) If l > 0, then |l| = l, hence Let {xn } be a sequence which properly converges to a. Since lim xn = a, there
is a positive integer r such that |xn − a| < δ for all n ≥ r. Also, xn 6= a, for all
1 1
x ∈ N ′ (a, δ) ⇒ f (x) > l − l = l > 0, n, hence we have |xn − a| > 0 for all n. Hence
2 2
′ n ≥ r ⇒ 0 < |xn − a| < δ.
so that f (x) is also positive in N (a, δ). (2)
If l < 0, then |l| = −l, hence
Combining (1) and (2) we obtain
1 1
x ∈ N ′ (a, δ) ⇒ f (x) < l − l = l < 0,
2 2 n ≥ r ⇒ 0 < |xn − a| < δ ⇒ |f (xn ) − l| < ǫ.
so that f (x) is also negative in N ′ (a, δ).
(ii) We have So, lim f (xn ) = l as n → ∞.
1 Conversely, (ii) implies (i). For this we will prove the contrapositive, namely
x ∈ N ′ (a, δ) ⇒ ||f (x)| − |l|| ≤ |f (x) − l| < |l| if (i) is false, then (ii) is false. So suppose that (i) is false. Then there is a number
2
1 ǫ′ > 0 such that for every δ > 0, there is at least one point x0 such that
⇒ ||f (x)| − |l|| < |l|
2
1 0 < |x0 − a| < δ and |f (x0 ) − l| ≥ ǫ′ .
⇒ |l| − |l| < |f (x)|
2
1 Here the point x0 depends on δ, in general. Hence, foe every positive integer n,
⇒ |f (x)| > |l| > 0. corresponding to δ = 1/n, we can find a point xn such that
2
In the next theorem we give an alternative definition of limit of a function in 1
terms of sequences. Using this sequential definition, we can easily derive many 0 < |xn − 1| < (3)
n
properties of limits of functions from the corresponding properties of limits of
sequences. and
Recall that when we consider the behaviour of a function f as x → a, we |f (xn ) − l| ≥ ǫ′ . (4)
consider values of x near a, but different from a. That is, we consider values
of x in a deleted neighbourhood of a. For this purpose we make the following So, we have produced a sequence {xn } which, by (3), converges properly to a but
defintion: for which, by (4), lim f (xn ) 6= l. So, (ii) is false and the proof is complete.
We say that a sequence {xn } is properly convergent to a, if
We now apply the last theorem to deduce some properties of limits of functions
xn 6= a for all n and lim xn = a.
n→∞ from the corresponding properties of limits of sequences.
Calculus 13 14 Dr. V. V. Acharya

Theorem 1.6 Suppose l, m are real numbers and Proof: By the data, it follows that, for every sequence {xn } which properly con-
verges to a, we have
lim f (x) = l, and lim g(x) = m. (5)
x→a x→a
lim f (xn ) = l and lim g(xn ) = l.
n→∞ n→∞
Then as x → a,
(i) lim[f (x) ± g(x)] = l ± m, Therefore, since f (xn ) ≤ h(xn ) ≤ g(xn ), for all n, it follows that limn→∞ h(xn )
(ii) lim cf (x) = cl, c = a constant, exists and is l.
(iii) lim f (x)g(x) = lm,
f (x) l Notes: (i) The results (i) and (ii) of the above theorem can be extended to any
(iv) lim = ,
g(x) m finite number of functions f1 , f2 , . . . , fn by induction. That is, if for i = 1, . . . , n,
(v) lim |f (x)| = |l|. fi (x) → li as x → a where li are real numbers, then as x → a,
Proof: (i) By (5), we see by the last theorem, that for every sequence {xn } which h i
properly converges to a, we have lim f1 (x) + · · · + fn (x) = l1 + · · · + ln ,
x→a
h i
lim f (xn ) = l and lim g(xn ) = m. and lim f1 (x) · · · fn (x) = l1 · · · · ln .
n→∞ n→∞ x→a

So limn→∞ [f (xn ) + g(xn )] = l + m. Hence, again by the last theorem, the (ii) We can use Theorem 5 to prove the non-existence of a limit. For example,
function f + g has limit l + m i.e. limx→0 sin(1/x) does not exist. For this, consider the sequence {xn } where

lim [f (x) + g(x)] = l + m. 1


x→a xn = .
(n + 12 )π
Similarly, the remaining parts of the theorem can be proved.
Clearly, {xn } converges to zero properly. But {sin(1/xn )} does not con-
n
Theorem 1.7 Suppose as x → a, lim f (x) = l and lim g(x) = m where l, m   for all n, sin(1/xn ) = cos(nπ) = (−1) . Hence by Theorem 5,
verge because
are real numbers and f (x) ≤ g(x) in some deleted neighbourhood of a. Then 1
lim sin does not exist.
l ≤ m. x→0 x
Proof: By the data, it follows that, for every sequence {xn } which properly con- (iii) Limits of polynomials and rational functions: As seen in Example 2, §3.2,
verges to a, we have we know that limx→c x = c. Hence by (i), above we see that,
 n
lim f (xn ) = l and lim g(xn ) = m. lim xn = lim x = cn ,
n→∞ n→∞ x→c x→c

Therefore, since f (xn ) ≤ g(xn ), for all n, it follows that for every positive integer n.
So, again by (i), we see that if p(x) = a0 + a1 x + · · · + an xn , is a polynomial,
lim f (xn ) = lim g(xn ), i.e. l ≤ m. then as x → c,
n→∞ n→∞

Theorem 1.8 (Sandwich Principle) Suppose lim f (x) = l and lim g(x) = l lim p(x) = lim a0 + lim a1 x + · · · + lim an xn
x→a x→a
where l is a real number and f (x) ≤ h(x) ≤ g(x) in some deleted neighbourhood = a0 + a1 lim x + · · · + an lim xn
of a. Then the limit, = a0 + a1 c + · · · + an cn
lim h(x), exists and is l. = p(c). (6)
x→a
Calculus 15 16 Dr. V. V. Acharya

Hence if q(x) is a polynomial such that q(c) 6= 0, then by Theorem 5 (iv), we can Condition (2) is equivalent to saying that f (a+) = f (a), i.e.
evaluate the limit of the rational function p(x)/q(x) thus: Since limx→c q(x) =
q(c) 6= 0, lim f (x) exists and equals f (a).
x→a+
p(x) limx→c p(x) p(c)
lim = = . (7) If f is defined on the interval (a−k, a] for some k > 0, we say that f is continuous
x→c q(x) limx→c q(x) q(c)
from the left at a if for every ǫ > 0, there is a number δ > 0 such that
Exercise Set 1.1 a − δ < x ≤ a ⇒ |f (x) − f (a)| < ǫ. (3)
1. Prove the following either using the definition of limit or using theorems on Condition (3) is equivalent to saying that f (a−) = f (a), i.e.
limits.
x2 − 3x + 2 lim f (x) exists and equals f (a).
(i) lim (5x + 4) = 9 (ii) lim =1 x→a−
x→1 x→2 x−2
5x + 7 Further, we say that f is continuous on a closed and bounded interval [a, b], if
(iii) lim x3 = 1 (iv) lim =7 f is continuous at every point of the open interval (a, b) and f (a+) = f (a) and
x→1 x→0 3x + 1
f (b−) = f (b).
2. Show that the following limits do not exist:
1 x 1 Graphically, a function f is continuous at a point a means that there is no gap
(i) lim 2 (ii) lim (iii) lim sin 2 in the graph of f at x = a.
x→0 x x→0 | x | x→0 x
1 A function f is said to be discontinuous at a point a of the domain of f if f is
(iv) lim √ (v) lim e1/x (vi) lim (x − [x])
x→0+ x x→0 x→1 not continuous at a. Thus f is discontinuous at a point a of its domain if either
(i) the limit of f as x → a does not exist or
3. Show that if as x → a, lim f (x) exists but lim g(x) does not exist, then
(ii) the limit of f as x → a exists but lim f (x) 6= f (a).
lim [f (x) + g(x)] does not exist. x→a
x→a
Example 1.7 Consider the function f defined on R as follows :
1.2 Continuity (
2x + 3 if x ≤ −2
f (x) =
Definition 1.4 Let f be a function defined on the neighbourhood (a − k, a + k) x2 − 5 if x > −2.
of a point a for some k > 0. Then f is said to be continuous at the point a if for
every ǫ > 0, there is a number δ > 0 such that To test the continuity of f at x = −2, note that f is defined by different formulas
for x ≤ −2 and for x > −2. Therefore, we must find the one-sided limits of f at
|x − a| < δ ⇒ |f (x) − f (a)| < ǫ. (1) −2 by using the corresponding formulas thus :

Since f is defined on a full neighbourhood of a, condition (1) is equivalent to f (−2−) = lim f (x) = lim (2x + 3) = −1
x→−2− x→−2−
saying that
and f (−2+) = lim f (x) = lim (x2 − 5) = 22 − 5 = −1.
x→−2+ x→−2+
lim f (x) exists and equals f (a).
x→a
Also, f (−2) = 2(−2) + 3 = −1. Hence
If f is defined on the interval [a, a+k) for some k > 0, we say that f is continuous
from the right at a if for every ǫ > 0, there is a number δ > 0 such that f (−2) = −1 = lim f (x) = lim f (x).
x→−2− x→−2+

a ≤ x < a + δ ⇒ |f (x) − f (a)| < ǫ. (2) Hence f is continuous at x = −2.


Calculus 17 18 Dr. V. V. Acharya

Example 1.8 Let f (x) = [x] be the greatest integer function. We shall prove that This condition is very useful in solving problems of the type given in Exam-
f is continuous at every point a except when a is an integer. ples 4,6 below.
Case i) Suppose a is not an integer. Let [a] = n so that n < a < n + 1. Now Note that it follows from the above Theorem that
f (x) = [x] = n for n ≤ x < n + 1 i.e. f is constant on the interval [n, n + 1). (i) if we can find a sequence {xn } such that xn → a as n → ∞ and limit of f (xn )
Hence as n → ∞ does not exist, then f is discontinuous at a and
lim f (x) = n = [a] = f (a). (ii) if we can find sequences {xn }, {yn } such that xn → a yn → a as n → ∞ but
x→a
lim f (xn ) 6= lim f (yn ), then f is discontinuous at a. See Exercise 1.2, problems
Therefore f is continuous at a. 3,4.
Case ii) Suppose a is an integer. Then It was seen in Note 2 to Theorem 6 of §2.3, that given any real number a, we
 can find a sequence {xn } of rational numbers and a sequence {yn } of irrational
a − 1 if a − 1 ≤ x < a numbers such that
f (x) =
a if a ≤ x < a + 1. lim xn = a and lim yn = a.
n→∞ n→∞
Hence
This property is often needed in applying the above Theorem.
lim f (x) = a − 1 while lim f (x) = a.
x→a− x→a+
The following theorem shows that the usual algebraic combinations of contin-
Thus the one-sided limits of f at a are unequal, hence f is discontinuous at a. uous functions are continuous:

Example 1.9 Find constants α, β if the function f defined below is continuous in Theorem 1.10 Suppose functions f and g are continuous at a. Then the functions
(−2, 2). f + g, f − g, kf where k is a constant, and f g are continuous at a. The function
f /g is continuous at a if g(a) 6= 0. Further, |f | is continuous at a.

 x + α if −2 < x < 0 Proof: Let us prove that if g(a) 6= 0, then f /g is continuous at a. Since f, g are
f (x) = 2x + 1 if 0 ≤ x < 1 continuous at a, we have by definition,
β − x if 1 ≤ x < 2

lim f (x) = f (a) and lim g(x) = g(a).
Note that the formula for f changes at x = 0 and at x = 1. Now by data, the x→a x→a
function f is continuous on (−2, 2) and hence, in particular, at 0 and 1. Hence Hence, if g(a) 6= 0, by Theorem 6 of §1.1.1 we get
lim f (x) = f (0) i.e. lim (x + α) = 2(0) + 1 or α = 1.
x→0− x→0− f (x) lim f (x) f (a)
Similarly, lim f (x) = lim f (x) i.e. lim (β − x) = lim (2x + 1) lim = = .
x→1+ x→1− x→1+ x→1− x→a g(x) lim g(x) g(a)
or β − 1 = 3 or β = 4.
Hence f /g is continuous at a. The reader can similarly prove the remaining parts
A very convenient necessary and sufficient condition for a function f to be contin- of the theorem.
uous at a point a can be obtained using sequences. This is given in the following It is important to note that all the standard functions such as polynomials, rational
Theorem: functions, root functions, the logarithmic function, the exponential function, the
trigonometric functions and their inverses, and combinations of these functions
Theorem 1.9 A function f is continuous at a point a in its domain S, if and only
are continuous on their respective domains. We now state this fact in more detail
if for every sequence {xn } of points in S,
as follows:
we have lim f (xn ) = f (a) whenever lim xn = a. (i) If p(x) = a0 + a1 x + · · · + an xn , is a polynomial, and c is any point, then it
n→∞ n→∞
was proved in §1.1.1 that

Proof: Follows from Theorem 5 of §1.1.1. lim p(x) = p(c).


x→c
Calculus 19 20 Dr. V. V. Acharya

√ √
Hence the polynomial p(x) is continuous at c. Thus a polynomial is continuous So as x → 0, x → 0 = 0, and again (∗) holds.
at all points i.e. on R. In particular, the constant function f (x) = k = a constant (v) The functions sin x, cos x are continuous on R;
and the identity function g(x) = x are polynomials, hence they are continuous on tan x and sec x are continuous on R except at odd multiples of π/2; cot x and
their domains. cosec x are continuous on R except at integral multiples of π; F (x) = ex is con-
(ii) If q(x) is a polynomial such that q(c) 6= 0, then for the rational function tinuous on R; G(x) = log x is continuous on (0, ∞), and φ(x) = ax where a 6=
f (x) = p(x)/q(x) we have proved that 1 is a positive constant is continuous on R.
lim p(x) Note: Since lim x = a, we see that a function f is continuous at a point a of its
p(x) p(c) x→a
lim f (x) = lim = x→c = . domain
x→c x→c q(x) lim q(x) q(c)
x→c
if and only if lim f (x) = f (a)
Hence the rational function f is continuous at every point c at which the denomi- x→a

nator q(x) is non-zero. if and only if lim f (x) = f ( lim x). (4)
x→a x→a
(iii) Let n be a positive integer. Then the nth root function f given by
Hence f is continuous at a if and only if the operations of applying f and taking
f (x) = x1/n , for x ≥ 0, limit can be interchanged. For example, the square-root function is continuous at
every point a ≥ 0, means that we can write
is continuous on [0, ∞). √ q
(iii) As seen in Theorem 6, §1.1.1, we have for every a ∈ R, lim x = ( lim x).
x→a x→a

lim |x| = |a|. Let S, T be subsets of R and let f and g be real-valued functions such that f :
x→a
S → R, g : T → R and f (S) ⊆ T. Then we define the composite function g ◦ f
Hence the function |x| is continuous on R. on R thus:

(iv) The function f (x) = x is continuous at every a ≥ 0. for x ∈ S, (g ◦ f )(x) = g(f (x)).
Here we have to show that for all a ≥ 0, Regarding the continuity of the composite function, we have the following result:
√ √
lim x = a. (∗) Theorem 1.11 If f is continuous at a point a in S, and g is continuous at the point
x→a
√ √ √ f (a) in T, then the composite function g ◦ f is continuous at a.
Let ǫ > 0. First let a > 0. Then x + a ≥ a > 0 for x ≥ 0. Hence for x ≥ 0,
Proof: Let ǫ > 0 be given. Then since g is continuous at b = f (a), there is δ1 > 0
1 1 such that
√ √ ≤ √
x+ a a
|y − b| < δ1 ⇒ |g(y) − g(b)| < ǫ. (5)
√ √ |x − a| |x − a|
and so | x − a| = √ √ ≤ √ . Now since f is continuous at a, corresponding to δ1 > 0, there is δ2 > 0 such that
x+ a a
√ if y = f (x), then
So, taking δ = ǫ a, we get
|x − a| < δ2 ⇒ |f (x) − f (a)| < δ1 ⇒ |y − b)| < δ1 .
√√ |x − a| δ
|x − a| < δ ⇒ | x − a| ≤ √ < √ = ǫ. Hence we see that
a a
|x − a| < δ2 ⇒ |y − b| < δ1
Hence (∗) follows. Again, if a = 0, then given ǫ > 0, take δ = ǫ2 . Then
⇒ |g(y) − g(b)| < ǫ [by (5)]
√ √ √
0 ≤ x < δ ⇒ | x − 0| = x < δ = ǫ. ⇒ |g(f (x)) − g(f (a))| < ǫ.
Calculus 21 22 Dr. V. V. Acharya

So, g ◦ f is continuous at a. the composite function f is continuous at every point a where a 6= 0. But f is not
continuous at 0. For this, we apply Theorem 1 above. Let
This theorem allows us to infer the continuity√ of many complicated functions.
Thus for example, the function F (x) = sin x is continuous on [0, ∞) because 1 1
√ xn = < δ and yn = π < δ.
F = g ◦ f is the composite of the functions f (x) = x and g(x) = sin x and 2nπ 2nπ + 2
both f and g continuous on [0, ∞). Similarly, it is easy to see that the functions
esin x , log(x2 + 1) and | cos x| are continuous on R. Then xn → 0 and yn → 0 as n → ∞ but the sequences {f (xn )} and {f (yn )}
have different limits: lim sin x1n = lim sin 2nπ = lim 0 = 0 and lim sin y1n =
Example 1.10 The function f is defined on [0, 3] thus: lim sin(2nπ + π2 ) = lim 1 = 1.

x
2
if 0 ≤ x < 1 1
Example 1.12 Let f (x) = x sin( ) if x 6= 0 and f (0) = 0. As in Ex. 4), it

f (x) = x + 1 if 1 ≤ x ≤ 2 x
6
 is clear that f is continuous at every point a where a 6= 0. Further, f is con-
x 2<x≤3 tinuous at 0 also. For this, let F (x) = x, G(x) = sin(1/x), for x 6= 0. Then
limx→0 F (x) = 0 and G(x)| ≤ 1. Hence
Discuss the continuity of the function f on [0, 3].
Note that f (x) = x2 is a polynomial, hence continuous on [0, 1); similarly, lim f (x) = lim F (x)G(x) = 0,
f (x) = x + 1 is a polynomial, hence continuous on (1, 2). Further, f (x) = 6/x is x→0 x→0
a rational function whose denominator, x, is non-zero on (2, 3), hence continuous
by Cor. to Theorem 5, of §1.2. Also, f (0) = 0 by data. Hence f is continuous at
on (2, 3). So, the only points where we have to test continuity are 1 and 2, since
0.
the definition of f changes at these points. Now f (x) = x2 when x is near 1 and
is less than 1. So Example 1.13 Let f : [0, 1] → R be defined by
lim f (x) = lim x2 = 1
x→1− x→1− (
x if x is rational
while f (x) = x + 1 when x is near 1 and is greater than 1. So f (x) =
1 − x if x is irrational.
lim f (x) = lim (x + 1) = 2.
x→1+ x→1+ We shall prove that f is continuous only at x = 12 . Let {xn } be any sequence
such that xn → 12 as n → ∞. By the definition of f, we have f (xn ) = xn or
Thus the one-sided limits of f at 1 are unequal. Hence lim f (x) does not exist so
x→1 f (xn ) = 1 − xn depending on whether xn is rational or irrational. So, in the first
that f is discontinuous at 1. case, | f (xn ) − f ( 12 ) |=| xn − 12 | and in the second case, | f (xn ) − f ( 21 ) |=
Similarly, |1 − xn − 12 | = |xn − 12 |. Hence in either case, we get
lim f (x) = lim (x + 1) = 3, lim f (x) = lim (6/x) = 3 1 1
x→2− x→2− x→2+ x→2+ | f (xn ) − f ( ) |=| xn − | .
2 2
so that the one-sided limits of f at 2 are equal with the common value 3. Hence
Thus
lim f (x) exists and is equal to 3. Also, f (2) = 2 + 1 = 3. So f is continuous at 1 1
x→2 xn → ⇒ f (xn ) → f ( ).
2. 2 2
1 So f is continuous at x = 12 . Now let a 6= 12 be any real number in [0, 1]. As seen
Example 1.11 Let f (x) = sin if x 6= 0 and f (0) = 0. Since sin x is continuous in Note 2 after Theorem 1 above, we know that there exists a sequence, say xn , of
x
at every point and 1/x is continuous if x 6= 0, it follows by Theorem 2 above, that rational numbers and a sequence, say yn , of irrational numbers in [0, 1] such that
Calculus 23 24 Dr. V. V. Acharya

xn → a and yn → a as n → ∞. Suppose f is continuous at a. Then as n → ∞, i.e. f (N (a, δ)) ⊂ (f (a) − ǫ, f (a) + ǫ). Hence f (x) > 0, ∀x ∈ N (a, δ). Thus
by theorem 3 above, we see that f (x) and f (a) have same sign for all values of x in N (a, δ).
f (a) = lim f (xn ) = lim xn = a, and Exercise Set 1.2
f (a) = lim f (yn ) = lim(1 − yn ) = 1 − a.
1. Test the continuity of the following function at 0 :
Thus we have, a = f (a) = 1 − a so that a = 1/2, a contradiction. Thus f is not e1/x − e−1/x
continuous at any point except for x = 12 . f (x) = , if x 6= 0 and f (0) = 1.
e1/x + e−1/x
e1/x − 1 2. Discuss the continuity of f at x = 0 if
Example 1.14 Discuss the continuity of f at x = 0 if f (x) = , if x 6= 0
e1/x + 1 xe1/x
and f (0) = 0. f (x) = , if x 6= 0, f (0) = 0.
e1/x+1
It is clear that f is continuous at every point other than 0. At 0, observe that
limx→0+ x1 = ∞ and so lim e1/x = ∞, as e > 1. Hence e−1/x → 0 as x → 0+ 3. Let f : R → R be defined by
x→0+
and so (
1 − e−1/x 0 if x is rational
f (0+) = lim f (x) = lim = 1. f (x) =
x→0+ x→0+ 1 + e−1/x 1 if x is irrational.
But limx→0− 1
x = −∞ and so lim e1/x = 0 so that
x→0− Prove that f is discontinuous at every point of R.
e1/x − 1 0−1 4. Test the continuity of the following function at x = 3 :
f (0−) = lim f (x) = lim = = −1.
x→0− x→0− e1/x + 1 0+1 (
|x−3|
So f (0+) 6= f (0−) i.e. lim f (x) does not exist. Hence f is not continuous at 0. x−3 if x 6= 3
x→0
f (x) =
0 if x = 3
Now we note two important properties of continuous functions.
1 1
5. Test the continuity of f at x = 0 if f (x) = sin , if x 6= 0 and f (0) = 0.
Theorem 1.12 (i) If f and g are continuous functions on R such that f (x) = g(x) x x
for every rational x, then f (x) = g(x), for all x ∈ R. 6. Define the functions f and g on R as follows:
(ii) Suppose f is continuous at a point a, and f (a) 6= 0, then there exists a δ > 0
such that f (x) and f (a) have same sign for all values of x in N (a, δ). ( (
Proof : (i) Let a be an irrational number. To prove that f (a) = g(a). As seen in x if x is rational x if x is rational
f (x) = g(x) =
Note 2 after Theorem 1 above, we know that there exists a sequence say, xn of 0 if x is irrational. −x if x is irrational.
rational numbers such that xn → a as n → ∞. As f and g are continuous at a,
we get, by theorem 3 above, that f (xn ) → f (a) and g(xn ) → g(a) as n → ∞. Prove that each of f and g is continuous only at x = 0.
But since xn are rational numbers, f (xn ) = g(xn ), ∀ n, as n → ∞ we get 7. Suppose the function f is defined on R as follows:
f (a) = g(a).
 x−|x|
(ii) Suppose f (a) > 0. Choose ǫ > 0 such that f (a) − ǫ > 0 [ we may take if x 6= 0
f (x) = x
ǫ = f (a)
2 ]. By continuity at a, there exists a δ > 0 such that 2 if x = 0.
| x − a |< δ ⇒| f (x) − f (a) |< ǫ. Test the continuity of f on R.
Calculus 25 26 Dr. V. V. Acharya

8. Suppose the function f is defined on [0, ∞) as follows: b, f (b) = a. Then clearly f has intermediate value property but f is discontinuous
( √ at a and b.
2 x−1
if x 6= 14 (2) Note that the Intermediate Value Property (I.V.P.) essentially means that a con-
f (x) = 4x−1
0 if x = 14 . tinuous function on an interval takes every intermediate value between any two
values it takes. Thus if f is a continuous function on an interval I, then f (I) i.e.
Test the continuity of f on [0, ∞). range of f is also an interval. But the reader should note in the following examples
that, in general, the interval f (I) and I may not be the intervals of same type.
9. Show that the following function defined on [0, ∞) is discontinuous at a
point a if and only if a is a positive integer:

5 if 0 ≤ x < 1
f (x) =
5 + 2n if n ≤ x < n + 1, n ∈ N .

10. Let f, g be real valued functions on R. Define the functions max(f, g), and
min(f, g) as follows :

max(f, g)(x) = max{f (x), g(x)}, min(f, g)(x) = min{f (x), g(x)}.

If f and g are continuous at a point x0 , then prove that both the functions
max(f, g) and min(f, g) are also continuous at x0 .
Fig.3
1.3 Continuous Functions on an Interval
Example 1.15 If f is a constant function f (x) = c on R, then certainly f con-
In this section we state (without proof), extremely important properties of con- tinuous everywhere and the range of f is the singleton set {c} which is also a
tinuous functions defined on special type of subsets of R, called intervals . We degenerate interval [c, c].
know that a subset I of R is called an interval if for any x, y ∈ I, if z is such that
x < z < y, then z ∈ I. Let a, b ∈ R. Then the bounded intervals are of the form : Example 1.16 Let I = (0, 1) and f : I → R be defined by f (x) = x1 . We
know that f is a continuous function on (0, 1). Further, for any real number a >
(a, b), (a, b], [a, b), [a, b]. 1, a1 ∈ (0, 1) and f ( a1 ) = a. Thus f ((0, 1)) = (1, ∞). Observe that the image of
a bounded interval is an unbounded interval.
The unbounded intervals are of the form:
Example 1.17 Let I = (−1, 1] and f : I → R be defined by f (x) =| x | . We
(a, ∞), [a, ∞), (−∞, a), (−∞, a]. know that f is continuous on I. Clearly, f ((−1, 1]) = [0, 1].

Theorem 5 (Intermediate Value Property ): Let I be an interval and f : I → R Example 1.18 Let f : R → R be defined by f (x) = sin x. We have assumed
a continuous function. If a, b ∈ I and y0 is any real number such that f (a) < that f is continuous on R. Note that the range of f is [−1, 1]. i.e. the image of the
y0 < f (b), then there is at least one point x0 ∈ (a, b) such that f (x0 ) = y0 . (See entire real line under f is a closed and bounded interval.
Fig. 3)
x
Example 1.19 Consider the function f (x) = defined on R. Observe
Remark : (1) Converse of the intermediate value property is not true. For example 1+ | x |
let f be a function defined on [a, b] as follows: f (x) = x on (a, b), f (a) = that | f (x) |< 1 and f (0) = 0. Also x and f (x) have same sign. Further, for
Calculus 27 28 Dr. V. V. Acharya

x y
0 < y < 1, we solve the equation = y for x to get x = . Thus for To prove 2), we argue by contradiction. If there exists no x ∈ [a, b] such that
1+x 1−y f (x) = M then M − f (x) is continuous at each x ∈ [a, b] and M − f (x) > 0 for
any y ∈ (0, 1), we get an x ∈ (0, ∞) such that f (x) = y. Similarly we can show
all x ∈ [a, b]. If we let g(x) := 1/(M − f (x) for x ∈ [a, b],, then g is continuous
that every y ∈ (−1, 0) is also in the range of f. Thus the range of f is (−1, 1).
on [a, b]. By 1), there exists A > 0 such that g(x) ≤ A for all x ∈ [a, b]. But
Note that sup f on R is 1, and inf f on R is −1 but the points 1, −1 are not in the 1 1
then we have, for all x ∈ [a, b], g(x) := M−f (x) ≤ A or M − f (x) ≥ A .
range of the function f. So neither supremum nor infimum is attained.
Thus we conclude that f (x) ≤ M − (1/A) for x ∈ [a, b]. This contradicts our
Remark: From the examples above, it is clear that even if f is a continuous hypothesis that M = sup{f (x) : x ∈ [a, b]}. We therefore conclude that there
function defined on an interval, we cannot guaranty that range of the function will exists c ∈ [a, b] such that f (c) = M . Arguing similarly we can find a d ∈ [a, b]
be a bounded interval. Further if the range is bounded we cannot guaranty that f such that f (d) = m.
will attain its supremum and infimum. In view of this, it is very important to note
the following property of the continuous functions where the domain is not just an Theorem 1.14 (Intermediate Value Theorem) Let f : [a, b] ⊂ R → R be con-
interval but a closed and bounded interval. In this special case we get that the tinuous. Assume that f (a) < 0 < f (b). Then there exists c ∈ (a, b) such that
range is again a closed and bounded interval. f (c) = 0.
Proof. Draw some pictures. We wish to locate the “first” c from a such that
Theorem 1.13 Let f : [a, b] ⊂ R → R be continuous. Then
f (c) = 0. Towards this end, we define
1. f is bounded.
E := {x ∈ [a, b] : f (y) ≤ 0 for y ∈ [a, x]}.
2. Let M := sup{f (x) : x ∈ [a, b]} and m := inf{f (x) : x ∈ [a, b]}. Then
there exist points c and d in [a, b] such that f (c) = M and f (d) = m. For ε = −f (a)/2, by continuity of f at a there exists δ > 0 such that f (x) ∈
Proof. Let (3f (a)/2, f (a)/2) for all x ∈ [a, a + δ). This shows that a + δ/2 ∈ E. Since E is
E := {x ∈ [a, b] : f is bounded on [a, x]}. bounded by b there is c ∈ R such that c = sup E. Clearly we have a+δ/2 ≤ c ≤ b
and hence c ∈ [a, b]. We claim that c ∈ E and that f (c) = 0.
We want to prove that b ∈ E. Since f is continuous at a, given ε = 1, there exists
Since c − 1/n is not an upper bound for E there is an xn ∈ E such that
a δ0 > 0 such that f (x) ∈ (f (a)− 1, f (a)+ 1) for all x ∈ [a, a+ δ0 ). Thus we see
c − 1/n < xn ≤ c. By sandwich lemma, lim xn = c. By continuity of f at c
that |f (x)| ≤ |f (a)| + 1 for x ∈ [a, a + δ0 /2]. Hence a + δ0 /2 ∈ E. Obviously
we have f (xn ) → f (c). As f (xn ) ≤ 0 for all n we conclude that f (c) ≤ 0. If
E is bounded by b. Let c = sup E. Since a + δ0 /2 ∈ E we have a ≤ c. Since b
f (c) 6= 0 then f (c) < 0. Arguing as in the first part of the proof, we can find an
is an upper bound for E, c ≤ b. Thus a ≤ c ≤ b. We intend to show that c ∈ E
η > 0 such that for x ∈ (c − η, c + η) we have f (x) ∈ (3f (c)/2, f (c)/2). As
and c = b. This will complete the proof.
lim xn = c, there is an N such that xN ∈ (c − η, c + η). But then we see that
For any n ∈ N, c − 1/n is not an upper bound for E. Therefore there is an
f (x) < 0 for x ∈ [a, xN ] ∪ (c − η, c + η/2]. Hence c + η/2 ∈ E. This contradicts
xn ∈ E such that c − 1/n < xn ≤ c. Since f is continuous at c, for ε = 1 there is
the fact that c = sup E. Hence we conclude that f (c) = 0.
a δ > 0 such that f (x) ∈ (f (c) − 1, f (c) + 1) for all x ∈ (c − δ, c + δ) ∩ [a, b]. By
Sandwich lemma, xn → c. But there exists an N ∈ N such that xN ∈ (c−δ, c+δ).
Since xN ∈ E there is an M such that |f (x)| ≤ M for x ∈ [a, xN ]. Also f is 1.4 Applications of Continuity
bounded by |f (c)| + 1 on (c − δ, c + δ) ∩ [a, b]. From these two facts we conclude
that In this section, we prove some important consequences of the properties continu-
1 ous functions.
|f (x)| ≤ max{M, |f (c)| + 1}, for all x ∈ [a, c + ] ∩ [a, b].
2N
1 Theorem 1.15 Let I be an interval and f : I → R be a continuous function. If
This shows that c ∈ E. Note that the above argument shows also that c+ ∈E there are points a and b in I such that f (a) and f (b) are of opposite signs, then
2N
if c 6= b. This contradicts the fact that c = sup E. Hence c = b. This proves 1). there exists a point c ∈ (a, b) such that f (c) = 0.
Calculus 29 30 Dr. V. V. Acharya

Proof : This follows immediately from the Intermediate Value Property. [This yields that f is decreasing]. Let x be any point such that a < x < b. To prove
means that if the points (a, f (a)) and (b, f (b)) are on opposite sides of the X- that f is increasing, it is sufficient to prove that f (a) < f (x) < f (b). Suppose
axis, then the graph of f must cut X- axis in at least one point c ]. See Fig.4. that this inequlity is not true. Then there are two possibilities: either f (x) ≤ f (a)
or f (x) ≥ f (b). But f (x) ≤ f (a) actually implies that f (x) < f (a), because f
is injective. Then we have f (x) < f (a) < f (b). By intermediate value property,
there exists a point c ∈ (x, b) such that f (c) = f (a), a contradiction to the
injectivity of f. By a similar argument, we arrive at a contradiction if we assume
that f (x) ≥ f (b). Hence we get the inequality : f (a) < f (x) < f (b).
Remark: Suppose f is an injective continuous function on an interval I. Suppose
range of f is J. By Intermediate Value Property, we know that J is also an inter-
val. So, f : I → J is a bijective continuous function. Thus f −1 is a function
on the interval J onto the interval I. The natural question here is, whether f −1 a
continuous function ? Note that, by Theorem 3 above, f must be a strictly mono-
tonic function. Hence f −1 is also strictly monotonic. To see this, let for example,
f be strictly increasing on I. Let x1 , x2 be in I, and let y1 = f (x1 ), y2 = f (x2 )
so that x1 = f −1 (y1 ) and x2 = f −1 (y2 ). Then y1 < y2 ⇒ x1 = f −1 (y1 ) <
x2 = f −1 (y2 ) because x1 ≥ x2 ⇒ f (x1 ) ≥ f (x2 ) ⇒ y1 ≥ y2 . Hence f −1 is a
strictly monotonic function on an interval onto an interval. The continuity of such
a function is asserted in the following theorem:
Fig.4 Fig.5
Theorem 1.18 If g is a strictly monotonic function defined on an interval J such
that the range of g is again an interval, then g is continuous.
Theorem 1.16 (Fixed Point Property) If f : [a, b] → [a, b] is continuous then
Theorem 1.19 (Inverse Function Theorem) Suppose f is an injective continu-
there exists a point c ∈ [a, b] such that f (c) = c. [A point c is called a fixed point
ous function defined on an interval I. If the range of f is the interval J, then
of f if f (c) = c.] See Fig.5.
f −1 : J → I is continuous.
Proof: Suppose g(x) = f (x) − x, ∀x ∈ [a, b]. Then g is continuous on [a, b]
Proof: By Theorem 3, f is monotonic. Then f −1 is also monotonic. Now by
and since a ≤ f (x) ≤ b, ∀x ∈ [a, b], we have g(a) ≥ 0 and g(b) ≤ 0. Hence
Theorem 4, f −1 is continuous.
g(a) = 0 gives f (a) = a and g(b) = 0 gives f (b) = b. If g(a) > 0 and g(b) < 0,
then by Theorem 1 above, there exists c ∈ (a, b) such that g(c) = 0, i.e. f (c) = c. The following theorem is an immediate application of the inverse function theo-
Hence in all cases, f has a fixed point in [a, b]. rem:

Recall that a function f on an interval I is strictly increasing if for all points Theorem 1.20 (Existence of nth root) Let n be a natural number. For every pos-
x1 , x2 in I, we have x1 < x2 ⇒ f (x1 ) < f (x2 ). itive real number a, there is a unique positive real number b such that bn = a.
Proof: Let f (x) = xn , ∀ x ≥ 0. Then f is continuous and strictly increasing on
Theorem 1.17 Let I be an interval and f : I → R be a continuous function. If f [0, ∞) with range [0, ∞). Hence by Theorem 5, the inverse of f is continuous and
is injective (i.e. one-to-one) then it is strictly monotonic i.e. either f is a strictly strictly increasing function defined on [0, ∞) i.e. given a ≥ 0, ∃ a unique b ≥ 0
increasing function or f is a strictly decreasing function. such that f −1 (a) = b. i.e. f (b) = bn = a.
Proof: Let a, b be arbitrary points of I such that a < b. Suppose f (a) < f (b). We conclude this section with an important application of the Intermediate Value
Now we shall prove that f is increasing [If f (a) > f (b), then a similar argument Property.
Calculus 31 32 Dr. V. V. Acharya

Theorem 1.21 A polynomial of odd degree with real coefficients has a real root. 1. Let f be a real-valued continuous function on (a, b). Show that if f (r) = 0
for each rational number in (a, b), then f (x) = 0 for all x ∈ (a, b).
Proof : Without loss, we may assume that the polynomial is monic i.e. the leading
coefficient is 1. Let 2. Find two consecutive integers n, n + 1 between which a real root of x3 +
x2 = 3 lies.
P (x) = a0 + a1 x + a2 x2 + · · · + xn
3. Prove that x = cos x for some x ∈ (0, π2 ).
be any polynomial where the coefficients ai are real and the degree n is an odd
positive integer. To prove that ∃ x0 ∈ R such that P (x0 ) = 0. The polynomial 4. Show that the equation 3x = 2x is satisfied by some x in (0, 1).
P (x) can be written as :
5. Show that there is at least one real number x between 0 and π2 such that
2
a0 + a1 x + a2 x + · · · + an−1 x n−1 cos x = x − 1.
P (x) = xn [1 + ].
xn (
x if x is rational
6. Show that the function g defined by g(x) =
a0 + a1 x + a2 x2 + · · · + an−1 xn−1 1 − x if x is irrational.
we denote the quotient by f (x), so that
xn maps [0, 1] onto [0, 1] but it is continuous only at x = 12 .
P (x) = xn [1 + f (x)]. (1) 7. Suppose that f is continuous on [0, 2] and f (0) = f (2). Prove that there
exist x, y in [0, 2] such that | y − x |= 1 and f (x) = f (y).
Now let M = 1+ | a0 | + | a1 | + · · · + | an−1 | . Then
8. Let A be a nonempty subset of R, and define f : R → R by the formula
| a0 + a1 x + a2 x2 + · · · + an−1 xn−1 |
| f (x) | =
| xn | f (x) = inf{| x − a |: a ∈ A}.
| a0 | + | a1 | | x | + · · · + + | an−1 | | xn−1 |
≤ Prove that f is a continuous function.
| xn |
9. Prove that
If | x |> M, then
(a) There is no continuous onto function f : [−1, 1] → (−1, 1).
[| a0 | + | a1 | + · · · + | an−1 |] | xn−1 |
| f (x) | ≤ (b) There exists a continuous one-one, onto function F : R → (−1, 1).
| xn |
n−1
Find F −1 and verify that F −1 is continuous.
|x||x |
< n (c) There exists a continuous onto function g : R → [−1, 1].
|x |
(d) There exists a continuous onto function G : (−1, 1) → [−1, 1].
Thus if
| x |> M, then | f (x) |< 1.
1.5 Uniform Continuity
Hence by equation (1), we get that if x > M, then P (x) > 0 and since n is an odd
integer, if x < −M, then xn < 0 and so P (x) < 0. Now by Intermediate Value Let f be a real-valued function on a subset S of R. Recall that by definition, f is
Property, there exists x0 ∈ [−M, M ] such that P (x0 ) = 0. Hence the polynomial continuous at a point x ∈ S if for every ǫ > 0, there exists a number δ > 0 such
P (x) has a real root. that

Exercise Set 1.3 for all y ∈ S, |x − y| < δ ⇒ |f (x) − f (y)| < ǫ. (1)
Calculus 33 34 Dr. V. V. Acharya


Further, f is continuous on S if f is continuous at every point in S. Example 1.22 Let F (x) = + x for x ∈ [0, ∞). Then F is uniformly continuous
So if f is continuous at x, then a number δ satisfying condition (1) exists. on [0, ∞). For this, note that
In general, this number δ depends on both the number ǫ and the point x under √ √ p
for all x, y ≥ 0, | x − y| ≤ |x − y|. (i)
consideration. We indicate this dependence by writing δ = δ(ǫ, x). Thus with the
same ǫ, continuity of f at another point x′ ∈ S requires that there is a number Hence given ǫ > 0, take δ = ǫ2 . Then for all x, y ≥ 0, using (i) we see that
δ(ǫ, x′ ) > 0 such that √ √ p √
|x − y| < δ ⇒ |F (x) − F (y)| = | x − y| ≤ |x − y| < δ = ǫ,
for all y ∈ S, |x′ − y| < δ(ǫ, x′ ) ⇒ |f (x′ ) − f (y)| < ǫ, (2)
as required.
and δ(ǫ, x′ ) may have to be different from δ(ǫ, x). Hence we must use the smaller
of the numbers δ(ǫ, x) and δ(ǫ, x′ ), say δ0 , which will “work” for both x and x′ Note that the property of uniform continuity of a function is defined on a set and
i.e. for which the following will hold: not at a point. Also, it is clear that if a function f is uniformly continuous on a set
S, then f is uniformly continuous on any subset T of S.
for all y ∈ S, |x − y| < δ0 and |x′ − y| < δ0
⇒ |f (x) − f (y)| < ǫ and |f (x′ ) − f (y)| < ǫ. By negating the property stated in the definition above, we see that a function
f is not uniformly continuous on a subset T of its domain if and only if there
Example 1.20 Let f (x) = x1 for x ∈ (0, 1). Then clearly, f is continuous on exists a number ǫ′ > 0 such that for every δ > 0, there exist points x, y in T such
(0, 1). So suppose ǫ > 0 is given and let 0 < x′ < x < 1. Then conditions that
(1),(2) hold for x and x′ corresponding to some positive numbers δ = δ(ǫ, x) and
δ ′ = δ(ǫ, x′ ). But δ ′ has to be smaller than δ. |x − y| < δ but |f (x) − f (y)| ≥ ǫ′ . (4)
Coming back to the general case, we note that the existence of a number δ for
which condition (1) holds for all points x in S, requires f to possess a property From this we get a test for f to be non-uniformly continuous on T as follows: For
each positive integer n, taking δ = 1/n, by (4) there exist points xn , yn in T such
which is stronger than continuity, namely uniform continuity on S.
that
1
Definition 1.5 Let f be a real-valued function on a subset S of R. We say that f |xn − yn | < and |f (xn ) − f (yn )| ≥ ǫ′ .
is uniformly continuous on S if for every ǫ > 0, there exists a number δ > 0 such n
that That is, there exist sequences {xn } and {yn } of points in T such that

for all x, y ∈ S, |x − y| < δ ⇒ |f (x) − f (y)| < ǫ. (3) lim (xn − yn ) = 0 and |f (xn ) − f (yn )| ≥ ǫ′ for all n ≥ 1. (5)
n→∞
Here the number δ depends only on the number ǫ.
Consider the function f (x) = 1/x on (0, 1). Let us show that f is not uni-
Example 1.21 Let g(x) = x2 for x ∈ [0, 1]. Then g is uniformly continuous on formly continuous on (0, 1). As indicated in Ex.1, we know that the numbers
[0, 1]. In fact, δ(x) decrease as x approaches 0. Hence we choose ǫ′ = 1 and use the sequences
xn = 1/n and yn = 1/2n in (0, 1). Then
x, y ∈ [0, 1] ⇒ |g(x) − g(y)| = |x2 − y 2 | = |x + y| · |x − y|
≤ |(|x| + |y|) · |x − y| ≤ (1 + 1)|x − y| = 2|x − y|. lim (xn − yn ) = lim (1/2n) = 0
n→∞ n→∞

Hence given ǫ > 0, take δ = 12 ǫ. Then for all x, y ∈ [0, 1], and 1 1
|f (xn ) − f (yn )| = − = n ≥ 1 = ǫ′ , for all n,

|x − y| < δ ⇒ |g(x) − g(y)| ≤ 2|x − y| < 2δ = ǫ, xn yn
as required. so that f is not uniformly continuous on (0, 1).
Calculus 35 36 Dr. V. V. Acharya

Problem. Prove that the following functions are not uniformly continuous: To prove this, note that since f ′ is bounded on I, there is a constant M > 0
(i) x2 on [0, ∞) (ii) sin(1/x) on (0, 1) (iii) x sin x on (0, ∞). such that |f ′ (x)| ≤ M for all x ∈ I. Take any x, y in I with x < y. Now, f is
derivable on [x, y] and so by the mean value theorem there is c ∈ (x, y) such that
Clearly, if f is uniformly continuous on S, then f is certainly continuous on S.
But the converse is not true: f (x) = 1/x is continuous on (0, 1) but f is not f (y) − f (x) = f ′ (c)(y − x).
uniformly continuous on (0, 1). But the following important result can be proved:
So |f (x) − f (y)| = |x − y| · |f ′ (c)| ≤ M |x − y|.
Theorem 1.22 Let f be a real-valued function which is continuous on a closed
Hence given ǫ > 0, we may take δ = ǫ/M. Then
and bounded interval [a, b]. Then f is uniformly continuous on [a, b].
Proof. Since f : [a, b] → R is a continuous function, given ǫ > 0 for every for all x, y ∈ I, |x − y| < δ ⇒ |f (x) − f (y)| ≤ M |x − y| < M δ = ǫ,
x0 ∈ [a, b], there exists δx0 > 0 such that
as required.
ǫ
|f (x) − f (x0 )| < whenever |x − x0 | < δx0 .
2 The condition in the above proposition is not necessary for f √to be uniformly
  continuous. Thus as shown in Ex.2, the function F (x) = + x is uniformly
δx 0 δx 0
Thus, x0 − , x0 + is an open cover of compact set [a, b]. Hence, there continuous on√[0, ∞) and hence on I = (0, ∞). But F ′ is not bounded on I since
2 2 F ′ (x) = 1/2 x → ∞ as x → 0.
exists finitely many sub-intervals, say
    We can now describe the geometric meaning of uniform continuity. Ex.4 says
δx δx δx δx that if the function f has bounded derivative, that is if the tangents to the graph
x1 − 1 , x1 + 1 , . . . , xn − n , x0 + n
2 2 2 2 of f are not too steep, then f is uniformly continuous. But this condition is too
strong. To see this, note that the graph of the square-root function, shown in Fig. 2
such that their union covers [a, b]. Now, let δ = min  {δx1 , . . . , δxn } .  above, has arbitrarily steep chords near the origin and yet the function is uniformly
δx δx continuous on (0, ∞).
If x ∈ [a, b] there exists xi such that x ∈ xi − i , xi + i . Thus,
2 2 The following result shows that the correct condition is that, for the graph of
δx j δx j δx j f, Steep chords are short.
|x − xj | < . If |x − t| < δ then |t − x| < and |x − xj | < . Hence,
2 2 2
Theorem 1.23 For a real function f on an interval I, the following conditions are
|t − xj | ≤ |t − x| + |x − xj | < δxj . equivalent :
(i) f is uniformly continuous on I i.e. for every ǫ > 0, there is δ > 0 such that
Hence,
|f (t) − f (x)| < ǫ whenever |x − t| < δ. for all x, y ∈ I, | x − y |< δ ⇒| f (x) − f (y) |< ǫ.

Hence, f is uniformly continuous function. (ii) For every ǫ > 0, there is N > 0 such that
f (x) − f (y)
It follows from the above theorem that a polynomial p(x) is uniformly continuous

for all x, y ∈ I, > N ⇒| f (x) − f (y) |< ǫ.

on any closed and bounded interval. x−y
The following result gives another sufficient condition for a function to be
uniformly continuous: [Ref. D. Paine, Visualizing uniform continuity, Amer. Math. Monthly, 1968, p.
44-45.]
Proposition 1.1 Let f be a function defined and derivable on an interval I and Proof : First assume (i) and let N = 2ǫ/δ. We shall prove the contrapositive of
suppose the derivative f ′ is bounded on I. Then f is uniformly continuous on I. (ii). Note that if | f (x) − f (y) |≥ ǫ, then | f (x) − f (y) |= kη for some positive
Calculus 37 38 Dr. V. V. Acharya

integer k and some η in [ǫ, 2ǫ). To see this, let a =| f (x) − f (y) | and k = [a/ǫ] 3. Suppose that functions f, g are uniformly continuous on S ⊆ R. Show that
so that 1 ≤ k ≤ a/ǫ < k + 1 ≤ 2k. Thus ǫ ≤ a/k < 2ǫ. So if η = a/k, then f + g, f − g, and kf (k, a constant) are uniformly continuous on S. Give an
ǫ ≤ η < 2ǫ, as stated. example to show that the product f g may not be uniformly continuous on
We can assume w.l.g. that f (x) < f (y). First let x < y. Then since f is S. Show that if f, g are both bounded on S, then the product f g is uniformly
uniformly continuous on I, f is continuous on I and hence on [x, y] ⊆ I. Also, continuous on S.
[x, y] is a closed and bounded interval. Therefore, since
4. Suppose that functions f, g are uniformly continuous on R. Show that the
f (x) < f (x) + η < f (x) + 2η < · · · < f (x) + kη = f (y), composite function f ◦ g is uniformly continuous on R.

the intermediate value property of continuous functions implies that there are 5. Suppose that function f is uniformly continuous on a bounded subset S of
numbers R. Show that f is bounded on S.
x = x0 < x1 < x2 < · · · < xk = y
Answers 1.1
such that f (xi ) = f (x) + iη for each i. If x > y, then the numbers are

x = x0 > x1 > x2 > · · · > xk = y. 1. (i) Here f (x) = 5x+4 near x = 1. Let ǫ > 0 be given. Then we want to find
the condition on x which gaurantees that |f (x)−9| < ǫ. Now |f (x)−9| < ǫ
In either case, for each i = 1, . . . , k, we have if |5x + 4 − 9| < ǫ i.e. if |5x − 5| < ǫ or |x − 1| < ǫ/4. Hence we may take
δ = ǫ/4. Thus
| f (xi ) − f (xi−1 ) |= η ≥ ǫ.
|x − 1| < δ ⇒ |f (x) − 9| = 5|x − 1| < 5δ = ǫ,
So by (i), for each i = 1, . . . , k, we have |xi − xi−1 | ≥ δ. Consequently, | x−y |≥
kδ and so that f (x) → 9 as x → 1.
1 1 Also, by Theorem 4, we have
| f (x) − f (y) |= kη and ≤ imply that
|x−y | kδ
f (x) − f (y) kη η 2ǫ lim (5x + 4) = 5 lim x + lim 4 = 5(1) + 4 = 9.
≤ = ≤ = N. x→1 x→1 x→1

x−y kδ δ δ

(ii) We first factorize the numerator and cancel the factor x − 2. Then apply
Conversely, assume (ii) and let δ = ǫ/N. To prove the contrapositive of (i), we Theorem 4 to obtain the result:
observe that if | f (x) − f (y) |≥ ǫ, then
x2 − 3x + 2 (x − 1)(x − 2)
f (x) − f (y) x−y 1 lim = lim = lim (x − 1) = 2 − 1 = 1.

≤ N so that



and so x→2 x−2 x→2 x−2 x→2
x−y f (x) − f (y) N

x−y 1 (iii) Let ǫ > 0 be given. We wish to examine the behaviour of f (x) = x3

| x − y |= · | f (x) − f (y) |≥ · ǫ = δ.

f (x) − f (y) N for x near 1. So consider f on the iterval [0, 2]. Now |f (x) − 1| = |(x −
Problems 1)(x2 + x + 1)| = |x − 1| · |x2 + x + 1| ≤ 7|x − 1| because on the interval
[0, 2], x is non-negative and has maximium value 2. So taking δ = ǫ/7 we
1. Show that the function f (x) = 1/x is uniformly continuous on the set see that
S = [a, ∞) where a is a positive constant.
|x − 1| < δ ⇒ |f (x) − 1| ≤ 7|x − 1| < 7δ = ǫ,
2. Show that the function f (x) = 1/x2 is uniformly continuous on the set
S = [1, ∞), but not uniformly continuous on T = (0, ∞). so that f (x) → 1 as x → 1.
Calculus 39 40 Dr. V. V. Acharya

5x + 7 | 16x | Answers 1.2


(iv) Let f (x) = near x = 0. Then | f (x) − 7 |= . Since
3x + 1 | 3x + 1 | 1
3x + 1 → 1 as x → 0, we see by Theorem 3(ii) that there is a number δ > 0 1. Discontinuous because f (0−) 6= f (0+). In fact, lim = ∞ and so
x→0+ x
such that | 3x + 1 |> 12 whenever x ∈ (−δ, δ). Therefore
lim e1/x = ∞ as e > 1. Therefore, lim e−2/x = 0.
x→0+ x→0+
1 −2/x
< 2 whenever |x| < δ. 1−e
| 3x + 1 | Hence lim f (x) = lim = 1.
x→0+ x→0+ 1 + e−2/x
ǫ 1
Hence given ǫ > 0, let δ1 = min{δ, }. Then |x| < δ1 implies Also, lim = −∞ and so lim e1/x = 0 as e > 1. Therefore,
32 x→0− x x→0−
2/x
lim e = 0.
| 16x | x→0−
| f (x) − 7 |= < |16x| · 2 = 32|x| < 32δ1 ≤ ǫ.
| 3x + 1 | e2/x − 1
Hence lim f (x) = lim = −1.
x→0− x→0− e2/x + 1
2 2

2. (i) Given ∆ > 0, we have 1/x > ∆ if x < 1/∆ i.e. if |x| < 1/ ∆. 2. Continuous.
Hence lim(1/x2 ) = ∞ as x → 0.
3. Let a be any real number. Let {xn } be a sequence of rational numbers con-
(iii) Let f (x) = sin(1/x2 ) for x 6= 0. To show that limx→0 f (x) does verging to a and let {yn } be a sequence of irrational numbers converging
not exist, it is enough to show that there is at least one number ǫ′ > 0 to a. If f is continuous at a, then the sequences {f (xn )} and {f (yn )} con-
such that for every number δ > 0, we can find a pair of points x1 , x2 in verge to the same limit. But {f (xn )} is the constant sequence < 0, 0, . . . >
N ′ (a, δ) = (−δ, δ) − {0} such that and {f (yn )} is the constant sequence < 1, 1, . . . > . So these sequences
converge to the unequal limits 0 and 1. Hence f is discontinuous at a. So f
|f (x1 ) − f (x2 )| ≥ ǫ′ . is discontinuous at every real number since a is arbitrary.

See the discussion at the end of §1.2. In fact, let ǫ′ = 1. So given any 4. Discontinuous.
1 1
number δ > 0, choose a positive integer n such that √ < δ. Then the 5. Discontinuous. For, if xn = , then xn → 0, but
nπ (2n + 12 )π
1 1
points x1 = √ and x2 = p lie in (−δ, δ) − {0} and f (x1 ) = 1 1 1
nπ 2nπ + π2 f (xn ) = (2n + )π sin(2n + )π = (2n + )π → ∞.
π
sin(nπ) = 0 and f (x2 ) = sin(2nπ + 2 ) = 1 so that |f (x1 ) − f (x2 )| = 2 2 2
|0 − 1| = 1 = ǫ′ . Hence f is discontinuous at 0.
1 6. Use the method of Example 6) above. Thus if a 6= 0, choose a sequence
(iv) As in (i), show that lim √ = ∞.
x→0+ x {xn } of rational numbers and a sequence {yn } of irrational numbers such
(v) Here f (x) = e1/x if x 6= 0. Now as x → 0+, 1/x → ∞, hence that xn → a and yn → a as n → ∞. Then for all n, f (xn ) = xn while
e1/x → ∞ as e > 1. Therefore the the one-sided limit f (0+) does not f (yn ) = 0. So, as n → ∞, lim f (xn ) = lim xn = a 6= 0 = lim f (yn ).
exist. Hence lim f (x) does not exist. Hence f is discontinuous at a. But f is continuous at 0 because |f (x)| ≤ |x|
x→0 for all x, so that limx→0 f (x) = 0 = f (0).
3. Let F (x) = f (x) + g(x). Then if we assume that, as x → a, lim F (x) 7. Observe that f (x) = 0 if x > 0 and f (x) = 2 if x ≤ 0. Thus if a 6= 0, then
exists, then by Theorem 4 (i), lim [F (x) − f (x)] i.e. lim g(x) would exist, f is constant in a neighbourhood of a, hence f is continuous at a. It is clear
x→a x→a
contrary to the data. Hence lim F (x) does not exist. that lim f (x) = 0 and lim f (x) = 2. Thus f is discontinuous at 0.
x→a x→0+ x→0−
Calculus 41 42 Dr. V. V. Acharya

1
√ a ≥ 0. If a 6= 4 , then f is the ratio of the continuous functions F (x) =
8. Let 7. Consider the function g(x) = f (x + 1) − f (x) on [0, 1]. Then clearly, g is
2 x − 1, G(x) = 4x − 1 and G(a) 6= 0. Hence f is continuous at a. Note continuous on [0, 1]. Further, g(0) = f (1) − f (0) and g(1) = f (2) − f (1).
that Thus g(0) = −g(1). Hence either g(0) = g(1) = 0 or g(0) and g(1) are
√ of opposite sign. In either case, there is a point say y ∈ [0, 1] such that
2 x−1 1 1 g(y) = 0. Hence f (y + 1) = f (y) and the conclusion follows.
lim1 f (x) = lim1 = lim1 √ = 6= f (1/4).
x→ 4 x→ 4 4x − 1 x→ 4 2 x + 1 2
8. Let y be any point in R. We prove that f is continuous at y. Note that
Thus f is discontinuous at 1/4. | x − a |≤| x − y | + | y − a | where a ∈ A. Thus f (x) ≤| x − y | + |
x − a | for any a ∈ A. Thus f (x)− | x − y | is a lower bound for the set
9. Note that f is a constant function on each of the intervals [n, n + 1). Thus {| y −a |: a ∈ A}. Hence inf{| y −a |: a ∈ A} = f (y) ≥ f (x)− | x−y | .
f is continuous at every non-integer point of [0, ∞) as well as at 0. If a is a By symmetry, we have f (x) ≥ f (y)− | x − y | . Combining these two
positive integer, then lim f (x) = 5 + 2(a − 1) and lim f (x) = 5 + 2a. inequalities we get that | f (x) − f (y) |≤| x − y | . So given ǫ > 0, choose
x→a− x→a+
Thus f (a−) 6= f (a+) so that f is discontinuous at a. δ = ǫ, then | x − y |< δ ⇒| f (x) − f (y) |< ǫ. Thus f is continuous at y.

10. Now by algebra of continuous functions it is clear that max(f, g) is contin- 9. (a) Image of closed interval under continuous function is a closed interval.
uous at a point where both f and g are continuous. Note that min(f, g) = 2
(b) Consider the function F (x) = tan−1 x.
− max(f, g) so that min(f, g) is also continuous at a point where both f π
x
and g are continuous. Consider the function F (x) = .
1+ | x |
Answers 1.3 (c) Take g(x) = sin x.
(d) From (b) and (c) we get that the required function is G = g ◦ F −1 .
1. Consider the function g(x) defined by g(x) = 0 ∀x ∈ (a, b) and use Theo-
rem 4 of §1.4. √ √
Using Sandwich Principle, show that lim x = a, where a > 0.
x→a
2. The polynomial f (x) = x3 + x2 − 3 is continuous on the interval [1, 2] Solution. We note that
and f (1) = −1 < 0 and f (2) = 9 > 0. Hence by the Intermediate Value √ √

x − a = √ x − a 1

Property, f (x) = 0 has a root in [1, 2]. √
x + a ≤ √a |x − a| .

3. Consider the function f (x) = cos x − x. Then f (0) = 1 > 0 and f ( π2 ) =


Hence, the result follows.
− π2 < 0. So by the Intermediate Value Property, there is a point x ∈ (0, π2 )
such that f (x) = 0 i.e. cos x = x.

4. Note that the function f (x) = 2x −3x is continuous. Further, f (0) = 1 > 0
and f (1) = −1 < 0. So by the Intermediate Value Property, there is a point
x ∈ (0, 1) such that f (x) = 0 i.e. 2x = 3x.

5. If f (x) = cos x − x + 1, then f (0) = 2 > 0 and f ( π2 ) = −π/2 + 1 < 0.

6. We have already seen that the function g is continuous only at x = 12 .


Further, let y ∈ [0, 1]. Then if y is rational, we have g(y) = y. If y is
irrational, we have g(t) = 1 − t = y where t = 1 − y is in [0, 1]. Hence f
is onto [0, 1].
44 Dr. V. V. Acharya

In the above definition, we have assumed that the domain of f is an open


interval J. But now suppose that f is defined on a closed interval [c, d]. Then f

is said to be derivable at the left end point c if f+ (c) exists (and then we write
Chapter 2 ′ ′
f (c) = f+ (c)). Similarly, f is said to be derivable at the right end point d if

f− (d) exists (and then we write f ′ (d) = f− ′
(d)). Further, f is said to be derivable
Derivative on [c, d] if f is derivable at every point of [c, d].
Example 2.1 Let f (x) = |x|. Show that f ′ (0) does not exist. Find f ′ (a) if a 6= 0.
2.1 Introduction (ii) Let f (x) = x for x ≥ 0 and f (x) = 0 for x < 0. Show that f ′ (0) does not
exist.
In this chapter, we begin by recalling the definition of the derivative and some Solution: (i) Note that k = f (0 + h) − f (0) = |h| − 0 = |h|. Hence k = |h| = h,
results which reader is already familiar with. if h > 0 and k = |h| = −h, if h < 0. This gives

Definition 2.1 We begin by recalling the definition of the derivative. Let y = k h


lim = lim= lim (1) = 1,
f (x) be a function defined on an open interval J = (c, d) and let a be a point in hh→0+ h h→0+
h→0+
(c, d). Let a + h a point in (c, d), where h 6= 0. Then the derivative of f w.r.t. x k −h
at a is given by the limit and lim− = lim− = lim+ (−1) = −1.
h→0 h h→0 h h→0
h df i f (a + h) − f (a)
f ′ (a) = = lim , (1) Since these one-sided limits of k/h are unequal, we see that lim k/h does not
h→0
dx x=a h→0 h
exist. Hence f ′ (0) does not exist. Thus,the slope of the curve y = |x| is +1 for
if the limit exists. Also, f is said to be differentiable (or derivable) on (c, d) if f x > 0 and −1 for x < 0 and the slope f ′ (0) does not exist at x = 0 where (0, 0)
is differentiable at every point of (c, d). is a corner point on the curve. So at this point the curve y = |x| does not have
tangent line. Next, let a 6= 0. Then
It is sometimes more convenient to write (1) in the form
f (x) − f (a) |x| − |a|
f (x) − f (a) f ′ (a) = lim = lim
f ′ (a) = lim . (2) x→a x−a x→a x − a
x→a x−a
x2 − a2 x+a

We also consider the right derivative f+ ′
(a) and the left derivative f− (a) of f at = lim = lim .
x→a (x − a)(|x| + |a|) x→a (|x| + |a|)
a, defined as follows:
x+a a
′ f (x) − f (a) ′ f (x) − f (a) Hence if a 6= 0, then f ′ (a) = lim = .
f+ (a) = lim , f− (a) = lim (3) (|x| + |a|)
x→a |a|
x→a+ x−a x→a− x−a (ii) Here k = f (h) − f (0) = f (h). So k = h if h > 0 and k = 0 if h < 0. Hence
Recall that the limit in (2) exists if and only if both the limits in (3) exist and
k h k 0
are equal. That is, f ′ (a) exists if and only if f+
′ ′
(a) and f− (a) both exist and are lim = lim+ = 1, while lim− = lim− = 0.
h→0+ h h→0 h h→0 h h→0 h
equal. Also,
Since these one-sided limits of k/h are unequal, we see that lim k/h does not
f ′ (a) = l if and only if f+
′ ′
(a) = f− (a) = l. h→0
exist. Hence f ′ (0) does not exist. Hence,the curve y = f (x) does not have
Geometrically, the right and the left derivatives of f at a exist but are unequal tangent line at the origin.
means that there is a sharp corner in the graph of f at x = a. Thus at such a point
the graph of f does not have tangent line. Theorem 2.1 If f is differentiable at a point a, then f is continuous at a.

43
Calculus 45 46 Dr. V. V. Acharya

Proof : Suppose f is differentiable at a. Then the limit 2.2 Mean Value Theorems
f (x) − f (a) Theorem 2.4 (Rolle’s Theorem) Suppose a < b and let f be a real valued func-
lim
x→a x−a tion on [a, b]. Suppose
exists and is f ′ (a). Hence (i) f is continuous on the closed and bounded interval [a, b],
h f (x) − f (a) i (ii) f is derivable on the open interval (a, b) and
lim f (x) = lim · (x − a) + f (a) (iii) f (a) = f (b).
x→a x→a x−a Then there exists at least one point x0 ∈ (a, b) such that f ′ (x0 ) = 0.
f (x) − f (a) Proof : If f is a constant function on [a, b] then f ′ (x) = 0 for all x ∈ [a, b]. This
= lim · lim (x − a) + lim f (a)
x→a x−a x→a x→a shows that in this case points x0 certainly exist as required.
= f ′ (a) · 0 + f (a) = f (a). So let f be a non-constant function. By (i) f is continuous on the closed and
bounded interval [a, b]. Hence there exist points c, d ∈ [a, b] such that M = f (c)
Therefore f is continuous at a. is the maximum value and m = f (d) is the minimum value of f on [a, b]. As
Note: The converse of the above theorem is not true. That is, even if f is continu- f is non-constant, we have m < M. Then since by (iii) f (a), f (b) are equal,
ous at a point a, f may not be differentiable at a. Thus consider f (x) = |x|. Here the numbers m and M cannot both be equal to these equal values f (a), f (b). So
first let M = f (c) 6= f (a). Then c 6= a and c 6= b so that c ∈ (a, b). Hence
f is continuous at 0 because
by (ii) f is derivable at c i.e f ′ (c) exists. We now show that f ′ (c) = 0. Since
lim f (x) = lim |x| = 0 = |0| = f (0). f (x) ≤ M = f (c) for every x ∈ [a, b], we have
x→0 x→0

But, as seen in the earlier example above, f ′ (0) does not exist. f (x) − f (c)
≥0 for x < c (4)
We recall the following theorems which reader has already studied: x−c
f (x) − f (c)
Theorem 2.2 Suppose that functions f and g are derivable at a point a. Then the and ≤0 for x > c. (5)
x−c
functions F = f + g, G = f − g, H = f · g, ψ = kf (k = a constant) and
φ = f /g (provided g(a) 6= 0) are derivable at a, and further, Hence as x → c−, (1) gives f ′ (c) ≥ 0 and as x → c+, (2) gives f ′ (c) ≤ 0.
′ ′
Therefore since f− (c) = f+ (c) = f ′ (c), we get f ′ (c) = 0. Thus x0 = c is a point
F ′ (a) = f ′ (a) + g ′ (a), G′ (a) = f ′ (a) − g ′ (a),
as required. A similar proof shows that if m = f (d) 6= f (a), then x0 = d is a
H ′ (a) = f (a)g ′ (a) + f ′ (a)g(a), ψ ′ (a) = kf ′ (a), point as required.
g(a)f ′ (a) − f (a)g ′ (a)
φ′ (a) = , if g(a) 6= 0. The following version of Rolle’s theorem is useful. It is obtained by substituting
g(a)2
b = a + h.
Theorem 2.3 (Chain rule) Suppose y = g(u) and u = f (x) are differentiable
functions of u and x respectively, such that the composite function y = F (x) = Theorem 2.5 (Rolle’s Theorem, another avatar) Suppose h > 0 and let f be a
g(f (x)) is defined. Then the composite function F is differentiable and its deriva- real valued function on [a, a + h]. Suppose
tive w.r.t. x is given by (i) f is continuous on the closed and bounded interval [a, a + h],
(ii) f is derivable on the open interval (a, a + h) and
d d dg df
F (x) = g(f (x)) = · . (iii) f (a) = f (a + h).
dx dx du dx
Then there exists at least one point θ ∈ (0, 1) such that f ′ (a + θh) = 0.
This is usually written as
dy dy du Remark 2.1 The conditions (i) and (ii) of the above theorem are called Rolle’s
= · .
dx du dx conditions. If in a given situation, the function f happens to be derivable on the
Calculus 47 48 Dr. V. V. Acharya

full interval [a, b] including the end-points a, b, then f is certainly continuous on Example 2.4 Verify Rolle’s theorem for f (x) = x3 + 4x2 − 7x − 10 on the
[a, b] and therefore satisfies Rolle’s conditions on [a, b]. interval [−1, 2].
Remark 2.2 Geometrically Rolle’s theorem says the following : if f is a contin-
uous function on [a, b] and if the graph of f has a tangent at each point between 20
the points (a, f (a)) and (b, f (b)), which are on the same level, then there exists
atleast one point between a and b such that the tangent to the graph of f at the
point (c, f (c)) is parallel to x-axis. 15

Remark 2.3 If a < x < b, then x is called an interior point of [a, b]. In the
above proof of Rolle’s theorem what we have actually shown is the following: if 10
f reaches its maximum (or minimum) value at an interior point x0 of the domain
[a, b] of f and if f ′ (x0 ) exists, then f ′ (x0 ) = 0. It is very important to note that
f ′ (x0 ) may not be zero if x0 is an end point of [a, b]. For example, let f (x) = x 5
on [a, b]. Then f is clearly minimum at x = 0 and maximum at x = 1 but f ′ (0) =
f ′ (1) = 1 6= 0.
Remark 2.4 Clearly, any polynomial p satisfies Rolle’s conditions on any interval −5
[a, b]. Hence by Rolle’s theorem we see that if p(a) = 0 and p(b) = 0, then
p′ (c) = 0 for some c ∈ (a, b). That is, between any two (distinct) roots of p lies
−5
a root of the derivative p′ . [Of course, if x = α is a repeated root of the equation
p(x) = 0, then also we get p′ (α) = 0.]

Example 2.2 Let f (x) = 9 − x2 , x ∈ [−3, 3]. Then f is the composite of the −10

polynomial function F (x) = 9 − x2 and the root function G(y) = y, y ≥ 0.
Hence f = G ◦ F is continuous on [−3, 3]. Also, G is derivable for y > 0 and
so f is derivable on (−3, 3). Finally, f (−3) = f (3) = 0. Thus f satisfies all the
conditions of Rolle’s theorem. To verify the conclusion of Rolle’s theorem, we Figure 2.1:
have√ to find a point x0 ∈ (−3, 3) such that f ′ (x0 ) = 0. For this, note that f ′ (x) =
−x/ 9 − x2 and so f ′ (x) = 0 gives x = 0. Therefore since x0 = 0 ∈ (−3, 3),
it follows that the point x0 = 0 is as required. Solution. The given function is a polynomial and is therefore derivable every-
where and, in particular, on [−1, 2]. Also, f (−1) = −1 + 4 + 7 − 10 = 0, f (2) =
Example 2.3 Show that for any real number a, the equation x3 − 3x + a = 0 has 8 + 16 − 14 − 10 = 0. Thus f (−1) = f (2) and so f satisfies all the condi-
atmost one root in the interval [−1, 1]. tions of Rolle’s theorem. To verify the conclusion of Rolle’s theorem, note that
Solution. The function f (x) = x3 − 3x+ a is a polynomial and is therefore deriv- f ′ (x) = 3x2 + 8x − 7 and so f ′ (x) = 0 gives the points
able everywhere and, in particular, on [−1, 1]. Let, if possible, f (c) = f (d) = 0 √ √
where −1 < c ≤ d < 1. Now first let c < d. Then f satisfies all the conditions of −4 + 37 −4 − 37
x0 = , x1 = .
Rolle’s theorem on [c, d] and so f ′ (x0 ) = 0 for some point x0 ∈ (c, d) ⊂ (−1, 1). 3 3
But this is a contradiction because f ′ (x) = 0 i.e. 3x2 − 3 = 0 holds only when Now x1 < −1 and x0 ∈ (−1, 2) because
x = −1, 1. Secondly, if c ∈ (−1, 1) is a repeated root of f (x) = 0, then we get √
f ′ (c) = 0 which is again a contradiction. So the equation x3 − 3x + a = 0 has at 36 < 37 < 49 ⇒ 6 < 37 < 7

most one root in [−1, 1]. ⇒ 2 < −4 + 37 < 3 ⇒ 2/3 < x0 < 1.
Calculus 49 50 Dr. V. V. Acharya

Hence the point x0 is as required. In this form, equation (6) holds even when h < 0. This means that if f satisfies
Rolle’s conditions on the interval [a + h, a] where h < 0, then there exists a
The first mean value theorem (briefly LMVT) is as follows : number θ such that 0 < θ < 1 and
Theorem 2.6 (Lagrange) Suppose a < b and let f be a real-valued function on
[a, b] such that f (a) − f (a + h) = (−h)f ′ (a + θh).
(i) f is continuous on the closed and bounded interval [a, b] and √
(ii) f is derivable on the open interval (a, b). Example 2.5 Verify Lagrange’s mean value theorem for f (x) = 25 − x2 on
Then there exists atleast one point x0 ∈ (a, b) such that the interval [−3, 4].

f (b) − f (a) 5
= f ′ (x0 ). (6)
b−a
A
Proof : Let for a ≤ x ≤ b, we define function g(x) by b
4

g(x) = f (x) − kx, (7)


3 b

B
where k is a constant to be chosen so that g(a) = g(b). In fact, we require
2
f (a) − ka = f (b) − kb
f (b) − f (a)
or k = . (8) 1
b−a
Now by (i), (ii) and (4) we see that g is continuous on [a, b] and for every x ∈
(a, b), g ′ (x) = f ′ (x) − k. Hence g satisfies the conditions of Rolle’s theorem on −3 −2 −1 1 2 3
[a, b] and so there exists x0 ∈ (a, b) such that g ′ (x0 ) = 0 i. e. f ′ (x0 ) = k. Hence
by (5), (3) holds.

Remark 2.5 Observe that the expression [f (b) − f (a)]/(b − a) gives the “mean Figure 2.2:
value” of the slope of the graph of f on [a, b] and mean value theorem says that the
derivative f ′ must attain this mean value at some point in (a, b). Geometrically this Solution. The given function is derivable on (−5, 5) and hence on [−3, 4] also.
means that there exists a point x0 ∈ (a, b) such that the slope of the chord joining Hence f satisfies the conditions of Lagrange’s
√ mean value theorem. Now clearly,
the points A(a, f (a)) and B(b, f (b)) on the graph of f equals the slope of the f (−3) = 4, f (4) = 3, f ′ (x) = −x/ 25 − x2 . To verify the conclusion of this
tangent to the graph at some point P (x0 , f (x0 )). theorem, we have to find a point x ∈ (−3, 4) such that

Remark 2.6 Put b − a = h and (x0 − a)/h = θ so that h > 0, b = a + h, x0 = f (4) − f (−3) = [4 − (−3)]f ′ (x)
a + θh. Then −x
i.e. 3−4 = (7) √
25 − x2
a < x0 < b = a + h ⇒ 0 < x0 − a < h ⇒ 0 < θ < 1 as h > 0.
1
i.e. 25 − x2 = 49x2 or x2 = .
Hence the MVT can be stated equivalently as follows : If f satisfies Rolle’s con- 2
ditions on the interval [a, a + h], then there exists a number θ such that 0 < θ < 1
and This gives the solutions x = ± √12 . Of these, x0 = ± √12 lie in (−3, 4) and
f (a + h) − f (a) = hf ′ (a + θh). (9) therefore is as required.
Calculus 51 52 Dr. V. V. Acharya

Example 2.6 If 0 < a < b < 1, prove that Thus,


b−a b−a
√ < sin−1 b − sin−1 a < √ . 3/8 = (1/2)(3θ2 /4 − 3θ + 2)
1−a 2 1 − b2 √
√ 6 ± 21
π 3 3 π 1 i.e. 3θ2 − 12θ + 5 = 0 or θ = .
Hence show that + < sin−1 < + . 3
6 15 5 6 8 √
Solution. Since 0 < a < b < √ 1, the function f (x) = sin−1 x is derivable Of these, θ = 6− 21
lies in (0, 1) and therefore the required number. (In fact,
3
on [a, b], and in fact, f ′ (x) = 1/( 1 − x2 ). Hence f satisfies the conditions of

Lagrange’s mean value theorem and so there is a number c ∈ (a, b) such that √ 1 6 − 21
16 < 21 < 25 ⇒ 4 < 21 < 5 ⇒ < < 1.)
f (b) − f (a) = (b − a)f ′ (x) 3 3
1 We now give some applications of the MVT. The power of the MVT lies in
i.e. sin−1 b − sin−1 a = (b − a) √ . (10)
1 − c2 the fact that merely the existence, and not the actual value, of a point x0 satisfying
(3) needs to be used in most of the applications.
But 0 < a < c < b < 1 ⇒ 0 < a2 < c2 < b 2 < 1
⇒ 0 > −a2 > −c2 > −b2 > −1 Theorem 2.7 Suppose a < b and the function f is derivable on the open interval
2
⇒ 1>1−a >1−c >1−b >0 2 2 (a, b). Then
p p p (i) f ′ (x) 6= 0, ∀x ∈ (a, b) ⇒ f is a one − one function on (a, b).
⇒ 1 − a2 > 1 − c2 > 1 − b 2 > 0
(ii) f ′ (x) = 0, ∀x ∈ (a, b) ⇒ f is a constant function on (a, b).
1 1 1
⇒ √ <√ <√ (iii) f ′ (x) > 0, ∀x ∈ (a, b) ⇒ f is strictly increasing on (a, b).
1−a 2 1−c 2 1 − b2 (iv) f ′ (x) < 0, ∀x ∈ (a, b) ⇒ f is strictly decreasing on (a, b).
b−a b−a b−a Proof : Take any x, y such that a < x < y < b. Then by data, f is derivable and
⇒ √ <√ <√ (as b − a > 0)
1 − a2 1 − c2 1 − b2 hence continuous on [x, y]. Hence by the MVT, there is x0 ∈ (x, y) such that
b−a b−a
⇒ √ < sin−1 b − sin−1 a < √ (by (10)) f (y) − f (x) = f ′ (x0 )(y − x). (11)
1−a 2 1 − b2
Example 2.7 Find the number θ of Lagrange’s mean value theorem for f (x) = (i) Suppose f ′ (x) 6= 0, ∀x ∈ (a, b). Suppose f is not one-one in (a, b). Then there
x(x − 1)(x − 2) on the interval [0, 1/2]. exist x, y ∈ (a, b) such that x < y and f (x) = f (y). Now there is x0 ∈ (x, y)
Solution. The given function f (x) = x3 − 3x2 + 2x is derivable everywhere such that (7) holds. This gives f ′ (x0 ) = 0, since f (y) = f (x) and y − x 6= 0.
and hence on [0, 1/2] also. Hence f satisfies the conditions of Lagrange’s mean This is a contradiction.
value theorem. Here a = 0, a + h = 1/2 so that h = 1/2. Now clearly, f (0) = (ii) Suppose f ′ (x) = 0, ∀x ∈ (a, b). Suppose f is not constant on (a, b). Then
0, f (1/2) = 3/8, f ′ (x) = 3x2 − 6x + 2. there exist x, y ∈ (a, b) such that x < y and f (x) 6= f (y). Now there is x0 ∈
(x, y) such that (7) holds. This gives
So, using MVT, there exists a number θ in (0, 1)
1
such that f ′ (x0 ) = [f (y) − f (x)]/(y − x) 6= 0,

f (a+h)−f (a) = hf ′ (a+θh), a+θh = (1/2)θ since f (y) 6= f (x). This is a contradiction.
1 2 (iii) Suppose f ′ (x) > 0, ∀x ∈ (a, b). Then for a < x < y < b, there is x0 ∈
i.e. f (1/2) − f (0) = (1/2)f ′ (θ/2). (x, y) such that (7) holds. Hence as f ′ (x0 ) > 0 and y − x > 0, (7) shows that
1
f (y) > f (x). Thus f is strictly increasing on (a, b). The proof of (iv) is similar.
Calculus 53 54 Dr. V. V. Acharya

Observe that if I denotes any one of the intervals Remark 2.8 Suppose a function f is derivable on [a, b] and f ′ (x) > 0 ∀x ∈
[a, b]. Then f is strictly increasing and therefore one-one on [a, b]. Hence f is
[a, b), (a, b], [a, b], [a, ∞), (a, ∞), (−∞, b], (−∞, b], (−∞, ∞), onto [f (a), f (b)] as f is continuous on [a, b]. Therefore the inverse function g of
and if f is continuous on I, and derivable on the corresponding open interval, f is defined on [f (a), f (b)]. It can be shown that g is derivable on [f (a), f (b)] and
then each of the above conclusions holds on I. For example, if f is continuous on if y ∈ [f (a), f (b)] and g(y) = x so that y = f (x), then
[a, b), and derivable on (a, b), and f ′ (x) > 0 on (a, b), then f is strictly increasing 1
on [a, b). For proving this we only need to replace the inequalities a < x < y < b g ′ (y) = . (12)
f ′ (x)
in the above proof by a ≤ x < y < b.
The parts (iii) and (iv) of the above theorem can be used to prove many in- This result is called the Inverse Function Theorem. If f is strictly decreasing on
equalities by noting that if a function f is increasing on [a, b], then f (a) is the [a, b], g is defined on [f (b), f (a)] and again satisfies (8).
minimum value of f on [a, b] and if f is decreasing on [a, b], then f (b) is the
minimum value of f on [a, b]. The following theorem is a generalization of Lagrange’s MVT.
Example 2.8 Show that for 0 < x < π/2,
Theorem 2.8 (Cauchy’s Mean Value Theorem) Suppose a < b and functions f
(i) sin x < x < tan x (ii) 2/π < (sin x)/x < 1.
and g are
Solution. (i) For x ∈ [0, π/2), let f (x) = x − sin x, g(x) = tan x − x. Then f, g (i) continuous on the closed and bounded interval [a, b] and
are continuous on [0, π/2) and for all x ∈ (0, π/2), we have (ii) derivable on the open interval (a, b) and
(iii) g ′ (x) 6= 0, ∀x ∈ (a, b).
f ′ (x) = 1 − cos x > 0, g ′ (x) = sec2 x − 1 > 0
Then there is a point x0 ∈ (a, b) such that
and so f, g are strictly increasing on [0, π/2). Hence for x ∈ (0, π/2),
f (b) − f (a) f ′ (x0 )
= ′ . (13)
f (x) > f (0) = 0 and g(x) > g(0) = 0, i.e. x > sin x and tan x > x. g(b) − g(a) g (x0 )
(ii) For x ∈ (0, π/2], let F (x) = (sin x)/x and F (0) = 1. Then F is continuous Proof : First note that g(b) − g(a) 6= 0. For, if g(a) = g(b), then by (i),(ii) Rolle’s
on [0, π/2]. Also, on (0, π/2), x < tan x by (i) and so theorem would be applicable to g and so there would exist a point c ∈ (a, b)
x cos x − sin x cos x(x − tan x) such that g ′ (c) = 0, contrary to assumption (iii). Now for a ≤ x ≤ b define the
F ′ (x) = = < 0. function F by
x2 x2
F (x) = f (x) − kg(x)
Hence F is strictly decreasing on [0, π/2) and so for (0, π/2),
where the constant k is to be chosen so that F (a) = F (b) i.e.
2
= F (π/2) < F (x) < F (0) = 1.
π f (a) − kg(a) = f (b) − kg(b).
Remark 2.7 Suppose a function f is derivable on (a, b). If f is increasing on
This uniquely determines k as
(a, b), then f ′ (x) ≥ 0 on (a, b). This is so because then
f (y) − f (x) f (b) − f (a)
a < x < y < b ⇒ f (x) ≤ f (y) ⇒ ≥ 0, k= , (14)
y−x g(b) − g(a)

so that as y → x, we get f ′ (x) ≥ 0. Similarly if f decreasing on (a, b), then since g(b) − g(a) 6= 0. With this choice of k we see by (i) that F is continuous
f ′ (x) ≤ 0 on (a, b). on [a, b] and by (ii) that F is derivable on (a, b) and F ′ (x) = f ′ (x) − kg ′ (x).
Calculus 55 56 Dr. V. V. Acharya

Thus F satisfies the conditions of Rolle’s theorem on [a, b] and hence there exists Exercise Set 2.1
x0 ∈ (a, b) such that F ′ (x0 ) = 0. Hence f ′ (x0 ) − kg ′ (x0 ) = 0 and so we get
1. Test whether the assumptions of Rolle’s theorem hold for the given function
f ′ (x0 ) on the given interval and if so, verify the theorem.
k=
g ′ (x0 )
(i) f (x) = (x − a)m (x − b)n , [a, b], m, n being positive integers.
and the result follows by (10).
(ii) f (x) = |x|, [−1, 1]. (iii) f (x) = e−x sin x, [0, π].
Remark 2.9 Taking g(x) = x on [a, b], in the above theorem we obtain La- (iv) f (x) = log[(x2 + ab)/x(a + b)], [a, b]; a > 0.
grange’s MVT.
(v) f (x) = 1 − (x − 1)2/3 , [0, 2].
Remark 2.10 The usefulness of the above theorem depends on the fact that in (9) (vi) f (x) = 2 sin x + cos 2x, [0, π/2].
above, f ′ and g ′ are taken at the same point x0 . If we apply the MVT separately
1 1
to f and g, then we get (vii) f (x) = + , on (0, 1), f (0) = f (1) = 0.
x 1−x
f (b) − f (a) = (b − a)f ′ (x1 ) and g(b) − g(a) = (b − a)g ′ (x2 ) p
(viii) f (x) = (x − a)((b − x), [a, b].
for some x1 , x2 in (a, b). But, in general, the points x0 , x1 and x2 are all differ- 1 1
(ix) f (x) = x sin , 0 < x ≤ , f (0) = 0.
ent. Hence we cannot deduce (9) from the last two equations, by division. For x π
4 3
example, let f (x) =p x and g(x) =px on [1, 2]. Then it is easy to see that
x0 = 45/28, x1 = 15/4 and x2 = 7/3. 2. If f (x) = x(x + 1)(x + 2)(x + 3), show that the equation f ′ (x) = 0 has 3
real roots.
Remark 2.11 Geometrically, consider a curve whose parametric equations are
x = g(t), y = f (t), a ≤ x ≤ b. Then the slope of the curve at any point is 3. Show that there is no real number k such that the equation
dy ′ x3 − 12x + k = 0 has 2 distinct roots in [0, 2].
= fg′ (t)
(t)
. Also, the slope of the chord joining the end points A(g(a), f (a))
dx
and B(g(b), f (b)) is the expression on the left hand side of (9). Hence under 4. Using the function f (x) = (4 − x) log x, show that x log x = 4 − x for
the assumptions of the theorem, there exists a number x0 in (a, b) such that the some x in (1, 4).
tangent to the curve at the point (g(x0 ), f (x0 )) is parallel to the chord AB.
5. Find c of Lagrange’s Mean Value Theorem for the given function on the
Example 2.9 Verify Cauchy’s mean value theorem for the functions defined on given interval:
the interval [−π/2, 0] by f (x) = sin x, g(x) = cos x. (i) f (x) = x3 , [−1, 2] (ii) f (x) = log x, [1, e]
Solution. The given functions are derivable on [−π/2, 0] and g ′ (x) = − sin x 6= 0 √
2
on (−π/2, 0). Hence f, g satisfy the conditions of Cauchy’s mean value theorem. (iii) f (x) = px + qx + r, [a, b] (iv) f (x) = 100 − x2 , [0, 8].
Now clearly, f (−π/2) = −1, f (0) = 0, f ′ (x) = cos x, g(−π/2) = 0, g(0) = (
3−x2
if 0 ≤ x ≤ 1
1, g ′ (x) = − sin x. To verify the conclusion of Cauchy’s theorem, we have to (v) f (x) = 1 2
find a point x ∈ (−π/2, 0) such that x if x > 1.

f (0) − f (−π/2) f ′ (x) cos x 6. Find θ of Lagrange’s Mean Value Theorem for the given function on the
= ′ i.e. 1= i.e. tan x = −1. given interval:
g(0) − g(−π/2) g (x) − sin x

(i) f (x) = 1 − x2 , [0, 1] (ii) f (x) = x3 − 3x, [−1, 1]
This gives the solutions x = −π/4 + nπ, where n = any integer. Clearly, x0 =
−π/4 lies in (−π/2, 0) and therefore is as required. (iii) f (x) = x(x − 1)(x − 2), [0, 12 ]
Calculus 57 58 Dr. V. V. Acharya

7. Applying the Mean Value Theorem to sin x and tan−1 x over [x, y], show 1 1
(iii) f (x) = √ , g(x) = , [a, b], a > 0.
that x 2 x
(i) | sin x − sin y| ≤ |x − y| (ii) | tan−1 x − tan−1 y| ≤ |x − y|. (iv) f (x) = x(x − 1)(x − 2), g(x) = x(x − 2)(x − 3), [0, 1/2].
b−a b−a (v) f (x) = sin x, g(x) = cos x, [−π/2, 0].
8. Show that, for 0 < a < b, < tan−1 b − tan−1 a < .
1 + b2 1 + a2 (vi) f (x) = x − x2 , g(x) = 3x2 − 4x3 , [0, 1].
√ √ 1
9. Show that if a > 0, then for x, y ∈ [a, ∞), | x − y| ≤ |x − y|. 14. If f satisfies conditions of Lagrange’s mean value theorem on [a, b], where
2a
0 < a < b, show that there exists a number x0 ∈ (a, b) such that
10. Use the Mean Value Theorem to show that
1 √ 1 b
(i) < 51 − 7 < . f (b) − f (a) = x0 f ′ (x0 ) log .
8 7 a
x
(ii) x ≤ sin−1 x ≤ √ , for 0 ≤ x < 1.
1 − x2
x 2.3 Indeterminate Forms
(iii) ≤ tan−1 x ≤ x, x ≥ 0.
1 + x2
√ sin x Suppose f (x) → l and g(x) → m as x → a. If m 6= 0, then we know that
(iv) 1 − x2 ≤ < 1. f (x)/g(x) → l/m as x → a. If l 6= 0, and m = 0, then limx→a f (x)/g(x) does
x
x not exist. But if l = 0 = m, then as x → a, f (x)/g(x) may tend to a finite limit
(v) x < tan x < √ , 0 < x < 1. or to ∞ or to −∞ or it may oscillate. For example, as x → 0,
1 − x2
1 sin x x −x
(vi) 1 − < log x < x − 1, for x > 0, x 6= 1. → 1, 3 → ∞, → −∞,
x x x x3
(vii) | cos ax − cos bx| ≤ |x| · |a − b|, x 6= 0.
(viii) | sin ax − sin bx| ≤ |x| · |a − b|, x 6= 0. while x/x2 oscillates. Hence when l = m = 0, the limit limx→a f (x)/g(x) is
called the “indeterminate form 0/0” if it is not immediately clear whether or not
11. Find the intervals on which the given function is monotonic: the limit exists. Similarly, if f (x) → ∞ and g(x) → ∞ as → a, then the limit
(i) f (x) = 2x3 − 9x2 − 60x − 60. limx→a f (x)/g(x) is called the “indeterminate form ∞/∞”. We now state some
results, called L’Hospital’s rules, which are useful in evaluating indeterminate
(ii) f (x) = 24x5 + 15x4 − 20x3 + 1. forms. In theorems 3.20 and 3.21 below, L denotes a real number or ∞ or −∞.
12. Suppose h > 0 and f satisfies Rolle’s conditions on [a − h, a + h]. Show
Theorem 2.9 (0/0 form) Suppose a < b and functions f and g are derivable in
that there exists a number θ such that 0 < θ < 1 and
(a, b] and g ′ (x) 6= 0, on (a, b]. Suppose f (x) → 000 and g(x) → 0 as x → a + .
f (a + h) − f (a − h) = h[f ′ (a + θh) + f ′ (a − θh)]. Then
f ′ (x) f (x)
lim ′ = L ⇒ lim = L.
x→a+ g (x) x→a+ g(x)
13. Find x0 of Cauchy’s Mean Value Theorem for the given function on the
given interval: Notes : (1) Results similar to the above theorem are true when we have x → a−
(i) f (x) = ex , g(x) = e−x , [a, b]. or x → a in place of x → a+. (2) If, as x → a+, the limit lim f ′ (x)/g ′ (x) is also
√ 1 indeterminate of the form 0/0, then it may be possible to apply the above theorem
(ii) f (x) = x, g(x) = √ , [a, b], a > 0. again to conclude that if lim f ′′ (x)/g ′′ (x) exists, then it equals lim f ′ (x)/g ′ (x).
x
Calculus 59 60 Dr. V. V. Acharya

Example 2.10 We have Example 2.12 If m is a real number, show that

ex − 1 − x ex − 1 xm (log x)m
lim = lim [First step] (i) lim =0 (ii) lim = 0.
x→0 x2 x→0 2x x→∞ ex x→∞ x
x
e Proof : (i) Clear if m ≤ 0. So, let m > 0. Let n = the integral part of m and
= lim [Second step]
x→0 2 m = n + r. Then 0 ≤ r < 1. Applying the above corollary n times we get
1
= [Third step].
2 xm m(m − 1) · · · (m − n + 1)xr
lim = lim .
x→∞ ex x→∞ ex
Here we have applied the above theorem in the first and second steps. But note
that, in the third step, the limit limx→0 ex /2 is not indeterminate and so the theo-
The last limit is clearly 0 if r = 0. If r > 0, then again by the above corollary,
rem in not applicable to it.
xr rxr−1
Example 2.11 We have lim = lim = 0 as x → ∞, since r − 1 < 0.
x→∞ ex x→∞ ex
log(1 + x) − x 1/(1 + x) − 1
lim = lim [First step] (ii) Follows from (i) by putting x = log t.
x→0 x2 x→0 2x
−x 1
= lim =− .
x→0 2x(1 + x) 2 In the above we saw how to deal with the indeterminate forms 0/0 and ∞/∞.
The other indeterminate forms are
Here we have applied the above theorem only in the first step.
0 · ∞, ∞ − ∞, 00 , ∞0 , and 1∞ .
Corollary 2.1 Suppose that the functions f and g are derivable on (a, ∞) and
g ′ (x) 6= 0 for x > a. Suppose f (x) → 0 and g(x) → 0 as x → ∞. Then
Here the first two forms can be converted into 0/0 form by rearrangement. For

f (x) f (x) the remaining forms we need to use the continuity property of the logarithmic and
lim =L ⇒ lim = L. exponential functions.
x→∞ g ′ (x) x→∞ g(x)

Note : A similar result holds when x → −∞.


Example 2.13 Consider limx→0+ xm · log x.
Theorem 2.10 (∞/∞ form) Suppose that the functions f and g are derivable in This is not indeterminate if m ≤ 0; for then the limit is −∞. So let m > 0.
(a, b] and g ′ (x) 6= 0, on (a, b]. Suppose f (x) → ∞ and g(x) → ∞ as x → a + . Then
Then
f ′ (x) f (x) lim xm · log x [0 · (−∞) form]
lim ′ = L ⇒ lim = L. x→0+
x→a+ g (x) x→a+ g(x)
log x
= lim [−∞/∞ form]
Corollary 2.2 Suppose that the functions f and g are derivable on (a, ∞) and x→0+ x−m
g ′ (x) 6= 0 for x > a. Suppose f (x) → ∞ and g(x) → ∞ as x → ∞. Then 1/x
= lim [by Theorem 3.21]
x→0+ −mx−m−1

f (x) f (x) −xm
lim =L ⇒ lim = L. = lim = 0, since m > 0.
x→∞ g ′ (x) x→∞ g(x) x→0+ m
Calculus 61 62 Dr. V. V. Acharya

Example 2.14 Consider Exercise Set 2.2


 x 1 
L = lim − [(∞ − ∞) form] 1. Prove that
x→1+ x − 1 log x
(1 − x)n − 1 xex − log(1 + x) 3
x log x − x + 1 (i) lim = n. (ii) lim 2
= .
= lim [0/0 form] x→0 x x→0 x 2
x→1+ (x − 1) log x eax − e−ax 2a ex − 2 cos x + e−x
(iii) lim = . (iv) lim = 2.
log x + 1 − 1 x→0 log(1 + bx) b x→0 x sin x
L = lim [by Theorem 3.21] x cos x − sin x 1 x cos x − log(1 + x) 1
x→1+ log x + (x − 1)/x
(v) lim =− . (vi) lim = .
x log x x→0 x3 3 x→0 x2 2
= lim [0/0 form] 1 − cos x 1 tan x − x 1
x→1+ x log x + x − 1 (vii) lim = . (viii) lim 2 = .
x→0 x2 2 x→0 x tan x 3
log x + 1 1 ax − b x a log(1 + x sin x)
= lim = . (ix) lim = log . (x) lim = 2.
x→1+ log x + 1 + 1 2 x→0 sin x b x→0 1 − cos x
log(1 + x3 ) x
e − 1 + log(1 − x) 1
Example 2.15 Consider lim xx [00 form] (xi) lim = 1. (xii) lim =− .
x→0+ x→0 sin3 x 2 x→0 tan x − x 2
Put y = xx . Now, log y = x log x → 0 as x → 0+ . Also, y = elog y . Hence as log(1 − x ) ex sin x − x − x2 2
(xiii) lim = 2. (xiv) lim 2 =− .
x → 0+ , x→0 log cos x x→0 x − x log(1 − x) 3
x sin x
lim y = lim elog y = elim log y = e0 = 1. e −e
(xv) lim = 1.
x→0 x − sin x
Example 2.16 We have
 log(1 + mx)  2. Prove that
lim (1 + mx)1/x [1∞ form] = lim exp log(x − π2 ) log(x − π)
x→0+ x→0+ x (i) lim =0 (ii) lim+ = 1.
π+
x→ 2 tan x x→π ex − eπ
 log(1 + mx) 
= exp lim = em . log(1 − x2 ) log x2
x→0+ x (iii) lim = 2. (iv) lim = 0.
x→0 log cos x x→0 cotx2
Example 2.17 For m > 0 consider lim (1 + mx)1/x [∞0 form] sec πx log sin 2x
x→∞ (v) lim1 = 3. (vi) lim = 1.
x→ 2 tan 3πx x→0+ log sin x
Put y = (1 + mx)1/x . Then tan 5x 1
log(1 + mx) m (vii) limπ = .
x→ 2 tan x 5
lim log y = lim = lim = 0.
x→∞ x→∞ x x→∞ 1 + mx
3. Prove thath
Hence as x → ∞, lim y = lim elog y = elim log y = e0 = 1. The same method 1 1 i 1 h1 log(1 + x) i 1
shows that (i) lim 2
− 2x = − 3 . (ii) lim − = .
x→0 x sin x→0 x x2 i 2 π2
lim (ex + x)1/x = e. ha xi hπ π
x→∞ (iii) lim − cot = 0. (iv) lim − = .
x→0 x a x→0 4x 2x(e πx + 1) 8
Example 2.18 We have 1 1  1  x
(v) lim x 1−x = . (vi) lim 1 + 2 = 1.
ex + e−x − 2 cos x x→1− e x→∞ x 2
lim [0/0 form]
x→0 x sin x (vii) limπ (sin x)2 tan x = 1. (viii) lim (cos x)cot x = e−1/2 .
x→ 2 x→0
ex + e−x − 2 cos x x 1 1  1
= lim (ix) lim (ax + x)1/x = ae. (x) lim − = .
x→0 x2 sin x x→0+ x 2
x −x
x→0
 1 x tan x 3
ex + e−x + 2 cos x
 1 2 sin x 1 1
e − e + 2 sin x

= lim = lim = 2. (xi) lim = 1. (xii) lim x − =− .
x→0 2x x→0 2 x→0+ x x→0 e − 1 x 2
Calculus 63 64 Dr. V. V. Acharya

 tan x 1/x2 2 1  1 so that (4) follows.



(xiii) lim = e1/3 . (xiv) lim − =− .
x→0 x x→1 −x2 1 x − 1 2 We can easily prove the following results using induction:
1  x tan(πx/2a) 2
(xv) limπ (tan x)tan 2x = . (xvi) lim 2 − = .
x→ 4 e x→a a π If y = eax , then yn = an eax . (5)
(xvii) limπ cos x log tan x = 0. (xviii) lim xn · log x = 0 if n > 0. If x x
y = a , then yn = a (log x) . n
(6)
x→ 2 x→0+
 ax + bx 1/x √
(xix) lim x log sin x = 0. (xx) lim = ab. π
x→0+ x→0+ 2 If y = sin(ax + b), then yn = an sin(ax + b + n ). (7)
1 2
4. Prove that (i) lim (cosec x)1/ log x = .
x→0+ e Proof : True for n = 1 since
(ii) lim (cos x)cos x = 1. (iii) lim (1 − x2 )1/ log(1−x) = e. π
π
x→ 2 − x→1− y1 = a cos(ax + b) = a sin(ax + b + ).
2
2.4 Successive Differentiation Assume the result for n and differentiate (7). Then
π π π
If y = f (x), we denote the nth derivative of f (x) as yn+1 = an+1 cos(ax + b + n ) = an+1 sin(ax + b + + ).
2 2 2
dn y Hence the result holds for n + 1 and the induction is complete.
f (n) (x) or yn (x) or Dn f (x) or .
dxn
π
Here we collect formulas for the nth derivative in some simple cases. These can If y = cos(ax + b), then yn = an cos(ax + b + n ). (8)
2
be proved by induction on n. Let y = (ax + b)m , where m is a real number. Then
yn = m(m − 1) · · · (m − n + 1)an · (ax + b)m−n . (1) If y = eax sin(bx + c), then yn = rn eax sin(bx + c + nα), (9)
p
Corollary : If m is a positive integer, then where r = a2 + b2 and α = tan−1 (b/a).

m! = aeax sin(bx+c)+eax cos(bx+c). Put a = r cos α and b = r sin α so


Proof : y1 √
yn = Dn (ax + b)m = an (ax + b)m−n , (2) that r = + a2 + b2 and α is given by cos α = a/r, sin α = b/r, − π < α ≤ π.
(m − n)!
Then
if n ≤ m. Also yn = 0 if n > m, since ym is a constant. If m = −1 in (1) then y1 = reax sin(bx + c + α).
we get the following:
Hence the result holds for n = 1. Differentiating (9) it is easy to see that
1 (−1)n n!an
If y = , then yn = . (3) yn+1 = rn eax sin(bx + c + nα + α).
ax + b (ax + b)n+1
Hence the result holds for n + 1 and the induction is complete.
(−1)n−1 (n − 1)!an
If y = log(ax + b), then yn = . (4)
(ax + b)n If y = eax cos(bx + c), then yn = rn eax cos(bx + c + nα), (10)
p
Proof : Here y1 = a/(ax + b). Differentiate n − 1 times i.e. apply (2) with n − 1 where r = a2 + b2 and α = tan−1 (b/a).
in place of n. Then
If y = af (x) + bg(x) where a, b are constants, then
(−1)n−1 (n − 1)!an−1
yn = a · , yn = af (n) (x) + bg (n) (x). (11)
(ax + b)n
Calculus 65 66 Dr. V. V. Acharya

Theorem 2.11 (Leibnitz Theorem) If y = uv where u = f (x), v = g(x) are Hence


functions which are n times derivable at x, then at x, dn 1 h 1 1 i 1 h dn 1 dn 1 i
yn = n
− = n
− n [by (11)]
 
n
 
n dx 2 x − 3 x − 1 2 dx x − 3 dx x − 1
yn = un v + un−1 v1 + · · · 1 h (−1)n n! (−1)n n! i
0 1 = − [by (3)]
      2 (x − 3)n+1 (x − 1)n+1
n n n
+ un−r+1 vr−1 + un−r vr · · · + uvn . (12)
r−1 r n (ii) Here we first convert the product into a sum:

Proof : We prove the formula by induction on n. The result is true for n = 1 since 1 1
y = sin 2x cos 3x = [sin 5x + sin(−x)] = [sin 5x − sin x].
    2 2
1 1
y1 = u1 v + uv1 = u1 v + uv1 . dn 1 1 dn dn
0 1 Hence, yn = [sin 5x − sin x] = [ sin 5x − sin x] [by (11)]
dxn 2 2 dxn dxn
Now assume the result for n and differentiate (12). Then 1
= [5n sin 5(x + nπ/2) − sin(x + nπ/2)]. [by (7)]
 
n
 
n 2
yn+1 = un+1 v + un v1 + un v1 + un−1 v2 + · · ·
1 1 Example 2.20 Find yn if y = 1/(x2 + a2 ).
   
n n Solution : We have
+ un−r+2 vr−1 + + un−r+1 vr
r−1 r−1 1 1 h 1 1 i
 
n
 
n y= = − .
+ un−r+1 vr + un−r+1 vr+1 + · · · + u1 vn + uvn+1 (x − ai)(x + ai) 2ai x − ai x + ai
r r
    So, by (3),
n n
= un+1 v + [ + ]un v1 + · · · (−1)n n! h 1 1 i
0 1 yn = − (i)
    2ai (x − ai)n+1 (x + ai)n+1
n n
+[ + ]un−r+1 vr + · · · + uvn+1
r−1 r Now, to convert the result into real form, put x = r cos θ, a = r sin θ. Then
(x − ai)n+1 = rn+1 (cos θ − i sin θ)n+1
     
n+1 n+1 n+1
yn+1 = un+1 v + · · · + un+1−r vr + · · · + uvn+1 ,
0 r n+1 = rn+1 (cos(n + 1)θ − i sin(n + 1)θ),
because       and similarly,
n n n+1 (x + ai)n+1 = rn+1 cos(n + 1)θ + i sin(n + 1)θ),
+ = .
r r−1 r
by DeMoivre’s theorem. Hence (i) becomes after substitution,
Hence the result holds for n + 1 and the induction is complete.
(−1)n n!
Example 2.19 Find the nth derivative of the function yn = 2i sin(n + 1)θ.
2airn+1
1
(i) y = 2 (ii) y = sin 2x cos 3x. Hence, as r = a/ sin(n + 1)θ, we get
x − 4x + 3
Solution : (i) Here we first express the function in partial fractions: (−1)n n!
yn = sinn+1 θ sin(n + 1)θ,
1 1 1h 1 1 i an+2
y= 2 = = − . where tan θ = a/x.
x − 4x + 3 (x − 1)(x − 3) 2 x−3 x−1
Calculus 67 68 Dr. V. V. Acharya

Example 2.21 Find yn if y = x2 ex cos x. Hence differentiating n times using Leibnitz’s theorem, we get
x n(n − 1)
Solution : We apply Leibnitz’s theorem to y = uv where u = e cos x and
(x2 − 1)yn+2 + nyn+1 · (2x) + yn · (2) − 2(n − 1)xyn+1
v = x2 . Note that u, v have been chosen in such a way that formulas for un , vn 2
are already known. Moreover, in this case, v = x2 is a polynomial of degree 2 − 2n(n − 1)yn · (1) − 2nyn = 0
and so the derivatives of v of order ≥ 3 are identically zero. Hence or (x2 − 1)yn+2 + 2xyn+1 − n(n + 1)yn = 0.
n(n − 1) Exercise Set 2.3
yn = un x2 + nun−1 · (2x) + un−2 · (2),
2 dn  1  n! h 1 1 i
1. = + .
and so by (10), dxn a2 − x2 2a (a − x)n+1 (a + x)n+1

yn = rn ex cos(x + nα)x2 + nrn−1 ex cos(x + (n − 1)α) 2. If y = sin2 x · sin 2x, then


1h nπ nπ i
+ n(n − 1)rn−2 ex cos(x + (n − 2)α), yn = 2n+1 sin(2x + − 4n−1 sin(4x + .
√ 4 2 2
where r = 2 and tan α = 1/2. 3. If y = eax cos x sin x, then
1
Example 2.22 Differentiate n times the following equation, with usual notation : yn = eax (a2 + 4)n/2 sin(2x + n tan−1 (2/a)).
2
(1 − x2 )y2 − xy1 + m2 y = 0. √
4. If y = log(x + 1 + x2 ), prove that
(i) (1 + x2 )y2 + xy1 = 0.
Solution : By (11), we have
(ii) (1 + x2 )yn+2 + (2n + 1)xyn+1 + n2 yn = 0.
dn 2 dn dn
[(1 − x )y 2 ] − (xy 1 ) + (m2 y) = 0 5. If y = ea sin
−1
x,
prove that
dxn dxn dxn
and hence by Leibnitz’s theorem, we get (1 − x2 )yn+2 − (2n + 1)xyn+1 − (n2 − a2 )yn = 0.

n(n − 1) 6. If y = sin mx + cos mx, prove that yn = mn (1 + (−1)n sin 2mx)1/2 .


(1 − x2 )yn+2 + nyn+1 · (−2x) + yn · (−2)
2 −1
7. If y = ea tan x, prove that
− xyn+1 − nyn (1) + m2 yn = 0,
(1 + x2 )yn+2 + (2nx + 2x − 1)yn+1 + n(n + 1)yn = 0.
or (1 − x2 )yn+2 − (2n + 1)xyn+1 − (n2 − m2 )yn = 0.
8. If y = sin−1 x, prove that (1 − x2 )yn+2 − (2n + 1)xyn+1 − n2 yn = 0.
th
Example 2.23 The n derivative of a composite function is usually difficult to
9. If y = (x2 − 1)n , prove that (x2 − 1)yn+2 + 2xyn+1 − n(n + 1)yn = 0.
find. One method is to obtain the differential equation satisfied by the function and
then applying Leibnitz’s theorem we get a recursive relation between the deriva- 10. If y = e−x xn , prove that
tives. For example, consider the function y = (x2 − 1)n . Differentiating once we (i) xy2 − (n − 1 − x)y1 + y = 0.
get
y1 = n(x2 − 1)n−1 · (2x), and so (x2 − 1)y1 = 2nxy. (ii) xyn+2 + (x + 1)yn+1 + (n + 1)yn = 0.

Differentiating again we get 11. If y = tan−1 x, prove that


(i) (1 + x2 )y2 + 2xy1 = 0.
(x2 − 1)y2 + 2xy1 − 2ny − 2nxy1 = 0
(i) (1 + x2 )yn+2 + 2(n + 1)xyn+1 + n(n + 1)yn = 0.
or (x2 − 1)y2 − 2(n − 1)xy1 − 2ny = 0.
70 Dr. V. V. Acharya

trigonometric, exponential and logarithmic functions are easily obtained by using


the corresponding power series. Also, when the power series expansion of a func-
tion is available, the series helps us in studying the properties of that function. In
Chapter 3 this chapter we show how Taylor’s theorem allows us to obtain such power series
expansions of standard functions.
Taylor’s theorem The series
a0 + a1 (x − a) + a2 (x − a)2 + · · · + an (x − a)n + · · ·
3.1 Introduction is called a power series about the point a. Thus the series (1) above is a power
series about the origin.
Suppose we have a function f (x). We are interested in writing it in terms of a
polynomial or in powers of x. In particular, we have power series or a series in
powers of x as follows: 3.2 Taylor’s Theorem

a0 + a1 x + a2 x2 + · · · + an xn + · · · (1) The following theorem generalizes Lagrange’s mean value theorem.

Here a0 , a1 · · · are real numbers, called the coefficients of the power series. Note Theorem 3.1 (Taylor) Let h > 0, and let f be a function defined on the closed
that the terms a0 , a1 x, a2 x2 , · · · of the series are functions defined on R. A simple interval [a, a + h] such that the (n − 1)th derivative f (n−1) of f is continuous on
example of a power series is the geometric series: [a, a + h] and the nth derivative f (n) of f exists on the open interval (a, a + h)
for some positive integer n. Let p be a given positive integer. Then there exists a
1 + x + x2 + · · · + xn + · · · (2) number θ such that 0 < θ < 1 and
h2 ′′ hn−1 (n−1)
We know that if |x| < 1 i.e. if x is in the open interval (−1, 1), then the above f (a + h) = f (a) + hf ′ (a) + f (a) + · · · + f (a)+ Rn , (3)
1 2! (n − 1)!
series is convergent and has the unique sum and it is divergent if |x| ≥ 1.
1−x hn (1 − θ)n−p (n)
1 where Rn = f (a + θh). (4)
Hence the sum of the series (2) is the function f defined by f (x) = if x ∈ (n − 1)!p
1−x
(−1, 1). Thus we write Proof : Since the derivative f (n−1) exists [a, a + h], it follows that the functions
f, f ′ , f ′′ , . . . , f (n−2) are all continuous on [a, a + h]. Choose a number A such
1
f (x) = = 1 + x + x2 + · · · + xn + · · · if x ∈ (−1, 1). that
1−x
h2 ′′ hn−1 (n−1)
and say that the series (2) is the power series expansion of f on (−1, 1) or that f (a + h) = f (a) + hf ′ (a) + f (a) + · · · + f (a) + Ahp . (5)
2! (n − 1)!
the series represents f on (−1, 1).
More generally, if a power series such as (1) converges for a non-zero value Define the function F on [a, a + h] as follows :
of x, then it can be shown that the series converges on an interval so that its sum 1
is a function of x defined on that interval. Conversely, it is very important to be F (x) = f (x) + (a + h − x)f ′ (x) + (a + h − x)2 f ′′ (x) + · · ·
2!
able to expand a given function f as a power series in some interval I. One rea- 1
son for this is that the function f may be very complicated and so its values may + (a + h − x)n−1f (n−1) (a) + A(a + h − x)p − f (a + h).(6)
(n − 1)!
not be easy to calculate. But the partial sums of a power series are polynomi-
als which are easy to evaluate at a given point and they give approximations to Then by data, F is continuous on [a, a + h] and derivable on (a, a + h). Also, by
the values of f at points in I. Thus the values of non-algebraic functions like the (6), f (a+h) = 0 and by (6) and (5), F (a) = 0. Thus F (a) = F (a+h) = 0, hence

69
Calculus 71 72 Dr. V. V. Acharya

by Rolle’s theorem there exists a number θ ∈ (0, 1) such that F ′ (a + θh) = 0. Rn (x) is called the remainder in the form given by Schlömilch and Roche. When
But differentiating (6) we get p = 1, we have Cauchy’s form of remainder :

F ′ (x) = f ′ (x) − f ′ (x) + (a + h − x)f ′′ (x) − (a + h − x)f ′′ (x) (x − a)n (1 − θ)n−1 (n)
Rn (x) = f (a + θ(x − a)). (10)
1 (n − 1)!
+ ···+ (a + h − x)n−1 f (n) (x) − pA(a + h − x)p−1 .
(n − 1)!
When p = n, we have Lagrange’s form of remainder, namely
Here all terms, except the last two, cancel in pairs and we get
(x − a)n (n)
1 Rn (x) = f (a + θ(x − a)). (11)
F ′ (x) = (a + h − x)n−1 f n (x) − nA(a + h − x)n−1 . n!
(n − 1)!
Note that the number θ in (9) depends on a, x and n when f is given. Hence the
Hence values of θ in (9) and (10) are different, in general.

(h − θh)n−1 (n) (iv) Suppose that f is a polynomial of degree m in x given by,


0 = F ′ (a + θh) = f (a + θh) − pA(h − θh)p−1 ,
(n − 1)!
f (x) = a0 + a1 x + a2 x2 + · · · + am xm .
hn (1 − θ)n−p (n)
so that A= f (a + θh),
(n − 1)!p Then since f (m) (x) = m! am and f (r) (x) = 0 for r > m, we see that the remain-
der Rn (x) in the finite Taylor expansion of f about any point a is independent of
since 1 − θ 6= 0 and h 6= 0. On substituting this value of A in (5) we get (3) where θ if n = m and zero if n > m. Hence if n ≥ m, we have from (7) the identity
Rn is given by (4).
m
X (x − a)r
Notes : (i) The above result reduces to Lagrange’s mean value theorem when f (x) = f (r) (a). (12)
r!
n = 1. r=0

(ii) Let h < 0, and suppose that the assumptions of the above theorem hold for f Here we write f (0) in place of f.
on [a + h, a]. Then it can be shown that (3) holds in this case also.
Example 3.1 let f (x) = x3 + 3x2 − 2x + 4. Then it is easy to see that
(iii) Suppose α < β and let f (n−1) be continuous on (α, β). Let a, x be points in
(α, β) and a 6= x. Then if f (n) exists on (a, x) or on (x, a), we see by (ii) that (3) f (−1) = 8, f ′ (−1) = −5, f ′′ (−1) = 0, f ′′′ (−1) = 6, and f (r) (x) = 0 if r > 3.
holds for x = a + h. That is, there exists a number θ in (0, 1) such that
Hence, with m = 3, (10) gives
f (x) = Pn (x) + Rn (x), (7)
(x + 1)2 ′′ (x + 1)3 ′′′
where f (x) = f (−1) + (x + 1)f ′ (−1) + f (−1) + f (−1)
2! 3!
h2 ′′ hn−1 (n−1) = 8 − 5(x + 1) + (x + 1)3 ,
Pn (x) = f (a) + hf ′ (a) + f (a) + · · · + f (a) + Rn , (8)
2! (n − 1)!
which is the Taylor expansion of f about −1.
(x − a)n (1 − θ)n−p (n)
and Rn (x) = f (a + θ(x − a)). (9) (v) When a = 0, (iii) gives the following theorem :
(n − 1)!p

Here (7) is called the finite Taylor expansion of f about the point a (or in powers Theorem 3.2 (Maclaurin) Suppose k > 0 and f is a function on [−k, k] such
of (x − a)). Pn (x) is called the nth order Taylor polynomial for f about a and that f (n−1) is continuous on [−k, k] and f (n) exists on (−k, k) for some positive
Calculus 73 74 Dr. V. V. Acharya

integer n. Let p be a given positive integer. Then for every x ∈ (−k, k), there is a where the remainder is
number θ ∈ (0, 1) such that
(−1)n−1  x n
2 n−1 Rn (x) = . (20)
x x n 1 + θx
f (x) = f (0) + xf (0) + f ′′ (0) + · · · +

f (n−1) (0) + Rn (x), (13)
2! (n − 1)!
Example 3.4 Let m be a given real number. Let f (x) = (1 + x)m for x > −1.
Then for all positive integers r we have
xn (1 − θ)n−p
where Rn (x) = f (n) (θx). (14) f (r) (x) = m(m − 1) · · · (m − r + 1)(1 + x)m−r .
(n − 1)!p
xn (1 − θ)n−1 (n) Hence substituting in (13) we get
or Cauchy′ s form : Rn (x) = f (θx) (15)
(n − 1)!
xn (n) m(m − 1) 2
or Lagrange′s form : Rn (x) = f (θx) (16) (1 + x)m = 1 + mx + x + ···
n! 2!
m(m − 1) · · · (m − n + 2) n−1
Here (13) is called Maclaurin’s formula with remainder. + x + Rn (x),
(n − 1)!
We note that p = 1 gives us Cauchy’s form of remainder and p = n gives La- where the remainder is
grange’s form of remainder.
m(m − 1) · · · (m − n + 1)
Let us apply (13) to some standard functions. Rn (x) = (1 + θx)n . (21)
n!
Example 3.2 Let f (x) = ex . Then for all x and for all positive integers r, we have
f (r) (x) = ex . Hence f (r) (0) = 1 and Maclaurin’s formula (13) with Lagrange’s Example 3.5 Let f (x) = sin x for all ( x. Then for all x and for all r ∈ N,
remainder is
 π  0 if r is even
f (r) (x) = sin x + r . Hence f (0) =
2 (−1)(r−1)/2 if r is odd.
x2 ′′ xn−1 (n−1) xn (n)
f (x) = f (0) + xf ′ (0) + f (0) + · · · + f (0) + f (θx), So (13) with remainder after 2n terms is
2! (n − 1)! n!
x3 x5 x2n−1 x2n
so that sin x = x − + − · · · + (−1)n−1 + (−1)n sin θx, (22)
3! 5! (2n − 1)! (2n)!
x2 xn−1 xn θx
ex = 1+x+ + ··· + + + e . (17) because sin(θx + nπ) = (−1)n sin θx.
2! (n − 1)! n!

Example 3.3 Let f (x) = log(1 + x) for x > −1. Then for all positive integers Example 3.6 Let f (x) = cos x for all ( x. Then for all x and for all r ∈ N,
r, we have  rπ  0 if r is odd
(−1)r−1 (r − 1)! f (r) (x) = cos x + . Hence f (0) = (r−1)/2
f (r) (x) = . (18) 2 (−1) if r is even.
(1 + x)r So (13) with remainder after 2n + 1 terms is
Hence f (0) = 1, f (r) (0) = (−1)r−1 (r − 1)! and for n ≥ 2, (13)becomes on
x2 x4 x6 x2n x2n
substitution, cos x = 1 − + − − · · · + (−1)n + (−1)n+1 sin θx, (23)
2! 4! 6! (2n)! (2n)!
x2 x3 xn−1
log(1 + x) = x − + − · · · + (−1)n−2 + Rn (x) (19) because cos(θx + nπ + π/2) = (−1)n cos(θx + π/2) = (−1)n+1 sin θx.
2 3 (n − 1)
Calculus 75 76 Dr. V. V. Acharya

x3 x5 1 xn
Example 3.7 For n = 3, (22) gives sin x = x − + + R, Thus, = 1 + x + x2 + · · · + xn−1 +
3! 5! 1−x (1 − θx)n+1
x6 1 − xn xn
where R = (−1)3 sin θx. Hence if we take approximately = + so that
6! 1−x (1 − θx)1/(n+1)
1h i
x3 x5 θ = 1 − (1 − x)1/(n+1) .
sin x = x − + , (i) x
3! 5!
(ii) f (r) (x) = (−1)r−1 (r − 1)!x−r . Hence (7), (13) give
6
|x|
then the absolute error is |R| < when x 6= 0. Thus to find sin 20◦ , put log x = log[1 + (x − 1)]
6!
x = π/9 in (i) [20◦ = π/9 radians] to get sin 20◦ = 0.34202. Here the error is (x − 1)2 ′′ (x − 1)3 ′′′
= f (1) + (x − 1)f ′ (1) + f (1) + f (1 + θ(x − 1))
1  π 6 2! 3!
less than = 0.000002, hence the above value is correct to 5 places of
6! 9 1 1 2
decimal. = (x − 1) − (x − 1)2 + (x − 1)3 .
2 6 [1 + θ(x − 1)]3
1h  8 1/3 i
1 So, when x = 3 we get θ = −1+ .
Example 3.8 For |x| ≤ π/2, show that (i) 1 − x2 ≤ cos x 2 log 27
2
1 1
(ii) cos x ≤ 1 − x2 + x4 . x
√= xe cos x. π
Example 3.10 Let f (x)
2 24 n
Note that f (x) = 2e cos(x + n 4 ). Hence, using (13), we get
For n = 1 and 2, (23) gives
x3 x4 x5
1 x3 ex cos x = 1 + x − − − + ··· .
(iii) cos x = 1 − x2 + sin θx, and 3 6 30
2 3!
1 x4 x5 3.3 Maxima and Minima
(iv) cos x = 1 − x2 + − sin θx.
2 24 5!
Taylor’s theorem allows us to extend the second derivative test for maxima-minima
π 3 x3 as follows:
When |x| ≤ x and sin θx have the same sign as θ > 0. Hence sin θx ≥ 0,
2 3!
x5 Theorem 3.3 Suppose (i) there is a positive integer n > 1 such that f (n) is con-
hence (iii) gives (i). Similarly, sin θx ≥ 0, hence (iv) gives (ii). tinuous on a neighbourhood of a and (ii) f (n) (a) 6= 0 and (iii) f (r) (a) = 0
5!
for r = 1, 2, . . . , n − 1. Then if n is even, then f is locally maximum at a if
Example 3.9 Let us find the number θ of Lagrange’s form of remainder in the f (n) (a) < 0 and minimum if f (n) (a) > 0. If n is odd, then f is not locally
following cases: extreme at a.
Proof: By (i), for every x near a we have by Taylor’s theorem,
(i) a = 0 < x < 1, n arbitrary, f (x) = (1 − x)−1 .
n−1
(ii) a = 1, x = 3, n = 3, f (x) = log x.
X (x − a)r (r) (x − a)n (c)
f (x) = f (a) + f (a) + f ,
r=1
r! n!
(i) Here f (r) (x) = r!(1 − x)r−1 , hence (7) and (13) give
for some c between a and x. Hence by (iii),
2 n−1 n
1 x x x (n) (x − a)n (c)
= f (0) + xf ′ (0) + f ′′ (0) + · · · + f (n−1) (0) + f (θx), f (x) − f (a) = f . (24)
1−x 2! (n − 1)! n! n!
Calculus 77 78 Dr. V. V. Acharya

Let k = f n (a). Then since f ( n) is continuous and non-zero at a, we see that for
every x in some deleted nhd I of a, f (x) and k have the same sign. Let n be even.
Then for every x ∈ I, (x − a)n > 0 and so by (24), f (x) − f (a) and k have the
same sign. Thus

k > 0 ⇔ f (x) > f (a), for everyx ∈ I,

hence f is locally minimum at a. Similarly, if k > 0, then f is locally maximum


at a.
Let n be odd. Let k > 0. Let J be any deleted nhd of a contained in I. Then
for every x ∈ J, (24) shows that f (x) − f (a) and (x − a)n have the same sign.
But as n is odd, (x − a)n is positive for x > a and negative for x < a. Hence
f (x) − f (a) changes sign in every nhd of a. This is also true if k < 0. Hence in
this case f is not locally extreme at a.

Example 3.11 Let us determine the nature of f at x = 0 in the following cases:

x3 x4
(i) f (x) = sin x − x + − ,
6 24
x3
(ii) f (x) = sin x − x + .
6
(i) Here

x2 x3 x2
f ′ (x) = cos x − 1 + − , f ′′ (x) = − sin x + x − ,
2 6 2
′′′
f (x) = − cos x + 1 − x and f (iv) (x) = sin x − 1.

Hence
f ′ (0) = f ′′ (0) = f ′′′ (0) = 0 and f (iv) (0) = −1,
so that the first non-zero derivative of f at 0 is of even (fourth) order and is nega-
tive. Hence by the above theorem, f is locally maximum at x = 0.
(ii) Here

f ′ (x) = cos x − 1 + x2 , f ′′ (x) = − sin x + 2x and f ′′′ (x) = − cos x + 2.

Hence f ′ (0) = f ′′ (0) = 0 and f ′′′ (0) = 1. Hence the first non-zero derivative of
f at 0 is of odd (third) order. Hence by the above theorem, f is not locally extreme
at x = 0.
80 Dr. V. V. Acharya

We have also seen some standard integrals and have solved problems using
these standard forms by applying the rules of integration. Most of the integrals
were obtained by direct use of standard forms or by simplifying the given inte-
Chapter 4 grands to get a combination of standard forms. However,
Z this method is not suit-
able for many functions. For example, the integral tan x dx cannot be found by
Integration Z
inspection and log x dx cannot be simplified further to obtain a standard form.
Hence, we need to develop methods to transform such integrals to standard forms.
4.1 Introduction We shall study the following methods:
We have already been introduced to the process of integration in Class XI and (A) Integration by the method of substitution.
XII. In fact, we defined the concept of indefinite integral as the inverse operation (B) Integration by formation of partial fractions.
to differentiation. More precisely, let f (x) and φ(x) be functions defined on an (C) Integration by parts.
interval [a, b]. Suppose that φ(x) is derivable on [a, b] and is such that
4.2 Integration by substitution
d
φ(x) = f (x) for all x in [a, b]. (1)
dx The following theorem, known as method of substitution or change of variable in
Then we say that φ(x) is an indefinite integral of f (x) on [a, b] and denote this an integral, follows from the chain rule of differential calculus.
fact by writing
Theorem 4.1 Suppose that on the interval [a, b], f (x) is a continuous function
Z and x = φ(t) is a function having continuous derivative. Then
φ(x) = f (x) dx. (2) Z Z
f (x) dx = f [φ(t)]φ′ (t) dt. (1)
Thus the statements (1) and (2) are equivalent. In words, “φ(x) is an indefinite
integral of f (x) if and only if the derivative of φ(x) is equal to f (x) on [a, b].” We
d
Z
do not usually mention the common domain [a, b] of f (x) and φ(x) because most Proof: Let f (x) dx = g(x). So, by definition, [g(x)] = f (x). Now, by the
of the time we deal with simple functions whose natural domains are known to dx
chain rule of differentiation, we have
us. For example, recall that the function log x is defined only for positive values
of x, and we have d d dx dx
[g(x)] = [g(x)] = f (x) = f [φ(t)]φ′ (t) as x = φ(t).
1 1
Z Z
dt dx dt dt
dx = log x on (0, ∞) and dx = log(−x) on (−∞, 0).
x x Hence, again by definition,
Z Z
We combine these two formulas into one by writing
g(x) = f (x) dx = f [φ(t)]φ′ (t) dt,
1
Z
dx = log |x| on [a, b], when 0 6∈ [a, b], i.e. when [a, b] consists
x
entirely of positive numbers or of negative numbers. Also, the results like which is the required result (1).
Z Z Z
2
3x dx = x , 3
sin x dx = − cos x, 2e2x dx = e2x , Notes: 1) The above result (1) is usually written as

dx
Z Z
are valid on any interval [a, b]. f (x) dx = f (x) dt.
dt

79
Calculus 81 82 Dr. V. V. Acharya
Z
2) Comparing the two sides of (1), the process can be viewed as follows: Example 4.1 Find sin2 x cos x dx.
Our intention is to simplify the given integral. So we choose a suitable substi-
tution, say x = φ(t) with inverse t = ψ(x). Solution: Here note that ‘cos x dx’ is the differential of ‘sin x’. So, hopefully, the
To obtain the integral on the right, we replace in the integral on the left, x substitution sin x = t will simplify the integral. Indeed, this does happen because
by φ(t) and the differential dx by φ′ (t) dt (or dxdt dt). Then we evaluate the new
the remaining factor of the integrand is a function of sin x only. So put t = sin x.
integral which is w.r.t. the new variable t ( and which is simpler than the original!) Then cos x dx = dt and sin2 x = t2 so that, by above theorem, we get
Finally, we replace the variable t by ψ(x).
t3 sin3 x
Z Z
3) Sometimes, it is convenient to make the substitution in the form g(x) = t with sin x cos x dx = t2 dt =
2
+c= + c.
3 3
inverse x = h(t). Now t = g(x) gives the differential dt = g ′ (x) dx. So we
dx
Z
replace dx by dt/g ′ (x) and then x by h(t). This gives
Example 4.2 Find I = .
x log x
1
Z Z
1
f (x) dx = f [h(t)] · ′ dt. Solution: Here we observe that dx is the differential of log x. So let log x = t,
g [h(t)] x
1
so that dx = dt. Then
4.2.1 Some standard forms x
1 dx 1
Z Z
We note the following integrals which may be seen as standard forms in applying I= · = dt = log |t| + c = log | log x| + c.
log x x t
the method of substitution: Z p
[f (x)]n+1
Z 3
(A) [f (x)]n f ′ (x) dx = + c, n 6= −1. Example 4.3 Find x 1 + x2 dx.
n+1
Solution: Observe that x dx is almost the differential of 1 + x2 ; in fact, 2x dx is
Proof: Let f (x) = t so that f ′ (x) dx = dt. So the differential of 1 + x2 . So put 1 + x2 = t. Then dt = 2x dx or x dx = 12 dt.
So
tn+1 [f (x)]n+1
Z Z
[f (x)]n f ′ (x) dx = tn dt = +c= + c. Z p Z √
1 1 t4/3 3
n+1 n+1 3
1 + x2 xdx =
3
t dt = · + c = (1 + x2 )4/3 + c.
2 2 4/3 8
f ′ (x)
Z
sin−1 x
Z
(B) dx = log |f (x)| + c. √
f (x) Example 4.4 Find dx.
1 − x2
Proof: Let f (x) = t so that f ′ (x) dx = dt. So 1
Solution: Let sin−1 x = t, so that √ dx = dt. So
Z ′ 1 − x2
f (x) 1
Z
dx = dt = log |t| + c = log |f (x)| + c. Z
sin−1 x
Z
t2 (sin−1 x)2
f (x) t √ dx = t dt = +c= + c.
1 − x2 2 2
Z
f ′ (x) Z √x
e
p
(C) p dx = 2 f (x) + c. Example 4.5 Find √ dx.
f (x) x

Proof: Let f (x) = t so that f ′ (x) dx = dt. So Solution: Let x = t, so that 2√1 x dx = dt i.e. √1x dx = 2dt. Hence,


Z ′
f (x) 1 e x √
Z p Z Z Z
dx = √ dt = 2 t + c = 2 f (x) + c. √ dx = 2et dt = 2 et dt = 2et + c = 2e x
+ c.
f (x) t x
Calculus 83 84 Dr. V. V. Acharya

Exercise Set 4.1 Let sec x + tan x = t so that (sec x tan x + sec2 x) dx = dt. So
2x (1 + log x)2 1
Z Z Z Z
Find the following integrals: i) 2
dx ii) dx sec x dx = dt = log |t| + c = log | sec x + tan x| + c.
1+x x Z t
√ ex + e−x
Z Z
iii) x cos(x2 + 1)dx iv) (4x + 2) x2 + x + 3 dx v) dx Alternatively, note that
exZ− e−x√
Z
cos x + sin x
Z
cos x
Z
dx sec2 x 1 + sin x (cos x2 + sin x2 )2
vi) dx vii) dx viii) √ ix) √ dx sec x + tan x = =
3 cos x cos2 x2 − sin2 x2
Z cos x − sin x Z x(1 − sin x) x+ x x
1 + cos x e (1 + x) sin(tan−1 x) x
Z
1 + tan 2 x π
x) √ dx xi) 2 dx xii) dx = = tan( + ).
3 x 1 + x2 1 − tan x2
Z x +2sin x Z sin (xe ) 2 4
x sin 2x dx
Z
x π
Z
xiii) dx xiv) xv) sin x · sec4 x dx. So sec x dx = log | sec x + tan x| + c = log | tan( + )| + c.
1 + x6 a sin2 x + b2 cos2 x
2
2 4
(4) Multiply the numerator and denominator by cosec x − cot x, to get
4.3 Trigonometric functions Z Z
cosec x(cosec x − cot x)
cosec x dx = dx
cosec x − cot x
Using the above formula (B), we can find integrals of trigonometric functions as
−cosec x cot x + cosec2 x
Z
standard results. These are listed below. = dx.
Z cosec x − cot x
(1) tan x dx = log | sec x| + c.
Z Let cosec x − cot x = t so that (−cosec x cot x + cosec2 x) dx = dt. So
1
Z Z
(2) cot x dx = log | sin x| + c.
cosec x dx = dt = log |t| + c = log |cosec x − cot x| + c.
Z
x π t
(3) sec x dx = log | sec x + tan x| + c = log | tan(
+ )| + c. Alternatively, note that
2 4
x
Z
(4) cosec x dx = log |cosec x − cot x| + c. = log | tan | + c. 1 − cos x 2 sin2 x2 x
2 cosec x − cot x = = = tan .
sin x 2 cos x2 sin x2 2
Proof: (1) Put cos x = t so that − sin x dx = dt. So
x
Z
So cosec x dx = log |cosec x − cot x| + c = log | tan | + c.
sin x 1 1 2
Z Z Z Z
tan x dx = dx = (sin x dx) = (−dt)
cos x cos x t sin(x − a)
Z
= − log |t| + c = − log | cos x| + c = log | sec x| + c. Example 4.6 Find I = dx
cos(x + b)
sin(x − a) sin[(x + b) − (a + b)]
(2) Let sin x = t so that cos x dx = dt. So Solution: We have = . Hence,
cos(x + b) cos(x + b)
Z Z
cos x
Z
1 sin(x − a) sin(x + b) cos(a + b) − cos(x + b) sin(a + b)
cot x dx = dx = dt = log |t| + c = log | sin x| + c. =
sin x t cos(x + b) cos(x + b)
= cos(a + b) tan(x + b) − sin(a + b)
(3) Multiplying the numerator and denominator by sec x + tan x, we get Z Z
Hence, I = cos(a + b) tan(x + b)dx − sin(a + b) 1dx
sec x(sec x + tan x) sec x tan x + sec2 x
Z Z Z
sec x dx = dx = dx. = cos(a + b) log | sec(x + b)| − x sin(a + b) + c.
sec x + tan x sec x + tan x
Calculus 85 86 Dr. V. V. Acharya

dx
Z
Example 4.7 Find √ . Solution: We find constants A and B such that
1 − sin x Numer. = A (Denom.) +B (derivative of the denom.) i.e.
Solution: We have sin x + 3 cos x = A(2 sin x − 5 cos x) + B(2 cos x + 5 sin x)
√ p √ π x = (2A + 5B) sin x + (−5A + 2B) cos x.
1 − sin x = 1 + cos(π/2 + x) = 2 cos + .
4 2 Comparing coefficients of sin x and cos x on both sides we get 2A + 5B = 1 and
Z
dx 1
Z  π x  −13 11
So √ = √ sec + dx −5A + 2B = 3. Solving we get A = and B = .
1 − sin x 2 4 2 29 29
√  π x  π x  Z −13 11
= 2 log sec + + tan + +c 29 (2 sin x − 5 cos x) + 29 (2 cos x + 5 sin x)

4 2 4 2 So I = dx
(2 sin x − 5 cos x)
dx
Z
−13 11 2 cos x + 5 sin x
Z Z
Example 4.8 Find I = . = 1dx + dx
a cos x + b sin x 29 29 2 sin x − 5 cos x
Solution: Put a = r sin θ, b = r cos θ. Then r2 = a2 + b2 , and tan θ = a/b so 13
Z ′
11 f (x)
that θ = tan−1 (a/b). Further, a cos x + b sin x = r(sin θ cos x + cos θ sin x) = = − x+ dx wheref (x) = 2 sin x − 5 cos x
29 29 f (x)
r sin(θ + x). So
13 11
= − x+ log |2 sin x − 5 cos x| + c.
29 29
1 dx 1
Z Z
I = √ =√ cosec (θ + x)dx a sin x + b cos x aex + b
Z Z
a2 + b2 sin(θ + x) a2 + b 2 Note: All integrals of the type dx and dx can be
c sin x + d cos x cex + d
1 a
evaluated using this method.
= √ log | cosec (θ + x) − cot(θ + x)| + c, where θ = tan−1 .
a2 + b 2 b Exercise Set 4.2
Z
dx Integrate the following w.r.t. x :
Example 4.9 Find I = . sin(x − a) cos(x − 2a) cos(x − a) 1
sin(x − a) · cos(x − b) 1) 2) 3) 4)
Solution: Put k = cos(a − b). Then we have sin(x + a) sin(x + a) cos(x − b) sin(x − a) · sin(x − b)
1 2 sin −3 cos x 1 + tan x 3ex − 5
5) 6) 7) 8) x
1 cos(a − b)dx 1 cos[(x − b) − (x − a)] cos(x − a) · cos(x − b) 4 sin x + 5 cos x 1 − tan x 2e + 3
Z Z
I= = dx ex + 1 e3x + 3 1 1
cos(a − b) sin(x − a) · cos(x − b) k sin(x − a) · cos(x − b) 9) x 10) 3x 11) √ 12) √
e −1 3e − 4 1 − cos 3x 1 + cos 2x
1 sin x cos x cos 2x
1 cos(x − b) cos(x − a) + sin(x − b) sin(x − a)
Z 13) √ 14) √ 15) 16)
Hence, I = dx 1 − sin 4x 1 + sin x cos x + sin x (cos x − sin x)2
k sin(x − a) cos(x − b) 1 1 1 1
17) 18) 19) 20)
1 3 cos x + 4 sin x 2 cos x − 3 sin x cos x + sin x sin x − cos x
Z
= [cot(x − a) + tan(x − b)] dx
k
1 4.4 Some standard substitutions
= [log | sin(x − a)| + log | sec(x − b)|] + c
k
sin(x − a) There are some standard substitutions for quadratic expressions under square root
= sec(a − b) · log + c.

sign which
√ are given below:
cos(x − b)
For √a2 − x2 , put x = a sin θ (or x = a cos θ).
sin x + 3 cos x For √a2 + x2 , put x = a tan θ (or x = a cot θ).
Z
Example 4.10 Find I = dx. For x2 − a2 , put x = a sec θ (or x = acosec θ).
2 sin x − 5 cos x
Calculus 87 88 Dr. V. V. Acharya

Using
Z these substitutions we now derive a few more standard integrals. Proof: Put x = a sin θ. Then dx = a cos θdθ, and a2 − x2 = a cos θ. There-
dx √ fore,
(1) √ = log |x + x2 + a2 | + c..
2
x +a 2
dx a cos θ x
Z Z Z
Proof : Put x = a tan θ. Then dx = a sec2 θ dθ, and √ = dθ = dθ = θ + c = sin−1 ( ) + c.
2
a −x 2 a cos θ a
p p p
dx −1 x
Z
x2 + a2 = a2 tan2 θ + a2 = a tan2 θ + 1 = a sec θ. Hence, √ = sin ( ) + c.
Z
dx
Z
a sec2 θ
Z
a2 − x2 a
∴ √ = dθ = sec θ dθ
x2 + a2 a sec θ Z
dx 1 x
= log | sec θ + tan θ| + c1 . (4) = tan−1 ( ) + c.
x2 + a2 a a
√ Proof: Put x = a tan θ. Then dx = a sec2 θdθ and x2 + a2 = a2 tan2 θ + a2 =
x x2 + a2 a2 sec2 θ. Hence,
Now tan θ = so that sec θ = . So
a a
dx a sec2 θ 1 1
Z Z Z
x √x2 + a2 = dθ = dθ = θ + c

x2 + a2 a2 sec2 θ a a
I = log + + c1

dx 1 x
Z
a a
So = tan−1 ( ) + c.

p x2 + a2 a a
= log |x + x2 + a2 | + c, where c = c1 − log a.
dx 1 x − a
Z
dx
Z p
Thus, √ = log |x + x2 + a2 | + c. (5) 2 2
= log + c with |x| > |a|.
x2 + a2 x −a 2a x + a  
1 1 1 1 1
Z
dx Proof: Note that 2 = = − . Hence,
x − a2 (x − a)(x + a) 2a x − a x + a
p
(2) √ = log |x + x2 − a2 | + c..
x2 − a2
dx 1 1 1 i 1
Z Z h
Proof:
√ Put x√= a sec θ. Then dx = a sec θ tan θdθ, and so = − dx = [log |x − a| − log |x + a|] + c
x2 − a2 = a2 sec2 θ − a2 = a tan θ. Hence, 2
x −a 2 2a x−a x+a 2a

dx 1 x − a
Z
Z
dx
Z
a sec θ tan θ
Z Hence, = log x + a + c.

√ = dθ = sec θdθ = log | sec θ + tan θ| + c1 . x2 − a2 2a
x − a2
2 a tan θ Z
dx
1 a+x
√ (6) = 2a log a−x + c with |x| < |a|.
a2 − x2

x x2 − a2 1 1 1 h 1
Now sec θ = . So tan θ = . Therefore,
i
1
a a Proof: Now 2 = = + Hence,
a − x2 (a − x)(a + x) 2a a+x a−x
x √x2 − a2

dx 1 1 1 i
p Z Z h
I = log + + c1 = log |x + x2 − a2 | + (c1 − log |a|)

a = + dx
a a2 − x2 2a a+x a−x

p 1h log |a − x| i 1 a + x
= log |x + x2 − a2 | + c where c = c1 − log |a| = log |a + x| + +c= log + c.
2a −1 2a a − x
dx
Z
dx 1 a + x
p Z
Thus, √ = log |x + x2 − a2 | + c. ∴ = log + c.
x2 − a2 2
a −x 2 2a a − x
dx x
Z
(3) √ = sin−1 ( ) + c.. Thus we have the following results:
a2 − x2 a
Calculus 89 90 Dr. V. V. Acharya

dx 4ac − b2
Z p b
1. √ = log |x + x2 + a2 | + c Now if we put x + = t, then dx = dt and is a constant which is
x2 + a2 2a 4a2
2 2
taken as +k or −k ( with k > 0 ) according to whether it is positive or negative,
dx
Z p
2. √ = log |x + x2 − a2 | + c respectively. Then, by (1) above, the integral
x2 − a2 Z
dx 1
Z
dt 1
Z
dt
becomes either or
dx x p(x) a t2 + k 2 a t2 − k 2
Z
3. √ = sin−1 +c
a2 − x2 a which can be evaluated using the standard results in section 4.5. Similarly for

ax2 + bx + c we use the above procedure to simplify ax2 + bx + c and consider
dx 1 −1 x
Z  
4. = tan +c cases (i) a > 0 and (ii) a < 0. Of course, in any case, we assume that p(x) > 0 in
x2 + a2 a a the interval in which we consider the function p(x). If a > 0, then

dx 1 dt 1 dt
Z Z Z
dx 1 x − a
Z
5. = log x + a + c
becomes either √ √ or √ √ ,
x2 − a2
p
2a p(x) a 2
t +k 2 a t − k2
2

Z
dx 1

a + x
which can be evaluated using the standard results. If a < 0, put a = −α where
6. = log a − x + c.
α > 0. [Then p(x) = −αx2 +bx+c = αq(x) where q(x) = (c/α)+(b/α)x−x2 .]
a2 − x2 2a
Now (1) becomes
Exercise Set 4.3 " 2  #
A) Using the above results obtain the integrals of the following functions: b 4αc + b2
p(x) = − α x − − = α [K − t2 ], (2)
1 1 1 1 1 2α 4α2
1) √ 2) √ 3) √ 4) √ 5) √
x2 + 5 7 − x2 x2 − 8 x2 + 2 x2 − 6
1 cos x ex sec2 x 4αc + b2 b
6) √ 7) p 8) 2x 9) √ where K = , and t = x − 2α . Since by assumption, p(x) > 0 in its
9 − x2 sin2 x + 2 e + 10 4 − tan2 x 4α2
2
1 domain, we have K > 0. So let K = k , k > 0. Hence
10) . Z
dx 1
Z
dt
x[(log x)2 − 1] p becomes √ √ ,
p(x) α k − t2
2
B) Using the standard
r substitutions find2 the integrals
r of the following:
1 a−x x x which can be evaluated using the standard result.
1) √ 2) 3) √ 4) px + q
2
a +x 2 a + x 2
a −x 2 a√ x
− (ii) We rewrite by finding constants A and B such that px + q =
1 1 x3
a2 − x2 ax2 + bx + c
5) 2 6) √ 7) √ 8) . d
2
(x + a ) 2 2
x x −a 2 1−x 2 x2 A (ax2 + bx + c) + B i.e. px + q = A(2ax + b) + B. Hence,
dx
px + q A(2ax + b) B
Z Z Z
4.5 The integrals involving quadratic polynomials dx = dx + dx
ax2 + bx + c ax2 + bx + c ax2 + bx + c
Now we shall see the method to find integrals of = I1 + I2 , say.
1 px + q 1 px + q
, , √ , √ , Now if f (x) = ax2 + bx + c, then we get I1 as
ax2 + bx + c ax2 + bx + c 2
ax + bx + c ax2 + bx + c
where a 6= 0. (i) As a 6= 0, we may write Z ′
f (x)
I1 = A dx = A log |f (x)|,
f (x)
" 2  #
2 2 b c b 4ac − b2
p(x) = ax +bx+c = a[x + x+ ] = a x + + . (1)
a a 2a 4a2 and I2 can be evaluated by the procedure given in (i) above.
Calculus 91 92 Dr. V. V. Acharya

5x − 2
Z
(iii) Next, we have Example 4.14 Find I = dx.
3x2 + 2x + 1
px + q A(2ax + b) B Solution: We find constants A and B such that
Z Z Z
√ dx = √ dx + √ dx
2
ax + bx + c 2
ax + bx + c 2
ax + bx + c d
5x − 2 = A (3x2 + 2x + 1) + B or 5x − 2 = A(6x + 2) + B.
= J1 + J2 , say. dx
5 11
Z
f ′ (x) Thus 6A = 5 and 2A + B = −2 ⇒ A = , B = − .
6 3
p
Now we get J1 as J1 = A p dx = A[2 f (x)], and J2 can be evaluated
f (x) 5 11
6 (6x + 2)
Z Z
3
by the procedure given in (i). So I = dx − dx = I1 − I2 say. Hence,
3x2 + 2x + 1 3x2 + 2x + 1
dx 5 f ′ (x)
Z
5 6x + 2
Z Z
Example 4.11 Find . I1 = dx = dx where f (x) = 3x2 + 2x + 1
2x2 + 3x + 5 6 3x2 + 2x + 1 6 f (x)
 2 5 5
3 31 = log |f (x)| + c1 = log |3x2 + 2x + 1| + c1 ,
Solution: We ‘complete the square’ to obtain 2x2 + 3x + 5 = 2 x + + . 6 Z 6
4 8
11 dx 11 dx
Z
Hence, and I2 = =
! 3 3x2 + 2x + 1 9 x + 23 x + 13
2

dx 1 dx 1 1 x + 34
Z Z  
11 dx 11 3x + 1
Z
−1
= = · √ tan √ +c = = √ tan −1
√ + c2 .
2x2 + 3x + 5 2 (x + 34 )2 + 31 16
2 ( 31 ) 31
9 (x + 13 )2 + 29 3 2 2
4 4
 
2 −1 4x + 3 Hence,
= √ tan √ + c.  
31 31 5 2 11 −1 3x + 1
I = I1 − I2 = log |3x + 2x + 1| − √ tan √ + c.
6 3 2 2
dx
Z
Example 4.12 Find √ . 4x + 1
Z
3x2 + x − 1 Example 4.15 Find I = √ dx.
5x2 + 2x − 3
Solution: We have 3x2 + x − 1 = 3 (x + 16 )2 − 13
 
36 . Hence, Solution: We find constants A and B such that 4x + 1 = A(10x + 2) + B. By
2 1
comparing the coefficients, we get 10A = 4 and 2A + B = 1 ⇒ A = , B = .
dx 1 dx
Z Z
√ √ 5 5
= q 2
Z
(10x + 2) 1
Z
dx
3x2 + x − 1 3 (x + 16 )2 − 13 Hence, I = √ dx + √ = I1 + I2 , say.
36 5 5
r 5x2 + 2x − 3 5x2 + 2x − 3
1 1 1 2 13 2
Z
10x + 2 2
Z
f ′ (x)
= √ log (x + ) + (x + ) − + c.

3 6 6 36 I1 = √ dx = dx where f (x) = 5x2 + 2x − 3
5 5
p
5x2 + 2x − 3 f (x)

2 p 4p 2
dx
Z
= [2 f (x)] + c1 = 5x + 2x − 3 + c1 ,
Example 4.13 Find I = √ . 5 5
7 + 5x − 2x2
1 dx 1 dx
Z Z
Solution: We have 7 + 5x − 2x2 = 2 81 5 2
 
16 − (x − 4 ) . Hence, I2 = √ = √ q
5 2
5x + 2x − 3 5 5 (x + 15 )2 − 16
  25
1 dx 1 4x − 5
Z
= √ sin−1

I = √ + c.
r
2
q
2 9 1 1 1 16

81
− (x − 54 )2 = √ log (x + ) + (x + )2 − + c2 .

16 5 5 5 5 25
Calculus 93 94 Dr. V. V. Acharya
Z
Hence, dx
4.6 Integrals of the type
r a + b cos x + c sin x
4p 2 1 1 1 2 16
I= 5x + 2x − 3 + √ log (x + ) + (x + ) − + c.

dx
Z
5 5 5 5 5 25 Integrals of the type I = can be reduced to the form
a + b cos x + c sin x
dx
Z
Z r
x+1 by using the substitution tan x2 = t. Here we use the relations
Example 4.16 Find I = dx. px2 + qx + r
x+3 1 − t2 2t x 1 x
Solution : cos x = , and sin x = . Also, tan = t gives sec2 dx =
1 + t2 1 + t2 2 2 2
2dt
Z r
x+1
Z
x+1 1
Z
2x + 2 dt i.e. dx = . Hence,
dx = √ dx = √ dx 1 + t2
x+3 x2 + 4x + 3 2 x2 + 4x + 3
1 2dt
Z
1 2x + 4 dx
Z Z
= √ dx − √ = I1 − I2 , say. I =
2 2
x + 4x + 3 2
x + 4x + 3 a + b[ 1+t2 ] + c[ 1+t2 ] + t2
1−t2 2t 1
1 2
Z Z
= 2 2
2dt = 2
dt.
a(1 + t ) + b(1 − t ) + 2tc (a − b)t + 2tc + (a + b)
1 2x + 4 1 f ′ (x)
Z Z
I1 = √ dx = dx where f (x) = x2 + 4x + 3 This integral can be evaluated by the method discussed earlier.
2 2
p
x2 + 4x + 3 f (x)
dx
Z
1 p p Example 4.17 Find .
= [2 f (x)] + c1 = x2 + 4x + 3 + c1 , 2 + sin x + cos x
Z2 x 1 x 2dt
dx dx Solution: Put tan = t so that sec2 dx = dt or dx = .
Z
I2 = √ = √ 2 2 2 1 + t2
x2 + 4x + 3 x2 + 4x + 4 − 1 2t 1−t 2
Z
dx Also sin x = , cos x = . Therefore,
1 + t2 1 + t2
p
= p = log |(x + 2) + x2 + 4x + 3| + c2
2
(x + 2) − 1 2
dx 1+t2 dt dt
Z Z Z
= 2t 1−t 2 = 2 2
Thus, 2 + sin x + cos x 2 + 1+t2 + 1+t2 t + 2t + 3
p p
I= x2 + 4x + 3 − log |(x + 2) + x2 + 4x + 3| + c. √ 1 + tan x2
 
dx dt
Z Z
−1
= 2 = 2 tan √ + c.
2 + sin x + cos x (t + 1)2 + 2 2
Exercise Set 4.4
dx
Z
Integrate the following w.r.t. x : Example 4.18 Find .
1 1 1 1 5 − 4 sin x
1) 2 2) 2 3) 2 4) √ x x 2dt
2x − 4x + 1 3x + 2x + 1 4x − x + 2 2
3x − 4x − 5 Solution: Put tan = t so that 12 sec2 dx = dt or dx = . Hence,
1 2x + 5 1−x 2 2 1 + t2
1
5) √ 6) √2x2 −4x+7 7) 2 8) 2
1 + 4x − 5x2 3x + x + 1 2x + 3xr+ 5 Z
dx
Z 2dt Z
2dt 2
Z
dt
1+t2
3x + 2 6x + 7 2x + 5 x+1 = 2t = 2 + 5 − 8t
= 4 2 9
9) 2 10) √ 11) √ 12) 5 − 4 sin x 5 − 4( 1+t 2 ) 5t 5 (t − 5 ) + 25
4x + x − 1 2
2x − 4x + 5 1 − 3x + 2x 2 9−x
5 tan x2 − 4
   
1 sec2 x cos x 2 5t − 4 2
13) p 14) 15) p . I = tan−1 √ + c = tan−1 √ + c.
x (log x)2 + 3 2 tan2 x + 3 tan x + 5 3 9 3 9
9 + 8 sin x − sin2 x
Calculus 95 96 Dr. V. V. Acharya

dx
Z
Example 4.19 Find . is called a proper fraction if degree of N (x) < degree of D(x). And if degree of
1 + 2 cos x N (x) ≥ degree of D(x), then by long division we get N (x) = Q(x)D(x)+R(x)
x 2dt 1 − x2 where the quotient Q(x) and the remainder R(x) are polynomials in x and either
Solution: Put tan = t ∴ dx = , cos x = . Therefore,
2 1 + t2 1 + t2 R(x) is the zero polynomial or the degree of R(x) < degree of D(x). Hence in
N (x) N (x) R(x)
2dt this case, the improper fraction can be expressed as = Q(x)+
dx dt D(x) D(x) D(x)
Z Z Z
1+t2
= =2 R(x)
1 + 2 cos x 1 + 2( 1−t
2
1+t2 )
3 − t2 such that where the degree of R(x) < degree of D(x) i.e. such that is a
√ √
3 + tan x
D(x)
2 3 + t 1 2 proper fraction.
= √ log √ + c = √ log √ + c.

2 3 3 − t 3 3 − tan x2 Therefore, since integrating polynomials of the type Q(x) is easy, integrating a
rational function is reduced to integrating a proper fraction.
dx N (x)
Z
Example 4.20 Find . Suppose that f (x) = is a proper fraction. Then it can be shown that we
1 + cos α cos x D(x)
1−t2 can factorize its denominator D(x) as a finite product of real linear factors like
Solution: Put tan x2 = t Hence, dx = 1+t2dt
2 , cos x = 1+t2 .
ax + b and quadratic factors like px2 + qx + r. Here it must be remembered that
2dt we assume that such a quadratic factor is not further factorisable into real linear
dx dt
Z Z Z
1+t2
I= = =2 . factors: we say that such a quadratic factor is irreducible. Thus the quadratic
1 + cos α cos x 1+ cos α( 1−t
2
1+t2 )
(1 + cos α) + (1 − cos α)t2
polynomial x2 + 1 is irreducible; it cannot be factored into real linear factors
(although x2 + 1 factors as x2 + 1 = (x + i)(x − i), using complex numbers).
dt 1 dt We shall state the method of expressing such a proper fraction into partial
Z Z
I = 2 α 2 α 2 = 2 α 2 α fractions in the following three cases according to whether the denominator D(x)
2
2 cos 2 + 2(sin 2 )t sin 2 cot 2 + t2
  consists of
1 1 −1 t
= × tan +c 1) Non-repeated linear factors 2) repeated linear factors
sin2 α2 cot α2 cot α2
 x α 3) Non-repeated irreducible quadratic factors.
= 2 cosec α · tan−1 tan · tan + c.
2 2 In these cases, expressing f (x) into partial fractions means expressing f (x) as a
sum of related proper fractions as follows:
Exercise Set 4.5
Find constants A, B, C etc. such that
Integrate the following w.r.t. x : px + q A B
1 1 1 a) f (x) = = + , a 6= b.
1) 2) 3) (x − a)(x − b) x−a x−b
3 sin x + 4 cos x + 5 3 + 2 sin x + cos x 13 + 3 cos x + 4 sin x
1 1 1 1 px + q A B
4) 5) 6) 7) b) f (x) = 2
= + .
5 − 13 sin x 3 + 2 cos x 1 + 2 sin x 5 − cos x (x − a) x − a (x − a)2
1 1 1
8) 9) 10) px2 + qx + r A B C
4 + 3 cos 2x cos α − cos x 1 + sin α sin x c) f (x) = = + + .
(x − a)(x − b)(x − c) x−a x−b x−c
( Here a, b, c are all distinct.)
4.7 Integration by partial fractions px2 + qx + r A B C
d) f (x) = = + + . ( Here a 6= b.)
Recall that if N (x) and D(x) are polynomials with real coefficients, then a func- (x − a)2 (x − b) x − a (x − a)2 x−b
N (x) N (x) px2 + qx + r A Bx + C
tion of the form f (x) = is called a rational function. This fraction e) f (x) = = + .
D(x) D(x) (x − a)(x2 + bx + c) x − a x2 + bx + c
Calculus 97 98 Dr. V. V. Acharya

3x − 1
Z
(Here x2 + bx + c is supposed to be irreducible.) Example 4.23 Find I = dx.
px2 + qx + r A Bx + C Dx + E (x − 5)2
f) f (x) = 2 2
= + 2 + 2 . 3x − 1 A B
(x − a)(x + bx + c) x − a x + bx + c (x + bx + c)2 Solution: Let = + . Thus, 3x − 1 = A(x − 5) + B.
(x − 5)2 x − 5 (x − 5)2
(Here x2 + bx + c is supposed to be irreducible.)
Putting x = 5 we get B = 14. Also comparing coefficients of x weZget A = 3.
The following points should be noted regarding the factors of D(x) : 3x − 1 3 14
Z
dx dx
(i) As in a),c),d),e), to each non-repeated linear factor like ℓx + m, there corre- Thus, 2
= + 2
. Hence, I = 3 + 14 .
(x
Z − 5) x − 5 (x − 5) x−5 (x − 5)2
sponds a single fraction A/(ℓx + m) where A is a real number. 3x − 1 14
(ii) As in b),d), to each repeated linear factor like (ℓx + m)r , there corresponds a Hence, dx = 3 log |x − 5| − + c.
(x − 5)2 (x − 5)
sum of r fractions Ai /(ℓx + m)i (i = 1, . . . , r), Ai being real numbers.
(iii) As in e), to each non-repeated quadratic factor like ℓx2 + mx + n, there Z
x2 + 5x + 3
corresponds a single fraction (Ax + B)/(ℓx2 + mx + n) where A, B are real Example 4.24 Find I = dx.
(x − 2)2 (2x + 1)
numbers. x2 + 5x + 3 A B C
(iv) It can be shown that an expression of f (x) into partial fractions is unique. Solution: Let = + + . Hence,
(x − 2)2 (2x + 1) x − 2 (x − 2)2 2x + 1
2 2
 
dx 1 x−a x + 5x + 3 = A(x − 2)(2x + 1) + B(2x + 1) + C(x − 2)
Z
(i)
Example 4.21 Show that = log + c. 17
x2 − a2 2a x+a Putting x = 2 we get, 17 = 5B ⇒ B = −1
. Putting x = 2 we get,
Solution. Note that  2 5
1 3 25 3
. Also comparing the coefficients of x2
 
dx 1 1 1 C − −2 ⇒ = C ⇒ C = 25
= − . 2 4 4
x2 − a2 2a x − a x + a 1
on both sides of equation (i) we get 1 = 2A ⇒ A = .
Integrating, both sides, we get the required result. 2
2 1 17 3
x + 5x + 3
... = 2 + 5
+ 25
. Hence,
(2x2 + 3)dx
Z
Example 4.22 Find I = . (x − 2)2 (2x + 1) x − 2 (x − 2)2 (2x + 1)
(x − 1)(x − 2)(x + 3)
1 dx 17 dx 3 dx
Z Z Z
Solution: Note first that the integrand is a proper fraction.
I = + + .
2x2 + 3 A B C 2 x−2 5 (x − 2)2 25 2x + 1
Let = + + .
(x − 1)(x − 2)(x + 3) x−1 x−2 x+3
 
1 17 −1 3 log |2x + 1|
On clearing the denominators (i.e. on multiplying by the common denominator ... I = log |x − 2| + + + c.
2 5 x−2 25 2
(x − 1)(x − 2)(x + 3)), we get the identity:
1 3 17
∴ 2x2 + 3 = A(x − 2)(x + 3) + B(x − 1)(x + 3) + C(x − 1)(x − 2). ... I = log |x − 2| + log |2x + 1| − + c.
5 2 50 5(x − 2)
Putting x = 1 in this identity, we get 5 = −4A ⇒ A = − .
4
x2 + 3
Z
7
Putting x = 2 we get 7 = 5B ⇒ B = . Putting x = −3 we get 21 = 20C ⇒ Example 4.25 Find I = dx.
5 (x − 1)(x2 + 4)
21 2x2 + 3 5 1 7 1 21 1 2
x +3 A Bx + C
C= . Hence, = − x−1 + + . Thus, Solution: Let = + 2 . Hence,
20 (x − 1)(x − 2)(x + 3) 4 5 x − 2 20 x + 3 (x − 1)(x2 + 4) x−1 x +4
−5 dx 7 dx 21 dx
Z Z Z
I = + + x2 + 3 = A(x2 + 4) + (Bx + C)(x − 1) = (A + B)x2 + (C − B)x + (4A − C).
4 x − 1 5 x − 2 20 x + 3
−5 7 21 Comparing coefficients of x2 , x and constants on both the sides we get
= log |x − 1| + log |x − 2| + log |x + 3| + c
4 5 20 A + B = 1, C − B = 0 and 4A − C = 3. Since, B = C we get A + C = 1,
Calculus 99 100 Dr. V. V. Acharya

4 1 1 A B
4A − C = 3. Solving these two equations we get 5A = 4 ⇒ A = and = = + . Hence,
5 (x2 − 1)(x2 + 2) (t − 1)(t + 2) t−1 t+2
1 x2 + 3 4 1 1 (x + 1)
C= = B. Thus, = + . Hence, 1 = A(t + 2) + B(t − 1).
5 (x − 1)(x2 + 4) 5 x − 2 5 x2 + 4
Thus, by comparing the coefficients, we get A + B = 0 and 2A − B = 1. Hence,
x2 + 3 4 dx 1 x+1 1 1 1 1 1 1
Z Z Z
I = dx = + dx. A = −B = . Thus, 2 = − . Hence,
(x − 1)(x2 + 4) 5 x−1 5 x2 + 4 3 (x − 1)(x2 + 2) 3 x2 − 1 3 x2 + 2
Z  
4 1 2x 2 Z
dx 1
Z
dx 1
Z
dx
Thus, I = log |x − 1| + + dx = −
5 10 x2 + 4 x2 + 4 (x2 − 1)(x2 + 2) 3 x2 − 1 3 x2 + 2
4 1 1 x 1

x − 1

1

x

= log |x − 1| + log |x2 + 4| + tan−1 ( ) + c. = log − √ tan −1
√ + c.
5 10 10 2 6 x + 1 3 2 2
R dx Z
x2
Example 4.26 Evaluate I = . Example 4.28 Evaluate I1 = dx.
x3+1 x4 − x2 − 2
Solution. Note that 2
Solution. Note that if we substitute z = x then
1 A Bx + C x2 z z A B
= + . = 2 = = +
x3 + 1 x + 1 x2 − x + 1 x4 − x2 − 2 z −z−2 (z − 2)(z + 1) z−2 z+1
Hence, 1 = A(x2 − x + 1) + (Bx + C)(x + 1).
Thus, z = A(z + 1) + B(z − 2). Comparing the coefficients, we get A + B = 1
1 1 2
Putting x = −1, we get 3A = 1. Hence, A = . Comparing the coefficient of and B = 2A. Hence A = , B = . Hence, the given integral
3 3 3
2 1 1
Z
dx 2
Z
dx
x , we get A + B = 0. Thus, B = − . Similarly, comparing the constant term, I1 = +
3 3 2
x +1 3 2
x −2
2
we get A + C = 1. Thus, C = . Hence, √ !
3 1 −1 2 1 x− 2
= tan (x) + √ log √ + c.
1
Z
dx 1
Z
x−2 3 32 2 x+ 2
I = − dx
3 x+1 3 x2 − x + 1 Z
x2
1 1
Z
(2x − 1) − 3 Example 4.29 Evaluate I2 = dx.
= log(x + 1) − dx x + x2 − 2
4
3 6 x2 − x + 1 Solution. Note that if z = x2 then
1 1 2x − 1 1 dx
Z Z
= log(x + 1) − dx + x2 z z A B
3 6 2
x −x+1 2 2
x −x+1 = 2 = = +
x4 2
+x −2 z +z−2 (z + 2)(z − 1) z+2 z−1
1 1 1 dx
Z
2
= log(x + 1) − log(x − x + 1) + √ Thus, z = A(z − 1) + B(z + 2). Comparing the coefficients, we get A + B = 1
3 6 2 (x − 2 ) + ( 23 )2
1 2
2 1
and 2B = A. Hence A = , B = . Hence, the given integral
1 1 1 2 2x − 1 3 3
= log(x + 1) − log(x2 − x + 1) + √ tan−1 √ + c.
3 6 2 3 3 2 dx 1 dx
Z Z
I2 = +
Z
dx 3 x2 + 2 3 x2 − 1
Example 4.27 Find I = .
   
(x2 − 1)(x2 + 2) 2 x 11 x−1
= √ tan−1 √ + log + c.
Solution: For the sake of partial fractions only, take x2 = t. Hence, 3 2 2 32 x+1
Calculus 101 102 Dr. V. V. Acharya
Z
dx cos x x
Example 4.30 Evaluate I = . 13) 14)
(x2 + a2 )(x2 + b2 ) (1 − sin x)(2 − sin x) (x2 + 3)(x2 − 5)
1 A B 3x + 1 sec2 x
Solution. Put z = x2 . Note that = + . Hence, 15) 16) 2
(z + a2 )(z + b2 ) z + a2 z + b2 (x − 1)(x − 2)(x − 3) 3 tan x + 4 tan x + 1
2 2
1 = A(z + b ) + B(z + a ). Comparing the coefficient of z and constant term, sin 2x 1 + log x
17) 18)
we get cos2 x + 5 cos x + 6 x(2 + log x)(3 + log x)
1 1
A + B = 0 and 1 = Ab2 + Ba2 . 19) 20)
Z  sin x + sin 2x x(x − x3 − 12)
6
1 dx dx
Z
Hence, I = − 2. Show that
a2 − b 2 x2 + b2 x2 + a2
2x2 − 3x + 3 2
 Z
1 1  
−1 x 1 −1 x
 
= tan − tan + c. (a) 3 2
dx = 3 log |x| − − log |x − 1| + c.
a2 − b 2 b b a a x − 2x + x x−1
x 3 1 1
Z
x2 + 1
Z
(b) dx = − log |x + 1| −
Example 4.31 Evaluate I = dx (x + 1)2 (x + 2)(x − 1) 4 2x+1
x − 11x2 + 1
4
2 1
Solution. Note that here if we want to use partial fractions, then we have to ‘ + |x + 2| + log |x − 1| + c.
factorize a polynomial of degree 4. Instead of this, we divide numerator and 3 12
1 1 1 1 1
Z
denominator by x2 . Thus (c) dx = log |x + 1| − + tan−1 (x) + c.
x3 + x2 + x + 1 2 4 1 + x2 2
1 + x12 1 + x12
Z Z
I = dx = dx (d)
x2 − 11 + x12
2
x − x1 − 9  
x+1 2 2x + 1
Z
−1
  dx = √ tan √
1 1 (x2 + x + 2)(x2 + 4x + 5) 3 7 7
Put u = x − . Thus, du = 1 + 2 dx.
x x 1 −1
− tan (x + 2) + c.
3

du 1 u − 3
Z
I= = log u + 3 + c.

u2 − 9 6 3. Evaluate
Z the following:
1 1
Z
We substitute the value of u in terms of x and simplify to get (i) dx (ii) dx
3 2 2
Z (x − 1)2 (x + 1) Z (x2 −2 7x +2 12)
2
1 x − 3x + 1
I = log 2
+ c. x − 3x + 3 x (x + a )
6 x + 3x + 1 (iii) 3 2
dx (iv) dx
Z x − 7x 2 + 16x − 12 x2 + b2
x 1
Z
Exercise Set 4.6
(v) 2 + 1)(2x2 + 1)
dx (vi) 2 + 1)
dx
1. Integrate the following w.r.t. x : (x x(x
2
x +2 (x + 1)2
Z Z
1 x − x2 2x2 − 1 (vii) dx (viii) dx
1) 2 2) 2 3) 4 x + x2 + 1
4
x + 5x + 4 x − 2x − 3 (x2 + 4)(x2 − 5) Z x(x2 − 1)
x +1 x2
Z
x3 + 4x2 + 7x + 2 x2 + 1 x (ix) dx (x) dx
4) 5) 2 6) 2 4 2 (x − 1) (x2 − 2x + 4)
2
2
x + 3x + 2 x − 5x + 6 (x + 1)(x − 1) Z x −x + 1
x4 1
Z
3x + 5 x2 + x + 1 5x + 1 (xi) dx (xii) dx
7) 2
8) 2 9) (x − 1) 2 (x2 + 4) (x 2 + 1)(x − 1)2
(x + 1)(x − 1) x (x + 2) (x + 2)2 (x + 3)
1 1
Z Z
ex + 1 2x2 + 5 1 (xiii) dx (xiv) dx
10) x x 11) 12) 4 3 2
x (x + a 2 x(x + 1)2
2
e (e − 3) (x + 2)(x2 + 2) x −1
Calculus 103 104 Dr. V. V. Acharya
Z
4.8 Integration by parts Example 4.32 Find x log x dx.
1 x2
Z
In this section, we shall see a method to integrate a function which is expressed ′
Solution: Here let u = log x, v = x. Hence, u = , vdx = . Hence,
as a product of two functions. This method is known as method of integration by x 2
parts and is given in the following theorem.
Theorem 4.2 Suppose that on the interval [a, b], u = f (x) and v = g(x) are
Z Z Z Z
functions having continuous derivative. Then x log xdx = u vdx − (u′ · vdx)dx

x2 log x 1 x2 x2 log x 1
Z  Z  Z  
du
Z Z Z
uvdx = u vdx − vdx dx. = − · dx = − xdx
dx 2 x 2 2 2
x2 log x x2
Z Z
Proof: Let w = vdx so that dw/dx = v. By the rule of differentiation of ... x log dx = − +c
2 4
product of two functions we have
Z
d du dw
(u w) = w+u . Example 4.33 Find I = x tan−1 xdx.
dx dx dx
1 x2
Z
Integrating both sides w.r.t. x we have Solution: Let u = tan−1 x, v = x. Hence, u′ = , vdx = .
Z   Z   1 + x2 2
du dw
uw = w dx + dx
dx dx Z Z Z
Z
x tan−1 xdx (u′ · vdx)dx
   
dw du
Z Z
= u vdx −
i.e. u w dx = uw − w dx.
dx dx
x2 1 x2
Z
Substituting for w in the above equation we get = tan−1 x · − 2
· dx
Z  Z  2 1+x 2
du
Z Z
x2 tan−1 x 1 h 1 i
Z
uvdx = u vdx − vdx dx = − 1− dx
dx 2 2 1 + x2
which is the required result. This result is also stated as: x2 tan−1 x x tan−1 x
= − + + c.
Z Z Z h Z 2 2 2
i
f (x)g(x) dx = f (x) g(x) dx − f ′ (x) g(x) dx dx. (1) Z
Example 4.34 Find I = x sin x dx.
Here u = f (x) is the first function and v = g(x) is the second. In words, in (1), Z
we have the first function times the integral of the second minus the integral of Solution : Let u = x, v = sin x. ... u′ = 1, vdx = − cos x. Hence,
‘the derivative of the first function times the integral of the second’.
In the given product, uv, we have to choose the first function and the second Z Z Z Z Z
function suitably. For example, for x sin xdx = u vdx − (u′ · vdx)dx = −x cos x − (1)(− cos x)dx
Z Z
x log x dx, we take u = log x, v = x. log x dx, we take u = log x, v = 1. = −x cos x + sin x + c.
Z Z
x sin xdx, we take u = x, v = sin x. ex cos xdx, we take u = cos x, Z
x
v = e . The method is illustrated in the following examples: Example 4.35 Find I = x2 ex dx.
Calculus 105 106 Dr. V. V. Acharya
Z
Solution: Let u = x2 , v = ex . ... u′ = 2x, vdx = ex . Hence,
Z Z Z Z
Z Z Z Z sin−1 xdx = u vdx − (u′ · vdx)dx
2 x ′
x e dx = u vdx − (u · vdx)dx Z
x
= x sin−1 x − √ dx
1 − x2
Z Z
= x2 ex − 2xex dx = x2 ex − 2 xex dx (i) 1
Z
−2xdx
= x sin−1 x + √ . (i)
Z Z 2 1 − x2
−2xdx f ′ (x)
Z Z
Now for xex dx, let u = x, v = ex . Hence, u′ = 1, vdx = ex . Hence, Now √ = p dx [where f (x) = 1 − x2 ]
Z Z 1 − x2 f (x)
xex dx = xex − ex dx = xex − ex . Substituting in (i) we get p p
Z = 2 f (x) = 2 1 − x2 .
x2 ex dx = x2 ex − 2[xex − ex ] + c = x2 ex − 2xex + 2ex + c. Z

Substituting in (i) we get sin−1 xdx = x sin−1 x + 1 − x2 + c.
Z Z
log x dx.
p
Example 4.36 Find Example 4.39 Find I = log(x + x2 + a2 )dx.
1
Z √
Solution: Let u = log x, v = 1. ... u′ = , vdx = x. Solution: Let u = log(x + x2 + a2 ), v = 1.
x
Z   "√ #
1
Z Z Z Z
log xdx = u vdx − (u′ · vdx)dx = x log x − · x dx ′ 1 x2 + a2 + x 1
x u = √ × √ = √ .
Z x + x2 + a2 2
x +a 2 x + a2
2

... log xdx = x log x − x + c.


Therefore
Z p Z Z Z
I = x2 + a2 )dx = u vdx − (u′ · vdx)dx
log(x +
Z
Example 4.37 Find I = x3 log x dx.
x
p Z
1 x4 x log(x + x2 + a2 ) −
Z
= √ dx
Solution: Let u = log x, v = x3 . ... u′ = , vdx = . x2 + a2
x 4
1 2x
p Z
= x log(x + x2 + a2 ) − √ dx.
Z
3
Z Z

Z 2 x + a2
2
x log xdx = u vdx − (u · vdx)dx
Thus,
x4 x3 x4 x4
Z
= · log x − dx = log x · − + c. Z
1
Z
f ′ (x)
4 4 4 16
p p
log(x + x2 + a2 )dx = x log(x + x2 + a2 ) − dx
2
p
f (x)
Z p
Example 4.38 Find sin−1 xdx. where f (x) = x2 + a2
Z p p p
1
Z
Solution: Let u = sin −1 ′
x, v = 1. u = √ , vdx = x. Hence, ... log(x + x2 + a2 )dx = x log(x + x2 + a2 ) − x2 + a2 + c.
1 − x2
Calculus 107 108 Dr. V. V. Acharya

√ eax a2
Z
Example 4.40 Find sin xdx I = (b sin bx + a cos bx) − 2 I.
b 2 b

Solution: Put x = t i.e. x = t2 . Hence, dx = 2tdt. Thus,  a2  eax
Hence, 1 + 2 I = (b sin bx + a cos bx).
Z

Z Z b b2
sin xdx = 2t sin tdt = 2 t sin tdt eax
Thus, I = (b sin bx + a cos bx) + c.
a + b2
2
= 2[−t cos t + sin t] + c [Refer to Example 4.34]
Z
√ √ √ √ If a 6= 0, then putting a = t cos α, b = r sin α, we get the second form of the
... sin xdx = 2[− x cos x + sin x] + c. result.
eax
Z
(2) ea x sin bxdx = 2 · (a sin bx − b cos bx)
Exercise Set 4.7 a + b2
eax
 
Integrate the following w.r.t. x : b
log(log x) = √ sin bx − tan−1 , a 6= 0.
1) (log x)2 2) x sec2 x
3) xe−x 4) x2 tan−1 x 5) a2 + b 2 a
√ x Z Z
x Proof: Let J = eax sin bx and I = eax cos bx. Integrating by parts we get
6) x2 log(x + 2) 7) tan−1 x 8) 9) x2 cos2 x 10) esin x · sin 2x.
1 + cos x
cos bx cos bx
Z Z
4.9 Standard Integrals J = eax sin bx dx = eax · (− ) − aeax · (− ) dx
b b
cos bx a cos bx a
Z
The following are some standard integrals which can be obtained by the method = −eax · + eax · cos bx dx = −eax · + · I.
of integration by parts: b b b b
Again integrating by parts, we have
ax
e
Z
(1) eax cos bxdx = · (a cos bx + b sin bx) + c
a2+ b2 sin bx sin bx
Z
I = eax · − aeax · dx
eax b b
 
b
= √ cos bx − tan−1 + c, where a 6= 0.
sin bx a sin bx a
Z
a2 + b 2 a
= eax · − eax · sin bx dx = eax · − · J.
Z b b b b
Proof: Let I = eax cos bxdx. Integrating by parts we get cos bx a sin bx a
h i
So J = −eax · + eax · − ·J
b b b b
ax 2
sin bx a
Z e a
I = eax · − eax · sin bx dx. = (−b cos bx + a sin bx) − 2 J.
b b b2 b
 a2  eax
Again integrating by parts, we have ... 1+ J = (−b cos bx + a sin bx).
b2 b2
Z
cos bx
Z
cos bx eax
J = eax sin bx dx = eax · (− ) − aeax · (− ) dx ... J = (a sin bx − b cos bx).
b b a + b2
2

cos bx a cos bx a
Z
If a 6= 0, then putting a = t cos α, b = r sin α, we get the second form of the
= −eax · + eax · cos bx dx = −eax · + · I.
b b b b result.
xp 2 a2
Z p
sin bx a h cos bx a i p
So I = eax · − − eax · + ·I (3) x2 + a2 dx = x + a2 + log(x + x2 + a2 .)
b b b b 2 2
Calculus 109 110 Dr. V. V. Acharya

dx
Z p Z p p Z p Z
Proof: Let I = x2 + a2 dx = x2 + a2 · 1 dx. Integrating by parts we Hence, I = x a2 − x2 − a2 − x2 dx + a2 √
a − x2
2
get
dx
p Z
Z 2 So I = x a2 − x2 + a2 √ −I
x x + a2 − a2
Z
a2 − x2
p p
I = x x2 + a2 − √ · xdx = x x2 + a2 − √ dx.
x2 + a2 x2 + a2
p x
Thus, 2I = x a2 − x2 + a2 sin−1 + 2c
a
dx
Z p Z
2
xp 2 a x
p
Hence, I = x x2 + a2 − x2 + a2 dx + a2 √ I = a − x2 + sin−1 + c.
x + a2
2
2 2 a
dx
p Z
So I = x x2 + a2 + a2 √ −I Alternatively, put x = a sin θ, so that dx = a cos θdθ. Hence
x2 + a2
p p
Thus, 2I = x x2 + a2 + a2 log(x + x2 + a2 ) + 2c. Z
a2
Z
xp 2 a2 I = a2 cos2 θ dθ = (1 + cos 2θ) dθ
2
p
Hence, I = x + a2 + log(x + x2 + a2 ) + c.
2 2 a2 h sin 2θ i a2 a2
Z p
x a2 = +θ = sin θ cos θ + θ.
2 2 2 2
p p
(4) x2 − a2 dx = x2 − a2 − log(x + x2 − a2 .)
2 2 r
Z p Z p
x2 x
Proof: Let I = x2 − a2 dx = x2 − a2 · 1 dx. Integrating by parts we Now sin θ = x. So cos θ = 1− = sin−1 . So
a 2 a
get
x
Z r
a2 x x2 a2 x xp 2 a2 x
p
I = x x −a −2 2 √ · x dx I= · 1− + sin−1 + c = a − x2 + sin−1 + c.
x − a2
2
2 a a 2 2 a 2 2 a
x2 − a2 + a2
p Z
= x x2 − a2 − √ dx
x2 − a2
Z
(6) To find I = (a2 − x2 )3/2 dx.
dx
p Z p Z
= x x2 − a2 − x2 − a2 dx − a2 √ Using integration by parts, we get
x − a2
2

dx
p Z
So I = x x2 − a2 − a2 √ −I 3
Z
x − a2
2 I = x(a2 − x2 )3/2 − x(a2 − x2 )(−2x)dx
p p 2
2I = x x2 − a2 − a2 log(x + x2 − a2 ) + 2c Z
2 2 3/2
= x(a − x ) −3 (a2 − x2 )(−x2 )dx
xp 2 a2 p
I = x − a2 − log(x + x2 − a2 ) + c.
2 2
Z
= x(a2 − x2 )3/2 − 3 (a2 − x2 )(a2 − x2 − a2 )dx
x a2 x
Z p p
(5) a2 − x2 dx = a2 − x2 + sin−1 + c. Z p
2 2 a
Z p Z p = x(a2 − x2 )3/2 − 3I + 3a2 a2 − x2 dx.
Proof: Let I = a2 − x2 dx = a2 − x2 · 1 dx. Integrating by parts we
x 2 3a2
Z p
get Hence, I = (a − x2 )3/2 + a2 − x2 dx.
4 4
−x a2 − x2 − a2
p Z p Z
I = x a2 − x2 − √ · x dx = x a2 − x2 − √ dx. By applying the above formula, we get
2
a −x2 a2 − x2
Calculus 111 112 Dr. V. V. Acharya

For I1 , put t = x2 − x + 1. Then dt = (2x − 1)dx so that


Z √
x 2 3a2 x p 2

a2 x
 t3/2 2
I = (a − x2 )3/2 + a − x2 + sin−1 +c I1 = t dt = = (x2 − x + 1). Also,
4 4 2 2 a 3/2 3
x 2 3a2 x p 2 3a4 x
s 2
= (a − x2 )3/2 + a − x2 + sin−1 + c 1 3
Z p Z
4 8 8 a I2 = 2
x − x + 1 dx = x− + dx
2 4
ByZapplying the above formulas we can evaluate integrals of the type r r
1 1 1 2 3 3 1 1 3
(x − ) (x − ) + + log (x − ) + (x − )2 +
p
ax2 + bx + c dx and = + k.

(i) 2 2 2 4 8 2 2 4
Z
Thus I can be found by substituting in (A).
p
(ii) (px + q) ax2 + bx + c dx where a 6= 0. Thus for (i), we proceed as in
Exercise Set 4.8
§4.6 to express f (x) = ax2 + bx + c as a[(x + r)2 ± s2 ] if a > 0 or as −a[s2 ±
Integrate the following w.r.t. x :
(x + r)2 ] if a < 0. For (ii), find constants A, B such that
d 1) e√3x cos 4x 2) e2x sin
√ 3x cos x 3) √ e4x cos2 x 4) e−3x sin 2x √
px + q = A (ax2 + bx + c) + B i.e. as px + q = A(2ax + b) + B. 5) 4x2 + 5x + 1 6) √ 2x − x2 7) 3x2 − 2x − 1 8) (x + 1) 5 + 4x − x2
dx
Then the integral can be written as 9) cos(3 log x) 10) ex 5e2x − 4ex + 1.
Z p Z p
A (2ax + b) ax2 + bx + c dx + B ax2 + bx + c dx.
ex [f (x) + f ′ (x)] dx
R
4.10 Integrals of the form
For the first integral on the right, put t = ax2 + bx + c; the integral is of type (i). Z
Z p We will show that ex [f (x) + f ′ (x)] dx = ex f (x) + c. For this, let u = f (x),
Example 4.41 Find I = 3x2 + 5x − 1 dx. Z
Solution: We have v = e . Then u = f (x) and v dx = ex . Hence integrating by parts, we get
x ′ ′

√ Z √ Z
r r
5 1 5 37 Z Z Z Z
I = 3 x2 + x − dx = 3 (x + )2 − dx ex f (x) dx = u v dx − u′ v dx dx
3 3 6 36
√ h1
r
5 5 37
Z
= 3 (x + ) (x + )2 − = e f (x) − ex f ′ (x) dx + c
x
2 6 6 36
r
37 5 5 37 i ex f ′ (x) dx to the left we get
R
− log (x + ) + (x + )2 − + k.
So transferring the term
72 6 6 36 Z
Z p ex [f (x) + f ′ (x)] dx = ex f (x) + c.
Example 4.42 Find I = (x + 1) x2 − x + 1 dx.
To apply this result, note the form of the integrand: the multiplier of ex is the sum
Solution: Let A, B be constants such that
d of the function f (x) and the derivative of f (x). Sometimes the integrand can be
x + 1 = A (x2 − x + 1) + B i.e. x + 1 = A(2x − 1) + B. reduced to this form, if it is not already stated.
dx
Thus 1 = 2A and 1 = −A + B. So A = 1/2 and B = 3/2. Hence
Z
Example 4.43 Find I = ex (sin x + cos x) dx.
1 3
Z p Z p
I = 2
(2x − 1) x − x + 1 dx + x2 − x + 1 dx Solution: Here f (x) = sin x, so that f ′ (x) = cos x. Hence
2 2 Z
1 3
= I1 + I2 , say. (A) I = ex [f (x) + f ′ (x)] dx = ex f (x) + c = ex sin x + c.
2 2
Calculus 113 114 Dr. V. V. Acharya
Z
Example 4.44 Find I = ex (1 + tan x + tan2 x) dx. Exercise Set 4.9
Integrate the following w.r.t. x :
Solution: Here f (x) = tan x, so that f ′ (x) = sec2 x = 1 + tan2 x. 2−x 1 − sin x
1) ex (sin x − cos x) 2) ex (cot x + log(sin x)) 3) ex 2
4) ex ·
(1 − x) 1 − cos x
Z
So I = ex [f (x) + f ′ (x)] dx = ex f (x) + c = ex tan x + c. x2 + 1 2
x cos x + sin x x (x − x + 1) x (x − 1)
5) ex 6) e 7) e 8) e .
(x + 1)2 cos2 x (x2 + 1)3/2 (x + 1)3
xex
Z
Example 4.45 Find I = dx.
(x + 1)2
Solution: We have 4.11 Reduction Formulae
(x + 1) − 1
Z
I = ex · dx Integration by parts plays an important role in deriving reduction formulae. Note
(x + 1)2 that
h 1 Z Z
1 h 1 −1 i
Z i Z
= ex − 2
dx = ex + dx. udv = uv − vdu
x + 1 (x + 1) x + 1 (x + 1)2
1 −1 is a useful way to use integration by parts. We illustrate this with few examples.
Here f (x) = ⇒ f ′ (x) = so that
x+1 (x + 1)2 Z
ex sinn xdx
Z
I = ex [f (x) + f ′ (x)] dx = ex f (x) + c = + c. 4.11.1 I =
x+1
Z h 2 + sin 2x i R
We note that sin xdx = − cos x + c. Assume that n be a natural number greater
Example 4.46 Find I = ex dx.
1 + cos 2x than 1. Then, we write sinn x = sin x sinn−1 x. Then suing integration by parts,
2 + sin 2x 2 + 2 sin x cos x we get
Solution: We have = = sec2 x + tan x. So
1 + cos 2x 2 cos2 x Z Z
I = sinn xdx = sin x sinn−1 xdx
Z
I = ex [tan x + sec2 x] dx = ex tan x + c. Z
Z
log x = − cos x sinn−1 x − (− cos x)(n − 1) sinn−2 x(cos x)dx
Example 4.47 Find I = dx.
(1 + log x)2 n−1
Z
= − cos x sin x + (n − 1) sinn−2 x(cos2 x)dx
t
R t t
Solution: Put t = log x or x = e . Then we get I = (1+t) 2 e dt.

1 x Z
Hence by Example 4.46 above, I = et · = . = − cos x sinn−1 x + (n − 1) sinn−2 x(1 − sin2 x)dx
1+t 1 + log x
1 + x + x2
Z Z Z
−1
Example 4.48 Find I = etan x · dx. = − cos x sinn−1 x +(n − 1) sinn−2 xdx − (n − 1) sinn xdx
1 + x2
d tan−1 x −1 1
Solution: Note that e = etan x · . Hence, integrating by parts,
dx 1 + x2 Z Z
we get
Hence, n sinn xdx =
− cos x sinn−1 xdx + (n − 1) sinn−2 xdx
1
Z h i
−1
tan−1 x
I = etan x + x e dx Z
1 n−1
Z
1 + x2 Thus, sinn xdx = − cos x sinn−1 xdx+ sinn−2 xdx
Z
−1 −1
Z
−1 −1 n n
= etan x dx + x · etan x − (1)etan x dx = x · etan x + C.
We also have
Calculus 115 116 Dr. V. V. Acharya

sinn x cosm xdx


R
π π
4.11.2
n−1 n−1
Z 2
Z 2
In = sinn xdx = sinn−2 xdx = In−2 . Here, m, n are positive integers and we write Im,n = sinn x cosm xdx. There
R
0 n 0 n
are four cases:
Continuing in this way, we get Case (i) m, n are both even.

n−1 n−3 2 Put m = 2r, n = 2s. In this case, we can express the given integral in terms of
Z π2 
 × × ··· × if n is odd, either sin x or cos x. Thus,
In = n n
sin xdx = n − 1 n − 3 n − 2 3
1 π
× × ··· × × if n is even.
Z
0 
Im,n = sinm x(1 − sin2 x)s dx.

n n−2 2 2
cosn xdx can be integrated in a similar manner. We get
R
Case (ii) m, n are both odd.
1 n−1 Put m = 2r + 1, n = 2s + 1 and cos x = t. Hence, − sin xdx = dt. Thus,
Z Z
cosn xdx = sin x cosn−1 xdx + sinn−2 xdx
n n Z
Im,n = − (1 − t2 )r tn dt.
In this case also, we have
Since r is a positive integer, we can expand (1 − t2 )r using binomial theorem and
Z π2 Z π
n−1 2 n−1
In cosn xdx = cosn−2 xdx = In−2 . integrate term by term. We may also use the substitution, sin x = t.
0 n 0 n
Case (iii) m is even and n is odd
Continuing in this way, we get Put m = 2r, n = 2s + 1 and sin x = t. Hence, cos xdx = dt. Thus,
n−1 n−3 2
 Z
Z π2 
 × × ··· × if n is odd, Im,n = tm (1 − t2 )s dt.
In n
cos xdx = n − n n−2 3
1 n−3 1 π
0 
 × × ··· × × if n is even.
n n−2 2 2 Case (iv) m is odd and n is even Put m = 2r + 1, n = 2s and cos x = t.
Hence, − sin xdx = dt. Thus,
R4 √
Example 4.49 Evaluate x3 4x − x2 dx. Z
0 Im,n = (1 − t2 )r tn dt.
Solution. Let x = 4 sin2 θ. Hence, 4x − x2 = 16 sin2 θ cos2 θ and
dx = 8 sin θ cos θdθ. Hence, Another way to deal with this integral directly. Let u = sinm−1 x cosn x and
Z4 Zπ/2 dv = − sin xdx.
p
3
x 4x − x2 dx = 64 sin6 θ(4 sin θ cos θ)(8 sin θ cos θ)dθ
Z
Im,n = − udv
0 0
Z
Zπ/2 Zπ/2 = −uv + cos x (m − 1) sinm−2 x cosn+1 x − n sinm x cosn−1 x dx

8
= 2 11 2
sin θ(cos θ)dθ = 2 11
sin8 θ(1 − sin2 θ)dθ
0 0 = − sinm−1 x cosn+1 x + (m − 1)Im−2,n − (m − 1)Im,n − nIm,n
Zπ/2  
11 8 10 11 7 5 3 1 9 π
Hence, (m + n)Im,n = − sinm−1 x cosn+1 x + (m − 1)Im−2,n
= 2 (sin θ − sin θ)dθ = 2 1−
8642 10 2 1 m−1
0
i.e. Im,n = − sinm−1 x cosn+1 x + Im−2,n .
= 28π. m+n m+n
Calculus 117 118 Dr. V. V. Acharya

Exercise Set 4.10 Thus, In = x(x2 + a2 )n/2 − nIn + na2 In−1 . Hence,
I. Evaluate the following integrals:
R π/2 (n + 1)In = x(x2 + a2 )n/2 + na2 In−1 .Thus,
1. sin8 xdx 2. cos7 xdx 3. sin5 xdx 4. cos6 xdx 5. 0 sin6 xdx
R R R R

tann−1 x 1 n
II. Let In = tann xdx. Show that if n ≥ 2 then In =
R
− In−2 . In = x(x2 + a2 )n/2 + a2 In−1 .
n−1 n+1 n+1

xn eax dx.
R
Z
1 4.11.5
4.11.3 dx, n a positive integer
(x2 + a2 )n xn eax dx.
R
We start by setting: In =
n+1

1 Integrating by substitution: xn dx = d(x )


n+1 ,
Let u = 2 , dv = dx. Hence, integrating by parts, we get
(x + a2 )n 1
Z
In = eax d(xn+1 ).
Z
1 x
Z
x2 n+1
In = dx = + 2n dx Now integrating by parts:
(x2 + a2 )n (x2 + a2 )n (x2 + a2 )n+1
Z 2
x x + a2 − a2
= 2 2 n
+ 2n dx
(x + a ) (x2 + a2 )n+1
Z Z
ax n+1 n+1 ax
e d(x )=x e − xn+1 d(eax ) (3)
x 1
Z
2
= 2 2 n
+ 2nIn − 2na dx Z
(x + a ) (x + a2 )n+1
2
= xn+1 eax − a xn+1 eax dx, (4)
x
= + 2nIn − 2na2 In+1
(x + a2 )n
2

x (n + 1)In = xn+1 eax − aIn+1 ,


Hence, 2na2 In+1 = + (2n − 1)In . Thus,
(x2 + a2 )n shifting indices back by 1 (so (n + 1) → n, n → n − 1):
1 x 2n − 1 nIn−1 = xn eax − aIn ,
In+1 = + In .
2na2 (x2 + a2 )n 2na2
solving for In :
2 n/2
R 2
4.11.4 (x + a ) dx, n a positive integer 1 n ax
In = (x e − nIn−1 ) ,
2 2 n/2
Let u = (x + a ) , dv = dx. Hence, integrating by parts, we get a
Z so the reduction formula is:
In = (x2 + a2 )n/2 dx  
1
Z Z
xn eax dx = xn eax − n xn−1 eax dx .
n a
Z
2 2 n/2
= x(x + a ) − 2x2 (x2 + a2 )n/2−1 dx
2
Z
= x(x2 + a2 )n/2 − n (x2 + a2 − a2 )(x2 + a2 )n/2−1 dx
Z
= x(x2 + a2 )n/2 − nIn + na2 (x2 + a2 )n/2−1 dx
120 Dr. V. V. Acharya

7 22
5) log | sec(x − b)| − log | sec(x − a)| 6) − x − log |4 sin x + 5 cos x|
41 41
5 19
Hints and Answers 7) − log | cos x − sin x| 8) − x +
3 √6
log |3ex − 5| 9) 2 log |ex/2 − e−x/2 |
Exercise Set 2.1 3 13 2 3x 3x
10) x + log |3e3x − 4| 11) log |cosec − cot |
4 36 3 2 2
mb + na m 1 1
1. (a) x0 = ,θ = 12) √ log | sec x + tan x| 13) √ log | sec( π4 + 2x) + tan( π4 + 2x)|
m+n m+n 2 2 2
x x √ π x π x
(b) Rolle’s theorem is not applicable as the function is not differentiable 14) 2(sin − cos ) − 2 log | sec + + tan + |
at 0. 2 2 4 2 4 2
1
π 1 15) [x + log |(cos x + sin x)|] 16) − 21 log |1 − sin 2x|
(c) x0 =4 , θ = 4. 2
√ 1
(d) x0 = ab. 17) log |cosec (x + α) − cot(x + α)| where α = tan−1 (3/4)
5
(e) Rolle’s theorem is not applicable as the function is not differentiable 1
18) √ log |cosec (x + α) − cot(x + α)| where α = tan−1 (3/2)
at 1. 13
π 1 1 π π
(f) x0 = 6,θ = 13 . 19) √ log |cosec (x+ π4 )−cot(x+ π4 )| 20) √ log |cosec (x− )−cot(x− )|
2 2 4 4
(g) Rolle’s theorem is not applicable as the function is not continuous at
0 as well as at 1. Exercise Set 4.3
a+b 1 √ x √ 1 x
(h) x0 = 2 ,θ = 2. A] 1) log |x + x2 + 5| 2) sin−1 ( √ ) 3) log |x + x2 − 8| 4) √ tan−1 ( √ )
√ 7 2 2
2. Note that f satisfies conditions of Rolle’s theorem on the intervals [−3, −2], 1 x− 6 1 3+x p
2
[−2, −1] and [−1, 0]. Hence, in each of these intervals there is a root of 5) √ log | √ | 6) log | | 7) log | sin x + sin x + 2|
2 6 x+
 x6  6 3−x
f ′ (x) = 0.

1 e 1 (log x) − 1
8) √ tan−1 √ 9) sin−1 ( 12 tan x) 10) log
10 10 2 (log x) + 1
Note that in all the answers of Exercise sets 3.1 to 3.9 constant of integra-
tion c should be added. √ √ a2 x√ 2
B] 1) x2 + a2 2) a2 − x2 − a cos−1 (x/a) 3) sin−1 (x/a) − a − x2
Exercise Set 4.1 2 2
√ 1 h  x  ax i 1 −1 x
 
4) −a cos−1 ( xa )− ax − x2 5) 3 tan−1
p
1 1 4 + 2 6) sec
(i) log(1 + x2 ) (ii) (1 + log x)3 (iii) sin(x2 + 1) (iv) (x2 + x + 3)3/2 √2a a x + a2 a a
3 2 3 1 2 3/2
√ a 2 − x2
−1
1 7) (1 − x ) − 1 − x 8) −2 − a sin (x/a)
(v) log(ex + e−x ) (vi) − log | cos x − sin x| (vii) 3 x
2(1 − sin x)2
√ √ 3 Exercise Set 4.4
(viii) 2 log | x + 1| (ix) 2 tan x (x) − cot(xex ) (x) (x + sin x)4/3 √
4    
(xi) Put xex = t. Hence, I = cosec 2 tdt = − cot t + c. (xii) − cos(tan−1 x) 1 2x − 2 − 1 1 3x + 1 2 8x − 1
R
1) √ log | √ | 2) √ tan−1 √ 3) √ tan−1 √
1 log(a2 sin2 x + b2 cos2 x) 1 2 2 2x − 2 + 1 2 2 31 31
(xiii) tan−1 x3 (xiv) (xv) sec3 x r
3 2
a −b 2 3 1 2 4 5 1
4) √ log (x − ) + x2 − x − 5) √ sin−1 [(5x − 2)/3]

Exercise Set 4.2 3 3 3 3 5
r
1) x cos 2a − sin 2a log | sin(x + a)| 2) x sin 3a + cos 3a log | sin(x + a)| 1 7

6) √ log (x − 1) + x2 − 2x +

3) x cos(a − b) + sin(a − b) log | tan(x − b)| 4) log | sin(x − a)| − log | sec(x − a)| 2 2

119
Calculus 121 122 Dr. V. V. Acharya

1 10

6x + 1

13 1 1 √
7) log |3x2 + x + 1| + √ tan−1 √ 11) log |x + 2| − log |x2 + 2| + √ tan−1 (x/ 2)
3 11 6 12 3 2
11
1 x−1 1 1 − sin x 1 x2 − 5

1 2 7 −1 4x + 3 −1
8) − log |2x + 3x + 5| + √ tan √ 12) log | | − tan x 13) log | | 14) log | 2 |
4 2 31 31 4 x+1 2 2 − sin x 16 x +3
3 2 13 8x + 1 − √65 15) 2 log |x
− 1| − 7 log |x − 2| + 5 log |x − 3|
1 3 tan x + 1

9) log 4x + x − 1 + √ log √

8 8 65 8x + 1 − 65 16) log 17) 4 log | cos x + 2| − 6 log | cos x + 3|
r 2 tan x + 1
p 13 5 18) 2 log |3 + log x| − log |2 + log x|
10) 3 2x2 − 4x + 5 + √ log (x − 1) + x2 − 2x +

2 1 1 2
2 19) − log |1 + cos x| + log |1 − cos x| + log |1 + 2 cos x|
  2 6 3
p 13 4x − 3 1 1 1 1
11) 1 − 3x + 2x2 + √ sin−1 √ 20) [− log |x | + 3 3
log |x + 3| + log |x3 − 4|]
2 2  17 3 12 21 28
p
2 −1 x−4 p
12) − 9 + 8x − x + 5 sin 13) log | log x + (log x)2 + 3| Exercise Set 4.7
  5
2 4 tan x + 3 1) x(log x) − 2x log x − 2x 2) x tan x − log | sec x| 3) −xe−x − e−x
2
14) √ tan −1 √ 15) sin−1 [(sin x − 4)/5]
71 71 1 x2 1
4) x3 tan−1 x − + log(x2 + 1) 5) log x · log | log x| − log | log x|
Exercise Set 4.5 3 6 6
1 3 x3 x2 4x 8
6) x log |x + 2| − + − + log |x + 2|
2 x 1 x 3 √ 9 √3 3 3
1) − 2) tan−1 (tan + 1) 3) tan−1 ([5 tan + 2]/6) −1
7) (x + 1) tan ( x) − x 8) x tan(x/2) − 2 log | sec(x/2)|
3 + tan x2 2 6 2
tan x + 2 − √3
x3 x2 x
x x 9) + sin 2x + cos 2x − 18 sin 2x 10) esin x [cos(sin x) + sin(sin x)]

5 tan − 25 tan

1 2 2
5) √ tan−1 √ 2 6) √ log 1 2 6 4 4

4) log √

12 5 tan x2 − 1 5 5 3 tan x2 + 2 + 3
1 √ x √ 1 h x √ i Exercise Set 4.8
7) √ tan−1 [( 3 tan )/ 2] 8) √ tan−1 tan / 7 1 3x 1
6 2 7 2 1) e [3 cos 4x + 4 sin 4x] 2) e2x [2 sin 4x + 4 cos 4x + 5 sin 2x + 5 cos 2x]
tan x − tan α 25 40

tan x2 + sin α
 
2 2 −1 1 1 1
9) cosec α log
x α 10) 2 sec α tan
3) e4x + e4x [4 cos 2x − 2 sin 2x] 4) e−3x [2 cos 2x − 3 sin 2x]
tan + tan cos α 8 40
r 13 r
2 2 5 5 2 9 9 5 5 9
5) (x + ) (x + ) − − log |(x + ) + (x + )2 − |
8 8 64 64 8 8 64
Exercise Set 4.6 (x − 1) √ 1
1 x+1 1 1 x−3 6) (2x − x2 ) + sin−1 (x − 1)
1) log | | 2) −x − log |x2 − 2x − 3| − log | | √ 2 2
h √ √ i
3 x+4 2 √ 4 x+1 7) 3 12 (x − 13 ) [(x − 13 )2 − 49 ] − 29 log |(x − 13 ) + [(x − 13 )2 − 49 ]|
1 x 1 x − 5
x2 1 h (x − 2) 9 (x − 2) i
3) tan−1 tan + √ log √ 4) + x − 2 log |x + 1| + 4 log |x + 2| 8) − (5 + 4x − x2 )3/2 + 3 (5 + 4x − x2 )1/2 + sin−1
2 2 4 5 x + 5 2 3 2 2 3
1 1 1 x
5) x + 10 log |x − 3| − 5 log |x − 2| 6) log |x − 1| − log |x2 + 1| + tan−1 x 9) [cos(3 log x) − 3 sin(3 log x)]
2 4 2 10 h x
√ (5e − 2) √ 2x 4ex 1 1 2 √ 2 1 i
1 x + 1
− 4 1 3 1 10) 5 [e − + ]+ log |(ex − )+ [(ex − )2 + ]|
7) log 8) log |x| + log |x + 2| − 10 5 5 50 5 5 25
2 x − 1 x − 1
4 x 4 2x
x + 2 9 4 e − 3 1
9)14 log + 10) log x + x
x + 3 x + 2 9 e 3e
Calculus 123

Exercise Set 4.9


ex

x−1
 Index
1) ex cos x 2) ex · log(sin x) 3) 4) −ex cot x2 5) ex Leibnitz Theorem, 65
(1 − x) x+1
x ex ex
6) e sec x 7) √ 8)
x2 + 1 (x + 1)2

124

You might also like