ADVANCED CALCULUS.

SMTA 021
University of Limpopo (Turfloop Campus)

DR L. Rundora
School of Mathematical and Computer Sciences
Department of Mathematics and Applied Mathematics
University of Limpopo (Turfloop Campus)
Contents

1 Limits: A Review 1
1.1 What this Unit is all About . . . . . . . . . . . . . . . . . . . . . 1
1.2 Limits: A Review . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 Limits of Indeterminate Type 6

2.1 What this Unit is all About . . . . . . . . . . . . . . . . . . . . . 6
2.2 Limits of Indeterminate Type . . . . . . . . . . . . . . . . . . . . 6
2.2.1 L’Hospital’s Rules . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.2 Other Indeterminate forms . . . . . . . . . . . . . . . . . . 11

3 Sequences 14
3.1 What this Unit is all About . . . . . . . . . . . . . . . . . . . . . 14
3.2 Properties of convergent sequences . . . . . . . . . . . . . . . . . 16
3.2.1 Boundedness . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3 The Bolzano-Weierstrass Theorem . . . . . . . . . . . . . . . . . . 25

4 Infinite series: Convergence and Divergence 29

4.1 What this Unit is all About . . . . . . . . . . . . . . . . . . . . . 29
4.2 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.3 Tests for Convergence and divergence . . . . . . . . . . . . . . . . 34
4.3.1 Comparison test . . . . . . . . . . . . . . . . . . . . . . . . 34
4.3.2 The ratio test . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.3.3 The Integral Test . . . . . . . . . . . . . . . . . . . . . . . 39
4.3.4 Alternating Series Test . . . . . . . . . . . . . . . . . . . . 41
4.3.5 The nth root test . . . . . . . . . . . . . . . . . . . . . . . 44

5 Power Series 48
5.1 What this Unit is all About . . . . . . . . . . . . . . . . . . . . . 48
5.2 Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.3 Taylor and Maclaurin Series . . . . . . . . . . . . . . . . . . . . . 51

6 Limits and Continuity 55

6.1 What this Unit is all About . . . . . . . . . . . . . . . . . . . . . 55
6.2 Limits of functions . . . . . . . . . . . . . . . . . . . . . . . . . . 55

i
 6.2.1 Characterization of limits of functions in terms of conver-
gence of sequences . . . . . . . . . . . . . . . . . . . . . . 56
6.3 Continuous functions . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.3.1 Some properties of continuous functions . . . . . . . . . . 59

7 Differentiation and Partial Differentiation 63

7.1 What this Unit is all About . . . . . . . . . . . . . . . . . . . . . 63
7.2 Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
7.2.1 Inverse trigonometric functions . . . . . . . . . . . . . . . 64
7.2.2 Hyperbolic functions . . . . . . . . . . . . . . . . . . . . . 66
7.2.3 Inverse Hyperbolic Functions . . . . . . . . . . . . . . . . 69
7.2.4 Parametric Equations . . . . . . . . . . . . . . . . . . . . . 71
7.3 Partial Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . 73
7.3.1 First partial derivatives . . . . . . . . . . . . . . . . . . . . 73
7.3.2 Second partial derivatives . . . . . . . . . . . . . . . . . . 75
7.3.3 Applications: Differentials . . . . . . . . . . . . . . . . . . 77

8 Integration 81
8.1 What this Unit is all About . . . . . . . . . . . . . . . . . . . . . 81
8.2 The Riemann Integral . . . . . . . . . . . . . . . . . . . . . . . . 81
8.2.1 Riemann Integrable Functions . . . . . . . . . . . . . . . . 84
8.2.2 Properties of the Riemann integral . . . . . . . . . . . . . 85
8.2.3 The fundamental theorem of integral calculus . . . . . . . 87
8.3 Multiple integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
8.3.1 The Double integral . . . . . . . . . . . . . . . . . . . . . 88
8.3.2 Triple Integrals . . . . . . . . . . . . . . . . . . . . . . . . 89
8.3.3 Transformations of multiple integrals . . . . . . . . . . . . 92
8.4 Line Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
8.5 Surface integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
8.5.1 Gradient, Divergence and Curl . . . . . . . . . . . . . . . . 100

9 Ordinary differential equations 106

9.1 What this Unit is all About . . . . . . . . . . . . . . . . . . . . . 106
9.2 Basic concepts and definitions . . . . . . . . . . . . . . . . . . . . 107
9.3 Types of first order differential equations . . . . . . . . . . . . . . 108
9.3.1 Equations with separable variables . . . . . . . . . . . . . 108

ii
 9.3.2 Equations reducible to variables separable type . . . . . . 112
9.3.3 Homogeneous equations . . . . . . . . . . . . . . . . . . . 114
9.3.4 Linear differential equations . . . . . . . . . . . . . . . . . 117
9.4 Second order linear differential equations with constant coefficients 120
9.4.1 Homogeneous equations of first order . . . . . . . . . . . . 120
9.4.2 Homogeneous equations of second order . . . . . . . . . . . 120

iii
Preface
In this module, we study sequences, infinite series, power series, limits and con-
tinuity, partial derivatives, Riemann integrals, multiple integrals, Line integrals
and surface integrals. These major areas are subdivided into units and subunits.
We start off with a review of limits and limits of indeterminate forms. These
first 2 units are included only for revision purposes and will not be discussed in
class. They were covered in SMTH011. The last unit is on ordinary differential
equations. It is most unlikely that we will have time to discuss this last unit in
class.
In each unit we incorporate several worked examples, individual class activities
and tutorial activities. A thorough study of the worked examples should enable
a student to work through the tutorial questions with little difficulty. Students
are encouraged to make sure that they read and understand their lectures before
attempting tutorial questions. Students are further encouraged to make use of
their lecturer should they have any queries arising from their studies.
It should be noted that the learning of Mathematics, unlike other subjects, is in
the ’doing of the Mathematics’ rather than just reading through.
This study guide, as the name suggests, is just a guide through the lectures. It is
still important to attend all the lectures. Again, the study guide is not meant to
replace any textbook. Thus, the study guide should be used in conjunction with
the relevant prescribed textbooks. A habit of visiting the library is encouraged.
It is my hope that you will enjoy going through this module.

L. Rundora January 2017

iv
Recommended Textbooks
1. Murray R. Spiegel, ”Advanced Calculus, Schaum’s Outline Series”, McGraw-
Hill, 1981.
2. Murray R. Spiegel, ”Advanced Mathematics for Engineers and Scientists”,
McGraw-Hill, 1971.
3. Buck R.C, ”Advanced Calculus”, McGraw-Hill.
4. Finney Thomas, ”Calculus”, 2nd edition, Addison Wesley.
5. Burkill J.C., ”A first Course in Mathematical Analysis”, Cambridge Univer-
sity Press,1970.
6. Rudin W., ”Principles of Mathematical Analysis”, McGraw-Hill, 1974.
7. Leonard F. Richardson, ”ADVANCED CALCULUS: An Introduction to Lin-
ear Analysis”, John Wiley & Sons, Inc., New Jersey, 2008.

v
Unit 1
Limits: A Review

1.1 What this Unit is all About

In this unit we revise the concept of a limit as discussed in MATH111. We en-
courage every student to understand and appreciate that limits are the backbone
of the study of Calculus.

By the end of this unit, you should be able to:

1. State the ε − δ definition of a limit.

2. State the definition of a limit in terms of the left hand and right hand limits.
3. State the theorem on algebra of limits and use it.
4. Prove various limits using the ε − δ definition of a limit.
5. Calculate/evaluate various limits.

1.2 Limits: A Review

Calculus is built on the notion of a limit. Consider f : R −→ R. One of the
important things to know in order to determine the behaviour of this function f
is how its outputs will change when the inputs change. Consider for example the
function f : R −→ R where f (x) = 4 − x2 . One may ask the question: ”What is
exactly happening to the values of f (x) as x approaches 3, say?” Or ”What is
the limit of the values of f (x) as x approaches 3?”

Definition 1.1 Let f (x) be a function, not necessarily defined at a point x = x0

but well defined at every point in some neighbourhood (nbd) of the point x = x0 .
Then, we say that a number l is the limit of f (x) as x approaches x0 if for every
ε > 0, there is a δ > 0 such that |f (x) − l| < ε whenever |x − x0 | < δ.

Notation:
Write l = lim f (x) or f (x) −→ l as x −→ x0 .
x−→x0

Remark 1.1 If lim f (x) = l1 and lim f (x) = l2 , then we call l1 and l2
x−→x+
0 x−→x−
0
respectively the right hand limit and the left hand limit of f (x) at x0 . In this way,

1
2 Limits: A Review

we say that lim f (x) exists if lim + f (x) = lim− f (x), i.e., if l1 = l2 .
x−→x0 x−→x0 x−→x0

Example 1.1 1. If f (x) = |x|, find lim f (x).

x−→0

Solution:
Use the graph of f (x) = |x| to observe that lim+ |x| = 0 = lim |x|.
x−→0 x−→0−
Therefore lim |x| = 0.
x−→0

|x|
2. If f (x) = , x 6= 0, what is lim f (x)?
x x−→0

Solution:  
x, if x > 0, 1, if x > 0,
Recall that |x| = Therefore f (x) =
−x, if x < 0 −1, if x < 0
Now, use a graph to observe that lim− f (x) = −1 and lim+ f (x) = 1. So
x−→0 x−→0
|x|
we have that lim− f (x) 6= lim+ f (x). Therefore lim , x 6= 0 does not
x−→0 x−→0 x−→0 x
exist.

 3 − x, if x < 2,
3. f (x) = 2, if x = 2 Does lim f (x) exist?
 x x−→2
2
, if x>2
Solution
Exercise!!
4. Prove that
(a) lim (7x − 1) = 13.
x−→2

x2 − 4
(b) lim = 4, x 6= 2
x−→2 x − 2

Solution:
(a). Here f (x) = 7x − 1, x0 = 2 and l = 13. So, given ε > 0, we need to
find δ(ε) > 0 so that ∀ x, 0 < |x − 2| < δ =⇒ |f (x) − 13| < ε. Now,

|f (x) − 13| = |7x − 1 − 13|

= |7x − 14|
= 7|x − 2|

So 0 < |x − 2| < δ =⇒

|f (x) − 13| = 7|x − 2|

< 7δ
ε
= 7·
7
= ε
Limits: A Review 3

ε
where we have chosen δ = . So 0 < |x − 2| < δ =⇒ |f (x) − 13| < ε, i.e.
7
lim (7x − 1) = 13.
x−→2

x2 −4
(b). Let ε > 0 be given. We have that x−2
− 4 = |x + 2 − 4| = |x − 2|,
x2 −4
so that if 0 < |x − 2| < δ, then x−2
− 4 = |x + 2 − 4| = |x − 2| < δ. So
x2 − 4
we choose δ = ε to conclude that lim = 4, x 6= 2.
x−→2 x − 2

5. Prove that lim x2 = 4

x−→2

Solution
Here f (x) = x2 , l = 4 and x0 = 2. Now,

|x2 − 4| = |(x + 2)(x − 2)| = |x + 2||x − 2|.

The desired factor is |x−2|. We need to bound |x+2|. For the factor |x+2|,
if x is close to 2, x + 2 is close to 4. So we should be able to replace |x + 2|
by a constant close to 4. More precisely, we observe that if |x − 2| < 1, then
|x + 2| < 5.

(|x − 2| < 1 =⇒ −1 < x − 2 < 1 =⇒ 1 < x < 3 =⇒ 3 < x + 2 < 5).

Thus, we have

|x2 − 4| = |(x + 2)(x − 2)| = |x + 2||x − 2| < 5|x − 2|,

so that given ε > 0, let δ = min 1, 5ε . This ensures that 0 < |x−2| < δ =⇒


|x2 − 4| = |(x + 2)(x − 2)| = |x + 2||x − 2| < 5|x − 2| < 5δ = 5 · 5ε = ε. Thus

|x2 − 4| < ε whenever 0 < |x − 2| < δ. This proves that lim x2 = 4.
x−→2

1 1
6. Prove that lim = .
x−→2 x + 1 3
Solution
1 1
Here f (x) = , x0 = 2 and l = . So
x+1 3
1 1 3 − (x + 1) 1 2−x 1 |x − 2|
|f (x) − l| = − = = = .
x+1 3 3(x + 1) 3 x+1 3 |x + 1|

Now, |x − 2| is the desired factor. We must find a bound on the factor

1
. We arbitrarily assume that |x − 2| < 1. Then −1 < x − 2 <
|x + 1|
1 =⇒ 1 < x < 3, so that 2 < x + 1 < 4. Thus |x + 1| > 2 if |x −
1 1
2| < 1 =⇒ < when |x − 2| < 1. Hence in this case we have
|x + 1| 2
4 Limits: A Review

1 |x − 2| 1 |x − 2| 1
|f (x) − l| = < = |x − 2|. Therefore, given ε > 0,
3 |x + 1| 3 2 6
we can take δ = min{6ε, 1} to ensure that if 0 < |x − 2| < δ we have
1 1 |x − 2| 1 1
x+1
− 31 = 1
< |x − 2| ≤ · 6ε = ε. So x+1 − 13 < ε whenever
3 |x + 1| 6 6
1 1
0 < |x − 2| < δ and so lim = .
x−→2 x + 1 3

Theorem 1.1 (Algebra of limits)

Suppose lim f (x) = l and lim g(x) = m. Then
x−→x0 x−→x0

1. lim (f (x) ± g(x)) = lim f (x) ± lim g(x) = l ± m.

x−→x0 x−→x0 x−→x0

2. lim (f (x) · g(x)) = ( lim f (x)) · ( lim g(x)) = lm.

x−→x0 x−→x0 x−→x0

  lim f (x)
f (x) x−→x0 l
3. lim = = , m 6= 0.
x−→x0 g(x) lim g(x) m
x−→x0

4. lim (c · f (x)) = c · lim f (x) = cl, c − any constant.

x−→x0 x−→x0

Activity 1.1 1. Find the following limits.

2x2 − 3x + 4 x2 − 9 x4 + x3
(a) lim 3 (b) lim 3 (c) lim
x→1 x √+ 5x + 1 x→3 x − 27 x→∞ 12x3 + 128
2
x + 4x − 2x x−1 x2 − 4
(d) lim (e) lim √ (f ) lim
x→−∞ 2x x→1 x2 + 24 − 5 x→2 x − 2

2. Find the following limits.

x3 − 1
(a) lim e−x sin x lim e−x sin x (c) lim 2
(b)
x→∞ x→−∞ x→0 x + 1
sin(x sin x1 )
     
πx 2 1 1
(d) lim 1 (e) lim cos (f ) lim x cos (g) lim 2x sin .
x→0 x sin x x→1 1+x x→0 x x→0 x

3. Find the following limits.

sin x sin−1 x 1
(a) lim (b) lim (c) lim x sin
x→0 x x→0 x x→∞ x
sin−1 (2x) sin 3x 1 − cos x
(d) lim (e) lim (f ) lim
x→0 sin 3x x→0 x x→0 x2
1 − cos x 1 − 2 cos x + cos 2x
(g) lim (h) lim .
x→0 x x→0 x2
4. Use the  − δ definition of a limit to prove that
x2 − 1 1 1
(a) lim −3x + 7 = 10 (b) lim = 2 (c) lim =
x→−1 x→1 x − 1 x→4 x 4
1 1
(d) lim = .
x→2 x + 3 5
Limits: A Review 5

5. The numbers x0 , l and  are given. Determine a number δ such that

|f (x) − l| <  for all x such that 0 < |x − x0 | < δ. √
(a) f (x) = 2x + 3, x0 = 1, l = 5,  = 0.001. (b) f (x) = 2x, x0 =
2, l = 2,  = 0.002.
Unit 2
Limits of Indeterminate Type

2.1 What this Unit is all About

In this unit we study limits of indeterminate type. Specifically, we study L’Hospital’s
rules.

By the end of this unit you should be able to:

1. Prove and apply Rolle’s theorem.
2.Prove and apply the Generalized Mean value theorem.
3. Prove L’Hospital’s rules.
4. Apply L’Hospital’s rules to evaluate limits of indeterminate type.

2.2 Limits of Indeterminate Type

If functions f (x) and g(x) are both 0 (or ∞) at x = x0 or as x −→ x0 then
f (x)
lim cannot be found by substituting x = x0 or by letting x −→ x0 . The
x−→x0 g(x)
0 ∞
substitution produces or , meaningless expressions known as indeterminate
0 ∞
forms. We have already encountered these limits but have manipulated them in
a way that avoided these forms.

1 − cos x 0
Example 2.1 1. lim 2
takes the form but can be solved as:
x−→0 x 0

1 − cos x 1 − (1 − 2 sin2 x2 )
lim = lim
x−→0 x2 x−→0 x2
2 x
2 sin 2
= lim
x−→0 x2
sin2 x2
= 2 lim
x−→0 x2
2 sin2 x2
= lim
4 x−→0 ( x2 )2
1
=
2

6
Limits of Indeterminate Type 7

x2 − 9 0
2. lim again takes the form but can be solved as:
x−→3 x − 3 0
x2 − 9 (x − 3)(x + 3)
lim = lim = lim x + 3 = 6.
x−→3 x − 3 x−→3 x−3 x−→3

Problem!! These (such) techniques do not work for some types of functions.
For such functions, we will need differential techniques in order to evaluate the
associated limits. This is done via L’Hospital’s Rules, which we are now set
to explore.
Recall that

Definition 2.1 A function f , defined on (a, b) is said to be differentiable at

f (x) − f (x0 )
x0 ∈ (a, b) if f 0 (x0 ) = lim .
x−→x0 x − x0

Definition 2.2 A function f is monotone increasing for a ≤ x ≤ b if f (x1 ) ≤

f (x2 ) for all x1 , x2 such that a ≤ x1 ≤ x2 ≤ b. If f (x1 ) < f (x2 ), then we say
that f is strictly monotone increasing.

Definition 2.3 The function f is said to be strictly increasing at c if there is a

nbd of c in which f (x) < f (c) for x < c and f (x) > f (c) for x > c.

There is a corresponding definition for ”strictly decreasing at c”.

Activity 2.1 Give this definition.

Theorem 2.1 If f 0 (c) > 0, then f is strictly increasing at c.

f (x) − f (c)
Proof 2.1 f 0 (c) = lim > 0 =⇒ f (x) > f (c) for x > c and f (x) <
x−→c x−c
f (c) for x < c. ♠

Theorem 2.2 (Rolle’s)

If (1) f is continuous in [a, b];
(2) f 0 exists in (a, b); and
(3) f (a) = f (b),
then there exists a value c with a < c < b, for which f 0 (c) = 0.

Proof 2.2 Either f is constant on [a, b] or f is not a constant on [a, b].

If f is a constant on [a, b], then f 0 (c) = 0 ∀ c ∈ (a, b).
Suppose f is not a constant on [a, b]. Since f is continuous on [a, b], f is bounded
on [a, b] and it attains its maximum and minimum values M and m respectively
on [a, b]. Moreover M 6= m since f is not a constant on [a, b]. If M 6= f (a) =
8 Limits of Indeterminate Type

f (b), then f (a) and f (b) are less than M and ∃ a point c ∈ (a, b) such that
f (c) = M . Since f (c) is the least upper bound of the set f ([a, b]), it implies that
f (x) ≤ f (c) ∀ x ∈ [a, b]. This means that f has a local maximum at c. Hence
f 0 (c) = 0.
Similarly, if m 6= f (a) = f (b), then f (a) and f (b) are greater than m and ∃ a
point c ∈ (a, b) such that f (c) = m. Since f (c) is the greatest lower bound of the
set f ([a, b]), it implies that f (x) ≥ f (c) ∀ x ∈ [a, b]. This means that f has a
local minimum at c. Hence f 0 (c) = 0. ♠

Theorem 2.3 (Mean Value Theorem)

Suppose f is continuous on [a, b] and differentiable on (a, b). Then there is a
f (b) − f (a)
point c ∈ (a, b) such that f 0 (c) = .
b−a
Geometrically, this theorem says that the gradient of the line joining (a, f (a))
and (b, f (b)) is equal to the gradient of the tangent to the curve at some point c
lying between a and b.
The Mean Value Theorem is a special case of
Theorem 2.4 (The Generalized Mean Value Theorem)
Suppose f and g are continuous on [a, b] and differentiable on (a, b). Then there
f 0 (c) f (b) − f (a)
is a point c ∈ (a, b) such that 0 = , provided g(b) − g(a) 6= 0 and
g (c) g(b) − g(a)
f 0 (x) and g 0 (x) are not simultaneously 0.

Remark 2.1 If g is chosen to be the identity function g(x) = x, then the Gen-
eralized Mean Value Theorem becomes the Mean Value Theorem above.

Proof 2.3 (Proof of The Generalized Mean Value Theorem)

We construct a special function F to which we apply Rolle’s theorem. Let
F (x) = f (x) − kg(x)
where k is a constant to be selected later. The function F is continuous on
[a, b] and differentiable in (a, b). (WHY?) To apply Rolle’s Theorem, we want to
have F (a) = F (b). Forcing this, we must have f (a) − kg(a) = f (b) − kg(b) or
f (b) − f (a)
f (b) − f (a) = k(g(b) − g(a)) =⇒ k = , and F satisfies the hypotheses
g(b) − g(a)
for Rolle’s Theorem. So there is a point c with F 0 (c) = 0. Since F 0 (x) = f 0 (x) −
f 0 (c)
kg 0 (x), this tells us that f 0 (c) = kg 0 (c), =⇒ k = 0 and the theorem is proved.
g (c)
♠

Remark 2.2 Suppose g(b) − g(a) = 0. Then we cannot solve for k. However,
we do not need to since we see that the theorem will hold if we can find a c with
g 0 (c) = 0 so that both sides will be 0. This we can do by applying Rolle’s theorem
to g.
Limits of Indeterminate Type 9

2.2.1 L’Hospital’s Rules

L’Hospital’s Rule 1
Case 1
Let f and g be functions that are differentiable on the interval (a, b) with g 0 (x) 6= 0
f 0 (x)
on (a, b). Let c ∈ (a, b). Suppose lim f (x) = lim g(x) = 0 but lim 0 = l.
x−→c x−→c x−→c g (x)
f (x)
Then lim = l.
x−→c g(x)

Proof 2.4 Suppose f and g are differentiable in (c, x) and g 0 6= 0 in (c, x). It
follows from the Generalized Mean Value Theorem that there is a point y ∈ (c, x)
f 0 (y) f (x) − f (c) f 0 (y)
such that 0 = . Since f (c) = g(c) = 0, this reduces to 0 =
g (y) g(x) − g(c) g (y)
f (x)
. Since y ∈ (c, x), x −→ c implies y −→ c which in turn ensures that
g(x)
f (x) f 0 (y)
lim = lim 0 = l which proves. ♠
x−→c g(x) x−→c g (y)

Case 2
Suppose, instead, that x −→ ∞. We show that the same result as in Case 1 is
1
obtained. We introduce the substitution x(t) = and apply the chain rule of
t
differentiation. We see that as x −→ ∞, t −→ 0+ . Now,

f (x) f ( 1t )
lim = lim+ 1
x−→∞ g(x) t−→0 g( )
t
(f ox)(t)
= lim+
t−→0 (gox)(t)

(f ox)0 (t)
= lim+
t−→0 (gox)0 (t)

f 0 (x)x0 (t)
= lim+ 0
t−→0 g (x)x0 (t)
f 0 ( 1t )
− t2
= lim g 0 ( 1t )
t−→0+
− t2
f 0 ( 1t )
= lim
t−→0+ g 0 ( 1t )
0
f (x)
= lim
x−→∞ g 0 (x)
= l

L’Hospital’s Rule 2
Suppose f and g are functions differentiable in (a, b) with
10 Limits of Indeterminate Type

g 0 (x) 6= 0 ∀ x ∈ (a, b). Let c ∈ (a, b). If lim f (x) = lim g(x) = ∞ but
x−→c x−→c
f 0 (x) f (x)
lim 0 = l. Then lim = l.
x−→c g (x) x−→c g(x)

Proof 2.5 We shall prove the theorem for the case x → +∞.
From lim f (x) = lim g(x) = ∞, pick a sufficiently large real number y such that
x−→c x−→c
the functions f, g and g 0 are non-degenerate on (y, ∞). Then for any x ≥ y, the
Generalized Mean Value Theorem ensures the existence of a point λ ∈ (y, x) such
that " f (y) #
0
f (λ) f (x) − f (y) f (x) 1 − f (x)
0
= = .
g (λ) g(x) − g(y) g(x) 1 − g(y)
g(x)

Equivalently, this means that

" g(y) #
f (x) f 0 (λ) 1 − g(x)
= 0
g(x) g (λ) 1 − f (y)
f (x)
" f (y) f (y) g(y) #
f 0 (λ) 1 − f (x) f (x)
− g(x)
= 0 +
g (λ) 1 − f (y) f (y)
1 − f (x)
f (x)
" f (y) g(y) #
f 0 (λ) f (x)
− g(x)
= 0 1+
g (λ) 1 − f (y)
f (x)

Observe that the proof will be complete if we show that

" f (y) g(y) #
f (x)
− g(x)
lim = 0.
x−→c 1 − ff (x)
(y)

But then this is so since we assumed that lim f (x) = lim g(x) = ∞ and y is
x−→c x−→c
f (x) f 0 (λ)
fixed. Hence lim = lim 0 = l. ♠
x−→c g(x) λ−→c g (λ)

sin x cos x
Example 2.2 1. lim = lim = 1.
x−→0 x x−→0 1
1 − cos x sin x cos x 1
2. lim 2
= lim = lim = .
x−→0 x x−→0 2x x−→0 2 2
3.
xn nxn−1
lim = lim
x−→∞ ex x−→∞ ex
n(n − 1)xn−2
= lim
x−→∞ ex
n(n − 1)(n − 2)xn−3
= lim
x−→∞ ex
Limits of Indeterminate Type 11

.
= ..
n!
= lim
x−→∞ ex
= 0

x2 − 9 2x
4. lim = lim = 6.
x−→3 x − 3 x−→3 1

2.2.2 Other Indeterminate forms

0 ∞
Other than the indeterminate forms and , there are others such as ∞0 , 0∞ ,
0 ∞
∞∞ , 00 , 1∞ , etc. The good news here is that these forms can be converted to
either of the two forms encountered earlier. Thus, we do not need another version
of L’Hospital’s rule to evaluate the limits involved under such cases.
Consider h(x) = [g(x)]f (x) and take logarithms:

ln g(x)
ln h(x) = f (x) ln g(x) = 1 .
f (x)

So " #
ln g(x)
lim ln h(x) = lim 1
x−→c x−→c
f (x)

∞
which is a limit of the type . So L’Hospital’s Rule 2 is applicable. Consequently,
∞
" #
ln g(x)
lim 1
x−→c
lim h(x) = e f (x) .
x−→c

Example 2.3 1. To evaluate lim+ xx , we observe that it is a limit of the

x−→0
form 00 and consider

ln x
y = xx =⇒ ln y = x ln x = 1
x

and
  1
ln x x
lim+ ln y = lim+ 1 = lim+ = lim+ (−x) = 0.
x−→0 x−→0
x
x−→0 − x12 x−→0

So
lim ln y
lim+ xx = lim+ eln y = ex−→0+ = e0 = 1.
x−→0 x−→0
12 Limits of Indeterminate Type

1
2. Find lim (cos x) x2
x−→0

Solution
1
Since lim cos x = 1 and lim 2 = ∞, the limit in question takes the inde-
x−→0 x−→0 x
1
terminate form 1 . Now, let y = (cos x) x2 . Then ln y = ln(cos
∞
x2
x)
. We now
apply L’Hospital’s rule to this. We have
ln cos x
lim ln y = lim
x−→0 x−→0 x2
sin x
−
= lim cos x
x−→0 2x
− sin x
= lim
x−→0 2x cos x
− cos x
= lim
x−→0 −2x sin x + 2 cos x
= − 12
1 1
Hence lim (cos x) x2 = e− 2 .
x−→0
 x
1
3. Find lim 1 + .
x−→∞ x
Solution
x
Observe that we have a limit of the form 1∞ . Let y = 1 + x1 . Then
 
1
ln y = x ln 1 +
x
ln 1 + x1


= 1
x
ln 1+x


x
= 1
x
ln 1+x


x
∴ lim ln y = lim 1
x−→∞ x−→∞
x
x −1


x+1 x2
= lim −1
x−→∞
x2
x
= lim
x−→∞ x + 1
= 1
 x
1
∴ lim 1+ = e.
x−→∞ x

Activity 2.2 1. For each of the following functions find, if possible, a number
η satisfying Rolle’s theorem in the given interval. If not possible explain why
Limits of Indeterminate Type 13

Rolle’s theorem is not applicable.

2
(i) f (x) = x(x + 1)(x − 2); [0, 2] (ii) f (x) = 4 − (8 − x) 3 ; [0, 16]
1
(iii) f (x) = ; [0, 4].
(x − 3)3
2. Prove that between any two real roots of sin x = e−x , there is a real root of
ex cos x = −1.

3. Suppose the functions f and g are differentiable with f (a) = g(a) and
f 0 (x) < g 0 (x) ∀ x > a. Show that f (x) < g(x) ∀ x > a.
[Hint: Let h(x) = f (x) − g(x) and apply the Mean Value theorem.]

4. Show that ln(1+x) < x ∀ x > 0. Apply the exercise above with appropriate
f (x) and g(x).

5. Verify the Generalized Mean Value Theorem for the functions

f (x) = 3x3 − 2x − 7 and g(x) = x2 + 2x + 5 on the interval [0, 2].

6. Find the following limits.

1 − cos x x−1 2 ln(sec x) − x2
(a) lim (b) lim (c) lim
x→0 x x→1 ln x x→0 x4
6x − sin 2x 1 − 2 cos x + cos 2x
(d) lim (e) lim
x→0 2x + 3 sin 4x x→0 x2
ln(sin x)
(f ) lim x(tan−1 (2x) − tan−1 (x)) (g) limπ
x→∞ x→ 2 (π − 2x)2
√
x 1 − x2
(h) lim+ √ √ (g) lim √ .
x→0 1+x− 1−x x→1 1 − x3
7. Find the following limits.
1 1
(a) lim+ (sin x) ln x (b) lim (1 + 2x) 3x (c) lim (1 + cot x)tan x
x→0 x→∞ x→ π2
 x
1
(d) limπ (tan x)tan 2x (e) lim+ xsin x (f ) lim cos
x→ 4 x→0 x→∞ x
1
(g) lim (x2 + 1) x .
x→∞
Unit 3
Sequences

3.1 What this Unit is all About

In this Unit we are going to study sequences of real numbers. Within this broad
subject we look at convergent sequences and their properties, we look at various
theorems including the ”Principle of Monotone Bounded Convergence” and the
Bolzano-Weierstrass theorem. We also define a Cauchy sequence and prove the
”Cauchy Convergence Criterion”.

By the end of this unit, you should be able to:

1. define a convergent sequence and several other given definitions.

2. prove the convergence of various sequences.
3. prove various theorems and propositions on sequences.
4. State and understand several given results on sequences.
5. find limit points, cluster points and subsequences of given sequences.
6. state and understand various given relationships on sequences.

Definition 3.1 A sequence of real numbers is a function defined on N whose

range is R where f (n) = un , n = 1, 2, 3, · · ·.
Intuitively, a sequence is a set of numbers u1 , u2 , · · · , un , un+1 , · · · in a definite
order of arrangement and formed according to a definite rule.

Example 3.1 1. 1, 2, 4, 8, 16, · · ·

2. 1, −2, 3, −4, 5, −6, · · ·

3. 0, 1, 0, 21 , 0, 31 , 0, 14 , · · ·

4. 2, 32 , 43 , 54 , 65 , · · ·

One can deduce a formula or an algorithm or a defining rule for any given se-
quence. In the above example

1. f (n) = un = 2n−1

14
Sequences 15

2. Exercise

3. Exercise
n+1
4. f (n) = un = .
n
We can also define sequences by giving a relation or a formula that connects
successive terms of the sequence and specifying the value or values of the first
term or the first and the second terms etc. The formula or relation linking the
terms is called a recursive formula or a recurrence relation. Thus, in this case the
sequence is said to be defined recursively.

Example 3.2
2
un = 1 + , u1 = 1
un−1
un = un−1 + un−2 , n ≥ 3, u1 = 1, u2 = 1
3
un+1 = un + 1, u1 = 1
4

Activity 3.1 Obtain the first five terms for each of the sequences defined above.

Definition 3.2 A number l is the limit of a sequence u1 , u2 , u3 , · · · if ∀ ε >

0 ∃ N (ε) ∈ N such that |un − l| < ε ∀ n > N (ε).

Write lim un = l or un −→ l as n −→ ∞.
n−→∞

Remark 3.1 • A sequence that has a limit as n −→ ∞ is said to be convergent

to that limit or is just said to converge. Thus, if un −→ l as n −→ ∞, then un
is said to be convergent to l or just to converge to l.
• A sequence that does not have a limit as n −→ ∞ is said to be divergent or
to diverge.

Activity 3.2 Give the ε − δ definition of a divergent sequence {un }.

Hint: The answer is the negation of the definition of a convergent sequence.

4 − 2n
Example 3.3 1. Find the number l to which the sequence {un } =
3n + 2
converges and prove, using the definition, that this number is indeed the
limit.

Solution
2
Here, it is clear that l = − . To prove this: Given ε > 0, we look for
3
16 Sequences

N (ε) ∈ N such that |un − l| = |un − (− 23 )| < ε, ∀ n > N (ε). Now,

|un − (− 32 )| < ε =⇒ 4−2n
3n+2
+ 23 < ε

12 − 6n + 6n + 4
=⇒ < ε
9n + 6
16
=⇒ < ε
9n + 6
16
=⇒ < ε
9n + 6
=⇒ 16 < 9nε + 6ε
16 − 6ε
=⇒ < n
9ε
16 − 6ε
So N = will do. e.g. if ε = 0.1, N = 18. What is N if ε = 0.001?
9ε
2n − 7
2. Let {un } = Find how large n must be so that
3n + 2
(a) |un − l| < 0.1
(b) |un − l| < 0.001
(c) |un − l| < ε.

3. Prove that lim xn = 0 if |x| < 1.

n−→∞

Solution
If x = 0, then clearly lim xn = 0. Now, assume that x 6= 0. Then we
n−→∞
must show that given ε > 0 ∃ N ∈ N such that |xn − 0| < ε ∀ n > N .
log ε
Now, |xn − 0| < ε =⇒ |xn | < ε =⇒ n log |x| < log ε =⇒ n > since
log |x|
log ε
log |x| < 0 as |x| < 1. Thus, if N is the smallest integer greater than
log |x|
then |xn − 0| < ε ∀ n > N and so lim xn = 0.
n−→∞

3.2 Properties of convergent sequences

Recall from MATH111 that ∀ x, y ∈ R the triangle inequality |x+y| ≤ |x|+|y|
holds.

Theorem 3.1 The limit of a sequence is unique.

Proof 3.1 Suppose {un } −→ l and {un } −→ s. We show that l = s. Now,

ε
{un } −→ l =⇒ ∀ ε > 0 ∃ N1 (ε) : |un − l| <
2
Sequences 17

for all n > N1 and

ε
{un } −→ s =⇒ ∀ ε > 0 ∃ N2 (ε) : |un − s| <
2
for all n > N2 . Now,

|l − s| = |l − un + un − s|
≤ |l − un | + |un − s| by triangle inequality
ε ε
< + by definitions above
2 2
= ε

∴ |l − s| < ε This implies that l = s for all n > N = max{N1 , N2 }. i.e. l and
s differ by at most ε which is arbitrary, i.e., can be chosen to be as small as we
like. ♠

3.2.1 Boundedness
Definition 3.3 A sequence {un } is said to be bounded above if ∃ a number α
such that un ≤ α ∀ n.

Example 3.4 • The sequence 1, 1 12 , 1 23 , 1 34 , · · · is bounded above since all the

terms of the sequence are below any number greater than or equal to 2.
• The sequence 1, −1, 1, −1, · is also bounded above.
• The sequence 2, 4, 6, 8, · · · is not bounded above.

Remark 3.2 • The number α in the definition above is often called an upper
bound for the sequence {un }.
• Whenever {un } is bounded above, there are uncountably many upper bounds to
the sequence.
• If {un } is bounded above, then the graph of the sequence would show, the points
of the sequence all lying below a specific horizontal line.

Definition 3.4 A sequence {un } is said to be bounded below if ∃ a number β

such that un ≥ β ∀ n.

Remark 3.3 • The number β in the definition above is often called a lower
bound for the sequence {un }.
• Whenever {un } is bounded below, there are uncountably many lower bounds to
the sequence.
• If {un } is bounded below, then the graph of the sequence would show, the points
of the sequence all lying above a specific horizontal line.
18 Sequences

Definition 3.5 A sequence {un } is said to be bounded if it is both bounded above

and below, i.e. if ∃ a number γ such that |un | ≤ γ.

Example 3.5 • 1, 2, 3, 4, · · · is not a bounded sequence.

• 1, 23 , 53 , 47 , 59 , · · · is bounded.
• 1, −1, 1, −1, 1, −1, · · · is bounded.

Definition 3.6 Let {un } be a sequence. Then a number α is said to be the the
least upper bound or supremum or simply lub of {un } (a) if α is an upper
bound of {un }; and (b) if η is an upper bound of {un }, then α ≤ η.

Definition 3.7 Let {un } be a sequence. Then a number β is said to be the the
greatest lower bound or infinimum or simply glb of {un } (a) if β is a lower
bound of {un }; and (b) if γ is a lower bound of {un }, then γ ≤ β.

Theorem 3.2 A convergent sequence is bounded.

Proof 3.2 Suppose {un } is a convergent sequence and suppose that un −→ l.

Then
∀ ε > 0 ∃ N (ε) ∈ N : |un − l| < ε ∀ n > N.
If ε = 1, we can find N ∈ N such that |un −l| < 1 ∀ n > N. Thus l−1 < un < l+1
for n = N + 1, N + 2, N + 3, · · ·. Consequently, we have m < un < M where
m = min{u1 , u2 , · · · , uN , l − 1} and M = max{u1 , u2 , · · · , uN , l + 1}. Thus {un }
is bounded. ♠

Theorem 3.3 Suppose {un } and {vn } are sequences of real numbers such that
lim un = u and lim vn = v. Then
n−→∞ n−→∞

1. lim {un ± vn } = u ± v.
n−→∞

2. lim {cun } = cu, c ∈ R.

n−→∞

3. lim {un vn } = uv.

n−→∞
 
un u
4. lim = , provided v 6= 0.
n−→∞ vn v

Proof 3.3 1.
ε
{un } −→ u =⇒ ∀ ε > 0 ∃ N1 (ε) : |un − u| <
2
for all n > N1 and
ε
{vn } −→ v =⇒ ∀ ε > 0 ∃ N2 (ε) : |vn − v| <
2
Sequences 19

for all n > N2 . Now, choosing N : N = max{N1 , N2 } we have that for all
n>N

|(un + vn ) − (u + v)| = |un − u + vn − v|

≤ |un − u| + |vn − v|, by triangle inequality
ε ε
< + by definitions above
2 2
= ε (3.1)

=⇒ un + vn −→ u + v. ♠

2.
{un } −→ u =⇒ ∀ ε > 0 ∃ N (ε) : |un − u| < ε
for all n > N . Now

|cun − cu| = c|un − u| < cε ∀ n > N

=⇒ cun −→ cu since c is constant and ε is arbitrary. ♠

3. We need to show that ∀ ε > 0 ∃ N ∈ N such that |un vn −uv| < ε ∀ n > N .
Now,

|un vn − uv| = |un (vn − v) + v(un − u)|

≤ |un (vn − v)| + |v(un − u)|
= |un ||vn − v| + |v||un − u|
≤ M |vn − v| + (|v| + 1)|un − u| (3.2)

by Theorem 3.2.
Since un −→ u and vn −→ v, then ∀ ε > 0 we can find N1 and N2 such
that
ε
|un − u| < ∀ n > N1
2(|v| + 1)
and
ε
|vn − v| < ∀ n > N2
2M
Hence, from (3.2)we have
ε ε
|un vn − uv| < M · + (|v| + 1) ·
2M 2(|v| + 1)
ε ε
= +
2 2
= ε

for all n > N = max{N1 , N2 } ♠

1 1
Lemma 3.1 Suppose un −→ u where u 6= 0, then −→ .
un u
20 Sequences

Proof 3.4 By hypothesis, ∀ ε > 0 ∃ N1 (ε) :

1
|un − u| < u2 ε (3.3)
2
for all n > N1 . Now
1 2
1 1 un − u |un − u| uε
− = = < 2 .
un u un u |un ||u| |un ||u|
Also, since u 6= 0, we can find N2 :
1
|un | > |u| (3.4)
2
∀ n > N2 . Thus
1 2
1 1 uε
− < 2 1 =ε
un u |u| · 2 |u|
1 1
for all n > N = max{N1 , N2 }. Hence −→ . ♠
un u
4.
un u un v − vn u
− =
vn v vvn
(un − u)v + u(v − vn )
=
vvn
v(un − u) u(v − vn )
≤ +
vvn vvn
un − u u(v − vn )
= +
vn vvn
2|un − u| 2(|u| + 1)|vn − v|
< + (3.5)
|v| |v|2
by the lemma above. Now, by hypothesis
|v|ε
∀ ε > 0 ∃ N1 ∈ N : |un − u| < ∀ n > N1
4
and
|v|2 ε
∀ ε > 0 ∃ N2 ∈ N : |vn − v| < ∀ n > N2
4(|u| + 1)
and choosing N = max{N1 , N2 } gives
un u 2 |v|ε 2(|u| + 1) |v|2 ε
− < · + ·
vn v |v| 4 |v|2 4(|u| + 1)
ε ε
= +
2 2
= ε
(3.6)
un u
for all n > N . =⇒ −→ . ♠
vn v
Sequences 21

Remark 3.4 The results in the previous theorem can be extended by Mathemat-
ical Induction to a finite number of convergent sequences.
• If an , bn , cn , · · · , zn are convergent sequences, then their sum (an + bn + · · · + zn )
is also a convergent sequence and lim(an + bn + · · · + zn ) = lim an + lim bn +
lim cn + · · · + lim zn .
• Their product an bn · · · zn is also a convergent sequence and lim(an bn · · · zn ) =
(lim an )(lim bn )(lim cn ) · · · (lim zn ).
• Hence, if k ∈ N and if {un } is a convergent sequence then lim(un )k = (lim un )k .

Definition 3.8 • A sequence of real numbers {un } is called an increasing se-

quence if un+1 ≥ un for all n, and is called a decreasing sequence if un+1 ≤ un
for all n.
• A sequence of real numbers {un } is called a strictly increasing sequence (strictly
decreasing sequence) if un+1 > un (un+1 < un ) for all n.

Definition 3.9 A sequence of real numbers is called a monotonic sequence if

it is either an increasing sequence or a decreasing sequence. Thus, we talk of a
monotone increasing sequence and a monotone decreasing sequence.

Theorem 3.4 Let {un } be a monotone increasing sequence of real numbers.

Then either
(a). {un } is bounded above and {un } tends to a limit as n −→ ∞, or
(b). {un } is not bounded above and un −→ ∞ as n −→ ∞.

Corollary 3.1 A monotone increasing sequence of real numbers converges if and

only if it is bounded above.

Theorem 3.5 Let {un } be a monotone decreasing sequence of real numbers.

Then either
(a). {un } is bounded below and {un } tends to a limit as n −→ ∞, or
(b). {un } is not bounded below and un −→ −∞ as n −→ ∞.

Corollary 3.2 A monotone decreasing sequence of real numbers converges if and

only if it is bounded below.

Theorem 3.4 and Theorem 3.5 taken together give the following theorem which is
referred to as ”The Principle of Monotone Bounded Convergence (PMBC)”:

Theorem 3.6 A bounded monotonic sequence converges.

22 Sequences

n
Example 3.6 We want to show that the sequence an = 1 + n1 is monotonic
increasing.
 n
1
an = 1+
n
   2  n
1 n(n − 1) 1 1
= 1+n + + ··· +
n 2! n n
      
1 1 1 1 2 r−1
= 1+1+ 1− + ··· + 1− 1− ··· 1 −
2! n r! n n n
   
1 1 n−1
+ ··· + 1− ··· 1 −
n! n n
 n+1
1
an+1 = 1+
n+1
      
1 1 1 1 2 r−1
= 1+1+ 1− + ··· + 1− 1− ··· 1 −
2! n+1 r! n+1 n+1 n+1
       
1 1 n−1 1 1 n
+ ··· + 1− ··· 1 − + 1− ··· 1 −
n! n+1 n+1 (n + 1)! n+1 n+1

Observe that the first two terms of an and an+1 are the same: the rth term of
an+1 is greater than the rth term
of an for 3 ≤
r ≤ n + 1. Also, an+1 has an
1 1 n
extra positive term (n+1)! 1 − n+1 · · · 1 − n+1 at the end. Hence an < an+1
1

n
and an = 1 + n is a strictly increasing sequence.
NB: We have already (in the tutorial) shown that an −→ e.

Example 3.7 The sequence {an } is defined by

5
a1 = , 5an+1 = a2n + 6, n = 1, 2, · · · .
2
Using induction it can be proved that

2 < an < 3 ∀ n.
5
Proof 3.5 a1 = , and so the result is true for a1 . Now, suppose the result is
2
true for n = k − 1, where k is some integer such that k ≥ 2. Then

2 < ak−1 < 3.

We have
5ak = a2k−1 + 6 < 32 + 6 = 15
by above. Also
5ak = a2k−1 + 6 > 22 + 6 = 10
again by above. Hence 10 < 5ak < 15 =⇒ 2 < ak < 3.
Sequences 23

Since the result is true for n = 1, n = k − 1, n = k, it follows by induction that

2 < an < 3 ∀ n.
We now show that {an }, as defined, is decreasing:

1 2
an+1 − an = (a + 6) − an
5 n
1 2
= (a − 5an + 6)
5 n
1
= (an − 3)(an − 2)
5
< 0

So an+1 < an , since by above 2 < an < 3. Hence {an } is decreasing.

Since {an } is bounded below, by PMBC it must converge. Let a be its limit. Then
an −→ a as n −→ ∞ and so an+1 −→ a as n −→ ∞. But
1 1
an+1 = (a2n + 6) −→ (a2 + 6)
5 5
as n −→ ∞. Since this limit is unique, we have that
1
a = (a2 + 6),
5
i.e.,
a2 − 5a + 6 = 0
which solves to give a = 2 or a = 3.
5
Now, since the sequence is decreasing, a ≤ a1 = and so the only possibility is
2
a = 2.

Activity 3.3 The sequence {an } is defined by

1
a1 = 1, an+1 = 4 + 2an3 , n ∈ Z+ .

Use Mathematical Induction to prove that 1 ≤ an ≤ 8 ∀ n ∈ Z+ . Show also that

an+1
> 1 ∀ n. Deduce that {an } converges and find its limit.
an

Theorem 3.7 (Squeeze theorem)

Suppose that {un }, {an } and {vn } are sequences such that un ≤ an ≤ vn ∀ n ∈ N,
and that lim un = lim vn . Then {an } converges and lim un = lim an = lim vn .

Proof 3.6 Let un −→ l and vn −→ l. Then given ε > 0, ∃ N ∈ N such that

|un − l| < ε ∀ n > N

24 Sequences

and
|vn − l| < ε ∀ n > N.
This implies that
−ε < un − l < ε ∀ n > N
and
−ε < vn − l < ε ∀ n > N.
Now, un ≤ an ≤ vn ∀ n ∈ N =⇒

un − l ≤ an − l ≤ vn − l ∀ n > N.

This implies that

−ε < an − l < ε,
i.e.,
|an − l| < ε ∀ n > N
which means an −→ l. ♠

sin n
Example 3.8 To show that lim = 0, we observe that
n−→∞ n

−1 ≤ sin n ≤ 1 ∀ n ∈ N.

This implies
−1 sin n 1
≤ ≤ ∀ n ∈ N.
n n n
−1 1 sin n
Since lim = 0 = lim we conclude that lim = 0 by the squeeze
n−→∞ n n−→∞ n n−→∞ n
theorem.
1
Activity 3.4 Show that lim (2n + 3n ) n = 3.
n−→∞

Activity 3.5 1. (a) Show that if un → u, then |un | → |u|.

(b) Give an example to show that if |un | → |u|, then {un } does not neces-
sarily converge to u.
(c) Show that if |un | → 0, then un → 0.

2. For the following

(a) state the limit of the sequence.

(b) verify that the stated value is indeed the correct limit by using the
definition of a limit.
n 2o
(i) 5 − 21n (ii) 1+n
  sin n
n2
(iii) n
.
Sequences 25

 n
1
3. Prove that lim 1 + = e.
n→∞ n
[Hint: Use the Binomial expansion.]

4. For {sn } given by the following formulae, establish either the convergence
or divergence of {sn }.
(−1)n n nn 1 + (−1)n
(i) (ii) (iii)
n+1 (n + 1)n+1 n√
1 n


(iv) ln n − ln(n + 1) (v) 1 − n2 (vi) n n + 1.

5. Given that {sn } = {(−1)n }, show that {sn } is divergent but the sequence
s1 + · · · + sn
{tn } defined by tn = is convergent.
n
6. (a) Suppose {sn } is a sequence converging to 0, and {tn } another sequence
such that 0 ≤ tn ≤ sn for every index n. Show that {tn } converges to
0.
(b) Investigate
 n the behaviour (by sketching a graph) of the sequence
n+1
, when n is odd,
sn = 1
1 − n , when n is even

3.3 The Bolzano-Weierstrass Theorem

Definition 3.10 Given a point or a number x0 ∈ R, the set of all points x such
that 0 < |x − x0 | < δ or x ∈ (x0 − δ, x0 + δ) is called a δ-nbd of the point x0 .

Definition 3.11 Let D be a set of real numbers. A point x0 is called an accu-

mulation point or a cluster point of D if every δ-nbd of x0 contains at least
one point of D different from x0 .
 
1
Example 3.9 Let D = : n ∈ N . Cluster point x0 = 0.
n

Theorem 3.8 (Bolzano-Weierstrass Theorem for sets)

Every bounded infinite subset of real numbers has at least one cluster point.

Definition 3.12 Let {an } be a sequence of real numbers and let n1 < n2 <
n3 < · · · < nk < · · · be a strictly increasing sequence of positive integers. Then
the sequence an1 , an2 , an3 , · · · , ank , · · · is called a subsequence of {an } and is
denoted by {ank }

Example 3.10 If {an } = {a1 , a2 , a3 , a4 , · · ·}. Then two subsequences which im-
mediately come out are
26 Sequences

• {a2n } = {a2 , a4 , a6 , · · ·} and

• {a2n−1 } = {a1 , a3 , a5 , · · ·}

obtained by even and odd indexing respectively.

Example 3.11 {an } = (−1)n . Let nk = 2k. Then the constant sequence

{a2k } = {1, 1, 1, · · ·}

is a subsequence of {an } = (−1)n . What is {a2k−1 }?

1 3n 1 2n 1 n




Example 3.12 1 + 3n
, 1+ 2n
are both examples of subsequences of 1 + n
.

Theorem 3.9 A sequence {an } converges to a limit l if and only if every subse-
quence of {an } converges to l.

Theorem 3.10 (Bolzano-Weierstrass Theorem for sequences)

A bounded sequence has a convergent subsequence.

Definition 3.13 A sequence {un } is called a Cauchy sequence if, given ε > 0
∃ N ∈ N such that ∀ m, n > N |um − un | < ε.

Proposition 3.1 A convergent sequence is a Cauchy sequence.

Proof 3.7 Let un −→ l. Then given

ε
ε > 0, ∃ N ∈ N : ∀ n > N |un − l| < .
2
Now, let m, n > N, then

|um − un | = |um − l + l − un |
≤ |um − l| + |l − un |
ε ε
< +
2 2
= ε

Thus |um − un | < ε ∀ m, n > N and {un } is Cauchy. ♠

Theorem 3.11 A Cauchy sequence is convergent.

Proof 3.8 Let {xn } be a Cauchy sequence. Then given ε > 0 ∃ N ∈ N such that
∀ m, n > N |xn − xm | < ε. In particular, if we take ε = 1 and m = N + 1 we
obtain that ∀ n > N, |xn − xN +1 | < 1. In other words, for n > N ,

xN +1 − 1 < xn < xN +1 + 1.
Sequences 27

Now, let
M = max{x1 , x2 , · · · , xN , xN +1 + 1}
and
m = min{x1 , x2 , · · · , xN , xN +1 − 1}.
Then, clearly m < xn < M . We have proved that {xn } is bounded. It remains to
prove that {xn } is convergent.
Now, since {xn } is bounded, it has at least one convergent subsequence {xnk }.(This
is by which result?) Let xnk −→ l. Then given ε > 0 ∃ N1 ∈ N such that
∀ n > N1 |xnk − l| < 2ε . But, since {xn } is Cauchy, ∃ N2 ∈ N such that
ε
∀ m, n > N2 and ∀ nk > N1 , |xm − xn | < .
2
Let N = max{N1 , N2 }. Then for n > N, we have

|xn − l| = |xn − xnk + xnk − l|

≤ |xn − xnk | + |xnk − l|
ε ε
< +
2 2
= ε

and xn −→ l. ♠

In Proposition 3.1 and Theorem 3.11 together we have proved:

Theorem 3.12 (Cauchy convergence criterion)

A sequence is convergent if and only if it is Cauchy.

Activity 3.6 1. (a) Definitions to understand and apply:

(1) x is the limit of the sequence {xn }, i.e. xn converges to x or
lim xn = x, if ∀ε > 0 ∃ N (ε) ∈ N : n ≥ N =⇒ |xn − x| < ε.
n→∞
(2) x is a limit point of the sequence {xn } if there is some subsequence
xn(k) −→ x.
(3) x is a cluster point of the sequence {xn } if every ε-nbd of x con-
tains an infinite number of points of the sequence.
Find all the Limit points and cluster points for each of the follow-
ing sequences and give the convergent subsequences where appropriate.
Find also, if they exist, supxn , infxn and lim xn .
n→∞
(i) (−1)n (ii) 1 + n1 (1 + (−1)n ) (iii) n1 + sin nπ


2
1 nπ 1 3n


(iv) n sin 2 (v) sin n (vi) 1 + 3n

3n
n
n
(vii) 1 − n1 (viii) 2 +
n1 (ix) 1 + n3 .
n
You may assume that 1 + n1 −→ e.
(b) Decide what implications exist between the properties (1), (2) and (3).
28 Sequences

2. Definition to understand and apply:

{xn } is a Cauchy sequence if ∀ε > 0 ∃ N (ε) : n, m ≥ N =⇒ |xm − xn | < ε.
Show that each of the following is a Cauchy sequence
(i) xn = n+1n
(ii) xn = 1 + 14 + 19 + · · · + n12 .
3. Investigate the monotonicity and convergence of
1
an = n n .

4. If √
un+1 = 3un , u1 = 1,
show that un converges and find the limit.

n
5. By using the fact that 1 + n1 → e as n → ∞, show that the following
sequences

are convergent and find
their limits in terms

of e:
1 n 1 n 1 n 1 1 n


1 + 3n , 1 − n , 1 + 2n , 1 − 2n − 2n2 .
6. The sequence {an } is defined by
r
a2n
a1 = 1, an+1 = 1+ , n = 1, 2, 3, · · ·
2
Show that (a) a2n − 2 < 0, (b) a2n+1 − a2n > 0. Deduce that {an }
converges and find its limit.

Definition 3.14 A sequence {In } of intervals In = [an , bn ] is a sequence of

nested intervals if In+1 ⊆ In and lim (bn − an ) = 0.
n−→∞

Theorem 3.13 To every sequence of nested intervals there corresponds one and
only one real number in common to all of them.

Proof 3.9 Let {In } be a sequence of nested intervals where In = [an , bn ]. Then
an ≤ an+1 and bn+1 ≤ bn and a1 ≤ a2 ≤ · · · ≤ an ≤ an+1 ≤ · · · ≤ bn+1 ≤
bn ≤ · · · ≤ b2 ≤ b1 . Thus, {an } is monotonic increasing bounded by a1 and b1
and {bn } is monotonic decreasing bounded by b1 and a1 . Hence the sequences
{an } and {bn } converge to a and b respectively. So ∀ ε > 0 ∃ N1 , N2 ∈
N s.t. |an − a| < 3ε ∀ n > N1 and |bn − b| < 3ε ∀ n > N2 . Further, since
ε
lim (bn − an ) = 0, ∃ N3 ∈ N s.t. |bn − an | < ∀ n > N3 . Now, let
n−→∞ 3
N = max{N1 , N2 , N3 }. Then ∀ n > N ;
|a − b| = |a − an + an − bn + bn − b|
= |(a − an ) + (an − bn ) + (bn − b)|
≤ |a − an | + |an − bn | + |bn − b|
ε ε ε
< + +
3 3 3
= ε
⇒a=b ♠
Unit 4
Infinite series: Convergence and
Divergence

4.1 What this Unit is all About

In this unit we study the convergence and divergence of infinite series. We discuss
the comparison test, the limit form of the comparison test, the ratio test, the nth
root test, the integral test as well as the alternating series test.

By the end of this unit you should be able to:

1. use all the different tests to test various series for convergence.
2.identify which test to use on any given series.
3. demonstrate various studied proofs.

4.2 Preliminaries
Let {un } be a sequence of numbers. Then
∞
X
u1 + u2 + · · · = un
n=1

is called an infinite series.

n
X
un = u1 + u2 + · · · + un
i=1

is called a finite series.

Let
s1 = u1 , s2 = u1 + u2 , s3 = u1 + u2 + u3 ,
sn = u1 + u2 + · · · + un .
Then
{s1 , s2 , s3 , · · ·}
is referred to as the sequence of partial sums.

29
30 Infinite series: Convergence and Divergence

Note:

• If the sequence of partial sums {sn } = {s1 , s2 , s3 , · · ·} is convergent, i.e., if

X∞
there exists s such that lim sn = s, then the infinite series un is said to be
n−→∞
n=1
convergent and s is the sum of the series.
• If lim sn does not exist then the infinite series diverges.
n−→∞

Example 4.1 Consider the geometric series

∞
X
axn−1 , a, x, constants.
n=1

In order to determine when the series converges or diverges, we find an expression

for the sequence of partial sums and compute the limit. The nth partial sum is

sn = a + ax + ax2 + · · · + axn−1 (4.1)

xsn = ax + ax2 + ax3 + · · · + axn−1 + axn (4.2)

a(xn − 1)
(4.2) − (4.1) =⇒ (x − 1)sn = a(xn − 1) =⇒ sn = , x 6= 1.
x−1
Therefore

a(xn − 1)
lim sn = lim
n−→∞ n−→∞ x−1
a
= lim (xn − 1)
x−1 n−→∞
a

1−x
, if |x| < 1,
=
∞, if |x| > 1
∞ ∞
X
n−1 a X
Hence ax = if |x| < 1, i.e., the series converges and axn−1
n=1
1 − x n=1
diverges if |x| > 1.

∞
X
Activity 4.1 Determine whether or not the series (−1)n+1 converges or di-
n=1
verges.

Example 4.2 Consider the geometric series

∞
X 3 3 3
n
= 3 + + 2
+ ···.
n=0
10 10 10
Infinite series: Convergence and Divergence 31

1
Here x = < 1 =⇒ the series converges and
10
1 n
a(xn − 1) 3( 10 ) −1 10
lim sn = lim = lim 1 = .
n−→∞ n−→∞ x−1 n−→∞
10
−1 3

Hence ∞
X 3 10
n
= .
n=0
10 3

∞
X ∞
X
Let an and bn be convergent series. Write
n=1 n=1

s n = a1 + a2 + · · · + an ,

tn = b1 + b2 + · · · + bn .
Then
(a1 + b1 ) + (a2 + b2 ) + · · · + (an + bn ) = sn + tn
∞
X ∞
X
and sn + tn −→ s + t as n −→ ∞, where s = an and t = bn . Hence
n=1 n=1
∞
X ∞
X ∞
X ∞
X
(an + bn ) converges and (an + bn ) = an + bn . Thus, there is no
n=1 n=1 n=1 n=1
difficulty about adding convergent series.
∞
X 3n + 4n
Example 4.3 Consider the series . Observe that
n=0
5n

∞ ∞ ∞ ∞  n ∞  n
X 3n + 4n X 3n X 4n X 3 X 4
= + = + .
n=0
5n n=0
5n n=0
5n n=0
5 n=0
5

3 4
Both series are geometric with r = < 1 and r = < 1 respectively. This
5 5
implies that both series converge and hence the original series converges also.
Now,
( 3 )n − 1 ( 4 )n − 1 15
lim sn = lim 53 + lim 54 = .
n−→∞ n−→∞
5
−1 n−→∞
5
−1 2
∞
X 3n + 4n 15
Hence = .
n=0
5n 2

Example 4.4 (Telescoping Series)

1 1 1
Let an = . Then an = − and a1 = 1 − 12 , a2 = 1
2
− 31 , a3 =
n(n + 1) n n+1
32 Infinite series: Convergence and Divergence

1
3
− 14 , · · · , an−1 = 1
n−1
− n1 , an = 1
n
− 1
n+1
. So
s n = a1 + a2 + a3 + · · · + an
         
1 1 1 1 1 1 1 1 1
= 1− + − + − + ··· + − + −
2 2 3 3 4 n−1 n n n+1
1
= 1−
n+1
and sn −→ 1 as n −→ ∞, so that the series
∞ ∞
X X 1
an =
n=1 n=1
n(n + 1)
∞
X 1
converges and = 1.
n=1
n(n + 1)

  ∞
1 X
Activity 4.2 Find an explicit expression for the partial sums of log 1 + .
n=1
n
∞  
X 1
Deduce that log 1 + diverges.
n=1
n

Example 4.5 Consider the series

∞
X 20
.
n=1
(7n − 3)(7n + 4)

Observe that using partial fractions, we can rewrite the series as

∞ ∞  
X 20 X 20 20
= −
n=1
(7n − 3)(7n + 4) n=1
7(7n − 3) 7(7n + 4)
∞  
20 X 1 1
= −
7 n=1 7n − 3 7n + 4
n  
20 X 1 1
sn = −
7 r=1 7r − 3 7r + 4
 
20 1 1 1 1 1 1 1 1
= − + − + − + ··· + −
7 4 11 11 18 18 25 7n − 3 7n + 4
 
20 1 1
= −
7 4 7n + 4
 
20 1 1 5
and lim sn = lim − = . So, the series converges and
n−→∞ n−→∞ 7 4 7n + 4 7
∞
X 20 5
= .
n=1
(7n − 3)(7n + 4) 7
Infinite series: Convergence and Divergence 33

∞
X 1 1 1 1
Consider the series = 1+ + + · · · + + · · ·. This series is called the
r r=1
2 3 n
Harmonic series. Now,
1
s2 = 1 +
2
1 1 1
s4 = 1+ + +
2  3 4 
1 1 1
= 1+ + +
2 3 4
 
1 1 1
> 1+ + +
2 4 4
1 1
= 1+ +
2 2
1 1 1 1 1
s8 = 1 + + + + + ··· +
2  3 4 5  8 
1 1 1 1 1
= 1+ + + + + ··· +
2 3 4 5 8
   
1 1 1 1 1 1 1
> 1+ + + + + + +
2 4 4 8 8 8 8
1 1 1
= 1+ + +
2 2 2
1 1 1 1 1 1 1
s16 = 1 + + + + + ··· + + + ··· +
2  3 4 5  9 10  16 
1 1 1 1 1 1 1 1 1
> 1+ + + + + + + + + ··· +
2 4 4 8 8 8 8 16 16
1 1 1 1
= 1+ + + +
2 2 2 2
Let us have a summary of these observations:
s21 > 1
1
s21 ≥ 1 +
2
1 1
s22 > 1+ +
2 2
1 1 1
s23 > 1+ + +
2 2 2
1 1 1 1
s24 > 1+ + + +
2 2 2 2
..
.  
1 1 1
s2 n > 1+ + + ··· + n − times
2 2 2
 
1
= 1+n
2
34 Infinite series: Convergence and Divergence

n
= 1+
2
1
= (n + 2) −→ ∞
2

∞
X 1
as n −→ ∞. Thus, lim s2n does not exist and hence the harmonic series
n−→∞
r=1
r
diverges.

This example must be remembered. The harmonic series is a divergent series

that is frequently used for reference.
Consider the series
∞
X 1 1 1
p
= 1 + p + p + ···
n=1
n 2 3

where p is a constant. It will be shown later that this series converges for p > 1
and diverges for p ≤ 1. This series is called the p-series. Observe that when
p = 1, this series becomes the harmonic series.

∞
X 1
Example 4.6 The series is a p-series with p = 2.
n=1
(2n)2

4.3 Tests for Convergence and divergence

It rarely happens that we are able to find an explicit expression for the partial
sum sn and then check whether sn tends to a finite limit as n −→ ∞. For this
reason, we must build up a collection of tests which can be used to decide whether
a series converges or not.

4.3.1 Comparison test

(for series of non-negative terms)
∞
X ∞
X
(a) If 0 ≤ un ≤ vn ∀ n > N ∈ N and if vn converges, then un converges
n=1 n=1
also.
∞
X ∞
X
(b) If un ≥ vn ∀ n > N ∈ N and if vn diverges, then un diverges
n=1 n=1
also.
Infinite series: Convergence and Divergence 35

∞
X
Proof 4.1 (a) The partial sums of the series un are bounded above by
n=1

∞
X
M = u1 + u2 + · · · + uN + vn .
n=N +1

∞
X
They therefore form a non-decreasing sequence with a limit L ≤ M . Thus un
n=1
converges.
∞
X
(b) Clearly, the partial sums of un are not bounded from above. If they were,
n=1
∞
X
the partial sums for vn would be bounded by
n=1

∞
X
0
M = v1 + v2 + · · · + vN + un
n=N +1

∞
X ∞
X
and vn would have to converge instead of diverge. Thus, un diverges. ♠
n=1 n=1

1 1 1 1
Example 4.7 Let un = and vn = . Then we see that > for all
∞
ln n n ∞
ln n n
X1 X 1
n ≥ 2. Now, since diverges (harmonic series), diverges also by
n=2
n n=2
ln n
the comparison test.
∞
X 1
Example 4.8 Test the series for convergence.
n=1
n(n + 2)
∞
1 1 1 1 X 1
Here un = . Let vn = 2 . Then < 2 ∀ n and is
n(n + 2) n n(n + 2) n n=1
n2
∞
X 1
convergent (p-series). =⇒ converges by comparison test.
n=1
n(n + 2)

Theorem 4.1 (Limit form of comparison test (asymptotic test))

∞
X ∞
X n o
Let an and bn be series of positive real numbers such that abnn tends to
n=1 n=1
∞
X ∞
X
a finite non-zero limit or ∞ as n −→ ∞. Then either an and bn both
n=1 n=1
∞
X ∞
X
converge or an and bn both diverge.
n=1 n=1
36 Infinite series: Convergence and Divergence

n o
an
Proof 4.2 Let bn
−→ l 6= 0. Note that l > 0.

an 1
=⇒ ∃ N ∈ N : −l < l ∀ n>N
bn 2
i.e.,
1 an 1
=⇒ − l < −l < l
2 bn 2
1 an 3
=⇒ l < < l
2 bn 2
1 3
lbn < an < lbn ∀ n > N.
2 2
∞
X ∞
X
Now, if bn converges, an must also converge by the comparison test since
n=1 n=1
∞ ∞
3 X X
0 < an < lbn ∀ n > N . Moreover, if an converges then bn must also
2 n=1 n=1
2an
converge by the comparison test since 0 < bn < ∀ n > N . Thus if one of
l ∞ ∞
X X
the series converges, then the other also converges, i.e., either an and bn
n=1 n=1
∞
X ∞
X
both converge or an and bn both diverge. ♠
n=1 n=1

1 1
Example 4.9 Let an = and bn = . Then for all n > 1, an > 0, bn > 0
ln n n
and
an n
=
bn ln n
and
an n
lim = lim = lim n = ∞.
n−→∞ bn n−→∞ ln n n−→∞
∞
X1 ∞
X 1
Now, since diverges (harmonic series), it follows that diverges also
n=2
n n=2
ln n
by the limit form of the comparison test.
1 1
Example 4.10 Let an = , and bn = 2 . Then For all n ≥ 1, an >
n(n + 2) n
0, bn > 0 and
an n2
= 2 −→ 1
bn n + 2n
∞
X 1
as n −→ ∞. Now is a convergent p-series, so by the limit form of com-
n=1
n2
∞
X 1
parison test, also converges.
n=1
n(n + 2)
Infinite series: Convergence and Divergence 37

X n X ln n
Activity 4.3 Test for convergence: (a). √ , (b). √ .
3n5 − 8 n+1
Example 4.11 (p-series)
1 1
1. Let an = p
where p ≥ 2 and bn = 2 . For all n ∈ N, an > 0, bn > 0 and
n n
1 1
0 < an = p ≤ 2 = bn .
n n
∞
X ∞
X
Now, since bn converges, it follows by the comparison test that an
n=1 n=1
converges. So
∞
X 1
p
converges ∀ p ≥ 2.
n=1
n
1 1
2. Let an = , bn = p where p ≤ 1. For all n ∈ N, an > 0, bn > 0 and
n n
1 1
0 < an = ≤ p = bn .
n n
∞
X ∞
X
Since p ≤ 1, by the comparison test bn diverges, since an diverges
n=1 n=1
(harmonic series). Thus
∞
X 1
diverges ∀ p ≤ 1.
n=1
np
∞
 
X1
Activity 4.4 Test for convergence: sin
n=1
n

4.3.2 The ratio test

Using the comparison test, we examine a series for convergence by comparing it
with another series whose behaviour is already known. It is, however, possible
to test for convergence by studying only the behaviour of the ratio of successive
terms.
Theorem 4.2 (D’Alembert’s Test - Ratio test)
∞
X an+1
Let an be a series of positive real numbers such that −→ l as n −→ ∞.
n=1
an
X∞
(a). If l < 1, then an converges.
n=1
∞
X
(b). If l > 1, then an diverges.
n=1
38 Infinite series: Convergence and Divergence

Proof 4.3 (a). Case l < 1

an+1
Choose a number k such that l < k < 1. Since −→ l as n −→ ∞, ∃ N such
an
that
an+1
− l < k − l ∀ n > N.
an
an+1
This gives < k for all n > N =⇒ an+1 < kan ∀ n > N . Thus, for all
an
n>N +1
a 
N +1
0 < an < kan−1 < k 2 an−2 < · · · < k n−N −1 aN +1 = k n N +1
k
∞
aN +1 X
Now, is just a constant. Moreover, k n converges since 0 < k < 1. By
k N +1 n=1
X∞
comparison test, an converges also.
n=1

(b). Case l > 1

an+1
Since −→ l > 1 as n −→ ∞ ∃ N1 such that
an
an+1
> 1 ∀ n > N1 .
an
Thus for n > N1 + 1

an > an−1 > an−2 > · · · > aN1 +1 .

∞
X
Hence an does not converge to zero as n −→ ∞, and therefore an diverges.♠
n=1

∞
X an+1
Corollary 4.1 Let an be a series of positive terms such that −→ ∞ as
n=1
an
∞
X
n −→ ∞. Then an diverges.
n=1

Remark 4.1 If l = 1 in the theorem, then either convergence or divergence is

∞
1 X
possible. For instance, we have already seen that if an = , then an diverges
n n=1
an+1 n
(harmonic series), but = −→ 1 as n −→ ∞. On the other hand if
an n+1
∞
1 X an+1 n2
an = 2 then an converges (p-series, p = 2), but again = −→ 1
n n=1
an (n + 1)2
as n −→ ∞. The ratio test is therefore inconclusive if l = 1.
Infinite series: Convergence and Divergence 39

[(2n)!]2
Example 4.12 Let an = for all n ∈ N. Then an > 0 and
(4n)!
an+1 [(2n + 2)!]2 (4n)!
=
an (4n + 4)![(2n)!]2
(2n + 2)(2n + 1)(2n + 2)(2n + 1)
=
(4n + 4)(4n + 3)(4n + 2)(4n + 1)
(2n + 2)(2n + 1)
=
2(4n + 3)2(4n + 1)
4n2 + 6n + 2 1
= −→
4(16n2 + 16n + 3) 16
∞
X [(2n)!]2
as n −→ ∞. Since the limit is less than 1, converges.
n=1
(4n)!

Example 4.13 Let an = n2 e−n(n+1) for all n ∈ N. Then an > 0 and

an+1 (n + 1)2 e−(n+1)(n+2)
=
an n2 e−n(n+1)
2
en(n+1)

n+1
= · (n+1)(n+2)
n e
 2
1
= 1+ e−2(n+1) −→ 0
n
∞
X ∞
X
as n −→ ∞. Hence an = n2 e−n(n+1) converges.
n=1 n=1

∞ ∞
X 1 X n7
Activity 4.5 Test for convergence using the ratio test: (a). (b)
n=1
n2 n=1
2n
∞
X 2n
(c) 3 (n + 4)
.
n=1
n

4.3.3 The Integral Test

Suppose f (x) is a positive strictly decreasing function for x ≥ N and is such that
f (n) = un , n = N, N + 1, N + 2, · · · ,
X
then un converges or diverges according as the integral
Z ∞ Z M
f (x)dx = lim f (x)dx
N M −→∞ N

converges or diverges. In particular, we may have N = 1, as is often true in

practice.
40 Infinite series: Convergence and Divergence

Proof 4.4 f is a decreasing function with f (n) = un , ∀ n. We are led to

observe that, according to figure 1,
Z n+1
f (x)dx ≤ u1 + u2 + · · · + un .
1

Alternatively, considering figure 2, we see that, disregarding the first rectangle:

Z n
u2 + u3 + · · · + un ≤ f (x)dx.
1

If we include u1 , we have
Z n
u1 + u2 + u3 + · · · + un ≤ u1 + f (x)dx.
1
Infinite series: Convergence and Divergence 41

Combining these results gives

Z n+1 Z n
f (x)dx ≤ u1 + u2 + · · · + un ≤ u1 + f (x)dx.
1 1
Z ∞ X
If the integral f (x)dx is finite,the right-hand inequality shows that un is
1 Z
∞
also finite. But if f (x)dx is infinite, then the left hand inequality shows that
1
the series is also infinite. Hence the series and the integral are either both finite
or both infinite. ♠
∞
X 1
Example 4.14 Consider the p-series p
, p a constant. Note that if
n=1
n
1 1
f (x) = p , then p = f (n). Clearly f (x) is a strictly decreasing function for x
x n
positive. Now,
Z ∞ Z ∞
1
f (x)dx = dx
1 1 xp
Z M
1
= lim dx
M −→∞ 1 xp
 M
1 1−p
= lim x if p 6= 1 or
M −→∞ 1 − p
1
= lim [ln x]M
1 if p = 1
M −→∞
 
1 1
= lim − 1 for p 6= 1 or
1 − p M −→∞ M p−1
= lim ln M, p = 1
M −→∞

1
• If p > 1, lim = 0 and series converges.
M −→∞ M p−1
1
• If p < 1, lim diverges and the series diverges.
M −→∞ M p−1
• If p = 1, lim ln M diverges and series diverges.
M −→∞

∞
X 1
Activity 4.6 Use the integral test to test the series for convergence.
n=1
n2

4.3.4 Alternating Series Test

Theorem 4.3 (Alternating Series Test)
Let {an }∞
n=1 be a decreasing sequence of positive real numbers such that an −→ 0
X∞
as n −→ ∞. Then the series (−1)n+1 an converges.
n=1
42 Infinite series: Convergence and Divergence

Proof 4.5
sn = a1 − a2 + a3 − a4 + · · · + (−1)n+1 an first n terms
s2n = a1 − a2 + a3 − a4 + · · · + a2n−1 − a2n first 2n terms
s2n+2 = a1 − a2 + a3 − a4 + · · · + a2n−1 − a2n + a2n+1 − a2n+2
∴ s2n+2 − s2n = a2n+1 − a2n+2 ≥ 0,
since {an } is decreasing and, therefore, a2n+1 ≥ a2n+2 . Thus s2n+2 ≥ s2n and the
sequence {s2n } is increasing. Moreover,
s2n = a1 − a2 + a3 − a4 + a5 − · · · − a2n−2 + a2n−1 − a2n
= a1 − (a2 − a3 ) − (a4 − a5 ) − · · · − (a2n−2 − a2n−1 ) − a2n
≤ a1
because a2 ≥ a3 ≥ a4 ≥ · · · ≥ a2n−2 ≥ a2n−1 and a2n > 0. Thus s2n ≤ a1 for all n
and the sequence {s2n } is bounded above. Since it is also increasing, by PMBC,
it must converge to some number s, i.e., s2n −→ s as n −→ ∞. Now, if we
consider the first 2n + 1 number of terms, s2n+1 , we have
s2n+1 = a1 − a2 + a3 − a4 + · · · + a2n−1 − a2n + a2n+1
= s2n + a2n+1
As n −→ ∞, a2n+1 −→ 0 and s2n −→ s. It follows that s2n+1 −→ s as n −→ ∞.
Since s2n −→ s and s2n+1 −→ s as n −→ ∞, we see that sn −→ s as n −→ ∞
∞
X
and the series therefore converges and (−1)n+1 an = s. ♠
n=1

If, however, an 9 0 as n −→ ∞, then (−1)n+1 an 9 0 as n −→ ∞ and the series

X∞
(−1)n+1 an diverges.
n=1

1


Example 4.15 Let an = sin n
, n = 1, 2, 3, · · ·. For
1 1 π
n ≥ 1, 0 < < ≤1< ,
n+1 n 2
and therefore    
1 1
an+1 = sin < sin = an ,
n+1 n
i.e., {an } is
a decreasing sequence of positive real numbers. Furthermore,
an = sin n1 −→ 0 as n −→ ∞. By the alternating series test,
∞  
X
n+1 1
(−1) sin
n=1
n
converges.
Infinite series: Convergence and Divergence 43

Activity 4.7 What happens to the series in example 4.15 if we remove the factor
(−1)n+1 ?
∞  
X
n+1 1
Activity 4.8 Test for convergence: (−1) cos .
n=1
n

Definition 4.1 Absolute convergence

∞
X ∞
X ∞
X
Let an be a series of real numbers. If |an | is convergent, then an is
n=1 n=1 n=1
said to be absolutely convergent.

The phrase ’absolute convergence’ immediately suggests convergence plus some-

thing more. We would, therefore, expect that an absolutely convergent series is
certainly convergent.

Theorem 4.4 Every absolutely convergent series of real numbers is also conver-
gent.
∞
X
Proof 4.6 Let an be an absolutely convergent series of real numbers, i.e., it
n=1
∞
X ∞
X
is a series such that |an | converges. Now, some of the terms of an may
n=1 n=1
be positive and some may be negative. We therefore
 separate the positive and
an , if an ≥ 0,
negative terms in the following way: Let bn = ,
 0, if an < 0
0, if an ≥ 0,
cn = Then bn ≥ 0, cn ≥ 0 for all n, and an = bn − cn .
−an , if an < 0
Write

sn = |a1 | + |a2 | + · · · + |an |,

tn = a1 + a2 + · · · + an ,
rn = b1 + b2 + · · · + bn ,
un = c1 + c2 + · · · + cn .
∞
X
Since |an | converges, then sn −→ s as n −→ ∞. Moreover |an | ≥ 0 ∀ n (by
n=1
definition of modulus) and so the sequence {sn } is increasing. Thus sn ≤ s ∀ n.
From the definition of bn , we see that

rn = b1 + b2 + · · · + bn ≤ |a1 | + |a2 | + · · · + |an | = sn ≤ s ∀ n

i.e., rn ≤ s ∀ n. But bn ≥ 0 for all n, and therefore the sequence {rn } is

increasing. Since it is bounded above by s, {rn } must converge to some limit r as
n −→ ∞. Similarly {un } converges to some limit u as n −→ ∞.
44 Infinite series: Convergence and Divergence

Now, tn = rn − un −→ r − u as n −→ ∞. Since tn tends to a finite limit as

∞
X
n −→ ∞, an converges as required. ♠
n=1

Definition 4.2 A series of real numbers which is convergent, but not absolutely
convergent, is called a conditionally convergent series.

∞  
X
n+11 2
. We notice that (−1)n+1 sin2 1


Example 4.16 Consider (−1) sin n
=
n=1
n
1
sin2 n1 . Write an = sin2 n1 , bn = 2 . Then an > 0, bn > 0 and



n

!
!
sin2 n1 sin n1 sin n1


an
= 1 = 1 1 −→ 1 · 1 = 1
bn n2 n n

∞
X 1
as n −→ ∞. Now 2
converges, and therefore by the limit form of the compar-
n=1
n
∞   ∞  
X
2 1 X
n+1 2 1
ison test, sin is convergent, i.e., (−1) sin is convergent.
n=1
n n=1
n
∞  
X
2 1
Thus sin is absolutely convergent and therefore convergent.
n=1
n

The notion of absolute convergence allows us to extend the ratio test in the
following way:

∞
X an+1
Theorem 4.5 Let an be a series of non-zero real numbers. If an
−→ l as
n=1
n −→∞∞ then
X
(a). an is absolutely convergent (and so convergent) if l < 1.
n=1
∞
X
(b). an is divergent if l > 1.
n=1

4.3.5 The nth root test

p X
n
Theorem 4.6 Let lim |un | = l. Then the series un
n−→∞
(a). converges (absolutely) if l < 1.
(b). diverges if l > 1.
If l = 1, the test fails.
Infinite series: Convergence and Divergence 45

Proof 4.7 (a). l < 1.

√ √
Choose ε > 0 so small that l + ε < 1. Since n un −→ l, the terms n un eventually
√
get closer than ε to l. In other words ∃ M ≥ N such that n un < l + ε when
X∞
n
n ≥ M . Then it is also true that un < (l + ε) for n ≥ M . Now (l + ε)n ,
n=M
∞
X
a geometric series with ratio l + ε < 1, converges. Hence by comparison un
n=M
converges, from which it follows that
∞
X ∞
X
un = u1 + u2 + · · · + uM −1 + un
n=1 n=M

converges.

(b). l > 1.
√
For all indices beyond some integer M , we have n un > 1, so that un > 1 for
n > M . The terms of the series do not converge to 0. The series diverges. ♠

∞
th
X n2
Example 4.17 Use the n root test to test the series for convergence.
n=1
2n

r √
n √
n n2 n2 ( n n)2 1
n
= √
n
= −→ < 1.
2 2n 2 2
Thus, the series converges by the nth root test.

∞
X 2n
Activity 4.9 Test for convergence: .
n=1
n2

1
Activity 4.10 1. Let an = , n = 1, 2, 3, · · ·. Let sn = a1 +
n(n + 1)(n + 2)
a2 + · · · + an . Express an in partial fractions and hence, or otherwise, show
that
1 1 1
sn = − + .
4 2(n + 1) 2(n + 2)
∞
X 1
Deduce that converges and find its sum.
n=1
n(n + 1)(n + 2)

2. Find an explicit expression for the partial sums of

∞  
X 1
log 1 + .
n=1
n
46 Infinite series: Convergence and Divergence

Deduce that
∞  
X 1
log 1 +
n=1
n
diverges.
∞
X 1
3. Determine the sum, if it exists, of the series .
n=1
(5n − 2)(5n + 3)

4. Determine whether the following series converge or diverge:

∞ ∞   ∞  
X n+1 X 1 1 X
2 1
(a) 2
(b) sin (c) sin
n=1
n +1 n n n
∞ √ √ n=1 ∞
n=1
∞
X n+1− n−1 X 1 X 3 n1
(d) √ (e) (f )
n ( n+1 ) 1
n=1 n n=1 n
n 3
n=1

5. Determine the convergence or divergence of the following series:

∞ ∞ ∞
X 1 X 1 X 1
(a) 1 (b) √ (c) 4
n=1 n
3
n=1
3
n n=1 n
3
∞
X 1 ∞
X 1
(d) (e) .
n=1
n2 n=1
nπ

Activity 4.11 1. Verify the convergence or divergence of each of the series:

∞ ∞ ∞
(3n)!3−n
 
X X sin n X 1 (2n − 1)π
(a) (b) (c) √ tan
n=1
n!(2n)! n=1
n2 n=1
n 4

∞ ∞ ∞
n2 + 2
 
X X 1 X
n 1
(d) (e) (f ) (−1) sin √
n=1
3n3 + 4n n=2
n ln n n=1
n

∞ ∞ ∞
X (5n)! X (n2 + 1)4 X 2 + (−1)n
(g) (h) (i) 3
n=1
(30)n (3n)!(2n)! n=1
(n3 + 1)3 n=1 n2

∞ ∞ ∞
X n2 + 1 X (2n)! X sin(n!)
(j) 4 2
(k) 2
(l)
n=1
4n + n − n − 1 n=1
(n!) n=1
n2 +n+1

∞ ∞ ∞
X (−1)n n2 X 1 X 9n (3n)!n!
(m) (n) (o)
n=1
n3 + 1 n=2
n(ln n)3 n=1
(4n)!

∞ ∞ ∞
2 + (−1)n n2 + n + 1 (−1)n
X X X  
(p) √ (q) √ (r) sin
n=1
n n=1
n7 + n4 + 1 n=1
n
Infinite series: Convergence and Divergence 47

∞ ∞ ∞
X 1 X (−1)n+1 X (−1)n
(s) 3 (t) (u)
n=2 n(ln n) 2
n=1
2n n=1
n2 + 2n + 2

2. Test for absolute or conditional convergence:

∞ ∞ ∞
X (−1)n−1 X (−1)n−1 n X (−1)n
(a) (b) (c)
n=1
n2 + 1 n=1
n2 + 1 n=2
n ln n

∞ ∞ ∞
X (−1)n n3 X (−1)n−1 1 X (−1)n−1 n3
(d) 4 (e) sin √ . (f )
n=1 (n2 + 1) 3 n=2
2n − 1 n n=1
2n − 1
Unit 5
Power Series

5.1 What this Unit is all About

In this unit we study power series. We study the Taylor and the Maclaurin series
of given functions.

By the end of this unit you should be able to:

1. find the interval of convergence of a given power series.
2. find the Taylor polynomial (approximation) of a given function.
3. find the Maclaurin series (approximation) of a given function.

5.2 Power Series

Definition 5.1 A series having the form
∞
X
an x n = a0 + a1 x + a2 x 2 + · · · + an x n + · · ·
n=0

or
∞
X
an (x − c)n = a0 + a1 (x − c) + a2 (x − c)2 + · · · + an (x − c)n + · · ·
n=0

where a0 , a1 , a2 , · · · are constant coefficients and c is the centre, is called a power

series in x.

A power series converges for |x| < R and diverges for |x| > R, where the constant
R is called the radius of convergence of the series. For |x| = R, the series may
or may not converge.
The interval |x| < R or −R < x < R, with possible inclusion of end points, is
called the interval of convergence of the series.

Theorem 5.1 A power series converges uniformly and absolutely in any interval
which lies entirely within its interval of convergence.

Theorem 5.2 (Abel’s)

When a power series converges up to and including an end point of its interval of

48
Power Series 49

convergence, the interval of uniform convergence also extends so far as to include

this end point.

Theorem 5.3 (Abel’s limit theorem)

∞
X
If an xn converges at x = x0 which may be an interior point or an end point
n=0
of the interval of convergence, then
(∞ ) ∞ ∞
X X X
n n
lim an x = lim an x = an xn0 .
x−→x0 x−→x0
n=0 n=0 n=0

Remark 5.1 Usual algebraic operations, i.e., addition, subtraction, multiplica-

tion, division, multiplication by a constant etc can be applied to power series.

Example 5.1 For what values of x do the following series converge?

∞
X xn−1
(a).
n=1
n · 3n
∞
X (−1)n−1 x2n−1
(b).
n=1
(2n − 1)!
X∞
(c). n!(x − a)n
n=1
xn−1
(a). Let fn = . Assuming x 6= 0 (if x = 0, the series converges) we
n · 3n
have
fn+1 xn n · 3n
lim = lim ·
n−→∞ fn n−→∞ (n + 1) · 3n+1 xn−1
n
= lim |x|
n−→∞ 3(n + 1)

|x|
=
3
|x| |x|
and by the ratio test the series converges if < 1, and diverges if > 1. If
3 3
|x|
= 1, i.e., x = ±3, the test fails.
3 ∞ ∞
X 1 1X1
• If x = 3, the series becomes = which diverges (harmonic se-
n=1
3n 3 n=1 n
ries).
∞ ∞
X (−1)n−1 1 X (−1)n−1
• If x = −3, the series becomes = which converges
n=1
3n 3 n=1 n
by the alternating series test. Thus the interval of convergence is −3 ≤ x < 3.
50 Power Series

(−1)n−1 x2n−1
(b). Here fn = . Then
(2n − 1)!
fn+1 (−1)n x2n+1 (2n − 1)!
lim = lim ·
n−→∞ fn n−→∞ (2n + 1)! (−1)n−1 x2n−1
(−1)n x2n x(2n − 1)!
= lim
n−→∞ (2n + 1)(2n)(2n − 1)!(−1)n (−1)−1 x2n x−1

−x2
= lim
n−→∞ 2n(2n + 1)
= 0
and the series converges (absolutely) for all x. So the interval of (absolute) con-
vergence is −∞ < x < ∞.

(c). fn = n!(x − a)n .

fn+1 (n + 1)!(x − a)n+1
lim = lim
n−→∞ fn n−→∞ n!(x − a)n
= lim (n + 1)(x − a)
n−→∞

which is infinite if x 6= a. Thus the series converges for x = a.

∞
X
Example 5.2 Prove that both the power series an xn and the corresponding
n=0
∞
X
series of derivatives nan xn−1 have the same radius of convergence.
n=0

∞
X
Proof 5.1 Let R > 0 be the radius of convergence of an xn . Let 0 < |x0 | < R.
n=0
Then we can choose N so that
1
|an | <
|x0 |n
for n > N . Thus the terms of the series
X X
|nan xn−1 | = n|an ||x|n−1

can, for n > N , be made less than corresponding terms of the series
X n|x|n−1
|x0 |n
X
which converges, by the ratio test, for |x| < |x0 | < R. Hence nan xn−1 con-
verges absolutely for all points x0 (no matter how close |x0 | is to R). If, however,
Power Series 51

X
|x| > R, lim an xn 6= 0 and thus lim nan xn−1 6= 0, so that nan xn−1 does
n−→∞ n−→∞ X
not converge. Thus R is the radius of convergence of nan xn−1 . ♠

∞
X xn
Example 5.3 Illustrate Example 5.2 by using the series 2 3n
.
n=1
n
Here
xn+1 n2 3n n2 |x| |x|
lim · = lim = .
n−→∞ (n + 1)2 3n+1 xn n−→∞ (n + 1)2 3 3
Thus, by an earlier example, the radius of convergence is −3 ≤ x < 3. Now,
∞
X xn−1
consider the series of derivatives . This has already been found to have
n=1
n3n
radius of convergence −3 ≤ x < 3. This proves.

5.3 Taylor and Maclaurin Series

Most applications of Mathematics ultimately require the precise evaluation of
functions by numerical means at some points of their domains.
The task of computing a value f (x) to a prescribed accuracy calls for a com-
bination of Arithmetic and Approximation; the approximation of functions by
rational functions. We approximate functions using polynomials of suitable
degrees.
The local approximation of f near a uses values of f and its derivatives at a.
Now, in order to consider the local approximation of f near a, we put x = a + h.
Approximating f (a + h) near h = 0 by the polynomial
a0 + a1 h + a2 h2 + · · · + an hn
gives the error
en (h) = f (a + h) − (a0 + a1 h + a2 h2 + · · · + an hn )
and lim en (h) = 0 translates to
h−→0

a0 = lim f (a + h)
h−→0

which is the first of the following series of observations:

1. lim en (h) = 0 if and only if f is continuous at a and a0 = f (a).

h−→0

2.
en (h) f (a + h) − f (a)
= − (a1 + a2 h + · · · + an hn−1 )
h h
and so en (h) = (h) if and only if f 0 (a) exists and a1 = f 0 (a).
52 Power Series

3. If f 0 exists on a nbd of a, then differentiating the formula for en gives

e0n (h) = f 0 (a + h) − f 0 (a) − (2a2 h + 3a3 h2 + · · · + nan hn−1 )

and so step 2 gives

e0n (h) f 0 (a + h) − f 0 (a)

= − (2a2 + 3a3 h + · · · + nan hn−2 )
h h
1
Thus e0n (h) = (h) if and only if f 00 (a) exists and a2 = f 00 (a).
2
(m−1)
4. By induction en (h) = (h) if and only if f (m) (a) exists and

1 (m)
am = f (a).
m!

These observations lead to

Definition 5.2 For any f such that f (n) (a) exists, the nth Taylor polynomial
of f at a is
∞
X f (k) (a)
Tn (f ; a) = (x − a)k
k=0
k!
f 00 (a) f (n) (a)
= f (a) + f 0 (a)(x − a) + (x − a)2 + · · · + (x − a)n + · · ·
2! n!
The coefficients are the Taylor coefficients of f at a. The error when f is
approximated near a by Tn satisfies

e(k)
n (h) = (h), k = 0, 1, 2, · · · , n − 1

Remark 5.2 Geometrically, the tangent line

f (a) + f 0 (a)(x − a)

which is T1 (f ; a) is the best linear approximation of f locally at a.

The Taylor quadratic
1
f (a) + f 0 (a)(x − a) + f 00 (a)(x − a)2
2
gives a second order approximation to f near a.
If the values of f and its derivatives to order n are known at a then Tn (f ; a)
expresses the arithmetic that will give approximate values for f (x) when x is
near a.
Power Series 53

Definition 5.3 The Maclaurin series(polynomial) of f is obtained by putting

a = 0 in the Taylor series; i.e.,
∞
X f (k) (0) f 00 (0) 2 f (n) (0) n
xk = f (0) + f 0 (0)x + x + ··· + x + ···
k=0
k! 2! n!

Activity 5.1 1. Find sin 1 approximately√ using a fourth-degree polynomial for

the arithmetic and the values sin π3 = 23 , cos π3 = 12 . Express your answer
√
in terms of 3 and π.
2. Find the general Taylor polynomial for ln(1 + x) at x = 0. Derive six
successive approximate values for ln 2. Comment on their accuracies given
that ln 2 = 0.693147 · · ·

Theorem 5.4 (Taylor’s Theorem)

If f has n derivatives that are continuous on the closed interval joining a and
a + h and if f (n+1) exists on the interior, then there is a value ζ in the interior
such that
0 f 00 (a) 2 f (n) (a) n f (n+1) (ζ) n+1
f (a + h) = f (a) + f (a)h + h + ··· + h + h .
2! n! (n + 1)!
The formula is called Taylor’s formula with the Lagrange form of remainder.
Write:
f 00 (a) f (n) (a) f (n+1) (ζ)
f (x) = f (a)+f 0 (a)(x−a)+ (x−a)2 +· · ·+ (x−a)n + (x−a)n+1
2! n! (n + 1)!

Remark 5.3 If we take the case n = 0, then Taylor’s Theorem is just the Mean-
Value Theorem. Thus, Taylor’s Theorem gives generalizations of the Mean-Value
Theorem.
Z 1 2
1 − e−x
Example 5.4 Evaluate dx.
0 x2

u2 u3 u4 u5
eu = 1 + u + + + + + · · · , −∞ < u < ∞
2! 3! 4! 5!
Then if u = −x2 , we have
2 x4 x6 x8 x1 0
e−x = 1 − x2 + − + − + · · · , −∞ < x < ∞
2! 3! 4! 5!
Thus
2 x4 6 8 x10
1 − e−x 1 − (1 − x2 + − x3! + x4! −
2! 5!
+ · · ·)
=
x2 x2
2 4 6
x x x x8
= 1− + − + + ···
2! 3! 4! 5!
54 Power Series

Since the series converges for all x and so, in particular, converges uniformly for
0 ≤ x ≤ 1, we can integrate term by term to obtain
1 2
1 − e−x
Z
1 1 1 1
2
dx = 1 − + − + − ···
0 x 3 · 2! 5 · 3! 7 · 4! 9 · 5!

Activity 5.2 1. Prove:

x3 x5 x7
tan−1 x = x − + − + ···
3 5 7
where the series is uniformly convergent in −1 ≤ x ≤ 1.

2. Prove that
π 1 1 1
= 1 − + − + ···
4 3 5 7
Unit 6
Limits and Continuity

6.1 What this Unit is all About

In this unit we study limits of functions, continuity and uniform continuity.

By the end of this unit you should be able to:

1. Define a cluster point and the limit of a function.
2.Characterize limits of functions in terms of convergence of sequences.
3. Define continuity and uniform continuity.
4. Prove various theorems on continuity and uniform continuity of functions.
5. Prove that given functions are continuous and/or uniformly continuous.
6.Prove that a uniformly continuous function preserves Cauchy sequences

6.2 Limits of functions

A function f assigns a numerical value f (x) to each point x in the domain Df
on which f is defined. We wish to define the concept of the limit of f (x) as
x −→ x0 .

Definition 6.1 A point x0 is called a cluster point of the set D if and only if
∀ δ > 0 ∃ x ∈ D such that 0 < |x − x0 | < δ.

Thus, a cluster point x0 of a set D has the property that it is always possible to
find points x ∈ D for which x 6= x0 and yet x is as close to x0 as we like.

Example 6.1 Let D = [0, 1). The set of cluster points of D is the interval [0, 1].

Example 6.2 Let D = n1 , n ∈ N . Then D has only one cluster point; namely


the point 0. Note also that in this example 0 ∈

/ D.

Definition 6.2 Let x0 be any cluster point of the domain Df of a function f .

Then lim f (x) = l if and only if ∀ ε > 0 ∃ δ > 0 such that x ∈ Df and
x−→x0
0 < |x − x0 | < δ ⇒ |f (x) − l| < ε.

Example 6.3 • ∀ m, b ∈ R we have lim (mx + b) = mx0 + b. Observe

x−→x0
that the domain of mx + b is the whole real line R. Now, let ε > 0 be given.

55
56 Limits of Indeterminate Type

We need δ > 0 such that 0 < |x − x0 | < δ ⇒ |(mx + b) − (mx0 + b)| < ε ⇔
ε ε
|m(x − x0 )| = |m||x − x0 | < ε ⇒ |x − x0 | < |m| , provided m 6= 0. So δ = |m| .
(Notice that if m = 0, then all x would satisfy the required inequality).
x2 − 1 2 −1
• ∀ x 6= 1, lim = 2, since xx−1 = x + 1 −→ 2 as x −→ 1.
x−→1 x − 1

0, if x 6= 0,
• Let f (x) = Then lim f (x) = 0 6= f (0).
1, if x = 0 x−→0

1, if x ≥ 0,
• f (x) = Then lim f (x) does not exist. In fact, no matter
0, if x < 0 x−→0
how small we make δ > 0, the inequality 0 < |x − 0| < δ will be satisfied by
points x for which f (x) can be either 1 or 0, and we cannot force 1 and 0
to be within arbitrarily small ε > 0 of an one number l.
T
Theorem 6.1 Suppose x0 is a cluster point of the domain Df Dg , lim f (x) =
x−→x0
l and lim g(x) = M . Then
x−→x0

• lim (f ± g)(x) = l ± M
x−→x0

• lim (f g)(x) = lM
x−→x0
 
f l
• lim (x) = , provided M 6= 0 and that x0 is a cluster point of D f .
x−→x0 g M g

Proof 6.1 Exercise!!, cf SMTH011.

6.2.1 Characterization of limits of functions in terms of

convergence of sequences
Theorem 6.2 Suppose x0 is a cluster point of the domain Df of f . Then the
following two statements are logically equivalent.

• lim f (x) = l.
x−→x0

• For any sequence xn ∈ Df \{x0 } such that xn −→ x0 , we have the sequence

f (xn ) −→ l.

Proof 6.2 (⇒), i.e.(a) ⇒ (b)

Suppose lim f (x) = l and let xn ∈ Df \{x0 } such that xn −→ x0 . Then ∀ ε >
x−→x0
0 ∃ δ > 0 such that x ∈ Df and 0 < |x − x0 | < δ ⇒ |f (x) − l| < ε. But, since
xn −→ x0 , ∃ N such that n ≥ N ⇒ |xn −x0 | < δ ⇒ |f (xn )−l| < ε ⇒ f (xn ) → l.
Limits of Indeterminate Type 57

(⇐), i.e.(b) ⇒ (a). Suppose (b) is true. We need to show that lim f (x) = l. Now
x−→x0
suppose this conclusion were false, i.e. suppose that lim f (x) 6= l. Then ∃ ε > 0
x−→x0
such that ∀ δ > 0 ∃ x ∈ Df such that 0 < |x − x0 | < δ and yet |f (x) − l| ≥ ε.
In particular, if we let δn = n1 , then we get xn such that 0 < |xn − x0 | < n1 and
yet |f (xn ) − l| ≥ ε. Now, xn ∈ Df \{x0 } and xn −→ x0 , yet f (xn ) 9 l. This
contradicts (b) which was assumed to be true.♠

sin x1 , if x 6= 0,

Example 6.4 Let f (x) = Show that lim f (x) does not ex-
0, if x = 0 x−→0
ist.
2
Let xn = , n = 1, 2, 3, · · ·. Observe that xn → 0. Now f (xn ) =
 (2n
 + 1)π
sin (2n+1)π
2
= (−1)n which diverges. We conclude that f has no limit at x = 0.

xn − xn0
Activity 6.1 • Find lim , n ∈ N.
x−→x0 x − x0
1
• Prove: lim f (x) = ∞ ⇔ lim = 0.
x−→∞ x−→∞ f (x)

• Prove: lim (x2 − 3x + 7) = 11.

x−→−1

6.3 Continuous functions

Definition 6.3 A function f is called continuous
T at a point x0 ∈ Df provided
that ∀ ε > 0 ∃ δ > 0 such that x ∈ Df (x0 − δ, x0 + δ) ⇒ |f (x) − f (x0 )| < ε.

Remark
If f is continuous at every point x0 ∈ Df , we say f ∈ C(Df ), the family of all
continuous functions.

Definition 6.4 A function f is continuous at a cluster point x0 ∈ Df ⇔ lim f (x) =

x−→x0
f (x0 ).

T
Theorem 6.3 Suppose f and g are each continuous at x0 ∈ Df Dg . Then

• f ± g is continuous at x0 .

• f g is continuous at x0 .
f
• is continuous at x0 provided g(x0 ) 6= 0.
g
58 Limits of Indeterminate Type

• If, moreover, h is continuous at f (x0 ), then the composition hof is contin-

uous at x0 , where hof (x) = h(f (x)).

Proof 6.3 We prove (d) and leave the rest as exercises.

(d). lim (hof )(x) = lim h(f (x)) = h( lim f (x)), since h is continuous at
x−→x0 x−→x0 x−→x0
f (x0 ). But lim f (x) = f (x0 ), since f is continuous at x0 . Thus lim (hof )(x) =
x−→x0 x−→x0
h(f (x0 )) = (hof )(x0 ), and so hof is continuous at x0 .

Example 6.5 • Let i(x) = x, ∀ x ∈ R. Then i is continuous

T at every point
x0 ∈ R. In fact, given ε > 0, we take δ = ε, and ∀ x ∈ Di (x0 − δ, x0 + δ)
we have |i(x) − i(x0 )| = |x − x0 | < δ = ε.
• Let f (x) = x2 , ∀ x ∈
mathbbR. Then f ∈ C(R). This follows easily from the previous theorem,
i.e. f, g continuous implies f g continuous.
Further, this generalises easily to show that g(x) = xn is in C(R), ∀ n ∈ R.
• (Exercise!!) Apply the theorem to see that every polynomial p(x) = an xn +
· · · + a1 x + a0 is continuous on all of R as well. Moreover, every ratio-
p(x)
nal function Q(x) = , where p and q are polynomials, is continuous
q(x)
wherever defined.

1, if x ≥ 0,
• Let f (x) = Then f is discontinuous at 0 but continuous
0, if x < 0
everywhere else. why?

1, if x ≥ 0,
• Let f (x) = Then f is continuous wherever it is defined.
0, if x = −1
Note that −1 is an isolated point of Df .

Theorem 6.4 If f (x) is continuous in a closed interval, it is bounded in the

interval.

Activity 6.2 1. Prove part (c) of Theorem 6.3. The domain D f is that subset
T g
of Df Dg consisting of points x for which g(x) 6= 0.
2. Prove that if a(x) = |x|, the absolute value function, then a ∈ C(R).
√
3. Suppose Q(x) = x is defined ∀ x ≥ 0. Prove that Q ∈ C([0, ∞)).

f be continuous at x0 , and ε > 0. Show ∃ δ > 0 such that x1 , x2 ∈

4. Let T
Df (x0 − δ, x0 + δ) ⇒ |f (x1 ) − f (x2 )| < ε.
 2
x, if x ∈ Q,
5. Let f (x) = 2 Prove that f is continuous at p ⇔ p = 0.
−x , if x ∈/Q
Limits of Indeterminate Type 59

 xn −xn
x−x0
0
, if x 6= x0 ,
6. Let f (x) = Find the value of c which makes
c, if x = x0 , n ∈ N
f ∈ C(R)

6.3.1 Some properties of continuous functions

Lemma 6.1 (Preservation of sign)
SupposeTf is continuous at p and suppose f (p) > 0. Then ∃ δ > 0 such that
x ∈ Df (p − δ, p + δ) ⇒ f (x) > 0.

f (p)
Proof 6.4 Let ε = , which is positive. Then ∃ δ > 0 such that x ∈
2
T f (p) 3f (p)
Df (p − δ, p + δ) ⇒ |f (x) − f (p)| < ε ⇒ 0 < < f (x) < , which
2 2
proves the lemma.

Theorem 6.5 (Intermediate value theorem)

Suppose f ∈ C([a, b]) and suppose k lies strictly between f (a) and f (b). Then
∃ c ∈ (a, b) such that f (c) = k.

Proof 6.5 We will suppose that f (a) < k < f (b). We will let a1 = a and b1 = b
a1 + b 1
and we will use the method of interval halving. Let m = . If f (m) > k, let
2
b2 = m and a2 = a1 . Otherwise let a2 = m and b2 = b1 . Then halve the interval
and proceed as above in that fashion. We generate two sequences of endpoints
both of which are guaranteed to be Cauchy since they are sequences of left and
right endpoints respectively, generated by interval halving. Thus an → Ca and
bn → Cb . But {an } is monotone increasing bounded above by Ca and {bn } is
b−a
monotone decreasing bounded below by Cb . So |Ca − Cb | ≤ |bn − an | = n−1 → 0.
2
so Ca = Cb = c. Since c ∈ [a, b], f is continuous at c. Thus f (c) = lim an ≤ k
n−→∞
and f (c) = lim bn ≥ k of which the two inequalities can be satisfied if and only
n−→∞
if f (c) = k ♠

Remark
If f ∈ C([a, b]), for a closed finite interval [a, b], we can conclude more than just
the continuity of f at each point c ∈ [a, b]. The next theorem shows that f will
be uniformly continuous on [a, b].

Definition 6.5 We say f is uniformly continuous on a domain D if and only if

∀ ε > 0 ∃ δ > 0 such that ∀ x0 ∈ D and ∀ x ∈ D such that |x − x0 | < δ,
we have |f (x) − f (x0 )| < ε.
60 Limits of Indeterminate Type

1
Example 6.6 Let f (x) = on (0, 1). We will show that f is continuous on
x
(0, 1). Let x0 be any point in (0, 1). We want to show that given ε > 0 ∃ δ > 0
such that |f (x) − f (x0 )| < ε whenever 0 < |x − x0 | < δ. Now
1 1 |x − x0 |
|f (x) − f (x0 )| = − =
x x0 |x||x0 |
Our problem is how to control the denominator. We see that as x becomes small,
the fraction becomes large. However, if x is reasonably close to x0 , for example,
x0
if we take x > , we have
2
1 1 |x − x0 | 2|x − x0 | 2δ
|f (x) − f (x0 )| = − = < <
x x0 |x||x0 | |x0 |2 |x0 |2

2δ ε|x0 |2
In other words we want = ε, i.e. δ = . Thus if |x − x0 | < δ, then
|x0 |2 2
|f (x) − f (x0 )| < ε.
We see that δ depends on both the point x0 and ε. Then, given ε > 0, the same
δ DOES NOT work for all points.

Example 6.7 Let f (x) = x2 . We will prove that f (x) is uniformly continuous
on [0, 4].
We want to show that ∀ ε > 0 ∃ δ > 0 such that ∀ x1 ∈ [0, 4] and ∀ x ∈ [0, 4]
|f (x) − f (x1 )| < ε whenever |x − x1 | < δ. Now,

|f (x) − f (x1 )| = |x2 − x21 |

= |(x − x1 )(x + x1 )|
= |x − x1 ||x + x1 |
≤ (|x| + |x1 |)|x − x1 |
≤ (4 + 4)|x − x1 |
= 8|x − x1 |
< 8δ
ε
since |x|, |x1 | ≤ 4. We have 8δ = ε ⇒ δ = . Thus if |x − x1 | < δ, then
8
|f (x) − f (x1 )| < ε. Hence f is uniformly continuous on [0, 4].

Theorem 6.6 If f ∈ C([a, b]), i.e. if f is continuous on [a, b], then f is uniformly
continuous on [a, b].

Proof 6.6 We suppose that the theorem were false, and we will deduce a contra-
diction. Thus we suppose f is not uniformly continuous on [a, b]. Then ∃ ε > 0
such that no δ > 0 will suffice to meet the uniform continuity condition. Hence
1 1
if δn = , ∃ xn , yn ∈ [a, b] such that |xn − yn | < but |f (xn ) − f (yn )| ≥ ε. Now,
n n
Limits of Indeterminate Type 61

since {xn } is a bounded sequence, the Bolzano-Weierstrass theorem guarantees

the existence of a convergent subsequence xnk → p ∈ [a, b] as k → ∞. Now,

|ynk − p| ≤ |ynk − xnk | + |xnk − p| → 0 + 0 = 0

Thus ynk → p as well. Therefore, since f is continuous at p, f (p) = lim f (xnk ) =

k−→∞
lim f (ynk ). This means that
k−→∞

|f (xnk ) − f (ynk | ≤ |f (xnk − f (p)| + |f (p) − f (ynk | → 0 + 0 = 0

But |f (xnk ) − f (ynk | ≥ ε, ∀ k, and so we have a contradiction.

Theorem 6.7 If a function f is uniformly continuous on an interval (a, b) and

if {xn } is a Cauchy sequence in (a, b), then {f (xn )} is a Cauchy sequence.
[i.e. A uniformly continuous function preserves Cauchy sequences.]

Proof 6.7 Suppose f is uniformly continuous on (a, b) and {xn } is a Cauchy

sequence in (a, b). Since {xn } is a Cauchy sequence in (a, b), given any δ > 0
(taking ε as δ), there exists a positive integer N such that for all m, n > N we
have |xm − xn | < δ. Moreover, since f is uniformly continuous on (a, b), then
given ε > 0 ∃ δ > 0 such that ∀ x, y ∈ (a, b) with |y − x| < δ, |f (y) − f (x)| < ε.
Thus, given any ε > 0, we can find a positive integer N (via δ > 0) such that if
m.n > N, |f (xm ) − f (xn )| < ε ⇒ {f (xn )} is a Cauchy sequence.

1
Example 6.8 Let f (x) = on (0, 1]. We show that f is not uniformly contin-
x
1
uous on (0, 1] using the above theorem. Let xn = . Then clearly xn ∈ (0, 1] and
n
∀ m, n > N we have
 
1 1 1 1 1
− < max , < <ε
m n m n N

⇒ {xn } is Cauchy. However f (xn ) = n which is clearly not Cauchy. Thus f

cannot be uniformly continuous on (0, 1].

Activity 6.3 1. Suppose f ∈ C([a, b]) and f (a) > k > f (b). Prove: ∃ c ∈
(a, b) such that f (c) = k. [Hint: Consider g(x) = −f (x).]

2. Let p(x) = x3 + 3x2 − 2x − 1. Prove that the polynomial equation p(x) = 0

has a root somewhere in the interval (0, 1).

3. Let p(x) = x4 +x3 −2x2 +x+1. Prove that the polynomial equation p(x) = 0
has a root in (−1, 0).
62 Limits of Indeterminate Type

4. Let p(x) = a2n+1 x2n+1 + · · · + a1 x + a0 be any polynomial of odd degree.

Prove: The equation p(x) = 0 has at least one real root. (Hint: Prove
p(x) → ±∞ depending on whether x → ±∞ and consider then that p must
have both positive and negative values.)

5. Prove the following fixed point theorem: Suppose f ∈ C([0, 1]) and suppose
0 ≤ f (x) ≤ 1 ∀ x ∈ [0, 1]. Then ∃ c ∈ [0, 1] such that f (c) = c. (The
point c is then called a fixed point for the function f .)[Hint: Consider
g(x) = f (x) − x].

6. Suppose f is uniformly continuous on (a, m] and also on [m, b). Prove that
f is uniformly continuous on (a, b).
1
7. Let f (x) = , ∀ x ∈ (0, 1). Is f uniformly continuous on (0, 1)? Justify
x
your conclusions.

8. Let f (x) = x2 , ∀ x ∈ R. Is f ∈ C(R)? Is f uniformly continuous on R?

Justify your conclusions.

9. Suppose f is uniformly continuous on D and suppose E ⊂ D. Prove: f is

uniformly continuous on E as well.

10. Is f (x) = x2 uniformly continuous on (0, 1)? why or why not?

√
11. Let f (x) = x. Prove each of the following conclusions:

(a) f is uniformly continuous on [0, 1].

√ √
(b) f is uniformly continuous on [1, ∞). (Hint: Consider x− x1 and
rationalise the numerator.)
(c) f is uniformly continuous on [0, ∞).

12. Give examples of the following:

(a) A function that is not uniformly continuous on [0, 2].

(b) A continuous function that is not uniformly continuous on [0, 1).
(c) A continuous function that is not uniformly continuous on [0, ∞).
(d) A bounded function on (0, 1) that is uniformly continuous.
Unit 7
Differentiation and Partial
Differentiation

7.1 What this Unit is all About

In this unit we discuss differentiation of inverse trigonometric functions, hyper-
bolic functions, inverse hyperbolic functions and parametric equations. We end
the unit by discussing partial derivatives up to second order derivatives and ap-
plication to rates of change and small changes.

By the end of this unit you should be able to:

1. differentiate inverse trigonometric functions.
2. differentiate hyperbolic functions and inverse hyperbolic functions.
3. differentiate parametric equations.
4. perform partial differentiation.
5. solve problems involving rates of change and small changes.

7.2 Differentiation
The concept of differentiation was introduced in SMTH011 and there you dealt
with problems of finding derivatives of various functions. In this unit we deal
with differentiation as a continuation from where it was left there in SMTH011.
Suffice to say that the techniques and differentiation formulae that you have
learnt remain the same, the only new thing is that this time around we will be
applying these to different type of functions. In particular we focus on inverse
trigonometric functions, hyperbolic functions, inverse hyperbolic functions and
parametric equations.
Perhaps it will be worthwhile to start by recalling the techniques/formulae that
you have used before.

Differentiation from first principles

If y = f (x), then
dy f (x + h) − f (x)
= lim
dx h→o h

63
64 Differentiation and Partial Differentiation

Derivative of a constant
d
If k is a constant, (k) = 0
dx

The power rule for differentiating polynomials

d
(axn ) = anxn−1 , a is a constant.
dx

Product Rule for differentiating products

d
(f (x) · g(x)) = f (x) · g 0 (x) + g(x) · f 0 (x)
dx

Quotient Rule for differentiating rational functions

g(x) · f 0 (x) − f (x) · g 0 (x)

 
d f (x)
=
dx g(x) (g(x))2

Chain Rule or function of a function rule

d dy dy du
[f (x)]n = n[f (x)]n−1 · f 0 (x) and if y = f (u(x)), then = ·
dx dx du dx

dy
You are also urged to recall implicit differentiation for finding when y is an
dx
implicit function of x. This, as well as logarithmic differentiation, will also be
encountered in the foregoing.

7.2.1 Inverse trigonometric functions

Recall that the only functions for which inverse functions are defined are one-
to-one functions. Trigonometric functions, however, are not one-to-one and thus
they do not have inverse functions. However, if we restrict their domains, they
will be one-to-one in the restricted domains and thus we can find inverses of such
restricted trigonometric functions. In particular the sine function
π π
y = sin x, − ∞ < x < ∞ is not one-to-one, whereas y = sin x, − ≤ x ≤
2 2
is one-to-one and so the later has an inverse. This inverse is denoted by sin−1 or
arcsin. Now, in general f −1 (x) = y ⇐⇒ f (y) = x. In this way
π π
sin−1 x = y ⇐⇒ sin y = x and − ≤y≤ .
2 2
Please note that
1
sin−1 x 6=
sin x
Differentiation and Partial Differentiation 65

and this applies to all other inverse functions. If −1 ≤ x ≤ 1, sin−1 x is the

π π
number between − and whose sine is x. Observe that
2 2
π π
sin−1 (sin x) = x for − ≤ x ≤
2 2
and
sin(sin−1 x) = x for −1≤x≤1
These cancellation equations correspondingly apply to all other inverse functions
as well.

Activity 7.1 Find the exact value of each expression:

√   
−1 3 −1 −1 √1
(i) sin 2
(ii) cos (−1) (iii) tan 3

(iv) sec−1 2 (v) tan−1 tan 3π (vi) cos arcsin 12

4
√

(vii) tan sin−1 2
(viii) cosec arccos 35 sin 2 tan−1




3
(ix) 2

Now, since the inverse sine function sin−1 x is continuous in the interval in which it
is defined, it is differentiable. We use implicit differentiation to find its derivative.

dy
Let y = sin−1 x, then sin y = x =⇒ cos y = 1 and
dx
dy 1 p √
= . Now, since cos y = 1 − sin2 y = 1 − x2 then
dx cos y
dy 1 1
= =√
dx cos y 1 − x2
and so
d 1
(sin−1 x) = √ −1<x<1
dx 1 − x2
dy
In general if y = sin−1 (f (x)), then sin y = f (x) and cos y = f 0 (x), so that
dx
dy f 0 (x) f 0 (x)
= =p
dx cos y 1 − [f (x)]2

3x2 −1
 dy
Example 7.1 If y = arcsin 2
find and simplify the answer.
dx
Solution
3x2 − 1
Here f (x) = =⇒ f 0 (x) = 3x and therefore
2
dy 3x 3x 6x
=q =q =√
dx 2 2 4−(3x2 −1)2 3 − 9x4 + 6x2
1 − 3x 2−1


4
66 Differentiation and Partial Differentiation

Example 7.2 Differentiate y = arcsec(2x2 )

Solution  
2 1 1 0 1
First note that arcsec(2x ) = arccos . Thus f (x) = =⇒ f (x) = −
2x2 2x2 x3
and
− − x13 1


dy x3 2
=q = √ = √
dx 4x4 −1 x 4x4 − 1
1 − 4x14 2x2

d
Example 7.3 Find (cot−1 (cosec x))
dx
Solution   
d −1 d −1 1 d cos x
(cot (cosec x)) = tan = (tan−1 (sin x)) =
dx dx cosec x dx 1 + sin2 x
Activity 7.2 1. Differentiate the following and simplify where possible
(a) y = arctan x+1


2
(b) y = (tan−1 x)2
(c) y = tan−1 x2
1
(d) y =
tan−1 x
(e) y = cos−1 (sin−1 x)
q
(f ) y = arctan 1−x
1+x
b+a cos x


(g) y = arccos a+b cos x
 
1+x2
(h) y = arccos 1−x2
x


(i) y = arctan 16−x2
(j) y = cos−1 (e ) 2x

(k) y = earccot 2x

d2 y dy
2. If y = earctan x , show that (1 + x2 ) 2
+ (2x − 1) =0
dx dx
2arcsin x d2 y2 dy
3. If y = e , prove that (1 − x ) 2 − x − 4y = 0
dx dx

7.2.2 Hyperbolic functions

Recall that trigonometric functions are defined in terms of the cosine and the
1 1 cos x
sine functions. For instance sec x = , cot x = = , etc,etc.
cos x tan x sin x
Correspondingly, hyperbolic functions are all defined from the hyperbolic cosine
function (cosh) and the hyperbolic sine function (sinh).
Differentiation and Partial Differentiation 67

ex + e−x ex − e−x
Definition 7.1 1. cosh x = 2. sinh x =
2 2

sinh x ex − e−x
From these two, we can define tanh x = = x ,
cosh x e + e−x
cosh x ex + e−x 1 2
coth x = = x −x
, sech x = = x ,
sinh x e −e cosh x e + e−x
1 2
cosech x = = x , etc,etc.
sinh x e − e−x
Hyperbolic functions also play an important role in integral calculus.

To each trigonometric identity there is a corresponding hyperbolic identity. To

obtain the hyperbolic identity from the trigonometric identity, we use Osborne’s
rule which says; simply replace the trigonometric function with the corresponding
hyperbolic function and in front of a product or an implied product of sine put a
negative sign. For example, from sin2 x+cos2 x = 1 we obtain cosh2 x−sinh2 x = 1
and from 1 + tan2 x = sec2 x we obtain 1 − tanh2 x = sech2 x.

Activity 7.3 Complete the following table

Trigonometric Identity Hyperbolic Identity

2 2
sin A + cos A = 1 cosh2 A − sinh2 A = 1
1 + tan2 A = sec2 A 1 − tanh2 A = sech2 A
cosec2 A = 1 + cot2 A
sin(−A) = − sin A
cos(−A) = cos A
tan(−A) = − tan A
sin(A + B) = sin A cos B + cos A sin B
cos(A + B) = cos A cos B − sin A sin B
tan A + tan B
tan(A + B) =
1 − tan A tan B
sin 2A = 2 sin A cos A
cos 2A = cos2 A − sin2 A
1
sin2 A = (1 − cos 2A)
2
2 1
cos A = (1 + cos 2A)
2
1
sin A cos B = [sin(A − B) + sin(A + B)]
2
1
sin A sin B = [cos(A − B) − cos(A + B)]
2
1
cos A cos B = [cos(A − B) + cos(A + B)]
2

d ex + e−x ex − e−x
 
d
Now, (cosh x) = = = sinh x
dx dx 2 2
68 Differentiation and Partial Differentiation

ex − e−x ex + e−x
 
d d
and (sinh x) = = = cosh x.
dx dx 2 2
From these two derivatives, and in conjunction with the laws or techniques of
differentiation, we can obtain the derivative of any given hyperbolic function.

dy
Example 7.4 If y = cosh3 x, find .
dx
Solution
dy
The chain rule gives = 3 cosh2 x sinh x
dx
dy
Example 7.5 If y = sinh2 x cosh x, find .
dx
Solution
The chain rule and product rule give
dy
= 2 sinh x cosh x cosh x + sinh2 x sinh x = 2 sinh x cosh2 x + sinh3 x
dx
dy
Example 7.6 If y = x3 etanh x , find .
dx
Solution
dy
= 3x2 etanh x + x3 sech2 xetanh x
dx

Example 7.7 Differentiate y = sin−1 (cosh x) + cosh(sin−1 x)

Solution
dy sinh x 1 sinh x sinh(sin−1 x)
=p + sinh(sin−1 x) · √ =p + √
dx 1 − cosh2 x 1 − x2 1 − cosh2 x 1 − x2

Activity 7.4 Differentiate the following:

2
1. y = cosh
x
1
2. y = sinh2
x
√
3
3. y = tanh x2
1+tanh x


4. y = ln 1−tanh x

5. y = sinh(cos−1 x) + sin−1 (sech x)

6. y = sinh(cos−1 x) + cos−1 (sinh x)

√ √
7. y = xcosech x
Differentiation and Partial Differentiation 69

8. y = xsinh 3x

9. y = sech3 (2x )

10. y = sech(sinh 2x − sin 3x)

7.2.3 Inverse Hyperbolic Functions

In this section we look at inverse hyperbolic functions. Since the discussion here
follows more or less the discussion under inverse trigonometric functions, similar
details will be omitted. If
y = cosh−1 x
then
cosh y = x
and similarly
y = arcsinh x =⇒ sinh y = x.
Now, since hyperbolic functions are defined in terms of the exponential function,
it is reasonable to expect inverse hyperbolic functions to be defined in terms of
the natural logarithmic function.

Let y = cosh−1 x. Then cosh y = x =⇒

ey + e−y
=x
2
=⇒ ey + e−y = 2x
=⇒ e2y − 2xey + 1 = 0
=⇒ z 2 − 2xz + 1 = 0, z = ey
√
2x ± 4x2 − 4
=⇒ z =
√2
=⇒ z = x ± x2 − 1
√
=⇒ ey = x ± x2 − 1
√
=⇒ y = ln(x ± x2 − 1)
√
=⇒ cosh−1 x = ln(x + x2 − 1), x ≥ 1

Activity 7.5 Show that

√
arcsinh x = ln(x + x2 + 1) −∞<x<∞

and  
1 1+x
arctanh x = ln −1<x<1
2 1−x
70 Differentiation and Partial Differentiation

The process of obtaining derivatives of inverse hyperbolic functions is as explained

under inverse trigonometric functions. In general if
y = sinh−1 f (x),
then
sinh y = f (x).
Differentiating implicitly gives
dy dy f 0 (x)
cosh y = f 0 (x) =⇒ = .
dx dx cosh y
From the hyperbolic identity
cosh2 y − sinh2 y = 1,
we have q p
cosh y = 1 + sinh2 y = 1 + [f (x)]2 ,
so that
d f 0 (x)
(sinh−1 f (x)) = p .
dx 1 + [f (x)]2
Similar formulae can be deduced for other inverse hyperbolic functions.
dy
Example 7.8 If y = arctanh x2 , find
dx
Solution
Here, f (x) = x2 and f 0 (x) = 2x so that
dy 2x
=
dx 1 − x4
dy
Example 7.9 If ln y = cosh−1 x, find
dx
Solution
1 dy 1
ln y = cosh−1 x =⇒ =√ and so
y dx x2 − 1
−1
dy y ecosh x
=√ =√
dx x2 − 1 x2 − 1

Example 7.10 Differentiate y = cosech−1 (sech x3 )

Solution
y = cosech−1 (sech x3 ) = sinh−1 1
= sinh−1 (cosh x3 ). Thus,


sech x3

dy 3x2 sinh x3
=p
dx 1 + cosh2 x3
Differentiation and Partial Differentiation 71

Activity 7.6 1. Differentiate the following:

cot(tanh x)
(a) y =
sinh(cosh−1 x)
(b) y = tanh−1 (cosh x) · sech(sinh x)
(c) y = coth−1 (tan−1 2x)
(d) y = sech−1 x32

(e) y = ln(arccosh x)arcsinh x

(f ) y = sinh−1 (sin x) + sin(arcsinh x)

dy
2. Determine if:
dx
−1
(a) (cosh x)sinh y
= (tanh−1 y)coth x
−1 x
(b) (cot−1 y)ln y = xcosh x e 2
 
(c) arccoth xy = ln x + ln y

d2 y dy
3. If ln y = cosh−1 x, find the value of (1 + x2 ) 2
+x −y
dx dx

7.2.4 Parametric Equations

Sometimes it is not possible to express y explicitly in terms of x, i.e. it may
not be possible to have y = f (x). An alternative to this representation is to
relate the variables x and y through another independent variable, say θ, i.e.
x = x(θ), y = y(θ). This third variable θ is called a parameter and

x = x(θ), y = y(θ)

are called parametric equations. Now, given such parametric equations, our ob-
dy d2 y
jective here is to find and 2 .
dx dx

dy dy dθ dθ 1
Let x = x(θ), y = y(θ). Chain rule gives = · . Since = dx , we
dx dθ dx dx dθ
have that
dy
dy dθ
= dx .
dx dθ

d2 y
   
d dy d dy dθ
Further 2 = = · , which gives
dx dx dx dθ dx dx
d dy


d2 y dθ dx
= dx .
dx2 dθ
72 Differentiation and Partial Differentiation

dy d2 y
Example 7.11 Given x = 2 cos 2t, y = 2 sin t, determine and .
dx dx2
Solution
dx dy
= −4 sin 2t, = 2 cos t and
dt dt
dy
dy dt 2 cos t cos t cosec t
= dx
= =− =−
dx dt
−4 sin 2t 2 sin 2t 4
 
d dy cosec t cot t
Now, = and
dt dx 4
d dy


d2 y dt dx cosec t cot t cosec3 t
= dx = − =−
dx2 dt
16 sin 2t 32

Example 7.12 Obtain an equation of the tangent line to the parametric curve

x = 2 sin 2t, y = 2 sin t

√
at the point ( 3, 1).

Solution
dx dy
= 4 cos 2t, = 2 cos t. The slope is
dt dt
dy
dy dt 2 cos t cos t
= dx
= =
dx dt
4 cos 2t 2 cos 2t
√ π
The point ( 3, 1) corresponds to the parameter value t = , so that the slope of
6
the tangent at this point is
√
cos π6


dy 3
|t= π6 = π

=
dx 2 cos 3 2
√ √
3 √ 3 1
Thus, the equation of the tangent line is y − 1 = (x − 3) =⇒ y = x− .
2 2 2
Activity 7.7 1. Determine an equation of the tangent line to the curve
x = t + 1, y = t3 + t at the point t = −1.
4

2. A curve C is defined by the parametric equations x = t2 , y = t3 − 3t. Show

that C has two tangents at the point (3, 0) and find their equations.

3. Obtain an equation of the tangent to the cycloid

π
x = r(θ − sin θ), y = r(1 − cos θ) at the point where θ = .
3
Differentiation and Partial Differentiation 73

dy d2 y
4. Determine and for each of the following:
dx dx2
(a) y = sinh t, x = t + cosh t
(b) x = cos−1 (ln t), y = −2 sin−1 (3t )
(c) y = sin θ + θ cos θ, x = cos θ − θ sin θ
π
(d) y = 3 sinh θ, x = cosh3 θ, at θ =
6
t t
(e) x = e cos t, y = e sin t
(f ) x = cos3 2θ, y = sin3 2θ
(g) x = a cos 2θ, y = a sin θ, a-constant.

7.3 Partial Derivatives

7.3.1 First partial derivatives
dy
Recall that if y = f (x), then = f 0 (x) can be obtained from first principles
dx
using the definition
dy f (x + h) − f (x)
= lim .
dx h→o h
In this section we begin by adapting this definition to the case of a function of
two variables f (x, y).

Definition 7.2 If f is a function of two variables, its partial derivatives are

the functions fx and fy defined by

f (x + h, y) − f (x, y)
fx (x, y) = lim
h→o h
and
f (x, y + h) − f (x, y)
fy (x, y) = lim
h→o h

Notations for partial derivatives

If z = f (x, y), we write

∂f ∂ ∂z
fx (x, y) = fx = = f (x, y) =
∂x ∂x ∂x
and
∂f ∂ ∂z
fy (x, y) = fy = = f (x, y) =
∂y ∂y ∂y
74 Differentiation and Partial Differentiation

Rule for finding partial derivatives of z = f (x, y)

∂z
To find , regard y as a constant and differentiate f (x, y) with respect to x. To
∂x
∂z
find , regard x as a constant and differentiate f (x, y) with respect to y. All
∂y
the usual rules/laws and properties of differentiation apply.

Example 7.13 If f (x, y) = x3 − x2 y 3 + 2y 2 , find fx (2, −1) and fy (2, −1)

Solution
fx (x, y) = 3x2 − 2xy 3 and therefore fx (2, −1) = 3 · 22 − 2 · 2 · (−1)3 = 16
fy (x, y) = −3x2 y 2 + 4y and so f − y(2, −1) = −3 · 22 · (−1)2 + 4 · (−1) = −16

x
 ∂f ∂f
Example 7.14 If f (x, y) = cos 1+y
, calculate and .
∂x ∂y

Solution
       
∂f x ∂ x x 1 1 x
= − sin = − sin · =− sin
∂x 1+y ∂x 1+y 1+y 1+y 1+y 1+y
and
       
∂f x ∂ x x −x x x
= − sin = − sin · = sin
∂y 1+y ∂y 1+y 1 + y (1 + y)2 (1 + y)2 1+y

∂z ∂z
Example 7.15 Find and if z is defined implicitly as a function of x and
∂x ∂y
y by the equation x2 + y 2 + z 2 + 3xyz = 1

Solution

∂z ∂z
2x + 2z
+ 3yz + 3xy =0
∂x ∂x
∂z
=⇒ (2z + 3xy) = −2x − 3yz
∂x
∂z 2x + 3yz
=⇒ =−
∂x 2z + 3xy

Similarly,

∂z ∂z
2y + 2z + 3xz + 3xy =0
∂y ∂y
∂z 2y + 3xz
=⇒ =−
∂y 2z + 3xy
Differentiation and Partial Differentiation 75

Example 7.16 If f (x1 , x2 , x3 ) = x1 sin(2x2 −x3 ), find the first partial derivatives
∂f ∂f ∂f
, and .
∂x1 ∂x2 ∂x3
Solution
∂f
= sin(2x2 − x3 )
∂x1
∂f
= 2x1 cos(2x2 − x3 )
∂x2
and
∂f
= −x1 cos(2x2 − x3 )
∂x3

7.3.2 Second partial derivatives

If f is a function of two variables, then we have seen that its partial derivatives
∂f
fx =
∂x
and
∂f
fy =
∂y
are also functions of two variables. We can thus consider their partial derivatives
∂ 2f
(fx )x = ,
∂x2
∂ 2f
(fx )y = ,
∂y∂x
∂ 2f
(fy )x =
∂x∂y
and
∂ 2f
(fy )y =
∂y 2
which are called second partial derivatives of f . The notation used, and
indeed which also provides a way of calculating second partial derivatives, is the
following:
∂ 2f
 
∂ ∂f
(fx )x = fxx = =
∂x2 ∂x ∂x
2
 
∂ f ∂ ∂f
(fx )y = fxy = =
∂y∂x ∂y ∂x
∂ 2f
 
∂ ∂f
(fy )x = fyx = =
∂x∂y ∂x ∂y
76 Differentiation and Partial Differentiation

∂ 2f
 
∂ ∂f
(fy )y = fyy = 2
=
∂y ∂y ∂y
xy
Example 7.17 Given z = , find all the second partial derivatives.
x−y

Solution

∂z y(x − y) − xy y2
= = −
∂x (x − y)2 (x − y)2
∂z x(x − y) + xy x2
= =
∂y (x − y)2 (x − y)2
∂ 2z y2 2y 2
   
∂ ∂z ∂
= = − = −
∂x2 ∂x ∂x ∂x (x − y)2 (x − y)3
∂ 2z y2 −2y(x − y)2 − 2y 2 (x − y) 2xy 2 − 2x2 y
   
∂ ∂z ∂
= = − = =
∂y∂x ∂y ∂x ∂y (x − y)2 (x − y)4 (x − y)4
∂ 2z x2 2x(x − y)2 − 2x2 (x − y) 2xy 2 − 2x2 y
   
∂ ∂z ∂
= = = =
∂x∂y ∂x ∂y ∂x (x − y)2 (x − y)4 (x − y)4
∂ 2z x2 2x2
   
∂ ∂z ∂
= = = −
∂y 2 ∂y ∂y ∂y (x − y)2 (x − y)3

Activity 7.8 1. Determine the first partial derivatives of the following func-
tions:

(a) f (x, y) = y 5 − 3xy

x−y
(b) f (x, y) =
x+y
(c) z = tan(xy)
(d) w = ln(x + 2y + 3z)
(e) u = xy sin−1 (yz)
(f ) w = zexyz
∂z ∂z
2. Determine and
∂x ∂y
√ q
3
(a) z = tanh xy − ln xy
(b) x2 + y 2 + z 2 = 3xyz
(c) x − z = arctan(yz)
(d) sin(xyz) = x + 2y + 3z
(e) yz = ln(x + z)
Differentiation and Partial Differentiation 77

(f ) z = f (x + y)
(g) z = f (x)g(y)
(h) z = f (xy)
(i) z = f ( xy )
3. Obtain all the second partial derivatives
 
−1 x+y
(a) z = tan
1 − xy
1
(b) z = p
x2 + y 2
(c) z = cos xy
p
(d) z = ln x2 + y 2
x+y+z ∂u ∂u ∂u
4. If u = p , show that x +y +z =0
x2 + y 2 + z 2 ∂x ∂y ∂z
∂z ∂z
5. If z = ln (ex + ey ) show that + =1
∂x ∂y
∂ 2z ∂ 2z
6. Show that the equation + = 0 is satisfied by
∂x2 ∂y 2
1 y
z = ln x2 + y 2 + tan−1
p
2 x
∂z ∂z
7. If z = f yz , show that x


+y =0
∂x ∂y
8. If z = z(u, v) where u = u(x, y), v = v(x, y), prove that
∂z ∂z ∂u ∂z ∂v
(a) = +
∂x ∂u ∂x ∂v ∂x
∂z ∂z ∂u ∂z ∂v
(b) = +
∂y ∂u ∂y ∂v ∂y
9. If x = ρ cos φ, y = ρ sin φ, prove that if U is a twice differentiable function
of x and y,
∂ 2U ∂ 2U ∂ 2U 1 ∂U 1 ∂ 2U
+ = + +
∂x2 ∂y 2 ∂ρ2 ρ ∂ρ ρ2 ∂φ2

7.3.3 Applications: Differentials

For a function of one variable y = f (x), the differential of y is defined as
dy = f 0 (x)dx. For a differentiable function of two variables, z = f (x, y), the
differential dz, also called the total differential, is defined by
∂z ∂z
dz = dx + dy
∂x ∂y
78 Differentiation and Partial Differentiation

This expression is the basis for solving problems involving small increments and
rates of change. From
∂z ∂z
dz = dx + dy
∂x ∂y
we get
∂z ∂z
∆z ≈ ∆x + ∆y
∂x ∂y
which gives an approximation of the small change in z, ∆z as a result of small
changes in x and y, ∆x and ∆y respectively. We also have
dz ∂z dx ∂z dy
= + ,
dt ∂x dt ∂y dt
dz
where t is time. This expression gives a calculation of the rate of change of z,
dt
dx dy
given the rates of changes and of x and y respectively. If z is a function
dt dt
of only one variable or a function of more than two variables, the formulae are
adapted accordingly.

Small increments

Example 7.18 The dimensions of a rectangular box are measured to be 70cm, 60cm
and 50cm, and each measurement is correct to within 0.2cm. Estimate the largest
possible error when the volume of the box is calculated from these measurements.

Solution
If the dimensions of the box are x, y and h, then its volume is V = xyh
∂V ∂V ∂V
∆V ≈ ∆x + ∆y + ∆h = yh∆x + xh∆y + xy∆h
∂x ∂y ∂h
Here |∆x| ≤ 0.2, |∆y| ≤ 0.2 and |∆h| ≤ 0.2. Thus

∆V ≈ (60)(50)(0.2) + (70)(50)(0.2) + (70)(60)(0.2) = 2140

Thus, an error of only 0.2cm in measuring the dimensions leads to an error of

2140cm2 , i.e. about 1% of the volume.

Example 7.19 The lift, L, of a body in a fluid of density ρ is given by

1
L = CρV 2 A
2
where C is the lift coefficient, V is the free stream velocity and A is the area. If
C can be measured to within 1%, ρ to within 0.5%, V to within 0.6% and A to
within 0.1%, find the largest percentage error in the value of L.
Differentiation and Partial Differentiation 79

Solution

∂L ∂L ∂L ∂L
∆L ≈
∆C + ∆ρ + ∆V + ∆A
∂C ∂ρ ∂V ∂A
1 1 1
= ρV 2 A(±0.01C) + CV 2 A(±0.005ρ) + CρV A(±0.006V ) + CρV 2 (±0.001A)
2 2 2
1 1 1 1
= CρV 2 A(±0.01) + CρV 2 A(±0.005) + CρV 2 A(±0.012) + CρV 2 A(±0.001)
2 2 2 2
1
= CρV 2 A(±0.01 ± 0.005 ± 0.012 ± 0.001)
2
= L(±0.028)
and the largest percentage error in the value of L is 2.8%

Rates of change

Example 7.20 The altitude of a right circular cone is 20cm and is increasing
at 0.1cms−1 . The radius of the base is 10cm and is decreasing at 0.2cms−1 . How
fast is the volume of the cone changing?

Solution
1
The volume of the right circular cone is given by V = πr2 h where r is the radius
3
dh dr
and h is the altitude. = 0.1 and = −0.2
dt dt
dV ∂V dr ∂V dh
= +
dt ∂r dt ∂h dt
2 dr 1 2 dh
= πrh + πr
3 dt 2 dt
2 1
= π(10)(20)(−0.2) + π(100)(0.1)
3 3
= −73.3
and the volume is decreasing at a rate of 73.3cm3 s−1 .

Activity 7.9 1. The radius of a cylinder increases at the rate of 0.3cm/s

while the height decreases at the rate of 0.4cm/s. Find the rate at which
the volume is changing at the instant when r = 0.06m and h = 14cm.
2. The frequency, f , of an LC circuit is given by
1
f= √
2π LC
where L represents an inductance and C a capacitance. If L is decreased by
1.5% and C is decreased by 0.5%, find the approximate percentage change
in f .
80 Differentiation and Partial Differentiation

3. A square pyramid reservoir has a base of side 12m and height 18m. It
contains oil that is flowing out at a rate of 2m3 s−1 . At the moment when
the depth of oil is 8m, find the rate at which the depth is decreasing.
Hint:The volume of a square pyramid with a base of side a and height h is
1
given by V = a2 h.
3
4. The perimeter of a rectangle is 25m and one side decreases at a rate of
0.02m/s. At the moment when the side is 25cm, at what rate does the area
change if the perimeter remains constant?

5. The height of a cone of volume 20cm3 decreases at a rate of 0.15ms−1 while

the volume remains constant. At what rate does the radius increase when
the altitude is 2.4cm?

6. The moment of inertia, I, of a body is given by

1 3
I= bd
12
where b and d are dimensions of the body. If b = 5m, d = 0.8m and the
change in b is 5cm and the change in d is 0.8cm, find the approximate
change in I.
Unit 8
Integration

8.1 What this Unit is all About

In this unit we discuss the Riemann Integral and the properties of Riemann
integrable functions, multiple integrals, line integrals and surface integrals.

By the end of this unit, you should be able to:

1. Prove that a given function is Riemann integrable.

2. State and prove theorems on properties of Riemann integrable functions.
3. Evaluate double integrals and triple integrals.
4. Evaluate line integrals.
5. State, verify and use Green’s theorem and Stoke’s theorem.

8.2 The Riemann Integral

Definition 8.1 A partition P = {x0 , x1 , · · · , xn } of [a, b] is a finite set of points
such that a = x0 < x1 < x2 < · · · < xn = b.

In this notation, the (closed) ith subinterval is [xi−1 , xi ] and ∆xi = xi − xi−1 is its
length, i = 1, 2, · · · , n. We want to define integrability for a bounded function f
on a closed interval [a, b]. Using the boundedness of f , we introduce, as heights
of rectangles, mi = inf {f (x); xi−1 ≤ x ≤ xi } and Mi = sup{f (x); xi−1 ≤ x ≤ xi }
and form n n
X X
the lower sum L(f, P ) = mi (xi − xi−1 ) = mi ∆xi and
i=1 i=1
Xn Xn
the upper sum U (f, P ) = Mi (xi − xi−1 ) = Mi ∆xi .
i=1 i=1
Clearly L(f, P ) ≤ U (f, P ) for any f and P . Any partition of [a, b] that contains
P is called
S a refinement of P . Two partitions P an Q have common refinements;
i.e. P Q is one.

Example 8.1 Suppose [a, b] = [0, 3]. Let P = {0, 1, 2, 3} and Q = {0, 12 , 1, 2, 2 12 , 3}.
Then Q is a refinement of P .

81
82 Integration

Example 8.2 If [a, b] = [0, 4], P = {0, 1, 2, 3, 4} and Q = {0, 2, 2 12 , 3, 4}. Q is

not a refinement of P .

Now, observe that since f is bounded on [a, b], ∃ numbers m and M such that
m ≤ f (x) ≤ M ∀ x ∈ [a, b]. Thus m ≤ mi (f ) ≤ Mi (f ) ≤ M, ∀ i = 1, 2, · · · , n.
So
n
X n
X n
X n
X
m(b − a) = m∆xi ≤ mi (f )∆xi ≤ Mi (f )∆xi ≤ M ∆xi = M (b − a)
i=1 i=1 i=1 i=1

In other words

m(b − a) ≤ L(f, P ) ≤ U (f, P ) ≤ M (b − a).

Lemma 8.1 For partitions P and Q of [a, b], L(f, P ) ≤ U (f, Q)

Proof 8.1 We shall show that under refinement, lower sums increase (wide
sense) and upper sums decrease. This gives the conclusion since then
[ [
L(f, P ) ≤ L(f, P Q) ≤ U (f, P Q) ≤ U (f, Q)

Suppose a point u is adjoined to (xk−1 , xk ). The term mk (xk − xk−1 ) is re-

placed, in forming the lower sum for the refinement {x0 , · · · , xk−1 , u, xk , · · · , xn },
by inf {f (x) : xk−1 ≤ x ≤ u}(u−xk−1 )+inf {f (x) : u ≤ x ≤ xk }(xk −u). But mk
is less or equal to each of these infinima, since it is the infinimum over the larger
range [xk−1 , xk ] of x. Thus, this expression exceeds mk (u − xk−1 ) + mk (xk − u) =
mk (xk − xk−1 ), the term it replaces when u is adjoined. This shows that lower
sums increase under refinement and similarly upper sums decrease under refine-
ment. in other words L(f, P ) is increasing or non-decreasing, and U (f, P ) is
decreasing or non-increasing.

We are interested in the supremum of L(f, P ) and infinimum of U (f, P ) as P

varies.
Remark
In the intuitive, area case, lower sums are underestimates and upper sums are
overestimates. The area is defined precisely when there is no gap between the
two sets of estimates; i.e. when there is exactly one number that is approximated
by both L(f, P ) and U (f, P ).
Guided by this intuition; no ε gap for any ε > 0, we make the

Definition 8.2 The bounded function f is Riemann integrable on [a, b] if, for
every ε > 0, ∃ a partition P such that U (f, P ) − L(f, P ) < ε. Now, Let
Z b
f dx = supP ∈P L(f, P )
a
Integration 83

and Z b
f dx = infP ∈P U (f, P )
a
Z b Z b
Then f dx is called the Lower Riemann integral and f dx is called the
a a
upper Riemann integral.

Z b Z b
Theorem 8.1 f dx ≤ f dx and equality holds if and only if f is Riemann
a a
integrable on [a, b].
S
Proof 8.2 Let P1 and P2 be any two partitions of [a, b]. Let P = P1 P2 and so
P is a refinement of both P1 and P2 . So

L(f, P1 ) ≤ L(f, P ) ≤ U (f, P ) ≤ U (f, P2 ).

Thus, for any partition P2 of [a, b], U (f, P2 ) is an upper bound of {L(f, P1 ) : P1 ∈
P} where P is the set of all partitions of [a, b]. Hence supP1 ∈P L(f, P1 ) ≤ U (f, P2 ).
Z b Z b
Thus f dx ≤ U (f, P2 ), since f dx = supP1 ∈P L(f, P1 ).
a a
Z b
Since P2 is an arbitrary partition of [a, b] in P, this implies that f dx is a
a
lower bound of {U (f, P2 ) : P2 ∈ P}. Hence
Z b Z b
f dx ≤ infP2 ∈P U (f, P2 ) = f dx
a a

Z b Z b
Thus f dx ≤ f dx.♠
a a

Definition 8.3 A bounded function f defined on [a, b] is Riemann integrable on

Z b Z b
[a, b] if the lower and upper Riemann integrals are equal, i.e. f dx = f dx.
a a
Z b
The common value, denoted by f dx is called the Riemann integral of f over
a
[a, b].

Example 8.3 Show that f (x) = x2 is Riemann integrable on [0, 1].

1
Let Pn = {0, n1 , n2 , · · · , nn }, so ∆xi = for each i.
n
   2
2 i−1 i i−1
mi (f ) = inf {x ; x ∈ , = }
n n n
84 Integration

and   2

i−1 i
2 i
Mi (f ) = sup{x ; x ∈ , = }
n n n
and
n  2 n   
X i−1 1 1 X 2 1 1 1 1 1
L(f, Pn ) = · = 3 (i−1) = 3 · (n−1)(n)(2n−1) = 1− 2−
i=1
n n n i=1
n 6 6 n n
Z 1   
1 1 1 1
and f dx = lim 1− 2− = . Moreover,
0 n−→∞ 6 n n 3
n  2 n   
X i 1 1 X 2 1 1 1 1 1
U (f, Pn ) = · = 3 i = 3 · (n)(n+1)(2n+1) = 1+ 2+
i=1
n n n i=1
n 6 6 n n
Z 1    Z 1 Z 1
1 1 1 1 1
and f dx = lim 1+ 2+ = . Hence f dx = f dx =
0 n−→∞ 6 n n 3 0 0 3
and so f is Riemann integrable over [0, 1].

1, if x is rational,
Example 8.4 Let f (x) = Show that for any a < b,
0, if x is not rational
f is not Riemann integrable on [a, b].
Let Pn = {x0 , x1 , · · · , xn } be any partition of a, b. Any subinterval [xi−1 , xi ] con-
tains both rationals and irrationals. Thus mi (f ) = 0 and Mi (f ) = 1 ∀ i =
1, 2, · · · , n.
Xn Xn
L(f, Pn ) = mi (f )∆xi = 0 · (b − a) = 0
i=1 i=1

and n n
X X
U (f, Pn ) = Mi (f )∆xi = 1 · (b − a) = b − a
i=1 i=1
Z b Z b
Thus f dx = 0 and f dx = b − a and f is not Riemann integrable over
a a
[a, b].

8.2.1 Riemann Integrable Functions

Theorem 8.2 If f is continuous on [a, b], then f is integrable on [a, b].

Proof 8.3 f is continuous on [a, b] ⇒ f is uniformly continuous on [a, b] by an

ε
earlier theorem. Now, given ε > 0 ∃ δ > 0 s.t. |f (x)−f (x0 )| < whenever
b−a
|x − x0 | < δ ∀ x, x0 ∈ [a, b]. Let P be any partition such that ∆xi < δ for all
ε
i = 1, 2, · · · , n. Then Mi − mi < , ∀ i = 1, 2, · · · , n. This is so since f
b−a
Integration 85

attains its maximum and minimum on the interval [xi−1 , xi ] as f is continuous on

[xi−1 , xi ]. In other words there are points x0 , x00 ∈ [xi−1 , xi ] such that mi = f (x0 )
and Mi = f (x00 ). Thus
n
X
U (f, P ) − L(f, P ) = (Mi − mi )∆xi
i=1
n  
X ε
≤ ∆xi
i=1
b − a
n
ε X
= ∆xi
b − a i=1
ε
= · (b − a)
b−a
= ε

i.e. U (f, P )−L(f, P ) < ε which is condition for Riemann integrability. Therefore
f is Riemann integrable.

Theorem 8.3 If f is monotonic on [a, b], then f is integrable on [a, b].

Proof 8.4 Assume f is increasing and let ε > 0 be given. Let P be a partition
b−a
such that ∆xi = . Since f is monotonic increasing, we have Mi − mi =
n
f (xi ) − f (xi−1 ), so that
n
X
U (f, P ) − L(f, P ) = (Mi − mi )∆xi
i=1
n
X b−a
= (f (xi ) − f (xi−1 )) ·
i=1
n
n
b−aX
= (f (xi ) − f (xi−1 ))
n i=1
b−a
= (f (b) − f (a))
n
< ε

for n large enough, and f is Riemann integrable.

8.2.2 Properties of the Riemann integral

Theorem 8.4 If f and g are Riemann integrable on [a, b], then
Z b Z b Z b
1. f + g is Riemann integrable and (f + g)dx = f dx + gdx,
a a a
86 Integration

Z b Z b
2. αf is Riemann integrable for every real number α and (αf )dx = α f dx.
Z b Z b Z b a a

Moreover (αf + βg)dx = α f dx + β gdx for all real numbers α

a a a
and β. This says Riemann integration is a linear operation.

Theorem 8.5 A function f is Riemann integrable on [a, b] if and only if f is

Z b Z c
Riemann integrable on [a, c] and [c, b] for every c ∈ [a, b] and f dx = f dx +
Z b a a

f dx.
c

Z b
Theorem 8.6 If f is Riemann integrable on [a, b] and f ≥ 0, then f dx ≥ 0.
a

Corollary 8.1 If f and g are Riemann integrable on [a, b] and f (x) ≤ g(x) ∀ x ∈
Z b Z b
[a, b], then f (x)dx ≤ g(x)dx.
a a

Theorem 8.7 If f is Riemann integrable, then |f | is Riemann integrable and

Z b Z b
f dx ≤ |f |dx.
a a

Corollary 8.2 If f is Riemann integrable on [a, b] and m ≤ f (x) ≤ M on [a, b].

Z b
Then m(b − a) ≤ f dx ≤ M (b − a).
a

Theorem 8.8 Mean Value theorem for integrals Z b

If f is continuous on [a, b], then there is a point ξ ∈ (a, b) such that f dx =
a
f (ξ)(b − a)

Proof 8.5 Since f is continuous on [a, b], it is integrable on [a, b]. Moreover,
since [a, b] is closed and bounded, it means that f attains its maximum and min-
imum values M and m, respectively on [a, b]. In other words, there are points x0
and y0 in [a, b] such that m = f (x0 ) and M = f (y0 ). Thus
Z b
m(b − a) ≤ f dx ≤ M (b − a)
a

1

Rb
by the corollary above. Therefore m ≤ b−a a
f dx ≤ M . Since f is continuous
on [a, b], it attains all values between m and M on [a, b] and in particular the value
1

Rb 1

Rb
f dx. Hence there is a point ξ on [a, b] such that f (ξ) = f dx ⇒
Zb−a
b
a b−a a

f dx = f (ξ)(b − a). ♠
a
Integration 87

8.2.3 The fundamental theorem of integral calculus

The fundamental theorem of integral calculus gives a method for evaluating in-
tegrals.

Theorem Z x8.9 Let f be a bounded Riemann integrable function on [a, b] and let
F (x) = f (t)dt ∀ x ∈ [a, b]. Then F is continuous on [a, b].
a

Proof 8.6 Let M = sup{f (x) : x ∈ [a, b]} and let x0 be any point of [a, b]. Now,
let ε > 0 be given. Suppose x ∈ [a, b]. Then
Z x Z x0
|F (x) − F (x0 )| = | f (t)dt − f (t)dt|
a a
Z x
= | f (t)dt|
x0
Z x
≤ |f (t)|dt
x0
≤ M |x − x0 | < ε
ε
provided |x − x0 | < δ = and F is continuous at x0 .♠
M
Z x
Theorem 8.10 Let f be a continuous function on [a, b], and let F (x) = f (t)dt
a
for x ∈ (a, b). Then F is differentiable at x and F 0 (x) = f (x).

Proof 8.7 Let c be any point in (a, b). Then

Z c+h Z c 
F (c + h) − F (c) 1
lim = lim f (t)dt − f (t)dt
h−→0 h h−→0 h a a
Z c+h 
1
= lim f (t)dt
h−→0 h c
f (ξ)h
= lim , where c < ξ < c + h
h−→0 h
= lim f (ξ)
h−→0
= f (c), by cty of f.
Z c+h
Note that f (t)dt = hf (ξ) by use of Mean Value theorem for integrals. Thus,
c
F is differentiable and F 0 (c) = f (c) for all c ∈ (a, b).

Theorem 8.11 The Fundamental Theorem for Integral Calculus

Let f be a continuous function on [a, b] and let F be any function such that
F 0 (x) = f (x), i.e. F is an indefinite integral or antiderivative of f . Then
Z b
F (b) − F (a) = f (t)dt.
a
88 Integration

Z x
Proof 8.8 Let F be any function whose derivative is f . Then F (x) = f (t)dt+
a
C, C a constant. This is so since all indefinite integrals
Z x of f differ by a
constant and we know from the previous theorem that f (t)dt is an indef-
Z a a

inite integral of f (x). Now, F (a) = f (t)dt + C = 0 + C = C. Thus

Z x a Z b
C = F (a). Hence F (x) = f (t)dt + F (a) and F (b) = f (t)dt + F (a)
Z b a a

or F (b) − F (a) = f (t)dt ♠

a

8.3 Multiple integrals

In this section, we extend the idea of a definite integral to double and triple
integrals of functions of two or three variables. These ideas can then be used to
compute volumes, surface areas, masses, and centroids of some regions.
It is difficult to evaluate single integrals directly from the definition of an integral.
The Fundamental Theorem of Calculus provides a much easier method that is
used in everyday practice. The evaluation of double integrals from first principles
is even more difficult and we do not even attempt to discuss it in this module.
Rather, we will use the technique of expressing a double integral as an iterated
integral, which can then be evaluated by calculating two single integrals.

8.3.1 The Double integral

Let f be a function
Z d of two variables that is integrable on the rectangle R =
[a, b] × [c, d]. f (x, y)dy in this case means that f (x, y) is integrated with
c
respect to y from y = c to y = d while x is held constant. One Z d can think of
this as partial integration with respect to y. The result is that f (x, y)dy is a
c
number that depends on the value of x, as such it defines a function of x. Hence
we write
Z d
A(x) = f (x, y)dy
c

If we now integrate the function A with respect to x from x = a to x = b, we get

Z b Z b Z d 
A(x)dx = f (x, y)dy dx
a a c
Integration 89

The integral on the right side is called an iterated integral. In this way, we
have the two iterated integrals:
Z bZ d Z b Z d 
f (x, y)dydx = f (x, y)dy dx
a c a c

and Z d Z b Z d Z b 
f (x, y)dxdy = f (x, y)dx dy
c a c a
where the former means that we first integrate with respect to y from c to d and
then with respect to x from a to b and the later means that we first integrate
with respect to x from a to b and then with respect to y from c to d. The two
iterated integrals are always equal.

Theorem 8.12 Theorem:(Fubini)

If f is continuous on the rectangle R = {(x, y)|a ≤ x ≤ b, c ≤ y ≤ d}, then
Z Z Z bZ d Z dZ b
f (x, y)dA = f (x, y)dydx = f (x, y)dxdy
R a c c a

This theorem gives a practical method for evaluating a double integral, as ex-
plained earlier, by expressing it as an iterated integral.
Now, if f is positive, then the double integral
Z Z
f (x, y)dA
R

can be interpreted as the volume V of the solid S that lies above R and under
the surface Z = f (x, y).

8.3.2 Triple Integrals

The above results are easily generalised to closed regions in three dimensions. If
we construct a grid consisting of planes parallel to the xy, yz and xz planes, the
region R is subdivided into subregions which are rectangularZ and
Z Zparallelepipeds.
In such case we can express the triple integral over R by f (x, y, z)dV ,
R
and this can be expressed as an iterated integral of the form
Z Z Z Z b Z g2 (x) Z f2 (x,y)
f (x, y, z)dV = f (x, y, z)dxdydz
R x=a y=g1 (x) z=f1 (x,y)
"Z (Z ) #
Z b g2 (x) f2 (x,y)
= f (x, y, z)dz dy dx
x=a y=g1 (x) z=f1 (x,y)

where the innermost integral is to be evaluated first. The integration can also be
performed in any other order to give an equivalent result. This can be extended
to higher dimensions.
90 Integration

Example 8.5 Sketch the regionZ RZ in the xy plane bounded by y = x2 , x = 2, y =

1. Evaluate the double integral (x2 + y 2 )dxdy
R
The double integral can be expressed as the iterated integral
Z 2 Z x2 Z 2 (Z x2 )
2
y 3 x2
Z
2 2 2 2
(x + y )dydx = (x + y )dy dx = [x2 y + ] dx
x=1 y=1 x=1 y=1 x=1 3 y=1
2
x6
Z  
4 2 1
= x + −x − dx
x=1 3 3
 5 2
x x7 x3 1
= + − − x
5 21 3 3 1
1006
=
105
The double integral can also be expressed as an iterated integral
Z Z 4 2 Z (Z 4
)
2

√
(x2 + y 2 )dxdy = √
(x2 + y 2 )dx dy
y=1 x= y y=1 x= y
4 2
x3
Z 
= + xy 2 dy
y=1 3 √
x= y
3
!
Z 4
8 y2 5
= + 2y 2 − − y2 dy
y=1 3 3
 4
8 2 2 5 2 7
= x + y3 − y 2 − y 2
3 3 15 7 1
1006
=
105

Example 8.6 Find the volume of the region R bounded by the parabolic cylinder
z = 4 − x2 and the planes x = 0, y = 0, y = 6, z = 0.
Z Z Z Z 2 Z 6 Z 4−x2
dxdydz = dzdydx
R x=0 y=0 z=0
Z 2 Z 6
= (4 − x2 )dydx
x=0 y=0
Z 2
= (4 − x2 )y|6y=0 dx
Zx=0
2
= (24 − 6x2 )dx
x=0
2
= 24x − 2x3 0


= 32
Integration 91

Activity 8.1 1. Calculate the following iterated integrals:

Z 3Z 1
(a) (1 + 4xy)dxdy
1 0
Z 2 Z π
(b) x sin ydydx
0 0
1 2
xex
Z Z
(c) dydx
0 1 y
Z 4Z 2 
x y
(d) + dydx
1 1 y x
Z 2Z π
(e) r sin2 θdθdr
0 0

2. Calculate the double integral.

Z Z
(a) (6x2 y 3 − 5y 4 )dA, R = {(x, y)|0 ≤ x ≤ 3, 0 ≤ y ≤ 1}
Z ZR
π
(b) cos(x + 2y)dA, R = {(x, y)|0 ≤ x ≤ π, 0 ≤ y ≤ }
R 2
Z Z 2
xy
(c) 2
dA, R = {(x, y)|0 ≤ x ≤ 1, − 3 ≤ y ≤ 3}
R x +1
Z Z
x
(d) dA, R = [0, 1] × [0, 1]}
R 1 + xy

3. Evaluate the iterated integral.

Z 4 Z √y
(a) xy 2 dxdy
0 0
Z 2 Z 2y
(b) xydxdy
0 y
Z 1 Z 2
(c) (x − y)dydx
0 2x

4. Find the volume of the solid.

(a) Under the plane x + 2y − z = 0 and above the region bounded by y = x

and y = x4 .
(b) Under the surface z = 2x + y 2 and above the region bounded by x = y 2
and x = y 3
(c) Under the surface z = xy and above the triangle with vertices (1, 1), (4, 1)
and (1, 2)
(d) Enclosed by the paraboloid z = x2 + 3y 2 and the planes x = 0, y =
1, y = x, , z = 0
92 Integration

(e) Bounded by the coordinate planes and the plane 3x + 2y + z = 6

5. Evaluate the integral by reversing the order of integration.

Z 1Z 3
2
(a) ex dxdy
0 3y
√ √
Z π Z π
(b) cos x2 dxdy
0 y
Z 4 Z 2
1
(c) √
dydx
0 x y3 + 1
π
Z 1Z 2 √
(d) cos x 1 + cos2 xdxdy
0 sin−1 y

6. Evaluate the double integral.

Z Z
(a) y 2 dA, D = {(x, y)| − 1 ≤ y ≤ 1, − y − 2 ≤ x ≤ y}
D
Z Z
y
(b) 5
dA, D = {(x, y)|0 ≤ x ≤ 1, 0 ≤ y ≤ x2 }
D x +1
Z Z
(c) x3 dA, D = {(x, y)|1 ≤ x ≤ e, 0 ≤ y ≤ ln x}
D
Z 1 Z 1 Z 2
7. Evaluate √
x=0 y=0 z= x2 +y 2

8.3.3 Transformations of multiple integrals

In evaluating a multiple integral over a region R, it is often convenient to use
coordinates other than rectangular, such as the curvilinear coordinates. If we
let (u, v) be curvilinear coordinates of points in a plane, there will be a set of
transformation equations x = f (u, v), y = g(u, v) mapping points (x, y) of the xy
plane into points (u, v) of the uv plane. In such case the region R of the xy plane
is mapped into a region R0 of the uv plane. We then have
Z Z Z Z
∂(x, y)
F (x, y)dxdy = G(u, v) dudv
R R0 ∂(u, v)

∂x ∂x
∂(x,y)
where G(u, v) = F {f (u, v), g(u, v)} and ∂(u,v) ≡ ∂u ∂v
∂y ∂y is the Jacobian
∂u ∂v
of x and y with respect to u and v. Similarly if (u, v, w) are curvilinear co-
ordinates in three dimensions, there will be a set of transformation equations
Integration 93

x = f (u, v, w), y = g(u, v, w), z = h(u, v, w) and we can write

Z Z Z Z Z Z
∂(x, y, z)
F (x, y, z)dxdydz = G(u, v, w) dudv
R R0 ∂(u, v, w)
∂x ∂x ∂x
∂u ∂v ∂w
∂(x,y,z) ∂y ∂y ∂y
where G(u, v, w) = F {f (u, v, w), g(u, v, w), h(u, v, w)} and ∂(u,v,w)
≡
∂u ∂v ∂w
∂z ∂z ∂z
∂u ∂v ∂w
is the Jacobian of x, y and z with respect to u, v and w.
Double Integrals in Polar Coordinates
If f is continuous on a polar rectangle R given by 0 ≤ a ≤ r ≤ b, α ≤ θ ≤ β,
where 0 ≤ β − α ≤ 2π, then
Z Z Z βZ b
f (x, y)dA = f (r cos θ, r sin θ)rdrdθ
R α a

where we have used the transformation x = r cos θ and y = r sin θ. r in the

formula is called the Jacobian of the transformation calculated as
∂x ∂x
∂r ∂θ cos θ −r sin θ
J= ∂y ∂y = = r cos2 θ + r sin2 θ = r
sin θ r cos θ
∂r dθ
Z Z p
Example 8.7 Evaluate x2 + y 2 dxdy, where R is the region in the xy
R
plane bounded by x2 + y 2 = 4 and x2 + y 2 = 9
The presence of x2 + y 2 suggests the use of polar coordinates. It follows that
Z Z p Z 2π Z 3
2 2
x + y dxdy = r2 drdθ
R
Zθ=0
2π 3
r=2
r 3
= |2 dθ
θ=0 3
Z 2π
19
= dθ
θ=0 3
38π
=
3
Z Z Z
Example 8.8 Express F (x, y, z)dxdydz in cylindrical coordinates.
R
The transformations equations in cylindrical coordinates are x = r cos θ, y =
r sin θ, z = z. The Jacobian of the transformation is thus
cos θ −r sin θ 0
∂(x, y, z)
= sin θ r cos θ 0 =r
∂(r, θ, z)
0 0 1
94 Integration

Z Z Z Z Z Z
Thus F (x, y, z)dxdydz = G(r, θ, z)rdrdθdz where G(r, θ, z) =
R R0
F (r cos θ, r sin θ, z)

Activity 8.2 1. Evaluate the given integral by changing to polar coordinates.

Z Z
(a) xydA where D is the disk with centre the origin and radius 3.
D
√
Z 3 Z 9−x2
(b) sin(x2 + y 2 )dydx
−3 0
Z aZ 0
(c) √ x2 ydxdy
0 − a2 −y 2
Z 1 Z √2−y2
(d) (x + y)dxdy
0 y

2. Use polar coordinates to find the volume of the given solid.

p
(a) Under the cone z = x2 + y 2 and above the disk x2 + y 2 ≤ 4
(b) Enclosed by the hyperboloid −x2 − y 2 + z 2 = 1 and the plane z = 2.

3. Use a double integral to find the area of one loop of the rose r = cos 3θ.
Z Z p
4. Evaluate x2 + y 2 dxdy where R is the region x2 + y 2 ≤ a2 .
R
Z Z
2 +y 2 )
5. Evaluate e−(x dxdy where R is the region x2 + y 2 ≤ a2 .
R
Z 1 Z 1−z y
6. By using the transformation x+y = u, y = uv, show that e x+y dydx =
x=0 y=0
e−1
2

8.4 Line Integrals

Let C be a curve in the xy plane which connects points A(a1 , b1 ) and B(a2 , b2 ).
Let P (x, y) and Q(x, y) be single-valued functions defined at all points of C. Sub-
divide C into n parts by choosing (n−1) points on it given by (x1 , y1 ), (x2 , y2 ), · · · , (xn−1 , yn−1 ).
Let ∆xk = xk −xk−1 , ∆yk = yk −yk−1 , k = 1, 2, 3, · · ·. (a1 , b1 ) ≡ (x0 , y0 ), (a2 , b2 ) =
(xn , yn ). Suppose that points (ξk , ηk ) are chosen so that they are situated on C
between points (xk−1 , yk−1 ) and (xk , yk ). Form the sum
n
X
{P (ξk , ηk )∆xk + Q(ξk , ηk )∆yk }
k=1
Integration 95

n
X
Definition 8.4 The limit lim {P (ξk , ηk )∆xk + Q(ξk , ηk )∆yk }, if it exists,
n−→∞
k=1 Z
is called a line integral along C and is denoted by [P (x, y)dx + Q(x, y)dy] or
Z (a2 ,b2 ) C

[P dx + Qdy].
(a1 ,b1 )

The limit in the definition above will exist if the functions P and Q are continuous
(or piecewise continuous) at all points of C. The definition above can be extended
to the definition of a line integral along a curve C in three dimensional space:
Xn Z
lim {A1 (ξk , ηk , ζk )∆xk +A2 (ξk , ηk , ζk )∆yk +A3 (ξk , ηk , ζk )∆zk } = [A1 dx+A2 dy+A3 dz]
n−→∞ C
k=1

where Ai = Ai (x, y, z), i = 1, 2, 3. This line integral can be expressed in vector

notation (for brevity):
Z Z Z
[A1 dx + A2 dy + A3 dz] = (A1 i + A2 j + A3 k) · (dxi + dyj + dzk) = A · dr
C C C

where A = A1 i + A2 j + A3 k and r = dxi + dyj + dzk. If we set z = 0, we go back

to the two dimensional case. If a force filed F is defined, then
Z
F · dr
C

represents physically the total work done in moving the object along the curve
C.
How do we evaluate line integrals?
If the equation of a curve C in the plane z = 0 is given as y = f (x), the line
integral in definition 8.4 is evaluated byZ placing y = f (x), dy = f 0 (x)dx in the
a2
integrand to obtain the definite integral [P {x, f (x)}dx + Q{x, f (x)}f 0 (x)dx].
a1
Similarly if C is given as x = g(y), then dx = g 0 (y)dy and the line integral
Z b2
becomes [P {g(y), y}g 0 (y)dy + Q{g(y), y}dy]. If C is given in parametric form
b1
x = φ(t), y = ψ(t), the line integral becomes
Z t2
[P {φ(t), ψ(t)}φ0 (t)dt + Q{φ(t)ψ(t)}ψ 0 (t)dt]
t1

where t1 , t2 denote the values of t corresponding to points A and B respectively.

Definition 8.5 A curve in the xy plane is defined by the parametric equations

x = φ(t), y = ψ(t) where φ and ψ are single-valued and continuous in an interval
t1 ≤ t ≤ t2 . If φ(t1 ) = φ(t2 ) and ψ(t1 ) = ψ(t2 ), the curve is said to be closed. A
simple closed curve is a closed curve which does not intersect itself anywhere.
96 Integration

Definition 8.6 If a plane region has the property that any closed curve in it can
be continuously shrunk to a point without leaving the region, then the region is
called simply-connected, otherwise it is called multiply-connected.

Theorem 8.13 Green’s Theorem in the plane

∂P ∂Q
Let P, Q, , be single valued and continuous in a simply-connected region
∂y ∂x
R bounded by a simple closed curve C. Then
I Z Z  
∂Q ∂P
[P dx + Qdy] = − dxdy
C R ∂x ∂y
I
where is used to emphasize that C is closed and that it is described in the
C
positive direction (anticlockwise direction).

Proof 8.9 Refer to the figure. Let the equations of the curves AEB and AF B
be y = Y1 (x) and y = Y2 (x) respectively. If R is the region bounded by C, we
have
Z Z Z b "Z Y2 (x) #
∂P ∂P
dxdy = dy dx
R ∂y x=a y=Y1 (x) ∂y
Z b Z b
Y2 (x)
= P (x, y)|y=Y1 (x) dx = [P (x, Y2 (x)) − P (x, Y1 (x))]dx
x=a a
Z b Z a I
= − P (x, Y1 (x))dx − P (x, Y2 (x))dx = − P dx
a b C

Then I Z Z
∂P
P dx = − dxdy.
C R ∂y
Similarly let the equations of curves EAF and EBF be x = X1 (y) and x = X2 (y)
respectively. Then
Z Z Z f "Z X2 (y) #
∂Q ∂Q
dxdy = dx dy
R ∂x y=e x=X1 (y) ∂x
Z f Z f
X2 (y)
= Q(x, y)|x=X1 (y) dx = [Q(X2 (y), y) − Q(X1 (y), y)]dy
y=e e
Z e Z f I
= Q(X1 (y), y)dy + Q(X2 (y), y)dy = Qdy
f e C

Then I Z Z
∂Q
Qdy = dxdy.
C R ∂x
Adding the two line integrals gives
I Z Z  
∂Q ∂P
[P dx + Qdy] = − dxdy
C R ∂x ∂y
Integration 97

Z (1,2)
Example 8.9 Evaluate [(x2 − y)dx + (y 2 + x)dy] along (a). a straight line
(0,1)
from (0, 1) to (1, 2), (b). straight lines from (0, 1) to (1, 1) and then from (1, 1)
to (1, 2), (c). the parabola x = t, y = t2 + 1.
(a). An equation for the line joining (0, 1) and (1, 2) in the xy plane isZy = x + 1.
1
In this way, dy = dx and the line integral becomes, after substituting, [(x2 −
1 x=0
Z 1 
2 2 5
(x + 1))dx + ((x + 1)2 + x)dx] = (2x2 + 2x)dx = x3 + x2 = + 1 =
0 3 0 3 3
(b). Along the straight line from (0, 1) to (1, 1), y = 1, dy = 0 and the line integral
Z 1  1
2 1 3 1 2
equals (x − 1)dx = x − x = − 1 = − .
x=0 3 0 3 3
Along the straight line from (1, 1) to (1, 2), x = 1, dx = 0 and the line integral
Z 2  2
2 1 3 14 4 10
equals (y + 1)dy = y + y = − = .
y 3 1 3 3 3
Z 1
(c). Since t = 0 at (0, 1) and t = 1 at (1, 2), the line integral equals [(t2 −
Z 1 t=0
2 2 2 5 3 2
(t + 1))dt + ((t + 1) + t)2tdt] = (2t + 4t + 2t + 2t − 1)dt = 2.
0
Z
2 2
Activity 8.3 If A = (3x − 6yz)i + (2y + 3xz)j + (1 − 4xyz )k, evaluate A · dr
C
from (0, 0, 0) to (1, 1, 1) along the following paths C:

1. x = t, y = t2 , z = t3 .
2. the straight lines from (0, 0, 0) to (0, 0, 1) then to (0, 1, 1), and then to
(1, 1, 1).
3. the straight line joining (0, 0, 0) and (1, 1, 1).
I
Example 8.10 Verify Green’s theorem in the plane for [(2xy − x2 )dx + (x +
C
y 2 )dy] where C is the closed curve of the region bounded by y = x2 and y 2 = x.
The plane curves y = x2 and y 2Z = x intersect at (0, 0) and (1, 1). Along
1
y = x2 , the line integral equals [(2x(x2 ) − x2 )dx + (x + (x2 )2 )d(x2 )] =
Z 1 x=0 Z 0
3 2 5 7 2
(2x + x + 2x )dx = . Along y = x, the line integral equals [(2(y 2 )y −
0 6 y=1
Z 0
17
(y 2 )2 )d(y 2 ) + (y 2 + y 2 )dy] = (4y 4 − 2y 5 + 2y 2 )dy = − The required line
1 15
integral is thus 76 − 17 15
1
= 30
Z Z   Z Z  
∂Q ∂P ∂ 2 ∂ 2
− dxdy = (x + y ) − (2xy − x ) dxdy
R ∂x ∂y R ∂x ∂y
98 Integration

√
Z Z Z 1 Z x
= (1 − 2x)dxdy = (1 − 2x)dydx
R x=0 y=x2
Z 1 √
Z 1 √
x 3 1
= (y − 2xy)|y=x2 dx = ( x − 2x 2 − x2 + 2x3 )dx =
x=0 0 30
Hence Green’s theorem is verified.
Z
Theorem 8.14 A necessary and sufficient condition for [P dx + Qdy] to be
C
independent of the path C joining any tow given points in a region R is that in
R
∂P ∂Q
=
∂y ∂x
where it is supposed that these partial derivatives are continuous in R.

The condition in the theorem above is also the condition that P dx + Qdy is an
exact differential, i.e. that there exists a function φ(x, y) such that P dx + Qdy =
dφ. In this way if the end points of curve C are (x1 , y1 ) and (x2 , y2 ), the value of
the line integral is given by
Z (x2 ,y2 ) Z (x2 ,y2 )
[P dx + Qdy] = dφ = φ(x2 , y2 ) − φ(x1 , y1 )
(x1 ,y1 ) (x1 ,y1 )

Further, if C is closed, we have x1 = x2 , y1 = y2 and

I
[P dx + Qdy] = 0.
C

Activity 8.4 Let P (x, y) and Q(x, y) be continuous and have continuous first
partial derivatives at each point of a simply connected region R. Prove that a
necessary and sufficient condition that
I
[P dx + Qdy] = 0
C

around every closed path C in R is that

∂P ∂Q
=
∂y ∂x
identically in R.

Proof 8.10 Sufficiency

∂P ∂Q
suppose = , then Green’s theorem,
∂y ∂x
I Z Z  
∂Q ∂P
[P dx + Qdy] = − dxdy = 0
C R ∂x ∂y
Integration 99

where R is the region bounded by C.

NecessityI
∂P ∂Q
Suppose [P dx + Qdy] = 0 around every closed path C in R and that 6 =
C ∂y ∂x
∂P ∂Q
at some point of R. In particular suppose − > 0 at the point (x0 , y0 ). By
∂y ∂x
∂P ∂Q
hypothesis and are continuous in R, so that there must be ome region
∂y ∂x
∂P ∂Q
R0 containing (x0 , y0 ) as an interior point for which − > 0. If Γ is the
∂y ∂x
boundary of R0 , then by Green’s theorem
I Z Z  
∂Q ∂P
[P dx + Qdy] = − dxdy > 0
Γ R0 ∂x ∂y
I
contradicting the hypothesis that [P dx + Qdy] = 0 for all closed curves in R.
∂P ∂Q ∂P ∂Q
Thus − cannot be positive. Similary we can show that − cannot be
∂y ∂x ∂y ∂x
∂P ∂Q
negative, and it follows that it must be identically zero, i.e. = identically
∂y ∂x
in R.
Z (3,4)
Example 8.11 (a). Prove that [(6xy 2 − y 3 )dx + (6x2 y − 3xy 2 )dy] is inde-
(1,2)
pendent of the path joining (1, 2) and (3, 4)
(b). Evaluate the integral in (a).
∂P ∂Q
(a). P = 6xy 2 − y 3 , Q = 6x2 y − 3xy 2 . Then = 12xy − 3y 2 = , and by
∂y ∂x
theorem, the line integral is independent of the path.
∂P ∂Q ∂φ ∂φ
(b). Since = , we must have = 6xy 2 − y 3 and = 6x2 y − 3xy 2 . In
∂y ∂x ∂x ∂y
this way we have φ = 3x2 y 2 −xy 3 +f (y) and φ = φ3x2 y 2 −xy 3 +g(x). The only way
these can be equal is to set f (y) = g(x) = c, a constant. Hence φ = 3x2 y 2 −xy 3 +c,
Z (3,4) Z (3,4)
2 3 2 2
and thus [(6xy − y )dx + (6x y − 3xy )dy] = d(3x2 y 2 − xy 3 + c) =
(1,2) (1,2)
(3,4)
3x2 y 2 − xy 3 + c|(1,2) = 236. The arbitrary constant c is omitted in this evaluation.

Z (4,2)
Activity 8.5 1. Evaluate [(x + y)dx + (y − x)dy] along (a). the parabola
(1,1)
y 2 = x, (b). a straight line, (c). straight lines from (1, 1) to (1, 2) and then
to (4, 2), (d). the curve x = 2t2 + t + 1, y = t2 + 1.
Z
2 2
2. If F = (x − y )i + 2xyj, evaluate F · dr along the curve C in the xy
C
plane given by y = x2 − x from the point (1, 0) to (2, 2).
100 Integration

I
3. Verify Green’s theorem in the plane for [(x2 − xy 3 )dx + (y 2 − 2xy)dy]
C
where C is a square with vertices at (0, 0), (2, 0), (2, 2), (0, 2).
Z (2,1)
4. (a). Prove that [(2xy − y 4 + 3)dx + (x2 − 4xy 3 )dy] is independent of
(1,0)
the path joining (1, 0) and (2, 1). (b). Evaluate the integral in (a).
I
5. (a) Evaluate F · dr where F = yi + (x + z)2 j + (x − z)2 k from (0, 0, 0)
C
to (2, 4, 0) along the straight line y = 2x and along the parabola y =
x2 , z = 0.
(b) Show that the area A bounded by a simple closed curve C is given by
I
A= xdy − ydx.
C

Hence determine the area of a planar region R bounded by an ellipse

C whose major and minor axes are 2a and 2b respectively.

8.5 Surface integrals

8.5.1 Gradient, Divergence and Curl
Consider the vector operator ∇ (del) defined by
∂ ∂ ∂
∇=i +j +k
∂x ∂y ∂z
Then if φ(x, y, z) and A(x, y, z) have continuous first partial derivatives ina re-
gion, we can define the following:

Definition 8.7 The gradient of φ is defined by

 
∂ ∂ ∂
grad φ = ∇φ = i +j +k φ
∂x ∂y ∂z
∂φ ∂φ ∂φ ∂φ ∂φ ∂φ
= i +j +k = i+ j+ k
∂x ∂y ∂z ∂x ∂y ∂z
Remark
If φ(x, y, z) = c is the equation of a surface, then ∇φ is a normal to this surface.

Definition 8.8 The divergence of A is defined by

 
∂ ∂ ∂
div A = ∇ · A = i +j +k · (A1 i + A2 j + A3 k)
∂x ∂y ∂z
∂A1 ∂A2 ∂A3
= + +
∂x ∂y ∂z
Integration 101

Definition 8.9 The curl of A is defined by

 
∂ ∂ ∂
curl A = ∇ × A = i +j +k × (A1 i + A2 j + A3 k)
∂x ∂y ∂z
i j k
∂ ∂ ∂
=
∂x ∂y ∂z
A1 A2 A3
∂ ∂ ∂ ∂ ∂ ∂
= i ∂y ∂z − j ∂x ∂z + k ∂x ∂y
A2 A3 A1 A3 A1 A2
     
∂A3 ∂A2 ∂A1 ∂A3 ∂A2 ∂A1
= − i+ − j+ − k
∂y ∂z ∂z ∂x ∂x ∂y

Activity 8.6 If φ = x2 yz 3 and A = xzi − y 2 j + 2x2 yk, find (a). ∇φ, (b). ∇ · A,
(c). ∇ × A, (d). div(φA), (e). curl(φA)

Let S be a two-sided surface having projection R on the xy plane (see figure).

Assume that an equation for S is z = f (x, y), where f is single-valued and
continuous for all x and y in R. Divide R into n subregions of area ∆Ap , p =
1, 2, · · · , n, and erect a vertical column on each of these subregions to intersect
S in an area ∆Sp . Let φ(x, y, z) be single-valued and continuous at all points of
Xn
S. Form the sum φ(ξp , ηp , ζp )∆Sp where (ξp , ηp , ζp ) is some point of ∆Sp . If
p=1
the limit of this sum as n −→ ∞ in such a way that each ∆Sp −→ 0 exists, the
resulting limit is called the surface integral of φ(x, y, z) over S and is designated
by Z Z
φ(x, y, z)dS
S

If we assume that z = f (x, y) has continuous (or r

sectionally continuous) deriva-
 2
∂z 2 ∂z


tives in R, then it can be shown that dS = 1 + ∂x + ∂y dA where
dA = dxdy. In this way
s  2  2
Z Z Z Z
∂z ∂z
φ(x, y, z)dS = φ(x, y, z) 1 + + dxdy
S R ∂x ∂y

Theorem 8.15 The Divergence Theorem

Let S be a closed surface bounding a region of volume V . Choose the outward
drawn normal to the surface as the positive normal and assume that α, β, γ are
the angles which this normal makes with the positive x, y and z axes respectively.
Then if A1 , A2 and A3 are continuous and have continuous partial derivatives in
102 Integration

the region, we have

Z Z Z   Z Z
∂A1 ∂A2 ∂A3
+ + dV = (A1 cos α + A2 cos β + A3 cos γ)dS
V ∂x ∂y ∂z S
Z Z
= [A1 dydz + A2 dzdx + A3 dxdy]
S
In vector form with A = A1 i + A2 j + A3 k and n = cos αi + cos βj + cos γk, these
can be simply written as
Z Z Z Z Z
∇ · AdV = A · ndS
V S
In other words, the divergence theorem states that the surface integral of the
normal component of a vector A taken over a closed surface is equal to the integral
of the divergence of A taken over the volume enclosed by the surface.
Example 8.12 Evaluate the integrals
Z Z Z
1. ∇ · FdV
V
Z Z Z
2. ∇ × FdV
V

where F = (x2 − z)i + 2xyj − 2xk and V is the closed region bounded by the planes
x = 0, y = 0, z = 0, 2x + 2y + z = 6
Solution
∇ · F = 2x + 2x + 0 = 4x. Thus
Z Z Z Z 3 Z 3−x Z 6−2x−2y
∇ · FdV = 4xdzdydx
V x=0 y=0 z=0
Z 3 Z 3−x
= 4x(6 − 2x − 2y)dydx
x=0 y=0
Z 3
3−x
24xy − 8x2 y − 4xy 2

= 0
dx
Zx=0
3
24x(3 − x) − 8x2 (3 − x) − 4x(3 − x)2 dx
 
=
Zx=0
3 3
(36x − 24x2 + 4x3 )dx = 18x2 − 8x3 + x4 0 = 27

=
0

i j k
∂ ∂ ∂
∇×F =
∂x ∂y ∂z
x2 − z 2xy −2x
∂(x2 − z) ∂(−2x) ∂(2xy) ∂(x2 − z)
     
∂(−2x) ∂(2xy)
= − i+ − j+ − k
∂y ∂z ∂z ∂x ∂x ∂y
= (0 − 0)i + (−1 + 2)j + (2y − 0)k = j + 2yk
Integration 103

and therefore
Z Z Z Z 3 Z 3−x Z 6−2x−2y Z 3 Z 3−x Z 6−2x−2y
∇×FdV = j dzdydx +k 2ydzdydx
V x=0 y=0 z=0
| {z } | x=0 y=0 z=0
{z }
I1 I2

Z 3 Z 3−x Z 3 3−x
6y − 2xy − y 2 0 dx

I1 = j (6 − 2x − 2y)dydx = j
x=0 y=0 0
3 3
x3
Z 
2 2
= j (9 − 6x + x )dx = j 9x − 3x +
0 3 0
= j(27 − 27 + 9) = 9j
Z 3 Z Z 3 Z 3−x
3−x
I2 = k y = 0 2y(6 − 2x − 2y)dydx = k (12y − 4xy − 4y 2 )dydx
x=0 0 0
Z 3 3−x Z 3
4 2
= k 6y 2 − 2xy − y 3 dx = k (18 − 18x + 6x2 − x3 )dx
0 3 0 0 3
 3
2 81 27
= k 18x − 9x2 + 2x3 − x4 = (54 − 81 + 54 − )k = k
12 0 6 2
Z Z Z
27
Hence ∇ × FdV = 9j + k
V 2

Z Z
Activity 8.7 State the divergence theorem. Hence evaluate r · ndS where
S
S is a closed surface.

Theorem 8.16 Stokes’s theorem

If S is a smooth oriented surface with piecewise smooth oriented boundary C
and if A1 , A2 , A3 are single-valued, continuous, and have continuous first partial
derivatives in a region of space including S, we have

Z  Z Z 
∂A3 ∂A2
[A1 dx + A2 dy + A3 dz] = − cos αdS
C S ∂y ∂z
Z Z   Z Z  
∂A1 ∂A3 ∂A2 ∂A1
+ − cos βdS + − cos γdS
S ∂z ∂x S ∂x ∂y

In vector form with A = A1 i + A2 j + A3 k and n = cos αi + cos βj + cos γk, this

is simply written as
Z Z Z
A · dr = (∇ × A) · ndS
C S
104 Integration

Stoke’s theorem states that the line integral of the tangential component of a
vector A taken around a simple closed curve C is equal to the surface integral of
the normal component of the curl of A taken over any surface S having C as a
boundary. The orientation of C is determined from the orientation of S according
to the right hand rule.

Example 8.13 Show that Green’s theorem is a special case of Stoke’s theorem.
Let A = P i + Qj be a vector point function which is continuously differentiable
in a simply connected region S in the xy-plane whose boundary C is a piecewise
smooth simple closed curve. Now,

j i k
∂ ∂ ∂
∇×A =
∂x
∂y ∂z
P
Q 0
     
∂Q ∂P ∂Q ∂P
= − i+ j+ − k
∂z ∂z ∂x ∂y
∂Q ∂P
and n · (∇ × A) = k · (∇ × A) = − since the unit normal vector to the
∂x ∂y
xy-plane is k. Also

A · dr = (P i + Qj) · (dxi + dyj) = P dx + Qdy

Therefore Stoke’s theorem takes the form

I Z Z   Z Z  
∂Q ∂P ∂Q ∂P
(P dx + Qdy) = − dS = − dxdy
C S ∂x ∂y S ∂x ∂y
which is Green’s theorem for the plane and is a special case of Stoke’s theorem.

Example 8.14 Verify Stoke’s theorem for F = (2x − y)i − yz 2 j − y 2 zk, where S
is the upper half of the sphere x2 + y 2 + z 2 = 1 and C is its boundary.
The boundary C of S is a circle in the xy plane of radius 1 and centre at the
origin. Let x = cos t, y = sin t, z = 0, 0 ≤ t ≤ 2π be parametric equations of C.
Then
I I
F · dr = [(2x − y)dx − yz 2 dy − y 2 zdz]
C
ZC2π
= (2 cos t − sin t)(− sin t)dt = π
0

Also,
i j k
∂ ∂ ∂
∇×F= =k
∂x ∂y ∂z
2x − y −yz 2 −y 2 z
Integration 105

Then Z Z Z Z Z Z
(∇ × F ) · ndS = k · ndS = dxdy
S S R
since k · ndS = dxdy and R is the projection of S on the xy plane. In this way,
√ √
Z Z Z 1 Z 1−x2 Z 1 Z 1−x2
dxdy = √
dydx = 4
R x=−1 y=− 1−x2 0 0
Z 1√
= 4 1 − x2 dx =π
0

and Stoke’s theorem is verified.

Unit 9
Ordinary differential equations

9.1 What this Unit is all About

In this unit, we discuss first order separable equations as well as first order lin-
ear equations.We also discuss equations reducible to variable separable type and
homogeneous ordinary differential equations. We conclude the unit by discussing
second order linear constant coefficient homogeneous ordinary differential equa-
tions. We will be using the expressions ordinary differential equation, differential
equations, equations and ODEs interchangeably.

By the end of this unit, you should be able to:

1. explain what a solution of a differential equation is.

2. classify differential equations according to type, order and linearity.
3. solve separable variable first order ODEs.
4. solve first order linear ODEs.
5. transform to variable separable form.
6. determine whether an ODE is homogeneous or not.
7. solve homogeneous ODEs
8. solve second order linear constant coefficient homogeneous ordinary differential equations

Introduction

You have already acquired knowledge of various methods of integration. You are
urged to review these methods and get ready to use them to solve simple ordinary
differential equations.
In this unit, we introduce the standard form of a first order differential equation:
P (x, y)dx + Q(x, y)dy = 0 (9.1)
In the domain in which Q(x, y) is not zero, equation 9.1 can equivalently be
written as;
dy
= f (x, y) (9.2)
dx
where
P (x, y)
f (x, y) = − (9.3)
Q(x, y)

106
Ordinary differential equations 107

We will discuss some methods which allow us to solve some of the equations of
this form and those equations which reduce to the first order form.

9.2 Basic concepts and definitions

Definition 9.1 Any equation of the form:

F (x, y, y 0 , y 00 , · · · , y (n) ) = 0,

where F is a real valued function of the n + 1 variables x, y, y 0 , y 00 , · · · , y (n) is

called an nth order ordinary differential equation where x is the independent
variable and the variables y, y 0 , y 00 , · · · , y (n) depend on x.
Any equation of the form:
F (x, y, y 0 ) = 0,
where F is a real valued function of three variables x, y and y 0 is called a first
order ordinary differential equation where x is the independent variable and
y and y 0 depend on x.

The following are examples of ordinary differential equations:

dy
Example 9.1 1. = ky, k is a constant.
dt
dy xy − y 2
2. = 2 , x2 − xy 6= 0
dx x − xy
p
3. x 1 + (y 0 )2 = 0

dy d5 y
4. x4 + = y + x7
dx dx5

One can rewrite a first order equation in the form:

 
dy
F x, y, = 0,
dx

which is often rearranged into the form:

dy
= f (x, y).
dx

Definition 9.2 The order of a differential equation is that of the highest deriva-
tive occurring in the differential equation.
108 Ordinary differential equations

Example 9.2
y 0 + ay 2 = bx2
is a first order differential equation, while
(y 000 )2 − 2xy 000 + y(y 0 )3 = 0
is a third order equation.

Definition 9.3 A solution of a first order differential equation is a function

y = φ(x), with first derivatives defined such that when substituted into the o.d.e
it yields an identity on the domain of definition of φ(x).
The general solution of a first order o.d.e is one with the arbitrary constant
that describes all possible solutions of the equation, while a particular solution
is determined by certain initial conditions, and is free of any arbitrary constants.

Exercises

1. Determine the order of each of the following o.d.e’s.

4 dy d5 y
(a) x + 5 = y + x7
dx dx
d3 y dy
(b) 3
+ x − y = x6
dx dx
0 2 32
(c) [x + (y ) ] = y 00
(d) y 000 + cos y 00 + x2 y 0 = x3
(e) (x0 )2 x000 = x4 x00 + t5 x0
2. Check whether the given function y = φ(x) is a solution of the given o.d.e.
1−x
(a) φ(x) = ; (x2 + 1)y 0 + y 2 + 1 = 0
1+x
d3 y d2 y
(b) φ(x) = e3x ; − 9 =0
dx3 dx2
d2 y dy
(c) φ(x) = xe3x ; 2
−9 = 6e3x
dx dx
(d) φ(x) = ln(−x), x < 0; xy 0 = 1

9.3 Types of first order differential equations

9.3.1 Equations with separable variables
We say that the first order differential equation of the form 9.2 is one of separable
variables if the function f (x, y) can be expressed as a product of a function which
Ordinary differential equations 109

solely depends on the variable x, g(x), and another function which solely depends
on the variable y, h(y), such that the differential equation then assumes the form:
dy
= g(x) · h(y) (9.4)
dx
dy
= g(x)dx, h(y) 6= 0 (9.5)
h(y)
If the differential equation can be presented in the form 9.1 where
P (x, y) = P1 (x) · P2 (y)
and
Q(x, y) = Q1 (x) · Q2 (y)
then we also consider such a differential equation to be of variables separable
type. At all points (x, y) where the functions P (x, y) and Q(x, y) are defined, the
equation can still be written in the form
M (x)dx + N (y)dy = 0 (9.6)
with
P1 (x) Q2 (y)
M (x) = , N (y) = , Q1 (x) 6= 0, P2 (y) 6= 0
Q1 (x) P2 (y)
The variables in equations 9.5 and 9.6 are said to be separable where 9.5 is the
separated form, hence the name separable variables. Now, suppose that the
function g(x) is continuous in an interval I1 and that the function h(y) is also
continuous and never zero in the interval I2 . Then the solution of 9.5 is given by
Z Z
dy
= g(x)dx + C
h(y)
where C is an arbitrary constant. Note that it is not necessary to introduce a
constant of integration for the first integral since this will be combined with the
one from theZsecond integral to produceZ one constant C.
dy
Let H(y) = and let G(x) = g(x)dx, then
h(y)
H(y) = G(x) + C (9.7)
is the general solution of 9.5. The differential equation of the form
dy
= g(x) (9.8)
dx
dy
is a particular case of = g(x)h(y) where h(y) = 1 and g is continuous in the
dx
dy
interval I. Similarly if g(x) = 1 in = g(x)h(y), we obtain
dx
dy
= h(y) (9.9)
dx
110 Ordinary differential equations

The class of equations 9.8 and 9.9 is called autonomous differential equations.
Autonomous equations are such that the right side of the equations does not
depend explicitly on the independent variable x. In Physics, the independent
variable is usually the time component t.

dy x2
Example 9.3 Show that the equation = is separable and then find an
dx 1 − y2
equation for the integral curves.

Solution
If we rewrite the equation as
dy
−x2 + (1 − y 2 ) = 0,
dx
then it has the form
dy
M (x) + N (y) =0
dx
and therefore separable. Solving:

(1 − y 2 )dy = x2 dx
y3 x3
y− = +C
3 3
Example 9.4 Show that the following differential equation is separable and find
a solution that satisfies the given initial condition.
xy − y
y0 = , y(2) = 1
1+y

Solution

y 0 (y + 1) = xy − y
y 0 (y + 1) = y(x − 1)
 
0 1
1−x+y 1+ = 0
y
 
1
(1 − x)dx + 1 + dy = 0
y
Z Z  
1
(1 − x)dx + 1+ dy = C
y
1
x − x2 + y + ln |y| = C
2
Now, y(2) = 1, =⇒ 2 − 2 + 1 + 0 = C, =⇒ C = 1 Hence
1
x − x2 + y + ln |y| = 1
2
Ordinary differential equations 111

Activity 9.1 Show that the differential equation

y 0 = xey−x

is separable and find the general solution.

Example 9.5 Life scientists have used the equation

dy
= ky(m − y),
dt
called the logistic equation to model the spread of a disease or information through
a population. If m denotes the total number of people in the population and y(t)
is the number of infected people at time t, then the o.d.e. states that the rate of
growth of the disease is proportional to the product of the number of people who
dy
are infected and the number who are not. i.e. ∝ y(m − y). Now
dt
dy
= kdt
y(m − y)
Z Z
dy
=⇒ kdt − = C
y(m − y)
Z 1 1 
m m
kt − + dy = C
y m−y
1 1
ln |y| − ln |m − y| = kt + C
m m
1 y
ln = kt + C
m m−y
y
= C2 emkt
m−y
mC2 emkt
=⇒ y =
1 + C2 emkt
C2 m
=⇒ y =
C2 + e−mkt
Suppose we are given that y(0) = R < m, then we can find C2 :
C2 m R
R= , =⇒ C2 = , so
1 + C2 m−R
mR
y(t) =
R + (m − R)e−mkt
Observe that as t −→ ∞, y(t) −→ m which means assuming no cure is found,
eventually all the people will be infected.

Example 9.6 A rumour spreads through a population of 5000 people at a rate

proportional to the product of the number of people who have heard it and a
112 Ordinary differential equations

number who have not. Suppose that 100 people initiate the rumour and that a
total of 500 people know the rumour after two days. How long will it take for half
the people to hear the rumour?

Solution
100(5000) 5000
Here m = 5000 and R = y(0) = 100, so y(t) = −5000kt
= .
100 + 4900e 1 + 49e−5000kt
Now, we are given that y(2) = 500. We determine k using this.

5000
500 = , =⇒ k = 0.00017
1 + 49e−10000k
Therefore
5000
y(t) =
1 + 49e−0.85t
Now, to determine how long it will take for half the population to hear the
rumour, we solve
5000
2500 =
1 + 49e−0.85t
which gives t = 4.58. Thus, it will take slightly more than 4 12 days for half the
population to hear the rumour.
Exercises

1. Solve the differential equation.

(a) y 0 = y sin(2x + 3) (b) y 0 = (x2 + 1)(y 2 + y)
0 3 0
(c) y = (xy) (d) y = 3x(1 + y 2 )
sin 1 y2 + 1
(e) y 0 = 2 x (f) y 0 =
x y cos y y + xy
(g) y 0 = xey−x (h) y 0 = xy 2 − x − y 2 + 1
0 (y + 1)2
(i) (y ln x)y = (j) ey sin 2xdx + cos x(e2y − y)dy = 0.
x
2. Solve the initial value problems.
q
1−y 2 dy ex−y
(a) y 0 = x 1−x 2 , y(0) = 0. (b) = , y(1) = 0.
dx 1 + ex
dy x2 y − y dy
(c) = , y(3) = 1. (d) x2 = y − xy, y(−1) = −1
dx y+1 dx
(e) (xy 2 + y 2 + x + 1)dx + (y − 1)dy = 0, y(2) = 0
π
(f) cos ydx + (1 + e−x ) sin ydy = 0, y(0) = .
4

9.3.2 Equations reducible to variables separable type

These are some equations which can be transformed by certain substitutions or
transformations into the variables separable form.
Ordinary differential equations 113

Example 9.7 The differential equation

dy
=x+y
dx
can be transformed to one of variables separable by the substitution u = x + y,
noting therefore that y = u − x so that the o.d.e takes the form
du
= 1 + u.
dx
All first order o.d.e.’s of the form
dy
= F (a1 x + a2 y + a3 ),
dx
where a1 , a2 and a3 are constants can be reduced to variables separable type by
introducing the substitution
z = a1 x + a2 y + a3
dz dy
Now, = a1 + a2 and substituting for z into
dx dx
dy
= F (a1 x + a2 y + a3 )
dx
we obtain
dz
= a1 + a2 F (z).
dx
Separating the variables we obtain
dz
= dx
a1 + a2 F (z)
and on integration we obtain the general solution
Z
dz
x= + C.
a1 + a2 F (z)
We can have the case where a1 = 0; which implies that the differential equation
dy
= F (a2 y + a3 )
dx
is already of the variables separable type with the general solution given by
Z
dz
x= + C.
F (a2 y + a3 )
We may also have the case where a2 = 0 and similarly we obtain the general
solution after separating the variables;
Z
y = F (a1 x + a3 )dx + C.
114 Ordinary differential equations

dy
Example 9.8 Prove that the differential equation = (x+y)2 can be reduced to
dx
one of variables separable type by the substitution v = x + y. Obtain the solution
if y(0) = 0.

Solution
The substitution v = x + y allows us to obtain
dv dy
=1+
dx dx
or
dy dv
= −1
dx dx
after we differentiate with respect to x, and the differential equation takes the
form
dv
− 1 = v2
dx
or
dv
= v2 + 1
dx
which is separable. Its general solution is

v = tan(x + C)

which becomes
y = tan(x + C) − x
after back substitution. Now, y(0) = 0 gives the particular solution

y = tan x − x.

Activity 9.2 Use the substitution xy = t to solve (x2 y 2 + 1)dx + 2x2 dy = 0.

9.3.3 Homogeneous equations

dy
Definition 9.4 The differential equation = f (x, y) is called a first order
dx
homogeneous equation if it can be written in the form:
dy y
=F , x 6= 0. (9.10)
dx x

y 0 y y y−x
; y = log x−log y; y 0 = +sin and y 0 =
Differential equations like y 0 =
x x x y+x
2
y x
are all homogeneous. However, the equation y 0 = + is not homogeneous.
x y
The form (4.10) suggests the introduction of a new variable by the substitution
Ordinary differential equations 115

y
v = that is, y = vx. This substitution, first used by the Mathematician Leibniz
x
in 1961, transforms the differential equation 9.10 into one of variables separable.
dy dv
= x + v = F (v),
dx dx
or
dv F (v) − v
= , x 6= 0.
dx x
We separate variables and then integrate;
Z Z
dv dx
= , x 6= 0, F (v) 6= v, =⇒ G(v) = ln |x| + C.
F (v) − v x
for which y
G = ln |x| + C.
x
Definition 9.5 A function f (x, y) is said to be a homogeneous function of
degree r with respect to the variables x and y if

∀ t ∈ R, f (tx, ty) = tr f (x, y)

If the differential equation 9.10 is expressible in the form

P (x, y)dx + Q(x, y)dy = 0,

where P (x, y) and Q(x, y) are homogeneous functions of the same degree, then
it is a homogeneous differential equation.
y−x
Example 9.9 Solve y 0 = .
y+x

Solution
The differential equation is homogeneous given that:
y
y−x 1− x
y
f (x, y) = = y =F
y+x 1+ x
x
is a homogeneous function of degree r = 0. To determine its solution we use the
substitution y = vx. Thus
dy dv 1−v
=v+x =
dx dx 1+v
which is variable separable. Separating the variables we obtain:
(1 + v)dv dx
2
= , with x 6= 0
1 − 2v − v x
The solution to this equation is x2 − 2xy − y 2 = C
116 Ordinary differential equations

y 2 + x2
Example 9.10 Solve y 0 = , xy 6= 0.
yx

Solution
dy x y
Rewrite the equation in the form = + , with x, y 6= 0. It is clear that this
dx y x
equation is homogeneous. Let y = vx to obtain

dy dv
=v+x
dx dx
which reduces to
dv 1
v+x = + v, v 6= 0,
dx v
or
dv 1
= , vx 6= 0.
dx vx
Separating variables and integrating we obtain

1 2
v − ln |x| = C, x 6= 0.
2
Since y = vx, the general solution of this equation is given by

y 2 = Cx2 + x2 ln x2 .
y
Activity 9.3 Solve (xe x + y)dx − xdy = 0, x 6= 0.

Exercises

1. Use the substitution xy = t to solve (1 + x2 y 2 )y + (xy − 1)2 xy 0 = 0.

2. Use the substitution xy = t to solve (x2 y 3 + y + x − 2)dx + (x3 y 2 + x)dy = 0.

3. Use the substitution y = tx to solve

(x6 − 2x5 + 2x4 − y 3 + 4x2 y)dx + (xy 2 − 4x3 )dy = 0.

4. Use the substitution x ln y = t to solve

(xy + 2xy ln2 y + y ln y)dx + (2x2 ln y + x)dy

5. Solve the differential equations

p
(a) xy 0 = y + y 2 − x2
(b) 2xy 0 (x2 + y 2 ) = y(y 2 + 2x2 )
p
(c) xy 0 = y 2 − x2
Ordinary differential equations 117

9.3.4 Linear differential equations

A first order differential equation is linear if it has the form

a1 (x)y 0 + a0 (x)y = f (x) (9.11)

The characteristic element for linearity for this equation is in y and y 0 . Linear
differential equations are important given the frequency in which they occur in
applications.
The linear
R equation 9.11 requires an integrating factor of the form
v(x) = e P (x)dx where
a0 (x)
P (x) =
a1 (x)
for which the general solution becomes
Z
R
− P (x)dx
R R f (x)
y=e Q(x)e P (x)dx dx + Ce− P (x)dx ; Q(x) =
a1 (x)
1
Example 9.11 Solve y 0 + y = 2, x 6= 0
x
Solution
1 R 1
From the general form, P (x) = so that v(x) = e x dx = eln x = x. We multiply
x
the differential equation by the integrating factor to get:
 
0 1
x y + y = 2x
x
or
xy 0 + y = 2x
which is then written as
d
(xy) = 2x
dx
=⇒ yx = x2 + C
or
y = x + Cx−1 , x 6= 0.

Example 9.12 Solve y 0 + y = x.

Solution
Here P (x) = 1, so that v(x) = ex and multiplying the differential equation by
this integrating factor and integrating we obtain

ex y = (x − 1)ex + C

which simplifies to
y = Ce−x + x − 1.
118 Ordinary differential equations

Activity 9.4 1. Obtain the general solution of the linear differential equation

y 0 + ay = b

where a, b are constants with a 6= 0.

2. Determine the general solution of

y 0 + 2xy = x

and evaluate the particular solution that satisfies y(0) = 2.

3. Solve the initial value problem

xy 0 − 2y = 3x4 , y(−1) = 2.

Example 9.13 Newton’s Law of Cooling states that the rate of change of the
temperature T of an object is proportional to the difference between T and the
(constant) temperature τ of the surrounding medium, called the ambient temper-
ature. Assuming that the temperature of the object is a differentiable function,
the mathematical formulation of Newton’s Law is
dT
∝T −τ
dt
dT
=⇒ = −k(T − τ ), k > 0 constant
dt
dT
=⇒ + kT = kτ
dt
which is a first order linear equation with P (x) = k and f (x) = kτ . We solve
this equation.
R
v(t) = e kdt
= ekt
dT
ekt + ekt kT = kτ ekt
dt
d
=⇒ (ekt T ) = kτ ekt
dt
=⇒ ekt T = τ ekt + C
=⇒ T = τ + Ce−kt

An example to apply Newton’s Law of cooling is the following:

Example 9.14 A cup of coffee is served to you at 185◦ F in a room where the
temperature is 65◦ F . Two minutes later, the temperature of the coffee has dropped
to 155◦ F . How many more minutes would you expect to wait for the coffee to
cool to 105◦ F ?
Ordinary differential equations 119

Solution
Here τ = 65 and T (0) = 185. So by Newton’s Law of cooling
T = τ + Ce−kt
we find C:
185 = 65 + C, =⇒ C = 120.
Thus
T = 65 + 120e−kt
Now, using T (2) = 155, we determine k.
155 = 65 + 120e−2k
This gives k = 0.144. Hence
T = 65 + 120e−0.144t
Finally
105 = 65 + 120e−0.144t
gives t ≈ 7.63 minutes. Thus you should wait for another 7.63 minutes.
Exercises

1. Determine the general solution of the following differential equations.

(a) y 0 − 2y = 1 (b) xy 0 − 2y = −x
0
(c) 2y + 5y = 2 (d) y 0 − y = −2e−x
cos x
(e) y 0 − 2y = 1 − 2x (f) xy 0 + 2y =
x
0 0 2
(g) xy − y = 2x ln x (h) y + y=0
x+1
2 5
(i) y 0 + y = (x + 1) 2 .
x+1
2. Obtain the particular solution determined by the given conditions.
(a) y 0 + y = x, y(0) = 1. (b) xy 0 + 2y = xe−x , y(1) = −1.
1
(c) y 0 + y = , y(0) = e. (d) xy 0 − 2y = x3 ex , y(1) = 0
1 + ex
3. An object falling from rest in air is subject not only to the gravitational force
but also to air resistance. Assume that the air resistance is proportional
to the velocity and acts in a direction opposite to the motion. Then the
velocity of the object at time t satisfies
v 0 (t) = 32 − kv(t), k > 0 constant, v(0) = 0.
Here we are measuring distance in metres and the positive direction is down.
(a) Determine v(t).
32 32
(b) Show that v(t) cannot exceed and that v(t) → as t → ∞.
k k
(c) Sketch the graph of v(t).
120 Ordinary differential equations

dB
4. Suppose that a certain population P has a known birth rate and a
dt
dD
known death rate . Then the rate of change of P is given by
dt
dP dB dD
= − .
dt dt dt
dB dD
(a) Assume that = aP and = bP , where a and b are constants.
dt dt
Find P (t) if P (0) = P0 > 0.
(b) Analyze the cases (i) a > b, (ii) a = b, (iii) a < b.
Find lim P (t) in each case.
t→∞

9.4 Second order linear differential equations with

constant coefficients
The differential equations we consider in this section take the general form:
dn y dn−1 y d2 y dy
an n
+ a n−1 n−1
+ · · · + a 2 2
+ a1 + a0 y = f (x),
dx dx dx dx
where the coefficients ai , i = 0, 1, · · · , n are real constants and an 6= 0. If f (x) =
0, the equation is said to be homogeneous. Solutions of such equations are of
exponential form.

9.4.1 Homogeneous equations of first order

dy
a+ by = 0
dx
dy b
To solve this equation, write = − y and we see that this is a separable
dx a
equation.
Z Separating
Z the variables and integrating both sides:
dy b b b
=− dx =⇒ ln y = − x + C =⇒ y = C1 e− a x
y a a

9.4.2 Homogeneous equations of second order

d2 y dy
a + d + cy = 0
dx2 dx
If we try a solution of the form y = emx , then y 0 = memx and y 00 = m2 emx .
Substituting into the equation we get
am2 emx + bmemx + cemx = 0
=⇒ emx (am2 + bm + c) = 0
=⇒ am2 + bm + c = 0.
Ordinary differential equations 121

The last equation is called the auxiliary equation or the characteristic equa-
tion for the second order ode. Solving this equation leads to 3 cases of solution:
Case 1: Distinct real roots:
If the roots are m1 and m2 , m1 6= m2 , then we get two solutions y1 = em1 x and
y2 = em2 x . The general solution is thus

y = c1 em1 x + c2 em2 x .

Case 2: Repeated real roots:

When m1 = m2 , we obtain only one Zexponential solution y1 = em1 x . A second
e2m1 x
solution of the equation is y2 = em1 x dx = xem1 x . The general solution
e2m1 x
then becomes
y = c1 em1 x + c2 xem1 x
Case 3: Complex conjugate roots:
If m1 and m2 are complex conjugates, then we can write m1 = α + jβ, m2 =
α − jβ where α, β > 0 are real and j 2 = −1. In this way we have

y = c1 e(α+jβ)x + c2 e(α−jβ)x .

However this solution is not practically useful as it is in terms of j. We need to

get rid of the j. Using Euler’s formula, we write ejβx = cos βx + j sin βx and
e−jβx = cos βx − j sin βx. Adding and subtracting the two equations results in

ejβx + e−jβx = 2 cos βx

and
ejβx − e−jβx = 2j sin βx
Since y = c1 e(α+jβ)x + c2 e(α−jβ)x is a solution, the choices c1 = c2 = 1 and
c1 = 1, c2 = −1 give, in turn, 2 solutions:

y1 = e(α+jβ)x + e(α−jβ)x = eαx ejβx + e−jβx = 2e αx
cos βx
−jβx
(α+jβ)x (α−jβ)x αx jβx


and y2 = e −e =e e −e = 2jeαx sin βx
We have thus, without loss of generality, shown that the real functions eαx cos βx
and eαx sin βx are solutions of the second order equation under this case 3. Con-
sequently, the general solution is

y = Aeαx cos βx + Beαx sin βx = eαx (A cos βx + B sin βx).

Example 9.15 Solve the following differential equations:

d2 y dy
1. 2 2
− 5 − 3y = 0
dx dx
d2 y dy
2. 2
− 10 + 25y = 0
dx dx
122 Ordinary differential equations

3. y 00 + y 0 + y = 0

Solution
The solution process involve giving the auxiliary equations, the roots and the
corresponding general solutions.

1
1. 2m2 − 5m − 3 = 0 =⇒ (2m + 1)(m − 3) = 0, =⇒ m1 = − , m2 = 3 and
x
2
y = c1 e− 2 + c2 e3x

2. m2 − 10m + 25 = 0 =⇒ (m − 5)2 = 0 =⇒ m1 = m2 = 5 and

y = Ae5x + Bxe5x
√ √
1 3 1 3
3. m2 + m + 1 = 0 =⇒ m1 = − + j, m2 = − − j and
√ 2 !2
√ 2 2
x 3 3
y = e− 2 c1 cos x + c2 sin x .
2 2

Activity 9.5 1. Determine the general solution of the given differential equa-
tion:

(a) y 00 − y 0 − 6y = 0
d2 y dy
(b) 2
+ 8 + 16y = 0
dx dx
00 0
(c) y + 3y − 5y = 0
(d) 12y 00 − 5y 0 − 2y = 0
d2 y dy
(e) 2
− 3 + 2y = 0
dx dx
00
(f ) y + 9y = 0
(g) 4y 00 + y 0 = 0

2. Solve the given differential equation subject to the indicated initial condi-
tions:

(a) y 00 + 16y = 0, y(0) = 2, y 0 (0) = −2

(b) y 00 − y = 0, y(0) = y 0 (0) = 1
(c) y 00 + 6y 0 + 5y = 0, y(0) = 0, y 0 (0) = 3
(d) y 00 − 8y 0 + 17y = 0, y(0) = 4, y 0 (0) = −1