UNIVERSITY OF NEW SOUTH WALES

SCHOOL OF MATHEMATICS AND STATISTICS

MATH1141 Higher Mathematics 1A

CALCULUS
Semester 1 2015

Professor Wolfgang K. Schief

Chapter 1
Sets, inequalities and functions
1.1

Sets of numbers

A set is a collection of distinct objects. The objects in a set are called the elements
or members of the set.
The set N of natural numbers is given by
N = {0, 1, 2, 3, 4, . . .}.
The set Z of integers is given by
Z = {. . . , 3, 2, 1, 0, 1, 2, 3, . . .}.
3

 The set Q of rational numbers is given by

Q = {fractions of integers} =

p
: p, q Z, q 6= 0 .
q

The set R of real numbers may be represented as the collection of points lying on
the number line.
|

Notation. If x is a member of a set A, then we write x A. If x is not a member

of A then we write x
/ A.
Example.

22

/ N,
7

22

/ Z,
7

22
Q,
7

Remark. The set

is the set of irrational numbers.

{x R : x
/ Q}

22
R
7

Notation for intervals. Suppose that a and b are real numbers and that a < b.
Then
(a, b) = {x R : a < x < b}

(open)
a

[a, b] = {x R : a x b}

(closed)
a

[a, b) = {x R : a x < b}

(???)
a

(a, b] = {x R : a < x b}

(???)
a

An interval [a, b] that includes its endpoints a and b is called a closed interval, while
an interval (a, b) that excludes its endpoints is called an open interval. The intervals
[a, b) and (a, b] are neither open nor closed.
5

Rays of the real line using the symbol .

[a, ) = {x R : a x}
a

(, b) = {x R : x < b}

(, ) = R
Definition. We say that a set A is a subset of a set B if every element of A is an
element of B. If A is a subset of B then we also say that B contains the set A.
Examples.
N is a subset of Z, and Z is a subset of Q, and Q is a subset of R.
{0, 2, 3} is a subset of {0, 1, 2, 3, 5}.

(1, 2] is not a subset of [0, ).

Any set is a subset of itself.
(1, 3) is a subset of [1, 3).

1.2

Solving inequalities

Inequalities. Remember the following rules for manipulating inequalities.

If x > y then x + z > y + z.

If x > y and z > 0 then xz > yz.

If x > y and z < 0 then xz < yz.

Here, x, y and z are real numbers.

Two types of inequalities deserve special attention: quadratic inequalities and rational
inequalities.

Examples. (a) Solve the quadratic inequality

x2 + 4x 21.

(b) Solve the rational inequality

1
1
<
.
x + 1 (x 2)(x 3)

1.3

Absolute values

The magnitude, or absolute value, of a real number x is denoted by |x| and defined
by
(
x
if x 0
|x| =
x if x < 0.
Properties. Suppose that x and y are real numbers. Then
| x| = |x|,

|xy| = |x||y|,
 
 x  |x|
  =
provided that y 6= 0,
y
|y|
|x + y| |x| + |y|,

|x y| |x| |y|.

The fourth property is known as the triangle inequality.

9

The following facts are useful for solving inequalities.

For every real number x,
|x| =

|x|2 = x2.

x2 ,

Geometrically, |xa| represents the distance between x and a on the real number
line.
For any positive real number a,
|x| < a

x2 < a2

a < x < a.

For any positive real number a,

|x| > a

x2 > a2

10

x < a or x > a.

Examples. Solve the following inequalities.

(a)
|3x + 1| 4

(b)
|x + 5|
<1
|x 11|

11

1.4

Functions

A function
f :AB
is a rule which assigns to every element x belonging to a set A exactly one element
f (x) belonging to a set B, that is
x 7 f (x).

Terminology.

A is called the domain of the function f , that is

A = Dom(f ) = {all allowable inputs}.

B is called the codomain of f , that is

B = Codom(f ) = {all allowable outputs}.

The range of f is

Range(f ) = {f (x) : x A}
= {all outputs that actually occur}.
12

Example.
f : [0, ) R

x 7 2 + x
Remarks.

Dom(f ) = [0, ),

Codom(f ) = R,

Range(f ) = [2, )

f denotes a function, while f (x) B is a number, namely the value of f at the

point x A.
The codomain of f may be changed but it must contain all the outputs of f .
The statement

f (x) =

f (x) =

may be abbreviated as

x for all x in [0, )

x

x [0, ).

Functions which are defined by the same rule but have different domains are not
the same.
13

Example.

f : R R, f (x) = x2
g : [0, ) R, g(x) = x2
f (x)

g(x)

g is invertible, while f is not!

Natural domain. If, for whatever reason, the domain of a function is not defined
then we may choose the natural domain or maximal domain, that is the largest possible domain for which the rule makes sense (for real numbers).

14

Examples. Find the maximal domain for

(a)
f (x) =

(b)
f (x) =

x 1/x.

1
.
2
x +x2

Remark. We distinguish between the range and the codomain of a function since
we may not actually be able to determine the range of a function.
Exercise. What is the range of the function f defined by
f (x) = x2 + ex ?

f : R [0, ),
15

Combining functions. If f and g are two functions with the same domain, then
one can combine f and g to form new functions.
Definition. Suppose that f : A B and g : A B are real-valued functions.
Then, the functions f + g, f g, f.g and f /g are defined by the rules
(f + g)(x) = f (x) + g(x)
x A
(f g)(x) = f (x) g(x)

x A

(f.g)(x) = f (x)g(x)

x A

f (x)
(f /g)(x) =
g(x)
where A = A \ {x A : g(x) = 0}.

x A,

Example. Determine the maximal domains of f /g and g/f , where

f (x) = 1 + x2, g(x) = cos(x).

16

Another way of constructing new functions is given below.

Definition. Suppose that
f :CD

and

g:AB

are functions such that Range(g) is a subset of Dom(f ). Then the composition
is defined by the rule

f g :AD
(f g)(x) = f (g(x))

x A.

Example. Suppose that the functions f and g are given by the rules
p
f (x) = ln x, g(x) = 4 x2
What is the maximal domain of the composition f g?

17

1.5

Polynomials and rational functions

Polynomials. A function f : R R is called a polynomial if

f (x) = anxn + an1xn1 + + a2x2 + a1x + a0,

where n N is the degree and the coefficients a0, a1, . . . , an are real numbers with
the leading coefficient an 6= 0.
Rational functions. Suppose that p and q are polynomials. The function f defined
by the rule
p(x)
f (x) =
q(x)
with
Dom(f ) = {x R : q(x) 6= 0}
is called a rational function.

18

Previous example. f defined by

1
,
f (x) = 2
x +x2
is rational.

Dom(f ) = R \ {2, 1}

Remark. The function defined by

3
(x2 + 3)(x 1) + 3
f (x) = x 1 + 2
=
x +3
x2 + 3
is also rational and Dom(f ) = R.

19

1.6

The trigonometric functions

Definition. The trigonometric functions

are defined by

sin : R R

cos : R R

and

sin x = s
and
cos x = c,
where s and c are defined in a geometric manner:
s

Question. Is it obvious that

d
sin x = cos x ?
dx
20

The following properties are evident.

Range(sin) = Range(cos) = [1, 1].

sin and cos are periodic of period 2, that is

sin(x + 2) = sin x,

cos(x + 2) = cos x.

cos is an even function, that is

cos(x) = cos x.
sin is an odd function, that is
sin(x) = sin x.
sin2 x + cos2 x = 1.

21

Other trigonometric functions with suitable domains

sin x
,
cot x =
tan x =
cos x
1
sec x =
, cosec x =
cos x

are defined by
cos x
sin x
1
.
sin x

The six trigonometric functions are related by various identities and formulae (which
you are supposed to know):
complementary identities


x = cos x
sin
2


cos
x = sin x
2
Pythagorean identities
cos2 x + sin2 x = 1
1 + tan2 x = sec2 x
cot2 x + 1 = cosec2 x
22

 the sum and difference formulae

sin(x y) = sin x cos y cos x sin y
cos(x y) = cos x cos y sin x sin y
tan(x y) =

tan x tan y
1 tan x tan y

double-angle formulae
sin(2x) = 2 sin x cos x
cos(2x) = cos2 x sin2 x

2 tan x
.
tan(2x) =
1 tan2 x

23

1.7

The elementary functions

The elementary functions are all those functions that can be constructed by combining
(a finite number of) polynomials, exponentials, logarithms, roots and trigonometric
functions (including the inverse trigonometric functions) via function composition,
addition, subtraction, multiplication and division. Hence the following expressions
give rise to elementary functions:
f (x) = esin x + x2,
ln x tan x

,
x
p
3
h(x) = x4 2x2 + 5,

k(x) = |x| = x2.

g(x) =

It also follows that every rational function is an elementary function.

However, there exist important functions which are not elementary!

24

1.8

Implicitly defined functions

Many curves on the plane can be described as all those points (x, y) on the plane
that satisfy some equation involving x and y. For example, consider the equation
(x2 + y 2 1)3 x2y 3 = 0.

()

The set of points (x, y) satisfying this equation are shown on the graph below.
y

x
Properties.
There exist several y-values for some x-values. Hence, the curve cannot be the
graph of one function of x.
The curve may be decomposed into two curves which may be regarded as the
graphs of two functions, f and g, say.
25

y = f (x)

b
|

y = g(x)

b
|

We say that the functions f and g are implicitly defined by the relation ().
[In this case, can one determine formulae for these functions f and g?]

26

Chapter 2
Limits
Question. How would you define continuity?
Intuitive (incorrect) answer. The function is continuous if its graph can be
drawn without lifting the pencil off the page.
Rigorous (correct) answer. Via limits.

27

f (x)

g(x)

h(x)

bc

bc

f has a removable discontinuity at x = a.

g has a jump discontinuity at x = a.

h has an essential discontinuity at x = a.

28

Counterexample. Consider the function

tan : A R
with
A = Dom(f ) = {x R : x 6= 2 + n, n Z}.
tan x

The function tan is continuous everywhere! The break in the graph is merely due to
the missing points in the domain A.
29

2.1

Limits of functions at infinity

In this section, we examine some techniques for calculating limits of the form lim f (x)
x
1
and postpone their precise definition.
x2
ln x
1

Notation.
x
If f (x) gets arbitrarily close to some real number L as x tends to infinity, then
we write
lim f (x) = L
x
or
f (x) L as x .
If f (x) gets arbitrarily large (that is, approaches ) as x tends to then we
write
f (x) as x .
We never write lim f (x) = since is not a real number.
x

30

2.1.1

Basic rules for limits

Elementary rules.
If f is a constant function, that is f (x) = c for all x, then
lim f (x) = c.

If f (x) as x then

1
= 0.
x f (x)
lim

These are intuitively obvious and give limits such as

1
1
lim = 0,
lim ex = lim x = 0.
x x
x
x e

31

Theorem. Suppose that

lim f (x) = a,

lim g(x) = b

for some functions f and g. Then

lim [f (x) + g(x)] = a + b
x

lim [f (x) g(x)] = a b

x

lim [f (x)g(x)] = ab
x

f (x) a
=
x g(x)
b

lim

provided that b 6= 0.

Proof. Later.

32

Example. Determine the limit of

as x .

2
1+
3x + 4
f (x) =
5 6ex

33

2.1.2

The pinching theorem

Idea. Assume that two functions f and h have the same limit as x and the
graph of a function g lies between the graphs of f and h (if x is large enough). Then,
g has the same limit as f and h.
y
y = h(x)

y = g(x)
L
b
|

y = f (x)

34

The pinching theorem. Suppose that f , g and h are three functions such that
f (x) g(x) h(x)
on an interval (b, ) for some b R and
lim f (x) = lim h(x) = L.

Then
lim g(x) = L.

Example. Determine the limit of

2

g(x) = ex 2sin 3x
as x .

35

2.1.3

Limits of the form f (x)/g(x)

Suppose that we want to calculate a limit of the form

f (x)
lim
,
x g(x)
where both f (x) and g(x) tend to infinity as x .
Problem. We cannot apply the preceding rules since f and g do not have limits.
Idea. Divide both f and g by the leading term, that is the fastest growing term
appearing in the denominator g (if it exists).

36

Examples. Find the following limits (if they exist).

(a)
5x3 + 6x2 4 sin x
lim
x cos 3x + 5x 2x3

(b)

x2 + 3x
lim
x
2x4 + 3 4x

37

2.1.4

Limits of the form

f (x)

g(x)

In this case, one needs to be very careful since a similar leading term argument may
be erroneous!
In fact, it is the lower order terms that may determine the limit!
Example. Determine the limit of
p

2
f (x) = 2x + 3x 2x2 x
as x .

Exercise. Does

4
3
lim x x x4 + 1

exist?

38

2.1.5

Indeterminate forms

but each displays a very different limiting

The following limits have the form
behaviour as x :
x2

x
x
20
x
2x2
2 2
x
Since we cannot determine in advance what kind of limiting behaviour something of

has, we say that

the form
is an indeterminate form.

Other types of indeterminate forms are

00

0
and appear in various applications.
39

2.2

The definition of lim f (x)

x

Example. Why do we believe that

1
lim f (x) = 0 for f (x) = ?
x
x
Consider the distance between f (x) and 0 denoted by
d(x) = |f (x) 0|.
Facts.
d(x) < 1 whenever x > 1.
y

y = f (x)

1
0 1
1

x
40

 d(x) < 0.2 whenever x > 5.

y
y = f (x)
0.2
0
0.2

d(x) < 0.1 whenever x > 10.

d(x) < 0.01 whenever x > 100.

d(x) < 0.0001 whenever x > 10000.

Set = 1/M . Then, d(x) < whenever x > M .

y
y = f (x)

M
41

Definition. Let f be a function defined on some interval (b, ) and let L be a real
number. We say that
lim f (x) = L
x

if
for every > 0, there exists a real number M such that
if x > M then |f (x) L| < .
y y = f (x)
L+
L
L

42

2.3

Proving that lim f (x) = L using the limit definition

x

Example. Prove that

5x
= 5.
x x + 3
lim

Proof. We consider the distance



 5x


|f (x) L| = 
5
x+3


 5x 5(x + 3) 

= 

x+3


 15 

= 
x + 3
15
=
x+3
15
<
x
43

for x > 3
for x > 0.

In summary,

15
|f (x) L| < .
x

This inequality gives an upper bound for the distance between f (x) and L!
Accordingly,
15
< .
|f (x) L| < whenever
x
The latter condition is equivalent to
15
x>

and hence if we set

15
M=

then
|f (x) L| < whenever x > M,
as required.
44

General strategy. Given , we need to find a number M such that

|f (x) L| < whenever x > M.
The number M can be found by following the procedure below.
1. Find a good upper bound for |f (x) L|.

2. Find a simple condition on x such that this upper bound is less than .
3. Use this condition to state an appropriate value for M (in terms of ).
Example. Show that

x2 cos x
lim
= 1.
x x2 + 1

45

Remarks.
In general, M depends on but it is not uniquely defined.

The definition of the limit does not tell you what the limit is. The definition may
be used to prove theorems which allow you to justify methods of finding limits.
Applying the definition to verify an educated guess for a limit is usuallly the last
resort. Make use of the theorems unless you are specifically asked to apply the
definition.

46

2.4

Proofs of basic limit results

Proof of the limit of a sum of two functions. Suppose that

lim f (x) = L1,

lim g(x) = L2

and set
1 = 2 =

for any given > 0.

By definition of the limit, there exist numbers M1 and M2 such that
|f (x) L1| < 1 whenever x > M1
and
|g(x) L2| < 2 whenever x > M2.
47

Hence, by the triangle inequality,



[f (x) + g(x)] (L1 + L2) |f (x) L1| + |g(x) L2|
< 1 + 2
=

whenever both x > M1 and x > M2.

If we set M = max{M1, M2} then


[f (x) + g(x)] (L1 + L2) < whenever x > M

so that

lim [f (x) + g(x)] = L1 + L2.

The other rules may be proven in a similar manner.

Proof of the pinching Theorem. See page 37 of the Calculus Notes.
48

2.4.1

Left-hand, right-hand and two-sided limits

Example. Consider the function f whose graph is shown below.

|

bc

bc

|
|

f (x)
4
3
2
1

With reference to this graph, we will discuss the behaviour of f (x) when x is near
the points 2, 4, 6 and 8.

49

Notation.
Let f be a function defined on an interval (c, a). We say that L is the left-hand
limit of f (x) as x approaches a if f (x) gets closer and closer to L when x gets
closer and closer to a from the left. We write
lim f (x) = L.

xa

Let f be a function defined on an interval (a, b). We say that L is the right-hand
limit of f (x) as x approaches a if f (x) gets closer and closer to L when x gets
closer and closer to a from the right. We write
lim f (x) = L.

xa+

50

 Let f be a function with the property

lim f (x) = lim f (x) = L
xa+

xa

for some point a Dom(f ). Then, we say that L is the (two-sided) limit of f (x)
as x approaches a and we write
lim f (x) = L.

xa

Remark. In all cases, the value of f at x = a (if defined) is irrelevant!

51

 For a = 2:

lim f (x) = 3,

x2

lim f (x) = 1

x2+

The two-sided limit does not exist.

For a = 4:

f (x) as x 4

No limit exists.
For a = 6:

lim f (x) = lim f (x) = 2,

x6

x6+

f (6) = 4

The two-sided limit exists but does not coincide with the value of f at a = 6.
For a = 8:

lim f (x) = lim f (x) = f (8) = 1

x8

x8+

The two-sided limit exists and coincides with the value of f at a = 8.

Question. What is so special about the above function at x = 8?

52

2.4.2

Limits and continuous functions

Definition. Let f be defined on some open interval containing the point a. We say
that f is continuous at a if
lim f (x) = f (a);
xa

otherwise we say that f is discontinuous at a.

If f is continuous at every point of its domain, we simply say that f is continuous.
Previous example. The function f is continuous everywhere except at x = 2 and
x = 6.
Note that x = 4 is not part of the domain of f and hence asking whether or not f
is continuous at x = 4 does not make any sense.
Remark. A rigorous definition of the preceding limits exists but, here, we will focus
on rules for calculating limits.

53

2.4.3

Rules for limits at a point

Polynomials, rational functions, the trigonometric functions, exponentials and logarithms are continuous at every point in the respective domains.
Limits at a point also have nice arithmetic properties ...
Theorem. Suppose that
lim f (x) = L1,

lim g(x) = L2

xa

xa

for some functions f and g. Then

lim [f (x) + g(x)] = L1 + L2
xa

lim [f (x) g(x)] = L1 L2

xa

lim [f (x)g(x)] = L1L2

xa

f (x) L1
=
lim
xa g(x)
L2

provided that L2 6= 0.
54

... and interact nicely with function composition.

Theorem. If lim f (x) = L and g is continuous at L then
xa

lim g(f (x)) = g(L).

xa

Example. Find the limit

lim

x5

2x + cos(x2) .

55

The pinching theorem. Suppose that f , g and h are all defined on an open interval
I containing the point a. If
f (x) g(x) h(x)

x I

and
lim f (x) = lim h(x) = L

xa

xa

then
lim g(x) = L.

xa

y
y = h(x)
L
a
|

y = g(x)
x
y = f (x)

56

Example. Is the function f defined by

sin x
x
f (x) =

1
continuous?

if x 6= 0
if x = 0

Concluding example. Suppose that f : R R is defined by the formula

|x 9|
if x 6= 3
x3
f (x) =

6
if x = 3.
Discuss the limiting behaviour of f (x) as x approaches 3.

57

58

Chapter 3
Properties of continuous functions
3.1

Combining continuous functions

Question. If two continuous functions are combined via function addition, subtraction, multiplication, division or composition, when is the resulting function also
continuous?
Theorem. Suppose that the functions f and g are continuous at a point a. Then
f + g, f g, f g
are continuous at a. If g(a) 6= 0 then
f /g
is also continuous at a.
59

Proof. Suppose that f and g are continuous at a. Then,

lim f (x) = f (a),

lim g(x) = g(a)

xa

xa

by the definition of continuity at a point. Therefore,

lim (f + g)(x) = lim (f (x) + g(x))

xa

xa

(def. of f + g)

= lim f (x) + lim g(x)

(limit rule)

= f (a) + g(a)

(f, g cont.)

xa

xa

= (f + g)(a)

(def. of f + g).

Hence f + g is continuous at a.
The proofs that the functions f g, f g and f /g are continuous at a are similar.

60

Example. Based on the fact that constant functions and the function f : R R
defined by f (x) = x are continuous at every point, show that polynomials and rational functions are continuous at every point of their respective domains.
Solution. Any polynomial can be obtained from f and constant functions via addition and multiplication, e.g.
x3 4x2 + 5 = [(x x x)] + [(4) x x] + 5,
and hence is continuous everywhere.
Any rational function is of the form
p(x)
,
q(x)
where p and q are two (continuous) polynomials, and is therefore continuous at every
point a for which q(a) 6= 0.

61

If we can prove that trigonometric functions are continuous at every point of their
domain then functions defined by rules such as
1 + tan x
f (x) =
1 + x2 + x6
are likewise continuous on the appropriate domain.
Thus, it suffices to verify the following Lemma:
Lemma. Show that the functions sin and cos are continuous everywhere.

62

Interlude. Why is the function f : R R defined by

f (x) = x
continuous everywhere?
Answer. Prescribe a R and > 0. Define = and assume that
|x a| < .
Then,
Thus, we conclude that

|f (x) f (a)| < .

for every > 0, there exists a number > 0 such that

|f (x) f (a)| < whenever |x a| < .

The latter is the formal definition of continuity of any function f at a point a.

Accordingly, the particular function considered above is continuous everywhere.

63

Even larger classes of continuous functions may be obtained in the following manner:
Theorem. Suppose that f is continuous at a and that g is continuous at f (a).
Then g f is continuous at a.
Proof.
lim (g f )(x) = lim (g(f (x))

(def. of g f )

= g(lim f (x))

(Th. of Ch. 2)

xa

xa

xa

= g(f (a))

(f cont.)

= (g f )(a).

(def. of g f )

Hence g f is continuous at a.

64

Example. Suppose that the function g : R R is continuous at the point a and

let f : R R be the function defined by
f (x) = |g(x)|.
Show that f is continuous at the point a.

65

3.2

Continuity on intervals

Strictly speaking, the previous example requires a natural extension of continuity to

endpoints. We first recall the standard definition ...
Definition. Suppose that f is a real-valued function defined on an open interval
(a, b). We say that f is continuous on (a, b) if f is continuous at every point in the
interval (a, b).
... and make the following generalisation.

66

Definition. Suppose that f is a real-valued function defined on a closed interval

[a, b]. We say that
f is continuous at the endpoint a if
lim f (x) = f (a),

xa+

f is continuous at the endpoint b if

lim f (x) = f (b),

xb

f is continuous on the closed interval [a, b] if f is continuous on the open interval

(a, b) and at each of the endpoints a and b.
Further generalisation. The function f : [0, ) R defined by f (x) =
continuous on [0, ).

67

x is

Example. Consider the functions f , g and h, whose graphs are shown below.
f (x)

g(x)

h(x)

bc

bc

b x

b x

b x

All three functions are defined on the interval [a, b].

f is continuous on the open interval (a, b) and at the endpoint b.

g is continuous at the endpoints a and b but not continuous on the open interval
(a, b).

h is continuous on the closed interval [a, b] (and, by implication, on the open

interval (a, b) and at both endpoints a and b).
68

3.3

The intermediate value theorem

z
f (a)

f (b)

f (x)
b

The intermediate value theorem. Suppose that f is continuous on the closed

interval [a, b]. If z lies between f (a) and f (b) then there exists at least one real
number c in [a, b] such that f (c) = z.

69

Proof. Omitted but based on the the least upper bound property of the real numbers:
If a1 a2 a3 M is a non-decreasing bounded sequence of real numbers
then ak converges to some real number L M as k .
(It is this property that distinguishes the real numbers from the rational numbers!)

70

Remarks.
The number c in [a, b] may not be unique.
|

z
f (b)

f (a)

f (x)

c1

c2

c3

There exists three numbers ci with f (ci) = z.

71

 Continuity of f is crucial.
f (x)

f (a)

bc

f (b)

For all c [a, b], we see that f (c) 6= z.

72

 The intermediate value theorem is a theorem about real numbers. Consider, for
instance, the function
f : [0, 2] R,

f (x) = x2 2

so that, in particular, f (0) = 2 and f (2) = 2.

The intermediate value theorem says that there exists a real number c [0, 2]
such that f (c) = 0. In fact,

c = 2 6 Q

and, hence, an analogous theorem for rational numbers cannot exist.

73

Application. Show that there exists a solution c [1, 2] of the equation

c = c2 1
and approximate its value.

Example. Suppose that f : R R is continuous and

lim f (x) = 1,

lim f (x) = 1.

Show that there exists a point c R such that f (c) = 0.

74

3.4

The maximum-minimum theorem

Definition. Suppose that f is defined on a closed interval [a, b].

We say that a point c in [a, b] is an absolute minimum point for f on [a, b] if
f (c) f (x)

for all

x [a, b].

The corresponding value f (c) is called the absolute minimum value of f on [a, b].
If f has an absolute minimum point on [a, b] then we say that f attains a minimum
on [a, b].
We say that a point d in [a, b] is an absolute maximum point for f on [a, b] if
f (x) f (d)

for all

x [a, b].

The corresponding value f (d) is called the absolute maximum value of f on

[a, b]. If f has an absolute maximum point on [a, b] then we say that f attains a
maximum on [a, b].
An absolute maximum point and an absolute minimum point are sometimes referred
to as a global maximum point and a global minimum point.
75

1 2 3 4 5 x

|
|

h(x)
4
3
2
1

g(x)
4
3
2
1

Example. Consider the functions g and h, whose graphs are illustrated below.

1 2 3 4 5 x

The absolute minimum and maximum points of g and h on [1, 5] are recorded in the
following table.
g
h
Absolute minimum points none
3
Absolute minimum value
n.a.
1
Absolute maximum points 2, 4 none
Absolute maximum value
4 n.a.
76

The above example shows that a function f : [a, b] R need not have an absolute
maximum point (or an absolute minimum point) on a closed interval [a, b].
The maximum-minimum theorem. If f is continuous on a closed interval [a, b]
then f attains a minimum and maximum on [a, b]. That is, there exist points c and
d in [a, b] such that
f (c) f (x) f (d)
for all x in [a, b].

Proof. Omitted but, once again, the least upper bound property of the real numbers
is used. The proof relies on the crucial fact that f is continuous on the closed interval
[a, b]!

77

Example. Suppose that f : R R is continuous and

lim f (x) = lim f (x) = 0

Show that f attains a maximum or a minimum (or both) on R.

Question. Is there a link between continuity and boundedness?

78

Definition. A function f is said to be bounded on an interval I if there exists some

positive number M such that
|f (x)| M

for all

x I.

In other words, f is bounded if the y-values of its graph lie between M and M for
some positive number M .
Theorem. If f : [a, b] R is continuous on [a, b] then f is bounded on [a, b].
Proof. Suppose that f is continuous on [a, b]. By the maximum-minimum theorem,
f attains a maximum and minimum on [a, b], that is
f (c) f (x) f (d)

for some c and d and all x in [a, b]. If we set

then
and hence f is bounded.

M = max{|f (c)|, |f (d)|}

|f (x)| M

79

Remark. A function may be bounded without having a maximum or minimum value.

Example. Is the function f defined by

 

1 sin 1
x
f (x) = x2 + 1

if x 6= 0
if x = 0

bounded on [2, 2]? Does it attain a minimum or maximum on [2, 2]?

80

Chapter 4
Differentiable functions
9

s = t2

t [0, 3].

What is the snowboarders instantaneous

speed when t = 1?

displacement (m)

1
0

81

Typical question. A snowboarder accelerates from rest in such a way that her displacement s (in metres) from her starting position
after t seconds is given by

1 2 3
time (s)

Intuitive solution. Calculate the average velocity v of the snowboarder over any
time interval according to
s
,
v=
t
where s denotes the change in displacement corresponding to a change t in time ...
Time interval t Average velocity
[1, 1.5]
0.5
2.5
[1, 1.4]
0.4
2.4
[1, 1.3]
0.3
2.3
[1, 1.2]
0.2
2.2
[1, 1.1]
0.1
2.1
[1, 1.01]
0.01
2.01
[1, 1.001] 0.001
2.001
... and let t approach 0. It appears that
v 2

as

t 0

and hence define 2 (meters/second) to be the instantaneous speed at t = 1.

82

Geometric interpretation. Find the slope (gradient) of the (tangent line to the)
graph at t = 1.
Thus, calculate the gradient of the secant intersecting the graph at (1, 1) and some
other point P ...
s = t2
secant
P

2.25

s
1

grad. of secant =
b

1 1.5
t
83

s
t

(1.5)2 (1)2
1.51

= 2.5

Intersection points
t = 1, t = 1.5
t = 1, t = 1.4
t = 1, t = 1.3
t = 1, t = 1.2
t = 1, t = 1.1

t Gradient of secant
0.5
2.5
0.4
2.4
0.3
2.3
0.2
2.2
0.1
2.1

... and let P approach the point (1, 1).

84

s = t2
secants
tangent

1.21
1

|
|

1.44

1.96
s
1.69

2.25

1 1.1 1.2 1.3 1.4 1.5

t 85

4.1

Gradients of tangents and derivatives

Example. Find the gradient of the tangent to the graph of the function f : R R
defined by f (x) = x3.

y
y = f (x)
secant
(x + h, (x + h)3)

(x + h)3 x3
(grad. at x) = lim
h0
h
3x2h + 3xh2 + h3
= lim
h0
h
x3
= lim 3x2 + 3xh

(x, x3)

h0
2

= 3x .
|

x
86

x+h

tangent

Definition. Suppose that f is defined on some open interval containing the point x.
If
f (x + h) f (x)
lim
h0
h
exists then it is called the derivative of f at x and we say that f is differentiable at x.
The derivative of f at x is denoted by
f (x)

or

df
(x)
dx

Remark. The ratio

or

d
f (x).
dx

f (x + h) f (x)
h
is called the difference quotient for f at the point x.

87

Example. Show that

Exercise. Write
and conclude that

d
cos x = sin x.
dx

cos2 h 1
sin2 h
cos h 1 =
=
cos h + 1
cos h + 1
cos h 1
lim
= 0.
h0
h

88

Example. For which n N is the function f defined by

 

xn sin 1
if x 6= 0
x
f (x) =

0
if x = 0
continuous or differentiable at x = 0?

Exercise. Show that

d
c = 0, where c is a constant.
dx
d m

x = mxm1, where m is an integer

dx
[without using the product rule!]
d

sin x = cos x.
dx

89

4.2

Rules for differentiation

Many differentiable functions may be constructed via function addition, subtraction,

multiplication and division ...
Theorem. Suppose that f and g are differentiable at x. Then,
(f + g)(x) = f (x) + g (x)
(cf )(x) = cf (x), where c is a constant
(f g)(x) = f (x)g(x) + f (x)g (x)
f (x)g(x) f (x)g (x)
(f /g) (x) =
g(x)2

(product rule)
(quotient rule)

provided that g(x) 6= 0.

... and function composition:

90

Theorem. Suppose that g is differentiable at the point x and f is differentiable at

the point g(x). Then,
(f g)(x) = f (g(x))g (x) (chain rule).
Exercise. Use the product rule and induction on n to prove that
d n
x = nxn1
dx
for n N.
Example. Find the derivative of


f (x) = sin

91

x
x2 + 1

2

4.3

Proofs of results in Section 4.2

Proof of product rule. Suppose that f and g are differentiable at the point x.
The difference quotient of f g at x gives
(f g)(x + h) (f g)(x)
h
f (x + h)g(x + h) f (x)g(x)
=
h
g(x + h) g(x)
f (x + h) f (x)
+ f (x)
.
= g(x + h)
h
h
We will show later that if a function is differentiable at a point then it must be
continuous at that point! Hence,
g(x + h) g(x) as h 0.
Accordingly,
(f g)(x + h) (f g)(x)
g(x)f (x) + f (x)g (x)
h
as h 0.
92

Proofs of the other differentiation rules are found in most undergraduate calculus
textbooks.

93

4.4

Implicit differentiation

Idea. On use of the chain rule, determine the derivative of a function which is
implicitly defined.
Example. Determine the tangent line to the curve defined by
at the point (2, 1).

x4 x2y 2 + y 4 = 13

Write
and differentiate both sides:

x4 x2y(x)2 + y(x)4 = 13

4x3 2xy(x)2 2x2y(x)y (x) + 4y(x)3y (x) = 0.

Evaluate at (x, y) = (2, 1) and solve for y (2):

32 4 8y (2) + 4y (2) = 0

Equation of tangent line at (2, 1):

y 1 = 7(x 2).
94

y (2) = 7.

Theorem. Suppose that q is a rational number. Then

d q
x = qxq1.
dx

Proof. Since q is a rational number, there exist integers m and n such that
m
q= .
n
Write
y = xq = xm/n
and take the nth power of both sides, leading to
y n = xm .
Differentiation of both sides with respect to x yields
ny n1

dy
= mxm1.
dx
95

Hence,
dy mxm1
=
dx
ny n1
m xm1
=
n xq(n1)
= qx(m1)qn+q
= qxq1
as required.

96

4.5

Differentiation, continuity and split functions

Theorem. If f is differentiable at a then f is continuous at a.

Proof. It is seen that


f (a + h) f (a)
h
h0
h


f (a + h) f (a)
= lim
lim h
h0
h0
h

lim f (a + h) f (a) = lim

h0

= f (a) 0 = 0
since f is differentiable at a. Accordingly,
lim f (a + h) = f (a).

h0

Conversely: If f is not continuous at a then it is not differentiable at a!

97

Previous example. At x = 0, the function f defined by

 

xn sin 1
if x =
6 0
x
f (x) =

0
if x = 0
is continuous for n 1 and differentiable for n 2.

Extra care is required if one deals with split functions. Whether or not the function is
differentiable at the split point can be determined by calculating left- and right-hand
limits of the difference quotient. ...

98

Example. Suppose that f : (0, ) R is defined by

(
4 x
if 0 < x 1
f (x) =
bx2 + c
if x > 1,
where b and c are real numbers. Find all possible values of b and c such that f is (i)
continuous at x = 1 and (ii) differentiable at x = 1.

... However, in many cases, this may be avoided by using the following theorem:

99

Theorem. Suppose that a is a fixed real number and that a function f is defined
by the rule
(
p(x)
if x a
f (x) =
q(x)
if x < a,
where p and q are defined on some open interval containing a. If f is continuous at
a and p(a) = q (a) then f is differentiable at x = a.
Remark. Note that the requirement of f being continuous at a is equivalent to
demanding that p(a) = q(a) since p and q are continuous at a.
Previous example. Consider the function f : (0, ) R defined by
(
if 0 < x 1
4 x
f (x) =
bx2 + c
if x > 1,
where b and c are real numbers.

100

4.6

Derivatives and function approximation

By definition, if a function f is differentiable at a, we have

f (x) f (a)
xa
xa

f (a) = lim
so that

f (x) f (a)
f (a)
xa
if x is sufficiently close to a. Thus,

f (x) f (a) + f (a)(x a)

and the right-hand side may be regarded as an approximation of f (x) in a neighbourhood of x = a.

101

Example. Suppose that the horizontal distance L covered by a tennis ball is given
by
v2
L = sin 2,
g
where v and are the initial speed and angle respectively and g is the constant
vertical acceleration due to gravity. How does L change if either v or are changed
by a small amount?

102

4.7

Derivatives and rates of change

Many physical processes involve quantities (such as temperature, volume, concentration, velocity) that change with time but may not be independent of each other.
Their rates of change may then be obtained by careful application of the chain rule.
Example. A spherical balloon is being inflated and its radius is increasing at a constant rate of 6 mm/sec. At what rate is its volume increasing when the radius of the
balloon is 20 mm?
Let V (t) be the volume of the balloon and r(t) be its radius at time t. Alternatively,
let V (r) be the volume of the balloon as a function of its radius r given by
4
V = r3
3
so that

dV
= 4r2.
dr
103

Then, the chain rule implies that

dV
dV dr
2 dr
=
= 4r .
dt
dr dt
dt
We are told that
so that
at r = 20.

dr
=6
dt
dV
= 4(20)2 6 = 9600
dt

Hence, the volume is increasing at a rate of 9600 mm3/sec when the radius is 20 mm.

104

The above example illustrates an approach to solving such problems.

1. Define variables for the quantities involved.
2. Write down what is known in terms of these variables and their derivatives.
3. Write down what you need to find in terms of these variables and their derivatives.
4. Write down anything else you know that relates the variables (for example, a
volume or area formula).
5. Use the chain rule (or implicit differentiation) to find the relevant derivative.

105

4.8

Local maximum, local minimum and stationary points

In this section, we begin to develop a systematic approach to locating maxima and

minima. A complete approach will be presented in the next chapter.

Local minimum point c

b

ch

c+h

Definition. Let f be defined on some interval I.

We say that a point c in I is a local minimum point if there exists an h > 0 such
that
f (c) f (x) for all x (c h, c + h) I.

We say that a point d in I is a local maximum point if there exists an h > 0 such
that
f (x) f (d) for all x (d h, d + h) I.
106

Theorem. Suppose that f is defined on (a, b) and has a local maximum or minimum
point at c for some c in (a, b). If f is differentiable at c then f (c) = 0.
Definition. If a function f is differentiable at a point c and f (c) = 0 then c is
called a stationary point of f .
Example. Find all the stationary, maximum and minimum points of the function
f : [3, 4] R defined by
f (x) = 4x5 5x4 40x3 2
f (x)
200

3
200

107

Differentiation yields
f (x) = 20x4 20x3 120x2
= 20x2(x2 x 6)
= 20x2(x + 2)(x 3).
Result:
x = 3: local minimum point

x = 2: stationary point: local maximum point

x = 0: stationary point: point of inflection

x = 3: stationary point: global minimum point

x = 4: global maximum point

The proof of the above theorem may be found in the Calculus Notes.

108

Chapter 5
The mean value theorem and its applications
The mean value theorem is one of the most important results for establishing the
theoretical structure of calculus.
Applications of the mean value theorem include
identifying where a function is increasing or decreasing,
identifying different types of stationary points,

determining how many zeros a polynomial has,

evaluating limits which are indeterminate forms of type

proving useful inequalities and

estimating errors in approximations.

109

and 00 ,

5.1

The mean value theorem

f (x)
C

f (a)

|
|

f (b)

A
|

The mean value theorem. Suppose that a function f is continuous on [a, b] and
differentiable on (a, b). Then, there exists at least one real number c in (a, b) such
that
f (b) f (a)
= f (c).
ba
110

Remark. In the above theorem, it is required that f is continuous on the closed

interval but differentiable only on the open interval!
Example. Find counterexamples which demonstrate that the continuity and differentiability requirements must be met.

Application. A car enters a tunnel at a speed of 30km/h and after 1 minute leaves
the tunnel at a speed of 40 km/h. The length of the tunnel is 1km. Did the driver
break the speed limit of 50km/h?

111

Example. Apply the mean value theorem to the function f defined by f (x) = x5
and a = 1, b = 4. Find the value(s) of c which satisfy the conclusion of the theorem.

112

5.2

Proof of the mean value theorem

If f (a) = f (b) then the mean value theorem reduces to Rolles theorem.
g(x)
C

A
a

B
b x

Rolles theorem. Suppose that a function g is continuous on [a, b] and differentiable

on (a, b). If, in addition,
g(a) = g(b) = 0
then there exists a c in (a, b) such that g (c) = 0.

113

Proof of Rolles theorem.

Case 1: Suppose that
g(x) = 0.
Then, g (c) = 0 for every c in (a, b).
Case 2: Suppose that there exists a point d in (a, b) such that
g(d) > 0.
By the max-min theorem, g attains a maximum value at some point c in [a, b].
Moreover,
g(c) g(d) > g(a) = g(b) = 0

so that c must lie in (a, b).

Since g is differentiable on (a, b), we know that

g (c) = 0.
114

Case 3: Suppose that

g(x) 0

for all x in [a, b] and that g is not constant on [a, b].

Then, g attains a minimum at a point c in (a, b) and hence g (c) = 0.
In order to prove the mean value theorem, we merely deform the graph of f in such
a way that Rolles theorem applies:

115

Proof of the mean value theorem. Suppose that f is continuous on [a, b] and
differentiable on (a, b).
We consider the function g defined by


f (b) f (a)
(x a) + f (a) .
g(x) = f (x)
ba
It is evident that g is continuous on [a, b], differentiable on (a, b) and that
g(a) = g(b) = 0.
By virtue of Rolles theorem, there exists a c in (a, b) such that g (c) = 0, that is,
such that


f (b) f (a)
f (c)
= 0.
ba
Hence,
f (b) f (a)

f (c) =
ba
as required.
116

5.3

Proving inequalities using the mean value theorem

An important application of the mean value theorem is in proving inequalities.

First example. Show that
ex > 1 + x

for all

x > 0.

Hint: Make the identification [a, b] = [0, x] for any fixed x > 0.

General philosophy. Apply the mean value theorem to an appropriate function f

and find a lower or upper bound for f (c) on (a, b).

117

Example. It is known that any polynomial grows faster than the logarithm. For,
instance, show that
ln x < x 1 for all x > 1.
Suppose that x > 1 and consider the closed interval [1, x]. We define a function
f : [1, x] R
by
f (t) = ln t.
Now, f is continuous on [1, x] and differentiable on (1, x) so that the mean value
theorem implies that
f (x) f (1)
= f (c)
x1
for some c in (1, x). Accordingly,
ln x
1
=
x1 c
for some c between 1 and x.
118

An upper bound of f (c) on (1, x) is given by

f (c) =

1
<1
c

since c > 1. We therefore conclude that

ln x
<1
x1
or, equivalently,
ln x < x 1.

119

5.4

Error bounds

A second application of the mean value theorem is in calculating error bounds.

Example. How good an approximation of cos 1 is cos 3 ?
Since 1 3 , we are inclined to believe that cos 1 12 .
We apply the mean value theorem to the function
f : [1, 3 ] R,

f (x) = cos x

which is continuous and differentiable. Thus,

cos 3 cos 1
= f (c) = sin c

3 1

for some c in (1, 3 ).

120

Accordingly,


0 < cos 1 cos =

1 sin c 1 < 0.05
3
3
since < 3.15 so that
0.5 < cos 1 < 0.55.

Conclusion: The error of approximating cos 1 by cos 3 is at most 0.05.

121

5.5

The sign of a derivative

Definition. Suppose that a function f is defined on an interval I. We say that

f is increasing on I if for every two points x1 and x2 in I,
x1 < x2

implies that

f (x1) < f (x2).

f is decreasing on I if for every two points x1 and x2 in I,

x1 < x2

implies that

f (x1) > f (x2).

Example. The function f : R R given by f (x) = x2 is increasing on [0, 10) and

decreasing on [5, 0].
f (x)

increasing

decreasing

x
122

Theorem. Suppose that f is continuous on [a, b] and differentiable on (a, b).

If f (x) > 0 for all x in (a, b) then f is increasing on [a, b].

If f (x) < 0 for all x in (a, b) then f is decreasing on [a, b].

If f (x) = 0 for all x in (a, b) then f is constant on [a, b].

Proof. Suppose that f (x) > 0 for all x in (a, b) and choose two points x1 and x2
in [a, b] such that x1 < x2.
Since f is differentiable on I, it is continuous on [x1, x2] and differentiable on (x1, x2).
Hence, by the mean value theorem,

for some c in (x1, x2).

Accordingly,

f (x2) f (x1)
= f (c)
x2 x1

f (x2) f (x1) = f (c)(x2 x1) > 0.

The remaining two statements are proven in a similar manner.

123

The above theorem may be directly used to classify stationary points.

Example. Find and classify all stationary points of the function f : R R whose
derivative is given by
f (x) = (x 4)(x 1)(x + 5)2.
Stationary points: Set f (x) = 0.
Result: x = 4, x = 1 and x = 5.
Classification: Investigate f (x) in a small neighbourhood of any stationary point.
5
x4
+
x1
+
(x + 5)2 +
f (x)
+
Gradient 

5 5+
+ +
+ +
0
+
0
+

124

+
+


0
+
0

1+

+
+

+
+

4
0
+
+
0

4+
+
+
+
+


Result:
x = 4: Local minimum point
x = 1: local maximum point

x = 5: horizontal point of inflexion

125

5.6

The second derivative and applications

Another (potential) method for classifying the stationary points of a function f involves the second derivative of f , which is denoted by
f

or

d2 y
,
2
dx

or

if we set y = f (x).
Theorem (The second derivative test). Suppose that a function f is twice
differentiable on (a, b) and that c (a, b).
If f (c) = 0 and f (c) > 0 then c is a local minimum point of f .

If f (c) = 0 and f (c) < 0 then c is a local maximum point of f .

126

The proof of this theorem uses the following result.

Lemma. Suppose that g is differentiable at a point c.
If g (c) > 0 then g(c h) < g(c) < g(c + h) for all positive h sufficiently small.

If g (c) < 0 then g(c + h) < g(c) < g(c h) for all positive h sufficiently small.

Exercise. Find and classify the stationary points of the function f : R R given
by
f (x) = x3 6x2 + 9x 5.

127

Remark. If f (c) = f (c) = 0, no conclusion may be drawn!

x4

x4

x3

If f (x) = x4 then f (0) = f (0) = 0 and there is a local minimum at 0.

If f (x) = x4 then f (0) = f (0) = 0 and there is a local maximum at 0.

If f (x) = x3 then f (0) = f (0) = 0 and there is a horizontal point of inflexion

at 0.

Hence if f (c) = f (c) = 0 then it is best to classify the stationary point c by

examining the sign of the derivative on either side of c!
128

5.7

Critical points, maxima and minima

Question. How does one find global maxima or minima?

Definition. Suppose that f is defined on [a, b]. We say that a point c in [a, b] is a
critical point for f on [a, b] if c satisfies one of the following properties:
c is an endpoint a or b of the interval [a, b]
f is not differentiable at c
f is differentiable at c and f (c) = 0.
f (x)
b
b

4
b

3
2

4
129

5 x

Theorem. Suppose that f is continuous on [a, b]. Then, f has a global maximum
and global minimum on [a, b]. Moreover, the global maximum point and the global
minimum point are both critical points for f on [a, b].
Example. Suppose that the function f : R R is given by the rule
f (x) = |x2 3x 4|.

Find the absolute maximum and absolute minimum values of f on the interval [0, 5]
(cf. previous graph).

130

5.8

Counting zeros

Example. Determine the number of real zeros of

f : R R,

f (x) = x4 x3 3x2 8x 5

and give an approximate location for each zero.

Differentiation yields
so that f (2) = 0 and

f (x) = (x 2)(4x2 + 5x + 4)

f (x) < 0 on (, 2)

and

f (x) > 0 on (2, ).

Now, since f (2) = 25, we can roughly sketch the graph (exercise!) and conclude
the following:
f is decreasing on (, 2) and therefore cannot have more than one zero on this
interval.
But f (1) = 2 and f (0) = 5 and, hence, the intermediate value theorem
implies that f has at least one zero on (1, 0).
131

 f is inreasing on (2, ) and therefore cannot have more than one zero on this
interval.
But f (3) = 2 and f (4) = 107 and, hence, the intermediate value theorem
implies that f has at least one zero on (3, 4).
Result: f has two real zeros, one in the interval (1, 0) and one in the interval (3, 4).

132

5.9

Antiderivatives

Remark. The velocity of a particle is the time-derivative of its position. Accordingly,

the position of a particle may be regarded as an antiderivative of its velocity.
Definition. Suppose that f is continuous on an open interval I. A function F is
said to be an antiderivative (or a primitive) of f on I if
F (x) = f (x)

for all

x I.

The process of finding an antiderivative of a function is called antidifferentiation.

Remark. It is evident that if F is and antiderivative of f then G defined by
G(x) = F (x) + C,
where C is an arbitrary constant, is also an antiderivative of f .

133

Example. Let

f(x) = xex ,

0.

Show that if F2 is an antiderivative of f2 for some 2 then F defined by

x1 x2 1
F(x) =
e
F2(x)
2
2
is an antiderivative of f.

134

Theorem. Suppose that f is a continuous function on an open interval I and that

F and G are two antiderivatives of f on I. Then, there exists a real constant C such
that
G(x) = F (x) + C
for all x in I.
Proof. Let H denote the function given by
H(x) = G(x) F (x)
for all x in I. Then, H is differentiable on I and
H (x) = G(x) F (x)
= f (x) f (x)
=0
for all x in I. Hence, there exists a constant C such that H(x) = C for all x in I so
that
G(x) = F (x) + C
for all x in I.
135

Some well-known antiderivatives are given below.

Function

Antiderivative

xr , where r is rational and r 6= 1

1
r+1
r+1 x

+C

sin x

cos x + C

cos x

sin x + C
1 ax
ae

eax
f (x)
f (x)

+C

ln |f (x)| + C

136

5.10

LH
opitals rule

Question. What is the limit of the indeterminate expression

x sin x
lim
?
x0 1 cos x
Theorem (lH
opitals rule). Suppose that f and g are both differentiable functions
in a neighbourhood of some a R and that either one of the two following conditions
hold:
f (x) 0 and g(x) 0 as x a

If

f (x) and g(x) as x a.

exists then

f (x)
lim
xa g (x)
f (x)
f (x)
= lim .
lim
xa g (x)
xa g(x)
137

Remark. The theorem also holds for

limits as x or x

one-sided limits (as x a+ or x a).

LHopitals rule is proved using the mean value theorem!

Example. Determine the limit
x sin x
.
x0 1 cos x
lim

138

Remark. lHopitals rule may be applied iteratively.

Example.

sin x x cos x
x sin x
= lim
= ... see above.
x0
x0
x sin x
1 cos x
lim

Remark. It is important that the limit exists after a finite number of applications of
lHopitals rule!
Exercise. What is the limit

2x sin x
lim
x 3x + sin x

and why can lHopitals rule not be applied?

139

140

Chapter 6
Inverse functions
Problem. Given a function
if we set

f : A B,

y = f (x),
under what circumstances is it possible to express x as a function of y, that is, to
find a function
g:CA
such that
x = g(y) ?
Remark. The above process of expressing the input of a function in terms of its
output is commonly referred to as inverting a function.
141

6.1

Some preliminary examples

Examples. Is it possible to invert the following functions?

(a)
x
f : [0, ) R,
y = f (x) =
x+1

(b)
f : [1, 1] R,

y = f (x) =

142

1 x2

Question. How do we make sure that a function is invertible and what is the domain
of any inverse function?
Remark. It is evident that the domain and range of a function play a crucial role in
answering this question! For example, formal inversion of c = c(t) given by
c = cme(ttd)
leads to

1 cm
t = td + ln .

But for which t and c does the latter make sense?

143

Standard example. Consider the rule

y = x2 .
Whether any function defined by this rule is invertible depends on the domain:
f1 : [0, ) R,

y = f1(x) = x2

If we take into account that Range(f1) = [0, ) then the inverse function is given
by

g1 : [0, ) [0, ),
x = g1(y) = y.
f2 : (, 0] R,

y = f2(x) = x2

If we take into account that, again, Range(f2) = [0, ) then the inverse function
is given by

g2 : [0, ) (, 0],
x = g2(y) = y.
144

 f3 : R R,

y = f3(x) = x2

The latter is not invertible since for any output y 6= 0 there exist two inputs

x = y and x = y.
f1(x)

f2(x)

f3(x)
output

output

output
input x

input

input

input x

Remark. It is evident that it might be possible to construct an invertible function

by restricting the domain of a given function.
Conclusion. The main criterion for invertibility is the existence of a one-to-one
correspondence between inputs and outputs.
145

6.2

One-to-one functions

Idea. A function is one-to-one if every output corresponds to a unique input.

Definition. A function f is said to be one-to-one
if

f (x1) = f (x2)

implies that

x1 = x2

for all x1, x2 Dom(f ).

Terminology. One-to-one functions are also called injective functions.
Remark. An injective function is equivalently characterised by
f (x1) 6= f (x2)

for all

in the domain of f .

146

x1 6= x2

Example. For which R is the function

f : R R,

f (x) = x3 + x + 1

one-to-one?

147

The horizontal line test. Suppose that f is a real-valued function defined on some
subset of R. Then, f is one-to-one if and only if every horizontal line in the Cartesian
plane intersects the graph of f at most once.
Examples.
f1(x)

f2(x)

f3(x)

f1 is not one-to-one.

f2 is one-to-one.

f3 is one-to-one (even though it is neither strictly increasing nor strictly decreasing).

148

Although not every one-to-one function is strictly increasing (or strictly decreasing),
it is true that every strictly increasing function is one-to-one.
Theorem. If a function f is either strictly increasing or strictly decreasing then f is
one-to-one.
Previous example. Show that the function
f : R R,

f (x) = x3 + x + 1

is one-to-one for > 0?

Since
f (x) = 3x2 + > 0
for > 0, f is strictly increasing and is therefore one-to-one for > 0.
Exercise: How does one argue in the case 0?

149

Example. Is the function f : (1, 1) R defined by

 
2
f (x) = 3 + x + 2 tan x
2
one-to-one?

Remark. Not every function whose derivative is only positive (or only negative) on
its domain is one-to-one. For example,
d
tan x = sec2 x 1
dx
but tan is not one-to-one on its maximal domain!

150

6.3

Inverse functions

Theorem. Suppose that f is a one-to-one function. Then, there exists a unique

function g satisfying
g(f (x)) = x

x Dom(f )

for all

and
f (g(y)) = y

for all

Moreover,
Dom(g) = Range(f ),
and g is one-to-one.

y Range(f ).
Range(g) = Dom(f )

Proof. Set D = Dom(f ) and R = Range(f ) and define the function

g:RD

by choosing as g(y) the unique x D for which y = f (x).

It is then left as an exercise to show that g has the properties listed above.
151

The theorem allows us to define the term inverse function.

Definition. Suppose that f is a one-to-one function. Then the inverse function of
f is the unique function g given by the above theorem. The inverse function for f is
often denoted by f 1.
Remark. If f 1 denotes the inverse function of a one-to-one function f then the
relations in the above theorem may be expressed as
f 1(f (x)) = x

for all

x Dom(f )

and
f (f 1(y)) = y

for all

y Range(f )

so that f may also be interpreted as the inverse of the function f 1.

Note. f 1(y) does not mean 1/f (y)!

152

Remark. Since f 1 is a function just like any other function, we regard it as a

function
x 7 f 1(x)
so that we can graph f 1 in the usual manner.
Example. Determine f 1, where
f : R R,
Set
so that

1 3
f (x) = 4 x .
3

1
y = 4 x3
3
p
x = 3 12 3y.

153

Hence, (interchanging x and y),

f

: R R,

(x) =

12 3x.

y=x

y = f 1(x)
x
y = f (x)

154

6.4

The inverse function theorem

Question. If the derivative of an invertible function exists, under what circumstances

is the inverse function likewise differentiable?
Subtlety. Consider the function
Its inverse is given by

f (x) = x3.

f : R R,

g : R R,
but g is not differentiable at x = 0!

g(x) =
y

g(x)
f (x)
x

Conclusion. Points at which the derivative of f vanishes must be excluded.

155

The inverse function theorem. Suppose that I is an open interval, f : I R

is differentiable and
f (x) 6= 0
for all x in I. Then,

f is one-to-one and has an inverse function

g : Range(f ) Dom(f )
g is differentiable at all points in Range(f )

The derivative of g is given by

1
g (y) =
f (g(y))

for all y Range(f ).

156

Proof.
Since f (x) 6= 0 on I, f is one-to-one (mean value theorem!).
g is differentiable ... too hard!
Differentiation of

f (g(y)) = y

with respect to y yields

f (g(y)) g (y) = 1.
Remark. Once again, we usually write the derivative of the inverse function g as
1
g (x) =
f (g(x))
for x Range(f ).

157

Example. Consider the function

f : (0, ) (0, ),

f (x) = x3

and its inverse

Then,

g : (0, ) (0, ),

g(x) =

x.

1
1
1
g (x) =
=
=
f (g(x)) 3[g(x)]2 3x2/3

as expected.
Previous example. Determine the derivative of the inverse of the function f :
(1, 1) R defined by
 
f (x) = 3 + x2 + 2 tan x
2
at the point f (x) = 3.

158

6.5

Applications to the trigonometric functions

The inverse sine function. We consider the restricted sine function

sin : [ 2 , 2 ] [1, 1].
This function is one-to-one and therefore has an inverse
sin1 : [1, 1] [ 2 , 2 ].
sin x

sin1 x

( 2 , 1)

(1, 2 )

( 2 , 1)

x
b

x
(1, 2 )

159

On (1, 1), according to the inverse function theorem, the derivative of sin1 is given
by
1
d
1
(sin x) =
.
1
dx
cos(sin x)
Since cos is positive on ( 2 , 2 ), we conclude that
d
1
1
1
(sin x) = p
.
=
2
1
2
dx
1x
1 sin (sin x)

Note.

d
(sin1 x) as x 1.
dx

160

Table of inverse trigonometric functions.

Function
sin
sin1
cos
cos1
tan
tan1

Domain

Range

Derivative

d
(sin x) = cos x
dx
d
1
1

[1, 1]
[ 2 , 2 ]
(sin x) =
dx
1 x2
d
[0, ]
[1, 1]
(cos x) = sin x
dx
d
1
1
(cos x) =
[1, 1]
[0, ]
dx
1 x2
d
( 2 , 2 ) (, )
(tan x) = sec2 x
dx
1
d
1

(, ) ( 2 , 2 )
(tan x) =
dx
1 + x2
[ 2 , 2 ]

[1, 1]

161

cos1 x

(1, )
b

tan1 x

(0, 2 )
b

(1, 0) x

x
b

Remark. Even though

sin(sin1 x) = x
for x [1, 1], in general,
unless x [ 2 , 2 ].

sin1(sin x) 6= x

162

Example. Determine
(a)

cos 2 sin

(b)

sin1 sin

163

3
.
5

5
.
6

164

Chapter 7
Curve sketching
y

x
x

165

In this chapter, we study curves in a plane which are given in terms of

a Cartesian equation such as
y = x2 4

or

a parameter such as

y=

1
.
2
1x

x(t) = sin t cos t ln |t|,

y(t) =

t [1, 1],

t 6= 0.

y = r sin ,

where

|t| cos t

polar coordinates such as

x = r cos ,

166

r = cos 4.

7.1

Curves defined by a Cartesian equation

In this section, we survey techniques for sketching curves which are described by a
Cartesian equation of the form y = f (x).
Checklist. Consider the following items:
Identify the domain of f .
Identify any symmetries:
f is even if
f is odd if

f (x) = f (x)
f (x) = f (x)

for all
for all

f is periodic of period T if

f (x + T ) = f (x)

for all

167

x Dom(f )
x Dom(f )
x, x + T Dom(f )

 Find x- and y-axis intercepts.

Identify vertical asymptotes.

Examine the behaviour of f (x) as x and identify existing (oblique)

asymptotes.
If necessary, identify stationary points and other features using calculus.
In the following, the above checklist is illustrated by various examples.

168

Symmetries. Consider the function f : R R defined by

f (x) =

| sin x|
.
2 + cos(2x)

f is even since f (x) = f (x) for all x R.

f is of period since f (x + ) = f (x) for all x R.

Intercepts. Sketch the graph of the function f : R R defined by

sin x for x 6= 0
x
f (x) =

1
for x = 0.

169

Oblique asymptotes. The graph of a function f may be asymptotic to a line

y = ax + b as x or x .
y y = ax + b

y = f (x)

Definition. Suppose that a 6= 0 and b are real numbers. We say that a straight line
given by the equation
y = ax + b
is an oblique asymptote for a function f if


lim f (x) (ax + b) = 0
x
or


lim f (x) (ax + b) = 0.
x

170

Remark. If f is a rational function with

p(x)
f (x) =
,
q(x)

deg(p) = deg(q) + 1

then the oblique asymptotes of f may be determined by polynomial division.

Example. Find the oblique asymptotes to the function f defined by
f (x) =
for all x 6= 2.

(x 2)|x| + x
x2

If x > 0 then
(x 2)x + x
x2
(x 2) + 2
=x+
x2
2
=x+1+
.
x2

f (x) =

171

Thus, the line y = x + 1 is an oblique asymptote.

If x < 0 then
(x 2)x + x
f (x) =
x2
(x 2) + 2
= x +
x2
2
= x + 1 +
.
x2
Thus, the line y = x + 1 is an oblique asymptote.

172

Example. Sketch the function f defined by

on the maximal domain.

x2 3
f (x) =
2x 4

173

7.2

Parametrically defined curves

Parametrically defined curves in a plane

x(t), y(t) ,

t A,

where t is the parameter and A is a given domain arise in various contexts such as
the planar motion of a point particle (or body) under the influence of forces.




Newtons 2nd law: m x(t), y(t) = F x(t), y(t)
a suitable parametrisation of a curve defined implicitly.




Ellipse: x(t), y(t) = a cos(t), b sin(t)

In principle, the analysis of parametrically defined curves is analogues to that outlined

in the previous section.
174

Example. Sketch the curve

(t) = x(t), y(t) = (t2 1, t3 1).

Possible values of x(t) and y(t):
It is evident that x(t) and y(t) assume any value in the intervals
x(t) [1, ),

y(t) (, )

respectively.
Intercepts:
x(t) = 0 if and only if t = 1 so that we obtain the points
(1) = (0, 2),
In addition, y(t) = 0 if and only if t = 1.
175

(1) = (0, 0).

Vector derivatives:
Consider the tangent vector

(t) = x (t), y (t) .

If one interprets (t) as the position of a particle at the time t then (t) is the
velocity of the particle at that time.

We will justify later that the slope of a parametrised curve (t) = x(t), y(t) is
given by
dy y (t)
=
dx x(t)
at all points with x(t) 6= 0.

176

Here,

dy y (t) 3
=
= t
dx x(t) 2

(t) = (2t, 3t2),

so that
and

(t) points

for

t<0

(t) points

for

t > 0.

Note that (0) = 0 so that the particle stops at t = 0!

In fact, there exists a cusp at (0) = (1, 1) since the normalised tangent vector
has the property
(2t, 3t2)
(t)
= lim
lim
= (1, 0).

2
4
t0
t0 | (t)|
4t + 9t
Conclusion: The curve does not have a proper tangent vector at the point (1, 1)!
177

Limiting behaviour:
As t ,

(2t, 3t2)
(t)
= lim
lim
= (0, 1).

2
4
t
t | (t)|
4t + 9t

Accordingly, the normalised tangent vector becomes vertical at infinity.

178

Justification of slope formula: By definition,

dy
y(t + h) y(t)
= lim
dx h0 x(t + h) x(t)

.

y(t + h) y(t)
x(t + h) x(t)
= lim
h0
h
h

 

y(t + h) y(t) .
x(t + h) x(t)
= lim
lim
h0
h0
h
h
y (t)
=
x (t)
provided that x(t) and y(t) are differentiable and x(t) 6= 0.

179

7.2.1

Parametrisation of conic sections

The ellipse
x2 y 2
+ 2 =1
2
a
b
with semi-axes a and b admits the parametrisation
x(t) = a cos t,

y(t) = b sin t,
y(t)

<t<
t=

<t<

t=

0<t<

t=0
a x(t)
b

a
3
2

0 t < 2.

b
b

t=

3
2

3
2

< t < 2

Each point (x, y) of the ellipse corresponds to a unique t [0, 2).

180

The table below lists some commonly used parametrisations of conic sections.
Conic section Cartesian equation Parametric equation
Parabola

4ay = x2

x(t) = 2at
y(t) = at2

Circle

x2 + y 2 = a2

x(t) = a cos t
y(t) = a sin t

Ellipse

x2 y 2
+ =1
a2 b2

x(t) = a cos t
y(t) = b sin t

Hyperbola

x2 y 2
=1
a2 b2

x(t) = a sec t
y(t) = b tan t

Note that there exist other useful parametrisations for each of these curves.
181

7.2.2

The cycloid and curve of fastest descent

Question. Find the shape of a curve that a particle should follow if it is to slide
without friction in the minimum time from a higher point A to a lower point B (not
directly beneath it) under the influence of gravity.
A
b

Curve of fastest descent.

Such a curve is known as a curve of fastest descent or a brachistochrone (which, in
Greek, means shortest time).

182

Answer. The curve of fastest descent from A to B is the unique arc of an (inverted)
cycloid whose tangent at A is vertical.
y

P
P

The cycloid.
Exercise. A circle of radius r rolls along the x-axis, starting from the origin as shown
above. Show that the locus (x(t), y(t)) of the point P on the edge of the circle which
satisfies (x(0), y(0)) = (0, 0) is given by
x(t) = r(t sin t)
y(t) = r(1 cos t),
where t 0.
Remark. If the particle moves on an inclined plane (surfer) then the trajectory is
still given by an arc of a cycloid!
183

7.3

Curves defined by polar coordinates

Many problems in mathematics are easier to solve if one chooses a suitable coordinate
system. Here, we focus on polar coordinates.
7.3.1

Polar coordinates

Every point P in a plane can be specified by, for example,

(x, y), where x is the horizontal distance of P from the origin, and y is the vertical
distance, or
(r, ), where r is the distance of P from the origin and is the angle (taken in
the anticlockwise direction) between OP and the positive horizontal axis.
x
b

P
b

y
O

O
184

The pair (x, y) is called Cartesian coordinates of P .

The pair (r, ) is called polar coordinates of P .
Note. If P is the origin then r = 0 and is not defined.
Polar coordinates (r, ) and Cartesian coordinates (x, y) of a point P are related by
x = r cos ,
and
r=
provided that x 6= 0.

x2

y 2,

y = r sin
y
tan =
x

Note. Finding Cartesian coordinates of a point P given in terms of polar coordinates

is easy but care must be taken in the opposite case.

185

Example. Find all polar coordinates of the point P with Cartesian coordinates

P = (x, y) = (3, 3).

Remark. Polar forms of equations may be simpler or more involved compared to

their Cartesian counterparts.
Examples. Find the polar forms of the
(a) straight line
y=

3 x.

(b) circle
x2 + y 2 = 4.

186

7.3.2

Basic sketches of polar curves

Many curves can be described by equations of the form

r = f ()

or

= g(r)

so that we obtain the parametrically defined curves

() = (r cos , r sin ) = f () cos , f () sin
or

(r) = (r cos , r sin ) = r cos g(r), r sin g(r) ,

where or r plays the role of the parameter.
Previous examples.
(a) =

corresponds to y =

3 x for x 0.

(b) r = 2 corresponds to x2 + y 2 = 4.

187

Remark. In order to sketch a curve represented by an equation in polar form, it may

be helpful to begin with an r vs sketch.
Example. Sketch the polar curve defined by
r = 2 cos ,

0 2.

We first graph r against ...

2
2 12
1

d
b

2 + 2

r
3
1

b
b

r = 2 cos

h
b

3
4

5
4

3
2

7
4

188

... and then mark the corresponding points on the (x, y)-plane:

y =

3
4

c r = 2
b

a r = 1
i

r=3

=0

g r = 2
b

5
4

=
=

3
2

189

7
4

Now:
As increases from 0 to , r increases from 1 to 3. Hence, the distance r from
the origin to points on the curve increases from 1 to 3.
As increases from to 2, r decreases from 3 to 1. Hence, the distance r from
the origin to points on the curve decreases from 3 to 1.
These considerations lead to the final sketch.
y
=

3
4

2
b

3
b

1
b

r = 2 cos

5
4
190

7
4

Symmetries.
If f () = f () then the polar curve is symmetric about the x-axis.

If f ( ) = f () then the polar curve is symmetric about the y-axis.

If f is 2periodic then it suffices to consider in the range 0 < 2.

Exercise. Sketch the curve described by the polar equation

r = 2 sin ,

and show that it constitutes a circle.

191

7.3.3

Sketching polar curves using calculus

Suppose that a curve can be expressed in polar form as

r = f ().
Since the curves parametric form is given by


() = x(), y() = (r cos , r sin ),

the tangent vector reads

() = x (), y ()


dr
dr
=
cos r sin , sin + r cos .
d
d
Thus, horizontal tangents are obtained by solving
y () = 0

but

x() 6= 0,

but

y () 6= 0.

while vertical tangents correspond to

x() = 0

192

Example. Sketch the curve described by the polar equation

r = 1 + cos .

193

Concluding remark. In connection with representing and sketching curves in terms

of polar coordinates, it is often convenient to allow negative values of r. Thus, if r
is negative, we make the identification
(r, )

(r, + )

and hence (r, ) is obtained from (r, ) by reflection in the origin. This is consistent
with the formulae
x = r cos = (r) cos( + )
y = r sin = (r) sin( + ).

Example. Sketch the curve described by the polar equation

1
r = + cos .
2

194

Chapter 8
Integration
Problem. How does one find/measure/define areas (of regions with curved boundaries)?
The main ideas can be traced back to
Archimedes (the great Greek mathematician of antiquity!) and then
Isaac Barrow (Isaac Newtons mentor)

Isaac Newton (one of the greatest mathematicians and physicists of all time)
Gottfried Leibniz (Newtons contemporary)

Bernhard Riemann (a German mathematician of the nineteenth century)

Henri Lebesgue (a French mathematician of the early twentieth century).

195

One (but not the only) method for calculating the area of regions with curved boundaries is known as Riemann integration.
Note. A priori, calculating areas (integration) and antidifferentiation are two separate problems but the remarkable fundamental theorem of calculus shows that these
are essentially the same!
This theorem was actually known to Barrow but its implications were developed by
Newton, Leibniz and their disciples.
Remark. The problem of calculating area is much the same as that of calculating
mass, volume, work and probability. The unifying feature is the so-called integral
calculus.
Here, we confine ourselves to the determination of the area under the graph of a
function via the Riemann integral. This may then be generalised to the area of sets
the boundary of which cannot be represented in this simple manner.
196

8.1

Area and the Riemann integral

We all (think that we) know how to calculate the area of a rectangle, a triangle or
even general polygons by partitioning the polygon into triangles.
But what are the rules that we take for granted?
The area of a rectangle is the product of its length and height.
Areas of congruent regions are equal.

The area of a whole region is the sum of the areas of its parts.
Thus, from the area of a rectangle (bh), we can derive the formula for
the area of a right-angled triangle ( 21 bh), then
the area of an arbitrary triangle ( 21 (b1 + b2)h) and then
the area of an arbitrary polygon.

h
b

b1

b2
197

Formally, we demand that any definition of an area satisfy the following axioms:

(A1) If is a region of the plane then

area() 0.
(A2) If one region 1 is contained in another region 2, then
area(1) area(2).
(A3) If the area of a region is partitioned into two smaller disjoint
regions 1 and 2, then
area() = area(1) + area(2).

198

2
1

1 2

(A4) If 1 and 2 are congruent regions then

area(1) = area(2).

2
1

(A5) If is a rectangle of length a and height b then

area() = ab.

b
a

199

8.1.1

Area of regions with curved boundaries

Question. How could one calculate the area of the region below?
1

Answer. Apply axiom (A3) and conclude that

area() = area(1) + area(2) + area(3).
Then, rotate and translate each subregion, so that application of axiom (A4) implies
that each of area(1), area(2) and area(3) is equal to the area under the graph of
a function.
y
y
y
f1
f2
f3
x

x
200

This procedure can be done for any region in the plane with a reasonable boundary.
Hence, we are left with defining what is meant by the area under the graph of a
function.

201

8.1.2

Approximations of area using Riemann sums

Example. Consider the function f : R R given by the rule

f (x) = x2.

Let denote the region bounded by the graph of f , the x-axis and the lines x = 0
and x = 1.
f (x)

f (x)

f (x)
1

1x

2
1x

1x

Idea. Find lower and upper bounds for area() by choosing appropriate approximations 1 and 2 of the region in terms of n rectangles.
202

It is evident that
area(1) area() area(2).
If limn area(1) = limn area(2) then
A = area() = lim area(1) = lim area(2).
n

Remark. One of Archimedes best ideas was to estimate areas in circles and parabolas by approximating them by inner and outer polygons!

203

y = f (x)

Explict evaluation of the bounds.

We begin by subdividing the interval
[0, 1] into n subintervals
[0, n1 ], [ n1 , n2 ], [ n2 , n3 ], . . . , [ n1
n , 1].
Rk

4 k1 k n1 n

1 2 3
n n n n

The set Pn given by

n n

n n

Pn = {0, n1 , n2 , n3 , . . . , n1
n , 1}

which divides the interval [0, 1] into these subintervals is called a partition of [0, 1].
Let Rk denote the area of the kth rectangle. Then
1
k2
k
Rk = width height = f ( n ) = 3 .
n
n
204

If S Pn (f ) denotes the total area of the shaded region then

n
n
X
1 X 2
k.
S Pn (f ) =
Rk = 3
n
k=1

k=1

One may show by induction (exercise!) that

n
X
1
2
k = n(n + 1)(2n + 1)
6
k=1

and hence

S Pn (f ) =

1
1
1
+
+ 2.
3 2n 6n

The quantity S Pn (f ) is called the upper Riemann sum of f with respect to the partition Pn.
Axiom (A2) now implies that
1
1
1
A S Pn (f ) = +
.
+
3 2n 6n2
205

In a similar manner, a lower bound is obtained:

y

y = f (x)

Rk

4 k1 k n1 n

1 2 3
n n n n

n n

n n

The area Rk of the kth rectangle is given by

1
(k 1)2
k1
Rk = f ( n ) =
.
3
n
n
206

The sum of all the areas of the rectangles is called the lower Riemann sum for the
function f over the partition Pn and is denoted by S Pn (f ). We obtain
S Pn (f ) =
so that

n
X
k=1

1 X
(k 1)2
Rk = 3
n

S Pn (f ) =

k=1

1
1
1

+ 2.
3 2n 6n

Axiom (2) therefore implies that

1
1
1
A S Pn (f ) =
+
.
3 2n 6n2

207

Hence, for every positive integer n, the inequality

1
1
1
1
1
1

+
A +
+
3 2n 6n2
3 2n 6n2
gives an upper and a lower bound for A.
Conclusion. In the limit n , we obtain
1
A= ,
3
regardless of the actual definition of A as long as it is compatible with the axioms
(A1)-(A5)!

Remark. The process of calculating upper and lower Riemann sums and taking a
limit of the above type (provided it exists) is called integration.

208

8.1.3

The definition of area under the graph of a function and the Riemann
integral

Suppose that f is a bounded function on [a, b] and that f (x) 0 for all x in [a, b].
Definition. A finite set P of points in R is said to be a partition of [a, b] if
P = {a0, a1, a2, . . . , an}
and
a = a0 < a1 < a2 < . . . < an = b.

209

Suppose that P is a partition of [a, b]:

f (x)

a0 a1 a2

a3 ak1ak an2 an1 an

Note. The points of P need not be evenly spaced.

The area of the kth rectangle in the above figure is
width height = (ak ak1) f k ,
where
f k = the maximum value of f on the subinterval [ak , ak1].
210

Note. Even though f is bounded, the maximum value of f on a given subinterval

may not exist. Then, f k is the least upper bound (MATH1241) of f on that subinterval.
The upper Riemann sum S P (f ) for f with respect to the partition P is defined by
S P (f ) =

n
X
k=1

(ak ak1)f k

which is the total area of the rectangles in the above figure.

Likewise, the lower Riemann sum S P (f ) for f with respect to the partition P is
defined by
n
X
S P (f ) =
(8.1)
(ak ak1)f k ,
k=1

where

f k = the minimum value of f on the subinterval [ak , ak1]

or the greatest lower bound (MATH1241).
211

Definition. Suppose that a function f is bounded on [a, b] and that f (x) 0 for
all x in [a, b]. If there exists a unique real number A such that
S P (f ) A S P (f )
for every partition P of [a, b] then we say that A is the area under the graph of f
from a to b.
In general, if we remove the condition f (x) 0 then the following definition is still
sensible:

212

Definition. Suppose that a function f is bounded on [a, b]. If there exists a unique
real number I such that
S P (f ) I S P (f )
for every partition P of [a, b] then we say that f is Riemann integrable on the interval
[a, b].
The unique real number I is called the definite integral of f from a to b and we write
Z b
f (x) dx.
I=
a

Remark. The function f is called the integrand of the definite integral, while the
points a and b are called the limits of the definite integral.
Exercise. Ponder about why this definition of area is consistent with the axioms
(A1)-(A5).
213

Historical remark. The notation

Z

f (x) dx
a

is due to Leibniz. It evolved from a slightly different way of writing down lower and
upper Riemann sums. For example, S P (f ) may be written as
S P (f ) =

n
X

f (xk )xk ,

k=1

where xk = ak ak1 and f attains its maximum value on [ak1, ak ] at the point xk .
When taking a limit as before, xk was replaced with dx and the symbol
replaced with an elongated S (S stands for sum).

214

was

8.2

Integration with Riemann sums

Even in the case f (x) = C, where C is a positive constant, the actual determination
of the Riemann integral is somewhat tedious.
Indeed, let P = {a0 = a, a1, . . . , an = b} be a partition of [a, b]. Since f is constant,
we have
fk = fk = C
for every k between 1 and n and hence
S P (f ) =

n
X
k=1

=C

f k (ak ak1)

n
X
k=1

(ak ak1)

= C(b a).
A similar calculation leads to
S P (f ) = C(b a).
215

Accordingly,
S P (f ) = C(b a) = S P (f )

for every partition P of [a, b] so that we conclude that f is Riemann integrable and
Z b
f (x) dx = C(b a).
a

Question. Can we find simple sufficient conditions which guarantee that a Riemann
integral exists?

216

Definition. A function f : [a, b] R is said to be piecewise continuous if it is

continuous on [a, b] at all except perhaps a finite number of points.
Examples.
f (x)

g(x)
b

bc
b

b x

b x

Both functions f and g are piecewise continuous but f is bounded while g is not!
Theorem. If f is bounded and piecewise continuous on [a, b] then f is Riemann
integrable on [a, b].
Proof. Difficult!
217

Remark. If we know for some reason that f is Riemann integrable on [a, b] (e.g.
because the above conditions are satisfied) then it is sufficient to show that
lim S Pn (f ) = lim S Pn (f )

for some sequence of partitions Pn of [a, b]. If I denotes this common limit, then
Z b
f (x) dx = I.
a

Example (Fermat). Find

x4 dx,

using a clever partition, i.e. rectangles of varying width!

218

Exercise. Find a Riemann integrable function which has infinitely many discontinuities.
Example. The function f : [0, 1] R defined by
(
1 if x Q
f (x) =
0 if x
/ Q.

is not Riemann integrable.

Proof. Let Pn be an arbitrary partition of [0, 1]. Then,

f k = 1,

fk = 0

for every k between 1 and n so that

S Pn (f ) = 1,

S Pn (f ) = 0.

Hence, there is no unique number I satisfying

S P (f ) I S P (f )

so that f is not Riemann integrable.

219

Remark. There exist more sophisticated ways of measuring areas, volumes etc.
such as Lebesque integration.
If f is Riemann integrable then it is Lebesque integrable.

If f is not Riemann integrable, it may still be Lebesque integrable. The Lebesque

integral of the above example is 0!
However, there exist regions in R2 to which it is impossible to assign an area in
any meaningful way! (Cf. Banach-Tarski paradox in three dimensions).

220

8.3

The Riemann integral and signed area

If f is a function which is Riemann integrable but not necessarily non-negative then

it is natural to refer to
Z c
f (x) dx
a

as the signed area under the graph of the function f from a to c and to
Z c
area() =
|f (x)| dx
a

as the unsigned area under the graph of the function f from a to c provided that the
latter integral exists.

221

Consider the following example:

f (x)
1
a

c
b

In this case, the unsigned area is

area() =

b
a

f (x) dx

222

f (x) dx.
b

8.4

Basic properties of the Riemann integral

y

f (x)

f M
m
a

bx

bx

bx

For brevity, from now on, we refer to Riemann integrable functions as merely integrable.
Theorem. Suppose that f, g : [a, b] R are integrable. Then,
(i) (Linearity) f + g is integrable for any , R with
Z

(f + g)(x) dx =
a

223

f (x) dx +
a

g(x) dx.
a

(ii) If a < c < b then

Z

f (x) dx =
a

(iii) If f (x) 0 for all x in [a, b] then

Z

f (x) dx +
a

f (x) dx.
c

f (x) dx 0.

(iv) If f (x) g(x) for all x in [a, b] then

Z b
Z b
f (x) dx
g(x) dx.
a

(v) If m f (x) M for all x in [a, b] then

Z b
m(b a)
f (x) dx M (b a).
a

(vi) If |f | is integrable on [a, b] then

Z
 Z
 b

b


|f (x)| dx.
 f (x) dx
 a

a
224

Sketch of proof.
(iii) If f (x) 0 then
0 S P (f )

f (x) dx.
a

(iv) Apply (iii) to h = g f .

(v) From (iv), it follows that
Z b
Z b
Z b
M dx.
f (x) dx
m dx
a

(vi) From
|f (x)| f (x) |f (x)|

x [a, b]

and (iv), we deduce that

Z b
Z b
Z b
|f (x)| dx.
f (x) dx
|f (x)| dx

225

It is convenient to introduce the following definition ...

Definition. Suppose that b < a and that f is integrable on [b, a]. Then,
Z a
Z b
f (x) dx
f (x) dx =
b

and

f (x) dx = 0.
a

... since, for instance, the following version of (ii) is still valid.
Remark. Suppose that a, b and c are real numbers and that f is integrable over
some interval containing a, b and c. Then
Z b
Z c
Z b
f (x) dx.
f (x) dx +
f (x) dx =
a

226

8.5

The first fundamental theorem of calculus

Idea (Leibniz, Newton). How does the integral (area) change as a boundary
changes?
y

y = f (t)

y = f (t)

F (x)
|

b t

x
b t
x+h

Suppose that a function f is continuous and therefore integrable on an interval [a, b].
We define the area function F by
F (x) =
for all x [a, b].

f (t) dt
a

227

Then,

F (x + h) F (x)
f (x),
h
where h is a small positive number.
Thus, we are tempted to believe that F is differentiable and
F (x + h) F (x)
= f (x).
h0
h

F (x) = lim

The first fundamental theorem of calculus. Suppose that f : [a, b] R is a

continuous function. Then, the function F : [a, b] R defined by
Z x
F (x) =
f (t) dt
(8.2)
a

is continuous on [a, b] and differentiable on (a, b) with

F (x) = f (x)
for all x in (a, b).
228

Implications.
Every continuous function f : [a, b] R has an antiderivative F on (a, b) given
by
Z x
F (x) =
f (t) dt.
a

Since two antiderivatives of f differ by a constant, integration and antidifferentiation are essentially the same procedures!
Differentiation undoes what integration does to f since
f (x) =

d
dx

Z

Is the converse true?

229

f (t) dt .
a

Proof of the 1. FTC.

y
Mh
mh

y = f (x)

xx+h

Differentiability.

Suppose that x (a, b) and choose an h > 0 such that x + h (a, b). Then,
Z x+h
F (x + h) F (x) =
f (t) dt.
x

Since f is continuous on [a, b], it attains a minimum value mh and a maximum value
Mh on [x, x + h], that is,
mh f (t) Mh
for all t [x, x + h].
230

We therefore conclude that

mh h
and hence
so that

x+h
x

f (t) dt Mhh

mhh F (x + h) F (x) Mhh

F (x + h) F (x)
mh
Mh
h

since h > 0.
Now, since f is continuous on [x, x + h], we know that
lim mh = f (x) = lim Mh

h0+

h0+

and hence

F (x + h) F (x)
= f (x)
+
h
h0
by virtue of the pinching theorem.
lim

231

In a similar manner, one can show that

F (x + h) F (x)
lim
= f (x).
h
h0
Hence, F is differentiable on (a, b) and
F (x) = f (x)
for all x in (a, b).
Continuity.
F is differentiable on (a, b) and is therefore continuous on (a, b).
Since f is continuous on [a, b], it is bounded on [a, b], that is, there exists an M > 0
such that
|f (x)| M
for all x in [a, b].

232

Hence, for x (a, b), we obtain:

Z x

Z a


|F (x) F (a)| = 
f (t) dt
f (t) dt
a
a
Z x




=
f (t) dt
a
Z x
|f (t)| dt

Za x
M dt

= M |x a|.

Accordingly,
|F (x) F (a)| M |x a| 0

as x a+. Thus, F is continuous at a.

Similarly, F is also continuous at the endpoint b.

233

Remark. The above proof of continuity of F at the endpoints could be applied to

any point on [a, b]. Thus, for continuity of F on [a, b], it is actually only required
that f is integrable and bounded!
Note. Many important functions are actually defined as area functions such as
Z x
1
ln(x) =
dt
t
1
Z x
2
2
erf(x) =
et dt
0
Z x
sin t
dt.
Si(x) =
t
0

The first fundamental theorem of calculus then implies that these functions are continuous and differentiable on the respective (closed or open) intervals.

234

8.6

The second fundamental theorem of calculus

There exists a fast way of calculating an integral if an explicit antiderivative is known.

The second fundamental theorem of calculus. Suppose that f is a continuous
function on [a, b]. If F is an antiderivative of f on [a, b], that is,
F (x) = f (x)
for all x [a, b], then
Z b
f (t) dt = F (b) F (a).
a

Proof. We know that two antiderivatives differ by a constant and hence

Z x
f (t) dt + C
F (x) =
a

for some constant C. Accordingly,

Z b
Z b
f (t) dt.
f (t) dt + C 0 C =
F (b) F (a) =
a

235

Notation. We frequently use the notation

b

F (x)
or
a

for the expression F (b) F (a).

236

F (x)

ib

Example. Calculate the area under the curve

y = f (x) = 3x2ex

between x = 0 and x = 1.
It is easy to verify that F given by
F (x) = ex

is an antiderivative of f , that is, F (x) = f (x), and hence

A = F (1) F (0) = e 1.
Question. How did we find the antiderivative of the above function f ?

237

8.7

Indefinite integrals

Notation. We denote the complete class of antiderivatives of a function f defined

on a suitable interval by
Z
f (x) dx.

The latter is called the indefinite integral of f . If F is an antiderivative of the function

f then
Z
f (x) dx = F (x) + C,

where C is an arbitrary constant of integration.

238

Examples. It is easy to verify that

Z
xn dx =
Z

1 n+1
x
+ C,
n+1

n 6= 1

sin x dx = cos x + C
Z

1 ax
e dx = e + C,
a
ax

a 6= 0

2x + 1
2

3x + 2 (1 + 6x) + C
dx =
27
3x + 2
Z
f (x)
dx = ln |f (x)| + C.
f (x)

239

8.8

Integration by substitution

Question. What is the analogue of the chain rule

d
F (g(x)) = F (g(x))g (x)
dx
in the context of integration?
Formally, if F is an antiderivative of a function f then
Z
f (g(x))g (x) dx = F (g(x)) + C,

since

d
F (g(x)) = F (g(x))g (x) = f (g(x))g (x).
dx

Previous example.
Z

2 x3

3x e dx =

x3


240

d 3
x3
x dx = e + C.
dx

Example. Find

1
dx.
x ln x

It is not always easy to recognise an integrand as being of the form

F (g(x))g (x).
In this case, the mechanical procedure which might lead to a successful integration
is called integration by substitution.

241

Thus, if we set
u = g(x)
then
Z

f (g(x))g (x) dx = F (g(x)) + C

= F (u) + C
Z
= f (u) du.

The shorthand for the above operation is

u = g(x),

du = g (x) dx

x = h(u),

dx = h(u) du.

or, if appropriate,

242

Example. Find
I=
Set
u=

3x + 2

2x + 1

dx.
3x + 2

3
du =
dx,
2 3x + 2

u2 2
.
x=
3

Hence,
I=

Z 

2(u 2)
2
+1
du
3
3

4 2 2
u du
=
9
9
4 3 2
= u u+C
27
9
2
4
3/2
3x + 2 + C.
= (3x + 2)
27
9
Note. The last step always consists of rewriting everything in terms of the original
variable!
243

The precise statement in the case of definite integrals is the following:

Theorem. Suppose that g is a differentiable function such that g is continuous on
[a, b]. If f is continuous on any interval I containing g([a, b]) then the change of
variables formula
Z g(b)
Z b
f (u) du
f (g(x))g (x) dx =
g(a)

holds.
Proof. Exercise.
Example. Evaluate the definite integral
Z 1
1

I=
dx.
2
4x
0

244

f (x)
a
|

Application. Suppose that f is a continuous function and a is a real number.

(i) If f is even then

f (x) dx = 2
a

(ii) If f is odd then

f (x) dx.
0

f (x) dx = 0.
a

(iii) If f is periodic with period T then

Z
Z a+T
f (x) dx =
a

245

f (x) dx.
0

Example. Find
Z

37
51

17

sin99 3x dx.

246

8.9

Integration by parts

Question. What is the analogue of the product rule

[f (x)g(x)] = f (x)g(x) + f (x)g (x)
in the context of integration?
Integration of both sides leads to
Z
Z
f (x)g(x) = f (x)g(x) dx + f (x)g (x) dx.
Hence, we obtain the integration by parts formula
Z
Z
f (x)g (x) dx = f (x)g(x) f (x)g(x) dx.
The corresponding formula for definite integrals reads
Z b
h
ib Z b
f (x)g (x) dx = f (x)g(x)
f (x)g(x) dx.
a

247

Sometimes, it may be relatively easy to apply the above formulae...

Example. Find the indefinite integral
Z
I = x cos x dx.
It is evident that
Z

d
I= x
sin x dx
dx

Z 
d
= x sin x
x sin x dx
dx
= x sin x + cos x + C.

... on other occasions, it may not be so easy:

248

Example. Find the definite integral

I=

ln x dx.
1

249

8.10

Improper integrals

Question. In many applications (e.g. probability theory, quantum mechanics etc.),

the class of integrals of bounded functions over finite intervals is too restrictive.
For instance, we could consider the improper integral
Z
f (x) dx,
a

but what is its exact definition?

f (x)

R
250

Definition.
(a) Suppose that there exists a real number L such that
Z R
f (x) dx L
a

as R . Then, f is said to be integrable over [a, ). We say that the

improper integral
Z
f (x) dx
a

is convergent and write

(b) Suppose that

f (x) dx = L.

f (x) dx
a

does not have a limit as R . Then, we say that f is not integrable over
[a, ) and the improper integral
Z
f (x) dx
a

is said to be divergent.

251

Remark. A similar definition for improper integrals of the form

Rb

f (x) dx

applies.

Example. The volume of a solid bounded by a curve y = f (x) rotated about the
x-axis and the planes x = a and x = b is given by
Z b
V =
[f (x)]2 dx.
a

Show that the volume of the body generated by the curve y = 1/x between 1 and
is
Z
1
V =
dx =
2
x
1
and, hence, finite.

252

A paradox. The surface area of a solid of the above type (excluding the vertical
boundaries) is given by
Z b
q
A = 2
f (x) 1 + [f (x)]2 dx.
a

Hence, for the above example, we obtain

Z b r
1
1
A = 2
1 + 4 dx
x
x
1
#b
"




x4 + 1
2
4
= ln x + x + 1
x2

[Verify this!]

Thus, the improper integral

b .

as

1
x

1+

diverges and the surface area is infinite!

253

1
dx
4
x

Question. How would one define the improper integral

Z
f (x) dx ?

Definition. We say that f is integrable over (, ) if f is integrable over both

(, 0] and [0, ). In this case, we write
Z
Z 0
Z
f (x) dx.
f (x) dx +
f (x) dx =

If f is not integrable on either of the intervals (, 0] or [0, ) then we say that

the improper integral
Z
f (x) dx

diverges.

254

Example. Does the improper integral

Z
converge?
Since

R
0

x dx

1
x dx = x2
2

as R , the improper integral

diverges by definition.

R
0

R2
=

x dx

Remark. The fact that

lim

is irrelevant

x dx = lim 0 = 0
R

R
255

Theorem (Convergence and divergence of p-integrals). The improper integral

Z
1
dx
p
x
1
is convergent if p > 1 and divergent if p 1.
Proof. If p 6= 1 then
Z

x
1

1p

x
dx =
1p

R
1

R1p 1
=
1p
(
1
p1 when 1 p < 0

when 1 p > 0
as R .
256

If p = 1 then
Z

R
1

h
iR
1
dx = ln x = ln R ln 1 as R .
1
x

Remark. If the logarithm is defined as an area function then it is required to prove

(using Riemann sums) that it is unbounded!

257

8.11

Comparison tests for improper integrals

Some important improper integrals such as

Z
2
ex dx

(probability theory, quantum mechanics) do not admit elementary antiderivatives

but one may still be able to discuss convergence/divergence by comparing them with
known improper integrals.
Theorem (Comparison test). Suppose that f and g are integrable functions and
that
0 f (x) g(x)
for x a.
Z
Z
g(x) dx converges then
(i) If

(ii) If

f (x) dx diverges then

f (x) dx converges.

g(x) dx diverges.

a
258

Sketch of proof. This follows from

Z
Z R
f (x) dx
0
a

g(x) dx
a

and the Least Upper Bound Axiom (MATH1241).

259

Example. Discuss the convergence of

Z
I=

x2

dx.

The integrand is even and hence

Z
Z 1
Z
2
2
I=2
ex dx = 2
ex dx + 2
0

On the other hand, we know that

x2

e >x
for x 6= 0 and that the p-integral
converges.

1
dx
x2

We therefore conclude that I converges.

260

x2

1
< 2
x

ex dx.

Note. One may show (2nd year) that I =

Example. Discuss the convergence of

Z
I=

1
dx.
ln x

261

Instead of using inequalities to estimate integrands, one often uses a dominant term
analysis such as

sin x + x2
f (x) =
2x4 1
behaves like
1
g(x) = 3
2x
for large x and hence one expects the convergence of the two associated improper
integrals to be the same.
The precise formulation of this idea is as follows:

262

Theorem (Limit form of the comparison test.) Suppose that f and g are
nonnegative and bounded on [a, ). If
f (x)
=L
lim
x g(x)

and 0 < L < then either

Z
both
f (x) dx and
a

or

both

f (x) dx and

Proof. See calculus notes.

263

g(x) dx converge

g(x) dx diverge.

Example. Discuss the convergence of

Z
sin x + x2
I=
dx.
4
2x 1
2
Set
f (x) =
Since
and the p-integral

sin x + x2
,
2x4 1
f (x) 1
=
lim
x g(x)
2
Z

1
dx
x3

converges, we conclude that I converges.

264

1
g(x) = 3 .
x

Example. Discuss the convergence of

Z

I=
( x x 3 ) dx.
3

265

266

Chapter 9
The logarithmic and exponential functions
In the preceding, we have manipulated functions such as
ln x,

ex ,

even though we have not defined them formally.

In particular, we are familiar with the important formula
ln(xy) = ln x + ln y.

267

Question. Consider the functional equation

f (st) = f (s) + f (t),

(9.1)

where s and t are independent variables. It is evident that f = ln constitutes one solution of this functional equation. Are there other functions f obeying this functional
equation?
Answer. We first note that (9.1) evaluated at s = t = 1 yields
f (1) = 0.
Moreover, differentiation with respect to s leads to
tf (st) = f (s)
so that

1
f (t) = f (1)
t

at s = 1.

If we now demand that f (1) = 1 then f is uniquely determined via integration since
f (1) = 0.
268

Conclusion. A function f is uniquely defined by the functional equation

f (st) = f (s) + f (t),
subject to
f (1) = 1.
It is given by
f (x) =

x
1

269

1
dt.
t

9.1

Powers and logarithms

See calculus notes.

9.2

The natural logarithm function

Definition. The natural logarithm function

1/t

ln : (0, ) R
is defined by the formula
ln x =

x
1

1
dt.
t

It is evident that ln x is simply the area of the shaded region shown below in the case
when x 1.

270

Theorem. The function ln : (0, ) R has the following properties:

(i) ln is differentiable on (0, ) and

1
d
ln x = .
dx
x

(ii) ln x > 0 for x > 1,

ln 1 = 0,
ln x < 0 for 0 < x < 1.
(iii) ln x as x 0+,
ln x as x .
(iv) For all x, y > 0:
(v) For all x, y > 0:
(vi) For all x > 0 and r Q:

ln(xy) = ln x + ln y.
 
x
ln
= ln(x) ln(y).
y
ln(xr ) = r ln x.
271

Proof.
(i) Apply the first fundamental theorem of calculus to the definition of ln.

(ii) This follows from the definition of ln and the fact that

1
t

> 0 when t > 0.

(iii) The diagram below shows that

Z

2
1

1
dt
1 ,
t
2

y=

1
t

4
2

1
dt
2 ,
t
4

8
4

8
272

1
dt
4 .
t
8

In general,
Z

as n .
Hence the improper integral

2n
1

1
dt 1 1
+ + +
t
2}
|2 2 {z
n terms
n
=
2

1
dt
t
1
diverges to infinity and, therefore, ln x as x .

This argument can be adapted to show that

Z 1
dt

t
n
2
as n . Hence ln x as x 0+.
273

(iv) Suppose that y is some fixed positive number and that x > 0. Then, the chain
rule implies that
d
1 d
y
1
d
[ln(xy)] =
(xy) =
= =
ln x.
dx
xy dx
xy x dx
Accordingly,
ln(xy) = ln(x) + C
for some constant C.
Evaluation at x = 1 leads to
ln(y) = C
and hence
ln(xy) = ln(x) + ln(y).
(v) Similar technique.
(vi) Similar technique.

274

Remark. The above properties imply that

Range(ln) = R, and

ln is increasing and hence invertible so that

ln x = 1 has a unique solution.

Definition. The real number e is defined to be the unique

number x satisfying
Z x
1
dt = 1.
1 t
ln x
1
d
1
Note.
ln x = 0 as x .
dx
x

275

9.3

The exponential function

Question. It is recalled that we have already defined xr for r Q. Can this definition be extended to r 6 Q?
Definition. The function
exp : R (0, )

is defined to be the inverse function of ln : (0, ) R.

y
(1, e)
b

y = exp x

1
b

(e, 1)
1
y = ln x
276

Remark. For any rational number r, we can evaluate both

exp r

and

er

but are these two numbers the same?

Theorem. The function exp : R (0, ) has the following properties:

(i) exp(ln x) = x for all x (0, ),
ln(exp x) = x for all x R.
(ii) exp(1) = e,
exp(0) = 1.
(iii) exp x as x ,
exp x 0 as x .
277

(iv) exp is differentiable on R with

d
exp x = exp x.
dx
(v) For all x, y R:

(vi) For all x > 0 and r Q:

exp(x + y) = exp x exp y.

exp(rx) = (exp x)r .

278

Proof.
(i) - (iii) Follows from the definition of exp.
(iv) The function exp is differentiable on R by virtue of the inverse function theorem
and differentiation of
ln(exp x) = x
produces
1 d
exp x = 1
exp x dx
as required.
(v) For any x and y, we have


exp(x + y) = exp ln(exp x) + ln(exp y)



= exp ln(exp x exp y)
= exp x exp y

as required.
279

(vi) Suppose that r is a rational number and x is a real number. Then,



exp(rx) = exp r ln exp x)



r
= exp ln (exp x)
= (exp x)r .

Remark. In particular, the above theorem implies that

exp(r) = (exp 1)r = er .
for every rational number r. It is therefore consistent to make the following definition.

280

Definition. For any x 6 Q, we define the number ex to be

ex = exp x.

Note. The above definition merely means that ex is the unique real number R such
that
Z R
1
dt = x
t
1
for any x R. By construction, the function
f : R R,

f (x) = exp x = ex

is differentiable (and continuous) and is called the exponential function.

281

9.4

Exponentials and logarithms with other bases

Question. How would one define bx for x 6 Q and b > 0?

Since, for any rational number r,
r

b = exp ln(b ) = exp r ln b = er ln b,

the following definition is natural.

Definition. Suppose that b > 0 and x 6 Q. Then, the number bx is defined by

bx = exp(x ln b) = ex ln b.

Note. By combining the defintions of bx for rational x and irrational x, we now

obtain a well-defined function
fb : R (0, ),

fb(x) = bx = exp(x ln b)

for any b > 0.

282

Example. It is seen that

f3(2) = 32 = exp(2 ln 3) = exp(ln 3) exp(ln 3) = 3 3 = 9
as one would expect!
Remark. Since fb is a combination of continuous and differentiable functions, it is
also continuous and differentiable with
fb(x) = (ln b) exp(x ln b) = (ln b)bx.
Accordingly,
if b > 1 then fb(x) > 0 for all x R,

if 0 < b < 1 then fb(x) < 0 for all x R

so that fb is invertible for b 6= 1.

283

Definition. Suppose that b is a positive real

number with b 6= 1. Then, the logarithm
function to the base b
logb : (0, ) R

is defined to be the inverse of the function

fb : R (0, ),

y = 3x
y = 2x

y = 8x
y

fb(x) = b = exp(x ln b).

y = log2 x
y = log3 x
y = log8 x
x

In particular, loge x = ln x.
Remark. The above definition may be stated as
y = bx

x = logb y.

The following theorem demonstrates that all logarithm functions are just scaled versions of the natural logarithm function ...
284

Theorem. Suppose that b is a positive real number with b 6= 1. Then

logb x =

ln x
ln b

for all x > 0.

Proof. Since
x = blogb x = exp(logb x ln b),
we conclude that
ln x = logb x ln b.
... and, hence, they share all the properties of ln such as


d ln x
1
d
log x =
=
dx b
dx ln b
x ln b
or
logb(xy ) = y logb x.
285

Remark. The above theorem implies that the class of logarithm functions represents
the general solution of the functional equation (9.1).

286

9.5

Integration and the ln function

Since

d
1
ln(x) =
dx
x
for x < 0, the function ln(x) is also an antiderivative of 1/x. Thus,
Z
1
dx = ln |x| + C
x
provided that x is restricted to an interval which does not contain 0.
Moreover, if a function f is positive and differentiable then

so that

f (x)
d
ln f (x) =
dx
f (x)
Z

f (x)
dx = ln f (x) + C.
f (x)
287

This may be generalised to

Z

f (x)
dx = ln |f (x)| + C
f (x)

provided that f is differentiable and does not vanish on the interval of integration.

Example. On any interval not including zeros of cos x, we have

Z
Z
sin x
tan x dx =
dx = ln | cos x| + C.
cos x
Example. Find the indefinite integral
Z

1
dx.
2 sec x + tan x

288

9.6

Logarithmic differentiation

Logarithms are powerful in that they transform powers into products, products into
sums and quotients into differences.
Example. Find the derivative of
 2
3/5
(3x + 4)(x + 2)
.
y=
x3 + 5x
The idea is to take ln of both sides of the equation to obtain
 2

3
(3x + 4)(x + 2)
ln y = ln
5
x3 + 5x


3
2
3
= ln(3x + 4) + ln(x + 2) ln(x + 5x) .
5
Differentiating both sides with respect to x is relatively easy and leads to


2
1 dy 3
1
3x + 5
6x
=
+

.
y dx 5 3x2 + 4 x + 2 x3 + 5x
289

Hence, we obtain
dy 3
=
dx 5

(3x + 4)(x + 2)
x3 + 5x

3/5 

6x
1
3x + 5
+

.
2
3
3x + 4 x + 2 x + 5x

Remark. The above procedure is only valid for intervals on which y > 0.

Example. Consider the function

f : (0, ) R,
Determine its derivative.

290

f (x) = xsin x.

9.7

Indeterminate forms with powers

Example. Find
L = lim xsin x.
x0+

Note that the above limit is of the indeterminate form 00.

Since the logarithm function is continuous, we conclude that




sin x
sin x
.
= lim ln x
ln L = ln lim x
x0+

x0+

Now,

lim | ln x

sin x

x0+

| = lim | sin x ln x|
x0+

lim |x ln x|
x0+

=0
so that
ln L = 0

291

L = 1.

The following example is of the indeterminate form 1.

Example. Show that
lim

t
1+
x

where t is a constant real parameter.

292

x

= et ,

Chapter 10
The hyperbolic functions
10.1

Hyperbolic sine and cosine functions

Definition. The hyperbolic cosine function cosh : R R is defined by

1 x
cosh x = (e + ex).
2
Definition. The hyperbolic sine function sinh : R R is defined by
1
sinh x = (ex ex).
2
293

Note. cosh and sinh are differentiable with

d
d
(sinh x) = cosh x,
(cosh x) = sinh x
dx
dx
so that cosh x and sinh x obey the differential equation
d2 y
= y.
2
dx
y

Properties of the cosh function.

cosh is an even function.

cosh 0 = 1.

cosh is decreasing on (, 0), stationary at 0

and increasing on (0, ).
cosh x 1 for all x in R.

cosh x gets arbitrarily close to 21 ex as x .

294

y = cosh x
y = 21 ex
x

Properties of the sinh function.

sinh is an odd function.

sinh 0 = 0.

sinh is increasing on (, ) and has a point

of inflexion at 0.
sinh x < 0 for x < 0 and sinh x > 0 for x > 0.
sinh x gets arbitrarily close to 21 ex as
x .

295

y = 12 ex

y = sinh x
x

Theorem. The hyperbolic functions are related by

cosh2 x sinh2 x = 1.
Proof. By definition,
cosh2 x sinh2 x =

ex + ex
2

2

ex ex
2

2


1  2x
2x
2x
2x
= (e + 2 + e ) (e 2 + e )
4
= 1.

Remark. The similarity to relations such as

d
d
cos x = sin x,
sin x = cos x
dx
dx
explains the words cosine and sine in the hyperbolic functions.
cos2 x + sin2 x = 1,

296

The term hyperbolic is motivated in the following manner:

Example. Sketch the curve (t) defined by


(t) = x(t), y(t) = (cosh t, sinh t).
Elimination of the parameter t leads to
[x(t)]2 [y(t)]2 = cosh2 t sinh2 t = 1

so that parametrises the branch of the hyperbola

x2 y 2 = 1,

x > 0.

The other branch of the hyperbola is parametrised by

x(t), y(t) = ( cosh t, sinh t).
297

10.2

Other hyperbolic functions

Other hyperbolic functions are defined in analogy with the trigonometric functions
according to

sinh x
tanh x =
,
cosh x
1
sech x =
,
cosh x

cosh x
coth x =
,
sinh x
1
cosech x =
.
sinh x

298

Recall that

sinh x ex ex
tanh x =
= x
cosh x e + ex

Properties of the tanh function.

tanh is an odd function.
tanh 0 = 0.

tanh is increasing on (, ) and has a

point of inflexion at 0.
tanh x < 0 for x < 0 and tanh x > 0 for
x > 0.

lim tanh x = 1.
x

tanh x = sech2 x > 0.

dx

299

tanh x
1
x
1

Proof of the derivative of tanh. By definition,



d
d sinh x
(tanh x) =
dx
dx cosh x
d
d
sinh x sinh x
cosh x
cosh x
dx
dx
=
cosh2 x
cosh2 x sinh2 x
=
cosh2 x
1
=
cosh2 x
= sech2 x.

Note. The slope at the point of inflexion is



d
tanh x
= sech2 0 = 1.
dx
x=0

300

10.3

Hyperbolic identities

Difference of squares identities.

cosh2 x sinh2 x = 1

1 tanh2 x = sech2 x

coth2 x 1 = cosech2 x

Sum and difference formulae.

sinh(x y) = sinh x cosh y cosh x sinh y

cosh(x y) = cosh x cosh y sinh x sinh y

tanh(x y) =

tanh x tanh y
1 tanh x tanh y

301

Double-angle formulae.
sinh(2x) = 2 sinh x cosh x
cosh(2x) = cosh2 x + sinh2 x
2 tanh x
tanh(2x) =
2 .
1 + tanh x

Exercise. Prove the first two sum and difference formulae and, hence, derive the
third.

302

10.4

Hyperbolic derivatives and integrals

The following derivatives may be readily verified:

d
sinh x = cosh x,
dx

d
cosh x = sinh x
dx

d
d
tanh x = sech2 x,
coth x = cosech2 x
dx
dx
d
sech x = sech x tanh x
dx
d
cosech x = cosech x coth x.
dx
Corresponding indefinite integrals are, for instance,
Z
Z
sinh x dx = cosh x + C,
sech2 x dx = tanh x + C.
303

Example. Determine the definite integral

Z (ln 2)2
2
sech x

I=
dx.
x
0

304

10.5

The inverse hyperbolic functions

The hyperbolic sine and tan functions are increasing and therefore invertible. In the
case of the hyperbolic cosine function, one usually restricts the domain to [0, ) to
guarantee invertibility.
y

1
x

sinh1

cosh1

305

tanh1

Definition. The functions

cosh1 : [1, ) [0, )

sinh1 : R R

tanh1 : (1, 1) R

are defined to be the inverse functions of the (restricted) hyperbolic functions

cosh : [0, ) [1, )

sinh : R R

tanh : R (1, 1).

306

Example. Evaluate
sinh

cosh1 34

Evaluation of
sinh2 t = cosh2 t 1

at t = cosh1 34 yields
sinh

cosh1 34

Accordingly,
sinh

cosh1 34

since t > 0.

307

4 2
3

1 = 97 .

=+

7
3

Remark. Note that

cosh1(cosh x) = x
for x 0 only!
Example. Simplify
cosh

x
1
+
,
2 2x

308

x > 0.

The relation

ex ex
y = sinh x =
2

may be written as
so that

(ex)2 2yex 1 = 0
4y 2 + 4
e =
2


p
x = ln y + y 2 + 1 .
x

since ex > 0. Hence,

2y +

In a similar manner, the remaining identities of the following theorem may be proven.

309

Theorem. The following identities hold:




1
sinh x = ln x + x2 + 1


p
cosh1 x = ln x + x2 1


1
1+x
tanh1 x = ln
.
2
1x
The derivatives of these functions are therefore given by
1
d
sinh1 x =
dx
x2 + 1
1
d
1
cosh x =
dx
x2 1
d
1
tanh1 x =
.
2
dx
1x
310

Example. Use the inverse function theorem to confirm that

d
1
1
.
sinh x =
2
dx
x +1

If we set y = sinh1 x then the inverse function theorem implies that

dy
1
1
.
=
=
dx cosh y cosh(sinh1 x)
On the other hand, since
cosh t =
we obtain
cosh(sinh
so that

1 + sinh2 t,

x) =

1 + x2

dy
1

.
=
2
dx
x +1

311

10.6

Integration leading to the inverse hyperbolic functions

From the previous considerations, it follows that

Z

dx
x
= sinh1 + C
x2 + a2

 a

= ln x + x2 + a2 + C,

dx
1 x

+C
= cosh
2
2
a
x a


p

= ln x + x2 a2 + C,

312

a>0

xa>0

1 x

Z
+ C, |x| < a
tanh
dx
a
a
=
x
1
a2 x2

coth1 + C, |x| > a > 0

a
a



1  a + x 
ln 
=
+ C,

2a
ax

x2 6= a2.

[These formulae are included in the table of standard integrals which is issued at the
final examination.]

313

Example. Determine the indefinite integral

Z
dx

.
2
x 2x + 10
If we complete the square, that is,
x2 2x + 10 = (x 1)2 + 9 = u2 + 32,
where u = x 1, then
Z

du
u2 + 3 2
1 u
+C
= sinh
3
x1
= sinh1
+ C.
3

dx

=
2
x 2x + 10

314