Engineering Mathematics

Part I
2023-2024

School of Physics, Engineering & Technology

University of York

[ELE00038C] J. J. Bissell
 3

Preface
The purpose of this course is to introduce those mathematical tools that are
most relevant to solving practical problems in engineering and the physical
sciences. Where feasible, core concepts will be motivated using ‘real-life’ con-
texts; however, to develop competence and fluency we will often need to ‘drill’
technique using examples that are in themselves somewhat abstract.

Although our main goal might be practical application, it is important to under-

stand that mathematics underpins any serious attempt to describe the world,
and provides unparalleled insights into its beauty. So let us enjoy it!

Course delivery

Delivery of first year mathematics (ELE00038C) will be a mixture of in-person

lectures, and independent study based on a comprehensive set of course notes.
Material in these notes should be followed according to the weekly schedule
posted on the Virtual Learning Environment (VLE) - you will need to check
this schedule regularly.

Mathematics is (and always has been) a subject that is best learned by doing,
and our emphasis will be on active participation through problem solving. This
approach will require more effort, but it should also be more effective, and
ultimately more rewarding.

Lectures

Your timetable should include four slots each week. Although these are listed
as ‘lectures’, they will—as far as possible—be run more like workshops, with
lots of examples and exercises, with a basic structure as follows:

1. A short introduction to a new mathematical topic or method.

2. Active problem solving, with opportunities to ask questions.

3. Repeat.

In this way I hope to maximise our overall contact time, and help to set the
pace of learning. Remember to bring pencil and paper with you.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

4

Workshops

In addition to the lecture sessions, I will be holding a weekly workshop (as per
your timetable). These workshops will mainly involve practicing techniques by
solving problems on the course exercise sheets, but they can also be used as
an opportunity for you to ask me further questions about course material.

In previous years students have also set-up and used their own online forum to
discuss course content (e.g., via Discord). I encourage this kind of collaborative
engagement; however, remember that you are all coming from different back-
grounds, with different life experiences, so please be respectful of one-another.

Exercise sheets

Almost all of the techniques that we will be studying in this course can be mas-
tered by practice, and it is essential that you work on course exercise sheets
throughout term. Our ultimate goal is skilled creative work in applied con-
texts; however, to reach that point we must first hone our craft, and this will
often mean working on more abstract problems. As a rule-of-thumb, keep
mathematically fit by solving at least 10 exercise problems each week.

How to use these notes

These notes have been written as an independent guide, and (in principle)
you should be able to learn the course content by studying each section, and
completing the worked examples, and course exercises. Reading mathematics
requires active participation, so keep pencil and paper to hand to check the
steps in an argument. For example, a mathematical statement like

x2 + 3x + 2
f (x) = , (x 6= −2) ⇒ f (x) = x + 1
x+2
might not seem obvious; however, if you have a pencil and paper to hand, then
you can confirm that it is correct by noting that

x2 + 3x + 2 (x + 1)(x + 2)
f (x) = = = x + 1. (0.1)
x+2 (x + 2)

In general, therefore, reading these notes will require application - do not be

surprised if you need to deduce some of the steps in an argument by yourself.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

 5

Assessment

The course will be assessed by two components with the following weightings:

• 20% continuous assessment by fortnightly multiple choice quizzes (MCQs),

with five quizzes in total, each worth 4%:
– Week 02 : Quiz 01 (Differentiation).
– Week 04 : Quiz 02 (Integration).
– Week 06 : Quiz 03 (Complex Numbers).
– Week 08 : Quiz 04 (Ordinary Differential Equations).
– Week 10 : Quiz 05 (Vectors and Vector Algebra).

• 80% end of course assessment by a closed book, written examination.

Further information about the format and timings of the multiple choice quizzes
is summarised on the Virtual Learning Environment.

Will it be on the exam.?

Unless explicitly marked ‘(optional)’ any section in these notes could be ex-
amined. However, it would not be a very fair examination if I were to set very
difficult problems, or ask for long, and technical derivations. Clearly I cannot
tell you what will be on the exam., but I can tell you that I will try to be fair.
The best guide to the sorts of questions you might be asked in an examination
is to look at past papers (see the course wiki-pages). If you can solve most of
the course exercises, then you will do very well in the exam. - so practice!

Further reading

The following texts are recommended as supplementary reading:

• J. Gilbert and C. Jordan, Guide to Mathematical Methods, 2nd Edition, Palgrave

Macmillan (2002).
• E. Kreyszig, Advanced Engineering Mathematics, 7th Edition, Wiley (1993).
• K. F. Riley, M. P. Hobson, and S. J. Bence, Mathematical Methods for Physics and
Engineering, 2nd Edition, Cambridge University Press (2002).

The simplest of these is probably Gilbert and Jordan, but the other two texts
will also be useful for second year mathematics and beyond.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

6

Rules of Maths Club

1. The first rule of Maths Club is: you do to talk about Maths Club.

2. The second rule of Maths Club is: you do not talk about Maths Club.

3. Third rule of Maths Club: If someone yells “stop!”, goes limp, taps out,
or fails to draw a diagram, then the maths is over.

4. Fourth rule: All angles are in radians.

5. Fifth rule: One line of mathematics at a time people, no skipping steps.

6. Sixth rule: The maths is bare knuckle. No shirt, no shoes, no calculators.

7. Seventh rule: The maths will go on as long as it has to.

8. The eighth and final rule: If this is your first time at Maths Club, then
you have to math.

Why we are here

It has always taken a certain amount of skill to navigate university life, but
things have arguably become more difficult over recent years. Let us try to
remember, then, that fundamentally we are here as a community of scholars to
engage joyfully with the process of learning and discovery. So let’s treat each
other with goodwill, and give it our best shot.

I look forward to seeing how we get on.

J. J. Bissell
(Summer 2023)

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Contents

1 Functions and Algebra 13

1.1 Sets, Numbers, and Intervals . . . . . . . . . . . . . . . . . . 14

1.1.1 Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.1.2 Intervals . . . . . . . . . . . . . . . . . . . . . . . . . 17

1.2 Definition of a Function . . . . . . . . . . . . . . . . . . . . . 18

1.3 Graphing Functions . . . . . . . . . . . . . . . . . . . . . . . 19

1.4 Polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

1.4.1 Straight lines . . . . . . . . . . . . . . . . . . . . . . . 21

1.4.2 Quadratics . . . . . . . . . . . . . . . . . . . . . . . . 21

1.4.3 Polynomials of arbitrary degree . . . . . . . . . . . . . 24

1.4.4 Roots and factors . . . . . . . . . . . . . . . . . . . . 25

1.4.5 Algebraic division . . . . . . . . . . . . . . . . . . . . 27

1.4.6 Remainder theorem . . . . . . . . . . . . . . . . . . . 28

1.4.7 Comparing coefficients . . . . . . . . . . . . . . . . . . 29

1.4.8 Rational functions . . . . . . . . . . . . . . . . . . . . 30

1.5 Partial Fractions . . . . . . . . . . . . . . . . . . . . . . . . . 30

1.6 Trigonometric Functions . . . . . . . . . . . . . . . . . . . . . 33

1.6.1 Pythagoras . . . . . . . . . . . . . . . . . . . . . . . . 35

1.6.2 Compound angle formulae . . . . . . . . . . . . . . . . 36

7
8 CONTENTS

1.6.3 Identities . . . . . . . . . . . . . . . . . . . . . . . . . 38

1.7 Composite Functions and Inverse Fucntions . . . . . . . . . . . 40

1.7.1 Composite functions . . . . . . . . . . . . . . . . . . . 40

1.7.2 Inverse functions . . . . . . . . . . . . . . . . . . . . . 41

1.8 Odd and even functions . . . . . . . . . . . . . . . . . . . . . 43

1.9 Exponentials and Logarithms . . . . . . . . . . . . . . . . . . 44

1.9.1 Laws of logarithms . . . . . . . . . . . . . . . . . . . . 46

1.9.2 Orders of magnitude . . . . . . . . . . . . . . . . . . . 47

1.9.3 Logarithmic plots . . . . . . . . . . . . . . . . . . . . 48

1.9.4 Natural logarithm . . . . . . . . . . . . . . . . . . . . 50

1.10 Exponential Function and Hyperbolic Functions . . . . . . . . . 50

1.11 Explicit and Implicit Functions . . . . . . . . . . . . . . . . . . 53

1.12 Parametric Equations . . . . . . . . . . . . . . . . . . . . . . 54

1.13 Plane Polar Coordinates . . . . . . . . . . . . . . . . . . . . . 56

2 Differentiation 59

2.1 Limits and Continuity . . . . . . . . . . . . . . . . . . . . . . 60

2.1.1 Limits . . . . . . . . . . . . . . . . . . . . . . . . . . 60

2.1.2 Continuity . . . . . . . . . . . . . . . . . . . . . . . . 63

2.1.3 Rules for evaluating limits . . . . . . . . . . . . . . . . 64

2.1.4 Asymptotic limits (optional) . . . . . . . . . . . . . . . 65

2.2 Foundations of Differentiation . . . . . . . . . . . . . . . . . . 66

2.2.1 Differentiation from first principles . . . . . . . . . . . 66

2.2.2 Derivatives of standard functions . . . . . . . . . . . . 70

2.2.3 Gradients and tangents . . . . . . . . . . . . . . . . . 71

2.2.4 Differential element . . . . . . . . . . . . . . . . . . . 73

2.3 Rules for Differentiation . . . . . . . . . . . . . . . . . . . . . 73

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

CONTENTS 9

2.3.1 Linearity of differentiation . . . . . . . . . . . . . . . . 73

2.3.2 Chain rule . . . . . . . . . . . . . . . . . . . . . . . . 74

2.3.3 Products and quotients . . . . . . . . . . . . . . . . . 76

2.3.4 Implicit differentiation . . . . . . . . . . . . . . . . . . 78

2.3.5 Inverse function rule . . . . . . . . . . . . . . . . . . . 79

2.3.6 Parametric differentiation . . . . . . . . . . . . . . . . 80

2.4 Higher Order Derivatives . . . . . . . . . . . . . . . . . . . . . 81

2.4.1 Second-order derivative . . . . . . . . . . . . . . . . . 81

2.4.2 Newton’s notation . . . . . . . . . . . . . . . . . . . . 82

2.4.3 Derivatives of arbitrary order . . . . . . . . . . . . . . 83

2.4.4 Stationary points . . . . . . . . . . . . . . . . . . . . . 84

2.5 Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

2.5.1 Maclaurin series . . . . . . . . . . . . . . . . . . . . . 88

2.5.2 Binomial series . . . . . . . . . . . . . . . . . . . . . . 92

2.5.3 Convergence . . . . . . . . . . . . . . . . . . . . . . . 93

2.5.4 Taylor series . . . . . . . . . . . . . . . . . . . . . . . 95

2.5.5 Linear approximation by Taylor series (optional) . . . . 96

2.6 L’Hôpital’s Rule . . . . . . . . . . . . . . . . . . . . . . . . . 97

3 Integration 101

3.1 Motivating Integration . . . . . . . . . . . . . . . . . . . . . . 102

3.1.1 Integration to find the area under a curve . . . . . . . . 102

3.1.2 Integration to find the mass of a solid . . . . . . . . . . 104

3.1.3 Setting up integrals informally . . . . . . . . . . . . . . 105

3.1.4 Integration to find the distance travelled by an object . 106

3.2 Evaluating Integrals . . . . . . . . . . . . . . . . . . . . . . . 107

3.2.1 Integration by first principles (optional) . . . . . . . . . 107

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

10 CONTENTS

3.2.2 Fundamental theorem of calculus . . . . . . . . . . . . 109

3.2.3 Integration by inspection . . . . . . . . . . . . . . . . . 111

3.3 Methods of Integration . . . . . . . . . . . . . . . . . . . . . . 115

3.3.1 Integration by substitution . . . . . . . . . . . . . . . . 115

3.3.2 Integration using trigonometric identities . . . . . . . . 117

3.3.3 Checking indefinite integrals . . . . . . . . . . . . . . . 118

3.3.4 Logarithmic integration . . . . . . . . . . . . . . . . . 118

3.3.5 Integration by completing the square . . . . . . . . . . 119

3.3.6 Integration by partial fractions . . . . . . . . . . . . . . 120

3.3.7 Tangent half-angle substitution . . . . . . . . . . . . . 121

3.3.8 Integration by parts . . . . . . . . . . . . . . . . . . . 124

3.4 Area and Symmetry . . . . . . . . . . . . . . . . . . . . . . . 127

3.5 Improper Integrals . . . . . . . . . . . . . . . . . . . . . . . . 129

3.6 Mean Value of a Function . . . . . . . . . . . . . . . . . . . . 130

4 Complex Numbers 133

4.1 What is a Complex Number? . . . . . . . . . . . . . . . . . . 134

4.1.1 Cartesian form . . . . . . . . . . . . . . . . . . . . . . 135

4.1.2 Complex plane and Argand diagrams . . . . . . . . . . 136

4.2 Algebra of Complex Numbers . . . . . . . . . . . . . . . . . . 137

4.2.1 Addition and subtraction . . . . . . . . . . . . . . . . 137

4.2.2 Complex numbers as vectors . . . . . . . . . . . . . . . 137

4.2.3 Multiplication . . . . . . . . . . . . . . . . . . . . . . 138

4.2.4 Equating real and imaginary parts . . . . . . . . . . . . 139

4.2.5 Complex conjugate . . . . . . . . . . . . . . . . . . . . 140

4.2.6 Modulus . . . . . . . . . . . . . . . . . . . . . . . . . 142

4.2.7 Division and quotients . . . . . . . . . . . . . . . . . . 143

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

CONTENTS 11

4.2.8 Square roots . . . . . . . . . . . . . . . . . . . . . . . 145

4.3 Polar Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

4.4 Exponential Form . . . . . . . . . . . . . . . . . . . . . . . . 149

4.4.1 Euler’s formula . . . . . . . . . . . . . . . . . . . . . . 149

4.4.2 Complex exponential . . . . . . . . . . . . . . . . . . . 150

4.4.3 Exponential form of a complex number . . . . . . . . . 151

4.4.4 Multiplication and division in exponential form . . . . . 151

4.4.5 Multiplication by unity . . . . . . . . . . . . . . . . . . 153

4.5 Computing general roots . . . . . . . . . . . . . . . . . . . . . 153

4.5.1 Cubed roots of unity . . . . . . . . . . . . . . . . . . . 153

4.5.2 Roots of arbitrary complex numbers . . . . . . . . . . . 155

4.6 Solving Polynomial Equations . . . . . . . . . . . . . . . . . . 157

4.7 Complex Functions . . . . . . . . . . . . . . . . . . . . . . . . 160

4.7.1 Complex trigonometric functions . . . . . . . . . . . . 161

4.7.2 De Moivre’s Theorem and Trigonometric Identities . . . 162

4.7.3 Logarithms . . . . . . . . . . . . . . . . . . . . . . . . 163

4.7.4 Powers . . . . . . . . . . . . . . . . . . . . . . . . . . 164

4.8 Oscillatory Systems . . . . . . . . . . . . . . . . . . . . . . . 165

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

12 CONTENTS

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 Engineering Mathematics
J. J. Bissell

Functions and Algebra

An essential task in engineering and the physical sciences is to answer questions

of the form “what happens to quantity y if quantity x changes?”. In mathe-
matical language questions of this kind are posed in terms of functions, i.e.,
expressions that describe a relationship between two or more variables. From
an engineering perspective these variables may be the inputs and outputs of
a system. For example, Ohm’s law V = IR expresses a relationship between
the voltage V , current I, and resistance R in a circuit. Mathematically we say
that V (I, R) is a function of two variables I and R (the arguments of V ).
The purpose of this section is to review some of the most important functions
for applications in engineering and the physical sciences. Many of the topics
in this section are covered by school mathematics; indeed, some of the more
elementary concepts are included only briefly for completeness, and will be left
as revision exercises. By the end of the section you should be able to. . .
Learning outcomes:
◦ Use open and closed intervals to define sets of numbers.
◦ Use Cartesian coordinates to represent functions as graphs.
◦ Find the intersection of the graph of a function with coordinate axis.
◦ Determine the roots of simple polynomials.
◦ Apply algebraic division to divide one polynomial by another.
◦ Simplify rational functions using the method of partial fractions.
◦ Use identities to manipulate expressions involving trigonometric functions.
◦ State whether a function is odd or even.
◦ Determine the inverse function of simple functions.
◦ Manipulate expressions involving logarithms and exponents.
◦ Express functions in parametric form.
These learning outcomes must be reinforced by completing course exercises.

13
14 Section 1 : Functions and Algebra

1.1 Sets, Numbers, and Intervals

The most important functions in engineering and the physical sciences are func-
tions of numbers, and to discuss how they work we must first define the basic
terminology of numbers. Here we review sets of numbers (§1.1.1), especially
the real numbers, and subsets of real numbers defined by intervals (§1.1.2).

1.1.1 Sets

A set is the mathematical name for a collection of objects known as elements.

Sets can be used to describe aspects of the world, including abstract mathe-
matical objects. For example, the colours red, white, and blue are all colours
used in the union flag of the United Kingdom; in this way we can say that the
set of colours in the union flag contains the elements red, white, and blue.
In mathematical contexts a set is often denoted using upper-case, while the
elements of a set are denoted using lower case. In particular, we can use ‘curly
bracket’ notation to describe the elements a1 , a2 , . . . , an of a set A, thus

A = {a1 , a2 , . . . , an }. (1.1)

This expression may be read as “A is the set containing the elements a1 , a2 ,

etc. all the way up to the element an ”. Here the set A has n elements,
where n is a finite number; in this way, the ‘three-dot’ notation ‘. . . ’ is used
as shorthand to mean “all the elements that have not been written explicitly”.
A set with an infinite number of elements, say B, can be denoted

B = {b1 , b2 , . . . }, (1.2)

where b1 , b2 , etc., are the elements of B.

The numbers 1, 2, 3, etc., used for counting form a particularly important set
known as the natural numbers; this set is denoted using the symbol N, i.e.,

N = {1, 2, 3, . . . }. (1.3)

We can extend the natural numbers to include the number 0, and the negative
numbers −1, −2, −3 etc.; this new set is known as the set of integers, and
denoted Z, thus

Z = {. . . , −3, −2, −1, 0, 1, 2, 3, . . . }. (1.4)

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 15

If some object x is an element of a set A, then we write

x∈A (shorthand for ‘x is a member of the set A’). (1.5)

For example, we may write 339648 ∈ N (‘the number 339648 is a member of

the set of natural numbers’).

If every element of a set B also happens to be a member of the set A, then

we say that B is a subset of A, and write

B⊂A (shorthand for ‘B is a subset of A’). (1.6)

For example, every natural number is also an integer; this means that

N ⊂ Z, (1.7)

i.e., the natural numbers are a subset of the integers.

Two other important sets are the rational numbers, and the real numbers:

• The set of rational numbers is denoted Q; these are the numbers formed
by quotients of integers and natural numbers, e.g., 12 , and − 53 .

• The set of real numbers is denoted R; these are the numbers that can
√
be approximated by rationals to arbitrary precision, e.g., π, e, and 2.

Real numbers R are essentially those numbers that can be expressed in decimal
form, irrespective of whether such representation requires an infinite number of
digits. Hence, the rational numbers Q, the integers Z, and the natural numbers
N are all subsets of the real numbers R. Example real numbers include:

1= 1.0 (a natural number N) (1.8a)

0= 0.0 (an integer Z) (1.8b)
−2 = −2.0 (a negative integer Z) (1.8c)
− 23 = −1.5 (a rational Q, and finite decimal) (1.8d)
20
11
= 1.81 (a rational Q, and repeating decimal) (1.8e)
π≈ 3.14159296535 . . . (an irrational number) (1.8f)

(see figure 1.1). A real number R that is not a rational number Q is known as
√
an irrational number, e.g., 2, π, and e.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

16 Section 1 : Functions and Algebra

Figure 1.1: Real numbers represented as points on the real line R.

The set R is often referred to as the real line since any real number x ∈ R
can be represented as a point on a straight line. The reference point on the
real line is the number zero, with positive numbers (x > 0) to the right, and
negative numbers to the left (x < 0), as depicted in figure 1.1.

The relationship between the natural numbers N, the integers Z, the rational
numbers Q, and the real numbers R can be expressed using subset notation as

N ⊂ Z ⊂ Q ⊂ R. (1.9)

Historically speaking it has been necessary to introduce ever more general sets
of numbers to solve ever more general problems. To this end we introduce a
further set of numbers called the complex numbers C in Section 4.

Example 1.1 Let A be the set defined by A = {1, 2, −3, 8}. Which elements
of A are less than 2?

I Solution: The elements of A are 1, 2, −3, and 8. Of these elements, only 1

and −3 are less than 2. J

Example 1.2 Let A be the set defined by A = {1, 2, −3, 8}. Which of the
following statements are true?

(a) The number −3 is a member of the set A.

(b) The elements of A are natural numbers.
(c) The elements of A are a subset of the integers.
(d) If x ∈ A, then x < 10.

I Solution: Considering each of these statements in turn we have

(a) This statement is true; in set notation we may write e number −3 ∈ A.

(b) This statement is false. The natural numbers N are all positive, since
−3 < 0, the elements of A = {1, 2, −3, 8} are not all natural numbers.
(c) This statement is true, since all the elements of A are integers.
(d) Every element of A is less than 10, so the statement is true.

Note: we may express the fact that A is a subset of the integers as A ⊂ Z. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 17

1.1.2 Intervals

A convenient method for defining subsets of real numbers R is to use a subset of

the real line known as an interval. An interval is defined such that it contains
all the numbers between two endpoints. For example, the interval

−1≤x≤6 (1.10)

is the set of all the real numbers x ∈ R that are greater than or equal to −1,
and less than or equal to 6, where −1 and 6 are the intervals endpoints.
An interval that includes its endpoints is known as a closed interval. In
contrast, an interval that excludes its endpoints is known as an open interval.

Definition 1.1 Let a and b be real numbers, with a ≤ b then we may

define the following kinds of interval using bracket notation:

• The closed interval [a, b] is the set of numbers x ∈ R that satisfy

a ≤ x ≤ b. (1.11)

• The open interval (a, b) is the set of numbers x ∈ R that satisfy

a < x < b. (1.12)

A half-open interval includes only one of its endpoints, and is denoted

using mixed parentheses. In particular:

• The right closed interval (a, b] is the set of numbers satisfying

a < x ≤ b. (1.13)

• The left closed interval [a, b) is the set of numbers satisfying

a ≤ x < b. (1.14)

Intervals without endpoints can be denoted using the symbol ∞ for infinity;
e.g., the interval [0, ∞) means the set of numbers x that satisfy 0 ≤ x.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

18 Section 1 : Functions and Algebra

Example 1.3 Express the following intervals using bracket notation:

(a) − 2 < x ≤ π; (b) − 1 ≤ x ≤ 9; (c) x < −3. (1.15)

I Solution: The intervals are: (a) (−2, π]; (b) [−1, 9]; and (c) (−∞, −3). J

1.2 Definition of a Function

We began this section by describing a function as a mathematical object that

expresses a relationship between two or more variables. Such a relationship can
sometimes be stated using mathematical formula, but this is not always the
case. Some examples of functions include the following:

• If a current I is passed across a resistor of resistance R, then the potential

V across the resistor satisfies the equation V = IR. For any given current
I the voltage V is well defined, and we say that V (I) is a function of I.

• The surface area S of a sphere of radius r > 0 is given by the equation

A = 4πr2 , where r > 0 is positive. For any given radius r, the area S is
well defined, and we say that S(r) is a function of r.

• The mass M of carbon dioxide emitted by the United Kingdom each year
Y since 1950 can be recorded in a table, where Y > 1. For each year Y
the mass M is well defined, and we say that M (Y ) is a function of Y
(even though there is no obvious formula for computing M given Y ).

In the context of engineering and the physical sciences, a function f can be

thought of as a mathematical machine that takes an input x, and returns an
output y, as illustrated schematically in figure 1.2. The value y returned by
the function depends on the value x passed to the function, so we say that
“y is a function of x”, and write this as y = f (x).

Figure 1.2: Analogy between a function f , and a machine with inputs and outputs.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 19

We may define a function more formally as follows:

Definition 1.2 A function f is a rule that defines a relationship between

each element x of a set A with a single element y of another set B.

• A is called the domain of f , and we say that f is defined on A.

• B is called the codomain of f .

• We say that f maps A to B, and write this as f : A → B.

The elements x ∈ A and y ∈ B are known as variables:

• x is known as the independent variable, or argument of f .

• y is known as the dependent variable, or value of f at x.

In particular, we say that y is a function of x, and write y = f (x). This

dependence is often also expressed by writing y = y(x).

1.3 Graphing Functions

We are typically concerned with functions whose variables x and y are real
numbers, i.e., x, y ∈ R. Such a function f (x) returns a unique value y = f (x)
for each x in its domain, and can therefore be visualised as a set of points
(x, y) plotted on perpendicular x and y-axis known as the Cartesian plane.
As an example, consider the function f (x) defined by the rule

1
f (x) = , for x 6= ±1, (1.16)
x2 −1

which is plotted in figure 1.3. Here the domain of f (x) is not stated explicitly,
but by convention we can say that the domain is the set of numbers x ∈ R
for which the rule applies. Since division by 0 is not defined, the function f (x)
is undefined when (x2 − 1) = 0, i.e., when x = ±1. Indeed, because the
magnitude of f (x) becomes infinitely large as x approaches either x = 1 or
x = −1, we say that f (x) diverges at x = ±1.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

20 Section 1 : Functions and Algebra

4
3
2
1
0
-1
-2
-3
-4
-3 -2 -1 0 1 2 3

Figure 1.3: Function y = f (x) = 1/(x2 − 1) graphed (plotted) on Cartesian axis.

p √
Example 1.4 The functions f (x) = 2x − 1 and g(x) = f (x) = 2x − 1
are sketched in figure 1.4. Determine the domains of (i) f (x) and (ii) g(x).

I Solution: In each case we take the domain of the function to be the set of
values of x ∈ R for which the rule of the function applies.

(i) The function f (x) = 2x − 1 can be evaluated for all values of x ∈ R, so

the domain of f (x) is R.
√
(ii) The square root operation . . . only applies to non-negative numbers, so
√
the function g(x) = 2x − 1 is only defined when 2x − 1 ≥ 0, that is, x ≥ 12 .
We therefore write
√
g(x) = 2x − 1, for x ≥ 12 , (1.17)

where the domain of g(x) is the interval [ 21 , ∞). J

2 3

1 2

0 1

-1 0

-2 -1
-1 0 1 2 3 -1 0 1 2 3
(a) (b)

p √
Figure 1.4: Two functions: (a) f (x) = 2x − 1; and (b) g(x) = f (x) = 2x − 1.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 21

1.4 Polynomials

Polynomials are a particularly important class of functions that involve terms

in integer powers of x; these functions include straight lines, and quadratics.

1.4.1 Straight lines

Perhaps the simplest function involving real numbers is

f (x) = c, (1.18)

where c ∈ R is a constant. When graphed, this function yields a horizontal

straight line y = f (x) = c. Note that y is still said to be a function of x, even
though it always takes the same value y(x) = c.
The general equation for a straight line is given by the function

y = f (x) = bx + c, (1.19)

where b, c ∈ R are both constants. The line intersects the y-axis when x = 0,
that is, when y = c; the constant c is therefore known as the y-intercept.
The constant b determines how ‘steep’ the line is, and is called the gradient.

Example 1.5 Determine where f (x) = 2x − 1 intersects the x and y-axis.

I Solution: The function y = f (x) = 2x − 1 is sketched in figure 1.4. This

function intersects the y-axis when x = 0, that is, when

y = f (0) = −1; (1.20)

the y-inctercept is thus (0, −1). The intersection with x-axis occurs when
y = 0, i.e., when x = 12 . The intersection with the x-axis is thus ( 12 , 0). J

1.4.2 Quadratics

The general form of the quadratic function is given by

f (x) = ax2 + bx + c, (1.21)

where a, b, c ∈ R are constants, with a 6= 0. Two example quadratic functions

f (x) = x2 − x − 2 and g(x) = 4x2 + 12x + 9 are plotted in figure 1.5.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

22 Section 1 : Functions and Algebra

10 10
8 8
6 6
4 4
2 2
0 0
-2 -2
-4 -4
-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
(a) (b)

Figure 1.5: Two quadratics: (a) f (x) = x2 − x − 2; and (b) g(x) = 4x2 + 12x + 9.

The roots of a quadratic are those values of x for which f (x) = 0; in this way
the roots of a quadratic are the solutions to the quadratic equation

f (x) = ax2 + bx + c = 0. (1.22)

Quadratic equations may be solved either by factorising, or completing the

square. Indeed, in general it may be shown that the solutions are given by
√
−b ± b2 − 4ac
x = x± = . (1.23)
2a
This equation is known as the quadratic formula (see Exercise 1.7).
Example 1.6 Determine the roots of the quadratics (a) f (x) = x2 − x − 2,
and (b) g(x) = 4x2 + 12x + 9.

I Solution: The roots are those values of x for which f (x) = 0, and g(x) = 0.
In this case the roots may be found neatly by factorising.
(a) The roots of f (x) = x2 − x − 2 occur when

x2 − x − 2 = (x + 1)(x − 2) = 0. (1.24)

Thus, the roots of f (x) are x = −1 and x = 2 (two distinct roots).

(b) The roots of g(x) = 4x2 + 12x + 9 occur when

4x2 + 12x + 9 = (2x + 3)(2x + 3) = (2x + 3)2 = 0. (1.25)

Thus, g(x) has a double root when 2x + 3 = 0, i.e., at x = − 32 .

The roots of a quadratic correspond to the intersections of the quadratic with
the x-axis. Distinct roots imply two intersections, while a double root implies
that the quadratic only just ‘touches’ the x-axis (see figure 1.5). J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 23

Definition 1.3 The general form for the quadratic function is

f (x) = ax2 + bx + c, (1.26)

where a, b, c ∈ R are constants, with a 6= 0. The roots of the quadratic

are the solutions x = x± to the quadratic equation f (x) = 0, namely
√
−b ± b2 − 4ac
x± = (1.27)
2a
(this equation is called the quadratic formula). These solutions yield:

• Two distinct roots (x+ 6= x− ) when b2 − 4ac > 0.

• A double root (x+ = x− ) when b2 − 4ac = 0.

• Complex conjugate roots when b2 − 4ac < 0 (see Section 4).

The quadratic function is also called the quadratic polynomial.

Example 1.7 Show that the roots of f (x) = ax2 + bx + c are

√
−b ± b2 − 4ac
x= . (1.28)
2a

I Solution: The roots of f (x) are the values for x that satisfy f (x) = 0, i.e.,
 
2 2 b c
ax + bx + c = 0 ⇔ x + x+ =0 (1.29)
a a
 2  2  
b b c
⇔ x+ − + = 0, (1.30)
2a 2a a
2
b2 − 4ac

b
⇔ x+ = , (1.31)
2a 4a2
√
b ± b2 − 4ac
⇔ x+ = , (1.32)
2a √ 2a
−b ± b2 − 4ac
⇔ x= , (1.33)
2a
as required. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

24 Section 1 : Functions and Algebra

1.4.3 Polynomials of arbitrary degree

The general definition of a polynomial function is as follows.

Definition 1.4 The general form of a polynomial of degree n ∈ N is

P (x) = an xn + an−1 xn−1 + · · · + a2 x2 + a1 x + a0 , (1.34)

where a0 , a1 , . . . , an are constants known as the coefficients, and an 6= 0.

Notice that the degree of the polynomial function is the highest power of x.
Thus, a polynomial of degree 2 is a quadratic function, viz.

P (x) = a2 x2 + a1 x + a0 . (1.35)

Likewise, a polynomial of degree 3 is a cubic function

P (x) = a3 x3 + a2 x2 + a1 x + a0 . (1.36)

The general polynomial of degree n may be written in summation notation as

n
X
P (x) = ar xr = a0 + a1 x + a2 x2 + · · · + an−1 xn−1 + an xn , (1.37)
r=0

where r is known as the summation index, or dummy index.

Example 1.8 State the degree of the following polynomials: (a) f (x) = a0 ,
and (b) f (x) = a1 x + a0 , where a0 , a1 ∈ R are constants.

I Solution: (a) Since x0 = 1 we may write f (x) = a0 x0 ; the highest power of

x in this equation is 0, so the polynomial is of degree 0 (or zeroth-degree).

(b) In this case we may write f (x) = a1 x1 + a0 x0 ; the highest power of x in

this equation is 1, so the polynomial is of degree 1 (or first-degree).

Note: These polynomials are both examples of straight lines (see §1.4.1). J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 25

1.4.4 Roots and factors

The roots of a polynomial P (x) are those values of x for which P (x) = 0;
thus, if x = α is a root of P (x), then

P (α) = 0. (1.38)

The roots of a polynomial are closely related to what are called the linear
factors of the polynomial. To illustrate this idea consider the cubic polynomial

P (x) = 2x3 + x2 − 2x − 1. (1.39)

Here P (x) can be factorised into the product of a linear factor and a quadratic
factor, viz.

P (x) = 2x3 + x2 − 2x − 1 = (x − 1) (2x2 + 3x + 1) . (1.40)

| {z } | {z }
linear factor quadratic factor

If we set x = 1, then the linear factor vanishes, i.e., (x − 1) = (1 − 1) = 0,

meaning that P (1) = 0; hence, x = 1 is a root of P (x). More generally:

Definition 1.5 If x = α is a root of P (x), then

P (α) = 0. (1.41)

It may be shown that if P (α) = 0, then (x − α) is a linear factor of P (x),

i.e.,
P (x) = (x − α)Q(x), (1.42)

where Q(x) is a polynomial of degree one less than the degree of P (x).
This result is a consequence of the remainder theorem (see §1.4.6)

In our example above the quadratic factor can also be written as the product
of linear factors, i.e, (2x2 + 3x + 1) = (2x + 1)(x + 1), hence

P (x) = 2x3 + x2 − 2x − 1 = (x − 1)(2x + 1)(x + 1). (1.43)

The polynomial P (x) thus has three real roots x = 1, x = − 21 , and x = −1.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

26 Section 1 : Functions and Algebra

In general, it may be shown that a polynomial of degree n has n roots; this

result is known as the fundamental theorem of algebra:

Theorem 1.1 A polynomial P (x) of degree n has exactly n roots. [This

is not to say that the roots are necessarily real numbers, or distinct.]

We shall not prove this theorem, but observe that the note in parenthesis is
important: a polynomial of degree n does not necessarily have n distinct
real roots. To illustrate this consider the cubic polynomial defined by

Q(x) = x3 + x + 3x2 + 3 = (x + 3)(x2 + 1). (1.44)

The linear factor is (x + 3), meaning that x = −3 is one of the roots of Q(x).
However, the quadratic factor is (x2 + 1), so the other roots occur when

x2 + 1 = 0. (1.45)

Since x2 > 0 for all x ∈ R, this equation does not have any solutions that
are real numbers. Indeed to solve this equation we must introduce complex
numbers (see Section 4). It follows that Q(x) has only one real root x = −3.

Example 1.9 Consider the cubic polynomial P (x) = x3 − 3x2 − 4x + 12.

(a) Given that P (3) = 0, write down one of the linear factors of P (x).
(b) Factorise P (x) into its linear and quadratic factor.
(c) Determine all the real roots of P (x).

I Solution: (a) Since P (3) = 0, we have that x = 3 is a root of P (x); this

means that (x − 3) is a linear factor (see Definition 1.5).

(b) Factorising P (x) we have P (x) = x3 − 3x2 − 4x + 12 = (x − 3)(x2 − 4),

where (x − 3) is the linear factor, and (x2 − 4) is the quadratic factor.

(c) The quadratic factor may itself be factorised as (x2 − 4) = (x − 2)(x + 2);
hence, P (x) may be factorised as

P (x) = (x − 3)(x − 2)(x + 2). (1.46)

Thus, P (x) has three real roots x = 3, x = 2, and x = −2. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 27

1.4.5 Algebraic division

It is not always easy to ‘spot’ the factorisation of a polynomial. However, if

one factor is known, then others may be found using algebraic division.
Example 1.10 At the beginning of §1.4.4 we considered the cubic polynomial
defined in equation (1.39) as

P (x) = 2x3 + x2 − 2x − 1. (1.47)

Use algebraic division to show that (x − 1) is a factor of P (x).

I Solution: Dividing P (x) = 2x3 + x2 − 2x − 1 by (x − 1) yields

2x2 + 3x + 1. (1.48)


x−1 2x3 + x2 − 2x − 1
− 2x3 + 2x2
3x2 − 2x
− 3x2 + 3x
x−1
−x+1
0

Since the remainder is 0, we have that (x − 1) is a factor of P (x) as required.

Note: It follows from this division that

P (x) = 3x3 + x2 − 3x − 1 = (x − 1)(2x2 + 3x + 1), (1.49)

which is the factorisation we had in equation (1.40). J

Algebraic division can also be used when the divisor is not a factor of the
polynomial. In such cases the division leaves an algebraic remainder.
Example 1.11 Consider the polynomial defined by

P (x) = x3 − 3x2 + x − 1 (1.50)

Use algebraic division to show that

P (x) = (x − 1)Q(x) + R, where Q(x) = x2 − 2x − 1 (1.51)

is a quadratic, and R = −2 is a constant remainder.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

28 Section 1 : Functions and Algebra

I Solution: We divide P (x) by (x − 1) using algebraic division, viz.

x2 − 2x − 1. (1.52)


x−1 x3 − 3x2 + x − 1
− x 3 + x2
− 2x2 + x
2x2 − 2x
−x−1
x−1
−2

Thus division division by (x − 1) leaves a quotient Q(x) = x2 − 2x − 1, and

a remainder R = −2. In this way we have as required that

P (x) = (x−1)Q(x)+R, with Q(x) = x2 −2x−1, R = −2. (1.53)

Note: Observe as a point of interest that P (1) = −2 = R. J

1.4.6 Remainder theorem

In Example 1.11 we found that the cubic P (x) = x3 − 3x2 + x − 1 could be

written in the form

P (x) = (x − 1)Q(x) + R where Q(x) = x2 − 2x − 1 (1.54)

is a quadratic , and R = −2 is a constant.

More generally, if P (x) is a polynomial of degree n, then given some constant

α it is possible to write

P (x) = (x − α)Q(x) + R, (1.55)

where Q(x) is a polynomial of degree (n−1), and R is a constant, i.e., division

of P (x) by (x − α) leaves a constant remainder R.

Informally, if we put x = α into equation (1.55), then we see that the value of
the remainder must be
P (α) = R. (1.56)

This result is known as the remainder theorem.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 29

Theorem 1.2 (remainder theorem) Division of a polynomial P (x) by

(x − α) leaves a constant remainder R = P (α). In particular

P (x) = (x − α)Q(x) + R, with R = P (α), (1.57)

where Q(x) is a polynomial of degree (n − 1).

Corollary: If α is a root of P (x), then by definition P (α) = R = 0; it

therefore follows by equation (1.57) that (x − α) is a factor of P (x).

1.4.7 Comparing coefficients

The following result is often useful when solving polynomial equations:

Result 1.1 Two polynomials P (x) and Q(x) are equal to one another for
all values of x ∈ R if and only if

(i) P (x) and Q(x) are polynomials of the same degree.

(ii) The coefficients of P (x) are identical to the coefficients of Q(x).

This result may be proved using mathematical induction.

Example 1.12 Consider the quadratic polynomials defined by

P (x) = ax2 + bx + 5 and Q(x) = 2x2 − 3x + c, (1.58)

where a, b, c ∈ R are constants. Given that P (x) = Q(x) for all x ∈ R,

determine the values of a, b and c.

I Solution: Setting P (x) = Q(x) we have

ax2 + bx + 5 = 2x2 − 3x + c. (1.59)

Hence, by comparing coefficients, we require a = 2, b = −3, and c = 5. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

30 Section 1 : Functions and Algebra

1.4.8 Rational functions

Let P (x) and Q(x) be polynomials, then a function of the form

P (x)
f (x) = (1.60)
Q(x)

is known as a rational function. A rational function is said to be proper if

the degree of P (x) is less than the degree of Q(x); for example,

x3 + 2x + 3
f (x) = (1.61)
x4 + 2x3 − x − 4
is a proper rational function. A rational function that is not proper is said to
be improper (see Example 1.13).
Example 1.13 Use the fact that (x + 1)(x − 2) + 3 = (x2 − x + 1) to write

x2 − x + 1
g(x) = (1.62)
x−2
as the sum of a polynomial and a proper rational function.

I Solution: Setting (x2 − x + 1) = (x + 1)(x − 2) + 3 we may write g(x) as

x2 − x + 1 (x + 1)(x − 2) + 3 3
g(x) = = = (x + 1) + , (1.63)
x−2 (x − 2) (x − 2)

i.e., as the sum of a linear polynomial and a proper rational function. J

1.5 Partial Fractions

When the denominator of a rational function can be factorised, the rational

function can be broken down into a sum of simpler fractions known as partial
fractions. To see how this works, consider the rational function defined by

5x + 1 5x + 1
f (x) = = , (1.64)
x2+x−2 (x + 2)(x − 1)

where the factors in the denominator are (x + 2) and (x − 1). The basic idea
of partial fractions is to find some constants A and B such that f (x) can be
written in terms of these factors as
A B
f (x) = + . (1.65)
(x + 2) (x − 1)

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 31

Now let us find A and B. For equation (1.65) to hold, we require

5x + 1 A B
= + . (1.66)
(x + 2)(x − 1) (x + 2) (x − 1)

Thus, multiplying through by (x + 2)(x − 1) we have

5x + 1 = A(x − 1) + B(x + 2); (1.67)

this equation must be true for all x. To find A we eliminate B by setting

x = −2, since this gives (x + 2) = 0; indeed, putting x = −2 into equation
(1.67) gives
− 9 = −3A, that is, A = 3. (1.68)

Similarly, to find B we eliminate A from equation (1.67) by setting x = 1,

since this gives (x − 1) = 0; indeed, putting x = 1 into equation (1.67) gives

6 = 3B, that is, B = 2. (1.69)

With these values for A and B, it follows from equation (1.66) that f (x) may
be written as
5x + 1 3 2
f (x) = = + , (1.70)
(x + 2)(x − 1) (x + 2) (x − 1)

where the second form is f (x) expressed as partial fractions.

Successful application of the method of partial fractions relies on making

a good starting guess, as we did in equation (1.66). Table 1.1 suggests good
trial forms for the partial fractions given various kinds of rational function.

Rational function Partial fractions

ax+b A B
(x+α)(x+β) (x+α) + (x+β)

ax2 +bx+c A
+ B
+ C
(x+α)(x+β)(x+γ) (x+α) (x+β) (x+γ)

ax2 +bx+c A
+ Bx+C
(x+α)(x2 +βx+γc) (x+α) (x2 +βx+γc)

ax2 +bx+c A
+ B
+ C
(x+α)(x+β)2 (x+α) (x+β) (x+β)2

Table 1.1: Trial forms for partial fractions given various kinds of rational function. The
constants a, b, c, α, β, and γ are knowns; the coefficients A, B and C are to be determined.
This table is not exhaustive, but hints at suitable partial fractions for other rational functions.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

32 Section 1 : Functions and Algebra

Example 1.14 Express the rational function

3x2 − x − 5
f (x) = (1.71)
x3 − 3x + 2

as partial fractions given that x3 − 3x + 2 = (x + 2)(x − 1)2 .

I Solution: According to the hint f (x) may be written as

3x2 − x − 5
f (x) = . (1.72)
(x + 2)(x − 1)2

Thus, f (x) is a rational function of the kind listed in the final row of Table
1.1, suggesting that we seek partial fractions of the form

3x2 − x − 5 A B C
f (x) = = + + , (1.73)
(x + 2)(x − 1)2 (x + 2) (x − 1) (x − 1)2

with A, B, and C as constant coefficients to be found. [To see this, set a = 3,

b = −1, c = −1, and α = 2, β = −1 in the final row of Table 1.1.]

Multiplying equation (1.73) through by (x + 2)(x − 1)2 yields

3x2 − x − 5 = A(x − 1)2 + B(x + 2)(x − 1) + C(x + 2). (1.74)

We can eliminate A and B from this equation by setting x = 1 to give

− 3 = 3C, that is, C = −1. (1.75)

Likewise, setting x = −2 in equation (1.74) eliminates B and C to give

9 = 9A, that is, A = 1. (1.76)

The coefficient B may be found by selecting any other value for x, and noting
that A = 1 and C = −1. For example, setting x = 0 in equation (1.74) gives

− 5 = A − 2B + 2C = −2B − 1 that is, B = 2. (1.77)

Thus, substituting A = 1, B = 2, and C = −1 into equation (1.73) we obtain

3x2 − x − 5 1 2 1
f (x) = 2
= + − , (1.78)
(x + 2)(x − 1) (x + 2) (x − 1) (x − 1)2

as f (x) in the form of partial fractions. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 33

1.6 Trigonometric Functions

Figure 1.6 depicts a circle of radius r = 1, i.e., a unit circle, centred at the
origin O of the x-y plane; the circumference of this circle is 2πr = 2π.

Let P be an arbitrary point on the circumference of the unit circle, and let θ
denote the angle that the line OP makes with the positive x-axis, with s as
the distance along the circumference of the circle from the x-axis to P .

The angle θ is measured in radians, which are defined such that there are
exactly 2π radians in a circle. Hence, by the equality of proportions

θ s
= , that is (since r = 1), θ = s. (1.79)
2π 2πr
The trigonometric functions sine and cosine may then defined as follows. The
cosine of the angle θ, denoted cos θ, is the value of the x coordinate of P .
Likewise, the sine of the angle θ, denoted sin θ, is the y coordinate of P , viz.

P = (x, y) = (cos θ, sin θ). (1.80)

We can think of the point P as being a function of θ: as θ is increased, P

orbits the circle in the anti-clockwise sense. Since there are 2π radians in a
circle, increasing θ by 2π returns P to its starting position. It thus follows that

cos(θ + 2π) = cos θ and sin(θ + 2π) = sin θ; (1.81)

in this way the trigonometric functions are said to be periodic. Such periodicity
is clear from the plots of sin θ and cos θ in figure 1.7.

Figure 1.6: Unit circle. The point P has coordinates (x, y) = (cos θ, sin θ).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

34 Section 1 : Functions and Algebra

2 2

1 1

0 0

-1 -1

-2 -2
-2 -1 0 1 2 -2 -1 0 1 2
(a) (b)

Figure 1.7: Trigonometric functions (a) cos θ and (b) sin θ. The horizontal (abscissa) axis
is normalised to units of 2π radians; thus, one full cycle (θ = 2π) corresponds to θ/2π = 1.

Other properties of sin θ and cos θ include the following (see figure 1.7).

Result 1.2 The functions sin θ and cos θ are related by

cos θ = sin(θ + π2 ) and sin θ = cos(θ − π2 ). (1.82)

Observe by figure 1.7 that these functions satisfy (cf. §1.8).

cos(−θ) = cos(θ) and sin(−θ) = − sin(θ). (1.83)

For any integer n ∈ Z we have

cos(nπ) = (−1)n and sin(nπ) = 0. (1.84)

From these equations it follows by equations (1.82) and (1.83) that

cos(nπ + π2 ) = 0 and sin(nπ + π2 ) = (−1)n . (1.85)

Our definitions of sin θ and cos θ allow us to further define the tangent (tan θ)
of an angle θ as
sin θ
tan θ = . (1.86)
cos θ
We also define the secant (sec θ), cosecant (csc θ), and cotangent (cot θ):

1 1 1
sec θ = , csc θ = , cot θ = . (1.87)
cos θ sin θ tan θ
These functions are plotted in figure 1.8.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 35

4 4

2 2

0 0

-2 -2

-4 -4
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
(a) (b)
4 4

2 2

0 0

-2 -2

-4 -4
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
(c) (d)

Figure 1.8: Trigonometric functions (a) tan θ, (b) sec θ, (c) csc θ, and (d) cot θ.

1.6.1 Pythagoras

Returning to our diagram of the unit circle in figure 1.6, consider the right-
angle triangle with hypotenuse OP ; this hypotenuse has length 1. The side
adjacent to the angle θ has length x = cos θ, while side opposite the angle θ
has length y = sin θ. Thus, according to Pythagoras’s theorem we have

x2 + y 2 = 12 , that is, cos2 θ + sin2 θ = 1. (1.88)

[In fact, this equation is itself a statement of Pythagoras’s theorem.]

Equation (1.88) may be used to derive other relationships between the trigono-
metric functions; for example, dividing through by cos2 θ yields

sin2 θ 1
1+ = , that is, 1 + tan2 θ = sec2 θ. (1.89)
cos2 θ cos2 θ
Similarly, we may show that cot2 θ + 1 = csc2 θ (see Exercise 1.15).
Example 1.15 Show that cot2 θ + 1 = csc2 θ.

I Solution: Dividing equation (1.88) through by sin2 θ yields

cos2 θ 1
2 +1= , that is, cot2 θ + 1 = csc2 θ (1.90)
sin θ sin2 θ
as required. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

36 Section 1 : Functions and Algebra

1.6.2 Compound angle formulae

The diagram in figure 1.9 shows a right-triangle (grey) with hypotenuse of

length 1 inscribed within a rectangle. The triangle has been labelled to define
an angle β, and sits such that the side adjacent to the angle β makes a
further angle α with the bottom side of the rectangle. In this way, each of the
lengths associated with the intersection of the triangle with the rectangle can
be determined using trigonometry as labelled.

Comparing opposite sides of the rectangle in figure 1.9 we deduce that

cos(α + β) = cos α cos β − sin α sin β, (1.91)

sin(α + β) = sin α cos β + cos α sin β; (1.92)

these equations are known as the compound angle formulae and hold for all
α, β ∈ R. Indeed, since cos(−β) = cos(β) and sin(−β) = − sin(β), we may
use equations (1.91) and (1.92) to write

cos(α − β) = cos α cos β + sin α sin β, (1.93)

sin(α − β) = sin α cos β − cos α sin β. (1.94)

As we illustrate throughout the examples in this section, a vast range of other

trigonometric formulae may be derived by combining the above expressions.

Figure 1.9: Geometric mnemonic for the compound angle formulae.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 37

Example 1.16 Use the compound angle formulae to show that

tan α + tan β
tan(α + β) = (1.95)
1 − tan α tan β

where α, β ∈ R.

I Solution: With reference to equations (1.91) and (1.92) we have

sin(α + β)
tan(α + β) =
cos(α + β)
sin α cos β + cos α sin β
= (1.96)
cos α cos β − sin α sin β

Hence, dividing the top and bottom by cos α cos β we obtain

sin α/ cos α + sin β/ cos β tan α + tan β

tan(α + β) = = (1.97)
1 − sin α sin β/ cos α cos β 1 − tan α tan β

as required. J
Example 1.17 Use the compound angle formulae to show that

2 cos α cos β = cos(α + β) + cos(α − β), (1.98)

2 sin α cos β = sin(α + β) + sin(α − β), (1.99)

where α, β ∈ R.

I Solution: Adding equation (1.93) to (1.91) yields

cos(α + β) + cos(α − β) = 2 cos α cos β (1.100)

as required. Likewise, adding equation (1.94) to (1.92) yields

2 sin α cos β = sin(α + β) + sin(α − β). (1.101)

as required. J
Other formulae may be obtained by repeated application of the double angle
formulae. For example, suppose that we define

A+B A−B
α= and β= (1.102)
2 2
where A and B are any two real numbers.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

38 Section 1 : Functions and Algebra

It follows that A and B may be expressed in terms of α and β as

(α + β) = A and (α − β) = B (1.103)

Thus, putting equations (1.103) and (1.102) into equation (1.100) we obtain.
   
A+B A−B
cos A + cos B = 2 cos cos . (1.104)
2 2

A similar expression for sin A + sin B is derived in Example 1.18 below.

Example 1.18 Demonstrate that

   
A+B A−B
sin A + sin B = 2 sin cos , (1.105)
2 2

where A, B ∈ R.

I Solution: Putting equations (1.103) and (1.102) into equation (1.101) we

have    
A+B A−B
sin A + sin B = 2 sin cos , (1.106)
2 2
as required. J

1.6.3 Identities

If two mathematical expressions are equal for all values of the arguments,
then we say that they are equivalent, or identical. Equations which convey
equivalence are known as identities, and use an equivalence symbol ‘≡’ in
place of the equality symbol ‘=’. For example, equation (1.88) holds for all
values of θ ∈ R, and may be written as the identity.

cos2 θ + sin2 θ ≡ 1. (1.107)

The compound angle formulae are also identities, i.e.,

cos(α + β) ≡ cos α cos β − sin α sin β, (1.108)

sin(α + β) ≡ sin α cos β + cos α sin β. (1.109)

Two commonly used identities known as the double angle formulae may be
derived by considering the identities above.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 39

To see this, let α = β = θ in equation (1.109), then we have

sin(2θ) ≡ 2 sin θ cos θ. (1.110)

Likewise, setting α = β = θ in equation (1.108) yields

cos(2θ) ≡ cos2 θ − sin2 θ. (1.111)

Alternative forms of this identity are derived in Example 1.19.

Example 1.19 Beginning with equation (1.111), use the fact that

cos2 θ + sin2 θ ≡ 1 (1.112)

to show that (i) cos(2θ) ≡ 2 cos2 θ − 1; and (ii) cos(2θ) ≡ 1 − 2 sin2 θ.

I Solution: (i) By equation (1.112) we have sin2 θ ≡ 1−cos2 θ; hence, putting

this relation into cos(2θ) ≡ cos2 θ − sin2 θ yields

cos(2θ) ≡ 2 cos2 θ − 1. (1.113)

(ii) Similarly, equation (1.112) means that cos2 θ ≡ 1 − sin2 θ; hence, putting
this relation into cos(2θ) ≡ cos2 θ − sin2 θ gives

cos(2θ) ≡ 1 − 2 sin2 θ (1.114)

as required. J
The most important identities derived in this section are summarised below.

Result 1.3 The following trigonometric identities are often helpful:

sin2 θ + cos2 θ ≡ 1 (1.115)

cos(α + β) ≡ cos α cos β − sin α sin β (1.116)
sin(α + β) ≡ sin α cos β + cos α sin β (1.117)
cos A + cos B ≡ 2 cos A+B cos A−B



2 2
(1.118)
cos A − cos B ≡ −2 sin A+B sin A−B



2 2
(1.119)
sin A ± sin B ≡ 2 sin A±B cos A∓B



2 2
(1.120)

These identies are valid for all α, β, A, B ∈ R.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

40 Section 1 : Functions and Algebra

1.7 Composite Functions and Inverse Fucntions

A mathematical function can be thought of as an operator that performs an

operation on a variable. For example, the function f (x) = 2x is said to
operate on x; the operation f performs is to multiply x by 2.

We can also use functions to operate on other functions, this leads us to the
idea of composite functions (§1.7.1), and inverse functions (§1.7.2).

1.7.1 Composite functions

Given two functions f (x) and g(x), we may define a third function

h(x) = g(f (x)). (1.121)

A function h(x) of this kind is known as a composite function and means

apply the operation g to the output of the function f (see figure 1.10).

In practice the form of a composite function h(x) = g(f (x)) is usually found
by replacing each instance of the variable x with f (x) in the rule for g(x).

Example 1.20 Let f (x) = x2 and g(x) = cos(x). Write out following com-
posite functions in terms of x: (a) h(x) = g(f (x)); and (b) l(x) = f (g(x)).

I Solution: (a) Taking the output of f (x) and using it as the input argument
of g yields
h(x) = g(f (x)) = cos(f (x)) = cos(x2 ). (1.122)

(b) Similarly, taking the output of g(x) and using it as the input argument of
f yields
l(x) = f (g(x)) = (g(x))2 = cos2 (x). (1.123)

Note: This example demonstrates that in general g(f (x)) 6= f (g(x)). J

Figure 1.10: Composite function h(x) = g(f (x)): the output of f used as the input of g.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 41

1.7.2 Inverse functions

The idea of an inverse function may be understood informally as follows. Sup-

pose that f (x) is a function of x; if some other function g(x) can be found
such that
g(f (x)) = x (1.124)

then we say that g(x) is the inverse function of f (x), and write

g(x) = f −1 (x). (1.125)

The basic idea of an inverse function is to allow us to infer x given y = f (x).

For example, the function f (x) = 2x has inverse function f −1 (x) = 21 x, i.e.,

1 1
f −1 (f (x)) = f (x) = (2x) = x. (1.126)
2 2
Here f (x) = 2x is said to be a one-one function because each value of x is
paired with a distinct value of y = f (x); thus, x can be inferred from the value
of y = f (x). The domain of f is the codomain of f −1 , and vice versa.
A function f (x) that maps several values of x to the same value of y = f (x)
is known as a many-one function. An example of a many-one function is

f (x) = cos(x); (1.127)

for instance, there are many values of x (e.g., x = 0, 2π, 4π, . . . ) that map
to the value y = cos(x) = 1. Thus, even if we know y = 1, we cannot infer
whether x = 0, or x = 2π, or x = 4π (etc.). Indeed, in general many-one
functions do not have unique inverses because the value of x cannot always be
inferred from the value of y = f (x).
Many-one functions can, however, be ‘converted’ to one-one functions by re-
stricting the domain. For example, the function defined by

f (x) = cos(x) for x ∈ [0, π] (1.128)

is a one-one function since each value of x ∈ [0, π] is paired to a single value

of y ∈ [−1, 1] and vice versa. Restricting the function in this way means that
we can define its inverse, namely f −1 (x) = arccos(x) (see figure 1.11).
Note that if f (x) has inverse function f −1 (x), then graphs of y = f (x) and
y = f −1 (x) are reflections of one-another in the line x = y (see figure 1.11).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

42 Section 1 : Functions and Algebra

1 8

6
0.5
4

0 2

0
-0.5
-2

-1 -4
-1 -0.5 0 0.5 1 -4 -2 0 2 4 6 8
(a) (b)

1
Figure 1.11: Inverse functions: (a) cos(x) and arccos(x); (b) 27 (x − 1)3 and 3x1/3 + 1.

Example 1.21 Determine the inverse function of

1
f (x) = (x − 1)3 , (1.129)
27
and sketch both f (x) and f −1 (x) on the same axis.

I Solution: Let us write

1
y = f (x) = (x − 1)3 , (1.130)
27
then rearranging this equation we obtain

x = f −1 (y) = 3y 1/3 + 1, (1.131)

where the right-hand-side describes the rule of f −1 .

In equation (1.131) we have expressed the rule of f −1 using y to denote the
input, and x to denote the output. If instead we denote the input as x and the
output as y, then the inverse function is expressed as

f −1 (x) = 3x1/3 + 1. (1.132)

The functions f (x) and f −1 (x) are plotted in figure 1.11(b).

Observe that the graphs y = f (x) and x = f −1 (y) are identical to each
another. The graph of y = f −1 (x) is obtained from the graph of x = f −1 (y)
by swapping the x and y coordinates; this is why y = f (x) and y = f −1 (x)
are reflections of one-another in the line x = y. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 43

1.8 Odd and even functions

Many functions can be classified as to whether they are odd or even (see
figure 1.12); such characteristics are known as symmetry properties

Definition 1.6 A function f (x) is said to be either odd (anti-symmetric)

or even (symmetric) if it satisfies one of the following conditions:

f (x) is odd about x = 0 if and only if f (−x) = −f (x), (1.133)

f (x) is even about x = 0 if and only if f (−x) = +f (x). (1.134)

It may be shown that the product of two odd or even functions is also
either odd or even according to the following rules:

(odd function) × (odd function) = (even function), (1.135a)

(even function) × (odd function) = (odd function), (1.135b)
(odd function) × (even function) = (odd function), (1.135c)
(even function) × (even function) = (even function). (1.135d)

For example, the trigonometric functions both have symmetry properties:

cos(−x) = + cos(x) is an even function, (1.136)

sin(−x) = − sin(x) is an odd function. (1.137)

These properties are clear from the plots in figure 1.7.

(a) (b)

Figure 1.12: Symmetry about the y-axis: (a) an even (symmetric) function f (−x) = f (x);
and (b) an odd (anti-symmetric) function f (−x) = −f (x).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

44 Section 1 : Functions and Algebra

Example 1.22 Classify the following functions in terms of whether they are
odd or even: (a) f (x) = x4 −4; (b) g(x) = sin(x3 ); (c) h(x) = (x4 −4) sin(x3 ).

I Solution: Let us take each of the functions in turn.

(a) f (−x) = (−x)4 − 4 = x4 − 4 = f (x), that is, f (x) is even.
(b) g(−x) = sin((−x)3 ) = sin(−x3 ) = − sin(x3 ) = −g(x), i.e., g(x) is odd.
(c) Since f (x) is even and g(x) is odd, h(x) = f (x)g(x) is odd. J

1.9 Exponentials and Logarithms

Consider the expression

8 = 23 . (1.138)

When 8 has been written in this way we say that the base is 2 and the power
(or exponent) is 3. An equivalent way of expressing the relationship above is

log2 8 = 3, (1.139)

which reads as “the logarithm to base 2 of 8 is 3”; indeed, we say that “log
to base 2 of 8 is 3”. To emphasise,

8 = 23 and log2 8 = 3 (1.140)

are equivalent ways of expressing the same relationship.

Definition 1.7 For some numbers x, a, n ∈ R we say that

x = an and loga x = n (1.141)

are equivalent statements, where a is known as the base, and loga x = n

means “the logarithm to base a of x is n”. Notice by this definition that

a = a1 and loga a = 1 (1.142)

are equivalent statements. The definition also means that

1 = a0 and loga 1 = 0 (1.143)

are equivalent, i.e., the logarithm of unity (1) in any base is zero.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 45

Example 1.23 Write the following statements in terms of logarithms:

(a) 83 = 512; (b) 34 = 81; (c) 210 = 1024. (1.144)

I Solution: (a) log8 512 = 3; (b) log3 81 = 4; and (c) log2 1024 = 10. J

Our definition of the logarithm means that it is a function. To see this let
a, x, y ∈ R, and consider the equivalent statements

x = ay and loga x = y. (1.145)

We thus have two functions, y = f (x) and x = g(y), where

y = f (x) = loga x, and x = g(y) = ay . (1.146)

Hence, if we know x, then we can compute y, and vice versa; that is, f and g
are inverse functions of each other, viz.

x = g(f (x)) = aloga x and y = f (g(y)) = loga ay . (1.147)

Notice that we have expressed the rule for g using y as the independent variable;
to express the rule using x as the independent variable we simply write

g(x) = ax . (1.148)

The functions f and g are plotted in figure 1.13 using bases a = 10 and a = e,
where e ≈ 2.718281828459 is called Euler’s number (see §1.9.4).

2 2

1 1

0 0

-1 -1

-2 -2
-2 -1 0 1 2 -2 -1 0 1 2
(a) (b)

Figure 1.13: Functions f (x) = loga x and g(x) = ax for bases: (a) a = 10; (b) a = e.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

46 Section 1 : Functions and Algebra

Result 1.4 The functions f (x) and g(x) defined by

f (x) = loga x and g(x) = ax (1.149)

are inverse functions of each other. Thus,

loga ax = x and aloga x = x. (1.150)

These results are useful for solving equations involving powers.

Example 1.24 Solve the following equations: (a) log3 x = 2; (b) log2 x = 4.

I Solution: The inverse function of f (x) = loga x is g(x) = ax , thus:

(a) log3 x = 2 means that x = 3log3 x = 32 = 9.
(b) log2 x = 4 means that x = 2log2 x = 24 = 16. J

Example 1.25 Solve the following equations: (a) 3x = 9; (b) 2x = 16.

I Solution: The inverse function of g(x) = ax is f (x) = loga x, thus:

(a) 3x = 9 means that x = log3 3x = log3 9.
(b) 2x = 16 means that x = log2 2x = log2 16. J

1.9.1 Laws of logarithms

The following properties of logarithms are often useful for solving equations.

Result 1.5 Let a, n, x, y ∈ R, then we may deduce the following laws:

loga (xy) = loga x + loga y (first law) (1.151a)

loga xn = n loga x (second law) (1.151b)
loga (x/y) = loga x − loga y (third law) (1.151c)

These laws are deduced from the rules for exponents (e.g., Example 1.26).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 47

Example 1.26 Suppose that x, y ∈ R be real numbers that can be written in

terms of a base a ∈ R as

x = an , and y = am (1.152)

where n, m ∈ R. Show that loga (xy) = loga x + loga y.

I Solution: First observe by the definition of logarithms (Definition 1.7) that

x = an and y = am may be written in logarithm form as

loga x = n and loga y = m. (1.153)

Now notice that the rules for exponents mean that xy = an am = an+m , so in
logarithm form (Definition 1.7) we have

loga (xy) = n + m. (1.154)

Thus, substituting equations (1.153) into equation (1.154) we obtain

loga (xy) = loga x + loga y (1.155)

as required. This equation is the first law of logarithms listed in Result 1.5. J

1.9.2 Orders of magnitude

Logarithms in base 10 are important because we use a decimal number system,

and because base 10 is convenient for expressing data spread over what are
termed orders of magnitude. In particular, given two quantitates x1 and x2 ,
we say that x2 is n orders of magnitude bigger than x1 if

x2 ∼ 10n x1 , (1.156)

where ∼ means similar to (or ‘about the same size’). For example, the average
weight of a female human in the United Kingdom is mH ≈ 70kg, whereas the
average weight of a male African elephant is mE ≈ 6000 kg; thus,

mE ∼ 102 mH , (1.157)

and we say that mE is n = 2 orders of magnitude larger than mH .

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

48 Section 1 : Functions and Algebra

1.9.3 Logarithmic plots

Logarithms are extremely useful for revealing trends in data over several orders
of magnitude. For example, consider some kind of exponential dependence

y = y 0 ax , (1.158)

where a and y0 are constants. Dividing by y0 and taking logarithms yields

log10 (y/y0 ) = log10 ax = x log10 a, (1.159)

which is a straight line relationship of the form

Y = mx, with Y = log10 (y/y0 ), and m = log10 a. (1.160)

Thus, plotting log10 (y/y0 ) against x yields a straight line whose gradient m
can be used to infer the value of a (see Example 1.27 and figure 1.14). The
same visual effect can also be achieved using logarithmic axis. On logarithmic
axis numbers are spaced evenly according to their magnitude, e.g., the sepa-
ration between 1 and 10 is the same as the separation between 10 and 100.
Logarithmic axis are best understood by plotting some example data-series.

Example 1.27 Moore’s law is an empirical observation that the number of

transistors N that can be fit on a microchip doubles every two years, and has
been in operation since about 1960. Show that Moore’s law can be written
√
log10 (N (n)/N0 ) = n log10 2, (1.161)

where n is the number of years since 1960, and N0 = N (0) (see figure 1.14).

10 8
12 10 10 10

10 8 10 8
8
6 10 6
6
4 10 4
4

2 2 10 2

0 0 10 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60
(a) (b) (c)

n
√
Figure 1.14: Moore’s law:
√ (a) on standard axis as N/N0 = ( 2) ; (b) on standard
√ axis
as log10 (N/N0 ) = n log10 2; (c) plotted using logarithmic y-axis as N/N0 = ( 2)n .

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 49

I Solution: According to Moore’s law the number of transistors N on a mi-

√
crochip increases by a factor of 2 every two years, that is, by a factor of 2
every year. Thus, letting n denote the number of years since the advent of the
law, and N0 = N (0) denote the value of N in that advent year, then
√ √
after n = 1 years: N (1) = N0 × 2 = N0 ( 2)1 , (1.162a)
√ √ √
after n = 2 years: N (2) = N0 × 2 × 2 = N0 ( 2)2 , (1.162b)

and so on. In general, therefore, we have the exponential form of Moore’s law
√
after n years: N (n) = N0 ( 2)n . (1.163)

Thence, dividing by N0 , and taking logarithms to base 10, we obtain

√ √
log10 (N (n)/N0 ) = log10 ( 2)n = n log10 2 (1.164)

as required. These expressions for Moore’s law are plotted in figure 1.14. J

Logarithms can also be used to transform a power law into a linear relationship.
For example, let
y = y 0 xn , (1.165)

where n and y0 are constants. Dividing by y0 and taking logarithms yields

log10 (y/y0 ) = log10 xn = n log10 x, (1.166)

which is a straight line relationship of the form

Y = nX, with Y = log10 (y/y0 ), and X = log10 x. (1.167)

An example relationship of this kind is plotted in figure 1.15.

10 5 6
10 10 6

8 3 10 3

6
0 10 0
4
-3 10 -3
2

0 -6 10 -6 -2
0 20 40 60 80 100 -2 -1 0 1 2 10 10 -1 10 0 10 1 10 2
(a) (b) (c)

Figure 1.15: Power law y = x3 : (a) plotted on standard axis as y = x3 ; (b) plotted on
standard axis as log10 y = 3 log10 x; (c) plotted using logarithmic axis as y = x3 .

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

50 Section 1 : Functions and Algebra

1.9.4 Natural logarithm

Base 10 is said to be a standard base in the sense that it is used very com-
monly. The other standard base is base e, where
∞
X 1
e= = 2.718281828459 . . . , (1.168)
n=0
n!

is Euler’s number. Base e is important because e arises everywhere in nature.

Taking a logarithm in base e is called “taking the natural logarithm”.

Result 1.6 The standard bases are base 10 and base e (see figure 1.13).
Logarithms in these bases are often written using the following notation:

• log10 x is sometimes written as log x.

• loge x is called the natural logarithm, and sometimes written ln x.

This notation is not universal; for example, mathematicians tend to write

the natural logarithm loge x as log x (because it is the most important).
To avoid ambiguity, it is usually prudent to specify the base explicitly.

1.10 Exponential Function and Hyperbolic Functions

Any function of the form

f (x) = ax (1.169)

where a is a constant, is known as an exponential function. A particularly

important case of such a function is the natural exponential function, or
simply the exponential function

f (x) = ex , (1.170)

where e is Euler’s number. There are many equivalent ways of defining the
exponential function, but in this course we favour a power series representation
(see Definition 1.8). We discuss power series in further detail in §2.5.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 51

Definition 1.8 The exponential function is denoted ex or exp(x), and

defined by the power series
∞
x
X xn x2 x3
e ≡ exp(x) ≡ =1+x+ + + ... (1.171)
n=0
n! 2! 3!

Putting x = 1 into this series gives Euler’s number e = 2.7182818284 . . .

Hyperbolic functions

The exponential function allows as to define the so-called hyperbolic func-

tions: hyperbolic cosine, or cosh (read as ‘cosh’), hyperbolic sine, or sinh (read
as ‘shine’); and hyperbolic tangent, or tanh (read as ‘tanch’). The hyperbolic
functions are closely related to the trigonometric functions.

Definition 1.9 The hyperbolic functions cosh(x), sinh(x), and tanh(x)

are defined by

1
cosh(x) ≡ (ex + e−x ),
2
1
sinh(x) ≡ (ex − e−x ), (1.172)
2
sinh(x)
tanh(x) ≡ (1.173)
cosh(x)

(see figure 1.16). We also define the reciprocal hyperbolic functions

1
sech(x) ≡ ,
cosh(x)
1
csch(x) ≡ , (1.174)
sinh(x)
1
coth(x) ≡ ; (1.175)
tanh(x)

these functions are read as ‘sesh’, ‘cosesh’, and ‘coth’ respectively.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

52 Section 1 : Functions and Algebra

8 10

6 5

4 0

2 -5

0 -10
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
(a) (b)
10 2

5 1

0 0

-5 -1

-10 -2
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
(c) (d)

Figure 1.16: Hyperbolic functions (a) ex , (b) sinh x, (c) cosh x, and (d) tanh x.

As with the trigonometric functions, we write the inverse hyperbolic functions

using the notation arccosh, arcsinh, and arctanh.

Definition 1.10 It may be shown (see, e.g., Example 1.29) that the hy-
perbolic functions have inverse functions given by
 √ 
sinh−1 (x) ≡ arcsinh(x) ≡ loge x + x2 + 1 ,
−1
 √ 
cosh (x) ≡ arccosh(x) ≡ loge x + x − 1 , 2 (1.176)
r !
1+x
tanh−1 (x) ≡ arctanh(x) ≡ loge . (1.177)
1−x

Example 1.28 Show that cosh2 (x) − sinh2 (x) ≡ 1.

I Solution: According to Definition 1.9 we have

1 1
cosh2 (x) − sinh2 (x) ≡ (ex + e−x )2 − (ex − e−x )2
4 4
1 2x 1
≡ (e + 2 + e−2x ) − (e2x − 2 + e−2x )
4 4
≡1 (1.178)

as required. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 53

Example 1.29 Let x(y) = sinh(y); determine y(x) as a function of x.

I Solution: By Definition 1.9 we have that x = 12 (ey − e−y ). Hence, multiply-

ing this equation through by ex , we find upon rearranging that

1
x = (ey − e−y ) ⇔ e2y − 2xey − 1 = 0. (1.179)
2
Now let ey = z, then e2y = z 2 , and equation (1.179) can be expressed as

z 2 − 2xz − 1 = 0. (1.180)

Equation (1.180) is a quadratic in z, and therefore has solution

√
2x ± 4x2 + 4 √
ey = z = = x + x2 + 1. (1.181)
2
where we have retained the positive square-root only because ey > 0. Hence,
taking logarithms we obtain
 √ 
y(x) = loge z = loge x + x2 + 1 . (1.182)

This equation is the rule for the inverse function of sinh x in Definition 1.10.J

1.11 Explicit and Implicit Functions

A function of the form y = f (x) is known as an explict function in the sense

that the dependent variable y can be obtained by applying an explicit rule to
the independent variable x. For example, the function

y = f (x) = 2ex sin x (1.183)

is an explicit function. Not all functions can be expressed in this way; indeed,
it is sometimes necessary (or desirable) to express the relationship between two
variables x and y in the form

g(x, y) = c (1.184)

where c is a constant. Such a function is called an implicit function: the rule

applies to x and y to give c, meaning that y is related to x through c implicitly.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

54 Section 1 : Functions and Algebra

A simple example of an implicit function is

xey + sin(xy) = 1. (1.185)

Implicit functions can sometimes be rearranged to take the form of an explicit

function. For example, the expressions

2
xy = 2 (implicit form) ⇔ y(x) = (explicit form) (1.186)
x
both describe the same relationship between x and y (they are equivalent).

Example 1.30 Express the implicit relation xe−y = 1 in explicit form.

I Solution: Rearranging xe−y = 1 we obtain the explicit form y = loge x. J

1.12 Parametric Equations

It is sometimes convenient to express the relationship between two variables x

and y through parametric equations governed by a third variable known as
a parameter. To illustrate this idea, consider the parametric equations

x(t) = cos t and y(t) = sin t (1.187)

where t ∈ [0, π] is a parameter. For every value of t we can compute x(t)

and y(t), the parametric equations therefore describe an implicit relationship
between x and y. Indeed, in this case the parameterisation implies that

x2 + y 2 = cos2 t + sin2 t = 1, (1.188)

i.e., (x, y) are points on the circumference of a unit circle (see figure 1.17).

Figure 1.17: Unit semi-circle parameterised by x(t) = cos t, y(t) = sin t, with t ∈ [0, π].

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 55

Example 1.31 The position (x, y) of a particle at time t ∈ R is given by the

parametric equations

x(t) = a cos(ωt) and y(t) = b sin(ωt) (1.189)

where a, b, and ω are constants. Show that particle follows an elliptical path.

I Solution: Observe that x and y are related by

 x 2  y 2
+ = cos2 (ωt) + sin2 (ωt). (1.190)
a b

Hence, letting θ = ωt, we have by the identity cos2 θ + sin2 θ ≡ 1 that

 x 2  y 2
+ = 1. (1.191)
a b
This is the equation of an ellipse with semi-major axis a and semi-minor axis
b; the particle therefore follows an elliptical path (see figure 1.18).

Note: By the periodicity of the trigonometric functions we have

cos(ωt + 2π) = cos(ωt) and sin(ωt + 2π) = sin(ωt), (1.192)

it then follows from these equations that

2π
x(t + T ) = x(t) and y(t + T ) = y(t), where T = . (1.193)
ω
Thus, the particle returns to its original position when t is incremented by T ,
i.e., T is the time period for the particle to complete an orbit of the ellipse. J

Figure 1.18: Elliptical path traced out in the anticlockwise sense.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

56 Section 1 : Functions and Algebra

1.13 Plane Polar Coordinates

Let P be a point in the Cartesian plane, then P can be specified uniquely by

its Cartesian coordinates (x, y), as shown in figure 1.19. The coordinates
(x, y) are said to be an ordered pair in the sense that the order is important,
e.g., (2, 1) 6= (1, 2). Since two real numbers R are used to define a point, we
sometimes write (x, y) ∈ R2 , where R2 is shorthand for the Cartesian plane.

Consider the line from the origin O to P . Let r denote the length of this line,
and let θ denote the angle that the line makes with the positive x-axis. We call
r the radial coordinate of P , and θ the azimuthal coordinate of P ; in this
way P may be expressed in plane polar coordinates as (r, θ). By inspection
of figure 1.19 we see that polar coordinates are related to Cartesians by

x = r cos θ and y = r sin θ, with r 2 = x2 + y 2 (1.194)

By restricting r and θ to the set of values r ∈ [0, ∞), and θ ∈ [0, 2π), the
plane polar coordinates (r, θ) of a point are unique.
√
Example 1.32 A point P has plane polar coordinates (r, θ) = (2 2, 3π
4
);
what are the Cartesian coordinates of P ?

I Solution: According to equation (1.194) we have

√ √
x = 2 2 cos( 3π
4
) = −2 and y = 2 2 sin( 3π
4
) = 2. (1.195)

Thus, the Cartesian coordinates of P are (x, y) = (−2, 2). J

Figure 1.19: Point P in plane polar coordinates specified by P = (r, θ).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 1 : Functions and Algebra 57

Example 1.33 Polar coordinates and Cartesians are related according to equa-
tions (1.194) by x = r cos θ and y = r sin θ. Show that r2 = x2 + y 2 .

I Solution: Using the identity cos2 θ + sin2 θ = 1 we obtain

x2 + y 2 = r2 (cos2 θ + sin2 θ) = r2 (1.196)

as required. J

Now suppose that we have some function f (θ). If we set

r = f (θ), (1.197)

then the function f (θ) may be depicted using plane plane polar coordinates by
the curve produced by the locus of points (r, θ), with r = f (θ). Curves of this
kind are sometimes called polar curves; classic examples include:

r(θ) = a, (a circle) (1.198a)

r(θ) = 2a(1 − cos θ), (a cardioid) (1.198b)
r(θ) = aθ + b, (an Archimedean spiral) (1.198c)
r(θ) = a sin(nθ), (a rhodonea curve, or rose) (1.198d)

where a, b ∈ R, and n ∈ Z are constants (see figures 1.20 and 1.21).

It is sometimes possible to convert between polar form and Cartesian form using
the fact that the two coordinate systems are related by equations (1.194).

2 2

1 1

0 0

-1 -1

-2 -2
-2 -1 0 1 2 -3 -2 -1 0 1
(a) (b)

Figure 1.20: Example polar curves: (a) a unit circle r = 1; (b) a Cardioid r = 2(1−cos θ).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

58 Section 1 : Functions and Algebra

2 2

1 1

0 0

-1 -1

-2 -2
-2 -1 0 1 2 -2 -1 0 1 2
(a) (b)

Figure 1.21: More example polar curves: (a) an Archimedean spiral; (b) a rhodonea curve.

Example 1.34 A curve C is defined in polar form by

r = 2 cos θ. (1.199)

Express the curve in Cartesian form. What is this curve?

I Solution: The basic strategy when attempting problems of this kind is to

substitute for the trigonometric terms using the fact that x = r cos θ and
y = r sin θ; it is also often useful to note that x2 + y 2 = r2 .

Multiplying equation (1.199) by r we have

r2 = 2r cos θ. (1.200)

Thus, noticing that r2 = x2 + y 2 , and x = r cos θ we may write

x2 + y 2 = 2x. (1.201)

It is not obvious what curve this is; however, progress may be made by adding
unity (1) to both sides to give

x2 + y 2 + 1 = 2x + 1, that is, (x2 − 2x + 1) + y 2 = 1. (1.202)

Hence, factorising the first term on the left-hand-side, we obtain

(x − 1)2 + y 2 = 1. (1.203)

This is the Cartesian equation for a circle of radius 1 centred at (1, 0). J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 Engineering Mathematics
J. J. Bissell

Differentiation

The modus operandi of scientific enquiry could justifiably be summarised as

the study of how quantities vary with respect to one-another; or put another
way, the study of rates-of-change. In mathematical notation, rates-of-change
are expressed using derivatives, and the process of finding derivatives is known
as differentiation. Differentiation (and differential calculus) forms one branch
of an enormously important field of mathematics known as the calculus.

Our goal in this section is to review the main ideas and rules of differentiation,
so that these can be applied with confidence to problems in engineering and
the physical sciences. As with our work on functions, much of what follows
is revision of school mathematics, although we shall go somewhat deeper by
considering foundational aspects, such as limits and continuity. By the end of
the section you should be able to. . .

Learning outcomes:

◦ Determine the limit lim f (x) of a function f (x) as x approaches a value a.

x→a
◦ Use the product and quotient rules for evaluating simple limits.
◦ Describe whether a function is continuous or discontinuous at a point.
◦ Write down the definition for the first derivative of a function.
◦ Determine the derivatives of simple functions from first principles.
◦ Use the product, chain, and quotient rules to evaluate simple derivatives.
◦ Determine the derivatives of simple parametric functions.
◦ Define the second order derivative of functions.
◦ Use differentiation to identify and classify the stationary points of a function.
◦ Determine the Maclaurin and Taylor expansions of simple functions.

These learning outcomes must be reinforced by completing course exercises.

59
60 Section 2 : Differentiation

2.1 Limits and Continuity

The formal study of the foundations of differentiation is known as analysis,

and though we shall not be studying analysis proper, we will need to draw upon
some of its ideas to better understand what differentiation means. We begin
here by introducing some basic concepts about limits and continuity.

2.1.1 Limits

The function f (x) depicted in figure 2.1 is defined for x ∈ R by the rule

2x for x ≤ 1
f (x) = (2.1)
x for x > 1;

because this rule is different for different parts of the domain, f (x) is known
as a compound function. Notice that as we approach x = 1 from the left
(i.e., with x < 1) that f (x) approaches the value f (x) = 2. In mathematical
notation we write this as the left-hand-limit

lim f (x) = 2, (2.2)

x→1−

and say that ‘the limit of f (x) as x approaches 1 from the left (−) is 2’. [The
minus-sign (−) superscript on 1− signifies that the limit is left-handed.]
Similarly, as we approach x = 1 from the right (wwith x > 1), the function
f (x) approaches the value f (x) = 1. In this case we have a right-hand-limit

lim f (x) = 1, (2.3)

x→1+

and say that ‘the limit of f (x) as x approaches 1 from the right (+) is 2’.
[The plus-sign (+) superscript on 1+ signifies that the limit is right-handed.]
Observe here that the left and right-hand limits of f (x) at x = 1 are not
equal; this is because there is a ‘break’ in the function at x = 1, i.e., we are
concerned with the limiting behaviour of f (x) as x → 1 from either the left
or right, rather than the value of the function f (x) at x = 1. Indeed, this
distinction is most clear in equation (2.3): the limit of f (x) as x → 1+ is 1,
but the value of the function at x = 1 is f (1) = 2. To emphasise:

In general, the limit of the function f (x) at a point x = a is not

identical to the value of f (a) unless f (x) is continuous (see §2.1.2).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 61

Figure 2.1: Function f (x) of equation (2.1).

Now consider the behaviour of f (x) as x → 2, in this case we find that the
left and right-hand limits are identical, viz.

lim f (x) = lim+ f (x) = 2, (2.4)

x→2− x→2

(there is no ‘break’ in the function at x = 2). In this instance, therefore, we

can write
lim f (x) = 2 (2.5)
x→2

as the limit of f (x) as x → 2, where we do not use the ‘±’ notation, because
it is implied that the limit can be taken from either direction. More generally:

Definition 2.1 (Informal) Let f (x) be a function defined on an interval

containing the point a ∈ R. If f (x) approaches some value L− as x
approaches a from the left (x < a), then the left-hand-limit of f (x) is

lim f (x) = L− . (2.6)

x→a−

Similarly, if f (x) approaches some value L+ as x approaches a from the

right (x > a), then the right-hand limit of f (x) is

lim f (x) = L+ . (2.7)

x→a+

If the left and right-hand limits are equal (i.e., if L+ = L− ) then we say
that the limit of f (x) as x → a is L, and write

lim f (x) = L, where L = L+ = L− . (2.8)

x→a

If L+ 6= L− at a, then we say that the limit of f (x) does not exist at a.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

62 Section 2 : Differentiation

Note that when evaluating the limit of a function f (x), neither the limiting
value of x, nor the value of the limit itself need be finite (see Example 2.1).

Example 2.1 Evaluate the following limits:

1 1
(a) lim x2 ; (b) lim e−x ; (c) lim− ; (d) lim+ .
x→2 x→∞ x→0 x x→0 x

I Solution: We take each of the limits in turn (cf. figure 2.2).

(a) As x approaches 2 we have that x2 approaches 4, i.e.,

lim x2 = 4. (2.9)
x→2

(b) As x tends towards infinity, the function e−x vanishes, i.e.,

lim e−x = 0. (2.10)

x→∞

1
(c) As x < 0 approaches 0 from the left, x
diverges negatively, i.e.,

1
lim− = −∞. (2.11)
x→0 x
1
(d) As x > 0 approaches 0 from the right, x
diverges positively, i.e.,

1
lim+ = +∞. (2.12)
x→0 x
Graphs of the functions associated with the above limits are depicted in fig-
ure 2.2. In formal analysis one cannot appeal to the graph of a function to
determine a limit, but graphs are nonetheless useful for illustrative purposes.J

4 6 4

3 2
4

2 0

2
1 -2

0 0 -4
-2 -1 0 1 2 -2 0 2 4 -4 -2 0 2 4
(a) (b) (c/d)

Figure 2.2: Graphs of the functions associated with limits in Example 2.1.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 63

2.1.2 Continuity

We began our discussion in §2.1.1 by considering a compound function f (x)

which had a ‘break’ at x = 1, whereas elsewhere the function exhibited no
‘breaks’. More formally, we refer to these features in terms of continuity:

Definition 2.2 A function f (x) is said to be continuous at a point x = a

if and only if

lim f (x) = lim+ f (x) = lim f (x) = f (a). (2.13)

x→a− x→a x→a

If a function is not continuous at a point a, then it is discontinuous at

a. Notice that if f (x) is not defined at the point x = a, then f (a) is not
meaningful; hence, f (x) must be defined at x = a to be continuous there.

Exercise 2.1 State whether the function defined by


2x for x ≤ 1
f (x) = (2.14)
x for x > 1

is continuous or discontinuous at: (a) x = 1; and (b) x = 3.

I Solution: (a) With reference to our discussion at the start of 2.1.1 we have

lim f (x) = 2 6= 1 = lim+ f (x). (2.15)

x→1− x→1

It follows by Definition 2.2 that f (x) is discontinuous at x = 1.

(b) Evaluating the left and right-hand limits at x = 3 we find that

lim f (x) = 3 and lim f (x) = 3. (2.16)

x→3− x→1+

It follows that the limit of f (x) at x = 3 is

lim f (x) = 3 = f (3), (2.17)

x→3

i.e., f (x) is continuous at x = 3. J

1
Exercise 2.2 State whether the function f (x) = x
is continuous at x = 0.

I Solution: The function f (x) is discontinuous at x = 0 (cf. figure 2.2). J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

64 Section 2 : Differentiation

2.1.3 Rules for evaluating limits

The following rules frequently prove useful when evaluating limits:

Result 2.1 Let f (x) and g(x) be continuous functions whose limits at the
point x = a are

lim f (x) = f (a) and lim g(x) = g(a). (2.18)

x→a x→a

The standard rules for evaluating the limits of sums, products, and quo-
tients of these functions are then
 
sum rule: lim f (x) + g(x) = f (a) + g(a); (2.19a)
x→a
 
product rule: lim f (x)g(x) = f (a)g(a); (2.19b)
x→a
 
quotient rule: lim f (x)/g(x) = f (a)/g(a). (2.19c)
x→a

Note: If f (a) and g(a) are both zero or infinity, then f (a)/g(a) is not
well defined, and is said to be in indeterminate form; in these cases the
quotient rule fails, and other methods must be used (see §2.6).

We shall not prove these results here, but instead illustrate their application.

Example 2.2 Evaluate the following limits:

(a) lim (x2 − sin x); (b) lim xe−x .

x→π x→0

I Solution: (a) By the sum rule for limits we have

   
lim (x − sin x) = lim x − lim sin x = π 2 + 0 = π 2 .
2 2
(2.20)
x→π x→π x→π

(b) By the product rule for limits we have that

   
lim xe−x = lim x × lim e−x = 0 × e0 = 0. (2.21)
x→0 x→0 x→0

It may be shown that the functions f (x) = (x2 − sin x) and g(x) = xe−x are
continuous, so these results are consistent with f (π) = π 2 and g(0) = 0. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 65

Example 2.3 Let f (x) = (x2 − 9) and g(x) = (x − 3); evaluate the limit

lim [f (x)/g(x)] = L. (2.22)

x→3

I Solution: Notice here that both lim f (x) = 0 and lim g(x) = 0, meaning
x→3 x→3
that the quotient rule will not work (see Result 2.1). However, notice that

f (x) (x + 3)(x − 3)
= = (x + 3), for x 6= 3. (2.23)
g(x) (x − 3)

In this way we have that

lim [f (x)/g(x)] = lim (x + 3) = 6. (2.24)

x→3 x→3

Here division by (x − 3) is acceptable when taking the limit: even though we

are interested in what happens as x gets closer and closer to 3, we implicitly
assume that x is not equal to 3. J

2.1.4 Asymptotic limits (optional)

Often we are faced with algebraic expressions that cannot be solved analytically,
but which nevertheless yield useful and exact asymptotic limits. For example,
Bissell1 has investigated a convection system described by parameters C, CB ,
P1 , RS and Rc . The system exhibits ‘over-stability’ whenever C exceeds a
threshold CT (P1 ) (see figure 2.3), where CT is given by the implicit relation
s
2π 2 CT3 (P1 + 1)
CT2 − = . (2.25)
Rc 4Rc P12

In general this expression must be solved numerically for CT ; however, analytical

insight can be gained by considering the asymptotic (limiting) dependence, i.e.,

1 2π 2
lim CT (P1 ) = √ = ∞, and lim CT (P1 ) = . (2.26)
P1 →0 2 Rc P1 P1 →∞ Rc

Thus, one can check the accuracy of any numerical solution for CT (P1 ) by
ensuring that the numerical solution approaches these limits as P → 0 and
P → ∞. A graph of numerical values for CT (P1 ) is plotted in figure 2.3
alongside dash-dotted curves for the asymptotic limits, or asymptotes.
1
See J. J. Bissell, Proceedings of the Royal Society, Series A, 471:20140845 (2015).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

66 Section 2 : Differentiation

1
10
C T ( P 1)
C B ( P 1)

0
RS < Rc
10
C

−1
RS > Rc
10

−2
10 −2 −1 0 1 2
10 10 10 10 10
P1

Figure 2.3: Numerical values for CT (solid curve) with asymptotes (dash-dotted curves).

2.2 Foundations of Differentiation

Differentiation is the mathematical process of determining how quickly a func-

tion f (x) changes as its argument changes, i.e., it is a way of answering the
question ‘if I change x by a small amount, then how much does f (x) change?’,
or ‘what is the rate-of-change of f (x) with x?’. In this way differentiation is
key to formulating mathematical descriptions (or models) of physical processes
when one quantity varies with respect to another.
Instinctively, the terminology ‘rate-of-change’ encourages us to think of differ-
entiation in time-dependent contexts, and yet differentiation is not restricted
to temporal phenomena: for example, in a mechanical system we may be in-
terested in the rate-of-change of velocity v(t) with respect to time t (that
is, acceleration), while in thermodynamic system we might want to know the
rate-of-change of pressure P (T ) with respect to changes in temperature T .
In this section we consider the foundations of differentiation as a limiting pro-
cess (i.e., differentiation by first principles), and discuss the relationship be-
tween differentiation, and the gradient of a function.

2.2.1 Differentiation from first principles

The basic idea behind differentiation is well illustrated by the problem of deter-
mining the instantaneous velocity v of an object moving in a straight line. Let

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 67

us suppose that the velocity varies with time t according to some function v(t),
then we can denote the position—or displacement—of the particle at time t as
s(t). Now let δt represent a short period of time; it follows that the distance
δs travelled by the particle during the time interval [t, t + δt] is

δs = s(t + δt) − s(t). (2.27)

By definition, therefore, the average velocity hvi of the particle during the time
interval [t, t + δt] is

change in displacement δs s(t + δt) − s(t)

hvi = = = . (2.28)
change in time δt δt

Since hvi is the average velocity of the particle during the time interval [t, t+δt],
to compute the instantaneous velocity v(t) of the particle at time t we simply
need to let the width δt of the interval vanish, i.e.,
   
δs s(t + δt) − s(t) ds
v(t) = lim = lim ≡ . (2.29)
δt→0 δt δt→0 δt dt

This limit, which we denote in shorthand by ds/dt, is called the first derivative
of s(t) with respect to t; evaluating ds/dt is known as differentiating s(t).

Definition 2.3 Let f (x) be a function defined on an interval containing

x, then the first derivative (or rate-of-change) of f (x) at x is

df δf f (x + δx) − f (x)
≡ lim ≡ lim , (2.30)
dx δx→0 δx δx→0 δx
provided this limit exists. [If the limit does not exist, then f (x) is not
differentiable at x.] Derivatives are also denoted using the ‘prime’ notation

df
f 0 (x) ≡ . (2.31)
dx
Furthermore, in operator notation the differential operator d/dx can be
used to write a derivative as
d df
(f ) ≡ , (2.32)
dx dx
where the left-hand-side means ‘the derivative of f with respect to x’.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

68 Section 2 : Differentiation

Differentiating a function f (x) by evaluating the limit in Definition 2.3 is known

as differentiation by first principles. We illustrate this idea below.
Example 2.4 The function f (x) is defined by f (x) = kx, where k is a con-
stant. Evaluate the first derivative of f (x) by first principles.

I Solution: According to Definition 2.3 we have with f (x) = kx that

df f (x + δx) − f (x)
= lim
dx δx→0 δx
k(x + δx) − kx
= lim
δx→0 δx
kδx
= lim
δx→0 δx

= lim k
δx→0

= k. (2.33)

Hence, the first derivative of f (x) = kx with respect to x is f 0 (x) = k. J

Example 2.5 Let f (x) be the function defined by f (x) = x2 . Evaluate the
first derivative of f (x) by first principles.

I Solution: According to Definition 2.3 we have with f (x) = x2 that

 
df f (x + δx) − f (x)
= lim
dx δx→0 δx
(x + δx)2 − x2
 
= lim
δx→0 δx
 2
x + 2xδx + (δx)2 − x2

= lim
δx→0 δx
2xδx + (δx)2
 
= lim
δx→0 δx
= lim [2x + δx]
δx→0

= 2x. (2.34)

Hence, the first derivative of f (x) = x2 with respect to x is f 0 (x) = 2x.

Note: In operator notation we write

d 2
(x ) = 2x, (2.35)
dx
where the left-hand-side means ‘the derivative of x2 with respect to x’. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 69

Example 2.6 It may be shown that

sin(δx/2)
lim = 1. (2.36)
δx→0 δx/2

Use this result to find the first derivative of f (x) = sin x by first principles.
[Hint: You may wish to use the identity sin A−sin B ≡ 2 sin( A−B 2
) cos( A+B
2
).]

I Solution: By Definition 2.3 we have with f (x) = sin x that

 
df f (x + δx) − f (x)
= lim
dx δx→0 δx
 
sin(x + δx) − sin(x)
= lim . (2.37)
δx→0 δx

Now, according to the hint we have

   
x + δx − x x + δx + x
sin(x + δx) − sin(x) ≡ 2 sin cos
2 2
   
δx δx
≡ 2 sin cos x + . (2.38)
2 2

Thus, putting this identity into equation (2.37) we obtain

 
df 2 sin (δx/2) cos (x + δx/2)
= lim
dx δx→0 δx
   
δx sin (δx/2)
= lim cos x + ×
δx→0 2 δx/2
    
δx sin (δx/2)
= lim cos x + × lim
δx→0 2 δx→0 δx/2
| {z }
=1 (by equation (2.36))
  
δx
= lim cos x +
δx→0 2
= cos(x), (2.39)

where we used the product rule for limits. Hence, the derivative of f (x) = sin x
is f 0 (x) = cos x.
Note: A similar argument may be used to show that

d
(cos x) = − sin x. (2.40)
dx
The derivatives of other trigonometric functions are described in §2.3. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

70 Section 2 : Differentiation

2.2.2 Derivatives of standard functions

It will be apparent from our examples in the previous section that differentiating
functions from first principles is often a laborious task, and for this reason it is
more common to appeal to standard results, rather than Definition 2.3. Table
2.1 lists the derivatives of some commonly occurring functions, all of which
may (in principle) be determined by Definition 2.3. The results from this table
may then be used to evaluate derivatives in specific cases (see Example 2.7).

Example 2.7 Use table 2.1 to determine the first derivatives of:

(a) x4 ; (b) e2t ; (c) cos(5x); (d) arctan(2θ). (2.41)

I Solution: Appealing to table 2.1 we have:

d
(a) By row three, column one: dx (x4 ) = 4x3 .
d
(b) By row two, column two: dt (e2t ) = 2e2t .
d
(c) By row four, column one: dx cos(5x) = −5 sin(5x).
d 1/2 2
(d) By row six, column two: dθ arctan(2θ) = 1/4+θ 2 = 1+4θ 2 .

Note: In part (d) we used the fact that arctan(2θ) = arctan(θ/ 12 ). J

Function Derivative Function Derivative

df
f 0 (x) d2 f
f (x) f 0 (x) ≡ f 00 (x) ≡
dx dx2
a 0 eax aeax
1
xn nxn−1 loge (x + a)
x+a
1
sin(ax) a cos(ax) arcsin(x/a) √
a − x2
2
1
cos(ax) −a sin(ax) arccos(x/a) −√
a − x2
2
a
tan(ax) a sec2 (ax) arctan(x/a)
a 2 + x2
1
sinh(ax) a cosh(ax) arcsinh(x/a) √
a + x2
2
1
cosh(ax) a sinh(ax) arccosh(x/a) √
a − x2
2
a
tanh(ax) a sech2 (ax) arctanh(x/a)
a − x2
2

Table 2.1: Derivatives of some common functions, where a and n are taken as constants.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 71

2.2.3 Gradients and tangents

Geometrically it is common to interpret the derivative of a function f (x) as the

gradient (or ‘slope’) of the tangent to the curve y = f (x). To illustrate how
this works, consider the curve y = f (x) depicted in figure 2.4, and let P be a
point on y = f (x) with coordinates (x, y). In this figure the straight solid line
through P is the tangent to the curve at P , where by ‘tangent’ we mean ‘the
line through P that has the same gradient (or ‘slope’) as the curve at P ’.

Now consider a second point Q further along the curve, and with coordinates
(x + δx, y + δy), where

δy = f (x + δx) − f (x). (2.42)

The straight dashed line through P and Q is called a secant because it ‘cuts’
the curve. [In Latin the word secans means ‘cutting’ (thus secant), and the
word tangens means ‘touching’ (thus tangent).] The gradient of the secant is

δy f (x + δx) − f (x)
= . (2.43)
δx δx
Now, as δx is decreased, the point Q approaches the point P , such that the
secant (dashed line) gets closer-and-closer to coinciding with the tangent (solid
line). Indeed, in the limit that δx → 0, the secant and the tangent are identical.
It follows by equation (2.43), therefore, that the gradient of the tangent is

δy f (x + δx) − f (x) df
lim = lim ≡ , (2.44)
δx→0 δx δx→0 δx dx
i.e., the gradient of the curve y = f (x) at x is f 0 (x).

Figure 2.4: Tangent to f (x) at P (solid line), and a secant through P and Q (dashed).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

72 Section 2 : Differentiation

Example 2.8 The function f (x) = x2 has first derivative

f 0 (x) = 2x. (2.45)

Calculate the gradient of the function at the point P = (3, 9). Hence determine
the equation for the tangent to f (x) that passes through P .

I Solution: At the point P = (3, 9) we have x = 3; hence, the gradient of the

function f (x) at P is f 0 (3) = 2 × 3 = 6.

Now the tangent is a straight line, so it has an equation of the form

y = mx + c, (2.46)

where m is the gradient, and c is the line’s intercept with the y-axis. By
definition, the gradient of the tangent to f (x) at the point P is the same as
the gradient of the function f (x) at P . Hence, we have

m = f 0 (3) = 6. (2.47)

The constant c is found by the fact that the tangent passes through P , where
x = 3 and y = 9. Thus, with m = 6, we require by equation (2.46) that

9 = 6 × 3 +c,
|{z} that is, c = −9. (2.48)
| {z }
y mx

The equation of the tangent at P is therefore y = 6x − 9 (see figure 2.5). J

24

21

18

15

12

0
-5 -4 -3 -2 -1 0 1 2 3 4 5

Figure 2.5: Tangent y = 6x − 9 to the function f (x) = x2 at the point P = (3, 9).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 73

2.2.4 Differential element

The ‘d’ notation in the derivative of a function refers explicitly to the idea
of a differential element, or infinitesimally small change (a vanishingly
small change). In particular, we say that df is the differential of f and dx
is the differential of x. Indeed, we can see how differentiation corresponds to
rates-of-change by noting that the infinitesimal change df in f caused by an
infinitesimal change dx is

df
df = f 0 (x)dx ≡ dx. (2.49)
dx
This notation suggests that derivatives work in similar way to fractions; how-
ever, this interpretation is not quite right because the notation df /dx is defined
as the limit of a ratio (see Definition 2.3), not as a fraction proper.

2.3 Rules for Differentiation

It is often necessary to determine the derivate of various combinations of the

functions listed in table 2.1 (for example, the sums, and products of such
functions). As we illustrate below, rather than working from the definition of
the derivative, differentiation of combinations of functions can be accomplished
using various rules.

2.3.1 Linearity of differentiation

Let u(x) and v(x) be two functions, then any expression f (x) of the form

f (x) = au(x) + bv(x), (2.50)

where a and b are constants, is called a linear combination of u(x) and v(x).
It may be shown that derivative of such a sum is equal to the sum of the
derivatives; that is, differentiation is a linear operation:

Result 2.2 Differentiation is a linear operation, i.e., given two functions

u(x) and v(x), and two constants a and b, it may be shown that

d du dv
[au(x) + bv(x)] = a +b . (2.51)
dx dx dx

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

74 Section 2 : Differentiation

Example 2.9 Determine the first derivative of f (x) = (3x2 − 2x + cos x).

I Solution: By the linearity of differentiation we have

df d  2 
= 3x − 2x + cos x
dx dx
d d d
= (3x2 ) − (2x) + (cos x)
dx dx dx
= 6x − 2 − sin x, (2.52)

where we evaluated the derivatives using the standard results in table 2.1. J

2.3.2 Chain rule

It is often necessary to determine the derivative of a function of the form

y(x) = f (u) were u = u(x). (2.53)

For example, the function y(x) = cos(x2 ) may be written in the composite
form y = f (u) = cos(u), with u(x) = x2 . To differentiate a function of this
kind we proceed as follows. Suppose that x changes by δx, then u(x) changes
by δu, and f (u) changes by δf , where

δu = u(x + δx) − u(x), and δf = f (u + δu) − f (u). (2.54)

In this way we may write

δf δf δu f (u + δu) − f (u) u(x + δx) − u(x)

= = × . (2.55)
δx δu δx δu δx
As δx → 0 these ratios become derivatives, i.e.,

df δf δf δu δf δu df du
≡ lim = lim = lim × lim = , (2.56)
dx δx→0 δx δx→0 δu δx δu→0 δu δx→0 δx du dx
where δu → 0 because δx → 0. This expression is known as the chain rule.

Result 2.3 Let y = f (u(x)), where f is a function of u, and u(x) is a

function of x. The chain rule states that
dy df df du
= = . (2.57)
dx dx du dx

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 75

Example 2.10 Determine the first derivative of y(x) = sin(x4 ).

I Solution: The function of interest is effectively

y = f (u) = sin u, where u(x) = x4 . (2.58)

Hence, according to the chain rule we have

df df du
y 0 (x) = = = (cos u) × (4x3 ) = 4x3 cos(x4 ), (2.59)
dx du dx
where we used the fact that
d d 4
(sin u) = cos u and (x ) = 4x3 . (2.60)
du dx
Notice that it is essential to write the expression for f 0 (x) in terms of x only.J
Example 2.11 Determine the first derivative of v(t) = (2t + 1)3 .

I Solution: The function of interest is effectively

v = f (z) = z 3 , where z(t) = (2t + 1). (2.61)

Since f 0 (z) = 3z 2 and z 0 (t) = 2, according to the chain rule we have

df df dz
v 0 (t) = = = 3z 2 × 2 = 6(2t + 1)2 , (2.62)
dt dz dt
As ever, we have written the expression for v 0 (t) in terms of t only. J
Example 2.12 Determine the first derivative of y(x) = sin(exp(x2 )).

I Solution: In this example the function of interest is effectively

y = f (g) = sin(g), where g(h) = exp(h), with h(x) = x2 . (2.63)

To handle a problem of this kind we must apply the chain rule repeatedly, i.e.,

df df dg df dg dh
y 0 (x) = = = . (2.64)
dx dg dx dg dh dx

Thus, evaluating the respective derivatives we obtain

df
y 0 (x) = = 2xeh cos(g) = 2x exp(x2 ) cos(exp(x2 )), (2.65)
dx

where we used the fact that f 0 (g) = cos(g), g 0 (h) = eh , and h0 (x) = 2x. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

76 Section 2 : Differentiation

2.3.3 Products and quotients

Consider the function f (x) formed by the product of two functions u(x) and
v(x), that is,
f (x) = u(x)v(x). (2.66)

If x changes by an amount δx, then f (x) changes by an amount δf , where

δf = f (x + δx) − f (x)
= u(x + δx)v(x + δx) − u(x)v(x)
= u(x + δx)v(x + δx) − u(x + δx)v(x) + u(x + δx)v(x) − u(x)v(x).
= u(x + δx) [v(x + δx) − v(x)] + v(x) [u(x + δx) − u(x)] . (2.67)

Now, dividing by δx we obtain

   
δf v(x + δx) − v(x) u(x + δx) − u(x)
= u(x + δx) + v(x) . (2.68)
δx δx δx

Thus, taking the limit as δx → 0 we have

df δf
= lim
dx δx→0
δx   
v(x + δx) − v(x) u(x + δx) − u(x)
= lim u(x + δx) + lim v(x)
δx→0 δx δx→0 δx
   
v(x + δx) − v(x) u(x + δx) − u(x)
= u(x) lim + v(x) lim
δx→0 δx δx→0 δx
dv du
= u(x) + v(x) . (2.69)
dx dx
This result is known as the product rule for differentiation.

Result 2.4 Let f (x) = u(x)v(x), then the product rule states that

df d dv du
= (uv) = u +v . (2.70)
dx dx dx dx
This rule may be used in combination with the chain rule to derive the
quotient rule
d  u  vu0 − uv 0
= , (2.71)
dx v v2
where u0 = du/dx and v 0 = dv/dx (see Example 2.14).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 77

Example 2.13 Determine the first derivative of f (x) = 2 sin x cos x.

I Solution: By the product rule we have

d
f 0 (x) = (2 sin x cos x)
dx
d d
= 2 cos x (sin x) + sin x (2 cos x)
dx dx
= 2 cos x cos x − 2 sin x sin x
= 2(cos2 x − sin2 x). (2.72)

Note: By the double angle formula we have sin(2x) ≡ 2 sin x cos x, and
cos(2x) ≡ (cos2 x − sin2 x). As expected, therefore, this result is equivalent
to saying that f (x) = sin(2x) with f 0 (x) = 2 cos(2x). J
Example 2.14 We now illustrate how to derive the quotient rule. Let

u(x) 1
f (x) = with v(x) = , (2.73)
v(x) w(x)

where u(x), v(x), and w(x) are all functions of x.

(a) Show that f 0 (x) = u(x)w0 (x) + w(x)u0 (x).
(b) Given that v = 1/w, use the chain rule to show that w0 (x) = −v 0 /v 2 .
(c) Hence express f 0 (x) in terms of u(x), u0 (x), v(x), and v 0 (x).

I Solution: (a) Because w(x) = 1/v, the product rule yields

d u d du dw
f 0 (x) = = (uw) = w +u . (2.74)
dx v dx dx dx
(b) Since w = v −1 , we have that

dw d −1 1
= (v ) = −v −2 = − 2 (2.75)
dv dv v
Thus, according to the chain rule

dw dw dv 1 dv v0
= =− 2 = − 2. (2.76)
dx dv dx v dx v
(c) Putting equation (2.76) into equation (2.74) we obtain

0 d u du v0 u0 v0 vu0 − uv 0
f (x) = =w −u 2 = −u 2 = , (2.77)
dx v dx v v v v2
where we substituted for w = 1/v. This result is called the quotient rule. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

78 Section 2 : Differentiation

2.3.4 Implicit differentiation

When dealing with implicit functions of the form

f (x, y) = constant (2.78)

it usually necessary to differentiate the entire expression, and then rearrange

for the derivative of interest. We illustrate this idea by example.

Example 2.15 Let y(x) and x be two variables related by

x4 − 3xy + y 3 = 2. (2.79)

Determine the first derivative of y with respect to x.

I Solution: Differentiating equation (2.79) with respect to x we have

d d
x4 − 3xy + y 3 =


(2) = 0. (2.80)
dx dx
Evaluating the left-hand-side of this equation we find

d d 4 d d 3
x4 − 3xy + y 3 =


(x ) − (3xy) + (y )
dx dx dx dx
d d d 3
= 4x3 − 3x (y) − y (3x) + (y )
dx dx dx
dy dy
= 4x3 − 3x − 3y + 3y 2 , (2.81)
dx dx
where we used the chain rule to write
d 3 d 3 dy dy
(y ) = (y ) = 3y 2 . (2.82)
dx dy dx dx

Thus, putting equation (2.81) into equation (2.80) we obtain

dy dy
4x3 − 3x − 3y + 3y 2 = 0, (2.83)
dx dx
whereupon solving for y 0 (x) we have

dy 3y − 4x3
y 0 (x) = = 2 . (2.84)
dx 3y − 3x

Since y is implicitly a function of x, this is an implicit expression for y 0 (x). J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 79

2.3.5 Inverse function rule

Let y = f (x) be some function of x, such that

y = f (x) and x = f −1 (y) = g(y), (2.85)

where g is the inverse function of f . If we differentiate x = g(y) with respect

to x, then we have by the chain rule that

d dg dy
g(y) = ; (2.86)
dx dy dx

however, because x = g(y) we may replace g with x to give

d dx dy dx dy
(x) = that is, 1= . (2.87)
dx dy dx dy dx

This equation is known as the inverse function rule.

Result 2.5 Let y = f (x) have inverse function f −1 such that x = f −1 (y);
then the inverse function rule states that
 −1
dx dy dx dy
1= , i.e., = . (2.88)
dy dx dy dx

This rule is useful for determining the derivatives of inverse functions.

Example 2.16 Determine the derivative of y(x) = arcsin(x).

I Solution: Observe that if y(x) = arcsin(x), then

dx
q √
x(y) = sin y, with = cos y = 1 − sin2 y = 1 − x2 (2.89)
dy

where we used the identity cos2 y + sin2 y ≡ 1. Hence, by the inverse function
rule we obtain
 −1
dy dx 1 d 1
= =√ , i.e., arcsin(x) = √ . (2.90)
dx dy 1 − x2 dx 1 − x2

Note: A similar method may be used to differentiate f (x) = arccos(x). J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

80 Section 2 : Differentiation

Example 2.17 Let y(x) = loge x; show that y 0 (x) = 1/x.

I Solution: Observe that if y(x) = loge x, then

dx
x(y) = ey , with = ey = x (2.91)
dy

Hence, by the inverse function rule we obtain

 −1
0 dy dx 1 d 1
y (x) = = = , i.e., loge x = (2.92)
dx dy x dx x

as required. J

2.3.6 Parametric differentiation

Suppose that we have two variables y and x related implicitly through some
parametric equations, that is,

y = y(t) and x = x(t), (2.93)

where t is the parameter (see §1.12). By the chain rule and inverse function
rule we have that
 −1
dy dy dt dy dx dy/dt
= = = . (2.94)
dx dt dx dt dt dx/dt

Expression allows us to determine the rate-of-change of one variable with re-

spect to another when the variables are related parameterically.
Example 2.18 An ellipse is given by the parametric equations

x(t) = a cos(ωt) and y(t) = b sin(ωt), (2.95)

where t ∈ R is a parameter, and a, b, and ω are constants. Determine the rate

of change of y with respect to x.

I Solution: Differentiating y(t) and x(t) we have y 0 (t) = bω cos(ωt) and

x0 (t) = −aω sin(ωt). Hence, the rate-of-change of y with respect to x is
 
dy dy/dt bω cos(ωt) b
= =− =− cot(ωt), (2.96)
dx dx/dt aω sin(ωt) a

where we used the fact that cos(ωt)/ sin(ωt) = 1/ tan(ωt) = cot(ωt). J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 81

2.4 Higher Order Derivatives

The first derivative f 0 (x) of a function f (x) is obtained by differentiating

f (x). In this section we introduce the idea of differentiating multiple times
(i.e., differentiating derivatives) to obtain derivatives of arbitrary order.

2.4.1 Second-order derivative

The second-order derivative of a function is the derivative of the derivative.

Definition 2.4 The second derivative of a function f (x) is denoted

d2 f /dx2 , and given by the derivative of the first derivative f 0 (x), thus

d2 f f 0 (x + δx) − f 0 (x)
 
d df
2
≡ ≡ lim . (2.97)
dx dx dx δx→0 δx

The second-order derivative is also denoted using the ‘prime’ notation

d2 f
f 00 (x) ≡ . (2.98)
dx2
In operator notation the differential operator d/dx can be used to write
the second derivative as
 2
d2 f
   
d d d d df
(f ) ≡ (f ) ≡ ≡ 2, (2.99)
dx dx dx dx dx dx

where the left-hand-side means ‘operate on f (x) with d/dx twice’.

Example 2.19 Determine the second derivative of f (x) = x2 ex .

I Solution: We differentiate f (x) twice. The first derivative is

df d d 2 x
= f (x) = (x e ) = 2xex + x2 ex = (x2 + 2x)ex . (2.100)
dx dx dx
Likewise the second derivative is

d2 f
 
d df d 2
2
= = (x + 2x)ex = 2ex + 2xex + 2xex + x2 ex . (2.101)
dx dx dx dx

Hence, factorising this expression we have f 00 (x) = (x2 + 4x + 2)ex . J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

82 Section 2 : Differentiation

2.4.2 Newton’s notation

When the argument of a function is a time variable t, it is common to denote

derivatives using Newton’s notation.

Definition 2.5 Let x(t) be a function of t. In Newton’s notation the

derivatives of x(t) are expressed using ‘dots’ above the variable x as

dx d2 x
ẋ(t) ≡ , and ẍ(t) ≡ . (2.102)
dt dt2

[The ‘d’ notation dx/dt and d2 x/dt2 is called Leibniz’s notation.]

Example 2.20 The position x(t) of a particle at time t is given

x(t) = A cos(ωt + φ) (2.103)

where A, ω, and φ are constants. Show that the motion of the particle satisfies

ẍ(t) + ω 2 x(t) = 0. (2.104)

[This equation is called a differential equation because it involves derivatives.]

I Solution: Evaluating the first derivative of x(t) we have

d d
ẋ(t) = (x) = [A cos(ωt + φ)] = −ωA sin(ωt + φ). (2.105)
dt dt
Likewise, the second derivative is
 
d dx d
ẍ(t) = = [−ωA sin(ωt + φ)] = −ω 2 A cos(ωt + φ). (2.106)
dt dt dt

Since A cos(ωt + φ) = x(t), we may express this equation as

ẍ(t) = −ω 2 A cos(ωt + φ) = −ω 2 x(t). (2.107)

Thus, rearranging we obtain

ẍ(t) + ω 2 x(t) = 0. (2.108)

as required. This is the equation for simple harmonic motion. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 83

2.4.3 Derivatives of arbitrary order

Derivatives of arbitrary order may be obtained by repeated differentiation.

Definition 2.6 Let f (n−1) (x) denote the (n−1)th order derivative of f (x),
where n ∈ Z is an integer, then the nth derivative of f (x) is defined by

dn f f (n−1) (x + δx) − f (n−1) (x)

f (n) (x) ≡ ≡ lim . (2.109)
dxn δx→0 δx
This means that to obtain the nth order derivative of a function f (x)
we must differentiate n times. Indeed, in operator notation
n
dn−1 f dn f
  
d d
(f ) ≡ ≡ , (2.110)
dx dx dxn−1 dxn

where the left-hand-side means ‘operate on f (x) with d/dx n times’.

Example 2.21 Determine the fourth derivative of f (x) = x2 ex .

I Solution: We found in Example 2.19 that the second-derivative of f (x) is

d2 f
f (2) (x) ≡ = (x2 + 4x + 2)ex . (2.111)
dx2
Thus, the third derivative is

d3 f d (2) d 2
f (3) (x) ≡ 3
= f (x) = (x + 4x + 2)ex
dx dx dx
d d
= ex (x2 + 4x + 2) + (x2 + 4x + 2) ex
dx dx
= (2x + 4)e + (x + 4x + 2)e = (x + 6x + 6)ex .
x 2 x 2
(2.112)

The fourth derivative is therefore

(4) d4 f d (3) d 2
f (x) ≡ 4 = f (x) = (x + 6x + 6)ex
dx dx dx
d d
= ex (x2 + 6x + 6) + (x2 + 6x + 6) ex
dx dx
= (2x + 6)e + (x + 6x + 6)e = (x + 8x + 12)ex .
x 2 x 2
(2.113)

It may be shown that the nth derivative is f (n) (x) = (x2 + 2nx + n2 − n)ex .J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

84 Section 2 : Differentiation

2.4.4 Stationary points

Consider the function f (x) depicted in figure 2.6. The points A, B, and C on
the curve are those for which the gradient of f (x) is zero, that is, where the
tangents (dashed lines) to the curve are horizontal. We say that these points
are stationary (unchanging) because the rate-of-change is f 0 (x) = 0.

Let us examine the curve at the point A. Immediately to the left of A the
function is increasing with increasing x, i.e., the gradient is positive f 0 (x) > 0;
likewise, immediately to the right of A the function is decreasing with increasing
x, i.e., the gradient is negative f 0 (x) < 0. Thus, the point A is a local
maximum of f (x), with the gradient changing from positive to negative.

Now consider the point B. Although the gradient of the function at B is zero,
i.e., f 0 (x) = 0, the gradient either side of B is negative, i.e., f 0 (x) < 0. We
call such a point a stationary point of inflection.

Finally consider the point C. Immediately to the left of C the function is

decreasing with increasing x, i.e., the gradient is negative f 0 (x) < 0; likewise,
immediately to the right of C the function is increasing with increasing x, i.e.,
the gradient is poistive f 0 (x) > 0. Thus, the point C is a local minimum of
f (x), with the gradient changing from negative to positive.

Because stationary points of a function are defined by the change in gradient,

they can sometimes be classified by examining the second-derivatives.

Figure 2.6: Stationary points of f (x) at A (maximum), B (inflection), and C (minimum).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 85

Definition 2.7 A function f (x) is said to have a stationary point at

x = a if f 0 (a) = 0. We classify stationary points as follows:

• f (x) has a local maximum at x = a if f 0 (a) = 0, and the gradient

changes from positive to negative as x increases through x = a.

• f (x) has a stationary point of inflection at x = a if f 0 (a) = 0,

but the gradient does not change sign as x increases through x = a.

• f (x) has a local minimum at x = a if f 0 (a) = 0, and the gradient

changes from negative to positive as x increases through x = a.

Maxima and minima are referred to collectively as turning points or ex-

trema; it is sometimes possible to determine the turning points of a func-
tion f (x) by examining the second derivative:

• If f 0 (a) = 0 and f 00 (a) < 0, then x = a is a local maximum;

• If f 0 (a) = 0 and f 00 (a) > 0, then x = a is a local minimum.

The nature of extrema must be determine by other methods if f 00 (a) = 0.

Example 2.22 The height h(t) of a projectile as a function of time t is given

by the function
h(t) = 34 + 2t − 4t2 (2.114)

(see figure 2.7). Show that the projectile reaches a maximum height hM = 1.

0.5

0
0 0.25 0.5 0.75

3
Figure 2.7: Height of a projectile h(t) = 4 + 2t − 4t2 as a function of time t.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

86 Section 2 : Differentiation

I Solution: There are two stages to the solution of this problem: first (i) we
identify the stationary points of h(t); and second (ii) we classify the stationary
points as to whether they are maxima or minima (or other).
(i) The function h(t) is stationary whenever h0 (t) = 0, i.e., when

dh 1
= 2 − 8t = 0, that is, t= . (2.115)
dt 4
1
(ii) Evaluating the second derivative at t = 4
we have

d2 h
= −8 < 0; (2.116)
dt2 t=1/4

since this is negative, t = 14 corresponds to a maximum of h(t). It follows that

the maximum height reached by the projectile is

3 1 2
hM = h( 14 ) = + 2( 41 ) − 4


4
=1 (2.117)
4
as required. J
Example 2.23 The perimeter p of a rectangle is given by p = 2(a + b), where
a is the width, and b the length. Express the area A = ab of the rectangle
as a function of p and b. Hence, show that if the perimeter is fixed, i.e., if
p = constant, then A maximised when b = a.

I Solution: Suppose that we vary b, then for fixed p we have a = ( 12 p − b),

and the area is given by

A(b) = ab = ( 12 p − b)b = 12 pb − b2 . (2.118)

Now A takes a stationary value as we vary b whenever dA/db = 0, that is, if

dA
A0 (b) = = 12 p − 2b = ( 21 p − b) − b = a − b = 0. (2.119)
db
Thus, A takes a stationary value when a = b. Evaluating the second derivative
of A(b) at b = a we have

d2 A
= −2 < 0. (2.120)
db2 b=a

since this is negative, b = a corresponds to a maximum of A(b). Observe that

this configuration is that of a regular rectangle, or square. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 87

2.5 Power Series

Recall that the general form of a polynomial of degree n ∈ N is

n
X
2 n−1 n
Pn (x) = a0 + a1 x + a2 x + · · · + an−1 x + an x = ar x r , (2.121)
r=0

where the ar are constant coefficients (see §1.4.3), and the summation includes
a finite number of terms. Polynomials are attractive because they can be
differentiated easily, and evaluated using multiplication only.
Now consider the infinite series
∞
X
P (x) = a0 + a1 x + a2 x2 + a3 x3 + · · · = ar x r . (2.122)
r=0

This expression looks like a polynomial with an infinite number of terms, and is
known as a power series (a series of powers of x). Power series are important
because they can be used to represent functions. Indeed, we saw in §1.10 that
the exponential function is in fact defined by a power series, i.e.,
∞
x2 x3 X xr
ex ≡ exp(x) ≡ 1 + x + + + ··· = . (2.123)
2! 3! r=0
r!

Power series representations of functions are especially useful in methods of

approximation due to the rapidly diminishing magnitude of terms when x is
small. For example, we can approximate ex by a quadratic if we exclude terms
of order x3 or higher, viz.

x x2
e ≈1+x+ , for |x|  1. (2.124)
2!
Here the symbol ‘≈’ means ‘approximately equal to’, while ‘for |x|  1’ means
‘the approximation works provided |x| is much less than 1’ (so x3 is small
compared to 1). For example, putting x = 0.1 into this approximation gives

0.1 (0.1)2
e ≈ 1 + 0.1 + = 1.105, (2.125)
2!
which is accurate to within 1 part in 1000 (e0.1 = 1.105170918 . . . ).
In this section we explore how to find such power series representations of
functions using what are termed Maclaurin series and Taylor series.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

88 Section 2 : Differentiation

2.5.1 Maclaurin series

To represent a function f (x) in the form of a power series we need to find

some set of coefficients ar such that
∞
X
2 3 4 5
f (x) = a0 + a1 x + a2 x + a3 x + a4 x + a5 x + · · · = ar x r . (2.126)
r=0

If we assume that f (x) can be differentiated as often as we please, then these

coefficients ar may be determined as follows. First notice that setting x = 0
in equation (2.126) gives
a0 = f (0). (2.127)

Now observe that differentiating both sides of equation (2.126) gives

df
≡ f (1) (x) = a1 + 2a2 x + 3a3 x2 + 4a4 x3 + 5a5 x4 + . . . ; (2.128)
dx
hence, setting x = 0 in this equation we have

f (1) (0)
a1 = . (2.129)
1
Differentiating a second time then gives

d2 f
= f (2) (x) = 2a2 + (3 × 2)a3 x + (4 × 3)a4 x2 + (5 × 4)a5 x3 + . . . ; (2.130)
dx2
hence, setting x = 0 yields
f (2) (0)
a2 = . (2.131)
2×1
Similarly, differentiating a third time we find

d3 f
= f (3) (x) = (3 × 2)a3 + (4 × 3 × 2)a4 x + (5 × 4 × 3)a5 x2 + . . . ; (2.132)
dx3
while setting x = 0 yields
f (3) (0)
a3 = . (2.133)
3×2×1
Continuing in this pattern, it may be shown that the nth coefficient is given by

f (n) (0)
an = , (2.134)
n!
where n! = n × (n − 1) × · · · × 2 × 1 is n factorial.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 89

Thence, substituting these coefficients into equation (2.126) we have

∞
(1) x2 x3 X xr
f (x) = f (0)+xf (0)+ f (2) (0)+ f (3) (0)+· · · = f (r) (0). (2.135)
2! 3! r=0
r!

This power series is called the Maclaurin series for f (x).

Definition 2.8 Let f (x) be a function that may be differentiated as many

times as we please at x, then the Maclaurin series for f (x) is given by

x2 (2) x3
f (x) = f (0) (0) + xf (1) (0) + f (0) + f (3) (0) + . . . , (2.136)
2! 3!

where the notation f (0) (x) means ‘differentiate f (x) zero times’ such that

f (0) (x) ≡ f (x) and f (0) (0) ≡ f (0). (2.137)

In this way the Maclaurin series may be written in summation notation as

∞
X xr
f (x) = f (r) (0). (2.138)
r=0
r!

Note: A Maclaurin series is only a valid when it converges (see §2.5.3).

Example 2.24 Determine the Maclaurin series for f (x) = ex .

I Solution: We look for a Maclaurin series expansion of the form

∞
x (0) (1) x2 x2 X xr
e =f (0)+xf (0)+ f (2) (0)+ f (3) (0)+· · · = f (r) (0). (2.139)
2! 3! r=0
r!

Since f 0 (x) = f (x) = ex in general it follows that

dn x
f (n) (x) = (e ) = ex , and thus f (n) (0) = e0 = 1. (2.140)
dxn
Hence, putting this result into equation (2.139) we obtain the Maclaurin series
∞
x x2 x3 X xr
e =1+x+ + + ··· = . (2.141)
2! 3! r=0
r!

Note: This example is circular because we used the series to define ex . J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

90 Section 2 : Differentiation

Example 2.25 Determine the first four non-zero terms of the Maclaurin series
for f (x) = sin x. Hence, by using a quinitic approximation, estimate sin(0.1)
to within 1 part in 1,000,000,000.

I Solution: We look for a Maclaurin series expansion of the form

x2 (2) x3
sin x = f (0) (0) + xf (1) (0) + f (0) + f (3) (0) + . . . (2.142)
2! 3!
Evaluating the first few derivative of f (x) = sin x we have

f (0) (x) = + sin x ⇒ f (0) (0) = 0 (2.143a)

f (1) (x) = + cos x ⇒ f (1) (0) = +1 (2.143b)
f (2) (x) = − sin x ⇒ f (2) (0) = 0 (2.143c)
f (3) (x) = − cos x ⇒ f (3) (0) = −1 (2.143d)
f (4) (x) = + sin x ⇒ f (4) (0) = 0 (2.143e)
f (5) (x) = + cos x ⇒ f (5) (0) = +1 (2.143f)
f (6) (x) = − sin x ⇒ f (6) (0) = 0 (2.143g)
f (7) (x) = − cos x ⇒ f (7) (0) = −1 (2.143h)

Hence, putting these values into equation (2.142) we obtain

x3 x5 x7
sin x = x − + − + ... (2.144)
3! 5! 7!
Retaining the terms up to quintic order only (i.e., ‘truncate after x5 ’) we obtain
the approximation

(0.1)3 (0.1)5
sin (0.1) ≈ 0.1 − + = 0.099833416̇, (2.145)
3! 5!
which is accurate to 1 part in 1,000,000,000 (sin 0.1 = 0.0998334166468 . . . ).

Note: It may be shown that the following series expansions are valid for all x
∞
x3 x5 x7 X (−1)n
sin(x) = x − + − + ··· = x(2n+1) , (2.146a)
3! 5! 7! n=0
(2n + 1)!
∞
x2 x4 x6 X (−1)n
cos(x) = 1 − + − + ··· = x(2n) , (2.146b)
2! 4! 6! n=0
2n!

The first few terms of these expansions are depicted in figure 2.8.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 91

2 2

1 1

cos(x)
sin(x)

0 0

-1 -1
ñ = 1 ñ = 1
-2 -2
-2 -1 0 1 2 -2 -1 0 1 2
x/π x/π
2 2

1 1

cos(x)
sin(x)

0 0

-1 -1
ñ = 2 ñ = 2
-2 -2
-2 -1 0 1 2 -2 -1 0 1 2
x/π x/π
2 2

1 1
cos(x)
sin(x)

0 0

-1 -1
ñ = 3 ñ = 3
-2 -2
-2 -1 0 1 2 -2 -1 0 1 2
x/π x/π
2 2

1 1
cos(x)
sin(x)

0 0

-1 -1
ñ = 4 ñ = 4
-2 -2
-2 -1 0 1 2 -2 -1 0 1 2
x/π x/π
2 2

1 1
cos(x)
sin(x)

0 0

-1 -1
ñ = 5 ñ = 5
-2 -2
-2 -1 0 1 2 -2 -1 0 1 2
x/π x/π

Figure 2.8: Maclaurin series (solid curves) for sine (left column, dash-dotted curves) and
cosine (right column, dash-dotted curves) according to the expansions in equations (2.146),
with ñ as the number of terms retained in the series. Notice the symmetry of the expansions.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

92 Section 2 : Differentiation

2.5.2 Binomial series

The Maclaurin series for (1 + x)r , is known as the Binomial series.

Definition 2.9 The binomial series is a power series defined by

r(r − 1) 2 r(r − 1)(r − 2) 3

(1 + x)r = 1 + rx + x + x + ..., (2.147)
2! 3!
where x, r ∈ R. The series converges for |x| < 1, and terminates if r ∈ N.

The binomial series provides a natural framework for approximating roots.

√
Example 2.26 Compute 3 9 to one part in 500 without using a calculator.

I Solution: We begin by noticing that 9 = (8 + 1) = 8(1 + 18 ) such that

√ 1 1/3 1 1/3
9 = 81/3 1 +
3



8
=2 1+ 8
. (2.148)

Now for r = 1/3 the binomial series yields

1 1
1/3 1 ( − 1) 2 13 ( 13 − 1)( 13 − 2) 3
3 3
(1 + x) =1+ x+ x + x + ...; (2.149)
3 2! 3!
thus, setting x = 1/8 in this expression we have
1 1 1 1
1 1/3

1 1

(
3 3
− 1) 1 2

(
3 3
− 1)( 13 − 2) 1 3


1+ 8
=1+ · 8
+ 8
+ 8
+ ...,
3 2! 3!
(2.150)
such that evaluating the fractions we find
 1/3
1 1 1
1+ =1+ − + smaller terms
8 24 576
 
25 1
= +O , (2.151)
24 500

where O(1/500) means ‘of order (size) 1/500 compared to unity’. Hence,
substituting this expression for (1 + 18 )1/3 into equation (2.148) we have
1/3
√

3 1 25 25
9=2 1+ ≈2× = (2.152)
8 24 12
√
3
as an approximation for 9 to an accuracy of 1 part in 500. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 93

2.5.3 Convergence

Generally speaking the Maclaurin series for a function will have an infinite
number of terms. For example, the Maclaurin series for f (x) = ex is
∞
x x 2 x3 X xr
e =1+x+ + + ··· = . (2.153)
2! 3! r=0
r!

This means that for some given value of x, the series involves adding together
an infinite series of numbers; for example, when x = 2 the series for ex is
∞
2 22 23 X 2r
e =1+2+ + + ··· = . (2.154)
2! 3! r=0
r!

If such an ‘infinite summation’ tends towards a unique value (limit), then we

say that the series converges, otherwise we say that the series diverges.
It turns out that not all power series converge for all values of x, so we
specify the range of values for which a series does converge by a convergence
condition (e.g., an inequality). For instance, the binomial series is

r(r − 1) 2
(1 + x)r = 1 + rx + x + ..., |x| < 1, (2.155)
2!
where the condition ‘|x| < 1’ means ‘(1 + x)r is identical to the power series
expansion provided |x| < 1’. [If the series converges for all values of x, then this
is sometimes written as x ∈ R.] Crucially, a function and its series are only
identical for those values of x that satisfy the convergence condition.
Determining convergence is beyond the scope of this course; however, the
convergence conditions for the series described so far are listed below.

Result 2.6 Useful power series expansions include:

x x2 x3
e =1+x+ + + ... x ∈ R, (2.156a)
2! 3!
x3 x5 x 7
sin x = x − + − + ... x ∈ R, (2.156b)
3! 5! 7!
x2 x4 x6
cos x = 1 − + − + ... x ∈ R, (2.156c)
2! 4! 6!
r(r − 1) 2
(1 + x)r = 1 + rx + x + ... |x| < 1. (2.156d)
2!

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

94 Section 2 : Differentiation

Example 2.27 Obtain a cubic polynomial P3 (x) approximation for

1
f (x) = . (2.157)
2−x

Sketch f (x) and P3 (x), and comment on the suitability of the approximation.

I Solution: Let α = −x/2, then we may write f (x) in the form

1 1 x −1 1
f (x) = = 1− = (1 + α)−1 . (2.158)
2−x 2 2 2
Now, by the binomial series with r = −1 we have

(1 + α)−1 = 1 − α + α2 − α3 + α4 + . . . for |α| < 1. (2.159)

such that upon substituting for α = −x/2 we obtain

 x −1 x x2 x3 x4
1− =1+ + + + + ... for |x| < 2, (2.160)
2 2 4 8 16
where we have expressed the convergence condition as 2|α| = |x| < 2. Hence,
putting this series into equation (2.158) we have

1 x −1 1 x x2 x3 x4
f (x) = 1− = + + + + +... for |x| < 2. (2.161)
2 2 2 4 8 16 32
To obtain a cubic approximation P3 (x) we neglect terms of order x4 or higher:

1 x x2 x3
f (x) ≈ P3 (x) = + + + . (2.162)
2 4 8 16
We expect the approximation to hold provided |x|  2. (see figure 2.9). J

2
P3 (x)
1.5 f (x)
f (x)

0.5

0
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
x

Figure 2.9: Cubic approximation P3 (x) (dashed curve) to f (x) = 1/(2 − x) (solid curve).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 95

2.5.4 Taylor series

The Maclaurin series expansion in powers of x can be generalised to a series

expansion in powers of (x − x0 ), where x0 is a constant, according to what is
known as a Taylor series. In particular, it may be shown that if f (x) is well
behaved2 in the region of x = x0 , then the Taylor series of f (x) about x0 is

(x − x0 )2 (2)
f (x) = f (x0 ) + (x − x0 )f (1) (x0 ) + f (x0 ) + . . . ; (2.163)
2!
we can therefore obtain expansions for functions that are not well defined at
x = 0, because ‘good behaviour’ is required near x0 only.

An alternative expression for the Taylor series is

∆x2 (2)
f (x0 + ∆x) = f (x0 ) + ∆xf (1) (x0 ) + f (x0 ) + . . . , (2.164)
2!
where ∆x is the difference between x and x0 , i.e.,

∆x = (x − x0 ). (2.165)

This alternative expression makes it clear that a Maclaurin series is simply the
special case of a Taylor series with x0 = 0, and thus ∆x = x.

Definition 2.10 Let f (x) be a function that may be differentiated as many

times as we please at x = x0 , then the Taylor series for f (x) about x0 is

∆x2 (2)
f (x) = f (x0 + ∆x) = f (x0 ) + ∆xf (1) (x0 ) + f (x0 ) + . . . , (2.166)
2!
where ∆x = (x − x0 ). In summation notation this series is written as
∞
X (∆x)r
f (x) = f (x0 + ∆x) = f (r) (x0 ). (2.167)
r=0
r!

Equation (2.166) is called the Taylor series for f (x) even if the series does
not converge on f (x). A Maclaurin series is a Taylor series with x0 = 0.

2
Broadly speaking, we say that f (x) is ‘well behaved’ if we can differentiate f (x) as often as
required at x = x0 .

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

96 Section 2 : Differentiation

Example 2.28 Determine the Taylor series of f (x) = loge x about x = 1.

I Solution: By Definition 2.10, the Taylor expansion of f (x) = loge x about

the point x0 = 1 is

∆x2 (2) ∆x3 (3)

loge (1 + ∆x) = f (1) + ∆xf (1) (1) + f (1) + f (1) + . . . , (2.168)
2! 3!
and so forth. Evaluating the first few derivatives of f (x) we have

f (0) (x) = loge x ⇒ f (0) (1) = 0 (2.169a)

1
f (1) (x) = ⇒ f (1) (1) = +1 (2.169b)
x
(2) 1
f (x) = − 2 ⇒ f (2) (1) = −1 (2.169c)
x
(3) 2
f (x) = 3 ⇒ f (3) (1) = +2! (2.169d)
x
(4) 3×2
f (x) = − 4 ⇒ f (4) (1) = −3! (2.169e)
x
Thence, putting these values into equation (2.168) we obtain

∆x2 ∆x3 ∆x4

loge (1 + ∆x) = ∆x − + − + ...; (2.170)
2 3 4
in fact, in general we may write
∞
X (−1)r
loge (1 + ∆x) = ∆x(r+1) . (2.171)
r=0
r+1

Note: It is more common to write this series in the form

∞
x2 x3 x 4 X (−1)r
loge (1 + x) = x − + − + ··· = x(r+1) . (2.172)
2 3 4 r=0
(r + 1)

It may be shown that the convergence condition is −1 < |x| ≤ −1. This series
may also be obtained from the Maclaurin series for f (x) = loge (1 + x). J

2.5.5 Linear approximation by Taylor series (optional)

Like Maclaurin series, Taylor series are particularly useful for problems involving
approximation (see figure 2.10). To illustrate this idea, suppose that we know
the value of a function at x = x0 , and wish to estimate its value at some nearby

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 97

Figure 2.10: By geometry we expect f (x + ∆x) ≈ f (x0 ) + ∆x tan φ, where φ is the

angle that the tangent makes with the x-axis. However, since the gradient of f (x) at x = x0
is f 0 (x0 ) = tan φ, we may write this approximation as f (x0 + ∆x) ≈ f (x0 ) + ∆xf 0 (x0 ).

point x = x0 + ∆x, then according to the geometric argument described in

figure 2.10 we expect

f (x0 + ∆x) ≈ f (x0 ) + ∆xf 0 (x0 ). (2.173)

This approximation is identical to expression found by truncating the Taylor

expansion of f (x) at first-order, i.e., by neglecting terms in ∆x2 or higher.
Truncation of the series at first-order is sometimes referred to as linearisation,
or a linear approximation, because the first-order series is the equation of
a straight line. Provided that ∆x is small compared to unity, the truncation
error will usually diminish rapidly with the order of the approximation.

2.6 L’Hôpital’s Rule

Taylor series may be used to derive L’Hôpital’s rule for the limits of quotients
in indeterminate form (cf. §2.1.3). Very informally, the rule is as follows:

Result 2.7 Let f (x) and g(x) have limits at x = a that satisfy either:

lim f (x) = 0 and lim g(x) = 0; or lim |f (x)| = ∞ and lim |g(x)| = ∞.
x→a x→a x→a x→a

If the derivatives f 0 (x) and g 0 (x) exist, then L’Hôpital’s rule states that

lim [f (x)/g(x)] = lim [f 0 (x)/g 0 (x)]. (2.174)

x→a x→a

Note: This rule may be used repeatedly if lim f 0 (x) = lim g 0 (x) = 0 etc.
x→a x→a

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

98 Section 2 : Differentiation

Example 2.29 Evaluate the limit L = lim [sin x/x].

x→0

I Solution: This limit is of the form

sin x
L = lim , with lim sin x = 0 and lim x = 0. (2.175)
x→0 x x→0 x→0

Since the limit of the top and bottom of the quotient are both zero, we may
use the formulation of l’Hôpital’s rule in Result 2.7 to obtain
d
sin x dx
(sin x) cos x
L = lim = lim d
= lim = 1, (2.176)
x→0 x x→0 (x) x→0 1
dx

where the final step is obtained by the quotient rule for limits (see §2.1.3). J
Example 2.30 Evaluate the limit L = lim xe−x .
x→∞

I Solution: This limit is of the form

x
L = lim , with lim x = ∞ and lim ex = ∞. (2.177)
x→∞ ex x→∞ x→∞

Since the limit of the top and bottom of the quotient are both infinite, we may
use the formulation of l’Hôpital’s rule in Result 2.7 to obtain
d
x (x) 1
L = lim = lim dx
d
= lim = lim e−x = 0. (2.178)
x→∞ ex x→∞ (ex ) x→∞ ex x→∞
dx

Note: It may be shown by repeated application of l’Hôpital’s rule that

L = lim xn e−x = 0, (2.179)

x→∞

where n ∈ N is a positive integer. J

Example 2.31 Use l’Hôpital’s rule to show that L = lim x2 e−x = 0.
x→∞

I Solution: The limit L is in indeterminate form, so we use l’Hôpital’s rule to

give
d
x2 (x2 ) 2x
L = lim x = lim dx d
= lim . (2.180)
x→∞ e x→∞
dx
(ex ) x→∞ ex
Since this form is still indeterminate, we apply l’Hôpital’s rule a second time
to obtain
d
2x (2x) 2
L = lim x = lim dx d
= lim x = 0. (2.181)
x→∞ e x→∞ (e ) x→∞ e
x
dx

as required. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 2 : Differentiation 99

L’Hopital’s rule by Taylor series (optional)

The origin of l’Hopital’s rule may be understood as follows. Suppose that we

wish to evaluate the limit

L = lim [f (x)/g(x)], with lim f (x) = 0 and lim g(x) = 0. (2.182)

x→a x→a x→a

Clearly we cannot use the quotient rule to write L = f (a)/g(a) because the
quantity 0/0 is not defined (see §2.1.3).

Nevertheless, by Taylor expanding f (x) and g(x) about x = a we may write

∆x2 (2) 3
f (x) f (a) + ∆xf (1) (a) + 2!
f (a) + ∆x 3!
f (3) (a) + . . .
= ∆x2 (2) ∆x3 (3)
, (2.183)
g(x) g(a) + ∆xg (1) (a) + 2!
g (a) + 3!
g (a) + . . .

where
∆x = (x − a). (2.184)

Because f (a) = g(a) = 0, the ratio in equation (2.183) reduces to

2 3
f (x) ∆xf (1) (a) + ∆x
2!
f (2) (a) + ∆x
3!
f (3) (a) + . . .
= 2 , (2.185)
g(x) ∆xg (1) (a) + ∆x g (2) (a) + ∆x3 g (3) (a) + . . .
2! 3!

whereupon dividing the top and bottom by ∆x we obtain

2
f (x) f (1) (a) + ∆x
2!
f (2) (a) + ∆x
3!
f (3) (a) + . . .
= . (2.186)
g(x) g (1) (a) + ∆x g (2) (a) + ∆x2 g (3) (a) + . . .
2! 3!

Now, since ∆x = (x − a), we have that

if x → a, then ∆x → 0, (2.187)

meaning that the limits x → a and ∆x → 0 are equivalent. Taking this limit
we obtain
" #
f (x) f (1) (a) + ∆x
2!
f (2)
(a) + . . .
lim = lim
x→a g(x) ∆x→0 g (1) (a) + ∆x g (2) (a) + . . .
2!

f (1) (a)
= , (2.188)
g (1) (a)

where we noticed that all the terms in powers of ∆x vanish as ∆x → 0. This

is the expression l’Hôpital’s Rule given in Result 2.7.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

100 Section 2 : Differentiation

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 Engineering Mathematics
J. J. Bissell

Integration

In the previous section we considered differential calculus as the branch of cal-

culus concerned with the study of rates-of-change. We now conclude our survey
of preliminary calculus by considering its other branch: integral calculus.
Integral calculus, or more simply integration, is the study of the summation
R
(‘ um’) of parts to make wholes, and is named after the Latin word integralis
meaning ‘complete’. Since there exist many situations in engineering and the
physical sciences in which one must evaluate the cumulative effects of quanti-
ties, integration is of great practical importance, and provides us with a precise
mathematical language in which to express physical concepts.
The purpose of this section, therefore, is twofold: first, we explore the founda-
tions of integration as a method of computing summations over infinitesimally
small quantities; and second, we review important methods of integration from
school mathematics. By the end of the section you should be able to. . .
Learning outcomes:
◦ Describe the meaning of integration as a method for computing summations.
◦ Distinguish between a definite integral and an indefinite integral.
◦ Write down an expression for the fundamental theorem of calculus.
◦ Evaluate the integrals of standard functions by inspection.
◦ Identify basic substitutions to simplify integrals.
◦ Convert integrals into more recognisable forms by completing the square.
◦ Integrate rational functions using the method of partial fractions.
◦ Apply the tangent half-angle substitution.
◦ Use integration by parts to evaluate simple integrals.
◦ Evaluate infinite integrals formally using limits.
◦ Express the mean value of a function using integration.
These learning outcomes must be reinforced by completing course exercises.

101
102 Section 3 : Integration

3.1 Motivating Integration

Broadly speaking integration is the process of summing together lots

of infinitesimally small quantities to give a finite result. Here we in-
troduce this idea by describing how the need for integration arises in some
common applications. We then motivate the connection between integration
and differentiation enshrined by the fundamental theorem of calculus.

3.1.1 Integration to find the area under a curve

Consider a function f (x) defined on an interval x ∈ [xa , xb ], such as that

depicted in figure 3.1(a). Now suppose that we wish to determine the area A
under the curve y = f (x) between x = xa and x = xb . One way to do this
is to approximate the area as a rectangle of width ∆x = (xb − xa ) and height
f (xb ), as depicted in figure 3.1(b); we then have

A ≈ f (x1 )∆x, where ∆x = xb − xa , and x1 = xb = xa + ∆x. (3.1)

For most curves this approximation is likely to be rather poor; however, we

can make the approximation better by using more rectangles. For example, in
figure 3.1(c) we have approximated the area using two rectangles each of width
∆x = (xb − xa )/2. In this way we have a slightly better approximation
2
X
A ≈ f (x1 )∆x + f (x2 )∆x = f (xk )∆x, (3.2)
k=1
 
xb − xa
where ∆x = , and xk = xa + k∆x.
2

Likewise, in figure 3.1(d) we use four rectangles to obtain an even better

approximation
4
X
A ≈ f (x1 )∆x + f (x2 )∆x + f (x3 )∆x + f (x4 )∆x = f (xk )∆x, (3.3)
k=1
 
xb − xa
where ∆x = , and xk = xa + k∆x.
4

Indeed, in general, if we divide the interval [xa , xb ] into n sub-intervals, each

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 103

(a) (b)

(c) (d)

(e) (f)

Figure 3.1: (a) Area A beneath a curve y = f (x) between xa and xb (light grey). This
area may be approximated by the combined area of N rectangles (dark grey). Here we show
such approximations with: (b) n = 1; (c) n = 2; (d) n = 4; (e) n = 8; and (f) n = 16.
The approximation becomes exact in the limit that n → ∞ and ∆x = ((xb − xa )/n) → 0.

of width ∆x = (xb − xa )/n, then we may approximate the area as

n  
X xb − xa
A≈ f (xk )∆x, where ∆x = , and xk = xa + k∆x.
k=1
n
(3.4)
The sum in equation (3.4) is known as a Riemann sum, and the approximation
given by the Riemann sum gets better as n gets larger (see figure 3.1). Indeed,
in the limit that n → ∞, the approximation becomes exact, and we write
n
X
A = lim f (xk )∆x. (3.5)
n→∞
k=1

Notice that as n → ∞, the width ∆x = (xb − xa )/n → 0 of each rectangle

vanishes. Informally, then, equation (3.5) suggests that the sum of an infinite
number of vanishingly small quantities can—in some cases—converge on a
finite limit. We use this idea to define what is known as a definite integral.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

104 Section 3 : Integration

Definition 3.1 (informal) Let f (x) be a function defined on an interval

x ∈ [xa , xb ], and divide [xa , xb ] into n subintervals each having width
 
xb − xa
∆x = . (3.6)
n

Then we may define a Riemann sum of f (x) on [xa , xb ] as

n
X
S= f (xk )∆x, with xk = xa + k∆x. (3.7)
k=1

If S converges on a unique limit as n → ∞ and ∆x → 0, then we define

the definite integral I of f (x) on [xa , xb ], as
Z xb n
X
I= f (x)dx ≡ lim S = lim f (xk )∆x. (3.8)
xa ∆x→0 n→∞
k=1

We call the function f (x) the integrand. The interval bounds xa and xb
are known as the lower and upper limits of the integration respectively.
Note: Very informally, the limiting process of turning a Riemann sum into
P R
an integral means making the replacements → , and ∆x → dx.

Returning to equation (3.5), therefore, it follows from Definition 3.1 that the
area A under a curve y = f (x) between xa and xb is given by
n
X Z xb
A = lim f (xk )∆x = f (x)dx. (3.9)
n→∞ xa
k=1

Hence, integration can be used to determine the area under a curve;

now let us consider some other applications.

3.1.2 Integration to find the mass of a solid

Figure 3.2 depicts a rod of length l and cross-sectional area A, lying parallel
to the x-axis. One end of the rod is at x = 0, while the other is at x = l; the
density ρ(x) of the rod varies as a function of position x. Notice that we can
divide the rod into n segments, each of width ∆x = l/n, and volume

∆V = A∆x. (3.10)

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 105

Figure 3.2: Rectangular bar of length l, cross-sectional area A, and density ρ(x).

Now, the density ρ(x) varies over the length of each segment ∆x; however,
if n is made sufficiently large, such that each segment ∆x = l/n becomes
very thin, then the density does not vary much. In this way the mass of each
segment may be approximated by

∆mk ≈ ρ(xk )∆V = ρ(xk )A∆x, where xk = k∆x. (3.11)

Adding up all these little masses, the total mass m of the rod is then
n
X n
X
m= ∆mk ≈ ρ(xk )A∆x. (3.12)
k=1 k=1

Hence, defining a function f (x) = ρ(x)A, we may approximate the mass m by

a Riemann sum, viz.
n
X
m≈ f (xk )∆x, where f (x) = ρ(x)A. (3.13)
k=1

Thus, in the limit that n → ∞, for which the width ∆x = (l/n) → 0 of

each segment vanishes, this approximation becomes exact, and we have by
Definition 3.1 that the total mass of the rod is given by the integral
Z l Z l
m= f (x)dx = ρ(x)Adx. (3.14)
0 0

Notice that we have (broadly speaking) got from the sum in equation (3.13) to
P R
the integral in equation (3.14) by setting → and ∆x → dx. This gives
us an informal way for thinking about integration (see §3.1.3).

3.1.3 Setting up integrals informally

An informal approach to setting up integrals is to think of integration as being

R
a sum ( um) over infinitesimal elements. To illustrate this idea, suppose that

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

106 Section 3 : Integration

the rod in figure 3.2 is divided into lots of infinitesimal elements, each of width
dx, and infinitesimal volume element

dV = Adx. (3.15)

Then the volume element dV corresponding to the interval [x, x + dx] has
infinitesimal mass dm given by

dm = ρ(x)dV = ρ(x)Adx. (3.16)

The total mass of the rod is then given by integrating up all these elements,
that is, Z Z Z l
m= dm = ρ(x)dV = ρ(x)Adx, (3.17)
rod rod 0

where the integration is over the whole rod, i.e., from x = 0 to x = l.

3.1.4 Integration to find the distance travelled by an object

Suppose that s(t) is the distance travelled at time t by an object moving in a

straight line with time-dependent (i.e., non-constant) velocity v(t). Then the
total distance sT travelled by the object in the time interval t ∈ [ta , tb ] is

sT = s(tb ) − s(ta ). (3.18)

We now consider an informal argument for computing the distance sT by inte-

grating up the velocity v(t).
Although the velocity v(t) is time-dependent, over a sufficiently short period
of time it does not change very much. Indeed, over an infinitesimal period of
time dt, the velocity may be considered constant. Thus, during a time interval
[t, t+dt] of infinitesimal duration dt, the object travels an infinitesimal distance

ds = v(t)dt (3.19)

(where we used the fact that distance = speed × time). The total distance sT
R
travelled by the object is therefore found by integrating up (‘ umming up’) all
these infinitesimal distances, i.e.,
Z Z tb
sT = ds = v(t)dt, (3.20)
interval ta

where the integration is over the total time interval [ta , tb ].

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 107

3.2 Evaluating Integrals

The processes of expressing a problem as an integration is known as ‘setting-up’

an integral. So far we have ‘set-up’ integrals to express: (i) the area under a
curve (§3.1.1); (ii) the mass of a bar (§3.1.2); and (iii) the distance travelled by
an object (§3.1.4). Our next task is to explore how such integrals are evaluated.

There are at least two methods for evaluating integrals. The first method is to
begin with Definition 3.1 by treating the integration as the limit of a Riemann
sum; this procedure is known as integration by first principles (§3.2.1). The
second method is to think of integration as a kind of anti-differentiation; this
approach relies on the fundamental theorem of calculus (§3.2.2).

3.2.1 Integration by first principles (optional)

The process of integrating a function by first setting up a Riemann sum, and

then considering a limit, is known as integration by first principles. We
illustrate this idea in the following example.

Example 3.1 In this problem we explore how integration by first principles can
be used to evaluate the definite integral of

f (x) = x2 (3.21)

on the interval x ∈ [xa , xb ], where xa = 0 and xb = b, with b constant. The

interval [0, b] and curve y = f (x) is depicted in figure 3.3.

(a) Suppose that the interval [0, b] is divided into n subintervals of equal
width; what is the width ∆x of each of these subintervals?
(b) Given that xk = xk−1 + ∆x, with x0 = 0, demonstrate that xk = k∆x.
(c) Use the fact that
n
X n
k 2 = (2n + 1)(n + 1) (3.22)
k=1
6

to write the Riemann sum SR of f (x) on [0, b] as

n
b3
  
X 1 1
SR = f (xk )∆xk = 2+ 1+ . (3.23)
k=1
6 n n
Rb
(d) Hence determine the value of the definite integral I = 0
f (x)dx.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

108 Section 3 : Integration

Figure 3.3: Interval x ∈ [0, b] divided into n subintervals of width ∆x = (b/n).

I Solution: We consider each part of the problem in turn.

(a) If we divide [0, b] into n equal subintervals, then the width of each is
∆x = (b − 0)/n = (b/n), as depicted in figure 3.3.
(b) The kth subinterval is [xk−1 , xk ], with xk = xk−1 +∆x, and x0 = 0. Hence

x0 = 0 x1 = x0 + ∆x = ∆x, x2 = x1 + ∆x = 2∆x, (3.24)

and so forth. Continuing with this pattern we have xk = k∆x.

(c) By Definition 3.1 the Riemann sum of f (x) = x2 on [0, b] is
n n n n
X X X b3 X 2
S= f (xk )∆x = x2k ∆x = 2 3
k (∆x) = 3 k , (3.25)
k=1 k=1 k=1
n k=1

where we used xk = k∆x, and ∆x = b/n. Thus, by equation (3.22) we obtain

n
b3 X 2 b3 n
S= 3 k = 3 × (2n + 1)(n + 1)
n k=1 n 6
b3
  
1 1
= 2+ 1+ . (3.26)
6 n n

(d) Now, the limit that ∆x = (b/n) → 0 is equivalent to the limit that n → ∞.
By Definition 3.1, therefore, we have
Z b
I= f (x)dx ≡ lim S = lim S
0 ∆x→0 n→∞

b3 b3
  
1 1
= lim 2+ 1+ = . (3.27)
n→∞ 6 n n 3

The definite integral of f (x) on [0, b] is thus I = b3 /3. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 109

3.2.2 Fundamental theorem of calculus

In §3.1.4 we considered the motion of an object travelling in a straight line

with velocity v(t). In particular, we found in equation (3.20) that if s(t) is the
position of the object at time t, then the total distance sT travelled by the
object between t = ta and t = tb is given by the definite integral
Z tb
sT = s(tb ) − s(ta ) = v(t)dt, (3.28)
ta

where ta and tb are the upper and lower limits of the integral respectively.
Now recall from our work on differentiation that the velocity of the object at
time t is given by the derivative of the displacement s(t), that is,

ds
v(t) = = s0 (t). (3.29)
dt
It follows that the definite integral in equation (3.28) can be written as
Z tb
ds(t)
s(tb ) − s(ta ) = s0 (t)dt where s0 (t) = . (3.30)
ta dt

By direct analogy, therefore, if F (u) is a function with derivative f (u) = F 0 (u),

then the definite integral of f (u) on some interval u ∈ [a, b] is
Z b
dF (u)
F (b) − F (a) = f (u)du, where f (u) = . (3.31)
a du

In this equation u is known as a dummy variable.

Result 3.1 If f (x) is the derivative of F (x), then the definite integral
of f (x) on the interval [a, b] is
Z b
dF (x)
f (x)dx = F (b) − F (a), where f (x) = . (3.32)
a dx

It is common to write the right-hand-side of this equation using the notation

h ib
F (x) ≡ F (b) − F (a), (3.33)
a

where a and b are the limits of the integration. The integral is said to be
definite because it has definite limits.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

110 Section 3 : Integration

Now consider equation (3.31), and suppose that we choose an arbitrary value
for the lower limit, say a = α, but allow that upper limit to vary by setting
b = x, where x is a variable. In this way, equation (3.31) may be written
Z x
dF (u)
F (x) + c = f (u)du, f (u) = , with c = −F (α) (3.34)
α du

as an arbitrary constant that is fixed by whatever arbitrary value we choose for

α. In this equation the value of the integral is allowed to change by varying x,
meaning that instead of taking a definite value, the integral returns a function
of x; for this reason we call equation (3.34) the indefinite integral of f (x).
Informally, it is customary to reflect the fact that x varies (with α arbitrary) by
omitting the limits, and writing the integral in terms of x rather than u, viz.
Z
dF (x)
F (x) + c = f (x)dx, where f (x) = . (3.35)
dx

This equation is an expression of the fundamental theorem of calculus, and

states that if f (x) is the derivative of F (x), then F (x) + c is the integral of
f (x), i.e., integration is a bit like the reverse process of differentiation.

Theorem 3.1 The fundamental theorem of calculus states that if f (x)

is the derivative of F (x), then the indefinite integral of f (x) is
Z
dF (x)
f (x)dx = F (x) + c, where f (x) = , (3.36)
dx

c is an arbitrary constant of integration, and f (x) is called the integrand.

Similarly, the definite integral of f (x) on some interval [a, b] is
Z b h ib
f (x)dx = F (x) ≡ F (b) − F (a), (3.37)
a a

where a and b are the limits of the integration.

We shall see in §3.2.3 that the fundamental theorem of calculus allows us to

evaluate a many integrals simply by considering the derivatives of functions.
In this way the integral of a function is sometimes referred to as its anti-
deriviative; however, differentiation does not have a unique inverse,
because the constant of integration c in equation (3.36) is arbitrary.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 111

3.2.3 Integration by inspection

The fundamental theorem of calculus allows us to easily integrate a function

f (x) if we know that it is the derivative of another function F (x); integrating
a function in this way is known as integration by inspection.

In §2.2.2 we tabulated a set of derivatives of standard functions (see table 2.1);

this table may be used in reverse to compute the integrals of standard functions
by inspection. We illustrate this idea in the Examples below.

Example 3.2 Let f (x) = xn , where n 6= −1 is a constant. Show that f (x)

is the derivative of
1
F (x) = xn+1 , (3.38)
n+1
and hence write down the integral of f (x) = xn by inspection.

I Solution: Differentiating F (x) we obtain

 
dF d 1 n+1 n+1 n
= x = x = xn = f (x), (3.39)
dx dx n + 1 n+1

as required. It follows from the fundamental theorem of calculus that the

indefinite integral of f (x) is
Z Z
1
f (x)dx = F (x) + c, that is, xn dx = xn+1 + c, (3.40)
n+1

where c is an arbitrary constant of integration, and n 6= 1. J

Example 3.3 Let f (x) = eax , where a is a constant. Show that f (x) is the
derivative of F (x) = eax /a, and hence write down the integral of f (x) = eax .

I Solution: Differentiating F (x) we obtain

d eax aeax
 
dF
= = = eax = f (x), (3.41)
dx dx a a

as required. It follows from the fundamental theorem of calculus that the

integral of f (x) is
Z Z
1
f (x)dx = F (x) + c, that is, eax dx = eax + c, (3.42)
a

where c is an arbitrary constant of integration. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

112 Section 3 : Integration

Example 3.4 Let f (x) = sin(ax), where a is a constant. Show that f (x) is
the derivative of
1
F (x) = − cos(ax), (3.43)
a
and hence write down the integral of f (x) = sin(ax).

I Solution: Differentiating F (x) we obtain

 
dF d 1 a
= − cos(ax) = sin(ax) = sin(ax) = f (x), (3.44)
dx dx a a

as required. It follows from the fundamental theorem of calculus that the

integral of f (x) is
Z Z
1
f (x)dx = F (x) + c, i.e, sin(ax)dx = − cos(ax) + c, (3.45)
a

where c is an arbitrary constant of integration. J

Example 3.5 Write down the indefinite integral of f (x) = 0.

I Solution: Observe that f (x) = 0 is the derivative of F (x) = c, where c is

an arbitrary constant. Hence, by the fundamental theorem of calculus we have
Z Z
f (x)dx = F (x) + c, that is, 0dx = c, (3.46)

where c is an arbitrary constant of integration. J

Example 3.6 Write down the integral of f (x) = 1/(x + a), with a as a
constant, given that (x + a) > 0.

I Solution: Observe that

d 1
loge (x + a) = , (3.47)
dx x+a

i.e., f (x) = 1/x is the derivative of F (x) = loge x. Hence, by the fundamental
theorem of calculus we have
Z Z
1
f (x)dx = F (x) + c, that is, dx = loge (x + a) + c, (3.48)
x

where c is an arbitrary constant of integration. J

Proceeding in this way—working in reverse from the derivatives of standard
functions—we may compile a list of standard integrals (see table 3.1).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 113

Function Integral Function Integral

Z Z
f (x) f (x)dx f 0 (x) f 0 (x)dx = f (x) + c
1 ax
0 c eax e +c
a
1 1
xn (n 6= −1) xn+1 + c loge |x + a| + c
n+1 x+a
1 1
sin(ax) − cos(ax) + c √ arcsin(x/a) + c
a a − x2
2
1 1
cos(ax) sin(ax) + c √ − arccos(x/a) + c
a a − x2
2
1 a
sec2 (ax) tan(ax) + c arctan(x/a) + c
a a + x2
2

1 1
cosh(ax) sinh(ax) + c √ arcsinh(x/a) + c
a a + x2
2
1 1
sinh(ax) cosh(ax) + c √ arccosh(x/a) + c
a a − x2
2
1 a
sech2 (ax) tanh(ax) + c arctanh(x/a) + c
a a2 − x2

Table 3.1: Integrals of common functions, with c as an arbitrary constant of integration.

Z π/4
1
Example 3.7 Show that cos(2x)dx = .
0 2

I Solution: Integrating by inspection (see table 3.1) we have

Z
1
cos(2x)dx = sin(2x) + c. (3.49)
2

Hence, the definite integral of interest is

Z π/4  π/4
1 1 1 1
cos(2x)dx = sin(2x) = sin(π/2) − sin(0) = , (3.50)
0 2 0 2 2 2

as required. J
Z 8
1
Example 3.8 Evaluate the definite integral I = dx.
2 x

I Solution: Integrating by inspection (see table 3.1) we have

Z 8
1 h i8
I= dx = loge x = loge 8 − loge 2 = loge 4 = 2 loge 2, (3.51)
2 x 2

where we used the rules of logarithms to write loge 4 = loge 22 = 2 loge 2. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

114 Section 3 : Integration

In addition to the standard integrals listed in table 3.1, we also often need to
make use of the following rules.

Result 3.2 Integration is a linear operation, that is, given two functions
f (x) and g(x), and two constants a and b, it may be shown that
Z Z Z
 
af (x) + bg(x) dx = a f (x)dx + b g(x)dx. (3.52)

For limits a and b, it may be shown that

Z b Z a Z a
f (x)dx = − f (x)dx and f (x)dx = 0. (3.53)
a b a

For three constants a, b and c satisfying a < b < c, it may be shown that
Z c Z b Z c
f (x)dx = f (x)dx + f (x)dx. (3.54)
a a b

These rules all follow from the results listed in Theorem 3.1.

Z 2
Example 3.9 Demonstrate that (4x3 − 8x + 1)dx = 2.
0

I Solution: Integrating term by term (due to the linearity of integration) we

obtain
Z 2
2
(4x3 − 8x + 1)dx = x4 − 4x2 + x 0 = 24 − 4 × 22 + 2 = 2

(3.55)
0

as required. J
Z
6
Example 3.10 Show that I = du = 3 arctan(u/2) + constant.
4 + u2

I Solution: Re-arranging the integrand, and setting a = 2 we obtain

Z Z Z
6 2 a
I= = 3 = 3 , (with a = 2)
4 + u2 22 + u2 a2 + u 2
= 3 arctan(u/a) + c, (with a = 2)
= 3 arctan(u/2) + c, (3.56)

where c is an arbitrary constant, and we used table 3.1.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 115

3.3 Methods of Integration

If integration by inspection is not possible, then other methods must be used.

In this section we discuss some common approaches to handling such integrals,
mainly centred around the idea of making judicious substitutions.

3.3.1 Integration by substitution

The basic idea of integration by substitution is to convert an unfamiliar inte-

grand into a form that can be integrated by inspection:

Method 3.1 Integration by substitution involves three basic steps:

1. Choose a substitution u(x) that transforms a ‘difficult’ looking inte-

R R
gral f (x)dx to an equivalent but simpler integral g(u)du, i.e.,
 
du
write f (x)dx in terms of u(x) and du = dx.
dx
2. Evaluate the resulting expression for the integral in terms of u.

3. Re-subsitute for u(x) in terms of x.

This approach is formalised using the chain rule for differentiation; however,
in practice we learn what types of substitutions are effective by experience.

Z
Example 3.11 Evaluate the indefinite integral I = 7(x + 1)6 dx.

I Solution: Let us try the substitution u(x) = (x + 1), then we have

 
6 6 du
7(x + 1) = 7u , and du = dx = dx. (3.57)
dx

Putting these expressions into our integral, we have by inspection that

Z Z
I = 7(x + 1) dx = 7u6 du = u7 + c.
6
(3.58)

Hence, re-substituting u = (x + 1) we obtain I = (x + 1)7 + c. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

116 Section 3 : Integration
Z
1
Example 3.12 Evaluate the integral I = dt.
2t + 3
I Solution: Let us try the substitution u(t) = (2t + 3), then we have
 
1 1 du 1
= , and du = dt = 2dt, i.e., dt = du. (3.59)
2t + 3 u dt 2

Putting these expressions into our integral, we have by inspection that

Z Z
1 1 1 1
I= dt = du = loge |u| + c. (3.60)
2t + 3 2 u 2

Hence, re-substituting u = (2t + 3) we obtain I = 21 loge |2t + 3| + c. J

Z
Example 3.13 Evaluate the integral I = 4x cos(x2 )dx.

I Solution: Let us try to eliminate the problem caused by x2 by using the

substitution u(x) = x2 , then

du 1
cos(x2 ) = cos(u), and = 2x, i.e., dx = du (3.61)
dx 2x
Thence, putting these expressions into our integral, we have
Z Z Z
2 du
I = 4x cos(x )dx = 4x cos(u) = 2 cos(u)du, (3.62)
2x

whereupon integrating by inspection we obtain

Z
I = 2 cos(u)du = 2 sin(u) + c = 2 sin(x2 ) + c, (3.63)

where we re-substituted u = x2 . Z J
Example 3.14 Evaluate the integral I = 30x2 (x3 + 8)4 dx.

I Solution: Let us try the substitution u(x) = (x3 + 8), then

du 1
(x3 + 8)4 = u4 , and = 3x2 , i.e., dx = du (3.64)
dx 3x2
Thence, putting these expressions into our integral, we have
Z Z Z
2 3 4 2 4 1
30x (x + 8) dx = 30x u du = 10u4 du = 2u5 + c. (3.65)
3x2

Hence, re-substituting u = (x3 + 8) we obtain I = 2(x3 + 8)5 + c. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 117

3.3.2 Integration using trigonometric identities

For integrands involving trigonometric functions, it is often necessary to make

use of identities before possible substitutions present themselves. Particularly
useful identities include
sin2 θ + cos2 θ ≡ 1, (3.66)

and the double angle formulae

cos(2θ) ≡ 2 cos2 θ − 1 and sin(2θ) ≡ 2 sin θ cos θ. (3.67)

It is sometimes necessary to apply these identity repeatedly.

Z
Example 3.15 Evaluate the integral I = cos3 θdθ.

I Solution: Notice that we may use the identity 1 ≡ (sin2 θ + cos2 θ) to write
Z Z Z
I = cos θdθ = cos θ × cos θdθ = (1 − sin2 θ) cos θdθ.
3 2
(3.68)

Because cos θ = is the derivative of sin θ, it is prudent to try the substitution

u = sin θ. Indeed, in this way we have

du
(1 − sin2 θ) = (1 − u2 ), and du = dθ = cos θdθ, (3.69)
dθ
Thence, putting these expressions into our integral, we have
Z Z
1
I = (1 − sin θ) cos θdθ = (1 − u2 )du = u − u3 + c.
2
(3.70)
3

Hence, re-substituting u = sin θ we obtain I = (sin θ − 31 sin3 θ + c). J

Z
Example 3.16 Evaluate the integral I = 2 cos2 (3φ)dφ.

I Solution: By the double angle formulae 2 cos2 θ ≡ [cos(2θ) + 1] we have

Z Z Z
2
I = 2 cos (3φ)dφ = [cos(2 × 3φ) + 1]dφ = [cos(6φ) + 1]dφ. (3.71)

Hence, completing integrating by inspection we obtain

1
I= sin(6φ) + φ + c, (3.72)
6
where c is an arbitrary constant (see table 3.1). J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

118 Section 3 : Integration

3.3.3 Checking indefinite integrals

The fundamental theorem of calculus states that

Z
dF (x)
f (x)dx = F (x) + c where = f (x). (3.73)
dx

When computing an indefinite integral, therefore, one can always check one’s
work by ensuring that F 0 (x) = f (x).
Example 3.17 In Example 3.15 we found that
Z
1
I(θ) = cos3 θdθ = sin θ − sin3 θ + c, (3.74)
3

where c is an arbitrary constant. Confirm this result by differentiating I(θ).

I Solution: Differentiating I(θ) we have as required that

   
dI d 1 3 d 1 3
= sin θ − sin θ + c = cos θ − sin θ
dθ dθ 3 dθ 3
= cos θ − cos θ sin2 θ = cos θ(1 − sin2 θ) = cos3 θ, (3.75)

where we differentiated 1
3
sin3 by the chain rule. J

3.3.4 Logarithmic integration

A special situation arises when integrand may be written in terms of a function’s

derivative u0 (x) divided by the function u(x) itself. In such cases one has

u0 (x)
Z Z Z
1 du 1
dx ≡ dx = du = loge u(x) + c, (3.76)
u(x) u dx u

where the final step follows by inspection (see table 3.1). This method of
integration is sometimes called logarithmic integration.

Result 3.3 If an integrand is of the form u0 (x)/u(x), then

u0 (x)
Z
dx = loge u(x) + c, (3.77)
u(x)

where c is an arbitrary constant.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 119
Z
Example 3.18 Evaluate the integral I = tan θdθ.

I Solution: By the definition of tan θ that this integral may be written as

Z Z
sin θ
I = tan θdθ = dθ. (3.78)
cos θ

Let u(θ) = cos θ, such that u0 (θ) = − sin θ, then

u0 (θ)
Z Z Z
1 du 1
I=− dθ = − dθ = − du = − loge |u| + c. (3.79)
u(θ) u dθ u

Thence, re-substituting for u(θ) = cos θ we have

I = − loge | cos θ| + c = loge |1/ cos θ| + c = loge | sec θ| + c, (3.80)

where we used the rule for logarithms that − loge a = loge a−1 = log(1/a). J

3.3.5 Integration by completing the square

Sometimes it is necessary to perform algebraic manipulations before an obvious

substitution presents itself; one useful approach is to complete the square.
Z
6
Example 3.19 Evaluate the integral I = 2
dt.
t + 2t + 5
I Solution: Notice that by completing the square in the denominator we have
Z Z
6 6
I= dt = dt. (3.81)
t2 + 2t + 5 (t + 1)2 + 4

Now let us try the substitution u(t) = (t + 1), for which

 
6 6 6 du
2
= 2 = 2
and du = dt = dt. (3.82)
(t + 1) + 4 u +4 4+u dt

Putting these expressions into our integral, we have

Z Z
6 6
I= dt = du. (3.83)
(t + 1)2 + 4 4 + u2

This final form for I may be evaluated using the method in Example 3.10, i.e,
Z
6
I= du = 3 arctan(u/2) + c. (3.84)
4 + u2

Thus, re-substituting for u = (t + 1) we have I = 3 arctan[(t + 1)/2] + c. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

120 Section 3 : Integration

3.3.6 Integration by partial fractions

Recall that the quotient of two polynomials is known as a rational function

(see 1.4.8 and §1.5); for example, the function

P (t) t−7
φ(t) = = 2 (3.85)
Q(t) t − 2t − 3

is a rational function because it is the quotient of two polynomials P (t) = t−7

and Q(t) = t2 −2t−3. Integrands involving rational functions may be simplified
using the method of partial fractions (see 1.4.8 and §1.5).

Example 3.20 Use the method of partial fractions to express the function

t−7
φ(t) = (3.86)
t2 − 2t − 3
R
as the sum of two fractions. Hence evaluate the integral I = φ(t)dt.

I Solution: By factorising the denominator we have

t−7 t−7
φ(t) = = ; (3.87)
t2 − 2t − 3 (t + 1)(t − 3)

this expression suggests that we look for partial fractions of the form (see §1.5)

t−7 A B
= + , (3.88)
(t + 1)(t − 3) (t + 1) (t − 3)

where A and B are constants. Multiplying through by (t + 1)(t − 3) we have

t − 7 = A(t − 3) + B(t + 1). (3.89)

The constants A and B may be found as follows (see §1.5):

(i) Setting t = −1 in equation (3.89) we obtain −8 = −4A, that is, A = 2.

(ii) Setting t = 3 in equation (3.89) we obtain −4 = 4B, that is, B = −1.

With A = 2 and B = −1, equation (3.88) then gives

t−7 2 1
φ(t) = = − (3.90)
t2 − 2t − 3 (t + 1) (t − 3)

as φ(t) expressed in the form of partial fractions.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 121

When expressed as partial fractions, φ(t) may be integrated by inspection, viz.

t−7
Z Z Z Z
2 1
I = φ(t)dt = 2
dt = dt − dt
t − 2t − 3 (t + 1) (t − 3)
= 2 loge |t + 1| − loge |t − 3| + c, (3.91)

where c is an arbitrary constant (see table 3.1). J

3.3.7 Tangent half-angle substitution

The so-called tangent half-angle substitution is a special kind of substitution

that may be used to express trigonometric functions as rational functions; in
this way any method for evaluating integrals of rational functions can then be
used to integrate trigonometric functions. We motivate the background to this
substitution in in the following example.

Example 3.21 Figure 3.4 depicts a right angled triangle with an adjacent side
of length 1, and an opposite side of length t. If we denote the angle of such a
triangle as θ/2, then
t
t = = tan(θ/2). (3.92)
1
This expression can be used as a substitution to evaluate integrals.
(a) With reference to figure 3.4, use a geometric argument to show that

1 t
cos(θ/2) = √ , and sin(θ/2) = √ . (3.93)
1 + t2 1 + t2

(b) The double angle formulae can be expressed in the form

cos θ ≡ 2 cos2 (θ/2) − 1, and sin θ ≡ 2 sin(θ/2) cos(θ/2); (3.94)

demonstrate that
1 − t2 2t 2t
cos θ = , sin θ = , tan θ = . (3.95)
1 + t2 1 + t2 1 − t2

(c) Given t = tan(θ/2), show that if the magnitude of t changes by an

infinitesimal amount dt, then the angle θ changes by an infinitesimal
amount  
dθ 2
dθ = dt = dt. (3.96)
dt 1 + t2
[Hint: You may wish to use the fact that sec2 φ ≡ (tan2 φ + 1).]

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

122 Section 3 : Integration

Figure 3.4: Triangle with adjacent side 1, and opposite side t, subtending an angle θ/2.

I Solution: (a) With reference to figure 3.4, if the adjacent side is length 1,
and the opposite side is length t, then by Pythogoras’s theorem the hypotenuse
√ √
has length 12 + t2 = 1 + t2 . It then follows by definition that

1 t
cos(θ/2) = √ , and sin(θ/2) = √ . (3.97)
1 + t2 1 + t2

(b) Using the double angle formulae, we have by equation (3.97) that

1 2 1 + t2 1 − t2
cos θ ≡ 2 cos2 (θ/2)−1 = 2× −1 = − = , (3.98)
1 + t2 1 + t2 1 + t2 1 + t2
and similarly

t 1 2t
sin θ ≡ 2 sin(θ/2) cos(θ/2) = 2 × √ ×√ = . (3.99)
1+t2 1+t2 1 + t2

Thence, with these expressions for sin θ and cos θ,

sin θ 2t 1 + t2 2t
tan θ = = 2
× 2
= . (3.100)
cos θ 1+t 1−t 1 − t2

Equations (3.98), (3.99), and (3.98) are the required forms of equation (3.95).
(c) With t = tan(θ/2) we have by the chain-rule that

dt d d 1
= tan(θ/2) = sec2 (θ/2) (θ/2) = sec2 (θ/2). (3.101)
dθ dθ dθ 2
Appealing to the hint sec2 φ ≡ (tan2 φ + 1) this equation may be written

dt 1 1 dθ 2
= (tan2 (θ/2) + 1) = (t2 + 1), that is, = 2 . (3.102)
dθ 2 2 dt t +1
It therefore follows that
 
dθ 2
dθ = dt = dt (3.103)
dt 1 + t2

as required. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 123

Summarising the results of Example 3.21 we have the following:

Result 3.4 The tangent half-angle angle substitution is

2
t = tan(θ/2), with dθ = dt, (3.104)
1 + t2
and may be used to express trigonometric functions of θ in the form of
rational functions of t. In particular, the substitution yields

2t 1 − t2 2t
sin θ = , cos θ = , tan θ = , (3.105)
1 + t2 1 + t2 1 − t2

(see Example 3.21). Use of this substitution is illustrated in Example 3.22.

Z
3
Example 3.22 Let I = dθ, use the tangent half-angle
sin θ + 2 cos θ + 3
substitution (Result 3.4), and the solution from Example 3.19 to show that

I = 3 arctan[(tan(θ/2) + 1)/2] + constant. (3.106)

I Solution: Using the substitution t = tan(θ/2) we have by the expressions

for sin θ, cos θ, and dθ in Result 3.4 that
Z
3
I= dθ
sin θ + 2 cos θ + 3
Z  
3 2dt
= 1−t2
2t
+ 3 1 + t2



2 + 2 2
Z 1+t 1+t
3
= 2dt
2t + 2(1 − t ) + 3(1 + t2 )
2
Z
6
= 2
dt. (3.107)
t + 2t + 5

In Example 3.19 we found that this integral may be evaluated to give

I = 3 arctan[(t + 1)/2] + c, (3.108)

where c is an arbitrary constant. Hence, re-subsituting for t we have

I = 3 arctan[(tan(θ/2) + 1)/2] + c, (3.109)

as required. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

124 Section 3 : Integration

3.3.8 Integration by parts

Suppose that f (x) is the product of two functions u(x) and v(x), that is,

f (x) = u(x)v(x), (3.110)

then differentiating f (x) by the product rule gives

df d dv du
= (uv) = u +v . (3.111)
dx dx dx dx
Integrating this expression, therefore we have
Z Z Z
df dv du
f (x) = dx = u dx + v dx. (3.112)
dx dx dx

Thence, rearranging we obtain

Z Z Z
dv du du
u dx = f (x) − v dx = uv − v dx, (3.113)
dx dx dx

where we used the fact that f (x) = u(x)v(x). This equation forms the basis
for the method of integration by parts.

Result 3.5 If an integrand can be written in the form u(x)v 0 (x), then it
may be shown that
Z Z
dv du
u dx = uv − v dx. (3.114)
dx dx

This expression may be used to perform integration by parts.

The objective of integration by parts, therefore is to express an integrand as the

product of two functions u(x) and v 0 (x), where v 0 (x) is the derivative of some
function v(x), and then to apply equation (3.114); knowing which sorts of
products will work requires experience (and occasionally some trial and error).
Since equation (3.114) replaces one integral (the left-hand-side) with another
(the right-hand-side), it is important to make a choice which leaves an integral
that is easier to evaluate than the one we are initially faced with.
Note: It is sometimes necessary to use the expression for integrating by parts
successively before the integral can be evaluated by inspection.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 125

Example 3.23 Use the formula for integration by parts, i.e.,

Z Z
dv du
u dx = uv − v dx, (3.115)
dx dx
Z
to evaluate I = 2x cos xdx.

I Solution: To use the formula for integration by parts we must write the
dv
integrand 2x cos x in the form u dx . One possibility is to set

dv
u = 2x and = cos x; (3.116)
dx

this is a good choice because dudx

= 2 is a constant, which will simplify the
integral on the right-hand-side of equation (3.115). With this choice

du
=2 and v = sin x. (3.117)
dx
Hence, appealing to equation (3.115), we have
Z Z Z
dv du
I = 2x cos xdx = u dx = uv − v dx
dx dx
Z
= 2x sin x − 2 sin xdx

= 2x sin x + 2 cos x + c, (3.118)

where c is an arbitrary constant. J

Example 3.24 By writing u(x) = loge x and v(x) = x, show that
Z Z
dv
loge xdx = u dx. (3.119)
dx

Hence use integration by parts to determine the indefinite integral of loge x.

I Solution: By writing u = loge x and v = x, we have du dx

= x1 and dx
dv
= 1.
As required, therefore, the integral of loge x may be evaluated by parts, viz.
Z Z Z Z
dv du 1
loge x = u dx = uv − v dx = x loge x − x × dx
dx dx x
Z
= x loge x − 1dx = x loge x − x + c, (3.120)

where c is an arbitrary constant. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

126 Section 3 : Integration

As we illustrate in the following example, it is sometimes necessary to apply

the formula for integration by parts repeatedly.

Example 3.25 Use the formula for integration by parts to show that
Z Z
˜ 2 2
I = x sin xdx = −x cos x + 2x cos xdx. (3.121)

Hence complete the integration by using integration by parts a second time.

dv
I Solution: Suppose that we choose to write x2 sin xdx as u dx by setting

dv
u = x2 and = sin x; (3.122)
dx

this is a good choice because du

dx
= 2x is linear in x (a linear term is simpler
2
than the quadratic term x in our original integral). With this choice we have

du
= 2x and v = − cos x, (3.123)
dx

such that I˜ may be evaluated using integration by parts as

Z
˜
I = x2 sin xdx
Z
dv
= u dx
dx
Z
du
= uv − v dx
dx
Z
2
= −x cos x + 2x cos xdx (3.124)

as required. Here we may use integration by parts a second time to evaluate

the the final term; indeed, by our solution to Example 3.23 we have
Z
2x cos xdx = 2x sin x + 2 cos x + c, (3.125)

where c is an arbitrary constant. Thence, combining equations (3.124) and

(3.125) we obtain

I˜ = −x2 cos x + 2x sin x + 2 cos x + c. (3.126)

Note: For integrals involving an arbitrary index n ∈ N, integration by parts

may be used repeatedly to derive what are called reduction formulae. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 127

3.4 Area and Symmetry

In §3.1.1 we motivated the geometric interpretation for the value of an integral

Z b
I= f (x)dx (3.127)
a

as the area bounded by the curve y = f (x) and the x-axis on [a, b]. As
illustrated by the odd and even functions depicted in figure 3.5, portions of the
curve above the x-axis have y = f (x) > 0, and yield a positive integral; for this
reason we say that the area above the x-axis is positive (‘+’). Similarly,
portions of the curve below the x-axis have y = f (x) < 0, and yield a negative
integral; in this case we say that the area below the x-axis is negative (‘−’).

Now suppose that we integrate an even function on x ∈ [−a, a]. It follows by

figure 3.5 (a), and the meaning of integration as the area under the curve, that
the integral on x ∈ [−a, a] is twice the value of the integral on x ∈ [0, a], i.e.,
Z +a Z +a
for f (x) even: f (x)dx = 2 × f (x)dx. (3.128)
−a 0

In contrast, if we were to evaluate the integral of an odd function on the interval

x ∈ [−a, a], then it follows by figure 3.5 (b) that the integral on x ∈ [−a, 0]
will cancel-out the value to the integral on x ∈ [0, a], i.e.,
Z +a
for f (x) odd: f (x)dx = 0. (3.129)
−a

Thus, we can exploit the symmetry of a function to evaluate definite integrals.

(a) (b)

Figure 3.5: An even function (a), and an odd function (b). The area bounded by the
curves (shaded) is positive (‘+’) above the x axis, and negative (‘−’) below the x axis.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

128 Section 3 : Integration

Result 3.6 The geometric interpretation for the integral

Z b
I= f (x)dx (3.130)
a

is the area bounded by the curve y = f (x) and the x-axis on the interval
[a, b]. Above the x-axis the area is positive, whereas below the x-axis the
area is negative; in this way it may be shown that
Z +a Z +a
for f (x) even about x = 0: f (x)dx = 2 f (x)dx. (3.131)
−a 0
Z +a
for f (x) odd about x = 0: f (x)dx = 0. (3.132)
−a

Note: These results can be generalised to symmetry about a point.

Example 3.26 Show that area A bounded by the curve y = 8x3 − 6x2 and
the x-axis on the interval x ∈ [−1, 2] is A = 12.

I Solution: By the geometric meaning of integration we have

Z +2 +2
8x3 − 6x2 dx = 2x4 − 2x3 −1

A=
−1

= 2(+2)4 − 2(+2)3 − 2(−1)4 − 2(−1)3

   

= 24 − 4 = 12, (3.133)

as required. J
Z +9
2
Example 3.27 Evaluate the integral I = e−x sin(5x)dx.
−9

I Solution: Although this integral does not look easy to evaluate, notice that
2
e−x is an even function, and sin(5x) is an odd function. We thus have
Z +9
I= (even function) × (odd function)dx
−9
Z +9
= (odd function)dx = 0, (3.134)
−9

where we appealed to the symmetry result in equation (3.132), and used the
fact that (even func.) × (odd func.) = (odd func.), as in Definition §1.6. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 129

3.5 Improper Integrals

When one (or both) of the limits on an integral is infinite the integral is called
an improper integral. For example, the integral
Z ∞
I= f (x)dx, (3.135)
a

is an improper integral. Since ‘infinity’ is not a number, the precise meaning

of this expression is “evaluate the integral of f (x) on [a, b] in the limit that
b → ∞”; the notation should therefore be interpreted as
Z ∞ Z b
f (x)dx ≡ lim f (x)dx. (3.136)
a b→∞ a

Likewise, a lower limit of ‘minus infinity’ should be interpreted as

Z b Z b
f (x)dx ≡ lim f (x)dx. (3.137)
−∞ a→−∞ a

Improper integrals may therefore be evaluated by taking an appropriate limit

of the corresponding definite integral.

Example 3.28 Evaluate the improper integral

Z ∞
F (s) = e−st dt, (3.138)
0

where s > 0 is a real number.

I Solution: The meaning of the improper integral is

Z ∞ Z b
−st
F (s) = e dt = lim e−st dt
0 b→∞ 0
 b
1 −st
= lim − e
b→∞ s
 0  
1 −sb 1 0
= lim − e − − e
b→∞ s s
1 1
= − lim e−sb
s s b→∞
1
= (s > 0). (3.139)
s
Note: The integral F (s) = 1/s is an example of a Laplace transform. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

130 Section 3 : Integration

A second kind of improper integral (optional)

Another kind of improper integral occurs when a function f (x) diverges at

some point c in the interval [a, b] of interest. For instance, the integral
Z 4
I= f (x)dx, with f (x) = x−1/2 , (3.140)
0

is improper because f (x) diverges at c = 0 ∈ [0, 4]. In these cases the integral
must be evaluated by excluding the unbounded point c ∈ [a, b], i.e., by writing
Z b Z c−ε Z b
f (x)dx = lim f (x)dx + lim f (x)dx, (3.141)
a ε→0 a ε→0 c+ε

where ε ∈ R is a vanishingly small number.

Z 4
Example 3.29 Evaluate the improper integral I = x−1/2 dx.
0

I Solution: Notice that the integrand x−1/2 diverges at x = 0, so this point

must be excluded from our integration interval [0, 4]. Following the approach
summarised in equation (3.141), therefore, we have
Z 4  √ 4 √ √
I = lim x−1/2 dx = lim 2 x ε = 2 4 − lim 2 ε = 4, (3.142)
ε→0 ε ε→0 ε→0

where ε ∈ R is a vanishingly small number. J

3.6 Mean Value of a Function

The mean value hf i of a function f (x) on an interval x ∈ [a, b] is defined by

the integral Z b
1
hf i = f (x)dx. (3.143)
(b − a) a
In this way, the mean value hf i may be thought of as the height of a rectangle
with base length (b − a) which has the same area as that bounded by the curve
f (x) on [a, b] (see figure 3.6). It may be shown that if f (x) is a continuous
function on x ∈ [a, b], then there exists some value c ∈ [a, b] such that

hf i = f (c). (3.144)

This result is known as the mean value theorem for definite integrals.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 3 : Integration 131

Figure 3.6: Mean value of a function hf i = f (c) as the height of a rectangle with base
length (b − a) which has the same area as that bounded by the curve f (x) on x ∈ [a, b].

Definition 3.2 The mean value hf i of a function f (x) on an interval

x ∈ [a, b] is defined by
Z b
1
hf i = f (x)dx. (3.145)
(b − a) a

According to the mean value theorem, if f (x) is a continuous function

on x ∈ [a, b], then there exists some value c ∈ [a, b] such that hf i = f (c).

Example 3.30 Determine the mean value hf i of f (x) = 15 x2 + 2x on the

interval [2, 5]. Hence find some value c ∈ [2, 5] such that f (c) = hf i.

I Solution: By definition we have that

Z 5
1 5 1 2
Z
1
hf i = f (x)dx = ( x + 2x)dx
(5 − 2) 2 3 2 5
11 3 5 1  125  48
= 15
x + x2 2 = 8
( 15 + 25) − ( 15 + 4) = . (3.146)
3 3 5
To find value of c ∈ [2, 5] such that f (c) = hf i we must therefore solve

f (c) = 51 c2 + 2c = 48
5
= hf i, that is, c2 + 10c − 48 = 0. (3.147)
√
This is a quadratic equation with solutions c = (± 73 − 5). Thus, selecting
√
the value that lies in the interval [2, 5], we have c = ( 73 − 5) ≈ 3.54. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

132 Section 3 : Integration

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 Engineering Mathematics
J. J. Bissell

Complex Numbers

Complex numbers were developed in response to the search for solutions to

equations such as z 2 = −1. In particular, because any real number z ∈ R
satisfies z 2 > 0, solutions to z 2 = −1 require a new kind of number j with the
property j 2 = −1. The number j is called the imaginary unit, and belongs to
a set of numbers called the imaginary numbers. Real numbers can be combined
with imaginary numbers to define composite numbers that contain both a real
and imaginary part; we call such numbers complex numbers C.

The terminology imaginary and complex is a result of historical legacy, and it

is important to realise that neither of these kinds of numbers are any less ‘real’,
or more ‘complex’ than real numbers. Thus, complex numbers can be used to
describe objects in the physical world in much the same way as other numbers,
but with a far greater range of application. In this section we illustrate the
power of complex numbers; by the end of the section you should be able to. . .

Learning outcomes:

◦ Write down a complex number z ∈ C in the form z = x + yj, with x, y ∈ R.

◦ Add, subract, multiply, and divide complex numbers.
◦ Write down the complex conjugate z = x − yj of a complex number z.
◦ Represent complex numbers graphically using Argand diagrams.
◦ Determine the modulus r = |z| and argument θ of a complex number.
◦ Express complex numbers in polar form, and exponential form z = rejθ .
◦ Use Euler’s formula ejθ = cos θ + j sin θ to obtain trigonometric identities.
◦ Derive de Moivre’s theorem (cos θ + j sin θ)n = (cos(nθ) + j sin(nθ)).
◦ Use complex numbers to compute roots, and solve simple equations.

These learning outcomes must be reinforced by completing course exercises.

133
134 Section 4 : Complex Numbers

4.1 What is a Complex Number?

We introduced this section by discussing the quadratic equation

z 2 + 1 = 0, or equivalently z 2 = −1; (4.1)

this equation does not have any solutions given by real numbers because z 2 > 0
for all z ∈ R. To circumvent this problem we introduce a mathematical object
j known as the imaginary unit, which is defined such that

j 2 = −1. (4.2)

The imaginary unit j behaves like any other algebraic quantity—it can be
added, subtracted, multiplied, etc.—we just have to remember that j 2 = −1.
In this way equation (4.1) can now be written in the form

z 2 = −1 = j 2 , that is z = ±j. (4.3)

The equation therefore has two solutions, namely z = −j and z = +j, just
like any other quadratic equation.

Definition 4.1 The imaginary unit is denoted i, and defined such that

i2 = −1. (4.4)

Engineers often denote the imaginary unit using the letter j, that is,

j 2 = −1, (4.5)

so that i can be used to denote current. Either notation is acceptable.

Example 4.1 Determine both solutions to the quadratic equation z 2 + 4 = 0.

I Solution: We notice that the quadratic may be written as

z 2 = −4 = j 2 22 = (2j)2 . (4.6)

Thus, taking the square-root, the solutions are z = ±2j. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 135

4.1.1 Cartesian form

Now consider the following quadratic equation

z 2 − 2z + 2 = 0. (4.7)

To solve this equation, we complete the square to give

(z − 1)2 = −1, that is (z − 1)2 = j 2 , (4.8)

since j 2 = −1. Thus, taking the square-root, the two solutions for z are

z = 1 ± j. (4.9)

These solutions lead us to the idea of a complex number.

Definition 4.2 A complex number is a mathematical object of the form

z = x + yj, where x, y ∈ R (4.10)

are real numbers, and j is the imaginary unit. The right-hand-side of

equation (4.10) is called the Cartesian form, in particular, we say that

◦ the real part of z is x ∈ R, and is written Re{z} = x.

◦ the imaginary part of z is y ∈ R, and is written Im{z} = y.

The set of complex numbers is denoted C.

Example 4.2 Determine the real and imaginary parts of the complex numbers:
√ √
(a) z1 = 4 + 52 j; (b) z2 = −2 2 + 2 2j; (c) z3 = −π − 32 j; (d) z4 = −2j.

I Solution: According to Definition 4.2, the real and imaginary parts are:

(a) Re{z1 } = 4 and Im{z1 } = 25 .

√ √
(b) Re{z2 } = −2 2 and Im{z2 } = 2 2.

(c) Re{z3 } = −π and Im{z3 } = − 23 .

(d) Re{z4 } = 0 and Im{z4 } = −2.

These numbers are depicted graphically in figure 4.1. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

136 Section 4 : Complex Numbers

4.1.2 Complex plane and Argand diagrams

Every complex number z = x + yj may be represented by a unique point (x, y)

in a two-dimensional space known as the complex plane (see figure 4.1). In
this system the x-axis is known as the real axis (Re), and the y-axis is known
as the imaginary axis (Im). When complex numbers are depicted in this way
the resulting graphic is called an Argand diagram.

The fact that each complex number describes a unique point (x, y) in the
complex plane reflects the fact that two complex numbers are equal if and only
if they have identical real (x) and imaginary (y) parts. This property is known
as the law of equality, and is considered in more detail in §4.2.4.

Notice that a complex number without an imaginary part (Im{z} = 0) is a

real number, and lies on the real axis; in this way the real numbers are a
subset of the complex numbers. Similarly, a complex number without a real
part (Re{z} = 0) is called an imaginary number, and lies on the imaginary
axis; for example, z4 = −2j is an imaginary number (see figure 4.1). Thus,
the imaginary numbers are also a subset of the complex numbers.

The zero complex number

z=0 (4.11)

has real part Re{z} = 0 and imaginary part Im{z} = 0. By convention we

say that the number z = 0 is both real and imaginary.

Figure 4.1: Argand diagram of the complex numbers defined in Example 4.2.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 137

4.2 Algebra of Complex Numbers

Arithmetic operations on complex numbers—addition, subtraction, multiplica-

tion, and division—follow as with real numbers, and use the same symbols;
however, the symbol j is treated algebraically, according to j 2 = −1.

4.2.1 Addition and subtraction

In general, if z1 = x1 + y1 j and z2 = x2 + y2 j, then

z1 + z2 = (x1 + x2 ) + (y1 + y2 )j (4.12)

We illustrate this idea in the following exercise.

Example 4.3 Given that z1 = 2 + 3j and z2 = 5 + j, determine the following

complex numbers: (a) z1 + z2 ; (b) z2 + z1 ; (c) z1 − z2 ; and (d) z2 − z1 .

I Solution: Computing each complex number in turn we have

(a) z1 + z2 = (2 + 3j) + (5 + j) = (2 + 5) + (3 + 1)j = 7 + 4j.

(b) z2 + z1 = (5 + j) + (2 + 3j) = (5 + 2) + (1 + 3)j = 7 + 4j.

(c) z1 − z2 = (2 + 3j) − (5 + j) = (2 − 5) + (3 − 1)j = −3 + 2j.

(d) z2 − z1 = (5 + j) − (2 + 3j) = (5 − 2) + (1 − 3)j = 3 − 2j.

Notice that z1 + z2 = z2 + z1 is true for any pairs of complex numbers. J

4.2.2 Complex numbers as vectors

In section §4.1.2 we saw that an Argand diagram can be used to represent a

complex number z = x + yj as a point (x, y) in the complex plane. This idea
can be given an alternative interpretation by thinking of a complex number as
a a vector with one component that points in the direction of the real axis,
and another component that points in the direction of the imaginary axis. In
this way the addition and subtraction of complex numbers works in the same
way as vector addition, i.e., by real and imaginary components. We illustrate
this idea in figure 4.2 using the complex numbers z1 = 2 + 3j and z2 = 5 + j.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

138 Section 4 : Complex Numbers

Figure 4.2: Complex addition as vector addition, (2 + 3j) + (5 + j) = (7 + 4j).

4.2.3 Multiplication

Complex numbers may be multiplied together according to the standard rules

of arithmetic, using j symbolically, and with the property that j 2 = −1. In
general, therefore, if z1 = x1 + y1 j and z2 = x2 + y2 j, then

z1 z2 = (x1 + y1 j)(x2 + y2 j)
= x1 x2 + x1 y2 j + x2 y1 j + y1 y2 j 2
= x1 x2 + (x1 y2 + x2 y1 )j − y1 y2
= (x1 x2 − y1 y2 ) + (x1 y2 + x2 y1 )j. (4.13)

Rather than commit this expression to memory, it is much easier to multiply

complex numbers in context.

Example 4.4 Let z1 = 1 + 3j, z2 = 4 − j, z3 = √12 (1 + j). Compute the

following complex numbers: (a) z1 z2 ; (b) 2z1 + z2 ; (c) z32 ; and (d) z34 .

I Solution: Computing each complex number in turn, with j 2 = −1 we have

(a) z1 z2 = (1 + 3j)(4 − j) = 4 − j + 12j − 3j 2 = 4 + 11j + 3 = 7 + 11j.

(b) 2z1 + z2 = 2(1 + 3j) + (4 − j) = (2 + 6j) + (4 − j) = 6 + 5j.

(c) z32 = z3 z3 = 21 (1 + j)(1 + j) = 12 (1 + 2j + j 2 ) = 12 (1 + 2j − 1) = j.

(d) By part (c) we obtain z34 = (z32 )2 = j 2 = −1.

Notice here that part (c) implies z3 = √1 (1 + j) is the square root of j. J

2

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 139

4.2.4 Equating real and imaginary parts

We now show that two complex numbers are equal if and only if they have
identical real and imaginary parts. To this end let z = x + yj and w = u + vj
be complex numbers with x, y, u, v ∈ R, and suppose that z = w; then

z=w ⇔ (x + yj) = (u + vj)

⇔ (x − u) = (v − y)j
⇔ (x − u)2 = (v − y)2 j 2
⇔ (x − u)2 = −(v − y)2 . (4.14)

Since x, y, u, v ∈ R are real numbers, we have (x−u)2 ≥ 0 and −(v −y)2 ≤ 0.

It follows, therefore, that the final line of equation (4.14) is true if and only if
x = u and y = v. We have therefore proved the following law of equality

x + yj = u + vj ⇔ x=u and y = v. (4.15)

Identifying x = u and y = v is known as equating real and imaginary parts.

Result 4.1 Let z1 = x1 + y1 j and z2 = x2 + y2 j be complex numbers with

x1 , x2 , y1 , y2 ∈ R. The law of equality states that z1 and z2 are equal if
and only if Re{z1 } = Re{z2 } and Im{z1 } = Im{z2 }, that is

x1 + y 1 j = x2 + y 2 j ⇔ x1 = x 2 and y1 = y2 . (4.16)

If z1 = z2 , then the process of inferring that Re{z1 } = Re{z2 } and

Im{z1 } = Im{z2 } is called equating real and imaginary parts.

Example 4.5 Let z1 = 1 + 2j and z2 = 3 − j. Determine the values of

a, b ∈ R which satisfy (a + bj) = z1 − 2z2 .

I Solution: If (a + bj) = z1 − 2z2 , then

a + bj = (1 + 2j) − 2(3 − j) = −5 + 4j. (4.17)

Hence, equating real and imaginary parts we require a = −5 and b = 4. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

140 Section 4 : Complex Numbers

Example 4.6 Let z1 and z2 be the complex numbers z1 = (a + 2b) − 3j and

z2 = 1 + (a + b)j, where a, b ∈ R. Given that z1 = z2 , determine a and b.

I Solution: By the definitions of z1 and z2 we have

z1 = z2 ⇔ (a + 2b) − 3j = 1 + (a + b)j. (4.18)

Hence, equating real and imaginary parts we require

a + 2b = 1 and a + b = −3. (4.19)

These equations may be solved simultaneously to give a = −7 and b = 4. J

4.2.5 Complex conjugate

The imaginary unit j allows us to perform operations on complex numbers that

are not possible with real numbers; one of these is the complex conjugate.

Definition 4.3 Let z = x + yj be a complex number, with x, y ∈ R.

Then the complex conjugate of z is denoted z, and defined by

z = x − yj. (4.20)

In this way the complex conjugate z̄ is defined such that

Re{z} = Re{z} and Im{z} = −Im{z}. (4.21)

The complex conjugate of z is also sometimes denoted as z ∗ . When we

write down z̄ we say that we are taking the complex conjugate of z.

Example 4.7 Write down the complex conjugates of the complex numbers:
√ √
(a) z1 = 4 + 52 j; (b) z2 = −2 2 + 2 2j; (c) z3 = −π − 32 j; (d) z4 = −2j.

I Solution: By definition the complex conjugates are given by: (a) z 1 = 4− 52 j;

√ √
(b) z 2 = −2 2 − 2 2j; (c) z 3 = −π + 32 j; (d) z 4 = 2j. J

Example 4.8 Let z = x + yj with x, y ∈ R; show that (z) = z.

I Solution: Since z = x − yj we have (z) = (x − yj) = (x + yj) = z. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 141

Example 4.9 Let z be the complex conjugate of z = x + yj. Show that: (i)
z = z if and only if Im{z} = 0; and (ii) z = −z if and only if Re{z} = 0.

I Solution: Beginning with (i) we have

z=z ⇔ x + yj = x − yj. (4.22)

By equating real and imaginary parts we require y = −y, but this is true if and
only if y = Im{z} = 0. Likewise for (ii) we have

z = −z ⇔ x + yj = −x + yj. (4.23)

Thence, we require x = −x, which is true if and only if x = Re{z} = 0. J

Example 4.10 Let z1 , z2 ∈ C be the complex numbers z1 = x1 + y1 j and

z2 = x2 + y2 j; demonstrate that z1 + z2 = z 1 + z 2 .

I Solution: Observe that z1 + z2 = (x1 + x2 ) + (y1 + y2 )j, so we have

z1 + z2 = (x1 + x2 ) + (y1 + y2 )j
= (x1 + x2 ) − (y1 + y2 )j
= (x1 − y1 j) + (x2 − y2 j)
= z1 + z2. (4.24)

The result z1 − z2 = z 1 − z 2 may be proved in an equivalent fashion. J

Example 4.11 Let z1 = x1 + y1 j and z2 = x2 + y2 j; show that z1 z2 = z 1 z 2 .

I Solution: Observe that z1 z2 = (x1 x2 − y1 y2 ) + (x1 y2 + x2 y1 )j, so we have

z1 z2 = (x1 x2 − y1 y2 ) + (x1 y2 + x2 y1 )j
= (x1 x2 − y1 y2 ) − (x1 y2 + x2 y1 )j
= x1 (x2 − y2 j) − y1 y2 − x2 y1 j
= x1 (x2 − y2 j) + y1 y2 j 2 − x2 y1 j
= x1 (x2 − y2 j) − y1 j(x2 − y2 j)
= (x1 − y1 j)(x2 − y2 j)
= z1z2. (4.25)

The related result for quotients z1 /z2 = z 1 /z 2 is proved in Example 4.15. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

142 Section 4 : Complex Numbers

4.2.6 Modulus

Notice that the product zz is a real number given by the sum of the squares
of the real and imaginary parts of z, i.e.,

zz = (x + yj)(x − yj) = x2 − y 2 j 2 = x2 + y 2 ; (4.26)

this suggests a way to quantify the absolute size of z.

Definition 4.4 The modulus of a complex number z = x+yj is denoted

|z| and defined by
√ p
|z| = |x + yj| ≡ zz = x2 + y 2 (4.27)

such that |z|2 = x2 + y 2 . It follows from this definition that

|z| = |z|. (4.28)

The modulus is also sometimes referred to as the absolute value of z.

When interpreted as a vector, the absolute value of a complex number is
the length of its corresponding the vector (see §4.2.2 and figure 4.2).

Example 4.12 Determine the absolute value of the following complex num-
√
bers: (a) z1 = 1 − j ; (b) z2 = − 3 + j; and (c) z3 = 5j.
p
I Solution: We compute |z| = x2 + y 2 for each z = x + yj, thus
p √
(a) |z1 | = 12 + (−1)2 = 2.
q √ √ √
(b) |z2 | = (− 3)2 + 12 = 1 + 3 = 4 = 2.
√
(c) |z3 | = 52 = 5.
√
We consider z1 = 1 − j and z2 = − 3 + j again in Example 4.19. J
Example 4.13 Let z = x + yj; show that |z| ≥ Re{z} and |z| ≥ Im{z}.

I Solution: For the first inequality we have

p √
|z| = x2 + y 2 ≥ x2 = x = Re{z}. (4.29)

The second inequality |z| ≥ Im{z} follows in a corresponding fashion. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 143

4.2.7 Division and quotients

A complex number z1 = x1 + y1 j may divided by another non-zero complex

number z2 = x2 + y2 j to form a quotient, thus

z1 x1 + y 1 j
= for z2 6= 0 (4.30)
z2 x2 + y 2 j

(the quotient is only defined if z2 6= 0). To represent such a quotient in

Cartesian form we need to find real numbers u, v ∈ R such that
z1
= u + vj. (4.31)
z2

To find these numbers we multiply the top and bottom by z 2 = x2 − y2 j, thus

   
z1 x1 + y1 j x2 − y2 j x1 x2 + y 1 y 2 x2 y1 − x1 y2
= × = + j. (4.32)
z2 x2 + y2 j x2 − y2 j x22 + y22 x22 + y22

Rather than memorise this result, it is easier to follow the procedure in context.
Example 4.14 Express the following quotients in Cartesian form:

1 3 + 4j
(a) z1 = ; and (b) z2 = . (4.33)
1 + 2j 1−j

I Solution: (a) We multiply the top and bottom by 1 + 2j = 1 − 2j to give

 
1 − 2j 1 − 2j 1 2
z1 = z1 × = = − j. (4.34)
1 − 2j (1 + 2j)(1 − 2j) 5 5

(b) In this case we multiply the top and bottom by 1 − j = 1 + j to give

 
1+j (3 + 4j)(1 + j) 1 7
z2 = z2 × = = − + j. (4.35)
1+j (1 − j)(1 + j) 2 2

This proceedure for handling quotients is summarised in Method 4.1. J

Example 4.15 Show that z1 /z2 = z 1 /z 2 , where z2 6= 0.

I Solution: Taking the complex conjugate of the quotient we have

       
z1 z1 z 2 z1 z 2 z 1 z2 z 1 z2 z1
= = = = = , (4.36)
z2 z2 z 2 |z2 |2 |z2 |2 z 2 z2 z2

where we used the fact that z1 z 2 = z 1 z 2 = z 1 z2 (see Example 4.11).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

144 Section 4 : Complex Numbers

Method 4.1 The quotient of z1 = x1 + y1 j and z2 = x2 + y2 j 6= 0 is

z1 x1 + y 1 j
= for z2 6= 0. (4.37)
z2 x2 + y 2 j

This quotient may be converted to Cartesian form by multiplying the top

and bottom by z 2 = x2 − y2 j, thus
   
z1 x1 + y 1 j x2 − y 2 j x1 x2 + y 1 y 2 x2 y1 − x1 y2
= × = + j. (4.38)
z2 x2 + y 2 j x2 − y 2 j x22 + y22 x22 + y22

This approach is similar to the process of ‘rationalising the denominator’

when dealing with the quotients of surds.

Example 4.16 Use the result |z| ≥ Re{z} (see Example 4.13) to show that
|z1 | + |z2 | ≥ |z1 + z2 | for any two complex numbers z1 , z2 ∈ C.

I Solution: The inequality |z1 | + |z2 | ≥ |z1 + z2 | is clearly true in the case
that |z1 + z2 | = 0. Otherwise, if |z1 + z2 | =
6 0, then we have

|z1 | + |z2 | |z1 | |z2 | z1 z2

= + = +
|z1 + z2 | |z1 + z2 | |z1 + z2 | z1 + z2 z1 + z2
     
z1 z2 z1 + z2
≥ Re + Re = Re = 1 (4.39)
z1 + z2 z1 + z2 z1 + z2

as required. This result is known as the triangle inequality, and corresponds

to the fact that the sum of the lengths of any two sides of a triangle is greater
than the length of the remaining side (see figure 4.3). J

Figure 4.3: Triangle equality |z1 | + |z2 | ≥ |z1 + z2 |.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 145

4.2.8 Square roots

Square roots of complex numbers may be found by appealing to the idea of

comparing real and imaginary parts. We illustrate the basic procedure below.
Example 4.17 Determine the square roots of the complex number −3 + 4j.

I Solution: Let z = x + yj be a square-root of −3 + 4j, where x, y ∈ R are

real numbers. In this way we have

z 2 = −3 + 4j. (4.40)

Putting z = x + yj into this equation yields

(x + yj)2 = (x2 − y 2 ) + (2xy)j = −3 + 4j. (4.41)

Thus, comparing real and imaginary parts we have

x2 − y 2 = −3 and 2xy = 4. (4.42)

Equations (4.42) are simultaneous equations in x and y; the second equation

yields y = 2/x, and this may be substituted into the first equation to give
 2
2 2 4 2
x − = x2 − 2 = −3, with y= . (4.43)
x x x

By rearranging we obtain a quadratic in x2 , viz.

x4 + 3x2 − 4 = (x2 − 1)(x2 + 4) = 0. (4.44)

It follows that either x2 = 1 or x2 = −4, that is, x = ±1 or x = ±2j. Since

x ∈ R is a real number only the solutions x = ±1 are permitted, i.e.,

2
x = ±1 and y = = ±2. (4.45)
x
Hence, the square roots of −3 + 4j are

z = x + yj = ±(1 + 2j). (4.46)

This solution may be checked by confirming that z 2 = −3 + 4j. J

Example 4.18 Confirm that z 2 = −3 − 4j is satisfied by z = ±(1 − 2j).

I Solution: By direct substitution z 2 = (1 − 2j)2 = −3 − 4j as required. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

146 Section 4 : Complex Numbers

4.3 Polar Form

Let z = x + yj be the complex number depicted by the Argand diagram (x, y)

in figure 4.4, and consider the line joining z to the origin. By Pythagoras’s
theorem this line has length
p
r= x2 + y 2 = |z|, (4.47)

and subtends an anti-clockwise angle θ ∈ R with the positive real-axis (x-axis).

It follows that the x and y coordinates of z can be expressed as

x = r cos θ and y = r sin θ. (4.48)

Thus, z = x + yj may be written in terms of r and θ only as

z = r(cos θ + j sin θ). (4.49)

When z is expressed in this way we say that it has been written in polar form.

Observe that adding integer multiples of 2π to θ does not change the value of
either sin θ or cos θ. Thus, we may also write z in polar form as


z = r cos(θ + 2πk) + j sin(θ + 2πk) , where k∈Z (4.50)

is an integer. Each angle (θ + 2πk) ∈ R is called an argument of z, denoted

arg(z). Of these arguments, there exists a unique argument that lies in the
range (−π, π]; we call this unique argument the principal argument of z.

Figure 4.4: Polar form z = r(cos θ + j sin θ) of a complex number, with r = |z|.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 147

6 0).
Definition 4.5 Let z = x + yj be a non-zero complex number (|z| =
Then an argument arg(z) of z is an angle θ ∈ R that satisfies

x = r cos θ and y = r sin θ. (4.51)

If θ is an argument of z, then the polar form of z is

z = r(cos θ + j sin θ), where r = |z|. (4.52)

If θ is an argument of z, then so to is (θ + 2πk) ∈ R, where k is an integer.

The principal argument of z, denoted

θ = Arg (z), (4.53)

is defined as the unique argument of z that lies in the range −π < θ ≤ π.

√
Example 4.19 Consider the complex numbers z1 = 1 − j and z2 = − 3 + j.

(a) Compute |z1 | and |z2 |.

(b) Determine an argument arg(z1 ) of z1 and an argument arg(z2 ) of z2 .

(c) Express z1 and z2 in polar form.

(d) Determine the principal arguments Arg (z1 ) and Arg (z2 ) of z1 and z2

I Solution: The complex numbers z1 and z2 are sketched on an Argand dia-

gramin figure 4.5. In this way, arguments may be found using trigonometry.

(a) The moduli of z1 and z2 are

p √ q √
|z1 | = 12 + (−1)2 = 2 and |z2 | = (− 3)2 + 12 = 2. (4.54)

(b) Let us denote our arguments of z1 and z2 as θ1 = arg(z1 ) and θ2 = arg(z2 ).

Observe by figure 4.5 that z1 makes a clockwise angle
π
arctan(1/1) = (4.55)
4
with the positive x-axis. Thus, the anti-clockwise angle with the positive x-axis
is − π4 , that is, arg(z1 ) = θ1 = − π4 is an argument of z1 .

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

148 Section 4 : Complex Numbers

√
Figure 4.5: Argand diagrams of the complex numbers z1 = 1 − j and z2 = − 3 + j.

Now observe by figure 4.5 that z2 makes an angle

√ π
arctan(1/ 3) = (4.56)
6
with the negative x-axis. Thus, the anti-clockwise angle that z2 makes with
the positive x-axis is
π 5π
θ2 = π − = , (4.57)
6 6
5π
that is, arg(z2 ) = θ2 = 6
is an argument of z2 .

(c) According to Definition 4.5, and our answers to parts (a) and (b), the
polar forms of z1 and z2 are
√
2 cos(− π4 ) + j sin(− π4 ) , z2 = 2 cos( 5π 5π



z1 = 6
) + j sin( 6
) . (4.58)

These expressions can be checked by evaluating the trigonometric functions.

(d) By the periodicity of the trigonometric functions in equations (4.58), we

may also write z1 and z2 in polar form as
√
2 cos(− π4 + 2πk) + j sin(− π4 + 2πk) ,


z1 = (4.59a)
z2 = 2 cos( 5π 5π


6
+ 2πk) + j sin( 6
+ 2πk) , (4.59b)

where k ∈ Z is an integer. Thus, we may use any of the arguments

arg(z1 ) = − π4 + 2πk and arg(z2 ) = 5π

6
+ 2πk, with k ∈ Z. (4.60)

Of these only θ1 = − π4 and θ2 = 5π 6

lie in the range (−π, π]; hence we have
Arg (z1 ) = θ1 and Arg (z2 ) = θ2 as the principal arguments of z1 and z2 . J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 149

4.4 Exponential Form

Having introduced the basic arithmetic of complex numbers, we now need

to consider functions of complex numbers. In this section we introduce the
complex exponential function, and some of its applications.

4.4.1 Euler’s formula

MacLaurin series expansions may be used to write many functions in terms of

power series. For example, the exponential function, and the sine and cosine
functions may be written as

φ2 φ3 φ4 φ5
eφ = 1 + φ + + + + + ... (4.61a)
2! 3! 4! 5!
φ2 φ4 φ6 φ8 φ10
cos(φ) = 1 − + − + − + ... (4.61b)
2! 4! 6! 8! 10!
φ3 φ5 φ7 φ9 φ11
sin(φ) = φ − + − + − + ... (4.61c)
3! 5! 7! 9! 11!
These series are known to be valid for any real numbers φ ∈ R.
Let us suppose that the expansion for the exponential function eφ works for
imaginary numbers, i.e., for φ = jθ, where θ ∈ R is a real number, then
equation (4.61a) gives

j 2 θ2 j 3 θ3 j 4 θ4 j 5 θ5
ejθ = 1 + jθ + + + + + ... (4.62)
2! 3! 4! 5!
Hence, because j 2 = −1, by collecting real and imaginary parts we may write

θ2 θ4 θ3 θ5
   
jθ
e = 1− + + . . . +j θ − + + ... , (4.63)
2! 4! 3! 5!
| {z } | {z }
cos θ sin θ

where we observed that the bracketed terms on the right-hand-side are the
series expansions for sin θ and cos θ listed in equations (4.61b) and (4.61c).
Equation (4.63) suggests that if θ ∈ R is a real number, then one way to
interpret the meaning of the function ejθ is as

ejθ = cos θ + j sin θ; (4.64)

this equation is known as Euler’s formula.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

150 Section 4 : Complex Numbers

Definition 4.6 Let θ ∈ R be a real number, then the result of raising e

to the power of jθ is defined by Euler’s formula

ejθ = cos θ + j sin θ, for θ ∈ R. (4.65)

By setting θ = π we obtain Euler’s identity ejπ = −1, that is, ejπ +1 = 0.

4.4.2 Complex exponential

Now suppose that z = x + yj is a complex number with x, y ∈ R. Using the

standard rules for combining indices we may write

ez = ex+yj = ex ejy = ex (cos y + j sin y), (4.66)

where we used Euler’s formula ejy = (cos y +j sin y). Equation (4.66) suggests
the following definition for the complex exponential function.

Definition 4.7 Let z = x + yj be a complex number with x, y ∈ R; then

the complex exponential function is defined by

ez ≡ exp(z) ≡ ex (cos y + j sin y). (4.67)

This expression reduces to the real exponential function if Im{z} = y = 0.

Example 4.20 Let z = ρeφj , with ρ, φ ∈ R as real numbers; use Euler’s

formula to show that |z| = ρ.

I Solution: According to Euler’s formula we have

z = ρeφj = ρ(cos φ + j sin φ) = ρ cos φ + jρ sin φ. (4.68)

Hence, by the definition of the modulus of z (see Definition 4.4) we obtain

|z|2 = ρ2 cos2 φ + ρ2 sin2 φ = ρ2 (cos2 φ + sin2 φ) = ρ2 , (4.69)

where we used (sin2 φ + cos2 φ) ≡ 1. It follows that |z| = ρ as required. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 151

4.4.3 Exponential form of a complex number

In Section 4.3 we saw that a complex number z = x + yj may be written in

polar form as
z = r(cos θ + j sin θ), (4.70)

where r = |z| and θ = arg(z) are the modulus and arguments of z. Thus,
using Euler’s formula to write (cos θ + j sin θ) = ejθ we have

z = rejθ ; (4.71)

this expression is known as the exponential form of the complex number z.

Definition 4.8 Let z 6= 0 be a complex number with modulus r = |z| and

argument θ = arg(z); then z is written in exponential form as

z = rejθ . (4.72)

This form greatly simplifies multiplication and division (see §4.4.4).

√
Example 4.21 Express the complex numbers z1 = 1 − j and z2 = − 3 + j
in exponential form.
√
I Solution: In Example 4.19 we found that |z1 | = 2 and |z2 | = 2. We also
found that Arg (z1 ) = − π4 and Arg (z2 ) = 5π
6
. Thus, by Definition 4.8, z1 and
z2 may be written in exponential form as
√
z1 = 2e−(π/4)j and z2 = 2e(5π/6)j . (4.73)

We have used the principal arguments, but any valid arguments will suffice. J

4.4.4 Multiplication and division in exponential form

The exponential form of a complex number provides a powerful means of com-

puting the products and quotients of complex numbers. To illustrate this idea,
consider the complex numbers z, w ∈ C defined by

z = rejθ and w = ρejφ , (4.74)

where r = |z| and θ = Arg (z), and ρ = |w| and φ = Arg (w) (see figure 4.6).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

152 Section 4 : Complex Numbers

(a) (b)

Figure 4.6: Product wz of z and w. (a) Multiplying z by w = ρejφ means scaling by

the factor ρ = |w|, and rotating by the angle Arg (w) = φ. (b) Equivalently, multiplying w
by z = rejθ means scaling by the factor r = |z| and rotating by the angle Arg (z) = θ.

If we multiply z = rejθ by w, then we have

wz = ρejφ × rejθ = ρre(θ+φ)j . (4.75)

Here the modulus of wz is |wz| = ρr, while the argument is arg(wz) = θ + φ.

Geometrically, then, the effect of multiplying z = rejθ by w is to scale by the
factor ρ = |w|, and rotate by the angle Arg (w) = φ, as depicted in figure 4.6.

Now consider the process of dividing z by w, that is,

z rejθ
= jφ = (r/ρ)e(θ−φ)j . (4.76)
w ρe

Here the modulus and argument are |z/w| = (r/ρ) and arg(wz) = θ − φ. In
this case, therefore, the geometric effect of dividing z = rejθ by w is to scale
by the factor 1/ρ = 1/|w|, and to rotate by the angle −φ = −Arg (w).

In summary, therefore, multiplication and division by complex numbers can be

thought of as performing rotations and scalings. Crucially, the exponential
form simplifies the process of multiplication (division) to one of computing the
products (quotients) of moduli, and adding (subtracting) arguments.
√
Example 4.22 In Example 4.21 we found that z1 = 1 − j and z2 = − 3 + j
could be written as z1 = 21/2 e−(π/4)j and z2 = 2e(5π/6)j ; show that z26 /z14 = 16.

I Solution: Using the exponential form of z1 and z2 we obtain

z26 (2e(5π/6)j )6 26 e(5π/6)j×6 26 e5πj

4
= 1/2 −(π/4)j 4
= 2 −(π/4)j×4
= 2 −πj
= 24 e6πj = 16e6πj . (4.77)
z1 (2 e ) 2e 2e

Since e6πj = cos(6π) + j sin(6π) = 1, we thus have z26 /z14 = 16 as required.J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 153

4.4.5 Multiplication by unity

Euler’s formula means that we can write unity (the number 1) as

e2πkj = cos(2πk) + j sin(2πk) = 1. (4.78)

Thus, multiplying a complex number z = reθj by unity corresponds to per-

forming rotations by 2πk (full rotations), viz

z = rejθ = 1 × rejθ = e2πkj × rejθ = rej(θ+2πk) . (4.79)

This property is another way of indicating that if θ is an argument of z, then so

too is (θ + 2πk); it also leads us to the idea of the general form for a complex
number in exponential form.

Definition 4.9 Let z 6= 0 be a complex number with modulus r = |z| and

argument θ = arg(z); then the general exponential form for z = rejθ is

z = rej(θ+2πk) , where k∈Z (4.80)

is an integer; this is because if θ is an argument of z, then so too is θ +2πk.

4.5 Computing general roots

We considered the problem of computing the square-root of a complex number

in §4.2.8. In this section we consider the more general problem of computing
the nth root of a complex number, where n ∈ N is a positive integer. We begin
by considering the specific problem of computing the cubed roots of unity.

4.5.1 Cubed roots of unity

The general exponential form for unity (z = 1) is

z = 1 = e2πkj , where k∈Z (4.81)

is an integer. Thus, denoting a cubed root of unity as w, we have

w3 = z = e2πkj , that is, w = z 1/3 = e(2πk/3)j . (4.82)

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

154 Section 4 : Complex Numbers

Since k ∈ Z can take any integer value, the numbers given by

(taking k = 0) w = w0 = e(0π/3)j = +1, (4.83a)

√
1 3
(taking k = 1) w = w1 = e(2π/3)j = − + j, (4.83b)
2 √2
1 3
(taking k = 2) w = w2 = e(4π/3)j = − − j, (4.83c)
2 2
etc., are all cubed roots of z = 1, where we have denoted the root w corre-
sponding to a particular value of k as

wk = e(2πk/3)j . (4.84)

The cubed roots given in equations (4.83) are sketched in figure 4.7; observe
that they all lie on a circle of radius |wk | = 1 (the unit circle), and are dis-
tributed evenly by the angle 2π/3. The reason for this even distribution follows
from the relationship between two consecutive roots wk and wk+1 , that is,

wk+1 = e(2[k+1]π/3)j = e(2kπ/3)j e(2π/3)j = wk e(2π/3)j . (4.85)

Thus, each root wk+1 is obtained by rotating the previous root wk through
by an angle of 2π/3. After three such rotations, therefore, we return to our
original root, i.e.,
wk+3 = wk e(2π)j = wk . (4.86)

It follows, therefore, that only three of the cubed roots are distinct. In
particular, taking k equal to any value other than k = 0, 1, 2 will simply repeat
the roots already listed in equation (4.83). This aspect is an instance of the
more general rule that a complex number has only n distinct nth roots.

Figure 4.7: Cubed roots of unity.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 155

4.5.2 Roots of arbitrary complex numbers

The procedure for computing nth roots of an arbitrary complex number z

generalises the approach taken above. Starting with the exponential form for
z we have
z = rej(θ+2πk) where k∈Z (4.87)

is an integer, r = |z| is the modulus of z, and θ is an argument of z. Thus,

taking the nth root of equation (4.87) we find that the number w defined by

w = z 1/n = r1/n ej(θ+2πk)/n (4.88)

is the nth root of z. Now k ∈ Z can take any integer value, so the numbers

(taking k = 0) w = w0 = r1/n ejθ/n , (4.89a)

1/n j(θ+2π)/n
(taking k = 1) w = w1 = r e , (4.89b)
(taking k = 2) w = w2 = r1/n ej(θ+4π)/n , (4.89c)

etc., are all nth roots of z, where we have written the root w corresponding
to a particular value of k as

wk = z 1/n = r1/n ej(θ+2πk)/n with k ∈ Z. (4.90)

Although it would seem that z has an indefinite number of nth roots, cor-
responding to the indefinite number of integers k, not all of these roots are
distinct. To see this, observe that

wn+k = r1/n ej(θ+2πn+2πk)/n (4.91)

= r1/n ej(θ+2πk)/n ej(2π) (4.92)
= r1/n ej(θ+2πk)/n = wk (since ej(2π) = 1), (4.93)

i.e., the root given by wn+k is identical to the root given by wk . Thus,
listing the roots wk in sequence, we find that only n of them are distinct, viz.

z 1/n = w0 , w1 , w2 , . . . , wn−2 , wn−1 . (4.94)

Other roots are simply repetitions of the roots listed above, for example,
wn+2 = w2 , wn+1 = w1 , wn = w0 , and w−2 = wn−2 , w−1 = wn−1 . In
summary, therefore, a complex number has only n distinct nth roots .

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

156 Section 4 : Complex Numbers

Result 4.2 Let z ∈ C a complex number with exponential form

z = rejθ . (4.95)

Then z has n ∈ N distinct nth roots

z 1/n = w0 , w1 , . . . , wk , . . . , wn−2 , wn−1 , (4.96)

where
wk = r1/n ej(θ+2πk)/n , with k ∈ Z. (4.97)

These roots are distributed evenly about a circle of radius |wk | = r1/n ,
where the regular angular separation between each of the roots is 2π/n.

Example 4.23 Determine the five distinct fifth roots of z = −32.

I Solution: Since eπj = −1 we have that z = −32 = 32eπj . Thus, the general
exponential form for z is

z = 32ej(π+2πk) , where k∈Z (4.98)

is an integer. In this way, the fifth roots of z are given by

wk = z 1/5 = 321/5 ej(π+2πk)/5 = 2ej(π+2πk)/5 . (4.99)

Hence, the five distinct solutions to w5 = −32 are

w0 = 2e(π/5)j , (4.100a)
w1 = 2e(3π/5)j , (4.100b)
w2 = 2e(5π/5)j = −2, (4.100c)
w3 = 2e(7π/5)j , (4.100d)
w4 = 2e(9π/5)j . (4.100e)

(there is no need to convert all these numbers into Cartesian form). These
roots are sketched on an Argand diagram in figure 4.8. Notice that the roots
are distributed evenly about the circle with radius |wk | = 2, where the regular
angular separation between each of the roots is 2π/5. J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 157

Figure 4.8: Fifth roots of z = −32.

It is important to emphasise that computing roots is equivalent to solving

simple polynomial equations. For example, computing the fifth roots of −32
is equivalent to solving the quintic equation

w5 + 32 = 0. (4.101)

We now extend these ideas to more general polynomial equations.

4.6 Solving Polynomial Equations

Consider the general form of a second-order polynomial

P (z) = az 2 + bz + c (4.102)

where the coefficients a, b, c ∈ C are complex numbers, with a 6= 0. A root of

this polynomial is some number z that satisfies

P (z) = az 2 + bz + c = 0, (4.103)

that is, the roots of P (z) are solutions to a general quadratic equation. By
completing the square, therefore, we find that the second-order polynomial has
two roots, z = z+ and z = z− given by

−b ± (b2 − 4ac)1/2
z = z± = , (4.104)
2a
where evaluating these roots may require computing the square-roots of a
complex number (b2 − 4ac)1/2 . More generally, any nth order polynomial has
n roots, a property known as the fundamental theorem of algebra.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

158 Section 4 : Complex Numbers

The quadratic equation is special because its general solution can be written
down in a concise form. Higher order polynomials are usually more difficult
to solve unless either: (i) factorisation is straightforward; or (ii) there exists a
simple reduction to a quadratic. In this section we illustrate how to determine
the n solutions to a polynomial of order n in these simple cases.
Example 4.24 Solve the quadratic equation w2 + 4w + 7 − 4j = 0.

I Solution: By completing the square the quadratic equation may be written

(w + 2)2 + 3 − 4j = 0, (4.105)

or equivalently (by denoting z 2 = −3 + 4j)

(w + 2)2 = z 2 , i.e., w = −2 ± z, with z 2 = −3 + 4j. (4.106)

Now, in Example 4.17 we found that if z 2 = −3 + 4j, then z = ±(1 + 2j). It

follows that the solutions (roots) of the quadratic are

w = w± = −2 ± z = −2 ± (1 + 2j). (4.107)

that is, the quadratic has roots

w+ = −1 + 2j or w− = −3 − 2j. (4.108)

Our solution may then be checked by confirming that the factorisation

(w − w+ )(w − w− ) = 0 (4.109)

is equivalent to our original equation w2 + 4w + 7 − 4j = 0. J

Example 4.25 Solve the quartic equation z 4 + 6z 2 + 25 = 0.

I Solution: This equation is effectively a quadratic in z 2 . Thus, completing

the square we have

(z 2 + 3)2 + 16 = 0 that is (z 2 + 3)2 = −16 = (4j)2 . (4.110)

Hence, taking the square root we find z 2 + 3 = ±4j, i.e.,

z 2 = −3 ± 4j. (4.111)

The solutions must therefore satisfy either z 2 = −3 + 4j or z 2 = −3 − 4j.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 159

In our earlier examples we found:

(Example 4.17) z 2 = −3 + 4j ⇔ z = ±(1 + 2j) (4.112a)

(Example 4.18) z 2 = −3 − 4j ⇔ z = ±(1 − 2j). (4.112b)

It therefore follows that the four solutions to the quartic are

z = 1 + 2j, z = 1 − 2j, and z = −1 + 2j, z = −1 − 2j. (4.113)

Observe here that our quartic equation has four roots (as expected) in the form
of two complex conjugate pairs.

Example 4.26 Let P (z) = az 2 + bz + c be a quadratic polynomial with real

coefficients a, b, c ∈ R satisfying 4ac > b2 . Show that the two roots of P (z)
are a complex conjugate pair.

I Solution: According to equation (4.104) the roots of P (z) are given by

−b ± (b2 − 4ac)1/2
z = z± = . (4.114)
2a
Now if 4ac > b2 , then (4ac − b2 ) is a positive real number. Let us signify this
property by defining
√
β= 4ac − b2 , where β∈R (4.115)

is the positive square root of (4ac − b2 ). In this way

(b2 − 4ac) = −β 2 = β 2 j 2 , (4.116)

and the roots of P (z) may be written as

−b ± (β 2 j 2 )1/2 b β
z = z± = = − ± j. (4.117)
2a 2a 2a
Hence the roots are complex conjugate pairs with z− = z + . J

Polynomials with real coefficients

Notice in our examples of polynomials P (z) with real coefficients that if w is

a root of P (z), then so too is its complex conjugate w. This property is a
consequence of the following result (Result 4.3).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

160 Section 4 : Complex Numbers

Result 4.3 Let P (z) be polynomial of order n with real coefficients, i.e.,
n
X
P (z) = ar z r = an z 2 + an−1 z n−1 + · · · + a1 z + a0 , (4.118)
r=0

where ar ∈ R, with r = 0, 1, . . . , n. If w is a root of P (z), i.e., P (w) = 0,

then so too is its complex conjugate w, i.e., P (w) = 0. Thus, the roots of
polynomial with real coefficients come in complex conjugate pairs.

This result may be proved using the properties of complex conjugates. In

particular, if we assert that w is a root of P (z), i.e., P (w) = 0, then we have

n
X n
X n
X n
X
r
P (w) = ar w = ar wr = ar wr = ar wr = P (w) = 0, (4.119)
r=0 r=0 r=0 r=0

that is, w is also a root with P (w) = 0.

Example 4.27 One of the solutions to P (z) = z 3 − 31z 2 + 304z − 274 = 0
is z = 15 + 7j, write down another solution.

I Solution: Notice that P (z) = z 3 − 31z 2 + 304z − 274 = 0 is a polynomial

with real coefficients. Hence, it follows by Result 4.3 that if z = 15 + 7j is a
root of P (z), then so too is its complex conjugate z = 15 − 7j.
[Notice that since P (z) is a cubic polynomial, it has a total of three roots. In
this instance the remaining root is z = 1.] J

4.7 Complex Functions

A complex function is a function f : C → C that maps a complex number

to another complex number. In practical terms we define such functions to
be consistent with the properties of real functions. For example, in §4.4.2 we
defined the complex exponential function as

ez = ex (cos y + j sin y), (4.120)

where f (x) = ex is the usual exponential function for x ∈ R. Thus, if z is a

real number (Im{z} = y = 0), then ez returns the same real value as f (x).

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 161

In this section we continue the process of defining complex functions that

extend the definitions of their corresponding real functions. We begin with the
complex trigonometric functions.

4.7.1 Complex trigonometric functions

Euler’s formula (see Definition 4.6) states that if θ ∈ R is a real number, then

ejθ = cos θ + j sin θ. (4.121)

Thus, because cos(−θ) = cos(θ) and sin(−θ) = − sin(θ), we have

e−jθ = cos θ − j sin θ. (4.122)

It follows from these equations that the real sine and cosine functions can
be written
1 +jθ 1 +jθ
e + e−jθ e − e−jθ .



cos θ = and sin θ = (4.123)
2 2j

These expressions motivate the following general definitions for the complex
sine and cosine functions.

Definition 4.10 For all z ∈ C the sine and cosine functions are defined
1 +jz 1 +jz
e + e−jz e − e−jz . (4.124)



cos z = and sin z =
2 2j

If Im{z} = y = 0, then these expressions reduce to the real sine and cosine
functions of equation (4.123).

Other complex trigonometric functions follow from this definition, for example:

Definition 4.11 For all z ∈ C the tangent function is defined

sin z
tan z ≡ , (4.125)
cos z
where the complex sine and cosine functions are given in Definition 4.10.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

162 Section 4 : Complex Numbers

Example 4.28 Express sin(j) and cos(π + j) in Cartesian form.

I Solution: By the definitions of the complex sine and cosine functions (Defi-
nition 4.10) we have

1  +j 2 2
 1 −1 j
e − e−j = e − e1 = 2 e−1 − e1



sin(j) =
2j 2j 2j
j 1
= − e−1 − e = e − e−1 j.



(4.126)
2 2
Similarly,

1 +(π+j)j
1  πj j 2 2

cos(π + j) = e + e−(π+j)j = e e + e−πj e−j
2 2
1 πj −1 1
e e + e−πj e = − e + e−1 ,



= (4.127)
2 2
where we used the fact that eπj = e−πj = −1. J

4.7.2 De Moivre’s Theorem and Trigonometric Identities

According to Euler’s formula ejθ = (cos θ + j sin θ), and the laws for indices,
if θ ∈ R is a real number, and n ∈ Z is an integer, then

n
(cos θ + j sin θ)n = ejθ = ejnθ = (cos(nθ) + j sin(nθ)) . (4.128)

This result is known as de Moivre’s Theorem:

Theorem 4.1 If θ ∈ R is a real number, and n ∈ Z is an integer, then

(cos θ + j sin θ)n = (cos(nθ) + j sin(nθ)) . (4.129)

This result is known as de Moivre’s Theorem (or De Moivre’s formula).

The complex trigonometric functions may be used in conjunction with de

Moivre’s theorem to find trigonometric identities, especially identities for si-
nusoidal functions raised to powers. Given some integer n ∈ Z, the basic
approach is to compare the real and imaginary parts of either side of equation
(4.1). We now illustrate this idea by example.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 163

Example 4.29 Use de Moivre’s theorem to express sin(2θ), cos(2θ), and

tan(2θ) in terms of sin θ, cos θ, and tan θ only.

I Solution: According to de Moivre’s theorem we have that

(cos(2θ) + j sin(2θ)) = (cos θ + j sin θ)2

= cos2 θ + 2j cos θ sin θ + j 2 sin2 θ
= (cos2 θ − sin2 θ) + 2j cos θ sin θ. (4.130)

Hence, by comparing real and imaginary parts we obtain

cos(2θ) = cos2 θ − sin2 θ and sin(2θ) = 2 cos θ sin θ (4.131)

as required. The equation for tan(2θ) then follows by definition, i.e.,

sin(2θ) 2 cos θ sin θ 2 tan θ

tan(2θ) ≡ = 2 = , (4.132)
cos(2θ) 2
cos θ − sin θ 1 − tan2 θ

where we divided the top and bottom of the penultimate fraction by cos2 θ.J

4.7.3 Logarithms

For real numbers x, y ∈ R the natural logarithm (loge ) is defined as the inverse
function of y = ex , that is, x = loge y. Here we seek a comparable definition
for the complex logarithm (log) for complex numbers z, w ∈ C; in particular if
z = ew , then we seek a complex logarithm function such that w = log z.
To this end let us suppose that w = u + vj, where u, v ∈ R are real numbers,
in which case our complex logarithm function must satisfy

w = log z = u + vj, with z = ew = e(u+vj) = eu ejv . (4.133)

Let us write z in the general exponential form z = re(θ+2πk)j , where r = |z|,

θ = Arg z, and Z is an integer. Then equation (4.133) entails that

z = re(θ+2πk)j = eu evj ⇔ r = eu and v = θ + 2πk, (4.134)

i.e., u = loge (r) = loge |z|, and v = Arg z + 2πk; hence,

log z = u + vj = loge |z| + (Arg z + 2πk)j. (4.135)

This equation serves as a suitable definition for the complex logarithm.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

164 Section 4 : Complex Numbers

Definition 4.12 Let z ∈ C be a non-zero complex number, then we define

the complex logarithm of z as the function

log z = loge |z| + (Arg z + 2πk)j, (4.136)

where loge |z| is the real natural logarithm of |z|, and k ∈ Z. Since any
integer k ∈ Z may be used, this logarithm is not unique. We thus define

Log z = loge |z| + jArg z. (4.137)

as the unique principal logarithm of z (denoted Log z).

Example 4.30 Let z1 = 1 − j, evaluate Log z1 in Cartesian form.

√
I Solution: By Example 4.19 we know that z1 has modulus |z1 | = 2 = 21/2 ,
and Arg (z) = − π4 . Hence, by Definition 4.12 we have

π 1
Log z1 = Log (1 − j) = loge (21/2 ) − j = (2 loge 2 − πj) (4.138)
4 4
as the principal logarithm of z1 . J

4.7.4 Powers

Recall that if x ∈ R is a positive real number, then for real powers a ∈ R we

have that
xa = ea loge x , (4.139)

where loge x is the real natural logarithm of x. This property prompts us to

make the following definition for complex powers.

Definition 4.13 Let z ∈ C be a non zero complex number, then raising

z to a complex power α is defined by the principal αth power

z α = eαLog z . (4.140)

This definition is consistent with our discussion of integer powers and roots.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

Section 4 : Complex Numbers 165

Example 4.31 Evaluate (1 − j)j in Cartesian form.

√
I Solution: In Example 4.30 we found that Log (1 − j) = loge 2 − π4 j. Thus,
by definition 4.13 we have
√ π π
√
(1 − j)j = ej(loge 2− 4 j) = e 4 ej(loge 2) (4.141)
π
 √ √ 
= e 4 cos(loge 2) + j sin(loge 2) , (4.142)

where the final line is obtained using Euler’s formula. J

4.8 Oscillatory Systems

Complex numbers are particularly useful for describing oscillatory systems,

that is, systems in which the values of the variables oscillate in time t.

As an example, consider the circuit formed by connecting an inductor of induc-

tance L in series with a capacitor of capacitance C (see figure 4.9). It may be
shown that the voltages across the inductor and capacitor are respectively

dI Q(t) dQ
VL (t) = −L and VC (t) = , where I(t) = (4.143)
dt C dt
is the current in the circuit, and Q(t) is the charge on the capacitor at time t.
Since these voltages must be equal, we have

dI Q(t) dI 1
−L = , that is + Q(t) = 0. (4.144)
dt C dt LC
Thus, differentiating this equation with respect to time we obtain

d2 I(t) 1
+ ω 2 I(t) = 0, where ω2 = , (4.145)
dt2 LC
and we used the fact that the current is I(t) = dQ/dt.

Figure 4.9: Schematic of an LC-circuit, i.e., an inductor L in series with a capacitor C.

Engineering Mathematics : ELE00038C J.J.B. 2023-2024

166 Section 4 : Complex Numbers

Equation (4.145) is known as a differential equation because it involves a

derivative. A solution to this equation is some function I(t) for which the left-
hand-side is equal to the right-hand-side. It may be shown that the general
solution to may be written as

I(t) = A cos(ωt + φ), (4.146)

where A and φ are constants known as the amplitude (A) and phase (φ).
The solutions therefore oscillate sinusoidally in time. In particular, notice that

2π
I(t + T ) = I(t) where T = ; (4.147)
ω
i.e., the system oscillates with time period T , and frequency ω = 2π/T .

The reason that complex numbers are useful for describing such oscillatory
solutions may be understood as follows. Observe that Euler’s formula allows
us to write I(t) as

I(t) = Re {A (cos(ωt + φ) + j sin(ωt + φ))} = Re{Aej(ωt+φ) }, (4.148)

that is,
I(t) = Re{IA ejωt }, where IA = Aejφ (4.149)

is a complex number. In this way information about the amplitude and phase
is contained within IA = Aejφ , while ejωt describes the frequency. Thus, the
complex form allows us to make a clear and compact distinction between the
constant part of the solution (IA = Aejφ ), and the time dependent part (ejωt ).

Example 4.32 Show by direct substitution that equation (4.145) is satisfied

by I(t) = IA ejωt , where IA ∈ C is a constant.

I Solution: The first and second derivatives of I(t) = IA ejωt are

dI(t) d2 I(t)
= jωI(t) and = j 2 ω 2 I(t) = −ω 2 I(t) (4.150)
dt dt2
respectively. Thus,

d2 I(t)
2
+ ω 2 I(t) = −ω 2 I(t) + ω 2 I(t) = 0 (4.151)
dt
as required by equation (4.145). J

Engineering Mathematics : ELE00038C J.J.B. 2023-2024