R EVISION N OTES

I MPERIAL C OLLEGE L ONDON

D EPARTMENT OF C OMPUTING

Complex Numbers

Instructors:
Marc Deisenroth

Version: August 1, 2017

Contents

1 Complex Numbers 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Imaginary Number . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.3 Complex Numbers as Elements of R2 . . . . . . . . . . . . . . 3
1.1.4 Closure under Arithmetic Operators . . . . . . . . . . . . . . 3
1.2 Representations of Complex Numbers . . . . . . . . . . . . . . . . . . 4
1.2.1 Cartesian Coordinates . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Polar Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.3 Euler Representation . . . . . . . . . . . . . . . . . . . . . . . 5
1.2.4 Transformation between Polar and Cartesian Coordinates . . 6
1.2.5 Geometric Interpretation of the Product of Complex Numbers 8
1.2.6 Powers of Complex Numbers . . . . . . . . . . . . . . . . . . 8
1.3 Complex Conjugate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.1 Absolute Value of a Complex Number . . . . . . . . . . . . . . 10
1.3.2 Inverse and Division . . . . . . . . . . . . . . . . . . . . . . . 10
1.4 De Moivre’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4.1 Integer Extension to De Moivre’s Theorem . . . . . . . . . . . 12
1.4.2 Rational Extension to De Moivre’s Theorem . . . . . . . . . . 12
1.5 Triangle Inequality for Complex Numbers . . . . . . . . . . . . . . . 13
1.6 Fundamental Theorem of Algebra . . . . . . . . . . . . . . . . . . . . 14
1.6.1 nth Roots of Unity . . . . . . . . . . . . . . . . . . . . . . . . 14
1.6.2 Solution of zn = a + ib . . . . . . . . . . . . . . . . . . . . . . 15
1.7 Complex Sequences and Series* . . . . . . . . . . . . . . . . . . . . . 16
1.7.1 Limits of a Complex Sequence . . . . . . . . . . . . . . . . . . 16
1.8 Complex Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.8.1 A Generalized Euler Formula . . . . . . . . . . . . . . . . . . 18
1.9 Applications of Complex Numbers . . . . . . . . . . . . . . . . . . . . 18
1.9.1 Trigonometric Multiple Angle Formulae . . . . . . . . . . . . 18
1.9.2 Summation of Series . . . . . . . . . . . . . . . . . . . . . . . 19
1.9.3 Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2
Chapter 1

Complex Numbers

1.1 Introduction
We can see need for complex numbers by looking at the shortcomings of all the
simpler (more obvious) number systems that preceded them. In each case the next
number system in some sense fixes a perceived problem or omission with the previ-
ous one:

N Natural numbers, for counting, not closed under subtraction

Z Integers, the natural numbers with 0 and negative numbers, not closed under
division

Q Rational numbers, closed under arithmetic operations but cannot represent the
solution of all non-linear equations, e.g., x2 = 2

R Real numbers, solutions to some quadratic equations with real roots and some
higher-order equations, but not all, e.g., x2 + 1 = 0

C Complex numbers, we require these to represent all the roots of all polynomial
equations.1

Another important use of complex numbers is that often a real problem can be solved
by mapping it into complex space, deriving a solution, and mapping back again: a
direct solution may not be possible or would be much harder to derive in real space,
e.g., finding solutions to integration or summation problems, such as
Z x n
X
aθ
I= e cos bθ dθ or S= ak cos kθ . (1.1)
0 k=0

1.1.1 Applications
Complex numbers are important in many areas. Here are some:
1 Complex numbers form an algebraically closed field, where any polynomial equation has a root.

1
1.1. Introduction Chapter 1. Complex Numbers

• Signal analysis (e.g., Fourier transformation to analyze varying voltages and

currents)

• Control theory (e.g., Laplace transformation from time to frequency domain)

• Quantum mechanics is founded on complex numbers (see Schrödinger equa-

tion and Heisenberg’s matrix mechanics)

• Cryptography (e.g., finding prime numbers).

• Machine learning: Using a pair of uniformly distributed random numbers (x, y),
we can generate random numbers in polar form (r cos(θ), r sin(θ)). This can
lead to efficient sampling methods like the Box-Muller transform (Box and
Muller, 1958).2 The variant of the Box-Muller transform using complex num-
bers was proposed by Knop (1969).

• (Tele)communication: digital coding modulations

1.1.2 Imaginary Number

An entity we cannot describe using real numbers are the roots to the equation

x2 + 1 = 0, (1.2)

which we will call i and define as

√
i := −1. (1.3)

There is no way√of squeezing this into R, it cannot be compared with a real number
(in contrast to 2 or π, which we can compare with rationals and get arbitrarily
accurate approximations in the rationals). We call i the imaginary number/unit,
orthogonal to the reals.

Properties From the definition of i in (1.3) we get a number of properties for i.

1. i 2 = −1, i 3 = i 2 i = −i, i 4 = (i 2 )2 = (−1)2 = 1 and so on

2. In general i 2n = (i 2 )n = (−1)n , i 2n+1 = i 2n i = (−1)n i for all n ∈ N

1 i
3. i −1 = i = i2
= −i

1 1
4. In general i −2n = i 2n
= (−1)n
= (−1)n , i −(2n+1) = i −2n i −1 = (−1)n+1 i for all n ∈ N

5. i 0 = 1
2 This is a pseudo-random number sampling method, e.g., for generating pairs of independent,
standard, normally distributed (zero mean, unit variance) random numbers, given a source of uni-
formly distributed random numbers.

2
Chapter 1. Complex Numbers 1.1. Introduction

Im

z = (x, y) = x + iy
iy

1 x Re

Figure 1.1: Complex plane (Argand diagram). A complex number can be represented
in a two-dimensional Cartesian coordinate system with coordinates x and y. x is the real
part and y is the imaginary part of a complex number z = x + iy.

1.1.3 Complex Numbers as Elements of R2

It is convenient (and correct3 ) to consider complex numbers

C := {a + ib : a, b ∈ R, i 2 = −1} (1.4)

as the set of tuples (a, b) ∈ R2 with the following definition of addition and multipli-
cation:

(a, b) + (c, d) = (a + c, b + d) , (1.5)

(a, b) · (c, d) = (ac − bd, ad + bc) . (1.6)

In this context, the element i := (0, 1) is the imaginary number/unit. With the
complex multiplication defined in (1.6), we immediately obtain

i 2 = (0, 1)2 = (0, 1)(0, 1) = −1, (1.7)

which allows us to factorize the polynomial z2 + 1 fully into (z − i)(z + i).

Since elements of R2 can be drawn in a plane, we can do the same with complex
numbers z ∈ C. The plane is called complex plane or Argand diagram, see Fig-
ure 1.1.
The Argand diagram allows us to visualize addition and multiplication, which are
defined in (1.5)–(1.6).

1.1.4 Closure under Arithmetic Operators

Closing R ∪ {i} under the arithmetic operators +, · as defined in (1.5)–(1.6) gives the
complex numbers, C. To be more specific, if z1 , z2 ∈ C, then z1 + z2 ∈ C, z1 − z2 ∈ C,
z1 · z2 ∈ C and z1 /z2 ∈ C.
3 There exists a bijective linear mapping (isomorphism) between C and R2 . We will briefly discuss
this in the Linear Algebra part of the course.

3
1.2. Representations of Complex Numbers Chapter 1. Complex Numbers

Im

z2
Re

z1 + z2

z1

Figure 1.2: Visualization of complex addition. As known from geometry, we simply add
the two vectors representing complex numbers.

1.2 Representations of Complex Numbers

In the following, we will discuss three important representations of complex num-
bers.

1.2.1 Cartesian Coordinates

Every element z ∈ C can be decomposed into
(x, y) = (x, 0) + (0, y) = (x, 0) + (0, 1)(y, 0) = (x, 0) +i (y, 0) = x + iy. (1.8)
|{z} |{z}
∈R ∈R

Therefore, every z = x + iy ∈ C has a coordinate representation (x, y), where x

is called the real part and y is called the imaginary part of z, and we write x =
<(z), y = =(z), respectively. z = x + iy is the point (x, y) in the xy-plane (com-
plex plane), which is uniquely determined by its Cartesian coordinates (x, y). An
illustration is given in Figure 1.1.

1.2.2 Polar Coordinates

Equivalently, (x, y) can be represented by polar coordinates, r, φ, where r is the
distance of z from the origin 0, and φ is the angle between the (positive) x-axis and
the direction 0z~ . Then,

z = r(cos φ + i sin φ), r ≥ 0, 0 ≤ φ < 2π (1.9)

uniquely determines z ∈ C. The polar coordinates of z are then
q
r = |z| = x2 + y 2 , (1.10)
φ = Arg z , (1.11)
~ (the distance of z from the origin) and φ is the argument
where r is the length of 0z
of z.

4
Chapter 1. Complex Numbers 1.2. Representations of Complex Numbers

z = (x, y) = r(cos φ + i sin φ)

iy

i r
φ

Figure 1.3: Polar coordinates.

Im

r sin φ

φ r
r cos φ Re

Figure 1.4: Euler representation. In the Euler representation, a complex number z =

r exp(iφ) “lives” on a circle with radius r around the origin. Therefore, r exp(iφ) =
r(cos φ + i sin φ).

1.2.3 Euler Representation

The third representation of complex numbers is the Euler representation

z = r exp(iφ) (1.12)

where r and φ are the polar coordinates. We already know that z = r(cos φ + i sin φ),
i.e., it must also hold that r exp(iφ) = r(cos φ+i sin φ). This can be proved by looking
at the power series expansions of exp, sin, and cos:
∞
X (iφ)k (iφ)2 (iφ)3 (iφ)4 (iφ)5
exp(iφ) = = 1 + iφ + + + + + ··· (1.13)
k! 2! 3! 4! 5!
k=0
φ2 iφ3 φ4 iφ5
= 1 + iφ − − + + ∓ ··· (1.14)
2! 3! 4! 5!

5
1.2. Representations of Complex Numbers Chapter 1. Complex Numbers

φ2 φ4 φ3 φ5
! !
= 1− + ∓ ··· + i φ − + ∓ ··· (1.15)
2! 4! 3! 5!
∞ ∞
X (−1)k φ2k X (−1)k φ2k+1
= +i = cos φ + i sin φ . (1.16)
(2k)! (2k + 1)!
k=0 k=0

Therefore, z = exp(iφ) is a complex number, which lives on the unit circle (|z| = 1)
and traces out the unit circle in the complex plane as φ ranges through the real
numbers.

1.2.4 Transformation between Polar and Cartesian Coordinates

Cartesian coordinates Polar coordinates

x, y r, φ

p
x = r cos φ r r = x2 + y 2
y
y = r sin φ tan φ = xy + quadrant

z = x + iy = r(cos φ + i sin φ)
Figure 1.5: Transformation between Cartesian and polar coordinate representations of
complex numbers.

Figure 1.5 summarizes the transformation between Cartesian and polar coordinate
representations of complex numbers z. We have to pay some attention when com-
puting Arg (z) when transforming Cartesian coordinates into polar coordinates.

Example: Transformation from Polar to Cartesian Coordinates

Transform the polar representation z = (r, φ) = (2, 2π 3 ) into Cartesian coordinates
(x, y).
It is always useful to draw the complex number. Figure 1.6(a) shows the setting. We
are interested in the blue dots. With x = r cos φ and y = r sin φ, we obtain
x = r cos( 23 π) = −1 (1.17)
√
y = r sin( 32 π) = 3 . (1.18)
√
Therefore, z = −1 + i 3.

Example: Transformation from Cartesian to Polar Coordinates

Getting the Cartesian coordinates from polar coordinates is straightforward. The
transformation from Cartesian to polar coordinates is somewhat more difficult be-
cause of the argument φ. The reason is that tan has a period of π, which means

6
Chapter 1. Complex Numbers 1.2. Representations of Complex Numbers

y
z r=2 Im

φ= 2π
3
Re

x
z = 2 − 2i
2π
(a) (r, φ) = (2, 3 ) (b) (x, y) = (2, −2)

z = −1 + i Im Im

1 1

Re Re
z= − 32 i

(c) (x, y) = (−1, 1) (d) (x, y) = (0, − 32 )

Figure 1.6: Coordinate transformations

that y/x has two possible angles, which differ by π, see Figure 1.7. By looking at the
quadrant in which the complex number z lives we can resolve this ambiguity. Let us
have a look at some examples:
√ √
1. z = 2 − 2i. We immediately obtain r = 22 + 22 = 2 2. For the argument,
we obtain tan φ = − 22 = −1. Therefore, φ ∈ { 34 π, 74 π}. We identify the correct
argument by plotting the complex number and identifying the quadrant. Fig-
ure 1.6(b) shows that z lies in the fourth quadrant. Therefore, φ = 74 π.
2. z = −1 + i.
√ √
r = 1+1 = 2 (1.19)
−1
tan φ = = −1 ⇒ φ ∈ { 34 π, 74 π} . (1.20)
1
Figure 1.6(c) shows that z lies in the second quadrant. Therefore, φ = 43 π.

3. z = − 23 i.
3
r= 2 (1.21)
− 32 π 3
tan φ = ⇒ φ ∈ { , π} (1.22)
0 2 2
Figure 1.6(d) shows that z is between the third and fourth quadrant (and not
between the first and second). Therefore, φ = 23 π

7
1.2. Representations of Complex Numbers Chapter 1. Complex Numbers

φ1 φ2 = φ1 + π

Figure 1.7: Tangens. Since the tangens possesses a period of π, there are two solutions
for the argument 0 ≤ φ < 2π of a complex number, which differ by π.

1.2.5 Geometric Interpretation of the Product of Complex Num-

bers
Let us now use the polar coordinate representation of complex numbers to ge-
ometrically interpret the product z = z1 z2 of two complex numbers z1 , z2 . For
z1 = r1 (cos θ1 + i sin θ1 ) and z2 = r2 (cos θ2 + i sin θ2 ) we obtain
z1 z2 = r1 r2 (cos θ1 cos θ2 − sin θ1 sin θ2 + i(sin θ1 cos θ2 + cos θ1 sin θ2 ))
= r1 r2 (cos(θ1 + θ2 ) + i sin(θ1 + θ2 )) . (1.23)
1. The length r = |z| = |z1 | |z2 | is the product of the lengths of z1 and z2 .
2. The argument of z is the sum of the arguments of z1 and z2 .
This means that when we multiply two complex numbers z1 , z2 , the corresponding
distances r1 and r2 are multiplied while the corresponding arguments φ1 , φ2 are
summed up. This means, we are now ready to visualize complex multiplication, see
Figure 1.8. Overall, multiplying z1 with z2 performs two (linear) transformations on
z1 : a scaling by r2 and a rotation by φ2 . Similarly, the transformations acting on z2
are a scaling by r1 and a rotation by φ1 .

1.2.6 Powers of Complex Numbers

We will encounter situations where we need to compute powers of complex numbers
of the form zn . For this, we can use some advantages of some representations of
complex numbers. For instance, if we consider the representation using Cartesian
coordinates computing zn = (x + iy)n for large n will be rather laborious. However,
the Euler representation makes our lives a bit easier since
zn = (r exp(iφ))n = r n exp(inφ) (1.24)

8
Chapter 1. Complex Numbers 1.3. Complex Conjugate

Im

z1

Re

z2
z1 z2
Figure 1.8: Complex multiplication. When we multiply two complex numbers z1 , z2 ,
the corresponding distances r1 and r2 are multiplied while the corresponding arguments
φ1 , φ2 are summed up.

Im
z = x + iy

Re

z = x − iy

Figure 1.9: The complex conjugate z is a reflection of z about the real axis.

can be computed efficiently: The distance r to the origin is simply raised to the
power of n and the argument is scaled/multiplied by n. This also immediately gives
us the result

(r(cos φ + i sin φ))n = r n (cos(nφ) + i sin(nφ)) (1.25)

which will later (Section 1.4) know as de Moivre’s theorem.

1.3 Complex Conjugate

The complex conjugate of a complex number z = x+iy is z = x−iy. Some properties
of complex conjugates include:
1. <(z) = <(z)

2. =(z) = −=(z)

3. z + z = 2x = 2<(z) ∈ R

9
1.3. Complex Conjugate Chapter 1. Complex Numbers

4. z − z = 2iy = 2i=(z) is purely imaginary

5. z1 + z2 = z1 + z2

6. z1 z2 = z1 z2 . This can be seen either by noting that the conjugate operation

simply changes every occurrence of i to −i or since

(x1 + iy1 )(x2 + iy2 ) = (x1 x2 − y1 y2 ) + i(x1 y2 + y1 x2 ) , (1.26)

(x1 − iy1 )(x2 − iy2 ) = (x1 x2 − y1 y2 ) − i(x1 y2 + y1 x2 ) , (1.27)

which are conjugates. Geometrically, the complex conjugate z is a reflection of

z where the real axis serves as the axis of reflection. Figure 1.9 illustrates this
relationship.

1.3.1 Absolute Value of a Complex Number

√
The absolute value (length/modulus) of z ∈ C is |z| = zz, where

zz = (x + iy)(x − iy) = x2 + y 2 ∈ R. (1.28)

Notice that the term ‘absolute value’ is the same as defined for real numbers when
=(z) = 0. In this case, |z| = |x|.
The absolute value of the product has the following nice property that matches the
product result for real numbers:

|z1 z2 | = |z1 | |z2 |. (1.29)

This holds since

|z1 z2 |2 = z1 z2 z1 z2 = z1 z2 z1 z2 = z1 z1 z2 z2 = |z1 |2 |z2 |2 . (1.30)

1.3.2 Inverse and Division

If z = x + iy, its inverse (reciprocal) is

1 z z x − iy
= = 2= 2 . (1.31)
z zz |z| x + y2

This can be written z−1 = |z|−2 z, using only the complex operators multiply and add,
see (1.5) and (1.6), but also real division, which we already know. Complex division
is now defined by z1 /z2 = z1 z2−1 . In practice, we compute the division z1 /z2 by ex-
panding the fraction by the complex conjugate of the denominator. This ensures that
the denominator’s imaginary part is 0 (only the real part remains), and the overall
fraction can be written as
z1 z1 z2 z1 z2
= = (1.32)
z2 z2 z2 |z2 |2

10
Chapter 1. Complex Numbers 1.4. De Moivre’s Theorem

Geometric Interpretation of Division

When we use the Euler representations of two complex numbers z1 , z2 ∈ C, we can

write the division as
z1   r
= z1 z2−1 = r1 exp(iφ1 ) r2 exp(iφ2 ) = 1 exp(i(φ1 − φ2 )) . (1.33)
z2 r2

Geometrically, we divide r1 by r2 (equivalently: scale r1 by r12 ) and rotate z1 by −φ2 .

This is not overly surprising since the division by z2 does exactly the opposite of
a multiplication by r2 . Therefore, looking again at Figure 1.8, if we take z1 z2 and
divide by z2 , we obtain z1 .

Example: Complex Division

Bring the following fraction into the form x + iy:

3 + 2i
z = x + iy = (1.34)
7 − 3i

Solution:
3 + 2i (3 + 2i)(7 + 3i) 15 + 23i 15 23
= = = +i (1.35)
7 − 3i (7 − 3i)(7 + 3i) 49 + 9 58 58
15 23
Now, the fraction can be written as z = x + iy with x = 58 and y = 58 .

1.4 De Moivre’s Theorem

De Moivre’s theorem (or formula) is a central result because it connects complex
numbers and trigonometry.

Theorem 1 (De Moivre’s Theorem)

For any n ∈ N

(cos φ + i sin φ)n = cos nφ + i sin nφ (1.36)

The proof is done by induction (which you will see in detail in the course Reasoning
about Programs). A proof by induction allows you to prove that a property is true for
all values of a natural number n. To construct an induction proof, you have to prove
that the property, P (n), is true for some base value (say, n = 1). A further proof is
required to show that if it is true for the parameter n = k, then that implies it is also
true for the parameter n = k + 1: that is P (k) ⇒ P (k + 1) for all k ≥ 1. The two proofs
combined allow us to build an arbitrary chain of implication up to some value n = m:

P (1) and (P (1) ⇒ P (2) ⇒ · · · ⇒ P (m − 1) ⇒ P (m)) |= P (m)

11
1.4. De Moivre’s Theorem Chapter 1. Complex Numbers

Proof 1
We start the induction proof by checking whether de Moivre’s theorem holds for n = 1:

(cos φ + i sin φ)1 = cos φ + i sin φ (1.37)

is trivially true, and we can now make the induction step: We assume that (1.36) is
true for k and show that it also holds for k + 1.
Assuming

(cos φ + i sin φ)k = cos kφ + i sin kφ (1.38)

we can write

(cos φ + i sin φ)k+1 = (cos φ + i sin φ)(cos φ + i sin φ)k

= (cos φ + i sin φ)(cos kφ + i sin kφ) using assumption (1.38)
= (cos(k + 1)φ + i sin(k + 1)φ) using complex product (1.23)

which concludes the proof.

1.4.1 Integer Extension to De Moivre’s Theorem

We can extend de Moivre to include negative numbers, n ∈ Z

(cos φ + i sin φ)n = cos nφ + i sin nφ

We have tackled the case for n > 0 already, n = 0 can be shown individually. So we
take the case n < 0. We let n = −m for m > 0.
1
(cos φ + i sin φ)n =
(cos φ + i sin φ)m
1
= by de Moivre’s theorem
cos mφ + i sin mφ
cos mφ − i sin mφ
=
cos2 mφ + sin2 mφ
= cos(−mφ) + i sin(−mφ) Trig. identity: cos2 mφ + sin2 mφ = 1
= cos nφ + i sin nφ

1.4.2 Rational Extension to De Moivre’s Theorem

Finally, for our purposes, we will show that if n ∈ Q, one value of (cos φ + i sin φ)n is
cos nφ + i sin nφ. Take n = p/q for p, q ∈ Z and q , 0. We will use both de Moivre’s
theorems in the following:
!q
p p
cos φ + i sin φ = cos pφ + i sin pφ (1.39)
q q
= (cos φ + i sin φ)p (1.40)

12
Chapter 1. Complex Numbers 1.5. Triangle Inequality for Complex Numbers

p p
Hence cos q φ + i sin q φ is one of the qth roots of (cosφ + i sin φ)p .
The qth roots of cos φ + i sin φ are easily obtained. We need to use the fact that
(repeatedly) adding 2π to the argument of a complex number does not change the
complex number.
1 1
(cos φ + i sin φ) q = (cos(φ + 2nπ) + i sin(φ + 2nπ)) q (1.41)
φ + 2nπ φ + 2nπ
= cos + i sin for 0 ≤ n < q (1.42)
q q
We will use this later to calculate roots of complex numbers.
Finally, the full set of values for (cos +i sin φ)n for n = p/q ∈ Q is:
pφ + 2nπ pφ + 2nπ
cos + i sin for 0 ≤ n < q (1.43)
q q

Example: Multiplication using Complex Products

We require the result of:
(3 + 3i)(1 + i)3
We could expand (1+i)3 and multiply by 3+3i using real and imaginary components.
Alternatively, we could tackle this in polar form (cos φ + i sin φ) using the complex
product of (1.23) and de Moivre’s theorem.
(1 + i)3 = [21/2 (cos π/4 + i sin π/4)]3
= 23/2 (cos 3π/4 + i sin 3π/4)
by de Moivre’s theorem. 3 + 3i = 181/2 (cos π/4 + i sin π/4) and so the result is
181/2 23/2 (cos π + i sin π) = −12
Geometrically, we just observe that the Arg of the second number is 3 times that of
1 + i, i.e., 3π/4 (or 3 · 45◦ in degrees). The first number has the same Arg, so the Arg
of the result is π. √ √
Similarly, the absolute values (lengths) of the numbers multiplied are 18 and 23 ,
so the product has absolute value 12. The result is therefore −12.

1.5 Triangle Inequality for Complex Numbers

The triangle inequality for complex numbers is as follows:
∀z1 , z2 ∈ C : |z1 + z2 | ≤ |z1 | + |z2 | (1.44)
An alternative form, with w1 = z1 and w2 = z1 + z2 is |w2 | − |w1 | ≤ |w2 − w1 | and,
switching w1 , w2 , |w1 | − |w2 | ≤ |w2 − w1 |. Thus, relabelling back to z1 , z2 :
 
∀z , z ∈ C :  |z | − |z |  ≤ |z − z |
1 2 1 2 2 1 (1.45)
In the Argand diagram, this just says that “In the triangle with vertices at 0, z1 , z2 , the
length of side z1 z2 is not less than the difference between the lengths of the other
two sides”.

13
1.6. Fundamental Theorem of Algebra Chapter 1. Complex Numbers

Proof 2
Let z1 = x1 + iy1 and z2 = x2 + iy2 . Squaring the left-hand side of (1.45) yields

(x1 + x2 )2 + (y1 + y2 )2 = |z1 |2 + |z2 |2 + 2(x1 x2 + y1 y2 ), (1.46)

and the square of the right-hand side is

|z1 |2 + |z2 |2 + 2|z1 ||z2 | (1.47)

It is required to prove x1 x2 + y1 y2 ≤ |z1 ||z2 |. We continue by squaring this inequality

x1 x2 + y1 y2 ≤ |z1 ||z2 | (1.48)

⇔ (x1 x2 + y1 y2 )2 ≤ |z1 |2 |z2 |2 (1.49)
⇔ x12 x22 + y12 y22 + 2x1 x2 y1 y2 ≤ x12 x22 + y12 y22 + x12 y22 + y12 x22 (1.50)
2
⇔ 0 ≤ (x1 y2 − y1 x2 ) , (1.51)

which concludes the proof.

The geometrical argument via the Argand diagram is a good way to understand the
triangle inequality.

1.6 Fundamental Theorem of Algebra

Theorem 2 (Fundamental Theorem of Algebra)
Any polynomial of degree n of the form
n
X
p(z) = ak zk , ak ∈ C, an , 0 (1.52)
k=0

possesses, counted with multiplicity, exactly n roots in C.

A root z∗ of p(z) satisfies p(z∗ ) = 0. Bear in mind that complex roots include all real
roots as the real numbers are a subset of the complex numbers. Also some of the
roots might be coincident, e.g., for z2 = 0. Finally, we also know that if ω is a root
and ω ∈ C\R, then ω is also a root. So all truly complex roots occur in complex
conjugate pairs.

1.6.1 nth Roots of Unity

In the following, we consider the equation

zn = 1 , n ∈ N, (1.53)

for which we want to determine the roots. The fundamental theorem of algebra tells
us that there exist exactly n roots, one of which is z = 1.

14
Chapter 1. Complex Numbers 1.6. Fundamental Theorem of Algebra

Im

1 Re

Figure 1.10: Then nth roots of zn = 1 lie on the unit circle and form a regular polygon.
Here, we show this for n = 8.

To find the other solutions, we write (1.53) in a slightly different form using the
Euler representation:

zn = 1 = eik2π , ∀k ∈ Z . (1.54)

Then the solutions are z = ei2kπ/n for k = 0, 1, 2, . . . , n − 1.4

Geometrically, all n roots lie on the unit circle, and they form a regular polygon
with n corners where the roots are 360◦ /n apart, see an example in Figure 1.10.
Therefore, if we know a single root and the total number of roots, we could even
geometrically find all other roots.

Example: Cube Roots of Unity

The 3rd roots of 1 are z = e2kπi/3 for k = 0, 1, 2, i.e., 1, e2πi/3 , e4πi/3 . These are often
referred to as ω1 ω1 and ω3 , and simplify to

ω1 = 1
√
ω2 = cos 2π/3 + i sin 2π/3 = (−1 + i 3)/2 ,
√
ω3 = cos 4π/3 + i sin 4π/3 = (−1 − i 3)/2 .

Try cubing each solution directly to validate that they are indeed cubic roots.

1.6.2 Solution of zn = a + ib
Finding the n roots of zn = a + ib is similar to the approach discussed above: Let
a + ib = reiφ in polar form. Then, for k = 0, 1, . . . , n − 1,

zn = (a + ib)e2πki = re(φ+2πk)i (1.55)

1 (φ+2πk)
i
⇒ zk = r n e n , k = 0, . . . , n − 1 . (1.56)
4 Note that the solutions repeat when k = n, n + 1, . . .

15
1.7. Complex Sequences and Series* Chapter 1. Complex Numbers

Example
Determine the cube roots of 1 − i.
√
1. The polar coordinates of 1 − i are r = 2, φ = 74 π, and the corresponding Euler
representation is
√
z = 2 exp(i 7π
4 ). (1.57)

2. Using (1.56), the cube roots of z are

1 1
z1 = 2 6 (cos 7π 7π 7π
12 + i sin 12 ) = 2 exp(i 12 )
6 (1.58)
1 1 1
z2 = 2 (cos 15π
6 15π 5π 5π 5π
12 + i sin 12 ) = 2 (cos 4 + i sin 4 ) = 2 exp(i 4 )
6 6 (1.59)
1 1
z3 = 2 6 (cos 23π 23π 23π
12 + i sin 12 ) = 2 exp(i 12 ) .
6 (1.60)

1.7 Complex Sequences and Series*

A substantial part of the theory that we have developed for convergence of sequences
and series of real numbers also applies to complex numbers. We will not reproduce
all the results here, there is no need; we will highlight a couple of key concepts
instead.

1.7.1 Limits of a Complex Sequence

For a sequence of complex numbers z1 , z2 , z3 , . . ., we can define limits of convergence,
zn → l as n → ∞ where zn , l ∈ C. This means that for all  > 0 we can find a natural
number N , such that

∀n > N : |zn − l| <  . (1.61)

The only distinction here is the meaning of |zn − l|, which refers to the complex
absolute value and not the absolute real value.

Example of complex sequence convergence Prove that the complex sequence

1
zn = n+i converges to 0 as n → ∞. Straight to the limit inequality:
 
 1  <  (1.62)
n + i 
|n − i|
⇔ 2 < (1.63)
p +1
n
(n − i)(n + i)
⇔ < (1.64)
n2 + 1
1
⇔√ < (1.65)
n2 + 1

16
Chapter 1. Complex Numbers 1.7. Complex Sequences and Series*

r
1
⇒n> −1 for  ≤ 1 (1.66)
2

Thus, we can set

 q 
1
−1  ≤ 1


2

N () =  (1.67)


 1 otherwise


We have to be a tiny bit careful as N () needs to be defined for all  > 0 and the
penultimate line of the limit inequality is true for all n > 0 if  > 1. In essence this
was no different in structure from the normal sequence convergence proof. The only
difference was how we treated the absolute value.

Absolute Convergence
Similarly, a complex series ∞
P P∞
n=1 zn is absolutely convergent if n=1 |zn | converges.
Again the |zn | refers to the complex absolute value.

Complex Ratio Test

A complex series ∞
P
n=1 zn converges if
 
zn+1 
lim  <1 (1.68)
n→∞  zn 

and diverges if
 
zn+1 
lim   > 1. (1.69)
n→∞  zn 

Example of Complex Series Convergence

Let us take a general variant of the geometric series:
∞
X
S= azn−1 (1.70)
n=1

We can prove that this will converge for some values of z ∈ C in the same way we
n
could for the real-valued series. Applying the complex ratio test, we get limn→∞ | azazn−1 | =
|z|. We apply the standard condition and get that |z| < 1 for this series to converge.
The radius of convergence is still 1 (and is an actual radius of a circle in the complex
plane). What is different here is that now any z-point taken from within the circle
centred on the origin with radius 1 will make the series converge, not just on the
real interval (−1, 1).
a
For your information, the limit of this series is 1−z , which you can show using
Maclaurin as usual, discussed below.

17
1.8. Complex Power Series Chapter 1. Complex Numbers

1.8 Complex Power Series

We can expand functions as power series in a complex variable, usually z, in the
same way as we could with real-valued functions. The same expansions hold in C
because the functions below (at any rate) are differentiable in the complex domain.
Therefore, Maclaurin’s series applies and yields
∞ n
X z z2 z3
exp(z) = = 1 + z + + + ... (1.71)
n! 2! 3!
n=0
∞
X z2n+1 z3 z5
sin(z) = (−1)n = z − + − ... (1.72)
(2n + 1)! 3! 5!
n=0
∞
X z2n z2 z4
cos(z) = (−1)n = 1 − + − ... (1.73)
(2n)! 2! 4!
n=0

1.8.1 A Generalized Euler Formula

A more general form of Euler’s formula (1.12) is

∀z ∈ C, n ∈ Z : z = rei(φ+2nπ) (1.74)

since ei2nπ = cos 2nπ + i sin 2nπ = 1. This is the same general form we used in the
rational extension to De Moivres theorem to access the many roots of a complex
number.
In terms of the Argand diagram, the points ei(φ+2nπ) for i ≥ 1 lie on top of each other,
each corresponding to one more revolution (through 2π).
The complex conjugate of eiφ is e−iφ = cos φ − i sin φ. This allows us to get useful
expressions for sin φ and cos φ:

cos φ = (eiφ + e−iφ )/2 (1.75)

sin φ = (eiφ − e−iφ )/2i. (1.76)

We will be able to use these relationships to create trigonometric identities.

1.9 Applications of Complex Numbers

1.9.1 Trigonometric Multiple Angle Formulae
How can we calculate cos nφ in terms of cos φ and sin φ? We can use de Moivre’s
theorem to expand einφ and equate real and imaginary parts: e.g., for n = 5, by the
Binomial theorem,

(cos φ + i sin φ)5 = cos5 φ + i5 cos4 φ sin φ − 10 cos3 φ sin2 φ (1.77)

− i10 cos2 φ sin3 φ + 5 cos φ sin4 φ + i sin5 φ

18
Chapter 1. Complex Numbers 1.9. Applications of Complex Numbers

Comparing real and imaginary parts now gives

cos 5φ = cos5 φ − 10 cos3 φ sin2 φ + 5 cos φ sin4 φ (1.78)

and

sin 5φ = 5 cos4 φ sin φ − 10 cos2 φ sin3 φ + sin5 φ (1.79)

Trigonometric Power Formulae

We can also calculate cosn φ in terms of cos mφ and sin mφ for m ∈ N: Let z = eiφ so
that z + z−1 = z + z = 2 cos φ. Similarly, zm + z−m = 2 cos mφ by de Moivre’s theorem.
Hence by the Binomial theorem, e.g., for n = 5,

(z + z−1 )5 = (z5 + z−5 ) + 5(z3 + z−3 ) + 10(z + z−1 ) (1.80)

5 5
2 cos φ = 2(cos 5φ + 5 cos 3φ + 10 cos φ) (1.81)

Similarly, z − z−1 = 2i sin φ gives sinn φ

When n is even, we get an extra term in the binomial expansion, which is constant.
For example, for n = 6, we obtain

(z + z−1 )6 = (z6 + z−6 ) + 6(z4 + z−4 ) + 15(z2 + z−2 ) + 20 (1.82)

26 cos6 φ = 2(cos 6φ + 6 cos 4φ + 15 cos 2φ + 10) (1.83)

and, therefore,
1
cos6 φ = (cos 6φ + 6 cos 4φ + 15 cos 2φ + 10) . (1.84)
32

1.9.2 Summation of Series

Some series with sines and cosines can be summed similarly, e.g.,
n
X
C= ak cos kφ (1.85)
k=0

n
ak sin kφ. Then,
P
Let S =
k=1

n
X 1 − (aeiφ )n+1
C + iS = ak eikφ = . (1.86)
k=0
1 − aeiφ

Hence,

(1 − (aeiφ )n+1 )(1 − ae−iφ )

C + iS = (1.87)
(1 − aeiφ )(1 − ae−iφ )
1 − ae−iφ − an+1 ei(n+1)φ + an+2 einφ
= . (1.88)
1 − 2a cos φ + a2

19
1.9. Applications of Complex Numbers Chapter 1. Complex Numbers

Equating real and imaginary parts, the cosine series is

1 − a cos φ − an+1 cos(n + 1)φ + an+2 cos nφ

C= , (1.89)
1 − 2a cos φ + a2

and the sine series is

a sin φ − an+1 sin(n + 1)φ + an+2 sin nφ
S= (1.90)
1 − 2a cos φ + a2

1.9.3 Integrals
We can determine integrals
Z x
C= eaφ cos bφdφ , (1.91)
Z0x
S= eaφ sin bφdφ (1.92)
0

by looking at the sum5

Z x
C + iS = e(a+ib)φ dφ (1.93)
0
e (a+ib)x −1(eax eibx − 1)(a − ib)
= = (1.94)
a + ib a2 + b2
ax ax
(e cos bx − 1 + ie sin bx)(a − ib)
= (1.95)
a2 + b2
The result is therefore
eax (a cos bx + b sin bx) − a + i(eax (a sin bx − b cos bx) + b)
C + iS = (1.96)
a2 + b2
and so we get

eax (a cos bx + b sin bx − a)

C= , (1.97)
a2 + b2
eax (a sin bx − b cos bx) + b
S= (1.98)
a2 + b2
as the solutions to the integrals we were seeking.

5 The reduction formula would require a and b to be integers.

20
Bibliography

Box, G. E. P. and Muller, M. E. (1958). A Note on the Generation of Random Normal

Deviates. Annals of Mathematical Statistics, 29(2):610–611. pages 2

Knop, R. (1969). Remark on Algorithm 334 [G5]: Normal Random Deviates. Com-
munications of the ACM, 12(5):281. pages 2

21