196 Chapter 8.

Complex numbers

8.8 De Moivre’s formula

De Moivre’s formula is useful to calculate powers and roots of complex numbers. It states:
If z = r [cos ω + i sin ω ] then z n = r n [cos(n ω ) + i sin(n ω )] .

This is a direct consequence of the exponential form:

z = r eiω =→
z et r le r e r cos It is not

8.8.1 Complex number multiplication as a transformation of the plane

Writing multiplications in exponential form shows that multiplying by a complex number
w = r eiω corresponds to performing a rotation by an angle ω = arg(w ) and scaling uniformly
by a factor r = |w |:
Z Rein
wz =
reit Reit rR e R e Irreilota

Going further
YEE
To see this using vector transformations,
! " consider w = a + i b and z = x + i y , and find the
x
transformation of the vector z = represented by z when it is multiplied by w :
y

wz fatib pitiy
an aig ibn i
by
a by i s
ay
0st
as 2D vector sina.at afm
971 by a b
betas a

air
f

a
foso
 sino

8.8. De Moivre’s formula 197

Combining rotations and scalings are only a subset of all the linear transformation of vectors that
matrices can perform, but manipulating complex numbers to rotate and scale objects is arguably
simpler than manipulating matrices. The development of quaternions was largely motivated by
generalising this idea to represent rotations of 3D vectors.

8.8.2 Exercises
Let z = 10 eiϵ/2 and w = 2 eiϵ/4 . Draw these numbers in the complex plane and compute the
following complex numbers:
1. eiω z
2. i z
3. z w yes
4. z /w
5. z + w
6. e2nϵi , n ↑ Z.

1 e'oz e
foe 10e
n

2 iz i toe
FIR
e toe
e't E Re
10 10 e 10 f
3 ZW toe ze 20 ei zoe
aim
r

Y
E
5 it E
se re

5 2 w n Ze 10 IE I E 2 is
D
10
f tiff r
Je t 10th I
 I
198 Chapter 8. Complex numbers

6 e n t I

it
47 2 21 4 2

8.9 Complex roots

Complex roots generalise square roots and n th root of real number. Complex roots always
exist! Finding the n th root of a number means solving the following:
Knowing the value of z n , what is z ?
It’s easy to solve this question in exponential form. If z n = R eiϑ , then there are always n distinct
roots z , given by:
↓ 2ϵ
R e i ωk ,
n ϑ
zk = ωk = +k , k = 0, 1, . . . , n ↔ 1.
n n

Let z = r eiω

z re ein Reid
r R r R real nth root
ofr
e e
no KIT KEI
t k 2 0
I 1,2 n I

unique argument
z Rei

↓
The roots are new complex numbers, evenly arranged around the circle of radius r = R .
n

These roots are therefore located at the vertices of a regular n polygon, with the first vertex
being at the angle ϑ/n (Fig. 8.7).
8.9. Complex roots 199

1 1
Fig. 8.7 The complex roots of z n are at the vertices of regular n polygons.

8.9.1 Exercise
Find all possible z for
1. z 2 = ↔1
2. z 2 = ↔r , r > 0
3. z 3 = 1

1 Z I le Re R 1 I
2m k
Z le e

n Im
Z re't Zo

I I

f
r
Re
k

k2I I kt
K O
I
K
31
Zo let let
200 Chapter 8. Complex numbers

2 Z r v70 z re

2 5 KEI th
Z Zz
Zo iJr
3 23 1 le R 1 0

RT I one
a
Eth
8.9.2 Roots of the quadratic equation
Complex numbers enables us to generalise the roots of the parabola
f (x ) = a x 2 + b x + c
as the complex numbers
↓
↔b ± b 2 ↔ 4a c
D 5 49
z± =
2a
.
512
Δ
EEE4m FIR
him

cnn.sk i tiene
Δ O b
a

CO Z
 u
rea
8.10. Fundamental theorem of Algebra 201

Remark
The roots of irreducible quadratics are always a pair of complex conjugate numbers.

8.10 Fundamental theorem of Algebra

Without complex numbers, the fundamental theorem of algebra states that any polynomial of
degree n with real coe!cients can be written in factored form with k ! n real roots and (n ↔k )/2
irreducible factors. Complex numbers give us a more powerful version of the fundamental
theorem of algebra:
Any polynomial of degree n with real coe!cients has exactly n roots (zeros). These
roots are either real numbers, or pairs of complex conjugate numbers.
In other terms, we can always factor

p (x ) = cn x n + cn↔1 x n↔1 + . . . + c1 x + c0 ,

where c1 , . . . cn ↑ R (cn ↗= 0), into

p (x ) = cn (x ↔ a 1 )(x ↔ a 2 ) · · · (x ↔ a n ),

where the roots a 1 , . . . , a n are not necessarily distinct, and may be complex-valued. If one root
is complex-valued, then its complex conjugate is also one of the roots.
Going further
The fundamental theorem of algebra generalises to complex polynomials

p (z ) = cn z n + . . . + c1 z + c0 , where c1 , . . . cn ↑ C.

Such polynomials can always be written in fully factored form with complex roots. The di"erence
with having c1 , . . . cn ↑ R is that these roots do not necessarily come in complex conjugate pairs.

8.10.1 Example
Factor the cubic polynomial

p (x ) = x 3 ↔ x 2 + x ↔ 1.
A a
x b Gi el
nano
PH A a
qp 1
PH I 0
pH 4 1
qc
a
202 Chapter 8. Complex numbers

i
e nli.it
i Is
1H 11

pH N 1
qty A 11 2 1

x
f it ti
1

real
Taxcoljugate
pair

8.11 Historical notes

↓
• ↔1 was introduced by Italian mathematicians in the 16th century. Cardano’s formula
provides an explicit expression for a real root of cubic polynomials.
• Imaginary was coined by Descartes (1596–1650);
• Complex numbers was coined by Gauss (1777–1855);
↓
• The notation “i = ↔1” was introduced by Euler (1707–1783) to avoid
↓ inconsistencies
↓ ↓
when
# manipulating↓ square
↓ roots of negative numbers: we have a b = a b , but
1 = (↔1)(↔1) =↓ ↔1 ↔1 = i 2↓ = ↔1? This ambiguity is lifted by considering i such that
i = ↔1, but i ↗= ↔1. We have ↔1 = ± i (two roots);
2

• Hamilton (1805–1865) defined an algebra for couples (x , y ) which coincides with the
algebra of complex numbers z = x + i y , i.e.: (a , b ) + (c , d ) = (a + c , b + d ) and
(a , b )(c , d ) = (a c ↔ b d , b c + a d ).
• The conceptualisation of a complex number as a point in 2D space helped remove the
mystery around imaginary numbers (Wessel, 1799). It also led to the development of 2D
vectors;
• Euler (1707–1783) showed that roots of complex numbers are vertices of regular polygons;
• Cauchy (1789–1857) initiated the theory of complex function analysis (not covered in this
unit).
8.11. Historical notes 203

Epilogue
Coming back to Cardano’s riddle, we can solve it as being the (complex) roots of the quadratic
x 2 ↔ 10x + 40, for which we now have an explicit formula, since it still works when the discrim-
inant is negative:
↓
10 ± 100 ↔ 160 ↓ ↓ ↓
x± = = 5 ± 25 ↔ 40 = 5 ± ↔15 = 5 ± i 15
2
Note that the quadratic equation x 2 ↔ 10x + 40 = 0 that a and b should be solutions of can be
found from Cardano’s riddle by substituting b = 10 ↔ a into a b = 40:

a (10 ↔ a ) = 40 =→ a 2 ↔ 10a + 40 = 0.

Since a and b play symmetric roles they are both solutions of this quadratic. Thus
↓ ↓
a = 5 + i 15, b = 5 ↔ i 15.

It is easy to check that a + b = 10, and

$ ↓ %$ ↓ %
a b = 5 + i 15 5 ↔ i 15 = 25 ↔ 15 i 2 = 40.

Going further
8.12 Imaginary in the real world
Now that we know complex numbers, we can gain more insights into how they arise in real-world
situations.

8.12.1 Stability of dynamical systems, bifurcation and control theory

Stability analyses of dynamical systems is beyond our scope, but is based on systems of di"erential
equations that depend on a parameter r , of the type:
d
x (t ) = f r (x (t ))
dt
When these equations are linearised about a steady state x0 , i.e., a zero of f r (x ), they can be
rewritten as systems of equations of the type
d
y (t ) = ϖy (t ) (8.3)
dt
for deviations y (t ) = x (t ) ↔ x0 from steady state. In this so-called eigenmode (or spectral)
decomposition, ϖ can be complex-valued, and in general, depends on the tunable parameter r .
The solution to the di"erential equation (8.3) is readily found by dividing by y (t ) and integrating
with respect to t by substitution, which gives:
y (t ) = Aeϖt ,

where A is a constant. If ϖ = a + i b , then, since exp((a + i b )t ) = exp(a t ) exp( i b t ) and using

Euler’s formula:
y (t ) = Aea t (cos(b t ) + i sin(b t )) . (8.4)
204 Chapter 8. Complex numbers

The behaviour of y (t ) = x (t ) ↔ x0 is dramatically di"erent depending on the sign of a = Re ϖ,

and depending on whether b = Im ϖ is zero or not, and these signs can depend on the tunable
parameter r . If a > 0, the solution is unstable since x (t ) moves away from x0 exponentially fast.
If a < 0, the solution is stable and returns to its steady state x0 exponentially fast. If b ↗= 0, these
behaviours are oscillatory.
When these behaviours change when the parameter r changes, due for example to a sign change
in a , or the appearance of a nonzero imaginary part b , we say that the system undergoes a
bifurcation. Examples of bifurcations include buckling (r =compressive force), caster wheel
instabilities (r =speed), unbalanced wheel wobble (r =rotation speed), airplane wing stall (r =wing
angle of attack), flag or sail flutter (r =wind speed, sail tension), etc. Control theory aims to add
components to an engineering system so that the state of the dynamical system can be controlled,
usually so it remains stable.

8.12.2 Making solutions real

An apparent issue in Eq. (8.4) above is that the solution y (t ) appears to be complex-valued. In
reality, if the initial problem is real, then the di"erent possible values of ϖ come in complex
conjugate pairs (they are zeros of real polynomials) and the solution can be written as

y (t ) = Aeϖt + B eϖt ,

where A and B are integration constants to be determined. These constants may be complex
valued, so we can always make y (t ) real-valued by choosing B = A :
$ %
y (t ) = Aeϖt + Aeϖt = Aeϖt + Aeϖt = 2 Re Aeϖt ↑ R

Often a shortcut notation is used by simply adding the complex conjugate of Aeϖt directly:

y (t ) = Aeϖt + Aeϖt = Aea t ei b t + c.c.

Using Euler’s formula, it can then be shown after a long but straightforward calculation that
& '
y (t ) = Aea t ei b t + c.c. = ea t ϑ cos(b t ) + ϱ sin(b t ) ,

where ϑ = 2 Re(A) and ϱ = ↔2 Im(A) are real constants. This final expression is now clearly
real-valued and exhibits the same roles described above for how a and b influence the solution.

8.12.3 Electrical circuits

When an alternating current I (t ) = I0 sin(ςt ) flows through an electrical circuit component, the
current may be modified in amplitude I0 , and by a phase shift ϕ .
We can easily understand that multiplying I (t ) = I0 eiςt (+c.c.) by a complex number can model
both of these changes. If z = z 1 + i z 2 = R eiϑ represents the e"ect of the electrical circuit
component onto the incoming current I (t ), then the outgoing current z I (t ) has both its amplitude
modified and phase shifted:

z I (t ) = R eiϑ I0 eiςt = (R I0 )ei(ςt +ϑ) .

8.11. Historical notes 205

8.12.4 Transmission and reflection of electromagnetic waves

The speed of light in a medium is slower than in the vaccum by the factor 1/n , where n " 1 is
the index of refraction of the medium (by definition). Generalising n to be a complex number is
useful to model the attenuation of electromagnetic waves in that medium.
When an electromagnetic wave hits the medium’s interface, it splits into a reflected wave, and an
incident wave (Fig. 8.8). The electromagnetic wave is represented by a spatially and temporally
varying function of the type

ei(ςt ↔k x ) (+ c.c),

where
ς
k =n
c
is the dispersion relation, representing the fact that di"erent wavelengths propagate at di"erent
speeds, and disperse as they get refracted with di"erent angles at the medium’s interface.
The absorption of the transmitted wave in the medium, i.e., the exponential attenuation of the
wave’s amplitude in space, can be represented by the imaginary part of the refraction index. If
n = n1 + i n2 , then
( ) ς *+ ( ς +
exp { i (ςt ↔ k x )} = exp i ςt ↔ n1 x exp ↔n2 x .
c c

Fig. 8.8 The incident electromagnetic wave is split into a reflected wave, and a transmitted wave.
The absorption of the transmitted wave by the medium is due to the imaginary part of the index
of refraction.

Fourier-Transform Infrared spectroscopy (FTIR) is an imaging technique that measures the

absorption spectrum of infrared light shone onto material specimens. Measurements of the
transmitted electromagnetic waves provide frequency-dependent information on the index of
refraction, and therefore on the molecular composition of the specimen.

8.12.5 Airplane wings and the Magnus e!ect

The mathematics of fluid flow in two-dimensional space is highly simplified by the use of complex
variables. The airflow around an airplane wing seen in cross section can be mapped onto the
airflow around a rotating cylinder (Fig. 8.9). This mapping relies on extraordinary properties of
functions of complex variables and their calculus, and it directly connects the lift force of a wing
to the Magnus e"ect describing lift on a rotating body.
206 Chapter 8. Complex numbers

Fig. 8.9 Complex transformations of space show that that lift generated by an airplane wing is
the same as lift generated by a rotating cylinder (Magnus e"ect).

The lift force on the rotating cylinder can be understood using Bernouilli’s principle that fluid
pressure is low where fluid velocity is high, and fluid pressure is high where fluid velocity is low.
The rotation of the cylinder creates a fluid velocity di"erence between the bottom and top of the
cylinder, and therefore a pressure di"erence generating lift. The Magnus e"ect is what curves
the trajectory of spinning balls in many sports such as soccer, tennis, cricket, etc.

8.12.6 Schrödinger’s equation of quantum mechanics

The evolution equation for the wavefunction ψ(x , t ) of a quantum mechanical particle of mass m
in an energy potential V (x ) is ruled by Schrödinger’s equation, in which the imaginary number
appears explicity:
, -
d h 2 d2
ħ
iħ
h ψ(x , t ) = ↔ + V (x ) ψ(x , t ),
dt 2m dx 2

where h ħ is a universal constant of quantum mechanics (Planck’s constant). The wavefunction

ψ(x , t ) is complex-valued, but all measurable properties are given by real properties related
to ψ. For example, the
.
probability
.2
of presence of the particle at position x at time t is given by
the square modulus .ψ(x , t ). .