Chapter 4

Complex Variables

Although all the quantities are real variables in the physical

world, however, it is son1etimes easy or even necessary to use
complex variables in mathematics and engineering. In fact,
the techniques based on complex variables are among the most
powerful methods for mathematical analysis and solutions of
mathematical models.

4.1 Complex Numbers and Functions

lviathematically speaking, a complex number z is a generalized
set or the order pair of two real numbers (a, b), written in the
form of
z =a+ ib, z·2 = -1' a, bE 'R, (4.1)
which consists of the real part ~(z) =a and the imagery part
SJ<( z) = b. It can also be written as the order pair of real nunl-
bers using the notation (a, b). The addition and substraction
of two complex numbers are defined as
(a+ ib) ± (c + id) =(a± c)+ i(b ±d). (4.2)
The multiplication and division of two complex numbers are in
the similar way as expanding polynon1ials
(a+ ib) · (c + id) = (ac- bd) + i(ad +be), (4.3)

61
4.1 Complex Numbers and Functions Complex Variables

and
a + ib ac + bd . be - ad
-c_+_i_d = c2 + d2 + z-:c2~+-d::::-2 · (4.4)
Two complex numbers are equal a + ib = c + id if and only if
a= c and b =d. The complex conjugate or simply conjugate
z (also z*) of z =a+ ibis defined as
z =a- ib. (4.5)
The order pair (a, b), similar to a vector, implies that a
geometrical representation of a complex number a + ib by the
point in an ordinary Euclidean plane with x-axis being the
real axis and y-axis being the imaginary axis (iy). This plane
is called the complex plane. This representation is often called
the Argand diagram (see Figure 4.1). The vector representation
starts from (0, 0) to the point (a, b). The length of the vector
is called the magnitude or modulus or the absolute value of the
complex number
r = lzl = v' a2 + b2 . (4.6)

Figure 4.1: Polar representation of a complex number.

The angle () that the vector makes with the positive real
axis is called the argument (see Fig 4.1),

() = arg z. (4.7)
In fact, we may replace () by () + 2n7r (n E N). The value range
-1r < () :::; 1r is called the principal argument of z, and it is

62
Complex Variables 4.1 Complex Numbers and Functions

usually denoted as Argz. In the complex plane, the complex

number can be written as

z = rei() = r cos( 0) + ir sin( 0). (4.8)

This polar form of z and its geometrical representation can
result in the Euler's formula which is very useful in the complex
analysis
ei6 = cos(O) + isin(O). (4.9)
The Euler formula can be proved using the power series. For
any z E C, we have the power series
z2 zn
ez = 1 + z + 2f + ... + ~ + ... , (4.10)

and for a special case z = iO, we have

ilJ 1 .0 02 i03
e = + z - 2! + 31 - ... ,
02 03
= (1- 2f + ... ) + i(O- 3f + ... ). (4.11)
Using the power series
03 os
sin 0 =0- -
3!
+ -51 - ... , (4.12)

and
02 04
cos 0 = 1 - - + - - ... , (4.13)
2! 4!
we get the well-know Euler's formula or Euler's equation

eilJ = cos 0 + i sin 0. (4.14)

For 0 = 1r, this leads to a very interesting formula

(4.15)

If we replace 0 by -0, the Euler's formula becomes

e-ilJ = cos( -0) + i sin( -0) = cos 0 - i sin 0. (4.16)

63
4.1 Complex Numbers and Functions Complex Variables

Adding this equation to (4.14), we have

ei0 + e-iO = 2 cos 8, (4.17)

or
eiO + e-iO
cos8 = (4.18)
2
Similarly, by deducting (4.16) from (4.14), we get
eiO- e-iO
sin8 = i (4.19)
2
For two complex numbers z1 = r 1ei 01 and z2 = r2ei 02 , it is
straightforward to show that

ZJZ2 = r1r2ei(o 1 +02 ) = r1r2[cos(a1 + a2) + isin(at + a2)],

(4.20)
which can easily be extended to get the well-known de l\1oivre's
formula

[cos( B)+ isin(8)r = cos(n8) + isin(n8). (4.21)

0 Example 4.1: Find z 4 if z = 1 + J3i. lVe can evaluate it by

direct calculation

z4 = (1 + J3i) 4 = [(1 + J3i) 2 ) 2 = [1 - 3 + 2J3i) 2

= 22 ( -1 + v'3i) 2 = 4(1 - 3- 2v'3) = -8- 8v'3i.
lVe can also use Moivre's formula. The modulus of z is r = lzl =

V1 2
2
+ J3 = 2. The argument()= tan- 1
f = 7r/3 = 60°. Thus,
z = 2e7rf 3 . We now have

z 4 = 24 e 4 7r/J = 16(cos 4 11" + isin 411")

3 3

= 16( -~- v; i) = -8- 8v'3i,

which is exactly the same result as we obtained earlier. The second
method becomes much quicker if you want to evaluate (say) z 100 .
0

64
Complex Variables 4.2 Hyperbolic Functions

5~---,~----~----~----~----~----~
\
\
~
\
\

''
''
''' ...

-
>C
~ 0
c
u;
-·-·---~-·-·-·- -

--sinh
--- cosh
·-·- tanh
-5
-3 -2 -1 0 1 2 3
X

Figure 4.2: Hyperbolic functions.

4.2 Hyperbolic Functions

Hyperbolic functions occur in many applications and they can
be thought as the complex analogues of trigonometric func-
tions. The fundamental definitions are
ex- e-x ex+ e-x
sinhx = coshx = , (4.22)
2 2
and
1 sinhx
tanhx = --h-~
(4.23) coth x = --h-.
COS X tan x
Figure 4.2 shows the variation of sinh, cosh, and tanh. If
we replace x by ix and use Euler's formula, then we have
eix _ e-ix
sinh ix =
2

= 1 [( COSX . . )-
+ ZSlnX (COSX . . )]
-ZSlnX . .
= ZSlllX. (4.24)
2
65
4.2 Hyperbolic Functions Complex Variables

Similarly, we have
1
+ e-zx)
0
°
cosh ix = -(ezx
2

= ~[(cosx + isinx) + (cosx- isinx)] = cosx. (4.25)

In a similar fashion, we can also prove that

cosix = coshx, sin ix = i sinh x. (4.26)

Some identities are as follows:

cosh2 x- sinh2 x = 1, (4.27)

sinh 2x = 2 sinh x cosh x, (4.28)

and
cosh 2x = sinh2 x + cosh 2 x. (4.29)

0 Example 4.2: Prove that cosh 2 x- sinh 2 x = 1. From the

definitions, we have

and

Thus, we have

1
= 4[2- (-2)) = 1.

0
The inverses of hyperbolic functions are defined in a similar
way as trigonometric functions. For example, y = cosh x, its
inverse is defined as x = cosh -t y. From the basic definitions,
we have
sinhx + coshx =ex. (4.30)

66
Complex Variables 4.3 Analytic Functions

Using sinhx = Vcosh2 x- 1, we have

Vcosh 2 x- 1 + coshx =ex, (4.31)

or
(4.32)
which gives

4.3 Analytic Functions

Analytic Functions
Any function of real variables can be extended to the function
of complex variables in the same fonn while treating the real
numbers x as x + iO. For example, f (x) = x 2 , x E 'R becomes
f (z) = z 2 , z E C. Any complex function f (z) can be written as

f(z) = f(x + iy) = ~(f(z)) + ir;s(J(z))

= u(x, y) + iv(x, t), (4.34)
where u(x, y) and v(x, y) are real-valued functions of two real
variables.
A function f (z) is called analytic at z0 if f' (z) exists for all z
in smne €-neighborhood of zo, that is to say, it is differentiable
in some open disk lz- z0 1< f. If f(z) = u + iv is analytic at
every point in a domain n, then u(x, y) and v(x, y) satisfying
the Cauchy-Riemann equations

8u 8v
(4.35)
8x 8y'

Conversely, if u and v of f (z) = u + iv satisfy the Cauchy-

Riemann equation at all points in a domain, then the complex
function f(z) is analytic in the same domain. For example, the

67
4.3 Analytic Functions Complex Variables

elementary power function w = zn, (n > 1) is analytic on the

whole plane, w = pei(j>, z = rei 9 , then

p = rn, 4> = n8. (4.36)

The logarithm is also an elementary function w = In z

In z =In lzl + i arg(z) =In r + i(8 + w1rk), (4.37)

which has infinitely many values, due to the multiple values of

8, with the difference of 21rik(k = 0, ±1, ±2, ... ). If we use the
principal argument Argz, then we have the principal logarithm
function
Ln(z) =In lzl + Argz. (4.38)
If we differentiate the Cauchy-Riemann equations, we have
fJ2uj8x8y = 8 2 uj8y8x. After some calculations, we can reach
the following theoren1. For given analytic function f(z)
u + iv, then both u and v satisfy the Laplace equations

(4.39)

This is to say, both real and imaginary parts of an analytic

function are harmonic.
A very interesting analytical function is the Riemann zeta-
function ((s), which is defined by
00 1
((s) = ~ -, (4.40)
L- ns
n=l

where s is a complex number with its real part more than unity.
That is s E C and ~(s) > 1. This function (infinite series) is
analytic, and it can be extended for all complex numbers s =I= 1.
For example,
1 1 7r
2
((2) = 1 + 22 + 32 + ... = 6' (4.41)

but
1 1 1
((1) = 1 + 2 + 3 + 4 + ... = 00. (4.42)

68
Complex Variables 4.3 Analytic Functions

This ((s) function has trivial zeros s = -2, -4, -6, ... and it
also has non-trivial zeros.
There is a famous unsolved problem, called the Riemann
hypothesis, related to this function. The Riemann hypothesis
conjectured by Bernhard Reimann in 1859 states that all real
parts of any non-trial zero of the Riemann zeta-function ((s)
are !· That is to say, all the non-trivial zeros should lie on a
!
straight lines= + iy. This is a-million-dollar open problem
as the Clay Mathematical Institute in 2000 offered a million
dollars to search for a proof, and yet it still remains unsolved.

Laurent Series
For an analytic function p(z), one of important properties is
the singularity such as the pole. If p( z) can be written as

q(z)
p(z) = ( z- zo )n' (4.43)

where n > 0 is a positive integer while q(z) =/= 0 is analytic

everywhere in the neighbourhood containing z = zo, we say
that p( z) has a pole of order n at z = z0 . The above definition
is equivalent to say that the following limit is finite

lim [p(z)(z- zot] = (,

z-zo
11(11 < oo, ( E C. (4.44)

Any analytic function f (z) can be expanded in terms of the

Taylor series

(4.45)

This expansion is valid inside the analytic region. However,

if the function f (z) has a pole of order n at z = zo and it is
analytic everywhere except at the pole, we can then expand the
function p(z) = (z-z 0 )n f(z) in the standard Taylor expansion.

69
4.4 Complex Integrals Complex Variables

This means that original function f (z) can be written as a

power series
O_n O_t
)
!( z=( ) + ... +( z-zo )
z-zon

oo(z- zo) + ... + ok(z- zo)k + ... , (4.46)

which is called a Laurent series, and it is an extension of the
Taylor series. In this series, it is often assumed that O-n =F 0.
The tern1s with the inverse powers a-n/(z-zo)n+ ... +a-t/(z-
zo) are called the principal part of the series, while the usual
terms ao(z- zo) + ... + ok(z- zo)k + ... are called the analytic
part.
Furthermore, the most in1portant coefficient is probably
O-t which is called the residue of f(z) at the pole z = zo.
In general, the Laurent series can be written as
00

f(z) = L ok(z- zo)k, (4.47)

k=-n
where n may be extended to include an infinite number of terms
n-+ -oo.

4.4 Complex Integrals

Given a function f (z) that is continuous on a piecewise smooth
curve r, then the integral over r, fr f(z)dz, is called a contour
or line integral of f (z). This integral has similar properties as
the real integral

frlof(z) + fig(z)]dz = o £ f(z)dz + .8£g(z)dz. (4.48)

If F(z) is analytic and F'(z) = f(z) is continuous along a curve

r, then
l f(z)dz = F[z(b)]- F[z(a)]. (4.49)

70
Complex Variables 4.4 Complex Integrals

Cauchy's Integral Theorem

We say a path is simply closed if its end points and initial
points coincide and the curve does not cross itself. For an
analytic function f(z) = u(x, y) + iv(x, y), the integral on a
sin1ply closed path

I= lr f(z)dz = fr (u + iv)(dx + idy)]

= fr(udx- vdy) +i fr (vdx + udy). (4.50)

By using the Green theorem, this becomes

[ 8u av . [ 8u 8v
I = Jn (- ay - ax )dxdy + zJn (ax - ay )dxdy. (4.51)

Fron1 the Cauchy-Riemann equations, we know that both in-

tegrals are zero. Thus, we have Cauchy's Integral Theorem,
which states that the integral of any analytic function f(z) on
a simply closed path r in a simply connected domain n is zero.
That is
fr
f(z)dz = 0.

SS(z)

Figure 4.3: Contours for Cauchy integrals.

This theorem is very important as it has interesting conse-
quences. If the close path is decomposed into two paths with

71
4.4 Complex Integrals Complex Variables

reverse directions r 1and r 2(see Figure 4.3), then r1 and -r2

form a close path, which leads to

{ f(z)dz = f f(z)dz. (4.52)

lr1 lr2
That is to say that the integrals over any curve between two
points are independent of path. This property becomes very
useful for evaluation of integrals. In fact, this can be extended
to the integrals over two closed paths r and -y such that -y is a
very small circular path inside r. Using a small cut with two
curves C1 and C2 so that these two curves combine with r and
-y form a closed contour (see Figure 4.3), the Cauchy integral
theorem implies that

fr f(z)dz = L f(z)dz, (4.53)

since the contribution from the cut is zero.

For an analytic function with a pole, we can make the con-
tour -y sufficiently small to enclose just around the pole, and
this makes the calculation of the integral much easier in some
cases.
For the integral of p( z) = f (z) / (z - zo) over any simply
closed path r enclosing a point zo in the domain n,

I= frp(z)dz, (4.54)

we can use the Laurent series for p( z)

p(z) = ( a-1 ) +ao(z-zo ) + ... +ak ( z-zo) k + ... , (4.55)

z- zo
so that the expansion can be integrated term by term around
a path. The only non-zero contribution over a small circular
contour is the residue a-1· We have

kf p(z)dz = 21ria-1 = 21ri Res[p(z)]j ,

I= (4.56)
~
72
Complex Variables 4.4 Complex Integrals

which can be written in terms of f(z) as

i
- 1 . -f(z)
- d z = f(zo).
21rz r z- zo
(4.57)

Similarly, this can be extended for higher derivatives, and we

have
j f(z) dz = 21rij(n)(zo).
Jr (z- zo)n+l nl

-00 00
Figure 4.4: Contour for the integral J(o, (3).

Residue Theorem
For any analytic f (z) function in a domain n except isolated
singularities at finite points Z1, z2, ... , ZN, the residue theorem
states
i r
f(z)dz = 21fi L
k=t
N
Resf(z)lzk'

where r is a simple closed path enclosing all these isolated

points. If f(z) has a pole of order Nat zo, the following forn1ula
gives a quick way to calculate the residue

_ 1 . dN-t [(z- zo)N f(z)]

Resf(z)lzo - (N- 1)1 }.!...~o dzN-1 . (4.58)

The residue theoren1 serves a powerful tool for calculating

some real integrals and summation of series, especially when

73
4.4 Complex Integrals Complex Variables

the integrand is a function of sin and cos that can be changed

into a complex integral. The real integral f~oo 1/J(x)dx becomes
27ri multiplying the sun1 of the residues of 1/J(x) at the poles in
the upper half-space.

0 Example 4,.9: In order to evaluate the integral

it is necessa1y to construct a contour (see Figure 4.4). As the function

</> = ei </ (/34 + ( 2 ) has two poles ( = +i/32 and -i/3 2 and only one
02

pole ( = +i/32 is in the upper half plane, we can construct a contour

to encircle the pole at ( = i{J2 by adding an additional arc at the
infinity (( --+ oo) on the upper half plane. Combining the arc with
the horizontal line from the integral limits from -oo to oo along the
(-axis, a contour is closed. Hence, we have

eio2c; /(( + ij32) f(()

4> = ( - i(32 = ( - ij32 '
2
where f(() = eio c; /(( + i/32 ). Using the residue theorem, we have

In a special case when o = 0, we have

oe 1 1r

f -00 (
2
+ j34 d( = fJ/l2.

Another important topic in complex variables is the confor-

mal mapping. The essence of a conformal mapping

w = f(z), z,w EC, (4.59)

is that this mapping preserves the angles between curves and
their orientations. One of the widely used mappings is 1\tlobius
linear fractional transformation
az+f3
W= ' (4.60)
"fZ + 8

74
Complex Variables 4.4 Complex Integrals

By choosing the appropriate coefficients o., (3, "/, {J E C, this

mapping can include all major geometrical transformations such
as translations, rotations, inversion, and expansions and con-
tractions. Conformal mappings are useful in solving steady-
state problems involving harmonic functions by transforming
the problem from a complicated geon1etrical domain to a reg-
ular domain such as circles and rectangles, and subsequently
the techniques based on conformal mapping are widely used in
solving Laplace's equation in engineering.

75