COMPLEX ANALYSIS

A Short Course

M.Thamban Nair

Department of Mathematics
Indian Institute of Technology Madras

January-May 2011
.
Contents

Preface v

1 Complex Plane 1
1.1 Complex Numbers . . . . . . . . . . . . . . . . . . . . 1
1.2 Some Definitions and Properties . . . . . . . . . . . . 3
1.2.1 Metric on C . . . . . . . . . . . . . . . . . . . 3
2. Polar representation and n th -roots . . . . . .. 5
3. Steriographic projection . . . . . . . . . . . . . 6
1.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . 7

2 Analytic Functions 9
2.1 Differentiation . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Holomorphic or Analytic Functions . . . . . . . . . . . 14
1. Analytic extension . . . . . . . . . . . . . . . . 15
2. Geometric representations . . . . . . . . . . . . 16
3. Curves in the complex plane . . . . . . . . . . 18
2.3 Fractional linear transformations . . . . . . . . . . . . 2.3.1 23
The map z ›→ 1/z . . . . . . . . . . . . . . . . 23
2.3.2 Extended plane . . . . . . . . . . . . . . . . . . 25
3. Fractional linear transformations . . . . . . . . 26
4. Image of inverse points . . . . . . . . . . . . . . 29
2.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . 32

3 Elementary Functions 37
3.1 Exponential Function . . . . . . . . . . . . . . . . . . 37
3.2 Hyperbolic and Trigonometric Functions . . . . . . . . 39
3.3 Logarithms . . . . . . . . . . . . . . . . . . . . . . . . 40
3.4 Branches of arg(z ) and log(z ) . . . . . . . . . . . . . . 41
3.5 Problems . . . . . . . . . . . . . . . . . . . . . . . . . 44

iii
iv Contents

4 Power Series 46
4.1 Convergence . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2 Problems . . . . . . . . . . . . . . . . . . . . . . . . . 56

5 Integration 59
5.1 Integrals along Piecewise Smooth Curves . . . . . . . . 59
5.2 Cauchy’s Theorem . . . . . . . . . . . . . . . . . . . . 65
3. Cauchy’s Integral Formulas . . . . . . . . . . . . . . . 71
5.3.1 Appendix . . . . . . . . . . . . . . . . . . . . . 76
4. Zeros of analytic functions . . . . . . . . . . . . . . . . 77
5.4.1 Identity theorem . . . . . . . . . . . . . . . . . 79
5.4.2 Maximum modulus principle . . . . . . . . . . 5.4.3 80
Schwarz’s lemma . . . . . . . . . . . . . . . . . 81
5.4.4 On harmonic functions . . . . . . . . . . . . . . 81
5.5 Problems . . . . . . . . . . . . . . . . . . . . . . . . . 82

6 Laurent Series and Isolated Singularities 87

6.1 Laurent Series . . . . . . . . . . . . . . . . . . . . . . . 87
6.2 Isolated Singularities . . . . . . . . . . . . . . . . . . . 92
6.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . 97

7 Residues and Real Integrals 98

7.1 Residue theorem . . . . . . . . . . . . . . . . . . . . . 98
2. Calculation of Residues . . . . . . . . . . . . . . . . . 99
3.Evaluation of Improper Integrals . . . . . . . . . . . . 101 7.4
Problems . . . . . . . . . . . . . . . . . . . . . . . . . 105

Index 106
Preface

• This book is primarily for the students and teachers of IIT

Madras.

• This is based on a Core Course that I have given for the sec-
ond semester students of M.Sc. (Mathematics) at IIT Madras.
The audience included some B.Tech. students and a faculty
member (Dr. Parag Ravindran) from Mechanical Engineering
department.

• The contents of the book is in the line of the well-written,

small book Complex Function Theory1 by Donald Sarason. I
fondly acknowledge some e-mail discussions that I had with
Prof. Sarason during the time of giving the course.

IIT Madras M. Thamban Nair

June 2011

1
Second Editin, Hindustan Book Agency (‘trim’series), New Delhi, 2008.

v
.
1
Complex Plane

1. Complex Numbers
After having the real field R, it is natural to look for a bigger field in
which algebraic equations such as
x2 + 1 = 0 (∗)

has a solution. Of course, the + sign here must be the symbol for
addition in the bigger field. Since two fields can be considered to be
identical if there is a surjective isomorphism between then, it is
enough to have a field which contains an isomorphic image of R and
having required properties such as solution to algebraic equations.
We shall define such a field with the intention of having a solution to
the equation (∗).
Deftnition 1.1.1 The set C of complex numbers is the set of all
ordered pairs (x, y) of real numbers with the following operations of
addition and multiplication:

(x 1 , y 1 ) + (x 2 , y 2 ) = (x 1 + x 2 , y1 + y 2 ),
(x 1 , y 1 ).(x 2 , y 2 ) = (x 1 x 2 − y 1 y 2 , x 1 y 2 + x 2 y 1 ).
♦
The proof of the following theorem is left to the reader.
Theorem 1.1.1 The following hold.
(i) C is a field with additive identity (0, 0) and multiplicative identity
(1, 0).
(ii) The map ϕ : R → C defined by

ϕ(x) = (x, 0), x ∈R,

is a field isomorphism.

1
2 Complex Plane

We observe that the multiplicative inverse of a nonzero complex

number z = (x, y) is given by
. Σ
x y
,− .
x 2 + y2 x 2 + y2
Writing
i = (0, 1)
and for x ∈R,
x˜= (x, 0),
we observe that
i 2 = −˜1,
and
C = {x˜ + iy˜ : x, y ∈ R}.
With the above notations, the addition and multiplication in C can
be written as

(x˜1 + iy˜1) + (x˜2 + iy˜2) = (x˜1 + y˜2) + i(y˜1 + y˜2)

= (x˜1 + y 2 ) + i(y˜1 + y 2 ),

(x˜1 + iy˜1).(x˜2 + iy˜2) = (x˜1x˜2 − y˜1y˜2) + i(x˜1y˜2 + x˜2y˜1)

= (x1x˜2 − y 1 y 2 ) + i(x1y˜2 + x 2 y 1 ).

Throughout this course, we identify x˜with x for every x ∈ R, so

tha t C is the set of all numbers of the form

a + ib with a, b ∈R.

T hus,
a + i0 = a, 0 + i1 = i and i 2 = −1,
and for nonzero z = x + iy,
1 x y
:= z −1 = −i
z 2
x + y 2 x + y2
2

One of the important properties of this field is that, not only equa-
tion (∗) has a solution in C, but every algebraic equation also has a
solution. This is the so called fundamental theorem of algebra which
we shall prove in the due course.
Some Definitions andProperties 3

1.2 Some Definitions and Properties

Deftnition 1.2.1 For a complex number z = x + iy, x is called the
real part of z and is denoted by Re(z), y is called the imaginary
part of z and is denoted by Im(z),

z¯= x − iy

is called the complex conjugate of z, and the non-negative number

√
|z| = x 2 + y 2

is called the absolute value or modulus of z. ♦

We observe that
2 1 z¯
Re(z) ≤ |z|, Im(z) ≤ |z|, z¯z= |z| , = .
z |z|2
Using the face that |z|2 = z¯z, it can be easily verified tha t
|z1 + z2| ≤ |z1| + |z2| ∀z 1 , z 2 ∈C.

The above inequality is called the triangle inequality.

1. Metric on C
Theorem 1.2.1 The function d : C × C → R defined by
d(z 1 , z 2 ) = |z1 − z2 |, z 1 , z 2 ∈C

is a metric on C.
Hereafter any metric property of C is referred to the metric de-
fined as in the above theorem. With respect to the above metric we
have the following:

• A sequence (z n ) in C converges to z ∈ C if and only if

|zn − z| → 0 as n → ∞.

• A point z 0 ∈ C is called an interior point of a set A ⊆ C if

there exists r > 0 such tha t

B(z 0 , r) := { z ∈ C : |z − z0| < r } ⊆ A.

The set B(z 0 , r) is called the open ball with centre z and radius
r, usually denoted by B(z, r).
4 Complex Plane

• A subset t t of C is open in C if and only if every point in t t is

an interior point of tt.

• A point z 0 ∈ C is call a boundary point of a set A ⊆ C if every

open ball containing z 0 contains some point of A and some
point of its compliment, i.e., for every r > 0, B(z 0 , r) ∩ A ƒ= ∅
and B(z 0 , r) ∩A c ƒ = ∅.

• A subset S of C is closed in C it contains all its boundary

points. It can be shown that a set

S ⊆ C is closed if and only if S c := C \ S is open if

and only if for every sequence (z n ) in S,

z n → z =⇒ z ∈S.

• A function f : A → C defined on a subset A of C is continuous

at a point z0 ∈ A if and only if for every ε > 0, there exists δ >
0 such that

z ∈ A, |z − z0| < δ =⇒ |f (z) − f (z0 )| < ε.

Equivalently, f is continuous a t a point z 0 ∈ A if and only if for

every ε > 0, there exists δ > 0 such that

z ∈ A ∩B(z 0 , δ) =⇒ f (z) ∈ B ( f (z 0 ), ε).

• A function f : A → C defined on a subset A of C has the limit

ζ ∈ C at a point z 0 ∈ C if and only if for every ε > 0, there
exists δ > 0 such tha t

z ∈ A, 0 < |z − z0| < δ =⇒ |f (z) − ζ| < ε,

and in that case we write

lim f (z) = ζ.
z→z0

• A subset A of C is bounded if and only if there exists α > 0

such that
|z| ≤ α ∀z ∈ A.

• A subset A of C is compact if and only if every sequence in A

has a convergent subsequence.
Some Definitions and Properties 5

Theorem 1.2.2 The set C is a complete metric space.

Proof. Let (z n ) be a Cauchy sequence in C. Writing z n = x n +iy n

with x n , yn ∈R, we have

|xn − x m| ≤ |zn − zm|, |yn − ym| ≤ |zn − zm|

for all n, m ∈ N. Hence, (x n ) and y n ) are Cauchy sequences in R.

Since R is a complete metric space with respect to the absolute-value
metric, there exist x, y ∈ R such that x n → x and y n → y. Then,
writing z = x + iy, we have
2 2 2
|zn − z| = (x n − x) + (y n − y) → 0.

Thus, (z n ) converges to z.

1.2.2 Polar representation and nth-roots

Let z be a nonzero complex number and let θ be the angle which the
line segment joining 0 to z makes with the positive real axis, and r =
|z|, the length of the line segment. Then it is clear from the geometry
that
z = r(cos θ + i sin θ). (∗)

Deftnition 1.2.2 The representation (∗) of a nonzero z ∈ C is called

its polar representation, and an angle θ for which (∗) holds is
called an argumet of z , denoted by arg(z ). ♦
Note that each (r, θ) with r > 0 and θ ∈ R represents a unique
nonzero z ∈ C with the representation (∗), but a non zero z ∈ C has
many polar representations, namely,

z = r(cos θ + i sin θ), θ ∈ {θ 0 + 2πk : k ∈Z},

where θ0 is one of the angles for which (∗) holds.

We note that if z 1 = r1 (cos θ1 + i sin θ 1 ) and z 2 = r2 (cos θ2 +
i sinθ 2 ), then

z 1 z 2 = r1 r2 [cos(θ 1 + θ 2 ) + i sin(θ 1 + θ2)].

Thus, if z = r(cos θ + i sin θ) and n ∈ N, then

z n = rn(cos nθ+i sin nθ) = r n [cos(nθ+2kπ)+i sin(nθ+2kπ)], k ∈Z.

6 Complex Plane

The above observation helps us in finding the n t h roots of complex

numbers.
Let ζ = ρ(cos α + i sin α), z = r(cos θ + i sin θ) and n ∈ N. Then
we have
α + 2kπ
z n = ζ ⇐⇒ r = ρ1/n and θ = , k ∈Z.
n
Thus, n-th roots of ζ are
Σ . Σ . ΣΣ
α + 2kπ α + 2kπ
zk = ρ 1/n
cos + i sin , k ∈Z.
n n
Note that for each k ∈ Z

zk = z k+ n for j ∈ {1, . . . , n}.

Thus, the only distinct n-th roots of ζ are z 1 , z 2 , . . . , z n .

If ζ = 1, then we have ρ = 1 and α = 0, and in this case we see
that the n-th roots of unity are

ω, ω2, . . . , ωn with ωn = 1,

where . Σ . Σ
2kπ 2kπ
ω = cos + i sin .
n n
1.2.3 Steriographic projection
Theorem 1.2.3 The complex plane is homeomorphic with the set S 2
\ {(1, 0, 0)}, where S 2 is the unit sphere in R 3 with centre as the
origin, i.e.,

S 2 = {(α, β, γ ) ∈ R 3 : α 2 + β 2 + γ 2 = 1}.

Proof. Clearly, C is homeomorphic with the set

X = {(x, y, 0) ∈ R 3 : x, y ∈ R }

with the homeomorphism being the map x + iy ›→ (x, y, 0). Hence,

we find a surjective homeomorphism from X to S 2 \ {(1, 0,0)}.
Note that the parametric representation of the straight line join-
ing the north pole u0 = (1, 0, 0) with a point u = (x, y, 0) ∈ X is
given by
u λ = (1 − λ)u 0 + λu, λ ∈R,
Problems 7

i.e.,
u λ = (λx, λy, 1 − λ), λ ∈R.
Clearly, for every u = (x, y, 0) ∈ X there exists one and only λ := λ u
such that u λ ∈ S 2 . We consider the map
u := (x, y, 0) ›→ (λ u x, λ u y, 1 − λ)

from X to S 2 \ {(0, 0, 1)}. Note that

2 2 2 2 2 2
uλ ∈ S ⇐⇒ λ x + λ y + (1 − λ) = 1

if and only if λ = 0 or λ(x 2 + y 2 + 1) − 2 = 0. The point λ = 0

correspond to u 0 . Hence,
2
λu = .
x2 + y2 + 1

Thus the map

. Σ
2x ,2x |z|
, −1
2
z := x + iy ›→
1 + |z|2 1 + |z|2 1 + |z|2

is a bijective continuous function from C onto S 2 \ {(1, 0, 0)} with its

inverse
α + iβ ,
(α, β, γ ) ›→
1 −γ
which is also continuous.

1.3 Problems
1. Show that C is a field under the addition and multiplication
defined for complex numbers.

2. Show that the map f : R → C defined by f (x) = (x, 0) is a

field isomorphism.

3. For a nonzero complex number x, show that z −1 = z¯/|z|.

4. Show that for z 1 , z 2 in C, |z1 + z2| ≤ |z1| + |z2|.

5. Show that d(z 1 , z 2 ) := |z1 − z2| defines a metric on C, and it is

a complete metric.
8 Complex Plane

6. Show that |z1 − z2| ≥ |z1| − |z2| for all z 1 , z 2 ∈C.

7. Suppose α, β, γ are nonzero complex numbers such that |α| =

|β| = |γ|.
1 1 1
Show that α + β + γ = 0 ⇐⇒ + + = 0.
α β γ

8. Suppose z 1 , z 2 , z 3 are vertices of an equilateral triangle. Show

that z 21 + z 22 + z 23 + = z 1 z 2 + z 2 z 3 + z 3 z 1 .

9. Show that the equation of a straight line using complex variable

z is given by α¯z+ αz¯ + γ = 0 for some α ∈ C and γ ∈R.
1+ z
10. If |z| = 1 and z =ƒ 1, then show that = ib for some b ∈R.
1−z

11. For n ∈ N, derive a formula for the n t h root of a complex

number z using its polar representation.

12. Let S 2 be the unit sphere in R 3 with centre at the origin, i.e.,
S 2 := {(α, β, γ ) ∈ R 3 : α 2 + β 2 + γ 2 = 1}. Show that the
steriographic projection
. Σ
2x ,2x |z|
, −1
2
z := x + iy ›→
1 + |z|2 1 + |z|2 1 + |z|2

is a bijective continuous function from C onto S 2 \ {(1, 0, 0)}

with its inverse
α + iβ ,
(α, β, γ ) ›→
1 −γ
which is also continuous.

13. Show that the functions z ›→Re(z), z ›→Im(z), z ›→|z|

are continuous functions on C.
z
14. Show tha t lim does not exist.
z→0 |z|
2
Analytic Functions

In this chapter we do calculus of complex valued functions of a com-

plex variable.

2.1 Differentiation
Let f be a complex valued function defined on a set Ω ⊆ C.
Deftnition 2.1.1 Let z 0 be an oneerior point of Ω. Then f is said to
be differentiable at z 0 if
f (z) − f (z 0 )
lim
z→z0 z −z 0
exists, and in that case the above limit is called the derivative of f
at z 0 , denoted by f j (z 0 ). ♦
Thus the following are equivalent:
( i ) f is differentiable at z 0 ∈ Ω.
(ii) There exists c ∈ C such that for every ε > 0, there exists δ > 0
satisfying

f (z) − f (z 0)
z ∈ Ω, 0 < |z − z0| < δ =⇒ .. − c..< ε.
z −z 0
(iii) There exists c ∈ C such that
f (z) − f (z 0 ) − c(z − z 0 )
→0 as z → z0.
|z − z0|
The equivalence in (iii) above shows that if f is differentiable at
z 0 , then f (z) is approximately equal to f (z 0 ) − c(z − z 0 ) whenever z
is in some neighbourhood of z 0 , which we write as

f (z) c f (z 0 ) − c(z − z 0 )

9
10 Analytic Functions

whenever z is in some neighbourhood of z 0 .

The following theorem can be proved (exercise) using arguments
similar to the real case of real valued functions of a real variable.
Theorem 2.1.1 Let z 0 be an interior point of Ω ⊆ C. Then the
following holds.
(i) If f differentiable at z 0 ∈ Ω, then f is continuous at z 0 .
(ii) If f and g are differentiable at z 0 ∈ Ω, then f + g and f g are
differentiable at z 0 , and

( f +g) j (z 0 ) = f j (z 0 )+g(z 0 ), (fg) j (z 0 ) = f j (z 0 )g(z 0 )+f (z 0 )g j (z 0 ).

(iii) If f and g are differentiable at z 0 ∈ Ω and if g(z 0 ) ƒ = 0, then

f / g is differentiable at z 0 , and
. Σj
f g(z 0) f j(z 0) − gj(z 0) f (z0)
(z 0) = .
g [g(z0 )]2

(iv) If f is differentiable at z 0 ∈ Ω and g is differentiable in a

neighbourhood of f (z 0 ), then g ◦ f is differentiable at z 0 and
(g ◦ f ) j (z 0 ) = g j (f (z 0 )f j (z 0 ).

Now, let us write f (z) as u(z) + iv(z), where u(z) = Ref (z) and
v(z) = I mf (z). Recall that f is differentiable at z0 ∈ Ω if and only if
there exists c ∈C such that
R (z )
→ 0 as z → z0,
|z − z 0|
where R(z) = f (z) − f (z 0 ) − c(z − z 0 ). Writing

z = x + iy, z 0 = x 0 + iy 0 , c = a + ib,
we have
R(z) = f (z) − f (z 0 ) − c(z − z 0 )
= [u(z) − u(z 0 )] + i[v(z) − v(z 0 )] − (a + ib)[(x − x 0 ) + i(y − y0 )]
= [u(z) − u(z 0 ) − a(x − x 0 ) + b(y − y0 )]
+i[v(z) − v(z 0 ) − b(x − x 0 ) − a(y − y0 )]
= R 1 (z) + iR 2 (z),
Differentiation 11

where
R 1 (z) = u(z) − u(z 0 ) − [a(x − x 0 ) − b(y − y0 )],
R 2 (z) = v(z) − v(z 0 ) − [b(x − x 0 ) + a(y − y0 )].
T hus,
R (z ) R 1 (z) R 2 (z)
→ 0 ⇐⇒ →0 & →0
|z − z 0| |z − z 0 | |z − z 0|
if and only if u and v are differentiable as a functions of two real
variables at (x 0 , y 0 ), and

a = ∂u (x 0 , y 0 ), −b =
∂u
(x , y ),
∂x ∂y 0 0

b = ∂v (x 0 , y 0 ), a = ∂v (x 0 , y 0 ),
∂x ∂y
i.e.,
∂u ∂v ∂u (x 0 , y 0 ) = − ∂v (x 0 , y 0 ),
(x 0 , y 0 ) = (x , y ) &
∂x ∂y 0 0 ∂y ∂x
and in that case
∂u ∂v
f j (z 0 ) = a + ib = (x 0, y0) + i(x 0, y0)
∂x ∂x
∂v ∂u
= (x , y ) − i (x 0 , y 0 ).
∂y 0 0 ∂y
Thus, we have proved the following theorem.
Theorem 2.1.2 The function f is differentiable at z 0 ∈ Ω if and only
if its real part u and imaginary part v are differentiable at (x 0 , y 0 )
and u x , uy , v x , vy satisfy the equations
u x (z 0 ) = vy (z 0 ), uy (z 0 ) = −v x (z 0 ), (∗)

and in that case

f j (z 0 ) = u x (z 0 ) + iv x (z 0 ) = vy (z 0 ) − iuy (z 0 ).
Equations in (∗) are called the Cauchy-Riemann equations,
or in short CR-equations.
Now, recalling from a sufficient condition for differentiability of a
real valued function of two variables, we have the following sufficient
condition of differentiability of f at z 0 ∈Ω.
12 Analytic Functions

Theorem 2.1.3 Let f = u + iv and z 0 = x 0 + iy 0 ∈ Ω. Suppose

u x , uy , v x , vy exist in a neighbourhood of z 0 and are continuous at
(x 0 , y 0 ), and suppose they satisfy the CR-equations at (x 0 , y 0 ). Then
f is differentiable at z = x 0 + iy 0 , and

f j (z 0 ) = u x + iv x .

EXAMPLE 2.1.1 Let us find out points at which some of the

simple functions are differentiable.
(i) Let f (z) = x. In this case we have u(x, y) = x and v(x, y) = 0.
Hence, u x = 1, uy = 0, v x = 0 = vy at every point. Since u and v do
not satisfy CR-equations, f is not differentiable at any z ∈C.
(ii) Let f (z) = z¯ = x − iy. In this case we have u(x, y) = x and
v(x, y) = −y. Hence, u x = 1, uy = 0, v x = 0, vy = −1 at every
point. Again CR-equations are not satisfied at any point. Hence, f
is not differentiable a t any z ∈ C.
(iii)Let f (z) = |z|2 = x 2 + y 2 . In this case we have u(x, y) = x 2 + y2
and v(x, y) = 0. Hence, u x = 2x, uy = 2y, v x = 0 = vy at every
point. Thus, partial derivatives of u and v exist and are continuous
on R 2 , and CR-equations are satisfied only at (0, 0). Hence, f is
differentiable only at 0.
(iv)Let f (z) = z¯2 = (x 2 − y 2 ) − i2xy. In this case we have u(x, y)
= x 2 − y 2 and v(x, y) = 2xy. Hence, u x = 2x, uy = −2y, v x =
−2y, vy = −2x at every point. Thus, partial derivatives of u and v
exist and are continuous on R 2 , and CR-equations are satisfied only
at (0, 0). Hence, f is differentiable only at 0. Q

CR-equations in polar coordinates:

Let f = u + iv where u = Ref and v = Imf . For z = x + iy with
x = Rez and y = Imz, writing
√
r = x 2 + y2, θ = tan −1 (y/x), x ƒ = 0,

and considering u and v as we have the following:

∂u ∂u ∂r ∂u ∂θ
= + ,
∂x ∂r ∂ x ∂θ ∂ x
∂v ∂v ∂r ∂v ∂θ
= + .
∂x ∂r ∂ x ∂θ ∂ x
Differentiation 13

Now, r 2 = x 2 + y 2 and tan θ = y / x so that

∂r ∂r
2r = 2x, 2r = 2y,
∂x ∂y
∂θ y ∂θ 1
sec2 θ = − 2, secθ = ,
∂x x ∂y x
i.e.,
∂r x ∂r y
= , = ,
∂x r ∂y r
x 2 + y 2 ∂θ = − y , x 2 + y 2 ∂θ = 1 ,
x2 ∂ x x2 x2 ∂y x
i.e.,
∂r = x , ∂r = y , ∂θ y ∂θ = x .
= − 2,
∂x r ∂y r ∂x r ∂y r2
Now,
∂u ∂u ∂r ∂u ∂θ ∂u ∂u ∂r ∂u ∂θ
= + , = + .
∂x ∂r ∂ x ∂θ ∂ x ∂y ∂r ∂y ∂θ ∂y
T hus,
∂u = x ∂u − y ∂u , ∂u = y ∂u + x ∂u .
(1)
∂x r ∂r r 2 ∂θ ∂y r ∂r r 2 ∂θ
Similarly,
∂v = x ∂v − y ∂v , ∂v = y ∂v + x ∂v .
(2)
∂x r ∂r r 2 ∂θ ∂y r ∂r r 2 ∂θ
Recall that the CR-equations in Cartesian coordinates are
∂u = ∂v , ∂u = − ∂v .
∂x ∂y ∂y ∂x
Hence, (1) − (2) give
x ∂u y ∂u y ∂v x ∂v
− 2 = + 2 , (5)
r ∂r r ∂θ r ∂r r ∂θ
y ∂u + x ∂u = x ∂v − y ∂v .
(6)
r ∂r r 2 ∂θ r ∂r r 2 ∂θ
The equations (3) − (4) imply that

r ∂u = ∂v , ∂u = −r ∂v .
∂r ∂θ ∂θ ∂r
These are the CR-equations in polar coordinates.
14 Analytic Functions

2. Holomorphic or Analytic Functions

Deftnition 2.2.1 Let Ω ⊆C.

(i) A function f : Ω → C is said to be analytic at a point z0 ∈ Ω if

there exists r > 0 such that B(z 0 , r) ⊆ Ω and f is differentiable
at every point in B(z 0 , r).

(ii) A function f : Ω → C is said to be holomorphic or analytic

on Ω0 ⊆ Ω if f is analytic at every point in Ω0 .

• If f : Ω → C is analytic on Ω0 ⊆ Ω, then Ω0 is an open set.

• f : Ω → C is not analytic at a point z 0 ∈ Ω if and only if for

every r > 0, there exists ζ ∈ B(z 0 , r) ∩ Ω such that f is not
differentiable at ζ.

Deftnition 2.2.2 A complex valued function defined and analytic on

the entire complex plane is called an entire function. ♦

• f : Ω → C is analytic at a point z0 ∈ Ω if it is analytic on some

open set containing z 0 .

• f : Ω → C is analytic on Ω if and only if u := Re(f ) and v :=

Im(f ) have continuous first partial derivatives in Ω and they
satisfy the CR-equations at every point in Ω.

Remark 2.2.1 In the subject of complex analysis, it is very common

to say a function
f is analytic at a point z 0 ∈ C

to mean that f is defined in an open neigbourhood of z 0 and f is

analytic at z 0 .
Usually, a function is given in terms of certain expression, and in
that case, the domain of definition of f is taken to be the largest sub-
set of C in which the expression makes sense. For example, consider
the expression
1
f (z ) = .
z
Holomorphic or Analytic Functions 15

In this case, the domain of definition of f is taken to be Ω := C\{0},

and f is not analytic at z 0 = 0, since 0 ƒ∈Ω. The function

f (z) = z¯
is not analytic at any point in C, since f is defined on C, but f is
not differentiable at any z 0 ∈ C. ♦
Deftnition 2.2.3 A set Ω0 ⊆ C is called the domain ofanalyticity
of a function f if Ω0 is the largest open set in which f is analytic. ♦
For example, the domain of analyticity of f (z) = 1/z is C \ {0},
whereas the domain of analyticity of f (z) = z¯is ∅.
Deftnition 2.2.4 A point z 0 ∈ C is a singularity of an analytic
function f if z 0 is not in the domain of analyticity of f . ♦
Deftnition 2.2.5 A point z 0 ∈ C is an isolated singularity of
a function f if the domain of analyticity of f contains a deleted
neigbourhood of z 0 . ♦
EXAMPLE 2.2.1 1. For a 1 , a 1 , . . . , an in C, let

f (z) := a0 + a 1 z + . . . + a n z n , z ∈C.

Then f is an entire function.

2. The function z ›→ 1/(1 − z) is analytic on C \ {0}.
Σ ∞ n
3. The function f defined by f (z) = n= 0 z is analytic on the
unit disc Ω := { z ∈ C : |z| < 1}.
Q

2.2.1 Analytic extension

Deftnition 2.2.6 Suppose Ω is an open subset of C and f : Ω → C is
an analytic function. Let Ω˜be an open set such that Ω˜⊇ Ω. We say
that f has an analytic extension to Ω ˜if there exists an analytic
function g on Ω˜ which is an extension of f , i.e., g : Ω˜ → C is analytic
and g(z ) = f (z ) for all z ∈ Ω. ♦
EXAMPLE 2.2.2 1. Let Ω = { z ∈ C : |z| < 1} and f (z) =
1 , z ∈ Ω. Then f is analytic on Ω, and has extension to
1 −z
the open set Ω˜:= C\{1}. Clearly, in this case, g(z) = 1/(1−z)
for all z ∈Ω˜.
16 Analytic Functions

Σ ∞ n
2. The function f defined by f (z) = n= 0 z is analytic on the
unit disc Ω := { z ∈ C : |z| < 1}, and cannot be extended to a
bigger open set.
Q

2.2.2 Geometric representations

We know that a real valued function of a real variable can be repre-
sented by its graph, as a subset of the plane. Such a representation is
not possible for a complex valued function of a real variable, as we
require at least four real dimensions. But, we can represent them as
mappings of C into itself as which specify changes taking place for
certain figures such as straight lines and circles. Let consider a few
simple examples.
EXAMPLE 2.2.3 Consider the function
z ›→ az
for some nonzero a ∈ C. Let us consider the following cases.
Case (i) a ∈R and a > 0.

We see that this maps scales the figures in the plane - if a > 1,
then it takes a circle to a bigger circle, and if a < 1, then it takes a
circle to a smaller circle. More precisely, consider a circle

C: |z − z0| = r. (∗)

Writing ζ = az, we have

ζ
|z − z 0 | = r ⇐⇒ .. − z 0 .. = r ⇐⇒ |ζ − az |0 = ar.
a
Thus above function transforms a circle with centre z 0 and radius r
into a circle with centre az 0 and radius ar.
A straight line has the equation
L: α¯z+ αz¯ + γ = 0. (∗∗)

Indeed, for a, b, c ∈R,

. Σ . Σ
z + z¯ z −z¯
ax + by + c = a +b +c
2 2i
. Σ . Σ
a −ib a + ib
= z+ z¯+ c.
2 2
Holomorphic or Analytic Functions 17

Now, under the map z ›→ ζ := az with a > 0, we have

. Σ . ζ¯Σ
ζ
α¯z+αz¯+γ = 0 ⇐⇒ α¯ +α + γ = 0 ⇐⇒ α¯ζ +αζ ¯ +aγ = 0.
a a
Thus, the function z ›→ az transforms a straight line into a a straight
line.
Case (ii) a ∈ C and |a| = 1.
In this case a is of the form a = cos θ0 + i sin θ0 for some θ ƒ = 0.
Hence, by the map z ›→ az, a point z = r cos0 θ+ i sin θ0 is rotated by
an angle θ0 . Thus, circles and straight lines are mapped onto circles
and straight lines, respectively.
Case (iii) a ∈ C, a ƒ = 0.
Since
. a Σ
a = |a| ,
|a|
this function is a composition of the functions considered in Case (i)
and Case (ii). Q

Exercise 2.2.1 Explain the last statement in Case(ii) analytically

using the circle and straight line given in (∗) and (∗∗). a

EXAMPLE 2.2.4 Consider the function

z ›→ z + b

for some nonzero b ∈ C. In this case the circle in (∗) is mapped into
the circle
C˜ : |ζ − (z 0 − b)| = r,
i.e., the original circle is translated by −b, and the straight line in
(∗∗)is mapped into the straight line

L˜ : α¯z + αz¯+ (β¯ + α¯b+ γ ) = 0.

Note that for γ ∈ R, β¯+ α¯b+ γ is also in R. Q

EXAMPLE 2.2.5 Consider the function

z ›→ az + b
18 Analytic Functions

for some nonzero a, b ∈ C. This case is combination of the function

considered in the previous two examples, i..e, f (z) = g(h(z)), where
h(z) = az and g(z) = z + b. Therefore, circles and straight lines are
mapped into circles and straight lines, respectively. Q
Remark 2.2.2 We observe tha t under the under the map z ›→ az +b
with a ƒ = 0, the tangent at a point z 0 on a curve Γ is mapped onto the
tangent at the point ζ := f (z 0) on the curve Γ ˜:= f (Γ). Of course,
we have not defined so far what we mean by a curve in C. We shall
do this now. ♦

2.2.3 Curves in the complexplane

Deftnition 2.2.7 A curve in the complex plane is a complex valued
function γ defined on an interval I .
If I = [a, b], then the point z 1 := γ(a) is called the initialpoint
of γ and z 2 := γ(b) is called the terminal point of γ. ♦
If γ is a curve in C, then we shall identify it with its image

Γ γ := { γ ( t ) : t ∈I } .

If Γγ lies in an open set Ω, then we say that the curve γ is in Ω.

The direction of a curve γ is along the direction in which the
points on Γ vary as t increases on I .
EXAMPLE 2.2.6 Given z 1 , z 2 ∈ C, the line segment joining z 1 to
z 2 is a curve given by
γ(t) := (1 −t)z 1 + tz 2 , t ∈ [0, 1].

This curve has the same image as the one given by

. Σ . Σ
b− t t− a
γ(t) := z1 + z 2, t ∈ [a,b].
b− a b− a

EXAMPLE 2.2.7 (i) The curve defined by

γ(t) := cos t + i sin t, 0 ≤ t ≤ 2π

traces the unit circle (with centre 0) once in anti-clockwise direction.

Holomorphic or Analytic Functions 19

(ii) The curve defined by

γ(t) := cos t − i sin t, 0 ≤ t ≤ 2π
traces the unit circle (with centre 0) once in clockwise direction.
(iii) The curve defined by
γ(t) := cos t + i sin t, 0 ≤ t ≤ 2nπ
traces the unit circle (with centre 0) n-times in anticlockwise direc-
tion. Q
EXAMPLE 2.2.8 Using the definition of a line segment in Example
2.2.6, it can be seen that the following curve traces the boundary of
the square with vertices at 0, 1, 1 + i , i once in anticlockwise direction:

t, 0≤t ≤ 1,
1 + (t − 1)i, 1< t ≤2,
γ(t) := 3 − t + i, 2< t ≤3,
(4 − t)i, 3< t ≤4.
Q
Deftnition 2.2.8 A curve γ : I → C is said to be differentiable at
a point t 0 ∈ I if
γ(t) −γ(t 0 )
lim
t→t 0 t −t 0
exists, and in that case the above limit is called the derivative of γ
at t 0 , denoted by γ j (t 0 ). ♦
If γ j (t 0 ) exists and is nonzero, then it represents the direction of
the tangent vector to the curve at the point z 0 := γ(t 0 ). In this case,
the direction of the curve γ at t 0 is specified by
γ j (t 0 )
arg γ j (t 0 ) or by the unit vector .
|γ j(t0 )|

Deftnition 2.2.9 A curve γ : I → C is said to be regular at a point

t 0 ∈ I if γ is differentiable at t 0 and γ j (t 0 ) =ƒ 0.
If γ : I → C is regular at t 0 ∈ I , then abusing the terminology,
we may say tha t γ is regular at z 0 := γ (t 0 ). ♦
Deftnition 2.2.10 We say that curves γ 1 and γ 2 (or their images Γ1
and Γ 2 ) intersect at a point z 0 ∈ C if there are points t 1 , t 2 such
tha t γ 1 (t 1 ) = γ 2 (t 2 ) = z 0 . ♦
20 Analytic Functions

Remark 2.2.3 (i) In the above definition the curves γ 1 and γ 2 may
be defined on different intervals I 1 and I 2 .
(ii) As a particular case of the above definition, if we take γ 1 = γ 2 ,
the we can say that γ is self intersecting at z 0 . ♦
Deftnition 2.2.11 Suppose that curves γ 1 and γ 2 intersect at a point
z 0 ∈ C and regular at z 0 . Then the angle between γ 1 and γ 2 a t z0 is
defined as

Θz0 (γ 1 , γ 2 ) := arg γ 2 j (t 2 ) − arg γ 1 j (t 1 ),

where t 1 , t 2 are such tha t γ 1 (t 1 ) = γ 2 (t 2 ) = z 0 . ♦

Note that
Θz0 (γ 1 , γ 2 ) = arg γ 2 j (t 2 )γ 1 j (t 1 ).

We observe that if γ : I → C is a curve with its image as Γ, and

if f is a continuous complex valued function on Γ, then γ˜ : I → C
defined by
γ˜(t) = ( f ◦ γ)(t), t ∈I
is also a curve in C. Note that (Exercise) if f is defined on an open
set containing Γ, γ is differentiable at t 0 ∈ I, and f is differentiable
at z 0 := γ(t 0 ), then γ˜ is also differentiable at t 0 and

γ˜ j (t 0 ) = f j (z 0 )γ j (t 0 ).
If, in addition, if γ is regular at t 0 and f j (z 0 ) ƒ = 0, then we obtain

arg γ f j (t 0 ) = arg f j (z 0 ) + arg γ j (t 0 ).

Deftnition 2.2.12 Let f be a continuous complex valued function

defined on an open set Ω and z 0 ∈ Ω. Let γ 1 and γ 2 be curves
intersecting at z 0 and regular at z 0 . Then f is said to preserve
angle between γ 1 and γ 2 at z 0 if γ˜1 := f ◦ γ 1 and γ˜2 := f ◦ γ 2 are
regular at z̃ 0 := f (z 0 ) and Θ z̃0 (γ̃ 1 , γ̃ 2 ) = Θ z0 (γ 1 , γ 2 ). ♦

Thus we have the following theorem.

Theorem 2.2.1 Let f be defined on an open set Ω and differentiable

at a point z 0 ∈ Ω and f j (z 0 ) ƒ = 0. Then f preserves angle between
curves which are regular at z 0 and intersecting at z 0 .
Holomorphic or Analytic Functions 21

In fact, with some additional conditions, we have the converse of

the above theorem as well. We shall state it without proof. For a
proof of this, see Sarason1

Theorem 2.2.2 Let f := u + iv be defined on an open set Ω such

that u and v are differentiable at a point z 0 ∈ Ω. If f preserves angle
between curves which are regular at z 0 and intersecting at z 0 , then f
is differentiable at z 0 and f j (z 0 ) ƒ = 0.

Deftnition 2.2.13 A function f defined on an open set Ω is said to

be a conformal map at z0 ∈ Ω if it preserves angles between any
two curves in Ω intersecting at z 0 and regular at z 0 . ♦
Thus, a holomorphic function is conformal at every point in its
domain of definition.
Now, let f be defined on an open set Ω. Let

u = Ref, v = Imf.

Considering f as a function of two real variables and assuming the

quantities involved are well-defined on Ω, we can write
∂f ∂u ∂v ∂f ∂u ∂v
= + i and = + i .
∂x ∂x ∂x ∂y ∂y ∂y
Hence, . Σ . Σ
∂f ∂f ∂u ∂v ∂v ∂u
+ i = − +i + .
∂x ∂y ∂ x ∂y ∂x ∂y
Thus, u and v satisfy CR-equations if and only if
∂f ∂f
+ i = 0
∂x ∂y

Notation:
. Σ ∂ 1. ∂ ∂ Σ
∂ 1 ∂ ∂
:= −i , := + i .
∂z 2 ∂x ∂y ∂z¯ 2 ∂x ∂y

∂f
Thus, u and v satisfy CR-equations if and only if = 0.
∂z¯
1
D. Sarason, Notes on Complex Function Theory, Hindustan Book Agency,
New Delhi, 1994.
22 Analytic Functions

The following would justify the introduction of the above nota-

∂ ∂
tions and : Recall that, for z = x + iy,
∂z ∂z
z + z¯ z −z¯
x= , y=
2 2i
so that
∂x 1 ∂x 1 ∂y −i ∂y i
= , = , = , = .
∂z 2 ∂z¯ 2 ∂z 2 ∂z¯ 2
T hus, . Σ
∂f ∂f ∂x ∂f ∂y 1 ∂f ∂f
= + = −i ,
∂z ∂x ∂z ∂y ∂z 2 ∂x ∂y
. Σ
∂f ∂ f ∂x ∂ f ∂y 1 ∂f ∂f
= + = +i .
∂z¯ ∂x ∂z¯ ∂y ∂z¯ 2 ∂x ∂y

We can also define higher partial derivatives:

. Σ . Σ
∂2f ∂ ∂f ∂2f ∂ ∂f
= , =
∂x 2 ∂x ∂x ∂y 2 ∂y ∂y
. Σ . Σ
∂2f ∂ ∂f ∂2f ∂ ∂f
= , = .
∂x∂ y ∂x ∂y ∂y∂x ∂y ∂x

Deftnition 2.2.14 Let f be defined on an open set Ω ⊆ C. Then

( i ) f is said to be of class C 1 if first partial derivatives of f exist and
are continuous.
( i i ) f is said to be of class C 2 if second partial derivatives of f
exist and are continuous.
( i i i ) f is said to be harmonic if f is of class C 2 and

∂2f ∂2f
O f := + = 0.
∂x 2 ∂y 2

The above equation is called the Laplace equation, and the oper-
ator O is called the Laplacian. ♦
The following can be easily verified:
• f holomorphic on Ω and is of class of C 2 =⇒ f is harmonic.
• f holomorphic on Ω =⇒ u and v are harmonic, i.e.,

Ou = 0 = Ov.
Fractional linear transformations 23

∂2f
• f harmonic on Ω ⇐⇒ = 0. In fact,
∂z¯∂z
. Σ
∂ 2 f = 1 ∂ 2 f + ∂ 2f .
∂z¯∂z 4 ∂x 2 ∂y 2
Exercise 2.2.2 Prove the above three statements. a
Deftnition 2.2.15 Let u and v be real valued functions of class C 2
defined on Ω (considered as a subset of R 2 ). Then v is said to be a
harmonic conjugate of u if f := u + iv is holomorphic on Ω. ♦
We observe the following:
• If u and v are of class C 2 on an open set Ω, then then v is a
harmonic conjugate of u if and only they and satisfy CR-equations,
and in that case both u and v are harmonic on Ω.
Exercise 2.2.3 Prove that v is a harmonic conjugate of u if and
only −u is a harmonic conjugate of v. a

3. Fractional linear transformations

1. The map z ›→1/z
Recall that, for nonzero complex numbers a and b, the functions

z ›→ az, z ›→ z + b

map circles onto circles and straight lines onto straight lines. Now let
us look at another simple function
1
z ›→ , z ƒ = 0.
z
Let us see the images of circles and straight lines under this function:
Consider the image of the circle:

C: |z − z0| = r.

Since
2 2 2 2
|z − z0| = (z¯− z¯0)(z − z 0 ) = |z| − (z¯0z + z0z¯) + (|z0| − r ),

the above equation can also be written as

2 2 2
|z| − (z¯0z + z0z¯) + ρ = 0, ρ := |z0| − r . (∗)
24 Analytic Functions

To find its image under z ›→1/z, let us write ζ = 1/z. Thus, the
image is given by the equation

1 − (z 0 ζ + z¯0ζ¯) + ρ|ζ|2 = 0.

Thus, if ρ = 0, i.e., if |z0| = r, i.e., if the circle passes through 0,

then the image is given by
1 − (z 0 ζ + z¯0ζ¯)= 0

which is an equation of a straight line.

Now, assume that ρ ƒ = 0 i.e., |z0| =ƒr. Then the equation of the
image takes the form
.
z 0ζ z¯ζ ¯Σ 1
|ζ|2− +
0
+ = 0. (∗∗)
ρ ρ ρ
Note that
.
. z0 . 2
. z0 . 2 Σ . Σ2
1 1 . z0 .2 r
= . . − . . − = . . − .
ρ ρ ρ ρ ρ ρ

Comparing (∗∗)with (∗), it follows that (∗∗)represents a circle with

centre at z¯0/ρ and radius r/ρ. Thus, under the function z ›→ 1/z,
(i) circles passing trough 0 are mapped onto straight lines, and
(ii) circles not passing trough 0 are mapped onto circles.
Next, consider a straight line

L: α¯z + αz¯ + γ = 0.

This is mapped onto

. Σ . Σ
1 1
L˜ : α¯ +α +γ = 0
ζ ζ¯

⇐⇒ α¯ζ¯+ αζ + γ|ζ|2 = 0.
If γ = 0, i.e., if the line L passes through 0, then its image also a
straight line passing through 0. If γ ƒ = 0, i.e., if L does not pass
through 0, then
. ¯ Σ
α¯ζ αζ
L˜ : |ζ|2+ + = 0
γ γ
Fractional linear transformations 25

which is an equation of a circle with centre at −α¯/γ and radius |α|/γ,

which also pass through 0. Thus, under the function z ›→1/z,
(i) straight lines passing through 0 are mapped onto straight lines
trough 0, and
(ii) straight lines not passing through 0 are mapped onto circles
passing through 0.
Thus, in general, under the function z ›→1/z,
• circles and straight lines are are mapped onto either circles or
straight lines.

2.3.2 Extended plane

Note that the function z ›→ 1/z is not defined at z = 0. However, we
know that
lim 1 = ∞,
|z|→0 |z|

i.e., for every M > 0, there exists δ > 0 such that

1
|z| < δ =⇒ > M.
|z|

We shall write this fact by

1
lim = ∞.
z→0 z
In the above, the ∞ is just a symbol which correspond to the north
pole in steriographic projection. Let us extend the complex plane C
by incorporating ∞, i.e., let us consider

C˜ := C ∪ { ∞ } .

Deftnition 2.3.1 The set C ˜:= C ∪{ ∞ } is called the extended

complex plane. ♦
1
In view of the above definition, we say that lim = ∞ in the
z→0 z
extended complex plane C.
˜
Since, we have
1
lim = 0,
|z|→∞ |z|
26 Analytic Functions

i.e., for every ε > 0, there exists M > 0 such that

1
|z| > M =⇒ < ε,
|z|

we also write
1
lim
= 0
zz→∞

in the extended complex plane. Thus, for the functions

1
f (z) := az, g(z) = z + b, h(z) =
z
for nonzero complex numbers a and b, we can write

lim f (z) = ∞, lim g(z) = ∞,

z→∞ z→∞

lim h(z) = 0, lim h(z) = ∞.

z→∞ z→0

Now, defining

f (∞) = ∞, g(∞) = ∞, h(∞) = 0, h(0) = ∞,

we can say that f , g, h are defined on the extended complex plane C.

˜

2.3.3 Fractional linear transformations

Now, for complex numbers a, b, c, d, consider the function

az + b
ϕ : z ›→ ϕ(z) := .
cz + d

Since,
. Σ . Σ
az + b a cz + bc/a a cz + d − [d −bc/a]
= = .
cz + d c cz + d c cz + d

T hus, . Σ. Σ
az + b a 1 ad −bc
= − .
cz + d c c cz + d
Thus, the function ϕ can be thought of as compositions of the func-
tions f , g, h we can write the function. Note that if ad − bc = 0,
then ϕ is a constant function. To avoid this case, we shall assume
that ad − bc ƒ = 0.
Fractional linear transformations 27

Deftnition 2.3.2 For complex numbers a, b, c, d with ad − bc ƒ = 0,

the function
az + b ˜
ϕ(z) = , z ∈ C,
cz + d
is called a fractional linear transformation or a linear frac-
tional transformation or a Möbius transformation. ♦

We observe that
a + b/z a
ϕ(∞) = lim ϕ(z) = lim = .
z→ ∞ z→∞ c + d/z c

Thus,
ϕ(∞) = ∞ ⇐⇒ c = 0.
Since ϕ is a composition of the functions z ›→ az, z ›→ z + b and
z ›→ 1/z, we can infer that under ϕ, the family of circles and straight
lines in C are mapped onto the family of circles and straight lines.
It can be easily seen that compositions of a finite number of frac-
tional linear transformations is a fractional linear transformation.

Exercise 2.3.1 Suppose ϕ 1 and ϕ 2 are fractional linear transforma-

tions. Prove that ϕ 1 ◦ ϕ 2 and ϕ 2 ◦ ϕ 1 are fractional linear transfor-
mations. a

Exercise 2.3.2 Consider a fractional linear transformation ϕ given

by
az + b ˜
ϕ(z) = , z ∈ C.
cz + d
(i) Show that ϕ is one-one and onto and its inverse is given by

−dz + b
ϕ −1 (z) = , z ∈ C˜.
cz − a
(ii) Show that ϕ is differentiable at every z ∈ C and its derivative is
given by
ad − bc
ϕ j (z) = , z ∈C.
(cz + d)2
a

Deftnition 2.3.3 A point z ∈0C is˜ said to be a ftxed point of a

function f : C˜ → C˜ if f (z 0 ) = z 0 . ♦
28 Analytic Functions

Consider a fractional linear transformation ϕ given by

az + b
ϕ(z) = , z ∈ C˜.
cz + d
We have already observed that ϕ(∞) = ∞ if and only if c = 0. Thus
∞ is a fixed point of ϕ if and only if c = 0. Suppose c ƒ = 0. Then we
have
ϕ(z) = z ⇐⇒ cz 2 + (d − a)z − b = 0.
Since the above equation has either two distinct roots or one repeated
root, we can say that
Theorem 2.3.1 A fractional linear transformation which is not an
identity function has either one or two fixedpoints.
In particular, the identity function is the only fractional linear
transformation having more than two distinct fixed points.
By the above theorem we can identity a fractional linear trans-
formation by requiring to map three distinct points z 1 , z 2 , z3 to three
distinct points w1 , w2 , w3 , respeoctively. This is done in the following
theorem.
Theorem 2.3.2 Given distinct points z 1 , z 2 , z 3 and distinct points
w 1, w 2, w 3in the plane C,˜ there exists a unique fractional linear trans-
formation ϕ such that ϕ(z j ) = wj for j ∈ {1, 2,3}.
Proof. First let us settle the uniqueness part. Suppose ϕ 1 and ϕ 2
are fractional linear transformations such tha t

ϕ(z j ) = wj = ϕ 2 (z j ), j ∈ {1, 2,3}.

Then the fractional linear transformation ψ := ϕ−21 ◦ ϕ 1 satisfies ψ(z j

) = w j for j ∈ {1, 2, 3}. Hence, ψ is the identity transformation, so
that ϕ − 1 = ϕ 2 .
Now, the question of existence. Let ϕ 1 and ϕ 2 be defined by
(z − z1 )(z 2 − z 3 ) (z − w1 )(w 2− w 3)
ϕ 1(z) = , ϕ 2(z) = .
(z − z 3 )(z 2 − z 1 ) (z − w3 )(w2 −w1 )
If none of the points z 1 , z 2 , z 3 , w1 , w2 , w3 is ∞, then we see that

ϕ 1 (z 1 ) = 0, ϕ 1 (z 2 ) = 1, ϕ 1 (z 3 ) = ∞,

ϕ 2 (w1 ) = 0, ϕ 2 (w2 ) = 1, ϕ 2 (w3 ) = ∞.

Fractional linear transformations 29

Hence, the transformation ϕ := ϕ−21 ϕ 1 satisfies the requirements. In

case one of the points z 1 , z 2 , z 3 is ∞, then we define ϕ 1 as follows:
(z − ζ)(z 2 − z 3 ) z 2− z 3
1 ∞, then ϕ (z)
(i) If z = 1 = lim = .
ζ→∞ (z − z 3 )(z 2 − ζ) z −z 3
(z − z1)(ζ − z 3 ) z −z 1
2 ∞, then ϕ (z)
(ii) If z = 1 = lim = .
ζ→∞ (z − z 3 )(ζ − z 1 ) z −z 3
(z − z1 )(z 2 − ζ) z −z 1
3 ∞, then ϕ (z)
(iii) If z = 1 := lim = .
ζ→∞(z − ζ)(z 2 − z 1 ) z2 − z1
Similarly, if one of the points w1 , w2 , w3 is ∞, then we define ϕ 2
as in the case of ϕ 1 by replacing z j by w j . Thus, in these cases also,
the transformation ϕ := ϕ−21 ϕ 1 maps z j onto wj for j ∈{1, 2, 3}.

In the above theorem, if we write w = ϕ(z), then it can be seen

that
(w − w1 )(w2 − w3 ) (z − z 1 )(z 2 − z 3 )
=
(w − w3 )(w2 − w1 ) (z − z 3 )(z 2 −z 1 )
with the convention that if any of the points z 1 , z 2 , z 3 , w1 , w2 , w3 is
∞, the limit is taken for the corresponding expression.

2.3.4 Image of inverse points

Deftnition 2.3.4 Points p and q in the extended complex plane C˜are
said to be inverse points with respect to a circle { z ∈ C : |z −z0| = r }
if
r2
z0 − p = ,
z¯0− q¯
and in tha t case we say that p is inverse to q (respectively q is inverse
to p). ♦

Thus, points p and q are inverse points with respect to a circle

{ z ∈ C : |z − z0| = r } if and only if
(i) they line on a ray emanating from 0 and have same arguments,
and
(ii) satisfies |z0 − p| |z0 − q| = r 2 whenever p and q are different from
z 0 , and in case one of them is z 0 , then the other is ∞.
We are going to show that

Theorem 2.3.3 Under a bilinear transformation, inverse points are

mapped onto inverse points.
30 Analytic Functions

First suppose that the bilinear transformation is linear, i..e, it is

given by
ϕ(z) = az + b
for some nonzero a, b ∈ C. Let p and q be inverse points with respect
to a circle { z ∈ C : |z − z0| = r}. Then, taking α = ap +b, β = aq +b,
we have p = (α − b)/a and q = (β − b)/a so that

r2 r2
z0 − p = ⇐⇒ z 0− (α − b)/a =
z¯0− q¯ ¯ .
z¯0− (β ¯−b)/a¯
Thus, α and β are inverse points with respect to the circle with
centre at −b/a and radius r. Clearly, p = ∞ (resp. q = ∞) if and
only if α = ∞ (resp. β = ∞).
Next we consider the situation under the transformation
1
ϕ (z ) = .
z
Suppose p is inverse to q with respect to the circle, |z − z0| = r.
Case (i): Assume first that both p and q are different from z 0 .
Then we have
r2
z 0− p = ⇐⇒ (z 0 − p)(z¯0 − q¯)= r 2
z¯0− q¯
⇐⇒
2 2
(|z0| − r ) − (z¯0p + z0q¯) + pq¯= 0.

Now, taking α = 1/p, β = 1/q and ρ = |z0|2 − r 2 , we have

r2
z 0− p = ⇐⇒ (|z0|2− r 2) − (z¯0p + z0q¯) + pq¯= 0
z¯0− q¯
z¯0 z 0 1
⇐⇒ ρ − ( + ¯) + ¯ = 0
α β αβ
. Σ. Σ
z¯0 z0
⇐⇒ −α − β = R2,
¯
ρ ρ
where
. z¯0 . 2 1 r2
R2 = . . − = 2.
ρ ρ ρ
Thus α := 1/p is inverse to β := 1/q with respect to the circle,
|z − w0| = R, where w0 = z¯0/ρ and R = r/|ρ|.
Case (ii): Suppose p = z 0 so that q = ∞.
Fractional linear transformations 31

Clearly, if z 0 = 0, then α := 1/p = ∞ is inverse to β := 1/q = 0

with respect to the circle, |z − w0| = R, where w0 = 0 and R =
r/|ρ| = 1/r.
Next assume that z 0 ƒ = 0. In this case, we have to show that
α = 1/z 0 and β = 0 are inverse points. Note tha t
. Σ. Σ . Σ. Σ
z¯0 z0 z¯0 1 z0 |z0|2 −ρ = r 2 .
−α −β =
¯ − =
ρ ρ ρ z0 ρ ρ2 ρ2
Thus, the point α = 1/z 0 is inverse to β := 0 with respect to the
circle, |z − w0| = R, where w0 = z¯0/ρ and R = r/|ρ|.
Case (iii): Suppose p = z 0 so that q = ∞: This case is similar to last
case.
Thus, we have proved Theorem 2.3.3.
Theorem 2.3.3 helps in finding a general bilinear transformation
that maps the open unit disc D = { z ∈ C : |z| < 1} onto itself.
Suppose the the required transformation is given by
az + b
ϕ(z) = , ad − bc ƒ = 0.
cz + d
Suppose z 0 ∈ D be such that ϕ(z 0 ) = 0, i.e.,
az 0 + b
= 0.
cz0 + d
Hence, z 0 = −b/a. Also we have
. a Σ z + b/a . a Σ z −z
0
ϕ (z ) = = .
c z + d/a c z + d/a
Note that z 1 := 1/z¯0 = −a¯/¯b is the inverse point of z 0 with respect to
the circle S := { z ∈ C : |z| = 1}. Since inverse points are mapped
onto the respective inverse points, z 1 must be mapped onto ∞, as 0
and ∞ are inverse points. Thus, z 1 = −d/a, and hence, d/a = −z¯0
so tha t . az¯ Σ z −z
0 0
ϕ(z) = .
c zz¯0 − 1
Again, since
1 −z 0
|ϕ (1)| = 1 and .. . = 1,
1 −z¯0 .
we have |az¯0/c| = 1. Thus, we obtain
. Σ
z −z 0
ϕ(z) = α with |α| = 1, |z0| < 1.
zz¯0 − 1
32 Analytic Functions

4. Problems
1. Suppose u is a real valued function defined on an open set Ω ⊆
R 2 . Let (x 0 , y 0 ) ∈ Ω.
(i) When do you say that u has partial derivatives u x and uy
at (x 0 , y 0 )?
(ii) When do you say that u is differentiable at (x 0 , y 0 )?
(iii) What is gradient of u at (x 0 , y 0 )?
(iv) What is the relation between gradient and derivative of u?

1. Show tha t a function f is differentiable at z 0 ∈ Ω if and only if

its real part u and imaginary part v are differentiable at (x 0 , y 0 )
and u x , uy , v x , vy satisfy the equations

u x (z 0 ) = vy (z 0 ), uy (z 0 ) = −v x (z 0 ),

and in that case

f j (z 0 ) = u x (z 0 ) + iv x (z 0 ) = vy (z 0 ) − iuy (z 0 ).

3. Find points at which the following functions are differentiable:

( i ) f (z) = x, (ii) f (z) = z¯= x − iy,

(iii) f (z) = |z|2, (iv) f (z) = z¯2.

4. Find points at which the functions in the last problem satisfy

CR-equations.

5. Prove that the CR-equations in polar coordinates are qru r =

vθ , uθ = −rv r .

6. Suppose f is holomorphic on an open set Ω. Prove that if f

satisfies any of the following conditions, then f is a constant
function.
(i)f j is constant on Ω,
( i i ) f is real valued on Ω,
(iii) |f | is constant on Ω,
(iv) arg(f ) is constant on Ω.
Problems 33

7. Suppose f is holomorphic on an open set Ω. Prove that the

function z ›→ g(z) = f (z¯) is holomorphic in Ω∗:= {z¯ : z ∈t t } .

8. (i) Show that the equation of a straight line is given by

α¯z + αz¯+ γ = 0

for some α, β in C and γ ∈ C.

(ii) Show that the above line passes through 0 if and only if
γ = 0.

9. Show that the equation of a circle with centre at z 0 and radius

r > 0 is given by
2 2 2
|z| − (z¯0z + z0z¯) + |z0| − r = 0

10. Prove the following:

(i)For nonzero a ∈ C, the function z ›→ az maps a straight line
into a straight line and a circle into a circle.
(ii)For nonzero b ∈ C, the function z ›→ z + b maps a straight
line into a straight line and a circle into a circle.
11. For nonzero a, b ∈ C, the function z ›→ az + b maps a straight
line into a straight line and a circle into a circle - Why?

12. Given a curve γ : [0, 1] → C, let γ˜ : [a, b] → C be defined by

γ˜(t) = γ(a + (b − a)t).

If Γ and Γ˜are the images of γ and γ˜ respectively, then show

that Γ and Γ˜ are homeomorphic.
13. Find the points at which the curve γ : [0, 4] → C defined in the
following are not regular. Justify your answer:
t, 0≤t ≤ 1,
1 + (t − 1)i, 1< t ≤2,
γ(t) :=
3 − t + i, 2< t ≤3,
(4 −t)i, 3< t ≤4.

14. Let f be defined on an open set Ω and differentiable at a point

z 0 ∈ Ω and f j (z 0 ) ƒ = 0. Then prove that f preserves angle
between curves which are regular at z 0 and intersecting at z 0 .
34 Analytic Functions

∂f ∂f
15. Define and , and show that the real and imaginary parts
∂x ∂y
of f satisfy the CR-equations if and only if

∂f ∂f
+ i = 0
∂x ∂y

∂f ∂f
16. Define the operators and and show tha t the real and
∂z ∂z¯
imaginary parts of f satisfy the CR-equations if and only if
∂f
= 0.
∂z¯
17. Show that . Σ
∂2f = 1 ∂ 2 f + ∂ 2f
,
∂z¯∂z 4 ∂x 2 ∂y 2
∂2f
and deduce that f harmonic on Ω ⇐⇒ = 0.
∂z¯∂z
18. Prove that v is a harmonic conjugate of u if and only if −u is a
harmonic conjugate of v.

19. Prove tha t v1 and v2 are harmonic conjugates of u if and only

v1 − v2 is a constant.

20. Prove that if u is a real valued harmonic function on an open set

Ω, then any two harmonic conjugates of u differ by a constant.

21. Prove tha t if u is a real valued on an open set Ω such that both
u and u 2 are harmonic on Ω, then u is a constant function.

22. Prove that if u and v are harmonic functions on an open set Ω

such that v is a harmonic conjugate of u, then uv and u 2 − v 2
are harmonic.

23. Prove that the Laplace equation Ou = 0 can be written in

polar coordinates as r 2 u rr + ru r + uθτ = 0.

24. Prove that if u is a real valued harmonic function on an open

∂u
set Ω, then is holomorphic on Ω.
∂z
25. Suppose ϕ 1 and ϕ 2 are fractional linear transformations. Prove
that ϕ 1 ◦ ϕ 2 and ϕ 2 ◦ ϕ 1 are fractional linear transformations.
Problems 35

26. Consider a fractional linear transformation ϕ given by

az + b
ϕ(z) = , z ∈ C˜.
cz + d
(i) Show that ϕ is one-one and onto and its inverse is given by

−dz + b
ϕ −1 (z) = , z ∈ C˜.
cz − a

(ii)Show that ϕ is differentiable at every z ∈ C and its deriva-

tive is given by
ad − bc
ϕ j (z) = , z ∈C.
(cz + d)2

27. Show that the set of all fractional linear transformations is in

one-one correspondence with the set of all 2 × 2 nonsingular
matrices with complex entries.

28. Let F be the set of all fractional linear transformations. Define

a binary operation on F so that F becomes a group.

29. Show a fractional linear transformation maps every circle and

straight line in C onto either a circle or a straight line.

30. Prove that the identity function is the only fractional linear
transformation having more than two distinct fixed points.

31. Given distinct points z 1 , z 2 , z 3 and distinct points w1 , w2 , w3

in the plane C,
˜ show that the fractional linear transformation
w = ϕ(z) defined by

(w − w1 )(w2 − w3 ) (z − z 1 )(z 2 − z 3 )
=
(w − w3 )(w2 − w1 ) (z − z 3 )(z 2 −z 1 )

map z 1 , z 2 , z 3 onto w1 , w2 , w3 , respectively.

32. Find the fractional linear transformation ϕ that maps −1, 0, i

onto the points 0, 1, −i, respectively.
What is the image of the circle passing through −1, 0, i? A
circle or a straight line? Why?
36 Analytic Functions

33. Find the fractional linear transformation ϕ that maps 0, 1, ∞

onto the points 1, i, −1, respectively.
What is the image of the real axis under this transformation?
Why?

34. Find the fractional linear transformation ϕ that maps 1, i, −1

onto the points i, 0, −i, respectively.
What is the image of the unit circle (with centre at 0) under
this transformation? Why?

35. Suppose
az + b ˜
ϕ(z) = , z ∈ C.
cz + d
is the fractional linear transformation that maps the real axis
onto the unit circle (with centre 0). Show that

|a| = |c| =ƒ 0, |b| = |d| =ƒ 0.

36. If the fractional linear transformation in the last problem maps

the upper half plane onto the open unit disk, then show tha t
it is of the form z −z 0
ϕ(z) = α
z − z¯0
for some α, z 0 in C such that |α| = 1 and Im(z 0 ) > 0.
3
Elementary Functions

In this chapter we define the complex analogues of real elementary

functions, namely, the functions such as polynomials, rational func-
tions, exponential functions, and the functions obtained from these
functions by applying the operations of addition, substraction, multi-
plications, divisions and compositions. We are already familiar with
polynomials

p(z) = a 0 z n + a 1 z n−1 + . . . + a n−1 z + an

and rational functions

a 0 z n + a 1 z n−1 + . . . + a n−1 z + an
f (z) = .
b0 z m + b1 z m−1 + . . . + bm−1 z + bm
Note that the above two types functions are holomorphic wherever
they are defined. Now we shall consider exponential function and
other elementary functions associated with it.

3.1 Exponential Function

Recall that the function exp(x), denoted by e x for real x is defined
by
Σ∞ x n
x
e := ,
n!
n= 1
and we know that, it satisfies the relation

e x 1 + x 2 = ex 1 ex 2 , x 1 , x 2 ∈R.

In view of this we may define

Σ∞ (iy)n
iy (∗)
e := ,
n= 1
n!

37
38 Elementary Functions

for real y. Note tha t the series in (∗) converge absolutely for every
y ∈ R so that

Σ∞ (iy) 2n Σ∞ (iy) 2n+ 1

eiy = +
(2n)! (2n + 1)!
n= 0 n= 0
∞
Σ (−1)ny2n Σ∞ (−1)ny2 n+ 1
= +i
(2n)! (2n + 1)!
n= 0 n= 0
= cos y + i sin y.

In view of the above we may define ez for z = x + iy with x, y ∈ R

as
ez := ex(cos y + i sin y).
We note that the function z ›→ez satisfies
• e z 1 +z 2 = ez1 ez2 for every z 1 , z 2 ∈C,
z+ 2 π i z
•e = e for every z ∈ C, and
• it is holomorphic on the entire complex plane.
Further, we observe that for x = Re(z) and y = Im(z),
z x
• |e | = e and arg(z) = y.
The last property shows that the function z ›→ ez maps straight
lines parallel to y-axis onto concentric circles with centre 0, and
straight lines parallel to y-axis onto rays emanating from 0. In fact,
• the strip { z ∈ C : −π ≤ Im(z) < π } is mapped onto the entire
complex plane.

Exercise 3.1.1 Given θ1 and θ2 ∈ [0, 2π) with θ1 < θ 2, , describe the
image of the strip { z ∈ C : θ1 ≤ Im(z) < θ 2 } under the map z ›→ ez .
a

Exercise 3.1.2 Given r, R0 with 0 < r < R, describe the image of

the strip { z ∈ C : r ≤ Re(z ) < R } under the map z ›→ ez . a

Exercise 3.1.3 For the function f (z) = ez , describe the curves

|f (z)| = constant, arg(f (z)) = constant.

a
Hyperbolic and Trigonometric Functions 39

Exercise 3.1.4 If f is is holomorphic on C satisfying f j (z) = f (z),

and g is defined by g(z) = e−z f (z), then show that g j (z) = 0, so that
f (z ) = cez for some constant c ∈ C. a

Exercise 3.1.5 Find the most general form of a holomorphic f on

C satisfying f j (z) = cf (z) for some constant c ∈ C. a
Having defined exponential function, we move on to define hy-
perbolic and trigonometric functions.

3.2 Hyperbolic and Trigonometric Functions

For z ∈C, we define hyperbolic and trigonometric functions.
Hyperbolic functions:
ez + e−z ez −e−z
sinh z = , cosh z = ,
2 2
sinh z cosh z
tanh z = , coth z = ,
cosh z sinh z
1 1
sechz = , cosechz = .
cosh z sinh z
Trigonometric functions:

eiz + e−iz eiz − e−iz

sin z = , cos z = ,
2 2i
sin z cos z
tan z = , cot z = ,
cos z sinz
1 1
secz = , cosecz = .
cos z sin z
One may observe that
sinh z
sin z = , cos z = cosh iz.
i
Exercise 3.2.1 Derive the identities:
(i) cosh(z1 + z 2 ) = cosh z 1 cosh z 2 + sinh z 1 sinh z 2 .
(ii) sinh(z1 + z 2 ) = sinh z 1 cosh z 2 + cosh z 1 sinh z 2 .
(iii) cos z = cos x cosh y − i sin x sinh y.
(iv) sin z = sin x cosh y + i cos x sinh y.
40 Elementary Functions

2 2 2
(v) |cos z| = cos x + sinh y.
2 2 2
(vi) |sin z| = sin x + sinh y. a
Exercise 3.2.2 Show that
cos(z + 2π) = cos z, sin(z + 2π) = sin(z)
for all z ∈C. a
Exercise 3.2.3 Find zeros of sin z and cosz. a
Exercise 3.2.4 Find all roots of the equation cos z = 1. a

3. Logarithms
Deftnition 3.3.1 Logarithm of a complex number a is a complex
number b such that eb = a, and in that case we write b = log a. ♦
Thus,
• a complex number b is a logarithm of a complex number a if and
only if b is a zero of the function f defined by f (z) = ez − a.
Clearly,
• if b is a logarithm of a, then b + 2nπi is also a logarithm of a
for every n ∈ Z.
Observe that if b = log a, then
b Re(b)
|a| = |e | = e and arg(a) = Im(b).
Hence,
log a = ln |a| + i arg(a).
The value of log a corresponding to the principal value of arg a is
denoted by Log a, i.e.,
Log a = ln |a| + iArg a.
Exercise 3.3.1 Show that, for n ∈ Z,
(i) 2nπi = log 1.
(ii) (2n + 1)πi = log(−1).
. Σ
(iii) 2n + 12 πi = log i. a
Exercise 3.3.2 Find all values of
(i) cosh(log 2),
(ii) log(log i). a
Exercise 3.3.3 Does the relation log a1 a2 = log a1 + log a2 hold for
all nonzero a1 , a2 in C? a
Branches of arg(z) andlog(z) 41

3.4 Branches of arg(z) and log(z)

Recall that
z ›→ arg(z) and z ›→ log(z)
are “multi-valued mappings”. So, by the strict definition of the term,
they are not functions. They can be considered as set-valued func-
tions, that is, arg(z) and log(z) are subsets of C containing more
than one elements rather than single numbers. Recall tha t

log(z ) = ln |z | + i arg(z ). (∗)

Now, the question is the following:

Given a multi-valued map f (z) on certain open set Ω,

can we identify certain values of f (z) for each z ∈ Ω, say
F (z) ∈ f (z) such that the F : Ω → C is a continuous
function?

For instance,
z ›→ Arg (z)
is a single valued function on C \ {0}. But, this function is not
continuous on the negative real axis:

z n = −1 + i / n → −1, znj = −1 − i / n → −1,

but,
Arg (z n ) → π, Arg (znj ) → −π.
However, if Ω0 is the set obtained by deleting the whole of negative-
real axis, including 0, from the complex plane, then the function
z ›→ Arg (z)

is continuous on Ω0. Similarly,

z ›→ Log (z)

is continuous on Ω0.
Deftnition 3.4.1 Suppose f is a multi-valued map defined on an
open connected set Ω ⊆ C. Then by a branch of f we mean a
continuous function f 0 : Ω0 → C defined on an open set Ω0 ⊆ Ω such
tha t for each z ∈ Ω0 , f 0 (z ) is one of the values of f (z ). ♦
42 Elementary Functions

Suppose f a and fA are branches of arg(z) and log(z), respectively,

on an open connected set Ω0.
In view of (∗),
f A (z) = ln |z| + if a (z)

and hence
f a (z) = Imf A (z).

It can be see that

• difference of any two branches of arg(z) is an integer multiple of
2π and
• difference of any two branches of log(z) is an integer multiple of
2πi.
By the discussion preceding the definition, it is clear that

f a (z) := Arg (z) and f A (z) := Log (z)

are branches of arg(z) and log(z), respectively, on the set

Ω := { z = reiθ : r > 0, −π < θ < π}.

Also, arg(z) and log(z) have branches on any open disc which
does not contain the point 0. If z 0 = r0 e iθ 0 is the centre of the disc,
then we can define

f a (z) = arg(z) with θ0 − π / 2 < arg(z) < θ0 + π/2.

The point 0 has special significance for the arg(z) and log(z) and
hence has a special name for it, the branch point.
Deftnition 3.4.2 Let f be a multi-valued map defined on an open
connected set Ω and f 0 be a branch of f on an open subset Ω0 of Ω.

(i) A curve Γ in Ω is called a branch cut for f if f has a branch

on Ω0 := Ω \ Γ.

(ii) A point z 0 ∈ Ω is called a branch point for f if z 0 is the

intersection of all branches of f .

♦
Branches of arg(z) and log(z) 43

Analyticity of branches of logarithm function

Let f a (z) be a branch of arg(z) in an open connected set Ω. We
know that Ω cannot contain 0, and f a (z) is a real valued continuous
function. Let f A (z) be the corresponding branch of log(z), i.e.,

f A (z) = ln |z| + if a (z), z ∈Ω.

We show that fA is analytic in Ω.

Clearly,

1
f A(z) = ln(r 2) + iθ, z := reiθ ∈Ω.
2
Recall that if f := u + iv is an analytic function on an open set, then
the CR-equations in polar coordinates is given by

ru r = vθ , uθ = −rv r .

In the case of fA we have

1
u(r, θ) = ln(r 2 ), v(r, θ) = θ.
2
Hence, we have
1
ur = , uθ = 0, v r = 0, vθ = 1.
r
Thus, the partial derivatives of u and v are continuous and they
satisfy CR-equations. Hence, fA is analytic in Ω.
It can be also, seen tha t

1
f A(z) = ln(x 2 + y2 ) + i tan −1 (y/x), z = x + iy ∈ Ω.
2
Hence,
x y −y x
u x= , u y= , vx = , vy = .
x 2 + y2 x 2 + y2 x 2 + y2 x 2 + y2
Again we see that partial derivatives of u and v are continuous and
they satisfy CR-equations. Further, we have
x − iy z¯ 1
f j (z) = u x+ iv x= = = .
2
x + y 2 |z|2 z
44 Elementary Functions

3.5 Problems
Σ ∞
1. For y ∈ R, show that the series e iy := (iy) n
n= 1 n! converges.
2. For y ∈ R, eiy = cos y + i sin y. Why?

3. Show that the function z ›→ ez is holomorphic on the entire

complex plane and satisfies the following:

(i) e z 1 +z 2 = ez1 ez2 for every z 1 , z 2 ∈C,

(ii) e z+2πi = ez for every z ∈ C,

(iii)if x = Re(z) and y = Im(z), then |ez | = ex and arg(z) = y,

and

(iv)the strip { z ∈ C : −π ≤ Im(z) < π } is mapped onto the

entire complex plane.

4. Given θ1 and θ2 ∈ [0, 2π) with θ1 < θ2, , describe the image of
the strip { z ∈C : θ1 ≤ Im(z) < θ 2 } under the map z ›→ez .

5. Given r, R0 with 0 < r < R, describe the image of the strip

{ z ∈ C : r ≤ Re(z) < R } under the map z ›→ ez .

6. For the function f (z) = ez , describe the curves

|f (z)| = constant, arg(f (z)) = constant.

7. If f is is holomorphic on C satisfying f j (z) = f (z), and g is

defined by g(z) = e−z f (z), then show that g j (z) = 0, so that
f (z) = cez for some constant c ∈C.
8. Find the most general form of a holomorphic f on C satisfying
f j (z) = cf (z) for some constant c ∈ C.

9. Derive the identities:

(i) cosh(z1 + z 2 ) = cosh z 1 cosh z 2 + sinh z 1 sinh z 2 .

(ii) sinh(z1 + z 2 ) = sinh z 1 cosh z 2 + cosh z 1 sinh z 2 .

(iii) cos z = cos x cosh y − i sin x sinh y.

(iv) sin z = sin x cosh y + i cos x sinh y.

Problems 45

2 2 2
(v) |cos z| = cos x + sinh y.
2 2 2
(vi) |sin z| = sin x + sinh y.
10. Show that

cos(z + 2π) = cos z, sin(z + 2π) = sin(z)

for all z ∈C.

11. Find zeros of sin z and cos z.

12. Find all roots of the equation cos z = 1.
13. Show that if b is a logarithm of a, then b + 2nπi is also a
logarithm of a for every n ∈ Z.

14. Show that if b = log a, log a = ln |a| + i arg(a).

15. Show that, for n ∈Z,
(i) 2nπi = log 1.
(ii) (2n + 1)πi = log(−1).
. Σ
(iii) 2n + 12 πi = log i.

16. Find all values of

(i) cosh(log 2),
(ii) log(log i).
17. Does the relation log a1 a2 = log a1 + log a2 hold for all nonzero
a 1 , a2 in C?
4
Power Series

4.1 Convergence
In this chapter we study convergence and other properties of series
Σ ∞
of the form a 0+ a (z − z 0) nwhich we write as
n= 1 n

Σ∞
a n (z − z 0 ) n (∗)
n= 0

whenever sequence (a n ) in C and z 0 ∈ C are given.

Clearly, the series (∗) is a special case of the series

Σ∞
f n (z ), (∗∗)
n= 0

where (f n ) is a sequence of complex valued functions defined on some

subset Ω ⊆ C.
Deftnition 4.1.1 The series in (∗∗) is said to converge at a point
z ∈ Ω if the sequence of its partial sums converge at z, i.e., if the
sequence (g n (z)) of complex numbers, where
Σn
gn (z) = f j (z), z ∈Ω, (+)
j=0

converges. ♦
Deftnition 4.1.2 The series in (∗∗)is said to converge

(i) point-wise to a function g on Ω if it converges to g(z) at every

point z ∈ Ω;
Σ ∞
(ii) absolutely on Ω if the series n= 0
|f n(z)| converges at every
point z ∈ Ω;

46
Convergence 47

(iii) uniformly to a function g on Ω if the sequence (gn) defined as

in ( + ) converge uniformly to g on Ω.

♦
Remark 4.1.1 It can be seen that if (∗∗)converges at a point z,
then
f n (z) → 0 as n → ∞.

This follows from the face that

|f n (z)| = |gn(z) − gn−1 (z)| → 0 as n → ∞.

Σ ∞
It should be noted that a series of the form n= 1 a n, where (a n)
is a sequence of complex numbers, is a particular case of (∗∗)with
f n (z) = an for all z ∈ C. ♦
Σ ∞ n
EXAMPLE 4.1.1 Consider the series n= 0 z . Note that

Σn 1 −z n+ 1
gn (z) = z j= , z ƒ = 1,
1 −z
j=0

so tha t
1 zn+1
− g n (z) = .
1−z 1−z
Thus, if |z| < 1, then

1
g n (z) → as n → ∞.
1−z

Also, for z| < 1, we have

Σn 1 − |z|n+ 1 1
|z j|= → as n → ∞.
1 − |z| 1 −|z|
j=0
Σ ∞ n
Thus, the series n= 0
z converges absolutely for |z| < 1. Further,
if 0 < r < 1, then for |z| ≤ r, we have

|z|n+1 rn+1
. 1 − gn(z) .= ≤ .
. 1 −z . |1 − z| 1 −r
Σ ∞
Since r n→ 0, it follows that n= 0 z nconverges uniformly on the set
{ z ∈ C : |z| < r}. Q
48 Power Series

As in the case of real valued functions of real variable, we have

the following result.

Theorem 4.1.1 (M-test) Suppose (f n ) is a sequence of complex

valued functions defined on some subset Ω ⊆ C. Suppose there exists a
sequence (M n ) of positive real numbers such that

(i) |f n (z)| ≤ M n for all n ∈ N and for all z ∈Ω, and

(ii) Σ∞n=1 M n converges.

Σ ∞
Then the series n= 1 f nconverges uniformly on Ω.

Proof. Note that for n > m,

. Σn . Σn Σn
|gn(z) − gm(z)| = f j (z) ≤ |f j (z)| ≤ Mj . (×)
. j = m+ 1 . j = m+ 1 j = m+ 1

Σ ∞
Now, let ε > 0, and let N ∈ N, Since n= 1 M n converges, from
the above it follows from (×) tha t (g n (z)) is a Cauchy sequence, and
hence it converges. Let

g(z) = lim g n (z), z ∈Ω.

n→∞

Again from (×),

Σ∞
|g(z) − gm(z)| = lim |g (z)
n − g (z)|
m ≤ Mj .
n→ ∞
j = m+ 1
Σ
Now, let ε > 0 be given and N ∈ N be such that n
j = m+ 1 Mj < ε
for all m ≥ N . Then we have

|g(z) − gm(z)| < ε ∀n ≥ N , ∀z ∈Ω.

Thus, (gn) converges uniformly to g.

Now, we prove an important theorem due to Abel1 in the theory

of power series.
1
Niels Henrik Abel (5 August 1802 –6 April 1829) was a noted Norwegian
mathematician who proved the impossibility of solving the quintic equation in
radicals - Curtsey Wikepedia.
Convergence 49

Σ ∞
Theorem 4.1.2 Consider a power series a (z − z 0) n. If this
n= 0 n
series converges at a point z 1 , then it converges at every point z such
that |z − z0| < |z1 − z0 |.
An immediate corollary:
Σ ∞
Corollary 4.1.3 If the power series a (z − z 0) ndiverges at a
n= 0 n
point z 2 , then it diverges at every point z such that |z −z0 | > |z2 −z0 |.
Σ ∞
Proof of Theorem 4.1.4. Suppose a (z − z 0) nconverges at
n= 0 n
z 1 . Let z be such that |z − z0| < |z1 − z0 |. Note that for every n ∈ N,
. Σ
n n |z − z 0| n
|an(z − z 0 ) | = |an(z1 − z 0 ) | .
|z1− z |0
Since |an(z1 −z0 )n| → 0, there exists M > 0 such that |an(z1 −z0 )n| ≤
Σ ∞ . |z−z 0| nΣ
M for all n ∈ N, and since |z − z | < 0|z − z |,1
Σ ∞ n= 1 |z1−z0| n
0
converges. Hence, by comparison test, the series n= 0
|an(z − z 0) |
converges.
Σ ∞
Theorem 4.1.4 Suppose a power series a (z − z0) nconverges
n= 0 n
for all z with |z − z0| < r for some r > 0. Then, for any ρ with
Σ ∞
0 < ρ < r, the series a (z − z 0) nconverges uniformly on the
n= 0 n
set { z : |z − z0| ≤ ρ}.
Σ ∞
In particular, a (z − z 0 ) n converges uniformly on every
n= 0 n
compact subset of the disc { z : |z − z0| < r } .
Proof. Let ρ < r 1 < r and let z 1 be such that |z1 − z0| = r 1 . Then
we have
. Σ . Σ
n n |z − z 0| n ρ n
|an(z − z 0 ) | = |an(z1 − z 0 ) | ≤ |an(z1 − z 0 ) |
n
.
r1 r1
Let M > 0 be such that |an(z1 − z0 )n| ≤ M for all n ∈ N. Thus,
. Σ
ρ n
|an(z − z 0 ) | ≤M
n ∀n ∈N.
r1
Σ ∞
Hence, by M-test, a (z − z 0) nconverges uniformly on the set
n= 0 n
{ z : |z − z0| ≤ρ}.

Suppose
Σ∞
R := sup{|z − z0 | : an(z − z0) n converges at z } .
n= 0
50 Power Series

By Theorem 4.1.4 and Corollary 4.1.3, it follows that

Σ∞
an(z − z 0) nconverges for |z − z |0< R
n= 0

and
Σ∞
an(z − z 0) ndiverges for |z − z |0> R.
n= 0
Σ ∞
Deftnition 4.1.3 For the series n= 0
an(z − z 0) n, the number

Σ∞
R := sup{|z − z0 | : an(z − z 0) nconverges at z }
n= 0

is called the radius of convergence of the series. The disc

{ z : |z − z0| < R }

is its disc of convergence and the set

Σ∞
{ z ∈C : an(z − z0) n converges at z }
n= 0

is called its region of convergence. ♦

Clearly, if Ω is the region of convergence, then

{ z : |z − z0| < R } ⊆ Ω ⊆ { z : |z − z0| ≤ R } .

Thus, region of convergence may include some of the boundary points

of the disc of convergence. See the following examples.
Σ∞
EXAMPLE 4.1.2 (i) Consider the series z n. Its radius of con-
n= 0
vergence is 1. Its disc of convergence and region of convergence are
the same, the disc:
{ z : |z − z0| < 1}.

Σ∞ z n
(ii) Consider the series 2
. Since |z n /n 2 | ≤ 1/n 2 for all z
n= 0
n
with |z| ≤ 1, the series converges absolutely on

Ω := { z ∈ C : |z| ≤ 1}.
Convergence 51

If z ƒ∈Ω, then |z| n /n 2 ƒ→ 0 (see Exercise 4.1.1). Thus, the series does
not converge outside Ω. Hence, the region of convergence is Ω, and
the disc of convergence is interior of Ω.
Σ∞ z n
(iii) Consider the series . Since |z n /n| ≤ |z|n for all n ,
n
n= 0
the series converges absolutely for |z| < 1. Also, the series does not
converge at z = 1. Hence, the radius of convergence is 1.
Now, let z be such that |z| = 1 and z ƒ = 1. Since
. Σn .
z j = . 1 −z
n+ 1 . 2
. . . 1 −z . ≤ |1 − z| ∀n ∈N.
j=0
.Σ Σ
n
Now, recall (see Theorem 3.42 in Rudin 2) tha t if j = 1 a j is bounded
and (bn) is a decreasing sequence of non-negative real numbers which
Σ ∞
converges to 0, then n= 1 anb n
converges. In the present example,
Σ
we have an = z n with |z| = 1, z ƒ = 1 and bn = 1/n. Thus, ∞n=0z n / n
Σ∞ n
converge. Thus, the region of convergence of n=0 z / n is

{ z : |z − z0| ≤ 1} \{1}.
Σ∞ z n
(iv) Consider the series . Since |zn/n!| ≤ 1/n! for all n
n!
n= 0
and for all z ∈C, the series converges absolutely in the entire plane.
Hence, the region of convergence is the entire C.
Σ∞
(v) Consider the series n!z n. Writing a =n n!z , we
n have for

n= 0
z ƒ= 0
. an+ 1 . = (n + 1)|z| → ∞ as n → ∞.
. .
an
Hence, for z ƒ = 0, an ƒ→0 (see Exercise 4.1.1). Thus, region of
convergence of the series is the singleton set {0}. Q
Σ n
2
W . Rudin, Principles of Mathematical Analysis, 1976: Let A : =n j=1 an
a n d M > 0 is such t h a t |An| ≤ M for all n ∈ N. For q > p, we have
Σq Σq q− 1
Σ
an bn = (A n− A n − 1 )bn = A n (bn − bn + 1 ) + A p b+
p A b.
qq
n= p n= p n = p− 1

q q− 1
Hence, . Σ a b| ≤ M Σ
(bn − b n + 1 ) + Mb p + Mb q = M (b p−1 + bq) → 0 as
. n n
n= p n = p− 1
p, q → ∞.
52 Power Series

an+ 1
n 0 for all n ∈ N and
Exercise 4.1.1 Suppose a > → α. Prove
an
that
(i) if α ≥ 1, then an ƒ→0, and
(ii) if α > 1, then an → ∞. a

For the next theorem, we recall the following from real analysis:

Theorem 4.1.5 Let (a n ) be a sequence of positive real numbers and

b = lim supn a n . Then the following hold:

(i) If b < A, then there exists k ∈N such that an < Afor all n ≥ k.

(ii) If b > A, then an > Afor infinitely many n ∈N.

Proof. Let bn := sup{a j : j ≥ n}. Then, we see that (bn) is a

decreasing sequence, and hence

b := lim sup an := lim bn.

n n→ ∞

(i) Suppose b < A. Then there exists k ∈ N such that bk < A. Hence,
an ≤ bk < Afor all n ≥ k.
(ii) Suppose b > A. Then bn > Afor all n ∈N. Hence for each
n ∈ N, there exists k n ∈ N such that k n > n and a k n > A.

Using the above theorem we have the following result on conver-

gence of a sequence of complex numbers.
Theorem 4.1.6 Suppose (a n ) is a sequence in C, and

b := lim sup |a n|1/n .

n
Σ ∞
Then the series n= 1 an

(i) converges absolutely if b < 1, and

(ii) diverges if b > 1.

Proof. (i) Suppose b < 1, and let A be such that b < A < 1. Then
by Theorem 4.1.5 (i), there exists k ∈ N such that |an|1/n < Afor all n
≥ k. Thus,
|an| < An ∀n ≥ k.
Convergence 53

Σ ∞ Σ ∞
Since 0 ≤ A< 1, the series n= 1 An, and hence the series n= 1
|an|
converges.
(ii) Suppose b > 1, and let Abe such that 1 < A< b. Then by
Theorem 4.1.5 (ii), there exists a sequence (k n ) in N such that
|akn |1/kn > Afor all n ∈ N. Thus,

|akn | > Akn > 1 ∀n ∈N.

Σ∞
Hence, an ƒ→0. Consequently, the series n=1 a n diverges.

Theorem 4.1.7 (Cauchy3–Hadamard4 theorem) For a sequence

(a n ) in C, let
β := lim sup |a n|1/n .
n
Σ∞
Then R := 1/β is the radius of convergence of n=0 a n (z − z0 )n.
Σ∞
Proof. By Theorem 4.1.6, the series n=0 a n (z − z0 )n
(i) converges if |z − z0| lim supn |an|1/n < 1, and
(ii) diverges if |z − z0| lim supn |an|1/n > 1.
From the above, the conclusion follows.

Using similar arguments as in the proof of Theorem 4.1.7, we

obtain the following.

Theorem 4.1.8 (Ratio test) For a sequence (a n ) of nonzero com-

plex numbers, let
γ := lim sup . an+ 1 . .
. .
n an
Σ
Then R := 1 / γ is the radius of convergence of ∞n=0 a n (z −z 0 ) n .

Exercise 4.1.2 Find the radius of convergence for each of the fol-
lowing series5:
Σ∞ Σ∞ 2n Σ∞ 2n Σ∞ n 3
2 n n n
(i) n z , (ii) z , (iii) z , (iv) z n. a
n! n2 3n
n= 0 n= 0 n= 1 n= 0

3
Augustine Luis Cauchy (21 August 1789 –23 May 1857)
4
Jaques H a d am ard (8 December 1865 - 17 October 1963)
5
W.Rudin, C h a p ter 3, Exercise 9.
54 Power Series

Theorem 4.1.9 Suppose R > 0 be the radius of convergence of the

Σ
series ∞n=0 a n (z − z 0 ) n , and

Σ∞
f (z) := an(z −z 0 ) n for |z − z0| < R.
n= 0
Σ ∞
Then f is differentiable and the series n= 1
na n (z −z 0 )n−1 converges
for |z| < R and

j
Σ∞
f (z) = na n(z − z 0) n−1
for |z − z0| < R.
n= 1

Proof. Since lim supn |nan|1/n = lim supn |an|1/n the radii of con-
Σ ∞ Σ ∞
vergence of n= 0 an(z − z0 ) nand na n(z −z 0 ) n−1 are the same.
Σ ∞ n= 1
n=1na n (z − z 0 )
Thus, n−1 converges for |z − z | < R. Let
0

Σ∞
g(z ) = na n (z − z 0 ) n−1 , |z − z0| < R.
n= 1

For the sake simplicity of presentation, without loss of generality, we

assume that z 0 = 0. Thus,

Σ∞ Σ∞
f (z ) = a n z n, g(z) = na nz n−1 for |z − z0| < R.
n= 0 n= 1

Now, let 0 < ρ < R and let z ƒ = z 1 with |z| ≤ ρ, |z1| ≤ ρ. Note that

Σ∞
f (z) − f (z 1 ) = an (z n− z )1n
n= 0
∞ n−1
Σ Σ
= an (z − z 1 ) z kzn−k−1
1
n= 0 k= 0

Hence,

f (z) − f (z 1 ) Σ∞ Σn−1
Σ Σ
− g(z 1 ) = an z kzn−k−1 − nz 1n−1
z − z1 1
n= 1 k= 0
Σ∞ ΣΣ
n−1 Σ
= an z 1n−k−1 (z k− z 1k)
n= 2 k= 0
Convergence 55

Now,
k−1
Σ
z 1n−k−1 (z k− z )1k = z n−k−1
1
(z − z 1) z j z 1k−j −1
j=0

so that
k−1
Σ
|z 1n−k−1 (z k− z )1k ≤ |z − z |ρ
1
n−k−1
ρ j ρ k−j−1 ≤ |z − z 1 |kρ n−2 .
j=0

T hus,
∞ Σn−1
. f (z) − f (z 1) − g(z ) . ≤ Σ Σ n−2
. 1 . |an| |z − z1 |kρ
z −z 1
n= 2 k= 0
Σ∞ n(n − 1) n−2
≤ |z − z 1 | |an| ρ .
2
n= 2

Σ ∞
Since n= 2
|an| n(n−1)
2 ρ n−2 converges (Why?), we have

f (z) − f (z 1)
→ g(z 1 ) as z → z1.
z −z 1

Thus, the proof is complete.

By the above theorem, a power series can be differentiated ter m

by term within its disc of convergence. Further, if f represents a
power series on its disc of convergence, i.e.,

Σ∞
f (z) := an(z − z 0 ) n for |z − z0| < R.
n= 0

then f is infinitely times differentiable, and for any k ∈ N,

Σ∞ n!
f (k)
(z) := an (z − z 0 ) n−k for |z − z0| < R.
(n −k)!
n= k

Hence, we have
f (k) (z 0 )
, k ∈N.
ak =
k!
The above discussion urges us to ask the following question:
56 Power Series

Suppose f is holomorphic in a disc centered at z 0 . Then,

does f have the representation

Σ∞ f (n) (z
0)
f (z) = (z − z0) n
n!
n= 0

in that disc?

We answer this question affirmatively.

Suppose f is holomorphic in a neighborhood of z 0 . Let Ω be the
largest open disc centered at z 0 in which f is holomorphic, i.e., if

ρ := sup{|z − z0| : f is holomorphic at z } ,

then Ω = { z ∈ C : |z − z0| < ρ}. We shall show that there exists

(a n ) in C such that

Σ∞
f (z) = an(z − z0) n ∀z ∈Ω,
n= 0

and hence
Σ∞ f (n) (z
0)
f (z) = (z − z0) n ∀z ∈Ω.
n!
n= 0

This is one of the biggest differences between

(i) a real valued differentiable function of a real variable on an open
set in R and
(ii) a complex valued differentiable function of a complex variable on
an open set in C.

4.2 Problems
Note: Problems from 1-3 and 9-12 are discussed in class, either by
proving them, or by way of indicating their proofs.
Σ ∞
1. Suppose a power series a (z − z 0 ) n converges for all z
n= 0 n
with |z − z0| < r for some r > 0. Then, prove that for any ρ
Σ ∞
with 0 < ρ < r, the series n= 0 an(z−z 0) nconverges uniformly
on the set { z : |z − z0| ≤ ρ}.
Problems 57

Σ ∞
2. Let R be the radius of convergence of n= 0
an(z − z 0) n. Prove
the following:
(i) If the series converges at z 1 , then R ≥ |z1 −z0 |.
(ii) If the series diverges at z 2 , then R ≤ |z2 −z0 |.
Σ
(iii) If ∞n=0 |an(z − z0 )n| diverges at z 2 , then R ≤ |z2 − z0 |.

3. If (a n ) and (bn) are sequences of complex numbers such that

|an| ≤ M |bn| for all n ∈ N, and if R 1 nd R 2 are the radius of
Σ ∞ Σ ∞
convergence of n= 1 an(z − z 0) nand b (z − z0) nrespec-
n= 1 n
tively, then prove that R 1 ≤ R 2 .
Σ ∞
4. Using Problem 3, show that radius of convergence of n= 1 n −n zn
is ∞.
Σ∞ nzn
5. Prove that radius of convergence of n=1 n is 0.

6. Find a power series in a neighborhood of z 0 = 1 which repre-

sents the function f (z) := 1/z.

7. Find the radius of convergence for each of the following series:

Σ∞ Σ∞ 2n Σ∞ 2n Σ∞ n 3
(i) n 2 zn , (ii) z n , (iii) z n
, (iv) z n.
n! n2 3n
n= 0 n= 0 n= 1 n= 0
Σ∞ (n!) 3 Σ∞ z n! Σ∞ Σ∞ n + 1 3
n n2
(v) z 3n , (vi) , (vii) n z , (viii) zn .
(3n)! n n!
n= 0 n= 1 n= 0 n= 0

8. Give one example each of a power series which

(a) converges only on the interior of the disc of convergence,

(b) converges diverges on a proper subset of the boundary of
the disc of convergence,
(c) converges on the closure of the disc of convergence.
Σ ∞ Σ ∞ n−1
9. Show that the series a (z−z 0) nand
n= 0 n n= 1
na n(z−z 0)
have the same radius of convergence.

10. If (a n ) and (bn) are sequences of complex numbers such that

Σ
lim supn |bn|1/n, then show that the series ∞ a (z−z 0) nand
n= 0 n
Σ ∞
n=1na n b n (z − z 0 )
n−1 have the same radius of convergence.
58 Power Series

11. For a sequence (a n ) of nonzero complex numbers, let γ :=

lim sup an + 1 . Then show that R := 1 / γ is the radius of
n . an .
Σ ∞
convergence of n= 0 an(z − z 0) n.
Σ ∞
12. If f represents a power series a (z − z 0) non its disc of
n= 0 n
f ( k) (z0 )
convergence, then a k= k!
for every k ∈ N. Justify.

13. Let f be a holomorphic function in an open set Ω such that f j

= f and f (0) = 1. Then show that f (z) = ez . Deduce that,
for all z ∈C,
Σ∞ z n
(i) ez=
n!
n= 0
Σ∞ z 2n+ 1
(ii) sin z = (−1)n ,
(2n + 1)!
n= 0
∞
Σ z 2n
(iii) cos z = (−1) n .
(2n)!
n= 0

14. Prove that, for |z| < 1,

1 Σ∞
(i) = (−1)nz2n
1 + z2 n= 0

1 Σ∞ z n
(ii) Log = n.
1 −z
n= 1
Σ∞
15. Find the function represented by the series n=1 n2 zn.
5
Integration

5.1 Integrals along Piecewise Smooth Curves

Recall that a curve in C is a continuous function from a closed (non-
degenerate) interval to C. Thus, a continuous function

γ : [a, b] → C

is a curve in C. If the range of γ is contained in a set Ω ⊆ C, then we

say that γ is a curve in Ω. The point z 1 := γ(a) is called the initial
point of γ and the point z 2 := γ(b) is called the final point or
terminal point of γ.
Deftnition 5.1.1 Given a curve γ : [a, b] → C, the corresponding
reversed curve γ˜: [a, b] → C is defined by

γ˜(t) = γ(a + b − t), t ∈ [a, b].

♦
Note that, γ˜ has the same range as that of γ, but its orientation
as t varies from a to b is reversed.
Deftnition 5.1.2 We shall call the range of a curve γ : [a, b] → C
with orientation as t varies from a to b as oriented range of γ, and
denote it by Γ γ or, simply, Γ.
If Γ is the oriented range of γ : [a, b] → C, then we say that γ is a
parametrization of Γ. ♦
Thus, oriented range of γ : [a, b] → C is

Γ γ := { γ ( t ) : a ≤ t ≤ b}.

with orientation as t varies from a to b.

Similarly, the oriented range of γ˜ will be denoted by Γ˜.

59
60 Integration

Convention: Hereafter, if we say Γ is a curve, then we mean that Γ

is the oriented range of a curve γ : [a, b] → C.
Deftnition 5.1.3 Curves γ 1 : [a1 , b1] → C and γ 2 : [a2 , b2] → C are
said to be equivalent, if there exists continuous increasing bijection
ϕ : [a1 , b1] → [a2 , b2] such that γ 1 = γ 2 ◦ ϕ, and in that case we write
γ1 ∼ γ2.
If γ 1 ∼ γ 2 , then for γ 2 is called a reparameterization of γ 1 . ♦
Exercise 5.1.1 (i) Show that equivalence of curves defines an equiv-
alence relation.
(ii) If γ 1 ∼ γ 2 , then show that Γ γ 1 = Γ γ 2 .
(iii)Given a curve γ : [a, b] → C and a closed interval [α, β], find a
curve η : [α , β ] → C such γ ∼ η. a
Exercise 5.1.2 Given a curve γ : [a, b] → C, show that the curve
η : [−b, −a] → C defined by
η(t) = γ(−t), −b ≤ t ≤ −a,
is equivalent to the reverse of γ, i.e., η ∼ γ˜. a

Given curves γ 1 : [a, b] → C and γ 1 : [c, d] → C, define γ 3 : [a, b +

d − c] → C by
.
γ 1 (t), a ≤ t ≤ b,
γ 3(t) =
γ 2 (t + c − b), b ≤ t ≤ b + d − c.
Then, γ 3 : [a, b + d − c] → C is a curve such that terminal point of
γ 1 is the initial point of γ 2 . This curve γ 3 is called the sum of the
curves γ 1 and g2 and is denoted by γ 1 + γ 2 .
In the sequel, we shall be dealing with piecewise smooth curves.
Deftnition 5.1.4 A curve γ : [a, b] → C is said to be piecewise
smooth if
( i ) γ is differentiable except possibly a t a finite number of points in
[a, b], and
(ii) right and left derivative of γ exists at every point in [a, b]. ♦
Deftnition 5.1.5 If γ : [a, b] → C is a piecewise smooth curve, then
its length is defined by
∫ b
A(Γ) := |γ j(t)|dt,
a
where Γ is the oriented range of γ. ♦
Integrals along Piecewise Smooth Curves 61

In order to define integrals of complex valued functions of a com-

plex variable along piecewise smooth curves, first we define integral
of complex valued functions on bounded intervals.
Suppose ϕ : [a, b] → C be a continuous function. Then the inte-
gral
∫ b
ϕ(t) dt
a
is defined in a natural way:
Deftnition 5.1.6 If ϕ = ϕ 1 + iϕ 2 , where ϕ 1 and ϕ 2 are real and
imaginary parts of ϕ, respectively, then
∫ b ∫ b ∫ b
ϕ(t) dt := ϕ 1 (t) dt + i ϕ 2 (t) dt.
a a a

♦
The following properties can be easily verified.
∫ b ∫ b ∫ b
• [ϕ(t) + ψ(t)] dt = ϕ(t) dt + ψ(t) dt,
a a a
∫ b ∫ b
• [αϕ(t)] dt = α ϕ(t) dt for all α ∈R.
a a

Exercise 5.1.3 Verify the above properties. a

Further, we have the following.

Proposition 5.1.1
∫ b ∫ b
. ϕ(t) dt . ≤ |ϕ(t)| dt. (∗)
. .
a a
∫ b
Proof. If ϕ(t)
dt = 0, then clearly (∗) holds. So, assume that
a ∫ b ∫ b
∫b . ϕ(t) dt ./ ϕ(t) dt. Then wehave
a ϕ(t) dt is nonzero and λ := . .
a a
∫ b ∫ b ∫ b
. ϕ(t) dt . = λ ϕ(t) dt = λϕ(t) dt
. .
a a a
∫ b ∫ b ∫ b
= Reϕ(t) dt ≤ |λϕ(t)| dt = |ϕ(t)| dt
a a a

This completes the proof.

62 Integration

Exercise
∫ a 5.1.4 For a > 0 and α > 0, show that
e−(α+ it) dt = 0 .
2
lim a
α→ ∞
0

We have the following analogue of the fundamental theorem of

integration.
Proposition 5.1.2 If ϕ : [a, b] → C is continuously differentiable,
then ∫ b
ϕ j (t) dt = ϕ(b) − ϕ(a). (∗∗)
a

Now we define the integral for a continuous complex valued func-

tion of a complex variable along a piecewise smooth curve.
Deftnition 5.1.7 Let γ : [a, b] → C be a piecewise smooth curve
with oriented range Γ and f : Γ → C be a continuous function. Then
we define ∫ ∫ b
f (z) dz := f (γ(t))γ j(t) dt.
Γ a
∫
We shall also denote the above integral by f (z)dz. ♦
γ

Proposition 5.1.3 The following hold.

∫ ∫ ∫
1. [f (z) + g(z)] dz = f (z) dz + g(z) dz,
Γ Γ Γ
∫ ∫
2. [αf (z)] dz = α f (z) dz for all α ∈ C.
Γ Γ
∫ ∫
3. f (z) dz = − f (z) dz.
Γ̃ Γ
∫ ∫ ∫
4. f (z) dz = f (z) dz + f (z) dz.
γ1+ γ2 γ1 γ2

Proof. Exercise.
∫
Remark 5.1.1 It is to be observe that the integral Γ γ f (z)dz de-
pends essentially on the way the point γ(t) moves along Γ γ as t varies
on [a, b]. It can happen
∫ that two∫ different curves γ and η can have
same range Γ, but γ f (z)dz =ƒ η f (z)dz. For example, consider the
curves
γ(t) = t 2 , η(t) = t for 0 ≤ t ≤ 1.
Integrals along Piecewise Smooth Curves 63

Clearly, the range of γ and η coincide and it is the line segment [0, 1].
However, if we take f (z) = z, then
∫ ∫ 1 ∫ ∫ 1
1 1
f (z )dz = t 2 (2t)dt = , f (z )dz = t(t)dt = .
γ 0 2 η 0 3
♦
Deftnition 5.1.8 (i) A curve γ : [a, b] → C is called a closed curve
if its initial and terminal points are the same.
(ii) A curve γ : [a, b] → C is said to intersect at a point z 0 if there
exists distinct t 1 , t 2 ∈ [a, b] such tha t γ(t 1 ) = z 0 = γ(t 2 ).
(iii) A closed curve γ : [a, b] → C is said to be a simple closed
curve if for distinct points t 1 , t 2 in [a, b], γ(t 1 ) = γ(t 2 ) implies
{t 1 , t 2 } = {a, b}. ♦
∫
f (z) dz ≤ MA Γ , where M = maxz∈Γ |f (z)|.
Proposition 5.1.4 . .
Γ
Proof. By (∗),
∫ ∫ b
. .
. f (z) dz . ≤ |f (γ(t)) | |γ j(t)| dt ≤ MAΓ.
Γ a

Thus, the proof is complete.

Proposition 5.1.5 Let f be continuously differentiable in an open
set containing a piecewise smooth curve with initial point z 1 and
terminal point z 2 . Then
∫
f j (z) dz = f (z 2 ) − f (z 1 ).
Γ
∫
In particular, if Γ is closed, then f j (z) dz = 0.
Γ
Proof. By (∗∗), we have
∫ ∫ b
j
f (z ) dz = f j (γ(t))γ j (t) dt
Γ a
∫ b
= ( f ◦ γ ) j(t) dt = ( f ◦ γ)(b) − ( f ◦ γ)(a)
a
= f (z 2 ) − f (z 1 ).

The particular case follows, since in this case z 2 = z 1 .

64 Integration

Corollary 5.1.6 Let Γ be a closed curve and f has a primitive on an

open set Ω containing Γ, i.e., there exists a continuously differ-
entiable function g such that gj = f on Ω. Then
∫
Γ
f (z) dz = 0.

Proof. Follows from Theorem 5.1.4.

Remark 5.1.2 We shall prove in the next section that if f is holo-

morphic on a simply connected domain Ω, then integral of f along
every closed curve in Ω is 0. This result is known as Cauchy’s theo-
rem. ♦
EXAMPLE 5.1.1 Let Γ be the circle with center at z 0 and radius
r given by
it
Γ := {z 0 + re , 0 ≤ t ≤ 2π}.
Then, we can take γ(t) := z 0 + re it , 0 ≤ t ≤ 2π so that
∫ ∫ 2π j ∫ b it
dz γ (t)dt rie dt
dz = = = 2πi.
Γ z − z 0 0 γ(t) − z 0 a re it

Note that the value if the above integral does not depend on the
centre and the radius. Q

Exercise 5.1.5 Let Γ n be the circle with center at z 0 and radius r

traced n∫ times, i.e., Γ n:= { z +0 re , it0 ≤ t ≤ 2nπ}. Then, show that
dz
(i) = 2nπi, and if p(z) is a polynomial, then
∫Γ n z −z 0

(ii) p(z) dz = 0. a
Γn

EXAMPLE 5.1.2 Let Ω be the disc of convergence of a power

Σ ∞
series a (z − z 0) nand let f (z) represent this series in Ω. Let
Σn=∞0 nan j = f (z). Hence, if
g(z) = n= 0 n+ 1
(z − z 0 ) n+ 1 . Then we have g (z)
Γ is a closed piecewise smooth curve in the disc of convergence, then
∫
Γ
f (z) dz = 0.

Thus, integral of a function which represents a power series, over any

closed piecewise curve in its disc of convergence, is 0. Q
Cauchy’s Theorem 65

Exercise 5.1.6 For a > 0 and α > 0, let Γ α be ∫the line segment
e−z dz = 0.
2
joining z 0 = α to z 1 = α + ia. Show that lim a
α→ ∞
Γa

Exercise
∫ ∫ Suppose f →
5.1.7 n f uniformly on Γ. Then show that
f n (z)dz → f (z) dz. a
Γ Γ

Exercise 5.1.8 If f is continuous on { z : |z − z0| ≤ 1} and if

γ(t) = z 0+ re it, 0 ≤ t ≤ 2π} and Γ ∫is the oriented range of γ for
1 f (z)
0 < r ≤ 1, then show that lim dz = f (z 0 ). a
r→0 2πi Γ z −z 0

5.2 Cauchy’s Theorem

Now, we prove one of the most important theorems in complex anal-
ysis, the so called Cauchy’s theorem. For its statement we shall use
the following definitions.
Deftnition 5.2.1 An open connected subset of the complex plane is
called a domain. ♦
It is easy to conceive the statement in the following theorem,
though its proof is much involved and beyond the scope of this course:

Theorem 5.2.1 (Riemann’s theorem) If Γ is a simple closed

curve, then it is the boundary of two disjoint domains one of which is
bounded and the other is unbounded.

Deftnition 5.2.2 If Γ is a simple closed curve, then the bounded

domain as in Definition 5.2.1 is called the domain enclosed by Γ.
♦
Deftnition 5.2.3 A domain Ω is said to be simply connected if for
every simple closed curve Γ in Ω, the domain enclosed by Γ is
contained in Ω. ♦
In the following, we shall call a simple closed piecewise smooth
curve as simple closed contour.
Deftnition 5.2.4 A simple closed contour γ is said to be positively
oriented if the domain enclosed by it is on the left while traversing
along γ.
More precisely, for each t ∈ [a, b], the normal vector γ j(t)eiπ/2 at
the point γ(t) must direct towards the domain enclosed by γ, i.e., for
66 Integration

each t ∈ [a, b], there exists ε t > 0 such that

γ(t) + εγ j(t)eiπ/2 ∈ Ωγ whenever 0 < ε < ε t,

where Ωγ is the domain enclosed by γ. ♦

Theorem 5.2.2 (Cauchy’s theorem - using Green’s theorem)

Let Γ be a positively oriented simple closed contour and let Ω be the
domain enclosed by Γ. Let f be holomorphic and its derivative
continuous on Ω ∪Γ. Then
∫
f (z)dz = 0.
Γ

Proof. Let u and v be the real and imaginary parts of f . Let Γ

have the parametrization γ on [a, b]. Let γ(t) = x(t)+iy(t), t ∈ [a, b].
Then
∫ ∫ b
f (z ) dz = f (γ(t))γ j(t) dt
Γ a
∫ b
= [u(x(t), y(t)) + iv(x(t), y(t))][x j (t) + iy j (t)] dt
a
∫ b ∫ b
= [ux j − vy j]dt + i [uy j− vx ]dt
j

∫a ∫ a
= [udx − vdy] + i [udy + vdx].
Γ Γ

Now, by Green’s theorem1 and using the CR-equations, we have

∫ ∫ ∫ . −∂v ∂u Σ
[udx −vdy] = − dxdy = 0,
Γ ∂x ∂y
Ω
∫ ∫∫ . Σ
∂u ∂v
[vdx + udy] = − dxdy = 0.
Γ ∂x ∂y
Ω

This completes the proof.

Exercise 5.2.1 Prove Cauchy’s theorem if Γ is any piecewise closed

curve which intersects itself only at a finite number of points. a
1
Green’s theorem, named after the British mathematician a.n d physicist
Σ
¸ ¸ ¸ ∂Q
George Green (14 July 1793 31 May 1841) : Γ Pdx+Qdy = ∂x
− ∂P
∂y
dxdy.
Ω
Cauchy’s Theorem 67

We shall see that the condition that the real and imaginary parts
of f have continuous partial derivatives is redundant. In fact, we
shall prove the following general version of Cauchy’s theorem.

Theorem 5.2.3 (Cauchy’s theorem) Let Ω be a simply connected

domain and f be a holomorphic function on Ω. Then for every piece-
wise smooth closed curve Γ in Ω,
∫
Γ
f (z) = 0.

Our proof involves the following steps:

1. The theorem holds with γ as a triangle.

2. The theorem holds if γ is a rectangle.

3. Every holomorphic function on a simply connected domain has

a holomorphic primitive.

4. Use Corollary 5.1.6.

First we require the following simple property of a domain. For

simplicity of expression, we shall call a curve a rook-path2 if it is a
polygonal curve consisting of line segments parallel to coordinate
axes.

Lemma 5.2.4 Any two points in an open connected set can be joined
by a rook-path.

Proof. Let Ω be an open connected set. If Ω is empty, then the

lemma holds vacuously. Hence assume that Ω is nonempty and z 0 ∈
Ω. Now, consider the set Ω0 of all those points in Ω which can be
joined with z 0 by rook-paths. We have to show that Ω0 = Ω. Since Ω
is connected and z 0 ∈ Ω0, it is enough to show that both Ω0 and its
compliment Ω1 := Ω \ Ω0 are open.
Let z ∈ Ω0, and let r > 0 be such that B(z, r) ⊆ Ω. Since every
point in B(z, r) can be joined to z by a rook-path, we have B(z, r)
⊆ Ω0. Thus, Ω0 is an open set.
2
borrowed from Persian rokh, in Sanskrit rath meaning “chariot”, is a piece in
the strategy board game of chess.
68 Integration

Next, let ζ ∈ Ω1, and let ρ > 0 be such that B(ζ, ρ) ⊆ Ω. Since
every point in B(ζ, ρ) can be joined to ζ by a rook-path, and since ζ
ƒ∈ Ω0, we have B(ζ, ρ) ⊆ Ω1. Thus, Ω1 is also an open set. This
completes the proof.

Now, we proceed to prove (1)-(3).

Theorem 5.2.5 (Goursat’s lemma)3 Suppose Γ is a positively

oriented triangle and f is holomorphic on and inside Γ. Then
∫
Γ
f (z)dz = 0.

Proof. By joining midpoints of the sides of Γ construct four pos-

itively oriented triangles, say Γ 0,1 , Γ 0,2 , Γ 0,3 , Γ 0,4 . Then we obtain
∫ Σ4 ∫
I := f (z)dz = f (z)dz.
Γ j=1 Γ 0,j

Σ4 . ∫
Since |I| ≤ . f (z)dz.. , it follows that there exists j0 ∈ { 1, 2, 3, 4}
j=1 Γ 0,j
such that ∫ ∫
1
. f (z)dz.. ≥ .. f (z)dz .
.
Γ 0 ,j 0 4 Γ

Denote this Γ 0,j 0 by Γ 1 .

Now, joining midpoints of the sides of Γ 1 construct four positively
oriented triangles, say Γ 1,1, Γ1,2 , Γ 1,3 , Γ 1,4. Following the same argu-
ment as above with Γ 1 in place of Γ, there exists j 1 ∈{1, 2, 3, 4} such
that ∫ ∫ ∫
f (z)dz. ≥ 1 . f (z )dz..≥ 1.
. f (z )dz .. (1)
. . .
Γ 1 ,j 1 4 Γ1 42 Γ
Denote Γ 1,j 1 by Γ 2 and continue the above procedure. Then we
obtain a sequence { Γ n } of positively oriented triangles such that
∫ ∫
f (z)dz. ≥ 1 . f (z)dz ..
. . .
Γn 4n Γ

3
Edo uar d Goursat (1858-1936, a French mathematician, was the first rec-
ognized, in 1800, t h a t continuity of the derivative is not required for proving
Cauchy’s theorem.
Cauchy’s Theorem 69

Let Ωn be the closure of the domain enclosed by Γ n . Then we see

that Ωn ⊇ Ω n+1 for all n ∈ N and
1
diam(Ω n ) ≤ A(Γn) = A(Γ). (2)
2n
T∞
Hence, by nested compact sets theorem in real analysis,
T n= 1 Ωnis a
singleton set, say ∞n=1Ωn = {z 0 }.
Now, let ε > 0 be given. Since z 0 ∈ Γ ∪ Ω and f is holomorphic at
z 0 , there exists δ > 0 such that

|f (z) − f (z 0 ) − f j (z 0 )(z − z0 )| < ε|z − z0| whenever |z − z0| < δ.

Let N ∈ N be such tha t Ωn ⊆ B(z 0 , δ) for all n ≥ N . Then, we have

∫
. ( f (z) − f (z 0 ) − f j (z 0 )(z − z 0 )dz. ≤ ε[A(Γn)]2. (3)
Γn

Note that, since the function z ›→ f (z 0 ) + f j (z 0 )(z − z 0 ) has a prim-

itive, by Corollary 5.1.6,
∫
Γ n [f (z 0 ) − f j (z 0 )(z − z 0 )]dz = 0 ∀n ∈N. (4)

Hence, by (1)-(4),
∫
∫
f (z)dz ≤ 4 n. f (z)dz ≤ ε4 n[A(Γn)] =
2 εA(Γ).
. . .
Γ Γn
∫
This is true for every ε > 0. Hence, f (z)dz = 0.
Γ

Corollary 5.2.6 Conclusion in Theorem 5.2.5 holds if the positive

oriented triangle Γ is replaced by a rectangle.
Theorem 5.2.7 Every holomorphic function on a simply connected
domain has a holomorphic primitive.
Proof. Let Ω be a simply connected domain and f be a holo-
morphic function on Ω. We show that there exists a holomorphic
function g on Ω such tha t g j (z) = f (z) for every z ∈ Ω.
Let z 0 ∈ Ω, and z be any arbitrary point in Ω. By Lemma 5.2.4
there exists a rook-path Γ z0,z joining z 0 to z. Define
∫
g(z) = f (ζ)dζ.
Γ z 0 ,z
70 Integration

In view of Corollary 5.2.6, the integral along any two rook-paths

joining z 0 to z will give the same value. Hence, g is well defined on Ω
as long as we restrict the integration along rook-paths joining z 0 to z.
Further, if δ > 0 and if B(z, δ) ⊆ Ω, then for any h ∈ C with z + h
∈B(z, δ), we have
∫
g(z + h) = f (ζ)dζ.
Γz 0,z +h

Hence, by Corollary 5.2.6, we obtain

g(z + h) − g(z) ∫
1
− f (z) = [f (ζ) − f (z)]dζ.
h h Γz ,z + h
∫
Here, we also used the fact that dζ = h, by Proposition 5.1.5.
Γz ,z + h
Also, by Theorem 5.2.5,
∫ ∫
1 1
[f (ζ) − f (z)]dζ = [f (ζ) − f (z)]dζ,
h Γz ,z + h h Ch

where Ch is the straight line segment joining z to z + h. Hence, by

Proposition 5.1.4,
g(z + h) − g(z) A(Ch)
. − f (z). = max |f (ζ) − f (z)|.
h |h| ζ∈Ch
Now, let ε > 0 be give. Since f is uniformly continuous on cl (Ω), for
every ε > 0, there exists δ > 0 such that

z + h ∈ Ω, |h| < δ =⇒ max |f (ζ) − f (z)| < ε.

ζ∈C h

Thus,
g(z + h) − g(z)
. − f (z). < δ whenever |h| < δ.
h
Consequently, g is differentiable at z and g j (z) = f (z). This com-
pletes the proof.

Proof of Theorem 5.2.3. The proof follows from Theorem 5.2.7

and Corollary 5.1.6.

If we observe the proof of Theorem 5.2.7, it is apparent that we

have actually proved the following.
Cauchy’s Integral Formulas 71

Theorem 5.2.8 Suppose f is a continuous function on a simply

connected domain Ω and
∫
Γ
f (z)dz = 0

for every positively oriented triangle Γ in Ω. Then f has a holomor-

phic primitive on Ω.

5.3 Cauchy’s Integral Formulas

Theorem 5.3.1 (Cauchy’s integral formula) Suppose f is an-
alytic on and inside a simple closed contour Γ, and let ΩΓ be the
domain enclosed by Γ. Then for every z ∈ΩΓ,
1 ∫ f (ζ)
f (z) = dζ.
2πi Γ ζ − z

Proof. Let z ∈ ΩΓ and let C := C r be a positively oriented circle

with centre at z and radius r such that C r ⊆ ΩΓ. Take two points on
C r and join with Γ by a line segments, say L 1 and L 2 . More
precisely, take two points z 1 , z 2 ∈C r and join with ζ 1 , ζ2 ∈Γ such
that
|z1 − ζ1| = dist (z 1 , Γ), |z2 − ζ2| = dist (z 2 , Γ).

Cut Γ into into two pieces Γ 1 , Γ 2 at the points ζ 1 , ζ 2 , and cut C

into into two pieces C 1 , C2 a t the points z 1 , z 2 retaining the original
orientations. Note that the function ζ ›→ f (ζ)/(ζ − z) is analytic in
the region between Γ and C, Hence by Cauchy’s theorem, integral of
f (ζ)/(ζ − z) over the curves Γ 1 + L 1 + C˜1+ L 2 and Γ2 + L˜2 + C˜2+ L˜1
are zeros. Therefore, we have
∫ ∫ f (ζ)
1 f (ζ) 1
dζ = dζ.
2πi Γ ζ − z 2πi C ζ − z
1
∫ dζ
Since 2πi Γ ζ−z = 1, we have
∫ ∫
1 f (ζ) 1 f (ζ) − f (z)
dζ − f (z) = dζ. (∗)
2πi C ζ −z 2πi C ζ −z

Since
f (ζ) − f (z)
. . → |f j(z)| as ζ → z,
. ζ −z .
72 Integration

there exists δ > 0 such that

f (ζ) − f (z)
. . → |f j (z)| + 1 whenever |ζ − z| < δ.
. ζ −z .

Now, taking r < δ, we have

∫
f (ζ) − f (z)
. dζ ..≤ (|f j(z)| + 1)2πr. (∗∗)
. C ζ −z

From (∗) and (∗∗), we have

∫
. 1 f (ζ)
dζ − f (z).. ≤ (|f j(z)| + 1)r.
. 2πi Γ ζ −z

This is true for all r such that 0 < r < dist (z, Γ). Hence,

1 ∫ f (ζ)
dζ = f (z),
2πi Γ ζ − z

completing the proof.

Corollary 5.3.2 Suppose f is holomorphic on { z ∈ C : |z −z0 | ≤ r } .

Then
1 ∫ 2π
f (z 0 ) = f (z 0 + re it) dt.
2π 0

Our next attempt is to show that if f is analytic at a point z 0 ,

then in neigbourhood of z 0 , f can be expressed as a power series.

Proposition 5.3.3 Let C be a circle with centre z 0 and ΩC be the

domain enclosed by C. Let g be continuous on C and let
∫ g(ζ)
ϕ(z) = dζ, z ∈ Ω C.
C ζ −z

Then for every z ∈ ΩC ,

Σ∞ ∫ g(ζ)
ϕ(z) = an(z − z 0) n, an:= dζ.
C (ζ − z 0 ) n+ 1
n= 0
Cauchy’s Integral Formulas 73

Proof. Let ζ ∈ C and z lies inside C. Since |z − z0| < |ζ − z0 |, we

have
1 1
=
ζ −z (ζ − z 0 ) − (z − z 0 )
1
=
(ζ − z 0 )[1 − z−z
ζ−z 0 ]
0

∞ . Σ
1 Σ z −z 0 n
= .
ζ − z 0 n= 0 ζ − z 0

Since the above series converges uniformly on C, we have

∫ Σ∞ ∫ g(ζ)
g(ζ)
dζ = an(z − z 0) n, an:= dζ.
C ζ −z C (ζ − z 0 )
n+ 1
n= 0

This completes the proof.

Theorem 5.3.4 Suppose f is analytic at a point z 0 . Then f is
infinitely differentiable in a neigbourhood D 0 of z 0 and we have the
following.
(i) (Talor series expansion)
Σ∞ f (n) (z
0)
f (z) = (z − z0 ) n, z ∈D0 (∗)
n!
n= 0

in that neigbourhood.
(ii) (Cauchy’s integral formula of higher orders)
n! ∫ f (ζ )
(n)
f (z 0) = dζ,
2πi Γ (ζ − z 0 ) n+ 1
where Γ is a simple closed curve enclosing z 0 such that f is analytic
on and inside Γ.
Proof. Using Proposition 5.3.3 taking g = f and applying Cauchy’s
integral formula (Theorem 5.3.1), we have
Σ∞ 1 ∫ f (ζ )
f (z) = an(z − z 0) n, an:= dζ,
2πi C (ζ − z 0 ) n+ 1
n= 0

for z in a neigbourhood of z 0 , where C is a circle with centre at z 0

lying inside that neigbourhood. Since the function
f (ζ)
ζ ›→
(ζ − z0) n+ 1
74 Integration

is analytic on Ω ∩ c Cauchy’s theorem (Theorem 5.2.3), we

Γ Ω , by
C
obtain (How?)
∫ 1 ∫ f (ζ )
1 f (ζ )
dζ = dζ.
2πi C (ζ − z 0 ) n+ 1 2πi Γ (ζ − z 0 ) n+ 1
Now, from the discussion at the end of Chapter 4, we know that
f (n) (z 0 )
an = .
n!
This completes the proof.

Deftnition 5.3.1 The series expansion (∗) of f is called the Taylor

series expansion of f around z 0 . ♦
By the above theorem derivative of a holomorphic function is
holomorphic. Hence, in view of Theorem 5.2.8, we have a converse to
the Cauchy’s theorem.
Theorem 5.3.5 (Morera’s theorem) Suppose f is a continuous
function on a simply connected domain Ω and
∫
Γ
f (z)dz = 0

for every positively oriented triangle Γ in Ω. Then f is holomorphic

on Ω.
Remark 5.3.1 In fact, the conventional Morera’s theorem is slightly
weaker form of the above theorem, namely, if∫f is a continuous func-
tion on a simply connected domain Ω and Γ f (z)dz = 0 for every
closed contour Γ in Ω, then f is holomorphic on Ω. ♦
Exercise 5.3.1 Suppose f is an entire function, M > 0, R > 0 and
n ∈ N such that
n
|f (z)| ≤ M |z | ∀z with |z| ≥ R.
Show that f is a polynomial of degree atmost n. a
Exercise 5.3.2 Suppose f is holomorphic for |z| < 1 and
|z| < 1 =⇒ |f (z)| < 1.
Show that 1
|f j(z)| ≤ .
1 −|z|
a
Cauchy’s Integral Formulas 75

Theorem 5.3.6 (Liouville’ therem) Suppose f is an entire func-

tion, i.e., f is holomorphic on the entire C. If f is bounded, then f
is a constant function.
Proof. Let z ∈ C. Then, by Theorem 5.3.4,

j
1 ∫ f (ζ)
f (z) = dζ,
2πi Γ R (ζ − z) 2

where Γ R is the circle with centre z 0 and radius R. Hence, we have

M A(Γ R ) M
|f j(z)| = = .
2πR 2 R
Now, letting R → ∞, we have f j (z) = 0. This is true for all z ∈ C.
Hence, f is a constant function.

Theorem 5.3.7 (Fundamental theorem of algebra) Suppose

p(z) is a nonconstant polynomial with complex coefficients. Then
there exists z 0 ∈C such that p(z 0 ) = 0.
Proof. For a0 ƒ = 0, let p(z) = a 0 z n + a 1 z n−1 + . . . + a n−1 z + a n .
Suppose p(z) ƒ = 0 for all z ∈C. Then, for z ƒ = 0,
. Σ
n a1 an−1 an
p(z) = a 0 z 1+ + ... + + .
a0z a 0 z n−1 a0zn
Hence,
. Σ
|a1| |an−1| |an|
|p(z)| ≥ |a0z | 1−
n
+ ... + + .
|a0z| |a0zn−1| |a0zn|
Let R > 0 be such that
|a1| |an−1| |an | 1
|z| ≥ R =⇒ + ...+ + ≤ .
|a0 z | |a0zn−1| |a0zn| 2
Then, for |z| ≥ R, we have
1 2 2
≤ ≤ .
|p(z)| |a0z n | |a0|R n
Since 1/p(z) is bounded for |z| ≤ R, it then follows that 1/p(z) is a
bounded entire function. Hence, by Liouville’s theorem, 1/p(z) is a
constant function, so that p(z) is a constant polynomial, which is a
contradiction to our assumption on p(z).
76 Integration

5.3.1 Appendix
We state again the Cauchy’s integral formula for the derivatives of
analytic functions, and give another proof for the same.

Theorem 5.3.8 Suppose f is analytic on an inside a positively ori-

ented simple closed curve Γ and z lies inside Γ. Then, for every n ∈
N ∪{0}, f (n) (z) exists and
n! ∫ f (ζ )
f (n) (z) = dζ.
2πi Γ (ζ − z) n+ 1
Proof. We know that the above result is true for n = 0. Assume
it for n − 1. We shall prove it for n. Let ρ > 0 be such that the circle
with centre z and radius ρ lies inside Γ. Then for h ∈C with |h| < ρ,
∫
f (n−1) (z + h) − f (n) (z ) n! f (ζ )
− dζ
h 2πi Γ (ζ − z) n+ 1

takes the form

∫ Σ Σ
(n − 1)! f (z) (ζ − z)n − (ζ − z − h) n n
− dζ.
2πi Γ h (ζ − z − h) n (ζ − z) n (ζ − z) n+ 1

Now, writing a = ζ − z and b = ζ − z − h), we have

(ζ − z ) n − (ζ − z − h) n n an − bn n
− − n+ 1 =
h(ζ − z − h) (ζ − z)
n n (ζ − z) n+ 1n
ha b n a
a(a − b ) −nhbn
n n
=
ha n+1 b n
Σ Σ n−1 n−j j
Since a(an −bn) = a(a −b) n−1
j=0 a
n−1−j j
b= h j=0 a b , we have

n−1
Σ
a(an − bn) − nhbn = hbn−1−j (a j + 1 − bj + 1 )
j=0
n−1
Σ
= h 2b n−1−j [aj + a j −1 b + . . . + abj −1 + bj]
j=0

Since |a| = ρ and ρ −|h| ≤ |b| ≤ ρ +|h|, taking α = ρ +|h| we obtain,

n−1
Σ
|a(a n − bn ) − nhbn | ≤ |h|2 α n−1−j [ρ
j
+ r j −1 α + . . . + ρα j −1 + α j ].
j=0
Zeros of analyticfunctions 77

Thus, absolute value of

Σ Σ
f (z) (ζ − z)n − (ζ − z − h) n n
−
h (ζ − z − h) n (ζ − z) n (ζ − z) n+ 1

is less than or equal to

Σ
M |h| n−1
j=0 α
n−1−j
[ρ j+ r j −1 α + . . . + ρα j −1 + α j ]
εh := .
ρ n+ 1 (ρ −|h|)n
Hence,
∫
.f
(n−1) (z + h) − f (n) (z ) − n! f (ζ )
dζ .≤ ε ρ.
h
. h 2πi Γ (ζ − z) n+ 1 .

Since εh → 0 as |h| → 0, it follows that f (n−1) is differentiable at z

and
∫
(n) (z) f (n−1) (z + h) − f (n) (z) = n! f (ζ ) dζ.
f = lim
|h|→0 h 2πi Γ (ζ − z) n+ 1

This completes the proof.

5.4 Zeros of analytic functions

Suppose f is analytic in a domain Ω and z 0 is a zero of f , i.e., f (z 0 )
= 0. Then, using the Taylor series expansion of f around z 0 , it
follows that
f (z) = (z − z 0 )g(z)
in a neigbourhood of z 0 , where g is analytic in a neigbourhood of z 0 .
In fact,
Σ∞ f (n) (z 0)
g(z) = f j(z 0) + (z − z0) n−1
n!
n= 2

in a neigbourhood of z 0 . Note that, if f j (z 0 ) = 0, then we can write

2
f (z) = (z − z 0 ) g 1 (z),

where
f (2) (z
0)
Σ∞ f (n) (z
0)
g(z) = + (z − z0) n−2
2 n!
n= 3
in a neigbourhood of z 0 .
78 Integration

Deftnition 5.4.1 (i) A point z 0 ∈ Ω is said to be a zero of f of

order m if f (k) (z 0 ) = 0 for k = 0, 1, . . . , m−1 and f (m) (z 0 ) ƒ = 0.

(ii) A point z 0 ∈ Ω is said to be a zero of f of ftnite order if it is a

zero of f of order m for some m ∈N.

(iii) A zero of f which is not of finite order is called a zero of f of

inftnite order.
♦
Deftnition 5.4.2 A zero z 0 ∈ Ω of an analytic function f is said to
be an isolated zero if there exists a r > 0 such that B(z 0 , r) ⊆ Ω
and f (z ) ƒ= 0 for every z ∈ B (z 0 , r) \ { z 0 } . ♦

Remark 5.4.1 We observe the following:

(a) If z 0 ∈ Ω is a zero of f of order m, then

f (z) = (z − z 0 ) mg(z)

in a neigbourhood of z 0 , where g is analytic in a neigbourhood

of z 0 and g(z 0 ) ƒ = 0.

(b) If z 0 ∈ Ω is a zero of f of infinite order, then f (k) (z 0 ) = 0 for

all k ∈ N; consequently, f = 0 in a neigbourhood of z 0 .

Exercise 5.4.1 Prove the statements in the above remark. a

Theorem 5.4.1 Every zero of finite order of an analytic function is

isolated.

Proof. Follows from Remark 5.4.1(a).

Theorem 5.4.2 Suppose f is analytic in an open connected set Ω. If

Ω contains a zero of f of infinite order, then f = 0 on Ω.

Proof. Follows from Remark 5.4.1(b) using the fact that Ω is open
and connected.
Zeros of analyticfunctions 79

5.4.1 Identity theorem

Theorem 5.4.3 (Identity theorem-I) Suppose f is analytic in an
open connected set Ω. If Ω contains a point which is the limit point of
a set of zeros of f , then f = 0 on Ω.

Proof. Follows from Theorem 5.4.2.

Theorem 5.4.4 (Identity theorem-II) Suppose f and g are an-
alytic in an open connected set Ω. If f = g on a set having a limit
point in Ω, then f = g on Ω.

Proof. Follows from Theorem 5.4.3.

EXAMPLE 5.4.1 Let f be analytic in { z ∈C : |z| < 1} and

. 1 Σ 1
f = ∀n ∈N,
n+ 1 n+ 1
and . 1Σ
g = 0 ∀n ∈N.
n +1
Then, by Theorem 5.4.4, f (z ) = z and g(z ) = 0 for all z . Q
EXAMPLE 5.4.2 We show that there is no analytic function f on
Ω := { z ∈ C : |z| < 1} satisfying
. 1 Σ (−1)n
f = ∀n ∈N.
n n2
Suppose there is an analytic function f satisfying the above require-
ments. Then we have
. 1Σ 1
f = ∀n ∈N,
2n (2n) 2
. 1 Σ 1
f = − ∀n ∈N.
2n −1 (2n −1)2
Then, by Theorem 5.4.4, we have f (z) = z 2 and f (z) = −z 2 for all
z ∈ Ω, which is not possible. Q
EXAMPLE 5.4.3 Suppose Ω is a connected (nonempty) open set
which is symmetric with respect to the real axis, i.e., z ∈ Ω ⇐⇒ z¯ ∈
Ω. Suppose f is holomorphic on Ω such that it is real on Ω ∩ R. We
show that f (z¯) = f (z).
80 Integration

It can be shown that g defined by g(z) := f (z¯) is analytic on Ω

(verify). For z ∈ Ω ∩ R, we have g(z) = f (z). Sine f is real on Ω ∩ R,
we have g(z) = f (z) for all z ∈ Ω ∩ R. Hence, by Theorem 5.4.4, f
= g on Ω so that f (z¯) = f (z ) on Ω. Q

5.4.2 Maximum modulus principle

Theorem 5.4.5 (Maximum modulus principle) Suppose f is
holomorphic in a connected open set Ω. If f is not a constant func-
tion, then |f | cannot attain maximum at a point in Ω.
In other words, if there exists z 0 ∈ Ω such that |f (z)| ≤ |f (z0 )|
for all z ∈Ω, then f is a constant function.
Proof. Suppose there exists z 0 ∈ Ω such that |f (z)| ≤ |f (z0 )| for
all z ∈Ω. Let r > 0 such that { z ∈C : |z − z0| ≤ r } ⊆ Ω. Recall
that
1 ∫ 2π
f (z 0 ) = f (z 0 + reit)dt.
2π 0
If f (z 0 ) = 0, then we have f (z) = 0 for all z ∈ Ω. Hence, assume
tha t f (z 0 ) ƒ = 0, and let λ := |f (z 0 )|/f (z 0 ). Then we have
∫ 2π ∫ 2π
1 1
|f (z0 )| = λ f (z 0 ) = it it
2π 0 λ f (z 0 +re )dt = 2π 0 Re[λf (z 0 +re )]dt.
Hence, ∫ 2π
1
[|f(z0 )| − Re(λf (z 0 + re it))]dt = 0.
2π 0
Since |Re(λf (z 0 + reit)| ≤ |f (z0 )|, we have
|f (z0 )| = Re(λf (z 0 + re it)) ∀t ∈ [0, 2π],
i.e.,
|f (z0 )| = Re[λf (z)] ∀z ∈ C r := {ζ : |ζ − z0| = r}.
Again, since |λf (z)| ≤ |f (z)| for all z ∈ C r , we have
|f (z0 )| = λ f (z) ∀z ∈C r .
Hence,
f (z) = f (z 0 ) ∀z ∈ C r .
By, identity theorem, f (z) = f (z 0 ) for all z ∈ Ω.
Exercise 5.4.2 Suppose f is holomorphic in a connected open set
Ω. If f is not a constant function and |f | attains minimum at z0 ∈ Ω,
then f (z 0 ) = 0. a
Zeros of analyticfunctions 81

5.4.3 Schwarz’s lemma

Theorem 5.4.6 (Schwarz’s lemma) Suppose f is holomorphic in
the open unit disc D := { z ∈ C : |z| < 1} such that

f (0) = 0 and f (z) ∈ D ∀z ∈ D.

Then |f (z)| ≤ |z| for all z ∈ D. Strict inequality follows unless f is

of the form f (z) = λ z for some λ ∈ C.

Proof. By the assumptions on f , we have

f (z) = zg(z), z ∈ D,

where g is holomorphic on D and g(0) = f j(0). Since f (z)| ≤ 1, we

have 1 1
|g(z)| ≤ = whenever |z| = r < 1.
|z| r
By maximum modulus principle,
1
|g(z)| ≤ whenever |z| ≤ r < 1.
r
Now, let z ∈ D, and 0 < r < 1 such that |z| ≤ r. By the above
arguments,
1
|g(z)| ≤ .
r
Letting r → 1, we obtain |g(z)| ≤ 1. Thus, |f (z)| ≤ |z| for all
z ∈ D.

5.4.4 On harmonic functions

Theorem 5.4.7 Suppose u is real harmonic in a simply connected
domain Ω. Then u has a harmonic conjugate which is unique up to
addition of an imaginary constant.

Theorem 5.4.8 Suppose u is real harmonic in a simply connected

domain Ω. Then u ∈ C ∞ (Ω).

Theorem 5.4.9 Suppose u is real harmonic in an open set Ω. Then,

for z 0 ∈Ω,
1 ∫ 2π
u(z 0 ) = u(z 0 + re it )dt, 0 < r < dist (z 0 , C \ Ω).
2π 0
82 Integration

Theorem 5.4.10 Suppose u is real harmonic in a connected open set

Ω. If u is not a constant function, then |u| cannot attain maximum at
a point in Ω.
In other words, if there exists z 0 ∈ Ω such that |u(z)| ≤ |u(z0 )|
for all z ∈Ω, then u is a constant function.

Theorem 5.4.11 Suppose u is real harmonic in an open connected

set Ω. If f = 0 on an open set Ω0 ⊆ Ω, then f = 0 on Ω.

5. Problems
1. A curve η : [α, β] → C is called a reparameterization of the
curve γ : [a, b] → C if there exists continuous increasing bijec-
tion ϕ : [α, β] → [a, b] such that η = γ ◦ ϕ, and in that case we
may say that η is equivalent to γ and we write η ∼ γ.

(a) Given a curve γ : [a, b] → C and a closed interval [α, β],

find a curve η : [α, β] → C such γ ∼ η.
(b) If γ˜ is the reverse of γ : [a, b] → C and η : [−b, −a] → C is
defined by η(t) = γ(−t) for −b ≤ t ≤ −a, then show that
η ∼ γ˜.
(c) If γ and η are piecewise smooth curves such tha t η ∼ γ,
then show that A(Γη ) = A(Γγ ).

2. Given a piecewise smooth curve γ : [a, b] → C and a parti-

tion Π := a = t 0 , < t 1 < . . . < t n = b of [a, b], let S n :=
Σ n n
j = 1 |γ(t j ) − γ(t j−1 )|. If max 1≤j≤n |t j − t j−1 | → 0, then show
∫
that S n → ab |γ j(t)|dt.

3. If η is a differentiable reparamterization of a piecewise smooth

∫
curve γ, and if f is continuous on Γ γ , then show that γ f (z)dz =
∫
η f (z)dz.
∫ ∫ ∫
4. Prove: f (z) dz = f (z) dz + f (z) dz.
γ1+ γ2 γ1 γ2

5. Given a piecewise smooth curve γ : [a, b] → C, a partition

Πn := a = t 0 , < t 1 < . . . < t n = b of [a, b], and a continuous
Σ n
function f on Γ, let S n( f ) := j = 1 f (γ(t j ))[γ(t j ) − γ(t
∫
j−1 )]. If
max 1≤j ≤n |tj − t j −1 | → 0, then show that S n ( f ) → Γ f (z)dz.
Problems 83

6. If ϕ : [a, b] → C is continuously differentiable, then show that

∫ b
ϕ j(t) dt = ϕ(b) − ϕ(a).
a

7. Let Γ be a closed curve and f has a primitive on an open set Ω

containing Γ, i.e., there exists a continuously differentiable
∫
function g such tha t gj = f on Ω. Then f (z) dz = 0. Justify.
Γ

8. Let Γ n be the circle with center at z 0 and radius r traced n

times,
∫ i.e., Γ n := { z 0+ re it, 0 ≤ t ≤ 2nπ}. Then, show that
dz
= 2nπi,.
Γn z − z0
∫
9. If p(z) is a polynomial, then prove that p(z) dz = 0.
Γn

10. If∫ Γ is the circle with centre z 0 and radius r, then show that
dz
= 0 for every n ∈ Z \{−1}.
Γ (z − z0 )n
11. Given distinct points α and β in C, evaluate the integrals
∫ n
∫ n
[α,β] z dz and [α,β] z¯dz, where [α, β] denotes the line segment
joining α to β.
12. If f is a real valued function defined on the interval [a, b] and
∫ ∫
if γ(t) = t, a ≤ t ≤ b, then show that [a,b] f (z)dz = ab f (t)dt.

13. Let Γ be a closed piecewise smooth curve in the disc of conver-

Σ ∞
gence of a power series a (z − z0 ) nand let f (z) represent
n= 0 ∫ n
this series in that disc. Then f (z) dz = 0. Justify.
Γ
∫
14. Evaluate the integrals γ f (z)dz in following, where

( a ) γ is the curve joining z 1 = −1 − i to z 2 = 1 + i consisting

of the line segment from −1−i to 0 and the . portion of the
1, y < 0, .
curve y = x 2 from 0 to 1 + i and f (z) =
4y, y > 0
Answer: − +5 i5
8 2
(b) γ is the curve consisting of the line segments joining the
points 0 to 1 and 1 to 1 + 2i and f (z) = 3x 2 − y + ix 3 .
Answer: 2 + 3i
84 Integration

(c) f (z) = z¯and γ is the curve joining 1 to 1 + i along the

parabola y = x 2 . Answer: 1 + 13 i
(d) γ is the positive oriented circle z − 1| = 4 traced once and
1 +
f (z) = z−1 6
(z−1) 2
Answer: −6πi

15. Show that ∫

.
z
γ z2+1
dz .≤ , 12where γ be the line segment joining
2 to 2 + i.
16. For piecewise smooth curve γ :: [a, b] → C and continuous
∫ ∫ b
function f : Γ → C, define f (z)|dz| := f (g(t))|γ j(t)|dt.
γ a
Show tha t ∫ ∫
f (z)dz ≤ |f (z)| |dz|.
. .
γ γ

17. If Γ r is the circle γ(t) = z 0 + re it , 0 ≤ t ≤ 2π and if f is

continuous on and inside Γ r , then prove tha t

1 ∫ f (z)
lim dz = f (z 0).
r→0 2πi Γ r z − z 0

∫
e−z dz = 0.
2
18. For every closed piecewise smooth curve Γ,
Γ
Why?

19. For positive real numbers, let I 1 , I 2 , I 3 , I 4 be the integrals of

e−z over the line segments
2

[−a, a], [a, a + ib], [a + ib, −a + ib], [−a + ib, −a],

respectively. Prove that

∫ a
e−x dx,
2
(i) I 1 =
−a

(ii) |I |2 ≤ be −a2 +b 2
.
∫ a
(iii) I 3 = −e b2
e −t cos(2bt)dt,
2

−a

(iv) |I |4 ≤ be −a2 +b 2
.
∫ ∞ ∫ ∞
−t 2
b2
e−x dx.
2
(v) e e cos(2bt)dt =
−∞ −∞
Problems 85

20. Suppose f n → f uniformly on Γ. Then show that

∫ ∫
f n (z)dz → f (z) dz.
Γ Γ

∫ . √
∞
−t 2 1√ 2
21. Prove tha t e cos t dt = π 1 + 2, by integrating
0 4
e −z 2
over the positive oriented triangle with vertices at 0, R, Re iπ/8
for R > 0 and letting R → ∞.
∫ ∞ ∫ ∞
2
22. Evaluate the integrals cos t dt, cos t2dt by integrat-
0 0
ing e−z over the positive oriented triangle with vertices a t
2

0, R, Re iπ/4 for R > 0 and letting R → ∞.

23. Let (Ωn ) be a sequence of nonempty compact sets in C such

that Ωn ⊇ Ω n+1 for all n ∈ N and diam(Ω n ) → 0 as n → ∞.
Let z0 ∈ ∩∞n=1Ωn. If f is a continuous function defined on Ω1,
show that

max |f (z) − f (z0 )| → 0 as n → ∞.

z∈Ωn

24. Suppose Ω be a simply connected domain and f be holomorphic

on Ω. Suppose integral of f over every positively oriented trian-
gle is zero. Prove that if Γ 1 and Γ 2 are two polygonal lines join-
∫ ∫
ing any two points z 0 and ζ0 in Ω, then Γ 1 f (z)dz = Γ 2 f (z)dz.

25. Let f be continuous in a neigbourhood of z 0 and Γ r := { z ∈

C : |z − z0| = r}. Show that

1 ∫ f (z)
dz → f (z 0 ) as r → 0.
2πi Γ r z −z 0

∫ 2π
dθ
26. Evaluate the integral (using complex in-
0 1 − 2r cos θ + r 2
tegrals).

27. .Let f . be an entire function such tha t for some n ∈ N and R > 0,
. z n . is bounded for |z| > R. Prove tht f is a polynomial of
f (z)

degree atmost n.
86 Integration

28. Let f be holomorphic and map Ω := { z ∈ C : |z| < 1} into

itself. Prove that |f jz)| ≤ 1/(1 − |z|) for all z ∈ Ω.

29. Prove tha t there is no analytic function f on Ω:= { z ∈ C :

|z| < 1} such that
(i) f (1/n) = (1/2 n for n ∈ N \ {1}.
(ii) f (1/n) = (−1) n /n 2 for n ∈ N \ {1}.

30. Let f be a nonconstant holomorphic function in a connected

opens set Ω. If z 0 ∈ Ω is such that |f (z0 )| ≤ |f (z)| for all z ∈ Ω,
then prove that f (z 0 ) = 0.

31. Let u be a (real valued) harmonic function in a connected opens

set Ω. Let g := u x − iuy on Ω. Justify the following:
(i) g is holomorphic on Ω.
(ii)There exists a holomorphic function f on Ω such that Ref
= u.
(iii) u is infinitely differentiable.
6
Laurent Series and Isolated
Singularities

6.1 Laurent Series

In this chapter we consider series expansions of functions which are
analytic in an annulus. Such series are known as Laurent series.
We know that
∞
1 Σ n
= z
1 −z
n= 0

whenever |z| < 1. We also have

∞
1 −1 Σ
= = (−1/z) (1/z n)
1−z z(1 −1/z)
n= 0

whenever |z| > 1. Note also tha t

1 1 1
= −
(z − 1)(z − 2) z − 1 z −2
Σ∞ (−1) Σ∞ (−1)
= + zn
zn 2 n+1
n= 1 n= 0

whenever 1 < |z| < 2.

In view of the above examples, for a function f which is analytic
in an annulus { D := { z ∈ C : R 1 < |z − z0| < R 2 } , we may look for
an expansion of f of the form

Σ∞ n
a n (z − z 0 ) .
n= −∞

87
88 Laurent Series and IsolatedSingularities

Σ ∞
n= −∞ a n(z −z 0) for a given
Deftnition 6.1.1 A series of the form n

z 0 ∈ C and sequences (a n ) and a −n ) in C is said to converge at a point

Σ Σ
z =ƒ z 0 if both the series ∞n=1 a −n (z − z 0 ) −n and ∞n=1 a n (z − z 0 ) n
Σ ∞
n= −∞ a n(z − z 0) converges for all z in
converge at z. If the series n

a subset Ω of C, then we say that it converges in Ω. ♦

Note that
Σ∞
• the series a−n (z − z 0 ) −n converge for all z such that
n= 1

|z − z0| > R 1 := lim sup |a−n| 1/n

n∈N

and
Σ∞
• the series an(z − z0) n converge for all z such that
n= 1

1
|z − z0| < R 2 := .
lim sup n∈N |an| 1/n

Σ∞ n
Here, R 1 can be 0 and R 2 can be ∞. Thus, the series a n (z−z 0 )
n= −∞
converges in the annulus z such tha t R 1 < |z − z0| < R 2 .
Questions: Consider an annulus

D := { z ∈ C : R 1 < |z − z0| < R 2 }.

Σ ∞
• Suppose a series n= −∞ a n(z − z 0) converges in D. Then, is
n

the function represented by this series is holomorphic in D? In

such case, are the coefficients uniquely determined?

• If f is holomorphic in D, does it have a series expansion of the

Σ ∞
form n= −∞ an(z − z 0) nin D?
We answer both the above questions affirmatively.
Σ ∞
n= −∞ a n(z − z 0) converges in an annu-
Theorem 6.1.1 Suppose n

lus D := { z ∈C : R 1 < |z − z0| < R 2 } and let

Σ∞ n
f (z) = a n (z − z 0 ) , z ∈D.
n= −∞
Laurent Series 89

Then f is holomorphic in D and

∫
1 f (ζ
an= dζ ∀n ∈Z,
2πi Γ (ζ − z 0 ) n+ 1

where Γ is any simple closed contour in D enclosing z 0 .

Proof. Let Γ r := { z : |z − z0| = r}, R 1 < r < R 2 . Since f is
uniformly continuous on Γ r ,
∫ Σ∞ ∫
1 1
f (z )dz = an (z − z0 ) ndz.
2πi Γr 2πi Γr
n= −∞

Hence, ∫
1
a−1 = f (z)dz.
2πi Γr
Also, for k ∈ N, we have

f (z ) Σ∞
= a n (z − z 0 ) n−k−1 , z ∈D.
z −z 0 ) k+ 1 n= −∞

Hence,

1 ∫ f (z ) Σ∞ 1
∫
dz = an (z − z0) n−k−1 dz
2πi Γr z − z 0 ) k+ 1 2πi Γr
n= −∞

so that (since n − k − 1 = −1 ⇐⇒ n = k)
1 ∫ f (z )
ak= dz.
2πi Γ r z −z 0 ) k+ 1

By Cauchy’s theorem, Γ r can be replaced by any cure Γ as in the

theorem.

By the above theorem the coefficients of the convergent series

Σ ∞
n= −∞an(z − z 0) nis uniquely determined by the function which it
represents. In view of this fact, we have the following definition.
Σ ∞
n= −∞ a n(z − z 0) converges
Deftnition 6.1.2 Suppose the series n

in an annulus D := { z ∈ C : R 1 < |z − z0| < R 2 } and let f be the

function represented by the series in D. Then the series is said to be
the Laurent series of f in D . ♦
90 Laurent Series and IsolatedSingularities

Proposition 6.1.2 Suppose ϕ is continuous on piecewise smooth

curve Γ and
∫ ϕ(ζ)
ψ(z) := dζ ∀z ƒ∈Γ.
Γ ζ −z

Then ψ is holomorphic in C \ Γ and ψ has the series expansions

Σ∞
an(z − z 0) n whenever |z − z0| < inf |ζ − z0 |,
ζ ∈Γ
n= 0

Σ∞
bn(z − z 0 ) −n whenever |z − z0| > sup |ζ − z0 |,
ζ ∈Γ
n= 1

where ∫
f (ζ
an= dζ ∀n ∈Z,
Γ (ζ − z 0 ) n+ 1
with a−n = bn for n ∈ N.

Proof. We note that for ζ ƒ = z,

1 1 −1
= =
ζ −z (ζ − z 0 ) − (z − z 0 ) (z − z 0 ) − (ζ − z 0 )
−1
= Σ Σ.
−z0
(z − z 0 ) 1 − ζz−z 0

Hence, for |z − z0| > maxζ∈Γ |ζ − z0 |,

∞ ∞
1 = −1 Σ . ζ − z 0 Σn Σ (ζ − z 0 ) n−1
= (−1)
ζ −z z −z 0 z −z 0 (z − z0) n
n= 0 n= 1

T hus,
∞
ϕ (z ) = Σ (−1)ϕ(z) (ζ − z 0 ) n−1
ζ −z (z − z0) n
n= 1

so tha t
∫ Σ∞
ϕ (z ) −n
dζ = bn(z − z 0)
Γ ζ −z
n= 1

where ∫
bn = − ϕ(z)(ζ −z 0 ) n−1 dζ, n ∈N.
Γ
Laurent Series 91

Also, for ζ ƒ = z,
1 1 1
= = Σ Σ.
ζ −z (ζ − z 0 ) − (z − z 0 ) (ζ − z0) 1− z−z0
ζ −z0

Hence, for |z − z0| < infζ∈Γ |ζ − z0 |,

1 = 1 Σ∞ . z −z 0 Σn 1 Σ
∞
(z −z 0 ) n
= .
ζ −z ζ −z0 ζ − z0 ζ −z0 (ζ − z0) n+ 1
n= 0 n= 0

T hus,
∫ Σ∞
ϕ (z )
dζ = an(z − z 0) n
Γ ζ −z
n= 1
where ∫
ϕ (z )
an= dζ, n ∈N0:= N ∪{0}.
Γ (ζ − z 0 ) n+ 1
This completes the proof.
Theorem 6.1.3 Suppose f is holomorphic annulus D := { z ∈ C :
R 1 < |z − z0| < R 2 } . Then f has a series expansion
Σ∞
a n (z − z 0 ) n on D,
n= −∞

where ∫
1 f (ζ) dζ ∀n ∈Z.
an=
2πi Γ (ζ − z 0 ) n+ 1
Proof. Let R 1 < r < R < R 2 and let C := {ζ : |ζ − z0| = r } and
Γ := {ζ : |ζ − z0| = R } . Then it can be seen that (verify!) for every
z with r < |z − z0| < R,
1 ∫ f (ζ) 1 ∫ f (ζ)
2πi Γ ζ − z dζ − 2πi C ζ − z
f (z) = dζ = f 1(z) + f 2(z).

By Proposition 6.1.2, we have

∫ Σ∞
1 f (ζ ) −n
f 1 (z) = dζ = bn(z − z 0)
2πi Γ ζ − z
n= 1

and ∫
1 f (ζ ) Σ∞
f 2 (z) = dζ = an(z − z 0) n,
2πi C ζ −z
n= 1
92 Laurent Series and IsolatedSingularities

where, ∫
bn = f (z)(ζ − z 0 ) n−1 dζ, n ∈N,
Γ
1 ∫ f (ζ)
an= , n ∈ N0.
2πi Γ (ζ − z 0 ) n+ 1
T hus,
Σ∞ n
a n (z − z 0 ) on D,
n= −∞

where ∫
1 f (ζ)
an= dζ ∀n ∈Z.
2πi Γ (ζ − z 0 ) n+ 1
This completes the proof.

2. Isolated Singularities
Deftnition 6.2.1 Let f be analytic in an open set Ω. A point z0 ƒ∈Ω
is said to be a singularity of f if z 0 is a limit point of Ω and f cannot
be extended to an open set Ω̃ which contains Ω and z 0 . ♦
• z 0 is a singularity of an analytic function f : Ω → C if and only if
z 0 is a limit point of Ω and for every r > 0, f cannot be extended
analytically to Ω ∪ B(z 0 , r), i.e., there is not analytic function g :
Ω ∪B(z 0 , r)τ C such that g(z) = f (z) for all z ∈ Ω.
Deftnition 6.2.2 Let Ω be an open set in which a holomorphic
function f is defined and z 0 ∈ C\Ω. Then z 0 is said to be an isolated
singularity of f if a deleted neigbourhood of z 0 is contained in Ω.
Let z 0 be an isolated singularity of a holomorphic function f , and
let
Σ∞ n
a n (z − z 0 )
n=−∞
be its Laurent series expansion in a deleted neigbourhood D 0 of f .
Then z 0 is said to be

• a removable singularity of f if a−n = 0 for all n ∈N;

• a pole order m of f if a−m ƒ = 0 and an = 0 for all n < −m;

• an essential singularity of f if a−n ƒ = 0 for infinitely many
n ∈N.
Isolated Singularities 93

If z 0 is a pole of order 1, then z 0 is called a simple pole. ♦

EXAMPLE 6.2.1 (i) For the function f (z) := 1/z 3 , z 0 = 0 is a
pole of order 3.
(ii) For the function f (z) := sin z / z , z 0 = 0 is a removable sin-
gularity.
2
(iii) For the function f (z) := e1/z , z 0= 0 is an essential singu-
larity. Q
Theorem 6.2.1 Let z 0 be an isolated singularity of a holomorphic
function f . Then the following are equivalent.
(i) z 0 is a removable singularity of f .
( i i ) f can be extended holomorphically to a neigbourhood of z 0 .
( i i i ) f is bounded in a deleted neigbourhood of z 0 .
(iv) limz→z0 (z − z 0 )f (z) = 0.
Σ ∞
Proof. Let a (z − z 0) nbe the Laurent series expansion f
n= 0 n
in a deleted neigbourhood D 0 of z 0 .
(i) ⇐⇒ (ii): Suppose z is0 a removable singularity of f . Defining
.
f (z), z ƒ = z 0 ,
f ˜ (z 0 ) = , we see tha t f ˜ is a holomorphic extension
a0, z ƒ = z0 ,
of f to a neigbourhood of z 0 .
Conversely, suppose f ˜ is a holomorphic extension of f to neig-
Σ
bourhood D of z 0 . Let ∞n=0 a n (z − z 0 ) n be the Taylor series ex-
Σ ∞
pansion of f ˜ in D. Then, a (z − z 0 ) n is the Laurent series
n= 0 n
expansion of f in the deleted neigbourhood D \ {0}.
(i) ⇐⇒ (iii): Suppose z 0 is a removable singularity of f . Since
limz→z0 f (z) = a 0 , there exists δ > 0 such that
0 < |z − z0| < δ =⇒ |f (z) − a0| ≤ 1.

Hence,
0 < |z − z0| < δ =⇒ |f (z)| ≤ |a0| + 1.
Conversely, suppose that f is bounded in a deleted neigbourhood
D 0 of z 0 , say |f (z)| ≤ M for all z ∈ D 0 . Let r > 0 be such that Γ r
:= { z ∈ C : |z − z0| = r } ⊆ D 0 . Then, for each n ∈ N,
∫
1
|a −n | = . (z − z 0 ) n−1 f (z)dz . ≤ M r n.
2πi Γ r
94 Laurent Series and IsolatedSingularities

Letting r → 0, we obtain a−n = 0.

(i) ⇐⇒ (iv): Suppose z 0 is a removable singularity of f . Since
limz→z0 f (z) = a 0 , we have limz→z0 (z − z 0 )f (z) = 0.
Conversely, suppose limz→z0 (z − z 0 )f (z) = 0. Let ε > 0 and let
δ > 0 be such that

0 < |z − z0| < δ =⇒ |(z − z 0 )f (z)| < ε.

Let 0 < r < min{δ, 1}. Then, for each n ∈ N,

∫
1
|a −n | = . (z − z 0 ) n−1 f (z)dz .
2πi Γ r
∫
1
=
. 2πi (z − z 0 ) n−2 (z − z0 ) f (z)dz .
Γr
n−1
≤ r ε ≤ ε.

Hence, a−n = 0 for all n ∈ N, and hence z 0 is a removable singularity

of f .

Theorem 6.2.2 Let z 0 be an isolated singularity of a holomorphic

function f . Then the following are equivalent.

(i) z 0 is a pole of f .

(ii) There exists m ∈ N and a holomorphic function ϕ in a neig-

bourhood D of z 0 such that f (z) = (z − z 0 ) − mϕ(z) for all z ∈
D \ { 0 } with ϕ(z 0 ) ƒ = 0.

(iii) |f (z)| → ∞ as z → z 0 .

(iv) There exists m ∈ N such that z 0 is a removable singularity of

ϕ(z) := (z − z0 )m f (z) with lim z→ z0 ϕ(z) ƒ= 0.

(v) The function 1 / f defined in a deleted neigbourhood of z 0 can be

extended holomorphically to a neigbourhood of z 0 and z0 is a
zero of order m of the extended function.
Σ ∞
Proof. Let a (z − z 0) nbe the Laurent series expansion f
n= 0 n
in a deleted neigbourhood D 0 of z 0 .
(i) ⇐⇒ (ii): This is obvious from the definition of the pole.
Isolated Singularities 95

(i) ⇐⇒ (iii): Suppose z 0 is a pole of order m of f . By (ii), There

exists m ∈ N and a holomorphic function ϕ in a neigbourhood D of
z 0 such that

f (z) = (z − z 0 ) −mϕ(z) ∀z ∈ D 0 := D \ { 0 }

with ϕ(z 0 ) ƒ = 0. Since ϕ(z) → ϕ(z 0 ) ƒ = 0, we obtain

|f (z)| = |(z − z 0 ) −mϕ(z)| → ∞ as z → z0.

Conversely, suppose |f (z)| → ∞ as z → z 0 . Then f is nonzero in

a deleted neigbourhood D 0 of z 0 , and hence, the function defined by
g(z) = 1 / f (z) for z ∈D 0 , satisfies

|g(z)| → 0 as z → z0.

Define g(z 0 ) = 0. Then g is holomorphic in D := D 0 ∪ {0}. Hence,

there exists m ∈ N and a holomorphic function ϕ in D such that
m
g(z) = (z − z 0 ) ϕ(z) ∀z ∈D

ϕ(z 0 ) ƒ = 0. Thus, for z ∈D 0 ,

1 1
f (z) = = (z − z 0 ) −m ψ(z ), ψ(z) :=
g(z ) ϕ (z )

and ψ(z 0 ) ƒ = 0. Hence, z 0 is a pole of order m of f .

(i) ⇐⇒ (iv): Suppose z 0 is a pole of f . By (ii), there exists
m ∈ N and a holomorphic function ϕ in a neigbourhood D of z 0 such
that
f (z) = (z − z 0 ) −mϕ(z) ∀z ∈ D 0 := D \ { 0 }
with ϕ(z 0 ) ƒ = 0. Then, ϕ(z) = (z − z 0 ) m f (z) in D 0 . Clearly,
limz→z0 ϕ(z) = ϕ(z 0 ). Hence, z 0 is a removable singularity of ϕ.
Conversely, suppose there exists m ∈ N such that z 0 is a remov-
able singularity of ϕ(z) := (z − z 0) mf (z) with lim z→ z0 ϕ(z) ƒ= 0. Let
ϕ˜ be the holomorphic extension of ϕ to a neigbourhood D of z 0 .
Then we have

f (z) = (z − z 0 ) −mϕ(z) ∀z ∈ D 0 := D \ { 0 }

and ϕ˜(z) ƒ = 0. Hence z 0 is a pole of order m of f .

96 Laurent Series and IsolatedSingularities

(i) ⇐⇒ (v): Suppose z 0 is a pole of f . By (ii), there exists m ∈ N

and a holomorphic function ϕ in a neigbourhood D of z 0 such that

f (z) = (z − z 0 ) −mϕ(z) ∀z ∈D 0 := D \ { 0 }

with ϕ(z 0 ) ƒ = 0. Then,

1
= (z − z0 )mψ(z) ∀z ∈ D 0:= D \{0},
f (z)

where ψ, defined by ψ(z) := 1/ϕ(z), is analytic on D. Thus, the

function 1 / f can be extended holomorphically to D by defining 0 at
z 0 , and z 0 is a zero of order m of the extended function.
Conversely, if g := 1 / f can be extended holomorphically to a
neigbourhood D of z 0 and z 0 is a zero of order m of the extended
function g˜,then there exists a an analytic function ψ in D such that

g˜(z) = (z − z 0 ) mψ(z) ∀z ∈ D.

Thus, we have

f (z) = (z − z 0 ) −mϕ(z) ∀z ∈ D 0 := D \ { 0 }

where ϕ(z) = 1/ψ(z) on D and ϕ(z 0 ) ƒ = 0 so that z 0 is a pole of f

order m.

The following theorem shows that if z 0 is an essential singularity

of a holomorphic function f , then in a neigbourhood of z 0 , there are
values of f which are arbitrarily close to any complex number.

Theorem 6.2.3 (Casorati-Weierstrass theroem) Suppose z 0 is

an essential singularity of a holomorphic function f . Then for every
w ∈ C, there exists a sequence (z n ) in the domain of analyticity of f
such that f (z n ) → w as n → ∞.

Proof. Suppose for a moment that the conclusion in the theorem

does not hold. Then there exists w ∈C such that for any sequence
(z n ) with z n → z 0 , f (z n ) ƒ→w. Then the function g(z) := f (z)−w
1
is
analytic in a deleted neigbourhood of z 0 . Further, |g| is bounded in a
deleted neigbourhood of z 0 . (If |g| is not bounded in any deleted
neigbourhood of z 0 , then there exists a sequence (z n ) such that zn →
z 0 and |g(zn)| → ∞.) Hence, by Theorem 6.2.1, z 0 is a removable
Problems 97

singularity of g. Suppose g(z) → α 0 as z → z 0 . We have the following

two cases:
Case (i): α 0 ƒ = 0: In this case, f (z) = w + 1/g(z) → w + 1/α 0 as
z → z 0 , so that by Theorem 6.2.1, z 0 is a removable singularity of f
as well.
Case (ii): α 0 = 0: In this case, we have |f (z)| = |w+1/g(z)| → ∞
as z → z 0 , and hence by Theorem 6.2.2, z 0 is a pole of f .
Since z 0 is an essential singularity, cases (i) and (ii) can not occur.
Thus, our assumption that the conclusion in the theorem does not
hold is not true.

6.3 Problems
1. For 0 < a < 1, find the annulus of convergence of the series
Σ ∞ n2 n
n= −∞ a z .

2. Locate and classify the isolated singularities of the following

functions:

z5 1
(i) , (ii) , (iii) sin(1/z).
1 = z + z2 = z3 + z4 sin2 z
Also, check whether z 0 = ∞ is an isolated singularity (i.e.,
w0 = 0 is an isolated singularity of f (1/z)) in each case.
3. If f and g are holomorphic functions having z 0 a pole of the
same order for both, then prove that

f (z) f j (z)
lim = lim j .
z→z0 g(z) z→z0 g (z)
7
Residues and Real Integrals

7.1 Residue theorem

Deftnition 7.1.1 Suppose f is holomorphic in a deleted neigbour-
hood D 0 of z 0 ∈ C and Γ r := { z ∈ C : |z − z0| = r } ⊆ D 0 . Then
residueof f at z 0 is defined by
∫
1
Res(f, z0 ) = f (z)dz.
2πi Γ r
♦
Recall that if f is holomorphic in a deleted neigbourhood D 0 of
z 0 ∈ C, then f has Laurent series expansion
Σ∞ n
f (z) = a n (z − z 0 ) , z ∈D 0 ,
n= −∞

and we know that

∫
1 f (z)
an= dz, n ∈Z.
2πi Γr (z −z 0 ) n+ 1
T hus,
Res(f, z 0 ) = a−1 .
The following theorem, known as residue theorem follows from
Cauchy’s theorem.
Theorem 7.1.1 (Residue theorem) Suppose Γ is a simple closed
contour and z 1 , . . . , zk are points in Γ Γ which are the only singular
points of f in Γ ∪ΩΓ. Then
∫ Σk
f (z)dz = 2πi Res(f, z j ).
Γ j=1

98
Calculation of Residues 99

7.2 Calculation of Residues

Suppose z 0 is a pole of order m of a holomorphic function f . Then we
know that there exists a holomorphic function ϕ in a neigbourhood
D of z 0 such that

f (z) = (z − z 0 ) −mϕ(z) ∀z ∈D 0 := D \ { z 0 } .

Let
Σ∞
ϕ(z) = α n(z − z 0) n, z ∈D.
n= 0

Then we have

ϕ (n) (z 0 )
αn = , n ∈ N0 := N ∪{ 0 }
n!
so that

−m
Σ∞ Σ∞
f (z) = (z − z 0) ϕ(z) = α n(z − z 0 ) n−m
= an(z − z 0) n.
n= 0 n= −m

Hence,
ϕ (m−1 ) (z 0 ) .
Res(f, z 0 ) = a−1 = α m−1 =
( m −1)!
Thus, we have proved the following theorem.

Theorem 7.2.1 Suppose z 0 is a pole of order m of a holomorphic

function f . Then the function

z ›→ ϕ(z) := (z − z 0 ) m f (z)

defined in a deleted neigbourhood of z 0 has a holomorphic extension

to a neigbourhood D of z 0 , again denoted by ϕ, and

ϕ (m−1 ) (z 0 ) .
Res(f, z 0 ) =
( m −1)!

In particular, if z 0 is a simple pole of f , then

Res(f, z 0 ) = lim (z − z 0 )f (z).

z→z0
100 Residues and Real Integrals

Corollary 7.2.2 Suppose g and h are holomorphic in a neigbourhood

D of z 0 and z0 is a zero of h of order m. If h(z) = (z − z 0 ) mh 0 (z)
with h 0 (z 0 ) ƒ = 0, then

Res( g, z 0 ) = ϕ
(m−1 ) (z )
0 ,
h ( m −1)!

where ϕ(z) = g(z)/h 0 (z). In particular, if m = 1, then

g g(z 0 ) g(z 0 )
Res( , z )0 = = j .
h h 0 (z 0) h (z0)
∫
1
EXAMPLE 7.2.1 Let us find f (z)dz, where f (z) =
Γ z(z −1)
and Γ is the positively oriented circle with centre 0 and radius 2.
By residue theorem,
∫
dz
= 2πi[Res(f, z 1 ) + Res(f, z2 )], z 1 = 0, z 2 = 1.
Γ z(z −1)

Since z 1 = 0 and z 2 = 1 are simple poles of the function

Res(f, z 1 ) = lim z f (z) = −1,

z→ 0

Res(f, z 2 ) = lim(z − 1)f (z) = 1.

z→ 1
∫
dz
Hence, = 0. Q
Γ z(z − 1)
∫
1
EXAMPLE 7.2.2 Let us find f (z)dz, where f (z) =
Γ z(z −1)2
and Γ is the positively oriented circle with centre 0 and radius 2. By
residue theorem,
∫
dz
= 2πi[Res(f, z 1 ) + Res(f, z2 )], z 1 = 0, z 2 = 1.
Γ z(z −1)
2

Since z 1 = 0 is a simple pole and z 2 = 1 is a pole of order 2,

Res(f, z 1 ) = lim z f (z) = 1, Res(f, z 2 ) = ϕ j (1)

z→0
∫ dz
where ϕ(z) = 1 so that ϕ j (1) = −1. Thus, = 0. Q
z
Γ z(z − 1)2
Evaluation of ImproperIntegrals 101

Exercise 7.2.1 1. Find Res(f, z 0 ), where

(a) f (z) = ze1/z , z 0 = 0.

(b) f (z) = z + 2 , (i) z 0 = 0, (ii) z 0 = −1.
z(z + 1)
∫
2. Evaluate f (z)dz, where
Γ

(a) 3z + 1 and Γ = { z : |z| = 2}.

z(z − 1)3
z+1
(b) and Γ = { z : |z| = 1}.
− 3z 2 − 2z
2z 3
(c) z + 1/z and Γ = { z : |z| = 1}.
z(2z − 1/(2z)
log(z + 2)
(d) and Γ = { z : |z| = 1}.
2z + 1
cosh(1/ z )
(e) and Γ = { z : |z| = 1}.
z
a

7.3 Evaluation of Improper Integrals

In this section we shall evaluate integrals of the form
∫ ∞ ∫ ∞
f (x)dx and f (x)dx,
0 −∞

where f is a continuous function.

∫
∞
dx
EXAMPLE 7.3.1 Let us evaluate . For this consider
−∞ 1 + x2
the function f (z) = dx for z ƒ = 0. Note tha t z = i is the only
1 + z2
singularity of f in the upper half plane and it is a simple pole. Con-
sider the positively oriented curve Γ R consisting of the semicircle
with centre 0 and radius R, i.e., S R := { z : |z| = R, Im(z) > 0} and
the line segment L R := [−R, R]. Then, by Cauchy’s theorem,
∫ ∫
f (z)dz = f (z)dz,
ΓR Cr
102 Residues and Real Integrals

where C r = { z : |z − i| = r } with 0 < r < R. But,

∫
f (z)dz = 2πi Res(f, i) = 2πi lim(z − i ) f (z) = π.
Cr z→ i

T hus, ∫
f (z)dz = π.
ΓR
Also, we have
∫ ∫ ∫ ∫ ∫ R
f (z)dz = f (z)dz + f (z)dz = f (z)dz + f (x)dx.
ΓR SR LR SR −R

But, for z ∈ S R ,
1
|f (z)| ≤ .
R2 − 1
Hence, ∫
A(SR ) πR
. f (z)dz. ≤ = 2 .
SR R2 − 1 R −1
Hence,
∫ ∫ ∫
lim f (z )dz = lim f (z)dz + lim f (z)dz
R→ ∞ ΓR R→ ∞ SR R→ ∞ LR
∫ ∞
= 0+ f (x)dx.
−∞

T hus, ∫ ∫
∞
f (x)dx = lim f (z)dz = π.
−∞ R→ ∞ ΓR

∫ Q
cos x ∞
EXAMPLE 7.3.2 Let us evaluate 2
dx. Since
−∞ 1 + x
∫ . ∫ ∞ Σ
∞ cos x e ix
2 dx = Re 2 dx
−∞ 1 + x −∞ 1 + x

we consider the function

eiz
f (z ) = , z ƒ∈{i, −i}.
1 + z2
Following the arguments as in the previous example, one arrive at
∫ ∞
e ix π
1 + x 2 dx = e .
−∞
Evaluation of ImproperIntegrals 103

But, . ∫ Σ ∫
∞ e ix ∞ sin x
Im dx = dx = 0.
−∞ 1 + x2 −∞ 1 + x2
Hence, ∫
∞ cos x π
2
dx = .
−∞ 1+ x e
Q
∫ ∞
sinx
EXAMPLE 7.3.3 Let us evaluate dx. Since
−∞ x
∫ ∞ . ∫ ∞ ix Σ
sinx e
dx = Re dx
−∞ x −∞ x

we consider the function

eiz
, z ƒ = 0.
f (z ) =
z
For r > 0, let S r := { z : |z| = r } with positive orientation. Then, tak-
ing 0 < ε < R and Γ as the curve consisting of S R, [−R, −ε], S˜ ε, [ε, R],
using Cauchy’s theorem,
∫
0= f (z)dz
Γ
∫ ∫ −ε ∫ ∫ R
= f (z)dz + f (x)dx + f (z)dz + f (x)dx
SR −R S̃ε ε
∫ ∫ −ε ∫ ∫ R
= f (z)dz + f (x)dx − f (z)dz + f (x)dx
SR −R Sε ε

But,
∫ −ε ∫ R
cosx cosx
dx + dx = 0
−R x ε x
and ∫ ∫ ∫
−ε R R
sinx sinx sinx
dx + dx = 2 dx.
−R x ε x ε x
Hence,
∫ R ∫ −ε ∫ R sinx
f (x)dx + f (x)dx = 2i dx.
ε −R ε x
T hus,
∫ R ∫ ∫
sinx
2i dx = f (z)dz − f (z)dz
ε x Sε SR
104 Residues and Real Integrals

Now, we note that with the parametrization γ(t) = Re it , 0 ≤ t ≤ π

of S R ,
∫ ∫ π iR(cos t + i sin t) ∫ π
e iR t
f (z)dz = iR e dt = i eiR(cos t + i sin t) dt.
SR 0 Re iRt 0

Hence,
∫ ∫ π ∫ π/2
.
. f (z )dz .. ≤ e −R sin t
dt = 2 e−R sin t dt.
SR 0 0

sint sin t sin π / 2

Since is decreasing in [0, π/2], we have ≥ so that
t t π/ 2
sin t ≥2t/π.Thus,
∫ ∫ π/2
. f (z)dz.. ≤ 2 e−2Rt/π dt =
π (1−e −2R ) → 0 as R → ∞.
. 2R
SR 0

Next, we observe that

eiz 1
= + ϕ(z)
z z
where ϕ is an entire function. Hence, there exists M > 0 such that
|ϕ(z)| ≤ M for all z with |z| ≤ 1. Thus,
∫ ∫ dz ∫
f (z)dz = + ϕ (z )dz,
Sε Sε z Sε

where ∫
ϕ(z)dz ≤ Mπ ε , 0 < ε ≤1.
. .
Sε
Hence,
∫ R ∫ ∫ ∫
sinx dz + ϕ(z)dz − f (z)dz
2i dx =
ε x Sε z Sε SR
∫ ∫
= πi + ϕ(z)dz − f (z)dz,
Sε SR

where
∫ ∫
ϕ (z )dz → 0 as ε → 0 and f (z)dz → 0 as R → ∞.
Sε SR

T hus,
∫ R sinx π
dx = .
ε x 2
Q
Problems 105

4. Problems
1. Find the residues of the following functions:
z3 z3
(i) (ii)
z −1 (z −1)2
2. If f and g are holomorphic in a neigbourhood of z 0 , and z 0 is a
simple a pole of g, then prove that R es ( f / g , z 0 ) = f (z 0 )/g j (z 0 ).

3. Determine the residues of each of the following functions a t

each of their singularities:
z3 z5 cos z
(i) , (ii) , (iii) .
1 − z4 (z 2 − 1)2 1 + z + z2

4. If f is holomorphic in a neigbourhood of z 0 , and z 0 is a zero of

f order m, then prove that Res(f j / f , z 0 ) = m.

5. Evaluate the following using complex integrals:

∫ ∞ ix ∫
e dx, ∞ dx
(i) (ii) ,
0 x 0 1 + x2
∫ ∞ ∫
sin2x ∞ cos ax
(iii) dx, (iii) dx, a ≥ 0, b > 0.
0 x 0
x 2 + b2
Index

C, 1 imaginary part, 3
intersect, 17, 52
absolute value, 3 isolated singularity, 75
analytic, 13 isolated zero, 65
angle, 17
annulus, 71 Laplace equation, 20
argument, 5 Laplacian, 20
Laurent series, 70, 72
branch, 34 linear fractional transforma-
tion, 24
Cauchy’s theorem, 53, 55
Cauchy-Riemann equations, Logarithm, 33
10
M¨obius transformation, 24
closed curve, 52
modulus, 3
complex conjugate, 3
conformal map, 18 open ball, 3
converge, 37 oriented range, 48
absolutely, 37
point-wise, 37 parametrization, 48
uniformly, 38 piecewise smooth, 49
CR-equations, 10 polar representation, 5
curve, 15 pole
of order m, 75
differentiable, 8, 16 positively oriented, 54
disc of convergence, 41 preserve angle, 18
domain, 54
radius of convergence, 41
extended complex plane, 23 real part, 3
fixed point, 25 region of convergence, 41
fractional linear transforma- regular, 17
tion, 24 reparameterization, 49
residue, 80
harmonic conjugate, 20 Residue theorem, 80
holomorphic, 13 reversed curve, 48

106
Index 107

rook-path, 56
roots, 5

self intersecting, 17
simple closed curve, 52
simple pole, 75
simply connected, 54
singularity
essential, 75
removable, 75
Steriographic projection, 6

tangent vector, 17
triangle inequality, 3

zero of f , 65