i

COMPLEX ANALYSIS
Dr. ZIAD KHALIL
May 31, 2020

ii
 CONTENTS

RECALL iv

1 COMPLEX INTEGRATION 2

2 ZERO’S AND SINGULARITIES OF HOLOMORPHIC FUNCTIONS 12

3 CAUCHY THEORY AND ITS APPLICATIONS 17

iii
 RECALL

Lecture 1.
C will denote the field of complex numbers and Ω will denote an open subset of C.
0.1 Definition. Let a ∈ C and r > 0.
1. D(a, r) = {z ∈ C; ∣z − a∣ < r} is the open circular disc with center a and radius r.

2. D[a, r] = {z ∈ C; ∣z − a∣ ≤ r} is the closed circular disc with center a and radius r.

3. D′ (a, r) = {z ∈ C; 0 < ∣z − a∣ < r} is the punctured disc with center a and radius r.
Note that D′ (a, r) = D(a, r) ∖ {a}

Holomorphic Functions
Let f ∶ Ω ↦ C be a function.
f (z) − f (z0 )
1. If z0 ∈ Ω and lim exists, we donte this limit by f ′ (z0 ) and we call it the derivative
z→z0 z − z0
of f at z0 . We also say that f is C− differentiable at z0 .

2. We say that f is holomorphic on Ω if f is C− differentiable at each point of Ω.

3. The class of all holomorphic functions in Ω will be denoted by H(Ω). i.e f ∈ H(Ω) iff f is
C− differentiable at each point of Ω.

4. A function f ∈ H(C) is called an entire function.

0.2 Property. Let z0 ∈ Ω and f ∶ Ω ↦ C be a function .
1. If f is differentiable at z0 , then there exists a function ε ∶ Ω ↦ C continuous at z0 with
ε(z0 ) = lim ε(z) = 0 such that f (z) = f (z0 ) + (z − z0 )f ′ (z0 ) + (z − z0 )ε(z).
z→z0

2. If there exist a function ε ∶ Ω ↦ C continuous at z0 with ε(z0 ) = 0 and λ ∈ C such that

f (z) = f (z0 ) + (z − z0 )λ + (z − z0 )ε(z), then f is differentiable at z0 and f ′ (z0 ) = λ.

Power Series
. For a power series centered at a ∈ C ,
∞
∑ cn (z − a)n (1)
n=0

there exists R = 1
1 ∈ [0, ∞] called the radius of convergence of the power series such that
lim sup ∣cn ∣ n

iv
 1. The p.s (1) is absolutely convergent on the disc of convergence D(a, R) = {z ∈ C; ∣z − a∣ < R}.

2. The p.s (1) is divergent on (D[a, R])c = {z ∈ C; ∣z − a∣ > R}.

3. The p.s (1) is uniformly convergent on the disc D[a, r] for each 0 < r < R.

4. Convergence and divergence both may take place on ∂D(a, R) = {z ∈ C; ∣z − a∣ = R}.

0.3 Theorem. (Weierstrass Test)

Let fn ∶ Ω ↦ C be complex valued functions on any subset Ω of C and ∣fn (z)∣ ≤ an for all z ∈ Ω, , n ∈
∞ ∞
N. If ∑ an is convergent, then the series ∑ fn (z) is uniformly convergent on Ω.
n=0 n=0

0.4 Property. If R ∈]0, ∞], then the p.s (1) defines a holomorphic function on the disc of
∞
convergence D(a, R), i.e ∑ cn (z − a)n = f (z) ∈ H(D(a, R)). Moreover,
n=0

f (n) (a)
1. cn = n! , ∀n ∈ N∗ .

2. f is infinitely differentiable on D(a, R) and

∞ ∞
f (k+n) (a)
f (k) (z) = ∑ (n−k)! cn (z
n!
− a)(n−k) = ∑ (n)! (z − a)(n)
n=k n=0

for all z ∈ D(a, R), k ∈ N∗ .

Analytic Functions
Let Ω ⊂ C be an open set , f ∶ Ω ↦ C be a function.

1. f is said to be analytic in Ω if for each a ∈ Ω there exists D(a, r) ⊂ Ω and a sequence (cn )n ⊂ C
∞
such that f (z) = ∑ cn (z − a)n for all z ∈ D(a, r).
n=0

2. f is said to be representable in Ω by a power series if for each D(a, r) ⊂ Ω, there exists a

∞
sequence (cn )n ⊂ C such that f (z) = ∑ cn (z − a)n for all z ∈ D(a, r).
n=0

3. f is analytic in in Ω if and only if f is representable in Ω by a power series.

0.5 Property. If f is analytic in Ω, then f ∈ H(Ω) and f (k) ∈ H(Ω) for all k ∈ N.

In the next chapter we will show that if f ∈ H(Ω), then f is analytic in Ω and so f (k) ∈ H(Ω).

v
vi
1
CHAPTER 1
COMPLEX INTEGRATION

C will denote the field of complex numbers and Ω will denote an open subset of C.
1.1 Definition. Let a, b ∈ R.

1. A curve in C is a continuous map γ ∶ [a, b] ↦ C.

2. A path γ in C is a piece-wise C 1 − curve i.e, there exist finitely many points sj such that
a = s0 < s1 < ⋯ < sn−1 < sn = b and γ/[sj−1 , sj ] is of class C 1 .
However at the points s1 , ⋯sn−1 the left and right hand derivatives of γ may differ.

ˆ [a, b] is called the parameter interval of γ.

ˆ γ = γ([a, b]) = {γ(t) ∈ C; t ∈ [a, b]} will denote the range of γ.
∗

ˆ γ is a compact subset of C. (Because the continuous image of a compact set is a compact

∗

set).

ˆ The points γ(a) and γ(b) are called respectively the initial and final points of γ or γ . ∗

ˆ A curve (or path) with γ(a) = γ(b) is called a closed curve (or path).
1.2 Example. Let γ ∶ [0, 1] ↦ C defined by γ(t) = cos3 (2πt) + i sin3 (2πt) = x(t) + iy(t).

ˆ γ is a path.(Because x and y are of class C 1

on [0, 1]).

ˆ γ is a closed path.( Because γ(0) = γ(1)).

1.3 Definition. (Integration Over Paths)
Let γ ∶ [a, b] ↦ γ ∗ ⊂ C be a path and f ∶ γ ∗ ⊂ C ↦ C be a continuous function on γ ∗ . Then the path
integral of f along γ is defined by
b
∫ f (z) dz = ∫ f (γ(t))γ ′ (t) dt.
γ a

1.4 Definition. (Reparametrization)

Let γ ∶ [a, b] ↦ γ ∗ ⊂ C be a path in C. If φ ∶ [a1 , b1 ] ⊂ R ↦ [a, b] is C 1 ,(or piece wise C 1 ), onto
function with φ′ (t) > 0. Then γ1 = γ ○ φ is a path in C with parameter interval [a1 , b1 ] and range
γ ∗ . The path γ1 is called a reparametrization of γ and φ is called the reparametrization map.

Note that since φ′ (t) > 0, then φ(a1 ) = a and φ(b1 ) = b . Hence the reparametrization map φ
preserves orientation.

1.5 Property. Reparametrization doesn’t change integral.

2
Proof. ∫γ1 f (z) dz = ∫a11 f (γ1 (t))γ1′ (t) dt = ∫a11 f (γ(φ(t))φ′ (t)γ ′ (φ(t)) dt = ∫a f (γ(s))γ ′ (s) ds =
b b b

b ′
∫a f (γ(t))γ (t) dt.

We shall regard γ and γ1 as equivalent. Thus we can replace a path by an equivalent one with
new parameter interval.
For instance, if the end point of a path γ1 coincides with the initial point of another path γ2 , then
we may locate that parameter intervals so that γ1 and γ2 join to form a path γ with the following
property:
∫ f (z) dz = ∫ f (z) dz + ∫ f (z) dz.
γ γ1 γ2

γ will be denoted by γ = γ1 ⋃ γ2 or γ = γ1 + γ2 .

1.6 Definition. (Reverse Path)

Let γ ∶ [a, b] ↦ C be a path in C. The opposite or reverse path γ − ∶ [a, b] ↦ C of γ is given by
γ − (t) = γ(a + b − t).

Note that γ − is γ in the opposite direction. Hence γ − ∗ = γ ∗ .

1.7 Property. Let γ be a path and f, g ∈ C(γ ∗ ). Then

1. ∫γ (αf (z) + βg(z)) dz = α ∫γ f (z) dz + β ∫γ g(z) dz, ∀α, β ∈ C.

2. ∫γ − f (z) dz = − ∫γ f (z) dz.

Proof. 1. Evident. Use the definition of path integral.

2. ∫γ − f (z) dz = ∫a f (γ − (t))γ − ′ (t) dt = − ∫a f (γ(a + b − t))γ ′ (a + b − t) dt = − ∫a f (γ(s))γ ′ (s) ds =

b b b

− ∫γ f (z) dz.

Lecture 2.
b
1.8 Definition. L(γ) ∶= ∫ ∣γ ′ (t)∣ dt represents the length of the path γ.
a

1.9 Property. (Basic Estimate)

Suppose that γ ∶ [a, b] ↦ C is a path and f is a complex function continuous on γ ∗ . Then,

∣ ∫ f (z) dz∣ ≤ M L(γ), where M = max {∣f (z)∣}.

γ ∗ z∈γ

Proof. Since ∣f ∣ is continuous on the compact set γ ∗ , then ∣f ∣ attains its maximum M on γ ∗ .
b b b
∣ ∫ f (z) dz∣ = ∣ ∫ f (γ(t))γ ′ (t) dt∣ ≤ ∫ ∣f (γ(t))∣ ∣γ ′ (t)∣ dt ≤ M ∫ ∣γ ′ (t)∣ dt ≤ M.L(γ)
γ a a a

Special Cases:

1. The positively oriented circle ⟲ with center a ∈ C and radius r > 0.

γ ∶ [0, 2π] ↦ C or γ1 ∶ [0, 1] ↦ C

t ↦ a+reit t ↦ a+re2iπt

● If f ∈ C(γ ∗ ), then ∫γ f (z) dz = ∫0 f (γ(t))γ ′ (t) dt = ir ∫0 f (a + reit )eit dt.

2π 2π

● L(γ) = ∫0 ∣γ ′ (t)∣ dt = ∫0 ∣ireit ∣ dt = ∫0 r dt = 2πr. ( As expected )

2π 2π 2π

3
 2. The oriented segment [a, b] with initial point a ∈ C and final point b ∈ C.

γ ∶ [0, 1] ↦ [a, b] ⊂ C
t ↦ a+(b−a)t

● ∫[b,a] f (z) dz = − ∫[a,b] f (z) dz, ∀f ∈ C([a, b]).

● L(γ) = ∫0 ∣γ ′ (t)∣ dt = ∫0 ∣b − a∣ dt = ∣b − a∣. ( As expected )
1 1

● Reparametrization of γ:

φ γ
[α,β] ↦ [0,1] ↦ [a,b]⊂C
t ↦ t−α ↦ a+(b−a)φ(t)
β−α

φ is a reparametrization map because φ ∈ C 1 [α, β], φ′ (t) = β−α

1
> 0, φ(α) = 0 and φ(β) = 1.
Hence the path γ1 = γ ○ φ is a reparametrization of the path γ, the segment [a, b] , with
parameter interval [α, β].
Note that, a path can be parametrized by any given interval by the following parametrization
map:
φ
[α,β] ↦ [c,d]
t ↦ rt+s

where r = c−d
α−β and s = αd−βc
α−β can be obtained by solving the linear system φ(α) = c and
φ(β) = d.

3. The positively oriented triangle ∆ = ∆(a, b, c) of vertices a, b and c in positive orientation ⟲.

We may regard the boundary of the triangle ∂∆ as a path joining [a, b] to [b, c] to [c, a].
●∫ f (z) dz = ∫ f (z) dz + ∫ f (z) dz + ∫ f (z) dz.
∂∆(a,b,c) [a,b] [b,c] [c,a]
● If {a, b, c} permute cyclically, then ∫ f (z) dz does not affected. i.e.
∂∆(a,b,c)

∫ f (z) dz = ∫ f (z) dz = ∫ f (z) dz.

∂∆(a,b,c) ∂∆(b,c,a) ∂∆(c,a,b)

● The integral over the reverse path,

∫ f (z) dz = − ∫ f (z) dz
∂∆(a,c,b) ∂∆(a,b,c)

with all possible cyclic permutations on both sides.

1.10 Definition. Let γ be a closed path in C and z ∈ C ∖ γ ∗ = Ω, then the index of γ with respect
to z is defined by
1 dξ
Indγ (z) = ∫ , ∀z ∈ Ω = C ∖ γ ∗ .
2πi γ ξ − z
Note that, since ξ ∈ γ ∗ and z ∈ C ∖ γ ∗ , then ξ − z ≠ 0. Hence the map ξ ↦ ξ−z
1
is continuous on γ ∗
and so the above integral is well defined . Thus Indγ ∶ C ∖ γ ∗ ↦ C defines a function.
Recall: Let (Y, d) be a metric space and y ∈ Y .
● C(y), the component of y in Y , is the union of all connected subsets of Y containing y.
● It is evident C(y) is connected. (because y ∈ ⋂ of all connected subsets of Y containing y.)
● C(y) is a maximal connected set in Y , i.e. no connected subset of Y contains C(y).
● C(y) is closed in Y . ( Since C(y) is connected , then C(y) ⊂ C(y).
● The set of all components of y form a partition of Y .
1.11 Remark. Let γ be a closed path in C . Then Ω = C ∖ γ ∗ has only one unbounded component.
Proof. γ ∗ is compact Ô⇒ γ ∗ is bounded Ô⇒ γ ∗ ⊂ D(a, r) Ô⇒ D(a, r)c is a subset of some
component of Ω( since it is connected) Ô⇒ Ω has precisely one unbounded component.

4
 O
1.12 Property. 1 The index is an integer valued function on Ω = C ∖ γ ∗ ,i.e. Indγ ∶ Ω ↦ Z 2 O
O
which is constant in each component of Ω 3 and zero in the unbounded component of Ω.

O
Proof. 1 Suppose γ is parametrized by [α, β], then Indγ (z) =
1
∫
β γ ′ (s)

2πi α γ(s) − z
ds, ∀z ∈ Ω.
● Recall that; ew1 = ew2 ⇐⇒ w1 = w2 + 2ikπ, k ∈ Z.
● Note that; 2πi
w
∈ Z ⇐⇒ w = 0 + 2ikπ, k ∈ Z ⇐⇒ ew = eo = 1.
β γ ′ (s) t γ ′ (s)
● Indγ (z) ∈ Z ⇐⇒ e∫α γ(s)−z = 1 ⇐⇒ ϕ(β) = 1, where ϕ(t) = e∫α γ(s)−z = eg(t) ∀t ∈ [α, β].
ds ds

Since γ = x + iy is a path in C, then γ ′ is continuous on [α, β] ∖ S where S is a finite subset of the

interval ]α, β[ with jump discontinuity if it happens for x(t), y(t). Thus g and ϕ are continuous on
γ ′ (t)
[α, β] and of class C 1 on [α, β] ∖ S with ϕ′ (t) = γ(t)−z ϕ(t) on [α, β] ∖ S.
ϕ(t)
Hence the function t ↦ γ(t)−z is continuous on [α, β] with derivative zero on [α, β] ∖ S.
ϕ(t) γ(t)−z
Therefore γ(t)−z = c on [α, β]. But ϕ(α) = 1, thus c =
γ(α)−z and ϕ(t) = c(γ(t) − z) = γ(α)−z .
1

γ(β)−z γ(α)−z
In particular; since γ is a closed path, then ϕ(β) = γ(α)−z = γ(α)−z = 1.
Thus Indγ is an integer valued function on Ω = C ∖ γ ∗ ,i.e. Indγ ∶ Ω ↦ Z .
O2 We will see in the next Lemma that Indγ ∈ C(C∖γ ∗ ). Since the continuous image of a connected

O3 Ind (z) = 0 on the unbounded component ⇐⇒ ∣Ind (z)∣ < 1 for ∣Z∣ sufficiently large.
set is connected, then Ind is constant on every connected component of Ω.
γ
γ γ
Since γ is piece wise C 1 on [α, β], then ∃M > 0 such that ∣γ(s)∣ < M and ∣γ ′ (s)∣ < M . Now for
∣z∣ > M > ∣γ(s)∣ we have

1 β γ ′ (s) 1 β ∣γ ′ (s)∣ 1 β ∣γ ′ (s)∣

∣Indγ (z)∣ = ∣ ∫ ds∣ ≤ ∫ ds ≤ ∫ ds
2πi α γ(s) − z 2π α ∣γ(s) − z∣ 2π α ∣∣γ(s)∣ − ∣z∣∣
1 β ∣γ ′ (s)∣ 1 β M 1 M
≤ ∫ ds ≤ ∫ ds ≤ (β − α) < 1
2π α ∣z∣ − ∣γ(s)∣ 2π α ∣z∣ − M 2π ∣z∣ − M
M (β−α)
provided that ∣z∣ > M + ∣z∣−M . Thus Indγ (z) = 0 on the unbounded component.

1.13 Fundamental Lemma. Let γ be a closed path in C and ϕ ∶ γ ∗ ↦ C be a continuous (or piece
wise continuous) function. For each m ∈ N, define fm ∶ C ∖ γ ∗ ↦ C .
z ↦ ϕ(ξ)
∫γ (ξ−z)m
dξ
∗ ′
Then fm ∈ H(C ∖ γ ) and fm = mfm+1 , ∀m ∈ N. In particular Indγ ∈ H(C ∖ γ ∗ ) ⊂ C(C ∖ γ ∗ ).
Proof. Fix m ∈ N, a ∈ C ∖ γ ∗ and r > 0 such that D(a, r) ⊂ C ∖ γ ∗ .
∞
(z−a)n
● Let ξ ∈ γ ∗ , then 1
ξ−z = ∑ (ξ−a)n+1
∀z ∈ D(a, r).
n=0
∞
(m−1)!
● Differentiate (m − 1) times w.r.t z,then (ξ−z)m = ∑ (n−m+1)! (ξ−a)n+1 (z − a)
n! 1 n−m+1
∀z ∈ D(a, r).
n=m−1
∞
ϕ(ξ) (z−a)n−m+1
∀z ∈ D(a, r), ∀ξ ∈ γ ∗ .
ϕ(ξ)
● Accordingly, (ξ−z)m = ∑ n!
(n−m+1)! (m−1)! (ξ−a)n+1
n=m−1
● Hence,
ϕ(ξ) ∞
ϕ(ξ) (z−a)n−m+1
fm (z) = ∫ dξ = ∫ ∑ n!
(n−m+1)! (m−1)! (ξ−a)n+1
dξ
γ (ξ − z) m γ n=m−1
∞
ϕ(ξ) (z−a)n−m+1
= ∑ ∫γ n!
(n−m+1)! (m−1)! (ξ−a)n+1 dξ
n=m−1
∞
ϕ(ξ)
= ∑ [ (n−m+1)!(m−1)!
n!
∫γ (ξ−a)n+1 dξ](z − a)
n−m+1
n=m−1
∞
(n+m−1)! ϕ(ξ)
= ∑ [ (n)!(m−1)! ∫γ (ξ−a)n+m dξ](z − a)n ∀z ∈ D(a, r).
n=0

● The interchange between summation and integral is due to the uniform convergence of the series:
O1 ϕ is continuous (or piece wise continuous) on the compact set ∶ γ ∗ , Hence ∣ϕ(ξ)∣ ≤ β.

5
O2 ∣ξ − a∣ ≥ r.
O3 ∣ ϕ(ξ) (z−a)n−m+1
(n−m+1)! (m−1)! (ξ−a)n+1 ∣ ≤ (m−1)!∣z−a∣m ∣ [ (n−m+1)!
n! β n!
( ∣z−a∣
n+1
) ] ∶= Mn independent of ξ.
O4 lim = lim
r
∞
n+1 ∣z−a∣ ∣z−a∣
Mn+1
M n n−m+1 r = r < 1 Ô⇒ ∑ Mn is convergent.
n→∞ n→∞ n=m−1
Hence by Weierstrass Test, the above mentioned series of functions in ξ is uniformly convergent.
● Thus fm is representable by a power series in C ∖ γ ∗ . Consequently, fm ∈ H(C ∖ γ ∗ ) and
∞ ∞
′ (n+m−1)! ϕ(ξ) (n+m−1)! ϕ(ξ)
fm (z) = n ∑ [ (n)!(m−1)! ∫γ (ξ−a)n+m dξ](z − a)n−1 = ∑ [ (n−1)!(m−1)! ∫γ (ξ−a)n+m dξ](z − a)n−1
n=1 n=1
∞
(n+m)!
∀z ∈ C ∖ γ ∗ .
ϕ(ξ)
= ∑ [ (n)!(m−1)! ∫γ (ξ−a)n+m+1
dξ](z − a)n = mfm+1 (z)
n=0

Lecture 3.
1.14 Property. If γ is the the positively oriented circle C(a, r) with center a ∈ C and radius r > 0,
then
1, if ∣z − a∣ < r
Indγ (z) = {
0, if ∣z − a∣ > r.

Proof. Note that C ∖ γ ∗ has two components. The unbounded component U = {z ∈ C; ∣z − a∣ > r}
and the bounded component D(a, r) = {z ∈ C; ∣z − a∣ < r.
● Indγ (z) = 0 in the unbounded component i.e. for ∣z − a∣ > r.
● Since Indγ is constant on each component, hence to compute Indγ in the bounded component it
is sufficient to compute Indγ (a).

1 dξ 1 2π γ ′ (t)
Indγ (a) = ∫ = ∫ dt γ(t) = a + reit , 0 ≤ t ≤ 2π
2πi γ ξ − a 2πi 0 γ(t) − a
1 2π ireit
= ∫ dt = 1.
2πi 0 reit
Hence Indγ (z) = Indγ (a) = 1 in the bounded component i.e. for ∣z − a∣ < r.

Notice that if γ is the the positively oriented circle C(a, r) looping three times in the positive
3, if ∣z − a∣ < r
direction , then Indγ (z) = { Because
0, if ∣z − a∣ > r.

1 dξ 1 6π γ ′ (t)
Indγ (a) = ∫ = ∫ dt γ(t) = a + reit , 0 ≤ t ≤ 6π
2πi γ ξ − a 2πi 0 γ(t) − a
1 6π ireit
= ∫ dt = 3.
2πi 0 reit
1.15 Property. Let Ω be an open subset of C and γ ∶ [a, b] ↦ γ ∗ ⊂ Ω be a path in Ω. If F ∈ H(Ω)
and F ′ ∈ C(γ ∗ ), then ∫ F ′ (z) dz = F (γ(b)) − F (γ(a)).
γ
Moreover, if γ is a closed path then ∫ F ′ (z) dz = 0.
γ

b b
Proof. ∫ F ′ (z) dz = ∫ F ′ (γ(t))γ ′ (t) dt = ∫ (F ○ γ)′ (t) dt = F (γ(b)) − F (γ(a)).
γ a a

● We will see later; if F ∈ H(Ω) then F ′ ∈ C(Ω) and hence F ′ ∈ C(γ ∗ ) is always satisfied.
● Let z1 = γ(a), z2 = γ(b) be the initial and final points of the path. Then ∫ F ′ (z) dz =
γ
F (z2 ) − F (z1 ) is independent of the choice of the path inside Ω.

6
1.16 Definition. Let D ⊂ C be a domain, i.e. open and connected, and f ∶ D ↦ C be a continuous
function. A map F ∶ D ↦ C satisfying F ′ = f is said to be a primitive for the function f .

1.17 Remark. If F and G are two primitives of f on a domain D, then (F − G)′ = f − f = 0 on a

connected set D and hence F − G = c ∈ C . In other wards, the difference between two primitives is
a constant.
∞ ∞
1.18 Example. Let f (z) = ∑ cn (z − a)n on D(a, r), then F (z) = ∑ n+1 cn (z − a)
1 n+1
is a primitive
n=0 n=0
of f on D(a, r).

1.19 Corollary. If F is a primitive of f on a domain D and γ ∗ ⊂ D, then ∫ f (z) dz = F (γ(b)) −

γ
F (γ(a)). In particular ∫ f (z) dz = 0 if γ is a closed path in D.
γ

1.20 Example. Let γ be a closed path in C.

● ∫γ z n dz = 0, ∀n ∈ N, γ ∗ ⊂ C. Because F (z) = zn+1 is a primitive of f (z) = z n on the domain C
n+1

and γ ∗ ⊂ C.
● ∫γ z n dz = 0, ∀n = −2, −3, ⋯, ∀γ ∗ ⊂ C∗ . Because F (z) = zn+1 is a primitive of f (z) = z n on the
n+1

domain C∗ and γ ∗ ⊂ C∗ .
● ∫γ z1 dz = ∫γ z−0
dz
= 2πi Indγ (0).

Note that Log(z) is a primitive of 1

z on C ∖ R− where R− = {z = x + iy ∈ C; x ≤ 0, y = 0} .

1.21 Theorem. (Cauchy’s theorem for a triangle)

Let Ω ⊂ C be an open set, p ∈ Ω and ∆ = ∆(z1 , z2 , z3 ) be any triangle subset of Ω. If f ∈
C(Ω) ⋂ H(Ω ∖ {p}), then ∫ f (z) dz = 0.
∂∆

O
Proof. We will discuss four cases:
1 p ∉ ∆:
Let a, b and c be the midpoints of [z1 , z2 ], [z2 , z3 ] and [z3 , z1 ]. Then ∆ is subdivided into four
sub triangles ∆1 (z1 , a, c), ∆2 (z2 , b, a), ∆3 (z3 , c, b) and ∆4 (a, b, c). Since the integrals along common
sides ”path and reverse path” of the triangles ∆i cancel out, then we have

∣∫ f (z) dz∣ = ∣ ∫ f (z) dz + ∫ f (z) dz + ∫ f (z) dz + ∫ f (z) dz∣

∂∆ ∂∆1 ∂∆2 ∂∆3 ∂∆4

≤ ∣∫ f (z) dz∣ + ∣ ∫ f (z) dz∣ + ∣ ∫ f (z) dz∣ + ∣ ∫ f (z) dz∣

∂∆1 ∂∆2 ∂∆3 ∂∆4

≤ 4∣ ∫ f (z) dz∣ where ∆1 is one of the four triangles ∆i

∂∆1

≤ (4)2 ∣ ∫ f (z) dz∣ subdivide ∆1 into 4 triangles and repeat the same steps
∂∆2

≤ (4)n ∣ ∫ f (z) dz∣ proceeding inductively.

∂∆n

Note that ∆ ⊃ ∆1 ⊃ ∆2 ⋯ ⊃ ∆n , where ∆ is a compact set and {∆n } is a sequence of closed subsets
in ∆ that has the finite intersection property. Hence ⋂ ∆n ≠ ∅ .
n≥1
Also diam (∆n ) = 1
2 diam (∆n−1 ) = 1
22
diam (∆n−2 ) = ⋯ = 1
2n−1
diam (∆1 ) = 1
2n diam (∆) → 0.
n→∞
Hence ⋂ ∆ = {z0 } ∈ Ω. Since otherwise z1 ∈ ⋂ ∆ and diam (∆ ) ≥ ∣z1 − z0 ∣.
n n n
n≥1 n≥1
Since ∆ ⊂ Ω, f ∈ H(Ω ∖ {p}), p ∉ ∆ and x0 ∈ ∆, then f is differentiable at x0 . So there exists a
function ε ∶ Ω ↦ C continuous at z0 with ε(z0 ) = 0 such that f (z) = f (z0 )+(z−z0 )f ′ (z0 )+(z−z0 )ε(z).

7
Now let  > 0,

∣∫ f (z) dz∣ ≤4n ∣ ∫ f (z) dz∣ ≤ 4n ∣ ∫ [f (z0 ) + (z − z0 )f ′ (z0 ) + (z − z0 )ε(z)] dz∣

∂∆ ∂∆n ∂∆n

≤ 4n ∣ ∫ (z − z0 )ε(z) dz∣ ≤ 4n max

n
{∣(z − z0 )ε(z)∣}L(∂∆n ) Basic Estimate
∂∆n ∂∆
1 1
≤ 4n max {∣ε(z)∣}L(∂∆n )diam (∆n ) ≤ 4n max {∣ε(z)∣} n
L(∂∆) n diam (∆)
n
∂∆ n ∂∆ 2 2
≤ max
n
{∣ε(z)∣}L(∂∆)diam (∆) < ,
∂∆

because ε(z) → 0, and 

> 0, there exists δ > 0 such that ∣z − z0 ∣ < diam (∆n ) =
z→z0 L(∂∆)diam (∆)
1
2n diam (∆) < δ, then ∣ε(z)∣ < 
Thus ∫ f (z) dz = 0.
.
O L(∂∆)diam (∆) ∂∆
2 p is a vertex: Assume p = z1 and let a ∈]p, z2 [, b ∈]p, z3 [. Then ∆(p, z2 , z3 ) is subdivided into
three triangles ∆1 (a, b, p), ∆2 (a, z2 , z3 )) and ∆3 (a, z3 , b). Let  > 0,

∣∫ f (z) dz∣ = ∣ ∫ f (z) dz + ∫ f (z) dz + ∫ f (z) dz∣

∂∆ ∂∆1 ∂∆2 ∂∆3

= ∣∫ f (z) dz∣ ( since p ∉ ∆2 , p ∉ ∆3 )

∂∆1
≤ M L(∂∆1 ) ( since f is continuous on the compact set ∂∆1 )
≤ M ∣p − a∣ + M ∣a − b∣ + M ∣b − p∣ ≤ M ∣p − a∣ + M ∣a − p∣ + M ∣p − b∣ + M ∣b − p∣
≤ 2M ∣p − a∣ + 2M ∣b − p∣ ( let a, b → p)
 
≤ 2M + 2M < .
4M 4M
O3 p lies on one edge: Assume p ∈]z1 , z3 [. Then ∆(z1 , z2 , z3 ) is subdivided into two triangles
∆1 (p, z1 , z2 ) and ∆2 (p, z2 , z3 ). Hence

∫ f (z) dz = ∫ f (z) dz + ∫ f (z) dz = 0

∂∆ ∂∆1 ∂∆2

O
because p is a vertex of ∆1 and ∆2 .
4 p is in the interior of ∆: Then ∆(z1 , z2 , z3 ) is subdivided into three triangles ∆1 (p, z1 , z2 ), ∆2 (p, z2 , z3 )
and ∆2 (p, z3 , z1 ). Hence

∫ f (z) dz = ∫ f (z) dz + ∫ f (z) dz + + ∫ f (z) dz = 0

∂∆ ∂∆1 ∂∆2 ∂∆3

because p is a vertex of ∆1 , ∆2 and ∆3 .

This Theorem plays the same role as Chasles relation for Riemann integral. In other wards,
if f ∈ C(Ω) ⋂ H(Ω ∖ {p}), a, z, z0 ∈ Ω, then

∫ f (ξ) dξ = 0 ⇐⇒ ∫ f (ξ) dξ − ∫ f (ξ) dξ = ∫ f (ξ) dξ.

∂∆(a,z,z0 ) [a,z] [a,z0 ] [z0 ,z]

1.22 Definition. A set E ⊂ C is convex if for each pair of points a, b ∈ E we have [a, b] ⊂ E.

1.23 Theorem. (Cauchy’s theorem for a convex set)

Let Ω be a nonempty open convex set in C and p ∈ Ω. If f ∈ C(Ω) ⋂ H(Ω ∖ {p}), then there exists
F ∈ H(Ω) such that F ′ = f on Ω. Moreover, ∫ f (z) dz = 0 for every closed path γ in Ω.
γ
Proof.
● Fix a ∈ Ω and let F (z) = ∫ f (ξ) dξ, ∀z ∈ Ω.
[a,z]
● Since a, z ∈ Ω and Ω convex, then [a, z] ⊂ Ω. Hence f ∈ C([a, z]) and F is well defined.

8
● Claim F is a primitive of f on Ω: Let  > 0, z0 ∈ Ω such that z0 ≠ z.
O1 Since f is continuous at z0 , then there exists δ > 0 such that ∣f (z)−f (z0 )∣ <  whenever ∣z−z0 ∣ < δ.
O2 For ∣z − z0 ∣ < δ,
F (z) − F (z0 ) ∫[a,z] f (ξ) dξ − ∫[a,z0 ] f (ξ) dξ ∫[z ,z] 1 dξ
∣ − f (z0 )∣ = ∣ − f (z0 ) 0 ∣ (∵∫ 1 dξ = ξ∣zz0 = z − z0 )
z − z0 z − z0 z − z0 [z0 ,z]

∫[z ,z] f (ξ) dξ ∫[z0 ,z] f (z0 ) dξ

=∣ 0 − ∣ ( ∵ f ∈ C(Ω) ⋂ H(Ω ∖ {p}))
z − z0 z − z0
1
= ∣∫ [f (ξ) − f (z0 )] dξ∣ (∣ξ − z0 ∣ ≤ ∣z − z0 ∣ < δ Ô⇒ ∣f (ξ) − f (z0 )∣ < )
∣z − z0 ∣ [z0 ,z]
1
≤ (L([z0 , z])) ≤  (Basic Estimate)
∣z − z0 ∣
Hence F is differentiable at z0 and F ′ (z0 ) = f (z0 ), ∀z0 ∈ Ω. Thus F ∈ H(Ω) and F ′ = f on Ω.
1.24 Theorem. (Cauchy’s integral formula)
Let γ be a closed path in a convex open set Ω and f ∈ H(Ω). Then
1 f (ξ)
f (z)Indγ (z) = ∫ dξ, ∀z ∈ Ω ∖ γ ∗ .
2πi γ ξ − z
Proof. Let z ∈ Ω ∖ γ ∗ .
1 f (ξ) f (ξ) − f (z)
● Claim: f (z)Indγ (z) = ∫ dξ, ∀z ∈ Ω ∖ γ ∗ ⇐⇒ ∫ dξ = 0.
2πi γ ξ − z γ ξ−z
f (ξ)−f (z)
, if ξ ∈ Ω and ξ ≠ z
● Define g on Ω by g(ξ) = { ′ ξ−z
f (z), if ξ = z.
Note that g ∈ C(Ω) ⋂ H(Ω ∖ {z}). Then by the previous theorem ∫ g(ξ) dξ = 0.
γ

1.25 Theorem. (Holomorphy implies Analyticity)

∞
If f ∈ H(D(a, R)), then f (z) = ∑ an (z − a)n ∀z ∈ D(a, R). In other words
n=0
f ∈ H(Ω) Ô⇒ ∀D(a, R) ⊂ Ω, f ∈ H(D(a, R)) Ô⇒ f is representable by a power series in Ω.
Proof. Let γ(t) = a + reit , t ∈ [0, 2π] and r < R. We have γ is a closed path in the open convex set
D(a, R) and f ∈ H(D(a, R)). Then
1 f (ξ)
f (z) = f (z)Indγ (z) = ∫ dξ ∀z ∈ D(a, r) (by previous theorem )
2πi γ ξ − z
∞
f (ξ)
= ∑ ( 2πi
1
∫γ (ξ−a)n+1
dξ)(z − a)n (f ∈ C(γ ∗ ), by fundemental lemma f is rep. by p.s on C ∖ γ ∗ )
n=0
∞
= ∑ cn (z − a)n ∀z ∈ D(a, r) (∵D(a, r) ⊂ C ∖ γ ∗ )
n=0

1 f (ξ) f (n) (a)

where cn = ∫ dξ = . Since the power series is obtained for all r < R, then
2πi γ (ξ − a)n+1 n!
the representation is valid for all z ∈ D(a, R).
1.26 Corollary. Let k ∈ N.
f ∈ H(Ω) ⇐⇒ f is representable by a p.s in Ω ⇐⇒ f is analytic in Ω ⇐⇒ f (k) ∈ H(Ω).

O
1.27 Corollary.
1 If Ω is open convex and f ∈ C(Ω) ⋂ H(Ω ∖ {p}), then f ∈ H(Ω).
O2 If Ω is open and f ∈ C(Ω) ⋂ H(Ω ∖ {p}), then f ∈ H(Ω).
1.28 Theorem. (Morera’s theorem)
Let Ω ⊂ C be open. If f ∈ C(Ω) and ∫∂∆ f (z) dz = 0 for every closed triangle ∆ ∈ Ω, then f ∈ H(Ω).
The proof of the last theorem and corollary left as exercises.

9
Exercise 1. Let D ⊆ C be a domain, f ∈ H(D) and f ′ ≡ 0 on D. Show , using the integral over a
path, that f is a constant function on D.
Exercise 2. Let n ∈ N∗ and γ be the positively oriented circle of center a ∈ C and radius r > 0
looping n−times. Calculate Indγ on C ∖ γ ∗ .
zez
Exercise 3. Let f (z) = 2 .
z +1
1 ez ez
1. Show that f (z) = [ + ].
2 z−i z+i

2. Compute ∫ f (z) dz, where γ ∶ [0, 2] ↦ C .

γ ↦ 2iπt t 2e

3. Compute ∫ f (z) dz, where γ = γ1 ∪ γ2 with γ1 ∶ [0, π] ↦ C and γ2 = [−2, 2].

γ ↦ it t 2e

Exercise 4. Let 0 < r < R and γ(t) = reit , 0 ≤ t ≤ 2π.

1 R+z
1. Calculate I = ∫ dz.
2πi γ z(R − z)

1 2π R2 − r 2
2. Deduce that ∫ dt = 1.
2π 0 R2 − 2rR cos t + r2
Exercise 5. Corollary 1.27.
Exercise 6. Theorem 1.28.
Exercise 7.(Used in the proof of the fundamental lemma)

1. Let {sn }n be a sequence of continuous functions converging uniformly on Ω to a function f

and γ ∶ [a, b] ↦ Ω be a path in Ω. Show that lim ∫γ sn (z) dz = ∫γ f (z) dz.
n→∞
∞
2. Let f = ∑ fn be a series of continuous functions converging uniformly on Ω and γ ∶ [a, b] ↦ Ω
n=0
∞
be a path in Ω. Show that ∫γ f (z) dz = ∑ ∫γ fn (z) dz.
n=0
∞
Exercise 8. Let Ω be open subset of C, {fn }n≥1 ⊂ H(Ω), ∑ fn is a convergent series in Ω and
n=1
∞ ∞ ∞ ∞
∑ fn′ is uniformly convergent on Ω. Show that ∑ fn ∈ H(Ω) and ( ∑ fn )′ = ∑ fn′ .
n=1 n=1 n=1 n=1
ez
Exercise 9. Let γ ∶ [0, π] ↦ C be a path. Show that ∣ ∫γ z dz∣ ≤ πe.
↦ it t e
Exercise 10. Let f (z) = Log(3 + z).

1. Determine the domain of holomorphy of f .

f (z)
2. Compute ∫γ z dz where γ is a circle centered at the origin of radius 2 and looping three
times in the positive direction.

3. Same question, but γ looping three times in the negative direction.

Exercise 11.Let γ ∶ [0, 1] ↦ C be a path and D any domain containing containing γ ∗ but
t ↦ ei2πt
excluding zero.

1. Show that f (z) = 1

z has no primitive on D.

2. Deduce that there is no branch of log on D.

3. Deduce that there is no branch of log on C∗ .

10
Exercise 12. Let D be an open convex subset of C, z0 ∈ D, f ∈ H(D) and f ′ (z) ≠ 0 on D.
f′
1. Show that f has a primitive F on D.

2. Let ψ(z) = f (z)e−F (z) ∀z ∈ D. Show that ψ ∈ H(D) and ψ ′ ≡ 0 on D.

3. Deduce that f (z) = eg(z) where g(z) = F (z) − F (z0 ) + log(f (z0 )).

4. Deduce that ∀n ∈ N, ∃h ∈ H(D) s.t hn = f on D i.e. h is an nth root of f .

1
5. Give another (n − 1) branches of f n .

11
CHAPTER 2
ZERO’S AND SINGULARITIES OF HOLOMORPHIC FUNCTIONS

Lecture 4.
2.1 Definition. Let Ω ⊂ C be open and f ∈ H(Ω). The zero set of f is
Z(f ) = {a ∈ Ω; f (a) = 0} = f −1 ({0}).
Note that:
● If f is a continuous function, then Z(f ) is a closed set.
● Z(f ) = Ω ⇐⇒ f = 0 on Ω, Z(f ) ≠ Ω ⇐⇒ f is not the zero function on Ω.
2.2 Theorem. Let Ω be a domain and f ∈ H(Ω). If there exists a point a ∈ Ω such that f (n) (a) = 0
for all n ∈ N, then f = 0 on Ω i.e. Z(f ) = Ω.
Proof. Suppose that there exists a ∈ Ω such that f (n) (a) = 0 for all n ∈ N. Define
−1
A = {a ∈ Ω; f (n) (a) = 0 ∀n ∈ N} = ⋂ f (n) ({0}).
n∈N

● A is an open subset of Ω:
∞
f (n) (a)
a ∈ A ⊂ Ω Ô⇒ ∃D(a, r) ⊂ Ω Ô⇒ f (z) = ∑ n! (z − a)n = 0 on D(a, r) Ô⇒ D(a, r) ⊂ A.
n=0
● A is a closed subset of Ω:
−1
f ∈ H(Ω) Ô⇒ f (n) ∈ H(Ω) ⊂ C(Ω) ∀n ∈ N Ô⇒ f (n) ({0}) is closed Ô⇒ A is closed .
Hence A is a clopen subset of Ω. But Ω is a connected set, then the only clopen subsets in Ω are
∅ and Ω. Since A ≠ ∅ (a ∈ A), then A = Ω. Thus f = 0 on Ω.

2.3 Corollary. Let Ω be a domain, f ∈ H(Ω) and Z(f ) ≠ Ω. Then for all a ∈ Z(f ), there exists a
unique positive integer m ∈ N∗ such that f (z) = (z − a)m g(z) ∀z ∈ Ω, where g ∈ H(Ω) and g(a) ≠ 0.
m is called the order of the zero a and it is characterized by; m is the smallest positive integer such
that f (m) (a) ≠ 0.
Proof. Let a ∈ Z(f ).
∞
f (n) (a)
a ∈ Z(f ) ⊂ Ω Ô⇒ ∃D(a, r) ⊂ Ω Ô⇒ f (z) = ∑ cn (z − a)n on D(a, r), cn = n! , c0 = f (a) = 0.
n=0
f (n) (a)
Suppose to the contrary that cn = n! = 0 for all n ∈ N∗ , then by the previous theorem Z(f ) = Ω
which contradicts the assumption Z(f ) ≠ Ω. Hence there exists a smallest positive integer m ∈ N∗
such that cm ≠ 0 i.e.
∞ ∞ ∞
f (z) = ∑ cn (z − a)n = ∑ cn+m (z − a)n+m = (z − a)m ∑ cn+m (z − a)n ∀z ∈ D(a, r)
n=m n=0 n=0
⎧
⎪ f (z)
⎪ m , if z ∈ Ω ∖ {a}
. Define g(z) = ⎨ (z−a) f (m) (a) It is clear that g ∈ H(Ω ∖ {a}) ⋂ C(Ω), hence g ∈ H(Ω).
⎪
⎪ c = , if z = a.
⎩ m m!
Thus f (z) = (z − a)m g(z) ∀z ∈ Ω, where g ∈ H(Ω) and g(a) = cm ≠ 0.

12
If m = 1, then the zero is called a simple zero.
2
2.4 Example. Let f (z) = ez − 1. We have 0 ∈ Z(f ). The order of the zero m = 2:
We have f (0) = 0, f ′ (0) = 2zez ∣z=0 = 0 and f ′′ (0) = 2ez + 4z 2 ez ∣z=0 = 2 ≠ 0. Thus m = 2.
2 2 2

∞ n
z2
One can deduce the order from the p.s.r. f (z) = ( ∑ n! ) − 1 = 0 + 0z + 1z 2 + 0z 3 + 12 z 4 + ⋯.
n=0
Hence f (0) = f ′ (0) = 0 and f ′′ (0) = 2!c2 = 2 ≠ 0. Thus m = 2.
2.5 Definition. Let Ω ⊂ C be an open set. D ⊂ Ω is called discrete in Ω if and only D has no limit
point in Ω. i.e
x ∈ Ω Ô⇒ x ∉ D′ ( i.e ∃D(x, r) s.t D(x, r) ⋂ D ∖ {x} = φ).
Note that if x ∈ D, then x is an isolated point of D. i.e x ∈ D ∖ D′ ( ∃D(x, r) s.t D(x, r) ⋂ D = {x}).
2.6 Example. Let D = {1, 21 , 13 , ⋯}. Then D is discrete in Ω = C∗ but D is not discrete in Ω = C.
2.7 Theorem. Let Ω be a domain, f ∈ H(Ω) and Z(f ) ≠ Ω. Then Z(f ) is closed discrete in Ω,
hence all zeros of f are isolated points of Z(f ). Moreover Z(f ) is at most countable.
Proof. Let a ∈ Ω and claim a ∉ Z(f )′ :
● If a ∉ Z(f ) = Z(f ) = Z(f ) ⋃ Z(f )′ , then a ∉ Z(f )′ .
● If a ∈ Z(f ), there exists a unique positive integer m ∈ N∗ such that f (z) = (z − a)m g(z) ∀z ∈ Ω,
where g ∈ H(Ω) and g(a) ≠ 0.
Now; g ∈ C(Ω) and g(a) ≠ 0 Ô⇒ g(z) ≠ 0 on D(a, r) Ô⇒ f (z) ≠ 0 on D(a, r) ∖ {a}.
Hence there exists D(a, r) s.t D(a, r) ⋂ Z(f ) ∖ {a} = φ. Thus a ∉ Z(f )′ .
Claim Z(f ) is countable:
Let K be a compact set in Ω, then Z(f ) ⋂ K is a compact and discrete set in Ω thus finite. In fact,
n
Z(f ) ⋂ K ⊂ ⋃ D(x, εx ) Ô⇒ Z(f ) ⋂ K ⊂ ⋃ D(xi , εxi ), xi ∈ Z(f ) ⋂ K .
x∈Z(f ) ⋂ K i=1
Since the zeros of f are isolated, then Z(f ) ⋂ K ⊂ {xi }ni=1 is finite. We know that C ≡ R2 is
○
σ−compact i.e. there exists a sequence {Kn }n∈N of compact sets in Ω such that Kn ⊂ Kn+1 and
Ω = ⋃ Kn . Hence Z(f ) = Z(f ) ⋂ Ω = ⋃ (Kn ∩ Z(f )) is a countable union of finite sets and thus
n∈N n∈N
countable.

Note that if Ω is a domain, f ∈ H(Ω) and Z(f ) has a limit point in Ω, then Z(f ) = Ω.
2.8 Example. The set of zeros of the function f (z) = sin z is Z(f ) = {nπ ∶ n ∈ Z} and The set of
zeros of the function g(z) = cos z is Z(g) = { π2 + nπ ∶ n ∈ Z}. In fact ,
sin z = 0 ⇐⇒ e −e = 0 ⇐⇒ eiz = e−iz ⇐⇒ e2iz = e0 ⇐⇒ 2iz = 0+2inπ, n ∈ Z ⇐⇒ z = nπ, n ∈ Z
iz −iz
2i
2.9 Theorem. (Identity Theorem for holomorphic functions on Domains)
Let Ω be a domain and f, g ∈ H(Ω). If f = g on a set A which has a limit point in Ω, then f = gon Ω.
Proof. Let h = f − g on Ω. Then A ⊂ Z(h). Since Ω is a domain, h ∈ H(Ω) and Z(h) has a limit
point in Ω, then Z(h) = Ω or equivalently f = g on Ω.

● For the identity theorem, the limit point of A must be in Ω as the following example shows:
Ω = C∗ , f (z) = sin z1 , g(z) = 0 and f = g on A = Z(f ) = { nπ1
∶ n ∈ Z}. A has the limit point 0 ∉ Ω.
● It becomes clear now that the real functions sin, cos, exp, sinh, cosh ∶ R ↦ R can be extended
uniquely to C.
● Functional identities can be transformed from real to complex analysis:
We know that sin2 z + cos2 z = 1, ∀z ∈ R. Let f (z) = sin2 z + cos2 z and g(z) = 1. We have Ω = C is
a domain, f, g ∈ H(Ω) and f = g on A = R which has a limit point in Ω = C.(In fact all points of R
are limit points of R in C) Thus by the identity theorem we have f (z) = g(z) ∀z ∈ C.
We know that cos x cos y = 21 ( cos(x + y) + cos(x − y)) ∀x, y ∈ R. Then by the identity theorem we
obtain cos z cos y = 12 ( cos(z + y) + cos(z − y)) ∀z ∈ C, y ∈ R and again by the identity theorem we
obtain cos z cos w = 21 ( cos(z + w) + cos(z − w)) ∀z, w ∈ C.

13
2.10 Definition. Let Ω ⊂ C be open, a ∈ Ω and f ∈ H(Ω ∖ {a}), then f is said to have an isolated
singularity at the point a. The singularity is said to be removable if f can be defined at a so that
the extended function is holomorphic on Ω.
In other words, f has an isolated singularity at the point a if there exists r > 0 such that f ∈
H(D′ (a, r)). The singularity is said to be removable if f can be defined at a so that the extended
function is holomorphic on D(a, r).

2.11 Recall. If f has an isolated singularity at the point a(i.e f ∈ H(Ω ∖ {a})) and f can be
extended by continuity at a (i.e f ∈ C(Ω) can be regarded as continuous function on Ω), then
f ∈ H(Ω) (i.e a is a removable singularity).
In other words, the isolated singularity a is removable if f can be extended by continuity at a.

2.12 Examples.

1. Let f (z) = 1
(z−1)(z−3) . z = 1 and z = 3 are isolated singularities because f ∈ H(D′ (1, 21 )) and
f∈ H(D′ (3, 41 )).

2. Let f (z) = Logz. z = −1 is not an isolated singularity, because f ∉ H(D′ (−1, r) for all r > 0.

2.13 Theorem. Let Ω ⊂ C be open, a ∈ Ω and f ∈ H(Ω ∖ {a}). If f is bounded on a punctured disc
D′ (a, r) for some r > 0, then f has a removable singularity at a.

(z − a)2 f (z), if z ∈ Ω ∖ {a}

Proof. Let h(z) = { It is clear that h ∈ H(Ω ∖ {a}). Also
0, if z = a.
h(z) − h(a)
lim = lim (z − a)f (z) = 0 ∵f is bounded on D′ (a, r). Hence h ∈ H(Ω) and h′ (a) = 0. So
z→a z−a z→a
∞ ∞ ∞
h(z) = ∑ cn (z − a)n on D(a, r). consequently, f (z) = ∑ cn (z − a)n−2 = ∑ cn+2 (z − a)n on D(a, r).
n=2 n=2 n=0
Define f (a) = c2 . Thus f ∈ H(Ω).

2.14 Remark. The converse of the above theorem is also true. In fact, if a is a removable
singularity of f , then f ∈ H(D(a, 2r)) for some r > 0. Hence f is continuous on the compact
set D(a, r) ⊂ D(a, 2r) and thus bounded on D′ (a, r) ⊂ D(a, r).

2.15 Theorem. Let f ∈ H(D′ (a, r)). Then the following are equivalent:

1. f has a removable singularity at a.(i.e f can be extended by holomorphy at a)

2. ∃` ∈ C s.t lim f (z) = ` .(i.e f can be extended by continuity at a)

z→a

3. lim (z − a)f (z) = 0 .

z→a

Proof. Exercise.

2.16 Theorem. Let Ω be open, a ∈ Ω and f ∈ H(Ω ∖ {a}). Then one of the following must occur:

i. f has a removable singularity at a.

ii. f has a a pole of order m at a:

There are complex numbers c1 , . . . , cm , where m is a uniquely determined positive integer and
m
cm ≠ 0, such that f (z) − ∑ ck
(z−a)k
has a removable singularity at a.
k=1
m ∞
Equivalently, f (z) − ∑ (z−a)
ck
k = ∑ an (z − a)n , ∀z ∈ D(a, r) ⊆ Ω.
k=1 n=0
m
The function z ↦ ∑ (z−a)k is called the
ck
principal part of f at the pole a.
k=1

14
 iii. f has an essential singularity at a:(Casorati-Weierstrass Theorem)
If r > 0 and D(a, r) ⊂ Ω, then f (D′ (a, r)) is dense in C.
Equivalently, ∀w ∈ C, ∃{zn } ⊂ C such that zn → a and f (zn ) → w.
n→∞
Proof. Suppose iii. fails. Claim i. or ii. occurs:
If iii. fails, then ∃r > 0 s.t D(a, r) ⊂ Ω and f (D′ (a, r)) is not dense in C i.e. C ⊈ f (D′ (a, r)) i.e.
∃w ∈ C s.t w ∉ f (D′ (a, r)). Hence
∃r > 0, ∃δ > 0, ∃w ∈ C s.t D(w, δ) ⋂ f (D′ (a, r)) = φ i.e.
∃r > 0, ∃δ > 0, ∃w ∈ C s.t ∀z ∈ D′ (a, r), f (z) ∉ D(w, δ) i.e.
∃r > 0, ∃δ > 0, ∃w ∈ C s.t ∀z ∈ D′ (a, r), ∣f (z) − w∣ > δ.
1
Define g(z) = (z ∈ D′ (a, r)). We have g ∈ H(D′ (a, r)) and ∣g∣ < 1δ on D′ (a, r), then by the
f (z) − w
previous theorem g has a removable singularity at a i.e. g ∈ H(D(a, r)). We have two possibilities:
P1. If g(a) ≠ 0, then f (z) = g(z)
1
+ w is bounded in D′ (a, ρ) for some ρ > 0.(∵f (z) → 1
g(a) + w).
z→a
Hence i. holds by the previous theorem.
P2. If g(a) = 0, then ∃m ∈ N∗ s.t g(z) = (z − a)m g1 (z) where g1 ∈ H(D(a, r)) and g1 (a) ≠ 0. Note
′
that g1 (z) = (z−a)
1
m f (z)−w has no zero in D (a, r). put h = g in D(a, r). Then h ∈ H(D(a, r))
1 1
1
∞
and h has no zeros in D(a, r) i.e. h(z) = ∑ bn (z − a)n on D(a, r) with b0 ≠ 0. Now,
n=0
∞ ∞
−m
f (z) − w = (z − a) h(z) = ∑ bn (z − a) n−m
= ∑ bn+m (z − a)n on D(a, r). Hence ii. holds,
n=0 n=−m
with ck = bm−k , k = 1, . . . , m and cm = b0 ≠ 0.
2.17 Theorem. Let f ∈ H(D′ (a, r)). Then the following are equivalent:
1. f has a pole of order m at a.
g(z)
2. ∃g ∈ H(D(a, r)) with g(a) ≠ 0 s.t f (z) = (z−a)m .

3. lim (z − a)m f (z) = `, ` ≠ 0 .

z→a
Proof. Exercise.
sin(z)
sin(z)
2.18 Example. f (z) = z4
has a pole of order 3 at 0. In fact, f (z) = z
z3
or lim z 3 f (z) = 1 ≠ 0.
z→0
2.19 Theorem. Let f ∈ H(D′ (a, r)). Then f has a pole at a if and only if lim ∣f (z)∣ = ∞.
z→a
Proof. Exercise.
2.20 Remark.
● Let f ∈ H(D′ (a, r)). Then f has essential singularity at a if and only if lim f (z) fails to converge
z→a
and lim ∣f (z)∣ fails to diverge to ∞. How bad can this failure be? To show that a is an essential
z→a
singularity, it is sufficient to find two sequences {un }n , {vn }n s.t un Ð→ a and vn Ð→ a but
n→∞ n→∞
lim f (un ) ≠ lim f (vn ).
n→∞ n→∞
● Big theorem of Picard: If f has an essential singularity at a and r > 0 then f (D′ (a, r)) = C or
f (D′ (a, r)) = C ∖ {c} for some c ∈ C.
2.21 Examples.
1. Let f (z) = e z . Note that, z = 0 is an isolated singularity because f ∈ H(D′ (0, 1)). This
1

singularity is essential. In fact, un = n1 Ð→ 0 and vn = − n1 Ð→ 0 but lim f (un ) ≠ lim f (vn ).

n→∞ n→∞ n→∞ n→∞

2. Let f (z) = sin( z1 ). Note that, z = 0 is an isolated singularity because f ∈ H(D′ (0, 1)). This
singularity is essential. In fact, un = n1 Ð→ 0 and lim f (un ) = lim sin n ∄. Hence lim f (z)
n→∞ n→∞ n→∞ z→0
fails to converge and lim ∣f (z)∣ fails to diverge to ∞.
z→0

15
Exercise 1. Let f (z) = z sin( z1 ) for all z ∈ D′ (0, 1).

1. Determine Z(f ).

2. Show that Z(f ) has a cluster point in C.

3. We have f ≡ 0 on a set which has a cluster point. Justify why f ≢ 0 on D′ (0, 1).

4. Does f have a removable singularity at 0. Justify your answer.

Exercise 2. Let Ω be open convex, z0 ∈ Ω, γ is a closed path s.t z0 ∉ γ ∗ and f ∈ H(Ω) s.t z0 is a
f ′ (z)
∫γ f (z) dz = m Indγ (z0 ).
1
zero of order m of f . Show that 2πi
Exercise 3. (Complex L’Hospital Rule)
Let Ω be open z0 ∈ Ω, f, g ∈ H(Ω) s.t z0 is a zero of order m for both f and g. Show that h = fg has
f (z) f (m) (z0 )
a removable singularity at z0 and lim = (m) .
z→z0 g(z) g (z0 )
Exercise 4. Let Ω be a domain in C, f, g ∈ H(Ω) and Z(f ) = Z(g) = φ on Ω. Suppose that there
f ′ (a ) g ′ (a )
exists a sequence {an }n ⊂ Ω and a ∈ Ω s.t an Ð→ a and f (ann) = g(ann) for all n. Show that f = λg
n→∞
on Ω for some λ ∈ C.
Exercise 5. Theorem 2.15
Exercise 6. Theorem 2.17
Exercise 7. Theorem 2.19

16
CHAPTER 3
CAUCHY THEORY AND ITS APPLICATIONS

Lecture 5.

17