I

Lecture Notes on Complex Analysis I (2017)

BH
Author:
DA Prakash A. Dabhi

Department of Mathematics, Sardar Patel University, Vallabh Vidyanagar - 388120, Gujarat

E-mail address: lightatinfinite@[Link]
PA
PA
DA
BH
I
 I
Contents

BH
1. Complex numbers 4
2. Modulus and conjugate of a complex number 4
3. Argument and Principal Argument 7
4. n-th root of a complex number 10
5. Some Topological Aspects 10
6. Functions and Their Limits 11
7. Differentiability of Complex Functions 16
8. Cauchy-Riemann Equations 18
9. Analytic Functions 21
10. Harmonic Functions
DA 24
11. Derivative of w(t) 27
12. Integral of w(t) 27
13. Contours 29
14. Integral of a function 31
15. Cauchy and Cauchy-Goursat Theorems 35
16. Cauchy Integral Theorem 36
17. Maximum Modulus Principle 41
18. Taylor’s Theorem 44
19. Power series an analytic function 47
20. Laurent’s Theorem 50
PA

21. Singularities of a function 54

22. Evaluation of Improper Integrals 61
23. Bilinear Transformations 71

3
4 CONTENTS

1. Complex numbers
An ordered pair z = (x, y) is called a complex number, where x and y are real numbers. Let
C denote the set of all complex numbers.
A complex number of the form z = (0, y) is called purely imaginary. A real number x can be
identified with (x, 0). When z = (x, y) is a complex number, x is called the real part of z and y is

I
called the imaginary part of z and we write Re z = x and Im z = y. Thus z = (Re z, Im z). Two
complex numbers z1 and z2 are equal if Re z1 = Re z2 and Im z1 = Im z2 .

BH
If z1 = (x1 , y1 ) and z2 = (x2 , y2 ) are complex numbers, then we define their addition and
multiplication as follows.
z1 + z2 = (x1 + x2 , y1 + y2 ),
z1 z2 = (x1 x2 − y1 y2 , x1 y2 + x2 y1 ).
The above operations become usual when they are restricted to the real numbers, i.e.,
(x1 , 0) + (x2 , 0) = (x1 + x2 , 0) and (x1 , 0)(x2 , 0) = (x1 x2 , 0).
One may see easily that (0, 1)(y, 0) = (0, y). Thus (x, y) = (x, 0) + (0, y) = (x, 0) +
(0, 1)(y, 0). Thus if we think of a real number as x or (x, 0) and if we write i = (0, 1), then z = x+iy.
DA
One can see easily that i2 = ii = (0, 1)(0, 1) = (−1.0) = 1. Thus i is a solution of the equation
x2 + 1 = 0. If we represent z1 = x1 + iy1 and z2 = x2 + iy2 , then z1 + z2 = (x1 + x2 ) + i(y1 + y2 )
and z1 z2 = (x1 x2 − y1 y2 ) + i(x1 y2 + x2 y1 ).
Exercise 1.1. If z ∈ C, then show that Re (iz) = −Imz and Im (iz) = Re z.
Since C is a field, it has no zero divisors. It means that if z, w ∈ C and zw = 0, then z = 0 or
w = 0. In other words, the multiplication of two nonzero complex numbers is nonzero.
Examples 1.2.
(1) Solve the equation z 2 − 2z + 2 = 0.
Suppose that z = (x, y) is a solution of z 2 − 2z + 2 = 0. Then (x, y)(x, y) − 2(x, y) + (2, 0) =
(0, 0), i.e., (x2 − y 2 , 2xy) − (2x, 2y) + (2, 0) = (0, 0). This gives x2 − y 2 − 2x + 2 = 0 and
PA

2y(x − 1) = 0. If y = 0, then x2 − 2x + 2 = 0. It means that (x − 1)2 = −1 for some real x,

This is not possible. Hence y 6= 0. Since y 6= 0, x = 1. Since x = 1, 1 − y 2 − 2 + 2 = 0, i.e.,
y = ± = 1. Thus z = (x, y) = (1, ±1) or z = 1 ± i is a solution of z 2 − 2z + 2 = 0.

2. Modulus and conjugate of a complex number

We may consider a complex number z = x + p iy as a directed line segment or a vector.
If z = x + iy is a complex number, then |z| = x2 + y 2 is called the modulus or the absolute
value of z. p
Geometrically, |z| = x2 + y 2 is the distance between the points (x, y) and origin. We note
that if z = (x, 0), then |z| = |x| and if z = (0, y), then |z| = |y|.
2. Modulus and conjugate of a complex number 5

p
If z1 = x1 + iy1 and z2 = x2 + iy2 are in C, then |z1 − z2 | = (x1 − x2 )2 + (y1 − y2 )2 is the
distance between the points (x1 , y1 ) and (x2 , y2 ).
Examples 2.1.
(1) If z ∈ C, then Re z ≤ |Re z| ≤ |z| and Im z ≤ |Im z| ≤ |z|.

I
We have z = Re z +iIm z. Therefore |z|2 = (Re z)2 +(Im z)2 . This gives (Re z)2 ≤ (Re z)2 +
(Im z)2 and (Im z)2 ≤ (Re z)2 + (Im z)2 . Hence Re z ≤ |Re z| ≤ |z| and Im z ≤ |Im z| ≤ |z|.

BH
(2) Let R > 0 and z0 = a0 + ib0 ∈ C. What does the set {z ∈ C : |z − z0 | = R} represent?
Let z = x + iy ∈ C. Then |z − z0 | = R if and only if |(x − a) + i(y − b)| = R if and only if
(x − a)2 + (y − b)2 = R2 . Thus the above set is a circle with center z0 = x0 + iy0 and radius
R.
Thus the locus of a point z satisfying |z − 1 + 3i| = 2 is a circle with center 1 − 3i and radius 2.
Definition 2.2. Let z = x + iy ∈ C. Then the complex number z = x − iy is called the complex
conjugate or conjugate of z.
Exercise 2.3. Let z, z1 and z2 be in C. Show that
(1) z = z
(2) (z1 ± z2 ) = z1 ± z2
DA
(3) z1 z2 = z1 z2
(4) If z2 6= 0, then ( zz12 ) = zz12
(5) |z| = | − z| = |z|
(6) zz = |z|2
(7) If z2 6= 0, then zz12 = |z 1|
|z2 |
(8) z + z = 2Re z and z − z = −2i Im z
Example 2.4. If z1 , z2 , . . . , zn are in C, then z1 z2 · · · zn = z1 z2 · · · zn .
Proof. We shall prove it by using Principle of Mathematical Induction. Let z1 = x1 + iy1 and
PA

z2 = x2 + iy2 are in C. Then

z1 z2 = (x1 x2 − y1 y2 ) + i(x1 y2 + x2 y1 ) = (x1 x2 − y1 y2 ) − i(x1 y2 + x2 y1 )
= (x1 − iy1 )(x2 − iy2 ) = z1 z2 . (2.4.1)
Assume that the result is true for n − 1, i.e., z1 z2 · · · zn−1 = z1 z2 · · · zn−1 for all z1 , z2 , . . . , zn−1 ∈ C.
Let z1 , z2 , . . . , zn ∈ C. Then
z1 z2 · · · zn = (z1 z2 · · · zn−1 zn ) = z1 z2 · · · zn−1 zn [ by 8.3.1]
= (z1 z2 · · · zn−1 )zn [ by Induction hypothesis]
= z1 z2 · · · zn .
This proves the result. 
6 CONTENTS

Theorem 2.5 (Triangle Inequality). If z1 and z2 are complex numbers, then |z1 + z2 | ≤ |z1 | + |z2 |.
Proof. We see that
|z1 + z2 |2 = (z1 + z2 )z1 + z2 = (z1 + z2 )(z1 + z2 )
= z1 z1 + z1 z2 + z1 z2 + z2 z2

I
= |z1 |2 + z1 z2 + z1 z2 + |z2 |2
= |z1 |2 + 2Re (z1 z2 ) + |z2 |2

BH
≤ |z1 |2 + 2|z1 z2 | + |z2 |2
= |z1 |2 + 2|z1 ||z2 | + |z2 |2
= |z1 |2 + 2|z1 ||z2 | + |z2 |2
= (|z1 | + |z2 |)2 .
Hence |z1 + z2 | ≤ |z1 | + |z2 |. 
Corollary 2.6. If z1 and z2 are complex numbers, then ||z1 | − |z2 || ≤ |z1 − z2 |.
Proof. Using triangle inequality, we have |z1 | = |z1 − z2 + z2 | ≤ |z1 − z2 | + |z2 |. Therefore
DA
|z1 | − |z2 | ≤ |z1 − z2 |. Changing roles of z1 and z2 , we get |z2 | − |z1 | ≤ |z2 − z1 | = |z1 − z2 |, i.e.,
−(|z1 | − |z2 |) ≤ |z1 − z2 |. Hence ||z1 | − |z2 || ≤ |z1 − z2 |. 
Example 2.7.
√
(1) If z ∈ C, then |Re z| + |Im z| ≤ 2|z|.
Let z = x + iy. Since (|x| − |y|)2√≥p 0, x2 + y 2 = |x|2 + |y|2 ≥ 2|x||y|.
√ Therefore 2x2 + 2y 2 =
|x|2 + |y|2 + 2|x||y|. This gives 2 x2 + y 2 ≥ (|x| + |y|)2 , i.e., 2|z| ≥ |Re z| + |Im z|.
(2) Find the locus of |z − 4i| + |z + 4i| = 10. Thep above equation can be written as |z − 4i|2 =
2 2 2
(10−|zp+4i|) . Hence x +(y −4) = 100−20 x2 + (y + 4)2 +x2 +(y +4)2 . Simplification
gives 5 x2 + (y + 4)2 = 25 + 4y. Therefore 25x2 + 25y 2 + 200y + 400 = 625 + 200y + 16y 2
2 2
or x32 + y52 = 1. Hence the above equation represents an ellipse.
PA

(3) Characterize the set of all complex number z satisfying z 2 + z 2 = 2.

Let z = x + iy. Then z 2 + z 2 = 2 if and only if z 2 − y 2 + 2ixy + x2 − y 2 − 2ixy = 2 if and
only if x2 − y 2 = 1. Hence the given set represents hyperbola.
(4) What is the locus of a point z satisfying |z − 1| = |z + i|?
Squaring both the sides, (x − 1)2 + y 2 = x2 + (y + 1)2 , i.e., x + y = 0. Hence the set of complex
numbers z = x + iy satisfying |z − 1| = |z + i| is a line passing through origin have slope −1.
(5) If |z| < 2, then |z 3 + 3z 2 − 2z + 1| ≤ |z 3 | + |3z 2 | + |2z| + 1 = |z|3 + 3|z|2 + 2|z| + 1 ≤
8 + 12 + 4 + 1 = 25.
z+1 1
(6) If |z| = 2, then show that z4 −4z 2 +3 ≤ 3 .

We notice that z 4 − 4z 2 + 3 = (z 2 − 3)(z 2 − 1). Here |z| = 2. Then |z 2 − 1| ≥ ||z 2 | − 1| =

z+1 |z|+1 3
||z|2 − 1| = 3. Similarly, |z 2 − 3| ≥ ||z|2 − 3| = 1. Hence z4 −4z 2 +3 ≤ |z 2 −1||z 2 −3| ≤ 3 = 1.
3. Argument and Principal Argument 7

(7) A complex number z is real if and only if z = z.

We see that z = z if and only if z − z = 0 if and only if 2i Im z = 0 if and only if Im z = 0 if
and only z is a real number.
(8) A complex number z is real or purely imaginary if and only if z 2 = z 2 .
Let z = x + iy. Then z 2 = z 2 if and only if x2 − y 2 + 2ixy = x2 − y 2 − 2ixy if and only if

I
xy = 0 if and only if x = 0 or y = 0 if and only if z is real or purely imaginary.
(9) Let 0 6= z ∈ C. Then z = z −1 if and only if |z| = 1.

BH
Note that z = z −1 if and only if z = z1 if and only if zz = 1 if and only if |z|2 = 1 if and only if
|z| = 1.
(10) Sketch the points determined by (a) Re(z − i) = 2; (b) |2z − i| = 4.

3. Argument and Principal Argument

Let r and θ be polar coordinates of the point (x, y) that corresponds to a nonzero complex
number z = x + iy. Since x = r cos θ and y = r sin θ, the number z can be written in polar form
as z = r(cos θ + i sin θ). If z = 0, the coordinate θ is undefined; and so it is always understood that
z 6= 0 whenever it is discussed.
√
DA
We observe that |z| = r2 cos2 θ + r2 sin2 θ = r ≥ 0. Hence r is a nonnegative real number
and it is the length of z. The real number θ represents the angle, measured in radians, that z makes
with the positive real axis when z is interpreted as a radius vector. The number θ has an infinite
number of possible values, including negative ones, that differ by integral multiples of 2π. Those
values can be determined from the equation tan−1 θ = y/x, where the quadrant containing the point
corresponding to z must be specified. Each value of θ is called an argument of z, and the set of all
such values is denoted by arg z. The principal value or principal argument of arg z, denoted by
Arg z, is that unique value Θ such that −π < Θ ≤ π. Notice that

arg z = Arg z + 2nπ (n ∈ Z).

PA

If z is a negative real number Arg z = π.

We observe that θ is a value in arg z if and only if arg z = {θ + 2nπ : n ∈ Z}.

Example 3.1. Let z = 1 + i = (1, 1). Then z is lying in the first quadrant and z makes an angle π4
√ √
with the positive real axis. Since −π < π4 ≤ π, Arg (1 + i) = π4 . Also, |z| = 1 + 1 = 2. Hence
√
arg z = π4 + 2nπ, n ∈ Z. Thus 1 + i = 2(cos( π4 + 2nπ) + sin( π4 + 2nπ)), n ∈ Z.

Lemma 3.2. If z1 and z2 are nonzero complex numbers, then arg(z1 z2 ) = arg z1 + arg z2 .

Proof. Let z1 = r1 (cos θ1 + i sin θ1 ) and z2 = r2 (cos θ2 + i sin θ2 ), where θ1 and θ2 are values in
arg z1 and arg z2 respectively. Then arg z1 = {θ1 + 2nπ : n ∈ Z} and arg z2 = {θ2 + 2mπ : m ∈
Z}. Since z1 z2 = r1 r2 (cos θ1 + i sin θ1 )(cos θ2 + i sin θ2 ) = r1 r2 (cos(θ1 + θ2 ) + i sin(θ1 + θ2 )),
8 CONTENTS

θ1 + θ2 is a value in arg(z1 z2 ). Therefore

arg(z1 z2 ) = {θ1 + θ2 + 2nπ : n ∈ Z}
= {θ1 + θ2 + 2(n + m)π : n, m ∈ Z} = {θ1 + 2nπ + θ2 + 2mπ : n, m ∈ Z}
= {θ1 + 2nπ : n ∈ Z} + {θ2 + 2mπ : m ∈ Z}

I
= arg z1 + arg z2 .
Hence the result. 

BH
Notice that Arg(z1 z2 ) need not be same as Arg z1 + Arg z2 .
π
Take z1 = −1 and z2 = i. Then Arg z1 = π and Arg z2 = 2
. Now z1 z2 = −i and
Arg(z1 z2 ) = − π2 . Hence Arg(z1 z2 ) 6= Arg z1 + Arg z2 .
Lemma 3.3. Let 0 6= z ∈ C. Then arg(z −1 ) = arg(z) = − arg(z).
Proof. Suppose that z = r(cos θ + i sin θ), where θ is a value in arg z. Then
1 1 1 1 1
z −1 = = = (cos θ − i sin θ) = (cos(−θ) + i sin(−θ)).
z r cos θ + i sin θ r r
DA
It means that −θ is a value in arg z. Therefore
arg(z −1 ) = {−θ + 2nπ : n ∈ Z} = {−θ − 2nπ : n ∈ Z} = −{θ + 2nπ : n ∈ Z} = − arg z.

 
z1
Corollary 3.4. If z1 , z2 are nonzero complex numbers, then arg z2
= arg z1 − arg z2 .

Proof. We shall use the last two results. We see that

 
z1
arg = arg(z1 z2−1 ) = arg z1 + arg z2−1 = arg z1 − arg z2 .
z2

PA

Example 3.5. If θ1 , θ2 , . . . , θn are real numbers, then we show that

n n
! n
!
Y X X
(cos θj + i sin θj ) = cos θj + i sin θj .
j=1 j=1 j=1

We shall prove it using induction on n. Let θ1 , θ2 ∈ R. Then

(cos θ1 + i sin θ1 )(cos θ2 + i sin θ2 ) = cos θ1 cos θ2 − sin θ1 sin θ2 + i(sin θ1 cos θ2 + sin θ2 cos θ1 )
= cos(θ1 + θ2 ) + i sin(θ1 + θ2 ).
Qn−1 P  P 
n−1 n−1
Assume that the result is true for n−1, i.e., j=1 (cos θj +i sin θj ) = cos θ
j=1 j +i sin θ
j=1 j
for all θ1 , θ2 , . . . , θn−1 .
3. Argument and Principal Argument 9

Let θ1 , θ2 , . . . , θn ∈ R. Then
n
Y n−1
Y
(cos θj + i sin θj ) = (cos θj + i sin θj )(cos θn + i sin θn )
j=1 j=1
n−1
! n−1
!

I
X X
= cos θj + i sin θj (cos θn + i sin θn ) [∵ Induction hypothesis]
j=1 j=1

BH
n
! n
!
X X
= cos θj + i sin θj . [∵ the result is true for n = 2]
j=1 j=1

Hence the result.

If θ is a real number, measured in radians, then by Euler’s formula
exp(iθ) = eiθ = cos θ + i sin θ.
Thus a nonzero complex number z = r(cos θ + i sin θ) can be written as z = reiθ . The above
is called an exponential representation of z. Since both cos and sin are 2π- periodic functions,
DA
z = rei(θ+2nπ) , n ∈ Z, are all exponential representations of z.
π
If a point is lying on a unit circle, i.e., if |z| = 1, then z = eiθ . It is obvious that ei0 = 1, ei 2 = i,
3π
eiπ = 1 and ei 2 = √ −i. √
Since 1 + i = 2(cos( π4 + 2nπ) + sin( π4 + 2nπ)), n ∈ Z, 1 + i = 2 exp(i π4 ).
Lemma 3.6.
(1) If θ is a real number, then (eiθ )−1 = eiθ = e−iθ .
(2) |eiθ | = 1 for all θ ∈ R.
(3) If θ1 , θ2 , . . . , θn are real numbers, then eiθ1 eiθ2 · · · eiθn = ei(θ1 +θ2 +···+θn ) .
iθ
(4) If θ1 and θ2 are real numbers, then eeiθ12 = ei(θ1 −θ2 ) .
Proof. (i) By definition
PA

1 1
(eiθ )−1 = iθ = = cos θ − i sin θ = cos(−θ) − i sin(−θ).
e cos θ + i sin θ
Therefore (eiθ )−1 = eiθ = e−iθ .
(ii) We see that |eiθ | = | cos θ + i sin θ| = 1.
(iii) If θ1 , θ2 , . . . , θn are real numbers, then
n
Y
iθ1 iθ2 iθn
e e ···e = (cos θj + i sin θj ) = cos(θ1 + · · · + θn ) + i sin(θ1 + · · · + θn )
j=1

= ei(θ1 +θ2 +···+θn ) .

eiθ1
(iv) We note that eiθ2
= eiθ1 (eiθ2 )−1 = eiθ1 e−iθ2 = ei(θ1 −θ2 ) . 
10 CONTENTS

If z1 = r1 eiθ1 and z2 = r2 eiθ2 are exponential representations of z1 and z2 respectively , then

z1 z2 = r1 r2 ei(θ1 +θ2 ) is an exponential representation of z1 z2 .
Note that two complex numbers z1 = r1 eiθ1 and z2 = r2 eiθ2 are equal if and only if r1 = r2
and θ1 − θ2 is an integral multiple of 2π.

Lemma 3.7. If 0 6= z = reiθ , then z n = rn einθ for all n ∈ Z.

I
Proof. We first show that z n = rn einθ for all n ∈ N by using Principle of Mathematical Induc-

BH
tion. Clearly, z 2 = (reiθ )(reiθ ) = r2 ei2θ . Suppose that it is true for n − 1. Now z n = z n−1 z =
(rn−1 ei(n−1)θ )(reiθ ) = rn einθ .
Now z −2 = z12 = r2 e1i2θ = r−2 e−i2θ . Using induction it follows that z −n = r−n e−inθ for all
n ∈ N. Hence z n = rn einθ for all n ∈ Z. 

4. n-th root of a complex number

Definition 4.1. A complex number z is called an n-th root of a complex number w if z n = w.

Example 4.2. Given w ∈ C and n ∈ N, we shall find all complex numbers z such that z n = w.
DA
If w = 0, then obviously z = 0 is the only solution of z n = w.
Let w 6= 0, and let w = r0 eiθ0 . Let z = reiθ be a solution of z n = w. Then rn einθ = r0 eiθ0 .
1
Therefore rn = r0 and inθ = θ0 + 2kπ, k ∈ Z, i.e., r = r0n and θ0 = θ0 +2kπ
n
, k ∈ Z. But the above is
1
θ0 +2kπ
same as r = r0n and θ0 = n
, k = 0, 1, . . . , n − 1. Hence the set of all z ∈ C such that z n = w
1 θ0 +2kπ
is {r0n e n : k = 0, 1, . . . , n − 1}.

We note that if r0 eiθ0 = w 6= 0, then n- th roots of w are lying on the circle with center 0 and
1/n
radius |w|1/n = r0 and they are equally spaced by 2π n
radians.

Example 4.3. Compute n-th roots of unity.

i2π i4π i2(n−1)π
We note that 1 = ei0 . Hence all complex numbers z satisfying z n = 1 are 1 = ei0 , e n , e n , . . . , e n
PA

5. Some Topological Aspects

Definition 5.1. Let  > 0. Then by an -neighborhood of z0 ∈ C we mean the set of all complex
number z lying inside a circle of radius  and center z0 . By a neighborhood of z0 ∈ C we mean an
-neighborhood of z0 for some  > 0.

We shall denote -neighborhood of z0 by N (z0 , ). Thus

N (z0 , ) = {z ∈ C : |z − z0 | < }.
A deleted neighborhood of z0 ∈ C is a neighborhood of z0 except the point z0 .
6. Functions and Their Limits 11

Definition 5.2. An open set S is called connected if each pair of points z1 and z2 in it can be joined
by a polygonal line, consisting of a finite number of line segments joined end to end, that lies entirely
in S.
A nonempty open set that is connected is called a domain.

I
Note that any neighborhood is a domain. A domain together with some, none, or all of its
boundary points is referred to as a region.
A set S is called bounded if every point of S lies inside some circle |z| = R; otherwise, it is

BH
unbounded.

6. Functions and Their Limits

Definition 6.1. Let S be a nonempty set of complex numbers. A function f defined on S is a rule
that assigns to each z in S a complex number w. The number w is called the value of f at z and is
denoted by f (z); that is, w = f (z). The set S is called the domain of definition of f .

Let S be a nonempty set, and let f : S → C be a function. If z = x + iy, then f (z) = f (x + iy)
DA
is a complex number, say, u(x, y) + iv(x, y). Then u, v : S → R are functions. The function u and
v are called real part and imaginary part of f respectively (in rectangular coordinates).
If z = reiθ is written in polar coordinates, then there are real valued functions u and v such
that f (z) = f (reiθ ) = u(r, θ) + iv(r, θ) for all z ∈ S. The functions u and v are called real part
and imaginary part of f (in polar coordinates). Sometimes we may write u = Re f and v = Im f .

Examples 6.2. (1) Let f : C → C be f (z) = z 2 .

If z = x+iy ∈ C, then f (z) = f (x+iy) = (x+iy)2 = x2 −y 2 +i2xy. Thus u(x, y) = x2 −y 2
is the real part of f and v(x, y) = 2xy is the imaginary part of f .
If z = reiθ , then f (z) = f (reiθ ) = (reiθ )2 = r2 ei2θ = r2 (cos 2θ + i sin 2θ). Thus u(r, θ) =
r cos 2θ is the real part of f and v(r, θ) = r2 sin 2θ is the imaginary part of f .
2

(2) Let f : C → C be f (z) = |z|2 .

PA

If z = x + iy ∈ C, then f (z) = f (x + iy) = |x + iy|2 = x2 + y 2 + i0. Therefore

u(x, y) = x2 + y 2 and v(x, y) = 0.
If z = reiθ , then f (z) = f (reiθ ) = |reiθ |2 = r2 + i0. This gives u(r, θ) = r2 and v(r, θ) = 0.
(3) Let f : C \ {0} → C be f (z) = z1 .
1 x y x
If z = x + iy, then f (z) = f (x + iy) = x+iy = x2 +y 2 − i x2 +y 2 . It means that u(x, y) = x2 +y 2
y
and v(x, y) = − x2 +y 2.

If z = reiθ , then f (z) = f (reiθ ) = re1iθ = 1r (cos θ − i sin θ). Thus u(r, θ) = 1r cos θ and
v(r, θ) = − 1r sin θ.

Definition 6.3. Let a function f be defined at all points z in some deleted neighborhood of z0 . The
statement that the limit of f (z) as z approaches z0 is a number w0 , or that limz→z0 f (z) = w0 , means
12 CONTENTS

that the point w = f (z) can be made arbitrarily close to w0 if we choose the point z close enough to
z0 but distinct from it. We now express the definition of limit in a precise and usable form.
The above means that for each positive number  > 0, there is a positive number δ such that
|f (z) − w0 | <  whenever 0 < |z − z0 | < δ.
Lemma 6.4. Let f be defined in a deleted neighborhood of z0 . If limz→z0 f (z) exists, then it is

I
unique.

BH
Proof. Suppose that w0 and w1 are limits of f at z0 . Take any  > 0. Then there are positive δ1
and δ2 such that |f (z) − w0 | < /2 whenever 0 < |z − z0 | < δ1 and |f (z) − w1 | < /2 whenever
0 < |z − z0 | < δ2 . Take δ = min{δ1 , δ2 }. Then both above hold for δ. Take any z ∈ C with
0 < |z − z0 | < δ. Then |w0 − w1 | ≤ |w0 − f (z)| + |f (z) − w1 | < /2 + /2 = . Since  > 0 is
arbitrary, it follows that w0 = w1 . 
A function f is said to be bounded on the set S ⊂ C if the set f (S) is a bounded subset of C.
Lemma 6.5. Suppose that lim f (z) = w0 ∈ C. Then f is bounded in a neighborhood of z0 .
z→z0
DA
Proof. Since limz→z0 f (z) = w0 , there is δ > 0 such that |f (z) − w0 | < 1 for all 0 < |z − z0 | < δ.
If z ∈ C with 0 < |z − z0 | < δ, then |f (z)| ≤ |f (z) − w0 | + |w0 | < 1 + |w0 |. Hence f is bounded
on a deleted neighborhood 0 < |z − z0 | < δ of z0 . 
Lemma 6.6. Suppose that limz→z0 f (z) = w0 ∈ C. Then for given  > 0 there is δ > 0 such that
|f (z1 ) − f (z2 )| <  whenever 0 < |z1 − z0 | < δ and 0 < |z2 − z0 | < δ.
Proof. For given  > 0 there is δ > 0 such that |f (z) − w0 | < /2 whenever 0 < |z − z0 | < δ.
Let z1 , z2 ∈ C be such that 0 < |z1 − z0 | < δ and 0 < |z2 − z0 | < δ. Then |f (z1 ) − f (z2 )| ≤
|f (z1 ) − w0 | + |w0 − f (z2 )| < /2 + /2 = . 
It follows from Lemma 6.6 that if there is  > 0 such that for every δ > 0 there are z1 , z2 ∈ C
with 0 < |z1 − z0 | < δ and 0 < |z2 − z0 | < δ but |f (z1 ) − f (z2 )| ≥ , then limz→z0 f (z) does not
PA

exist.
Example 6.7.
(1) Let f : C \ {0} → C be f (z) = zz . We show that limit of f at 0 does not exist by using Lemma
6.6.
Take  = 2. Let δ > 0. Then 0 < | 2δ − 0| < δ and 0 < | iδ2 − 0| < δ. But |f ( 2δ ) − f ( iδ2 )| =
|1 − ii | = |1 − i2 | = 2 = . Hence limz→0 f (z) does not exist.
(2) Now we show that limz→2i (2x + iy 2 ) = 4i.
Take any  > 0. We want to find δ > 0 such that if 0 < |z − 2i| < δ, then |2x + iy 2 + 4i| < .
We may assume that δ ≤ 1. Let z = x + iy ∈ C be such that 0 < |x + iy − 2i| < δ. Then
|x| < δ and |y − 2| < δ. Here δ is to be determined. Now |2x + iy 2 − 4i| ≤ |2x| + |y 2 − 4| =
6. Functions and Their Limits 13

|2x| + |y − 2||y + 2|. Since |y + 2| < δ < 1, |y − 2| = |y + 2 − 4| ≤ |y + 2| + |4| ≤ 5. Thus

|2x + iy 2 − 4i| ≤ |2x| + 5|y − 2| < 2δ + 5δ = 7δ = . Thus take δ = min{ 7 , 1}.
(3) We prove that limz→1−i [x + i(2x + y)] = 1 + i.
Take any  > 0. Let z = x + iy ∈ C such that 0 < |x + iy − 1 + i| < δ, where δ is
to be determined. Then |x − 1| < δ and |y + 1| < δ. Now |x + i(2x + y) − 1 − i| =

I
|x − 1 + i(2x + y − 1)| ≤ |x − 1| + |2x − 2 + y + 1| ≤ 3|x − 1| + |y + 1| < 4δ. Take δ = 4 .
Lemma 6.8. If limz→z0 f (z) = w0 , w0 ∈ C and w0 6= 0, then there is c > 0 and δ > 0 such that

BH
|f (z)| ≥ c for all z satisfying 0 < |z − z0 | < δ.
Proof. Take  = |w20 | > 0. Then there is δ > 0 such that |f (z)−w0 | <  whenever 0 < |z −z0 | < δ.
Let 0 < |z − z0 | < δ. Then |w0 | − |f (z)| ≤ |f (z) − w0 | <  = |w20 | . This gives |f (z)| ≥ |w20 | . 
Lemma 6.9. Suppose that limz→z0 f (z) = w0 and limz→z0 F (z) = W0 , where w0 , W0 ∈ C. Then
(1) limz→z0 [f (z) + F (z)] = w0 + W0 OR limz→z0 [f (z) + F (z)] = limz→z0 f (z) + limz→z0 F (z).
(2) limz→z0 [f (z)F (z)] = w0 W0 OR limz→z0 [f (z)F (z)] = (limz→z0 f (z))(limz→z0 F (z)).
1
(3) If W0 6= 0, then limz→z0 F (z) 1
= W10 OR limz→z0 F (z) = limz→z1 F (z) .
0
limz→z0 f (z)
(4) If W0 6= 0, then limz→z0 Ff (z) w0
=W OR limz→z0 f (z)
= .
(5)
DA (z) 0
limz→z0 αf (z) = α limz→z0 f (z) for all α ∈ C.
F (z) limz→z0 F (z)

(6) limz→z0 |f (z)| = |w0 |.

(7) limz→z0 f (z) = w0 .
Proof. (i) is easy and left to the reader.
(ii) Since limz→z0 F (z) exists, there are positive constants M and δ1 such that |F (z)| ≤ M for all z

satisfying 0 < |z − z0 | < δ1 . Take any  > 0. Then there are δ2 , δ3 > 0 such that |f (z) − w0 | < 2M

whenever 0 < |z − z0 | < δ2 and |F (z) − W0 | < 2(1+|w0 |) whenever 0 < |z − z0 | < δ3 . Take
δ = min{δ1 , δ2 , δ2 }. Let 0 < |z − z0 | < δ. Then |f (z)F (z) − w0 W0 | = |f (z)F (z) − w0 F (z) +
 
w0 F (z) − w0 W0 | ≤ |F (z)||f (z) − w0 | + |w0 ||F (z) − W0 | < M 2M + |w0 | 1+|w0|
≤ . Thus
limz→z0 [f (z)F (z)] = w0 W0 .
PA

(iii) There are positive constants c and δ1 such that |F (z)| ≥ c for all z satisfying 0 < |z − z0 | < δ1 .
Take any  > 0. Then there is δ2 > 0 such that |F (z) − W0 | < c|W0 | whenever 0 < |z − z0 | < δ2 .
1
Take δ = min{δ1 , δ2 }. Let z ∈ C with 0 < |z − z0 | < δ. Then F (z) 1
− W10 = |W0 ||F (z)|
|F (z) −
W0 | < c|W1 0 | c|W0 | = .
(iv) Use (ii) and (iii).
(v) and (vii) are easy.
(vi) If  > 0, then there is δ > 0 such that |f (z) − w0 | <  whenever 0 < |z − z0 | < δ. If
0 < |z − z0 | < δ, then ||f (z)| − |w0 || ≤ |f (z) − w0 | < . Hence limz→z0 |f (z)| = |w0 |. 
Theorem 6.10. Suppose that f (z) = u(x, y) + iv(x, y), (z = x + iy) and z0 = x0 + iy0 , w0 =
u0 +iv0 . Then lim f (z) = w0 if and only if lim u(x, y) = u0 and lim v(x, y) = v0 .
z→z0 (x,y)→(x0 ,y0 ) (x,y)→(x0 ,y0 )
14 CONTENTS

Proof. Suppose that limz→z0 f (z) = w0 . Then for given  > 0 there is δp> 0 such that
p
(u(x, y) − u0 )2 + (v(x,py) − v0 )2 = |f (z) − w0 | <  whenever 0 < (x − x0 )2 + (y − y0 )2 =
|z − z0 | < δ. Thus if 0 < (x − x0 )2 + (y − y0 )2 < δ, then |u(x, y) − u0 | < |f (z) − w0 | <  and
|v(x, y) − v0 | < |f (z) − w0 | < . So, lim u(x, y) = u0 and lim v(x, y) = v0 .
(x,y)→(x0 ,y0 ) (x,y)→(x0 ,y0 )
Conversely, assume that lim u(x, y) = u0 andv(x, y) = v0 . Take any  >
lim

I
(x,y)→(x0 ,y0 ) (x,y)→(x0 ,y0 )
p
0. Then there are δ1 , δ2 > 0 such that |u(x, y) − u0 | < 2 whenever 0 < (x − x0 )2 + (y − y0 )2 <

BH
p
δ1 and |v(x, y) − v0 | < 2 whenever 0 < (x − x0 )2 + (y − y0 )2 < δ2 . Take δ = min{δ1 , δ2 }. If
p p
0q< |z−z0 | = (x − x0 )2 + (y − y0 )2 < δ, then |f (z)−w0 | = (u(x, y) − u0 )2 + (v(x, y) − v0 )2 <
2 2 √
4
+ 4
= 2
< . 
It follows from above results that limz→z0 c = c, where c is a constant, limz→z0 Re z = Re z0 ,
limz→z0 Im z = Im z0 , limz→z0 |z| = |z0 |, limz→z0 z = z0 and limz→z0 z n = z0n for all n ∈ N.
If a0 , a1 , . . . , an are in C and an 6= 0, then p(z) = a0 + a1 z + · · · + an z n is called a polynomial
of degree n.
Let z0 ∈ C. Then
lim p(z) =
z→z0
DA
lim (a0 + a1 z + · · · + an z n )
z→z0

= lim a0 + lim a1 z + lim a2 z 2 + · · · + lim an z n

z→z0 z→z0 z→z0 z→z0

= 2
lim a0 + a1 lim z + a2 lim z + · · · + an lim z n
z→z0 z→z0 z→z0 z→z0

= 2
lim a0 + a1 lim z + a2 ( lim z) + · · · + an ( lim z)n
z→z0 z→z0 z→z0 z→z0

= a0 + a1 z0 + a2 z02 + · · · + an z0n
= p(z0 ).
Let R > 0. Then an open set of the form {z ∈ C : |z| > R} is called a deleted neighborhood
of ∞. In fact, if K is a compact set, then a set C \ K is referred as a neighborhood of ∞.
PA

Definition 6.11.
(1) Let f be defined on {z ∈ C : |z| > R} for some R > 0. Let w0 ∈ C. We say that f (z)
approaches to w0 as z approaches to ∞ if for given  > 0 there is δ > 0 such that |f (z)−w0 | < 
whenever | z1 | < δ. In this case we write limz→∞ f (z) = w0 .
(2) Let f be defined on {z ∈ C : |z| > R} for some R > 0. We say that limz→∞ f (z) = ∞ if for
1
given  > 0 there is δ > 0 such that | f (z) | <  whenever | z1 | < δ.
(3) Let f be defined in a deleted neighborhood of z0 . We say that limz→z0 f (z) = ∞, if for given
1
 > 0 there is δ > 0 such that | f (z) | <  whenever 0 < |z − z0 | < δ.
Exercise 6.12.
(1) Show that limz→∞ f (z) = w0 if and only if limz→0 f ( z1 ) = w0 .
6. Functions and Their Limits 15

1
(2) Show that limz→∞ f (z) = ∞ if and only if limz→0 f ( z1 )
= 0.
1
(3) Show that limz→z0 f (z) = ∞ if and only if limz→z0 f (z) = 0.
(4) Evaluate the following limits.

(a) limz→−1 iz+3

(b) limz→∞ 2z+i 2z 3 −1
z+1 z+1 (c) limz→∞ z 2 +1

I
(5) Let p(z) = a0 + a1 z + · · · + an z n and q(z) = b0 + b1 z + · · · + bm z m , where ai , bj ∈ C, an 6= 0

BH
and bm 6= 0. Prove the following.
(a) limz→∞ p(z)
q(z)
= ∞ if n > m.
p(z) an
(b) limz→∞ q(z)
= bn
if n = m.
(c) limz→∞ p(z)
q(z)
= 0 if n < m.
Definition 6.13. A function f is said to be continuous at z0 if limz→z0 f (z) exists, f (z0 ) exists and
limz→z0 f (z) = f (z0 ).
Note that if f : S → C is a function and z0 ∈ C, then f is continuous at z0 if for given  > 0
there is δ > 0 such that |f (z) − f (z0 )| <  whenever z ∈ S and |z − z0 | < δ.
DA
A function of a complex variable is said to be continuous in a region R if it is continuous at each
point in R.
Exercises 6.14. Let f and g be continuous at z0 . Prove the following.
(1) f ± g is continuous at z0 .
(2) αf is continuous at z0 for all α ∈ C.
(3) fg is continuous at z0 if g(z0 ) 6= 0.
Theorem 6.15. Composition of continuous functions is a continuous function.
Proof. If f is continuous at z0 and g is continuous at f (z0 ), then we need to show that g ◦ f is
continuous at z0 . Take any  > 0. Then there is δ > 0 such that |g(w) − g(f (z0 ))| <  whenever
PA

|w − f (z0 )| < δ. Since f is continuous, there is γ > 0 such that |f (z) − f (z0 )| < δ whenever
|z − z0 | < γ. Let z ∈ C be such that |z − z0 | < γ. Then |f (z) − f (z0 )| < δ. But then |g(f (z)) −
g(f (z0 ))| < . Hence g ◦ f is continuous at z0 . 
Theorem 6.16. If a function f is continuous and nonzero at a point z0 , then f (z) 6= 0 throughout
some neighborhood of that point.
Proof. Here f (z0 ) 6= 0. So,  = |f (z20 )| > 0. By continuity of f at z0 , there is δ > 0 such that
|f (z) − f (z0 )| <  whenever |z − z0 | < δ. Let z ∈ C with |z − z0 | < δ. Then |f (z0 )| − |f (z)| ≤
|f (z0 ) − f (z)| <  = |f (z20 )| . This gives |f (z)| ≥ |f (z20 )| > 0 for all z with |z − z0 | < δ. 
Exercise 6.17. Let f = u+iv be a complex function, and let z0 = x0 +iy0 . Show that f is continuous
at z0 if and only if both u and v are continuous at (x0 , y0 ).
16 CONTENTS

Theorem 6.18. If a function f is continuous throughout a region R that is both closed and bounded,
then there exists a nonnegative real number M such that |f (z)| ≤ M for all points z in R, where
equality holds for at least one such z.

Proof. Let f = u + iv. Then both u and√v are continuous on R. Since R is closed and bounded
subset of C, it is compact. We see that u2 + v 2 is a continuous function on a compact set R.

I
Hence it is bounded and it attains the
p p bound, i.e., there is z0 = x0 + iy0 ∈ R such that |f (z0 )| =
2 2 2 2
pu(x0 , y0 ) + v(x0 , y0 ) = sup{ u(x, y) + v(x, y) : z = x + iy ∈ R}. Let us take M =

BH
u(x0 , y0 )2 + v(x0 , y0 )2 . Then M ≥ 0 and
p p
M = |f (z0 )| = u(x0 , y0 )2 + v(x0 , y0 )2 ≥ u(x, y)2 + v(x, y)2 (z ∈ R).
This completes the proof. 
We note that
(1) f (z) = z is continuous on C.
(2) f (z) = z is continuous on C.
(3) f (z) = |z| is continuous on C.
(4)
DA
f (z) = z1 is continuous on C \ {0}.
(5) Polynomials are continuous on C.
(6) If f is continuous, then both Re f and Im f are continuous.

7. Differentiability of Complex Functions

Definition 7.1. Let f be a function whose domain of definition contains a neighborhood |z − z0 | < δ
of a point z0 . The derivative of f at z0 is the limit
f (z) − f (z0 )
f 0 (z0 ) = lim ,
z − z0
and the function f is said to be differentiable functiondifferentiable at z0 when f 0 (z0 ) exists. The
PA

number f 0 (z0 ) is called the derivative of f at z0 .

Let ∆z = z−z0 (z 6= z0 ). If f is differentiable at z0 , then limz→z0 f (z)−f

z−z0
(z0 )
= lim∆z→0 f (z0 +∆z)−f
∆z
(z0 )
.
0 f (z+∆z)−f (z)
Thus if f is differentiable at z, then f (z) = lim∆z→0 ∆z
. Let ∆w = f (z+∆z)−f (z),
dw 0 dw ∆w
and let dz = f (z). Then dz = lim∆z→0 ∆z .

Examples 7.2. We shall check the differentiability of the following functions.

(1) Let f (z) = z 2 , z ∈ C.
Let z ∈ C. Then
∆w f (z + ∆z) − f (z)
lim = lim
∆z→0 ∆z ∆z→0 ∆z
7. Differentiability of Complex Functions 17

(z + ∆z)2 − z 2 z 2 + 2z∆z + ∆z 2 − z 2
= lim = lim
∆z→0 ∆z ∆z→0 ∆z
= 2z.
Hence f is differentiable at every z ∈ C and f 0 (z) = 2z for all z ∈ C.
(2) Let f (z) = z for all z ∈ C.

I
The similar process above shows that f is differentiable at every z ∈ C and f 0 (z) = 1 for all
z ∈ C.

BH
(3) Let f (z) = z, z ∈ C.
We observe that z+∆z−z∆z
= ∆z
∆z
. We know that lim∆z→0 ∆z∆z
does not exist. Thus the function f
is nowhere differentiable.
(4) Let f (z) = |z|2 , z ∈ C.
2 −|z|2
Here |z+∆z|
∆z
= z + ∆z + z ∆z
∆z
. Thus lim∆z→0 ∆w
∆z
exists if and only if z = 0 and f 0 (0) = 0.
Lemma 7.3. Let f be defined in a neighborhood of z0 . If f is differentiable at z0 , then f is contin-
uous at z0 .
Proof. We need to prove that limz→z0 f (z) = f (z0 ). We see that
DA
f (z) − f (z0 ) f (z) − f (z0 )
lim [f (z) − f (z0 )] = lim (z − z0 ) = lim lim (z − z0 ) = 0.
z→z0 z − z0
z→z0 z→z 0 z − z0 z→z0

Therefore limz→z0 f (z) = f (z0 ) and hence f is continuous at z0 . 

The converse of Lemma 7.3 is not true. For example the function f (z) = z is continuous at
every z ∈ C but it is nowhere differentiable.
df
We shall use the symbol dz (z) to denote the derivative of f at z.
Theorem 7.4. Let f and g be complex functions. Suppose that f and g are differentiable at z.
Then
df
d
(1) dz (f + g)(z) = dz (z) + dgdz
(z).
d df
(2) dz (cf )(z) = c dz (z).
PA

d
(3) dz (f g)(z) = f (z) dg (z) + g(z) dz
dz  
df
(z).
df
d f g(z) (z)−f (z) dg (z)
(4) If g(z) 6= 0, then dz g
(z) = dz
g(z)2
dz
.

Theorem 7.5. If f is differentiable at z and g is differentiable f (z), then g ◦ f is differentiable at

z and (g ◦ f )0 (z) = g 0 (f (z))f 0 (z).
Proof. Here f is differentiable at z0 and g is differentiable at f (z0 ) = w0 . In a neighborhood of
w0 , say, |w − w0 | < , define Φ(w) = g(w)−g(w w−w0
0)
− g 0 (w0 ) if w 6= w0 and Φ(w0 ) = 0. Then
Φ is continuous at w0 . We may write g(w) − g(w0 ) = [g 0 (w0 ) + Φ(w)](w − w0 ) for all |w −
w0 | < . Since f is differentiable at z0 , it is continuous at z0 . Therefore there is δ > 0 such that
|f (z) − w0 | <  whenever |z − z0 | < δ. Thus if 0 < |z − z0 | < δ, then g(f (z))−g(f z−z0
(z0 ))
=
18 CONTENTS

[g 0 (f (z0 )) + Φ(f (z))] f (z)−f

z−z0
(z0 )
. As f is continuous at z0 and Φ is continuous at w0 = f (z0 ), the
map Φ ◦ f is continuous at z0 . This gives limz→z0 Φ(f (z)) = Φ(f (z0 )). Taking limit z tends to
z0 , we obtain limz→z0 g(f (z))−g(f
z−z0
(z0 ))
= g 0 (f (z0 ))f 0 (z0 ). Hence g ◦ f is differentiable at z0 and
0 0 0
(g ◦ f ) (z0 ) = g (f (z0 ))f (z0 ). 

I
8. Cauchy-Riemann Equations

BH
Theorem 8.1 (Cauchy - Riemann Equations). Suppose that f = u + iv and that f is differentiable
at z0 = x0 + iy0 . Then the first-order partial derivatives of u and v must exist at (x0 , y0 ), and they
must satisfy the Cauchy-Riemann equations ux (x0 , y0 ) = vy (x0 , y0 ), uy (x0 , y0 ) = −vx (x0 , y0 ).
Also, f 0 (z0 ) = ux (x0 , y0 ) + ivx (x0 , y0 ).

Proof. Let ∆z = ∆x + i∆y. Then obviously ∆z → 0 if and only if (∆x, ∆y) → (0, 0). Since f is
differentiable, the limit lim∆z→0 f (z0 +∆z)−f
∆z
(z0 )
exists and it is f 0 (z0 ). Since this limit exists, it is same
as lim∆x→0 f (z0 +∆x)−f
∆x
(z0 )
and lim∆y→0 f (z0 +∆y)−f
∆y
(z0 )
. It means that
f (z0 + ∆z) − f (z0 ) f (z0 + ∆x) − f (z0 ) f (z0 + ∆y) − f (z0 )
lim = lim = lim .
∆z→0 ∆z
DA
∆x→0 ∆x ∆y→0 ∆y
f (z0 +∆z)−f (z0 )
Now we shall simplify z−z0
. Therefore
f (z0 + ∆x) − f (z0 )
f 0 (z0 ) = lim
∆x→0
 ∆x 
u(x0 + ∆x, y0 ) − u(x0 , y0 ) v(x0 + ∆x, y0 ) − v(x0 , y0 )
= lim +i
∆x→0 ∆x ∆x
u(x0 + ∆x, y0 ) − u(x0 , y0 ) v(x0 + ∆x, y0 ) − v(x0 , y0 )
= lim + i lim
∆x→0 ∆x ∆x→0 ∆x
Here we may separate the limit as the limits of real part and imaginary part exist. Thus the partial
derivatives of u and v with respect to the first variable at (x0 , y0 ) exist. Hence f 0 (z0 ) = ux (x0 , y0 ) +
PA

ivx (x0 , y0 ).
Similarly,
f (z0 + ∆y) − f (z0 )
f 0 (z0 ) = lim
∆y→0 ∆y
 
u(x0 , y0 + ∆y) − u(x0 , y0 ) v(x0 , y0 + ∆y) − v(x0 , y0 )
= lim +i
∆y→0 i∆y i∆i
 
u(x0 , y0 + ∆y) − u(x0 , y0 ) v(x0 , y0 + ∆y) − v(x0 , y0 )
= −i lim +
∆y→0 ∆y ∆y
u(x0 , y0 + ∆y) − u(x0 , y0 ) v(x0 , y0 + ∆y) − v(x0 , y0 )
= −i lim + lim
∆y→0 ∆y ∆y→0 ∆y
8. Cauchy-Riemann Equations 19

Again the limits of real part and imaginary part exist as f 0 (z0 ) exists. Therefore f 0 (z0 ) = −iuy (x0 , y0 )+
vy (x0 , y0 ). It follows that f 0 (z0 ) = ux (x0 , y0 ) + ivx (x0 , y0 ) = vy (x0 , y0 ) − iuy (x0 , y0 ). Hence
ux (x0 , y0 ) = vy (x0 , y0 ) and uy (x0 , y0 ) = −vx (x0 , y0 ). 
Theorem 8.1 says that if the Cauchy-Riemann equations are not satisfied at z0 = x0 + iy0 , then
f is not differentiable at z0 .

I
The equations ux = vy and uy = −vx are the Cauchy-Riemann equations, so named in honor
of the French mathematician A. L. Cauchy (1789-1857), who discovered and used them, and in honor

BH
of the German mathematician G. F. B. Riemann (1826-1866), who made them fundamental in his
development of the theory of functions of a complex variable.
Satisfaction of the Cauchy-Riemann equations at a point z0 = (x0 , y0 ) is not sufficient to ensure
the existence of the derivative of a function f at that point. But, with certain continuity conditions it
is true. We see it soon.
Examples 8.2.
(1) We know that the function f (x + iy) = f (z) = z 2 = x2 − y 2 + 2ixy is differentiable at every
z = x + iy ∈ C.
DA
Here u(x, y) = x2 − y 2 and v(x, y) = 2xy. We note that ux (x, y) = 2x = vy (x, y) and
uy (x, y) = −2y = −vx (x, y) and f 0 (z) = f 0 (x + iy) = ux (x, y) + ivx (x, y) = 2x + i2y = 2z
for all z = x + iy ∈ C.
2
(2) Now we shall show that the converse of Theorem 8.1 is not true. Let f : C → C be f (z) = zz
if z 6= 0 and f (0) = 0.
2 2 2 2 3 2 −2xy 2 2 3 −2x2 y
If z = x + iy 6= 0, then zz = (x−iy)
x+iy
= (x −y x−i2xy)(x−iy)
2 +y 2 = x −xyx2 +y 2
+ i −x y+y
x2 +y 2
.
x3 −xy 2 −2xy 2 −x2 y+y 3 −2x2 y
Therefore u(x, y) = x2 +y 2
if (x, y) 6= 0, u(0, 0) = 0, v(x, y) = x2 +y 2
and
v(0, 0) = 0. Now
u(δ, 0) − u(0, 0) δ3 u(0, δ) − u(0, 0) 0
lim = lim 3 = 1, lim = lim 2 = 0,
δ→0 δ δ→0 δ δ→0 δ δ→0 δ
PA

v(δ, 0) − v(0, 0) 0 v(0, δ) − v(0, 0) δ3

lim = lim 3 = 0 and lim = lim 3 = 1.
δ→0 δ δ→0 δ δ→0 δ δ→0 δ

Therefore ux (0, 0) = 1 = vy (0, 0) and uy (0, 0) = 0 = −vx (0, 0). It means that the Cauchy -
2
Riemann equations are satisfied at 0 = 0 + i0. Now if z 6= 0, then f (z)−f
z−0
(0)
= zz2 . We see that
2 2
δ δ+iδ (1−i)2
limδ→0 δ2
= 1 and limδ+iδ→0 (δ+iδ)2
= (1+i)2
. It means that the limits along these paths are
2
different, i.e., the limit limz→0 zz2 does not exist. Hence the function f is not differentiable at 0.
It follows from last example that Cauchy - Riemann equations are not sufficient for differentia-
bility of a function. The following gives a sufficient condition of differentiability of a complex function.
Theorem 8.3. Let the function f = u + iv be defined throughout some  neighborhood of a point
z0 = x0 + iy0 , and suppose that
20 CONTENTS

(a) the first-order partial derivatives of the functions u and v with respect to x and y exist every-
where in the neighborhood;
(b) those partial derivatives are continuous at (x0 , y0 ) and satisfy the Cauchy˛ Riemann equa-
tions ux = vy , uy = −vx at (x0 , y0 ).
Then f 0 (z0 ) exists and f 0 (z0 ) = ux (x0 , y0 ) + ivx (x0 , y0 ).

I
Proof. Let f be defined in a neighborhood |z−z0 | <  of z0 . Let ∆z = ∆x+i∆y and 0 < |∆z| < ,
let ∆w = f (z0 + ∆z) − f (z0 ). Thus ∆w = ∆u + i∆v, where ∆u = u(x0 + ∆x, y0 + ∆y) − u(x, y)

BH
and ∆v = v(x0 + ∆x, y0 + ∆y) − v(x, y). Since the the partial derivatives of u and v are continuous
at (x0 , y0 ),
∆u = ux (x0 , y0 )∆x + uy (x0 , y0 )∆y + 1 ∆x + 2 ∆y
and
∆v = vx (x0 , y0 )∆x + vy (x0 , y0 )∆y + 3 ∆x + 4 ∆y,
where 1 , 2 , 3 , 4 tend to 0 as (∆x, ∆y) → (0, 0). By using above expressions we have
∆w = ∆u = ux (x0 , y0 )∆x + uy (x0 , y0 )∆y + 1 ∆x + 2 ∆y
+i[∆v = vx (x0 , y0 )∆x + vy (x0 , y0 )∆y + 3 ∆x + 4 ∆y].
DA
Since Cauchy-Riemann equations are satisfied at (x0 , y0 ), by simplifying the above equation we have
∆w ∆x ∆y
= ux (x0 , y0 ) + ivx (x0 , y0 ) + (1 + i3 ) + (4 + i4 ) . (8.3.1)
∆z ∆z ∆z
Since |∆x| ≤ |∆z| and |∆y| ≤ |∆z|, |(1 +i3 ) ∆x ∆z
| ≤ |1 |+|3 | and |(2 +i4 ) ∆y
∆z
| ≤ |2 |+|4 |. Now
∆w
applying limit ∆z → 0 in equation 8.3.1, we get lim∆z→0 ∆z = ux (x0 , y0 ) + ivx (x0 , y0 ) (because
i → 0 as ∆z → 0). Hence f is differentiable at z0 and f 0 (z0 ) = ux (x0 , y0 ) + ivx (x0 , y0 ). 
Example 8.4. Let f (z) = exp(z) = ez , z ∈ C.
If z = x + iy, then f (z) = f (x + iy) = ex eiy = ex cos y + iex sin y. Therefore u(x, y) =
ex cos y and v(x, y) = ex sin y. Now ux (x, y) = ex cos y, uy (x, y) = −ex sin y, vx (x, y) = ex sin y
and vy (x, y) = ex cos y. We note that all the functions ux , uy , vx and vy are continuous, ux = vy and
PA

uy = −vx . Therefore f is differentiable at every z = x + iy ∈ C and f 0 (z) = ux (x, y) + ivx (x, y) =

ex cos y + iex sin y = ez = f (z).
8.1. Cauchy-Riemann Equations in Polar Coordinates. Suppose that f (reiθ ) = u(r, θ) + iv(r, θ)
and partial derivatives of u and v with respect to r and θ exist. Differentiating f (reiθ ) = u(r, θ) +
iv(r, θ) with respect to r and θ respectively, we get eiθ f 0 (reiθ ) = ur (r, θ)+ivr (r, θ) and rieiθ f 0 (reiθ ) =
uθ (r, θ)+ivθ (r, θ). Comparing real and imaginary parts of f 0 (reiθ ), we get rur = vθ and rvr = −uθ .
Also, f 0 (reiθ ) = e−iθ (ur (r, θ) + ivr (reiθ )). Thus if f is differentiable at z = reiθ , then rur = vθ and
rvr = −uθ at that point.
The equations rur = vθ and rvr = −uθ are called Cauchy-Riemann equations in polar coordi-
nates.
9. Analytic Functions 21

If the partial derivatives of u(r, θ) and v(r, θ) in a neighborhood of z0 = r0 eiθ0 exist, they
are continuous and if they satisfy Cauchy-Riemann equations at z0 , then f is differentiable at z0 and
f 0 (z0 ) = e−iθ0 (ur (r0 , θ0 ) + ivr (r0 eiθ0 )).
Examples 8.6.

I
(1) Let f (z) = z12 , z 6= 0.
If 0 6= z = reiθ , then f (reiθ ) = r12 e−i2θ = cosr22θ − i sinr22θ . Therefore u(r, θ) = cosr22θ and
v(r, θ) = sinr22θ . Now ur = −2 cosr32θ , uθ = −2 sinr22θ , vr = −2 sinr32θ and vθ = 2 cosr22θ . Clearly,

BH
rur = vθ and rvr = −uθ and the partial derivatives are continuous at every 0 6= reiθ . There-
fore f is differentiable at every nonzero point and f 0 (reiθ ) = eiθ (ur + ivr ) = eiθ (−2 cosr32θ −
2i sinr32θ ) = − z23 .
Let z = x + iy. Then x = z+z 2
and y = z−z2i
. Thus if we are given a complex function
f (x + iy) = u(x, y) + iv(x, y) it can be considered as a function of z and z. Suppose that the
functions u and v satisfy the Cauchy-Riemann equations. Since x = z+z 2
and y = z−z
2i
, we have
∂x 1 ∂x 1 ∂y 1 ∂y 1
∂z
= 2 = xz , ∂z = 2 = xz , ∂z = 2i = yz and ∂z = − 2i = yz . Now by chain rule
∂f
DA
∂z
= [ux xz + uy yz ] + i[vx xz + vy yz ]
1 i 1 i
= [ux + uy ] + i[vx + vy ]
2 2 2 2
1
= [(ux − vy ) + i(vx + uy )]
2
= 0.

9. Analytic Functions
Definition 9.1. A function f is said to be analytic at z0 if it is differentiable at each point in a neigh-
borhood of z0 .
A function f is said to be analytic on a domain D if it is analytic at every point of D.
PA

A function f is called entire function if it is analytic at every point of C.

Analytic functions are also called a holomorphic functions.
If D is domain, then by definition f is differentiable on D if and only if f is analytic on D.
By definition if f is analytic at z0 , then f is differentiable at z0 . The converse is not true. For
example the function f (z) = |z|2 is differentiable at 0 but it is not analytic at 0 as f is not differentiable
at any nonzero point.
Definition 9.2. If a function f fails to be analytic at a point z0 but is analytic at some point in every
neighborhood of z0 , then z0 is called a singular point, or singularity, of f .
Examples 9.3.
22 CONTENTS

(1) Let f (z) = z1 , z 6= 0. Then f is differentiable at every nonzero complex number. Thus f is
analytic on C \ {0}.
(2) The functions z 2 , ez are analytic at every point of C. Therefore they are entire functions.
(3) Let f (z) = cosh x cos y + i sinh x sin y.
Then u(x, y) = cosh x cos y and v(x, y) = sinh x sin y. The functions u and v are differen-

I
tiable. We note that ux (x, y) = sinh x cos y, uy (x, y) = − cosh x sin y, vx (x, y) = cosh x sin y
and vy (x, y) = sinh x cos y. The functions ux , uy , vx and vy are continuous on R2 and the

BH
Cauchy-Riemann equations are satisfied. Therefore f is differentiable at every z ∈ C, i.e., f is
an entire function.
Theorem 9.4. Let f and g be analytic at z0 . Then
(1) f ± g is analytic at z0 .
(2) cf is analytic at z0 .
(3) f g is analytic on z0 .
(4) fg is analytic at z0 if g(z0 ) 6= 0.

Proof. (ii) If f is analytic at z0 , then f is differentiable on N (z0 , δ) for some δ > 0. But then cf is
DA
differentiable on N (z0 , δ). Therefore cf is analytic at z0 .
(i) and (iii) If f and g are analytic, there f and g will be differentiable on N (z0 , δ1 ) and N (z0 , δ2 )
respectively for some δ1 , δ2 > 0. Then f g and f ± g are differentiable on N (z0 , δ), where δ =
min{δ1 , δ2 }. This implies analyticity of f g and f ± g at z0 .
(iv) Since g(z0 ) 6= 0 and g is continuous at z0 , there is δ1 > 0 such that g(z) 6= 0 for any z ∈ N (z0 , δ1 ).
Also, there is δ2 > 0 such that g is differentiable on N (z0 , δ2 ) as g is analytic at z0 . The analyticity of
f at z0 implies the differentiability of f on N (z0 , δ3 ) for some δ3 > 0. If δ = min{δ1 , δ2 , δ3 }, then fg
is differentiable on N (z, δ). Therefore fg is analytic at z0 . 
Corollary 9.5. Let f and g be analytic on a domain D. Then
(1) f ± g is analytic on D.
PA

(2) cf is analytic at on D for all c ∈ C.

(3) f g is analytic on on D.
(4) fg is analytic on D if g(z) 6= 0 for any z ∈ D.

Theorem 9.6. If f is analytic at z0 and g is analytic at f (z0 ), then g ◦ f is analytic at z0 .

Proof. Since g is analytic at f (z0 ), g is differentiable in a neighborhood of f (z0 ) say N (f (z0 ), )
for some  > 0. Since f is continuous, there is δ1 > 0 such that |f (z) − f (z0 )| <  whenever
|z − z0 | < δ1 . Since f is analytic at z0 , there is δ2 > 0 such that f is differentiable on N (z0 , δ2 ).
Take δ = min{δ1 , δ2 }. If z ∈ N (z0 , δ), then f is differentiable at z and since f (z) ∈ N (f (z0 ), ),
g is differentiable at f (z). Therefore g ◦ f is differentiable at every point of N (z0 , ), i.e., g ◦ f is
analytic at z0 . 
9. Analytic Functions 23

Examples 9.7.
3
z +4
(1) Let f (z) = (z2 −3)(z 2 +1) . Determine the points in C where f is analytic also determine the points

in C which are singularities of f .

We note that the functions z 3 + 4, z 2 − 3 and z 2 + 1 are analytic at every complex number. If
z 3 +4
z0 ∈ C such that z02 −3 6= 0 and z02 +1 6= 0, then the quotient (z2 −3)(z 2 +1) is analytic at z0 . Thus

I
√ √
f√ is analytic
√ on the set C \ {− 3, 3, −i, i}. We notice that f fails to be analytic at the points
− 3, 3, −i, i (in fact the function is not defined at these points) but any neighborhood √ √of any of

BH
these points contains a point where f is analytic. Thus the singularities of f are − 3, 3, −i, i.
z 2 +1
(2) Let f (z) = (z+2)(z 2 +2z+2) .

We observe that z +1, z+2 and z 2 +2z+2 are polynomials. Therefore they are entire functions.
2

If z0 ∈ C, z0 6= −2 and z0 6= −1 ± i, then (z0 + 2)(z02 + 2z0 + 2) 6= 0. Therefore f (z) =

z 2 +1
(z+2)(z 2 +2z+2)
is analytic at z0 . Thus the set of points where f is analytic is C \ {−2, −1 ± i}.
Notice that f is not analytic at −2 and −1 ± i but any neighborhood of these points contain a
point where f is analytic. Thus the set of singularities of f is {−2, −1 ± i}.
Theorem 9.8. Let f be analytic on a domain D. If f 0 (z) = 0 for all z ∈ D, then f = 0.
DA
Proof. Let f = u + iv. Since f 0 (z) = 0 for all z ∈ D, ux (x, y) = vx (x, y) = 0 for all x + iy =
(x, y) ∈ D. By Cauchy - Riemann equations, we get ux (x, y) = uy (x, y) = vx (x, y) = vy (x, y) = 0
for all (x, y) ∈ D.
We show that u and v are constant functions. Fix a point P = (x0 , y0 ) ∈ D. Let P 0 = (x, y) be
a point in D such that there is a line segment joining P and P 0 , which lies in D. Define ϕ : [0, 1] → R
by ϕ(t) = u((1 − t)x0 + x, (1 − t)y0 + y). Then ϕ is continuous on [0, 1] and differentiable on
(0, 1). Let t ∈ (0, 1). Then by using chain rule, we have
ϕ0 (t) = ux ((1 − t)x0 + tx, (1 − t)y0 + ty)(x − x0 ) + uy ((1 − t)x0 + tx, (1 − t)y0 + ty)(y − y0 ).
Since ux and uy are zero at every element of D, we get ϕ0 (t) = 0 for all t ∈ (0, 1). Since (0, 1) is
connected, ϕ is constant on (0, 1). Continuity of ϕ on [0, 1] implies that ϕ is constant on [0, 1].
PA

Now if P, Q ∈ D, then there is there is a polygonal path joining P and Q which lies in D.
Since u is constant on each of the segment and these segments have common points, it follows that
the value u at P and Q is same. It means that u is constant, say u(x, y) = a for all (x, y) ∈ D.
Similarly, there is a constant b ∈ R such that v(x, y) = b for all (x, y) ∈ D. Thus f (z) = a + ib for
all z ∈ D, i.e., f is a constant map. 
Corollary 9.9. Suppose that a function f = u + iv and its conjugate f are both analytic in a
domain D. Then f is constant throughout D.
Proof. Since f = u + iv, f = u − iv. Since both f and f are analytic, we get ux = vy , uy = −vx ,
ux = −vy and uy = vx . These equations imply ux = uy = vx = vy = 0 on D. Therefore by
Theorem 9.8, f is a constant map. 
24 CONTENTS

Corollary 9.10. Suppose that f is analytic in a domain D. If |f | is constant on D, then f is a

constant map.
Proof. Let f = u + iv. Since |f | is constant, there is a constant k such that |f (z)|2 = k for all
z ∈ D. It means that u(x, y)2 + v(x, y)2 = k for all x + iy ∈ D. If k = 0, then both u and v are
zero functions. Suppose that k 6= 0. Then f (z) 6= 0 for all z ∈ D. Therefore f1 is analytic on D and

I
hence fk is analytic on D. Since f f = k, f = fk . This implies analyticity of f in D. Therefore f is a
constant map. 

BH
Lemma 9.11. Suppose that f is analytic in a domain D. If f is real valued, then f is a constant
map.
Proof. Let f = u + iv. Since f is real valued, v = 0. Since f is analytic on D, ux = vy = 0 and
uy = −vx = 0 on D. The continuity of partial derivatives of u on D implies that u is a constant map.
Thus f is a constant map. 

10. Harmonic Functions

DA
Definition 10.1. Let H be a real valued function defined on a domain D. If H has continuous partial
derivative of first and second order throughout D and if H satisfies the equation Hxx + Hyy = 0 for
every point of the domain D, then H is called a harmonic function in D.
The equation Hxx + Hyy = 0 is called the Laplace’s equation. Thus a function H is harmonic
if it satisfies the Laplace’s equation on D.
Examples 10.2.
(1) Let H(x, y) = ey sin x, where (x, y) ∈ R2 .
We see that Hx (x, y) = ey cos x, Hxx (x, y) = −ey sin x, Hy (x, y) = ey sin x and Hy (x, y) =
ey sin x. Thus Hxx (x, y) + Hyy = −ey sin x + ey sin x = 0. Thus H is harmonic on R2 .
Therefore H is harmonic on any open subset of R2 .
PA

We shall prove later that if f = u + iv is analytic in a domain D, then first and second order
partial derivatives on u are continuous on D. We shall use this result to prove
Theorem 10.3. Let f = u + iv be analytic in a domain D. Then both the functions u and v are
harmonic in D.
Proof. Since f = u + iv is analytic on a domain D, then functions u and v have continuous partial
derivatives of first and second order. Since f is analytic, the Cauchy-Riemann equations are satisfied
on D, i.e, ux = vy and uy = −vx on D. Differentiating these equations partial with respect to x and
y respectively, we have uxx = vyx and uyy = −vxy . The continuity of partial derivatives of u and v
imply uxy = uyx and vxy = vyx . Using the last equation we get uxx + uyy = 0 on D. Hence u is
harmonic on D. Similarly, v is harmonic on D. 
10. Harmonic Functions 25

The converse of Theorem 10.3 is not true. For example consider f (x + iy) = x2 − y 2 − i2xy.
Then u(x, y) = x2 − y 2 and v(x, y) = −2xy. Notice that u and v are harmonic function on R2 .
Now ux = 2x, uy = −2y, vx = −2y and vy = −2x. The above implies that the Cauchy-Riemann
equations are not satisfied at point unless it is (0, 0). Thus the function is not analytic at any point in
C but its real and imaginary parts are harmonic functions.

I
Definition 10.4. If u and v are harmonic on a domain D and if they satisfy the Cauchy-Riemann

BH
equations, then v is called a harmonic conjugate of u.

Theorem 10.5. A function f = u + iv is analytic in a domain D if and only if v is a harmonic

conjugate of u.

Proof. Suppose that f is analytic in a domain D. Then both u and v are harmonic on D and they
satisfy the Cauchy-Riemann equations. Therefore v is a harmonic conjugate of u.
Conversely, assume that v is a harmonic conjugate of u. Then by definition both the functions
u and v have continuous partial derivatives and they satisfy the Cauchy-Riemann equations at every
point of D. Therefore f is differentiable at every point of D, i.e, f is analytic on D, 

Examples 10.6.
DA
(1) Find a harmonic conjugate of y 3 − 3x2 y. Also, find an analytic function f whose real part is
y 3 − 3x2 y.
Let u(x, y) = y 3 −3x2 y. Then ux = −6xy, uxx = −6y, uy = 3y 2 −3x2 and uyy = 6y. Clearly,
the partial derivatives are continuous and uxx + uyy = 0 at every point x + iy ∈ C. Let v be a
harmonic conjugate of u. Then vy = −6xy and vx = −3y 2 +3x2 . Integrating vy with respect to
y, we get v(x, y) = −3xy 2 +ϕ(x) for some function ϕ. Therefore vx = −3y 2 +ϕ0 (x). This gives
ϕ0 (x) = 3x2 and hence ϕ(x) = x3 +c for some real constant c. Thus v(x, y) = −3xy 2 +x3 +c.
Define f (z) = u(x, y) + iv(x, y) = y 3 − 3x2 y + i(−3xy 2 + x3 + c). Then f is analytic. Now
we shall express f in terms of z. We have
PA

f (z) = y 3 − 3x2 y + i(−3xy 2 + x3 + c)

= i(x3 + 3x2 (iy) + 3x(iy)2 + (iy)3 ) + ic
= iz 3 + c.

(2) Find a harmonic conjugate of u(x, y) = y/(x2 + y 2 ) in some domain. Also, find an analytic
function f whose real part is y 3 − 3x2 y.
We note the functions u is defined on the domain D = R2 \ {(0, 0)}. Now ux = − (x22xy+y 2 )2
,
2 2 2 2
uy = (x (x
+y )−y(2y)
2 +y 2 )2
−y
= (xx2 +y 2 )2 . Clearly, both the functions ux and uy are continuous on D. You

can verify that the function uxx + uyy = 0. Therefore u is harmonic on D. Let v be a harmonic
2 −y 2
conjugate of u. Then vy = ux = − (x22xy +y 2 )2
and vx = −uy = − (xx2 +y 2xy
2 )2 . Integrating − (x2 +y 2 )2
26 CONTENTS

with respect to y we get v(x, y) = − (x22xy

R
+y 2 )2
dy + ϕ(x) for some function ϕ. Then
Z Z
2xy
v(x, y) = − 2 2 2
dy + ϕ(x) = −x (x2 + y 2 )−2 (2y)dy + ϕ(x)
(x + y )
x
= 2 + ϕ(x).

I
x + y2
(x2 +y 2 )−(2x)x y 2 −x2 y 2 −x2
Therefore vx = (x2 +y 2 )2
+ ϕ0 (x) = (x2 +y 2 )2
+ ϕ0 (x). This implies that (x2 +y 2 )2
+ ϕ0 (x) =

BH
2 2
−y
− (xx2 +y 0
2 )2 , i.e., ϕ (x) = 0. It means that ϕ is a constant map say c for some c ∈ R. Thus
x
v(x, y) = x2 +y 2 + c is a harmonic conjugate of u. If z 6= 0, then

y x
f (z) = f (x + iy) = u + iv = 2 + i + ic
x + y2 x2 + y 2
 
x − iy z
= i 2 2
+ ic = i 2 + ic
x +y |z|
z i
= + ic = + ic.
zz z
DA
(3) If v and V are harmonic conjugates of u in a domain D, then we show that v and V can differ
at most by an additive constant.
Since both v and V are harmonic conjugates of u, they are harmonic. Therefore their partial
derivatives are continuous. Since ux = vy , uy = −vx , ux = Vy and uy = −Vx , we get
(v − V )x = and (v − V )y = 0 on a domain D. Again, since the partial derivatives of v − V are
continuous on D, v − V is a constant map, i.e., v − V = c for some c ∈ R. It means that v and
V can differ at most by an additive constant.
(4) Suppose that v is a harmonic conjugate of u in a domain D and also that u is a harmonic
conjugate of v in D. Then we show that both u and v are constant throughout D.
Since v is a harmonic conjugate of u and u is a harmonic conjugate of v, we have ux = vy ,
uy = −vx , vx = uy and vy = −ux on D. These equations imply that ux = uy = vx = vy = 0.
PA

Since ux , uy , vx , and vy are continuous on D, the functions u and v are constant.

(5) Construct an analytic function having the imaginary part v(x, y) = e2x sin 2y − y.
We note the function v is differentiable on R2 and v is harmonic. We want to construct a function
u so that v is a harmonic conjugate of u. Then ux = vy and uy = −vx . Now vx = 2e2x sin y
and vy = 2e2x cos 2y − 1. Therefore ux = 2e2x cos 2y − 1. Integrating with respect to x we get
u(x, y) = e2x cos 2y − x + ϕ(y) for some y. Then uy = −2e2x sin 2y + ϕ0 (y). Since uy = −vx ,
we have ϕ0 (y) = 0, i.e., ϕ is a real constant, say, c. Therefore u(x, y) = e2x cos 2y − x + c.
Therefore required function f = u + iv will be

f (z) = f (x + iy) = u(x, y) + iv(x, y) = e2x cos 2y − x + c + i(e2x sin 2y − y)

= e2x (cos 2x + i sin 2x) − (x + iy) + c = e2z − z + c.
12. Integral of w(t) 27

(6) If a harmonic function is a harmonic conjugate of itself, then it is a constant map.

Suppose that a harmonic function v on a domain is a harmonic conjugate of itself. Then vx = vy
and vy = −vx imply vx = vy = 0 on D. Therefore v is a constant map.

11. Derivative of w(t)

I
Let w be a complex valued function of a real variable t. Then there are real valued functions u
and v of real variable t such that w(t) = u(t) + iv(t). The derivative of w at a point t is defined by

BH
w0 (t) = u0 (t) + iv 0 (t) provided both the derivatives u0 (t) and v 0 (t) exist.
We remark that w is continuous at t if and only if both u and v are continuous at t and w is
differentiable at t if and only if both u and v are differentiable at t.
Lemma 11.1. (1) Suppose that w = u + iv is differentiable at t and z0 = x0 + iy0 . Then
(z0 w)0 (t) = z0 w0 (t).
(2) If z0 = x0 + iy0 ∈ C, then dtd (ez0 t ) = z0 ez0 t .
Proof. We observe that
(z0 w)0 (t) = [(x0 + iy0 )(u + iv)]0 (t) = ([x0 u − y0 v] + i[x0 v + y0 u])0 (t)
DA
= x0 u0 (t) − y0 v 0 (t) + ix0 v 0 (t) + iy0 u0 (t)
= (x0 + iy0 )(u0 (t) + iv 0 (t)) = z0 w0 (t).
Now
d z0 t d (x0 +iy0 )t d
(e ) = (e ) = (ex0 teiy0 t )
dt dt dt
d x0 t
= (e (cos y0 t + i sin y0 t))
dt
= ex0 t x0 cos y0 t − ex0 t y0 sin y0 + i[ex0 t x0 sin y0 t + ex0 t y0 cos y0 t]
= x0 [ex0 t (cos y0 y + i sin y0 )] + iy0 [ex0 t (cos y0 y + i sin y0 )]
PA

= z0 ez0 t .

Remark 11.2. Mean Value Theorem is not true for complex valued functions of real variable.
Let w : [0, 2π] → C be w(t) = cos t+i sin t. Then w is continuous on [0, 2π] and differentiable
on (0, 2π). Note that w0 (t) = − sin t + i cos t for all t ∈ (0, 2π). Therefore w0 (t) 6= 0 for any
t ∈ (0, 2π). We note that there is no c ∈ (0, 2π) such that 0 = w(2π)−w(0)
2π
= w0 (c).

12. Integral of w(t)

Let w : [a, b] → C be a map. Then there are functions u, v : [a, b] → R such that w(t) =
u(t) + iv(t) for all t ∈ [a, b]. If the functions u and v are integrable over [a, b], then the integral of w
28 CONTENTS

over [a, b] is defined as

Zb Zb Zb
w(t)dt = u(t)dt + i v(t)dt.
a a a
It is obvious from definition of integral of w that

I
 b   b 
Z Zb Z Zb
Re  w(t)dt = Re w(t)dt and Im  w(t)dt = Im w(t)dt.

BH
a a a a
Remark 12.1.
(1) Let w(t) = u(t) + iv(t), t ∈ [a, b]. Suppose that U 0 = u and V 0 = v. Then by Fundamental
Rb Rb
Theorem of Integral Calculus, a u(t)dt = U (b) − U (a) and a v(t)dt = V (b) − V (a). Let
W (t) = U (t) + iV (t), t ∈ [a, b]. Then W 0 = w and
Zb Zb Zb
w(t)dt = u(t)dt + i v(t)dt = [U (b) − U (a)] + i[V (b) − V (a)] = W (b) − W (a).
a a a
DA
(2) The mean value theorem for integral calculus is not true for complex valued functions of real
Rb
variables. It states that if f is integrable over [a, b], then there is c ∈ (a, b) such that a f (t)dt =
(b − a)f (c). R 2π
Take w(t) = eit , t ∈ [0, 2π]. Then 0 w(t)dt = 0. Since w(t) 6= 0 for any t ∈ (0, 2π), there
is no c ∈ (0, 2π) such that eit (2π − 0) = 0.
Lemma 12.2 (Integral Inequality). Suppose that w : [a, b] → C is integrable. Then
Zb Zb
w(t)dt ≤ |w(t)|dt.
a a
Rb Rb
Proof. If a w(t)dt = 0, then there is nothing to prove. Let a
w(t)dt = r0 eiθ0 . Then r0 =
PA

Rb
a
w(t)dt . Also,
Zb Zb Zb
−iθ0
w(t)dt = r0 = e w(t)dt = e−iθ0 w(t)dt
a a a
 
Zb Zb
−iθ0
= Re  e w(t)dt [∵ e−iθ0 w(t)dt ≥ 0]
a a
Zb Zb
= Re(e−iθ0 w(t))dt ≤ |e−iθ0 w(t)|dt
a a
13. Contours 29

Zb
= |w(t)|dt.
a

This proves the result. 

I
Examples 12.3.
(1) Let w = u + iv. If w is differentiable t, then dtd [w2 ](t) = 2w(t)w0 (t).

BH
Since w = u + iv, w2 = u2 − v 2 + i(2uv). Thus
(w2 )0 (t) = 2u(t)u0 (t) − 2v(t)v 0 (t) + i(2u0 (t)v(t) + 2u0 (t)v(t))
= 2(u(t) + iv(t))(u0 (t) + iv 0 (t)) = 2w(t)w0 (t).
R2
2 R∞
(2) We shall evaluate 1 1t − i dt and Ą 0 e−zt dt, where Re z > 0.
Let z = x + iy and x > 0. Then
Z∞ Z∞
−zt
e dt = e−xt (cos yt − i sin yt)dt
0 0

e−xt
DA ∞  −xt
e
∞
= (−x cos xt + y sin yt) − i 2 (−x sin yt − y cos yt)
x2 + y 2 0 x + y2 0
x y z
= 2 −i 2 =
x + y2 x + y2 zz
1
= .
z
We observe that
Z2  2 Z2    2
1 1 2i −1 1
− i dt = 2
− − 1 dt = − 2i log(t) − t = − − i ln 4.
t t t t 1 2
1 1
PA

13. Contours
Definition 13.1. A set of points z = (x, y) in the complex plane is said to be an arc if x = x(t),
y = y(t), (a ≤ t ≤ b), where x(t) and y(t) are continuous functions of the real parameter t.
Equivalently, an arc in a complex plane is an image of continuous function z : x + iy : [a, b] →
C.
It is convenient to describe the points of an arc C by means of the equation z = z(t), (a ≤ t ≤
b), where z(t) = x(t) + iy(t).
Definition 13.2. An arc C is a called a simple arc or a Jordan arc if it does not cross itself, i.,e., C is
simple if z(t1 ) 6= z(t2 ) when t1 6= t2 . (It means that z is a one one function.)
30 CONTENTS

When an arc C is simple except for the fact that z(b) = z(a), we say that C is a simple closed
curve, or a Jordan curve.
A simple closed curve is called positively oriented when it is in the counterclockwise direction.

Definition 13.3.

I
(1) A curve z(t) = x(t) + iy(t) on [a, b] is called differentiable arc if z 0 exists and z 0 is continuous

BH
on [a, b].
(2) A differentiable arc z(t) = x(t) + iy(t) on [a, b] is called smooth arc if z 0 (t) 6= 0 for any
t ∈ (a, b).
(3) A contour, or piecewise smooth arc, is an arc consisting of a finite number of smooth arcs joined
end to end.
(4) When only the initial and final values of contour are the same, a contour is called a simple closed
contour.

Examples 13.4.
(
DA
x + ix when 0 ≤ x ≤ 1,
(1) Let z(x) = .
x+i when 1 ≤ x ≤ 2
Then z is an arc consisting of a line segment from 0 to 1 + i followed by one from 1 + i to 2 + i.
(2) Let z : z(θ) = eiθ , 0 ≤ θ ≤ 2π.
Then z is a circle about the origin with radius 1. It is a simple closed curve, oriented in the
counterclockwise direction.
(3) Let z : z(θ) = z0 + Reiθ , 0 ≤ θ ≤ 2π.
Then z is a circle about the z0 with radius R. It is a simple closed curve, oriented in the
counterclockwise direction.
(4) The arc z : z(θ) = e−iθ , 0 ≤ θ ≤ 2π is not the same as the arc described above. The set of
PA

points is the same, but now the circle is traversed in the clockwise direction.
(5) The points on the arc z : z(θ) = ei2θ 0 ≤ θ ≤ 2π are the same as those making up the arcs
above. The arc here differs, however, from each of those arcs since the circle is traversed twice
in the counterclockwise direction.
(6) It is known from Differential Calculus that if γ(t) = (x(t), y(t)) is a differentiable curve, then
the length of the curve between the points γ(a) and γ(b) is

Zb Zb p
kγ 0 (t)kdt = x0 (t)2 + y 0 (t)2 dt.
a a
14. Integral of a function 31

Thus if z(t) = x(t) + iy(t), t ∈ [a, b] is a differentiable arc, then the length of the arc between
the points z(a) and z(b) is
Zb p Zb
x0 (t)2 + y 0 (t)2 dt = |z 0 (t)|dt.

I
a a

14. Integral of a function

BH
Definition 14.1. Suppose that the equation z = z(t), (a ≤ t ≤ b) is a contour C, extending from a
point z1 = z(a) to a point z2 = z(b) and f [z(t)] is piecewise continuous on the interval [a, b]. Then
the line integral or contour of f along C is defined by
Zb
f [z(t)]z 0 (t)dt.
a
R
We shall denote it by C
f (z)dz.
Remarks 14.2. Some properties of the integral
R
DA
R
(1) If z0 ∈ C, then C z0 f (z)dz = z0 C f (z)dz.
Take g(z) = z0 f (z). Then
Z Z Zb Z Z
0 0
z0 f (z)dz = g(z)dz = z0 f [z(t)]z (t)dt = z0 f [z(t)]z (t)dt = z0 f (z)dz.
C C a C C
R R R
(2) C
[f (z) + g(z)]dz = C
f (z)dz + C
g(z)dz.
By definition
Z Zb Zb Zb
[f (z) + g(z)]dz = [f [z(t)] + g[z(t)]]z 0 (t)dt = f [z(t)]z 0 (t)dt + g[z(t)]z 0 (t)dt
C a a a
PA

Z Z
= f (z)dz + g(z)dz.
C C

(3) Associated with the contour C, is the contour −C, consisting of the same set of points but with
the order reversed so that the new contour extends from the point z(b) to the point z(a). If the
contour C has parametric representation z(t), a ≤R t ≤ b, then the contour
R −C has a parametric
representation z = z(−t), −b ≤ t ≤ −a. Then −C f (z)dz = − C f (z)dz.
By definition
Z Z−a Z−a
f (z)dz = f [z(−t)](z(−t)) dt = f [z(−t)](−z 0 (−t))dt
0

−C −b −b
32 CONTENTS

Za Zb
0
= − f [z(t)]z (t)(−dt) = − f [z(t)]z 0 (t)dt
b a
Z
= f (z)dz.

I
C

(4) Consider now a path C, with representation, that consists of a contour C1 from z1 to z2 followed

BH
by a contour C2 from z2 to z3 , then C is called the sum of the contours C1 and C2 . We shall
write
R it as C = CR1 + C2 . If f isR piecewise continuous
R on both C1 and C2 , then we show that
C1 +C2
f (z)dz = C f (z)dz = C1 f (z)dz + C2 f (z)dz.
Here C1 : z1 (t), a ≤ t ≤ c, C2 : z2 (t), c ≤ t ≤ b and z1 (c) = z2 (c). Then we take z(t) = z1 (t)
if a ≤ t ≤ c and z(t) = z2 (t) if c ≤ t ≤ b. Then z is piecewise smooth. Since f ◦ z1 and f ◦ z2
are piecewise continuous on [a, c] and [c, b] respectively, f ◦ z is piecewise continuous on [a, b].
Therefore

Zb Zc Zb
Z
f (z)dz =
DA f [z(t)]z 0 (t)dt = f [z(t)]z 0 (t)dt + f [z(t)]z 0 (t)dt
C a a c
Zc Zb
= f [z1 (t)]z10 (t)dt + f [z2 (t)]z20 (t)dt
a c
Z Z
= f (z)dz + f (z)dz.
C1 C2

ExamplesR 14.3. For the functions f (z) = (z + 2)/z and contours C described below, we shall
evaluate C f (z)dz.
PA

(1) the semicircle z = 2eiθ , (0 ≤ θ ≤ π)

(2) the semicircle z = 2eiθ , (π ≤ θ ≤ 2π)
(3) the circle z = 2eiθ , (0 ≤ θ ≤ 2π).
(i)

Z Zπ Zπ
f (z)dz = f [z(θ)]z 0 (θ)dθ = f (2eiθ )2eiθ idθ
C 0 0
Zπ
2eiθ + 2 iθ
= 2i e dθ
2eiθ
0
14. Integral of a function 33

Zπ π
eiθ

iθ
= 2i (e + 1)dθ = 2i +θ
i 0
0
= −4 + 2πi

(ii)

I
Z Z2π Z2π

BH
0
f (z)dz = f [z(θ)]z (θ)dθ = f (2eiθ )2eiθ idθ
C π π
Z2π
2eiθ + 2 iθ
= 2i e dθ
2eiθ
π
Z2π  iθ 2π
iθ e
= 2i (e + 1)dθ = 2i +θ
i π
π
= 4 + 2πi
R
DA
(iii) C f (z)dz
R = 4πi.
(4) Evaluate C f (z)dz when f (z) = z − 1 and C is the arc from z = 0 to z = 2 consisting of (a)
the semicircle z = 1 + eiθ , (π ≤ θ ≤ 2π); (b) the segment z = x (0 ≤ x ≤ 2) of the real axis.
(a) Here z(θ) = 1 + eiθ , π ≤ θ ≤ 2π. By definition
Z2π Z2π 2π
e2iθ
Z 
0 iθ iθ
f (z)dz = f (z(θ))z (θ)dθ = e ie dθ = i = 0.
2i π
C π π

(b) Here z(t) = t, 0 ≤ t ≤ 2. By definition

Z2 Z2
PA

Z
0
f (z)dz = f (z(t))z (t)dt = (t − 1)dt = 0.
C 0 0

(5) f (z) = π exp(πz) and C is the boundary of the square with vertices at the points 0, 1, 1 + i
and i, the orientation of C being in the counterclockwise direction.
Here the contour is the sum of the four contours given by C1 = z1 (t) = t, 0 ≤ t ≤ 1,
C2 : z(t) = 1 + it, 0 ≤ t ≤ 1, C3 : z3 (t) = 1 − t + i, 0 ≤ t ≤ 1 and C4 : z4 (t) = i(1 − t),
0 ≤ t ≤ 1. Thus
Z 4 Z 1
X
f (z)dz = f [zi (t)]zi0 (t)dt
C i=1 0
34 CONTENTS

Z1 Z1
= π exp(πt)dt + π exp(π(1 − it))idt
0 0
Z1 Z1
+ π exp(π(1 − t − i))(−dt) + π exp(−πi(1 − t))(−idt)

I
0 0
 1  1
exp(πt) exp(π(1 − it))

BH
= π + iπ
π 0 −iπ 0
 1  1
exp(π(1 − t − i)) exp(−iπ(1 − t))
−π − iπ
−π 0 iπ 0
= exp(π) − 1 − (exp(π − iπ) − exp(π))
+(exp(−iπ) − exp(π − iπ)) − (1 − exp(−iπ))
= 4(exp(π) − 1).
Theorem 14.4 (ML- inequality). Let C denote a contour of length L, and suppose that a function f
DA
is piecewise continuous on C. If M is a nonnegative constant such that |f (z)| ≤ M for all points
z on C at which f (z) is defined, then
Z
f (z)dz ≤ M L
C
Rb
Proof. Let the contour C be parameterized as z = z(t), a ≤ t ≤ b. We note that a
|z 0 (t)|dt = L.
Now
Z Zb Zb
f (z)dz = f [z(t)]z (t)dt ≤ |f [z(t)]||z 0 (t)|dt
0

C a a
PA

Zb
≤ M |z 0 (t)|dt = M L.
a


Examples 14.5.
(1) Let
R Cz+4 be the arc of the circle |z| = 2 from z = 2 to z = 2i that lies in the first quadrant. Then
C z 3 −1
dz ≤ 6π7
.
If z is on the arc, then |z| = 2. We note that the length of the arc is π. If |z| = 2, then
|z + 4| ≤ |z| + 4 = 6. Also, z31−1 ≤ ||z|31−1| = 17 . Thus if |z| = 2, then zz+4 6
3 −1 ≤ 7 . It follows

from ML - inequality that C z3 −1 dz ≤ 6π

R z+4
7
.
15. Cauchy and Cauchy-Goursat Theorems 35

(2) If C is the boundary of the triangle

R with vertices at the points 0, 3i, and −4, oriented in the
z
counterclockwise direction, then C (e − z)dz ≤ 60.
We note that the length of C is 12. If z ∈ C, then |z| ≤ 4. Also, |ez | = |ex+iy | = ex ≤ 1 as
z z
R z =z x ≤ 0. Thus if z ∈ C, then |e − z| ≤ |e | + |z| ≤ 5. Now ML - inequality implies that
Re
C
(e − z)dz ≤ 60.

I
15. Cauchy and Cauchy-Goursat Theorems

BH
Definition 15.1. Let f be a function defined on a domain D. A function F defined on D is called an
antiderivative of f if F 0 (z) = f (z) for all z ∈ D.
We note that an antiderivative of a function is unique except for an additive constant, i.e., if F
and G are antiderivatives of a function f on a domain D, then F − G is a constant function.
Theorem 15.2. Suppose that a function f is continuous on a domain D. The following statements
are equivalent.
(1) f has an antiderivative F throughout D.
(2) The integrals of f along contours lying entirely in D and
R z extending from any fixed point z1
DA
to any fixed point z2 all have the same value, namely z12 f (z)dz = F (z2 ) − F (z1 ), where
F is the antiderivative of f .
(3) The integrals of f around closed contours lying entirely in D all have value zero.
Theorem 15.3 (Green’s Theorem). Suppose that two real-valued functions P (x, y) and Q(x, y),
together with their first-order partial derivatives, are continuous throughout the closed region R
consisting of all points interior to and on the simple closed contour C. Then
Z ZZ
(P dx + Qdy) = (Qx − Py )dxdy.
C R

Theorem 15.4 (Cauchy’s Theorem). Let C be a simple closed contour. Let f be analytic within and
PA

0
on C. Let
R R be the closed region consisting of all points within and on C. If f is continuous on
R, then C f (z)dz = 0.
Proof. Let f = u + iv. Then f 0 = ux + ivx . Since f is analytic on R, ux = vy and uy = −vx on
R. Since f 0 is continuous on R, the functions ux , uy , vx and vy are continuous on R. Now
Z Z Z
f (z)dz = (u + iv)(dx + idy) = [(udx − vdy) + i(vdx + udy)]
C C C
ZZ ZZ
= (−vx − uy )dxdy + (ux − vy )dxdy [By Green’s Theorem]
R R
= 0.
36 CONTENTS


TheoremR15.5 (Cauchy-Goursat Theorem). Let f be analytic within and on a simple closed contour
C. Then C f (z)dz = 0.
If f is analytic within and on a simple closed contour C, then by Cauchy-Goursat Theorem

I
f (z)dz = 0. The converse is not true. For example consider f (z) = z12 , z 6= 0. Let C be the unit
R
C
circle. Then

BH
Z Z2π
1 iθ
f (z)dz = ie dθ = 0.
e2iθ
C 0

Lemma 15.6. Let C1 be a positively oriented simple closed contour, and let C2 be a negatively
oriented simple closed contour
R such that C2R is interior to C1 . If f is analytic on the closed region
bounded by C1 and C2 , then C1 f (z)dz + C2 f (z)dz = 0.
Proof. We introduce two polygonal segments joining C1 and C2 shown in the figure.R Let C10 , C100 ,
C 0 , C 00 , L and L2 be as shown in the figure. Then by Cauchy-Goursat Theorem, C 0 f (z)dz +
R2 2 1 R 1
R−L2
f (z)dz +
R
C20 f
DA
(z)dz +
R
−L1
f (z)dz = 0 and
R
00
RC1
f (z)dz +
R
RL1
f (z)dz + 00 f (z)dz +
RC2
f (z)dz = 0. Adding these two equations, we have C 0 f (z)dz + C 00 f (z)dz + C 0 f (z)dz +
RL2 R R 1 1 2

C 00 f (z)dz = 0, i.e., C1
f (z)dz + C2
f (z)dz = 0. 
2

Corollary 15.7 (Principle of deformation of paths). Let C1 and C2 be positively oriented simple closed
contours Rsuch that C2 lies
R interior to C1 . If f is analytic on the closed region bounded by C1 and
C2 , then C1 f (z)dz = C2 f (z)dz.
R
Proof. Since C2 is positive oriented, −C2 is negatively oriented. By last Lemma, we have C1 f (z)dz+
R R R R R
−C2
f (z)dz = 0. But −C2 f (z)dz = − C2 f (z)dz. Hence C1 f (z)dz = C2 f (z)dz. 
The above result says that if C2 is continuously deformed into C1 by passing through points
PA

where f is analytic, then the value of integral of f along C2 does not change.

16. Cauchy Integral Theorem

Theorem 16.1 (Cauchy Integral Formula). Let f be analytic within and on a simple closed contour
C, taken in the positive sense. If z0 is any point interior to C, then
Z
1 f (z)
f (z0 ) = dz.
2πi z − z0
C

Proof. Let ρ > 0 be small enough so that the circle with center z0 and ρ lies interior to C. Let
Cρ denote a positively oriented circle |z − z0 | = ρ. Since the quotient f (z)/(z − z0 ) is analytic
between and on the contours Cρ and C, it follows from the principle of deformation of paths that
16. Cauchy Integral Theorem 37

R f (z) R f (z) R 1
C z−z0
dz= Cρ z−z 0
dz. We note that Cρ z−z0
= 2πi. Take any  > 0. Then by continuity of f at
z0 , there is δ > 0 such that |f (z) − f (z0 )| <  whenever |z − z0 | < δ. Let 0 < ρ < δ. Note that the
length of Cρ is 2πρ and the function f (z)−fz−z0
(z0 )
is continuous on Cρ . If z ∈ Cρ , i.e., |z − z0 | = ρ < δ,
f (z)−f (z0 )
then z−z0
≤ ρ . Therefore by ML- inequality, we have

I
f (z) − f (z0 )
Z

dz ≤ 2πρ = 2π. (16.1.1)

BH
z − z0 ρ
Cρ

Thus if 0 < ρ < δ, then

Z Z
1 f (z) 1 f (z)
dz − f (z0 ) = dz − f (z0 )
2πi z − z0 2πi z − z0
C Cρ

Z Z
1 f (z) 1 f (z0 )
= dz − dz
2πi z − z0 2πi z − z0
DA Cρ Cρ

f (z) − f (z0 )
Z
1
≤ dz
2π z − z0
Cρ

1
≤ 2π = .
2π
1
R f (z)
Since  > 0 is arbitrary, we get 2πi C z−z0
dz = f (z0 ). 
Examples 16.2.
z
R
(1) Let C be the positively oriented circle |z| = 2. We evaluate C (9−z 2 )(z+i)
dz.
2
Here (9−z2z)(z+i) = z/(9−z )
. Note that the function f (z) = z
within and on is analytic
PA

z+i 9−z 2
C : |z| = 2 and −i is interior to C. Thus by Cauchy integral formula
−i
Z Z
z f (z) π
2
dz = dz = 2πif (−i) = 2πi 2
= .
(9 − z )(z + i) z − (−i) 9 − (−i) 5
C C

(2) Let C denote the positively oriented boundary of the square whose sides lie along the lines
x = ±2 and y = ±2. We evaluate the following.
(a) C exp(−z) dz. Note that exp(−z) is analytic within and on C and πi
R
z− πi 2
is an interior to C.
2
Therefore
Z
exp(−z) πi
πi dz = 2πi exp(− ) = 2π.
z− 2 2
C
38 CONTENTS

cos z
√
dz. Since the singularities of z21+8 are ±2 2 and
R
(b) C z(z 2 +8)
they are outside C, the function
cos z
z 2 +8
is analytic within and on C. The point 0 is interior to C. Thus

cos z/(z 2 + 8)
Z Z
cos z cos 0 iπ
dz = dz = 2πi 2 = .
z(z 2 + 8) z−0 0 +8 4
C C

I
(3) We find the value of the integral of g around the circle |z − i| = 2 in the positive sense when
g(z) = z21+4 . Here the singularities of z21+4 are ±2i. Note that −2i lies outside the contour

BH
1
C : |z − i| = 2. Therefore z+2i is analytic within and on C. Since −2i is interior to C, by using
Cauchy integral formula we have
Z Z 1
1 z+2i 1 π
2
dz = dz = 2πi = .
z +4 z − 2i 2i + 2i 2
C C

Theorem 16.3. Let f be analytic within and on a simple closed contour C. Then f is differentiable
on the interior of C and Z
0 1 f (z)
f (z0 ) = dz
DA 2πi (z − z0 )2
C

for all z0 in the interior to C.

Proof. Fix a point z0 which is interior to C. We show that f is differentiable at z0 and f 0 (z0 ) =
1
R f (z)
2πi C (z−z0 )2
dz. Let d = inf{|z0 − z| : z ∈ C}. Since z0 is interior to C and C is compact,
d > 0. If 0 < |∆z| < d, then z0 + ∆z is interior to C and hence by Cauchy integral formula
1
R f (s)
f (z0 + ∆z) = 2πi C s−z0 −∆z
ds. Now
Z  
f (z0 + ∆z) − f (z0 )
Z
1 f (s) 1 1 1 1 ∆z
− ds = − − f (s)ds
∆z 2πi (s − z0 )2 ∆z 2πi s − z0 − ∆z s − z0 (s − z0 )2
C C
Z
1 ∆zf (s)
PA

= ds. (16.3.1)
2πi (s − z0 − ∆z)(s − z0 )2
C

Let M = sup{|f (s)| : s ∈ C}. Since f is continuous and C is compact, 0 < M < ∞. Let L
be the length of the contour C. Since |s − z0 | ≥ d and 0 < |∆z| < d, we have |s − z0 − ∆z| =
|(s − z0 ) − ∆z| ≥ ||s − z0 | − |∆z|| ≥ d − |∆z| > 0. Using equation (16.3.1) we get

f (z0 + ∆z) − f (z0 )

Z Z
1 f (s) 1 ∆zf (s)
− ds = ds
∆z 2πi (s − z0 )2 2πi (s − z0 − ∆z)(s − z0 )2
C C
1 |∆z|M L
≤ → 0 as ∆z → 0.
2π (d − |∆z|)d2
16. Cauchy Integral Theorem 39

R f (z)
Thus f is differentiable at z0 and f 0 (z0 ) = 1
2πi C (z−z0 )2
dz. Assume that the result is true for n, i.e.,
(n) n!
R f (z)
n- th derivative of f exists and f (z0 ) = 2πi C (z−z0 )n+1
dz. 
Theorem 16.4. Let f be analytic within and on a simple closed contour C. Then all the derivatives
of exist at every point of interior to C and

I
Z
(n) n! f (z)
f (z0 ) = dz,
2πi (z − z0 )n+1

BH
C

for all n ∈ N and for all z0 in the interior to C.

The proof of the Theorem 16.4 can be given by using Mathematical Induction and by a tedious
computation.
Corollary 16.5. If f is analytic on a domain D, then all the derivatives of f are analytic on D.
Proof. Let z0 ∈ D. Then there is ρ > 0 such that circle with center z0 and radius ρ lies inside D.
Then f will be analytic within and on the circle |z − z0 | = ρ. Thus by above Theorem f (n) (z0 )
exists for all n. Since z0 is arbitrary and D is a domain, it follows that f (n) is analytic on D for all
n ∈ N.
DA 
The following is the converse of Cauchy’s Theorem.
R
Theorem 16.6 (Morera’s Theorem). Let f be continuous on a domain D. If C f (z)dz = 0 for
every closed contour C in D, then f is analytic throughout D.
R
Proof. Since f is analytic and C f (z)dz = 0 for every closed contour C in D, f has an antiderivative
on D. That is, there is a function F : D → C such that F 0 (z) = f (z) for all z ∈ D. Since F is
differentiable at every point of D, F is analytic on D. By Cauchy Integral Formula all the derivatives
of F are analytic on D. In particular, F 0 = f is analytic on D. 
Theorem 16.7 (Cauchy’s Inequality). Suppose that a function f is analytic inside and on a positively
PA

oriented circle CR , centered at z0 and with radius R. If MR denotes the maximum value of |f (z)|
on CR , then
n!MR
|f (n) (z0 )| ≤ (n ∈ N ∪ {0}).
Rn
Proof. Since f is analytic within and on CR and z0 is interior to CR , by Cauchy integral formula
n!
R f (z)
f (n) (z0 ) = 2πi CR (z−z0 )n+1
dz for all n ∈ N ∪ {0}. Since |f (z)| ≤ MR for all z ∈ CR , we have
f (z) MR
(z−z0 )n+1
≤ Rn+1
for all z ∈ CR . As the length of CR is 2πR, it follows from ML-inequality that
n! MR n!MR
|f (n) (z0 )| ≤ n+1
2πR = for all n ∈ N ∪ {0}.
2π R Rn

40 CONTENTS

Theorem 16.8 (Liouville’s Theorem). If a function f is entire and bounded in the complex plane,
then f is constant throughout the plane.

Proof. Since f is bonded, there is M > 0 such that |f (z)| ≤ M for all z ∈ C. Take any z ∈ C. Let
R > 0. Since f is an entire function, f is analytic within and on a circle CR with center z and radius
R. It follows from Cauchy’s Inequality that |f 0 (z)| ≤ M . Since R > 0 is arbitrary, we get f 0 (z) = 0.

I
R
Since z ∈ C is arbitrary, f 0 (z) = 0 for all z ∈ C. Since C is a domain and f 0 (z) = 0 for all z ∈ C,
f is a constant function. 

BH
Theorem 16.9 (Fundamental Theorem of Algebra). Any polynomial P (z) = a0 +a1 z +a2 z 2 +· · ·+
an z n (an 6= 0) of degree n, n ≥ 1, has at least one zero. That is, there exists at least one point z0
such that P (z0 ) = 0.
1
Proof. Suppose that P (z) is not zero for any z ∈ C. Then the function f (z) = P (z) , z ∈ C,
a0 a1
is an entire function. Since P is nonconstant, f is nonconstant. Let w(z) = zn + zn−1 + · · · +
an−1 ai
z
, z ∈ C \ {0}. Since each of zn−i → 0 as z → ∞, there is R > 0 such that |w(z)| =
a0 a1 an−1 |an |
zn
+ zn−1 + · · · + z < 2 whenever |z| > R. Notice that P (z) = (an + w(z))z n . If |z| > R,
DA
then |P (z)| = |an + w(z)||z|n ≥ (|an | − |w(z)|)|z|n ≥ (|an | − |a2n | )Rn = |a2n | Rn . Thus |f (z)| =
1
|P (z)|
≤ |an2|Rn whenever |z| > R. Since f is continuous and the set |z| ≤ R is compact, there is
M > 0 such that |f (z)| ≤ M for all |z| ≤ R. Thus |f (z)| ≤ M + |an2|Rn for all z ∈ C. Thus the
function is nonconstant, entire and bounded. This is a contradiction to Liouville’s Theorem. Hence
there is z0 ∈ C such that P (z0 ) = 0. 

Corollary 16.10. Let P be a polynomial of degree n, n ≥ 1. Then P has exactly n zeros, counted
according to their multiplicities.

Proof. Since P is a polynomial of degree n, by the Fundamental Theorem Algebra there is z1 ∈ C

such that P (z1 ) = 0. But then P (z) = (z − z1 )P1 (z) for some nonzero polynomial P1 . If P1 (z) is
constant (i.e., deg P1 = 0), then we are done. Otherwise P1 is a polynomial of degree n − 1 ≥ 1.
PA

Again applying the same result to P1 , there is z2 ∈ C such that P1 (z) = (z − z2 )P2 (z) for some
polynomial. Therefore P (z) = (z − z1 )(z − z2 )P2 (z). Applying induction we get P (z) = (z −
z1 )(z − z2 ) · · · (z − zn )c for some nonzero constant. Thus P has exactly n zeros. 

Examples 16.11.
(1) Let f be an entire function. If the real part of f is bounded above, the f is a constant map.
Define g : C → C by g(z) = exp(f (z)), z ∈ C. Since both exp and f are entire functions, g
is an entire function. Let f = u + iv. Since u = Re f is bounded above, there is M ∈ R such
that u(x, y) ≤ M for all (x, y) ∈ R2 . Let z = x + iy ∈ C. Then |g(z)| = | exp(u(x, y) +
iv(x, y))| = exp(u(x, y)) ≤ exp(M ). Thus g is a bonded entire function and Liouville’s
Theorem implies that g(z) = exp(f (z)) is a constant function. Then exp(f (z))f 0 (z) = 0 for
17. Maximum Modulus Principle 41

all z ∈ C. Since exp(f (z)) 6= 0 for any z ∈ C, f 0 (z) = 0 for all z ∈ C. Therefore f is a
constant map.
(2) Let f be an entire function. If the real part of f is bounded below, the f is a constant map.
(3) Let f be an entire function. If the imaginary part of f is bounded above, the f is a constant map.
(4) Let f be an entire function. If the imaginary part of f is bounded below, the f is a constant map.

I
17. Maximum Modulus Principle

BH
Lemma 17.1. Suppose that |f (z)| ≤ |f (z0 )| at each point z in some neighborhood |z − z0 | <  in
which f is analytic. Then f has the constant value f (z0 ) throughout that neighborhood.
Proof. Let 0 < ρ < . Then the circle Cρ with center z0 and radius ρ lies inside N (z0 , ). We may
parametrize cρ as z(θ) = z0 + ρeiθ , 0 ≤ θ ≤ 2π. Since f is analytic on N (z0 , ), f is analytic within
an on Cρ . Therefore Cauchy Integral Formula gives
Z2π Z2π
f (z0 + ρeiθ ) iθ
Z
1 f (z) 1 1
f (z0 ) = dz = iθ
iρe dθ = f (z0 + ρeiθ )dθ. (17.1.1)
2πi z − z0 2πi ρe 2π
Cρ 0 0

Therefore |f (z0 )| ≤ 2π 1
R 2π
DA
|f (z0 + ρeiθ )|dθ. Since |f (z)| ≤ |f (z0 )| for all z ∈ N (z0 , ), |f (z0 +
0 R 2π
1
ρeiθ )| ≤ |f (z0 )| for all θ ∈ [0, 2π]. Takin integration, we have 2π 0
|f (z0 + ρeiθ )|dθ ≤ |f (z0 )|.
1
R 2π
Thus 2π 0
|f (z0 + ρeiθ )|dθ = |f (z0 )|. Now
Z2π Z2π
iθ
|f (z0 + ρe )|dθ = 2π|f (z0 )| = |f (z0 )|dθ
0 0
R 2π
gives 0 (|f (z0 )| − |f (z0 + ρe )|)dθ = 0. Since |f (z0 )| − |f (z0 + ρeiθ )| is a nonnegative continuous
iθ

function of θ and its integral is zero, we have |f (z0 + ρeiθ )| = |f (z0 )| for all θ ∈ [0, 2π]. Since
0 < ρ <  is arbitrary, |f (z)| = |f (z0 )| for all z ∈ N (z0 , ). Thus f is analytic on N (z0 , ) and |f |
is constant. Therefore f is constant and f (z) = f (z0 ) for all z ∈ N (z0 , ). 
PA

Lemma 17.2 (Gauss’s mean value theorem). If a function f s analytic within and on a given circle
Cρ : z(θ) = z0 + ρeiθ , 0 ≤ θ ≤ 2π, then
Z2π
1
f (z0 ) = f (z0 + ρeiθ )dθ.
2π
0

Proof. Cauchy Integral Formula gives

Z2π Z2π
f (z0 + ρeiθ ) iθ
Z
1 f (z) 1 1
f (z0 ) = dz = iθ
iρe dθ = f (z0 + ρeiθ )dθ.
2πi z − z0 2πi ρe 2π
Cρ 0 0
42 CONTENTS


The above result says that is a function is analytic within and on a given circle, then its value at
the center is the arithmetic mean of its values on the circle. This result is called Gauss’s mean value
theorem.

I
Theorem 17.3 (Maximum Modulus Principle). If a function f is analytic and not constant in a given
domain D, then |f | has no maximum value in D. That is, there is no point z0 in the domain such

BH
that |f (z)| ≤ |f (z0 )| for all points z in it.
Proof. Suppose that there is z0 ∈ D such that |f (z)| ≤ |f (z0 )| for all z ∈ D. Let P be any
point in D. Let L be a polygonal path between z0 and z1 . Let d be shortest distance between
L and the boundary of D. If D is an entire plane, then we take d to be any positive real. We
observe that there is a finite sequence of points z0 , z1 , . . . , zn in L such that |zk − zk+1 | < d for all
k = 0, 1, . . . , n and P = zn . Let Ni = N (zi , d), i = 0, 1, 2, . . . , n. We see that f is analytic in
each of these neighborhoods, which are all contained in D, and that the center of each neighborhood
Nk , k = 1, 2, ..., n lies in the neighborhood Nk−1 . Since |f | was assumed to have a maximum value
in D at z0 , it also has a maximum value in N0 at that point. Hence, according to the preceding
DA
lemma, f has the constant value f throughout N0 . In particular, f (z1 ) = f (z0 ). This means that
|f (z)| ≤ |f (z1 )| for each point z in N1 ; and the lemma can be applied again, this time telling us that
f (z) = f (z1 ) = f (z0 ) for all z ∈ N1 . Since z2 is in N1 , f (z2 ) = f (z0 ). Hence |f (z)| ≤ |f (z2 )|,
when z is in N2 ; and the lemma is once again applicable, showing that f (z) = f (z2 ) = f (z0 ), when
z is in N2 . Continuing in this manner, we eventually reach the neighborhood Nn and arrive at the
fact that f (zn ) = f (z0 ). Since P = zn is an arbitrary point of D, f (z) = f (z0 ) for all z ∈ D. 
Corollary 17.4. Suppose that a function f is continuous on a closed bounded region R and that it
is analytic and not constant in the interior of R. Then the maximum value of |f | in R, which is
always reached, occurs somewhere on the boundary of R and never in the interior.
Proof. Since |f | is continuous on R and R is compact, there is z0 ∈ R such that |f (z)| ≤ |f (z0 )|
PA

for all z ∈ R. Suppose that z0 is not on the boundary of R, i.e., z0 is in the interior of R. Since the
interior of R is a domain and |f (z)| ≤ |f (z0 )| for all z ∈ R, Maximum Modulus Principle implies
that f (z) = f (z0 ) for all z in the interior of R. Since f is continuous on R, we get f (z) = f (z0 ) for
all z ∈ R. This is a contradiction to the fact that f is nonconstant. Hence the maximum value of |f |
is attained on the boundary only. 
Examples 17.5.
(1) Let R denote the rectangular region [0, π] × [0, 1]. Let f (z) = sin z. Since f is an entire
function, f is continuous on R and is analytic in the interior of R. Note that f is nonconstant.
By Maximum Modulus Principle, f has a maximum
p value in R that occurs somewhere on the
boundary only. We observe that | sin z| = sin x + sinh2 y for all z. Since the maximum of
2
17. Maximum Modulus Principle 43

sin2 x is attained at π/2 and the maximum of sinh2 y is attained at 1, the maximum of |f | is
attained at the point (π/2, 1) = π/2 + i
(2) Let a function f be continuous on a closed bounded region R, and let it be analytic and not
constant throughout the interior of R. Assuming that f (z)| 6= 0 anywhere in R, prove that |f |
has a minimum value in R which occurs on the boundary of R and never in the interior.

I
Since f (z) 6= 0 for any z ∈ R and f is continuous R, f1 is continuous on R. Also, f1 is analytic
in the interior of R. Since f is nonconstant, f1 is nonconstant. By Maximum Modulus Principle

BH
| f1 | has a maximum value on the boundary of R only. But the maximum of | f1 | is the minimum
of |f |.
(3) The above is not true if f (z) = 0 for some z ∈ R. For example, let f : N (0, 1) → C be
f (z) = z. Then |f | has the minimum value in the interior of N (1, 1), namely at 0.
(4) Let R be a closed and bounded region. Suppose that f is continuous on R, analytic on the
interior of R and nowhere zero in the interior of R. If |f | is constant on the boundary of R, then
f is a constant map.
Since f is nowhere vanishing in the interior of R, there is z1 ∈ ∂R such that |f (z1 )| ≤ |f (z)|
for all z ∈ R, by Minimum Modulus Principle. Also, by Maximum Modulus Principle there is
DA
z2 ∈ ∂R such that |f (z)| ≤ |f (z2 )| for all z ∈ R. Since |f | is constant on ∂R, |f (z1 )| = |f (z2 )|.
Thus |f | is constant on R. In particular, |f | is constant on the domain R◦ . Therefore f is constant
on R◦ . Let z ∈ ∂R. Then there is a sequence (zn ) in R◦ such that zn → z. Since f is continuous,
f (zn ) → f (z). As (f (zn )) is a constant sequence, the value of f at z is same as the value of f
at any point of R◦ . Therefore f is constant on R.
(5) Consider the function f (z) = (z + 1)2 and the closed triangular region R with vertices at the
points 0, 2, and i. Find points in R where |f | has its maximum and minimum values.
We observe that |z+1|2 is the square of the distance of z from −1. The function f is nonconstant
and analytic on R and f (z) 6= 0 for any z ∈ R. Thus the maximum value and minimum value
of |f | are attained at boundary of R only. Note that 0 is closest to −1 and 2 is farthest from
origin. Thus the minimum and maximum of |f | are attained at 0 and 2 respectively.
PA

(6) Let f = u + iv be a function that is continuous on a closed bounded region R and analytic
and not constant throughout the interior of R. We show that the component function u has a
minimum value in R which occurs on the boundary of R and never in the interior.
Define g(z) = exp(f (z)). Then g is continuous on R and it is analytic in the interior of R. Since
exp(f (z)) 6= 0 for any z, g(z) 6= 0 for all z ∈ R. Since f is nonconstant, g is nonconstant.
By above g has minimum on the boundary of R only. Therefore there is z0 = (x0 , y0 ) in the
boundary of R such that |g(z0 )| ≤ |g(z)| for all z ∈ R. This gives u(x0 , y0 ) ≤ u(x, y) for all
(x, y) ∈ R.
Suppose that u(x0 , y0 ) ≤ u(x, y) for all (x, y) ∈ R and (x0 , y0 ) is in the interior of R. If
z = (x, y) ∈ R, then |g(z0 )| = u(x0 , y0 ) ≤ u(x, y) = |g(z)|. This is a contradiction (by (ii)
above).
44 CONTENTS

(7) Let f be the function f (z) = ez and R the rectangular region [0, 1] × [0, π]. We find the points
in R where the component function u(x, y) = Re f (z) reaches its maximum and minimum
values.
Here u(x, y) = ex cos y. By above exercises u has minimum and maximum value on the
boundary of R only. The function cos x is maximum at 0 and ex is maximum at 1. Thus the

I
maximum value of u is attained at (1, 0). Now the minimum value of cos x is −1 and in attained
at π. Thus the minimum value of u is attained at (1, π) = 1 + iπ.

BH
Exercise 17.6. Let the function f = u+iv be continuous on a closed bounded region R. Suppose that
it is analytic and not constant in the interior of R. Show that the component function v has maximum
and minimum values in R which are reached on the boundary of R and never in the interior.

18. Taylor’s Theorem

We notice that if |z| < 1 and N ∈ N, then
∞ N −1 ∞ N −1 ∞ N −1
1 X
n
X
n
X
n
X
n N
X
n
X zN
= z = z + z = z +z z = zn + .
1−z 1−z
n=0 n=0
DA
n=N n=0 n=0 n=0

Theorem 18.1 (Taylor’s Theorem). Suppose that a function f is analytic throughout a disk |z −
z0 | < R, centered at z0 and with radius R, where 0 < R ≤ ∞. Then f has the power series
representation
∞
X f (n) (z0 )
f (z) = (z − z0 )n (|z − z0 | < R).
n=0
n!

Proof. If z = z0 , then the required equality holds. Let z ∈ N (z0 , R) and z 6= z0 . Let |z − z0 | = r.
Then 0 < r < R. Let r0 ∈ R be such that 0 < r < r0 < R. Let C0 be the positively oriented circle
with center z0 and radius r0 . Then f is analytic within and on C0 . Note that z is interior to C0 . Take
any N ∈ N. Using Cauchy Integral Formula we get
PA

Z
1 f (s)
f (z) = ds.
2πi s−z
C0

We also observe that

Z
(n) n! f (s)
f (z0 ) = for all n ∈ N ∪ {0}.
2πi (s − z0 )n+1
C0

z−z0
Notice that if s ∈ C0 , then |z − z0 | < |s − z0 |, i.e., s−z0
< 1. So, if s ∈ C0 , then
1 1 1 1
= =
s−z s − z0 − (z − z0 ) s − z0 1 − z−z
s−z0
0
18. Taylor’s Theorem 45

−1 
N n  N !
1 z − z0
X z − z0 1
= +
s − z0 n=1
s − z0 s − z0 1 − z−z
s−z0
0

N −1 N
(z − z0 )n

X z − z0 1
= n+1
+ .
n=1
(s − z0 ) s − z0 s−z

I
Therefore
Z

BH
1 f (s)
f (z) = ds
2πi s−z
C0
Z "N −1 N #
(z − z0 )n

1 X z − z0 1
= + f (s)ds
2πi n=0
(s − z0 )n+1 s − z0 s−z
C0
N −1 N
(z − z0 )n
Z 
z − z0
Z
X f (s) 1 f (s)
= n+1
ds + ds
n=0
2πi (s − z0 ) 2πi s − z0 s−z
C0 C0
N −1 N
f (n) (z0 )
Z 
=
X
DA n!
(z − z0 )n +
1
2πi
z − z0
s − z0
f (s)
s−z
ds.
n=0 C0

f (n) (z0 ) R  z−z0 N f (s)

Hence to show f (z) = ∞ n
P
n=0 n!
(z−z0 ) we need to show that ρN (z) = 2πi C0 s−z0 s−z
ds →
0 as N → ∞. If s ∈ C0 , then |s − z0 | = r0 . Now |s − z| = |s − z0 − (z − z0 ) ≥ |s − z0 | − |z − z0 | =
r0 − r and z−z 0
s−z0
= rr0 < 1. Let M be the maximum of |f | on the circle C0 , i.e., |f (s)| ≤ M for all
 N
1 (z−z0 )N f (s) M r
s ∈ C0 . Thus 2πi (s−z0 )N s−z
≤ (r0 −r)2π r0
. Since the length of C0 is 2πr0 , the ML-Inequality
gives
 N  N
M r M r0 r
|ρN (z)| ≤ 2πr0 = .
(r0 − r)2π r0 r0 − r r0
PA

 N
f (n) (z0 )
Since rr0 → 0, i.e., ρN (z) → 0, it follows that the sum of the series ∞ (z − z0 )n is
P
n=0 n!
f (z). Since z ∈ N (z0 , R) is arbitrary, we have
∞
X f (n) (z0 )
f (z) = (z − z0 )n (z ∈ N (z0 , R)).
n=0
n!

Note that the Taylor series of f is unique. That is, if f is analytic in N (z0 , ), f (z) = ∞
P
n=0 an (z−
n
P ∞ n
z0 ) , z ∈ N (z0 , ) and f (z) = n=0 bn (z − z0 ) , z ∈ N (z0 , ) for some sequences (an ) and (bn )
in C, then an = bn for all n.
46 CONTENTS

Examples 18.2.
(1) Let f (z) = ez . Then f is entire function. If z ∈ C, then there is R > 0 such that |z| < R. We
note that f is analytic in N (0, R). We also note that f (n) (0) = 1 for all n ∈ N ∪ {0}. Thus by
Taylor’s Theorem f (z) = ez = ∞ zn
P
n=0 n! . Since z ∈ C is arbitrary, we get
∞

I
X zn
ez = (z ∈ C).
n=0
n!

BH
(2) If z0 ∈ C, then
∞
z z−z0 +z0 z0 z−z0
X ez0
e =e =e e = (z − z0 )n for all z ∈ C.
n=0
n!
P∞ (3z)n P∞ 3n
(3) It follows that e3z = n=0 n!
= n=0 n! z
n
for all z ∈ C.
(4) We see that
∞
X 5n
z 2 e5z = z n+2 (z ∈ C).
n=0
n!
DA
(5) Let f (z) = sin z. We know that sin is an entire function. We observe that f (2n) (0) = 0 and
f (2n+1) (0) = (−1)n for all n ∈ N ∪ {0}. Since f is an entire function, it follows from similar
arguments above that
∞
X (−1)n 2n+1
sin z = z (z ∈ C).
n=0
(2n + 1)!
(6) Show using similar argument above that if z ∈ C, then
∞ ∞ ∞
X (−1)n 2n
X 1 2n X 1
cos z = z , cosh z = z and sinh z = z 2n+1 .
n=0
(2n)! n=0
(2n)! n=0
(2n + 1)!
1
(7) Let f (z) = 1−z
Then f is analytic on the open disc N (0, 1). Note that f (n) (0) = n! for all
.
PA

n ∈ N ∪ {0}. Thus
∞
1 X
= zn (|z| < 1).
1−z n=0
Replacing z by −z in above expression, we have
∞
1 X
= (−1)n z n (|z| < 1).
1+z n=0

(8) Let f (z) = z1 . Then f is analytic in N (1, 1). It follows from above example that
∞
1 1 X
= = (−1)n (z − 1)n (|z − 1| < 1).
z 1 + (z − 1) n=0
19. Power series an analytic function 47

1
(9) We shall derive the Taylor’s series of 1−z about i.
√ 1
√
Since the distance of i and 1 is 2, the function 1−z is analytic in the disc N (i, 2). Thus
1
√ √
f (z) = 1−z has the power series representation in N (i, 2). Let z ∈ N (i, 2), i.e., |z − i| <
√
2. Then
1 1 1 1

I
= =
1−z 1 − i − (z − i) 1 − i 1 − z−i
1−i

BH
∞  n
1 X z−i z−i
= [∵ < 1]
1 − i n=0 1 − i 1−i
∞
X (z − i)n
= n+1
.
n=0
(1 − i)

(10) Let f (z) = sin(z 2 ). We find the Maclaurin series of f .

If z ∈ C, then
∞ ∞
2
X (−1)n 2 2n+1
X (−1)n 4n+2
sin z = (z ) = z .
DA
n=0
(2n + 1)! n=0
(2n + 1)!

It follows that f (4n+1) (0) = f (4n+3) (0) = f (4n) (0) = 0 for all n ∈ N ∪ {0}.

19. Power series an analytic function

Lemma 19.1. If a power series ∞ n
P
n=0 an (z − z0 ) converges when z = z1 (z1 6= z0 ), then it is
absolutely convergent at each point z in the open disk |z − z0 | < R1 where R1 = |z1 − z0 |.

Proof. Here ∞ n n
P
n=0 an (z1 − z0 ) converges. Therefore an (z1 − z0 ) → 0 as n → ∞. Consequently,
n
there is M ≥ 0 such that |an (z1 − z0 ) | ≤ M for all n. Let z ∈ C with |z − z0 | < R1 = |z1 − z0 |.
Then zz−z 0
< 1. So,
PA

1 −z0

∞ ∞
X X (z − z0 )n
|an (z − z0 )n | = |an (z1 − z0 )n |
n=0 n=0
(z1 − z0 )n
∞ n
X z − z0 z − z0
≤ M < ∞. [∵ < 1]
n=0
z1 − z0 z1 − z0

This proves the result. 

Theorem 19.2 (Abel’s Lemma). Let (an ) be a sequence of complex numbers, and let z0 ∈ C. Then
there is 0 ≤ R ≤ ∞ such that the power series ∞ n
P
n=0 an (z − z0 ) converges if |z − z0 | < R and
it diverges if |z − z0 | > R.
48 CONTENTS

The number R is called the radius of convergence of the power series.

It follows from Abel’s Lemma and previous result that if R is the radius of convergence of
P∞ n
P∞ n
n=0 an (z −z0 ) , then n=0 an (z −z0 ) converges absolutely for every z ∈ C satisfying |z −z0 | <
R.
We use the following well known result to prove the next result.

I
Theorem 19.3 (Weierstrass M - Test). Let E be a nonempty subset of C, and let (fn ) be sequence
complex valued functions defined on E. If there is a sequence (Mn ) of nonnegative real such that

BH
|fn (z)| ≤ Mn for all z ∈ E and for all n, then the series ∞
P
n=1 fn converges uniformly on E.

Lemma 19.4. If R > 0 is the radius of convergence of the power series ∞ n

P
P∞ n=0 an (z − z0 ) , then
n
n=0 an (z − z0 ) converges uniformly on N (z0 , R1 ) for every 0 < R1 < R.

Proof. Let 0 < R1 < R. Let z1 ∈ N (z0 , R) be such that |z0 − z1 | = R1 . Then ∞ |an ||z1 − z0 |n
P
n=0
converges. We see that sup |an ||z − z0 |n ≤ |an |R1n for all n ∈ N ∪ {0}. Since ∞ n
P
n=0 |an |R1 =
z∈N (z0 ,R1 )
P∞ n
it follows from Weierstrass M - test that the series ∞ n
P
n=0 |a n ||z1 − z0 | converges, n=0 an (z − z0 )
converges uniformly on N (z0 , R1 ). 
DA P∞
Let R be the radius of convergence of the power series n=0 an (z − z0 )n . Let C be a com-
pact subset of N (z0 , R). Then C will be contained in N (z0 , R1 ) for some 0 < R1 < R. Since
P∞ n
n=0 an (z − z0 ) converges uniformly on N (z0 , R1 ), it converges uniformly on C.
Thus a power series converges uniformly on every compact subset contained in the disc of
convergence of the power series.
Let N (z0 , R) be the disc of convergence of the power series S(z) = ∞ n
P
n=0 an (z − z0 ) . For
N ∈ N, let
N
X −1 ∞
X
n
SN (z) = an (z − z0 ) and ρN (z) = an (z − z0 )n for all z ∈ N (z0 , R).
n=0 n=N

Notice that
PA

S(z) − SN (z) = ρN (z)for all z ∈ N (z0 , R) and for all N ∈ N.

Lemma 19.5. Let N (z0 , R) be the disc of convergence of the power series S(z) = ∞ n
P
n=0 an (z−z0 ) .
Let K be a compact subset of N (z0 , R). Then for given  > 0 there is M ∈ N such that |ρN (z)| < 
for all z ∈ K and for all N ≥ M .
Proof. We know that ∞ n
P
n=0 an (z − z0 ) converges uniformly on K. That is, SN → S uniformly
on K. Therefore for given  > 0 there is M ∈ N such that |SN (z) − S(z)| <  for all z ∈ K and
N ≥ M . This is the same as |ρN (z)| <  for all z ∈ K and for all N ≥ M . 
P∞
Corollary 19.6. Let N (z0 , R) be the disc of convergence of S(z) = n=0 an (z − z0 )n . Then S is
continuous on N (z0 , R).
19. Power series an analytic function 49

Proof. Let w ∈ N (z0 , R). Then w is contained in a compact set N (z0 , R1 ) for some 0 < R1 < R.
PN −1 n
Since SN (z) = n=0 an (z − z0 ) are continuous and uniform limit of continuous functions is
continuous, it follows that S is continuous on N (z0 , R1 ). In particular, S is continuous at w. Since w
is an arbitrary point of N (z0 , R), the function S is continuous on N (z0 , R). 

I
Theorem 19.7 (Term by term integration of power series). Let N (z0 , R) be the disc of convergence
of the power series S(z) = ∞ n
P
n=0 an (z − z0 ) . Let C be a contour lying in N (z0 , R), and let g be
a continuous function on C. Then

BH
Z ∞
X Z
g(z)S(z)dz = an g(z)(z − z0 )n dz.
C n=0 C

Proof. We first notice that S, SN and ρN are continuous on N (z0 , R) for all N ∈ N. Let N ∈ N.
Then
Z Z
g(z)S(z)dz = g(z)(SN (z) + ρN (z))dz
C C
N −1 Z Z
DA =
X
an n
g(z)(z − z0 ) dz + g(z)ρN (z)dz.
n=0 C C
R
Thus to prove the result we need to show that C g(z)ρN (z)dz → 0 as N → ∞. Let M =
sup{|g(z)| : z ∈ C}. Since C is compact and g is continuous, 0 ≤ M < ∞. Let L be the length
of C. Take any  > 0. Since C is compact, there is M ∈ N such that |ρN (z)| < (1+M )(1+L) for all
z ∈ C and for all N ≥ M . Thus if N ≥ M , then
Z

g(z)ρN (z)dz ≤ M L < .
(1 + M )(1 + L)
C

Hence C g(z)S(z)dz = ∞
R P R n
n=0 an C g(z)(z − z0 ) dz. 
PA

P∞
Corollary 19.8. Let N (z0 , R) be the disc of convergence of the power series S(z) = n=0 an (z −
z0 )n . Then S is analytic on N (z0 , R).
Proof. Notice that S is continuous on N (z0 , R). Let C be any simple closed contour lying in
N (z0 , R). Take g(z) = 1 for all z. Then g is continuous. It follows from term by term integra-
tion of power series that
Z X∞ Z
S(z)dz = an (z − z0 )n dz.
C n=0 C
n
R
Since (z − z0 ) is an entire function for all nR ∈ N ∪ {0} and C is a simple closed contour, C (z −
z0 )n dz = 0 for all n ∈ N ∪ {0}. Therefore C S(z)dz = 0. It follows from Morera’s Theorem that
S is analytic on N (z0 , R). 
50 CONTENTS

Theorem 19.9 (Term by term differentiation of power series). Let N (z0 , R) be the disc of conver-
gence of the power series S(z) = ∞ n
P
n=0 an (z − z0 ) . Then
∞
X
S 0 (z) = nan (z − z0 )n−1 for all z ∈ N (z0 , R).
n=1

I
Proof. Take any z ∈ N (z0 , R). Then there is ρ ∈ R such that |z − z0 | < ρ < R. Let C : z(θ) =
1 1
z0 +ρeiθ , 0 ≤ θ ≤ 2π. Let g(s) = 2πi . Then g is continuous on C. By term by term integration

BH
(s−z)2
we have
∞ ∞
(s − z0 )n
Z Z Z Z
1 S(s) X
n
X 1
ds = g(s)S(s)ds = a n g(s)(s − z0 ) = a n ds.
2πi (s − z)2 n=0 n=0
2πi (s − z) 2
C C C C

1 S(s)
R 0 1
R (s−z0 )n
It follows from Cauchy Integral Formula for derivatives that 2πi C (s−z) 2 ds = S (z), 2πi C (s−z)2 ds =
d n n−1 1
R 1 0
P∞
ds
(s − z0 ) |z=z 0
= n(z − z0 ) for all n ∈ N and 2πi C (s−z) 2 ds = 0. Thus S (z) = n=1 nan (z −
n−1
z0 ) for all z ∈ N (z0 , R). 

Corollary 19.10. Let N (z0 , R) be the disc of convergence of the power series S(z) = ∞
P
n
z0 ) . If N ∈ N, then
DA n=0 an (z −

∞
X
(N )
S (z) = n(n − 1) · · · (n − N + 1)an (z − z0 )n−N for all z ∈ N (z0 , R).
n=N
P∞ n
Corollary 19.11 (Uniqueness of the Taylor series). Let f (z) = n=0 an (z − z0 ) for all z ∈
(n)
N (z0 , R), then an = f n!(z0 ) for all n ∈ N ∪ {0}.

Proof. If N ∈ N ∪ {0}, then f (N ) (z) = ∞ n−N

P
n=N n(n − 1) · · · (n − N + 1)an (z − z0 ) for all
(n)
f (z0 )
(N )
z ∈ N (z0 , R). In particular, f (z0 ) = N !aN . Thus an = n! for all n ∈ N ∪ {0}. 

Corollary 19.12. If a function f is analytic on a domain D, then all its derivatives are analytic on
PA

D.

Proof. Take any z ∈ D. Since D is an open set, there is R > 0 such that N (z0 , R) ⊂ D. Then
f (n) (z0 )
by Taylor’s Theorem f can be written in the power series form f (z) = ∞ (z − z0 )n for
P
n=0 n!
all z ∈ N (z0 , R). But then by above all the derivatives of f are analytic on N (z0 , R). Hence the
proof. 

20. Laurent’s Theorem

Theorem 20.1 (Laurent’s Theorem). Suppose that a function f is analytic throughout an annular
domain R1 < |z − z0 | < R2 , centered at z0 , and let C denote any positively oriented simple
20. Laurent’s Theorem 51

closed contour around z0 and lying in that domain. Then, at each point in the domain, f has the
series representation
∞ ∞
X
n
X bn
f (z) = an (z − z0 ) + (R1 < |z − z0 | < R2 ),
n=0 n=1
(z − z0 )n

I
where
Z Z
1 f (z) 1 f (z)
an = dz, (n = 0, 1, 2, . . .) and bn = dz, (n = 1, 2, . . .).

BH
2πi (z − z0 )n+1 2πi (z − z0 )−n+1
C C

Proof. We notice that if γ is any positively oriented (simple) circle surrounding z0 in R1 < |z −z0 | <
R2 , then by the principle of deformation of paths γ (s−zf (s) f (s)
R R
0)
n+1 ds = C (s−z0 )n+1
ds for all n ∈ Z
and for any simple closed contour C in R1 < |z − z0 | < R2 surrounding z0 . Fix z ∈ C with
R1 < |z −z0 | < R2 . Then there are real numbers r1 and r2 such that R1 < r1 < |z −z0 | < r2 < R2 .
Let C1 : z(θ) = z0 + r1 eiθ , 0 ≤ θ ≤ 2π and C2 : z(θ) = z0 + r2 eiθ , 0 ≤ θ ≤ 2π. Let γ be a
positively oriented circle about z such that it lies in r1 < |w − z0 | < r2 . Then by Cauchy Integral
formula, we have
DA
f (z) =
1
2πi
Z
f (s)
s−z
ds.
γ
f (s)
Since the function is analytic within and a region bounded by C1 , C2 and γ, Cauchy’s Theorem
s−z
for multiply connected region implies that
Z Z Z
f (s) f (s) f (s)
ds − ds − ds = 0.
s−z s−z s−z
C2 C1 γ

Therefore
Z Z
1 f (s) 1 f (s)
f (z) = ds − ds. (20.1.1)
2πi s−z 2πi s−z
PA

C2 C1

Notice that if s ∈ C2 , then | z−z

s−z0
0
| < 1. Let s ∈ C2 and N ∈ N. Then
1 1 1 1
= =
s−z s − z0 − (z − z0 ) s − z0 1 − z−z
s−z0
0

N −1 N
(z − z0 )n

X z − z0 1
= + .
n=0
(s − z0 )n+1 s − z0 s−z
Thus
−1
N N !
(z − z0 )n

z − z0
Z Z
1 f (s) 1 X 1
ds = f (s) + ds
2πi s−z 2πi n=0
(s − z0 )n+1 s − z0 s−z
C2 C2
52 CONTENTS

N −1 N
f (s)(z − z0 )n
Z 
z − z0
Z
X 1 1 f (s)
= ds + ds
n=0
2πi (s − z0 )n+1 2πi s − z0 s−z
C2 C2
N −1 Z  N
X 1 z − z0 f (s)
= an (z − z0 )n + ds.
2πi s − z0 s−z

I
n=0 C2

So,

BH
N −1 Z  N
z − z0
Z
1 f (s) X 1 f (s)
ds − an (z − z0 )n = ds.
2πi s−z n=0
2πi s − z0 s−z
C2 C2
Let M = sup{|f (s)| : s ∈ C2 }. Since C2 is compact and f is continuous, 0 ≤ M < ∞. If s ∈ C2 ,
then |s − z| = |s − z0 − (z − z0 )| ≥ |s − z0 | − |z − z0 | = r2 − |z|. Using M L- Inequality, we get
Z  N N
1 z − z0 f (s) 1 z − z0 2πr2 M z − z0
ds ≤ → 0 as N → ∞ [∵ < 1].
2πi s − z0 s−z 2π r2 r2 − |z − z0 | r2
C2

1
R f (s) P∞ 1
R f (s) P∞ bn
Thus 2πi ds = an (z−z0 )n . Using similar arguments we get 2πi ds =− n=1 (z−z0 )n .
Thus
C2 s−z n=0
DA C1 s−z

∞ ∞
X X bn
f (z) = an (z − z0 )n + .
n=0 n=1
(z − z0 )n
Since z is an arbitrary point satisfying R1 < |z − z0 | < R2 , the above holds for all z with R1 <
|z − z0 | < R2 . 
Like Taylor’s series the Laurent series of a function is unique.
1 1
R
It follows from Laurent’s Theorem that 2πi C f (z)dz = b1 , i.e., the coefficient of z−z0
in the
1
R
Laurent series expansion of f is 2πi C
f (z)dz.
Examples 20.2.
PA

(1) Find the Laurent’s series of f (z) = e1/z , z 6= 0.

Then function f is analytic on C \ {0}. Let z ∈ C \ {0}. Then
∞
1/z
X 1
e = .
n=0
n!z n
The coefficient of z1 in the Laurent series of f is 1. Thus C e1/z dz = 2πi for any simple closed
R

contour C surrounding 0.
1
(2) Let f (z) = (z−i) 2 , z 6= i.
1
Then (z−i)2 is itself is the Laurent series of f . Therefore an = 0 for all n ∈ N∪{0}, b1 = bn = 0
for all n ≥ 3 and b2 = 1. Let n ∈ N ∪ {0}. Let C be any simple closed contour in C \ {i}
1
R f (z) 1
R 1
R 1
surrounding i. Then 0 = an = 2πi C (z−i) n+1 dz = 2πi C (z−i)n+3 dz = 0, i.e., C (z−i)n
dz = 0
20. Laurent’s Theorem 53

1 1 1
R R
for all n ≥ 3. Now 0 = b1 = 2πi C
f (z)dz = 2πi C (z−i)2
dz = 0. If n ≥ 3, then 0 = bn =
1
R f (z) 1
R 1
R 1
2πi C (z−i)−n+1
dz = 2πi C R(z−i)−n+3
dz. Since n ≥ 3, −n + 3 ≤ 0. Thus C (z−i) n = 0 for all
1 −2+1 1
R
n ≤ 0. Since b2 = 1, 2πi = C (z−i)2 (z − i) dzWe note that C (z−i)2 dz = 0 for all simple
closed contour C surrounding i.
1
(3) Let f (z) = (z−1)(z−2) .

I
We note that 1 and 2 are singularities of f . Thus the function f is analytic on C \ {1, 2}. We
find the Laurent series expansion of f in the regions |z| < 1, 1 < |z| < 2 and |z| > 2. We note

BH
1 1
that f (z) = z−2 − z−1 .
z
Let |z| < 1. Then | 2 | < 1. Now

1 1 1 1 1
f (z) = − =− z +
z−2 z−1 21− 2
1−z
∞ ∞
1 X zn X n
= − + z
2 n=0 2n n=0
∞  
X 1
= 1 − n+1 z n .
DA n=0
2

Let 1 < |z| < 2.

Then | z1 | < 1 and | z2 | < 1. Therefore

1 1 1 1 1 1
f (z) = − =− z − 1
z−2 z−1 21− 2
z1− z
∞ ∞
1 X zn 1 X 1
= − −
2 n=0 2n z n=0 z n
∞ ∞
X zn X 1
= − n+1
− n+1
2 z
PA

n=0 n=0
∞ ∞
X zn X 1
= − n+1
− n
.
n=0
2 n=1
z

Let |z| > 2.

Then | z1 | < 1 and | z2 | < 1. Therefore

1 1 1 1 1 1
f (z) = − = 2 − 1
z−2 z−1 z1− z
z1− z
∞ ∞
1 X 2n 1 X 1
= −
z n=0 z n z n=0 z n
54 CONTENTS

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

I
21. Singularities of a function

BH
We recall that z0 ∈ C is a singular point or a singularity of f if f is not analytic at z0 and every
neighborhood of z0 contains a point where f is analytic.
Definition 21.1. A singularity z0 ∈ C of f is called an isolated singularity if there is a deleted
neighborhood of z0 in which f is analytic.
Examples 21.2.
(1) Let f (z) = z1 . Then 0 is a singularity of f and 0 is an isolated singularity.
z+1
(2) Let f (z) = z2 (z 2 +1) . Then 0, −i and i are singularities of f and they are singularities of f .

(3) Let f (z) = sin π . If z 6= n1 for any n ∈ Z and z 6= 0, then sin πz 6= 0. Since sin πz is analytic
1
z
DA
at all such points, the set of singularities of f is {0, n1 : n ∈ Z}. The singularities 1
are isolated
n
singularities and 0 is a nonisolated singularity.
If z0 ∈ C is an isolated singularity of f , then there is R > 0 such that f is analytic in 0 <
|z − z0 | < R. It follows from Laurent’s Theorem that f has the representation
∞ ∞
X
n
X bn
f (z) = an (z − z0 ) + (0 < |z − z0 | < R),
n=0 n=1
(z − z0 )n
where
Z Z
1 f (z) 1 f (z)
an = dz, (n = 0, 1, 2, . . .) and bn = dz, (n = 1, 2, . . .)
2πi (z − z0 ) n+1 2πi (z − z0 )−n+1
PA

C C

and C is any simple closed contour in 0 < |z − z0 | < R surrounding z0 .

The series ∞ bn
P
n=1 (z−z0 )n is called the principal part of f .

Definition 21.3. Let z0 be an isolated singularity of f .

1
(1) The coefficient of z−z 0
is the Laurent series of f is called the residue of f at z0 . We denote it by
Res f (z) or Res f .
z=z0 z=z0
(2) If the principal part of f at z0 is zero, then z0 is called a removable singularity of f .
(3) If the principal part of f at z0 has finitely many terms and at least one nonzero term, then z0 is
called a pole of f . The largest integer m in the principal part of f for which bm 6= 0 is called
the order of the pole at z0 .
21. Singularities of a function 55

(4) If the principal part of f at z0 contains infinitely many terms, then z0 is called a essential singu-
larity of f . If m = 1, then z0 is called a simple pole of f .
If z0 is a removable singularity of f , then f can be made analytic at z0 by defining appropriate
value at z0 . In fact, one should define f (z) = limz→z0 f (z).
z
For example consider f (z) = e z−1 . Then 0 is an isolated singularity of f . If z 6= 0, then

I
ez − 1 z2 z3 z2
 
1 z
= 1+z+ + + ··· − 1 = 1 + + + ··· .

BH
z z 2! 3! 2! 3!
Therefore the principal part of f contains no term, i.e., 0 is a removable singularity of f . If we define
f (0) = 1, then f is analytic at 0.
Let f (z) = sinz z . Then 0 is a singularity of f as f is not defined at 0. If z 6= 0, then
z3 z5 z7 z2 z4 z6
 
sin z 1
= z− + − + ··· = 1 − + − + ··· .
z z 3! 5! 7! 3! 5! 7!
Therefore 0 is a removable singularity of f . The function f will be analytic at 0 if we define f (0) = 1.
2 +1
Let f (z) = zz−2 . One observes that 2 is an isolated singularity. If 0 < |z − 2| < ∞, then
z2 + 1
=
(z − 2 + 2)2 + 1
DA
=
1
(5 + 4(z − 2) + (z − 2)2 ) =
5
+ 4 + z − 2.
z−2 z−2 z−2 z−2
Therefore 2 is a simple pole of f .
1
Let f (z) = e z . Then 0 is an essential singularity of f . If z 6= 0, then
1 1 1 1
ez = 1 + + 2 + 3 + ··· .
z z 2! z 3!
That is the principal part of f contains infinitely many terms. Therefore 0 is an essential singularity
of f .
Example 21.4. Find the residue of the following functions at given singularity.
(1) f (z) = z + z1 at 0.
PA

We see that z + z1 is the Laurent series expansion of f in 0 < |z| < ∞. The coefficient of 1
z
is
1. Thus Resz=0 f = 1.
(2) f (z) = cosh
z4
z
at 0.
We observe that f is analytic in the region 0 < |z| < ∞. If z ∈ C \ {0}, then
∞
1 X z 2n z2 z4
 
cosh z 1
= 4 = 4 1+ + + ··· .
z4 z n=0 (2n)! z 2! 4!
1
We see that the coefficient of z
is 0. Thus Resz=0 f = 0.
Theorem 21.5. Let z0 be an isolated singularity of f . Then z0 is a pole of f order m if and only
if there is a function ϕ which is analytic at z0 , ϕ(z0 ) 6= 0 and f (z) = (z−z10 )m ϕ(z) for all z in
56 CONTENTS

some neighborhood of z0 . If m = 1, then Resz=z0 f = limz→z0 (z − z0 )f (z) and if m > 1, then

1 dm−1
Resz=z0 f = (m−1)! dz m−1
[(z − z0 )m f (z)]|z=z0 .

Proof. Assume that there is a function ϕ which is analytic at z0 , ϕ(z0 ) 6= 0 and f (z) = (z−z10 )m ϕ(z)
for all 0 < |z − z0 | < R for some R > 0. Since ϕ is analytic in N (z0 , R), it has the Taylor series
00

I
representation ϕ(z) = ϕ(z0 ) + ϕ0 (z0 )(z − z0 ) + ϕ 2!(z0 ) (z − z0 )2 + · · · for all |z − z0 | < R. If
0 < |z − z0 | < R, then

BH
∞
1 X ϕ(n) (z0 )
f (z) = m
(z − z0 )n .
(z − z0 ) n=0 n!
This is the Laurent series of f (by uniqueness of Laurent series). Since ϕ(z0 ) 6= 0, the coefficient
ϕ(z0 ) of (z−z10 )m is nonzero. Hence z0 is a pole of f of order m.
Conversely, assume that z0 is a pole of f of order m. Then f has the Laurent series represen-
tation
∞ m
X
n
X bn
f (z) = an (z − z0 ) + (0 < |z − z0 | < R).
n=0 n=1
(z − z0 )n
DA
Since z0 is a pole of order m, bm 6= 0. We may write the above expression as
" ∞
#
1 X
f (z) = bm + bm−1 (z − z0 ) + · · · + b1 (z − z0 )m−1 + an (z − z0 )m+n ,(21.5.1)
(z − z0 )m n=0

for all 0 < |z − z0 | < R. Define ϕ : N (z0 , R) → C by

∞
X
m−1
ϕ(z) = bm + bm−1 (z − z0 ) + · · · + b1 (z − z0 ) + an (z − z0 )m+n (|z − z0 | < R).
n=0

Since the above power series is convergent for every z ∈ N (z0 , R), ϕ is analytic in N (z0 , R). Also,
ϕ(z0 ) = bm 6= 0. We also note that f (z) = (z−z10 )m ϕ(z) for 0 < |z − z0 | < R.
PA

1
If z0 is a simple pole of f , then it follows from equation (21.5.1) that the coefficient b1 of z−z 0
is
limz−z0 f (z)(z − z0 ), i.e., Resz=z0 f = limz−z0 f (z)(z − z0 ).
Let m > 1. If 0 < |z − z0 | < R, then
1
f (z) = ϕ(z)
(z − z0 )m
" ∞
#
0 (m−1) (m+n)
1 ϕ (z 0 ) ϕ (z0 ) X ϕ (z0 )
= ϕ0 (z0 ) + (z − z0 ) + · · · + (z − z0 )m−1 + (z − z0 )m+n
(z − z0 )m 1! (m − 1)! n=0
(m + n)!

1 ϕ(m−1) (z0 )
Therefore the coefficient of z−z0
in the Laurent series of f about z0 is (m−1)!
. But f (z)(z −z0 )m =
ϕ(m−1) (z0 ) 1 dm−1
ϕ(z) (0 < |z − z0 | < R). Hence Resz=z0 f = (m−1)!
= (m−1)! dz m−1
[(z − z0 )m f (z)]|z=z0 
21. Singularities of a function 57

Exercise 21.6. Let f and g be analytic at z0 . Suppose that f (z0 ) 6= 0, g(z0 ) = 0 and g 0 (z0 ) 6= 0.
Then show that z0 is simple pole of fg and find the residue of fg at z0 .
g (n) (z0 )
Let R1 > 0 be such that g(z) = ∞ (z − z0 )n for all z ∈ N (z0 , R1 ). Since g(z0 ) = 0,
P
n=0 n!
P∞ g(n) (z0 )
we have g(z) = (z − z0 ) n=1 n! (z − z0 )n−1 for all z ∈ N (z0 , R1 ). Define h : N (z0 , R1 ) → C
g (n) (z0 )
by h(z) = ∞ (z − z0 )n−1 . Then h is analytic at z0 and h(z0 ) = g 0 (z0 ) 6= 0. Choose

I
P
n=1 n!
0 < R < R1 such that f is analytic on N (z0 , R) and h0 (z) 6= 0 for any z ∈ N (z0 , R). Then
f 1 f
(z) = z−z (z) for all 0 < |z − z0 | < R, fh is analytic at z0 and fh (z0 ) 6= 0. Therefore z0 is a

BH
g 0 h
f (z0 )
simple pole of fg . So, Resz=z0 f
g
= limz→z0 (z − z0 ) fg (z) = limz→z0 fh (z) = g 0 (z0 )
.
Definition 21.7. A simply connected domain D is a domain such that every simple closed contour
within it encloses only points of D.
Theorem 21.8 (Cauchy Residue Theorem). Let C be a simple closed contour, described in the posi-
tive sense. If a function f is analytic inside and on C except for a finite number of singular points
zk (k = 1, 2, . . . , n) inside C, then
Z n
X
DA f (z)dz = 2πi Res f.
k=1
z=zk
C

Proof. Let the points zk (k = 1, 2, . . . , n) be centers of positively oriented circles Ck which are
interior to C and are so small that no two of them have points in common. The circles Ck , together
with the simple closed contour C, form the boundary of a closed region throughout which f is analytic
and whose interior is a multiply connected domain consisting of the points inside C and exterior to
each Ck . Hence, according to the adaptation of the Cauchy – Goursat theorem to such domains, we
have Z Xn Z
f (z)dz − f (z)dz = 0.
C k=1 C
k
PA

R
But Ck
f (z)dz = 2πi Resz=zk f for k = 1, 12, . . . , n. Hence
Z X n
f (z)dz = 2πi Res f.
z=zk
C k=1


Exercise 21.9.
R 5z−2
(1) Let C : |z| = 2 be the positively oriented circle. We shall evaluate C z(z−1) dz.
5z−2
Let f (z) = z(z−1) . Then f is analytic within and on C except the singularities 0 and 1. Note that
ϕ1 (z) = 5z−2
z−1
is analytic in a neighbourhood of 0 and ϕ1 (0) 6= 0. Also, ϕ2 (z) = 5z−2
z
is analytic
in a neighbourhood of 1 and ϕ2 (1) 6= 0. Therefore points 0 and 1 are simple poles of f and
58 CONTENTS

5z−2
hence Resz=0 f = limz→0 z z(z−1) = limz→0 5z−2
z−1
5z−2
= 2 and Resz=1 f = limz→1 (z − 1) z(z−1) =
limz→1 5z−2
R
z
= 3. Thus by Cauchy Residue Theorem C f (z)dz = 2πi(2 + 3) = 10πi.
(2) We find the residue at 0 of the functions

1 z−sin z sinh z
(a) z+z 2 (c) z (e) z 4 (1−z 2 )

I
cot z
(b) z cos z1 (d) z 4
(f) eiz
2
.
(1−cos z)2

BH
1
(a) We note that 0 is an isolated singularity of f (z) = z+z 2
. If 0 < |z| < 1, then
1 1 1 1
f (z) = = (1 − z + z 2 − z 3 + · · · ) = − 1 + z − z 2 + · · · .
zz+1 z z
Therefore 0 is a pole of order 1 and Resz=0 f = 1.
(b) Clearly, 0 is an isolated singularity of z cos z1 . If z 6= 0, then
 
1 1 1 1 1 1 1
z cos = z 1 − 2 + 4 − 6 + · · · = z − + 3 − 5 + ··· .
z z 2! z 4! z 6! z2! z 4! z 6!
DA
Therefore 0 is an essential singularity of z cos z1 and Resz=0 z cos z1 = − 12 .
(c) If z 6= 0, then
z3 z5 z2 z4
 
z − sin z 1
= z−z+ − + ··· = − + ··· .
z z 3! 5! 3! 5!
Therefore 0 is a removable singularity of z−sinz
z
and its residue at 0 is 0.
cot z
(d) Notice that z4 = z5 sin z cos z, the function sinz z cos z is analytic at 0 and its value at 0 is
1 z

nonzero. Therefore 0 is a pole of cot

z4
z
of order 5.
d4
cot z
One way to find the residue of z4 at 0 is to compute 4!1 dz 4 (z cot z)|z=0 .
2 4
The following is another way. Let δ > 0 be sufficiently small (so that | z3! − z5! + · · · | < 1
whenever 0 < |z| < δ). If 0 < |z| < δ, then
PA

z2 4
cot z 1 1− 2!
+ z4! + · · ·
= z3 5
z4 z4 z − 3!
+ z5! + · · ·
2 4
1 1 − z2! + z4! + · · ·
=
z 5 1 − ( z3!2 − z5!4 + · · · )
2 4
1 z z z2 z4 z2 z4
= (1 − + + · · · )(1 + ( − + · · · ) + ( − + · · · )2 + · · · )
z5  2! 4! 3! 5! 3! 5! 
1 1 1 2 1 1 1 1 4
= 5 1 + (− + )z + (− + − + )z + · · · .
z 2! 3! 5! 4! 2!3! (3!)2
Therefore 0 is a pole of order 5 and Res cot
z4
z
= − 5!1 + 1
4!
− 1
2!3!
+ 1
(3!)2
(simplify it).
z=0
21. Singularities of a function 59

(e) Notice that z4sinh z

(1−z 2 )
= z13 sinh z 1
z 1−z 2
, sinh z 1
z 1−z 2
is analytic at 0 and sinh z 1
z 1−z 2
is nonzero at 0.
Therefore 0 is a pole of order 3. Therefore
1 d2 sinh z 1
 
sinh z
Res 4 = (you need to compute it).
z=0 z (1 − z 2 ) 2! dz 2 z 1 − z 2 |z=0

I
Alternately, if 0 < |z| < 1, then
sinh z 1 z3 z5

BH
= (z + + + · · · )(1 + z 2 + z 4 + · · · )
z 4 (1 − z 2 ) z4 3! 5!
 
1 1 3 1 1 5
= 4 z + (1 + )z + ( + + 1)z + · · · .
z 3! 5! 3!
sinh z
Therefore 0 is a pole of z 4 (1−z 2 )
of order 3 and Res z4sinh z
(1−z 2 )
=1+ 1
3!
= 76 .
z=0
2 2
eiz eiz 1 z4 iz 2 z4 2 z4 2
(f) Note that (1−cos z)2
= 4 sin4 z = z 4 4 sin4 z e , 4 sin4 z eiz is analytic at 0 and 4 sin4 z eiz is
2 2 2 2
2
eiz 3 4 2
nonzero at 0. Therefore 0 is a pole of (1−cos z)2
of order 3. Hence Res 3!1 dz
d z
3 ( 4 sin4 z e
iz
)|z=0 .
z=0 2
(3) Use Cauchy’s residue theorem to evaluate the integral of each of these functions around the circle
DA
|z| = 3 in the positive sense: (a) exp(−z)
z2
(b) exp(−z)
(z−1)2
(c) z 2 exp z1 (d) zz+1
2 −2z .

(a) We note that 0 is the only singularity of exp(−z)

z2
lying inside C. If z 6= 0, then
z2 z3
 
1 1 1 1 1 z
2
exp(−z) = 2 1 − z + − + ··· = 2 − + − + ··· .
z z 2! 3! z z 2! 3!

Thus the residue of exp(−z) at 0 is −1. By Cauchy’s Residue Theorem |z|=3 exp(−z)
R
z2 z2
dz =
2πi(−1) = −2πi.
(b) We see that 1 is the only singularity of exp(−z)
(z−1)2
lying inside C. If 0 < |z − 1| < 1, then
 
exp(−z) exp(−(z − 1) − 1) 1 1 1 1 z−1
= = − + − + ··· .
PA

(z − 1)2 (z − 1)2 e (z − 1)2 (z − 1) 2! 3!

exp(−z)
Therefore the residue of (z−1)2
at 1 is − 1e . Using Cauchy’s Residue Theorem we have
Z
exp(−z) 1 2πi
2
dz = 2πi(− ) = − .
(z − 1) e e
|z|=3

(c) The only singularity 0 of z 2 exp z1 lies inside C. If z 6= 0, then

1 1 1
z 2 exp = z2 + z + + + ··· .
z 2! z3!
Therefore Resz=0 z 2 exp z1 = 1
and hence |z|=3 z 2 exp z1 dz = 2πi 16 = πi
R
6 3
.
60 CONTENTS

z+1
(d) The singularities 0 and 2 of the function z 2 −2z
lie inside C. If 0 < |z| < 2, then
∞ ∞
z+1 1 1 1 1 1 X zn 1 X zn
= − z − z =− − .
z(z − 2) 21− 2
2z 1 − 2
2 n=0 2n 2z n=0 2n
Notice that − 12 is the coefficient of z1 . Hence Resz=0 zz+1 1
2 −2z = − 2 .

I
Let 0 < |z − 2| < 2. Then
z+1 1 1 1 1 1 1

BH
= + = +
z(z − 2) z − 2 z − 2 2 + (z − 2) z − 2 2(z − 2) 1 + z−2
2
∞ n
1 1 X (z − 2)
= + (−1)n
z − 2 2(z − 2) n=0 2n
∞
3 1 X (z − 2)n−1
= + (−1)n .
2 z − 2 n=1 2n+1
z+1
Therefore Resz=2 z(z−2)
= 23 . Now
Z  
z+1 z+1 z+1
dz = 2πi Res 2 + Res 2 = 2πi.
|z|=3
DA
z 2 − 2z z=0 z − 2z z=2 z − 2z

(4) In each case, write the principal part of the function at its isolated singular point and determine
whether that point is a pole, a removable singular point, or an essential singular point:

(a) z exp( z1 ) z2 (c) sin z

(d) 1
(b) 1+z z (2−z)3

(a) Here 0 is the only isolated singularity of z exp( z1 ). If z 6= 0, then

   
1 1 1 1 1 1
z exp =z 1+ + 2
+ 3
+ ··· = z + 1 + + + ··· .
z z 2!z 3!z 2!z 3!z 2
1
+ 3!z1 2 + · · · = ∞ 1
P
Therefore the principal part of the function at 0 is 2!z n=1 (n+1)!z n . It means
PA

that 0 is an essential singularity of the function z exp( z1 ).

z2
(b) Note that −1 is the only singularity of z+1 . If 0 < |z + 1| < ∞, then
z2 (z + 1 − 1)2 1
= =z+1−2+ .
z+1 z+1 z+1
1 z 2 z 2
Therefore z+1 is the principal part of z+1 and hence −1 is a pole of z+1 of order 1.
sin z
(c) Observe that 0 is the only singularity of z . If 0 < |z| < ∞, then
∞ 2n+1 ∞
sin z 1X n z
X
n z 2n
= (−1) = (−1) .
z z n=0 (2n + 1)! n=0 (2n + 1)!
sin z sin z
Therefore the principal part of z
is 0 and hence 0 is a removable singularity of z
.
22. Evaluation of Improper Integrals 61

1 1 1
(d) Here 2 is the only singularity of (2−z) 3 . We see that (2−z)3 = − (z−2)3 for all 0 < |z −2| < ∞.
1 1 1
Therefore the principal part of (2−z) 3 is − (z−2)3 . Hence 2 is a pole of (2−z)3 of order 3.

(5) Show that the singular point of each of the following functions is a pole. Determine the order m
of that pole and the corresponding residue B.

I
1−cosh z 1−exp(2z) exp(2z)
(a) z3 (b) z4
(c) (z−1)2

BH
(a) Here 0 is the only isolated singularity of 1−cosh
z3
z
. If 0 < |z| < ∞, then
z2 z4 z6 z3
 
1 − cosh z 1 1 z
= 1 − 1 − − − − · · · = − − − − ··· .
z3 z3 2! 4! 6! z2! 4! 6!
Therefore 0 is a simple pole, i.e., m = 1 and Resz=0 1−cosh z3
z
= − 21 .
(b) The point 0 is the only singularity of 1−exp(2z)
z4
. If z 6= 0, then
22 z 2 23 z 3 24 z 4 25 z 5
 
1 − exp(2z) 1
= 4 1 − 1 − 2z − − − − + ···
z4 z 2! 3! 4! 5!
DA
= −2 3 −
1 22 1
−
23 1 24 25
− − z + ··· .
z 2! z 2 3! z 4! 5!
So, 0 is a pole of order 3 and Resz=0 1−exp(2z)
z4
= − 43 .
(c) Here 1 is a singularity of exp(2z)
(z−1)2
. If 0 < |z − 1| < ∞, then

exp(2z) exp(2(z − 1) + 2) e2
= = exp(2(z − 1))
(z − 1)2 (z − 1)2 (z − 1)2
∞
e2 X 2n (z − 1)n
=
(z − 1)2 n=0 n!
e2 2e2 22 e2 23 e2 24 e2
= + + + (z − 1) + (z − 1)2 + · · · .
PA

(z − 1)2 z − 1 2! 3! 4!
exp(2z) exp(2z)
Thus 1 is a pole of (z−1)2
of order 2 and Resz=1 (z−1)2
= 2e2 .

22. Evaluation of Improper Integrals

If f is a continuous function on [0, ∞), then the improper integral of f over [0, ∞) is defined
by
Z∞ ZR
f (x)dx = lim f (x)dx,
R→∞
0 0
provided the limit exists.
62 CONTENTS

If f is continuous on R, then the improper integral of f over R is defined by

Z∞ Z0 ZR2
f (x)dx = lim f (x)dx + lim f (x)dx
R1 →∞ R2 →∞
−∞ −R1 0

I
provided both the limit exist.
If f is continuous on R, then the Cauchy principal value of the integral of f is defined by

BH
Z∞ ZR
P. V. f (x)dx = lim f (x)dx,
R→∞
−∞ −R

when the limit exists.

We notice that if the integral of f over R exists, then the integral of f over R is same as the
Cauchy principal value of the integral.
R∞ RR
The converse is not true. For example consider f (x) = x. Then P.V. −∞ xdx = limR→∞ −R xdx =
RR
0 but limR2 →∞ 0 2 xdx = ∞.
DA
Theorem 22.1 (Jordan’s Lemma). Suppose that a function f is analytic at all points in the upper
half plane y ≥ 0 that are exterior to a circle |z| = R0 . Let CR denote a semicircle z(θ) = Reiθ ,
0 ≤ θ ≤ π, where R > R0 . Suppose that there is MR > 0 such that |f (z)| ≤ MR for all z ∈ CR
and limR→∞ MR = 0. Then
Z
lim f (z)eiaz dz = 0 for all a > 0.
R→∞
CR

Theorem 22.2. Suppose that

(1) a function f has a simple pole at a point x0 on the real axis, with a Laurent series represen-
tation in a punctured disk 0 < |z − x0 | < R2 and with residue B0 ;
(2) Cρ denotes the upper half of a circle |z − x0 | = ρ, where ρ < R2 and the clockwise direction
PA

is taken.
Then Z
lim f (z)dz = −B0 πi.
ρ→0
Cρ

22.1. Evaluation of Integrals of functions not having singularity on the real line.
Example 22.4. We evaluate the following integrals.
R∞ 2 −1
(1) 0 x42x +5x2 +4
dx.
2 −1 2z 2 −1
Let f (z) = z42z+5z 2 +4
= (z2 +1)(z 2 +4) . We notice that ±i and ±2i are simple poles of f . For R >

2, consider the simple contour CR made up of the semicircle CR0 : z(θ) = Reiθ (0 ≤ θ ≤ π)
22. Evaluation of Improper Integrals 63

followed by the segment joining the points −R and R. Then f is analytic within and CR except
the points i and 2i, which are interior to CR . It follows from Cauchy’s Residue Theorem that

2z 2 − 1
Z
dz = 2πi[Res f + Res f ].
z 4 + 5z 2 + 4 z=i z=2i

I
CR

2z −1 2 i
Since i and 2i are simple poles of f , Resz=i f = limz→i (z − i) (z−i)(z+i)(z 2 +4) = and

BH
2
2z 2 −1
Resz=2i f = limz→2i (z − 2i) (z−2i)(z+2i)(z 2 +1) = − 3i4 . Thus

2z 2 − 1
Z
i 3i π
4 2
dz = 2πi[ − ] = .
z + 5z + 4 2 4 2
CR

If z ∈ CR0 , i.e., |z| = R, then |(z 2 + 4)(z 2 + 1)| = |z 2 + 4||z 2 + 1| ≥ (R2 − 4)(R2 − 1) and
2 −1 2R2 +1 0 0
|2z 2 − 1| ≤ 2R2 + 1. Thus z42z +5z 2 +4
≤ (R2 −1)(R 2 −4) for all z ∈ CR . The length of CR is πR.

Thus
DA
2z 2 − 1 2R2 + 1
Z
dz ≤ πR → 0 as R → ∞.
z 4 + 5z 2 + 4 (R2 − 1)(R2 − 4)
0
CR

Therefore

ZR
2z 2 − 1 2z 2 − 1 2x2 − 1
Z Z
π
= dz = dz + dx.
2 z 4 + 5z 2 + 4 z 4 + 5z 2 + 4 x4 + 5x2 + 4
CR 0
CR −R
PA

R∞ 2 −1 2x2 −1
Applying limit R → ∞, we obtain −∞ x42x +5x
π
2 +4 dx = 2 . Since x4 +5x2 +4 is an even function,
R ∞ 2x2 −1 π
2 +4 dx = 4 .
R0∞ x4 +5x
dx ∞
and −∞ (x2 +1)(xx2 +2x+2) dx.
R
(2) −∞ x2 +2x+2
1
Let f (z) = z2 +2z+2 . We see that z 2 + 2z + 2 = (z + 1)2 + 1. Thus the zeros of z 2 + 2z + 2
1
are −1 ± i. The complex numbers −1 ± i are simple poles of f . So, Resz=−1+i z2 +2z+2 =
1 i
√
limz→−1+i (z + 1 − i) (z+1−i)(z+1+i) = − 2 . For R > 2, consider CR described above. Then
by Cauchy’s Residue Theorem

−i
Z
dz
= 2πi = π.
z 2 + 2z + 2 2
CR
64 CONTENTS

√
If z ∈ CR0 , then |z 2 +2z+2| = |(z+1−i)(z+1+i)| ≥ (R−|−1+i|)(R−|−1−i|) = (R− 2)2 .
The length of CR0 is πR. Thus C 0 z2 +2z+2
dz
≤ πR (R−1√2)2 → 0 as R → ∞. Now
R
R

Z Z ZR
dz dz dx
π= 2
= 2
+ .
z + 2z + 2 z + 2z + 2 x2 + 2x + 2

I
CR 0
CR −R
R∞ dx
Applying limit R → ∞, we get −∞ x2 +2x+2 = π.

BH
z
For the second integral, take f (z) = (z2 +1)(z2 +2z+1) . Then −1 ± i and ±i are simple poles of f .
z z
We see that Resz=−i+i f = limz→−1+i (z+1−i) (z+1−i)(z+1+i)(z 2 +1) = limz→−1+i (z+1+i)(z 2 +1) =
3i−1 z
10
and Resz=i f = limz→i (z − i) (z−i)(z+i)(z 2 +2z+2) =
√
limz→i (z+i)(z2z+2z+2) = −2i+1
10
. For R > 2, consider CR described above. Then by Cauchy’s
Residue Theorem Z
dz i π
= 2πi = − .
z 2 + 2z + 2 10 5
CR

If z ∈ CR0 , 2
then |z + 2z + 2| = |(z + 1 − i)(z + 1 + i)| ≥ (R − | − 1 + i|)(R − | − 1 − i|) =
√ 2
DA
(R − 2) and |z 2 + 1| ≥ R2 − 1. The length of CR0 is πR. Thus C 0 (z2 +1)(z
R zdz
2 +2z+2) ≤
R
R √
πR (R2 −1)(R− 2)2
→ 0 as R → ∞. Now
Z Z ZR
π zdz zdz xdx
− = = + .
5 (z + 1)(z 2 + 2z + 2)
2 (z + 1)(z 2 + 2z + 2)
2 (x2 + 1)(x2 + 2x + 2)
CR 0
CR −R
R∞ xdx π
Applying limit R → ∞, we get −∞ (x2 +1)(x 2 +2x+2) = − 5 .
R ∞ dx
(3) 0 x4 +1 .
π 3π 5π 7π
Let f (z) = z41+1 . Then z0 = ei 4 , z1 = ei 4 , z2 = ei 4 and z3 = ei 4 are simple poles of f .
For i = 0, 1, 2, 3, Resz=zk = limz→zk z−z k
z 4 +1
zk
= limz→zk 4z13 = 4z13 = 4z zk zk
4 = −4 = − 4 . Thus
PA

k k
Resz=z0 f + Resz=z1 f = − 14 (z0 + z1 ) = − 41 √2i2 = − 2√i 2 . For R > 1, consider CR described
above. Then z0 and z1 are interior to CR . By Cauchy’s Residue Theorem
−i
Z
dz π
4
= 2πi √ = √ .
z +1 2 2 2
CR

If z ∈ CR0 , then |z 4 + 1| =≥ R4 − 1. The length of CR0 is πR. Thus dz

≤ πR R41−1 → 0
R
0 z 4 +1
CR
as R → ∞. Now
Z Z ZR
π dz dz dx
√ = 4
= 4
+ 4
.
2 z +1 z +1 x +1
CR 0
CR −R
22. Evaluation of Improper Integrals 65

R∞ dx π
R∞
Applying limit R → ∞, we get −∞ x4 +1
= √
2
. Since x41+1 is an even function, 0
dx
x4 +1
= π
√
2 2
.

22.2. Evaluation of Integrals of functions containing trigonometric functions.

Examples 22.6. We find the (Cauchy principal) value of the following integrals
R∞
(1) 0 (xcos 3x

I
2 +1)2 dx.
R ∞ i3x i3z
We observe that the required integral is the real part of 0 (xe2 +1)2 dx. Let f (z) = (z2e+1)2 .

BH
Then ±i are singularities of f . In fact, they are poles of f order 2. Therefore Resz=i f =
1 d i3z
1! dz
[(z − i)2 (z2e+1)2 ]|z=i = − ei3 . For R > 1, consider the simple contour CR made up of the
semicircle CR0 : z(θ) = Reiθ (0 ≤ θ ≤ π) followed by the segment joining the points −R and
R. Since f is analytic within and on CR except for the pole i, which is interior to CR , we get
ei3z ei3z
Z
2π
dz = 2πi Res = .
(z 2 + 1)2 z=i (z 2 + 1)2 e3
CR

If z ∈ CR0 , i.e., |z| = R, then |z 2 + 1| ≥ R2 − 1. Also, |ei3z | = e−3y ≤ 1 for all z = x + iy ∈ CR0 .
The length of CR0 is πR. Therefore
DA
ei3z
Z
1
2 2
dz ≤ πR 2 → 0 as R → ∞.
(z + 1) (R − 1)2
0
CR

So,
ZR
ei3z ei3z ei3x
Z Z
2π
= dz = dz + dx.
e3 (z 2 + 1)2 (z 2 + 1)2 (x2 + 1)2
CR 0
CR −R

R∞ i3x
Applying limit R → ∞, we have −∞ (xe2 +1)2 dx = 2π . Comparing real and imaginary parts,
PA

R ∞ cos 3x e3 R∞
we get −∞ (x2 +1)2 dx = e3 . Since (x2 +1)2 is an even function, 0 (xcos
2π cos 3x 3x π
2 +1)2 dx = e3 .
R∞
(2) 0 (x2 +acos x
2 )(x2 +b2 ) dx, where a > b > 0.

eiz
Take f (z) = (z 2 +a2 )(z 2 +b2 )
. Then ±ia and ±ib are simple poles of f . Note that Resz=ia f =
eiz eia e−a −b
limz→ia (z − ia) (z2 +a2 )(z2 +b2 ) = limz→ia (z+ia)(z and Resz=ib = 2ib(ae 2 −b2 ) .
2 +b2 ) = − 2ia(a2 −b2 )

For R > a, consider the simple contour CR made up of the semicircle CR0 : z(θ) = Reiθ (0 ≤
θ ≤ π) followed by the segment joining the points −R and R. Since f is analytic within and
on CR except for the pole ia and ib, which are interior to CR , we get
 −b
eiz e−a
Z 
π e
dz = 2πi[Res f + Res f ] = 2 − .
(z 2 + a2 )(z 2 + b2 ) z=ia z=ib a − b2 b a
CR
66 CONTENTS

If z ∈ CR0 , i.e., |z| = R, then |z 2 + a2 | ≥ R2 − a2 and |z 2 + b2 | ≥ R2 − b2 . Also, |eiz | = e−y ≤ 1

for all z = x + iy ∈ CR0 . The length of CR0 is πR. Therefore

eiz
Z
1
2 2 2 2
dz ≤ πR 2 → 0 as R → ∞.
(z + a )(z + b ) (R − a )(R2 − b2 )
2

I
0
CR

By similar arguments given above we get

BH
Z∞
e−b e−a
 
cos x π
2 2 2 2
dx = 2 − .
(x + a )(x + b ) a − b2 b a
0
R∞ x sin x
(3) −∞ (x2 +1)(x2 +4)
dx.
zeiz
Let f (z) = (z2 +1)(z 2 +4) . Then ±i and ±2i are simple poles of f . We see that Resz=i f =
zeiz ie−11 ze iz
2ie −2
limz→i (z+i)(z2 +4) = 2i(3) = 6e and Resz=2i f = limz→2i (z2 +1)(z+2i) = (−3)4i = − 6e12 . For
R > 2, let CR be the simple closed contour consisting of the semicircle CR0 : z(θ) = Reiθ ,
DA
0 ≤ θ ≤ π followed by the segment joining −R to R. Then f is analytic within and on CR
except the points i and 2i, which are interior to CR . Therefore by Cauchy’s Residue Theorem

zeiz πi(e − 1)
Z
1 1
2 2
dz = 2π[Res f + Res f ] = 2πi( − 2 ) = .
(z + 1)(z + 4) z=i z=2i 6e 6e 3e2
CR

Thus
Z∞
zeiz xeix πi(e − 1)
Z
dz + dx = . (22.6.1)
(z 2 + 1)(z 2 + 4) 2 2
(x + 1)(x + 4) 3e2
0
CR −∞
PA

z
Let g(z) = (z2 +1)(z 2 +4) . Then g is analytic at every point on the upper half plane which are

exterior to the circle |z| = 2. If z ∈ CR0 , i.e., |z| = 1, then |g(z)| ≤ (R2 −1)(R R
2 −4) → 0 as
R ze iz
R → ∞. It follows from Jordan’s Lemma that limR→∞ C 0 (z2 +1)(z 2 +4) dz = 0. Applying limit
R
R∞ xeix πi(e−1) R∞ x cos x
R → ∞ in the equation (22.6.1), we get −∞ (x2 +1)(x2 +4) dx = 3e2 , i.e., −∞ (x2 +1)(x 2 +4) dx+
R∞ x sin x π(e−1)
i −∞ (x2 +1)(x2 +4) dx = i 3e2 . Hence

Z∞
x sin x π(e − 1)
dx = .
(x2 + 1)(x2 + 4) 3e2
−∞
R 2π dθ
(4) 0 1+a sin θ
, where −1 < a < 1.
22. Evaluation of Improper Integrals 67

dz
Take z = eiθ . Then dz = eiθ idθ = izdθ, i.e., dθ = iz
. Therefore

Z2π Z2π Z
dθ 1 1 dz
= iθ −iθ dθ = z+ 1
1 + a sin θ 1+ a e −e2i 1+ a 2iz iz
0 0 |z|=1

I
Z
2/a
= dz.
z2 + (2i/a)z − 1

BH
|z|=1
 √   √ 
R 2/a −1+ 1−a2 −1− 1−a2
Let f (z) = |z|=1 z 2 +(2i/a)z−1
. Then z1 = a
i and z2 = a
i. Note that
|z2 | > 1. Since z1 and z2 are zeros of z 2 + (2i/a)z − 1, z1 z2 = −1. Since |z2 | > 1, |z1 | < 1.
2/a 1
We see that Resz=z1 f = limz→z1 (z − z1 ) z2 +(2i/a)z−1 = i√1−a 2 . Thus by Cauchy’s Residue

Theorem
Z2π Z
dθ 2/a 2π
= dz = 2πi Res f = √ .
1 + a sin θ z2 + (2i/a)z − 1 z=z1 1 − a2
0 |z|=1
Rπ dθ
DA
(5) −π 1+sin2 θR
.
πdθ
Rπ dθ
Rπ 1
Rπ dθ
Note that −π 1+sin 2θ = 2 0 1+sin2 θ
=2 0 1+ 1−cos 2θ =4 0 3−cos 2θ
. Taking t = 2θ in the
2
last integral we get
Zπ Z2π
dθ dθ
=2 .
1 + sin2 θ 3 − cos θ
−π 0

Now
Z2π Z Z
dθ dz/iz 2 dz
= 1 1 = − .
3 − cos θ 3 − 2 (z + z ) z2 − 6z + 1
PA

i
0 |z|=1 |z|=1
√ √ √
we see that z 2 − 6z + 1 = 0 if and only if z = 6±42 2 = 3 ± 2 2. Therefore 3 − 2 2 is a
1 1
simple pole of z2 −6z+1 lying inside |z| = 1 and Resz=3−2√2 z2 −6z+1 = limz→3−2√2 (z − 3 +
√ 1 1 1
2 2) z2 −6z+1 = 3−2√2−3−2 √ = − √ . Thus
2 4 2

Z2π Z Z
dθ dz/iz 2 dz
= 1 1 = −
3 − cos θ 3 − 2 (z + z ) i z2 − 6z + 1
0 |z|=1 |z|=1
2 1 π
= − 2πi(− √ ) = √ .
i 4 2 2
68 CONTENTS

Thus
Zπ Z2π √
dθ dθ π
=2 = 2 √ = 2π.
1 + sin2 θ 3 − cos θ 2
−π 0

I
R 2π cos2 3θ
(6) 0 5+4 cos 2θ
dθ.
R π cos2 3θ R π cos2 3θ R 2π cos2 3θ R 2π cos2 3θ
Take t = θ+π in 0 5−4 cos 2θ
dθ. Then 0 5−4 cos 2θ
dθ = π 5−4 cos 2θ
dθ. Therefore 0 5−4 dθ =
R π 1+cos 6θ cos 2θ

BH
R π cos2 3θ R π 1+cos 6θ
2 0 5−4 cos 2θ dθ = 0 5−4 cos 2θ dθ. Take t = 2θ in the last integral. Then 0 5−4 cos 2θ dθ =
1 2π 1+cos 3θ
R
2 0 5−4 cos θ
dθ. Thus
 
Z2π 2 Z2π Z2π i3θ
cos 3θ 1 1 + cos 3θ 1 1+e
dθ = dθ = Re  dθ .
5 − 4 cos 2θ 2 5 − 4 cos θ 2 5 − 4 cos θ
0 0 0

R 2π 1+ei3θ
We evaluate 0 5−4 cos θ
dθ. Take eiθ = z, then

Z2π
1 + ei3θ
dθ =
DA
Z
1 + z3
1
dz
=−
1
Z
1 + z3
dz
5 − 4 cos θ 5 − 2(z + z ) iz i 2z 2 − 5z + 2
0 |z|=1 |z|=1
" #
1 + z3 1 + z3
Z
1 1
= − dz = − 2πi Res1
2i (z − 12 )(z − 2) 2i 1
z= 2 (z − 2 )(z − 2)
|z|=1
" #
1 1 + z3 3π
= − 2πi lim1 = .
2i z→ 2 z − 2 4
R 2π cos2 3θ
Thus 0 5+4 cos 2θ
dθ = 3π
8
Rπ
(7) 0 (sin2n θ)dθ.
PA

Rπ Rπ R 2π Rπ
Take t = θ + π in 0 sin2n θdθ. Then 0 sin2n θdθ = π sin2n θdθ. Thus 0 sin2n θdθ =
1 2π
sin2n θdθ. Now
R
2 0

Z2π 2n
z − z1 (z 2 − 1)2n
Z  Z
2n dz 1
sin θdθ = = 2n dz.
2i iz i2 (−1)n z 2n+1
0 |z|=1 |z|=1

(z 2 −1)2 n
We see that 0 is a pole of z 2n+1
of order 2n + 1. Thus

(z 2 − 1)2 n 1 d2n 2 2n
1 d2n
 
2n+1 (z − 1)
Res = z = [(z 2 − 1)2n ]|z=0 .
z=0 z 2n+1 (2n)! dz 2n z 2n+1 |z=0 (2n)! dz 2n
22. Evaluation of Improper Integrals 69

2
d2n 2n
(2n)! = (−1)n ((2n)!)


We observe that dz 2n
[(z 2 − 1)2n ]|z=0 = (−1)n n (n!)2
. Therefore

Z2π
1 (2n)! 1 (2n)!
sin2n θdθ = 2πi(−1)n = π 2n−1 .
i22n (−1)n (n!)2 2 (n!)2
0

I
Thus
Zπ Z2π
1 π(2n)!

BH
2n
sin θdθ = sin2n θdθ = .
2 22n (n!)2
0 0

22.3. Evaluation of integrals of functions having singularity on the real line.

Examples 22.8. We shall evaluate the following integrals.

R∞
(1) 0 sinx x dx
iz
Let f (z) = ez . Then 0 is simple pole of f and Resz=0 f = limz→0 eiz = 1. Let 0 < ρ < R.
Let CR be the simple closed contour consisting of the semicircle CR0 : z(θ) = Reiθ , 0 ≤ θ ≤ π
followed by path from L1 from −R to −ρ followed by negatively oriented semicircle Cρ0 with
center
DA
R 0 and radius ρ followed by the path L2 from ρ to R. Since f is analytic within and on
CR , CR f (z)dz = 0, i.e.,
Z Z Z Z
f (z)dz + f (z)dz + f (z)dz + f (z)dz = 0, i.e.,
0
CR L1 Cρ0 L2

Z−ρ ZR
eiz eix eiz eix
Z Z
dz + dx + dz + dx = 0.
z x z x
0
CR −R Cρ0 ρ

Now
PA

Z−ρ ZR ZR ZR ZR
eix eix e−ix eix sin x
dx + dx = − dx + dx = 2i dx.
x x x x x
−R ρ ρ ρ ρ

Therefore the above equation may be written as

ZR
eiz eiz
Z Z
sin x
2i dx = − dz − dz. (22.8.1)
x z z
ρ 0
CR Cρ0

eiz eiz
= −πi(1) = −πi. If z ∈ CR0 , i.e., |z| = R, then
R
Since Resz=0 z
= 1, limρ→0 Cρ0 z
dz
1 1 iz
→ 0 as R → ∞. Hence by Jordan’s Lemma limR→∞ C 0 ez dz = 0. Applying limit
R
|z|
= R R
70 CONTENTS

R∞ sin x
ρ → 0 and R → ∞ in the equation (22.8.1), we get 2i 0 x
dx = πi. Therefore
Z∞
sin x π
dx = .
x 2
0

I
Let m > 0. Then
Z∞ Z∞ Z∞
sin mx sin mx sin x π

BH
dx = m dx = dx = .
x mx x 2
0 0 0
R∞
(2) Ą 0 cos ax−cos
x2
bx
dx, where a, b ≥ 0.
eiaz −eibz
Let f (z) = z2
. If z 6= 0, then
eiaz − eibz (iaz)2 (ibz)2
 
1
= 2 1 + iaz + + · · · − 1 − ibz − − ··· .
z2 z 2! 2!
Therefore 0 is a simple pole of f and Resz=0 f = i(a−b). Let 0 < ρ < R. Let CR be the simple
closed contour consisting of the semicircle CR0 : z(θ) = Reiθ , 0 ≤ θ ≤ π followed by path from
DA
L1 from −R to −ρ followed by negatively oriented semicircle Cρ0 with center R 0 and radius ρ
followed by the path L2 from ρ to R. Since f is analytic within and on CR , CR f (z)dz = 0, i.e.,
Z Z Z Z
f (z)dz + f (z)dz + f (z)dz + f (z)dz = 0, i.e.,
0
CR L1 Cρ0 L2

Z−ρ ZR
eiaz − eibz eiax − eibx eiaz − eibz eiax − eibx
Z Z
dz + dx + dz + dx = 0. (22.8.2)
z2 x2 z2 x2
0
CR −R Cρ0 ρ

R −ρeiax −eibx
RR e−iax −e−ibx
One has dx = dx. Therefore
PA

−R x2 ρ x2

ZR ZR ZR
e−iax − e−ibx eiax − eibx cos ax − cos bx
dx + dx = 2 dx.
x2 x2 x2
ρ ρ ρ

eiaz −eibz eiaz −eibz

R
BY Jordan’s Lemma, limR→∞ z2 0
CR
dz = 0. Since 0 is a simple pole of z2
and its
R eiaz ibz
residue at 0 is i(a − b), limρ→0 Cρ z−e 2 dz = −iπi(a − b) = π(a − b). Applying limit
R → ∞ and ρ → 0 in equation (22.8.2) we get
Z∞
cos ax − cos bx π(b − a)
2
dx = .
x 2
0
23. Bilinear Transformations 71

R∞ sin2 x
(3) 0 x2
dx.
We may write it as
Z∞ 2 Z∞ Z∞
sin x 1 1 − cos 2x 1 cos 0x − cos 2x 1 π2 π
2
dx = 2
dx = 2
dx = = .
x 2 x 2 x 2 2 2
0 0 0

I
23. Bilinear Transformations

BH
Definition 23.1. The transformation
az + b
w(z) = , (ad − bc 6= 0),
cz + d
where a, b, c and d are complex constants, is called a linear fractional transformation, or Mobius
transformation or bilinear transformation.
Exercise 23.2. The set of all Mobius transformations is a group (with multiplication defined as com-
position).
The map w(z) = az+b is analytic on C \ {− dc } and − dc is an isolated singularity of w. After
simplifying w(z) = az+bcz+d
cz+d
DA
, we may write z = −dw+bcw−a
. Therefore the inverse transformation of az+b
cz+d
is
−dw+b
cw−a
.
Given a 6= 0, then map Da (z) = az is called a dilation. Given α ∈ C, the map Tα (z) = z + α
is called a translation. The map R(z) = z1 is called a reciprocation.
If c = 0, then w(z) = ad z + db , i.e., w(z) = T b ◦ D ad (z).
d

If c 6= 0, then w(z) = ad − (ad−bc)/c

cz+d
= (T ad ◦ D−(ad−bc)/c ◦ R ◦ Td ◦ Dc )(z).
Thus any bilinear map can be written as a composition of dilation, translation and reciprocation.
(z0 −z2 )(z1 −z3 )
Definition 23.3. The expression (z 0 −z3 )(z1 −z2 )
is called the cross ratio of the four distinct z1 , z1 , z2 , z3
in the extended complex plane and it is denoted by (z0 , z1 , z2 , z3 ).
Note that if zi = ∞ for some i, then we shall not write the term containing zi in the cross ratio.
PA

−z3
For example, if z0 = ∞, then (z0 , z1 , z2 , z3 ) = zz11 −z 2
.
Theorem 23.4. Mobius transformations preserve cross ratios; that is, if T is a Mobius transforma-
tion, then (T z0 , T z1 , T z2 , T z3 ) = (z0 , z1 , z2 , z3 ) for all distinct points z0 , z1 , z2 , z3 .
It follows from above Theorem that a mobius transformation can be uniquely determine by
knowing its image on three distinct points.
Examples 23.5.
(1) We find the linear fractional transformation that maps the points z1 = 2, z2 = i, z3 = −2 onto
the points w1 = 1, w2 = i, w3 = −1. Let T be the required linear fractional transformation.
Then (T z0 , T z1 , T z2 , T z) = (z0 , z1 , z2 , z) for all z which is not equal to any of z1 , z2 and
72 CONTENTS

(1+1)(i−T z)
z3 . Thus (1, i, −1, T z) = (2, i, −2, z), i.e., (1−T z)(i+1)
= (2+2)(i−z)
(2−z)(i+2)
. Simplifying it we have
3z+2i
T z = iz+6 .
(2) We find the bilinear transformation that maps the points z1 = ∞, z2 = i, z3 = 0 onto the
points w1 = 0, w2 = i, w3 = ∞. Let T be the required linear fractional transformation. Then
(T z0 , T z1 , T z2 , T z) = (z0 , z1 , z2 , z) for all z which is not equal to any of z1 , z2 and z3 . Then

I
i−T z
(0, i, ∞, T z) = (∞, i, 0, z), i.e., 0−T z
= i−z
i−0
. Simplification gives T (z) = − z1 .

BH
Definition 23.6. A point z0 of an extended complex plane is called a fixed point of a linear fractional
transformation T is T (z0 ) = z0 .
Examples 23.7. We shall find fixed points of the linear fractional transformations below.
z−1
(1) w = z+1 .
z−1
If z is a fixed point of w, then z = w(z) = z+1 . This gives z 2 + z = z − 1, i.e., z = ±i.
(2) w = 6z−9 z
.
If z is a fixed point of w, then z = w(z) = 6z−9
z
, i.e., (z − 3)2 = 0, i.e., z = 3.
DA
PA