Endre Pap Auth - Complex Analysis Through Exambookzz-Org PDF

Complex Analysis through Examples and Exercises
Kluwer Text in the Mathematical Sciences

VOLUME21
A Graduate-Level Book Series
The titfes published in this series are listed at the end 0/ this vofume.
Complex Analysis
through Examples
and Exercises
by
Endre Pap
Institute of Mathematics,
University of Novi Sad,
Novi Sad, Yugoslavia
SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

A c.I.P. Catalogue record for this book is available from the Library of Congress.
ISBN 978-90-481-5253-7 ISBN 978-94-017-1106-7 (eBook)

DOI 10.1007/978-94-017-1106-7
Printed an acid-free paper
AII Rights Reserved

© 1999 Springer Science+Business Media Dordrecht
Originally published by Kluwer Academic Publishers in 1999
N o part of the material protected by this copyright notice may be reproduced or
utilized in any form or by any means, electronic or mechanical,
inc1uding photocopying, recording or by any information storage and
retrieval system, without written permission from the copyright owner.
Contents
Contents v
Preface IX
1 The Complex Numbers 1

1.1 Algebraic Properties 1
1.1.1 Preliminaries 1
1.1.2 Examples and Exercises 2
1.2 The Topology of the Complex Plane 32
1.2.1 Preliminaries ...... 32
2 Sequences and series 37

2.1 Sequences ..... 37
2.2 Series . . . . . . . . . 44
3 Complex functions 53
3.1 General Properties 53
vi CONTENTS
3.2 Special Functions .. 64

3.3 Multi-valued functions 68
4 Conformal mappings 73
4.1 Basics ....... 73
4.2 Special mappings .. 74
5 The Integral 103

5.1 Basics • • • 0 •• 103
6 The Analytic functions 129

6.1 The Power Series Representation 129
6.1.1 Preliminaries ...... 129
6.2 Composite Examples ...... 162
7 Isolated Singularities 171

7.1 Singularities . . . . 171
7.2 Laurent series .... 177
CONTENTS vii
7.2.2 Examples and Exercises . . . . . . . . . . . . . . . . . . . . . 179
8 Residues 191
8.1 Residue Theorem · 191
8.1.1 Preliminaries · 191
8.1.2 Examples and Exercises · 193
8.2 Composite Examples . . . . . . .206
9 Analytic continuation 227

9.1 Continuation ... .227
9.1.1 Preliminaries .227
9.1.2 Examples and Exercises .228
10 Integral transforms 255

10.1 Analytic Functions Defined by Integrals. .255
10.1.1 Preliminaries . . . . . . .255
10.1.2 Examples and Exercises .256
11 Miscellaneous Examples 313
Bibliography 333
List of Symbols 335
Index 336
Preface
The book Complex Analysis through Examples and Exercises has come out from the
lectures and exercises that the author held mostly for mathematician and physists .
The book is an attempt to present the rat her involved subject of complex analysis
through an active approach by the reader. Thus this book is a complex combination
of theory and examples.
Complex analysis is involved in all branches of mathematics. It often happens
that the complex analysis is the shortest path for solving a problem in real circum-
stances. We are using the (Cauchy) integral approach and the (Weierstrass) power
se ries approach .
In the theory of complex analysis, on the hand one has an interplay of several
mathematical disciplines, while on the other various methods, tools, and approaches.
In view of that, the exposition of new notions and methods in our book is taken step
by step. A minimal amount of expository theory is included at the beinning of each
section, the Preliminaries, with maximum effort placed on weil selected examples
and exercises capturing the essence of the material. Actually, I have divided the
problems into two classes called Examples and Exercises (some of them often also
contain proofs of the statements from the Preliminaries). The examples contain
complete solutions and serve as a model for solving similar problems given in the
exercises. The readers are left to find the solution in the exercisesj the answers, and,
occasionally, some hints, are still given. Special sections contain so called Composite
Examples which consist of combinations of different types of examples explaining,
altogether, some problems completely and giving to the reader an opportunity to
check his entire previously accepted knowledge.
The necessary prerequisites are a standard undergraduate course on real func-
tions of real variables. I have tried to make the book self-contained as much as
possible. For that reason, I have also included in the Preliminaries and Examples
some of the mathematical tools mentioned.
The book is prepared for undergraduate and graduate students in matheniatics,
physics, technology, economics, and everybody with an interest in complex analysis.
We have used for some calculations and drawings the mathematical software
ix
x PREFACE
packages Mathematica and Scientific Work Place v2.5.
I am grateful to Academician Bogoljub Stankovic for a long period of collabo-
ration on the subject of the book, to Prof. Arpad TakaCi for his numerous remarks
and advice about the text, and to Ivana Stajner for reading some part of the text.
I would like to express my thanks to MarCicev Merima for typing the majority of
the manuscript. It is my pleasure to thank the Institute of Mathematics in Novi
Sad for working conditions and financial support. I would like to thank Kluwer
Academic Publishers, especially Dr. Paul Roos and Ms. Angelique Hempel for their
encouragement and patience.
Novi Sad, June 1998 ENDRE PAP

Chapter 1
The Complex N umbers
1.1 Algebraic Properties
1.1.1 Preliminaries
The field of complex numbers Cis the set of all ordered pairs (a, b) where a and b
are real numbers and where addition and multiplication are defined by:
(a, b) + (c, d) = (a + c, b + d)
(a, b)(c, d) = (ac - bd, bc + ad).

We will write a for the complex number (a,O). In fact, the mapping a 1-+ (a,O)
defines a field isomorphism of IR into C, hence we may consider IR as a subset of C. If
we put z = (0,1), then (a, b) = a + bz. For z = a + zb we put Re z = a and Imz = b.
Real numbers are associated with points on the x-axis and called the real axis .
Purely imaginary numbers are associated with points on the y-axis and called the
imaginary axis .
Note that z2 = -1, so the equation z2+ 1 = 0 has a root in C. If z = x+zy (x, y E
IR), then we define
Izl = Jx 2 + y2
to be the absolute value of z and z = x - zy is the conjugate of z. We have Izl 2 = zz
and the triangle inequality
Iz + wl s:; Izl + Iwl (z,w E C).
By the definition of complex numbers, each z in C can be identified with a unique

point (Rez,Imz) in the plane IR2 •
E. Pap, Complex Analysis through Examples and Exercises

© Springer Science+Business Media Dordrecht 1999
2 CHAPTER 1. THE COMPLEX NUMBERS
The argument of z i- 0, denoted by arg z, is the angle 0 (modulo 271") between

the vector from the origin to z and positive x-axis, i.e., which satisfy
Rez . 0 Imz
cosO = Tz! and sm = Tz!.
The point z = x+zy =I- 0 has polar coordinates (r, 0) : x = r cos 0, y = r sin O. Clearly
r = Izl and 0 is the angle between the positive real axis and the line segment from
° to z. Notice that 0 plus any multiple of 271" can be substituted for 0 in the above
equations. The angle 0 is called the argument of z and is denoted by 0 = argz.
Let Zl = rl ( cos Ol + zsin Ol and Z2 = r2 ( cos O2 + i sin ( 2 ) then
In particular, if z = r( cos 0 + z sin 0), then
zn = rn(cos(nO) + zsin(nO». (1.1 )
As a special case of (1.1) we obtain DeMoivre's formula:
(cos 0 + z sin Ot = cos nO + z sin nO. (1.2)
The n-th root of z = r( cos 0 + zsin 0) are
Zk = y'r(cos o+n2h + zsm

. 0 + 2h
n
)
for k = 0,1, ... , n - 1.

The complex number z = r( cos O+z sin 0) also has the exponential representation
z = r exp( zO).
For more explanations see the chapter on power series.
1.1.2 Examples and Exercises

Example 1.1 Find the real numbers p and q such that the complex numbers
1
z = p + zq, W = P + z- be equal.
q
1.1. ALGEBRAIC PROPERTIES 3
Solution. We have that z = w is equivalent with Rez = Rew and Imz =

Immw. Therefore p = p and q = !,
q
pEllt, q2 = 1, i.e., ql = 1, q2 = -1 and pEllt.
Example 1.2 For z = 1 + z find w such that the real parts of the following numbers
are equal to zero a)z + Wj b) Z· Wj c) fuj d) ~.
Solution. Let z = 1 + z = (1,1) and w = x + zy = (x, y). Then we have

a)
Re(z + w) = ° {=} 1 + x = 0.
Hence x = -1 and y E llt is arbitrary.
b)
Re (z . w) = °
{=} Re (x - y + z( x + y)) = 0.
Hence x - y = 0. Therefore w is given by w = x + zx = x(l + z) = x· z, for arbitrary
xE llt.
c) We have fu = 0. Hence
z 1 + z x - zy x + y + z( x - y)
- = - - . - - = ---'--:--'--::--~
W X + zy x - zy x 2 + y2
Therefore
Re(-=-) =o{=}
w x +y
~+Y2
=O:::}x=-y,xi=0.
Finally we have w = x(l - z) for x E llt and x i= 0.

d) From Re °
(~) = it follows that for every x E llt :
w x + zy 1 - z x + y + z(x - y)
-=--.--=
z l+z 1-z 2
Hence W = x(l - z) for every x E llt.
Example 1.3 Prove that
Hint. It is easy to prove the case n = 2 and then use mathematical induction
to prove the general case.
Example 1.4 Find for z = 1 + 2z the following numbers

a) zn j b) l/zj c) l/z n j d) Z2 + 2z + 5 + z.
Solution. a) We have
zn (1 + 2zt
E(~)(2z)k
IJ-l)k(n)22k+Z 't(-l)k(
[.!!.) [n-l)
n )22k+1,
k=O 2k k=O 2k +1
where [xl is the greatest integer part of x.
b) We have
-1 = -1- . -1 - 2z
-=- 1 - 2z
- = -1 - -2 z.
z 1 + 2z 1 - 2z 5 5 5
e) Sinee ~
zn
= (~)n,
Z
we have by b)
(1
1 (1
+ 2z)n = 5" - 5" z
2)n 1
= 5n (1 - 2z) .
n
Applying the same proeeclure as in a) we obtain (using that the imaginary part of
z is -2) :
cl) Putting z = 1 + 2z in Z2 + 2z + 5 + z, we obtain 4 + 9z.
Example 1.5 Find the positions of the following points in the complex plane:
a + za, a - za, -a + za, -a - za fOT a E lR?
Solution. Using the trigonometrie representation (p,O) wc obtain

V2
a + za V2aT + zv'2a TV2
(v'2' a '4~) '
a - za V2a . -
v'2 - zV2a-
V2
2 2
(v'2a , -~)
4'
-a + za (V2a, 3:),
-a - za (V2a, 3:).
Hence the four given points are the corners of the square in the circle with the center
at origin and the radius J21a1-
Example 1.6 Which subsets of the complex plain correspond to the complex num-
bers with the following properties:
a) Rez = Imz; b) Rez< 1; c) -1:::;Rez:::;l;
d) Imz 2 0; e) Izl :::; 2; f) 1 < Izl < 3;
g) Izl > 2; h) -7r < argz < 7r; i) ~ < argz < f?
Solution.
a) Rez = Imz {::::=:} x = y, where z = x + iy.
The desired sub set consists of the points of the straight line y = x (Figure 1.1).
Figure 1.1 Rez = Imz
b) Rez < 1 {::::=:} x < 1 and y is an arbitrary real number, where z = x + iy. The
desired subset is the half plane left from the straight li ne x = 1 (without the points
of this straight line), Figure 1.2.
Figure 1.2 Rez < 1
c) -1 :::; Re z :::; 1 means -1 :::; x :::; 1 and y is an arbitrary real number. The
desired subset is the strip between straight lines x = -1 and x = 1, Figure 1.3.
:..1 x
Figure 1.3 -1 :::; Rez :::; 1
d) Im x ~ 0 means y ~ 0 and x is an arbitrary real number. The desired subset

is the upper half plane with respect to the x- axis, Figure 1.4.
Figure 1.4 Imz ~ °

e) The eondition JzJ ~ 2, using the trigonometrie representation of the complex
number z = (p, ()), reduced on JzJ = p ~ 2 and () is an arbitrary angle from [0, 27l'J.
The desired sub set is the disc with center at (0,0) and radius r = 2, Figure 1.5.
2 X
Figure 1.5 JzJ ~ 2
f) The ease 1 < JzJ < 3 reduces in a sirnilar way as in e) on 1 < JzJ = P < 3,
where () is an arbitrary angle from the interval [0,27l'J. The desired sub set is the
annulus between the circles Izl = 1 and Izl = 3 without these circles, Figure 1.6.
Figure 1.6 1 < Izl < 3
g) For the case Izl > 2 we have JzJ = p > 2, () E [0, 27r]. Therefore the desired
subset is the whole complex plane without the disc Jzl ~ 2, Figure 1.7.
Figure 1.7 Izl > 2
h) The condition -7r < arg z < 7r implies that p is arbitrary and -7r < () < 7r.
Therefore the desired subset is the whole complex plane without the negative part
of the real axis, Figure 1.8.
Figure 1.8 -'Fr < arg z < 'Fr
i) The condition 'Fr /6 < arg z < 'Fr /4 implies that p is arbitrary and e E ('Fr /6, 'Fr /4).
The desired sub set are the points between half straight lines y = x . tan ~ and
y = x . tan iwithout these half straight lines, Figure 1.9.
Figure 1.9 'Fr /6 < arg z < 'Fr /4
Example 1.7 Find the conditions for the real and imaginary part of a complex
number z so that z belongs to the triangle with vertices 0,1 + zY3, 1 + z/ Y3, Figure
1.10.
Solution. We denote the vertices as:
Zl = 0, Z2 = 1 + zV3, Z3 = 1 + z/V3.
O=z 1 x
Figure 1.10
If we denote z = x + zy, then Rez = x and Imz = y and
-Y = tan (), () E [1C'

-, -1C'] .
x 6 3
Therefore Im z = Re z . tan (). Since the function tan () is monotone increasing we

have
1
y'3Rez ::::; Imz ::::; V3Rez.
Together with the condition 0 ::::; Re z ::::; 1 we have completely described the points
in the given triangle with vertices Zt, Z2 and Z3.
Example 1.8 Prove that tor every z E C
Izl ::::; IRezl + IImzl ::::; -12 ·14

Solution. The left part of the inequality follows from
Izl = v'Re 2z + Im 2 z::::; y'(IRezl + IImzl)2 = IRezl + IImzl

The right part of the inequality follows from the following obvious inequality
(IRezl-IImzr ~ O.
Then
IRezl 2 + IImzl 2 - 2lRezl·IImzl ~ 0,
Izl2 ~ 2IRezl·IIm z l·
Adding Izl 2 to the both sides of the last inequality we obtain
21z1 2 ~ 2lRezl'IImzl + Iz 12,
21z1 2 ~ (IRezl + IImzlt Hence v'2lzl ~ IRezl + IImzl·
Example 1.9 Find the complex numbers which are the corners 01 the triangle with
equal sides with vertices on unit circle and whose one vertex is on the negative part
the real axis.
Solution. Let us put Zj = pj(cos8j + zsin8j ),j = 1,2,3, for the soughtafter
points.
/
/
Figure 1.11
We have for alt three vertices Pj = 1 and

Zl : 81 = 7r /3, z2: 82 = 7r, Z3: 83 = -7r /3.
Therefore:
Zl = (1, 7r/3) = 1/2 + zV3/2; Z2 = (1, 7r) = -1; Z3 = (1, -7r/3) = 1/2 - zV3/2,
Figure 1.11.
Exercise 1.10 Find the real and imaginary parts, modulus, argument and the eom-
plex conjugate for the following numbers:
a) 51; b) 7r', e) 1/2 + 1/2; d) 1 + l/VJ;
e) 1
1 + 1VJ;
1) (3 + 13)(1 + 1VJ); g) C +11VJ f; h)
1+ 1
1- 1
Answers.
Rez Imz Izl argz z

a) 0 5 5 7r/2 -51
b) 7r 0 7r 0
e) 1/2 1/2 ../2/2 7r/4 1/2 - 1/2
d) 1 1/v3 2../3/3 7r/6 1-1/../3
e) 1/4 -V3/ 4 1/2 57r/3 1/4 + 1V3/4
1) 3-3V3 3+3V3 V72 arctan( -2 - V3) 3 - 3V3 - (-3 + 3V3)1
g) -1/2 3 0 1/23 7r -1/23
h) 0 1 1 7r/2 -1
Example 1.11 Find t and 0 so that the eomplex numbers
z = t+ tf) and w = t(cosO + lsinO)
would be equal.
Solution. The equality z = w implies
Izl = Iwl and tan(argz) = tan(argw).

The first condition Izl = Iwl implies t 2 = t 2 + 02 , i.e., 02 = O. Putting 0 = 0 in z and
w we obtain z = t and w = t for t an arbitrary real number.
Example 1.12 Let C* be the set of alt eomplex numbers different from zero.
a) Prove that the set T of all eomplex numbers with modulus 1 is a multiplicative
subgroup of the group (C*,.).
b) The multiplieative group C* is isomorphie with jR+ X T.
Solution. a) We have T = {zllzl = I} C C. Tunder the multiplication by

the equality
IZI . z21 = IZII·l z21·
The associativity follows by the associativity of multiplication in C. The neutral
element is 1. The inverse of z = (a, b) E T is
C2 : b 2' a2 - : b2) E T.
b) It is easy to check that an isomorphism is given by
J(z) = (lzl,cosO+zsinO),
where 0 = argz. nopagebreak c) We define an equivalence relation'" on R by
rl '" r2 <===} rl - r2 = 2br, k is an integer. Let i be the corresponding quotient
set. Prove that R+ x T isomorphie with R+ xi. The group c· is isomorphie with
R+ x i (check). Therefore by b) and transitivity of the isomorphisms of groups it
follows c).
Example 1.13 Find the sum of complex numbers which are the vertices of a n-
polygon in circle with radius r with the center in (0,0) for n = 4,6, ... , 2p, p E F:!,
Figure 1.12.
Solution. Using the geometrical interpretation of the addition we see that the
surn is O.
Figure 1.12 hexagon
Namely, adding the pairs of complex numbers on main diagonals we always

obtain 0 and then the total sum is 0 for n = 4,6, ... , 2p.
Second method. Using the Euler representation of complex number Z = re"" we
obtain
Example 1.14 Prove that the condition Jor Jour points Zl,Z2,Z3 and Z4 be conse-
quent vertices oJ a parallelogram is the Jollowing
Solution. The equality Zl - Z2 + Z3 - Z4 = 0 implies
Exercise 1.15 For three given complex numbers Zl, Z2 and Z3, find Z4 such that the
corresponding points in the complex plane will be the corners oJ a parallelogram ..
Rint. Complex number Z4 can be obtained by Example 1.14:
Example 1.16 Starting with a complex number Z i- 0, find where the complex
numbers 2z,3z, ... ,nz are?
Solution. For Z = p( cos () + z sin ()) we have
nz = np( cos () + z sin ()),

the complex numbers nx, for n = 1,2, ... are on the half straight line y = tan () . x
with modulus np.
Example 1.17 Find the length oJ a side oJ a pentagon and the length oJ its diagonal
iJ it is inside the unit circle.
Solution. We have Zl = 1. Then
27r . 27r 47r . 47r

Z2 = cos -
5
+ z sm -
5
and Z3 = cos - + z sm - .
5 5
The length of the side Zl - Z2 is
I1 - 2~
cos 5" - . 5"
z sm
2~1
2~ . 2~
= (1 - cos 5" )2 + (sm 5")2
= )2(1 _ 2; ) cos
.
2sm 5 ,
~
where we have used the identity 2 sin 2 ~ = 1- cos t.
_-+-+-__~I--_ _ _~Z' ____ )

1 X
Figure 1.13
We obtain analogously IZ2 - z31 2 sin 2;, Figure 1.13. Find the length of
diagonals.
Exercise 1.18 Find where the points Z are for a fixed Zo :
Iz + zol = 1;
~
a) Iz-zol=l; b) c) arg (z . zo) = 4'
Answers. a) The set {zllz - zol = I} is the circle with center Zo and radius 1,
Figure 1.14.
y
o x
Figure 1.14 {zllz - zol = I}
b) {zllz + zol = I} is the circle with center -Zo and radius 1, Figure 15.
Figure 1.15 {zllz + zol = I}

c) {z 1 arg (z . zo) = !}, the corresponding complex numbers are on the half
straight line y = tan (! - O)x, where argzo = 00 , This follows by arg(z· zo) =
7r
00 + 0 = 4'
Exercise 1.19 Find where the points in the complex plane are if
a) Iz - 11 ::; 2; b) Iz + zl > 1; c) Iz - zol < r;
7r 7r 7r
d) arg (z - 1) = -; e) 4 < arg (z + z) ::; 4; f) larg (z - zo)1 < 0;
4
g) Re (zo . z) = 0, for Zo E li; h) III < r; i) Re(zz) = !;

z
j) -z = z.
Answers. a) The closed disc with center at (1,0) and radius r = 2, Figure 1.16.
--~~~~~~r--'
-1 X
Figure 1.16 Iz - 11 :s: 2
b) Wh oIe complex plane without the disc with center -2 and radius r = 1, Figure
1.17.
Figure 1.17 Iz + 21> 1
c) The disc with center Zo, and radius r (without the circle Iz - zol = r), Figure
1.18.
Figure 1.18 Iz - zol < r
cl) The straight half-li ne y = x-I, y 20, Figure 1.19.
y
/
/
./"---
//
/// //
/' /./
./ /./
---------<-/'-/-O~-ij--/~--·--·-·-x- .
Figure 1.19 y = x-I, Y2 0
e) The region between the straight half-lines

Y = x-I, y = ffx -1 ancl Y 2 1 inclucling the last straight half-line, Figure 1.20.
r
-I
Figure 1.20
f) For Zo = a + zb the points z :
-0 < arg (z - zo) < 0
are in the region between straight half-lines
y - b= (x - a) tan 0 and y - b = (x - a) tan( -0), x ~ a
and y ~ b (y ::; b for the second), 0 E [0, i-l, and x ~ a and y ~ b (y ::; b for the
second), Figure 1.21.
~Y
,
Figure 1.21
g) The points z are on the imaginary axis fOT Zo #- O. FOT Zo = 0 the points z are
arbitrary complex numbers (Examine the case Zo E C; Y = kx, k = ~~:O).
h) From I!I < r we obtain Izl > ;, the points z are outside of the cirele Izl = ~.
i) The points are on the straight li ne y = -i.
z
j) Using the Euler representation z = pe'tp we obtain from ~ = z that

e'tp . 2,tp
--
e-'tp
= z, z.e., e = l.
Therefore the points z = pe'1r/4 satisfy the condition
z
-= = z, y = x.
z
Example 1.20 Prave for u, v E C the relation
and give a geometrie interpretation.
Solution. Since u . u = lul 2 we can easily obtain the desired equality. Narnely,
(u + v)· (u + v) + (u - v)· (u - v) = 2u· u + 2v· v,
where we have used :u:E""V = u ± v.

The geornetrical interpretation: the surn of squares of diagonals of a square is
equal to the surn of the squares of its sides.
Exercise 1.21 Let t be a real parameter ZI = (PI, 81 ) and Z2 = (P2' 82 ) fixed complex
numbers. Which curves are given in the complex plain by the following relations?
a) Z = (2, t), 0::; t < 271";
b) Z = (t,~), t ~ 0;
c) Z = ZI + cost + zsint, 0::; t < 271";
d) Z = ZI + t ( cos T+ z sin~), t ~ 0;
e) Z = ZI + Z2(COS t + zsin t), 0::; t < 271";
J) Z = ZI + t(Z2 - zt), 0 ::; t ::; 1;
g) Z = ZI (cos t + z sin t) + Z2( cos t - z sin t), 0 ::; t < 271";
h) Z = t + (cos t + i sin t), t > O.
Answers.
a) The circle with the center 0 and radius r = 2.
b) The half straight line y = V3x, y ~ o.
c) The circle with the center Z and radius r = 1.
d) The straight half-line y = x, y ~ 0 translated for PIon the half straight line
y = tan ()I . X, Y ~ 0, Figure 1.22.
o
Figure 1.22
e) The circle with the center ZI and radius r = P2'

f) The line segment [zt, Z2], Figure 1.23.
f"Y
r:/Z'
.
I /
.' ,/
2 - 22 /" /" ..... 0

1 /"/" I x
0" I
Figure 1.23 [zt, Z2]
g) The circle with parametrie equations
x = PI COS(()I + t) + P2 COS(()2 - t)
Y = PI sin(()I + t) + P2 sin(()2 - t), 0::; t < 211",
where
ZI = PI( COS ()I + zsin ()I), Z2 = P2( tos ()2 + zsin ()2)'
In particular, for ZI = Z2 we obtain
x = 2PI COS ()I COS t and y = 2PI . sin ()I cos t, 0::; t < 211".
Eliminating the parameter t we obtain apart of the straight line
y = tan 01 . x, - 2Pl cos 01 :::; X :::; 2Pl cos 01.
h) The curve is given by the following parametrie equations
x = t + cos t, Y = sin t, t > 0,

Figure 1.24.
Figure 1.24
Example 1.22 Find
a) 0; b) 0=1; c) ~; d) N.
{ji = y!(cosi+zsini)
"Fr
- + 2k7r "Fr + 2k7r
cos 2 3 + Z sin 2" 3 '
for k = 0, 1, 2.
b) We have
77r +
~( cos 4 . 77r)
v2 5
5
ZS1ll
= v2 cos 1f + 2k7r + zsm

10~( . 1f + 2k7r) ,
5 5
for k = 0,1,2,3,4.
c) We have
lJlO(cos arctan(-1/3) + zsinarctan(-1/3))

417fi ( arctan(-1/3)+2k1r . arctan(-1/3)+2k1r)
v 10
IU COS 2 + z sm 2 .
Second method:
Starting from ~ = x + zy, find x and y taking equal the real parts and then the
imaginary parts.
d) We have
4/""7
V -1 =
~ .
cos 7r + z sm 7r = cos
(7r"4 + 2k7r) + z sm. (7r"4 + 4k7r) '
for k = 0,1,2,3.
Write down in all examples all cases in the form a + zb, a, b E ~.
Example 1.23 Solve the following equations in C :
a) x 8 - 16 = 0; b) x 3 +1 = 0; c) x 6 + z + 1 = O.
Solution.
a) The zeroes of the equation x 8 - 16 = 0 are the values of m,
m = y'16( cos 0 + z sin 0) = h( cos k: + z sin k47r),
for k = 0,1,2, ... , 7. Hence
Zl = V2, Z2 = 1+ Z, Z3 = V2z, Z4 = -1 + z,
Zs = -V2, Z6 = -1 - Z, Z7 = -V2z, Z8 = 1 - z.
b) The zeroes of the equation x 3 + 1 = 0 are the values of A,

r-I {j .
-1= Cos7r+zsm7r = cos
7r + 2k1r
3
. 7r + 2k1r
+zsm 3 ' fork=0,1,2.
Hence
Zl = 1, Z3 =
1
-+z-.
v'3
2 2
c) The zeroes of the equation x 6 + z + 1 = 0 are values of ~ -1 - z ,
571"
571" +2h -+2h
~ (cos T 6 + z sin 4 6 ),
for k = 0,1, ... ,5.
Example 1.24 Using the equality cos t = sin t = y.} find:

a) cos TI;; b) sin f6-.
Solution.
a) Starting from the equality
(cosi+ zsin i)2 =cosz+zsinz,

we obtain
. 2 . z z d
sm 2 cos 2
2 Z • 2 Z
sm z = an cosz = cos 2- sm 2'
i + zsin i
Since
cos E {z Ilzl = I},
we have cos 2 t + sin2 ~ = 1. Putting this in the second identity we obtain
cos ~ =
2
J cos z + 1 .
2
Applying the last formula two times on cos ~ we obtain
b) In this case we start from the equality

z
( cos 3" .
+ zsm 3"
Z)3 = cos z + zsm. z.
Example 1.25 Prove for x =I- 0 :

n cos R±J_ x sin nx n. sin~x. sin~
a) ~
L.J cos kx = ~ • X 2"" ., b) E smkx = . x .
k=l sm '2" k=l sm '2"
Solution. a) and b): We consider the sum
n n n
S =E cos kx + tE sin kx = E(cos kx + tsinkx),
k=l k=l k=l
Then
n 1 mx
'" ,kx =e'x
S =L...Je - e ,
k=l 1 - e'X
where we have used the Euler formula cos kx + tsin kx = e,kx. Using the formula
e1X _ e- tX
Slnx = - - - -
2t
we can transform S in the following way:
If we apply the Euler formula on e(n+1},x/2 and compare with the starting sum we
obtain
n + 1 sin nx +1
= cos - - x . T + z sin _n__ x T·
n n sin nx
E cos kx + zE sin kx .
k=l k=l 2 sm "2" 2 sm "2"
Putting equal the real parts and then the imaginary parts we obtain the desired
equalities.
Remark. Find in an analogous way the more general equalities for the sums
n n
E cos(a + kx); E sin(a + kx).
k=O k=O
Example 1.26 Find the position 0] the vertices 0] the triangle with equal sides i]
the two vertices are -1 and 2 + t, Figure 1.25.
JfA.y
I
/
-~
I2 x
I
'" 'CI
Figure 1.25
Solution. First we shall show that for any triangle with equal sides we have
where Zl, Z2, Z3 are the vertices of the triangle with equal sides. We have (see Figure
1.25)
Z2 - Zl = e 1r· /3 (z3 - zt} and Zl - Z3 =e 1r
·/
3 (z2 - Z3).
Dividing these two equalities we obtain
Since Zl and Z2 are known we can find Z3 by the last equality (we obtain two solu-
tions).
Exercise 1.27 Find the following sums:

n n n
a) L cos(2k - l)x; b) L sin(2k - l)x; c) L (_l)k-l sin kx.
k=l k=l k=l
Answers.
a)
sin 2nx
-2-'-; b)
sin 2 nx
c)
sin n..tl cos T
smx smx cos~
1.1. ALGEBRA1C PROPERT1ES 27
Exercise 1.28 Find the following sums for real eonstants m and n 1= 2k7r
T
sin(m + kn).
T
a) L cos(m + kn); b) L
k;O k;O
Answers.
sin~n rn sin!±!.n rn
a) . n cos(m+-
2 ); b) . 2n sin(m + -).
sm2 sm 2 2
Exercise 1.29 Solve the equation z= zn-l (n is a natural number).
Exercise 1.30 Let m and n be integers. Prove for z 1= 0 :

a) that (y'Z)m has ~ different values where (m, n) is the greatest eommon
~n,mJ
divisor of the numbers m and n.
b) That the sets of values of (y'Z)m and yrzm are equal, i.e.,
(y'Z)m = ;:;zm if and only if(n,m) = 1, i.e., n and m has no non-trivial eommon
divisors.
Exercise 1.31 Prove the identity

11 - ZlZ21 2 - IZI - z21 2 = (1 - IZlI2) . (1 - IZ212).
Exercise 1.32 Prove the inequality
Exercise 1.33 Find the vertices of regular n- polygon if its center is at Z = 0 and
one vertex is known.
Answer . The vertices are

Zk = Zl· e'2k1n, k = 0,1, ... ,n -1,
where Zl is the given vertex.
Example 1.34 Prove:

a) 1f Zl + Z2 + Z3 = 0 and IZII = IZ21 = IZ31 = 1 then the points are vertices of a
triangle with equal sides whieh is in the unit eirele.
b) 1f Zl + Z2 + Z3 + Z4 = 0 and IZII = IZ21 = IZ31 = IZ41 = 1, then the points
Zl, z2, Z3, Z4 are either vertiees of a triangle with equal sides or they are equal in
pairs.
Solution. a) The length of si des of the triangle with vertices Zb Z2 and Z3 are
IZ3 - zll, IZ2 - zt\, IZ3 - z21·

By the given conditions: Zl + Z2 + Z3 = 0 and IZil = 1, i = 1,2,3, we have
IZ3 - zll2 = 12z1 + z21 2 = (2z1 + z2)(2Z1 + Z2) = 5 + 2(ZlZ2 + ZlZ2),
and analogously
IZ3 - z21 2 = 12z2 + zll2 = 5 + 2(ZlZ2 + ZlZ2)'
Therefore IZ3 - zll = IZ3 - z21·
We can prove in a quite analogous way that IZ2 - zll = IZ3 - z21. Therefore the
tri angle with vertices Zb Z2, Z3 is with equal sides.
Example 1.35 Let the points Zb Z2, ... , Zn be on the same side with respect 0/ a
straight line wh ich cross (0,0). Prove that the points
1 1 1
, , ... ,
Zl Z2 Zn
have the same property and that Zl + Z2 + ... + Zn # 0, and that

1 1 1
-+-+
Zl Z2
... +-
Zn
#0.
Solution. We can suppose without weakening the generality that the straight
line from this example is just the imaginary axis and that all points are right from it
(in the opposite it is enough to multiply Zk by some convenient number eil"). Then
it is obvious that Re Zk > 0 and Re lk > 0 for all k, which implies the desired
properties.
Exercise 1.36 Solve the equation (1 + ~) 3 = z.
Example 1.37 Prove that the field 0/ complex numbers is the smallest field which
contains the field 0/ real numbers and the solution 0/ the equation x 2+ 1 = O.
Solution. We will prove the desired result by reductio ad absurdum. Suppose

the contrary, that there exists a field T which contains IR, the solution of the equation
x 2 + 1 = 0 and is smaller than the field C, C :J T and C # T. Then there exists
Zo = a + zb, z E C, such that a, b E IR and Zo (j. T. On the other side, since T contains
the solution of the equation x 2 + 1 = 0 we have z E T. Therefore by T :J IR and the
fact that T is a field we have x + yz E T for every x, y E IR. Hence Zo ETaIso, which
is a contradiction.
Example 1.38 Prove that there does not exist a total order in the field of complex
numbers whieh is compatible with the opemtions in this field and which extend the
usual order of reals.
Solution. Suppose that there exists a total order ~ in the field C. We shall
compare z = z and z = O. Suppose that z ~ O. Then Z2 ~ 0, -1 ~ O. Contradiction.
Suppose now that z ~ O. Then 0 = z - z ~ i. Multiplying both sides by -z (-z ~ 0),
we obtain (-z)(-z) ~ 0, i.e., -1 ~ O. Since we have obtained contradiction in both
cases, we condude that such a total order ~ can not exist.
Example 1.39 Let p be the eommutative ring of all polynomials with real eoeffi-
eients endowed with the usual addition and multiplieation.
Let J be a ideal ofthe elements oftheform (1+x 2 )Q(x), where Q is a polynomial,
in the ring P. We define in P an equivalenee relation'" in the following way:
Prove that
a) the set of all polynomials of first order with respeet to + is isomorphie with
the set PI J of all equivalenee classesj
b) PIJ is a fieldj
c) the field PI J is isomorphie with the field of alt eomplex numbers C.
Solution. a) Each polynomial P from P can be written in the polynomial

form
P(x) = (x 2 + 1)Q(x) + ax + b (x E lR)
for some a, b E lR. Therefore an equivalence dass from PI J has the form J + ax + b
for some a, b E lR. The addition + in PI J is defined by
(J + (ax + b)) + (J + (ex + d)) = J + (a + e)x + b + d. (1.3)

This implies the isomorphism between (PI J, +) and (PI, +), where pI is the set of
all polynomials of the first order.
b) By (1.3) the operation + in PI J is an inner operation. The neutral element
is J + 0 and the inverse element of J + ax + b is the element J - ax - b. The
multiplication in PI J is given by
(J + (ax + b)) . (J + (ex + d)) = J + ae(x 2 + 1) + x(be + ad) + bd + ae,

i.e.,
(J + (ax + b)) . (J + (ex + d)) = J + x(be + ad) + bd - ae. (1.4)
The unit element is J + 1. The inverse element of J + ax + b for a 2 + b2 =I 0 is

J + a' x + b', where a' and 1/ are the unique solutions of the system of the equations
b . a' + a . b' =0
b . b' - a . a' = 1.
c) Comparing the usual operations in C :
(az + b) + (cz + d) = (ae)z + b + d

(az + b) . (cz + d) = (bc + ad)z + bd - ae
with (1.3) and (1.4), respectively, follows the isomorphism between (C, +,.) and
(PfJ, +, .). .
Example 1.40 Prove the isomorphisms:
a) the group (C \ {O},·) with the group of matrices of the following form
[_~ !], a, b E IR and a 2 + b2 =I 0
with the matrix multiplication;

b) The group of quaternions
3
K \ {O} = {wo + w1i + w2j + w3kl Wi E IR,Ew~ =I 0, i = 0,1,2,3},
;=0
where
ij = -ji = k, jk = -kj = i, ki=-ik=j;

1.i = i . 1 = i, 1· j = j . 1 = j, 1· k = k·1 = k,
with multiplication and the group of real matrices
:~ -:: :: 1
W3 Wo -Wl
-W2 Wl Wo
with matrix multiplications.

c) The group of quaternions (K \ {O},·) with the group of matrices of the second
order
(u,v E C \ {O})
with matrix multiplication.

Rint. For c). Use a) and b) in the deeomposition of the following matrix
where u = Wo + iWI and v = W2 + iW3.

z-c
Example 1.41 Let Icl < 1 and w =
1 - cz
--_-.
What are the following sets in the z-complex plane:

Al = {zllwl < I}, A 2 = {zllwl = I}, and A 3 = {zllwl > I}?
Solution. For set AI, the eondition Iwl < 1is equivalent with the inequality
Iz - cl 2 < 11 - czl\ i.e., (1 -lcI 2 )(lzI 2 - 1) < O.
Henee Al = {zllzl < I}. In a quite analogous way we obtain
A2 = {zllzl = I} and A3 = {zllzl > I}.
Exercise 1.42 Prove for arbitrary Zl, Z2, Z3 E C that
1 Zl Zl
1 Z2 Z2
1 Z3 Z3
is areal number.
Exercise 1.43 Prove the inequality

Iz -11 ~ Ilzl-11 + Izllargzl (z E C).
Rint. Use the geometrie interpretation of the complex number.
Exercise 1.44 Which curves are given by the following sets

a) {zllz - 111z + 11 = const},
1= const}?
b) {ziliz - 11
z+l
Exercise 1.45 Prove the equality
(n - 2) E + IE lakl 2 akr E Elak +ail

l::;k< i::;n
2
1.2 The Topology of the Complex Plane
1.2.1 Preliminaries
We denote by D(zo, r) the open disc centered at Zo with radius r,
D(zo,r) = {z Ilz - zol < r}.

We call D(zo, r) also a neighborhood of zoo We denote by C(zo, r) the circle centered
at Zo with radius r,
C(zo,r) = {z Ilz - zol = r}.
The function d : IC -+ 1m. given by
is ametrie on IC, i.e., d(zt, Z2) 2:: 0;

d(ZI' Z2) = 0 if and only if ZI = Z2;
d(Zt,Z2) = d(Z2,ZI)
d(ZI,Z2):::; d(Zt,Z3)+d(Z3,Z2)'
A set 0 C IC is open if for every z E 0 there exists c: > 0 such that D(z, c:) C O. A
set Ais closed if its (set) complement AC is open.
A point w is an accumulation point for a set A C IC if in every disc D (w, r) there
are infinite elements from the set A.
The boundary of a set AC IC, denoted by BA, is the set of complex numbers whose
every neighborhoods have a nonempty intersection with both A and AC.
A set A is bounded if A C D(O, M) for some M > O. A set A is compact if it is
closed and bounded. A set A is connected if there do not exist two disjoint open
sets 0 1 and O2 whose union contains A while neither 0 1 nor O2 alone contains A.
We denote by [ZIZ2] the line segment with endpoints ZI and Z2' A set Ais polygonally
connected if any two points Zo and Zn of A can be connected by a polygonal li ne
[zn, ZI] U ... U [Zn-I, Zn] contained in A. A polygonally connected set is connected.
A set A is a region if it is open and connected.
Let
S = {(a, b, e) I a2 + b2 + (e -1/2? = 1/4}. (1.5)
The stereographie projection of S \ {(O, 0,1)} on IC ( of S on IC U {oo}) is a 1-1
correspondence obtained taking that the plain e = 0 coincides with the complex
plane IC, and that a and b axes are the x and y axes, respectively, and we associate
to (a, b, e) E S the eomplex number where the ray from (0,0,1) intersects c. We
have
x
a= . b= Y ;
x 2 + y2 + l' x 2 + y2 + 1
1.2. THE TOPOLOGY OF THE COMPLEX PLANE 33

Exercise 1.46 Prove that the function d : !C -+ Im. given by
is ametrie on !C.
Exercise 1.47 Prove that the lunction d: !C -+ Im. given by
is ametrie on !C, where Zi = Xi + ZYi, i = 1,2.
Exercise 1.48 Which 01 the lollowing sets is open
a) {zllzl<3};
b) {zl1 < Rez < 3};
c) {zIRez<3}U{3}?
Answers. a) Yes. h) Yes. c) Not.
Example 1.49 Give an example lor a connected set which is not polygonally con-
nected.
Solution. The set of points
is connected hut not polygonally connected, since the set A contains no straight line
segments.
Exercise 1.50 Prove that a connected subset 01 Im. is an interval.
Hint. Prove that a suhset A of Im. is an interval if and only if for every two points
rl and r2 in Im. with rl < r2 we have trI, r2] C A.
Exercise 1.51 Which 01 the lollowing subsets 01!C are connected:
a) {zllz-11<1}U{zllz-31<1};
b) {-~lnEN}U(0,11?
Example 1.52 Prove that a region is polygonally connected.
Solution. Let A be a region in IC and Zo E A. We denote by 0 1 the set of

points of A which are polygonally connected by Zo in A and by O 2 the set of points
in A which are not. Both sets 0 1 and O 2 are open, e.g., 0 1 is open since every point
z can be connected by every point in D (z, c:). Since A = 0 1 U O 2 and A is connected
it follows that O2 is empty (we have Zo E Od. Now we have that every point in
A can be polygonally connected by zoo Therefore every pair of points in A can be
polygonally connected through the point zoo
Exercise 1.53 Prove that a polygonally connected subset of IC is connected.
Exercise 1.54 Prove that for stereographie projection the following formulas hold
a b
x=--; y=--.
1-e 1-e
Hint. Use that the points (O,O,l),(a,b,e) and (x,y,O) are collinear.
Exercise 1.55 For the points 0 and 1 + z in IC give the corresponding points of S
by stereographie projeetions.
Exercise 1.56 Let S be given by (1.5) and C be a circle on S, i.e., the intersection
of S with a plane of the form Aa+ Bb+ Ce = D. Prove that ifC. is the stereographie
projection of C on IC then
a) for (0,0, 1) E C the projection C. is a line;
b) for (0, 0, 1) rf. C the projection C. is a circle.
Exercise 1.57 Prove that the function d : IC U {oo} -t ~ given for ZI, Z2 E IC by
and by
2
d(z, 00) = (1 + IzI2)1/2
for z E IC is ametrie on IC U {oo }.
1.2. THE TOPOLOGY OF THE COMPLEX PLANE 35
Exercise 1.58 Prove that the function d : C - t lR given by
is ametrie on C, where Zi = Xi + zy;, i = 1,2.
lu-vi
Example 1.59 Prove that by d( u, v) = u, v E C is given a
metric on C.
-/1 + lu1 -/1 + Iv1
2 2'
Solution. We shall prove only the triangle inequality d( u, v) d( u, w) :s +

d( w, v), since other properties can be easily verified. We start with the identity
(u - v)(l + ww) = (u - w)(l + vw) + (w - v)(l + uw)

for u,v,w E C.
This implies
lu - vl(l + Iw1 2 ) :s lu - will + vwl + Iw - vIII + uwl. (1.6)
For 11 + vwl we have the following equality
which is equivalent with the inequality Iv - wl 2 ~ O. Hence
(1 + vw)(l + vw) :s (1 + IvI 2 )(1 + Iw1 2 )

(1 + IvI 2 )(1 + IwI 2 ).
Analogously we obtain
(1.7)
Using the previous two inequalities in (1.6) we obtain the desired inequality
lu - vi lu - wl Iw - vi
-/(1 + luI 2 )(1 + Iv1 2 ) :s -/(1 + luI 2 )(1 + Iw1 2 ) + J(1 + IwI )(1 + Iv1
2 2) •
It is easy to prove that d( u, v) E [0, 1] for every u, v E C.

Chapter 2
Sequences and se ries
2.1 Sequences
2.1.1 Preliminaries
Let {Zn} be a sequence of complex numbers. A sequence {zn} is bounded if there

exists M > 0 such that IZnl < M for all n E N.
A complex number V is an accumulation point of {zn} if for every c > 0 there exist
infinitely many integers n such that
Definition 2.1 We say that {zn} converges to w E C, in the notation
lim Zn = W
n-+oo
if for every c > 0 there exists an integer no such that for all n ?: no
We say that {Zn} is a Cauchy sequence if for every c > 0 there exists an integer no
such that for all n, m ~ no
Theorem 2.2 A sequence {zn} of complex numbers is convergent if and only if it

is a Cauchy sequence.
37

38 CHAPTER 2. SEQUENCES AND SER1ES

Example 2.1 Find which of the following sequences are bounded:
b) {C ~zf};
c) {G::f}; d) {2n:1 +zn:1};
Solution. a) The sequence is bounded since Iznl = Izln = l.

b) The sequence is bounded, since
I( -1- )nl_ ( - - 1 - < -1- rwr every n E N.

1 - )n _ - -
1 + z - 11 + zl - (v'2)n - ..;2'
c) The sequence is bounded, since I(i :t !fl = 1, for every n E N.
d) The sequence is bounded since
I- n- + z
n-- -1 = V n
(--)2
1
+ (__
n-l )2 < -5 for every n E N.
2n + 1 n 2n + 1 n - 4
e) The sequence is unbounded since n i= 4k, k = 1,2, ... ,
Example 2.2 1f the sequence {an} is bounded, prove that the following sequences
are also bounded
LPiai
n }
b) { i=~ , Pi > 0;
LPi
i=1
Solution. We shall use that there exists M > 0 such that lail :s M for every
i E N. a) We have
-n1 Lai
1 :s -n L lail:S -n1 nM =
nil n
;=1 ;=1
M.
2.1. SEQUENCES 39
b) We have
c) We have
Example 2.3 Find the accumulation points of sequences fram Example 2.1.
Solution. The accumulation points are
a) l,-l,z,-lj b) Oj
c) 1,-1,z,-zj d) ~+Zj
e) O.
Example 2.4 Which sequences fram Example 2.1 converge?
Solution. a) Does not converge.

h) The sequence is hounded and it has one accumulation point O. Therefore it
converges to O.
c) Does not converge. It is hounded hut with few accumulation points.
d) The sequence is bounded and it has one accumulation point ~ + z. Therefore
it converges to ~ + z.
e) Does not converge. It has one accumulation point hut it is unhounded ..
Example 2.5 Prave that ifthe sequence {w n } converges to w, then {Iwnl} converges
Iwl. The opposite is not true.
Solution. First we shall prove that the inequality
lu - vi ~ Ilul-lvll for u,v E C. . (2.1)
Since u = (u - v) + v we have
lul ~ lu - vi + lvi·
Hence
lul-Ivl ~ lu - vi.
40 CHAPTER 2. SEQUENCES AND SERIES
Starting from v = (v - u) + u, analogously we obtain

Ivl-Iul ~ lu - vi,
-lu - vi ~ lul-Ivl ~ lu - vi·

Since {w n } converges to w, we have that for every e > 0 there exists no E N such
that IWn - wl < e for every n ~ no. Putting v = Wn in (2.1) we obtain
for every n ~ no, i.e., Iwnl -7 Iwi as n -7 00. The following example shows that
the opposite is not true.
Exercise 2.6 Prove that a sequence {wn} converges to w, if and only ifthe sequence
{Re w n } converges Re wand the sequence {Im w n ) converges to Im w.
Exercise 2.7 Find a sequence with one accumulation point which does not converge.
Answer. {(( _l)n + l)n}.

lim Iwnl = r
n-+oo
(r > 0) lim argw n = cp
and n-+oo
imply
lim Wn = re"",
n-+oo
where arg W is the main value.
Solution. For Wn = Un + ZVn , we have

un = Re Wn = Iwnl cos{arg wn)
Vn = Im Wn = Iwnl sin(arg w n ).
Therefore
lim U n = n-+oo
n ....... oo
lim Iwnllim
n-+oo
cos(arg wn ) = rcoscp,
lim IWn In-+oo
lim V n = n--+oo
n--+oo
lim sin{ arg wn ) = r sin cp.
Hence
lim
n-+oo
Wn = r(coscp + zsincp).
We have used that {w n } converges, if the sequences {u n } and {v n } converge.
2.1. SEQUENCES 41
Example 2.9 Let nlim

....oo
W n = wo. Is it true then that also
lim arg
n .... oo
Wn = arg wo?
Solution. It is not true ! Counter-example: Wo = 0, Wn = -1 + (_l)n;. We

have
1 1
Ji..~ Wn = -1, and arg W2k = 7r - arctan 2k' arg W2k+l = 7r + arctan 2k + 1 .
Hence {arg w n } diverges.
Remark. If {w n } converges to Wo =I 0, then for every value 'P = IArgwo there
exists a sequence 'Pn = Arg Wn which converges to Arg wo. If Wo =I 0 is not a negative
number, then we have also limn .... oo arg W n = argwo.
Example 2.10 Prove that if n-+oo

lim W n = wand n-oo
lim w~ = w' then for an arbitrary
but fixed k E N :
a) n-+oo
lim (wn)k = w\
k k
c) n-+oo "w
lim L...J 1'
n "w
= L...J 1' •
1'=0 1'=0
Solution. a) and b) The proof goes by induction. We known that for k = 2
lim (w n )2 = w2 and lim (W n ) 2 _ (:::.-) 2

n-+oo n-+oo W~ - W' .
Suppose that the desired equalities hold for k - 1,
lim (W n ) k- 1 = Wk- 1 , lim (Wn)k-l

- = (W
- )k-l .
n-+oo n-+oo w~ w'
Then we have for k
( Analogousl y also for b) ).

c) By induction we can prove that the surn of finite number of convergent se-
quences converges to the sum of their limits. Therefore by a) we have
"w
k k k
lim L..J
n-+oo
1'
n lim w 1'n
= "L...J n-+oo = "w
L.J
1' •
p=o 1'=0 1'=0
Example 2.11 If lim Wn = w, and {U n } does not converge, can the following
n--+oo
sequences can be convergent?
Wn
C) - .
Un
Solution. a) Never. If {u n } is unbounded then the inequality
implies that IW n + U n I is also unbounded, and the sequence {w n + u n } is not conver-

gent. Examine the case when {u n } is bounded but it has few accumulation points.
b) Take Un = n and Wn = A.
n
Then {w n u n } converges to O.
c) Take U n = n and {w n } arbitrary convergent sequence. Then {w n u n } converges

to O.
Example 2.12 Find

a) one value of
J+ -1 (1- z)J-1 + (1 - 2z)V-1 + (1 - 3z)vf-1 + ... ;
b) one real value of
zJ -1 + 2zJ -1 + 3zV-1 + 4zvf-1 + ....
Solution.
a) and b) are complex generalizations of the equality
a) For n E N we have
n(n - 2z) nV-1 + (n - z)(n - 3z)
nJ-1 + (n - z)J-1 + (n - 2z)(n - 4z)
nJ-1 + (n - z)J-1 + (n - 2z)J-1 + (n - 3z)(n - 5z) .

2.1. SEQUENCES 43
Letting this proceded to infinity we obtain
n(n - 2z) = nJ-l + (n - z)J-l + (n - 2z)V-l + (n - 3z)V-1 + ....

Putting n = 1, we obtain
J -1 + (1 - z)V -1 + (1 - 2z)V-l + ... = 1 - 2z.
b) For z E C we have
z(z - 2z) = zV-l + (z - z)(z - 3z)
= zJ-1 + (z - z)V-l + (z - 2z)(z - 4z)
= zV-1 + (z - z) J -1 + (z - 2z )V-1 + (z - 3z)( z - 5z) .
Letting this proceded to infinity we obtain
z(z - 2z) = ZV-l + (z - z)J-l + (z - 2z)V-l + (z - 3z)V-l + ....
Putting z = z we obtain
zV-1 + 2zJ-1 + 3zV-1 + 4zV-I + ... = -3.
Remark. For real case we have that
Jl + xVI + (x + 1)~ = f(x) (x E IR)
satisfies the functional equation
1 + x f(x + 1) = f2(x),
whose solution is f(x) = x + 1.

Example 2.13 Let the members of the sequence ao, al, ... , an, ... can take only the
values -1,0,1. Prove that
Solution. The form
is meaningful, since
By induction we prove the equality
= 2 sm "4 E
• ( 11" n aOal··· a k )
Zn 2k • (2.2)
k=O
We have
sign Zn = sign ( 2 sin ( 4"11" E
n aOal··· a k ) )
2k = ao·
For ao ":f 0 we have
Z~ - 2 = al J+ 2 a2,/2 + ... + an V2 ,
and on other side by (2.2)
Z~ - 2
Exercise 2.14 Prove that for x, -2 :$ x :$ 2 we have
x = ao ,/1 + an/2+~ ,
where ao, ab ... ,a n , ••• take only one of the value either -1 or 1.
2.2 Series
2.2.1 Preliminaries
An infinite series L:~=l Zn is called (ordinary) convergent if the sequence {Sn} of its
partial sums given by
2.2. SERIES 45
converges. If E~=l IZn I converges we say that E~=l Zn absolutely converges.

A series E~=l Zn is convergent if and only if for every c > 0 there exists no E N
such that for all n, m ~ no
Theorem 2.3 (Cauchy criterion) IflimsuPn--+oo yllznl < 1, then the series
E~=l Zn converges absolutely. If limsuPn--+oo y'lznl > 1, then the series E~=l Zn
diverges absolutely.
Theorem 2.4 (D'Alembert criterion) IflimsuPn--+oo IZ:!11 < 1, then the series
L~=lZn converges absolutely. Ifliminfn--+oolz:!ll > 1, then the series E~=lZn di-
verges absolutely.
We have for any sequence {zn}

liminf IZn+ll :$ liminf
n-+oo Zn n-+oo
y'lznl :$limsup
n-...oo
yllznl :$limsup
n--+oo
I. Zn+ll·
Zn
Theorem 2.5 (Raabe criterion) IflimsuPn--+oo n (IZ~n-ll-1) < -1, then the se-
ries E:: 1 Zn converges absolutely.
Für two series E~=o U n and E~=o V n we define their multiplication as aseries

Example 2.15 Examine the absolute and ordinary convergence of the following se-
ries with respect to the real parameter a.
1 + zn an + zn
E-,-;
00 00
E~();
a) n=2 n n
c)
n.
n=O
E
00 ,
d) n.
n=O (n + z)n'
00 eIn zn
E-; E n·
00
g) h)
n=l n n=lln nTI
Solution. a) an b). Diverges absolutely and ordinary for all values of a.

c) and d). Converges absolutely for aB a.
e) Diverges absolutely, but converges ordinary, since
as k ----+ 00.
f) and g). Diverges absolutely, but converges ordinary.
h) Diverges.
Example 2.16 Let

(
w) = w( w - 1) ... (w - n + 1)
n n!
for w E !C, n E N. Find for which complex w the series
converges.
Solution. Let w = x + zy. We obtain by D' Alembert criterion
w(w - 1) .. · (w - n - l)(w - n)
lim
n--+oo
(n + I)!
w(w-1)· .. (w-n-l)
.
I1m --
n +1
n--+oo
IW - nl
n!
. Jx 2 - 2xn + n 2 + y2
= IIm - - - - - - - -
n--+oo n +1
1.
Therefore we can not decide by D'Alembert criterion about the convergence of the
given series except we aBow that w depends of n. Namely, taking w(n) = n + zy for
an arbitrary but fixed real y the series
converges absolutely.
Taking the Raabe criterion we obtain
n(I:~~I-l)
_n_ (J'-x- 2 -+-y-2 - (n
2 ---2-n-x-+-n- + 1))
n+l
2.2. SERIES 47
n Iwl 2+ n 2 - 2nx - (n 2 + 2n + 1)
= n+l Vlwl2 + n 2 - 2nx + n + 1
Iwl 2
1 n - 2(x + 1) - n 1
as n - t 00. By Raabe criterion the series absolutely converges for -(x + 1) < -1,
i.e., Rew > O. The series converges ordinary for -1 < Rew ~ O.
Remark. Analogously the hypergeometrie series
absolutely converges for Re (w + U - v) < O.

Example 2.17 Prove that the series
converges if the se ries Ek:OUk absolutely converges and the sequence {ak} mono-
tonically decreasingly converges.
Solution. We have
= I'f
k=n+l
uk(ak - ak+l + ak+ 1 )1
< I'f
k=n+l
uk(ak - a k+1 )1 + 'f Uk ak+ll·
Ik=n+l
Since the series Ek:o Uk converges, we have IUkl ~ M and for every e > 0 there
exists no E N such that E~!~+1lukl < e for all n ~ no(E) and pE N. Therefore
(2.3)
since lan+11 ~ lanI for k = n + 1, ... , n + p. The series Ek~O( ak - ak+d absolutely
converges by
n+p n+p
L lak - ak+ll = L (ak - ak+d = an+l - an+p+1 -t 0 as n - t 00,
k=n+l k=n+l
since the sequence {an} converges and therefore it is a Cauchy sequence. Hence by
(2.3) the series
is convergent.
Exercise 2.18 Prove that if the series
converges and Iarg ukl :::; <P < ~, then this series absolutely converges.
Exercise 2.19 Let 00 00
L Un and LU!
n=l n=l
be convergent series. Prove the for Re U n ;::: 0 the series
00
converges.
Exercise 2.20 Examine the convergence of the following series

00 III
a) L~j b) c)
n=l n
Answers.
a) Diverges. b) Diverges. c) Converges absolutely.
Example 2.21 Prove that the set of all complex series S with respect to addition
of series, i.e.,
00 00 00
LUk+ LVk = L(Uk+Vk),

k=O k=O k=O
is an abelian group.
00 00
Solution. By the definition the sum of series E Uk and E Vk is again a complex
k=O k=O
E (Uk + Vk).
00
series
k=O
2.2. SERIES 49
Associativity:
E (E E
Uk + Vk + Wk)
00
L Uk
k=O
00
00
+ :~:JVk + Wk)
k=O
k=O
00
k=O
00 00
L(Uk+Vk)+ LWk
k=O k=O
(E E E
Uk + Vk) + Wb
where we have used the associativity of the addition of complex numbers.

The neutral element is the zero series, L:~=o 0 = o. The inverse element of the series
L:k=O Uk is the series L:~o( -Uk).
Example 2.22 11 one series converges and the other diverges can their sum con-
verges?
Solution. No. Apply Example 2.11.
Example 2.23 11 both series diverges can their sum converges?
Solution. Yes, e.g., L:k=o t and L:~o( -t)·

Example 2.24 Find two convergent (divergent) series whose sum is the series
Solution. We have
00 (l-z/ cos 71rk
00 sin 71rk
00
L k =L~+zL~'
k=O 2 k=O 2 2 k=O 22
where on the right side are two convergent series.
We have
where on the right side are two divergent series.

Exercise 2.25 Prove D'Alembert and Cauchy criterions for the absolute conver-
gence of aseries.
Example 2.26 Prove that the condition limsuPn->oo IW~±l1

n
< 1 in D'Alembert cri-
terion of the convergence of series is not necessary.
Solution. Take the real series
a + b2 + a3 + b4 + ... for 0 < a < b< 1.
Then lim SUPn->oo IW~;;l I > 1 hut the series converges, since
Ea + E b2k ,
00 00
a + b2 + a3 + b4 + ... = 2k+1
k=O k=l
and hoth series on the right side converge.
Exercise 2.27 Apply Cauchy and D'Alembert criterions on the series
Z
2+ (Z)2
3" + (Z)3
2 + (Z)4
3" + ....
a) The series E~
k +1
k=O
not converges absolutely, but it converges ordinary.
b) Using
for p ~ k, prove that

1
Ek
>1.
p=oV(k-p+l)(p+l) -
c) Multiplying the series E~

k=O
k+l
by itself we obtaining a divergent series.
Solution. a) Series Lk>O k not converges ahsolutely, since Ikl

k+l >
1
7C'
t - k+l
I~ .;2, and k '\. o.

and the series Lk::l diverges.
The ordinary convergence follows by Lk=O zk I
2.2. SERIES 51
b) We obtain
t 1
p=oy'(k-p+1)(p+1)
~ t ~=
p=02"-1
2( k
k-2
+ 1) ~ 1 for k > 2.
c) Since
we obtain by b) for the member of series
Hence it does not converge to zero what is a necessary condition for the convergence
of aseries.
Example 2.29 Find the sums 0/ the /ollowing series:
Solution. a) Since
~
~ (~+2kl)k = ~
~ (~)
2 ,ancl I~I
2 < 1,
k=O k=O
the series converges and has the sum 1 ~ ~ = 1 + ~.

b) Multiplying the series
with itself we obtain

Therefore by a)
c) Using a) and b) we obtain

+ 1)" + 1)"-1
E 00 (
l k • E k·
00 (.
• = E p. (.• + 1)"-1 = -1 E k(k -
E k-l
00 00 ( + 1)"-1
1 )-,--l----o--''----
"=1 2 k=1 2"-1 k=2 p=1 2k - 1 2 "=2 2k - 1 '
where we have used the equality
n(n + 1)
1+2+···+n= .
2
Hence
(l + 1)"-2
E k( k -
00
k=2
1)
2
k-2 = -4 + 4l.
d) Since
z+ 1 (f k(k + 1) (z + 1)"-1 _ f k (z +"~lk-1) f k = 2 • (z +,,1)k,

2 k=1 1 k - 2 k=1 k=1 2
we obtain by b) and c)
fk 2 .(z+1)" =-3-z.
2k "=1
Remark. The preceding sums can be obtained also using the power series
zk
E k2
k=O
for z = 1 + z,
see Chapter 6.
Exercise 2.30 For a real find:
Ek
00 2 (
a )k
k=1 a +z
Answer. a(a + z)( -2a 2 - 2az + 1).
Exercise 2.31 Prove that there exists a sequence {zn} of complex numbers such
that the senes
n=O
converge and absolutely diverge for all k.
Answer . Zn = e 27r1a / ln( n + 1) for airrational.

Chapter 3
Complex functions
3.1 General Properties
3.1.1 Preliminaries
Let z(t) = x(t) + zy(t), a ~ t ~ b. The curve z(t) is a path if z'(t) = x'(t) + zy'(t)
exists and it is continuous on each subinterval of a finite partition of [a, b], and
z'(t) "1= 0 except at a finite number of points.
Let J : A ...... C and Zo is an accumulation point of A. We say that w is a limit of

J at Zo through A, in the notation
lim J(z) = w,
z.....,.zo,zEA
if for every e > 0 there exists 6 > 0 such that if z E A and Iz - zol < 6, then
IJ(z) - wl < e. We usually omit the symbol z E A under the limit sign.
Let Zo E A. We say that J is continuous at Zo if z-+zo
lim J(z) = J(zo).
Definition 3.1 A complex Junction J defined in a neighborhood oJ a point z E C is

differentiable at z iJ
r J(z+h)-J(z)
h~ h
exists and then is denoted by f'(z).
If J and gare both differentiable at z then also the functions J + g, Jg, ~ (for
g(z) "1= 0) are differentiable at z and
(J + g)'(z) = j'(z) + g'(z),
53

54 CHAPTER 3. COMPLEX FUNCTIONS
(fg)'(z) = J'(z)g(z) + J(z)g'(z),

(L)'
9
(z) = J'(z)g(z) - J(z)g'(z).
g2(z)
Theorem 3.2 IJ J = u+w is differentiable at z = (x, y), then there exist the partial
derivatives
äu äu äv äv
äx ' äy , äx ' äy
and they satisJy the Cauchy-Riemann equations at z
äu äv
äx äy'
äu äv
,
äy äx
or, equivalently
If the partial derivatives

äu äu äv äv
äx ' äy , äx ' äy
are continuous at z and satisfy the Cauchy-Riemann equations at z, then J is
differentiable at z.
Definition 3.3 A Junction is analytic at z iJ J is differentiable in a neighborhood

oJ z. J is analytic on a set Ace iJ J is differentiable at all points oJ an open set
containing A. An everywhere differentiable Junction is called an entire Junction.
Example 3.1 Find the real and imaginary parts oJ the Jollowing Junctions:
a) Z2 + Zj b) v'Z+Tj c) zn j d) yZj e) 2z 2 - 3z + 5j J) (z - l)/(z + 1).
Solution. a) We write the function J(z) = Z2 + z in the form

J(z) = u(x, y) + w(x, y),
where u(x, y) = x 2 - y2 and v(x, y) = 1 + 2xy for z = x + zy.
b) We write the function J(z) = v'Z+T in the form J = u + w, where u and v
are given by the system of equations
u2 - v2 = X + 1 and 2uv = y
3.1. GENERAL PROPERTIES 55
for z = x + zy.
c) We have f( z) = zn = u(p, tJ) + zv(p, tJ) for z = p( cos tJ + z sin tJ). We have
u(p, tJ) = pR cos ntJ and v(p, tJ) = pR sin ntJ.

d) Taking
f(z) = vrz = u(p, tJ) + zv(p, tJ),
for z = p( cos tJ + z sin tJ), we obtain
u(p, tJ) = ~ cos tJ +n2k1r and v(p, tJ)

. tJ + 2k1r
= ~ sm ---
n
for k = 0,1,2, ... , n - 1.
e) Taking
f(z) = 2z 2 - 3z + 5 = u(x,y) + w(x,y)
we obtain
u(x,y) = 2x 2 - 2y2 - 3x + 5 and v(x,y) = 4xy - 3y,
where z = x + zy.
f) Taking
z-l
f(z) = z+ 1 = u(x,y) + zv(x,y),
we obtain
z-l (z-l){z+l)
z+l = Iz + 11 2
for z = x + zy.
Example 3.2 Find the inverse mappings for the following functions
a) l/z, z 1= 0; b) (z -l)/(z + 1), z 1= -1; c) 2z2 + zz - z + 1; d) Z7 + 1 + z.
Solution. a) The inverse function for the function 1/ z, z 1= 0, is the same function.
b) The inverse function for the function (z - l)/(z + 1), z 1= -1, is the function
(-z -l)/(z - 1), z ~ 1.
c) The inverse mapping for the function 2z 2 + zz - z + 1 is the multi-valued
function
where the square root is taken in the complex plane.

d) The inverse mapping for the function Z7 + 1 + z is the multi-valued function
w=V'-1-z+z
for k = 0,1, ... ,6.
Exercise 3.3 Let I(z)~.

a) Find the images by the lunction 1 01 the lollowing lines:
1. x = C (constant), 2. y = C (constant), 3. Izl = R, 4. argz = a, 5. Iz -11 = 1.
b) Find the lines which are mapped by Ion: 1. u = C (constant), 2. v = C
(constant).
Hint. Find the functions u and v in the representation 1 = u + zu.

Answers. a)
1. Circle u 2 + v 2 - V = 0 for C # 0, and for C = 0 the axis u = o.

2. Circle u 2 + v 2 f; = 0 for C # 0, and for C = 0 the axis v = o.
3. Circle Iwl = l
4. argw = a.
5. The straight line u = ~.
b)
1. x 2 + y2 - V = 0 for C # 0, and x = 0 for C = o.

2. x 2 + y2 + -t = 0 for C # 0, and y = 0 for C = o.
Exercise 3.4 Explain geometrically the lollowing curves:

a) z = ut + vt\ 0 ~ t ~ 1;
b) z= l+zt tEIl~:,
1 +t +t2 '
V;t+1+,V;, 0~t~1
c) z(t) = {
v'2 v'2
T t + 1 - 'T' -1 ~ t ~ O.
Answers.
a) The straight line segment [0, u + v].
b) A closed curve.
c)
Example 3.5 Find the images of the circle Izl = R by the functions fl(z) = z +}
and fl(z) = z - }.
Solution. The function !t maps the circle Izl = R :f:. 1 on an ellipse

u2 v2
(R+ *)2 + (R- *)2 = 1,
and the circle Izl = 1 on the straight line segment v = 0, -2 :::; u :::; 2.
The function f2 maps the circle Izi = R :f:. 1 on an ellipse
u2 v2
1)2
(R-R + (R+R1)2 = 1,
and the circle Izl = 1 on the straight li ne segment u = 0, -2 :::; v :::; 2.
Example 3.6 Suppose that there exists lim z ..... zo fez) = w, then
lim fez) = Wj
a) %-+%0 lim Ref(z) = Rew
b) %-+%0 and lim Imf(z) = Imwj
%-+%0
c) limz ..... zo If(z)1 = Iwl·
Solution. The condition limz ..... zo fez) = w implies that for every c: > 0 there exists
S> 0 such that Iz - zol < Simplies If(z) - wl < c:.
a) We have
c: > If(z) - wl = If(z) - wl = If(z) - wl for Iz - zol < S(c:).
b) We have
c: > If(z) - wl ~ IRe (f(z) - w)1 = IRef(z) - Rewl,
and
c: > If(z) - wl ~ Im (f(z) - w)1 = IImJ(z) - Im wl, for Iz - zol < d(c:).
c) We have
c: > If(z) - wl ~ IJ(z)I-lwll for Iz - zol < S(c:).
Example 3.7 Suppose that there exists limz ..... zo Ij(z)1 = Iwl. For which values there
exists also limz ..... zo j(z)?
Solution. We have that limz ..... zo j(z) exists for w = 0, since e > Ilf(z)11 = If(z)1
for Iz - zol < b(e).
We shall show that this is the only case in general. Suppose that there exists
limz .....zo j(z) = A for w ~ O. Then by Example 3.6 we have lAI = w. We shall
examine when we have equality in the following inequality
Ilf(z)l- All::; If(z) - AI

(since then t: > Ilj(z) -lAll = If(z) - AI for Iz - zol < b(t:)). The equality would
imply
(lf(z)l- IAI)2 = (f(z) - A)(f(z) - A) = If(zW + IAI 2- 2Re (f(z) . A),
Re (f(z)A) = Ij(z)AI = Ij(z)AI,

and therefore
Re (f(z)A) ~ 0 and Im (f(z)A) = O.
Then f(z)A have to be real and non-negative, which is impossible for a general
complex function f. This is possible specially when j(z) = A or f real function with
a constant sign.
Example 3.8 Find m and n such that there exists
Solution. Taking z = l/w we obtain
For the limit to exist we have to have p - m ::; p - n, m ~ n.
Example 3.9 Find which of the following limits exist .
a) lim-
l-z b) l i m
Z2 _ Z2
l -;
z ..... 1 - z z-+1
l z -_;
+z c) l i m - - .
z-+o z
Solution. a) The limit does not exist, since for z = 1 - x we have

· 1 - z 1·Im 1 - 1 + x = 1 ,
1Im--=
z--+ 1 1 - z z--+o 1 - I + x
and for z = 1 + zy we have

lim 1 - z = lim I - I + zy = -1.
1- z
z--+ 1 y--+O 1 - I - zy
b) The limit exists and it is 1 (prove that using the polar form of the complex
number).
c) The limit exists and it is 0 (prove that).
Exercise 3.10 Can we modify the following functions at z = 0 such that the new
functions will be continuous at z = 0 ?
a) Rez; ) zRez
z c -I-zl-.
Answers. a) No. b) No. c) Yes, f(O) = O.
Exercise 3.11 Examine the continuity of the following functions in the unitdisc
a) _1_. b) _l_
1- z' 1 + Z2·
Answers. a) and b). They are continuous.
Example 3.12 Prove that x 2 - y2 + zxy is a polynomial in z = x + zy (analytic

polynomial) and x 2 + y2 - 2zxy is not.
Solution. We can write

x 2 _ y2 + 2zxy = (x + zy)2.
On the other side, if we suppose that
x 2 + y2 - 2zxy = an(x + zyt + ... + al(x + zy) + aa, (3.1 )
then for y = 0 we would obtain
Hence aa = 0, al = 0, a3 = 0, ... , an = 0 and a2 = 1. Therefore we would have by

(3.1)
wh at is a contradiction.
Second method. Apply the Cauchy-Riemann equations on the previous polynomials.
Example 3.13 Let f : Q -4 C be an analytic function, where Q is a region in

C. Prove that the functions u(x,y) = Ref(x + zy) and v(x,y) = Imf(x + zy) for
x + zy E Q satisfy
a) the Cauchy-Riemann equations

8u 8v 8u 8v
- and - .
8x - 8y 8y - - 8x'
b) the Laplace equation

8 2u 82 u
8x2 + 8 y2 = 0,
ij, additionally, u and v are from the class C2(Q).
Solution. a) Since the function fis analytic, there exists
J'(z) = lim fez + h) - fez)

h-O h
for z = x + zy E Q. We evaluate the preceding limit in two ways. First let h -+ 0
for real h. We have for h 1- 0
fez + h) - fez) fex + h + zy) - fex + zy)

h = h
u(x + h,y) - u(x,y) v(x + h,y) - v(x,y)
= h +l h .
Taking h -+ 0 we obtain
f '( ) _ 8u(x,y) 8v(x,y)

z - 8x +z 8x . (3.2)
Now let zh -+ 0 for real h. We have for h 1- 0
fez + zh) - fez)

=
fex + zeh + y)) -
fex + zy)
zh zh
u(x,y+h)-u(x,y) v(x,y+h)-v(x,y)
= -t h + h .
Letting h -+ 0 we obtain
f '( ) 8u(x,y) 8v(x,y)

z = -z 8y + 8y . (3.3)
Since both the real and imaginary parts of (3.2) and (3.3) must be equal, we obtain
the Cauchy-Riemann equations.
b) Differentiating the first Cauchy-Riemann equation with respect to x and the

second one with respect to y we obtain
(PU (Pv
- and
ox 2 - oxoy
Adding the obtained equalities we obtain that the function u satisfies the Lap1ace
equation. Differentiating now the first Cauchy-Riemann equation with respect to y
and the second one with respect to x and repeating the preceding procedure we find
that the function v satisfies the Lap1ace equation, too.
Example 3.14 Let a complex function f be dilferentiable at z. Find

. f(z) - f(w)
11m '--'---'---'--'--'-
z-+w Z - W
when z belongs to the straight line Re z = Re wand then when z belongs to the
straight line Im z = Im w. Prove that J = u + zv satisfies the Cauchy-Riemann
equations.
+ zy and w = a + zb, then the limit on Re z = Re w is given by

Solution. If z = x
· J(z)-J(w) = l'1m u(a,y)+zv(a,y)-u(a,b)-zv(a,b)
11m
a + zy - a - zb
--'--:"'::-'-_-O...~'"--_":"""";--:-'-_---''-'---'-
%-+w z- W y_b
· u(a,y)-u(a,b) l' y(a,y)-v(a,b)

11m
y_b z(y - b)
+ 1m ::""':"-:"::-'---::--'-'--'-
y-+b Y- b
= _z(OU)
oy x=a,y=b
+ (OX)
oy x=a,y=b
.
Analogous1y we obtain on Im z = Im w :
. J(z) - J(w)
11m '--'---'---'--'--'- 1. u(x, b) + zv(x, b) - u(a, b) - zv(a, b)
X + zb -
Im~~-'---o"'~~-~~----'~~
z-+w Z - W x-+a a - zb
= (OU)
ox x=a,y=b
+ z(OX)
ox ",=a,y=b
.
We obtain the Cauchy-Riemann equations taking equal the real and imaginary
parts, respective1y.
Example 3.15 Prove that the Junction VlZ2 - z21 satisfies the Cauchy-Riemann
equations at z = 0, but it is not dilferentiable at this point.
Solution. The function VI Z2 - z2 1 is identically zero on the real and imaginary

axes and therefore trivially satisfies the Cauchy-Riemann equations at z = O.
Taking z = r( cos 6 + z sin 6) and r -+ 0 we obtain
. VIZ2_Z21 . VI4cos6sin61
hm = hm .'
r_O Z r-O (cos6 + zsm6)
Since the last limit depends on (), take () = 0 and ()7r / 4, we conclude that the function
VIZ2 - z2 1 is not differentiable at z = o.
Example 3.16 Find by the definition the derivatives (in the regions where they
exist) of the following functions
n
z b) l+z.
a) Z2 + 1j z3 '
c) EUiZij d) Imzj e) Zj J) 14
i=O
z w
lim Z2 +1 +1 w2
%-+w w Z -
lim -wz{z-w)+z-w
z-+w (z - w)(z2 + 1)(w 2 + 1)
= lim (z - w){l - wz)

%-+w (z - W)(Z2 + 1)(w 2 + 1)
1- w 2
=
(w 2 + 1)2
for all w =I- z, -z.
Exercise 3.17 Find which of the following functions can be real or imaginary part
of a complex function f which is differentiable in the region Izl < 1.
Solution. a) We have for w(x, y) = x 2 - axy + y2 that
Therefore
()2w 82 w
8x 2 + 8 y2 = 2 + 2 =I- 0,
and the given function w can not be the real or imaginary part of a differentiable
function.
b) No. c) No. d) Yes.
Exercise 3.18 Find the constants a, b, and c such that the following fundions would
be analytic
a) fez) = x + ay + l(bx + cY)j

b) fez) = cosx(coshy + asinhy) + lsinx(coshy + bsinhy).
Answers.
a)c=l,b=-a, then f(z) = (l-al)z; b)a=b=-l, then f(z) =e
U •
Exercise 3.19 Find the region of analyticity of the function
Answer . The function is analytic for

0< argz < 7r/4 and 7r < argz < 57r/4,
then fez) = Z2 j and for
7r/2 < argz < 37r/4 and 37r/2 < argz < 77r/4
and then fez) = _z2.
Exercise 3.20 If we represent a complex analytic function f as
w(z) = f(x,y) =f z+ z-
( -2-'~i i) = F(z,z),_
and considering z and z as independent variables prove that
aw= -
1 (aw
- - law) aw _ ~ (aw aw)
-
az 2
-
ax ay
and ai - 2 ax + l ay .
Example 3.21 (Hadamard) Show that the function u = u(x,y) given by

e ny _ e- ny
u(x,y) = 2n 2 sinnx
for n E N is a solution on D = {(x, y) I x 2 + y2 < I} of the Cauchy problem for the

Laplace equation
au(x,O) _ sin nx
u(x,O) = 0, ay - n
Prove that this problem is not well-posed, i.e., a problem with a partial differential
equation is well-posed in a dass of functions C, if the following three conditions are
satisfied:
(i) there exists a solution in Cj
(ii) the solution is uniquej
(iii) the solution is eontinuously dependent on the given eonditions, e.g., initial~
values, boundary eonditions, eoeffieients, ete.
Solution. It easy to check that the function Un for an arbitrary but fixed n E N
given by
e ny _ e~ny
un(x, y) = sin nx 2
2n
is a solution of the given Cauchy problem. Letting n ---t 00 we obtain for x, y E D,
x f- 0, y f- 0
IUn(x, y)1 ---t 00.
On the other side, the given Cauchy problem for n ---t 00 reduces to the problem
82u 82u
8x2 + 8 y 2 = 0,
( ) = 0, 8u(x,0) =0
U x,O 8y ,
which has only a trivial solution U = O. Therefore the considered solution of the
given Cauchy problem does not depend continuously of the initial condition. Hence
it is not well-posed (it is ill-posed).
Remark . In contrast with this simple problem of a partial differential equation
which is not well-posed, let us remind some weIl known results from the theory of
ordinary differential equations where for general classes of problems with ordinary
differential equations the well-posedness can be ensured. For example, the well-
posedness of the Cauchy problem
y' = J(x,y), y(xo) = Yo

is ensured supposing that the function J is continuous and satisfies the Lipschitz
condition in some region which contains the point (xo,Yo).
3.2 Special Functions
3.2.1 Preliminaries
The function eZ •
The function e Z is the analytic solution of the functional equation
3.2. SPECIAL FUNCTIONS 65
which for every z = x real reduces on the function eX • This gives
eZ = eX cos y + ze X sin y
for z = x + zy. The function has the following properties

(i) lezi = e X ;
(ii) eZ =I- 0;
(iii) e'Y = cosy + zsiny;
(iv) the equation eZ = a has infinitely many solutions for any a =I- O.
(v) (e z )' = eZ •
The functions sin z and cos z.

We define
1
cos z = -( elZ + e-'Z).
2
We have (sinz)' = cosz and (sinz)' = -cosz.

Example 3.22 Prove that f(z) = eX cos y + zex sin y is the only analytic solution of
the functional equation f(zdf(Z2) = f(ZI + Z2) which satisfies the condition f(x) =
eX fOT all real x.
Solution. It is easy to check that the given function satisfies the functional equation
f(ZI)f(Z2) = f(ZI + Z2) and f(x) = eX for all real x.
Suppose now that f is an analytic solution of the functional equation which satisfies
the condition f(x) = eX for all real x. Then
f(z) = f(x + zy) = f(x)f(zy) = eXf(zy).

Taking f( zy) = u(y) + w(y) we obtain
f(z) = eXu(y) + zeXv(y).
Since f is analytic it satisfies the Cauchy-Riemann equations and therefore u(y) =

v'(y) and u'(y) = -v(y). Hence u" = -u. This ordinary differential equation has
the general solution
u(y) = a cos y + b sin y
for a and b real constants. Since f(x) = eX and
v(y) = -u'(y) = asiny - bcosy
we obtain for y = 0
u(O) = a = 1 and v(O) = -b = O.
Therefore v(y) = sin y and finally
f (z) = eX cos y + zex sin y.
Exercise 3.23 Prove that eZ is an entire function.
Exercise 3.24 Show that lezi = eX and eZ =I- O.
Hint. The equality follows by Example 3.22. For the second property use the
preceding equality and that eX =I- 0 for x real.
Example 3.25 Find the real and imaginary parts and modulus of
Solution. Since
We have arg e(2+"·/4)2 = 7r. Since e'" = cos 7r + zsin 7r = -1 we obtain
Exercise 3.26 Solve the equations: a) eZ = 1; b) eZ = 2; c) e% = 1 + 1.

Answers. a) (7r/2 + 2h)1, k = 0, ±1, ±2, ... ; b) In2 + 2hz, k = 0, ±1, ±2, ... ; c)
~(ln 2 + 7r /2) + 2h1, k = 0, ±1, ±2, ....
Exercise 3.27 Prove the identities
a) sin2 + cos 2 z = 1;
b) sin2z = 2sinzcosz;
c) (cos z)' = - sin z.
3.2. SPECIAL FUNCTIONS 67
Example 3.28 Find the real and imaginary parts, and modulus of the following
functions: sin z, cos z, tan z, cot z.
Solution. For sin z and cos z see chapter 3.

It is easy to prove that
tan u + tan v
tan (u + v ) = - - - - - - for u, v E (C.
1 - tan u . tan v
Therefore
tanx + tan iy
tan z = tan( x + zy) = ,
1 - tan x . tan zy
and since tan zy = z tanh y we obtain
Re( tanz ) = tanx(1+tanh2y)

2 ,
2 1 + tan xtanh y
Im (tanz ) = tanhy(1-tan2x) 2 ,
1 + tan 2 xtanh y
Itan zl = tan 2 x + tanh 2 y.
Exercise 3.29 Prove that
sin(zz) = zsinhz, cos(zz) = coshz,
tan(zz) = danhz, cot(zz) = -zcothz.
Exercise 3.30 Solve the equation sin z = 3.
Hint. Solve first the equation v = e'Z for v.
Exercise 3.31 Prove that for every u, v and z

a) sin( u + v) = sinu cos u + sin v cos Uj
b) cos(u + v) = cosucosv - sinusinv.
Hint. a), b) . Putting
elZ e- IZ e lZ + e-' Z
= = ----
_
smz and cosz
2 2z
we directly verify the desired equalities.
3.3 Multi-valued functions
3.3.1 Preliminaries
A point Zo is a branching point of a multi-valued function w = j(z) if taking a
closed path around Zo the image of this path in the w-plane is not a closed path. If
after going n times around a branching point Zo the path in w-plane is closed, then
Zo is an algebraic branching point of order n. If after going an unlimited number
of times around a branching point Zo the path in w-plane is never closed, then Zo
is a transcendental branching point of order n. Cutting is the straight line segment
connecting two branching points ( we can go through z = 00).
The junction Log z
The multi-valued function Logz as an inverse mapping of eZ is given by
Logz = logz+ 2kn, k = 0,±1,±2, ... ,

where log z is the logarithm principal value (branch) of Logz given by
log z = log Izl + zarg z, -7r < arg z ~ 7r.
for -7r < arg z < 7r when the cutting is the negative real axis or 0 < arg z < 27r
when the cutting is the positive real axis.
For each k we have a branch of the multi-valued function Log z, which is a
function.
We define for w f- 0 and z the multi-valued function

Example 3.32 Find Log z and log z, jor
a) 1+z; b) 1-z; c) -1+z; d) -1-z.
Solution. In the complex plane without the positive real axis we obtain:
1 7r
a) log(l + z) = "2 log 2 + "4z,
Log (1 + z) = ~ log 2 + (~+ 2h) z (k = 0, ±1, ±2, ... ).

1
"2 log 2 + TZ,
77r
b) log(l - z) =
3.3. MULTI- VAL UED FUNCTIONS 69
Log (1 - z) = ~ log 2 + C: + 2k1r) z (k = 0, ±1, ±2, ... ).

1 311"
c) log(-l + z) = 2 log2 + 4z,
Log ( -1 + z) = ~ log 2 + (3: + 2k1r) z (k = 0, ±1, ±2, ... ).

1 511"
d) log(-l-z) = 2 1og2 + 4z,
1
Log(-1-z)=2Iog2+ (511"
4+2k1r ) z (k=0,±1,±2, ... ).
Example 3.33 Find the mistake in the J.I.Bernoulli paradox chain of reasoning
i) Log (_Z)2 = Log(Z2).
ii) Log (-z) + Log (-z) = Logz + Logz.
iii) 2Log (-z) = 2Log z.
iv) Log (-z) = Log z for z =I- 0.
Solution. We have
Logz = log Izl + zarg z + 2k1l"z,
and
Log( -z) log I - z I + zarg z + zarg ( -1) + 2k1rz

log Izl + zarg z + (2k + 1)1I"Z, k = 0, ±1, ±2, ...
We see that no value of Logz coincides with any value of Log( -z).
The mistake was made in going from ii) to iii) since Log(-z) + Log(-z) is not
equal 2Log( -z). Namely, Log( -z) + Log( -z) is sum of any two numbers from the
set of values Log(-z), and 2Log(-z) is a sum of the same value of Log(-z) with
itself.
We can illustrate this property with the following simple example. Take A =
{O, I}, then the set A + A consists of three elements: °°
°°
+ = 0, 0+1 = 1 and
1 + 1 = 2 and the set 2A consists of two elements: 2· = and 2· 1 = 2.
Exercise 3.34 Is it true the equality log( u . v) = log u + log v? Find log( -1 - Z)2.
Is it true the equality log( -1 - Z)2 = 2log( -1 - z)?
Example 3.35 Prove that wz, w =I- 0, for z = p/q rational has most q values.
Solution. Follows from
Example 3.36 Find F(O), F(l), F( -1) where F(z) is the bmnch of the function
~z -z which for z = 1 + z takes the value 1.
Solution. First we shall find the desired branch. We have
for k = 0,1,2,3. The desired branch for which Fk (l + z) = 1 is given by

Fo(z) = e(ln!z-.!+ar g (z-.).)/4.
Therefore
Fo(O) = e37r'/ 8 ;
Fo(l) = e(1n2)/8 e71n/16;
Fo ( -1) = e(1n2)/8 e 57r ./ 16 •

1
Arccos z = -Log
z
z(z ± Vz2=1).
Find Arcsin z.
Exercise 3.38 Prove that all bmnches of Arccos z are given by
Arccos z = arccos z + 2br for k = 0, ±1, ±2, ....
Hint. Use Example 3.37.
Example 3.39 Find the real and imaginary parts, and modulus of the following
functions: sin z, cos z, tan z, cot z.
Solution. For sin z and cos z see chapter 3.

It is easy to prove that
tan u + tan v
t an ( u + v ) = .,.------ for u,v E C.
1 - tan u . tan v
3.3. MULTI- VALUED FUNCTIONS 71
Therefore
tanx + taniy
tan z = tan( x + zy) = 1 '
- tanx· tan zy
and since tan zy = z tanh y we obtain
2
Re( tanz ) = tanx(I +tanh 2y) ,
2
1 + tan x tanh y
I m (tanz ) = tanhy(I-tan2x) 2 ,
1 + tan2 x tanh y
Jtan zJ = tan 2 x + tanh 2 y.

sin(zz) = zsinhz, cos(zz) = coshz,
tan(zz) = ztanhz, cot(zz) = -zcothz.
Exercise 3.41 Find the branching points 0/ the /ollowing /unctions a) VZj b)
Vz2=1j c) Arcsinzj d) Logz(1 + zz)J(I- zZ)j e) J(z - 5Hz - zHz - 2z + 3).
Answers. a) z = 0 and z = 00 are algebraic branching points of the order n.

b) z = 1 and z = -1 are algebraic branching points of the order 2.
c) z = 1 and z = -1 are algebraic branching points of the order 2, and z = 00
is a transcendental branching point.
d) z = z and z = -z are transcendental branching points.
a) z = 5, z = z, z = 2z - 3 and z = 00 are algebraic branching points of the order
2.
Exercise 3.42 Find the sets where are the branches 0/ the multi-valued /unction
W= J(1 - z2)(1 - p z 2), 0< k< 1.
Answer. The branching points are ±I and ±IJk. The point z = 00 is not a
branching point. We can take the cuttings [-l,-I] and [1, l] from the complex
plane. The second solution: we can take the cuttings [00, -7}], [-1,1] and [l,oo]
from the complex plane.

z z+ z
Arctgz = -Log--.
2 z- z
Example 3.44 The brunches of the function
z-l
Log--
z+l
are separuted by the cuttings [=, -1] and [1, =]. Let w( z) be the brunch of the
preceding function which is real on the upper part of the cut [1, =]. Find w( -2z).
Solution. To find w( -2z) we have to calculate arg (z-l) and arg (z+ 1) at z = -2z
when z moves through a path from a point from (-1, 1) up to the point - 2z not
crossing the cuttings. So we obtain arg(z - 1) = 57r/4 and arg(z + 1) = -7r/4.
Therefore
z -1 3
arg-- = -7r.
Z +1 2
Hence w( -2z) = In 2 + Z~7r.
Exercise 3.45 The brunches of the function ZZ are separuted by the cut [-=,0].
Let w(z) be the brunch of the preceding function for wh ich w(l) = 1. Find w( -e+zO)
and w( -e - zO), where -e + zO and -e - zO denotes the value at the brunch upper
and down of the cut, respectively.
Answer. w( -e + zO) = a- e (l+l) and w( -e - zO) = e-e(l-I).

Chapter 4
Conformal mappings
4.1 Basics
4.1.1 Preliminaries
Let z(t) = x(t) + ty(t), a $ t $ b, be a path. We suppose throughout this chapter
that z'(t) #- 0 for all t. Let 1 be defined in a neighborhood of a point Zo E ((:. 1 is
conformal at Zo if 1 preserves angles at Zo, for any two paths PI and P2 intersecting
at Zo the angle from PI to P2 at Zo , the angle oriented counterclockwise from the
tangent line of PI at Zo to the tangent line of P2 at Zo, is equal to angle between
I(Pd and I(P2 ). The function 1 is conformal in a region 0 if it is conformal at all
points from O.
Theorem 4.1 111 is analytic at Zo and f'(zo) #- 0, then 1 is conlormal at zoo
Definition 4.2 A 1 - 1 analytic mapping is called a conlormal mapping. Two

regions 0 1 and O2 are conlormally equivalent il there exists a conlormal mapping
lrom D 1 onto D 2 •
Theorem 4.3 (Riemann Mapping Theorem) For any simply connected region
o #- ICand Zo E 0, there exists a unique conlormal mapping 1 010 onto the unit
disc D(O, 1) such that I(zo) = 0 and f'(zo) > O.

Example 4.1 Examine the conlormality ollunctions
a) I(z) = z2 j b) eZ •
73

74 CHAPTER 4. CONFORMAL MAPPINGS
Solution. a) The function j(z) = Z2 is not conformal at z = 0, it maps the angle

between positive real axis and positive imaginary axis on the angle between positive
real axis and negative real axis.
Since 1'(z) = 2z -:j; 0 für z -:j; we conclude by Theorem 4.1 that f(z) = Z2 is conformal
for all z -:j; o.
Second method. Taking f = u + zu we have that the preimages of the curves u = rl
and v = r2 for rl and r2 different from zero have to be orthogonal. Namely, since
we have for z = x + zy
u(z) = x 2 - y2 and v(z) = 2xy,
the desired preimiges are the orthogonal families of hyperbolas
b) Since 1'(z) = eZ and it is always different from zero we conclude by Theorem 4.1
that j(z) = e is everywhere conformal.
Z
Exercise 4.2 Prove that if f is analytic at Zo and !'(zo) -:j; 0, then j is conformal
at zoo
Example 4.3 Prove that a 1 - -1 analytic function a region 0 is conformal as weil

its inverse function f- I which is also analytic.
Solution. Since f is 1 - -1 we have l' -:j; O. Hence f is conformal and f- I IS

analytical. Since (1-1)' = 1/1' we obtain (I-I)' -:j; O. Hence f- 1 is conformal.
Exercise 4.4 Prove that the relation conformal equivalence is an equivalence rela-
tion.
4.2 Special mappings
4.2.1 Preliminaries
1) Linear transformation
W = az +b
can be viewed as a composition of three mappings: WI = lalz (magnification for laD;
W2 = e· 9 z, () = arga (rotation through the angle theta); and W3 = Z + b (translation
for b); as W = W3 0 W2 0 WI.
4.2. SPECIAL MAPPINGS 75
Power transformation
w = za, areal.
It maps the wedge {z I ()l < arg z < ()d onto the wedge {w I a()l < arg w < a()2}. For
()2 - ()l ::; 2: it is a conformal mapping.
3) Exponential transformation
It maps the strip {x + ty IYl < Y < Yd onto the wedge {wl Yl < arg w < Y2}. For
Y2 - Yl ::; 21T it is 1 - -1.
4) The bilinear (Möbius) transformation
az + b
w - -- ad - bc -:j:. O.
- cz + d'
It is conformal and 1 - -1.
Theorem 4.4 Bilinear transformation maps circles and lines onto circles and lines.

Example 4.5 Consider the linear fractional transformation
t-z
w=--.
t+z
Find the regions in w-plane which are images of the following regions in the z-plane
a) Imz ~ 0; b) Imz ~ O,Rez ~ 0; c) Izl < 1.
Solution. The linear fractional transformation w = (t -z)j(t+z) or represented

by the matrix
[~1 :]
can be written in the form
2t
w=-l+--
z+t
or in the matrix form
[ ~1 :] = [~t ~ 1 ] [~ ~] [~ :].
a) First the region {z I Im z :::: O} (Figure 4.1)
o
Figure 4.1 Imz :::: 0
is translated for l (Figure 4.2) by the transformation
/ '\
I \
I u,
/
./
Figure 4.2 Im Wl :::: 1

in the wrplane.
Then the transformation
maps the Ene Im Wl = 1 on the circle ( Figure 4.3)
IW2 + z/21 = 1/2,

since the line has common point z = z with the unit circle (this point goes in the
point z = -z) and the point z = 00 go es in the point O.
Figure 4.3 IW2 + z/21 = 1/2

The image in the w2-plane is
(Figure 4.4). Finally, the transformation
maps the region

on the unit disc Iwl ~ 1 ( first rotating for ~ the starting region, then multiplying
it by 2 and translating it by -1, Figure 4.4).
~"-,.'-.Q----------- -- - - .
Figure 4.4 IW2 + z/21 ~ 1/2 and Iwl ~ 1
b) In the analogous way as in the preceding example we obtain
Figure 4.5 Imz ~ O,Rez ~ 0
that the linear fractional transformation
z-z
w---
- z+z·
maps the region Imz ~ 0, Re z ~ 0 ( Figure 4.5) in the z-plane

on the half-disc Iwl < 1,Im ~ 0 (Figure 4.6).
0, u
i
Figure 4.6 Iwl < 1, Imz ~ 0
c) The unit disc Izl ::; 1 (Figure 4.7)
"'y
!
z
Figure 4.7 Izl ::; 1
first is translated by z with the transformation

(Figure 4.8) in the wl-plane.
u,
Figure 4.8
The transformation
[~z ~1]
maps the circle (Figure 4.8) IWI - zl = 1 on the line Im W2 = -1/2 since the circle
cross the point 0 (and this point goes to z = 00) and met the unit circle in the
points ../3/2 + z/2 and -../3/2 +z/2 which are transformed in the points ../3/2 - z/2
and -../3/2 - z/2, respectively.
Figure 4.9 Imw2 ~ -1/2
In this way the disc IWI - zl ~ 1 is mapped on the region Im W2 ~ -1/2 (Figure
4.9). Finally, the transformation
maps the region Im W2 ~ -~ on the half-plane Re w ~ 0 ( rotated by ~, then

multiplying it by 2 and then translated for -1, Figure 4.10).
Figure 4.10 Rew ~ 0
Exercise 4.6 Find the regions in w-plane which are images of the following regions
in the z-plane by the following maps
a) Half-disc Izl < 1, Imz > 0 by transformation w = ~z+-;;.
b) Half-disc Izl < R, Imz > 0 by transformation w = ~:; ~.
c) The region 0 < 0 < 'Ir / 4 by transformation w = z ~ 1.
Figure 4.11 Izl < 1, Imz > 0
d) The ring 1 < Izl < 2 by transformation w = z/(z -1).

Answers. a) The half-disc Izl < 1, Imz > 0 (Figure 4.11) is transformed on
the region which contains the point w = 0 and it is bounded by circles Iwl = 1 and
Iw + 5t/41 = 3/4 (Figure 4.12).
1 u
Figure 4.12
b) The half-disc Iz/ < R, Imz > 0 (Figure 4.13) is transformed
Figure 4.13 /z/ < R, Imz > 0

on the region Rew < 0, Imw > 0 (Figure 4.14).
Figure 4.14 Re w < 0, Im w > 0
c) The region 0 < 0 < 7r/4 (Figure 4.15) is transformed
o
Figure 4.15 0 < 0 < 7r/4
on the half-plane Im w < 0 without the disc Iw -1 + 2-1 ::; {} (Figure 4.16).
Figure 4.16
d) The ring 1 < Izl < 2 (Figure 4.17) is transformed
Figure 4.17 1 < Izl < 2

on the region bounded by the line Rew = ~ and the circle Iw - 4/31 = 2/3
(Figure 4.18).
f
o
Figure 4.18
Exercise 4.7 Using the progmm Mathematica or Maple V (or some similar com-
puter program) plot the image of the lines z = x + zy, x = k!:l.x, y = k!:l.y for fixed
!:l.x and !:l.y in the w-plane by the function a) w =~; b) w = 2z/{1 + Z2).
Answer. a) w = 1/ z
Figure 4.19
Example 4.8 Prove the uniqueness of the bilinear (Möbius) transformation which
maps the points Zb Z2, Z3 on the points Wb W2, W3 (i =/:- j => Zi =/:- Zj 1\ Wi =/:- Wj),
respectively.
Solution. We introduce the so called anharmonic relation between four points

Zi, i = 1,2,3,4:
Z3 - Zl Z4 - Zl
(Zl>Z2,Z3,Z4) = - - - : - - - .
Z3 - Z2 Z4 - Z2
Abilinear transformation saves the anharmonic relation.
Namely, let
aZi + b . az + b
Wi = ---d for z = 1,2,3,4, where W = - -
CZi + CZ + d
is the given bilinear transformation.
Then we have
(ad - bC)(Z3 - Zl) (ad - bC)(Z3 - Z2)
W3- W1 = and
W3 - W2 = ( CZ3 + d)( CZ2 + d)'
(CZ3 + d)(CZ1 + d)
Therefore
W3 - W1 _ (Z3 - Zl)(CZ2 + d)
W3 - W2 - (Z3 - Z2)(CZ1 + d)'
We obtain analogously
W4 - W1 (Z4 - Zl)(CZ2 + d)
W4 - W2 = (Z4 - Z2)(CZ1 + d)'
The last two equalities imply (Zl,Z2,Z3,Z4) = (W1,W2,W3,W4)' Putting Z4 = Z (z is
an arbitrary point) and exchange the positions of Z3 and Z4 we obtain
Z - Z2 Z3 - Z2 W - W2 W3 - W2
" 1mp I'les t h at theb'I'

ThlS l mear translOrmatIon
1"
W
az-+db 1S
= - . umque.
.
CZ+
Example 4.9 a) Find the image of the unit circle Izi = 1 by the following trans-
formation
Z-v
w=u·----
vz - 1
' u,vEC and Izl=/:-l.
b) Which bilinear transformation maps the unit circle Izi = 1 on the unit circle
Iwi = I? What is the image of the unit disc Izi :5 1 by this transformation?
Solution. a) Since
zz - vz - vz + vv
w·w=u·iJ.·-------
vvzz - vz - vz + 1 '
for z . z = 1 we have that w . W = u· iJ.. Therefore the unit circle Izl = 1 is mapped
on the circle Iwl = lul.
b) Using a) we obtain that the bilinear transformation
z-v
w=u·-_-- v=/:-l,
vz -1
maps the unit circle Izl = 1 on the unit circle Iwl = 1, if and only if lul = 1, for
u = eO. (0 :::: B :::: 211"). The desired bilinear transformation is
z- v
w = e0. . - - lvi
vz -1'
=/:- 1.
The unit disc Izl :::: 1, for lvi< 1, is mapped on the unit disc Iwl :::: 1, and for
lvi> 1 on the region Iwl 2 1.
Exercise 4.10 Find abilinear transformation whieh maps points -l,z, 1 +z on the
following points: a) 0, 2z, 1 - Zj b) z, 00, 1, respectively.
Hint. See Example 4.8.

Answers.
a) w = -2z(z + 1) j b) w=(1+2z)z+6-3z.
-4z - 1 - 5z 5(z - z)
· 4.11 A bz'mear
E xerClse 1'· 1
transjormation az-+
w = - db' 1a db 1-1
-r 0 is perio die if
ez + c
there exists a natural number n such that
[; ~r=[; ~r=[~ ~].

Find whieh of the following bilinear transformations are periodie
u z+l 2zz - 3
a) - , b) c) ---j
z z z-z
d) w = az + b,
cz+d I; ~ 1 =/:- 0, (a + d? = 3(ad - bc).
Answers.
a) Yes. b) Yes. c) No. d) Yes.
Example 4.12 Find all bilinear transformations with fixed points z = 1 and z =
-1.
Solution. Using the anharmonic relation from Example 4.8 we find that the
desired transformations are
w-l z-l
--=k·-- (k # 0 and k E C).
w+l z+l
Example 4.13 Prove that two bilinear transformations are commutative (with re-
spect to function composition) if they have a common fixed point.
Solution. Let
and
with I:: ~ 1·1:: ~ I# o.

A fixed point for both transformations will be determined from
CIZ 2 + (d - al)z - b1 = 0, C2z2 + (d 2 - a2)z - b2 = o.

The zeroes of these quadratic equations will be equal if and only if
Cl dl - al bl
C2 = d2 - a2 = b2 •
This condition implies
what we had to prove.
Exercise 4.14 Prove that J(z) = l/z maps circles and lines on cireles and lines,
respectively.
Example 4.15 Prove that the bilinear transformation maps cireles and lines onto
circles and lines.
Solution. We have for c =1= 0
f(z) = az + b = ~ (a _ (ad - bC)) ,

cz+d c cz+d
i.e., f = ho f2 °f1, where
f1(Z) = CZ + dj h(z) = -1j h = -a - (ad - bC) z.

z c c
Then use Exercise 4.14.
Exercise 4.16 Let us consider the following sets of bilinear transformations
G={W= u~+v
-vz+u
I U,V,EIC,luI2+lvI2=1},
G2 =
{ W =
()
cos 2 . z - z . sm 2
. ()
.
()
()
I
-zsm 2 + cos 2
a) Prove that (G,o) is a group, and (G1 ,o) and (G2 ,o) are its subgroups, where
°is the operation of the composition of functions.
b) Prove that for every T E G there exist Tl, T3 E Gi and T 2 E G 2 such that
c) Apply b) on
l+z
[ -l+z 1-Z]
1-z .
Answer.
b) We have
[ -v~ ~]
u
= [ e,,,,,j2
0
o ]. [
e,,,,,j2
cos ~
. ()
-zsm2
Z· sin ~
. 7)
Z· sm2
j. [
e''''2/ 2
0
0
e''''2/2
]
,
where
and
Exercise 4.17 Find the images oj the jo/lowing regions by the given junetions:
a) The angle 0 < argz < 0, Figure 4.20, by thejunetion w = z'1f/8.
b) The strip 0 < Imz < 71', Figure 4.22, by the junction w = eZ •
e) The region between two eireles between points Zl and Z2, Figure 4.24, by the
z- Zl
junction w = - - .
Z - Z2
d) The region between two eireles, jrom whieh one is inside oj other and they
have one eommon point zo, Figure 4.26, by the junction w = _1_.
Z - Zo
Answers.
o
Figure 4.20 0 < arg Z < 0
a) The image is the region Im w > 0, Figure 4.21.
Figure 4.21 Im w > 0
b) The image is
y
z
Figure 4.22 0 < Im w < 1f'

the region Im w > 0, Figure 4.23.
Figure 4.23 Im w > 0
c) The image is
o x
Figure 4.24
the angle on Figure 4.25.
v w
Figure 4.25
cl) The image is
Figure 4.26
the strip between two parallel stright lines on Figure 4.27.
Figure 4.27
Example 4.18 Construct a Junction w = J(z) which maps the region Oz : /z/ <
1, Imz > 0 on the region Ow : /w/ < 1.
Solution. Up to now our task was to find the image Ow of a region Oz under
the given mapping f.
Now we have two given regions Oz and Ow, and we have to construct a mapping
from Oz onto Ow.
The Riemann Mapping Theorem ensures the existence of the desired conformal
mapping and with some additional suppositions (that a point from 0% and a line
through it are mapped on a given point and a line in Ow), we have uniqueness.
There is no general algorithm for finding the desired conformal mapping. We
can use for the construction of the desired function the functions from Exercise 4.17.
-1 o 1 x
Figure 4.28 /z/ < 1, Imz > 0
For our case this can go in the following way. First we map with the function
z-l
w=--
z+l
(see Exercise 4.17 c) ) the region Izl < 1, Imz > 0 (Figure 4.28) on the inner part
of angle 7r /2 with the corner at the point (0,0)
V,
w,
o u,
Figure 4.29 RewI < 0, Im WI > 0
in the WI- plane and because of WI(O) = -1, on the region ReWI < 0, ImwI > 0,
Figure 4.29.
Then we map with the function W2 = w; the region Re WI < 0, Im WI > 0 on
the half-plane Im W2 < 0, Figure 4.30.
+Vz
I
Figure 4.30 Im W2 <0
Now using the function

W2 + t
W=--,
W2 - t
we map the half-plane Im W2 < 0 on Iwl < 1. So finally, we obtain the desired
transformation
z - 1)2
w-
( z:tT +Z - Z2 - 2z + z
- (Z - 1)2 _ Z - Z2 - 2zz + 1 .
z:tT
-1
Figure 4.31 Iwl < 1
Example 4.19 Construct a Junction w = J(z) which maps conJormally the region
0% : Izl < 1, Imz > 0 Rez > 0 on the region Ow : Iwl < 1.
Solution. We map with the function WI = Z2
Figure 4.32 Izl < 1, Imz > 0" Rez > 0

the quarter of the unit disc Izl < 1, Im z > 0" Re z > 0, Figure 4.32, on the
half-disc IWll < 1, Im Wl > 0, Figure 4.33.
U1
Figure 4.33 IWll < 1, Im Wl > °

Using now the function from Example 4.18 we obtain the function
Z(Z2)2 - 2z 2 + Z ZZ4 - 2z 2 + Z
W = J(z) = (Z2)2 _ 2zz2 + 1 = Z4 - 2zz2 + 1·
Example 4.20 Construct afunction W = J(z) which maps the angle 'Ir/6 < argz <
/3 on the unit disco
'Ir
Solution. Firstly we rotate the given angle (Figure 4.34)
o x~
Figure 4.34 'Ir /6 < arg z < 'Ir /3
for -'Ir /6 by the function Wl = e- l1r / 6 . z, Figure 4.35.
o u,
Figure 4.35
Next we map with the function W2 = w~
o Uz
Figure 4.36 Im W2 > 0
the angle obtained on the half-plane Im W2 > 0, Figure 4.36.

Finally, we map by the function w = W2 - l (see Example 4.5) the region Im W2 >
W2 +l
o onto Iwl < 1, Figure 4.37.
~~~~*M~-----~
-1 1 u
Figure 4.37 Iwl < 1
So we obtain
w = f(z) =
Exercise 4.21 Construct a function w = f(z) which conformally maps

a) the region Izl < 1, Iz - 1/21 > 1/2 onto the half-plane Im w > 0;
b) the region Izl < 1,0< argz < rr/n (n E N) on the half-plane Imw > 0;
c) the region Izl < 1,0< argz < rr/3 on the unit disc Iwl < 1.
Answers.
a) w = e-'7l"(Z+l)/(Z-l);
Example 4.22 Prove that the function w = cos z maps the lines Re z = const on
hyperbolas in the plane w, and the lines Imz = const on ellipses in the w-plane.
Solution. We have by definition

e'(X+'Y) + e-'(x+,y)
w=u+w= ,
2
for z = x + zy. This implies

eV + e- Y e- Y - e V •
u= cosx and v = 2 smx.
2
Then the lines x = const are mapped in w-plane onto the hyperbolas
u2 v2
----.--=1,
cos 2 x sin 2 x
and the straight lines y = const onto ellipses
u2 v2
(e ile-vr + (e- :2 eyr = 1.
V Y
Exercise 4.23 Find the sur/ace area 0/ the region onto which the /unction w = e Z
maps the rectangle
K = {z I a - b ~ Re z ~ a + b, - b ~ Im z ~ b},
where a is real and 0< b< 'Ir.
Find the limit 0/ the /raction 0/ the sur/ace area 0/ the obtained region and sur/ace
area 0/ rectangle as b _ O.
Answer: The desired region is bounded by straight lines arg w = band arg w =
-b and by circles w = eaH , w = ea - b• The limit fraction of surface areas is e2a as
b_ O.
Exercise 4.24 Find cireles which are invariant under the mapping z I-t w given by
1 1
---= --+a,
W-Zo Z-Zo
where Zo and a are arbitrary but fixed points on C.
Answer. The desired circles crosses the point Zo where they have a tangent
given by 0.
Example 4.25 Find the curves on which the /unction w = (z+ 1/z)/2 maps circles
Izl = r, r < 1.
Solution. Since we have for z = re,a that
l+r l-r
q = Tcosa-zTsina,
we obtain that the circles Izl = r, r > 1, are mapped on the ellipses with axes
(l/r + r)/2 and (l/r - r)/2 and same focuses 1 and -1.
Exercise 4.26 Prove that the function w = (e-· a z)21r/(b-a) for 0 :::; a < b < 271',
maps the angle a < arg z < b onto the whole w-plane without the non-negative part
of the x-axis.
Hint. We have
2
argw = -b-a
-(argz - a), 0< argw < 271'.
Exercise 4.27 Find the function which maps the annulus 0< rl < Izl < r2 on the
region between two ellipses Iz - 21 + Iw + 21 = 4b1 and Iw - 21 + Iw + 21 = 4b 2 for
1<b2 <b1 •
Hint. Use Example 4.25. The desired function is w = cz + ~, for

cz
Chapter 5
The Integral
5.1 Basics
5.1.1 Preliminaries
For a continuous complex-valued function J(t) = u(t) + w(t), a S; t S; b, we take
l b
J(t) dt = t u(t) dt +z t v(t) dt.
Definition 5.1 The (line) integral oJ a continuous function J at all points of a path
P given by z(t), is given by
h, fez) dz = t f(z(t))z'(t) dt.
Two curves PI : z(t), a S; t S; b, and P2 : w(t), c S; t S; d, are smoothly equivalent if

there exists an 1-1 and Cl mapping s: [c,d] ~ [a,b] such that s(c) = a,s(d) =
b, S'(t) ~ 0 for t E [c, d], and w = z 0 s. We have
( fez) dz = ( fez) dz
}p, }P2
for smoothly equvalent paths PI and P2 . We have for a path P : z(t), a S; b, contin-
uous function J on P
{pJ(z)dz = - h,J(z)dz,
where -P is defined by z(b+ a - t),a S; t S; b. We have for a path P: z(t),a S; b,
continuous functions J and 9 on P
k(f(z) + g(z)) dz = k J(z) dz + kg(z) dz.
103

104 CHAPTER 5. THE INTEGRAL
We have for a path P: z(t),a S; b, continuous functions f on P and A E IC
We have for a path P : z(t), a S; b, continuous functions f on P
Theorem 5.2 If {fn} is a sequence of continuous functions which uniformly con-

verges to f on a path P, then
Ern r
n-+oo}p fn(z)dz= r
}p f(z)dz.
Theorem 5.3 If f is the derivative of an analytic function F (analytic on a path

P) then
i f(z) dz = F(z(b)) - F(z(a)).
Theorem 5.4 (Cauchy's Theorem) Let f be analytic in a simply connected re-

gion 0 and P is a closed path contained in O. Then
i f(z) dz = O.
Theorem 5.5 (Generalized Cauchy Theorem) Let f be analytic inside of a re-

gion 0 bounded by a closed path P. If f is continuous on C and in 0, then
if(z)dz=O.

Example 5.1 Find the following integrals:
a) 1 1
e'!· cosatdt, a E lW.;
b)
j 1
_1t 2 +Z'
dt
1 1
e'! . cos at dt = 1
1
cos t . cos at dt +z 1
1
sin t . cos at dt.
5.1. BASICS 105
Since
cost· cosat = (cos(t + at) + cos(t - at))/2,
sin t . sin at = (sin( t + at) + sin( t - at)) /2,
1
we obtain
1
e't . cos at dt
= ~ (sin(a + l)t + sin(l - a)t) 11 _ z~ (cos(a + l)t + cos(l - a)t) 1

1
2 a+1 1-a 0 2 a+1 1-a 0
= ~ (sin(a + l)t
2 a+1
+ sin(l -
1- a
a)) _ ~ (cos(a + 1)
2 a+1
+ cos(l -
1- a
a) _ _ 2_), a
1 - a2
E IR.
b) We take
1 A B
t 2
+l t- 4 + z4 + + 4 - z4' t
Then we obtain
A=
../2 + z../2 and B
-../2 - z../2
4 = 4 .
Therefore
1 t dt+
-1
1
2 Z
= ../2 + z../2 log

4
(t _../22 + z../2)
2
1
1
-1
_../2 + z../2 log
4
(t + ../22 _ V2)
2
1
1
-1
.
Example 5.2 For which reals a and b do there exist the following integrals
a) 11
-1
cos it dt.
t a '
c) 1 --dt?
ta
+t
o z
1
Solution. The integral f~l c~sa it dt is improper because of the point O. It con-
verges absolutely far a < 1, what follows by
1 -rl- 1 1COS
-1 t
it I d t<
-
1 -Itrll-t
1
-1
cosh d t<2 1 - rl '
1
-1
dt
t
where we have used Icos ztl ::::; cosh t < 2 far -1 ::::; t ::::; 1. The last integral converges
for a < 1.
We remark that in a quite analogous way we can consider also the general case
when a is a complex number using that WI = tRe a. The integral converges in this
case for Re a < 1.
b) The integral 11
00 ta
--b
z+t
dt converges absolutely for a - b < -1, what follows
from
[00 I~I dt < [00 ta-bdt, or a < -1,
1t z+t J1
smce
I ~I
Z+t ,. , t a as t --t 00.
For complex numbers a and b using the fact that Ita-bl = tRe(a-b) we obtain the
convergence of the integral for Re (a - b) < -1 or Re a < -1.
c) The integral
1 ta
1
- - b dt converges absolutely for a > -1, since
o z +t
[11~1 dt <
Jo z+ t
[1 t a dt or a - b > -1.
Jo
Investigate the convergence of this integral for a, b E !C.
Example 5.3 Find for which complex numbers z the integral 1e1
dt exists, and
give abound for the modulus of this integral.
Solution. Since we have
the considered integral converges for Re z > -1. We have for the modulus of the
integral
1= 1 ,
1Jo[1 t z dtl:::; [1 WI dt = Jo[1
Jo
tRez dt = tRez+I
1 + Re z
101 + Re z
for Rez > -1.
[b [b_
Ja f(t) dt = Ja f(t) dt,
for any integrable function f : [a, b] --t !C, what will imply that a complex function
of real variable 7 is integrable on the interval [a, b] if the function f is integrable on
[a,b].
5.1. BASICS 107
Solution. Taking J(t) = u(t) + zv(t), we obtain

t J(t) dt t u(t) dt + z t v(t) dt
= l bu(t) dt - l bv(t) dt
z
= l bCuCt)-zvCt))dt
= l bJCt) dt.
Example 5.5 Find the values 0/ the /ollowing integrals
(i) fL Re zdz; (ii) fL z 2 dz; (iii) fL dZ, Jrom the point Zl = 1 to the point Z2 = i
z
in the positive direction on the /ollowing curves L :
a} the boundary 0/ the square: 0 :::; x :::; 1, 0 :::; y :::; 1 (Figure 5.1);
-
t
-o~--~~--------- .~-.
Figure 5.1
b} the part 0/ the circle: z = e't, 0 :::; t :::; i (Figure 5.2);
~----~-------+
o X
Figure 5.2
c) the straight line segment z = (1 - t) + zt, 0:::;: t :::;: 1, Figure 5.3.

+y
o x
Figure 5.3
Solution. (i) a) We have
1Rezdz = ll.zdt+ 1° Re (z+t)dt=z-ltdt=-1/2+z;
(i) b) We obtain
I Re z dz = Z 1,,/2 Re (e,t)e,t dt
= ZJo
r/ cost(cost+zsint)dt
2
r/2 ["/2
Z Jo cos 2 t dt - Jo cos t sin t dt
Z7I" 1
(i), c) We have
lRezdz = lRe[(l-t)+zt](z-l)dt
= (-1 +z) 1(1-t)dt

1
(-1+z)/2;
(ii), a) We obtain
(ii), b) We have
[ ["/2 2t t r/2
JL z 2dz = ZJo e 'e ' dt = ZJo (cos 3t + Zsin 3t) dt = -(1 + z)/3.
5.1. BASICS 109
(ii), c) We obtain
kZ2dz = k((1-t)+zt)2(-1+Z)
(-1 + z) (l dt - 211 t dt + 2z j1 Ot dt - 2z l 2
t dt)
= -(1+z)/3.
We remark that in (i) we have obtained different values of the integral on different
curves in the contrast to the case (ii) where we have always obtained the same value.
Why? The case (iii) can be considered in an analogous way as the previous two ones.
Exercise 5.6 Prove that k zndz = 0, for every n ~ 0 and any closed path L.
Example 5.7 Find the integral k Izlz dz on the path L which consists of the half-
circle z = Re t', 0 ~ t ~ 7r, and the straight line segment: -R ~ Rez ~ R, Imz = 0,
Figure 5.4.
-R R
Figure 5.4
Solution. We have
kIzlz dz
Example 5.8 Find the integral k z dz on the closed path L wh ich consists 01 the
semi-circle L 1 : z = e,t, -7r /2 ~ t ~ 7r /2 and the straight Une segment L 2 : z =
zt, -1 ~ t ~ 1, Figure 5.5.
L,
-i
Figure 5.5
Solution. We have
Ir L
Z dz = 11'/2
-,,/2
e-Itze,t dt + 1-1
1
(-zt)z dt = 7rZ.
Example 5.9 Let a lunction 1 be continuous in a neighborhood 01 z = O. Prove
1
that
21'
a) lim
T-+O 0
I(re't) dt = 27r 1(0); b) lim f I(z) dz = 27rZ/(0),
T-+olL z
where L is the circle Izl = r.

Solution. Since the function 1 is continuous in a neighborhood of z = 0 we
have that for any c: > 0 there exists a 8 > 0 such that
I/(r. ed ) - 1(0)1 < 2:' for r < 8(c:), 0 ~ t ~ 27r.
a) Therefore
5.1. BASICS 111
for r < 6(e), 0::; t::; 211".

b) Analogously as in a), putting z = re t ', 0 ::; t ::; 211", we obtain
11 /~z) dz - 211"z/(0)1 = 11 2
,. /~.·e~~') r· ehz dt -1 2
" /(0) dtl
= I, 10 2
,. (J(r . et ') - /(O))dtl
< 10 2
,. I/(r . et') - /(0)1 dt
< -e 10 2 ,.
dt
211" 0
= e,
for r < c5(e), 0 ~ t ~ 211".
Remark. We have a more general result if we assume continuity of the function
/ in a neighborhood of a point z = w. In this case we have
lim r /(z) dz = 211"/(w),

r ..... O}L Z - W
for L: Iz - wl = r.
Example 5.10 Let / be a continuous /unction in the region D,
D = {z 10 < Iz - wl < R, 0 ~ arg (z - w) ~ 0 (0 < 0 ~ 211")},
and there exists lim(z - w)/(z) = k. Prove that then liml /(z)dz = zOk, where L
z-+w r-+O L
is an arc 0/ the circle Iz - wl = r, which lies in the region D, positively oriented,
Figure 5.6.
x
Figure 5.6
Solution. The condition zlim(z

.... w
- w)J(z) = k implies that for every c > 0 there
exists b > 0 such that
c
(j> I(z - w)J(z) - kl = IrehJ(w + re t.) - kl
for r < b(c), 0::::; t ::::; 271". Therefore we have
liJ(Z)dZ-dJkl = 11°J(w+re t·).et·zdt- z1°kdtl

1°(J(w + reti)ret• - k) dtl
Iz
< 1°IJ(w + r· kl dt
eti)re tl -
< ~ fO dt
() Jo
c,
for r < b(c), 0::::; t ::::; ().
Example 5.11 Let J be a continuous Junction in the region
IJthere exists zlim

....
zJ(z) = k Jor z in the region D, then
OO
lim
R'-+ OO JfL J(z)dz = i()k,
where L is the part oJ the circle Izl = R' which lies in the region D.
Solution. The condition lim zJ(z) = k means that for every c > 0 there exists
'-00
zED
R o > 0 such that
~ > IzJ(z) - kl = IR'rt'J(R'r to ) - kl,
for R' > Ro, 0::::; t ::::; (). Therefore
1i J(z) dz - Z()kl 11 J(R' etl)R' et·z dt - Z()kl

9
< 1°IJ(R'et')R'et• - kl dt
c f9
< (j Jo dt
c,
for R' > Ra, 0 ::::; t ::::; ().
5.1. BASICS 113
Example 5.12 Let J be a continuous function in the region Oz containing a path

L connecting the points z(a) and z(b). If a function 9 is a bijection 0/ a path C /rom
a region Ow (for z = g( w) on a path Land 9 and g' are continuous on C) prove
that then
1 J(z) dz = fc J(g(w))g'(w) dw,
where w(a) = g-l(z(a)) and w(b) = g-l(z(b)); g-l is the inverse /unction 0/ the
function g.
Solution. The proof follows by the definition of the integral of a complex

function.
Let (Zk), k = 0,1, ... , n, be a partition of the path L. The function w = g-l(Z)
maps it on partition (Wk), k = 0,1, ... n, of the path C. Then we have
n-1 n-1
L f(Zk)(Zk+1 - Zk) = L J(g(Wk))(g(Wk+1) - g(Wk))
k=O k=O
We shall show that the last sum tends to zero as n -+ 00 for finer partition.
Then the preceding equality will imply the desired conclusion.
° °
Since g' is continuous we have that in every closed region which lies in Ow by
uniform continuity for every c > there exists 8 > such that
Ig(Wk+d - g(Wk)
Wk+1 - Wk
- 9
'()I
Wk < C,
°
whenever IWk+1 - wkl < 8(c). This is possible always to achieve for any small 8 > 0,
since max IWk+1 - wkl-+ as n -+ 00, together with IZk+1 - zkl,since
!Zk+1 - Zk! = 19(Wk+1) - g(Wk)I!Wk+1 - Wk!, and exists g'(w).

wk+1 - Wk
Let M = max If(z)l. Then
IE J(9(Wk))(g(w~:~~ =~~Wk)
zEL
- 9'(Wk)) (Wk+1 - Wk)1

n-1
<€ L !f(9(Wk))!!Wk+1 - Wk! :::; €. M . C',

k=O
where C' is the length of the path C.
Example 5.13 Let L be a path and L the path which is the image oj L by the
junction z f-4 z. Let j be a continuous junction on L. Prove:
a) The junction z f-4 j(z) is continuous on the path Land
[j(z) dz = k j(z) dz.
b) Ij L is a circle z = e lt , 0 :::; t :::; 271", positively oriented, then
iL
j(z) dz = - j
L
-dz
j(z) 2"
Z
Solution. The function z f-4 j(z) is continuous on L, since for z E L we have

z E L, and by the supposition j(z) is continuous on L we have that j(z) is also
continuous as composition of continuous functions. We have used that the map
z f-4 Z is a bijection and continuous function. This follows by the fact that for every
c > 0 there exists 8 > 0 such that
Iz - zol = Iz - zol = Iz - zol < c

for Iz - zol < 8 = c.
Now we shall prove the desired equality. We remark that by the preceding result
follows the existence of the integral from the right part of the desired equality. We
can put w = z in the integral in the right part of the desired equality
k j(z)dz = [j(w)dw = [j(w)dw.
Putting w(t) = z(t) = x(t) + ly(t) and j(z(t)) = u(t) + zv(t), we obtain
[j(z)dz = l b
j(z(t))z'(t) dt
= l (u(t) + iv(t))(x'(t) + iy'(t)) dt
= l (u(t) - zv(t))(x'(t) - ly'(t)) dt
= l b
(u(t)x'(t) - v(t)y'(t)) dt + l l(v(t)x'(t) + u(t)y'(t)) dt
= l(u(t) +zv(t))(x'(t) +ly'(t))dt
= [j(z)dz.
5.1. BASICS 115
b) By the proof of a) we have
1 J(z) dz = 1 21r
J(et')ze" dt
= - Jf21r-
o J(et')ze-hdt
= - 1
o
21r_-zeh
J(e h )2t- dt
e •
-dz
= - /,L J(Z)-;2.
Example 5.14 Let
and let J be a continuous Junction on Izl ::; r. Prove that
JLf J(z)dz = n-+oo

lim f J(z)dz.
JL n
Solution. We have
= 1 J(z)dz,
where we have used the continuity of the integration.
Exercise 5.15 Find the integral
on the Jollowing paths:

a) semi-circle Izl = 1, Y 2': 0, VI = 1;
b) semi-circle Izl = 1, Y ::; 0, VI = -1;
c) semi-circle Izi = 1, Y ::; 0, VI = 1.
Answers.
a) -2(1 - t); 2) 2(1 - t); c) -2(1 + t).
Exercise 5.16 Prove that Jor lai =j; R we have
Hint. Put in the integral z = Re'8 (0 ~ () ~ 27r) and use the inequalities for the
modulus.
Exercise 5.17 Prove:

IJ J is a continuous function in the region {zllzl ~ Ra, Imz ~ a} (a is a fixed
real number) and in this region it holds J(z) --+ 0 as z --+ 00, then Jor every positive
1
integer m
lim e,mz J(z)dz = 0,
R_oo eR
where C R is the part oJ the circle Izl = R which belongs to the considered region
(Figure 5.7).
Im 2 - Cl
Figure 5.7
5.1. BASICS 117
Hint. To find the bound for modulus of the integral on the semi-circle Izl =
R, Im z > 0 use the inequality
• () 2() 'Ir
sm >-
- 'Ir
for 0<
- () -< -,
2
and the paths of the circles L l and L 2 which lies in lower the half-plane, see Figure
5.7 . In the case a < 0 use that the lengths of each of them tends to lai as R ~ 00).
Example 5.18 Using the connection between integral on a closed path and the dou-
ble integral in ]i2 prave
Let J be an analytical Junction in the simple connected region 0 C C with a
continuous derivative in this region. Then Jor every closed path L C 0 we have
L J(z)dz = o.
Solution. The proof follows by Green's formula which relates the line integral
on a closed path Land the double integral on the region D which is bounded by
the path L. Putting J(z) = u(x,y) + w(x,y), we obtain
L J(z) dz = L (u(x, y) dx - v(x, y) dy) +z L

(v (x, y) dx + u(x, y) dy),
where dz = dx + zdy.
Since f' is continuous, we have that also the partial derivatives of the functions
u(x,y) and v(x,y) are continuous in D. Applying Green's formula we obtain
f J(z) dz = _]. f (8v + 8U) dx dy + z]· f (8u _ 8v) dxdy.

JL JD 8x 8y JD 8x 8y
Therefore, by the Cauchy-Riemann equations
8u 8v 8u 8v
and -=--,
8x = 8y 8y 8x
L
we obtain
J(z)dz = O.
Remark. The proved statement is known the name of the Cauchy theorem. The
condition of the continuity of the first derivative f' is not necessary (Goursattheo-
rem). Pollard proved that even the differentiability on the path L is not necessary
(it is enough to assume only continuity).
Figure 5.8
[00 sin x dx,

Jo x
taking for the path for the integration that from the Figure 5.8.
Hint. The given path can be replaced with the following path:
- R ~ z ~ -rj Z = reD., 0 ~ () ~ 71'j r ~ Z ~ Rj and z = Re 8i , 0 ~ () ~ 71',
Figure 5.9.
Figure 5.9
5.1. BASICS 119
Then, we obtam
a
.1
sinx
--dx
x
00
= 'Ir /2.
Find the given integral just on the given path, directly.
Example 5.20 Let J be an analytical Junction on the strip 0 ~ Im z ~ k, and
i:
lim J(x + zy) = 0 on this strip. Prove that iJthe integral [00 J(x)dx exists, then
x-+±oo 1-00
also the integral J(x + zk)dx exists, and these two integrals coincides.
Solution. By the analycity of the function J we have fL J(z) dz = 0, where L

is the path from Figure 5.10.
~----.....(I------
ki
. . . ---
I
t
I
t t
--~------~----~---~
-R 0 R X
Figure 5.10
We have
[ J(z)dz = jR J(x)dx+z [k J(R+zy)dy+

JL -R Ja
r R J(x+zk)dx+z Jk[a J(-R+zy)dy.
JR
Letting R - 00, the second and fourth integrals are zero (by the condition
lim J(x + zy) = 0).
x--+±oo
Therefore
Since we have
1
00
J(x)dx= 1 00 k
J(x+zk)dx+z1 J(zy)dy,
and the function J is regular on the given strip the last integral is finite. Therefore
i:
by the existence of the integral
J(x) dx, the integral 1

00
J(x) dx,
it follows the existence of the integral
10'' ' J(x+zk)dx.
It can be proved analogously the existence of the integral 1 J(x + zk) dx.
0
00
1° e-
00
x
2 Vi
cos 2bx dx = - e - ,
2
b2
iJ it is known that
[00 2
i-oo e- t dt = Vi·
Hint. Take the integral of the function J(z) =e z2 on the rectangle:
IRezl ::; R, 0::; Imz ::; b (Figure 5.11).
bi
,I
-R --- 0/
Figure 5.11
R -.
X
Solution. We have
5.1. BASICS 121
Since
'10fb e-(R+.y) 2 fb
dy = z 10 e(R
2
-y
2)
e- 2R • y dy
we obtain
Iz !ob e-(R+.y)2 dyl < e- R2 !ob ey2 dy < 15M,
as R -+ +00, for arbitrary small 15 > 0, since
/l e
y2
dY/ ::::; M, for finite b.
We can prove analogously that
Iz 1° e-(-R+.y)2 dyl-+ 0 as R -+ +00.
1: 1:
Therefore we obtain for R -+ 00
VP e- x2 dx - VP dx = 0,
1:
e-(xH)2
VP e-(x+.bj2 dx = ,fii,
where we have used

1 00
-00
2
e- t dt = ,fii.
Further we have
eb2 V P 1 e-
00
-00 x 2 (cos 2bx - z sin 2bx) dx = ,fii,
1 e-
which implies
00 2 2
VP -00 x cos 2bx dx = ,fiie- b •
Since
we obtain
1: e- x2 cos 2bx dx = Vii e- b2 .
Since the function e- x2 cos 2bx is even we have
Second method. Using the representation

and interchanging the order of sum and integration (why is it possible?) we obtain
Since
t'"
10 xe- X
2
X 2n - 1 dx
1
-e _x x 2n_ 1 1
2
2
00
+--
2n -
2
11 00
e _x x 2n-2 d x
2
1
0 0
2n --1 e _x2 x 2n-2 dx

00
-
20'
2n -1
Jn = - 2 - Jn - 1 for n = 1,2, ... ,
and so
J = (2n -l)!!J = (2n -I)!! . V7r
n 2n 0 2n 2 .
Finally, we obtain
[00 2
10 e- X cos 2bx dx
Third method. The desired real integral we can find also using only the real
analysis, differentiating the following integral with respect to the parameter z.
J(z) = 1
-00
00 2
e- X coszx dz for z E IR.
5.1. BASICS 123
It is easy to check that the conditions for such differentiation are satisfied. Therefore
we have
j'(z) = i:(-xe- X2 )sinzxdx
= 1 e-
2"( X 2.
SIn zx) 1-00
00
- 2"z 1 00
-00 eX 2 cos zx dx
z
= -2"f(z).
The general solution of the obtained differential equation j'(z) = -~f(z) is
1:
where C is an arbitrary real constant.
Putting z = 0 we have C = f(O) = e- x2 dx = ..;:;r, and so we obtain
Putting z 2b and using that the function under integral is even we finally
obtain
1 o
00 2
e- X cos 2bx dx =
y'7i
_ eb
2
2
•
Example 5.22 Taking the function ez2 , and choosing a convenient path and know-
ing that
1-00
00 2
e- t dt = ..;:;r,
p1'Ove that
[00 cos x 2 dx = [00 sin x 2 dx =~

10 10 2v2
(Fresnel integrals).
Solution. We take the following path

z = x, 0 ~ x ~ R;
(see Figure 5.12).
~~~------~---~
o -- R X
Figure 5.12
By the analicity of the function e- z2 in the region bounded by the path L we

have
Therefore we obtain
f
J{R
o e- X
2
dx + Jo
{
e- R
2 2<,
e zR eh dt + J(o
R
e-· t
2
e'7r/4 dt = O.
We have for the second integral
["/4
Jo
22·
~ R le- R e t'ldt
["/4 R2
~ R Jo e- cos2t dt
~ R Jo["/4 e k(1-4t/7r) dt,

5.1. BASICS 125
since cos2x ~ 1 - 4x/7r for 0 ~ x ~ 7r/4 (see Figure 5.13).
Figure 5.13
Therefore
I11r/4 e- R2e2" zRet. dt I < Re- R2 1 1r

/
4
e4R2t/1r dt
< Re- R2 ~(eR2 - 1)

4R2
< ~(1 _ e- R2 )
4R
- 0,
as R - 00. Letting R _ 00 we get
e- 1r /4 1 00
e- x2 dx - 1
00
e-· t2 dt = o.
Separating on real and imaginary parts and taking into account that
1 o
00
e -x2d x =.Ji
-
2 '
we finally get
1 o
00 cos e dt = 1 sin t0
00
2 dt = v;".
2y2
Example 5.23 Knowing that

1, dz = 2n
L Z
on the circle L : z( t) = r . et., 0 ~ t ~ 27r, find a path which will enable us to prove
1
that
211" _ _ _d_t_ _-,,- _ 27r
o a2 cos 2 t + b2 sin 2 t - ab·

Solution. We shall choose for the path the following ellipse
C:x=acost, y=bsint, 0~t~2'11".
Using theorem on the equality of integrals on these paths we have
f dz = f dz = 2'11"z,
Je z iL z
f21r -a·sint+zbcost f21r sint cost(b2 -a 2 ) f21r abdt
Jo a cos t + z b sin t dt = Jo a2 cos 2 t + b2 sin 2 t dt + z Jo a2 cos 2 t + b2 sin 2 t
2'1TZ.
Taking equal separately the real and imaginary parts, respectively, we find that the
imaginary part of the integral on the left side is equal to 2'11". Hence
Example 5.24 Prove the basic theorem oj algebra using the Cauchy theorem.
Solution. Take an arbitrary but fixed polynomial P(z) = Lk=O ak zk. Suppose
the contrary, that P(z) f: 0 for every z E c. We introduce the following polynomial
Q(z) = Lk=Oak zk. Then also Q(z) f: 0 for every z E C (in the opposite it would be
P(z) = 0). We apply Cauchy's theorem on the following regular function
1
j(z) = P(z)Q(z) '
on the path L given on the Figure 5.14.
Figure 5.14
5.1. BASICS 127
Therefore
r dz
JL
P(z)Q(z) = 0,
r dz jR dx r Reet dO
JL P(z)Q(z) = -R [P(x)]2 + Z Jo P(ReO·)Q(Re Ot )·
The second integral tends to zero as R -+ 00 (prove !). Hence
j oo
-00
dx
(P(x))2 = 0,
which is impossible, since the function under integral is positive for all 00 < x < 00.
Exercise 5.25 Let log z be the principal value oJ the complex logarithm and L
ett,O :::; t :::; 27r. Find h log zdz Jor starting points: a) z = 1, b) z = Z, c) z = -1.
Chapter 6
The Analytic functions
6.1 The Power Series Representation
6.1.1 Preliminaries
Theorem 6.1 Let J be an analytic Junction in the disc D(zo, r). Then Jor every
a E D{zQ,r)
(i) there exist analytic Junctions in D Fand G such that
F'(z) = J(z) and G'(z) = J{z) - J{a) ;

z-a
(ii) Jor every closed path P contained in D(zQ, r)
f J(z) dz = f J(z) - J(a) dz = O.

}p }p z- a
Theorem 6.2 Let J be an analytic Junction in D(zo, r), 0 < r < R, and la- zol < r.
1
Then
J(a) = _1 J(z) dz,
211"~ er z - a
where er
is the circle ZQ + re'B, 0 :::; () :::; 211".
Theorem 6.3 (Cauchy's Integral Formula) Let J be analytic in the disc

D(zQ, r), 0 < p< r, and la - zol < p. Then
J(a) = _1 f J(z) dz.

211"z }av(zo,p) z- a
129

130 CHAPTER 6. THE ANALYTIC FUNCTIONS
Theorem 6.4 (Taylor Expansion) If f is an analytic function in D(zo, r), then

there exist constants Ck, k E EN u {O}, such that it represented by power series
L Ck(Z -
00
f(z) = zo)k
k=O
for alt z E D(zo,r). Moreover, f is infinitely differentiable at Zo and
Ck =
f(k)(zo)
k' = -2
1 1 f(z)
(_ )k+l dz,k E ENu {O}.
• ll'Z er Z Zo
Ifmaxlzl=r If(z)1 = M(r), then the coefficients Ck satisfy Cauchy's inequality
M(r)
lekl ~ -n-'
r
r < R.
Theorem 6.5 (Uniqueness Theorem) If two functions fand g, analytic in a
region D are equal on a subset of D with an accumulation point in D, then fand g
are equal on the whole D.
Theorem 6.6 (Liouville's Theorem) A bounded entire function is constant.
Theorem 6.7 If f is an entire lunction and I( z) -+ 00 as z -+ 00, then I is a

polynomial.
Theorem 6.8 (Mean Value) Let I be an analytic lunction in a region D and

Zo E D. Then
lor D(zo,r) C D.
Theorem 6.9 (Maximum Modulus) III is a non-constant analytic lunction in

a region D, then for every z E D and every c > 0 there exists z' E D(z, c) U D such
that
I/(z')I> I/(z)l·
Theorem 6.10 (Open mapping theorem) The image of an open set under a
nonconstant analytic mapping is an open set.
Theorem 6.11 (Schwarz Lemma) Prove that if I is an analytic function on

Izl < 1, I/(z)1 ~ 1 and 1(0) = 0, then If(z) I < Izl for Izl < 1 and 11'(0)1 ~ 1,
excluding the case f(z) = e,8 z (0 real).
Theorem 6.12 (Morera's Theorem) Let f be continuous on an open set O. il
kl(z)dz=O
whenever R is the boundary of a closed rectangle in 0, then f is analytic on O.
6.1. THE POWER SERIES REPRESENTATION 131

Example 6.1 If exists
lim I
W n+ll,
n--+cx:> Wn
then this limit is the reciprocical value of the radius of convergence of the power
series a L:~=o wn(z - Wt.
Solution. The power series L:~own(z - wt. absolutely converges by D'Alembert

criterion when
n~oo
(
lim I Wn+l z-w
)
wn(z-w)n
n+l
I< 1,
Jz - wJ lim wn+1 1 < l.
n----+(X)
I Wn
Therefore we have for the radius of convergence
1
Jz-wJ< rIffin -+ oo
JW n +1 J =R.
Wn
If we suppose that for sorne z such that Jz - wJ > R the power series converges we
obtain a contradiction with the previously proved part.
Exercise 6.2 Find the radius of convergence for the following power series
(z wt (z w)n
a) L
00
-a ,aE~; b) L 00
-2 ,aE~;
n=l n n=2 In n
c) L(3 + (-ltt(z - wt;

00
d)
n' - wt·
L -;;-(z
00
n=O n
n=l
Answers. a) The radius of convergence is 1 for all a E ~.
b) The radius of convergence is 1 for all a E ~.
c) The radius of convergence is 1/4.

d) The radius of convergence is e.
Example 6.3 Find the radius of convergence for the following power series
L00 ( w) n
n=l n z
with respect to the complex parameter w.

Solution. By D'Alembert's criterion we have R = 1. On the boundary circle Izl = 1

of convergence the power series converges absolutely for Re w > 0, and ordinarily for
-1 < Re w ::; O. The result about ordinary convergence follows from the following
facts. For z = emS , -7[' < () < 7[', we can write
t Ukl = letS I:( _l)k e.ksl = 11 -

Ik=1 (_l)n-l e,(n-l l SI < _1_
k=O eSt /2 2 cos ~ - I cos ~I
I
the sum EI:=1 ukl is bounded for -7[' < () < 7['. Since
converges by Raabe's criterion we obtain the ordinary convergence for -1 < Re w ::;
o of the series E~1 (:)zn.
Example 6.4 Compare the radiuses 01 convergence lor the power series
00 00
Lun(z-wt and LnB.un(z-wt,

n=1 n=1
where a is real.
Solution. Let R be the radius of convergence of the power series E~=1 un(z - w)n.
Then En~l un(z - w)n
independently of a. We have used that
limsup(
n-+oo
vnt = 1 B = 1.
Exercise 6.5 Find the radius es 01 convergence 01 the lollowing power series
1 1 oon-1
= L(-; + _)zn =L
00
A and B _ _ zn.
n=l 2 n n=1 n
Then find the radii 01 convergence lor A + Band A - B.
Answer. We have R(A) = R(B) = R(A + B) = R(A - B) = 1.

. 1
Example 6.6 Takzng r;:,o zn = -1- prove that
-z
~n(n -1)· ··(n - k+ l)zn-k = (l_k~)k+l' for Izl < 1.
Solution. The proof goes by induction. For k = 0 we obtain r;~=ozn = 1/(1- z).
Supposing that the desired equality is true for k and differentiating it we obtain
'"' n(n - 1)··. (n - k + 1)(n _ (+1) + 1) . zn-(k+l) = k!(k + 1)

L...J (1 _ z)(k+l)+l '
n;?:k+l
the desired equality for k + 1.
Example 6.7 Prove that the radiuses of the following power se ries
00 00
A = ~ un(z - w)"j B = ~ vn(z - w)"j

n=O n=O
00 00
c= ~(Un)P(Z - w)"j D = ~ unvn(z - W)",

n=O n=O
satisfy the equalities

a) R(G) = RP(A)j b) R(D) = R(A) . R(B).
RtC) = lie.s~p v'lu~1 = li~-!~p ( v'lunlY = (lie.s!P v'lunlY = RP~A)'

since lim SUPn_oo v'lu I =
n RlA)"
b) We have
1
R(D)
= limsup
n_oo
v'lunvnl
< limsup
n-+oo
v'lunl ·lirrisup
n-+O
v'lvnl
1 1
=
R(A) . R(B)'
Example 6.8 Let

Sn = LUk.
k=O
Prove that if the radius of convergence R(A) of the power se ries A = ~k=O ukzk, is
one then the radius of convergence R(C) of the power series C = ~k=O Skzk is also
one.
Solution. Since
L zk . L Uk zk = L L Ui zk = L Sk zk = C,
00 00 001;; 00
k=O k=O k=O i=O k=O

we have R(C) 2:: 1, since both series which are in the product have radius one.
It can not be R(C) > 1. Namely, supposing ISnl = ~k=O IUkl (prove without this
supposition!), we have
n+p n+p
t> L ISkllzl k 2:: L (n +p - k + 1)lukllzlk for R > Izl > 1
k=n+l k=n+l
and for all p E 11:1 and n 2:: no(c). Hence
n+p n+p
L (n +p - k)lukllzl k ::; L (n +p - k + 1)lukllzlk < c (6.1)
k=n+l k=n+l
for R> Izl > 1. Therefore for R > Izl > 1
n~ n~ n~
t> L (n +p - k + 1)l u kll z lk = L (n +p - k)lukllzl k + L IUkllzl k

k+n+l k=n+l k=n+l
for p E 11:1 and n 2:: no(c). Then by (6.1) we obtain
n+p
L IUkllzl k < c
k=n+l
for R > Izl > 1 and all p E 11:1 and n 2:: no, what is impossible, since R(A) = 1.
Contradiction. Therefore R( C) = 1.
Exercise 6.9 Let the radius of convergence of the power series ~~=o unz n be R, 0 <
R < 00. Find the radiuses of convergence of the following series
00
a) L(2n - l)u n z n j
n=O
Answers. a) R/2; b) 00; c) O.
Example 6.10 Find two power series A and B such that the radius 0/ convergence
0/ their product AB is strictly greater than their radiuses R(A) and R(B).
Solution. We can take the power series of the following functions

1-z 2-z
..,-----:-;----,- and ..,---...,.-:----,-
(5-z)(2-z) (1-z)(3-z)
Exercise 6.11 Let U n = an /or n 2: 1 and V n = bn /or n 2: 1, a f: b, and if:

a = (1 - vo)(l - b) and b = (uo - l)(a - b). Prove that then the series A . B, /or
00 00
A = L: unz n and B = L: vnz n , converges /or alt z.
n=O n=O

z (tz)2k
00
J(z) = "2 {; 2k k!(k + I)!
is the solution 0/ the ordinary differential equation
Example 6.13 Find apower se ries expansion in a neighborhood 0/ zero /or the
/ollowing /unctions and find their radius es 0/ convergences.
1 1 1
a) - -, u f: 0; b) c)
uz +v z2 - 5z + 6' (z+l)2'
1+z
e) 1og--'
d) sin 2 Z;
1- z'
f) v'ZTI, (0=1)j
g) i Z
o eZ 2 dz; h) i
o
Z
-
z
- d Zj
sinz i)
1
er=z·,
j) (log(l - Z))2.
Solution. All functions are analytic at the point z = 0 and can be expanded in
apower series at the point z = o.
1 _ 1 1 1 00 n (uz)n 00 n un n
a) uz + v -;;. 1 + uz = ;; I)-l) --;.;- = L:(-1) v n+! z ,
v n=O n=O
Slnce
1
- =L
00
zn, Izl< 1.
1- z n=O
The radius of convergence is I~I.

b) Put
1 1 A B
= = - - + --.
Z2 - 5z + 6
-::----:---=-
(z - 3)(z - 2) z- 3 z- 2
Then we obtain A = 1 and B = -1. Therefore we have
1
Z2 - 5z + 6
The radius of convergence is 2.

c)
L( -lt+1nz
00
= n- 1
n=O
00
= L(-lt(n + l)zn.
n=O
The radius of convergence is 1. Why is it possible to exchange the order of differen-

tiation and SUffi ?
d) S'mce we h
ave ' 2 2z ,and t h e senes
sm 2 z = 1 - cos . Eor t h e f '
unctlOn .
cos 2z IS
(2z)2n 1 00 22n 2n
t -(-)-" t
00
cos2z = L(-l we obtain sin 2 z = -2 L(-l +1-(z)'

n=O 2n . n=O 2n .
The radius of the convergence is 00.
e) Since
zn zn
+ z) = L( -lt+1-, L -,
00 00
log(l and log(l - z) = -

n=O n n=O n
we obtain
logl+z = log(l+z)-log(l-z)
1-z
The radius of convergence is 1.

f) The function j(z) = (z + l)a, a E ~, has power series expansion at the point
z=O
the binomial series, with the radius of convergence 1. For a E ~ \ Z, the point z = -1
is the branching point of the function (z + l)G. In our case we have a = 1/2. Since
we want to expand the branch for which y'(O+l)
= 1, we obtain
Vz+T = f (!)zn,
n=O n
with the radius of the convergence 1.
g) Since
Z2n
-"
n.
1z 1L L -n!1 1z z 2n dz = L --:-;-----,-
we obtain
2 Z 00 z2n 00 z2n+l00
eZ dz = -dz =
o 0 n=O n! n=O 0 n!(2n + 1)·
n=O
The radius of converges is 00. Explain why it was possible to exchange the order of
integration and SUffi
h) Since
z z2n
E(-l t (2n + I)!'
sin 00
-z- =
we have
=
00 (
-
l)n 1z z 2n dz
E(2n+1)! 0
E(-l t (2n + 1)!(2n + 1)

00 z2n+l
=
The radius of the convergence is 00.
i) The desired expansion is
1 = 1+L
e1-z
00 (
Ln 1(n - 1)) zn.
k' k _ 1
n=l k=l .
j) The desired expansion is

1
!; (; k(k _ n) zn
00 n
(Iog(l - Z))2 =
= 2(z22 + (1 + ~)23z3 + (1 + ~2+3~)4z4 + ... + (1 + ~2+3~ + ... + !..)

n z
zn + ... )
Example 6.14 Find the power series expansion at the point z = v for the next
functions with their radius of convergence
z
a) z + z j c) cos Zj e) log(l+z).
Solution. a) To find the power series of the function f(z) = _z_ at thepoint
z+z
z = v (v =I- -z) we put z - v = w. Then we have
z
f(z) =
z+z
l _ _z_
z+z
= 1-----
w+v +z
00 wn
1- z E(-lt (v + i)n+1
00 n (z-v)n
= 1-z E(-l) (v+z)n+l'
The radius of the convergence is Iv + zl.

b) First we shall find the power series expansion of the function for _z- at the
z+l
point z = v. We have z - v = w (v =I- -1)
z 1
1---
z+l z+l
1
1----
w+v+1
Then
(n
E(v + l)k+l (v + l)n-k+l
(_l)k (_l)n-k)
=,?;
00
wn
Finally, we obtain
2 1
= 1 - - - + ..,.------,c-
z+l (z+l)2
(_l)n n +1
-- 1- ~
" (v + 1 )n+1 (2 - - ) (z -
00
v)n
V +1 .
The radius of the convergence is Iv + 11.

c) Putting z - v = w we obtain
00 w2n 00 W2n+1
cosv E(-l t -2n()'. - sinv E(-lt (
+ 1)'.
smwsmv
n=O n=O 2n
cos v E( -1) n(z-v)2n

00
(2n)!
.00
- sm v E( -1) n{z-v)2n+1
+ I)! .
(2n
The radius of the converges is 00.
e) We have for v -:f -1 :
log(l + z) log(l + v + w)
= {W 1 dw
10 1+v +w
= lo
w 00 wn
I) _l)n (V + 1)n+l dw
n=l
00 wn+1
~(_l)n (v + 1)n+1(n + 1)
00 wn
~( l)n-l----;-_:-
~ n(v + l)n
00 (z _v)n
= ~ (-1 ) n-l ---'-:-----',:--
~ n(v+1)n·
The radius of the converges is Iv + 11.
or more general
Solution. Using the power series representation of the function eZ we obtain
and
Example 6.16 Prove that the Junction J(z) = _z_ can be represented by
eZ -1
z ~ Bn n
- - - L.J - z
eZ -1 - n=O n".
where B n are the Bernoulli numbers Jor which it holds
Bo = 1, ( n +
0 1) Bo + (n +1 1) BI + ... + (n +n 1) B n = O.
Solution. We shall prove that
1= ( f: B;n. zn) . (~)

n=O z
with the prescribed properties of B n • Then we can easily obtain the desired power
series as inverse of (eZ - 1)Jz. The inverse exists since Uo = 1 ~ 0 :
n B
=EE k!(n + 1 -
00
k n
n=Ok=O k)!z
=1.
Comparing the coefficients by the same power z we obtain
B - 1 Bo BI Bn = O.
0- , O!(n+1-0)!+1!(n+1-1)!+···+n!(n+1-n)!
Multiplying the last equality by (n + 1)! we have
(n +
o 1) Bo + (n +1 1) B + ... + (n +n 1) B
b n = O.
Izl < p.
00
Exercise 6.17 Let J be given by J(z) = E unz n , which converges Jor
n=O
a) Prove the inequality
00
Izol = ro < Pi
b) prove that the series
has a radius oJ convergence ~ p - ro;

c) prove that
Example 6.18 Let 9 be a continuous Junction on the path Land let the region 0
be in the region whose boundary is the path L such that inf Iz - wl = p =I 0 Jor
z E Land w E O. Prove:
a) J(z) = r
g(u) du is an analytic Junction in O. Expand _1_ zn apower
ftu-z u-z
series with respect to z - w, w E O.
b) J{n}(w) = n! r g(u)
JL (u - w)n+1
du Jor n ~ 1, w E O.
Solution. a) The analyticity of the function J follows from
J(z)
r g(u) du
JL U - Z
J, g(u)--·
L
1
u-w
1
-_-du
~
u-w
r g(u).
lL w 1_
U -
1
z-w
du
u-w
=
JL
r Jt0!l f (~)n du
U - W n=O U - W
00 r g(u)du n
'foJL (u _ v)n+1 (z - w) .
Explain the exchange of L: and f.

b) By a) and the power expansion of the function J at 0 we have
J{n}(w)
I =
J, (
g(u)
) +1 du for n ~ 1, w E O.
n. L u- W n
Example 6.19 Let J be an analytic Junction in Izl < R (R > 1). Starting Jrom the
integral
where L is the circle z( t) = eh, 0 S t S 211", prove that
:;2 Jar
21f t
f(et')cos 2 (2) dt = 2f(0) + 1'(0),
~ 1 21f
f(e ti ) sin 2 (~) dt = 2j(0) - 1'(0).
Solution. We have
r (2±(z+~))f(z)dz= r (2±z)f(z)dz+ r j(z) dz.

JL Z Z JL z-o JL (z-O)2
i (2 ±
Therefore
(z + ~)) f~z) dz = (2f(0) ± j'(0))27rZ. (6.2)
On the other side, putting z = et ., 0 S t S 211", we obtain
z+ 1 dz
cos t = __ z, dt = -
2 zz
i (2 +
Therefore
(z + ~)) f~z) dz = 1
z
21f
(2 + 2cost)j(e tl ) dt.
i +
Therefore
(2 (z + ~)) f~) dz = 2z 10 21f 2 cos 2 (~)f(et.) dt. (6.3)
Comparing (6.2) and (6.3) we have
We can analogously prove the second desired equality taking in the starting integral
minus sign.
Exercise 6.20 How many different values can have the integral
r dz
Je jn(z)'
where j(z) = (z - Zl)(Z - Z2)··· (z - zn), Zi ::I Zj for i ::I j, and the path C does not
cross any point zi,i = 1, ... ,n
Answer. Using the Cauchy integral formula we obtain that the desired number
is 2n - 1.
. . 1
Exercise 6.21 Fznd the zntegral -2
1I'Z
1e z (
eZdz
1- z
.
)3' iJ:
a) the point z =0 is inside, and z =1 is outside 0/ the region whose boundary

is Cj
b) the point z = 1 is inside and z = 0 is outside the region whose boundary is Cj
c) the points z = 0 and z = 1 are inside oJ the region whose boundary is C.
Answers. Using the Cauchy integral formula we obtain

e
a) 1j c) 1 - -
2'
where it is used the representation of ( 1 )3 by simple fractions.

z1-z
Exercise 6.22 Let / be an analytic /unction in a region containing the point (0,0)
and bounded by the closed path C. Prove that Jor any choice 0/ the branch oJ Ln z
we have
1 . [ j'(z)Ln zdz = J(zo) - J(O),
-2
'n Je
where Zo is the starting point 0/ the path C.
Hint. Use partial integration.
Exercise 6.23 Prove (Cauchy integral /ormulas Jor an unbounded region):

Let C be a closed path which is the boundary 0/ a bounded region D. Let the
/unction J be analytic outside 0/ the region D and lim J(z) = A. Then
z""'OO
_1_ [ J(u) du = { - J(z) + a, z outside D

211'z Je u - z A tor z E D
The path has positive orientation with respect to the region D.
Hint. First consider the case A = O.
Exercise 6.24 Find the power senes expansion 0/ the /unction
21-
J(z) = -?z 3"1
( VI = -1 +2 zJ3)
at the point z = 1 and find the radius oJ the convergence oJ the obtained series.
Answer.
-1 ~ zV3 E(!) (z -Ir,

with the radius of the convergence 1.
Example 6.25 Prove that the coefficients of the expansion

1
E
00
---- = anzn
1- z - Z2 n=O
satisfy the equality an = an-l + an-2 (n 2:: 2). Find an and the radius of the conver-
gence (an are the Fibonacci numbers).
Solution. To solve the difference equation
we put an = qn. This gives q2 - q - 1 = 0, with the solutions ql,2 = 1 ±2.J5. Then
the general solution is given by
an
_ C
- 1
(1 + .J5)n + (1 - .J5)n
2
C
2 2 ' (6.4)
for arbitrary real constants Cl and C 2 •

On the other side, we have
11 00
-z2- z +I = (z-z~) (l-ZLy'K) = ~anzn.
n= _1
.J5 ((1 +2.J5)n+l _ (1 _2
Then
.J5)n+l) >
a ' n - 0,
which is of the form (6.4). The corresponding radius of convergence R is given by
R- l = .J5 + 1.
2
Example 6.26 For the functional series
00 zn
F(z)=E--
n=l 1- zn
00
find its power series 2: anz k and find its radius of convergence.
k=l
zn
Solution. The functions i(z) = -1--'
- zn
n = 1,2, ... , are analytic functions in
the disc Izl < 1.
The series
00 zn
EI-zn
is uniformly convergent on every closed subset F from Izl < 1. This follows for
z E F (Izl = p < 1) by the inequality
zn I pn
I- - <- - < -pn-
1 - zn - 1 - pn - 1 - p'
since the series
is convergent.
We find for the function in its apower series in the form
L
00
in(z) = zn + z2n + z3n + ... = zkn (n = 1,2, ... ).

k=l
We see that the coefficient by zr is 0 if n is not a divisor of rand equal 1, if n is a

divisor of r.
We shall now find the coefficients a r for the function F
L L L
00 00 00
F(z) = zkn = arz r.

n=l k=l r=l
By the preceding reasoning ar is equal to the sum of such a number of 1 which is

equal to the number of all divisor of the number r. We denote this number by 7(r).
Specially,
7(1) = 1, 7(2) = 2, 7(3) = 2, 7(4) = 3, ...
So we have
L
00
F(z) = 7(k)· Zk.

k=l
Since F is an analytic function in the unit disc we obtain that the preceding series
converges in the unit disc {z Ilzl < I}.
For z = 1 the obtained series diverges, since it reduces on the series whose
numbers are natural numbers 00
L 7(k).
k=l
Hence the radius of the convergence of this series is one.
00
Example 6.27 Find Jor the Junction J(z) = e1/(1-z) its power se ries L anz n and
n=O
find the radius oJ the convergence oJ this series.
Solution. We shall use the equality e1/(1-z) = e·ez/(1-z). We have the expansions
z
L
00
w = <p(z) = - - = zr for Izl < 1,

1- Z r=1
and so
wk
= ew +1 = e· L -,
00
J(w) for Izl < 00.

k=O k.
We have for Izl < 1
( Z+Z 2 +Z3 + ... )k
Since
k(k+1)···(k+n-1) (k +n - 1)(k +n - 2) .. · (n + 1)
n! (k - 1)!
(n +
k -1
k-1) '
we obtain
(z + Z2 + z3 + .. .)k = f:
n=O
(n +k -
k - 1
1)zn+k.
Replacing the expansion of w = <p( z) in J(w) we obtain
1 00 1 00
J(z) = e(1 + lf E zn+1 + 2! E(n + 1)zn+1
+-1 ~
~
(n + 2)(n + 3) z n+3 + ... + -1 ~
~
(n +k - 1) z n+k + ... )
3! n=O 2! k! n=O k - 1
e(1 + z + ( -1!1 + -2!1) z 2 + (1-1! + 2-

1 + -1) z 3 + ...
2! 3!
+ ( 1!1 + (n -1 1) 2!1 + (n -2 1) 3!1 + ...

+ ( n -1) 1
n-2 (n-1)!
+1)
- zn
n!
+ ... ) .
Since the function j is analytic in the unit disc {z I Izl < I} the obtained series
converges in this disco The point z = 1 is a singular point of the function j and
therefore R = 1.
Example 6.28 Let

00
Lan(z- at
j (z) = ~n:,,-::=O,----_ __
L bn(z - at
n=O
jor Iz - al < R, be such that in the given disc both series converge an the series in the
denominator has no zeros and bo -=J O. Express the coefficients Cn in the expansion
=L
00
j(z) Cn(z - at
n=O
by the coefficients ak and bk in the disc Iz - al < R.
Solution. By the given conditions the fraction
n=O
is an analytic function in the disc {zl Iz - al < R} and therefore there exists its
Taylor series
00
The equality
00
Lan(z-at 00
n:,o = L cn(z - at,

L bn(z - at n=O
n=O
implies
00 00
n=O n=O
00 n
= E ECkbn-k(Z - at·
n=O k=O
By the uniqueness of the expansion in power series in the disc {zllz - al < R} the
coeflicients by the same powers of z - aare equal
Cobo = ao,
Cob l + clbo = ab
Cob2 + Cl bl + C2bo + a2,
This is an infinite system of linear equations with unknown Co, Cl, C2, ... , cn • ••
This system is of special form since for every n (n = 0,1,2, ... ) the first n + 1
equations contain only the first n + 1 unknown Ck. Therefore we obtain (the solution
by determinants)
bo 0 0 ao
bl bo 0 al
1
b 2 bl bo a2
Cn = bgH
bn bn - l bn - 2 ... an
where
bo 0 0 0
bl bo 0 0
~ bl bo 0 = b~H f:. o.
bn bn - 1 bn - 2 bo
Remark. The result obtained can be useful for finding power series of the frac-
tion oftwo analytic functions, e.g., we obtain for the function J(z) = _z_, Co = 1,
e Z -1
1 1 0 0
2r
1 1 1 0
3r 21
1 1 1 Bn
Cn = (_1)n 4T 31 2r 0
-"
- n.
1 1 1 1
(n + 1)! nr (n -I)! 2r
where B n are the Bernoulli numbers.
Finally, we have
Z Z ~ B 2k 2k
eZ -1 = 1 - "2 + ~ (2k)!z .
Example 6.29 Let
=L
00
<p(z) an(z - zot

n=O
be apower series with the radius of convergence R. Let w be a point fram the bound-
arg and let Zl be an arbitrary point on the straight line segment [zo, w] different from
Zo and w. Prave that the point w is singular for <p if
and it is an ordinary point if
1
ß < ------,===;'=;===
( ) ,
.
11m n l<p n (Zl)1
sup ,
n~oo n.
Hint. Use Figure 6.1 and the expansions of the function <p at the points Zl and w
and the fact that for the point Zl
00
<p(z) = L bn(z - zd n
n=O
6.1. THE POWER SERlES REPRESENTATION 151
for n = 0,1,2, ....
Figure 6.1
Example 6.30 (Pringshajm theorem) Prove that if for the power series cp(z) =
L: anzn we have R
00
=1 and an ~ n for n = 0,1,2, ... , then the point z =1 as a

n=O
singular point for the function cp.
Solution. Let x be an arbitrary point from the open interval (0,1). If, we
suppose that z = 1 is not a singular point of the function cp, then by Exercise 6.29
we have
1
A = 1 -Ix - zol = 1 - x< ,
cp(nl(x)
lim sup n _ __
n .... oo n!
Further, let w be an arbitrary point on the unit circle Izl = 1, and Zl a point on
the straight line segment [0, w) and which is also on the circle Izl = x. Then
R-lzI-Zol = I-x.
On the other side we have
Icp(nl(zl)1 = la n + n ~ 1 an+1 z 1 + (n + l~n + 2) an+2Z~ + ... 1

n+l (n+l)(n+2)
+ -1- an+1 X + + ...
2
:5 an 2 an+2 X
= cp(nl(x).
Therefore
1 1
---r==;;=:==>-----;==::;:=:::::::=
• n Icp(n)(zdl . n Icp(n) (x) I
hm sup , hm sup ,
n--+oo n. n-+oo n.
By the last inequality and the consequence of the supposition that the desired state-
ment is not true we obtain
Since w was an arbitrary point on the circle we obtain by the last inequality and
Exercise 6.29 that every point of the circle Izl = 1 is an ordinary point for cp, which
is a contradiction with R = 1. Hence the desired statement is true.
Exercise 6.31 Prove that in Example 6.90:

a) It is enough to suppose an ~ 0 for n ~ no.
b) The point w on the unit circle Izl = 1 is singular if anw n ~ 0 for n ~ no.
Hints. a) Represent the function cp in the form

no-l 00
~ akzk + ~ ak zk .
k=O k=no
b) Follow the proof for z = 1.
Example 6.32 Let

2k
;k2'
00
cp(z) = ~
k=O
prove that
a) the function cp is continuous and differentiable infinitely many tim es on the
disc Izl 5 1;
b) the function cp has no ordinary points on the circle Izl = 1.
Solution. a) We have for Izl 5 1

1 z2 2
L L
00 00
Since the series "k2 is convergent we have that the series 2k2 converges ab-
k=O 2 k=O
solutely and uniformlyon the closed unit disc Izl 1. s:
Differentiating the function <p (the series under the surn)
This series converges uniforrnly in the disc Izl s: 1, since
b) Since the coefficients in the series of the function 0 for k = 0,1,2, ... , we obtain by Pringshajrn's theorern-Exarnple 6.30
that the point z = 1 is singular for <p. By Exercise 6.31 we obtain that also singular
are the points {j = 20, n is an arbitrary natural nurnber, since
for k 2': n.
Therefore the set of all singular points is dense on the circle Izl = 1. Hence on
the unit circle there is no ordinary points of <p.
Remark. It is interesting that the singular points frorn the boundary have no
influence on the rnain properties of the considered function, it is continuous and
differentiable at the boundary points.
Example 6.33 Expand the function e,mArcsinx in apower series with respect to x.
Solution. The function y = e,mArcsinx satisfies the differential equation

y'/ y = ~m/~, which after squaring and integrating reduces on the equation
The given function forces the following initial conditions y(O) = 1 and y'(O) = ~m.
This initial problem has a unique analytic solution in the neighborhood of zero
L
00
and which has the power series form anx n . So we obtain the following recursive
n=O
formula
(n + 2)(n + l)a n +1 - n(n - l)a n - nan + m 2 an = 0 (n E NU {O}).

Hence
n2 _ m2
an+2 = (n + l)(n + 2) an (n E NU {O}). (6.5)
Therefore by the initial conditions we obtain
and
k ~m(m2 -1)(m 2 - 9)··· (m 2 - (2k - 1)2)
a2k+l = (-1) (2k + I)! .
The equality (6.5) implies that the radius of the convergence is 1.
Exercise 6.34 Let f be an analytic function in Izl < R ' . Prove that
a) f(a) = -2
7r~
1
(
1
c z - a
)(R2
R2 - aii.
_/(z) dz, for lai< R < R, and C is a circle
- za
I
Izl=R.
b) (Poisson's formula)
for r < R.
Hints.a) Apply the Cauchy integral formula to
lai< R< R'.
b) Put a = re e., 0< r < R, in the equality in a).

Exercise 6.35 Let D be an open disc with the center Zo and radius ro. Let Uk(Z)}
be a bounded sequence of of analytic functions on D, which satisfy the condition:
For every fixed n ~ 0 the sequence u1n )(zo)} converges.
=L
00
a) Let fk(Z) an.k(z-ZO)'" What are the discs of convergence ofthese series'?
n=O
Find the upper bound of an.k which is independent of k.
b) Prave that an.k - an.h ~ 0 for a fixed n, as k and h tend to infinite independent
of each other.
c) Using b) prave that
lim !fk(Z) - fh(Z)! = 0 for every Z E D.

k.h ..... oo
Example 6.36 We know that E~=o zn = 1/(1 - z) for Izi < 1. Examine the situa-
tion re 'D ~ e'D as r ~ 1 for () =1= 0, 27r.
Solution. We have for Z =1= 1
lim-1- = _1_.
1- Z
r ..... 11 - e,D
On the other side the series E:'o eonD diverges.
Example 6.37 (Tauber theorem) The radius of convergence of apower series

E~=o unz n is one and limn ..... oo nUn = O. Prove that
b) if there exists lim.,..... 1 f(x) = A, where f(z) = E~=o unz n for Izl < 1, then the
series E:'o U n converges (to the sum A).
Solution. We have for every f > 0 nlunl < c/2 for n > N(c). We fixe one
n = no which satisfies the preceding condition. Then we have the estimation for
n=1
m
for m> no:
n=1
= :..:..n=-=1'---_ + n=no+1
m m m
no
Lnlunl
n=l (m - no)
< m + 2m c:
For enough big m > M we have
n=l c:
m < 2'
and so
n=l
< c:,
m
for m > max(n(c:), M).
b) We have
n =
< (1 - x) L IUkl(l + x + ... + X k- 1 ) + L IUkl xk
k=O k=n+l
Let mluml < c:/2 and so by a)

m
Lklukl
k=O m c: l'lor m > M( c.
<2 )
Then we obtain by the preceding inequality for n > M(c:)
< n(l - x)- + -

c c:
L= xk
2n k=n+1
2
C 1 x n +1
- (n(1 - x) + - . - )
2 n 1- x
c 1
< 2(n(1- x) + n(l _ x))'
Then for 1 - x = 11n and n > M(c:)
It
k=O
Uk - f(l - ~)I < c,
n
which implies
lim
n-+oo
t
k=O
Uk = n-+oo
lim f (1 - ~)
n = A.
Remark. Hardy and Litlewood proved more general Tauber theorem, where
instead of the condition limn -+ oo nUn = 0 they required only boundedness of the
sequence {nun}.
Example 6.38 (Schwarz lemma) Prove that if f is an analytic function on Izl <
1, If(z)1 ~ 1 and f(O) = 0, then If(z)1 < Izl for Izl < 1 and 1f'(0)1 ~ 1, excluding
the case f(z) = e,9 z (0 real).
Solution. By f(O) = 0 it follows the existence of an analytic function on Izl < 1

such that f(z) = zg(z). We have for 0< r < 1
M(r) = max Ig(z)1 = max Ifl(zl)1 ~ ~.

Izl=r Izl=r z r
The function M = M (r) is monotone increasing by the maximum principle and so
M(r) ~ limM(t)
t-+l
~ lim ~ = 1.
t-+l t
Therefore Ig(z)1 = 1. By the principle of the maximum there is no interior point

where Ig(z)1 = 1, except g(z) = k and Ikl = 1.

a) For every complex number U
(U
-
n!
n )2
= -
1 1
une"Z
- - d z for n = 0,1,2,···,
2n c n!zn+1
where C is a closed path and the region (C) contains the origin.
b)
Hints.a) Use the Cauchy integral formula for the n-th derivative.
b) Expand the function e U/ z in apower series with respect to u and then apply
a) on the function !e
z
u (z+1/ z ).
Exercise 6.40 Let f be analytic on Izl :$ r. Prove that
1 (27f
f(O) = 271' Jo f(re,t) dt,
i.e., the value of an analytic function in the center of a disc is equal to the arith-
metical mean on the boundary of the circle
1. f(r) + f(rw n ) + ... + f(rw~-l)

1m ,
n-+oo n
Hint. Use in the function under the integral the representation
= I: anrne,nt
00
f(re lt )
n=O
and the equality

for k =I m
- 1 127f zkZ""dt = { 0
271' 0 1 for k = m.
Exercise 6.41 Let f be analytic on D(O, r) = {z Ilzl :$ r} and f(z) =I 0 for z E

D(O, r ). Prove that
If(O) 1 = exp (-d; i 27f In f(re lt ) dt)
Ji.~ .jf(r)f(rwn ) · · · f(rw~-l),
Hint. Apply on the function Inf(z) Exercises 6.40.
Exercise 6.42 Let Un} be a sequence of analytic functions on the disc D(O, r)
which is uniformly bounded, i.e., there exists M > 0 such that If(z)1 < M for
n E N, Izl < r. Prove that i f the sequence {fn} converges on a set with accumulation
point 0, then
b) the power series Ek:o akzk defines in the disc D(O, r) an analytic function
fez);
c) limn _ oo fn(z) = fez) uniformlyon Izl ~ rI·
Hints.
a) Apply Schwarz' lemma to the function fn(z) - fn(O). Hence
for Izl ~ rI· Choosing a point ZI f:. 0 in which the sequence {in} converges, such
that the right side in the last inequality is enough small we can easily prove that
laQ' - aöl = Ifm(O) - fn(O)1 tends to zero as m, n - t 00, limn _ oo aö = ao. Applying
the preceding procedure on the sequence {(Jn(z)-a ö)/ z} we obtain limn _ oo a]'= al.
Continuing this procedure we obtain the desired statement.
b) Use the Cauchy inequality and the Weierstrass criterion for uniform conver-
gence.
c) Since
rn-I 00
fn(z) - fez) = L (al: - ak)zk + L (al: - ak)zk,

k=O k=rn
we have to estimate each sum on the right side.
Exercise 6.43 (Vitali theorem) Let {in} be a sequence of analytic functions on

the region 0, which is uniformly bounded on O. Prave that if {In} converges on the
set with an accumulation point in 0, then it converges uniformlyon every compact
subset of 0 to an analytic function.
Hint. AppIy the Heine-Borel theorem related to compact sets and reduce to Exer-
eises 6.42.
Exercise 6.44 (Hadamard Three Circle Theorem) Let f be analytic on the

annulus rl ~ Izl ~ r3, centered at the origin. let M(r) = sUPlzl=r If(z)l. Ifrl < r2 <
r3, then
In M(r2) ~ In r2 -In rl In M(r3) + In r3 - In r2 In M(rI),
In r3 - In rl In r3 - In rl
i.e, InM(r) is a convexfunction oflnr.
Hint. Consider the function za J(z) on the annulus A = {z Irl ::; Izl ::; r3}' The
maximum of modulus of zaJ(z) in Ais either r~M(rd or r~M(r3)' Choose a such
that r~M(rl) = rgM(r3), and put this value in the inequality
Exercise 6.45 The function In M(r) form Example 6.44 is strictly convex function
of In r, if the function is not of the form bz a, where a and b are real constants.
Hint. Use the preceding Example 6.44 and the fact that za J(z) attends its maxi-
mum of modulus on the circle Izl = r2 only if za f(z) is a constant.
Example 6.46 (Montel theorem) Let Un} be a sequence of analytic functions

on a region 0, which is uniformly bounded on O. Then there exists of it a subsequence
which uniformly converges on every compact subset of O.
Solution. Let {zn} be a sequence from 0 which has an accumulation point in O.

Since Ifn(Zl)1 < M,n E N, for some M > 0, there exists a subsequence Unl(i)} of
{fn} such that there exists
Now in the same way we extract a subsequence {fn2(i)} of Unl(i)} such that there
exists
Continuing this procedure we obtain a sequence of subsequences {fnk(i)} such that

Unk(i)} is a subsequence of Unk(i)} and there exists
Then the desired sequence is the diagonal sequence {fnk(k)}'
Exercise 6.47 (Riemann's Mapping Theorem) Let 0 be a simply connected

region whose boundary is with minimum two points in the extended complex plane.
Let Zo E O. We denote by F the set of all functions f which are analytic and univa-
lent, i.e., J(zt) = f(Z2) implies Zl = Z2, on 0 and If(z)1 < 1 Jor z E 0, f(zo) = 0
and !'(zo) > O. Prove that
a) the set F is non-emptYj
b) sUPfE:!"f'(zo) = M < 00 and there exists in F a function with a maximal

derivative at Zoj
c) (Riemann's Mapping Theorem) there exists an analytic Junction J wh ich con-

Jormaly and bijectively maps the region 0 on the unit disc D(O, 1) such that
J(zo) =0 and !'(zo) > 0;
d) the Junction Jrom c) is unique.
Hint.
a) Choose the branch of the function Log (z - a), where a f. 00 is the boundary
point, which is uniquely determined on O. Let Wo = log(zo - a). There exists
s > 0 such that
Ilog(z - a) - Wo - 27rZI ~ s.
Then the analytic function
J(z) = slg'(zo)l(g(z) - g(zo)) ,
1 + sg(zo
where g(z) = (log(z - a) - Wo - 2n)-1, belongs to F.
b) Consider the supremum of numbers !,(zo), J E F, and use Example 6.46.
Prove that the limit of the obtained subsequence is the desired function, using
Rouche's theorem 8.6.
c) Use the function J from b). To prove that J(O) = {w Ilwl < I} suppose that
this is not true, i.e., that there exists e E oj(O) such that lei< 1. Then the
function
(p(z) - y'-C)e,argFc
h(z) = ,
1 - y'-Cp(z)
where p(z) is one branch of the function
~ J(z) - e
y~= l-cJ(z)'
belongs Fand
h'(zo) > sup !,(zo),
JEF
what is impossible.
d) Use that the only automorphism of the unit disc with J(O) = 0 are given by
J(z) = e'oz. This fact follows by Schwarz' lemma.
Exercise 6.48 Let A be a simply connected region and 0 f/. A. Taking Zo E A fix a
value Log Zo and put
J(z)
w
1.
dw + logzo.
=
zo
z
Prove that J is an analytic brunch oJ Log z in A.

6.2 Composite Examples

Example 6.49 Prove:
1. 11 1 is an analytic lunction in the region 0 C C and il this lunction and all
its derivatives are zero at the point a E 0, then I(z) = 0, z E O.
2. Let M(R) = sup I/(z)l. 11 1 is an analytic lunction in the entire complex
Izl<R
plane and lor fixed n we have
M(R)
- - -+ 0 as R -+ 00,
Rn
then
a) l(k)(O) = 0 lor k ~ nj
b) the lunction
n-l
F(z) = I(z) - 1(0) - z 1'(0) _ ... _ z I(n-l)(o)
(n - 1)!
and all its derivatives are zero lor z = O.
c) 1 can be only a polynomial 01 order not greater than n - l.
3. Let 1 be an analytic lunction in 0 and the circle C with mdius Rand center
a E 0 lies in O.
a) Prove that lor all z which are inside 01 C and z =F a
I(z) = I(a) - (z _ a)f'(a) _ ... _ (z - a)n-l I(n-l)(a)

(n - 1)!
(z - a)n { I(u)du
- 2n Jc (u - a)n(u - z)"
b) 11 1 has a zero 01 order n in the point a and
1 (z - a)n { I(u)du
(z) = 27rz Jc (u - a)n(u - z)'
then
Iz - al)n R
I/(z)l:::; ( ~ R-Iz-al M,
where M = sup
zEC
I/(z)l.
4. 111 has zero 01 order n prove that P(z) has zero 01 order n· k and that
I/(z)1 :::; M(lz - al)n .k! R .
R VR-Iz-al
6.2. COMPOSITE EXAMPLES 163
Solution.
1. We shall expand the function j in a neighborhood of the point a in apower
series (what is possible since j is an analytic function)
j(n)()
L
00
j(z) = _,_a (z - at, (Iz - al < r).

n=O n.
Since by the suppositions
j(z) = j'(a) = '" = j(n)(a) = 0
we have j(z) = 0 for Iz - al < r, where r > 0 and the circle C(a, r) is in the region
O. Therefore the functions j and O(z) = 0 are equal on the whole region 0 since
both are analytic on 0 and equal on C(a, r).
2. a) Let r < R. Since j is analytic in the entire complex plane we have
Therefore
Ij(n)(O)1 < n! r2" Ij(r eBi)1 dB< M(R) n! = M(R) n!, (6.6)
- 271" Jo rn - rn (R - 8)n
where r < Rand R = r+8 (8 > 0). Since
M(R) M(R)
(R - 8)n Rn - (7) Rn- 8 + ... + (-8)n
1
MAl!)
1- (~) R{j + ... + ( 7r8)n,
M(R) M(R) ..
we have for R ........ 00 by ~ ........ 0 that also (R _ 8)n ........ O. Then by the mequahty
(6.6) we have j(n)(o) = O.
Since for k ;::: n we have
M(R) = M(R) _1_ ........ 0

Rk Rn Rk-n
as R ........ 00, we obtain in an analogous way as before for n that j(k)(O) = 0 for k ;::: n.
2. b) and c) By a) we have
n-l
fez) = f(O) + z 1'(0) + ... + z f(n-l)(o).
(n -1)!
Hence
F(O) = F'(O) = ... = 0
and the function f is a polynomial of the order not greater than n - 1.

3. a) We have for z =I- a :
~
{ f(u)du (z-a)n { ~ du
Je (u - a)n(u - z) 2n Je u- z
(z - at fez)
(z - a)n
= fez).
We have on the other side for z =I- a :
~
{ f(u)du = (z-a)n { ~ du
Je (u-a)n(u-z) 27r~ Je u - z
= (z - a)n (f(U) ) (n-l)
(n - 1)! u - z u==a
( )n-l
f(a) - (z - a) f'(a) _ ... _ z - a f(n-l)(a).
(n - 1)!
So we have proved the both equalities.

3. b) Since the function f has a zero of order n at the point a we have in a
neighborhood of a that
fez) = (z - at h(z),
where h is an analytic function and h(a) =I- O. Then
hz _ _1_ f(u) du
( )- 27r~ Je
{
(u - a)n(u - zr
Further we have
l1(z)1 < Iz - al n f If(u)lldul

- 27l' Jc lu - al n lu - zl
Mlz - al n f21r IzR eO'dBI
:S 27l' Ja la+Reo'-alnla+Reo'-zl
_ (Iz-al)n R M
- R R-Iz - al '
where M = sup If(z)l.
zEC
4. If f has a zero in a of order n, then f(z) = (z - a)n h(z). Therefore
where hk is an analytic function and hk(a) i=- 0, what follows by the same properties
of h. The last equality implies that fk has zero at a of the order nk.
Applying the inequality from 3. b) on the function r we obtain
Since Ir(z)1 = 11(z W, taking the k- root ofthe last inequality we obtain the desired
inequality.
Example 6.50 I 1. Prove that the integral
fc Idzl
is the length 01 the path C.
I 2. Let f be an analytic function in the region 0 which contains the disc
Izl :SR. Suppose that 1 maps the circle Izl = R onto bijectively the path L. Prove
that the length L 01 the path L is given by
II 1. Suppose that the sequence 01 complex numbers {u n } converges to u. Prove

that the sequence {u n } converges to U.
2: Un zn,
00
II 2. Let the function 9 has apower senes representation g( z) =

n=O
which converges for Izl < r.
112. a) Prove that for r <R
II 2. b) Find for which r the function is J(r) is continuous. Explain.

II 3. Let M(r) = max Ig(z)l, where 9 is a function given in II 2.
Izl=r
II 3. a) Prove that IUnl S; M(r) for r < R, where M(r) = max Ig(z)l.
rn Izl=r
II 3. b) If for sorne n = k is IUkl = M~r), then g(z) = Uk zk (for that proof use
r
II 2. a)).
Solution. I 1. Let C: z(t) = x(t) + ly(t), t 1 S; t S; t 2 • Then
fc Idzl = l21(x'(t) + Y'(t))ldt = l2 V(x'(t))2 + (y'(t))2 dt = C

l
where C is the length of the path C.

I 2. By the given conditions, w = f(R e O.), 0 S; B < 27l", gives the path L. By
the analyticity of f in the region 0 we have
By I 1, we have
- f f2~
= JL Idwl = R Jo 1f'(R eBi)ldB.
L
(f maps bijectively the circle Izl = R on the path L). Cauchy's formula implies
F(O) = _1 f F(z) dz.

27l"z Jlzl=R Z
Taking F(z) = J'(z) we obtain
1'(0) = _1_ f2tr f'(R eO.) lR eO. dB = ~ f2tr f'(R eO')dB.

2n Jo R eO. 27l" Jo
Since
finally we obtain
11 1. Let c > O. Then by the convergence of the sequence {u n } there exists no E N

such that JU n - uJ < c for n ~ no. Then it follows c > JU n - uJ = JU n - uJ = JUn - uJ
for n ~ no(c).
00
11 2. a) Since g(z) = 2: unz n for JzJ < R, we have

n=O
00
g(r e9.) = 2: unrne nO , for r < R.

n=O
Then
Jg(r eO·W = g(r eO') g(r eO,)
2: unrne nO , 2: unrne- nO ,
00 00
=
n=O n=O
2: 2: Un_krn-ke(n-k)O·ukrke-kO.
00 00
=
n=O k=O
Applying the integral with respect to () we obtain
J(r) =~ [2" Jg(r eO'Wd()

27l" Jo
1
(2: 2:
2, , 0 0 0 0
= - [ Un_kukrne(n-2k)O,) d().
27l" Jo n=O k=O
We can exchange the order of integration and summing since the series under integral
converges in disc JzJ ~ r < R.
Then
00
= 2: JU n J2 r 2n ,
n=O
sm ce
for n = 2k
for n =I- 2k.
11 2. b) The function J(r) is continuous for r < R, since for such real numbers
r, the function J(r) has a convergent power series representation.
L
00
11 3. a) We have for the coefficients Un of g(z) = unz n

n=O
Un = ~
21Tt
f
Jlzl=r
g~:~
Z
dz for r < R.
Hence
1 f21r M(r) D, M(r)
IUnl $ 21T Jo rn+Ile(n+I}D'1 r I z e I d() = --;:;;-.
11 3. b) The equality
implies
J.- f21r M(r?d()
21T Jo
~ f: lunl2r 2n .
n=O
Since for some n = k we have IUkl = M~r)

r
, we obtain
f
k-l 00
(M(r) ~ L lunl2r 2n + IUkl2r2k + L lunl2r 2n

n=O
~l 00
= L lunl2r 2n + (M(r)? + L lunl2r 2n .

n=O n=k+l
Since r > 0, the preceding inequality reduces on equality for IUnl = 0, n i:- k. This
implies g(z} = Uk zk.
Example 6.51 I. Let f an analytic on the disc Izl $ Rand on the circle Izl = R
we have
M
If(z)1 $ TYf for z = x + zy.
I 1. Prove that on the disc Izl $ R
I 2 Using I 1. prove that on the disc Izl < R/2 we have
If(z)1 $ ~~
I 3. Let f be an entire function and
M(r) = sup Iy f(z)l·

Izl=T
Using I 2. prove that if f is not a polynomial then
M(r)
- - -+ +00 as r -+ +00,
rP
for an arbitrary real number p ~ 1.
Hint. Suppose the contrary and then use the Cauchy inequality.
II Suppose that the function 9 is analytic for Izl < R.
II 1. Prove that
g(a) = -2
1
1I"Z
1 C
(Z -
R 2 - aa
a )(R2 - za
_) g(z) dz
lai< R < R' and C is the circle Izl = R.

II 2. Prove that for r <R
g(r eS') = -
1 1 211" (R 2 - r 2 )g(R eICI')
----'------'-'('-'---)-'--- dcp.
211" 0 R2 - 2Rr cos () - cp + r2
Solution.
I 1. We have for z = R e's = x + zy
Therefore by the given conditions on f on Izl = R
Then by the maximum principle
I(Z2 - R 2 )r(z)1 :::; 2 M R on the disc Izl:::; R.

I 2. We have for Izl :::; R/2 that
Iz 2 _ R2 1 ~ 3:2
•
Then by I 1. we have
2MR 8 M
If(z)l:::; ~ = 3· ii"
34
I 3. Suppose that M(r) does not converges to +00 as r ---. +00. Then there
rP
exists a sequence {rn} such that r n ---. +00 as n ---. +00 and
M(r n ) < C (n E w),

r~ -
for some C > o.

Hence by I 2. we have for R = rn
Ij(z)1 ~ 8~ r~-l on Izl ~ T n /2.
L:
00
Take j(z) = amz m (J is an analytic function on the disc Izl ~ R). The last
m=O
inequality and the inequality on Izl ~ ~ r n imply
8C
la m I<
- -3 r nP- m- 1 2m (m,
nE.
W)
Letting n ---. +00 we obtain a m = 0 for m ;::: p. This shows that j is a polynomial
(of the order less or equal then p - 1).
11 1. Use the definition of the integral and in 11 2. put a = res •.
Chapter 7
Isolated Singularities
7.1 Singularities
7.1.1 Preliminaries
A function 1 has an isolated singularity at Zo if 1 is analytic in {z I 0 < Iz - Zo I < r}

for some r > 0 hut is not analytic at zoo
The isolated singularities of 1 are classified in the following way.
(i) If there exists an analytic function at Zo, denoted by g, such that I(z) = g(z)
for all z E {z 10 < Iz - zol < r} for some r > 0, we call Zo a removable
singularity of f.
(ii) If 1 can be written in the form
I() =g(z)
z h(z)
for z =I zo, where 9 and h are analytic functions at Zo, g(zo) =I 0 and h(zo) =I 0,
we call Zo a pole of f. If h has zero of order n at Zo, we call Zo a pole of order
n of f.
(iii) If 1 has neither a removable singularity nor a pole at Zo, we call Zo an essential
singularity of f.
Theorem 7.1 I1 a lunction 1 has an isolated singularity at Zo and
lim(z - zo)l(z)
z-+zo
= 0,
then the singularity Zo is removable.
171

172 CHAPTER 7. ISOLATED SINGULARITIES
Theorem 7.2 11 1 is analytic in {z I0 < Iz - Zo I < r} lor some r > 0, and there
exists n E N such that
lim (z - zot I(z)

z~zo
f. 0 and lim (z - zo)n+1 I(z) = 0,
z--+zo
then the singularity Zo 01 1 is a pole 01 order n.
Theorem 7.3 11 1 has an essential singularity at Zo, then the range
{f(z) I z E {z I 0 < Iz - zol < r}

for r > 0, is dense in the complex plane.
We eall 1 meromorphic in a domain A if 1 is analytic there exeept at isolated poles.

Example 7.1 Find the singular points in the extended complex plane (with z = 00)
of the following lunctions
1 Z5 eZ
a) Z-Z3; c) (1 _ z)Z' d) 1 + zZ'
Solution. a) z = 0, z = 1 and z = -1 are poles of first order, z 00 IS

removable singular point.
b) z = l~Z and z = -~ Z are poles offirst order, z = 00 is removable singular

point.
e) z = 1 is pole of first order, z = 00 is pole of third order.
d) z = Z, Z = -1 are poles of first order, z = 00 is essential singular point.
Example 7.2 Find the singular points in the extended complex plane (with
Z = 00) of the following functions
1
1 1 e·- 1
b) - - - - . e) eZ _ l'
eZ - 1 z'
Solution. a) z = 00 is an essential singular point.

b) z = 2k1ri, k = ±1, ±2, ... , are poles of first order, z = 00 is a limit point of
poles.
7.1. SINGULARITIES 173
c) Z = (2k + 1)7l"i, k = 0, ±1, ±2, ... are poles of first order, Z = 00 is a limit
point of poles.
d) = 0 is a essential singular point, Z = 00 is a removable singular point.
Z
e) Z = 1 is an essential singular point Z = 2k7l"i, k = 0, ±1, ±2, ... , are poles of

first order, Z = 00 is a limit point of poles.
Example 7.3 Find the singular points in the extended complex plane (with
Z = 00) of the following funciions
1 cos Z 1
a) . , b)
smz Z2 '
c) tan Z; d) sm--'
1- z'
Solution. a) z = k7l", k = 0, ±1, ±2, ... , are poles of first order, z = 00 IS a

limit point of poles.
b) z = 0 is a pole of second order, z = 00 is an essential singular point.
c) z = (2k + 1)7l" /2, k = 0, ±1, ±2, ... , are poles of first order, z = 00 is a limit
point of poles.
d) z = 1 is an essential singular point, z = 00 is a removable singular point.
2
e) z = = 0, + 1, +2, ... , are essential singular points, z = 0 is
(2k + 1)7l"' k the
limit of essential singular points, Z = 00 is the removable singular point.
Example 7.4 Prove that if apower series has the radius of convergence one and
on the unit circle it can have of singular points only poles of first order, then the
sequence of coefficients of the power series is bounded.
L
00
Solution. The power series an zn which satisfies the described condit.ions

n=O
has the following representation
~
L..J anz
n b1
= -- - + ... + bp
+ ~ n
L..J cnz ,
n=O 1- ZIZ 1 -zpz n=O
where
IZil = 1 (i = 1, ... ,p) and limsup
n-+oo
~ < 1.
Therefore
an = blZ~ + ... + bpz; + Cn ·
The last equality enables to prove easily the boundedness of the sequence {an}.
Example 7.5 Let J be an analytic Junction in annulus
A = {z 11' ::; Izl ::; R}.

Prove that the surJace area 0/ the image oJ the annulus by the /unction J(z) =
L
00
anz n (counting the parts 0/ the regions as many timesas they are covered) is
n=-oo
given by
L
00
7r nlanl2(R 2n - r 2n ).
n=-oo
Solution. Let J(z) = u + zv for z = x + zy. Then the desired area is
where z = pe,a. Since

8 (u'v)1 = 1J'(z)1 2 ,
18(x, y)
we have
ir
R
Jf21r 1
o J'(pe
.
la
)
12
p dp da 27r i R
( f
n=-(X)
n2IanI2p2n-l)dp
00
n=-oo
Example 7.6 Prove that i/ the /unction J has zero 0/ order k, k 2: 2 at z = w,

then l' has at z = w zero 0/ the order k - 1.
Solution. The function / has in the neighborhood of w the representation

J(z) = (z - w)kg(z), k 2: 2, where 9 is an analytic function in the neighborhood of
wand g(w) :I O.
Then
J'(z) = (z - w)k-l(k g(z) + (z - w) g'(z)),
which implies the desired statement.
Example 7.7 Prove that iJ the Junction / has pole 0/ the order k at z = w, then
l'has pole 0/ the order k + 1 at z = w.
7.1. SINGULARITIES 175
Solution. The function 1 has in the neighborhood of w the representation

I(z) = ( g(z) )k' where 9 is an analytic function in the neighborhood of z = w and
z-w
g(w) ::I O.
Then
J'(z) = (z _ ~)k+l (-k g(z) - g'(z) (z - w)),
which implies the desired statement.
i I(
Example 7.8 Prove that ill is an entire lunction, then all its derivatives and the
integrals on arbitrary path L U )du are entire lunctions.
Solution. Since the function 1 has apower series representation with infinite
radius of convergence the corresponding power series of derivatives and integrals
have also infinite radius of convergence.
Example 7.9 Prove that ill is a meromorphic lunction on the region 0, then I'
is also meromorphic on 0 and has the same poles as f.
Solution. The function 1 has the following representations in the neighborhoods

of its poles Ul, U2, • .• , U n
j(z) = ( 9k(Z\ ' k = 1,2, ... ,n,

z - Uk Pk
where gk are regular functions and gk( Uk) ::I 0 for k = 1,2, ... , n, and Pk E N is the
order of pole.
Then
j'(z) = (z _ :k)Pk+ 1 ( - Pk gk(Z) - (z - Uk) g'(z»)
in the neighborhood Uk for k = 1,2, ... , n, which implies the desired statement.
Example 7.10 Let 1 be an analytic lunction, which is not dejined at the point
z = wand which the is analytic in the neighborhood 01 w. Prove that a sufficient
and necessary condition that the lunction 1 has a removable singularity at z = w
isthat it is bounded in so me neighborhood 01 the point w.
Solution. If the function 1 has a removable singularity at the point z = w, then

it has apower series representation in a disc Iz - wl < R. Hence for some M > 0
I/(z)1 ~ M (Iz - wJ < R).

Suppose now that J is a function such that IJ(z)1 ::; M for Iz - wl < R. Then J
has no singular point at z = w, since in the opposite it would be unbounded in the
neighborhood of this point.
Example 7.11 Prove that an entire Junction which at the point z = 00 has a pole
oJ order k is a polynomial oJ the order k.
Solution. Since the function J is entire it has apower series representation

00
J(z) = ~ unz n which converges for every z E !C.

n=O
On the other side the function J has at z = 00, a pole of the order k, which
t u:
implies that the function
J( ~) =
z n=O Z
will have a pole of the order k at z = O. Then by the definition of pole we have
Un = 0 for n = k + 1, k + 2, ...
Therefore
k
J(z) = ~ Un zn.
n=O
Example 7.12 Let J be an entire Junction and let there exist a natural number n
and real numbers M > 0 and R > 0 such that
IJ(z)1 :s M Izln Jor Izl;:::: R.

Prove that J is a polynomial 0/ the order not greater than n.
Solution. We have for the entire function J
Uk -
_ J(k)(z) _
-k-'• - - -2
1 1, J(u)
k+l du,
'lrZ L U
where L is the circle z = R et ., 0 ::; t :s 2 'Ir •
Then IJ(u)1 :s M lul n, for lul ;: : R implies

/(k)(Z) I MR"-k
IUkl = 1k!::; In - kl .
For k > n the right hand side can be managed small enough . Therefore
00
Uk = 0
for k = n + 1, n + 2, .... Hence in the representation J(z) = ~ Uk zk we have
k=O
Uk i- 0 for k ::; n.
7.2. LAURENT SERIES 177
Example 7.13 Find the order of the zero z = 0 for the following functions
Answer.
a) Fourth. b) 15th.
Example 7.14 Find the zeros and their orders for the following functions
b)
Z2 +9 c) zsinz.
Z4 '
Answers.
a) z = ±3z are zeros of first order.
b) z = ±3z are zeros of first order, z = 00 is a zero of second order.
c) z = 0 is zero of second order, z = k-rr, k = ±1, ±2, ... are zeros of first order.
Example 7.15 Find the zeroes and their orders for the following functions
Answers.
a) z = ±7r are zeros of third order, z = k7r, k = 0, ±2, ... , are zeros of first
order.
b) It has no zeros.
c) z = 0 is zero of third order, z {jh and z ~ ifh (1 ± zV3), k

±1, ±2, ... , are zeros of first order.
7.2 Laurent series
7.2.1 Preliminaries
If both series
LZn and LZ-n
00 00
n=O n=l
L Zn as
00
converge, then we define

-00
00 00 00
LZn = LZ-n+ LZn.

-00 n=1 n=O
Theorem 7.4 We have that L:~oo anz n converges in the domain
A = {z 1r < Izl, Izi < R},

where
1 1
R = 1.1m SUPn-+oo 1an 11/n and r = 1·1m SUPn-+oo 1a_ n 11/.
n
Moreover, ij r < R, then A is an annulus and J given by
-00
is an anytic junction in A.
Theorem 7.5 IJ J is analytic in the annulus A = {z 1 r < Izi < R}, then J has a
(unique) Laurent expansion
n=oo
J(z) = L anz n
-00
in A.
Theorem 7.6 IJ J has an isolated singularity at Zo, then Jor some c > °we have
00
J(z) = L an(z - zo)"

n=-oo
on A = {z 10< Izl < c}, where
1 f J(z) d
an = 27l"z Je (z _ zo)n+I z
and C = C(O, R), °< R < c.

Characterization of singu1arities by Laurent expansion:
(i) If j has a removab1e singu1arity at zo, then all coefficients an, n < 0, of its
Laurent expansion about Zo are zero.
(ii) If f has a pole of order k at Zo, then all coeflicients an, n < -k, of its Laurent
expansion about Zo are zero and a-k -=I- O.
(iii) If f has an essential singularity at Zo, then infinitely many coefficients an, n <
0, of its Laurent expansion about Zo are nonzero.
Theorem 7.7 (Mittag-Leffier) Let {zn} be a sequence 0/ distinct complex num-

bers such that IZnl --+ 00. If {Pn } is a sequence of polynomials without constant term,
then there exists a meromorphic function f whose only poles are at {zn} with
All such functions f have the following representation

1
=L
00
f(z) (Pn ( - - ) - P:(z)) + e(z),

n=O Z - Zn
where P;:' is some polynomial, and e is an entire function. The series converges
absolutely and uniformlyon every compact set not containing the poles.

Example 7.16 Expand in the Laurent senes the following functions in the annulus
where they satisfies Laurent theorem:
1 2z~
a) b)
Z2 - 3z + 2' (1-z)(Z+t)2'
1 1
c) d) sm-j
(zz+2)n' z
1
e) sinz sin-,
z
first in a neighborhood of zero and then zn a neighborhood of an arbitrary point

wEC,
Solution. a) Since
- -1- - = -1- - --,1

Z2 - 3z + 2 z - 2 z - 1
we obtain for Izl <1
1 1 1 1
Z2 - 3z +2
which is the desired expansion in the neighborhood of the point w = o.

For 1 < Izl < 2 we have
1 1 = 1 1 1 1
Z2 - 3z +2
--L:-._--.-
z z 1- 1 2 1-!
n=O z 2
the expansion in w = o.
For Izl > 2 we have
1 1 1 1 1
Z2 - 3z +2 -;·--1 +;·--2
1-- 1--
z z
1
z
f=
n=O
In
Z
+~ f= (~r
Z n=O Z
-1 1
= - L: (2 n+I -1)zn,
n=-oo
expansion in the neighborhood of the point w = o.

Case I: For arbitrary w (w i- 1,2) we have for
Iz - wl < min(lw - 11, Iw - 21),

Figure 7.1,
0
~
0/ 2 X
Figure 7.1
1 1 1
Z2 - 3z + 2
= - - - -
1-z 2-z
-
1 1 1 1
1-w·1-z -2-w· 2-z
l-w 2-w
1 1 1 1
= -----,- - - - . - - -
1-w 1_ z - w 2-w 1_ z - w
l-w 2-w
= 1
1- w
t(Z-Wr
1- w
1
2- w
t (Z-Wr
2- w
E(1 1)
n=O n=O
=
00
(1- w)n+1 - (2 _ w)n+1 (z - wt·
for min(lw -11, Iw - 21) < Iz - wl < max(lw -11, Iw - 21).

Case 11: For
min(lw -11, Iw - 21) < Iz - wl < max(lw -11, Iw - 21).

Figure 7.2
Find the expansions for cases Iw - 21 < Iz - wl < Iw -11 and Iw -11 < Iz - wl <
Iw -21, Figure 7.2 .
Case III: For
Iz - wl > max(lw - 11, Iw - 21),
Figure 7.3,
Figure 7.3
we have
111
Z2 - 3z + 2 = - z - 1 + z - 2
1 1 1 1
= --_. +-_._.....".--
2 w
z-w l_l-w z-w 1_ -
z-w z-w
= - z ~ w f: (~ =
n=O
:r + z ~ w f: (! =
n=O
:r
-1 1 1
= n200 ( - (1 - W)n-1 + (2 _ w)n-J . (Z - wt·
Find the expansions for w = 1 and w = 2.

b) The consideration is analogous to the previous example a). For arbitrary
w (w =11, -z) the regions are:
Case I: Iz - wl < min(lw - 11, Iw + zl)j
Case 11: min(lw -11, Iw + zl) < Iz - wl < max(lw -11, Iw + zl)j
Case 111: Iz - wl > max(lw - 11, Iw + zl).
Find the expansions for w = 1 and w = -z.
c) We have for n = p :
dP ( 1) (-1 )PzPp!
dzp zz + 2 = (zz + 2)P+1 .
For Izl < 2 we have
Therefore
-(-)
d P 1
dzp zz + 2 --
Comparing the obtained results for n = p, we have
(zz
1
+ 2)P+1 = zpp! E
(-l)P 00 (k + p) (k + p - 1) ... (k + 1) . zk
2k +p+1 zk+p ,
for Izl < 2.
We have for Izl > 2
Using this result we obtain
-1 k(k-1) ... (k-p+1)zk- P

= - E
k=-oo
2k-1zk
Finally for n = p we have
1 _ (-1)1'+1 -EI (k+p)(k+p-1)· .. (k+1)zk

(zz + 2)1'+1 - Zpp! k=-oo 2k+p+1 zk+p ,
for Izl > 2.

Find the Laurent expansion at the point w (w ~ 2z).
Example 7.17 Prove that the principal value 0/ log _z- satisfies the Laurent the-
z-l
orem in the region Izl > 1 (prove that log _z_ is analytic in this region and on any
z-l
path which lies in this region we stay always on the same branch) and expand it in
Laurent series in w = o.
Solution. The function log _z_ is regular in the region Izl > 1, since it can
z-l
be expand in power series at an arbitrary point w (Iwl > 1) :
z
log-- log z - log( z - 1)
z-l
z-w z-w
logw(l + - w- ) -log(w -1)(1 + - -)
w-1
w
Iog--+ ~ (-l (z-w)n
L.,;-- - -
t ~ (_1)n(Z-w)n
- L.,; - - - -
w - 1 n=1 n w n=1 n w- 1
w 00 (_1)n 1 1
l o g - + " - ( -R- )(z-wt.
w - 1 n=1
L.,; n w (w - l)n
The branching points of the function

z
log - - = log z -log(z - 1)
z-l
are z = 0 and z = 1. Since they are outside of the considered region Izl > 1, we stay
always on the principal value going through any path in this region.
The Laurent expansion is obtained by
z z-l
log-- -log--
z-l z
-log (1 - ~)
-L (l)n
00
- -n1
n=1 Z
-1 n
L~
n=-oo n
for Izl > 1.

Example 7.18 Prave that on both circles which are boundaries ofthe annulus where
the series
n=-oo
converges is surely a singular point of the function which is represented by this series.
Solution. We have for r < Iz - ul < Rj

-1
L L +L
00 00
un(z - ut = un(z - ut un(z - ut

n=-oo n=-oo n=O
00 1 00
= ~U-n(z_u)n + ~ un(z-ut
=L
00
L
00
u_nw n + un(z - u)n

n=1 n=O
1
where we have put w = - - .
z-u
The desired statement now follows from the property of the power series that on
the boundary circle of the convergence there is always a singular point.
Example 7.19 Find by the definition the coefficients in the Laurent series of the
function
f(z) = cos z
(z - U)2
the point u.
Solution. Using the formula
Un -
- -
1
271"Z
i L(W-U)n+1
f(w) d w, n = 0,±1,±2, ...
on the path L : z(t) = U + reh, 0 ~ t ~ 271", we obtain the Laurent series
cos Z _
00
'"
L...J (_l)n+I z -
()2n
U rlor Iz - U I > O.
(z-u)2 - n=-I (2n+2)!'
Exercise 7.20 Find the Laurent series of the function ~z e I/z2 in the region Izl > O.
Answer.
~
z
eI/z2 =~
Z n=O
f: (~)n n!1 f:Z2 = n=O z2n+I
1
n!'
for Izl > O.
Example 7.21 Let
00
for a E ~. Find the Laurent series ~ UkZ k of the function f using the following
k=-oo
steps.
a) Find the annulus of convergence of this series;
b) Prove that Uk = U-k, k ~ 1;
c) Prove that Uk = -1
271"
10
2
1<
eacostcos ktdt, k ~ 0;
d) Prove that
(k + 2p)!
forn=k+2p, pENU{O}
p!(k+p)!
otherwise
e) Expand the coefficient Uk in apower series with respect to a.
Solution. a) The desired annulus is the whole complex plane without zero and
z = 00;
1
=-
00
b) Putting z in ~ Ukzk, we obtain

w k=-oo
1 a 1
+ - )) .
00
~ Uk k = exp ( - ( w
k=-oo W 2 w
Comparing these two expansions (using the uniqueness of the Laurent expansion)
we obtain
E E
00 00
'Uk zk = 'U-k w k ,
k=-oo k=-oo
'Uk = 'U_k, for k 2: 1.

c) We have by the definition
'Uk = 2~t k (w ~(:;k+1 dw, k = 0, ±1, ±2, ....
Let 'U = O. We take for the path L the unit circle z( t) = et ., 0 ~ t ~ 271". Then
= -.!...
271" Jo
r'" eacost cos kt dt - t r'"
Jo
eacost sin kt dt.
By b) we have 'Uk = 'U_k and hence
(z2+1)n
d) By the definition 21 { zn+k+1 dz is the coefficient 'Uk of the Laurent
71" J1z1=1
series for the function
The expansion
gives Uk i= 0 only as a coefficient by zk for n = k + 2p and i = k + p. Then

U = (k+2 P) = (k+2p)! >0
k k +p p!(k + p)!' p- .
e) We have
00 an
e acost = ~ nt
L..J '" cos .
n=O n.
Therefore
-
1
271"
1
0
2 ".
eacost cos kt dt = -
1
271"
1
0
2"
L ,ann.
00
n=O
cosn t cos kt dt.
We can exchange the order of integration and surn since the series converges
uniforrnly in every closed region ( for Izl > 0) by the estirnation
I ,ann. cosnt cos kt ::::; ,anI (n E N).

n.
We have
Uk = -1 L ,ann.
00 12" cosnt coskt dt.
271" n=O 0
By d) we have
(k+2p)!
for n = k + 2p, p ~ 0
(Z2 + l)n dz ={ p!(k + p)!'
zn+k+l
o otherwise
= ~ [2" 2n COSn t( COS kt - zsin kt) dt.

271" Jo
Cornparing the obtained results we have
-
1
271"
10
2" n
cos t cos kt dt n
+ 2p)!
= 2 (kp.'(k + p )'.
for n - k = 2p, p ~ 0, and for other cases the integral is zero. Therefore
an n!
E n! 2
00
Uk= n (~)! (~)!'

where n runs through natural nurnbers n for which n - k = 2p, pE NU {O}.
Example 7.22 Find the Laurent se ries for the function
z2 - 2z +5
(z - 2)(Z2 + 1)
in the annulus 1 < Izl < 2. Then find the expansion at z= 2.
Answer. The desired Laurent series on annulus 1 < Izl < 2 is

00 1 00 zn
2 I:(-lt--I:-.
n==l n 1 z2n n==O 2 +
The expansion at Z = 2, or more precise for 0 < Iz - 21 < v's is

1 ~ ( )n (2 + zt+ 1 - (2 - i)n+1 ( 2)n
--+z L.J -1 z- .
z- 2 n==O 5n +1
Exercise 7.23 Prove the Mittag-Leffier theorem.
Hints. Suppose that Zn i 0 for all n. Expand Pn ( z ~ zJ in apower series of z / Zn

at the origin.
Exercise 7.24 Prove that for given entire functions fand 9 without common zeroes
there exist entire functions hand e such that
hf + eg = 1.
Hint. Use the Mittag-Leffler theorem to obtain a meromorphic function F whose

part with negative powers (principal part) occur only at zeroes of g, and the principal
part of F at a zero Zn of 9 is the same with the principal part of 1/ f 9 at Zn' Take
h = Fg.
Example 7.25 Find the Laurent series for the function
1
f(z) = (Z2 + 1)2
in the neighborhood of the points z = z and z = 00.
Answer. The expansion in 0 < Iz - zl < 2 is
The expansion for Izl > 1 is

00 n
I: ( - 1 t - ·
n==l 2n + 2
Example 7.26 Examine which o/the/ollowing multi-valued/unctions have branches

which can be expanded in Laurent se ries (especially in power series) in the neigh-
borhood 0/ the given point
a) yZ, z = 0; b) vz(z - 1), z = 00;
c) f}(z - 1)(z - 2)(z - 3), z = 00; d) VI + yZ, z = 1.
Answers.
a) Not possible;
b) Both branches can expand;
c) All three branches can expandj
d) Two branches can expand, and they are given by the condition
Chapter 8
Residues
8.1 Residue Theorem
8.1.1 Preliminaries
Definition 8.1 The residue oJ J at Zo given by the Laurent expansion

00
J(z) = L an(z - zot

n=-oo
is the coefficient a_l> denoted by Res(f(z))z=zo'
If J has a simple pole at Zo (pole of first order),

g(z)
J(z) = h(z)'
where 9 and h are analytic at Zo and h has a simple zero at Zo, then
. g(zo)
a_l = Z--+ZQ
hm(z-zo)J(z) = -, h
Zo
()'
If J has a pole of order k at zo, then

1 dk - 1
a_l = (k -1)! dZ k - 1 ((z - zo)k J(z))z=zo·
Theorem 8.2 (General Residue Theorem) Let J be analytic in a region D ex-

cept Jor isolated singularities at Zl> Z2, ... , ZS' IJ C is a closed path not intersecting
any oJ the singularities, then
10 J(z) dz = ; 10 z :ZZn Res(f(z))Z=Zn'

where C is a closed path arround Zl, Z2, . .. ,Z•.
191

192 CHAPTER 8. RESIDUES
A closed path C is called regular if C is a simple closed path with
1 dz
- - = 0 or 1 for a
ez-a
~ C.
1 dz
Then we call
{ a jI- ---1}
27rt e z - a
the inside of C.
Theorem 8.3 (Residue Theorem) Let J be analytic in a region D except Jor

isolated singularities at Zl, ZZ, • .. ,ZS' IJ C is a closed regular path not intersecting
any oJ the singularities, then
t 1e ResU(z))Z:=Zn,
je J(z) dz = 27n n:=l
where C is a closed path arround ZI, Zz, ... ,Zs'
Theorem 8.4 Let C is a regular closed path. IJ J is meromorphic inside and on C

and contains no zeros or poles on C, then
1 ( J'(z)
27rt Je J(z) dz = ZU) - PU),
where ZU) and PU) are the number oJ zeros (counted with their multiplicity) and
poles oJ J inside C, respectively.
Theorem 8.5 (Argument Principle) Let C is a regular closed path. IJ J is an-

alytic inside and on C and contains no zeroes on C, then
_1
27rt
1J'(z) dz
e J(z)
= ZU),
where ZU) is the number oJ zeroes of J inside C.
Theorem 8.6 (Rauche) Let J and g be analytic inside and on a regular closed
path C and jJ(z)j > jg(z)j for all z E C. Then
ZU + g) = ZU) inside C.
Theorem 8.7 (Generalized Cauchy Integral Formula) Let J be analytic in a

simply connected domain A and let C be a regular closed path contained in A. Then
for every Zo inside C
f (n)( Zo ) -
-
~
27rt
1z-
e
(
J(z)) +1 d Z
Zo n
r
lOr n -- 0,1,2, ....
8.1. RESIDUE THEOREM 193

Exercise 8.1 Find the residues of the following functions at the isolated singular
points in the extended complex plain (with (0) :
1 Z2n Z2 + z -1.
a) -3--5; b) nE N;
z - z (1 + z)n' c) Z2(Z _ 1) ,
Answerf' 1 1
a) Res(--)
Z3 - Z5 z=±1
= --,
2
Res (Z3-- -z5) %=0 = 1, Res(+s)
z - z z=oo
=0.
b) Re ( Z2n) (lt+1 (2n)!
s (l+z)n z=-1 = - (n-l)!(n+l)!'
R ( z2n) (lt (2n)!

es (1 + z)n z=oo = - (n - 1)!(n + I)!"
e s ( Z2 + z - 1)
) Re _ ,0 eRs (Z2 + z - 1 )
- _ ,1 eRs (Z2 + z - 1)
- - -1
- .
Z2(Z - 1) %=0 z2(Z -1) z=1 Z2(Z - 1) z=oo
~ 1 ~ 1
d) Res( z 2( z 2 + 9») z=o = -,
9 Res( z 2( z 2 + 9») z=3. = --5
4 (sin3 - z cos3),
~ 1 ~ 1
Res( z 2( z 2 + 9») z=-3. = --(sin3
54 + zcos3), Res( z 2( z 2 + 9») z=oo = 27(sin3 - 3).
Example 8.2 Find Res (~{:j) z=w if:

a) w is a zero of n-th order of the function f;
b) w is a pole of n-th order of the function f.
Solution.
J'(z) n g'(z). . .
a) We have by -f() = - -
z z-w
+- ()'
9 z
where 9 IS a regular functlon m the
neighborhood of wand g(w) -=F 0, that Res (~{:n z=w = n.

J'(z) -n
n
g'(z)
b) We have by f( ) = - -
z z-w
+- ()'
9 z
where 9 is a regular function in the
neighborhood of wand g( w) -=F 0, that Res (j~ z=w = -no

1"
o
tan(t + az) dt = { 'In for a > 0
-:rrz for a < 0,
for a E lR.
Solution. It is easy to prove that
1" o
tan(t + at) dt = -1
2
1 0
2
" tan(t + at) dt.
Taking z = eh, 0 ::; t ::; 2:rr, we have
. 1 1 1 1 dz
sm t = - (z - -), cos t = - (z + -), dt = - ,
2t z 2 z zz
1"tan(t+at)dt==--2 1
and we obtain
1 z2 - e2a dz
o JzJ=l Z
2
+ e2a · -Z ·
We have for a > 0 inside of the path /z/ = 1 only one pole at the point z == 0 ;
Z2 _ e2a
Res (
(Z2 + e2a )z ) z=o == -l.
Therefore
tan(t + az) dt == -2"1 2:rrz( -1) == 1rZ for a > O.
We have for a < 0 inside of the path /zl == 1 three poles of first order at the points
Z2 _ e 2a )
(
Res e2a )z == l.
(Z2
+ z=:b eG
Therefore
Jo
r tan(t + az) dt == _!2 2:rrz( -1 + 1 + 1) == -1rZ for a < O.
Exercise 8.4 Find the integrals
a) 1 21r
(1 + 2 cos tt cos nt dt, n E Nj b) 1 2"
o
dt
1 - 2a cos t + a 2
, 0< a < 1j
c) r"
Jo a + bcost
dt for a > b.
Answers.
27r 27r
b) 1 _ a2; c) ~.
a)
(n)k -__1 [(1 + z)n dz

27rz Je zk+l '
where C is a simple path surrounding the originj

b)
Solution. a) (~) is the binomial coefficient of zk in (1 + z)n and by the theorem on

(n)k -__
residue
1 r (1 +z)n dz
2n Je zk+! .
b) By a) we have
(2n)
n
= _1 r (1 + z)2n dz.
2n Je zn+!
Choosing for C the unit circle we obtain
Example 8.6 Find L00 (2n) 1

n --;.
n=O 7
Solution. By Example 8.5 we have
(2n)
n
= _1 r (1 + z)2n dz,
27rz Je zn+l
where C is an arbitrary contour surrounding the origin. Therefore
00 (2n) 1 1 00 r (1 + z)2n dz
~ n 7n = 2n ~Jo (7z)n -;.
Taking for C the unit circle surrounding origin we have
1(1+Z)21
7z
<~
- 7' z E
C
.
Then the convergence of the considered series is uniform and therefore
f (2n)
n=O n 7
~ n
= .!...- f
27rZ Jlzl=l 5z - 1 -
dz Z2
= 7Res ( 1
5z - 1 - Z2
)
z=(5-01)/2·
Find this residium.
Exercise 8.7 Prave that
a )1 +ax
00
-00
cos
-- x = -
1 x4 V2 e
d 1r -aV2/2 (aV2 aV2)
cos -2- + sin -2-
E
l'
or a> 0;
21r cos(2p + 1)~

b) VP 10 00 Xp-l
dx = - . for 0< p < 2;
o x2 +X +1 v'3 sm p 1r
sin 2 x dx = ~.
x2 2
Hints. a) Use the function e'U' 4 on the path in Figure 8.1;

l+z
~---.----~--------~
-R ~ 0 R ~
Figure 8.1
p-l
b) Use the function 2 Z on the path in Figure 8.2 ;
Z +z + 1
-R R x
Figure 8.2
x
Figure 8.3
e2z1 - 1
c) Use the function 2 on the path in Figure 8.3.
z
Example 8.8 Taking the integral
f In 2 Z d
lL (z2 + a 2)2 z fo1' a > 0,
and the path L f1'om Figu1'e 8.4 find which real integrals can be calculated fo1' R - t 00
and l' - t O.
-R R x
Figure 8.4
Solution. We obtain by Theorem 8.3 on residues
1, In2 z
2)2 dz
lR In 2 x
+Z
1 211" (In Re tl )2 tI
L
(2
Z +a
= T x +a 2)2 dx
(2 0
(R2 e 2t'+a 2)2 Re dt
We estimate the second integral:
<
<
<
as R ~ 00. We obtain in an analogous way for the fourth integral
I1[21f
0
(In r + (t + 211")z)2 reh dtl
r 2e2h + a2)2
< {21f (llnr+(t+211")zl)2 rdt
10 la 2 + re 2t.12
< [21f (Iln rl + It + 211"1)2
Jo (a 2 - r 2 )2
~ 0
as r ~ O. The desired residues are
2
=lim.!!...- (
~)
z->.a dz z + za
= z
-3
( 2ln a - In 2 a + -11"2 + n - 1I"Z In a) ,
4a 4
R 2 = Res ((Z2 ~2a2)2) z=-.a = - 4: 3 (2 lna -ln 2 a + a: + 3n - 311"z lna) .

2
Therefore we have for R --+ 00 and r --+ 0 :
-2z V P lo
OO
(2
Zn x
x +a
2)2 dx + 211" V P
10
00
(2
dx
x +a
2)2 = z(R 1 + R 2 ).
Putting the values of residues R 1 and R 2 into the last equality and comparing the
real and imaginary parts we obtain
1 00
o
(2
In x
x +a
2)2 dx = --3
11"
4a
(1 -lna);
since the integrals obtained converge absolutely.

r
Jo
x sin x d tor a >
1 + a2 - 2a eos x x
o.
Hint. Take the integral of the function t(z) = z

a - e- az
on the reet angle:
-7r :::; Rez :::; 7r, 0:::; Imz:::; h,
and let h - t 00, Figure 8.5.
hi
f
-11 o
Figure 8.5
7r
Answer. - In(l
a
+ a).
Exercise 8.10 Calculate the value ot the integral 1 00
-00 (ex
eax
+ 1)( e + 2)
X dx, tor 0 <
a < 2.
Hint. Take the integral of the function

eaz
t(z) = (ez + l)(ez + 2)
on the rectangle with the vertiees - R, R, R + 27rz, - R + 27rZ.
7r(1 - 2"-1)
Answer . --'-----'-
sm a 7r
Example 8.11 Find the integral _1_

27rz Je z
r
sin ~ dz, where C is the circle Izl = r.
Solution. The point z = 0 is an essential singularity of the function sin~. The

z
corresponding Laurent series is
00
(_1)n
~ (2n + 1)! z2 n +t'
Hence U-l = 1. Therefore
_1_ [ sin ~ dz = Res (sin~) = 1.

211"z Je z z z=o
Exercise 8.12 Find the value of the integral -2 1 [ ~z , where a = e In a Z Z
1I"Z Je a sm 1I"Z Z
for a > 0, and C is the straight line x = d (0 < d < 1), oriented from below to above.
Hint. Take first the integral on the path from Figure 8.6 and let b -+ 00.
,
.0IIII ~y
d+ib ~
~
---~
0 d 1 1+<1 X
I
d-ib
Figure 8.6
1
Answer . ----:-:------:-
11"(1 + a)'
Exercise 8.13 Find the real integral
for -1< p < 2.

Hints. Take the integral of the function
on the path C given on the Figure 8.7 and then take R --+ 00.
Answer. 1rp(l ~ p)
23 - p sm 1rp
~y
-_R+---HlIJi(4~~R ---~
Figure 8.7
1 1 dx
n~'
-\}l-x n
n=2,3, ...
Hints. Take the integral of the function
J(z) = vr=zn
1
on the path C given on the Figure 8.8 which consists of the cuttings on radius
2:'
through the points 1, W, w 2 , ..• , w n , where w = e and the circle Izl = R > 1.
-R
Figure 8.8
Answer. ~.
nstn-
n
Example 8.15 Let J be an analytic Junction in the region which contains the disc
Izl S 1 Prove that
Jor lai< 1
Jor lai> 1,
where L : z(t) = eh, 0 S t S 211".
Solution. By Example 5.13 b) we have on the path L : Izl = 1
-dz
/, J(z)dz = - /, J(Z)2'
L L z
Therefore
1=_1_ ( J(z) dz = _1_ I J(Z)Z2 . dz = __1_ ( J(z) . Z2 dz.

2n lL z - a 2n lL z - a Z2 211"zlL z - a
Since z2 = e- 2ts = _1_

e 2t•
= ..!..,
z2
Ost 5 211", we have
__1 f j(z) dz
I
211"zJL z2(z - a)
= 1
__ f j(z) dz
211"zJL z(l - äz)
__1 f j(z) dz
21l"ZiLäz(k- z ) ,
where Z· z= 1 on the path L.
Since
Res (_ ~(z) ) = _j(~)j
az(-
a - z) z_"
-1. a
Res ( j(z) ) = ( j(z) ) = j(O)

äz(k - z) z=o z(l - äz) z=o '
we obtain
211"zj(0) for lai< 1,
1
j1(z) dz = {
äZ(a - z) 2.. (1(0) -1 m) for lai> 1.
Finally
for lai< 1,
for lai> 1.
Example 8.16 Find the number oj zeroes oj polynomials in the disc Izl < 1 :
a) Z4 - 5z + 1 = 0; b) Z9 - 2z6 + Z2 - 8z - 2 = Oj
c) z6 - 6z + 10 = 0; d) z7 - 5z 4 + z2 - 2 = O.
Solution. We shall use Rouche's theorem. We put in Izl < 1 that j(z) = -5z
and g(z) = Z4 + 1. We have on the circle Izl = 1 polynomial Ij(z)1 = 15z1 = 5 and
Ig{z)1 = Iz 4 + 11< Iz4 1+ 1= 2. Since the polynomial (-5z) has one zero in Izl < 1,
then also the (Z4 - 5z + 1) has one zero in the disc Izl < 1.
b) One zero.
c) There are no zeros in Izl < 1. Put j(z) = 10, and g(z) = Z6 - 6z.
d) Four zeros.
Example 8.17 Prove that the equation zn = ez - k for k > 1, has exactly n zeros in
the disc Izl < 1.
Solution. It follows by Rouche theorem putting fez) = _zn and g(z) = eZ-k,
where on the unit circle Izl = 1 we have If(z)1 > Ig(z)1, i.e., 1 > ex - k for k > 1 and
-1 ~ x ~ 1. Since f has n zeros in Izl < 1, the considered equation has also n zeros
in Izl < 1.
Example 8.18 Prove that the equation z = h(z), where h is an analytic function
in the disc Izl ~ 1 such that Ih(z)1 < 1, has in the disc Izl < 1 exactly one zero.
Solution. It follows by Rouche theorem putting fez) = z and g(z) = h(z),

since If(z)1 > Ig(z)1 on the circle Izl = 1, and f has one zero in Izl < 1.
Exercise 8.19 How many zeroes has the equation

eZ - 4z n +1= 0 (n E N)
in the disc Izl < 1?
Answer. It has n zeroes.
Exercise 8.20 How many zeroes in the disc Izl < R has the equation
e Z = a zn, n E N,
eR
for lai> Rn?
Answer. It has n zeros.
Example 8.21 Prove that for enough small p > 0, and enough big n all zeros of
the function
111
fn(z) = 1 + - + -2!z2 + '" + -n!zn
Z
are in the disc Izl < p.
Solution. Since the sequence of functions {In} converges to the function e~ every-
where except of the point z = 0 we can find for every disc K, with the center in
z =F 0 and which does not contain z = 0 for every c > 0 a natural number no such
that
for all points from K and n > no. The desired conclusion follows by Rouche theorem
taking
c< rnin le I
zEC
1/ z
'
where C is a path in K.
206 CHAPTER 8. RESID UES

Example 8.22 Let A be an open subset of C and f a continuous function on A.
Let 1+ be the region Im z > 0, and 1- the region Im z < O. Let f be analytic on
A n 1+ and A n 1- .
1) Prove that for every real number a E A there is a disc D( a, r) with the center
at a and radius r, wh ich lies in A.
2) Let C+ be the path
z(t) = a + (2t + l)r, -1:::; t:::; 0; z(t) = a + r et ... , 0:::; t:::; 1.
Let C- be the path:
z(t)=a+re t1rl , -1:::;t:::;O; z(t)=a+(1-2t)r, O:::;t:::;1.
Find the values of integrals
1 f f(u) du 1 f f(u) du.

2n lc+ u - z ' 2n lc- u - z
a) For z inside oJ the path C+.

b) For z inside oJ the path C-.
3) Using 2) prove that the Junction f is analytic on the whole set A.
L akz k. Denote
00
4) Let A : Izl < r, and J(z) =

k=O
L lakl
00
MI(r,J) = Tk
, M(r,J) = sup IJ(z)l·
k=O Izl=r
Prove that Jor 0 < r < r +8 < R :

r+8
M(r, J) :::; MI (r, J) :::; -8- M(r + 8, J).
5) Using 4J prove
Solution. 1) Let a E A and Im a = o. Since A is an open subset of C it is a

neighborhood of each its points and so also of the point a. Therefore there exists a
disc D(a, r) which lies in A, D(a,r) C A.
A..y
I _ _ _.... ~
x
--~~--~----~.-.
Q-r -~ o
o+r o
Figure 8.9
2) The paths C+ and C- given by
C+:z(t)=a+(2t+1)r, -l$t$O; z(t)=a+re t 1l"', O$t$l.
C- : z(t) = a + (1 - 2t)r, 1 $ t $ 1; z(t) = a + ret 1l"', -1 $ t $ 1

are shown on Figures 8.9 and 8.10, respectively.
""y,
Q-r Q ..--
Figure 8.10
Consider first the integral on the path C+. The only singular point of the function
J(u) is the point u = z, which is a pole of first order. We have
u-z
_1 f J(u) du = { Res
Jc+ u - z
(!~~) _ for z inside of the path C+
21rZ
o u-z
for z inside of the path C-,
since for z outside of path C+ the function J(u) is an analytic function inside of
u-z
the path C+.
Since
Res (J(u)) = lim(u _ z) J(u) = J(z)
u- z u-+z U - Z
u=z
(the function J is continuous everywhere on C+), the value of the integral is

for z inside of the path C+
_1 f J(u) du = { J(z)
21rZ Jc+ u - z 0 for z inside of the path C-.
We obtain in an analogous way for the integral on the path C-
for z inside of the path C-

_1 f J(u) du = { J(z)
2n Jc- u - z 0 for z inside of the path C+.
3) We shall prove that the function J is analytic on the whole set A. By the
supposition it is analytic on An 1+ and An /-. We have to prove that it is analytic
on the set {z I Imz = 0, z E C} n A, on the path of the x-axe which lies in A.
By 1) for each real point a E A always there exists a disc D( a, r) with the center
in a which lies in A. We shall prove the analicity of the function J on the part of
x-axis which lies in D (a, r). Since J is continuous on C+ and C- we have that
_1 f J(u)
2n Jc- u - z
define analytic functions for z E C+ and z E C- , they are analytic functions on the
part of x-axis which is in D(a,r). Then we have
_1
2n
(1 J(u) du +
c+ u - z
1 J(u) dU)
c- u - z
= _1 f
2n JK
f(u) du,
u- Z
where K is the boundary of the disc D( a, r).

By 2) we have
-1
27rl
1 -f(u)- du
K u - Z
= f(z) for z E D(a,r), (8.1 )
without x-axis. The integral on the left side of (8.1) defines an analytic function for
all z E D(a, r), and it is an analytic extension of f on real axis in D(a, r). By the
uniqueness of the analytic continuation it follows that (8.1) holds on the part of real
axis in D(a,r). Since this holds for every point a E A from real axis we obtain that
f is analytic function on the whole set A.
4) We shall prove first that M(r,l):S; M 1 (r,l). Namely, since
If(z)1 = I~ akzkl :s; ~ lakllzl k = ~ lakl rk

for every z for which 0 < Izl = r < R, we have
00
M(r,J) = sup If(z)l:S; L lakl rk = M1 (r, f).

Izl=r k=O
We shall prove now the second part of the inequality,

r+8
M1 (r,J) :s; -8- M(r + 8,1).
We have 00
M 1 (r, I) = L lakl rk
k=O
= ~ 1_1
L.J 2
k=O 7rl
1
Izl=r
f(z) d I k
zk+l Z r
= f
k=O
1_1 r
27rl 1Izl=r+5 zk+l
f(z) dzl r k
'
by the equivalence of integrals on paths Izl = rand Izl = r + 8 (0 < r < r + 8< R).
Since
_1 r f(z) dzl < 1 r If(z)1 d

1
27rl 1lz l=r zk+l - 2n 1lzl=r+o Izlk+1 z
sup If(z)1
Izl=r+o r27r r +8
< 2 10 (r + 8)k+1 dt
sup If(z)1
Izl=r+5
(r + 8)k+l '
we obtain
sup If(z)1
E
00
M(r,f) < Izl=r+6 rk

k=O
(r + 6)k
00 r k
= Izr~!~)f(z)1 ~ (r + 6)
1
= sup
Izl=r+6
If(z)1 1
- r+6
r
r+6
= -1:- sup If(z)l,
u Izl=r+6
since ~ < 1, which implies the inequality and the convergence of series. So we
r+u
have proved that
r+6
M(r,f)::; M 1 (r,f)::; -6- M(r + 8,f) for 0< r < r + 6< R.
5) Taking
M 1 (r,j n ) =E 1-1
00
k=O 21ft
1 r(z) dz 1r k ,
~
Izl=r z
we can easily prove in an analogous way as in 4) that
Putting 6 = .!.. and taking the n-th root we obtain

n
since
1 -< Y'1 + nr -< Y'n 2 + n 2 = Y'2n 2 -+ 1 as n -+ 00.
Therefore by the inequality (8.2) using the comparison criterion for sequences we
obtain
lim r!M1 (r,r) = M(r,f).
n-+oo
Example 8.23 Let

1 1
F(z) = -.- - -.
smz z
a) Find the singular points ofthe function F(z). Is the function

F(z) for z '" 0
f(z) = { 0
for z = O.
analytic at the point z = O?
b)Find the residues of the Junction J in all poles.
c) Show
1 /2 2
/sin z = cosh 2y - cos 2x .
d) Show that there exists a natural number no such that the Junction J is bounded
independently of n ~ no on the boundary of a the square K n given by
z = ±(n + 1/2)7r + ~t, -(n + 1/2)7r + 1/2)7r;
~ t ~ (n
z= t ± t(n + 1/2)7r, -(n + 1/2)7r ~ t ~ (n + 1/2)7r.
e) For an arbitrary but fixed n E N find the integral
In = _1 { f(u) du.
27ft lK n u(u - z)
f) Prove that liIIln-+oo In = 0 and the equality
F(z) = 2 E(_l)n
00
z2 _ (n7r)2'
g) Find Jor the preceding series:

(i) the domain oJ the convergence,.
(ii) the domain oJ the uniform convergence;
(iii) the domain oJ the absolute convergence.
Solution. a) The singular points of the function J are z = k7r for k = ±1, ±2, ...
We shall examine the point z = O. We have
. z - sin z
11m
lim J(z) .
z-+o z-+O Z Sin Z
1 - cosz
= lim -c.:------
z-+O sm z + z cos z
smz
lim - - - - - - - -
z-+O cos Z + cos Z - Z cos z
= 0,
limz-+oj(z) = j(O). Hence the function j is continuous at the point z = O. The

function F has the same singular points as the function j, and at z = 0 it has
removable singularity.
Now we shall show that the function j is analytic at the point z = O.

r j{O + ~z) - j(O)
z~ ~z
r ßz - sin~z
z~ ~z2sin~z
r 1- cos~z
= z~ 2ßz sin ~z + ~Z2 cos ~z
r cos~z
= z~ 2 sin ~z + 2ßz cos ~z + 2~z cos ~z - ß Z2 sin ~z
r cos~z
z~ 2 cos ~z + 4 cos ~z - -4~z sin ~z - 2~z sin ~z - ~z2 cos ßz
1
- < 00.
6
Since the preceding limit exists we conclude that the function j has a derivative at
z = 0, it is analytic at the point z = O.
b) The function j has at z = 0, k = ±1, ±2, ... poles offirst order. We calculate
the residues of the function f at z = k1r
Res{f (z)} z=k1r = (--:._Z_-_sI_n_Z_) k1r _ (_l)k

Sin z + z cos Z z=k", k1r cos k1r - .
c) Using the equality sin z = elZ 2ze -1% we obtain
2 2z
4
= (e'X-Y _ e-,x+Y)(e'x+y - e-'x- y)
2
2 2
2
= cosh 2y - cos 2x .
d) The path K n is given in Figure 8.11.
(n#-jJ1i!
Kn
~
-"''''JIN" 0 {n#-IPi X
,
-(n-#T]1i"f.
Figure 8.11
We shall show that there exists a natural number no such that the function f is
bounded independently of n ;::: no on the boundary of a square K n • We have on the
sides of K n
z = ±(n + 1/2)7r + tt, -(n + 1/2)7r $ t $ (n + 1/2)7r;
the following inequalities
If(z)1 = I±(n + ~)7r + tt - sin(±(n + ~)7r + tt) I
(±(n + ~)7r + tt) sin(±(n + ~)7r + tt)
< I ± (n + ~)7r + ttl + I sin(±(n + ~)7r + ti)1
7r(n + ~)I sin(±(n + ~)7r + tt)1
< (n + ~)7r + Itl + cosh Itl
(n + ~)7r1 sin(±(n + ~)7r + tt)1
< 2(n + ~)7r + cosh Itl 2
(n + ~)7r cosh2t - cos2(±(n + ~)7r)
< 2(n + ~)7r + cosh(n + ~)7r 2
(n+~)7r 1+1
< M
for a natural number no, fixed M > 0 and n ~ no.

In a quite analogous way we can prove the boundedness of f on the paths
e) The function f for a fixed n is analytic on K n • Its singular points (poles of

the first order) inside the square are at the points u = k7r, k = ±1, ±2, ... , ±n. The
function under the integral in In additionally has a pole of first order at the point
u = z when the point z is inside the square K n (taking n enough big we can always
manage this), and at the point u = 0 we have a removable singular point. Hence
I - 1 r f(u) d - t
R ( f(U))
n - 271"~ lKn u(u - z) u - k=-n es u(u - z) u=k1r +
R
es
( f(u) )
u(u - z) u=z·
Since
R ( f(u) ) R ( u - sinu )
es u(u - z) u=k1r es u2(u - z) sin u z=k1r
h
(h)2(h - z) cos h
( _l)k
h(h-z)
and
R ( f(U)) f(z)
es u(u - z) u=z z
we finally obtain (the term in the sum with the index 0 is equal 0)
(8.3)
f) We shall show that limn _ ao In = O. We have by d) If(u)1 < M for u E K n and

n ~no. Hence
1
IInl < 271" 1 lu(uIf(u)1- z)11 dul
Kn
< M 1 lulluIdul- zl
i lul(lul-lzl)·
271" Kn
M
< 271" du
Kn
Since we have minuEK n lul = (n + D1I" and lul- Izl 2: (n + !)11" -Izl we obtain
< M [ Idul
211" JK n (n + ~)1I"((n + ~11" -Izl)
= M [ Idul.
+ !)1I"((n + })11" -Izl) lK
211"(n n
Using that one side of the square K n is equal 2(n + ~)11" we obtain
111 M8(n+ t)1I" _ 4M

n < 211"(n + ~)1I"((n + ~)71" - jzl) - (n + !)11" - JzJ·
Taking n ---t 00 we obtain that the last member in the preceding inequality tends to
zero, what implies limn->oo In = 0.
To prove the desired equality we shall take n ---t 00 in (8.3) from e)
After some transformations we obtain

-1 (_l)k (-l)k
+ .t; h(z -
00
z k~OO h(z - h) h)
(_l)k (_l)k
-E +E .
00 00
k=1 h(z + h) k==1 h(z - h)

Finally we have
E----;;;-.
00 -z+h+z-h
(-l)k
(z + h)(z - h)
00 (_1)k
2E
k==1 Z2_(k 71" F·
This implies the desired equality.
g) The series
E
00 (-1 )k z
z 2_(h)2
converges absolutely in the disc D(O, 71"), and uniformly in every closed region in
D(O, 11"). The absolute convergence of the series follows by the inequality
z I R
IZ2(-1)k
- (h)2 :::; (hF - R2
for jzl :::; Rand k 2: ~, since J( n1l")2 - Z2J 2: j( n1l"? - JzJ 2 1holds.
216 CHAPTER 8. RES1DUES
z+u
Example 8.24 I) Let W = J(z) = k - - Jor k 1= o.
z+v
+u
Z·
1) We denote by Wi = k -'--, i = 1, 2, 3. Prave that
Zi+ W
---=------
Prave that iJ there are given three different points Zl, Z2, and Z3 in the z-plane and
three different points Wb W2, W3 in the w-plane then there exist a unique bilinear
transformation J wh ich map Zi on Wi, i = 1,2,3, respectively. Find the bilinear
transformation J which maps the unit disc Izl < 1 on the half-plane Im W > o.
2) Find the image in w-plane by the Junction J{z) = 2 z +1 oJ the circle from
Z
z- plane, wh ich does not contain the point (0,0)7 For which points is this Junction
1""
a conformal mapping?
a-l
II) 1) For 0 < a < 1 find the value oJ V P -IX dx. We are taking Jor za-l
o -x
the bmnch which is real for real z.
2) 1f 0 < a < 1 and 0 < b < 1, then
1""
o
xa-l - X b- 1
- - - - dx = 7l'(cot a7l' - cot b7r)
1- x
Solution. I) 1) See Example 4.8. Since the bilinear transformation is uniquely

determined by the given images of three points we shall find the desired transfor-
mation by three points.
Taking Zl = O,Z2 -- 12' z3 = :4'
1
we h ave
u _ k 1 + 2u 1 + 4u
Wl = k ;-, W2 - 1 + 2v' W3 = k 1 + 4v '
respectively.
The condition Im Wi ~ 0, i = 1,2,3, implies
ku v-ku v
Im Wl = Ivl 2 2z ~ 0, Im (k u v) ~ 0, (8.4)
Im W2 -- 2ku(I +112v) - 2ku(I

+ 2vl 2z
+ 2v) 0 I (k) > 21 (k-)
~ , m u _ - m uv (8.5)
2
and
Imw3 =
4ku(1 + 4v) - 4ku(1 + 4v) ~ 0, (8.6)
11 + 4vl 2 2z
Im(k u) ~ -4Im(kuv).
Since (8.6) is a consequence of (8.4) and (8.5) we can omit it. Therefore the
desired bilinear transformation is
w_k
-
Z
z+v'
+u ki=0
,
where Im(kuv) ~ °and Im (ku) ~ -21m (kuv), see also Example 4.5,
-1
Figure 8.12
where the bilinear transformation w' = ~ - Z maps the unit disc Izi < 1 on the
~+ z
half-plane Re w' > 0, see Figures 8.12 and 8.13. If we rotate for the angle 7r /2, w =
e7r • j2 w, we obtain the desired transform (see Example 4.18 )
7r.j2 ~ - z
w=e --.
~+z
Figure 8.13
z+l
2) For fez) = 2 -2- we have
Therefore we obtain the desired image in the following way. The transformation
[~ ~] maps the circle azz + bz + cz + d = a which does not cross (0,0) from the
z-plane onto the circle dWIWI + CWI + bW1 + a = 0 in the w-plane.
Applying now the transformation [~ ~] on dWIWl +CWl +bWl +a = 0: we first

translate the circle by 2 and enlarge the radius by 2. Write down the equation of
the obtained circle in the w-plane. For every point z E ewe have J'(z) #- O. This is
not true only in the extended complex open at z = 00. The bilinear transformation
fez) = 2 z + Z is a conformal mapping at all points of the complex plane except at
z
z = 0 and z = 00.
za-l
II 1) The function - - has the branching points 0 and 00. Therefore the cutting
l-z
za-l
[0,00) divide the branches of - - . This function has a pole of first order at z = l.
l-z
a-l
Therefore we shall integrate the function fez) = _z__ on the path L which is
l-z
given at Figure 8.14.
-R
Figure 8.14
We have
a-1
f _z_dz = 0,
JL 1- z
since the function f is analytic in the region bounded by L.
We have (see Figure 8.14)
1 za-1
--dz 1 1- E xa-1
--dx+
1 za-1
--dz+
JR x a- 1
--dx
L 1- z 1- X kl 1 - Z 1+E 1 - x
1
r
+ a-1 dz
_z__ + J,1+E _x__
a-1 e2 (a-1)1I"' dx
k(R) 1- Z R 1- X
za-1
_z_dz + f _x__ e2 (a-1)1I"' dx +f
a-1 r a-1
+ f --dz.
lk 2 1- z lt+< 1 - x Jk3(r) 1-z
a-1
Since 0 < a < 1, we have lim z _z__ = O. Therefore the value of the fourth integral
z-+oo 1- z
a-1
is zero as R --+ 00. Since lim z _z__ = 0, the value of the last integral is also zero
z-+o 1- z .
as r --+ O. On the other side, since z = 1 is a pole of first order the values of second
and sixth integrals are
1 za-1
- - dz = -1l"Z Res (-lZ)
kl 1 - z
a-1
- Z z=1+.0
= 1l"Z,
{ za-1 dz = -'In Res ( za-1 ) = 'In e2"a.

1k2 1- z 1- Z z=1-.0
(since 1 + z = eO., and 1 - zO = e2"') as c --t O. Letting R --t 00, r --t 0 and c --t 0
we obtain
( 1- e2(a-1)1r,)VP {OO x - dx = -7l"z(l + e2,..a,).
a 1
10 1- x
1
Hence
00 xa-1 e2 ,..a. + 1
VP - - dx = 7l"Z 2 = 7l" cot a7l" for 0 < a < 1.
° 1- x e "a. - 1
2) By 1) we have for 0 < a < 1, 0< b< 1,

x q-
VP f'X> x q - 1 dx _ VP {OO X -
1 _ X b- 1 b 1
1- x dx 10 1 - x 10 I-x
= 7l"(cot a7l" - cot b7l").
Then V P 10
00 xa-1 -
----
I-x
X b- 1
dx
= {01 x a- 1 - X b- 1 ],C x a- 1 - X b- 1 {OO x a- 1 _ X b- 1

1n 1- x dx + 1 1_ x dx + 1c 1_ x dx.
Examine the integral also at the point o.

First and second integrals are not improper at x = 1, since at x = 1
. xa-1_xb-1 . (a-1)x a- 2 -(b-1)x b- 1
11m
",-+1
-----
1- X
= hm
",-+1 -1
= -a + b.
The last integral is absolutely convergent since 0 < a < 1 and 0 < b < l.
Then
1 00
X
a-1
x-
b-1
dx = V P
100 x a-l
- x
b-1
dx.
o I-x 0 I-x
Example 8.25 Suppose that J is an analytic Junction in the region D which con-
ta ins the point a. Let
F(z) = z - a - q J(z),
where q is a complex pammeter.
1) Let K C D be a circle with the center at the point a on wh ich the Junction J
is zero. Prove that the Junction F has one and only one zero z = w on the closed
disc (K) whose boundary is the circle K, iJ
. Iz - al
Iql < ~W J(z)1 .
8.2. COMPOS1TE EXAMPLES 221
2) Let G be an analytic function on the disc (K) together with the boundary.
Using the theorem on residues prove that
G(w) = _1 ( G(z) dz
F'(w) 2?ri lK F(z) ,
where w is the zero jrom 1).
. 1
3) 1f z E K, prove that the functlOn F(z) can be represented as a convergent
series with respect to q :
4) Using 3) and 2) prove

G(w)
F'(w) = G(a) +; 00 qn dn n
n! dan (G(a) j (a)).
5) Prove that ij G is oj the form

G(z) = H(z) (1 - q !,(z)),
where H is an analytic junction on (K), then
H(w) = H(a) + f
n=l
~~ ::n-~l (H'(a) j(n)(a)).
Solution. 1) We shall apply the Rouche theorem. Let
F(z) = z - a - q j(z) = cp(z) + '!fJ(z),

where cp(z) =z - a and '!fJ(z) = -q f(z). We shall prove that:
(i) cp(z) and '!fJ(z) are analytic functions for all z E (K)j
(ii) cp(z) f 0 for z E Kj
(iii) Icp(z)1 > 1'!fJ(z) I for z E I<.
(i) is obvious since cp is a polynomial of first order and so an analytic function
in the whole complex plane and the function f is analytic by the supposition in the
region D which contains (I(). Hence -qf is also analytic function in D.
(ii) follows because it can not be z = a on the path K, since K is the circle with
the center at a.
(iii) will be proved using the fact
. Iz - al
Iql < ~f{ If(z)l·
Iz -al
Namely, Iql < If(z)1 for z E K. Hence
11/J(z) I = Iqllf(z)1 < Iz - al = Icp(z)l·

The conditions (i) , (ii) and (iii) by Rouche theorem imply that the number of zeroes
of the function F = cp + 1/J and cp are equal. Since cp( z) = z - a is a polynomial of
first order it has only one zero z = a, we obtain that the function F has also one
and only one zero z = w on the closed disc (I<).
2) Since the function F has a zero of first order at z = w, the function ~ has a
pole of first order at z = wand also the function ~, since G is an analytic function
in the disc (I<).
We have by Residue Theorem
r G(z) (G(z))
JK F(z) dz = 2n Res F(z) z=w·
Since ~ has pole of first order at z = w
( G(Z)) G(w)
Res F(z) z=w = F'(w) ,
and so
r G(z) G(w)
JK F(z) dz = 27rZ F'(w)'
which gives us the desired equality.
3) Since F(z) = z - a - q fez), we have
1 1 1 1
F(z) = z-a-qf(z) z - a 1 _ qj(z) .
z-a
Since
qf(Z) / < 1,
/ z-a
for z E I< (this inequality is equivalent to Iql < 1~(z~I, which holds by the condi-
tion in 1)) we can represent the second factor in the last equality as an absolutely
convergent series
1 1 (qf(z))n
= -z - L ( )
00
F( z ) for z E I<,
a n=O Z - a n
8.2. OOMPOSITE EXAMPLES 223
which implies the desired equality.

4) By 2) we have
G(w) 1 [ G(z) d
F'(w) 27r~ JK F(z) Z.
Using 3) we have
G(w)
F'(w) =
1
27r~ JK
[
G(z) ::a
~ (q f(z))n d
(z _ a)n+l z.
By the uniform convergence of the series for z E I<, we can exchange the order of
the integration and sumo Therefore
G(w)
F'(w)
= _1
27r~ n=O
f JK[ G(z) (zqn- r(a)
a)n+l
dz
= ~ qn _1 [G(z)r(z) dz.
L..J
n=O 2n JK (z - a)n+l
Then by the Cauchy integral formula we have
Therefore
G(w) 00 qn n
F'(w) = G(a) + ~ n! (G(a) f (a)).
5) By G(z) = H(z) F'(z) and 4) we easily obtain the desired equality.
Example 8.26 Let f be a meromorphic function in the whole complex plane for
which there exists an increasing sequence { r n } which tends to +00 as n ~ 00 such
that
(8.7)
where M > 0 is a constant independent of n and (). We denote by u a point in the
complex plane which is not a pole of the function fand On is the circle Izl = r n ,
positively oriented.
I) 1. Prove that the integral
converges to zero as n - t 00.

I) 2. Prove that
f(u) - f(-u) = -2u L Res (

aEP Z
!(z)
- U
2) ,z=a
where P is the set of all poles of the function f.

II) 1. a) Prove that for the complex number a > 0 there exists A> 0 such that
z = x + zy and Iy I ~ a imply 11 + e2•z 1~ A.
II) 1. b) Prove that for the complex numbers z which satisfy the inequality
there exists a number B > 0 such that
II) 2. Prove that the function tan z satisfies (8.7) taking r n = mr (n E N).
II) 3. Using I) 2. and II) 2. prove that
00 -2u
tanu=L
n=l
2 ((2 n
U - + 1)")2·
2"
II) 4. Examine the uniform convergence of the senes from II) 3.
Solution. I) 1. For rn ~ lul we have
Letting n --+ +00 we obtain that the integral tends to zero.

I) 2. By the theorem of residues we have
f ~dz = 2n "Res ( f(z) )

Je n 2
Z2 - u L..J
aEGn
Z2 - u 2
z=a
where Cn is the closed region bounded by Cn. Letting n --+ 00, we obtain by I) 1.
(8.8)
. . J(z)
where S IS the set of all poles of the functlOn 2 2.
Z -u
For J(u) =I- 0 and u =I- 0, u is a pole of the first order for the function !(z) 2'
z -u
(JJ!L)
and therefore
R = J(u)
es z -u
2 2 2u
z=u
We have analogously for J( -u) =I- 0
Res C!~ ~2 ) z=-u = - J~~u) .

Putting this in (8.8) we obtain
J(u) - J( -u) = -2u

aEP
L Res ( !(z)
Z - U
2) . z=a
The previous equation is true also for u = 0, or J(u) = 0 or J( -u) = 0, what

can be easily checked.
11) 1. a) Let z = x + zy. Then le2, z l = e- 2y . For y ~ a we have
11 + e2, z l ~ 11 _le 21Z 11 = 1- e- 2y ~ 1_ e- 2a •

For y :::; -a we have
We take
11) 1. b) The function 11 + e2, z l is periodic with the period 'Ir. Therefore it is
enough to consider the case 0 :::; x :::; 'Ir.
By 11) 1. a) we have for lyl > a that 11 + e21Z 1~ A. For lyl :::; a, 0 :::; x :::; 'Ir, Iz-
'Ir/21
~ a the function 11+e2, z l is different from zero. Therefore inf 11+e21Z 1= m > 0,
where z belongs to the given region. Finally, taking B = min (m, A) we obtain the
desired inequality.
11) 2. By the definition of the function tan z we have
e2•z - 1 2z
tan z = -z e2•z - = -z + .
1 e 21Z +1
Then for z E Cn : z = n'lr, we obtain by 11) 1. b) 1 tanzi:::; l+i = M independently
of n (0 < a < 'Ir /2).
11) 3. By 11) 2. the condition (8.7) is fulfilled for the function f(z) = tg z and
rn = mr (n E ~). Therefore by I ) 2.
tanu -tan(-u) = -2u L

+00
n=-oo
Res
( tanz)
Z
2
- U
2
z=(2n+l)"./2
.
Since
Res ( tan z ) _ 1
Z2 - u2 z=(2n+1)"./2 - u 2 - ((2n + 1)11"/2)2
we obtain (tan u is an odd function)
+00 1
tanu = -2u L
n=O u 2 - ((2n + 1)11"/2)2·
11) 4. Let F be an arbitrary closed bounded region of the complex plain without
the points z = (2n + 1)11"/2 for n = 0, ±1, ±2, ....
We shall prove that the functional series from 11) 3. is uniformly convergent on
F.
Let d = max
uEF
lul. For u E Fand (2n + 1)11"/2 > d we have for n 2: no
1
IIul 2 - ((2n + 1)11"/2)21
1
<
u 2 - ((2n + 1)11"/2)2 - d2
1
as n --+ 00.
Since the series

+00 1
~ ((2n + 1)11"/2)2
is convergent, also the series
+00 1
~ ((2n+ 1)11"/2)2 -d2
is convergent.
Hence the series representing the function tg z is uniformly convergent on F.
Chapter 9
Analytic continuation
9.1 Continuation
9.1.1 Preliminaries
Definition 9.1 Let f be an analytie function in a region A. An analytie function 9
on a region Al that intersects A is an (direet) analytie eontin uation
of f from region A to region Al if f = 9 throughout Al n A.
Such analytic continuation is uniquely deterrnined.
Definition 9.2 Let f be analytie in a dise D and Zo E oD. f is regular at Zo if f

ean be eontinued analitieally to a region A with Zo E A. Otherwise, we say that f
has singularity at zoo
Theorem 9.3 If apower series E~=o an zn has a positive radius of eonvergenee R,

then the function
00
n=O
has at least one singularity on the eirc1e \z\ = R.
In partieular, if R < 00 and an ~ 0 (n E NU {O}), then f has a singularity at z = R.
Definition 9.4 If f( z) = E~=o anz n has a singularity at every point on its eirc1e of
eonvergenee, then that eircle is a natural boundary of f.
Theorem 9.5 If
00
f(z) = L ankz nk and · . f-

1ImIn
k_oo
- > 1,
nk+1
nk
k=o
227

228 CHAPTER 9. ANALYTIC CONTINUATION
then its circle oJ convergence is a natural boundary Jor f
Theorem 9.6 (Schwarz' reflection principle) Let A be a bounded region that

is contained in either the upper or lower half-plane and whose boundary contains a
segment L on the real axis. IJ f is analytic in A and continuous on AC, and f(z)
is real for real z, then it can be defined an analytic extension g oJ J to the region
AULUA*, where A* = {zlzE A}, by
J(z) for zEAUL

g(z) ={ _
f(z) for z E A*.
Let D o be a disc centered at a point zo0 Let z(t), t E [a, b) be a path beginning at
Zoand whit the end point W. If
is a partition of the interval [a, b), the we denote by D i a disc containing z(ai)' We say
that the sequence {D o, D h D 2 , ••• ,Dn } is connected by the path along the partition
if the image z([ai' ai+l]) is contained in D i . Let J be analytic on D o. An analytic
continuation of (J, D o) along a connected sequence (along a path C) [D o, . .. ,Dn ) is
a sequence of pairs (canonical elements)
such that (Ji+1' D i+1) is a direct analytic continuation of (Ji, D i ), i = 0,1, ... ,n-1.

Example 9.1 The Junction J is defined on the disc Izl < 1 with the power senes
00
L( -1t(2n + 1)zn
n=O
and Jor other values Jrom the complex plane by analytic continuation. Find this
continuation summing the power series.
Solution. The radius ofthe convergence ofthe given power series is 1. Therefore
for z = w 2 with Iwl < 1 we have
00
J(w 2 ) = L( -lt(2n + 1)w2n ,

n=O
9.1. CONTINUATION 229
00
= L(-1t w2n+1
n=O
W
=
1 +w2 '
where we have exchanged the integral and the sumo We have by the analyticity of
the function f( w 2 ) in Iwl < 1
putting z = w 2 we obtain
l-z
f(z) = (1 + Z)2'
The function obtained is the analytic continuation of the function f given by power
series in Izl < 1, on the whole complex plane without the point z = -1, where the
function f has a pole of the second order.
The desired analytic continuation can be obtained also by another method. We
have for Izl < 1
Since
00
(obtained by differentiating the series L(-1)n z n), we obtain

n=O
2z 1 1- z
f(z) = -(1+Z)2 + 1+z = {1+Z)2'
Example 9.2 The funclion f is defined on the disc Izl < 1 by the power series
Find its analytic continuation outside the disc Izl :5: 1.

Solution. Let x be real and -1 < x < 1. Then
(x f(x))' = f) -lt x3n

n=O
= A·
+ X
Hence
x f(x) = J 1+x3dx
- - = - - log
1
3
Vx 2 -
x+1
X + 1 + -v'3
3
2x - 1
arctan-- + C.
v'3
We obtain the constant C by the condition f(O) = 1: C = 7rv'3/6. Therefore for
-1 < x < 1 :
~ (_l)n
L...J - - x
3n
= -
1 ( 7r v'3
-- -
I
og
vx 2 -x+1;;;
+V3 2X-l)
arctan - - .
n=O 3n +1 3x 6 x +1 v'3
By the uniqueness of the analytic continuation this equality holds also for Izl < 1.
Since on the right side is an analytic function in the whole complex plane cutted
from the points -1, e"r/3, e-"r/3 in the radial direction to z = 00, Figure 9.1, so in
this region this is the analytic continuation of the left side of the equality.
Figure 9.1
Example 9.3 The power series

zn (z 2)n
L -
00
Z7r + L(-lt -
00
and
n=l n n=l n
have no common region 01 convergence. Prove that they are analytic continuations
of each other.
9.1. CONTINUATION 231
Solution. Both series define the same function J(z) = -log(1-z) which is analytic
outside of the cut from the point 1 to 00.
Example 9.4 The Junction 1 is defined in the disc Izl < 1 by the power series
(_1)n-l
L 00
z2n.
n=l n(2n - 1)
Find its analytic continuation outside 01 the disc Izl ~ 1.
Solution. For real x with -1 ~ x ~ 1 we have
I'(x) = 2 t (_1)n-l x 2n- \

n=l 2n - 1
and
Hence by 1'(0) = 1(0) = 0 we have 2x arctan x - log(1 + x 2 ) the obtained formula

holds also for Izl < 1, and this gives the desired analytic continuation in the whole
complex plane without radial sections from the points z, -z till 00, Figure 9.2.
I
/' -- - "-
'\
\
I
, 0 I
-
\ X
/
"- ./
./
-I
Figure 9.2
Example 9.5 Do there exist functions which are analytic at z = 0, and which
satisfy the condition
Solution. a) On the set

{~,nEN},
which has the accumulation point 0, the function f(z) = Z2 satisfies the given
condition.
b) There is no function which satisfies the given condition since on the set
{~, nE N} it would be f(x) = f(-x) = x 3 •
Exercise 9.6 The function j dejined on the unit disc Izl < 1, by the power series
00 ( l)n 00 z4n
a) ~ - z2n. b) ~ -I'
n==2 n(n-1) , n==O n.
Find their analytic continuations outside of the disc Izl :::; 1.
Answers.
a) j(z) = (Z2 + 1) log(l + Z2) - Z2 in the complex plane with cuttings from points
l, - lfix 00, Figure 9.2.
b) The radius of convergence of the given power series is 00. Therefore this power
series extend analyticly itself on the whole complex plane. We have
00 z4n
~ - , = (coshz + cos z)/2
n==O n.
which easily follows by the fact that the given power series satisfies the following
differential equation
j(iv)(Z) - f(z) = 0
and the initial conditions flll(O) = 1"(0) = 1'(0) = 0 and f(O) = 1.
Example 9.7 The function j is dejined in the disc Izl < 1 by
Prove that it can not be analytically extended outside the disc Izl :::; 1.
Solution. The function satisfies the functional equations
Then, since z = 1 is a singular point of the function I, the solutions of z2 = 1, Z4 =

1, z8 = 1, ... are also singular points of the function f. The set of all singular points
is dense on the circle Izl = 1 and they form the natural boundary of the function I.
L
00
Exercise 9.8 Let I(z) = unz n , with the radius 01 convergence R = 1. Putting
n=O
z = _z_ we have
l+z
I(z) = I (_Z_) = F(z) = f vnz n .

1+z n=O
Denote by p the radius 01 the convergence 01 the last power senes. Prove that
a) p ~ 1/2, where il z = -1 is a singular point 01 the [unction I we have p = 1/2.
b) 111/2 < P < 1, then the equality I(z) = F(z) = I (_Z_)
l-z
allows the
Izl < o[ the
I-z-I
analytic continuation 01 the [unction I outside o[ disc 1, and inside
circle = p.
l-z
c) II p = 1, then by the equality [rom b) the [unction F analytically extend I on
the half-plane Re z < 1/2, (see Example 9.11 )
Exercise 9.9 Let land 9 be arbitrary entire [unctions and let
00 (I-zn l_Z n
- 1)
Sz=
() ~ ---
1 + zn 1 + zn-l .
Prove that
I(z) = (f(z) + g(z)) /2 + S(z) (f(z) - g(z)) /2
reduces to the function I in the regionlzl < 1 and to the function g in the region
Izi > 1.

Example 9.10 1. Prove that lor Izl < 1
Iln(l + z)1 $ -ln(1 -Izl).
z
2. Find the singular points of the function In - - and the regions of analyticity
z-l
for each bmnch.
S. The bmnch of the function Ln_ z _ which is positive for z > 1 expand in a
z-l
Laurent series in the annulus 1 < Izl < R.
a) Find the coefficients of this Laurent se ries by the formula
b) Find the coefficients by some other method.

4. The functions
and 12 = 7l'Z + f( -lt (z -n 1t

n=l
analytically extend each other. Prove that.
Solution. 1. We have for Izl < 1
zn
+ z) = L( -lt -.
00
-In(l
n=l n
Therefore
For 2. and 3. see Exampie 7.17.

4. The both functions f1 and 12, see Exampie 9.3, are expansions in power series
of the functiün fez) = -ln(l-z), where first in the neighbürhood üf zero für Izl < 1,
and the second in the neighborhood of the point 2 for Iz - 21 < 1, Figure 9.3,
y
Figure 9.3
smce
-ln(l- z) = -ln(l - 2 - (z - 2))

= -ln(-1)(1+(z-2))
= -lne- 1r' + In(l + (z - 2))
(z l)n
+ E(-lt -n
00
7rZ
n=1
It is interesting to remark that the regions of functions 11 and 12 have no common

points, but they analytically extend each other (since they have common represen-
tation -ln(l - z)).
Example 9.11 1. Let ~ > 0 lor -1 < x < 1, and a is a complex number.
Prove that the integral
1=
1 dx
, 1-1 (x-a)~
is given by:
I = - 'Ir for a > 1,
";a2 -1
I =± 'Ir e·(31r-2t)/4 for a = ±et • for 0<t< 'Ir,
";2 sin t
I = ~t . ~
J'1I"":"":2 slgn ny lOr a = tY,
vI + y-
~
1=+ ~ for a < -1,
a2 -1
and for -1 < a < 1 the prineipal value (VP) is I = O.

2. The dise Izl ::; 1 map with the function w = _z_. What is its image in the
l-z
w-plane ?
Map by the inverse function the eireles Iwl ::; r, 0 < r < 00, from the w-plane
into the z -plane. What are the images?
L
00
3. Let f(z) = anz n be a eanonieal element of the eomplex analytie function

n=O
W
F with mdius R = 1. Put z = - - and let F(z) = G(w). Let g(w) = J(z) be a
I+w .
L enw n and let r > 0 be its mdius
00
eanonieal element for G( w) of the form g( w) =

n=O
of eonvergenee. Prove that r ~ 1/2. The analytie function g (_Z_)
I-z
analytieally
extends f from the region Izl < 1 to a region O. Find this region 0 for the eases
1/2 < r < 1, r = 1, r > 1.
Solution. 1. We start with the function
1
J(z) = y'l-Z2' a 7' ±1,
(z - a) 1- Z2
for which VI - x 2 > 0 for -1 < x < 1.

The function f has pole of the first order at the point z = a (a 7' ±1) and
algebraic branching points of the first order at z = 1 and z = -1. The cut is ( -1, 1),
since z = 00 is an ordinary point. We shall first consider the case a = ty, a< -1,
taking the path L from Figure 9.4.

y
-R R
Figure 9.4
We have
i J(z) dz = 27rZ Res(J(z))z=a. (9.1)
The integral on the path L can be written in the following form
j
= 1
[ dz 1- e dx dz
(9.2)
lL (z - a)Vf=Z2 -1+. (x - a)v'f=X2 + "11 (z - a)Vf=Z2
[R dx [dz
+ A+e (x - 2h!(1- x)(x + l)c"" + lK (z - a).Jf=Z2
[He dx j dz
+ lR (x - a).J(1- x)(x + l)e31r' + "12 (z - a)v'f"=Z2
r1+ e dx j dz
+ A-. (x - a).J(1 - X)(X + l)e 21r' + 1'3 (z - a)Vf=Z2·
Since e- 1r• = e31r' , the third and fifth integrals on the right side of (9.2) are zero.
Analogously we obtain that the integrals on /1 and /2 give zero since the functions
under integrals are two different branches with opposite signs.
Since
r
11 (z-a~nl
21f IRehz dtl
< 10 Re- t - lall hjl - R2e2.tl
2'1rR 1
<
IR -lall J(R - 1)(R + 1)
2'1rR
<
IR -lall(R -1)'
the integral on K tends zero as R -+ 00.
If in the preceding inequality we take the path /3 instead of K and e instead of

R we obtain again a true inequality, which implies
1 f(z)dz-+O
'Y3
as e-+O.
Therefore by (9.1) and (9.2) we obtain for R -+ 00 and e -+ 0,
2 1 (x-a)
1
-1
dx
~ = 2'1rz
l-x 2
Res(f(z))z=a.
Since
Res(f(z))z=a = (h) z=a ,

1- Z2
we obtain for a < -1 that I =
va 'Ir
2 -1'
smce
arg (1 + a) = 'Ir, arg (1 - a) = 0,

and so
~= J(a 2 - l)e1f••
In the case a = zy we have on the upper side of the cutting the positive branch
and on the down side the negative branch of the function~, and so we obtain
I = 'lrZ •
+ y-
~slgnny.
vI
In the case a = et ., 0 < t < 'Ir, Figure 9.5,
/ 1 x
;1
I
I
Figure 9.5
we have
1= 'Ir
VI - e 2t' e h / 2e- 1r./ 4 V2 sin t
'Ir e,(31r-2t)/4.
V2 sin t
In the case a = _e t • for 0 < t < 'Ir, we put in the preceding case a = e·(t+ 1r) and we
obtain
I = _ 'Ir e'(311"-2t)/4.
V2 sin t
In the case a > 1 we take the path L as in Figure 9.6,
-R R x
Figure 9.6
and we obtain in an analogous way as earlier
1= 7r Res(J(z))z==a, a> 1.
Since for a > 1,
arg (1 + a) = 0, arg (1 - a) = -7r,
we have
1=- 7r .
y'ä2=t
For -1 < a < 1, we take the path as in Figure 9.7,
,"y
I
-R R x
Figure 9.7
and we obtain
vp {lI dx
(x - a)Vf=X2 -
- 0
,
since by the analyticity of the function j in the region bounded by the path L we
have
f j(z)dz+
1s
1 1s
r j(z)dz=O.
2
Remark 1. The path L for the case a < -1 can also be taken as in Figure 9.8,
-R R x
Figure 9.8
and also analogously for a > 1.

Remark 2. The following general theorem holds:
Theorem. If:
(i) the function J analytic in the complex plane with singular points ZI, Z2,.·. ,Zn;
(ii) the function J is analytic for a < z < b;

lim ZT+s+IJ(Z) = A f-
(iii) z---+oo = (Irl < 1, Isl < 1), then
l (x - s)'(b - x)J(x) dx =
b
si:'lrs (E Res(z - a)'(b - z)'J(Z))Z=Zk - A) .
The condition (ii) can be relaxed to:
(ii)' If the function J has finitely many singular points aI, a2, ... ,am on a < Z < b,
then V P l (x - a)'{b - x)' J(x) dx
'Ir n m
= - . - L Res((z-a)'{b-zYJ(z))Z=Zk+Cot'lrsL Res((z-a)'(b-z)sJ(z)L=Uk·
SIn 'Ir S k=I k=I
2. We can write
z 1
w=--=-l+--.
1-z 1-z
Then starting from the z-plane we map the disc Izl ~ 1, Figure 9.9,
Figure 9.9
first by the function Wl = 1- z (rotate for 7r and translate for 1), and we obtain the
disc IWl - 11::; 1, Figure 9.10.
v,
Figure 9.10
Further, we map by W2 = l/Wl and we obtain the desired region Rew2 ;::: 1/2,
where W2 : 0 I-t 00, 2 I-t 1/2, and the border li ne have to be normal on the real
axis, see Figure 9.11.
01
Figure 9.11
Finally, we take w = -1 + W2, translation for -1, and so the desired region is
the half-plane
Rew;::: -1/2,
Figure 9.12.
Figure 9.12
The inverse function is given by

w
Z=--.
1 +w
We have z = -w1 = 1 - - 11 . So we ean in an analogous way as for the previous

+w +w
mapping find the images of dis es Iwl :$ r, 0 < r < 00.
Second method. Sinee I-z-I
1-z
= Iwl :$ r, we have for z = x + zy and so
We have for r < 1
and for r > 1 :
(x + 1 ~2r2) 2 + y2 2: (1 ~2r2)2.
We have for r = 1 the image x = 1/2.
1
2
Figure 9.13
L
00
3. Since J(z) = anz n is the canonical element of the analytic function F with
n=O
the radius of convergence R = 1,and we have by 2. that the unit disc Izl < 1 is
mapped by the function w = _z_ on the region Rew > -1/2, Figure 9.13, we find
1-z
that the dosest positive singular point for the canonical element g( w) (= J(z)) for
G(w) is the point w = -1/2. Therefore r ~ 1/2, z = -l.
Now we shall ex amine how the analytic function 9 (_Z_)
1-z
analyticallyextend
the function J from the region Izl < l.
(i) Let 1/2 < r < 1. By 2. the disc Iwl < r is mapped by the function z = ~
1 +w
on the disc
(x + 1 ~2 r 2 ) 2 + y2 < (1 ~2r2)2'
which gives us the region of the analytic extension of the function J by the function
9 (_Z_)
1-z
, Figure 9.14.
--~~~~~~~~---~
X
Figure 9.14
If the set of singularities of J( z) = g( w) are not dense on the new cirde we can
continue with the analytic continuation.
(ii) Let r = 1. By 2 the disc Iwl < r is mapped by the function z = ~ on the
l+w
region Re z < 1/2, which gives us the region of analytic continuation of the function
f by the function 9 (_Z_)

1- z
,Figure 9.15.
1 x
Figure 9.15
If the set of all singular points of the function f (z) = 9 (w) is not dense on the
circle obtained the procedure of the analytic continuation can be continued.
iii) Let r > 1. By 2. the disc Iwl < r is mapped by the function z = ~ on
1 +w
the region
which gives us the region of analytic continuation, Figure 9.16.
+y
_.-e:.
1_ r'l
Figure 9.16
For further analytic continuation it holds the same as for the previous cases.
Example 9.12 1) Let f be a mtional function which has on the positive part of
real axis only poles of first order ih, b2 , ••• , bm • If there exist other poles, denoted by
ab a2, ... , an, they are all different from zero. Let p be areal number such that
limzPHf(z) = lim zP+lf(z) =0.

%-+0 z--+oo
Prove that for p real which is not integer

n
VP 10'>0 xPf(x)dx = _ _'lr_
sin 'lrp
e- 1rp•
2: Res(zP f(z))z=ak
k=l
m
-'Ir cohp 2: 14Res(f(Z))z=bk,
k=l
where x P > 0 for x > o. Prove that for pinteger
[00 n m
VP 10 xPf(x)dx = - 2: Res(zPf(z)Ln z)z=ak - 2: 14(lnbk + n) Res(f(z))z=bk'
o k=l k=l
where Ln z = In z + z arg z, 0 ~ arg z < 21l".

00
2) Let g(z) = :LAiZ;, where

;=1
k-l
Al = 1, A k = :LA;Ak - l + 1, k = 2,3, ....
;=1
The funciion h is given by
Z
h 2 (z) - h(z) +- - = o.
1-z
Prove that one branch of the funciion h is an analytic extension of the funciion 9
and find from which region to which region.
S) Map the region -1l"/n < argz < 1l"/n, Izl < 1, by thefunciion
z
w(z) = 2' nE N,w(z) > 0 for z> o.
(1 + zn)n
Solution. 1) i) Let p is areal number different of a integer. The function under

the integral is f(z)zP and the path L as in Figure 9.17, where R is enough big that
all poles of f ; ak, k = 1,2, ... , n and bk , k = 1,2, ... , m, will be inside. The points
z = 0 and z = 00 are branching points of zP (p is not an integer) and we have a cut
from 0 to 00.
Using the partition of the path L from Figure 9.17 we represent

y
-R R x
Figure 9.17
the integral on the path L in the following way
kzPJ(z)dz =
where fi is the upper semi-circle and "ti is the lower semi-circle around bio
Because of lim zpH J(z) = 0 we have
z-+oo
l iR zP J(z) dzl S; RPH L IJ(Re ')1 dO

21r 9 -+ 0 as R -+ 00
(exchange of the limit and integral). We obtain in an analogous way
{ zP J( z )dz -+ 0, as c:-+ 0,
lK,
since limz ..... o Zp+1 J( z) = O. Since the points bk, k = 1,2, ... , m, are poles of the first
order we have for € -+ 0
1 zP J(z) dz = -7rlll,;Res(J(Z))z=bk'
Ik
and on lower semi-circles
1 zP J(z) dz
Ik
= -7rZ eP(lnb k+2,n)Res(J(z))z=b k
By Residue Theorem we have
t Res(zPJ(Z))z=ak'
1, zPJ(z) dz = 27rZ k=l
L
Therefore we have for € -+ 0 and R -+ 00
(1 - e 27r ')V P 10 00
J(x) dx = 27rZ t
k=l
Res(zP J(Z))z=ak
m
+7rz(l + e27rp,) L 14Res(J(z))z=bk'

k=l
Then
vP 10 00
x PJ ( x) dx
m
-7r cot7r L lI,;Res(J(z))z=bk'

k=l
where x P > 0 for x > O.
1) ii) pis an integer. We apply the integral on the function 'ljJ(z) = zP J(z)Lnz on
the same path L, where Ln z = In Izl + zarg z, 0::; arg z < 27r. Using the partition
of the path we obtain
i zPJ(z)Lnz dz = tl-~ xPJ(x)lnx dx

+f LzPJ(z)Lnzdz+ 'Y: l b'+I-<xPJ(x)lnxdx
1
i=1 I. b.+< ;=1
+ [R xPJ(x) lnxdx + [ zPJ(z)Lnz dz

lbm+< lK R
+
R
l
bm +< x PJ(x)(ln x + 27rz) dx +
i=l "
f 1
zP J(z)Lnz dx
+I: 1
[b i-< x PJ(x)(lnx+27rz)dx+ [ zPJ(z)Lnzdz.
i=1 lbi+I+< lK,
We have by supposition zP+1J(z) -t 0 as z -t 0 and also as z -t 00. Therefore also

zP J(z)Ln z -t o. Hence the integrals on ](R and ](. tends to zero as R -t 00 and
E -t 0, respectively.
We have at poles of first order bk , k = 1,2, ... , m, on the upper semi-circles as

c-tO
1
zPf(z)Lnz dz = -7fZ bpnbkRes(J(z))z=bk'
"Ik
and on the lower semi-circles
~ zP J(z)Ln z dz = (-7fZ b~ + 27fz)Res(J(z))z=b

Jik k •
By Residue Theorem we have
1zPf(z)Lnz dz =
L
27fZ tRes(zPf(z)Lnz)z=ak.
k=1
Letting R -t 00 and E -t 0 we obtain
lX> xPJ(x)dx
n
-27fZVP = 27fZ LRes(zpJ(z)Lnz)z=ak
k=1
m m
+27fZ L Res(J(Z))z=bk~ lnbk - 27f 2 L ~Res(J(Z))z=bk'
k=1 k=1
which gives us the desired equality.
L aif Then
00
2. Let h(z) =
i=1
h2 (z) _ h(z) + _z_ = 0 (9.2)
1-z
implies
00 00
h(z)h(z) + Lz i = Laizi,
i=1 ;=1
Hence
k-1
a1 = 1, ak = L ajak_j + l.
;=1
Therefore the power series of the functions h (one branch) and gare equal.
We find h solving (9.2)
The branching points of the function h are 1/5 and 1. Then the radius of convergence
of the power series of h, and also for g, is 1/5, we have
E Aii
00
h(z) = g(z) = for Izl < 1/5.

;=1
The function h from (9.2) is an analytic extension of 9 on the whole complex plane
out of the cut from 1/5 to 1.
3) The given function can be written in the following form
Remark. If n is odd the function w = w( z) has n branching points of (n - 1)-th

order in the points
z (e,,·/n)
= k, k = 1,2, ... , n.
For n even, n = 2m, w(z) has the representation by two function

m
wm _± _z__
1,2 - 1 + z2m
When w(z) > 0 for z > 0 we have the behavior of the maps in Figures 18 and 19.
y
•
Figure 9.18
Figure 9.19
Chapter 10
Integral transfornns
10.1 Analytic Functions Defined by Integrals
10.1.1 Preliminaries
Let 0 be a region in rc. We denote by A(O) the set of all analytic functions in O.
Definition 10.1 A sequence {fn} from A(O) converges in A(O) to f E A(O) if

for every closed subsei F of 0 and every c > 0 there exists no E N such that
max Ifn(z) - f(z)1 < c

zEF
for every n ~ no.
Theorem 10.2 If a sequence {fn} from A( 0) converges uniformlyon every closed

and bounded subset of 0, then the limit function is also regular on O.
Theorem 10.3 If a family {fq}qEQ, Q C lW. from A(O) converges in A(O) to f E

A(O) as q ~ qo, then the family {f~}qEQ converges to f' in A(O).
Theorem 10.4 Let 0 be a simple connected region in C and 9 = g( z, t), 9 : 0 X L ~

C, for L : Re z ~ a, Im z = 0 satisfies the following conditions: (i) 9 is a continuous
function of t on for fixed z E 0;
(ii) 9 is an analytic function of z E 0 for fixed t;
(iii) the family {f:
g(z, t) dth~a converges in A(O). Then the function
f(z) = t g(z, t) dt
255

256 CHAPTER 10. INTEGRAL TRANSFORMS
is analytic in 0 and we have
J'(z) = l a
b
g(z, t) dt.
Bt
Theorem 10.5 Let A be a region in C and 9 = g(z, t),g : A x L -+ C, be a

continuous function of t onto L : a ::; t ::; b, for fixed z and an analytic function of
z E A for fixed t. Then the function
f(z) = t g(z, t) dt
is analytic in A and we have
J'(z) = t g(~/) dt.

Example 10.1 Prove that the gamma function
r(z) = 1'XJ e- t t z- 1 dt
for Re z > 0 is an analytic function.
Solution. The given integral is improper because of the point 0 and the infinite
interval. Therefore we shall write
1 00
e- tt z - 1 dt = 11
e- tt z - 1 dt + 1 00
e- t t z - 1 dt.
The integral 1 1
e- tt Z - 1 dt converges absolutely for Re z > 0 and uniformlyon every
bounded and closed region G in Re z > o. Namely, if mi[l x
zEO
= 8 for z = x + zy, we
have
!e- t t Z - 1 ! = e-ttRez-l ::; t o- 1 for z E G.
Since the integral 1t1 O 1

- dt converges, then by Weiersrtass' criterion follows the uni-
form convergence of the integral on G. The integral 1 1

e- t e- 1 dt is then an analytic
function for Re z > o. On the other hand 1 00
e- t e- 1 dt converges uniformly and
absolutelyon every bounded region G. Namely, if Gis in the disc Izl ::; M, then for
z E G,
!e -ttZ-1! _< e-t eM+1 ,
10.1. ANALYTIC FUNCTIONS DEFINED BY INTEGRALS 257
and the integral 1 00

e-ttM-1dt converges, and therefore by Weierstrass' criterion the
integral 1 00
e-te-1dt is an analytic function in the whole complex plane.
So it follows that the gamma function r is a regular function in half-plane Re z >
o.
Example 10.2 Prove that the gamma function r(z) satisfies
a) r(z + 1) = z r(z) for Rez > 0;
b) r(n + 1) = n! for nE N.
Then extend analytically the function r outside of Re z > o.
Solution. Applying the partial integration on r(z) = 100

e-te-1dt for Rez > 0
we obtain
r(z) 100
e-tt z - 1 dt
= e_t~ICO + ~ (CO e- t t(z+1)-l dt

z 0 z Jo
r(z + 1)
= z
b) Putting z = n and applying a) we obtain

r(n + 1) = n(n - l)(n - 2)···3·2· r(I).
Since r(l) = 1 00
e-tdt = 1, we obtain r(n + 1) = n! for n E N.
We can find the analytic extension of r by the equality in a). Namely, we have
for -1 < Rez < 0 that r(z) = r(z + 1) is an analytic function, since Re (z + 1) >
z
O,Rez> -1. Further, we have for -2< Rez ~ -1 without -1
r(z) = r(z + 1) = r(z + 1).

z z(z+l)
So for the general case -( n + 1) < Re z ~ -n without -n we have
r(z)= r(z+(n-l)) .
z(z + 1) ... (z + n)
In this way we obtain the analytic continuation of r on the whole complex plane
without the poles at z = 0, -1, -2, ... For those points we have
Remark. The gamma function is a special case of the Mellin transform. This
transform is defined for complex functions of real variable <p(t) which satisfy the
condition
1<p(t)1 ~ cr'" for 0< t < 1; 1<p(t)1 ~ c1C ß for t> 1, and a < ß.
Then the Mellin integral transform for the function <p(t) is given by
4>(z) = 1'' ' <p(t)t Z

-
1 for a< Rez < ß·
Example 10.3 Prove that the beta funciion
B(p, q) = la oo P
t - 1(1 - t)q-l dt
/or Re p > 0 and Re q > 0 satisfies
a) B(p, q) = B(q,p); b)B(p, q) = r(p)f(q).

f(p + q)
Solution. a) Putting t = 1 - u we obtain
B(p, q) = la t
1
P- 1(1 - t)q-l dt = J1
uq-1(1 - U)p-l du = B(p, q).
o
b) We take in f(z) = la oo e-te-1dt the substitution t = x 2 (and y2). Therefore

for z = p and z = q we obtain
(OO 2 2p (OO 2
f(p) = 2 10 e- X X - 1 dx, f(q) = 2 10 e- Y y2 q-l dy,
respectively.
Taking the polar coordinates:
x=pcos<p, y=psin<p, O<p<oo, O<<p<1f/2,

8(x,y) .
and so 8(p, <p) = p, we obtam
f(p)f(q) = 41 00
la oo
e-(x2+y2)x2p-ly2q-l dxdy
= 4 1 00
la 7r
/
2e- p2 p2P+2q-2p( cos <p ?p-l(sin <p )2 q -l d<pdp
= 2 1 00
e- p2 p2 P+2Q-l dp· 2la 7r / 2(cos <p )2P-l(sin <p )2 q -l d<p
= f(p + q)B(p, q),
where we have taken t = cos 2 cp in
B(p, q) = la 1 tp-1(I - t)q-l dt,
B(p, q) = 2la f (cos cp?p-l(sin cp)2 q -l dcp.
Example 10.4 Solve the Bessel differential equation 0/ n-th order
Z2W" + zw' + (Z2 - n 2)w = 0,
when n is not an integer.
Solution. We shall solve the differential equation in the reals and then using
analytic continuation we shall obtain the complex solution. So we shall find the
solution of
x 2 y" + xy' + (x 2 - n 2 )y = 0 (10.1 )
in the following power series form
Putting this series in (10.1) and taking the coefficients of the same power of xequal
to zero
(2n + I)al = 0, (2n + k)ak + ak-2 = 0, k = 2,3, ....
Hence the coefficients by odd powers are zero and even coefficients are given by
(-I)k ao
a2k = 22kk!(n + I)(n + 2) ... (n + k) for k = 1,2, ...
Putting ao = 2nr(~ + 1)' we obtain a particular solution of the equation (10.1) for
nE ~ (n can be also integer):
x n 1 (X)2k
(2) E(-I)k k!r(n + k + 1) 2
00
Jn(x) =
Extending the obtained solution on the whole complex plane (the series has radius
of convergence 00) we obtain the general solution of the Bessel differential equation
Z2W" + zw' + (Z2 - n 2)w = 0 (n is not an integer).
It is given by
for A,B E C.
The function
z)n k 1 (Z)2k
= ( 2 {;( -1) k!r(n + k + 1) 2
00
Jn(z)
is the Bessel function of first order with index n. Why is the general solution not a
solution for n integer?
Example 10.5 Prave that
E
00
a) ez(t-l/t)/2 = Jn(z)tnj
n=-oo
Solution. We shall find the Laurent series of ez (t-l/t)/2 in the neighborhood of

w=O:
with Un = -
1 1 ez (t-l/t)/2
dt.
n=-oo 27rZ Itl=l t n +1
. m
P uttmg . t he I
ast'mtegraI t = -2u
z
we 0 b '
tam
Un = 2~Z (~) n jUI=tlzl eu - z2 /4u u:+l du for n = 0, 1,2, . .. .
By the theorem on residues have
To find the preceding residue we shall start from the following expansion
The essential singularity of the above expansion of eu-z2/4U at point u = 0 will give
us the coefficient by u- 1 (the residues). Therefore
z)n ~ k 1 (Z)2k
un = ( 2 f:'o(-l) k!r(n+k+1) 2 =Jn(z), n=0,1,2, ...
It is easy to check that
Ln(z) = (-ltJn (z) for n=0,1,2, ....

b) By the preceding a)
L
00
ez (t-l/t)/2 = Jn(z)t n.
n=-oo
Then
e z (t-l/t)/2 (~ (t + ~) )
=
n=-oo
Hence 00 00 00
n=-oo n=-oo n=-oo
Remark. The function ez (t-l/t)/2 is the generator of the Bessel function Jn(z).
Exercise 10.6 Let us consider the Wright function in the integral form
where u- a is the principal branch 0: > -1, and the path C is a on Figure 10.1
Figure 10.1
(starting at -00 on the real axis, going around (0,0) in the positive direction and
coming back to -00).
Prove that: If 0: > 0, ß > 0, then the Wright function has the following repre-
sentation in every finite part of the complex plane
cI>(ß, a, z) = -1 lXo+,oo
w- ß eW +z .w
-Q
dw for Xo > 0,
27rZ Xo-'OO
l
where
XO+,y lxo+.oo
lim = .
y--+oo xo-"Y xo-too
Hints.Use the path C' from Figure 10.2.
y
___.0-----.. E
w
A.
~ F
Figure 10.2
Remark. The Wright function has the following power series representation
Ef(k + l)f(ß + ak)'

00 zk
cI>(ß, 0:, z) =
where ß is a complex number and 0: > -1 (see E.M. Wright: On the coefficients of
power series having exponential singularities, J. Lond. Math. Soc. S., 1953, 71-79).
Exercise 10.7 Prove that the gamma function f can be extended over the whole
region of the existence by the equality
(( _wY- 1 = e(z-l)ln(-w)), where C is the path given on Figure 10.3.

Figure 10.3
Example 10.8 Let 0 and a E ~ and <pet) = 0, for t < O. The Laplace transform of
'P is given by
1
4>(z) = 00 <p(t)e- zt dt,
(if it exists, generally in the sense of the Lebesgue integral) which is an analytic
function in the half-plane Re z > Q.
Find the Laplace transforms of the following functions
a) <pet) = 1 for t;::: 0; b)<p(t) = ekt for t;::: 0;

c) <pet) = t -
n 1 for t;::: 0, n > 0; d)<p(t) = 1 - ate-at, t > -a, a E ~;
e) <pet) = 1 - (1 + at + a2 t 2 /2)e- at for t;::: -a, a E~.
Solution. As usual, we write 4>(z) = L(<p(t)).

a) 4>(x) = {OO e- xt dt = ~ for x> O. We obtain by the analytic extension
Jo x
4>(z) = 1 e-
o
00
zt dt
1
=-
z
for Re z > 0,
L(I) = l/z.
b) 4>(x) = ['>0 e-xtektdt = {OO e(k-x)tdt = _1_ for x > k.
Jo Jo x- k
With the analytic continuation we obtain L(e kt ) = ~k for Rez > k.

z-
c) In an analogous way we obtain
a2
cl) We have L(cp) = ( )2 for Rez > -a
zz+a
a3
e) We have L(cp) = ( )3 for Rez > -a.
zz+a
Example 10.9 Prave that if for the function cp there exist cp(n)(t) and l cp(t)dt
then.
a) L( cp(n)(t)) = zn L( cp(t)) - zn-lcp(O) _ zn-2cp'(0) _ ... _ cp(n-l)(O).
b) L (l cp(t) dt) = L(:(t)).
Solution. The proof goes by incluction. Let n = 1. Then
L(cp'(t)) 10'00 cp'(t)e- zt dt

cp(t)e-ztl: +z 1 cp(t)e-ztdt
00
zL(cp(t)) - cp(+O).
Supposing now that a) is true for n = k we can prove a) in an analogous way for
n=k+1.
b) Using a) we obtain
L ((l CP(t)dt),) L(cp(t)) zL (l cp(t)dt) ,

= =
L (l cp(t) dt) L(:(t)). =
Example 10.10 Let cp(t) = L-1(<P(z)) be the inverse Laplace transform. Find with
it the solution of the following differential equation
w'(z) + aw(z) = a(l - aze- az _ e- az ),
for w(O) = 0, a E llt.

Solution. We shall solve first the equation over the reals and then by analytic
extension we shall obtain the its solution in the complex plane. We start from the
problem
y'(t) + a y(t) = a(l - at e- at - e- at ), y(O) = O.
Applying the Laplace transform and using Example 10.9 and Example 10.8 d)
L(y' + a; y) = aL(l - ate-at _ e- at ),
L(y') + aL(y) = aL(1 - ate-at _ e- at ).

Further , we have
a3
xL(y) - y(O) + aL(y) = xx+a
( )2'
Hence
a3
(x + a)L(y) = (
xx+a
)2'
a3
L(y) = x(x + a)3
Applying now the inverse Laplace transform L- 1 we obtain
Since by Example 10.8 e)
we have
y(t) = 1 - (1 + a + at + a 2 t2/2)e- at .
With the analytic extension we obtain the complex solution
w(z) = 1 - (1 + az + a 2 z 2 /2)e- az for Rez> -a.
F(n)(z) = L((-ttcp(t)),
for n E N. Using this result find a particular solution of the Bessel differential
equation
Z2 W " + zw' + (Z2 - n 2 )w = 0,
with w(O) = 0, for n E NU {O}.

Solution. By the uniform convergence of the integral we can exchange the deriva-
tive and the integral
F'(z) = 10 00
e- zt ( -tcp(t)) dt = L(-tcp(t)).
Then the proof follows by the induction.

We shall find now a particular solution of the differential equation in real
We put u(t) = tny(t), and then we obtain
tu"(t) + (1 - 2n)u'(t) + tu(t) = 0, u(O) = o.

Applying the Laplace transform on the preceding equation, taking
dF
F(x) = L(u(t)), L(tu(t)) =- dx' L(u'(t)) = xF(x),
L(u"(t)) = x 2F(x) - u'(O), L(tu"(t)) = _~(x2 F(x) - u'(O)),

dx
we obtain
(x 2 + I)F'(x) + (2n + l)xF(x) = O.
The general solution with respect to F is given by
1) -(n+I/2)
F(x) = C(x 2 + 1)-n-1/2 = CX 2n - 1 ( 1 + x 2 '
where C is an arbitrary real constant.

By the power series expansion we have
L(u(t)) = F(x) = n!C ~ (-I)k (2n + 2k)!

(2n)! t:o
22k k!(n + k)! X2n +2k +I .
Applying the inverse Laplace transform we obtain
_L- 1 (n!C 00 (_I)k (2n+2k)!)

u(t) - (2n)!Ek!(n+k)! x2nHk+1
n!C 00
= (2n)! E (_I)k
22k k!(n + k)!L
-1 ((2n+2k)!)
X2n+2k+1'
Since
L-1 ((2n + 2k)!) _ 2n+2k . h () m!
X2n+2k +1 - t ,wIt L t m = Xn+1'
we obtain {or
n!C 1
(2n)! = 2n' {or C = 1 :
(t) _ ~ ( -1 )k t 2n +2k
u - f:'o 2"k!(n + k)! .
Since it was u(t) = tny(t), we have
( -1 )k (t) n+2k
y(t) = E
00
k!(n + k)! 2" .
With the analytic continuation we obtain the complex solution
(_l)k (Z)n+2k
L klr( k )
00
w(z) = -2 = Jn(z) {or Rez > O.
k=o·n++1
Exercise 10.12 Let f be a function defined by 10 00

e-zte t sin etdt.
a) Find the region where f is an analytic function.

b) Extend analytically the function f on the whole half-plane Re z > -1.
Solution. a) Re z > O.
b) Putting et = x the given integral reduces to f(z) = [00 sinx dx {or X Z = ezlnx .
J1 X
Z
Applying the partial integration we obtain

[00 cos x
f(z) = cos 1 - A x z +1 dx.
The last integral converges {or Re z > -1.
Exercise 10.13 The Riemann zeta fundion is given by

1
= L --; = L
00 00
((z) e- zlnn •
n=1 n n=1
a) Expand the function (z) in a Taylor series in a neighborhood of z = 2 and

find the corresponding radius of convergence.
b) Prove that for Re z >1
1 [00 w z - 1
((z) = r(z) Jo ew - 1 dw.
c) Extend analytically the fundion ((z) on the whole complex plane without the
point z = 1.
Answers.
and
00 1 11'2
UO=2:-=-.
k=l k2 6
The radius of the convergence is R = 1.
b) and c) For the analytic extension consider the integral
1 (_WY-l
~-'--dw,
c eW -1
where C is the path given on Figure 10.3.

Example 10.14 1. Find lor the function
its singular points and their nature lor different values 01 the complex parameter a.
For a = 2 expand the given lunction in apower senes in the neighborhood of z = 1.
Find the disc 01 convergence.
2. The lunction f is given by
Prove:
2 a) f can be represented by two senes
00 00
fez) = 2: Un(Z) + 2: vn(z) for Rez > 0,

n=l n=l
where
2b) Both senes converges uniformly in every closed region 01 the form
0< Xo:::; Rez:::; Xl< 00,

and r is an analytie function at the right half - plane.

3) Starting from the integral
where L is the path in Figure 10.4,
~y
I
I
,I
Figure 10.4
represent the following two integrals by r -sine and -eosine functions given by
1= ca cos t dt, 1= ca sin t dt for 0 < a < 1,
respeetively, and prove that they exist.
Solution. 1. (i) a E M. Then the point Zl = 0 is a pole of order a.

(ii) a is a negative integer. The point Z2 = 00 is an essential singularity.
(iii) ais a rational number, i.e., a = E., where (p, q) = 1. Then points Zl = 0 and
q
Z2 = 00 are branching points of order q.
(iv) ais a rational number or a complex number al + W2 (a2 :j:. 0).
Then the points Zl = 0 and Z2 = 00 are branching points of infinite order.
For a = 2 the given function is J(z) = -i-. The expansion in the neighborhood at
Z eZ
z = u (u i= 0) in power series we obtain in the following way. Let w = z - u, z =

w+u. Then
On the other side we have
A=
1
Z2
1,
= (w + U)2 = U (w); u(w)
1
= - w + u = -;:
1
E
00 (-1 )nwn
un '
and so
Then for u = 1
1
= (E(-l t (n + l)(z -l t ) (e- E(_l)n~ _l)n)
1
1)"
e- 1 L L T,(-l)n-k(n
00 n (
- k + l)(z _l)n
n=O k=O .
1
e-1L L
00 n
= k ,(-lt(z- l t(n-k+1).
n=O k=O .
The radius of convergence is Rand
R ~ min[R(A), R(B)] = 1.
2. a) We have for Rez >0
r(z) fooo e- t t z- 1 dt
= fo1 e- t t z-1 e- e- dt1 dt + 00

t 1
1 e- tz- dt + 1 e-~u(z+1) du,

00
t 1 00
1
where we have taken u = t'
Then
and
So we obtain
L un(z) + L vn(z)
00 00
f(z) = for Rez > o.

n=O n=O
2. b) We shall prove that both series from 2. a) converges uniformly in every

closed region: 0 < Xo :::; Re z :::; Xl < 00.
We have
Since the series

Ee-n(n+1Y' _n X1
n=O Xl
L
00
is convergent we have by Weierstrass criterion that Un(Z) is absolutely convergent

n=l
for 0 < Re z :::; Xl. what implies its uniform convergence on the given closed regions.
We obtain in an analogous way
Ivn(z)1 Il n+1 e-l/tr(z+l) dtl
< e- l /(n+1) l n +1 C(xo+l) dt
= e- l /(n+1) (n + 1)-"'0 - n- XO
-xo
Since the series
E e-l/(n+l) (n + l)-XO - n XO
n=l -x
L vn(z)
00
is convergent we have by Weierstrass' criterion that the series is abso-

n=l
lutely convergent for 0 < Xo :::; Rez, which implies the uniform convergence in the
previously mentioned closed region.
By Theorem 10.2 we have {(vn } C A(O), n E N) and Ek=l Vk(Z) converges

uniformlyon every closed and bounded subset of 0, then E:=l Vn E A(O); and
«u n ) c A(O), n
E N) and E k=l Uk(Z) converges uniformlyon every closed and
bounded subset of 0, then E:=l u" E A(O). Then by 2. a) we have
(Eun E A(O) and ~vn E A(O))::::} r = Eun+ EVn E A(O), i.e. ,
r is an analytic function on 0 = {z I Rez > O}.

3. See next Example 10.15.
e'"
Example 10.15 I) Let F(z) = - .
zP
1. Find the zeros, singular points and periods of F for different values of the
parameter p. For p = 2, find the real and the imaginary parts of F.
2. Let 0 r > 0. These points make a closed path
P which consists of straight parts AB,BC,CD and a part DA on a circle with the
center (0,0), Figure 10.5.
A B x
Figure 10.5
Using the path P find the integrals
1 00
o
cos x
--dx and
xP
10
00 sin x
--dx,
xP
i ~nowing that 1000
e- II yP- 1 dy = r(p) (gamma function).
3. Using 2. find also the following integrals
and prove that they exist.

II.) Let fand 9 be two given complex functions such that f(a) =f 0 and 9 has a
zero of second order at z = a.
/1.1) Find the residues of L at the point z = a with the values of these functions
9
and their derivatives at z = a.
II.2) Find the coefficients with negative indices in the Laurent series ofthe func-
tion L using the values of fand 9 at the point z = a.
9
Z2 elZ
II.3) Apply the results from 1) and 2) on the function (2
Z + 1)2'
Solution.
= -ezp
iz
1. (i) For pE N the function F(z) has a pole of the order p at the point
z = 0, since
e'z 1 00 znzn 00 1
-;; = zp ~ --;:J = n~p (n + p)!zn+pzn.
It has no zeroes since e'Z f. 0 for all z E C and for z = 00 it has an essential
singularity.
1. (ii) For p = -k (k E N) the function F has a zero of the order k at z = 0,
and for z = 00 it has an essential singularity.
1. (iii) For p = 0 the function F(z) =e lZ has an essential singularity at z = 00.
It is periodic with the period 271".
1. (iv) For p = ~,
n
and n are integers different of zero without common
where k
divisors, then z = 0 is an algebraic branching point of the order n.
1. (v) For p an irrational number the point z = 0 is a transcendental branching
point.
1. (vi) For p = Pt + ZP2 (P2 f. 0) be a complex number the point z = 0 is an
infinite branching point of F.
For p = 2 we have for z = x + zy
e'Z e'("'+'Y) e- Y e''''(x 2 - y2 - 2zxy)
F(z) = -z2 = (x + .y)2
• = (x2 - y 2)2 +xy 4 2 2
Hence
-y
Re F(z) = (x 2: y2)2 (x 2 - y2) cos X + 2xy sin x),
12. The integral ip ~zP dz < <

.z
for 0 p 1, can be written in the following form
ipzP-dz= iTR -dx-;/-

eU e'X
xP
1 e,(,Rt+(I-t)R) (zR - R)dt
(zRt+(I-t)R)p 0
1
+z i --dy+
T e-Y
(zy)p R
1f e'T erPe,9p
9'zre 9'dO
.
0
(10.2)
The function e' has no singularities in the region bounded by the path P and
Z
p
therefore by Cauchy theorem we have
i -dz
p zP
e'Z
= O. (10.3)
Letting R -+ 00 the integral on the path Be tends to zero, since
(zR - R) dt I < e- Rt ..,j2Rdt

IJ[1o e,(,Rt+(I-t)R)
(zRt + (1 - t)R)p
[1
Jo (zRt + (1 - t)R)p
< [1 e- Rt .j2Rdt
Jo RP:iiB.
1 e- Rt dt
2
1
= 2R-P
-+ 0
as R -+ 00, where we have used that
IzRt + (1 - t)RI -09:9

> min IzRt + (1 - t)RI = .j222R
_ e'z
for 0 :::; t :::; 1. The integral on the path DA tends to zero as r -+ 0, since z zP -+ 0
as z -+ 0 for 0 < P < 1.
Therefore by 10.2 and 10.3 we have for R -+ 00 and r -+ 0
VP1o 00 cosx + zsinx d x- VP1°O ze-- Y d

y=O.
xP 0 zPyP
10.2. GOMPOSITE EXAMPLES 275
Since the integrals in the previous equality converge absolutely we can ornit the
symbol V P. Since
we obtain
1o
00 cos x + zsin x dx =
xP
z(cos -7fp - z sm
2
.
-7fP )r(1 - p ).
2
Therefore
1 o
cos x
00 . 7rp
- - dx = sm _ . r(l- p)
xP 2
1
and
sin x 7rp
= cos -
00
- - dx . r(1 - p),
o xP 2
for 0 Z
Remark. By the analyticity of the function - outside of the point z = 0, it
zP
was possible to interchange the path BG by one on circle Re B• (0 ~ () ~ 27r), and
then to prove that the integral an BG tends to zero as R ---7 00.
3. We easily obtain the desired integrals putting p = ! in 2.
1 00
o
cosr,:;;x dx = 2
yX
1 0
00
cos e dt = sin -7r . r ( -1 ) ,
4 2
where we have taken x = t 2 • Therefore
1 00
o
COS x 2 dx
y'2
== -r(1/2) = - -
4
,j2i
4
and
The convergence of the first integral follows by
1""
o
cosx 2 dx =L
00
k=O
jV(k+ 1 )1f/2
Vkrr/2
cosx 2 dx =-
1
L 1(k+1)1f/2
00
2 k=O krr/2
COS x
r:::: dx.
Y X
Since this series converges by Leibnitz criteria since the coefficient
1(k+krr/2
1 )1f/2 cos x
--dx
yX
converges monotonically zero as k ---7 00 and the sign is alternatively changed~
We can prove analogously the existence of the integral 1

00
sin x 2 dx.
11.1) The function F(z) = :~;j has at point z = a a pole of second order, since
9 has a zero of second order at z = a and J(a) i= O.
Therefore
R (F( » _ = ( _ )2J(Z»)' = (f'(Z)h(Z) - J(Z)h'(Z»)

es z z_a Z a (z) h2(z) ,
9 z=a z=a
where g(z) = (z - a)2h(z), h(a) i= 0, since 9 has zero and order u at z = a. We find
by the last equality
g'(z) 2(z - a)h(z) + (z - a?h'(z),

g"(z) 2h(z) + 4(z - a)h'(z) + (z - a?h"(z),
g"'(z) 6h'(z) + 6(z - a)h"(z) + (z - a)2h"'(z).
Therefore
h(a) = g"~z), h'(a) -_ g"'(z)
6 .
Putting the obtained results in residues of F we obtain
Res(F(z»z-a = 2f'(a) _ 2J(a)g"'(z).

- g"(a) 3(g"(a»2
11 2. Since F has a pole of second order at z = a, the Laurent series of this

function has the form
00
F(z) = L un(z - at·

n=-2
We shall find U-l and U-2.
We have that 11.1. implies
2f'(a) 2J(a)g"'(a)
U-l = Res(F(z»z=a = g"(a) - 3(g"(a»2 .
Further integrating through circle J( with the center at the point a, taking J( inside
of the annulus of the analicity of the function F, we have
= 1 [ F(z) dz
2n lK (z - a)-l
1 [ J(z)dz
2n lK (z - a)h(z)
27rz R ( J(z) )
27rz es (z-a)h(z) z=a
( f(z) )
h(z) + (z - a)h'(z) Z=(l
f(a)
h(a)
2f(a)
g"(a) ,
where we have used the procedure for finding the residue of the fraction of two
functions from which the function in denominator has a pole of first order.
2f(a)
Hence U_2 = -(-) .
g" a
Z2 eiz
11.3. For F(z) = (2
z + 1)2 we take
The points Zl = ~, Z2 = -~ are zeros of second order for the function g.

Since
f'(z) = ze'Z(2 + zz), g'(z) = 4z(z2 + 1),
and
g"(z) = 12z2 + 4, g"'(z) = 24z,
we obtain by 11.1. that
By the same formula we obtain
By II.2 we have for z = Z, U-2

2f(~)
= -(-)
g" Z
= -4e1 and for z = -Z, U-l = ~e/2, U-2 = e/4.
Exercise 10.16 Let a function F : IC X IC --+ IC be defined by

w
F(z,w) =-
e
- z.
W -
a) Find the branching points and on fixing one variable find the analyticity do-
main with respect to the other variable.
b) Let q(z) = 10'' ' F(z, t) dt. Find the analyticity region of q.
n
E(
00
c) Let u(z) = )2' Prove that q is an analytic continuation of u(z).

z
n +1
n=O
Specify from which region to which region.
d) Let Cn : (2n + l)11"e t., 0::; t ::; 211". Find
X(z) = r
JC n
F(l,w) dw
w(w - z)
(z is inside the region bounded Cn). Find n_oo
lim X(n).
e) Prove that
w 2w 2
+E
00
F(1,w)=1--
2 2 4 22 '
n=l W + n 11"
For which w does the preceding equality hold?
Hints.a) Use the analyticity regions of the elementary functions involvecl.

b) Use Theorem 10.4.
c) Write the function F(z, t) for a fixecl t as power series with respect to z.
cl) Apply Residue Theorem.
Example 10.17 I Let the function f be represented by apower senes f(z) =

E anz
00
n with the radius of convergence R = 1, and let the function g be given
n=O
by
I a) Prove that g is an entire function.

I b) Prove that for Izl < 1
1 00
e-tg(zt) dt = f(z),
(apply the partial integration n times and then let n _ +00).

II Let the region G be obtained in the following way: through each singular point
v of the function f take a normal straight line on Ov. G is the convex region which
contains Izl < 1, and with the boundary consisting of described straight lines.
II a) Prove that for z E G, t ;::: 0 :
1
g(zt) = -2
nJc.
r f(u)ezt/u du,
u
where Cz is the circle u = z/2 + (lzl/2 + b)e,q, 0:5 q< 211" (b> 0 enough smalI).
II b) For z E G the integral 1 00

e-tg(zt) dt converges.
III For
zn Z _
L- + L (-)
00 00 Z n
f(z) = and fl(Z) = 1I"z/4 -ln2/2 /4,
n=l n n=l 1- z
prove that they analytically extend each other. Specify from which region to which
region. Give the function which cover all analytic extensions of the function J.
L
00
Solution. I a) Since f(z) = anz n is apower series with radius of convergence

n=O
R = 1 we have
1
RJ = -1
lim lan+11- .
n~oo an
L
00
Since g(z) = a~ zn we obtain by R J = 1

n=O n.
111
an+l l'1m -an+11 1-I
Rg = ----;;---- = ----,..,.-------,.---.....-- = 00.
- lim - -
lim (n + I)! n-+oo an n-+oo (n + 1)
n~oo an
n!
Hence the function 9 is analytic in the whole complex plane, 9 is an entire
function.
I b) To prove that for Izl < 1 we have fooo e-tg(zt) dt = J(z) we introduce a
1
function F by
F(z) = 00
e-tg(zt) dt.
Applying the partial integration on F we obtain
F(z) = -e-tg(zt)l: + z l'XJ e-tg'(zt) dt.

Applying again the partial integration we obtain
F(z) = -e-tg(zt)l: + z( - e-tg'(zt)l: +z 100

g"(zt)e- t dt).
We can conclude that after applying the partial integration n times we shall obtain
Prove this by mathematical induction.

Since
1) ... (p - k + 1)
9
( k)
(z) = I: a p(p -
00
p , zp
-k
p=k p.
we obtain for z = 0
(k)(O) _ akk! _
9 - k! - ak·
Therefore
We shall prove that the second summand on the right hand side tends to zero as
n -+ 00. Namely, we have
Since
from laHm I < M for m E N we obtain
Since Izl < 1, the right hand side of the preceding inequality tends to zero as n -+ 00.
Therefore for n -+ 00 we obtain F = f.
II a) The circle Cz : u = z/2 + (lzl/2 + b) eq·, 0 ~ q < 27r, Z E G, and small
enough b > 0, completely belongs to convex region G since its center is at the point
z/2 with a radius Izl/2 + b, Figure 10.6.
Figure 10.6
Hence the function J is analytic in the disc bounded by C•. So we have
J(z) = E anzn
00
.
wüh an
1
= -2
1I"Z
1J(u) du.
~
GUn
n=O
Therefore
where we have interchanged the order of integration and the series, since the series
(z E G, t 2::: 0) uniformly converges on Cz •

11 b) We shall prove that 1 e-tg(zt)dt
00
converges for z E G and t 2::: o.
Since
max Re - (z) = -I-IIzl-b = S < 1,

uEe. U Z +
we obtain by 11 a)
Ig(zt)1 < ~ f /f(u)ezt/u/ldul < Ae st ,

271" Je. lul
where A is areal constant.

10
Remark. The function F(z) = 00 e-tg(zt)dt analytically extends the function
f on the whole region G. This is the so called Borel method of extension. So we
can apply this method on the following power series a) E:=ozn; b) E:=oz2n; c)
E:=oz4n j (given regions Gon Figures 10.7, 10.8 and 10.9, respectively). Find the
corresponding functions 9 and F.
Figure 10.7
-1
Figure 10.8
Figure 10.9
III
n
E ~ the radius of convergence is R =
00
For the function 1 given by I(z) = 1.

n=1 n
We shall find the function represented by this power series.
Differentiating the given power series and using its uniform convergence for Izl <
1, we obtain
j'(z) = f
n=1
nzn-1 =
n
f: zk = _1_.
k=O 1- z
Integrating we obtain
I(z) = -ln(l - z).
We represent now the function z f--t In(l - z) by its power series at u = z
-ln(l-z) -ln(l-z-(z-z))
z-z
= -ln(l-z)(l-l_z)
-ln(l - z) -ln (1 _ z - Z)
t .!.(z - zr
1- z
= -ln V2e-,n/4 +
n=1 n 1- z
7rZ
- - In 2/2 + L.J - - -
~ 1 (z - Z) n
4 n=1 n 1- z
11 (z).
The new power series converges for
Iz - zl < 11 - zl = V2,
in the disc with center at z, and radius .J2, Figure 10.10.
Figure 10.10
We have in the region
0= {z Ilzl < 1, Iz - zl < 11 - zl}

that
fez) = ft(z) = -ln(1 - z).
Hence the functions fand ft analytically extend each other.
All possible analytic extensions of the function f from the region Izl < 1 on
the whole complex plane out of the cut from 1 to 00 are given by the function
z I--t -ln(! - z).
Example 10.18 Let C be a straight Une Re z = a > 0 and cp is an analytic function

on the half - plane Re z :::; a, except at finite number of poles and essential singu-
larities ab a2, ... , an which are in Re z < a. If cp( z) --. 0 as z --. 00 in the region
Re z :::; a, prove that
where C is oriented from below to above.

z-a
2. Let F(z, t) = ezt In --b' t > O.
z-
2. a) Find the singular points of the function F.
2. b) Find the value of the integral
Jc +-1 d z
e zt Inz -
z -1 '
t> 0,
where C is the straight line Re z = a > 1, which is oriented from below to above.
z-a
2. c) Find the Laurent series in the neighborhood of 00 of In --b and specify
z-
the region where this representation is meaningful.
Solution. 1. The straight line C is exchanged by the path Cl given by

Re e., 11" /2 :::; () :::; 31r /2, part of the straight line Re z for Re z = a for - R :::;
Im z :::; Rand the part of the straight line from x + Rz to x - Rz (0 :::; x :::; a) (Figure
10.11).
Figure 10.11
Then by the theorem on residues (for enough big R) we have
We have to prove now only that the integral on KR : Res. ('Ir /2 :::; () :::; 3'Ir /2) tends
to zero as R -+ 00, as weH as the integral on AB and CD. By
and the fact that <p( z) -+ 0 as z -+ 00 in the region Re z :::; a, easily follows the
convergence of the above integral to zero. Analogously there follows the convergence
to zero of the integrals on AB and CD.
2. a) We shall find the singular points of the function
z-a
F(z, t) = ezt In --b' t > o.
z-
The points z = a and z = bare the branching points of infinite order. The point
z = 00 is an essential singularity.
2. b) Using 1. and 2. a) and the path C' from Figure 10.12
c' t
o~ oe x
Figure 10.12
we obtain for R -+ 00
-1 1
27rz c'
z +-
ezt 1n - 1 dz = Res (zt
Z - 1
z +-
e 1n - 1)
z - 1 z=<Xl
.
Letting R -+ 00, and using
z +_
Res ( eztln _ 1) e - e-
-
t 2. h
- - = -sln
t
t
t,
t
z - 1 z=<Xl
we obtain
-1 1e
27rZ c'
zt z +-
In - 1 dz = - 1
Z - 1 27rz
1<>+·00 ezt In -
0-'00
z +-1 dz =
Z - 1
-2 sinh t
t '
for t > O.
2. c) We shall expand in the neighborhood of the point z 00 the branch
z-a
In --b in a Laurent series. P '
uttmg z = -1 we 0 b '
tam
z- w
z-a
In-- In(z - a) -ln(z - b)
z-b
In(l - aw) -ln(l - bw)
(aw)n 00 (bwt
-2:-+2:-,
<Xl
n n
n=l n=l
where we have used the power se ries representation of In(l - z) at z = 0 (jzj < 1).
Taking w = ~ we obtain
z
Since jawj < 1 and jbwj < 1, we have jzj > max(jaj, jbl).
Remark. The integral of the function J on the straight line Re z = a
a+. la+.oo
i~~ la-.b J(z) dz =
b
a-.oo J(z) dz
is the so called Brounwich-Wagner integral. The considered integral
(given by some additional suppositions on <p) is the inverse Laplace transform of

So we have obtained in 2. b) for the function <p( z) = In z + 1 that its inverse
<po
z-l
Laplace transform is J(t) = 2 sinh t/t, t > O.
Example 10.19 Let

[CO dt a>0
Ja ts(t - x)' - ,
where X is a real parameter, s a complex number and t S = es/nt.

1. a) Investigate the convergence of the given integral with respect to a, X and s.
1. b) For a > 0, x < a prove that the given integral defines an analytic function
in region 0 and find this region O.
2. Let
(a)k
Gs =
()
'Ir
Ixl s sin 'lrS
+a- s
E
co 1
---
k- s k '
a> 0, x< -a.
Find the region of the analicity of the function G(s).

3. Prove that for s > 0 different from integer (for parameters a > 0 and x < -a)
we have F(s) = G(s).
Hints. Use the integral
[ dz
Je zs(z - x)'
where the path C is given on the Figure 10.13 and let R --+ 00.
Figure 10.13
4. Using the principle of analytic continuation find the region where F = G and
the region from wh ich the function G extends the function F. Find the region of this
extension.
5. Prove that for lxi< a, a > 0, and sE 0
= 1 x
F(s)=a- s 2:-
k -(-).
k=O+ a S
k
Solution. 1. a) We shall consider the following cases.

Ia > O.
I (i) x < a, and s = m + zn. Then
IJa[= ts(tdt- x) I ~ Ja[= tmltdt- xl

and
1 1
--,----,- '" - - as t ~ 00
tmlt - xl t mH '
and so the integral converges for m + 1 > 1, i.e., Re s = m > O.
I (ii) x = a. In this case the integral diverges, since
[= dt [b [=
Ja t s(t - a) = Ja + A '
where the first integral on the right side diverges.
I (iii) x > a. In that case we have
[= dt [b [=
Ja ts(t - x) = Ja + Jb '
where a < x < b, and therefore the first integral on the right side diverges for every
sEC.
II a = O.
II (i) x < O. We have
1=ts(t -
o
dt
x)
= 1+ 1=
0
b
b
,b > 0,
The first integral on the right side converges for Re s < 1 and the second integral
for Res> 0 (as in I d)). Therefore the integral converges for 0< Res< 1.
II (ii) x = O. Then the integral
[00 dt [b [=
Jo tsH = Jo + Jb ' b > 0,
diverges since we can not find such sEC for which both integrals on the right side
would converge.
II (iii) x > o. Then
{CO dt {b t (CO
Jo ts(t - x) = Jo + A + Je '
where for b < x < c, diverges for every sEC.
1. b) Let a > 0, x < a and F(s) = {CO s(dt ). To find the region 0 of the
Jo t t - x
analicity of the function F we shall prove that the function under integral g(s, t) =
(1 ) has the following properties.
ts t - x
(i) g(s, t) is continuous with respect to for every sEC, since t- S = e- slnt . The
function g(s,t) is continuous for every tE [a,oo) since x < a ~ t.
(ii) The function g(s, t o) for t o E [a,oo) is a analytic function for every sEC,
since toS = e- slnto is analytic.
(iii) The family Fq(s) = Jar t s(dt

x t-
) uniformly converges on every closed and
bounded subset A of the half-plane Re s > 0, since for s = m + zn
IFq(s)I ~ [ tm(td~ x) ~ JtxO(:~ x)'

where Xo = min Re s. The function under integral on the right side has the following
seA
behavior
1 1
xo l (q -+ 00).
txo(t-x) '" t +
Hence Res ~ Xo > o.
Then (i), (ii), (iii) by Theorem 10.4 implies that the function F is analytic for
Res> o.
2. We shall find the region of the analyticity of the function
1 (a) k
G(s) = Ix I·
'Ir
SSln'lrS
ECO -k-
+ a -8 k=l -
- S X
for a > 0, x< -a.
The first summand on the right side is an analytic function in the whole complex
plane without integer points s = 0, ±l, ±2, ... for which sin 'Ir s = O. The second
summand is an analytic function on C \ {I, 2, ... }, since on every closed and bounded
subset A of the given region by
f: _l_(~)k,
the series
k=lk- s x
for a > 0, x< -a, uniformly converges, since 1;1 < l.

3. We have
[dz [R dt [ dz
(10.4)
Je zs(z - x) = Ja ts(t - x) + JKR zs(z - x)
By Residue Theorem
1zs(zdz- x)
C
= 27rzRes ( 1)
Z8(Z -x) z=", -
27rZ
lxises", ' smce x< -a, a> O.
Letting R -+ 00 we obtain (where x < -a, a > 0)
IJ[2" Re· dO 9 I
o R e, (Re· s s9 9 - x)
< 1 r" dO
Rs-l Jo R - lxi
2
(for s > 0).
RS-l(R + x)
Therefore by (10.4) letting R -+ 00 we obtain
(10.5)
Then
[ dz [ dz
JJzJ=a zs(z - x) JJzJ=a zs( -x)(l - z/x)
tZJ=a C8(~X) E(;)) dz
--
11 00 zk-.
E-dz
x JzJ=a k=O x k
= ~
- L....ti
k=O x
1
k+l
1 JzJ=a
Z
k-SdZ,
where the interchange of the integral and series is allowed because of the uniform
convergence of the series for Izl < lxi.
Therefore
r dz
11z1=" zs(z - x)
Putting this in (10.5) we have
F(s)(l - e- 2 '11"") = ~ + (1- e- 2'11"'")a- s _l_(~)k.

Ixl s e''1I"s k=l k - s k
f:
Hence by
e'l7rB _ e - I lI"S
Sin 7rS =
2z
we obtain
1 (a)k
F s
()
=x S
7r
sin 7rS
+ a- s E - - -
k- s X
00
•
k=l
4. The function F(s) is an analytic function on the whole half-plane Res> 0

(see 1. b)) for a > 0 and x < a. The function G (s) is analytic on the whole complex
plane without integers. The equality F(s) = G(s) holds in the region
1= {s I s > 0, s =I- 0,±1,±2, ... }
Therefore the analytic function G( s) analytically extends the function F( s) through

the region I on the whole complex plane without integers for a > 0 and x < -a.
5. We shall start from the same integral and path as in 3. Therefore (10.4) holds.
Since lxi< a, a > 0 in the region bounded by C there are no singular points of the
function under the integral. Hence by Cauchy's theorem
1CZS(z-x) dz
=0.
Letting R - t 00, the integral on the path KR tends to zero (as in 3.) and we obtain
F(s)(l - e- 2'11"1S) = r dz (s E 0).

1Iz l=" zs(z - x)
We have for the integral on the right side
r dz r dz
1Izl=" zs(z - x) = 1Izl=" zS+I(l -;)
where we have exchanged the order of the integration and series using the uniform
. smce
convergence senes . -lxi < 1.
a
Finally we have
00 1 x k
F(s)=a-'L- k -(-) ,
k=O +s a
for lxi< 1, a> 0 and s E O.
00 n
= 1.../3/2 x cot 7rX dx
+ -7r cot 7r (2 - vt;:;)
3 .
: ; (n 2 - 3)V4n2 - 3 0 (3 - x 2 h/3 - 4x 2 6
2. Let
and
J(z) = L n.(z
00
,-
(l)n
+ 100 e-te- 1 dt.
n=O + n) 0
Prove that J analytically extend rand find Jrom which region on wh ich region. Find
the singular points oJ the Junction J and the corresponding residues.
7r
3. Find the image oJ the region 0 ~ Re z ~ a, by the Junction w( z) = tan 2 4a z.
Solution. 1. To prove the desired equality we shall take the integral of the
function
7rzcot7rZ
J(z) = (3 _ z2)v'3 _ 4z2
on the path L given on Figure 10.14.
-R -2
-Ri
I
Figure 10.14
The function J has the branching points of second order at V3/2 and V3/2, and
we take the cut [-V3/2, V3/2]. The function J has poles of first order at points
±V3,±1,±2, ... We have
iJ(z)dz = j-V3/2+E2
V3/2- 7rX cot 7rX dx
-;:-----;:~~~ +
El i J( z) dz
(3 - x2)V3 - 4x 2 Kl
+ jO 7rX cot 7rX dx +[ J(z) dz + [ J(z) dz

V3/2-El (3 - x 2)e"'v'3 - 4x 2 JI 1 JK
+ 12J(z)dz+ la -V3/2+E2
1 0
7rxctg 7rX dx
(3 - x 2 )e"'V3 - 4x 2
27rZ( :E
~-~
Res(J(z))z=k + Res(J(z))z=V3 + Res(J(z))Z=_V3)'
.
It is easy to prove that for R ~ 00 we have fK J(z) dz ~ 0, for Cl ~ 0 we have
r J(z) dz ~ 0, and for

k
C2 ~ 0 we have
~
r J(z) dz ~ O. Integrals kr J(z)dz and
r J(z)dz together give zero.
Jl 2
Since J is an even function we have
00 00 n
'" Res(J(z)) -k = -2z ' " --;:-~-r=~===
n~oo n- ~ (n 2 - 3)J4n 2 - 3'
and therefore
4
o
V3/ 2
i 7rX cot 7rX
J
(3-x 2 ) 3-4x 2
dx = 27rz (00
- 2z 2:
n=1(n 2 -34n 2 -3
n
)J -
7r
6 cot(7rV3) ,
2z-
)
smce
-R R x
Figure 10.15
Remark. We can obtain the same equality using the same function j and the
path C from Figure 10.15. 2. See Examples 10.1 and 10.2. So it is easy to obtain that
j(z) = f(z) for Rez > 0 and j(z) analytically extend f(z) on the whole complex
plane without points 0, -1, -2, ... , which are poles of first order of the function f.
The corresponding residues are given in the following way (k E N U {O} )
L: (-l)n(z
00 + k) + lim(z+k) 100 e- t e- dt
Res(f(z))z=-k = lim l
z-+-k n =1 (n + z)n! z-+-k 1
1. ~ (-l)n(z+k) (_l)k
= 1m L."
z-+-k n=I,njk n.'( n + z) + -k'-
.
( _l)k
kl'
3. We have
2 7r 2
( e1rzt / 4a _ e- 1rZl / 4a ) 2
w(z)=tan -z=(-z) 2'
4a (e1rZ./4a _ e-1rZ./4a)
Taking the sequence of transformations
w1 =~zz=~e1r·/2z·
4a 4a w - eW1 ,• w3 -_(w)2.
,2 - 2,
1
W4 = W3 + 1j Ws = -j
W4
Ws = 2zwsj
Figure 10.16
o o
Figure 10.17
we obtain the final image (Figures 10.16, 10.17, 10.18, 10.19, 10.20).
/-
t! ........
I/ II "
Figure 10.18
Figure 10.19
iI w ~v
7
o u
-1.
Figure 10.20
In this way the region 0 ::; Re z ::; a, Figure 10.16 (left), is mapped by the
function w(z) = tan 2 :a z on the unit disc Iwl ::; 1, Figure 20 (right).
Example 10.21 The closed regions D n , n = 0,1, ... , are given by
Dn = {z = x+zy I lxi::; n+ 1/2, Iyl::; n + 1/2}.

1. Prove that
1. a) Isin 2 1l"(x + zy)1 = sin 2 1l"x + sinh 2 1l"yj

1. b) for lxi = n + 1/2 we have Isin 1l"ZI > 1/2 e1l"IYl j
1. c) for lyl = n + 1/2 we have I sin 1l"zl ~ 1/4e1l"(n+t).
Hints. Use that et > 2e- t for t ~ 1l" /2.
2. Let f be an entire function. Prove that for any arbitrary but fixed n E N
[ !(z) dz
JD SIn 1l"Z
= 2z t
k=-n
(-ll f(k)
(D n is the border of D n ).
3. Let the entire function f satisfies the condition
e-aIY1Izf(z)1 < M (U)
for every z = x + zy and 0 S a < 1l", where M is areal constant. Prove that
n
lim
n--++oo
L (-l)kf(k) = O.
k=-n
4. Let g(t) be a complex function of real variable on the interval [-a, a] for which
there exists
Prave that the function

f(z) = iaa e,tzg(t) dt
is an entire function which satisfies (U).
Solution. 1. a) See Chapter 3.

1. b) By 1. a) we have for lxi = n + 1/2
Isin2 1l"(n + 1/2 + zy)1 = sin 2 1l"(n + 1/2) + sinh 2 1l"Y = 1 + sinh2 1l"Y = cosh 2 1l"Y.
Since cosh 1l"Y > 1/2 e1l"IYI, the preceding equality implies the desired inequality.
1. c) By 1. a) we have for lyl = n + 1/2
Isin 1l"zl ~ sinh 1l"(n + 1/2).
Since
et
sinh t > 4' i.e., et > 2e- t
for t ~ 'Ir /2, we have

sinh 'Ir(n + 1/2) ~ e,,(n+l/2) /4.
The preceding two inequalities imply the desired inequality.
2. The desired equality follows by Residue Theorem and
Res (f(Z)) = (-l)kf(k)

Sin 'Ir z z=k 'Ir
for k = O,±1,±2, ....

3. The path D n is divided on finite sides. On sides parallel to x - axis (one of
them we denote by D~) we have lyl = n + 1/2 and Izl ~ n + 1/2, and therefore
Ifv~ s{~;z dzl < fv~ I~~;!lldzl

< 4e- 7r (n+!) [lf(z)lldzl
JD~
< 4e-,,(n+!)Mea(n+~) [ Idzl
lD~ Izi
< 4Me(a-1r)(n + ~)2(n + ~)
n+!2
where we have used 1. b) and (U). Letting n -+ +00 the right side of the preceding
inequality tends to zero since a < 'Ir.
On sides parallel to y - axis (one of them we denote by D~) we have lxi = n+1/2
and Izl ~ n + 1/2, and therefore
[ .ß:ldzl
IlD~ sin 'lrZ
<
<
('Ir - a)(n + ~)'
where we have used 1. a) and (U). Letting n -+ +00 the right side of the preceding
inequality tends to zero.
The preceding considerations imply
[ f(z) dz -+ 0 as n -+ +00.
lDn sm 'lrZ
Therefore by the equality from 2. we obtain the desired result.
4. The function f is an entire function by Theorem 10.4.
We shall prove that the function J satisfies the condition (U). Applying the
partial integration on the integral which defines J we obtain
and then
IzJ(z)l::; eaIY1(lg(a)1 + Ig(-a)1 + iaa Ig'(t)ldt).
Taking
M = Ig(a)1 + Ig( -a)1 + iaa 19'(t)1 dt,
we obtain that the function J satisfies the condition (U).
L anz
00
Example 10.22 Let J(z) n be an analytic Junction in the disc Izi <
n=O
R, R > 0, and let
F(z) = f
n=O
a~ zn.
n.
I 1. Let M(r) = max IJ(z)l, 0< r < R.

Izl=r
I 1. a) Prove that F is an entire Junction.
I 1. b) Prove that Jor every real number r, 0 < r < R,
IF(z)1 ::; M(r)e 1zl / r .
12. Let Kr be a circle Izl = r, 0 < r < R, oriented positively. Prove that
F(z) = _1 [ J(u)e z / u du.

27l'Z lKr U
L
00
II Let G(z) = bnz n be an entire Junction which satisfies the inequality

n=O
IG(z)1 ::; Bi 1zl
(B and k are real constants).

II 1. Prove that for Izl = r
II 2. Using II 1. prove that the series
1
converyes jor k > Iz!-
II 3. Taking that jor n E N
prove that
1
jor Re - > k.
z
Solution. I 1. a) See Example 10.17 I a).

Second method. The Cauchy inequality implies
(10.6)
1
2: ,anzn converges absolutely for every z
00
Therefore the series and therefore the
n=O n.
sum is an entire function.
11. b) Using (10.6) we obtain
lF(z)1 :s M(r) f ~(l:lr = M(r)e 1zl /r •

n=O n. r
I 2. Residue Theorem implies
_1
2:n
r j(u)ez/u du = Res (! j(u)ez/u) u=o .
JKr U U
The desired residue is the zero coefficient in the Laurent expansion of the function
j(u)e z/ u with respect to u
and this is just the expansion of the function F.

II 1. The desired inequality follows by the Cauehy inequality applyied to the

function Fand eircle Kr.
II By II 1. we have
Ibnl ~ Br-ne kr = hn(r).
We want to find the minimum of hn(r) (the preceding inequality holds for 0 ~ r <
+00). The minimum is at r = n/k and it is
Then
Ibnl ~ min hn(r). (10.7)
The series (by D'Alembert's eriterion
L
00
n!n-nknenz n
n=O
L n!bnzn also eonverges

00
eonverges for Izl < l/k, and beeause of (10.7) the series
n=O
for Izl < l/k.
II 3. By the uniform eonvergenee of the series whieh represents G on [0, 1] we
have
Sinee
/'0
we obtain that the integral
e-x/zG(x) dx
uniformly eonverges on the set whieh satisfies Re (1/ z) ~ k + e for e > O.

The series
uniformly eonverges for Re (l/z) ~ k + e and e > O.

Therefore, letting L _ +00 we obtain
for Re (1/ z) > k, Le., on dise x 2 + y2 < x/k whieh is eontained in the dise Izl < l/k.
Example 10.23 The function F is given by F(z) = {OO s~nhzt dt.

10
sm t
a) Find the region where F is a regular analytic function.
b) For z real find the function F integrating the function elZU / sinh u on the path
C given on Figure 10.21 and letting that Cl and C tend to zero, and R tends to
infinity.
t
x
Figure 10.21
c) Extend F maximally and find the singular points of this analytic extension U.
d) Map the region Rez > 0, 0< Imz < 1/2 by the function U.
Solution. a) We shall find the region 0 C C where the function

{OO sin zt d
F( z) = 10 sinh t t
is analytic.
(i) The functions~nhzt is continuous on C x L, where L is the whole real line

sm t
(Imz = 0). We have for t = 0
sinzt
z--
· sin zt I· sin zt I·
I1m - - = Im -.-- = Im ~-z
t_O sinh t t_O sm zt t_O sin zt - .
z --zr-
(ii) The function sin zt is analytic with respect to z E C for every tEL, and
t h erelOre
r
t h e functIon
. sin zt.
-:--h . IS a1so anaIytIc.
·
sm t
s~nhzt dt} converges in A(O) where

(iii) We shall show that the family { {q
sm t 10 q
0: -1 < Imz < 1. Namely, on a closed bounded sub set Fe 0 always exist
max Imz = Yo, minImz = Y1.
zEF zEF
Then
zt I etz, - e- tz , I
Sin
Isinh t I et _ e- t
letz'l + le-tz'l
<
et - 1
since e- t :<:; 1 for t ~ O.
Further we have for z =x + zy
s~n zt I !ext'e-ytl + le-xt'eytl

Ismht <
et - 1
e-Y1t + eyot
<
e t -1
for -1 < YI :<:; Yo < 1.
b) Let z be areal number. Then by the theorem on residues
e'ZU
i ( e'ZU )
-.-- du = 21l'zRes -.--
c smh u smh u U=1I"'
.
e'ZU ) 1
Since z = 7rZ is pole of first order we have Res ( -'-h- - - . Further, we
e1l"Z
SIn u u=1rt
have
i e'ZU
--du j -q etzX
--dx+
etzU
--du
1
c sinh u -R sinh x K«,) sinh u
+ l
R e'ZX
-.-- dx + .
1
211" e'Z(R + zy)
z dy + .
e,z(x+211"')
dx
1<
<I smhx 0 smh(R+zy) R smh(x+ 21l'z)
[ e lZu j-R etz (x+21r') [0 e,z(-R+,y)
Z dy.
+ JK«) sinh u du + -E sinh(x + 21l'z) dx + J21r sinh( -R + zy)
The residues at the poles z = 0 and z = 21l'Z of the first order give us
lim [ ~ du = -1l'zRes (~) = -1l'Z

<1 .... 0 JK(.,) sinh u sinh u u=o '
lim [ ~ du = -1l'zRes (~)

JK«) sinh u
< .... 0 sinh u u=211"1 - e21rZ '
respectively.
Letting R ~ 0 we obtain
[211" e,z(R+,y) I
IJa sinh(R + zy) z dy IeR+,y 2 e- R-. y I
2e- zY eR
e2R _ 1 dy
Analogously we have
1
o
211"
.
e,z(-R+,y)
smh( -R + zy)
z dy ~ 0 as R ~ 00.
Putting the obtained results in the equality where we have represented the integral
on C by the integral on parts of C and letting 15 ~ 0, Cl ~ 0 and R ~ 00, we
obtain
VP 1
00
-00
eZ'"
- - dx
sinh x
1-
e,z(x+2"')
+ V p .smh(
00
x + 211"z)
00
dx
(1)
= 7l"Z 1 + -
e2 11"Z
27l"Z
-- ,
e11"Z
( 1 - - 1) V P 1 00
- - dx
e lZX (e11"Z _1)2
= 1I"Z ~---::----''-
e21rZ -00 sinh x e
21rZ
Since e'zx = cos zx + Z sin zx, we have
100
-00
sinzx
--dx =
sinhx
11" - - -
e'll"Z
e11"Z
+1 = 11" tanh -2 .
1I"Z
The last integral is without V P since we have proved in a) the existence of this
integral for every real z.
Since the function under the integral is even we have
F(z) = [0 00 sinzx dx = ~1°O sinzx dx = ~ tanh 1I"Z

in sinhx 2 -00 sinhx 2 2
c) By a), the function F(z) = 10 sm

s~nhzt
t
[00 dt is analytic for IImzl < 1. The
function
11" 1I"Z
U(z) = '2 tanh 2"""
is an analytic function on the whole complex plane except at the points
z = (1 + 2k)z, k = 0, ±1, ±2, ....
Since F(z) = U(z) for z real, functions Fand U are equal on real axis, the function
U is an analytic extension of the function Fon the whole complex plane excluding
the points z = (1 + 2k)z, k = 0, ±1, ±2, .... .
d) Now we shall apply the function
11" 1I"Z 11" e11"Z - 1

w = U(z) = '2 tanh 2""" = '2 ell"Z +1
on the region Rez > 0, 0< Imz < 1/2, Figure 10.22.
Figure 10.22
Let Wl = e1fZ = e1fZ: e'1fY. Applying it on zo, where x> 0, 0< y < 1/2, we obtain
in Wr plane the point with radius p = e1fX 2:: 1 and 0 < arg Wl = 7rY < 1/2, Figure
10.23.
Figure 10.23
We have
7r Wl - 1 7r ( 2)
W = "2 Wl + 1 ="2 1- Wl +1 .
We take that W2 = Wl + 1, and W3 = ~, then W4 = 1 - 2W3. Finally W =~ W4.
i, will be mapped onto the

W2 2
Then the quarter of the circle: p = 1 and 0 < () <
following lines see Figures 10.24, 10.25, 10.26.
----------~/~/_/_/~~/--~--_\~~~---~------------~
-1' \ 0 /
,I U,
~
" ....... ./ /
Figure 10.24
/' --- "-

I
I t \\
\ Q
\. _1- /
'- Z /
---
Figure 10.25
o o u
Figure 10.26
The part of real axis Ul > 1 is mapped on the part of real axis 0 1 is mapped in the w -plane on
'Ir
w--·-----·--+---
ZVl - 1 'Ir vf - 1 t'lr 2Vl
- 2 ZVl +1 - 2 v~ +1 2 v~ + 1.
Since w = U + zv we have
'Ir v? - 1 'lrVl
U = 2" v? + l' v = v~ + 1 .
Hence u 2 +v 2 = 'Ir 2 /4 for U > 0 and v > O. This means that the part oft he imaginary
axis Vl > 1 is mapped on the quarter of the circle p = 'Ir /2, 0 < arg w < 'Ir /2.
Since the point Wl = 00 is mapped on the point w = 0, we conclude that the
desired region in the w-plane is the quarter of the disc:
w = pes., 0 < P < 'Ir /2, 0 < () < 'Ir /2.

w
11"
2
o u
Figure 10.27
Chapter 11
Miscellaneous Examples
Example 11.1 Give.an example of a continuous, closed curve which does not in-
tersect itselj in a bounded region and whose length is not finite.
Solution. We start from a tri angle with equal sides ßABC, Figure 11.1 (left)
whose one side length is 1 and which lies in the circle Izl = V;.
E
Figure 11.1
Dividing each side on three equal part we construct equal sided triangles (see
Figures 11.1 and 11.2). Taking out the sides DF,FG,GJ,KM,MD, we obtain a
313

314 CHAPTER 11. MISCELLANEOUS EXAMPLES
closed curve.
Figure 11.2
Continuing this procedure ad infinitum we obtain a continuous clos~d, not inter-

secting itself in a bounded region, in the disc Izl $ R, where R is given by
R =
Since the length of this curve is given by
1
E 3 . 22n 3n '
00
L = 3+1+
n=l
and this series diverges, we conclude that the length of this curve is not finite.
Example 11.2 Find the region on which the function w(z) = Z2 maps the region
bounded by straight lines x = 1, Y = 1 and x + y = 1, Figure 11.3.
315
Figure 11.3
Solution. Since w = z2 is equivalent with
u + w = (x + zy)2 = x 2 _ y2 + 2zxy,
we obtain u = x 2 - y2, V = 2xy. Then the following straight lines are mapped onto
the corresponding parabolas
v
2
X = 1 on u = 1 - y2, V = 2y, u = 1 - 4'
v2
Y = 1 on u = x2 - 1, v = 2x, U = - - 1,
4
x +Y = 1 on u = x 2 - (1 - x)2 = 2x - 1, v = 2x(1 - x) = 2x - x 2 , i.e.,

v = ~(1 - u 2 ), Figure 11.4.
Figure 11.4
Remark that the map considered is conformal.
L
00
Example 11.3 If laml < 00, find

m=-oo
00 00
Solution. Since L lam1 = (J < 00, there exists L am = S.

m=-oo m==-oo
Let
1
21 L
00
Cn = lam- n + ... + am+nl·

n + m=-oo
lim Cn = ISI. Let e > O. Then there exists a natural number

We shall prove that n-+oo
M such that ~ laml < e. Then for n :2: M we have
Iml>M
(2n + 1)Cn = L lam- n + ... + am+nl

Iml>n+M
317
+
n+M~lml>n-M
+ L lam- n + ... + am+nl·

Iml~n-M
We have
L la m - n + ... + am+nl < L (Iam-nl + ... + lam+nl)

Iml>n+M Iml>n+M
< (2n+1) L laml
Iml>M
< (2n + 1)5
and
(1 :::; 4M (1.
n+M~lml>n-M n+M~lml>n-M
Since Iml:::; n-M implies m-n:::; -M:::; M:::; m+n, we havefor Iml :::; n-M
/Iam- n + ... +am+nl - ISI/:::; L laml < 5,
Iml~M
and so
1
L la m - n + ... + am+nl- (2n - 2M + l)'SII
Iml~n-M
< L /Iam- n + ... + am+nl-ISII

Iml~n-M
< (2n-2M+1)5.
Therefore
/(2n + l)Cn - (2n - 2M + l)ISI\ :::; (2n + 1)5 + 4M(1 + (2n - 2M + 1)5.
Dividing by (2n + 1) and taking n - t 00, we obtain

lim sup 1Cn
n-oo
- ISII:::; c: + c: = 2c:.
Since c: > 0 was arbitrary we have n_oo

lim Cn = ISI.
Exarnple 11.4 Find the integral,,;

Izl=l 4z 2
1 dz
+ 4z + 3
starting with the positive value
of the square raot at the point z = 1.
Solution. The zeros of the equation 4z 2 + 4z + 3 = 0 are Zl,2 = -1/2 ± zV2/2,

where IZl,21 < 1.
Taking a cut ZlZ2, we can take in the complex plane outside of this cutting an
analytic part the function
1 1
J(z) = V4z2 + 4z + 3
2J(z + 1/2 - zV2/2) (z + 1/2 + zV2/2)'
The condition that we are starting with positive value of the square root at the
point z = 1 means that J(l) > O.
Therefore
So we can take for z = 1 that

arg(z - zd = -(Y and arg(z - Z2) = (Y,
Figure 11.5. When the point z moves upto any position on the right part of the cut
without crossing the cut arg(z - zd continuously changes to -7r /2, and arg(z - Z2)
to 7r/2.
y
Z1
.
--- J:;--
-
1
---- ----
-1 -f 0 ..-- -----
..-- -----
l~_
Zt
Figure 11.5
319
Therefore on the right part of the cut arg((z - Zl)(Z - Z2)) = O.

We have (independence of the path from Figure 11.5)
f dz
Jlzl=l J 4z 2 + 4z + 3
where
I2Z je; rel' dt

-f vreh(zl - Z2 - reh)
I
1 ~ rdt
< 2"1!§ VrVl z z21- r 1 -
=
7rVr
~ 0, as r ~ O.
Analogously we can prove that
(/21 = I.: f~ rehdt 1~ 0 as r ~ O.

2Jf jreh(z2-z1+ret') ,
Taking r ~ 0 we finally obtain
f dz j:f. dy
Jlzl=l J4z 2 + 4z + 3 = Z -:f. j~ _ y2 = 1rl.
Example 11.5 Find the value 01 the integral I: e-s2/2elSZ ds.
Solution. We have to find the function
F(z) = 1 00
-00
2
e- s /2 elSZ ds.
First we shall find the corresponding real function for z = t real, and then with an
analytical continuation we shall obtain F for z E (C. We shall prove that
(11.1)
where L is the straight line u = s - zt (-00 < S < +00). From u = s - zt we have
s = u + zt and so
u2 + 2iut - t2
=----
2 2 2
i
Therefore
vP e-t2/2-u2/2 du
e- t2 / 2 V P i e- u2 / 2 du.
We can remove V P, since jetSt-s2 /21 = e- s2 /2, and the integral of this function abso-
lutely converges. Since e- u2 / 2 is an entire function its integral on a closed path C is
zero
fc e- u2 / 2 du = o.
We take the path C as in Figure 11.6.
Figure 11.6
l
We have
1 s -. t
e- U 2 /2 du +
S
e- U 2 /2 du
1'-'t
S-tt
1
-s-~t
s
__ se- S 2/ 2ds+ -s e- U 2/ 2du
o.
321
We have
The last integral is bounded and therefore
lim
Isl ....oo
11
s-d
s
e- u2 / 2 dul = 0,
the second and fourth integrals are zero as Isl --t 00. So we have
[ e- u / 2 du = 1-00
00
e-·/ 2 ds.
1:
JL
The last integral is e- s / 2 ds = ."fi;, and therefore by (11.1)
F(t) = ."fi;e- t2 / 2 •
With an analytical continuation on the whole complex plane we obtain
F(z) = .j2;e-z2 / 2 (z E C).
Remark. The integral transform
cfl(z) = ~1°O f(t)e lZt dt

v27r -00
is the Fourier transform of the functions (see [17]). The case f(t) = e- t2 / 2 is impor-
tant in probability theory. In this way we obtain the characteristic function cp(t)
= ~F(t) for the normal distribution N(O, 1) :
v 27r
za
Exercise 11.6 Let w(z) = - - , a E ~.
1+z
a) At which points and what kinds of singularities does this function depending
on a?
b) For a = 1 prove that this function is a conformal mapping on the whole
complex plane except at z = -1. What is the imageof the part of the unit disc from
the first quadrant under the given function '?
c) For 0 < a < 1 expand one part of the given function in apower senes in a
neighborhood of the point z = -2.
d) For -1< a < 0, using the function w(z), prove that
[00 x a 7r
Ja 1 + x dx = - sin a7r .
Exercise 11.7 Let VI - Z2 be the branch of this function which is positive on the
upper part of the cut [-1,1].
a) Prove that
r dz r dz
JK JI=Z2 = Je JI=Z2'
where K is the circle with the center at (0,0) and radius 2, and C is the path on
Figure 11.7.
-1 o
Figure 11.7
b) Using a) prove that r~

JK
1- z
= -211".
Exercise 11.8 Let f(z) 1+~

= v'ZLn 1- z
a) Expand the function f with respect to powers of z for Izl < rand Izl > r.
Find for which z this is possible.
b) Find the singular points of f.
c) Let h(z) = In 11 +v'Z.

_ v'Z' where y'Z zs the
.
branch for whzch vI
11
= 1. .
Map wzth h
the disc Izi < 1.
Exercise 11.9 Let

323
a) Find the region of analyticity of G.

b) Find the analytic form of G.
Example 11.10 The function f(z) = u(x,y) + w(x,y) is analytic at the point
Zo = Xo + lYo and f(zo) = Co. Prove that
f(Z)=2U(z~zO, z;/o)_Co.
Solution. We have
=L
00
f(z) an(z - zo)n.

n=O
Hence
Co+Co
+L + l(Y -
00
u(X,y) -2- (an((x - xo) yo)t
n=1
+an((x - xo) - l(Y - yoW)·

Putting
w - Zo w-zo
x = xo+ --2-' Y=Yo+-2-
and taking w enough dose to z we obtain
w - Zo w - zo)
u ( xo+ --2-' Yo + -2l- = (Co + f(w))/2.
Exchanging w with z we obtain the desired equality.
II Pn
00
Exercise 11.11 Prove that for -'Ir < argPn ::::; 'Ir, the infinite product either
n=1
L
(Xl
converges or diverges simultaneously with the series lnpn.

n=1
Hints.The infinite product
II Pn
00
PI . P2 ... Pn ... = (11.2)

n=1
converges if there exists a finite limit
IIp; =
00
lirn
n ..... oo
p
;=1
different frorn zero.

If P = 0 and no factor Pn is zero, then we say that (11.2) diverges to zero Cn, in
the opposite case we call it convergent to zero.
Exercise 11.12 Prove that a necessary and a sufficient condition lor the absolute
n (1 + E ln(1 + u
00 00
convergence 01 U n ), i.e., absolute convergence 01 the series n ), is

n=l n=l
Eu
00
that the series n is absolutely convergent.

n=l
Exercise 11.13 Suppose that n

00
n=l
Un and n
00
n=l
Vn converge.
Examine the convergence 01 the lollowing products:
a) n (u
00
n=l
n + vn)j b)
c) n (u
00
n=l
n . vn)j
a) Divergesj b) Diverges (to zero);

Answer.
c) Converges; d) Converges.
Example 11.14 Let Pl>P2,'" ,Pn,'" be a sequence 01 all prime numbers and
1
E---;
00
(s) =
n=l n
the Riemann zeta lunction (which is analytic in Re s > 1). Prove that
1
a) ((s) = 00 j
I1 (1 - p~")
n=l
b) The Iunction (( s) has no zeros in the half-plane Re s > 1.
Solution. a) Substracting from the series

1
(s) = 1 + -
2"
+ -3"1 + ...
the series 2- 0 • (s), we obtain
(1- 2- 0 )(s) = 1 + ~ + ~ + ~ + ....

3" 5" 7"
In the last sum there are missing the term displaystyle ;. with n which are divisible
with 2.
Analogously we obtain
325
where in the last surn thcre are rnissing the membcrs ~

n~
with n which are divisible
with 2 or 3. Generally we have
(1- pt")(l - P2"S) . '" • (1 - p;;'S)(s) = 1 + t~,

n
where in the surn E' we are surnming through indices n (greater than one) which
are not divisible with PIJP2, ... ,Pm' It is easy to prove that for Res> 1 +~, ~ >
1
0, E' - -+ 0 as m -+ 00, and therefore
n"
n (1 - p;;'S) = 1.
00
(s)
m=l
b) Since the series f

m=:I P
1.. converges for Re s > 1 + 5, 5 > 0 we have that the
product
m=1
n (1 - p;;'S)
00
also converges. Therefore ( s) has DO zeroes in Re s > 1.
Example 11.15 Prove that the series f:~, where {Pn} is the sequence 01 alt prim
n=1 Pn
numbers, diverges.
Solution. By Exarnple 11.14 we have for every 5 > 0
n (1 _ Pn
°o
n=1
-(1+6» _
- (1
1
+ 8)"
Therefore
lirn n(1 - p~(1+Ö»
00
6-+0 n=1
== O.
Since (1- p;;l) < (1- p;(I+ö», the product E:=l (1- p;;l) diverges. Hence the series
1
L-
00
also diverges .
n=1 Pn
Example 11.16 Let I be a bounded analytic lunction in the unit disc and 1(0) i= o.
11 {an} is the sequence 01 zeros 01 f in an open disc where the zeros appear in the
sequence so many times as is their order, then
n lanl converges, i.e., L(1 - lanD<

00 00
00.
n=1 n=1
Solution. For simplicity we suppose that I/(z)1 ~ 1, for Izl ~ 1.
In the function I has only finite number of zeros then the statement is trivially
true. In the opposite I has countable many zeros al, a2, a3, . .. . Let B n(z) be the
finite product
II t -_a kz .
n
Bn(z) =
k=lak -
The function Bn(z) is a rational function, analytic in Izl ~ 1 and IBn(eo')1 = 1, since
every function z - _a k has modulus one on the circle Izl = 1. Therefore BI((z)) is a
1- akZ n Z
bounded analytic function in Izl < 1.
Since
Specially
n
0< 1/(0)1 ~ IBn(O)1 = II lakl
k=l
Since lakl < 1 for k = 1,2, ... and every partial product n lak I is not less than
n
k=l
1/(0)1, the infinite product converges.
Remark. The analytic function
B(z)=zPll (an. an-=z)PR,

n=l lanl 1 - anz
where:
(i) p, PI, P2, ... , are non-negative integers,
(ii) an are different non-zero numbers from the unit disc ,
n lanl PR converges,
00
iii) the product
n=l
is the Blaschke product.
Example 11.17 Let f be a bounded analyticallunction in the unit disco Prove that
I can be in a unique way represented as I(z) = B(z)g(z) where B is the Blaschke
product, 9 a bounded analyticallunction without zeroes.
Solution. Since fez) "I- 0, we have I(z) = zPh(z), where h(O) "I- O. Let B be the
product zP with the Blaschke product of zeros {an} of h.
Then g( z) = I (z) / B( z) (see the preceding example) is an analytic and bounded
function in the disco The representation I(z) = B(z)g(z) is unique since the
Blaschke product is uniquely determined by the zeros of the function.
327
Exercise 11.18 Find the integral J n = ~zl=2 1 ~ z el/z dz, n = 0,1,2, ... , and find
A = lim J 2n
n~~
and B = n-+oo
lim J 2n - I .
Hints.The function 1 ~ z el/z in the disc Izl :::; 2 has a pole of first order at z = -1
and an essential singularity at z = O. Use Residue Theorem.
Solution.
1 oe (_l)k) n (_l)k
J n =2n(-lt ( - - : L - , - = 27rx(-lt:L-,-,
e k=n+1 k. k=Q k.
27rX · J 2n
11m 27rX
lim J 2n
n-+oo
= -e and
n-oo
= --.
e
Exercise 11.19 Function F( z) is defined by the following integral
F(z) = (oe sinzt dt.

10 e27rt - 1
a) Find the region of analytieity of the function F.

eUU
b) For z real find F integrating the function 2 on the rectangle with ver-
e 7rU_1
tiees 0, R, R + x, x, and letting R --+ 00.
e) Analytieally extend the function F as for as it is possible and examine the

singularities of the analytieal extension .
d) Map the region Rez,O < Im < 7r/2 by the function w = (z) + 1/2z.
Hints.a) Prove that for Im z :::; 27r - ö,O < Ö < 27r, and big enough t :
Iesinzt- 1 I<-
27rt e
-M/ 2.
Therefore F is analytic in the region IImzl < 27r.

b) Take the path in Figure 11.8 and apply Cauchy theorem.
R~I
~~------------------~
t
o R x
Figure 11.8
Then let R --+ 00 and r --+ o. So we obtain
1 + e- Z 1
F(z) = 4(1 _ e- z ) - 2z·
1 + e- z 1 .
c) The function (z) = () - -2 analybcally extend F(z) on the whole
4 1- e- Z z
complex plane without the points Zk = 2k7n, k = ±1, ±2, ... , where it has poles of
first order and z = 00 is an essential singularity. d) The region Rez > 0, 0<
Imz< i, Figure 11.9 (left) is mapped by (z) + 1/2z onto the region in Figure 11.9
329
(right).
z w
Figure 11.9
Exercise 11.20 Find the real integral
("/2
I = 10 cos a x cos dx dx (b> a > -1).
Hints.Integrate the function J(z) = (Z2 + l)a z b-l-a along the path in Figure 11.10
( section at [-1, 1]) and let r ---t O.
Answer.
f(a+ l)f(~)
1= sin~(b-a).
2f(1 + 1 +~) 2
o 1 x
-L
Figure 11.10
Answer.
r(a + 1)r(~)
1= sin:::'(b-a).
2r(1 + 1 +~) 2
Exercise 11.21 Let

(Z2 + 3)(e a (1-Z)/2(I+z) - 1)
F(z,a) = ,
(z - 1F{jz(z + 1F
(a is a complex number).
a) Find the singularities oJ this multivalued Junction.
b) Choose the branch J(z,a) oJ the multivalued Junction F(z,a) which at the
point z = 2 takes the value
J(2,a) = m18
7(e- a / 6 - 1)
and find the value J(z -l,a).

c) Let
w(a) = [ J(z, a) dz, R>2.
Jlz+1I=R
Find the region oJ analyticity oJ the Junction w( a).
331
d) Prove that in the region Iz + 11 > 2 the function f has the following represen-
tation
00 a
f(z,a) = ~ (z+n 1)n'
What is the cut? Find the coefficient al.
e) Find the analytical form of w(a).
j) Map the region bounded by curves x = O,y = O,xy = f, with the junction
w(a 2 ).
Example 11.22 Let LI and L 2 be two closed paths with a common path L. Let j
be a continuous junction on LI and L 2 and in regions bounded by LI and L 2 , where
it is also analytical. Prove that then j is analytical also on L.
Solution. Let 0 be the union of regions bounded by LI and L 2 • Then we have
!cj(z)dz = 0
for any closed path contained in O. Namely, if C is completely in a region bounded

either by LI or L 2 , then the preceding equality foliows by Cauchy's theorem 5.4. In
the opposite case, Figure 11.1,
Figure 11.11
we have by the generalization of Cauchy's theorem 5.5 that
1c, j(z)dz = 0 and 1 j(z) dz

C2
= 0
Therefore
1c j(z)dz=l C,
j(z)dz+l j(z)dz=O.
C2
Then by Morera's theorem f is analytic in O. Since L c 0 we obtain that j is
analytic also on L.
Exercise 11.23 Find
where an are the maximum points 0/ the function sin x 2 / x 2 at (-00, +00).
Hint. Take for the path P the square of side 2'Ir R centered at the origin and
integrated on this path the function
zsinz
(1 + z2)(sinz - zcosz)"
Then let R --+ 00.
Exercise 11.24 (Prime Number Theorem) Let 'Ir(N) denote the number 0/ pri-
mes less that 01' equal to N. P1'Ove that
. 'Ir ( N) log N
11m
N_oo
N = 1,
i.e., 'Ir(N) rv N/logN.
Hint. Let
1 z- 2 )
g(z) = N ( + R2 + 1 . Z
:;;
Integrate the function

1
L(z) = L -;/(z)
pprime P
on the path P consisting of C the right semi-circle of radius R with center at
z = 1; D the continuation of C for 1 - [j < Re z < 1; E the vertical segment
Re z = 1 - [j from Im z = J R2 - [j2 to the x-axis; F the segments from 1 - [j to 1
and the circle around z = 1 clockwise; G : the refIection of E across the x-axis; H :
the reflection of D across the x-axis. Use Cauchy's theorem and let R --+ 00. Prove
that
1a
L(z)/(z) dz = 2'1rz'Ir(N).
Use the estimations
" 1 < e log 4x r 1

L.J - - ( _ 1) x-I lor x > ;
p prime,p>N P" X N log N
and
L ~ = 0 (Nl-x~Og(2 + lxI) + N(1-x)!210g N 10g(2 + lXI))
p prime,p<N P" (1 x) log N
for x < 1, where 0 is the small "0" limit sign.
Bibliography
[1] Ahlfors, L., Complex Analysis, McGraw-Hill Book Co., New York, 1966.
[2] Bak, J., Newman, D.J., Complex Analysis, Springer-Verlag, New York, 1997.
[3] Carathoodory, C., Theory of Functions of a Complex Variable, Vol II, Chelsea
Publishing Co., New York, 1954.
[4] Conway, J., Functions of One Complex Variable, Vol. 1,11, Springer-Verlag,
New York, 1995.
[5] Courant, R., Hilbert, D., Methods of Mathematical Physics I, II, Interscience
Publishers, 1953, 1962.
[6] Evgrafov, M.A., Problem Book on Analytical Functions (in Russian), Moscow,
1973.
[7] Feyel, D., De la Pradelle, A., Exercices sur les fonctions analytiques, Arman
Colin, Paris, 1973.
[8] Fuchs, W.H.J., Topics in the Theory of Functions of One Complex Variable,
D. Van Nostrand Co., 1967.
[9] Hörmander, L., The Analysis of Linear Partial Differential Operators -I-IV,
Springer-Verlag, Berlin, 1983-1985.
[10] John, F., Partial Differential Equations, Springer-Verlag, 1982.
[11] Ladyzenskaya, O. A., Uraltseva N. N., Linear and Quasilinear Elliptic Equa-
tions, Academic Press, New York, 1968.
[12] Lakshmikantham, V., Leela, S., Differential and Integral Inequalities,· Aca-
demic Press, New York, 1969.
[13] Lang, S., Complex Analysis, Springer-Verlag, New York, 1993.
[14] Lelong, J., Ferrand, Problemes d'analyse Cl, Paris, 1967.
333
334 BIBLIOGRAPHY
[15] Markushevich, A. 1., Theory 0/ Functions 0/ a Complex Variable, Prentice-

Hall, Englewood Cliffs, N.J., 1965.
[16] Polya, G., Szegö, G., Problems and Theorems in Analysis (2 vols), Springer-
Verlag (vol. I, 1972; vol. 2, 1976).
(17] Pap, E., TakaCi, A., TakaCi, Dj., Partial Differential Equations through Exam-
pies and Exercises, Kluwer Academic Publishers, Dordrecht, 1997.
[18] Rubinstein, Z., A Course in Ordinary and Partial Differential Equations, Aca-
demic Press, 1969.
[19] Rudin, W., Principles 0/ Mathematical Analysis, McGraw-Hill Book Co., New
York, 1964.
[20] Rudin, W., Real and Complex Analysis, McGraw-Hill Book Co., New York,
1966.
[21] Schmeelk, J., TakaCi, Dj., TakaCi, A., Elementary Analysis through Examples
and Exercises, Kluwer Academic Publishers, Dordrecht 1995.
[22] Sobolev, S. 1., The Equations 0/ the Mathematical Physics, Nauka, Moscow,
1966 (in Russian).
[23] Strang G., Introduction to Applied Mathematics, Wellesley-Cambridge Press,
1992.
[24] Taylor M.E., Partial Differential Equations I,II, III, Springer-Verlag, 1996.
[25] Titchmarsh, E.C., The Theory 0/ Functions, Oxford University Press, London,
1939.
[26] Tikhonov A., Samarski A. A., Equations 0/ Mathematical Physics, Pergamon

Press, New York, 1963.
[27] Vvedensky, D., Partial Differential Equations with Mathematica, Addison-

Wesley, 1993.
[28] Vladimirov, V., S.,Equations 0/ Mathematical Physics, Nauka, Moscow, 1976.

[29] Volkovyskii, L.I., Lunts, G.L., Aramanovich, I.G., Problem Book on Functions
0/ a Complex Variable (in Russian), Moscow, 1961.
List of Symbols
set of natural numbers

set of integers
set of real numbers
n-dimensional real Euclidean space
set of complex numbers
.;=I imaginary unit
z conjugate of the complex number z
Rez real part of a complex number z
Imz imaginary part of a complex number z
Izl absolute value of the complex number z
argz argument of the complex number z
AC complement of the set A
XA characteristic function of the set A
n intersection of sets
u union of sets
\ set difference
ao border of the region 0
Q closure of the region Q
liminf limit inferior
limsup limit superior
Logz multivalued complex logarithm
logz principal value of the complex logarithm
Res(f(z))z=zo residue of a function f at Zo
335
Index
Absolute value 1 compact set 32
analytic continuation 227 complex
along a path 228 number 1
direct 227 conformal mapping 73
analytic function 54 conjugate 1
anharmonic relation 87 connected set 32
argument 2 curves
Argument Principle 192 smoothlyequivalent 103
aXlS cutting 68
imaginary 1
real 1 DeMoivre
formula 2
Bernoulli numbers 141
entire function 54
Bessel function 260
beta function 258 Fibonacci numbers 145
bilinear transformation 75 function
Blaschke product 326 analytic 54
bounded set 32 conformal 73
branch of the logarithm 68 continuous 53
branching point 68 differentiable 53
algebraic 68 entire 54
transcedental 68 meromorphic 172
Brounwich-Wagner integral 288
Gamma function 256
Cauchy
Generalized Integral Formula 192 Hadamard
inequality 130 example 63
integral formula 129 Three Circle Theorem 159
sequence 37 Integral
theorem 104 Fresnel123
theorem generalization 104 line 103
Cauchy-Riemann equations 54,60 isolated singularity 171
closed path
regular 192 Laplace
closed set 32 equation 60
336
INDEX 337
transform 263 sequence

Laurent expansion 178 accumulation point 37
limit 53 bounded 37
Liouville Cauchy 37
theorem 130 convergent 37
logarithm series
principal value 68 absolutely convergent 45
convergent 44
Maximum Modulus Theorem 130 power 130
Mellin transform 258 set
rnerornorphic function 172 accurnulation point 32
metric 32 singularity 227
Mittag-LefHer essential 171
theorem 179 pole 171
Mittag- LefHer removable 171
theorem 189 stereographic projection 32
Montel
theorem 160 Tauber
Morera theorem 155
theorem 130 Taylor expansion
for analytic function 130
Natural boundary 227 theorem
general residue 191
Open mapping theorem 130
residue 192
open set 32
Uniqueness theorem 130
Path 53
Poisson Vitali
formula 154 theorem 159
power series 130
Prime Number Theorem 332 Well-posed problem 63
Wright function 261
Region 32
regular point 227
residue 191
Riemann
Mapping Theorem 73, 160
zeta function 267
Rouche
Theorem 192
Schwarz
lemma 130, 157
reßection principle 236

Endre Pap Auth - Complex Analysis Through Exambookzz-Org PDF

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Endre Pap Auth - Complex Analysis Through Exambookzz-Org PDF

Enviado por

Direitos autorais:

Formatos disponíveis

Complex Analysis through Examples and Exercises

Kluwer Text in the Mathematical Sciences

A Graduate-Level Book Series

SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

ISBN 978-90-481-5253-7 ISBN 978-94-017-1106-7 (eBook)

Printed an acid-free paper

AII Rights Reserved

1 The Complex Numbers 1

2 Sequences and series 37

3.2 Special Functions .. 64

5 The Integral 103

6 The Analytic functions 129

7 Isolated Singularities 171

7.2.2 Examples and Exercises . . . . . . . . . . . . . . . . . . . . . 179

9 Analytic continuation 227

10 Integral transforms 255

11 Miscellaneous Examples 313

List of Symbols 335

Novi Sad, June 1998 ENDRE PAP

The Complex N umbers

1.1 Algebraic Properties

(a, b)(c, d) = (ac - bd, bc + ad).

Iz + wl s:; Izl + Iwl (z,w E C).

By the definition of complex numbers, each z in C can be identified with a unique

E. Pap, Complex Analysis through Examples and Exercises

The argument of z i- 0, denoted by arg z, is the angle 0 (modulo 271") between

In particular, if z = r( cos 0 + z sin 0), then

zn = rn(cos(nO) + zsin(nO». (1.1 )

As a special case of (1.1) we obtain DeMoivre's formula:

(cos 0 + z sin Ot = cos nO + z sin nO. (1.2)

The n-th root of z = r( cos 0 + zsin 0) are

Zk = y'r(cos o+n2h + zsm

for k = 0,1, ... , n - 1.

For more explanations see the chapter on power series.

1.1.2 Examples and Exercises

Solution. We have that z = w is equivalent with Rez = Rew and Imz =

Solution. Let z = 1 + z = (1,1) and w = x + zy = (x, y). Then we have

Finally we have w = x(l - z) for x E llt and x i= 0.

Example 1.3 Prove that

Example 1.4 Find for z = 1 + 2z the following numbers

cl) Putting z = 1 + 2z in Z2 + 2z + 5 + z, we obtain 4 + 9z.

Solution. Using the trigonometrie representation (p,O) wc obtain

d) Imz 2 0; e) Izl :::; 2; f) 1 < Izl < 3;

Figure 1.1 Rez = Imz

of this straight line), Figure 1.2.

Figure 1.2 Rez < 1

Figure 1.3 -1 :::; Rez :::; 1

d) Im x ~ 0 means y ~ 0 and x is an arbitrary real number. The desired subset

Figure 1.4 Imz ~ °

Figure 1.5 JzJ ~ 2

Figure 1.6 1 < Izl < 3

Figure 1.7 Izl > 2

of the real axis, Figure 1.8.

Figure 1.8 -'Fr < arg z < 'Fr

Figure 1.9 'Fr /6 < arg z < 'Fr /4

Solution. We denote the vertices as:

If we denote z = x + zy, then Rez = x and Imz = y and

-Y = tan (), () E [1C'

Therefore Im z = Re z . tan (). Since the function tan () is monotone increasing we

Example 1.8 Prove that tor every z E C

Izl ::::; IRezl + IImzl ::::; -12 ·14

Izl = v'Re 2z + Im 2 z::::; y'(IRezl + IImzl)2 = IRezl + IImzl

We have for alt three vertices Pj = 1 and