Escolar Documentos
Profissional Documentos
Cultura Documentos
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
Contents v
Preface IX
3 Complex functions 53
3.1 General Properties 53
3.1.1 Preliminaries 53
3.1.2 Examples and Exercises 54
vi CONTENTS
4 Conformal mappings 73
4.1 Basics ....... 73
4.1.1 Preliminaries 73
4.1.2 Examples and Exercises 73
4.2 Special mappings .. 74
4.2.1 Preliminaries 74
4.2.2 Examples and Exercises 75
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
Bibliography 333
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.
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)
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
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
z = r exp( zO).
1
z = p + zq, W = P + z- be equal.
q
1.1. ALGEBRAIC PROPERTIES 3
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) ~.
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.
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.
Hint. It is easy to prove the case n = 2 and then use mathematical induction
to prove the general case.
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) :
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?
a - za V2a . -
v'2 - zV2a-
V2
2 2
(v'2a , -~)
4'
-a + za (V2a, 3:),
-a - za (V2a, 3:).
1.1. ALGEBRAIC PROPERTIES 5
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;
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).
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
6 CHAPTER 1. THE COMPLEX NUMBERS
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
is the upper half plane with respect to the x- axis, Figure 1.4.
2 X
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
8 CHAPTER 1. THE COMPLEX NUMBERS
annulus between the circles Izl = 1 and Izl = 3 without these circles, Figure 1.6.
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.
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
1.1. ALGEBRAIC PROPERTIES 9
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.
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.
10 CHAPTER 1. THE COMPLEX NUMBERS
Zl = 0, Z2 = 1 + zV3, Z3 = 1 + z/V3.
O=z 1 x
Figure 1.10
Together with the condition 0 ::::; Re z ::::; 1 we have completely described the points
in the given triangle with vertices Zt, Z2 and Z3.
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
Exercise 1.10 Find the real and imaginary parts, modulus, argument and the eom-
plex conjugate for the following numbers:
e) 1
1 + 1VJ;
1) (3 + 13)(1 + 1VJ); g) C +11VJ f; h)
1+ 1
1- 1
Answers.
would be equal.
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.
1.1. ALGEBRAIC PROPERTIES 13
C2 : b 2' a2 - : b2) E T.
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.
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
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 ..
Example 1.16 Starting with a complex number Z i- 0, find where the complex
numbers 2z,3z, ... ,nz are?
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.
I1 - 2~
cos 5" - . 5"
z sm
2~1
2~ . 2~
= (1 - cos 5" )2 + (sm 5")2
= )2(1 _ 2; ) cos
.
2sm 5 ,
~
Figure 1.13
We obtain analogously IZ2 - z31 2 sin 2;, Figure 1.13. Find the length of
diagonals.
Iz + zol = 1;
~
a) Iz-zol=l; b) c) arg (z . zo) = 4'
16 CHAPTER 1. THE COMPLEX NUMBERS
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.
Answers. a) The closed disc with center at (1,0) and radius r = 2, Figure 1.16.
--~~~~~~r--'
-1 X
b) Wh oIe complex plane without the disc with center -2 and radius r = 1, Figure
1.17.
c) The disc with center Zo, and radius r (without the circle Iz - zol = r), Figure
18 CHAPTER 1. THE COMPLEX NUMBERS
1.18.
y
/
/
./"---
//
/// //
/' /./
./ /./
---------<-/'-/-O~-ij--/~--·--·-·-x- .
Figure 1.19 y = x-I, Y2 0
r
-I
Figure 1.20
1.1. ALGEBRAIC PROPERTIES 19
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
Solution. Since u . u = lul 2 we can easily obtain the desired equality. Narnely,
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
1.1. ALGEBRAIC PROPERTIES 21
o
Figure 1.22
f"Y
r:/Z'
.
I /
.' ,/
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".
22 CHAPTER 1. THE COMPLEX NUMBERS
Figure 1.24
a) 0; b) 0=1; c) ~; d) N.
Solution. a) We have
{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
for k = 0,1,2,3,4.
c) We have
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 ~.
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.
Hence
Zl = 1, Z3 =
1
-+z-.
v'3
2 2
24 CHAPTER 1. THE COMPLEX NUMBERS
571"
571" +2h -+2h
~ (cos T 6 + z sin 4 6 ),
i + zsin i
Since
cos E {z Ilzl = I},
cos ~ =
2
J cos z + 1 .
2
Applying the last formula two times on cos ~ we obtain
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
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.
26 CHAPTER 1. THE COMPLEX NUMBERS
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).
Since Zl and Z2 are known we can find Z3 by the last equality (we obtain two solu-
tions).
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 1.33 Find the vertices of regular n- polygon if its center is at Z = 0 and
one vertex is known.
Solution. a) The length of si des of the triangle with vertices Zb Z2 and Z3 are
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
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.
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.
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.
where
:~ -:: :: 1
W3 Wo -Wl
-W2 Wl Wo
Rint. For c). Use a) and b) in the deeomposition of the following matrix
1 Zl Zl
1 Z2 Z2
1 Z3 Z3
is areal number.
1.2.1 Preliminaries
We denote by D(zo, r) the open disc centered at Zo with radius r,
is ametrie on !C.
a) {zllzl<3};
b) {zl1 < Rez < 3};
c) {zIRez<3}U{3}?
Example 1.49 Give an example lor a connected set which is not polygonally con-
nected.
is connected hut not polygonally connected, since the set A contains no straight line
segments.
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.
a) {zllz-11<1}U{zllz-31<1};
34 CHAPTER 1. THE COMPLEX NUMBERS
b) {-~lnEN}U(0,11?
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
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
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'
(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) •
2.1 Sequences
2.1.1 Preliminaries
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
37
b) {C ~zf};
c) {G::f}; d) {2n:1 +zn:1};
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
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.
a) l,-l,z,-lj b) Oj
c) 1,-1,z,-zj d) ~+Zj
e) O.
Example 2.5 Prave that ifthe sequence {w n } converges to w, then {Iwnl} converges
Iwl. The opposite is not true.
Since u = (u - v) + v we have
lul ~ lu - vi + lvi·
Hence
lul-Ivl ~ lu - vi.
40 CHAPTER 2. SEQUENCES AND SERIES
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.
lim Iwnl = r
n-+oo
(r > 0) lim argw n = cp
and n-+oo
imply
lim Wn = re"",
n-+oo
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
lim arg
n .... oo
Wn = arg wo?
k k
c) n-+oo "w
lim L...J 1'
n "w
= L...J 1' •
1'=0 1'=0
"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
42 CHAPTER 2. SEQUENCES AND SERIES
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) and b) are complex generalizations of the equality
a) For n E N we have
b) For z E C we have
Putting z = z we obtain
1 + x f(x + 1) = f2(x),
is meaningful, since
= 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·
Z~ - 2 = al J+ 2 a2,/2 + ... + an V2 ,
and on other side by (2.2)
Z~ - 2
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
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.
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
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
46 CHAPTER 2. SEQUENCES AND SERIES
as k ----+ 00.
f) and g). Diverges absolutely, but converges ordinary.
h) Diverges.
converges.
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
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
48 CHAPTER 2. SEQUENCES AND SERIES
since the sequence {an} converges and therefore it is a Cauchy sequence. Hence by
(2.3) the series
is convergent.
converges and Iarg ukl :::; <P < ~, then this series absolutely converges.
L Un and LU!
n=l n=l
be convergent series. Prove the for Re U n ;::: 0 the series
00
converges.
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
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
Example 2.22 11 one series converges and the other diverges can their sum con-
verges?
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
Exercise 2.25 Prove D'Alembert and Cauchy criterions for the absolute conver-
gence of aseries.
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.
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
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
Hence it does not converge to zero what is a necessary condition for the convergence
of aseries.
Solution. a) Since
~
~ (~+2kl)k = ~
~ (~)
2 ,ancl I~I
2 < 1,
k=O k=O
Therefore by a)
k=2
1)
2
k-2 = -4 + 4l.
d) Since
see Chapter 6.
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.
Complex functions
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.
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).
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
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
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).
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
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
56 CHAPTER 3. COMPLEX FUNCTIONS
b)
V;t+1+,V;, 0~t~1
c) z(t) = {
v'2 v'2
T t + 1 - 'T' -1 ~ t ~ O.
3.1. GENERAL PROPERTIES 57
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 - }.
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
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:).
58 CHAPTER 3. COMPLEX FUNCTIONS
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
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
3.1. GENERAL PROPERTIES 59
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·
wh at is a contradiction.
Second method. Apply the Cauchy-Riemann equations on the previous polynomials.
60 CHAPTER 3. COMPLEX FUNCTIONS
Taking h -+ 0 we obtain
Letting h -+ 0 we obtain
Since both the real and imaginary parts of (3.2) and (3.3) must be equal, we obtain
the Cauchy-Riemann equations.
3.1. GENERAL PROPERTIES 61
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.
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.
. VIZ2_Z21 . VI4cos6sin61
hm = hm .'
r_O Z r-O (cos6 + zsm6)
62 CHAPTER 3. COMPLEX FUNCTIONS
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
Solution. a) We have
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)
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.
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.
3.1. GENERAL PROPERTIES 63
Exercise 3.18 Find the constants a, b, and c such that the following fundions would
be analytic
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 .
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:
64 CHAPTER 3. COMPLEX FUNCTIONS
(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
3.2.1 Preliminaries
The function eZ •
The function e Z is the analytic solution of the functional equation
3.2. SPECIAL FUNCTIONS 65
eZ = eX cos y + ze X sin y
(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 •
1
cos z = -( elZ + e-'Z).
2
We have (sinz)' = cosz and (sinz)' = -cosz.
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
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.
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
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.
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
Im (tanz ) = tanhy(1-tan2x) 2 ,
1 + tan 2 xtanh y
Itan zl = tan 2 x + tanh 2 y.
elZ e- IZ e lZ + e-' Z
= = ----
_
smz and cosz
2 2z
we directly verify the desired equalities.
68 CHAPTER 3. COMPLEX 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
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
Solution. In the complex plane without the positive real axis we obtain:
1 7r
a) log(l + z) = "2 log 2 + "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
and
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.
70 CHAPTER 3. COMPLEX FUNCTIONS
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.
Therefore
Fo(O) = e37r'/ 8 ;
Fo(l) = e(1n2)/8 e71n/16;
Fo ( -1) = e(1n2)/8 e 57r ./ 16 •
Find Arcsin z.
Example 3.39 Find the real and imaginary parts, and modulus of the following
functions: sin z, cos z, tan z, cot z.
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.
Exercise 3.42 Find the sets where are the branches 0/ the multi-valued /unction
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-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.
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 4.3 (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.
73
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
Exercise 4.4 Prove that the relation conformal equivalence is an equivalence rela-
tion.
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.
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.
[~1 :]
can be written in the form
2t
w=-l+--
z+t
or in the matrix form
[ ~1 :] = [~t ~ 1 ] [~ ~] [~ :].
76 CHAPTER 4. CONFORMAL MAPPINGS
o
Figure 4.1 Imz :::: 0
/ '\
I \
I u,
/
./
in the wrplane.
Then the transformation
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----------- -- - - .
z-z
w---
- z+z·
0, u
i
Figure 4.6 Iwl < 1, Imz ~ 0
"'y
!
z
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.
In this way the disc IWI - zl ~ 1 is mapped on the region Im W2 ~ -1/2 (Figure
4.9). Finally, the transformation
4.2. SPECIAL MAPPINGS 81
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.
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
o
Figure 4.15 0 < 0 < 7r/4
84 CHAPTER 4. CONFORMAL MAPPINGS
on the half-plane Im w < 0 without the disc Iw -1 + 2-1 ::; {} (Figure 4.16).
Figure 4.16
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).
86 CHAPTER 4. CONFORMAL MAPPINGS
Answer. a) w = 1/ z
Figure 4.19
4.2. SPECIAL MAPPINGS 87
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.
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
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?
88 CHAPTER 4. CONFORMAL MAPPINGS
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.
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
d) w = az + b,
cz+d I; ~ 1 =/:- 0, (a + d? = 3(ad - bc).
4.2. SPECIAL MAPPINGS 89
Answers.
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
Cl dl - al bl
C2 = d2 - a2 = b2 •
This condition implies
Exercise 4.14 Prove that J(z) = l/z maps circles and lines on cireles and lines,
respectively.
90 CHAPTER 4. CONFORMAL MAPPINGS
Example 4.15 Prove that the bilinear transformation maps cireles and lines onto
circles and lines.
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
]
,
4.2. SPECIAL MAPPINGS 91
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
92 CHAPTER 4. CONFORMAL MAPPINGS
b) The image is
y
z
c) The image is
o x
Figure 4.24
94 CHAPTER 4. CONFORMAL MAPPINGS
v w
Figure 4.25
Figure 4.26
4.2. SPECIAL MAPPINGS 95
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
96 CHAPTER 4. CONFORMAL MAPPINGS
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
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
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.
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
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
o x~
Figure 4.34 'Ir /6 < arg z < 'Ir /3
4.2. SPECIAL MAPPINGS 99
o u,
Figure 4.35
o Uz
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
So we obtain
w = f(z) =
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.
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
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.
102 CHAPTER 4. CONFORMAL MAPPINGS
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 •
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
( 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
103
Ern r
n-+oo}p fn(z)dz= r
}p f(z)dz.
i f(z) dz = O.
if(z)dz=O.
a) 1 1
e'!· cosatdt, a E lW.;
b)
j 1
_1t 2 +Z'
dt
Solution. a) We have
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
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
-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.
106 CHAPTER 5. THE INTEGRAL
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
Example 5.3 Find for which complex numbers z the integral 1e1
dt exists, and
give abound for the modulus of this integral.
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
[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
= 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
~----~-------+
o X
Figure 5.2
108 CHAPTER 5. THE INTEGRAL
o x
Figure 5.3
(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)/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 =
110 CHAPTER 5. THE INTEGRAL
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.
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
a) Therefore
5.1. BASICS 111
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
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
112 CHAPTER 5. THE INTEGRAL
< 1°IJ(w + r· kl dt
eti)re tl -
< ~ fO dt
() Jo
c,
for r < b(c), 0::::; t ::::; ().
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
< 1°IJ(R'et')R'et• - kl dt
c f9
< (j Jo dt
c,
for R' > Ra, 0 ::::; t ::::; ().
5.1. BASICS 113
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
IE J(9(Wk))(g(w~:~~ =~~Wk)
zEL
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
iL
j(z) dz = - j
L
-dz
j(z) 2"
Z
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 b
(u(t)x'(t) - v(t)y'(t)) dt + l l(v(t)x'(t) + u(t)y'(t)) dt
= [j(z)dz.
5.1. BASICS 115
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.
= 1 J(z)dz,
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.
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
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
118 CHAPTER 5. THE INTEGRAL
Figure 5.8
Hint. The given path can be replaced with the following path:
Figure 5.9.
Figure 5.9
5.1. BASICS 119
Then, we obtam
a
.1
sinx
--dx
x
00
= 'Ir /2.
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.
~----.....(I------
ki
. . . ---
I
t
I
t t
--~------~----~---~
-R 0 R X
Figure 5.10
We have
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
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·
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.
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,
-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
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
[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
= 1 e-
2"( X 2.
SIn zx) 1-00
00
- 2"z 1 00
-00 eX 2 cos zx dx
z
= -2"f(z).
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
(Fresnel integrals).
z = x, 0 ~ x ~ R;
(see Figure 5.12).
~~~------~---~
o -- R X
Figure 5.12
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.
["/4
Jo
22·
~ R le- R e t'ldt
["/4 R2
~ R Jo e- cos2t dt
Figure 5.13
Therefore
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
1
that
211" _ _ _d_t_ _-,,- _ 27r
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) '
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
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)
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".
129
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.
lor D(zo,r) C D.
Theorem 6.10 (Open mapping theorem) The image of an open set under a
nonconstant analytic mapping is an open set.
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
then this limit is the reciprocical value of the radius of convergence of the power
series a L:~=o wn(z - Wt.
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
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
n=O n
n=l
Example 6.3 Find the radius of convergence for the following power series
L00 ( w) n
n=l n z
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
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
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.
6.1. THE POWER SERIES REPRESENTATION 133
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
Example 6.7 Prove that the radiuses of the following power se ries
00 00
00 00
Solution. a) We have
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)'
134 CHAPTER 6. THE ANALYTIC FUNCTIONS
Solution. Since
L zk . L Uk zk = L L Ui zk = L Sk zk = C,
00 00 001;; 00
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
6.1. THE POWER SERIES REPRESENTATION 135
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).
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
136 CHAPTER 6. THE ANALYTIC FUNCTIONS
Slnce
1
- =L
00
zn, Izl< 1.
1- z n=O
1
Z2 - 5z + 6
L( -lt+1nz
00
= n- 1
n=O
00
= L(-lt(n + l)zn.
n=O
(2z)2n 1 00 22n 2n
t -(-)-" t
00
e) Since
zn zn
+ z) = L( -lt+1-, L -,
00 00
we obtain
logl+z = log(l+z)-log(l-z)
1-z
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
=
138 CHAPTER 6. THE ANALYTIC FUNCTIONS
1 = 1+L
e1-z
00 (
Ln 1(n - 1)) zn.
k' k _ 1
n=l k=l .
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'
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 .
(2n)!
.00
- sm v E( -1) n{z-v)2n+1
+ I)! .
(2n
log(l + z) log(l + v + w)
140 CHAPTER 6. THE ANALYTIC FUNCTIONS
= {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·
or more general
and
6.1. THE POWER SERIES REPRESENTATION 141
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.
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
142 CHAPTER 6. THE ANALYTIC FUNCTIONS
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.
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) .
Example 6.19 Let J be an analytic Junction in Izl < R (R > 1). Starting Jrom the
integral
6.1. THE POWER SERIES REPRESENTATION 143
:;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
i (2 ±
Therefore
(z + ~)) f~z) dz = (2f(0) ± j'(0))27rZ. (6.2)
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.
144 CHAPTER 6. THE ANALYTIC FUNCTIONS
. . 1
Exercise 6.21 Fznd the zntegral -2
1I'Z
1e z (
eZdz
1- z
.
)3' iJ:
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.
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.
6.1. THE POWER SERIES REPRESENTATION 145
Answer.
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).
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)
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
146 CHAPTER 6. THE ANALYTIC FUNCTIONS
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'
is convergent.
We find for the function in its apower series in the form
L
00
L L L
00 00 00
L 7(k).
k=l
Hence the radius of the convergence of this series is one.
6.1. THE POWER SERIES REPRESENTATION 147
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
and so
wk
= ew +1 = e· L -,
00
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.
1 00 1 00
+-1 ~
~
(n + 2)(n + 3) z n+3 + ... + -1 ~
~
(n +k - 1) z n+k + ... )
3! n=O 2! k! n=O k - 1
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.
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.
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 n=O
00 n
= E ECkbn-k(Z - at·
n=O k=O
6.1. THE POWER SERIES REPRESENTATION 149
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.
150 CHAPTER 6. THE ANALYTIC FUNCTIONS
Finally, we have
Z Z ~ B 2k 2k
eZ -1 = 1 - "2 + ~ (2k)!z .
=L
00
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
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
Figure 6.1
Example 6.30 (Pringshajm theorem) Prove that if for the power series cp(z) =
L: anzn we have R
00
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.
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.
~ akzk + ~ ak zk .
k=O k=no
;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.
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)
for n = 2k
we obtain R = 1, since
lirn sup
n-+oo
ilanl = lirn
k-+oo
2k[l2~2
V~ = 1.
Since 2~2 > 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.
154 CHAPTER 6. THE ANALYTIC FUNCTIONS
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
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.
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
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.
lim-1- = _1_.
1- Z
r ..... 11 - e,D
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
156 CHAPTER 6. THE ANALYTIC FUNCTIONS
no
Lnlunl
n=l (m - no)
< m + 2m c:
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
Lklukl
k=O m c: l'lor m > M( c.
<2 )
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).
M(r) ~ limM(t)
t-+l
~ lim ~ = 1.
t-+l t
(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)
158 CHAPTER 6. THE ANALYTIC FUNCTIONS
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 ).
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
= I: anrne,nt
00
f(re lt )
n=O
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
6.1. THE POWER SERIES REPRESENTATION 159
b) the power series Ek:o akzk defines in the disc D(O, r) an analytic function
fez);
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
Hint. AppIy the Heine-Borel theorem related to compact sets and reduce to Exer-
eises 6.42.
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.
Now in the same way we extract a subsequence {fn2(i)} of Unl(i)} such that there
exists
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
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
(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
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
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
as R ........ 00, we obtain in an analogous way as before for n that j(k)(O) = 0 for k ;::: n.
164 CHAPTER 6. THE ANALYTIC FUNCTIONS
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
~
{ f(u)du (z-a)n { ~ du
Je (u - a)n(u - z) 2n Je u- z
(z - at fez)
(z - a)n
= fez).
~
{ 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)!
hz _ _1_ f(u) du
( )- 27r~ Je
{
(u - a)n(u - zr
6.2. COMPOSITE EXAMPLES 165
Further we have
_ (Iz-al)n R M
- R R-Iz - al '
where M = sup If(z)l.
zEC
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.
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
2: Un zn,
00
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)).
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
finally we obtain
=
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
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.
168 CHAPTER 6. THE ANALYTIC FUNCTIONS
L
00
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
f
k-l 00
~l 00
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.
If(z)1 $ ~~
6.2. COMPOSITE EXAMPLES 169
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
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
Then by I 1. we have
2MR 8 M
If(z)l:::; ~ = 3· ii"
34
170 CHAPTER 6. THE ANALYTIC FUNCTIONS
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
L:
00
Take j(z) = amz m (J is an analytic function on the disc Izl ~ R). The last
m=O
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
(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.
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.
lim(z - zo)l(z)
z-+zo
= 0,
then the singularity Zo is removable.
171
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
1 Z5 eZ
a) Z-Z3; c) (1 _ z)Z' d) 1 + zZ'
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'
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
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'
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
~
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}.
174 CHAPTER 7. ISOLATED SINGULARITIES
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
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 7.7 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
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.
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.
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.
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
Uk -
_ J(k)(z) _
-k-'• - - -2
1 1, J(u)
k+l du,
'lrZ L U
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.
7.2.1 Preliminaries
If both series
LZn and LZ-n
00 00
n=O n=l
178 CHAPTER 7. ISOLATED SINGULARITIES
L Zn as
00
00 00 00
-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
1 f J(z) d
an = 27l"z Je (z _ zo)n+I z
(i) If j has a removab1e singu1arity at zo, then all coefficients an, n < 0, of its
Laurent expansion about Zo are zero.
7.2. LAURENT SERIES 179
(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.
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
Solution. a) Since
1 1 1 1
Z2 - 3z +2
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
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·
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
7.2. LAURENT SERIES 183
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·
Therefore
-(-)
d P 1
dzp zz + 2 --
(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
184 CHAPTER 7. ISOLATED SINGULARITIES
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
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
n=-oo
converges is surely a singular point of the function which is represented by this series.
00 1 00
= ~U-n(z_u)n + ~ un(z-ut
=L
00
L
00
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.
186 CHAPTER 7. ISOLATED SINGULARITIES
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.
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
Solution. a) The desired annulus is the whole complex plane without zero and
z = 00;
1
=-
00
1 a 1
+ - )) .
00
~ Uk k = exp ( - ( w
k=-oo W 2 w
7.2. LAURENT SERIES 187
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
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.
(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
e) We have
00 an
e acost = ~ nt
L..J '" cos .
n=O n.
188 CHAPTER 7. ISOLATED SINGULARITIES
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
By d) we have
(k+2p)!
for n = k + 2p, p ~ 0
(Z2 + l)n dz ={ p!(k + p)!'
zn+k+l
o otherwise
-
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
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.
1
f(z) = (Z2 + 1)2
in the neighborhood of the points z = z and z = 00.
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
8.1.1 Preliminaries
191
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.
t 1e ResU(z))Z:=Zn,
je J(z) dz = 27n n:=l
where C is a closed path arround ZI, Zz, ... ,Zs'
where ZU) and PU) are the number oJ zeros (counted with their multiplicity) and
poles oJ J inside C, respectively.
_1
27rt
1J'(z) dz
e J(z)
= ZU),
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.
f (n)( Zo ) -
-
~
27rt
1z-
e
(
J(z)) +1 d Z
Zo n
r
lOr n -- 0,1,2, ....
8.1. RESIDUE THEOREM 193
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)!'
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).
Solution.
J'(z) n g'(z). . .
a) We have by -f() = - -
z z-w
+- ()'
9 z
where 9 IS a regular functlon m the
n
g'(z)
b) We have by f( ) = - -
z z-w
+- ()'
9 z
where 9 is a regular function in the
1"
o
tan(t + az) dt = { 'In for a > 0
-:rrz for a < 0,
for a E lR.
1" o
tan(t + at) dt = -1
2
1 0
2
" tan(t + at) dt.
. 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.
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.
8.1. RESIDUE THEOREM 195
Answers.
27r 27r
b) 1 _ a2; c) ~.
a)
b) By a) we have
(2n)
n
= _1 r (1 + z)2n dz.
2n Je zn+!
(2n)
n
= _1 r (1 + z)2n dz,
27rz Je zn+l
00 (2n) 1 1 00 r (1 + z)2n dz
~ n 7n = 2n ~Jo (7z)n -;.
196 CHAPTER 8. RESIDUES
1(1+Z)21
7z
<~
- 7' z E
C
.
f (2n)
n=O n 7
~ n
= .!...- f
27rZ Jlzl=l 5z - 1 -
dz Z2
= 7Res ( 1
5z - 1 - Z2
)
z=(5-01)/2·
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;
~---.----~--------~
-R ~ 0 R ~
Figure 8.1
8.1. RESIDUE THEOREM 197
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
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.
198 CHAPTER 8. RESIDUES
-R R x
Figure 8.4
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
<
<
<
8.1. RESIDUE THEOREM 199
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
2
=lim.!!...- (
~)
z->.a dz z + za
= z
-3
( 2ln a - In 2 a + -11"2 + n - 1I"Z In a) ,
4a 4
-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);
r
Jo
x sin x d tor a >
1 + a2 - 2a eos x x
o.
hi
f
-11 o
Figure 8.5
7r
Answer. - In(l
a
+ a).
-00 (ex
eax
+ 1)( e + 2)
X dx, tor 0 <
a < 2.
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)'
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, ...
J(z) = vr=zn
1
8.1. RESIDUE THEOREM 203
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".
-dz
/, J(z)dz = - /, J(Z)2'
L L z
Therefore
__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
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.
8.1. RESIDUE THEOREM 205
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.
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?
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
L akz k. Denote
00
L lakl
00
MI(r,J) = Tk
, M(r,J) = sup IJ(z)l·
k=O Izl=r
5) Using 4J prove
A..y
I _ _ _.... ~
x
--~~--~----~.-.
Q-r -~ o
o+r o
Figure 8.9
""y,
Q-r Q ..--
Figure 8.10
208 CHAPTER 8. RESIDUES
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
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
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
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
we obtain
sup If(z)1
E
00
= 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
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
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(z) = 2 E(_l)n
00
z2 _ (n7r)2'
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,
212 CHAPTER 8. RESIDUES
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
2 2z
4
= (e'X-Y _ e-,x+Y)(e'x+y - e-'x- y)
2
2 2
2
= cosh 2y - cos 2x .
8.2. COMPOSITE EXAMPLES 213
(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
214 CHAPTER 8. RESIDUES
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)
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
8.2. COMPOSITE EXAMPLES 215
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
z k~OO h(z - h) h)
(_l)k (_l)k
-E +E .
00 00
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
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)
and
Imw3 =
4ku(1 + 4v) - 4ku(1 + 4v) ~ 0, (8.6)
11 + 4vl 2 2z
8.2. COMPOSITE EXAMPLES 217
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
218 CHAPTER 8. RESIDUES
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.
-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,
220 CHAPTER 8. RESIDUES
(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
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
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 :
H(w) = H(a) + f
n=l
~~ ::n-~l (H'(a) j(n)(a)).
Iz -al
Namely, Iql < If(z)1 for z E K. Hence
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·
( 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
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
Therefore
G(w) 00 qn n
F'(w) = G(a) + ~ n! (G(a) f (a)).
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
I) 2. Prove that
II) 1. b) Prove that for the complex numbers z which satisfy the inequality
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"
where Cn is the closed region bounded by Cn. Letting n --+ 00, we obtain by I) 1.
(8.8)
8.2. COMPOSITE EXAMPLES 225
. . 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 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).
226 CHAPTER 8. RESIDUES
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.
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.
+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.
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
227
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.
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
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
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
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
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)
and
I
/' -- - "-
'\
\
I
, 0 I
-
\ X
/
"- ./
./
-I
Figure 9.2
232 CHAPTER 9. ANALYTIC CONTINUATION
Example 9.5 Do there exist functions which are analytic at z = 0, and which
satisfy the condition
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.
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.
Prove that it can not be analytically extended outside the disc Izl :::; 1.
9.2. COMPOSITE EXAMPLES 233
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
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 )
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.
zn
+ z) = L( -lt -.
00
-In(l
n=l n
Therefore
and the second in the neighborhood of the point 2 for Iz - 21 < 1, Figure 9.3,
y
Figure 9.3
smce
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
236 CHAPTER 9. ANALYTIC CONTINUATION
I = ~t . ~
J'1I"":"":2 slgn ny lOr a = tY,
vI + y-
~
1=+ ~ for a < -1,
a2 -1
L
00
1
J(z) = y'l-Z2' a 7' ±1,
(z - a) 1- Z2
-R R
Figure 9.4
We have
i J(z) dz = 27rZ Res(J(z))z=a. (9.1)
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.
238 CHAPTER 9. ANALYTIC CONTINUATION
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.
1 f(z)dz-+O
'Y3
as e-+O.
2 1 (x-a)
1
-1
dx
~ = 2'1rz
l-x 2
Res(f(z))z=a.
Since
smce
~= 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
9.2. COMPOSITE EXAMPLES 239
/ 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
240 CHAPTER 9. ANALYTIC CONTINUATION
-R R x
Figure 9.6
1= 7r Res(J(z))z==a, a> 1.
we have
1=- 7r .
y'ä2=t
9.2. COMPOSITE EXAMPLES 241
,"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
242 CHAPTER 9. ANALYTIC CONTINUATION
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
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
9.2. COMPOSITE EXAMPLES 243
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
244 CHAPTER 9. ANALYTIC CONTINUATION
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
9.2. COMPOSITE EXAMPLES 245
(x + 1 ~2r2) 2 + y2 2: (1 ~2r2)2.
We have for r = 1 the image x = 1/2.
1
2
Figure 9.13
246 CHAPTER 9. ANALYTIC CONTINUATION
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
9.2. COMPOSITE EXAMPLES 247
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
248 CHAPTER 9. ANALYTIC CONTINUATION
+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
k-l
Al = 1, A k = :LA;Ak - l + 1, k = 2,3, ....
;=1
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
-R R x
Figure 9.17
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
{ zP J( z )dz -+ 0, as c:-+ 0,
lK,
9.2. COMPOSITE EXAMPLES 251
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
1 zP J(z) dz
Ik
= -7rZ eP(lnb k+2,n)Res(J(z))z=b k
t Res(zPJ(Z))z=ak'
1, zPJ(z) dz = 27rZ k=l
L
(1 - e 27r ')V P 10 00
J(x) dx = 27rZ t
k=l
Res(zP J(Z))z=ak
m
vP 10 00
x PJ ( x) dx
m
+I: 1
[b i-< x PJ(x)(lnx+27rz)dx+ [ zPJ(z)Lnzdz.
i=1 lbi+I+< lK,
252 CHAPTER 9. ANALYTIC CONTINUATION
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)
9.2. COMPOSITE EXAMPLES 253
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
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
When w(z) > 0 for z > 0 we have the behavior of the maps in Figures 18 and 19.
y
•
Figure 9.18
254 CHAPTER 9. ANALYTIC CONTINUATION
Figure 9.19
Chapter 10
Integral transfornns
10.1.1 Preliminaries
Let 0 be a region in rc. We denote by A(O) the set of all analytic functions in O.
f(z) = t g(z, t) dt
255
J'(z) = l a
b
g(z, t) dt.
Bt
f(z) = t g(z, t) dt
r(z) = 1'XJ e- t t z- 1 dt
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.
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
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
B(p, q) = la oo P
t - 1(1 - t)q-l dt
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
respectively.
Taking the polar coordinates:
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),
10.1. ANALYTIC FUNCTIONS DEFINED BY INTEGRALS 259
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
It is given by
260 CHAPTER 10. INTEGRAL TRANSFORMS
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?
E
00
a) ez(t-l/t)/2 = Jn(z)tnj
n=-oo
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, ...
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
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).
262 CHAPTER 10. INTEGRAL TRANSFORMS
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
y
___.0-----.. E
w
A.
~ F
Figure 10.2
Remark. The Wright function has the following power series representation
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
Figure 10.3
Example 10.8 Let <p : ~ -+ c be a complex function which satisfies the inequality
1<p(t)1 < ce"'t for t > 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
L(I) = l/z.
b) 4>(x) = ['>0 e-xtektdt = {OO e(k-x)tdt = _1_ for x > k.
Jo Jo x- k
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).
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
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
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)
we have
y(t) = 1 - (1 + a + at + a 2 t2/2)e- at .
With the analytic extension we obtain the complex solution
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,
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)).
dF
F(x) = L(u(t)), L(tu(t)) =- dx' L(u'(t)) = xF(x),
1) -(n+I/2)
F(x) = C(x 2 + 1)-n-1/2 = CX 2n - 1 ( 1 + x 2 '
Since
L-1 ((2n + 2k)!) _ 2n+2k . h () m!
X2n+2k +1 - t ,wIt L t m = Xn+1'
10.1. ANALYTIC FUNCTIONS DEFINED BY INTEGRALS 267
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
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
((z) e- zlnn •
n=1 n n=1
c) Extend analytically the fundion ((z) on the whole complex plane without the
point z = 1.
268 CHAPTER 10. INTEGRAL TRANSFORMS
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.
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
where
2b) Both senes converges uniformly in every closed region 01 the form
~y
I
I
,I
Figure 10.4
represent the following two integrals by r -sine and -eosine functions given by
(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
270 CHAPTER 10. INTEGRAL TRANSFORMS
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 .
R ~ min[R(A), R(B)] = 1.
r(z) fooo e- t t z- 1 dt
1
where we have taken u = t'
Then
10.2. COMPOSITE EXAMPLES 271
and
So we obtain
L un(z) + L vn(z)
00 00
n=O Xl
L
00
= 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
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 < P < 1. We take Jour points in the complex plane A = (r,O), B =
(R,O), C == (O,R) and D = (O,r) JorR > 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
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).
10.2. COMPOSITE EXAMPLES 273
/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
274 CHAPTER 10. INTEGRAL TRANSFORMS
Hence
-y
Re F(z) = (x 2: y2)2 (x 2 - y2) cos X + 2xy sin x),
+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)
_ 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
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 < P < 1.
e>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
1 00
o
COS x 2 dx
y'2
== -r(1/2) = - -
4
,j2i
4
and
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
1(k+krr/2
1 )1f/2 cos x
--dx
yX
converges monotonically zero as k ---7 00 and the sign is alternatively changed~
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
Therefore
h(a) = g"~z), h'(a) -_ g"'(z)
6 .
Putting the obtained results in residues of F we obtain
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
10.2. COMPOSITE EXAMPLES 277
( 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
a) Find the branching points and on fixing one variable find the analyticity do-
main with respect to the other variable.
278 CHAPTER 10. INTEGRAL TRANSFORMS
b) Let q(z) = 10'' ' F(z, t) dt. Find the analyticity region of q.
n
E(
00
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"
1
g(zt) = -2
nJc.
r f(u)ezt/u du,
u
10.2. COMPOSITE EXAMPLES 279
where Cz is the circle u = z/2 + (lzl/2 + b)e,q, 0:5 q< 211" (b> 0 enough smalI).
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
L
00
1
function F by
F(z) = 00
e-tg(zt) dt.
We can conclude that after applying the partial integration n times we shall obtain
280 CHAPTER 10. INTEGRAL TRANSFORMS
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
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
10.2. COMPOSITE EXAMPLES 281
Figure 10.6
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
282 CHAPTER 10. INTEGRAL TRANSFORMS
we obtain by 11 a)
Figure 10.7
10.2. COMPOSITE EXAMPLES 283
-1
Figure 10.8
Figure 10.9
284 CHAPTER 10. INTEGRAL TRANSFORMS
III
n
E ~ the radius of convergence is R =
00
j'(z) = f
n=1
nzn-1 =
n
f: zk = _1_.
k=O 1- z
Integrating we obtain
-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).
Iz - zl < 11 - zl = V2,
10.2. COMPOSITE EXAMPLES 285
Figure 10.10
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.
Figure 10.11
10.2. COMPOSITE EXAMPLES 287
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
288 CHAPTER 10. INTEGRAL TRANSFORMS
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
.
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
(a)k
Gs =
()
'Ir
Ixl s sin 'lrS
+a- s
E
co 1
---
k- s k '
a> 0, x< -a.
Figure 10.13
290 CHAPTER 10. INTEGRAL TRANSFORMS
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
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,
10.2. COMPOSITE EXAMPLES 291
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.
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
292 CHAPTER 10. INTEGRAL TRANSFORMS
f: _l_(~)k,
the series
k=lk- s x
By Residue Theorem
1zs(zdz- x)
C
= 27rzRes ( 1)
Z8(Z -x) z=", -
27rZ
lxises", ' smce x< -a, a> O.
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)
--
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.
10.2. COMPOSITE EXAMPLES 293
Therefore
r dz
11z1=" zs(z - x)
Sin 7rS =
2z
we obtain
1 (a)k
F s
()
=x S
7r
sin 7rS
+ a- s E - - -
k- s X
00
•
k=l
1CZS(z-x) dz
=0.
Letting R - t 00, the integral on the path KR tends to zero (as in 3.) and we obtain
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
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
10.2. COMPOSITE EXAMPLES 295
-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
27rZ( :E
~-~
Res(J(z))z=k + Res(J(z))z=V3 + Res(J(z))Z=_V3)'
.
296 CHAPTER 10. INTEGRAL TRANSFORMS
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
10.2. COMPOSITE EXAMPLES 297
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)
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
298 CHAPTER 10. INTEGRAL TRANSFORMS
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
10.2. COMPOSITE EXAMPLES 299
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).
[ !(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
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
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
10.2. COMPOSITE EXAMPLES 301
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.
302 CHAPTER 10. INTEGRAL TRANSFORMS
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.
12. Let Kr be a circle Izl = r, 0 < r < R, oriented positively. Prove that
L
00
1
converyes jor k > Iz!-
II 3. Taking that jor n E N
prove that
1
jor Re - > k.
z
(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
_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
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
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
for Re (1/ z) > k, Le., on dise x 2 + y2 < x/k whieh is eontained in the dise Izl < l/k.
10.2. COMPOSITE EXAMPLES 305
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.
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
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
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
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
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
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 .
----------~/~/_/_/~~/--~--_\~~~---~------------~
-1' \ 0 /
,I U,
~
" ....... ./ /
Figure 10.24
---
Figure 10.25
310 CHAPTER 10. INTEGRAL TRANSFORMS
o o u
Figure 10.26
The part of real axis Ul > 1 is mapped on the part of real axis 0 < U < 'Ir /2 in
the w-plane
'Ir Ul -1
w = - -- = U and v = O.
2 Ul +1
The part of imaginary axis Vl > 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
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
closed curve.
Figure 11.2
R =
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
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
316 CHAPTER 11. MISCELLANEOUS EXAMPLES
Figure 11.4
L
00
00 00
Let
1
21 L
00
+
n+M~lml>n-M
We have
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
< (2n-2M+1)5.
Therefore
/(2n + l)Cn - (2n - 2M + l)ISI\ :::; (2n + 1)5 + 4M(1 + (2n - 2M + 1)5.
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
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
where
=
7rVr
~ 0, as r ~ O.
f dz j:f. dy
Jlzl=l J4z 2 + 4z + 3 = Z -:f. j~ _ y2 = 1rl.
F(z) = 1 00
-00
2
e- s /2 elSZ ds.
320 CHAPTER 11. MISCELLANEOUS EXAMPLES
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.
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
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).
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 .
322 CHAPTER 11. MISCELLANEOUS EXAMPLES
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
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
II Pn
00
Exercise 11.11 Prove that for -'Ir < argPn ::::; 'Ir, the infinite product either
n=1
L
(Xl
II Pn
00
IIp; =
00
lirn
n ..... oo
p
;=1
Exercise 11.12 Prove that a necessary and a sufficient condition lor the absolute
n (1 + E ln(1 + u
00 00
Eu
00
n=l
Un and n
00
n=l
Vn converge.
a) n (u
00
n=l
n + vn)j b)
c) n (u
00
n=l
n . vn)j
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.
n (1 - p;;'S) = 1.
00
(s)
m=l
m=1
n (1 - p;;'S)
00
Example 11.15 Prove that the series f:~, where {Pn} is the sequence 01 alt prim
n=1 Pn
numbers, diverges.
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
00.
n=1 n=1
326 CHAPTER 11. MISCELLANEOUS EXAMPLES
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
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
Hints.a) Prove that for Im z :::; 27r - ö,O < Ö < 27r, and big enough t :
Iesinzt- 1 I<-
27rt e
-M/ 2.
R~I
~~------------------~
t
o R x
Figure 11.8
1 + e- Z 1
F(z) = 4(1 _ e- z ) - 2z·
1 + e- z 1 .
c) The function <I>(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 <I>(z) + 1/2z onto the region in Figure 11.9
329
(right).
z w
Figure 11.9
("/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
330 CHAPTER 11. MISCELLANEOUS EXAMPLES
o 1 x
-L
Figure 11.10
Answer.
r(a + 1)r(~)
1= sin:::'(b-a).
2r(1 + 1 +~) 2
(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)
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.
!cj(z)dz = 0
Figure 11.11
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.
332 CHAPTER 11. MISCELLANEOUS EXAMPLES
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,
Hint. Let
1 z- 2 )
g(z) = N ( + R2 + 1 . Z
:;;
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.
[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.
333
334 BIBLIOGRAPHY
[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.
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
Schwarz
lemma 130, 157
reßection principle 236