Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
OF PROBLEMS ON
COM PLE X ANA
-
LYS IS
L. I. Volko vyskii , G. L. Lunts ,
.and I.G. Aram anovi ch
A Collection of Problems
on
Complex Analysis
A Collection of Problems
on
COMPLEX ANALYSIS
L.I. VOLKOVYSKII, G.L. LUNTS,
I. G. ARAMANOVICH
Translated by
J. BERRY
Translation edited by
T. KOVAR!, PH.D.
CHAPTER II
Conformal mappings connected with elementary functions
1 Linear functions (linear furiptions; bilinear functions) 21
2 Supplementary questions of the theory of linear transforma-
tions (canonical forms of linear transformations; some ap
proximate formulae for linear transformations; mappings of
simply connected domains; group properties of bilinear trans-
formations; linear transforma~ions and non-Euclidean geometry) 26
3 Rational and algebraic functions (some rational functions;
mappings of circular lunes and domains with cuts; the func-
tion !(z+ ~); application of the principle of symmetry;
the simplest non-sohlicht mappings) 33
4 Elementary transcendental functions (the fundamental tran
scendental functions; mappings leading to mappings of the
strip and half-strip; the application of the symmetry prin-
ciple; the simplest many-sheeted mappings) 42
5 Boundaries of univalency, convexity and starlikeness
CHAPTER III
Supplementary geometrical questions. Generalised analytic functions
1 Some properties of domains and their boundaries. Mappings
of domains 51
2 Quasi-conformal mappings. Generalised analytic functions 55
CHAPTER IV
Integrals and power series
1 The integration of fw1ctions of a complex variable 64
v
vi CONTENTS
2
Cauchy's integral. theorem 68
3
Cauchy's integral formula 70
4
Numerical series 72
5
Power series (determination of the radius of convergence;
behaviour on the boundary; Abel's theorem) 74
6 The Taylor series (the expansion of functions in Taylor series;
generating functions of systems of polynomials; the solution
of differential equations) 78
7 Some applications of Cauchy's integral formula and power
series (Cauchy's inequalities; area theorems for univalent
functions; the maximum principle; zeros of analytic functions;
the uniqueness theorem; the expression of an analytic function
in terms of its real or imaginary part) 83
CHAPTER V
CHAPTER VI
CHAPTER VII
CHAPTER XI
Applications to mechanics and physics
1 Applications to hydrodynamics 243
2 Applications to electrostatics 258
3 Applications to the plane problem of heat conduction 271
ANSWERS AND SOLUTIONS
Chapter I 274
Chapter n 285
Chapter m 307
Chapter IV 310
Chapter V 317
Chapter VI 324
Chapter VD 328
Chapter vm 339
Chapter IX 349
Chapter X 374
Chapter XI 395
FOREWORD
A Collection of Problema on Oomple:x: Analysis is intended mainly
for students of the physico-mathematical and mechanico-mathemat-
ical faculties of universities and also for students of the physico-
mathema.tical faculties of teachers training colleges.
The authors believe also that the collection will be useful for
those specialising in the mechanics of continuous media (hydro-
dynamics, the theory of elasticity) and electrical engineering, since
it contains a large number of problems either on the direct appli-
cation of the theory of functions of a complex variable to the given
subject, or on questions forming their mathematical basis (conformal
mappings, harmonic functions, potentials, integrals of Cauchy type,
etc.).
Problems connected with the compulsory university course on
the theory of functions of a complex variable are arranged ma.inly
in Chapters I, II ( 1, 3 and 4), IV, V, VII, X ( 1).
The book also contains groups of problems which go beyond the
range of the syllabuses. Some of these may be used as the basis of
course work and as material for study in seminars on the theory of
functions of a complex variable.
It seems to us that the book deals with the main sections of the
theory of functions of a complex variable approximately in accord-
ance with the curriculum.
For the convenience of those using the book, in the contents
besides the titles of the chapters and sections, the main groups of
problems contained in them have sometimes been enumerated (this
refers ma.inly to the basic course material).
It is assumed that the users of the collection are acquainted with
the corresponding sections of the course in the theory of functions
of a complex variable contained in the book by A. I. Markushevich
A Short O<YUrse on tke Theory of Analytic Functions, and therefore
as a rule references are not given to this textbook. If additional
material is used the necessary information of the source and also re-
ferences to the literature are given. The following notation is used
for the books most :frequently cited:
[l] A. I. Markushevich, Theory of Analytic Functions (Teoriya
analitickeskikk funktsii), Gostekhizdat, Moscow, 1950.
ix
x FOREWORD
I. Complex numbers
I. Perform the operations indicated:
2
(1) l/i; (3) l-3i;
1-i
(2) l+i ; (4) (l+iy3)s.
I
2 PROBLEMS ON COMPLEX ANALYSIS
internal angle of the triangle, with vertices at the points -1, I and
z, which passes through the vertex z.
5. Let m and n be integers. Prove that (ii'z)m has n/(n, m) different
values, where (n, m) is the greatest common divisor of the numbers
m and n. Verify that the sets of values of (yz )m and j/ (zDI) are iden-
tical if and only if (n, m) =I, that is, if n and mare coprimes.
A1 B1 01 D 1)
( A2 B2 02 D2
A3 B3 03 D3
A4 B4 04 D4
with integer elements which has the same sum of squares of elements
along the rows, columns and principal diagonals and satisfies the
relations
AkA1+BkB1+0k01+DkD1 =0 (k :F l; k, l = 1, 2, 3, 4).
2. Elementary transcendental functions
By definition
expz =es= e" (cosy+isiny);
els+e-ls es+e-s
COSZ= 2 ; cosh z = - -2- ;
es-e-s
sinhz=---;
2
sinz sinhz
tanz = 00s7; tanhz=--;
coshz
cosz coshz
cotz=-.-; cothz=--
SlnZ sinhz
72. Show that in the case of a rational index (a.= m/n) the
general definition of the power z is identical with the usual definition:
z"n"' = (Y'z)"'
(see also problem 5).
73. Are the sets of values of the following identical:
a, (a)11, (a2)!
By definition, the equation w = cos-1 z is equivalent to the equation
z = cosw. The functions sin-1 z, tan- 1 z, cot-1 z and the inverse hyperbolic
functions cosh-1 z, sinh-1 z, tanh-1 z, coth- 1 z are defined similarly.
74. Prove the following equalities (all the values of the roots
a.re ta.ken):
(1) cos-1z = - i Log(z+y(zll-1));
(2) sin-1z = -iLogi(z+y(zll-1));
93. Into what are the segments of the straight lines x = 0 and
the straight lines y = 0, which lie in the strip 0 ~ y ~ :n; trans-
formed by means of the function w = e:r: +z 1
94. What corresponds in the z-plane to the polar net lwl = R,
arg w = ix in the transformations:
1
(1) w=e-=; (2) w=e:r:1
sufficient that the limits lim lznl '# 0 and (for a suitable definition
n-+OO
limit does not depend on the choice of the sequence and that if f (z)
is defined on the boundary circle by means of passing to the limit,
it is continuous throughout the whole of the closed disk Jzl ~ 1.
14 PROBLEMS ON OOMPLEX ANALYSIS
Jim [Re
.dz-+0
~w]
LJZ
exists for the function w = f (z) then the partial derivatives Uic
and v1 exist and are equal.
(2) If the limit lim [rm
.dz....0
~w] exists then the partial derivatives
LJZ
u1 and Vicexist and u 1 = -Vic.
(3) If it is assumed in advance that the functions u and v
are differentiable then the existence of either of the limits mentioned
in parts (1) and (2) ensures the existence of the other and consequent-
ly the differentiability of the function f(z).
FUNOTIONS OF A OOMPLEX VARIABLE 15
112. At the point z0 the function w = f(z) possesses the following
properties: (1) u, v a.re differentiable; (2) the limit lim
.dz-+0
I~wl exists.
LJZ
au av
-=z-,
am 8y
are said to be conjugate.
115. Prove the following propositions:
n
(1) A linear combination of harmonic functions}; c1u1 (x, y)
i=l
is a harmonic function.
(2) H the inversion transformation
v(x, y) = J-
(x,, y,)
au au
ay dx+ ax dy+O.
Or,
Cf
c:Jn 0
FIG. 1
v(x, y) = J
(x,, y 0)
au au ~
- ay ax+ ax dy+ L,.; ma::n:1,+0.
k-1
The integral is ta.ken along a path which lies in the domain G,
the ma: are integers and
:n;k = J au au
--dx+-dy,
ay ax
YA:
where the /'A: are simple closed contours each of which contains
within it a single connected part of the boundary (I'a:) (the numbers :n;"
are known as the periods of tke integral or the cyclical conBtants).
FUNCTIONS OF A OOMPLEX VARIABLE 17
For the function v(x, y) to be single-valued it is necessary and
sufficient that all the numbers nk should be zero.
REMARK. The contour I'0 may not exist provided that the function u(a:, y)
is harmonic at the point at infinity. By definition this means that the function
U(~, fJ), obtained from the function u(a:, y) by the inversion transformation
(see problem 115, (2)), will be harmonic at the coordinate origin. It can be
n
proved that in this case 2 :n:k = 0.
k=l
117. Assuming it to be Jinown that an analytic function is infi.
nitely differentiable, prove the following theorems:
(1) The real and imaginary parts of the analytic function
/(z) = u+iv are conjugate harmonic functions.
(2) The derivatives (of any order) of a harmonic function
are also harmonic functions.
118. (1) Is u 2 a harmonic function if u is a harmonic function 1
(2) Let u be a harmonic function. For what functions J is
the function f(u) also harmonic 1
119. Are lf(z)I, argf(z), log IJ(z)I, harmonic functions if f(z) is
an analytic function 1
a2u a2u.
120. Transform Laplace's operator Liu = aaf' + ay2 mto the polar
system of coordinates (r, c/J) and find the solution of Laplace's
equation Liu= 0, which depends only on r.
121. Calculate for n = 1, 2, 3, 4 the harmonic polynomials Pn(x, y)
and qn(X, y), defined by the equation zn = Pn+iqn. Write down the
genera.I form of Pn and qn in polar coordinates.
Using the formulae of problem 116 in problems 122--126 find the
functions conjugate to the given harmonic functions in the domains
indicated.
122. u(x, y) = x2-y2 +x, 0 ~ lzi < oo.
x
123. u(x, y) = x2+y2 , 0 < lzl ~ CX>.
1
124. u(x, y) = 2 1og(x2+y2);
(a) in the domain obtained from the plane by the deletion
of the semi-axis: y = 0, -oo < x ~ O;
(b) in the plane with the coordinate origin deleted (O<lzl <oo).
1
125. u(x, y) = 2 1og(x2+1f)-log[(x-1)2 +y2]:
(a) in the plane with the points z = 0 and z = I deleted;
18 PROBLEMS ON COMPLEX ANALYSIS
134. u = <P(!).
135. u = c/J(zy).
136. u = c/J(z2+y2).
z2+y2)
137. u=cf> (- z-.
FUNOTIONS OF A OOMl'LEX VARIABLE 19
138. u = </>[x+y(x2+1f)].
139. u = ef>(x2+?f).
In problems 140-143 prove the existence of and find the analytic
functions f(z) from the given modulus or argument.
e
140. = (xt+ys)eX.
e
141. = e""cos8'[>.
142. 0 = ey.
143. 0 = ef>+rsinef>.
144. Prove that for the family of curves cf>(x, y) = 0, where cf> is
a twice continuously differentiable function, to be the family of
level lines of some harmonic function it is necessary and sufficient
that the ratio LJcf>/(gra.dcf>) 2 should depend only on ef>.
HINT. As a preliminary establish that the required harmonic function is of
the form u =/[(a:, y)].
In problems 145-149 find analytic functions of which the real
part, the imaginary part, the modulus or the argument preserves
a constant value along any line of the corresponding family.
145. x = 0.
146. y = o.
147. y =Ox.
148. x11+y11 = 0.
149. xl+y2 =Ox.
Phe geometrical meaning of the modulUB and argument of a derivative
150. Mappings are made using the functions w = z2 and w = z8
For each function find the angle of rotation (0) of the directions
issuing from the point z0 , and the magnification (k) at the following
points:
(1) 210 = 1; (2) z0 = -1/4; (3) z0 = l+i; (4) z0 = -3+4i.
151. Which part of the plane is compressed and which is stretched
if the mapping is effected by the functions:
(1) w = z2; (4) w = e:r:;
(2) w = z8 +2z; (5) w =log (z-1) 1
(3) w = l/z;
152. The domain G is mapped by means of the function f(z)
conformally and one-one onto the domain G'. Find formulae for
the calculation of the area S of the domain G' and the length L of
the arc onto which some arc l belonging to the domain G is mapped.
153. Find the length L of the spiral onto which the function e:I:
maps the segment y = x, 0 .:::;; x .:::;; 23t (see problem 91).
20 PROBLEMS ON OOMPLEX ANALYSIS
154. Find the area of the domain onto which the function e"
maps the rectangle 1 < z < 2, 0 ,.s;; y < 4.
155. Find the domain D, onto which the function e" maps the
rectangle 1 < z < 2, 0 < y < 8. Calculate the area of the domain
D by means of the formula obtained in the solution of problem 152
and explain why this formula gives an incorrect result.
OHAPTER II
1. Linear functions
Linear function!J
156. Find the linear function which maps the triangle with ver-
tices at the points 0, 1, i onto a similar triangle with vertices at
0, 2, l+i.
157. Find the linear transformation with fixed point 1+2i whioh
transforms the point i into the point -i.
158. For the transformations given find the finite fixed point z0
(if it exists), the angle of rotation round it, (), and the magnification le.
Reduce these transformations to the canonical form w-z0 = A.(z-z0).
(1) w = 2z+l-3i; (4) w-w1 = a(z-z1 ) (a =F O);
(2) w = iz+4; (5) w = az+b (a =f:: 0).
(3) w = z+l-2i;
159. Find the general form of the linear transformation which
transforms :
(1) The upper ha.If-plane into itself;
(2) The upper half-plane into the lower ha.If-plane;
(3) The upper half-plane into the right ha.If-plane;
(4) The right ha.If-plane into itself.
Show that in each case the transformation is uniquely determined
by specifying a single pair of corresponding interior points or two
pairs of boundary points.
160. Find the general form of the linear transformation which
transforms :
(1) The strip 0 < z < 1 into itself;
(2) The strip -2 < y < 1 into itself;
(3) The strip bounded by the straight lines y = z and
y = z-1 into itself.
Explain which pairs of points can correspond to one another in
these mappings and in which case this correspondence determines
the mapping uniquely.
21
22 PROBLEMS ON OOMPLEX ANALYSIS
161. Find the linear function w(z), which maps the strip, con-
tained between given straight lines onto the strip 0 < u < 1 with
the given normalisation:
(1) x =a, x = a+k; w(a) = O;
(2) x=a, x=a+k; w(a+ ;)= ! +i;
Z-Z1
165. Given the function w = - - :
Z-Z2
(1) Prove that the original of the family !w! =A (O <
< A< oo) is a family of circles (the circles of Apollonius). For
a given A find the radius and the position of the centre of the
corresponding circle in the z-pla.ne.
(2) Find the original of the ray arg w = ().
(3) Construct the net in the z-plane which corresponds
to the polar net in the w- plane.
CONFOBMAL MAPPINGS 23
(4) Find the domain of the z-plane which corresponds
to the semicircle lwl < 1, Im w > 0.
In problems 166-170 explain what the domains indicated are
transformed into with the given mapping functions.
z-i
166. The quadrant :x: > 0, y > O; w = --
..
z+i
167. The semicircle lzl < 1, Imz w = 22 z-:i.
> O;
+iz
168. The angle 0 < </> < :rc/4; w = z/(z-1).
169. The strip 0 < :x: < 1; (1) w = (z-1)/z; (2) w = (z-1)/(z-2).
170. The ring 1 < lzl < 2; w = z/(z-1).
171. Map onto the vertical strip 0 < Rew < 1:
(1) The half-plane Re z > 0 with the disk lz-d/21 d/2 <
deleted;
(2) The lune contained between the circles
lz-d1 /21 = d1 /2, lz-d2 /21 = d2 /2 (d1 < d2 );
(3) The exterior of the circles lz+d1 /21 = di/2, lz-d2/21
= d2 /2 so that w(d2 ) = 0.
172. Find the bilinear function which transforms the points -1,
i, 1+i respectively into the points:
(1) 0, 2i, 1-i; (2) i, oo, 1.
173. Find the bilinear function which transforms the points -1,
oo, i respectively into the points :
(1) i, 1, l+i; (2) oo, i, 1; (3) 0, oo, 1.
174. Find bilinear functions from the following conditions:
(1) The points 1 and i are fixed and the point 0 passes into
the point -1;
(2) The points 1/2 and 2 are fixed and the point 5/4+3i/4
passes into oo;
(3} The point i is a double fixed point and the point 1 passes
into oo.
175. Find the bilinear function which transforms the points -1,
0, 1 respectively into the points 1, i, -1, and explain what the upper
half-plane becomes in this mapping.
176. Find the general form of the bilinear transformation whioh
transforms:
(1) The upper half-plane into itself;
(2) The upper half-plane into the lower half-plane;
(3) The upper half-plane into the right half-plane.
24 PROBLEMS ON COMPLEX ANALYSIS
177. Find the mapping of the upper half-plane into itself \\ith
the given normalisation:
(1) w(O) = 1, w(l) = 2, w(2) = oo;
(2) w(O) = 1, w(i) = 2i.
RBMABK. For mappings of the upper halfplane into itself for other nor-
malisations see problem 188.
178. Find the function w(z) which maps the circle lzl < R onto
the right half-plane Re w > 0 in such a way that w(R) = 0,
w (-R) = oo, w (0) =I. What is the image of the upper semi-
circle in this mapping 1
Two points P 1, P 1 are said to be symmetrical with respect to the circle
K with centre 0 and radius R, if they lie on the same ray issuing
from 0 and
OP1 .0P1 = R 8
179. Find the points symmetrical to the point 2+i with respeot
to the cii-cles:
(1) lzl = 1; (2) lz-il = 3.
180. Find the symmetrical image with respect to the unit circle
of the following curves:
(1) lzl = 1/2; (4) lz-zol = l:<iol (zo = xo+iyo):
(2) lai-11 = 1; (5) lz-z0 1= y(lz0 12 -l) (jz0 1> 1);
(3) y = 2; (6) the hyperbola. x2-y2 = 1;
(7) The perimeter of the rectilinear triangle with vertices
Zt Z,, Z:1(Z1 ': 0).
181. Prove that for the symmetry of the points P 1 and P 2 with
respect to K it is necessary and sufficient that one of the following
two conditions should be satisfied:
(1) Every circle K 1 passing through the points Pv P 2 , is orthogonal
to K;
(2) For all points M of the circle K
MP1
--=Const
MP2
(that is, K is a circle of Apollonius with respeot to the points P 1
and P 2).
if z.=oo.
A bilinear transformation with two distinct fixed points z1 and :i:s has the
canonical form
w-z1 z-z1
--=k--,
w-z. z-z.
w-z 1 = k (z-z1 )
if z1 = oo; the transformation with two distinct fixed points is hyperbolic if
k> O, elliptic if k = elar: and ct;f: 0 and lOlllO!lromic if k = aelar:, where a;f: I
and ct#:O (ct and a real numbers, a>O).
201. Prove the following assertions:
(1) The general bilinear transformation
az+b
W==--
cz+d
can be reduced to the form
where
I~ ~I= 1.
(2) If a+<5 is a real number, the transformation is elliptic
when !a+'51 < 2, hyperbolic when la+<51 > 2, and parabolic when
1a+<5I = 2.
(3) If Im(0!+'5) '# 0, then the transformation is loxodromic.
202. Prove that if a linear transformation has two fixed points
the product of the derivatives at these points is equal to unity.
203. Find the circles which in the parabolic transformation
1 1
--=--+h
W-Zo Z-Zo
(1) Any circle passing through the two fixed points is trans-
formed into itself, the direction of traversal being preserved.
(2) Any circle orthogonal to the circles passing through the
fixed points is transformed into a circle having the same property.
(This property follows directly from property (1).)
HINT. As a preliminary consider the case when the fixed points are 0
and oo.
206. Prove the following properties of the elliptic transformation:
(1) Any circle orthogonal to the circles passing through two
fixed points is transformed into itself with preservation of the direc-
tion of traversal.
(2) A circular arc connecting the fixed points passes into a
circular arc connecting the fixed points and making an angle IX with
the first arc (IX= arg k).
207. (1) Prove that in the loxodromic transformation the prop-
erties (2) of the hyperbolic (see problem 205) and of the elliptic
transformation (see problem 206) are preserved.
(2) Prove that, provided that ix =f:: n (ix= argk), fixed circles
do not exist in the loxodromic transformation. Prove that if
oc = n, then the circles passing through the fixed points transform
into themselves with change of the direction of traversal.
208. Prove that in the loxodromic transformation w = ae1z the
10110"
logarithmic spirals r = Ae...- {A > 0) pass into themselves.
209. Prove that the linear transformation
z-a
w= e"----
1-az
(a= lale1, lal < 1),
which transforms the unit circle into itself can only be elliptic, hyper-
bolic or parabolic. Explain for what values of a each of the given
cases holds. Find the fixed points of the transformation and reduce
it to the canonical form.
Some approximate formul,ae for linear transformations
210. The upper half-plane is mapped onto the unit disk so that
the point z = ki (k > 0) passes into the centre of the circle. Find
the length I' of the image of the segment [O, a] of the real axis (a> 0)
and obtain linear approximate formulae for I' for small a/k and
for small k/a.
211. The unit disk is mapped onto itself in such a way that the
original of the centre of the disk, the point x 0 , is on the real axis.
Find the length I' of the image of the arc 0 ~ cf> ::( y of the unit
CONFORMAL MAPPINGS 29
circle (y::,;: n). How does the quantity I'/r vary depending on the
sign of x0 1
212. With the conditions of problem 211 obtain the formulae:
l+xo
(1) I'= - 1 - r+o (y8) for small 'Y;
-Xo
where e = l-r0
Mappings of simply connected domains
214. Prove that if the linear mapping of the disk izl < 1 onto
itself does not reduce to a rotation, then no concentric ring with
centre at the coordinate origin passes into a concentric ring.
REMARK. This proposition is a particular case of the following theorem:
For the conformal mapping of the ring r 1 < lzl < r 8 onto the ring R 1 < lwl
< R 1 it is necessary and sufficient that the condition R 1 /R1 = r 1 /r1 should
be satisfied. In this case the mapping function can only have two forms:
w = az or w = a/z. The mapping is uniquely determined by the specification
of a single pair of boundary points (see, for example, [3, Chapter II, 3]).
2Ui. (1) Map the ring 2 < jzj < 5 onto the ring 4 < lwl < 10 so
that w(5) = -4.
(2) Map the ring 1 < jz-2il < 2 onto the ring 2 < jw-3+2il
< 4 so that w(O) = -l-2i.
The following theorem holdst:
Every doubly connected domain, the boundaries of which do not shrink
T,(z) = a1z+b1'
c1z+d1
LI,=
c, 1
la,
db11 : 0 (i = 1, 2, ... ).
Prove the following assertions:
(1) T = T 1T 2 is a linear transformation with determinant
LI = Ll 1 L1 2
(2) The product of the transformations is associative, that is,
(T3 T 2)T1 = T 3 (T2T 1 ).
(3) Every transformation T 1 has an inverse Tj 1, that is,
T 1Tl 1 = T'j' 1T 1 = J,
where J(z) = z is the identical transformation.
(4) The product of transformations, generally speaking, is not
commutative (give examples).
222. Prove that the transformations
1 1
T2 = - T 3 = 1-z, T , = - - ,
z 1-z,
z-1 z
T,,=--, T a--z-1
--
z
form a group (the group of anharmonic ratioB).
223. Prove that the set of linear transformations which consists
of the rotation of a plane round the coordinate origin by angles which
a.re multiples of cc, forms a cyclical group. In which case does this
group consists of a finite number of transformations 1
224. (1) Prove that the set of transformations of the form
w = (az+b)/(cz+d) where a, b, c, d a.re real integers and
ad-be= 1, forms a group (this group is said to be modular).
(2) Prove that if a, b, c and d a.re considered as complex
integers (that is, numbers of the form m+ni, where m and n a.re
real integers), which satisfy the condition ad-be= 1, then the
set of transformations of pa.rt (1) also forms a group (Picard'B group).
225. Find the fundamental domains of the groups generated
by the transformations :
ll1fl
(1) T(z) =en z (n a natural number);
32 :PROBLEMS ON COMPLEX ANALYSIS
27d
(2) T 1 (z) = enz, T 2 = 1/z;
(3) T(z) = z+w;
(4) T 1 (z) = z+w, T 2 (z) = -z;
(5) T 1 (z) = z+ro1 , T 2(z) = z+ro 2 (Imro 2/ro1 =F 0) (the
doubly periodic group) ;
(6) T 1 (z) = z+ro1 , T 1 (z) = z+w1 , T 3 (z) = -z;
(7) T 1 (z) = z+w; T 2 (z) = iz;
(8) T 1 (z) = z+w, T 2 (z) = e9"'18 z;
(9) T 1 (z) = z+w, T 2 (z) = a21Cil&z.
226. Find the group of linear transformations corresponding
in stereographic projection to the rotation of the sphere:
(1) Round the vertical diameter;
(2) Round the diameter parallel to the real axis;
(3) Round the diameter parallel to the imaginary axis;
(4) Round the diameter, the stereographic projection of
one of the ends of which is the point a.
HINT. If z1 , z1 are the images of diametricaJly opposite points on the sphere,
then z1ii = -1 (see problem 48 of Chapter I).
227. (1) Prove that the group of linear transformations which
correspond to a rotation of the sphere, and transform the points
with stereographic projections a and b into one another, is defined
by the relation
w-b z-a
---=e'--
l+bw l+az
(2) Prove that the differential
ldzl
ds = 1+1z12
is invariant with respect to the transformations of this group and
represents the spherical length of the element of arc dz (that is, the
length of the image of this element on the sphere).
Linear transformations and non-Euclidean geometry
In the realisation of non.Euclidean geometry in the unit circle the part
of straight lines is played by arcs of circles in the unit circle and orthogonal
to it; the role of motion is played by linear transformations of the unit circle
onto itself, the role of distance between the points Zi and z1 by the quantity
1
!,>(Z1 , Z 2 ) = 2log(a, p, z., z1 ),
where a and p are the points of intersection of the "straight lines" passing
through the points Zi and z2 , with the unit circle (the order of the points is
CONFORMAL MAPPINGS 33
a, z1 , z1 , fl), and (a, fl, z9 , z1 ) is the anharmonic ratio of the given points. The
angles are measured as in Euclidean geometry (see, for example, [I, Chapter II,
4, section 8]).
228. Prove that Q(Z1, Z2) > 0 if Z1 i: Z2 and Q(Z, z) = 0.
229. Prove that Q(Z1, Z3) ~ Q(Z1, Z2)+Q(Z2, Z3) where the sign
of equality applies only when the point z3 lies on the "segment",
connecting the points z1 and z2
230. Prove that if one of the points 211 and z2 tends to a point of
the unit circle (or both of them to different points of the unit circle),
then the non-Euclidean distance e(z1 , z2) tends to infinity (that
is, the points of the unit circle correspond to the infinitely distant
points of the non-Euclidean plane).
231. Prove that the differential
remark in mind when finding a normed mapping, that is, a mapping which
satisfies certain supplementary conditions. If the condition of normalisation
is not given then one of the mapping functions is usually indicated in the
answer.
In the practical construction of conformal mappings an important part
is played by certain general principles (see, for example, [2, Chapter IX, 5
and Chapter X, 7] or [3, Chapter II, 1 and 3].
The Riemann-Schwar7J principle of symmetry
Let the boundary of the domain D 1 contain an arc of the circle 0 (in par-
ticular a rectilinear segment), and let the function w =f1 (z) effect the conformal
mapping of this domain onto a domain D~ so that the arc 0 passes again into
an arc of a circle or rectilinear segment O*. Then the function J1 (z) which at
points symmetrical with respect to 0 assumes values symmetrical to the values
of f 1 (z) with respect to O*t, will be analytic in the domain D 1 symmetrical
to the domain D 1 with respect to 0 and will map it onto the domain n:, sym-
metrical to D~ with respect to O*.
The function
l f1(z) in D 1 ,
w = f1(z) = f 1(z) on 0,
J1 (z) in D 1
effects the conformal mapping of the domain D 1 +0+D1 onto the domain
n~+o+n:tt.
The principle of corresponding boundaries
Let D and D* be simply connected domains with boundaries 0 and o
the domain D* being situated entirely in the finite part of the plane. If the
function w = f(z) is analytic in D, continuous on 15 and maps 0 onto O* one-one
with prese~ation of the direction of traversal then it effects the oneone and
conformal mapping of the domain D onto D*.
In the solutions of the problems of this and the following sections in cases
when the mapping is effected by a branch of a many-valued function, it is
recommended that the correspondence of points on the boundaries of the
mapped domain and its image should be traced out (this especially refers to the
mapping of domains with cuts).
234. By means of the function w = z2 and its inverse find the
conformal mapping of the following domains:
(I) The interior of the right hand branch of the equilateral
hyperbola x2-y2 = a 2 onto the upper half-plane;
(2) The exterior of the parabola y2 = 2px, p > 0 (that
is the domain bounded by this p~rabola and not containing its
focus) onto the right hand half-plane.
t If 0 and O* are segments of the real axis (it is always poSBible to arrange
this by making an additional bilinear transformation), then fz(z) = f 1 (z).
tt The mapping will be one-one if the domains D 1 and D 1 and also D~
and n: do not intersect.
CONFORMAL MAPPINGS 35
REIU.BX. For the mapping of domains bounded by curves of the second
order, see also problems 164, HS, 291, 212, 293, 328.
235. By means of the functions indicated in the preceding problem
map:
(1) The interior of the circle r = a cos cf> (a > 0) onto the
exterior of the cardioid e = ~ (1 +cos fJ);
(2) The interior of the same circle onto the interior of the
right hand branch of the lemniscate e = y (cos 20);
(3) The disk lzl < 1 onto the interior of the cardioid
e = A(l+cosO), A> 0 so that w (0) = A/8, w'(O) > O.
238. Find the domain onto which the disk lzl < 1 is mapped
by means of the function
w = R(z+mz), R > 0, 0 ~ m ~ j.
Find the image of the polar net of the z-plane.
!37. Find the domain onto which the semicircle lzl < 1, Re z > 0
is mapped by the function
w=z+z.
238. (1) Find the domain onto which the disk lzl < 1 is mapped
by the function
w = R(z+(zn/n)), R > 0,
where n is an integer, n > 1.
(2) Find the domain onto which the exterior of the unit
disk lzl > 1 is mapped by the function w = R(z+ (1 /nzn)), R > 0, n
an integer, n > 1.
RE111A.:ax. For mappings involving the function
w = B(z+ (1/z))
(Zhukovskii's function) see problem 261 et 11eq.
239. (1) Explain for what values of m the function w = R(z+mzn),
where n is a natural number, effects the conformal mapping of the
disk jzj <1 onto a certain domain, and find this domain.
(2) Explain the same questions for the mapping of the exterior
of the disk lzJ <1 by means of the function w = R(z+ (m/zn)) and of
the interior of the same disk by the function w = R((l/z)+mzn).
240. (1) Map the angle 0 < argz < nx, (0 < '
~ 2) onto the
upper half-plane.
(2) Map the angle -n/4<argz<n/2 onto the upper half.
plane so that w(l-i) = 2, w(i) = -1, w(O) = 0.
36 l'ROBLEMS ON COMPLEX ANALYSIS
241. Find the function w(z), which maps the semicircle lzl < 1,
Im z >0 onto the upper half-plane with the conditions:
(1) w(-1) = 0, w(O) = 1, w(l) = oo;
(2) w(l) ==fl, w(O) = oo;
(3) w(i/2) = i, argw'(i/2) = -7'/2.
242. Find the function w(z), which maps the semicircle lzl<l,
Im z>O onto the disk lwl<l with the conditions:
(1) w(l) = 1, w(O) = -i;
(2) w(i/2) = 0, argw'(i/2) = 7'/2.
243. Find the function w(z), which maps the domain lzl >I,
Imz>O onto the upper half-plane.
244. Map onto the upper half-plane:
(1) The sector lzl <R, 0<argz<7'1X (O<ix ::;;;2);
(2) The domain lzl >R, O<argz<'J"Ccx (O<cx ~2).
In problems 245-250 map onto the upper half-plane the circular
lunes (two-angles).
245. lzl<l, lz-il<l.
246. lzl<l, lz-il>I.
247. lzl>l, jz-ij<l.
248. lzl>l, !z-ij>l.
249. lzl >2, lz-Jf21 <J!2.
250. The exterior of the upper unit semicircle.
In problems 251-259 map the given domains onto the upper
half-plane.
251. The plane with a cut along the segment [-1, I].
252. The plane with a cut along the segment [-i, i].
253. The plane with a cut along the segment [z1' z1].
254. The plane with cuts along the rays (-oo, -R], [R, oo)
(R>O).
255. ';['he plane with a cut along the ray in the first quadrant
which issues from the point i and is parallel to the straight line
y=x.
256. The half-plane Imz>O with a cut along the segment [O, ih],
h>O.
257. The half-plane lmz>O with a cut from ih to oo along the
positive imaginary semi-axis (h>O).
258. The disk lzl<l with a cut along the radius [O, I].
259. The exterior of the unit circle with a cut along the ray [I, oo).
260. Find the mapping of the disk lzl <I onto the w-plane with
a cut along the ray (-oo, -1/4] subject to the conditions w(O) =0,
w'(O)>O.
OONFOBMAL M.Al'l'INGS 37
261. Find the transformation of the polar net lzl = R, arg z =et:
by the function w = ! ~
(z+ ).
onto a concentric circular ring with centre at the coordinate origin, and
find the modulus of the given doubly connected domain (seep. 29).
266. Find the domain onto which the function w = ! (z+-~-)
38 PROBLEMS ON COMPLEX ANALYSIS
maps the disk lzl < 1 with a cut along the segment [a,l] (-1
<a < 1). Consider the cases a > 0 and a < 0.
In problems 267-270 map the given domains onto the upper
half-plane.
267. The disk lzl < 1 with a out along the segment [1/2, l].
268. The disk lzl < 1 with cuts along the radius [- 1, OJ and
the segment [a,l] (0 < a < 1).
269. The upper half of the disk lzl < 1 with a cut along the segment
[O, exi] (0 < ex < 1).
270. The upper half of the disk lzl < 1 with a cut along the segment
[ixi, i] (0 < ex < 1).
271. Map the disk lzl < 1 with a cut along the segment [(l-k)e1,
e'J onto the unit disk of the w-plane.
272. Map the disk lzl < 1 with a cut along the segment [a, l],
0 <a< 1, onto the disk lwl < 1 80 that w(O) = 0, w'(O) > 0.
Find w'(O) and the length of the arc corresponding to the cut. For
what value of a does the cut pass into a semicircle 1
HINT. It is an advantage first to map both the given domain and the
disk lwl < 1 onto the exterior of the segment.
273. Map the disk lzl < 1 with outs along the segments [a, l],
[- l, -b] (0 <a < 1, 0 < b < 1) onto the disk lwl < 1 80 that
w(O) = 0, w' (0) > 0. Determine w' (0) and the length of the arc
corresponding to the cuts.
274. Represent Zhukovskii's function in the form
and find:
:+~ =(~+~r
(1) The image of the circle 0 which passes through the points
z = 1 at an angle ex (- n <IX < n) to the real axis at the point 1,
and the domain onto which the exterior of this circle is mapped;
(2) The image of the circle O' which passes through the point z = 1
at an angle ex to the real axis and contains within it the point -1,
and also the domain onto which the exterior of this circle is mapped.
275. (1) Find the images of the circles and domains of the z-plane
mentioned in problem 274 if the mapping function w(z) is given
by the equation
281. (1) Using the solution of problem 280 and the 'principle
of symmetry find the image of the unit circle in the mapping
z
w=---2-
(l+z")"
(2) Find the function which maps the interior (and
exterior) of the unit circle onto the exterior of the "star" :
2:n;lc
jwj <l, a.rgw = - - (I~= 0, 1, 2, ... , n - 1).
n
282. Map onto the exterior of the unit circle:
(1) The whole plane with cuts a.long the segments [-1, l]
and [ - i, i] (the outside of a. cross);
(2) The whole plane with cuts a.long the rays (-ex>, -1 ],
[l, + ex>), (- i=, - i] and [i, + i=).
283. (l)* Using the function of problem 279, map the sector
lzl < 1, 0 < a.rg z < :n/n (n an integer) onto itself so that the radial
segments lzl <a, a.rg z = 0 and izl <a, a.rg z = :n/n (0 < a < 1)
pass into the corresponding radii.
(2) Map the exterior of the unit circle with cuts a.long the
segments 1 < <IX,
lz! a.rg z = 2lc:7t/n (le= 0, 1, 2, ... , n -1) onto the
exterior of the unit circle.
HINT. In the solution of (2) use the principle of symmetry.
40 PBOBLEMS ON COMPLEX ANALYSIS
284. Map onto the upper half-plane and onto the exterior of
the unit circle the exterior of the cross consisting of the segment
[-a, b] of the real axis and the segment [-ci, ci] of the imaginary
axis (a~ 0, b ~ 0, c ~ 0, a+ b2 + c2 :0).
Hnm Find the function which maps the upper half-plane with a cut
along the segment (0, ci] onto the upper half.plane and use the principle of
symmetry in virtue of which the exterior of the cross is mapped onto the whole
plane with a cut along a segment of the real axis.
285. Map the plane with outs along the ray [-a, + oo) (a ;;;;i: 0)
and along the segment [- ci, ci] (c > 0) onto the upper half-plane.
HINT. See the hint to problem 184.
286. Map onto the exterior of the unit circle the plane with outs
along the negative part of the imaginary axis and along the lower
half of the unit circle.
HINT. It is reduced to problem 182, (1) by a linear transformation.
287. Map the plane with outs along the segment [- ai, O] (a> 1)
and along the lower half of the unit circle (Fig. 2) onto the upper
half-plane.
HINT. It is reduced to problem 284 by a linear transformation.
-t eiA
-1
-ce'l
"
Fm. 2 FIG. 3
288. Map onto the upper half-plane the plane with outs along
the segment [-1, b] (b > -1) and along the circular arc with ends
at the points e:l:lai, which passes through the point z = -1 (Fig. 3).
OONFORMAL M.Al'l'INGS 41
289. Map onto the upper half-plane the exterior of the unit circle
with cuts along the segments: [i,bi], [-bi, - i], [l, a], [-a, -1]
(a> 1, b > 1).
HINT. Zhukovskii's function maps the domain considered onto the domain
of problem 284. '
290*. Map onto the exterior of the unit circle the exterior of
the "star" represented on Fig. 4.
FIG. 4
291*. Map onto the upper half-plane the interior of the right-
hand branch of the hyperbola
~-1--1
coslla sin2 oc -
292*. Map onto the upper half-plane the exterior of the right-
ha.nd branch of the hyperbola
x2 y2
-----=l.
cos2 ix sin2 oc
293. Map onto the upper half-plane the domain included between
the branches of the hyperbola
x2 y2
aa--=1.
297. (1) w
298. w
= y[z(z2+1)];
= y(z9-1).
(2) w = -v( z2~ I ) .
a a
Fm. 5
asinh~ a sin 17
x= cosh~+cos17 ' y = cosh~+cos17 '
(x-acoth~o)2+y2 = ( sin~~o r
(the original of the segment ~ = 0, -n ~ 17 ~ n is the ordinate
axis) (Fig. 6).
44 PROBLEMS ON COMPLEX .ANALYSIS
(4) Prove that the originals of the lines 'Y/ = 'Y/o are arcs of
circles passing through the points a,
a:l+(y+acot'Yj0 ) 2 = (~)
Slll 'YJo
2
,
which lie in the upper half-plane for 'Y/o > 0 and in the lower half.
plane for 'Y/o < 0. The line 'Y/ = 0 corresponds to the segment [-a, a].
The arcs corresponding to the values 'YJ = 'Y/o and 'Y/ = 'Y/o-n ('l}o > 0),
supplement one another and make up a complete circle (Fig. 7).
!I
:r
FIG. 6
!I
z
acot'lo .r
FIG. 7
II
Fm. 8
48 PROBLEMS ON COMPLEX ANALYSIS
FIG. 9
HINT. Map the function which gives the solution of problem 329 onto the
plane with cuts along the rays
( -co -
'
l ]
sin-1 sech a ' [sin-1 ~ch a' +oo] '
332. Map the plane with cuts along the rays (-oo,p], [q, +oo}
(-n/2 ~ p < q ~ n/2) and along the segments -a ~ y ~a,
x = n/2+kn (k = 0, 1, 2, ... ) onto the upper half-plane
(Fig. 10).
FIG. 10
333*. Map the plane with cuts along the rays 0 ~ y < oo,
x = kn/2 (k = 0, 1, 2, ... ) onto the upper half-plane.
CONFORMAL MAPPINGS 49
In problems 334-337 the images are many-sheeted domains (see
the footnote on p. 41).
334. Find the domains onto which the following are mapped by
means of the function w = es:
(1) The rectangle 0 < x <a, 0 < y < b;
(2) The half-strip 0 < x < a, y > 0;
(3) The strip 0 < x < a.
335. Find the domains onto which the following are mapped by
means of the function w = cos z:
(1) The strip -n/2 < x < :i/2;
(2) The strip 0 < x < 2n.
336. Find the domain onto which the strip 0 < x < 2n is mapped
by means of the function w = tanz.
337. Construct the Riemann surface onto which the function
.!
w =es maps the z-plane.
5. Boundaries of univalency, convexity and starlikeness
In problems 889-345 r 1 denotes the maximum radius of a circle with centre
at the coordinate origin within which the function w = f(z) is univalent;
r 1 denotes the maximum radius of a disk with centre at the coordinate origin
which the function w = f (z) maps schlicht onto a convex domain and r 3 the
maximum radius of a disk with centre at the coordinate origin mapped by
the function w = f(z) schlicht onto a domain which is sta.rlike with respect
to the point w = O. (A domain is atarlike with respect to a given point, if
any point of the domain can be joined to the given point by a rectilinear
segment lying wholly within the domain.) It is obvious that r 1 .;:;:;; r 3 .;:;:;; r 1
338. Find r 1 , r 2, r 3 for the function w = z/(1-z) and construct the
images of the disks izl < rv izl < r 2, izl < r 3
339. Find r 1 for each of the following functions:
(1) w = z+z2;
(2) w = z+a.z2, (a is a real number);
z
(3) w = (l-z>9.
340. Prove that in the mapping w = f(z) the curvature of the
image of the circle izl = r is expressed by the formula
l+Re [z7fz)
f"(z)]
k=
jzj'(z)j
50 PROBLEMS ON COMPLEX ANALYSIS
341. Prove that an analytic function /(z) maps the circle Jzl = r
onto a convex curve when and only when
a [n
a<P 2+</J+argf'(z)
]= l+Re [zf"f'(z)(z)] ;;,:: O
for all <P (z = re14>)t.
342. Prove that the disk Jzl < r is mapped by the analytic function
f (z) onto a starlike domain when and only when
SUPPLEMENTARY G:mOMETRICAL
QUESTIONS. GENERALISED ANALYTIC
FUNCTIONS
350. Prove that every closed polygonal curve, that is, a continuous
curve consisting of a. finite number of segments, can be divided into
a finite number of aimple closed polygonal curves, that is not pos-
sessing multiple points.
51
52 PROBLEMS ON COMPLEX ANALYSIS
351. Prove that every polygon can be divided into a finite number
of convex polygons.
352. Prove that every simple closed polygonal line divides the
points of the plane not belonging to it into two domains, an interior
and an exterior, and is identical with the boundary of each of these
domains (a particular case of Jordan's theorem, which makes the
same assertion for any simple closed continuous curve).
HINT. Let us fix a direction (ray) which is not parallel to any side of the
polygon P considered and divide all the points not belonging to P into two
classes: we refer to one class those points for which a ray starting from them
and parallel to the given direction intersects P in an even number of points
(0, 2, 4, ... ), and to the other class the points for which the same ray gives
an odd number of intersections with P. A vertex of P is considered to be
a point of intersection if the two sides of P approaching it are situated on
different sides of the ray. Let us prove that two points of one class can be
connected by a polygonal line which does not have points in common with P,
and if two points belong to different classes then any polygonal line connecting
them has points in common with Pt.
353. Prove that in every simple polygon it is possible to draw
a diagonal, all the points of which excluding the ends lie within
the polygon.
354*. Prove that a Jordan curve does not contain interior points.
A set a S G is said to be open with respect to G if together with each of its
points z0 it contains all the points ze G lying in a sufficiently small neigh-
bourhood Ua (Zo) of the point z0 (lz - z0 1 < e). The set e S E is said to be
olosed. with respect to E if it contains all its limit points belonging to E.
355. Let G be a domain in the z-plane and E =f(G) the image
of G in the w-plane in the mapping w = f(z).
Prove the equivalence of the following definitions of the conti-
nuity of the mapping:
(1) The mapping w = f (z) of the domain G is continuous at the
point Zoe G iffor every e-neighbourhood u.(wo) of the point Wo = f(zo)
it is possible to find a .5-neighbourhood U.,(:210) of the point z0 , situated
in G such that its image belongs to U8 (w0 ), that is, for ze U.,(z0 )
we have w = f(z)e U8 (w0). The mapping w = f(z) is said to con-
tinuous in the domain G, if it is continuous at every point of this
domain (Cauchy).
(2) The mapping w = f(z) of the domain G is continuous
if the original J-1 (e) of every set e c E, E = f(G), closed with
respect to E, is closed in G.
(3) The mapping w = f(z) of the domain G is continuous if the
originalf-1 (e) of every set e c E, open with respect to E is open in G.
t See R. CoURANT and H. ROBBINS, (1942). What iB mathematics?
SUPPLEMENTARY GEOMETRICAL QUESTIONS 53
(4) The mapping w = f(z) of the domain G is continuous
if for every set g s;; G the image of the closure of g with respect
to G is contained in the closure of f(g) with respect to E.
A mapping (single.valued) w = f(z) is said to be one-one or ~. if two
distinct points z always correspond to two distinct points to, that is f (z1 ) of: /(z 1)
if z1 of: z1 A one-one and mutually continuous mapping is said to be topolo
gical or homeomorphic.
356. Construct a topological mapping of the unit disk lzl < 1 onto
the whole finite w-plane.
357. Is the following theorem correct~ [See, for example, 1, Chap-
ter I, 6.1.]
If w = f(z) is a function continuous in the domain G and maps
this domain one-one on some set E, then the latter is also a domain,
the mapping w = f(z) of the domain G on the domain E being topo-
logical (homeomorphic).
Using this theorem prove that if the function f(z) is continuous
in the closed region G and is one-one within G then the boundary
of the domain G is transformed into the boundary of the domain
f(G).
A path r is defined as a continuous image of a directed segment 0 ~ t ~ I,
z = z(t) <r may have multiple points). The point z(O) is said to be the beginning
of the path, z(l) is its end. IfT = ~(t) is a homeomorphic mapping of the segment
0 ~ t ~ I on the segment 0 ~ T ~ I and t/>(O) = O, t/>(l) = 1, then the paths
z{t) and z[t/>{'r)l are considered identical. Two paths ,,0 , r 1 with common ends
a, b are said to be cominuoualy deformable into one another or homotopic in the
domain G, if there exists in G a family of paths r.,a
0 ~ .il ~ I, with ends
a, b, continuously varying with .il (this means that m:x lt/>A1(t) - ~A. (t)I -+ 0
as .ili -+ .il1) and such that the given paths ro r1 correspond to the values
.il = 0, .il = I. The closed path r is said to be continuoualy deformable into
a point or homotopic to zero in the domain G if it is homotopic in the domain
G to a null path, that is, a path consisting of a single point.
358*. Prove the following assertions:
(1) Every path yin an arbitrary domain is homotopio to a poly-
gonal path.
(2) Every closed path in a circle is homotopio to zero.
(3) Every closed path in a circular ring is either homotopic to
zero or homotopio to any boundary circle, traversed in a definite
direction one or several times.
359. Let y be an arbitrary path, closed or not. Prove the fol-
lowing assertions :
e
(1) arg (z-a), z e y, a y, are uniquely defined as continuous
functions of z, if arg (zti-a) is given for an arbitrary point z0 e y.
54 PROBLEMS ON COMPLEX ANALYSIS
(2) The quantity LI,. arg (z-a), equal to the increment of arg (z-a)
along y, does not depend on the choice of z0 or arg (z0 -a).
(3) H the path r lies within an angle with vertex at the point
a of magnitude IX (0 < IX < 2n) then LI.a arg (z-a) < IX.
(4) For a smooth (or even only rectifiable) path r the equality
is said to be the indez of the point a with respect to y (or the winding number
of y with respect to the point a).
360. Prove the following assertions:
(1) In the homotopio deformation of the path r taking place
outside the point a the index n,.(a) is unchanged.
(2) In every closed path r it is possible to inscribe a closed poly-
gonal curve n such that n 7 (a) = nn(a) (a is a given point outside y).
(3) If the points cii, a1 can be connected by a path which does
not encounter the path y, then n 7 (a1 ) = n 7 (a9).
(4) If the point a and the path rare separated by some straight
line then n 7 (a) = 0.
361. Prove that for a simple closed polygonal path n the index
n,.(a) = 1, if a lies within n (the sign depends on the direction
of traversal), and n,.(a) = 0 if a lies outside n.
362*. Let the function w = f(z) effect the topological mapping
of the domain G on the domain G'. Let Q be a small circle in G, r
a closed path in Q and a a point of Q which does not belong to y.
Let, finally, r' and a' be the images of r and a in G'. Prove that
n 7 ,(a') = n7 (a)c5,
where c5 equals either +1 or -1, independently of r and a. In par-
ticular prove that if the mapping is continuously differentiable
and the Jacobian
J = au !!!.. - ~ !!!.. =I= 0
ax oy oy ax '
then the sign of c5 is identical with the sign of J.
SUPPLEMENT.A.BY GEOMETRIC.AL QUESTIONS 55
363*. Let w = /(z) effect the continuous mapping of the closed
region G bounded by the Jordan curve y. Prove that if f(z) a
on y and for the image y' = f(y) of the curve y the index n 7,(a) 0,
then within G, f(z) assumes the value a.
364. Prove the equivalence of the following three definitions of
a. simply connected domain:
(1) The domain G is said to be simply connected if every closed
pa.th lying in G is homotopic to zero in this domain.
(2) A domain G which does not contain the point at infinity is
simply connected if together with every simple polygon situated
in G it also contains its interior.
(3) A domain G different from the extended complex plane is
simply connected if it is bounded by a. single continuum or point.
RElllABX. The extended complex plane is simply connected by the dret
definition.
2. Quasi-conformal mappings. Generalised analytic functions
If in the relation
au
8111 = "'x
etc. will be used.
365. Prove the following relations:
(4) ux = ! (w +wi+w.+w
11 11 );
56 PROBLEMS ON COMPLEX ANALYSIS
(5} u1 = 2i(Wz-Wi-wz+wr:
- -} ;
(6) Vx
i
= -2(w.z-wi+wz-wr:};
- -
(7) v1 = 2I (wr:+wz--wi-wr:
- - }.
J w/r: = D(u,v}
D(:.i:, y} = IWz 11 -,wi"
' 1 .
maxl*I lw.l+lw.I
where
p= minl~~I - liw.1-lwill
where
r = psin18+2-cos11 8.
p
REMARK. The quantities at, (J, y are also 1ialled characteristics of the ellipse.
They are connected by the relation
rx.y-(J = 1.
'l'he function w == /(z) effects an affine (linear) mapping of the z-plane on the
w-plane, if
where
() P1(pl-l) el"'
q1 z = - (PP1+l)(p+P1)
Pwtt ~ Pw1111P111tt
FIG. 11
(the condition of ellipticity is here a{l-y 2 > 0; in addition at: > 0) can
be reduced to the form w.-q1(z)w111 -q2 (z)wi+.Aw+BW = F,
where lq1 (z)l+lq1 (z)I < 1, if at:1 >0, and jlq1 (z)l-lq2 (z)I\ > 1 if 1 <0.
HINT. See problem 181. The case at1 < 0 is reduced to the case at1 > 0
by the replacement of w ... u+w by iii = u-w.
393. Prove that if w(z) is a continuous differentiable solution
of the equation
ios-q2(z)w = o,
where q2 (z) is an analytic function of z and lq2 (z)I =f:. 1 then
w(z) = </>(z)+q2 (z) ~(z) ,
l-lq2(z)!
where <f>(z) is an arbitrary analytic function.
394. Prove that if w(z) is a twice continuously differentiable
solution of the equation
ios-q2(z)w. = o
where q2 (Z') is an analytic function of z and lq2 (z)I .P 1, then
w(z) = </>(z)+ Jq2 (z)~sdz,
where <f>(z) is an arbitrary analytic function of z.
HINT. It is first necessary to prove that w(z) is the sum of an analytic
function of z and an analytic function of z.
OHAPTER IV
f
Z1
(1)
Zo
dz = z1 - z0 ; (2) J
z,
zdz = ! M-zU .
396. Let 0 be a simple closed contour bounding an area S. Prove
the following equalities:
(1) fc zdz = iS; (2) fc ydz = - S; (3) Ic zdz = 2iS.
397. Evaluate the integrals
11 = Jzdz, 12 = Jydz
along the following paths:
(1) Along the radius vector of the point z = 2+i;
(2) Along the semicircle lzi = 1, 0::::;;; arg z :s:;;;:n; (the commence-
ment of the path is at the point z = 1);
(3) Along the circle lz - al = R.
J
398. Evaluate the integral lzldz along the following paths:
(1) Along the radius vector of the point z = 2-i;
(2) Along the semicircle lzl = 1, 0 ::::;;; arg z ::::;;; :n; (the
commencement of the path is at the point z = 1);
(3) Along the semicircle izl = 1, -:n;/2 ::::;;; arg z ::::;;; :n;/2 (the
commencement of the path is at the point z = -i);
(4) Along the circle lzl = R.
J
399. Evaluate the integral izlzdz, where 0 is the closed contour
c
consisting of the upper semicircle izl = 1 and the segment -1 ::::;;; z
::::;;; 1, y=O.
64
INTEGRALS AND POWER SERIES 65
FIG. 12
J
401. Evaluate the integral (z-a)ndz (n is an integer):
(1) Along the semicircle lz-al = R, 0 ::::;;; a.rg (z-a) ~ :r:
(the commencement of the pa.th is at the point z = a+R);
(2) Along the circle lz-al = R;
(3) Along the perimeter of the square with centre at the
point a and sides parallel to the coordinate axes.
In problems 402-405 the branch of the many-valued function
which occurs as the integrand is selected by the specification of
its value at some point of the contour of integration. If the contour
is closed the initial point of the pa.th of integration is always consid-
ered to be that point at which the value of the integrand is specified
(it must be remembered that the value of the integral may depend
on the choice of this initial point).
402. Evaluate the integral J ~: a.long the following contours:
(1) Along the semicircle lzl = 1, y ~ 0, yl = 1;
(2) Along the semicircle lzl = 1, y ~ 0, yl = -1;
(3) Along the semicircle lzl = 1, y ~ 0, yl = 1;
(4) Along the circle lzl = 1, f.'l = 1;
(5) Along the circle izl = 1, y(- 1) = i.
J
403. Evaluate the integral Log z dz, where:
c
(1) 0 is the unit circle and Log 1 = 0;
(2) 0 is the unit circle and Logi = ni/2;
(3) 0 is the circle lzl = R and LogR =log R;
(4) 0 is the circle izl = R and LogR =log R+2ni.
66 PROBLEMS ON COMPLEX ANALYSIS
J adz = 0.
lzl=l
407. For what values of oi (0 ,,;;; oi < 2n) do the following integrals
exist:
(pis a natural number), taken along the radius vector of the point
z = e"'1
f
lzl~R
jdzl
iz-all 21 +al <
2nR
IR2-lal 2 I'
409. Prove the following assertions:
(1) If f(z) is continuous in the neighbourhood of the
coordinate origin then
2..
lim
, r-0 0
Jf(re1~)d4' = 2n/(O).
(2) If f (z) is continuous in the neighbourhood of the point
z=athen
lim
r-0
J
lz-al=r
f(z) dz = 2nif(a).
z-a
410. Prove the following assertions:
(1) If f(~) is continuous in the half-strip z ~ z 0 , 0,,;;; y,,;;; h
and the limit lim f(z+iy) =.A, not depending on y exists, then
lim
x-+00 Px
Jf(z)dz = i.Ah,
INTEGRALS AND POWER SERIES 67
where Px is a segment of the vertical straight line 0::::;;; y ~ h, trav-
ersed from below upwards.
(2) If f (z) is continuous in the sector 0 < lz - al ~ r0 ,
0 ~ arg (z - a) ~ cc, (0 < cc ~ 2n) and the limit
lim [(z-a)f(z)] = A
z_,.a
exists, then
lim
r...O 1,
Jf(z) dz = iAcc ,
where i'r is the arc of the circle lz-al < r which is present in the
given sector, traversed in the positive direction.
(3) If f(z) is continuous in the region !zl;;:?:: R 0 , 0 ~ arg z ~cc
(0 < cc ~ 2n) and the limit
lim zf(z) = A ,
Z-+00
exists, then
lim
R"""oorR
Jf(z)dz = iAcc,
where JR. is the arc of the circle lzl = R, which lies in the given
region traversed in the positive direction with respect to the coor-
dinate origin.
411. Prove the following theorems:
(1) If f(z) is continuous in the region lzl;;:?:: R 0 , Im z;;:?:: a (a is
a fixed real number) and lim /(z) = 0, then for any positive
number m
lim Je',,,.f(z) dz = 0 ,
R....oorR
where I'R is the arc of the circle lzl = R, which lies in the region
considered (Jordan's lemma).
HINT. In the evaluation of the modulus of the integral along the semicircle
lzl = R, Im z > 0 make use of the inequality sin 6;;;;.: 28/n for 0 =s;;; 6 =s;;; n/2,
and in the evaluation along the a.res lying in the lower half-plane (in the case
a < 0), use the fact that the length of each of them tends to !al as R -+ co.
(2) If f(z) is continuous in the half-plane Re 21;;:?:: <1 (<1 is a fixed
real number) and lim/(z) = 0, then for any negative number t
lim
R-+oorR
Jff'f(z)dz = 0,
68 PROBLEMS ON COMPLEX ANALYSIS
where I'R is the arc of the circle izl = R, Re z ";;!:;<1. If f(z) is contin-
uous in the half plane Re z ~Cl, the assertion is true if tis positive
and I'R is the circular arc izi = R, Re z ~Cl.
REMABK. The proof of both theorems Vi given, for example, in [3, Chapter V,
2, 73].
2. Cauchy's integral theoremt
412. Show that if the path does not pass through the coordinate
origin, then
Ji
z
= In r+i<P+2nik
1
where k is an integer indicating how many times the path of inte-
gration has encircled the coordinate origin (z =re'"'>
413. Show that if the path does not pass through the points
i, then
where k is an integer.
414. Show that if 0 is an arbitrary simple closed contour which
does not pass through the point a, and n is an integer, then
f
c
(z-a)ndz = I 0, if n =F -1 ,
2ni, if n = -1, a is inside 0,
0, if n = -1, a is outside 0.
415. Cauchy's theorem holds in the following stronger form:
if f (z) is continuous in the closed region G, bounded by a simple
rectifiable contour 0, and analytic within G, then f (z)dz = 0. Jc
Prove this for the case of a starlike contourtt.
HINT. Assuming 0 to be starlike with respect to the coordinate origin
consider the contour OJ,: C= .M (0 <A< 1, ze 0) and pass to the limit
as .A -+ 1 (see, for example, [2, Chapter V, sec. 8] or [3, Chapter I, 4,
sec. 12]).
t The problems on the evaluation of integrals given in this and in the
following section are mainly of an illustrative nature. The majority of problems
of this type are located in Chapter VII on the calculus of residues.
tt A contour is said to be starlike with respect to some point if every ray
issuing from this point meets the contour in a single point.
INTEGRALS AND POWER SERIES 69
418. Prove the following asse1 tions:
(1) Ifj(z) is analytic in the strip 0 ~ y:,;;;; k, liin f(x+iy) = 0
~:!:co
00 co
and the integral Jf(x)dx exists, then the integral -coJf(x+ik)dx
-co
also exists and they are both equal.
(2) If f(z) is analytic in the angle 0 :,;;;; arg z ~ ex (0 < ex ~ 2n),
00
Jf(z)dz along the ray z = re1, 0 ~ r < oo, also exists and they
are equal to one another.
HINT. Use the results of problem 410.
417. Prove that
co
J e-x"cos2bxdx =
0
yn e-b'.
2
HINT. Integrate the functionf(z) = e-zl along the boundary of the rectangle
l:i:I ~ R, 0 ~ 1J ~ b and use Laplace's integral
00
J e-t'dt=Jl':i.
-oo
HINT. Integrate the function j(z) = el along the boundary of the sector
0 ~ lzl ~ R, 0 ~ arg z ~ : snd use the result of problem 411 (1) (put
z = C).
419. Prove that
co
0
J sinx n
---dx=-
x 2
(Dirichlet's integral).
420. Prove that the following equations hold for 0 < 8 < 1:
00
HINT. Integrate the function f(z) = z&-1e-is along the boundary of the
domain r ~ lzl ~ R, - ; ~ arg z ~ O; use the results of problems 410 (2)
and 4ll (1) and the integral representation of the Gamma function:
00
I'Ce) = J a:t-ie-tde.
0
(1) The point 3i lies inside the contour 0 and the point
-3i outside it;
(2) The point -3i lies inside the contour 0 and the point
3i outside it;
(3) The points 3i lie inside the contour 0.
4!2. Calculate all the possible values of the integral J z(z~l)
c
for different positions of the contour 0. It is assumed that the contour
0 does not pass through any one of the points 0, 1 and -1.
4!3. How many different values can the integral Jw~~z)
c
assume, where ron(z) = (z-z1)(z-z1) ... (z-zn) (z1 .P ZJ) and the
contour 0 does not pass through any of the points z1
424. Evaluate the integral
f
lz-J- o
zdz
z4 - l '
a> 1.
INTEGRALS AND FOWER SERIES 71
-5.
,... E v aluate t h e mtegra1 -1.
211:i z2 a11
f +
edz , if t h e contour O
c
contains within itself the circle izl ~ a.
. 1
426. Evaluate the mtegral - 2 .
11:i
f (
zedz
z-a
)3 , if the point a lies
c
within the contour 0.
HINT. Use the formulae for the derivatives of Cauchy's integral.
!i cf
2 J'(z)Logzdz =f(z0 )-f(O),
-2 1 . f z Log--
2 z+l dz,
z- 1
:rci
c
if Log a = log a for a > 0, and for the contour 0 we have:
(1) The circle izl = 2;
(2) The circle iz-11=1 and the initia.l point of integration
is z = 1 i. +
430. By Liouville's theorem a function f(z) which is analytic
and bounded throughout the whole plane is a constant. Prove this
theorem after calculating the integral
f
lzl=R
f(z)dz
(z-a)(21-b)
(!al < R, !bl < R) and taking its limit as R --. oo.
72 PROBLEMS ON COMPLEX: ANALYSIS
1
2"7
J f(C)
-,-de=
1-f(z)+A, if the point z belongs to the exterior
of the domain D,
:n:i c -z A, if the point z belongs to the domain D.
The contour 0 is traversed in the positive direction with respect
to the domain D.
HINT. First consider the oase A = 0.
433. Let the function f (z) and the contour 0 satisfy the conditions
of the preceding problem.
Prove that if the coordinate origin belongs to the domain D,
then
1
2:rti
J zf(C)
Cz-C 2
dC = { 0, if zeD,
f(z), if zeD.
c
4. Numerical series
00
434. Prove that if the series J; en converges and larg cnl ~ oc < :rt/2,
n=l
then the series converges absolutely.
00 00
436. The series 2 en possesses the property that the four parts
n=l
of it, each of which consists of the terms contained in one and the
same closed quadrant of the plane, converge. Prove that the given
series converges absolutely.
437. Prove the formula (Abel's transformation)
n n-1
439. Prove that for the convergence of the series 2 anbn where
n-1
00
2 akbk converges.
k=l
00
and
cc+pi+cca+p'+ ... (0 < ex < p < 1),
show that a series may converge even when lim Ic:: 1 I > I.
2z"
00
455.
n-1
-;;; 457. 2n"-a".
nl
2n z.
00
2z"
00
2 :! z".
ClO ClO
459.
n=l
463. 2
n-0
[3+(-l)"]"z".
ClO
2
ClO
460.
n=O
znl, 464. 2
n=o
(cos in)z".
2 2nznl.
ClO ClO
461.
n=O
465. 2
n=O
[n+an]z".
ClO
462
'1 z2" .
.L.J
n-0
ClO
ClO ClO
(2) }; (2"-l)cnz";
n=O
(5) 2 c~z";
n=O
ClO
ClO
(3) 2 ~z";
n.
n=O
(6) }; (l+z3)c4 z".
n=O
ClO ClO
2 :n z"?
ClO ClO
(1) 2
n=O
(a4 bn)zn; (2) }; anbnz"; (3)
n-o n=O n
76 PROBLEMS ON COMPLEX ANALYSIS
473. }; (-l)nzn/n.
n=l
00
By Abel's theorem if the series 2 c,, converges then
n=O
00 00
lim
r-+1
2 o,,r" = n=O
n=O
2 Cn (0 < r< 1).
477. Show that the converse to Abel's theorem is not true, that
00
is, give an example of a divergent series 2 Cn, for which there exists
n-0
00
00
(4) ~ sin(2n+l)</>
,L,; 2n+l
=-
4
(O < </> < :n;);
n-0
2
00
2
00
00
~ (-l)lt'n)zn
4'79*. Prove that the series ~ - converges, but condi-
n-1 n
tionally only at all the points of the circle of conv-ergence.
480. Prove that if the sequence of real positive numbers {a.} tends
monotonically to zero and the radius of convergence of the series
co
2 anz" is equal to 1, then this series converges everywhere on the
n=O
circle izl = 1, excluding, perhaps, the point z = 1.
HINT. Use Dirichlet's test of convergence (see problem 488).
00
516. Prove that all the Bernoulli numbers with odd subscripts,
except for B1, are equal to zero.
z (-z)
HINT. Use the identity &11-l - e--l = -z.
80 PROBLEMS ON COMPLEX ANALYSIS
n-o
satisfy the relation en= Cn-i+en-2 (n ~ 2). Find en and the radius
of convergence of the series.
REMABB:, The numbers Cn are known as Fibonacci numbera.
520. In the expansion
find e 0, o1, e 8, and also the recurrence relation between e., Cn-1, Cn-s
Cn-8 (n ~ 3).
F(t, z) = 2 In (z)t
n=O
iB satisfied, then the function F(t, z) iB said to be a generating functidn. for
the sequence {/n(z)}. It is frequently possible to prove some of the properties
of the sequence of functions {/n(z)} from the properties of its generating
function.
521. The Bernoulli polynomials efJ.(z) are defined by the expansion
00
/"'-1 = ~ _tPn(z_)_t.
e'-1 .L.J
n=l
nl
INTEGRALS AND FOWER SERIES 81
l 2
y(l-2tz+t)
=
11 = 0
L
P11 (z)t".
Using the integral formulae for the coefficients of the Taylor series
establish that for n ;):. 2
4T11 + 1 (z)-4zT11 (z)+T11 _ 1 (z) = 0.
82 PROBLEMS ON COMPLEX ANALYSIS
e2t:r:-i =
" Hn(z) en.
"1
L.; n!
n=O
Prove the following relations:
(1) Hn+ 1(z)-2zH.(z)+2nHn_ 1(z) = 0 (n ~ l);
(2) H~(z) = 2nH._ 1 (z) (n ~ 1);
(3) H~'(z)-2zH~(z)+2nHn(z) = 0 (n ~ O);
(4) H (z) = (-l)e=~ce-"'')
n dzn
526. The Laguerre polynomials can be defined by the equation
d(ze-"')
Ln(z) = e= dz
Find a genera.ting function for the sequence {L.(z)} and by means
of it obtain the recurrence formula. connecting
Ln- 1 (z), Ln(z) and Ln+ 1 (z).
REMARE. In problems 522-526 only a few of the particular properties
of the given systems of polynomials have been considered. For other important
properties of them which play an important part in the solution of various
problems of mathematical physics, see, for example, [3, Chapter VII, 2]
or R. CoURANT and D. HILBERT, Methods of Mathematical Physica, (Vol. I,
Chapters II and VII, Interscience Publishers, New York).
In problems 52'7-529 find the solutions of the given differential
equations, which satisfy the conditions w(O) = 0, w' (0) = 1.
52'7. w"-z2 w = 3z2 -z4.
528. (1-z2)w"-2zw'+n(n+l)w = 0.
529. (l-z2)w"-4zw'-2w = 0.
530. Expand the function cos (m sin-1 z) (sin-1 0 = 0) in a series
00
dF(adb, c, z) = ab F(a+I,b+l,c+I,z).
z c
7. Some applications of Cauchy,s integral formula and power series
mu. Let the expansion of the function /(z) in the circle lzl < R
be of the form
00
/(z) = 2
.... o
011 z11
2~ Jj/(re')illd,P =
0
,21c 12rfln
m=O
11 (r < R).
2 c~~n
00
00
S=n };nlcnl2
n-1
HINT. Write the formula for the calculation of the area S in polar coordi
nates (see problem 152).
Rmt:ABJC. If we omit the condition that the mapping is schlicht the separate
parts of the domain G must be counted as many times as the corresponding
values of the function /(z) are ta.ken in the circle Ir.I .s;;; 1.
537. Prove that if in the conditions of the preceding problem the
function /(z) is analytic only in the open disk izl < 1 and if the
finite limit lim Sr = 8 then exists, where Br is the area of the open
r--.1
00
equals 8 /:n;. Prove also that if lim 8, = <X>, then the series 2 n lcnl2
r-1 n-1
diverges.
REMARE. See, for example, [4, Chapter xm. l].
538. (1) Using the solution of problem 536 prove that ifj'(O) = 1
and the functionf(z) maps the disk lzl :,;;;; 1 conformally and one-one
t See problem 480 where another method of proof of this theorem is
given.
INTEGRALS AND l"OWER SERIES 85
onto some domain G, then the area of the domain G is not less than
the area of the mapped disk (the extremal property of mapping
onto a disk).
(2) Prove that of all the functions f(z) analytic in the disk
lzl ~ R and satisfying the condition
2n
f
0
if(Re1"')! 2 dcp = M
the linear function reali3es the mapping of the disk onto the domain
of least area. Find this area if f(O) = 0.
I
lf(z)i ~ 1-az
Z-IX I
is satisfied.
86 PROBLEMS ON COMPLEX .ANALYSIS
1-ii.z
HINT. Consider the function --/(z).
Z-13'
1 2 3 4 5 6 _n_ ~
<4> 2' 3' 4' 5' 6' 7' ... , n+l ' ...
(1) 1(.!..)
n
=!(-.!..)n = ~.
n"'
(n a na.tura.l number).
567*. The function f(z) = u (a:, y)+iv (a:, y) is analytic at the point
z0 = z0 +i110 a.nd f(z0 ) = c0 Prove that
z+z-,~
f(z)=2u (- z+zo) -c0
0
2
88 PROBLEMS ON COMPLEX ANALYSIS
1. Laurent series
In problems 573-588 expand the given function in a Laurent
series either in the given ring or in the neighbourhood of the given
point. In the latter case determine the domain within which the
expansion holds.
573. l/(z-2) in the neighbourhood of the points z = 0 and z = oo.
574. l/(z-a)" (a ..P 0, k a natural number) in the neighbourhood
of the points z = 0 and z = oo.
575. l/(z(l-z)) in the neighbourhood of the points z = 0, z = I,
Z= oo.
576. l/((z-a) (z-b)) (0 <!al< !bl) in the neighbourhood of the
points z = 0, z =a, z = oo and in the ring !al< izl <lb!.
577. (z1L2z+5)/((z-2)(z11 +1)) in the neighbourhood of the point
z = 2 and in the ring I < izl < 2.
578. I/ (z1 +I )11 in the neighbourhood of the points z = iand z = oo.
579. y((z-a) (z-b)) (!bl ~ !al) in the neighbourhood of the point
z = oo (consider both branches of the function).
580. /(z) =-VCz-l;(z- 2)) (Imf(i) >0) in the ring l<lz1<2.
1
581. z11 e in the neighbourhood of the points z = 0 and z = oo.
1
582. e1=s" in the neighbourhood of the points z = 1 and z = oo.
1
583. es+ in the domain 0 < izl < oo.
584. sin z sin 1/z in the domain 0 < izl < oo.
585. sin (z/(1-z)) in the neighbourhood of the points z = 1 and
z = oo (in the latter case restrict yourself to the first four terms
of the series).
89
90 PROBLEMS ON OOMPLEX ANALYSIS
(2) Prove that the formula for the area S still applies when
f(z) is analytic only in the domain r < lzl < R; in this case both
sides of the equation may tend simultaneously to oo.
HINT. See problems 536 and 537.
592. The functionf(z) is univalent in the domain lzl >1 and in
this domain can be expanded in the Laurent series
C-1
f( z) =z+-+-C-2
2 + ...
z z
Prove that
co
_}; n lc- 11 12 =::;; 1 ,
n=l
and explain the geometrical meaning of this inequality (the area
theorem).
HINT. Use the fact that the area S,, bounded by the image of the circle
lzl= r > I, is given by (f(:z:) = u + w)
o~s, = J udv = J2" !+l ~ (of - o</Iaf)d.
22io</I
Izj=r 0
Explain onto what domain the unit disk 1211 < 1 is mapped by this
function.
595*. Prove that if the function f(z) is schlicht in the disk lzl < 1
andf(O) = O,J'(O) = 1, then the domain G onto which this function
maps the unit disk always contains the circle with centre at the
coordinate origin and of radius 1/4 (Kobe's theorem).
596. (1) Prove tha.t if the function f(z) satisfies the conditions
of the preceding problem and the points w1 and w1 lie on the boundary
of the domain G, where arg w1 = arg w1 +n, then
lw1 1+ lw1 1 ~ 1 (Szego's theorem).
HINT. Apply Kobe's theorem to the function l-~~:~/wi .
(2) Prove that if the points w1 and w8 satisfy the cofulition
arg w 8 = argw1 +cc(O ~cc~ n), then
lw1l+lwsl ~sin~.
REMABK. This result strengthens Kobe's theorem only for ' > n/3.
2. Singular points of single-valued analytic functions
In problems 597-630 find the singular points of the functions,
explain their nature and investigate the behaviour of the function
at infinityt.
1
597. --8.
Z-2'
z'
598. l+z'.
z&
599. (1-z)I
2
619. cotz--.
1 1 z
604. S-1
e- - -z. 1
620. - - - -
605.----
as sinz-sina
21(1-e-=) 1
621. .
1-e" cosz+cos a
606. l+e=.
. 1
622. sm--. 1 -z
607. tanhz.
1 1
608. e-~. 623. cot-.
1
z
609. ze. 1 1
z 624. cot- - -.
610. el-. z z
:-.!.
611. e " 625. sin..!:..+~.
z z
1
ez-1 626. e-z cos-.
1
612. -.--- .
e-1 z
1
1
613. -.-.
627. e00 t.
smz 1
628. etan.
614. co:z.
z
615. tanz.
629. sin(~). Slll-
z
616. tan1 z.
1
617. co~z . 630. s i n (l-)
z cos-
21
1
618. cotz--.
z
631. Let P,.(z) and Qm(z) be polynomials of the nth and mth
degree respectively. Describe the behaviour at infinity of the fol-
lowing functions:
(1) P,.(z)+Qm(z); (2) P,.(21)/Qm(z); (3) P 11 (21)Qm(z).
94 PROBLEMS ON COMPLEX ANALYSIS
n=O
Prove that if lim !JI 1 = R-r, then the point z = Re'11 is a sin-
~ r lb.I
gularity of the function f(z), and if lim !J/ > R-r then it is
~ r Jb.I
a regular point.
HINT. Use the fact that a power series always has at least one singular
point on the boundary of its circle of convergence.
642. Prove Pringsheim's theorem:
00
of the series 2 c,,z!' = f(z) there is just one pole of the function
n=O
f(z), then the series diverges at all the points of this circumference.
HINT. Using the fact that from the convergence of the series at just
one point of the circumference of the circle of convergence it follows tha.t
lim c,,R11 = 0 (R is the radius of convergence), prove that if (z'I = R then
n-+O
the radial limit lim (z' -z)/(z) == O.
z-+-z 1
00
64lS. Prove that if the power series f(z) = 2 c,,z" has on the cir-
n=O
cumference of the circle of convergence only one singular point z0 ,
Anm-1
a. pole of order m, then c,, = 11 [I +<P (n)] where lim t/J(n) = 0,
Zo n-o<>o
and A is a constant.
HINT. Expand in a series of powers of z the difference between the function
/(z) and the principal part of the expansion of /(z) in the neighbourhood of
the pole z..
00
c11
RElllARK. Hence, in particular, it follows that the formula R = lim - - -
n--.oo 0 n+l
for the radius of convergence always holds if the only singular point on the
circle of convergence is a pole.
98 l'ROBLEMS ON CO:Ml'LEX ANALYSIS
/(z) = 2 CnZ"
n=O
t On the subject of this group of problems see, for example, [I, Chapter
VII I] or B. YA. LEVIN, The diatribution of the roots of integral functiona
(Raspredeleniye lcornei tselylch funktsii), Gostekhizdat, 1956.
INTEGRAL FUNOTIONS 99
(3) if /(z) is a transcendental integral function, then
660*. 2 (2m~nn)
n=O
I (m a natural number).
661. es.
1
662* JezPdt.
0
663. The integral function/(z) is of order(! and type a (0~0'~ oo).
Prove that the function P(z)/(z)+Q(z), where P(z) and Q(z) a.re
any polynomials is also of order (! and type a.
864. The integral functions f 1 (z) and f 2 (z) are of orders (!1 and (! 2
respectively, where (!1 ::p (!a What can be said of the order e of
the functions Ji (z) f 2(z) and f 1 (z) +f2 (z) 1
665. The integral functions f 1 (z) and f 2 (z) are of the same order
(! and are respectively of type& equal to a 1 and 0'2 , where 0'1 #: a 2.
What can be said of the order (!* and type a of the functions:
(1) f 1(z)f2 (z); (2) / 1 (z)+/2{z)1
666. The integral functions Ji {z) and f 2 (z) a.re of the same order
(! and of the same type a. What can be said of the order e and type
a of the functions :
(1) f1{z)f2(z); (2) f1{z)+/2{z}1
667*. Prove that the order and type of an integral function a.re
not changed on differentiating the function.
Solve problems 668-675 by the use of the following theorem.
If the Taylor expansion of an integral f1mction is of the form
100 PROBLEMS ON OOMPLEX ANALYSIS
00
f(z) = 2 onzn, then the order and type of this function a.re given by
nO
the formulae :
.!.
11 - ( .!.
11 n )
(O'ee) = lim n ylOnl .
D-+00
670. f(z) = 2(
n=l
lo!n ) 0 zn (a> O).
oo n
2
00
~ zn
673. f(z) = L.J n,r.+n (a> 0).
n-1
INTEGRAL FUNCTIONS 101
co
co
-v ~ (-l)nzlln
675. z Jv(z) = .L.J nlI'(n+,,+l) (11 > -1; Jv (z) is a Bessel
n-0
function of the 11th order).
I. Series of functions
In problems 683-692 find the domains of convergence of the
given series.
00
00
'\1 Zn
690. L,; 1-zn
n=l
2
00 00
'\1 .,.n
686. es ios n 691. L,; 1 ~zll"
n-t n-o
687. 2 s~nz.
n-t
00
692
200
n-1
n
z
(4+z)(4+z2) ... (4+zn)
00
102
VA.BIOUS SERIES OF FUNCTIONS 103
00 00
00
697 2
n-1
(1-z");;_zn+l) (izl : 1).
698. 200
n Z
2"
n=l fl (l+zs")
k-0
104 PROBLEMS ON COMPLEX ANALYSIS
00
12 /rc(z)I < B is satisfied for all n > N(e) and for all points z of the set E.
k=n
700. Prove the propositions:
00
should be satisfied.
00
702. Prove:
In order that the sequence of continuous functions {!n(z)} should
converge uniformly on the bounded closed set E, it is necessary
and sufficient that this sequence should converge at all the points
of this set and that it should converge continuously at all the limit
points of the set E, that is, that for every sequence of points Zn,
belonging to the set E and converging to the point z0 ,
703. 2 ~8 (zn+ ~ ).
n-1
706. 2n-1
(sin nz)/n1
2
00 00
704.
n-0
e-ns. 707. 2
n~l
(sinnz)/n.
2
00
705. e-slosn
n=l
n-1
circle Iz I : : ; ; 1. Does the series obtained by term-by-term differen-
tiation converge uniformly in the circle Iz I < 1 ~
oo (-l)n-1
709. Prove that the series ,l: z+n converges uniformly in
n=l
any finite pa.rt of the plane, from which there has been removed
disks of arbitrarily small radius e with centres at the points z = 0,
-1, -2, ...
Prove that this series does not converge absolutely at a single
point.
oo n
710. Prove that the series }; : converges uniformly in the
n=l
interval (-1, 0), and the series }; 1 1' ~I converges in the same in-
n=l I
n=l
n
cannot be ma.jorised
711. (1) The series ,L: (I ~z2f converges absolutely for izl ;;:,: 0,
n=O
largzi ::::;;; n/4 (these values of z do not exhaust the whole of the domain
of absolute convergence which, as is easily seen, consists of the point
z = 0 and the outside of the lemniscate I I+z2 1 =I). Prove that
the series converges non-uniformly in the given domain.
106 PROBLEMS ON COMPLEX ANALYSIS
REMA.RX. This shows that uniform convergence does not follow even
from the absolute convergence of a series in a closed region.
(lO
(2) Prove that in the same region the series ,L; c\-l~:;n
+
con-
n=O
verges uniformly and absolutely, but not absolutely uniformly
(that is, the series of moduli does not converge uniformly).
(lO
every closed region inside the domain G, then the series 21/~(z)i
n=l
also possesses the same property.
2. Dirichlet seriest
(lO
Series of the form 2 ane -A,,z, where the an are complex coefficients
n-1
and the An are real non-negative numbers, satisfying the conditions
A1 < A1 < ... and lim An = oo,
II-co
are known as Diriokke Bef'iea.
713. Prove that if a Dirichlet series converges at the point
z0 = :r:0 +iy0 , then it converges at all the points of the half-plane
Re z > Re z0 , the convergence being uniform in every angle
Iarg(z-z0 )1 ~ () < n/2.
HINT. Apply Abel's transformation to the sum
q q .. )
~ -A,,z _ ~ -Anzo -AnCz-z0
L.J ane - L.J ane e
11-=p n==p
and use the inequality (a< b, te =a:+ iy)
b
je-az-e-111 = jz J e-l:tdtl~~(e-u-e-llx).
a a;
t On the subject of this group of problems see, for example, [l, Chapter IV,
I].
VARIOUS SERIES OF FUNCTIONS 107
is a half-plane Re z > a:0 (a:a;;;a. - oo), the series either converging absolutely
on the whole of the straight line Re z = a:0 , or not converging absolutely at
a single point of this straight line. The numbers a:c and a:0 are called respectively
the abBciaBa of oonoorgenoe and the absciaaa of abaolul.e cont1ergence of the Dirichlet
series.
In problems 715-721 find the abscissas of convergence (zc) and
the abscissas of absolute convergence (z0 ) of the given series.
2 2
00 00
2(
00
2 ~.
00
2 (Y~n
00
2
00
2
00
718. (-l)ne-Zlo1101n,
n-2
722. Prove that if lim (log n) /An :-- 0, then
2 2(
00 00
2(
00 00
725. 2 ~3
n=l
e-s 101 n 728.
n=l
-I)lvnJ
n
e-"'n.
00
726. 2~
n=l
e-zn.
-A z
The series Ioo ane n , where tJ.e An are complex numbers (ll1 1< IA.I <
n-1
<: ... ; lim A.a = oo ), is known as a generalised Dirichlee series.
n-+00
-
Prove that if lim (logn)/An = 0, then the series converges absolutely
within the domain G, the points z = x+iy of which satisfy fo1
any <f> the condition zcosef>-ysin<f>-k(c/J) >0, and diverges at
every point lying outside G.
3. Parametric integrals
731. Prove the theorem:
Let 0 be a simple contour (closed or not), possessing a finite
length, and f(T, z) a function, analytic with respect to the variable
z and continuous with respect to T for every z of some domain D and
for all the points T, belonging to the contour 0. Then the function
represented by the integral
F(z) = Jf(r:, z)dr:,
c
is an analytic function of the variable z and
If the integral Jf (T, z)dT is improper, that is, if the integrand has dis
c
continuities at some isolated values TE 0 or the contour of integration includes
the point at infinity, then the definitions of convergence and uniform conver
gence of such an integral are exactly similar to the corresponding definitions
given in books on mathematical analysis.
732. Prove that for the uniform convergence of the integral
J f(T, z)dT on the set E with respect to some point To#= oo of the
c
contour 0 it is necessary and sufficient that for any e > 0 there
should exist a number o(e) such that
for all the points z of the set E and for every arc 0 a of the contour 0,
lying in a a-neighbourhood of the point To and not containing this
point either as an interior or as an end point.
733. Formulate and prove a similar criterion for uniform conver-
gence of the integral if To = oo. Consider the cases when the contour
0 is not bounded at one end, or at both ends.
734. Prove that if lf(T, z)j ~ IVJ(T)I for all the points z of the set
f f
E and if IVJ(T)ldT converges then the integral f(T, z)dT converges
c c
uniformly on the set E.
735. Let f(T, z) be a function analytic with respect to z and con-
tinuous with respect to T for all the points z, belonging to some
domain D, and points T, belonging to the contour 0, with the
exception of some isolated points of it where the conditions imposed
on the function f(T, z) are not satisfied either for all points z, or
only for some of them.
Prove that if the improper integral
F' (z) =
c
Jazaf dT '
Jts-
00
J
00 c+loo
e'
737.
0
e-t'dt. 741.
f
c-loo
-dt
t
(c:FO).
738. J t00
sint dt
. 742. J
c+loo
-e'
t
dt (c:F 0).
0 c
J c;t J z'
c+loo
00
740. J 00
-sintz
, - dt.
(c ::/: 0, z'=etlOIS).
where the function /(t) is integrable on the segment [O, a] for any positive
a< oo is called a Laplace integral.
744. Prove the following propositions:
(1) If the integral (1) converges at the point z = z0 then it
converges in the half-plane Re z >Re z0 , the convergence being
uniform in the angle la.rg (z-z0)1 :,; ; (J < n/2.
(2) If the integral (1) converges absolutely for z = z0 then it
converges absolutely and uniformly in the half-plane Re z ~Re z0
(3) If lim loglf(t)I = p, then the integral (1) converges
t-+OO t
absolutely in the half-plane Re z >fl and uniformly in every half-plane
Rez ~ fl+e (e > 0) (constructa.n example of a La.place integral which
converges absolutely in the whole plane and for which fl= oo).
(4) If lim log lf(t)I =ex, then the integral (1) does not con-
t-.oo t
verge absolutely at a single point of the half-plane Re z < oc.
INFINITE PRODUCTS 111
It follows from the theorem formulated in problem '744 that the domains
of convergence and of absolute convergence of a Laplace integral (if they
exist) are the half-planes Re z > a:c and Re z > a:0 ; the number a:c is called
the absciasa of convergence, and a:0 is called the abacisaa of absolute convergence
of the Laplace integral.
co
In problems 745-751 find Xe and x 0 for the integral [ e_.' /(t) d.t,
where f(t) is the given function.
745. /(t) = 1. 746. /(t) = e-t. 747. /(t) = e'".
748. f(t) = e-t for 0 < t < log log 3 and
<
log log 2le t <log log (2le+l) (le= 2, 3, ... ) and f(t) = -e-t for
<
log log (2le+l) t <log log (2le+2) (le= 1, 2, ... ).
<
749. /(t) = eie' for 0 t < log log 3 and
<
log log 2le t <log log (2le+l) (le= 2, 3, ... ) and f(t) = -eie' for
log log (2le+l) < t <log log (2le+2) (le= 1, 2, ... ).
750. /(t) = e8 ' for 0 < t < log log 3 and
log log 2le ~ t < log log(2le+ 1) (le = 2, 3, ... ) and f(t) = -ee' for
log log (2le+l) ~ t <log log (2le+2) (le= 1, 2, ... ).
751. /(t) = e' for log (2le-l) ~ t <log 2le (le= 1, 2, ... ) and
f(t) = -e' for log 2le <
t <log (2le+l) (le= 1, 2, ... ).
In problems 752-755 investigate the convergence of the Laplace
00
756. n(
=2
00
1- :. ) = !.
112 PROBLEMS ON COMPLEX ANALYSIS
757. n(1+
00
n
n=l
758.
n=S
00
n2_4 1
n2-l = 4. 761. nn=l
oo [
l+ (-1)+1]
n = l.
00
constant.
n=l
o: sm o:
oos 2n = -o:-
'J'C J::OI ( 2n 2n )
2 =
n=l
2n-l 2n+l
766. Prove that if, as usual, we take --'J'C < argpn ~ 'J'C, then
00
series l; logpn.
n=l
767. Examine whether the assertion of the preceding problem
is still true if it is assumed that:
(l) 0 ~ argpn < 2'J'C; (2) o: < argpn o:+2n (o: < 0). <
INFINITE PRODUOTS 113
product
n-1
n (1 +an) (that is, for the absolute C!>nvergence of the
00
00
(1) n00
n-1
(Pn+qn); (2) n00
11=1
(pn-qn); (3) n00
n=l
p 11 q11 ;
00
(4) nPn.
n=l qn
In problems 770-774 investigate the convergence and absolute
convergence of the given infinite products.
771. nn-1
00
774, n
n=l
00
n 00
n=O
11
(l+zll ) = - - - ,
1
1-z
776. 11
11=1
(1-zn). 777. fJ(1--;:).
11-1
114 :PROBLEMS ON OOMPLEX ANALYSIS
00
r
n=l
784. n 00
n=l
(l+cnz), if it is known that the series
00
2: lcnl converges.
n=l
.fi'(z) = n00
n=l
[l+fn(z)]
I'(z-!--1) = n-- --
00
n=l
n
z+n
( n+l
n
)%
INFINITE PRODUCTS 115
converges absolutely in the whole plane, except for negative integral
values of z, and represents a function analytic in the whole plane
except at the points z = - 1, - 2, ...
788. Prove Euler's formula
.
r(z) = lun n! n" (n=e
r:lo1n)
n-+oo z(z+l)(z+2) ... (z+n)
and show that:
(1) I'(z + 1) = zI'(z);
(2) r (m + 1) =ml, if m is a natural number.
789. Prove that
n 00
n-1
n(a+/J+n)
(a+nH/J+n) -
_ I'(a+l)I'(/J+l) _
I'(a+/J+l)
(a,p #: -1, -2, ... ).
1
-=---=ec"
I'(z+l)
noo (l-1-- e _!..n, z)
' n
n=l
where 0 is Euler's constant.
HINT. Use the solution of problem '762.
791*. Let p 1 , p 2, Pn ... be the sequence of all the prime
00
n=l
n c1-p;)
(2) The function C(a) has no zeros in the half-plane
Rea> 1.
REHAB.JC. An extensive literature is devoted to Riemann's Zeta function.
See, for example, the monograph: E. C. TITOBMARSH, The zeta-function,
Cambridge Tracts, No. 26.
00
792*. Prove that the series .L; _!_, where {Pn} is the sequence
n=l Pn
of prime numbers, diverges.
CHAPTER Vll
116
RESIDUES AND TREIB APPLICATIONS 117
In problems 814-821 it is required to find the residues of each
of the single-valued branches of the corresponding many-valued
functions at the given points.
814. lyz at the point z = 1.
-z
825. Find res [ <f>(z) ~(~i l=a if <f>(z) is analytic at the point a and:
(1) a is a zero of order n of the function f(z);
(2) a is a pole of order n of the function f(z).
826. Find res{f[</>(z)]}s=a if the function <f>(z) is analytic at the
point a and </>'(a)= 0, andf(C) has a simplepoleatthepoint C= <f>(a)
with residue A.
82'7. At the point a the function <f>(z) has a simple pole with
residue A, and J(C) has at infinity a simple pole with principal part
BC. Find
res{f[<f>(z)]}z=a.
828. The function f(z), which assumes real values on the arc l
of the circle Iz-a I = R, is continued analytically through this arc
by the symmetry principle. Let the point z = fJ ({J = a) be for /(z)
a pole of order k with principal part
k
'\, c_.
LJ
11=1
(z-fJ).
832. f 2~:1
c
where a is the circle lzl =I.
883. f z(z~9)
c
dz, where a is the circle lzl =I.
834. 21 .
ni
c
f sin2:..dz, where
z
a is the circle lzl = r.
1 j dz
889. 2n;, c (z'+l)y'(z+l) (y'l = 1), where 0 is the pa1abola
'If"= x, traversed in the direction of increasing y.
840 - 1 -J dz
2ni c a sin nz
(a= e.1 0 ), wherel a> 0, and 0 is the
straight line x = IX, 0 < IX < 1, traversed from below upwards.
HINT. Consider _l_ J ck
!:iii y a=sinnz
, where the contour y is indicated in
841. 21 . J
e=dz , where the contour of integration 0 is indicated
nt c cosz
in Fig. 14.
120 PROBLEMS ON OOMPLEX ANALYSIS
If the function f(a:) becomes infinite at a: = c (a < c < b), then the Cauchy
b
principal value of the integral Jf(a:) dz is defined by
a
-+0
lim [7/(a:)dz+
a
j f (a:)dz]
c+
This definition is generalised to the case of a curvilinear integral in an
obvious manner.
g
tt+t -r
D c
!/
ft I z Z+tt
:& -
z
t
A
tZ-i,8
- 8
FIG. 13
-c
FIG. 14
-ot
If the functionf(a:) is continuous on the whole of the real axis, the principal
oo N
value of the integral J f(a:) dz is defined as lim J f(a:) dz.
-oo N-+oo -N
In problems 842-849 find the definite integrals. If an integral
is improper and diverges, find its principal value (if it exists).
f
In
.,.
HINT. Put eld> = z.
843. f
0
def>
(a+b COS c/>)1
(a> b > 0).
,. def>
844
J (a+b cos 2 cf>)8
(a > 0, b > 0).
0
.,.
845. J
0
1 _ 2a:!cf>+as (a is a complex number and a~l).
RESIDUES AND THEIR .APPLICATIONS 121
846
J 2n
cos1 3</> def>
l-2acoscf>+a1
(a is a complex number and a#: 1) .
0
2n
84'7. J
0
e008 ~cos (ncf>-sin </>)def> (n is an integer).
..
848. J
0
tan (x+ia)cb; (a is a real number).
2n
849. J
0
cot (x+a)cb; (a is a complex number and Im a#: 0).
FIG. 15
where 0 is the contour represented in Fig. 15, and let the radii of the arcs
of the small circles tend to zero. In calculating the integral along the vertical
segment divide it into two and by corresponding substitutions reduce them
122 PROBLEMS ON OOMPLEX ANALYSIS
to Eulerian integrals of the first kind; also use the known relation
B(p, q) = i;![)I'(~) and the formula I'(p)I'(l-p) = ~.
p+q smnp
00
zldx
852.
f 0
(zl+a11) 1
(a> 0).
co
853.
f
0
(zl:l)"
(n is a natural number).
co
854. J (zl+a~za+bl)
-oo
(a > 0, b > 0).
00
855. f zl+l
x'+l dx.
0
1 f
856. Prove that
dT ilc-n-l(n+lc-2) I
2ni T"r = (2k)n+lc-l(lc-l)! (n-l)! (na.ndlca.rena.tura.lnumbers),
c
where 0 is a straight line parallel to the real a.xis which cuts off on
the imaginary a.xis a segment equal to k (k > 0).
857. Evaluate the integral
1
2ni
f dT
T"(T-z) (le is a natural number),
c
where 0 is the contour of the preceding problem.
f
00
858 (l)
-oo
zcoszdx
z 11_2z+IO ' (2) f
co
zsinzdx
zl-2z+10.
RESIDUES AND THEIR APPLIOATIONS 123
f
00
xsinxdx
859.
x8+4x+20
-oo
f
00
cosa:x:
860. xB+b dx (a and b are real numbers).
-00
00
861
.
-oo
f x sin a:x: d
x8+b1 x (a and b are real numbers).
862. Let f(z) = e1m"F(z), where m > 0 and the function F(z) pos-
sesses the following properties :
(1) In the upper half-plane it has a finite number of singu-
larities a 1 , a8 , , a,. ;
(2) It is analytic at all the points of the real axis, except
the points x1 , x2 , , Xm, which are simple poles;
(3) F(z)-+ 0 if z-+ oo and Imz ~ 0.
Prove that
oo n m
-oo
J f(x) dx = 2:n:i{2 resU(z)]:r=ak + ! }; resU(z)]:r-~t},
k-1 k=l
f
00
863.
_ 00
J e''~
- d x (t is a real
x number).
865.
-oo
sinxdx
(x1 +4)(x-1) '
864 1 x -5x+6
00
xcosxdx.
2
866. f
00
cosax
l+x3 dx (a~ 0).
-oo -co
In problems 867-872 evaluate the given integrals (a and b are
real numbers).
f
00
sina:x:dx
868 x(x1 +b8) '
0
124 PROBLEMS ON OOMPLEX ANALYSIS
co
8'71.
0
f s:x ch.
S'70. J co
cos2aa:;;cos2bx dx.
0
!I
FIG. 16.
BniT. Use the integral Jells-l dz, where the contour 0 is indicated
C z
in Fig. 16.
co
8'72. f fi:8
0
sin8 x
ch .
co
(2) Jzl'- sin aa:ch (a> 0, -1<p<1).
0
1
FIG. 17.
co
874. f cos x"dx
0
(p > 1).
co
875. f sin x"clx
0
(IPI > 1).
877. Let the rational function f(z) have no poles on the positive
part of the real axis and
lim [z"f(z)] = lim [z"f(z)] = 0.
z-0 Z-+CO
Prove that
CO 11
J
o
zP-1f(x)dx = ~
smpn k~l
.2 res [(-z)"-1 f(z)]z=et1:
:where ix1 , ix1 , ... , 'n are the poles of the function f(z), pis an integer.
IIINT. Consider the integral
878. Evaluate J
0
x"(:l) (0<p<1).
n
879. Prove that I'(a)I'(l-a) = - . - - (0 < a < 1).
Blll na
126 PROBLEMS ON COMPLEX ANALYSIS
FIG. 18
HI:NT. Use the following known relation between the Beta and Gamma
functions:
I'(a)I'(b) == I'(a+b)B(a, b)
and in the integral defining the Beta funotion
I
B(a,b) = J:i;0-1(1-z)ll-ldz,
0
880. f
0
x"dx
l+xl (-l<p<l).
CCI
881. f
0
x"dx
(I+x9)1 (-1 <p< 3).
f
CCI
882. x-Pdx 1 A )
xl+2xcosii.+I (-l<p< ,-n< <n.
0
RESIDUES .AND THEIR .APPLIOATIONS 127
883. On the positive real axis let the rational function /(z) have
only poles of the first order and let
lim [z"f(z)] = lim [z"/(z)] = 0.
z...0 z~oo
Prove that
884. Jr W:/(1-a:)
0
1 co < p < I).
00
886. J
0
x1-P(l-z)P
(l+x)a cb (-1 < p < 2).
!I
-
FIG. 19
128 PROBLEMS ON OOMPI.EX ANALYSIS
887. f xl-P(l-x)"
1+xl dx (-l<p<2).
HINT. Prove
0
. Jzl-P(l-z)P
that llin I I
.
dz = 2me-.rmJ, where a is the
R-+oocR +z
circle lzl = R traversed in the positive direction.
1
888. f
0
xl-P(l-x)"
(l+x) dx (-1<p<2).
890.
0 (x+l)
dx
(x1 (1-x))
.
891. Evaluate the integral J l dx
(x-a)y'(l-xl)' where y'(l-xl)>O
when -1<x<1, a is a complex number and a =F 1. In par-
ticular, find the values of the integral for a= e"' (0 <ix< n),
a= iy, a< -1 and - l <a< 1 (the principal value).
1
893. In=
f
0
x2ndx
(l+xl)y'(l-xl) (n = 0, 1, 2, ... ).
895. In = fV
0
dx
(1-z")
(n = 2, 3, ... ).
....
sisting of segments along the radius vectors of the points 1, co, co, .. ., con-l,
where co= en, and along the circle lzl = R.
896
f log xdx
xl+a1
0
...._,_ e use or t he mte
HINT........... gral f logzdz
zl+a , wh ere t he contour a .
1s given
c
in Fig. 20.
g
FIG. 20
897 J 00
log1 xdx
xl+a1
I
J ez+1) x
00
898. f
0
logxdx
(xl+al)I VX
901.
I
oo ( dx
log - - - --
eX-1
00
f
logxdx
899. o (x+l) yx.
902. Let f(z) be a rational function which has no poles on the
positive part of the real axis or at the point z = 0, and is such that
130 PROBLEMS ON COMPLEX ANALYSIS
J oo f(x)dx
-logl xl+n1 =
~
L.J
[ f(z)
res Log z-ni
Jz-ak
0 k-1
where iii= -1, and a2, aa ... ,an are poles of the function f(z)
different from -1, and Log z =log lzl+i arg z, 0::::;;; arg z < 2n.
HINT. Consider the mtegral ~
1
:inc
J L f(z) wo
ogz-:in
-"- ~here the contour 0 IS
FIG. 21
903. (1) f
0
dx
(x+a) (loglx+31:11 )
j'
00
904 ----,---,,----dx_ _ __
(xl+a1)[log x+ (2n+ 1)131;1]
0
HINT. Make use of the integral
r 1 [ I
z1 +a1 Log z- (2n+ l)ni
1
+ Log z- (2n-l)ni
c
+ + Logz+(~n-l)m]m
RESIDUES AND THEIR APPLICATIONS 131
where the contour 0 is given in Fig. 21, and the branch of Log z is chosen as
in problem 902.
co
905 J
0
dz
(xl+a1)([log x]1 +4n1 xl)
HINT. Make use of the integral
c
Jz+a I [ I I
Logz-2mri+ Logz-(2n-2):ni +
FIG. 22 FIG. 23
906. Let f(z) be a rational function which has no poles on the open
contour 0, the first point of which is a and last b. Prove that
where the summation extends over all the poles of the function f(z)
132 PROBLEMS ON COMPLEX ANALYSIS
907 f co sinaxdx
sinhx
0
-- -
- f
co
' -c
FIG. 24
co
908. f
0
xcosaxdx
sinhx 909.
0
J cosh a.=_dx
coshnz
(-n <a < n).
.
HINT. Make use of the mtegral J-col$hns
c
e"dz
- - , where 0 is the perimeter
2~i J
c,
e"t<f>(z)dz = 2
k=l
res[e"t</>(z)]z,.ak.
2ni
I J~Y+l ezdz I
= I'('1+1)'
)'
FIG. 25
(z+I)y(z+2)
""' .
927. _ I
2:n:i
f etdz
z(I-e- 0 ")
(a> 0).
c,
HINT. Make use of the expansion 1
1-e oz
= I+ e-oz + e-llloz + ...
928. -
I f etz-Jtlfz dz (x > 0).
2:n:i z
c,
RESIDUES AND TREIB APPLICATIONS 135
HmT. Replace 0 1 by the contour shown in Fig. 26.
FIG. 26
929 . _1_
2ni
J fl' sinh ry21 dz
rz sinhayz
931. ~ffl'log(z+l)dz.
2ni z-1
c, c,
(a >r >0). co ot
930 ~
2m
J lo~(l+z)as' dz.
z
932. J Je:~
0
dt
c,
dz (a >0).
c,
HINT. Change the order of integration.
933.
0
HINT. Make use of the fact that J-z
e- -dz = 0 for u > 0.
c,
934. From the series expansion of the Bessel function
J,,(z) =
~
L.J
(-l)t
klI'(k+v+l)
(z2 )v+lt '
k=O
deduce the following integral representations ( y is the contour given
in problem 920):
1
(I) 2ni f c+i.. dC (z)
eC-it
=
2
J,(z);
136 PROBLEMS ON COMPLEX .ANALYSIS
HINT. Expand the function e - '' into a power series and use the solu
tions of problems 910 and HI.
-n
n
FIG. 27
f
J,(z} = __!__ eliratnt-fvtdC,
2:n; II
where ll is the contour given in Fig. 27, and hence obtain that for
evecy integer n
.
J.(z} = ~ J
0
cos(zsinC-nC}dC.
J
00
J J
00 00
J
00
sin t JI (xi- a2)
938. cos bx V(xD-aB) dx (t > lbl}.
0
RESIDUES A.ND THEm APPLIOATIONS 137
HINT. Make use of the fact that
ut
sinut
u
= y/(!!!...)J
2u l
(ut) = y'(nt) _l_
2 2ni
Jez-~ dz
?!la
c,
(see problem 93'), and change the order of integration.
The asymptotic behaviour of integral,at
939. Let the analytic function </>(z) have on the left of 0 1 only
a finite number of singularities, all of them being poles, and let
<f>(z) -+ 0 as z -+ oo and Re z :,;;;; ct. Let us put
J
2~ic, e%'</>(z)dz"' .2res [e%'<f>(z)],
holds for large values of t, the summation extending over all the
singularities of </>(z) with negative real part.
REMA.Rx. The functions f (t) and F (t) are asymptotically equal as t -+ co
(f(t)-F(t)), if lim f(t)/F(t) = 1.
f-+00
941. Investigate the asymptotic behaviour as t-+ oo of the
function
l e:r:'dz
f(t) = 2ni 2!2(z+a)3
J
(Rea> 0).
c,
t On the subject of this group of problems and also on the question of
the application of asymptotic estimates and other methods for obtaining them
see, for example, (3, Chapter V, 3]; B. A. FucHs and V. I. LEvlN, Fuoo-
tionB of a oomple3: variable and some of their applications (Funktaii komplekmogo
peremennogo i nekotoryye ikh prilozheniya), Chap. IV, Gostekhizdat, 1951
English translation published by Pergamon Press (1961); M. A. EvGRAFOV
Asymptotic e8timatu and integral Junetions (Asimptoticheskiye otaenki i taelyyl!i
Junktaii), Gostekhizdat, (1957).
138 PROBLEMS ON COMPLEX ANALYSIS
f()
t =-
I
2ni
f zen-"(z+aa)
(z-wi)y(z2+2az)
dz (w > 0, a> 0).
C1
(From this it does not even follow that the series converges!)
Asymptotic expansions of a more general form are also frequently consider-
ed. Let {q11 (z)} be a.n arbitrary sequence of functions such that lim q11+1( ()z) = O,
Z-+00 qll z
and { 11 (z)} a sequence satisfying the conditions:
lim 1'11+1(z)
Z-+00 q11(z)
= 0, lim ,. 11(z)
Z-+00 q11(Z)
I> 0.
The series 1:"" c11 11 (z) is said to be a.n asymptotic expansion of the function
n=O
00
j(z),.... 2 c11 11 (z),
n=O
if
lim - 1 -[J(z)-
z-+oo q1:(z) n=O
i; ~11(z)] = 0 (n = O, 1, 2, ... ).
J
00
e-xt I 2! 4! 11 (2n)!
I+t2 dt.....,-;-7+7- ... +(-l) :x:Dn+i + ...
0
RESIDUES AND THEIR APPLICATIONS 139
HINT. Make use of the expansion
I (-I)n+lt2+2
- -=
I+t l-tD+t'- ... +(-I)nt2n+-'---'----
I+t
and estimate the remainder term.
f -,-dt,....,-;-7+-;a-- ...
00
e"-t 1 1 2 n!
11
+(-1) z11+1 + ...
h
f
00 00
e-'"dt ~ (2n) ! 1
t-x ,...,-Ji:n; .L.J 22n.n! x2n+l'
-co n-0
where for :x: > 0 the integral is understood as its principal value.
948*. Prove that
HS. For small t it is necessary to choose 0 1 such that at is greater than co,
and expand l/(11 +w8) in a series.
-
FIG. 28
1
2ni
J pJt
y(z)(z1 +l)dz-sm
( n)
t-4,..., yn .Lt (-l)
2 ~ n (4n)I ( 1
(2n)I 2yt
)'n+l
C1 n=O
951. Find the asymptotic expansion of the function
1
f(t) = 23ii
J endz
!. (i
8
-
FIG. 29
958. How many roots has the equation z = </>(z) in the circle
lzl < 1, if for izl ~ 1 the function </>(z) is analytic and satisfies the
inequality l</>(z)I < 1?
959. How many roots has the equation ez = az" in the cii'clti Jzj < R
(n a natural number), if !al> eR/R"?
960. Prove that in the right half-plane the equation
z = A.-e-z (A> 1)
has a unique (and hence real) root.
961*. Prove that no matter how small is (! > 0, for sufficiently
great n all the zeros of the function
I 1 I
f,.(z) =I+ z + 2!z2 -1 + n!z".
are situated inside the circle izl < (!
962. Prove that if (! < I then the polynomial
P,.(z) = I+2z+3z2+ . +nz"-1
for sufficiently great n has no zeros in the circle izi < (!.
HINT. Use the method of solution of problem 961.
963. The function </>(z) is meromorphic in the domain G and ana-
lytic on its boundary 0. Prove the following assertions:
(I) If l</>(z)i < 1 on 0, then the number of roots of the equa-
tion </>(z) = I situated inside the domain G is equal to the number
of poles of the function </>(z) in the domain G.
(2) If l</>(z)i >I on 0, then the number of roots of the equa-
tion <f>(z) =I situated in the domain G is equal to the number of
zeros of the function t/>(z) in the domain G.
(3) The assertions (I) and (2) remain true if the equation
</>(z) = I is replaced by the equation </>(z) =a, where Jal ?- I in case
(1) and 1~! < l in case (2).
Pn(Z) = zn ( l + : 1 + .. + :: )
and apply the principle of the argument to the semicircle lzl < R, Res > O
for a sufficiently large R.
96lJ. Find the number of zeros of the polynomial
z8 +z6 +6z4+1W+Sz1 +4z+1
in the right half-plane.
966. Find the number of roots of the equation
z'+2zs+sz+z+2 = 0
in the right half-plane and in the first quadrant.
967. How many roots in each quadrant has the equation
2z4-3z3 +3z1 -z+ l = 0?
968. In which quadrants are the roots of the following equation
to be found1
z4+zs+4z1 +2z+3 = 0.
969. Prove that the number of roots of the equation
z11n+'z-1+p = 0
(' and p are real numbers, ' = 0, p = O; n is a natural number),
which have positive real part is equal to n if n is even. However,
if n is odd, the number of them is n-1if'>0 and n+l if'< 0.
lhNT. Consider the increase of arg (zll+oczln-1+/JI) when the point 11.
describes the boundary of a right semicircle of large radius.
f(z) =f(O)+
~
~ An(-1-+_!_),
z-a,. an
n-1
2m
I f C<C-z>
J(C)dC
cm
and pass t.o the limit as m-+ oo.
REMARK. For various generalisations of this theorem see, for example,
[2, Chapter VIII, sec. 4].
In problems 995-1002 prove the validity of the expansions.
= -+ 2
00
1 2z
995. cotz z z1 -nBn
n-1
1
998 - = - +
sin z
1
z
2 00
(-l)n2z
z" -n11i"
.
n=l
00
~ 2z
99'7. tanz = ~-l
[ (2n-l)3t ]
2
I
n- - - - -z
2
00
~
1
9118. oosz ~"ft "-[ (2" 2 l)n
(-l)n(2n-l)
J.
00
~ 2z
999. tanhz = ~
_ ..
[
(2n-1)3t ] 11
n-
1
z+ 2
RESIDUES AND THEIR Al'PLIOATIONS 149
co
1002. -.-=
1 ~ 1 .
sm1 z (z-nn)1
n=--oo
max lf'(z)
zECm /(z)
I= o(Rm).
j(z) =f(O)e/CO>r:
/'(0)
nco (
n-1
1- :n ) "
ell,,
n-1
1 4:.n')
then
co m
}; f(n) = -n}; res [f(z) cotnz]z-arc.
n=-OO k=l
1012. Prove that if in the conditions of the preceding problem
the requirement (1) is replaced by the condition
lim
n-+00
Jf ~z) dz = 0 '
smnz
en
then
1013. ~
.L.J
n==-co
1 .
(a+n)1 1016. 2 n!a
n-o
00 00
~ (-l)n 1017. ~i (-l)n
1014. ,L.,; (a+n) .L.J n1+a1
n--oo n-o
00 00
1015. 2 (2n~l)ll.
n=O
1018
~
.L.J
n-o
(-l)n
(2n+l)B
RESIDUES AND THEm APPLIOATIONS 151
(-:n<b<:n).
J(a -z1
zellr:dz
1) sin nz'
c
1020. Prove that
co
"1
LJ
n
(n-3)y(4n-3) =
J
rata
xoot:nx l
(3-xl)y(3-4x8) rlx+6oot[:n(2-y(3)].
n=l o
!I
FIG. 30
On the contour 0 let <Cl satisfy a Lipschitz condition of order or: (0 < or;~ 1
(for brevity, '1(C) e Lip or:), that is,
l<Cil-tfi(Call < Tc!C1-Cal 11
where the points C1 and Ca belong to the contour 0, and le is a constant. Then
if a point of the contour Co is not an end point of it, there exists the singula
W6grGl
.F(Col = ~ j<CldC
2ni C C-Co
defined as the principal value of the integral of Cauchy type.
152
INTEGRALS OF OAUOHY TYPE 153
This principal value can be expressed in terms of ordinary improper integrals
by the formula
.F(Co) = ~J l/l(C}-1/l(Co) dC +~(Co>+ l/l(C~) Log b-Co (2)
2m c C-Co 2 2m a-C0
where the points a and b are the ends of the contour 0, if it is open. The single-
valued branch of Log is chosen so that for a closed contour (a = b) the term
with the logarithm vanishes and the formula assumes the form
If we denote by .F+(C0 ) and .F-(C0 ) the limiting values of the Cauchy type in-
tegral .F(z) as z-+ Co on the left of a and on the right of a respectively, then
I
by Sokhotskii's formulae
.F-(Co) = .F(Co)-21/l(Co)
or
l
.F"o) =2 [.F+(Co>+.F-(Co)], .F+(Co)-.F-(Co) = l/l(Co) (5)
If the contour 0 is closed and is traversed in the usual direction, then .F+(C)
is the limiting value of the function .F+(z), determined within the contour
(the domain D+), and .F-(C) is the function determined outside the contour
(the domain D-)t. (See, for example, [l, Chapter III, 3] or [3, Chapter m, 3]).
1021. Prove the formula (1) for n = 1, find the limit of the difference
F(z+k)-F(z) __1_
k 2ni c
Jc/J(C)df_
(C-z) 2
as k-+ 0.
1022. Prove formula (1) for any n by the method of mathematical
induction.
1023. Prove that if 0 is a closed contour and the density of the
Cauchy type integral
<P(z) = ~
2ni
f </>(C)
c
C-z
dC
can be represented in the form
c/J(C) = cfJ+(C)+c/J-(C)
t The case when the contour is an infinite curve dividing the plane into
two domains is also referred to this one.
154 PROBLEMS ON COMPLEX .ANALYSIS
where cf>+(C) and cf>-(C) are the boundary values offunctions analytic
respectively inside and outside the contour 0, then
= CS+iC2-C+4i
c/>(C) + log[(C-2)/(C-3n
C"-3'2-4 '2-4
and 0 is the circle !Cl = 3/2.
1027. Find F+(z) and F-(z), if c/>(C) = cot C and 0 is the circle
ICl=5.
1028. Find F+(z) and F-(z), if c/>(C) = C2/(C2+1) and 0 is the real
axis traversed from left to right.
REMABK. By an integral of Cauchy type taken along the real axis
F(z) =
l
-:n:i
2 J--dT,
oo l/>(T)
T-Z
-00
ef>(C) =a;+ .2
n=l
(ancosnO+bnsinnO)
11 (z) = 21 .
ni c
Jf(C,) cot(C-z)dC,
1
J 9 (z) =2 -.
f f(C)dC
---;---(,.-),
ni c sm .,-z
in domains, the points z of which possess the property that not
one of the points z+lcn(lc is an integer) lies on 0.
(2) Solve the problems formulated in part (1) on the assumption
that 0 is the circle !Cl = n.
1031. Let 0 be the segment [-1, l], traversed from left to right,
==
and ef>(C) 1. Find F(z) outside 0, the limiting values F::l:(C) and
the principal value of F(C) on 0. Evaluate in particular F(i),F::1:(0)
and F(O).
1032. Let 0 be the semicircle ICI = R, 0 < arg C< n (commencing
=
at the point R) and ef>(C) 1. Find F(z) outside 0, the limiting
values F::l:(C) and the principal value of F(C) on 0. Evaluate, in
particular, F(O), F=(iR) and F(iR). Find also F'(O).
1033. Let 0 be the semicircle IC! = R, -n < arg C< 0 (com-
mencing at the point R) and ef>(C) = 1. Find F(z) outside 0, the
limiting values of F1=(C) on 0, F(O) and F'(O).
1034. Let the density of a Cauchy type integral ef>(C) = 1gn. Find
F(z) outside 0, if the contour 0 is given by:
(1) The boundary of the ring r < !z! < R;
(2) The straight line Im C= n, traversed from left to right;
(3) The boundary of the strip [Imz[ < n;
(4) The semicircle !Cl = R, 0 < arg C< n (commencing at
the point R);
(5) The semicircle !Cl = R, -n < arg C< 0 (commencing at
the point R).
156 PROBLEMS ON OOMPLEX ANALYSIS
In parts (4) and (5) find the limiting values of F*(C) on 0 and
evaluate F(O).
In problems 1035-1040 find F(z) outside 0, assuming that the
contour 0 is an arc connecting the points a and b, and c/>(C) is the
given function.
1035. efJ(C) = I.
1036. t/J(C) = c.
1037. </J(C) = C".
00
HINT. In order to obtain the first formula consider the Cauchy integral
l
23ii
f (c-1C )ii C-z
dC
'
c
where (-C-)il
c-1 ill a function single-valued in the plane with the straight line
cut (0,1] and equal to lat oo, and 0 is the boundary of the doubly connected
domain: the disk !Cl < R (R > 1) with a cut along the segment [0,1]. The
second formula is obtained from the first by means of Sokhotskii's formulae.
1000. Evaluate the singular integral
1
_!_
n
J--JII( 1-t) t2+a dt
I+t t-x
<-1<x<1).
-1
1051. Find the integrals:
1
(1) J
0
I-t dt
log --t- t-z (ze[O,l]);
158 PROBLEMS ON' COMPLEX ANALYSIS
(2) f
0
1-t dt
log -t- t- (i:e (0, 1)).
HI"T Use the method recommended in problem 1049 11ud the limit
( _t
1-t
-)..t-1 t
lim-----
..t-.0 A
= log--
1-t
.
1
F(zt)=-
0
f di:- -
(y = rx+i{J, 0 ~a < 1),
-
' 23ti (T-t0)11 (i:-z)
c
where 0 is an arc connecting the points a and b, t0 a point of thit:1
arc and (- t 0 )Y is a single-valued branch in the plane with a cut
joining the points t0 and oo. If the point t0 coincides with one of
the ends of the arc 0, we shall consider that the cut extends along
the whole of the contour 0; if, however, the point t0 is internal, the
cut extends along the arc (t0 , b) of the contour 0.
Prove the following assertions:
(1) In the neighbourhood of the point a
(zeO),
(t E 0),
F1(z) = 2m
I J d-r
(-r-a)Y(-r-z) -
ell'"
2i sin y:ll' (z-a)-Y
c
and using Sokhotskii's formulae prove that F 1 (z) is an analytic function in
the neighbourhood of the point a.
(2) In the neighbourhood of the point b
e-11"'
F(z, b) = -- .,. . -(z-b)-Y+FQ(z) (zeO),
-ism 7'3t ~
INTEGRALS OF OAUOHY TYPE 159
_1_
2ni c
J
log-C- _dC __
C-l C-z
close to the pointR z = -R and z = R if 0 is the semicircle ICI = R
(R > 1), in the upper half-plane (commencing at the point R).
HINT. See problem 1043.
REMARK. For the behaviour of integrals of Cauchy type close to a singular
line see N. I. MusBKELIRHVILI, (1951), Singular Integral Equations (Bingidyar-
fll/l!/6 integral'nyye uratmemya) Chap. I, Gostekhizdat, or F. D. GAKHOV, (1958).
Boundary Problems (Krat'IJ1fY6 wdachi) Chap. I, Fizmatgiz.
dz= tl:i:-idy = dz ,
(see page 55)
IOMt. Using Green's formula prove the following relations:
(3) f(z) = ~
2m c
Jf(C)dC _
C-z
_!_
1'
JJ G
of() ~d7J
oC C-z
(C = ;+iri).
ff
G
f'udxdy = ! f fUdE,
C
in particular
JJ
G
f'g'dxdy = i
2C
J
fdg.
1= 2i J- fdf
c
is equal to the area 8 of the dome.in G'; if, however,f(z) conformally
maps the domain G onto the exterior of the contour O', then
l=-8.
INTEGRALS OF CAUCHY TYPE 161
IOOSt. Prove that if the function cf>(C) is defined and continuous
in the bounded closed region G, then the function
ff d~d17
IC-zal IC-z1I
IC - Z1~Z11 < R
where R is the radius of the circle with centre at the point (z1 +:i:a}/2, which
contains within it the closed region G. By a simple substitution we arrive
at the last integral in the form
ff ~d17
IC-al IC+al =
2 ff d~d17
IC-al lf+ai'
where a = lzz-Z1I.
2
~<R ~<R
l>O
From geometrical considerations it is clear that
ff IC-ald~d17IC+al + a<IC-aJ<Rt'l
ff ~'!L__ < IC-aJ<a
ICl<R IC-al IC+al
ff IC-ald~d17IC+al
1>0 E>O
Noting that in the first integral IC+al > a and in the second IC+al >IC-al,
and passing to polar coordinates (C-a = rel<fl), we obtain the required
estimate.
1059. Let G be the circle izl < R '< 1 and
e2'1J
c/>(C) = --
1 CC= ee18 , z = re1"').
log-
e
t For problems 1058-1062 see I. N. VEKUA, GsnBralisBd .Analytic Functions
(ObobshchennyyB anaJitioh68kiys funktaii), Fizma.tgiz, 1959. English translation
published by Pergamon Press (1962).
162 PROBLEMS ON COMPLEX ANALYSIS
F(z) = --;-
I ff ef>(C)d~d'YJ
C-z
G
f(z) = ~
2:n:i
f f(C)dC
C-z
_ _!_ff ef>(C)~d'f/
:n: C-z
(Pompeius's formula.).
C G
J J vLludxdy+D(u, v) = J v ~= ds
G C
(3) If u1 = u 2 on 0, then u 1 u 2 in G; =
(4) If OU1/on = a~/on on 0, then U1-U2 =constant in G.
164 :PROBLEMS ON COMPLEX ANALYSIS
1065. Prove that if u is harmonic in the circle izl <R and con-
tinuous in the closed circle lzl ~ R, then
an
u(O) = 2~ Ju(Re' )d0 = n~s lzlJJu(re ")rdrd<f>.
0
9
<R
1
JJe<C>
G
log IC 1 zl ~ d17,
J
c
a 1
P(C)Tnlog-IC-zl ldCI
>R
-
(I) 2n
l_
J2:r _ _l _ -!log 1! 1 , if lzl
log IC-zl dO- 1 (C = Re19);
o loglf, if lz! <R
l l /log l!I , if lz! > R,
<2 > 7&R 2 fJ
ICl<R
log IC-zl d~d1J = 1 1[ {lzl ) 2]
loglf+2 1- 1f ,
if izl <R;
(3) J a:
c
P(C) log IC 1 zj ldCI = J
c
P(C)d arg (C-z)
<4 >
Ja
-a
1 { </>, if Im z > 0 ,
a'YJ log IC-zl ag = -</>, if Im z < o,
INTEGRALS OF CAUCHY TYI'E 165
where <P (0 < <P < 3t) is the angle subtended by the segment
(-a, a) at the point z (a > 0).
Gt'een'B junction g(a:, y, ~. 11> of the domain G (abbreviated to g(z, CJ,
z = a:+iy, C= ~+i1J) is defined as a harmonic function of both pairs ofvaria-
bles a:, y and~. fl which is equal to zero on the boundary of the domain G.
and has a singularity at z = C, where
Green's function is symmetrical with respect to its arguments, that is, g(z,
g(C, z) (see, for example, (1, Chapter VI, I]).
C> =
1067. Formulate the Dirichlet problem for harmonic functions
which is equivalent to the finding of Green's function g(z, C).
1068. Let the function w = f(z, C) conformally map the simply
connected Jordan domain G onto the disk jwj < 1, so that f(C, C) = 0
(Ce G).
Prove the relations
g(z, C) = - log lf(z, C)j, (1)
f(z, C) = e-<+lh), (2)
lim
r-+0
f g(z, C) ou(z) ds
on
= o.
Yr
(3) If u(z) is harmonic in G and continuously differentiabfo
on 0, then
u(C) = - 1-
23t c
J u(z) og~, C) ds
cn
(.;: e G).
166 PROBLEMS ON OOMPL'IDX .ANALYSIS
C
on on = J(u~-go")da
f ( u~-gou)da an on l'r
(I) 2n
I J In
R"-r8
Jll-2Rrcos(8-4>)+r8dr=l;
0
= 2n
I f In
R1 -r8
[u(R,O)-u(R,Oo)] Jll-2Rrcos (0-cf>)+rl d(J;
0
9
I J
.-n 111-0,1 <er
iu(R, 8)-u(R, 00)1 R 2_ 2R
Jll-r8
r cos
(0-.1.) +-- dO
'I' r-
< e;
ct
(4) If 10-80 1 >ct and lcf>-0 0 1 < 9 , then
-
B1 -2Rroos (0-cf>)+rl > 4Rrsin2 : ;
then
M(Jll-r8)
iu(r,,P)-u(R,0 0)1 < e+ 23tA -
where
f
2
u(z) = u(O)+
n=l
r
co
2 (;r
co
where
! J
I
a.= !J0
u(R, 0) cosnOdO, bn =
0
u(R, 8) sin nOdO,
In
- - a.-ibn
On- R- -- _I_f
3tR- u (R ,ve
a) -'"'dav.
0
D(u) = JJ(u~+u~)dxdy =
~<R
3t ~ n(a:+~)
n-1
THE INTEGRAL FORMULAE OF POISSON AND SOHW.A.BZ 169
(the coefficients a,, and b,. are defined in the preceding problem).
It is possible that both sides of the equation are equal to infinity.
HINT. Pass to polar coordinates (r, ~) and prove that for r <R
00
00
f(z) = _.;.
ni
J u(C)dC -f(O).
C-z
ICl=R
HINT. Use the equality
C+z 2 1
C<C-zl = C-z --r
1085. Obtain formulae similar to Poisson's integral and Schwarz's
formula for the upper half-plane Im z > 0, that is, express the
harmonic function u(z) and the analytic function f(z) = u(z)+iv(z)
in terms of u(t) (-oo < t < oo).
HINT. Make use of the conformal mapping of the half-plane onto the
circle.
1086. Derive Schwarz's formula for the strip 0 < Im z < 1.
HINT. Use the conformal mapping of the strip onto the half-plane.
The harmonic measurs w(z, oi:, G) of the boundary arc oi: at the point z with
respect to the domain G is defined as a bounded function, harmonic in G, which
is equal to 1 at the interior points of the a.re oi: and to 0 at the interior points
of the remaining pa.rt of the boundary. The harmonic measure w(z, oi:, G)
is an invariant in conformal mappings.
TH'E INTEGRAL FORMULAE OF POISSON AND SCHWARZ 171
J 1
ru(z,0 1 82)= 2n
o,
l-r2
l-2rcos(O-<f>)+r2d0,
o,
and in particular, that c.o(O, 01 , 02 ) = (08 -01)/2n.
1088. Find the level lines of the function dc.o (z, 01 , 0) /d(J for a fixed
value of 6 (z = re' is a variable point).
HINT. Prove that
~: = 2~ I::=: I
where w' is the end of the chord from w which passes through z.
1089. Let us denote by w' the end of the chord from w which
passes through the point z. Let IX be the arc (0 1 , 61), and IX'(z)
the arc described by the point w', when w traverses the arc IX. Prove
that the length of the arc IX'(z) equals 2nc.o(z, Ov 02).
1090. Find the level lines of the harmonic measure c.o(z, 01 , 01 )
of the arc (01 , 02). Using this prove that the integral defining the
harmonic measure (see problem 1087), actually has the limiting
values 1 on (61 , 02 ) and 0 on the complement (interior points of the
arc are considered).
1091. For the half-plane Im z > 0 determine the harmonic measure
c.o(z, a, b) of the segment (a, b), of the ray ( -oo, b) and of the ray
(a, oo). What is the geometrical meaning of these harmonic measures 1
1092. Find the harmonic measures of the boundary rays of the
angular sector.
O<argz<y.
1093. For the semicircle lzl < R, Im z > 0 find the harmonic
measures of the diameter LI and of the semicircu]ar arc I', and also
the level lines of these harmonic measures.
1094. Find the harmonic measure of the bounding semicircular
arc I' of the domain lzl > R, Im z > 0.
1095. Find the harmonic measure of the boundary circular arc
I' of the domain lzl > R, 0 < arg z < 2n.
1096. Find the harmonic measure of the boundary circumferences
of the annulus r < lzl < R.
172 PROBLEMS ON COMPLEX ANALYSIS
-f~:da.
r,.
1098. Prove that for the complex Green function g + ik of the
domain G (g(z, C) is the Green function of the domain G, k(z, C)
its harmonic conjugate) the period along r.
equals 2mo,.(z), where
m,,(z) is the harmonic measure of I',. with respect to the domain G.
Prove that
2" w,.(C) = 1.
HINT. The function u(C) harmonic in G can be represented in the form
p,, = - f ()aw..
r
(z)
u z - 0- w s .
n
..3-
"
2
-I
tppy = 0 ( = 1, 2, ... , n)
2" p.,pC.Op ~0 1
... "
the case of equality occurring only when allthe " are equal.
llDrT. Apply the formula
D(w) = -
r
f OJ : da
to the harmonic function w(z) ... I" C,,OJy (1) (see problem 1084; the afgn is
..-1
ravened because n is now the inward normal).
1103. Prove that the quadratic form
is positive definite, that is, positive for all sets of values {c.}, ex-
cluding 0i = 0 11 = = o...1 = 0.
116'. Prove that the system of equations
11-I
2 p.,".A" = B"
.. -1
( = 1, 2, ... , n - 1)
174 PROBLEMS ON COMPLEX ANALYt!!IS
(the A,, a.re unknowns) has a. unique solution for any B,.. Using this,
prove that for any function u(z) harmonic in G, generally speaking
not single-valued, it is possible to select constants A 1 , A 2 , ... , A 11_ 1
in such a. way that the harmonic function
11-l
1
c;-
..n
J
an elt+z
h(t)-1-,- d t
e -z
(z = rellf>). (1)
0
The function h(), defined by means of a singular integral with the Hilbert
kernel - 1- cot !-t
2n 2
~
h(of>) = T[h(of>)] ~ !n
1
J... cf>-t
h(t) cot-
l
2 -dt = !n ~
(.__.
J + JIn) , (2)
0 0 ~
is said to be conjugate to the function h(cf>). The series
00
-i +}; n-1
00
0 0
1106. Let k(tP) be continuous and have a piecewise continuous
derivative k'(q,). Prove the following assertions:
(1) h(tP) = limf(re1"') = lim f 1 (re1"') (the definition of the
r-1 r-+1
functions f(z) and f 1 (z) are given at the beginning of the section).
HrNT. Show that
00
(4) h(tP) =}; (-bn cos n<P+an sin n<P) and the series con-
n-1
verges uniformly.
Bn
(5) f h(tP)dtP =
0
0.
In problems 1108-1116 find the function k(<f>) from the given func-
tion k(<f>).
1108. cos n<f>.
1109. sin n<f>.
1-acos q,
1110. 1- 2acos <P+ a1 (-l<a<l).
1111. log (l-2 a cos q, + a 1) (-1<a<1).
1112. cos <P (m is a natural number).
sin (n+i><P .
1113. 2 sin l<P (n is a natural number).
1114. '"'' (-:n: ~ "' ~ :n:).
1115. "' (-:n: ~ "' ~ :n:).
1116. "' (0 ~ "' ~ 2:n:).
1117. Prove the following assertions:
(1) If k(<f>) is an even (odd) function, then k(<f>) is an odd
(even) function.
(2) If k(<f>+ro) = k(<f>), then k(<f>+ ro) = k(<f>).
(3) If k(<f>+ro) = -k(<f>), then k(t/>+ro) = -k(<f>).
1118. Let F(t, </>), k(<f>) and kt(<f>) be periodic functions (F(t, t/>)
with respect to both variables) with period 2:n:, belonging to the
class Lip ct (0 <ct ~ 1).
Prove the following assertions:
llll llll I In
(1) JJ
0
d<f>
0
F(t, <f>) cot q, 2 t dt = JJ
0
dt
0
F(I, <P> cot t/> 2 t dtf>
00
2
= 2:n:8
-I
(A 8 -B11) ,
where
I I
A.= ~ JJFcosntsinn<f>dtd<f>,
0 0
B 11 = ~1 ff
0 0
F sin nt cos n<f>dt dq,;
SINGULAR INTEGRALS 177
2's In
J{2~ J
(2)
0 0
k(t) cot </> 2 t dl}k,.(</>)def>
~ ~ ~
= f { 2~ f
0 0
k1 (</>) cot </> 2 t d<f>}k(t)dl =n 2 (anb~1)_bna~1)),
n=l
where (an, bn) and (a~1>, b~1>) are the Fourier coefficients of the
functions k(</>) and k1 (</>);
(3) 21n I 1
"{ 1
2n
fl>I k(t)cot-
T-t </>-T
-dl} cot-
2 -dT 2
0 0
111:
= -k(</>)+ 2~ f
0
k(tf>) def>
or, if k= T(k),
In
(5)
where
then
- I[
hk = -- 10 (h11:+i-hk 1)
n
cot 20 +(krc+s-hk-a) cot 00
3n
5n 7n
+(hrc+6 -k1 _ 6) cot 00 +(h1:+7-h1:_7 ) cot 00
1 t
1120. Let (l>(t) = - 231: [h(cf>+t)-h(cf>-t)]cot 2
Assuming that h(</>) can be continuously differentiated three times,
prove that
where
1}Irc = max jh,<l:>(cf>)I (k = 1, 3).
o.;;q,.;;;n
In particular, if n = 5
IR!< l 65X l0- 2{M1 +7 6Uf3 ). (8)
1-a: cot a:
a: cot a: .:;;; I, . ~ 1
sm1 a:
and
a:Bcosa: ~ 1 fi
sins a: """ or
SINGULAR INTEGRALS 179
1121. Prove that if k(<f>+n) = k(<f>), then formula. (6) (see problem
1119) acquires the form
f
00
__!_
ni
f00
k(t)dt
t-z
(12)
-oo
180 PROBLEMS ON COMPLEX ANAL"!SlS
If the integral in the usual sense diverges, its principal value must
be considered. The function h = T(k) is said to be conjugate to
the function k.
1129. Prove the following assertions:
(1) The principal value of the Schwarz integral (12) at the
point x, if it exists, is equal to ih(x).
(2) If k(x) = const, then h(x) =
0, if the principal value
of the integral (11) is considered not only at the point x, but also
at oo;
oo .x-e R
h(x) = _ _!_
n
J00
k(t)-k(x) dt =
t-x
_ _!_J
n
00
(4) Prove that with the conditions of section (3) h(x) is con-
tinuous and in addition, if IX< 1, h(x) e LiplX.
(5) If k(-x) = k(x), then h(-x) = =Fk(x).
(6) If k(x+ro) = k(x), then h(x+ro) = =F k(x).
1130. Let /(z) and fi(z) be analytic functions defined by the
Schwarz integral (12) in the half-planes Im z > 0 and Im z < 0
respectively. Prove that if k (x)e LiplX (0 <IX ~ 1) uniformly
co
and divide the integral J= Re/(:i:+iy)-h(:i:) into the integrals
-co
x-1 x+I co co
f +f+f. To estimate the integral J= Im/(:i:+iy)-h(:i:)
-co x-1 x+ I 0
I co
(see (13)) subdivide it into the integrals J+ J and in the first of these
0 I
make the substitution e = wy.
Rmu.Rx:. The assertion of the problem oa.n be significantly strengthened.
1131. (1) Let <f>(z) and 'J'(Z) be analytic functions in the half-planes
Im z > 0 and Im z < 0 respectively, bounded at oo and continuous
in the closed half-planes. Prove that if <f>(x) and 'l'(x) e Lip ix (0 < oi~l)
and there exist transformations T[</>(x)], T['J'(z)] (T(ki, +ik8)
= T(k1 )+iT("9)) and the limits
n n
lim
R-+00
_!_
no
f </>(Re11)d0 = </>"', lim
R-+00
_!_
no
f 'J'(Re- 11)d0 = "'"',
then
T[<f>(x)] = -i[<f>(x)-</>"'], T['J'(x)] = i['J'(Z)-'J'"'].
HINT. Let </l(z) = u(z)+iv(z) and "'"' = "'"'+"'"'' Applying the results
of problems 1085 and 1129, we have (z = :i:+iy)
.
v(:i:) = lim[~(s)-u(:i:)] =Jim
. {
--
J u(e)-u(:i:)
I
00
}
de +""''
y-+11 ,....0 :Tr -oo -
!l'[u(:i:)] = _ _!_
:Tr
J u(t)-u(:i:)
00
t-:i:
de.
-oo
Estimating the difference of integrals it is possible to obtain
e-:i: = v(:i:)-v"''
-co
which reduces to the first proved relation. The remaining one is proved similarly.
182 PROBLEMS ON OOMPLEX .ANALYSIS
(2) In the conditions of part (1) let efJ(z) and 1jl(z) have
respectively the isolated singularities a (Im a > 0) and b (Im b < 0).
Find T[efJ(x)] and T[1j1(x)].
In problems 1132-1137 find h(x), /(z) and / 1 (z) for the following
functions k(x) (the parameters A., a, b are real numbers, A. > 0,
b > 0, n is a natural number):
1132. COB Ax. 1
1133. sin A.a:. 1135. l+x'
x 1136. Re l/[x-(a+ib)].
1134. xl+bl . 1137. Re 1/[x-(a-ib)].
-!J 0
00
dt
tA(t-x) =
1-
_
cotA.n
---;r-'
1
l:tl"sinm'
HINT, Consider the integral of l/[C"(C-111)) along the boundary of the
annulus r < !Cl < R (deleting the circle IC-a:I < e. If a:> O) with a. cut
along the re.dial segment r <Re C= t < R, Im C= 0 (or the remaining
pa.rt of it if a: > 0) and pass to the limits e.s r -+ 0, R -+ oo (and e -+ O, if
a: > 0). On the out CA = tA above and C" = e.\e8'1U below.
1139. Prove that
(:t = a, a > 0, n is a natural number). For !xi < a the integral can
be represented in the form -
'j(,Q,
i
~ k-o ___..!:_k
n+ a
(~)k. Both expansions
are also valid for x = -a.
HINT, Use the expansion
I 1 ( 1 :r a:n-1 ) 1
"(I-a:) =- a:" ,-+,.+ ... +-en- + :i:"{e-:r.)
SINGULAR INTEGRALS 183
1140. Prove that
co
_ cotn _ _
1_ ~-1-(~)k if x>a,
:if- 'j(,Q,fl. .L.i k-a. x
k-l
if x< -a,
~-1- ~),
co ( k
1-
-- if lxl<a
na .L.i k+cc a
k-0
(a> 0, cc> 0 are not integers). Both the latter expansions a.re
valid also for x = -a, where the function k(x) defined by the integral
standing on the left remains continuous.
HINT. For la:!> a consider the integral of l/[Co'(C-111)) along the bound
ary of the ring a < IC! < R with a out along the re.dial segment a < Re C
= t < R, Im C= 0 (and deleting the circle IC-a:! < (/, if a: > a), and pass
to the limit as R-+ oo (and (!-+ 0, if a:> a). For la:I <a expand the
integrand in powers of a:/t. The latter assertion is established by means of
Abel's theorem on power series (for the latter expansion) and by means of
the expansion of I/sin or:n in partial fractions (see problem 998).
RllMABK, If 0 < or: < I, then for la:I > a it is possible to use the integral
of problem 1111, from which it is necessary to subtract the integral _..!._
"o
J
evaluated by expanding the integrand into a series of powers of t/a:. The
resulting expression retains its meaning (analyticity) also for or: ;;;i. I, and
hence by analytic continuation, it is valid also for all or; > O.
1141. Prove that if k(x) = 0 outside the interval (-a, a), then
for !xi> a
a n
h(x) = _..!_
n
j' k(t)dt
t-x
= ~ Ak +Rn(x),
.L.i x
-a k-l
where
a
Ai=
-a
"
! f et- 1 k(t)dt, IR
n
I< 1
lxl~(lxl-a) :re
1 f
-a
ltlnlk(x)I dt.
For an approximate ca.lcula.tion of h(a:) for sma.11 va.lues of a:, la:I < a, the
integrals a.long the infinite intervals (- oo, -b), (b, oo) and a.long the interva.1
(- b, b), where b >a a.re considered separately. For the interva.1 (-b, b)
the formula
_ _!_
:ri
J e-a:lk = _..:_ .L..J'1
b
h(e)dt
:ri
n-1
h 81+1
2i-21c+I
+Rak
-b i=-n
where
!J
00
ok = - e-k-1 h(e)dt.
b
The integral a.long the infinite interva.1 (- oo, -b) is treated simila.rly.
1142. Evaluate h(x) for k(x) = e--' at the points x = l, 2, 3, 4, 5.
Take a = 5, b = 6, n = 15. The calculation is to be taken to five
significant figures. Show that the integrals along the infinite inter-
vals can be neglected. For comparison evaluate h(x) at the points
x = 2, 3, 4, 5, by the asymptotic expansion given in problem 946
(the terms of the series are to be used as long as they decrease).
1143. Evaluate h(x) for k(x) = l/(l+x2) at the points x = 1,
x = 5, taking b =IO, n = 25. Compare the results with the exact
values.
connected with Schwarz's integral for the strip 0 <Im z < n (see
problem 1086), and
1
2ni
J t-z
h(t) coth ~ dt, (15)
(~denotes
-
....
the boundary of the strip).
The function h = T(h) is said to be conjugate to the function k.
1144. Prove the following assertions:
(1) The principal value of the integral (15) at the boundary
point T of the strip 0 <Im z < n, if it exists, equals ih(-r).
(2) If h = const, then h 0. =
(3) If h(t) e Lip a (0 < a ::=:;; 1) on any finite interval and
the integral (9) converges to oo (for example, if k(t) is bounded
00
we have
f(t) = limf(z) = h.(t)+ih(t),
(Im t = 0, Im t = n) .
then
f ~ cf
c
11-7:
F(l;, 11) dl;
l;-a
= -nl F(T T)
' +
f f
c
dl;
c
F(l;, 11) d11
(a-T) (~-a)'
SINGULAR INTEGRALS 187
where 0 is a smooth contour, closed or open (in the J.a.tter case a point Te 0
differs from an end point), the function F(C, a) e Lip et with respect to C and
a and if 0 goes off to infinity, decreases sufficiently rapidly at infinity (see the
book: N. I. MusKHELISBVILI, Singular Integral Equations (Singulyamwe
integral'nwe uravneniya), 23, Gostekhizdat, 1946 or I<'. D. GAKBOV, Boundary
Probkma (Kraevyye zadachi) Chapter I, 7, Fizmatgiz, 1958).
In problems 1148-1164 verify the Poinca.rlS-Bertra.nd formula.
for the given functions F(l;, a) (l; = x+iy) by the actual evaluation
of both sides of it. In these problems: 0 1 is the closed contour
enclosing the origin of coordinates, 0 2 is the horizontal a.xis, 0 3
is the segment [O, 1] and 0 4 is the semicircle ICI = R, Im l; ~ 0,
n = 0, 1, 2, ...
1148. t;n, the contour 0 1
1149. 1/t;n, the contour 0 1
1150. an, the contour f!11.
1151. 1/an, the contour 01"
1152. l;"a", the contour 0 1 (p, q = 0, 1, 2, ... ).
1153. 1/(x+bi), the contour 0 2
1154. cos A.x cos a (A. > 0, ~ 0), the contour 0 2
cos A.x
1155. --b-. (A. > 0), the contour 0 2
a+ i
1156. x, the contour 0 3
llmT. Make use of the formula
1 I
J
0
1-z dz
log---
x x--t
1 (
- - =-
2
t -) - -
log-
1-t
x1
2
.
f ~=logR+r
O'-T R-T
=logjR+rj+i~,
R-T 2
C:,
fc,
R+a
log R-a da = 0, f
c,
1 R+a
-log--da= - - ,
a R-a 2
:i21
j lo g
R+a
- -da
- -_- -
R-a a-T
1 [log
2
2 R+T
---~
R-T
2]
c,
(for the last equality see the hint to problem 1051).
188 PBOBLEMS ON COMPLEX ANALYSIS
w=-~ff'm~~
:ri (C-z) <~
E
are considered below, where C= e+lf'/, extending to the entire C-plane
and defined as the principal value
Sw(z) = lim -
e.....O
~
:ri 1~
f w(C) d;df'/
<C-z)
(17)
-Z >I!
The transformation SOJ is closely connected with the integral transformation
S
eC.O
( ) = _ _!__
z :ii
JI c.o(C)d~d.,,
(C-z)2 '
IC-zi>e
190 PROBLEMS ON COMPLEX ANALYSIS
g(l) ff 4(;Md11
((;1-1)2 +g(-1)
ff 4(;dEd11
((;-1)1
ReC>O ReC>O
I T 0 g(z) = _ _!_ff
n
(-C-z
1- _ _!.)g(C)dEd11
C
B
(in comparison with Tg the passage to T 0 g ensures integrability at
infinity). Prove that
_..!...ff
n
(-c-1
1--_!.)dEd11 =
C
l (20)
B
and using this fact, that
ITog(l)-g(O)-Sg(O)I ~ Bcx!Y!cx,
where
B" =-;;
1 ff IC-II
1ci11-1
dEd'fJ.
B
HINT. Multiply g(O) by the left; side of (20).
1178. Let w(C) e H 11 Putting
ro(C) = w[u+C(u-v)],
prove that
IToro(u)-Toro(v)-w(v)(ii-ii)-Sw(v)(u-v)I ~ B 11 \wl 11 \u-1J[H 11,
and deduce from this the relation:
(T0 co(z)k= w(z), (T0 w(z))s = Sw(z).
1174. Show that the search for the quasi-conformal mapping
w(z) of the whole w-plane onto itself, which satisfies the equation
ANALYTIC CONTINUATION.
SINGULARITIES OF MANY-VALUED
CHARACTER.
RIEMANN SURFACES
I. Analytic continuation
00
l .2
00 00
l zn
fi(z) = - and / 2 (z) = i:n:+ (-l)R (z~ 2 )n
n
n=l n-1
have no common domain of convergence. Prove that the functions
/ 1 (z) and f 2(z) are nevertheless analytic continuations of one another.
1179. Prove that the functions defined by the series
1 (1-a)z (1-a)2z2
1+az+a1z2+ ... and - - -
1-z (l-z)2 (1-z) 3
+... '
are analytic continuations of one another.
192
ANA.LYTIO OONTINUATION i93
1180. Let the power series
f(z) = a0 +a1z+ ... +anzn+ ...
have the radius of convergence R = 1. By carrying out the change
of variable z = Z/(i+Z) let us transform it into the form
f(z) =J[Z/(i+zn = F(Z) = ea+eiz+ ... +c.Z"+ ...
Denoting bye. the radius of convergence of the series obtained
prove the following assertions:
(i) e~ !' and if the point z = - i is a singularity of the
lonius lz zii= e
(3) If e = i, the equality given in part 2 analytically
continues the function /(z) into the half-plane Re z < i/2.
(4) If e > i, then the function f(z) is analytically conti-
nuable into a domain exterior to the circle of Apollonius z z i = I I e
00
prove that for any point of the form C= vt1 (le is a natural number)
J(tC> -+ oo as t -+ 1 co
< t < I).
In problems 1182-1184 prove that the functions represented
by the given power series are not continuable beyond the unit circle.
194 PROBLEMS ON COM"PLEX .ANALYSIS
00
~-, ~2n
1182. f(z) =- L.J --~,,-
n=O
HINT. Use Pringsheim's theorem (see problem 642).
00
S(z)
oo (---------
=,}; 1-zn l-zn-1) . Prove that the expression
n=l I+zn I-j-zn-1
1 1
1j!(z) = 2 [f(z)+</>(z)]+2S(z)[f(z)-<f>(z)]
represents in the domain JzJ < 1 the function /(z) and in the domain
[z[ > 1 the function <f>(z).
1187. (1) Prove that if oi is a real irrational number the series
represents in the domains JzJ < 1 and Jzj > 1 analytic functions
for each of which the circle [z! = 1 is a natural boundary.
ANALYTIC CONTINUATION 195
HINT. Prove that the sum of the series increasel'I without limit as z __,. eimtzn
along a. radill8 vector.
REMARK. This problem is a. parlicula.r case of the following general
theorem:
Let L be a curve, closed or open, which has at every point a definite radius
00
.F(z) = 2 ---"-
n-1
00
an-z
c
1188. Prove that the series J;-,,-1- (11" =- e" 101 ") converges for
n=l n +n
Re z > 1 and its sum has the straight line Re z = 1 as its natural
boundary.
00
f(z) = L ane-Anz
n=l
where an= (-l)n+i, A.21c_ 1 = 2k, A.21c = 2lc+e- 2" (le= 1, 2, ... ), can
be continued analytically into the half-plane Re z > - l.
HINT. Write /(z) in the form
00
/(z) = J;(1-e-ze- 2")e--llcz
k-1
and prove that in any finite domain belonging to the half-plane Re z . --1,
11-e-ze-llcl <Me-lie,
where M is a constant for the domain considered.
196 PROBLEMS ON COMPLEX .ANALYSIS
REMA.BE. The given problem shows that on the straight line bounding
the half-plane of convergence the sum of a Dirichlet series may not have
singular points.
*.
1191 The function /(a) defined in the half-plane Re a > 0 by
means of the Laplace integral
00
I'(z) = ~ (-l)n
.L.J nl(z+n)
+f e-'F-ldt,
n-o 1
1
HnrT. In the integral Je-tt-ldt replace the function e-r by its expansion
0
into a power series.
1194. In problem 420 it was proved that for 0 <x< 1
00
f
0
F-1 sin t dt = I'(x) sin ~
I'(z) = -.-i-J
2smnz
c
e-w(-w)-ldw ((-w)-1 = e<-..101(-w>)'
ANA.LYTIO OONTINUATION 197
where the contour 0 consists of a cut a.long the positive part of the
real axis, the detour round the coordinate origin being traversed
anticlockwise.
1196. Let C(z) be Riemann's Zeta function (see problem 695)
co
C(z) = ~ _!_
L.J n:r:
(Rez >I).
n=l
Prove that for Re 21 >I
C(z) = I'(z)
I J w-
co
1 dw
(ew-1)-
o
and obtain from this the analytic continuation of the function
C(z) to the whole plane, excluding the point 21 = I; explain the
character of the singularity of the function C(z) at the point z = I.
HINT. For the analytic continuation consider the integral J(-w):r:~i dw
C eW-
Where 0 is the Contour of problem 1195.
1197*. Let the function /(z) be expanded into the power series
co
/(z) = 2 anzn which has the radius of convergence R = I. Let us
n=O
co n
denote by efJ(z) the sum of the series <fo(z) = J; a~~ (the function
n=O
t/J(z) is the Borel transform of the function /(z) - it is an integral
function, see problem 535).
Prove that for izl <I the equality
co
Je~t q, (zt) dt = J(z)
0
co
is satisfied. Prove also, that the functionf e-'q,(zt)dt effects the ana-
o
lytic continuation of the function /(z) into the domain G defined
as follows: A straight line is drawn through every singularity of
the function /(z), perpendicular to the segment connecting this
singularity to the coordinate origin: G is the convex domain con-
taining the circle izl <I, the boundary of which consists of points
of these straight lines; if the number of these straight lines is finite,
then G is a polygon (Borel's method of continuation).
198 PBOBLEMS ON OOMPLEX ANALYSIS
(3) 2z'".
11-0
F(z) = ~f ,P(C)dC
2ni C-z c
let 0 be a. simple closed contour and ,P(C) a function oontinuou
a.long 0. Prove that in order that one of the functions F+(z) and
F-(z) (see page 153) should be the analytic continuation of the
other through an arc ye 0, it is necessary and sufficient thatf(z) = 0
on the arc y.
1200. Prove that if the function ,P(C) is not analytiot a.t even
one point of the simple open a.re 0, then all the points of the arc 0
are singularities of the integra.l of Cauchy type
_I_f ,P(C)dC
2:n:i C-z .
c
HINT, Commence with Sokhotskii'a formulae for the limiting valuea
of a Cauohy type integral.
1201. Let y be a simple closed contour tra.versed in the positin
direction, consisting of arcs y 1 , y 8 with common end points Zi and z1
(Fig. 31), let G+ be the domain within y, G- the domain outside y.
rz
z, Zt
tr
FIG, 31
t That is, there does not exist any analytic function identical with (C)
on any arc belonging to O.
SINGULABITIBS OF MANYVALUED ClHABA.CTEB 199
t For tbis section see [l, Chapter VUI]; V. V. GoLUBEV Lecffwea on eke
Analytic Theory of DiffereMal Eqwtiona, (Lektrii po analiUchealcoJ teorU rllffe-
renttal'nylch. twa11716nU). Gostekhizdat, 1951; R. NEvANLINNA, Umformien.mg,
Berlin, 19153.
200 l'BOBLEMS ON OO:Ml'LEX ANALYSIS
w = 2 0ntn,
n=-N
where
(a of: oo),
(a= oo).
A point z = a or z = oo is said to be a logarithmic branch point if it is such
that any branch w(z) admits of unbounded analytic continuation in the domain
0 < lz-al < r (or R < lzl < oo) and is infinitely many valued there. Such
a branch w(z) in the neighbourhood of a logarithmic branch point becomes
a single-valued analytic function on passing to the parameter t = Log (z - a),
Re t < e (or t = Log z, Re t >(!)It must be remembered that on the Riemann
surface above one and the same point z, in addition to various logarithmic
branch points there may also be found other points of single-valued and
many-valued character.
1203. Explain for which values of z the values of w(z) a.re identical
on all the sheets of its Riemann surface above the z-pla.ne, if:
(1) w= (z2-9)y(z);
(2) w =sin z + (z2 + 4) Log z;
(3) w = sin z + (z + 4) Log z.
Are the values of w' (z) identical at the same points 1
_.1204. For ea.oh of the functions:
(1) w = zyz.
(2) w = z1 Log z, w(O) = 0,
verify that at the point z = 0 there exists a first derivative which
is the same for all the branches and that a finite second derivative
does not exist.
In problems 1205-1212 expand ea.oh of the given functions w(z)
into a series of powers of the local para.meter t in the neighbour-
hood of all the points of its Riemann surface situated above the
given z points; indicate the domains of convergence of the resulting
series.
1
1205. w= l+y( 2 -z), z = 1, z = 2.
If the function w = /(z) is single valued and the function z(w) inverse
oo it is many valued, for the determination of the algebraic branch points
of the function z(w) it is neceBBary to find the zeros of /'(z), the multiple poles
of /(z) and investigate the behaviour of /(z) at infinity. In this, a point z0 #: oo
corresponds oo an algebraic branch point of order k-1 of the function z(w),
if in the neighbourhood of Zo the Laurent expansion of the function /(z) has
the form
00
(ore #: 0)
or
-00
r
r
1220. w = (I + : Consider the limiting case n -+ oo.
1221. W= ( : :
(3) w = VCft
even and odd n).
(z-a")) (consider separately the ca~s of
(3) w =
n>3.
j/ [(z-a)(z-b)(z-c)]; 4) w = VLU. (z-a,,,)}
l+z
1253. w8 = z2 - -
1-z
HINT, Put
zn
1254. uJ' = (1 +zn)I
HINT, Consider separately the cases of even and odd n.
1279. w=Log(z- !)
1280. w =Log (e"-1). Construct the Riemann surface.
1281. w = Log sin z. Construct the Riemann surface.
1282. w = Log tan z. Construct the Riemann surface.
1283. w = cos-1 (cosz) (compare with cos(cos-lz)).
1284. w = tan-1 (tan z) (compare with tan (tan-1 z)).
1285. (1) w = (z')' (r1 , r 2 are rational numbers). Compare with
(z')'. Consider, in particular, r 1 = 2/3, r 2 = 3/2;
(2) w = z'z'; (3) w = z'+z'.
1286. w = y z j/ (1-z).
1287. w = y[j/(z)-1].
1288, W=
yz ,
v(z)+l
1289. w = y
(Log z). Construct the Riemann surface.
1290. w = Log Log z. Construct the Riemann surface.
1291. w = [Log (z - 1)]1
206 PBOBLEMS ON OOMPLEX .ANALYSIS
y
1292. w = (sin-1 z). Construct the Riemann surface.
1293. w = sin-1 Log z. Construct the Riemann surface.
1294. w =Log (j/(z)-1). Construct the Riemann surface.
1295. w =Log [(1-yz)/(l+Jlz).
1296. w = sin-1 [yz/(l+z)].
1297. w = Log z (o: is a real number).
1298. w = y (z)+Log z. Construct the Riemann surface.
+
1299. w = Logz Logz.
1300. w = sin-lz + cos-1 z .
1301. w = ta.n-lz + cot-lz.
1302. w = ta.nh-1z - coth-lz.
1303. Construct the Riemann surface of the function w = (Log z) 1
and investigate the set of limiting values of w for one logarithmic
branch point above the point z = 0, obtained for: (1) r-+ 0,
</> = const; (2) r-+ 0, o: < q, < {J; (3) r = const, q,-+ oo;
(4) r-+ 0, q, -+ oo.
1804. Let x(z) be a single-valued analytic function in the circle
lzl < 1, which cannot be continued beyond the circle lzl = 1.
Explain for what values of a the given fV.n,ctions resolve into
distinct analytic functions, and for what values they do not.
(1) w = x(z)+v (z-a); (2) w = x(z) r.og (z-a):
(3) w = x(a+z"); (4) w = x(a+e11 ).
1305. Explain the problem of the factorisation of the functions:
c
(1) w = V(C-a) (2) w = Log(C-a), where = x-1(z) and
xis the function of problem 1804.
1306. Investigate the behaviour of the separate analytic functions
determined by the equations:
(1) w = x(z)(Log z)1; (2) w = x(z)[Log (z-1)]1, where x(z)
is the function of problem 1304.
Find, in particular, the set of indeterminateness in the neighbour-
hood of the logarithmic branch point.
HINT. Use the solution of problem 1808.
1307. Let f(z) be an integral function.
Construct the Riemann surfaces of the functions:
(1) w= y(f(z)); (2)w=Logf(z); (3)w=[/(z)J1 (0: is an
irrational number).
1308. Construct the Riemann surface of the function
SINGULARITIES OF MANY-VALUED CHABAOTEB 207
1309. Let
HINT. For /e.AD consider a.Bf?-+0 the Dirichlet integral fJ l/'l8 rdrd
C1<fzi<C11
(a singularity at zero), starting from the Laurent expansion
.f(z)
"" Cnzn
= .J:
-00
/(z) = { a+~.
_z-_a
if lz-al > (!,
a+z-a, if lz-al< fl
1316. Prove that if E e NF, then the plane measure mE = 0
(a stronger form of the result of problem 1313).
HINT. On the assumption that E is bounded (this is not essential) and
mE > 0, consider the function (see problem 1058)
/(z) = ff d~d1/
(;-z
CC = ~+i11).
E
where Z(y1c) is the length of Yt and inf is applied to all possible systems {Y1c},
is lmown as the length Z(E) of the set E.
1318. Prove the following assertions:
(I) If Eis a perfect everywhere discontinuous set of finite
length Z(E) < oo, then E e N A.c.
HINT. For the domain G=>E and e> 0 there exists,d(e)> 0 such that
for z', z" E G, lz' -z"I < d we have lf(z')-/(z")I < e. For the system of
contours { Yt} with diameters less than d, containing E and having the sum
of lengths l:Z(y1c) < l(E) +I, the expansion
k
of problem 1312 is constructed and by means of the relations (Ctc is inside ;'le
z is in G outside {;'le}),
J(z)- J~.
t-z
E
1320. Prove that if Ee N .AD then every function f(z), which is
schlicht and conformally maps the domain D which is the complement
of E, is linear.
HINT. Use the AD-removableness of E. To prove that f(z) is schlioht
on E use the principle of corresponding boundaries, after enclosing E by some
contour.
CHAPTER X
CONFORMAL MAPPINGS
(CONTINUATION)
(1)
where - oo < IJi < a 1 < .. . < an < oo a.re points on the m-e.xis which cor-
respond to the vertices Air A 1 , ... , An of the polygon P; 0 and 0 1 a.re complex
constants.
The mapping of the upper half-plane onto the exterior of the same polygon
ia effected by the function
f n(z-ai/k-
z n
1
F(z) = 0 dz _ +Ou (2)
o k-1 (z-b)(z-b)
t For thia section see [I, Chapter VIII, 7], [3, Chapter II, 3]; B. A. FucHs
and B. V. SHA.BAT, l!'imotWna of a Oomple:a Variable, (l!'unlatsii laomplelamogo
peremennogo), hd ed., Chap. VIII, Fizmatgiz, 1959. English translation
published by Pergamon Press (1962).
211
212 PROBLEMS ON COMPLEX ANALYSIS
stants 0 and 0 1 (in all n+I real parameters) are determined from
the relations
0 1+1
J lf'(x)ldx =
~
101
1+1
J
~
nn
t-1
lx-atjt-ldx = l1
a1 oo a1
where the l 1 are the lengths of the sides A 1A 1+1 (An+i = A1), 0 1 = f(a 1)
= w 0 (w0 is the affix of the vertex A 1) and arg 0 = () (() is the
angle of inclination of the side AnA1 of the polygon to the real axis),
The given relations exceed the unknown parameters in number,
however: they determine these parameters uniquely. Explain this
and show which of the relations can be discarded.
1322. Prove that if one of the vertices of the polygon is the image
of the point at infinity, for example an= oo, then the mapping
function is of the form
z n-1
f(z) = 0 Jn (z-at) t-ldz+01. 11
0 t=I
A.y
2)
FIG. 32
CONFORMAL MAPPINGS (CONTINUATION) 213
HINT. Let At= oo. If OCA: < 1, then consider the polygon P', cut off from
P by the segment Ai,At' where Ai, and At' lie sufficiently far from the sides
At-iAk and AtHAt (Fig. 32 (1)) and in formula (1) for P' pass to the limit as
Al: -+ oo, A/: -+ oo. If, however, oc1c ;;;i:: 1, then join A_tA/i' in P by a polygonal
l.irie (see Fig. 32(2)) and by enlarging it similarly extend it to infinity.
1324. Determine the quantities '1:, which enter into formula
(I) for the infinitely distant vertex of the "polygon" formed from
parallel rays represented in Fig. 33.
- A
A,=======
A,~~Az
A 1 ~Aa
1) 2)
A,-A3
A AJ
z 3}
Az
A,=====~
A*
A,=======;:
-5)
A+
Az
6)
--- ~ ,_A
A,
A,c::=======!
A1
A,
--
As======;;
A
'\ tZ/(
-- A
-
11.
A3=======?Az
7)
Aa A~'
8)
FIG. 33
f n(z-ak)
z n 1:-l
/(z) = 0 dz+ 01 '
0 k=l
214 PBOBLEMS 0111' COMPLEX .ANALYSIS
"'t (
where arc == e1 </>1 < </> 1 < ... < <f>n) are the points on the circle
lzl == 1, corresponding to the vertices At, traversed in the positive
direction and the IXtn are the internal angles of the polygon P.
(2) Prove that the function which maps the unit disk izl < 1
onto the exterior of the same polygon P, subject to the condition
that the point z == 0 passes into the point w == oo, is of the form
/(z) == 0 j fl (z-<t,/k-i:
1 k-1
where ai == e11fii: (</>~ > </>~ > ... > </>i), and the f3tn are the external
angles of P.
1326. Find all the cases of single-valued transformation of the
Schwarz-Christoffel formula. (1), that is, explain for which polygons P
the inverse function z == z(w) is defined and single valued in the
whole w-plane.
HINT. Polygons obtained from P by any number of reflections in the
sides, must without omissions or overlapping cover the whole 1.0-plane.
In problems 132'7-1332 map the upper half-plane Im z > 0 onto
the given domain P, located in the w-plane with the given corres-
pondence of the vertices of P and the points of the real axis. Deter-
mine also the period or periods of the inverse function z(w), the
group G of its invariant linear transformations of the w-plane and
the fundamental domain B of this group (see page 30).
132'7. (1) P is the strip 0 <" < k; w(- oo, oo) -+Z (0, oo)t.
(2) P is the strip 0 <" < k; w(- oo, oo)-+ z(-1, 1).
1328. P is the half-strip 0 < u < n; v < O; w(O, n, - i oo) -+Z
(1, -1, oo).
1329. (1) P is the right-angled triangle with acute angles n/3,
n/6; w(o, co, co+ ~~ )-+ z(O, 1, oo),
(2) P is the right-angled isosceles triangle;
w(O, co, co + ico) -+ z(O, 1, oo).
(3) P is the equilateral triangle;
1) 2) 3)
FIG. 34
1331. (1) Find the domains of the to-plane onto which the function
z
w(z) = f
0
1-).
y[l(l-1)] de
= y[z(z-1)]+(1-2).)(log [y(z)-y(z-1))-tin)
maps the upper half-plane. Consider the oases: ). < 0, 0 <). < !.
). = t.
i < ). < 1 and ). > 1.
216 l'ROBLEMS ON' COl\.PLEX ANALYSIS
1
w(z) = 2ni
Jz -i/[t(t-1)]
1-A.
tdt = ni1 [ log[y(z)+y(z-l)]-A.JI -z-
, /(z-1)]
1
maps the upper half-plane. Consider the oases: A. < 0, 0 <A. < 1
and A.> 1.
1332. Map the upper half-plane Im z > 0 onto the domains of
the w-plane indicated in Fig. 34 with the given correspondence
of points.
(1) w (A= 0, B = 1, 0 = oo)-+ z (0, 1, oo);
(2) w (A = 0, B = 1, 0 = oo) -+ z (0, 1, oo);
(3) w (A= 0, B = oo, 0 = oo)-+ z (0, 1, oo);
(4) w (A = 0, B = oo, 0 = oo) -+ z (0, 1, oo);
(5) w (A = 0, B = ia, 0 = oo) -+ z (0, 1, oo).
1333. Map the upper half-plane Im z > 0 onto the domains in
the w-plane indicated in Fig. 35 (0 < 0 < 1).
(1) w (A, B = 0, 0 = oo)-+ z (0, 1, oo).
(2) w (A, B = 0, 0 = oo) -+ z (0, 1, oo).
C B
~
C
ti
A
Fm. 35
FIG. 36
FIG. 37 Fm. 38
FIG. 39
FIG. 40
f(z) = ~ n
k=l
n
(z-a1;)rx,.'
220 PROBLEMS ON OOMPLEX ANALYSIS
where 0 is a. complex constant and tit = e14'1c a.re the points on the
circle lzl =I which correspond to the A1c. The points b1c = e1"1c which
correspond to the vertices B1c a.re the roots of the equation
n
~~-_!__=0. (3)
k=I z-a" z
How a.re the para.meters 0, a", b1c determined 1
HINT. In continuation by the symmetry principle the function f(s) is
multiplied by a constant factor, hence the function
4>(z) = f'(z)
f(z)
is single valued in the z-ple.ne.
REMARK. This formula immediately gives the solution of a number of
problems of Chapter II, for example, problems 281, (2), 282(1), 284, 290.
It is recommended that these problems should be solved again e.nd the
constants entering into the general formula determined.
1343. Map the exterior of the double angle (Fig. 41) onto the
exterior of the circle lzl >I subject to the conditions: w(oo) = oo,
w'(oo) > 0.
1344. Prove that the function w = f(z), which maps the upper
half-plane Im z > 0 onto the exterior of the star of problem 1342
is of the form
n
f(z) = 0
n
k-1
(z-aa:)11"
- '
(z-z0 ) (z-z0 )
where z0 is the point of the upper half-plane which is transformed
into oo.
FIG. 41
FIG. 42
f (z) = 0
n n
k=l
(z-ak)a,
(4)
n m
j=l
(z-e1)'JIJ
where ak, e1 are the points on the x-axis which correspond to the
vertices At, 01. The points b,, a. on the x-axis which correspond
to the vertices B,, D. are the roots of the equation
2, 2
n m
-rtk- - - 'JIJ
-=0 (5)
Z-!Zt Z-C
k=l j=l J
a.
How are the parameters 0, ak' bi. Cj, determined 1 What is changed
if one of the parameters ak, b,, Cj, a8 is equal to 001
HINT. Prove that
/'(z)
n
'\'"1 IXk
m
'\'"1 "/J
n n-1
i-1
(z-bi) n
m-1
s=l
(z-ds)
k=l
(z-ak) nm
1-1
(Z-Cj)
222 PROBLEMS ON COMPLEX ANAJ,YSIS
JJ, /'
81
c,
c,[
~ FIG.44
c c, A,1~
FIG. 46
c,
C 8,~Bz
c
FIG. 46 FIG. 47
/ '\\
I
I
I \
-1 I
1) 2)
FIG. 48
f(z) = 2~
k=l
log (z-a1c)- 2 ~log
J=l
(z-c1)+0, (6)
where a1c, c1 are the points on the x-axis corresponding to the vertices
A1c and 0 1 of the domain P; the h1c are the distances between the
cuts on the left, and the l1 are the corresponding distances on the
right. The points b1(i = 1, 2, ... , n - 1) and d8 (8 = 1, 2, ... , m - 1)
on the x-axis corresponding to the vertices B 1 and D,, are the roots of
the equation
n m
'1 h1c '1 l1 - 0 (7)
.L.J z-a1c - .L.J z-c1 -
k-1 J=l
How are the parameters 0, a1c, b1o c1, ds determined 1 What is changed
if one of the parameters a1c, b1o c1, a.
becomes equal to 001
FIG. 49
~!!I!..
f (z) = L.J 1'
log (z-a1;)- ~!:!..log (z-c)+Az+B,
L.J 1' J
(8)
k=l J=l
FIG. 50
How a.re the para.meters A, B, a", b1, cj, a. determined 1 Show that
if A 0 passes into a. point a0 #= oo, then
n m
L-H A
+--log (z-a0)+--+B, (10)
n z-a0
226 PROBLEMS ON OOMPLEX ANALYSIS
where
n m
H= };h1" L= };zi, A <0 and ImB = L.
k=l J-1
FIG. 61
where the a1c are points on the x-axis corresponding to the vertices
A1c of the strips of the domain P, and k1c are the distances between
the cuts. The point A 0 maps into oo, A> 0 and Bis a real number.
The points b1 (i = 1, 2, .. ., n) corresponding to the vertices B 1
are zeros of the derivative J'(z). How are the parameters A, B,
0, ~. b1 determined 1
OONFOBMAL MAl'l'INGS (OONTINUATION) 227
FIG. 52
w = /(z), which maps the upper half-plane Im z > 0 onto the domain
P, is of the form
/(z) = 2
n-1
k"
-log(z-arc)-
:n;
2l1
m-1
-log(z-c1
:n;
)+--+--+B,
A
z-a
0
z-eo
0 (ll)
k-0 J-0
/(z) =
n-1
2n k"
-log (z-arc)- 2l1n
m-1
0
-log (z-c1 )---+Az+B,
z-c0
(12)
k-1 J-0
228 PROBLEMS ON OOMPLEX ANALYSIS
FIG. 53
is known as the normal elliptic integral of the first kind in Legendre'a form. The
inverse function
z = sn (u, le) (2)
is one of the fundamental elliptic functiona of Jacobi, known as the sn function
of Jacobi. Two other elliptic functions of Jacobi a.re connected with it:
en (u, le)= y[l-sn1 (u, le)], dn (u, le) = y[l-lc1 sn1 (u, le)], (3)
In what follows, if nothing is said to the contrary, it is assumed that 0 < le < I.
The integrals
(5)
1
are known as complete elliptic intll{l1"ala of the jirat lcind, and le' = y (l - lc2 )
as the complementary parameter.
w= f J"[(l-z)~'+klz)]'
z
Find the inverse function and prove that it is doubly periodic
with periods 4K, 2K + 2iK'.
HINT. Make the substitution 111 == y(l-P).
1380. Investigate the mapping of the z-pla.ne by means of the
elliptic integral
u+iK'
iu
I
ksnu
. sn'u
idnu
-1snt:i""
l
---
. onu
snu
dn'u
t---
on'u on'u on'u
TABLJ!I 2
2i~'}
K+i11 I tk' sn' 1/ k' on' 1/
dn' 1/ - - dn' 1/ dn' 1/ -{z'<11>+
-K+i71 - dn'71
l ik' sn' 1/
dn' 1/
k' on' 1/
dn' 1/
_ {z'<11>+
__!E_}
2KK'
;+iK' kmE
1
I
-
idn;
ksnE
idnE
--- ,on;
snE
ion;
Z(E)-~
2K
in
E-iK'
ksn; ksnE snE zm+ 2K
im'71 l dn'71 -i{Z'(71)+ dn'11cn'11
i71 on' 1/ on' 1/ on'71 sn' 1/
n71 }
+2KK'
l i
iK' =t=- =Fi
k k
'--
Residues
232 PROBLEMS ON OOMPLEX ANALYSIS
ik ik'
k' , k' and verify Table 3, where the notation
K 1 = K(ki), Ki= K(lci) (lei= y(I-kf)) is used.
HINT. To establish the connection between z1 and z = sn (u, Tc) compare
the mappings of the first quadrants of the zrplane and the z-plane by means
of the functions F(1 , Tc1 ) and F(, Tc) (in the case of k1 = Tc' take the second
quadrant of the z1-plane). The necessary relationships are given in the third
column.
TABLE 3
K'l
-
k
I
1cu Tc sn (u, Tc) I dn (u, Tc) en (u, Tc) Tc(K+iK') TcK'
The integral
j Ji( l;~;:)dz
0
= jy'(I-Tcsins)d = E(,Tc)
0
(6)
is known as the normal elliptic integral of the second kind. By the substitution
z = sn u it is reduced to the form
Two functions of Jacobi, Z(u) and O(u) are connected with this integral:
E O'(u)
Z(u) == E(u)--u
K
=- -,
O(u)
(8)
CONFORMAL MAPPINGS (CONTINUATION') 233
where
(9)
is the complete elUptic integral of the second kind. The quantities K, E and K'
E', obtained by p&BBing to the complementary parameter Tc', are connected
by the relations
EK' +E'K-KK' = n/2. (10)
D _ K-IC _
- ~-
f 1
z1 dz
f[(l-zl)(l-lclz 1 )]
(11)
0
(12)
1
j v( 1;__k;:9)dz = i(K'-JC')
+i
(ft,JII( l-Tc't)
l-t dt-
ft dt )
y[(l-t )(1-Tc' P]
1 1
(13)
0 0
Ma.king use of the notation for the elliptic integrals given by (1) and (7)
these formulae can be written briefly as:
where
v
1
k
f (1 ;~: )
1
dz = i(K' -JC') .
234 PROBLEMS ON COMPLEX ANALYSIS
The integral
z
J
0
(l+sizl)y'[(l ~z9)(1-klz1 )]
4'
=J
d4'
(J +11sin1 4') y'(l-k8 sin1 4') = ll(4', 11' k) (14)
0
is called the normal elliptic mtegral of the third kitld. By the substitution
z = sn u it is reduced to the form
II
J
0
du
1+11snu' (l 5 )
is known as the oomplet.e eUiptio integral of the third kitld. The quantity ll1
can be expressed in terms of elliptic integrals of the first and second kindst.
1363. Consider the mapping of the z1-plane onto the w1-plane
by means of the elliptic integral of the second kind w1 = E(</J1, lei)
(z1 = sin </J1 = sn u1) for the following values of the para.meter
k1 : ~ k', ~' ; ~~ (0<k<1) and verify Table 4 (page 235).
HINT. See the hint to problem 1362.
1364*. Map the upper half-plane Im z > 0 onto the upper ha.lf-
plane Im w > 0 with two vertical cuts a.long the segments Re w
= a, 0 < Im w < k with the normalisation: w(O) = 0, w( oo) = oo,
w'(oo) > 0.
HINT, Use the Schwarz-Christoffel formula in the form
z
w(z) =Oki
0
J b1 -z1
y'[(l-zl)(l-klz)] dz
= 0 [Jz-.JIf( l-k z
1-z
1 1 )
dz-(1-klb)
Jz y'[(l-z)(l-klz)]
dz ]
;
0 0
t For the calculation of elliptic integrals of the third kind and of related
quantities see ERDELYI, MA.GNUB, OBERHlllTTINGER, Tm:com, (1953), Higher,
Tramllllndemal Funotiom, vol. II. Note that for "e (-oo, -1), (-1, -k1 ),
(-k, 0),(0, oo) the formulae for the calculation of (15) and (16) have different
forms.
TABLE 4
I
k' E(u1 , k1 ) I Ei I E'1
I Ui I
I
l/k I ku I - [E(u)-k'2u] I _!_ [(E-k'2K)-i(E'-k2K')] (l/k)E' 0
k I k I 0
z
~
!;lj
~
I
k' ! iu I 4u-E(u)+ snudnu] E' E ~
cnu I I Is:
~
I'd
....
z
14'.l
rJl
l/k' I ik'u l ~[k'2u-E(u)+ sn u dn u] ~, [(E'-k2K')-i(E-k'2K)] (l/k')E
k' cnu I I 0
0
z
'"3
z
ik ~
'"3
:, [ E(u)-k sn~c= u] (I/k')E ~, [(E' -k2K')-i(E-k'2K')] ....
v I k'u I 0
~
ik' . I
I iku ~ [ E'(u)-k'2 sn:i:~:' u] (l/k)E' _!_[(E-k' 2 K)-i(E'-k2K')]
T k
~
i:..:I
Ot
236 PROBLEMS ON OOMPLEX .ANALYSIS
for the determination of the constants use the values of a and h with the equa-
tion w{ ~ )-w(l) = 0 (it determines b in terms of k) and the condition
w'(oo)> O.
1365. Verify the last column of Table 2, (see problem 1361) and,
using it, consider the mapping of the rectangle R: !El < K, 1111 < K',
in the u-pla.ne (u = E+ifJ), by means of the function
dnuonu -
w=Z(u)+-
snu
HINT. Use the equations
J
II
'-'I
If; dt
--=If; '-' u- E(u+
) Bn u dn u ,
0 en t en u
II
The integral
FIG. 54
(21)
1370. Map the doubly connected domains 1-15 of the z-plane shown
in Fig. 55 onto the circular ring (h < lwl < es and determine the
modulus = e.le1 (see page 29).
In problems 1371-1373 map the given domains onto the unit
circle ltl < 1.
1371. The rectangle Re luj < K, Im jitl < K' (0 < Ii < 1). Find
the positions of the vertices in the mapping.
1372. The interior of the ellipse jz - lj + jz + 1J = 2a (a> 1)
with the cuts [-a, - 1], [1, a].
1373. The interior of the ellipse jz - lj + jz + lj = 2a (a> 1).
Find the positions of the foci in the mapping.
1374. Map the exterior of the unit circle jtj > 1 onto the do-
mains 1 - 3, of the z-plane indicated in Fig. 56.
1375f. Let R be the rectangle IEI < K, l'l'/I < K' in the plane
of u = E+in. Using Table 2 (see problem 1361) and the principle
of corresponding boundaries, prove that the functions
z dnu
-l = sn u-J-ig en u '
FIG. 55
2)
FIG. 56
240 PROBLEMS ON COMPLEX .ANALYSIS
11+ + +
~III
~nrro ~~H
~ILLi-i-
~IJJ ~ ~
Jn 111Ln n
~-_
FIG 57
242 PROBLEMS ON COMPLEX ANALYSIS
APPLICATIONS TO MECHANICS
AND PHYSJCSt
1. Applications to hydrodynamics:
Steady flows
The established plane irrotational flow of an incompressible fluid is described
by the analytic function
w(z) = t/l(a:, y)+iV'(a:, y) (1)
known as the complea: potential or chamcteriBticfunction of the flow; t/I is known
as the potential function, V' as the atream function. The curves t/I = const
are the equipotential linea, the curves V' = oonst a.re the stream linea. The
stream velocity V is connected with w(z) by the relations:
V = Vel = Vx+iV1 = w'(z)}
V = lw'(z)I, oi: = - arg w'(z), (2)
V=gradt/I.
Let 0 be a closed contour traversed in the positive direqtion (the contour
0 may also consist of the two sides of an arc traversed in opposite directions).
The quantity
I'= f Vada= f Vxda:+ V1 dy =f dt/I (3)
c c c
is known as the circulation of the vector V with respect to the contour O.
The quantity
Q= fVnda= f<-V1 d11+Vxd11)= fd'P (4)
c c c
(n is the outward normal to the closed contour 0, traversed in the positive
direction) is called the flow of the vector V through the contour O. Similarly
the flow of the vector V through the arc AB is defined as the integral J
Vn dB
AB
(the direction of the normal n must be indicated).
Combining formulae (3) and (4), WA obtain
243
244 PROBLEMS ON COMPLEX ANALYSIS
It is said that the term I'~Q log (z-a)(I', Q, real numbers) defines a vortear
BOUrce Of strength Q and intensity I', denoted by (a; Q, I')t, the term 2P _l_
nz-a
is a doublet with moment p, denoted by (a; p) (p is a complex number;
the argument of p determines the direction of the axis of the doublet passing
through the point a in the direction of a stream line), the remaining terms
(0-k)/(z-a)li define at the point a multipletB of order 2k.
Correspondingly, if at oo
n P I'+iQ IJ-1
w(z) = CnZ + ... +2- z + - 2 -.-logz+c0 + - - + ... ,
:n; m z
I'+iQ
then the term 2:n;i log z defines at oo a vortea: BOuroe of strength Q and
intensity r,
the term pz/2:n; defines a doublet with moment p (the direction
of the stream line at infinity coincides with the argument of p), the remaining
terms o,.zk define multipletB of order 2k.
~oints at which V = 0, and consequently w'(z) = 0, are known as /ltag
nation pointB of the fl.ow; from these points stream lines and equipotential
lines issue alternately. If a stagnation point is a zero of order (n-1),
these curves enclose an angle :n;/n. This branching of the curves may also
occur at infinity.
.In problems 1376-1389 it is required from the given complex
potential of the fl.ow to construct the equipotential lines and stream
lines, to determine V, the singularities and stagnation points, the
strength and intensity of the vortex sources, the moments of the
doublets and also to investigate the behaviour of the fl.ow at oo.
1376. W := CZ (c = IX + i{J).
1377. w = zn (in particular, n = 2, 3).
r+iQ
1378. w = - 2-.- log z. Consider, in particular, the cases I'= 0
:n;s
and Q = O.
I'+iQ z-a
1379. w =~log z-a
1385. w =!?.__log
2:n;
(z- _!_)z
1386. w=log(1+ : 2 )
Q
1388. w = az+ 2:n; log z (a > 0, Q > 0).
I'
1389. w = az+- 2:n;i. log z (a >0, I'>O).
1391. Find the complex potential w(z) of the flow in the whole
plane formed by the vortex sources {(ak;Qk,I'k)} (k=l,2, ... ,n)
and having at infinity the given velocity V00 = Ve1.
1392. Is it possible for stream lines to emerge from a point hav:ing:
(1) a vortex, (2) a doublet, (3) a vortex and a doublet together~
1393. Find the law of variation of a vortex source, doublet
and multiplet, situated at a point a or a.t oo, for the following
schlicht conformal mappings of the neighbourhood of these points
(Ci =fi 0, C-1 =fi 0):
(1) C= a+Ci.(z-a)+ ... ;
246 PBOBLEMS ON COMFLEX .ANALYSIS
a'(z)
C<z>=-=-+
a(z)
I
z
2'(--+-+-,
I
z-D
I
D
z )
D
(7)
where the summation extends to all D different from zero. The function
C(z) is odd.
The Weierstrass function p(:z:) with periods 2co and 2<.o' (see pa.ge 236)
is connected with C(z) by the relation p(:z:) = - C'(:z:).
Since
[C(z+2<.o)-C(:z:)]' == p(:z:)-p(z+2ro) = 0,
it follows that
APPLICATIONS TO MECHANICS AND PHYSICS 251
and similarly
where '1 and 71' are constants. Using the oddness of the function CCz), it is easy
to prove that '1/ = CC<) and 71' = CC<').
The quantities 71, 71', < and <' are connected by Legendre's relation
11<'-11'< = :n.i/2.
Let us use the notation
11 = 11u 11' = 71a and 11+11' = 11
< = <u <' = <a and <+<' = <2
The functions aa:Cz) are determined by the relations
flt,"a(z-<1:)
<>"kCz) = -e Ck= I, 2, 3). (8)
a(<t)
Correspondingly
C C) - alcCz) C9)
k z - <>"tCz)
The functions <>"tCz) are connected wit~ Weierstrass's pCz) and the functions
sn z, en z, dn z of Jacobi by the following formula.et.
a11:Cz)
vr pCz)-e11:] = aCz) (10)
a(z) a1 (z)
snu=y(ei-e8 ) - - , cnu=--, dnu = <>"2Cz) , (11)
<>"a(z) <>"aCz) <>"a(z)
where u = zy(e1-e3 ).
It is possible to express any elliptic function in terms of a(z) and C(z).
If f(z) has only the simple poles b11: with residues At (k = I, 2, . , n), then
n
f(z) = _2 AtC(z-bkl+O. CI2)
k-1
If f(z) has the zeros a11: and poles bt (of any multiplicity), then
The Theta functions of Jacobi are the functions 81M (i = 1, 2, 3, 4), defined
by the relations:
00 ( 1)"
fli(v) = i .J; (-l)q n-2 e(lt-l)nlu
n--oo
where q = em-r, T
= (J) 1 /(J).
The Theta functions are connected with the Sigma functions by relations
of the form
(16)
.,... 0 (
al( ~~) -- elma ~ ("i = l ' 2, 3) , (17)
81+1(0)
where z = 2'1v.
The advantage of the Theta functions is the rapidity of the convergence
of the series defining them. Using formula (16), the representation (13)
can be written in the form
(z-a1)o1(z-a)
01 2
2
(z-an)
... 01 2
f(z) =0 ' ' ' . (18)
z;:1 )o1(z;:) ... z;;:)
81 ( 81 (
By the use of formulae (11), (16) and (17) it possible to write the is
expression for the Jacobi functions sn z, on z, z in terms the Theta
dn of
functions.
In what follows it is a88Uilled that '
is a real number, '' is purely imagi-
nary, that is. the period parallelograms are rectangles.
(k = 2, 3, 4) and Z(u)
(see page 232), are reduced to the flows of problem 1423, (2)
by means of linear transformations.
1426. Show that the fl.ow defined by the complex potential
E(u) (see page 232) is reduced to the fl.ow of problem 1423, (1)
by means of linear transformations.
\HINT. As a preliminary prove the relations
E e ro
K 1 = (e 1 -e3 )ro and - =
K
--+-1].
e -e
2
1
K 3 2
t For this section see: N.E. KooHIN, I.A. KlBEL' and N.V. RoZE, TheO'l"lltfuaJ,
hydromechanics (Teoreticheakaya gidromekhanika), Chapter V, Gostekhizdat,
1948. For problems 1448-1450 see: F. OBERHETTINGER and w. MAGNUS,
..i.lnwendungen der elliptiBohen Funotionen in Physik und Teohnik, Chapt.er IV,
1949.
256 PROBLEMS ON COMPLEX ANALYSIS
where w11:(z) = w(z) - S(z, a1:), S(z, a11:) is the principal part of the expansion
of w(z) close to a,., the logarithmic term also being included in this if a11: is
a vortex source, that is, the velocity of the point ak is determined by the
action on it of all the remaining singularities. In this case ak = a11:(t) and
the complex potential w(z) depends on the time.
1440. Prove that two vortices (z1 = x1+iy1; I'1), (z 2 = x2
+iy2 ; I'2), acting on one another rotate round a "centre of
. t " I'1X1 +I'2X2 I'1Y1 +I'2Y2 .f r +r .J. 0 d
mer 18. Xe= r1+r2 ' Ye= --I'1+r--;:- l 1 2 r an
move steadily perpendicular to the straight line connecting
them if I'1+r2 = O.
1441. Along the real a.xis there is an impenetrable wall.
At a distance h from it there is located a vortex of intensity
I'. Prove that it moves parallel to the x-axis with velocity
I'
4nh
HINT. Apply the symmetry principle.
1442. Prove that a vortex located in a circle with an impene-
trable boundary circumference, moves a.long a concentric
circumference.
1443. Prove that n vortices of the same intensity I'; arranged
at a certain instant at the vertices of a regular n-gon inscribed
in a given circle of radius R, move along this circle with con-
I' (n- l)
stant speed 4 nR
1444. Prove that a vortex located in the first quadrant
x > 0, y > 0 with impenetrable walls moves a.long the curve
1 1
-+-=0.
xs y2
1445. The fl.ow of a fluid is ca.used by a vortex cha.in consist-
ing of an infinite number of vortices of identical intensity
I' situated at a certain instant at the points Zt = z0 +kl (k
= 0, 1, 2, ... ). Prove that the fl.ow is steady with com-
plex potential
I' 81 (z 4k a)
w(z) = - 2 . log-(---'---..,..-),
m i-2k+a
01 -4-.,,,-
where for the function (;Ji we have 2ro = 4k and 2ro' = il, and
that the vortex chain moves steadily parallel to the walls of
w(z) = -2:ni.
r [
log
o,(~)
( z-2h+a ) - log (
o, (+:{) ]
il. )
01 - - - - - z+2h-a- -
4h 0 2
1 4h
and that the vortex chains move steadily parallel to the walls
of the channel with the constant velocity
r [2
-- 2:n
a
C< 2a)- 217 k"+2
1 .p'(2a)
.p(2a)-e1 +2
1 .p'(2a)
.p(2a)-e3
J
E = - grad11 = -iw'(z)
E = lw'(z)I, a: = -n/2- arg w'(z),
Ex = -avfa:i: = au/CJy, E 1 = -a11/8y = -au/ax.
In all the problems of this section where it is a question of electrostatic
fields in domains bounded by one or several boundary contours it is assumed
that the potential function is constant along each simple contour (that is,
every such contour is a conductor).
If a is a pole of w' (z) and close to a the function w has the expansion
pi "l l
w(z) =
C-n
(z-a)n + ... + z=a +2qi og z-a + c0 +c1 (z-a) + ... ,
then the term 2qi log - 1- defines at the point a a plane point charge of
z-a
magnitude (! = 2q, denoted by (a; 2q) (on unit length of a rectilinear con-
ductor perpendicular to the z-plane at the point a, there occurs the charge
q); the term pi/(z-a) defines at the point a a dipole of moment p, denoted by
(a;p) (pis a complex number; its argument defines the direction of the
axis of the dipole); the remaining terms (z".::)k (k = 2, 3, ... , n) define at
the point a a multipole of order 2k.
Correspondingly, if at oo
2qi"log z+c0 +-z-
w (z) = cnzn + ... +piz+ C-1
+ ... ,
then the term 2qi log z defines at infinity a plane point charge of magnitude
(! = 2q, and the term piz a dipole of moment p.
If the function w = u+i11 is considered as the complex potential of an
electrostatic field E = -iw'(z) and at the same time as the fl.ow of a fluid
with velocity V = w' (z), then this leads to the following electrohydro
dynamical analogy:
1458. w = piz+2i .2
k=l
qk log-1-
z-a1c
(p > 0, q,. > 0,
al <as< ... <an).
1459. Find the law of variation of the point charge (a; 2q)
and of the dipole (a; p):
(1) In a single-valued conformal mapping;
(2) In continuation by the symmetry principle across
a rectilinear or circular segment of an equipotential line.
APPLICATIONS TO MECHANICS AND PHYSICS 261
1460. Show that the complex potential of the electrostatic
field formed by the point charge (a; 2q) in the arbitrary single-
va.lued domain D is determined by the formula
Let g(C,z) be the Green's function of the domain D (see page 165), the
r
boundary of which consists of the piecewise smooth simple contours I'u r,,
r r
... ' I'n; also let ti be the inward normal to and let be traversed in the
positive direction with respect to D. If u(z) is a function harmonic in the
domain D and continuous on I', then it follows from Green's formula that
u(z) = - 1-
2nr
J
u(C) 8g(C, z) 'da
an.
or
u(z) =
l J[u<C> ana log IC-zl
2nr
l 1 aum]
-log IC-zl --a,;- da.
If the domain D contains the point at infinity and the function v(z) is
harmonic there, u (oo) must be added to the right hand sides of the given for
mulae. Then in the neighbourhood of the point at infinity the Green's func
tion g (z,oo) can be represented in the form
The function
l
tJ0 (z) -2qlog-1- -
1 z-a
is known as the logarithmic potential of the point charge (a; 2q). In the
extended z-pla.ne tJ0 (z) represents the logarithmic potential of the two point
charges: (a, 2q) and (oo; -2q).
Let the contour I' satisfy the conditions indicated on page 262, and let
am and <C> be real and continuous on
The integral
r.
"(z) = J
r
(!{C) log IC~zl da
is known as the logarithmic potential of the simple layer wUh density (in am
three dimensions there corresponds to it the potential of a charged cylindri
cal surface with base r and surface density of charge : that is bearing the
(this means that the potential "(z) corresponds to a charge (oo; -2q)).
The integral
tJi.{Z)= J 11m an
8 1
log IC-zi ds
r
is known as the logarithmic potential of the double layer with density 11((:) (if r
is the boundary of a domain there is on I' a distribution of dipoles with axes
directed along the inward normal to I'; 11((:) is the density distribution of
dipole moments). If 8((:,z) is the angle between n and a vector going from
Cto z, and d<J> ((:, z) is the angle subtended by the element of arc ds at z, then
111(21) = J 11((:)
cos 8((:, z)
IC-zl ds =
J 11((:)d<1>((:, z).
r r
In particular, for a closed contour I' and 11 ((:) = 1
f :n
l'
(see also problem 1066).
log IC~zl ds = I 2n, if z is inside
n,
0,
if z is on
if z is outside
r,
r,
r
The Green function g(z, a) of the domain D can be considered as the po
tential of the electrostatic field formed by the point charge (a; 1) ifthe bound-
ary I' of the domain D is earthed. Problem 1475, (1) shows that in the
case a * 00 the earthing of r is e~uivalent to the placing on r of a charge
v(z) = ~ f
-oo
v(t) a~ log It I zj dt.
1
v(z) = v(oo)--
31;
f 00
ov(t)
- 0- l o g -
n 1 -
1t-z
I dt.
1
-00
Wn(z) = 1- 2n
1 f OWn(C)
on 1
log IC-zl d8.
r
If the domain D contains the point at infinity, then
P11< =- _1_
2n
f a<t(C) dB
an
= - _1_
2n
f <1 aro,,(C) ds
an
r1 r
a.re known as the mutual capacities of the boundary contours (some properties
of the numbers Plk are considered in problems 1101-1104).
268 PROBLEMS ON COMPLEX ANALYSIS
1495. Find the harmonic measures wk(z), and also the quantities
ek(C) and p 1k, defined in the remark to the preceding problem, for:
(1) The circular ring 1 < fzl <;
(2) An arbitrary doubly connected domain D, assuming that
the function which maps this domain onto the ring is known.
1496. Let D be the domain of problem 1494 and v(z) the bounded
potential of an electrostatic field which assumes the constant values
ak (k = 1, 2, ... , n) on the boundary contours (conductors) rk. Prove
the following assertions :
n
(1) v(z) =2 IXkwk(z).
k=l
(2) If the domain D is bounded and I'n is the external
contour, then
1
v(z) = 1Xn- 2:7t
J----a;-
av(C)
log IC-zl
1
ds;
r
if, however, D contains the point oo, then
1
v(z) = v(oo)- 2:7t
J----a;-
av(C)
log IC-zl
1
d.s.
I'
Prove that the right sides of these formulae are equal to IXk (k
= 1, 2, ... , n) in the domains complementary to D bounded by the I'k.
(3) The values of the charges of the layer induced on the
I'1 are given by
2qi =- 1
2:7t
J----an-
av(C)
ds
~
n
where
(4) 2~ JJ(gradv)
D
2 dxdy= 2Pik1Xi1Xk.
i,k=l
HINT. See problem 1102.
APPLICATIONS TO MECHANICS AND PHYSICS 269
(5) If w(z) is the complex potential of the field the density
of the induced layer is
(a#: oo),
(a= oo).
(2) Onto the plane with radial cuta (this refers to the plane with
cuts along segments on rays originating at the coordinate origin, or with cuts
along concentric circular arcs with centre at the coordinate origin). The
function f(z) is determined by its zero a, the pole b and the coefficient A of
the expansion
f(z) = Iz~b
Az+
+ei(z-b)+ ...
C-1
z
+ ...
(b #: oo),
(b = oo).
(3) Onto a clialc with radial cuts or with cuts along concentric circular
area (centre at the coordinate origin). The function f(z) is determined by
the conditions
f(a) = 0, f'(a) = l
J
c
Qncls = -k J: ds = - cJdv
c 0
k
w(i) = [ ...
C-1
+--+eo+ci(z-a)+ ...
111-a
]
+-q 1
2n k log--,
111-a
The following similarities exist with fluid flow and the electrostatic fileld:
Complex
potential w(z) = u+i11 w{z) = u+i11 =
iw(z) -11+iu
Vector field Q = -Tcgradu V=gradu E = - gradu
= -Tcw'{z) = w'(z) = w'(z)
u Temperature Potential Potential func-
function tion
u = const Isothermals Equipotential Equipotential
lines lines
17 Stream function Stream function -11 is the force
function
11 = const Stream lines Stream lines Lines of force
Source (a; q) Source (a; - ! ) Point charge
(a; 2~)
Doublet Doublet Dipole
A thermal field A fl.ow deter- A field with given
with given sources, mined by the com- charges, dipoles
doublets and iso- plex potential and equipotential
thermal boundary iw(z), with given boundary con-
contours vortices and doub- tours.
lets, streamUning
the boundary
contours
CHAPTER I
1. (1) -i; (2) -i; (3) ! (1+3i); (4) -8.
2. (1) 3, n/2 (here and below only the values of a.rgz a.re given);
(2) 2, n; (3) y'2, n/4; (4) y'2, -3n/4; (5) y'29, tan-15/2;
(6) y29, -tan-1 5/2; (7) y'29, n-tan-15/2;
b
ta.n-1- - n for a
a
< 0 and b < 0.
1 iy'3 y'3 i
3. (1) 1,-2-2-; (2) 2+2 -i;
(2k+: )n (2k+{)n]
(8) y'2 [ cos 3 +isin 3 (k = 0, 1, 2);
(k = O, 1, 2, ... , n-1).
274
ANSWERS AND SOLUTIONS 275
13. z, = z1 +z 3 -z 2
14. The ratio (z8 -z1 )/(z1 -z1 ) must be a real number (the condition is
necessary and sufficient).
34. (1) Each curve is the geometrical locus of points, the product of the
distances of which from the points z =--L.ancLz.-= l is constant (lemniscates
with foci z = 1). For ..1. > 1 the curves of the family are simple closed
curves, for ..1. < 1 they separate into two simple closed curves which for ..1. -+ 0
shrink to the points 1. For ..1. = 1 we have the lemniscate of Bernoulli;
its equation in polar coordinates is rl = 2 cos 2.
(2) Lemniscates with foci at the points z1 and z20 where z10 z1 are the
of the equation z1+az+b = 0. The lemniscates consist of a single curve if
..1. > V(. lza;zil), and of two curves if ..1. < V(lza;zil). For
..1. = v(
(z1 +z1 )/2.
lzi;Zil) we have a lemniscate of Bernoulli with the double point
-11 -77
59 28 -23
8
61
49
sin~{J ( n{J}
(2)
sin~{J
sin{J/2
. (
sm oi:+T .
n{J}
61. (1) sin{J/2 cos oi:+-2- ;
65. (1) sinz = sin(x+iy) = sinxcoshy+icosxsinhy,
lsinzl = y(sinh1y+sin1x);
(2) cosz = cosxcoshy-isinxsinhy, lcoszl = y(cosh111-sin1x);
(3) tan z -_ -l sin 2x+isinh 2y I _ y(sin22x+sinhl 2y) .
. , ltan z - . ,
2 cos1x+sinh2y 2(cos2x+sinhly)
(4) sinhz = sinhxcosy+icoshxsiny, lsinhzl = y(sinh1x+sin1y);
278 PROBLEMS ON COMPLEX ANALYSIS
(4 > 8+ 15i
sinh 8-i sin 4
17 ' (5) 2(cosh2 4-cos1 2)' (6) '
Retanz = 0, if Re z = T;
kn Im coshz = 0, if lmz =Ten, or Rez = O;
87. (1) The images of the straight lines a:= 0 are the circles u 1 +v- ~ = 0,
for 0 = 0, the axis u = O; the images of the straight lines '!/ = 0 are the
circles u 1 +v8+ ~ = O, for 0 = O the axis "= O; the images of the circles
lzl =Rare the circles lwl = l/R; the images of the rays arg z =a: are the rays
arg z =-a:; the image of the circle lz-11=1 is the straight )ine u = ~;
280 PROBLEMS ON COMPLEX ANALYSIS
(2) The originals of the straight lines U=O are the circles w+y- ~ = O,
for 0 = O, the a.xis a:= O; the originals of the straight lines t1 =0 are the
circles a:1 +y1 + ~ = 0, for 0 = O, the a.xis y = O.
= z+ _!._
88. The function w maps the circles lzl = R + I onto the ellipses
z
u8 t1 8
Fm. 58
ANSWERS AND SOLUTIONS 281
90. On the ray going along the negative part of the real axi11 from the
.
point w = - 41 to t h e pomt
. w = oo.
91. (1) The circles / = eC, the rays (J = 0, the logarithmic spiral ! = e;
(2) The curves y = e""+2kn.
92. (1) The family of straight lines a: = 0 is transformed into the family
v1 =-4a1 (u-a+ !) (parabolaswithfociatthepointw= -!,a=O+ !),
1
to which also belongs the ray going from the point w = - 4 along the negative
part of the real axis (a = 0); the family y = 0 is transformed into the family
of confocal parabolas 118 = 401 ( u+o+ !), to which also belongs the ray going
from the point w = - ~ along the real axis in the positive direction;
(2) The family a: = 0 is transformed into a family of circles of Apollonius
with respect to the points w = - 1 and w = 1 (including also the imaginary
axis); the equation of the family of circles of Apollonius is: (u - a)1 +v = a
-1; !al > l (a = coth 20); the family y = 0 is transformed into a family
of circular arcs with ends at the points w = -1 and w = 1, including also
the corresponding parts of the real axis; the equation of the pencil of circles
is ul+(v+b) = l+bl (b =cot 20); ,.
(3) The family a: = 0 is transformed into the family of spirals(/ = eCl- Co.
the axis a: = 0 corresponding to the segment (J = 0, 0 < (! ~ 1 ; the family
~-Cl
y = 0 is transformed into the family of spirals (/ = e' , the axis y = 0
corresponding to the ray (J = 0, 1 ~ (! < oo.
98. To the straight lines y = 0 correspond the curves u = a:+e"" cos 0,
I}= o+e"" sin 0, to the rectilinear segments a:= 0 there correspond arcs
of the curves u = o+ec cosy, v = y+ec sin y (Fig. 59).
94. (1) To the family lwl = R there correspond the circles r = (cos ~)/log B
I
~.._~~~~ ......~~~~-ci--~-+--i----~~
I
I
I
:r
0
y=~ 1/
......
......
c ' ',:r=f
'\
\
' \
,y=1
I \
l:r=-4 \
I I
\vO
FIG. 59
ANSWERS AND SOLUTIONS 283
109. o.
118. (1) No, if u #: const; (2) /(u) = au+b.
119. lf(z)I is not a harmonic function, arg/(z) and log IJ(z)I are harmonic.
88u 1 8u l 8 2u
120. Liu = a;a +-;-a;-
+-;:a at1i ; u = 01 log r+o
UU. P1 =a:, qi = y; P1 = a: 2 -y, qi = 2a:y; Pa = a:8-&::y1, qa = 3a:ly
-'If'; P = a:t-6a:ly1 +y; q, = 4a;ly-4a:y8; 'Pn = ,.n cos 114', qn = ,.n sin r&t/I.
122. tJ (a:, y) = 2a:!f+y+0.
y
128. v(a:,y) = - a:+y +o.
124. (a) v(a:,y) = argz+O; (b) v(a:,y) = argz+2mn+O (the expression
for arg z in terms of a: and y is given in problem 2, (10)).
125. (a) v(a:,y) = argz-arg(z-1)+2mn+O; (b) v(a:, y) = argz-
arg (z-1)+0; (c) v(a:, y) = arg z-arg (z-1)+2mn+0.
n n
126. (a) v(a:, y) = 2 cx1c arg (z-z1c)+2n 2 mu1+0;
k-1 k=l
n n n
(b) v(a:, y) = 2 cx1c arg (z-z1:)+2nm 2 cx1:+0 {if 2 cx1: =0 the function
k-1 k-1 k-1
v(a:, y) is single valued in the given domain!).
127. (1) It exists; (2) It exists; (8) It does not exist.
135. u = 0 1a:y+02
186. u = 0 1 log (z8 +y1 )+01
01a:
187. u = - 1- - 2 +oz.
a: +Y
188. u = 0 1 y[a:+ J"(za+y)] +01
189. It does not exist.
140. /(z) = el:z: 2 e.
141. f (z) = ele'.
4
(4) (J = :ri - tan-1 8 , Tc = IO.
For w = zl: (I) 0, Tc= 3; (2) (J =- 0, Tc= 3/16; (3) (J = :ri/2, Tc=- 6;
(J =
4
(4) 9 = - 2 ta.n-13' Tc = 75.
I I
UH. (1) A compreBBion for lzl < 2' a stretch for lzl > 2;
I I
(2) A compression for lz+ll < 2' a stretch for lz +II> 2;
(3) A compression for lzl > I, a stretch for lzl < I;
(4) A compreBBion for Re z < 0, a stretch for Re z > O;
(5) A compression for lz - II > 1, a stretch for lz - II < I.
152. S = Jf1f'(z)l 8 da:dy, L =f 1/'(z)lds.
G I
ua. y2(e8"- I).
154. 2e 2 (e1 - I).
155. The domain D is the ring e < lwl < e. It is not possible to apply
the formula of problem 152 since the mapping is not one-one.
CHAPTER II
1156. w = (l+i)(l-z).
157. w = (2+i) z+l-3i.
158. (1) z0 = -1+3i, 8 = 0, k = 2, w+l-3i = 2(z+l-3i);
(2) z0 = 2+2i, 8 = n/2, k = 1, w-2-2i = i(z-2-2i);
(3) there is no finite fixed point;
(4) if a = 1 there is no finite fixed point; if a+. 1, then
w1 -az1 w1 -az1 w1 -az1 )
Zo = - - - , 8 = arga,k = lal,
1-a
w---- l-a
(
=a z - - -- ,
l-a
y(l+k1 ) -i(-j-+tan-ik)
(3) w = b e z,
y(l+k1 ) -1(.!!.+tan-1k) .
(4) w = e 2 (z-ib1 ).
b1-b1
162. w = el 11 Rz+w0
163. (1) The family of straight lines u = l/a parallel to the imaginary
axis (not including the imaginary axis itself);
(2) The family of straight lines 11 = - l /b, parallel to the real axis
(not including the real a.xis itself);
(3) The family of circles b(u+v)+u+v = 0, touching the straight
line tJ = -u at the origin (also including this straight line);
(4) The pencil of straight lines v = -ku;
(5) The family of circles passing through the coordinate origin and
through the point Wo = l/Zo (the straight line which passes through the points
w = 0 and w = w0 also belongs to this pencil);
(6) The cissoid u 1 = -vl/(v+ 1).
285
286 PROBLEMS ON COMPLEX ANALYSIS
164. (1) Into a family of circles touching at the point w = h the straight
lines, respectively parallel to the imaginary and real axes (including also
these straight lines); the equations of these families a.re;
+ (v-hg)8]-(u-h1) = O;
(O-:i:o)[(u-h1 ) 1
(0-y0 )[(u-h1 ) 1 + (v-h8 ) 1l+ (v-h1 ) = 0,
J..<I
I
I I
I 'i I
i--'-1
I 1i.
FIG. 60
segment a:i z1 internally and externally in the ratio .A. If use is ma.de of the no
tation indicated in the diagram (0 is the centre of the circle with diameter
AB), and we put lz1 -z1 1= d. then for A< 1 we have the relations:
r1 R .Ad. _Aid. d.
A= cos cc=-=-, B =--.-, r1 =--.-, "1 = - -1;
R " 1-,.. l-A1 l-.A
. . . z-z1 (J
(2) Circular a.res passmg through the pomts z1 and z1 : arg-- = .
Z-Zz
The arcs corresponding to the values (J = cc and (J = n-cc are complements
and form a complete circle;
(3) To the polar net there corresponds (Fig. 61) a net consisting of the
.ANSWERS AND SOLUTIONS 287
z-z
circles of Apollonius - - 1
Iz-z. I== z-z1
R and the arcs arg--
z-z.
= 8 orthogonal to
them (if 8 > 0, then the area a.re situated on the right of the direction z~~
if8 < 0, they are on the left of it);
B
FIG. 61
(4) to the upper semicircle there corresponds the right angle indicated
in the figure.
166. The semicircle lwl < 1, Im w < O.
167. The domain containing the point w = 0 and bounded by arcs of the
170. The doubly connected domain the boundary of which coneists of the
straight line Re 10 = ! and the circle / 10 - : / = :
(2) 10 = ~(~-1)+hi
da-~ z
or 10 = ~(~-1)+1+hi;
~-da z
(3 ) 10 = d1(z-da) .
z(d1+d2 )
(1) =_ 2i(z+l) (2 ) = (1+2i)z+6-3i
172. 10 4z-l-5i ' 10 5(z-i)
(l+i)z+l+3i iz+2+i 1-i
173. (1) w = (l+ilz+ 3 +i ; (2) 10 = z+l ; (3) w = - 2 -(z+l).
174. (l) 10
= (-1+3i)z+l-i.
(l+i)z-l+i '
(2) 10 = z2z(l-i)-(4-i)
(l-4i)-2(1-i)
. ;
(3) 10 = z (3-i)- (1 +i)
(l+i)(l-z)
175. 10 = IZ-1
.z-i ; the upper half.plane maps onto the unit disk.
176. (1) 10 = (az+b)/(oz+d), where a,b,o, dare real numbers and ad-bo> O;
(2) 10 = (az+b)/(cz+d), where a,b,o, d are real numbers and
ad-bo < 0;
(3) 10 = i(az+b)/(cz+d), where a,b,o, d are real numbers and
ad-bo < O.
2 2z+l)
177. (1) 10 = 2 _ 21 ; (2)10= -2 ( - -
z-2
178. w = (B-z)f(B+z); the image of the upper semicircle is the quadrant
u> O,v < O.
179. (1) 2 ~i ; (2) : +i.
180. (1) lzl = 2; (2) the straight line a: = ! ; Iz - ~ I= ~ ;
(3)
identical with the sign of arg z1 /z1 Rays issuing from the point w = 0 in the
half-plane Re w > 0 correspond to circular arcs, situated within the disk
lzl < l and passing through the points z1 and z1 in the z-plane; the semicircles
with centre at the point w = 0 in the halfplane Re w > 0 correspond to
arcs of the circles of Apollonius with respect to the points z1 and z1 situated
within the disk lzl < l.
w-b z-a
188. ----
w-b
= el11.-=
z-a
w-ii z-a
189. - - =i--_-.
w-a z-a
sin<1>-Asin8
190. (1) O (</>) = oi:-<J>+2arg(el<f>-a) = oi:-<1>+2tan-1 A
COS</>- COS 8
, where
a= Aelll.
el
(2) w'(O) = (l-lal 1 )el, w'(a) = l-ja!"
(3) Ifa#= O, then the domain lying inside the disk lz - !I< V(,:i. -1}
is stretched, and the domain outside this disk is compressed. (The circle
z-2+i
19'. w = 2 .
u+ 2 - 2
i
z-a w-b . z-a z-a
195 (1) w = R 8e11 - - (2) - - = e' 11 - - (3) w = R 1 - -
. R-az' R-bio RB-az' RB-az'
where a is a real number and Jal < I.
196. w = (l+k)z-(za+kzi) a, where k =-. f(l-lzsl 1)elar1(1-i.z,),
(1-k)z- (z1 -kz1) l-Jz1J1 JI
Jal= lz1-z1l - - - -
Jl-z1z1J+y[(l-Jz1J2)(1-lz111)]
az-l+Jl'(l-a2) l-y'(l-a2)
197 w = [l-y(l-a1 )]z-a' !I= 2 a
z+2-y3 . z-2+)1'3
200. (1) w = z; (2) w = l+( 2 -y 3 )z; (3) w = t l-( 2 -y'3 )z"
toa. Circles passing through the point Zo and having at this point a tan
gent determined by the vector h.
(R-lci)z-R1
204. w = z-(R+lci) , where k is a real number, k O (fork= oo it +
is necessary to put w = 1).
209. If Jal <sin ; , then the transformation is elliptic. If we use the
.a 1(11+~-.!.) fJ , 1(11+.!_.!!.) fJ
notation: Jal = sin 2 sin(J, 11 = e II II tan 2 z1 = e I 2 oot 2
Ao
tan T = tan
.a cos (J, then the transformation can be written in the form
2
w-zi = elA. I-Ii .a
If Jal = sin 2 , then the transformation is parabolic and
W-11 1-1.
1
is of the form - -
1
= - - +iki0 where z0 = e
I(11+ .!_.!.)
11 I ,
A
h = tan -
W-Zo Z-Zo 2
If Ja I > sin ; , then the transformation is hyperbolic. If we use the notation:
J. I I . fJ 1 =e
mn2=asm,z 1(11+f-P) zI = e 1(11+-i-,.+.8) ,
W-Zi
form - - -
w-z1
= KZ-Z1
---
z-z 1
ANSWERS AND SOLUTIONS 291
211. r
~ + 0 [(:
= Y+2 tan-1Zo
n sin,,
for small : .
I ~~-~ I ~~-~ d
Z1 = z1+ d Zs= z1+ d , = lz1-z1I
I
~= 2ffM+d1-f1-f-y([d2-(r1+r1)I] [dl-(r1-r2)1]);
I)
( or-
=
I (d+r1-u1) (r1-u1)
--------
(d+r8 -u1 ) (r1-u1)
I
I"
.HO. (1) = 2; (2) = 5+2 y6.
223. The group will be finite if a/n is rational.
225. The fundamental domains (some of their possible forms) are shown
shaded in Fig. 62. Equivalent boundary sides are connected by arrows.
Points with figures are fixed points of the rotations leading to the groups (the
figures indicate the number of rotations). For the last five examples the paral-
lelogram of the doubly periodic semigroup is indicated; in example (7)
this is a square, in examples (8) and (9) it is a rhombus with angles 120
and 60.
RElllARX. It can be shown that apart from a linear transformation groups
(3)-(9) exhaust the groups of linear transformations with one limit point
(known as the limit point of a set of points equiva.lent to one another)t.
w-l z-1 w-i z-i
226. (1) w = elotz; (2) - - = el--; (3) - - = el--.
w+l z+l w+i z+i '
4 w-a =e"'~
( ) l+aw l+az
232. (1) and (2). The construction is obvious; (3) The equidista.nts of the
"straight line" a.P (a. and p a.re the "infinitely distant" points of this straight
line) are circular arcs with ends at a., 8 (known as hypercycles); (4) The li
miting lines of a pencil of "parallel straight lines" with common point "at
infinity" a. are circles touching (internally) the unit circle at the point a. (they
are known as oricyclea).
288. (1) For the construction of a "rectilinear" triangle with angles tf> 1,
r1> 1 , r1> 8 , construct a circular sector OAB with central angle LI = n- (r1> 1 +r1> 1
+r1> 8 ) draw the "straight line" AB, and through the points A and B "straight
lines" at angles t/> 1 and t/> 3 to AB, intersecting at the point O. The triangle
ABO is the one required.
234. (1) w = i(z1 -a1 ); (2)w = y(z- ~ }-iy'(p/2).
235. (1) w = z1/a1 ; (2) w = y(z/a); (3) w = 2A (z+l
21 _ 2
)'
I)
I J)
I
2 2
4)
l
7) 8)
FIG. 62
ANSWERS AND SOLUTIONS 293
286. The domain is bounded by Pascal's lim~on u = B(ooslfi+moos 21/1),
"= R(sinlfi+m sin 21/1). If the coordinate origin in the w-plane is
transferred to the point w =-Rm, then we obtain the equation of the ~n
in the usual form (in polar coordinates): (! = B(l+2m cos 0). For m = 0
Pa.sea.l's lim~on becomes a circle, for m == 1/2, a cardioid with cusp w
= -B/2. The images of the circles lzl = r < 1 are also lim~ns of Pa.seal
the polar equations of which are easily obtained by transferring the coordi-
nate origin to the point u = -Rmr2 : (! = Br{l+2mr cos 0). The images
of the radii of the circle arg z = a: are parabolas passing through the coordi-
nate origin: m(u sin 2a:-tl cos 2a:) 2 +R sin a:(u sina:-t1 cos a:) = 0. To the radii
a= 0 and a = 1r correspond the segments of the real axis O~ u~ B(l +m)
and R(m-1) ~ u ~ O.
237. The domain bounded by the parabola u = -t18 and the curve
Q = 2 cos (0/3), 101 < Sn/4 (Fig. 63).
Fm. 63
a ici '
2 (y(4)+l)ea za
l
240. (1) w =z; (2) w = ---""---:-""""'----
(V
c4)-2) e
1 +3 j/4 ~J
241. (1) w = ( _ z )' ;
l-J-_ (2) w =- z+ l ; (3) w = -2zl+3z-2 .
1-z 2z 2zl-J-3z+2
z1 -J-2iz-J-l 2z1 -J-3iz-J-2
242. (1) w = iz+2z +i; <2> w = 2z1 -3iz+2
243. w = [(z-1)/(z-J-1)]8.
l 1
244. (1) w = z +R
( .!. .!. ): (2) w =( z~ -R~).
z-R z+R
a
245 w = -(2z-J-y(3)-i)I.
2z-y(3)-i
250. w=e~'(z+l
z-1
)f.
246. w = - ( 2z-J-y(3)-i
2z-y(3)-i
)a. 151.
152.
w = y[(z+ 1)/(1-z)].
w = y[(z-J-i)/(i-z)].
2z+ y(3)-i) 153. w = y[(z-z 1 )/(z1 -z)].
(
247 " w = 2z-y(3)-i 254. w = y[(z-J-R)/(z-R)].
a tel
.. ~ 8 = (2z-J-y(3)-i)I. 255. w = e - ... y(z-i)
Q w i 2z-y(3)-i 256. w = y(z1 -J-h1 ).
z-y(2)(1-i) ]' y(zl-J-hl)
249. w = [ z-y(2)(l-J-i) 157. "' = -----"--
'Z
FIG. 64
4t1 4111
162. (1) and (2). The exterior of the ellipse 1 + 1 = 1
(R+ !) (R- ! )
(Fig. 65, (1), (2)) ;.
(3) and (4). The whole plane with a cut along the segment [-1, l]
(Fig. 65, (3), (4));
(5) and (6). The wholepla.newithcutsalong the rays (- oo, -1] and
[1, oo], on the real axis;
(7) The lower half-plane;
(8) The upper half-plane;
(9) The upper halfplane;
(10) The upper half of the interior of the ellipse
4'11 4111
(R+
4u1
! r 4v1
+ (R- ! )' = 1
FIG. 65
4ul 4tJ!
263. (1) To the circles lzl =R correspond the ellipses
(R- ~ r
+ ----,-,-
(R+ ~r
= l (to the circle lzl = l corresponds the segment u = 0, -1 =::;;;; v =::;;;1), to
the rays arg z = r1. correspond the branches of the hyperbola v1 /sin1 r1.-u,lfcosr1.
= l (to the rays = 0 and = :n; corresponds the axis v = 0, to the ray
= :n;/2 corresponds the ray u = 0, v ~ 1, to the ray = -:n;/2 corresponds
the ray u = 0, v=::;;;; -1);
(2) Tothecircleslzl=Rcorrespondtheellipses (
4ul
1a) +( as)
4tJI
R+R R-R
= l (to the circle lzl = a corresponds the segment v = O, -a=::;;;; u =::;;;;a), to the
rays arg z = r1. correspond the branches of the hyperbola
a 1 cos1 r1.
= l (to the ray arg z = 0 corresponds the ray v - 0, u ~ a,
to the ray arg z = :n; corresponds the ray v = 0, u =::;;;; -a,
to the rays arg z = ; corresponds the axis u = 0);
(3) The families of confocal ellipses and hyperbolas obtained from
the corresponding families of problem 261 by a rotation through an angle y
and a magnification with coefficient lol (the centre of the magnification is
the coordinate origin).
ANSWERS AND SOLUTIONS .297
ef
264. (1) w = -(z+
c
y(z1 -c1));
(2) w = _2_b (z+ y[z1 - (a1 -b1 )]); in both cases for one choice of the
a+
branch of the root we obtain a mapping onto the exterior of the unit circle,
and for the other onto the interior;
az-by[z1 -(a1-b)]
(3) w = a-b .
265. w = A(z+ y[zl- (a1 -b1 )]), where A is an arbitrary comple:ir num
a-b
ber; = y(al+b1 )-y(bl+kl)
266. The whole plane with a cut along the segment [-1, ! (a+ !)] a>O; if
the whole plane with cuts along the rays [-oo, ~ (a+ !}] [-1,+oo)
and
if a< O.
the lengths of the arcs corresponding to the cuts are equal respectively to:
274. (1) The image of the circle 0 is a circular arc with ends at the points
1, inclined at the point l at an angle 2ct to the real axis; the exterior
of the given circle is mapped onto the whole plane with a cut along the given
arc;
!I
u
Fm. 66
(2) The image of the circle 0 is (see Fig. 66) a closed curve (Zhu
kOtJBkii profile) with cusp at w = 1, the tangent at this point ma.king an
angle 2ct with the real axis; the circular arc with ends l discussed in
part (1), is contained in the domain bounded by the Zhukovekii profile; the
exterior of the circle 0 is mapped onto the exterior of the Zhukovekii profile.
276. (1) The image of the circle 0 is a closed curve consisting of two cir-
cular arcs with common ends at the points 1, the tangents to these arcs
at the point l ms.king with the real axis angles respectively equal to 2ct-M
and 2ct+ (n-cx)c5; the exterior of the circle is mapped onto the exterior of the
domain bounded by the given arcs; the image of the circle 0' (see Fig. 67)
is a closed curve with an angular point at w = 1, the tangents at this point
being inclined to the real axis at angles respectively equal to 2ct-M and
2oi:+ (n-cx)c5; the image of the circle 0 is contained in the domain bounded
by the image of the circle O'; the exterior of the circle O' is mapped onto the
exterior of the image of this circle.
(2) The interior of the circle 0 is mapped onto the exterior of the
domain bounded by a.res of circles passing through the points -1,1, the tan-
gents to these circles s.t the point l forming with the real a.xis angles respec-
tively equal to: (a) 2oi:+ (n-cx)c5, 2oi:+ (2n-oi:)c5, if the function w(z) is
defined in the z-pl.a.ne with a cut a.long an arc of the circle 0 lying in the
ANSWERS AND SOLUTIONS 299
lower half-plane, (b) 2oc-d, 2oc- (n+oc)d, if the cut defining the func-
tion w(z) is made sloug an arc of the circle 0, which lies in the upper hslf.
plane.
lJ
FIG. 67
w-l
276. w+l =
( zelY+ie'~ )
1
_,1.. ,wherey= oc, if P> O,andy-oc+nif P<O.
zel7+ie
277. The whole plane with cuts along the rays y = 0, a: ~ -1 /2 and
y = 0, a:~ 1/2.
278. The hslf-plane a:> 1/2 with a cut along the segment y = 0,
1/2~ a:~ I.
279. The whole plane with cuts slong the rays y = O, l ~a:< oo and
y = 0, -oo < a:~ -1.
280. The angle -n/n < arg z < n/n with a cut a.long the ray y = O,
Solution. The function C= ! (z + 21~) maps the sector onto the lower hslf.
300 PROBLEMS ON OOMPLEX ANALYSIS
plane, the points oi: a.ndoi:e :'passing into the points 2-(oi:n+_.!._), Next,
2 ix"
it is necessary to compress the half-plane 1' = 1 C and map it onto
2 (ix"+ ix-")
the unit semicircle
required.
T = 1'+ y(pl-1). The function w = y T is the one
(2) w
n n )-
= ( oi:I +oi:-s n
.!.( zl +z-I + V![(zR +z-S)a- (oi:I +oi:-8)']).!n
n 11 .. n n n 11
184. w = ..
JI
I[ y(b
Ji'(z'+c')+ y(a+o)] and
1 +o y(z +c')
1) - 1
I
w =- p {y(z+o)+oi:+y[y(z1 +c'+oi:)-tJ8]}.
I I
where oi: = 2" [t'(a1 +o')-y(b'+cl)], tJ = 2" [y(a1 +o1)+ y(b1 +o1)].
185. w = y[y(z1 +c')+ y(a1 +o1)].
188. W=-*(v[(:::r+1J+v[(:::r-1J)
i ~ ..!!..
= y2 [(z+y(z1 -1))2 + (z+ y(z 1 -1)- ]
.!.
SOLUTION. By means of the function T = ~, where C= z+y(z- I) is the
inverse function of z = ~ (z + !), the upper half-plane is mapped onto the
(2) The logarithmic spiral e = e-A:- (for Ii= 0 the ray tJ = b);
(3) The angle oc < tJ < p (for IX = 0 and p = 2n the plane with a cut
along the positive part of the real axis);
(4) The whole plane with a cut along the logarithmic spiral Q = e;
(5) The sector e < 1, O < tJ < IX (for IX = 2n onto the unit disk with
a cut along the radius " = 0, 0 ~ u ~ l);
(6) The domain e > 1, 0 < tJ < IX (for IX = 2n onto the exterior
of the unit circle with a cut along the ray"= o. 1 ~ u < oo);
(7) The domain e < (! < efl, y < tJ < d (for d-y = 2n this domain
is a concentric ring with a cut along the segment tJ = y, e ~ (! ~ efl).
100. The angle 0 < arg (z+n) < n/n; the strip 0 < '!/ < n.
301. (1) The rectangular Cartesian net u = 0, "=O; (2) A straight line;
(3) The strip 0 < " < IX; (4) The half-strip u < O, 0 <" < 1X;
(5) The rectangle log r 1 < u <log r 1, 0 <" < 2n.
I ~. I '1o
802. (5) b ==a tanhT , l =a ta.n2 I I
.ANSWERS .AND SOLUTIONS 303
303. (1) The family a:= 0 is transformed into the family of confocal hy-
perbolas with foci at the points 1, ( c::: 0 - si:: 0 = l); the family y = 0
into the family of confocal ellipses with the same foci (cos~ 0 +~: 0 = l) ;
(2) The upper half-plane; (3) The fourth quadrant;
(4) The right half-plane with a cut along the segment [O, l];
(5) The whole plane with cuts along the real axis along the rays
(- oo,- l] and [1, oo);
u t1
(6) the interior of the ellipse cosh h +Binh h = 1 with cuts along
the segments [ - cosh h,-1] and [1, cosh h].
804. (1) The half-strip - ; <u < ~ , v> 0;
2nl.r
.r-2 y(2;-1 (y(.r)- l)
111. w = e 81'. w =e
= _ 1 +iy'S m
(z+Si), "' !..!
.r-1
815 (l)
w 2 tanh 4 (z - i) , (2) w = - e +2 -
"'.!...!.1 '
e z- +2+i
ni.r+I
,., __
2e z-I
(3) w =---.-+...,I
l+e .r-I
323. w= V( l+oos 4;
4n
cos-21- - cos a4n
)
z
points oosh a. Hence it is easily obtained that the function w = sin-1 sin z
oosha
will map the given half-strip onto itself so that the rays a: = n/2, a.::;;; y < oo
will correspond to the rays u = n/2, 0.::;;; " < oo. On applying the symmetry
principle an infinite number of times we see that the function found is
the one required.
b sin-l i sinh z
330. w = cosha '
sin-1 (l/oosh a)
1
. 1 1 .. / [( . 1 1 )1 ( 1 sin z ) ]
331. w = sm- "'CciSiil.I +V sm- "'CciSiilJ - sm- "'CciSiil.I
. 1 sinz
sm- oosha
332. w = ..
V!['"'-:~ -~-:~].
sm- cosh a - sm- cosh a
333. w == sin-1 e21s . Solution. The function C= ells maps the strip 0 < a:
< n/2 onto the upper half-plane and the function w = sin-1 Cmaps the upper
half-plane onto the half-strip - n/2 < u <n/2, "> 0. The function w = sin-1ells
which maps the strip onto the half-strip transforms the rays a: = 0, - oo < y
< O; a:= n/2, - oo < y < 0 into the rays a:= n/2, 0 < y < oo; x = n/2,
0 < y < oo, respectively. Applying the symmetry principle an infinite number
of times we see that this function effects the required mapping.
134. (1) If b < 2n the curvilinear rectangle: 1 <!I< e", 0 < (J < b; if
b = 2n the ring 1 <!I < e" with a out along the segment [l, e"]; if b = 2Tcn
(Tc= 2, 3, ... ) a many-sheeted domain consisting of k rings 1 <!I < e", out
along the segment [l, e"] and attached in such a way that the lower edge
of the out of the first ring is attached to the upper edge of the out of the
second ring, the lower edge of the out of the second ring to the upper
edge of the out of the third ring and so on; if b = 2Tcn+P (Tc= 2, 3, ... ,
0 < p < 2:ir), then to the free lower edge of the last ring of the surface being
constructed it is necessary to attach along the segment [l, e"] the curvilinear
rectangle 1 < !I < ea, 0 < 0 < /J;
(2) The domain with an infinite number of sheets consisting of rings
1 < !I < e", out along the segment [l, e"] and attached by the method
indicated above;
(3) The infinitely many-sheeted domain consisting of the rings
1 < !I < e" out along the segment [l, e"], enumerated by means of the integers
( ... , - 2, - 1, 0, 1, 2, ... ) and attached so that tile lower edge of the out
of each ring is attached to the upper edge of the out of the ring the number
of which is one unit higher. This domain is the part of the Riemann surface
of the function Log w which lies above the ring 1 <!I< e".
381i. (1) The two-sheeted domain obtained by joining together two right
half-planes, each of which is out along the ray " = 0, 1 .::;;; u < oo; the edges
of the outs are attached criss cross, that is, so that the lower edge of the out
of the first sheet is attached to the upper edge of the out of the second sheet,
and conversely;
306 PROBLEMS ON OOMPLEX .ANALYSIS
(2) The two-sheeted domain consistlng of two planes with cuts along
the real axis along the rays -= < u =s;;;; -1 and 1 ~ u < CXl and attached
criss cross along the cuts -= < u =s;;;; -1. The edges of the cuts 1 =s;;;; u < oo
remain free.
336. A two-sheeted domain consisting of two planes cut along the imaginary
axis along the segment -1 =s;;;; " =s;;;; 1 and attached so that the left-hand edge
of the cut of the first sheet is attached to the right-hand edge of the cut of the
second sheet. The remaining edges are free.
887. The Riemann surface is infinitely many-llheeted and has two loga-
rithmic branch points above the points w = 0 and w = = Domains of single
valuedness in the 111-plane corresponding to sheets of the w-plane with cuts
along the positive real axis are bounded by the circles 2Tcn(x'+y)+y = O
(Tc= 0, 1, 2, ... ).
888. r 1 = <Xl, r 1 = r 8 = 1; the image of the disk 11111 < r 1 is the whole
plane with the point w = -1 deleted; the image of the disk lzl < r 1 (and of the
disc 11111 < r 8 ) is the half-plane Re w > - !.
1 1
889. (1) r1 ="j"; (2) r1= 2 lal; (3) r1=l.
1
848. (1) r 1 = 1/4; (2) r 1 = 4 lal; (3) r,
= 2 - y3.
144. r 1 = n, r1 = 1.
1
345. (1) r8 = 1/2; (2) r1 = 2 lal; (3) ra = I.
CHAPTER III
847. (1) For example, the whole plane with a segment removed;
(2) For example, the domain bounded by two circles touching internally.
354. BoL'O'TION. The .Jordan curve y can be considered as the topological
image of the segment LI (or, if the curve is closed of the circle LI). Let us assume
that y contains the circle K : lz - z0 1< (I. On LI the circle K corresponds to
some continuum, consequently, the segment [a, b]. The arc Yab CK joining
the images of the points a and b on y, also corresponds on LI to a conti-
nuum, moreover it belongs to [a, b] and contains the points a, b, consequently
it coincides with [a, b]. But this is impossible. However, continuous curves
exist which contain interior pointst.
858. BoL'O'TION. (1) We divide up y into sufficiently small parts, draw chords,
map them linearly onto the corresponding parts of the segment 0 :s;;;; t :s;;;; 1
and obtain the polygonal path n; then map the segments connecting points on
y and n which correspond to identical values oft, continuously onto the segment
OE;; A:s;;;; l;
(2) and (8). We first replace y (in the circle or in the ring) by a homo-
topic polygonal path.
862. SoL'O'TION. Let 0 be the centre of Q, y an arbitrary closed path in
Q not passing through 0, p an auxiliary circle with centre at the point 0 such
that the path y lies in the ring between p and Q, and lJ = ""fl' (0'), where
p', 0' are the images of p, O. In the circle Q the path y is homotopic to an
.,,,,. (0) times traversed circle p (see, 858, (3)) consequently, y' the image of
y is homotopic in Q' to an .,,,,.(0) times traversed path P', hence .,,,,.,(0')
= ni.(O) lJ. In particular, choosing as y the original of some circle y' around
0', we have ni.(0') = 1, whence we conclude that lJ = 1. Now let a be an
arbitrary point of Q. As above we conclude that fly(a') = .,,,,.(a)lJ4 , where
da has a meaning similar to lJ for 0. We show that lJa = lJ. For this we draw
in Q the circle p, enclosing 0 and a. Then: f6fJ(a) = f6fJ(O) = 1, f6JJ (a') = .,,,,. (0),
""fl' (a') = iJ0 , ""fl' (0') = lJ, whence it follows that iJ4 = lJ. In the case
of a continuously differentiable mapping with a Jacobian different from
zero the infinitesimal behaviour is equivalent to an aftine transformation
(see problem 875).
363. SOL'O'TION. Divide up the domain G by Jordan arcs into the domains
Gt, (i1 = 1, 2, ... , tii) with diameters less than 1. If /(z) of:. a, on these arcs
then denoting by y11 the boundary of G1,1 , we have
n,
f d arg [f(z)-a] = 2 J
d [arg /(z)-a] of:. 0,
,, l,=111,
hence a domain <P.1 (s1 is the value of i 1 ) bas been found, such that
307
308 PROBLEMS ON COMPLEX ANALYSIS
FIG. 68
2 Po
cos 2 Po
= - - - ~ cos2- (prove this!). In the resulting half-plane we widen the
cos 2 J!.. 2
2 -
vertical half-strip on the segment AB as base to the length of the arc AB,
and moreover in such a way as in the result to preserve lengths everywhere
on the arc AB. The resulting quasi-conformal mapping has the charac-
te1'1B _..-
. t'ic p""=' (i +-n/30) sec 22
/Jo
.ANSWERS AND SOLUTION'S 309
i in n
197. (I) 1 1 = ~a+i, 1 1 = 1+2; (2) 1 1 =2' 1 1 = -2; (3) 11 ,. id',
1. = -nB1
198. (1) vs( 1- ~ ); (2) 2; (3) 2i; (4) o.
899. m.
400. 4/3.
Bn+1
401. (1) n+l [(- l)n+l_ I], if n + -1; m, if n = -1;
(2) and (3). 0, if n +
-1; 2ni, if n - -1.
402. (1) -2(1 - i); (2) 2(1 - i); (3) -2(I+i); (4) -4; (5) 4i.
408. (1) 2ni; (2) -2n; (3) 2di; (4) 2di.
404. (1) 2ni, if n
n+I
+ -1; -2n', if n = -1; (2) (-l)n+l 2ni if n
n+l
+ -1;
-2n1 if n = -1.
elGuri-I
405. I+et , if IX+ -1; 2ni, if ct= -1,
407. (1) letl::s;;;;n/2; (2) sinpet;;;::.O.
421. (1) n/3; (2) -n/3; (3) o.
422. If the contour 0 contains the point 0 as an interior point and does
not contain I and -1 then I= -2ni; if it contains only one of the points
-1 or I and does not contain the point 0 then I = ni. It is clear from this
that the integral may assume five different values (-2m; -m; 0; ni; 2ni).
423. 2n - I, if n > I; 2, if n = I.
424. ni/2. 429, (1) 2/3; (2) 1 - 2i/3.
425. (sin a)/a. 445. Convergent.
a} 446. Convergent.
(
426. e" 1 + 2 447. Divergent.
427. (1) I; (2) -le; (3) I-le. 448. Non-absolutely convergent,
449. Non-absolutely convergent for ~ :;(; 2lm (k = O, 1, 2, ... ),
divergent for ~ = 2kn.
450, Converges absolutely. 459. e.
451. Diverges. 460. I.
452. Absolutely convergent. 461. I.
458. Divergent. 462. I.
454. Absolutely convergent. 468. 1/4.
455. B =I. 464. l/e.
456. 00, 465. I, if !al ::s;;;; I; Iflal, if !al> 1. -
457. o. 466. I.
458. 2.
467. (1) B; (2) B/2; (3) oo; (4) 0; (5) B"; (6) B, if !Zol ::s;;;; I, and B/llol
if lz.I > I. .
310
ANSWERS AND SOLUTIONS 311
468. (1) R ;;;:i:: min (r1, r 1); (2) R ;;;:i:: r 1r 1; (3) R r1/r2 <
1
469. (1) z/(l-z) ; (2) -log (1 - z); (3) 2 log [(l+z)/(1-z)]; (4)log (l+z).
1
[ 1 1 ] 1 1
... + (k+l) 2 - l - (k+2)1 -3 - (k+2)1 -2 - (k+2) 2 - l
1 1 ] 1 1
> (2k+l) [ (k+l)-1 - (k+2)1 -3 - (k+2)1 -2 - (k+2)2 - l > O.
'l'his proves that the given series converges for z = I. If lzl = 1, but z #: 1,
then use must be made of the convergence test of problem 440, putting
a,,= (-l)[vnlzn, bn = .!.. We have
n
1-zl l-z 1-z' 5
Sn= -z-z1 -zl+z'+ ... +(-1)
[yn)
zn = -z--+z'---z--+
1-z 1-z 1-z
...
1-zSP+l
... zP'. (z(p+t>'+ +z")
1-z
where p = [yn] - I.
Hence lsnl < ll~zl +2p+3, consequently for every z there exists a k such
that lsnl < kp < kyn and this completes the proof.
00 00
"'1 212n "'1 212n +1
482. ~ (2n)!, R = oo. 483. ~ (2n+l)!, R = oo.
n~o n=O
312 PROBLEMS ON COMPLEX .ANALYSIS
00 00
"1 21111-1z "1 2-lz'rt
484. ,L.J (-1)11+ 1 ( 2n)! , R =co. 485. 1/2+ ,L.J (2n)! , R =co
r
n=O n=O
00
487. l+i
y2
[1+_!_~+
2 i
2
n=l
(-l)n-1 1.3 ... (2n-3)(~)1, R
2. 4 ... 2n 1
= 1.
f, anzn R =
488. ,L.J (-1)" bn+i, Ia.I
b
n=O
[";1]
wherec,.= "1 (-1)'11( n }2-tm-131111, R=yl3.
,L.J 2m+l
m=O
00
492. 2 (-l)-
2- -,
n+l
zsn+1
R =I.
n=O
00
497. ! +2 2(-1)11+1
11=1
<;:-;t, B = 3.
oosin(1+tm)
511. ,2; n! 2 (z - 1)111, B = oo,
n=O
z'
508. I+zI + 3 + ...
z z'
504. 1-4-96+ ...
I 5 3
505. I+z- 2 z1+ 6 z'-4z1 + ...
5 5 13
506. 1+z+z1+ 6 z11- 8 z1+ 30 z11+ ...
z 2zl 9z5
507. z+ 21 + 31 + 51 + ...
00
z
2
zl
8
z'
508. log2+-+----+
192 ... = log2+
2"
--z11+1,.
n+I
n-o
I
Co= 2' 00 +nc1+n(n-I)o1 + ... +nlo,.- 1+2nlo11 =I, n;;;;i. I.
xza+ ...
ab a(a+l)b(b+l) 1
581. w = F(a, b, c, z) = 1+ T.Cz + 21 c(c+l) z +
... (b+n-1) 11 I I
... + a(a+l) ...n!(a+n-l)b(b+l)
c(c+l) ... (c+n-1)
z; z < 1 .
d1C dC
z(l-z) dzl +[(c+l)-(a+b+3)z]dZ-(a+l)(b+l)C =- O. (1)
Since the function ! F(a, b, c, z), being the derivative of the function
F(a, b, c, z) which is analytic at the point s = 0 is also analytic at the point z = 0,
and every solution of equation (1) analytic a.t the point z = 0 must be of the
form kF(a+l, b+l, o+l, z) (see problem 581), where k is a constant, it
d
follows that dZ F(a, b, o, z) = kF(a+ 1, b+ 1, o+ 1, z). Putting z = O, we
find that k = ab/o.
538. (2) 2n!M.
545. (1) 4; (2) 15; (3) 3.
546. (1) A zero of order k+Z; (2) A zero the order of which is not lower
than min (k, Z); (3) A zero of order k - l, if k > l; a regular point not being
a. zero if k = l, and a. singularity if k < l.
547. The points z = 3i a.re zeros of the first order.
548. The points z = 3i a.re zeros of the first order; the point at infinity
is a. zero of the second order.
549. z = 0 is a zero of the second order; z =kn (k = 1, 2, . ) are
zeros of the first order.
550. z = 2 a.re zeros of the third order; z = 2kni (k = 0, 1, 2, ... )
are zeros of the first order.
551. z = 2kn (k = 0, 1, 2... ) are zeros of the second order.
651. z = n are zeros of the third order; all the remaining points of the
form z =kn (k = 0, 2, 3, ... ) are zeros of the first order.
553. z = n/4+kn (k = 0, 1, 2, ... ) are zeros of the first order.
564. It has no zeros.
555. z =kn (k = 0, 1, 2, ... ) are zeros of the third order.
556. z = 0 is a zero of the second order; z = kn (k = 1, 2, ... ) are
zeros of the third order.
316 PROBLEMS ON COMPLEX .ANALYSIS
168. (1), (2) and (3). It does not exist; (4) It exists {/(z) = z~ 1 )
164. (1) It exists (J(z) = z1 ); (2) it does not exist.
161. There is no contradiction, as the point z = 1 does not belong to the
domain of analyticity of the function.
166. The limit point can only be the point at infinity.
00
167. SOLlJ'TION. From the expansion /(z) = 2 c,,(z - z0) it follows that
n-o
- 00
u(z, 71) = co+co +2 {(c.[(z-:t:oHi(1/-11o)]n +c;;[(z-zo)-i(1/-yo>l"}
2 n=l
Putting here z = 1110 + C~ Zo, y =-Yo+ C~ Zo, where Cis suftl.ciently close
to z, we obtain (establish this!)
!27 (; r
co
575. ! +2 n-o
zn for lzl < 1; - z~l + 2
n=O
(-1) (z-1)
00
22
n=l
(-1) z!n - 2 ;:
n=O
1 for 1<lzl<2.
co
578. - _ i _ - __l __ + ~ (n+S)i (z-i) for 0 < lz-il < 2;
4(z-i) 4(z-i)B .LJ 2n+c
n=O
co
2
n=l
(-1) z=+a for lzl > I.
317
318 PROBLEMS ON COMPLEX ANALYSIS
00
00 00
Cn=2-"c-a(n=l,2,. .. ), C0 = : 2 [1+,k:(-})2-].
00
oo sin{1+ n2n)
585. - ,J;ni-(z-l)" for 0< lz-11 < oo;
n=O
. cosl , sinl-2 cosl 6 sinl-5 cos2
-sml- --ZT 21 zl + 3! zs ~+ ... or lzl > I.
00
3 ,J; nan
(see problem 615), for 0 < lzl < 2n; - +2
z --+-
zan 1
n=l
+ ~[(-1)" 22nR1n
.L.J (2n) !
+-2-]zan-1
nn
for n < lzl < 2n .
n=l
00
687. \-, b" - all for fzf > max (la!, !bl) .
.L.J nzn
n=l
k-1
where o- 1 = 2i tan-1 1/2, 0-1Tc = 21lci ( tan-1 1/2- ,2: ( 2m1;l~):m+t),
m=O
'-c1t+11 = 20- 11: (k = 1, 2, .. ), for 1 < lzl < 2.
689. (1) Yes; (2) Yes; (3) No (the point z = 1 is not an isolated singularity);
(4) No; (S) No; (6) No; (7) No (in any ring round the point z = 0, the function
is not continuous); (8) No; (9) Yes; (10) Yes, if ex is an integer or zero; no
in all the remaining cases.
590. (1) No; (2) Yes, both branches can be expanded; (3) No; (4) Yes, all
three branches have expansions; (5) No; (6) Two branches of the four determined
y
by the conditions (l+yl) = y'2 have expansions; (7) No; (8) No; (9) No;
(10) Yes, all six branches have expansions; (11) No; (12), (13) and (14) Yes,
all the branches have expansions; (15) No; (16) No; (17) All i;ne branches have
expansions except for two determined by the values sin-1 y'2/2 = n/4.
594. Onto the whole plane with a cut along the ray e-loit, : ~t < oo.
595. SOLUTION. If f(z) = + ... , then denoting by w 0 the value not
z+c 2z2
assumed by the function f(z) in the unit circle, we consider the funct.ion
320 PROBLEMS ON COMPLEX ANALYSIS
f(z) / ( 1- f :: } It is single valued in the circle lzl <l (why?) and has the
expansion
1
~~ = z + ( o1 + ~o } z1+ .. . By the theorem of problem
Wo
ao+a1z+ +anzn
of the numbers am different from zero) or
(Z-C1:1) (z-ix1 ) ... (Z-IXn-1)
(an #- 0, IXk #- a1 for k #- l).
636. (1) z0 is a removable singularity; (2) z0 is a pole of order n, if cf>(z)
is single-valued in the neighbourhood of the point z0 , and a pole of order nm,
if cf>(z) is m-valued in the neighbourhood of this point; (3) z0 is an essential
singularity.
637. (1) The point z: is a pole of order n if y' is a rectilinear segment, and
a regular point of multiplicity n, if y' is a circular arc, that is, J(z) - f(zt)
322 PROBJ,EMS ON COMPLEX ANALYSIS
= (z-zcT)" t/>(z) where t/>(z) is analytic in the neighbourhood of the point zf and
t/>(zf) < o. If z: = oo, then this condition is written in the form /(z) - /(oo)
= z-nt/>(z), where t/>(z) is analytic at infinity and t/>(oo) .fi O;
(2) Z: is an essential singularity.
689. (1) -1; (2) O; (3) O; (4) O.
Ha. (1) An essential singularity at z = oo; the exceptional value is 0 (and
ro!); ez-+ 0 if a:-+ - oo (art:-+ oo, if a:-++ oo);
(2) An essential singularity at z = O; the exceptional value 0 (and oo);
-1
e 11 -+ 0, if z -+ 0, for example, along the path y = 0, a: < o (e 11 -+ oo as
-1
z -+ 0, along the path y = 0, a: > 0);
(3) An essential singularity at z = 0; there are no exceptional values
(not counting oo); cos .!._ -+ oo for a:
z
== 0, y -+ 0;
(4) An essential singularity at z == oo; the exceptional values are i and
-i; tan z-+ i, if y-+ + oo, and tan z-+ -i, if y-+ - oo;
(5) An essential singularity at z == oo; the exceptional value is -1;
tan z-+ - 1, if '!J-+ oo.
1
6'8. For example, /(z) = (l+z)I = 1-2z+3zl- ... ; 11111 = -n, 'an+i
= n+l.
650. (! = o. 655. (! = 2, '1 == Jl'5.
61H. (! = n, a= a. 656. (! = 1, '1 = 1.
652. (! = 1, '1 = 3. 657. (! = 1, '1 = 1.
658. (! = 1, '1 = 3. 658. (! = 1, '1 = y2.
654. (! = 3, '1 = 2. 659. (! = 1/2, '1 = 1.
1
660. (! = 2m, '1 = l .
oo ztn I
SOLUTION. 2 ---
n=O (2 .n)I8
=-
2
(cos yz + COB iyz),
'Vz)+cos i j/z).
00
whence
"\"'1
nL::.o zn
(2.n)! = 21 (cos
Similarly
n=O
J; (2~n 1
.n).
= 1/4 (cos otVz+cos ot1 Yz+cosot3 V'z+cos otVz),
V-1, 2~ = -4l~
00
zn
where ot = whence ,,c.; cos afc 'l /z and so on.
n=O (.....-.n).
k=l
661. (! = 00.
662. SOLUTION. It is easily seen that it is sufficient to consider values of
z> O.
I I
f e.z:t'dt Jart:''dt
Then _o_ _
art:
<I
'
on the other hand, if 0 < ot < 1, then e"%
ANSWERS AND SOLUTIONS 323
1 1
= J ez(t1 -1Z)dt;;;;i: J eZ(t 1 -1Z)dt-+oo ifz-+OO,
o v
Consequently, fl= 1, a= I.
66'. fl* = max (fl1 fl>
665. (1) fl*= fl a .::;;;a1+a1; (2) fl*= fl, a= max (a1, <11)
666. (1) fl*.::;;; fl a.::;;; 2a; (2) e.::;;; fl a.::;;; a.
667. SOLUTION. Let M(r), fl a be the maximum of lf(z)I on the circle lzl = r,
the order and type of the function /(z), and let M 1 (r),(11 and a1 be the correspond
If
ing characteristics for the function f'(z). From the equality /(z) = Jf'(t) di
0
+J(O) it follows that M(r).::;;;rM1 (r)+IJ(O)i and, consequently, e.::;;;e1 Since
J' (z) == ~
2m
f 'c;c>d~
.,-z)
where as r can be taken the circle with centre at z and
r
radius r+~ (~ > 0 is arbitrary), it follows that M 1 (r).::;;; M(2r+:: (r+~l ,
that is fll.::;;; fl and hence, fl1 = Cl Hence also from the inequalities given
above we conclude that a 1 = a. Another possible method of proof is based
on the theorem given on page 99.
669. fl = 1, a = l/e. 675. fl = 1, a = 2.
670. fl - a, a"" oo. 676. h(</I) = cos"'
671. fl= a, a= 0. 677. h(</I) = !sin"''
672. fl = o. 678. hC> = !sin"''
678. fl = o. 679. h(</I) = !cos .,.
674. e == 1, a..,. 1. 680. h(</I) = cos n</I.
681. h(<fl) = cos </I, if cos <fl;;;;:, O; h(<fl) = 0, if cos <fl < 0.
682. h*(<fl) = h(</I), if h(</I) > 0; h*(<fl) = 0, if h(</I) < 0; h*(<fl).::;;; 0,
if h(</I) = o.
CHAPTER VI
gence of the series ~ 1Gnzn is obvious and from the identity ~ anzll
~ -zn ~ 1-zn
=- 2 .2;
00
an-
00
n-1
an(_!_
zl
rn it follows that the series also converges for
n=l
lzl > 1.
n=l n=l 1-h-)
co co
If the series J; an diverges then the series J; anzn has the radius of conver
n=l n-1
co
gence R :s;;; I. For lzl > 1 the divergence of the series ~ anzn follows from
~ 1-zn
n=l
00
the fact that if this were not so the series ~ - 1 Gn would converge, and
~ -zn
n-1
co co
consequently also the series 2( ~ 1 zn - ;.:_z:n } = 2 an. If, however lzl < 1,
n=l n-1
2
00
then the modulus q of the ratio of the general terms of the series anzn and
n=l
co
L ~z~ is contained within the limits 1 - lzl :s;;; q :s;;; 2 and consequently
n-1
both series converge or diverge together.
324
ANSWERS AND SOLUTIONS 325
co
694. (1) 2 bnzn, where bn = 2a,,, the summation extending to those
n=l
indices p, which are divisors of the number n, including 1 and n. The radius
of convergence R =min (r, 1), where r is the radius of convergence of the
co
series 2 anzn .
n-1
co co
~ (-l)D ~ (logk)D
695. ~ aa(z-2)n, where an = ~~ kl
n=O k=l
R = l.
696. 1/2, if lzl < 1, and - 1/2, if lzl > I.
'748. The integral converges uniformly in the intervals 0 < :s;;; z :s;;; {J < 1 '
and 1 < y :s;;; z < oo.
'744 .An example is:/(t) =- e11 for n < t < n+e-n' (n =I, 2, . ) and/(t) = 0
for all other values of t.
'74li. :I:c = a:a = 0. '749. flJc = - oo; a:a =-+ oo.
'746. :l:c = :I:a = - 00. '750. :l:c = - l; a:a =+ oo.
'74'7. :I:c = a:a = + oo. '751. :l:c = 0; flJa.,. I.
'748. a:c = - oo; a:a =I.
'752. Divergent at a.11 the points of the boundary.
'753. Absolutely convergent.
'714. Convergent at the point z = 0, convergent (non.absolutely) at the
remaining points of the boundary.
'755. Non-absolutely convergent at all the points of the boundary.
'765. SOLUTION. On evaluating the integrals present in the obvious ine-
qualities
:!.. ~ n
I I I
f smn+l:i;da; < f sinna;da; < f sinn-la;da;,
0 0 0
. [ (2n)!! ] 1 l n [ (2n)ll ] 1 1
we obta.m (2n-l)ll 2n+l < 2 < (2n-l)ll .2n. In order to prove
Wallis's formula it remains to establish that the difference between the extreme
members of this set of inequalities tends to zero as n .,.. oo.
'76'7. (1) No; (2) Yes.
'769. (1) Divergent; (2) Diverges (to zero); (3) Convergent; (4) Convergent.
'7'70. Converges non-absolutely.
'7'71. Divergent.
'7'72. Convergent if p > !, the convergence being absolute if p > l;
divergent if p .s;;; 1/2.
> 1; divergent if p .s;;; 1.
'7'78. Absolutely convergent if p
'7'74. Absolutely convergent. '780. lzl < l/e.
'7'76. lzl < I. '781, lzl < oo.
'7'7'7. lzl < 2. '781. lzl < oo.
778. lzl < oc. '781. 11 < oo.
'7'79. lzl > 1. '784. lzl < oo.
'791. SOLUTION. (1) From the series C(a) =- I+ ~ + ~ + ... by subtract
the terms !, for which n is divisible by 2 a.re absent from the right-hand
m=l
(1-p;;;) converges for Re a;;;;i. l+c5, the function C(a)
n-1
(1-p;Ci+" = C(l~c5) From this it is easily
00
concluded that lim fl (1-p;<i+">) = O. Hince (l-p;;-1) < (1-p;<i+">),
1-oO n=l
00
it is obvious that the product fl (I-p;;1) diverges (to zero), and conse
n=l
00
quently, the series I p;;-1 also diverges.
n=l
CHAPTER VII
res [/(z)]:r=-al = - 514 (sin 3+i cos 3); res [/(z]:r=co = 217 (sin 3~3).
800. res [/(z)]:r= d:+i,. = -1 (le= 0, 1, 2, ... ).
I
co
=-sin 2 [2co (2n-l)!
41n
(2n)!
")1 41n+1 ]
+ -.J (2n)I (2n+l)I
n-1 n-0
l
809. res [/(z)]:r=o = 1/2; res [/(z)]
z--.-
d:ni = -2le
:n:i
(le= 1, 2, ... ).
810. res [/(z)]z=o = 0, if n < 0, and also if n > 0 is odd; res [/(z)]z=o
..!!.
(-1) . .
= (n+l)I' if n = 0 or n > 0 is even; res [/(z)]:r=co = -res [/(z)]:r=o
328
ANSWERS AND SOLUTIONS 329
l
811. res [J(z)] 1
z-bi"
= (-l)k+1-k
1 1 (k
:rr;
= 1, 2, ... );
00
2 ~ (-l)k 1
res [J{z)]:r=oo =
:rr; ,L.J ---,CS- = - 6.
k=l
812. res [f(z)]z=fcl"' = (-l)k2k8:rr;1 (k = 1, 2, ... ).
. . 2tk(211k-l)
813. res [f(z)],.=o = 0, if n is odd, res [J(z)],.=o = (-l)k+l (2k)! B 1t,
if n = 2k (k = 0, 1, 2, ... ), where B1 t are the Bernoulli numbers (see
1
problem IH5); res [f(z)] (k+ 1) = - (k = 0, 1, 2 ... ).
z= I" ( k+2
l )":rr;n
814. l; - 1.
815. O; 2.
816. -2e2knai, if yl = 1 and Logl = 2km; 0 for the branch defined
by the value yl = -1.
(a-b)2
s17. -- 8-
818. (1) a.-{J (for all the branches); (2) e'% - efl (for all the branches).
if Tan-1 oo = ( 2 k+l):rr;.
2
821. res [f(z)],.=o = 0, if n;;;.. O; res [J(z)]z=o =Log P' if n = -1, and
2n+1
836. - - - i f n;;;:i: -1, and 0 if n < -1.
(n+I)I
838. 0, if,. < I; I if,. > l (the sign depends on the choice of the branch
of the integrand).
I 841. entl/(l+e").
839. 7v(I+y'2). 842. 2n/v(a1 -I).
843. 2na/(a1 -b1 ) 111
840. n/(l+a). 844. (2a+b)n/[a(a+b)]111
845. 231/(1-a1 ) if lal < l; 231/(a8 -l) if !al> l; 0 (the principal value)
if !al = 1, a + l (for a = I the principal value does not exist).
. n(a1 -f-l) . n l-a11
846. n(a'+l)/(1-a1 ), if lal <I; a'(a-I), if !al> I; 2 a'(a-I)
(the principal value), if !al = 1, a + I (for a = I the principal value
does not exist).
847. 2n/nl, if n ;;;:i: O; 0, if n < 0.
848. nisigna (for a = 0 the principal value of the integral equals 0).
849. -2ni signima.
851. -n/27. 852. n/4a.
8..,. 3 . 1.3.5 ... (2n-3) n , if n> 1 ,. n /2 , if n -_I .
2.4.6 ... (2n-2) 2
854. n 855. n/y2.
ab(a+b) 857. ( - l)k-1/(2ih-z)k.
858. (1) ~(cos 1-3 sin I); (2) ~ (3 cos I+sin 1),
3e' 3e1
ne-lallJ
860. 861. ..:!.. e-Jalllsign a.
2 lbl 2
859. ~ (2cos2+sin2)
2e'
863. ni, if t> O; 0, if t = O; -ni, if t < 0.
864. n (2sin2 - 3sin3).
865 ..:!.. (cos l - l/e8).
5
866 ..:!.. [sina+e-Cav'a)/I (sin (a/2)+y3 cos (a/2))].
3
867. n(e-labl-1/2) sign a.
labl 869. ~ [2- (2+ labl)e-fallf] sign a.
868. ~ (1-e- ) sign a. 4b'
2b8 870. n (lbl-lal)
871. n/2. 872. 3n/8.
873. (1)
I'(p) cos n: I'(p) sin n:
; (2) - - - - ; i n order to verify the correctness
aP aP
ANSWERS AND SOLUTIONS 331
the answer for -1 < p < 0, it is sufficient to note that for these values
of p the integral converges and the function in the answer is analytic.
87'. - 1
p
r (1)
-p cos -2p
n 875. -1
p
r (1).
-p ~m -2p
n
876. _l_ r
p-l
(2-) cos n/2p (for p = l the integral is equal to n/2).
p
878. n/sin p:;i.
880. n/2 cos np
2
n(l-p)
881. - - - (for p = 1 the integral equals 1/2).
4cos np
2
882. ~ ~pl , if l #:- o, and~ if l = o.
smpn sm l smpn
892. If b does not belong to the interval (0,1), then I = ,..:!-bll- 1 (b- l)-ll
smp:;g
where (b-1)-11 > 0, bll- 1 > O for b > l; if 0 < b < 1, then I = -nb11-1
X(l-b)-czcotp:;g (the principal value).
T' ,r -,:..J.
k+ -
2
:ri;2
2n-l
905. - l { l :ri; ~
--+- (-I)t+1 2k+I }.
2n l + a 2 2a ( l )2
k=O log2 a+ k+ 2 :ri;2
907. 2
n :ri;a
tanh 2 . 909. 1/(2cos ; ) .
:ri;2e""'
908. (e""'+l)B
~
1
929. -+2
a
2oo
( -l)ne -- ntnt sin
a - -a-
nnr
931. 2sinh t
t
n-1
-oo
du. 932. 2'"
a
J -e-x-t
x
1'3. SOLUTION. Let us represent the integral in the form -dt
-x t
00 _,._,
+ J _e- - dt. Since on asymptotic expansion in negative powers of a;, e-x....., 0,
x t
then by the solution of problem 144 it follows that the second integral is
asymptotically equal to zero. By the definition of the principal value of an
integral,
= e-x lim
B-+0
J
x
- - -et
e-t
t
- dt = e-x J x
e-t et
--=--dt
t
0
334 PROBLEMS ON COMPLEX ANALYSIS
J~
JC
-
g
z
'I
FIG. 69
0 0 JC
The first integral on the right-hand side equals yn as"; let us write the
.2
l800Jld in the form
I
--
2
Jt+ig
co
I
- - das-cr+11>1 = -
1
2z
_ -12 J 00
as-(t+l1)
(t+ig)B
dt.
JC JC
1.3 ... (.2n-l) Jco as-(t+l1) dtl < 1.3 ... (2n-l)Jco eX"-tl dt.
1 2" JC
(t+ig) I 2" JC
It+'1/l18
ANSWERS AND SOLUTIONS 335
IJoo es-(t+l7)
----dt
(t+iy)llt
I< Joo y e-t _ y (n)
-de - - -
2y11t
0 0
r(3n+ ! )
1 co
951. /(t),.,., 1+-
n
2
(-1)"
0 t3n+-}
;
8 I I
e1 t e1 4t8
for small t,f(t) = r(:) - I'(4) +I' (~l) - ... 111:1 3yn.
952. 1. 957. 0.
953. o. 958. 1.
95'. 4. 959, n.
956. l; 3.
961. SOLUTION. Since the sequence offunctions /n(z) converges to the func-
tion ell everywhere except at the point z = 0, then for any small circle K
with centre at a point 111 0 and not containing the coordinate origin either
within it or on its boundary, for a given e > 0 it is possible to find an N
such that the inequality f/,.(111)-el/lll < e is satisfied for all the points of the
small circle K for n > N. It remains to choose B < min!ellf, where 0 is the
EC
circumference of the circle K, and apply Roucbe's theorem.
REMARK. The assertion of the problem follows directly from Hurwitz's
theorem (see, for example, (2, Chapter VIII, 2].
965. O. 966. 2; 1.
967. One root in each quadrant,
968. Two roots in ea.ch of the second and third quadrants.
970. In the domain fl> O, or:> +vfl (the domain I in Fig. 70) m = O;
in the domain fl> O, or:< + y{J (the domain II) m = 2; in the domain
fJ < 0 (the domain III) m =- I.
336 FROBLEMS ON COMPLEX ANALYSIS
971. In the domain ex > O, fJ > _!._ (the domain I in Fig. 71) m = O; in
ex
the domain where either ex ~ 0, or ex > 0, fl < _!._ (the domain II), m = 2.
ex
~
+-i/({J +!) (the domain II) m=2;
-v
m,=0; in the domain O<a:< 2
in the domain where either ex<! (P2+ !). or ! -y({J +4) ~ex< O, 1
ANSWERS AND SOLUTIONS 337
= y'(r*2 -l) = 06627 ... Thus, the required radius of convergence is not
less than 06627 ... At the same time the point w, at which dw/dz = 0, that is,
tan z = z -n/2, cannot be inside the circle of convergence of the expansion
of z(w). Putting z- ; = t we transform the equation dw/dz = 0 into the
form cot t = -t or e11r = (it+ l)/(it-1). Consequently, it = r* is a root of
the equation dw/dz = O. The corresponding value is given by w = t/cos t
2r*
= -2ir*/(e'+e-r), whenJe lwl = e' +e-r = y'(r*2 -l) = 0-6627 ... Thus,
this point is located on the circumference of the circle of convergence a.nd
the radius of convergence is consequently equal to y'(r*1 - l) = 0-6627 ..
1018. n 1 /(sin2 na).
1014 n 8 .cot na
smna 1017. 2~2 ( l+ m:
na).
1015. n 1 /8. 1018. n 8 /32.
1
1016. 2a2 (l+na coth na). 1019. n2 ~ ab
smna
CHAPTER VIII
.F-(z) =- (z) + -1 -)
"\"1 9k (z-ak Uk { -1 - ) is the principal part
+ u(z).
L.J z-ak
of the expansion of the function (z) in the neighbourhood of the pole ak;
g(z) is the principal part of the expansion of (z) in the neighbourhood of
the point at infinity, in which the constant term is included.
z-2
z+logz-3 i
1026 .F+(z) = z- 4 , .F-(z) = - z+l
"
= ao + ..!..2 (Ref9) + .i "\"1 (-b cosn8+~ sin n8);
11
4 2L.i
11=1
00
lOao. (1) If the point z lies in the circle Qk: lz-Tc:nl < ~, then l1(z)
2
339
340 PROBLEMS ON COMPLEX .ANALYSIS
Log : : : = log
of arg <C-z> along
I:::a. I
+iLlc arg CC-z), where Lie arg (C-z) is the increment
F+(C) = ~1ogJC+B1
2ni C-B
+.!.,
4
F-(C) = ~logJC+BJ
2m ~-B
_ _!_,
4
1
F(C)= 2mlog 1C+B1 1
C-B +7; 1
F(0)=2
F+(iB) = 3/4, F-(iB) = -1/4, F(iB) = 1/4; F'(O) = -i/nB.
F(O) = ~ l+(-l)n-l ; (5) the function F(z) is the same as in the pre-
2ni nRn
ceding section, only Log RR+z is the single-valued branch in the z-plane
-z
with a cut along the semicircle 0, with the same value Log 1 = 0.
The limiting value of this branch on 0 has imaginary part 3n/2 on the
left and ( - ; ) on the right. This defines the limiting values of F (C) on 0;
F (O) = _1_ 1+ (- l)n-1.
2ni nRn
REMARK. identical in
In parts (4) and (5) the branches of Log R+z are
R-z
the disk izl < R, generally they are different; thus, for example, at oo in
the first case the branch has the value -ni and in the second +ni. However,
F (oo) = O.
1035. Log b-z = log / b-z1 + iLlc arg (C- z) is the single-valued branch
a-z a-z
in the z-plane with a cut along 0, defined by the value Log I = 0 (the
branches are defined in the same way in problems 1036-1040).
b-z
1036. b-a+z Log a-z.
n
'\1 b-z bn-k+l-an-k+l
1037. L.J Ankzk-1 +zn Log a=z , Ank = n _ k -i- 1
k=l
oo ~n
1038. ~ b-z
cnAnkzk-1 + q, (z) Log - - .
a-z
n=lk=l
b-z
1039. -l - ( Log - b-z..)
- - Log - - .
z-z., a-z a-z0
n
1040.
~-Zo)
b-z
n [ Log -
l b-z + ~ Ak (z-z0 ) k-
- - Log--"
a-z a-z.,
1] ,
k=2
l [ l l ]
At= k-1 (b-z0 )k-1 - (a-z 0 )k-f ' F (zo) = -An+i
342 PROBLEMS ON OOMPLEX .ANALYSIS
= logC-I' F-(C) = O.
1 ~ 1 [ R+z
1043, F (z) = 2ni ,.:::- ";;i' Log R-z - 2 ,.:::- (2k- l) Rd-l
~ zlll:-1 ]
where m
n=l k=l
=[;], and Log [(R+z)/(R-z)] has the same value as in problem 1034,
(4). This determines the limiting values .ll':i:( C) on O. If lzl < R, then
00
~
1l' (z) -= ,.:CJ onzn, where On = 2 ni log C- l
If ~ .
C C-+1
n-o c
1044 F+ (z) = - :i [ 2+y'z log ~~~: ] ,
F- (z) .... - :, [ 2+y'z log :~:: ] - rz.
1045. F+ (z)seO, F-(z) =Log z-b (the branch of the logarithm is de-
z-a
termined by the conditions of the problem).
1046. Jl'+(z) ==
1, F- (z) == 1- y'[(z-a)/(z-b)].
1047. F+(z) = z-la- (1-A)b, .ll'-(z) == z-Aa- (l-A)b-(z-a)A(z-b)l-.1.,
1048. F+ (z) =Log z-Cbo, F- (z) =Log z-Co,
z- z-a
1050. -:i:l+z-35.
t If Log z and log z are the branches indicated in the conditions of the
problem, then Log z =log (-z)+m.
ANSWERS AND SOLUTIONS 343
~ 1 11 R+z
1058. If lI> 1 and zeO, then F' () =2-, log--Log-R +F'1
m z-1 -z
(),
where F'1 (z) is an analytic function for lI > 1, and the branch of Log BB+z
-z
is chosen just as in problem 1084, (4). From this the behaviour of F(z) at the
points B at the ends of the arc 0 is obvious.
1059. F'(z) = - 211 log log 1/r+2 log log l/B,
BF
Ba: = - 2 log log l/r
- 2 1 ~~
ogr
+m.oglogl/R, u.ll.. = -2'loglogl/r-2 1
ogr
+ 2'loglogl/R, ~
BF' elll<f> BF' 1 1 1
;;z= of>(Z) =logl/r' az=-2loglog-;--logr+2 loglog1f,
1068. D (/)is the area of the domain (}', onto which/ () maps the domain (Jo
I
Rl-C ; (2) log
1069. (1) log B(z-C) I 1-c1
-.C ; ,e.. -e"'I
(3) log e _ eC.
1
1976" "() = 2n
J "(C)
2K
.Bl-2Br
rl-B
~ (0- 4>)+rl dO'
0
2n
" (oo) = 2~
0
J "m d8.
1078. (1) /() - 4>()+'1' ('!'), / 1 () = -4> (':)- 'P(z), where
1085. u(z) = n
J u(t)
1
00
yc:U
(t-a:)+y , /() = ni
1 J u(t)c:U + iO.
00
~
-oo -oo
(for the existence of the first integral the piecewise continuity and bounded-
ness of u(t) in the whole interval (- oo, oo) is sufficient, for the second, in
344 :PROBLEMS ON COMPLEX ANALYSIS
addition, for example, the function u(t) must be of order It~ (a.> 0) at infi.
nity).
1086. /(z) == u(z)-f-it1(z)
=
1
2f J ~
n(t-z)
u(t) coth - -
2-
1
dt -2f
J
~
n(t-z)
U1(t) te.nh-2- dt,
-~ -~
where u 1(t) =- u(t+i) (for the existence of the integrals it is sufficient, for
example that u(t) and u 1(t) should decrease at infinity like lti~+ , a.> 0)
1088. Circles inside the disk lzl < 1, tangential to the circle lzl == 1 at
the point el9.
1090. Arcs of circles, connecting the points el9,, el91 inside the circle lzl < 1.
1 z-b 1
1091. <(z; a, b) = -arg - - , <(z; -oo, b) = - arg (z-b); <(z; a, oo)
n z-a n
= 1- 2-arg
n
(z-a). The geometrical meaning of these harmonic measures
is the angle divided by n subtended by the segment or ray at the point z.
1 1
1092. 1- - arg z for the ray arg z
y
= 0 and-
y
arg z for the ray arg z = 1"
_ 2 z-R 2 _1 2Ry
1093. <(z, LI) - -;:; arg z+R -1 - 1--;:; tan R- izl , <(z, I')
= 1-<(z, LI). The level lines are circular arcs connecting the points R,
from the points of which the diameter LI subtends an angle; (l+<) in the
2 z-R
1094. <(z, I') = -n arg - B
z+
2 yz-yB
1095. m(z, I') = -;:; arg yz+yR
loglzi-logr loglzi-logR
1096. for lzl = R and for izl = r.
log R-log r log r-log R
n ll09. -cos nl/>.
1099. u(z) == 2Cy<y(z). lllO. a sin 4>
Y-1 l-2a COS 4>+a1
nos. sin,,,,,..
llll. arg (1-ae14') = -sin-1 y(l
-
:asin 4>4>+a
COB
1)
m-1
Ill!. 21 !_ .J: otm
1 sin 2(m-k) 4>.
k=O
nl/> (n+ 1)4>
sm 2 sm 2
n1a.
ANSWERS AND SOLUTIONS 345
0 000000 0 000000
1 058779 1 080903
2 095106 2 095107
3 095106 3 030902
4 058779 4 -058779
5 -100001
0 000000 0 -100000
1 095104 1 -095107
2 058777 2 -080902
3 -058778 3 -058779
4 -095104 4 -0-30902
5 0.00000 5 0-00000
k h11; .Approx.
0 -100002
1 -058780
2 030902
3 095108
4 080903
5 0.00000
346 PROBLEMS ON COMPLEX ANALYSIS
k hk Approx. hk Actual
0 -084925 -085247
1 -078768 -078556
2 -064004 -064039
3 -044958 -044940
4 -023107 -023120
5 -000000 -000000
- 4 ( cos3tf> cos51/> )
1128. h(tf>) = --;;- COSl/>--3-1 -+-5-1--
k hk Approx. hk Actual
0 -11868 -11662
1 -11240 -1.1345
2 -10611 -10815
3 -08430 -08585
4 -06248 -05694
5 00000 00000
No.of
prob- h(x) f(z) f1(z)
lem
U.'1. If h (:i:) == 1 -
in the interval (-1, 1), then h (:i:) = - -1
n
:i:-1
log--
:i:+l
for l:i:I > l; if h(:i:) = 1-l:i:I, then h (:i;) = .!.. [:i: logw-l -log:i:-l]
n r :i:+l '
- 1 :i;I
if h (:i:) = sign :i:, then h (:i:) = -; log w-1
of the functions If> (z) and 'I' (z) near the points a and b.
1147. (The notation is the same as in problem 1146)
(1) l/l(z)-i Im lf>oo+2lf>-oo; (2) 'I' (i)+i Im '1'oo+2'1'-oo;
(5)
1
z::a- I
coth-
I z-a z-o
2 2-+ 2 coth-
2- ;
:ii
ll5S. - T+bi
I
llH. -:n1 oosAToosT,ifA > ;:ii 1 sinATsinT, if A<; - ~ cos 2AT,
if A= O.
ll55. =F i:n1 :~; (minus, if b > O, plus, if b < 0).
... + ~; 1 )] -1!
p=O
T-p-l Lpl 1)1 - i1 ~ (~~1~:.J.
k-0
T [ R+T ] R+T
1180, "j" log8 R-T-:n1 -2R log R-T
ll61. j[tog:~;-:n],
ll62. 2R [m- logR+T] + .!.[1ogR+T -:ii].
T R-T 2 R-T
R+T ] . R+T
ll63. 21 [ logR-:n-:n1 +2:n.Rt+TlogR-T'
00 (~)"(z+~)"
1176. /(z) = log -32 + .L.i
~
n
; the circle of convergence of this
n-1
1191. SOLUTION. The substitution e' =a: reduces the integral to the form
00
/(a) = J sin; da; (11:8 = eB 101 X), Integrating by parts, we obtain /(a)
1
00
1197. SoLUTION. Let us use the notation / 1 (z) = J e-tl/l(zt)dt. The ana-
o
lyticity of the function / 1 (z) in the circle lzl < 1 follows from the result of
problem 535 and the genera.I properties of Laplace integrals (see page 110).
Integrating by parts (n+l) times and using the inequalities of problem 535,
we obtain for lzl < 1
n oo
/i(Z) = - 2,1 zk [e-tlfi(lc)(zt)r +zn+ 1 f e-tlfi(n+ll(zt)dt
k=O O O
n oo
= J; anz"+zn+i J e-tlfi(n+l(zt)dt.
k-o o
From the estimate for llfi(n+l)I it follows that the second term on the right
hand side of the last equality tends to zero as n-+ oo (lzl < r). In order to
prove the second assertion let us take any point z e G. Then, as is not diffi-
cult to prove, inside and on the boundary of the circle on Oz as diameter
there a.re no singularities of the function /{z). Hence for a sufficiently small
d > 0 within and on the boundary 0 of a circle of radius (lzlf2)+d, concen-
tric with the one already constructed, the function /(z) is also analytic. Thus
3-19
350 PROBLEMS ON COMPLEX ANALYSIS
the equality o,, = 2~ j ~:~ dC holds for the ooemcients On of the expansion
00
co
Since the series ~ znt",!(C) converges uniformly on 0, it follows that lfl(d)
..I n!1o+i
n-o
=~ J!(C)entcdf. The maximum of the quantity Re (z/C) on Oie equal to
me ..
jzi'~,,,.- q < 1 (for the proof of this assertion it is sumoient to consider the
case when z is real and positive as a rotation round the coordinate origin
does not change the quantity Re (z/C)) and consequently, ll/l(d)I < .Aett (..4.
00
is a constant). It follows from this that the integral J e-tl/l(zt)dt converges.
0
1201. If Log<1 l z-za is the branch of the function Log z-zi, which isana
z-z1 z-z1
lytio in the Zpla.ne with a out along the arc y 1 and which tends to 0 at
111 ... oo, then .F-(z) - (a-b) Log<1 lz-za, F+(z)
111-1111
= (a-b) Log<ll z-zi +2nbi;
z-z1
within G+: F-(z) = (a-b) Log<1l z-z. -2m (a-b) for analytic continuation
z-z1
through y 1 and F-(111) = (a-b) Log<1l z-za for analytic continuation through
z-z1
Ya
1208. (1) z = 3; (2) z = 2i; (3) z = 2i. In all three oases the va.luee
of w' (z) are different.
1205. Above z = 1 there are two elements: z = I+t,
1-(1-t)i 1
wi ----t-- = 2 +se+ ... , ltl < 1,
and
Wz=
I+(I-t)i = .!_ _ _!_ _ _!_t+ 0 < ltl < 1;
t t 2 8 ....
algebraic element: z = o+t, to= ~ (1- ~ t'+ ...). ltl < 2, and two reg
there is one algebraic element: z = t-', to= +-t+ ... , 0 < ltl < ~(1/5),
1207. Above z == 1 there are two algebraic elements: z = l+t,
is one less than the order of the pole. If at co the function w(z) = w0 + ;;k + ...
(la> 1), then z = oo corresponds to an algebraic branch point of order (la - 1)
abovew = w0
II
1220. The surface for z(w) is the same as for yw; its branch points are
situated above w = O, and w = oo and correspond to z = -n, z = oo. To
sheets of the w-plane with the cuts 0 < u < oo, ti = 0 there correspond angles
2n/n with vertices at the point z = -n. As n .,.. oo these angles are trans-
formed into horizontal strips of width 2n, the function w(z) into eS, and the
surface for z(w) into the surface for Logw.
II
1221. The surface for z(w) is the same as for yw. To the sheets of the
w-plane with the cuts - oo < u < 0, ti = 0 there correspond figures formed
from two circular segments with angles 2n/n at the points z = a, z = b
(Fig. 73, where n = 3)t.
1222. The surface for z(w) is obtained by joining together two sheets of
the w-plane with the cuts lul < 1, ti =- 0, corresponding to the domains lzl < 1
and lzl > 1. The branch points are situated above w = 1 and correspond
to z = I. To the polar net lzl = r, arg z -= 4' there correspond the ellipses
and hyperbolas with foci 1 :
:ii y
-----=l
cos 4' sin1 4'
FIG. 74
354 PROBLEMS ON COMPLEX ANALYSIS
1
points
Wk=
(2n-1)
2n Zkcorrespondingtothe
points
Zk = ny(l/(2n-ll) rut. The
basic mapping is shown in Fig. 75. The rest is obtained by continuation by
the symmetry principle.
FIG. 75
1226. The surface for z(w) has n sheets with a branch point of the (n-1)-th
order above w = oo, corresponding to z = oo, and n - 1 branch points of the
first order, situated above Wk = Zk ( 1- !) and corresponding to Zk = wit
1nl
(w = en- 1 , k = 0, I, ... , n-2). In order to construct the surface it is necessary
to attach to its zero sheet (the w-plane with n - I radial cuts originating
from the WJt) along each cut criss cross one sheet (the w-plane with one radial
cut). '!'he basic mapping is shown in Fig. 76. To the disk lzl < I there
corresponds the interior of an epicycloid (for n = 2, a ce.rdiod).
FIG. 76
1227. The surface for z(w) has (n+l) sheets with a branch point of the
(n - l)th order above w = oo, corresponding to z = oo, and n+ I branch
ANSWERS AND SOLUTIONS 355
points of the first order situated above w1c = i1c ( 1 + ~) and corresponding
to z1c = ~ (ro = e n+i, Tc = O, I, ... , n). The basic mapping is shown in Fig. 77
The domain lzl < 1 corresponds to the exterior of a hypocycloid (for n = I
a segment).
(I) Zo (2) !J
FIG. 77
1228. The surface for z(w) has n-sheets with a branch point of the (n - l)th
order above w = oo corresponding to z = oo, and n - I branch points of the
Ten
(-I)k
first order situated above Wk = ~ and corresponding to Zk = COS n
(Tc= 1, 2, ... , n - 1). To construct the surface it is necessary to attach to
its zero sheet (the w-plane with the cut - oo 1 , t1 = 0) sequen
< u::;;;; - -2n-1
tfall.y n-2 sheets with the two cuts 2 L 1 ::;;;; lul < oo, t1 = 0, then to the last
ot them to attach another sheet having the cut - oo < u::;;;; - 2.~ 1 , t1 = O,
1 ::;;;; u < oo, t1 = 0 if n is odd. In the
if n is an even number, and the cut -2n-t
mapping w(z) ellipses and hyperbolas with foci 1 pass into ellipses and
hyperbolas with foci 2.~ 1 The change of the semi-axes is obtained from the
relations z = const, w = const. Fig. 78 shows the dissection of the z-plane into
domains corresponding to the half-planes t1 > 0 and t1 < 0 (the former are
shaded) for n = 5.
356 PROBLEMS ON COMPLEX .ANALYSIS
FIG. 78
U' IJ'
FIG. 80
358 PROBLEMS ON COMPLEX ANALYSIS
1235. z = ta.n-110 = 21. Log ~+-w The surface for z(w) has an infinite
' 1D
number of sheets and two logarithmic branch points above w = i; it is
obtained by joining together an infinite number of 10-pla.nes with the cut
u = 0, lt1I.;;;;; 1, corresponding to the vertica.Istrips kn< a: < (k+l)n (Fig. 81).
R' D'
FIG. 81
1238. z = cot-lw = n/2 - tan-110. The mrfa.ce for z(w) is the same aa
for Ta.n-1w (Fig. 82).
D' D'
FIG. 82
ANSWERS AND SOLUTIONS 359
+
1237 z - cosh-lw = L og (w l(w8-l))
f ' coehz =
COHZ. The surface for
z(w) is the ;rune as that for cortw (Fig. 83).
FIG. 83
.l(w+l)) sinh z = - t sm
u.
The surface
for 1238. z = sinh-1 w =Log (w+rding one 'by a rotation through n/2 about
z(w) is obtained ~rom t?e prece
the coordinate origm (Fig. 84).
Flo. 84
360I l'ROBLEMS ON COMPLEX AN
Fxo. 86
ANSWERS AND SOLUTIONS 361
1241. The surface for z(w) is constructed as follows: we make on thew-plane
the horizontal cuts -'.Xl<u.;;;;;1, v=(2k+l)n, (k=0,1, 2, ... )
md along each of these we attach one sample of the w-plane with the same
(single) cut. The surface constructed for z(w) is based on the fact that w(z)
maps ea.ch strip 2kn < y < (2k+l)n onto the strip 2kn < v < (2k+2)n
carrying the Wplane joined along the cut -1.;;;;; u < oo, tJ = (2k+ l)n (see
Fig. 87; the sign + denotes that the domains have to be joined together).
JV
(8) A{-/+!!/)
~~~+~~- V=O
-------
FIG. 87
12,2. (1) Let F' be the Riemann surface onto which the function w = R(C)
maps the C-plane. In order to construct the Riemann surface of the function
z(w) it is necessary to join together infinitely many specimens of the surface
F' with a cut along the arc connecting on F' the points w(O) and w(oo) (similar
to the construction of the Riemann surface of the function Log w). The re-
sulting Riemann surface has two logarithmic branch points at the ends of
the arc of attachment and an infinite number of algebraic branch points be-
longing to the surfaces F'.
(2) To construct the Riemann surface of the function z(w) we join together
an infinite number of specimens of the surface F' alternately along the cuts
going from the points w(l) to the point w(oo) (similar to the construction
of the Riemann surface of the function sin-1 w). The resulting Riemann surface
has two logarithmic branch points above w(oo) and in addition to the algebraic
branch points of part (1) it has also an infinite number of algebraic branch
pointsofthefirstorderatthepoints w(l) (ifw(+l)orw(-l)isan algebraic
branch point of order k of the surface F, then for z(w) it will be an algebraic
branch point of order 2k+ 1). The investigation of the function z(w) can also
be reduced to the preceding case by the substitution z1 = iz.
1243. All the Riemann surfaces of w(z) have two sheets and have algebraic
branch points of the first order above the points: (1) z =a, z = b; (2) z =a,
z = b, z = c, z = oo; (3) z = ak and z = oo, if n is odd. For the construction
of the surfaces we take two sheets of the z-plane with cuts going from the
points indicated above to oo, and join them along identical cuts.
362 PROBLEMS ON COMPLEX ANALYSIS
1244. All the Riemann surfaces for w(z) have three sheets and have algebraic
branch points of the second order above the points:
(1) z =a, z = oo; (2) z =a, z = b, z = oo; (3) z =a, z = b, z = e; (4) z = ak
and z = oo, if n is not a multiple of 3. For the construction of the surfaces
we take three sheets of the z-plane with cuts going from the given points
anl
to oo, on which we define the three single-valued branches w, (J)'W, w'w (w = e 8 ).
On going round the branch points w(z) acquires a factor won account of one
of the factors of the root, hence the order of the attachment of the cuts is
the same, cyclic, along all the cuts (see the sketch in Fig. 88).
FIG. 88
1245. The Riemann surface for w(z) has n sheets with algebraic branch
points of the (n - l)th order above z =a, z = b, z = e, z = oo. The surface
is obtained by attaching n sheets of the z-plane with cuts going from the
points z = a, z = b, z = e to oo. The attachment is cyclic, simultaneous on
all three cuts. The sheets correspond to the single-valued branches of the
function wkw (w = e2 711/n, k = 0, 1, ... , n-1).
1246. The Riemann surface of w(z) has six sheets with an algebraic branch
point of the fifth order above z = oo, two algebraic branch points of the second
order above each of the points z = a, z = b and three algebraic branch points
of the first order above z = e. The surface is obtained by joining together
six sheets of the z-plane with a cut along a curve going from a to b, from b to
c and from e to oo. These sheets correspond to the single-valued branches:
W1 +w1, (J)'Wl +w2, ro9w1 +w1 , W1 -w2, WW1 -w2 , ro9w1 -w2, where w = etni/a
3
and w1, w1 are the single-valued branches y[(z-a)/(z-b)2] and y(z-c).
The detour round a cyclically connects sheets 1, 2, 3 and 4, 5, 6, that round
b sheets 1, 3, 2 and 4, 6, 5 (connected twice in half-sheets), that round o the
half-sheets 1, .2, 4, 5 (shown in Fig. 89); 2, 3, 5, 6; 3, 1, 6, 4; that round oo cy-
clically the sheets 1, 6, .2; 4, 3, 5 (see the diagrams in Fig. 89).
1247. The Riemann surface for w(z) has she sheets with two algebraic
branch points of the second order above z = 0, one algebraic branch point
of the first order above z = 1 and an algebraic branch point of the fifth order
above z = oo. In order to construct it two samples of the surface of the function
8
yz, have to be joined together, each of them having on one of the sheets a out
along the ray y = 0, 1 =:;;;; :c < oo.
1248. The Riemann surface of w(z) has two sheets with algebraic branch
points of the first order above the points z = kn (k = 0, 1, 2, ... ) ; above
z = oo the surface has a transcendental singularity the limit of algebraic branch
points. In order to construct the surface take two sheets of the z-plane with
cuts going from the algebraic branch points to oo (for example, along rays
parallel to the imaginary axis) and join the sheets together along identical cuts.
ANSWERS AND SOLUTIONS 363
1) c
w1+Wz
:
w1+w1 @\ w1+w1
4) c
w,-w,
:
w,-w,
@ w,-w,
w
t.JW1-Wz tiJW,-Wz cuw,-w,
5) c :
a (J)ZWt -W3 b W,-Wz w,+Wz
FIG. 89
364 PROBLEl\IS ON OOMPLEX ANALYSIS
1249. t F':z and F'w are each obtained by joining together two planes with
cuts [ -1, l] (Fig. 90).
FIG. 90
1250. F's is obtained by joining together two z-planes with the cuts (-oc
< z :s;;;; O, y = 0), F'w two w-planes with the outs [ - ! , !] (Fig. 91).
~
~
1
0 I
FIG. 91
1251. F' and F'w are each obtained by joining together three planes having
a a a
in cyclic order outs along two of the segments [O, y4], [O, roy4], [O, ro8y'4J,
where <.o = e11tl/1 (Fig. 92).
FIG. 92
FIG. 93
l!li3. Fz is obtained by joining together two Zplanes with cuts [-1, I],
Fw by joining together three wplanes having respectively one cut
[-ib, ib], two cuts [-a, a], [O, ib] and two cuts [-a, a], [O, -ib], where
y'5-l J1'5+1
a= - 2-y'(y'5-2), b = - 2-y(Jl'5+2). The mapping and the three
Wplanes with cuts are shown in Fig. 94 (the :figures are not to scale).
366 PROBLEMS ON COMl'LEX ANALYSIS
FIG. 94
1254. (a) n odd. F 11 has nsheets and has n alirebraic branch points of the
(n-l)th order above the points 11 = wk(w = e"Ttn, k = 1, 2, ... , n), Fw has
Jin sheets and n 1 algebraic branch points of the first order, n over each of
the points w = r{'y(l/4) (17 = ealfn, v = 0, 1, 2, ... , n-1), corresponding to
the points of F11 above z = 17"', and n algebraic branch points of the first order
above w = oo, corresponding to the algebraic branch points of F 11 above the
1 = wt. The lD6pping is shown in Fig. 95 (for the investigation use is made
A(cfl)
0 B(I) 0~,;,.;:,.-c~===:~
c
FIG. 95
or the substitutions c= zn, w = w"). The function w(z) maps the circle 1111
n
< 1 onto thew-plane with n radial cuts from the points 17"'y (1/4); this is a well
known extremal function from the theory of single-valued conformal map
pingst.
the limit of its simple poles at the pointszrc = u~::r. a1; = ; +nk (k - o.
1, ... ).By problem 638, the domain of indeterminateness at the point z = I
is identical with the whole plane.
1260. If n = 0, then z = 0 and z = oo are removable singularities and
w == 1; if n < 0, then z = 0 and z = oo are logarithmic branch points where
lim w = lim w = 1 and z = 1 is an essential singularity for one of the branches
,, .... o z-oo
of the function; if n = 1, then w = z; ifn > 1, then z = 0 and z = oo a.re
logarithmic branch points, the domain of indeterminateness of w (z) at these
points coinciding with the whole extended w-plane.
1261 and 1262. z = 0 and z = oo are logarithmic branch points with the
domain of indeterminateness for w(z) coinciding with the extended w-plane.
1263. z = 1 and z = oo are logarithmic branch points where lim w
,, .... 1
=Z-+00
lim w = oo and z ... 0 is a simple pole for all the branches of w(z) except one.
1264. z = 0 and z = oo are logarithmic branch points and lim w
z....O
... umw = oo.
Z-+00
1265. The branch points a.re the same as for sin-1z (that is, an infinite set
of algebraic branch points of the first order above z = 1 and two loga-
rithmic branch points above z = oo, lim w = O; z = O is a pole of the first
Z-+00
order for all the branches of the function except one).
1266. z = i are logarithmic branch points, lim w = oo; z = 0 is a pole
z .... :I: I
of the second order for all the branches of the function.
1267. The surfaces w(z) and z(w) are the same as for the logarithmic func-
tion (logarithmic branch points above the points 0 and oo ). The mapping
is easily obtained by means of the parametric representation z = eC. w = ~.
368 PROBLEMS ON COIVIPLEX ANALYSIS
1268. The Riemann surface for w(z) has an infinite number of sheets with
one logarithmic branch point above each of the points z = a, z = b and two
logarithmic bmnch points above z = oo. The surface is obtained by joining
together an infinite number of sheets of the z.plane with cuts going from the
points z = a, z = b to co. These sheets correspond to the single-valued branches
of the function w+2:iin(n = 0, 1, 2, ... ). On going round the points
z = a and z = b these branches pass successively into one another and this
determines the character of the attachment of the sheets.
1269. The Riemann surface for w(z) has an infinite number of sheets with
one logarithmic branch point above each of the points z = a, z = b, z = c
and three logarithmic branch points above z = oo. The construction is the
same as the preceding.
1270. The Riemann surface for w(z) has an infinite number of sheets with
one logarithmic branch point above each of the points z = kn (k = 0, I,
2, ... ). As sheets, z-planes with cuts going from the points z = k"l to oo
(for example, along vertical straight lines) can be taken. Two sheets are joined
together simultaneously along all the cuts, on one side, as in the construction
of the surface of the logarithmic function. At oo there is a transcendental
singularity, the limit point of the logarithmic branch points.
1271. (1) In every connected part of the Riemann surface of the function
!; = <t>(z) above the z-plane to which corresponds a connected part of the
Riemann surface of the inverse function z = 4>-1 (/;) above Gt, w(z) repre-
sents a single analytic function; (2) In every connected part of the Riemann
surface of the function !; = lf>(z), located above Gz: (this is the domain Gt,
transferred to the z-plane), w(z) represents a single analytic function; (3)
The same as in part (2). The particular case of w(z) indicated in the condi-
tions of the problem always represents a single analytic function.
1272. w(z) consists of the two analytic functions z.
1273. w(z) consists of the two analytic functions zt/s.
(w = e21ti/P, m n1
p = H.O.F of (m,n); k = 0, 1, ... ,p-1; m 1 = p' = -p
n )
1111
FIG. 96
@ (4aew) :
Q ~ Q' ~I
Q
-
-~
(2)~g:
(1) D'
FIG. 97
1291. An infinitely many valued function with the same Riemann surface
as Log Log z.
1292. The Riemann surface has infinitely many sheets and has an a.lgebraic
branch point of the (n-l)th order above z = o. only algebraic branch points
of the first order above z = 1 and 2n logarithmic branch points above z = oo.
In order to construct the surface it is necessary to join n surfaces for sin-1 111
with the cut [0,1] on one of the sheets of each of them. The curves n8
= Tc:iifn and en cos n8 = Ten (Tc = 0, 1, ... ) divide the w-plane into do-
mains corresponding to the half-planes y ~ 0 (Fig. 98, where n = 2, C= w'
is the auxiliary plane).
ANSWERS AND SOLUTIONS 371
FIG. 98
(C-sln w) 1
I
ltll
FIG. 99
1324. (1) At the vertex ..4. 1, 111: = O; (2) At the vertices ..4., and ..4. 1, 111:1 = lits
= O; (3) At the vertex ..4.8 , 111:8 = O; (4) At the vertices ..4. 1 and ..4.4, 111:1 = 4
= -1; (5) At the vertex ..4. 1, 111:1 = -2, at the vertex ..4.4 , 4 = O; (6) Atthe
vertices ..4. 1, A8 and ..4. 4 , 111:1 = 111:8 =4 = O; (7) At the vertex A 1, 111:1 = -2,
at the vertices ..4. 4 and ..4. 8 , 4 = 111:8 = O; (8) At the vertex A 1, 111:a = -2, at
the vertex A,, 4 ... 111:-2.
1826. It is necessary and sufficient that a1: = 2..
n1:
(n1: is a natural number
"i na na p
1 00 00 a strip
2 2 00 a half-strip
2 3 6 a right-angled triangle
2 4 4 an isosceles right-angled triangle
3 3 3 an equilateral triangle
t Diagrams of the fundamental domains are given in Fig. 62, page 292.
374
.ANSWERS AND SOLUTIONS 375
ita fundamental domain B conaiata of the doubled triangle and two different
boundary aides;
z
(2) w = 0 Jz-f(l-z)-l dz, where 0 = ro/[B(l/4, 1/2)]; z(w) iaa doubly
0
periodic function with periods 2w and 2wi; the group G ia generated by the
transformations: !Z'(w) = w+2w, S(w) == iw; ita fundamental domain B con
aiata of a square with aide w and two boundary aides of one of the triangles
forming the square;
z
(3) w = 0 J:1rl(l-z)-fdz where 0 = ro/[B(l/3, 1/3)); z(w) is a doubly
0
periodic function with periods 2hl and 2hed/e, where h = wy'3/2; the group
Gia generated by the transformations: !l'(w) = w+2hi, S(w) = fQ8l'Cf/1; its
fundament.al domain B conaiata of the doubled triangle and two dift'erent
boundary sides.
1330. (1) The "triangle" with two vertices at the points w = 0, w =cl
= I'(rr.)I'({J)/I'(rr.+/J) and angles nrr., nfJ at these vertices. If rr.+fJ < l then
the three vertices are finite; if rr.+fJ ';?!: l, the three vertices lie at infinity; if
rr.+fJ = l, then cl = n/ain rr.n, and the "triangle" haa the shape of a half-strip,
oblique, if rr.:f./J; in the case rr.+/J = 2 the aides of the "triangle" which issue
from the vertices of the base are parallel, directed in opposite directions and
h == s~nfJ r~:~:; l) from the origin. In the case rr. - 1 a strip of width n
is obtained; in the case rr.+/J = l two aides are parallel and h = n; in the
case rr. == 2 a half-plane is obtained with a cut along the real positive semi-axis
(4) w = -2h
n
u u
[tan-1 t' a+tanh-1 t' z-2 z];v
(5) w = ia ( -yz -z-3 )
2 --1
ANSWERS AND SOLUTIONS 377
z q-1
1333. (1) w = - -naJ zi- 11 (dz
z-1 )II If 8 h
= -,
q
then w = -na~l
-p
tv
X
l v-o
1 llYftl
1337. (l)w=
,..
! {Htan-1 y(a~z)+htanh-1 ! v(a~z)}' whmi
a= l+ HI;
(") - _!
"' w - n
{H tan -1 z
y(a1 -z8) where
A
ol= l+ HI;
z
ytdt
(3) w = 0 J (t-l)(t-a)(t+b), whmi 0, a, b are determined from
0
the equations:
On OnJfa OnJfb
(a-l)(a+b) =hi, (a-l)(a+b) =ha, (b+l)(a+b) =A,.
1338. (1) w = ( 1
"'
2 ) J (l-tn) ndt;
Z I
--
B -, 1-- o
n n
l .!.
(2) w =0 J (l-tn) n
- 11- dt+ 1, C-1 = 0 = (l
2-1/"nsin n/n
l l )
z B -;-+22
378 PROBLEMS ON COMPLEX ANALYSIS
1389. (1) w =
5 21/I
. f (1-t1)-&/1 (l+t1)1/1dt;
Z
BUo!)o
(2)
_J
w - 0
1
(1-t1)&/1(l+t')-l/1
t
_ 2-l/II'(7/10)
dt+l, C-1 - 0 = I'(g/lO)I'(4 / 5)
z
1840, Onto a many pointed star with angles :n;- 2" -i.n and :n:+i.:n: alter.
n
ilately, with centre at the coordinate origin and one of the vertices of the
first type of angle at the point
1341. w = 000 f
,
z
{z n
k-1
m-1
(z-tan1 :n-1/dz, if n = 2m, and
"' = 001 J
,
z
{z n( m
k-1 \
z1- tan1 :k)}-1/n dz, if n - 2m+ 1, where
1345. The parameters a!.'e determined from the values of lf(bi)I, lf(ota)I,
known from P and one of the directions of the sides of P being given. Three
of the parameters (ak, b,, "J d8 ) are chosen arbitrarily. If one of the parame-
ters t or "J is equal to infinity, then (4) and (5) apply if the factor and term
with this parameter are neglected. If one of the parameters b1 or d8 is equal
to oo then (4) and (5) remain in force without alteration.
1847. w = O(z+l)1(z-1)1, 0 = h(b+l)-1(1-b)--., b = 1-1 ,
1+1
1348. w = ay'{;~:~.). o = J"(hH), a= J"(h/H).
b = v(2hih:hs):
(3) w = -h11og--- - og -z-aa
z+a1 +h11 - - . The parameters a 1, a 1 are
n l+a1z n 1-a1z
determined from the system of equations: a/11a.'2 = e-,.,
z+I 2 . .
(7) w =log I-z +Az,A- b'-I' where b is determined from the
. b+I 2b
equation log b-l + b'-I =cl;
ned from the equation log a+ ~(a-~) +cl= O. In particular, if cl= 0, then
a- I;
(9) w = h/n log (z-a)+O/(z-a)+Az+const, where A = h/2na,
0 = 27:m (l-a 8) and a is determined from the equations
C' Ill
---- D~C
A~1 8
iK'
@
I
I C' '
B' JV D c" B'
/(
-f
-if('
FIG. 102
z
1158. 10 = .ii J
0
dz
y[(I-z1 )(1-klzl)]'
z = sn ( ': , k) , where .ii and k are
"'
determined from the relations K'/K = b/a, a= .ilK.
ANSWERS AND SOLUTIONS 381
1859. See Fig. 103. The inverse function is z = en (w, k).
JJ" D'
@)
Ill
k'
I If C' D
IY
A' D" A' D'
21< ])"' -1
D Ill
]) c D" IV
-ti<'
FIG. 103
2il<'~----
@)
IV
-tl<''----'----1
-2tK'1---...i..---1
FIG. 104
1362. The mapping of the z1 -plane onto the Ui-plane is shown in Fig. 105
(the w 1 -plane on this same diagram refers to the answer to problem 1868).
1368. The relation between the z1 -plane and the w1 -plane is shown in Fig.
105. The expressions for l and h are indicated in Table 5 on page 383.
1864. SOLUTION. After applying formulae (12), (13) oo the expression w(z)
given in the hint oo the problem we obtain:
a~r
l
FIG. 105
-:Jr -b -1 0 I b if'
FIG. 106
.ANSWERS AND SOLUTIONS 383
TABLE D
I -~
---~ ____l _____________ h
k E K'-E'
1
7C
~
k
(E-k' 1K) I ~
k
(E'-k 1K')
v
ik 1
k' E !, (K'-E')
FIG. 107
t For details of the solution and graphs for the determination of the
parameters see: A. BETZ, KonfONne Abbildu11g, Berlin, 1948.
384 PROBLEMS ON COMPLEX ANALYSIS
l 1-(E-k'1 K)
= -k'I h 1-(E'-klK')
= -k'I
'
and for 11 = -k1 :
_ ...!._ [
w - k' E(u)
-k1 sn u u] -_...!._k' E ( k , u, ikk' ) ' z -_ ...!._
en
dn u k' (
K' -E' H _ _!_JC
), - k'
1867. SOLUTION. It follows from the condition .1 > 0 that e1, e1, Sa are
real and different and g8 > 0. Let us consider that e1 > e1 > e1 The upper
half-plane Im z > 0 is mapped onto the rectangle with vertices 0, co,
co- co', - w' (it is assumed that Im co' > 0), corresponding to the points
co
oo, e1 , e1 , Sa The median line of the rectangle corresponds to two semicircles
(Fig. 111) : the first with centre at the point ea, with respect to which e1 and
e1 are symmetrical (consequently, its radius is y[(e1 - e1) (e1 - e1)], and
the second with centre at the point e1 , with respect to which e8, e1 are sym-
metrical (its radius is y[(e1 - e8) (e1 - e8)]. Continuing the mapping of p (w)
by the symmetry principle we arrive at the halfperiods co, w' of this function:
00
w=J~~~~~-dx~~~~~
y'[(x-e1) (x-e2) (x-ea)]
ANSWERS AND SOLUTIONS 385
@) IY
II
f
JI
Ill
IY 18
/Y
m
II
2)
Fm. 108
JV t /JI
z)
FIG. 109
'11~-1;
lJ c
@
A 8
0 I<
FIG. 110
TABLE 6
1' I H H I l I h
l
I
V< -1 I n-,/ Iv! 1
I k'i:+ v nc'v. k '>
1- - n 1 (v,k) I H-H
2 lv+Il lv+k'I
I i ~
n -,/ jvj I ! k' II (I I i H=H t::I
v>O I I II1 (v, k) i k+v i v k)
2 fv+ll lv+k'j I I Cll
' 0
t"
I
q
( ,. 1-3
V= -1 I - - I F1 E-k K) ;,, (E' -k 2 K') ....
i 0
z
Cll
1 ' I (K' E1
v = -k 2 ! - i -k'' E k' - )
!
REMARK. In the column "l" the integrals are underst.ood in the sense of the principal value.
The quantity v' v -k'.
= -- __
k'+v ~
00
~
388 PROBLEMS ON COMPLEX ANALYSIS
p(w) = ea + e,. - 6a , k = ~ .
sn8 ( row,
K
k
) e1 - 6a
If g1 > 0, then ea < e1 < 0 < e1 and co< lw'I (for k < k' and consequently
K < K'); if Ua < 0, then ea < 0 < e. < e,., and consequently O> > IW'I
FIG. 111
2(J)
----,
! @)
I
IV I
i
i
FIG. 112
//"
F10. 113
390 PROBLEMS ON OOMl'LEX ANALYSIS
A
Az
"FIG. 114
arg dw = - ( n) 1
n+ aa+2 -2(-111:1+ix1+a) n
= -2 +
and so on. For the determination of the complex conjugates of the half.
c
e,
FIG. 115
A,
FIG. 116
A,
FIG. 117
ANSWERS AND SOLUTIONS 393
K'
1170. (1) z = sn (u, la), w = fJffu/K', log= 2n K; !Reul< K, llmul < K';
(2) the mapping function is the same as in part (1) except that
0 < Reu < K, llmul < K', log= nK'/K;
(3) reduces to part (1) by means of a linear transformation; then we
have la= ~.il~~~!=~~ where .il = (a, b, c, d) is the cross-ratio of the given
point.a;
(4) reduces to part (3) by means of the mapping I = y'(z1 +h1 );
nK'
(5) z = sn(u,la), w = i,elnu/2K, log="'! K ; - 3K <Re u < K,
0 < Imu < K';
(6) reduces to part (5) by means of the mapping I= y'(l+z); here
la= 008 Ot;
(7) the mapping function is the same as in part (5) except that in
nK'
onecaseO < lmu < !K',andintheother !K' <Imu < K'; log=-; K;
-ir K' 1
lo isdeterminedfromtheequationsq ... e K = (a+b)I' whereb = y'(a'-1).
I - dnu 2K .
1373. t = k , where u = - sm-J z, IRe u) < K, llm ul < K'.
snu :n;
The parameter lo has the same value as in the answer to problem 1371,
The foci z == l pass into the points k/(I+k').
dnucnu] I+cnu .
137,. (1) z = 0 [ Z(u) + , t = - - - ; k IS determined as
snu snu
in problem 1370, (Iii). See also problems 1385 and 1371;
+on(u,snk)(u,k)
dn (u, k)] t 1 + on u h
= sn u w ere
lo
IS
determined f
rom
th t'
e aqua ion
J!J' - k 1K'
J!J _ Tt'K =
b
a(see problem 1383 and Table 6).
CHAPTER XI
'\\
\
\
\\
FIG. 118
118); log IJ = - ~ 8+0 are equipotential lines; loge= ~ 8+0 are stream
z-a
lines ( --b- = eel ) .
The velocity is V = - 2- . _
r-Q ;_;;
z- - :n:i (z- a) (z-b)
1380. At the point z = 0 there is a doublet with moment p = 2n;
r = 0 cos If> are equipotential lines; r = 0 sin </> are stream lines. V = -elf</> /r1
Vco = 0. At the points 2 i the velocity equals - (3 4i) /25.
1881. (1) and (2). At the points 0 and oo there are doublets with moments
2:n:R8 and 2:n: (the upper sign refers to part (1). the lower to part (2)). The
field lines are cubic curves: ~ R1~/(a:&+y1 ) = 0 are equipotential line11.
395
v~
N COMFLEX ANALYSIS
PROBLllMS O 1 The -
96 Jl' ,,. - .
3 =F Ry/(zl+y) = O are stream rm.._
_ Ri part1 'F7
forv- '.. Pomts 1- F>g.
(2 ) are cnt1cal
z = R for part (1) and z -
Y
V la = ..;
a
the coordinate origin is a critical point (Fig. 122).
o a
FIG. 122
FIG. 123
1384. At the points a, ai there are sources: (a; 2n), ( ai; -2n);
re+oatr2 cos 2e+a' = O (le!> 2) are the equipotential lines (the straight
lines y = x also belong to them); re+oa2r2sin 2<1>-a' = 0 are the stream
4.az 2
lines (the coordinate axes also belong to them). V = ~ V00 = 0, the
..--a
point z = 0 is a critical point (Fig. 123).
1385, There are sources at the points 1, 0 and at oo; ( l; Q), (C,;-Q)
X ANALYSIS
OBLEMS ON COMPLE tential lines (for
PB the eqwpo
398
(
-Q); ..+7-
1 - o+2cos 21/> (0 > 0) are. clear= '0
....... to ... = I'
"" ..............
and r = l/y'O);
' alues or o the curves elude
lines (they m
Jarge.. vl( o+tan<J>) ... ... ....... The ...... l
r =- JI 0-tan</> Q [ z1+1 ]. Voo == o.
d the circle
. r = 1). V = 2n
=-~~~'~1~%,~
z(zl-1)
'w$%~--.
~.~~.'l.h, 0,
ANSWERS AND SOLUTIONS 399
1887. At the points (1 i)/y2 there are sources of strength 2n. at the
point z = 0 a source of strength - 4n, at oo a source of strength - 4n; r'+ ~
= a - 2 cos~ (0 > 0) are equipotential lines (for a < 4 the curves sepa
rate into four components, for 0 > 4 into two; for large values of a they
t a
are "almost" the circles r = yO and r = l/yO); tan24> = (Or'-1)/(r'+l)
are stream lines (they also include the coordinate axes, the bisectors of the
1888. At the point z = 0 there is the source (O; Q), at the point z = oo a dou-
~(C-11.1r)
blet and the source (oo; -Q); y 1 =ea -a; ( Q
cos-1 4> + 2n log r = a)
are the equipotential lines; r= { 0 - ! 4') /a sin 4' are the stream lines; the
o-!L
stream rmes have thehonzont
. al asymptotes: YJt:-+OO-. -;
a 2
YJt:-+-oo-.--a-'
V =a+ Qel4> V: Q ' " al ' (F'ig. 127)
2nr, oo =a; z = - 2na lB a critic pomt
1889. At the point z = 0 there is the vortex (0; I'), at oo the doublet
and vortex (oo; -I'); r == (O-I'4>/2n)/acos4' are equipotential lines;
400 PROBLEMS ON COMPLEX ANALYSIS
~=er
~(ay-C) ( I'logr )
-y sin-l<fl- - - - = 0 are stream lines; V
r
=a+ --X
2n 2nr
1( ")
Xe <fl+ I , V00 = a; the critical point is z = 2I'"
,;, (Fig. 128).
FIG. 128
1390, The fiuid stream lines the circle of radius B; Voe =a, the circulation
is I'; the critical points are determined by the equation Zkr = 4 (11!a
y'(16nla Bl-I'9)). If I'< 4-naR, then lztrl = B, that is, both stagnation
1
points lie on the circle lzl = B; if I' = 4naR, then the stagnation points merge
into one; if I'> 4nq.R, then IZtrl > B (the second stagnation point is in this
case inside the circle lzl = B). See, for example, [3, Chapter III, 49].
sources within lzl = R and Qj are the strengths of the sources on lzl = R;
Iri = 0 where the rj are the intensities of the vortices on 1z1 = R.
(2) The singularities of the flow must be symmetrically situated with
respect to the circle lzl = R (see problem 1875). In particular, the axes of
doublets situated on lzl = R must be orthogonal to it. If there are sources
on lzl = R they must also have doublets. Moreover the sums of the strengths
and of the intensities must be equal to zero, for which: 2 I't+ ! Ir';
= O, where the I't are the intensities of the vortices inside lzl = R and
the rj are the intensities of the vortices ~n lzl = R; 2
Qj = 0, where the Qj
are the strengths of the sources on lzl = R.
1897. (1) w = Vz+c (c is a constant); (2) w = -2r:n:i.log--_-+c;
z-a
z-a
(3) w = ~ log [(z-a) (z-ii)]+c (at oo there is a source of
strength -2Q);
p 1 p 1
(4) w= 2n z::a+ 2n z-a+o;
402 PROBLEMS ON COMPLEX .ANALYSIS
p 1 p 1
+--+--+liz+c
2:n z-a 2:n z-i
7
n
(at oo there is a source of strength - .J; Q1c);
k=l
(6) w = -2-
:nz 2:ii
P
+ -r+Q
where Imp= O. A flow with
-.-logz+ c,
a vortex without a doublet is impossible.
r z-a
1398. (1) w = - 2 log R _ +c;
:it' -az
pl p*l .
(2) w = - - - + - - - + c , if
2:n z-a 2:n z-a*
a+ 0
k-1
m
= 2~
Q 1z -l
1404. w= -
2n
log --+c.
z1 +I
1405, w log (I+!) +c.
n
The flow is possible if tX = ~ {if I I't iF 0, then at oo there is
2 1'=1
n
a vortex of intensity -2 ,2; I't) .
k-1
n
1407. w ~ [r"+'Q"
= .J 2ni r"-'Q"
log (z-ak)+ ~-log (Rl-atz) - ]
k ... 1
1' I p I RI Vel
+-. --+-. --+
2n z-a 2n z-a
ve-lz- -z- +a
RI 1
( a = ii' p = Ba p )
n
The flow is possible if 2 I't = 0
k-1
1408. Let t =- /(z) ma.p D conformally onto the unit disk !ti < I. Then
to ""' fl)[J(z)], where
n
11: = /(at), with the indispensable conditionQ" = 0 2
k-1
1409. Let I = /(z) map D conformally onto the domain ltl > I with the
n
normalisation /(oo) = oo, /'(oo) > 0. Then with the condition 2 Q" = 0
k-1
we have w =- !J>[J(z)], where
404 PROBLEMS ON COMPLEX ANALYSIS
For I' = O, w(z) maps the exterior of 0 onto the exterior of the segment
[- !'~:), /'~:)] of the real axis of the w-plane with the normalisation
w(oo) = oo, w'(oo) = Ve-lex.
1411. (I) w(z) = _!_b [(az-by(z1 -c1)cosa:+i(bz-a y(z1 -c1))sina:]+const
a-
(Voo = Velex, c = y(a1 -b1));
(2) w(z) = ...!'.'.._ [(az-b y(z1 -c1)) oos a:+i(bz-a y'(z1 -c1)) sin a:]
a-b
+ 2
rm log (z+ y(z -c )}+ const.
1 1
1412. (I) w(z) = V(z cos GC-iy(z1 -c1 ) sin GC)+const (Voo = Vefex);
(2) w(z = V(zcosa:-iy(z1 -c1 ) sin GC)+ {m log (z+v (z -c ))+ const,
1 1
where I'= - 2n BV sin (or:+ {J)(I' is determined from the condition w'(I)
= 0 corresponding to the Zhukovskii-Chaplygin postulate).
1414. w(z) = y[z - (p/2)]+o is the streamlining of the parabola from
outside; w(z) = i cosh y(2z-p)/2yp is the streamlining of the parabola from
inside.
I 1!. _ mcx 1!. mcx
1415. w(z) = y.I [(z+ y(z1 -c1 ))1P e IP -(z-y(z-08))1/1 e I/I ]+const
is the streamlining of the right hand branch of the hyperbola from outside
b
(tan GC = - , fJ = n-a:, c = y(a1 -f-b1 ) and
a
i ..!!... 1!..
w(z) = y 2 [(z+y(z1 -c1 )).., + (z- y(z1 -c1 )).,. l+const
is the streamlining of the right hand branch of the hyperbola from inside.
1416. w(z) is determined from the equation z = enwto+nw/v (the values
of the stream function on the streamlined half-lines are taken as equal to v).
.ANSWERS AND SOLUTIONS 405
1417. w = cosh-1z =Log [z+Jl'(z1 -l)] (the values of the stream func
tion on the streamlined half-lines are taken as equal to 0 and n).
1418. (1) A fl.ow with period n; at the points kn (k an integer) there
are sources of strength Q; the points ; +kn are stagnation points. The ve-
FIG. 129
1419. A fl.ow with period n; at the points Tm there are doublets with
moment p; the velocity is V(:i: i oo)""' O. For the stream lines see Fig.
130.
1420. If V1 = iV, then the solution is possible and unique for V1
=i{v- 2~):
_ I'+iQ 1
"' - 2ni
n(z-a) + -I'+iQ 1
ogsm 2w 2ni
. n(z+ti) +
ogsm 2w '
(v--;;-Q) +
1 0
f>O
arg pf
FIG. 130
ANSWERS AND SOLUTIONS 407
A.
~B
]J
A ~B A
Cc:e, A Cce8 A
1) z)
~8 .8A
~ ]) c =
A
e1 cCc:e1
3)
FIG. 181
40S PROBLEMS ON OOMPLEX ANALYSIS
1413. (1) It is necessary and sufficient that the numbers M and 0 should
be real. In this case the curves Re u = < are equipotential lines. In Fig.
131 the mapping t = J(u) = C(u)+Ou is shown for various real a. Figure
131 corresponds to the case < > I<'[. By the solution of problem 1367, we
1
then have ea<O<e1 <e1 and k8>2
o'(~)
(2) JM= : 2< +c. For or:= 0 and M == 2n the mapping
< o~( "2:,or:)
t =JM= 71
CM--u = -
1 ~(2:) is shown i::1. Fig. 132
(note that
< 2< 01 ( 2: )
FIG. 132
1427. J(u) = ':! C(u-or:)+ !': C(u-P)+Ou+c. For the function /(u) to
(i. ;, lo are permutations of I, 2, 3). The points u = ~" (mod <, <') a.re
ANSWERS AND SOLUTIONS 400
critical, that is. f'(u) = 0 there. For the basic mappings see Fig. 133. The
rectangles given there are mapped onto the half-plane bounded by the
horizontal straight line (k = 1), the half-plane bounded by the vertical
FIG. 133
straight line (k = 2), and onto a two-sheeted quadrant joined along the
horizontal half-line corresponding to the dotted curve on the rectangle
(k = 3). These mappings are continued by the symmetry principle.
410 PROBLEMS ON COMPLEX ANALYSIS
! r,
1428. The periods of the B.ow are 4K and 2iK', doublets 2mK+ (2n+l)iK'
with moments 2n( - critical points (2m+ l)K+niK' (m,n integers).
For the mapping see Fig. 134.
FIG. 134
1429. The periods of the flow are 4K and 2K+2iK', the doublets are the
(-l)n+m
same as in problem 1428 with moments 2n ik , critical points 2mK
+2niK' and (2m+l)K+(2n+l)iK' (Fig. 135).
i i.
i i
FIG. 135
1430. The periods of the flow are 2K and 4iK', the doublets are the same
as in problem 1428 with the moments 2n(-I)n+mi. the critical points are
mK+2niK', The basic mappings are shown in Fig. 136.
ANSWERS AND SOLUTIONS 411
@ii])
I !
FIG. 136
I'+iQ a(u-oc) .
1431. /(u) =--.,.-log ( /J) +ou+o. In particular, for ix= 0, w,
...m au-
co+w', and {J = w' neglecting the additive constant and the factor r:;:.Q,
varying 0 and transforming to the a-functions, we obtain respectively,
log a(u) +Ou, log a 1 (u) +Ou, log aa(u) +Ou. If /(u+2w) = /(u) then
~M ~M ~M
r "Q 81(u;a)
f (u) = _t.;_ log ..w + e.
2m 81( u2:!)
1432. (1), (2), (3) doubly periodic flows with sources of strength 2n, and
-2n at the zeros and poles of the functions sn u, en u, dn u (Fig. 137).'
1433. A doubly periodic flow with quadruplets at the zeros of p(u) (Fig.
138).
143'. Periodic flows with the period 2w (period of the velocity!) with
sources of strength 2n at the zeros of 81<M (Fig. 139, (1) for 81 and for 81 on
displacing to the right by !; Fig. 139, (2) for 61 and for 68 on displacing
I) ~)
FIG. 139
The function /(z) maps B onto the exterior of two parallel segments, at a dfs.
tance l~I from one another (Fig. 140). The ends of the segments are
determined from the condition !Zi'(u) = O.
414 PROBLEMS ON COMFLEX ANALYSIS
@
"'' D C
:~
A 8
C D
FIG. I40
t
A = - 0 -.w
:n;
,B= ~
:n;
(c_ 8 -c_ 1 ). The problem is possible if C- 1 is a real
number and the difference c--a-c..1 is purely imaginary. If A of: 0, then the
function /(z) maps R onto the exterior of a horizontal ray and a segment
parallel to it and distant 1-; 0 - 11 from it. The ends of the segment and
the beginning of the ray are determined ftom the condition !li'(u) = 0 (Fig.
HI, (I)), (the case B = 0 is shown in Fig. HI, (2)). If, however, A = 0,
that is, there is only a doublet then R is mapped onto the halt".plane
bounded by the horizontal straight line and having a cut along a horizontal
segment at a distance '0; 1 1 from the straight line (Fig. HI, (8)).
ANSWERS AND SOLUTIONS 415
q b
a b
FIG 141
416 PROBLEMS ON OOMPLEX ANALYSIS
1439. (1) A solution is possible if I'1 -I'1 =I'; with this condition
f(z) = 4> ( :i log z ), where
(2
f~-a:)
01 - -
r.
4>(u} = ~log cu +-
2cu u+c,
cu
a = - . log a
2m 01 (-
u+a:)
-
2cu
m
[O { u-a: )]
1
(it 18
. necessary to b ear m
. mm
. d t'--t
,.... the mcremen
. t of 1og - - \ 2W
---
[ 01 ( u~a: )]
on varying u from 0 to 2cu equals 2m, and on varying u from 2w+iw' to
iw' it equals 0). The stagnation points of the flow are determined from the
equation
where A = I'1 /I', and are situated on the sides of the rectangle with ver-
tices (0, w, w+w', w') and the rectangles symmetrical to it. In the case I'1
= 0 the function 3(z} maps R onto a circle with a cut along an arc of the
circumference (Fig. 142; I'> 0). In the case I'1 = -I'1 =- ~ the function
FIG. 142
3(z) maps R onto the two-sheeted domain formed by joining the exteriors
of the circles 131 > 1 and 131 > ~
a
along the segments from - oo to
&ml'o
3o ""' - e--Y- where 'l'o is the value of 'I' at the stagnation point. The function
11(3) maps this two-sheeted domain onto the exterior of two lemniscates, and
ANSWERS AND SOLUTIONS 417
/(z) = - ~I'.
..:n;i
log [s(z)-s0 )(s(z)+s0 )] (Fig. 143; I'> O, a< e; it is neceesa.ry
to join the half-strips in the e-plsne along the common cut). In the general
case 'I'= 0 on the lower base of the rectangle in the u-plsne and varies
FIG. 143
r r
from - 2 to 2 +r1, snd on the upper base "'= 2nr log a+
1 r." loge i
2
and is varied from 0 to (I'> Q). r.
418 PROBLEMS ON COMPLEX .ANALYSIS
u-a)
(J (
2W r. o
+~
I'1-I'1 l i
og-(--::.-) + 2 ,.. u+ ,
(Ju-a ""
1--
2w
and u(z) - a+!:_+ ... maps the domain D onto a circular ring.
z
14150. The trajectories of the vortex are closed curves within the rectangle
with centre of symmetry at the point a+bi. For a vortex close to the centre
of symmetry these trajectories are nearly ellipses. If the vortex is at the point
a+bi itself these trajectories are stationary.
Solution. Let Zo = m0 +iy0 be the position of the vortex in the rectangle con-
sidered. Continuing the flow by the symmetry principle we obtain a doubly
periodic flow with periods 2w == 4a, 2w' = 4bi, having in the rectangle of
periods two vortices of intensity r at the points Zo and two vortices of
intensity -I' at the points i 0 Hence
(J (z-zo)(J (z+zo)
w(z)
r
= - - . log
1' 1'
2ni 91 (z~0 }(J1 (z!zi0 )
consequently
Sb
u tractmg tirom t his -2r . -1- and passmg
to t h e lim'it as z -+ z.,, we
m z-zo
obtain:
dmo . dyo I' ,. 2a; ,. 2i '"(Om I' p'(2mo)-p'(2iYo)
""dt-'dT = 2ni [,( o+ 2Yo)-,( Yo)-, -o>l = 4m p(2lllo)-p(2iyo) '
whence
ANSWERS AND SOLUTIONS 419
(since p(2iy0 ) is a real number, p'(2iy0 ) is a purely imaginary number). (From
() it follows that p' (2m0 )d3:0 = ip' (2iy0 ) dy0 Butfor the analytic function p(z)
the derivative p'(z) = apa =.;. apa and it follows from the preceding that
a: i y
dp(2111o) = dp(2iy0 ), consequently,
p(2111o)-p(2iyo) = a. ()
Such are the trajectories of the vortex within the rectangle. In order to in-
vestigate them, let us note, that when a:0 increases from 0 to a, the value of
p(2m0 ) decreases from +oo to 8 1 = p(2a), then increases to +oo, and when
y0 increases from 0 to b, the value of p(2iy0 ) increases from - oo to 8 1
= p(2bi), then decreases to -oo. We conclude from this that for C > 8 1 - 88
the trajectories () represent closed curves symmetrical about the point a+bi;
for 0 = 8 1 -e1 the curve shrinks to a point - the stationary case. In parti-
cular, if 1110 = a+,, Yo = b+17 and E, 17 are small, then p(2a + 2E)
AS e8 + ;! (2E}1 p''(2a), as p'(2a) = O; p(2bi+2Ei) ""'e8 + ;! (217i)1 p"(2bi), as
p'(2bi) = 0 and the equation (.. ) assumes the form p"(2a),1 +p"(2bi)17'
""' ! [0-(e1 -e8 )] that is the trajectories of the vortex are close to ellipses.
1455, u = J!1
r
sin (<t>-cx)," = E1 cos (<fl-ex); E =
r
2 1PI el(4>-);
~
a dipole (O; p) (Fig. 144).
/
i
i
FIG. 144
Fm. 145
ANSWERS AND SOLUTIONS 421
1469. (1) The value of the point charge is preserved; the law of varia-
tion of the moment of the dipole is the same as in problem 1893;
(2) The sign of the charge is reversed; the law of variation of the
moment of the dipole is the same as in problem 1896 on continuation across
the equipotential line.
1460. " ""' 2q g(z,a).
1464. w = 2qi log f;z) +c, where f(z) = [z- y(z1 -R1 )]/R.
1 1-cn(!z,k)
1466. w = 2qi log /(z) +c, where /(z) = and k is de-
sn ( ! z, 1c)
termined from the equation K' /K = b/a (see problem 1871).
6 ( Z-Zo) (J ( Z-Z9 )
1 1 4a 1 4a
1467. w = 2qi log/-() +c, where f(z) =
z 61 ( z;az1) 61 ( z;aza)
and w = 2a, w' = 2ib, z1 = (4a-a:0 )+iy0 , z1 = (4a-a:0 )+i(4b-Yo),
z1 = a:0 +i(4b-y0 ) (see problem 1460).
- + - -~
1468. w = -pi .RI
. + c ( a+ 0, a*= -,p* ~) ,
=-::a
z-a z-a q a
w = ~ _ii.z!. z+c (a =0), c is a real number. Compare with problem 1868, (2).
pi p*i ( R
1469. w = - - + - - . + c a+ oo, a*=-, p* =-=a p ,
RB)
z-a z-a q q
pi pi
w = z--Ri"z+c (a= oo). See problem 1400, (2).
1470. w = e(z cos a.+i sin a. y(z8-R1))+const.
422 PROBLEMS ON COMPLEX ANALYSIS
where pi= <!8'' The functions in the square brackets effect the normalised
conformal mapping of D onto the exterior of the horizontal and vertical segmenta
respeotivelyt.
n
1474. w = ,J; 2q1ci log /(z,1a1c) +if'(a, a)~(:.
a)-J/(z, a)] +o,
k-1
I-
where /(z, a1c) and /(z, a) conformally map D onto the unit disk with the
normalisation /(a1c, a1c) =/{a, a) = 0, /'(a, a)> 0 and o is a real number.
v(z, a) = log I
RI-_
azl =
,-[log/
1
z- i.i
I
.Rl +log l;I] if a O,
R
logR, ifa = 0,
I
and the value log lzl outside the circle.
IR1 -iiil
log R , a of:. oo, lzl ;;;;;. R,
(2) v(z, a)=
- log-1- 1- 1 , a of:. oo, lzl
z-a
~ R;
log lzl, if lzl ;;;;;. R,
v(z, oo) = { .
log R, if lzl ~ R;
l !all-RI
e(Re19, a)= - 2nR Rl-2Rlal cos (8-0t)+lal1 (a"" lalef and a of:. oo).
U a = oo the same potential is induced all in the preceding case for a = O.
lz+y(z1-R1 )1 1
1478. v(z, oo) =log 2 , e(a:, oo) = - 23tf(R-a;I) (la:I< R).
lz+y(z1-ol)l
1479. v(z, oo) =log 2 , outside the ellipse, "(z, oo)
.... - log 2(0t-/1) inside the ellipse. The density is e(z, oo) = - 2ny(l~-cI)
CC is on the ellipse, e' = 0t-f18).
a real additive constant): (1) w = ..1og (z-a)," = </lc.o(z; LI), LI= (- oo, a);
3t
z-b
<fl 1 [
(2) w=-log--,t1=</lc.o(z;LI), LI= (a, b); (3) w=- </11 log(z-a1)
n z-a n
z-a1 z-an ] .n
+<fl1 log--+ ... +<fin log , ti= 2 </IJ;C.O (z; Ll1;), 11; =- (a1;-1, a1;),
z-a1 z-an-1 k=l
a_1 = - oo; (4) w and " are obtained from the expressions indicated in
the answer to part (3) by replacing <flt by </11;-</10 , by the addition of
i</1 0 1io w and of </lo to v. The potential can also be represented in the form
n
2 </11;ro(z; 11;), Lio= (!Jn, oo).
k-o
424 PROBLEMS ON COMPLEX ANALYSIS
id d
1491. (1) w(z) = - 1- - logz+const, t1(Z) =- - - log r+const, where
og log
r1
=-;
rl
id d
og og
It (z)I + const,
(2) w(z) = - 1- - log t(z)+const, t1(z) = - 1- - log
where is the modulus of the domain D (see page 30) and t(z) conformally
maps D onto the circular ring.
1493. w(z) is determined by the formula indicated in the text of problem
ei(C) = -e1(C) = I -
1
2n log
1
2iilog
on lzl = 1,
on lzl = ;
Pu= PH= -Pi= -P11
l
=log;
;, ;,
el( ,.) = -ea (") = II Ilt'(C),_l
-
log
t(C)
onI'to
t'(C) 1
t(C) log onI'1;
Pu= Pza = -P11 =-Pu= log'
1
where t(z) conformally maps D onto the ring 1 < !ti < n, the contour I'1
corresponding to the circle !ti = 1.
ANSWERS AND SOLUTIONS 425
n
1497. t1(z) =Im f(z)- J; oc1cw1c(z), where f(z) conformally maps D onto
k-1
the plane with horizontal outs, and
f (z) =
I pi
Z-iJ + ...
piz + ... ,
if
if a
a #: oo,
= oo.
k=l r1c
1499. If /(a) = oo, then the field is formed by the dipole (a; p) where
I' is determined from the expansion off (z) close to the point a;
'( ) 1--1!!.__
z =z-a
+ ... , if a #: oo,
piz+ . , if a = oo.
1500. (1) If /(a) == 0, f(b) = oo, then the field is formed by the point
charges (a; 2q), (b; -2q) the flux of the vector intensity through each boundary
contour being equal to irero;
(2) If /(a) = 0, then the field is formed by the point charge (a, 2q),
the flux of the vector intensity through a boundary contour, corresponding to
a circle, in the direction of the outward normal to the domain D being equal
to 4:.irq, end through every other contour being equal to zero;
(3) The field is regular everywhere. The flux of the vector intensity
through boundary contours, transformed into circles, in the direction of the
outward normal to Dis equal to 4:.irq (+for the contour, transformed into
the outer circle), end through every other contour it is equal to zero.
1501. See problems 1487 and 1489.
n-1
1502. (1) t1(Z) = J; or:lc'1c(z)+c, where the or:1c a.re uniquely determined from
k=l
n-1
the system J; Plkott = 2qi (i = 1, 2, ... , n-1) (see problem 1104) e.nd c is
k=l
an arbitrary real number. The problem is equivalent to the construction of
the :Bow in D, streamlining the boundary contours I'1c with circulations 4:.irqrc
(k = 1, 2, ... , n), if oo e D, and by the circulations 4:.irq1c (k = 1, 2, .. , n - 1),
e
-4nqn, if oo D (I'n is the external contour).
(2) t1(z) = t10 (z)-2qg(z, a), where t10 (z) is determined as in part (1),
form the charges of the layer 2q1c+2qt, where
!lt = _ _!_
2:.ir
f og(C,a) dB.
on
r1c
uoa. "(z) = 2q1 log ~ +c, if !l = 0, and
r1
t1(z) = 2(q1-.Aq)log"jZj'"-2qg(z,a)+c,
426 PROBLEMS ON COMPLEX ANALYSIS
8 (log z+log a)
l 1 2:.iri
where A - og a g(z, a) = lzlA log - - - - - - 1
- loge' 8 (logz-loga)
l 2:.iri
for.,,.= (loge)/:.iri, where e < a < 1, if q:;tO (in the notation of problem 1439
the Green's function g(z, a) = Im 4i ( : log 111) for I'= 2:.ir and I'1 = -2:.ir;
the latter from the condition 'P = 0 on the boundary of the ring).
15H. The source (a; q) is transformed into the source (a*; -q), where
a is the point symmetrfoal to a. The function u = -2q log -11 1 +c, where
:ii (111, all
f (111, a)
conformally maps the domain D onto the unit disk (here and in what
follows the coefticient of heat conductivity Tc is assumed to be equal to 1).
measure of I'1;.
q U1-U1 lzl
1510, u = -2 g(z, a)+ 1 ( I )log- +uv where g(z, a) is the Green
:ii og r 1 ,., f't
function (see the answer to problem 1508).