286121-Functions of A Complex Variable - Text

FUNCTIONS OF
A COMPLEX VARIABLE
FOR
POST-GRADUATE AND HONS. STUDENTS
Dr. R. K. GUPTA, J. N. SHARMA,

M. A., Ph. D. M. A.
Head of the Deptt. of Mathematics Associate Professor of Maths.
Balwant Rajput College, Agra Meerut College, Meerut
Third Revised and Enlarged ]§ditioo~-

1966 * \ * C n,
\
KRISHNA PRAKASHAN MANDIR

EDUCATIONAL PUBLISHERS
MEERUT (U. P.)
OUR MATHEMATICS PUBLICATIONS
FOR
Post-Graduate and B. A. (Honours) Students
1 . Co-ordinate Geometry (Dr. S. C. Mittal M. A. Ph. D.)
2 . Integral Calculus (Dr. S. C. Mittal &J. N. Sharma)
3 . Differential Calculus (Dharma Vira & J. N. Sharma)
4 . Partial Differential Equations and Spherical Harmonics
(Dharma Vira & Dr. S. C. Mittal)
5. Differential Equations (Dr. S. C. Mittal & J. N. Sharma)
6 . Statics (Profs Brahamanand, B. S. Tyagi & B. D. Sharma)
7 . Dynamics (Profs. Brahamanand, B. S. Tyagi & B. D. Sharma)
8 . Theory of Equations (Dr. S. C. Mittal & J. N. Sharma)
9 . Algebra (Infinite Series) (J. N. Sharma)
10. Vector Analysis (Dr. S. C. Mittal)
11. Hydro Statics (Dr. R. K. Gupta)
12. Dynamics of Rigid Bodies (Dr. R. K. Gupta & J. N. Sharma)
13. Function of a Complex Variable
(Dr. R. K Gupta & J. N. Sharma)
14 . Theory of Functions of Real Variable
(Dharma Vira & Dr. S. C. Mittal)
15 . Determinants (J. N. Sharma)
16 . Co-ordinate Solid Geometry II (Differential Geometry)
(Dr. S. C. Mittal)
17 . Attraction and Potential (Dr S. C. Mittal & T. Singh)
18. Sequence and limits and Uniform Convergence of Series and
Power Series (J. N. Sharma)
19. Theory of Aggregates (Dharma Vira & Dr. S C. Mittal)
20. Theory of Statistics (J. N. Sharma & J. K Goyal)
21 . Spherical Trigonometry (G S. Malik)
22. Mathematical Theory of Electricity and Mp^nîsm
23 . Convergence of Infinite Series /

24 . Dynamics of Regid Bodies for B. A. Hops.
.
v.
25 Spherical Astronomy
Price Rs. TEN Only

All Rights Reserved by the Authors
Published by Printed by
B. D. RASTOGI, M. A., B. T. H. C. GOAL
Proprietor Proprietor
KRISHNA PRAKAHAN MANDIR LALITLOK PRINTING PRESS
Meerut. Meerut.
PREFACE TO THE FIRST EDITION
This book on ‘Complex Variable* has been written for the use
of students preparing for M. A., M. Sc. and Honours classes of
various Universities. The book will also be found useful for
various competitive examinations.
No claim to originality can be made but the authors have put
the subject matter in their own way after consulting all available
standard books on the subject. An effort has been made in the
direction of clarity without sacrificing any rigour. Obscure points
which are a source of confusion to students have been added
after each article.
Authors will feel rewarded if the book is found useful by those
whom it is intended to serve. Suggestions for improvement of
the book shall be gratefully received.
Authors also feel grateful to their publishers and printers for
their keen interest in the book.
—THE AUTHORS
PREFACE TO THE SECOND EDITION

It gives us great pleasure to being out the second edition of
the book in such a short period. In this edition the book has
been thoroughly revised and enlarged. Every care has been taken
to remove the misprints which had crept in the first edition. We
shall, however, feel greatly obliged if the readers point out any
misprints or send their suggestions for the improvement of the
book.
Three new chapters—Power Series, Uniform Convergence and
Elementary Trancendental Functions—have been added in the end
to make the book complete in itself and to serve the purpose of
the students of all Indian Universities.
We are grareful to the publishers and printers for their keen
interest in the book.
—THE AUTHORS.
PREFACE TO THE THIRD EDITION
This edition is almost a reprint of the second edition with
some minor changes. —THE AUTHOURS/
CONTENTS
Chapter Pages
I. Complex Numbers and their Geometrical
Representation - •1
II. Functions of a Complex Variable 34
III. Conformal Representation 70
IV. Conformal Representation (Continued) 116
V. Complex Integration 151
VI. Calculus of Residues 251
VII. Power Series 391
VIII. Uniform Convergence 402
IX. Elementary Transcendental Functions 449
CHAPTER I
COMPLEX NUMBERS AND THEIR GEOMETRICAL

REPRESENTATION
1*1. Introduction. The extension to the concept of complex

numbers from that of real numbers was first necessitated by the
solution of algebraical equations. The equations A*-fl — 0 and
**-2jc+3.*0, for example, cannot be satisfied by any number
in the domain of real numbers. Euler (1707—1783) was first
to introduce the symbol i for the square root of —1 with the
property, /*= - 1. It was Gauss (1777—1855) who first proved
in a satisfactory manner that every algebraic equation with real
coefficients has complex roots of the form a+ bi, the real roots
being a particular case of complex numbers for which the coeffi¬
cient of i is zero. His approach to the concept of complex numbers
was geometrical. Hamilton (18-5—1865) also made a great
contribution to the development of the theory of complex numbers.
His approach was arithmetical.
The concept of complex variable was usefully applied to great

variety of problems formerly studied through real functions.
Faulty proofs were replaced by more satisfactory ones and a
foundation for further progress was thus secured. By means
of complex integral calculus it was possible to evaluate a large
number of new definite integrals. The theory of differential
equations was marvellously developed. The use of complex
variable was also made in other sciences. For example, the flow
of electricity and the conduction of heat in two dimensions was
studied through this medium As a matter of fact the application
of mathematics to other sciences has made it indispensable to
have some knowledge of the theory of analytic functions of a
complex vaiiable.
Amongst the great mathematicians who have done immense

work towards the development of the theory of analytic functions,
the names of the French mathematician Cauchy (1789-1857) and
the two Germans Riemann (1836-66) and Weierstrass (1815-97)
are worth mentioning.
2 Complex Variable
Now assuming the theory of the aggregate of real numbers;

we proceed to define the complex numbers and the four funda¬
mental algebraic operations.
1-2. Definitions. A complex number may be defined as an
ordered pair of real numbers and may be denoted by the symbol
(x, y). If we write z~(x. y), then a is called the real part and y
th; imaginary part of the complex number z and may be denoted
by R (z) and I (z) respectively.
It is clear from the definition that two complex numbers
( v, y) and (.v\ y') are equal if and only if
x=*' and y=y'.
Sum of two complex numbers. The sum of two complex
numbers (x, y) and (x't y') is defined by the equality
(x, y) + (x\ y')=(x+x', y+y').
Product of two complex numbers. The product is defined by
the equality
(x, y) (x\ y') = (xx -vy , xy +yx ).
The symbol i. It is customary to denote the complex number
(0, 1) by the symbol i. With this notation,
/- = (0, I).(0, 1) = (0.0-1.1, 1.0 + 0.1) = (-l,0)
so that i may be regarded as the square root of the real number
-1.
Using the symbol /, we may write the complex number (x, y)
as x + /y. For, we have
(x+/y)=(x, 0) + (0, l).(y, 0)
= (.x, 0) + (0. y — 1.0, 0 0+1.y)
=-(-v, 0) + (0, y)
=U+U, 0+y)=(x, y).
1*3. Fundamental laws of addition and multiplication.
Commutative law of addition.
We have
(v, y) + (.x\ y') = (.v + .x', y+y')
=(*' + *, y'+y)
= (*',>•') + (*, y).
Associative law of addition.
We have
[(x, y)4 t*\ y'>] + (*". /')
= ( y + x', y+y') + (x'\ y")
= (.x+.x'-f A-",y4-y'+y")
= (A+x'+.x">y+y'+y")
Complex Numbers and Their Geometrical Representation 3
=(a, y)-r(x'+x\y'+y")
=(*, y)+[(x'+y')+(x",y")].
Additive identity.
We have
(x, y) + (Ot 0) = (*+0, + >•)•
Therefore complex number (0, 0) is the additive identity
and consequently it is called zero for the system of complex
numbers.
Additive inverse.
We have
(*,.y)+( -y)=(x-xty-y)=(0, 0)
The number ( -xf - y) is, therefore, the additive inverse of
(x, y) and is called the negative of the complex number (a, y) and
we write - (x, _y) = ( - x, -y).
Commutative law of multiplication.
We have
Cx,y) (x't y')=(xx'-yy\ xy'+yx')
= (x'x-y'y, y’x+x'y)
= (*', y') (x. y).
Associative law of multiplication.
We have
fU, y) (x\y')) (x". y1)
<=(yx'—yy\ xy'+yx') (x", y”)
= [(xx'-yy') x"-{xy'+yx')y"% (xx'-yy') y"~Hxy' + yx’) x')
= (* (x'x"-y'y")-y (x'y'+y'x"),
. x (x'y"-j y'xn)+y (x'x"-y'y")]
= (*, y) (x'x'-y'y, x'y”+/x")
“(*» y) [ix't y') (x\ /')]•
Multiplicative identity.
We have
(x,y) (1, 0) = (x.l+y 0,x.0+l y)
= (x, y).
Therefore the cotnplex number (1, 0) is the multiplicative
identity and is called unity for the system of complex numbers.
Multiplicative inverse. The complex number (a', y') is called
the inverse of the ccmplex number (a, y) if
(a, y).(x\ y') = (1» 0). ...(1)
We have from (I),
(xx'-yy', a>'' + >-a') = (1, 0)
4 Comi lex Variable
so that xx-yy' = 1,
xy'+yx'=0.
These give x' = —^—., ,>■'=- » provided (x, y) does not

x -\-y■ -Y ■+■>
represent the complex number (0, 0).

Thus a non-zero complex number (at, y) has a unique multi¬
plicative inverse, the complex number
and is denoted by (a, v) l.

Distributive law of multiptication.
We have
(v, y) [(a\ y') + (x" y"))
=(*, y) [x'+x", y'+y"]
= [.v (x'+x")-y (y'+y"), x (y'+y")+y (-v'4-a'')]
= [xx' + xx" - yy' -yy", xy'-t-xy"+yx' + yx"]
= [ax' - vy' -f xx" - yy", xy' -f yx' + xy" -f >’a''J
=(xx' —yy', xy' 4 yx') 4 (xx" - yy", xy"4-yx")
= ( v, y) (a\ y')+(x\ y') (a", y"u
We now define difference and division of two complex
numbers.
1*4. Difference of two complex numbers.
The difference of two complex numbers -=(a, y) and
r' = (x', y') is defined by the equality
=(v. y)+(-x\ ->■')

= [*+( - a'), >’-r(
= (a -a', y-y').
Division. It is defined by the equality r/c'=r(r'r\ provided
r'#(0, 0).
We have ‘, = (x, y) (a', y'r1
“(A,;)Gt-y2’ a:"+/*)
-( xx> . yy' _ x-v' + y^L ^
\x's+y'- x'-+y'* ’ x‘-+y * a-'24T V
/ xx' -f yy' yx' - x>'\
V x “+y “ a £4yj
provided x'2 f i
1*5. Modulus and argument of a complex number.

Let z—x-f iy be any complex number.
If x=r cos 0, y=r sin d, then r= + y/(x*+y*) is called the
modulus of the complex number 2 written as | z | and
y
0=tan-1 — is called argument or amplitude of z written as arg z.
•A
It follows that I 2- I =0 if, and only if, xr = 0 and y= 0.

Also argument of a complex number is not unique, since
*fe be a value of the argument, so also 2nn + 6, where /r = 0,
± 1, ±2,...
The value of argument which satisfies the inequality
-7T ^ 6 < 7T,
is called the Principal value of the argument.
1*6. The geometrical representation of complex numbers.

We represent the complex number z = x+iy by a point P
whose cartesian co-ordinates are (x. y)
referred to rectangular axes OX and OY
usually called the real and imaginary
axes respectively. Clearly the polar
co-ordinates of P are (r, 0), where r is
the modulus and 6 the argument of the
complex number 2. The plane whose
points are represented by complex
numbers is called Argand plane or Argand diagram.
It is also called Complex plane or Gaussian plane.
• ,N,°'C- , TJle 1comP|')l number z is known as the affix of the

point (x, y) which re/resents it.
17. Vector representation of complex numbers.
f lhIf/>,S tf!e Point (** y) on the Argand plane corresponding

to the complex number z~x + ly referred to OX and OK as
co-ordinate axes, then the modulus and argument of 2 are
represented by the magnitude and direction of Ihe vector
OP respectively and vice versa.
difference, product and division oHwoiompllVZmte7st,n8 ^ ^
the points P and Q on°theA^rgan^plMc^ ^ ** ^ rCpreSen'Cd by

6 Coni pi ex Variable
Complete the parallelogram

OPRO. Then the mid-points of
PO and OR must be the same.
But mid point of PQ is
(xj+*2 yx±y%\
V 2 * 2 )
so that the co ordinates of R are
(*i+*2, yi+rz)- Thus the point
R coiresponds to the sum of the complex numbers zx and z2.
In vector notation, we have
OR=OP+ PR
= OP+ OQ
= OR+OQ.
Difference. We first represent — zx by Q' so that QQ is
bisected at O. Complete the parallelogram OPRQ' and join
QP. Then the point R repre¬
sents the complex number
zx- r„ since the mid-points of
PQ' and OR are the same.
As OQ is equal and parallel
to RP, we see that ORPQ is a
parallelogram, so that
OR = QP.
Thus wc have in vectorial
notation
OP-OQÔP+QO
= OP+PR
= OR—QP.
It follows that the complex number rx-r2 is represented
by the vector OP where the points P and Q represent the comp¬

lex numbers zx and r2 respectively.
Obviously ! rx — r2 | —OP and arg (rx -rt) is the angle
through which OX has to rotate in anti clockwise direction, so
as to be parallel to the line QP,
• Another interpretation of (z,- z2).

Let the polar co ordinates of P and
Q be (rIt and (r2, 02).
Also let \zx-22\=a, i. e. PQ = a
and let PQ make an angle 0 with the
real axis. Then
zl-z2 = rl (cos 0! + / sin 6X)
—r2 (cos 02 + i sin 02)
= {rl cos 6X — r2 cos 02)
-f / (rx sin 0X -r2 sin 02)
= (OM-OL)-h/ (PM—QL) = LA/-f /. PN
= QN+i.PN
i6
=a cos 0 + ia sin 0 = a.e . ...(3)
Thus zx—z2 is a complex number whose modulus and argu¬
ment are a and 0 respectively.
• Product. Writing r, (cos 0!+/ sin 0,) and r, (cos 0t + i sin 0,)
for z, and z2 respectively, we have
ZlZ2=r\r2 fcos (0i4 ^21+/‘ sin (0| + ^2l]
W'c see that rxr2 is the modulus and 01 +02 the argument of
zxr2. The point R representing the product zxz2 may now be
obtained in the following manner :
Make OP=rx, OQ = r2, i_POX=Ox and £QOA = 0t.
Take a point A on the real axis OX, such that OA — 1.
Join PA,
Now rotate OP in the anticlock¬
wise direction through an angle
arg z,—02, Let OS be its new posi¬
tion. Through Q draw a line making
an angle equal to the /_OAP with OQ
and meeting the line OS (produced if
necessar)) in the point R. Then R is
the point representing product of
OR __OP
and OOR,
OQ OA‘
OR — OP.OQ, since OA—\
= fxr2,
and /_ROX= LROQV LQOX
~ Z POX i LQOX =t>x-\-(*2
8 Complex Variable
Note. Multiplication by i.
Since 2=r (cos Q+i sin 0),
and i—(cos l^-j-isin
we get iz=r £cos + + / sin
Hence multiplication of z with / rotates the vector for z

through a right angle in the positive direction
• Division. Since^-=^-[cos (Bt -*,)+/sin (0,-6,)], we obtain
the point R representing the complex number in the following
manner.
Rotate the line OP in the clock-wise direction through an
angle arg z2 = V2.
Let OS be new position of OP.
Take 0/1=1 and through A draw a
line making with OA an angle equal to
the LOQP and meeting OS' in R. Then
R is the point representing §-. For, we

r2
have from similar A'" OPQ and OAR,

ORÔP
OA ' OQ
or 07?=—, since OA — 1.
r2
Also /_AOR = /_POR-/_POX
= 02 — $i,
so that the vectorial angle of is - {6, -01)=61-6g.
19.
Conjugate Complex Numbers. If z=x+iy, then the
complex number x-iy is called the conjugate of the complex
number z and is written as is. It is easily seen that the numbers
conjugate to 2,+72 and zxz2 are isx + ?2 and 3j22 respectively. Also,
we have | r 12 = z3, 2R (z)—z+s
and 2/(2) = r-3.
Geometrically the conjugate of z

is the reflection (or image) of z in the
real axes. If (r, 0) are the polar co-or¬
dinates of P, then the polar co-ordinates
of P‘ are (r, -0j, so that we hate
amp z = — amp z, i
Complex Numbers & Their Geometrical Representation 9
1*10. Properties of Moduli. We now prove some basic

results on moduli.
Theorem I. The modulus of the sum of two complex numbers
can never exceed the sum of their moduli.
We shall prove that
! zi+*2 I < | *» | + I z2 |.
We have \ Zl + z2 |2=(z14-z2) (*1+*2)
= Z1~1 + Zl%2 + ~\Z-i +^2^2
= 2l?l4-^2 4-(^l32 + 2xZ2). ...(1)
Now ZlZi — \Zi I2, Z 2^2=1 z2 I2
and Zj32 + slz2=(x1 4- iyt) (,v2 - iyt)+(*, —iyx) (x2 + iy2)
= 2 (*1*24->>1;'2)=2/* (zi3?2)
< 2 | ZjZ2 |,
since the real part of a complex number can never exceed its
modulus.
Then (I) gives |z,4-z2 |2 < | z, |* + | z2 |*+2 zx32
= 1 zi l2+l z2 |24-2 zxz2
smce I z'i 1=1 ~2 I* so that |zxz2 |=] ZjZ21.
zi + z2 la < (1 z, 1-4-1 Z2 | )*.
Hence | zt+z2 < Izi!+|z2 I.
Alternative Method.
Writing Zj = rx (cos 0x-{-i sin 0X) and z2=r2 (cos 02 4-/sin 02),
wc get zl+z2 = (rl cos 0l + r2 cos 02)+i (r1 sin 0, + r2 sin 02), so
that I zi+z* 1= y/[ri*-\-r2*+2rxr2 cos (Ox—02)]
< V('i2+'a*+2rIr2] [V cos (0X - 02) ^ 1]
= ri +'•-/ zi I + |z*|.
H^ce Iz1+zt|<|z1| + |*2|.
Geometrical Interpretation.
Let P, Q be the points of affix
Complete the parallelogram R(l.+h}
OPRQ. Then R is the paint of
affix (z,-f z,).
Now | Zj | = OP,
= OQ= PR
and | z, -f z2 | = OR.
We know that in any triangle,
sum of any two sides is greater than the third side.
Hence OP-\-PR > OR
or I zi I + | z2 | > | zx4-z2 |
or zi + z2 I < | Zj I 4- I z2
10 Complex Variable
Theorem II. The modulus of the difference of two complex

numbers can never be less than the difference of their moduli.
Proceeding as in theorem I, we get
|*i-r, |2=|~i |2-2i? (*!*,)+! r212
> |z, l2 —2 I Z,22 l + l“2 I2
=1 zi I*-2 \zxz2 | + ]r, I2
=(\zx\~\zr I)2
so that [ zt - r2 I > | I - I r2 I.
Alternative. We have
zx-z2 = (rl cos 0, -r2 cos 02) + i (ri s*n -r2 sin 02)
so that | zx-z2 | = V[/'i*+/,2* -2rjr2 cos (0! -02)]
> V(r^+zy5 “ 2r,r2) [V cost^-^l < I]
=r i~r2=l z\M ~~ I*
Hence \ r, -r2 | > | zx |-| z2 I-
Geometrical Intemretation.
Here OP= | r, |, OQ=\ r2 |, OP=\ zx - r2 j.
(See figure of Tneorem I)
Since in a triangle difference of any two sides is less than
the third side, we have from A OPQ,
OP-OQ< QP
or I z\ |-| r2 | <| zx-z2 |
or I “i — —2 I >1 -i I — 1 za I*
Theorem III. The modulus of the product of two complex
numbers is the product of their moduli.
We have | zxz, (*~z,?23|38
= ’l7ir2-2
_| - 12 1-2
— I -1 !
i T 1i
so that | zxz2 1=1 zx 11 r21.
In the same manner it can be proved that
I’ll. Properties of the Arguments. We now prove the follow¬

ing theorems on arguments of complex numbers.
Theorem I. The argument of the product of any number of
complex numbers is equal to the sum of their arguments.
Let .j, Zn i —3... - n be the com pi ex num bers. L? t rj, r2,... r „
denote their moduli and 0X, P2...0n their arguments.
We then have
Z\-a•. .-„ = r, (cos P, +/* sin fix) r2 (cos P24-i sin ft,)...
rn (cos 0n+i sin 0„)
= r,r2[cos (01 + 02+..-+*») + « sin (*i+** + • • • + *•)!•

Arg (z,z3.. .zn) = 0i-\ 02+ •• *en
= arg z, + arg z2 + .. • + arg
which proves the theorem.
Theorem II. The argument of the quotient of two complex
numbers is equal to the difference of their arguments.
, Zi r, (cos 0, + / sin 0X)
We have ^ ^ (cos 0*+/ sin 02)
= -1 (cos 0x + i sin 0,) (cos 02-i sin 02)

r2
= ^ [cos (0,-00) f f sin (0,—02)J.
r3
Arg îl^=0,-02=arg z,-arg z2.
1-12. Solved examples.

1. Show that if a and bare real, the principal value of arg a
is 0 or n according as a is positive or negative and that of bi is v\2
or — ir/2 according as b is positive or negative.
Let a=r cos 0 and b = r sin 0 so that r=\ a |.
Then cos 0 = 1 or - 1 according as a is + ve or -ve and
sin 0=0.
Hence 0=0 or n according as a is + ve or -ve.
Again let 0=r' cos 0' and b = r' sin O' so that r =[ b \
Then cos #'=0 and sin0'=l or - 1 according as b is + ve
or — ve.
Hence 0' = tt/2 or -w/2 according as b is + ve or -ve.
2. Find the moduli and arguments of the following complex
numbers :—
1 -i 1 + 2/
(i) (>») 1 — (1 —«)" *
1 +/'
3-1,3 + /
(iii)
2+7 2—7*
We have
!-/_(!-/)»_-2f . .
(,) l+i 1-i* 2
1_—i
= + V(0 + ( -1)2 = I
1+/.
l-l* , a
and arg pp%=arg( -0
[see Ex. 1 above].

12
Complex Variable
(ii) _ 1+2/ _ 1-f 2/ _ I + 2/

1-(I-/)* 1 —(1 —2/+/*) \+Ji
= 1.
1 +2/ I ,
I 1 1= 1
and
arg (rr(T-,f=)=arg 1
=0. [see Ex. 1].
(iii) tz! + 3±l^(3-i) (2~/)+(34-f) (2 + /)
2+i 2- i 4__/2
= 162=2.
* 3-'4-3 + /i o
•• 2tp7+2=?' = 2
and
arg *
= 0.
3. (a) that
arg z+arg ■s = 2mrt
where n is an integer or zero.
Let z=x + iy, then ?=x—iy, where x, y are real.

Now arg z + arg z
= arg zu=arg (x-\-iy) (x - iy)
= arg (.v2+j*)
= arg a, where a=x2+y*.
Clearly a is positive and real.
Let a = r cos 6t 0=r sin 0

s° that r=Q
and cos 0=1, sin 0=0.
Therefore the general value of 0= 2mr where//k

or zero, 9 ncrc n ,s an integer
It follows that arg c+arg 2=2™-.
(b) If P, Q, R are the points of affix z - - ,
show that OPRQ is a pa-allelogram. ' " ~‘+ 3 re*Pec,'ye,P-
Let -i—-vi + b'i, ~z -- r. -f- /Vj, so that
ri+zt=(*i+ *,( + / (v,+ r.,)
so that the co-ordinates of O, P, Q r a'" ,n n. ,
and +,, ,, + ,,> respectively. “ ^ (°’ °K <**• <*» *>
13
Now mid-point of OR is
0+*,+*, 0+)\+y2 (xx+x2 yl+yi\
2 * •he- \ 2 ’ 2 )
and the mid-point of PQ is also
*i-i -*2 yx +y2\
2 1 2 )’
Hence OPRQ is a parallelogram.
4. Prove that
i *1+ z21*+| Zi - ^2 \2=2 I zx I2 +2 I z212.
Interpret the result geometrically and deduce that
I a+ y/(*x-P*)\ +1 a— v(a2“^l
==1 a-f-/3 |-f | a— P |,
all the numbers involved being complex. (Agra 1963)
We have
i*x+*2 r+i z\ ~z2 r
= (21 + 32) (*i + Z*)+(2i-Z*) (3i-32)
= rzjS, + 2z232=2 I Zj I2 f 2 I z2 I2. ... (1)
Geometrical Interpretation. Let P and Q be the points of affix

z, and z2 respectively.
Complete the parallelogram OPRQ.
Then we have
I zi \ = OP, | z2 1=00.
\zl + z2\ = OR, 1*1 — \=QP-
Now from a property of a para¬
llelogram, we have
OR2 + 0Pi'=20P* + 20Q*
or 1 *i + *212-H *i — *212 = 2 |zt |2-f 2 |z2 |2.
Deduction.
Now let zx = a-f V(a2 — /3*) and z2 = a- \/(y.2 -/32).
We then have
I *» l* + l ** l2 = ^ I *i4-*2 r-f-i 1*1 1* from (1)
=i|2«|2+||2V(*-£2) i2
= 2 | a l2 -f 2 1 a2 — /52 1
and so [ zx H-|z2l]fHZ| l*H~l *21*4-2 |ztz21
= 2|«|‘+2|a2-02| + 2|/3|2,
= | a+/3|*+| «-/3|!+2 1 a2 — /32 1 uîng (1 >]
= [|«+/3|+|«-/?||2.
Hence I «-f V(*2 -02) 1+1 * - \/(a? - ft2) 1 = 1 a + Z3 I + 1 a ~P !•
14 Complex Variable
5. Show that arg ^ is ^te ongle between the lines joining
the points a to z and b to z on the Argand plane.

Let A, B, P be the points on
the Argand diagram representing
the complex numbers at b and z
respectively.
Then the complex numbers z—a
and z—b are represented by the vec-
—>• —>
tors AP and BP respectively.
[See equation (2) of § 1 8]
z—a .
— tr B ^ n through which the vector BP has to rotate to coin¬
cide with the direction of the vector AP. For the adjoining figure
it is clear that this argument is positive.
Thus arg is the angle between the lines joining a to z

+0 “ Is
and b to r taken in (he proper sense.

Note. If AP is perpendicular to BP, then
z—a 77
ars rr*- ± 2
2_Cl
so that \~2fy ls Pure,y imaginary.
If AP coincides with BP, then

z—a
arg “ =0 or tt
Z—u
so that \^ is purely real.
It follows that if ^ is purely real, the points A, B, P are
collinear.
6. Prove that the area of the triangle whose vertices are the
points zlt z2, z3 on the Argand diagram is
2{(z3- z9)\zt \*l4izt}.
Show also that the triangle is equilateral if
=i2^-tiA-^2 = zlz2-j-z2z3Az3zl. (Agra 60)
Let zv z2. z3 represent the

points A, By C on the Argand
diagram.
Also let z^Xi + tyx.
z2=-x2-\-iy2,
z3==x3-\- iy3.
Then the required area
=£ *1 yx 1
*2 y2 1
*3 J’3 1
*i+*yi 1 1_
2/ 2/
x2 x2+iy2 1
*3 *3 4- iy3 1 1
= A- (**“**> ^ i (Î + »l) (Z2-Z3>
=5JSz' (*.-*.>+£ SM*.-*,)

1 z,-s, , . v ^ U. !2(^2-^3)
=--0+47s ir (2 3> -47*
Now the triangle ABC will be equilateral if
AB=BC=CA
i.e. if | Zj—72 | = | z2 “2a I —I z3—zi I (
i.e. if | — ^2 i2=l z2 _z3 l2==l z3"zi I"
/.e. if (z,-z2) (2X—22)=| ^2—^3) (32-23>
= (z3 —Zj) (23 — 2i). ..(1)
From first two of (1), we get
z, — Z2_ __Zg- z3 __(zt—z2) + (z2 ~z3)
i’g - 2# “»1 - s* (2t-33> + (3l "3a)
zi ~ Z2 31 “ ^3 ..(2)
or
~2 ~ S3 '2l - 23*
Again from last two of (1), we get
(Z2 - Z3) («2—23) ■=(Z3 - z,) (23 - Si).
Multiplying (2) and (3), we get

(Zj — z2) (z2 — z3) = (zt — z3)
or Z|Z2 —ZjZ3 — z2a-Fz2z3 = z12+z3 2zjZ3
or Zj* -f-z22 -f" z3*=Zj z2 ~F z2 z3 -+■3 3 z 11
which is the required condition.
16 Complex Variable
Alternative. The triangle is equilateral if

AB—BC=CA—a, say.
Let AB make an angle 0 with the real axis, then AC makes
an angle with it.
We then have,
\e+(7T/3)}i
^3 *1 _ ae niJ3
=e
r2 — ~1 6i
ae
[See equation (3) of § 1*8].
*. o f 7r/3) i
Similarly
-3 ^2
- ^
-n ** l =zi
Zn — zx
or
or ri:+ r22^ -32—rir2 + r2r3 + ::2r'l*
7. If rlf rs, r3 ///t? vertices of an isosceles triangle, /vg//f-
angled at the vertex r2, prove that
zf + 2:f + :f=2:., (z^za).
Let /I, B, C be the points of
affix r,, .-2, r3 respectively.
Let BC=BA = a and let BC
make an angle 0 with the real
axis so that BA makes aa angle
0+~ ^ with it.
We then have
/{(-/2| + 6}
*1-2 ae irr/2
—e [§ 1**J
~3 ^2 it)
ae
= cos — -f / sin — =/.

a.
Squaring both sides, we get

(- 1 “2 ) = ( ~~ Zn)~
or •1 T -2 * - 1 “ 2--3 -2 +
or - 2 i ^~ 2 - 2 -) _ / _ i-\
* *■ *• 2 T-a —~-2 ( - 1 “f" —3 )•
Alternative. Since f ABC=9GU, we have

_ <>
arg -i_ 7T
->-r -3> 2 9
r., — r,
so that —~ is purely imaginary.
*■ 3
17
Complex Numbers and their Geometrical Representation
Hence r (z-^f y 0
\Zt —ZZJ
i rz2~7*4-Eg—-i_l=o
or * U2-Z3 22-23J
2; — Zj __ _ 22 ~~”l _
_TV
...(1)
or Zo “ 23 ~2 ^1
AB — BC
Again
so that I *2“Z1 I H -2_Z3
Zo-Z 1 = (z2-z312
or (2)
(z2 —Z,) (2o —2,)= -(22—2j) (z2 —Z3h
or
Multiplying (I) and (2), we get
(z2-z1)2 (S2_?i) = _(S2-5i) (z2-Z3)
(Zg — ^3)
or (z8-z,)s+fe-z3)2=0
or Zl>+2zIt + Z3,= 2z2<zi+z3>'asbefore-
8. SAoiv that the triangles whose vertices are z„ z2, za and z,’,
Zo', z3' are directly similar if

A
1 =0.
Z2 1
Let A, B, C be the
points of affix zlt z2, z3
and A\ B\ C' the points
of affix Zj » z2, z3.
A'(fc)
Then the triangles
are similar if
AB=A'B'
..(1)
/!C ,4'C'
and l.BAC=B'A'C'=d, say.
z* - z, AC id ...(2).
Now
z2-z, /1Z?
Zs'-z/^^C' r0 .. .(3)
and , e
z2'-z/
From (1), (2) and (3), we have
*i_*3 “*i
z2 Z/-Z/ *
(Z2'—Zi')—Zl (Zz'—Zi ) — Z2 (z3 — z, ) zt (Z3 — Zj )
or Zy (z2'-z3') + z2 (Za'-z/l + Za (Z1'-Z2') = 0

18 Complex Variable
1 1=0.
7 *
~2
T '
-3
Alternative. Two triangles are similar if
AB A'B'
_^ - - _/ _7
" . “3 -2 *-3 ”*2
AC AC'
which is the sime as before.
9. Show that the triangle whose vertices are zlt z2, z3 and
V. -s'. -3 are equilateral if
(Zi-=z) (r/-r2')=(z8-r3) (zf-z/)
From the given condition, we have

z\ ~t _ *-2 ~_ (~i ~ r2)-f (z., — za)
-2 -3 -1 -2 1-2 “"3 J T“V~1 —-2 1
«*• _ «•
-1 -2 _ -1 — -3
** ...(1)
Also (z2-z3) (>V-z3') = (r3-r1) (V-V). ...(2)
Multiplying (I) and (2), we get
-r2)(r2-r3) = (r1-r3)=
°r *|2+Z22 + V='l-2+^2-3 + -3'l-
Hence the triangle whose vertices are zl% z2, z3 is equilateral
(see Ex. 6 above). Similarly it can be shown that the triangle
whose vertices are zx\ z2', zf is equilateral.
10. Prove that the centroid of the triangle whose vertices are
~ , ;c ri + -2 + ^3
-li -2* -3 -j--
Let A, B, C be the points of affix zXt z«, z3.

Also let z,=A'1-f iylt z2=.v2-f/j2 and z3 =.v3-f- iy9.
Then the co-ordinates of the centroid G are
/-XV±AV±*3
V 3 3 /
so that the affix of the point G
—xi ~F x2 4- xn , . y\ y9 -f y3
3 3
= i ['Xj+lrJ+fx: f i>2)+(at3 + iv3)J
= a (Zj -f-Z2-f Z3).
Complex numbers and their Geometrical Representation 1y
11. The roots zx, z2, z3 of the equation
x'i-\-2>uxz-\-3bx-\-c=Qi
in which a, by c are complex numbers, correspond to the points

A, By C on the Gaussian plain. Find the centroid of the triangle
ABC and show that it will be equilateral if and only if a~ = b.
Since z,, z2, z3 are the roots, we have
zizz== 3a or —--^- — — a and ZjZ2-}- z2z3-f- z3zx = 3b
so that the centroid is the point of affix -a.
Now the triangle will be equilateral if
2\ + *2*'+ Z2 = ZXZ2 + % + -3-1

or (*i + r2-f-r3)2^3 (zxz2+z2z3+z3zx)
or 9a2 = 9b
or a2 = b.
12. Find the circum-centre and ortho-centre of the triangle
whose vertices are given by the complex numbers zlt z2, z3.
If z denotes the complex number corresponding to the circum-

centre, then
| 2-7x |=| Z -Z2 |=| Z Z3 |
or I z—Zx |2—, z—z2 j2=| z-z3 j*

or (2-r1)(3--?l)=(2-r2)(2-^)=(z-r3)(3-33) ...(1,
whence from first two of (1),

2 (Z2 - Zl)=2l (Z — Zj) - Z2 (Z- Z2) . . .(2)
and from last two of (1),
2 (z3 — z2) = 22 (r-72)-u3 (z-z3). . ..(3)
Eliminating 2 from (2) and (3), we get
(Zz - Z, ) [22 (Z Z2) — 23 (Z — Z8)]
— (?3- zi) [~i (2-Z,)— 22 (z -z..)j

or Z [22 (~2 - Zl) - -3 (z2 -Zl)~zl (z3-z2)-f i:2 (z3-z2)J
= z222 (22-Z1)-Z323 (Zjj — Z,)
21-I (Z3 ~ Z..) + Z2S-, (Z3 - Z2)

or Z r 2, (z2 Z.j) = £ z,2, (Z2 - Z3)
or T Z)2| (z.j z:)

z=
2 2, (Zn — Z3)
20 Complex Variable
—2,
From (1), 7=5, - -—~ (5g—53). ...(3)
*3
•y _ **
-
From (2), 5 = 32 — - (53-5,). ...(4)
*3 “ Z1
Eliminating 3 from (3) and (4), we have
2 • ri / V
-1 ~2—“ ~ (So ~ <?3^
to “ —*3•» ^3 ~ -1
or (»!-»*) (r2-r3) (r3-z1) = (z- r2) »2?2-23) (28-rt)

(r r2) (jg • 2j) (r2 r3)
or r [i?223 ^2** i ^3-3 “4" ^3-1 — "4- 23r3 -f- ?jZo ^1^3]
= r r12(-2-23M-£r12l (r2-z3)
_^ 2i ^ (*^2 ^ 3 ^ “4* £ ~iî (Zo “
OF ^ (3,32-223,)
13. 5/;o»r //wf the equation of a straight line in the Argand

plane can be put in the form zb-\-b2=c, where bis a non-zero
complex constant and c is real.
Let 2,, 2n be two given points A, B on the Argand plane.
Let 2 be an> point P on the line AB. We then have
•* *• 1 r\
arg —— — 0 or n A P B
2. — -j
it t fa
so that f—~ is purely real.
21 ” r2
V-J
f
i.e.
II
v-l ~ *-2 S
or — _ *"2
*t » _ rs‘°2
-1 -1
Complex Numbers and their Geometrical Representation 21
or (Z—Zx) (3j — Z2)-(2-2i) (Zj-Zj) = 0

or 2 (3x-s2) -2 (2i-22) + (r122-21r2) = 0. . . .(1)
Now since z{2% is conjugate to 3xz2, the number ZjS*-?xzz is

purely imaginary.
Let zi-2 - ~Z\Zo — ic.
Multiplying (1) by /, we get
Z.i (Zj-Sg) —3.1 (z1-22)-c = 0.
Now if / (z2 - zx - —b, then / (2i~32) = 6.

Hence the equation takes the from
zb + zb=c.
14. that the point a' is the reflection of the point a in
the line
zb + zb=c,
if a'b + ab = c.
Let Py Q be the points of affix P(CL)
a, a'.
Now the point Q will be the
reflection of P in the given line if
the line is the right bisector of the
join PQ. Let R be any point z on
the line.
We then have
PR=QR
or | z—a |=| z-rt'l
or (z-a) (z-a) = (z-a') (z-a')
or z (a'-o)+z <a'-a) + (aa- ara') = 0. ...(1)

Since 7? is any point on the line, the equation (I) may be
regarded as the equation of the given line. Hence comparing (1)
with the given line, we have
a‘ — a _a' — a a'a'—aa
l b-^—“*• say
so that a' - a=bk,
a' - a=bk
and a'a' -aa—ck.
22 Complex Variable
Now a'b-\-ab= -j^[a' (a' - a)+a{a'—a)]
=T fa a'-<*<*]= f-ck=c
or a'b + ab = c,
which is the required condition.
15.
Show that the equation of a circle in the Argand plane can
he put in the form
:? + b'5 + bz-\-c=0,
where c is a real and b a complex constant.
v'»
Let r be the radius and a the affix
of the centre of the circle. Let z be
the affix of any point P on the circle.
Then we have
|r-a I =r
or jz - a |* = r2
or (z -a) (5 —a)=r2
or Z5~o3 — a'z-f-ai —r2 = 0
which is of the from
zt?+£- + £z + c=0
where c is real since aa—r2 is real and b
Clearly the centre and radius of the circle of form (1) are -b
and \/(bb — c) respectively provided bb—c -> 0.
16. Show that zx> z2 are the inverse points with respect to the
circle
bz + bz + c=0
m-2+te2 + ^z1 + c=P.
Let /, Q be the points of affix z,,
z2 and 0 the centre of the circle.
Then affix of O is —b and radius
of the circle
= \/(bb- c).
Now P, Q will be inverse points
with respect to the circle if
OP. OQ = (radius)2
and the points O, P, Q are collinear.
First of these conditions gives

| zx + b i | z2 + b \=bb-c ...(1)
and second condition gives
arg (*! + £) —arg (z2+ b). ...(2)
Since arg (z2+6) = -arg (32 + 6),
we can write (2) as
arg (z,+6) = - arg (32 + b )
or arg (r, + 6) (52 + 6) = 0
so that (Zi + 6) (32+&) ...(3)
is a positive real number.
Also (1) may be written as
I zi + t> I I ?2 + b \ = bb—c,
since the moduli of a complex number and its conjugate are the
same.
Fium (3) and (4), we conclude that
(?i + b) (z2-f 6)=66~—c
or z12?2 + 6jo4-^zl + c=0.
17. Show that inverse of any point a with respect to the unit
circle J z |=1 is 1 | a~
Let b be the inverse of the point a with respect to the circle
z|=l.
We then have
a 11 b |= l
or ...0)
and arg 6 = arg a— - arg a
or arg ba = 0,
so that ba is purely real and positive.
Hence ()) may be written as
ba= 1
or 6=1 | a.
18. Show that if the points zltzt,z9,z4 taken in order are
coneyclic, then the expression
~ zt) (z4 — z2)
( *3 z2) (z4 — zx)
is purely real.
24 Complex Variable
Since the points A% B, C, D are

concyclic, we have
/_ADB*= /_ABC
in the same segment.
Z3--1
Hence arg *4-*: =arg
r4-z2 *3-*2
(See Ex. 5 § 112)
or
■" (k)| <H)-°
or arg
(z4-z2> iZj-Zj)
(Z4-Z,) (z3-z2)
so that is purely real
(z4-z2) (Za-Zj)
or that its reciprocal J——-1-} —-- is purely real.

[Z3 — Zo) (-4 -1 )
19. A, B, C, D are the points z,, z«, z3, z4 respectively. Show

that if z1r2-f-z3r4=0 tf/;d zi+z2=-0, r/i? po/nte A,B,CtD
are concyclic.
The points z4, z2> z3, z4 are concyclic if
(^3-î)U4-^)
(z3-z2) (z4-z,)
is purely real (see Ex. 18).
~3 Z1 ) (~4 ~ -2)
Now
( "3 - *2.) (Z4 - Z2 I
(Z3Z4 +rtrs) - *Z1Z4 + "2‘3)
l ~3~4 "t zi "2) ” (z2z4“hziz3

— * -T -*
M^4T A’2*'3 [V z1z2 + z3z4 = 0J
~2Z4*1Z3
Z1Z4 ~ Z1Z3
- ~ ~m + so that z2=-z,J
*■ 1 “4 '
= — 1, which is real
Hence the result.
20. Find the loci of the points z satisfying the following
conditions :—
Z--J
O’) arS jx} — -y . OH
z+7
2 (///) |z*-l | <1.
O'v) 2—a cos t-\-b sin t, where t is real parameter and a,b
are complex constants.
z- 1 I)-f iy
(i) Now
z+1 (x+l)+i>
_[(*— l) + iy] [(*+1) -/>]
(AT+ l)2 + >’2
_(x2+y*-l)+2iy
(x+\)2+yz '
z—l . 2>>
arg^j-T=tan-1
a2+>’2- 1
_ 7T
Hence tan -l
x2-^*- 1 3
2y
or -5-—5—: = tan -- =v/3
x* + y*— 1 3
or -v24->'2 - ^3 > - 1=0,
which is a circle.
z-/
(ii) > 2
zp
Z -l
or > 4
*+ »
fz —l) (3 + 1) > .
or
(Z + 0 (3-0 ^
or 23 i / (z — 3) 4-1 > 4Z3 4-4/ (3 - z)-f 4
or 3zz+5/ (3—2)4- 3 ^ 0
or 3 (*•+>•) +5/ ( -2/» + 3 < 0
or 3*2 + 3>’2-4- lOj'4-3 < 0.
which represents the interior and frontier (or boundary) of thr
circle 3x24-3j>2 +10;>-4-3=0.
(iii) We have | z2— 1 | < 1
or (Z2— 1) (32 — 1) < 1
or Zz32 - Z2 - 32-f 1 < 1
or <x* + y*)*-2 (x2-y2) < 0
or r4 - 2r* cos 20 < 0
or r2 < 2 cos 26,
which represents the interior of the curve
r*=2 cos 20.
Alternative. | z* - 1 | < 1
or j r2 (cos 20 + i sin 20) — I | < I

or (r* cos 26— I)2-f-r4 sin2 20 < 1
26 Complex Variable
or r4 - 2r* cos 20 < 0

or r2 < 2 cos 20.
(iv) Here z = a cos /+£ sin t
or x+iy={al + aii) cos t-\-(bx-\-bti) sin f,
where a=ax + azi, b=bx +bz i.
Equating real and imaginary parts, we get

x=ax cos t-\-bx sin t,
y=a2 cos t+b2 sin t.
These give sin

bxa2 - o2a.
and xbo—yb
cos / = - * ^ .
.
Squaring and adding, we get
2 _(*0g— 3’gi)24 - jfti)2
(Vs—Vi)*
or .x2 (<v + V)+>J (0^+ b*)-2xy {axa2+bxb2)=(bxaz - Vi>*
which is an ellipse since
(Vs + ^s)*— (<?a2 + V) lV +V)
=2axa2bxb2 - t/22V - 6,2<V
= - MrMi)2
=a - ve quantity.
Note. a.v2+2/Mry+V+2g*+2fy + c = 0
will represent an ellipse if
hz -ab < 0.
21. Determine the regions-of Argand diagram defined by
(/') z3-z|<l. (//) |r-l |+|*+i | <4.
(i) We have
ZJ-Z | < 1
or r~ (cos 20 + / sin 20)—r (cos 0 + 1 sin 0) • < 1
or (r2 cos 20-r cos 0;2 + (r2 sin 20-r sin 0)2 < 1
or r* -2r* (cos 20 cos 0+sin 20 sin 0) + r8 < 1
r4—2r3 cos 0 + r2 —1 < 0.
Hence the region represented is the interior of the curve
r4—2r3 cos 0 + r8—1=0.
(ii) Here | z— 1 |+| 2+1 | ^ 4
or (*-!)+/> |+|(.x+ l)+/y ( ^ 4

Complex Numbers and their Geometrical Representation 27
or
or V[(x- 1 )2+/2] < 4-y/[{x+\)2+y2]
or (x-V* + y* < 164-(x4-I )2 + y* - 8\/[(x4- !)*+/]
or -4x-16< -Vl(*4-l)*+>'2]
or (*+4)2 < 4 [(x4-l)24->>lJ
or 3x2 + 4/2- 12^0
•2 -,2
X* V:
or
7+7 > >■
The required region is therefore the boundary and exterior
of the ellipse
X2 V2
T + T-1-
11 can be easily seen that the foci of this ellipse are the
points (1,0) and (— 1, 0) and major axis is 4.
22. Find the regions of the z-plane for which
I ^- i

where the real part of a is positive.
The regions are given by
I z—a i2 < = > | z+fl]2
or (z-a) (z-a) < = > (z+a)(z+a)

or zz-(as+az) < = > zz+za + za
or z (a + a 4-2 (a+a) > = < 0
or (z4z) (a+a) > = < 0

or 2x. 2 R (a) > .— <()
or
x > = < o [v R (a) is positive].
Therefore the required regions are the right half of the
Z'p ane: ,he ,ma8lnary axis and the left half of the e.plane
respectively. piane
A point P (a+ib) lies on the line AB, where A is

z = f> and B is z = 2ip. If Q is A ,s
-(-O' \
\a+tb )’
CO °,di"a‘eS °f P a"d<3 referred ,o ,he origin O as
ard ot an 7 aT.raS M"al tlne- ,ndicale 'he positions of P
ad Q in an Argon I dtagratn. If P moves along the line AB and
similar nn°ni, ,7 7’ ^ ,lle lrlan8,es OAP and OQC are

similar and that the locus of Q is a circle. ^
28 Complex Variable
Let a=rx cos 0,,

b=rx sin 0X
so that the polar co-ordi¬
nates of P are (r„ 0X), where
r1 = V(a* + b2)
and 0,^tan-1 —.
a
Now
PiK = -/+Ji
a-\-ib
= -~ (cos 0X - i sin 0.)

ri
so that the polar co-ordinates of Q are , tt-61
In the figure, we have

OA — OC=pt
OP=rl,
=?
ri
LAOP=Bx and /_AOQ=tt-6x,
so that LQOC=Bx.
OA p
Again
OP~rx
and OQ Pt/rx_ P
OC~ p ~~rx *
OA_°Q
*’* OP=OC
and LAOP=l_COQ.
Hence the triangles AOP and OQC are similar.
To obtain the locus of Q, let
-=_ r _ P2_{a-ib)
a-\-ib a2 b2
or p~ (a—ib)
*+/>=-
fl --f b* *
P2a
a*+b2 ..(I)
P’b
}'=
a*+b-' ...(2)
Complex Numberc and their Geometrical Representation 29
Also the point P ia, b) lies on the line
P 2P
so that
P 2p
or 2a + b = 2p. ..(3)
Squaring and adding (1) and (2), we get
. -H)
a* 4 b2
Also from (1) and (2),
y — 2x= (6 + 2*)
a2 t b
2p3
or >> — 2*=q2_^2 from (3). . .(3)
Dividing (4) by (5*, we get

x*+yi_P_
y - /x 2
or x2+y2+px-lpy=0, which is a circle.
EXERCISES
1. Find the moduli and arguments of the following complex
numbers : —
24 i
ii) o•
U) (tty 4i+( I +i)z
2. Find the foci of points z satisfying the following
conditions :—
2- 1
(i) | 2 - 1 1 ^ 2, tii)
z+\
^ 2 ,
(iii) i z2-l 1 > 2,
(iv) Z««f+y,
where / is a real parameter and a, b are complex constants.

3. Show that if the equation z2 + az-f 0=0 has a pair of conju¬
gate complex roots, then a, /3 are both real and
a2 < 4ft.
4. Show that the locus of the points z where z = — , js £

c-t df
straight line or cifcle according as j is real or imaginary
where a, b, c, d are complex constants such that ad -bc^ti)

and t is a real parameter.
30 Complex Variable
\
5. Find the equation of the circle described on the join of the

points a, b as diameter.
6. Find the equation of the circle through the points 1, /', l+i.
Find its centre and radius.
7. If the sum and product of two complex numbers are both
real, then the two numbers must either be real or conjugate.
[Let zl=xl + iyl and z2 -= x2+ iy2 ; then the sum 7j+:2 is
real if J'1+r2 = 0 and the product zxz2 is real if xx 12-f j1.v2 = 0.
These give y2= —yx and x2 = xx so that r2 = xx -0’j=21.]
8. The one and only case in which
' ri l + i "2 1+ • • • H M + -2 + • • • I*
is that in which the numbers r„ z2; ... have all the same
amplitude. Prove this both geometrically and analytically.
[Let zx=rx (cos flj + i sin 6X), z2 = r2 (cos 02 + /' sin 02), etc.,
then | zx | + | r2 \ = rx+r2
and | 2x-f r2 | = \/[/■12 + r22+2r1r2 cos (6l -02)J
=r, + r., if cos (0X -6»2)=1.
| r,4-z2 !==rj + if 0X = 02.
Hence | zx j + | z2 | = | zx+z2 | only if 9x = e2.
Similary
| zx + z2 + z3 1 = 1 (ri-Ko) cis 0x + r3 cis 6Z \
where cis 9X = cos 0, + / sin 0X
= \/I(ri-hr2)2-Fra2-4-2 (rj + ro) ra cos (0, —M
= ri+r2+rz if 0X = 03 etc ]
9. If r,, 22 are the roots of o.z* + 2f3z+y = 0, prove that
I -i I+ ~s i=r—, f —0+V(a'O l+,_ P - V(«-y) ]•

a !
Hint. See Ex. 4 of § 112 ]

10. If A, D, C D are four points in a plane, then prove that
AD.BC < BD.CA + CD.AB.
Let 2lt 22, z3, zt be the affixes of points A, B, C, D. We
have the identity
(~1 ““«) (“2 ~ "3^ "b (Z2 — -4) (-*3 “ -1) + (-3 — ”4) (2j —22) = 0
whence
1
(Z1 ~ri) (r2~ rn) i = | (-2 (-3—:i)t(:3 ~ zi) (Z1 — Zi) t
^ ! ("a ~i) (-3 ~ -1) |+i (-3 —(Z1 ” Zi) i

.*. AD.BC < BD.AC+CD.AB]
Complex numbers and their Geometrical Representation 31
11. Similarly triangles BCX, CAY, ABZ are described on the sides
XYZ are
XYZ arT"8 6 M£C'
coincident. Sh°W the c^!roids of ,f BC and
xI'yz'LT the affiXeS 0{A’B’C and *• y.* those of
b-c c-a~a-b~X
when (x-c)+(y-a) + (z_b) = \z (b-c) = 0
..
x-\-y+z = a-\-b-\-c
or x-\-y+z a-\~b + c
3 = r~-
Hence the centroids of ABC and XYZ concidej
ANSWERS
‘•® J2 ■ T Cli> -Ian-1 2.

2. (i) The exterior and frontier of the circle
(*-I 2 +y*=4.
(ii) The interior and boundary of the circle
3 (*2+yO-HO* + 3 = 0.
(iii) The exterior of the curve
r4 - 2r2 cos 20 - 3 = 0.
(iv) The frontier of a hyperbola.
5. 2zz-(a + b)z-ia+b) 3 + (ab +cb) = 0.
6. 22s+(/_il4_((+1)3=0>i(/+])> J
numbers can Tlso" b^repre””11 'he P°im “* infinil>’- c°mpl ex

sented by points on a
sphere. For this, it is suffi.
V (o,o,i)
cent to establish a one-one
correspondence between the
points on a sphere and the
points on a plane.
This correspondence is ^ ,
established by means of c#
take a point V on the sphere”?1^ Pr°;!eclion- F" this wt
equatorial plane as plane of nm , °f proJceli°n and its
32 Complex Variable
A' of the plane and, conversely, to each point of the plane, there
corresponds a unique point of the sphere.
Analytically we may proceed as follows :
Let the sphere be
X2 + Y*+Z*= 1
and the plane of projection be
Z = 0. ..-(2)
We take the vertex V of projection as (0,0, 1).
Let X, Z) be the co-ordinates of any point A on the sphere
and (x, v, 0) the co-ordinates of the corresponding point A', where
the line VA meets the plane of projection.
Since the points (0, 0, 1), (*, Y. Z), (x, y, 0) are in a straight
line, we have
X__ Y Z-l ...(3)
x~ y ~~ -1 *
These ive x= ^,y= 1 -Z

1-Z
X+iY
z—x+iy= (4)
1-Z *
And conversely, we get from (3) with the help of (1)
2.v --f 5
A'=
x*+vi+\=zz + 1
2y ...(5)
T=
A-24-;,a-fl 1(234-1)
x'+y'—l 23- l
and Z=
x*+3,I+ l“r3 + l • J
The equations (4) and (5i establish a one-one correspondence
between the complex numbers and the points on a spherical
surface except that no point of the complex plane correspond*
to the vertex V (0,0, 1) of the sphere. This is accomplished
by defining the point at infinity of the complex plane.
Point at infinity is defined to be the point corresponding to
the origin of the complex plane by the transformation
2=1/2'.
Again if the point (A', Y, Z) corresponds to the point of

affix r, then the point (X, Y, - Z) corresponds to the point of
affix z' where
X+iY {X+iY) {X - iY)
1+Z ~il+Z)(X-iY)
Xi+Y2 1-Z2
from (1)
(1 + Z) {X - in - (1 + Z) (X - iY)
Complex Variables end tkefr Geometrical Representation 33
1-Z_ 1
from (4).
X-iY 2
But
and arg z--^ - arg 3=arg
so that z, z' are inverse points with respect to the unit circle with
centre at the origin. Therefore the points of the Argand plane
which are inverse with respect to the circle J z J = 1 correspond to
points of the Reimann sphere which are symmetrical with respect
to the plane Z=0.
As a particular case, when z=0, we get from (5),
^=0, y=0 and Z = — 1.
and z' is the point at infinity by definition, which therefore corres¬
ponds to the point (0, 0, 1) of the sphere.
CHAPTER H
FUNCTIONS OF A COMPLEX VARIABLE
21. Sets of points in the complex plane. In Chapter I, we

defined complex numbers and their geometrical representation on
the Argand plane. We now consider sets of points on this plane
and defined certain terms connected with the theory of these point
sets.
Neighbourhood of a point. By a neighbourhood of a point z0
in the Argand plane, we mean the set of all points z defined by the
circular domain.
\ Z~zq \ < C ...(1)
We call e the radius of this neighbourhood
By the deleted neighbourhood of z0 we mean the set of all
points given by (1) except the point z0 itself.
Neighbourhood of the point at infinity. The set of all points
r such that | r | > K, where K is any positive real number what¬
soever, is called a neighbourhood of infinity.
*2+y8- 1
Since (see (5) of § 1*13]
*2+>‘+l
and x*+y* > A'!,
2 2
we get
Z=l~*‘+r+1 <
or z > x2+r
Hence the neighbourhood of infinity defined by \ z | > K cor¬
responds to the portion of Riemann sphere which is interior to the
small circle cut off by the plane
y_K2-\
A'2+r
Limit point. A point z0 is said to be a limiting point of a set
of points 5 in the Argand plane if every neighbourhood of z0, how¬
ever small, contains an infinite number of points of S. For exam¬
ple, the limiting point of the set of points of affix
Functions of a Complex Variable 35
is 2+/ and the limiting points of the set
+( -l)n.i, (/i=l, 2, 3. ...) are 1±/.

n+\
Interior and boundary points. A limit point r0 of set 5 is said
to be an interior point of the set if there exists a neighbourhood
of z0 which consists entirely of the points of S.
A limit point which is not an interior point is called a
boundary point of the set S.
For example, all the points of the set of points defined by
I z | < 1 are interior points of the set and the points defined by
| z |«=1 are its boundary points.
Open and closed sets. A set is said to be open if it consists
entirely of the inner points e.gthe set defined by | z \ < 1 is
open.
A set which contains all its limiting points is said to be
closed.
It should be noted that there are sets which are neither open
nor closed. For example, the set of points of affix
iTT-,1 (" = 1. 2, 3,...)

( -ir + n+
is such a set. It is not closed since its two limiting points ± 1 + i
do not belong to it. Neither it is open, since none of its points is
an inner point.
Bounded sets. A set of points is said to be bounded if there
exists a positive real constant K such that | z | ^ K for all points
z of the set.
If there exists no such point K, the set is said to be unbounded,
e.g. the set defined by
ni ,
Z~n+1 ’ 2» 3»
is bounded whereas the set z=/i2+2/n (//= 1, 2, 3, ...) is un¬
bounded.
Domain. A set of points in the Argand plane is said to be
convex (or connected) if every pair of its points can be joined by
means of a continuous curve which consists only of the points of
the set. An open domain is an open convex set of points. If we
add boundary points of a set to an open domain, it is called
closed domain.
2*2. Jordan curves. The equation z=x (t) + iy (0, where
x (/) and y (/) are real continuous functions of the real variable /,
35 Complex Variable
defined in the range a < / ^ b, determines a set of points in the

Argand diagram which we call a continuous arc.
A point z1 is a multiple point of the arc if the equation
zx=x (/)+iy (0 is satisfied by more than one value of / in the
given range.
A continuous arc without multiple points is called a Jordan
arc.
If the points corresponding to the terminal values a and b of
t coincide, the arc has only one multiple point, the double point
corresponding to these terminal values of t. Such an arc is
called a simple closed jordan curve.
For example, the arc.
r = cos / + / sin t
where 0 < t < 2n, is a simple closed jordan curve, the double
point being z= \ corresponding to / = 0 and t = 2ir.
The Jordan curve theorem. It states that a simple closed jordan
curve divides the Argand plane into two open domains which have
the curve as common boundary.
One of these domains is bounded and is called the interior
domain. The other is unbounded and is called the exterior domain
e g, the circle |r' = r, divides the Argand plane into two open doma¬
ins defined by | z | < r which is the interior and I z \ > r which is
the exterior. The circle ) r l = r is the common boundary of the two
domains.
2*3. Functions of a complex variable.
Definition. If a rule or set of rules be given by means of
which it is possib’e to find one or more complex numbers w for
every value of : in a certain domain D, we say that w is a
function of r defined for the domain D and write
Since 0\/(r) will be of the form u + iv where u and v

are functions of two real variables x and y.
We may then write
w = u (x, >•) \ iv (x, >-)•
Single-valued and multiple-valued functions, w is said to be
a single-valued or multiple valued function of c according as for
a given value of r there correspond one or more than one
values of »■.
2*4. LIMIT AND CONTINUITY. Limit of a function. If
w=f (r) any Junction of z defined in a bounded and closed
Function of a Complex Variable 37
domain D, we say that f (z) tends to the limit l as z tends to a

along any path in Z), if to each positive arbitrary number e, however
small, there corresponds a positive number 8 such that
\f(z)-l\ <
for all points z of D for which
0 < | z-a | < 8.
We write
Lim / (2)=/.
z-*a
The above definition means that there exists a deleted neigh¬
bourhood of the point in which | /(z) -/1 can be made as
small as we please.
Continuity. A function w=f(z) defined in the bounded
closed domain D, is said to be continuous at a point z=a of D,
if, given any positive number e, we can find a positive number 8,
such that
\f(z)-f{a)\<e
for all points z of D for which
0 < | z-a | < 8.
It follows from the above definition that/(2) will be conti¬
nuous at z = a if
Lim/ (2)=/ (a).
z-+a
Again a function / (2) is said to be continuous in a domain D
if it is continuous at every point of D.
Continuity io terms of real and imaginary parts of f (z). It
can be easily shown that if
f(z) = u (x,y) + iv (x, y)
is a continuous function of z, then u (x, y) and v (.v, y) are
separately continuous functions of x, y, and conversely if u (*,;•)
and v (x, y) are continuous functions of x, y, then /(z) is a
continuous function of 2.
Also if Lim/(2)=/, then Lim/(z) = h
Uniform continuity. A function f (z) is said to be uniformly
continuous in the domain D if for a given e, it is possible to find a
pcs!tive number 8 iniependent of a and depending upon e only such
that the inequality \f (z) -f (a) | < e holds for every pair of points
z, a of the domain D for which
0 ^ \ z-a\ < 8.
38 Complex Variable
It should be noted that continuity at every point of an open

domain D does not necessarily imply uniform continuity in D. If,
however, the function /(r) is continuous at every point of a
closed and bounded domain D. then it is uniformly continuous
in D. The proof of this is omitted.
2*5. Differentiability. Let w—f (r) be a singled-valued
function of z defined in a domain D. Then/(r) is said to be
differentiable at z=o if the increment-ratio
Aw f (a+Az)-f (a)
Jr = Jr
tends to a unique finite limit as r tends to a (i.e. as Jr tends to
zero) along any path of the domain D, ar.d this unique limit is
called the derivative of/(r) at z = a and is denoted by/' {a).
If we get different values of this limit as z->a from different
paths of D, i.e. if the limit depends upon ainp Jz, we say that
the derivative of/(r) at z — a does not exist and the function
/ (r) is said to be non-differcntiable at r — a.
Hence in order to prove non-differentiability of /(z), we
should try different paths for Jr. Convenient paths for Jr are
along real and imaginary axes i.e. by taking Jr either wholly real,
or wholly imaginary.
Just as in the case of real variable, we also have in the case
of complex variable the theorem which states that
Continuity is the necessary but not a sufficient condition for
the existence of a finite derivative.
It means that if a function is differentiable at a point, it is
necessarily continuous there but the converse is not necessarily
true. For example, consider the function
The function is obviously continuous everywhere because of

Funct ions of a Complex Variable 39
At a — 0, this limit is zero so that

/'(0) = 0.
When a-^t0, let Az=p (cos / sin ); then
Ais—p (cos </> —/ sin <j>)
Az _cosj/> — / sin </>
so that
zlr cos <A + / sin •/>
= cos 2j)—i sin 2$
which does not tend to a unique limit since this limit depends
upon arg Az. Hence the function f (z) is non-differentiable for
any non zero value of z.
Another form for the derivative of f (z).
The function f(z) defined in a domain D, is said to be
differentiable at a point a of D if
f(z)-fia)
z~a
tends to a unique limit as z-+a, provided that z is also a point
of D.
2*6. Analytic functions in a domain.

If a function is one-valued and differentiable at each point of
domain D, then wc say that the function is analytic or regular or
liolomorphic in the domain.
Again a function is said to be analytic at a point, if it is
analytic in some neighbourhood of the point.
A function may possess finite derivatives at all points of a
domain except at a finite number of points. These exccptioral
points are called the singular points or singularities of the
function.
2 7. The necessary condition for f (z) to he regular.
(Raj. 1958. 59; Agra 58)
Cauchy-Riemann Partial Differential Equations.
Theorem I. If a function f(z) = u(x, y) + iv(x,y) is
differentiable a! any point z — x + iy, then the four partial derivatives
ur, vz, uv, i y should exist and satisfy the equations
= ^y» =
Since f(z) = u (x, y)-\-iv (x, y) is differentiable at a point r,

the ratio
fiz-t-Az) -f(z)
Az
must tend to a unique limit as Az->0 in any manner.
40 Complex Variable
Since Az-Ax+iAy, we may write (1) as

ru (*-+ Ax, y+Ay)-u (*,) y
(_ Ax-riAy
. v (x+Ax, y+Ay)-v (x. >>)
Ax-\-iAy
]. ...(2)
Take Az to be wholly real so that Ay=0.
In this case the limit of the ratio given in (2) takes the form
Lim
im u(x+Ax, >•)-« (x,y)
Ax:->0 _ Ax
. V (x-fJ.v, y)~V (x, y)
Ax
]. ...(3)
Now since/(::) is differentiable, the limit given in (3) should
also exist, i. e. the partial derivatives and should exist and
the limit is w* f i•• ^

Similarly taking Az to be wholly imaginary, we see that the
partial derivatives i/„ and vy also exist and the limit of (2) in this
case is r9—iu9. ...(5)
As the limits given in (4) and (5) must be identical, we get
on equating real and imaginary parts
ux=vt and uy — -vx. ...(6)
The equations given in (6) are known as the CaucbyRientann
partial differential equations.
Sufficient condition for f (z) to be regular.
[Raj. 1958 (statement), Agra 58]
Theorem II. The continuous one valued function f (z) is
regular in a domain D if the four partial derivatives ux, vxt uv, %
exist, are continuous and satisfy the Cauchy-Rien ann equations at
each point of D.
We have
u = u (x,y) and u-f Au=u (x-\-Ax v + Ay),
so that Au = u (x-\-Ax. y-\-Av) - u (.v, ;•)
= M (a + Ax, y + Ay) - u (x-f Ax, y)
-fw (*+j.y, >•)-« (*» y)
= Ay vy (x-f J.v, y+0 -J.V)
-1- J.v ux (x-f O' Ax, y), ... (I)
where O<d<l;O'<0 <1*
by the mean value theorem*. _
*Mcan value theorem states that if (i) f (x) is continuous in a < x < b,
(ii) differentiable in a<J.r<6. then /(<j + /i)-/(a) =/;/' (fl+OA), where 0<0<J.
Now and — are continuous in the given region, and

by
therefore from the property of uniform continuty, we have
| w, (a 4-Ax, y + 0Ay)-uv (a, y) < e
and | ux (x+0'Ja>, y)—ux (x, y) | < e
provided that | Ax | < 8, | Ay [ < 8,
/. e. uv (a- + A x, y -f 9Ay) — uv (a-, y)=a
and ux (x+6'Ax, y) — ux (x, y) = fi
where | a | < e and | /5 < c.
Then, we have from (1),
Au=[uv (a, y) fa] Ay+[ux (a, y) + p] Ax.
Similarly, we shall get
Av=*[vv (a, y) + a') Ay+[vx (a,y)+p’\ Ax,
I a' | < €' and | /3' | <
Hence, we get
Aw_ Au + iAv
Az Ax+iAy
_ (uvAy + uxAx + aAy + PAx) + i (rvAy+ vxAx+ a'Av + fi'Ax)
Ax+iAy
= - v^y-f u.Ax+iu.Ay + ivÂx + ctAy+ /?Ja + /a'J>>4-//?'Ja
Ax+iAy
[V Uz = vv and vx= -*/v)
Aw = (ux + ivx) <Ax+iAy)+aAy+0Ax+ia.'Ay+ i(l'Ax
or
Az Ax+iAy
-{u, /, ) , (a + /a) Ay <P + i(i')Ax
'(+ ' x) + ^x+Tnr + Ax + iAy '
div
or j7 -(w» + i»W
^ | a-f/V 1 1 Jv|+1 p+ift' | | d*

Ax+iAy | ' | zia+ /dy
< Ml +1 a' l-f-l /3 |-H /S' |,
since Ax\ Ax + iAy \
and Ay | < | Ax + iAy |
Aw
^-(ux\ivx) <> + 2«\
Lim Aw
Henee
Az->0 Az
= ut+ivx. ...(2)
Alternative. Let w=f z) = u (a, y)+iv (a, y).
42 Complex Variable
Since u (a*, j') is continuous and differentiable in domain D, we

have by the mean value theorem* for functions of two variables,
Au = u (*-f-Ax, y+Ay)—u {x, y)
= ["* (x. >’) + <] Ax-\-[uv (.y, Ay, ...(1)
where e and r, tend to zero as Ax and Ay tend to zero.
Similarly applying this theorem for v (x, v», wc get
Av= v (x+Ax, y+Ay) - v (.v, y)
= fvx (-v, r)-f z] Ax+[vv (a-, y)-f r/J Ay,
where e' and r/ also tend to zero.
J»r = Ji/-f/Jv
~ fWx (*.->’) + fv* (X, ;•)] J.v+K (A*. V) 4 />•„ (x, y)] Ay
-f (c-rfV.) Ax+ (>) + />/') Ay
= ["* (v, >')4 i'x {x. y)] J.y + [ - v* (a-, y)+ru, (x, >-)] Ay
+ (* + /V) J.y-Ht,-j- it/) Ay fusing Cauchy]
Riemann equations ux = 'v, r/tf= — vr]
— ["* (V, y) -f-Mx (v, ;■)] (J.v-f iAy)
4-(e+/e/J* + (r( + ir/) J>’.
We then have
J»r _ J//-4-/Jr
Jr Jjc-t /Jr
= "* (AT, y. + ivxts. y) +

J v-t iAy Jv-f /Jr
Jit
. - {ux i.Y.y) + /vx (x, r)}
J:
f + « 1 ! J-';j . 1•') + />/ | ! Jr ‘

I J.v-t-/Jr ’ j J.v-i iAy
< I«!+I«'I + IV + I VI.

since | J.v I < | J v + iAy i
and j Jr | < | J.v+ /Jr j.
Lim Jtr
Hence
Jr-*0 Jr =I/x ( v* 4) + /»’r ( v, .v).
Thus/(r) is differentiable at each point of D.

Note. We have
(In ?ir
* ="' + f,'=0.v
l or proof see 5 154 page 27 S of Hardy’s Pure Mathematics*.

Funciions of a Complex Variable
d\v
Also
= y (wu+/vv)
_ 1 0H>
”T
so that / ' (r) is either equal to
d\v 1
— or —
Da i 0y'
2 8. An Important Observation.
Since x=h (z-fs), y=jr (--s), u and v can be regarded as
functions of two independent variables z and 3. If u and v have

first order continuous derivatives, the condition that w shall be
independent of z is
a-"'=o
°r ~ (M + /V)«0
or (— SZV/ ^ d£,3v D>’ \ n

\dx • 53 •+> • \da • 53 • air
or _L du _ 1 8m i 9v 1 5v
2 5a 2/ 5j> + 2 0a: T 0^ =0
or + / 5- + / 5Z. =n
CV 0>> 0.v *
whence equating real and imaginary parts to zero, we get
0« Dp
Da Dy
and dv_ _ 0W
0A-“ay’
which are the Cauchy Riemann equations.

Ii follows that iff (z) is a regular function of z, then x and y
can occur inf(z) only in the combination of x-\-iy.
Note. If/(z)=/(x-f iy)= u-\- iv, where /(z) is regular, then

the functions u and v of two real variables x and y are called
Conjugate Functions.
2*9. Laplace’s Equation. Let the function
/(z) = w + /v
Complex Variable
be regular in some domain D ; then Cauchy-Riemann equations

give
dii dv_ c)u__ 8v
dx=dy * dy~ ~dx '
Now assume that the second order partial derivatives of u

and v exist and are continuous functions of .v, y.
We then have from (1),
a ?u arv a*M a-v
dx2 ~ 3* ay' dy2 dy a.*’
so that ?!?=0 ...(2)
6.v=4 By2
Similarly, alv+§lv-=o ...(3)

dx~+ dy2
The equations (2) and (3) show that the functions u and v
satisfy the equation
c2<l> , dH ...(4)
a.v>+?j2'
The equation (4) is known as Laplace's differential equation.
2*10. Harmonic Functions. Any function of x, y which has
coutinuous partial derivatives of the first and second orders and
satisfies Laplace’s equation is called a Harmonic Function.
If/(r) = M + i> be regular and u, v both satisfy Laplace’s
equation, then u and v are called conjugate harmonic functions
or simply conjugate functions.
2T1. Orthogonal System. Two families of curves
a ( v, >') = cx and v (x, y)=c2
are said to form an orthogonal system, if they intersect at right
angles at each of their points of intersection.
Differentiating // (.v, y) = <T, we get
du du tfy_Q
dx + cv * dx
C_u
dy ?x
or say.
dy
Similarly from v (.v, y) = c2, we get
a«-
dy dx
or — =---= /;/.>. say
dx cv
Function of a Complex Variable
Now the two families of curves will intersect orthogonally if

m1m2= — 1
du dv.du 9^0. ...(1)
or 3* dx 3y'dy
Again if /(z)=u+iv is a regular function of 2, the functions
u and v will satisfy Cauchy-Riemann equations,
3u=3v
dx 3y
Multiplying these, we get

du 3v , du 9v__q
dx'dx dy'dy
which is the same as (1).
Hence iff (z) is a regular function of 2, then the curves
R [f (z)] = constant
and 1 [f (z))~constant
form an orthogonal system.
Example. If vv=/(--) = « + /v be ati analytic function of
z = x+iy. show that the curves u = constant, v=cons,a„t represented
on the z-plane intersect at right angles. (Agra 1952, 63)
2*12. Method of constructing a regular function.
Milne-Tliomson’s Method.
Since f (z) — u (x, y) + iv (x. y)
and x=\ (2+2), y=2i (z "
we may write
/(Z)=U [-i (Z+3). '2j (Z - 2)] + iv [i (z + 2), J' (z - 2)] . . .0)
We may regard (1) as a formal identity in two independent

variables 2, 3.
On putting 3 = 2,we get

/(2j = u (2, 0) + /v (2, 0).
dw . .
Now f (2) =
ai=“*+'v*
=ut-iuv, by Cauchy-Riemann equations.
Let ='M*»>’), uv=‘t2{x, y);

then /' (-) = '/*, {x,y)-Hz (x, y)
— '/’i (2, 0) — i<f>2 (2, 0).
46 Complex Variable
Integrating, we get
/ 0)=J -K (:. 0) d:-i j fa (z, 0) </z+c,

where c is an arbitrary constant.
Similarly if v (x, y) be giveD, we shall have
f&=j [0i 0) + /02 (r, 0)j dz+c,

where vv=^ (.v, y) and vx=^ (*, y).
Example. Find the analytic function of which the real part
is e~x {(x2~y2) cos y-\-2xy sin >}.
Here u {x, y)=e~* {(.v2 — yl) cos y-{-2xy sin >•},
0u
then */>!= — = e~x {2x cos y-\-2y sin y)
—e~x {(.v2— yz) cos y + 2xy sin y}

and * = ~=e~x {- 2y cos y-(x2-y*) sin y
2 2
-f 2.v sin y-t-2.vy cos y).

Hence /' (r) = «/»1 (z, 0)—(r, 0)
= <r‘ (2r-r-)
or /(‘)=J (2z-z2)
= c+(2r-r=)(-e-2) + | (2-2r) e~: dz
= c - (2z — zz) (2 - 2z) ( — e~:) — J 2<r2

= r*r: + r
= (at+/»2 e-(2+,‘', -f-c
— OH- «»2. e~* [cos >’-/ sin

Another Mtthod.
The regular function of which either real part or imaginary
part is known, can ulso be found by using Cauchy-Riemann
equations.
For suppose, u (x, y) is known. We then have to find
v (*, y).
Now dv=^- dx+ -v dv
dx dy
=— dx^ ~ dv,
dy 0-V '
by Cauchy-Riemann equations.
The equation (1) is of the from

dv=M dx+N dy.
Now 9A/=_92u
by by2
and
djf=dfu
dx a.x2'
. dN_bM_Vu a-u
dx dy dx*+dy*~0’
since u satisfies Laplace’s equation.
Therefore right hand side of (1) is an exact differential and
so v can be found.
Example. Let u = e* (x cos y-y sin y) ; then
dv , ,dv,
dv dx+ ~ dy
dx dy
dU .3 U .
“5- dx+— dy
dy dx *
e1 (x sin y+sin y+y cos y) dx
+ ** (x cos y-y sin .y+cos y) dy
v=
Jy cents. * (X S‘n y + Sin y+y cos y) dx
+ | (those terms which do not contain *) dy-\-c

= (x sin y+sin y+y cos y).ex—e* sin y+c
= e' lx sin y+y cos y)+c, where c is a constant
.*. f(z)=u+iv
= ez [x cos y->> sin y + lx sin y+ iy cos >*]-f ci
— e' (x+iy) (cos y+i sin y) + ci
=c*+“'.(x+0)+c/
= zez + ci.
2*13. Polar Form of Cauchy-Riemann equations.
c. , (Delhi 1959, Agra ’63)
Since Ar = r cos 9, y=r sin 9, we have
y2—r2 and 0 = tanyjx.
dr x
• • -=t = c°s9
and 0r
sin 0
dx i+/v*(-SO— >
r
dd
—=sin f> d0 = = cos fl
dy r r •
48 Complex Variable
„ am a« ar, auafl
Hence dx dr'dx^dOdx
du a da sin B
= -.cosB-
ar 30 ^
Similarly a>,= a7 sin tf+09-— *

av 0V n av sin B
^=3r-C0S(>-it —
dv dv . „ , av cos B
and
a.y a/- 3^ r
Therefore the equations
du = dv da av
dx dy * dy dx
become in polar from as
3« „ 0usin0_0v . , Srcosfl
57 cose-9^ — -0>S,n"+09 r
...(1)
0« . 0u cos 9-0v e sv sjntf
and ...(2)
3,s,n*+0TT arcosff+0tf- <• '
Multiplying (I) by cos 9 and (2) by sin 9 and adding, we get
3“=i 3v ...(4)
dr r dB'
Again multiplying (1) by -sin 6 and (2) by cos » and
adding, we get
r cB cr
Equations (3) and (4) are the Cauchy-Riemann equations
in polar form.
We can now obtain the derivative of/(r) in the polar form.
Wc have
dw dr , dw dB
dr ’dx dO ’dx
(du .3v\ fdu . 0v\ sin 6
cos B —|
\dr+‘*r) KdB*'dO)' r
(da . . tv\ ( av , . aw\ sin
i cos B-
\dr+‘ dr) C_rar+/rar>*7
= (cosf>-/sin 0) Q“ + ifr)-
dw , . . nJir
r = (cos B-I sin B) — ...(5)
dz o'
SOLVED EXAMPLES
1. Show that the functions u=x2 - 3xy2 is harmonic and find
the correspond ng analytic function. (Agra 1949)
We have
?_m=3*8_3v2
3a ’
— 6a.
dx*
du d2u ^
0— - 6^. sy«= - 6*.
d2u fi2u
so that u is a harmonic function.
c* oy
Now f' (z)=ux- iuv
= {3x2-3y*)-i(-6xy)
= (3z2 —0) —/ (—6z.O)
= 3r2.
f(z>=z3 + c.
2. Show that the function
u=sin x cosh y+2 cos x sinh y-\-x2 - y2 + 4xy
satisfies Laplace s equation, and find the correspond ng regular
function u + iv.
J",crc ^x~COS X COS^ y~ ^ s'n x S‘n^ >,+ 2Ar-f 4>’,
a2«
*= — sin x cosh y -2 cos a sinh >•+2,
dx
du
^ = sin at sinh >-+2 cos a cosh — 2^-*-4x^.
2 = sin at cosh y-f 2 cos a: sinh > - 2 ;

dv
3
.z r
az ■ a;
Now /' (Z) = ux~iuv
=cos a- cosh y — 2 sin v sinh y-\-2x-{-4y
— i (sin a sinh y +2 cos a cosh y—2y+4a)
= (cos z + 2z) -i (2 cos z + 4z).
f{z)={i\n z-\-z2) - i (2 sin z-\-2z2)-\-c,
which is the required analytic function
3. If u’=/(2) = m4/V and u - v=ex (cos y — sin y)t find u>
in terms °f z* (Agra 1952)
We have. u-v = e* (cos y - sin y).
Complex Variable
50
?lf =€* (cos y—sin >0 ...(1)

ax ax
— = — ex (sin y+cosy)
and ay dy
_9w= _ ex (Sjn j._|_cos y)
9x ax
[using Cauchy-Riemann equations]
d- +^* = ex (siny+cos y). —(2)

ax^0x ‘
Solving (1) and (2), we get
—=e* cos y=4>i (x, >’)«
ax
--=e* sin y—(*. >')-

ax
r(2)s=*?+/^
J ax^ ax
= </•! (X, >') + '■/'2 (x» y)*
... /(,)=( [#, 0)+i*. U, 0>] <fe
= J^<fc+c=*=+c.
Alternative* We have as above

^-=ex sin y ...(3)
ax
■ ?isse*cosy. ...(4)
and 1 ax
We get on integrating (3),
v=ex sin )•+/(}')-
Differentiating this w. r. t. y, we get
av =ex___ i f' (>’)•
cos V + i .a ...(5)
Also (4) gives

e* cosy=^=^.
3u a»* ...(6)
Then (5) and (6) give
f (y) = 0 or f(y) = c
v=e* sin y+c
and u = v + ex (cos y-sin y)
= ex cos y+c.
f fr)=e* cos y+c+i (e* sin y + c)
= ex.(cos y+i sin y)+c+ic

=e7V'+a.
or f(z) = ez + a.
Another useful device. We have
/ (z)=u + iv
so that if (z)=iu-v.
(\+i)f(z)={u-v)+Hu±v),
which shows that w-v and w+v are conjugate functions.
Let U=u—v and F=u+v;
then [e* (cos>,”sin yfi’

=ex (cos j'-sin jO — ’/’i (-*, ;■)
_0K__3f/__e;e ^ _sjn ^ _ CQS y)
and
dx dy
= -£* (sin >> + cos y)
= ^2 (*. y).
(1 + 0/(z) = J [<t>x (2, 0) - (-, 0)] * + c
= | (e*+fez) dz + c
= | (1 + 0 c*+c
or f (z)=ez+cl as before.
4 .
If u-v=(x—y) (x2 + 4xy+y2) and f(z)=a + iv is an
analytic function of z=x-\-iy, find f (z) in terms of z. (Agra ’42)
Now / (r) = u + i‘v, so that if (z) = iu—v.
.'. (l+0/U) = (“ ->■) + / (i<+v) = C/+iT, say
Here £/=u - v = (x—y) (x2 + 4xy+y2)
%=&~¥x=x'l+*xy+y,+(x~y) (2x+4>,)-
= 3x* + 6xy- 3y*
9C/_3u dv
and =" - *7 = -(x2+4+V+>2) + (X-y) (4A-+2y)
dy ~ dy dy
= 3x2 - 6xy - 3y2.
dU
Let
ox =f/»i (*>y),
dJJ
and ='h (x, y);
dy
then (1 + 0/(2)=j [<f>i (2, 0) - i'1'2 (2, 0)] dz-\-c

52 Complex Variable
= f (3z2 - 3iz2) dz+c
= (1 - /') 23 + c.
• •
or / (z)= -;z3 + /3.

Note. Students are advised to solve this qaestion by other
methods discussed in the previous example.
5* v 'findt,w analyiicfunction
f(z) = u + iv.
We have (1 + 0/(-) = (" - v) + * («+»’>•
Let u- v=U and w+ v=V, so that
(1-f/)/ z)=V+iV.
■ d + 0/'
= '^/ + /9-. bv Cauchy-Riemann equations

dy nv
0X
aK__ -2 sin 2.v (2c'-'v — 2e~-‘J)
Now
d.v ~~ (e2,J + e-”v -2 cos 2xf
= •/■! ( v, ;•)
and
aK
Sx~2 I
r2 cos 2.v (eZl> + e-~v-2 cos 2v)- 4 sin- 2v
+ - 2 cos 2x)- ]
_ rcos e2j+e |
= 4 l_(e2v+e“-y - 2 cos 2.v)-J
= ‘/>a »*, )’)■
(1 + 0/(-)=| ['/•, 0) + / '2 (z, (-)] +c
_ fT ( .4 (2 cos 2r- 2H
+c
-J [° + ' (2~2 coT2r)*
= 2/ Jf cos *r- ,1 f C
= — j J cosec3 z </z + <-
— i cot z-t-c.
/(r)=]J_.cot
= J (1 + 0 COt 2 + a.
6. f (z)—P+iQ is an analytic function of z=x+iy and

P_q^cos x+sin x — e~v
2 cos x — ev — e~v *
the condition f (7t/2) = 0.
2 cos x-\-2 sin x-2e~v
Now P-Q=t
2 cos x~ev - e~v
2 sin Ar+e*
+ ev -e“*'
-e-^l
-]
1+
2 cos jc —e*' - e-*J
or sin x+sinh y
t-v • rr»
r-Q-l 1 + cos x - cosh
. • . t ... .]■ ...d)
^ne^tiat,£8 (I) W* r‘ t0 * 3nd y P-A. nave
dJL - = £ | cos x (cos x—cosh jQ + sin a: (sin Ar+sinh y)l
dx dx [_ (cos x—cosh y)* J
_i fi —oos x cosh >>-fsin x sinh y~1
* L (cos x - cosh y)2 J ...(1)
and *--*8-1 [~cosh y (cos -E-cosh v)+sinh y (sin x+sinh >>n
dy dy [_ (cos x—cosh~j»)2 ~ j
or
dP __dQ = _dQ _dP
dy dy ~qx dx
_ l Tcos x cosh y+ sin x sinh y~ 1 "|
l x—cosh v)2
(cos AT— J* ...(2)
Solving (1) and (2), we get
d£, 1 - cos x cosh y
dx ’ (cos x — cosh y)*— ^x*
and dQ__y sin x sinh y
dx **(cos a:—cosh yf~^2 y)
• f'(z'_. . dQ
" ' (z'-a3f+,a7
*i (z. o)+/02 (z, 0)
A 1 — cos z _ I
* (cos z-1)»~rn^cosT)•
/(*)=if cosec2 -j dz+c
= -i cot yM.
\Vben z=tt/2,/(2)=0. A=l
?en“ a, /(z)=i (1 - cot z/2).
7. /f and fjj are functions of x and y satisfying Laplace's
equations, show that s-f- it is holomorphlc where
„_d<t> 00
dy di
and /=00,00
dx^dy (Agra 1936)
54 Complex Variable
It is only necessary to show that 5 and t satisfy Cauchy-

Riemann equations, i. e. we have to show that
ds d1 . ds_d[_
■ VX 3y a d3y <>x'
Since •/> and * satisfy Laplace’s equations, we have
3^4-al,'=0=a^+—• ...d)
3*i+0y* 3-v* 0y2
05 = 3'^__0*0
Now dx dx dy dx-
dr
_= j_0^
and dy dy dx dy2 *
3/ _ 05 __ , 9“*A
These give
3r~0.y dy* <> x2
=0 by (1).
• dS =r9/ .
0-V 0>‘
, 3J.0t_0V_ 020
Similarly, + 3>- 0* 3.v2 d.v 0v
4-^-0 by (1).
0A“ 0T‘
05 0r
Hence 0.v 0A
Therefore 5 and / satisfy Cauchy-Riemann equations and
consequently 5 4 it »s regular.
S. Show that the function e* (cos y+i sin y) is holomorphic
anil find its derivative.
/(:)=«+ /v
Let
= ex cos y+ i.e* sin y,
We have
-~=e* cos v
dx
du — ex sin v
dy
0V 3v
= <4 sin y,z-=e* cos y.
dx dy
These relations show that
?u dv Adu_0v
= . and—=-T7
dx
- x dy dy
so that u, v satisfy Cauchy-Ritmnn equations.

Also u and v are clearly continuous functions of x, y for a
finite values of at, y. Hence / (r) is regular.
Now f'(z) = <^ + j d-L
dx dx
= e* cos y+ie* sin y
=e* (cos y+i sin y) = .*+»■'
— e\
Here the derivative is identical with the given function. This
result is similar to that of the real function e*.
9. Shew that the function
J{z) = y/\xy\
is not regular at the origin, although the Can hy-Riemann equations
are satisfied at that point. (Delhi 1959)
Let f(z)sszu (x, y)+iv (x, y) so that u (at, T) = v'| xy and
»’ (*,T) = 0.
We tnen have at the origin,
9M=Lim U ■ — 0) ”f/
ox
jc->0
i 0-0 A
= Lim-= 0.
Similarly ^ = Ljm “ (0,

oy y
y-> 0
LimÔ - 0
y~* U y
and ?v i• —
r-=Lim o-° =0,
„
0* A-.0 *
dv I-
-=Lim °-0 „
— =0.
a>’ y-+ o y
Hence Cauchy-Riemann equations are satisfied at the origin
Now f (0) = Lim
z—»-0 z
=Lim yLiw±r_o
z-+ o i*+iy) *
Now if z—►O along y = mx, we get
/' (O)--Lim St. I
x-+0 xdfim) ,
Ay »
^ /
y/\m\
1 l
^ *f -#' * J
'
(!+/>») ‘
• cr «** si
A » a «* A
56 Complex Variable
Now this limit is not unique siace it depends on m. Hence

/' (0) does not exist.
/(*)=“ + v,
where /(*) =-jp+y*

/ (0)=0.
is continuous and that the Cauchy-Riemann equations are satisfied
at the origin, yet f' (0) does not exist. (Agra 1956, ’59, ’62)
x*-y* j x*+y*
Here —^ and v=-+?.
When 27^0, u and v are rational functions of x and y with
non-zero denominators. It follows that they are continuous when
2^0. To test them for continuity at r=0, we get on changing, to
polars, , a , . a
u=r (cos3 0 — sin3 9) and v=r (cos3 0 + sin3 6).
each of which tends to zero as r-*0 whatever value 9 may have.
Now the actual values of u and v at the origin are zero since
/(0)=0.
Since the actual and limiting values of u and v are equal at
the origin they are continuous there. Hence f (z) is a continuous
function for all values of z.
Now at the origin,
0u .. u (*, 0)-« (*\ 0)
— = Lim -
x
dx x-+0
. *-0
= Lim - 1.
*
du — v—0
Similarly — = Lim = -l.
W y-+ 0
0v «• x-0 .
]r- = Lim —— =1
0a: -
0 -x
9v . . y-o
— = Lim = i.
dy y o y
0U_0V
Hence
0X-0
0V^=_0U
and
dx dy'
The Cauchy-Riemann equotions are therefore satisfied.
Again / ' (0)=Lim /(r)
=-+Q
T.(x*-y*\±i (x*+y*) 1
— him x2-vyr *x+iy*
z-*0
Now let z—>0 along .y=x ; then
2»
/' (0) = Lim 2 (1 + 0 = i (*+')•
X->0 '
Again let z-*0 along x-axis. then

/'(0)-Lim^-Tr^[v y=o]
x-*0
-a+o. , « «
Since the two limits obtained above are different, the function
f(z) is not differentiable at z=0.
f (z) = e “ (z-jÔ) and f (0) = 0
is not analytic at z=0 although the Cauchy-Riemann equations
are satisfied at that point. (Agra 1936, ’54 ; Allahabad 47)
1 (x - tyr
-z-* ~(x+iy?^ ”(**+>V
Here /(z)=e =e —e
_i 6x2>’2 - 4ix3<y+4/xy3)
=e
where r2«=xa-f>2
_ -7, <** +y* - 6*V> ^
"ee x*+/-6 wxVC0S

-( *
r vu (v2_
+ (lin{«5*£=2!>}]
Hence we have at the origin,

aU . u (x, 0) - u (0, 0)
— = Lim-3
0ac *
e“*"4-0
Urn
x-*0
1
Urn
1/x*
x-*0 xe
Lim“T—r—i-l
*->0x[l+-4+2I-,+ ...J
1
Lim —
1 . 1
x-Ox+i
58 Complex Variable
=— =
OG
0.
du u (0, v)—u (0, 0)
Similarly — = Lim--
>.-o y
, .-^-0 n
>’->0 y
0v .. v(x,0)-r(fi,0)
— = Lim -
dx x—>-0 X
= Lim-=0
a--*0 A
dv 0-0
and —= Lim -=0.
*y .v->o y
Hence Cauchy Riemann equations are satisfied at the origin.
- -r-4
Now f (0)= Lim 6-—
r->0 r
exp. (— /—4.e i-O i*/4

= Lim —J -:—Ti-, taking z — re
in {4
0 re
= Lim IA = oo
—0 re'”l4
Hence f (z) is not regular at the origin.
Note. For a function to be regular, the first order partial deri
vatives of u and v must be continuous in addition to Cauchy'
Riemann equations.
12. lff(z)J?y£Z£*> (z^0),f(0)=0, prove that
fhhfjo) ^ 0 us
along any radius vector but not as z-*0 in any manner.
Let z—*0 along y =■ mx.
Then Lim^-^°)
Z-* 0 “
= Lim fy (y-ix)
=-+o i*+y )(x+iy)
— Lim ** mx (n x- ix)
~~ Y_^(j (.t'+wV) (x-timx)
m (/«—/)..v2
=0.
,v^0 (",2+*4) +
Functions of a Complex Variable
Now let 2=0 along y = x3.
Then Lim
2—>0
i — I V)
==Lim(!(x6+^) (at+7x3)
T._ (X*-l> _ _ 1 ,
“"Ll™2(l + iV) 2
x-*0
Hence the result.
13. For what value af z do the functions w defined by the
following equations cease to be analytic :
(i) z—log n + iF where w = f> (cos 4>-f / sin <f>),
z=sinh u cos v+ / cosh u sin v,
(//)
where w = u~riv.
(i) We have
$-=(cos +-/ Sin.A)®- [see (5) of §2'13)

0P
= (cos <l>—i sin <£).—
dw_p_
or c?2 cos <f> - i sin <f>
or ^=p (cos ^ + i sin «£) = w. . ..(1)
Therefore in order that w may be a regular function of 2, we

conclude from (1) that w should be finite. Now w will be finite,
so long as p is finite, /. e. so long as 2 finite. Hence w is an
analytic function of 2 in any finite domain.
(ii) Here ==— [see note of § 2* 7]

dw du
= cosh u cos V + / sinh u sin v.
Now 22 = sinh2 u cos2 v—cosh2 u sin2 v
-t*2/ sinh u cosh u sin v cos v
=(cosh2 u - l) cos3 v - (1-fsinh2 u) sin2 v
+ 2/ sinh u cosh u sin v cos v
= (cosh u cos i’4-1’ sinh u sin v)2- 1
=(-Y-
' dwj
= ±V(z +l)2
dw_^_j_
dz ~ + I)
60 Complex Variable
Hence w will not be regular when rl-f-l=0,

/. e. when z = ± i.
14. For what values of z, the function w defined by
z = e~v (cos u-\- i sin u), w=u + iv,
ceases to be regular ?
We have
dz 92 r . . , . x
— = — — e~v ( —sin uy i cos u).
dw du
= ie~r (cos i/ + / sin u) = iz
or ^ =Hence w ceases to be regular at 2=0.

dz iz
15. Show that an analytic function with constant modulus is
constant. (Allahabad 46)
Let /'(r) = u-f/'v.
Since \f (z) =constant = c say, we have
«!-f v2 = r*. ...(1)
••• +*•£-•
and 2M?“+2y|!» o.
3y dy
ILing Cauchy-Rieraann equations, these become
3u du
u --v — = 0,
3-v dy
and «_+,|?.0L
3.V 3*
du
Eliminating we get
?v
(“■ + •*> 3-“ = °-

du
Thus —- = 0 provided «•= w-f ;r^0,
dx
Similarly, ^=0,51 = 0
dy dx
and a-
dy = °-
Since the four partial derivative of u, v are Zefo, thd
functions u, v are constant and consequently w==u-\-iv is also
constant.
16. If f (z) — uyiv is on analytic function of z = x+iy\ and ip
any function of x and y with differential coefficients of the first and
Funct ions of a Complex Variable 61
second orders, then
!/'<-'
and (Agra 1954)

(i) We have
90 _ 90 9m 90 3_v
9a:-9m * 9*+9v * 0x ..(1)
and 90__90 9m 90 3v
9>>“9u ' 9y f 9v • dy
or
90 _ 30 9v 90 3«
9y 9m * 93c+,9v • ax • •
•(<)
by using Cauchy-Rieraann equations.
Squaring and adding (1) and (2), we get
©Xf Me)'*©'] r®
since
(ii) Again we have

920 920
9xz_f9y2 V9 x dy) \9x 9y)‘ • • • (1)
Now i-+« 1
= 3m 9 9v 9 / 9y a du 9\
dr 3-v ) m + 3x ' 3jA V"9-v * 3m+^ * fv)
[uîng Cauchy*Riemann equations).
-(«-/ 8V>) *- + / (?L_/ 3^ 3_
\9x 3x/9m \9x 9x / ' 9v
8-^V3 +/ m
\9a* 9xA9m+ 9v> *..(2)
Similarly
3
-/ 9 _fa“+/3lV3 • a N
dx 3y V9x+,a*yvaM a^y* ...(3)
Now u-\-iv=f (z)=f (x-j-iy). *..(4)
-c/ » -/ r ■■■■■ *^ • i • i
i -/•(*+«(£+'to
62 Complex Variable
0* 9^1 3^_92L
gv ”h‘ 9v 9m 3m '
a* 9y . cy dx
Then ...(5)
dv "~du an 9v "~9m'
\ . /9m . Sv \
i+i i
3v )*' Ca5+' iJj
Yd,u,.&yydx sy .dx dy\
='W+' 8*,Ar»+' Tu 3V-0.J
= 0 by (5).
Hence we have from (1), (2), (3) and (5)
9_V' , d±YJ.+, l-')]

5lYi.+/ A")]
9 a-2 + ar~l\9-v
9Aa_f 9v2- V9.v 9* A?m ^
dxAdu gv / I
dv )j
X^9M.+ . 3v
■XI-1 If)]
/3m_. dvXfd4> _i 90V_3 .
+ \3* * A
9* 9m 0v A 9m
-gj+”) !/■ (••>!■•
Aliter. Since m-*m+/v, *r»« - iv, we have
M=J (»f+n), v = 2T )
such that
a_ __ a 9m a ‘A _ J ( A 9_^
an,-9M dw dv 9h~ 2 \9« * 9v /
9 _ A dji , 3 3v __ _]/A_ a. /. 9
a iv 9m 9w 9)’ 9*v 2 \3m 1 dv
where u and v are treated as functions of two independent
variables m* and iv.
9 <L_ L(JL-i iYi + / ±\
*• dw 9~ — 4 \9m 0v A3« 3v )
, 1__L YAl
4 0H’0IV ~”\0Ma ‘ a»8/*
/av 3VN 9V
4 A!i
Hence
Van- avv 9h-
9w a»v*
9 V , ?V_4 320
aV
Similarly
a.v2+?r 9- 95*
Also h=/(r) and »r=/(2).
Functions of a Complex Variable 6 3
. _ 9^_ __4 / 3
0»v 9w \dz d\v) \02 dw)
"4(/' '?) dz) (/' <s)3*)

1
=4
f (Z)f' (3) 02 02*
since z and 2 are independent variables.

. /0V , 0jA\ 4 0V' __\_RV , 3V1
" V0U8+0vV 0VV* 0 w /' (2)/( 2) L0A2 0>-2J
or £+iHP+py <-)/■« )
17. Iff (z) is a regular function of z, prove that
(a>S 0
,/, = |/(D |* = u2 + v2,

Let
then
to to’
8K- 0X2
Similarly, we have
P4-SMg)>[>8KS)']-
— (S+S)
=4[®AI)>4'“
using Cauchy-Riemann equations and the condition that tt, »■
satisfy Laplace's equation.
Hence ’V!3
8
0*2 0y -4,/'wp.
Alternative. Since x-f i>=z and x—iy=ist we have
x=y(z-fz) and - y (2—2)
so that A=2 J to=A(A_iA^

0z 0x * 0z 0>» ’ 0z 2\0* dy)
64 Complex Variable
and
a_ a_ dx.d
as dx as dy • 03 2Va*+#aW’
x and v being treated as functions of two independent variables
r and 3.
• 0 3 • J3 VJL > /
as ' as 4 Va* 1 dy)\dxr ay)
or 4 _9-=-0!+ll
eras a a -2 a.)-1*
Hence (S+S«) ^ '°=4 a/L I/W/Wl

-4l [/<*!/'<3)]
= 4/'(z)/'(s)=4|/'(z)|2.
18. If w=f (z) is a regular function of z such that /' (z)^0,

prove that
If \f' (z) is the product of a function of x and a function of

V, show that
f (z)=exp.(a='-+pz+y)
where a is a real and ft and y are complex constants. (Agra 1960)
Now log !/' (r) | = i log |/' (z) |*
= * log f ' (z)f’ (3)
=4 log/' (r) + * log/' (s).
Hence (a0>3?) io8l/'(-^1
= 4 r~rz [£ log/' (s) + £ log/' (?)]
[see Ex. 17 alternative method]

=0.
For the second part we proceed as follows :—
Let \f (=) 1 = 0 (-v) 0 (>’).
Then, since log \ f (z) 1 = 0,
+ flog (v)-log 0 00] =0.

2
i. e. log 0 (.v) + jp log 0 00=0.
Functions of a Complex Vanab.e
As there is no relation between * and y, let

d2 , . / ,
log «/. U) = c,
a real constant. Then

jp log«A(^)=-c.
These give, log </> (x) = |cx2-t-f/.v + e,

i. e., </■ <*)—exp. (£cx2-f </*+<?)
and 0 >>)=exp. (-\cy2+d'y+e'),
where dt «?, d’, e' are real constants.
Hence \ f' (z) l=exp. (x2-y2)-{ dx+d'y + e + e' J. ...(I)
Now I exp. (az2+/Jz + 7) |

= | exp. {a (Jf+iF)2 + (a+6/) (x+f» + (c-f <#)} |,
where (3=a+ib and y=c4-/</
= exp. {a (a- ->’sJ+o.x—
[V \eA+lB\ = eA]
which is of the same form as (1).
Hence we can write/' (z) = exp. (v.z2 + (3z +7).
20. If f (z) = u-\- .v is a regular function of z in any domain,
prove that
<0 (fr.+|?) \f(*> 1 ’=p‘ 1 fm I-21 f w i!-
co 1«1 r=p (p-n i«1-21 r (*) i*.

(i) Let «/>= | f (z) |*=(u2-rv2)»"2.
^=2 (m* + v2)<,,m''1 (2«m,+2vvx)

w=/> (M24- V,)tvli>-1 (MW*+VV*)*
and |ur-f vvx,2

-4/> (w* + v2)<',*,-1.(wwa*-f wx*+ vr„-f vx2). .. (1)
(1)
Similarly, (-| - l) («*+ v*/*«-*.2 <1/1/,+ » vv)2

+/> + (wrv„-f i/„H »'vWf+ »■„*). ...(2)
.(2)
Adding (I) and (2), we get
\x2+\y'i=P {p~2) (u2+v)‘"2,'5!

x (u2 (u2-}- wv2) 4- vl <»•,*+ »•„*)+ 2«v (1/^4 tV'vJ
+ P (n’+v2/"2'-1 [»/ (»,,4 ww)-l-v (v^.^-4 t'w)
+ «,,4^4«,S + ^J
66 Complex Variable
=p 07-2) (*++)**- [(«■+*■> (ux2+vv2)]

+p («!-fv2)<*,/8,-1.2 (iix ■+■vv2)
(using Cauchy-Riemann and Laplace’s equations)
= (/>*-2-p+7p) (tf+v*)**'2*-1 <i/x2+w,2) .
=/>a(H2-fv2)<*-2>'2 (»«*+0
=?2 i/(*> r2 \r w i2.

(ii) Let |m|=A
and 0= | M |*«A*.
0X ^ 3*
and g-^-w^v^S- (1)
Simihriy |4=P 0>-D A-2 (|A) +pk^' ~2. ...(2)
Adding (1) and (2), we get
-[(!)•*©•]
+Â‘”1 @+l?} •••(3)
Now obviously A will so satisfy Laplace’s equation, so that
a2A 3*A
a*2^2
and
/aAY=/a«\2
\9y) \dyJ *
Hence, we get from (3),
S+fr» "-> [<SHBG

=/> (p-Di«r# irooi*.
Alternative. We have
® ©+I !/(2)|P ;) ! / (*) I

0
=4aTSl/«|B=4aTk[/(2,/(J,ir"
[see Ex. 17 alternative)
=4 Jj.[/(2)]"2 y [{/(s)}"'2-1/' (2)]
= 4 i [{/(z)}"12-1/' (--)] f [{ / (2)P'’-'/ ' (2 )

=P2 [/(*)/(*)p/,~1/' <*)/'(*)

=p2i/(*>r* i/'wi*.
<”> (aS+3?0'“ 11=4 air,[IH/(2)+/(;,}11'
(V /(z)=w+iv and /(3)=u—ivj
= 4 gj^-, [i {/(*)+/(»>} {/(2)+/(2)}]W2
-45rFH{'w+/w>1'
= 24-p| {/(z)+/(3))M/'(3)
= 2îP IP- 1) {/(Z)+/W}--2/' (S)/'(2)

=/>(/>-D [i {/(z)+/(3)}]’-*/' (2)/' (Z)
=/> (/> - 1) |u|”-2!/'(z)la.
Exercises
1. Find the regular functions »■*«+/» where
(i) u = e x {(x2 -y2) cos >’-F?.\'y sin y}.
(ii) u=x2 - 3xy2-|-3x2 —3y2+1. (Raj. 1959, Agra 1955)
...._sin 2x
111 U~cosh 2y+cos 2x*
(iv) u—ex (x cos y-y sin y). (Agra 1957)
2. Show that the functions
(i) «sJlog(i, + >,,)l
(ii) w=cos x cosh y are harmonic, and find their harmonic
conjugates.
3. If /(r)= —(zt^0),/(0) = 0. prove that
/ (z)—f (0) _
-~">0 as z->0,
along any radius vector, but not as z-+0 In any manner.

4. Show that the function/ (r) = xy-f iy is continuous everywhere
but is not analytic.
5. Un is real, show that the function rn (cos nO+i sin nQ) is
analytic except possibly when r=0, and that its derivative
is nrn~x {cos (//-I) 0+/ sin (n-1) 6).
f (z) = s\n x cosh y+i cos x sinh y
is continuous as well as analytic everywhere
Complex Variable
68
For what value of r does the function w=u + iv, where

7.
sin u cosh v+i cos u sin v,
cease to be regular ?
8. If / (z) is an analytic function of z, prove that
ANSWERS
e-x (x—iyY tcos y -i sin y).
1. (i)
(ii) (x -V /)•)*+3 (* + '-v )2 + 1 + ic‘
(iii) tan (x+iy).
(iv) (x+iy) er+iv + c.
2. (i) tan-1 yfx+c, (ii) -sin x sinh y + c.
2*15. Multiple Valued Functions. While defining regular-

function, we have seen that a regular function must be o re \iue
or uniform. There are, however, a large number of functions whic
are many-valued eg. tan"1 z, log z, zn (n not an integer .
explain the idea of many-valuedness, we consi Jer the simple case
of the function zl
Let u'—z1'2 or n-2 = z.
Putting z = r ices 0+/sin r is fixed and 0 lies
where
between 0 and 2*. we see that two obvious values of w are given
^ it'! = v'r e*
H/tf .
and tr2-= ,\ r e
. , ,
— v ’ _
^/r\eYl°
and are called the two branches of the two-valued function

Branch-cut or Cross cut. By making a narrow slit, extending
from the origin to infinity A '
along the positive real
oc -X
axis, we see that each of
ft
the branches and »»v
is a one-valued function in the z-plane if we distinguish between
the values of the function at points on the upper and lower edges
of the cut.
If OAthe value of »»•, at A (t> -0) is | \'a \ and the value
of H'i at B (0 = 2*0 is -| |. Nov on account of the cut, a
complete circuit about the origin is not possible.
Hence if we start with a value of z belonging to the branch
uV we can never change over to the branch wt Thus \vx (and
similarly w«) is a one-valued function on the cut plane.
The part of the real axis from x=0 to .v=oo, is called a
branch cut or cross-cut for the branch »»•,. Cross-cut is a boundary

so introduced that the corresponding branch is single valued and
regular throughout the cut-plane.
It is conventional to choose the positive real axis as the
branch cut. We might, if we like, take the negative real axis, or
more generally, any simple curve connecting the point z = 0 and
r=x>.
Branch-point. If z describes a circle of radius r about the
Xf)i
origin, ^ = 1 Vr | e varies continuously and the final value of n-,
is | \fr | e^'*ni= -| \/r | = »r2 at 0 = 0.

Hence the function n\ is apparently discontinuous along the
positive real axis since the values just above and just below the
real axis differ in sign and are not zero execnt at the origin itself
If, however, z describes a circle round the origin a second time,
the values of iv, continue those of »»•« and at the end of second
round, we have w2 = wx along the positive real axis.
We now distinguish the point z = 0 from all other points-
We see that if z describes a circle about any point a and the origin
lies outside this circle, then arg z is not increased by 2n but
returns to its initial value. Hence the values of and n\, are
exchanged only when z turns about the origin.
On account of this fact, the point z = 0 is called a branch¬
point of the function \v=\/z and as we have already stated, it*,
and w>2 are called its two branches.
Definition If a point is such that some of the branches of a
many-valued function interchange as the independent variable des-
aibes a closed path about it, then it is called a branch-point.
The branch-points of a function are always to be found among
those points for which two or more of the values of the function
become equal. This common value of the various branches of the
function may be finite or infinite.
CHAPTER Ilf
CONFORMAL REPRESENTATION
3*1. Let the two complex variables r=.v4-/> and u =«4-/v

be represented by some relation \v=f (z) where f(z) is an analytic
function of z. For different values of z which vve represent as
the points in the rplane, there are corresponding values of w
which are represented as points in the w-plane. It can also be
said that a point in the w-plane represents its corresponding
point in the r-plane, and the regions of the r-piane are represented
by or mapped on the corresponding regions of the w-plane. We
discuss this typs of representation or mapping in this chapter.
If u -u , y), vs=v(.y, y) are continuous and possess

continuous partial derivatives of first order at each point of a
domain D, and for any given curve in this domain laving
turning tangent, there corresponds a curve in the domain D'
of w-plane possessing the same property, then it is not necessary
that the coriespondence between the two domains is one-one.
A simple example for it is given by */=.v*, v=y-. The cir¬

cular domain < 1 corresponds to the triangle bounded by
the line w = 0, v = 0, u-\- v= 1. But there are four points of the
circle corresponding to each point of the triangle.
Definition. Let the two curves Clt C2 in the domain D

intersect at the point P <.r0y0)at an angle a, and if the two
corresponding curves F„ To in the domain £>' intersect at P'
(u0y vo) a* die same angle ux the transformation is said to be
isogonal.
If the sense of the rotation as well as the magnitude of the

angle is preserved, the transformation is said to he conformal.*
In the following figure (i) the tiansformation is isogonal

and in (ii) it is conformal.
It by the given transformation w=/(r> any figure in the

region D is mapped on the corresponding figure in the
•Some writers do not distinguish between isogonal and conformal trans-

formation.
>
Conformal Representation /I
domain D', isogonally or conformally, it is not necessary that the
O ^ plane x O ur plane Xa-
correspondence between the domains, is one-one. The correspon¬

dence is said to be one-one or biuniforni if to each point in domoin
D there corresponds one and only one point of domain D’ or vice
versa.
3*2. The necessary and sufficient condition that w ■=/(z)
represents a conformal transformation is that \v—f(z) is analytic
one- valued and regular. (Agra 1943, 47, 50, 53, 58, 59. Luck. 49)
Condition is Sufficient. Let us suppose that u»*-/(z) is an
analytic function of z, one-valued and regular in any region D
of the z-plane. Let z0=^» + 0'o be ony interior point of the
domain D through which the two continuous curves are passing,
having definite tangents at this point and make angles a,, and
<*2. say, with the real axis. The aim of this analysis is to discover
as to what figure is represented in the iv-plane to a corresponding
figure in the z-plane. For a certain reason which will be known
afterwards we make the restriction that f (z0)r£0.
Let z0. z,,z2 be represented by the points P, Px% P2 of the
z-plane on the curves Clt C2; and let the corresponding points
w0, wn w2 be represented by Qt Qlt 02 on the curves rlt r». of the
»v plane, such that
id. iPn
z1-zJ=re ; z2 - z0=re
where z, and z2 are at the same distance r from z0.
12 Complex Variable
Then as r->0, 01-»-a1 and Q2-*ai

iV-i i'h
and »'i - »‘o = Pie > u2-w0=^p2e
If we suppose zx — rp and 22—z0 to be small, then, since

i- u/Sr is independent of direction, we have
•llô=/l£!Wi£5)=/, ( } approx.
-o -0
ltg~^° = (-uJ =/' tr0) approx.

“2 — -0 r2 ~ ~0
Hence Lim Lim (*.)-£. .(I)

- -1 “0 ~ - -2 “ -0 ( "
V *0 “2 r“0
D 'A
Since/' (z0) is not zero, we may write n as /<e ,
Hence we pet
ibi
Lim a* = RJX'
n iff i
whence Lim (»/• j — and Lim -1 —
/. f’. Lim '/‘j — A + ai — arg_/ t-o)"bai»

/. <>. the cutve Tj has a definite tangent at n0 ; and makes an
angle a, + A with the real axis. Similarly, it can be proved that
r, has a definite tangent at h0, making an angle oc2 + A with the
same real axis.
Hence we conclude that the curves F F2 intersect at the
same angle as (Ire curves C, and Ca ; ai d the angle between the
curves // F., has the same sense, as the angle between Ct and C,
i. e. a regular function w-f(z) for which /'(r)^0 represents a
conformal transformation.
And this establishes the similarity of the corresponding
infinitesimal elements of the two planes, though corresponding
unit areas of the two planes are not similar.
To obtain one figure from the other it must be rotated
through the angle A = urg {/' (r0 } and subjected to a certain
magnification
Conformal Representation 73
Lim pJ = JR = |/' Wl.
And it is quite obvious that the magnification is indepen¬

dent of the direction, but it varies from point to point.
Aliter. Suppose that (z), where J (z) is a regular func¬
tion, maps a region D of the z-plane isogonally on a certain
domain D' of the tv-plane Let C be the regular arc z = x (ff-j-iy (t)
in D. Then the transformation is w=f [x (/H iy (0}, so that C
is mapped on a regular arc, say r. Let us now suppose that <}>
and 0 are the angles which the tangents to C and F at the points
of the same parameter t make with the real axis. Hence we get
</> = tan-1 =arg {x (t) + iy (0}}

•V
and 0 = argjj^/{* (t)+iy (r)}J
=arg [/' (z) {x (/)+/>’ (0)]

= </>+arg/' (z)
i. e. 0-0 = arg/' (z) and hence depends only on the point itself
and not on the particular curve through that point, /. e. indepen¬
dent of directions of approach to that point. Hence it implies
that not only the magnitude of the angle of intersection of two
curves but also the sense of rotation is conserved.
Now we find the mapping of the regular arc C and T. Let
s, a be the lengths of C and r respectively upto the point of
parameter t. Thus, we have
and (z) !•
h— £-i rw\\%\
Thus it has been shown that a small neighbourhood of the
point z0 is mapped conformaily on a neighbourhood of the point n 0
by the transformation \v=f(z) and this, correct to the first order,
is obtained by a magnification in the ratio | /' (z0) | : 1 and a rota¬
tion through the angle arg/' (z0).
Note. It should not, however, be concluded that »t’=/(z)
maps the domain D of the z-plane isogonally on a domain T of
the H-plane if/'(z) is non-zero and finite at each point of the
domain D. For, although a small neighbourhood of each point
74 Complex Variable
of the domain of the r-p!ane is mapped isogonally, but it is not

necessarily true that the mapping of the whole domain D has a
one-to-one correspondence with that of -T. It is quite evident
from the following example.
Consider
The domain D is defined by
< | 2 | < 4.
The corresponding domain, in iv-plane is
1 < | w | < 16.
The function has a derivative 2z which is finite and non¬
zero in the domain D. But the mapping is not one-one. Particu¬
larly we see that
r = 3 and r = — 3 are both mapped on u*=9, i. e. a neighbour¬

hood of »’ = 9 is the representation of the neighbourhood of 3
and —3, both lying in the domain D, i. e. the domain D of
2- plane is mapped on an overlapping region T of w-plane. Hence
the transformation is not one-one.
The sufficient condition can also be stated as below :
Iff(z) be a uniform analytic function of z in a region D
of the z-plane, then the relation u-=/(r) determines a conformal
transformation provided that f iz0)r£0, z0 being an interior point
of D.
Example. Define conformal transformation, and deduce suffi¬
cient conditions to be satisfied by f (e), so tint the relation w=f(z)
may define a conformal transformation on the plane in the neigh¬
bourhood of a point zQ in the plane.
(Agra 1943, 47, 50, 60, Jaipur 1959)
Condition is necessary. We are given a pair of differentiable

relations of the type
u=x (x, y) ; V— v (x, y)
defining a transformation from (x, y) plane to (u, v) plane. Does
there correspond an analytic function w=f (z) ? (Agra 1950)
Proof. Suppose da and ds are the elements of length in (u, v)
and (x, y) planes respectively.
Then do2=(dU)2+ (dv)2,
ds2=dx2+dy2.
.3m ou ,
We know that du=~d<-t-~ dy,
dx dy
, 3m 3v ,
)ydy-
3v
{© - ©■mgshen*1
+J
= £ dx* + 2F dx dy-\-G dy2,
©£+££)**
where E=(du Y+rtL'V -r--8" av 0M
vs* J + W ’
.a
a* - a^+s;^ 3*
W+ ‘
do: ds is independent of direction if
fi F G
i = tî =A2 ( say ),
/3v -
V3*J +
and ,+©‘=^^
writing 'Sjn0:m?r'r«;St eqUa,i0D6 are “‘MVdS

substituting these valued Z ^
.• (cos a cos 0 + sin « sin 0j = O
H T cos(«-/3)=0 or a —/3 = + 7t/2
,hC f°Ur first order Partial derivatives satisfy one of

76 Complex Variable
the following equations :

(a) ux — Yu. vx=—uv, or (b) vy,vxs=uy.
The results given in equation (a) are the well known Cauchy-
Riemann equations, and we can write
u + iv =/ (*+/»=/(*),
where/(2) is a regular function of 2.
We see that the equations (b) reduce to (a) if -vis substi¬
tuted for v, that is. by taking the image figures found by reflections
in the real axis of w-plane. Thus it is evident that (b) corres¬
ponds to an isogonal but not a conformal transformation. Hence
it follows that the only conformal transformations of a domain
in 2-plane into that of it-plane are of the form ie=/(z), where
f{z) is a regular function of 2.
Note. The relations (a) and (b) can also be obtained as
follows. From (2), we get
— = - X-~ = A, say.
v„ uv
The from (1).
(A2 - 1) ['V* + vtf8J — 0 i.e. A = + I.
Aliter. The condition is necessary.
Here we will prove that if the angle of intersection of the
cur\es T,, A is the same as that of C1 and C.2 both in magnitude
and sense, then f (z) is regular, ie the Cauchy-Riemann equa¬
tions are satisfied. .
I el r —,y (t) + iy (t) be any regular arc in the r-plane and its
image curve F through (//, v) be given by
11 •*»/{* (r) + />’ (03.
a = u {x (/),>• (/)>-!# (n,
v = v {.V u).y (/)> == v (/).
For the curve C to have a definite tangent at any point lx. y)
dx and l-y should not be zero. Hence we have
both
at ut
du dudxbudy_ dy
-dTtxdi + ,)v d,-"‘
dv 3v dx 0V dy__ ,
dt +"“ dt
r/A’ ^ dy
...d)
dt= dx 7/7 + dy 7r ~~V e dt ^ * di *
We also suppose that
J —, 1/ v. i=0.
...(2)
V w
which means that the neighbourhood of a point zQ is mapped in

a one-to one correspondence on to a neighbourhood of w. From
(2) it follows that curve F has a definite tangent as both ~ and
(h
-r are not zero.
dt
Hence if 7^0 and we get

dt 7 dt ' ’
dy
dv
.-(3)
du
W, +
dx
Let 'A, <A' be the angles which the tangents to the curves C
and T make with the real axis. Hence we have
tan ■/•'=11 1 ”tan<4

wx.+tan <f>
tan (f>' - tan </»
tan ('A' — '/-)=
1 + tan 'A tan </>'
__ (v.r+rtf tan «A)/(i/, +1/„ tan -A)-tan <A
1 + tan <£ (v,+ v„ tan <f>)/(urA uv tan •/>)
_ (vr-f tan <f>. —tan -A (ur+u,j tan <A)
(Wri-t'v tan «A)+tan *A tan <A)
__ ~ P_ (w* + £«»)
(Wx+^l + P (V* + FV'
where /> = tan «A.
Now let us suppose that'A'- *A is constant for every corres¬
ponding pair of curves C and F and put
tan (•//-</.) = /
i e { Av*+PvJ-P (Ur+puj
•u* + puv\ + p {vx+pvv) '
for each and every value of i.e. independent of direction,
We get
F2 (/>’v4 */„)+/> (/wy+/»'x-vtf) + /ue- ^=0.
Hence the coefficients of /72, /? and the constant should sepa¬
rately be zero.
/v„ + "v=0, ...(6)
/ ("v+»•*) + "*-0, ...(7)
lux-vx = 0. ...(8)
Subtracting (6) from (8), we have
l (Uz-vv)- (vx+uy)=0. ...(9)
78 Complex Variable
Eliminating /, which is not zero, from (9) and (7), we get

(wv+v,)2+(wx-rv)2=0, .(10)
i.e. */y+vx = 0,l
Ux - v„ = 0. f
• 4 .(11)
Hence, we get
ux= - vx, ux=vv.
i. e. the Cauchy-Riemann equations are satisfied and if 1=0, we
get from (5)
p*uv + (ux-vv' p-vx=0,
which being an identity gives
r/v=0 = vx and ux=vv.
i. e. again the Cauchy-Ricmann equations are satisfied. It means
the function / (z) is analytic.
It can be easily verified from (5) that if the Cauchy-Riemann
equations are satisfied, then
tan (•//-*) = Q^ = tan fare/ ' (*)],

which is constant at any point r, being independent of the
direction.
3*3. There are transformations where the magnitude of the
angle is conserved, but the sign is changed For example,
w=2 = (x— />•),
where 2 is conjugate of r. It replaces every point by its reflection
in the real axis. Here the angles are conserved but their signs
are changed, and this is true for every tiansformation of the type
w=f0s), ...(1)
where f (z) is an analytic function of r, as it is the combination
of two transformations :
(i) ; = 2, (ii) w=/<a
In (i) the angles are conserved, while their signs are changed,
ai d in (ii) both the angles and signs are conserved, i.e. in the
ghen transformation the angies arc conserved and their signs are
changed.
3 4. f' (z0>— 0 In § 3 2, we laid down the restriction that
/■' (ro)=£0. Let us now suppose that f‘(z> has a zero of order
/; in the neighbourhood of the point z0; we know from Taylor’s
theorem,
/(--)=/i-.H (--„)+•• •
o) yyiyr + • • •
When/'(z) has a zero of order n at r0./'(z0),/" (z0\....

fn (z0) are all zero and /n+1 (z0)=£0.
Hence in the neighbourhood of z0.
/(*)=/U0)+« (r -z0)n+l-f..
Thus, we get
>i’i-H>0=a (Zl-z0)n+l+...,
= ! a | r"+l exp. {(rt-F l) 0,+A} i f..

where A=arg a. Hence
lim f/»1=!im {A + («-fl) ^} = A + (/z-t-1) «x.
Similarly, lim f2 = X+(«+ 1) «2,
i. e. lim f^2-</>,)=(// + |) (**-«,). ...(I)
Hence the curves r,, r2, still have definite tangents at u0,
but the angle is given by the equation (1). And the linear
magnification lim ~ is zero becaus /' z0)=0. Hence the con-
formal property does not hold at such a point.

For example, consider »v = z”, n > I. Here the arg w=n arg z
and the angle between the line which joins the origin to n0 and
the-f ive real axis is n times the angle between the line joining
the origin to the corresponding point z0 and positive real axis in
the z-plane. Hence the corresponding angles at the origin are
not equal because at z=0, --—0
dz
The points where ^=0 or oo are called critical points and

.fdw
11 dz^0 P°*nt *s called an ordinary point. These critical
points play an important role in the transformations.
Note.
We have proved that an analytic function w=fiz),
provided f {z0)zfi.O, gives a one-to-one continuous representation
of a certain neighbourhood of the point z0 in the z plane to a cor res¬
ponding neighbourhood of a point »v0 in the w plane which is called
a conformal representation.
It can also be put in another form, by saying that such a
transformation tranforms infinite small circles of the z-plan into
infinitely small circles of the w-plane.
3*5. Superficial Magnification. It has already been discussed
that the linear magnification at any point z0 in a neighbourhood
80 Complex Variable
of the corresponding point m>0 in a conformal transformation

»»’—/(z) is obtained in the ratio of j/' (z0) |: 1 it being supposed,
that/' (zo)y±0.
Now here we have to prove that the superficial magnification

is / (7o) I2 : 1* Let A be the closed domain of the w-plane which
corresponds to a closed domain D in the z-plane.
Area A of A is given by
clu <fv= fJ Jf D d (x, J-)

i/.v dr.
We know that
3 (u, v)_9t/0v _ 9v3w
3 (v, y)~~dxdy dxdy’
But ux—vv, vx=—uv are the Cauchy-Riemann differential
equations. Hence
|^=â+v,*=i ur+iVx i*=i r & i2

and so A=^ f (z) \* dx dyt
which proves the theorem.

36 Equation of Circle.
Let us suppose that A, B, C are the three real or complex
constants and a, b, C their conjugates, z is a complex variable
and ? its conjugate. Then the equation,
(A + A) z2+Bz + B?+C+C= 0 ...(1)

represents a real circle or a straight line provided that
BB > (A + A) (C + C) ...(2)
For, if we write
A —a + ia', B=b + ib\ C=c-f ic\ z=x + iy.
then A + A —2a, C+C=2c, Z2 •=(*+/» (x - iy)=x*+ys.
Hence (1) is transformed into

a (x2 -f >’’) + bx -b'y + c=0
which is a circle. It will reduce to a straight line* if
A + A = 2a=0.
ma.z~c+Bz-\ (C-fC) = 0 will be a circle of infinite radius or straight

line through zôo and zif and only if a=0 which can be verified if
we divide by zz throughout and put z=oo and z — oo which gives «=0,

Conformal Representation M
If r be the radius of this circle,

r2=b2/4a2 +b'2/4a2 — c!a ...(3)
and the circle will be real if radius is real,
APIA'S c
i.e. if —> — or b* + b's > 4ac
which is the same condition as (2).

Note. We have already found the equation of a circle in
Ex. 15 page 22.
We have seen in the previous theorems that the infinitely
small circles of z-plane can be transformed to infinitely small
circles of the iv-plane by the conformal transformation of a
regular function w—f (z) provided/' (zJÔ. There exist, however,
non-trivial transformations which are true of finite circles. The
transformations are discussed here.
3*7. The Bilinear Transformation .
Consider w=a:Jr^
cz+d * ...d)
which is equivalent to
c»rz4 dw — cz—b — 0
where a. 6, c, d are complex constants. It is evident that (2) is
linear both in w and z. Hence one and only one value of z
corresponds to each value of w and conversely. It is due to this
fact ‘hat the transformation (1) is called the Bilinear Transform.!-
Hon or Mobms Transformation after Mobius (1790-1868) who
first studied the same. J
We shall suppose that ad-bc^0 which is called the determi¬

nant of the transformation. If the determinant vanishes then,,,
is merely a const, as shown below. S. ‘ben it-
The„Lwehavend ”'2 ^ ““ C°rreSP0I>di"® to z, and z, in (I).
- W, = ■ah±2=(ad—f-c) (z2 - z.)

. r cz2+d czj-fc/ (czo-f-c/i (cz, 4-d)
means'either °w ^consôr "’lean’ °f ** ^

jj . meaningless provided ad - bc=0.
to w forndmerent'valws°of z.See ‘hat<U associales di,Tcren< va,ues

3-8. Critical points. Equation
cz-\- d
82 Complex Variable
and its inverse

— dw+b
z=
cw — a
show that if the r-plane is closed by the addition of r=oo, every
Mobius or Bilinear transformation is a one-one transformation of
whole closed z-plane into itself.
If ad-bczpL0 and c^f0, we find that to each value of zzf - d/c
a value of w is associated, and to each wzfzajc there is a corres-
ponding value of z. The correspondence of w and z is one-one
except these points.
az+b
From ...d)
W cz-\-d'
dw _ ad—be
we see that ...(2)
dc (cz+d)2*
From (2) we find that
dw d
— =oo if z --
dz c
and zero at z= oo, where the conformal property does not hold
good and these are the only critical points.
If we extend the definition of the conformal mapping in the
following way, these critical points cease to be exceptional points.
A function w=f (z) is said to transform the neighbourhood of a
point z0 conformally into the neighbourhood of u’=oo, provided
the function transforms the neighbourhood of z0 con¬
formally into a neighbourhood of £=0. Also w=f(z) is said to

transform the neighbourhood of z—oo conformally into a
neighbourhood of >v0 if w=<f> (t)=/(1/t) transforms the
neighbourhood of t=0 conformally into a neighbourhood of
H’o• Here in the above definition wQ may have the value oo.
It may now be said in view of the above extensions that every
Mobius transformation gives a one-one conformal representation
of the whole closed z-plane on the whole w-plane, i, e. it can be
said that the transformation is biuniform for the complete planes
of z and w.
3*9. Resultant or Product of two bilinear transformations.

Suppose we are given any two bilinear transformations
azA-b
w
cz4- d .(1)
such that ad—bcz£0.

and r=aiw+h
CyW + di ...(2)
where axdx-bxcxÔ.
The equations (1) and (2) associate to every value of z
a value of Eliminating tv from the above equations we get a

relation
a az+b b
*;= 1 cz+d~i~-(gQl-\-blc) z-\-bxd \-atb
...(3)
c, az+b | d (C\<*+diC) z t dxd+cxb
1 cz+d^ 1
with the determinant
(aax + bxc) (dxd+cxb) - (bxd+axb) (cxa-\-dxc)
=aaxdxd-\-bbxccx—bxcxad~axdxbc
.. . . . =(Vi- Vi)
which shows that the transformatton given by (3) is also bilinear.
auM1.,mear trans/orâtion is called the Resultant or Product
ot the bilinear transformations (1) and (2).
3*10. Bilinear Transformation as the Resultant of Bilinear
transformations with Simple Geometric Imports. (1944, 1957)
Given that »=g±* ad-bc^0 ...(,)
is any bilinear transformation.
If cz£0, we write
... a ,bc-ad 1
_Z-d
c c *~~T; ^ = so that z ...(2)
the transformations (2) 'hat (° '* <° the succession of
W= Z+a; \v= pZ> W=l{z.

(i) Translation w = z-f-a.
Let Px and A be the points z Qt4^)
and a, where A is a fixed point and
Px, a variable point.
By the addition law of vectors Q
is the point
iv=z+a,
where p?q=2
84 Complex Variable
Hence we obtain the image Q of any point by subjecting the

plane to a translation which is defined by any number a.
(ii) Rotation and Magnification.
Px and B represent the points
r and p. The point Q which repre¬
sents w is such that
/-QOPx= f_AOB=arg p
OQ=OB. OPx,
Hence we find that Q is ob¬
tained from Px on rotating OPx
through an angle which is equal to arg P and a magnification of
OPx in the ratio | p |.
Or, analytically, we have
w = pz
which gives arg u=arg p + arg z l,
and . ^ |
w (=| p | z | J j
from which it is concluded that w=pz corresponds to a magnifi¬
cation as well as rotation. [If p is real and +ve there will be
simply a magnification and simply rotation if j p J = I J.
(iii) Lastly w —[Inversion in the real axis and the unit

mij
circle.]
If !"! = /• and arg z = 6, then taking modulus and argument
- 1
of —, we get
arg ir—arg 1 - arg
What see that n ==—, is the resul-

a*
tant of two transformations of the form.
Z=2 and »v=2*

The former is the reflection in the
real axis while the latter is the reflec¬
tion in the unit circle. Hence to pass
from the w-figure to the r-figurc we
invert the former with respect to the origin of the w-plane; with
unit radius of inversion; and then construct the image figure in the
real axis of the w plane.
Conformal Representation
Conclusion. The sequence of transformation (2) of § 3*10

consists of the combinations, we have first considered, of which
the important is that of inversion which effects the shape of
the figure. As the inverse of a circle is a circle or a straight line
(circle of infinite radius), we conclude that the Mobius transfor¬
mation transforms circle^ into circles ; which we prove in the
following article.
3 11, Every Mobius or Bilinear transformation transforms

cue es into circles or a straight line into a straight line. (Agra 1951)
The equation of a circle or a straight line is given by

/1z3+5z + B3-1-C=0 ...0)
wêre ^ Rb, A and C being real. ..(2)
Now the bilinear transformation is
az-\-b
w—
cz+d
and iis inverse is
__— dw-\- b
cu' — a , (ad -bc^zO)
*ênce substituting for
(^)
+i(#5rV-°
or / W/lw -bb-bdi- uw) f fi (_*lt -+~adw+A- -_ b-}
A-B ( - cdww-\rcbw +ad\v — ab)

_ "fC (c c\vw — caw - acw+aa ) = 0
or (Add -BUc-Bca+Ccc)+ w (- Ahd+ B~ad+~Bcb _Cf-}
+ w ( “ Ahd - Bbc+Bad - Cacj
which is of the form + ^ + Caa ~ Dba -*£)-0
a lvw+pw+pn+y^
••.(3)
86 Complex Variable
oL=Add -\-Ccc—Bdc - Bed

($= — Aid+Bad+Bcb - Cca
/3= - Abd + Bbc -\- ~Bad - Cac
y=-Abb + Caa - Bba — Bab
Clearly a, y are real ; and the coefficients of w and w are
conjugate complex numbers.
Also /3/3 - ay=BB (aa dd+bbcc -adbc—bcad)

— AC {aadd-f bbcc — ad be - bead)
the other terms clearly vanish.
= (BB—AC) (aadd+b~bcc -ad be -bcadft
z=(BB - AC ) (be —ad) (be—ad) f ...(5)
= (B'B-AC) \bc-ad\* ^ 0
Hence equation (3) represents a circle or a straight line,
since it is of the same form as (1).
3*12. Inverse points with respect to circle.
If M be any circle with centre O
and radius a, then the two points R
and S collinear with O, such that
OR.OS=a2, are called inverse points
with respect to the circle.
Let the points O, R, S be ky zt zx
respectively, then by definition,
arg (zi-£)=arg *) •••(')

The second condition of (1) is the condition of collinearity
of the two points with O, which can also be written as
arg (zx—k) = - arg (z-k),
since arg (z — k)= arg (js—k)t
i.e,, we have arg {z1—k) = — arg (3—k)
or arg (Zj -A) (3 A)=0. ...(2)
This equation gives that
(rx—A) (3 — A) ...(3)
is a positive real number.
Also first condition of (1) can be written as
j (zl—k) || (z-kj\=a2.
The conditions (3) and (4) are given by one single condition
(z, -k) '2 -k)=o2. ...(5)

The equation of the circle M is given by
Azjs -f Bz -f- i?2 + C=0, ...(6)

which may be writen in the form
5\/
BB-AC B\
M)K-)= A2
where A and C are real quantities, i. e., (6) represents a circle
.. ~B ,
with centre-^ and radius-—.
.. £5 -^C
Equation (5) can then be represented by
B\ BB-AC
/l2
...(7)
or Azl-2 + Bzl + Bis+C=0. ...(8)

Hence we have got a relation between z and its inverse zt
from (6) by substituting zx for z and 2 unchanged, /. e. from (8),
we get
_ fl2+C ...(9)
Z,_ /fz+S‘
Now we prove an important theorem.
3*13. The bilinear transformations transform two points which
are inverse w. r. t. a circle into two points which are inverse
with respect to the transformed circle.
It is quite clear from the preceding article that if z, zx are
the inverse points with respect to the circle (6) of § 3*12, then (8)
of the same article is satisfied.
Now we make use of the transformation
az+b
1V= ——,
cz-\-d’
If w and M'j be the transformed points, we have
_ ~ dwx -f- b —dw-\-b

*i = ,2
cw, - a civ—a
If these values are substituted in equation (8) of the previous
article, we find an expression of the form
*wlw+(Swl+pw+y = 0,
88 Complex Variable
where a, (3, 0, y are the same as in (4) of § 3*11.

In fact, we get (3) of article 3-11 if w is replaced by wu which
is the condition that w and wx are inverse points with respect to
the tiansformed circle
a»vw' + j8w' +/Siv + y* 0.
Hence the theorem.
The equation (9) of article 3 12 can be written as
-Bw-C
'v-s;rx- aw + B
in which first is the reflection while the second is a Mobius
transformation. In first the angle is preserved but the sign is
changed, and the second is a conformal transformation. Hence
the inversion is the isogonal transformation, but not conformal.
As we have seen that the inversion is an isogonal but not
conformal ; it can be concluded that the result of two or an even
number of inversions is a one-one conformal transformation
because both the sign and the magnitude of the angles are
preserved. That is, we can say that the successive transformation
of even number of inversions is a Mobius transformation.
3*14. The equation z’p'=A, where A is a positive number

z-q
represents a family of circles for every member of which p, q are
inverse points end conversely.
Let r0 be the centre and a the radius
of the circle whose inverse points are
p and q. Hence if
• i
p—zQ |=/ i. e. p = zQ+lj , ...(1)

a2 a2 .,
then <7 ro \~~/ i'C.q—z0 + . t'
/
...(2)
id
If r=20-f ae be any point of the
circle, then
Z-P - ~n
—'
•
=—q\ t> 1
1
1
z-z0-(afl) e }
id . ty . id . i<f>
ae —le 1 ae -Jer
aeld — (a*//) e^ a , ^ it>
le -ce
l | (a cos 0 — l cos </»)-+-/ {a sin 0-— / sin '/>)
<7 (/ cos 0 - a cos 1>) + / (/ sin 0 - a sin )
I / pa + /g-2a/ cos (fl - <A)~1

a V L
2-H«2- 2a/ cos(0 — </*)J
/
where | (a cos 0-/cos *) + i (a sin 0-1 sin <f>) |
= cos 0 — 1 cos tf)2+(a sin 0 - / sin '/>)*}
= v/fa2-: /* -2.7/ (cos 0 cos </>-fsin 0 sin <£)}
= \/{a2+/2 - 2a/ cos (0 - '/»)}.
Similarly for the denominator.
z-p
Hence — — which is therefore the new form of the circle.
z-q
Conversely, any equation
z ~P
z-q ^ =klt ...([)
represents a circle with respect to which p and q are the inverse
points.
Equation (I) can be written as
(z-p) 2
(z-q) (is-qi 1
or Z?- p? - pz + pp = kxz (zs-q? — qz + qq)
or (l-kS)_z (p-qicl*)-S(p-qkli)+pp-qqk*=0.mm{2)
To show that p, q are inverse points with respect to the circle
(2), we replace z by p and 2 by q. (§3 12] This gives,
pq (l-k*)-p (p - qk*) -q(p- qkp) f pp-qqk* = 0,

which is identically true. Hence p, q are inverse points with res¬
pect to the circle (2).
315. Another proof of theorem of § 3*13.
In a bilinear transformation, a circle transforms into a circle;
and inverse points transform into inverse points. In the particular
case in which the circle becomes a straight line, inverse points
become symmetrica! about the line.
Given :
Zj-Pi = k
\Z-qx t
...G)
a circle (or a st. line for k= 1) wit h pv and qx as the inver
points.
Let -_t
w= a z+ft z__# ow—ft
yz-\ 5’ — yw>4 a*
If we substitute the value of z in (1) we have
?‘\v—(3 - px (a—yit>)
90 Complex Variable
w (8 -*-pxy) -p—pxa.
or =k
w (s + <7iy) - 0 “ <h*
w - /? + />,«
= k S + giY
P+J?xCt 8+Ay
u—
8 + qiy
which is clearly a circle whose inverse points are /?+A« and

5 +AY
P-\-qyo.
corresponding to the inverse points px and qx of (1).
8 + 9i y
3*16. Special Mobius Transformations.

1. To fir.d all the Mobius transformations of the half plane
/m (z) ^ 0 into the unit circle | w | ^ 7. (Agra 1961)
»r, ~ which are inverse points with respect to the unit circle
ii’
in the n -plane clearly correspond to the inverse points z, is which
are inverse points with respect to the real axis [see § 3 13] Parti-
culary the origin >c=0 and w=oo , which are the inverse points of
| iv >= 1, correspond to the conjugate points in the z-plane.
Suppose that
az+b
)v = ——
cz-yd
is the required bilinear transformation.
Obviously cÔ; otherwise the points at infinity would corres¬
pond. The points »v = 0, iv=oo in the u -plane are the correspon¬
ding points to z= — b/a, -die in the r-plane ; but as these are
inverse points with respect to the real axis, we may write
_± = ±Z-=P ...(2)
c Z'P
c*+4
When z is real and | w |= 1, we see that
z-p
= 1 provided = 1
z-p
or in particular 2=0 must correspond to a point on the circle
| w | = 1, i.e.
(fX-f)HT = 1.
id
Now putting c=ce , where 6 is real, we get
id z-p
\v=e ...(3)
z-p
As it is obviously true that z=p gives w=0, p should be the
point of upper half plane i.e. Jm (p) > 0.
Hence with this condition (3) is the required transformation.

Conversely. It can be shown that the transformation
Z~P
where/) is any number, does map / (2) > 0 on the circle | id | ^ 1.
Clearly, we have
m -1 = eWe ~18 U~p) (~ . ,

(z-p) (2-p)
= Z~ + PP -pz-pz —Zg-f PJ2 — pp +pz

\ Z-P l*
_»2-2) (p-p)
!2 -p I2
or ww - 1 = - 4 -- ^ I,n
u.hu ■ \(z-P)\2
which shows that the transformation (3) maps /„, (Z)=0 on
I »v I—1 and lm (2) > 0 will be mapped on to | w | > 1 0r
I .V | < 1 according as Up) < o or /(p) > 0. but / („) > 0
as we have proved above. Hence / (a) > 0 lor any value of - js
mapped on I < 1.
2. The general linear transformation oj the half plane
R (z) > 0 on the circle | w | .< 1 is
92 Complex Variable
>' -e'A ~f , (a) > 0)}.

Let us suppose in the same way
as before that
az-\-b
h=
cz-f r/ ...d)
is the required transformation.
The points z and —s symmetrical
about the imaginary axis correspond to
the points w and l/w which are inverse
points vv. r. t. the unit circle in the
»»’ plane Clearly cef.0 ; otherwise the
points at infinity would correspond. Particularly the points tv=0
and iv=co in the tv-plane correspond to
2= - bja, —djc.
Hence we may write

- b/a = xt — d[c = - a.
Thus we get
a z—x
tr=
C * Z-f-a’
Clearly the point z = 0 must correspond to a point of the circle

tr ! = 1, i.e..
-a ' - X
i 1 a_
I c : a , "1 c
.
or putting a = ce iX , we may write
i-\ 2 — x
w—e
Z+ a *
With the same argument as in Ex. J, z=a gives tv = 0 and
hence R t«) > 0 i.e. a must be a point of the right half plane.
Conversely, tnr — 1 = - (a-fa)

i :+a *
= _4/?(r) R (a)
| r-f-ot |2~*
The left hand side of the above equation is negative, i.e.
! “■' I < 1 if R (-) 7* Hence R (z) ^ 0 is mapped on | w | ^ 1
and R (r) < 0 on ; tv j ^ 1 as before.
3. find all the Mubins T, ansformations which transform
| z | < 1 //; the z plane into the unit circle | tr j ^ 1 of the plane.
(Punjab 1958j Agra 1952, 44, 60 j Luck. 1958)
Let the required transformation be

az + b
V cz-\-d* •••(!)
which transforms | z | < 1 into | w | < 1. Now u=0 and iv=oo,
the inverse points of the circle | >v | < 1, are the transforms of
b d
z ——— . —- respectively.
TT b d .
Hence are inverse points of [ z | = I.
Put - a =« ; then --^=

* c x
, w-herej a | < 1. i-
Hence we can write (1) as
— - z~a a~* z — CL
~c r=~c • rz-i • ...(2)
a
The point z=l corresponds to a point on the circle | iv = i

Hence 1=| iv |=|0"a I"a| 1 ' '[ V | 1 —a | = | i- 1 1
C x- 1 c
a* id
ie. u~ =e° , say, where 6 is real.
Thus the required transformation is

id z—a
w—e ~-
»z - 1 ‘ ...(3)
It can clearly be seen that the transformation (3) maos
21 = 1 on | w |=1. Weseethat ' P
id z~ a
IV
aZ - 1
z—a
az - 22 f * 12 \ ~ 1, i.e. 22 = 1J
z—a - l
"a-2 I 1 z
=1.
Now we shall examine the corresnnnH™,^ .. •
and the interiors of the circle | z | = l. ie cxtenors
We get
iO Z — C/.
IV iv -1= -r—r e~ 6 * ■ i
a2__1 2a —1 1*
(z - a) (2 - q) __ j
(a2- 1)~
94 Complex Variable
_ ~ - otz-f-aa - oca zg — 1 -f- as-j.az

(az — I) (a5— 1)
(25- 1) (1 — aa) __ (zs - 1) (1 - aa)

= (az- 1) (az - 1) ““ | (az - 1) !2
_(1 z j2 ~ i) (1 —I a l2)
“ | az— 1 I2
Hence (4) shows that the transformation (3; will map

j 2 | < 1 on I iv | < 1 conformally or on I w | > 1 according as
| a | < 1 or | a I > 1.
Alternative. Put z=re‘ , a =/?e , where p < 1, r < 1 ; then
2- a ;2 - I az —1 *
»• i2 - 1 =
l«z-l i2
/A 0
re — Be |»- | j8r.e,(A“W -1 I*
Numr-(r cos A -(3 cos <£)2-f-(r sin A— (3 sin </»)*

-{0r cos (A -- 1 }• - j8*r* sin2 (A - </)
= r2-f/32 — 2r(3 cos (A - 0)-(j8*r*+ 1) + 2/3r cos (A—«/>)
= r2 + p*-p-r2-l
= (r2-l)( 1-/32) < 0.
Hence »v | < 1, i.e. the interiors correspond.
4. The general linear transformation of the circle j z | ■=• p into
the circle j »»• I = (>' is w=pp'e ~-- , (| a | < p).

az—p*
(Agra 1956, Optional Paper)
Proceeding in the same way as in example 3, let us suppose
that the transformation is
az + b ...
W~cz + d•
The points n =0 and u’=oo which are the inverse points
with respect to the circle w\ = p' correspond to the points
z= — — : —— ; ie. and —— are the inverse points with

a c a c
respect to the circle | z =p.
, dp2
Put - ^ = a, where j a P. then-- .
C a
Hence we get
__ a z—a act / z—a \
...(2)
c * z - pl/a e * \xz- p2'
We require | w | = p', when | z |—p, i.e. Z3=
Hence p' = u, |_|oa z —a ax ! z-a

■ c * aZ—Z3 C 1 x —3
a<x 1
i.e.
c z ~p
or ax
7 =pp.
, i\ . ^
Putting —:= pp e in (2), we get
c
, /A 2—*
w-pp e - ...(3)
az — p2
which is the required transformation.
We show in a similar way that the interiors correspond.

HMD—
W\v —n'l
p'i= ^ 2 „' 2
— p2p'2 (2 — a) S
if-TLzi ~—~ a)
-P'2
(»2-p2) (a3 — p2)
_ P*2 [aa —p2) ( -Z3-f p2)]

|(iz-P2)l2 *
We see from (4) that | w \ < P\ when | z | < p and | x | < p,
16‘ * e *nteri°rs correspond. It is obvious that if p'=p= 1, the
results agree with that of example 3.
5. Show that the region \z—a\ ^ R is mapped conformally on
\»\^lbvw=._ R (z-cl «/ , „ .
y R* -~(Z-a) (c-a) *e * iere C ls t/,e P0,nt n/,‘c/l
is transformed into the origin. (Delhi 1959)
Or
Show that there is a unique function w=f (z) which maps
(z a) ^ p conformally on | w | ^ I and also transforms an inte¬
rior point z=c into the origin and a given direction at c into the
positive direction of the real axis.
Let the reqnired transformation be
w= <^+/3
...
Vz+S
96 Complex Variable
« z+ft/q
y (z+8h) *
z=—(3Joi. gives iv=0. Put —/3/a=c. As we know that the

bilinear transformation transforms circles into circles and inverse
points w. r. t. the circles are also transformed. For the point
z—c the inverse point with respect to the circle \z—a\ < R is
R2
a-fr—~ [see (5) of § 3*12] which must correspond to»v=oo,the
inverse point of w=0 with respect to the circle in the w-plane.

Clearly r=-6/y is the point which transforms to H’=» in the
tv-plane.
Hence
Substituting these values in (1),

...
a z—c
y * \z-a-R21 (c-a)'
We require | w j=l, when | z-a | =R,
i.e. when (z - a) (z—a) = Rs.
Substituting for R2, we get
a Z—c
1 = | w| =
y 1 _
z~a~JZ*(?-">(*-*)
a (c-a) (z - c)
y (z—a) (c-j?)
1 « 1 | c -a | a | c—a)
| y | | z-a | “ y R
a R id
i.e. ~—e
y (c-a)
Conformal Representation V/
Hence the transformation becomes

_R id z—c R (z—c) id
c—a'C . a -K2 "(z-a) (c-a)-Rz 6
c -a
R (z—c) _ /a
R2-(z-a) (c- a)'€
where a=0-jr, a real quantity.
Aliter. First let us find a bilinear transformation in which
the centre of \ z -a | ^ R is transformed into the origin in the
z'-plane and the interior point c of the circle in the r-plane into
c\ the interior point of the circle \ z' | < 1.
Since the centres of | z—a | < R and | z' | < 1 correspond,
the inverse points of the centres must also correspond. But the
inverse point of the centre of a circle is infinity. Hence infinities
in both the planes correspond.
cf.z-\- (3
Hence the transformation z' = — rk 8*ves y*=0 and - Pi*=a,
yZ-i-b
such that
(z-a).
a 1 JL I p
We require | z' (Z-fl)
T | S 1 R
ot 1 a 1 M
which gives
R r R
Hence the required transformation is
. z—a M
Z=~R e >
where the inner point c of the circle | z - a \ ^ R is the corres¬

ponding inner point c' of | z' | < I, given by - e/A t
Now the second part of the problem is that 1 z' | < 1 may
be transformed to | w | < 1, such that c' is transformed into the
origin.
With the same procedure as in equation (3) of Ex. 3 above

we get
2 -c in
1 -c'z' e ' [Here fj = 0 —»r.]
Substituting for z' and c' in terms, of z, we get
98 Complex Variable
/z-a c-a \ iX iu
■ -m
R (-- c) <_
R- — (c—a) (z-a)
where x is the real constant.
It should be noted that, by suitably choosing a, we can make
any prescribed direction at z=c correspond to the positive direction
of the real axis at w = c.
6. Find the radius and the centre of the circle in w-pfane which
corresponds to the real axis in the 2-plane, where
a .
ze —i
tv
a
z - ie
x being a real constant.
**ca — |
Solving w=-r~a for z, we have

z - it
-1 \ w-ea )
.(1)
or x+<>.=/ (w^)~i
\ »v — e J (w — ea) (w — ea)
i [wwea-\-ea— w— e*aw]
...(2)
[inr-ea (w+wH-e2®]
The real axis is given by the equation _>■ = 0, i.e. the real
part within the brackets on the R. H. S. of (2) must be zero,
i.e, wwea-j-ea - u - ue2a = 0
or H’H’ — uea—ue~a-\-1 = 0
or m2-J- va — 2u cosh a-f-1 =0
or (r/-cosh a)2-f-v2 = cosh2 a—l=sinh8 a.
Hence the real axis in the r-plane is transformed into a
circle whose radius is sinh a and centre is (cosh a, 0).
7. Find the condition that the transformation
az + b
U~~cz-\-d (Agra 1940. 65)
transforms the unit circle in the w-plane into a straight line.
The unit circle in the H’-plane is ^.iven by | w | = 1
Conformal Representation
- /'oz+b\/'<*2±b\= i
"w~\a+d)\cJ!+rt'
(az-\-b) {a?-\-b) = {cz+d) (ci + d)
(aa — cc) zs-\-z (ab-cd)+z (ab-cd)-\- bb .. .(2i
Now if (2) represents a straight line, the coefficient of z; roust

be zero, i.e.
ua*=cc or
is the required condition. If a and c are real, the condition be¬
comes a=c.
2z ' 3
8. Show that the transformation iv=——~changes the circle
x*+y*- 4*=0 into the straight line 4n4-3 = 0, and explain why the
curve obtained is not a circle.
2-+ 3
H' =
z —4
...d)
_4»v+3
4»v+3 __4iv4-l
TP — _
Z it> —2 ’ 3_ T^2 -
iv-2
m ■n-'mhGm-'
The given circle is Z2-2 (z4-3) = 0
- 4w-f 3
Solving, we get
4w + 3 f/4iv4-3N
1 6ww+ 9-+ 12 (if 4- w)-2 [4mv - 8iv- 64-3m>4-41vw- 8u:-6 4- 3m ] = 0

or 22 (H’4->v)4-33=0
or 44«4-33c=0
or 4m4-3=0.
It will represent a straight line if the coefficient of int is zero

which is clearly zero. Hence the circle in the z-plane is trans¬
formed into a straight line.
9. Discuss the application of the transformation

, Jz+ 1
v z+i *
to the area in the z-plane which are respectively in-side and outside
the unit circle with its centre at the origin. (Agra 1948)
Let z=re<9t where r ^ 1.
100 Complex Variable
IV — rL±l
rJ\ J
(1 -f rie°) (re "ld - i) i (re'9 - /) (re ~‘6-i)

iff -id id -W
(re -f-i) (re —i) (re -hi) (re —i)
i (r2 - 1 —2ir cos d)
“ (r*-f 1)+ 2r sin 0
_ _ 2r c?9 L_ , /(/■»-!)
— J + 2r sin 0+rx’*'/,2-f- l+2r sin d
= M-f f'v, say.
Equating real and imaginary parts, we have
_ 2r cos d _r2 - 1 _
j -+-2r sin d f r2 ’ r- + 2r sin 0*
...(2)
Clearly the boundary of the circle | z | =r= I is represented
by v = 0 which is the real axis in the w-plane.
A1 COS 6 /3\
A,so 1 -f-sin d'
From (3) we see that for the right hard semi-circle
/. e. — y < d ^ is positive. Hence the right hand semi¬
circle in the r-plane is transformed into the positive part of the
rr-axis whereas the left hand semi-circle, e. ~ < d ^ into
the negative part of real axis in the w-plane.

Again when r < 1, v is negative showing that the points
inside the circle ! r '== 1 are represented by points in the lower
half of the H’-plane, and r > 1 gives v to be -J-ve.
The upper half of the tv-plane is the transform of the
out-side of the circle \ z | = 1 in the r-plane.
Aliter. Consider directly
I - 1=1 »v - i
< = > 1
i 4- ic
/. e. = r, where r < = > 1.
i - iv
/-fiv|* /—iv I2
(/ —F IV) (II -/)—A* (/ - H’) ( -IV-/)
h iv (1 — r2)+\ — rJ —/iv-f/iv-f r*/iv — r*/iv=-0 . .(1)
(l—rs) (Hiv+l) + (l-f r8) / ( — iv-fiv) —0

(1 - r-) (u2 + v3-f 1)-;-2v (I 4- r*)=0
or 2v=^-| (w2+v2+ 1).
Now clearly if r= 1, v—0 and v is negative or positive if

r < 1 or >1.
Hence the result.
10. Show that the transformation
. 1 -2
w=i
' 1 +2
transforms the circle | z |= 1 into the real axis of the w-plane and
the interior of the circle ) z | < 1 into the upper half of the w-plane.
. l—z
Consider the inverse transformation of iv—i
1+
l —W
t. e.t Z=
7+iv
l~W
i e, z == '- => 1
l-f-»V
I-w 2
= ra, where r < = > 1
i+iv
(i - »v) ( — i — »v) = r2 (/ + »v) (»v -1)
O'—r2) (>v»v+ l) + i»v — iw=r2 (iw-iw)
or (1 -r2) (»vw*+1)+(1 +r2) / (vv-»v) = 0

or (1 - r2) (W*+-v*-|-1)—2 (1 + r2) v=0
or 2v=|^|(u‘+v*+I)
which clearly gives v = 0 if r= I and if r < 1, v is positive and for

r > 1, v is negative.
Hence the real axis in the w-plane is the transform of | z ] = 1
in the z-plane and the interior of the circle in the z-plane is
transformed into the upper half of the »v-plane as v is positive for
r < 1. The lower half of the »v-plane is the transform for the
outside of the circle \ z | = 1 in the z-plane.
Aliter. Proceeding in the same way as in question 9 page 99
we see that
id
„.»/ LlH_2r*in 0 ■ ■ (1 -r2)_
l+reid 1 +2r cos 0+r2"1- 1 +2r cos 0+rz *
id
if z=re .
If r < 1, the imaginary part is +ve, and so a point inside
| z | < 1 corresponds to a point in the upper half of the w-plane.

For a point on the circumference of the circle | z | = I, viz.
r= 1, w is real /. e. v=0
2 sin 6 0
and
U~2 (1+cos y)"tan 2*
/. e.t we see that the upper semi-circle 0 ^ 6 ^ rr is transformed
into the positive half of the real axis and the lower semi-circle
(n ^ e < 27t), into the negative half of the real axis.
11. Find a Mobius* transformation which maps the circle
| w | = 7 into the circle | z-7 | = 7 and maps w = 0, and w=l into
z= and z = 0 respectively, and show that the transformation is
uniquely determined.
Examine the correspondence between the interiors and the
exteriors of the circles.
Hint. Proceeding as in Q. 5, by putting a=7r, c==^"»

1 -»v"|
7?= 1, a = \, we get a transformation z
2 + w_ '
3* 16. Invariencc of the Cross Ratio Property :
Theorem. If any bilinear transformation
az y b
w=
czy-d
such tint ad-bcÔ transforms zl% z2, z3, z4 to corresponding point
u»jf u’a, it’3, h*4 in the w-plane, then
(wx, wt, w3, wj = (zx. z2, r3, zj
H’i — W4 U'3 - W2 __ Zy - Zx Z3 ~ C2
or
»•’. - ^2 * - “'4 -1 ~ r2 ^3 ri
Let us suppose that zr, wr where r=l, 2, 3, 4 arc all finite.

az.yb azs-\-b
W, - H'. =
czr-f <7 czs + <7
(ad-be) (zr- z,'
(czr+d\ (cz.y-dY
In particular.
(ad-be) (z2—z,)
w2 - H’j =
(czx + d) (cz2 + r/)'
Similarly we can write for
(ad - 7>z) (z3 -z4)
U'3 ’‘'‘“(cz.+rf) (cz3 + <7)’
w. _M. - Lad *0-
3 2 (cz3 + <7) (cz2 -f- <7)’
(ad-be) (z, - z4)

\V, — H-. =
1 d) (czx-\ d)
and hence we get
( H’| ~ >V4) (>V3 ~ »Vg) (Z,-Zd) (Z3 — Zz)
(w-w2) (w3- »v4) (Zl~Z2) (z3 -z4)' • • *(2)
The R. H. S. of the equation (2) is called the cross-ratio
of the four points zl5 z2, z3>z4. Hence we have the invariance
of the corres-ratio.
The equation (2) is also true if any one of the numbers zr or

one of numbers tv, becomes infinite. This equation is modified.
Suppose Zo —oo f 1^!=oo . We get
w» ~ »*g=z, -z.,
,v3~m-4 z3-z4 • • *(3>
If there are three points zr> ur, where r-1,2, 3 and w is
substituted for w4 and z for z4, then also the equation (2) is true
and is modified w
K-w) (H’3-tv8) = (z, - z) (za - z2)
(H’l - W2) (u-3 - vv) u7-z2) (z3 —z/ •...(4)
• • (4)
transforms -nsformation which

nant of the transfo^tiJn
Od-ic = (w1-,l,!) (W.-W,) (w2-„.3) , ,Q
We can also write (4) in the form
— ^ -*2 Zjj-Zj (z-z.)

w - w.
‘"3- »’2 Zj * (Z — Z3)
Z-z.
where /c = f-lTK’2 z* ~ z2\

Vh’3-iv2*z1_2J-
* r/2c/r there is one only if

<a-df+4bc~0.
transformation Jay be put “in 'the form k*"’'®" p 9. t/,e
w~<i~
a"d V there is only one invarian, ^ (°P'i0nal Papcr> l95»]

trtvanant , o,„, „,c transformation
may be put in the form
_L_ = _L+Ar
w—p z—p'
(The points which coincide with their tranforms are called
invariant points or fixed points.)
(i) Consider the bilinear transformation
ar-\-b , ,
»v=-., ad—bc^tO.
cz-\-d
For the invariant points \v—z, i. e.
az-\-b
2=
cz-\-d
or cz2+{d-a) 2-6=0 ...(2)
Equation (2) is an identity, i. e. satisfied by each value of 2 if
c=0, d-a = 0, b=0.
This gives w=z, i. e. all the points of the plane are invariant.
If 0, in which case equation (2) has two roots z„ z2. such that
a - d± y/M
2r~ 2c •
where M=(a — df-Âbc and r= I, 2.
Thus the number of fixed points is one or two according as
M = 0 or M-fz.0, where the two invariant points are finite.
Let us suppose that c—0 and d^0 in which case we have
a b
w=nz+-d-
Clearly we see that one of the fixed points is oc and the 2nd
fixed point is given by
(d—a) z=b.
-GH) ■
which is finite if (d-a)^0 and if d—a=0 the transformation has
got only one fixed point, i. e. oo in which case
, b
w-~z-\-.
a
Hence we have the following results to be calculated.

(1) If czfb and (a — d)2+4bczf:0, two non-infinite invariant
points.
(2) If c=£0, (a—d)* + 4bc=0, one finite invariant point.
(3) If c=0, d-aÔy one non-infinite fixed and the other

infinite.
(4) If c=0, d—at there is only one fixed point as in (2), /. e.
at infinity.
For the 2nd part, we have to find the form of = , if (1)
and (2) are the conditions given above.

az + b
w-p cz+d P az+b - pcz - pd
w-q— az-\-b ~~az +b — qcz - qd
cz+d ^
(a— pc) z 1- b-pd
~ (a -qc) z-\-b- qd
pd—b
z - --
a—pc a- pc
= ” qd—b X a — qc
a—qc
qd — h
As p, q are invariants, the equation (2) gives p= — a - qc'
Hence (3) may be written in the form
'^p=k z-P
w-q z-q
where A:=-——.
a - qc
When there is one invariant point,

I_1_cz + d_
w—p^/az + b \" az+p—p [cz+d)
\cz+d~P)
cz+d
(a-
->(-&
cz + d 1
\z— pY a-d
[V (a-d)2+Abc=0 gives only one invariant point]

a—d
and
1 2 cz+d
/. e.
\\’—p u-\-d z-p
1c Z+-.+P-P
a-\-d : —P
p+d/cl
-Mi' ♦ z-p J
a-d+2d
2c , 2c 2c
a-\-d a + d' z-p
2c 1
.+
<* + d z—p
= *+-*-
w It ere
13. Show (hot if (a - d)2 + 4bc^£0, then for the transfor¬

mation
az-\- b
U'=-,
cz -f- d
there are two unequal numbers a, ft such that
H' — a_, z— a
— ft z — ft
where k is a constant. Show also that the radius of the circle in
the w-plane corresponding to the circle in the z-plane which has
as diameter the line joining the point z —a, z —ft is
j y-j- ft l
2 cos 0
where 0 is the argument of k.
For the first part see (ii) of Ex. 12 above.
For AP and BP to be at right
P z
angles, we have
arg
z-ft r
z—a . , .
/. e. ^ is purely imaginary. B
/3
Hence we obtain the equation
of the circle having AB as diameter
as
a 2 — a
or ?-5 + *=?«0. ...(I)
ft
z—p z—ft ~ - /]
Now the equation of the transform (I) in iv-plane is to be

formed with the help of the following transformation :
tr a ,. -z—a
w-x V.
...(2)
iv-,8 z-/T
Hence, we get
1 »v — a , 1 iv - a n
• • (3)
**-0 k 7v—/3~
which is also a circle.
Now (3) can be written as
_ p _
(w-a) (w-/3)-fj (iv—a) (iv-/3) = 0
or (»v-a) (iv-0} + e2* (iv —a) (\v — fi) =0,

where k ce2<9
or
M-H-(l+e«»)_M>(«*'•+W-w(* + j3c.<V-(-a|8+i18e=1. = 0
or iviv (e~<94-e<e)—iv (ae<0 +
- IV (ae-'fl + -f + a/3e,e
-0
or u^4/3V^) u, (2f_l±Pe^)
2 cos 0 ? cos 0
, *(3e~i9+ oii3ei9
2 cos 0 ..(•0
which is of the form
mv-Mii'-M^4_c=0
whose radius r is | V(^-C)|.
Hence radius of the circle (3) is given by
r=
2 cos u (v/U*e<<,4-(ie-*0) (xe-i0-\-pei0)
-(e<e+e-‘9) (.^••+;/Je..)>1 ,
= 1 r ' ' ~ 1
2 cos 0 -(«/? + ^/?)}j
" 2cos0 {|*I* + I0|*--2|a | |îj»<2

= 1
2 cos 0
where 8 is the argument of k.
IOcS Complex Variable
14. If y is real a, b are complex numbers such that

a ' > | b\, show that the bilinear transformation
iy az + b
w = e ' -—=—
a + bz
maps the inside of the circle | z |= 1 on the inside of \w=\.
If further the point z= 1 is the only invariant point, show that

the transformation may be written in the form
1 1 1 . , , <7
-, =-, + r » where A: = 1 +1.
w-1 z —\'k 'b
Let the transformation be the required transfor-
mation. It will transform | r |= 1 to | w j = 1 if

°~ + h |
cz + d |
1# ...(1)
Making use of the conjugate property, we have
(a:-\-b) {“ z+b) — {cz + d) (cz-\-d)t
i.e. (aa -cc) zz+ bb - dd+a7>z+abz — cdz - czd= 0. ...(2)
It will be of the form = 1, provided the coefficients of z and s
are zero, i.e. ab — cd ; ab = cd 1
and aa — cc= — bb ■+ dd J
°r °b = cd, aa -f- bb = cc-\-dd ... (5)

provided od - bc-^0.
Clearly from 1st equation of (3), b ——- if a is not zero.

a
cd
Hence b—
a
If these values are substituted in the 2nd equation of equation

(4), we get
a a +Mcc=c(rc/d'
ca
dd
{aa - cc) — — {aa - cc)
aa
or (aa-cc) {aa -dd) = Q, .. (5)
i.e. either |a| = |c| or \a\=\d\.
If we take o c , it can be written in the form c = ae^1.
where ft is real. Substituting it in the equation ab=cd. we find

that b = de^ or b=de~^ i.e. d—be${.
az-\-b
Substituting these values In ’ we get
az+b az-\-b e~nj(=
w~~ae& z+be& (az-\-b)'
which is not the bilinear transformation.
Hence we take |o|=|^ 1=1^1 and write d=ae‘t‘ i.t.

d=ae~Y< , where y is a real quantity. Substitute this value again
• •
in (ib=cd and we get b=ceyi or c=6e_Y' •

Hence the required transformation becomes
02-f b _ yi /gz-\-b\
w ~ bzer~’ti+ae~'fi \bz + a)
provided aa—bb^t0 i.e. | a & | b |.
Now to show that the interiors correspond, we find
=-- {(1 - zs) (aa - bb)). .. (6)

I a + bz |2
Clearly the left hand side of equation (6) will be positive if

22 < 1 and | a | > | b \ and negative if | a | < | b |, i.e. (5)
maps | 2 | ^ 1 on | w \ < 1 conformally if and only if
a | > | b !.
(ii) If 2=1 is the only invariant point of the transformation
; az+b
w= 1
bz + a '
such that 1 is the root of the equation
bz,-\-(a—ae'fl) z~be^‘ =-0=6 (2 - 1 f
» e. a - oeY' = — 2b and - be'(i= b
or a-\-b—ae^1 - b ; b — —be'*'.
1_ fbz+a)_
Hence
w—\ eyl (az + b)—(bz+a)
b+a + (z- 1) b
be'*1-
(ae^-b) ( z+
oeyl -b)
=_\ (z-1) b+a_1

~ae'(i — b U- \ V ae** -b'iz-U
[V from the inversion of the equation (5), we get
i. e. a -ben = tfel< — b or a-\-b = (a-\-b) e*1].
Hence (8) takes the following form with the help of (7),
1 b a+b 1
it- 1 a+b+a+b (z~ 0
1 J_
z— 1 + k '
where A = ^ I 4-
15. Prove that the transformation
awz — bw — bz-\- a — 0
maps the circle | r |= 1 on the circle | w =1 if b |^| a |.
Find the condition that the interior of the first circle may be
mapped on the interior of the second. Show also thary in this
transformation, the fixed points either are inverse points with respect
to the unit circle or lie on that circle.
The transformation becomes
a—bz
For the first and second parts if we substitute -b for, a, a

for b, a for c, - b for d in the equation (1) of question 14 above,
we get the condition corresponding to a \^f\ b | as i b \^t\ a ■.
By the same substitution, we get
| b | > 1 a | for the second part.
In the third part, we have for fixed points
oz‘ - (b + b) z + a = 0
which being a quadratic equation gives two roots of r (which are
fixed) given by
(by-h) +\f{ (b + h)' - 4aa} _ _\by-b)— \ ■+■ b)" - 4ca)}

• 2
2a 2 /
If Zj and z2 are inverse points with respect to the eircle

| z |= 1, we get z132= 1,
/. c. if <b+t>) + y/{<b+b)2-4aa} (b+b)-y/{tb + b)2 - A a a) 4 aa

'I/, * --=• = i
provided (b+b)*-4aa > 0.
Hence the points z, and z2 are inverse points with respect to

the circle I z != 1.
If (b+b)2-~4aa < 0, we have
z, = ^ + 6) + iV {4aa-(b+6)2}
2a
and z _(b + b)~ i\/{4aa-(b-\-bY-)

2a
Here, we have
=n iVi4aa-(b + 6)2)1
2aa
° 1(6+6) - iV{4<7« - (/> + £)2>}]

2 aa
1
~4aa {{h + h)2 + 4a°-(h + b)2]
= 1.
Hence z, lies on the circle \z\—i cirv,-, ,
= 1. |Z|~1- Similarly z2 will lie on
16. Express the relation w-U!Z±21 ■ ,

17^57" m the form
\v-b
*2 ~ * wl,ere °* b- k are constants. Show „,al ,he cirde ”' ‘
2-plane whose centre is z=0 5 . /e
C,r<7* in the w plane on the line joining w- tm,'Sfonne(l "”o the
and the points in the z.pfone which nrJ °nd *~b as diaêter
are
(Alld. 1943, Agra 1949)
From w= 13/z-f 7 >
3z-5i
we get ^=I3fe+7S-0 (32—5/)
1^+75^673^
(13/ — 3a) 2 + 75+5/0

(13/ — 3b) z + 75+5/5
(75+5/1/)
13/ — 3^ Z 3a —13/
...(2)
13/ - 35* 75 I- 55/
35-13/
75 + 5a/
Now if
3o-13/
= 0,'
i. e. 3o2— 18o/—75 = 0.
giving o=(3/± 4). Similarly for 5.
Hence, taking o = 3/-4, and 5=4 + 3/, ...(3)
. 13/ - 3o__ 13/ -9/+ 12__4/+12_/+3
13/ — 35 13/ — Vi* — 12 4/ — 12 i-3
= by rationalising the denominator
_4 f 3/
...(4)
-5
and arg (2) = argA:.
\v - a u z —o
Hence ...(5)
>T-5 F^5’
where o, 5. A: are given by equations (3) and (4).
Taking argument of equations (5), we have
»r — o 2-0
ars 6=ar* *+ar8

...(6)
If the point z moves on the circle | 2 | = 5,
arg ?r^ = arg (r-o)-arg (2-5)
= arg [5eie —( — 4 + 3/)} —arg (5e<f - 4— 31)

= arg {i5 cos 0 + 4) + / (5 sin 0 — 3)}
— arg {(5 cos 0 - 4) + / (5 sin 0-3)}
5 sin 0 — 3 . 5 sin 0 - 3
= tan“l -z-- tan-1 ^——~u—.
5 cos 0 + 4 5 cos 0 — 4
= tan_1 [8 (3 — 5 sin 0)]/[6 (3-5 sin 0)
tan -1 43 «
Hence from (6), we get

w-o . _ . .
arg -i = tan~1 J + tan 3
IV —
tan"1
which means a circle in the ir-plane drawn on the line joining the
points a and b as diameter.
Again if we take a point z outside thj circle z -= 5, /. e.
| z j=5/x where p > 1, then
z=5/«^>'<^
z-a 5/x sin P — 3 . 5p sin 0 - 3
Hence arg -—7 = tan~l - tan'1 ----—,
z-b 5/x cos 0+4 SpeosO -4
_ — 8 15// sin S - 3)
n 25/xa - 7 - 30u sin w
_! ~8J5/i sin 6-3)
< tan /‘
! 8 - 30/t sin 6
= tan-a £.
z—
/. e. arg < tan-1 3
z—b
= tan_l * - a, say.
• • arg ^—^=tan_1 2 +tan-1 j} - a
— tt/2 —a.
Hence the points in the u-plane corresponding to the points
outside | z | = 5 shall lie interior to the circle
w — a tt
arg
( ds
it— /> 2'
— is an invariant with respect to the trans¬
formation where a, bt c, d are real numbers satisfying
the relation ad - bc — 1, where ds = V (dx‘ + d) ~). (Alld. 1946)

The inverse of the given transformation can be written as
b-dZ
z= .(1)
cZ -a
Differentiating (I) with respect to Z, we have
dz — —(ad - be) dZ/(cZ -
-dZ
or dz— ■ •
ad— be — 1.
(cZ — a)-'
dZ\
Hence ds= y/(dx2 dy2) = | dz j =
cZ - a ,a
dZ I
vjtv. where Z=A'+ / >'.
(cX-a)* + c2Y*
/>-</Z />-JZ
Again 2iy = z-z =
rZ-fl fZ-tf
__(h - dZ)(cZ — a\ - (b—dZ){cZ—a)

(cZ—a)(cZ — a)
_ (ad-bc)(Z-Z)
r-( .V2+y2H- uz—2aXc
2iY
...(2)
c- (.V2 + Yx) +u*-2aXc
Y
}~~{cX-a)*+c*Y*'
Js _j dZ \_\'{dX2+JY-)_da
/. e.
v9
Y Y Y' Say*
‘ as
(Is . .
Hence — IS invariant.
y
Exericses
iyb
Prove that if -, 0 < a < h, the inside of the circle
a
x2 + y2 = a9 corresponds to the inside of an ellipse in the w
plane, but the transformation is not conformal.
Find the Mobius transformations which make the sets of
Points in the r-plane (i) a, b. c, (ii) 2, 1 -f/, 0 correspond to
the points 0, 1, =c of the ir-plane. In case (ii). show by
sketches the domains of the u*.plane and r-plane which
correspond.
Find a Mobius transformation which maps the circle | z | < 1
on ii’ - 1 | ^ 1 and makes the points r = 0, I correspond to
H'= A, 0, respectively. Is the transformation uniquely deter,
mined by the data ?
Âns. >nter>or corresponds to the interior. J
Find the transformation which maps the outside of the circle

j r | = 1 on the half plane R\v > 0, so that the points z=l, —/,
— 1 correspond to \v=i, 0, — / respectively. What corresponds
in the ir-plane to (i) the lines arg r = const, | z ! ^ 1*
(ii' concentric circles | z\ = r, (r > 1) ?
[ads.
Prove that it = .—^ maps the part of the real axis
between r= I and r = — 1 on a semi circle in the w-plane.

Find what regions of the ir-plane correspond by the transfor-
mation tv=-—to (i) the interior of a circle of centre z= —/'

z+i
(ii) the region y > 0, .x > 0, | z+i \ < 2.
7. Prove that if tv = -2^^ and aS— /3V = 1, then the linear and
>z-t-6
superficial magnifications are | (yz+S) ~2 and | yz+5
8. Show that the relation
, fe + 2
* 4z -4- /
transforms the real axis in the z-plane to a circle in the
wplane. Find the centre and the radius of the circle and the
point in the z-plane which is mapped on the centre of the
circle.
9. Prove that the bilinear transformation u = J maps I 2 | < 1
on to the half plane R (tv) ^ 0. Find also the region in

the tv-plane corresponding to the region
\z \ 1. If u=a"
yZ-j- 6
and z describes the upper half of the circle | z |= 1, show that
tv describes the upper half of the circle descibed on the line
joining the points and as diameter.

/'j-d y —6
11. Determine the most general bilinear transformation which
transforms the circle | z |=1 into | tv | = 1. What is the most
general transformation when the further restriction is
imposed that the points z= 1, —1 are to have images at
tv=l and tv= — 1 respectively.
(For the first part see the solved example no. 3).
12. If a, ft are two fixed points of the transformation
a-\-bz
tv= , adzy bc
c-\-az
tv - a z a
show that -A V
IV
-ft
ft
where A is a root of the equstion
t2 — (b + c) t + (be - ad)—0.
CHAPTER IV
CONFORMAL REPRESENTATION (Continued)
4*1. The fundamental theorem of existence as stated by

Reimann is as follows :
Any region with a suitable boundary can be conformally repre¬

sented on a circle by a biuniform transformation. •
We generally see in the practical applications of the confor¬

mal transformation that the problems which arise are to find the
function which transforms a given domain D in one plane to a
given domain D' in the other plane.
Although we can conclude from the above existence theorem

due to Reimann that there exists a function/(r) for each such
problem, hut it is difficult to find the particular function f (z)
for each and every problem. It can be stated that we have
considered the bilinear or Mobius’ transformation which trans¬
forms the circles in one plane into circles in other plane. As there
is no general method of finding the appropriate analytic function
/(“) for any arbitrary shape, we shall consider as to which of the
domains corresponds a given domain when f (z) is one of the
functions or a combination of several functions.
If we are given any function f (z), we can have sufficient

idea of the mapping by the consideration of the images of the
following curves :
R (c) = const. / (r) = const.
r =const., arg z = const.
We already know that 7?(c) = const , and I (r) = const. are

the orthogonal curves of one system. They shall in general
be transformed into families of mutually orthgonal curves. Here
v.c shall consider some special transformations whieh are of
great importance in practical problems such as Hydrodynamics
Heat conduction. Electrostatics etc.
4‘2. Some special transformations.
]. a =r" (/; being a positive integer).
Conformal Representation (continued) 117
Clearly the mapping is confoimal everywhere except at
the point z=0 where l = 0, and it is noted that the
correspondence between 2 and it- is also not one-one but n to 1.

Suppose w—u-\-iv=pei(? ; z=x+iy=re'°.
Hence it follows that
r (cos «/> +1 sin </<) = rn (cos 0 + i sin 0)"=/-" (cos /i0 + i sin nO).
P cos <f>=rn cos nd 1
p sin <l>=ru sin nO J •••(!)
or P—r" 1
= /!0 J ...(2)
circle,firstr0f ecluations <2) we see that the

circle r-c of the 2-plane for different values of c corresponds to
the circle P = c"in the w-plane, particularly the points on r= z 1=1
are transformed into the points at unit distance from the origin
in the w-plane.
(2) The curves 0=const., emanating from the origin of the

r-planeare transformed into the similar radial1 turves 0=consr
However, we note that the line in (he z plane having 0 as its
slope is transformed into a line whose slope is nd = <j. in the
•‘-plane ; which means that the mapping at the origin is not
conformal, because ^=„2»->=0 at zÔ. Hence .-=0 being a
critical point, the conformal property does not hold good at the
origin, and the angle between the two lines at this point in the
“•-plane is muluplied by n.
(3) Clearly a wedge (circular sector) having the origin as its
vertex is transformed into a wedge with its vertex at the origin
ange,eTn ;.pa,a' ee ^ ""P,ane fa * that central
(4)
If vertex angle in the a-plane is 0=n,„ Hz. the domain,
0 < arg z < TT//7,
then the interior of this wedge is conformally mapped on the
indented" Whe" /(M) > °- as ,he ori8in has been

(5) Similarly the domain defined by
t , . 0 < arg z < 2*/n
of the wedge in a z-plane transforms conformally on to the entire
M-plane and cuts along the positive real axis. This cut is often
called a slit and the plane is referred to as slit along the positive
real axis.
II. A special case of (1) when n=2. Here we have the trans¬
formation U’ = Z2.
G‘'Lo=(2z)-o=°.
Hence the conformal property does not hold good at the
origin or oo .
\v = u+iv=(x+iyy=x2 -y2+2ixy.

' u = x2—y*,v=Zxy. ...(1)
First of all we shall find as to what is the image in the

tv -plane of the lines parallel to the imaginary axist in the z-plan.
Let us take x = dx and x = d» as the st. lines parallel to the imagi¬
nary axis in the z-plane.
(i) Suppose that dx > 0. Then we get from the equation (1)
u = d*-y*; v=2dxy. ...(2)
Clearly on the line x=dlty varies from — oo to oo . Eliminat¬
ing v from the two equations in (2), we get
v*=4<V (</,* — u) ...(I)
which is clearly a parabola having its vertex at u—dx2t v=0 and
the locus at the origin and is given by the following figure (2) in
the n-plane :—
Fig-(0 Fig. (2)

When y varies from - oo to 0, v varies from —oo to 0 and v
varies from 0 to oc as y varies from 0 to oo. Also we find that
is positive so along as y is less than dx\ for y > dx, u becomes
negative. Hence we find that the part of the line * = </, below the
Conformal Representation (Continued) 119
real axis in the 2-plane corresponds to the lower part of the

parabola lying below the real axis in the w-plane and the parts
above the real axis in the two planes similarly correspond.
(ii) IfrfjCO,
We see that the same parabola in the u-plane is the imace
of the line x—dx\ the two parts of this line above and below the
real axis correspond to the two parts of the parabola below and
above the real axis in the w-plane.
(ii') We can verify very easily that the shaded area in the
r-plane between the lines x=dx, and x=d2 (0 < dx < d,) as well
as x—~dx, x= - d2 maps the region between the two parabolas
v*=4dx* (dx2-u) and v* = 4d22 (d2z-u) conformally.
It is quite obvious from (i) that the shaded area between the
lines in the 2-plane is developed by the line x=k in the r-plane
by a parabola v2=4k* (k* - u) in the plane for different values
of k i.e. as k varies from dx to d2. Similarly for the second area
in the 2-plane.
We also see that if d£-+oo the area developed in the it -plane

is the area outside the parabola (not containing the origin)
v*=4dxz (dxz — u)
which corresponds conformally to the semi-infinite plane to the
right of x=dXt i e. the domain represented by R (z) > dx. The
same exterior part of the parabola is mapped by R (2) < —dx.
Hence the semi-plane R (z) > dx > 0 is mapped conformally
on the exterior of the praboh v*=4d* (dx~-u) by the transforma¬
tion »V=2*.
(iv) Now we shall consider the interior of the parabola.

If the parameter *-+0; the parabola v-=4.v2 (x°--u) narrows
down, (because the latus rectum of the parabola decreases], until
it becomes a slit along the negative real axis in the n-plane from
0 to so. Hence we see that the portion between the lines a:=0
and x=dx in the --plane corresponds to the portion of the w-plane
between the parabola.
va—!</,* W-k)
and the slit or cut along the negative real axis from 0 to — ;o.
y
Mill 'll." ••
■ 1 •’ ' 1
at*• -1■ inil •
•f•*«’•if.>||>
V*
■WMl,
i.i i.
o . i X
. J ..l',' ■'
it
X
2°LANt
The two results given in (iii) and (iv) can be enunciated as

follows :
Tne portion of the z plane R (z) > 0 corresponds to the whole

ir-plane cut along the negative real axis from 0 to — oo .
3. Discuss the application of the transformation \v=z2 to the

area in the Jirst quadrant of the z-p/ane bounded by the axes and
the circles.
| r \ = a and | r |=b (a > b > 0).
(Agra 1947, Rajasthan 1958)
Is the transformation conformal ?
Now w=z' — r'euo. where z=rei9.
Taking iv=*Rci?, we get the relation
R=r2, and <t> = 20.
Thus the given quadrant | z \=a, 0 < 0 ^ tt/2 in the r-plane
is transformed into the semi-circle.
| w =R = a2, 0 < <£ < it
in the it-plane and similarly the quadrant | z [ = />, 0 < 0 < *72
is transformed into the semi-circle.
| It' I =b\ 0 < if, < 7T.
Thus the area in question is transformed into the angular

area between the two semi-circles |H-| = a8 and | >v!=68in the
upper half of the u -plane. For the given region the transfor-
Fig. (1) Fig. (2)

(i) By means of the first of the equations (1) we see that
x=a > 0 corresponds to the rectangular hyperbola
u2 — v2=a ...(2)
in the H’-plane having its transverse axis along the real axis and
the conjugate axis as the imaginary axis.
It is quite obvious that either branch is the complete mapping
of the line and depends on the branch of y/z, we have taken.
(ii) We see from equation (1) that x=constant and y=cons¬
tant are two orthogonal families of ractangular hyperbolas in the
tv-plane, as is quite natural, because x = constant and >’=constant
both also intersect orthogonally in the z-plane.
(iii) The shaded area bounded by x = a, x=b represented

by
a < R (z) < b (lying to the right of x=a) Fig. (2)
is transformed conformally on any of the shaded portion repre¬
sented by
u* — V2=a ; u2 — v* = b
in the tv-plane and these lie separately to the light as well as to
the left of the imaginary axis in the H’-plane.
Conformal Representation (Contiuned) 123
Clearly the domain obtained to the right of the imaginary

axis in u-plane is the branch of the hyperbola for which w is
\/a for z=a > 0, and the area on the left is obtained for the
branch for which w is -\'a for z = a > 0.
(iv) Similarly for >> = constant, we get the branches given in
the shaded parts given in the figures (3) and (4).
(v) As b->oo in third part of this question, the domain is
represented by
R (z) > a . . .(3)
and we see that the interiors of the two branches of the hyperbola
«l-vi = «aie the conformal mapping cf the domain given by (3)
separately ; the two transformations correspond to the two
branches of y/z. We see that the critical points w—0, z=0 at
which the conformal property does not hold good are not included
in the domains.
»t’-plane z-plane
6. (a)Show that the transformation w (z-\-i)*=l maps the
interior of the circle | z \= I in the z-plane on the exterior of the
parabola
=2(1- cos <f>) where pe9 —
in the w-plane. (Agra 1942)
Now 2+i=4 =4 y/W y/p ...(1)

i.e. e" + i=- e-i*l2 •/ on \z\ = l.z=e'*
VP
cos 9-\-i (1 -f-sin 0)=— e~i9/2 .
VP
= VpVC0S 2 "/s,n 2)
Separating real and imaginary parts, we have
(1 + sin 0) =-sin ^
VP
a 1 0
...(2)
cos 6= — cos
Vp 2
Eliminating 6 from the equations (2), we have
/I 0'
( —r COS -r-
Wp 2 y
2 . «/>
or —— sin ...(3)
Vp 2 P
1
or 2 sin ...(4)
VP
Squaring, we get
— =4 sin2 =2(1 - cos 0)
p = i cosec2 </>/2 .. .(5)

which is a parabola. Thus it is shown that when r describes a
unit circle in the 2-plane, w describes a parabola having its focus
at the origin.
Now if we take curves P= ^ cosec2 where ^ > 1, they are
confocal parabolas exterior to the parabola (5;,

1 2.0
i. e. —r --sin
VP P 2
[mark that it is not -f ve. otherwise, we get from (2)
2 (fa
sin 6= - 1-sin2 — which is not possible!
P 2
Hence from equation (1), we have
x+i<y+ 0=- 2 sin

• ^
e--19/2
2.0 0 I .
•*=-sin — cos
p 2
2.00,
2
--sin A,
fi. r
>’= — sin2 4,- - 1 = —- cos 0-f-1

1 1
1r ..(6)
P 2 n /x
Eliminating 0 from the
- equations
- (6) above,
w !
we have \ * / — ^
A,+(r+1-i)!=
1
P2
which represents circles with centre at fo, - - A and radius- all

\ P S p
touching at (0, —1) covering the entire interior portion of the
circle | z |= 1.
Hence the exterior of the parabola
— =l_cos‘/' correspond to the

2p
interior of the circle | z |= I.
or Directly, we get. For the interior
1 2 1 < 1
’!=(1'1‘7+7‘psin''')
or \ < - 2 sin y
vp 2
or — < 4 sin-
P 2
or p > i cosec2 (•/»/2),
a curve exterior to the parabola.
6. (b) //(-+ lj2 = 4r. show that the unit circle in the z-plane
corresponds to the parabola R cos2 (0/2)= 1 in the Z-plane and the

inside of the circle to the outside oj the parabola.
(Alld. 1942, 1947 ; Agra 1932, 1956, 1961)
The correspondence gives
r _i_ i — — — — —— e 1012 where Z = Re10

yj Z. \J
i ( e .. e\
or T
x+i +{==Vr\0S
v * \ 2 "z s,n Jj
liquating real and imaginary parts
2 0
*+1 = V/*C0S7 ..(2)
2 . 0
>,== ~7/r s,n 7
whence
/ 2 0 ty / 2 . o y i
A I+^=(7«cosT-1)n-7«sin7)
4
or
R-VRCO*h° i e RC°S'-T = U
Hence the circle lr| = l in z plane corresponds to the
parabola R cos2 — = 1 in the Z-planc.

Again if we take a curve outside the parabola given by
R=n* uc2 0/2 where /x > 1, then from (2)
«i ^
a + 1= - cos2 4-=— (l+cos(y)»
/X
2 6.6 1 . .
y=-cos sin -=■ =-sin 6
n 2 2 fjL
whence eliminating 6t we have
(x+1 - — ^ +}'*=—,2 where -- < 1.

\ V- / M ^
a series of circles all touching at (-1, 0) f/t varying from 1 to
infinity], covering the whole interior of *v +y~ = 1.
Hence outside the parabola R cos2 (6j2) = I corresponds to the
inside of the circle I1 c $! = 1.
Directly another method.
For the tnside of the circle \ rj=l, we have .\'-+v2 < 1
i.e. from (U), R cos2 (6J2) >1 or R > sec2 (6/2)
i.e. outside the parabola R cos2 (0/2)= 1.
7. Prove that the quadrant Z 1, 0 < arg Z ^ rr/2 is
4
mapped conformally on a domain Z in the w-plane by >r = —=—- 2 .
(^ I * )
bind Z and determine the length of its boundary. (Agar 1945)
(Z+l)*’
Differentiating (1), we have
du’_8
dZ~ <Z+])3*
which is :+0 or °c for any point of
4
the quadrant. Hence ^ )g maps
the quadrant conformally on a domain

Z if u» does not take any value twice in Z-plane
the quadrant.
Also if Z=Zj satisfies

(Z+l)2
so does
4
since
(Z1+l)2*
But if Z—Z, is a point on the quadrant, then -Z,-2 is
obviously not the solution of (1) [suppose Z=Z1=rei0, r < 1,
0 ^ 6 < 71 /2 then - rei0 2 does not lie within the quadrant].
, —
Thus »»■ does not take any value twice in the quadrant.
Hence the quadrant is mapped conformally on a domain Z in the
w-plane.
Now we find r the contour of T. But since w is infinite

only when Z=-1, the quadrant is mapped on the domain
within T.
(1) When Z moves along the real axis from 0 to 1, w moves
along the real axis from 4 to 1, and hence the length of this part
of r is 4 — I =3.
(2)
When Z moves on the arc of the quadrant | Z i = l.
i.e. Z=e‘°, 0 < 6 ^ tt/2.
4 1 e-'°
)v~Rei? =
(1 -f eie,2 ei9Se’i0U + e+i9'*\2 cos- (tt/2)
(- ■
1
i.e. R= and <f>= — 0,
cos2 (0/2)
2 2
so that R= __ =see2 —
14-cos 0 14-cos 7> 2
i.e., it- moves on a parabola from
7T
“ -y < '/■ < 0.
Length of this part of 27 is
r^vMsyM’- sec3 / c//, where —= /

,1,
sec3 — dh
2
-i:
—
f sec / tan /-Hog (sec /4-tan /)
I7"'1
= V2-flog (14-V2).
(3) Finally when Z moves on the imaginary axis i.e. Z=iY,
4 __ 4(1 — iT)2 _4(1-/T)2
iv = —=u4-/>.
(1-WIT* (1 4*/>V( 1 -iY)2 (1 4- Yz)
4(1 — y2) < -8K
re.
M '(14-Z2)2’ V (l4-y*>*\
/? = >/ (M14- V2) = (yy psp [(1 - z *)2 4- 4 Z 2|1' * * (J^yTy
2Y i - Y*
tan •!> = - so that cos </>= (-ypa
1 -Y
or 1 4-cos •/»= 1 4*
I 4- Y2 1 4- F2*
7? = 2 (1 4-cos 74.
77
Hence w> moves on a cardioide from «/•=-- to 0 fas V moves
from 1 to 0].
♦ r_
Length of this part of the curve is
<!> .
ncos'2d
= 4 y/2.
Hence the total length of the boundary of 2" is
3 + {x/2 + log (l-f v'2.} + 4v/2
= 3-j-5\/2-|-log (1 -f \/2).
n
8. Discuss the transformation w=tans ~
(Agra 1943, 54, 55, 58, 62)

2
We have seen in Q. 6, that u == —-1 cannot be used to
\ *
map the rigion inside the parabola R cos2 P/2=\ on the unit circle
| w | ^ 1. Now here we have considered a transformation which
enables us to do so.
The given transformation w= tan2 - VK can be considered
as the combination of the following three transforms :

(a) H’=tan2 r/2, (b) z-\nt, (c) t=
where = u+i
»r =
»•; if; t g
£ £ -f- = iy
-f t/; z =x -f
(Agra 1943, 55, 58, 62)

(a) Let us consider the first transformation
>v=tan2 (z/2)
1 - cos r
1 + COS 2*
M-plane r-plane
On the line x=ir/2, we write
Then n =
1 + / sinh y
1 - i sinh y
so that 1 + i sinh y
w =
1 —i sinh .;
_ VO +sinh2 y _
VU + sinli* y)~ '
Hence the line x=~ jn the z-plane corresponds to the unit

circle] w| = l in the tv-plane.
Again on x = 0, z=iy so that w = —~Cosh y

1+cosh y
= a real quantity.
Now asjt changes from - *> to 0, tv changes from - 1 to 0
[ Lim 1 - cosh y_ J/cosh y - 1

--——
-►—oo i+coshj' 1/cosh y+1

-r
as cosh y->x when y-> -oc J

Again as y changes from 0 to oc, w changes from 0 to -1
so that as z travels on x=0, from — oc to ©o, w travels from — I
to 0, and again to - 1, thus making a cut along the real axis from
H’= —1 tO H’ = 0.
Again to show that the interior of the infinite strip between
*=0 and *=~ corresponds to interior of the circle
putting z=-2~a + iy where a is real and <-, we have
\v— 1 - sin -yy)_ 1 - sin g cosh y-\-i cos a sinh y

1 + sin (a — iy) 1 + sin a cosh y - i sinh y cos a
N
D'
Now I N |* - | D |2=( I - sin a cosh >-)* + (cos a sinh y)2
-(1+sin a cosh y)2 — (cos a sinh y)*
= - 4 sinh a cosh y < 0 as 0 < sin a < l.
.*. I w | < 1.
Hence, if »v = tan2 r/2, the strip in the z-plane between the
lines *=0, * = - is represented on the interior of the unit circle

in the w plane, cut along the real axis from w= - 1 to tr=0.
(b) The strip in the /-plane is obviously that between the

lines <t=0 and <r = 1.
(c) Now we discuss the transformation
t=VK ...(l)
/. e. (,7+ /t)2 ==£ = (£+ />}).
<T2 - T2 = £ ; 2<rr = ...(2)
On eliminating tore from (2), we have
r* =4o2 (a2 — E) or ri2 = 4-2 (r2-f-£) respectively. ...(-3)
We see that, by the first of the equations (3), to the straight
lines '7=const., in the /-plane correspond parabolas with vertex
at E=g2 and the focus at the origin of £-plane.
Let us consider a particular parabola of the 1st system
which corresponds to <r=l,
i.e. rl2 = 4 (1 —s)»
whose transform in the /-plane is the line through /=1, which is
paralled to the imaginary 7-axis. We easily verify as in question
no. 2 (iv) P. 121 that the shaded areas correpond. If the para-
/-plane
meter (7-^0, the parabola r,2 = 4(l— 4) narrows down until it
becomes a slit along the negative real axis which is a branch
line.
Hence the transformation u =tan2 \/Z, represents the
domain which is inside the parabola ABC on the interior of the

unit circle | tr I = 1 in a one-one correspondence, for the real axis
of vr-plane between — 1 to 0 corresponds to the real axis of the
c-plane between — 00 to 0.
Also,
d\\' r 7r 7T
VZ sec2 T>/
2c L,an 4 0
f 4v%
which tends to a finite non zero limit as C-»0. Hence the points
£=0 and «'=0 are not critical points. Thus we see that the
transformation is conformal and one one.
9. Show that by means of the transformation w=*(-——

\z -f- ic )
the V
upper half of the w-plane may be made to correspond to the interior
of a certain semi-circle in the z-plane.
(Alld. 1944; Agra 1951, 57; Luck. 1950, 60)
.z+ic
\z+icj
or u+i,4*±yyzsyr
U+i'Cy+cJ
{-y+/ (.v-c)}2 {x—i O’+ci}*
{*+*' (y+c)}* {x-i(v + c)}2
_{(*2-f >2 - c*) - 2icx}2
[x* + (y+c)*y
(x* + y*-c*y -4c*x*-4icx (A2-f _y2-r2)
[x* + {y+c)z)z • • .(i)
w = fAT*4->2-ci)2-4c,,jr2
_l*‘ + (.V+c)2J2
4cX (-C8-fAT2-f V2 • • .(2)
[**+(>'+r)#]2
When at=0, v=0, *=[ T /fcfY
As z passes from A' to B

where OA = OB=c, y changes from
~c to -f c, and u changes from -f <»
to 0.
Again as z moves over the
circle x*+y2=c*,
v=0 Is' *
and u= — 4c2*2
(2cz-y2cy)*
x2
_c2~yt
(y + c)* (c+yf
\c+yj'
Putting y=c cos 6, we have
u= ~~cos A ) ;
(1-f cos 6) -
= - tan* ~ .
Hence as z describes the semi-circulur arc BB' A', i.e, as 0

varies from 0 to tt, u changes from 0 to - oo, so that the whole,
of the w axis corresponds to the boundary OBBfA'.
For any point lying within the z-semi-circle in the z-plane
both AT and c--x*—y* are positive, so that v is positive. Hence
the given relation transforms the interior of the z-semi-circle into
the upper half of the u>-plane.
10. Show that the transformation

___ -f Ws
maps the interior of the positive quadrant of the unit circle in the
w-plane conformally an the inteior of the positive quadrant of the
z-plane. Discuss also the correspondence between the boundaries
of the two domains.
Let us suppose
'=•»*. ...d)
so that we have
1+/
==l^t ...(2)
/=(«e'v0’ = «^'9i where w=Re,,
and if we take !=pei\ we get the relations
Hence the interior of the positive quadrant |iv| = l,

\ '/* ^ rr/2 in the iv-plane is transformed into the semi-circle of
w-plane /-plane
the '-plane and the interiors correspond conformally except at the

ongi n t
The inverse of (2) can be written as
,=£nl
*+r
Now for the interiors, we clearly have
1 - t 7=1 - ini 1
*+!’*+!
_(*+l)(2+l)-tz-l)(?-n
(Z4- I)(2+l)
_2(z + 2) 4/?fz)
*+l j2“| (z + 1) |2* --.(3)
I 11 < the R- H. S. of (3) is positive, i.e. R (z) > 0.
Now '/£ri *_=I\
21 2/Vz+l 2+1/
_
4 [ Z2- 1 - 2 + 2 - Z2—2 fZ+ I
2 (z —2)
(Z+1) (2+1) ]
\ (2T+- 1),*
2/(z)
2+I |2* ...(4)
From (4), we conclude that if /(,) > 0, so is /(z)
where and/(„ both are + ve , e. the first quadrant" of
Alos ~ = 4--
iv
clw f 1 — w*)2 = 0 when w=0.
Hence the conformal property does not hold good at w=0.

(ii) Correspondence between the boundaries.
(a) If w varies from 0 to 1, z varies from I to oo along the
real axis in the z-plane. ^
(b) When tv varies from t to along the arc of the unit
z varies - - - ° -* - «■£
<c) When .V varies from / to 0 along the imaginarv axis
:Zc'rP : Z Va"CS fr0m 0 10 1 a,0ng “he

z-plane
11. If w=2z+z2, prove that the circle } z \ = 1 corresponds
to a cardioide in the w-plane, and that the vertices of an equilateral
triangle in the circle correspond to the points of contact of parallel
tangents of the cardioide. (Agra 1941, 53, 59)
We have, u-s*2z+z*
or U>+1=(2+1)2.
Let u’-fl = pe,?, where the pole has been transferred to

(- 1,0). Now for the circle | z |= 1, we can write z=ei0.
Hence pe'?=( 1+ **•)*
pcr? = 4 cos2 j eie, V eie,i+e~i9l2=2 cos
Am
Hence P=4 cos2 and </> = 0
i e. p = 2‘( 1 -f cos 0)
= 2(14-cos </*), which is a cardioide.
It means that the boundary of the unit circle in the z-plane
corresponds to the boundary of the cardioide in the tv-plane hav¬
ing its pole at tv= - 1.
We see that the conformal property does not hold good at
(^) 2+2r=° or
=
Hence at z= - 1, the conformal property does not hold good.

(ii) The vectorial angles for the vertices, of the equilateral
triangle in a circle |rj=l are 0,
0 + 2tt/3, 0+4*13 respectively. [Since
angle at the circumference is half the
angle at the centre on the same arc,
hence the angles at A, B, C are each
equal to tt/3, as the angle at the centre
is 2*/3 on various arcs which subtend
angles A, B, C ]
Hence the corresponding points on the cardioide are
P [2 (1 +cos 0), 9], 2 [~2 {l + cos (e + tf+H
'[’{'«-(*♦?)}• ,t?].
The angle which the tangent at these points makes with the
_ . dv dvIdO
axis
is of u is — = where u=P cos 0, v= P sin 0
/>. u=2 (1 4-cos 0) cos 0
v= 2 (1 4-cos 0) sin 8
2 cos 64-2 cos 28
}
Hence
du du/d8 —2 sin 8 - 4 cos 8 sin 8
= cos 14-cos 28_ _ 2 cos 30/2 cos 8/2
sin 04-sin 28 2 sin 30/2 cos 0/2
30
= - cot ~2 .
Similarly, we get the slopes of the other tangents at the points

Q and R as follows :—
Since the slopes of the three tangents, are the same, they are
clearly parallel. Hence the result.
12. Prove that \ zA- y/fz* - c2) I4-| z— y/(z' - c*) |= | z+c |
4-1 z~c I- Hence show that if the relation between z and w-p/anes
be given by
2*4-211*2 cos A 4-wz = J
and if z describes an ellipse whose foci are branch points in the
w-plane, w describes an ellipse whose foci are points in the w-plane.
(Agra 1946, 53)
We have
I ?! 4-2,1 HI -Ti-2j |*--(2, 4-Z2) (2, 4-32)4- 2,-Zs) (2,-5,)
= 2 (2,^4-22 -f-2)
=2 12,1*4-2 |z21*,
Substituting \/(z+c\ y/(z—c) for z1 and z2 in the above,

we get
I V(z+c)+y/(z-c) l8+l \/{z+c)-J(z-c) I2
=2 I y/(Z-\-c) 1*4-2 I y/(z—c) \2
i.e. 2 l z+y/(z*-c2) I +2| z—y/(z*+c*) |=2 { \ z+c |+| z-c !},
since j z |2=| z2 |
or | z+y/(z2-c2) 1 + ] z-y/(z*-c*)\=\ z+c\+\z-c\. ...(1)
Now iv2-f 2tvz cos A+z2—1 =0 gives
iv = -z cos A + V(l —z* sin* A) ...(2)
and also z=— tv cos A + VO - tv2 sin* A). •••(3)
(2) gives the branch points in the z plane, i.e. z= + cosec A.
(3) gives the branch points in the iv-plane, i.e. iv= ±cosec A.
Let an ellipse in the z-plane be (the foci being the branch
points)
| z+cosec A | + | z-cosec A \=2a. ...(4)
From (1), we can write (4) in the form
| z -b V (z2 - cosec2 A | +1 z - V (z2 - cosec2 A 1 = 2a .. .(5)
i.e. | z sin A-fiV(l— z2 sin2 A) |-}-| z sin A-zV(l -z* sin2 A) |
—2a sin A.
(2) gives | z sin A+i (w+z cos A) | + | z sin A
-/ (h>4-z cos A) |=2a sin A
or | ize~ix-f iiv | + | —ize*^' - iw \=2a sin A
or | z + weix |-t-( z-b we"1*' |=2a sin A
or | - tv cos A+ \/(1 — w2 sin2 A)+ we** |
+ | - iv cos A±v/(1—>v2 sin2 A)-f we~ix \—2a sin A
or | wi sin A ± i\ (w2 - cosec* A) sin A | f | —iw sin A
+/V (»v2 - cosec* A) sin A) sin A |=2 a sin A
or | iv± \/(iv2—cosec2 A) | + | »v + V,(H’2'_cosec5 I —2n.
We get | »v±cosec A |+| iv + cosec A|=2n from (1),
i.e. | iv-f cosec A | + | yv- cosec A |=2a,
giving an ellipse in the iv-plane with its foci as the branch points
± cosec A. Hence the result.
Definition of Branch Points.
Wheniv=/(z) is a multiple-valued function of z, then any
point z for which two values of iv become equal is called a branch
point of iv.
A line which connects two and only two branch points is
called a branch line or a cross line.
The simplest example is iv2=z. The two-valued function has
two branches iv, = V*, >r2= - y/z.
Conformal Representation (<Continued) 137
The points 2 = 0, and 2 = 00 are branch points and the posi¬

tive part of the x-axis joining 2 = 0 and 2 = 00 is a bianch line.
13. Discuss the transformation involving confocal conics.
2»v=(a - b)
Suppose z = zei9 ; then we get
2u=(a_-
=-(a — b) r cos 9-r*—.— cos ^ 1 . ..(1)
= (a — b) rsin#— sin 0 j
Eliminating 6 from these equations, we get

u
■ = 1 -.-(2)
Hence we see that the concentric circles (/.*. r = r) with origin

as their centre in the z-plane are mapped into the confocal ellipses,
the distance between the foci being 2y/(A--B2), where A and B
are the semi-major and minor axes in the »v-plane.
Here A*-=-4 [ (a-b) ,
1 T aA-b~\*
B*=-$a-b)r- — J .
Hence \/(A2 — £2) = V(°2 “

Eliminating r from the equations (1), we find that
4 u2 4va
4 <a2-62).
cos'" 6 sin2 0
When 0 = constant. the lines emanating from the origin are
transformed into the hyperbola, a result to be expected, because
both the families of curves in each plane must cut orthogonally.
If we take 6 = 0, n = 1, there is no loss of generality and we
consider in detail the transformation
2 iv
=( + 2 0- .. .(3)
(Agra 1950)
Clearly tv becomes infinite at 2 = 0, and
B-O-n)
which is zero at z=± 1. Hence the two points 2= ± 1 are critical
points, where the conformal property does not hold good.
On solving (3) for 2, we get

z~w± y/(w2—$
and the inverse function is a double-valued function of \v.
From (1), we get
= cos 6
• • •
r- sin U
(a) When radius of the circle in the r-plane is unity, i.e.

r=l, we have w=cos 6, v-0.
As 6 varies from 0 to 2n in describing the unit circle in the
r-plane, the domain described in the w-pJane is the segment of
-1 I u
H’-plane
the real axis between the points 1 and -1 twice, or the ell
minor axis 0 and semi major axis equal to 1.
(b) Eliminating 6 between the equations in (4), we get
u v-
+ = 1 ...(5)
iK-)
which is an ellipse in the tv-plane and it corresponds to each of
the two circles | z | = — as the equation of the ellipse does not
I
change on changing r to — .
(c) When r increases, both-—the
axis of the ellipse, increase, that is, the exterior of the circle |2j=r.
and interior of | z | = _L corresponds to the exterior of the ellipse
i.e. the two shaded domains, viz.,
r <\z \ < R, <\z \< j
are mapped conformally on the region bounded by the confocal

ellipses.
‘Ty- plane
Hence as r->oo both the semi-axes tend to infinity and we see

that each of the regions
| 2 | > /• and | 2 ! < —
are transformed conformally on the region outside the ellipse.
(d) As we have seen that after solving (3) for 2, we get

3 = »v± \f (iv2 — 1),
which is an inverse function and two-valued. We note that the

lower sign i. e. the transformation
z=w — y/(\v2 — 1)
transforms the area outside the ellipse conformally into the inside
of the circle z \ = r. Hence the lower sign is the correct one to
select because the point z = 0 inside the circle must correspond
to the point tv = oo. Had the other sign been taken, the points
r = » and =oo would correspond. Hence the region between
two concentric circles in the z-plane is being transformed into the
region between two confocal ellipses in the tv-plane.
Thus we see that if r is taken as near to unity as we like, the

domain | z j > 1 as well as the region I - | < 1 is transformed
conformally on the whole tv-plane cut along the real axis from
-1 to 1.
In this part we consider the region mapped by the radial lines

0 = const.
Let S be equal to ±a such that 0 < a < ttJ2. Then
r/= ~ ^r + * ^ cos a, v=y (V— yj sin a .. .(6)
where r varies from 0 oo.
Eliminating r, we get w*/cos2 a — v*/sin2'a= 1 ...(7)

which is a hyperbola having its foci at tv— + I.
(ii) When r varies from 0 to oo, the point (u, v) in the
n -plane describes the branch of the hyperbola (7) on the right
hand side of the imaginary axis. As it is quite obvious that v
becomes negative if r lies between 0 and 1, hence as r varies
from 0 to 1, the lower part of the hyperbola is described and when
r varies from 1 to oo , the upper part.
Similarly the second branch of this hyperbola is the total

mapping of either of the two radial lines 6=-.-n - a or 0=*w + a.
iv plane r-plane
(iii) Consider the wedge defined by a < arg 2 ^ ft. This
is mapped conformally on the domain between the two branches
lying on the right of the imaginary axis of the hyperbolas
given by
z-plane
(iv) When
0, Jr).
When r->ao the branch of the hyperbola

m2 v*
cos2/? sin2/?-1
tends to the entire imaginary axis in the iv-plane.
Hence the wedge-shaped region defined by
a < arg z ^ n/2
is transformed conformally on the region bounded by the branch
|42 Complex Variable
of the hyperbola to the right of the imaginary ;axis and the imagi¬
nary axis itself.
H’-plane 2-plane
(v) Clearly with the combination of the results (ii) and (iii)
we find that the region defined by
a < arg Z < ir-a
bounded is mapped conformally on the region between the two
branches of the hyperbola as given below
w-plane z-plane
We similarly conclude that the area exterior to the hyperbola
(7) is the image of the region defined by
■* f a < arg z < 2tt —a
in the 2-plane.
Remark. Similar to that as in ellipse we see that the region
outside the hyperbola in the w-plane is transformed conformally
on the rigion defined by
a < arg 2 < it—a.
in the 2-plane by that branch of
2=H' + V'(tt'!“ 1)
for which \/(w2— 1)—W for h>=0.
The second branch for which VOr2- 1)=—* for w=0 maps
the same region outside the hyperbola on

it-fa < arg z < 2tt — a.
(vi) When 6->0, + and v->0.
Hence the point (u, v) describes the part of the positive real
axis varying from (1) to <x> twice as r varies from 0 to >=. Thus
we see that tlie domain defined by
0 < arg z < a
in the z-plnne is mapped on the interior of the branch which lies to
the right of the imaginary axis of the hyperbola
u2/cos2 a —v2/sin2 a=l
with a cut along the positive real axis from 1 to ».
Z~ PLANE
,v.. PLANE
14. The transformation z=c sin »•’.

We have
x+iy=c sin (u+iv)t c being real ...(1)
—c sin u cosh v + ic cosh u sinh v.
x=c sin u cosh v, y — c cos u sinh v.
(i) When v = con$t., the point (.v, >•) describes
Xz V“_ ^
c* cosh2 v c2 sinh2 v ’ **
which are confocal ellipses for different values of the parameter v.
Let us take a rectangle in the
w-plane defined by u=±n/2, and
v= ±A. As cos u is -j-ve between 0
to tt/2 or -77/2 to 0, then for v = -J-A.
y is positive. The value of x: lies
between c eosh A and — c cosh A,
that is, the half of the ellipse on the
+ ve side of axis of x in the z-plane
is covered.
Suppose now u= - then >•=0 and x=-c cosh v. Thus if

v varies from A through 0 to - A along BC, the side of the rectangle,
x takes the values from — c cosh A to —c. and back to —c cosh A ,’
i e. it passes from A' to the focus H and back from H to A'
(as given in the figure).
Again when v=—A, r-describes half of the ellipse on the
-ve side of x-axis. When u=£tt, again ;- = 0, and x=c cosh v,
so that x takes the value c cosh v to c and back from c to c cosh v,
i e.t it moves from A to the focus S and back from S to A.
Thus to the boundary of the rectangle in the iv-plane, the
corresponding boundary is the ellipse in the r-plane with two cuts
from extremities of the major axis each to the nearer focus.
2-plane
Clearly the lengths of the axes increase with the increase in

the value of v, and decrease with the decrease in the value of’.
Hence the interiors of the rectangle and the ellipse correspond.
The inverse function of (1) is
H’=sin-1 (z/c)
which is a many-valued function. Had the cut plane been used, we
would have used the cuts from S to infinity along the+ve real
axis and from H to infinity along the negative real axis.
15. Similarly we can deal with a different transformation
with z = c cos it-.
16. (i) Wedge or Sector or Semi-plane.
Discuss the transformation
H’ = 2l /P. . . . ( I)
We see that pc'9 = (re19)x/P

where z=rei9, »t-=p>
or p=rl P, <!> = 6lP.
or ...(2)
Conformal Representation (ContinueJ) 145
Now the infinite wedge

0 < arg r < pit
v//^/ O / / ' # V// ^ ' ' ' ^ ✓ / //

/
>
,,,,
;;;;
. / / // / - y / ^ /*
'""'s’, s''' ’ ' f'

,, " " 's'',/ '"
' s s ' ' / S * s ' ' ' y ',
"'y' S'"" ''I,'''
: s' '".'" ' '<' ''
n-plane
having its vertex at the origin and one arm of the angle along
the positive *-axis is transformed into the upper half u-plane
i. e. / (w) > 0.
We see clearly that the arc cut off from the circle j z\=[
by the wedge is transformed by (1) and (2) into a unit semi¬
circle in the w-plane [because if r=l, p=\ and ^ = tt, hence
the resuit].
(ii) Semi-circular domain on half plane.
Let us consider the transformation
...(1)
Clearly it can be the resultant combination of two
transformations
w = ” and £ = p)
* 14-z •••K*)
From the first equation of (2) the quarter plane becomes

the half plane with the help of the transformation (1)
and by the second transformation the semi-circle of radius
unity and the centre at the origin lying in the upper half plane
is transformed into the first quadrant of the £-plane. Hence
the result.
17. Exponential and Logarithmic Transformations.
Consider the transformation
w—e*.
(I) • •
Making use of the cartesian co-ordinates and polar co-ordina-

tes in the z and w-planes respectively, we get
Pei*=ex+ly.
Equating real and imaginay parts, we have
P =e X
'('=y
>’=a in the z-plane gives the line </»=constant radiating

from the origin provided 0 < a < 2ir.
y
•* *e' *' S.' s* / ■S'"',/""//

' ' y// s / * ' *s " '
'Vs,,,// -
T=c
• •
o Jt
z. PLANE
(a) Now the region bounded by the lines y—c, y—d in

r-plane, where | c—d | < 2tt is mapped into the wedge
shaped region of the w-plane, the angle of the wedge
being y=| ! = | c-d\. Since which is never zero,
the representation is conformal throughout the interior of

these regions.
Remark. The particular region defined by_v=0, y—n in
the r-plane so that | c—d \=n, i.e. the wedge becomes a half plane
in the »r-plane, /. e. I (»r) > 0.
(i) The semi-infinite strip on the negative side of the
axis of-y i.e. -oo < x < 0,0 < .F < * is 'clearly seen to
correspond to a unit semi-circle in the upper half of the ti'-plane.
Now x=c gives a circle in the w’-plane. Particularly
when x=0, the imaginary axis is the boundary of the unit circle
y*
2-plane
(ii) Consider the inverse transformation /. e.

H’=log 2.
If 2=re<ff, then a value of w is
M + /v«*log r+iO.
Hence r/=log r, v=6.
When r goes from Otoco, u goes from -=*> to oo ; the
147
Conformal Representation (Continued)
angular region a < arg 2 < P corresponds to the infinite stnp

a < v < P in the w-plane.
If the general value of log 2 is being considered
• :r ivslog r-M (04-2kn), .
r sts
logarithm. ^ remember that though is one-valued
function but log 2 is an infinitely many-valued function of-
*18. Consider the transformation
l±=
- € •
.14 2 1
=1
1 rrz= / 2-i*
1 l4e'® ^0
If z=e<tf, then w = - -co* 2 *
Hence, when 2 describes the unit circle in the 2-plane
i. e. as 0 varies from 0 to 2^. then

w4iv= - cot 0/2
gives w= — cot 0 or m = - cot it
f. e. iv varies from — to +« . or « varies from — to> +«■ -
Hence iv describes the real axis from to + «>. Theonly
in the z-plane^s^represented“sin.pl^ on the upper
pagVstyîtf ihe^'chapueô^bifinear^ransformations^but here

we
" ~ have given a different method.]
"-C- . v
4aw cot a (rs

fd . . f \]
19. If z=i_i_2li’ cot a - »v* ’ V 4/
Prove that when w describes the unit circle, z describes m ice
over an arc of a certain circle subt ruling an angle 4a al{l^n^S)
Since w describes the unit circle, we have,

| vv |=1 /. e. *v = e,<?, (0< </* < 2rr).
Let z=rei0. Then the given transformation may be written as
4ae,l? cot «
re*9
re'9 =
= f+ 2el* cot a - e2if
1 _ 1 42e<? cot a-e2,?
<r‘# =
or r 6 4ae'7 cot a
1
1 .
(cos „
e-i •
sin ,an a
0)=~ — +
[c-,? _L 0 rrst
- f>‘7
2 cot a - e»]I
/. e.
1 (cos « -1 sin »t = ,aA? (2 cot * - 2/ sin ■/ ].

r. e.
Equating real and imaginery parts, we get

cos 8 I
-= «— i. e. r=2a cos 8
r 2a
1 . . 1
and — sin B=— tan a sin 4>
r 2a
sin B 1 .
or tan a sin from (1)
2a cos B~~2a
or tan # = tan a sin <£. ...(2)
N’ow (2) shows that as <!> varies from 0 to nf 8 varies from 0 to
“ and back from a to 0 and as varies from n to 2^, 6 varies
from 0 to - « and back from -a to zero as is evident from the
following table.
1
77 77 377 377 In
0 0 7T - 277
4 2 4 4 2 T
<!> 0 tan-1(
'tan a'*
^ \/2 >
) a tan-1(
' tan
< V2 >> ° tan l V2 ,
\1 — a_ *rln-1/
lan 1
^
S “
tan a
V2 ,n
\J
Hence (1) represents an arc cf a ciicle subtending an angle
2a at the pole which is at the circumference /. e. the arc subtends
an angle -Ja at the centre, (see the figures below).
Exercises
1. If H's - /f cot (2/2), where c is real, show that the
rectangle bounded by .v=0, x=n; y=oo is conformally
represented on a quarter of the u’-plane.
22
2. Show that = t maps two of the four domains into
which the circles lr-l| = V2 and | r+1 | = V2 divide the

r-plane conformally on | »»■ | < 1.
3. Prove that, if 3r* — 2nr-f 1 = 0, the annulus I/-y/3 arg z = a.
[Hint. 2>v=3z-|--j , and proceed on as in question no. (13)].
Show that the relation »v=cosh z transforms lines parallel

to the axes in the 2-plane in general intQ confocal ellipses
and hyperbolas in the w-plane and determine the exceptional
cases.
Prove that there is one correspondence between the semi¬
infinite strips defined by
x > 0, 0 < ;- < 77
and the upper half vv-plane.
[Proceed as in question no. 14.]
Show that the function w = sinh r; takes every value w such
that R (iv) > 0 exactly once in the domain
R (2) > 0, - \ < J(z) <
Prove that if u =i ^z + then (in general) circles | z ^cons¬
tant and lines arg 2 = constant correspond to conics with foci

at w— ± 1, in the tv-plane. What exceptions are there ?
u— 1
By considering the argument of , or otherwise, prove
that in general any circle through two points w = ± 1 corres¬

ponds to two circles in the 2-plane. In particular, show that
| w |= 1 correspond to
x'+y- + xy= 1
and indicate in a diagram the regions in the 2-plane which
correspond to 1 »v I < 1.
Map the outside of the ellipse
x*la-+y*ibt = 1
into 1 »v | < 1.
Show that the transformation
(l-j-23)8—I (1 -zV
(n-z3)2+ru - 2y
maps the domain
| z | < I; 0 < arg 2 < tt/3
conformally on | w | < 1.
Discuss also the correspondence between the boundaries
of the two domains. (Agra, 1963)
^Hint. Make the following substitutions Z=z3.
v 1-fZ t—l
Show that iV=log W(z/a)-l) represents the strip v=0,

*=oo, tfBir on the interior of the cardioide
r=2a (1+cos 8) in the z-plane cut along the real axis from
the cusp to x=a.
Show that, if c is real,
Z’2~=ieiw
z+c
conformally transforms the strip v= — oo, v=oo, u=0,
into the circle | z | < c.
Discuss the transformations given below,
(a) H-(i±fV
(b) w=(jH)'
« ~G3'
and show that the certain semi-circle in the r-plane is the
conformal equivalent of the positive half of the w-plane.
Find those semi-circles.
[Proeceed as in Ex. 9J.
CHAPTER V
COMPLEX INTEGRATION
5*1. Rectifiable Curves.

Let us first consider briefly how
the length of the Jordan arc given by
*=* (')+/>(/)
where a < / < /?, may be defined,
before we discuss the Cauchy’s
Theorem of integrations.
Let the equations
x=<f> (/), y = if> (f),
wbeJre * < < !*• rePresent a plare curve. If the interval (x, B) be
subdivided into subintervals by points
a = 'o. *1* ^2*. »tr » • .»
corresponding to of the given arc, then the points

on the curve corresponding to these values of t may ba denoted
y o. Pi, ...Pn. The length of the polygonal arc P0...Pn which
is obtained by drawing straight lines from P9 to /> P. to P2 and
so on is measured by
I (Xr - ■*'-,) + ' O’.-yr-l) I
***r?S« the Iparticular mod= of subdivsion of the

u ' (■ the arc be of the type such that this summation
has got definite upper bound /, then the curve is said to be
rectifiable and / is called the length of the curve.
th». !h Ca" b! S the nccessary and sufficient condition

h“ ,h'ar.C Sh0U,d be ^ctifiable is that (,) and * u) should be
of bounded variation in P). „ ,y (l) and 0. *n
tcan be proved that the curve which is defined by *=</, (,)'
*().*< r < A ■s rectifiable and that its length s is given by
[{«/■' (/)}*+{*' (r)}2]11- dt.
Proof. We have first to show that s can be made to differ

by as little as we please from the sum
| ’I* (tf) I Ur tf-1^ ...(1)

by making the greatest of the tT—tT-x to be sufficiently small.
Now we start with the important inequality
|{ \ s\ ~ I I) I } I
= 1 {(^v — *r-\) + l Ov-lr-l) I “I*' + Ur) i ('r-'r—l)} I
<|{(*f-*r (M (/r-^r-l)} I
+ KOY-;Y-l)-0' Ur) Ur-tr-1 >} I ffr0m (3>i
where *,. = ■/> (fr )etc.
By the Lagrange’s mean value theorem of Differential
Calculus, we see that there exists a number " beloning to the
interval (tr. tr.x) such that
Xr-xr.x=<f> (tr) —A (tr-l) = fA, ( ”r) (*r ” *r-l' 1 • • •
;v— J’r-15^ Or) — 1/* Ur-U-'P' ( >) (/r”rr-l) J

and hence
i\/r < { | <!' ( ~T)~^' (Ml + I ^ ( "r)~*P' Ur) I } Ur-tr-l)y -(5)
But by hypothesis the functions <f>’(t),•/>'(t) are continuous,
and hence uniformly continuous. Then we can find an arbitrarily
positive number e, as small as we please, and then we can
choose another positive number 5 which depends only on €,
such that
| $ (/)-«// (Ol < €,| f (/)—(01 <
xvhere j < S. If the greatest of the numbers rf-/r_, is
less than S, such that
Mr < 2e(tr-tr.x), by (6), ...(7)
Thus we see that
Complex Integra ion 153
n
< 2e V Ur-fr-ih from (7)
r= 1
— 2z {t i — t^ t% ~~ t \ t-f ~ t'l • • T t n t,,-i )
= 2e [tn -t0]=2e (B-x).
Hence by definition of the integral of a continuous function,
the sum 5-0 tends to the limit
JW (/) + ty'(/)} I </'=]%{'>' ('if+(f (Of]1'' dt ...(8)

as n tends to infinity, and the greatest of the numbers tT — tr-\ tends
to zero. However, | s-s0' can be made as small as we please,
by making 8 sufficiently small; s must also tend to the same limit
as defined by (8), i.e. we have thus proved that a regular arc is
rectifiable and the length is defined by
(Of+«' <')f! dt.

5*2. Contours. Let there be two arcs
AB and BC having B as the only common
point; and also let their lengths be s and s’
The complete arc AC is also rectifiable whose
length is j+j'. Hence we conclude from it
that a Jordan arc consisting of finite number
of regular arcs is also rectificable. Such an
arc is called a contour.
By a contour we mean a continuous

Jordan curve consisting of finite number of
regular arcs. Clearly a contour is rectifiable.
5*3. Connectivity of a domain. Simply connected and multi¬
ply connected regions. A region in which every closed curve
can be contracted to a point without passing out of the region is
called a simply connected region ; otherwise the space is multiply
connected.
S.MPLr CONNLCTEO MULTIPLY CONMLCT ED

5*4. Riemann’s definition of integration.

Now we will define the integral of a function of a complex
variable along a regular arc L which is being defined by x—<f> (t),
y=<l> (t)y « < / < P.
Given, a function /(z) of the complex variable y, which

may not be analytic necessarily, but continuous (along L, a
regular arc having its end points A and B), such that
/(z)=w (x, y) + iv (*,>>).
Let z0, zt...zn be points on the curve L, z0 being A and z„

being B. Now consider the sum
ZfKr) — 0)
r= 1
Kr being the point of the curve between rr_, and zf, where
Kr—?r+t>3f. For brevity we write u(Er, rlf)=ur and »'(£,, r(f.)=vr.
Hence (1) equals to
n
^ W (£r» C^r» *Jfl) (“ r — *-r—j)
r= l
=£ {ur+/vr} {^r-arf-,)+i 0»r —>V-i>}

r= 1
Now, as before, by the mean value theorem of differential
calculus,
xr-xr_x=<1> (tr) -4> (tr_x)=<t>' (tr)(tr-tr.x),
>’r-yr-1=^ (/r-i' = ^' (V)
when /r_j ^ Tf ^ /r_j ^ Tr ^ fr.
Hence we can write the sum as
^{(wr+/vf)} [{*' + (r/)} {(/r-/rM)}j. ...(3)

r~l
Since all the functions concerned are continuous and hence

uniformly continuous, we can, given e, find a S which depends
on s only, such that
I M' (*r) - w (xr, vr) </*' (tr\ | < € etc. .. .(4)
for every r, such that each
| tf — tr-x | < 5.
Now/(£r) (zr-zf_1)={ur^'(Tr)-vr{/»' (t/>
+ '{V*' (Tr)-f Wr^'(V)} (/r-'r-l> ...(5)

Complex Integration 155
and/(2) ((,)+//' (/,)-!/' (ir)}+i[r.,/ (,r)+„0' (,r);
Hence |/(Cr) (zr - +-,)-/(2) {*' (2r) + /0'(,)} (2,- ‘ (6’

« ! M' (t,)-vr0' (Tf)-u^' (/,)+»*' (/,)] (,r _,r , (
+ 1 fV*' <f,) + <M4' (rr'i- V*' f/,>J )

< [ I “//■' (*,)-«*' (»,) l+l y,0' (tr')-Vf (fr) |
+1 (T,)-V*' (f,) 1 + 1 H0' (Tr')-B0< (,r) |j )
/.e., we get ^ <£+e + £+‘) ) = 4e </,-/_,) by (4). ...(7,
fl
f/(W - -r ,)} -/ (a) (;,)+/0- </J} (2, - , )
< rS 4e ('r- 'r-,) by (7)
= 4e (/B-/0)«4e C/3 — a).
A$ and ,he norm of we get in the limit for

the whole sum as
fp
•P
J,a (*» ^+/v (*. jO} {*' (O+/0' {/)} di. ...(8)
Thns hmuis taken as the definition of the complex integral
of/(r) along the regular arc L and is written as
/(-) dz — J^ (u + iv) (dx-j-i dy)
I
~\ L (u v dy) +1 (v dx+ u dy).
of . T"r a" ^nte8ral °‘ /fZ) al°ng a contour C- "-hich consists

Of a finite number of regular arcs Lr, is given by
fc/(z)dz=z f(z)dz.
1. Evaluate /,-z* dz along the line 0.4 of the adjoining

figure between the limits x = 0, p=2+i.
z*=(x+ iy)* = at* -y*+2xiy
and dz=dx+i dy.
+' L, o. I2*.*’*+(•«’->*) 4-)l.

|*V3*4&P -
Since x=2y on the line OA, then dx=-2 dy.
and /^j1 (3y2.2 dy-4y* dy)+i J (8y2 dy\3y2 dy)
= j1 2yi dy\ i j* I Iy* dy
=(f+^131)-
Now let us take the curve C as the path OBA as shown in
the figure ; then we have
^ 2' dZ+\liA Z' ^

z = x along OB so that dz=dx and z=2+iy along OA so that
dz=i dy.
/2=|2 a:2 dx+i j' {(4-/-)+4/» dv
= f + /(4-i)-2
2 Hi
“"3+ 3 *
Hence we see that /*«/1 In fact
ri T+< 2 .11
-3+,T-
Also the integral when evaluated around the closed curve
OB AO is zero, which is clearly the property that the integrand
22 is regular throughout the region including the AOAB.
2. As another example, letf(z)=is.
/3 = | ^ 2 dz.
=f (x - iy) (dx -f- i dy)
consists of the upper half of the circle

r= 1.
Let a:=cos 0, j>=sin 0 along C\, that
is,
z=e’0, dz=iei9 dd.
/3 = J° e i0 ieia dQ-i J_ d0=— m.
Similarly the line integral between the same limits along the
lower semi-circle C2 is given by
Thus /39^/4. Hence the integral around the complete circle

C in a counter-clockwise direction does not vanish because
— /3 = W - ( — 7r/) = 2rr/.
For the unit circle, we have zn=l, hence ; we can write
~ = y» J*e* =
Note. Generally we do not write
jr/(r) </z=fZJ‘f(z) dz.

...G)
as the value of the integal \Tf{z) dz depends on its form and
not in general on the initial and final points of the curve. We
can write into the from (1) only when |r/(r) dz is independent
of the path connecting r0 and which is evident from the above

examples.
Evaluation of the integral by ab initio method.
3. Evaluate /= d-
J/'t?
We have ab initio
/“Lint v (zT-zr_j
H-*cc r~ 1
= Lim [(:,-z0)
n—> oo
+ (z2 — -l)+...(2n-Zn—j)J
= Lim (-i, z0]
«->oo
«) since zn=/?. z0=a
= length />£>.
If both « and /J coincide, i.e. if ,he eune is a closed cne, /= 0

4. /=
-f)PQ'dz''
n
~Lim £ I c. -
n—>oc r= 1 1
-Lirn^ n c, - c. l+|z, - 7, |+j ,3 - ,, |+...
••• + 1 !j
Lim [chord zxz0-f chord
I_*
//—>*00 * ' • • 4 chord rnrr,_j)
arc />£?.
Complex Variable
158
5. Integrate I=z\Pq_z
/=Lim £ (^r) {(zr“2r-iH

n-> oo »=i
= Lim 2? ^(Zf-^r-i)}. taking k-*f

n-> oo r=l
Also/= Lim Z {z,-, (rr--%-i)i. taklnS

n->oor=l
adding, /=i Lim .£ (zr2 —-r-i^*

n-*<x>r=\ *
= * Lim {zt‘-V+z1I--12 + ---<:»2"”n-‘)}
«-*■ oo
=i Lim (zn2-z02>
=?!z^2 as r„=iS,2o=a-
Clearly if the contour is closed, the integral is zero.

5-6. An inequality for complex 'ntegrals length l
Theorem. ffW - C '
where it satisfies the inequality
. r I
<: Ml.
>JC IIc'W
r « numbers cannot
Since the modulus of a sum of n real n
exceed the sum of their moduli, we have
I 2 F (*,) (w-i) I < Z ! F (*) 11 <*' ' * *
and taking limit, we get
ft
Lim
n->*,
[V(x)rfx]<[VWI^
i. e.
such that norm of {xr—xr-x)-+0. ,
Proceeding in the sum way, we shall have the relat.on,
|jc/W*|<[cl/WII*l J ^
M jc | dz |, since |/(-) I ^ A*
< M jc y/{dx*+dy')t since dz**dx+i dy
=M [ ds==Ml.
| /(--)</--1 < ML
Alitcr. With the notation of article 5 4,
j*c/(r)<fe (=||^ (m + i'v) {f (/) + «<//(/)} dt

j‘ | m + /v 11 </>' (O + /0' (0 1 dt
^ >/(«*+**) ({-/>' (O}* + {0' (0>,],,i dt
< M %/[{*' (O}* + {0' (/)}*) <*
< .1//.
Note. If C is a contour, we make
it a convention that the positive sense
of description of the contour is anti¬
clockwise.
5-7. Cauchy’s Theorem. (Elementary Proof)

lff(z) is a regular/unction of z and iff' (:) is continuous at
each point within and on a dosed contour C, then
The proof which has been given below is based on the two-
dimensional form of the Green’s theorem and it requires the
assumption of the continuity of/' (z).
Let D be any closed domain which consists of all the points
on and within the contour C.
Now we write the left hand integral of the equation (1) into
the curvilinear integrals as
+ (dx f / dy)
= |c (w </* — v dy)+i (v dx+u dy).. .(2)
Making use of Green’s Theorem, which states that if

dP dQ
are all continuous functions within the given domain D, then
\ypdx+Qdy)=\ID(g-yr) * ’
Since/ (z) -itx'MLr=v„ — in* (by Cauchy-Riemann equations)
and/' (:) is continuous, it follows that ux, vx, //„, vy are all conti¬
nuous in the domain D. Hence the conditions of Green’s theorem
are satisfied. Then we get from (2), with the help of Green’s
theorem
i /(--) dz=
■ J Jo ■ ©+s) *1*+1JId (&-%)•**■
— 0 fusing Cauchy-Riemann equations].
Remark. Goursat first proved that the assumption of the
continuity of /' (r) was unnecessary and that Cauchy’s Theorem
still holds if we only assume that/' (r, exists at all points within
and on C.
The difTerentiablity of the function/ (r) are the consequences
of this theorem.
5*8. Cauchy’s Integral Theorem* :
Ij f (z) is regular at all points within and on the closed contour
C, then
\cf <-)
(Luck. 44, 47, 48, 49; Agra 40, 43, 55, 58; Raj. 59)
First we prove that the integral exists. It certainly exists,
since a function which is analy tic or regular is continuous and
a continuous function is integrable.
Let the contour C be y
divided into a network of
squares by drawing lines
parallel to the axes of .v and
y, with the contour C as its
outer boundary. The whole
of the contour is then divided
into complete squares or G
part of the squares such that
jcf(:)dz=sjf(z)dz ...(1)
•It is also called Caucby-Goursat theorem.

Complex Int egret ion 161
where y denotes the boundary of a mesh described in the anti¬

clockwise sense.
Let z0 lie inside the square of

contour S having its side equal
to a ; then
< 4\/2as=4\/2 (area of .S').. .(2)

[V I 2— z0 | < the diagonal of
the square = -v/2« and f\dz\=fds
— the whole length of the contour=4a, because the sum of the
four sides of the square^4a, hence the result (2)].
Before proving the main theorem, we prove the following

preliminary theorem or lemma called as Goursat's Lemma.
Lemma. Given any positive number it is always possible

to divide the region inside C into a finite number of squares and
partial square whose boundaries are denoted by C0 such that a point
r„ exists within each C0 such that
f (Z) — f ( Zq)
z-z. -/' (V <*
for every point z within C0.
We shall prove this
lemma by the well known
process of subdivisions.
Suppose the domain C
is covered by a set of equal
squares which have been
formed by drawing lines
parallel to the real and
imaginary axes. We remove
those portions of any square
that lie outside the closed contour C as in the adjoining figure.
Let us suppose that, for a given «, there is at least one region

of these sub-regions in which (1) does not hold good. If it is a
square, let it be divided into four equal squares and if the square
is not a complete square, let the whole square be divided in such
a manner that the portions that lie outside C be discarded If
again in any of these smaller subregions, there does not exist any
point ?„ such that (I) is true, we should again subdivide that

region in the same way, and so on.
There are now a priori two possiblities : (i) process of sub*
division may terminate after a finite number of steps, (ii) it
may go on indefinitely. In the first case the result is established.
But in case (ii) there is a least one region, say ct0 which can be
subdivided indefinitely without obtaining the required result.
Then after <r0 is subdivided once, there is at least one of the
smaller regions oi in which (l) is false. So in this way we shall
have an infinity of regions a0, alt a2.. ,cn.... where each of these
regions is contained in the preceding one, and for which (1) is not
true. Hence there is a point z0 common to all the regions on.
Now the dimensions of g„ tend to zero as n increases indefinitely,
i e. for each positive number 8 > 0, there exists an integer//0
such that all points a„, from one onwards, are the interior points
to the circle
r0 : < 8 when n 2 n0, say.
But / ir) is differentiable at z0, i.e.
fiz) -f(zQ)
Lim =/' <"o)
- -0 z-z 0
which can be written in the form
< € for j z — z0 | < 8.
Hence a contradiction. Thus we have proved the lemma.

With the assumption of this lemma, we shall prove the
Cauchy’s theorem.
Proof of the main theorem. Let the given closed contour C
be sub-divided into squares and partial squares whose boundaries
are Cr (r = 1,...«), and so a point ru exists for which

f<z)~~f(zo) * (-j <c by the previous lemma. ...(1)
Z Zo
Let e be any point on the boundary Cr. The value of/(r)

at ativ point r of Cr can be given as follows :
/(c)=/(r0) —(r„ -z>/' (r0)-Kr-r0Wr ...(2)
where I cr 1 < €-
Suppose the integral has been
taken in the counter-clockwise sense
around each Cr Then the sum of
these integrals will be the integral
around the closed curve C in the
anti clockwise sense, i.e.
f 4m a -\ J n, r
Since the line integrals along the boundary which is the

common boundary line of every adjacent subregion cancel each
other, the sense of the integral in a region is opposite to that of
the other as seen in the figure. Hence the only parts of the
sum that remain are the integrals along the arcs which are the
parts of the curve.
Substituting for/fz) from (2) in (3). we have
f
J C
f(z)dz= 2 f
r= 1 i Cr
{/(z„)-(r0-2)/' (*„)+(*-*„) £r} dz
- ^ ["{/(*•)-*•/' (*0»f dz+f{zA zdz

r=\L Jcr J Cr
+ *r Jc (Z-Z0) rfzj. ...(4)
From the definition of the integral, we have
dz~'L\mn_yx I as-max. | zf-r,_x I-+0

J C, ‘-1
= LinW> f {(2,-r0) + .. (r. + r.-j)}
= LinV»0(z" ~z°) = 0> because for a closed curve z„=z0. .. (5)
Similarly
z dz= Lim *->cc i t z$ — z) — Lim (z$mmZ9^)

f Cr f-.l
oo
9m 1
i
= J Lim,,^^ 27 (z,* — z*,_g)-*-0, for a closed curve.

I«1
Then from (4), we get
j f (z) dz= 2 f (z - z0) f dz. ..(6)

>C r-iJCr
and hence J f(z)dz < 27

f-1
f (~ Zq) cr dz
< € e[ |*-r0||*-
r“‘J Cr
since j «r | < e. ...(7)
point r„ such that (I) is true, we should again subdivide that

region in the same way, and so on.
There are now a priori two possibilities : (i) process of sub¬
division may terminate after a finite number of steps, (ii) it
may go on indefinitely. In the first case the result is establishad.
But in case (ii) there is a least one region, say <r0 which can be
subdivided indefinitely without obtaining the required result.
Then after cr0 is subdivided once, there is at least one of the
smaller regions ol in which (l) is false. So in this way we shall
have an infinity of regions <r0, ct, <t2- • .<?»• • where each of these
regions is contained in the preceding one, and for which (1) is not
true. Hence there is a point z0 common to all the regions on.
Now the dimensions of an tend to zero as n increases indefinitely,
i e. for each positive number 5 > 0, there exists an integer n0
such that all points on, from one onwards, are the interior points
to the circle
I z-z0 | < 8 when n > n0, say.
But /ic) is differentiable at z0, ie-
Urn. . f,Z\ i z„)

z-r
which can be written in the form
| (-o) < € for 1 z — zQ | < 8.
Hence a contradiction. Thus we have proved the lemma.

With the assumption of this lemma, we shall prove the
C auchy’s theorem.
Proof of the main theorem. Let the given closed contour C
be sub-divided into squares and partial squares whose boundaries
are Cr (r = 1,.. .w), and so a point r0 exists for which
i f(z) —/(z0) | < e by the previous lemma. .. .(1)
z rn
Let z be any point on the boundary Cr. The value of f (z)
at any point r of Cr can be given as follows :
f (Z) —f (C0) — (Z0 — Z' f' (Zo) + (r“r0Wr ...(-)
where I «r I < €-
Suppose the integral has been
taken in the counter-clockwise sense
around each Cr. Then the sum of
these integrals will be the integral
around the closed curve C in the
anti clockwise sense, i.e.
fJ ( /(r) </r= r-l

r JO
f /(z) dz . .(3)
where 0 is areal number between 0 and I, describes a closed

curve C' situated inside D.
Then = \cfOUI:-\c,f(l)'K
= jc/(z)d:-^c/{a + e (:-«)]«*
V C = <7+0 (Z-<7).
I.
= fc[/(z)-0/{z-<l-0) (--0))1*
= J c<f>{z)dzt say
f
We can make the difference (1—0)
so small that | 8 j will be arbitrarily
small, for we can write
<f> (*)=/(*) -/ [z ~(z~a) (1-0)]
+ 0 -0)f[z-(z-a)(\-d) J.
Since 0 also.
fc f(2)
In the case of a boundary of any form whatsoever we can

replace this boundary by a succession of the curve fulfiling the
preceding conditions by drawing suitable transversals.
5*10. Extension of Cauchy’s Theorem.
Corollary 1. Two paths AMB and ANB. having the same
extremities, give the same value for the integral ff(z) dz if we
can pass from one to the other by a continuous deformation without
encountering any point where the function ceases to be regular.
Or
. f (z) is regular in a simply connected region D, then the
integral along every rectifiable curve in the region D joining any
"°VVen PUUUL
the °f
twothe reg,°n D is the sawe '• independent of
curve joining points. J the
Suppose the two points A and B
of the domain D are joined by the
curves of C0 and Ct. Let the closed
curve consisting of C0 and Ct be deno¬
ted by C, Then by Cauchy’s theorem
f(z) dz=0
j
i,e’ \c f(z)dz-\c /(z)dz= 0
ot fc, flz) dz=\c /(z) dz-

Corollary 2. Extension to any connected region.
If any domain D be bounded by a system of simple rectifiable

Jordan curves C0, C,, C2.. .Ck. .and if f (z) be any function of z
which is analytic and one-valued everywhere in this region, then
\c/^ dz+\cJ{2) dz+\c/(kZ)dz’>r '"+\cj^z) ^+--*=0

where each of the boundary curve is so described that the domain
constantly lies on the left.
Let us take cross-cuts AB, CD, EF,...; then
f(z)dz=0
such that this integral is being

considered along the complete
boundary, so that the domain
considered is always towards the
left. The path has been shown in
the adjoining figure,
(i) along C in the anticlock¬
wise direction,
(ii) along the cross-cuts AB, CD...taken along these

curves once along AB. CD..., and once in the opposite directidn
/. e. BA, DC.. .etc, and
(iii) in the last, along the closed curves C0, C,, C2t.. .in the
clockwise direction.
Hence,
+ fc/W4fc+-"fc./(z)*]“0'
because the integral along the cross-cuts is zero, as it has been
taken once along AB, CD., etc. and once in the opposite direc¬
tion BA, DC. .. etc,
/. c. fc/(r) dz=\c f(z) t/r-f J Q/(z) dz
-f . .|c f(z)dz,
provided the integrals along the curves C0, C„ C2 are also taken
in the anti-clockwise direction.
Particular Case. If there is only one contour,
f(z) dz.
5*11. Cauchy’s Integral formula.

If f (z) is analytic within and on any closed contour C and if a
he a point within the contour C, then
f (a) =~ [ dz.
Znl J z — a
(Pb. 1959, Agra 54, 56, 58, 59, 61; Delhi 59)
Let z=a be any point. Describe about
z — ay a small circle y whose radius is p
lying entirely within the contour C. >
The function <f> (z) = js analytic LJ^\ /

z-a
p) v
at all points in the ring-shaped domain D (
between C and y. Now we take a cross- *
cut by joining any point of C to any point ^^
of y. Thus we get a closed contour r as
defined and shown in the figure. Hence by Cauchy’s theorem,
Jj. <A (z) dz=0. ...(1)

By corollary (2), § 510, we have
°r {/— dz-(
1C z-a JTz-a ■
Now 2 hjr A
“5i/t as 1 *•
Since/(z) is continuous at z=a, for every given < > 0. thee
corresponds a 5 > 0, such that
\f{z)~f(a) | < c> when \z-a\ < b. # (3)
Now putting z-a^pe", vve get
U ts. „
=!/(.,) jo « +21 Jo {/«-/«}*
=/ (<0 + /a*
where i;’ {/(*)-/(*)} de
2*
4!=jr| J” /(*)-/(«) !/(*>-/<«) I de
1€J dd, from (3)

2?r
or 277 ...(4)
/a I < € 2w===£’
But the left hand side is independent of e and so /a must
vanish.
Hence we deduce
1 f f(z) dz
*n J cT? =/(a)-
5*12. Poisson’s Integral Formula.
//■/ (z) be analytic in the region \ z | ^ p, then
2T [R* - r?] / (Rei(*)
f (re19) =*u(r,») + iv (/•, 0) = ~ J o R*-2Rrcos (V-)+r*d*'
where re{0 is any point of the domain \ z\ <Z p and R is any number
such that r < p.
(Agra 1955, 58, 63} Punjab 59; Aligarh 46, 50)
Suppose z = re'9 is the point within the domain | z | < p,
where \z =R is any circle C within D jy
such that r < R < p.
Hence by Cauchy’s integral formula,
we get
f «-k J /<$>
Clearly RzJz is the inverse point of

the point z w. r. 1.1 z | = R, and hence lies
out side the circle and so is regular within C.
Thus we get by Cauchy’s theorem.

f(0
°=kr\ r- R72
dZ- ...(2)
Subtracting (2) from (1), we have
2*i Ju
=re'°, we obtain
1 /(/te,<?) (/?2-r2) fo1? id*
’ - M
2*1 , 1”
J 0 (Re‘? - re'0) (/?2 - tfe»?. re"'0)
~2rr H /(7?e/(?) (R2 - r2)

(/?e,<?—re'°) (/te”f?-re“'°)
fI
~2tt ,
l.
r /(*/»> (jp-H)
i?a-t-r- — 2/?r cos (6 — «/*)
,,,
...(3)
Putting / (z)=f (re,0) = u (r, 0) + iv (r. 0),

f(Re^)=u(R, (R, </>)
and equating real and imaginary parts, we get
»(r,*)=JL r
R*-2Rr cos(V-<f» + r-
0
" <*• *>
..</ />
" (R2-r2) v(*. 0) ..(4)
V (r, 0) <//.
-±j
R*-2Rr cos (y—*/») + r2
5*13. Theorem : Differentiation under the integral sign.
If f(z) is analytic in a domain D, its derivative is represented by
rtu-± f UlLdl
3 {K)-lm)c (z-Z)"
C being any simply connected closed contour in D surrounding the
point %. (Punjab, 1958, 61)
Suppose d is the shortest distance
of the point % from the given curve
C, /. e. d is the lower bound of
\z—%\ for each and every point on
the curve C. Hence
\z-%\ > d
for each and every point on the curve
C, /. e. the circular domain C',
I z~% I ^ will not contain any of
the points of the curve C.
Now from § 5*11, we have
/ (0 = — f —- d- •U)
2*1)02-% ’
f&+h) =1. jc ...(2)
where l+h is any point within the circle C', i. e. it does not lie on
the curve C.
Subtracting (1) from (2), we get
«+»-/
= J_/2f ffz)
2 mj C (z~Z) (Z-XîT) dz
-A f (_z-g-A + A)/(z) -
2Wjc <z-S>a (r-<;-/*)
or 1 f f f(z) _hf(z)_1
h 2ni]c KZ-Z-n))2
or fJkt!ArJAl_ if fM
h 2ni J c(z-$)~ 2
1 f / /f*>
e=27T/Jc^.J-^2 {z-Z-h)dZ
From the property of the moduli,
> I H ...(4)
Also/(r) being continuous and regular in and on C, it is
bounded so that |/(z) | < A/ on C. From (3) with the help of (4),
\f'W) -/(€> J_ f /fr) I '
! h ~2niJc(F^dZ\
-LJL f_m ^
Uw/' J c (r — ^ i2 (z-Z-h)
*\±\[ _\f(z)\\dz\
^2tt JciU-5>*I|*-*-A|
mf I </g I , A/
^ 2tt | // |)
IJJ A/ /
^ 2ir
/ being the length of C. Now it is quite evident that the R.H,S-»-0
as | h |-»0. Hence *
m±^-sc^
/. e. f (£> is differentiable and
dz.
f {®~iciz-l)'dz'
Conformal Representation 1 71
5*14. Generalization. If f(z) is regular in a region D, then

f (z) has at every point ? <f Dy derivatives of all orders, their
values being given by
We shall prove this theorem by mathematical induction.

2 f f(z)
3 </z, we have
2 W , 1 Cir-E)
f'(Z+h)-r to 1
f /'*>f 1 1
| dz
h 2w/ \c h [ (z-Z,-h)*~(z-^
J_ f /'*> (r-^2-(z-E-/j)s A-
Q~
2"/, Ic //
1 2 (z-^ /,_//- A-
f /,z-(
2rr/ .Ic A liz-O* U-5-A)*]
2 ! r z-5-/i/2 i
* dz
2tt/ •fc HlU-4riz-c-/o2J
2[ f z-E-/i
2 TTI fc IU-5)1
l
+ **«»-sec*:-l-h)
.2! i /'*>
CU-5)1 (z-Z-h)
+ dz'
Lim,o r /rn
/'->0 h Ini) C{Z- <fc+Lim A •••(!)
where / = yr /<*>_
»/] C (z-W(z-i-h)zaz'
which tends to zero as |/»l“ *0 as shown in § 5 13.
Hence f /'*> ...(2)

JC(Z-^
Since/' ($) has a unique differential coefficient, it is an analytic
function of E ; its derivative which is given by the expression (2)
is called the second derivative of/(?).
Similarly it can be shown that/" (£) is an analytic function
of 2 possessing a derivative equal to
L! f /<*> .
2*/ J c (7^)4
and so we have the general derivative,
/» (0= — f ■ ^^ */z
y 2ir/ J (z - ^)«+*
A function which possesses a first derivative which is analytic
with respect to the complex variable z at all points of a closed
domain in the z-plane, possesses derivatives of all orders and these
are themselves all analytical at all the points inside the region.
Suppose the above formula is true for w=m, i.e, we can
assume that
ml f _ f
fm it) dz;
2 -ni JC (z— 5r+l
then mll( f '<z) f<2) 1 1-
...(3)
Consider the identity
__L // (z-g-/;+/;)
z-^-h z-% (z—£)* (z-£-//)
h{z-Z-h)+H*
“(z-5)a (z-*-A)
/i /?*
...(4)
Differentiating (4) w. r. t. z, w times we get
1 1 (w+Orr^s+^mW,
(z-S-/*/"*1 (z-5r+1 (z-^/«T*
where | <£„, (z) I is bounded on C. Hence /(z) (z) dz will
be finite. Thus (3) becomes
h ~2nr W+I;#W • Jc (z-^r+a

W I ^ |c (*)/(*)
/fz)
Lim dz
/j->0 /; 2»ri c (z - £r+*
{ V Lim^ ~ m ! h J <f>n (z) / (z) </z->0 which proves the
relation for 1.
Clearly it is true for w*0, 1, 2 etc. Hence it is true for
every value of m, i.e.
='Ll f
/" (S) = ~
/(*)
2w/J C (z — 5)"+*
Aliter. Consider the expression

/” (S-M)-/«ft) m! r r_ 1 i ~i
h 2irih)cJ 2 L(z-5"(z-|V',+lJ </J
f(z) !{(:r-£ - + ,
~2nih C (- - (z-^-Ar+» *
[(*-5-A)1"*1+ (»*+!) // (r — - —//,’■*
(m-f-1) »/A* . „ . , .
+-2~\— hr
m! f ., v
~2irih \c^{Z) U-5)m+I (z-c-//)n,+1 12
= f \_(w (w+I)
+ 1) f m {m+ 1) /, i
2 ni J c\(z —Eim+l
$r+l ir-e _ /.\ +
(z-z-h) Vi-x ——?
hm \
+...
(z-5)m+1'r-5-//)"•+» fVfiz)<k
(//;+!) !
2 iri
f r_i_
JcL(z-£ri+1
_i_^ fm .
+(i - y«+* ,z _ 5r*js+. {r;(z ■;
+ ~3 i ~ *
(«+i)!f r i '”+1
}]
2tt/ /,{
lrl(2-0'^7T-l^)
+ U-5)*+l (2h — ^ — * (*>j/U)

1
where* W-2-,U-?-/;)».+i+!lfc±>
m-f- 1
which is a polynomial in z of degree (n, - 1).
.
^ u~! fr r~/(z)
(»,+ !) | r
2ni' I (Z~~7~, /.
r
a
/fz)
JcL(z-^r+st,"~^/
hj*(z)f (rj
+.~
(r -ii",+\2-/i;-/n'
_fw+ 1) | »>
2w7 /( ¥(-)
j CIZ-5,«+»+(7Z|-t)771.rJ */4 • J
A'A (*)/(*)
+ (2-yw+i
i] <h ...(5)
where we have used the result
1+
-4 ,+rî+u
l Z-C)“ /
174 Complex Variale
=1+
(*-5) 1
2-1
(*.* in the bracket the series is an infinite G. P.)
h
= H-
iz-Z-fi)'
Hence (5) is given by
fm tt + h)-/" (g)_(m+ 1
h 2 TTl -\cy^+i}^
(m fl)! f f f(z)
where
2ni n JcLu-e-ZOU- £)m+l
_+ (z)f(z) 1 *
-f
Now / (z) is regular and <f> (z) being a polynomial will also
be regular in C and hence both are bounded.
Let |/(z) | < M and | <t> (z) | < A/'.
Then
(m+1) ! | h | M*r i i_ i -M:_ I*
1 l< Jc Jz—5—/0w+l
2irdm+l
(m+1) ! I //1 71/ Z' 1 , A/'
V (2-^-/?) ^ \z-Z\-\h\> d-\h\> ldif\/t\ > \d,

l being the length of the contour, and | z—£ | = </. Hence
\I\<K\h\
where K is some finite constant.
| /1 -*0 as | /; | -* 0.
Hence
±±LIf -Iim ./w <*+*)-/"* (0-^+1
<JH
2tt/ C J f^dz
— £)”*+* k,m***° h J W
which proves the relation for «=m-f-l.
The result is then quite obvious from induction.
*5*15. Morera’s Theorem. If f (z) be a continuous single¬
valued function of z throughout a two-dimensional region and if
Jc/(z) *=o
for all closed contours lying inside the region, then f (z) is an
analytic function of z throughout the interior of the regign,
(Agra, 1940, 1949, 1953, 55, 63; Utkal 1960, Punjab, 1958)
In this thorem the precise sence of the word ‘Contour*
does not matter. The result holds if we

restrict ourselves say to convex polygon.
Let zQ be any fixed and z, any
variable point of D and let C\, C2 be
any two continuous rectifiible curves in
D joining z0 to z. Then the curve
consisting of the curves Cx and —C2 is a closed rectifiable
continuous curve C in D so that, as given
\cf(z)dz=\c/(z'>dz-\cJ(2)dz=0 i.e.^cf{z)<k = ^cf(z)dz.
Hence the integral along every curve in D joining zQ to r is the

same.
Thus we may write
FM-rrtM.
J zo
...d)
where \ is the variable of integration.
Then fl» + *>-*<*)_/w
1 f*+A
-r>«]
{/(*)-/(*)} <*5.
when the path of integration may be taken to be the st. line
between z and z+A. The integral on the R. H. S. of (2) tends
to zero with A, as/(£) is continuous, so that^
for an arbitrary real c > 0, a number t? can be found such
that 1 /(?) ~f (z) | < €f where | c, — z | < vj.
Hence
'| A j: !/«>-/(*) I I I
< . |A|- ...(2)
h I
Prodded that | A | < t?.
Thus Lim FJl + h)-F(z) ^
h-*-§ h
Thus F (z) being dififerentiable, is regular and has the deriva¬
tive/^). But the derivative of a regular function is regular.
Hence f(z) is regular.
Note. 1. The theorem serves as a converse of the Cauchy’s

theorem. In view of this theorem, just proved, we can enunciate
that a neeessay and sufficient condition for a continuous function
to be analytic in a region is that every integral of the function,
taken round the entire boundary of each part of the region should
vanish.
2. We have incidentally proved here that / (z) is regular

within a closed domain; the integral [ f (z) dz taken along any
J *0
contour within D is analytic function within C, and on C and
has a derivative f{z). This is the fundamental theorem of the
Integral Calculus of analytic functions and asserts that the
integration and differention are inverse operations.
5T6. Cauchy’s Inequality. Suppose f (z) is analytic within
a circle C, \ z~z0 | = /? < pt and if\f(z) | -■> M on C, then
I /" (*-o)l
Let i z-z0 | = i? be the circle C; we have
. f I f(z)\ds\
^ 2n J C \(z - zQ)u+l ’
Putting z—z0=Re'°, we have
n l M C2V Rdd
Rn+l
n \ M 2n M (n !)
. ti —
2tt * R»+' R
n ! M
Hence I fn (*o> f< Rn
5*17. Liovi lie’s Theorem.Jf f (z) is on integral function
w hich satisfies the inequality \f(z)\^M for all values ofzt M
being a constant, then, f (z) is a constant.
The function which is analytic in every finite region of the
r-plane is called an integral function.
The theorem is readily deducible from the Cauchy’s inequality,
concerning the derivative of an analytic function.
Let a be any point of the r-plane. If/; > 0, it follows
from § 5* 16 that
f" (a) M
an i =
n ! ^ Rn
whatever be the value of R. Suppose n= 1.
\f ■
Making /?->■o© we find that /' (a) tends lo zero. However,
a is any arbitrary point. Hence we deduce that /' (a) is zero,
/. e. the differential coefficient of f (z) vanishes everywhere.
Thus / (z) is constant.
C
Aliter. Suppose we are given any
two distinct but arbitrary points zlt z2
in the z-plane. Draw a circle C with
centre, the origin, so that z, and z2 both
are included in the circle C, such that
R > | Zj | and also > | z2 '. Then
fM-fw-£,lcgrlc£g*
= JL f . . .(l)
2m Jc (z Zj) (z-z2) •
If (*,)-/M l-U, f
|2tt/Jc
~f‘7 f (2),dz
(z — z2) (z-Zi)
I - ^ I/(g) I I dz
<£fc z -z2 I z—z
< 1_ 2rcRM 1 z,— z2 |
2w ,77--^) 11 z-77
[v |/(z) | < M and Jc | dz
the length of the contour]

RM I z. - z2 I
- zl{)(R-\ z21)
^(R~\ ...(2)
because j z-zr | > z — Zr| > *-|zr|, r-1,2.
Clearly when /?->oo the right hand side of (2) tends to zero.
Hence/(Z|)=/(z2), and zx% z2 being any two distinct arbit¬
rary points, / <?. /(z) is constant.
5*18. Taylor’s Theorem.
U f (z) is regular tn circular domain D having its centre at
z—at then for every point £ within the domain, we have
/(!■)= 27 le-u)*
n=U)
where an=fn (") | l-a |=r < R, | z-a | < i?.

* * •
(Raj. 1958; Agra 59. 62; Delhi 59; Utkal 61))

Let us take circle C of radius
R' with its centre at z=a such that
r < R’ < R. /^-\
From Cauchy’s integral formula, ( (/&\
we have /^\l^ ill
/ ^)=2^/- |c j^g/ (z)dz.
Now consider the identity

_i _
-z—a—(t -ay
r -<? J
~a-t/?~aY i
-a+\z-a) +-‘-\z~aJ
z-
Y_-—
I $-<* ($-*)" _1
“z - fl^Cz - O)3 ^ * * * (Z-tf)" (Z - •
Hence f dz
- 2tti JJc
2w/ z—
C 2 -£
^
=/(o)+<5-oi/'(a) + ^j-^V (5-0)"-*

<&-«)
-fR„say, ...(1)
r? — n)n f I 1
where /?„=—=—r- I :-r~.;-— / (z) </z,
;2tt/ J c (z-a)" (z-t,)J }
Now j z — £ '=| z — a -Z+a \ > | z-o | —| £ — a I > R’ -r.

and !/(z) | < A/ on C.
Hence we see that
ft)’
where k is a constant independent of n. Since r is less than R',
we find that I Rn |->0 as «->oo, so that we obtain
/«)=/»+«-*)/' (aH-«-a)/^>+.. « - a)" + ...

00
= ^ <*-«)■
fl=0 A I
=/(*)+? (5-0)»/ip
n=l
//" the centre is at the origin it reduces to Maclaurins series

00 fn
/(S)=/(0)+ z
n-1 n !
Corollary I. If/(z) and its first (n -1) derivatives vanish at

a, while / " (a) is not zero, a is a zero of order n.
Corollary II. If/(z) and <f> (z) and also their first (// — 1)
derivatives vanish at a, while </>" (a)^0, then
f(z)__ /" (a)
Lim z->a */» (r) </>" (a)
[This can be easily proved by the theory of limits J

Corollary III. If /• l«)=0 (n=0, 1, 2...),/(«) vanishes
identically at points m the domain of a. [The circle of conver¬
gence of the Taylor’s series is called the domain of a.]
\--'■'■'Example. The function f (z) is regular when | z\<R and
has the Taylor's expansion £ anzn. Show that if r < R

o , J '
J
2,„' J o i* <t0=X
e I an I8 r*». ^
Hence prove that tf\f(z)\^M when | z | /?,
OO
f I a" ^ r'n ^ M2- (Agra, 1945; Punjab 1960)
1 f2ir
We have, ^ I '/ (rei0) j2 d0
27T . oo 2 CO
=1 f 27 anrnenie dd v f (z)=f(rei0)=Z an (reiB)n
o ' o
J! f27T i oo CO i2
= «-' l S anrn cos «0 + i 27 sin //0 <70. ^
Xt7- J o o • - o
1 f2TT
— I (.ao~^~air cos $4- • .onrn cos nB + ...)2 | dB

27r J *
+J [ I (Oo+^r sin 6+ ... + anrn sin n0-f . .)z | dO

*^7r J o
OO CO
= i S \ an I2 (rn)2+ 27 27 | avaQ | {r*+« (cos/>0 cos qd)
j-77 0 Ln=0 . * J>= 0 3=0
-f-sin pB sin qd)}J dd, p^tq
1 » f2-»r oo 1 OO OO
|*(/-"J2<0 + 2^ 27 27 I aMn
PUQ I rvJ,Q
-i I
^ Jo fj=o
s I 0 r= 0 r/=3
X cos (p - q) 0 dB, pz+q
1 OO f27r I 00 OO
= 2’ | a„ I2 (/•")* <0 + ~ £ 2? flA r»+«
2tT w==0 J 0 ^ p=z0 47=20
f21r •
XJ cos (p—q) dB,pz£q,
since integration and summation can be interchanged because

of uniform convergence.
= ±.2* £\a„\'r'"+\-£ £
2tt „ô j-7T a=o r=o
bk
= 27 ( | *„ |2 r2") 4-0.
71=0 ^
\/
oo
1 f2ir
Hence, |27;*H" = ^- l/(re«®)|2 </*.
0 ^ J0
4 fci/w i**
2TT
(i/2 r
-t7r Jo
= A/1, since | /(r) | < M when | z | = r < /?,
CO
/.e. r | *n |2 r2,‘ ^ .V/2.
0
Aliter.
Since /(z) is regular in the region | z !=r < J<, a
can be expanded in a Taylor's series,
°C 00
i.e. J {z)= E anzT1= E Qurntlng% where z=re,g.

n=0 n=z 0
If an be the conjugate of an, we have
/(z)=S
T-0
Hence |/(*)|2=/(z ./(r)= £ anein9 rn £ aDrê~»'•.

n=l 1=1
The two series for/(z) and/ul are absolutely convergent

and hence their product is uniformly convergent for the range
0 < 6 < 2*. Thus the term by term integration is justified.
On integration, all the terms for which n^p vanish, for
rzn
e'"i9 dtf=0, mÔ.
J0
Hence we have
±r'/«rj>
2n
fi-o
°r „£ |ff. |/(.)| 'jo
1 f2?r
d6=M\
IT t 2 Jo
Hence the result.
This is another method of deriving Cauchy's inequality.
S'19. Laurent’s Theorem.

Let f (z) be regular in the annuals
(ring-shaped region) bounded by two concentric circles C, C, of
centre z0 and radii ,, and TAen, ij z is any point of the annulus,
we have
Cc
f <z> =JL °»<- - +aE
n—1
a - r.>-
where
2nijc. a-zjn+i
«=/, 2, 2, • •
each integral being taken anticlockwise.

(Agra 1940, 44, 47, 50, 51, 54, 60, 65 ; Luck. 49, 58 ; Raj. 59;
Utkal 1960)
According to Cauchy’s integral
formula,
-jc. m «] • •<■>
Using the same method for the
identities as in Taylor’s theorem,
we g6t
•••+£$ *=?]• •«>
Similarly 1
l-z Z £0 — ( £ — Zq)
1 . jr.-o ■ , a-z0)»-' (5 - z0)n

=-z^(z—zQy ; •..(3)
(“ “~0)n ’r(r-r0)'*(z-5>
fn~x (za)
Suppose an^=J--^1 = _L f _/(§.L_ ,._i 2
(n-j)! 2ni Jc* (£-z0)n 2* ‘ *
Now
J_.f f r/ci T J , z—zo . .

27njc, 5-r 2^71 Jc, h~^+(T^)’ +’ ‘ ■ + WôPi
+aô)-"jz-z0ri dl
■“/(*•) + (*—z,)/' (r0)+t^-^r <Z°’+ • • •

f',_l
• . . + — -——°4 (■? _ T \n- 1 I /
('»-!)! 1 “o) ^7|
where
...(4)
'-a Jc
Now
Since l lies on C2 and z is within the annulus, we have
r——°|=£> where A: is a constant < 1

^ ~zo 1
and
I?-Z| > !S-70|-!z-z„I
^ r* ” I Z ~r0 (
Also |/(£) | < ,l/2 for every \ on C.,, where A/2 is any finite
number.
Hence 1f hhMl , Arr

2" J Ct r2 - | z - z0 I 1 ;
I 2irrt.knM2
^ 24- rT-| z- z0 | =/CA;r,> ^ being a finite ccnstant.
Thus /„ tends to zero as n-*oo, since k < 1.
Then, we get Jdz
= a»+ai (z-*0)+a2 (z-z0)?+... (z-z0)" +.(5)
Similarly wriiing dt we obtain from (3),
Now f fjv
2nl Jc, E-z
ô+(z-^+(7^+ • • • + Si L rHfe:J ^
n-l Ar
...(6)
whcre '-±iL8=?Jf!S*
For > on Cu we have Z~zn —/ < 1 where / is a constant.

z-z0
Again z~Z H z Zq — (' Zq) I
>\z-z0\-\Z-z0\
>\z-z0\-ri
and | f (z) i < A/j for every £ in the circle C1#
Hence | /. | -1 L f (|^)“ ~ AO A
—z, n
l
< dl I
sfc,''®'!*- I z-%.
MJn
. 2777-j
2tt! z-z0 | -z,
= S.ln where S is some finite constant.
Now as n-+r., j Jn |->0 as / < 1, so that
2rri J Cl 5-Z a*>-rix (Z-Z0y ...(7)

Hence from fl), (5) and (7), we have
OO DO
f (z)= 2S on (r -zt)n + - bn (z-z0)-M, ..(B)

n—0 i<—l
where „ -1. f - 4i b __L f /(O rfg

47,1 ~2W Jc. (5--0)n+1 * ” 2*/JCl (4-^0rn+1
Note 1. Since / (z) is regular in rx < | z — z0 | < r2, the
integrals for and bn can be replaced by integrals round any
concentric circle | z - z0 j ^ R such that rx ^ R < r2.
Further since we have
dl. ...(9)
2s(c
Hence we may write
/ (z) = 27 (2 - z0)M, where an is given by (9)

— CO
Note 2. f (z) =</» (zj + ^r (z), where
•/* (2)=S (z-z0)n

0
and 4* (-’) = ? (z - z0)-n.

O
Then •/> (z) is regular in | z—z0 j r2 < and ^ (z) is regular
outside | z-z0 1 > rx.
Remark on Laurent’s Series.
Suppose that we have obtained in any manner or as the defini-
tion of f (z) the formula
/(z)= £ An(z-z0)n. f R'

n ■= — » ---o < R]
Is the series necessarily identical with the Laurent's series for

J ^ (Agra, 1950)
Answer. Yes, for if C is the circle | r—r0 | = />, R‘ < p < Rt
the coefficient of an in the Laurent’s expansion is
‘■■S/fc (5-t!r+‘ =b,'i A'■■■ u-

_ r ^ f — 20)m
w=-oo 2ir/Jc (5-r0)n+1 4
°° f2- p»»e”»«•
= £ J0 ^Tf^rprn# putting ;-r0=P<?'«
-=c 2»r/Jo P'**1*?
oc J /*2n
= 2’ prn-»e<m-«) .• ^
— CO *’7T J 0
which is zero for every value of m^/i. If/;/=//, it becomes
tl (i6 = An>'.e. an=An.

Hence the result.
5*20. Zeros and Singularities.
Definitions Zeros. A function/(z) which is regular in a
domain D can be expanded in a Taylor series about any point
- = ro of D in the form /(z) = f an (z-z0r.

0
If further a0=al =0,-0^ . =0, a,,,^0, the first term in
this expansion is an (z—z0)"\ In this case z = r0 is called a zero
of order m.
Singularity. Ti e point, where the function ceases to be regu¬

lar, is the singularity of the function.
Insulated Singularity. z=-0 is said to be an isolated singularity
if there exists a neighbourhood of the point z=z0 which does not
contain any other singularity except zQ
Consider now a function which is regular in a domain D except
at the isolated singularity say z=z0 then it can be enclosed in a
ring-shaped region such that P2 < | 2 - z0 | < Pp where the radius
of the smaller circle may be as small as we please; and radius of
the laiger circle may be of any dimensions with the condition
that it should lie in a given domain D.
In this annulus or ring-shaped region the function can be
expanded by Laurent’s Theorem in the form

f(z)=£ an (z-z0)"+r bn (z-z0)~n.
o 1
The second term on the right is called the principal pari of

the function at z—z0.
Now there are the following three possibilities :
1. The principal part may contain only a finite number of
terms, i.e. all the terms after bm are zero, i.e. ^m+1=^m+2=... =0
and bm^0. Then the point z=z0 is called a pole of order m of
/(z) and the coefficient bx which in certain cases may be zero is
called the residue off (z) at the pole z=Zq,
If w = 1, then the pole is called a simple pole or pole of order
°ne and bx = Umz^ (z-z0)/(z).
2. Isolated Essential Singularity. If the principal part in the
Laurent’s expansion contains an infinite number of terms, then the
singularity is called an isolated essential singularity.
3. In case the principal part does not contain any term, /. e.
bx = b2=b3=... —0 for 0 z —z. r, then
/(z)=a04-a1 (z-z0)-f... +an (z-z0)"+...
In this case the singularity is called a removable singularity
and the function can be made regular there.
5*21. Solved Examples.
l + 2z 1
1. Prove that -/ + Z-Z*+23-...
z2 + z3 =7*+
z‘ z
when o z \ < 1.
Let f (z)= =p (2- j—which can be expanded in a
series of positive and negative powers of z; we cannot apply

Maclaurin's series to/(z) itself; since this function is not analytic
at z=0, but we can apply it to the function ,-L. Thus when

1
0 < 1 z ] < 1 it is true that
y+2?3==L. (2- i-f2-z2+23-...)

z*+z
~!*+4-~l+r.-.2r*+r*-
z. •••
2. Prove that when 0 I* 4,
1 00
4 Z-Z2
t 1
Let /(*) =
4z—z* 4z (1 —z/4)*
Since the function is not analytic at z=0, Maclaurin’s

theorem cannot be applied; but it can be applied to the function
(1 -z/4)-*. Thus when 0 < | z| < 4, it is true that
1 1 1 ,
4^=47(l-z/4)=4l [l+(*/4) +(*/4)H ... j
= 2 ?Ll.
n=0 4n+1
3. Represent a,function,by a series of
positive and negative powers of (z-1) „ inch converges toffs)

when 0 < \ z - 1 \ < 2. J y’
In the given region the function is not regular at s=l
is not valid. However, it can be
expanded in the form of Laurent’s series.
Here / (Z)= z
2 (z-1)
__I_
4 \ 2 y
= ~2T7r7)-|{I+(L^-1)+(-f-,),+ ... J
_ __[ 3 «> /z-iy
2(z-J) 4nf0 2~ J •
This expanion is valid, when 0 <

z < 2, since
^—(z—1 j~Cz~-3) is analytic in this region.
4. M /•» «*, expansions in powers of z of ,he

function f (z) given by f(z) =_/
Th, f ,• 2 (l+*>'
(0
(° WeTke
We take S"* '° bV'8Ular
the reg.on 0 <|z| a!
< 2=±'
j, then•"<« *-0.
^ 2 (I -f-Z*)“z ^ *f Z2)_1 = - j
=%-2+z3-z«+...
=i-+2 (-1 )-*»-* (0 < \z\ < 1)

z n—1
which is the Laurent’s expasion of the function/(z) in powers of

z in the region indicated here, as this series is in positive and
negative powers of z; which converges to ~rj~i in the region

~ * ~i~z )
0 < | z\ < 1.
(ii) When | z | > 1, we can write
1
/w-i
Z8 -•]+z-2
=L(i-*-*+* "f" . . . )
i
= £ ( -1 )- 7-n+l . I*I > 1
n=l
which is a Laurent’s expansion in the region defined.

2z* + l
5, For the function f iz) = , find a Taylor series valid in
the neighbourhood of the point z—1 and a Laurent's series valid

within the annudus of which the centre is the origin. (Agra, 1949)
(a) (i) Taylor’s expanion of the function/(r).
_ \
/(*) = ^£#=2 (z-I)-f 1
z‘+z z+ 1 + z
=0 (z) + 0 (r)+Z (')>
where 0 (z)=2 (z- 1), 0 (z)-=—j,* y. (r)=y.
Taylor’s expansion for 0 (z).

OC
2 (z —1)= Zan(z-i)\
nz=0
0<«) (/)
where «n=-j— =0 except at // = 0 and 1,
i.e. ^0=2 (i-1). ax = 2.

Because all the derivatives after n=\ vanish identically, the
Taylor’s series for 0 (z) is
2 (/— 1)4-2 (z—/) . .(1)
(ii) For the function =•/» (r).
n !
■&<»> (z)=(- l)w -—-
y) K ’ (z +!)"+»’
1
so that ^)=(-ir
/I! ' (l + l)n+1*
Hence the Taylor’s series is
oc 1
...(2)
+ (z’=£o(“1 r (T+Trf*~ '>"■
(i) Again for -=y (z), we have
n l
X(") (-)"( — l)n zn+l *
so that •/<n) (0
= t - 1 )n — •
n\ y ’ i"+1 ’
CO
•/(*>- ^(-|)n1^c-on. ...(3)

n=l
Hence the Taylor's expansion for f (r) is the sum of (1), (2)
and (3), i.e.
/(,)=2 (.-i)+2(2-o+jo{(-,»" +(-r+V+.}(-"')'•
(b) /,«)-2 (z- l)+|^z)=2 (z-D+’ + jI, .

The Laurent’s expansion valid in the region 0 < | z | < | is
2(z-l)-f-*+(l+z)-’
= 2 {z — l)4--f(i — z -f z2 — ...)
= 2 (2-1) + -^ + f (- 1)»Z\ 0 < I Z I < 1.

n=0
6. Expand f (z) = powers of z
(i) within the unit circle about the origin,

(ii) within the annular region between the concentric circles
about the origin having radii 1 and 2 respectively,
(Hi) I z | > 2, i,e. exterior to the circle of radius 2.
(Agra, 1937)
Here -+3
/(-’) =
z(z-2) (2f 1)
3 5 2
2z+6 (z-2;+3 (z+1)'
(i) When 0 < | z | < 1, we have
—s-n I. (O’+tI
which is a Laurent’s series representation of/(z) in powers of z
in the region indicated.
(ii) 1 < | z | < 2.
1
5+
-“(■-D’-O+D
1 S 00 /z\n 2 I
-S-kG) +55^
__ A A 2 00 1
2z—12 iuy+
which is clearly Laurent’s expansion valid in the region noted,

(lii) When J z I > 2, we get
2 1
'<*>—5+
*0-1)*0+-J)
=_2+a jf n‘(-i
2z ^6Z n=o\ 2 J ^3z n=0V z )
—E+5i.(T)+35£.(-,>-^
which is also the Laurent’s expansion valid within the region
given.
7. Find the Taylor and Taurent's series which represent
the function (z* — l)/(z + 2) (z+3) in the regions (i) j z | < 2,
(ii) 2 <\z\< 3, (Hi) | z | > 3.
(i) When | z | < 2, we get
J (z+2)(z+3) (z j-2) (z-f-3)
= i + _i_L
^z+2 z+3’
Complex Integration 19 j
Clearly the function is regular at every point of the region

| 2 I < 2, so that it can be expanded by the method of Taylor’s
series or binomial series.
/(2)=l + rL-8 (2 + 3)-‘.

z+2 ...(1)
Hence /(0) = l+f * _1
Li 6
and
n • ( z + 2)n+' 8(“I) -(2+3)m+i. ...(2)
Evidently /(0) cannot be obtained from (2) by putting
/>=0.
f(z)= -J + S (-\)»(* IV,

n=l• \2n+1 j"+V“ •
(ii) Region 2 3
/(*) = 1 + 8
2 I > 3
(,+i) *(‘+i)
Aliter We can also find the coefficients by evaluating

the integral k
o. f(2)
dz.
-WC (z-z0)n+l
By ab initio method.
1 8
L( 1+
"•"■SFrJ C zn+l \ z+2 z+3 ) dz
Z2-f . . .
2w/*Jc zn+1 ' 2w/Jc zn‘l’1\2 4 t8
, (-on (-1 >n+1 l
••• ' 2n+1 * "r 2n+1 *z+2j
1
• • •
-Js-Jc <fe{3F+i-3i-iii+ 3V-

(-l )w+1 J l
•••■' 3»+iz‘*’ 3n+l 'z+3J
where we have used the property

1 ir Z Z\ (- 1)" 2n(, z 2s, Y|
+a=«L1-«+s+---^' U-*+««+”vJ
.in-*+*?+ (-1)* — —provided I r| a.
at a + a!+“'^ ' fl"l+*/aJ
We know that every integral of the type
*-0t Ml-
f C zK
Also and f
JL are each 0 as the integrands
t
j C Z+2 Jc Z + 3
are regular on the circle C defined in the region 0 <\z\< 2.
For wÔ,
„ . r /3(-1)” s* l2«
2^=jc^-^Ti-3"+i Jz'1 ' L2n+1 3"+^
«r J_ J_1
/. <?. —( *) L2n+1 3n+1
For //—0,
8_8
f r dz
1
°°“2m C [z + 2z
1 3
2(z+2)'"32 + 3(zF3). j
_J_f
27t/ JC 6z 6
since other integrals are zero as before.
Note that a0 can be most easily evaluated in this case when
QO
/(z)= E a„z».
0
z2- 1
=/ (0)=(
V(z+2)(z+3) L-l
Hence/(*)= - J + 5 (- 1)"-j*,} z».
(ii) Region 2 < Iz

oo
Here /(z) = 27 *nz"+ % bnz~
*1=0 n=l
.(i)
where <?B =s JL f fSllAl b ^n~l fiz)d-

2tt/ J c z"+» • ®"“2ir/Jc " yU)^
where C is any circle, | z j=r, where 2 < r < 3. • (-)
When nÔ.
All the integrals of the type J *=0. when kzpt 1.
Also
/Cz-f-3 t/z=0, as the integral is regular within the
region defined in (2).
So there remain only two integrals
dz
= 2tt/
Ic T-^JczTjMc *+2
where C' is the circle | z + 2 | =/>, p beiDg any small radius.
[ *•’ Jc z+2^\c' z^2 by the PrinciPIe °r deformation of

contour. J
Hence for n-^0, we have

3 _J Af * . 3 (-!)»+* _dz_
r\2TI+1" 3"+ 0/c7 2»+i L z+2
( 1L2',+1 “ 3"+i ~ 2^+i] '27r/ ‘
°«=(-Dn+l-3^*
*“sJ(t+S -2Tz3+2)-|)*
=__l r—- f — - r
2w/'L 6 J Z “2 Jc z + 2J
- I f 2_:1 ^
2rr/Jc Z + 2*
3 f2w
= -2^J0 (-2 + pC")«-W^
putting 2+2«f»ew, p being small

2*=3.(-2 )-*,
since (* eni9 dQ=0, if nj±0.

Jo
-3V*-1 J_
Hence /=(z)=-f + 2 (-()”+* JL2-+ 2 3.(-2)
w=i 3n+I n=i 7n
(iii) Region | z \ > 3, put z — ~T'

z2—l
The function /(z)= - . . is transformed into
(2 + Z) (Z-f-
'2
1 -2 1 8
* (z')« (1+2*0 (1 + 3*0“ 6
2<l+2*0+3(l + 3*0
in the region | 2' | < J.
The function being regular in this region can be developed
as a Taylor’s series. Hence
* (z') = - f (l-l-2zfr1+f (1 + -V)-1
OO oo
= -£-f 27 (-l)n2".z'"+t 27 (- l)n.3n.z'n
n=o - n=0
oo
= 14- 27 ( - 1 )n (8.3"-1 — 3.2n~1) z n
r.=l
= 1+ 27 ( -l)n (8 3n-1-3.2n~1)4
n=l -
or an etc, can be evaluated as is done in part (i-.
The alternative method in this section has been simply given
as a model answer, as it is sometimes useful though here it is
much more complicated than the direct method adopted before
hand.
8. Find the Taylor's or Laurent's series which represent the
function (j + z*J(~l^2) when W I 2 I <■ A ('*)

(iii) | 2 | > 2. . ^
In the region i z\ < 1, /(r) is regular.
Hence /<z) = 27 anzn, a Taylor’s expansion,

ffc=0
where 0 =/ (0) = 2.
tln = 2dz> where C= I “/ = r < l> n^°

1 i r 1/1 2—2\ ,
2-rri 5jc zn+1\:t2 r2+l/
By the method of partial fractions,
1 D" n
Now
1
zn+1’z-f 2 -2n+i|_2 22 + 23 '
(“
2»+i
. (-1)"+1.^+1
+ 2"+».(r+2> ]
1 2-2
and zn+l*z2+ !
-2m+2
it'*-f 1
„J^.(r-2) [l —2*+24— ...(— 1)” z"‘+(- I) .!+ ,

i
z-2 -5m+2-|
|"i-z*+z4...(-ir z8"*+(-ir+1.^ TiJ if n----2m
1 z—2
and
~zWz*+ 1
_2fn+2
1
=pKF.[«-**+2*-f-ir z!”+(-ir+1 p+1J
o
#
if bs(2w+ 1).
Now fc *-0. k > 1.
Also : > 2.
iC z-f 2 °' lc4-i=0and
Again J d~ = 2iri, J —-? clz=2ni, j 2-2

clz 4rri
2n+i dz=2nian
(-1)" 1 (-1V*1 J 1 Jz
-;ic[ 2"+l z ' 2n+l ’z+2J
-1)
-H C (Z-2J [f j)m
z + —^ if" = 2"'
n
(-D 1
5.2^*27r/“5(’I)m(-47r/)
and if w=2w+l,
(—ir _ir
2-rria
5.2"+ '2W - 5 fc (^2>[(‘ 22 + Z2+l J
,2ni.
.2»+»
00
/(z)=o0+r <vn
1
co CO
~+C. z"+ 2 £ ( - lrz‘-*"-2’ (- 1 )mz2"‘!1
]
3S6 Complex Variable
=3+5 [f (-pTr z"+2 | (-1)“ z’"‘-z f (- i)m z2“J
= 1 -1_- 2 C_~ ? zn_i_2_~2 J ( —1)» Z2n

2 + 5 7 2n+1 + 5 f 1 i; 2 *
which can be easily verified by the binomial expansion as given
below.
(i) When 0 < j z
f(z)=
-ifeh-r+7.)
=H( +l 1 )"+¥(1+22)-1
| oo 2n 9-7 CO
-IS
(ii) When 1 0~ 5
-iio+r+w+?r
(iii) /(z)-± (J_ _ when j 2 j > 2.
Put z=
z'3 1 1 z'2 + 2z'+l

/(-) =
(1+*'2) (1+2*')“2 5(1+*'*) (1 + 22')
1 1 ='+2 1 1 ■ ..
2 — 5 1 +z'* J C (1 +2z') ,D the reg,0n i 2 \ »i -' n
!
2 5
+ £ (- I)"
o 10 o
Z (->
1\ *(-l)?‘ 1 v5 (— 1 )'*. 2"*

^ 2/i io 0
- ri
ir, the region | z \ <2. Here the constant term is zero.

For the part (iij, we can also use the alternative method.
In the region 1 < | z j < 2, f (r) can be expanded as a
Laurent's series
zr.
±..L=T^ = Z c
;c + 2 rM-iy 0
i
where an f , ■' *.
11
c
2w/ C' z’1'1

C being any circle z 1 = t> such that 1 _ I 1 P £ *1 .1
2n+i-52,Mi[_2-2i+23 2n+l 2',+1 iz-f 2) J
■»_? r i _ lr -n z2'"+1
- L 1 - 2?+2‘ 1 r *l",+——{— ]■
if n = 2m
f±z) _ I J I 1 z_ zj_ (_i)"£’•( -_1V+1 r"+n
1 zn+l 5 i[ 2-4 23 • ' z"+l 4 (z+2) 2M>1 J
-2-~r
2~
i 2"
4-1 -T Vm-4 2
( - 1 >
. ( - 1)’" z2 M l
5.z*m+M 1 z-+ 1
if // = 2/»-f 1.
( - 1)» </r 1 t - ir tz -2)
Hence 2w/an=^ ,
2n+1 z -:w (z - 2) (-li »r. - 1 _
- “J ‘Iz, it = 2m
zM-1
. . if (-i)Vz if r (- 1)'“ (z— 2)
and
27r,‘7" = 5jc'>^z -5} A
(Z~ 2)
Jz, n=2m-f 1
r
i
( - 11” ( - 1 VL ff 4
i. e. 2= . 2tt/-—- ( - 2) 2rri
" 5.2”+*
f (r“_2) z
—i (-ir+* </z, if n = 2m
• Jc' z‘M-l
(- l)n „ . (-1 m
and (-2). 2 W
5.2n+1
“3 jc- zmH *•il "=2"'+1

Now putting z=*-„ <fe=—rfz', we get
f (z —2) r , f 1 ((1/z*) — 2} (1//) rf„.

Jc-rf---Jc’_z'2 {(i/z'2)+n
where C" is i (?’2” 1+-'2)
= _2| since the other integrals vanish.
Similarly dz, transfarms by the same change to
-J ^(1-2^(T-.4n)rfz'=!c-f-=2"'
Hence 2«fl„-^+(-^.4-»+i (- ir<-4«),if *-2»
(— 1)”})Z.2*ri + i ( - 1 )m.2iri, if n—2m +1

" 5.2n+l 5
= ( - 1 )nl5 2"+1 in each case.
Similarly,
2rr,7,„=ifc.--(rf2-^)*.C' = |z| = P. > < !<■! < 2
Now rJC' r+2 r/r = 0, for /i = l, 2, 3... as the integrand is
re gular and Jc# dz' b> putting r=7' becomes

1
-Jc''1 •7-^(-.-)*'=/lSay
Complex Integration 1V9
where C" being the circle, 
= |c,(1“2zy-jL- dz' if, — 2m,
other integrals vani.'hing
and
I=\c - (1“2z,)F((1 if" = 2m+1
2nibn=— (- l/"-1 v — 2). 2ni, if n—2m
=-j- (— 1 )”* 2wif if « = 2/?/ + 1.
Hcnc „ 1 f — !)*»»+* z 2 « ( - 1)
wz"-5„.o 2*«** ~5„^~
1 f „ % (~-Qw+1 0 £ (-i)«*+n
5 L “« 2 „to J
Hence in the region 1 < | z | < 2,
(-!)'• — 1 >*« + J “I
M:-t) F2"^ J
which is the same as before.
9. /Vm-e that near z= I, the function

tog (l + z:)
may he expanded in a series of the form
t<>g 2 + £ a„ (z-J)"
"=1
and find the value of an, where the value of logarithm taken is not
general.
Find also the radius of convergence.
, , e i i *2 arfX t—4-1 being the two branch

Obviously the zeros of 1+z are z 5 ^
points of the function /(z)=log (l+z*)*
Hence in the circular region whose
centre is z=l, the nearest singularities
in the circle of convergence are z= ±i,
and radius of convergence=V2. The
function can be considered as a single
valued function, particular branch be¬
ing defined by giving a value to the
function for any z in the region.
Particular branch which has been referred is given as the
principal value i.e. for z= 1 the function is log 2.
Hence for | z—1 |< 72, we have
l0g(H-Z*W (1)+^(1)(j-1)+...-^t (“ )+‘**
where </> (l)®[log (l +r2)]r=i=log 2»
# (-1 = (j+z*) = Tî+ z+f
from which we get ^ -
*<*> tz)-(-i)"-1 (»-')! [(7^rr+(^+o"J
Hence •/.<”' (D-t-l)-1 («-» ! [(T^”+(TT7)\T
By ,he well known property of Trigonometry (De Moivre's
Theorem). r| i T“
(1 _ir»-Hl+0-"=(V2)-" [72-72] +(v/2)"" LV2+V2J
1 72)-" (cos J -I sin j)

T
v , . . rr V"
+ (V'lr"[cos — +1 sin 4 J
r, ww
W2)'n I 2 cos — 1
4j
(r,-2) ;,7T
7(S=*»r*,C0S 4 *
(// — 1) ’ ///n-N
Hence *.•(", (l)-(-l)-1 2n?3nrcosV4/‘
Thus we get </» (z)=log (l-fz2) = log 2-f E an(z-\)n,

n= l
(- 11"-1 1 njr
v\ here • 2<» cos 4 •
°"= ”7,
Clearly the radius of convergence is \/2.
10. Show that ecli <*—>»=> = 2 anZn>

— 30
27T
where cos (ntf - c sin 0) d&.
(Punjab 1958 ; Agra 1944, 52, 60)
The function is regular at every point except atz = 0, hence

it can be expanded by Laurent’s series ; in the region r C r.
(where rt may be taken as small as we please] in the form
f(Z)=e'lz <'-»'*>= £ anz",
where _ 1 f /<*> t
°n Jc z"*1 * ~
I f e'n u i
J
2*ri C dZf
where C is the circle I z 1=1.
Putting z=e,e and z— l/z=2f sin 0, we get

I 2 7r eci sin 0
°n = 2 ni 0 e(»+l)/0 le’° (iB
1 \t2n cHc sin 0—liOfjfj
= \
2n ' 0
= 2ir cos * -,,0 + c sill 0)-f / sin (e sin 0 — nd) jn
1 f2w
= 2^ J cos(-csin0-f/i0) </0,
V if <f> (0) = sin («6»—c sin 0).

'/• (2rr —={sin n (2tt_0) sin (2ir-(y)}
= —{sin (2/fir-nt/j- c sin 0}

/.e. .,
1 (0) = _ ^ (2tt .6-).
Hence the second integral is zero.
10. Prove that cosh =tfo+ f an(^nJrz^) wêre
a = f cos nd cosh (2 cos 0) dd.

' 277 Jo
(Alld. 1942; Agra 51, 57, 62; Delhi 59; Luck. 58)
Here /(z)=cosh fz+ is a regular function throughout
the finite part of the plane except at z=0. Therefore it can be

expanded in a Laurent’s series at any point z save the origin.
.*. /(-)=*cosh (z + anzn + Z^ bn rn
=ff0+2 anzn+2 bnz- n
where °n==^j }c z^ <iz' 31ld ^"=2!»/ jc ^ ^

C being j z | = r. Wetaker=l.
Since the function remains unaltered if we substitute - for z
hence an=bn.
Hence we may write
/r-)=fl„ + f a„ •
Then putting z=e<0,

1 f2ir cosh (2 cos 6) eie i dO
g(n+l) /•
°nZ=2"i J,0
__-L [** cosh (2 cos Q) e~ni0 dd
~2tt J0
_ -1 P* cosh (2 cos 0) (cos nd - i sin //0) d6

2* Jo
—— f cosh (2 cos 6) cos //0 </0,

2* Jo
the second integral vanishes if
if /(0) = cosh (2 cos 0) sin nO, then
/(2tt -0) = cosh [2 cos (2?r-0)] sin /; (2n-0)
s - cosh 2 cos 0 sin «0= ~/(0).
Complex Integral i on 203
11. Show that if c > 0, ez+u3i-z'> — r nn:n, where

n=- x
p-ciz r2ir
on=-—- ec cos 0+cos2 0) cos (c sin B i / — cos 0)—n9\ (id.
2tTCn } 0
(AIM. 1947)
/ (c)=ez+c3/2-'' is a regular function of r throughout the finite

part of the plane except at z = 0.
Hence it can be expanded in a Laurent’s series at any point

r save the oriein.
•*. /(->«£ nnzn,

—X
where
*•- a fc 'M *■
c being any circular path of finite radius with origin as centre.

Let C be taken as | j | = c, c >0 (given).
Putting z = ce‘•,
exp ( ce'9-\-cê ue
' \ __ 'ft* _ ,,
yr»*Hetn+l)ifl uv
cxpf cei9+2e~2
")
= 2^T« f„ cxp ( ~ «W+2«"f" ) <h
~2vc^ }„ exp ( c cos # + cos lO^.e-’M+d sin <)-/ (c/2; sin It*
>-cl* rnr .
2»tc" l„ exp- <c COs 6 + r cosI 0)|cos sin 0-f ^ sin 20^
— / sin ^ nB - r sin 0 + ^ sin 20^1 (/a.

(> r/l tlTT ~ ^ i
^ l^cu J„ eXp fc cos 0+c cos2 cos [nB-c sin 0 (1 - cos 0)}
sin 20 = 2 sin 0 cos 0.
Since/ (2*-0)= -/(#,, ,hc jnte„ra|

27T
er cos 0 + c cos2 U
1 0 sin [nB~.c sin 0 - (I —cos 0)J </0 = o.
Complex Variable
204
bo . h2
12. Show that e'#/-+v“ • • • y +fT + • • •
f f27r „(m+v) cof 0 {(v — u) sin d-nd} dd.

w/iere *
h -J- [** e(£/+v) ca5° ^ {(“—v) sin e~ne) dd-

n~~2*)o
f ,^-jilz+vz u resular throughout the region in the r-plane
except at"r==0, hence 4 can expand it by the Laurent's theorem
in the form
/(r)==e(vr+lf/z,2 anzn+Z bnz~n
fjz)
where
°n ~~ 2tT! j C z"+l
«r a _ 1 f
Hence fln-^ Jc*
Jys+w) -L
-"-l
dz.
Taking C as the circle \z\^\ and putting r=c‘9, ue get

•27T (rert + ue~',fl) je-n0i JQ
ml
2wi J 0
1 e
‘27T (veie -+ ue“

1 1 /y i0 - n0i) dd
"27T j 0
•snr
(,•+//) cos o e(i\- sin 0-iu sin 0 —"oO dd
'2ir J •27T
0
etv+«) COS o [cos (O’ — w) sin d-nd}]
*
2rr
1
J0
i sin {(v-m) sin 0—«0}]
before. Hence
1 integral vanishes as
e'//+v) cos e‘.cos {(v—m) sin 0-;i0} dd,
a
1
2tt , 0
r
)
1 ulz -n—1 dz
"2m 0 J
rzit (!•*■* +1/*- putting z=e'°
_1_ "0) /e"«' dO,
c
0 J
___ 1 g(w+v) cos 0 expH {( -v sin 0-f w sin d)-nd)]dd
r
~~ 2-n J 0
[cos {(m-v) sin 0 - «0}
1
r
2tt ) o
g(M+v) cosO
— i sin {(« - v) sin 0 - dd
1 C(«+V) COS 0 [iu — v) sin d 10} dd.

"2"- Jr
0
The other integral vanishes because of the property
Hence the result.
13. Find the function f(z) which is analytic throughout the

circle C and its interior, whose centre is at the origin and whose
ladius is unity and has the value
a—cos 6 sin d
+i
<r ~ cos 0+1 ~r ‘ a- - 2a cos 0-\-1
n here a > 1 and d is the vectorial angle at points on the circum¬

ference on C. .. „ ,
(Alld. 1947)
If/(2) is anaIy,ic inside circle | z |-1, it can be expanded
at any point inside it by Maclaurin’s theorem as
/(;>=.? £L<0)
n«-o n \ *
where y>* (0) f /(-) <i=

2iri Jc zn + ‘
Now on the circle so that dz=if* d0
and ft7)^?~C0s fl + / sin U

a2 - 2a cos 0+1
a -e~10
a--a + +i *
_a - e~t0
~ (a-e~'9)
=_1_
ta~t,9y
(0) 1 f2* l
n ! ~2ni j 0
| rZT,
—/iC #
Jo
= -— f »—n9( f I ,e'* t-‘® X

~7Ta Jo Vk, + F+?'+--î + ...) dd
an+l v j e“>d0~Oirk?±O.
oo i 1 z2 z
Hence f (z)= 2 zn==a+at+a*+ "'
T2=*U M
1
"a(l-zla) (a—z)
which is the required analytic function.
14. Determine a function which shall be regular within the

circle \z\=*l and shall hove on the circumference of this circle the
value
(a2-l) cos 6+i (a2+l) sin 6
a*-2a2 cos 20-R *
where a* > 1,0 denoting the vectorial angle. (Agra 1930)
The function / (z), being regular inside | z \ = 1, can be expan¬

ded inside it in powers of z by Maclaurin’s Theorem,
i.e. /(z)= 2 *" where /” (0)=^' |c ■

n-0 fl •
On the circumference of the circle | z 1=1, he. z—e ,

fcos 04-/ sin *)-<- / sin *+cos 6)
,.2
and /(z) =-:-ai_at (^Tj^p757+l
fl2 e o-e-i0
< _
=a‘-aUeiie + P*i0)-t 1
- <rw __ g*~* (fl* -g~2<<>)
(a»_e>«> («2- a’ -e!"
Z
fl--z 2
„ I (2* ete jV* </0
. /w jQ 5rr?» • e-p+H»
V f27T / «2t0‘ -1
-HI. ^"O-sO
.jLiri (i+£+£+...)*
2mj’Jo o2 \ a1 «* /
_n • or 0. according as n is odd or even.

~2na*'an-1
f" (0)
Hence/(z)=/(0)+/#(0)z4-2T 22+...
Z z3 z6 z/fl* _ z
== —o ~l i 1 I- ~R 4- • • ■ I -l /fl2
//>2 a2 -* z2‘
a* o4 <2° l — ^
Complex Integration -07
Aliter. /•'Oi-ij/j::*
-Sic kbWn*
1
^2" Q2 -Z-Jô
=[£M^-^)SL
=1 r /i! _>i !(-!)"]
/l
[l-(-Dn]
2 an+l
= 0 if /; is even.
// i
= an+l
—rz when n is odd.
15. Find the singularities of the function
indicating the character of each singularity. (Agra 1944)

c/(5 -a)
Here
/W“5=T?
The function f (z) ceases to be regular at r = 0 because the
numerator is finite and the denominator has a simple zero.
Hence the function has simple pole at z=0.
The denominator vanishes where
ecla = | =e'nvi
,'e‘ z=2niria,
where // = (() ±1. ±2. .). The numerator is finite for these
values, and the denominator does not vanish for any other value
of z except z=?niria (/i = 0, ±1.. ). Hence there is a simple
pole at each of these points.
At r~a, the numerator has an isolated essential singularity
because it does not tend to a unique limit at z=a. Hence a
Laurent’s expansion is possible, and the coellicients in the
Laurent’s expension may be obtained.
n
c- I
rl (r-o)
r>+—
L
+
2—a (=-a - 2 ! + ...]
Hut el +(:-a)/a_j
(e elz~a)la - 1)
[\+cUz-a)+c*l(z-o)*.2 If-. .)
=-0+z-^+<^2V--)
= -(I+z4-fl+(7^y»21!+-”)x
[Ô+^+Tfp-- ■)+ei{'+zir+(Tn?+~ } +- -]
which clearly gives an expansion involving all positive and
negative powers of (z-a). So there is an essential singularity
at z=a.
oc ^2
16. If f(z) = Z f r 83-t j//ou- that f (z) is finite and couth
nuous for all real values of z but f (z) cannot be expanded in a

Maclaurins series. Show that f (z) possesses Laurent's expension
valid in a succession of the ring spaces.
oo Z2
Here f (z) = Z -.■-.I*
„=i 4+ri-z
When 2 is real, every term of the series on the right is +ve.
z2 r2 1
A,so <
1
Hence the sum function f (z) < Z -t, a convergent series
whose sum is ir2/6.
Hence Z * -=-« is a uniformly convergent series by

n=i 4 + nV
Weierstrass’s M test.
0 < / (z) < ^'2/6, when z is real. #
Again each term is continuous function of z as the denomina¬

tor 4-f-«2z2 docs not vanish for any real value of z. Hence
f (z) is finite and continuous if z is real.
If /(z) is considered as a function of complex variable, each
term possesses two singularities ±— {n= 1, 2, 3...). Hence
however small a neighbourhood of zero may be taken, there

Complex Integral ion
are singularities of the function, having r = 0 as the limiting

point. Hence no Maclaurin's or
Laurent’s expansion can be formed
for the function valid throughout
any region of which the origin is
an interior point. We can howe\er
expand / (z) in Laurent's expansion_
in the ring shaped region defined
by
|zl«l,|r|-2, |z | = §, | z | = |,
2
| z |= — in general.
17. Discuss the function repre¬
sented by the series
z
OO
}_ __l _ a > 1.
ft=o n ! 1 +a'inzr
The function represented by this series has singularities
at z= ±i/an {n= 1 2,3,.. ), since at each of these points the deno-
ninator of one of the terms vanishes. All these singularities are
situated on the imaginary axis, having r=0 as their limiting
point; hence neither Maclaurin’s nor Laurent’s expansion is valid
throughout any region of which the origin is an interior point.
The function being an even function of r tends to zero
as I z |-*-oo and is analytic on and outside of a circle C of radius
greater than unity, having its centre at origin. Hence for points
outside the cirele \z\-ktk > 1, it has the Laurent's expansion
^2 bt bQ
Zi + -Zi + —>j-t-* • •
where £2m+l = 0 since the function is even.
A — J_ f T2m-1 y
a~2n .
and u2m _on f U 4m
2m Jc “ nt, n ! a Zn-\-zl
CO q—2n 2% ft*—l OO q—2n 2^m—3

Now E
n—0 («-*"+z*) n ! n=o’t ! (1 -J- a~2"z
GO
y &72m-
— . J
ft—0 n !
CO a~-n
E z2m~l [\—a-lnz-i+(a-2n*
ft—0 n\
= V 22m-3 .1. V (q-2»)P ( - 1 )*\ Z“2p. SiUCC j Z | > 1

n=0 . j)=o
__ -2m—3 I" rj _z-2 + z-4 - pj- (1—ar*2"2+ . •• •]
_-2m—3
+(-ir-* (z-=r-,î+?jy+..-)+-- •]
__ -2m-8 [^_2_ -—2 gfl-4_{_ __ (_ I)”1"1 Z~2m+Z. ea~2m + . . ■ ]•
This series converges absolutely because j r | > 1 and hence

conveges uniformly.
The only term in the expansion which furnishes non-zero

integral is the term z_1 whose coeff. is
00 1% . <r*mp
* <-*>
J/-0 jr
—2m
which is equal to ( —l;""1.*®
Hence *,..-53 Jc J 0 <" 1)"‘'1

a-2mp 1
f .7
co % , a-2"‘p
= £ (-1)""1 —»
l'-0 P l
putting z^ke10 and integrating,

i.e. b?in-( — h’n 1 e° .
Therefore when | z l > 1, the function can be expanded m
the f°rm ew ev« e'l°' _

ak _ - T— — ■ ■ • • •
—• r u
By putting z' = I/z, we see that the function in r' has a zero
of second order at r'=0; hence the given function has a zero of
second order at infinity.
18. Consider the singularities of the function represented by

the series
v
1 1
„t>n ! J+2"z>
and obtain its expansion by Laurent's Theorem.

[Proceed as in question 17, Here a = 2.J
Complex Integration 2II
19. A function w hich has no singularity in the finite part of

the plane or at infinity is constant. (^gra 1936, 43, 57)
As the function / (z) has no singularity in the finite part of
the plane, it can be expanded in Taylor's series in any circle
j z |- kt where k is arbitrarily large.
/(-)= Atzt
f—0
Also if / (z) has no singularity at z = o© , /(I/z) has none at

2 = 0. Moreover since /(z) has no singularity in the finite part
of the plane,/(l/z> can have none in the finite part of the plane
/. /(l/z)= E Btz'= E Atz't.

r-0 o
But this is not possible unless Br = Ar=0 for all values of r

except zero in which case /h,= #„ and/(z) = AQ= B0 — const.
20. Show that a function which has no singularity in the finite

part of the plane and has a pole of order n at infinity is a polynomial
of degree n. (Punjab 195N ; Agra 36, 52, 57)
As /(z) has no singularity in the finite part of the plane,

00
vve have / (zj — E ArzT.

r=o
Since / (z) has a pole of order n at infinity, /(1/z) has a pole

of order n at zero.
n B
e ■: -h (->,
i -
where </> (z) is a regular function of z.
Hence (f> {z)+ E ~= E Arz~r.

5—1 Z r~u
.’. <l> (z) = A0= a constant

and A/i+i= A n. j-2 = ... — 0.
.*. /(-)•= E Arzr ; which is a polynomial of degree //.

r-0
21. Iff (z) — uA-iv is a regular function of z and u is bounded

for all r, then show by considering the function e /(,) or otherwise
that f (z) is constant. (Alld. 1946)
Let '/> (z)=e'l:) ; then | <f> (z) |=j e/iz) \=\eu+iv |=eu which is
bounded for all r, as u is bounded.
By Lioville’s theorem, ef (t,=a constant, and consequently
f (z) is constant.
22. The function f (z) is regular in [ z — a | < R; prove that,

if 0 < r < R,
f =L
77 r P
JQ p (“) e~ia dB’
where P (6) is the real part of f (a + rei0). (Agra 1957)

If the function f (z) is regular in the region | z—a \ < R, it is
regular in C==| z — a \ = r < R. Hence by Taylor's theorem.
f (z) = aQ+ax (z- a)+ a2 (z—a)2+ ...
where «,.= & jc (£sj* ...(1)

Putting z — a = reie on C, 0 ^ 0 ^ 2nt we have
f(a+re")= 27 an (reiB)\
Similarly / (?)=f (a-\-re~'e)

CO
27 a„rne-in0.
00
/U)+/(2) = 2 real part of 27 anrn (ei0)n.

ti—0
= 2P (0), where P (0) is the real part of/(r).
Now for n > 0,
1 r m ^=j_ r
2rri/ JC (r—a)n+1ah1 d2~hi
2tt/ jc
J< mf0“m (~~ar (z-ar”
dz
the term by term

-i,L!
integration being
Q
rm rm am (3 - a)
valid
•
since
in
yZ — a)n+l9
the series
27 an (3 — a)n converges uniformly.
Hence
2TT J <
J_ f /<-> dz - 1 % a [2 rme-mio ^ ie-ni0 jq
277/JC (z-ar+l 27T/m=0 ”'J0 rn
[V is -a=re~l99 (z ~a)*=*rei0\
Complex Integration 2J3
’ $ om i f” r"-"e-('"+'■) "de
2*1 mÔ 0
=0 for every value of m and n. ...(2)

Hence adding (1) and (2), we have
Lf by Taylor’s Theorem
rr/Jc
2 TT (z-fl)n+1 n\
fn (a) 1 f 2 real f(z) dz
i e.
n ! 2Vi Jc (2—a)n+1
1 f2*
=7Tf" L real/(ZM —ni0 </*.
As a particular case, when n=I, we hane
/' (j)=— f*’ real f<z).e~iOd0

*r Jo
i r2?r
=- P(6)e-'9dd
nr J 0
or directly, we have
/(g) dz,
/'(a) 2w/ f
JC (z-a)
0= f ±I^Ldz
Jc 2m (r-u)2
Adding these we get the required result.
23.By using the integral representation of f" (6), prove that
(xn \2 1 f *" e*z
\n '.) ~2m JC /i!.z»+1 Z’
where C is any closed contour surrounding the origin. Hence show
that f (£\'_/ f2* ,lx cos 0 dd.

\n\) 2*r J0
(Punjab 58; Agra 56, 61; Delhi 59)

By Cauchy’s formula, we have
f(z)
q jn+J dz, . . .(1)
where \ z \ =r is the circle C and f (z)

has been expanded by Taylor’s
theorem.
Here, we have to prove
(x'
' 2m ' Jc zn+1 -1 J
The right hand side of (2) is evidently of the form (1), where
/(z)=e*2. Treating x as independent of z,
/" (z)=xneX3t so that/" (0)=a°.
n ! f
dz.
Hence *”=[/" (*)]*=o
2-ni 2"+1 J
Now summing up.
f. xVl
dz
n=0 Ic n !. zn+1
00
-i j 2 a Mil dz
[as the series is uniformly convergent]
.X2 » Xn 1
-u ,£>"; 2"+l
r x” L —
-if C
ezi Z t* * n
—n z
n=0
n .> t
~
1
—-i- f exz.ell\-L: </2
2tt/ J c
gz{s+1/2) -
1
= 2tt/J
-f c ^
Z1T
« JL f27 eZv cos 0 /
2 m' J o
taking r=l so that z=e'*

=-L f2,r *2.t cos0^t
2tt Jo
24. 0 fzj o«</0 ore /wo regular functions; z=a is a once
repeated root of 0 (z)=0 and 0 (a)^0. Prove that the residue oj
0 (z) ip (z) at z=-a is
[60' (a) iJj" (a)-24> (a) 0"' (n)]/[3 {0" (a)}2].
Since z=a is once repeated root of 0 (2)=0, let
0(z)=(z-n) f(z).
Required the residue of at z==fl*

Since <f> (a)^0 and 0 (2), 0 (2) are both regular at z=o, it
foilows that / (2) is also regular.
Now x (2) = ——wil1 haVC 3 doub,e poê at z==a>

where the residue is given by
j . d f»/» (2)1
function
[because the residue at z=u a pole of order m of the
F (z) .(fl)1
(z2-ar 9 1S (W-1)! J* •h' (a). f(a)-<t> (a)/’ (a)
the residue^
[/(«)]*
Now we have to evaluate/(a) and/' (a).
0 (z)=(z-a)2f(z).
0' (zMr-a)2/' (z)4 2 'z-a)f(z),
0" (z) = (z-0)2/" (z) + 4 {z-a)f (z)+2/(r),
0'" (z) = (z-fl)2/'" (r)46 (z-a)f (z)4 6/' (2).
Hence 0* (a)—0,
0" (a) = 2/ (a),
/(a) *=40" (a) ; of" («)=/' (<*)•
Substituting the value of/(o) and/'(u) in ll), we have the
0' (a) U" (a)-0 (a) W" (0)
iesidue =---
6-A' (a) 0" (a) — 2</» (a) 0'" (Q)

= 3 [0" (a)]2
25 nature and the location of the singularities oj the
function • Show that if0<\z\<2n, the function can
he expanded in the form j^+a<,+0^+042* +. • •Olid find the
values of a0 and az.
Here f (z) = —-—rr has singularities, where z=0

z (e1 1)
and e*= 1 = e*2nir< or z*=±2n*ri,
where n=0, ±1, ±2, ..
It means that z=0 is a double pole ; the other singularities
at z=2tt/, — 2tti are both simple poles.
Hence in the region 0 < | z | < 2-rr, the function can be
expanded by Laurent’s theorem in powers of z with its principal
part containing the first two negative powers of z,
i.e. f (z) = a0-\-alz-\-aiz1-\-... 4z142s»

To determine the coefficients of aQi alt a2.. .and bl9 b2t we can
1
write f (z) =
( Zz•2 z* \
2 ^1+^-1-2"jH-j , + —1 J
1
z,(1+r»+rf+—)
=?[1-(r!+r!+--)f(r(+Ji+--’)+ "]
1 1 1 1
-z*~2z + 12 J20“"i"*“
26. The only singularities of a single-valued function f (z)
are poles of orders 1 and 2 at z= —1 and z=2t with residues at
these poles 1 and 2 respectively. If f (0) = 7!4,f (l)=5/2, deter¬
mine the function and expand it in a Laurent's series valid in
1 < \z\ < 2.
Since f (z) has a pole of order 1 at z= — 1 and a pole of
order two at z=2, the only singularities, hence it is of the form
/u)=__
(*+!)(*-2)«'
JK)
Again since the residues at these poles and two functional
values are given, (z) may be taken as
<f> (z)â0+alz-\-a2zt+a3z3.
These four constants can be determined from the four con¬
ditions which are /(0)=£, /(l)=5/2 and residue at z= - 1 is
1 and at z=2, it is 2.
The first two give
/(0)=3=? or o0=7
r (| v go~l-gi-4-^2+^3 5
and J '1' —2'
2
oiA
i.e. a^~\- a2-\- a3 — S aQ-= —2. ...d)
Again residue at z= — 1,
=Limz_,_i {(z+U/(r))=Lim2^_1 1 (given)
i.e. °o““<7i +—^3=9

Complex Inegration 217
or -ax + a2~ a3 = 9-o0=9 — 7 = 2. ...(I)

.(I)
Also a1 + a2+<73=_2. ....((2)
2)
From (1) and (2;, at z = 2. we find that o2-0. and
°i + fl3=-2. ...(3)
.(3)
Again the residue at z = 2, a double pole, is
riftm
L^V2+l/Je=2
_rd
1^2 2+1
1
Ja=2
_r(f, + 3^**j (2+ 1 ) - (7 + £j|Z+t73Z3)'l
L iz+W J ,=2
— o [(fli + 12a-).3 —(7 + 2a1 + 8a3)J = 2, given,

/. e. ai + 28^=25. ...(4)
*• e‘ .(4)
From (3) and (4), we get on solving
tf3 = l, a1=~3.
Hence the required function is
fiZ)=J^±l-
(2-Hlj (2 —2j*•
Now this has to be expandedin powers of z in a Laurent’s
series valid in 1 < | z | < 2 i. e. in the form
J (z) = a0+ £ anzn-\-Z bnzm~n.

n-l 1
where c fM a-=21/ Ic
C being the circle in the ring space 1 < | z | < 2.
/W *.
We can separately calculate "0=2“-. j ~-Z) </z.
A liter. f(z)= 1 + + (T^2+» (by Partial fractions).

For 1 z 1 > 1 and j z 1 < 2, we have
1
2-2
= -i(1+2 + ^ + -.),
4 (1—z/2)s—4O 2)
1 +z+ —+.. .
=!(•■-■ 4
Hence combining these results, we have the requtired
expansion as
/(-)=!+
+ 5(1+r+3-4J+--’)
-K1-;+?+*")-G+l+-")+i(1+,+3-4+"0
is the required Laurent’s expansion.
Exercises
I. Obtain the Taylor’s expansion
€n! 2’ (—Dn («+1) (2 - i)n+1» (! 11 < *)•

z* n=0
2. Obtain Maclaurin’s expansion
z-i-'= -1 - 2
2—1
2
n=l
I 2 I < 1.
3. Obtain the first terms of the Laurent's expansion

e* l,i 1, _ 11,24. f) <: 2| (>»> I z | < V2.
f Ans. (i) ^ ( — 1>"^1 ” 2?^)*
co
(ii) ( - 1 )n+I p »»+2
(iii) 2
n=0
(- l)»z-*n-9 tl - 2W) 1
J
(a) Find the Laurent’s expansion of
(2*- 1)
(:+3) (2 + 2)
in the domain defined by 2 < j 2 j < 3.
(b) {z2 — 2 — 2)~’ in the domain defined by 1 < | z \ < 2
ADS. (a) -5.8 2 zn+ 3 2

n~ 1 ^ n+i H—0 n-f 1
1 00 (-1 ),l+1
(b) - 27
3 n=0 2"«‘ 2n+1
I
Expand - „
2 (2-2) (2 — 1) *
(a) 0 < |2| < 1, (b) 1 < \z \ < 2, (c) > 2
[a„, (a,
oo 1 oo
1
(b)
«=o 2,,+2 22 ,I=l 2,,+l ’
(C)
f 2'-1 , - 1 1
•-* *',+! J *
1
Find the different developments of in powers
of z according to the position of the point in the 2-plane.
Expand the function in Taylor’s series about z = 2 and*
indicate the circle of convergence. (Agra 1943)
Prove that cosh {A (2+1/2)}= X anzu

// = — oo
1 c2n
where a„ = — I cosh (2A cos 0) cos nd dd
for any finite values of z^L0.
Show that sin {c (2+ 1/2)} can be expanded in a series of the
type
a0 + aiz + asz2 + .. .bl + ^+ ...

— Zo
in which coefficients of both zn and z~’‘ are
I rin
2~ I sin (2c cos 0) cos nO dd.
< 1 and * < 121 <1/fe,,hcn
1 00 rn
= ±— 27 cos nO%/(\-2h cos 0 + b2) dd.
/;= —co J 0
The integral function/(2) satisfies everywhere the inequality

1/(2) < A 1 21 * where A and k are positive constants
Prove that f[z) is a polynomial of degree not exceeding k.
(Agra 1957)
.
12
function
{
In the annulus defined by | a/
h~
--z~n— x /
l1/z
(z-a)(b-z)J
*1 < 1*1. expand the
by Laurent’s theorem and prove
that it equals to
• *»+-s •*»(*» +10

, <s l. 3... (2/ -1) ■ ■ ■ (21+2n-1 y-aV
where ^ 2“+"./! (1+n) ! W'
13 . Establish the expansion
/«-/«+* {-i-V-C-f")r ■}
and show that the expansion is valid as long as it lies within
~2~ )’
14 . The function f(z) is regular when I z | < R\ Prove that
if | a I < R < R'.
J w 2tt/Jc (z-a) (tf2-za)

where C is the circle \z\=R. Deduce Poisson’s formula that.
if 0 < r < R,
nre^.-ir
J^re 2^ J 0
___,/(*«*)</*
/?2 — 2/fr cos (0 — 0) + r2'
|^Hint. Proceed from the R. H. S. by putting
— f / — 4-} / 00 dz] .
2ir/ Jc (r-a lR2-z*)J J
15 . Find out the zeros, and discuss the nature of the singularities
sr .\ z —2 .
2—2
1 •
of /(2)= — Sin z— 1
find also the residue at the singularities.
rAns. zeros 2, l-f- ; 0 is a pole of order two and 1 is

L f,7T
an essential singularity. The residue at z=0 is
2 cos 1 - sin ij.
16. Obtain the expansion for which are va,id (i) when
z | < 1, (ii) when I < | z | < 4, and (iii) | z | > 1.

(Agra 1950)
17. (a) Let D be a region bounded by a simple closed contour C.

lf/(z) < M on C and/(z) is not const, then |/(z) | < M
at all interrior points of D.
(b) If f (z) be an analytic function regular for j z j ^ R and
1/(2) for | z | = /?, and / (z) has a zero of order p
at 2 = 0 ; then prove that
Mrv
I fW) I ^ -jp* 0 < r < R. (Punjab 1958)
5-21. (i) Zeros are isolated : Let / (z) have z = a as a zero

of order m. Then /(z)=(z-a)"' (z) is regular in
the neighbourhood of z=u i.e. within | z—a | < r and </» {a)~± 0.
by Taylor's theorem,
f(z)=am (z-a)’n + am+l (z-a)m+l + am¥2 (z-a>"'+2 + ...
= (z-ar {am + aM+l (z-a)+am+o (z-a)?+..
= (z -a)"‘ <l> (z),
where (z: = am+amH (z-a) +...
Hence </> (a) = am.
Let us write </> (u) = 2c ; then since the function </> (z) is
continuous at z =a, there exists a region \ z-a I < 8 in the neigh¬
bourhood of z=a such that 1 •!> (z)-</» (a) | < | c |.
Hence | <l> (z) | = J </* (z>- (a) + «/-(a) |
= 1 7* (c(a)-<f> (z)} j
^ I '/» (a) H 7> (a) |
> I 2c |-1 c |
= 1 c |.
Hence •/> (z) does not vanish in | z-a | < S, /. e., the zeros
arc isolated.
(ii) Poles are isolated. Suppose f (z) is analytic everywhere

except at z=a where it has got a pole of order m ; then the
function is regular in the neighbourhood of the point z = a,
and has a zero of order m ; but the zeros are isolated ; so the
poles must also be isolated.
(iii) If f(z) has a pole at z=a, then |f(z)|->ao as z->a in

any manner. We know from Laurent’s theorem,
*
/fz)= Z an (z-a)"-f Z bn (z-o)-"
={z-a)~™ ^b„+bm-x (z-a) + bm-i (z-a)2+. ..b1 (z - a)m~l
+ 2a„(z-a)m+” V ...(])
n=0 |
. =(z — tf)“m 0 (j) ; ; .
where (z)=/?fn + ^_1 (z - o) + ... +£x (z—tf)™-1
+ 27 an (z-a)m+n.
ftzziQ
But bm^z0 and (z) is regular in | z—a | < p.

Hence <f> (a) = bmÔ.
Hence as in ti) we can find a neighbourhood [z-al<8,
of the pole at z=a, such that.
f(z)> | | z-a
r
bm~ |r 171 —\
2 |1 bn I (*-«)—»+ 2:
nznl n=0
(*-*)«+" }
1 JJ
as z->a in any manner.
tn-1 w _
the expression 2 bn (z — a)m~n+ 2 an (z—a)m+n-+ 0
n=1 n=0
and \ z - a\-™ \ bm \ -+oo as z-*-a.

Hence |/ (z) | to infinity, as z-*a in any manner.
Aliter. We have
I <f> (z) | = | * (z)—<t> {a) + 4> (a) |
^ | «Ma) M (z) |
>\bm |-1 !bm |,
fV | 6 (z) | < 1 | bm | because of continuity at z=a.]
Hence I <f> (z) ** a I bm |.
i.e. l/(z) !=l z-cr |-"» | <f> (z)| > | | 6m | (z~fl)-m.
Hence j/(z) j tends to ->» in any manner as z->a.
(iv) Limit points of zeros. (Utkai 1960)
Let f (z) be a function regular in a region D, and let a,,
a2...an... be a set of points having a limit point a. inside D.
Then iff(z)=0 at the points ax, a2.. ,a„. .it follows that either
f(z) vanishes identically or else f(z) has an isolated essential
singularity. <l
As J(z) is a continuous function, having zeros as near as
we please to a, f (a) must be zero.
As we have proved that zeros are isolated, it follows that a
cannot be a zero of f(z). Hence f (z) must be identically zero.
Comple v Integration 223
If f (z) is not identically zero in the region D, then z = a

is a singularity of /(r). Hence in this case the singularity is
isolated, but it is not a pole, since f(z) | does not tend to infinity
as z-+a in any manner. So the limit point of zeros must be an
isolated essential singularity of f(z).
(v) Limit point of poies. Let / (z) be analytic everywhere

except at a set of points which are the singularities, say
/>,, b2, bz.. b„. .infinite in number, having a limit points in
the region D. Thus b must be a singularity of/(r), as f <z) is
not bounded in the neighbourhood of h The point h cannot
be a pole as it is not isolated. Such a singularity is called
non-isolnted essential singularity or simply essential singularity.
Clearly sin -j is zero at- = /^r for different values of #f = (l, 2,
3...). These points have zero as their limiting point which is
an isolated essential singularity. But tan — has got poles at the
2
set of points given by 2=— for different values of // = ( + !,
±3. ±5...) and this set has got zero as its limiting point which
is a non isolated essential singularity.
^-^■^22. The behaviour of a function near an essential

singularity.
Weierstrass Theorem. In every neighbourhood of an isolated

essential singularity, there exists a point (and therefore an infinite
number of such prints) at whit It the function differs as little r.s ity
please from any previously assigned number or it can be stated
thus :
If z —a is an essential singularity of a function, then for any

arbitrary number C, arbitrary e > 0, and arbitrary y > 0, there
exists a point z such that for 0 < | z~a | < 7, we have
\f(z)-C\< €.
(Sugar 1951; Agra 1948; Punjab 1960; Raj. 1958)
Consider an analytic function f (z) which has an isolated

essential singularity at z=a. Let C be any given number real
or complex. We have to show that, given two positive numbers
c and y, no matter however small, we can find a point r in the
region | z - a | < y at which 1/ (r) - C | < e.
If a is a limiting point of the zeros of the function f (z) - C,

the theorem needs no demonstration for z, may be taken to be
any zero in the given neighbourhood.
But if a is not a limiting point of zeros the function f (z) - C,
has no zeros in the given neighbourhood for a sufficiently small y
and hence
8 (Z)=[7(z)-Cj
is regular in 0 < | z—a \ < y. We have to show that there exists
a point in this region at which | g (z) | > 1/e.
Let us prove it by supposing the contrary /. e. | g (z) 1 < 1/e
there. Then if
OO
g [z)=i}j (2)4-27 bn (z-a)~n,
1
where ip (z) is regular, we ha\e
b°—2 hL
where C is a circle \ z-a | = p < y.
* w ('■-
Hence j bn | =
2Tile8 (:) (z_a)n_1 d:
1 1
-L - pn-lc* <»-!> fl pC<#
2tt 6 Jo I
Pn/e.
Hence | | < pn/e.

Since, however, is independent of />, we find by making
p->0, that />„ is zero for all // and hence g (z) is regular at z=a.
This is impossible since it would imply that f (z) is regular at
z=a or else has a pole there. The assumption that | g (z) | < 1/c
when 0 < | r— a | < Y, is therefore untenable.
We have thus proved that
\g(z)\>l/€ or \f{z)—C | < €
at one point at least in the neighbourhood of a.
This property sharply distinguishes poles from essentially
singular points.
Example. 1-f *2 —,4-.. —n+... has an essen¬
tially singular point at z=0. (Refer also Q. 15 P. 207)

Picard’s Theorem. Every equation f(z)=A has an infinite
number of roots in the neighbourhood of an essentially singular
Complex Integration
225
value of T beinS n° eXCeplion except for a< most one particular

Let ,1 IS
= A.
neiglCuarhb:dShr,Lhr *, ^"îte nHD,ber »"

zero. ’ U 
satisfying this condition.3” mfin'te n°’ °f va,ucs of the integer m
for example, in IheâsTsin’l/zneaf“=0!" ^ "° CXCcplion'
2. Find the singularities of the function 1 1 ,
d,c°“ng the character of each singularity. L *‘/a- /J

fft has been given on page 207.J
s*23. The point at infinity.
in a sin”,ar ~
valurr0dUCe ,he ,ranSfb™a*i0n -h tha, the large
Ihcre is a ZZZTlol
Provided none of the points * °Ce
^h ^ °.--plane.
etween lilc points : and
to say that z is the “point a.S • Zfr°' Sometimes it is convenient
this Point is not a defin^ point.1 Whc“ ,S lhc °ri«in bul
hood onhCe5pointrz'=0.fUnCtiOn K’=/(2)='/' (z’>in ,hc oeighbour-
point z' = 0PPTheeI,hh fpnCtlon ^ (z'-> has a zero of order nt at the

1 “en by Taylor’s expansion.
and '' <Z>=“'mZ "' + am+'z'm+2 + a"‘+iZ'm1’ + ...

S0 ' e CXpansion of J (a) which is valid for sufficiently larye
values of I z \ will take the form
i.e. f(z) is said to have zero of order m at infinity.

r (ii) Let the function 0 (z') have a pole of order m at the
point z' =0. Then by Laurent’s expansion,
* (*') = F7n + ^T+ • • + +‘** .(I)

For sufficiently large values of | z |,/(z) will be of the form
/ (z)=&mzm4- bm-xzm~x +... bxz+a0+J1 +...
Here we see that the function / (z) has the pole of order m
at infinity.
(iii) Similarly the function/(z) is said to have an essential
singularity at infity provided <f> (z') has an essential singularity
at z' = 0. Thus f(z) = e* has an essential singularity at infinity,
as the function
ex ,s' = 1 + -p+^ 4-...
has got an essential singularity at z'—0.

(iv) Suppose that <t> (z'j has got a removable singularity at
z'=0 ; then all bt = b2.. .bm~ 0, hence
<f> (z')=a0 + alz' + a2z’2+.. ,
which is the Taylor’s expansion for 0 (z') in the deleted neigh¬
bourhood of z' = 0.
Hence the corresponding function/(z) is given by
/(z) = tfo+ * * •»
which is the Taylor’s expansion for /(z) about z=», i.e. it

consists of the negative powers of z.
Hence we conclude that f (z) is analytic, or has got a z;ro
of order m, or has got a pole of order mt or an essential siugurity
at infinity. provided the corresponding function tj> (z‘) has got the
same behaviour at z' = 0> i.e. it is analytic, or it has got a zero of
order m or a pole of order m or an essential singularity at 2'=0.
5*24. Rational Function.

Theorem. If a single valued function f (z) has no essential
singularities either in the finite part of the plane or at infinity, then
f (z) is a rationa function. (Agra 1959)
Let us suppose that/(z> is a function having poles as its only

singularities in the extended complex plane. Then/ (z) can have
only a finite number of poles, for otherwise, the infinite aggregate
of poles wOU|d have at |eas, one limjting pojnt which wQu|d a
non-isolated essential singularity. Hence a contradiction.
Suppose that /3„ ft,.. .ft, are the poles of order n „ m2. m
say Then the principal part of the function/(a) will consist o"f
the loUowing series :
<1*1 (z)=—^n- + . b\"h

(Z—Px) (2-/?,)* ’“(z-01ywl
..(1)
U) = -f. -—bflfTln
Clearly the principal part at infinity off(z) is
Let us now consider the function

0 <*)=/ W- K*. (*) + #, (*)+...<A. (Z;J + * (z)]
n.,n„N,°W ue runc,i°n 0 (Z) is eviden‘ly tegular everywhere in the
been tewo'vedJânâhâri^nity.'^Heîce^tzJvviM1 be bounded
f^Z) — 'I'l (z) + ‘/>8 (Z) K. (z)+<b (z)±C,
5-25. Zeros and Poles of Meromorphic functions.
Atom we shall prove on importer, theorem.

Theorem. If f(Z) is meromorphic, inside a dosed contour C
■ nd ts not zero at any point on the contour, then ’
snj'm *-"-*•
2WJ TG)Jz=N~P' .. (I)
..(i)
where N is the number of zeros and P the number of poles inside C.

(a pole or zero of order m must be counted m times.)
It is essential in the statement of the theorem that /(z) has

no zeros on the contour, otherwise ' (z)lf(z) will have infinities
on the path of the integration.
Let us suppose that z=ar is
any zero of order mr and bk be
any pole of order nk of the func¬
tion / (z).
Then in the neighbourhood
of aT which is of order wr, we have
f(z)=(z-ar)mrgr(z) ...(1)
where gr (z) is clearly analytic and
non-zero at z=or, and where (1)
holds in the neighbourhood
| z—ar | < <xr whose boundary
is Cr.
Taking logarithm of equation (1) on both sides, we have
log/ (z)=mr log (z - ar) -f log gr (z)-
Differentiating with respect to z, we have
f (z)_ "h ■ gr' (Z)
f(z) (z—ar) gr(z)'
p f (?)
Now sr is analytic at z=aT.
g (z)
Hence -yr— has a simple pole at z=ar and the residue
there is
(z—ar) mr_
Lim ==m.
z—a,
IfJ cr
2-ni
Ilil1 dz=—
f{z) 2<r/ JfCf \(~ “ ar) Sr (z) J
=mr + 0.
Hence the total sum of the residues corresponding to zeros of
/(z)is Zmr=N. ...(4)
Similarly if z~bk is a pole of orderof / (z), then in the
neighbourhood | z—bk j < dk of bk> we have
/ (z) = (z - bk)~nk <bk fz), ... (5)
where <f>k (z) is analytic at z=bk and does not vanish at z = bk. '
Hence by logarithmic differentiation of (5). we have

f'(z) nk <h'k (z)
f (z) (z-bky +k(z)
and as before,
1 f/'(r) , 1 ff ~'h , <jz=-n
2tSJ/<;)
2 rri dz~2m]\(z+bt)+ ^(z)fd “
Hence the total sum of the residues corresponding to the poles
of
of /U) is -Enk^—P.
Hence the result follows from the contour theorem
f fllZl dz=2ni (Zmr-2nk)=2ni(N-P).
Jc f(z,
Corollary : We derive an important elementary result from
the above theorem concerning the number of zeros of any regular
function.
jff (zj is regular in a domain D which is simply connected and
bonnded by a rectifirble Jordan curve C, is continuous on C, having
no zeros on C, then the number of zeros of (z) in the interior of C
is ^ orgf(z)t
where Ac argf(z) denotes the change in arg f (z) as z moves along

the contour in the positive sense. (Lucknow 1960)
If f(z) is regular within the contour C, then P=0 and we
have from the preceding theorem,
' f (LI*> dz =
2m ] f[Z)
— N,
N being the number of zeros of/(z) within C. We substitute

Z=log/(z), -^
dZ ftz) jC
dz - J\Z) • / a >v
Then 1 f /'<*>dz = f1 f[ dZ.

^ 1 \
2m J c J (z) 2m J c,
...d) \ I
where C, is the closed curve described X.
by Z as z describes C. -^
Since Z=log/(z)
=log l/(z) | + / arg/(2).
fJ c\ /
dZ={\og 1/(2) 1 + arg/(2)} on the contour C,
= iAc arg/(z). ...(2)

Because Cx is a closed curve [Z

being the starting point and coming
again to it after traversing the whole
closed contour] log \f(z) is single¬
valued and we see that A c log \f(z) [ = 0.
Hence we get from (1) and (2),
1 1
= br Ac *Tgf (z)==Ar*
the result being known as the principle of the argument.

5*26.Rouche’s Theorem. If f (z) and g (z) are analytic
inside and on a closed contour C and \ g (z) \ < \f (z) \ on C, then
f (z) and f (z) + g (z) have the same number of zeros inside C.
(Lucknow 1960, Delhi 59)
In the first place it is quite clear that neither / (z) nor
g (2)+/ (z) has zero on C, because if z=z0 is a zero of/ (z), then
/(z0)-0.
Since | g (z0) | < |/(z0) | and so g (z0)=0 also. Hence
| g (z) | < |/(z) | is not satisfied on the whole of the boundary.
Similarly if f(z0)+g (zo)=0, then |/(z0) H 8 (*o) I which again
contradicts the hypothesis.
Let <f> (z) = and let A' be the number of zeros of/(z)
and N' that of/(z)-}-g (z).

Then by § 5’25.
...d)
...(2)
*■
Now <t> {z).f(z)= g(z);
then (z).f(z)+f' (z).'A (z)=g' (z).
. ^_JLf C±K±Mld2
*• " 2»/Jc /+*/
_ X f /’(!+'*)+/*' ^
2ni J C /(I+«A)
= _L f + (!+«/')
2wi Jc 1/ }
The expansion by binomial theorem is valid as ! < 1

1
• *
=^+0,
as the integral of each term is zero by Cauchy's theorem, since
</> and its derivative <£' are analytic functions and the integration is
taken round a closed contour.
Remark. /and f-V g have no zeros on the contour as utilized
in (1) and (2), otherwise the denominator will vanish somewhere
on the path of integration.
Aliter. We know that from § r25,
J_ r
Ivi J c
rf (?) dz = N
/(z)
and
• •
1
^Jc7(f*=^Jc d['°ef(2)Uz-
=2ni Ohe variation of log /(z) round the contour C).
A' = ^- Jclog/(z).
4.7T •
Also log/(z) = log \J (z) | + i arg/(z) and log l/(z) | is

one-valued.
Now /-t- g=/(l 4 £//).

arg g)=arg/-f arg (1-f g//) and so if we prove that
•^C (1 -4 g/J ) = 0, the result will follow.
Now the point iv= 1 -j-gjf always lies interior to the circle of
radius unity in the w-plane whose centre is I [since 1 +g/f
then as z describes C, w describes a closed contour in the M-plane
about iv—1, not enclosing the origin, hence arg iv returns to its
original valuej and J iv— 1 | = | g / f\ < 1.
Thus if iv = pe^t then </> lies between
- tt/2 and tt/2.
arg (1 +g//) = arg h’=</.
must return to its original value as we go
round the contour because we do not
enclose the origin.
We can deal with it in another way also.

Let F(z)=l+g/f.
R {F (z)}=R (l +g/f) = 1+F (gif)
> i-\g/f\ > 0, V \g/f\ < 1.
Hence as z goes round C, F(z) describes a closed path entirely
to the right of the imaginary axis. Hence its argument returns to
its origin value. Since <f> cannot increase or decrease by a multiple
of 2rr, hence the theorem follows.
1. If a > et use Rouche’s theorem to prove that e3=azn has n
roots inside the circle \ z \=I. [Luck. 1958, Agra 1956 (Optional)}
Let/(z)=azn, g (z)= — e2, then /(z)-Fg (z)=nzn —e2.
Let C be the circle with centre at the origin and radius unity.
On C, |/(2) | = | a zn\=a. | z |”=0.
Further | g (z) | = | -e3 | = J e* |
=i1+z-*-r!+jrK--l
z* 1 I z3 |
< 1+1 z | + 21 I + I 3 ! I
< 1 + 1+2L!+31!+
Thus on C, 1 g (2) | < e,
ie., \ g (z) I < l/(z> I since e < a.
Hence |/(:)|>UWI on C, so by Rouche’s theorem
ftz)-g(z) will have the same number of zeros withinlz|— l
as/(z). But/(z)=azn has got n zeros all at the origin, hence
the polynomial azn - ez has n zeros within the circle I z 1 = 1.
2. Show that, if k > /, the equation zne*~*=l has n roots in
I z 1 < 1.
[Hint. 0. Put /(z)==efc2n. g(z)=-c2,
1/(2) | > | g(z) I if A: > 1]
$•27. Fundamental Theorem of Algebra.
Every polynomial of degree n has n roots. (Luck. 1958)
In the first place, z" has n zeros all at the origin. Now con-
sider any polynomial
*f> (z) = aQ-\-a±z + ... onzn, an^0.
Let /(z)=anzn and g (z)=a0-t axz+ ... -\-an-xzn~x.
Let C be a circle with centre at the origin and radius R > 1.

OnC, |/(2) Ma«l R".
Further | g (z) \ ^ | a0 | + | ai | R+ .. i«n-ii-^nl

<{'o0l + . | o„_, 1}/!”-■ on C.
Now choose R such that
I an I -H <?|1 -t- • -I °n-l |
l\ / -" j “ j *
I I
j e. i a0 | + ... | 0n-i I < a*> 1
|g <z) I < K I on C.
Hence | g | < i/| on C, so by Rouche’s theoiem/(z) and
/(z) + g (z) will have the same number of zeros in | z 1=/?. But
f (z) has n zeros all at the origin, hence the polynomial <f> (z) has n
zeros.
5*28. Theorem. If z be taken round any part of a large circle
with origin as centre and radius R, and if d be the change in arg z,
the change in the amplitude of f (z) will differ from nd by a quan¬
tity which lends to zero as R tends to infinity.
Let f(z)=zn (a0-{ y + .. .^7, )
Thus arg/(z)=arg zn4-arg + •••?")
= h arg z + arg ^0+T +

a2
Now ai ,aj_,
z + z* * ’*zn
°_n
^ ai 1+ 3
z
+•• • —
. flf*
< 1 jIlJ i <7o I gn I

--4-
ri I •
5s R + R R"
'i (say). .(1)
We choose R so large that
r, < \ a
Then the point
flo+<F +** *z~"'

should lie within the circle
I z-a0\=rA.
Since z-tf0 1 = G\ , °2 ,
T^z2 +*
it lies within the circle with centre A or a0 and radius rx.
Let OP' be the tangent to the circle (2).
Then arg (*0+^ + .. pf) - arg fl0=Y), say,

is not greater than the angle AOP\ and is such that it can
be made as small as we please by taking R sufficiently large

i.e. decreasing rx.
Now arg/(z)=/7 arg z+arg ^0±yj.
Thus Lim^^ arg f(z)=n arg z-f arg a0.
Hence we see that the change in arg/(z) tends to n times the

change in arg z when R-*-oo.
The theorem proved above will be found very useful in
locating the roots.
3. Show that
z«-f2z3+ 3z2-Mz+5=0
has no real or purely imaginary roots and that it has one complex
root in each quadrant.
Let <f> (z)^=z4 + 2z3 + 3z2+4z + 5.
The equation </» (z) = 0, obviously, has no real roots.
For if we put z=x, 4. (x)=x* + 2x*+ 3x:+4x + 5 = 0 has no
real positive roots. Also take z= — x, and write
«/> (-x)=x*-2x9+3x2-4x+5
(x2-2x+ l) + 2xz-4x+5
=**<*-l)*+2{(x-...(1)
Since <f> { - x) is greater than zero for x > 1 and also for
0 the given equation has no real negative roots.
Let us now consider z=iy. Then we get
</, (/>)=>*-2/>3-3j8-f4/y+5
= 0’4 — 3y*+5) + 2i (2y - y9)
which can vanish if the real and imaginary parts vanish separately
i.e. )A — 3_>2-j- 5 = 01
2y—y9=0y •••V)
But these equations do not hold good for the same value
ofy. Hence there are no purely
imaginary roots. Thus the given
equation has no roots either on the
real axis or the imaginary axis.
Now we shall examine the change
in arg/(z) when z moves over the
closed curve which is composed of
(i) part of the real axis, (0, R).
(ii) quadrant of the circle z = Rei0
where 0 < 6 ^ n/2.
(‘ii) Part of the imaginary axis (iR, 0), where R-+00
0) Clearly when z passes through all the values between

0 and oo along the real axis,/(zj remains real and positive, and
hence no change in arg /(z».
(ii) On the great circle z=Rel°, R-»ce, we have
/ {Re‘e, = z* (1 + 2/z + 3/z2 + 4/z3 + 5/z*)
= R*t*'* [1 + 0 |z)J
where 0 (z)-*0 uniformly as /?-*-» obviously.
Hence we conclude that while z moves along the positive

quadrant of the circle, the change in arg/(z) is
~~4‘2 = 27T>(^ V changes from 0 to
(iii) While z moves on the imaginary axis, we write z=iyt

f(z) = u + iv={y4 -3>-2 + 5) + 2/> (2-y2).
Hence u =y4 - 3y--r 5; v=2y (2-y1).
When y varies from to 0 in the z-plane, we have
Urn;-**} — = Lim>-*So 2y. (2 ~y2) = 0

u yl — 3^-d
* “ = ^-^ + V>0, for every v is pusilive for
y < V2 and negative for y > y/2 i.e. if y changes from so to
V2, the points (m, v) moves from oo , where the curve is parallel
to the w-axis and lies in the fourth quadrant, to the point (3 ()»
on the M-axis, which shows that there is no change in the argu¬
ment when y changes from oo to y/2, g
Also When y changes from y/2 to 0, the point in the iv plane

starts from (3. 0) upto (5, 0, on the real axis and remains in the
hrst quadrant. Again the arg/(z) remains zero.
Hence the number of roots in the first quadrant is
<0+27r + °) = l .
If we have to find the roots in the

second quadrant consisting of the imagi¬
nary axis (0, iR)t 77
and (-/?, 0) on the real axis letting

7?-*oo, we find the same number of
roots in this quadrant also by the method
discussed above.
As it is quite clear that the coeffi-
cients of the equation are all real, the conjugate imaginary roots
appear in pairs. Hence the remaining two roots lie in the third
and fourth quadrants.
4. Prove that one root of the equation z*+ z3+l=0 lies in the
first quadrant.
The equation 24+z3+l=0 clearly has no real roots. For if
we write z=x we have f {x)=x*+x3 +1 which does not vanish
for any positive real value of jc. Hence there is no positive real
root. Similarly if we put z= -x.
/(-*)=a4—*8+l
= a3 (x— 1) -f- 1 > 0 if x > 1.
/(-x) = *4 + (l — *)(l-t-*+A-2) > 0 if 0 < x < 1.
Hence there is also no -ve real root.
Suppose w=z* = z3-f-1, aDd the
contour described by z contains the
three portions
(1) a axis from 0 to R,
(2) first quadrant of the circle
z=Rei9t
(3) y-axis from iR to 0, R ten¬
ding to infinity.
(a) When z moves along the positive real axis from 0 to
oo, then n’=*4+*3+l passes along w-axis from 1 to °©. Hence
arg/(z) remains constant and equal to zero.
(b) On the circle z~Rei0.
/(*>=** (l+i+-j)
=*4e“9 [1 + 0 (-)]
where 0 (z)->0 uniformly if z-+oo .
This shows that if z moves along the first quadrant of the
7T
circle, the change in arg/(c) is 4.^ = 27r.
(c) At points on the .y-axis,

«=>4+l» ->-3,
so that we have Limy-ôo — = Lim.>->oc — —4—-j-^0.
When y varies from co to 0, w» passess through the curve in

the fourth quadrant from oo to 1.
Hence the arg /(z) at A/, the

find value of w is
Limô ^,4 j
i.e. the argument /|z)=0 in the
final stage as well as the initial
stage, i e. there is no change in the
argument.
Therefore the total change in
the argument is
(0+27r + 0) = 27r.
Hence the number of roots in the first quadrant is one.
5. In which quadrents do the roots of (he equation
z4 + 23442a + 22 + 3 = 0, lie ?
Let /(z) = z4 + z3 + 4z2+2z-f 3.
Obviously /(a)=a4 + a3+4a2 + 2a+3 does not vanish
for any positive value a so that /(z) has no real positive root.
Also putting z = - a, we have
/( — a) = a4 - a3+ 4 a" ~ 2 a + 3
*a» (A - 1) + 2a (*—!)+3 > 0 if a > I.
/(_a-) = aM-4a-+2-2a+1 - a3
= A«-f4A7 + 2 (l -A)+(l -a) (l-f* + x2)>0
if 0 < v < 1.
Hence the given equation has got no real negative root.
Let us write z=iy ; then the equation y
becomes
u + iv=>>4 - iy3 - 4 v* + 2iy -f 3 = 0,
i.e. m=>,4-4;-2+ 3 = 0,
v = — y3 + 2v =0.
Clearly these two equations do not
vanish together for the same value of.y.
Hence there is no purely imaginary root.
Now consider/(z) = z4 + z3+4z2-f 2z + 3 on the circle Re'*t
R tending to infinity.
/ (z)=z4 (l + - ■+ -2 + -I- * 4) ,
arg/(z)=arg (z4-f z3 +422-f 2z+ 3)
= argz4 arg (l+^+*2 + . • •)
taken round the first quadrant of the circle | z \=R.

238
Complex Variable
passe froL 0 To , Dg, axis is 2ero because as ,

°° u ®tue reai axis* * p^ ^ 310«
fiiV
(u)
on ,h ,Hen“ arg f(Z) along re“l îs is zero.
On the circle z=Reie.
ar8f(z)=4.-=2ir.
(iii) As z varies on the imaginary axis.
arg (z*+z°+4z*-h2z+3)=tan-> F/M
= tan-1 ^
y = J2"5andmtheatdr °f-the eXpression in> brackets vanishes at

and the denominator at y=.x/2 and v— 1 *.
rational fraction varies as follows as y varies from oo to 0 ^
3’=0° V3 y/2 , o
y/u =0 - OC + o - „ + 0
when j vanes from + oo to V3, v/« change from 0 to oo.
in h fn y " v^* w 1S ^"ve and v is negative, i.e. (u v) /ies
thi dequad« tqUal"Dt; ^Cn < y < ^ (w' ^ ]ies in *he

quadrant d H ’ Z ° * ^ fr°“ l° L ,ies in 2nd
q adrant, and when y changes from I to 0, (u, y) lies in the
bee qUa rant’ thaI ,s- ,tiere is a decrease in argument by 2n,
wise dl t- CUfVtSu haVC traversed round tbe "igin in the clock-
is -ero,
is -ero ifr /? is large
ThUS 'he ‘°tal varia,ion round <he first quadrant
enough.
Hence there are no zeros in the first quadrant.
Since the coefficients of the given equation are all real the
conjugate ,maginary roots will occur in pairs. It follows that
wre ore no zeros in the fourth quadrant, and two in each of the
second and third quadrants.
6. Prove that the equation
2*+fe+I®0
has a root in each of the first, second and fourth quadrants.
(i) Suppose r moves along the posi¬
tive real axis from 0 to oo . Now putting
z = x,
f (x)=u+iv=x3+ix+J,
u=x3+ 1,
v=x.
Hence - = Limv
-V-foo = 0.
Complex Inf eg rat ion 239
i.e. when z varies from 0 to along the real axis, we find that
the point vv remains in the first quadrant, moving from 1 to w
where the curve is parallel to u-axis. Hence there is no change
in the arg/(z) when z is moving along the real axis.
(ii) On the circle z = Re'e, R->°o , 0^0^ tt/2.
/ (r)=z3+iz + l
=z>(1+£+^)
are / (z) = arg (z3) + arg^l+^- + —3J
= 3.ir/2 + 0.
(iii) When z moves on the imaginary axis, we put z = iy.
f (iy) — -iy'-y-M=0,
u= 1 —y, v= -y*.
Again Lim> - = 30
u
Clearly when y > 1, the point (u, v) remains within third

quadrant. When 0 < y < 1, it lies
in the fourth quadrant. When y
varies from oo to 0, the curve in the
»v-plane passes from oo passing
through (i», - 1) to (1, 0).
v
Also Lim^-fO — =0.
u
Hence the change in argument
f(z) is(?-o)=4
Therefore the number of roots in the first quadrant is
iO+T+i)-'
(iv) Proceeding it the same manner when z moves ,n
the second quadrant from (0, iR) y
to (- R, 0), we see that the
change in the arguments along the
imaginary axis and the circle of the
second quadrant.
z = Re*•, tt/2 ^ 0 < n
tt ,3 n 1
are
—2 and 2 •
(v) Again on the negative real axis, put z=— x.

u= —x9.
v=
.*. Lim.v->co —=0.

u
When x changes from
co to 1, the point (w, v)
moves over to (0, —1)
starting from oo in the
third quadrant where the
curve is parallel to the u-axis and remains in the third quadrant.
When x varies from 1 to 0, the point moves from (0, l)and
remaining in the fourth quadrant moves on to (1, 0). Hence the
change in the argument is when z moves from —oo to 0 along
the real axis. Thus we get the number of roots in the second
quadrant
1 7T/ 3-rr\ ,
= 2^(-2 + .2+V = 1-
Similarly it can be shown that the third root lies in the
fourth quadrant.
Exercises
1. Show that the equation z4 + z-fl = 0 has one root in each
quadrant.
2. Prove that the equation z*+4 (1 + i) z-f 1 — 0 has one root
in each quadrant.
3. Show that the equation z,0-|-z2+1 =0 has two roots which
are purely imaginary and in addition two complex roots
in each quadrant. Show also that one of the complex roots
in the first quadrant lies in the sector n/10 < arg z < tt/8.
4. Show that the polynomial zB - 1 Oz3 + 5z3 - 3z - 1 has just
two zeros in the annulus 1 < | z | < 4£ and just three
zeros in the right half plane.
5. Prove that, if p is a positive integer, the polynomial
zp2-+-6z+ 10 = 0 has no real zeros. Find the number of zeros
which lie in each quadrant of the complex plane
(i) when p is even,
(ii) when p is odd.
Prove also that whatever be the value of the positive integer
p, all the zeros lie within the circle | z ( = 8.
6. Prove that in the ha f plane R (z) < 0 the function e*-z*T -2
has exactly 2r zeros, all simple ones if r is a positive integer.
7. Show that the equation e_I=r—(1 +/) has precisely one root
in first quadrant.
8. Show that the roots of the equation r5 -9r*+ 11 = 0 all lie
between the circles |z| = l, |z| = 3 and that two of them
lie to the right of the imaginary axis. [Applying Rouche’s
Theorem to the circle | z |= 1, take/(r) = 11, g (z) =z5 - 9z2,
and for the circle | z | = 3 take/ (z) = r5, g (z)— 11 - 9z J
5‘29. Analytic continuation. Suppose there exist two
functions fx (z), and/2 (z) such that they are regular in the domains
Di and D2 respectively and that Dx and £>., have a part in common
throughout which fx (z)=/2 (z).
Now we consider the aggre¬

gate of values of fx (z) and f2 (z)
at the points which are interior
to Dx or Z>2 as a single analytic
function ifi (z). Hence we see
that </. (z) is regular in the com¬
mon part A =z Dx-\- d.2 while
•P lz)=fi (z) in Dx and «/. (z)=/, (z) in D2,
Then we say that the function /2 (Z) extends the domain in
oftti W3S defined and We an !'na,>,ic con,‘nua<ion
method *of Tnalytic'conUnualio'n*'^ ZZlZ

which is described briefly as follows!- °f P°Wer series
Let there be any point P (z0) i„ ,he neighbourhood of which
the function /(z) is regular. Then by Taylor’s theorem,
expand the function in a series of ascend!™ can
powers of (z—z0) the coefficients of which involveUi'e ^ 'n,e*ral
derivatives of the function at the point z = 2 1 successive
within which the Taylor’s expansion is /(*’); then ,he cir^

valid is the circle of radius PS and having
P as its centre. 6 /
Now let us take any other point
I, within the circle of convergence not
on the line PS. Then we know the
value of the function at P, and all i,s
derivatives at P, can be found from the
series by the method of term by term differentiation. Hence we

can form the Taylor’s series for the same function / (z) with Px
as origin, which will define a function regular throughout some
circle having its centre at />,. This cricle will extend to region
as far as the singularity of the function defined by the new senes
which is nearest to P1 and which may or may not be 5. In any
case the new circle of convergence may lie partly outside the
old circle and for points in the domain which is included in
the new circle but not in the old circle the new series may be used
to define the values of the function, though the old series failed
to do so.
Tn a similar way we can take any other point P2 in the

domain for which the values of the function are known and form
the Taylor’s series having P, as origin. This in general will still
further extend the domain, where the values of the function were
not defined previously and so on.
Bv means of this method of continuation, starting from a

representation of a function by any one power series, we can
find any number of other power series which between them
define the value of the functional all points of a region, any
point of which can be reached from P without passing through
a singularity of the function. The aggregate of all the power
series we obtain, constitutes the analytical expression of the
function.
The continuation by two different paths PAM, PA'M will

give the same power series, if the function is regular inside and
on the closed curve PAMA'P.
Suppose we take any point P, on PAM, inside the circle C,

having its centre at P. Obtain the continuation of the function
with P, as origin and let it converge inside a circle C. Then
take any other point Pt’ inside both the circles and also inside the
curve PAMA'P.
Suppose that S,SitSx' be the power series with PtPltPx as the

origin. Then SX=Sf (because each

is equal to S) over a certain domain
which will contain Px if Px be
taken sufficiently near Px; and hence
Sx will be the continuation of Sx'.
If Tx were the continuation of S’,',
we have TX=S\ over a domain
containing Px and so on. By
carrying out such a process of conti¬
nuation a number of times, we de¬
form the path PAM into PA'M if no
singular point lies inside the path
PA' MAP.
5*31. There must he at least one singularity of the analytic
function on the circle of convergence C of the power series
§ a„ (z - z0)w.
o
(Punjab 1959 ; Delhi M. A. Final 1959)
Suppose there is no such singularity ; then by the method
of analytic continuation we can construct an analytic function
equal to/(z) within C and regular in another concentric circle f
larger than the previous one. The series expansion of this
function in Taylor's series in powers of (z-r0) would then
converge everywhere within T. This is impossible, as the series
would be the original series having C as its circle of convergence.
Let us take any point zx, say, lying inside C, and let T, be
the circle of convergence ot the power series
Z A„ (z — Zj)’*. . . .(1)
Let f2 be the circle with centre
zx touching the circle C internally.
Then the series defined by (1) would
definitely converge within Fz and will
have its sum=/(z). The radius of the
circle fx cannot be less than that of r2.
Now there arise three possibilities.
(i) rx has a larger radius than T2
in which case Px clearly lies outside C
and the analytic continuation is provided by the new series.
(ii) C may be the natural

boundary of f (z) outside C, and
the circle Tx touches Cx internally,
no matter what point z, within C
had been chosen.
(iii) rx may touch C internally, althuogh C is not a natural

boundary off (z) ; in this case the point of contact of rx and C
is a singularity of the analytic function which has been obtained
by the method of continuation of the original power series as
there is necessarily one singularity on rx and this cannot be
within C.
Now we illustrate the method by the following examples.

1. Show that, if b is real, the series
*log <i+^+'—■ *+!=£-* (££)+.-

is the analytic continuation of the function defined by the series
Z-\z*+\z*- ...
The series z — £z2f-£z3 —...
defines the function log (1-fz) in the region | z i= 1 and z= — 1
is the only singularity in the finite part of the z-plane.
Again
2 lo« <• +*2) + ‘- ta"-‘ b+l+%-\+ -
log (! + £*) + / tan'1 6-t-log^lf-^”-^
~2
• i
where < 1 i.e. I z — ib l ^ V(1 +b2)
a circle with centre ib, and radius V^+b2).

In this circle of covergence
the seiies of the function
log VO +b*i + i tan-1 b
-flog (If-z)- log (1 - ib)
= log (I f-Z)
which is the same as defined by 1st
series.
The second series in an
analytic continuation of the first
in a region | z—ib | = V(l + 62) which has a part in common

with the 1st region 121 =1 and also extends beyond it as shown
in the figure.
Mark that b should be real. If b is imaginary, say ci so that
ib = — c, it will be on the real axis and second region will not
extend outside | z | = 1 if c < 1.
2. The power series z-f iz2+|r3+ ...
and in-(z-2)+£ (z—2)x - ...
/lave no common region of convergence. Trove that they are
nevertheless analytic continuations of the same functions.
The series z-f- $z2-f |z3-f-...
defines the function —log (I—z) whose circle of convergence is
j z | = 1 and z= + 1 is the only singularity.
The second series in—(z- 2)+i (z ”2)* - ...
defines a function in—log [l + (z— 2)J
where | z-2 | < 1 i.e, a circle with centre z=2 and radius 1 and
thus touching the previous circle externally as shown in the figure.
The second series defines
the function
log (1 +z — 2)
—in—log (z—1)
=in-log ( —1) - log (1 -z)
— in — in — Jog (1 -z)
*=—log (1 -2).
Obviously they have no common region of convergence
though they are analytic continuations of the same function
-log (1 -z).
3. Show that the function f (z) = 1 -f z + z2 -f ... zu ... can
be continued analytically outside the circle of convergence.
The function defined by the series is (1— zrl provided
I z | < 1 which is the circle of
convergence. But this function is
regular in the domain which does
not contain its only singularity
z—1, and hence provides the
required analytic continuation.
Suppose we take any point a,
(where a < 1) inside the circle
of convergence. Then the series
oo (2' — o)n / Z —
S ^Y_-g^w+i defines a function ^1 — p—having Cj as its
circle of convergence given by
YZ~ < Le- (z~a) <1 1 - a |.

Now if 0 < a < 1 then C1 touches C0 at z=l which is
therefore a singularity of the complete analytic function defined
initially by the function (1 - z)_1. If however a is not real and
positive we have
| I - a j > 1 - | a |,
so that Cj crosses C0. In this case the new power series provides
an analytical continuation of /(z) outside C0. [See the figure
below].
Show that the function

1 z z2 z3
• •
can be continued analytically outside the circle of convergence.
The series is represented by the function /(z)=^-L_ only for

points within the circle of convergence | z | = | a |.
But there also exists another power series defined by
1 2 —b (z — b)2
S' / \ * V V
• « > ...(2;
... I z-bl-'
[_1 a —b_
provided the circle of convergence is Cj.
! z —Z> | = 1 a—b |.
Now if a and b both are real and positive such that
0 < b < a, the circle of convergence Cx of the second series
touches the first circle C internally at its only singularity z=a :

hence there is no analytic continuation in this case as in fig. (i).
However if b\a is not real and | / | < | a | and positive, the
series defined by (2) converges at points inside a circle which is
partly inside and partly outside the circle | z l = | a |. These series
represent the same function -~~z at points outside this circle and
hence can be continued analytically, as shown in fig. (ii).

Functions with Natural Boundary.
A closed curve is called a Natural Boundary of a function if the
function cannot be analytically continued to any point of the same.
5. Show that the circle of convergence of the power series
/(z; = / + z + 22+z1-M8+...
is a natural boundary.
The circle of convergence C0 of the power series is | z |= 1.

If the point P (ea>) on C0 is not a singularity of f (z), then f (rcai)
must tend to a finite limit as the real number r increases and tends
to unity.
Let us consider the behaviour of/ (r) as z moves up to C0

along the radius through the point of affix. etirpilZ'i. where />, q
are integers. Now we can express f (z) in the form
f(z)= 1 +24-£2-f... + r■-, + ...+/

r?=7+l
=fl (*)+/2 (*>.
Since/j [r exp. {2ni (p\2ry)}J is a polynomial in r of degree 2,/. it

tends to a unique limit as r-+1.
[It is, however, the case that iff (reai) tends to a finite limit,
then eai is not a singularity of/(r). For example, the binomial
248 Complex Variables
expansion of (1 converges when |z|< 1 and (I

tends to zero as r->l. Yet z= 1 is a singularity (a branch pt.)
of (i —z)v,a where p < q and p and q are relative primes.]
Q
But f2(re2I'n‘l2'‘) — £ (re2'*''“V

fu=a+i
_ F r2n p2iTi*{2ni2Q )P
= ^ r . c
= 27 Exp. (21+n-Q .p*i)

n=»2+l
= 2 r-n Y Exp. (2l+"-a. pvi)=I

n=^>l
which tends to infinity as r-> 1. Hence the point of affix.

e2p-nn2Q on co is a singularity of/(z) for each and every integral
value of p and q.
Thus any arc of C0, no matter how small its length may be,
contains a point whose affix is of the form where p
and q are integers. There are therefore points whose affix is of
this form within every circle which crosses C0, hence it is impossi¬
ble to continue/(z) analytically outside the circle of convergence,
i.e., the circle of convergence is a natural boundary whose every
point is a singularity of the function.
6. Show that the circle of convergence \ z \=1 is a natural
boundary for the function
OO ~ni
/(*)= n—0
*
Or
00 t along
Show that the series Z zn ! 30 as z
0 -
the radius through the point where q is any positive integer less than
m !. Deduce that the functions represented by the series can not
be continued analytically beyond the unit circle.
rrtoih; IQ5Q* Pnniab 1959)
Let f(z)=2z*1
0
Then/(z) is an analytic function, regular for |z|<
Complex Integration 2-19
| z |r=l is the circle of convergence.

[u zn *
7 Limn->» — = Limn->co
un+l z(«+D !
= Limn-»oo —> 1 if
zn • n
I2I < 1•
Hence convergent.
Let z=re2j,irihlt and let us consider the behaviour of f (z) as
r-> 1 through real values.
Now /(z) = Vzn ! + ^ zn ! =/i (z)+/2 (2), say.

n-0
Then fx (z) is a polynomial and tends to a finite limit as

r-*l. When n ^ q, q being a divisor of n ! and hence
. . I/ 2piri\n ! „, . (2pirin !\
z/i! exp =rn- since exp. ^1
[•/ M i/<7 > l, and q is the divisor of n, q can be chosen

such that /i l/q is an integral quantity.]
Hence /2 (z) = = as r->l.

n=î
i.e. f(z)-+°o and so z=e{iirj>ilQ) is a singularity. But points of
this kind are everywhere dense on the unit circle so that there
is no arc however small on which /(z) is regular. Hence it is
not possible to continue the function f (z) beyond the unit circle
and so the unit circle is a natural boundary of this function.
Exercises
The power series z-$z*+|z3 - ... may be continued by a
wider region by means of the series
.o0 1-* (i-z)2 • •
log 2 2 2.2* 3.23
The function defined by the series
1 + az+a2z2+...
. 1 (1 —a) z (1 -a)2 za
• •
and r^- d-z)2+ (i -z)3
are analytic continuations of each other.
3. If f (z)= Z zn !, show that

0
/(z) = z+/(z2)
and deduce from this that | z |= 1 is a natural boundary of
this function.
OO
4. Show that 27 zZn can be continued beyond the unit circle.

•
5. Iff(z) is regular in | z | < 1 and the only singularity of/(z)

on the cirumference |z|=l is a simple pole at z=z0,
prove that
where/W- jf a„z», | , | < 1.

f(z)=(l+z) (1-fz3) (1+z9) (1+z27)...
has z=l as a natural boundary. (Delhi B. Sc. 1959)
CHAPTER IV
CALCULUS OF RESIDUES
6*1. Residue at a simple pole. We have already seen that
if z=a is a simple pole of f (z), then the residue at z=a is given
by Lira (z - a) f (z).
Another form is obtained as follows
Let /(z) = ^-||j , where 0 (z) = (z-n) F (z), F fa)z£0.
0 (z)
Then Lim7_>a (z — a)
' 0 <*)
Let-_^fl
_ .
—Limz_*a
0
(z-a) [0 (a)-Hz-a) 0'(a)4-...]
0 («) + (z — <j) 0' (a) 4-...
(a)-f-(z - a) 0' (<7)4-...
0' (o)4-
.
’ Slnce ^
...
the form
n
a
0(a)
“0' (a)-
6*2. Residue at a pole of order greater than unity.
Let z—a be a pole of order m of /(z) and suppose
* /(2)—Jti£L
(z-a)”»*
We then have, by the definition of a pole,
>«-&-*»+,! ■
where 0 (z) is regular at z = <7.
-A (z)
or (T^r=0 (*>+<7^>+<*^+■• • * +(T^r-
0 (z)=0 (z) (z-fl)m+A/i (z -a)—14- A/2 (z-fl)'»-*4-... 4- A/,„.
Differentiating both the sides with respect to z, (m— 1) times,
we have
0(m-i) (2)=0(m~i) (z) (z-n)”*4-('w— 1) 0("'“''!, (z). w (z-a)"*-1
4-(w-l) (m-2).0("t_3) (m-1)

,0«"‘-3> (z)./w (m (z-a)"-2-f...
—1) (z—...
m I
4-0 (z). j-j-. (z - a) 4- Ml (m~\) \
0‘"*-1»(fir) = A/1(«i-l)»,
<£(m-1)
Hence A/.=-rf/ which is the required residue at z=a.
(m — I) !
' (a)
and if / (z) = ^Z]-a , the residue at z=a is etc.

(2 - o) 2 !
</> (z) , . (a)
Formula : Residue of .—at z=a is -r^r.
(z—a)m (m—1)!
</> (z)
Alternative proof. Let/(z)=^_^m. If z=u is a pole of
order m, then <f> (z) is regular at z=a.

Hence the residue at z=a
=coefficient of (z-a)-1 in/(z)
= „ „ (z - a)”1-1 in 4> (z)
„ (z - a}™-1 in |j/> (a) + (z - a) 4>' (a)
+ -+TOJT^,W+-]
^(m-1) (a)
(m — 1) ! *
Another method. Since residue at z=a is the coefficient
of .—-—. in the Laurent’s expansion of /(z), the residue is the
coefficient of l|r in the expansion of /(a-\-t) as a power series

where { / | is sufficiently small.
Ex. Find the residue of * at z = i.

(*+!)'
1 1
Let / (z) = (Z + l)3 where (z) =
(z-|)3-(2_/)3 ^ ^ (Z + i)3*
Here z=/ is a pole of order three and the residue is
21! *" «•
Now <t>" (-)=^^y5 so that F
1 12 3
• •
Residue =
2 ! * 32/ 161*
1 . 1
Alternative. Residue = coefficient of - in
= coeffi. of } in ((-2+12(-3
Calculus of Residues 253
-3
=coeffi. of 1 in-8^(l+2^)
= coeffi. of l„ in--L{|-3.£ + 6.Q )*+...}
_ 1_ _6_3_
“ Hi *4/2 16/*
6*3. Cauchy's Residue Theorem. If f (z) be regular,
a fini.e number of poles, iw/Ai/t a closed contour C and continuous
on the boundary C, then
f(z) dz = 2ni £ R,
\
where £ R is the sum of the residues off (z) at its poles within C.
(Agra 1962)
Let zlt z2t...zn be the n poles
within C. Let y,, y2 .. yn be the
circles with centres zlt z2. .zn and
each of radius f> which is so small
that all these circles lie entirely with¬
in C and do not overlap. Then / (2)
is analytic in the region between C
and these circles, so that we have by
Cauchy’s theorem (see § 5*10).
jc /,2,*=L ^z>d2+\yt /(*>*+... + /(*)*...,I)

Now if/(z) has a pole of order mx at z=zlt we have
wi fj
f (z) — <bx (z)+ Jz-z Y' where >s regular within and on y.
L /(*) dz= fyJO i 00 <fc + Jyi

f U-z,) fJ) "2
J _ V\2 «
Vi U-z.)
.
+ ... + [ bmi dz
Since <£, (z) is analytic within and on y1( therefore
fJ i ■l'l(z)dz=0.
7
And putting z— 21 = pe<*> vve get

f bmx , _ C2lt bmx. pie(* dd
J Vi (2 - zx)mx z~ J Q ~pmT.emx"
= [27T e-lm'-ViQ dd = 0, for

Pmi — l Jo
f bx j _[™ bx.piei0 dd
and = 27Tibx.
J VI (*-*!> Z Jo P*i§
Hence j* f{z) dz=2iriy.residue of /(z)atz=z1.
We then have from (I),
i /(z) dz=2ni x (sum of residues at z „ z8,.. ,zn)
=2*r/ 2: /?.
6*4. Residue at infinity. The definition of residue can be
extended so as to include the point at infinity. We define residue
at infinity as having given by the integral
St I c f iz) dz'
where C is a circle enclosing within it all other singularities of
f (z) and the integral along C is performed in a clockwise direction,
provided that this integral has a definite value.
Taking the integral along C in an anti-clokwise direction,
the residue at infinity
--Mef(z) dz-
An important observation : It can be easily seen tr.at the
residue at a finite point is zero if the fuuction is analytic at
that point but may not be so at infinity. For example,
consider the function ^-~a for residue at infinity. The only
singularity of the functions is a simple pole at z=o. The

function is analytic at infinity.
The residue at infinity =- dz
27r riei0 dO
tputting z—a = rei0
= 2tti J 0f r.e
where C is the circle , z—a j=r
= —^-..2 TTi = - 1.
2 tti
We have defined residue at infinity to be the value of the
integral 2~ | /(z) dz, in clockwise direction round a large

circle C which has in its interior all other singularities.
Putting z=^, we get
taken in a counter-clockwise direction round a small circle about

the origin.
Hence the residue of f (z) at infinity
= LimZ-.0 [ — -f^V-J=Lin._-_>x [-zf(z)],

provided that this limit has a definite value.
Another method. Let f (z) have a pole of order m at infinity;
then/(l/Z) has a pole of the same order at Z=0.
Expanding/(1/Z) by Laurent’s formula near Z=0, we get

/f 11 \
\ 5?
oo _™m h
b
Replacing Z by 1/z, the expansion of f (z) in the neighbour¬

hood of z — oo is given by
m oo
/(z)= Z buz"+ Z
a=l n-0
m f oo C n
=Z 6„z" t/z -f- Z “-2- dz,
n=i J C n n=0 J CZ
since the integral of the sum = sum of the integrals by the theorem
of uniform convergence.
/. / (z) dz=j^ since all other integrals vanish
2* P.ie^dQ . l9
=a'S 0 P.e>° • Pull"1g- = P-e
= 2nial.
4 •
Then the residue at infinity J^ f(z)dz=» — at.
Hence the residue of f (z) at infinity is the coefficient of 1/z

with its sign changed in the expansion of f (z) in the neighbourhood
of z=oo.
Iff (z) is analytic at z=*>, /.<?. if bx=b2... =bm=0, it still

has a residue at infinity which is zero only when ax= 0.
Cor. If an analytic function has sigularities at a finite

number of points, then the sum of the residues at these points
along with infinity is zero.
Let C be a circle enclosing within it all the singularities ex¬
cluding infinity.
Then, we have
2~. Jc f (z) £k=sum of the residues at all the finite
singular points
and - 2~. Jc /(*) = residue at infinity.
Adding these, we get the required result.

—3
Ex. Evaluate the residues of ,—. . -— at 1, 2, 3
[2—1) ( Z — 2) (2 — 2)
and infinity, and show that their sum is zero.
Here has simple poles at 1, 2 and 3.
The residue at (z= 1) is

-.3
Lirnî (z 1) /<*)-{r_2) {z_3)~2
Similarly residues at z — 2 and 3 are —8 and "22 respectively.

To find the residue at infinity, we expand/(z) in the neigh¬
bourhood of z=oo as follows :
= 1 -f
o-ro-r
higher powers of
Hence the residue at infinity= -6 (§ 6*3).

The sum of the residues at 1, 2, 3 and =>o .
= *-8+¥-6=0.
6*5. Evalution of real definite integrals. We have already

discussed the notion of complex integration and of residue at a
point. We now proceed to the evaluations of real definite integrals
with the help of contour integrations and the theorem of
residues by properly choosing the integrand and the contour. It
may be noted that the integrals which can be thus evaluated can
also be evaluated by other methods while there are many inte¬
J
grals which caDnot be evaluated by contour integration e. g.
OO
e~x' dx. Before proceeding to the evaluation of definite inte-

o
grals, we prove two useful theorems.
Theorem I. If Lim (z-a) f (z) = A, and if C is the arc

Z—+Q
< 0 < of the circle \ z—a |=r, then
Limrô Jc f(z) dz=iA (d2-dj.
For a given e, we can find 8 depending upon e, such that

(z-a)f(z) - A | < c, for | z—a | < 8.
Taking r less than 8, we have

| (z-a)/ (z)—A j < e on the arc C.
(z—a)f(z)=A + y), where | tj | < «
ci /(*) = z—a
Then f /(z) </z= f </z

JC JC z—a
f®. M + dd
= Je1-^-» putting z — a=re'°
~{fil-ex)iA+(et-01)
cr | jc /(z)f/z-M(^2-01)U(02-01)Mnl <
Hence Limr^0 /(z) (O2-0x).
In particular if z=o is a simple pole of f (z), A is the residue

°f /(z) at z=a. If C is a small circle | z—a \ = r, we have
(0i-0|)—2n and we get
|c /(*) dz=2niA.
Particularly if (z-a) f (z)-*-0 as z-*a,
the" |c /(*) dz-+U as z-»-o.
Theorem II. If C be an arc 6X ^ 6 < 02 of the circle

| z | = 7? and if Lim^^ zf (z) = A,
then
UmR ->oo JC (02-0i) ^
For sufficiently large values of R, we have

| zf (z)—A l < e
or zf (r) —where | -jq | < e.
A-\-t\
dz
\c /(2) *“j C z
fOs (A + rj) iRe'° do
=J0i --» putting z=Re-°
= Ai(02-ei) + r)i(di-dl)
or jc /(z> dz-jn^-e^ =1 71 I. («,-*,)
HtM*,-*,)
< <»*—»,) «-
Lim
In particular if z/(z)->0 as /?-><» , then
66.
i / (z) dz-+ 0 as R->x .
Integration round the unit circle. We consider the integ¬

rals of the type
2tt
i / (cos -in 0)
where the integrand is a rational function of sin 0 and cos 0. If

we write z=eie, then
i(2+-‘) = c°S0
and f/(z-i)“sin “• Tz=d<>-

We then have
J /(cos 0y sin 0) dd — ^c <f> (z) dz,
where </> (r) is a rational function of z and C is the unit circle

1-1=1.
Hence </» (~> dz = 2ni 27/?^,
where ZRC is the sum of the residues of <f> (z) at its poles inside C.
Ex. 1. Show that

Jn r dd 2n dd
a -f-b cos 0 -l 0 a + b sin 0
-
2*
, a > b > 0.
^(a2—b2)
Putting z=ei0, dd = j^y we get
=r- dd
Jo a + b cos 0 ='f —
i JC z [a + 4^ ( z~\~ i/2)]
dz
_2 f dz 2 f
/ J C bzi-\-2az-\-b~] J C '
where C is the unit circle ! z |—1. The poles of f (z) are given
by the roots of bzs-f 2az-f b=0
rtB. _ — 2a± — 4b*) -a—\/(a2-b2)
Z~ 2b = b ’
U - \/(n2 - b2) n -
Let a= - - , /? =-g-
-b
Since a > b > 0, | ft | > 1. But | a/3 |= 1 so that | a | < 1.

Hence z = a is the only simple pole inside C.
Residue (at z=a)=Lini2_>a (»-«>
_- . 2 2
-umw
2 I
”, 2y/ya*-b*rW\a*-F)
bi.-j-
r, i*2r
[2r. _d0_._1_2v_
Hence
enCC Jo
-—7 ...d)
Jo a + lA cos 0 b*) y/(a*—b'£V
Similarly j*2 - can be evaluated. This is left as an

Jo a + b sin 0
exercise for the students.
Ex. 2. Evaluate f2* — * , (a > b > 0).

Jo ("■+ b cos H)
Putting z = e‘®, we get
_1 f
Jo (ff-f-6 cos 0)*
fl)*—// J cC z [u-f b/2 (z 1- l/z)J
= -4,J
r f
JC
z dz
(6z*+2r;z + />)2
__4Jf z dz 4/ f
Jc [z2 + (2a/6j 2 -f 1 ]2— J C 7 U
where C is the circle lz 1=1.
Poles of order two of /(z) are given by the roots of
Z2+-r Z-f 1 = 0
or
”*[4V(:
—a±(a*-b2)
Let -tf + V'ta2-*’2) p_-a-V(aa-b2)
Since a > b > 0, | p | > 1. But a/?=l so that ( a | < I.

Hence a is the only pole of order two inside C.
°W b*f^~ b8 (z-a)2 (z-/3)2

1 4/z
(z), where (z)= -
(z-a)2 b* {z-W
Residue at the double pole z=a is
</>' (a)
_ _4/ r(z-p)2 — 2z (z-P) r
b2i (z-P)‘ Jz=
4/ Ha —/3) — 2a 1_4/ (a +
+ 0)
;“^L («-/?)8 J b («"
J~b* (a-/?)3
~ *
4/ - 2 alb ai
‘^•8/63 (a2-*2)3'2-(a2-£2)3'2,
/
ITT
dd . f ai 1 2
2rra
Hence -pfii-
o {a+b cos 0)a
Aliter. Differentiating (1) of Ex. 1 w.r. to a, we get the
required result.
2 IT
dd rra
Ex. 3. Prove that a— //,a_i\3/2 ^ >
0 (rt + COS <08 (tf3 —1)3/
■-*1
i
SW dd
We have /
o (a-j-cosfl) 0 (a-fcos ey~
1 dz
C [a+i (z+l/z)]**/z ’
where C is the unit circle | z | = 1.
41
C [z2 + 2az+l]2 dz=i \c f^Z) dZ'
Now f (z) has double poles for values of z given by
z2+2az+l=0 or z = ~.^<4a2 ~4)
= -a±V/(a2“l).
Let a= -a4-\/(«2- 1), and £= -a-\/(a2 - 1).
Since a > 1, we have | 0 | > I. But | a J | /3 | = 1 so that
a I < 1.
Hence a is the only pole at z=a within C.

Residue of /(z) (at double pole z=a) = </-' (a) where
i> U)=
(z—/3>2 - 2z (z—fi) '

Then <£' (z)=
(z-P)*
(z—P) - 2z
(Z-/?)3 *
a+0 2a a
/. («) = (a -py~b [a2 - 1)3/2—4 (a2 _])?/2-
2 a 7ra
Hence /=2tj7.— .
i • 4 J)3f2-(a/_1)3,2.
Clearly it is the particular case of Ex. 2 and ^ of it ; putting

b= 1, we get the result.
r 2TT d() 77
Ex. 4. Show that -=-7,= ^.
Jo 5 + 3 cos 0 2
Proceed exactly as in Ex. 1. by putting the values of a and b.
f* I 4- 2 CO 9 6
Ex. 5. Show that -- d& — 0.
J0 5 + 4 cos V
We have
, r 1 + 2 cos e ,n fIir (1+2 cos 6)
'"Jo 5 + 4c“oTfl a*-2 Jo (5 + 4 cos 9)d0
1 1 +2
-real part of ^ 5^4 cos 0 dd
Putting z—ei9t d0=^ and ^ (z+j^=COs 0, we get
r 1 . f 1 f 1 + 2z dz
/= real part of 2- c -7~T\ • 7z
5+2(z+r)
=real par( of 2| [c
=real part of jj j"c (2z+1) (z+2) dz
=real part of i-jc ^ .
Now the only simple pole of the integrand is at z = - 2 which

lies outside C and hence its integral is zero.
Equating real parts, we see that 7=0.

Ex. 6. Prove that f ^ n ta > q\

Jo a2+sm*
Jo 0
a2-{-sin2 0 VO +aa) »
V(l+flVv '
(Agra 1945)
Let /= r - aM V 2ade_
Jo tf^-fsin2 U Jo 2a2~Hl-cos 20)
(2ir * ^
= J0 2fl*+ J -cos0' putting 2d=4>
putting 20=^
= 7 jc z (2a2+l -t (z+l/z)] ’ PuttlnS z==£?'9 ,

where C is the unit circle | z |= 1
2o f dz
& / J [2z (2<32H- 1)—22 — 1]
“2a!Jc r~2-(2o=+i,zT7=2a/ |c f(2)dz-

Poles of/(r) are given by the roots of
z2 -2 (2al+ljr+l=0
or ...?,(2-+l)tVl^+l)‘-4].2fl.+ 1>±v,(^+4a0
= (2a2+l ±2aV(a2+l).
Let a = (2<i2-f 1) + 2a VO2-f-1), /?=(2a9 +1) - 2 a VO*+ 1).
Clearly | a | > 1 and since | a/3 |=J, we have | /3 | < 1. Hence
the only pole inside C is at z=/3.
Residue (at ;=/?)= Lim^
2a 2a
“ ~z(/3-«)-”i(_4a)V02+ 1)
1
2/V02+l)‘
. r * ^ ?jr, i
•• Jo a2+sin20 z ,*2/V(t?2+l)=Vfaa-f-D*
i/«2 0 2” , /. 2 .2v»
Ex. 7. /Var*> r/;a/ f” -ii
2*7^ = {« - A/(«a-**>}.
Jo «+*
where a > b (Agra 1936)
sin2 O
Let /= f —1‘1' P
Jo a + A cos 0
Putting z=f?<0, we get
,_1 f f(l/20.(z — 1/z)]2 dz , .
rr JC (z+l/J)- •'T' Wbere C ,S lhe c,rcle * Z ! =
(z2-\f dz
-if C z2 {2az-\-bz2+ b)
(.Z2~\y dz
—if
(zl4“*+i)
— 2ffifc r(Z)*'
Now/(z) has a double pole at z=0 and simple poles at r =
—a
and 2=/?, where a, £ are the roots of z2-f 4° z+ 1 = 0
b
or
_(a2—b2) n -a~V(o3-b2)
Let a t , p— 7
Since a > b > 0, | /3 | > I. But | a/31 = 1 so that | a | < 1.

Therefore the poles inside the circle are a double pole at the
origin and a simple pole at z = a.
Residue (at z = a) = Lim-_>a (z-aj.T -~ . — *——1

;L 2ib'zz (Z-a)(z-/3)J
(a2-!)2
lib. a2 (a—ft)
__1_ (a — 1/a)2
= lib* (a—/3)
1 («-/?)’ . 1 „
= -2f6f7^7’s,nce7=p
= 4 («-W —
=Tiy/^2—b3)
and residue at the origin
(z2 — 1 )z
coefficient of-! in- , where z is small
z
coefficient of 1 in'I -2z=+2‘) (l + -^z+z2Y‘
coefficient of’ in-jjgp (1 -2z«+...) (lz-... )

2 64 Complex Variable
:ir sin2 6 dd
■ fn a+b cos 0rw[^-«4]
f,r cas 20 J0 „ . 7TA2

Ex. 8. Show that
J0 l-2a cos6~+a^Tâ*'(a < ^
We have
cos 20 f/0 I f2* cos 20 dd
Kird 2a cos 0-j-a 2>'2j0 1- 2a cos 0 + a2
_1 f2,r
Now consider / dd.
2 Jo 1“ 2a cos 0 + a2
Putting r —e'9, dd=~ t we get
=i f __
1\ , o *IZ
2JC l-/z
zj + fl
Îf _£_
C (1 —az) ( 1 ~ z)
__ i r _£:i£_
2/ Jc (^ —oz) (z-a)
Now the integrand has simpe poles at 2=fl, I/fl of which

only z = a lies inside C since a2 < 1.
1 z3 a2
Residue at (z=a) =Limz->a (z - a).~. ({ ,C)|ZI7)-2<(1 -?)•
1 f27r
HoBce /=2j„ T= 2a cos 0-fa‘s
= 2 rri, —
" 2/(1 —a-) - 1 - a2
Equating real parts, we get
1 C2* cos 2d dd no2
2j0 1 -2a cos 0 + a8^ 1 —or
f17 cos 20 dd na:
Jn 1 -2a cos d + a%i~ I - a-*
Ex. 9.
Apply calculus of residues to prose that
* ccs 2d dd _tt
(Agra 1959)
•/ u, 5+4 cos 0 6
v cos 20 dd
We have I =f
5 4^4 cos 0*
,2 iO
_~_JQ
= real part of f
J 0 (5 + 4 cos 0)
2 TT Z2 dz
■-( o
Let I
(5 + 4 cos b) -I c [5 + 2 (z + 1/2)]’ <2 ’
putting z = ei0, where C is the unit circle, | z |= 1
z2 c/z If z2 dz
-u
C 2z»+5z+2~/ -i JC (2z+1) < z + 2)
z2 c/z
f/z If f(\>
-if c (z+!) (5+2) = 2/jc /(Z)
Now /(z) has simple poles at z=and z= - 2 of which
only z— — \ lies inside C.
Residue at z = -i is Limz__1/2 (r+*> 2717+1772+2)
J
12i*
•f 2tt £ i9
/- . ^
(5 + 4 cos 6)
1 7T
-rt dd = 2iri. — = 2*
12i
Now equating real parts on both sides, we get I=tt/6.
6
cos2 30 de 1 -/>+p2
Ex. 10. /Vove that
r- 2p cos 20 +
— 77
1 -/> *
(0 2
f2ff 1 +e Cflt
= real part of ,c/0.
*f. n 2/7 cos 20+/7
Now put z = ei9 and consider
1 I+Z‘ dz 1 f 1 + z'
dz
2 for. p (Z2+ 1/Z:~*)+/,2-/z“2/J C Z (1 ~/7Z“) (| -/7/Z2)
Z (1-f zfl) c/z
-if ( 1 +P2*7 (z2-p) = 2?fc /(z) '/z'
where C is the circle |z| = l. Now f(z) has simple poles at
2= ± 1 l\/p, ± y/p of which z= + y/p lie within the circle | z |= I,
since 0 Vp (z—y/p)f(z)+Limr^__v/p (z+ Vl»f(z)
r im 2 (l + z°) . .. z (I +z°)
(|_^J (Z+V/>)+Ll'/'(1-/B')(Z- Vp)
y/p (l+/>3) , -VP(1+P3) l+p:

(1 P2) (2VP)+(1 -P2) ( ~2V'^) 1 -/>2*
3
1 f27r 1
Hence
2 Jo 1 —2p cos20+/>2^ 2wi'2T1-P*
1 ~P
Equating real parts on both sides, we get
i-P+P2
/== —:-
1~P
Ex. II. Prove that
i
2* (1 + 2 cos 0)n cos nO .. 2w
3 + 2 cos 0 </0=;Z_5(3-V5)",
0
(w £e/wg o+ve integer).

We have1
f t•
(1+2 cos 0)n cos nO

/ d6
0 (3 + 2 cos 0)
**(1+2 cos 0)" e"0'
=real part of J </0
o (3+2 cos 0)
Putting z=efe, we get
2tr
(1+2 cos 0)" e< dd (l+Z+l/z)n.Zn dz
f
o (3 + 2 cos 0)
where C is the circle | z |= 1
-i C (3+z+l/z) * iz
_Lf (1 +z+z'rdz_i r ji+*+*g.1 f f(2)dz

“i Jc (1 + 3z+z2) -| Jc (r-aHz-P)^- / Jc H ’ '
where a, (3 are the roots of the equation z2+3z+1=0,
„ , -3 + V5 , -3-V5
whence a/+=l, a=-2-> P=-2-•
Clearly | 0 | > | a | so that | a | < 1.

The only pole of/(z) within C is at z=a.
Residue of/(z) at z=a is Limz_>a (z—a)/(z)
(1+z + z2)" (1 +a + a2)n

= Limz->a
(Z~P) (a—iff)
f, 3 ,V5 7-3VSV
V 2~*~ 2 + 2 y (3-V5)n
V5 ~ V5
Hence f” 0±2 cos el. e»'< d0=2ni l- <*=*9L
Jo (3 + 2 cos 0) “ */• y/5
=£5 <3-'/5)n-
r. »(l+2co.8).co.»«^_2!r
3 + 2 cos 0
2ir
V5
sin nO dd
12. Evaluate (a2 < 1 and n being a + ive
t + 2a cos 0+a2
integer).
sin n 0
, , dd
/-f 1 + 2a cos 0 + ai
C-* er,0i
= Imaginary part of ,—-- do.
J0 1 + 2a cos tf+u
Putting z = ei0, we get
f27r >n$l -n dz
o 1 + 2a cos 0+ C l+o (z+l/z) + o2 iz
n—1
-if C (1+02) (l + O./z)
dz
Zn
_i i - dz
or JC (z + 1/0) (z-f o)
= 4f / (z) </z.
or Jc
The poles of/(z) are at z = -1/a of which z= - o lies inside
C since o2 < 1.
Residuce of/(z) at z=-o is Lini _ (z+o)/(z)

z->—a
= Lim
z->-a (Z+|/a)
(-O)"
( o + l/o)
( - I)n.on+l
(I— a*) •
__1 (-l)".o”+»
Hence r-
Jo 1 + 2o cos 0 + o2 "S/ * (1—o2)
2n ( - 1 )”o"
- (l-o2) *
Equating imaginary parts on both sides, we get
/==J2,r sin nO
dO=0.
o l + 2ocos0+o2
Also equating real parts, we get
fzir
o
COS Z70
-— /It)- -__•
1 + 2o cos 0 + a*
( - 1 )non
(l-o2)
13. Prove that

‘lit
—cos 0 cos (nO+sin 0) <70=(—l)n
j; *• •
(n, being a positive integer).
Consider /= TV™*0 .e-("°+siD ‘dO

J0
_ r e— (cos 6+/ sin 6) e-rOi fiQ
f
J 0
21T
e-n9i
Putting z=e*9, we get

n &
-\c e ,Z~ ‘lz »
where C is the unit circle \ z | 1.
-4fc^*-Tfc/(r)A’
Now f (z) has a pole of order (w+1) at z=0.
1 (rf" _ \
Residue of/(z) at the origin=—, ja _ (
(-Dn
n
Hlence 7=2*7. 5-(-^y- = ( “ 1)n*« f*
j£ir f»-cos e cos (wfl-fsin 0) d0=(-\)n
and f2T e-cos0 sin (,,0+sin 0) d0=O.

J0
(air
£ cos 0 cos (n9-iin 0)
2w
<70=—..
ft •
(Agra 1952, 61)
Consider I=>f2ir ecos 0 .e—siD e>f dO

rnr
=I ^cos 0+i sin 0 'g—nWdQ
Jo
rzir pt®
= e . e~nQt dO * putting
Jo
where C is the circle | 2 |= 1
ez
dz /(z) dz.
=!f C zn+l -fj
i cJ
Clearly/(z) has a pole of order (/*+!) at the origin.

The residue of/ (z) at the origin
-If*
n ! Ldzn' Jj=0
= J_
n !*
Hence I = 2iri. J-. =

/ «! a !
Equating
”1 real and imaginary parts,
^coso cos (/10-sin e)dd=-nt
L
and f
J0
ecos 0 sin (nd - sin 0) dd=0.
15. Show that when R (a) > 0,
f tan (0+ai) d0=ni.

Jo
We have, tan (0 + q/)s=sin (d+ai)
* cos (0+a/)
_2 sin (6+ai) cos (0—a/)
2 cos (0 + a/) cos (0-a/)
_sin 20 + sin 2ai
cos 20+cos 2ai
_sin 20 + / sinh 2a
cos 20 +cosh la '
/—J tan (0 + az) d0= f sin 20+/ sinh 2a ,a

—T~ZZ- iW
0 cos 20 +cosh 2a
l1f sin </> + / sinh 2a
j 0 cos ‘/»+cosh 2a l>’
(J2i) (z - l/z) + / sinh 2a dz
putting 20 = •/>
-*i £ (z + l/z)+cosh 2a *iz’

putting z = e,?
where C is the unit circle | z | = 1

z2— 1 - 2z sinh 2a
—j C z (z2+ l + 2z cosh 2a)
z2 — 1 - 2z sinh 2a
</z
dz
—*f C z (z —a) [z- ft)
= lc /(*)*.
i.
Now /(z) bas simple poles at z=0, a, (3 where a, (3 are the

roots of the equation
z2-f2z cosh 2a +1 = 0
— 2 cosh 2a 4- \/(4 cosh2 2a -4) , - , . , -
or z=-~ ---= —cosh 2a ± sinh 2a.
Let a = — cosh 2a-f sinh 2a, (3= - cosh 2a—sinh 2a.

Obviously | [3 J > | a | and | a/31 = 1 so that I a | < 1.
Hence the poles within C are at z=0 and z=a.
Residue of/ (z) at the origin is Limô zf (*)**» a°d residue
at z=a is Limz_>a (z-a)/(z)

T . , z2—2z sinh 2a—1
=Limz_>a - 2.-Y(^)
! a2 -2a sinh 2a- I
“~2‘ M^/3)
a8-2a sinh 2a - 1'
a (a. —(3) J
_> r
- a/9-f 2a sinh 2a-f J~]
-2 [ a(a-/J) J
_,r
— 1-f 2a sinh 2a-f 1“|
CQ.
8
-,L
1
since a/3=l
sinh 2a_ sinh 2a _ A
— (a—/3) 2 sinh 2a" 2
Hence I=2î.l = T,i-
16. SV/aM’ ///at, // m he real and -7 < a < 7,
f
27r vn cos 0 r
0 {«* (m Sh' B)
in 0 +
— a sin (m sin d6 = 2tt cos ma.
■iv em cos 0
-q | sin (m sin 9)
^ f0 7 -\-a2-2asin
-fa cos (m sin 94 9)^d9= 2v sin ma.
(i) Consider the integrals

rzv .m cos 0 . . „
t I _£._pmi sin 0 dd
1 Jo 1 -+ a8 - 2a sin 9
2V
dd
-f o 1 ■+ a* - 2a sin 0
27T ,m cos 0
and .e (m sin 0+0) /
-i 0 \-\-az -2a bin 0
27T
eOT‘,’e.e0/
-i 0 1 + dL -2a sin 6
Putting z—ei9 in Ix, we get
j =f _en,z dz ( dz
1 Jc 1 +a‘-+-<7/' (z - \lz)‘iz “Jc ( 1 +<7/z) (l -ai/z)'iz
-if -r- dz= - - f

C (z — i/a) (z —ai)
-o/)
/, (z) </z
oJc 1
Now fx(z) has simple poles at z=i/a and z = ai of which
z—ai lies inside the unit circle C.
Residue of/t (z) (at z=ni) is Umz_âi {z-ai) /, (z)
(>mz gttiai
= Limz_*0,'
, - 1\ emai 2iremai
1 W \ a/* (a/—i/a)~[l-az)‘ •••(!)
Similarly, putting z=e,e in /2, we get
|-
rem: 1 f
-v dz — — I f (z) dz
« J1C (z-i/a) (z
,z-at) a Jc J K
The only pole of/2 (z) inside C is a simple pole at z=ai.
Residue of/2 (z) (at z=aO = Lim._*a, (z-ai)f2 (z)
die rnai
(ai - i/a)'
. / _ o •( 1\ aie,nai 2naie"iai
.(2)
Equaling real parts in (1), we get
•I* jn cos 0 2tt
I.. 1+-la
1+sin sin Becos
C0S <"■sin ») <W=0
sin *> COi ",J-
"IJ- • • -(2)
And equating imaginary parts in (2), we get
J„ i+u«-2flsta« sin (m sin e + 9) M= 1 -«• cos ",a- ■ -(4)

Multiplying (4) by -a and adding to (3), we get
f2w gOI cos 0
, , ~—r—-a {cos (m sin 0) - a sin (/» sin■ 0 + t))\ dd
Jo i ~ta — za sin u
— 2n cos ma.
Part (ii) can also be solved in a similar way. This is left as
can exerise for the students.
f2 _l_ J
z2 log -—J dz, taken round the circle
4rri
z |=2, to the value
3 *
1
Putting z—^, the integral is transformed to
'-ft'
3 •» °«f-
i-1
c- f*Io®
where |5|=f is the new circle C' over which we have to
evaluate.
Clearly the integrand has 5=0 the only pole of order four
within C\
Now residue at 5=0 is the coeff. of l in/(5) where
/«)=-£ log J±|
= -|. [log (1+5)-log (1—5)]
—?[25+t+-]
• • coeff. of | in /(5)= —
Hence f=—2m' (—f)

4wi
= 3 *
Note. Minus sign has been taken in the above because the
transformation r=l/5 gives isogonal transformation e. the
sign is reversed from anti-clockwise to clockwise.
Exercises
dd 2n__
Show that (Agra 33)
2-f cos 0—’y/3’
COS2 36
dd 3-rr
Prove that
5-4 cos 2d=~8 •
Show that, if | a \ < 1,

f27r dd_2tt
Jo 1 -j-a* - 2a cos 0“ 1 -a*'
and deduce the value of the integral when a > 1.
Evaluate
fr
J_7r a + cos d
dd, (a > 1)J*. (a > 1)
by putting z=e‘® and using the Theory of residues.
Ans.
Ads. 2«,[l---/--I
2a tt
Prove that
dd tt (2a-f b)
r -
Jo (a + 6 cos* + (a > ® > U)t
Show that
f°° _ dx tt Q- + ab-bb-
J-oo ix*+firrV2 d'b3 (a3-+ a*b+ab* + b*)
(a > 0, b > 0, ajhb).
6*7. Evaluation of the integral
J
Q (x) ax’
-CO
where (i) P (x) and Q (x) are polynomials,

(il) the equation Q (z)=0 has no real roots,
and (M) ‘he degree of P (x) is at least two less than that ofQ (x).
Conditions imposed upon P (x) and Q (*) ensure the conver¬
gence of the integral Jdx. Such integrals can be
evaluated dy integrating ^ (z) = ^-pj round a contour consisting
°f a semi circle r of radius R large enough to include all the

poles of F(z) in the upper half plane and the part of the real axis
from x= — R to x—R provided
that z F (z)-+ 0 uniformly, as
| z |-*.ao , for 0 ^ arg z ^ tt.
Condition (ii) shows that the fun¬
ction F (z) has no poles on the
real axis. Here we assume that
the only singularities of F (z) in
the upper half plane aie poles.
By the residue theorem, we have
ffjj F (x) dx+ fr F (z) dz=2ni ZR+, .. (I)
where £R+ denotes the sum of the residues of F (z) at its poles
in the upper half-plane.
Now since zF (z)->-0 as | z }-»*/, and hence from theorem II,
of § 6-5, Jj, F (z) dz=0.
Also Limjjôo F (x) dx=P J ^ F (x) dx,
where P denotes the principal value of the integral.

But since this integral is convegent, we can write
oo
i F (x) dx
=11f«
dx
Hence proceeding to the limits when R-*oo, we have

from (1),
1. Prove that
’oo dx rr
(» f (1 -h**)2-4*
f°° dx 3tt (Agra 59)
(I,) J -—«oo (*2+D: *
dz
(i) We consider f dz where C is the
C(l+z2
contour consisting of a large semi-circle T of radius R together
with the part of the real axis from *= -/? to R.
(See fig. of § 6’7)
We have by residue theorem,
« *
Jc /<*> *-J', fl&ji+f r - (I)

Now Lim.^^ zf (r) = Lim__>0O =0.
(1+r2)2
Hence Lim/jâo j*p ^ j =^
dx
and H(T^?=I_ oo (i +x*r
Then we have from (1),
dx
^=2tt» 27/?+
r. co (1 + **)»
Now f(z) has double poles at z= ± /, of which z=i lies inside C.
Residue (at z=/)=</>' (/') where <f> (z)= *

(z+/)2*
• • Residue at z=i is
2
(*+03-U
°° f
Hence we have from (1),
dx
-oc(H^j'2)0
f °o dx
(l-fJC*)2
^ . 1 *
2
00 dx
or
1° (H-^2)2'4’
dz
(ii) Consider the integral J C (z2+l)3 -f cf(z)dz- where C
is the same contour as in (1). We then have
Jc /<*) dz=JR_R /(*) /(*> <te- */rA+. ... (1)

2
Since Lim^^ z/(r) = 0, hence we have
Proceeding to the limits as R->oo, we get from (1),
/(*) dx=2m ZR+. ...(2)

Now/(z) has poles of order three at z= + /, of which z=i
ies in the upper half-plane.
Residue (at z=/) is 4>" (/), where <l> (^) = r^ra
-GvbSM..
6 3
“(2/)®“i6r
••• From (2), /(*) </x=2rr/.i.=^.
OO .x2— x-f2 , 57t

Ex. 2. Prove that
I—oo ^4 _l 7(?JC2-f
X4-f ;/t^2 HToP “x ~~ 72 *
(Delhi M. Sc. Final 1959, Alld. 1942)
We conside
u z‘~z+2
+ 102*4-9
contour discussed in § 6*7.
dz=jc /(2) dz. where C is the
We have by residue theorem,
jc f (*) dz=\R_R f(x)dx+jr f(z)dz=2ni ZR+ ...(1)

But j'^f(-) dz-*-0 as /?->», since Liraô z/(z)=0 or we
may proceed independently as follows :

On putting z=Rei0, where z lies on the semi-circle, we obtain.
-z + 2 Z2-Z-1-2
dz dz
ri
+ 10 Zz-r9 -fr (z2+9) (z2+l)
7r /? V*® - /?e‘® + 2
. /to" dd.
-I0 (R2e2i* + 9) (/?*e2<® + l)
f ; p /?zf
R2e2l9-Rei9 + 2
Jr/(z'l dz ,=j jo ^2^10 . /to'® </0
+ 9) (/?V*®+1)
1 /W" |+ 1 — /?e‘® 1 + 2
,2t0 /?V® I ^
o { | R2e2i0 | -1 -9 | }{ i /to2*®
[V j a + /> | < | a |+| b | and | a-/> | > | a , -| b | J,
f* /?22-f /?+2
/.e. .Rdd,
. o (*2- 9) (/?a-l)
^ 7T (R* + R- + 2R) n 0
/.e.
< <*■-!) <3rep° as *-*“ •
Hence we have from (1) on proceding to the limits as R-*■ oo,
f (a) dx=*2iri £R+.

/: OO
Now /(r) has simple poles at z = ± /, ±3/ of which z=/ and
z=3/ lie inside C.
Sum of residues at these poles
= Limzî (z—i)f (z) + Limz^3/ (z—3i)f(z)
/*-/+2_ 9/*—3/+2
(i2+9j (20 (9/2 + 1) 6/
_l-i 7 + 3/ 10 4
— i r : “4" 48/ ~48/~24/*
16/
|_oo ^V-2,,24/-~12*
a:2 dx 7r
Ex. 3. Show that
f: f*2+/; (x2-4)-y
Taking the contour of Ex. 2 and proceeding in the same way
we get
A3 UZ
fR —
J-/? (.v3 4-D (A2 + 4; + Jr (ii+Trfi+4,=w^+- •••(»
2
Since Lim-_^ Z U2+1) (z2 + 4)"'0’
• f _g!_fc— -Q
•* Jr (z2+l) (z*+4)_u-
Now the integrand has simple poles at z= ±i, ±2i of which

z=i, 2/ lie inside C.
Sum of the residues at these poles
= Limz_n (z - i)f(z) + Lim2_*2/ (2 -2/) / (2)
,•2
4/
4
(20(<*+4)^(4/*+0 (4iJ
I 41 /
= 6 +- 12 - ~ 6 ‘
Proceeding to the limits as R-+ 00, we have from (1),

a:2 dx
1 :00 (**+l)(x “+4>=’2’T'(~6)-“3‘
Ex. 4. Prove that
" X2 ^ Tr
« Jo
f oo ^.2 ^
**
-oo7xz + a-)*=8a* * provided that R (a) is +ive.
What is the value of this integral when R (a) is negative ?
Consider the integral j" > *=[c /<*) *. where C is

C Ua+1)
the contour of § 6*7.
We have by the residue theorem,
\c /(*)=}** /(*) dx+\r /(*) dz=2mZRK
Since Um^M 2/(z) = Limz_>0O ^~p=0, we have
Li*nj?->oo /(z) dz<= 0 [see theorem II of § 6*5J.
Hence proceeding to the limits as R-+00, we get
f(x) dx=2niER+ ...(0
Now /(z) has poles at z— ±1 of order two of which only z=»/

lies inside C.
2
Residue (at z=i)=<// (/), where (2):
(z+i)*'
Residu c=3[2l(L+i)'-2(z+i)S
L
_2i (—4) — 2 (2/) (- 1)
(*+o4 L
/
16 "" 6#
.*. We have from (1),
f(x) dx=
r /wa=3-
(ii) Consider J I / (z) dz, where C is the same

C (z: a2)3 JcJ
contour as in (1). By residue theorem,
j*c / (z) dz=-^R r f (x) dx+/(2) dz=2iri2R+. .. .(1)
Since Lim^^ z/(z) = 0, we have Lim^^ /(z)</z=0.
Proceeding to the limits, when R-+co , we get from (1),
J” fix) dx^ Itt'iZ R+. ... (2)
Now/ (z) has poles at z== + ai of order three of which z=a/

lies inside C provided 7? fa) is positive.
Residue of /(z) (at z=a/)=coeff of j in / (t + ai)
„ „ „ (/+«/)■
(*4-<w
Now (7^S-3 = -8^ (/»+2«/*-^(l+^)
6/_f
= ~'8^7?3 f,I+2a,/ ■ “2•ai~4a*
4a2 • • •J •
o «• r 1 • /->\ 1 6ai . 6fl21
Coeff- of f,n (2)= "IteSiL1 ~2S+4?J
16a3/'
Residue of/(z) (at z = a/) =

.. —-' 7 16a3/'
•• (2).
from (2), j fix) dx=
f(x) (lx=2î
2 ni j±-3.=£-3. ..•(3)
When R (a) is - ve, pole within C is at z— - ai.
in this case fix) dx— -
— oo 8a3
[replacing -a by a in (3‘J.
r=° dx
Ex. 5. Evaluate , (a~> 0). (Agra 1946, Punjab 58)
J —oo X y Q
Consider fcfWc f{z)dz’
where C is the contour of § 6 7.
|c f(z) dz=z\R_R /(*) dx+\r dz= 2ni Z 7?+
Sinoe Lim^^ zf(z) =0, we have f(z) dz=i).
Limêe f(x) dx = 2-rri Z /?+
f(x) dx=2ni Z R+, (1)

Now/ (z) has simple poles for values of z given by
z4-f-a4=0 or zi = ai.elZn+1) ni, n=0, 1,2, 3,
i. e. the poles are at z=aeni'*t ae3irili, atbnili, ae™1,4 of which
only first two lie in the upper half plane.
Let a denote any one of these poles ; then a4 = - a*.
Residue (at z=a)=Limr_>a(z-a)/(z)
=Lim[form 0/0J
— rim 1 1 a a
r">a 4z3 “4a3 “ <a4 " “ 4a4*
sum of the residues = — ^ (aenil* + ae3irili)
_J_ (pHU _,s-W</4\

- 4aa ^ e >
1 n. . ” y/2 i
«= —-r-^.2/ sin —~t-, •
4«3 • 4 4a3
■v/2w
II ^-K-£)- lo3
[°° dx y/2 7T
J-» x*+a*- '2a* *
tt 2n !
6. S/iow that
Up*+(2555+1 •(»!/»
f where n is a possitive integer).
Consider
jc (z*+A*)w+I“Jc û) </z’
where C is the contour consisting

of a large semi-circle T of radius
R in the upper half of the plane
and the part of the real axis from
x= - R to x = R.
By residue theorm, we have
\c f (z) dz=\R_R /(*> dx+\r /(*> dz=2m L /?+.
Since Limz^M zf (z) = 0, f (z) dz-+ 0 as R-+ oo,
Lim*-** /(*) dx=2ni 2 R+
or Pj’ f(x) (Jx=2niZ R+.
Now f(z) has poles of order («-f-1) at z= ±o/ of which z=-bi

lies in the upper half of the plane.
Residue (at z=bi)=^~ , where <f> (z)={z+\jyfii-
1 Cdn 1
residue = —L
n ! Ldzn * J==»<
_rj (-/»- n (-»-2)...(-k-1 -/;+!)]
L« r (z+6/)an+i j*=ib
_(- I)» (/i+l> (n + 2)...(2rr)

!.(2W"+»
1.2.3.. ./i (/»+- 1) (n +-2).. .(2«)
(n 0.1.2.3 w (2b)tn+l.i
[V i2n^=(- I)"./']
1 (2/i) !
J * (// !)2 (26)2n+i#
(2nJ !
J — OO
CO
f(x) dx=2iri. i
rt
r • (// !)2 (2&)2"+‘
(2n) !
or J ( / (* )<*r=
(2b)in+l ‘ (// !)2 *
f°° a- 2 dx
7. Evaluate
Jo ~x*+r
c* dz
Consider I ■j = /(z) dzt where C is the contour of
J C z6 +
example 6 above.
Calculus of Residues 28]
Then we have by residue theorem,
jc fix) dx+\r fiz) dz=2rri £ R+ .. .(1)
Since Lim^^ z/(z) = 0, we have Lim^ ^ /(z) <fc = 0
or we may proceed independently as follows :

Putting z=Rei9> we have
* /?a.e2'0
\r /W*r|f o P6.e6*‘-fl
. Rie10 de
r
^ r R3
*
J o R6 |
R3 I /e3,e | dd
de __
|- ! - 1
R3n
^Jo R6-l R°-l~*° as R^°°-
Hence Lim^^ |r /(z) dz=0.
proceeding to the limits as /?-►», we have from (1),
f / (x) dx—2-ni I R+. ...(2)

J —CO
Now/(z) has simple poles for values of r given by z°-f-1 =0
or z®=<?<*"+n *<t where *=0, 1, 2, 3, 4, 5.
Therefore the poles are at z=e,wo/6, ein,)l-t e,7,T,,/0.
e<nTM)i2( 0f which first three lie in the upper half plane.
Let a denote any one of these poles.
Residue (at z=a) = f .. . Z . —r-~1 (see § 6-1

l_t//i/z (z6 4-l)Jz=
a2 a3 a
6 a5 6a° 6
.*. sum of the residues at these poles
--i [/-/+/]=-//6.
We then have from (2),
Ex. 8.
r, Prove (hat
°r
r_2*u—f
7T
_j„ssee 40.
Jo /4-2x2 cos 0-f X* Jc
Consider the integral
\C 1 4-2zy cos U-j-z* \cf[Z)dZ

where C is the contour of 6‘7.
Jc /(2) d2=fR_R /(*) dx+jr /(2) dz=2ni SR. .. .(1)

Since Lirn^ zf (z) = 0, we have
frf(Z) dZ=°-
From (1),
/W dx=2mSRt
70
f(x) dx=2ni ER+. ...(2)
I —OO
Now/(z) has simple poles given by the roots of
z4-{-2z2 cos d+1=0,
i.e. z4 + z2 1=0
or (z* + eiB) (z2+e-i9)=0.
either z2=—
i.e. z=±iei9,i.
or z2= — e~i9 i.e. z— + ie~i912
Hence the poles are by z=iei9,tt iei9,t -iei912, — ie~i9n of
which first two lie within C. Denote these poles by ax and a2.
Residue at (z=ieiel2).
=Lin.^te(ie/2) (z - /«"») (7+;>./,}(zJU) (z'+e-J

2 1 1
2iei9n (- ei0-\-e~’e)~~iei9li ( - 2/ sin 0)-2ei9" 2 sin t>
Similarly residue (at z=ie~i9lt),
1
- 2e-«/2 sin 6 *
Sum of residues=^-4— I" e-i9,*-ei9,% 1

2 sin 0 L J
= ^Ttf[- ,sin ]
2 2 5
/ 0
= -T s,nr
Hence / (a) dx=2m. ( - y sec
=■* sec ^
e
or 2^ rr 0
Jo 1+2F cos 0 + z4~ 2sec2*
Ex. 9. Prove that

80 dx n (b+2c)
(b > O.c > 0).
j
-oo (*2 + 62) (x2+c2>2-26c3 (6 + c)3’
lc (za+/>2) (za+cV = lc /(2) *’

where C is the contour of § 6’7.
Jc / (z) dz = \R_R f M dx+lff(z) dz=2ni ZR+. ...(I)
Since Lizf(z) = 0, we have UmR_+x^rf(z)dz=0.
Lim R~> oo \*_Rf (x)dx=2*i ZR+
or f
J —OO
f (x) dx = 2*i £R+. ...(2)
Now / (z) has simple poles at z= ±bi and poles at z= ±ci
of order two of which z — bi and z—ci lie in the upper half
plane.
Residue (at z=bi) = L\mz_+bi (z-bi) f (z)
. . 1 1
1 z~*bi {z+bi) {zt + c*pm*2bi (c2-62)4
and residue (at z = ci) = <t>‘ (<?/),
where
* (Z,=(z! + 6s)(z + ciy
Now log <f> (z)= - log (z2+62)-2 log (z-f ci).
<!>' <z) 2z 2
z£+bx ~ z-\-ci
or
(C'-' = * <"» [ - b* - c2 - 2c/]
__i_c __?£L . n
(62-c2) (—4( 2j l_ 6*-c*+ c J
i r2c*/-6*/+c8n
4c2 (/,**-<*) L c (b'l-c*j J
(3c2 -b2) i
4c3 (62 -c2)2*
/ (3c2 -b2)i
sum of the residues = —
2b (?Zp)*+4c« (b*-c2)2
4W^v^3-3^+A3J
i(c-b)* (b+2c
“ 4bc3 (c2-b*)2
factorizing the numerator
_ i (6+2c)
4bc3 (c-f 6)2*

oo
f( x\ dx=2iri f _ _77 (b+2c)
i f(x)dx 2 ^ ^ {c+byJ -2bc* (6+c)2’
r°° xJ
1 dx 3V2 *
Ex. 10. Prove that (a > 0).
Jo <«* +*4)2~ 16a
*r6
Gonsider the integral dz=jc f(z)dz* where C is
l C (a4+z4)2
the same contour as in example 9.
By residue theorem, we get
Jc f(z) dz= /(x) dx+/(z) dz=2iri 27 /?+. .. .(1)
Since Lim,^^ z/(z)=0, we ihave Lim^_^ J*^, /(z) dz=0.
Therefore we get from (1),
LimR_+<x> ^ R f (x) dx=2irl 27 /?+
or f / (x) dx—2m 27 /?+. ...(2)

J OO
Now/(r) has poles of order two for values of z given by

z4-j-a4^0 or z*=a4.e(tn+1> vi, where n=0, 1,2,3. Therefore
the poles are at z=ae(1Ti)Ji, ae(3ri,,*l ae(6v<)U, ae(7iri),i of which
the first two lie in the upper half plane. Denote any one of these
poles by a.
Residue (at z=a) = <f>' (a),
, / / x z° (z —a)2 z® (z—a)2
[V a4 = -a1]
•8
(z + a)*U2+a2)2*
Now log «/» (z)=6 log z—2 log (z-fa) — 2 log (z2+a2).
. £_U)__6_ _2___4z_
• •
</» (z) 'z z-f-a Z2-f a2 *
a6 r6 2 4a 1
«/*' (a) =
4a1.4a4 La” 2a ~ 2a2 J
j_ 3_ 3_
'16 * a 16a *
sura of the residues= ~ fj_L_.l

16 [_ae {li ' ae3ir,,*J
= [p-Ttit* pHU4]
16a[e e J
=-,i.
16a
2/», nt-ba*
4 16a *
Hence from (2), we get
r. sO-’sr'
or
r
11. •S'/roH' that, if m and n are positive integers, and m
00 v2m
7r
dx = :
i **"+1 2//sin {(2w-f l)/2n} rr'
•2 m
f 22m f
Consider j ^ dz—J f (z) dz, where C is the same
contour as in Ex. 10. We then have from residue theorem,
Jc f(z) dz=J*R /^)<*f+|/./(2)*«2ir/rj?+. ...(1)

Since m and n are integers and m < /7> the value of 2m is at
least less by 2 than the value of 2n. Hence Lim zf{z)=o.
It follows that / (z) dz-+0 as R-+oo.
Proceeding to the limits, we get from (1)
fix) dx=2tt / 2? /?+.

i — 00 . .(2)
Now f (z) has simple poles for values of 2 given by
z2n-|-I=0 or 2*n=e<^^+1, *•
where r=0, 1, 2,...2/1-1.
Therefore the poles are given by
z=e*i,2nt ei%iri),tnt.. ,e
c[(2«+l) ni]l2n ^ c[(2/i+3) 7t/J/2« e[(4/1-1) 1W-J/2W
Out of these 2« poles first n poles lie in the upper half plane.
Denote these poles by alf a2, a8,.
•2m
Residue (at z=a) (see § 6*1)
-[dldz (1-f-i-jjz a
a 2m a2m+l
2n<xtn~1 2n<xZn
a 2«n+l
since a2n= - 1
2/i
1 2tn+lj
Sum of the residues = — ^ [a12m+1 -f a22T7l+1 + •. • + a
=
1 [g«4-e3.'fl+_.+c(2n-l) .9j
where
2/i
I e'» (1
2/i 1 - e2'®
1 c** (1 — einie) (I — e~Sie)
2/i* (1-e2") (I
1 £*e_£<S"+l) 19 _ g-i®_j_eC»n-l) <9
= __ 2-2 cos 20
1 2/ sin d — eZnie.2i sin 0

~~ ~~2n 4 sin* 6
1 j [1—cos 2nd—i sin 2nd]
~ ~2n 2 sin 0
= _ -——. [2 sin2 nd - 2/ sin nd cos n6]

4n sin d
1 sin n6
[cos nd+i sin nd]
2/i* sin 6
i sin2 nd . . 2m-f 1
since cos nd=cos —— rr = 0
2// sin d 2
. 4- 1
y-
2m
= -~——h V (sin/i0)2=f sin 77 (±1)?=1.
2/i sin 0 V
Hence, we have from (2),
P
J —oo
/(.V)^V=27rl('_—i— )
\ 2/i sin 0/
rr
/i sin t
f30 .v2m , 7T
or
„ . Z2///+1 V
2/1 sin rrj
12. Prove by contour integration that

f°° dx 1 1.3.5...(2n-3) 1
J0 (a+bx*)*“2*bl,tm 1.2.3.. .(n — 1)
(Agra 1954)
Consider |c ia+bzl)n^\c ^ ^ dZy vvhere c is the same

contour as in Ex. 11.
jc /(*) dz~\R_R /(*) <t*+lr /(*) <fe«2ir/r/?+ ...d)
Since Lim^*, z/(z) = 0, we get Lim^x /(z)=0, so that

we get from (1) when R->oo ,
/ (*) dx=2iriZR+. ... (2)
VC)
VO)
-VG')}i!5rnri*,“'{VC)
where </» (z) -vor
(a-f6z2jn
,
^F 76 )
1 (-/i) (-;;-)) .. f-/7-(/i- 1)+1]
Now </'(n_u (z) =
b*'
FMT
= (_Un-l n ^+0. . .(2/1-2)
4 +y©'}'.
Residue at the pole
_i_ , nn_4 w («+!).. (2n—2)
=<-1)! -TO}'"'1
« (/i+l)...(2/i—2) (V /Zn-2 _ ( — i )n-i
6" i («- 1)!

mr
288
Complex Variable
= - l»2.3...(/7- 1)./; (n-fl)t,.(2fl-2)
{V(!)j.■‘•-DK.-m
— —1-3.5... (2w 3).2n~1.1.2.3.. .(n—\\
2"-1 ■ o-'.'2’. . (n -Hn^TT!
__1.3.5...(2 n-3)
From (2), f f (x) dx= — 277 1.3.5...(2/1-3) 1

J-00 2”
w* * i.2.3
or r
J0
f(x)dx- _ 1.3.5.. .(2/7—3) 1
2"*1/f * T2.3...(/7-l) * ^=Tm»*
Exercises
Using the theory of residues evaluate the following integrals :
7T
1* If a > 0, prove that J
(z*+n2)8 4a3 •
2. 00
"y/2
/o°° *+r x* dx
Ads.
. 3\/2
3.
I0 C^+l)**
OO
Ans. —*r.
lo
4. or v
j: oo (x*+l) (x*+2x+2) Ans. —
5*
00 dx
5.
j0 1+*** Ans.
TT
3*
6.
00 a:* dx
j.o (*a+l)3 Ans. 6.
6'8. Jordan’s Inequality :

26 . _
— < J//7 0 < 0, >vAere 0 ^ ^ ^
We know that as 6 increases from 0 to cos 6 decreases

steadily and consequently the mean ordinate of the graph of
v=cos .V over the range 0 < x < 6 also decreases steadily. But
this mean ordinate is
1 (e , sin 6
J cos x dx>
e 6
If follows that when 0 < 6
2 < sin 6
TT ** 6 1
26
i. e. < sin 6 < 6,
IT
This is known as Jordan’s inequality.

Calculus of Residuces 289
6-9. Jordan’s Lemma.

If (i)f(z)-*0 unformaly as z-=c.
°nd (n) f(z) is meromorphic in the upper half plane, then
Lint jfeimif(z) dz=0, (m > 0)

R~>
where f denotes the semi-circle | z l=R, I (z) > 0.
Her' "e assume 'hat /(a) has no singularities on /-for

sufficiently large values of R.
Since (z)=0, we have for a given e, \f(z) | < c

when | z\=R ^ Rq where > 0
Putf.*0W lo I den°te 3ny semi_circ,e with radius R > /?,.

Cutting z = Re,$t we get 0
\r e*”»f (Z) dZ— [ eimRe,li f (Re"). Riei0 dd

J0
eimR C0S Q.e~mR siD 6 f (Rei0} Rieie ^
**• JJ y- eimz f(2) dzj^J | eimR ccs 0 | j e~„,R sin 6
x 1/ (Re*9) ) x | Ri || e*9 | dd
^ P e-mR sin 0 £ R Je
Jo
[ ■•• I/O) H/(*<?") | < €j
ftf/2
—2cR e~mR sin 0 dd
Jo
< 2e7? Pw e-2/»R o/tr ddf
by Jordan’s inequality
2cR (I — g~^R)
2/11 R/tt
C7T
—//jR\ -
’ m
Htnce Um;|^ JyemU f(z) dz=0.
6-1°. Evaluation of the integrals of the form

f? £i*> s. . f°® p (X)
J Q (x) Sln mx dx» J -c<> COS mx dx, (m > 0), where
(i) P (x) and Q (x) are polynomials,

(ii) degree of Q (x) exceeds that of P (x),
(Hi) the equation Q (x)=0 has no real roots,
Conditions imposed upon P (x) and Q (x) ensure the conver¬

gence of the above integrals.
i eimtf{z) dz,
P (z)
where /(*)=gr^y and
where C is the contour con¬

sisting of a semi-circle T of
radius R large enough to include all the poles of the integrand
in the upper half plane and the part of the real axis from x= — R
to x=R.
We then have by residue theorem,
Jc eims f (z) dz^^_R eimx f{x) dx+ eim2f(z) dz
=2*1 2R*. •••(^)
By Jordan’s Lemms (§ 6*9), we see that when R^>°o, f{z)-+ 0,
so that lreim*f(z)dz-+0.
Hence as R-+oo, we have from (1),

Principal value of
eimx f{x) dx~2ni 27R+
or 1°° fix) cos x dx+i fix) sin * dx=*2*i 2 R+.
Equating real and imaginary parts, we shall get the values of

the two given integrals.
Ex. 1. (i) Prove that (/) dx=Ta e"™'(W ^ °
(ii) Apply the calculus of residues to evaluate

f°° x sin x , . _ m
J 0 dx (a > °>-
(Raj. 1959, Alld. 46, Agra 49, 62)
(i) Consider
>m*f
dz=\cf(z)dz.
« C a2+z2
where C is the contour of
§ 6*10.
have
Jc/M *+J.
1
Since Lim =0, we have by Jordan’s lemma.
z~*°° z2 + a2
Lim^ Jr f(z) dz=0.
As R-+°o, we have from (1),
J / (a:) dx=*2nl ZR+. ...(2)

Now f (z) has simple poles at= ±ai of which only z=ai
lies in the upper-half plane.
Residue of (z)(at z=a/) = Limz->ol (z—ai)f(z)

»im: >-f7ia
Lim
z~*ai (z+ai) 2ia
OO
From (2), |
Equating real parts on both 6ides, we get
./x-77 e-"«
f: 008
a* t-x2 a
or
)°° cos mx , 7T ma
f COS X ,
c< w (Agra 1950)
In particular J j
+ x2 2c
(ii) Conside |c </z=Jc /(*) dz, where C is the same

C z2-f a
contour as of § 6*10.
Jc f{z) dz^_R f(x) dx+fr/(z) az=2niZR+. ...(1)

Since Lim z-»oo + a 2=0, we have by Jordan’s Lemma,
\rf{z)dz=0.
Hence at R->oo , we get from (I),
| ^ f(x) dx=2ni 27 R+. • .(!>

Now / (z) has simple poles atz=+fl/ of which only z=ai
lies within C, since a > 0.
Residue off(z) (at z=ai)=Um2_+ai (z-ai)f(z)
Lim2->ai z+ai~~'e a'

from (2), we get
j ^ f (x) dx=2iri \e~a=Txie~a.
Equating imaginary puts.

x sin x ,
r. —- dx=rre~a
oo xi-\-a£
x sin x , tt
*+Sdx=2e-
Ex. 2. Evaluate
cos ax
« f0 (X2+b2)
2 dx, (a > 0, b
(ii) r 00
J 00
cos2 X
(l+*‘l
2
(i) Consider J
w (2a-f-^2)2 ^=|c
J c /(z)
Jy j where C is
13 the
uic
same contour as in Ex. 1 above.
We, then have by the residue theorem,
Jc fWdz = \R_R /('V) <**+/,•/<*) dz=2niZR\ ...(1)
By Jordan’s lemma, Lim/^*, f (z) dz=0
or we can show it independently as follows :

Putting z=Rei0, we get
if | f” ea‘R (cos 0+< sin 0)
HL (/?=e2i0+^2
. R/e*®
^ fv e~aR sin 0 | eaiR cos 01 | Ri | | ei0 I dd

S Jo [Ri\e2i9\-\-b*\\i
n e~aRsin0 .R dd ^ n
^ J0 — iR^br —»0as*-^
a > 0 and sin 0 > 0 for 0 ^ 6 ^ n\.
fV
Hence as R->oo, we have from (J),
f(x)dx=2m£R+. ...(2)
. Now f (z) has double poles at z= ±bi of which only z = bi lies

in the upper half-plane.
Residue of/(z) at (z=-bi)
= </»' (hi) where <t> (z) = —iL_

(z+bi)2
f aie°il (z+bi)2-2(z+bi) ea‘n
L (z+bi)* J
=e-o> fai (-4b2)-4bil
L 16b* J
— e~at> (ab+l) .
4b3 '*
from (2), f f(x)dx=2ni\[-e~ab (-ab-+A)i

J -oo L 463
= 2^5(^+,)e-a6.
f°° cos ax , tt
J-oo (*+&)'dx=i
COS <7* 7T
o (**+6*)* dX~4b3 ( + *
00 (\4-cos 2x)
(ii) Here/—$ f dx.
J0 (1+*1)
Con;idcr fc (TtS *-/c*•

where C is the same contour as in § 6‘7.
**• fc /(*)dx+fr f (z) dz=2iri E R+. .. .(1)
By Jordan’s Lemma, jf f (z) dz->0 as R-*oo .

Hence when /?—► oo , we get
n. /«
dx~2ni£ R. -..(2)
N°w/(z) has double poles at z=±i of which only r=

lies within C.
Residue of/ (z) (at z=i)
= </>' (/) where (z)=j~~2

(z+i)
ie2i* (z+/)2-2 (l+e2'=) (z+/)“|
-P (*-W)4
2/>~2 fi-f-/)2 — 2 ( 1 -f e-3).2/
_L
16
2e 2_f“ 1 ~t~ e~2 .__ I +3e~2 .
4 1 4 7*
00 l+e2ix 7T
I -oo (r+x-f dx~~2(]+3e 9)
Equating real parts,

00 1+cos 2x , rr , 8
I _oo (1+A-2)2 ^-2(1+3 ^
it 7 1 f°° 1+cos 2* w , , 2,
Hence /=ij0 Tl+xy &=8 (1+3r 1
Ex. 3. Prove that
30 cos x dx 7r f B^ B~a \
\ (*»+<!*) (j:*+^) (r- / (°> *> 0)'
c°nsider fc u.+a.f'.,+^^=fc f(z) dz,

where C is the contour of Ex. 1 above. By residue theorem,
we get
jc / 00 ^z== |_r /(*) dx + j/' /(2) dz=2ni s • ••( 0
By Jordan's lemma, LimR.^ /(z) dz=0.
Therefore as R->=>o , we have from (1).

‘OO
/( Z 1 dz = 2rri Z /?+, ...(2)

— DO
Now / (z) has simple poles at z= ±ai and z= ±6/ of which

z = (7/ and z=bi lie in the upper half plane.
Residue of/(z) (at z=a/)=Limz_>o/ (z — ai) f (z)
ezi
= JLim-
:~*ai (z+aif Xz'+b*)
€T~a
2ai (62 — a2)'
-b
Similarly residue (at z = bi) =
2bi \a2—bl) '
Sum of the residues=27TJ-_^ [£-£].

J
1
ff0m<2>- \_x
“ * J*
Ex. 4. 7?>» integrating ei:j(z—ai)t (a > 0), round the contour
of Ex. 7 above, prove that
00
a cos x+x sin x
dx—2ne-'1.
i —oo x*+a
Consider f j2 = f / (z) Jr.
J C z — at JC
Jc dz=zJ*R /(■*)dx+\p f{<z) <iz=2r,i z R+- ••Ah
By Jordan’s lemma, LimR->30 f (z) dz = 0.
From (1), we get as 7?->=o,

00
i / (x) dx=2iri £ R+.
~co ...(2)
Now the only simple pole of/(z) in the upper half plane
is at z=ai.
Residue (at z=ai)
= Limr->a/ U-^/W-Um . ei: = e~a

OO
• t From (2),
-■i-00
| f(x)dx=2nie
00 (cos x-f /sin x) (x + ai) , - . „
oo
i:oo
-7-r——:—rr— dx=2nie~a
(jc—4//J (x-\-ai)
or (* cos x-r? sin x)+i (x sin x-M cos x) ,
f -oo -T‘Ta‘-d,=2*,e-°.
CO
a cos x-fx sin x
f -oo xz4-a*
dx = 2ire~a.
Ex. 5. Prove /A* f* -Sli* -' sin 2.

J-OO ^+4x4-5 e
c°nsider fc dz= Jc / (z) *.
where C is the contour of Ex. 1 above. By residue theorem,

we get
\iC /(*) /(*) dx+\r f(z) dz=bTi2R+. .. .(I)
By Jordan's lemma, jr f{z) dz-+0 as R-*oo .
Then as R->so , we get from (1),
f
J —DO
f(x) dx=2îER+, ..(2)
Now/(:) has simple poles
Let a-- —2-f-/ and /? = —2 —/.

The pole at z=a lies in the upper-half-plane.
The residue (at z=a) = Lim(z—a)f(z)
(z - a) _ e la
= Lim
z-+a (Z—a)(2-/?j (a-/?)
cK-2+« £-(1+20
2i 2i
From (2) f f (z) dz=ne~ll+2i) = - (cos 2 — i sin 2).

J —30 ^
30
Sin AT 7T
i-00 *2 + 4y+5
dx= — sin 2.
6. Prove that when ni > 0,

f30 cos w* . -n . / , n\ —(]/2)wv/3
J. tt?+idx=V3S,DVn+3)
tmz
Consider
Jc
24+Z2 +
where C is the contour of Ex. 1.
I *“jc f (z) dzy
By residue theorem, we have
[c /(*> ^ = J^R /(a:) ^ + dz=2rriER+. (1)

By Jordan’s Lemma, LimR_^ /(z) =0
and so when /?->oo, we get from (1),
j — OO
/ (»y) </.y= 2tt/ 2R+. ... (2)
Now f(z) has simple poles for values of z given by

z4 + z2-f-1 = 0
or (1 -z8) (l+za + z4)=0
or (l-z6) = 0
or *=**««»« k=0, 1, 2, 3, 4, 5.
Out of these e<T‘>'3, e'**"'3, g‘5iro/3 are the r00t$ of
l+z2+zi=0. The only poles within C are at
2_£(ir/)/3 an(j z = e(2iri}/3_
Denoted these poles by a and a2 respectively.
Residue (at z—a) =
*f,+z2+j‘) z=a
.twiCl
2a + 4a3 *
>m( a ,mW
sum of the residues = -K-.
2a + 4a3 2a2+4a«
,tn<a ,mid
2a+4a3
+ 2a*+4 (V *8=1]
<W«'/3) wf.* <2*0/3

-![ eC’ri)»3_|_2c-7r‘ 'ê<Z7r«)/3_|_2j
_ 1 rew/ t(l/2)+(v'3/2) /] em/ (—(1/2) +(43/2) 1] ->

2 Ll + (V3/2)/-2 + -i+(V3/2) i + 2~J
= e-m(v/3/2)f_£^! , g~w</2
LV3/-3^V3/+3j
=e-m(V3/2) f3 (e’"</2-e-w<'2/4v/3/ («»•«* +
L 3i* — 9 J
1 .
= _l2e m^3/2) [6, sin"V2V3, cos "'J
= —- Cv/3/2) f V3 . m
V3 l/F 2+*cos2 a
i_ p—m (V3/2) COS ~
V36 L
IT
6
. m
m ,
2
. r77-
sin --f sin , cos _
6 2 a
- i,
V3 ■wm•" (M>
From (2),
Sin (£+!).
Complex Variable
^ ■*-% - ™ - (M)
cos mx
or -ldxê-mwm sinM.
r, o !+A:2-t-A
7. If a ^ 0, prove that
r (^ “1“ ^ ) COS W , 7T n Q
(0
J0 -r+w £ K / ) cos 2 :
f°° x sin ax , n . /xn.
00
J0 7+^*“V3e hWsinI-
f (1 -j-r2) . r
(0 Consider . , , , ,
Jc ~i-4-ra+za
C l-fr^+z dz=)c /(z) ^z» where C is the
same contour as above.
By residue theorem.
|c /(z)<fe=J*s f(x)dx+\rf(z)dz=2mZ /?+ ...(1)
By Jordan’s Lemma, Lim^^ /(2) dz = 0.
When , we get from (1),
f / (a:) dx=2ni £ R+. ...(2)

J —00
Now /(2) has simple poles atz = eiri/3f ei1Ti/3, t47r<'3, e6ir</3
(see Ex. 6 above). Of these only two lie within C. Denote
any of these three by a.
Residue of/(2) at 2 = a is
(1 + z2) eia:
Lim,-*., j
-(H-22-j-24)
(1+22) eiaz (l *+- a2) eiaa

Limr_,a
2z+4z3 “2a (1 +2^)'
• • Residue (at z=eiri/z)
(] q-e2~,/3) eai jcos (r/3)+/ sin (tt/3)}
2e“;/3 (1 + 2e2"'/3)
J , ^ __/7* V3
0- 2 ‘ 2 Wf-^)
1 V3 .
(2+ 2 1
I (14 \/3<) e (v3/2) a (CQS 0/3.j./sjD aj2)
2* (I4-V3/JV37 "
737 e a(v/3/2,[cos?+. sinf]

2 y/3i
and residue at z=e2iri,i
(1 eai {cos (2r/3) + i sin (2tt/3)>
= 2e2*'/3 (1+2.^
^ ea/ {—1/2-Hv/3/2) />
3(-Hy»)(l-1-V/30
a (V3/2) fcos *_/ sin^l.

2i/3i
Sum of the residues^ * .e a (v/3/2) 2 cos 5.

2\/3/ 2
00 e-awm cos. a
1
-00 1 — V3 2-
/
•00
(12;, e —(v/3/2) cos«
— OO
j 00
(1+AT8) COS 05 ^
1+*• + *•
ze0*
ax
jr_
V3
fl (v/3/2) cos ?
2-
(ii) Here/(z)= Poles are the same.
ze aiz <xe ora

Residue at z=a is f i =
l_2z + 4z3Jz=a 2a (l + 2a*J
~”2 (1 + 2a2) *
e<7/ {cos (k/3) + * sin (?t/3)}
Residue at z=eni^ is *
2 (l+2e27T//3)
e-aWV2) ^1/2
= 2V3?
and similarly the residue at z=ein<13 is
e-(VV2) a e-ail2
—2y/3i
„-(V3/2)a ...
Sum of the residues =—. ... (eai/2-e ai12)
zy 3/
__ (V3/2) a sin fl/2
V3" '
Hence
r
J —GO 1
xeaix 2iri .
^ ) sinf •
0
Equating imaginary parts.

^ sin aa: , 2tt , ^
1
• [ sin w , tt / /->//!, sj
*• J0 H^-^v3^(V3/2)flsin|.
8. Calculate + ^ (Ara-f-Z)2) ^ (a > 0, b > 0) and
deduce the value °fl 00 a:3 a:

0 (AC84-a2J2
Consider Jc (1r_j. *=Jc / (z) rfz, where C is the

same contour as considered above.
By residue theorem,
Jc J/ <*)d*+^rflz) dz=imr«+, ...in
Since Lim/^*, (Z2_f_£*)=0> we have by Jordan’s

Lemma,
jrf(z) dz—0.
Proceeding to the limits when 7?—>-oo, we get from (1),
f — oo
f (x) dx=2iri £R+.
Now f(z) has simple poles at z=±ai and z= ±bi of which

...(2)
z=ai and z = bi lie within C.

Residue off(z) at z=ai is Lim (z- ai) f(z),
_T z3^*2 c2e-°
“ 1D3^fl' (z+m) (z2-f62) = “2(62-a2)*
Similarly residue at z=6/ is-
2 (a2-A2)’
Sum of the rcsidues = 1 (b*e~b-a*e-a).

2 (a--62;
From (2),
/!./(*>*-r oo (at-M2) (AT*-fA2J (a* — b2)^2e b~a'e °)'
J a:3 sin a: dx
or
r. x3 sin r cix
■°° (**+**) (x-+b£)
When a=bt we proceed as follows
77
2~ (a2e~a -b2e~bf.
2{a*-b-)
00 o • ,
! X3 Sin ^ dx n
When the r.h.s. is of the form ^ hence

*•00
sin x
xl a,n
J- ± •* /i„_I
(^M7? dx~Um^, -—4£-

• "■ (b'e~b — 2be-b)
^ (<7- 2) <r°
-4
a3 sin x dx
i: — x> a
(x2 -f- a*)*
"■ (a -2) t~a
4
9. Show (hat when m and a are positive,

x sin mx tt . , nin
i 1T^r*x=2*e '°IV2 ™ f•
f ze’"
zemiz C
Cons,der jcii+r*rfj:==Jc
C Z* + /(Z) *• where C is .he same
contour as of Ex. 1.
\c f{z)dz=\R_Rf<x>>dx+\rf(z)dz=2mZ:R+. ...(i
By Jordan’s Lemma, Lim*^ dz=*0.
When y?->». we get from (1),
r f(x)dx=2iriZR+. ...(2)
Now /(z) has simple poles for values of z given by
zi + ai=0 or 2*=a*el2n+l> vi
or (^)n/
= ^V 4 7 , "=0, I, 2, 3.
Of these poles only z=ae7,‘l* and z=ae3ntl* lie inside C
Denote these poles by and «t.
ze nit 2
Residue of / (z) at
at z=
z=«1 is |
o*)\
Jz = *,
a^rniCL gmiCL
4a3 ~ 4a2
1 remia e”**0*-!
Sum of the residues = —7-*+ —
4 L ai <*2 2 J
1 r miaeiTz/4
emioeir>l ~mi • ae3ni/4
_4- __
.2^<7r/2
=4[_ a2eilTl2 1 fl2^3lT</2
r mia mia
1 (1+/)
V2
4a2i [_ 2 -e
ma
ma
V2 m£7
2^2_ Sin V2'
ma
From (2), f(x)dx=^ e V2 sin ~2
Equating imaginary parts.

ma
x sin mx 7T —Try . ma
i: x* + a* '
dx=,e v-2 sin
V2
ma
x sin mx ma
Hence sin
x* + a* **-h *~V2 V2
10. Prove that
f50
o
cos mx ,
—-j-dx=-—~ e
^-fa
J^ + fl4
7r
2a3
ma
,7? . /mo , rr \
V2 «" (v2+4 >
Consider —4 dz=\ f(z).dz. where Cis the same

C z4 + a4 Jc J w
contour as above.
jc /(z) /(.v) dx+^r /(z> dz=2ni Z R+. ...(1)
By Jordan’s Lemma f (z) dz—0.
Hence when R-*oo , we get
| / (a) dx—2-rri Z R+. ... (2)

Now as in Ex. 8,/(z) has simple poles at z=aefir/4, aesirli*t

ae5ir<'4, ae7*!'4 of which first two lie within C. denote these poles
by alt a2.
Residue of / (z) at z=a is
ae mia
4a4
Sum of the residues
3/tc
~1
= -4^4 exp.( miae 4 ^+0e4 .(m/ae
exp.( miae 4
_ i r■ -t
j_ 4 ami (cos ^ +» sin -J-)
e * .e
4.3
-e~^.e~mai (C°S 4 ■'!ln0]

1*1 w
[v e«=-e 4]
e-mWv/2p| <ma tt >
=_ 4a* Lrxp<sV2+4 ^
p—mafy/ 2 f ma ir\
4fl3 •2'S,nl lv2+d
—maly/2 _ / ma
From (2), f
J —OO
f(x)dx=^r
Q° sin ^r/2 +
0-
OO cos mx
uus // . it —malv2 / ma tt\
f _oo x* + a* dx=*e
OO cos mx ,
Sln (v2+7j-
it — ma/y/2 f ma -rr \
Hence
11.
io *+al
Prove that
ax = —- e
2a8
sin 1 -^ + -4 J.
W2
00 a:3 sin mx , n —maly/2 ma . _

I . *13
Proceeding as in Ex. 9, we get
-1 ' to. ^ <">»>■
foo r oo y3pm2i
J-oo TO*-2"**-
...(1)
304
Complex Variable
Simple poles of/(7) within Care as before /. e. at

z=ae7ti,i and z=aeZlti‘*
.mat
Residue of f(z) at z = a is
!<**+-*'L 4
Sum of the rasidues

= 4 fexp {wme^j + exp {miae™*1*}]
“* [exp■ {^'<1 -«}+e*p j™'<-i+o}J

= $e~Ma/V2 [ewai/\/2 + e-maily/2j
= sf“'"<3/' 2 cos mall.

f 00 Y^fjTtUCi
■ * hrom <! J _oo dx=ine-”">IV2 cos ma
Equating imaginary parts,

-v3 sin m.Y r/.v n
/ — —cos
V2
Hence f°° * sin •»***_ * ma/v2 ma
Jo x* + a< 2 S y/2'
12. Prove that, if 8 is the unit of circular measure,
(i) r -
J — cc .V
sin a: 7t
(^2^+2j rf*“27 (ff_C0S 5 + Sin S':
nil r 1 —COS AT , IT ,
{ ) J-oo xTx*-2x+2) dx=2e (e~cos S “s,n a>-
Consider
1 —eiz C
/c 2- (2“—2r+2) <h = \c *•
of a large semi-circle in the upper
half plane and the real axis from
- R to R.
|c f{Z) dz=j f f(z) dz+fix) dx=l7ri £R+.
Now f(z) dz | ^ J" ! r//?e<0) Rie<•

ReJe {R^* - 2Re”‘° f2)
i 1 - exp (-7? sin 6) ,n

R'-lR-l
n
de~*0.
as R^zo since sin 6 is -f
Hence when R-+ oo, we get from (1)

•oo
1 — OO
f(x) dx=2iriZR+. ...(2)
Now f(z) has simple poles at z=-^-v^4—?)=i + /0f which
5*1+* lies within C. Leta=l+iand 3=1—i. Clearly 2=0

is not the pole since 2 is a common factor. If we write the
expansion of eix, we have 1 — eix= 1 +
Residue of /(2) at 2=a is

1—
Liniz-*a (*-«)/(*) Lira
2 (2-/3)
1 — ta 1 -
a (a -0) (I + l)#2l
=4/ (1 - 0 J (cos 1 + 1' sin 1)1
4~f (I—/) [e-cos 8—/ sin 8],
where 8 is the unit of circular measure.

From (2),
r 1 —e*
00 x (x2 — 2*-f-2) ^x—2e ^
Eqating real and imaginary parts, we get

^ ie cos 8 — i sin 8J.
f°° cos X1 J
J -00 x (X2- 2 x+ )dx~ e
2 2 ê~‘cosS~ sin SJ
and f°° sin * , tt r
J—00 x (at* — 2x-\-2) **x 2e -c°s 8 +sin 8].
13. Prove that
f°° sin x 2*r{v'3-F2} ,,,,
J-00 (**-fx+ 1)* dx— 3^73—^ e ^3/2 sjn j
Consider n).*=fc/(,)*.
where C is the same contour as above.
Jc f{z)dz=^_R f(X) dx+^f f(z) dz=2n/2R\

(1)
By Jordan’s Jemma, UmR^r f(z)dz=0.
When R-+ » , we get from (1),

OO
...(2)
f -OO
/(.v) dx=2ni Z R+.
Now /(z) has poles of order two for values of z given by

-2 —z+l=0
1 + iy/ 3
or z=
1 +iV3 „ 1-/V3
Let a =-2
Of these poles only z=a lies within C.
Residue at z=a is (a) where </> (z) — ^z_^t
Now log <j> (z) = iz—2 log (z-(3).

Differentiating, we get
_£_
 (a) (a_/3)J
=(^W)* L V3J
--p{'atxjl p-ifj
uV3;2
e-^t2e\/2i
(2+\/3) /
3v/3
From (2), jl„ /<*> rf*=3V36 V3/2<2+V^

x (cos $ + i sin §).
Equating imaginary parts on both sides,

00 sin *_e—\/V,2 (2+\/3)sin
1 -oc (.v2 + x -t-l)2 3\/3
14. Prove that
f00 cos fl.v 7T c~°6_ r(2afr)w-K {2ab)»-y

Jo [X* + t>*r+xdx'~n ! * i2£)2n+1 L( 1!
(/i i-2) (/H-l) n (n~ 1) (2
- ab)n~z+..
»>—- • •
+
In) n
+ 2 ! n\ J’
where a and b are real and positive.
Calculus of Residues
307
az~ f f(z)dz, where C is the

Consider jc [zt+bz,n+i ^")c ' ' ' '
fc
same contour as above.
JC ...n)
By Jordan's Lemma, Lim^_^ J /• ^z*0.

When /?-► oo, we get from (1).
f" f(X)dx=2niZR+. ...(2)

Now/lz) has poles of (n+l)th order at z= ±6/ of which
. =bi lies inside C.
Residue at z=bi is —. *An (bi)

rl •
...(3)
where 4> 7bi)n+l
Differentiating (3), n times by Leibnitz’s theorem, we get

., w x (oi )" ea1z , n(ai)n-x eM {—n -U
* ^)=(z+ft7prR+ 1 ! (2+W)"+*
"(n-l) (-if-|)(-w-2)
(fl<) 4-...
2 ! (z+6/)n+3
(/; + 1) (h+ 2). . . (2/i) ^
+ e°<'.(-l> (Z+bi)in+l
, fc r 6" , « (n-M) a”-1

•\ </>"
(6U = e-"L(.6)„t—,+ <2*)”+l.i
(/!•+• 2) 1/1+ 1)./I (/!" 1) ffi-a
-.4“ • • •
4--21- * (26)n+3.i
, (2w) } 1 1
n ! ’ (2bfn+l
rn+l.. i J J ’*
••• Residue= -- [(2a*)"+("+' — (2a*)"-
— , J^rTTi [(2^)n+("'(2fl6)n_1
(w+lM (*> 1) (2a&)"-»+... + ... 4(-^t‘].

Hence equating real parts, we get from (2),
r
J-~
cos ax_2*r °
Tj^yH\ax-~n ! # (26)^
(*’
Y;
r(2a*)"+^ V " (;2n*)->
+ -‘-+ n I J
or f- C°8ax dx= — e"a-- [gab)”+{n+ !?-/J (2ab)n~*

Jo (x*~tb*)'»1 n V{2by**\} * + 1 ! { >
(2^1
+ ni J*
Note. Contour of 6*7 can also be used to evaluate certain
other types of integrals. The following examples will illustrate
the method.
15. Prove by contour integration,
00 log (l+x~)
dx—TT log 2. (Agra 1932)
io (1+*)
Consider f |oj[_(£+0 j f f (z) dz, where C is the con-

JC 1-f-z2 JC J
tour of § 6’7.
Jc /(*) dz==\R_R /(*> dx

l-|r / (z) dz=2rri Z R+. . . .(1)
Putting : = R,0y we get
fr /(z) dz
r log (Rei0 i). Rieia
l -f-RKe2*'
. * I log Re" | + l log 11 + I/Re-") ■ R .Q
d0
I’
1 R2 - 1
[V | R2 e2i9+\ |=j R*.e*ie-(- 1)
> | R*.e*<* |-|-1 |= f?2- 1 ]
^ + 0) . i 1 I 1 , * M | rfQ
-»0 as /?->oo .
R log R , . log R R2 r% % n
Since Lim^^ = -•TjTTi ==0x 1—0
and Limôo ij ,og (l + ^",0)| = °-
Hence when R-+oo , we get from (1),
f(x) dx= 2ni Z R+. ...(2)

— OO
Now /(z)has a simple pole at z = + » and a logarithmic

singularity at z = ~/ of which z — i lies inside C.
309
log (z+i)
Residue of f (z) (at z=i) — Limz_»/ (- 0 ^ ^ (z- i)
U
Therefore from (2), we get
“ lo« t*+!) dx=„ [log 2 + '|].
I — oo A"-h

\
00 log(l+*2)
I
f°° log n
n+x^_)
+
J+*
2
dx = tt log 2
loc 2.
or
Jo »+**
16. Prove that

n 30 CO5 at24-sin X2 - 7
Zi
iz 2_ 1
(jx=o.
</z = f
( Agra 1945)
/(2) dz, where C is

Consider the integral |c " 2 2
a contour consisting of the real .F

axis from — R to R and a large
semi-circle r of radius R, centre
the origin in the upper half \
plane.
zlL \R
Within the contour thus o
defined the function / (z) is
regular and hence by Cauchy’s fundamental theorem, we get
f(z) dz=j^R f(x) </*+!r /(r) (/z = c- • • *(1)
Putting z = Rei0t we get
= - P e"<0 lexp [iR* cos 20 4 < sin 20)} - 1] dO

R Jo
— p ^ c-<0 [{exp (- R2 sin 20)}

x {exp (iR2 cos 20)} — 1]d0.
v ((,/(*)*!< Ajjir
-f [exp ( — R‘ sin 20) | exp (iR2 cos 20) | +1 - 1 |] ^
-i I* [exp (-A2 sin 20)+l]
—>0 as R-*x>.
Hence when R-+ co , we get from (1),
f /(*) dx=0
oo tx.2 r°° .IX'

—1
- 1
or
i — oo a:2
dx=*0 or 2
J. ■ 2
dx—0
Equating real and imaginary parts,
s. 00 cos *a -1 . n
o — dx=0
X-
00 sin xz
...(2)
...(3)
or
io *
dx=0.

03 cos A2+sin -y2—1
dx=0,
fo x*
17. If 0 a , s//oH> ///a/

oo
tan -i 7ra
d.r=
j
_oo A'2—2.y sin a + 1 2 cos a*
Consider
Consider
f _^ ^ ——-—
Jc z^—2z sin a+1
dz=
Jf C Jf (z) dz
round the contour of Ex. 16 above.
Then Jc / (2) <fe= /(2) + /(*) dx=2niSR+ ...(1)

Now / (2) has simple poles given by the roots of
z2-\-zi (eai — e~ai) + 1=0.
or (r + ieia) (z - ie~ia) = 0 or z=-ieai, /e"a< and - i,
a logarithmic sinularity of which z= —ie~ax lies within C.
Residue of f(z) at z=ie~at is :=ie-*‘
log (1 +5_aJ]-?_ l0p r j _|_COs a — / sin a}.

= / (e_a* + e+a<) 2i cos a e 1 r
Also J log (1 — ;7?g,g) R/V*9

r /(z) * | < | 0 , (i?e<fl + /6a‘) (Reie - ie-‘a)
i ™
71
log {I + /? sin 0 - iR cos #} \ p ^
1 (/?-!) (*-!)
311
. R cos o n R
Iog(l+2/?sin 6 + R2)+ -i tan J
(K-l~7
i? cos^ 6
r r iogft+*iogfi+J-sin 9+x0+tap 1i+* cos-e]Rde
J-- (R-D2
log R n
->■0 since —s- •
(
From (1) when R-*oo ,
oo
^ / (x) dx =2ni.—f—n
2/ cos u
log(l + cos a-isin *).
Equating imaginary parts, we get

rr sin a
oo — tan -1 x
dx = - tan -l
COS a 1 +COS a
_oo X2 -2x cos a+ 1
a
tan"1 tan ^=-2ôs a
COS a
Txa
or
r-?= tan-1 x
2x cos a-f 1
c/x=
2 cos a
Exercises
,-a
COS X
1. Prove that [°° </x= — (a > 0)
a
I — OO a '1
(Delhi 1959; Alld. 1943)
[Hint. See Ex. 1 § 6*10].

f°° sin dx = y-2 sin 1.
2. Show that
J — oo X* — i2x+5 2e2
f°° COS X , .
3. Show that d\ — ^
0 (x2 + I)2 2e
oo
cos ** cos n + sin a) e"a (a > 0).
4. Prove that
i x4 + 4 8
5. By integrating around a suitable contour, prove that
6. By integrating
r- -a cos x-fx sin x jx=_q
x2 + a-
the contOUr °f § 6'?

prove that
f00 x tan'1 x dx f* xsin^x , ? ,/2 _ i
J0 Ti + 2x«? -J. (l+x'r 8
log (1 — iz)
7. By integrating round the contour of § 6*7,
z2 — 2z sin a-f-1
prove tTiat
CO
tan —i 7TX
dx=
j x2—2x sin a-f-1 2 cos a*
6*11. Case of Poles on the real axis.
So far we have considered the case when the integrand of the
complex integral has no poles on the real axis. We shall now
explain the method when the integrand has simple poles on the
real axis. In such cases we shall follow the procedure known as
‘indenting at a point\ Method will be clear from the following
examples.
1. (i) By integrating — around a suitable contour, prove that

oo sin x ,
- dx=~.
f°o J
i x 2
- COS X , TT
-
o dxz=2*
Consider e— dz — j ^ f (z) dz,
where C is the contour consisting of

(1) real axis from p to R
where p is small and R is large,
(2) the upper half of the
circle j z \ = R,
(3) the real axis from — R to
-P,
(4) the upper half of the
circle y, { z \ = p.
Since there is no singularity inside C, we have by Cauchy’s
theorem,
jc f(z) dz = j* f(x) f (z) dz + f(x) dx+^f(z) dz
= 0. ...(1)
By Jordan’s Lemma, we have
Lim
R->oo \r f U) rf-==0-
Since Limz-W) z/(z) = 1, \ve have by Theorem I of § 6*5,

Unip^Q | J (z) dz=i (0 — w). I = — tti
or directly we get on putting z=>pei0,

f /(z) rfz= f° exP {'P (cos 9-j-i sin fl)}
Jr Jrr p
/. Linipô 1^/(2)^=Linip^Q i J exp {ip {cos 0-f/sin 6)}dd
=/ J° ^=-/tt.
Hence as p->-0 and /?->-oc , we get from (1),
J0 /(x)</x+J_^ f(x)d.x-7ri = 0
f00
/(x) dx — î
fOO <r
or — dx — rri.
J —oo x
Equating imaginary parts on both sides,
f°° sin x
x 2
1 -ei:
(ii) Consider Jc 22 ’) </2
- c /•
where C is the same contour as in 1 (i).
Within and on C, f (z) is regular and so by Cauchy’s theorem,
we have
jc /(*> /(x)Jx-f /(’) </z
fi* fix)
dx+\ f(z)dz=0. ...(1)
Now / (z) r/z j < I” I Rie» | <tt
fT ljf- exp ( - /? sin 0) ^

Jo R
-*0 as R->x since exp ( — R sin
and sin 6 being positive for 0 ^ 0 ^ n.
and
jT f(2) m: *- pit‘° i,°
-j: 2 iO dd
[1+0 (P)J dd
-K
Lirripô f(z)dz=j°d0=—v.
From (1), we get as R->oo and p->0,
|0 f(x) dx—tt=0
or f /(*) ^-x=7r-

CO
l - cos x
i CO
— OO
dX—TT
f°° 1— co
— COS X TT
or - dx= 2
J 0 X*
Ex. 2. Prove that
“ "»** * (l-e->.(«>0). (Alld. 1942)

jo *(*2+<,2)“2aa'..
Consider the integral Jc r (za^-a^)~fc ^ ^ dz' taken round

the contour C consisting of
(i) the real axis from p to R,
(ii) a large semi-circle T in the upper half plane given by
! z\=R.
(iii) the real axis from - R to — p, and
(iv) a small semi-circle y given by | z |=p.
Now / (r) has simple poles

at z=-0, ±ui of which only z = ai
lies within C.
The residue of f (z) at z=ai
is given by
e•<- c-a
Lim O x
z->ai z —2a~
f(z) </z=j** f(z) dz + ^r f(z) dz + /(.v) dx+j^ f (z) d:
. — iri
=27r/-^2 = -ê
...d)
By Jordan’s Lemma, Lim^^ /(z) dz = 0
we get directly,
- w fir 1 exp (/7?Ccos 0 + /' sin 0)}.Riei9 dd \
r J U) | < I /?ei9 (a2 + tfV“)
putting z = Rei9
exp ( — R sin 6) dO
J • R*euo -(-a2) l
i (R2 - a2)
^ ! R*\-\a*1]
—►O as R-+oo, [v 0 < 6 < 7T, sin 0 is -fvej.
And putting z = pei9, we get
f f° exP Vp (cos ^4-/ s'n 0)} pt'ei9 dd
J y J (Z) J* p.e" (a* + pV<*)
= a2 j- n+^P (cos *+* sin °)+- * •j[,+j2 dd
=^-J° {1 + /P (co*« + /sin 0)+...}[l-e2 e.«+. . J,/#
-I5- f° - as p->0
o
a . n
TTl
a2
Hence as p-*-0 and R-*■ >o , we get from (1)
C°° C° -rri ni
J.
7T/
n
00 jc («*-»-jc2)
dx=-t d-^'a).
sin x dx tt
= -r (1 —e~a)
i _oo x {x'+a^-a*
CO
sin x TT
1 0 x (x*+a*)(Jx-2ai{i~e^)‘
Ex. 3. Apply the calculus of residues to evaluate
00 cos ax - cos bx . , _
-dx, a > b > 0.
I ■ 2
(Agra 1948, Raj 1958)

Consider the integral K J

c ,aiz_Xtiz
Aai= — eliz f
- </r= I
^
f (z) dz, where C
is the same contour as in Ex. 2 above. Within the contour C,

f (z) is regular and so by Cauchy’s theorem, we get
jc f (z) <fe= f (x) dx+ jr f (z) dz+ j~eR f(x) dx
+JV /(*) *-0
or Jl + I2-\-l + 3 /4 = 0. ...(1)
Now
, r , fexn (aiR tcos 04-/ sin 0)) — exp {biR (cos 6-\-i sin 6)} D. ig ^
!4I^Jo| Reuo
putting z=Rt.'°
< r exp (-aR sin 0)4-exp (-bR sin 8) as
Jo R
[V I exp (aR s’n ^)-exp (aiR cos 0)
-h( — exp ( -bR sin 0).exp (biR cos 8)} \
<; exp (—aR sin 0) | exp (aiR cos 8) |
4-exp (—bR sin 8). | exp (biR cos 8) |
=exp (—aR sin 0)4-e*P ( — bR sin 0)].
And putting z=peiet we get
fn exp (aip (cos 0 + / sin 0)}—exp [bip (cos 04 /sin 0)} n/^t0 jq
/4==Jr o PV«
— —J° e~<0 (U-^)/p (cos 0-f/sin 0J
-T.T-, /2 P2 (“z-b2) (cos 0+/sin 0;2+ ...1 dd

+* • -J
sal J° 6-*'® {(a—b) i (cos 0-f / sin 0)4-O(p)} c/0.
/. Lim^0A=-j: (a-b)dO=*{a-b)n.
Hence when R->oo and p->0, we get from (1)
J°° /(at)</at4 0 4 /(*) <** + («-&) 77=0

-CO
or I /(a) c/a = - (a - £) 7r. ...( )

2
.. .(2)
Equating real parts on both sides of (2),

r°° cos ax - cos bx ,
-- - Jx=-n(a-b)
J -CO A
C°° cos ax—cos bx , tt

or -p- dx=-^(a—b).
4. Prove that
f00 sin2 mx dx
J0 =
*» (**+*«) “4a» (* 2ma — l + 2nia), (m > 0, a > 0).
Here /- P sin2 /wat dx

dx 1l ft 1 - cos 2mx
Jo x (a*+x‘) 2J *2 (d*+x*) dx.
| — glmzi
Consider the integral I* dz, taken round the con-
Cz3 (a2 + z2)
tour consisting of a large semi-
circle r of radius R, the real axis
from —R to - p, a small semi¬
circle y of radius p and centre the -R
origin and finally the real axis from
p to R. Here we indent the origin
in a small semi-circle to avoid the only singularity on the real
axis. Hence by Cauchy’s theorem on residues, we have
f f 1 plrnti C— c 1 pinixi
2 , (IX
x2 (a2 + A:2)
1 — e *wu< fR 1 _ gSnwrrf
+ 1V z2 (tf2 + 2 P A:2 (a* 4- a:-)
r. dx
= 2ni £R+ ...(1)

or A+A + /» + /«*=2»r/ -27/?+
Now I /, |=| f —CXP (2w<7? (cos Q + i sin 0)}
1 I Jo (a*+/?***<•) ™
f: TfSnJSJif -' i * i ~ ■»
1 -f exp ( — 2mR sin 0) D
i: R2(Rt-a2) 'R<
r* 60 r dO
J„ i
R (R2 - a2).exp (2mR sin 0)
0 as R-+x>.
and /. = f° J ~ exP {2wi> (cos d + i sin (/)} „ ,

J* p'e‘u(a’ + p’e’n) 1
ll + OWI*-^
=ai lies
Now the poles of /(*) are at z-±at of which z
inside the contour C. The residue of J (z) at z-ai
| e2mzi
= Limz_+aj pfZ+ai)
j _ e-2tna e~2ma ~~ 1 .
2fl3/
As and P-*0, we have, from (!)

* oo ,
1 —c j..
2mir -
9 . g~2mq - 1
.2
i 2"3'
1—e2m3'* __ 77 r„- Stna — J
or OO ^
Equating real parts, aie get

1 - cos 2mnx ^ 7T
0*
! _oo x2 (a2 + *2)
7T
j r<» 1—cos 2m* I ->r2ma]
J* [r:mo -
or
-H !?"(«*+**) 4*
13. By intergating ^ _ ,w (1_ft)
row//*/ the contour of $ 6 7 and p & ,/ia'

f^/2 r sin 2d_ q jo
.2
o 1 — 2r 2^+r
0
7T
• log (1 +r). ‘f ~ * < ' 1,
4
-1 or 1.
=VogC1'1'>)i/r<
r 2zr
zzr__ logOrlf) Jz= (
r^-r' / (z) ‘/z>
Consider j (_ r.-,- . âXTHTn* * 1 +z J
c z2 (H-0*+^l*r)
where C is the contour of § 6‘7.
we then have by the residue theorem.
...(2)
Now ■ | |r /(z> dz I
r i _2£^Mogii^'^!!) . 'ir..Rie“ \d0

< j 0 u+TT+O -n !+**
319
2rrt2 ^ log ( - iRei0) {1 + ^ e~<9 }

<
]0 {^(n-rr-o-'-)2) <**-“
2r/?2l log ^^/(e-^+log (l+^- e'*9)
dd
[ {R2 (1 -4-r)z_-'( 1 -O2} l**-7!)
^- \ - ' '
0 _ ^ \ *n
Y log (-ie") = log [exP {' (0 “ y ) }] =' (0 “ ^ )

. + • .1
<
■iO
2rRl [log ^+(fl--)+',log(l + -^
' --- ■■■■ *
< [R* (l+r)2-(! -r)8j (/?*- U
11
,„.[y.r^gy I,'.0,(1^.
= Jo -{R* a+rf-(l-r)'} (R2-1)
log /?
0 as /?-»•» since •0.
Hence when R , we get from (I),

00
/(x) dx=2*iERJr .(2)
i — oc
Now f(z) has simple poles at
1 r/ if - I < r < 1 ;
z=i, z=
1+r
1—r . if r < 1 and > - 1 ;
Z_ \+r‘
• •
and a bianch point at z= -/ which does not lie inside C.

We consider - 1 < r < 1.
Residue of/(z) (at z = i)
2zr log (I—12) 1
— Limz-W (1+r)*4-( 1-r)2} * (z + /)
2/r log 2 1
‘ log 2.
- _(l+r)* + (l-r) 2 • 2/ — — 4/* 4
And residue at z = a where *
2zr log (1 —iz\

= Lim Z—>a
(|+f,,{ + TTr),"+')
2 7
rl°gIT7 1 , 2
4r 4 Iog 1 +r‘
From (2), we get
f°° 2xr Jog (1 —ix) J 2rrir, 2 , .1
J-~ (l+r;*+(l-r)*}(l+ArJ) rfAr- 4 L og I + I- ~ log 2 J
= -?T Io£
f
°° — 2*r tan-1 x
—O {AT* ([+!■)*+(1 -r)*} (1 +JC*)<fa
. -rr
2 ,0g ( +,')-
Putting *=tan we get
f*/2 _2rO tan 5.sec* 0 itf_- iop n _l \
J —k/2 {tan2 5 (1 -j-r^-Hl —r)1} (l -+■ tan2 5) 2 8( + 9
f
«/2 2r sin 0 cos 5 a jq 17 , /1 , %
J-,r/2sin” tf (l + r)*+(I—r)* cos2 V

*/2 /-sin 25 „ w ,
_tt/2 1 - 2r cos :T+7» e "-5 log 11 + r)
2 log (,+r)
J — -
( 7T/2
f^'2 r sin 25 , it
01 J0
o 1 — 2r cos 20-fr* ° ‘“’“t '°g (1 + f>'
Similarly we may consider the case when r > 1 or < — I.
5. Pro ve that
r°° 77 V 3
^Jo *8-l“ 6 *
Consider Jc ^rz7\ dz = \c f<z)

where C is the contour as
shown in the figure. Denote
the radii of small semi-circles s'* "^vT*
with centres at — 1 and 1 by Pj / 's.
and p2 respectively. iR
-R
Then by residue theorem * O l x
|c f(z) <*z=f r f(z) ^z+/_^+P,) /<*) dx+jy1 f^dz
n a \
+ Jf1-n-Pt) f(x) dx~*~ Jy
f 2 -/r(z) dz+ Ji+p.
ff. dx
= 2 ni£R+ ...(1)
A+A+A 4~ A "F A+ As = 2niZR+.
| Rieu0. Riei9 r* R6
f /?5
Now
1 7* 1 ^ J0 | R*e*i0-l d ^ Jo *‘-l
Rl7T
= X*-l-+° as ^00*
. _ . (z-f I) zif--. 01
Since Limz_>_, (z+1) /(z) = Lim._>_1 ^ _ p [Form
_ . z*4-(z-t-l)4z3
— Li£n2_>_ i /-_5
lh.-» fvi /w
[see Theorem 1 of § 6*5J.
Similarly, since Lim^, (z-\)f(z)~\ , we have
Lim p2-+0 fy fWdz=\ i (0—) —^ ,

so that AFA->o
and Limpôpôôo [A + A + /J = |_oo f (x) dx
+ | /(*)</x+j^ / U) dx
= P |_^ /(x) dx.
Hence, we get from (1),
IL /« dX'=2iri£R+.
Now the simple poles of/(z) are given by

...(2)
z=exp (2niri/6), //=(), 1, 2, 3, 4. 5

of which exp (ni/3) and exp (2W/3) lie within C.
Denoting any of these by a,
. z4 a4 1
residue of/(z) at z=a is Limzâ ---= ^6 = ^•
<?z 2
Sum of the residues —& [exp (— «i/3) + exp ( - 2W/3)]
= « [exp ( - ni/3) - exp (w//3)J
2/ . * V3
6 sm 3~ ~ ~6 '•
From (2)» p J_oo /(*) dx=2iri(-^ 0=^

f°° 77
or /(*)</*=£ \/3.
S//I 7TJC
6. Prove that dx=TT.
0
j‘ A (J-X2) (Agra 1951, 58)
Consider }c rffriT) *”/c /(2) dz'

of a large serai-circle in the upper
half-plane indented at z= —1,0,1
as shown in the figure. Within
and on the contour C, f{z) is
regular and so by Cauchy’s theorem,
Jc / (*) dz-\rf & dz+jZR+Pl)f(x) dx+\yx f(z) dz

f_?2
+ J-(l-Pi) / (X) dx+ f f(z) dz+ P PV (-V)
J'/2 JPa
+ f f(z)dz+ [f /«</*=(). ...(1)

Jy3 Jl + Pa
Now dz I <
j
JQ | (l-R*eu*)
v exp ( -trR sin 0).dd
dd
fo (^-1)
0 as R-+oo
since 0 < 0 < tt, sin 0 is positive.
LimPl_»o J7 /<-) (0-rr), since LimM-i (z+l)/(*) —i;
L*mPa->0 jy2 /(’) (0-77).! = -77/,
since Limô z/(z)= 1
f , 77/
and LimPa_»o I /(z) </z=/ (0 — 77). §= —
J Vs
since Lim-^j (z — 1) /(z)« *
Hence as p,, p2, p3->0 and R-+00 , we get from (1),
0 + |~l f (at) r/.v - J77/ + J° / (a) dx - 77/4- / (a) dx
-i7ri+Ji /(x) </x=0

Calculus of Residues 3 23
j”oo f(X) dX=27,i-

f" Si1 % dx == 2tt

J _oo X (1 -
-X~)
*2)
f00 sin ttx ,

J0
[V /(-*)=/(*)]•
x-sin jf
7. Evaluate dx (a > 0).
'* - Jo ^(^+A'2)-'*
r z-i+ieix f
Consider j c 2a (o2+2») dz=)C f [Z) dz
along the contour as shown in the
figure. /
Now lc f (2> \r^(z)dz

-pop
+ fix) dx+fy:/(z)dz
+ J* f(x) dx=2ni ZR+. ...(1)

Simple poles of /(z) are at z=±ai, 0 of which z=ai lies
within C.
[Note that z = 0 is a simple pole and not a pole of third ordert
as may be seen by expanding the numerator.]
Residue at z — ai is given by
.. , .. z-Z-f/V5
z-*ai
z-+ai (z ~ al) zi
,3 (Z-j-ai) lz _ a,f
ai-i-\-it~
- a*i. 2ai
Now jr/U) dz <

J0 I
Rei0 — /-f/.exp IiRe‘°)
We*'9 (a'+RW)
Cn R4-1 4-exp ( — R sin 0)
Jo R3 (R*-a>)
0 as
pti0 —/4 ...)

-V Lim P->0
r <?2'0 (fl2 + pV<0)
(-4
P*eai» + ...) (id
— Linip^Q ^ eue {a*+p*e2i»)
I*0 d6 tt_
= * Jtt flr = ”2o-*

• • as /?->oo , p->0, we get from (1)
i
a: — i-\-ieiz tt 7t
or
f -co *»3 (<!» + **) dx = 2~2
/r,t i ^

f x x — sin x
.V -n
it
-co X* ((P + X*)dX~Q*â ~a+l~e )
oo
x- sin x , 77 /i 2
or
1
0 X* {a* + x2)dx'2a*{ia *+1 e 1
fV /(-*)=/(*)]
8. Prove that if a > 0.
_ f°° COS AC , TT sin a

(i) •
(ii) />r f* <*-<>.•

J - oo a2 - A-
f e<2
Consider 1 -r,az — f(z) dz.
JC c^-z2 lc
where C is the contour consisting of a large semi-circle in the
upper half plane indented at r = o and
z= — a as shown. Denote the radii of
small semi-circles as yx and y2 at the
points of indentation by px and p2
respectively.
Within and on C, / (z) is regular
and so by Cauchy’s theorem,
Jc /(’) <Iz=\r /(2> J_^+Pl) /<*) dx+fyi d2
+n;fl) /<*> *+jyi /w *+jr+P. /uwv=°-

By Jordan’s lemma, Lim^^ f^, /(z) </z = 0.
Putting zss—a + pxe<9 on ylt we get

f r, \ i f° exp [ — ia + ptiei0J . ,» ,a
ri“ Ha - n>) p',e de
-sH «p(-ia+^“'-('+ê,‘) dd
=h\l *“a<{exp «"+•••)<#
i fo _
= -x- e
“"2a J* * 2a
Similarly j*y f (z) dz-+^ itia.
Hence when /?-*■» , pj-»-0, p2-»- 0, we get from (1),

f—a iri fa _ f00
J _oo /(*) ^~2aC"<a+J /(*) <*f+“2 fe<a+Ja / W dx=0.
7T
f(x)dx=-2aHeia-e-°<)
n.
Df €*• J n .
or
p\-„ 5T^5^=T,,na-

o f°° cos at w .
^ J-«,S^rfjr=T s,no
f00 sin x .
and P I -=-7, dx=0.
J_oo a* - X*
Exercise
1. Prove that
00
sin nix t 7t ne-ma (m-f 2/a)
1 0 x(x*+a*f ax~2a*~ 4^
where m and n are real and positive. (Agra 1961)
,miz
Consider dz where f consists of semi-circle
r Sr z (2*+a2)*
,75
in the upper half plane together with the real axis indented
at the origin J.
2. Show that, if a and m are positive.

OO
sin* mx 77
/ . *» (a' + *-85* (2"w+3, + 4ma -3-

|_glmiz
[ Consider
Sr z2 (z2-{-a2)2
dz where f is the same as in
Ex. 1 above.J
3. Prove that when a > 0,

_ f°° r cos ax , tt
0) ^Jo -2 -7T dx=~2 Sin ar'
OO
(11) oJ
P s,n °x
—S- j
^ dx= n
^ cos ar.
x2 - r&
6*12. Integrals of many valued functions.

In the case of integrals involving many-valued functions such
as log z, za, where a is not an integer, we should use only those
contours whose interiors do not contain any branch points and
particular branches should also be specified. The method will
be clear by means of the following examples.
1. Prove that J ^ dx=^ sec y, ( — 1 < b < 1).
Consider f
. ** g dz=[ f (z) dz, where C is the contour
J c 1+ z2 JC
consisting of a large semi-circle in
the upper half plane indented at
the origin as shown in the figure.
Here we have avoided the branch
point 0, of zb by indenting the
origin.
The only simple pole of J (z)
within C is at z=i.
ib
The residue (at r = /) = Limz_>.i (z-i) / (z)=y.
—
e"2 —
I1 /
I
irb
7rb ., . . nb
n/>p I
7rb\
■ nt r»
2» 227/ (cos j + 'slnr>

Hence by Residue Theorem,
Jc / W *- f* f (v) dx+jrf (z) rfz + j^ f (xe") e‘” dx
+Jy f (z) dzt=Ti( cos T+'■sin t)

since z=xe<T on — ve’real axis.
, . . 7t£~]
“w[
cos 2 +i sin y . ..(1)
AbeM
Now | Jy/(z) <fe| < | .Rie" dO
V+R*eu9
<=1 o ^-\
RWn
0 as R-*>o since —1 0 as /?—►() since b+ 1 > 0.
(
Hence when p-+ 0 and R-+oo, we get from (1),
°°
f (x) dx+ I
f°
f(xeiv) (-dx)=ir{ cos
/ irb
y+/ sin
rrb\
yj
I oo . r°°
«•<» ,<*6 ^6
-<*6 / rrb . *b\
JC = ff^COS 2 +'Sin yj.
o i+?*+J. r+^<
Equating real parts.
OO irb
j-j-^ (1-fCOS Trb) dx = n cos y
0
00
i rr V ft 7T nb
or
i0 +**
dx=
2 cos (7r^/2) 2 SCw 2
2. Prove by contour integration
and deduce that

/: /o# (1 -\-xl) dx=„
, .
/og 2,
f1 I°Z(X+1J*) d”log 2. (Agr. 1952)
We have already shown that

log;.(!+**) dx=n log 2 (see ex 15 g 6.10)- ...(i)
f 0 * “T *
To deduce the second integral, we see that
dx
1+**
■°g (*+£) dx
,og (*+i) dx. ...(2)
1+X* OO 1+*1
, we get
i°g (*+J
[‘ 106 G'+;)/ in
L
J0 T+7F W~V
Jog (f-f-1//) dt
-i:
Therefore from (2), we get
o I+r*
= -/.
j_f°° Jog (*+ 1/*)

Jo i h-jc2
or Jog (1 +AT2) — Jog AT

'■*j: 1-f
= £7rlog2-*Jo ~~ dx from (1). ...(3)

/*» log *
Now to evaluate rxrr dx we consider the integral
o l+*!
log z
i.c l + z*dz=lc f(z)dz'
where C is the contour of example 1 above. Here we indent the
origin to avoid the only singularity (branch point) on the real
axis.
Now poles of/(r) are at z—±i of which the pole at z=i lies
inside C. ,
Residue of f(z) (at z=i) = Lim (z-i)./(z)
z—>/
=Limz->- =f, >°g '

1 ITT 7T_
==2/‘ • 2 = 4~#
Hence by residue theorem,
jc dz==\* ^{x) dx+\r dz+\*Rf(reiw)ei* dr

•f|y /(*) dz=2iri.^ = j i ...(4)
r:
or A 4- A+/3+A — 2” *•
-°f 23e22l Rie't putting z=R&*

•Al <
j o | /?*.*«•+! A,C
log /?-f 1 id
R dd
c R2-1
R+d
“Jo R R£-1
7r log R4- i71-2
Ra-1
__ _?L f\r I
-R*-l 'l R '2R
0 as R-+oc, since Lira12|_?=q

A
and r4|< J° M°e (pe»)

^Fe2« +1 p,e dd> putting z=pe‘#
|° (logp + 0)p
<
1 -p2
0 since LimP_*0 p log p=Lim. ->0 ,og p 1"=°

[5 1 IP
=Lim»-o ~jp=0.
Hence as p—>• 0 and /?->oc , we get from (4)
f/w
J00
dx +0
■fl /irei
n) dr+ 0 = ~ i
oo
or f°° log * ^ , f°° log (reiir) , .
J„0 l+x2<**
1+X + J00 l+rV”*^=2 1
oo
00 log X (log r+in) J ,
or --- dr=\TT*i.
f 1-fx2
dx+
\ \+r ...(5)

f°° 1logx
dx —0.
Jo r + x*
Hence (3) gives
1 log (x-f- 1/x)
i 0 1 +x
2 dx= ^ log 2.
00 log x dx v
3. Prove that (l+x»)*= ” ~4 » using QS a conlour a large
I
semi-circle in the upper half-plane indented at the origin.
jc +z*)* dz = Jc f{z)dz’
(1
where C is the contour as

shown in the figure.
By residue theorem, we
X
get
jc f (2) dz=\Rp f (x) d*+frf(z) dz
f(rei">eiirdr+J f (z) dz
= 277-/ £R+ (j)
or /i + /2 + /3 + /4 = 2 viZRi-.
Now |A j < log Rel°

Rie dd
I o | (l-[-R*eU0)
T log R+6
Jo (K°~i')z RM
0 as R->oo, since Limp.x = 0.

R
Similarly j JA | < J logp+0

(7=7F r*
->0 as p-r0, since Ump^0 P Jog P—0.
Hence as />->0 and R-+oo, we get from (1),

I f (x) dx-[ f{rtiv) dr=2ni ZR+.
Jo J oc
Now /(z) has double poles at z=±i of which z=i lies
inside C.
Residue of/(z) (at z=/)=0' (/)»
log z
where 0 (z) = ..(2;
(2-hi)2
or log 0 (z) = log log z-2 log (z-fi)
Differentiating this, we get
0* (z) 1 2
0 (Z)“Z log Z Z+r
,.(0=>£g;T >
— 4 |_z log i iJ
--jj p -1°« a
From (2), we get
r log_ dx i f” loe re<* ^_rri

Jo (l+jc’y^ +J (l + r*c«0lrf
0 2[_ ~2 J
or
f°° log x , f°° (log r+iV) ,
j.. 77 r. ml
Jo (1+^V^+J„ ^(7+75F
Equating real parts.
~ jgg* ,/v = -7r

2f (1 + ^)* 2
or r
J0 (1+JC*)*
dx=--
4-
/X
4. 5>> integrating row/jrf a suitable contour, prove that
I -\-x
(Luck. 1943, 47)

i: 'w
and
j: w
f c?+F)*-fc/(,)*-
where C is the same contour as in Ex. 3 above.
By residue theorem,
/ (z) dz = j^ f(x) flx bjr / «*) <fc+J'^ f{re{1t) e,TT dr
+j f (z) dz=2*i 2R+ .(I)
or li + Iz±I*+Ii=2iri ER+.
f* log Re")2 D im\ja
Now 1 /, I <
r | (log R+Wf I R do
Jo
<7f2“Ti][ [dog fl)* + G2+20 log R] dd
= j l" (log J?)* + |w3 + w* log R1
r r dog /?)*
_fl2-1 L 77 fl + 3-R- J
-►0 as fl-voo, since Lim^^
and Lim (jogjggaLim 2108 ^.(1/^,0.

I
(log p)*-M*4- 20 log p
Similarly 1 /4 ! ^ j 1-P*
P«0.
3 32 Complex Variables
since Limô p(logp)*=Lim„_,0 j70rm j]
=Limp_+0
=L'mP->0 -27^P[Form^]
= Llmf-+0 -I/p" =0‘
Hence as p->0 and R-+oo, we get form (1),
f
J0
/(at) </at- f° f(reiv) dr=2ni£R+.
J oo
..(2)
Now /(z) has simple poles at z=±i of which 2 = i lies
inside C.
Residue of/(z) (at 2=/)=Lim ^ (z-i)f(z)
r (log z)2
=L,m~' (F+7T
(jog/»2=i^Y- —
2i \ 2 /-25/
From (2), we get
f00 (log x)a (log re")* 7r
Jo (1+**) ** *+/o (l+r*e"<) ~ 4
f" (tog.*)' , f “ dog r+/V)» .
J0 !+*• *+J0 - 1+,, **•=—T.
Jo H--*2 J0 l-fA:2" 4
0 f°° (log at)2 j q T . *1°° 77
2 J„ tff *—* Ltan-1 *J0 =—4
^ f00 (log X)2 7r3 7T3 7T3
2 Jo ~T+^ ^=-4-+T=4
f00 (loe a:)2 7T8
Jo T+3F“ **-«-•
And equating imaginary parts, f°° J°iJ? dx=0

Jo 1 F*2
5. By integrating -- 7<*f * U) ■ round the contour of Ex. 3

or otherwise, prove that
[°° 2 cos x log x-\ tt sin x , , , , _
Jo -- dx=lire-* log 2.
ei: logf ~iz)
Consider
^V/UJ|VIV1 ■ „ , =JC /(-) dz, where C is the
JC 22 + 4
contour of Ex. 3. Then
[c fl2> /W dx+jr ftz) dz + jx dr

+ Jy f(z)dz-2niER+ .. (1)
A “I" E-\- f-\- R'r.
n°w ' v < j; |exp i

putting z=Reie on r
[* exp ( - R sin 6) | log (R sin B—iR cos B) | „ ,
Jo F
R- ^4
-4 Rde
I log R + i (},tt-yd) | exp ( - . sin 6) n

a A
Rz- 4
< ^^VexP(-.sin *) jj.

[v j lo8 *+/(s+*) < | log . | + |/ = log J
0 as R->oo j^V Lim^,, = °and Lim^^ exp ( - . sin ^)=0
sin 6 being positive when 0 ^ 6 < n J
Similarly |/4 | < J’ (jg£P +W2-f ^.exp (-/> sin 9) dQ
-*0 as p-+ 0 since Lim p log p=0.

P—>0
Hence we get from <1), when .->00,
|o /(*)</*f(rein) dr=2niZR+. ■(2)
Now /(z) has simple poles at z= ±2/ of which only z=2i

lies within C. The branch point at the origin has been avoided
by indentation.
Residue of f (z) (at z=2i) is Lim(z~20/(2)
e*2 log ( - iz)

= Lim-^2/ z-^2i
*47 e"2 loS 2*
Then (2) gives

f°° eff log (-/a) , , f00 exp (/re™) log (—ireiv) , 0
J0 a2 + 4 ^ + J0 r*«"‘+4 ^ 2 ,0g
°° *,x
/3t'X flog * —(iV/2)J
</a*
1 0 *2 + 4
+ r "" 3
CO
(cos a--}- / sin .v) (Jog x—(in/2)} ^
or
f a2+4
(cos r-/ sin r) (log r+'-infl)} , n
+J 0 r2+4 ~ 2 e 1 S *

2 cos a loe a+7t sin x , n
J -ta-.A-J.'lo,!.
-o
6. (i) integrating . r— ■->♦ ( - I < a < 1) round a large
± +-.Ti
semi-circular contour in the upper half-plane indented at the origin,
prove that
xa , _2tt j/j> (2an/3)
s0 A2 — A + 1 X~~y/3* sm an
Consider f Z.~—r ^-= f /U) ^/r, where C is the contour

J C -“+ 1 JC
of Ex. 3. Then
jc / (z) /(.v) rf* + |r /(z) d.
+ f(reiir) tin <Sr+Jy f(=) dz = 2ni 27R+. ...(I)
or 4+4+4+ 4=2rri zR+ •

If , I Rneia0 I
Now U =| Jr /<-) d- < Jo A272*e + Re«e + r7?/£,° I ^
f* K*x dd
<)Q Rz-R-V
[V | R*e™ + R.el°-(-1) I > I \-l> R2-R~\)
/2->-0 as R-+°o t since - 1 < a < 1.
f° P°fl /a
Similarly 1 h I < ]_ \~p-fdd
->0 as p—►O.
From (1), we then have as R->oo and p-*0,
f°° f(x)dx-\° f (reiV) dr=2rri ZR+. ...(2)

Jo J 00
Now /(z) has simple poles given by

— 1 + \/ 3 /
z2 + r-fl =0 or z =-^-*
Let a=-2+^3 , and/3=“i“'^'3/; °f lheSS tW° P°,CS

only a lies within C.
Residue of f (z) at z=x

,y . za (x)a
= Lim,_>a (2 -a)/(z) = Lim^_>a =a -/}
{ _^(4-(v/3/2) i}° (cos (2W3)-f i sin f27r/3))a

= V3» ” V3*
2ra . 2™~|
cos -=- + i sin
3 J*
-u
From (2), we get

C«o xa r00 ra.e,w,i
„ x‘+x+l dx+)<> r»e"'+re'>+l dr
2 nr 2-rra , . . 2?r /~1
=V3LC°S -r+,s,nTj
00 foo yCL (Si (a
or c/*+ dr
1 0 x2 + ;t+l M-l
2tt f 2?rd . . 2rr^"l
= — I cos - — + i sin — !.
V3L 3 3 J
r® sin an 2n . 2na
-— ar= — sin —-
j. r* - r+ 1
[°° *a . 2tt sin (2^/3)
or
Jo a2-*+1 Xt=V3 sin arr
(//) Apply the calculus of residues to evaluate
r°° (log x)'

dx (Agra 1965)
J0 x'+x + l
Equating imaginary and real parts in (3) of (i) above, we

get
f00 jx_ 2n_ sin (2^/3)

Jn -*2 —•*+! -v/3
V3 sin Q7T
qtt
cos an _ ?tt
2ira
and dx +
A2 + a+1 x2 -x+ 1 ax~~ 7/3
V3™ cos 3
2tt 2na27t cos a-n sin (2na>3)
CIX= COS
1+ V3 3 ~~ y/3 * sin fl7r
. In a
277^7 . 2tra
27TA 1
sin citt cos —— cos a-n sin -y-
-77
V3 ( Sin <77T
f°° x2 2tt sin (an/3)
i.e. ..(4)
J. r+7+
J0 Jr! rf-v= \/3
1+a+a‘ sin ATT '
Differentiating both sides of (4) w. r. to a twice and simplify¬

ing, we get
f°°
f x° (log
J?L ' J, •
a)2
Jo 1 T^+*dx
f°° Aa (log A' 2 . 47T3 r« . air - QTT .
or -7-7—-—s- dx=.7 3 cos y sin mr cos Ott
J0 14A-+-A- 9 \/3 sin3 att L9 sm j -
.. an . 1
—5 sin y sin2 an I
Now proceeding to the limits as a-*0, we get

(log a)2’ j, T
T •• 4-J.3
J7T3 fan
fo fa7T
fo a3n3
n' , ^
Jo 1 + a + a22 “A—Lima-»0 97/37a:i7r3\ 3 “27.3!+*‘V

'/wLimfl^09V3.fl;irr3L9V3 “27
3/ fl27T2 \/ 8r/37T3 \
"2 (, 18f + ' • • A “ “3T+ ■■■)

. fan a3n3 \ , , I
_ (y_ vX!+ •• ) (a'"
5 2 J
— Limô
9\/3.rJ [(
r 4
1 1, , -
~9V3l l8+6 + ““3
16 , lSv/3^
-77 *■
81V3 243
oo Xa-I _
7. (i) 0
y—
/+*
—
ATT
ro < ^ < 7;.
00 va—1
(ii) P f j— dx=rr cor an, (0 < a < 7J.
Jo 7 A
(Agra 1957, 62; Raj. 1953)
Consider
) 2o-l
^ i- dz
Cl-z
= |^/ (z) dzt where C is the
contour, as shown in the figure.

By Cauchy’s theroem, we get
Jc / (*) dz=jrf (z) /(re,w) </r+/ (2)
4-J* /(a) dx-(• |y^ / (z) </z + j^+^ /(^) </a = 0. ...(1)
J 1 + p. ' '
since z = reiv on — ve real axis.
Now dz I <:
fr/?°
0 as /?-ôo since a < 1.
R- 1
Similarly j / (zj rfz ^J j^ dd-+ 0 as p-*0,
and f f (z) dz = f dz= [ > ,, ,y* jq

Jy * Jy*l-z J* i-p»e##
putting z — 1 = pte*• on y2 in the above integral.
--j* 7 [1+0 (pa)J dO.
LimPtô [ f(z)dz=i\ dO^vi.

j y1 jo
Hence when p,-*-0 and R->oo, we get from (1),
0 r*~l exp {(a— 1) in) , „ fi xa~l r
°-i
. —r-,,« '*+''+(. E,*+■'+(
,+J. E </a=0
f°° A0-1 e0** f00 ro-l
. “TF7-^+J# ...(2)
Equating imaginary parts in (2),

00 r0-1 sin an
dx= 7r
j000 ^-1
1 +f
77
or
j0 i+x
dx =
sin an
(
And equating real parts,
eo ^-1
-- dx = cos an
roo 1
—— dx=cosan.
7r
= 77 COt
0 1-Jf Jo *+•* sin aw
Alternative. Use of double circle contour,

(i) The integral in (i) can also
fO— 1
be evaluated by integrating
1+r
round the contour consisting of a
large circle r with centre the origin R
and radius Ry a small circle y with
centre the origin and radius p joined
to the large circle r along the. real
axis from p to R by means of a cut
as shown in the figure. In this way we avoid the branch point
z = 0. It should be noted that z=x just above the cut and
z=rezlTi just below the cut.
Jr / (z) dz+ /(reswi) </r +/ (z> Jz+fp f (X) dx=l2niZR

...(I)
where HR denotes the sum of the residues inside the contour.
Now dz f Jj-» ^XP pQ,T 1-— Riei9 dd

( r/(z> Jo 1+ite"
2 TT ft*
dd
Jo X-i
2 nR°
0 as R->-o since a < 1.
R— I
Similarly | J / (z) dz ^| j* ~~ d0-+ 0 as p-*0
since a 0.
Hence when R->*> and p->0, we get from (I)

(o ra-i exp 2tt (a- 1) i . ^ , f°° x°-1
</x=2wj27/?*‘
°+l r+re5**—r/r+0+Jo T+i
roo -a-1 pivai xa~ 1
°r -J„ T+7“ dr + \„ i+i dx=2«iSR+ ...(2)

[V e~ivi= 1].
Now f (z) has a simple pole at z = — 1 = e7Ti which lies inside

the contour.
Residue off(z) (at 2=e‘w)=Lim._),e^' (r-e**) /(z)
= Lim_^« (z°-1j=exp {(a-1) "■/'}
= since e~iV= - 1.
■ 000 £a7r* — â—1
Hence from (2), dx — 7T
0 2/ 1 +*
cc Aa-1 77
or
j 1 -f- at sin utt‘
8. Prove that
°o x*-i
\ f+x2 dXs*\ C°SeC “J' <° < ° < 2)’

n j , f (lot* x)- dx it3
Deduce that —- - y z,
Jo * 1”x o
f°° *a“l-*6"1 dx . ( air I hn\

a"d J o -lifir • T+xt= 0g\ l°" ~4 l'an 4)
where 0 < b < 2.
Consider
wheie C is the contour as shown.

By residue theorem,
\c f (?) dz=\r f (z)dz
+ j/< €*ni dr
+ \y f f{x)dx*2niSR. ...(1)
Now I \r f(z) dz I < f7 I ^J;-1} dl)
Ra 2tt/?<* ^
I _ | as
since a < 2.
and (z) dz ^ | ^ </0->-O as p—>0 since a 0.

i' i
Hence when /?->oo and p tends to 0, we get from (1),
oo
f° r®-1 exp {2n (a — 1) i} , f
2
dx=2niSR
LOO
1 +r2e*vi l+x
oo 1 giiTai
f°° x°~l
or dx+ dx+2iriER
-
\+x2 Jo 1 +Vs
oo (| _ e1*ai) ^-1
or
l+x2
dx= liriER. ...(2)
[0
Now /(z; has simple poles given by z*= — 1
or z2=exp {(2«4-l) ni) or z=exp [{(2/7+1) 7t/}/2J, a=0, 1,
or z=£w*'2, e3ni'2
both of which lie inside C.
Let a denote any one of these; then a2= — 1.
z-1 a'0-1
Residue of f (z) at a = Lim,
-~+a d 2a
iu+*‘>
= *a°-2
Sum of the residues
= £ [(e^r-t+ie’1*"*)0-2]
= i fexp {tt/ (a/2 - l)}+exp {»/ (3a/2- 1)}]
= £ [ — exp (irai/2) — exp (3nai/2))
= — £ exp (irai) [exp ( — 7ria/2)+exp (irai/2)]
ira
= —exp (rria) cos y
From (2),
OO [1— exp..<2«rfj]^
/ -x cos y
r.exp (wai) ira
1 oo ,-o-l
l+x2
exp (irai) — exp ( - nai) â
—— -=ir cos —
or
I+* } 2/
ira 7ra
00 X°-l
IT COS y
Tt
COS y
or r=
/
dx
sin Tra — 7ra ’
cos
l+X na
2~
sin
y y
x—' ira
Hence
j: l+x:
j = ^ cosec
2 *
...(3)
Differentiating (3) twice with respect to a, we get

f°° (log x)* , _tt3 J"sin3 (ira/2)+2 sin (rra/2) cos1 (ira/2)l
a:0-1
J, “l + x* *Y" 8 L sin4 (na/ 2) J
Putting a=l in this, we get

00 (log x>*
i
l + xn- ax~ 8 *
Again integrating (3) w. r. t. a, we get
Tra
f ,—;—«-o~i- dx=\o° tan
0 (l+x'r log x 0
-r-.
4
Similarly j* 0 (1 + x~) 1°Tx“log ,an T-

OO
f
xa-l _ xb-l v a
0 (T+x2) log * dx=lo& ,an T “ og ,an 4

tan (rra/4)
~~ tan (77bj4)'
9. Prove that
00 ^-1
2-77 /2t7£7 + 77
v2 ■ v_i_ i dx= yr cos 1
f 0 x‘-r\4-i \/ 3 \
) cosec 77£7 (0 < a < 2).
Consider
f, rS5.-lc
where C consists of
(i) a large circle r, | z |=2? :
(ii) the radius vector 0^2*7 ;
(iii) a small circle y, | z l = p ;
and (iv) the radius vector 0 = 0.
By Cauchy’s residue theorem,

L /(*)*+I" exp(2W fa —1)}
J/ J /? I -j-re2w‘4-r «4W< ,e /r
•0—1
»a-l
+L/w*+f* 2 dx=x2niZR
fr l+z + z dz +
fp ra-l_ exp (2iria)
JX 1 + r+r2 *4-[y j
za-i
+z+T‘!h
0—1
+L”i +*+.*2
...(1)
Now lfr /(*) *1I < J0HIJ ^1 -fRtx01<*e'10 U< do
Ra
i 0 R*-R-\
do
2nRa n
/?*_y? — 1 as ^_>0° since a < 2
and |Jy /<*)) *

fe | ^ f[ r=£=? d°
il,/u
—>0 as p->0, since a > 0.
Hence when p->0 and R-+ oo, we get from (1),
„ • v»n f° *a-l.exp (2-nia) J , f60
27,,2/?“Jc i+^-&+J0 I+^'a
f00 [1 - exp (2w/a)J dx
a*®-1
...(2)
= Jo I+* + *a
Now / (2) has simple poles given by
22 + z+ 1=0
or (z-l) (z2 + z+l) = 0 or r3-l=0
or 2=exp (2/iw//3)f n = 0, 1,2
of which 2 = exp (2ir//3), exp (4ni/3) are the roots of 22+2+1=0.
Let a = exp (27r//3)=-1~f^/— and /3 = exp (4W/3) =-^
Both a and /? lie within C.

Residue of /(z) at a=Limz_>a (z-a)/(2)
-0-1 .0-1
aa-l
Sum of the residue= ^+ 1
a—p p— a
=o—
- vTi- [exP{t («-»*}-«P {y <a-J)'}]

=exp [e*p { '•(«-1)}-“p{s '' («-»}]
= 2_ex£îo) s,n
2 /7T 77 77*\
= “v"j cxp cos {2 + 3 a _ 3j
= --3exp (ma) cos^—— J.

V3 ' '
From (2),
, . 4rri . (iTra + TT
r .LL- exp (2™>J ^
.-1 Jx==-_ exp (ff/fl) cos ( —g—
xa-\
X (™) ('2^
Jn 1+A- + .X- v3 \ 6
I” t-Pi^^P^ cos (2--+S

a-l 2 77 277^4-77
r-*-r, (lx = -75 COSeC 77<7 COS
Jo H- A + A“ V 3
10. Prove that

’°° xa dx 7T (1 -O)
i; (1 +A'V 4 cos îra

, (- 1 < a < 3).
Consider (\-\-ziy^z = \c Nvêre *s the same

contour as in Ex. 9 above. Then by residue theorem, we get
Jc f(z) dz=ff f(z) dz+ f{r.exp(27ri,}.e2iri dr
+ {y /(*)<'*+J* f(x)dx = 2*iZR,... (1)
I \r“:
[2*
) dz
Ra+1
<
r Ra exp (aid)
—-2 Riei0
[l + R* exp (2/0))
do
5$ I —r.2 dO
2ttR*+' n
,—o->U as R—► oo , since a 3.
~iR*-l)*
Similarly |j f(z)dz
(0
J 2- (T^?p
p«-t 1
as p~>0'since
« + 1 > o.
Hence when />-►() and /?-*•<* , we get from (I),
2 niZR=[° rfjxpOnai) f
o (1+.xV</jc
(l—exp (7irai)} xa
dx
-fo 0+**)*
0 ra.exp (2nai)
[■•■ i 00 (14-
J. 1ha-v dx\-
Now f(z) has double poles given by
z2= ”1=exP "/} or r=exp {i {2n-\- 1) n), n—0, I.
Let a=exp {(w/)/2} = /, /3 = exp {(3w/)/2} =
Residue of/(z) at a = <// (a), where </> (’) = 2a
whence '/*' <z)

a 2
•l>(z)=z 2-/3*
.*. <// («) = *!_ f? 2 ~|

(a-/3)*L« a-pj*
Similarly <!>' (£)«=—^_f 5? 2 "1
W-a.)- (j-a J*
sum of the residues

a
= -^[exp {(*-1) yj+exp Jfa-1) yjj
[exp {(ani) 12}—exp {{3irai)/2}]

— O/
= — ^ exp {(a—1) w/} £exp | —(a- 1) yJ + exp|(«-l) y|J
-f exp (ani) [exp {-(ani)J2}—exp {(ani)j2}\
== — exp {(a - I) tti).2 cos ^-y /.exp (an*/) sin y
^ exp (a?r/) ^ sin y — sin yj=| exP (fl7T/) (a—s*n ~2
[\*. exp (-"/)=-! and cos7r)=sin yj.
Therefore from (2),

f°° 111 expdx=2iri.$a~l) {exp (air/)} sin y.
J (1 ^
exn — exp ( — -nai) f°° _3l_=i (l_fl sin ^
2/ Jo U+*3)2 2
r°° ^ _l (]
(sin t)
n) ^ 1 -
or Jo OH-x2)* 4 k ' sm rra
_t* (1 —a)
“4 * cos \ira’
11. Prove that if — 1 < p < 1 and 0 < A < tr,

f00 xp </x w ji/i p\
r,
0 I +2.v cos Ax2 sin pn sin A*
x~v dx * sin p\ (Agra 43; 44)
( f
I0 ] 4- 2x cos A -j- x1 sin A sin pn *
(i) Consider jc ]+2z col A+z» dz=*\c /(z) ** Where C

is the contour of ex. 9 above. By residue theorem,
L /u)*+|r /(*>*+ / (r.exp (2tt/)} exp (2tt/) </r

/?
+J / (r) / (x) </x—2*i2R.
Now f(z) has simple poles given by

z*+2z cos A+1=0 or z2 + z (>x, + e_x') +1 =0
or (z+eXl) (z + e~u) = Q
or z = —ell> -e~x'.
both of which lie within the contour since 0 < A < n.
Residue of f(z) (at z= - exi) = Li m _ ex' (c-f ^ )/U)
==C(z + e-x')]z= -ex'

(~$vePu ( - l )p <?/»>'
— ex,4-e—Xl —2/ sin A
.• / _ 1 )»> p—Pî
Similarly residue (at z= -e A )= ———:—-— .
J 21 sin A
Sum of the residues= - 1. [epyj - e~Pf l)

// sin a
Dsin pX ep7r< sin p\ . .
= -(-D sin A (V ew,= -l].
sin A
i r f!w I y?p pp9i Rieie
L/w </r <
NoW | {y /(z) * < f o | <Rem + e*> I ^
r2n np+l
<{0 (/?-!) (/?-l)rf®
(v
[V I|1 /?e'°- ( - eti) l ^ | Re® | -| - e* |=R - 1 etc.]
. If x j ^ 2ttRp+1
.. I JL/» *
rf<2) dz ^ (7? - J )*
-►0 as R-+oo since p+ 1 < 2.
Similarly | jy / <*) *| < ,-gp^
->0 as p—>0 sin />-+ 1 >0.

Hence when R-+ oo and p->0, we get from (1),
(0 f<» r *]
Jc* f(re2iri) <*'+] /(*) ^==27r/[__ihTA sin J
f°° rpg2»P« f°° XP_
Jo0 iH-2re'Z7r<
+ 2re2vi cos A-f.
A + r2e4w‘ r+ J 0 1 -f
+2 2.v cos A + **
a
_ 2«f .pni sin p\
~~ “sin A
r=° (l-e^)x^ ,2 rri

or
In I -+-2x cos X+x2 a—
sin Ai eT 1 sin p\
>TTiP __ yy-TTiP C CO
xp ir sin pX
or r.
2/ Jo 1 4-2a; cos A+A2=“sinT~
or _*P_ </v=_ *• s»n />A
Jo 1 + 2ar cos AA + *2
,x2 ' sin A sin pn
mr'
(ii) Proceed as in (i) above by integrating
_ Z~*
i -\-2z cos A-f zl
round the same contour. This is left as an exercise for [he
students.
12. Prove that, ij 0 < a < 3, b > 0, r > 0,
(i) f
Jo
*a_1 sin (£?r0—bx)
Ar--frs
bTra~-e~bT
‘ '
f30 dx
i) P\ :xa~1 sin (Wa-bx)
Oi) = rra~2 cos (hra—br)
J0 rz-x~
_ . f -0—1 r
Gons.der jc ^ *-Jc /(*) <fe.

of a large semi-circle in the upper
half plane indented at the origin.
By the residue theorem, we cet
fc/(*)*-/£ S(x)*+Jr
-h ) tiir dx+ /y/(c)d.- = 2rr/ 277?^
.(1)
owl ) dz
^ fr ; .ft*-1 exp {(a - 1) id) exp (r&R.e*10) Riei0

^Joi R-eii0-\-r‘ dd
-<
f" /?a exp (-/>/? sin 0) , 0 r/8/g° cxp(-2bR0/n)
Jo *** “J0 /?2 + rs ^
y?° 7T
~ R- -r£'bR (1 ~~e by Jordan's inequality
-►0 as /?->» since a < 3.
Similarly j f /(r) </r ■<" [° fi° exP sin 0)

Jy i ^ J :t - r2
—'0 as r>-»0 since a > 0.
Hence when R->x>, p->-0. we get from (1),
f30 -1 a0-1 exp {(/»— 1) iir}.e~ibx

dx dx = 2niZR±
Jo -r 00
x2 + r
â-1 e—ibxj
clx=2TTiZRJr.
or
( o ** + '•
...(2)
Now /(z) has simple poles at z=±/r of which only

z = ir = reinl2 lies inside C.
/ _<i_l
" -0-1 6t ibz
Residue off (z) (at z = rein/2) is
f^+r')\z=re,«,>
ra~l exp {wi fo — 1 )/2}.e~0r

= Yir
= - 2Tr ,a_1 CXP Kwa//2W • *•' e~illlS= ~ i.

r *a~l ie»*-€***.€-»•) 2tt,• „ ,
From (2). -zr-rzi-dx = - — r0-1 t' TTfll £—t>r
J 00 x* + r* 2r
or r
50 exp {(hm-bx) /} - exp {-(\c,TT — hx) /} x*-i
21 **+r
dx
= \ r°~2 e-*r
f» x
V« —1 sin (oir/2—xb) it
J 0
“•*—2 1 e T' ênce the result
(ii) Consider jc dz = \c /(z) d:<
where C is the contour consisting of a

large semi-circle in the upper half plane
together with the part of the real axis
indented at z = 0, r and ~r, as shown in
the llgure. Within and on this contour
J (z* is re8ular and so by Cauchy's Theorem, we get
fc f(Z) dz = \r f(z)dz+\+Pl f(xein)ei”dx+ j* f(z)dz

. ”" j yi
-f-
J* p* /(xe<W) e<* *+fy# A*) *+£"* / (*) r/.v
+!r3 fiz)dz+\Rr+n3 '<*>*-0. .(1)

By Jordan’s Lemma, f{z) dz=*0.
Putting z= —r-\- pxei0 on y„ we get
f f(z) dz
__ f° (-^+Pie,0;a-; exp + dO
J Pi**9 (2r-p1e*9)
• f° (-r)0-1 (\-Pleiolr)a^ tribr.exp (ibPle^)_dd
Jtt 2r {1 —(Pi/2r) ei0}
= Yr (-*-*£ e~ttr{l+0(Px)}de
-2 j" e-ibr 40 Where pj->0

2 (-1)0"1 ''
——2 (—I)0'1 ra-2.nc~ibr
_. __ ga7ri i&r
Similarly ra-2 ibr

L /{z) *=t •
and f
J >2
/ (z) dz- f° ^:l
J-
eXp {,a7 1)
r* — p2*e*'e
P (ibp>-eV Q'ie" d6
,2
=pi r«““ {i+o (*)}<«

r Jn
-►0 as p*—>0 since a > 0.
Hence as plt p2, p3->0 and 7?->-oo , we get from (I),
-f f(xe")dx+iZ rQ~2 gani> e~ibr - j* / (xtiV) clx
4 -Jo /(«*) <**+

fr fir
2" ra~2 e,tr4-J f°°
/(.V) </*=0
| f(xein) dx-\ J f(xeiir)dx + j f (x) dx
4
Jr
art
- f °° f{*)dx=-% r°-*{e
+*
&—ibr \ aibr
4- e”r}.
£ {f(x)+f(x*ilT)}dx4-J°° {/(-v)4-/(^}r/.v
= _ ~ ,a-2 +
OO
or
I {/(*)+/(**<*)} dx
_ _ ^
ITT
^(i^ra/2) â-2 ^(a^r/2-6r) i ^ e-{(nm-bT)
f°° p9-1., .x^-i. exp {ijt (0 — 1)}. e~ibt

or
Jo L '■a-*2 + ] dx
= — y r*-2. exp (ina/2). 2 cos - br ^

r oo
or -a—l .exp M2)p !:f (.^-^-exp {/ (»a/2 -6x)}j ^
J„
= - y /■a~26iTaJ2 x 2 COS
e?-*f
x0-1 sin (7rtf/2 -6x)
or
13.
J r* -x-
/Yovc that, if —1 < a
<^x = ^ r®"2 cos - 6r^.
I *'2 l-r cos 26

1 -2r-^T2i>+7‘ (,an °> dB
,
JL+is?‘2»> ,r™Vd»
[0 l+2r cos 20 + r* ' '
1 < r < /,
= 4 coj ±aw{/+0 + r) }' $ ~
7T
4 cos ârr or > I.
Deduce that, if —I < a J.
rvi 2 ftr/2
0^° </0 = I (cot0)nd6=--.
JO JO 2 COS $<J7T
Consider jr *-L /<*) *.
where C is the contour con¬

sisting of a large semi¬
circle in the upper half¬
plane indented at the origin.
By residue theorem.
+ J* f (*) dx=2iri £R+. ..(1)
Now fw|/?*«?«• (l+r) + (l-r) Raea«

frf(z) dz . Rie" dO
J 0 R*e'* (1 +/■)*-{-( 1 -rfRW-f-1
R- (l+r) + (l-r) R“+1

j
o R2 (l+r)2-(l-r/*.R2- 1
0 as /?->oo, since — 1 < a < I.
dd
Similarly j* j /(r) dz | ^ j"

* p2 (1 + r) + (l — r) p^1 dd
(1 - r)2 - (1 + r)2p2 * 1 - p2
0 as p—>0, since a + 1 >0.
Hence when R->x> and p->0, we get from (1),
_J° f(xeiV) </.x + J f (x) dx=2ni ZR+
or f°° *2 (l + r)+(l — r) _g!L (H-e,7ro) dx = 2rri ZR+. .. .(1)

Jn *2(l+r/+<l-r)2'l + **1
First we consider — 1 < r < 1.
Now/(z) has simple poles at z = /.
and z=+r Mf -1 < f < 1 and 2= i if r > 1 or <1

1 —r l-rr
. . . z2(I+r) + (l-0
Residue (at r = z) is Limz^./ ^{+r^q:^f_r)«*(Z+i)
-d+r) + (l-r) (eiV,2)a

[V 1W*'2]
-d + r)2 + (l-r)a‘ 21
_L
~~4i
and residue ^ at r = — £ z^ is
z2 (14-r) + (l-r) ( za
1 -rS’Kl+z*
1+r)/’+;—;<•) r=i-Hr /
1+r
-(^_rr)" + (l -r) ({-^7)° exP {(lV/2) a}

<l-r)2
1 -
d+^-2 hrr' (1+r)*
) exp.{(/V/2) fl}.
'°° a:2 (1 + r) + (l —r) A:a_

or
j a:2 (l + r/+(l - r)2* 1 + a:2
dx
CO +1
=l[i+G+>)]-
x3 (l+r) + d-r) _I
or
1
l+.x2'x* (1+r/-Hi -r>*
_ir
4 7ra
cos y
t »(&:)•]
Putting x-tan 0. 1= f ' 1 C°*2f (tan 6)° dd

Jo 1 —2r cos 2d-\-rl ’
■na
.
4 cos -J
170 {'♦(&)■]■
whence putting r =0, we get
rrri 2 »r/2
1
(tan 0)a dd (COt 0)° r/0—
J0 -i 0 2 77/
COS y
Similarly the second part may be proved.

14. Prove that
00 *1/6
„ J2i-£ dx = 2TT-—
! (1+*)2 \/3'
1,8 log z
Consider
Jf c -
(1 +Z)3
dz
= fc/(z) *•
where C is the contour as shown. Here
we see that origin is a branch point and
- 1 =e,7r is a pole of order two of /(z).
The branch point does not lie inside C.
We shall consider the branch of
log z=log | z | + id,
for which 0 lies between 0 and 2rr.
Residue of/(z) (at z=e'*)=<£' (e**) where <f> (z)=z»'• log r
Now <// (z)=J6 *-«»« logt z+

-r^z1'0 . 1-
r— —g
6?6/6 .
z
log ein + 6 in+ 6
'/*'(-!) =
6 (cO7j5/0 =ge(67r</tj)
=6^+in) (cos sin
o-
residue (at z=e") is _i2 (6+,V) (V3 + /).
Now w| pyfe1'6*9 log ,7?e"

dz < dd
i (1 -+■ Rei0)2
R7le I (log R+id) |
dd
I (R-D2
(lwRyic [log R + 6] , n n
Jo (R-l)* dU-*~0 as R-+oc
R7‘* log R rlog R

[ since Limyjôo
,. log R r .
—LlID^^00/?»/«(l — !//?)*
MR .. *
— Lim ^5/a — Ltm'/?-»oo *> p—1/6 — Lim ^ 1 -ol
* ^®76'"UJ *
C- -1 .
Similarly [if,/w
I / (z) dz Lf°
| < pm (log P+ft) dO
n. (1-P)*
log P
■0 as p—►oo since Limp _*o p7,e log p=Lim
P ’ — p ->0 P’7/8
1 Ip
=LimP-»o =Lim ^ -^,l5=s0-
By residue theorem, we have
jc f(z) dz=ff f(z) dz+jPR f(re"‘) e" i, /<■> dz

+ f(x) dx=2ni£R+. ...d)
Hence when p-*0 and R-+«o, we get from (1)
. . f« r1/a e<™8> [log r+2ir/] , f°° *»Mog
O + J^-+Jo (1+x)
= - 2»/. iXa (6+*”■) (V 3 + 0
f“ (J + /V3/2) (log jc+2iti) , r” X1'" log * ,v
or
-J0-(H^)5 +So 0+*>!
= -^- (6+i>)(V3 + »).
w- r ^
»d /: <iS>
(V37r + 6)
and 6V3+”)
2 o
whence -7=5 [- 12+2v/3tt] or 7=2*r-

o V3
f00 ^1/6 7T
and
Mo (T+x?^3-
Exercises
1. Show that
v* log a:
2. Prove that, if r > 0, p > 0, - 1 < k < 2,

,0° -vA' cos </\r- 4tt/->
clx = lTTrk~1~pr.
/,o *' -t- r*
3. By integrating a suitable function of z, round a circular
contour cut aiong the positive real axis or otherwise, prove
that f 00 loa x _
lQg* j. _ v
J0 ip* i-v/3*
2 a—l
4. By integrating round a large semi-circle in the upper
half-plane indented at the origin, prove that

/2rr/7-f-Tr\
r°° xp-i j cos ( — J
(i) | --K 6 '
Jo 1 + A -j- \/3 sin 77/?
/ 2 77/7 4 77 \
(ii)
f00
fJo —
yM
,’-1_ , _ 2tt s,nHr-)
l~-vr sin 77/?
where 0 0, q > 0 and 0 < 2, prove that
j: *“-*<"-osp* - Ct -- r* exp «•-)
[Integrate z— exp (e<»> (z’+ q'-) round the large semi-circle

in the upper half-plane indented at the origin].
6 13. Rectangular and other contours. Now we illustrate the

use of rectangular and other contours by means of the following
examples. ®
1. Prove that when 0 < a < 1,

30 /<*-!
J "/—. dt
l-Tt
=77 cosec <377.
Here w! ei y me use ot "

sc tf ***?' ^
rectangular ^
contour. 8 6',2)'
Putting /*»*•, We have
</at.
+ I
ow consider |c ^ + J dz~\c f (r) dz, where C is the rec¬

tangle ^CZ) with vertices at 7?. J?+2W, -S+ZW, - S. Here
R and S are positive (see the

figure). Clearly / (z) has simple C (-S,2tt) B(R.2a}
poles given by
ez= -l=exp {(2w+l) «■/}
or z=(2n +1) ni. OC-S.o) JA
n=0, + 1, +2, ±3,... CR»°) .v
The only pole inside the rectangle is z=ni.
Residue at 2=77-/is —--J
exp (am) _
exp (air/).
fc j (Z) *“flM / (Z) *+[« f (Z) *+j*C / (2) *

+Jcz> f (z) dz=~2ni exp (fl7r/
J/?
/(A) rfx+J
f2v f-S
/(R+iy) / 4y+/(x+2iri) dx
+ f _ / ( — £+(y) * 4y= - 2iri exp (mri)

J 2..
[V ^=x along TM, z=R+/y along /4R, z=x+2W along
BC, z——S+iy along CD\
r/? gM f2ir
2ir exp (a
{a (R+iy)} . , f—3 enz.exp (2ira/)
C—S eflz.exp (2iraf)
J_5ex+1 + J0o {1+exp (R+iy)} y+)R exp(a+2w/)+l
(R + iy)} 1 y~*~)R exp(a+2ir/)+l 6
f° exp {a(-*S'+/»} , . . .
r-rfys- 2ir/ exp (uiri)
+J 2je i+«FF~s+»
2a 1+exp (-£+/» dy=~ 2”‘exp { >
R {1 - exp (2a7r/)} e*x r2T (exp (aR+iay)

* ex+l *+/. Tl+exp(R+/y) 7
f2,r exp (-aS+/uy) . j „ . nVi /lx
-J0 1 + exp (-S+,,) ' -2m^‘ -(1>
2,r exp(aR+iy) ..
?+*» id \ [** q°R eiv
Now
1 +exp (R+iy)dy\< Jo -ir-T-1 dy
e .elv + l
f2* eaR
1
- 2neaR
^-*-0 as R —► oo since a < 1
eR — 1
and I r ™P(-°S+iy) . , , f- «—S

I Jo 1 -f exp ( — 5+ /» >|^J0 j'^T
2ne~aS
0 as »S ► ®c since a > 0,
l-e-S
Hence as R-*oo and S->oo, we get frora (1)
i — edx— — 2rricalJ<
*
— OO <?*+!
tto* e**
or
r.
oo 2/ l+ex
or r°° ea* TT
oo l+e1 sin qtt'
2. Integrating round the rectangle with vertices at
±7r, ±7r + i7?, /;rov<? that if a > 1.

a* j/m x , , / 1\
Jo (/ + J.
w the value of this integral when 0 < a < 7.
Consider f -—£—- dz
J C 1 - ae~is
= \c f(z) dz.
- -
Now f (z) has simple poles given by
e<I=a = exp (Jog a+2nnj)

0F z== ~’i a+2mr, n= 0, ± I, +2 etc.
C °-< * < ,*>thcre is on,y one pole z= —/ log a within

and there is no pole inside C when a > 1
Residue of / (r) (at z=-/Jog a)
_ — / log a
a/ exp (—log a)
log a)
= -Iog a.
Hence by residue theorem,
f / <z) dz= [" f(x) dx+ \X f(n+iy).i dy

J ^ J —*r jo
+ J~” f (x+iR) dx+j°R f (-*+/» i dy
= 2ni (—log a)
or /i + /2 + /3-+ /4= -2rri log a. (1)
Now /3 | < f *._1 r+iR 1-dx
J 7t | 1— a exp (— ix-t-Rt |
-i*|+L£| ^
J* | a exp ( - ix+R) |- 1
0 as .

J TT —l
, _ f* x dx [° x dx fn x dx
1 J — TT 1 — ue~u~~J -T. 1 —
0 1
Putting x = -t in the first integral, we get
/ _ f° 1 dt fK x dx
1 J 7t
w 1 -atu + Jo \-at,a
-I7T0 X fl - oe-ix~ \ - ae**\ d*
-f 0
n
x (ae~ix—Cciz)
1 —u (eix -te~,x)-ra2
—2/.r.7 sin x ,
,
1 — la cos x-f-u* Xm
Lim/2 + /4)=Jo yzra

__
(* + />) (-” + iy) ] f. dy
exp ( - iir+y) “ 1 - a exp (wi-fjOJ
=r [t&]-*
00 2rre~v
i dy
-j e y-t a
00
= - 2777 f log
= Ini {log (I-fa)— log a}=2*ri log .
We get from (I)

OO
—2ixa sin x
( 1 — 2a cos Ar-j-a
dx+2ni log ^1-f ^ = —2rri log a
when 0
35
or f xa sin x j , ,
Jo 1 — 2a cos dx=7T lo8 0 +«)
and when a > I, we get
roo ^ •
I - 2ixa sin x / I \
Jo l-~acosx+a* dx=~îlog +
since/(z) is regular within C when a < J,
f50 *** sin x , / j\
J„ *=-iog(i+').
Ex. 3. Apply the calculus of residues to prove that
r cosh ax l a
J0 cosh rrx UX~2 sec 2 *“w < a < n‘ (Agra 1958
COnSider \c coiFTz *-Jc /(*> *■

where C is the rectangle with vertices at -R, R, R+, and ~R+i
f(z) has simple poles given by
cosh ttz=0
irz—O (-R-vi) 1
(R+TJ
e**= - e“Vi
(-R)
(R)
This gives
gives r-^+O
z-~ 2 lt• n=l0t ±u ±2
Of these poles only zJ ,ies inside c
Residue ât z=^ — a3* eUat:)
| cosh ttz | i i77

U- J* = §/• * sinh-
e3"2 i
—j—- = — eai,t
in Sin hrr tti
By res.dueUieorem, we get
Jc /(*) * = /(AT) &+|‘ /(J» + M.,<fr+J-* /u +<) dx

eo J
or * tt/ • • •'A/
/i+/2 + /3 + /4 = 2ea</2.
Now |/t|= I «PiMV? + /»\ . , I
1 Jo cosh tt </?+/,,) 7dy I
^ f .—— 2fo/?|^n/ut.
exP {”■ (Kt/>;)-+-exp ( -77 (R+iyfiJ
leaR
_
r. (e*R
2eaR dy
- e~T'R)
0 as R->oo since — tt < a < tt.

.tt R -7zR
e**— —e
In the same way /4 can be shown to be zero.
Hence when R->oo, we get from (1).
r dx+ «P iff* dx=2elin

J -OO cosh 7TX JM cosh rr (* + /)
CO ,or roo or
or -f g
Ij 7TA:
— oo COSil J — oo —cosh W.x
[*.' cosh (7TA:-t-7n)=cos (7r — inx) =—cos i^x= - cosh
foo (l+g0<)£°* dx=^2e1,Ua.
or
J — CO cosh 77-.Y
(ea</2 + e-a</2)
or
r. CO cosh 7rA
,.az 1
eax dx=2
or
r. OO cosh tfA'
oo
cos (a/2)
eat 1
,ax
...(2)
or
t oo cosh ttx dxJr f
Jo cosh ttx
dx=
cos \a
f
Putting ,x= -/ in the first integral,
o eat ro c-at f co ,-ax
cosh ttx
dx.
— co COSh TTX ’ J OO cosh TTt J
■f
oo £-ox eax 1
From (2) d \ —• — — —
cosh ttx ’ cos

00 cosh ax jx__ _ 2 = $ sec \a.
or
4.
fcosh 7rx 2 cos
By integrating ez* round the rectangle whose vertices are
0, R, R+ ia. show that
n e~x\ ccs (2ax) dx=W”e~a'
and J*00 <?-** sin {2ax) dx = e~°a ^ e* dx.
Consider e~z* dz — / (z) dz.
where C is the rectangle with y

(Aa) B(R+«4
vertices at the points 0, R, C
R+ia, ia.
Since f (z) has no poles
within and on this contour.
we have by Cauchy’s theorem.
Jc f(z) = /(*) </-x'+Jo f(R+*y) 1 d>,Jr j‘ Rf(x+ia) dx
+J f(iy)i dy=0 ...(I)
or A A —0.
A
-if: exp {-(/?+/>)*} i dy | < |
f o
exp ( - R2+y* -2Ryi) i dy
I
^[ e R~ ey~ dy
Jo
-*■0 as R->x.
Hence when R-+ <», we get from (1)
r OO
exp {— (jr-f/Vi)*} dx-ij eyt dy=0
\/ 7r
or — e°3 I e~x*.(cos 2ax — i sin 2ax) dx — i j eyi dy= 0
Jo Jo
or
2
ea* |
j OO
0
e x1 (cos 2a x - i sin 2ax) dx—^-^r—i f ey'~ dy.
Jo
ea% [ 00 e — X* J V”
* cos 2ax dx=- —
Jo
or
f.-'- cos 2ax dx= e~a'
and [ °° _V »
e ■* si — n*
sin 2ox dx=e~° eyt
dy=e~a f e*3 dx.
Jo J0
5. Integrating e z* round the rectangle with vertices at
±R> ±R + ia, prove that
J* co e*p {-(Jf+flO*) dx=[°_ Me-Xt dx = V 7T
Deduce (i) [°°f?-*s cos 2ax dx^%^~e~at-

Jo 2
(H) {~(A'4-cj*} dx=\/nt where c is any constant.

Consider
y
(-R+VQ] (R+va*
u c
\c exp (-z2) dz==fc dz*
where C is the rectangle as shown.
Since / (z) is regular on and Al[-R) O B(R) T
within C, we have by Cauchy’s
theorem,
\c f(2) dz=\ AB f (z) dz + \Bcf(z) d:
+L f(z)
dZ+fDA f (z) dz=0
or
\-Rf{x) dx+\[ fW+Widy
+1^ f(x+ia)dx+J f( — R+iy)idy=0. ...((1)
Now | £/(* + *) * dy | < fJexP {-(*+0)2}. i dy ...(2)

< [ exp (y* — Rs) dy-*0 as R-+<x>.
J 0
ro
Similarly, J / ( — /?-+-/» i dy->0 as R-+*>.
Hence when R-+oo, we get from (])

OC
exp ( exp { - (x+ia)*} dx=0
oo -x'-)dx-\-~
or |*oo exP ( ~(x~\-ai)2} lix = J ~ ^ exp (-x2) dx = 2 J°°ex p ( - x”)dx
...(2)
To deduce the other results, we have from (2),
v-/:. exp {—(x+ai)*} dx
= \-oo exp (-(*+<*/)’}+exp {-(*+*/)*)</*
= ^ joo exp {—(-Jf+fl/J*} c/x + | 00 exp {-(* + <*/)*} dx

[changing x into -x in the first integral]
CO
-•* l
[exp (2aix)+exp ( -2aix)] dx
, .2 r°° 2
— 2ea e~x cos 2ax dx
Jo
_2 ~ . \/ 7T 2
£ * cos 2c7a: dx = -— e~G
Again putting ai=c in (2), we get
exp {—(Ar-j-r)2} dx- \/tt.

— oo
where c is any constant.

ciz
6. By integrating round the square with vertices
0, /?, /?+//?. //?, prove that

cos 50 \e~az
h-L. dx, (a > 0)
o J -f-X
f°° jr/'/i at f00

and ( 0).
Jo i+i*“J. 7+5**'
Consider luR) Bt
JC z-f-a
where C is the square as shown in the J
I
figure.
Since f {:) has no poles within and A(R)
on C, we have by Cauchy’s theorem.
\c fl«-\ OA f dz + jAB / (*> dz
\bc f[2> d: + fco f'z>d:=0

fJ0 / (*) dx+ fJo / (J? + iy) i dy+ °
JR
f f(x + iR) dx
* \R f(R + iy)idy ^ |£XP{/^-KV){ .j

Jo I R+iy+a 1
r0
« T- </)-*0 as /?-*• >3
R+a-y
and I f” / ix+ju,
/ (AT f/AT I <
-{-/?/) dx f0 | exp (Jx- RJ
JR
/? \ x-t ift + <?
X + i7f a
J 7? **&I-*Uoas *
R- x-a
Hence when /?-*oo, we get from (1),
f00 e** f°
I ——
~x~- /i </ya().
dy =» 0.
Jo * Jcocc 0
« +V
V
Putting y=ax in the second integral, we set
r — dx~ f 00 e~az. ia dx
a (1 -+■ ix)
=0
Jo X+a Jo
OO ix 00 / (1 -ix) e~ax dx
or
f x+a
dx
-f 1+x
Equating real and imaginary parts.
Jo *+* Jo ] +x
and r sit)~ <*= f" ^
Jo *+* Jo 1 + A
7 flv integrating 6Xp round the rectangle with vertices
J ° ° cosec nz
±R±lif show that, if 0 < a < *r,
^ dx= k caS ia
0
i
00
and
i sin (ax'> ^Va dX~i S'" ia‘
Consider Jc T&ET5T dz=Jc 1 (z) *■ where C is the

rectangle as shown. / (z) has (-R+i*) y (R<
simple poles given by D c
sinh 7tz=0
or e**-e-w‘=0 C-R) (R)
or evs=exp (-wz+2/nr/) O A*
or z=ni, n—0, ±1, ±2.

Only pole inside C is z=0. A ft
(-R-ji) (R- ix)

Residue (at z=0)
__ [ exp (iazz)
(sinh 7rz)
dz J *=o
{exp (/oz8)\ __ J
7T cosh 7TZJ — 7r'
By residue theorem,
fc/W*-J^ /(z)rfz
f (z) dz=2iri.~
+L DA J ’ 77
f/? fl/2
or I „/(*-¥) dx+ f(R + iy)idy
■* ~K
J—R
R
-
r-i/s
1/2
f{x + \i)dx + j ^ f { - R+ iy) i dy=2i ..(1)

(1/2 | 1*1/2
1,2 ! exp {/'a ■ R-\-iy)2} .
/ (/?+/» /</>> < / f/l'
-1/2 I J -112 I sinh 7T iR + <y)
(*1 / 2 exp {ia (/?-f />>*}
J -1/2 exp {it (R~\~iy)} —exp {-*■(/? + />•)} 2i* ' dy
f 1/2 2 Ray
<2 dy
J-l/ 2 e r.R _ —-R
[eRa —e—Ra
Ra )
_p-'*/£
->0 as R-*-oo since 0 < a < -
Similarly it can be shown that
/*—1/2
Lim/?-»oo fi — R+iyt i dy -O.
J 1/2
Hence when /?->>o , we get front (1»
f"exp {/a (x-*/)*} expf/afx f-A/)2} .
J _ ~7:~u—«*+ I
sinh-7—r —J a.V*2/
oo sinh tt (x- ki) tt ( x + h )
exp {ia (x2 - j)} ear _ r°° exn {ia <r ax

or dx dx—2i
i: OO — / cosh x
•50
oo COSl) /r.t
/ CO‘ J—
f50 exp {jg (a*- J) ^=2
or
J -oo cosh 77A‘
2 cosh ax .
n
^—(/a/4) ;ax2
or — e'aA J.v = 2
cosh 77X OO
cosh ax
or
00 cosh 77x
c
eia x*dx=eM*.
00 cosh ax a
_ cos ax1 dx—cos
1 — oo cosh ttx 4
or f°° cosh ax o.l a

Jo cosh~wx cos flx- dx — y cos 4
and
r
J0
cosh ax
îh^x
8. £>» integrating ze"* (I+*-’*)-’ w/iere 0 < a < 2. roa/i</

. 1 a
sin 4 .
the rectangle with vertices at z=~ ±R, =— ±R | rri, or otherwise

prove that
00
- dt — — COS
10 1+*2 4 j///2 (J,a77) '
Consider v
f za- r (-R+M) y
j c (T+V*) dz = \c /,z)^z»where
C is the rectangle as shown in the
figure. By residue theorem, A(-RJ
\cf(z) *-\AB /<z) dz+L *+ L /w *

+ jDA f(z)dz=2niZR+
or /(■*) f[R + iy)idy + f[x+ni)dx
+ |r / ( - R + iy). i dy=2iri27J?+.
Now | r
'Jo
/(/?+,» / dy j < f CTP.fel*±*j>
(i-fexp(2y?+2/»J
w (/? + y)
/ n ezR-i dy.
N°W exp {(2—a) exp {—aR)

[Form since 0 < a < 2]
_7T_
/?->» (2-o;exp {(2—a) = 0.
Hence [ f(R-h/y) i dy-+0 as 2? X .

J o
J" /(-■«+/» / </y=0.

Hence as , we get from (I)
n. ,5. *-*•*
or
i:.oo
at<°* (1 —ea1r*> — Tr/e^.e0^*
1 + e*x
Now/(r) has simple poles given by

dx=2rri£R+. ...(2)
e2*= - I=exp (2/7-f 1) rri

or Z=z\ (2/1+1) IT/, 71=0, ±1, ±2, etc.
Only pole inside C is at ar=$ir/.
•7
Residue of / (z) (at z=\ni)
z= 1/2 it/
).7Titalrilt 7T
ieaTin.
±ev* ~ 4
From (2),
OO
(I —<=<>*•) - . eair< 7T
- dx=— eaVi'\
i — oo i+£?*■*
(I — cos an) xe**4-neax sin an
A 77
•lx dx= — cos -1 an.
— OO \+e‘
dt
Putting ex=t, dx— —, we get
OO
(1 - COS an) ia los? t-f-ir sin anta dt
I l-w*
IT*
- = — cos - an
1
or 2 sin2 \an [ ‘ dl+n sin an f°° JUL dt

Jo l+i Jo iH-r*
n 1
= ~2 cos - an. ...(3)
and equating imaginary parts and putting <?*=/.
f* —sin an t~~x Jog t f°° ta~l n2 1
1+/ -,+pA-ySio
...(4)
Multiplying (3) by cos air and (4) by sin an and adding,
(^2 sin ? air cos air - sin2 an) ' dt
-?[ cos an cos J an-f sin an sin J an

1
or -2 sin* \an f°° ill 1
+ Io8 * dtf= - cos - an
Jo l+l*
or
Jo '"rtf
I” +tZ
1 +t *« - -
d ——
cin*
4 sin2 \na
e"
10. By integrating round a suitable contour, show
OO
that sin ay dy 1 .7 /
J = ? fo,A 2na~ 2a‘
Consider j*c (1z=\C ^(z) (in) (TT+tn)
where C is the contour as shown in the

figure. Within this contour / (2) is regular
so that we have by Cauchy’s theorem,
Jc / (2) f (x) dx+jy^ f (*) &
+ J" /(w+O') * dx
° /? A'
-f|Fl /(/»./+ f(z)dz=0 ...(1)
or A + A + A+A+A+A—O*
Now since Limz_»0 2/(2) = Lim-ô Z-u*Z\= "2/* WC haVC
Linifl_,o 4 = LimPi^0 J 2/)-4‘
Similarly, since Lim.^- (z-^)f (z)
. (2 -tt) e°*
— Limr_>- e_2lc _ |
= Lim^
ea-*+(2-7r) ae" <?°:
j^Form
a
— 2/ e -2t*5 — 2/
we have
Limpô A =LimPa_>0 |y2 Jr=/ (° " ”27)e

f-—Pi eax dx f* e°* dx_
A = LimP;_>0. p,->0 Jpi —J0
•jt eaz(\-eUx)dx
-j
frc gox (1 1 f* e“ (1 — e2,z)
If
</x,
J0 2 (1 - cos 2 a) 4J sin2 x
f/1 .... . . f" exp {a (*+iy)}idy
Lim /J->oo . Ps-X) J Pj /(*+«*) Jo exp {_2/ (tt+iy)} — 1
00 ear. eaiyi dy
f° ea
e (*+in)
-f 0 *2v “ 1
ân<
’
dx < </a
1=
J tt er« (x+tn) _ J (x+iM) | _ |
C° e°x dx
•0 as n—>00 .
J ein - 1
and f0 e°iy i dy f o eaiy i dy

Lim/?-»•» /5—Lim n_^
in «*'-l =J =o e*y-l *
Hence
- - we
-- - getk nv
from (1),
_ _ eax(\-eUx 00 (ea1T - 1) ieaiv ,
l(
4 i+o-H; sin2 a: 1 *+f 1—-
4 I j f~ -
~Jo 1 —
e<w (•—cos 2a:)
</a:
sin2 a:
+r(—
or
= ±(e°''-l)-1L(l+ean)'
2a
00
dv_ 7T
v l+ean
l+*aw1 ir Qtt ,1
I
4 r^*-2^=4coth~2-fa¬
ll. Integrating ~r~l (a real) round the rectangle of sides
AT-0, x=Rt y=0t y=\f indented at 0 and i, prove that

00 sin ax . 1 I i
r' 0 dx~4 COt ,2a~Ta'
Consider (c ?S^r*=[c /(*) <fe,

where C is the contour as shown in
the figure. Within the contour
/ (z) is regular and so by Cauchy’s
theorem, we get
f (z) dz
+ \Pi f(x)dx+ |o /(*+,»/f(x+i)dx
+ /ya </Z+f^p fVy)-idy=0. ...d)

or / , » Pt
/l + /2 + /34-/4 + /5-f-/„=0.
Now since Lim ~ w/-\ T • e0*1
nz-M) v (z)=Lim2_>0 7.
,2W*
«*"■- 1
<?a<‘-f m'zea<* 1
Lim
z^° lire™' 2tr *
we get Limô
(z -1) e°u e~*

and since Limz_>/ <z - /)/ (z.)=Lim v2_A =2w ’
_ in e~a i
we have LimPtô /5= - -y - — *
»-o
f1 exp {/c7 ((/?4-/v)) / </y i
Also I /
"Jo exp t2jr (R -tiy)) -l
C
c exp (2/^tt)—1 4k
0 as R-+x>.
Hence we get from (1),
Lim^„.Plô.ô[4H-J* /W*+JJ /«*+<>*

_ — *>“<* + fPl /(*» / a>] = 0
4 J 1—p* -I
or f“ /(x) rf.v4 J” /(*+0 dx+j° f(iy).dy
=-l (>+«->
00 ca* (*+<>
or
f0 exp (2-ttx) - 1
dx+ f
Joo
OO
£—<iy
exp {2tt (x-\-/)}-!
«/x
+r =
Ji exp (2w7y) —1
i</y=4 (1+c—)
CO /I --a-v.atr / f1 e~av.e~Viv . J
i: 11
e2**— 1
(l+e-)+J
^.-ay (cos ny — i sin ttv)
dy
- 4- (!+«-) +
j: 2 sin Try
dy.
Equating imaginary pats, we get

f00 sin"* J 1 1 f1
j. *-4 -T^F° J. T dy
1
4C0,h"2fl-27ii-O
1 t. 1 1
= - coth 2° 2a
12. Integrate eiat/sin/i z round suitably indented rectangle

with vertices at ±Rt ±R+iir and show that
rs!n «x a _j reav-j i
Jo Sinh * “2 W [_«*« +* J'
provided that the imaginary part of lies between — 1 and I.
Consider [ exP dz
J C sinh z (-R+7TI) CR+lA)
— /(z) </z, where C is the rect*
angle with vertices ± Rt ±R + in —I->_/vV _

° ^
/? (R) a
indented at the origin and z=iV as ^ 0 ^R)
Poles of f (z) are given by
sinh z=0
or -<?-*=0
or ^**=1 =exp (2/nri)
or z=niritn=0, ±1, ±2, ±3, etc.
None of these poles lie within the contour C and so by
Cauchy's theorem, we get
\c f(z) dz=\JRf{*) dx+\y /(*> dz+ JR fix) dx
+ f / (^ + iy). i dy-\- f f(x + in) dx

Jo JR
+ I f(z) dz+ f ~Rf (x + irr) dx

JTi J—P*
+ \nfl~R + iy) 1 dy = 0' .(1)

•••(!)
since Limô zf (z)=Limz_>0 -1, we get

J sinh z “*»
Lim Pi-*o[ / (z) 0 — 77) i. 1 = —77/

J Ti
Again since Lim***/ (z - ni)f(z)= we get
f (z) dz=—-rri ( — e~~x) = 77/e'TTi

and
I f~ /(^ + «» i dy I ^ ” I -XP {*«(* + /») ijy

IJu 1 Jo I sinh (/?+iy)
^ f" — e-*y dy_

Jo ! sinh R cos y-j-i cosh R sin y
^ f*__
Joo sinh R cos y
v—cosh itt.
—cosh R sin y Jy
u
71
2e~ax dy
/„ (eR — e R) cos y — (eR+e~R) sin y
2er*y dy
eR (cos y—sin y) -e-R (cos .y-f-sin y)
0 as R-r oo.
Similarly Lim^^ J° f(-R+iy)idy=0
Hence when /?-><* and px->0, p2->0, we get from (1),
f
J -0C
f(x)dx—ni+ f°° f{x) dx+ Jf
Jo oo
/(*+/*•)
+ir/ie-»«+ J °°/(A-f/7r)=0
<* exp f/a (A'-f/7T)} dx

or =77 i (1 — e-wr)
J_oo si nil a J- sinh (A-f nr)
or f“ e-xp. (?*x> dx+ f°° exp U«x±f?°dx=7,i(l_

J-oo sinh x J-oo sinb x
[7 sinh (A-f/V) = sinh a cos w-f-/ cosh a sin tt= - sinh aJ

j* OO (l-}_e--a
or exp (/ax) dx=ni (l — e **).
oo sinh a

00 1 4
f:
-
. sin cix—tt (] — e~â)

oo sinh a
foe si
sin ax dx=iT
or 2 (\+e~™) (1 - e_7rct)
sinh a
Jo 81
f<» si
sin aA 77(1 — e-Tra)_7T e~a—]
or dx =
J* sisinh a 12 +*-*«)“ 2'e»*+l
j>a2
13. (a) 2?y integrating ——^ round the rectangle with vertices
R, R+2iri, — R-\-2ni, —R indented at 0 2nit prove that
foo eax—exp (1—a) a , _ /A ^ ^ ..

-—r- dx—7T cot 77a (0 < a < 1).
Jo e 1
Consider fJc —, </z= fJc*'f (z) dz. where C is the contour

(r- 1
as shown.
Calculus of Residues 37 j
Poles off (z) are given by

**=1 =exp (2nni)
or z=2nm\n=0, ±1, ±2, etc.
Since none of these poles
lies within the contour, we have
by Cauchy’s theorem.
•«
Jc f{x)dx+ f f (z) dz+[R f (x) dx

C2n J 1 J P*
+ jV(/? + /v)i ^+|^ f(x+2rri) dx+ f /fz)*
-P, ^ ,x+2’"') dx + f 2n f<-R+iy).idy=*0

=0. ...fl)
Since LimM02/(*)=Lin>M0 2^-1. we have
Lim P.-+0 f / (Z) <fe— -IN. 1 =—7TI
and since LimMirf(z-2>r«)/(r|=UiD^2*f-^)—
[Form o]
exp (2wai)
“expT^T)
-r v- '/ =exp (2,ra')'
we have LimPi_0 J ^ /(z) -*/ exp (>«/,.
I f2r: i , #»->_
Again ; h f(i?+/y)/d> 277 exp fee </?+!»} jdv
J0 o' exp (/?-f iy) _ j ~
c 2« £,a/f
I
< 2neaR
eR ~ |
0- as Osince
JII v 0 < a 0 as /?-*« .
Hence as y?->oo, p,-*-0, p,-*0, we get from (1),
'-»> *-»/+£ /W*+J^/(jf+2w/)<fe-w/exp(2w«/)
+ | f{x+2tri) </*=0
J0
1°° / (x) dx- f(x+2iri) dx^ni [1 + exp (2jto/)]
r
J-oo^-i
*_r gxp (g ^+2^)}
exp (x+2iri) -
J _oo 1
. }
f {1—exp (2*r«i).<Q ^=7r/ {J +exp (2ira/)}
foo
J —oo
ax
•
f°°
i
, rri {1 + eXp (2irai)}
*+]. ?=1 -1-”* "
- — Wl{ .TTdi_e--X(H J
s= — ni.i cot 7ra=7r cot ir<x. ...(2)
Putting x— —y in the first integral, we get
L £r —L HT *■
f°° exp {(!-«) a 1 ——= =77rT cot ?ra
From (2), e*-l
f* egg—exp {(!-«) *}rfT_»

e»-l
f 00
.
SW/j aX
UnFnX d
, / tn„
2 tan 2*(
l rr (f)
")•
Jc £S *-Jc'<*’
where C is the rectangle with
vertices at /?,/?+/, -7?+/, -7? .R+£
R+ l
indented at z=*.
sinh air a f( *
Since L'm**o55fi^--f7W
-R
has no pole at the origin. Now
zeros of the denominator of / (z)
are given by sinh nz=0
or e**-e-»‘=0 or e^=l = e2nir<
or z—ni, n=0, ±1, ±2 etc.
Hence poles of /(z) are given by ±i, ±2i etc.
Since none of these poles lies within C, we get by Cauchy’s

theorem.
\c f(z) dz-\_RJ ix) dx-*[ f (R+iy) idy+[p f{x + i)dx

Jo JR
+j y f U)dz+\_( f(x+OJx+f° f(-R+iy)idy=v.
...(I)
Since LimM( (e-/,/W=Lim (,_«
Z^‘ sinh ttz
sinh a/ i
. = —— sin a,
W COSh ff/
we have Limôj^ /(2) A sin a)= _

sin a.
Now [ // </>,— |1 sinh g (/?+/>) / </>■

Jo Jo sinh 7r (/^-J-/>>
. f “**/ is.,ar8e. can write heaR for sinh xR and cosh xR
and .ze R for sinh rrR and cosh ttR.
Hence f1 s-™lgL(JLtlQJ * reduces

Jo sinh tt (R + iy) reauces 10
j*1 gq/? (cos ay 4- j sin ay) .
Jo L'VJi (cos vy+i sin try) 1 y'
which tends to 0 as R-*zc since a < tt.
Similarly, j /(-R+iy) i rfj,-o as R
Hence when 7?->oo, />-► 0, we get from (I),
J~» AW *+1 /(*+') </jr-sin a4- f"” /(*+/) A=0

or P sin dx~r sinh «(*+/)
s,nh w* J-oo sinh *■ (*+/) jA-s,n «
or r a* + sinh a (A-f n
sinh wx <7* = sin a
r°° sinh k i * sinh (~x-f 7t/)=.-sinh tt.v]

or I sinh aA+smh ax cos a-f/ cosh a.v sin a ,
J-cc sinh ttx rfA=sina.
r (N
(JL+cos_a) sinh ox
J —CO sinh ■nx' */x=sin a
sinh ax
or
/: _
°o sinh tt*
sin a
1+cos a
a
ta°2
00 sinh ax l a
or
I
o sinh wx A — 2 tan 2
Aliter. Consider
J c sinh ttz„ dz=\c f <z) dz>
X/\ZligiUVA I • |
where C is the rectangle with

corners at R.' R+i9 -R+i and VRtx) (Ri-t)
-R indented at 0 and i as
shown.
Poles off (z) are given by ~^-R) > o’ ' *—

ev*—e-"*=0,
or eZ7rz= 1 =eZnni
or z=ni, n—Oy ±1, ±2, etc.
Since none of these poles lies within C, we have by Cauchy’s

theorem,
\c /(*) dx=j f(z) dz+fYj f(z) *+\*/(x) dx
+/o /</?+/»./4v-ff(x+i)dx
+{y,/u) *+/:; /(jr+o *

+J /(-/?+!» i </y=0. ... (1)
Now since Limz_*0 r/ (z)= Lim^ , we get
LimPl->0 f(z) dz=(-rri)
Again since Limr-*/ (z - /) f (z)

ea< _ 1 ^
sinh 77Z —7r cosh m 7r ’
we get Linipô /(z) dz=(—iri)
and /(/?+/»./ rfy
f1 I exp {g f7?+;y)} . |
^ Jo 7T (R+iy)
J 0 I sinh it (R+iyj 1 ay
* I|
f1 _exp (aR) dy_
J o j sinh 7tR cos iry+i cosh ttR sin Try
exp (tf.K) dy
oo since a <. n.
Similarly, Lim/j_>x>j' f(-R+iy).i dy=0.
Hence when o, p,->0, p2->0, we get from (1).
j ^ f(x) dx-i+ | / far) dx+ J /(* + /) dx+icai
+ JQ /(■*+/) </.v=0
r.
>ax ea 1*4 O
or
00 sinh ttx
OO
dx
-i: OO sinh tt (x+i)
*r -f ear. eai
/ (I — eai)
dx=/ (1 — e0<)
— cc sinh ttx
[V sinh (TT.v-f 7r/)= - sinh tt.vJ
oo ar
e- . *(]-**» -i {et«<»> —g-(Q«*)} a
or
— oo sinh ttx ^ l+ea< e{at/2) _ê-(ail :) ' tan ^
I
or
I"—co sinh ttx
ax
dr+r i
J0 sinh ttx
.ax
r-r-— </x=tan ..
2
Putting x= -y in the first integral,

a
J a,oo — sinh Try
7rj; sinh^x<A
= tan
or
or
r o
eax
—r-
_ e-a.
sinh ttx
00 sinh ax 1
;rv— dx=-z tan
a
dx=tan -
2
a
I o sinh ttx 2 2
15 Apply the calculus of residues to prove dial

f°° coj// ax . ] a
-tt < r; < tt. (Agra 1958)
J0 cosh TTX dx=~
2 sec
cai
Consider dz = JC /(*) dz.
C Cosh 7T2 I
where C i:; the rectangle with
corners at RyR + \l, -rt+$/and
-R indented at \i. Within and
on this contour, f (z) is regular
and so by Cauchy’s Theorem,
we get
\c /(z) dz=\R-R f{x) dx+^
+J*£ /(*+$') /(r)</r+|~^/(jc+J/)<fe
+ji/2 fi-x+iy) idy=o. ...(i)
Since Umz_>i/2 (*-*/)/(z)=LimT_> ,• (—~ -*0 ^

Cosh 772
exp (q//2) exp(a//2)
_77 Sinh ^TT/- 77/ *
we have Limp->0 f (z) dz=*{-ni) , ~..exp (ai/2)= -exp(a//2),

1/2
exp {/z(.R + />)}
f {R + iv) i dy ' < J — — i dy ^
o , cosh n (/? + /» I
1/2
c0/? dy
0 cosh 77/? cos 7Tj’ —sinh ttR sin ny
[V | cosh 77-/? cos 7r_y-f / sinh irR sin ny
> cosh irR cos —sinh ttR sinh 77y>]
n/2_</3>
J0 (cos 77j — sin ttj)
fV R is iarge we can write evR for cosh ttR and sinh rrR]
->0 as R-+x, since a < n.
Similarly f /( - R-\-iy) i dy-*0 as R-*x>.

J1/2
Hence when R-*x and p-*0, we get from (I),
J ^ f{x)dx+^ /(*+$/) dx-ea{i* + j" f(x+j/) = 0
or f(x)dr~r f(x+ii) dx=<*•'*
_exp [a (x+Ji))l Jx=eam

or
/
-00 |_cosh 7rx cosh 7r (jr+tOJ
rr e** e°*.eai'2
e0* eai'z "1
or —r~ — u— dx—(cos \a + i sin ia).
J— 00 Lcosh rrx 1 Sinh ttXJ ~

â r poo
J" ^
~r-dx ^ - r sin \a.e**
//a = cos \a (i)
;h nx
00 cosh jtx J—,
j —00 sin 77a"
r 00
f °° p'*’ pnc
e9* ^ //
cos \a
and f/x=sin i/7
J —OO sinh 7tx
.CLX
oi
.. .{2)
/: oo sinh 7tx dx=\2Ln 2a.
From (I) and (2), we get
fIo<-^h^rfjr-sin*fltan odx
2fl = cos la
or
/•- dx=sec *“• (Raj. 1959)
00
e**
or
Llsars
.-oo cosh „„o cosh
-- ttx ‘''+1
dx=^sec \a.
Putting *= ->> in the first integral, we get
-r "• e'av r°"

cosh Try dy+ JI 0 C06h TTX
dx=sec \a
r
j 00
e^ + e-*^
or f/*=sec \a
J0 cosh ttx
or f 00 cosh ax , .
-—- </* = * sec \a.
Jo cosh TTX - 2
Similarly we shall get from (2),

f00 sinh ax , .
-r-r—— c/Ar = i tan ha.
J0 sinh ttx '
16. Prove that
OC
cos x2 dx
I, - j: * x,jx=ij(d-
(Lucknow 1945, 47)
Consider c-2* dz= \ cf{z) dz, B
where Cis the contour as shown. Within

and on this contour/(z) is regular and Ply
so by Cauchy’s theorem,
fc/(2' *“ L dz+ L/(z> * 0^-

+
f ao /,z) *=°
fo exp (r,enn) «■*"< Jr=0.. .(I)

Putting z=*Rei9 on F, we get
I r */i
r ***** exp ( - Rzeti9) Rieie | dO
J 0
>\. R exp (-R* cos 20) (18 = 1 f

j 0
exp ( - *» sin *) d*
Putting 20 = \n-2h
f*7s
J R exp {(— 2^!it) 7?2} d<J> by Jordan’s inequality
1 -e ]-->-0 as i? OO.
*T - *.
Hence when R-+ oe, we get from (1),
j;e-
j7 fcos sin r=)

(V2 +72‘)dr= f0 dx=~l-
y/TT
COS r2 dr + sin r2 dr— ...(2)
and J* cos r2 dr— J sin r3 dr=0 ...(3)
From (2) and (3), we get
| cos r2 dr= J sin r* dr = \J\-
17. By integrating e~=z over a contour consisting of

(/) the real axis from 0 to R;
(ii) the circle | z \— R, from 0 =0 to 0=a, when a ^ Jw,
(m) the line 0=a, from \z\=*R to 0,
prove that
J" exp (—xa cos 2a) cos (x~ sin 2a) dx— ^ cos a,
and f exp (— \? cos 2a) sin (x2 sin 2a) dx^^—sin a.
Deduce that cos A-2 dx = j°° sin x2 dx

si
Since within and on the contour f (z) is regular, we have, by

Cauchy’s theorem,
\cf(z)dz= \OAf(z)dz
\nofiz)dz-°
e-** dx+ | , f (z) dz
O
(re'a). eia dr=0.
t.> ...d)
N°W | \r f(z)dz\ = 11* exp R,eiad6
exp {—RW).R,e*\d6
Jo I
=Jo exp ( -Rl cos 20). fl rf6'->0 K-*x,

since cos 29 > 0.
Hence when R-+ oc , we get from (I),
f°° a f30
I e- dx- exp (—r:e~ix).eia dr=0
J0 Jo
or Jo exp {— r- (cos 2a+ / sin 2a)J.e,a <// = P » v/ TT
e dx— y~~
roo J°
or
exp { — r2 (cos 2a-j- / sin 2x)}.eia dr—^-^ e-*a
J o
y/TT ,
=~2 (cos a - i sin a)
Equating real and imaginary pails, we get
\o exp , -r= cos 2a).cos <r« sin V) c/r^” cos a,

roo
and J ^ exp (-r* cos 2a).sin (r* sin 2a) </r=^7 sin a.
a = 4* we get
X)
and
X)
IS. By in,egeing dong a suitable path. show ,hat
i: w m 00
«Vi a:
V*
dx
Consider Jc £ dz=^ /(„

dz
where C is the quadrant of a circle
indented at the origin as shown. Within
this contour / (z) is reguIar an(J SQ by
Cauchy’s theorem, we get
fc f (z) dz=l? dx+\r dz

+jfR f(re‘’"*).e"'*dr+ f f (z) dz=0. ...d)
Now
./wAkr^^*
J o i Rli2.eiel-
I* exp ( — R sin 0).Rm dd

f
Jo
ir/2
exp {-(IdIn) R}.&l*d0,
— —— (l-e~R )
2y/R ^ }
—>0 as R-+7o .
Similarly fJ y fiz)dz < fJ */2 exp (—p sin 0).pv* dd
< fir. pI"

since exp ( — p sin 8) < 1, when p is small
->0 as p~*~0t
Hence when R->oo, p->-0, we get from (1),
f°° fix) dx — f°° /(re**12) i dr=0

Jo Jo
00 e*
or [~ $ d*-t r
Jo y/x J0 Vr.eivl*
ie-ivl*
r e~r#r~1/2 dr
<72~‘-vd
Equating real and imaginary parts.
s
oo COS X oo
sin x
0
dx
Vx "" -I Jo Vx
dx
-JG>
19. By integrating e^.z*-1 round a quadrant of a circle of
radius Rt prove that when 0 < a < 1.
00 cos , cos 7TO
[ a:*"1 .
sin
x dx=r (a) .
sin 2
,
As in example 18, we have
/ (x) dx+^r f (2) dz+ /(reiv,2).el1Trj dr
H-|y f(z)dz—0. ...(1)

n£ 1
Now \rf^dz\
*12
exp {(//?)•«"} /P*”1.exp {fa - 1) 10} dd
TT/2
exp ( - /? sin 0)./?° </0
-i
exp [-(20/*) /?] dd.
J'-.R*-' (1 -e~R)
->0 as R-+oo, since a - 1 is negative
and J f (z) </z | < J°/2 Pa.exp ( —/? sin d) dd
< P' d0, since exp ( — p sin d) ^ 1,

when p is small
-*0 as p—>0 since a > 0.
Hence as /?-*-<» and p-v0, we get from (1),
exp [ir.e*<,2J ra"1.exp fi (fl-i).*/2].i dr=0

J 0°
j00 dx = exp ((fl7r/)/2] JJ ra-l.e~r dr^T(a) exp (a*»/2)]

cos cos 0*
«-» . xdx=r(a) . -y
sin sin 2
j .
20. By integrating Si round a suitable contour, prove that
Deduce
Consider
where C is the contour as shown. Within

and on this contour,/(z) is regular and
so by Cauchy's theorem, we have
Jc/(z)rfz=J* f{x)dx+^ff(z)dz
+ \r /(*)*-°
Now | Jj, f(z) dz | ^ j* o ^/al eiRx. eiie

Re“
Rie* dd
17/2
exp (■—/?* sin 29)
-i
77
->0 as oo, since sin 26 is 4-ve, when 0 < 9
2 *
f r/ \ j f° exp {ip*.ezi9} jd
and
\yf<Z)dz=U2 Je"--P'e dd
=/ f°,
Jn/2
[1 + 0 (p)] df>
ITT
2 ’
f
Hence when R^-oo and p->0f we get from (I)
°°
j(x)dx+ j
r°
y (/>) i dy—^ =u
ITT
oo
oo OO —IVs
or dx —— i dy=l~
i OO
-f n
,/** _ „-«»
^
177
2
or </x= ^
j
f°° 2i sin x* . ii7
or -dx—-^
Jo X 2
00 sin x2 J 77
..(2)
or
!0
Puttinc x2=r in (2), we get
x <^x==~4 *
f00
00 sin / I1 77
o ^7 * 2V' 4
or
21. 2?y integrating e~aszn~l

round a sector of the circle
z l = R from 9=0 to 9=a < \-n indented at the origin, prove when
a > > 0,
cos . ... _ ~ cos
J. xn_1 exp (—ax cos a) . (ax sin a.) dx—a n Tn

r ' sin sin
no..
Consider j*^ e~axrn_1 dz~fc / (z) ^2-
where C is the contour as shown. Within

and on this contour f (z) is regular and so
by Cauchy's theorem,
Jc / (?) dz==\Rp /<*) *

r • ^
+j£ f (reia) eia /(*) <&-0. (0

Now
rf(z)d2{ [ \ exp ( -aRe"). Rn~l. exp {(/7— 1) 16}. Rie*' j dd

Jo I |
< J exp (~aR cos ft).Rn dd
J0
•a Rn
f _ dd
J0 exp [aR cos ti)
R"
■0 as R->00, since Lim^^ = 0.
exp (aR cos d)
Similarly j f (z) dz j^ pn exp ( -ap cos 6) dd
-►0 as f>-*■() since // > 0.

Hence when R-*00 and p-*-0, we get from (1),
f°° f{x) dx+ f° f(rtia) eia dr= 0

^ Jo ’ OO
Jo dx-f exp ( - areia r"-1). exp {(// -1) a 1}. eia dr=0
fee °
exp { -ar (cos a+1 sin a)}.r"_1 dr
J0
f00 T/j
= exp(— #iai» I </*= —- (cos /ia-i sin «a).
Jo
r co t _
exp ( - ar cos a).r"-1 COS (or sin a) dr= — C0S («*).

J0 sin c/r* sin
, Exercises
Show that the simple poles of the function
f (z) — zRu — e~if)
are given by
z = i log a-\-2nn, « = 0, ±1, ±2, ...
By integrating f (z) round a rectangle with corners at ±nt
±n+ in, prove that
fw x sin x dx tt 14-a
Jo !+«*- 2a cos x
= a~ , if a > 1
TT
= ~ log (l+o) if a < 1,
By integrating log siD z round the rectangle with corners at
0, TTt n-\-jRt IR indented at a and w, prove that
J* log sin x dx= - n log 2.

3. By integrating (<a real and 0 < a < 1) round the rect-

l—ex
angle with corners at R. R+in, —R+in, — R indented at the
origin, prove that
[00 ecx
dx=ir cosec a-n,
and
f
J-oo
°°
-oo
-
l+e*
e**
--- dx=n cot air.
1 —ex
e*iz
4. By integrating —--L round the rectangle with
& 6 cosh z-cosh
vertices at R> R+iir, -R-\-in, — R, show that
00
cos ax 2n sinha(^-h)
dx=—
i — 00 cosh x-cos b sin h* sinh an
/•OO
I cos ax jX— S*P^
and
J— oo cosh x+cos b sin h‘sinh an
5. By integrating za~le~z round a sector indented at the origin,
prove that
cos
ix sin a)
ai fljc=i lai “s (fla).
i o
xa~1.exp i-x cos a) C?* {x
sia
dx=T(a)
sin
and provided that a > 0 and — ^ < 2* Show that this
formula still holds when a=±$n if 0 < a < 1.
6. By integrating rou°d an infinite quadrant, where a
and b are real and positive, prove that

00 cos bx ■n
dx=-^p e-*1 (sin ha+cos ba),
f o *4 + 4u4
°° .dx=-£-i er* (sin ha—cos ba).

i0 xi+4at 8a3
7. By taking a contour a square whose corners are at R, R+2Ri
-R+2Ri, -R and making i?-*cc, prove that
o (1+**)dX
cosh x;;=lo82-
fax
6*14. We were so far evaluating integrals with the help of
residues. We now illustrate by means of an example how residues
can be found by evaluating integrals.
Example. Find the residue of tan*-1 ttz at z=®£, where p is an
even positive integer.
Poles of tan nz are given by

/
z=±2, ±1 etc.
We enclose z=\ within the D £ ( 1 ■+■ i R 1
contour as shown in the figure. We (iR)

have by residue theorem,
the residue at {z = \) 0 \
’/2
-4. [i AB
f (z) dz
A B t$)
(oR)
+ \BC f (*) dz
f fz) r/zj ...(1)

/u) dz+\ DA
where /(z) = tan P_1
Since on BC, we have z=l-f/y and dz=i dy, so that
f tan p~l nz dz= f* ltanp-‘ w (1 + /y)J / </>'

J tfC J -/?
[tanp_l i»wy»] / dy.

-i —R
And on £M, z=iy
so that f tanp'1 nz d:= f R [tanp_1 miy)\ i dy
J DA JR
R
[tar.p_1 niy)J i dy.
-I -R
Hence
EC f
tanp_ 1 nz dz+ j
Therefore we get from (1),

J DA
tan1*-1 nz dz=0
the residue (at z= J)
tanp_1 nz dz-\—-—r f tanp_1 nz dz

AB Zni J CD
i,i JJ f°
r0
== — I tanp_1 n(x- iR) dx4- - — I tan”'1 n (x + iR) dx,
2tti J q 277/
2ni J J
Since z=x — iR on AB and z=x-riR on CD.

v, . ni 1 exp {in (x—//?!}-exp {-in (x—iR)}
Now tan (nx — in/<)=*- -■ — -—--7 :-7-.Di,
/ exp {/7T (x~/7?)} + exp {-m (x-iR\)
_ I I — exp f—2/w (x — iR))
i I 4-exp {— 2m (x—iR)}
_ I 1 - exp (- 2inx). exp ( - 2nR)
i J 4-exp ( — 2iirx).cxp {—2nR)
-*■- as R-yyo .
Similarly tan (nx+ivR)-* -1 asR-ôo.
Hence the residue (at z—h)
since ( l)p_1-1 ,p being an even positive interger
-L(v\v~l=l L
*7 V ) n ip
6*15. Expansion of a Meromorpbic Function. Let f (z)

he a function whose only singularities in the finite part of the plane
ore simple poles at ait with residues blt b2... respectively
and suppose that
0 < ! ax | < 1 | a2 | < j a3 | < ...
Consider a sequence of closed contours C„ C2, C3__ either
squares or circles such that
(i) Cn encloses alt o2,.. an and no other poles,
(ii) the minimum distance Rn of Cn from the origin tends to
infinity when n-*-x> .
("V the length Ln of Cn is O (Rn) f,
and(iv) f (z) = 0 (RJ on Cn. This will be satisfied iff (z) is
bounded on the sequence of contours.
Then for all values of z except poles,
_"='
„
/(*)=/«>)+11 (—L + I\
a„)
J The O and 0 noti^. If (4„) and („„) are two sequence, such-
.hat g is ultimately bounded, then a„ and *„ are said to be of the same
order when n is large and we write an=0 (bn).
Thus
(i) if Lim is constant, k ^0, we write a„=0 (bn),
(n) if Lim(^)=0, we write an=0 (b„).

Calculus of Residues 3S7
Suppose
’=.4- f
- ,C„
where z is a point within Cn. The poles of the integrand are
at i=am, m= 1, 2, 3,.. .n, £=0 and £=z.
Now the residue (at ^=am)=Limâm (5 - O
=-—-- since Lim:^,a -flfn)/(^)=^,„

Am ("m-2) " m
/(£» /(0)
The residue (at ^=0) = Linu_>0 f TIJ
. . .. . # , J • /(5> /(*)
and the residue tat ^ = z) = Lim^_>, (;-2) r
5 (5-*>
Hence by residue theorem, we have
m /(0) ,/CD
/= £ + ..(1)
m=l U?,* —
Nowi/i=! 4Jc„rffe)rf5
< J_ __n. Mn,
^ 'IrtRn • Rn-\z
where A/n is the maximum value of | /(*) I on Cn
—>0 as w-*oo .
Hence we get from (1),
n
b~ _/(0) ./(l^Q
Lim n-> oo 2-
m=l ^)
or /<«)-/«»+j[ 4.(ri5-+l).
1. Prove that
cosec z= z +2z n*n*—z*-
J z—sin z . ,
Consider the function/(z) = cosec z--- = - sin - (29*0) and
/(0)=0. Now simple poles of/(z) are given by sin z=0 or z=/m,
«= ± 1, ±2,...
Clearly there is no singularity at z=0.
.• 2 -sin z . . 1~0 - 31! 2,+ - - )

since Limz_>0 - gin - = Limô
z (z“3 ! 23+ *-)
Z- +
3 ! ' ***
= Lim
z~*° z%~ /rz4+”‘
The residue of f (z) (at z=mr)

(z—sin z) \
' "
— (z sin z)
dz ]z=trz
z-sin 2
sin z-\-z cos
nw-sin mr 1
sin nir+nir cos mr cos nn
Let Cn be the square with vertices at z = (n + \) (± 1 ±0 *•
Clearly - is bounded on these squares. Now we prove that
cosec z is also bounded. Consider separately the regions,

(i) > \tt, (ii)y < -i”, (iii) <y <
In (i) since y > K we have

_ 2_
cosec z = ~,iz _ p-»* >^/2_ *
[V iz-e-a |=| e-u—zis > i e~iz I—! M eV H e~v I

> evtt—e-v'*\.
Hence cosec z is bounded in this case.
A similar argument applies to the case (ii), since on writing
v= we get t > in-.
Incase (iii), let AB be the line joining the points \v±\iri.
Now | sin z |s*(cosh*.y-cos3 x)l,i. Then, since we get
on ABy \ sin z | = cosh y ^ 1, so that | cosec r | < 1.
Since cosec z has period w, it is bounded on all the lines
joining z=(n-\-h - ^ anc* - = («+i+i07r- It follows that
cosec z is bounded on all the squares Cn. We thus have
by §6-15,
cosec 2-1=* r (-1)"(—— + -\

^ ,,=-cc \z - mr J
the accent £ means that the term //=0 is omitted from the
summation.
Calculus of Residues 38'
Since the series is convergent when n > 0 or n < 0, we

may write the expansion, by adding the term, corresponding
to ±nt as follows :
1 „ £ ( - 1 )n~'
cosec z- - —2z 2 -J—.-
2 n=l " n
1 oo /
2. Prove that cot z=~-\-2z 2 —-—,.
„=1 z~-n-*-
_ . . , . .. . 1 r cos z —sir z
Consider the f unction / (z) cot z - - =
£ r sin r
Simple poles of/(z) are given by
_ . z cos z- sin z
Limz_>o z sin z
*—
Z(*~JT =*+■■■)
= Limz_>0
, 1 -
3 i2 +*-'
/(0) = 0.
Now residue (at : = mn)
z cos z - sin
-J- Z sill z) z = n\-

az
cos z - s \n_ z~I
-ftsin z+z cos zj z—m-
mrr cos mjr - sin
= 1
sin rmr + nm cos run
Let Cn be the square ABCD
with corners at c -,
D
z = (// + i) ( ± 1 ±0 n • n t (n + i)jc
The poles within this square 0
are at z--.±irt ±2.. , ±"7r
T> A
and there are no poles on the
square.
The minimum distance Rn of C„ from the origin is (/i-f-i) n
which tends to infinity when . The length of C„ is Xwr + 4n
and its ratio to Rn is 8 = a const.
. e° + e- u lit* + 1
Since cot Zts*i , we have along AB, where
*=(« + *)
390 Calculus of Residues
. exp {i (2nn+TT+2iy)}+\
cot 2 =
1 exp {/ (In-rr-^-ir-\-2iy)}— 1
• e~~2y-eXP O' (2/t7T + 7T)}+l
1 e_2y.exp {/ (2mr-t-ir)}— 1
-<r2*+I 1 -e~2v
< 1.
—e~2v - 1 l be-2v
Again on BC where >>—(«+£) ^r, we have

„ , | exp [2i{x+i (/*”-r *”•)}] +1
exp [2/ {x+i (nn+^)}J - 11
1 + exp {-(2/177 + 77)}
< ,-? -;—-r-*l as n->OO .
^ ] - exp { —(2n7r-\-7r)}
Similarly j cot z i~*l along DA.
Also - -*0 on Cn.

2
It follows that (cot z—l/z) is bounded on Cn for all
values of n.
Hence cot z-\ = S'(—

2 _«> V z-r.7T
-+—)
nn J
Accent on 2 indicates that the value n=0 is omitted from

summation. Adding together the terms corresponding to ±n.
we get
1 °° I
cot z= —2zE
22-n%2-
Exercises
Prove that
CO
sec 2 4tt 27 (-)» (2n+l)/ {<2zi+l)V-4z*}.

0
Show that
tan 2=2 27
i (n-*)2778 -z2*
3. Prove that
1 22
I 1 00
_L_ = _i4-i+ 27_
e2 — \ 2 2 i 22+4/22»t8
4. Prove that, if—77 < a < 77,
sin g2 _2 i)W n sin na.
Sin 772 77 2>-/22
cos na
f (_i).
Sin 772 TTZ TT n=\ 2»—n8
CHAPTER VII
POWER SERIES
7*1. A series of variable terms of the form Ecnzn or
Ean (z—a)n is called a power series where ant a and z are complex
numbers and an is independent of z. Clearly the second form of
the power series can be reduced to the first form by a simple
change of origin and consequently we shall use the first form in
our future discussions.
7*2. Absolute convergence of power series.
The series Eanzn will be absolutely convergent if the series
E | an | j z " is convergent. But the series £ \ an | | z |n is a
series of real positive terms. Hence all the tests for convergence
which are valid for the series of positive terms can be applied
to this series. Thus if we apply Cauchy's root test, we see that
the power series is absolutely convergent if Urn | an 1#n | z | is
less than unity. If we put Lim | an | 1,n=—; we see that the

K
power series Eanzn is absolutely convergent if | z | < Rand

divergent if | z | > R. If | z | we cannot say anything about
the convergence of this series. Its behaviour in this case may be
of the most diverse character
00
7*3. Theorem. If a power series 2/ anzn converges when

n=0
z = z0, it is absolutely convergent for every value of z such that

I z | < | z0 |.
Since the series E anz0n is convergent, each term of the series
must be bounded.
Hence | anzQn | < M (n=0, 1,2,....) where M is a positive
quantity.
Now let |4U k, where | z | < | z0 |; then

I zo I
anZn | =| anZ0n < MKn.
Now the series EMKn is convergent, being a geometric series

of common ratio k < 1. It follows from the comparison test
that the series.
2
ru=l
I V” I
is convergent and so the theorem is established.
7*4. Circle of convergence of a power series.
The number R defined as above is called the radius of conver¬
gence of the power series Sanzn and the circle whose centre is at
the origin and radius R is called the circle of convergence of the
power series.
Obviously, there are three possibilities to consider: (i) R=0,

lii) R is finite, (iii) R is infinite. In the first case, the series is
convergent only at the origin and so this case is not very impor¬
tant. In the second case the radius of convergence is finite and
the power series is absolutely convergent at all points within the
circle of convergence, and divergent at all points outside it. In this
case, no definite statement can be made about its convergence
at points of the circumference of the circle In fact, all conceiv¬
able types of behaviour are possible for points z on the circle
| z | = 7?.
In the third case, the series converges for all values ofz since
R is infinite in this case
7*5. Alternative formula for R. We have defined R, the
radius of convergence, by the formula
-^ = Lim \an | un.
But by Cauchy’s second theorem on limits, we have
Lim \an | »'"=Lim ^ , ...(1)
provided the limit on the right hand side of (1) exists.
Thus if Lim | exists, we have

t*n
Note. For the series San (2 - «)", the centre of the circle of
convergence is at z=a and the radius R is again given by
J- = Lim I an */".
Power Series 393

1. Find the radii of convergence of the following power
jer/es :
Zn
(0 00
^ • )" 7«
S'
(/«) 00 Z 2/;! * ’
(n+2)(/» + 3)** ’
(v) 27 (log n)n.zn. (W> 27 “U .

2'* 4-1
__ 1
(0 Herea„=-^- and fln+1 •
tl • («+D!
an . . (n+1) !
7?=Lim
*->00 <in+1_ ,,'>00 nl
= Lim,1_).oe(.T + l)= oo.
(ii) Here a«=(1 + •
• •
ft = *—“'/I-*
Lim/jv00ara1,n
oo“f» = Lim
—oo
o^y-
1
i.e. R=.
e
n+\ n+2
(iii) Here an= and an+l —
(« + 2)(m + 3> (n-\~3)(n + 4) ‘
a, (/f-HX'H-4) =1
Hence 7? = Lim„_,00 =Limn^,00 {n + 2)*
^fi+i
rw !)2 {<w + l> !)*
(iv) Here an
(2/i)T and fln+l (2/1+2)! ‘
(2/1+2) !
nln\..
y?=Lim >;-> oo — Limôo(2/^ , x l/I+l) , (/l+1)!
"n+t
_- . (2/;4-2)(2;/4-l) 4
-L,m«->oo («4-1)(«4- 1J
(v) Here £/„ = (log n)\
• •
(tffJ1<n = Lim„_>O0 log n= oo
so that /?=0.
(vi) Here and an+j—ji+i+T *

flfj . 2n+1+l
.. 7? = Lini/i_> qo —— — rj0 2"-\. 1
2-f-—
=Lim rt->co
1+2n
_2+_0
14-0
2. F/W the radii of convergence of the following power
series :—
(!) S(3 + 4iY 2", (U) Z z”.
('/Vi; Z(-^(z-2ir.
(i) Here 0„=(3+4/)»,

so that Lim | an |1,n=Lim j 34-4/ f = 5.
Hence
r.. TT ny/2+i
00 Here n~~\+2in*
(/ny/l + i) (1 - 2in)
so that I a» 1= 14-4«“
(24- y/2) n+i (1 - 2y/2/i8) j
14-4/22
= nl^V (2+V2)2«2+(l-2v/2«3)1]
= ]T^V[(6+4v/2) n= + l-4v/2/i2+8n‘J
V/(cS/244-6/22-{-1)
~ 14-4/2-
_ . , V[8 (/2-f 1)*4-6 (/i4-1 )2 + 1 ]

Then | o„+1 i- i+4
_ . • I a,
Hence A=Lim/;v:o
un*f 1 .
[1+4 (/2 + 1)2Jv'(8'24 + 6/22+!)
— Lm,/2->30 (1+4w-2)v/^ (*+l)* + 6 (/, + 1 + 1J
= 1.
(iii) Here the centre of the circle of convergence is at r—2i
and the radius of convergence R is given by
1 In
1
A = Lim | an |'»"=Limn^0O |
n
1
= Lim„_>.)0 -^ = 1 since Lim nl,n= 1.
Power Series 3*3
Hence R= 1.
3. Prove that
j.SlP■ a (a+1) b (b+1)

+ /.cZ+ J.2.c(c+1) Z+-*‘
has unit radius of convergence.
a (a +1) (tf + 2)...fa + n — 1) b (6 +1)...(6+//- 1)
Mere =-
1.2.3.. .n.c (c+ 1).. .\c+n- 1)
a {a+\)..Aa-*-n - 1) (a + n) b(b+ \ )...(h + n- 1) (b + n)
and a„+1 =
1.2.3...// (// + !).c (c+I)...(c+/i - 1) (c + //)
a //;+ 1) (c+/i)
7?=Lim //->00 — = Lim //—>00 = 1.
</„+! '' ~ (7/ + «)
4. .S/tow that the radius of convergence of the series
1 , 1.3 „ , 7.3.5 3 ,
2 Z+2.5 2 +2.5.S“ + *•*
ij I
_ 1.3. S... (2n - 1)
Here
*"_2.5.«...(2/;-l)
I.3.5.. .(2//-1) (2/?+h
and
°n+l 2.5.8...(3//- 1) (3/1+2)*
a 3/J + 2 3
Hence 7? = Lim^- = Lim
fln+1 — 2/1+1 2*
5. 7r//u7 the domains of convergence of the following series :
o> W(®)’ (ti) 27

n—1 ! (^y
» 7.3.5.. .(2/; - l) ( 1 -z
(i) Putzs=i;.
Then the series in £ is 27/r

cy-jr
(i+/)n*
n (/»+ n*
Now tfn+l 1 + /jn+l*
a"-(I+7T* and
a. n,2
Hence 7?=Lim n ' = Lim .-7,1 (1+0 i = l 1 + /| = V2.
(/»-+ 1)
Thus the radius of the circle of convergence for £ is y/2 and
centre is —1, so that the domain of convergence is given by
K+l | <. y/2 or | z’+l | < y/2.
(ii) Put ~ =£, so that the series becomes
£ 1.3.5...(2/i — 1) (^_I)W
n=i n !
1.3.5. .(2n— 1) 1.3.5.. .(2«—1) (2/1+1)

Here an = and tfn+1 =
n! («+!)!
«+l 1
Hence /?=Lim ^ |=Lim
2/2+1 rr
Thus the domain of convergence is given by
K-l| < h
or
T-'i<*
or |l-z|* < *|z|*
or (1-::) (1-2) < }zz
or 3r?-4 (z+2)+4 <0
or 22(Z+2)+f < 0
or 22—2 (z + 2) + i/ < |
*lso
or
ii
V
1
1
cor^
M
or
A
1
or l*-tl < f.
6. Examine the behaviour of the following power series on
the circle of convergence :—
OO oo 2U
(/) 2:
n=2 n (log n)2 ' ^ L n ’
f (-1 r -.
m n-1
co -4«
(/V) 27
ti-0 4//-f- 1
(i) First we find the radius of convergence.
Here , *«+.— 1
w (log //)* ,“"+1 (//+1) [log (n+ 1)J2 *
Then «=Lim -Lin.

an+1 n (log ny
Nov/ for any point z on the circle of convergence,
z" j_ 1
n (log ny i n (log /?)**
co J
But the series 27 —-is convergent by Cauchy's conden-
n=0«(log//)-
sation test. Hence the given series is absolutely convergent for all
points on the circle of convergence.
(ii) The radius of convergence is clearly 1. Also for z=l,
we have
~n
i
Power Series 397
But the series E - is divergent so that the given series is not
convergent for z — l.
To see that the series is convergent for every other point on
the circle of convergence, we apply Dirichlet’s test. For this we
take an=zn,un=lr
z — zn+1
Then | z+z2-fz3-f-... -fz" ' =
1 —z
I z l-H j ("+1
1-^1 i 1-?|'
so that the sequence of partial sums of Ean is bounded.
_1 J 1
Again
"" U^'-n~n+\-„(„+!)•
so that E | un -un+1 | is convergent.
1
Also Lim w„ —Lim- =0.
zn
Hence the series E anun i.e. E is convergent for every r
other than 1 on the circle of convergence.

(iii) The radius of convergence is 1. Also for z— - 1, we have
(- l)nzn _ 1
n ~n *
But the scries E ^ is not convergent so that the given series is
not convergent for z= — 1. To see that the series is convergent

for every other point on the circle of convergence, we apply
Dirichlet’s test. We take
Then | -z + zz-z*+ ... -f ( - |= —r)"*
l -zl+K-l)" z"+*| 2
ll-fz) "ll+r'l’
on the circle of convergence.
Hence the sequence of partial sums of Ean is bounded.
1 1 1
Again n -„+1 -W(/J+I)
so that E | i/„-i/n+1 | is convergent.

Also Lim yn=Lim^=0.

(— \ )n zn
Hence the series Zanunt i.a. Z--- is convergent for
n
every value of z other than — 1 on the circle of convergence.
(iv)Radius of convergence is 1 as may be easily seen.
An
l
Also for r= + l, +/, we see that But the series
4n -j-1 4//+1
1
Z - is divergent, so that the given series is also divergent for
4/2+ 1
the above values of r.
We now prove that the series convergees for all other
values of z on the circle of convergence. For this, we take
"n = 1 4/1+1*
i—z4n+1 [
Then | 1 + rl+r8 +... + r*n | =
1 —z* 1
1 + 1 z|4n+«
, z^±l, ±i
I 1 -z* 1
2
l-z* \ ’
so that the sequence of partial sums of Za„ is bounded.
. . 1 _ 1 __ 4_
Again un un+, 4//+1 4,;+5 (4//+l) (4/2 + 5)’
so that Z\un- unJrl | is convergent.
1
Also Lim w„= Lim =0.
4//+I
Hence the series Zanun% i.e. the given series converges for
every z other than ± 1, ±i on the circle of convergence.
Exercises
1. Find the radii of convergence of the following power series : —
(ii) Ciii) 2 ( 1-
oo —n
(iv) (v) Z -
2 log n
[Ans. (i) e, (ii) 1, (iii) e, (iv) 2, (v) 1.]
2. Show that the domain of convergence of the series
s is given by I-+ I <V5.

Power Series 399
CO ~n
Show that the series 27 converges for every value of
—•> n log n
z on the circle of convergence except z=l whereas the series

oo ( — !)’»zn .
27 —;- is convergent everywhere on the circle of convei -
n=2 n log n
gence except z= — 1.
4. Show that the series
2+£z2 + £z3+...
converges in | z | < 1 except at z= 1.
5. Find the radius of convergence of
CO Jzn
27
n=o 2" + !
and prove that
(2 — z) f (z) - 2->0 as z-*-2.
7*6. Sum function of a power series. If
f(z) = £ anzn,
0
then/(z) is called the sum function of the power series. We now
prove the following theorem on the analytic character of the sum
function.
CO
Theorem. The sum function /(z) of the power series ±'anzu

O
represents an analytic function inside its circle of convergence.
Further, every power series possesses derivati ves of all
orders within its circle of convergence and these derivatives are
obtained through term by term differentiation of the series.
Let the radius of convergence of the power series
OO
J (z) = £ anzn be /?, so that we have

0
~ = Lim I an |
Let <f> (z)=27 nanzn~l. The radius of convergence of this
series is also R for we have

Lim | nan l1,”=Lfm nlln \ an |,,n=Lim | an ,,,M (V Lim «l/n=lJ.
Suppose z is any point within the circle ol convergence, so
that | z | < /?. Then there exists a positive number r such that
1 z | < r < R.
For convenience, we write | z |=p, | h j=Y). Then p < r.

Also h may be so chosen that p+rj < r.
Since Z anzn is convergent in j z | < R, anrn is bounded for
0 < r < R, so that | anrn | < M where M is finite and positive.
We then have
f(z+h)-f(z)
h - <f> (-’)
OO
r • • •
n=o
co I\n fw—1)
z1 z\*-*\h\ + ...\h
n=0 1.2
°° M (n (n-1)
Pn“**)+. ..»/n_1
n=o rn \ 1.2
“Ml
{(p + ri)n- pn-”pn
_ n=0 rn '7j
...(1)
-I M©)"-©"-?0)'}
No. l(cay
n=0 V
r J ‘+©M©)'+-
1
H-r) r-p-f]
1-
To sum Z// , we put
*-'+!C),+3(0'+<;)‘+-
*,o *'-c)'+<y‘+<o‘+-
Subtracting, we get
Power Series 401
P
r p
j_P 1 -P*
r
pr
S=
(r-p)*'
Substituting these values in (1), we get
f(z + h)—f (z) , M f r r >jr J
~" (“) ^ 77\r-p-ri~r-p~'(r—p)*j
M r.V
»1 '(/■- p-7j) (r -p)*
Mrt\
(r-p- n) P>2
which tends to zero as
Hence Lim/,-»0 —Lfl) ~ ,f, (z).
It follows that / (z) has the derivative <l> (z). Thus/(z), which
is clearly one-valued, is also differentiable so that/(z) is analytic
within | z \ = R.
Again, since the radius of convergence of derived series
<f> (z)=2Jnanzn~l is also R (V Lira nlln \ an ^"^Lira | an |,,r*
since Lim nl/"=l], we see that </> (z) is also analytic in | z | < R.
Successively differentiating and applying the theorems, we sec
that the sum function / (z) of a power series possesse derivatives
of all orders within its circle of convergence and that these deriva¬
tives are obtained by term by term differentiation of the series.
Note. We proved Taylor's theorem in § 5 18 that a function
CO
f(z) can be expanded as a power series £ an (z—a)n about any

n=0
point a, provided that / (z) is regular in | z - a | < R. By combin¬

ing the above theorem with Taylor’s theorem, we may state the
result as follows :
The necessary and sufficient condition that a function f (z) may
be expanded in a power series is that it is regular in a region.
CHAPTER VIII
UNIFORM CONVERGENCE
8-1. Let 5, (2), Sa (z), S3 (2),... be a sequence of one-valued

functions, each defined in a bounded closed domain D. The
sequence of functions {Sn (2)} is said to be convergent at a point 20
of D, if the sequence of complex numbers Sx (z0), S2 {z0),... tends
to a definite finite limit. If the sequence converges at each point
of Dy it is said to be convergent in Dy and the limit function S (2)
of the sequence is defined at each point of D by the equation
S fz)=Lim Sn(z),
//-»oo
In other words, if z0 is any given point of Z>, for a given e,

however small, we can find an integer m such that
| S (Zq) — Sn (r0) | < e,
for every value of n ^ m.
The integer m in general depends on e as well as z0 so that

wc may write m=-m [e, z0). Thus to each point z of the domain
Dy and fixed e, there corresponds a value of m (e, z). The aggre¬
gate of the values of m (e, z) may not have an upper bound. If
this aggregate has an upper bound, say M, then for every value of
2 in Dy we have for n ^ M.
| S (z) - (z) I < €.
In such a case we say that the sequence {Sn (z)} converges

uniformly to S (z) in D.
Def. Uniform Convergence of a Sequence.
A sequence {5„ (2)} is said to converge uniformly to a function

S (2) of z in Dy if for a given e, however small but positive, there
can be found a poitsive integer m, independent of z, such that for
n > m, we have
! S (z)—S„ (2) | < €.
It should be noted carefully that the convergence of a

sequence {5n (z)} at every point of D does not necessarily ensure
its uniform convergence in D. A sequence may be convergent
at every point of an open domain and yet may not be uniformly
convergent in that domain.
Uniform Convergence 403
8*2. General Principle of uniform Convergence.
The definition of uniform convergence of a sequence given

in § 8'1 presupposes the knowledge of the limit function of the
sequence. It is, however, desirable to express the condition for
uniform convergence in a form which is independent of the limit
function. This is provided by the following principle of uniform
convergence.
The nectssary and sufficient con lit ion for the uniform conver¬
gence of the sequence (z), S2 (zj, S3 (z),... in the bounded
closed domain D is that, corresponding to any positive number c,
however small, there should exist an integer m (e) independent of x
and depending on e atone, such that for n ^ m (ej, we have
! Sn+J> (z)-Sn (z) | < e,
at each point z of D and for every positive integer p.
The condition is necessary. Suppose the sequence converges
uniformly to S (z) in D. Then given <=, we can find a positive
integer m, independent of z, such that for nil values of n ^ m
and for all points z in D, we have
| 5 (z)-S„ (z) | < U. .. .(I)
It follows that | S (z) -Sn+P(z)\ < U, ...(2)
for n ^ m and for every positive integral value ofp.
Hence | Sn+9 (z)-Sn (z) |
< | S„+p (z)-S(z) \ + \ S (z) -5. (z) f

< + from (1) and (2)
= €.
The condition is sufficient. If the condition is satisfied, we
have for every z in D,
\Sn+p(z)-Sn(z) | ^
• e. Sn (z) -*c < Sn+P (z) < Sn (z) + *€. ..
for every n ^ m and every positive integral value of p.
Now keep n fixed and let p-> oc, so that 6'„fp (z)-*-5 (z).

Sn(z)-k Srt (z) + \ r.
I S„(Z)-S(Z, | ^ |e < e.
for every n ^ m and every z in D.
Hence the sequence [S’* (z)J converges uniformly to the sum

of S (z) in D.
8*3. Uniform Convergence of Series. The series Eun (z) is

said to converge uniformly in the bounded closed domain D, if
the sequence {S„ (z)} where
Sn (2) = Mi (z)+u2 (z)+... (2),
converges uniformly.
Principle of uniform convergence of a series. The necessary
and sufficient condition for uniform convergence of the series Sun (z)
in the bounded closed domain D is that to every positive number e,
however small, there corresponds a positive integer m, independent
of 2, such that for every n ^ m, we have
I «*+l (Z) + » n+2 (Z)+ • • • (2) I <
where p may have any positive integral value.
The proof is the same as that of § 8 2, if we replace the
series Eun (2) by the sequence {Sn (2)}, where
Sn (2) =
(2) -fw2 <2) T. • • (z)»
8*4. Tests for Uniform Convergence. Below we give some
important tests for uniform convergence of a series.
8-5. Weierstrass’s M-test. The series Eun (z) converges uni¬
formly and absolutely in a bounded closed domain D, if .
(i) j Mu (2) | < A/„ for every n and 2 in D, where Mn is a
positive constant independent of and
(ii) the series E Mn is convergent.
Since EM„ is convergent, for a given €, there corresponds a
positive integer m, such that
| 3/,,+! + A/nj.2+ • • • + A/n+j> | < €,
for // ^ /», p > 0.
Then it follows from condition (i),
! M„+l (2)-T Mn+a (2)+ •• • + wn+p (z) I
+p (-) I
< Afn+i + A/n+2+ ...
< e.
for // > p ^ 0 and every 2 in D.
Hence the series is uniformly convergent in D.
Again since,
I Mn+1 U) I +1 Wn+2 U) |-f. . . +| Wn+J> (r) ! <
for n > w, p > 0,
we see that the given series is also absolutely convergent.
Note. The above test gives one of the simplest sufficient
conditions for uniform convergence.
8*6. Hardy’s tests for uniform convergence.

We first prove the following lemma :—
Lemma, If, in a bounded closed domain D, the function
un (2) vn (z) tends uniformly to a finite limit as n->x>, then if one
of the series
uo (z) vo (z)+% {//„ (z) - (z)} vB <Z)
OC
and ^ wn (z) {vn Iz) - vn+1 (z)}
0
converges uniformly in D, so also docs the other.

We have the identity
w0 (z) v0 (z)-\-£ {uT (z) - u (z)} i-r (z)

1
n-1
- 2’ ur (z) {vr (r)-vr+1 (z)} = t/nV
Proceeding to the limit as oo, we see that
Wo (*) Vo (2)+Lim„_>0O2 {i/r (z) //f—i (z)} vr (z)
- Lim,,^ i,,Mr (z) {vr (z) - vHl (z)}
= Lim,J_>a, <4nvn — a finite quantity.
Hence if one of the series tends to a finite limit, so does

the other. In other words, if one of the series is convergent, so
is the other.
lest I. If in a bounded closed domain D,
(i) the series 2' un (z) has uniformly hounded partial sums,
(H) the series £ {vn (z) - v„+, (z)) is uniformly and absolutely
convergent, and
(HI) vn (z) tends to zero uniformly as n-* cc ,
then the series £ un (z) v„ (:) converges uniformly in D.
Let Sn (z) = u0 (z)-f ux (z)-f-... -\-u„ (z). Then from condition
(i), it follows that | Sn (z) | < A', a finite constant, for all n and all
z in D. We then have
I Sn (z) vn (z) | < K | vn (z) |. ...(A)
Using condition (iii), we see from (A) that Sn (z) v,t (z) tends
uniformly to zero as //-*oo .
406 Calculus of Residues
Hence, by the lemma, 2 {Sn (z)-S„_j (z)} v„ (z), i.e.

S u« & vn (z) converges in D if 2 Sn (z) {vn (z) - vn+1 z} does.
Now from (ii) it follows that, for a given e, we can find an
integer m, independent of z. such that
Ti 1 ”r (Z)_V'+I (z)l <T

for every positive integer p and for all z in D.
n+p |
Hence 2 Sr (z) {vr (z) - Vf+1 (z)} |
n«f 1 j
< 21 | (z) | | vr (z)-vr+1 (z)

n+1
n-f-p
<AT i7 I vr (z)-vf+1 (z) I < €.

n+l
It follows that 2 Sn (z) {vn (z) - vn+1 (z)} converges uniformly

in Z). Hence the result.
Test II. If in a bounded closed domain D,

(/') r//e jer/ej 2 un (z) converges uniformly
(ii) the series 2 {vn (z)—vn4l (z)} is absolutely convergent and
has a bounded sum, and
(Hi) r0 (z) ij bounded, then the series 2 u„ (z) vn (z) converges
uniformly in D.
Let Rn (z) = un±\ (z) + ;/n+2 (r)-fi/n+3 (z)+...
Then from condition (i), /?„ (z) tends to zero uniformly as
M-* OC.
Again from (ii) and (iii), it follows that
2 | v„ (z)-vn4l (z) | < AT,,

0
and 1 v0 (z) I < K2,
where AT, and ATa are finite positive constants.
i
Hence ; v„ (z) 1 = {vr (z)-vr4l (z» j
< J v0 (z) ! 4- 2\ vr (z)—vr4, (2)|

0
< + AT2 < AT, ... (I)
where K is finite constant.
We then have
I Rn 00 v„ {z)\< K\Rn (z) |. ...(2;
Since Rn (z) tends to zero uniformly as n-*- ^, it follows from
(2) that Rn (z) (z) tends to zero uniformly in Z).
Hence u0 (z) v0 (z)-j- Z un (z) (z;

1
=wo (*) v-o (-) - I' {*» U/ - K-t (Z)j v„ U-)
converges uniformly in Z), provided

^ -ft* (*) K (2)-vnfl (z)J does so.
Since Rn (z) tends to zero uniformly, for a given £, we can

find an integer m, independent of z, such that
I (z) | < —, . . .(3)

for all n > //; and for all z in D.
i
/?r (z) [vr (z)- vf+1 (z)J

n+1
n-fp
€ vr (z) - vrU (z) I from (3)
"+'
€ r | V, (Z)-Vr+1 (Z) |
< K 0
< e from (1;,
for all z in D and for every integer p. It follows that
i K (2) (v„ (z>- vn+l (z)j
converges uniformly in D and the result follows.
8*7. Continuity of the sum function. If each term of the
infinite series Z un (z) be continuous in a hounded closed domain

0
A and if the series converges uniformly in D, then the sum function
S (z) of the series is also continuous in D. (Delhi M. A. Final 1959)
Given that the scries Z un (z) is uniformly convergent in D.
We have to show that the sum function £ (z)= Z un (z) is conli-

0
nuous in D.
Let Sm (z) = Mj fz) + Mg (z)+...+«« (2).
Then, since the series is uniformly convergent in D, we can,

for a given s, find an integer m independent of z, such that
j 5n (z)-S (z) | < ie.
for every /; > m and every z in D.
In particular, this gives
I Sm (z) - S (z) | < |e for every z in D .. .(1)
and | Sm (z0)—S (z0) j <32 ...(2)
where z0 is any point of D.
Now Sm (z) is the sum of a number m of continuous func¬
tions and so it is itself continuous. Hence there exists a positive
number § such that
I Sm (z) — (rol I < 3e •••(3)
for all values of z for which | z — z0 \ < 8 where z is a point
of D.
From (1), (2) and (3), we see that
15 (z)-S(zn) | = | S(z)-Sm(z)+Sm (Z)-Sm (z0)+sm (Z0)-S(z0) I
 1
-f1 Sm (Z0) — S (Z0) i
< |t+ae+a€ = e.
Hence S (z> is continuous at any point z0 and is therefore

continuous in D.
8*8. Term by term integration. If (i) the series Zun(z)
converges uniformly in a closed bounded domain D.
(ii) each un (z) is continuous in the closed domain D,
then the sum function S (z) of the series is iutegrable along f, a
contour of length l lying in D,
and Jf [ S un (r)J dz=Z J ()

un z dz,
/e the integral of the sum is equal to the sum of the integrals.

(M. A. Final Delhi 1959)
Using the notation of § 8*7, we have to show that
5 (r) dz - Lini/j^QoJ"j. Sn (z) dz.
The function S (z) has been shown to be continuous in D in

S 8*7 and so is certainly integrable along f. Since the series is
uniformly convergent in D. we can, for a given e, find a positive
integer m, independent of z, such that for n ^ w, we have
Uniform Convergence AO)
S (z) -Sn (z) <j.
For such values of //, we have
ij^ {5 (z)-S. <z|) dz I < 5 (z)—S„ (z) dz
< Adz
= /. T=€.
Hence {S (z) - (z)} as «-►^
i.e. S (z) Jz=Lim/i_>0O Sn iz) dz.
i.e. jr E »/„ (z)dz= E w„ (r) dz
The theorem is therefore cstabished.

8'9. Analyticity of the sum function of a series. Term by
term differentiation. If in a bounded closed domain D within a
closed contour C,
fi) the series E u„ (z) be uniformly convergent,
fii) each un (z) be an analytic function,
then
(i) the sum function S (z) of the series is analytic in D,
(ii)the term by term differentiations is justified,
i.e. S' (z)=Eun' (z),
and (Hi) the series Euf (z) of derivatives converges uniformly
to S' (z) in D,
Let z0 be any point of D.
Consider a circular neighbourhood
>
\ 2 - Z0 | < p
of z0 and let L be any closed con¬
tour in this circular domain.
Since L is a closed contour,
the series Eun (z) is uniformly
convergent on L. Also the sum
function S (z) is continuous in D.
Hence term by term integration is justified on L and we have
\L S(z)dz=jL ul(z)dz + j/ u2(z)dz+fL u3(z,dz-K.

410 Complex Varicble
Since each un (z) is analytic in the circular domain, we have

by Cauchy’s theorem.
JL un (z) dz=0.
for all values of n.

S (z) dz = 0.
1
Hence by Morera’s theorem, it is analytic in a circular neigh¬

bourhood of z0. m
As Zq is any point of Z), we see that S (z) is analytic in D.
This proves the first result.
To prove the second result, we take a simple closed curve
F lying in the neighbourhood! z-z0 | < p of z0 and having z0 in
its interiors.
Obviously the series
1_Siz) 1 Ml (Z)
+ 2?r
1 Uo (z)
(z-z0)
f, T...
2rri (z-Z0)z 2rri
is uniformly convergent on F.
Hence integrating term by term, we get
1
_L f su-i rf,= J_ f - dz T" f “a (z) dz 1
2-ni J / rT*-*o
2-m J F
(Z-Z0)a ~ 1 2”T]riz-Zor
i e. (z0) + u2'(z0)+...
by Cauchy’s integral formula for the derivatives (see § 5T3).
As z0 is any point of D, we see that
S' (z) = ux' (z) + m2' (z)+ • • •
for every z in D. This proves the second result.

Nowit remains to show that 5/(2) converges uniformly to
S’ (z) in D within C. For
this we construct a closed
contour H which lies within C
and yet is definitely out side
D. Let d denote the shortest
distance between r' and D.
Let Zq be any point of D\
then by Cauchy’s integral formula for the derivatives of an
analytic function (§5*13), we have
1 f S (z) ,
S' (Z0)—Sn (*o)= 2^rJr (z _ z0)* dz
___lf JsSEL.*
Irrj J T (2 - Z0)2 “
Uniform C on verge nee 411
= _Lf S(z)-Sn(z>
dz.
2m Jr (2-Z0)*
Since Sn (z) converges uniformly to S (z) within and on f\
we can, for a given find an integer m, independent of z, such
that
| (z)\ <€,
for every n ^ m and every z on f.
Also
so that we have
•S(2)-5n (z) , <. e
(z~zQf h2
Hence it follows from § 5*6 that
s' (z„)-.s-n- (z0) k-Lf/

2n 8*'
e/
2rr 5
where / is the length of f.
«/
since tcn,Js uniform,y to zero, being independent of.,,
it follows that 5V' (z) converges uniformly to 5' (z) in Z).
Proceeding in the same manner, it can be shown that Sn"(z)
converges uniformly to 51'' (z) in every closed domain, within C
and so on indefinitely.
Thus the theorem is completely established.
8*10. Uniform convergence of power series.
We have seen in the previous chapter that a power series
converges within its circle of convergence and diverges outside
it and that on the circle of convergence its behaviour is uncertain
We now prove some theorems on uniform convergence of power
series.
8-11. Theorem I. A power series is uniformly convergent

within its circle of convergence.
Let R be the radius of convergence of the power series
L a»z*' Let r be a positive number such that
I zl < r < /?.
We lher have
I onzn | < an i r
Since every power series converges within its circle of conver¬

gence, it follows that the series £ | an ( rn is convergent.
Hence by Weierstrass’s A/-test (see § 8*5), the series £ anzn
converges uniformly for | z j < r < R. /. e. within the circle of
convergence.
Note. The series £ anzn may or may not be uniformly
convergent on the circle of convergence, e. g. the series £ z"
,
is not convergent when | z j—1 /. e. on the circle of convergence
OO
_Z”_
whereas the series 2 ’ convergence for ail points on
n=1 n (log nV
which is the circle of convergence [see hx. 6 (i) $ 75} and hence
it is uniformly convergent on and within the circle ol convergence
by the above theorem.
Cor. (i Since each term of the power series 2 anz" is conti¬
nuous and the series is uniformly convergent, for ! * I ^ r < &
(R finite), it follows that the sum function of the power series is a
continuous function throughout the closed region formed by the
interior and boundary of any circle concentric with the circle oj
convergence and of smaller radius.
(ii) Again it follows from § 8S that the series f (z)=£ anz«
can be integrated term by term along any contour F lying within
the circle of convergence, i. e. we have
Jr /(x) *-ff, v*.

(ii) Finally the series f (z)=£ anzn can be differentiated term
by term for | z [ < r < R,
i.e. f' (*)-2 nanz”-\ [See §8-9]

1
We further see that the radius of convergence of the

differentiated series is the same as that of the original series, for
Lim nVn \ an |1'"=Lim | an |1,n [ V Lim =
It follows that the differentiated series is uniformly conver¬
gent within the same circle of convergence and therefore can be
differentiated term by term any number of times.
And so we can write
/' (z)=£ nanzn-\
f" (z)=£ n (n- 1) anzn~v. etc.
8*12. Theorem II. If a power series 2 jnzn vanishes identi¬

cally for values of z within its circle of convergence, then all
the coefficients in the power series vanish separately, i. e. a„ — 0 for
all n.
Let the circle of convergence of the power series be I z =R.
Hence the series is convergent for | z | < r < R, r > 0.
Now suppose, if possible, that all the coefficients do not vanish
and let am be the first coefficient which is not zero.
Then am + amHz-\-amx.iz'l+... vanishes for all values of r
except zero and converges absolutely when \z \ ^ r < R.
We then have
oc
I tf»n+i • .• I ^ E I am+n ^
Hence if '/* (z)=am+lz+am+2z* + ..., we can find a positive

numbes S ^ /• such that
| <!> (Z) | < ••• •••(!)
for all values of z for which | z 1 < $.
[It is sufficient to take 5 to be the smaller of the numbers r
and \\a„\[$ | I /■"-') •

2 / »=i
Then | am + + (z) | > | am |-|-* (z) \
(*) I
> h\am\ from (1).
Hence | an + >b (z) |^0, when | z | <5. We have therefore
arrived at a contradiction by supposing that some coefficient does
not vanish. Hence all the coefficients must vanish.
Cor. If one and the same function f (z) can be expandedin
a power series in two distinct ways, so that
f(z)= S °nz" = $ bnz\

0 n=D
both series being convergent, when | z | < R, then a„*=bn for all
values of n.
If we write an -bn=*an, then, when | z 1 < R.
0= 2 cnzn,
n=0
And so c„=0 for all values of n by the above theorem.

Hence Qn—bn for all n.
8*13. A note on absolute and uniform convergence.

It is to be noted carefully that the uniform convergence of a
series does not necessarily ensure its absolute convergence, nor
z2
conversely. For example, the series S^-jqpppr converges abso¬
lutely, but not uniformly (near z=0). Again the series
y (1 is not absolutely convergent, since the series

n=l z*+n
£ * is divergent. But for all real values of z, the terms

n=l | n + z2 |
of the series are alternately positive and negative and each term
is less than the preceding one and so the sum of the series lies
between the sum of fitst n terms and of its first («+l) terms.
Hence the remainder after n terms is numerically less than the
nth term,
1 I I
_
I = I <■» d- • • •
i.e. |z* + n-fl z2+n+2
1
z*+n
 —.

n €
Hence, taking an integer m > j- we have |i?n I < e> f°r ail
values of n ^ m and all values of z. And so the series is uniformly

convergent.
1. Show that near z=0, the series
oo_ -2z (l + z)"^_
Jo {l+d+^riu+dt^}
is non-uniformly convergent and its sum is discontinuous.
We have _ % ,
_2z (I +z')"~l 1-0+z)" l-(l+z)°~‘
»» (^“{l + d+zj-1} {1 + ll+m 1 + U+z)" 1+U+Z>"~'
!-(!+«)
• l+(l+z)
l-(H-z)*_l-(l+z)
1 + (1+z)’ l+(l+z)’
_l-(H-z)3_ l-(l+z)3
"3“H-(l+z)3 l+(l+z)3 '
Complex Variable 415
Hence £.
Therefore considering real values of z, we see that
1 when - 2 < z < 0,
S (z) = Lim Sn (r)= • 0 when z = 0,
n~*‘°° - 1 when z > 0.
Thus the sum function S’ (z) is discontinuous at z =0 and
consequently the series is non-uniformly convergent near z = 0.
Now uniform convergence may also be established indepen¬
dently as follows :—
We consider z > 0. Then
| s f-7' - S (z) | --= —1— 1 ~~ 0 ~^n

1 U 1 -H14 Z)n
_ I ~2
.. (I)
l+(l + z)n* •• (,)
I n particular, taking z= — we see that the right hand side of
(1) tends to j ! e~ as n-> oo. Thus however great n may be taken,
the right hand side of (1) cannot be made arbitrarily small. There¬
fore the series is non-uniformly convergent near z = 0.
oo 1
(1-2") (I -zn+')’
is equal to ^ , when I z | <1 and is equal to —- ^ . when

\ i - Z)~ -? M —
- (i ^ i-
Is this fact connected with the theory of uniform convergence ?

H __ zn~l __ _ I r 1 1 -j
ere u„—(^ (j _z»f») z (I - z) [_ I -z" 1 - "z«m J
Wi (z)=z{\-z) [i-z “r^]
“2 (Z)= z - z)\\^)~ T^z*~]
u*(z)= FT7-1) [i -73 ” H-z4] etc*
Hence S'„ (z) = —-[“■=-- * -!—1

z(l-z)Ll-z« 1
••• S M=LinWj S. W=r(-/r-2, (,~ - pro) when z < 1
and =_i_ (-L-cA when I z I > 1

Z(l-Z)\i—Z /
i.e. the series is equal to ^ __jj- when | z j < 1 and is equal to
_r (, _ z)t when | z | > 1.

As a matter of fact, the series is non-uniformly convergent
near z= 1 as may be seen by considering real values of z > 1.
Let c be a given positive number. Then
\S (z)- S„ (z) |=j z (i _z) (1 -z"*1) 1 < e

if z(z-l) (zn+l-T) K €
if 2"+1 > l+z(z^T)7\
log (
1+z(2-l)g)
i.e. if (/i+1) > log z
...(1)
Now when z tends to 1, it is evident from (1) that n-roc
and so it is not possible to determine an integer m, such that
( 5 (z)—Sn (z) | < 6,
for every n ^ m and every z > 1.
Thus the series is non-uniformly convergent near z=l.
3. Show that the following series for exp (z), sin z and cos z
are absolutely and uniformly convergent for all values of z, real
or complex :—
o') i+z+=n+3-!+--'+;rj+-'-
2’ 2* 2’ . , ( - D” 2Jn+1 ,
<"> z-3i+rrr!+"-+ u'n+ir+--*
2s Z4 Z6 ( —l)"Zln ,
• • •
(,7/) 1-2!+4'!-r!+,‘,+ (2n)! +
~n-1
(i) Here u„= *A»+1 M t•
Then Lim,,^ I Uf^ =Lim/J_0O - =0 < 1 for all finite
values of | z |.
Hence by D’Alemberts ratio test, the series is absolutely

cunvergent for all values of z.
Further the series is uniformly convergent throughout any
bounded region of values of z, for if the region be such that
| z \ < R when z is in the domain, then
\ Z1 \ ^
I n ! I ^ nl'
CO
But the series Z —, is convergent for all finite values of R

n-o n !
by D’Alembert’s ratio test as proved above.

OC 2n
And so by Weierstrass’s A/-test the series Z — is uniformly

n=0 n !
convergent.
Proofs of (ii) and (iii) are similar to that of (i) and are left
as an excercise for the students.
Note. Since the series for sin z is uniformly convergent,
sin z
and so defines a continuous function of z, and, in particu¬
lar, this function is continuous at z = 0, hence it follows that

sin z
Lim z-+ 0 — Lim._*o |^1 ^ j4-• • • 1.
4. Show that the series Z f f*zti,J is uniformly convergent
in the closed domain \z \ < r, where 0 r < 1.
We have 1
1— zin 1 —1 z
r*n
when | z | < r.
1 — r,n
In
Now the series £ l ~r,w is convergent when 0 r < 1.
2n r«n+a
For, if 2n , un+l = y—-lirFi , we have
u In j _ rm+i
r,n+*
so that Lim ^-=-|

4-, rB
1 since 0 1.
■in
Thus by D’Alembert’s ratio test, the series 2 is
1 — r2n
convergent.
Hence by Weierstrass’s A/-test, the given series is uniformly
convergent when [ z | < r, < 0 r < 1.
oo rn oo
uniformly convergent in \ z \ < 1.

rn zln rn
0) Here where I 2 l=r < 1.
I+*8
n 1-|2 2«=l-r 2fl
.n
But the series 2 ^ _^Sn is convergent when r < 1 by
D’AJemberi’s ratio test as may be verified by proceeding as in

the previous question.
It follows by A/-test that the given series is uniformly
convergent for j z | < 1.
>2n—1
(ii) Here ! ( — l)"+i
1 -r2"-1 1-1 Zj1”-1
-In-l
1 —fin—X when | z l = r < 1.

■ 2n-l
But the series E j _rgn-i *s convergent when r < 1 as may
be seen by applying D’Alembert's ratio test.

Hence the given series is uniformly convergent when
iz 1 < 1.
6. Show that (he series
2 sin ~ -f 4 sin ^ +... +2n sin ~ +...
converges absolutely for all values of z except zero, but does not
converge uniformly near z—0.
1
sin —
Now Limn^w 3” sin ^=Lim/J^0O
3«r
= *, z=£0 (see Note of Ex. 3 above).
Hence we can find a number A/, independent of at (but

depending on z), such that
Uniform Convegrence 419
I 3" sin 4r < w

I 3"z i
and therefore 2n sin < M

3"z,
But the series £ M (*)n is convergent. It follows that the

given series converges absolutely.
Now we prove that the series is non-uniformly convergent

near z=0. For this it is sufficient to show that z=0 is a point
of discontinuity of the sum function defined by the series. Now
the points of discontinuity of various terms are also the points of
discontinuity of the sum function. Clearly each term of the
series is discontinuous at r = 0 since Limz->0 2'* sin 4^ does not
exist. Hence the series is non-uniformly convergent near 7 = 0.
7. Show that the scries £ n~s converges absolutely ami uni•

i
fornily in any hounded closed region in which R (_) > /.

Since R (r) > 1. we can find a number o > 0 such that
R (z) ^ 1+5.
Now putting z=.y + />\ we have
«*«=cxp (7 log ;?) = exp {(.v+ iy) log /»}
= exp (x log «).cxp (iy log //)
[only principal values being considered)
— nx exp (iy log //).
7 hen | //* \ = n1 | e.vp (iy log n) \
=nc since | exp (iy log //) | =- 1.
And, since R (z)=x ^ 1 +5, we have
1 |_ 1 1
nM !“/i* ^ „'+*’
for every n and every r such that R (z) > 1 + 5.
OC ]
But the series E ■ —» is known to be convergent,

i «•+*
Hence the given series is absolutely and uniformly convergent

in and bounded closed region by Weicrstrass’s A/-test.
Note. The sum function of the series £ is known as

/r
Riemann’s zeta function and is denoted by the symbol ( (7).

Clearly l (z) represents an analytic function of 2 in the region

R (z) > 1, since the series for £ (z) is uniformly convergent and
each term ^ is analytic in R (z) > 1.
8. The power series 27 anzn, whose radius of convergence is
unityt converges at the point z= 1. Show that the series converges

uniformly in the domain \ I - z \ < cos 8,\arg (7-z)| < 8 if
0 < 8 < $7T.
Deduce that
CO oc
27 anzn-r 27 ant
o o
as z-+l along any path within the circle of convergence which does
not touch that circle.
We apply Hardy’s second test for uniform convergence given
in § 8*6.
For this we take un(z)=an, v„(z)=2n. Since the series
27 anzn converges when 2 = 1, the series Ean is convergent and so
the first condition of the test is satisfied. The third condition
is also satisfied since v0 (2)=1 so that v0 (z) is bounded.
To see whether the second condition is satisfied, we proceed
as follows :
When | z | < i, we get
•£K w-viw 1=1 i—z\s\zr=|^j-

Now the inequalities | 1—z\ < cos S, arg (\—z)\ < 8,
determine a sector of the circle with centre at 2=1 and radius
coso. When 0 < 8 < \ir, each point of the sector, except 2=1,
lies within the circle of convergence of the given power series.
In this sector, we put
1 -2=pe*9, where 0 < 8.
|
|2=1— 2p cos rf- + p2 < 1 -2p cos 5-f p*
2
^ 1 —2p cos 8 + p cos 8 ^ 1 - p cos 8 -f- £p2 cos* 8,
and so | z | < 1—ip cos 5.
Hence at each point of the sector, except z=l, we have
I Z ! V. fzj-■>.+, (z) I S =2 sec 5.
This inequality is also satisfied at 2= 1, since each term of

the series 271 vfl (2)—v*+1 (2) j vanishes there. Thus the second
condition of the test is also satisfiad. It then follows that the

series converges uniformly in the domain I 1 — z | < cos 8,
| arg (I — z) | < 8, when 0 < 8 <
Again by § <S-7, the sum function of the series Eanzn is
continuous in the sector under consideration.
Hence Eanzn->Ean, as 1 along any path in the sector.
This result also holds when 1 along any path within
1*1 = 1 which does not touch the circle of convergence; for by
choosing 8 sufficiently near to we can ensure that such a path
lies in the given sector.
f (~)= I-“—..
is analytic in the semi-plane R (z) > 0.
First we test the series for uniform convergence. For this
we apply Hardy’s first test given in § 8‘6.
We lake un (z)=( — l)n, v„ = The partial sums of
Eun(z) are alternately 1 and 0 and consequently the condition
(i) of the test is satisfied. Condition (ii) is also satisfied For
— V? (z) —5
vn (z) |=(«+1) <(//+!) ,6^0
and so vn (z) definitely tends to zero uniformly in D. Now we
show that the condition (ii) is also satisfied. For this we use
the formula
^ (z)-i'nu (z) = (n+\)-i-(n+z)-*'=z f+

J nfi
-C-l
dt.
This gives
I (z) - V„+1 (z) I < I Z I f

in ft
fi
f I r-*-1 I dt
n+2 _S-1 —S-l

: z
in+l
t dt < | Z | (// 4- | )
But the series E is convergent. It follows that the

(/i+l)s+ 1
scries 27 {v„ (z) — vn+1 (z)} is uniformly and absolutely convergent
in the closed domain R (z) < 8.
Thus it has been shown that the series
f(z)= 1-2-2~s + 3-‘-...
converges uniformly in any bounded closed region in which
R (z) < 0.
Also each term of the series is analytic in the domain
R (z) > 0. Hence we deduce that the sum function/(z) of the
Scr:es is analytic in the domain R (z) < 0.
10 . Test the following series for uniform convergence :—
('> * ian 2L» W * i TTF~ •
(m) 27 ~^rr t+f-

(i) We apply Hardy’s first test given in § 8*6. For this,
21 1 2
we take (z)=^r » vn (z)=~ tan ™ *
z .
Clearly the series 27 ^ is convergent in any bounded closed
domain and so its partial sums are bounded.

Therefore condition (i) of the test is satisfied. Condition (iii)
is also satisfied, for we have
1 Z
Limn->00 vn (z)=Lim„_>00 - tan^=0.
To see whether condition (ii) is satisfied, we have

Lim 2n+1 [vn (z)-vnH (z)]
2n+>
= Lim /J-*CO tan £ -lan^p
sin 2nf1
2n+i
= Lim n-+oo
z
COS tTT cos 2«+i
sin
2n+1
— Limôo
z z
cos 2" *COS 2"*1 2n+i
= 1
Hence we can find a number A/, independent of both n and
z such that | 2"+1 {v„ (2) -v„+1 (2)} | < M,
i.e. I Vn (z)-vn+1 (z) 1 < 2H+1.
But the series 27 ^ is convergent. It follows by Weiers-
strass’s M-test that the series 27 {»■„ (- ) — vn+1 (z)} is absolutely and
uniformly convergent in any bounded closed domain. Thus
condition (ii) of the test is also satisfied. It follows that the given
series converges uniformly in every bounded closed domain.
(ii) We apply Hardy’s second test given in § 8-6. For this

I
we take v»<z)=f+7»
Since Z is convergent, the condition (i) of the test is
satisfied. Also v0 (z) = ^ — ~ , so that v0 (z) is bounded. Hence
condition (iii) of the test is also satisfied. Condition (ii) is also

satisfied, for we have
Zn 2n+*
^ (z) - v
^= T+zn ~ 1 + zn+l
zn (l - z)
(1 +ZU> U4-zn+1)
I z\’'[\+\-z\)
rn (!+/■) , , ,
t-. wliere \ z \ = r
(1 — rn (1 — rn+1)
fl
Z v„ (Z) - V.+1 (z) I < (jtr)f ( 1 _ ?■ - iw>t|)
0
=£[rVr^r]-
1 + /T 1 _ 1 1
Hence Z '’n (z)- »•„+, (z) | =Lim,/_>0O
o 1 —r 1 —r 1 -rn+l_
1 4- r
if r > I.
and r < 1
r (14- r)
i.e.
(l—r)2
Thus Z | vn (z)-vn+1 (z) I is convergent and has a bounded
sum when | z | > 1 or | z | < 1. Hence condition (ii) of the test
is also satisfied. It follows that the given series is uniformly
convergent when | z | > 1 as well as | z | < 1 i.e. in the domain
\z | > I4-8 as well as | z j ^ 1—8, where 8 is any posili\e
number.
(iii) We apply Hardy’s first test in this case. For this,

we take
(— ii
un(z)^—\L, vn (z)=
n Z4~n
(-1)" .
The series 27 ——— is convergent since the terms of the
■* •
series are alternately positive and negative and Lim,/->30 un=0. In
fact the sum of the series is Hence the series 27 un (z) has
uniformly bounded partial sums in every bounded closed domain

and so the condition (i) of the test is satisfied.
I
Again Lim,,^ vn (c) = Lim^_> =0 for all z.
30 z+n
Thus condition (iii) of the test is also satisfied.
1
But the series 27 7 is convergent for all finite
(n-r) (/l-f-1 - r)
values of r. Hence the series 27 j vn (z) - vB+1 (r) | converges
uniformly in every bounded closed domain by Weierstrass’s Af-test.
Thus condition (iii) of the test is also satisfied. It follows that
the given series is uniformly convergent in every bounded closed
domain.
II. Show that the series 17 ze~nz converges absolutely but not
0
uniformly in the sector | z | R, \ arg z | ^ where 0 < 8 < Irr.

Is the convergence uniform when r ^ | z | < R, j arg z | ^ 8 ?
Putting z=pei<f, we have
ze~n: = pe /? e — n?eU — p 1 g n? C35 ?^ e— ni? si;i ? |
= pe-np cos ?
Also p cos <£ > 0 for 0 < | j < so that

—p cos ? 1.
Thus the series Z I ze n~ \ =p Z e "r cOS 9 js a G. P. with
common ratio less than 1 and consequently it is a convergent series

It follows that the given series is absolutely convergent in the
sector | z | < R, | arg z \ ^ 5 where 0 < 8 <
The series is not uniformly convergent in this regionjorj.
sum function 5 (z) is discontinuous at 2=0 \vliich i5"'a"poi nt of
the region. This can be show—beIow. We easily see that
S (z) is given by
0 if 2=0
S(z) = if 2^0.
1 -e‘s
Now Lim._>0t > (2)=Limz_>0
Thus 2=0 is a point of discontinuity of the sum function and

consequently the convergence is non-uniform in the region
l 2 | < Rt 0 < | arg 2 | < \tt.
The series is, however, uniformly convergent in the domain
< | 2 | ^ R, | arg 2 | ^ 5 .
F or, we have | ze nz \-=pe~"p cos 9 < Rc~'" cos 5
and the series Z Re~"rcosS is convergent. Thus by A/-test, the
given series is uniformly convergent in the above domain.
Exercises
1. Show that the series Z —-is uniformly convergent

zi — /r 77-
in every bounded closed domain not containing the points
±mr, n being any positive integer.
ei,,:
2. Test for uniform convergence, the serie sZ — .
Ans. Uniformly convergent in every bounded closed

domain defined by R (z) ^ MM (2) > rh where
8, y) are any positive numbers.
3. Show that, if Z an is convergent, the scries Z antr2 conver-

1
ges when R (z) > 0, but is not necessarily convergent

when R (z) — 0. Prove also' that the series is uniformly

convergent in any bounded closed region for which R (z) > 8
where 3 > 0.
8*15. Infinite Products.
Definition. If an infinite number of factors are multiplied
in a given order according to some definite law, then the products so
formed is called an infinite product.
The product axaza3.. .alt of finite number of factors alt a2.. .an
n
is denoted by TT ar and the product of infinite number of factors
r=l
oo
a\*ai,a2... is denoted by n oT. For convenience we take the
r=l
factors of the form 1 4- ar.

816. Convergence and Divergence of Infinite Products.
Let an be any real function of n defined for all positive integral
values of n and suppose
pn= ii (I+ar) = (l + ax) (l+a2)... (1+aJ.

f =i
If Pn tends a finite non-zero limit P, we say that the

infinite product II (1 converges to the limit P and we write
OO
n= 1
If Pn does not tend to a finite non-zero limit, then the product
n (1+ an) is said to be divergent. In this case Pn either tends to zero
or infinity. When Pn->0, we say that II diverges to zero.
It should be noted carefully that if each of a finite number
of factors has the value zero, the product will be said to be conver¬
gent if it converges when these factors are removed.
The phrase ‘diverges to zero' may appear to be strange to the
student at first sight but it will appear quite natural to him if he
observes that the behaviour of the product II (1 -fan), an > — 1, is
completely determined by the behaviour of the series 27 {log (1
Thus to say that the product II (1 +diverges to zero means
the same thing as saying that the series 27 log (1 diverges
to — oo.
OO
If the product II (l-ftfn) is convergent, then Pn and Pn-X

U=l
Complex Variable 427
must tend to the same limit as n-> oc, that is,
l/*2- = 1 + an -> 1 or an -> 0.

rn-1
Hence a necessary condition for the convergence oj II (1 + a,J
is that an ->-0.
I follows from above that, if n (1 -f a„) has an infinite number
of negative factors, it cannot converge.
8-17. General Principle of Convergence of an Infinite Product.
The necessary and sufficient condition for the converges of
the infinite product P is that corresponding to any positive number
e, however small, there exists an integer m such that for all value of
n ^ m, we have
- 11 <£.
whatever p may be. [-^8ra 6

The condition is necessary. Let Pn converge to a finite non¬
zero limit P. Then a positive number k can be found such that
| Pn\ > A; for every n.
Now a number wean be so chosen that
| Pm+p-Pm | < ke, for all integral values of/?.
u Pm+p 1 I < kz ie < e

Hence „ -■ i ^
/ „, 1 I Pm 1 K
The condition is sufficient. Let us assume that
Pn+v _ i < € for n m and for all values of p,
Pn
i.e. -«, ^
< * *»+p
p — 1 C e
^
1 n
or in particular, we have
i+«-
* m
Since € can be chosen as small as we please, we get

1 Ptnl pi > (!“«)! Pm !•
Hence [ Pn | > some fixed positive number vs hen n > w, and
therefore Pn cannot tend to zero.
AlSO (1 — C) Ptn < An + p < (•+«) P'"‘
But Pm is the product of finite number of factors and there¬

fore finite. Hence Pm+V lies between two finite quantities tl Ptn
and (1-fe) Ptn for all values of p. It follows that P„ tends to a
finite limit as n-*- co and therefore the product P is convergent.
Fxample. Show that the product
converges to i. (Agra 46, 57, Jaipur 54, Luck. 56)
Here .-4
" 1 n2 n2 n-
T, . D 1.3 2.4 3.5 4.6 (»— 1) (n+1)
Therefore F*n_j ^ • yi • r •> •
/2
-JO+ - 3
Hence as «->-oo.
n (w+2)
Also />n = ^n-1-
//
8*18. Wcierstrass’s Inequalities. The following inequalities
are due to Weiestrass :—
1
> n (!+«„) > l+Su*
^ 1 -Zu«
(b) rr1 > n (i-w„) >

1 -\-£un
where 0 < un < 1 and £ un < 1 for every value of n.

We have
(I+Wi) (1+«’*)*= I+ (Wi+wa)+“iw* > H-(Mi+w*>
so that (1+M,) (l-FWo) (I+w3) > H + O'i + M (H-w3>
> 1 +(W| +I/3 + W3).
Proceeding in this manner, we shall have

(H-Wj) (1 + Ws) (l+w3).--(l+w») > I+(wi+w*+w»+ *-+w")
.. (1)
or n (1+m») > 1+ s u»-
1 1
In the same way, we have

(1—Ut) ( 1 — Mj) — 1 -(Wl + W2) + Wll/2 > 1-(W1+W2)
and (1 -Wjl (1 -U2) (1 —1/3) > [1-(«1 + W2/] 0 -t's)
> 1—(Uj + Ma+Ma) and so on«

(1-Wi) (1—Ms) (1—1/3). . .(1 -l/„)
> 1— (Ul + U2 + Ua+,..+U»)
or n (i -«„) > 2 u„.

1 1
Again
<1+H>)=T^<T^sinccl"u,,< 1-
Hence (1+w,) (1+m2) (1+w3). . .(I +un)

_1_
<(l -i/,) (1 -w2)...(l - wn)
_1_
^ 1 — (W1 + W2-H ... -\-un)
*
or IT (l+«„) <
1 1- ^
1
Similarly U-».)=T^< ITv

1
(1 - Ux) (1 - Wo)...(1 - t/J <
( 1 + Mi) (1 -f W2)...(l + Wn*
I
<
1 -Hwi + ^2+ . .. -f-wn)
1
or n (i -un) < n ...(4;
H2«n
1
From (1), (2 , (3) and (4), we have
> II (1-MO > i + û„

l-2un
1
and > ri (i-on) > i-run.
l+Zu,,
8*19. Wc now give three theorems on special types of
products. The first two theorems deal with products in which
the an are all of the same sign and the third theorem is general
in which an may be of either sign.
Theorem I. If c/n ^ 0, the series £a„ and the product
FI (l-f <7n) converge or diverge together.
When a ^ 0, we have 1-fa < ea. Thus
tfi + a2+ ...+ an < 0+fli) (1+*»)...(1-f an) < e«i+fl»+-+®«
or Sn < Pu < es". .. .(1)
Since Pn and Sn are monotonic increasing functions of n, they
will have a limit finite or infinite. They cannot oscillate.
Hence from (1) it follows that if Sn~* a finite limit, then P„
also does so and if Sn-+ 00, Pn also -*•<» and vice versa.
That is, the scries £an and the product II (1 -f °n) converge
or diverge together.
<*30 Complex Variable
Theorem II. If - 1 < an < 0, the series Zan and the product
II (l-ftf,,) converge or diverge together.
Let an=-bn so that 0 < bn < I.
Since 1 -b < e~b when 0 ^ b < 1, we have
0 < (l-A.Ml—M...U-A) < e-(*.+A*+-+A")
or 0 < fl+o,) (l+«2).. {!+*„) < e-(*»+A<+-+M
or 0 < Pn ^ e° +«.+•••+«!».
Now if Zan diverges it will diverge, to —00 and consequently
Pn —>0. Hence the product diverges to zero.
Next suppose Zan is convergent. Then for a given e we can
find a positive integer m such that
0O
0 ^ Z b„ > e.
t»=m
Also (1 -bm) (1 —bn+j) ^ 1 — bm — Am+i*

(1 -bm) (l-^+i) (i -bm+2) ^ (l-bm-bmi.1)(l-hm+8)
7^ 1 —b„, — /’m+l — bm+2
and therefore for n > rr,
(1 — bm) (1 — b„, f.j). .. (1 - bn) ^ I — bm — bm+i.. bn > 1
Pn
or > 1—€.
P m—1
Since each factor in Pn is less than unity, Pn < Pw-i anc^
consequently is monotonic decreasing and hence it must

* in -1
have a positive lower bound.

It follows that Pn tends to a finite non-zero limit. In other
words. II is convergent.
Alternative. We give alternative proofs of theorems I and II
above This proof is based on Weierstrass’s inequalities. (§ 8‘18).
These theorems can be combined in one statement as follows.
If tfj, ci<, <73,.. ,an. ..., is a sequence of numbers 0 and 1, then
the infinite products
FI (1 + r7n) and £ (1 — an)

i i
are both convergent or divergent according as the series Z un is
convergent or divergent.
Let Pn= IT (I and On— H (l-tf„)

i i
Case I. Let Zan be convergent.
Then there can always be found a positive integer m such that
-/?,„ = ~b r7mj_2“b • • • 1 •
By Weierstrass’s inequalities, we have

1 _ 1
1 1 i^ro+1 + ^m+2 4~ • • • "H^n)
> ( (1 +^»n+2>- • •( 1 ~\~an)
P„
i.e. ...(!•
and (^ ^m+l 1 ( 1 ^m+2 )•••(! dn )
> i ”“(^w+l+^m+2"f • • • +^n)

> 1 - Rm
i.e. y=p- > 1— (^mri fflm+2+ • • • +^n) > 1 ~ Pm.
Hence from (1) and (2), we have

u ^
p
A m . •% v
*n ^| n •••*->)
1 *'tn
and Qn > QmU-Rm) -..(4)
for II > ///.
But Pu increases as n increases, always remaining less than a
finite positive quantity — ■*- .
And Qn decreases as n increases always remaining greater

than a finite positive quantity Qm (I — P,n).
Hence Pn and Qn approach finite limits P and Q so that
Pand Q > j-% .

I -Pm 1 —t<m
i.e. II(l + nn) and 11 (1— an) are both convergent if 23 an is
convergent.
Case Ii. Let 23 an be divergent.
Then P„ = II (1 + o„) > ... -\-an > 1+23 an.

i i
Since E an tends to oo as n-+oo f we have
Pn-*-oo as n->oo .
n
Hence H (1 + <7n) is divergent.
and
-«i)0- <*2). ..(1 -an)
> (1+<7l) ( 1+<72). . . ( 1 (§8*18
> 1 + (tfi + °2 + • • • + rr„)— 1 + 23 an.

\
But E an tends to co as n-> oo and hence it follows from

above that
(1 -ffj) (1— tf2)(l -a3)... (1 -ao)-*0 as n-> oo,
OO
i.e. the product n (1 -an) diverges to zero.
Theorem III. If the series E an2 is convergent, then the product

II fl-\- an) and the series E an converge or diverge together.
Since E arz is convergent, we can find m such that
a2 < \ or | an | < $ for n > m.
For such values of n, we have
, an2 a„3 . a*
I log (1+*»)-*,, I = -f -y
< ^na{l+knl + I^P+-}

*„2
2(1- 1^1)
Hence the series 27 | log (l4-^n)~anl *s convergent. It

follows that the series E {log (14-^n)-^n) is convergent. That is,
]0a Pn—Sn tends to a finite limit, Therefore the series E an and
the product II (1 +<?„) converge or diverge together.
Note. The theorem III gives an easily applied test for the
convergence of an infinite product in which the an may be of
either sign.
8 20. General Theorem. If — 1 an < 1 and if E an2 is
convergent, then
(i) FI (1+an) converges when E an converges,
(ii) n (/+//„)-► + 00 when E an->+ 00,
(i/7)
n (l+an)-*0 when E an->- 00,
(iv)n (l+an) oscillates if E an oscillates
and if E an2 is divergent, the infinite product II (l + an) diverges to
zero if E an is convergent or oscillates between finite limits.
This theorem has been partially discussed in theorem III of
§ 8*18. Here we give another proof of the theorem.
Since log Pn—E log (1+<?„), P„ in the limit will be conver¬
gent or divergent according as E log (l+o„) is convergent or
divergent.
By Taylor’s theorem, we have
log (l + 0n) = log l+fln-2

a,
where 0 < V < 1.
2(1 +0.an)*
j Taylor's theorem states
f (a+h)=f{a)+hf' {<1)+^ f“ (a)+ ... + , ■/"-* (a)+/?„.
where ^ , /<n> (a + 0. //), where 0 < 0 < 1 J.
Therefore an-Iog (1 + <?„) = =a + Ve quantity.
Since 27 i/n is convergent, we can find an integer m such that

| an | < \ for n > /;/.
Then 1 + d. an > 1 - | | > \-\ti.e. >\
or
(1 +0.an)* " *
so that 0 < —log (1+^n) < 2tf„2.
m+p m+p m+1
Therefore 0 < E an— E log (H-<7„) <2 E a„*.
m f-1 m+1 m +l
Case I. Let 27 an2 be convergent. Then we can find m such

that
m+P
£ an2 < i €.
m+l
rn fp m+p
Thus 0 < 27 on- E log (l+tf„) < e.
m+1 m+1
Hence 27 an-+ 27 log (l-fon) in the limit.

It then follows that
(i) if 27 an is convergent, log Pn and therefore Pn is conver¬
gent,
(ii) if an diverges to oc, log Pn and therefore Pn diverges
tO Or ,
(iii) if 27 an diverges to - oo, log Pn diverges to - oo and

therefore Pn diverges to zero,
(iv) if E an oscillates, log. Pn and therefore Pn oscillates.
Case II. Let 27 an2 be divergent.
Since 27 an is convergent, we heve
I an | < \ for n > m.
l-\-dan < 1 4- \an | 9-
Hence an~log (1 +on) > §a„2
tn-J-p m-tP
or £ 0n- £ Iog(l-ffln) > f £ af.

tnfl m+2 m+1
It follows that
(i) if £an converges, log Pn diverges to — » and therefore
Pn diverges to zero,
(ii) if £an oscillates between finite limits, log Pn diverges to
- » and therefore P diverges to rero.
Example. Discuss the convergence of the infinite product
(‘-DO+OO-DO+D--
Here £an=£ (—l)fi-! = —£ + £-•• •
which oscillates between -4 and zero.
But £an2 = £ 2? ==4+4‘ + 4“^ • * *
which is divergent.
Hence by the above theorem, the given product diverges to
zero.
8 21. The absolute Convergence of Infinite Products.
The product n (1 +an) is said to be absolutely convergent if
the product n [1 4-| an i] is convergent. It should be noted that
the infinite product n + is absolutely convergent if the series
£an is absolutely convergent. Now we prove the following
theorems.
Theorem I. A necessary and sufficient condition for
Z [ log (1 + alt) \ to be convergent is that £ | an | is convergent.
The condition is necessary. Let £ \ log (l4-tfn) I beconver-

gent so that log (I + and l+an->l. Hence an->0. We
can find m such that for n ^ m, | an | < i.
Therefore when n ^ /», we have
log (l+rt„)=0n-.|tf„2+itfn8-K • • 9
log (14 an)

and so -1 = 1 -\an + \an2-
an
1 1 log f I A-an\ 3
Then .. .< 1)
an ^ 2*
Hence I an | 2 | log (1+0 i, when n > m, and therefore
an | is convergent by comparison test.
The condition is sufficient. Let Z \ an ! be convergent. Then
^,,-►0 and again from (i), we have
I log (1 -\-an I ^ a ; wlien n ^ m.
Hence Z | log (1+fl,, |) is convergent.

Note. It follows from the above theorem that the infinite
product n (1+0 is absolutely convergent if the series
Z log (!+</„) is absolutely convergent.
Theorem II. An absolutely convergent infinite product is
convergent.
Since the product II (1 -f | +, 0 ,s convergent, the series Z \ an '
is convergent by the theorem of § 4'5.
Therefore by the preceding theorem, the series Z log (1 + 0
is convergent. Thus the series Z log (1 + 0 and hence the
product II (1+0 is convergent.
8*22. Derangement of Factors.

The factors oj an absolutely
convergent product II (i+O may be rearranged in any ordtr with¬
out affecting its convergence or sum.
Since the product ll(l+|tf„i) is convergent, the series

Z | an f is convergent by theorem I $819. 1 hen from theorem I
of§ 8*21, it follows that the series Z log (1 f a„) is absolutely con¬
vergent. Therefore the terms of this series may be deranged
without affecting its convergence or its sum. It follows that the
order of the factors of absolutely convergent product 11(1+0
may be altered without affecting its convergence or sum.
8*23. Semi-convergent Infinite Product. If Z an is conver¬
gent and Z | an | is divergent, II (1+0 is said to be semi-conver¬
gent.
We now give the following theorem

Derangement of the factors of a non-absolutely convergeDt

Infinite Product.
Theorem. If the factors of a non-absolutely convergent infinite
product are deranged, there is alteration in the value of the product
and the value of the product may be made to converge to any value
or to diverge or to oscillate.
We consider the case of the infinite product II [l + ( - l)n_1fln),
where an is positive ard lim (na„)=g, a finite quantity.
Let />= n [ I + ( - I )n~l an],
Suppose Q is the value of the product when the limit of the
ratio of the number of positive to the number of negative terms
is k. We then have
^ 4" ^2n+l 1 0 ^2n+3^ • • *0 4* ^2m~l)»
where
Now Zaf is clearly convergent and therefore it
follows that
(^72fi+l + f72n+3+ • • • 4-tfom-l) —log ((1 +°2r»+l) (1+^2i;+3)* • •04"*72in-l)J
can be made as small as we please by taking n sufficiently large.
Again by Pringsheim’s method it is evident that
Lim (tf2„+1+tf2«+3 +.. . + = log k,
and hence = *£-Iog
or
Hence to have the value Q of any infinite product having

positive and negative factors alternately, the factors should be
deranged in such a way that the ratio k of positive and negative
factors may be given where lim nan =g.
8‘24. Complex Factors. The absolute convergence of

infinite product of complex factors has the same definition as
that of real /'actors except that S | an I means the modulus of the
complex number an. Thus if au a2...are complex numbers
whose moduli are denoted by i ax |, \ a2\, ..., the infinite
product II (l-J-tf„) is said to be absolutely convergent when the
product II (14-1 an |) is convergent. Again the product II (I +tfn)
is absolutely convergent if and only if the series Zan is absolutely
convergent.

I. Prove that
converges obsolutely for any value of x, provided that c is not a

negative integer, and that
is absolutely convergent if \ x | < 1, (Agra 48, 56; Jaipur 55)

We have
0 1 -^X1 + • * •]
£_ .. . ”-c xti
= 1 + n \n-\-c) In1 (/i-+-<•)*
c (n—c) x2
Here an—-: -4- . . .
n <« + c) 2/r (n+ 1)
1
Take bnz= -j •
n1
a.
Now lim = | cx lt a finite quantity.
n
Therefore 27 an is absolutely convergent.
It follows that the product fl (I+d„) is also absolutely conver¬
gent.
Again the porduct n [^1 J is abs0,ute,y convergent

/ fix
if the series 27 ( J is absolutely convergent.
/ nx \"
Here
nx
(an)"n=
n+l'
(on)lln>=x.
Hence if | x | < 1, 27 an and therefore n (1 +an) is abselutely

convergent and if | x ) ^ 1, Z an and therefore the product is
divergent.
2. Test the convergence of the product of which the general
term is
(Agra 40)
Here 1 -f
\ Z2 1 23
— 1 ——. ~—o +=. • •
2'/i*w* 3Vir®
1 21 1
-
2’iiV +3#ii*ir® "**•
1
Comparing 2? /?„ with 27^ we see lât
Lim '-^-'=^1 r2 | , a finite quantity.

| 1 I I™*
\~^\
Hence E an ami therefore n (l+<7„) is absolutely convergent.
3. Discuss the convergence of the infintte products
to (/-/)(7+/)(7-f)(/+S)"-
(Delhi M. Sc. Final 1959)
OO /-// —z
oo rn >2zl
(Delhi M. Sc. Final 1959)
w ^ {" •1
(i) Here 27 an = z- z^\~\ + \~\+ - • •
We then have S2n —0
and S..nH —*0 as /?-> ©c.
Hence 27 converges to zero.

Also £i7(1*=22»(l + .i + 312 + ...)
which is convergent.
Hence the given product is also convergent.
Again the series E | an | =2 , r E ~ is divergent and the series
E | an i2 is convergent. Hence the product FI (1 -f | an 1) is diver¬

gent.
It follows that the given product is semi-convergeut.
Here 1 + <?„ = -—1- e2*ln

n+z
= \+-2z- + l*(z-z'l).+
2
^n (n + zy nUn + z)
Comparing £ an with £ bn = ~, , we have
Lim - ■. = 2 | z- | , a finite quantity.
But the series £ is convergent and therefore the series £au
is absolutely convergent.
It follows that the given product is absolutely convergent for
all values of z,
4. Test the convergence of inf nit e product
n \
(Agra 1938, Jaipur 1955)
, x + x*n x-1
Here , ■
l+^n= 1+X =1 + \+X ■it* ‘
1 .
Now when at | > I, £ — is convergent
X
and since X-l . _j.ll

£ la- I =£
1 + x 2n X •n
the scries £ an is absolutely.convergent. It follows that the given
product is absolutely convergent.
_ A ( 1 - A**1—1)
If | a- | < 1, l+an=
i+A 3 n
and an does not tend to zero. Hence the product is divergent.
If *=1, every factor is unity ; hence the product is convergent.
If a= —1. every factor is zero and an-> - 1^0 and hence

an is divergent. It follows that n (1-f an) is divergent.
5. Show that the inf nit e product
is absolutely convergent.
Here 1-fa,
=0 "5=r)[,+^3+«-/». 21—
i
]
4-
I • • • 9
2,,4/3
1
then 4-...
2n3li
1_
Take bn= We then have
n 4/3 *
Lim a finite quantity.

\bn I 2.
But the series 27 bn is convergent.
Hence the series 27 an is absolutely convergent. It follows
that the given product is absolutely convergent.
6. Show that
converges absolutely for all values of z situated outside a circle

whose centre is the origin and radius unity.
Here a»=(l
(• -;r —(n-fl)
.z.
.*. Lim -^2- =-. ! r

^n+l *
Hence if |z|> I, the series 27 and consequently the
given infinite product is absolutely convergent.
7. Express! (/i-M)* as a product of n factors. If

(«+x)!
the number of factors is increased indefinitely, show that the
product is absolutely convergent.
H f v f
m+ir
n l.x ! («+!)*
x !.(x + 1) (x+2). . .(x-f n)
1.2.3...n
(x+ 1) (x-f2). .(x-\-n)
_1_ 2 3* __3_ 4* n (n+1)*
"x+r *x4-2’2z*x+3* 3®* ’ * n-fx* /i*
=K^Xn^J=n(,-<,")-say-
then l + "-.= (“Xl+«) =(‘ + «) (* + 0
=0-^-
=>+^fin+oG0-
Comparieg Ean with E K, we have

Lim | nlan |=£ | a2— x j, a finite non-zero quantity.
if xÔ.
Hence E an and therefore n (!+*„) >s absolutely convergent.
»■ y'-C'-iX'+sX'-sX'+aX'-feX'+s)-
show that the product
( ~ l)(‘ -
:1 Sr)( 1 + IX11 “ 1 " *0(1 + >) ■ ‘ •
= P.exp^—^ log 2^.
z2 z2
Since the series 2*n*
z z z
and 77 + 77 277^ 2W
are both convereent, the product
is also convergent.
('-SO+SX'-sX'**)-
But this product is not absolutely convergent since the series
Li! . LfJ . lzJ_|Ai-l+ .. - is divergent.

TT ' TT 2™ ' Z7T
Further n„=^. or ~~ according as « is even or odd.
C^>
Hence Lim na.^=g. say, whether n is even or odd.
When we derange the factors such that one positive factor

follows two negative factors, we have
k = \.
Hence if Q denoted the changed value, we have

©i n-urw
= P.2~‘n
or Q=P.exp (-^Jog 2^.
9. If P=Yl ^1 —, show that if m—>oo and n—>oc in such
a way that
Lint ') = k> o finite quantity.
then Lim. oo f //->co ff# (\

\ Tit /
///e </««■/; indicating that the factor for which r=Ct is omitted.
Since 2? —-2 is convergent, the product />=I7 (1—pg^g) is
absolutely convergent.
-oo-s) (•+£)■■■
as shown in Ex 8 above.
But this product becomes semi-convcrgent if it is written in
the form
00 / z~ \
P— nil- -T7—r, ) can be written in the form
i V r'lr-J
O-DO+OO-Ô+s)
only if we assume that an equal number of positive and negative
values of r taken in the result
— Lim/>;_>co , M->oc
The product expressed in the form (2) above will also have
the value P.
But if m is positive and n negative, values of r are take where

m^tn, then the product thus obtained will be altered since it is
now semi-convergent.
In this case, un= -- when n is even

Wm
2
and , when « is odd.

(« + 1).T
2Z
Lini //—>y: nan= — —g, as in Ex. 8 above.
TT
Hence Lim inu-+x> iVmV(I -h '—^ =P.k‘n=P (kr,n

f-: - HI \
f-S '"V
since l im ^~^ = /c.
8*25. Uniform convergence of infinite products. The infinite

OO
product 11 /j to be uniform convergent in any
>i—i
hounded dosed domain D if Pn (z) = II (I +«r (j)} f<v/r/.c uniformly

r=l
to P (z) in D. Of course, Pn (z) must not tend to zero for any z

in D.
The following theorem can be used to test the uniform con¬
vergence of an infinite product.
A sufficient condition for the uniform convergence of an infinite

product.
Theorem. If the ser es 2' | //„ (r) | is uniformly convergent
in a bounded dosed domain D, then so (.Iso is the product
II {1+M„ (Z)}.
Since the series £ u„(z) | is uniformly convergent in Z),

we can find a positive integer m, independent of z, such that for
n > rn, we have
; utll (z) | + | umhl <z) |+ ...-H It,, <Z) I <
for all values of r in D,
Now | u,„ (z) |, i Mm+1 (z) I,.. .| u„ frt j are all positive, so that
we have
I uu (z) | < 4 . . (1)
for every n ^ ni and z in D.
For such value of n and z, we get
log (z)) (z) - } {un £ \u„ - ...
Hence | log {! +m„ (z)} | < I un (z)+ 1 i utt (z) f-f ...
< | un (7) ! + !n„ 1tc) i8+...
__ I un (z) |
1- I IZ) I
< 2 | u„ (z) | from (1) ...(2)
Since the series 27 | un (z) | is uniformly convergent in Dy it
follows from (2) that 27 | log {I +un (z)} |, and therefore
27 log {1 -f-un (z)} is uniformly convergent in D. In other words,
n
log Pn (z) where Pn(z)= U {I 4 un (z)} converges uniformly to
r= 1
a limit which we may cill log P (z), that is, Pn (z) converges
uniformly to P (z).
Note. If the infinite product II {l-ft/n (z)} converges
uniformly to P (z) in every closed region within a closed contour
C, and if each factor of the product is an analytic function,
regular within C, then it follows that P (z) is also regular within C.
8*26. Sohed Examples.
1. Show that the infinite product
0-£X'4)OS0-
converges absolutely and uniformly in any bounded closed domain
D which contains more of the points ±1, ±2,... whereas the
product
converges uniformly but not absolutely in the same region.

Since the region D is closed and bounded, there can be
found a constant/? such that | z| < R for all z in D. Then
~2 I fti z2
-r, But 27 -y is convergent. Hence the series 27 -y is
uniformly and absolutely convergent by /l/-test. It follows from

°° / Z2 N
§ 8*25 that the finite product n 1 —J is uniformly and abso¬
lutely convergent in D, not containing the points ±1, ±2,

±3...
On the other hand, the second product does not converge
absolutely in A since the series R (1+l-l-£ + £ + & + | + ...) Is
divergent. However, this product is uniformly convergent in D
as shown below.
n
Let Pn (z)= U
'=1 0 4)00
-
Then Q,n (z) = Pn (z) and Q2n+l (z) = ( 1 Pn (z) and
so the sequences Qx (z), Q3 (z), Oi (z)... and Qt (z),04(z),

Q6 (2) both converge uniformly to P <z). Hence the infinite
product n ( 1 - 1+ converges uniformly in D to P (z)
2. Show that the product
converges absolutely and uniformly in any bounded closed region D

which does not contain any of the points ±1, ±2. . . .
We have
= 1 -un (z),
OO
where «„ (z)= Z '-=4 (-X

r—i r ! \nj
r— 1 1
Now <
r ! r (r -2) l (r-2, !•
oo
Thu. I .. (z) I < E ^ '■ if z I </?.
( 1 +n) e~c'n==l (*>•
cR2
where l'n(Z> \<~1.
eR1 .
But the series £ is convergent. It follows that the given
product converges absolutely and uniformly in every bounded and
closed region not containing the points +1, ±2.
3. Prove that n n ) represents a regular fundr/Y>//

i
n=i \nx+nx — lj
•n the domain R (z) > 2.
» r 9 | 09
We have n \ 1+-- = n {I+*#, (ar)}.
where
“■ sqWr
Let D be any bounded closed domain R (z) > 2+8, where

8 is any positive number. For any z in D, we have
!«z+//2- 1 | > | ii* |-«*-1
= j U*+'» j -/22- 1
«JI*-/IS —1
> n2+s-«2-1
1
or
/l'+/*2 — 1
for every /; and every z in D.

But the series 2 ( & ^ is convergent as may be
seen by comparing it with 2 —Xr* • Hence by Af test, the series

/r+0
2 ——- —— is uniformly and absolutely convergent in D.
/r+/r- 1
It follows that the given product is uniformly convergent in D.
Also since each un (z) is analytic in D, the given product is analytic
in the domain R (z) > 2.
Exercise
1. Show that the product
converges to I. (JRsij. 50)

2. Show that the infinite product
e,+l. e-1-,'2. e*+l /a. e~l~m...
oscillates.
[Hint, log />=(! + !) - (l+|)+(l + i)-(l+i)+...which
oscillates between log 2 and l+log2, so that P oscillates
between 2 and 2e],
3. Test the convergence of the infinite products :—
(i) »(i+n4). ,ii, n(i+A).

Ans. (i) Convergent, (ii) Divergent.
Show' that the infinite product
oo r J _o/n
1 is convergent [a > - 1J.
nil (jog (l+tf/«)
Prove that H ^1 + jQ e~x,n is absolutely convergent for any

fixed value of (Agra 35, 45)
[Hint. Proceed as in Ex 2 of § 8-24.]
Test the convergence of the product
Ans. Absolutely convergent.

Discuss the convergence of the following products
r • -
:—
3in:l
(i) {frS}'
2+Z"1
1+2 (ii) Iljl + -
w-
fsin (z/n) |
Oionjcos^)}, (iv)
l U/'O j
Aus. (i) Convergent for z > 1 and divergent otherwise.
(ii) Absolutely convergent for I (z) > 0 and divergent for
for I (z) < 0.
(iii) Convergent for every r.
(iv) Absolutely convergent for every r,
Show that the product II ^1 +^-^ converges absolutely, if
7? (z) > 1. Also show that it is uniformly convergent in the

closed domain R (z) > 1+S, where 8 is any positive number.
[ I lint. We have shown in Ex. 7 of $ 8*14 that the series
E is absolutely and uniformiy convergent for ! z | ^ l -f s.
It follows that the product II (l+Jr) is absolutely and uni-

formly convergent for | z | > 1 + 8.]
9. Discuss the convergence of
where c, is constant.
Ans. Absolutely convergent for all values of r if r is a
constant other than a negative integer.
CHAPTER IX
ELEMENTARY TRANSCENDENTAL FUNCTIONS
9 1. In the earlier chapters we assumed certain properties

of the elementary transcendental functions. The results used
were some what intuitive. We now discuss some of the proper¬
ties of the Exponential, Trigonometric, Hyperbolic and Loga¬
rithm ic functions of a Complex Variable.
9*2. The exponential function, exp z. The exponential
function of a Complex Variable is defined by the power series
—2 -3 00 2n
exp z= 1 +n+2l+5“!+ ' • • n !
(i) The series given in (1) is absolutely convergent, for

n-f 1
u, n !
L im /;->cc = Lim /7->00
u nrl (n- l) !Xz" |
ln
= Lim - = OO
for all finite values of i z

> 1
and the absolute convergence of the series follows from
D’Alembert’s ration test.
Further, the above series is uniformly convergent throughout
any closed bounded domain of zy for if the domain is defined
by i z | < R, then
But the series Z — is convergent by D’Alembert’s ratio test.

n !
oo
Hence by Weierstrass’s A/-test, the series Z is uniformly con
nô ,l •
vergent in every bounded closed domain.
Again since each term —, is a continuous function of z for
fixed n, it follows from the uniform convergence of the series that

Elementary Transcendental Functions 449
the exponential function exp z is a continuous function for all

values of z and hence we can write
Limz-¥z0 exp (z)=exp z0.
The function exp z is also analytic for every value of z and

. d
we have — expz = expz.
dz
9*3. The addition theorem for the exponential function. We
have to prove exp (zj + z2)=(exp zA).(exp z2».
Now exp Zj^l-f zi + ^"j + • • • + + • ••

Zo* z n
and exp z2=\+zt+—: + .. * + ^77+ • • •
W • Ww •
Since these two series are absolutely convergent, it follows

from Cauchy's theorem on multiplication of absoluterly conver¬
gent series that
/ z % z n \
(exp zx) (exp + • + /pj+--«)
x(l + n+2i!+-"+?'!+
_—1, +, 1+i +, W + 2ZtZt
^1 ^2
or
+ Zt) ,
‘ 4
, . (*i + Z2) (Zi-t-Z,)*

=1+ rr+ 2s
= exp (Zj~f-z2).
From the above result, we sec by induction that
(exp zA) (exp zt).. .(exp zn) = exp (zx-\-zt+ ... + zn).
And in particular,
(exp z) {exp ( —z)}=cxp (0)= 1.
From the last result, it follows that exp zÔ, for if there
were such a value of z, since exp ( -z) would exist for this value
of z, we should have 0=1, which is absurd.
9*4. Trigonometrical functions sin z and cos z.
The functions sin z and cos z are defined in terms of the
exponential function as follows :
exp (/z)-exp ( — iz)
sin z
2/ - ’
...d)
exp (iz)-fexp (—iz)
cos z ...(2)
Using power series for the exponential functions on the

R. H. S. of (i) and (2), we easily see that
*r3
s,n 2 2 j J~^5 l • • • 4“ i\ i +••• »
(2/itO !
( — 1”) Z2”
cos z = l-r+r + ...+
2 n 4!. (2n) !
~h...
Just as in the case of the exponential functions the series for
sin z and cos z can converge uniformly in any bounded closed
domain of values of z and consequently sin z and cos z are
continuous functions of z for all values of z.
Again it may be proved in a similar way that the function
sin z
• •
~1 3!+5!
defines a continuous function of z for all z, and in particular, this
function is continuous at z=0 and so it follows that
.. sin: .
L'mz->0 —T=1-
Further the functions sin z and cos z are analytic in every
domain and so ~ (sin z)=cos z, (cos z)= -sin z.

az az
Note, sin (-z)= - sin z, cos (— z)=cosz, sin 0 = 0, cos 0 = 1.
Euler’s Equation.
From (\) and (2), we have
exp (/z)=cos z+i sin z,
which is known as Euler’s Equation.
Thus if z=x+i)\ we have
exp (x+/>')=exp (x) exp) (/»
=exp (x) (cos y+/ sin y)
and so exp x is the modulus and y the argument of exp (x + /».
9*5. Addition Theorems for sin z and cos z.
We have exp {/ (Zj + z2)}=exp [izx) exp {iz%).
Using Euler’s theorem, we get from (1),
cos (Zj+Zal + i sin (zl + z2) = icos zl+i sin zx) (cos z2+/sin z2)
= icos Zj cos z2-sin zx sin z2)
+ / (sinZj cos z2+cos zx sinz 2)....(2)
Changing zx into — zx and z3 into -z2, we get from (2),
cos (Zj + Zj) -/' sin (z4 + za) = (cos zx -i sin z,) (cos za-i sin z2)
=cos zx cos z2—sin zx sin z3
- i (sin zx cos za + cos zx sin z2).. .(3)
Elementray Transcendental Functions 451

cos (zx-\-z2) —cos 7j cos z2-sin z, sin z2, .. .(4)
and subtracting (3) from (2), we get
sin (Zi + r2) = sin zl cos z2-+ cos zx sin z2. .. .(5)
It may also be proved in a similar manner that
cos2 z + sin* z = 1. • (6)
...(6)
9*6. Hyperbolic Functions sinh z and cosh z.
These functions are defined by the relation.
l _ exp (~) + exp (-7)
cosh 7=
coslw=--- . ...(())
1)
. , exp (r)-exp ( -z)
sinh 7 =
s'„h-' =-2- ...(
...(2))
Writing power series for exp (7) and exp ( -7) on the R.H.S.
of (1) and (2), we easily see that
Z- 2* Zzn
cosh = 1 + +.-.+ • • • +• ...(3)
2a »1
•••(4)
sinh 7=7+ -r-j +5T + --*+j2w+"i)T.(4)
Clearly the power series for cosh z and sinh z iis uniformly
convergent in any bounded closed domain. It also follows that
these functions are continuous and analytic in every domain.
9*7. Relation between Hyperbalic and Trigonometrical
functions.
It can be easily verified that
sin (/z) — i" sinh z, sinh (/z)= i sin z,
cos (/7)=cosh z, cosh iz =■ cos 7.
9*8. The periodicity of sin z and cos z. A number A is said
to be a period of/(z) if/(z-(-A)=/(z), for every z.
If A be a period of sin z, then
sin (z+A)=sin z,
for every z. Putting z = 0, we see that A must satisfy
sin A=0
A = «7T.
Again sin (nn-\-z) = ( — 1)" sin z
= sin z, only if n is even.
Thus the periods for sin z are 2nn where n is any integer.
The number 2tt is called primitive period or simply period.
Similarly it can be proved that 2n is also a primitive period
of cos z.
9*9. Periodicity of exp (z).

Let A be a period of exp (z). We then have
exp (z+ A)=exp (z) for all z.
Putting z=0, we see that A satisfies
exp (A) = I.
If A=a-f-/,3, we have
exp (<*+//?) = 1
i-e' exp (a) fcos /?+/ sin /?]= I.
This gives
exp (a)= 1 and f3=2mr.
Thus we have
a=0, (3 = 2nir.
It follows that a-f-//?=2m77 is a period of exp (z). The number
2vi is the primitive period of exp (z).
9'10. Logarithmic Functions. {Inverse of exp (z)}.

If iv—exp (z), then z is called a logarithm of w written as
Log w=z.
From the above definition, it follows that Log w is defined
for every value of iv except 0 and that it is a many-valued
function.
Suppose now w is given, we are to examine the existence of
z-=zx+iy such that
exp (x+iy)=w.
This gives
exp (jr) fcos y+i sin y]=w
From the above equation, it follows that y is one of the
values of arg w», whilst I m \=exp (x) so that x=log I w |. Thus
every solution of the equation w=exp (z) is of the form
z=log I w | + ; arg w.
Since arg w is a many-valued function, it follows that there
are an infinite number of logarithms of the complex number wy
each pair differing by a multiple of 2n7.
We shall write
Log w=log j iv l+i arg w9
so that Log iv is a many-valued function for each value of iv.
Branches of Log w. Each value of Log iv, obtained by a special
choice of the many-valued function arg >v, is called a branch of
Elementary Transcendental Functions 453
the logarithm. The most important branch is the principal value

of Log w. We denote this principal value by log w since it is
identical with the ordinary logarithm when w is real and positive.
Now the function log | w | is continuous except at the
origin ; but when w-»-0, log | w | -*--00. Also the principal value
of arg w is continuous except at points on the negative half of the
real axis ; for, if u < 0, v > 0 where w=u+iv, we have
Limv_>0 arg (u-f/v)=tt
and Limô arg (u -iv)— - w.
Hence, if and w'-+w along any path which does not

cross the negative real axis, then log w'-v-log w. But if the path
from w' to w crosses the negative half of real axis once,
log w' ■►log w±2nit positive and negative sign being taken
according as the path crosses from above to below or from below
to above. Thus we can pass from the principal value to any other
branch of log w by making a number of circuits about the origin.
On account of this, the origin is called the branch point of log z.
Analyticity of Log w. We make a cross cut along the real
negative axis from - 00 to 0. Then if L>is any bounded domain
in this cut plane, so that no point of the cut belongs to D, log w
is one-valued and continuous in D. Let w, w' be any two points
of D. Let z% z' denote log \v and log w' respectively, then z'-*-z
as w'-+w along any path in D. Hence
_ z-z'
w—yv' exp (zj-exp (z')
^_z-z'_
iZ~Z'> + 2T (2’ ~ z’%) + JJ u’-- :'3) + •• •
_1_
(z + z')+ ~(za+zz'+z'a) + ...
1 + FT +Ti +••
z—z 1 1
Thus Lim
w - w' e* w
Hence log w is analytic function, regular in Z), which has the
function — for its derivative.

w
9*11. Complex Exponents. The function z°. We define the
function z° by the equation
z°=exp (a Log z), • • (1)
where z are any complex numbers whatsoever,
Now Log z = log | z l+i (a + 2mr),
where a denotes the principal value of arg z.
2°=exp [a log 1 z [ + iax-\-2nrrai]
=exp {a log | z ij.exp (fox).exp (2mrai). .-.(2)
Thus, in general, za is infinite-valued and different values of
are obtained by giving different integral values to n in (2).
m
When the exponent a is a real rational number, a=
formula (1) becomes

z»n/«=exp^log Z^, Zz£0. .. .(3)
In this case for various values of Log r, the right-hand

member of (3) takes on just n distinct values, the numbers
If z = r exp (W) and 60 is a real constant, the function
Log z^log r+W (r > 0, K < 6 < e°+2n' ; • •( )
is single-valued and analytic in the domain indicated, as is the
composite function exp (Log z). Thus the funcuon z^ e ned
by the equation (1), in which Log z .s given by eja.ioi ^ .s
singled-valued and analytic in the domain r > 0, 00 < 0 < W 2 •
The derivative of that branch of the multiple-valued powe
function (1), by written in terms of the logarithms defined by
can
equation (41. Thus exp (« log z)

— z°-(a log z).z a exp (i0g z)
= a exp [(a—1) log z]=aza~1.
In particular, when 80=-«, so that -n < » < ”, the
function z°=exp (u log z) (z-0) is called the princpal branch of
the multiple-valued power function given by (1). It is single¬
valued and analytic in the domain r > 0, —n < 6 < tt.
9*12. Inverse Trigonometric Functions.
We define the function n^sin-1 z as the inverse of the sine
function, so that we have
2= sin >n
Elementray Transcendental Functions 455
We can express the inverse function in terms of logarithms.

Thus (1) gives
eiiu>-2izeiw-l=0t
or eit0=iz+(l-z1)1'8,
or H,==sin~1 z = —i log (/z + (l-z2)l,aJ ... (I)
which is a multiple-valued function with infinitely many values.
en specific branches of the square root and the logarithm are
used, the function (1) becomes single-valued and analytic because
n is a composition of analytic functions.
In the same manner the inverse to functions cos w and tan w
are found to be
cos-1 / log [z+(z2- I)''*],

286121-Functions of A Complex Variable - Text

Uploaded by

Copyright:

Available Formats

286121-Functions of A Complex Variable - Text

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

286121-Functions of A Complex Variable - Text

Uploaded by

Copyright:

Available Formats

FUNCTIONS OF

Dr. R. K. GUPTA, J. N. SHARMA,

Third Revised and Enlarged ]§ditioo~-

KRISHNA PRAKASHAN MANDIR

23 . Convergence of Infinite Series /

Price Rs. TEN Only

PREFACE TO THE SECOND EDITION

COMPLEX NUMBERS AND THEIR GEOMETRICAL

1*1. Introduction. The extension to the concept of complex

The concept of complex variable was usefully applied to great

Amongst the great mathematicians who have done immense

Now assuming the theory of the aggregate of real numbers;

These give x' = —^—., ,>■'=- » provided (x, y) does not

represent the complex number (0, 0).

and is denoted by (a, v) l.

=(v. y)+(-x\ ->■')

We have ‘, = (x, y) (a', y'r1

1*5. Modulus and argument of a complex number.

It follows that I 2- I =0 if, and only if, xr = 0 and y= 0.

is called the Principal value of the argument.

1*6. The geometrical representation of complex numbers.

• ,N,°'C- , TJle 1comP|')l number z is known as the affix of the

f lhIf/>,S tf!e Point (** y) on the Argand plane corresponding

OP respectively and vice versa.

difference, product and division oHwoiompllVZmte7st,n8 ^ ^

the points P and Q on°theA^rgan^plMc^ ^ ** ^ rCpreSen'Cd by

Complete the parallelogram

by the vector OP where the points P and Q represent the comp¬

• Another interpretation of (z,- z2).

we get iz=r £cos + + / sin

Hence multiplication of z with / rotates the vector for z

• Division. Since^-=^-[cos (Bt -*,)+/sin (0,-6,)], we obtain

the point R representing the complex number in the following

R is the point representing §-. For, we

have from similar A'" OPQ and OAR,

Geometrically the conjugate of z

1*10. Properties of Moduli. We now prove some basic

Theorem II. The modulus of the difference of two complex

I’ll. Properties of the Arguments. We now prove the follow¬

= r,r2[cos (01 + 02+..-+*») + « sin (*i+** + • • • + *•)!•

= -1 (cos 0x + i sin 0,) (cos 02-i sin 02)

1-12. Solved examples.

[see Ex. 1 above].

(ii) _ 1+2/ _ 1-f 2/ _ I + 2/

Let z=x + iy, then ?=x—iy, where x, y are real.

Let a = r cos 6t 0=r sin 0

Therefore the general value of 0= 2mr where//k

Geometrical Interpretation. Let P and Q be the points of affix

5. Show that arg ^ is ^te ongle between the lines joining

the points a to z and b to z on the Argand plane.

— tr B ^ n through which the vector BP has to rotate to coin¬

Thus arg is the angle between the lines joining a to z

and b to r taken in (he proper sense.

If AP coincides with BP, then

so that \^ is purely real.

It follows that if ^ is purely real, the points A, B, P are

Let zv z2. z3 represent the

= A- (**“**> ^ i (^I + »l) (Z2-Z3>

=5JSz' (*.-*.>+£ SM*.-*,)

= r,r2[cos (01 + 02+..-+») + « sin (i+** + • • • + *•)!•

= A- (“> ^ i (^I + »l) (Z2-Z3>

=5JSz' (.-.>+£ SM.-,)

2\ + 2'+ Z2 = ZXZ2 + % + -3-1

xI'yz'LT the affiXeS 0{A’B’C and • y. those of