n a' s

ri s h
K TEXT BOOK on

Complex Analysis &

Calculus of Variations
rd
(For B.A. and B.Sc. III year students of all Colleges affiliated to
Shri Siddhartha University)

As per Latest Syllabus of Shri Siddhartha University

By

A.R. Vasishtha Kaushal Srivastava

Retired Head, Dep’t. of Mathematics M.Sc., Ph.D.
Meerut College, Meerut (U.P.) Retd. Head Dep’t of Mathematics
M.L.K (PG) College, Balrampur (U.P.)

Sanjay Kumar Singh Prateek Mishra

M.Sc., Ph.D. M.Sc., Ph.D.
Asst. Professor & Head Dep’t of Mathematics Balrampur (U.P.)
Shivpati (PG) College, Shohratgarh,
Siddharth Nagar (U.P.)

(Shri Siddhartha Edition)

,
KRISHNA HOUSE, 11, Shivaji Road, Meerut-250 001 (U.P.), India
 Jai Shri Radhey Shyam

Dedicated
to

Lord

Krishna
Authors & Publishers
 P reface
This book on COMPLEX ANALYSIS & CALCULUS OF VARIATIONS
has been specially written according to the Latest Syllabus to meet the
requirements of the B.A. and B.Sc. Part-III Students of all colleges affiliated to
Deen Dayal Upadhyaya Gorakhpur University, Gorakhpur and Shri
Siddhartha University, Siddharth Nagar in U.P.

The subject matter has been discussed in such a simple way that the students
will find no difficulty to understand it. The proofs of various theorems and
examples have been given with minute details. Each chapter of this book
contains complete theory and a fairly large number of solved examples.
Sufficient problems have also been selected from various university examination
papers. At the end of each chapter an exercise containing objective questions has
been given.

We have tried our best to keep the book free from misprints. The authors
shall be grateful to the readers who point out errors and omissions which, inspite
of all care, might have been there.

The authors, in general, hope that the present book will be warmly received
by the students and teachers. We shall indeed be very thankful to our colleagues
for their recommending this book to their students.

The authors wish to express their thanks to Mr. S.K. Rastogi (M.D.),
Mr. Sugam Rastogi (Executive Director), Mrs. Kanupriya Rastogi (Director) and
entire team of KRISHNA Prakashan Media (P) Ltd., Meerut for bringing out
this book in the present nice form.

The authors will feel amply rewarded if the book serves the purpose for
which it is meant. Suggestions for the improvement of the book are always
welcome.

— Authors

(vi)
 Syllabus
Complex Analysis &
Calculus of Variations
Deen Dayal Upadhyaya Gorakhpur University, Gorakhpur
& Shri Siddhartha University, Siddharth Nagar (U.P.)
B.A./B.Sc. IIIrd Year, Paper-IInd

Section A

Complex Analysis
Analytic function. Cauchy-Riemann equations. Harmonic functions. Complex
integration. Cauchy's theorem. Cauchy's integral formula. Derivatives. Taylor's series.
Laurent's series. Liouville's theorem. Morera's theorem. Zeros and singularities. Poles
and residues. Cauchy's residue theorem. Contour integration. Rouche's theorem.
Hurwitz theorem. Jensen's theorem. (4 questions)

Section B
Expansion of simple functions in Fourier series. Fourier transform and its simple
properties. (2 questions)

Calculus of Variations
Functionals. Variation of a functional. Euler's equation. Case of several variables.
Natural boundary conditions. Variational derivative. Invariance of Eular's equation
under transformation of coordinates. fixed end point problem for n unknown
functions. Variational problem with subsidiary conditions.
Isoperimetric problem. Finite subsidiary conditions. (2 questions)

(vii)
 B rief C ontents
Dedication.......................................................................................................(v)
Preface........................................................................................................(vi)
Syllabus (DDU Gorakhpur & Shri Siddhartha University)...........................................(vii)
Brief Contents.............................................................................................................(viii)

Chapter-1 : Analytic Functions .......................................................C-01—C-40

Chapter-2 : Complex Integration ...................................................C-41—C-96

Chapter-3 : Zeros, Singularities and Rouche's Theorem ................C-97—C-124

Chapter-4 : The Calculus of Residues.........................................C-125—C-178

Chapter-5 : Fourier Transforms ..................................................C-179—C-210

Chapter-6 : Finite Fourier Transforms ........................................C-211—C-220

Chapter-7 : Fourier Series ..........................................................C-221—C-246

Chapter-8 : Calculus of Variations ...............................................C-247—C-300

(viii)
 Krishna's

COMPLEX ANALYSIS AND

CALCULUS OF VARIATIONS
C hapters

1. Analytic Functions

2. Complex Integration

Zeros, Singularities and

3. Rouche's Theorem

4. The Calculus of Residues

5. Fourier Transforms

6. Finite Fourier Transforms

7. Fourier Series

8. Calculus of Variations
 1 Curves in the Argand Plane
e know that the equations of the type x = x (t), y = y (t), where t is the
W parameter, give the parametric representation of a curve in the plane. Using the
complex variable z, these equations can be written as a single equation
z = z (t) = x (t) + i y (t) where z = x + iy.
Definitions: (a) In the Argand plane, a continuous complex valued function
z (t) = x (t) + iy (t), where x (t) and y (t) are real valued continuous functions of a real variable t,
defined in the range α ≤ t ≤ β where α < β is called a continuous arc or a curve. We call z (α) and
z (β) the end points of the curve, z (α) is the initial point and z ( β) the terminal point of
the curve. If z (α) = z (β) i.e., if the initial and terminal points of a curve coincide, the
curve is said to be a closed curve.
A point z1 is a multiple point of the curve if the equation z1 = x (t) + iy (t) is satisfied by
more than one value of t in the given range. In particular the multiple point is called a
double point if the above equation is satisfied by two values of t in the given range.
(b) A curve Γ given by, z (t) = x (t) + iy (t), α ≤ t ≤ β is called a Jordan arc or a simple
 C-4

curve if t1 ≠ t2 implies z (t1) ≠ z (t2 ) i.e., z (t) is one-one. A Jordan arc is a curve without
multiple points.
(c) A closed curve Γ given by z (t) = x (t) + iy (t), α ≤ t ≤ β, is called simple if t1 < t2 and
z (t1) = z (t2 ) imply t1 = α and t2 = β.
We usually refer to such curves as simple closed Jordan curves.
Example: The circle z = cos t + i sin t, 0 ≤ t ≤ 2π is a simple closed Jordan curve since
the values of z (t) coincide only at the end points t = 0 and t = 2π.
The Jordan curve theorem: The theorem states that a simple closed Jordan curve divides
the Argand plane into two open domains which have the curve as common boundary.
Of these two domains one is bounded and it is called the interior domain ; the other is
unbounded and is called the exterior domain. For example, the circle| z | = r divides
the Argand plane into two open domains given by| z | < r and| z | > r. Out of these| z | < r
is bounded and is the interior of the circle ; the other | z | > r is unbounded and is the
exterior of the circle,| z | = r.The circle is the common boundary of the two domains.

2 Functions of a Complex Variable

If certain rules be given by means of which it is possible to find one or more complex numbers w for
every value of z in a certain domain D, w is said to be a function of z defined on the domain D and we
write
w = f (z ).
Since z = x + iy, f (z ) will be of the form u + iv where u and v are functions of two real
variables x and y. Thus we can write w = u ( x, y) + i v ( x, y), x, y are real.
Single valued and multiple valued functions: If w takes only one value for each
value of z in the region D, w is said to be a uniform or single valued function of z. If there
correspond two or more values of w for some or all values of z in the region D, w is called
multiple valued function of z.

3 Neighbourhood of a Point
A neighbourhood of a point z0 in the Argand plane is the set of all points z such that
| z − z0 | < δ, where δ is an arbitrary small positive number. The number δ is called the
radius of this neighbourhood.
Deleted neighbourhood: If from a neighbourhood of a point z0 , the point z0 itself is
deleted or excluded, we get a deleted neighbourhood of z0 .

4 Limit and Continuity

Let f (z ) be any function of the complex variable z defined in a bounded and closed domain D. Then
l is said to be the limit of f (z ) as z approaches a along any path in D if for any arbitrarily chosen
 C-5

positive number ε, however small but not zero, there exists a corresponding number δ greater than
zero such that
| f (z ) − l | < ε,
for all values of z for which 0 < | z − a | < δ.
lim
In symbols, we write f (z ) = l.
z→a

Continuity: A function f (z ) of a complex variable z defined in the closed and bounded domain
D is said to be continuous at a ∈ D if and only if for any arbitrarily chosen positive number ε,
however small but not zero, there exists a corresponding number δ > 0 such that
| f (z ) − f (a)| < ε whenever | z − a | < δ.
It follows from the definitions of limit and continuity that f (z ) is continuous at
lim
z = a iff f (z ) = f (a).
z→a

We say that f (z ) is a continuous function in a domain D if it is continuous at every

z ∈ D.
We can easily show that f (z ) = u ( x, y) + i v ( x, y) is a continuous function of z iff u and
v are continuous functions of x and y.
Uniform continuity: A function f (z ) defined in a domain D is said to be uniformly
continuous in D if given ε > 0, there exists a δ > 0 such that | f (z1) − f (z2 )| < ε whenever
| z1 − z2 | < δ, where z1 and z2 are points in D.
It should be noted carefully that uniform continuity is a property associated with a
domain and not with a single point of it.
If f (z ) is uniformly continuous in a domain D, it is continuous in D.
A function which is continuous in a closed and bounded domain D is uniformly
continuous in D, whereas a function continuous in an open domain D ′ may fail to be
uniformly continuous in the domain D ′⋅

5 Differentiability
Since the mode of definitions of continuity is the same both in case of the functions of
the real and complex variables therefore definition of differentiability of a complex
function is identical with that of the real function.
Let w = f (z ) be a function of a complex variable z defined in a domain D. Then f (z ) is said to be
differentiable at a point z0 of D iff
lim f (z0 + ∆ z ) − f (z0 ) lim f (z ) − f (z0 )
or exists uniquely and finitely
∆z→0 ∆z z → z0 z − z0
and this limit, if it exists finitely, is called the differential coefficient or derivative of f
with respect to z at z = z0 .
It is denoted by f ′ (z0 ) or by Df (z0 ).
 C-6

If the value of the above limit as z → z0 is not unique i.e., if the limit depends upon
amp ∆ z, we say that the derivative of f (z ) at z = z0 does not exist or the function f (z )
is non-differentiable at z = z0 .
Hence if we have to show that f (z ) is non-differentiable, we should try different paths
for ∆ z. Convenient paths for ∆ z are along real and imaginary axes i.e., we can take ∆ z
either wholly real or wholly imaginary.
Note: Since the derivative of a complex function has been defined in the same manner
as the derivative of a function of a single real variable, therefore all the rules of
differential calculus remain the same when applied to complex functions.
(Meerut 2002)

Theorem 1: Continuity is a necessary but not a sufficient condition for the existence of a finite
derivative.
Proof: Let f (z ) be differentiable at z0 . Then,
lim f (z ) − f (z0 )
z → z0 z − z0
exists and equals f ′ (z0 ).
Now we can write
f (z ) − f (z0 )
f (z ) − f (z0 ) = (z − z0 ) , if z ≠ z0 .
z − z0
Taking limit of both sides as z → z0 , we get
lim lim  { f (z ) − f (z0 )} 
[ f (z ) − f (z0 )] = (z − z0 )
z → z0 z → z0  z − z0


lim lim f (z ) − f (z0 )
= (z − z0 )
z → z0 z → z0 z − z0

= f ′ (z0 ). 0 = 0 ,
lim
so that f (z ) = f (z0 ).
z → z0
Hence f (z ) is continuous at z0 . Thus continuity is a necessary condition for
differentiability but it is not a sufficient condition for the existence of a finite
derivative.
The following example illustrates this fact :

Example 1: Prove that the function | z |2 is continuous everywhere but nowhere differentiable
except at origin. (Gorakhpur 2015)
2
Solution: Let f (z ) = | z | where z = x + iy.
 C-7

This function is continuous at every point because x2 + y2 is continuous at all points.

lim f (a + ∆ z ) − f (a) lim | a + ∆ z |2 − | z |2
Now f ′ (a) = ∆ z→ 0 ∆ z→ 0
∆z ∆z
lim (a + ∆ z ) (a + ∆ z ) − a a
= ∆ z→ 0
∆z
lim  a∆ z + a ∆ z + ∆ z∆ z 
= 
∆ z→ 0 
 ∆z 
lim  ∆ z 
= ∆ z→ 0  a +a+∆z 
 ∆z 
lim  ∆z 
= ∆ z→ 0 a +a [∵ ∆ z → 0 as ∆ z → 0 ]
 ∆z 
∵ at a = 0 , we have a = 0 , therefore f ′ (a) = 0 .
Again at a ≠ 0, let ∆ z = r (cos φ + i sin φ).
Then ∆ z = r (cos φ − i sin φ).
∆ z cos φ − i sin φ
∴ = = cos 2 φ − i sin 2 φ,
∆ z cos φ + i sin φ

which does not tend to a unique limit as ∆z → 0 since this limit depends upon arg
∆ z.
Thus f (z ) is not differentiable for any non-zero value of z, though it is continuous
everywhere.
Theorem 2. Rules of differentiation: If f (z ) and g (z ) are analytic functions in a domain
D, then their sum, product and quotient {provided g (z ) ≠ 0} are also analytic and we have
d d d
(i) [ f (z ) ± g (z )] = f (z ) ± g (z )
dz dz dz
d d
(ii) [cf (z )] = c f (z )
dz dz
d d d
(iii) [ f (z ) g (z )] = f (z ) g (z ) + g (z ) f (z )
dz dz dz
 d f (z ) g (z ) −  d g (z ) f (z )
d  dz   dz 
(iv) [ f (z ) / g (z )] = , [ g (z ) ≠ 0 ]
dz [ g (z )]2
d
(v) If f (z ) = F [ g (z )], then f (z ) = F ′ [ g (z )] g ′ (z ). [Chain rule]
dz

6 Analytic, Holomorphic and Regular Functions

Let f (z ) be a single valued function defined in a domain D. Then f (z ) is said to be analytic at a
point z0 of D if it is differentiable not only at z0 but also in some neighbourhood of z0 .
(Gorakhpur 2002, 03, 04, 07, 10, 16;
Rohilkhand 10; Bundelkhand 11; Purvanchal 10)
 C-8

A single valued function which is differentiable at each point of a domain D is said to be

analytic in the domain D.
A function, which is analytic, is also called a Holomorphic function.
If a function f (z ) is analytic at some point in every neighbourhood of a point z0 except
at z0 itself, then z0 is called an isolated singularity of f (z ).
A function f (z ) defined in a domain D is said to have removable singularity at a point
z0 of D if f (z ) is not analytic at z0 but can be made analytic by simply assigning a
suitable value to the function f (z ) at the point z0 .
A function f (z ) is said to be regular at a point z0 if it has a removable singularity at z0
and if f (z ) is analytic in some deleted neighbourhood of z0 . Some authors use the term
regular as a synonym for analytic.

7 Properties of Analytic Functions

If f (z ) and g (z ) are two analytic functions in a domain D, then
(i) f (z ) ± g (z )
(ii) f (z ) . g (z )
f (z )
(iii) provided g (z ) ≠ 0 at any point of D
g (z )

and (iv) k f (z ) where k is any constant are also analytic in D.

8 The Necessary and Sufficient Conditions for f ( z ) to be

Analytic. Cauchy Riemann Equations. (Cartesian Form)
(Rohilkhand 2008, 11; Purvanchal 10)
(a) The necessary condition for f ( z) to be analytic:
Theorem 1: If a function f (z ) = u ( x, y) + iv ( x, y) is differentiable at any point z = x + iy,
the partial derivatives u x , v x , u y, v y should exist and satisfy the equations u x = v y, u y = − v x .
(Meerut 2001; Garhwal 10; Gorakhpur 04, 10, 13)

Proof: Let w = f (z ) = u ( x, y) + iv ( x, y)
We have z = x + iy, then ∆ z = ∆ x + i∆y. …(1)
Since the function is differentiable at any point z, therefore the limit given by
lim ∆ f lim f (z + ∆ z ) − f (z )
∆ z→ 0 =
∆z ∆z→0 ∆z
must exist uniquely as ∆ z → 0 along any path we choose.
Using relations (1) the above limit can be written as
lim  u ( x + ∆ x, y + ∆ y) − u ( x, y) v ( x + ∆ x, y + ∆ y) − v ( x, y) 

∆ z→ 0 +i ⋅ …(2)
 ∆ x + i∆ y ∆ x + i∆ y 
 C-9

Taking ∆ z to be wholly real, we get ∆y = 0.In this case the limit given by (2) becomes
lim  u ( x + ∆ x, y) − u ( x, y) v ( x + ∆ x, y) − v ( x, y) 

∆ x→ 0 +i 
 ∆ x ∆ x 
∂u ∂v
= + i , since f (z ) is differentiable therefore the partial
∂x ∂x
∂u ∂v
derivatives , must also exist
∂x ∂x
= u x + iv x . …(3)
Again taking ∆ z to be wholly imaginary, we get ∆x = 0. In this case the limit given in (2)
becomes
lim  u ( x, y + ∆y) − u ( x, y) v ( x, y + ∆y) − v ( x, y) 
∆ y→ 0  +i 
 i∆y i∆y 
1 ∂u ∂v ∂ u ∂v
= + , since f (z ) is differentiable, therefore , also exist
i ∂y ∂y ∂y ∂y

= − iu y + v y = v y − i u y . …(4)
∆ f
Since the limit given by ∆lim
z→ 0 is unique, therefore equating real and imaginary
∆z
parts of (3) and (4), we get
ux = v y, vx = − u y.
These two equations are known as Cauchy-Riemann partial differential equations.
(b) Sufficient condition for f ( z) to be analytic:
Theorem 2: The single valued continuous function f (z ) is analytic in a domain D if the four
partial derivatives u x , v x , u y, v y exist, are continuous and satisfy Cauchy-Riemann equations at
each point of D.
Proof: Let w = f (z ) = u ( x, y) + iv ( x, y).
We have u = u ( x, y), so that u + ∆u = u ( x + ∆ x, y + ∆y)
∴ ∆u = u ( x + ∆ x, y + ∆y) − u ( x, y)
= u ( x + ∆ x, y + ∆y) − u ( x + ∆ x, y) + u ( x + ∆ x, y) − u ( x, y)
= ∆y u y ( x + ∆ x, y + θ1 ∆y) + ∆ x u x ( x + θ2 ∆ x, y), …(1)
where 0 < θ1 < 1, 0 < θ2 < 1, by the mean value theorem.
Since u x and u y are continuous in the given domain D, therefore by the definition of
uniform continuity, we have
| u y ( x + ∆ x, y + θ1∆y) − u y ( x, y)| < ε
and | u x ( x + θ2 ∆x, y) − u x ( x, y)| < ε, …(2)
provided | ∆ x | < δ and | ∆y | < δ.
Let u y ( x + ∆ x, y + θ1∆y) − u y ( x, y) = α1
and u x ( x + θ2 ∆ x, y) − u x ( x, y) = β1.
 C-10

Then from (2), we have | α1 | < ε1,| β1 | < ε1.

Putting these values in (1), we get
∆u = { α1 + u y ( x, y)} ∆y + {β1 + u x ( x, y)} ∆ x.
Proceeding in the same way, we shall get
∆v = {α2 + v y ( x, y)} ∆y + {β2 + v x ( x, y)} ∆ x,
where | α2 | < ε2 , | β2 | < ε2 .
∆ f ∆ u + i∆ v
Now =
∆z ∆ x + i∆ y
(u y ∆y + u x ∆ x + α1 ∆y + β1 ∆ x) + i (v y ∆y
+ v x ∆ x + α2 ∆y + β2 ∆ x)
=
∆ x + i∆y
− v x ∆y + u x ∆ x + iu x ∆y + α1 ∆y + β1 ∆ x + iα2 ∆y + iβ2 ∆ x
=
∆ x + i∆y
[∵ u x = v y , v x = − u y]
(u x + iv x )(∆ x + i∆y) + α1 ∆y + β1 ∆ x + iα2 ∆y + iβ2 ∆ x
=
∆ x + i∆y
(α1 + iα2 ) ∆y (β1 + i β2 ) ∆ x
= u x + iv x + +
∆ x + i∆y ∆ x + i∆y

 ∆f   (α + iα2 ) ∆y (β1 + iβ2 ) ∆ x 

or  − (u x + iv x ) = 1 + 
∆z   ∆ x + i∆y ∆ x + i∆y 
| α1 + iα2 || ∆y | | β1 + iβ2 || ∆ x |
≤ +
| ∆ x + i∆y | | ∆ x + i∆y |

≤ | α1 | + | α2 | + | β1 | + | β2 |
{∵ | ∆ x | ≤ | ∆ x + i∆y | and | ∆y | ≤ | ∆ x + i∆y |}

∴  ∆f − (u + iv )≤ 2 ε + 2 ε .
x x 1 2
 ∆z 
lim ∆f
Hence, u x + iv x = = f ′ (z ).
∆z → 0 ∆z

Note: (i) | f ′ (z )|2 = | u x |2 + | v x |2

= | u x |2 + | v x |2 = u x v y − v x u y, using Cauchy-Riemann equations

∂(u, v)
=
∂( x, y)
= Jacobian of u and v with respect to x and y.
(ii) Lagrange’s mean value theorem:
If f ( x) is defined and continuous in a ≤ x ≤ a + h, differentiable in a < x < a + h, then
there exists a point c = a + θh, (0 < θ < 1) in ] a, a + h [ such that
f (a + h) − f (a) = h f ′ (a + θh).
 C-11

(iii) From theorem 2, we have

dw ∂u ∂v
f ′ (z ) = u x + iv x ⇒ = +i
dz ∂x ∂x
dw ∂ ∂w
⇒ = (u + iv) =
dz ∂x ∂x
dw ∂w
Hence = ⋅
dz ∂x
An Important Observation:
1 1
Since x = (z + z ), y = (z − z ), u and v can be regarded as functions of two
2 2i
independent variables z and z. If u and v have first order continuous derivatives, the
condition that w shall be independent of z is
∂w ∂
=0 or (u + iv) = 0
∂z ∂z
 ∂u ∂x ∂u ∂y   ∂v ∂x ∂v ∂y 
or  ⋅ + ⋅  + i ⋅ + ⋅  =0
 ∂x ∂z ∂y ∂z   ∂x ∂z ∂y ∂z 

1 ∂u 1 ∂u i ∂v 1 ∂v
or ⋅ − + − =0
2 ∂x 2 i ∂y 2 ∂x 2 ∂y
∂u ∂u ∂v ∂v
or +i +i − = 0.
∂x ∂y ∂x ∂y
Whence equating real and imaginary parts to zero, we get
∂u ∂v ∂v ∂u
= and =−
∂x ∂y ∂x ∂y

which are the Cauchy-Riemann equations.

It follows that if f (z ) is an analytic function of z, then x and y can occur in f (z ) only in
the combination of x + iy.

Example 2: Show that the function f (z ) = sin x cosh y + i cos x sinh y

is continuous as well as analytic everywhere.
Solution: Let u ( x, y) = sin x cosh y, v ( x, y) = cos x sinh y. Here u and v are both
rational functions of x and y having non-zero denominators for all values of x and y,
therefore u and v are both continuous everywhere.
Hence f (z ) is continuous everywhere.
∂u ∂u
We have = cos x cosh y, = sin x sinh y
∂x ∂y
∂v ∂v
and = − sin x sinh y, = cos x cosh y.
∂x ∂y
 C-12

∂u ∂v ∂v ∂u
These relations show = , =− ⋅
∂x ∂y ∂x ∂y

∴ u and v satisfy Cauchy-Riemann equations.

Thus f (z ) is analytic everywhere.

Example 3: Show that the function f (z ) = √| xy | is not analytic at origin although the
Cauchy-Riemann equations are satisfied at that point.
(Meerut 2012; Kanpur 03; Rohilkhand 12; Purvanchal 10, 12; Agra 12)
Solution: Let f (z ) = u ( x, y) + iv ( x, y).
Then u ( x, y) = √| xy |, v ( x, y) = 0 .
At the origin, we have
∂u lim u ( x, 0 ) − u (0 , 0 ) lim 0 − 0
= = =0
∂x x → 0 x x →0 x
∂u lim u (0 , y) − u (0 , 0 ) lim 0 − 0
= = = 0.
∂y y→0 y y→0 y
∂v ∂v
Similarly = 0, = 0.
∂x ∂y

Hence Cauchy-Riemann equations are satisfied at origin.

lim f (z ) − f (0 ) lim √| xy | − 0
Now f ′ (0 ) = = ⋅
z→0 z z→0 x + iy

Suppose z → 0 along y = mx, then we have

2
lim √| mx | lim √| m |
f ′ (0 ) = = ,
z → 0 x + i mx z → 0 (1 + im)

which is not unique, since it depends on m.

∴ f ′ (0 ) does not exist.
d
Example 4: Show that (z ) does not exist anywhere.
dz (Kanpur 2000)
Solution: We have z = x + iy. Then z = x − iy.

d lim (z + ∆z ) − z
Now z = ∆ z→ 0
dz ∆z
lim ( x + iy + ∆x + i ∆y ) − ( x + iy)
= ∆ z→ 0
∆x + i ∆y
lim ( x + ∆x) − i ( y + ∆y) − ( x − iy)
= ∆ x→ 0
∆ y→ 0 ∆x + i ∆y

lim ∆x − i ∆y
= ∆ x→ 0 ...(1)
∆ y→ 0 ∆x + i ∆y
 C-13

Taking ∆z → 0 along real axis, we get ∆y = 0. In this case the limit given by (1) becomes
1.
Again taking ∆z → 0 along imaginary axis, we get ∆x = 0. In this case the limit given by
d
(1) becomes − 1. Since the value of the limit given by (1) is not unique so (z ) does
dz
not exist. Hence f (z ) = z is not analytic anywhere.
Example 5: Find whether the following functions are analytic.
z
(i) f (z ) = z (ii) f (z ) = e

(iii) f (z ) = cos x sin y + i sin x cos y.

Solution: (i) We have f (z ) = u + iv = z = x + iy = x − iy.
∴ u = x, v = − y.
∂u ∂v ∂u ∂v
∴ = 1, = 0, = 0, = − 1.
∂x ∂x ∂y ∂y
∂u ∂v
We see that ≠ ⋅
∂x ∂y
Hence one of the Cauchy-Riemann equations is not satisfied.
∴ f (z ) = z is not analytic.
(ii) We have f (z ) = u + iv = e z = e x + iy = e x . e iy

= e x (cos y + i sin y) = e x cos y + i e x sin y.

∴ u = e x cos y, v = e2 x sin y.
∂u ∂v ∂u ∂v
∴ = e x cos y, = e x sin y, = − e x sin y, = e x cos y.
∂x ∂x ∂y ∂y
∂u ∂v ∂v ∂u
We see that = and =− ⋅
∂x ∂y ∂x ∂y
Hence Cauchy-Riemann equations are satisfied.
∂u ∂v ∂u ∂v
Also , , and exist and are continuous functions.
∂x ∂x ∂y ∂y
z
Hence f (z ) = e is an analytic function.
(iii) We have f (z ) = u + i v = cos x sin y + i sin x cos y.
∴ u = cos x sin y, v = sin x cos y.
∂u ∂v
Now = − sin x sin y, = cos x cos y,
∂x ∂x
∂u ∂v
= cos x cos y, = − sin x sin y.
∂y ∂y
∂v ∂u
We see that ≠− ⋅
∂x ∂y

∴ One of the Cauchy-Riemann equations is not satisfied.

Hence f (z ) = cos x sin y + i sin x cos y is not analytic.
 C-14

9 Polar Form of Cauchy-Riemann Equations

(Purvanchal 2007, 09; Gorakhpur 07, 09, 11)
We have x = r cos θ, y = r sin θ.
∴ x2 + y2 = r2 …(1)
and θ = tan−1 ( y / x). …(2)
Differentiating (1) and (2) partially w.r.t. x and y, we get
∂r x ∂r y
= = cos θ, = = sin θ.
∂x r ∂y r
∂θ 1  y y sin θ
= − 2  = − 2 =−
∂x 1 + ( y / x)  x 
2
x + y 2 r
∂θ 1  1 = x cos θ
and =   = ⋅
∂y 1 + ( y / x)2  x  x2 + y2 r
∂u ∂u ∂r ∂u ∂θ ∂u ∂u sin θ
Now = + = cos θ − ,
∂x ∂r ∂x ∂θ ∂x ∂r ∂θ r
∂u ∂u ∂r ∂u ∂θ ∂u ∂u cos θ
= + = sin θ + ,
∂y ∂r ∂y ∂θ ∂y ∂r ∂θ r
∂v ∂v ∂r ∂v ∂θ ∂v ∂v sin θ
= + = cos θ −
∂x ∂r ∂x ∂θ ∂x ∂r ∂θ r
∂v ∂v ∂r ∂v ∂θ ∂v ∂v cos θ
and = + = sin θ + ⋅
∂y ∂r ∂y ∂θ ∂y ∂r ∂θ r
Cauchy-Riemann equations in cartesian form are
∂u ∂v ∂u ∂v
= and =− ⋅
∂x ∂y ∂y ∂x
Using the above relations, we get
∂u ∂u sin θ ∂v ∂v cos θ
cos θ − = sin θ + …(3)
∂r ∂θ r ∂r ∂θ r
∂u ∂u cos θ ∂v ∂v sin θ
and sin θ + =− cos θ + ⋅ …(4)
∂r ∂θ r ∂r ∂θ r
Multiplying (3) by cos θ and (4) by sin θ and adding, we get
∂u 1 ∂v
= ⋅ …(5)
∂r r ∂θ
Again multiplying (3) by sin θ and (4) by cos θ and subtracting, we get
1 ∂u ∂v
=− ⋅ …(6)
r ∂θ ∂r
Equations (5) and (6) are the Cauchy-Riemann equations in polar form.

10 Derivative of w = f ( z ) in Polar Form

dw ∂w ∂w ∂r ∂w ∂θ
We have = = +
dz ∂x ∂r ∂x ∂θ ∂x
 C-15

 ∂u ∂v  ∂r  ∂u ∂v  ∂θ
= +i  + +i 
 ∂r ∂r  ∂x  ∂θ ∂θ  ∂x [ ∵ w = u + iv ]
 ∂u ∂v   ∂u ∂v   sin θ 
= + i  cos θ +  + i  − ,
 ∂r ∂r   ∂θ ∂θ   r  [From article 9]
 ∂u ∂v   ∂v ∂u  sin θ
= +i  cos θ −  − r + ir  ,
 ∂r ∂r   ∂r ∂r  r

[ From (5) and (6) of article 9]

 ∂u ∂v   ∂u ∂v 
= +i  cos θ − i +i  sin θ
 ∂r ∂r   ∂r ∂r 

 ∂u ∂v 
= +i  (cos θ − i sin θ)
 ∂r ∂r 
∂w − iθ
= e .
∂r

Example 6: Show that the function f (z ) = z n, where n is a positive integer is an analytic

function.
Solution: We have f (z ) = z n.

lim f (z + ∆ z ) − f (z )
Now f ′ (z ) = ∆ z→ 0
∆z

lim (z + ∆ z )n − z n
= ∆ z→ 0
∆z
1
z n + nz n −1 ∆z + n (n − 1) z n − 2 (∆ z )2 + ... + (∆z )n − z n
lim 2
= ∆ z→ 0 ,
∆z
by binomial theorem
lim 1
= ∆ z→ 0 [nz n −1 + n (n − 1) z n−2
∆ z + ... + (∆z )n −1]
2
= nz n −1.

∴ f ′ (z ) exists for all finite values of z.

Hence f (z ) is an analytic function.
Note: Applying the above formula for z , z 2 , z 3 ,... and the rules of differentiation stated
in theorem II we see that a polynomial
f (z ) = a0 + a1z + a2 z 2 + … + anz n
is an analytic function of z. More generally, a rational function
 C-16

a0 + a1z + a2 z 2 + … + an z n
f (z ) =
b0 + b1z + b2 z 2 + … + bm z m

is an analytic function of z throughout any finite domain in the complex plane where
the denominator does not vanish.
x3 y( y − ix) f (z ) − f (0 )
Example 7: If f (z )= 6 2
, z ≠ 0 and f (0 ) = 0 , show that → 0 as
x +y z
z → 0 along any radius vector but not as z → 0 in any manner.
(Purvanchal 2008; Bundelkhand 11; Gorakhpur 11)
f (z ) − f (0 ) f (z ) − 0 f (z )
Solution: We have = =
z z z
3 3
x y ( y − ix) − i x y ( x + iy) x3 y
= = =−i ⋅
( x6 + y2 )( x + iy) ( x6 + y2 )( x + iy) x6 + y 2

Let z → 0 along the radius vector y = mx. Then we have

limf (z ) − f (0 ) lim − i x3 . mx lim − i mx2
= = = 0.
z→0 z x → 0 x + (mx)
6 2 x → 0 x4 + m2
Now let z → 0 along y = x3 . Then
limf (z ) − f (0 ) lim − i x3 . x3 i
= = − ≠ 0.
z→0 z x → 0 x6 + ( x 3)2 2

Hence the result.

Example 8: Show that the function f (z ) = u + iv,
x3 (1 + i) − y3 (1 − i)
where f (z ) = , z ≠ 0 and f (0 ) = 0
x2 + y2
is continuous and that Cauchy-Riemann equations are satisfied at the origin, yet f ′ (0 ) does not
exist. (Meerut 2001; Rohilkhand 08; Garhwal 10; Gorakhpur 07, 11, 14)
x3 − y3 x3 + y3
Solution: We have u = , v= , z ≠ 0.
x2 + y2 x2 + y2
Here u and v are rational functions of x and y with non-zero denominators. Therefore u
and v are continuous everywhere when z ≠ 0.
To test the continuity at z = 0, changing u and v to polars by putting
x = r cos θ, y = r sin θ, we get u = r (cos3 θ − sin3 θ), v = r (cos3 θ + sin3 θ), each of
which tends to 0 as r → 0 whatever may be the value of θ. Also we have f (0 ) = 0 .
Since actual and limiting values of u and v are same at origin therefore f (z ) is
continuous at origin. Hence f (z ) is continuous for all values of z.

Now at the origin, we have

∂u lim u ( x, 0 ) − u (0 , 0 ) lim x − 0
= = = 1,
∂x x → 0 x x →0 x
 C-17

∂u lim u (0 , y) − u (0 , 0 ) lim − y − 0
= = = − 1,
∂y y→0 y y→0 y
∂v lim x − 0 ∂v lim y −0
= = 1 and = = 1.
∂x x → 0 x ∂y y→0 y
∂u ∂v ∂u ∂v
Since = and =− , therefore u and v satisfy Cauchy-Riemann equations at
∂x ∂y ∂y ∂x
origin.
lim f (z ) − f (0 )
We have f ′ (0 ) =
z→0 z
lim ( x3 − y3 ) + i ( x3 + y3 ) 1
= ⋅ ⋅
z→0 2
(x + y ) 2 ( x + iy)

Taking z → 0 along y = x, we get

lim i2 x3 1 i 1
f ′ (0 ) = ⋅ = = (1 + i).
x → 0 2 x 2 ( x + ix) 1 + i 2

Again taking z → 0 along x-axis, we get

x3 + ix3 1
lim
f ′ (0 ) = ⋅ = 1 + i.
x→0 x2 x

Since f ′ (0 ) has different values along different curves therefore f ′ (0 ) is not unique. So
f ′ (0 ) does not exist.
−4
Example 9: Show that the function f (z ) = e − z , z ≠ 0 and f (0 ) = 0
is not analytic at z = 0 although the Cauchy-Riemann equations are satisfied at that point.
(Meerut 2003; Rohilkhand 10; Purvanchal 07, 11; Gorakhpur 12, 14)
−4 4
Solution: We have f (z ) = e − z = e −1 /( x + iy)
4
/( x2 + y2 )4
= e −( x − iy)
4
+ y4 − 6 x2 y2 − 4 ix3 y + 4 i xy3 ) /( x2 + y2 )4
= e −( x
4
+ y4 − 6 x2 y2 ) / r8 2
− y2 ) / r8
= e −( x e4 i xy ( x , where x2 + y2 = r2
4
+ y4 − 6 x2 y2 ) / r8  4 xy ( x2 − y2 ) 4 xy ( x2 − y2 ) 
= e −( x  cos 8
+ i sin .
 r r8 
We have at the origin
−4
∂u lim u ( x, 0 ) − u (0 , 0 ) lim e − x − 0
= =
∂x x → 0 x x→0 x
lim 1 lim 1
= =
x→0 1 / x4 x→0  1
1 
xe x  1 + 4 + 8 + ... 
 x 2x 
 C-18

lim 1
= = 0.
x→0 1 1
x+ + + ...
x3 2 x7
−4
∂u lim u (0 , y) − u (0 , 0 ) lim e − y
Similarly = = = 0,
∂y y→0 y y→0 y
∂v ∂v
= 0 and = 0.
∂x ∂y

∴ Cauchy-Riemann equations are satisfied at origin.

−4
lim f (z ) − f (0 ) lim e − z
Now f ′ (0 ) = =
z→0 z z→0 z
−4
e − iπ
lim e− r
= , taking z → 0 along z = re iπ /4
r→0 re iπ /4
−4 4
lim er lim e1 / r
= = → ∞.
r → 0 r e i π /4 r → 0 r e i π /4

∴ f (z ) is not analytic at z = 0.
Example 10: Find the analytic function whose real part is sin 2 x / (cosh 2 y − cos 2 x).
Solution: Let f (z ) = u + iv be the required analytic function.
sin 2 x
Then u= ⋅
cosh 2 y − cos 2 x
∂u 2 cos 2 x (cosh 2 y − cos 2 x) − sin 2 x (2 sin 2 x)
Now =
∂x (cosh 2 y − cos 2 x)2
2 cos 2 x cosh 2 y − 2
= = φ1 ( x, y)
(cosh 2 y − cos 2 x)2
∂u 2 sin 2 x sinh 2 y
and =− = φ2 ( x, y)
∂y (cosh 2 y − cos 2 x)2

The function f (z ) is given by

 2 cos 2 z − 2 
f (z ) = ∫ [φ1 (z , 0 ) − i φ2 (z , 0 )] dz + c = ∫ 
(1 − cos 2 z )
2
− i 0  dz + c

2
= ∫ −
1 − cos 2 z
dz + c = ∫ − cosec2 z dz + c

= cot z + c .

11 Orthogonal System
Two families of curves u ( x, y) = c1 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.
 C-19

Differentiating u ( x, y) = c1, we get

∂u ∂u dy ∂u
/ ∂∂uy = m1, say
dy
+ ⋅ =0 or =−
∂x ∂y dx dx ∂x
Similarly from v ( x, y) = c2 , we get
dy ∂u ∂v
dx
=−
∂x ∂y
/ = m2 , say

Now the two families of curves will intersect orthogonally if

∂u ∂v ∂u ∂v
m1m2 = − 1 or ⋅ + ⋅ = 0.
∂x ∂x ∂y ∂y
Theorem: If f (z ) = u + iv be an analytic function of z = x + iy, prove that the families of
curves u = c1, v = c2 are orthogonal to each other, where c1 and c2 are parameters.
(Rohilkhand 2008; Purvanchal 10)
Proof: It is given that f (z ) = u + iv is an analytic function therefore we have
u x = v y, v x = − u y. …(1)
The given systems are
u ( x, y) = c1, …(2)
v ( x, y) = c2 . …(3)
Let m1 and m2 be the slopes of curves (2) and (3) respectively.
dy ∂u / ∂x
From (2) =− = m1.
dx ∂u / ∂y
dy ∂v / ∂x
From (3) =− = m2 .
dx ∂v / ∂y
 ∂u / ∂x   ∂v / ∂x 
Now m1 m2 =  −  × −  = − 1, from (1).
 ∂u / ∂y   ∂v / ∂y 
Hence the systems u = c1 and v = c2 are orthogonal to each other.

Example 11: If f (z ) = u + iv is an analytic function, regular in D, where f (z ) ≠ 0 , prove that

the curves u = const.,v = const., form two orthogonal families. Verify this in case of f (z ) = sin z .
Solution: (i) Suppose f (z ) = u + iv is an analytic function of z.
To prove that u = const., v = const. from two orthogonal families.
Prove this as in Theorem.
(ii) To verify the result (i) by taking f (z ) = sin z .
u + iv = f (z ) = sin z = sin ( x + iy)
= sin x . cos iy + cos x . sin iy
= sin x . cosh y + i cos x . sinh y
 C-20

∴ u = sin x . cosh y = c1, say

v = cos x . sinh y = c2 , say.
Differentiating both w.r.t. x, we get
 dy 
cos x . cosh y + sin x . sinh y   = 0
 dx 1

 dy 
and − sin x . sinh y + cos x . cosh y   = 0
 dx  2

 dy  cos x . cosh y  dy  sin x . sinh y

or   = and   = ⋅
 dx 1 − sin x . sinh y  dx  2 cos x cosh y

Multiplying these two, we get

 dy   dy 
    = − 1.
 dx 1  dx  2

Hence the verification follows.

12 Harmonic Function
Theorem 1: Real and imaginary parts of an analytic function satisfy Laplace’s equation.
(Rohilkhand 2010)
Proof: Let f (z ) = u + iv be an analytic function. Then it satisfies Cauchy-Riemann
equations.
∂u ∂v ∂u ∂v
∴ = and =− …(1)
∂x ∂y ∂y ∂x
Since u and v are the real and imaginary parts of an analytic function therefore partial
derivatives of u and v of all orders exist and are continuous functions of x and y.
From (1), we have
∂2 u ∂2 v ∂2 u ∂2 v
= and = − ⋅
∂x2 ∂x ∂y ∂ y2 ∂y ∂x

∂2 u ∂2 u
∴ 2
+ = 0.
∂x ∂ y2

∂2 v ∂2 v
Similarly + = 0.
∂x2 ∂ y2

Therefore the functions u and v satisfy the Laplace’s equation

∂2 φ ∂2 φ
+ = 0.
∂x2 ∂ y2

Definition : Any function of x and y possessing continuous partial derivatives of the first and
second orders and satisfying Laplace’s equation is called a harmonic function.
 C-21

Harmonic Conjugate of a function:

Definition: Let u ( x, y) be a harmonic function. Then a function v ( x, y) is said to be a
harmonic conjugate of u ( x, y) if
(i) v ( x, y) is harmonic and
∂u ∂v ∂v ∂u
(ii) = and =− i.e., u and v satisfy Cauchy-Riemann equations.
∂x ∂y ∂x ∂y
Theorem 2: If f (z ) = u + iv is analytic in a domain D, then v is the harmonic conjugate of u.
Conversely, if v is the harmonic conjugate of u in a domain D, then f (z ) = u + iv is analytic in D.
Proof: Since f (z ) = u + iv is analytic in D, Cauchy-Riemann equations are satisfied i.e.,
∂u ∂v ∂u ∂v
= and =− ⋅
∂x ∂y ∂y ∂x
Differentiating partially with respect to x and y respectively and adding, we get
∂2 u ∂2 u ∂2 v ∂2 v
+ = − = 0.
∂x2 ∂ y2 ∂x ∂y ∂y ∂x

Similarly differentiating partially with respect to y and x respectively and adding, we get
∂2 v ∂2 v
2
+ = 0.
∂x ∂ y2

∴ u and v are harmonic functions in D and v is the harmonic conjugate of u because u

and v satisfy Cauchy-Riemann equations.
Conversely, let v be the harmonic conjugate of u.
Then by the definition of the harmonic conjugate of u, v is harmonic and
∂u ∂v ∂u ∂v
Cauchy-Riemann equations = and =− are satisfied. Also by the definition
∂x ∂y ∂y ∂x
of harmonic functions u and v possess continuous partial derivatives of the first and
∂u ∂u ∂v ∂v
second orders so that , , and are all continuous functions.
∂x ∂y ∂x ∂y

Hence, f (z ) = u + iv is analytic in D.
Remark: It is very important to note that if v is a harmonic conjugate of u in some
domain D, then it is always not true that u is also the harmonic conjugate of v in D.
We illustrate this by the following example :
Let u = x2 − y2 and v = 2 xy.
Then f (z ) = u + iv is analytic in D as shown below.
∂u ∂u ∂v ∂v
We have = 2 x, = − 2 y, = 2 y, = 2 x.
∂x ∂y ∂x ∂y
∂u ∂v ∂u ∂v
∴ = and =−
∂x ∂y ∂y ∂x
i.e., Cauchy-Riemann equations are satisfied by u and v.
 C-22

∂u ∂u ∂v ∂v
Also , , and are all continuous functions.
∂x ∂y ∂x ∂y
∴ f (z ) = u + i v is analytic in D.
Hence, both u and v are harmonic functions and they satisfy Cauchy-Riemann
equations
∂u ∂v ∂u ∂v
= and =− ⋅
∂x ∂y ∂y ∂x

∴ v is the harmonic conjugate of u.

But if we define φ (z ) = v + i u, we see that
∂v ∂v ∂u ∂u
= 2 y, = 2 x, = 2 x, = − 2 y.
∂x ∂y ∂x ∂y
∂v ∂u ∂v ∂u
We see that ≠ and ≠− ⋅
∂x ∂y ∂y ∂x
Thus if φ (z ) = v + i u, then v and u do not satisfy Cauchy-Riemann equations.
∴ φ (z ) is not analytic in D.
Hence, u is not the harmonic conjugate of v.
Theorem 3: Two functions u ( x, y) and v ( x, y) are harmonic conjugates of each other if and
only if they are constants.
Proof: Let u ( x, y) = c1 and v ( x, y) = c2 V x, y ∈ D, where c1 and c2 are constants.
∂u ∂2 u ∂v ∂2 v
Then = 0, = 0, = 0, = 0,
∂x ∂x2 ∂x ∂x2
∂u ∂2 u ∂v ∂2 v
= 0, = 0, = 0, = 0.
∂y ∂ y2 ∂y ∂ y2

∂2 u ∂2 u ∂2 v ∂2 v
∴ + =0 and + = 0.
∂x2 ∂ y2 ∂x2 ∂ y2

Thus both u and v are harmonic functions.

∂u ∂v ∂u ∂v
Also = and =−
∂x ∂y ∂y ∂x
i.e., u and v satisfy Cauchy-Riemann equations.
Hence v is the harmonic conjugate of u.
∂v ∂u ∂v ∂u
Again = and =−
∂x ∂y ∂y ∂x
i.e., v and u also satisfy Cauchy-Riemann equations.
Hence u is the harmonic conjugate of v.
Thus if both u and v are constants, they are harmonic conjugates of each other.
Conversely, let u ( x, y) and v ( x, y) be two harmonic functions such that they are
harmonic conjugates of each other.
 C-23

Then u and v satisfy Cauchy-Riemann equations i.e.,

∂u ∂v
= ...(1)
∂x ∂y
∂u ∂v
and =− ...(2)
∂y ∂x

Again v and u also satisfy Cauchy-Riemann equations i.e.,

∂v ∂u
= ...(3)
∂x ∂y
∂v ∂u
and =− ...(4)
∂y ∂x
From (1) and (4), we have
∂u ∂u
=−
∂x ∂x
∂u ∂u
⇒ 2 =0 ⇒ = 0 ⇒ u is independent of x.
∂x ∂x
From (2) and (3), we have
∂u ∂u ∂u ∂u
=− ⇒ 2 =0 ⇒ = 0 ⇒ u is independent of y.
∂y ∂y ∂y ∂y

∴u ( x, y) is independent of both x and y and consequently u ( x, y) is a constant function.

Similarly, we can show that v ( x, y) is also a constant function.
Hence if both u and v are harmonic conjugates of each other, they are constant
functions.
Determination of the conjugate function:
If f (z ) = u + iv is an analytic function, u and v are called conjugate functions. Being given one of
these say, u ( x, y), to determine the other v ( x, y).
∂v ∂v
We have dv = dx + dy , since v is a function of x and y
∂x ∂y
∂u ∂u
or dv = − dx + dy, …(1)
∂y ∂x
by Cauchy-Riemann equations.
The equation (1) is of the form dv = M dx + N dy,
∂u ∂u
where M=− , N = ⋅
∂y ∂x

∂M ∂2 u ∂N ∂2 u
Now = − 2 and = 2 ⋅
∂y ∂y ∂x ∂x

Since f (z ) is an analytic function therefore u is a harmonic function i. e., it satisfies

Laplace’s equation.
 C-24

∂2 u ∂2 u ∂2 u ∂2 u
∴ 2
+ 2
=0 or 2
=−
∂x ∂y ∂x ∂ y2
∂M ∂N
so that = ⋅
∂y ∂x
Thus equation (1) satisfies the condition of exact differential equation. Therefore v can
be determined by integrating (1).

Example 12: Show that the function u = x3 − 3 xy2 is harmonic and find the corresponding
analytic function. (Lucknow 2007, 13B, 14)
3 2
Solution: We have u = x − 3 xy .

∂u ∂2 u ∂u ∂2 u
= 3 x2 − 3 y2 , = 6 x, = − 6 xy and = − 6 x.
∂x ∂x2 ∂y ∂ y2

∂2 u ∂2 u
Now + = 0 , so that u satisfies Laplace’s equation.
∂x2 ∂ y2

Also first and second order partial derivatives of u are continuous functions of x and y.
Consequently u is a harmonic function.
∂u ∂u
Now f ′ (z ) = −i = 3 x2 − 3 y2 − i (− 6 xy)
∂x ∂y

= 3 ( x2 − y2 + 2 ixy) = 3 ( x + iy)2 = 3 z 2 .

Integrating f (z ) = z 3 + c .

Example 13: Show that the functions

1
(i) u = log ( x2 + y2 ) (Rohilkhand 2012; Garhwal 10) (ii) u = cos x cosh y
2
are harmonic, find their harmonic conjugates.
1
Solution: (i) We have u = log ( x2 + y2 ).
2
∂u x ∂2 u y2 − x2
= 2 , = 2
∂x x + y 2
∂x2
( x + y2 )2

∂u y ∂2 u x2 − y2
= 2 and = ⋅
∂ y x + y2 ∂y2 ( x2 + y2 )2
All the first and second order partial derivatives of u are continuous functions of x and y.
∂2 u ∂2 u
Also 2
+ = 0 i. e., u satisfies Laplace’s equation.
∂x ∂ y2

∴ u is a harmonic function.
 C-25

Let v be the harmonic conjugate of u.

∂v ∂v ∂u ∂u
We have dv = dx + dy = − dx + dy, by Cauchy-Riemann equations
∂x ∂y ∂y ∂x
y x
=− 2 2
dx + dy
x + y x + y2
2

x dy − y dx
or dv = ⋅
x2 + y2
y
Integrating, we get v = tan−1 + c , where c is a real constant.
x
(ii) We have u = cos x cosh y.
∂u ∂2 u
= − sin x cosh y, = − cos x cosh y
∂x ∂x2
∂u ∂2 u
= cos x sinh y and = cos x cosh y.
∂y ∂ y2

All the first and second order partial derivatives of u are continuous functions.
∂2 u ∂2 u
Also + = 0 , i. e., u satisfies Laplace’s equation.
∂x2 ∂ y2

∴ u is a harmonic function.
Let v be the harmonic conjugate of u.
∂v ∂v
We have dv = dx + dy
∂x ∂y
∂u ∂u
=− dx + dy, by Cauchy-Riemann equations
∂y ∂x
= − cos x sinh y dx − sin x cosh y dy
= − (cos x sinh y dx + sin x cosh y dy).
Integrating, v = − (sin x sinh y) + c , where c is a real constant.
Example 14: Show that the function u ( x, y) = e x cos y is harmonic. Determine its harmonic
conjugate v ( x, y) and the analytic function f (z ) = u + iv. (Bundelkhand 2011)
x
Solution: Here u = e cos y.
∂u ∂u
= e x cos y, = − e x sin y
∂x ∂y

∂2 u ∂2 u
so = e x cos y and = − e x cos y.
∂x2 ∂ y2

∂2 u ∂2 u
Now 2
+ = 0 , so u satisfies Laplace’s equation.
∂x ∂ y2

Also first and second order partial derivatives of u are continuous therefore u is a
harmonic function.
 C-26

Since v is the harmonic conjugate of u, therefore

∂v ∂v
dv = dx + dy, v is a function of x, y
∂x ∂y
∂u ∂u
=− dx + dy, by Cauchy-Riemann equations
∂y ∂x
= e x sin y dx + e x cos y dy.
Integrating v = e x sin y + c , where c is a real constant.

∴ f (z ) = u + iv = e x cos y + i (e x sin y + c ).
= e x (cos y + i sin y) + i c = e x + iy + d,
where d is a complex constant
= e z + d.

13 Milne-Thomson’s Method
(Method of Constructing a Regular Function)
(Meerut 2002)
We have f (z ) = u ( x, y) + iv ( x, y) and z = x + iy.
1 1
Then x = (z + z ), y = (z − z ).
2 2i
We can write

f (z ) = u  (z + z ), (z − z )  + iv  (z + z ), (z − z )  ⋅
1 1 1 1
…(1)
 2 2i 
 
2 2i 

This relation can be regarded a formal identity in two independent variables z and z.
Putting z = z in (1), we get
f (z ) = u (z , 0 ) + iv (z , 0 ).
∂f
We have f ′ (z ) = = u x + iv x = u x − iu y, by Cauchy-Riemann equations.
∂x
Let u x = φ1 ( x, y), u y = φ 2 ( x, y).
Then f ′ (z ) = φ1 ( x, y) − i φ 2 ( x, y) = φ1 (z , 0 ) − i φ 2 (z , 0 ).
Integrating, we get
f ( z) = ∫ φ1 ( z, 0) dz − i ∫ φ2 ( z, 0) dz + c,
where c is constant of integration.
Similarly if v is given, we have
f (z ) = ∫ ψ1 (z , 0 ) dz + i ∫ ψ2 (z , 0 ) dz + c ′ ,

where v y = ψ1 ( x, y), v x = ψ2 ( x, y).

Theorem: If the real part of an analytic function f (z ) is a given harmonic function u ( x, y),
f (z ) = 2 u (z / 2, z / 2 i) − u (0 , 0 ).
 C-27

Proof: Let f (z ) = f ( x + iy) = u ( x, y) + iv ( x, y).

Then f (z ) = f ( x + iy) = u ( x, y) − iv ( x, y).
Adding, we get
f ( x + iy) + f ( x + iy) = 2 u ( x, y). …(1)
Since f (z ) is a function independent of z, therefore it can be regarded as a function of z.
So we can write
f (z ) = f ( z ).
We can rewrite relation (1) as
1
u ( x, y) = [ f ( x + iy) + f ( x − iy)]. …(2)
2
We can regard (2) as a formal identity, therefore it holds even when x and y are
complex. Putting x = z / 2, y = z / 2 i, we get
1  z z   z − i z 
u (z / 2, z / 2 i) = f  + i  + f  
2   2 2i   2 2 i  
1
= [ f (z ) + f (0 )].
2
∴ f (z ) = 2 u (z / 2, z / 2 i) − f (0 ).
Since f (z ) is only determined upto a purely imaginary constant, we may assume that
f (0 ) is real. So we have
f (0 ) = u (0 , 0 ).
∴ f (z ) = 2 u (z / 2, z / 2 i) − u (0 , 0 ).
Adding a purely imaginary constant, we have
f (z ) = 2 u (z / 2, z / 2 i) − u (0 , 0 ) + ci, where c is real.

sin 2 x
Example 15: If u = ,
cosh 2 y + cos 2 x
find the corresponding analytic function f (z ) = u + iv.
sin 2 x
Solution: Here u = ⋅
cosh 2 y + cos 2 x
∂u 2 cos 2 x (cosh 2 y + cos 2 x) − sin 2 x (− 2 sin 2 x)
Then =
∂x (cosh 2 y + cos 2 x)2
2 + 2 cos 2 x cosh 2 y
= = φ1 ( x, y).
(cosh 2 y + cos 2 x)2
∂u 2 sin 2 x sinh 2 y
=− = φ2 ( x, y).
∂y (cosh 2 y + cos 2 x)2

By Milne-Thomson’s method the function f (z ) is given by

 C-28

f (z ) = ∫ [φ1 (z , 0 ) − i φ2 (z , 0 )] dz + c

 2 + 2 cos 2 z  2 dz
= ∫ 
(1 + cos 2 z )
2
− i0  dz + c =

∫ 1 + cos 2 z
+c

= ∫ sec2 z dz + c

= tan z + c .
Example 16: Find the analytic function whose real part is given and hence find the
imaginary part :
(i) e x sin y (ii) sin x cosh y (iii) x2 − y2 .

Solution: (i) Let f (z ) = u + i v be analytic.

Here, u = e x sin y.
∂u ∂u
∴ = e x sin y, = e x cos y.
∂x ∂y
We apply the Milne-Thomson method to find f (z ).
Since f (z ) = u + i v is analytic, therefore
∂u ∂v
f ′ (z ) = +i
∂x ∂x [See Note (iii) at the end of article 8]
∂u ∂u ∂v ∂u 
= −i [∵ By Cauchy-Riemann equations, =−
∂x ∂y ∂x ∂y 
x x
=e sin y − i e cos y.

Putting x = z and y = 0, we get

f ′ (z ) = e z sin 0 − i e z cos 0 = − i. e z .
Integrating with respect to z , we get
z
f (z ) = − i e + constant
x + iy
= − ie + constant
x iy
= − ie e + constant
x
= − ie (cos y + i sin y) + constant
x x
=e sin y − i e cos y + i c
x
= u + i (c − e cos y)
x
∴ v=c −e cos y.
Hence, the required analytic function is
z x x
f (z ) = u + i v = − i e + constant = e sin y + i (c − e cos y)
x
and the imaginary part v = c − e cos y.
(ii) Let f (z ) = u + i v be analytic.
 C-29

Here, u = sin x cosh y.

∂u ∂u
∴ = cos x cosh y, = sin x sinh y.
∂x ∂y
We apply the Milne-Thomson method to find f (z ).
Since f (z ) = u + i v is analytic, therefore
∂u ∂v
f ′ (z ) = +i
∂x ∂x
∂u ∂u
= −i
∂x ∂y [Using Cauchy-Riemann equations]
= cos x cosh y − i sin x sinh y.
Putting x = z and y = 0, we get
f ′ (z ) = cos z cosh 0 − i sin z sinh 0
= cos z [∵ cosh 0 = 1, sinh 0 = 0 ]
Integrating with respect to z , we get
f (z ) = sin z + constant
= sin ( x + iy) + constant
= sin x cos iy + cos x sin iy + constant
= sin x cosh y + i cos x sinh y + ic
= sin x cosh y + i (cos x sinh y + c )
= u + i v.
Hence, the required analytic function is
f (z ) = u + i v = sin z + constant = sin x cosh y + i (cos x sinh y + c )
and the imaginary part v = cos x sinh y + c .
(iii) Let f (z ) = u + i v be analytic.
Here, u = x2 − y2 .
∂u ∂u
∴ = 2 x, = − 2 y.
∂x ∂y
We apply the Milne-Thomson method to find f (z ).
Since f (z ) = u + i v is analytic, therefore
∂u ∂v
f ′ (z ) = +i
∂x ∂x
∂u ∂u  ∂v ∂u 
= −i ∵ =− , by Cauchy-Riemann equations 
∂x ∂y  ∂x ∂y 
= 2 x + 2 iy.
Putting x = z and y = 0, we get
f ′ (z ) = 2 z .
Integrating with respect to z , we get
 C-30

f (z ) = z 2 + constant

= ( x + iy)2 + constant

= x2 − y2 + 2 i xy + ic

= ( x2 − y2 ) + i (2 xy + c ) = u + i v.
Hence the required analytic function is
f (z ) = z 2 + constant = ( x2 − y2 ) + i (2 xy + c ) and the imaginary part v = 2 xy + c .

Example 17: Prove that the following functions are harmonic and find the harmonic conjugate.
(i) 2 x − x3 + 3 xy2 (ii) e − x ( x cos y + y sin y). (Agra 2012)
3 2
Solution: (i) Let u = 2 x − x + 3 xy .
∂u ∂u
Then = 2 − 3 x2 + 3 y2 , = 6 xy,
∂x ∂y

∂2 u ∂2 u
= − 6 x, = 6 x.
∂x2 ∂ y2

∂2 u ∂2 u
∴ 2
+ = − 6 x + 6 x = 0.
∂x ∂ y2

∴ u is harmonic.
Let v be the harmonic conjugate of u.
Then f (z ) = u + i v is analytic.
∂u ∂v
∴ f ′ (z ) = +i
∂x ∂x
∂u ∂u  ∂v ∂u 
= −i  ∵ By Cauchy-Riemann equations ∂x = − ∂y 
∂x ∂y  
= 2 − 3 x2 + 3 y2 − i 6 xy.

To apply Milne-Thomson’s method, putting x = z and y = 0, we get

f ′ (z ) = 2 − 3 z 2 .

Integrating with respect to z, we get

f (z ) = 2 z − z 3 + a constant

= 2 ( x + iy) − ( x + iy)3 + a constant

= 2 x + i 2 y − x3 + 3 xy2 − 3 x2 y i + iy3 + a constant

= (2 x − x3 + 3 xy2 ) + i (2 y − 3 x2 y + y3 ) + ic

= u + i (2 y − 3 x2 y + y3 + c ).

∴ v = 2 y − 3 x2 y + y3 + c .

Hence the harmonic conjugate of u is v = 2 y − 3 x2 y + y3 + c .

 C-31

(ii) Let u = e − x ( x cos y + y sin y).

∂u
Then = e − x (cos y) + ( x cos y + y sin y) (− e − x )
∂x
= e − x (cos y − x cos y − y sin y),
∂u
= e − x (− x sin y + y cos y + sin y),
∂y

∂2 u
= e − x (− cos y) + (cos y − x cos y − y sin y) (− e − x )
∂x2
= e − x (− cos y − cos y + x cos y + y sin y)

= e − x ( x cos y + y sin y − 2 cos y),

∂2 u
and = e − x (− x cos y + cos y − y sin y + cos y)
∂ y2

= e − x (− x cos y − y sin y + 2 cos y).

∂2 u ∂2 u
∴ 2
+ = e − x ( x cos y + y sin y − 2 cos y − x cos y
∂x ∂ y2
− y sin y + 2 cos y) = 0 .
∴ u is harmonic.
Let v be the harmonic conjugate of u.
Then f (z ) = u + i v is analytic.
∂u ∂v
∴ f ′ (z ) = +i
∂x ∂x
∂u ∂u  ∂v ∂u 
= −i ∵ By Cauchy-Riemann equations, ∂x = − ∂y 
∂x ∂y  
= e − x (cos y − x cos y − y sin y) − ie − x

(− x sin y + y cos y + sin y).

To apply Milne-Thomson’s method putting x = z and y = 0, we get
f ′ (z ) = e − z (1 − z − 0 ) − ie − z (0 ) = e − z (1 − z ).
Integrating with respect to z, we get
f (z ) = (1 − z ) (− e − z ) − ∫ − e − z (− 1) dz + constant

= − (1 − z ) e − z − ∫ e − z dz + constant

= − (1 − z ) e − z − (− e − z ) + constant

= e − z (− 1 + z + 1) + constant

= e − x − iy ( x + iy) + constant

= e − x (cos y − i sin y) ( x + iy) + constant

 C-32

= e − x ( x cos y + y sin y) + ie − x ( y cos y − x sin y) + ic

= u + i [e − x ( y cos y − x sin y) + c ].

∴ v = e − x ( y cos y − x sin y) + c .

Hence the harmonic conjugate of u is

v = e − x ( y cos y − x sin y) + c .

Example 18: Find the orthogonal trajectory of the family of curves

x2 − y2 + x = c .

Solution: Let u = x2 − y2 + x.
∂u ∂u
Then = 2 x + 1, = − 2 y.
∂x ∂y
Let f (z ) = u + i v be analytic.
∂u ∂v
Then f ′ (z ) = +i
∂x ∂x
∂u ∂u  ∂v ∂u 
= −i ∵ By Cauchy-Riemann equations, ∂x = − ∂y 
∂x ∂y  
= 2 x + 1 − i (− 2 y).
To apply Milne-Thomson’s method putting x = z and y = 0, we get
f ′ (z ) = 2 z + 1.
Integrating with respect to z, we get
f (z ) = z 2 + z + constant

= ( x + iy)2 + ( x + iy) + constant

= ( x2 − y2 + x) + 2 ixy + iy + ic

= u + i (2 xy + y + c )
= u + i v, where v = 2 xy + y + c .
We know that if f (z ) = u + i v is analytic, then the orthogonal trajectory of the family
of curves u = constant is the family of curves v = constant.
Hence, 2 xy + y = c where c is an arbitrary constant, is the orthogonal trajectory of the
family of curves x2 − y2 + x = c .
2 sin 2 x
Example 19: If u + v = , find the corresponding analytic function
e2 y
+ e −2 y
− 2 cos 2 x
f (z ) = u + iv.
Solution: We have f (z ) = u + iv, i f (z ) = iu − v.
∴ (1 + i) f (z ) = u − v + i (u + v) = U + iV , say.
2 sin 2 x sin 2 x
Here V =u+v= 2y = ⋅
e + e −2 y − 2 cos 2 x cosh 2 y − cos 2 x
 C-33

∂V ∂V
Let = ψ1 ( x, y) and = ψ2 ( x, y).
∂y ∂x
∂V sin 2 x (2 sinh 2 y)
Then ψ1 ( x, y) = =−
∂y (cosh 2 y − cos 2 x)2

∂V 2 cos 2 x (cosh 2 y − cos 2 x) − 2 sin2 2 x

and ψ2 ( x, y) = =
∂x (cosh 2 y − cos 2 x)2
2 cos 2 x cosh 2 y − 2
= ⋅
(cosh 2 y − cos 2 x)2
sin 2 z (2 sinh 0 )
Now ψ1 (z , 0 ) = − =0
(cosh 0 − cos 2 z )2
2 cos 2 z cosh 0 − 2 2 cos 2 z − 2
and ψ2 (z , 0 ) = 2
=
(cosh 0 − cos 2 z ) (1 − cos 2 z)2
−2
= = − cosec2 z .
1 − cos 2 z
By Milne’s method, we have
(1 + i) f (z ) = ∫ [ψ1 (z , 0 ) + i ψ2 (z , 0 )] dz + c
= ∫ (0 − i cosec2 z ) dz + c = i cot z + c
i c 1
∴ f (z ) = cot z + = (1 + i) cot z + d.
1+ i 1+ i 2
cos x + sin x − e − y
Example 20: If f (z ) = u + iv is an analytic function of z and u − v = , find
2 cos x − e y − e − y
π
f (z ) subject to the condition f   = 0 .
 2
cos x + sin x − e − y
Solution:Here u − v =
2 cos x − e y − e − y
1  2 cos x + 2 sin x − 2 e − y 
= 1 + y −y
− 1
2  2 cos x − e − e 
1  2 sin x + e y − e − y  1  sin x + sinh y 
= 1 + − y
= 1 + ⋅
2 y
 2 cos x − e − e  2  cos x − cosh y 

∂u ∂v 1 cos x (cos x − cosh y) + (sin x + sinh y) sin x

Now − =  
∂x ∂x 2  (cos x − cosh y)2 

1 1 − cos x cosh y + sin x sinh y 

=   …(1)
2  (cos x − cosh y)2 

∂u ∂v 1 cosh y (cos x − cosh y) + sinh y (sin x + sinh y)

and − =  
∂y ∂y 2  (cos x − cosh y)2 
 C-34

1 cosh y cos x + sinh y sin x − 1

=  
2  (cos x − cosh y)2 

∂v ∂u 1 cosh y cos x + sinh y sin x − 1

or − − =  , …(2)
∂x ∂x 2  (cos x − cosh y)2 
by Cauchy-Riemann equations.
Solving (1) and (2), we get
∂u 1  1 − cos x cosh y 
=   = φ1 ( x, y)
∂x 2 (cos x − cosh y)2 
∂v sin x sinh y
and =− = φ2 ( x, y).
∂x 2 (cos x − cosh y)2
∂u ∂v
∴ f ′ (z ) = +i = φ1 (z , 0 ) + i φ 2 (z , 0 )
∂x ∂x
1 1 − cos z 1 1 z
= = = cosec2 ⋅
2 (cos z − 1)2 2 (1 − cos z ) 4 2
1 1 1 1
∴ f (z ) =
4 ∫ cosec2
2
z dz + c = − cot z + c .
2 2
π
At z = , f (z ) = 0 .
2
π π 1
c = f   + cot = ⋅
1
∴
 2 2 4 2
1 1
∴ f (z ) = (1 − cot z ).
2 2
Example 21: If f (z ) = u + iv be an analytic function in a domain D, prove that f (z ) is
constant in D if any one of the following conditions holds:
(i) f ′ (z ) vanishes identically in D.
(ii) R ( f (z )) = u = constant. (Kanpur 2003)
(iii) I ( f (z )) = v = constant. (Garhwal 2000)

(iv) | f (z )| = constant.
(v) arg f (z ) = constant.
Solution: Since f (z ) = u + iv is analytic in D, therefore it satisfies Cauchy-Riemann
equations,
∂u ∂v ∂u ∂v 
= , =− ⋅ …(1)
∂x ∂y ∂y ∂x 
∂u ∂v ∂v ∂u
(i) We have f ′ (z ) = +i = −i , from (1)
∂x ∂x ∂y ∂y

∴ if f ′ (z ) = 0 , we have
∂u ∂v ∂v ∂u
+i = 0 and −i = 0.
∂x ∂x ∂y ∂y
 C-35

∂u ∂v ∂v ∂u
∴ = 0, = 0, = 0 and = 0.
∂x ∂x ∂y ∂y

Thus u and v are constants and consequently f (z ) is a constant function.

(ii) R ( f (z )) = u = constant
∂u ∂u
⇒ =0 = ⋅
∂x ∂y
∂u ∂v ∂u ∂u
Now f ′ (z ) = +i = −i , from (1)
∂x ∂x ∂x ∂y

=0
∴ f (z ) is a constant function.

(iii) I ( f (z )) = v = constant
∂v ∂v
⇒ = 0, = 0.
∂x ∂y
∂u ∂v ∂v ∂v
Now f ′ (z ) = +i = + i , from (1)
∂x ∂x ∂y ∂x

=0
∴ f (z ) is a constant function.

(iv) | f (z )| = constant ⇒ u2 + v2 = constant

∂u ∂v ∂u ∂v
⇒ u +v = 0 and u +v =0
∂x ∂x ∂y ∂y
∂u ∂v ∂v ∂u
⇒ u +v = 0 and − u +v =0
∂x ∂x ∂x ∂x
∂u ∂v
⇒ = 0 and = 0 , provided u2 + v2 ≠ 0 .
∂x ∂x
Therefore if u2 + v2 ≠ 0 , u and v are constants and consequently f (z ) is constant. If
u2 + v2 = 0 at a single point, it is constantly zero and so f (z ) is zero. Hence, in either
case f (z ) is constant.

v
(v) Here arg f (z ) = tan−1 ⋅
u
arg f (z ) = c (constant)
v
⇒ tan−1 = c ⇒ (v / u) = tan c
u
⇒ u = v cot c ⇒ u = k v, taking cot c = k .
We observe u − kv = 0 unless v is identically zero. But u − kv is the real part of (1 + ik ) f ,
therefore it follows from part (ii) that (1 + ik ) f is constant. But (1 + ik ) is a constant,
therefore f is also a constant.
 C-36

Example 22: If f (z ) is an analytic function of z, prove that

 ∂2 ∂2 
 2
+  | R f (z )|2 = 2 | f ′ (z )|2 .
 ∂x ∂ y2 
Solution: Let f (z ) = u + iv , where z = x + iy.
Then R f (z ) = u.
∂u2 ∂u
Now = 2u
∂x ∂x
2
∂2 u2  ∂u  ∂2 u
and = 2  + 2u 2 ⋅
∂x2  ∂x  ∂x
2
∂2 u2  ∂u  ∂2 u
Similarly =2  + 2u ⋅
∂ y2  ∂y  ∂y 2
 2
 ∂u  
2
 ∂2 u ∂2 u 
∂2 u2 ∂2 u2   ∂u 
∴ + = 2    +    + u  2 + 2  
∂x2
∂y 2
   ∂x   ∂y    ∂x ∂y 
  
 ∂2 ∂2   ∂u 2  ∂u  2 
  
or  2 + 2  u2 = 2    +    + 0,
 ∂x ∂y    ∂x   ∂y  

since u is a harmonic function
 ∂2 ∂2    ∂u  2  ∂v  2 
or  2 + 2  u2 = 2   +   ,
   ∂x  
 ∂x ∂y   ∂x 
using Cauchy-Riemann equa tions
 ∂2 ∂2  ∂u ∂v
or  2 + 2  | u|2 = 2 | f ′ (z )|2 , since f ′ (z ) = +i
 ∂x ∂y  ∂x ∂x

 ∂2 ∂2 
Hence,  2
+  | R f (z )|2 = 2 | f ′ (z )|2 .
 ∂x ∂ y2 

Comprehensive Exercise 1

1. Show that the function f (z ) = xy + iy is everywhere continuous but is not

analytic. (Rohilkhand 2011; Purvanchal 11, 12)

2. Show that the function e x (cos y + i sin y) is holomorphic and find its
derivative. (Lucknow 2006, 13)
2
xy ( x + iy) f (z ) − f (0 )
3. If f (z ) = , z ≠ 0 , f (0 ) = 0 , prove that → 0 as z → 0
x2 + y4 z
along any radius vector but not as z → 0 in any manner.
(Gorakhpur 2007; Purvanchal 12)
 C-37

x2 y5 ( x + iy)
4. Examine the nature of the function f (z ) = , z ≠ 0 , f (0 ) = 0
x 4 + y10
in a region including the origin. (Meerut 2002; Gorakhpur 09, 13)
5. Show that an analytic function with constant argument is constant.
6. Show that an analytic function with constant modulus is constant.
Or
Show that an analytic function cannot have a constant modulus without
reducing to a constant.
7. For what values of z the function w defined by z = e − v (cos u + i sin u), where
w = u + iv ceases to be analytic ?
8. Prove that u = y3 − 3 x2 y is a harmonic function. Determine its harmonic
conjugate and find the corresponding analytic function f (z ) in terms of z.
(Purvanchal 2010, 12)
9. If u = e x ( x cos y − y sin y), find the analytic function u + iv.
(Kanpur 2003; Gorakhpur 05; Rohilkhand 09, 11; Purvanchal 10; Meerut 12)
10. If u = ( x − 1)3 − 3 xy2 + 3 y2 , determine v so that u + iv is a regular function of
x + iy. (Meerut 2001; Gorakhpur 06, 09, 13)

11. Prove that if u = x2 − y2 , v = − y / ( x2 + y2 ) both u and v satisfy Laplace’s

equation but u + iv is not an analytic function of z. (Agra 2012)
2 2
12. Show that the function u = sin x cosh y + 2 cos x sinh y + x − y + 4 xy
satisfies Laplace’s equation and find the corresponding analytic function u + iv.
13. Find the analytic function of which the real part is
e − x {( x2 − y2 ) cos y + 2 x sin y}. (Purvanchal 2007; 09)
14. Construct the analytic function f (z ) = u + iv , where
(i) u = x3 − 3 xy2 + 3 x + 1, (Gorakhpur 2007)
3 2
(ii) u = y − 3 x y.
15. Prove that the function u = x3 − 3 xy2 + 3 x2 − 3 y2 + 1 satisfies Laplace’s
equation and determine the corresponding analytic function. (Meerut 2000)

16. If f (z ) = u + iv is analytic function and u − v = e x (cos y − sin y), find f (z ) in

terms of z. (Garhwal 2000; Purvanchal 09; Gorakhpur 11; Agra 12)

17. If u − v = ( x − y)( x2 + 4 xy + y2 ) and f (z ) = u + iv is an analytic function of

z = x + iy, find f (z ) in terms of z.
(Rohilkhand 2008; Purvanchal 08; Gorakhpur 10)

18. If f (z ) = u + iv is an analytic function of z = x + iy and

y
e − cos x + sin x π 3−i
u−v= , find f (z ) subject to the condition f   = ⋅
cosh y − cos x  2 2
(Gorakhpur 2008)
 C-38

19. If f (z ) = u + iv is an analytic function of z = x + iy and ψ any function of x and y

with differential coefficients of the first and second orders, prove that
2
  ∂ψ   ∂ψ  
2 2 2
 ∂ψ   ∂ψ  2
(i)   +  =   +   | f ′ (z )|
 ∂x   ∂y    ∂u   ∂v  

∂2 ψ ∂2 ψ  ∂2 ψ ∂2 ψ 
(ii) + = + 2  | f ′ (z )|2 .
∂x2 ∂ y2  ∂u
2
∂v 
 ∂2 ∂2 
20. If f (z ) is a regular function of z, prove that  2 + 2  | f (z )|2 = 4 | f ′ (z )|2 .
 ∂x ∂y 
(Kanpur 2001)
 ∂2 ∂2 
21. If w = f (z ) is a regular function of z, prove that  2 + 2  log | f ′ (z )| = 0 .
 ∂x ∂y 
(Kanpur 2002)
If | f ′ (z )| is the product of a function of x and a function of y, show that
f ′ (z ) = exp (αz 2 + βz + γ ),

where α is a real and β, γ are complex constants. (Rohilkhand 2012)

A nswers 1

7. z =0 9. f (z ) = ze z + c

10. y = 3 x2 y − 6 xy + 3 y − y3 + c 13. f (z ) = ( x + iy)2 ⋅ e − x [cos y − i sin y] + c

14. (i) z 3 + 3 z + 1 + ci , (ii) i (z 3 + c )

16. f (z ) = e z + c 17. f (z ) = − iz 3 + d
1 1
18. f (z ) = cot z + (1 − i)
2 2

Objective Type Questions

Multiple Choice Questions

Indicate the correct answer for each question by writing the corresponding letter from (a),
(b), (c) and (d).
1. Cauchy-Riemann equations for w = u + iv = f (z ) are :
(a) u x = v x , u y = v y (b) u x = v y , u y = v x
(c) u x = v y , u y = − v x (d) None of these.
(Rohilkhand 2011)
 C-39

2. Which of the following is not correct for analytic functions f (z ) and g (z ) in a

region R ?
(a) f (z ) + g (z ) is analytic in R
(b) f (z ) − g (z ) is analytic in R
(c) f (z ) g (z ) is analytic in R
(d) f (z ) / g (z ) is analytic in R .
3. Which of the following is correct for w = f (z ) ?
dw ∂w dw ∂w
(a) = (b) =−
dz ∂x dz ∂x
dw ∂w dw ∂w
(c) = (d) =− ⋅
dz ∂y dz ∂y
4. The derivative of a function w = f (z ) in polar form is given by :
dw ∂w iθ dw ∂w iθ
(a) = e (b) =− e
dz ∂r dz ∂θ
dw ∂w − iθ dw ∂w − iθ
(c) = e (d) =− e .
dz ∂r dz ∂θ
5. Any function of x and y possessing continuous partial derivatives of the first and
second orders is called a harmonic function if it satisfies :
(a) Euler’s equation (b) Laplace’s equation
(c) Lagrange’s equation (d) None of these.
6. An analytic function with constant modulus is :
(a) variable (b) constant
(c) may be variable or constant (d) None of these.
7. If f (z ) = u + iv is analytic function in a finite region and u = x3 − 3 xy2 , then v is:
(a) 3 x2 y − y3 + c (b) 3 x2 y2 − y3 + c

(c) 3 x2 y − y2 + c (d) None of these.

8. The analytic function whose real part is e x cos y is :

(a) e z + ci (b) e2 z

(c) xe z (d) None of these.

x
9. The analytic function whose real part is e ( x cos y − y sin y) is :
(a) ze z + ci (b) z 2 e z
2
+ iy
(c) ze x (d) None of these.
−v
10. The function w defined by z = e (cos u + i sin u) ceases to be analytic at z where
z is :
(a) 1 (b) 0
(c) ∞ (d) None of these.
 C-40

Answers

1. (c) 2. (d) 3. (a) 4. (c) 5. (b) 6. (b)

7. (a) 8. (a) 9. (a) 10. (b)

¨
 C-41

1 Introduction
e are familiar with the theory of integration of a real variable. In the case of a real
W variable, the integration is considered from two points of view, namely, the
indefinite integration as a process inverse to differentiation and definite integration as
the limit of a sum. There is a similar distinction between definite and indefinite
integrals of a complex variable. As in the case of real variables, the concept of indefinite
integral as the process inverse to differentiation also extends to a function of a complex
variable. The indefinite integral of a complex variable is a function whose derivative
equals a given analytic function in a region. However the concept of definite integral of
a real variable does not extend straightway to the domain of complex variables. For
b
example, in the case of real variables, the path of integration of ∫a f ( x) dx is always along the
real axis (x-axis) from x = a to x = b. But for a complex function f (z ), the path of the definite
b
integral ∫ f (z ) dz may be along any curve joining the points z = a and z = b, i.e., the value of
a
the integral depends upon the path of integration. However, this variation in the value of
definite integral will disappear in some special circumstances. Definite integrals of a
complex variable are usually known as line integrals.
 C-42

The theory of line integrals, along with the theory of power series and residues forms a
very useful and important part of the theory of functions of a complex variable. These
theories contain some of the most powerful theorems which have application in both
pure and applied mathematics.

2 Definitions
(i) Partition: Consider a closed interval [a, b], where a and b are real numbers.
Divide [a, b ] into n sub-intervals
[a = t0 , t1 ], [t1, t2 ],……,[tn − 1, tn = b ]
by inserting n − 1 intermediate points t1, t2 ,……, tn − 1 satisfying
a < t1 < t2 < …… < tn.
Then we call the set P = {t0 , t1, t2 ,……, tn}
a partition of [a, b]. The greatest number among t1 − t0 , t2 − t1,……, tn − tn − 1 is called
the norm of the partition P and is denoted by | P| .
(ii) Arcs and closed curves: We know that the equation
z = z (t) = x (t) + i y (t),
where a ≤ t ≤ b and x (t), y (t) are continuous functions, represents an arc L in the
Argand plane, i.e., an arc is the set of all image points of a closed finite interval under a continuous
mapping.
The equations x = x (t), y = y (t) give the parametric representation of the arc in the
plane.
If z ′ (t) exists and is continuous, the arc L is said to be differentiable or continuously
differentiable. If in addition to the existence of z ′ (t), we also have z ′ (t) ≠ 0, the arc L is
said to be regular or smooth. Geometrically, at every point of a smooth arc there exists
a tangent whose direction is determined by arg z ′ (t). As a matter of fact, as t increases
from a to b, z continuously traces out the arc L and at the same time arg z ′ (t) varies
continuously since z ′ (t) changes continuously without vanishing.
If among various representations of an arc L there exists at least one representation,
such that the interval [a, b] can be divided into a finite number of sub-intervals
[a , a1], [a1, a2 ], …… ,[an − 1, b ]
on each of which z ′ (t) exists, then the arc L is said to be piecewise differentiable. If in
addition to this we also have z ′ (t) ≠ 0 on any of these sub-intervals, the arc L is said to
be piecewise smooth.
If t1 ≠ t2 ⇒ z (t1) ≠ z (t2 ), the arc L is called simple or Jordan arc.
If the points corresponding to the values a and b coincide, the arc L is said to be a closed
curve.
If the arc L is defined by z = z (t), a ≤ t ≤ b, then the arc defined by
z = z (− t), − b ≤ t ≤ − a is called the opposite arc of L and is denoted by − L or L−1.
 C-43

3 Rectifiable Arcs
Consider the arc L defined by z = z (t) = x (t) + i y (t), a ≤ t ≤ b.
Let P = {t0 , t1, t2 ,……, tn} be any partition of [a, b].
Corresponding to this partition, dividing the arc L into n sub-arcs
Lk = arc z k − 1 z k , (k = 1, 2,……, n)
where z k = z (tk ), (k = 0 , 1, 2,……, n).
Joining each of the points z0 , z1, z2 ,……, z n to the
next point by straight lines, we obtain a polygonal
curve. The length of this polygonal curve is given
n
by Σ | z k − z k − 1 | .
k =1

The arc L will be rectifiable if the least upper

bound of the sum
| z1 − z0 | + | z2 − z1 | +……+ | z n − z n − 1 |
…(1)
taken over all partitions P is finite
n
i.e., if sup Σ | z k − z k − 1 | = l < ∞.
P k =1

The non-negative real number l is called the length of the arc L. The arc L is said to be
non-rectifiable if the sum (1) becomes arbitrarily large for suitably chosen partitions.
Contours: A contour is a continuous chain of a finite number of regular arcs.
If A is the starting point of the first arc and B the end point of the last arc, the integral of
a function f (z ) along such a curve is written as ∫ f (z ) dz .
AB

A contour is said to be closed if it does not intersect itself and the starting point A of the first arc
coincides with the end point B of the last arc.

The integral along such closed contour C is written as ∫ f (z ) dz . The boundaries of

C
triangles and quadrilaterals are examples of closed contours.
Simply connected region and multiply connected region:A region in which every
closed curve can be shrunk to a point without passing out of the region is called a
simply connected region otherwise it is said to be multiply connected.
(Meerut 2002)

4 Functions of Bounded Variation

We can easily show that
| x (tk ) − x (tk − 1)| ≤ | z (tk ) − z (tk − 1)|,
 C-44

| y (tk ) − y (tk − 1)| ≤ | z (tk ) − z (tk − 1)|,

| z (tk ) − z (tk − 1)| = | x (tk ) − x (tk − 1) + i { y (tk ) − y (tk − 1)}|
≤ | x (tk ) − x (tk − 1)| + | i|| y (tk ) − y (tk − 1)|
or | z (tk ) − z (tk − 1)| ≤ | x (tk ) − x (tk − 1)| + | y (tk ) − y (tk − 1)| [∵ | i| = 1]
From above inequalities we conclude that the sum
| z (t1) − z (t0 )| + | z (t2 ) − z (t1)| +……+ | z (tn) − z (tn − 1)|
and the sums
| x (t1) − x (t0 )| + | x (t2 ) − x (t1)| +……+ | x (tn) − x (tn − 1)| ,
| y (t1) − y (t0 )| + | y (t2 ) − y (t1)| + ……+ | y (tn) − y (tn − 1)|
are bounded at the same time.
The functions x (t) and y (t) are said to be of bounded variation if the latter two of the
above three sums are bounded for all partitions P of [a , b ].
It can be easily proved that an arc z = z (t) = x (t) + i y (t), a ≤ t ≤ b is rectifiable if and only
if the functions x (t) and y (t) are of bounded variation in [a, b ].
It is not hard to show that a smooth arc L is rectifiable and that its length l is given by the
familiar formula
b
l= ∫a [ x ′ (t)2 + y ′ (t)2 ] dt ...(5)
b
or equivalently by l = ∫a | z ′ (t)| dt. ...(6)

5 Complex Integrals
(Meerut 2002)
Let f (z ) be a function of a complex variable z defined
and continuous on an arc L , where L is a rectifiable
arc defined by
z = z (t) = x (t) + i y (t), a ≤ t ≤ b.
Let there be any partition
P = {a = t0 , t1, t2 ,……, tn = b } of [a, b ].
Form the sum
SP = (z1 − z0 ) f (ζ1) + (z2 − z1) f (ζ2 ) +……
+ (z k − z k − 1) f (ζ k ) +……+ (z n − z n − 1) f (ζ n)
n
= Σ (z k − z k − 1) f (ζ k ), …(1)
k =1

where z k = z (tk ), ζ k = z (α k ), tk − 1 ≤ α k ≤ tk
and ζ k is a point on each arc joining the points z k −1 to z k .
Thus to form the sum SP , we choose an arbitrary point ζ k on each arc joining the points
z k −1 to z k and add the terms of the form (z k − z k −1) f (ζ k ), where k varies from 1 to n.
 C-45

For convenience, we shall write z k − z k −1 = ∆ z k .

The function f (z ) is said to be integrable from a to b along the arc L if the sum SP taken
over all possible partitions P tends to a unique limit I as n → ∞ and | P| → 0.
lim n
∴ I= ∫L f (z ) dz = n → ∞
|P |→ ∞ k = 1
Σ (z k − z k − 1) f (ζ k ).

∫L f (z ) dz is called the complex line integral or simply line integral of f (z ) along L or

the definite integral of f (z ) from a to b along L.

6 Evaluation of Some Integrals ab-initio (By Definition)

Example 1:Using the definition of an integral as the limit of a sum evaluate the following integrals

(i) ∫L dz and (ii) ∫L z dz ,

where L is any rectifiable arc joining the points z = α and z = β.

Solution: Both the integrals exist since the integrand is a continuous function in each
case.
lim n
(i) We have ∫L dz =
n→ ∞
Σ (z k − z k − 1), by definition
k =1

lim
= (z − z0 + z2 − z1 +……+ z n − z n − 1)
n→ ∞ 1
lim
= (z − z0 ) = β − α. [ ∵ z 0 = α, z n = β ]
n→ ∞ n

Note: If L is a closed curve, we have α = β and ∫ dz = 0.

L

lim n
(ii) We have ∫L f (z ) dz =
n→ ∞
Σ
k =1
f (ζ k ) (z k − z k − 1) by def.

lim n
∴ ∫L z dz =
n→ ∞
Σ ζ k (z k − z k − 1)
k =1
…(1)

where ζ k is any point on the sub-arc joining z k −1 to z k . Since ζ k is arbitrary, therefore

taking ζ k = z k and ζ k = z k − 1 successively in (1), we get
lim n
∫L z dz =
n→ ∞
Σ
k =1
z k (z k − z k − 1)

lim n
and ∫L z dz =
n→ ∞
Σ
k =1
z k − 1 (z k − z k − 1).
 C-46

Adding these two integrals, we get

lim n
2 ∫ z dz = Σ (z k + z k − 1) (z k − z k − 1)
L n → ∞ k =1
lim n
= Σ (z k2 − z 2k − 1)
n→ ∞ k =1

lim n
= Σ (z12 − z02 + z22 − z12 +……+ z n2 − z 2n − 1)
n→ ∞ k =1

lim n
= Σ (z n2 − z02 ) = β2 − α2 , since z0 = α , z n = β.
n→ ∞ k =1

1 2
∴ ∫L z dz =
2
( β − α2 ).

If L is a closed curve, we have α = β, so in this case ∫ z dz = 0.

L

Example 2: Evaluate ∫ | dz | (ab-initio)

L

Solution: The above integral exists since the integrand is a continuous function.
Here f (z ) = 1 and we have | dz | in place of dz.
lim n
We have ∫L | dz | = n → ∞ k Σ=1 | z k − z k −1 |
lim
= [| z1 − z0 | + | z2 − z1 | +……+ | z n − z n − 1 |]
n→ ∞
lim
= [chord z1 z0 + chord z2 z1 +……+ chord z n z n − 1]
n→ ∞

= arc length of L .

7 Reduction of Complex Integrals to Real Integrals

Theorem: Let the arc L defined by
z = z (t) = x (t) + i y (t), a ≤ t ≤ b
be continuously differentiable and let
f (z ) = u ( x, y) + i v ( x, y)
be continuous over L. Then (i) L is rectifiable
b
(ii) ∫L f (z ) dz = ∫a { u ( x (t), y (t)) x ′ (t) − v ( x (t), y (t)) y ′ (t)} dt
b
+ i ∫ {v ( x (t), y (t)) x ′ (t) + u ( x (t), y (t)) y ′ (t)} dt.
a

Proof: (i) Let P = { a = t0 , t1 , t2 ,……, tn = b } be any partition of [ a, b ].

n n
We have Σ | z k − z k − 1 | = Σ | z (tk ) − z (tk − 1)| , where z k = z (tk )
k =1 k =1
 C-47

= Σ | x (tk ) − x (tk − 1) + i { y (tk ) − y (tk − 1)}| ,

since z (tk ) = x (tk ) + i y (tk )
≤ Σ | x (tk ) − x (tk − 1)| + Σ | y (tk ) − y (tk − 1)| . …(1)
Since x (t) and y (t) are continuously differentiable in [ a, b ], therefore by Lagrange’s
mean value theorem there exist real numbers γ k and δ k in ] tk −1, tk [ such that
x (tk ) − x (tk − 1) = (tk − tk − 1) x ′ (γ k ) …(2)
y (tk ) − y (tk − 1) = (tk − tk − 1) y ′ (δ k ). …(3)
Again z (t) is continuously differentiable in [ a, b ], therefore the derivatives x ′ (t)and
y ′ (t) are continuous in [ a, b ]. Consequently x ′ (t) and y ′ (t) are bounded in [ a, b ].
Therefore there exists a real number M such that
| x ′ (t)| ≤ M and
| y ′ (t)| ≤ M , V t ∈[a, b ]. …(4)
From (1), (2), (3) and (4), we have
Σ | z k − z k − 1 | ≤ Σ | tk − tk − 1 || x ′ (γ k )| + Σ | tk − tk − 1 || y ′ (δ k )|
n n n
≤ Σ M | tk − tk − 1 | + Σ M | tk − tk − 1 | = 2 M Σ | tk − tk − 1 |
k =1 k =1 k =1

= 2 M (t1 − t0 + t2 − t1 +……+ tn − tn − 1)
= 2 M (tn − t0 ) = 2 M (b − a).
n
∴ Σ | z k − z k − 1 | ≤ 2 M (b − a). …(5)
k =1

sup n
Thus we can say that Σ | z − k k − 1| < ∞.
P k =1 k

Hence the arc L is rectifiable.

(ii) Consider the sum S = Σ (z k − z k − 1) f (ζ k )
where ζ k is a point on the arc joining the points z k −1 and z k . Let τ k be the parameter of ζ k .
Then τ k lies between z k −1 and z k . We have
f [ z (t) ] = u [ x (t), y (t) ] + i v [ x (t), y (t) ]
= φ (t) + i ψ (t).
Then φ (t) = u [ x (t), y (t) ], ψ (t) = v [ x (t), y (t)]. …(6)
Now S = Σ [ x (tk ) + i y (tk ) − x (tk − 1) − iy (tk − 1)][ φ (τ k ) + i ψ (τ k )]
= Σ [ x (tk ) − x (tk − 1) + i { y (tk ) − y (tk − 1)}] [φ (τ k ) + i ψ (τ k )]
= Σ [(tk − tk − 1) x ′ (γ k ) + i (tk − tk − 1) y ′ (δ k )][φ (τ k ) + iψ (τ k )],
from (2) and (3)
= Σ (tk − tk − 1) x ′ ( γ k ) φ (τ k ) − (tk − tk − 1) y ′ (δ k ) ψ (τ k )
+ i Σ (tk − tk − 1) x ′ (γ k ) ψ (τ k ) + i Σ (tk − tk − 1) y ′ (δ k ) φ (τ k )
= S1 + S2 + i (S3 + S4 ), …(7)
where S1 = Σ (tk − tk − 1) x ′ (γ k ) φ (τ k ), S2 = − Σ (tk − tk − 1) y ′ (δ k ) ψ (τ k ),
 C-48

S3 = Σ (tk − tk − 1) x ′ (γ k ) ψ (τ k ),
and S4 = Σ (tk − tk − 1) y ′ (δ k ) φ (τ k ).
We can write
n
S1 = Σ (tk − tk − 1) x ′ (τ k ) φ (τ k )
k =1
n
+ Σ (tk − tk − 1) [ x ′ ( γ k ) − x ′ (τ k ) ] φ (τ k ). …(8)
k =1

Since x ′ (t) is continuous in the closed and bounded interval [a, b ], therefore it is
uniformly continuous in [a, b ], so that for given ε > 0 , there exists a δ > 0 such that
| x ′ (r) − x ′ (s)| < ε whenever | r − s | < δ
where r, s are in [a, b ].
Thus for any partition P of [a, b ] with norm ≤ δ, we have
| Σ (tk − tk −1) [ x ′ (γ k ) − x ′ (τ k )] φ (τ k )| ≤ ε (b − a) M1 …(9)
{Since z (t) is continuous over L therefore φ (t) and ψ (t) are continuous on [a, b] and
consequently they are bounded on [a, b ] . Therefore there exists a number M1 such that
| φ (t)| ≤ M1 , V t ∈[a, b ].}
As n → ∞ and| P| → 0, we conclude from (9) that the second term on the right side of
(8) tends to zero and the first term tends to
b
∫a φ (t) x ′ (t) dt.
b
∴ lim S1 = ∫a φ (t) x ′ (t) dt.
b b
Similarly lim S2 = − ∫a ψ (t) y ′ (t) dt, lim S3 = ∫a ψ (t) x ′ (t) dt
b
and lim S4 = ∫a φ (t) y ′ (t) dt.

Taking limit of both sides of (7) as n → ∞ and| P| → 0 and using the above results, we
get
b
∫L f (z ) dz = ∫a { φ (t) x ′ (t) − ψ (t) y ′ (t)} dt
b
+ i∫ {ψ (t) x ′ (t) + φ (t) y ′ (t)} dt
a
b
= ∫a [ u ( x (t), y (t)) x ′ (t) − v ( x (t), y (t) ) y ′ (t) ] dt

b
+ i∫ [ v ( x (t), y (t)) x ′ (t) + u ( x (t), y (t) ) y ′ (t) ] dt.
a

8 Some Elementary Properties of Complex Integrals

Prop. 1: ∫ L { f ( z) + φ ( z)} dz = ∫ L f ( z) dz + ∫ L φ ( z) dz.
We can generalize this property for a finite number of functions.
 C-49

Prop. 2: ∫ L f ( z) dz = − ∫− L f ( z) dz, where − L is the opposite arc of L .

Prop. 3: ∫L1 + L2 f ( z) dz = ∫ L1 f ( z) dz + ∫ L f ( z) dz,

where the end point of L1 coincides with the initial point of L2 .
This property can be generalized for a finite number of arcs provided the end point of
the preceding arc coincides with the initial point of the arc which follows it. Hence, if
L = L1 + L2 + … + Ln
where the final point of Lk coincides with the initial point of
Lk + 1 (k = 1, 2, ..., n − 1),
then ∫L f (z ) dz = ∫L f (z ) dz + ∫ L2 f (z ) dz + .... + ∫ Ln f (z ) dz .

Prop. 4: ∫L c f ( z) dz = c ∫ L f ( z) dz, where c is any complex constant.

These properties can be easily proved by the definition of a complex integral as the limit
of a sum.

Prop. 5: ∫ L [c1 f1 ( z) + c2 f2 ( z) + … + cn fn ( z)] dz

= c1 ∫ L f1 ( z) dz + c2 ∫ L f2 ( z) dz + … + cn ∫ L fn ( z) dz.
where c1, c2 , ……, c n are complex constants.
This property follows directly from properties 1 and 4.

Prop. 6: ∫ f ( z) dz ≤ ∫ L| f ( z) || dz | .
L 
   
Proof: We have R c
 ∫ L f (z ) dz  = R ∫ L c f (z ) dz  , prop. 4

where c is a complex constant.
Since c is arbitrary therefore taking c = e − i θ where θ is any real number.
   
∴ R e − i θ ∫L f (z ) dz  = R ∫ e − i θ f (z ) dz  = ∫L R [e − i θ f (z ) dz ]
   L 
−iθ
≤ ∫L | e f (z ) dz | [∵ R (z ) ≤ | z |]

= ∫ L | f (z )|| dz |. …(1)

Again since θ is any real number, therefore taking θ = arg ∫L f (z ) dz so that we can

write

∫ L f (z ) dz = ∫ L f (z ) dz ei θ .
 C-50

Also in this case, we have

   
R e − i θ ∫L f (z ) dz  = R e − i θ  ∫ f (z ) dz  e i θ 
    L  

= ∫ f (z ) dz . …(2)
 L 

From (1) and (2), we have  ∫ f (z ) dz ≤ ∫ L | f (z )|| dz |.

 L 

1
Example 3: Prove that the value of the integral of along a semi-circular arc | z | = 1 from
z
−1 to 1 is − πi or πi according as the arc lies above or below the real axis.
Solution: The given circle is | z | = 1. Parametric
equation of the circle is z = e i θ , where 0 ≤ θ ≤ 2 π.
iθ
We have dz = i e dθ.
As z moves from −1to 1 along the semi-circular arc
above the real axis, θ varies from π to 0. In this case,
we have
1 0 1 iθ
∫ CBA z dz = ∫ π e i θ i e dθ
0
= i∫ dθ = − iπ.
π

Again when z moves from −1to 1along the semi-circular arc below the real axis, θ varies
from π to 2π.
dz 2π i e i θ 2π
∴ ∫ CDA z = ∫ π e i θ dθ = i ∫ π dθ = πi.
dz
Note: We have ∫ = − i π,
CBA z
dz
therefore ∫ ABC z = i π.
dz
Also ∫ CDA z = i π.
dz dz dz
Hence ∫ ABCD z = ∫ ABC z + ∫ CDA z = 2πi.
1+i
Example 4: Find the value of the integral ∫0 ( x − y + i x2 ) dz .

(i) Along the straight line from z = 0 to z = 1 + i.

(ii) Along the real axis from z = 0 to z = 1and then along a line parallel to the imaginary axis from
z = 1 to z = 1 + i . (Meerut 2002; Rohilkhand 09; Gorakhpur 12, 14, 16)
 C-51

Solution: We have z = x + i y
∴ dz = dx + i dy.
Let A be the point of affix 1 and B be the point of affix 1 + i in the Argand plane. Join OB
and AB.
(i) OB is the straight line joining z = 0 to z = 1 + i.
Obviously on OB, we have y = x
∴ dy = dx.
Now ∫ OB ( x − y + ix2 ) dz

1
= ∫0 ( x − x + ix2 ) (1 + i) dx

1
1 1 
= ∫0 i (1 + i) x2 dx = (− 1 + i)  x3 
3  0
1
= (− 1 + i).
3
(ii) OA is the line from z = 0 to z = 1along the real axis and AB is the line from z = 1to
z = 1 + i parallel to the imaginary axis. On the line OA , y = 0,
∴ z = x + iy = x and dz = dx.
1
1 x 2 x 3 1 i
∫ OA ∫0
2 2
∴ ( x − y + ix ) dz = ( x + ix ) dx =  +i  = + .
 2 3 0 2 3

On the line AB, x = 1, therefore z = 1 + iy, dz = i dy.

1
∴ ∫ AB ( x − y + ix2 ) dz = ∫0 ( 1 + i − y ) i dy

1
 y2  i
= i (1 + i ) y −  = − 1+ .
 2 
0
2

1+i
Hence ∫0 ( x − y + i x2 ) dz along the contour OAB

= ∫ OA ( x − y + i x2 ) dz + ∫ AB ( x − y + i x2 ) dz

1 i i 1 5
= + − 1 + = − + i.
2 3 2 2 6

Example 5: Evaluate ∫C (z 2 + 3 z + 2) dz where C is the arc of the cycloid

x = a (θ + sin θ), y = a (1 − cos θ) between the points (0, 0) and (πa, 2 a).
Solution: Here f (z ) = z 2 + 3 z + 2 is a polynomial so f (z ) is analytic in z-plane,
therefore the integral of f (z ) between the points (0, 0) and (πa, 2 a) is independent of
the path joining these points.
The path of integration consists of :
 C-52

(i) the part of real axis from the point (0, 0) to the point (πa, 0 ).
(ii) a line parallel to y-axis from the point (πa, 0 ) to the point (πa, 2 a).

∴ ∫C (z 2 + 3 z + 2) dz
πa 2a
= ∫0 ( x2 + 3 x + 2) dx + ∫0 {(πa + iy)2 + 3 (πa + iy) + 2} i dy
πa 2a
1 3  1 3 
=  x3 + x2 + 2 x +  (πa + iy)3 + (πa + iy)2 + 2 iy
3 2 0 3 2 0
1 3 1 3
= (πa)3 + (πa)2 + 2 πa + (πa + i2 a)3 + (πa + i2 a)2
3 2 3 2
1 3
+ 4 ia − (πa)3 − (πa)2
3 2
1 3 3 2
= 2 πa + (πa + i 2 a) + (πa + i2 a) + 4 ia.
3 2

Comprehensive Exercise 1

(2, 5)
1. Evaluate I = ∫(0,1) (3 x + y) dx + (2 y − x) dy along

(i) the curve y = x2 + 1 (ii) the line joining (0 , 1) and (2, 5)

(iii) the line from (0 ,1) to (0 , 5) and then from (0 , 5) to (2 , 5).

2. Evaluate ∫ ( z )2 dz around the circle

(i) | z | = 1, (ii) | z − 1| = 1.
3. Evaluate ∫C ( x2 − iy2 ) dz along the parabola y = 2 x2 from (1, 2) to (2 , 8) ; and

along the line joining (1, 2) and (2 , 8).

1+i
4. Evaluate the integral ∫0 z 2 dz.

dz
5. (i) Evaluate ∫ , where L represents the circle | z − a | = r .
L z−a
(Purvanchal 2011; Gorakhpur 15)

(ii) Evaluate ∫ n
(z − a) dz when n ≠ −1 and C is the circle | z − a | = r.
C
(Gorakhpur 2007)
(iii) Evaluate the integral ∫ n
z dz , n ≠ −1, C : | z | = 1.
C (Gorakhpur 2014)

A nswers 1

88
1. (i) ; (ii) 32 ; (iii) 40 2. (i) 0, (ii) 4πi
3
 C-53

511 49 65 1
3. (i) −i ; (ii) − 14 i 4. (1 + i)3
3 5 3 3
5. (i) 2πi (ii) 0 (iii) 0

9 An Upper Bound for a Complex Integral

Theorem: If a function f (z ) is continuous on a contour L of length l and if M be the upper
bound of | f (z )| on L i.e., | f (z )| ≤ M on L , then
 f (z ) dz ≤ M l.
 ∫L  (Gorakhpur 2008, 10)

Proof: Proceeding as in article 5, we have

n
SP = Σ (z k − z k − 1) f (ζ k ).
k =1

Now | SP | = | Σ (z k − z k − 1) f (ζ k )|
≤ Σ |(z k − z k − 1) f (ζ k )| [ ∵| a + b | ≤ | a | + | b |]
= Σ | z k − z k − 1 || f (ζ k )|
≤ M Σ | z k − z k − 1 | , since ζ k is a point on L.
∴ lim| Σ (z k − z k − 1) f (ζ k )| ≤ lim M Σ | z k − z k − 1 |

 f (z ) dz ≤ M ∫ | dz |.
 ∫L
or
 L

Hence ∫ f (z ) dz ≤ M l, (see example 2 of article 6.)

 L 

10 Line Integrals as Functions of Arcs

We have seen that a line integral ∫ f (z ) dz over an arc L can be put in the form
L

∫L (u + iv) (dx + i dy).

General line integrals of the form ∫ p dx + q dy are often considered as functions (or
L

functionals) of the arc L under the assumption that p and q are defined and continuous
in a domain D and the arc L can vary freely in D. There exists a special class of integrals
characterized by the property that the integral over an arc depends only on its end
points. This means that if the two arcs L1 and L2 have the same initial point and the
same final point, then

∫ L1 p dx + q dy = ∫ L2 p dx + q dy.
 C-54

Theorem 1: The following statements are equivalent :

A line integral of f (z ) over an arc L depends only on the end points of L .
The integral of f (z ) over any closed curve is zero.
Proof: First suppose that the line integral of f (z ) over any closed curve is zero. Let L1
and L2 be any two arcs having the same end points. Then L1 − L2 is a closed curve.

∴ ∫ L1 −L2 f (z ) dz = 0

or ∫ L1 f (z ) dz − ∫ L2 f (z ) dz = 0 , by a property of complex integration

or ∫ L1 f (z ) dz = ∫ L2 f (z ) dz .

Hence the line integral of f (z ) over an arc depends only on its end points provided the
integral of f (z ) over any closed curve is zero.
Conversely, suppose the line integral of f (z ) over any two arcs with same end points be
same.
Consider a closed curve Γ. Then Γ and −Γ have the same end points, so that

∫Γ f (z ) dz = ∫ −Γ f (z ) dz

We know ∫ f (z ) dz = − ∫Γ f (z ) dz , by a property of complex integration.

−Γ

∴ ∫Γ f (z ) dz = − ∫Γ f (z ) dz

or 2∫ f (z ) dz = 0 or ∫Γ f (z ) dz = 0.
Γ

Thus the integral of f (z ) over any closed curve is zero.

Theorem 2: The line integral ∫ p dx + q dy, defined in a domain D, depends only on the end
points of Γ if and only if there exists a function U ( x, y ) in D such that
∂U ∂U
=p and = q.
∂x ∂y
∂U
Proof: The ‘if’ part. Let there exist a function U ( x, y ) in D such that = p and
∂x
∂U
= q. Also suppose a and b are the end points of Γ. Then we have
∂y
b  ∂U ∂U 
∫Γ p dx + q dy = ∫a 
 ∂x
dx +
∂y
dy 

b  ∂U dx ∂U dy  b d
= ∫a 
 ∂x dt
+  dt =
∂y dt  ∫a dt
U ( x (t), y (t)) dt

b
= [U ( x (t), y (t))] a = U ( x (b), y (b) ) − U ( x (a), y (a)),

which shows that the line integral depends only on the end points of Γ.
 C-55

The ‘only if’ part: Let us assume that the line integral ∫Γ p dx + q dy depends only on

the end points of Γ. Suppose ( x0 , y0 ) be a fixed point in D and ( x, y ) be any arbitrary

point in D. Join ( x0 , y0 ) to ( x, y ) by a polygonal arc Γ contained in D having its sides
parallel to coordinate axes.
Consider a function U ( x, y ) given by

U ( x, y ) = ∫Γ p dx + q dy.

Then the function U ( x, y ) is well defined since

according to assumption the integral depends
only on the end points of Γ. Also choose the last
segment of Γ parallel to x-axis so that y becomes
constant and dy = 0 and suppose that x varies
without changing the other segments. Choosing
x as a parameter on the last segment, we get
x
U ( x, y ) = ∫ p ( x, y ) dx + constant.

We have not specified the lower limit of x since it is insignificant for our purpose.
∂U
∴ = p.
∂x
∂U
Similarly choosing the last segment parallel to y-axis, we can show that = q.
∂y

 ∂U   ∂U 
Remark: (i) It is customary to write dU =  dx +   dy ...(1)
 ∂x   ∂y 

and we say that an expression p dx + q dy is an exact differential if it can be written in

the form (1). Using this terminology, the above theorem can be stated as :
An integral depends only on the end point if and only if the integrand is an exact differential.
(ii) We now determine the conditions under which
f (z ) dz = f (z ) dx + if (z ) dy
is an exact differential. By definition of an exact differential, there must exist a function
F (z ) in D such that
∂F (z )
= f (z )
∂x
∂F (z )
and = if (z ).
∂y
∂F ∂F
It follows that =−i , which is a Cauchy-Riemann equation.
∂x ∂y

Also f (z ) is by assumption continuous (otherwise ∫ f (z ) dz would not be defined).

L
Hence F (z ) is analytic.
 C-56

(iii) From the discussion in (i) and (ii), we conclude :

The integral ∫ f (z ) dz , with continuous f , depends only on the end points of L if and
L

only if f is the derivative of an analytic function in D.

For example, for n ≥ 0, the function (z − a)n is the derivative of (z − a)n +1 / (n + 1) which
is an analytic function in the whole complex plane. If Γ is any closed curve, then it
follows from theorem 1 and remark (iii) above that

∫L (z − a) n dz = 0.

If n is negative, but ≠ − 1, then also ∫ (z − a) n dz = 0 for all closed curve Γ which do not
L

pass through a, since in the complementary region of the point a the infinite integral is
still analytic and single-valued. If n = − 1, then (1) does not always hold. We have seen
in problem 13 after article 8 that
dz
∫ Γ z − a = 2πi ,
where Γ is any circle | z − a | = r.

11 Cauchy’s Fundamental Theorem

1
In example 3, after article 8 the integral of round the circle| z | = 1is 2iπ whereas in the
z
problem 8 after article 8 the integral of z 2 round the closed contour OLMO is zero. Here
1
we observe that the function is not analytic at z = 0 which is an interior point of the
z
circle | z | = 1 whereas the function z 2 is analytic throughout the interior and at the
boundary of the triangle OLM. Now we shall prove a very important theorem known as
Cauchy’s fundamental theorem which states :
If f (z ) is analytic at all points within and on the closed contour C, then

∫C f (z ) dz = 0.

Theorem 1: (Cauchy’s Theorem). Let D be a simply connected region and let f (z ) be a

single valued continuously differentiable function on D i.e., f ′ (z ) exists and is continuous at each
point of D. Then
∫ f (z ) dz = 0,
C

where C is any closed contour contained in D. (Rohilkhand 2008; Gorakhpur 13)

Proof: We have ∫C f (z ) dz = ∫C (u + i v ) ( dx + i dy )

= ∫C ( u dx − v dy ) + i ∫
C
( v dx + u dy ). …(1)
 C-57

To prove this theorem we shall use Green’s theorem for a plane which states :
∂Q ∂P
if P ( x, y ), Q ( x, y ), , are all continuous functions of x and y in the region D,
∂x ∂y
and C is any closed contour in D, then
 ∂Q ∂P 
∫C ( P dx + Q dy ) = ∫ ∫ D  ∂x −  dx dy .
∂y 

Since the function f (z ) is analytic in D therefore f ′ (z ) exists and

f ′ (z ) = u x + i v x = v y − i u y (by Cauchy-Riemann equations). Also f ′ (z ) is given to be
continuous at each point of D, therefore u, v, u x , v x , u y and v y are all continuous in D.
Thus all the conditions of Green’s theorem are satisfied. Hence from (1), we have
 ∂v ∂u   ∂u ∂v 
∫C f (z ) dz = ∫ ∫D −
 ∂ x
−
∂ y
 dx dy + i ∫ ∫ 
 D
−  dx dy
 ∂x ∂y 
 ∂v ∂v   ∂u ∂u 
= ∫ ∫ D  − ∂x + ∂x  dx dy + i ∫ ∫ D  ∂x − ∂x  dx dy,
by Cauchy–Riemann equations
= 0.
Note: This form of Cauchy’s theorem is quite useful in applied mathematics as the
continuity of the four partial derivatives u x , v x , u y and v y is generally assumed on
physical grounds.
Cauchy-Goursat theorem: In the statement of Cauchy’s theorem the function
f ′ (z ) is assumed to be a continuous function. It was Goursat who first proved the
theorem without considering the continuity of f ′ (z ). The revised form of the theorem
is known as Cauchy-Goursat theorem. We are giving here three independent proofs
of this important theorem.
Theorem 2: (Cauchy-Goursat theorem.) Let f (z ) be analytic in a simply connected
domain D and let C be any closed continuous rectifiable curve in D. Then

∫C f (z ) dz = 0.
(Meerut 2001; Gorakhpur 05, 08; Avadh 08; Purvanchal 08)
First we shall prove the following lemma known as Goursat’s lemma.
Lemma:Let f (z ) be analytic within and on a closed contour C. Then for every ε > 0, it is always
possible to divide the region within C into a finite number of squares and partial squares whose
boundaries are denoted by Si ( i = 1, 2,……, n ) such that there exists a point z i within each Si such
 f (z ) − f (z i) 
that  − f ′ (z i)< ε …(1)
 z − z i 
for each point z ( ≠ z i ) within or on Si ( i = 1, 2, ……, n ).
Proof of the lemma: Suppose the lemma is false. It means the lemma does not hold at
least in one mesh i.e., there exists ε > 0 such that in however small meshes (squares and
partial squares) we subdivide the region within C there will be at least one mesh (square
or a partial square) where the inequality (1) does not hold good.
 C-58

Let R denote the region within and on the closed

contour C. Cover the region R by a network of finite
number of meshes (squares and partial squares) by
drawing lines parallel to the coordinate axes. Then
as per assumption there is at least one mesh for
which (1) does not hold. Let us denote it by σ0 . It
may be a square or a partial square. Divide σ0 into
four equal squares. Then at least one of these
squares contains the points of R for which (1) is not
true. Suppose it is σ1. Quadrisect σ1 and repeat the above process. If this process comes
to an end after a finite number of steps we arrive at a contradiction and the lemma is
proved.
On the other hand if the process is continued indefinitely, we obtain a nested sequence
of squares σ0 , σ1, σ2 ,……, σ n … each contained in the previous one, for which lemma is
not true. Consequently there exists a point z0 common to all the squares of the above
sequence such that z0 is the limit point of the set of points in R. Also z0 ∈ R because R is
closed. Since f (z ) is analytic at every point which lies within and on the closed contour
C therefore f (z ) is differentiable at z0 . So for ε > 0, there exists a δ > 0 such that
 f (z ) − f (z0 ) 
 − f ′ (z0 )< ε …(2)
 z − z 0 
for all z for which | z − z0 | < δ.
We can choose a positive integer N so large that the diagonal of
the square σ N is less than δ. Then all the squares σ n ( n ≥ N ) are
contained in the circular neighbourhood
| z − z0 | < δ of z0 .
Also z0 ∈σ n for n.
Thus there exists a point z i ( here z i is z0 ) within each Si for
which inequality (1) is satisfied which contradicts the
hypothesis. Thus the lemma is true.
Proof of the main theorem: The inequality (1) can be written as
f (z ) = f (z i) − z i f ′ (z i) + z f ′ (z i) + ( z − z i ) ηi (z )
…(3)
where | ηi (z )| < ε.
Since (3) also gives the value of f (z ) at any point on the boundary of Si therefore
integrating (3) around Si , we get

∫ Si f (z ) dz = { f (z i) − z i f ′ (z i)} ∫
Si
dz + f ′ (z i)∫
Si
z dz + ∫ Si ( z − z i ) ηi (z ) dz

= ∫ Si ( z − z i ) ηi (z ) dz , …(4)
 
∵ ∫ Si dz = 0 = ∫ Si z dz 

 C-59

It is clear from the adjoining diagram that the

integral around the closed curve C is equal to the
sum of the integrals around all the Si because the
line integrals along the common boundaries of every
pair of adjacent meshes cancel each other. We are
left only with the integrals along the arcs which form
parts of C.
n
∴ ∫C f (z ) dz = Σ
i =1
∫ Si f (z ) dz . …(5)
From (4) and (5), we have
n
∫C f (z ) dz = Σ
i =1
∫ Si ( z − z i ) ηi (z ) dz

  n 
or ∫ f (z ) dz  = Σ ∫ Si ( z − z i ) ηi (z ) dz 
 C   i =1 
n
≤ Σ  ∫ ( z − z i ) ηi (z ) dz 
i = 1  Si 
n
≤ Σ
i =1 ∫ | z − z i || ηi (z )|| dz |
Si
n
<ε Σ
i =1 ∫ Si | z − z i || dz |. …(6)

The boundary Si of a mesh either completely or partially coincides with the boundary of a
square. Let ai be the length of a side of that square. The point z lies on Si and z i lies either
on the boundary of Si or inside Si therefore the distance between the points z and z i
cannot be greater than the length ai √ 2 of the diagonal of that square i.e.,
| z − z i | ≤ ai √ 2

∴ ∫ Si | z − z i || dz | ≤ ai √ 2 ∫
Si
| dz |. …(7)

Now ∫ | dz |represents the length of Si. This length is 4ai if Si is a complete square and
Si

it cannot exceed (4ai + li) if Si is a partial square where li is the length of arc of C which
forms a part of Si.
Hence if Si is a square, then inequality (7) gives

∫ Si | z − z i || dz | ≤ ai √ 2 . 4 ai = 4 √ 2 ai2 ...(8)

and ∫ Si | z − z i || dz | ≤ ai √ 2 (4 ai + li ) ≤ 4 ai2 √ 2 + a li √ 2 …(9)

if Si is a partial square,
where a denotes the length of the side of the square which encloses the entire curve C
together with the squares which are used in covering C. Obviously the sum of the areas
ai2 of all these squares cannot exceed a2 .
 C-60

If l denotes the arc length of C, we have from (6), (8) and (9)
n

∫ f (z ) dz < ε Σ (4 √ 2 ai2 + √ 2 a li) ≤ ε (4 √ 2 a2 + √ 2 al )
 C  i =1

= ε K, where K is a constant.
Since ε is arbitrary therefore we have ∫C f (z ) dz = 0.

Corollary 1: Let f (z ) be analytic in a simply connected region D. Then the integral along
every rectifiable curve in D joining any two given points of D is the same i.e., it does not depend on
the curve joining the two points.
Proof: Let Γ1 and Γ2 be any two curves in the domain D joining two given points z1
and z2 of D. Let Γ be the closed curve consisting of Γ1 and −Γ2 .
Then by Cauchy’s theorem, we have

∫Γ f (z ) dz = 0 or ∫ Γ1 +(−Γ2 ) f (z ) dz = 0

or ∫ Γ1 f (z ) dz + ∫ −Γ2 f (z ) dz = 0

or ∫ Γ1 f (z ) dz − ∫ Γ2 f (z ) dz = 0

or ∫ Γ1 f (z ) dz = ∫ Γ2 f (z ) dz .

b
Note: In view of the above corollary, we may use the symbol ∫a f (z ) dz for the integral

of f (z ) along any curve joining a and b.

Corollary 2: Extension of Cauchy-Goursat’s theorem to multiply connected
regions.
Let D be a doubly connected region bounded by two simple closed curves C1 and C2 such that C2 is
contained in C1 and f (z ) is analytic in the region between these curves and continuous on C1, then

∫ C1 f (z ) dz = ∫ C2 f (z ) dz

where both C1 and C2 are traversed in the positive sense i.e., in anti-clockwise direction.
Proof: Connect the curve C2 to C1 by making a narrow
cross-cut joining a point A of C1 to a point P of C2 . Then
ABCDAPQRPA is the simply connected region in the
interior of which f (z ) is analytic and on whose boundary
f (z ) is continuous. Hence by Cauchy-Goursat’s theorem,
we have ∫ ABCDAPQRPA f (z ) dz = 0

or ∫ ABCDA f (z ) dz + ∫ AP f (z ) dz

+ ∫ PQRP f (z ) dz + ∫ PA f (z ) dz = 0
 C-61

or ∫ ABCDA f (z ) dz + ∫ PQRP f (z ) dz = 0 ,

since ∫ AP f (z ) dz = − ∫ PA f (z ) dz

or ∫ C1 f (z ) dz + ∫ −C2 f (z ) dz = 0

or ∫ C1 f (z ) dz − ∫ C2 f (z ) dz = 0

or ∫ C1 f (z ) dz = ∫ C2 f (z ) dz .

In general if C is a closed curve and C1, C2 , C3 ,……are the closed curves which lie inside
C and if f (z ) is analytic function in the region between these curves and continuous on
C, then

∫C f (z ) dz = ∫ C1 f (z ) dz + ∫ C2 f (z ) dz + ∫ C3 f (z ) dz +……

where integral along each curve is taken in positive sense.

Remark 1: Cauchy’s fundamental theorem holds under less restrictive conditions. The
following is one of the versions of this theorem.
Let f (z ) be continuous on and analytic within a rectifiable Jordan curve C. Then

∫C f (z ) dz = 0 .

Note that in the above statement, it is not necessary for f (z ) to be analytic on C. Only
the continuity of f (z ) is essential on C. However, we shall not try to prove the above
assertion.
Remark 2: Cauchy-Goursat theorem gives only sufficient conditions for ∫ f (z ) dz
C

to become zero. In certain cases ∫ f (z ) dz may vanish even if f (z ) is not an analytic

C
function in C.
dz ,
Illustration: Evaluate ∫ , where γ is defined by | z | = d, d > 0.
γ z2
Solution: Let z = de i θ , 0 ≤ θ ≤ 2 π. Then dz = die i θ dθ.
1 2π die i θ i 2π − i θ
Now ∫γ z 2
dz = ∫0 d e 2 i2θ
dθ =
d ∫0 e dθ
2π
i  e− i θ  1 1
= −  = − [e − i 2 π − 1] = − (1 − 1) = 0 .
d  i 0 d d

∴ the integral of 1 / z 2 along the circle γ is zero but 1 / z 2 is not analytic at z = 0 which is
the centre of γ.
If, however, the function f (z ) is assumed to be continuous within and on the boundary
of C, vanishing of ∫ f (z ) dz will imply that f (z ) is an analytic function in C. This is
C
Morera’s theorem which will be proved later on.
 C-62

12 Cauchy-Goursat Theorem (Second proof )

Lemma: Let f (z ) be continuous on a domain D and let C be any continuous rectifiable curve in
D. Then for every ε > 0, there exists a polygon ∆ in D with vertices on C such that
 f (z ) dz < ε.
 ∫C ∫∆
f (z ) dz −

Theorem: Let f (z ) be analytic in a simply connected domain D and let C be any closed
continuous rectifiable curve contained in D. Then

∫C f (z ) dz = 0.

dz
Illustration: If C is the circle | z − 2 | = 5, determine whether ∫C z −3
is zero.

Solution: Putting z − 2 = 5 e i θ , dz = 5 ie i θ dθ, we get

dz 2π 5 ie i θ dθ 2π 1 − i θ −1
∫C z −3
= ∫0 5e i θ − 1
= i∫
0
(1 −
5
e ) dθ

2π 1 −i θ 1 
= i∫ 1 + 5 e + 2 e − 2 iθ + .... dθ.
0 5 

[
1 − miθ 2 π
] [ ]
2π 1
Now ∫0 e −miθ dθ = − e =− e − 2 πmi
− e0
mi 0 mi
1
=− [1 − 1] = 0 , when m ≠ 0.
mi
dz 2π
∴ ∫C z −3
= i∫
0
dθ = 2 πi ≠ 0 .

1
The reason that the integral is not zero is that is not analytic at z = 3 which is an
z −3
interior point of the circle | z − 2 | = 5.

13 Cauchy-Goursat Theorem (Third proof )

Step 1. Cauchy’s theorem for a rectangle:
By a rectangle R in the complex plane, we shall
mean a set of points ( x, y) such that
a ≤ x ≤ b, c ≤ y ≤ d.
We think of the perimeter of R as a simple closed
curve consisting of four line segments whose
direction is chosen in such a manner that the area R
lies to the left of these directed segments. The
vertices thus occur in the order (a, c ), (b, c ), (b, d),
(a, d). We shall refer to this closed curve as the boundary curve or contour of R and
shall denote it by ∂R. We now prove Cauchy’s theorem for a rectangle, namely :
 C-63

If the function f (z ) is analytic on a rectangle R, then

∫ ∂R f (z ) dz = 0 ,

where ∂R denotes the boundary curve of R.

Step 2: Cauchy’s theorem for a circular disc:
If f (z ) is analytic in the open disc ∆ defined by | z − z0 | < r, then

∫C f (z ) dz = 0

for every closed curve C in ∆ .

Step 3: Cauchy’s theorem for any closed curve:
If f (z ) is analytic in a simply connected domain D and C is any closed curve, then

∫C f (z ) dz = 0 .

14 Cauchy’s Integral Formula

Let f (z ) be an analytic function in a simply connected domain D enclosed by a rectifiable Jordan
Curve C and let f (z ) be continuous on C. Then
1 f (z )
2 πi ∫ C z − z0
f (z0 ) = dz

where z0 is any point of D. (Meerut 2001; Bundelkhand 01; Purvanchal 07, 09;
Kanpur 07; Gorakhpur 07, 08; Rohilkhand 12)
Proof: We describe a circle γ defined by the equation
| z − z0 | = ρ where ρ < d (the distance of z0 from C ).
Then the function
f (z )
φ (z ) =
z − z0

is analytic in the doubly connected region bounded by C

and γ.

∴ ∫C φ (z ) dz = ∫γ φ (z ) dz

1 f (z ) 1 f (z )
or
2 πi ∫ C z − z0
dz =
2 πi ∫ γ z − z0
dz ...(1)

where C and γ are both traversed in the counter-clockwise direction (See the figure).
It is evident that the integral on the right-hand side of (1) is independent of ρ and so we
may choose ρ as small as we please.
1 f (z ) 1 f (z ) − f (z0 ) 1 f (z0 )
Now
2 πi ∫ γ z − z0
dz =
2 πi ∫ γ z − z0
dz +
2 πi ∫ γ z − z0
dz ...(2)

Writing z − z0 = ρe i θ , dz = ρie i θ dθ, we have

 C-64

f (z0 ) 2 π ρie i θ 2π
∫γ z − z0
dz = f (z0 )∫
0 ρe i θ
dθ = f (z0 )∫
0
idθ = 2 πi f (z0 ).

Hence (2) can be written as

1 f (z ) 1 f (z ) − f (z0 )
2 πi ∫ γ z − z0
dz =
2 πi ∫ γ z − z0
dz + f (z0 )

1 f (z ) 1 f (z ) − f (z0 )
2 πi ∫ z − z0 2 πi ∫ γ
or dz − f (z0 ) = dz . ...(3)
z − z0

Since f (z ) is continuous at z0 , for a given ε > 0, there exists a δ > 0 such that
| f (z ) − f (z0 )| < ε ...(4)
for all z satisfying the inequality| z − z0 | < δ. Since ρ is at our choice, we can take ρ < δ so
that the inequality (4) is satisfied for all points on γ. Hence
 1 f (z ) − f (z0 )   1 2 π f (z ) − f (z0 ) 
 ∫γ dz  = ∫ . ρie i θ dθ
 2 πi z − z0   2 πi 0 ρe i θ 
1 2π 1 2π
2 π ∫0 ∫0
≤ | f (z ) − f (z0 )| dθ < ε dθ [by (4)]
2π
1
= . 2 πε = ε.
2π
 1 f (z ) 
Thus 
 2 πi
∫γ z − z0
dz − f (z0 )< ε.

...(5)

Since ε is arbitrary and the left-hand side of (5) does not depend upon ρ, we conclude
that
1 f (z )
2 πi ∫ γ z − z0
dz − f (z0 ) = 0

1 f (z )
2 πi ∫ γ z − z0
or dz = f (z0 ). ...(6)

Finally from (1) and (6), we obtain

1 f (z )
2 πi ∫ C z − z0
f (z0 ) = dz . ...(7)

Corollary 1. Extension of Cauchy’s integral formula to multiply connected

regions.
(We shall consider the case of doubly connected region)
If f (z ) is analytic in the region D bounded by two closed curves C1 and C2 , then
1 f (z ) 1 f (z )
2 πi ∫ C1 z − z0 2 πi ∫ C2 z − z0
f (z0 ) = dz − dz

where z0 is any point of D.

Make a cross–cut AP connecting the curves C1 and C2 .
Then f (z ) is analytic in the region ABCAPQRPA.
∴ By Cauchy’s integral formula, we have
 C-65

1 f (z )
f (z0 ) =
2 πi ∫ ABCAPQRPA z − z0
dz

1  f (z ) f (z )
= ∫ ABCA dz + ∫ AP dz
2 πi  z − z0 z − z0

f (z ) f (z ) 
+ ∫ PQRP z − z0
dz + ∫
PA z − z
0
dz 

1 f (z ) 1 f (z )
2 πi ∫ C1 z − z0 ∫ C2
= dz − dz ,
2 πi z − z0

the integrals along AP and PA cancel each other.

In particular, if C1, C2 are concentric circles with centre z0
and radii ρ1, ρ2 (ρ1 > ρ2 ), then for any point z in the annulus
(ring shaped) region, we have
1 f (z ) f (z )
f (z ) = ∫
2 πi C1 z − z0
dz − ∫
C2 z − z
0
dz .

Corollary 2. Gauss’s mean value theorem:

If f (z ) is an analytic function on a domain D and if the circular region| z − z0 | ≤ ρ is contained
in D, then
1 2π
f ( z0 + ρ e i θ ) dθ.
2 π ∫0
f (z0 ) =

In other words the value of f (z ) at the point z0 equals the average of its values on the boundary of
the circle | z − z0 | = ρ.
Proof: Let γ denote the circle | z − z0 | = ρ. The parametric equation of the circle is
z − z0 = ρ e i θ or z = z0 + ρ e i θ , 0 ≤ θ ≤ 2 π

and dz = ρ i e i θ d θ.
By Cauchy’s integral formula, we have
1 f (z ) 1 2π f ( z0 + ρ e i θ )
ρ i e i θ dθ
2 πi ∫ z − z0 2 πi ∫0
f (z0 ) = dz =
ρ ei θ
1 2π
= ∫0 f (z0 + ρ e i θ ) dθ .
2π

15 Derivative of an Analytic Function

Theorem: Let f (z ) be analytic function within and on the boundary C of a simply connected
region D and let z0 be any point within C. Then
1 f (z )
2πi ∫ C ( z − z0 )2
f ′ (z0 ) = dz .

(Kanpur 2003; Purvanchal 08, 09, 10; Gorakhpur 09, 11)

 C-66

Proof: Let z0 + h be any point within D in the neighbourhood of z0 . By Cauchy’s

integral formula, we have
1 f (z )
2 πi ∫ C z − ( z0 + h )
f ( z0 + h ) = dz

1 f (z )
2 πi ∫ C z − z0
and f (z0 ) = dz .

1  1 1 
2 πi ∫ C
Now f ( z0 + h ) − f ( z0 ) = f (z )  −  dz
 z − z 0 − h z − z0 
1 h f (z )
=
2 πi ∫ C ( z − z0 ) ( z − z0 − h )
dz .

f ( z0 + h ) − f (z0 ) 1 f (z )
2 πi ∫ C (z − z0 )2
∴ − dz
h

1  1 1 
= ∫
2 πi C
f (z )  −  dz
( z − z0 − h ) ( z − z0 ) ( z − z0 ) 
2

1 h f (z )
= ∫
2 πi C ( z − z0 )2 ( z − z0 − h )
dz . …(1)

In order to prove the desired result we have to show that

the right hand side of (1) approaches to zero as h → 0. For
this purpose draw a circle γ with centre z0 and radius r
lying entirely within C.
By Cauchy-Goursat theorem for multiply connected
region, we have
1 h f (z )
2 πi ∫ C ( z − z0 )2 ( z − z0 − h )
dz

1 h f (z )
=
2 πi ∫ γ ( z − z0 )2 ( z − z0 − h )
dz . …(2)

Since h is arbitrary therefore choosing h such that the point z0 + h lies within γ and that
1
| h | < r.
2
Equation of the circle γ is | z − z0 | = r.
∴ For any point z on γ, we have
| z − ( z0 + h )| = | z − z0 − h | ≥ | z − z0 | − | h |
1 1
≥ r − r = r.
2 2
Again the function f (z ) is analytic in D therefore it is bounded in D so that there exists
a positive constant M such that
| f (z )| ≤ M .
Using these facts, we get from (1) and (2)
 C-67

 f ( z0 + h ) − f ( z0 ) 1 f (z ) 

 h
−
2 πi ∫C ( z − z0 )
dz 
2

 1 h f (z ) dz 
=
 2 π i ∫ 0 0

γ ( z − z )2 (z − z − h)

| h| | f (z )|
≤
2π ∫ γ | z − z0 | 2 | z − z0 − h |
| dz |

| h| M | h | M . 2 πr
≤
2π ∫ γ 1
| dz | =
πr3
=| h | . constant
r2( r)
2
→ 0 as h → 0.
f ( z0 + h ) − f (z0 )
lim 1 f (z )
∴
h→ 0 h
=
2 πi ∫ C ( z − z0 )2
dz .

Hence f (z ) is differentiable at z0 and

1 f (z )
f ′ (z0 ) =
2πi ∫ C ( z − z0 )2
dz .

Note: The above formula for the derivative f ′ (z0 ) can be written formally by
differentiating the integral in Cauchy’s integral formula
1 f (z )
2 πi ∫ C z − z0
f (z0 ) = dz

with respect to z0 under the integral sign.

d  f (z )  1 f (z )
Thus f ′ (z0 ) = ∫C   dz =
dz0  z − z0  2πi ∫ C (z − z0 )2
dz .

16 Higher Order Derivatives of an Analytic Function

Theorem: Let f (z ) be analytic within and on the boundary C of a simply connected region D.
If z0 is any point within C, then f (z ) possesses derivatives of all orders at z0 and all these
derivatives are analytic at z0 . Also
n! f (z )
f n (z0 ) =
2 πi ∫ C ( z − z0 )n+1
dz .

(Garhwal 2000; Kanpur 07; Rohilkhand 12; Purvanchal 12)

Proof: Proceeding as in article 15 first prove that
1 f (z )
f ′ (z0 ) =
2πi ∫ C ( z − z0 )2
dz .

Now we shall show that f ′ (z ) is analytic at z0 .

We have
f ′ ( z0 + h ) − f ′ (z0 ) 1  1 1  f (z )
2 πi ∫ C ( z − z0 − h )2 ( z − z0 )2  h
=  −  dz
h
 C-68

1 2 h ( z − z0 ) − h2 f (z )
= ∫
2 πi C ( z − z0 − h )2 ( z − z0 )2
⋅
h
dz

1
( z − z0 ) − h
2!
2 πi ∫ C ( z − z0 − h )2 ( z − z0 )2
= 2 f (z ) dz .

f ′ ( z0 + h ) − f ′ (z0 ) 2 ! f (z )
Now
h
−
2πi ∫ C ( z − z0 )3
dz

 1 
 ( z − z0 ) − h 
2! 1
2 πi ∫ C
= f (z )  2 − dz
2 2 3
( z − z0 − h ) ( z − z0 ) ( z − z0 ) 
 

 3 
 h ( z − z0 ) − h2 
2!
2 πi ∫ C
= f (z )  2 dz
3 2
 ( z − z 0 ) ( z − z 0 − h ) 
 

h  ( z − z0 ) − h
3
2!  2 
2 πi ∫ γ ( z − z0 )3 ( z − z0 − h )2
= f (z ) dz,

by Cauchy-Goursat theorem for multi-connected region.

Here γ is the circle | z − z0 | = r lying entirely within C.
 f ′ ( z0 + h ) − f ′ (z0 ) 2 ! f (z ) 
∴ 
 h
− ∫
2πi C ( z − z0 ) 3
dz 

3
| z − z0 | + | − h |
2!
≤ | h| ∫ 2 | f (z )|| dz |
2π γ | z − z | 3 | z − z − h| 2
0 0
3
r + | h|
2!  
≤
2π
| h| 2 2
M . 2 πr  ∵ ∫ γ | dz | = perimeter of γ 
3 1 
r  r
2 
The right hand side of above inequality tends to zero as h → 0.
lim f ′ ( z0 + h ) − f ′ (z0 ) 2 ! f (z )
2 πi ∫ C ( z − z0 )3
∴ = dz .
h→ 0 h

Hence f ′ (z ) is differentiable at z0 i.e., derivative of an analytic function is analytic

2! f (z )
2πi ∫ C ( z − z0 )3
and f ′ ′ (z0 ) = dz .

∴ the result is true for n = 2.

To complete the induction we have to show that the result is true for n if it is true for
n − 1.
Now suppose that the result is true for n − 1 so that we assume that
 C-69

(n − 1)! f (z )
f (n − 1) (z0 ) = ∫ dz
2πi C (z − z0 ) n
(n − 1)! f (z )
and f (n − 1) (z0 + h) = ∫ dz
2πi C (z − z0 − h) n
∴ f (n − 1) (z0 + h) − f (n − 1) (z0 )
(n − 1)!  1 1 
=
2 πi ∫ C
f (z ) 
(z − z0 − h)
n
− n  dz
(z − z0 ) 

(n − 1)! (z − z0 )n − (z − z0 − h)n 

2 πi ∫ C
= f (z )  n n 
dz. ...(3)
 (z − z0 ) (z − z0 − h) 
Now (z − z0 )n − (z − z0 − h)n = [z − z0 − (z − z0 − h)] [(z − z0 )]n − 1

+ (z − z0 )n − 2 (z − z0 − h) + ...+ {(z − z0 ) − h}n − 1]

n
= h Σ (z − z0 )n − r (z − z0 − h)r − 1.
r =1

Hence we get from (3)

f (n − 1) (z0 + h) − f (n − 1) (z0 ) n! f (z )
h
−
2πi ∫C (z − z0 )n + 1
dz

n
Σ (z − z0 )n − r + 1 (z − z0 − h)r − 1 − (z − z0 − h)n
(n − 1)! r =1
2 πi ∫ C
= f (z ) dz
(z − z0 )n + 1 (z − z0 − h)n
n
Σ [(z − z0 )n − r + 1 (z − z0 − h)r − 1 − (z − z0 − h)n]
(n − 1)! r =1
2 πi ∫ γ
= f (z ) dz
(z − z0 )n + 1 (z − z0 − h)n
where γ is the circle | z − z0 | = ρ lying entirely within C
(n − 1)! n (z − z0 )n − r + 1 − (z − z0 − h)n − r + 1
= Σ
2 πi r = 1 ∫γ f (z )
(z − z0 )n + 1 (z − z0 − h)n − r + 1
dz

n− r
h Σ (z − z0 )n − r − s (z − z0 − h)s
(n − 1)! n s =0
= Σ
2 πi r = 1 ∫γ f (z )
(z − z0 )n + 1 (z − z0 − h)n − r + 1
dz

h (n − 1)! n n f (z ) dz
r =1 ∫ γ
= Σ Σ
2 πi s =0 (z − z0 )r + s + 1 (z − z0 − h)n − r − s +1
1
As before | z − z0 − h | ≥ ρ and so
2
(n − 1) (n − 1)
f (z0 + h) − f (z0 ) n! f (z )
h
−
2πi ∫C (z − z0 )n + 1
dz

| h |(n − 1)! n n M . 2 πρ
≤ Σ Σ
2π r =1 s = 0
ρ r + s +1 1 n − r − s +1
( ρ)
2
→ 0 as h → 0.
 C-70

(n − 1) (n − 1)
lim f (z0 + h) − f (z0 ) n! f (z )
2 πi ∫ (z − z0 )n + 1
∴ = dz
h→ 0 h
n! f (z )
∫
(n)
or f (z0 ) = dz .
2 πi C (z − z0 )n + 1

Hence the formula holds for all values of n. Thus f (z ) has derivatives of all orders and
these are all analytic at z0 . The theorem is thus completely established.
Another method to show that the result is true for n if it is true for n − 1. Suppose that
the formula is true for n − 1 i. e.,
(n − 1)! f (z )
f n − 1 (z0 ) = ∫ dz .
2 πi C ( z − z0 )n
(n − 1)! f (z )
Then f n − 1 ( z0 + h ) = ∫ dz .
2 πi C ( z − z0 − h )n

f n − 1 ( z0 + h ) − f n − 1 ( z0 )
Now
h
(n − 1)!  1 1 
=
2 πih ∫ f (z )  n
− n  dz
 ( z − z0 − h ) ( z − z0 ) 
C

(n − 1)! f (z )  1 
2 πih ∫ C ( z − z0 )n  
=  n
− 1  dz
h  
 1 − z − z 
 0 
−n
(n − 1)! f (z )    
 1 − h 
2 πih ∫ C ( z − z0 )n  
= − 1  dz
z − z0  
 
(n − 1)! f (z )  nh
2 πih ∫ C ( z − z0 )n
= 1 +
 z − z0


+ (terms containing higher powers of h ) − 1  dz,

by binomial theorem
(n − 1)! f (z )  n 
=
2 πi ∫ C ( z − z )n
0

 z − z0
+ terms containing h in Nr. dz

f n − 1 ( z0 + h ) − f n − 1 (z0 ) (n − 1)!
lim n f (z )
2 πi ∫ C ( z − z0 )n +1
∴ = dz
h→ 0 h
n! f (z )
2 πi ∫ C (z − z0 )n + 1
or f n (z0 ) = dz .

Hence the result holds for all values of n.

Therefore f (z ) possesses derivatives of all orders and these are themselves all analytic
at z0 .
 C-71

17 Morera’s Theorem
This theorem is a sort of converse of Cauchy-Goursat theorem.
Theorem 1: If f (z ) be continuous in a simply connected domain D and

∫Γ f (z ) dz = 0

where Γ is any rectifiable closed Jordan curve in D, then f (z ) is analytic in D.

(Meerut 2001; Kanpur 03, 04; Gorakhpur 07, 09; Purvanchal 08, 11; Rohilkhand 12)
Proof: Suppose z is any variable point and z0 is a fixed point in the region D. Also
suppose Γ1 and Γ2 are any two continuous rectifiable curves in D joining z0 to z and Γ is
the closed continuous rectifiable curve consisting of Γ1 and −Γ2 . Then we have

∫Γ f (z ) dz = ∫ Γ1 f (z ) dz + ∫ Γ2 f (z ) dz

and ∫Γ f (z ) dz = 0 (given)

∴ ∫ Γ1 f (z ) dz = − ∫ Γ2 f (z ) dz = ∫ Γ2 f (z ) dz ,

i.e., the integral along every rectifiable curve in D joining z0 to z is the same.
Now consider a function F (z ) defined by
z
F (z ) = ∫ z0 f (w) dw. …(1)

As we have discussed above the integral (1) depends only on the end points z0 and z.
If z + h is a point in the neighbourhood of z, then we have
z+h
F (z + h) = ∫ z0 f (w) dw . …(2)

From (1) and (2), we have

z+h z
F (z + h) − F (z ) = ∫z0 f (w) dw − ∫ z0 f (w) dw

z+h z0
= ∫ z0 f (w) dw + ∫z f (w) dw

z+h
= ∫z f (w) dw. …(3)

Since the integral on the right hand side of (3) is path independent therefore it may be
taken along the straight line joining z to z + h , so that

F (z + h) − F (z ) 1 z+h f (z )
h
− f (z ) =
h ∫z f (w) dw −
h
h

1  z+h z+h 
h  ∫ z
= f (w) dw − f (z ) ∫ dw 
z 
1 z+h
=
h ∫z [ f (w) − f (z ) ] dw. …(4)
 C-72

The function f (w) is given to be continuous at z therefore for a given ε > 0 there exists a
δ > 0 such that
| f (w) − f (z )| < ε …(5)
where | w − z | < δ.
Since h is arbitrary therefore choosing | h | < δ so that every point w lying on the line
joining z to z + h satisfies (5).
From (4) and (5), we have
 F (z + h) − F (z ) − f (z )≤ 1 z + h | f (w) − f (z )|| dw |
 h  | h| ∫ z
1 z+h
< ε∫ | dw |, [ From (5)]
| h| z

1
= ε | h | = ε.
| h|

Since ε is small and positive, therefore we have

 F (z + h) − F (z ) − f (z ) = 0 or F (z + h) − F (z )
lim
= f (z ).
 h  h→ 0 h
Hence F ′ (z ) = f (z )
i.e., F (z ) is differentiable for all values of z in D. Consequently F (z ) is analytic in D.
Since the derivative of an analytic function is analytic therefore f (z ) is analytic in D.
In view of Cauchy-Goursat theorem and Morera’s theorem, we may state the following
theorem.
Theorem 1(a): Let f (z ) be continuous in a simply connected domain D and let C be any
rectifiable closed curve in D. Then necessary and sufficient condition for f (z ) to be analytic in D is
that

∫C f (z ) dz = 0 .

18 Cauchy’s Inequality
Let f (z ) be analytic in a domain D and let D contain the interior and the boundary of the circle γ
defined by | z − z0 | = ρ and if | f (z )| ≤ M on γ, then
M
| f n (z0 )| ≤ n ! n
.
r (Kanpur 2008; Gorakhpur 09, 11;

Purvanchal 10, 11; Rohilkhand 12)

n! f (z )
2 πi ∫ γ (z − z0 )n + 1
Proof: We have f n (z0 ) = dz ,

 n! f (z )  n! | f (z )|
or | f n (z0 )| =
 2 π i ∫ γ (z − z )n + 1
0
dz ≤
 2π
∫γ | z − z0 | n + 1
| dz |
 C-73

n! M
≤ ⋅
2π r n +1 ∫γ | dz | [∵| f (z )| ≤ M ]

n! M
= ⋅ 2 πr.
2π r n +1
n M
Hence |f (z0 )| ≤ n ! ⋅ n
⋅
r

z 2 + 5z + 6
Example 6: If f (z ) = , does Cauchy’s theorem apply
z −2

(i) when the path of integration C is a circle of radius 3 with origin as centre.
(ii) when C is a circle of radius 1 with origin as centre.
z 2 + 5z + 6
Solution: Here f (z ) = ⋅
z −2
Obviously f (z ) is not analytic at z = 2.
(i) When the path of integration is the circle| z | = 3, the point z = 2 lies inside C so f (z )
is not analytic within C therefore Cauchy’s theorem is not applicable i.e.,
z 2 + 5z + 6
∫C z −2
dz ≠ 0.

(ii) When C is the circle| z | = 1, the point z = 2 lies outside C as a result f (z ) is analytic
within and on C. Hence Cauchy’s theorem is applicable i.e.,
z 2 + 5z + 6
∫C z −2
dz = 0.

Example 7: Verify Cauchy’s theorem for the function z 3 − iz 2 − 5 z + 2 i if C is the circle

| z − 1| = 2. (Garhwal 2010)
3 2
Solution: We have f (z ) = z − iz − 5 z + 2 i.

Since f (z ) is a polynomial in z therefore it is analytic within C.

On C we can choose z − 1 = 2 e iθ , 0 ≤ θ ≤ 2 π

or z = 1 + 2e i θ . ∴ dz = 2 ie i θ dθ.
2π
Now ∫C f (z ) dz = ∫0 [(1 + 2 e i θ )3 − i (1 + 2 e i θ )2 − 5 (1 + 2 e i θ ) + 2 i ] 2 ie i θ dθ

2π
= 2i ∫ [8 e4 i θ + 4 (3 − i) e3 i θ − 4 (1 + i) e2 i θ + (− 4 + i) e i θ ] dθ
0
2π
= 0, since ∫ e i k θ = 0 if k ≠ 0.
0

This verifies Cauchy’s theorem for the function f (z ) and contour C.

 C-74

z −3
Example 8: Evaluate ∫ 2
dz where C is circle
C z + 2z + 5
(a) |z|=1 and (b) | z + 1 − i| = 2 .
Solution: (a) We have
z2 + 2 z + 5 = z2 + 2 z + 1 + 4

= (z + 1)2 − (2 i)2 = (z + 1 + 2 i) (z + 1 − 2 i).

z −3 A B
Let = + ⋅
(z + 1 + 2 i) (z + 1 − 2 i) z + 1 + 2 i z + 1 − 2 i

∴ z − 3 = A (z + 1 − 2 i) + B (z + 1 + 2 i)
Putting z = − 1 + 2 i, we get
− 1 + 2 i − 3 = A (0 ) + B (− 1 + 2 i + 1 + 2 i)
i. e., − 4 + 2 i = 4 iB.
2i − 4 1
∴ B= = + i.
4i 2
Putting z = − 1 − 2 i, we get
− 1 − 2 i − 3 = A (− 1 − 2 i + 1 − 2 i) or − 4 − 2 i = − 4 i A.
4 + 2i 1
∴ A= = − i.
4i 2
1 1
z −3 −i +i
∴ = 2 + 2 ⋅
z 2 + 2z + 5 z + 1 + 2i z + 1 − 2i
z −3 1  1 1  i
∴ ∫C 2
z + 2z + 5
dz =  − i
2  ∫C z + 1 + 2i
dz +  + i
2  ∫C z + 1 − 2i
dz .

1
f (z ) = is analytic within and on the circle| z | = 1, as z = − 1 − 2 i lies outside the
z + 1 − 2i
circle | z | = 1.
1
∴ ∫C z + 1 + 2i
dz = 0, (by Cauchy’s integral theorem).

1
Similarly f (z ) = is analytic within and on the circle | z | = 1 as
z + 1 − 2i

z = − 1 + 2 i lies outside the circle | z | = 1.

1
∴ ∫ C z + 1 − 2i dz = 0 (by Cauchy’s integral theorem).
z −3
∴ ∫C 2
z + 2z + 5
dz = 0 .

(b) | z + 1 − i| = 2 is the circle with centre − 1 + i and radius 2 .

The point − 1 − 2i lies outside the circle| z + 1 − i| = 2 and the point − 1 + 2i lies inside the
circle | z + 1 − i| = 2 .
 C-75

1
∴ ∫C z + 1 + 2i
dz = 0, (by Cauchy’s integral theorem)

1
and ∫C z + 1 − 2i
dz = 2 πi (1), since f (z ) = 1 and f (− 1 + 2 i) = 1 = 2 πi.

z −3 1  1 
∴ ∫C 2
z + 2z + 5
dz =  − i 0 +  + i 2 πi = πi − 2 π = π (− 2 + i ).
 2  2 

Comprehensive Exercise 2

z2 − 4
1. Evaluate ∫C z (z 2 + 9)
dz , where C is the circle | z | = 1.

dz
2. Evaluate by Cauchy’s integral formula ∫ , where C is | z + 3 i| = 1.
C z (z + πi)
sin6 z
3. Find the value of ∫ dz if C is the circle | z | = 1.
C (z − π / 6)3 (Rohilkhand 2007)
2z
e
4. Evaluate ∫C (z + 1)4
dz , where the path of integration C is | z | = 3.

5. Using Cauchy integral formula, calculate the following integrals :

z dz
(i) ∫ , where C is the circle| z | = 2 described in positive sense.
C (9 − z 2 ) (z + i )

cosh (πz ) dz
(ii) ∫C z (z 2 + 1)
, where C is circle | z | = 2.

e a z dz
(iii) ∫C (z − πi)
, where C is the ellipse | z − 2 | + | z + 2 | = 6.

dz
(iv) ∫C z −2
, where C is | z | = 3.
(Kanpur 2003)

A nswers 2
8 πi 21 8 πi
1. − 2. 0 3. πi 4.
9 16 3 e2
π
5. (i) , (ii) 4πi , (iii) 0 , (iv) 2πi
5

19 Indefinite Integrals
Definition: Suppose f (z ) is a single-valued analytic function in a simply connected region D, then
a function F (z ) is called indefinite integral or primitive or anti-derivative of f (z ) if F (z ) is
single-valued and analytic in D and F ′ (z ) = f (z ) , z ∈ D.
 C-76

Theorem 1:A necessary and sufficient condition for the indefinite integral of a function f (z ) to
exist in a simply connected domain D is that the function f (z ) is analytic in D.
Also show that any two indefinite integrals of a function differ by a constant.
Proof: Condition is necessary: Let F (z ) be indefinite integral of f (z ). Then we
have F ′ (z ) = f (z ).
Therefore F (z ) is differentiable at every point z ∈ D. Consequently F (z ) is analytic in
D. Since the derivative of an analytic function is analytic therefore f (z ) is analytic in
D.
Condition is sufficient: Suppose f (z ) is analytic function in D. Take z0 a fixed
point and z any variable point in D. Then the integral of f (z ) along any path joining z0
to z is the same.
Consider a function F (z ) defined by
z
F (z ) = ∫ z0 f (z ) dz . …(1)

Now proceed as in Morera’s theorem and prove that F ′ (z ) = f (z ). Hence F (z ) given

by (1) is the indefinite integral of f (z ). Thus every analytic function possesses
indefinite integral in a simply connected domain.
Now we shall show that two indefinite integrals of a function differ by a constant.
Suppose φ (z ) and ψ (z ) are two indefinite integrals of f (z ).
Then φ′ (z ) = f (z ) and ψ ′ (z ) = f (z )
so that φ ′ (z ) = ψ ′ (z ) or φ′ (z ) − ψ′ (z ) = 0
or [φ (z ) − ψ (z )]′ = 0 or φ (z ) − ψ (z ) = c, where c is any constant.
∴ φ (z ) = ψ (z ) + c.
Hence the general indefinite integral of an analytic function f (z ) is given by
z
F (z ) + c , where F (z ) = ∫ z0 f (ζ) dζ.

Fundamental Theorem Of Integral Calculus For Complex Functions:

Theorem 2:If f (z ) is a single valued analytic function in a simply connected domain D, then for
a, b ∈ D we have
b
∫a f (z ) dz = F (b) − F (a) , where F (z ) is any indefinite integral of f (z ).

Proof: We have F (z ) as an indefinite integral of f (z ), therefore

F ′ (z ) = f (z )
z
and F (z ) is given by F (z ) = ∫ z0 f (z ) dz , …(1)

where z0 is any fixed point and z is any variable point in D.

b a
Now F (b) = ∫ z0 f (z ) dz and F (a) = ∫ z0 f (z ) dz

b a
∴ F (b) − F (a) = ∫ z0 f (z ) dz − ∫ z0 f (z ) dz
 C-77

b z0 b
= ∫ z0 f (z ) dz + ∫a f (z ) dz = ∫a f (z ) dz . Proved.

Since we have f (z ) = F ′ (z ) therefore we can also write

b
F (b) − F (a) = ∫a F ′ (z ) dz .

20 Integral Function: (Entire Function)

If a function f (z ) is analytic in every finite region of the z–plane, it is called an integral function or
entire function.
Liouville’s theorem: If f (z ) is an integral function satisfying the inequality| f (z )| ≤ M
for all values of z , where M is a positive constant, then f (z ) is constant.
(Meerut 2000, 01; Bundelkhand 01; Avadh 07; Rohilkhand 07; Purvanchal 10)
Proof: Suppose z1 and z2 are any two points in the z–plane.
With z1 as centre and radius r draw a circle γ whose equation is
| z − z1 | = r. Take r sufficiently large so that z2 lies inside γ and
1
| z2 − z1 | < r.
2
By Cauchy’s integral formula, we have
1 f (z )
2 πi ∫ γ z − z1
f (z1) = dz

1 f (z )
and f (z2 ) = ∫
2 πi γ z − z2
dz .

1  1 1 
2 πi ∫ γ
Now f (z2 ) − f (z1) = f (z )  −  dz
 z − z 2 z − z1 
z2 − z1 f (z )
=
2πi ∫γ (z − z1) (z − z2 )
dz ⋅ …(1)

We have | z − z1 | = r,
| z − z2 | = | z − z1 + z1 − z2 | = | z − z1 − (z2 − z1)|
1 1
≥ | z − z1 | − | z2 − z1 | ≥ r − r = r
2 2
and | f (z )| ≤ M (given).
Taking modulus of both sides of (1), we have
(z − z1) f (z ) 
| f (z2 ) − f (z1)| = 2
 2 πi
∫γ dz
(z − z1) (z − z2 ) 
| z2 − z1 | | f (z )|
≤
2π ∫ γ | z − z1 || z − z2 |
| dz |

| z2 − z1 | M
≤
2π
⋅
1 ∫γ | dz |
r⋅ r
2
 C-78

| z2 − z1 | M 2 | z2 − z1 | M
≤ ⋅ 2 2 πr =
π r r
→ 0 as r → ∞.
Consequently f (z2 ) − f (z1) = 0 or f (z2 ) = f (z1).
Hence f (z ) is constant.
Alternative proof: By Cauchy’s inequality, we have
M
| f n (z0 )| ≤ n ! n where z0 is any point in the z–plane and r is the radius of the circle γ
r
defined by | z − z0 | = r. For n = 1, we have
M
| f ′ (z0 )| ≤ ⋅
r
As r → ∞, f ′ (z0 ) = 0.
Since the point z0 is arbitrary therefore we conclude that f ′ (z ) vanishes at every point
in the z-plane. Hence f (z ) is constant.

21 Expansion of Analytic Functions as Power Series

Theorem 1: Taylor’s Theorem: Let f (z ) be analytic at all points within a circle C0 with
centre z0 and radius r0 . Then for every point z within C0 , we have
f ′ ′ (z0 )
f (z ) = f (z0 ) + f ′ (z0 ) (z − z0 ) + (z − z0 )2 + ……
2!
f n(z0 )
+ (z − z0 )n + ……
n !
∞ (z − z )n
= f (z0 ) + Σ 0
f n(z0 ) ⋅
n =1 n!
(Kanpur 2001; Purvanchal 07; Gorakhpur 10; Rohilkhand 09, 12; Garhwal 10)

Proof: Consider a circle C0 with centre z0 and radius r0 .

Suppose z is any point inside the circle| z − z0 | = r. Draw a
circle C with centre z0 and radius R such that r < R < r0 so
that the point z lies inside C. If w is any point on C, equation
of C is given by | w − z0 | = R.
Also by Cauchy’s integral formula, we have
1 f (w)
f (z ) =
2πi ∫ C w−z
dw. …(1)

Consider the identity

1 1 1
= =
w − z (w − z0 ) − (z − z0 )  z − z0 
(w − z0 ) 1 − 
 w − z0 
−1
1  z − z0 
= 1 − 
w − z0  w − z0 
 C-79

 2 n −1
1 z − z0  z − z0   z − z0 
= 1 + +  + …… +  
w − z0  w − z0  w − z0   w − z0 
 
 z − z0 
n
1 
+  
 w − z0  1 − − z0
z 
w − z0 
1 z − z0 (z − z0 )2 (z − z0 )n − 1
= + + + …… +
w − z0 (w − z0 )2 (w − z0 )3 (w − z0 )n
(z − z0 )n
+ ⋅
(w − z0 )n (w − z )
f (w)
Multiplying both sides by and integrating around C, we get
2πi
1 f (w) 1 f (w) (z − z0 ) f (w)
2 πi ∫ C ∫ ∫
dw = dw + dw
w−z 2 πi C w − z0 2 πi C (w − z )2
0

(z − z0 )n−1 f (w) (z − z0 )n f (w)

+ …… +
2πi ∫C (w − z0 )n
dw +
2πi ∫ C (w − z ) (w − z0 )n
dw

f ′ ′ (z0 )
or f (z ) = f (z0 ) + (z − z0 ) f ′ (z0 ) + (z − z0 )2 + ……
2!
f n − 1(z0 )
+ (z − z0 )n − 1 + Sn …(2)
(n − 1)!
(z − z0 )n f (w)
where Sn =
2πi ∫C (w − z ) (w − z0 )n
dw.

In order to get the desired result we have to show that

Sn → 0 as n → ∞.
We have | z − z0 | = r , | w − z0 | = R.
∴ | w − z | = | w − z0 − (z − z0 )| ≥ | w − z0 | − | z − z0 | = R − r.
If M denotes the greatest value of f (w) on C, we have
 (z − z0 )n f (w) 
| S n | =
 2πi
∫C (w − z ) (w − z0 )n
dw

| z − z0 |n | f (w)|
≤
2π ∫C | w − z || w − z0 |n
| dw |

rn M
≤ ⋅
2π ( R − r) R n ∫C | dw |

n
M⋅R r  
=
( R − r)
 
 R ∵ ∫ C | dw | = 2πR 
Since r < R therefore (r / R)n → 0 as n → ∞. Consequently Sn → 0 as n → ∞. Hence as
n → ∞, the limit of the sum of the first n terms on the right hand side of (2) is f (z ), so we
can represent f (z ) by the infinite series
 C-80

f ′ ′ (z0 )
f (z ) = f (z0 ) + (z − z0 ) f ′ (z0 ) + (z − z0 )2 + ……
2!
f n(z0 )
+ (z − z0 )n + ……
n!
∞ f n (z0 )
= f (z0 ) + Σ (z − z0 )n ⋅
n =1 n!
It is known as Taylor’s series.
If we put z0 = 0 in the above series, we get
∞ zn n
f (z ) = f (0 ) + Σ f (0 ),
n =1 n!
which is known as Maclaurin’s series.
Remark: For the validity of the expansion as a Taylor’s series, it is essential that f ( y)
be analytic at all points inside the circle C0 for then the convergence of Taylor’s series
for f (z ) is assured.
Hence the greatest radius of C0 is the distance from the point z0 to the singularity of
f (z ) which is nearest to z0 , since we require the function to be analytic at all points
within C0 .
Theorem 2: Laurent’s Theorem: Let f (z ) be analytic in the ring shaped region D
bounded by two concentric circles C1 and C2 with centre z0 and radii r1 and r2 (r1 > r2 ) and let z be
any point of D. Then
∞ ∞
f (z ) = Σ an (z − z0 )n + Σ bn (z − z0 )− n
n=0 n =1

1 f (w)
2 πi ∫ C1 (w − z0 )n +1
where an = dw ,

1
(w − z0 )n −1 f (w) dw , n = 1, 2, 3, ……
2 πi ∫ C2
and bn =

(Gorakhpur 2007, 13; Rohilkhand 11; Purvanchal 12)

Proof: The function f (z ) is given to be analytic in the ring
shaped region D bounded by concentric circles C1 and C2
with centre z0 and radii r1 and r2 (r1 > r2 ). Let z be any point in
the region D. Then by Cauchy’s integral formula for doubly
connected region, we have
1  f (w) f (w) 
2 πi ∫ C1 w − z ∫ C2
f (z ) =  dw − dw ⋅
w−z 
…(1)
For any point w on C1, we have the identity

1 1 1
= =
w − z w − z0 − (z − z0 )  z − z0 
(w − z0 ) 1 − 
 w − z0 
 C-81

n −1
1 
2
z − z0  z − z0   z − z0 
= 1+ +  + …… +  
w − z0  w − z0  w − z0   w − z0 


 z − z0 
n
1 
+  
 w − z 0  1−
z − z0 
w − z0 
1 z − z0 (z − z0 )2 (z − z0 )n − 1 (z − z0 )n
= + + + … + + ⋅
w − z0 (w − z0 )2 (w − z0 )3 (w − z0 )n (w − z0 )n (w − z )
Multiplying both sides by f (w) / 2πi and integrating around C1, we get
1 f (w) 1 f (w) (z − z0 ) f (w)
2 πi ∫ C1 w − z 2 πi ∫ C1 w − z0 2 πi ∫ C1 (w − z0 )2
dw = dw + dw + …

(z − z0 )n − 1 f (w)
…+
2 πi ∫ C1 (w − z0 )n
dw + Rn

(z − z0 )n f (w)
2 πi ∫ C1 (w − z0 )n (w − z )
where Rn = dw.

1 f (w)
2 πi ∫ C1 (w − z0 )n + 1
Putting an = dw in the above relation, we get

1 f (w)
2 πi ∫ C1 w − z
dw = a0 + a1 (z − z0 ) + a2 (z − z0 )2 + ……

+ an − 1 (z − z0 )n − 1 + Rn. …(2)
Now we shall show that Rn → 0 as n → ∞.
Suppose | z − z0 | = r so that r2 < r < r1.
We have | w − z0 | = r1.
∴ | w − z | = | w − z0 − (z − z0 )| ≥ | w − z0 | − | z − z0 | = r1 − r.
 (z − z0 )n f (w) 
Now | Rn | =
 2 π i ∫ C1 (w − z )n (w − z )
0
dw

| z − z0 |n | f (w)|
≤
2π ∫ C1 | w − z |n | w − z |
0
| dw |

r n . M1
≤
2 π r1n (r1 − r) ∫ C1 | dw |
where M1 is the greatest value of f (w) on C1
n
n
r . M1 M1 r1  r 
= 2 πr1 =   ⋅
2 π r1n (r1 − r) r1 − r  r1 

Since r < r1, therefore (r / r1)n → 0 as n → ∞. Consequently Rn → 0 as n → ∞.

∴ from (2), we have

1 f (w) ∞
∫
2 πi C1 w − z
dw = Σ an (z − z0 )n.
n=0
…(3)
 C-82

Again for the second integral in (1), consider the identity

1 1
− = , where w is any point on C2
w − z (z − z0 ) − (w − z0 )
1
=
 w − z0 
(z − z0 ) 1 − 
 z − z0 
n −1
1 
2
w − z0  w − z0   w − z0 
= 1+ +  + …… +  
z − z0  z − z0  z − z0   z − z0 


 w − z0 
n
1 
+  
 z − z 0  1−
w − z0 
z − z0 
1 w − z0 (w − z0 )2
= + + + ……
z − z0 (z − z0 )2 (z − z0 )3
(w − z0 )n − 1 (w − z0 )n
+ + ⋅
(z − z0 )n (z − z0 )n (z − w)
1 f (w) 1
2 πi ∫ C2 2 πi (z − z0 ) ∫ C2
∴ − dw = f (w) dw
w−z
1
+
2 πi (z − z0 )2 ∫ C2 (w − z0 ) f (w) dw + ……

1
+ n ∫ C2 (w − z0 )n −1 f (w) dw + Pn ,
2 πi (z − z0 )
1 (w − z0 )n f (w)
2 πi (z − z0 )n ∫ C2
where Pn = dw.
z−w
1
(w − z0 )n − 1 f (w) dw in the above relation, we get
2 πi ∫ C2
Putting bn =

1 f (w)
dw = b1 (z − z0 )−1 + b2 (z − z0 )−2 + ……
2 πi ∫ C2 w − z
−
+ bn (z − z0 )− n + Pn . …(4)
Now we have to show that Pn → 0 as n → ∞. We have| z − z0 | = r, | w − z0| = r2 for C2
where r2 < r.
∴ | z − w | = |(z − z0 ) − (w − z0 )| ≥ | z − z0 | − | w − z0 | = r − r2 .
1 | w − z0 |n | f (w)|
Now | Pn | ≤
2 π | z − z0 | n ∫ C2 | z − w|
| dw |

1 r2 n M2
≤
2π r n
⋅
(r − r2 ) ∫ C2 | dw |,

where M2 is the greatest value of f (w) on C2

n
M2 r2  r2 
=   ⋅
r − r2  r 
 C-83

∴ Pn → 0 as n → ∞, since r2 < r.
Thus from (4), we have
1 f (w) ∞
− ∫ dw = Σ bn (z − z0 )− n. …(5)
2 πi 2 w − z
C n =1

Substituting the values from (3) and (5) in (1), we get

∞ ∞
f (z ) = Σ an (z − z0 )n + Σ bn (z − z0 )− n,
n=0 n =1

1 f (w) 1
(w − z0 )n − 1 f (w) dw.
2 πi ∫ C1 (w − z0 )n + 1 2 πi ∫ C2
where an = dw and bn =

Remark: We have bn = a− n . Therefore if C is any circle of radius r and centre z0 such

that r2 < r < r1 then since the integrand is analytic in r2 < | w − z0 | < r1, we can write
1 f (w)
2 πi ∫ C (w − z0 )n + 1
an = dw

1
(w − z0 )n − 1 f (w) dw.
2 πi ∫ C
and bn = a− n =

In this case the resulting series becomes

∞ 1 f (w)
2 πi ∫ C (w − z0 )n + 1
f (z ) = Σ an (z − z0 )n where an = dw.
n= − ∞

Theorem 3: Uniqueness Theorem: Suppose that we have obtained in any manner or as

the definition of f (z), the formula
∞
f (z ) = Σ Pn (z − z0 )n , (r2 < | z − z0 | < r1)
n= − ∞

then the series is necessarily identical with Laurent’s series of f (z ).

Proof: Suppose C is the circle defined by | z − z0 | = r, where r2 < r < r1. Then the
Laurent’s series is given by
∞
f (z ) = Σ an (z − z0 )n
n= − ∞

1 f (w)
2 πi ∫ C (w − z0 )n + 1
where an = dw

1 1 ∞

2 πi ∫ C (w − z0 )n + 1
= Σ Pm (w − z0 )m dw
m=−∞

1 ∞ (w − z0 )m
=
2 πi
Σ
m=−∞
Pm ∫C (w − z0 )n + 1
dw

[term by term integration is possible since the series is

uniformly convergent on every closed subset of the annulus]
1 ∞ 2π r m
e imθ
= Σ Pm ∫0 n +1
r i e i θ dθ,
2 πi m=−∞ r e i (n + 1) θ
putting w − z0 = r e i θ
 C-84

1 ∞ 2π
= Σ Pm ∫ r m − n e i (m − n) θ dθ . …(1)
2π m=−∞ 0

When m ≠ n, we have
2π
2π i (m − n) θ  e i (m − n) θ  1
∫0 e dθ =   = [e i (m − n) 2 π − e0 ]
 i (m − n)  0
i (m − n)

1
= (1 − 1) = 0
i (m − n)
and when m = n, we have
2π 2π
∫0 e i (m − n) θ dθ = ∫0 dθ = 2 π.

1
∴ We have an = ⋅ Pn . 2 π, from (1)
2π
= Pn .
Hence the given series is identical with the Laurent’s series of f (z ).

1
Example 9: Expand f (z ) = in a Laurent’s series valid for the regions
(z + 1)(z + 3)
(i) | z | < 1 (ii) 1 < | z | < 3
(iii) | z | > 3 (iv) 0 < | z + 1| < 2.
(Purvanchal 2007, 09, 12; Gorakhpur 15)
1
Solution: We have f (z ) = ⋅
(z + 1)(z + 3)
Resolving into partial fractions, we get
1 1
f (z ) = − ⋅
2 (z + 1) 2 (z + 3)
(i) | z |< 1 .
−1
1 1 z
We have f (z ) = (1 + z )−1 − 1 + 
2 6 3
1 
2 3
1 z  z   z 
= [1 − z + z 2 − z 3 + ……] − 1 − +   −   + ……
2 6  3  3   3 
1 1 1 1 1 1 2
=( − )−( − )z +( − ) z − ……
2 6 2 18 2 54
1 4 13 2
= − z+ z − ……
3 9 27
(ii) 1 < | z | < 3
1 |z|
Then we have < 1 and < 1.
|z| 3
 C-85

−1
1 1 1  1
Now = = 1 + 
2 (z + 1)  1 2z  z
2 z 1 + 
 z
1  1 1 1  1 1 1
= 1 − + 2 − 3 + …… = − + − ……
2z  z z z  2z 2z 2
2 z3
−1
1 1 1  z 1 z z2 z3 
and = = 1 +  = 1 − + − + ……
2 (z + 3)  z 6  3 6 3 9 27 
6 1 + 
 3
1 1 1 2 1 3
= − z+ z − z + …… .
6 18 54 162
Thus the Laurent’s series valid for the region 1 < | z | < 3 is
1 1 1 1 1 1 2 1 3
f (z ) = …… + 3
− 2
+ + − z+ z − z + ……
2z 2z 2 z 6 18 54 162

(iii) | z | > 3 .Then (3 /| z |) < 1.

−1 −1
1  1 1  3
∴ f (z ) = 1 +  − 1 + 
2z  z 2z  z
1  3  3 
2 3
1  1 1 1   3
= 1 − + 2 − 3 + …… − 1 − +   −   + ……
2z      
z z z 2 z  z z z 
1 4 13 40
= 2
− 3
+ 4
− + ……
z z z z5
(iv) 0 < | z + 1| < 2 .
Let z + 1 = u. Then we have 0 < | u| < 2.
−1
1 1 1  u
∴ f (z ) = = = 1 + 
(z + 1)(z + 3) u (u + 2) 2 u  2

1  u  u
2
 u
3  1 1 u u2
= 1 − +   −   + …… = − + − + ……
2u  2  2   2  2 u 4 8 16

1 1 z + 1 (z + 1)2
= − + − + ……
2 (z + 1) 4 8 16
Example 10: Obtain the Taylor’s and Laurent’s series which represent the function
z2 − 1
f (z ) = in the regions
(z + 2) (z + 3)
(i) | z |< 2 (Garhwal 2010) (ii) 2 < | z | < 3 (Gorakhpur 2009, 11, 13)
(iii) | z | > 3. (Avadh 2008; Gorakhpur 09, 11, 13)

z2 − 1 5z + 7
Solution: Let f (z ) = = 1− ⋅
(z + 2) (z + 3) (z + 2)(z + 3)
Resolving into partial fractions, we get
 C-86

3 8
f (z ) = 1 + − ⋅
z +2 z +3
(i) | z | < 2. We have
−1 −1
3 z 8 z
f (z ) = 1 + 1 +  − 1 + 
2 2 3 3
3 z z2 z3  8  z z2 z3 
= 1+ 1 − + 2 − 3 + …  − 1 − + 2 − 3 + … 
2 2 2 2  3  3 3 3 
∞ n ∞ n
3  z  8  z 
= 1+ Σ (− 1)n   − Σ (− 1)n   ⋅
2 n=0  2 3 n=0  3

2 |z|
(ii) 2 < | z | < 3. Then < 1 and < 1.
|z| 3
−1 −1
3 2 8 z
∴ f (z ) = 1 + 1 +  − 1 + 
z  z 3 3
3 2 22  8 z z2 z3 
= 1+ 1 − + 2 − …… − 1 − + 2 − 3 + ……
z  z z  3 3 3 3 
∞ n ∞ n
3  2 8  z 
= 1+ Σ (− 1)n   − Σ (− 1)n   ⋅
z n=0  z  3 n=0  3
3
(iii) | z | > 3. Then < 1.
|z|
−1 −1
3 2 8 3
∴ f (z ) = 1 + 1 +  − 1 + 
z  z z  z
3  2  2 2  2 3 
= 1+ 1 − +   −   + ……
z  z  z   z  

8  3  3 2  3 3 
− 1 − +   −   + ……
3  z  z   z  
∞ n ∞ n
3  2 8  3
= 1+ Σ (− 1)n   − Σ (− 1)n   .
z n=0  z  3 n=0  z

1
Example 11: Find different developments of in powers of z according to the
(z − 1) (z − 3)
position of the point in the z-plane. Expand the function in Taylor’s series about z = 2 and indicate
the circle of convergence.
1
Solution: Let f (z ) = ⋅
(z − 1) (z − 3)
Resolving into partial fractions, we get
1 1
f (z ) = − + ⋅
2 (z − 1) 2 (z − 3)
Obviously f (z ) is regular everywhere except at z = 1 and 3.
 C-87

(i) 0 < | z | < 1.

−1 ∞ ∞ n
1 1 z 1 1  z 
f (z ) = (1 − z )−1 − 1 −  = Σ zn − Σ  
2 6 3 2 n=0 6 n=0  3
∞ 1 1  n
= Σ 1 − n + 1  z
n=0 2 3 

which is Taylor’s expansion of f (z ) in 0 < | z | < 1.

(ii) 1 < | z | < 3.
−1 −1 ∞ ∞ n
1  1 1 z 1 1 1  z 
f (z ) = − 1 −  − 1 −  =− Σ − Σ  
2z  z 6  3 2z n=0 zn 6 n=0  3

which is Laurent’s series in the positive and negative powers of z in the region1 < | z | < 3.
(iii) 1 z |> 3.
−1 −1
1  1 1  3
f (z ) = − 1 −  + 1 − 
2z  z 2z  z
∞ ∞ n ∞
1 1 1  3 1 1
=− Σ + Σ   = Σ (3 n − 1)
2z n=0 z n 2z n=0  z  2 n=0 z n +1

which is a Laurent’s series in the negative powers of z for | z | > 3.

Consider a circle with centre at z = 2. Then the distance of both the singularities z = 1
and z = 3 from the centre of the circle is 1. Hence if we draw the circle| z − 2 | = 1then the
function f (z ) is regular within this circle so that f (z ) can be expanded in a Taylor’s
series within this circle i. e., in the region| z − 2 | < 1. Consequently| z − 2 | = 1is the circle
of convergence.
1 1 1
Now f (z ) = = =
(z − 1)(z − 3) z 2 − 4 z + 3 (z − 2)2 − 1
∞
= − [1 − (z − 2)2 ] −1 = − Σ (z − 2)2 n,
n=0

which is a Taylor’s expansion of f (z ) about z = 2.

Example 12:Expand log (1 + z ) in a Taylor’s series about z = 0 and determine the region of
convergence for the resulting series.
Solution: Let f (z ) = log (1 + z ).
Taylor’s expansion for f (z ) about z = 0 is given by
z2
f (z ) = f (0 ) + z f ′ (0 ) + f ′ ′ (0 ) + …… . …(1)
2!
1 1
We have f ′ (z ) = , f ′ ′ (z ) = − ,
1+ z (1 + z )2
2 (n − 1)!
f ′ ′ ′ (z ) = 2 3
,……, f n
(z ) = (− 1)n − 1 ⋅
(1 + z ) (1 + z )n

∴ f (0 ) = 0 , f ′ (0 ) = 1, f ′ ′ (0 ) = − 1, f ′ ′ ′ (0 ) = 2, ……,
 C-88

f n
(0 ) = (−1)n − 1 (n − 1)!.
Substituting these values in (1), we get
1 2 2 3 (−1)n − 1 (n − 1)! z n
log (1 + z ) = z − z + z −…… + + ……
2! 3! n!
z2 z3 zn
+ − …… + (− 1)n − 1
=z− + ……
2 3 n
Let un be the nth term of the series. Then we have
zn (− 1)n z n + 1
un = (− 1)n − 1 , un + 1 =
n n +1
lim  un  lim  n + 1 1
∴  =  = ⋅
n→ ∞ u n → ∞  nz  | z |
 n +1 
Hence by D ′ Alembert’s ratio test the series converges for | z | < 1.
z
Example 13: Find the Laurent expansion of about the singularity z = − 2.
(z + 1) (z + 2)
Specify the region of convergence.
z 2 1
Solution: We have f (z ) = = −
(z + 1) (z + 2) z + 2 z + 1
2 1
or f (z ) = − ⋅ ...(1)
(z + 2) z + 1
1
To find Laurent expansion for φ (z ) = about z = − 2, we write
z +1
∞
φ (z ) = Σ an (z + 2)n ...(2)
n=0
(n)
φ (− 2) (− 1)n n !
where an = . But φ(n) (z ) = ⋅
n! (z + 1)n + 1
φ(n) (− 2) (− 1)n (− 1)n φ(n) (− 2)
∴ = n + 1
= n + 1
= − 1 or an = = −1
n! (− 2 + 1) (− 1) n!
Putting this in (2), we get
1 ∞ 1 ∞
= φ (z ) = Σ (− 1) (z + 2)n or − = Σ (z + 2)n.
z +1 n=0 z +1 n=0

2 ∞
Now (1) reduces to f (z ) = + Σ (z + 2)n. ...(3)
2+z n=0

This is the required expansion.

∞
Second Part: Let Σ (z + 2)n = Σun.
n=0

un + 1 (z + 2)n + 1

Then   =  = | z + 2 |.
n
 un   (z + 2) 
un + 1
Series will be convergent if < 1 i.e., if | z + 2 | < 1.
 un 
 C-89

∴ Radius of convergence = 1.
Series is convergent V z inside the circle whose centre is z = − 2 and radius = 1.
z3 z5
Example 14: Prove that tan−1 z = z − + − …… when | z | < 1.
3 5
Solution: Let f (z ) = tan−1 z .

By Taylor’s theorem, we have

(z − z0 )2
f (z ) = f (z0 ) + f ′ (z0 ) (z − z0 ) + f ′ ′ (z0 ) + ……
2!
Taking z0 = 0 , we get
z2
f (z ) = f (0 ) + f ′ (0 ) z + f ′ ′ (0 ) + …… . …(1)
2!
We have f (z ) = tan−1 z so that f (0 ) = tan−1 0 = 0 .
1 2z
f ′ (z ) = 2
, f ′ ′ (z ) = − ,
1+ z (1 + z 2 )2

2 (1 − 3 z 2 ) 24 (− z + z 3 )
f ′′′ z = − , f iv (z ) = − ,
(1 + z 2 )3 (1 + z 2 )4
24 (− 1 + 10 z 2 − 5 z 4 )
f v (z ) = − and so on.
(1 + z 2 )5

∴ f ′ (0 ) = 1, f ′ ′ (0 ) = 0 , f ′ ′ ′ (0 ) = − 2, f iv (0 ) = 0 , f v (0 ) = 24 etc.
Substituting all these values in relation (1), we get
1 1
tan−1 z = z − z 3 + z 5 − …… .
3 5
∞  1
 1
Example 15: Prove that cosh  z +  = a0 + Σ an  z n + n 
 z n =1  z 
1 2π
where an =
2π ∫0 cosh nθ cosh (2 cos θ) dθ.
(Gorakhpur 2007, 09; Rohilkhand 12; Purvanchal 08)
 1
Solution: The function cosh  z +  is analytic in every finite part of the z-plane except
 z
at z = 0. Thus the given function is analytic in the annulus r ≤ | z | ≤ R where r is small
and R is large so that we can expand f (z ) in a Laurent’s series in the annulus r < | z | < R.
∞ ∞
 1
∴ cosh  z +  = Σ an z n + Σ bn z − n,
 z n=0 n =1

1  1 dz 1  1
where an = ∫ cosh  z +  n + 1 and bn = ∫ cosh  z +  z n − 1 dz ,
2 πi C  
z z 2 πi C  z
C is a circle with centre at origin.
Let C be the unit circle defined by | z | = 1. Then z = e iθ , dz = ie i θ dθ.
 C-90

1 2π ie iθ
∴ an = ∫ cosh (e i θ + e − i θ ) i (n + 1) θ dθ
2 πi 0 e
1 2π −i n θ
2 π ∫0
= cosh (2 cos θ) e dθ

1 2π
2 π ∫0
= cosh (2 cos θ) (cos nθ − i sin nθ) dθ

1 2π
2 π ∫0
= cosh (2 cos θ) cos nθ dθ,

{since the other integral becomes zero by the property

2a
∫0 f (θ) dθ = 0 if f (2 a − θ) = − f (θ)}
 1
Now the function cosh  z +  remains unchanged by replacing z by 1 / z, therefore we
 z
have
1 2π
2 π ∫0
bn = a− n = cosh (2 cos θ) cos (− nθ) dθ

1 2π
2 π ∫0
= cosh (2 cos θ) cos nθ dθ = an .

∞ ∞
 1
Hence cosh  z +  = Σ an z n + Σ bn z − n
 z  n=0 n =1
∞ ∞
= Σ an z n + Σ an z − n [∵ bn = an]
n=0 n =1
∞
= a0 + Σ an (z n + z − n),
n =1

1 2π
where an =
2π ∫0 cosh (2 cos θ) cos nθ dθ.

1
c (z − 1 / z) ∞
Example 16: Show that e 2 = Σ an z n ,
n= − ∞

1 2π
where an =
2π ∫0 cos (nθ − c sin θ) dθ .
(Meerut 2001, 02; Gorakhpur 2004, 06, 08, 11; Avadh 07)
Solution: The given function is analytic at every point in the z-plane except at z = 0 so
it is analytic in the annulus r < | z | < R where r is small and R is large. Therefore it can be
expanded in a Laurent’s series in the region r < | z | < R.
1
c (z − 1 / z) ∞ ∞
∴ e 2 = Σ an z n + Σ bn z − n
n=0 n =1

1 dz 1
f (z ) z n − 1 dz
2 πi ∫ C 2 πi ∫ C
where an = f (z ) n + 1 , bn =
z
and C is any circle with centre at origin.
Let C be the unit circle defined by | z | = 1.
Then z = e i θ , dz = i e i θ dθ.
 C-91

1
1 2π c (e i θ − e − i θ ) ie iθ
2 πi ∫0
Now an = e2 dθ
e i (n + 1) θ
1 2π 1 2 π − i (n θ − c sin θ)
e i c sin θ e − i n θ
2 π ∫0 2 π ∫0
= dθ = e dθ

1 2π
2 π ∫0
= {cos (nθ − c sin θ) − i sin (nθ − c sin θ)} dθ

1 2π
2 π ∫0
= cos (nθ − c sin θ) dθ, …(1)

the second integral becomes zero by the property of definite integrals.

Since the given function remains unchanged if z is replaced by (− 1 / z ) therefore
bn = (− 1)n an.
1
c (z − 1 / z) ∞ ∞
Hence e 2 = Σ an z n + Σ bn z − n
n=0 n =1
∞ ∞
= Σ an z n + Σ (− 1)n an z − n
n=0 n =1
∞
= Σ an z n, where an is given by (1).
n= − ∞

Example 17: If the function f (z) is analytic when | z | < R and has the Taylor’s expansion
∞
Σ an z n, show that for r < R, we have
n=0
1 2π ∞
∫0 | f (r e i θ )|2 dθ = Σ | an |2 r2 n.
2π n=0 (Avadh 2007)
Hence prove that if | f (z )| ≤ M where | z | < R,
∞
Σ | an |2 r 2 n ≤ M 2 .
n=0

Solution: Here f (z ) is analytic within the circle | z | = r, (r < R) therefore it can be

expanded in a Taylor’s series within the circle | z | = r.
∞ ∞
∴ f (z ) = Σ an z n or f (r e i θ ) = Σ an r n e i n θ , where z = r e i θ .
n=0 n=0

If an is the conjugate of an , we have

∞
f (r e i θ ) = Σ am r m
e − im θ .
m =0
∞ ∞
Now | f (r e i θ )| 2 = f (r e i θ ) f (r e i θ ) = Σ an r n e inθ Σ am r m e − im θ .
n=0 m =0
2π 2π  ∞ n inθ  
∞
m − im θ 
∴ ∫0 | f (r e i θ )|2 dθ = ∫0  Σ an r e   Σ am r e  dθ.
 n=0   m =0 
…(1)
iθ iθ
Since the two series for f (r e ) and f (r e ) are absolutely convergent therefore their
product is uniformly convergent for 0 ≤ θ ≤ 2 π. Hence the term by term integration is
justified.
 C-92

Also we have
2π
∫0 e i(n − m) θ dθ = 0 if n ≠ m

= 2π if n = m.
Hence from (1), we have
2π ∞ 2π
∫0 | f (r e i θ )|2 dθ = Σ an an r 2 n ∫ dθ
n=0 0

∞
= 2π Σ | an |2 r2 n
n=0

1 2π ∞
or ∫0 | f (r e i θ )|2 dθ = Σ | an |2 r 2 n.
2π n=0

This proves the first result.

Again it is given that | f (z )| ≤ M when | z | < R.
∞ 1 2π
∴ Σ | an |2 r 2 n = ∫0 | f (z )|2 dθ [ ∵ z = r ei θ ]
n=0 2π
1 2π 1
≤
2π ∫0 M 2 dθ =
2π
M2 . 2π = M2 .

∞
Hence Σ | an |2 r2 n ≤ M 2 . Proved.
n=0

Comprehensive Exercise 3

(z − 2)(z + 2)
1. Expand for
(z + 1)(z + 4)
(i) | z | < 1 (ii) 1 < | z | < 4
(iii) | z | > 4. (Garhwal 2000; Kanpur 04)
1 5 7
2. Express f (z ) = 2 3
in a Laurent’s series in the region ≤ | z |≤ ⋅
z (z + 1) (z + 2) 4 4
1
3. Find the Laurent series of the function f (z ) = 2
about z = 0.
z (1 − z ) (Kanpur 2004)
4. Obtain the Taylor’s or Laurent’s series which represents the function
1
f (z ) = 2
(1 + z ) (z + 2)
when (i) | z | < 1, (ii) 1 < | z | < 2, (iii) | z | > 2. (Kanpur 2008)
z
5. If 0 < | z − 1| < 2, then express f (z ) =
(z − 1) (z − 3)
in a series of positive and negative powers of (z − 1). (Rohilkhand 2010)
 C-93

z −1
6. Expand f (z ) = as a Taylor’s series about
z +1
(i) z = 0 (ii) z = 1
(iii) its Laurent’s series for the domain 1< | z | < ∞. (Kanpur 2000)
1
7. Find Laurent’s series of the function f (z ) = valid in the region
(z 2 − 4) (z + 1)
1 < | z | < 2. (Kanpur 2001)
π
8. Expand sin z in a Taylor’s series about z = ⋅
4 (Kanpur 2002)
1
9. (i) Expand as a Taylor’s series about z = 1.
z
sin z
(ii) Determine Laurent’s expansion of the function f (z ) = 3
in the
 π
z − 
π  4
annulus 0 < z − < 1.
 4
∞ ∞
10. Prove that e u / z + v z = Σ an z n + Σ bn z − n,
n=0 n =1

1 2π
where an =
2π ∫0 exp {(u + v) cos θ} cos {(v − u) sin θ − nθ} dθ

1 2π
and bn =
2π ∫0 exp {(u + v) cos θ} cos {(u − v) sin θ − nθ} dθ.
(Rohilkhand 2011)
  1 
11. Show that sin  c  z +   can be expanded in a series of the type
  z 
∞ ∞
Σ an z n + Σ bn z − n,
n=0 n =1

where the coefficients of both z n

and z − n are
1 2π
2π ∫0 sin (2 c cos θ) cos nθ dθ.
(Garhwal 2000)
3 ∞
/2 z 2
12. Show that if c > 0, then e z + c = Σ an z n,
n= − ∞

e − c /2 2π 2
where an = n ∫0
θ)
e c (cos θ + cos
cos {c sin θ (1 − cos θ) − nθ} dθ.
2 πc
13. By using the integral representation of f n (0 ), prove that
2
 x n 1 e xz
2πi ∫ C
  = x n
dz ,
 n ! n ! z n +1
where C is any closed contour surrounding the origin. Hence show that
2
 x n 1 2π
Σ  = ∫0 e 2 x cos θ dθ.
 
n ! 2π (Kanpur 2002; Gorakhpur 10)
 C-94

A nswers 3

5 17 2 65 3
1. (i) f (z ) = − 1 − z− z − z − ……
4 16 64
1 1 1 1 z z2 z3
(ii) f (z ) = …… + 4
− 3
+ 2
− + − + − ……
z z z z 4 42 43
5 17 65
(iii) f (z ) = 1 − + − + ……
z z2 z3
3 2 ∞ 1
2. f (z ) =  2 + 3  Σ (− 1)n + 1 (n + 1) n
z z  0 z
n
1  17  ∞ z
+3 z + + 15 Σ (− 1) n + 1 (n + 1) (n + 2)  
16  z  0  2
1 1 ∞
3. f (z ) = 2 + + 1 + Σ z n
z z n =1

1 ∞ zn z −2 ∞
4. (i) f (z ) = Σ (− 1)n . n
− Σ (− 1)n z 2 n
10 n=0 2 5 n=0
1 ∞ zn z −2 ∞ 1
(ii) f (z ) = Σ (− 1)n − 2
Σ (− 1)n
10 n=0 2 n
5z n=0 z2 n
n
1 ∞
 2 1 1 2  ∞ (− 1)n
(iii) f (z ) = Σ (− 1)n   −  − 2  Σ
5z n=0 z 5 z z  n=0 z2 n
∞ n
1 3  z − 1
5. f (z ) = − − Σ  
2 (z − 1) 4 n=0  2 
∞ ∞ (− 1)n (z − 1)n
6. (i) f (z ) = 1 − 2 Σ (− 1)n z n (ii) f (z ) = 1 − Σ
n=0 n=0 2n
2 ∞ 1
(iii) f (z ) = 1 − Σ (− 1)n n
z n=0 z
n n
1 ∞
z 1 ∞
z 1 ∞ (− 1)n
7. f (z ) = − Σ   + Σ (− 1)n   − Σ
24 n=0  2 8 n=0  2 3z n=0 z n
n
 π
∞ z − 
 π nπ   4 
8. f (z ) = Σ sin  + 
n=0 4 2 n!
∞
9. (i) f (z ) = Σ (− 1)n (z − 1)n
n=0
∞ n ∞
 π bn
(ii) f (z ) = Σ an  z −  + Σ
n=0  4  π
n =1 n
z − 
 4
1 2π sin φ . cosh (sin θ) . cos (mθ) 
where an =
2π ∫0 
 + cos φ . sinh (sin θ) . sin (mθ)
dθ

π
φ= + cos θ, m = n + 3; and bn = a(− n).
4
 C-95

Objective Type Questions

Multiple Choice Questions

Indicate the correct answer for each question by writing the corresponding letter from (a),
(b), (c) and (d).
1. If C is closed contour | z | = r and n ≠ − 1, then ∫ z n dz =
C

(a) 2πi (b) 2π

(c) i (d) 0.
1
2. The value of ∫ dz where C is circle z = e i θ , 0 ≤ θ ≤ π is
C z
(a) πi (b) − πi
(c) 2πi (d) 0.
3. ∫L dz , where L is any rectifiable arc joining the points z = a and z = b is equal to

(a) z (b) b − a
(c) a − b − z (d) z − a − b.
4. ∫ L | dz |, where L is any rectifiable arc joining the points z = a and z = b is equal
to
(a) b − a (b) | b − a |
(c) arc length of L (d) 0.
dz
5. If C is circle | z − a | = r, then ∫C z−a
is

(a) 2πi (b) − 2πi

(c) πi (d) 0.
6. If f (z ) is analytic in a simply connected domain D enclosed by a rectifiable
Jordan curve C and f (z ) is continuous on C, then for any point z0 in D, we have
f (z0 ) =
1 f (z ) 1 f (z )
2π ∫ C z − z0 2πi ∫ C z − z0
(a) dz (b) dz

f (z ) f (z )
(c) 2πi ∫C z − z0
dz (d) 2π ∫C z − z0
dz .

7. Let f (z ) be continuous on a contour L of length l and let| f (z )| ≤ M on L, then

 f (z ) dz is
 ∫L 
(a) ≤ Ml (b) ≥ Ml
(c) > Ml (d) < Ml.
 C-96

8. If a function f (z ) is analytic within a circle C with its centre z = a and radius R,

∞
then for every point z inside C, f (z ) = Σ an (z − a) n, where an =
n=0
(n) (n)
f (a) f (a)
(a) (b)
n n!
(n)
(c) f (a) (d) none of these.

2! z 2 + 3z + 4
9. Value of
2 πi ∫| z | = 3 (z − 1)3
dz is

(a) 2 (b) 0
(c) πi (d) none of these.

A nswers
1. (d) 2. (a) 3. (b) 4. (c) 5. (a)
6. (b) 7. (a) 8. (b) 9. (a)

¨
 1 The Zeros of an Analytic Function
efinition: The value of z for which the analytic function f (z ) becomes zero is said to be the
D zero of f (z ).
If f (z ) is analytic in a domain D and z0 is any point of D, then we can expand f (z ) as
Taylor’s series about z = z0 given by
∞
f (z ) = Σ an (z − z0 )n.
n=0

If a0 = a1 = a2 = …… = am − 1 = 0 and am ≠ 0, f (z ) is said to have a zero of order m at

z = z0 .
In this case Taylor’s expansion of f (z ) reduces to
∞ ∞
f (z ) = Σ an (z − z0 )n = Σ an + m (z − z0 )n + m
n= m n=0
∞
= (z − z0 )m Σ an + m (z − z0 )n.
n=0

A zero of order one (m = 1) is said to be a simple zero.

 C-98

2 The Zeros are Isolated

(Gorakhpur 2004)
Theorem: If f (z ) is an analytic function in a domain D, then unless f (z) is identically zero,
there exists a neighbourhood of each point in D throughout which the function has no zero except
possibly at the point itself.
Proof: Let z = z0 be a zero of order m of the function f (z ).
Then we can write
∞
f (z ) = (z − z0 )m Σ an + m (z − z0 )n. …(1)
n=0
∞
Let φ (z ) = Σ an + m (z − z0 )n.
n=0
Then φ (z0 ) = am ≠ 0 .
Now the series (1) is uniformly convergent and its each term is continuous at z0 so that
φ (z ) is also continuous at z0 . Therefore for ε > 0 there will exist δ > 0 such that
| φ (z ) − φ (z0 )| < ε , …(2)
where | z − z0 | < δ.
a
Let ε = m and δ1 be the corresponding value of δ. Then we have from (2)
 2
a
| φ (z ) − am | < m, …(3)
 2
where | z − z0 | < δ1.
Thus φ (z ) is non-zero at any point in the neighbourhood of| z − z0 | < δ1. For if we have
φ (z ) = 0, (3) will not hold. The argument also holds when m = 0 in which case φ = f and
f (z0 ) ≠ 0 .
Hence the zeros of an analytic function are isolated.

3 Singularities of an Analytic Function

(Gorakhpur 2006)
If a function is analytic at all points of a bounded domain except at a finite number of
points then these exceptional points are called singular points or singularities. Thus
the singularity of a function is a point at which the function ceases to be analytic.

4 Isolated and Non-isolated Singularities

(Purvanchal 2010)
If z0 is a singularity of f (z ) and if f (z ) is analytic at each point in some neighbourhood of z0 , then
z0 is called an isolated singularity of f (z ) otherwise it is called non-isolated singularity.
 C-99

z2 + 5
For example the function f (z ) = is analytic at every point except
z (z − 3)(z 2 + 1)
z = 0 , 3, ± i. These are the isolated singular points of f (z ).
1
Consider another function f (z ) = ⋅ It has infinite number of isolated
tan (π / z )
singularities which lie on the real axis from z = − 1 to z = 1. These isolated singularities
1
are given by z = ± , n = 1, 2, 3, … . The origin (z = 0 ) is also a singular point but it is not
n
isolated since in every neighbourhood of 0 there are infinite number of other
singularities.
The function Log z has a non-isolated singularity at origin since every neighbourhood
of zero contains points on the negative real axis where Log z is not analytic.

5 Isolated Essential Singularities

Let f (z ) be an analytic function in a domain D except at the point z = z0 . Then there
exists a deleted neighbourhood 0 < | z − z0 | < R in which f (z ) is analytic. In the
annulus 0 < | z − z0 | < R the Laurent expansion of f (z ) is
∞ ∞
f (z ) = Σ an (z − z0 )n + Σ bn (z − z0 )− n.
n=0 n =1
∞
The term Σ bn (z − z0 )− n of the Laurent’s series is called the principal part of f (z ) at
n =1

z = z0 .
Now there arise three possibilities :
(i) The principal part contains infinite number of terms.
(ii) All the bn are zero i.e., there is no term in the principal part.
(iii) There are finite number of terms in the principal part. The above three
possibilities give rise to three types of singularities :
Isolated essential singularity:If there are infinite number of terms in the principal part of
f (z ) at z = z0 , then z0 is called an isolated essential singularity of f (z ).
(Gorakhpur 2007)
1
For example sin has an isolated essential singularity at z = 0 since
z
1 1 1 1 1 1
sin = + ⋅ 3
+ ⋅ + ……has infinite number of terms in negative powers of z.
z z 3! z 5 ! z5

6 Removable Singularity
If all the coefficients bn are zero i.e., if the principal part of f (z) at z = z0 consists of no terms, then
z0 is called removable singularity of f (z ).
 C-100

We can remove this singularity by defining the function f (z ) at z = z0 in such a way

that it becomes analytic at z0 .
sin (z − 1)
For example the function f (z ) = has removable singularity at z = 1because
z −1

sin (z − 1) 1  (z − 1)3 (z − 1)5 

= (z − 1) − + − ……
z −1 z − 1  3! 5! 
(z − 1)2 (z − 1)4
= 1− + − ……
2! 5!
contains no negative powers of (z − 1). This singularity can be made to disappear by
sin (z − 1)
defining = 1 at z = 1 so that the function becomes analytic at z = 1.
z −1

7 Pole
If the principal part of f (z) at z = z0 consists of a finite number of terms, say m, then z = z0 is said
to be a pole of order m of the function f (z ). (Gorakhpur 2006; Purvanchal 10)
For m = 1, the point z = z0 is said to be a simple pole. If z = z0 is a pole of order m, f (z )
has an expansion of the form
∞ b1 b2 bm
f (z ) = Σ an (z − z0 )n + + + …… + ⋅
n=0 z − z0 (z − z0 )2
(z − z0 )m

8 Residue at Pole
Let z0 be a pole of order m of the function f (z ). Then we have
∞ b1 b2 bm
f (z ) = Σ an (z − z0 )n + + + …… + , bm ≠ 0 .
n=0 z − z0 (z − z0 )2 (z − z0 )m

The coefficient b, which may also be zero is called the residue of f ( z) at z = z0 .

lim
If z = z0 is a simple pole, then we have b1 = (z − z0 ) f (z ).
z → z0

9 Meromorphic Function
A function which has poles as its only singularities in the finite part of the plane is said to be a
meromorphic function.

10 Entire Function
A function which has no singularity in the finite part of the plane is called an entire function.
(Purvanchal 2009)
 C-101

11 Polynomials
An expression of the form Pn(z ) = a0 + a1z + a2 z 2 + … + an z n, where a0 , a1, ……, an are
complex numbers and an ≠ 0 is said to be a polynomial of degree n.
In particular, every constant is a polynomial of degree 0. The degree of the constant
polynomial 0 remains undefined.
For example, z n is a polynomial of degree n, 5 + 4 z 2 + 2 z 3 is a polynomial of degree 3.

12 Behaviour of a Polynomial at Infinity

Let P (z ) be a polynomial of degree n defined by
P (z ) = a0 + a1z + a2 z 2 + …… + an z n, an ≠ 0 .
In order to discuss the behaviour of P (z ) at ∞, let us consider the function
 1
P   = P1 (z ). Then
z
a1 a2 a
P1 (z ) = a0 + + 2 + …… + nn
z z z
1
= (a0 z n + a1z n − 1 + …… + an).
zn
P1(z ) → ∞ as z → 0, because an ≠ 0. Consequently P (z ) has a pole of order n at infinity.
Hence every polynomial of degree n has a pole of order n at infinity.

13 Characterization of Polynomials
Theorem 1: The order of a zero of a polynomial equals the order of its first non-vanishing
derivative.
Proof: Suppose z = a is a zero of order m of a polynomial P (z ).
Then P (z ) = (z − a)m Q (z ), Q (a) ≠ 0 .

Differentiating both sides successively m times, we get

P ′ (z ) = m (z − a)m − 1 Q (z ) + (z − a)m Q ′ (z )

P ′ ′ (z ) = m (m − 1) (z − a)m −2 Q (z ) + 2 m (z − a)m − 1 Q ′ (z )

+ (z − a)m Q ′ ′ (z )

…… …… …… ……
…… …… …… ……
m m
P (z ) = m ! Q (z ) + C1 m !(z − a) Q ′ (z ) + ……

+ (z − a)m Q m (z ).
Putting z = a in above relations, we get
 C-102
m −1
P (a) = P ′ (a) = P ′ ′ (a) = …… = P (a) = 0

and P m (a) = m ! Q (a) ≠ 0 .

Hence the order of a zero of a polynomial equals the order of its non-vanishing
derivative.
Theorem 2: (Luca’s Theorem): If all the zeros of a polynomial lie in a half plane, then all
the zeros of its derivative also lie in the same half plane.
Theorem 3: 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.
Theorem 4: A polynomial of degree n has no singularities in the finite part of the plane but has
a pole of order n at infinity.
Theorem 5: A function f (z) is a polynomial of degree n if and only if f (z) has no singularities
in the finite part of the plane and has a pole of order n at infinity.
Theorem 6: If a function f (z) is analytic for all finite values of z and as
| z | → ∞,| f (z )| = a | z | k then f (z) is a polynomial of degree ≤ k.

14 Rational Function
A function R (z ) which is obtained by applying the rational operations of arithmetic (addition,
subtraction, multiplication and division) finitely many times is called a rational function.
Thus a rational function R (z ) is of the form
P (z )
R (z ) = ,
Q (z )
where P (z ) and Q (z ) are polynomials given by
P (z ) = a0 + a1z + …… + an z n, an ≠ 0

Q (z ) = b0 + b1z + …… + bm z m , bm ≠ 0
having no factors in common.
R (z ) will tend to ∞ at the zeros of Q (z ). The zeros of Q (z ) are called poles of R( z) and
the order of a pole of R (z ) is defined as the order of the corresponding zero of Q (z ).
P ′ (z ) Q (z ) − Q ′ (z ) P (z )
We have R ′ (z ) = , provided Q (z ) ≠ 0 .
Q 2(z )
Obviously numerator and denominator of R ′ (z ) are polynomials. Therefore the
derivative of a rational function is also a rational function having the same poles as R (z )
and order of each pole is increased by one.

Poles and Zeros of a Rational Function at Infinity:

Consider a rational function
P (z )
R (z ) =
Q (z )
 C-103

where P (z ) = a0 + a1 z + …… + an z n, an ≠ 0

Q (z ) = b0 + b1 z + …… + bm z m , bm ≠ 0 .
Let R (1 / z ) = R1 (z ). Then the order of zero or pole at ∞ of R(z ) is defined as the order of
the zero or pole of R1 (z ) at the origin. We have
a + a1z + …… + anz n
R (z ) = 0
b0 + b1z + …… + bm z m
 a z n + a1z n − 1 + …… + an 
so that R1 (z ) = z m − n  0 m ⋅
 b0 z + b1z
m −1
+ …… + bm 
Now there arise three cases :
Case 1: For m > n, R1(z ) has a zero of order m − n at the origin and consequently R (z )
has a zero of order m − n at infinity.
Case 2: For m < n, R1(z ) has a pole of order n – m at the origin and therefore R (z ) has a
pole of order n − m at infinity.
Case 3: For m = n we have R (∞) = R1(0 ) = (an / bm ) ≠ 0 or ∞, therefore R (z ) has
neither a zero nor a pole at infinity.
The rational function R (z ) has n zeros and m poles in the finite part of the plane.
Therefore the total number of zeros and poles of R (z ) are as given below :

Number of zeros Number of poles

In the finite At ∞ Total In the finite At ∞ Total

plane plane

m> n n m−n m m — m

m< n n — n m n−m n

m=n n — m=n m — n=m

Hence the number of zeros of a rational function is equal to the number of its poles.
The total number of zeros or poles (the number of zeros and poles is equal) of a rational function is
called its order.

15 Characterization of Rational Functions

Theorem 1: If a single valued function f (z) has no singularities other than poles in the finite
part of the plane or at infinity, f (z) is a rational function.
Proof: Suppose z1, z2 , ……, z k are poles of f (z ) of orders m1, m2 , ……, mk in the finite
part of the z-plane. Then we can write
P (z )
f (z ) =
(z − z1) (z − z2 )m2 …… (z − z k )mk
m1

where P (z ) is an analytic function for all finite values of z.

 C-104

∴ P (z ) = (z − z1)m1 (z − z2 )m2 …… (z − z k )mk f (z ).

Since P (z ) is analytic for all finite values of z therefore the Taylor’s expansion of P (z ) is
of the form
∞
P (z ) = Σ an z n. …(1)
n=0

 1 ∞ an
Then P  = Σ , where z = 1 / ζ.
 ζ n=0 ζn
The behaviour of P (z ) at infinity is the same as the behaviour of P (1 / ζ) at ζ = 0. Since
the singularity of P (z ) at z = ∞ is a pole therefore the singularity of P (1 / ζ) at ζ = 0 is
also a pole. As a result the expansion of P (1 / ζ) will consist of a finite number of terms.
Consequently the expansion (1) of P (z ) must contain a finite number of terms.
Therefore P (z ) is a polynomial.
Since the numerator and denominator of f (z ) are polynomials therefore f (z ) is a
rational function.
Theorem 2: A rational function has no singularities other than poles.
Proof: Let f (z ) be a rational function given by
P (z )
f (z ) = ,
Q (z )
where P (z ) and Q (z ) are polynomials having no factor in common.
The singularities of the function f (z ) in the finite part of the plane are given by
1
Q (z ) = 0 . We know that zeros of Q (z ) are the poles of ⋅ Hence the rational
Q (z )
P (z )
function f (z ) = has no singularities other than poles in the finite part of the
Q (z )
plane.
Now we shall discuss the behaviour of f (z ) near z = ∞.
a0 + a1z + …… + anz n
Taking f (z ) = , an ≠ 0 , bm ≠ 0 ,
b0 + b1z + …… + bm z m
 a0 a1 
 + + …… + an 
n −1
=z n− m zn z
 b0 b ⋅
 m
+ m1− 1 + …… + bm 
 z z 
The coefficient of z n − m is regular for large values of | z | , therefore we can write
 c c 
f (z ) = z n − m  c0 + 1 + 22 + …… , c0 ≠ 0 .
 z z 
The power series in the bracket converges for| z | > r if r is sufficiently large. As a result
the behaviour of f (z ) near z = ∞ depends on the value of n − m. For n − m ≤ 0, the
Laurent expansion of f (z ) near z = ∞ contains no positive powers of z therefore f (z ) is
analytic near z = ∞. For n − m > 0, f (z ) has a pole of order n − m at z = ∞. Hence all the
singularities of a rational function are poles.
 C-105

16 Theorems on Poles and Other Singularities

Theorem 1: A function which has no singularity in the finite part of the plane or at infinity is
constant.
Proof: Let f (z ) be the function which has no singularity in finite part of the plane or
at infinity.
Then f (z ) can be expanded as a Taylor’s series in any circle | z | = r, where r is large.
∞
∴ f (z ) = Σ an z n. …(1)
n=0

 1
Since f (z ) has no singularity at infinity therefore f   is analytic at z = 0. Also
 z 
f (1 / z ) has no singularity in the finite part of the plane because f (z ) has no singularity
in the finite part of the plane. Consequently f (1 / z ) can be expanded as a Taylor’s
series.
∞
∴ f (1 / z ) = Σ bn z n. …(2)
n=0

Replacing z by 1 / z in (1), we get

∞
f (1 / z ) = Σ (an / z n). …(3)
n=0

From (2) and (3), we get

∞ ∞
Σ bn z n = Σ an z − n.
n=0 n=0

For the above relation to hold we must have

bn = an = 0 , n = 1, 2, 3,…… and b0 = a0 .
Hence we have f (z ) = a0 = b0 = constant.

Theorem 2: If z0 is a pole of f (z), there exists a neighbourhood of z0 which contains no other

pole of f (z) i.e., poles are isolated.
Proof: If z0 is a pole of order m of f (z ), there exists a deleted neighbourhood
0 < | z − z0 | < r of z0 in which f (z ) is analytic and has a Laurent’s expansion of the form
∞ ∞
f (z ) = Σ an (z − z0 )n + Σ bn (z − z0 )− n.
n=0 n =1

Thus f (z ) contains no other pole in the neighbourhood 0 < | z − z0 | < r i.e., poles are
isolated.

Theorem 3:If f (z) is a function such that for some positive integer m, a value φ (z0 ) exists with
φ (z0 ) ≠ 0 such that the function φ (z ) = (z − z0 )m f (z ) is analytic at z0 . Then f (z) has a pole of
order m at z0 .
Proof: The function φ (z ) is given to be analytic at z0 , so that it can be expanded in a
Taylor’s series about z0 .
 C-106

∴ φ (z ) = (z − z0 )m f (z )

φ m (z0 )
= φ (z0 ) + (z − z0 ) φ′ (z0 ) + …… + (z − z0 )m + ……
m!
φ (z0 ) φ′ (z0 ) φ m − 1 (z0 ) 1
or f (z ) = + + …… + ⋅
(z − z0 )m (z − z0 )m − 1 (m − 1) ! (z − z0 )
∞ φ n (z0 )
+ Σ (z − z0 )n − m .
n= m n!
Since we have φ (z0 ) ≠ 0 therefore f (z ) has a pole of order m at z0 .
1 φ m − 1 (z0 )
Also the residue at z0 = coeff. of = ⋅
z − z0 (m − 1)!
Remark: It follows from the above theorem that if a function f (z ) can be put in the
φ (z )
form f (z ) = where φ (z ) is analytic at z0 with φ (z0 ) ≠ 0 , then f (z ) has a pole of
(z − z0 )m
order m at z0 .

Theorem 4: If f (z) has a pole of order m at z0 , then the function φ defined by

φ (z ) = (z − z0 )m f (z ) has a removable singularity at z0 and φ (z0 ) ≠ 0.

φ m − 1 (z0 )
Also show that the residue at z0 is given by ⋅
(m − 1)!
Proof: Since z = z0 is a pole of order m of f (z ), therefore there exists a deleted
neighbourhood of z0 given by 0 < | z − z0| < r (r > 0 ) in which f (z ) has a Laurent’s
expansion
∞ b1 b2 bm
f (z ) = Σ an (z − z0 )n + + + …… + , …(1)
n=0 z − z0 (z − z0 )2 (z − z0 )m

where bm ≠ 0.
Consider a function φ defined by
φ (z ) = (z − z0 )m f (z ). …(2)
Then φ (z ) is defined in the neighbourhood of z0 except at z0 .
From (1) and (2), we have
∞
φ (z ) = Σ an (z − z0 )m + n + b1 (z − z0 )m − 1 + b2 (z − z0 )m − 2 +…… + bm .
n=0

…(3)
Let us define φ (z0 ) = bm . Then φ (z0 ) ≠ 0 , so that the expansion of φ (z ) given by (3) is
valid throughout a neighbourhood of z0 including z0 . It can be easily shown that (3) is a
convergent power series. Thus φ (z ) is analytic at z0 . Therefore we have made φ (z )
analytic at z0 by setting φ (z0 ) = bm . Hence φ (z ) has a removable singularity at z0 .
Since φ (z ) has become analytic at z0 , therefore (3) represents a Taylor’s series for φ (z ).
Consequently coefficient of
 C-107

φ m − 1 (z0 )
(z − z0 )m − 1 = ⋅
(m − 1)!
But from (3) coefficient of (z − z0 )m − 1 = b1, which is the residue at z0 .

φ m − 1 (z0 )
Hence the residue at z0 = ⋅
(m − 1)!

In particular when z0 is a simple pole, then the residue at z0 is

lim
φ (z0 ) = (z − z0 ) f (z ) .
z → z0

Theorem 5: Let a function f (z) be analytic in an open domain D and let φ (z ) be defined by
1
φ(z ) = where f (z ) ≠ 0 . Then f has a zero of order m at a point z0 in D if and only if φ has a
f (z )
pole of order m at z0 .
Proof. The if part: Suppose the function φ (z ) has a pole of order m at z0 . Then we
have to show that f (z ) has a zero of order m at z0 .
Since φ (z ) has a pole of order m at z0 therefore we can write
g (z )
φ (z ) =
(z − z0 )m

where g (z ) is analytic function in a neighbourhood of z0 and g (z0 ) ≠ 0 .

1 1 (z − z0 )m
It is given that φ (z ) = , so that f (z ) = = ⋅
f (z ) φ (z ) g (z )
Because g (z ) is an analytic function and g (z0 ) ≠ 0 , therefore f (z ) has a zero of order m
at z0 .
The only if part: Again suppose f (z ) has a zero of order m at z0 .Then we can write
f (z ) = (z − z0 )m h (z ) …(1)
where h (z ) is analytic and h (z0 ) ≠ 0 .
1 (z − z0 )m
∴ =
h (z ) f (z )
is an analytic function in a neighbourhood of z0 so it can be expanded in the Taylor’s
series about z0 .
1
Thus = A0 + A1 (z − z0 ) + A2 (z − z0 )2 + …… + Am (z − z0 )m + ……
h (z )
…(2)
1
Now φ (z ) = therefore from (1) and (2), we have
f (z )
1 A0 A1
φ (z ) = = + + ……
h (z ) (z − z0 )m
(z − z0 )m
(z − z0 )m − 1
∞
+ Am + Σ Am + n (z − z0 )n.
n =1
Thus φ (z ) has a pole of order m at z0 .
 C-108

Theorem 6. (Riemann): Let z0 be an isolated singularity of f (z) and if| f (z )| is bounded

on some deleted neighbourhood of z0 , z0 is a removable singularity. (Kanpur 2008)
Proof: Suppose | f (z )| is bounded in a deleted neighbourhood N (z0 ) of z0 . Then
there exists a positive number M such that| f (z )| ≤ M where M is the greatest value of
f (z ) on a circle γ defined by | z − z0 | = ρ where ρ is so small that γ lies entirely within
N (z0 ). By Laurent’s theorem, we have
∞ 1 f (z )
f (z ) = Σ an (z − z0 )n, where an =
n= − ∞ ∫
2 πi γ (z − z0 )n + 1
dz .

 1 f (z )  1 | f (z )|| dz |
Now | an | =
 2 πi
∫ γ (z − z )
0
n + 1
dz ≤
 2π
∫ γ | z − z | n +1
0
M 1 M M
2 π ρn + 1 ∫ γ
≤ ⋅ | dz | = 2 πρ = n → 0 as n → ∞
2 π ρn + 1 ρ
i. e., an becomes zero when n is negative so that the principal part of f (z ) contains no
terms of negative powers of z − z0 in the Laurent’s expansion for f (z ). Hence f (z ) has
removable singularity at z0 .

Behaviour of a function in the neighbourhood of an essential singularity.

(Avadh 2008)
Theorem 7: (Weierstrass’s Theorem): Let z0 be an essential singularity of a function f
(z) and let c be an arbitrary constant. Then for every ε > 0 and every neighbourhood
0 < | z − z0 | < ρ of z0 , there exists a point z of this neighbourhood such that| f (z ) − c | < ε .
OR
In every arbitrary neighbourhood of an essential singularity there exists a point (and therefore an
infinite number of points) at which the function differs as little as we please from any pre-assigned
number.
Proof: We shall prove the theorem by contradiction. Let the theorem be false. Then
for a given ε > 0, there exists a constant c and a positive number ρ such that
| f (z ) − c | > ε, where z satisfies 0 < | z − z0 | < ρ.
1
Thus < ε for 0 < | z − z0 | < ρ.
| f (z ) − c |
1
By Riemann’s Theorem (Theorem 6) we see that the function has a
f (z ) − c
removable singularity at z0 so that the principal part of Laurent’s expansion for
1
contains no negative powers of (z − z0 ). Then we can write
f (z ) − c
1 ∞
= Σ an (z − z0 )n.
f (z ) − c n=0
1 1 1
If a0 ≠ 0 , defining = a0 or f (z0 ) = c + , then becomes analytic
f (z0 ) − c a0 f (z ) − c
and non-zero at z0 so that f (z ) itself is analytic at z0 . This contradicts the initial
assumption that z0 is not an essential singularity of f (z ). Again if we take
a0 = a1 = a2 = …… = am − 1 = 0, we can write
 C-109

1 ∞
= Σ an (z − z0 )n
f (z ) − c n= m

= am (z − z0 )m + am + 1 (z − z0 )m + 1 + ……
∞
= (z − z0 )m Σ am + n (z − z0 )n
n=0

1
which shows that z0 is a zero of order m of so that z0 is a pole of order m of
f (z ) − c
f (z ) − c . Since c is merely a constant therefore f (z ) has a pole of order m at z0 which
again contradicts the hypothesis. Hence the theorem is true.

17 Limit Points of Zeros

Theorem 1: Let f (z ) be analytic in a domain D and let E be a set of zeros of f (z ) having a limit
point α in D. Then f (z ) vanishes for all z ∈ D.
Proof: Since f (z ) is analytic in D therefore it is continuous in D. E is the set of zeros
of f (z ) and α is the limit point of E in D therefore f (z ) vanishes at infinite number of
points in every small neighbourhood of α. Since f (z ) is continuous at α therefore we
must have f (α) = 0 . But α cannot be a zero of f (z ) since zeros are isolated. Hence f (z )
vanishes identically in D i. e., f (z ) vanishes for all z ∈ D.
Remark: If f (z ) is not analytic in D and f (z ) is not continuous at z = α then f (z )
must have a singularity at z = α. In this case α is an isolated singularity but it is not a pole
since| f (z )|does not tend to ∞ as z → α in any manner. Hence z = α which is the limit
point of zeros is an isolated essential singularity of f (z ).
Theorem 2: (Identity Theorem): If f (z) and g (z) are analytic in a domain D and if f (z)
= g (z) on a subset E of D which has a limit point α in D, then f (z) = g (z) in the whole of D.
Proof: Let F (z ) = f (z ) − g (z ). Then F (z ) is analytic in D. Since f (z ) = g (z ) on E
therefore F (z ) vanishes at an infinite number of points in every arbitrary small
neighbourhood of α. The function F (z ) is continuous at α ∈ D therefore we have
F (α) = 0 . But zeros are isolated so that α cannot be a zero of F (z ) unless F (z ) vanishes
identically in D i. e., we must have f (z ) = g (z ) in the whole of D.

18 The Limit Point of Poles

Theorem 1: The limit point of a sequence of poles of a function f (z) is a non-isolated essential
singularity.
Proof: Let z0 be the limit point of a sequence of poles of f (z ). Then every
neighbourhood of z0 contains infinite number of points at which f (z ) becomes
unbounded so that f (z ) cannot be analytic at z0 . Thus z0 is a singularity of f (z ) which
is not isolated. Hence z0 is a non-isolated essential singularity of f (z ).
 C-110

1
Example: The zeros of the function sin are given by
z
1
z=± , n = 1, 2, 3,…
nπ
1
The limit point of these zeros is the point z = 0.Thus 0 is an isolated singularity of sin ⋅
z
1
Again the function tan has poles at points given by
z
2
z= , n = ± 1, ± 3, ± 5,…
nπ
The limit point of this sequence of poles is z = 0 which is therefore a non-isolated
essential singularity.

Theorem 2: (Picard's Theorem): An integral function which is constant takes every finite
value an infinite number of times with at most one possible exception.
Proof: Recall that a function f (z ) is called an integral function if f (z ) has no
singularities except at infinity. We shall not try to prove this theorem but only give an
example.
The equation e z = A has an infinite number of roots if A ≠ 0 as the reader can easily
verify. But if A = 0,this equation has no finite root. Thus 0 is an exceptional value of e z .

On the other hand, there exist integral functions with no exceptional values. The
function sin z provides a simple example of such a case.

Example 1: Show that the function e1 / z actually takes every value except zero an infinite number
of times in the neighbourhood of z = 0. (Gorakhpur 2004)

Solution: Let f (z ) = e1 / z .

To prove the required result we have to show that f (z ) has an isolated essential
singularity at z = 0.
1 1 1 ∞ 1 1
We have f (z ) = e1 / z = 1 + + 2
+ 3
+ …… = 1 + Σ ⋅ n⋅
z 2! z 3! z n =1 n ! z

The principal part of f (z ) contains infinite number of terms therefore z = 0 is an

isolated essential singularity of f (z ).
Example 2: Show that the function (z 2 + 4) / e z has an isolated essential singularity at z = ∞.

z2 + 4
Solution: We have f (z ) = ⋅
ez
1  1  1
Putting z = , we get f   = 4 + 2  e − 1 / y
y  y  y 
 C-111

 1  1 1 1 1 1 
= 4 + 2  1 − + . 2 − 3
+ …
 y   y 2! y 3! y 
 4 1  2 1  1 1 1 
= 4 + − + (1 + 2) 2 +  − 1 −  3 +  +  4 + ....
 y y  
3 y  
2 6 y 
 4 3 5 2 
= 4 + − + 2 − 3
+ 4
− ...... ⋅
 y y 3y 3y 
We have infinite number of terms in the negative powers of y in the principal part of
 1  1
the expansion of f   ,therefore f   has an isolated essential singularity at y = 0.
 y  y
Hence f (z ) has an isolated essential singularity at z = ∞.
Example 3: What kind of singularity have the following functions:
cot πz 1
(i) at z = 0 , z = ∞ (ii) sin at z = 1
(z − a)2 1− z
(Purvanchal 2012) (Gorakhpur 2010, 13)
1
(iii) sin z − cos z at z = ∞ (iv) cosec at z = 0
z
(Gorakhpur 2008)
1
(v) tan at z = 0 .
z
cot πz cos πz
Solution: (i) Let f (z ) = 2
= ⋅
(z − a) sin πz (z − a)2
Poles of f (z ) are obtained by equating to zero the denominator of f (z ).Then we have
(z − a)2 sin πz = 0

∴ sin πz = 0 or (z − a)2 = 0 .
Now sin πz = 0 gives πz = nπ or z = n, where n is any integer,
and (z − a)2 = 0 gives z = a.
Hence z = a is a double pole and z = 0 , ± 1, ± 2, …… are simple poles.
z = ∞ is a limit point of these simple poles therefore z = ∞ is non-isolated essential
singularity.
1
(ii) Let f (z ) = sin ⋅
1− z
Zeros of f (z ) are given by
1 1 1 ,
sin = 0 or = nπ or z = 1 − where n is any integer.
1− z 1− z nπ
z = 1 is a limit point of these zeros therefore z = 1 is an isolated essential singularity.
(iii) Let f (z ) = sin z − cos z .
Zeros of f (z ) are given by
sin z − cos z = 0 or sin z = cos z
 C-112
π
or tan z = 1 or z = nπ +
, n is any integer.
4
z = ∞ is a limit point of these zeros which is therefore an isolated essential singularity.
1 1
(iv) Let f (z ) = cosec = ⋅
z sin (1 / z )
Poles of f (z ) are given by
1 1
sin = 0 or = nπ or z = (1 / nπ), where n is any integer.
z z
Since z = 0 is a limit point of these poles therefore z = 0 is a non- isolated essential
singularity.
sin (1 / z )
(v) Let f (z ) = tan (1 / z ) = .
cos (1 / z )
Poles of f (z ) are given by
1 π
cos (1 / z ) = 0 or = 2 nπ ±
z 2
1
or z= , where n is any integer.
1
(2 n ± ) π
2
Since z = 0 is a limit point of these poles therefore z = 0 is a non-isolated essential
singularity.
e c /(z − a)
Example 4: Show that z = a is an isolated essential singularity of the function ⋅
ez / a − 1
Solution: We have
c /(z − a) c /(z − a)
e e
f (z ) = =
e z /a
−1 e 1 + (z − a) / a − 1
c c2 c3
1+ + + + ……
z − a 2 !(z − a)2 3 !(z − a)3
=
 z − a (z − a)2 
e 1 + + 2
+ …… − 1
 a 2! a 
 c c2 
= − 1 + + + ……
 z − a 2 !(z − a) 2

−1
  z − a (z − a)2 
× 1 − e 1 + + 2
+ …… 
  a 2! a  
 c c2   z − a (z − a)2 
= − 1 + + + …… 1 + e 1 + + + ……
 z − a 2 !(z − a)2
   a 2 ! a2

 
2 
2 z − a (z − a)2 
+ e 1 + + + …… + ……
 a 2 ! a2
 

Obviously in the expansion of f (z ) there are infinite number of terms containing
negative powers of (z − a). Hence z = a is an isolated essential singularity of f (z ).
 C-113

Example 5: Discuss the nature of singularities of the following functions :

1 z sin z
(i) tan z (ii) (iii) (iv) ⋅
z (1 − z 2 ) 1 + z4 (z − π)2
sin z
Solution: (i) Let f (z ) = tan z = ⋅
cos z

To obtain the singularities of f (z ) equating to zero the denominator of f (z ), we get

π
cos z = 0 or z = 2 nπ ± , n ∈ I
2
π π
or z = (4 n ± 1) , n ∈ I or z = (2 n + 1) , n ∈ I .
2 2
π
Hence z = (2 n + 1) , (n ∈ I ) give the simple poles of f (z ).
2
1
(ii) Let f (z ) = ⋅
z (1 − z 2 )
Singularities of f (z ) are given by
z (1 − z 2 ) = 0 or z = 0 , − 1, 1, which are the simple poles.
z
(iii) Let f (z ) = ⋅
1 + z4

Singularities of f (z ) are given by

1 + z4 = 0 or z = (− 1)1 /4

or z = (cos π + i sin π)1 /4 = {cos (2 nπ + π) + i sin (2 nπ + π)}1 /4

π
π π i (2 n + 1)
= cos (2 n + 1) + i sin (2 n + 1) = e 4 ⋅
4 4
Putting n = 0 , 1, 2, 3, we get
z = e iπ /4 , e3 πi /4 , e5 πi /4 , e7πi /4 which are the simple poles of f (z ).
sin z
(iv) Let f (z ) = ⋅
(z − π)2

Singularities of f (z ) are given by (z − π)2 = 0 .

Thus z = π is a pole of order two of f (z ).

Comprehensive Exercise 1

1. Show that the function e z has an isolated essential singularity at z = ∞.

(Gorakhpur 2016)
2. What kind of singularity have the following functions :
1
(i) cot z at z = ∞ (ii) sec at z = 0.
z
 C-114

3. Specify the nature of singularity at z = − 2 of

1
f (z ) = (z − 3) sin ⋅
z +2
2
 z +1
4. (i) Find zeros and poles of  2  .

 z + 1
ez
(ii) What kind of singularity has the function 2
?
z +4
5. Find the kind of the singularities of the following functions :
1− e z 1 π
(i) at z = ∞ (ii) at z =
1+ e z sin z − cos z 4
(Gorakhpur 2010, 13)
1
(iii) sin at z = 0 (iv) z cosec z at z = ∞.
z (Purvanchal 2007)
6. Find the zeros and discuss the nature of singularities of
z −2 1
f (z ) = 2 sin ⋅
z z −1 (Kanpur 2007)
−1 / z2
7. Show that the function e has no singularities. (Gorakhpur 2014)

A nswers 1

2. (i) non-isolated essential singularity

(ii) non-isolated essential singularity
3. isolated essential singularity
4. z = − 1, − 1 ; z = − i, − i, i, i ; z = 2 i, − 2 i are simple poles.
5. (i) non-isolated essential singularity ; (ii) simple pole.
(iii) isolated essential singularity
(iv) non-isolated essential singularity
6. zeros of order one; isolated essential singularity

19 Maximum Modulus Principle

Theorem: Let f (z ) be analytic within and on a simple closed contour C. Then| f (z )|reaches
its maximum value on C, unless f (z ) is a constant. In other words, if M is the maximum value of
| f (z )| on and within C, then, unless f is a constant, | f (z )| < M for every point within C.
(Rohilkhand 2009, 10)
(This is known as Maximum modulus principle.)
 C-115

Since f (z ) is an analytic function, it is continuous within

and on C so| f (z )| must reach its maximum value M at
some point on or within C. Let us assume that f (z ) is not
constant. Then we wish to prove that | f (z )| takes the
value M at some point on C. If possible, suppose this value
is not attained on the boundary of C but at a point a
within C, so | f (a)| = M . Let Γ be a circle inside C with
centre at a.
Since | f (z )| = M is the maximum value of | f (z )| and
f (z ) is not a constant, there must exist a point, b, inside Γ such that | f (b)| < M .
Let | f (b)| = M − ε, where ε > 0. Since | f (z )| is continuous at b, therefore for ε > 0,
there exists a δ > 0 such that
1
|| f (z )| − | f (b)|| < ε, …(1)
2
where | z − b | < δ.
We have | f (z )| − | f (b)| ≤ | f (z )| − | f (b)| therefore from (1), we have
1
| f (z )| − | f (b)| < ε
2
1 1 1
or | f (z )| < | f (b)| + ε = M − ε + ε = M − ε
2 2 2
1
∴ | f (z )| < M − ε for all points z satisfying | z − b | < δ.
2
Draw a circle Γ ′ with centre at a, which passes through b.
Since the arc QR of the circle Γ ′ lies inside γ therefore on this arc, we have
1
| f (z )| < M − ε.
2
On the remaining arc of Γ′, we have | f (z )| ≤ M .
Radius of the circle Γ′ = | b − a | = r, say.
By Cauchy’s integral formula, we have
1 f (z ) 1 2π f (a + re i θ )
rie iθ dθ,
2πi ∫Γ ′ z − a 2 πi ∫0
f (a) = dz =
re i θ
putting z − a = re i θ
1 2π
= ∫0 f (a + re i θ ) dθ.
2π
Measuring θ from PQ in the anti-clockwise direction and taking ∠ QPR = α,we have
1 α 1 2π
∫ f (a + re i θ ) dθ + f (a + re i θ ) dθ.
2π ∫ α
f (a) =
2π 0
1 α 1 2π
| f (a + re i θ )| dθ + | f (a + re iθ )| dθ
2 π ∫0 2π ∫ α
Now | f (a)| ≤

1 α  1  1 2π
<
2π ∫0 M −
 2
ε dθ +
 2π ∫α M dθ
 C-116
α  1  M αε
=  M − ε + (2 π − α) = M − ⋅
2π  2  2π 4π
αε
∴ M = | f (a)| < M − , which is absurd.
2π
Hence| f (z )|cannot attain its maximum value at any point within C, so it must attain
its maximum value on C.

20 Minimum Modulus Principle

Theorem 1: Let f (z ) be analytic inside and on a closed contour C and let f (z ) ≠ 0 inside C.
Then | f (z )| must reach its minimum value on C.
OR
If m is the minimum of | f (z )| inside and on C, then unless f is constant ,| f (z )| > m for every
point z inside C.
Proof: Since f (z ) is analytic inside and on C and f (z ) ≠ 0 inside C therefore1 / f (z )
is analytic inside C. By the previous theorem 1/| f (z )|cannot reach its maximum value
inside C so that | f (z )| cannot attain its minimum value inside C. Since f (z ) is
continuous on and within C therefore| f (z )| must attain its minimum value at some
point on C.

21 The Excess of the Number of Zeros over the Number of

Poles of a Meromorphic Function. The Argument
Principle
Theorem 1: If f (z ) is meromorphic inside a closed contour C and has no zero on C, then
1 f ′ (z )
2πi ∫ C f (z )
dz = N − P,

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). (Avadh 2007; Kanpur 08; Gorakhpur 09, 11, 12)

Proof: Let z = ai (i = 1, 2, ……, m) be the

zeros of f (z ) which lie inside C and ri be
the order of ai . Again suppose
bi (i = 1, 2, ……, n) be the poles of f (z )
inside C and si be the order of bi . Then we
have to show that
1 f ′ (z ) m n
∫
2 πi C f (z )
dz = Σ ri − Σ si .
i =1 i =1

Enclosing each zero and pole by

non-overlapping circles A1, A2 , ……, Am
 C-117

and B1, B2 , ……, Bn respectively each of radii ρ. Since poles and zeros are isolated, we
can always find such ρ. Therefore
1 f ′ (z ) m 1 f ′ (z ) n 1 f ′ (z )
∫
2 πi C f (z )
dz = Σ ∫
i = 1 Ai 2 πi f (z )
dz + Σ ∫
i = 1 2 πi B i f (z )
dz .

…(1)
Since ai is a zero of order ri of f (z ), we may write
f (z ) = (z − ai)ri φi (z ), where φi is analytic and non-zero at ai .
Taking log of both sides, we get
log f (z ) = ri log (z − ai) + log φi (z ).
Differentiating both sides w.r.t. z, we get
f ′ (z ) ri φ ′ (z )
= + i ⋅
f (z ) z − ai φi (z )
φi ′ (z ) φ ′ (z )
We have ∫ Ai φi (z )
dz = 0 , since i
φi (z )
is analytic at z = ai ,

ri 2 π ρ i ei θ
and ∫ Ai z − ai
dz = ri ∫
0 ρ e iθ
dθ = 2 πiri .

m 1 f ′ (z ) m 1 m
∴ Σ
i = 1 2 πi ∫ Ai f (z )
dz = Σ
i = 1 2 πi
2 πiri = Σ ri .
i =1
…(2)

Since bi is a pole of order si of f (z ), we may write

ψ i (z )
f (z ) = , where ψ i is analytic and non-zero at bi .
(z − bi)si
Taking log of both sides and differentiating w.r.t. z, we get
f ′ (z ) ψ i ′ (z ) si
= − ⋅
f (z ) ψ i (z ) z − bi
f ′ (z )
Proceeding as above, we have ∫ Bi f (z )
dz = − 2πisi .

n 1 f ′ (z ) n 1 n

i = 1 2 πi ∫ B i
∴ Σ dz = Σ (− 2 π i si) = − Σ si . …(3)
f (z ) i = 1 2 πi i =1

Hence from (1), (2), (3), we have

1 f ′ (z ) m n

2 πi ∫ C f (z )
dz = Σ ri −
i =1
Σ si = N − P.
i =1
…(4)

1
Corollary 1: N − P= ∆C arg f (z )
2π
where ∆C denotes the variation in arg f (z) as z moves once round C.
Proof: By the above theorem, we have
1 f ′ (z )
2πi ∫ C f (z )
N − P= dz .
 C-118

Let f (z ) = R e iφ . Then R = | f (z )|, φ = arg f (z )

and f (z ) dz = d f (z ) = d ( R e iφ ) = e iφ (dR + iRdφ).

1  dR  1 dR 1
2 πi ∫ C  R  2 πi ∫ C R 2 π ∫C
We have N − P =  + i d φ = + dφ.

dR
Now ∫C R
= [log R]C = 0 ,

since log R retains its original value if z moves once round C.

1 1 1
2π ∫ C
Also dφ = [φ]C = ∆C arg f (z ).
2π 2π
Hence we have
1
N − P= ∆C arg f (z ). …(5)
2π
Thus the excess of the number of zeros over the number of poles of a meromorphic
function equals (1 / 2π) times the increase in arg f (z ) as z goes once round C.
This is known as the argument principle.

Corollary 2: When f (z ) is analytic, we have P = 0 and in this case

1 f ′ (z ) 1
2 πi ∫ C f (z )
N = dz = ∆C arg f (z )
2π (Kanpur 2007)

i.e., the number of zeros of an analytic function f (z ) within C is (1 / 2π) times the
increase in arg f (z ) as z goes once round C.
This is known as the argument principle for an analytic function.

Remark: (i) We observe that the variation in arg f (z ) as z moves round C is always
equal to an integer. If z0 is any point on C, we have
∆C arg f (z ) = [arg f (z0 )]* − arg f (z0 )

where [arg f (z0 )]* is the value of the argument after the contour C has been traversed.
Since any two values of an argument differ by an integral multiple of 2π therefore we
have
1 1
∆C arg f (z ) = ⋅ 2 πm = m, where m is an integer.
2π 2π
(ii) We can use the formulae (4) and (5) to count the number of times, N α , a function
f (z ) takes the values α. f (z ) − α = 0 iff f (z ) = α and then (4) gives
1 f ′ (z )
2 πi ∫ C f (z ) − α
Nα − P = dz .

Similarly from (5), we get

1
Nα − P = ∆C arg [ f (z ) − α].
2π
 C-119

22 Rouche’s Theorem
(Gorakhpur 2007, 10, 14; Avadh 07; Purvanchal 09; Kanpur 07; Rohilkhand 08, 09)

Theorem 1: Let f (z ) and g (z ) be analytic inside and on a simple closed curve C and let
| g (z )| < | f (z )|on C. Then f (z ) and f (z ) + g (z ) have the same number of zeros inside C.
Proof: We observe that both f (z ) and f (z ) + g (z ) are non-zero on the boundary C.
If at some point a on C, we have f (a) = 0 , then | g (a)| < | f (a)| ⇒ g (a) = 0, which
contradicts the hypothesis that | g (z )| < | f (z )| on C. Similarly if we take
f (a) + g (a) = 0, then | g (a)| = | f (a)| which is again a contradiction.
Hence neither f (z ) nor f (z ) + g (z ) has a zero on C.
Let F (z ) = g (z ) / f (z ). Then g (z ) = f (z ) F (z ) so that
g ′ (z ) = f ′ (z ) F (z ) + f (z ) F ′ (z ).
Suppose M and N are the number of zeros of f (z ) and f (z ) + g (z ) inside and on C.
Since f (z ) and f (z ) + g (z ) are analytic within and on C, we have
1 f′ 1 f ′ + g′
M= ∫
2πi C f
dz and N = ∫
2πi C f + g
dz ,

 1 f ′ (z ) 
2πi ∫ C
Using the formula N − P = dz  ⋅
 f (z ) 
1 f′+ f′ F + f F′ 1 f′
2 πi ∫ C 2 πi ∫ C f
Now N − M = dz − dz
f + f F
1 f ′ (1 + F ) + f F ′ 1 f′
2 πi ∫ C 2 πi ∫ C
= dz − dz
f (1 + F ) f
1 f′ 1 F′ 1 f′
2 πi ∫ C ∫ 2 πi ∫ C
= dz + dz − dz
f 2 πi C 1+ F f
1
F ′ (1 + F )−1 dz
2 πi ∫ C
=

1
2 πi ∫ C
= F ′ (1 − F + F 2 − F 3 + ……) dz [∵| F (z )| < 1]

1 1 1
=
2 πi ∫ C
F ′ dz −
2 πi ∫ C
F ′ F dz +
2 πi ∫ C
F ′ F 2 dz − ……

= 0,
since f (z ) and g (z ) are analytic and g (z ) ≠ 0 at any point on C, so F and F ′ are also
analytic within and on C, consequently each integral is separately zero.
Hence N = M i.e., f (z ) and f (z ) + g (z ) have the same number of zeros inside C.
Alternative proof: First show that neither f (z ) nor f (z ) + g (z ) has zero on C.
(Proceed as above)
Suppose M and N are the number of zeros of f (z ) and f (z ) + g (z ) inside and on C.
Since f (z ) and f (z ) + g (z ) are analytic within and on C, by the argument principle for
analytic functions, we have
 C-120

1 1
M= ∆C arg f and N = ∆C arg ( f + g).
2π 2π
1
Now N −M= {∆C arg ( f + g) − ∆C arg f }
2π
1   g  
=  ∆C arg f 1 +  − ∆C arg f 
2π   f  

1    g   
= ∆C arg f + arg 1 +  − arg f  
2 π    f   

1  g  1
= ∆C arg 1 +  = ∆C arg w , where w = 1 + ( g / f ).
2π  f  2π

 g 
Since| g | < | f |, we have| w − 1| = < 1so that the point w always lies inside the circle
 f 
with centre w = 1 and radius unity. Thus the point w always lies to the right of the
 g  π π
imaginary axis consequently arg w = arg 1 +  always lies between − and ⋅ It
 f  2 2
 g 
follows that arg 1 +  returns to its original value when z describes C. Since arg
 f 
 g   g 
1 +  cannot increase or decrease by a multiple of 2π,we have ∆C arg 1 +  = 0.
 f   f 

Hence N − M = 0, which gives N = M .

23 Fundamental Theorem of Algebra

Theorem 1: (Fundamental theorem of Algebra):
(Gorakhpur 2004; Purvanchal 08; Rohilkhand 08, 09)
n
Let P (z ) = a0 + a1z + …… + anz , where n ≥ 1and an ≠ 0 so that P (z ) is a polynomial of degree
one or greater. Then the equation P (z ) = 0 has at least one root.
(We shall prove it with the help of Liouville’s theorem since its proof by purely
algebraic method is extremely difficult).
Proof: We shall prove it by contradiction. Suppose P (z ) is not zero for any value of z.
Then
1
f (z ) = , is analytic everywhere.
P (z )
1 1
We have f (z ) = =
P (z ) a0 + a1 z + …… + an z n
1
= → 0 as z → ∞.
n  a a 
z  n + n − 1 + …… + an
0 1
z z 
 C-121

∴ For every ε > 0, there exists a δ > 0 such that | f (z )| < ε for | z | > δ.
Since f (z ) is continuous in the bounded closed domain | z | ≤ δ therefore f (z ) is
bounded in the closed domain | z | ≤ δ so there exists a positive number K such that
| f (z )| < K for | z | ≤ δ.
 1 
If M = max (ε, K), then we have | f (z )| = < M , for every z.
P (z )
Hence by Liouville’s theorem f (z ) is constant. This gives a contradiction since P (z ) is
not constant for n = 1, 2, 3, …… and an ≠ 0. Thus P (z ) must be zero for at least one value
of z i.e., P (z ) = 0 must have at least one root.
Corollary: Every polynomial equation
P (z ) = a0 + a1z + a2 z 2 + …… + an z n = 0 ,
where n ≥ 1, an ≠ 0 has exactly n roots. (Gorakhpur 2006, 11, 13)

Proof: By the fundamental theorem of Algebra P (z ) = 0 has at least one root, say α1.
Then we have P (α1) = 0 .
Now P (z ) = P (z ) − P (α1) = (a0 + a1z + a2 z 2 + …… + an z n)

− (a0 + a1α1 + a2α12 + …… + anα1n)

= a1 (z − α1) + a2 (z 2 − α12 ) + …… + an (z n − α1n)

= (z − α1) P1 (z ), where P1 (z ) is a polynomial of degree n − 1.

Again by fundamental theorem of Algebra P1 (z ) = 0 must have at least one root, say α2 ,
(α2 may be equal to α1).
Proceeding as above, we have P (z ) = (z − α1) (z − α2 ) P2 (z ),
where P2 (z ) is a polynomial of degree n − 2.
Continuing in this way we see that P (z ) = 0 has exactly n roots.
Alternative proof: We shall use Rouche’s theorem to show that P (z ) = 0 has exactly
n roots.
Let f (z ) = an z n and g (z ) = a0 + a1z + …… + an − 1z n − 1.

g (z ) 1  a0 a an − 1 
Then =  + 1 + …… +  → 0 as z → ∞.
f (z ) an  z n z n − 1 z 

∴ there exists a δ > 0 such that

 g (z ) 
 < 1 for | z | > δ. …(1)
 f (z )
We now take the closed curve C as the circle| z | = δ + 1. Then from (1),| g (z )| < | f (z )|
on C. Also f (z ) and g (z ) are analytic on and inside C. Therefore by Rouche’s theorem
f (z ) and f (z ) + g (z ) have the same number of zeros inside C. Since f (z ) = an z n has n
zeros at the origin therefore f (z ) + g (z ) must also have n zeros. Hence the polynomial
equation P (z ) = 0 has exactly n roots.
 C-122

Example 6: Prove that all the roots of z 7 − 5 z 3 + 12 = 0 lie between the circles | z | = 1 and
| z | = 2. (Kanpur 2008; Gorakhpur 14)

Solution: Let C1 represent the circle | z | = 1 and C2 represent the circle | z | = 2.

Suppose f (z ) = 12 and g (z ) = z 7 − 5 z 3 .

We observe that both f (z ) and g (z ) are analytic within and on C1.

7 3
 g (z )   z 7 − 5 z 3  | z | + | − 5 z | | z |7 + 5 | z |3 1+ 5 1
Now   = ≤ = = = ,
 f (z )  12  12 12 12 2

since | z | = 1 on C.

 g (z ) 
∴  < 1 or | g (z )| < | f (z )| on C1.
 f (z )

Hence by Rouche’s theorem f (z ) + g (z ) = z 7 − 5 z 3 + 12 has the same number of zeros

inside C1 as f (z ) = 12. Since f (z ) = 12 has no zeros inside C1 therefore

7 3
f (z ) + g (z ) = z − 5 z + 12 has no zeros inside C1.

Now consider the circle C2 . Take F (z ) = z 7, φ (z ) = 12 − 5 z 3 . We observe that both

F (z ) and φ (z ) are analytic within and on C2 . We have

 φ (z )  |12 − 5 z 3 | |12 | + 5 | z |3 12 + 5 . 23 52
 = 7
≤ 7
= 7
= < 1,
 F (z )  |z| |z| 2 128

since | z | = 2 on C2 .
Thus on C2 ,| φ (z )| < | F (z )|. Hence by Rouche’s theorem F (z ) + φ (z ) = z 7 − 5 z 3 + 12
has the same number of zeros as F (z ) = z 7 inside C2 . Since F (z ) = z 7 has all the seven
zeros inside the circle| z | = 2 as they are all located at the origin therefore all the seven
zeros of z 7 − 5 z 3 + 12 lie inside the circle C2 .

Hence all the roots of the equation z 7 − 5 z 3 − 12 = 0 lie between the circles| z | = 1and
| z | = 2.

Example 7: Use Rouche’s theorem to show that the equation z 5 + 15 z + 1 = 0 has one root in the
3 3
disc | z | < and four roots in the annulus < | z | < 2.
2 2 (Kanpur 2007)
5
Solution: Let f (z ) = z and g (z ) = 15 z + 1.

Now on the circle | z | = 2, we have

| f (z )| = | z |5 = 25 = 32
 C-123

and | g (z )| = |15 z + 1| ≤ 15 | z | + 1 = 31.

∴ | g (z )| < | f (z )| on the circle | z | = 2.
Hence by Rouche’s theorem the function
f (z ) + g (z ) = z 5 + 15 z + 1

has as many zeros in| z | < 2 as the function f (z ) = z 5 . Since the function f (z ) has a zero
of order 5 at z = 0 therefore all the five roots of z 5 + 15 z + 1 = 0 must lie inside the disc
| z | < 2.
3
Again for | z | = , we have
2
243 45
| z 5 + 1| ≤ | z |5 + 1 =
+ 1< = |15 z |.
32 2
3
The function z 5 + 15 z + 1has as many zeros in| z | < as the function15z. Since15z has
2
exactly one zero in this region, so does z 5 + 15 z + 1. Hence four of the zeros of
3
z 5 + 15 z + 1 must lie in the ring < | z | < 2.
2

Comprehensive Exercise 2

1. If a > e, use Rouche’s theorem to prove that the equation

e z = az n
has n roots inside the circle | z | = 1. (Kanpur 2008)

2. Using Rouche’s theorem determine the number of zeros of the polynomial

P (z ) = z10 − 6 z 7 + 3 z 3 + 1 in | z | < 1 .

3. Apply Rouche’s theorem to determine the number of roots of the equation

z 8 − 4z 5 + z 2 − 1 = 0
that lie inside the circle | z | = 1.

4. Show that the polynomial z 5 + z 3 + 2 z + 3 has just one zero in the first
quadrant of the complex plane.

A nswers 2

2. seven zeros
3. five roots
 C-124

Objective Type Questions

Multiple Choice Questions

Indicate the correct answer for each question by writing the corresponding letter from (a),
(b), (c) and (d).
1. If a function is analytic at all points of a bounded domain except at finitely many
points, then these exceptional points are called
(a) zeros (b) singularities
(c) poles (d) simple points.
2. A function which has poles as its only singularities in the finite part of the plane is
said to be
(a) an analytic function (b) an entire function
(c) a meromorphic function (d) none of these.
3. For the function f (z ) = e z , z = ∞ is
(a) isolated essential singularity (b) pole
(c) ordinary point (d) none of these.
1
4. Number of poles of the function f (z ) = tan is
z
(a) 2 (b) 4
(c) infinite (d) none of these.
1
5. Number of zeros of the function f (z ) = sin is
z
(a) 3 (b) 4
(c) infinite (d) none of these.
z +3
6. The number of isolated singular points of f (z ) = 2 2 is
z (z + 2)
(a) 1 (b) 2
(c) 3 (d) 4.
f ′ (z )
7. If f (z ) = z 5 − 3 iz 2 + 2 z + i − 1 and C encloses zeros of f (z ), then ∫ dz
C f (z )
is
(a) 5πi (b) 0
(c) 10 πi (d) none of these.

A nswers
1. (b) 2. (c) 3. (a) 4. (c) 5. (c)
6. (c) 7. (c)

¨
 1 Residue at a Pole
efinition: Let z = a be a pole of order m of a single-valued function f (z ) and γ be
D any circle of radius r and centre z = a containing no other singularities except
z = a. Then the function f (z ) is regular within the region 0 < | z − a | < r so we can
expand f (z ) in a Laurent’s series in the region 0 < | z − a | < r.
∞ ∞
∴ f (z ) = Σ an (z − a)n + Σ bn (z − z0 )− n
n=0 n =1
1 f (z )
2 πi ∫ γ (z − a)n + 1
where an = dz

1
(z − a)n − 1 f (z ) dz .
2 πi ∫ γ
and bn =

1
2 πi ∫ γ
In particular, b1 = f (z ) dz .

The coefficient b1 is called the residue of f (z ) at z = a.

 C-126

2 Computation of Residue at a Finite Pole

(i) Residue at a simple pole: If z = a is a simple pole of f (z ) then the Laurent’s
expansion of f (z ) is of the form
∞ b1
f (z ) = Σ an (z − a)n + ⋅
n=0 z−a
lim
∴ b1 = (z − a) f (z ).
z→a
Hence the residue at the simple pole z = a is given by
lim ( z − a) f ( z)
z→ a
Another form is obtained as follows :
φ (z )
If f (z ) = where ψ (z ) = (z − a) F (z ), F (a) ≠ 0 , then
ψ (z )
lim lim φ (z )
(z − a) f (z ) = (z − a)
z→a z→a ψ (z )

lim (z − a) [ φ (a) + (z − a) φ′ (a) + ……]

= , by Taylor’s theorem
z→a ψ (a) + (z − a) ψ ′ (a) + ……

lim φ (a) + (z − a) φ′ (a) + ……

= , since ψ (a) = 0
z→a ψ ′ (a) + ……
φ (a)
= ⋅
ψ ′ (a)
φ (a)
Hence the residue at the simple pole z = a is ⋅
ψ ′ (a)
(ii) Residue at a pole of order greater than unity :
(Residue at a pole of order m) (Gorakhpur 2014)
Let z = a be a pole of order m of f (z ) and suppose
φ (z )
f (z ) = ⋅
(z − a)m
We then have, by the definition of a pole,
φ (z ) m Mr
f (z ) = m
= ψ (z ) + Σ ,
(z − a) r =1 (z − a)r
where ψ (z ) is regular at z = a,
φ (z ) M1 M2 Mm
or = ψ (z ) + + +…+ ⋅
(z − a)m (z − a) (z − a)2 (z − a)m

∴ φ (z ) = ψ (z ) (z − a)m + M1 (z − a)m − 1 + M2 (z − a)m − 2 + .... M m

Differentiating both the sides with respect to z , (m − 1) times, we have

 C-127

φ(m − 1) (z ) = ψ(m − 1) (z ) (z − a)m + (m − 1) ψ m − 2 (z ) . m (z − a) m − 1

(m − 1) (m − 2) (m − 3)
+ .ψ . (z ) m (m − 1) . (z − a)m − 2 + …
2!
m!
+ ψ (z ) (z − a) + M1 (m − 1) !
1!
∴ φ(m − 1) (a) = M1 (m − 1) !.

φ(m − 1) (a)
Hence M1 = which is the required residue at z = a.
(m − 1)!
φ (z ) φ (z )
In particular if f (z ) = 2
, the residue at z = a is φ′ (a) and if f (z ) = , the
(z − a) (z − a)3
φ′ ′ (a)
residue at z = a is etc.
2!
Alternative Proof (a): Suppose z = a is a pole of order m. Then f (z ) is of the form
φ (z )
where φ (z ) is analytic.
(z − a)m

Residue of f (z ) at z = a is given by
1 1 φ (z )
2 πi ∫ γ 2πi ∫ γ (z − a)m
b1 = f (z ) dz = dz

φ m −1 (a)
= , by Cauchy’s integral formula.
(m − 1)!
φ m − 1 ( a)
Hence the residue of f (z ) at the pole of order m is , where z = a is the pole of order m.
( m − 1) !
Alternative Proof (b): If z = a is the pole of order m of f (z ) then we have
φ (z )
f (z ) = where φ (z ) is analytic at z = a.
(z − a)m

The residue at z = a is the coefficient of (z − a)−1 in f (z )

= coeff. of (z − a)m −1 in φ (z )

= coeff. of (z − a)m −1 in [ φ (a) + (z − a) φ′ (a) + ……

(z − a)m − 1 m − 1
+ φ (a) + ……]
(m − 1)!
φ m − 1 (a)
= ⋅
(m − 1)!

Remark: We have seen that the residue of f (z ) at the pole z = a is the coefficient of
1
in the Laurent’s expansion of f (z ). If we put z − a = t or z = a + t, where t is small
(z − a)
then the Laurent’s expansion of f (z ) becomes
∞ b1 b2 b
n
f (a + t) = Σ an t + + 2 + …… + m ⋅
n=0 t t tm
 C-128

We see that b1 is the coefficient of 1 / t in the above expansion.

Thus to find the residue of f (z ) at z = a, put z = a + t in f (z ) and expand in powers of t, where t
is small, the coefficient of 1 / t is the residue at z = a.

3 Residue at Infinity
Definition: If the function f (z ) has an isolated singularity at infinity or is analytic there then
the residue of f (z ) at z = ∞ is given by
1
2πi ∫ C
f (z ) dz

where C is a large circle containing all the finite singularities of f (z ) and integral along C is
performed in a clockwise direction provided that this integral has a definite value.
If the integral along C is taken in anti-clockwise direction the residue at infinity is
1
2πi ∫ C
− f (z ) dz .

4 Computation of Residue at Infinity

1
2πi ∫ C
Method I: By definition the residue of f (z ) at z = ∞ is f (z ) dz taken in

clockwise direction round a large circle C which encloses in its interior all other
singularities.

Therefore the residue of f   at w = 0 is given by

1
 w
1  dw 
f    − 2  = − 2 f   dw,
1 1 1 1
2 πi ∫ γ  w   w  2 πi ∫ γ w  w 

taken in anti-clockwise direction around a small circle γ with centre at origin.

Hence the residue of f ( z) at infinity
 − w f (1 / w) 
= lim   = zlim [− z f ( c)] ,
w→0 w2 → ∞

provided the limit exists.

Method II: Suppose f (z ) has a pole of order m at infinity. Then f (1 / w) has a pole of
order m at w = 0.
By Laurent’s theorem the expansion of f (1 / w) at w = 0 is given by
∞
f   =
1 m bn
Σ an w n + Σ ⋅
 w n=0 n =1 wn
Therefore the expansion of f (z ) in the neighbourhood of z = ∞ is given by
m ∞
f (z ) = Σ bn z n + Σ an z − n .
n =1 n=0
 C-129

m ∞
Now ∫C f (z ) dz = ∫C Σ bn z n dz + ∫C Σ an z − n dz
n =1 n=0
m ∞
= Σ ∫C bn z n dz + Σ ∫ an z − n dz
n =1 n=0 C

a1
= ∫C z
dz , all other integrals vanish since
dz
each of them is of the form ∫C zk
, k ≠1

∵ dz
= 2πi
 ∫ C
= a1 ⋅ 2πi
z 
1
2 πi ∫ C
Thus the residue at infinity = − f (z ) dz = − a1, which is the coefficient of 1 / z

with sign changed in the expansion of f (z ) in the neighbourhood of z = ∞.

1
Hence the residue of f ( z) at infinity is the negative of the coefficient of in the
z
expansion of f ( z) in the neighbourhood of z = ∞.
An Important Observation: We can easily show that the residue of a function at a
finite point is zero if the function is analytic there. On the other hand a function may be
analytic at z = ∞ but yet has a residue there.
1
Consider f (z ) = ⋅ The only singularity of f (z ) is a simple pole at z = a. The
z−a
function is analytic at z = ∞.
1 dz
Now the residue at infinity = −
2πi ∫ C z−a

1 2 π rie i θ
dθ, putting z − a = re i θ
2 πi ∫0
=−
re i θ
1 2π
2 π ∫0
=− dθ = − 1.

5 Cauchy’s Residue Theorem

If f (z ) is regular, except at a finite number of poles within a closed contour C and continuous on the
boundary of C, then

∫C f (z ) dz = 2πi Σ R ,

where Σ R is the sum of the residues of f (z ) at its poles within C.

(Meerut 2001; Kanpur 01; Gorakhpur 06, 11; Rohilkhand 08, 12)

Proof:Let z1, z2 ,……, z n be the n poles within the closed contour C. Let γ1, γ 2 ,……, γ n
be the circles with centres z1, z2 ,……, z n respectively and each of radius r so small that
all the circles lie entirely within C and do not overlap. Then f (z ) is analytic in the region
lying between C and the circles. Then by Cauchy’s theorem
 C-130

n
∫C f (z ) dz − Σ
k =1 γ k∫ f (z ) dz = 0

or ∫C f (z ) dz = ∫γ1 f (z ) dz

+ ∫γ2 f (z ) dz + …… + ∫γ n f (z ) dz .
…(1)

Suppose f (z ) has a pole of order m1 at z = z1

then we have
m1
br
f (z ) = φ1 (z ) + Σ , where φ1 is analytic within and on γ1.
k =1 (z − z1)r
b1
Now ∫γ1 f (z ) dz = ∫γ1 φ1 (z ) dz + ∫γ1 z − z1
dz + ……

bm
+ ∫γ1 1
(z − z1)m1
dz . …(2)

We have ∫γ1 φ1 (z ) dz = 0 since φ1 is analytic within and on γ1.

b1 2π b1 rie i θ
Also ∫γ1 dz = ∫0 dθ, putting z − z1 = r e i θ
z − z1 r e iθ
2π
= ∫0 b1 i dθ = 2 π i b1

bm 2π bm r i e i θ
and ∫γ1 1
dz = ∫0 1
dθ, putting z − z1 = re i θ
(z − z1)m1 r m1 e i m1 θ
i bm 2π
= m1 −1
1
∫0 e − i (m1 −1) θ dθ = 0 , m1 ≠ 1.
r
Substituting these values in (2), we get

∫γ1 f (z ) dz = 2 π i b1 = 2 π i × residue of f (z ) at z = z1.

Proceeding as above, we have

∫γ2 f (z ) dz = 2 π i × residue of f (z ) at z = z2 and so on.

Hence from (1), we have

∫C f (z ) dz = 2πi (sum of the residues at z1, z2 ,……, z n)

= 2π i Σ R.
Corollary: If an analytic function has singularities at a finite number of points (including that
at infinity), then the sum of the residues at these points along with infinity is zero.
Let C be the circle enclosing within it all the singularities excluding infinity. Then by
the previous theorem, we have
 C-131

1
2 π i ∫C
f (z ) dz = sum of the residues at all the finite

singular points within C,

1
2 π i ∫C
also the residue at infinity is − f (z ) dz .

Hence the sum of the residues at all the finite poles along with infinity is zero.

z3
Example 1: Find the residue of at z = 1.
(z − 1)4 (z − 2) (z − 3) (Rohilkhand 2009)
3
z
Solution: Let f (z ) = ⋅
(z − 1)4 (z − 2) (z − 3)

z = 1 is the pole of order 4 of f (z ).

To find the residue at z = 1we shall put z = 1 + t in f (z ) then the coefficient of1 / t will be
the residue at z = 1.
z3 (1 + t)3
Now f (z ) = 4
= 4
, putting z − 1 = t
(z − 1) (z − 2) (z − 3) t (t − 1)(t − 2)
1 1 −1
= (1 + t)3 (1 − t)−1 (1 − t)
2t 4 2
1
= (1 + 3 t + 3 t2 + t3 )(1 + t + t2 + t3 + ……)
2t 4 1 1 1
(1 + t + t2 + t3 + ……)
2 4 8
1 2 3 3 7 2 15 3
= (1 + 3 t + 3 t + t )(1 + t + t + t + ……).
2t 4 2 4 8
The coefficient of 1 / t in the above expansion

= 
1 15 21 9  101
+ + + 1 =
28 4 2  16
which is the residue at z = 1.

z2
Example 2: Determine the poles of the function f (z ) =
(z − 1)2 (z + 2)
and the residue at each point.

Hence evaluate ∫C f (z ) dz where C is the circle | z | = 2 ⋅ 5.

z2 1 z2
Solution: We have f (z ) = = φ (z ), where φ (z ) = ⋅
(z − 1)2 (z + 2) (z − 1)2 z +2

Here z = 1 is a pole of order 2 of f (z ) and z = − 2 is a simple pole.

 C-132

Now residue at z = 1 is
1  d  z2    z 2 + 4z  5
[φ′ (z )]z = 1 =    = 2
= ⋅
1!  dz  z + 2   z =1 (z + 2)  z =1
9

lim lim z2 4
Residue at z = − 2 is (z + 2) f (z ) = = ⋅
z→ −2 z → − 2 (z − 1)2 9

The function f (z ) is analytic on| z | = 2 ⋅ 5 and at all points inside it except at z = 1, − 2

therefore by residue theorem, we have

∫C f (z ) dz = 2πi [residue at (z = − 2) + residue at (z = 1)]

= 2 πi  +  = 2 πi.
4 5
 9 9

z 2 − 2z
Example 3: Find the residues of at all its poles in the finite plane.
(z + 1)2 (z 2 + 4)

z 2 − 2z z 2 − 2z
Solution: Here f (z ) = = ⋅
(z + 1)2 (z 2 + 4) (z + 1)2 (z + 2 i) (z − 2 i)

Poles of f (z ) are given by (z + 1)2 (z + 2 i) (z − 2 i) = 0 .

f (z ) has a double pole at z = − 1 and simple poles at z = 2 i, − 2 i.

Residue at z = 2 i is
lim lim z 2 − 2z
(z − 2 i) f (z ) =
z → 2i z → 2 i (z + 1)2 (z + 2 i)

(2 i)2 − 2 . 2 i 7+i
= 2
= ⋅
(2 i + 1) (2 i + 2 i) 25

Residue at z = − 2 i is
lim lim z 2 − 2z
(z + 2 i) f (z ) =
z → − 2i z → − 2 i (z + 1)2 (z − 2 i)

(− 2 i)2 − 2 . (− 2 i) 7−i
= 2
= ⋅
(− 2 i + 1) (− 2 i − 2 i) 25

1d z 2 − 2z
Residue at z = − 1 is φ (z ) , where φ (z ) = is
1!  dz  z = −1 z2 + 4

 d  z 2 − 2z   2 z 2 + 8 z − 8  14
=  2  = 2 2 
=− ⋅
 dz  z + 4   z = −1  (z + 4)  z = −1
25

Example 4: Find the residues of e z cosec 2 z at all its poles in the finite plane.

ez
Solution: Let f (z ) = e z cosec2 z = ⋅
sin2 z
 C-133

The poles of f (z ) are given by sin2 z = 0

or z = mπ, m ∈ I are the poles of f (z ) of order 2.

The limit point of these poles is z = ∞ which is therefore a non-isolated essential
singularity.
Putting z = mπ + t in f (z ), we get
e mπ + t 1
f (mπ + t) = 2
= e mπ e t . 2
sin (mπ + t) sin (mπ + t)

e mπ 1 + t + t + ....
1 2
 2 ! 
= 2
 t − 1 t3 + 1 t5 − ......
 
 3! 5! 
−2
1  t2   t 2 t 4 
= e mπ 2
1 + t + + ..... 1 −  − + .... 
t  2!    3! 5!  

1  t2   t 2 t 4 
= e mπ 2
1 + t + + ..... 1 + 2  − + ....
t  2!    3 ! 5 ! 

t 2 t 4 
2 
+3 − + .... + ......
 3! 5!  

1
Now residue at z = mπ is the coeff. of in the above expansion = e mπ .
t

sin πz 2 + cos πz 2
Example 5: Evaluate ∫ dz , where C is the circle | z | = 3.
C (z − 1)2 (z − 2)

sin πz 2 + cos πz 2
Solution: We have f (z ) = ⋅
(z − 1)2 (z − 2)
The function f (z ) is analytic at every point within C except at the poles z = 1, 2.
lim lim sin πz 2 + cos πz 2
Residue at z = 2 is (z − 2) f (z ) = = 1.
z→2 z→2 (z − 1)2
Residue at z = 1 is
1d  d  sin πz 2 + cos πz 2  
(z − 1)2 f (z ) =   = 2π + 1 .

1!  dz  z −2
z =1  dz    z =1

∴ By residue theorem, we have

∫C f (z ) dz = 2πi [sum of residues at z = 1 and z = 2]

= 2 πi [1 + (2 π + 1)]
= 4 πi (π + 1).
 C-134

Comprehensive Exercise 1

1
1. Find the residue of at z = i.
(z + 1)3
2

1
2. Find the residue of at z = ia.
(z 2 + a2 )2
cot πz
3. Find the residues of the function ⋅
(z − a)2
z4
4. Find the residues of the function ⋅
(c + z 2 )4
2

A nswers 1
3i −i
1. − 2.
16 4 a3
1 i
3. 4. −
π (n − a)2 32 c 3

6 Evaluation of Real Definite Integrals by

Contour Integration
Now we shall evaluate the real definite integrals with the help of contour integration
and Cauchy’s residue theorem by properly choosing the integrand and the contour. It
should be noted that a large number of real definite integrals whose evaluation by usual
methods is difficult can be easily evaluated by using Cauchy’s residue theorem, yet
there are many integrals which cannot be evaluated by contour integration. Before
discussing the procedure of the evaluation of definite integrals we are going to prove
two useful theorems :
lim
Theorem 1: If (z − a) f (z ) = A and if C is the arc θ1 ≤ θ≤ θ2 of the circle| z − a | = r,
z→a
lim
then
r→0 ∫C f (z ) dz = i A (θ2 − θ1).

In particular, if (z − a) f (z ) → 0 as z → 0, then we have

∫C f (z ) dz → 0 as z → 0.
lim
Theorem 2: If C is an arc θ1 ≤ θ ≤ θ2 of the circle| z | = R and if z f (z ) = A then
R→ ∞
lim
R→ ∞ ∫C f (z ) dz = i (θ2 − θ1) A.
 C-135

7 Integration Round the Unit Circle

(Gorakhpur 2016)
We shall discuss here the method of evaluation, by contour integration, of the integrals
which are of the type
2π
∫0 f (cos θ, sin θ) dθ

where integrand is a rational function of cos θ and sin θ.

Putting z = e i θ , we have dz = i e i θ dθ
1 1 1  1
and cos θ =  z +  and sin θ = z −  ⋅
2 z 2i  z
2π 1 1  dz
f   z +  ,  z −  
1 1 1
Now ∫0 f (cos θ, sin θ) dθ =
i ∫ C 2  
z 2i  z  z

= ∫C φ (z ) dz , say,
where C is the circle | z | = 1.
It is obvious that F (z ) is a rational function of z.
Thus by residue theorem we have

∫C φ (z ) dz = 2πi Σ RC ,

where Σ RC is the sum of the residues of φ (z ) at its poles inside C.

Example 6: Show that

2π dθ 2π dθ 2π
∫0 a + b cos θ
= ∫0 =
a + b sin θ √ (a2 − b2 )
, a > b > 0.

2π dθ
Solution: (i) Let I = ∫0 a + b cos θ
2π dθ 1 dz
= ∫0 1 iθ −i θ
= ∫
1
,
) i
C
a + b (e + e z {a + b (z + 1 / z )}
2 2
putting e i θ = z , i e i θ dθ = dz
2 dz
= ∫ 2
, where C is the unit circle | z | = 1.
i C bz + 2 az + b
2
The poles of the integrand f (z ) = 2
are given by
i (bz + 2 az + b)
bz 2 + 2 az + b = 0
− 2 a ± √ (4 a2 − 4 b2 ) − a ± √ (a2 − b2 )
or z= = ⋅
2b b
 C-136

− a + √ (a2 − b2 ) − a − √ (a2 − b2 )
Let α= and β = ⋅
b b
Since a > b > 0 therefore | β | > 1. Also | αβ | = 1 so we have | α | < 1.
Thus z = α is the simple pole lying inside C.
Since α, β are the roots of bz 2 + 2 az + b = 0 , therefore we have
2
f (z ) = ⋅
i b (z − α)(z − β)
Residue of f (z ) at the simple pole z = α is
lim 2 lim 2
= (z − α) =
z→α i b (z − α)(z − β) z → α b i (z − β)
2 2 1
= = = ⋅
b i (α − β) 2
2 √ (a − b ) 2
i √ (a − b2 )
2
bi
b
2π dθ
Hence, ∫0 a + b cos θ
= 2 πi. sum of the residues of f (z ) at the poles inside C

1 2π
= 2 πi ⋅ = ⋅
i √ (a2 − b2 ) √ (a2 − b2 )
2π dθ
Similarly, we can evaluate ∫0 a + b sin θ
⋅

2π a cos θ
Example 7: Evaluate ∫ dθ, a > 1
−π a + cos θ

by putting e i θ = z and using the theory of residues.

π a cos θ π a cos θ 2π 2 a cos θ
Solution: Let I = ∫− π a + cos θ
dθ = 2 ∫
0 a + cos θ
dθ = ∫0 2 a + 2 cos θ
dθ

2π 2a e i θ
= real part of ∫0 2 a + 2 cos θ
dθ

2 az dz
= real part of ∫C ⋅
1 iz
2a + z +
z
2 az
= real part of ∫C i (z 2 + 2 az + 1)
dz

= real part of ∫C f (z ) dz , say, ...(1)

where C is the unit circle | z | = 1.

Solving z 2 + 2 az + 1 = 0 , we get
z = − a + √ (a2 − 1) = α , z = − a − √ (a2 − 1) = β,
which are the simple poles of f (z ).
Since | β | > 1 and | αβ | = 1 therefore | α | < 1. Thus z = α is the simple pole inside C.
 C-137

lim
Residue at z = α is (z − α) f (z )
z→α
lim (z − α) 2 az lim 2 az 2 aα
= = =
z → α i (z − α)(z − β) z → α i (z − β) i (α − β)

2 a {− a + √ (a2 − 1)}  a 
= 2
= ai  2
− 1 ⋅
2 i √ (a − 1)  √ (a − 1) 
∴ by Cauchy’s residue theorem, we have
 a   a 
∫C f (z ) dz = 2 πi . ai  2
− 1 = 2 aπ 1 − 2 ⋅
 √ (a − 1)   √ (a − 1)
π a cos θ
Hence, ∫− π a + cos θ
dθ = real part of ∫ f (z ) dz , from (1)
C

 a 
= 2 aπ  1 − 2 ⋅
 √ (a − 1)

2π cos2 3θ 1 − p + p2
Example 8: Prove that ∫0 1 − 2 p cos 2θ + p 2
dθ = π
1− p
, 0 < p < 1.

2π cos2 3θ
Solution: We have I = ∫0 1 − 2 p cos 2θ + p2
dθ

1 2π 1 + cos 6θ
=
2 ∫0 1 − 2 p cos 2θ + p2
dθ

1 2π 1 + e i6 θ
=
2
real part of ∫0 1 − 2 p cos 2θ + p2
dθ

1 1 + z6 dz
= real part of ∫ ⋅ , putting z = e i θ
2 C  2 1 
1 − p z + 2  + p2 iz
 z 
1 z (1 + z 6 )
= real part of ∫ dz
2 C i (1 − pz 2 )(z 2 − p)

1
= real part of ∫ f (z ) dz , say,
2 C

where C is the unit circle.

z = ± √ p, ± 1 / √ p are the simple poles of f (z ).
Since 0 < p < 1 therefore z = √ p, − √ p are the only simple poles which lie within C.
Residue at z = √ p is
lim (z − √ p) z (1 + z 6 ) 1 1 + p3 
(z − √ p) f (z ) = lim =  
z→√p 2 2 2
z → √ p i (1 − pz ) (z − p) 2 i 1 − p 
and residue at z = − √ p is
 C-138

lim lim z (1 + z 6 )
(z + √ p) f (z ) = (z + √ p)
z→−√p z→−√p i (1 − pz 2 )(z 2 − p)

1 1 + p3
= ⋅ ⋅
2 i (1 − p2 )

∴ Sum of the residues at the poles inside C

1 1 + p3 
=  ⋅
i 1 − p2 

2π 1+ e i6 θ 1 1 + p
3  1 + p3 
 ⋅
Hence, ∫0 1 − 2 p cos 2θ + p2
= 2 π i . 
i  1 − p2
 = 2 π
 1 − p2 
 
2π cos2 3θ 1 2π 1+ e i6 θ
∴ ∫0 1 − 2 p cos 2θ + p2
dθ =
2
real part of ∫0 1 − 2 p cos 2θ + p2
dθ

1 + p3  (1 + p + p2 )
= π   = π ⋅
 1− p
2
 1− p
π a dθ π
Example 9: (i) Prove that ∫ 2 2
= ,a>0
0 a + sin θ √ (1 + a2 ) (Avadh 2007)
2π adθ 2π
(ii) ∫0 a2 + sin2 θ
=
2 1 + a2
, a > 0.

π a dθ π 2 a dθ
Solution: (i) We have I = ∫0 2
a + sin θ 2
= ∫0 2 a2 + 2 sin2 θ
π 2 a dθ 2π a dt
= ∫0 2
2 a + 1 − cos 2θ
= ∫0 2
2 a + 1 − cos t
, putting 2θ = t

a dz 2 a dz
= ∫C 1
⋅ = ∫ 2 z (2 a + 1) − z 2 − 1
2
2 a2 + 1 − (z + 1 / z ) iz i C
2
dz
= 2 ai ∫
C z 2 − 2 (2 a2 + 1) z + 1 ∫ C
= f (z ) dz , say,

where C is the unit circle.

Solving z 2 − 2 (2 a2 + 1) z + 1 = 0 , we get

z = (2 a2 + 1) ± 2 a √ (a2 + 1)

or z = (2 a2 + 1) + 2 a √ (a2 + 1) = α,
z = (2 a2 + 1) − 2 a √ (a2 + 1) = β.
Thus z = α, β are the simple poles of f (z ).
Since a > 0 therefore | α | > 1. Also we have | αβ | = 1 therefore | β | < 1.
Thus z = β is the only simple pole inside C.
Residue at z = β is
 C-139

lim lim 2 ai 2 ai
(z − β) f (z ) = (z − β) =
z→β z→β (z − α)(z − β) β − α
2 ai i
=− 2
=− ⋅
4 a √ (a + 1) 2 √ (a2 + 1)
π a dθ
Hence, ∫0 a + sin2 θ
2
= 2 πi. sum of the residues inside C

 i  π
= 2 πi − 2 = 2
⋅
 2 √ (a + 1) √ (a + 1)
2π a dθ π a dθ 2π
(ii) ∫0 2
a + sin θ 2
= 2∫
0 2
a + sin θ 2
=
2 √ (1 + a2 )
⋅

Example 10: By the method of contour integration, prove that

2π 2π 2π
∫0 e cos θ cos (nθ − sin θ) dθ = ∫0 e cos θ cos (sin θ − nθ) dθ = ,
n!
where n is a positive integer. (Gorakhpur 2005; Kanpur 07, 08)

Solution: We have
2π 2π
∫0 e cos θ cos (nθ − sin θ) dθ = ∫0 e cos θ cos (sin θ − nθ) dθ.

[ ∵ cos (− θ) = cos θ]
2π
Now consider I = ∫0 e cos θ e i (sin θ − nθ) dθ
2π
= ∫0 e cos θ + i sin θ e − i n θ dθ
2π iθ
= ∫0 ee e − i n θ dθ

−n dz
= ∫C ez z . , putting z = e i θ
iz
ez
= ∫C i z n +1
dz

= ∫C f (z ) dz , say,

where C is the unit circle.

The function f (z ) has a pole of order (n + 1) at z = 0.
1  n ez  1
The residue at z = 0 is D  = ⋅
n!  i  in !
z =0

1 2π
Hence I = 2 πi ⋅ = ⋅
in ! n !
Equating real parts on both sides, we get
2π 2π
∫0 e cos θ cos (sin θ − nθ) dθ = ⋅
n!
 C-140

Comprehensive Exercise 2

π 1 + 2 cos θ
1. Prove that ∫ dθ = 0 .
0 5 + 4 cos θ (Kanpur 2007, Rohilkhand 12)

2π dθ 2π 2 π dθ π
2. (i) Show that ∫ = ,∫ = ⋅
0 2 + cos θ √ 3 0 5 + 3 cos θ 2
2π dθ 2π
(ii) Show that ∫ = , a2 < 1.
0 1 + a cos θ √ (1 − a2 )
π dθ 1 2π dθ π
(iii) Prove that ∫ = ∫0 = ⋅
0 a + b cos θ 2 a + b cos θ √ (a2 − b2 )

3. Use the method of contour integration to prove that

2π dθ 2π
∫0 1 + a2 − 2a cos θ = 1 − a2 , 0 < a < 1.
4. Apply the method of contour integration to prove that
2π cos 2θ π
∫0 5 + 4 cos θ dθ = 6 ⋅
5. By the method of contour integration prove that
π cos 2θ πa2
∫0 1 − 2 a cos θ + a2
dθ =
1 − a2
, (− 1 < a < 1).
(Gorakhpur 2006)
2π sin nθ
6. Evaluate ∫0 1 + 2 a cos θ + a2
dθ
(Rohilkhand 2010)
2π cos nθ
and ∫0 1 + 2 a cos θ + a2
dθ,

a2 < 1 and n is a positive integer.

π
7. Show that ∫ tan (θ + ia) dθ = iπ, where R (a) > 0 .
0
2π 2π
8. Prove that ∫ e − cos θ cos (nθ + sin θ) dθ = (− 1)n ,
0 n!
where n is a positive integer.

A nswers 2

2 π (− 1)n a n
6. ;0
1 − a2
 C-141

8 Evaluation of the Integral ∫−∞∞ f ( x ) dx.

Theorem: If the function f (z ) is analytic in the upper half of the z-plane except at a finite
number of poles in it, having no poles on the real axis and if further z f (z ) tends to zero as| z |tends
to infinity then by contour integration
∞
∫ −∞ f ( x) dx = 2πi Σ R +

where Σ R + represents the sum of the residues at the poles in the upper half plane.

Proof: Under the given conditions the integral

∞
∫− ∞ f ( x) dx is convergent. To evaluate such integrals

we shall integrate f (z ) round a contour C consisting of

a semi-circle Γ of radius R large enough to include all
the poles of f (z ) and the part of the real axis from
x = − R to x = R. The only singularities of f (z ) in the
upper half plane are poles.
∴ By Cauchy’s residue theorem, we have

∫C f (z ) dz = 2πi Σ R +

R
or ∫− R f ( x) dx + ∫Γ f (z ) dz = 2πiΣ R +
…(1)
where Σ R + represents the sum of the residues of f (z ) at the poles in the upper half
plane.
Since z f (z ) → 0 as | z | → ∞, therefore we have

∫Γ f (z ) dz = 0 . [ By theorem 2 of article 6]

lim R
Also
R→ ∞ ∫− R f ( x) dx

∞
=P ∫− ∞ f ( x) dx, where P stands for principal value of the integral

∞
= ∫− ∞ f ( x) dx, since the integral is convergent.

Now taking limit of both sides of (1) when R → ∞, we get

∞
∫− ∞ f ( x) dx = 2πi Σ R + .

Corollary: If the function f (z ) is of the form P (z ) / Q (z ) where P (z ) and Q (z ) are

both polynomials such that (i) Q (z ) = 0 has no real roots (ii) degree of P (z ) is at least
two less than that of Q (z ) so that z f (z ) → 0 as | z | → ∞ then
∞
∫− ∞ f (z ) dz = 2πi Σ R+ .
 C-142

∞ dx π
Example 11: (i) If a > 0, prove that ∫0 ( x2 + a2 )2
=
4 a3
⋅
∞ dx π
(ii) Prove that ∫ 2 2
= ⋅
0 (1 + x ) 4

Solution: (i) Consider

dz
∫C f (z ) dz = ∫C (a + z 2 )2
2
,

where C is the contour consisting of a large semi-circle

Γ of radius R together with real axis from − R to R.
z = ai, − ai are the poles of f (z ) of second order. Out of
these only z = ai lies inside C.
Residue at z = ai is
1
φ′ (ai) = [(d / dz ) (z + ai)−2 ]z = ai = [− 2 (z + ai)−3 ]z = ai = ⋅
4 a3 i
By Cauchy’s residue theorem, we have
R dx dz
∫ C f (z ) dz = ∫ − R (a2 + x2 )2 + ∫Γ = 2 πi Σ R + . …(1)
(a + z 2 )2
2

lim lim z
Now z f (z ) = = 0,
R→ ∞ z → ∞ (a2 + z 2 )2
lim dz
∴
R→ ∞ ∫Γ (a2 + z 2 )2
=0

lim R dx ∞ dx
and
R→ ∞ ∫− R 2
(a + x ) 2 2
= ∫− ∞ (a + x2 )2
2
⋅

Hence we have from relation (1)

∞ dx ∞ dx 1 π
∫ − ∞ (a2 + x2 )2 = 2 ∫ 0 (a2 + x2 )2 = 2πi ⋅ 4a3 i = 2a3
∞ dx π
or ∫0 (a2 + x2 )2
=
4 a3
⋅

(ii) Proceed as in part (i) taking a = 1.

∞ x2 π
Example 12: Prove that ∫ 2 2 3
dx = , provided that R (a) is +ve. What is the
−∞ (x + a ) 8 a3
value of this integral when R (a) is negative ?
z2
Solution: Consider the integral ∫ f (z ) dz = ∫C dz ,
C (z 2 + a2 )3

where C is the contour as described in Ex. 11.

By Cauchy’s residue theorem, we have
 C-143

R x2 z2
∫C f (z ) dz = ∫− R 2 2 3
dx + ∫Γ dz = 2 πi Σ R+ . …(1)
(x + a ) (z + a2 )3
2

lim lim z2
Since z f (z ) = z⋅ = 0 , therefore we have
R→ ∞ z→∞ (z 2 + a2 )3
lim z2
R→ ∞ ∫Γ (z + a2 )3
2
dz = 0 .

Also z = ± ai are the poles of f (z ) of order three. Out of these only z = ai lies inside C.
1  2 2 z2  1
Residue at z = ai is (d / dz )  = ⋅
2!  (z + ai)3  z = ai
16 a3 i

Now from (1), we have

∞ z2 1 π
∫− ∞ (z + a2 )3
2
dz = 2 πi ⋅
16 a3 i
=
8 a3
⋅

If R (a) < 0 then z = − ai is the pole inside C.

∞ z2 π
In this case, ∫− ∞ 2
(z + a ) 2 3
dz = −
8 a3
⋅

z 2 dz
Aliter: Consider ∫ = ∫C f (z ) dz , where C is the same contour as in Ex. 11.
C (z 2 + a2 )3
By residue theorem,
R
∫C f (z ) dz = ∫− R f ( x) dx + ∫Γ f (z ) dz = 2πi ΣR + .

lim lim
Since
z→∞
z f (z ) = 0 , we have
R→ ∞ ∫Γ f (z ) dz = 0 .

∴ Proceeding to the limits, when R → ∞, we get from (1),

∞
∫− ∞ f ( x) dx = 2πi ΣR + . ...(2)

Now f (z ) has poles at z = ± ai of order three, of which z = ai lies inside C provided R(a)
is positive.
1
Residue of f (z ) (at z = ai) = coeff. of in f (t + ai)
t
1 (t + ai)2
= coeff. of in ⋅
t [(t + ai)2 + a2 ]3
−3
(t + ai)2
(t2 + 2 ait − a2 ) 1 +
1 t 
Now =− 
2
(t + 2 ait) 3
8 a it 3 3  2 ai

1  3t 6 t2 
=− 3
[t2 + 2 ait − a2 ] 1 −
3
− 2 + ......
8 a it  2 ai 4 a 
 C-144

1 1  6 ai 6 a2  1
∴ coeff. of in f (t + ai) = − 3 1 − + 2 = ⋅
t 8 a i  2 ai 4 a  16 a3 i

1
∴ Residue of f (z ) (at z = ai) = ⋅
16 a3 i
∞ 1 π
∴ From (2), ∫ f ( x) dx = 2 πi . = ⋅ ...(3)
−∞ 16 a3 i 8 a3
When R (a) is − ve, pole within C is at z = − ai.
∞ π
∴ In this case ∫ f ( x) dx = − 3 ⋅ [Replacing − a by a in (3)]
−∞ 8a
∞ dx
Example 13: Evaluate ∫ , a > 0.
0 x4 + a4
(Meerut 2002; Purvanchal 09; Gorakhpur 14, 16)
dz
Solution: Consider the integral ∫ f (z ) dz = ∫C ,
C z + a4
4

where C is the contour as described in Ex. 11.

By Cauchy’s residue theorem, we have
R dx dz
∫ C f (z ) dz = ∫ − R x 4 + a 4 + ∫Γ 4 4
= 2 πi Σ R + .
z +a
lim lim dz
Since
R→ ∞
z f (z ) = 0 therefore
R→ ∞ ∫Γ z 4
+ a4
=0

lim R dx
so that ∫− R = 2πi Σ R +
R→ ∞ x 4
+ a4
∞ dx
or ∫− ∞ = 2πi Σ R + . …(1)
x4 + a4

Poles of f (z ) are given by z 4 + a4 = 0 or z = a e(2 n + 1) πi /4 , where n = 0 , 1, 2, 3.

∴ z = a e i π /4 , a e3 i π /4 , a e5 i π /4 , a e7i π /4 are the simple poles of f (z ).

Out of these only z = a e i π /4 , a e i3 π /4 lie inside C.

If α denotes any of these poles then residue at z = α is

lim lim z −α Form 0 
(z − α) f (z ) =
z→α z → α z 4 + a4  0 
lim 1 1 α α
= = = =− 4 ⋅ [ ∵α4 = − a4 ]
z → α 4z 3
4α 3
4α 4
4a
Now sum of the residues at poles inside C
1 1 2i i√2
=− {a e i π /4 + a e i 3 π /4 } = − . =− ⋅
4
4a 4a 3
√2 4 a3
Hence from (1), we have
 C-145

∞ dx  i √ 2 π √ 2
∫− ∞ x 4
+a 4
= 2 πi  −  =
 4 a3  2 a3
∞ dx π √2
or ∫0 x 4
+a 4
=
4 a3
⋅

Example 14: Prove by contour integration that

∞ dx 1 1 . 3 . 5 . …… (2 n − 3) 1
∫0 (a + bx2 )n = 2n b1 /2 ⋅ 1. 2 . 3 .…… (2n − 1) ⋅ a(n−1) /2 ⋅
(Rohilkhand 2011)
dz
Solution: Consider the integral ∫C (a + bz 2 )n
= ∫C f (z ) dz , where C is the same

contour as described in Ex. 11.

By Cauchy’s residue theorem , we have
R dx dz
∫ C f (z ) dz = ∫ − R (a + bx2 )n + ∫Γ = 2 πi Σ R + .
(a + bz 2 )n
lim lim dz
Since
R→ ∞
z f (z ) = 0 therefore
R→ ∞ ∫Γ (a + bz 2 )n
=0

so we have
lim R dx ∞ dx
∫− R = 2πi Σ R + or ∫− ∞ = 2πi Σ R + .…(1)
R→ ∞ (a + bx ) 2 n
(a + bx2 )n

z = ± i √ (a / b) are the poles of f (z ) of order n. The pole z = √ (a / b) i lies inside C.

Residue at z = i √ (a / b) is

1  1 
Dn−1
(n − 1)!  n 
 n  a  
b z + i   
  b   z = √(a / b) i


1 1  (− n)(− n − 1)……{− n − (n − 1) + 1} 
= ⋅ n 2 n −1 
(n − 1)! b 
 +  a  
  z i   
 b   z = i √(a / b)
 
(− 1)n − 1 n (n + 1)……(2 n − 2)
= ⋅
(n − 1)! b n
{2 i √ (a / b)}2 n − 1

(− 1)n − 1 1 . 2 . 3 . …… (n − 1) . n (n + 1) …… (2 n − 2)
=
22 n − 1 {√ (a / b)}2 n − 1 i2 n − 1 b n (n − 1)!. (n − 1)!

1 . 3 . 5 …… (2 n − 3) . 2 n − 1 1 . 2 . 3……(n − 1)
=− i
22 n − 1 a n − 1 /2 b1 /2 (n − 1)!. (n − 1)!
1 . 3 . 5 …… (2 n − 3) i
=− n − 1 /2
⋅
2 a n
b1 /2 1 . 2 . 3……(n − 1)
 C-146

Hence from (1), we have

∞ dx  1 . 3 . 5 …… (2 n − 3) i 
∫− ∞ (a + bx )
= 2 πi  − n n − 1 / 2 1 / 2
2 n 
b 1 . 2 . 3 …… (n − 1)
 2 a
∞ dx π 1 . 3 . 5 …… (2 n − 3)
or ∫0 (a + bx ) 2 n
=
2 a n n −1 /2 1 /2
b
⋅
1 . 2 . 3 …… (n − 1)
⋅

Comprehensive Exercise 3

∞ dx 3π
1. Prove that ∫ = ⋅
−∞ ( x2 + 1)3 8
∞ x2
2. Evaluate ∫0 (1 + x2 )3
dx.

∞ dx
3. Evaluate ∫0 1 + x2
⋅
(Kanpur 2007; Rohilkhand 10; Gorakhpur 05, 11, 13)

∞ dx π (b + 2 c )
4. Prove that ∫− ∞ 2 2
( x + b )( x + c ) 2 2 2
=
2 bc 3 (b + c )3
, b > 0, c > 0.

5. Use the method of contour integration to prove that

∞ x6 3π √ 2
∫0 (x 4
+a ) 4 2
dx =
16 a
, a > 0.

A nswers 3

π π
2. 3.
16 2

9 Jordan’s Inequality. 2θ ≤ sin θ ≤ θ, where 0 ≤ θ ≤ π ⋅

π 2
Let y = cos x. We know that as θ increases from 0 to π / 2, cos θ decreases from 1 to 0.
Consequently the mean ordinate of the graph y = cos x also decreases steadily over the
range 0 ≤ x ≤ θ.
1 θ sin θ
The mean ordinate is ∫ cos θ dθ = ⋅
θ 0 θ
π 2 sin θ 2θ
Hence when 0 ≤ θ ≤ , ≤ ≤ 1 or ≤ sin θ ≤ θ.
2 π θ π
This is known as Jordan’s inequality.
 C-147

10 Jordan’s Lemma
If f (z ) tends to zero uniformly as z → ∞ and f (z ) is meromorphic in the upper half plane then
lim
R → ∞ ∫Γ
e imz f (z ) dz = 0 , (m > 0 )

where Γ denotes the semi-circle | z | = R, I (z ) > 0 .

Proof: Here we assume that R is large enough so as to include within it all the
singularities of f (z ) and f (z ) has no singularity on Γ.
lim
Since f (z ) = 0 therefore for ε > 0 there exists R0 > 0 such that| f (z )| < ε when
R→ ∞
| z | = R ≤ R0 .
Now let Γ denote any semi-circle with radius R ≥ R0 . Putting z = Re i θ , we get
π iθ
∫Γ e imz f (z ) dz = ∫0 e imRe f ( R e i θ ) R i e i θ dθ

π cos θ − mR sin θ
= ∫0 e imR e f ( R e iθ ) i R e iθ dθ.

π
 e imz f (z ) dz ≤ imR cos θ
| e − mR sin θ | f ( R e iθ )|| R i e iθ| dθ
 ∫Γ  ∫0 | e
π
< ∫0 e − mR sin θ ε R dθ [∵ | f (z )| = | f ( R e iθ )| < ε]

π /2
=2ε R∫ e − mR sin θ dθ
0
π /2
≤2ε R∫ e −2 mR θ /π
dθ, by Jordan’s inequality
0

2 ε R (1 − e − mR ) επ επ
= = (1 − e − mR ) < ⋅
2 mR / π m m
lim
Hence
R→ ∞ ∫Γ e imz f (z ) dz = 0 .

11 Evaluation of the Integrals of the form

∞ P (x) ∞ P (x)
∫ − ∞ Q ( x ) sin mx dx ; ∫ − ∞ Q ( x ) cos mx dx, m > 0, where

(i) P ( x), Q ( x) are polynomials, (ii) deg Q ( x) > deg P ( x)

(iii) Q ( x) = 0 has no real roots.

Under the above mentioned conditions the given integrals are convergent. Consider
imz P (z )
∫ C e f (z ) dz = ∫ C e Q (z ) dz ,
imz
 C-148

where C is the contour consisting of a semi-circle Γ of radius R so large as to include all

the poles of the integrand in the upper half plane and part of the real axis from − R to R.
By Cauchy’s residue theorem, we have
R
∫C e imz f (z ) dz = ∫− R e imx f ( x) dx + ∫Γ e imz f (z ) dz = 2πi Σ R + .

lim
We have
R→ ∞ ∫Γ e iz f (z ) dz = 0 , by Jordan’s lemma.

lim R
∴ ∫− R e ix f ( x) dx = 2πi Σ R+
R→ ∞
∞ ∞
or ∫− ∞ f ( x) cos mx dx + i ∫ f ( x) sin mx dx = 2π i Σ R + .
−∞

Equating real and imaginary parts on both sides, we shall get the values of the
given integrals.

∞ cos mx π − ma
Example 15: Prove that ∫ 2 2
dx = e , m ≥ 0.
0 a +x 2a
∞ x sin mx π − ma
Deduce that ∫ 2 2
dx = e .
0 x +a 2 (Gorakhpur 2007, 09, 11, 15)

Solution: Consider the integral

e imz
∫C a2 + z 2
dz = ∫C f (z ) dz ,

where C is the contour consisting of a large semi-circle Γ of

radius R containing all the poles of the integrand in the
upper half plane and the part of real axis from − R to R.
By Cauchy’s residue theorem , we have
R e imx e imz
∫C f (z ) dz = ∫− R 2 2
dx + ∫Γ dz = 2πi ΣR + .
a +x a + z2
2

lim 1
Since = 0 therefore we have
z → ∞ z 2 + a2
lim
R→ ∞ ∫Γ f (z ) dz = 0 , by Jordan’s lemma.

lim R e imx
∴ ∫− R dx = 2πi ΣR +
R→ ∞ 2
a +x 2

∞ imx
e
or ∫− ∞ dx = 2πi ΣR + …(1)
a2 + x2

z = ± ai are the simple poles of f (z ). The pole z = ai lies inside C.

 C-149

Residue at z = ai is
lim lim e imz e − ma
(z − ai) f (z ) = (z − ai) = ⋅
z → ai z → ai z 2 + a2 2 ia

From (1), we have

∞ e imx 2 π i e − ma π − ma
∫− ∞ x2 + a2
dx =
2 ia
= e
a
.

Equating real parts on both sides, we get

∞ cos mx π − ma ∞ cos mx π − ma
∫ − ∞ x2 + a2 dx = a e or ∫0 x +a2 2
dx =
2a
e .

Differentiating both sides w.r.t. m, we get

∞ x sin mx π − ma ∞ x sin mx π − ma
∫0 − a2 + x2 dx = 2a (− a) e or ∫0 2
a +x 2
dx =
2
e .

∞ cos ax
Example 16: Evaluate ∫ dx, a > 0 , b > 0 .
0 ( x + b2 )2
2

e iaz
Solution: Consider ∫C (z 2 + b2 )2
dz = ∫C f (z ) dz ,

where C is the same contour as described in Ex. 15.

By Cauchy’s residue theorem, we have
R e iax e iaz
∫C f (z ) dz = ∫− R 2 2 2
dx + ∫Γ dz = 2 πi ΣR + .
(x + b ) (z + b2 )2
2

By Jordan’s lemma, we have

lim e iaz
R→ ∞ ∫Γ (z 2 + b2 )2
dz = 0 .

lim R e iax
∴ ∫− R dx = 2πiΣR +
R→ ∞ ( x + b2 )2
2

∞ e iax
or ∫− ∞ 2 2 2
dx = 2πi ΣR + . …(1)
(x + b )
z = ± ib are the double poles of f (z ).Out of these only z = ib lies in the upper half plane.
Residue at z = ib is
 e iaz   iae iaz (z + ib)2 − 2 e iaz (z + ib)
φ′ (ib) =  D 2
= 
 (z + ib)  z = ib  (z + ib)4  z = ib
− ab
e [ia (2 ib) − 2] e − ab
= 3
= (ab + 1).
(2 ib) 4 b3 i
Hence from (1), we have
∞ e iax e − ab
∫− ∞ ( x2 + b2 )2
dx = 2 πi ⋅
4 i b3
(ab + 1)
 C-150

∞ cos ax + i sin ax π e − ab
or ∫− ∞ 2
(x + b ) 2 2
dx =
2 b3
(ab + 1).

Equating real parts on both sides, we get

∞ cos ax π (ab + 1) ∞ cos ax π (1 + ab)
∫ − ∞ ( x2 + b2 )2 dx = 2b3 e ab or ∫0 ( x2 + b2 )2
dx =
4 b3 e ab
⋅

Example 17: Prove that

∞ sin x 2 π (√ 3 + 2) −√3 /2 1
∫ −∞ (1 − x + x ) 2 2
dx =
3 √3
e sin ⋅
2

e iz
Solution: Let ∫C f (z ) dz = ∫C (1 − z + z 2 )2
dz , where C is the same contour as in

Ex. 15.
By residue theorem, we have
R
∫C f (z ) dz = ∫ −R f ( x) dx + ∫Γ f (z ) dz = 2πi ΣR + .

By Jordan’s lemma, we have

lim
R → ∞ ∫Γ
f (z ) dz = 0 .

lim R
∴ ∫ −R f ( x) dx = 2πi Σ R +
R→ ∞
∞
or ∫ −∞ f ( x) dx = 2πi ΣR + . …(1)

z = (1 ± i √ 3) / 2 are the poles of f (z ) of second order. The only pole which lies within C
is (1 + i √ 3) / 2 = α, say.
Putting z = α + t in f (z ), we get
e i (α + t)
f (α + t) = 2
{t + (2α − 1) t + α2 − α + 1}2

e i (α + t)
= , since α2 − α + 1 = 0
{t + (2α − 1) t }2
2

−2
e i (α + t)  t 
= 1 + 
(2α − 1)2 t  2α − 1
2

e iα  2t 
= (1 + i t + ……) 1 − + …… ⋅
(2α − 1) t 2 2
 2α − 1 
Residue at z = α is the coefficient of (1 / t) in the expansion of
 i 2   i (2α − 1) − 2 
f (α + t) = e i α  2
− 3
= e iα  3 
(2α − 1) (2α − 1)   (2α − 1) 
[i (1 + i √ 3 − 1) − 2]
= e i (1 + i √3) /2
(i √ 3)3
 C-151

e −√3 /2 e i /2 (√ 3 + 2)
= ⋅
i3 √ 3
Hence from (1), we have
∞ e ix 2 πe −√3 /2
∫ −∞
i /2
dx = (√ 3 + 2) e .
2
( x − x + 1) 2
3 √3

Equating imaginary parts on both sides, we get

∞ sin x 2 π e − √3 /2 1
∫ −∞ 2
( x − x + 1) 2
dx =
3 √3
( √ 3 + 2) sin ⋅
2

∞ cos x2 + sin x2 − 1
Example 18: Prove that ∫ dx = 0 .
0 x2
2
e iz − 1
Solution: Let ∫C f (z ) dz = ∫C z2
dz , where C is the same contour as in

Example 15.
Since f (z ) has no poles in the upper half plane therefore by Cauchy’s residue theorem,
we have
R
∫C f (z ) dz = ∫ −R f ( x) dx + ∫Γ f (z ) dz = 0 . …(1)

π exp ( i R2 e i2 θ ) − 1
We have ∫Γ f (z ) dz = ∫0 R i e i θ dθ, putting z = R e i θ
R2 e i2 θ
i π
= ∫0 e − i θ [exp {i R2 (cos 2θ + i sin 2θ)} − 1] dθ
R
π i − iθ
= ∫0 R
e [exp (− R2 sin 2θ) exp (i R2 cos 2θ) − 1] dθ.

 f (z ) dz
 ∫Γ
Now

1 ∞
| i e − iθ |[exp (− R2 sin 2θ)|exp (iR2 cos 2θ)| + | − 1|] dθ
R ∫0
≤

1 ∞
R ∫0
≤ [exp (− R2 sin 2θ) + 1] dθ, which tends to zero as R → ∞.

∴ ∫Γ f (z ) dz = 0 when R → ∞.

Thus when R → ∞, we have from (1)

2
∞ ∞ ei x − 1
∫ −∞ f ( x) dx = 0 or ∫ −∞ x2
dx = 0
2
∞ ei x − 1
or 2∫ dx = 0 .
0 x2
Equating real and imaginary parts on both sides, we get
 C-152

∞ cos x2 − 1 ∞ sin x2
∫0 x2
dx = 0 and ∫0 x2
dx = 0 .

Adding these relations, we get

∞ cos x2 + sin x2 − 1
∫0 x2
dx = 0 .

∞ log (1 + x2 )
Example 19: Prove by contour integration ∫0 1 + x2
dx = π log 2.
(Kanpur 2008; Gorakhpur 09, 13)
Solution: Consider
log (i + z )
∫C 1 + z2
dz = ∫C f (z ) dz , where C is the contour of Ex. 15.

By residue theorem, we have

R
∫C f (z ) dz = ∫ −R f ( x) dx + ∫Γ f (z ) dz = 2πi Σ R + .
…(1)
lim lim z log ( i + z )
Now z f (z ) =
z→∞ z → ∞ (i + z )(z − i)
lim z lim log (i + z )
= ⋅ = 0.
z→∞ z−i z→∞ i+ z

∴ when R → ∞, we have from (1)

∞
∫ −∞ f ( x) dx = 2πi ΣR +
…(2)
z = ± i are the simple poles of f (z ) and only z = i lies inside C.
Residue at z = i is
lim lim log (i + z ) log 2 i log 2 + i (π / 2)
(z − i) f (z ) = = = ⋅
z→i z → i (z + i) 2i 2i
Hence from (2), we have
∞ log (i + x)
∫ −∞ 1 + x2
dx = π {log 2 + i (π / 2)}.

Equating real parts on both sides, we get

1
∞
log ( x2 + 1) ∞ log (1 + x2 )
∫ −∞ 2
x2 + 1
dx = π log 2 or ∫0 1 + x2
dx = π log 2.

Comprehensive Exercise 4

∞ sin x π
1. Prove that ∫ dx = − sin 2.
−∞ x2 + 4 x + 5 e
 C-153

∞ x sin x
2. (i) Apply the calculus of residues to evaluate ∫0 x2 + a2
dx, a > 0 .
(Avadh 2008; Gorakhpur 2011)
∞ π e− a
cos x
(ii) Prove that ∫ dx = , a > 0.
0 a2 + x2 a
∞ sin x π
3. Show that ∫ dx = 2 sin 1.
− ∞ x2 − 2 x + 5 2e
∞ cos x dx  e− b e− a 
π
4. Prove that ∫ =  −  , a > 0, b > 0.
−∞ 2 2 2
( x + a )( x + b ) 2
a − b  b
2 a 
2

5. Prove that when m > 0,

∞ cos mx 2π  m π
∫− ∞ dx = sin  +  e − (1 /2) m √3 .
x 4 + x2 + 1 √3  2 6
∞ cos mx π  ma π 
6. Prove that ∫0 e − ma / √2 sin 
dx = + ⋅
x +a 4
2a4 3  √ 2 2
(Gorakhpur 2008)
∞ x sin mx π ma
Deduce that ∫ dx = 2 e − ma / √2 sin ⋅
0 x4 + a4 2a √2
∞ x3 sin mx π − ma / √2 ma
7. Prove by contour integration that ∫0 x 4
+a 4
dx =
2
e cos
√2
⋅

8. If a ≥ 4, prove that
∞ (1 + x2 ) cos ax π − a(√3 /2) a
(i) ∫ dx = e cos
0 2
1+ x + x 4
√3 2
∞ x sin ax π − a(√3 /2) a
(ii) ∫0 1 + x2 + x 4
dx =
√3
e sin
2
⋅

A nswers 4
π −a
2. (i) e
2

12 Poles Lie on the Real Axis

We shall now discuss the case when the integrand has poles on the real axis as well as
within the semi-circle.
P (z )
Theorem: Let f (z ) = , where P (z ) and Q (z ) are polynomials such that Q (z ) has only
Q (z )
non-repeated real roots, that is f (z ) has only simple poles on the real axis. Let m > 0 and let the
degree of Q (z ) exceeds that of P (z ) , then
∞ p q
P ∫ −∞ e imx f ( x) dx = 2 πi Σ Res (ak ) + πi Σ Res(bk ),
k =1 k =1

where a1, a2 …, ap are the zeros of Q (z ) in the region Im z > 0 and b1, b2 , …, bq are its zeros in the
real axis, and Res (α) denotes the residue of e imz f (z ) at α.
 C-154

∞ cos mx
Example 20: If m > 0, show that P ∫ dx = − sin mb.
−∞ x−b
1
Solution: Referring to the above theorem, let f (z ) = ⋅ Here f (z ) has simple real
z−b
pole at z = b.
e imz e imz
∴ Res z = b = lim (z − b) ⋅ = e imz .
z−b z→b (z − b)

∞ e imz
∫ −∞ z − b dz = πie
imb
Hence P .

Equating real parts on both sides, we get

∞ cos mx
P ∫ −∞ x−b
dx = − π sin mb.

∞ sin mx π
Example 21: If m > 0, prove that P ∫0 x
dx = ⋅
2
1
Solution: Referring to the above theorem, let f (z ) = ⋅ Here f (z ) has simple real pole
z
at z = 0.
e imz e imz
∴ Res z = 0 = lim (z − 0 ) = 1.
z z→0 z
∞ e imx e imz
Thus P∫ dx = πi × Residue of at z = 0
−∞ x z
= πi × 1 = πi .
Equating real and imaginary parts on both sides, we get
∞ cos mx
P∫ dx = 0 …(1)
−∞ x
∞ sin mx
and P∫ dx = π. …(2)
−∞ x
sin mx
Note: The principal part ‘P’ has been dropped in (2), since lim = m,
x→0 x
whereas in the first integral the integrand becomes unbounded at the origin.
Hence from (2), we get the required result.

Indenting Method: We can avoid the poles which lie on the real axis by drawing
semi-circles of small radii about these poles as centres. This method is known as
‘indenting at a point’.
 C-155

∞ sin mx
Example 22: (i) Evaluate ∫ dx, m > 0 .
0 x (Gorakhpur 2007, 10, 13)

(ii) By integrating e iz / z around a suitable contour, prove that

∞ sin x π
∫0 x
dx = ⋅
2 (Rohilkhand 2011)

e imz
Solution: (i) Consider the integral ∫ f (z ) dz = ∫C dz
C z

where C is the contour consisting of (1) the upper half of the circle | z | = R

(2) real axis from r to R where r is small and R is large

(3) real axis from − R to − r

(4) upper half of the circle γ,| z | = r.

Obviously the function f (z ) has no singularity inside C, therefore by Cauchy’s residue

theorem, we have
R −r
∫C f (z ) dz = ∫r f ( x) dx + ∫Γ f (z ) dz + ∫ −R f ( x) dx + ∫γ f (z ) dz

= 0. …(1)
By Jordan’s lemma, we have
lim
R → ∞ ∫Γ
f (z ) dz = 0 .

lim
Again z f (z ) = 1 therefore
z→0
lim
r→0 ∫γ f (z ) dz = i (0 − π) = − iπ.

Thus when r → 0, R → ∞, we get from (1)

∞ 0
∫0 f ( x) dx + ∫ −∞ f ( x) dx − iπ = 0

∞ ∞ imx
e
or ∫ −∞ f ( x) dx = iπ or ∫ −∞ x
dx = iπ.

Equating imaginary parts on both sides, we get

∞ sin mx ∞ sin mx π
∫ −∞ x dx = π or ∫0 x
dx = ⋅
2
(ii) Proceed as in part (i) taking m = 1.

Example 23: Prove that if a > 0

∞ cos x π sin a ∞ sin x
(i) P ∫ −∞ a −x2 2
dx =
a
(ii) P ∫ −∞ a − x2
2
dx = 0.
(Gorakhpur 2008, 09)
 C-156

Solution: Consider the integral

e iz
∫C f (z ) dz = ∫C a − z2
2
dz ,

where C is the contour consisting of the large semi-circle

Γ of radius R, indented at z = a and z = − a. γ1 and γ 2 are
the small semi-circles with radii r1 and r2 and centres at
z = − a and z = a respectively.
The function f (z ) is regular within and on the contour C therefore by Cauchy’s residue
theorem, we have
− (a + r1)
∫C f (z ) dz = ∫Γ f (z ) dz + ∫ −R f ( x) dx + ∫γ1 f (z ) dz

a − r2 R
+ ∫ − (a − r1) f ( x) dx + ∫γ2 f (z ) dz + ∫ a + r2 f ( x) dx = 0 .

…(1)
2 2
Since 1 / (a − z ) → 0 as z → ∞, therefore by Jordan’s lemma, we have
lim
R→ ∞ ∫Γ f (z ) dz = 0 .

lim lim e iz e − ia
Also (a + z ) f (z ) = = ,
z→−a z→−a a−z 2a
lim e− i a π − ia
therefore
r1 → 0 ∫γ1 f (z ) dz = i
2a
(0 − π) = − i
2a
e .

lim π ia
Similarly
r2 → 0 ∫γ2 f (z ) dz = i
2a
e .

Hence when R → ∞, r1 → 0 , r2 → 0 , we have from (1)

−a  π − ia  a iπ ia ∞
∫ − ∞ f ( x) dx +  − i 2a e  + ∫ − a f ( x) dx + 2a e + ∫a f ( x) dx = 0

∞ π − ia
∫− ∞
ia
or P f ( x) dx = − i (e −e )
2a
∞ e ix π
or P ∫− ∞ a − x2
2
dx =
a
sin a.

Equating real and imaginary parts on both sides, we get

∞ cos x π ∞ sin x
P ∫− ∞ 2
a −x 2
dx =
a
sin a, P ∫− ∞ a2 − x2
dx = 0 .

∞ sin x π
Example 24: Prove that ∫ 2 2
dx = (1 − e − a ), a > 0 .
0 x (x + a ) 2 a2
e iz
Solution: Consider the integral ∫C f (z ) dz = ∫C z (z 2 + a2 )
dz ,

where C is the contour consisting of

 C-157

(i) a large semi-circle Γ,| z | = R in the upper half

plane
(ii) real axis from r to R
(iii) real axis from − R to −r
(iv) a small semi-circle γ,| z | = r.
z = 0, ± ia are the simple poles of f (z ). Out of these only z = ia lies within C.
Residue at z = ai is
lim lim e iz e− a
(z − ai) f (z ) = = ⋅
z → ai z → ai z (z + ai) − 2 a2
By residue theorem, we have
R −r
∫C f (z ) dz = ∫r f ( x) dx + ∫Γ f (z ) dz + ∫− R f ( x) dx + ∫γ f (z ) dz

= 2πi ΣR+ . …(1)

lim
By Jordan’s lemma, we have
R→ ∞ ∫Γ f (z ) dz = 0 .

lim lim e iz 1
Also z f (z ) = = 2 ,
z→0 z→0 z +a
2 2
a
lim 1 πi
∴
r→0 ∫γ f (z ) dz = i ⋅
a2
(0 − π) = −
a2
⋅

∴ when r → 0 and R → ∞, we have from (1)

∞ 0 iπ  e− a 
∫0 f ( x) dx + ∫− ∞ f ( x) dx −
a
= 2 πi  − 2 
2
 2a 
∞ iπ
or ∫− ∞ f ( x) dx = (1 − e − a ).
a2
Equating imaginary parts on both sides, we get
∞ sin x π −a
∫ − ∞ x ( x2 + a2 ) dx = a2 (1 − e ).
Example 25: Show that if a and m are positive,
∞ sin2 mx π
∫0 2 2 2 2
dx = 5
{e −2 am (2 am + 3) + 4 am − 3}.
x (a + x ) 8a
2
∞ sin mx 1 ∞ 1 − cos 2 mx
Solution: We have ∫0 2 2
x (a + x ) 2 2
dx =
2 ∫0 x2 (a2 + x2 )2
dx.

1 1 − e i 2 mz
Consider ∫C f (z ) dz =
2 ∫C z 2 (a2 + z 2 )2
dz ,

where C is the same contour as in Ex. 24.

z = 0, ± ia are the poles of f (z ) of order two. Only z = ai lies within C.
Putting z = ai + t in f (z ), we get
 C-158

1 {1 − e i 2 m (ai + t)} 1 (1 − e −2 am e i2 mt )
f (ai + t) = 2 2 2 2
=
2 (ai + t) {a + (ai + t) } 2 (ai + t)2 (2 ait + t 2 )2
−2 −2
1  t  t 
= (1 − e −2 am e i 2 mt ) 1 +  1 + 
8 a4 t 2  ai  2 ai
1  2t 
= {1 − e −2 am (1 + i2 mt + ……)} 1 − + ……
8 a4 t 2  ai 
 t 
1 − + ……
 ai 
1  2t   t
= (1 − e − 2 am − i2 m e − 2 am t) 1 −  1 − 
8a t 4 2  ai   ai
neglecting higher powers of t since t is small
1  3 
= 4 2 {(1 − e −2 am ) − i2 me −2 am t } 1 − t  ⋅
8a t  ai 
∴ Residue at z (= ai) is = coefficient of (1 / t) in f (ai + t)
1  3 −2 am 
=  − (1 − e ) − 2 im e −2 am 
8 a 4  ai 
1
= {3 i (1 − e −2 am ) − 2 iame −2 am} ⋅
8 a5
By residue theorem, we have
R −r
∫C f (z ) dz = ∫r f ( x) dx + ∫Γ f (z ) dz + ∫− R f ( x) dx + ∫γ f (z ) dz

= 2πi Σ R + . …(1)

By Jordan’s lemma, ∫Γ f (z ) dz = 0 , when R → ∞.

lim lim 1 (1 − e i 2 mz )  0
Since z f (z ) = Form 0 
z→0 z → 0 2 z (a2 + z 2 )2

lim − i 2 m e i 2 mz − im
= = 4 ⋅
z → 0 2 {(a + z ) + z 2 (a + z ). 2 z} a
2 2 2 2 2

lim  im  −mπ
∴
r→0 ∫γ f (z ) dz = i  − 4  (0 − π) =
 a  a4
⋅

Hence when r → 0, R → ∞, we have from (1)

∞ 0 mπ
∫0 f ( x) dx + ∫ − ∞ f ( x) dx − a 4
i
= 2 πi ⋅ {3 (1 − e −2 am ) − 2 ame −2 am}
8 a5
∞ π
or ∫− ∞ f ( x) dx = [4 am − 3 + e −2 am (3 + 2 am)].
4 a5
Equating real parts on both sides, we get
 C-159

1 ∞ 1 − cos 2 mx π
∫− ∞ 2 2 2 2
dx = [4 am − 3 + e −2 am (3 + 2 am)]
2 x (a + x ) 4 a5
∞ sin2 mx π
or ∫− ∞ dx = [4 am − 3 + e −2 am (3 + 2 am)]
x (a2 + x2 )2
2
4 a5
∞ sin2 mx π
or ∫0 2 2 2 2
dx = [4 am − 3 + e −2 am (3 + 2 am)].
x (a + x ) 8 a5
∞ x − sin x
Example 26: Evaluate ∫ dx, a > 0 .
0 x (a2 + x2 )
3

z − i + ie iz
Solution: Consider ∫ f (z ) dz = ∫C dz , where C is the same contour as
C z 3 (a2 + z 2 )
in Ex. 24.
z = 0 is the pole of f (z ) of order two and z = ± ia are the simple poles of f (z ). Out of
these only z = ai lies within C.
Residue at z = ai is
lim lim z − i + ie iz (a − 1 + e − a ) i
(z − ai) f (z ) = = ⋅
z → ai z → ai z 3 (ai + z ) 2a 4
By residue theorem , we have
R −r
∫C f (z ) dz = ∫r f ( x) dx + ∫Γ f (z ) dz + ∫− R f ( x) dx + ∫γ f (z ) dz

= 2πi ΣR + . …(1)

By Jordan’s lemma, when R → ∞, ∫ f (z ) dz → 0 .

Γ

lim lim z − i + ie iz  0
Since z f (z ) = Form 0 
z→0 z → 0 z 2 (a2 + z 2 )

 1 
z − i + i 1 + iz + i 2 z 2 + ……
lim  2 
=
z→0 z 2 (a2 + z 2 )
1 2
lim − 2 iz + ……
=
z → 0 z 2 (a2 + z 2 )
1
lim − 2 i + terms containing z in Nr. i
= =− 2 ⋅
z→0 a2 + z 2 2a
lim  i  π
∴
r→0 ∫γ f (z ) dz = i  − 2  (0 − π) = − 2 ⋅
 2a  2a
Hence when r → 0, R → ∞, we have
∞ 0 π (a − 1 + e − a ) i
∫0 f ( x) dx + ∫− ∞ f ( x) dx −
2a 2
= 2 πi
2a 4
 C-160

∞ π π
or ∫− ∞ f ( x) dx = 2
− (a − 1 + e − a ).
2a a4
Equating real parts on both sides, we get
∞ x − sin x π 1 2 −a
∫ − ∞ x3 (a2 + x2 ) dx = a 4 (2 a − a + 1 − e )
∞ x − sin x π 1
or ∫0 3 2 2
dx = 4
( a2 − a + 1 − e − a ).
x (a + x ) 2a 2
∞ x 4
π √3
Example 27: Prove that P ∫0 x 6
−1
dx =
6
⋅

z4
Solution:Consider ∫ f (z ) dz = ∫C 6
dz , where C
C z −1
is the contour consisting of semi-circle Γ of radius R in
the upper half plane indented at z = − 1, 1, r1 and r2 are
the radii of the small semi-circles γ1 and γ 2 with centres
at z = − 1 and z = 1 respectively.
z = e2 n π i /6 , n = 0 , 1, 2, 3, 4, 5

are the simple poles of f (z ) of which only z = e i π /3 , e i 2 π /3 lie within C. Let α denote
any of these poles.
lim 4 lim 4
z z 1
Residue at z = α is = = ⋅
z → α D (z 6 − 1) z → α 6z 5 6α

Sum of the residues at poles inside C is

1 − i π /3 1 2i π √3 i
= (e + e − i2 π /3 ) = (e − iπ /3 − e iπ /3 ) = − sin = − ⋅
6 6 6 3 6
By residue theorem, we have
− (1 + r1)
∫C f (z ) dz = ∫Γ f (z ) dz + ∫− R f ( x) dx + ∫γ1 f (z ) dz

1 − r2 R
+ ∫ − (1 − r1) f ( x) dx + ∫γ2 f (z ) dz + ∫1 + r2 f ( x) dx

= 2πi Σ R + . …(1)
π 4 i4 θ iθ π 5
 f (z ) dz ≤
R e Rie R
Now
 ∫ Γ  ∫0  R e 6 i6 θ  dθ ≤
−1  ∫0 6
R −1
dθ

R5 π
= → 0 as R → ∞.
R6 − 1
lim
∴
R→ ∞ ∫Γ f (z ) dz = 0 .

lim lim (z + 1) z 4  0
Since (z + 1) f (z ) = Form 
z → −1 z → −1 z6 − 1  0
 C-161

4
lim z + 4 z 3 (z + 1) 1
= =− ,
z → −1 6z 5 6
lim 1 iπ
∴
r1 → 0 ∫γ1 f (z ) dz = −
6
i (0 − π) =
6
⋅

lim lim 4
z  0
Since (z − 1) f (z ) = (z − 1) Form 0 
z →1 z →1 6
z −1
4
lim z + 4 z 3 (z − 1) 1
= = ,
z →1 6z 5 6

lim − iπ
∴
r2 → 0 ∫γ2 f (z ) dz =
6
⋅

Hence as r1 → 0 , r2 → 0 , R → ∞, we have from (1)

−1 iπ 1  iπ  ∞  √3 
∫− ∞ f ( x) dx +
6
+ ∫ −1 f ( x) dx +  −  +
 6 ∫1 f ( x) dx = 2 πi  −
 6 
i

∞ π √3 ∞ π √3
or P ∫− ∞ f ( x) dx =
3
or P ∫0 f ( x) dx =
6
⋅

13 Integrals of Many Valued Functions

a
Such integrals involve functions of the type log z , z where a is not an integer.
These integrals are not single valued. To evaluate such integrals we consider only those
contours whose interiors do not contain any branch point. For these integrals we
generally use double circle contour indented at the centre.

∞ log x π
Example 28:Prove that ∫ dx = − , using as a contour a large semi-circle in the
0 (1 + x2 )2 4
upper half plane indented at the origin.
log z
Solution: Consider ∫ C (1 + z 2 )2 dz = ∫C f (z ) dz ,

where C is the contour as given in the question. (See

figure).
z = ± i are the double poles of f (z ). Only z = i lies inside
C.
Residue at z = i is φ′ ( i), where
log z
φ (z ) = or log φ (z ) = log log z − 2 log (z + i).
(z + i)2
 C-162

Differentiating,
φ′ (z ) 1 2  1 2 
= − or φ′ (z ) = φ (z )  − 
φ (z ) z log z z + i  z log z z + i 
log i  1 1 (1 − log i)
∴ φ′ (i) =  − =−
−4  i log i i  4i
1 1
=− (1 − log e iπ /2 ) = − {1 − i (π / 2)}.
4i 4i
By residue theorem, we have
R r
∫C f (z ) dz = ∫r f ( x) dx + ∫Γ f (z ) dz + ∫R f ( x e iπ ) e iπ dx

+ ∫γ f (z ) dz = 2 πi ΣR + . …(1)

 π log ( R e i θ )  π log R + θ
f (z ) dz ≤ R i e iθ  dθ ≤
 ∫Γ ∫0  (1 + R2 ei2 θ )2 ∫0
Now R dθ
  ( R 2 − 1)2

 1 2
 π log R + π  R
 2  R2  π log R 1 2 1 
= = 2 
+ π 
2
( R − 1)2 2
( R − 1) R 2 R
lim log R
→ 0 as R → ∞, since = 0.
R→ ∞ R
π
 f (z ) dz≤ iθ
) rie i θ | dθ → 0 as r → 0 ,
 ∫γ ∫0 | f (re
Similarly

lim lim log r
since r log r = = 0.
r→0 r → 0 1/ r

Hence when r → 0, R → ∞, we have from (1)

∞ 0
∫0 f ( x) dx − ∫∞ f ( x e i π ) dx = 2 π i ΣR +

∞ log x ∞ log x e i π  1  iπ  
or ∫0 2 2
(1 + x )
dx + ∫0 (1 + x e 2 i2π 2
)
dx = 2 π i −
 4i
1 −  
 2 
π i π
=− 1 − ⋅
2  2
Equating real parts on both sides, we get
∞ log x π ∞ log x π
2∫ dx = − or ∫0 dx = − ⋅
0 (1 + x2 )2 2 (1 + x2 )2 4
∞ xb π πb
Example 29: Prove that ∫ 2
dx = sec , − 1 < b < 1.
0 1+ x 2 2
zb
Solution: Consider ∫C f (z ) dz = ∫C 1 + z2
dz , where C is the same contour as in

Example 28.
 C-163

b
Here we have avoided the branch point 0 of z by indenting at origin.
z = ± i are the simple poles of f (z ). Only z = i lies inside C.
Residue at z = i is
lim lim zb ib
(z − i) f (z ) = =
z→i z → i z + i 2i
b
 π π
cos + i sin 
 2 2 1  bπ b π
= = cos + i sin ⋅
2i 2i  2 2 
By residue theorem, we have
R r
∫C f (z ) dz = ∫r f ( x) dx + ∫Γ f (z ) dz + ∫R f ( x e iπ ) e iπ dx

+ ∫γ f (z ) dz = 2 πi ΣR + . …(1)

 π Rb e ibθ i R e i θ  π Rb + 1
f (z ) dz ≤
 ∫Γ ∫0  ∫0
Now  dθ ≤ dθ
 1 + R 2 e i2 θ  R2 − 1

Rb + 1
= π → 0 as R → ∞, since − 1 < b < 1.
R2 − 1

 r b +1
f (z ) dz ≤
0

 ∫γ ∫π
Similarly dθ → 0 as r → 0. since b + 1 > 0.
 1 − r2

Hence from (1), we have

∞ 0 1  πb πb 
∫0 f ( x) dx − ∫∞ f ( x e iπ ) dx = 2 πi ⋅ cos + i sin 
2i  2 2
∞ xb ∞ e ibπ x b
 πb πb 
or ∫0 1+ x 2
dx + ∫0 1+ x e 2 i2 π
dx = π cos
 2
+ i sin  ⋅
2

Equating real parts on both sides, we get

b
∞ x (1 + cos πb) πb
∫0 1+ x 2
dx = π cos
2
∞ x b
π cos (πb / 2) π  πb 
or ∫0 1+ x 2
dx =
1 + cos πb
= sec   ⋅
2 2

Example 30: Prove that

∞ x a −1 2π  2 aπ + π
∫0 2
x + x +1
dx =
√3
cos 
 6
 cosec aπ, (0 < a < 2).

z a −1
Solution: Consider ∫C f (z ) dz = ∫C z2 + z + 1
dz ,

where C is the contour consisting of

(1) a large circle Γ,| z | = R
(2) radius vector θ = 2 π
 C-164

(3) a small circle γ,| z | = ρ

(4) radius vector θ = 0.
By residue theorem, we have
ρ
∫C f (z ) dz = ∫Γ f (z ) dz + ∫R f (r e i 2 π ) e i 2 π dr
R
+ ∫γ f (z ) dz + ∫ρ f ( x) dx = 2 πi ΣR+ .

…(1)
The poles of f (z ) are given by
− 1± i √ 3
z2 + z + 1 = 0 or z = ⋅
2
−1 √3
Thus z = +i = e2 πi /3 = α
2 2
1 √3
and z = − −i = e4 πi /3 = β are the simple poles of f (z ) and both lie in C.
2 2
lim lim z a − 1 α a − 1
Residue at z = α is (z − α) f (z ) = = ⋅
z→α z→α z −β α −β

Sum of the residues at poles inside C is

α a −1 β a −1 1
= + = (α a − 1 − β a − 1)
α −β β −α α −β
1
= [e i 2 π (a − 1) /3 − e i4 π (a − 1) /3 ]
i√3
e i π (a −1) − i π (a − 1) /3
= [e − e i π (a − 1) /3 ]
i√3

2 e i πa (a − 1) π 2 i πa π π
= sin =− e cos  + (a − 1) 
√3 3 √3  2 3 
2 ia π  π + 2 aπ 
=− e cos  ⋅
√3  6 
2π  Ra − 1 exp {i θ (a − 1)} 
 f (z ) dz ≤ i R e i θ dθ
 ∫Γ ∫0
Now  iθ 2 i2 θ
 1 + R e + R e 
2π Ra 2 πR a
≤ ∫0 R2 − R −1
dθ =
R2 − R −1
→0

as R → ∞, since 0 < a < 2.

a
 0 ρ
f (z ) dz ≤
 ∫γ ∫2 π
Similarly dθ → 0 as ρ → 0.
 1 − ρ − ρ2

Hence when R → ∞, ρ → 0 , we have from (1)

∞ 0
∫0 f ( x) dx + ∫∞ f (r e i 2 π ) e i 2 π dr = 2 πi ΣR+
 C-165

∞ x a −1 ∞ x a − 1 e i (a − 1) 2 π
or ∫0 1 + x + x2
dx − ∫0 1 + xe i 2 π + x2 e i4 π
dx

 2 i πa  π + 2 aπ  
= 2 πi  − e cos  
 √3  6 
∞ x a −1 4 π i i πa  π + 2 aπ 
or ∫0 (1 − e2 iaπ ) dx = − e cos  
1 + x + x2 √3  6 

(e ia π − e − ia π ) ∞ x a −1 2π  π + 2 aπ 
or
2i ∫0 1+ x + x 2
dx =
√3
cos 
 6 


∞ x a −1 2π  π + 2 aπ 
or ∫0 1+ x + x 2
dx =
√3
cos 
 6 
 cosec aπ.

e iz log (− iz )
Example 31: By integrating round the contour consisting of a large semi-circle in
z2 + 4
the upper half plane indented at the origin or otherwise prove that

∞ 2 cos x log x + π sin x 1

∫0 2
dx = π e −2 log 2.
x +4 2

e iz log (− iz )
Solution: Consider ∫C f (z ) dz = ∫C z2 + 4
dz , where C is the contour as

given above. (See fig. of Ex. 28).

z = ± 2 i are the simple poles of f (z ). Only z = 2 i lies within C.
Residue at z = 2 i is
lim lim e iz log (− iz ) e −2 log 2
(z − 2 i) f (z ) = = ⋅
z → 2i z → 2i z + 2i 4i

By residue theorem, we have

R r
∫C f (z ) dz = ∫r f ( x) dx + ∫Γ f (z ) dz + ∫R f ( x e iπ ) e iπ dx

+ ∫γ f (z ) dz = 2 πi ΣR + …(1)

e iR e iθ 
 π log (− i R e iθ )
f (z ) dz ≤ i R e iθ
 ∫Γ ∫0
Now 
  dθ
 4 + R 2 e i2 θ
 
π log R + θ + (π / 2)
≤ ∫0 R2 − 4
R dθ

lim log R
→ 0 as R → ∞, since = 0.
R→ ∞ R
 f (z ) dz→ 0 as r → 0 .
 ∫γ
Similarly

Hence when r → 0, R → ∞, we have from (1)
 C-166

∞ 0 e −2 log 2
∫0 f ( x) dx − ∫∞ f ( x e iπ ) dx = 2 πi
4i

∞ e ix log (− ix) ∞ e − ix log (i x) π −2

or ∫0 2
x +4
dx + ∫0 x +42
dx =
2
e log 2

 π
(cos x + i sin x)  log x − i 
∞  2  dx
or ∫0 x2 + 4
 π
(cos x − i sin x)  log x + i 
∞  2 π −2
+ ∫0 x2 + 4
dx =
2
e log 2.

Equating real parts on both sides, we get

 π 
cos x log x + sin x dx
∞  2  π
2∫ 2
= e −2 log 2
0 x +4 2
∞ 2 cos x log x + π sin x π −2
or ∫0 x +42
dx =
2
e log 2.

Example 32: Prove that

a −1
∞ x π
(i) ∫0 1+ x
dx =
sin aπ
, 0 < a<1

∞ x a −1
(ii) ∫0 1− x
dx = π cot aπ, 0 < a < 1.

z a −1
Solution: Let ∫ f (z ) dz = ∫C dz ,
C 1− z
where C is the contour consisting of a large semi-circle
Γ,| z | = R in the upper half plane indented at
z = 0 , z = 1. γ1 and γ 2 are the semi-circles in the upper
half plane with radii ρ1 and ρ2 and centres z = 0 , z = 1
respectively.
By Cauchy’s residue theorem, we have
− ρ1
∫C f (z ) dz = ∫Γ f (z ) dz + ∫− R f ( x) dx + ∫γ1 f (z ) dz

1 − ρ2 R
+ ∫ρ1 f ( x) dx + ∫γ2 f (z ) dz + ∫1 + ρ2 f ( x) dx = 0 , …(1)

since f (z ) has no pole inside C.

π a −1 i (a − 1)θ 
 f (z ) dz ≤
R e
i R e iθ  dθ
Now
 ∫ Γ  ∫0  1− R e iθ

 C-167

∞ Ra Ra π
≤ ∫0 R −1
dθ =
R −1
→ 0 as R → ∞, since a < 1.

lim lim za
Since z f (z ) = = 0, a > 0,
z→0 z → 0 1− z
lim
∴
ρ1 → 0 ∫γ1 f (z ) dz = − i (π − 0 ) 0 = 0 ,

lim lim z a −1
(z − 1) f (z ) = (z − 1) = − 1,
z →1 z →1 1− z
lim
∴
ρ2 → 0 ∫γ2 f (z ) dz = iπ.

Hence as ρ1 → 0 , ρ2 → 0 , R → ∞, we have from (1)

0 1 ∞
∫− ∞ f ( x) dx + ∫0 f ( x) dx + iπ + ∫1 f ( x) dx = 0

0 ∞
or ∫ −∞ f ( x) dx + ∫0 f ( x) dx = − iπ

∞ ∞
or ∫0 f (− x) dx + ∫0 f ( x) dx = − iπ,
putting − x for x in the first integral
a −1 a −1
∞ (− x) ∞ x
or ∫0 1+ x
dx + ∫0 1− x
dx = − iπ

∞ (− 1)a − 1 x a −1 ∞ x a −1
or ∫0 1+ x
dx + ∫0 1− x
dx = − iπ

∞ (e i π )a − 1 x a −1 ∞ x a −1
or ∫0 1+ x
dx + ∫0 1− x
dx = − iπ

∞ e − i π ei a π x a −1 ∞ x a −1
or ∫0 1+ x
dx + ∫0 1− x
dx = − iπ

∞ e ia π x a − 1 ∞ x a −1
or − ∫0 1+ x
dx + ∫0 1− x
dx = − iπ.

Equating real and imaginary parts on both sides, we get

∞ cos aπ x a − 1 ∞ x a −1
− ∫0 1+ x
dx + ∫0 1− x
dx = 0

∞ x a −1 ∞ x a −1
or ∫0 1− x
dx = cos aπ ∫0 1+ x
dx
…(2)
sin aπ x a − 1 x a −1 π
and − ∫ 1+ x
dx = − π or ∫ 1+ x
dx =
sin aπ
⋅ Proved.

Substituting the above value in (2), we get

∞ x a −1
∫0 1− x
dx = π cot aπ.
 C-168

Comprehensive Exercise 5

∞ cos ax − cos bx
1. Apply the calculus of residues to evaluate ∫0 x2
dx, a > b > 0 .

∞ cos 2 ax − cos 2 bx
2. Prove that ∫0 x2
dx = − π (a − b), a > 0 , b > 0 .

∞ sin mx π π e − ma  2
3. Prove that ∫ dx = − m +  , m > 0 , a > 0.
0 2
x (x + a ) 2 2
2a 4
4 a3  a

∞ sin2 mx π
4. Prove that ∫ dx = (e −2 ma − 1 + 2 ma), m > 0 , a > 0 .
0 x (a2 + x2 )
2
4 a3
∞ sin πx
5. Prove that ∫ dx = π.
0 x (1 − x2 )
∞ log (1 + x2 )
6. Prove by contour integration ∫ dx = π log 2
0 1 + x2 (Gorakhpur 2009, 13)
1 log ( x + 1 / x) π
and deduce that ∫0 1 + x2
dx = log 2.
2

(log z )2
7. By integrating round a suitable contour prove that
1 + z2
∞ (log x)2 π3 ∞ log x
∫0 1+ x 2
dx =
8
and ∫0 1 + x2
dx = 0 .
(Gorakhpur 2003, 16)
∞ log x
8. Evaluate ∫0 (1 + x)3
dx.

A nswers 5

π 1
1. − (a − b) 8. −
2 2

14 Rectangular and other Contours

z
Example 33: Integrating round the rectangle with vertices at z = ± π, ± πiR, prove
1 − ae − iz
π ax sin x  1
that if a ≥ 1, ∫0 1 − 2 a cos x + a2
dx = π log 1 +  ⋅
 a

What is the value of this integral when 0 < a < 1 ?

 C-169

Solution: Consider the integral

z
∫ C 1 − ae− iz dz = ∫ C f (z ) dz ,
where C is the rectangle with sides x = ± π, y = 0 ,
y = R. f (z ) has simple poles given by
e iz = a = exp (log a + 2 nπi)

or z = − i (log a + 2 nπi), n = 0 , ± 1, ± 2
and so on.
If 0 < a < 1, there is only one pole z = − i log a which lies inside the rectangle C. If a > 1
there is no pole inside C.
Residue at z = − i log a is
 z  − i log a
 − iz 
= = − log a.
 D (1 − a e ) ai exp (− log a)
z = − i log a

By Cauchy’s residue theorem, we have

π R −π
∫C f (z ) dz = ∫− π f ( x) dx + ∫0 f (π + iy) i dy + ∫π f ( x + iR) dx

0
+ ∫R f (− π + iy) i dy = 2 πi (− log a). …(1)

| x + iR|
 − π f ( x + iR) dx≤ −π

 ∫π ∫π
Now dx
 |1 − a exp (− i x + R)|
−π | x| + R
≤ ∫π a e R −1
dx → 0 , as R → ∞.

π π x dx 0 x dx π x dx
∫− π f ( x) dx = ∫− π 1 − a e − ix
= ∫ −π 1 − a e − ix
+ ∫0 1 − a e − ix
⋅

Putting − x for x in the first integral, we get

0 x dx π x dx π  1 1 
I = ∫π 1 − ae ix
+ ∫0 1 − ae − ix
= ∫0 x 
1 − ae
− ix
−  dx
1 − ae ix 

π − x a (e ix − e − ix ) π − 2 ixa sin x
= ∫0 1 − a (e + e ix − ix
)+ a 2
dx = ∫0 1 − 2 a cos x + a2
dx.

lim  R 0 
R → ∞ ∫0 ∫R
Also f (π + iy) i dy + f (− π + iy) i dy

∞  π + iy (− π + iy) 
= ∫0  −  i dy
1 − a exp (− iπ + y) 1 − a exp (πi + y)
∞ 2π ∞ 2 πi e − y
= ∫0 1 + ae y
i dy = ∫0 e− y + a
dy

∞
= − 2 πi [log (e − y
+ a)]0 = 2 πi [log (1 + a) − log a].
 C-170

Hence as R → ∞, we have from (1)

π − 2 ixa sin x
∫0 1 − 2 a cos x + a2
dx + 2 πi {log (1 + a) − log a} = 2πi (− log a),
0 < a<1
π xa sin x
or ∫0 1 − 2 a cos x + a2
dx = π log (1 + a).

When a > 1, we have from (1)

π − 2 ixa sin x
∫0 1 − 2 a cos x + a2
dx + 2 πi {log (1 + a) − log a} = 0

π
dx = π log 1 +  ⋅
ax sin x 1
or ∫0 1 − 2 a cos x + a 2  a

Example 34: Apply the calculus of residues to prove that

∞ cosh ax 1 a
∫0 cosh πx dx = 2 sec 2 , − π < a < π.
Solution: Consider
eaz
∫C cosh πz
dz = ∫C f (z ) dz ,

where C is the rectangle with corners at

1 1 1
R, R + i, − R + i and − R indented at i.
2 2 2
Within and on this contour, f (z ) is regular
and so by Cauchy’s theorem, we get
R 1 /2
∫C f (z ) dz = ∫− R f ( x) dx + ∫0 f ( R + iy) i dy
ρ 1 −R 1
+ ∫R f (x +
2
i ) dx + ∫γ f (z ) dz + ∫− ρ f (x +
2
i ) dx
0
+ ∫1 /2 f (− R + iy) i dy = 0 . ...(1)

lim lim (z − 1 i) e az
1 2 exp (ai / 2) exp (ai / 2)
Since 1 (z − i) f (z ) = 1 = = ,
z→ i 2 z → i cos h πz 1
π sinh πi πi
2 2 2
lim 1
we have
ρ→ 0 ∫γ f (z ) dz = (− πi) .
πi
. exp (ai / 2) = − exp (ai / 2),

 1 /2 f ( R + iy) i dy ≤ 1 /2  exp {aR + aiy} 

 ∫0 ∫0  i dy
  cosh π ( R + iy) 
1 /2 e a R dy
≤ ∫0 cosh πR cos πy − sinh πR sin πy

[∵ |cosh πR cos πy + i sinh πR sin πy |

≤ |cosh πR cos πy | − | i sinh πR sin πy |]
 C-171

1 /2 e aR dy
∫0 e πR
(cos πy − sin πy)
→ 0 as R → ∞, a < π.

[∵ R is large we can write e πR for cosh πR and sinh πR]

0
Similarly ∫1 /2 f (− R + iy) i dy → 0 as R → ∞.

Hence when R → ∞, and ρ → 0, we get from (1),

∞ 0 1 −∞ 1
∫ − ∞ f ( x) dx + ∫ ∞ f ( x + 2 i) dx − e + ∫0
ai /2
f (x + i) = 0
2
∞ ∞ 1
∫ − ∞ f ( x) dx − ∫ − ∞ f ( x + 2 i) dx = e
ai /2
or

 exp {a ( x + i)}
1
∞  eax 
or ∫− ∞ cosh πx −
2
1 
dx = e ai /2
 cosh π ( x + i) 
 2 
∞  e ax e a i /2  1 1
∫ − ∞ cosh πx
ax
or − e ⋅  dx = (cos a + i sin a)
i sinh πx  2 2

Equating real and imaginary parts,

1
sin a . e ax
∞ e ax ∞ 1
∫− ∞ cosh πx
dx − ∫− ∞ 2
sinh πx
dx = cos a
2
...(1)

1
e ax cos a
∞ 2 dx = sin 1 a or ∞ e ax 1
and ∫− ∞ sinh πx 2 ∫− ∞ sinh πx
dx = tan a.
2
...(2)

From (1) and (2), we get

∞ e ax 1 1 1
∫− ∞ cosh πx
dx − sin a tan a = cos a
2 2 2
∞ e ax 1
or ∫− ∞ cosh πx
dx = sec a
2
0 e ax ∞ e ax 1
or ∫− ∞ cosh πx
dx + ∫0 cosh πx
dx = sec a.
2
Putting x = − y in the first integral, we get

0 e − ay ∞ e ax 1
− ∫∞ cosh πy
dy + ∫0 cosh πx
dx = sec a
2
∞ e ax + e − ax 1 ∞ cosh ax 1 1
or ∫0 cosh πx
dx = sec a
2
or ∫0 cosh πx
dx = sec a.
2 2

Example 35: By integrating e iz / √ z along a suitable path prove that

∞ cos x ∞ sin x π
∫0 √x
dx = ∫0 √x
dx =
2
⋅
 C-172

e iz
Solution: Consider ∫C √z
dz = ∫C f (z ) dz ,

where C is a quadrant of a circle indented at the origin

which is the centre of the circle. Since f (z ) is regular
within and on C,therefore by Cauchy’s theorem, we have
R
∫C f (z ) dz = ∫ρ f ( x) dx + ∫Γ f (z ) dz

ρ
+ ∫R f (r e i π /2 ) e i π /2 dr

+ ∫γ f (z ) dz = 0 . …(1)

 π /2  exp (i R e i θ ) 
f (z ) dz≤  1 /2 i θ /2 i R e i θ dθ
 ∫Γ ∫0
Now
  R e 
π /2
≤ ∫0 e − R sin θ R1 /2 dθ
π /2
≤ ∫0 e −(2 θ / π )R R1 /2 dθ, by Jordan’s inequality

π
= (1 − e − R ) → 0 as R → ∞.
2√R
 f (z ) dz ≤
0
e −ρ sin θ ρ1 /2 dθ ≤
0

 ∫γ ∫ π /2 ∫ π /2
Similarly, ρ1 /2 dθ,

since e − ρ sin θ ≤ 1 for small values of ρ
→ 0 as ρ → 0 .
Hence when R → ∞, ρ → 0 , we have from (1)
i π /2
∞ e ix ∞ e ir e
∫0 √x
dx − i ∫
0 √ r e i π /4
dr = 0

∞ e ix ∞
or ∫0 dx = i e − i π /4 ∫0 r − 1 /2
e − r dr
√x
π
= i 
1 i  1
−  Γ = (1 + i) ⋅
 √ 2 √ 2 2 2
Equating real and imaginary parts on both sides, we get
∞ cos x π ∞ sin x π
∫0 √x
dx =
2
, ∫0 √x
dx =
2
⋅

15 To find the Residue by Knowing the Integral First

So far we evaluated the integrals with the help of residues of the function f (z ) and
Cauchy’s residue theorem. Now we shall illustrate how can we find the residues of f (z )
by integrating f (z ) and then using the residue theorem.
 C-173

1
Example 36: Find the residue of tann −1 πz at z = , where n is an even positive integer.
2
n −1
n −1  sin πz 
Solution: Let f (z ) = tan πz =   .
 cos πz 
Poles of f (z ) are given by
cos πz = 0 or e iπz + e − iπz = 0

or e i 2 πz = − 1 = e(2 r + 1) πi
1
or z= (2 r + 1), r = 0 , ± 1, ± 2,……
2
Consider a rectangular contour C with sides x = 0 , 1,
1
y = ± R. The only pole which lies inside C is z = ⋅
2
1
By Cauchy’s residue theorem, we have the residue at (z = )
2
1  
2πi  ∫ AB ∫ BC ∫ CD ∫ DA
= f (z ) dz + f (z ) dz + f (z ) dz + f (z ) dz  ,

…(1)
n −1
where f (z ) = tan πz .

On BC, we have z = 1 + iy, dz = i dy,

R
∴ ∫ BC tann − 1 π z dz = ∫ −R tann − 1 {π (1 + iy)} i dy
R
= ∫ −R i tann − 1 iπy dy

and on DA, we have z = iy, dz = i dy,

−R
∴ ∫ DA tann − 1 πz dz = ∫R tann − 1 (πiy) i dy

R
=− ∫− R i tann − 1 (πiy) dy.

Thus ∫ BC tann − 1 πz dz + ∫ DA tann − 1 πz dz = 0 .

1
Hence from (1), we have the residue at (z = )
2
1  
tann − 1 πz dz + tann − 1 πz dz 
2 πi  ∫ AB ∫ CD
=

1  1 0 
tann − 1 π ( x − iR) dx + tann − 1 π ( x + iR) dx  ⋅
2 πi  ∫0 ∫1
=

 C-174

1 exp iπ ( x − iR) − exp {− iπ ( x − iR)}

Now tan π ( x − iR) =
i exp iπ ( x − iR) + exp {− iπ ( x − iR)}
1 1 − exp {− 2 iπ ( x − iR)}
=
i 1 + exp {− 2 iπ ( x − iR)}
1 1 − exp (− 2 iπx) exp (− 2 πR) 1
= → as R → ∞.
i 1 + exp (− 2 iπx) exp (− 2 πR) i
1
Similarly tan (πx + iπR) → − as R → ∞.
i
1
∴ the residue at z = is
2
n −1 n −1
1  1 1 0  − 1 
2 πi  ∫0  i ∫1
=    dx +   dx
 i 

1  1 (− 1)n − 1 
=  n −1 − 
2 πi  i i n − 1 
1  1 1 
= + , n is an even positive integer
2 πi  i n − 1 i n − 1 
1 1
= n = (− 1)n /2 .
πi π

16 Hurwitz Theorem
Theorem: Let the function f n (z ) be analytic and non-zero in a domain D and let f n (z )
converge to f (z ) uniformly on every closed and bounded subset of D. Then f (z ) is either
identically zero or never equal to zero in D.
Proof. If we assume that f (z ) is not identically zero then we have to show that f (z ) is
never equal to zero in D.
Let f (z0 ) = 0 for some z0 ∈ D. Since zeros of an analytic function are isolated, therefore
there exists a number r > 0 such that f (z ) is defined and non-zero for 0 < | z − z0 | ≤ r.
In particular, f (z ) ≠ 0 on the circle Γ,| z − z0 | = r.
Let ε = min {| f (z )|: z ∈Γ}.
Then ε ≤| f (z )| for all z ∈Γ.
Since f n (z ) converges to f (z ) uniformly on Γ therefore for sufficiently large n,we have
| f n (z ) − f (z )| < ε, for all z ∈Γ.
Therefore by Rouche’s theorem the functions f (z ) and f n (z ) have the same number of
zeros inside Γ. But f (z ) has a zero at z = z0 so f n (z ) must have a zero inside Γ. But
according to given statement, f n (z ) is non-zero in D. Therefore f (z ) can never be zero
in D.
 C-175

17 The Jensen and Poisson–Jensen Formulae

Theorem 1: Jensen’s Formula: Let f (z ) be an analytic function in the closed disc
| z | ≤ R, assume f (0 ) ≠ 0 and f (z ) ≠ 0 on| z | = R . If z1, z2 , ……, z n are the zeros of f (z ) in
the open disc | z | < R , each repeated as often as its multiplicity, then
n  R  2π
 + 1 log f ( R e i φ ) dφ.
log | f (0 )| = − ∑ log 
i =1
| z | 2 π
 i 
∫0
n R2 − z i z
Proof: Let F (z ) = f (z ) ∏ ⋅ …(1)
i =1 R (z − z i)

Then F (z ) ≠ 0 for| z | ≤ R and F (z ) is analytic in any domain in which f (z ) is analytic.

Therefore F (z ) is analytic and non–zero on an open disc | z | < ρ for some ρ > R.
 n R2 − z i z
Now | F (z )| = ∏ | f (z )|
 i = 1 R (z − z i)
 n R2 − z i Re i φ  iφ
= ∏ 2 iφ | f (z )|, putting z = R e
 i = 1 R e − Rz i 
n  R (R − z i e i φ) 
= | f (z )| ∏  iφ

i = 1 Re ( R − z i e − i φ )
n  R − z e i φ
= | f (z )| ∏  i
−iφ

i = 1 R − z i e 
=| f (z )| [ ∵ Moduli of the conjugates are same]
log F (z ) is analytic in the region| z | < ρ since F (z ) is analytic and non-zero in| z | < ρ. As
a result, the real part of log | F (z )| is harmonic in this region.
∴ by Gauss-Mean value theorem , we have
1 2π
log | F ( R e i φ )| dφ.
2 π ∫0
log | F (0 )| = …(2)

From (1), we have

n  R 
F (0 ) = f (0 ) ∏  − 
i =1  z i 

n  R n R
∴ | F (0 )| = | f (0 )| ∏ −  = | f (0 )| ∏
i = 1 z i  i =1 | z i |
n R
and log | F (0 )| = log | f (0 )| + ∑ log ⋅
i =1 | zi|

Substituting these values in (2), we get

n R 1 2π
log | f (0 )| + ∑ log = ∫0 log f |( R e iφ )| dφ
i =1 | z i | 2π
 C-176

n R 1 2π
or log | f (0 )| = − ∑ log + ∫0 log | f ( R e iφ )| dφ.
i =1 | z i | 2π

Corollary: In addition to the function f (z ) defined in theorem 1 above, let g (z ) also satisfy the
hypothesis of the above theorem with zeros in | z | < R at w1 , w2 , ……, wm and
h (z ) = f (z ) / g(z ). Then
m R n R 1 2π
log | h (0 )| = ∑ log
j =1
− ∑ log
| w j | k =1
+
| z k | 2π ∫0 log | h ( Re iφ )| dφ.

Proof: We have
n R 1 2π
log | f (0 )| = − ∑ log + ∫0 log | f ( R e iφ )| dφ
k =1 | z k | 2π
m R 1 2π
and log | g (0 )| = − ∑ log + ∫0 log | g ( R e iφ )| dφ,
j =1 | w j | 2π
by theorem 1, above.
R m R n
Then log | f (0 )| − log | g (0 )| = ∑ log − ∑ log
j =1 | w j | k =1 | z k|

1 2π 1 2π
+ ∫0 log | f ( R e iφ )| dφ − ∫0 log| g ( R e iφ )| dφ
2π 2π
 f (0 ) m R n R 1 2π  f ( R e iφ )
or log   = ∑ log
 g (0 )  j = 1
− ∑ log
| w j | k =1
+
| z k | 2π ∫0 log  iφ
 g ( R e )
 dφ

m R n R 1 2π
or log | h (0 )| = ∑ log
j =1
− ∑ log
| w j | k =1
+
| z k | 2π ∫0 log | h ( Re iφ )| dφ.

Comprehensive Exercise 6

e iaz
1. Integrating 2 πz
, (a is real) round the rectangle of sides x = 0, x = R,
e −1
y = 0 , y = 1, indented at 0 and i, prove that
∞ sin ax 1 1 1
∫0 e2 πx − 1 dx = 4 coth 2 a − 2a ⋅
2. Integrating log (1 − e i 2 z ) round a suitable contour, prove that
π
∫0 log sin x dx = − π log 2.
2
3. By integrating e − z round the rectangle whose vertices are 0, R, R + ia, ia, show
that
∞ 2 1 2
∫0 e − x cos (2 ax) dx = √ π e− a
2
∞ 2 2 a x2
and ∫0 e − x sin (2 ax) dx = e − a ∫0 e dx.
 C-177

z
4. By integrating round the rectangle with vertices at ± π, ± π + iR ,
a − e −iz
 π log (1 + a) , if 0 < a < 1
π x sin x dx a
prove that ∫0 =
1 + a − 2 a cos x  π log 1 + a  , if a > 1.
2
 
 a  a 
exp (iaz 2 ) 1
5. By integrating round the rectangle with vertices ± R ± i, show that
sinh πz 2
if 0 < a ≤ π,
∞ cosh ax 1 a ∞ cosh ax 1 a
∫0 cos (ax2 )
cosh πx
dx = cos
2 4
and ∫0 sin (ax2 )
cosh πx
dx = sin ⋅
2 4
2 ∞ sin x2 π
6. By integrating e iz / z round a suitable contour, prove that ∫0 x
dx = ⋅
4
∞ sin x π
Deduce that ∫0 x
dx = ⋅
2 (Gorakhpur 2004, 07, 10)

Objective Type Questions

Multiple Choice Questions

Indicate the correct answer for each question by writing the corresponding letter from (a),
(b), (c ) and (d).
1
1. The number of poles of f (z ) = inside the circle | z | = 1 is
z (z 2 + 3) (z 2 + 2)3
(a) 1 (b) 9
(c) 5 (d) 2.

dz
2. If C is circle | z | = 3, then the value of ∫ is
C (z − 1) (z + 1)
(a) 1 (b) 0
(c) 2 (d) none of these.
z +1
3. The residue of the function f (z ) = at z = 1 is
(z − 1) (z − 2)
(a) − 2 (b) 2
(c) 1 (d) − 1.
1
4. The residue of at z = ∞ is
z − z5
3

(a) 1 (b) 0
(c) − 1 (d) 2.
 C-178

 1 
5. The residue of z 3 cos   at z = 2 is
 z − 2
143 22
(a) (b) −
124 123
143
(c) − (d) none of these.
24
1
6. The residue of the function at z = i is
(z + 1)3
2

3 3i
(a) (b)
16i 16
3
(c) − (d) none of these.
16i
2π dθ
7. The value of ∫ , a > b > 0 is
0 a + b cos θ
π 2π
(a) 2 2
(b)
√ (a − b ) √ (a − b2 )
2

3π
(c) (d) none of these.
√ (a2 − b2 )
∞ dx
8. The value of ∫ is
0 1 + x2
(a) 0 (b) 1
π
(c) (d) none of these.
2
e miz
9. If f (z ) = , then residue of f (z ) at z = ai is
z 2 + a2
e− m a e ma
(a) (b)
2 ia 2 ia
e −m a
(c) (d) none of these.
2a
sin 2 z
10. If f (z ) = , the residue of f (z ) at z = − 1 is
(z + 1)3
(a) 2 sin 2 (b) − 2 sin 2
(c) 0 (d) none of these.

A nswers
1. (a) 2. (b) 3. (a) 4. (b) 5. (c)
6. (a) 7. (b) 8. (c) 9. (a) 10. (a)

¨
 C-179

1 Dirichlet’s Conditions
function f ( x) is said to satisfy Dirichlet's conditions in the interval (a, b), if
A (i) f ( x) is defined and is single-valued except possibly at a finite number of points
in the interval (a, b), and
(ii) f ( x) and f ′ ( x) are piecewise continuous in the interval (a, b).
These conditions play an important role in the study of Fourier series and Fourier
Transforms.

2 Fourier Series
(Meerut 2013, 13B; Rohilkhand 14)
If f ( x) is a periodic function with period 2 l i. e. f ( x + 2 l ) = f ( x) and satisfies
Dirichlet's conditions in the interval (− l, l ), then at every point of continuity we have
∞
1  nπx nπx 
f ( x) = a0 + ∑ an cos + bn sin …(1)
2 n =1  l l 
 C-180

1 l nπx
where an =
l ∫ −l f ( x) cos
l
dx …(2)

1 l nπx
and bn =
l ∫ − l f ( x) sin l dx. …(3)

The series (1) with coefficients an and bn given by (2) and (3) respectively is called the
Fourier series of f (x), and the coefficients an and bn are called the Fourier coefficients
corresponding to f ( x).
At a point of discontinuity
1
f ( x) = [ f ( x + 0 ) + f ( x − 0 )].
2
If the function f ( x) defined in the interval (− l, l ) be an even function of x i.e. if
f (− x) = f ( x), then
1 l nπx 2 l nπx
an = ∫ f ( x) cos dx = ∫ f ( x) cos dx
l − l l l 0 l
1 l nπx
and bn = ∫ f ( x) sin dx = 0 .
l − l l
Therefore in this case we get Fourier cosine series.
Again if f ( x) be an odd function of x i.e. if f (− x) = − f ( x), then
1 l nπx
an = ∫ f ( x) cos dx = 0
l −l l
1 l nπx 2 l nπx
bn = ∫
l ∫0
and f ( x) sin dx = f ( x) sin dx
l −l l l
and thus in this case we get Fourier sine series.
Note: If f ( x) is a function of period 2l but is defined only in (0 , l ), we can extend it to
(− l, 0 ) so as to be an even or an odd function of x in the interval (− l , l )

3 Fourier’s Integral Formula (Fourier’s Integral Theorem)

Let f ( x) be a function satisfying Dirichlet's conditions in every finite interval − l ≤ x ≤ l
1
and defined as [ f ( x + 0 ) + f ( x − 0 )] at every point of discontinuity.
2
∞
Further let ∫ − ∞ | f ( x)| dx converge i. e., f ( x)is absolutely integrable in − ∞ < x < ∞,then by
Fourier’s integral formula
1 ∞  ∞ 
f ( x) =
2π ∫ −∞ f (v)  ∫
 −∞
cos w ( x − v) dw dv

…(1)

1 ∞ ∞
or f ( x) =
2π ∫ −∞ dw ∫ −∞ f (v) cos w ( x − v) dv.

The representation (1) of f ( x) is known as Fourier’s integral formula.

The proof is out of scope of this book.
 C-181

Other Forms of Fourier Integral Formula (Fourier Integral Theorem):

1. Complex or exponential Form of Fourier Integral Formula:
We have
1 ∞  ∞ 
0=
2π ∫ −∞ f (v)  ∫
 −∞
sin w ( x − v) dw dv

…(2)

Multiplying both sides of (2) by i and then adding to (1), we get

1 ∞  ∞ iw ( x − v ) 
2 π ∫ −∞
f ( x) = f (v)  ∫ e dw dv
 −∞ 
1 ∞ ∞
or f ( x) = ∫ −∞ e iw x dw ∫ f (v) e − iwv dv.
2π −∞

2. General Form of Fourier Integral Formula:

1 ∞  ∞ 
(i) f ( x) = ∫
π −∞
f (v)  ∫
 0
cos w( x − v) dw dv.

∞
(ii) f ( x) = ∫0 { P (w)cos wx + Q (w)sin wx} dw

1 ∞
π ∫− ∞
where P (w) = f (v)cos wv dv

1 ∞
π ∫− ∞
and Q (w) = f (v)sin wv dv.

3. Fourier Sine Integral Formula:

If f ( x) is an odd function, then
2 ∞  ∞ 
f ( x) =
π ∫0
sin wx  ∫
 0
f (v)sin wv dv dw.

4. Fourier Cosine Integral Formula:
If f ( x) is an even function, then
2 ∞  ∞ 
π ∫0
f ( x) = cos wx  ∫ f (v)cos wv dv dw.
 0 

Example 1: Using Fourier integral formula prove that

2 ∞ (u2 + 2)cos ux
e − x cos x =
π ∫0
du.
(u4 + 4) (Gorakhpur 2007)

Solution: We know that the Fourier cosine integral formula is

2 ∞  ∞ 
f ( x) = ∫ cos ux ∫ f (v)cos uv dv du. ...(1)
π 0  0 
Now let f ( x) = e − x cos x
then f (v) = e − v cos v.
 C-182

∴ form (1), we get

2 ∞  ∞ 
e − x cos x = ∫ cos ux ∫ e − v cos v cos uv dv du
π 0  0 
1 ∞  ∞ 
= ∫ cos ux ∫ e − v (2 cos uv cos v) dv du
π 0  0 
1 ∞  ∞ 
cos ux ∫ e − v {cos (u + 1) v + cos (u − 1) v } dv du
π ∫0
=
 0 
1 ∞  ∞ ∞ 
cos ux ∫ e − v cos(u + 1) vdv + ∫ e − v cos (u − 1)v dv du
π ∫0
=
 0 0 
1 ∞  1 1 
π ∫0
= cos ux  +  du
2
1 + (u + 1) 1 + (u − 1)2 
 ∞ −a x a 
∵ ∫0 e cos bx dx = 
 a + b2 
2

1 ∞ (u2 + 2 − 2 u) + (u2 + 2 + 2 u)
π ∫0
= cos ux du
(u2 + 2 + 2 u)(u2 + 2 − 2 u)

1 ∞ (2 u2 + 4)cos ux 2 ∞ (u2 + 2)cos ux

π ∫0 (u2 + 2)2 − (2 u)2 π ∫0
= du = du.
u4 + 4
Example 2: Using Fourier integral method, prove that
2 (b2 − a2 ) ∞ u sin ux du
e − ax − e − bx = ∫ ⋅
π 0 (u + a2 )(u2 + b2 )
2

(Gorakhpur 2008, 11, 13)

Solution: We know that the Fourier sine integral formula is
2 ∞  ∞ 
f ( x) = ∫ sin ux ∫ f (v)sin uv dv du. ...(1)
π 0  0 
Let f ( x) = e − ax − e − bx then f (v) = e − av − e − bv .
∴ From (1), we get
2 ∞  ∞ 
e − ax − e − bx = ∫ sin ux ∫ (e − av − e − bv ) cos uv dv du
π 0  0 
2 ∞  ∞ ∞ 
= ∫ sin ux ∫ e − av sin uv dv − ∫ e − bv sin uv dv du
π 0  0 0 
2 ∞  u u 
π ∫0
= sin ux  −  du
 a2 + u2 b2 + u2 
 ∞ − ax u 
∵ ∫0 e cos bx dx = 
 a + b2 
2

2 ∞ u (b2 + u2 − a2 − u2 )
π ∫0
= sin ux du
(a2 + u2 )(b2 + u2 )
2 ∞ u (b2 − a2 ) sin ux
π ∫0 (a2 + u2 )(b2 + u2 )
= du
 C-183

2 (b2 − a2 ) ∞ u sin ux du
Hence, e − ax − e − bx = ∫ ⋅
π 0 (a + u2 )(b2 + u2 )
2

 0, when x ≤ 0 or x ≥ π
Example 3: Suppose a function f ( x) is given by f ( x) = 
 sin x, when 0 ≤ x ≤ π.
1 ∞ cos u (π − x) + cos ux
Using Fourier integral formula, show that f ( x) = ∫ du.
π 0 1 − u2
Solution: The Fourier integral formula is
1 ∞ ∞
f ( x) = ∫ [ f (v) ∫ cos u (v − x) dv] du, − ∞ < x < ∞. ...(1)
π −∞ 0
 0, when x ≤ 0 or x ≥ π
Given that f ( x) = 
sin x, when 0 ≤ x ≤ π.
 0, when v ≤ 0 or v ≥ π
∴ f (v) =  ...(2)
sin v, when 0 ≤ v ≤ π .
∞ 0
Now ∫− ∞ f (v)cos u (v − x) dv = ∫− ∞ f (v)cos u (v − x) dv

π ∞
+∫ f (v)cos u (v − x) dv + ∫π f (v)cos u (v − x) dv
0
π
= ∫0 sin v cos u (v − x) dv [From (2)]

1 π
2 ∫0
= [sin{(u + 1) v − ux } − sin { (u − 1) v − ux } ] dv

[∵ 2 cos A sin B = sin ( A + B) − sin ( A − B)]

π
1  − cos {(u + 1) v − ux} cos { u − 1} v − ux 
= +
2  u +1 u −1 
0
π
1 (u + 1)cos X − (u − 1)cos Y 
=   , where X = (u − 1) v − ux
2 (u − 1)(u + 1) 0 and Y = (u + 1) v − ux
π
 u (cos X − cos Y ) + (cos X + cos Y )
= 
 2 (u2 − 1) 0
X +Y Y − X  X +Y Y − X 
2 u sin   sin   + 2 cos   cos  
 2   2   2   2 
=
2 (u2 − 1)
1 π
= [u sin (v − x) u sin v + cos (v − x) u cos v]
2 0
(u − 1)
[ Putting the values of X and Y ]
1
= [cos (π − x) u cos π − cos (− xu)]
(u2 − 1)
1
= [− cos (π − x) u − cos xu]
(u2 − 1)
[∵ cos π = −1 and cos (− x) = cos x ]
 C-184

1
= [cos (π − x) u + cos xu].
(1 − u2 )
∞ cos u (π − x) + cos ux
Therefore, ∫− ∞ f (v)cos u (v − x) dv =
1 − u2
⋅ ...(3)

Hence from equation (1), we get

1 ∞ cos u (π − x) + cos ux
f ( x) = ∫ du.
π 0 (1 − u2 )

Comprehensive Exercise 1

1. Using Fourier cosine integral formula, prove that

∞ cos λx π −x
∫0 λ2 + 1 dλ = 2 e , x ≥ 0. (Gorakhpur 2016)

2. Using Fourier cosine integral formula, show that

2 a ∞ cos ux
e − ax =
π ∫ 0 u2 + a2
dx, a > 0 , x ≥ 0 .

 1, when | x|≤ 1
3. Express the function f ( x) =  as a Fourier integral.
0 , when | x|> 1
∞ sin λ cos λ x
Hence evaluate ∫ dλ .
0 λ
 1, when 0 ≤ x ≤ π
4. Express f ( x) =  as a Fourier sine integral.
0 , when x > π
∞ 1 − cos πλ
Hence evaluate ∫ sin xλ dλ .
0 λ
5. Using Fourier sine integral formula, show that
6 ∞ u sin ux du
e x − e2 x = ∫ , x ≥ 0.
π 0 (u2 + 1)(u2 + 4)

6. Using Fourier integral formula, show that

1 ∞ u [sin ux − sin u (π − x)]
f ( x) = ∫ du
π 0 u2 − 1
0 , when x ≤ 0 and x ≥ π
where f ( x) is given by f ( x) = 
cos x, when 0 ≤ x ≤ π .

A nswers 1

 π / 2, | x|< 1  π / 2, 0 ≤ x ≤ π π
3.  4.  ;
 0, | x|> 1  0, x > π 4
 C-185

4 Fourier Transform or Complex Fourier Transform

(Gorakhpur 2005, 07; Purvanchal 14)
Let f ( x) be a function defined on (− ∞, ∞) and be piecewise continuous in each finite
partial interval and absolutely integrable in (− ∞, ∞), then
1 ∞
F { f ( x)} =
√ (2π) ∫ − ∞
e ipx f ( x) dx

~
is called the Fourier Transform of f ( x) and is denoted by F { f ( x)} or f ( p).
~
The function f ( x) is called the inverse Fourier transform of f ( p) i.e.,
~
f ( x) = F −1 { f ( p)}.

5 Inversion Theorem for Complex Fourier Transform

~
If f ( p) is the Fourier transform of f ( x) and if f ( x) satisfies the Dirichlet's conditions in every
∞
finite interval (− l, l ) and further if ∫ − ∞ | f ( x)| dx is convergent, then at every point of
continuity of f ( x),
1 ∞ ~
f ( x) = ∫ −∞ f ( p) e − ipx dp.
√ (2π) (Gorakhpur 2007)
Proof: From Fourier integral formula, we have
1 ∞  ∞ iw ( x − v ) 
f ( x) =
2π ∫ − ∞
f (v)  ∫
 −∞
e dw dv

1 ∞ ∞
f (v) e − iwv dv
2 π ∫ −∞
= e iwx dw ∫
−∞
1 ∞ 1 ∞
= ∫ −∞ e − ipx dp ∫ −∞ f ( x) e ipx dx,
√ (2 π) √ (2 π)
putting w = − p so that dw = − dp
1 ∞ ~
= ∫ −∞ e − ipx f ( p) dp.
√ (2 π)
Note: Some authors also define the Fourier transform in the following forms :
~ ∞
(1) f ( p) = ∫ e − ipx f ( x) dx
−∞

1 ∞ ~
and f ( x) =
2π ∫ −∞ f ( p) ⋅ e ipx dp.

~ ∞
(2) f ( p) = ∫ e ipx f ( x) dx
−∞

1 ∞ ~
and f ( x) = ∫ −∞ e − ipx f ( p) dp.
2π
 C-186

~ 1 ∞
(3) f ( p) = ∫ −∞ e − ipx f ( x) dx
√ (2π)
1 ∞ ~
and f ( x) =
√ (2π) ∫ −∞ f ( p) ⋅ e ipx dp.

6 Fourier Sine Transform

~
The infinite Fourier sine transform of f ( x), 0 < x < ∞, is denoted by Fs { f ( x)} or f s ( p) , and is
defined as
~  2 ∞
Fs { f ( x)} = f s ( p) =  
 π ∫0 f ( x) sin px dx .

~
The function f ( x) is called the inverse Fourier sine transform of f s ( p)
~
i. e., f ( x) = Fs −1 { f s ( p)}.
Note: Some authors also define
~ ∞
f s ( p) = ∫ f ( x) sin px dx.
0

7 Inversion Formula for Fourier Sine Transform

~
If f s ( p) is the Fourier sine transform of the function f ( x) which satisfies the Dirichlet's conditions
∞
in every finite interval (0 , l ) and is such that ∫0 | f ( x)| dx exists, then

 2 ∞ ~
f ( x) =  
 π ∫0 f s ( p) sin px dp

at every point of continuity of f ( x).

This is an inversion formula for infinite Fourier sine transform.
Proof: From Fourier integral formula, we have
1 ∞ ∞
f ( x) =
π ∫0 dw ∫ −∞ f (v) cos w ( x − v) dv

1 ∞ ∞
=
π ∫0 dp ∫ −∞ { f (v) cos px cos pv + f (v) sin px sin pv} dv, where w = p

1 ∞ ∞ 1 ∞ ∞
=
π ∫0 cos px dp ∫ −∞ f (v) cos pv dv +
π ∫0 sin px dp ∫ −∞ f (v) sin pv dv

1 ∞ ∞
or f ( x) =
π ∫0 cos px dp ∫ −∞ f ( x) cos px dx

1 ∞ ∞
+
π ∫0 sin px dp ∫ −∞ f ( x) sin px dx …(1)
 C-187

Now define f ( x) in (− ∞, 0 ) such that f ( x) is an odd function of x in (− ∞, ∞). Then

obviously f ( x) cos px is an odd function of x and f ( x) sin px is an even function of x in
(− ∞, ∞).
∴ From (1), we have
 2 ∞  2 ∞
f ( x) =  
 π ∫0 sin px dp ×  
 π ∫0 f ( x) sin px dx

 2 ∞ ~
or f ( x) =  
 π ∫0 f s ( p) sin px dp .

Note: According to the authors who define

~ ∞
f s ( p) = ∫0 f ( x) sin px dx,

2 ∞ ~
we have f ( x) =
π ∫0 f s ( p) sin px dp.

8 Fourier Cosine Transform

The infinite Fourier cosine transform of f ( x), 0 < x < ∞, is denoted by Fc { f ( x)} or
~
f c ( p), and is defined as
~  2 ∞
Fc { f ( x)} = f c ( p) =  
 π ∫0 f ( x) cos px dx.

~
The function f ( x) is called the inverse Fourier Cosine transform of f c ( p)
~
i.e., f ( x) = Fc −1 { f c ( p)}
~ ∞
Note: Some authors also define f c ( p) =
0 ∫ f ( x) cos px dx .

9 Inversion Formula for Fourier Cosine Transform

~
If f c ( p) is the Fourier cosine transform of the function f ( x) which satisfies the Dirichlet's
∞
conditions in every finite interval (0 , l ) and is such that ∫0 | f ( x)| dx exists, then

 2 ∞ ~
f ( x) =  
 π ∫0 f c ( p) cos px dp

at every point of continuity of f ( x).

This is an inversion formula for infinite Fourier cosine transform.
Proof: Proceeding as in article 7, we have
1 ∞ ∞
f ( x) =
π ∫ 0
cos px dp ∫
− ∞
f ( x) cos px dx

1 ∞ ∞
π ∫0
+ sin px dp ∫ f ( x) sin px dx. …(1)
−∞
 C-188

Now define f ( x) in (−∞, 0 ) such that f ( x) is an even function of x in (− ∞, ∞). Then

obviously f ( x) cos px is an even function of x and f ( x) sin px is an odd function of x in
(− ∞, ∞).
∴ from (1), we have
 2 ∞  2 ∞
f ( x) =  
 π ∫0 cos px dp ×  
 π ∫0 f ( x) cos px dx

 2 ∞ ~
or f ( x) =  
 π ∫0 f c ( p) cos px dp.

Note: According to the authors who define

~ ∞
f c ( p) = ∫ f ( x) cos px dx,
0

2 ∞ ~
we have f ( x) =
π ∫0 f c ( p) cos px dx.

10 Linearity Property of Fourier Transform

~ ~
If f ( p) and g ( p) are Fourier transforms of f ( x) and g ( x) respectively, then
~ ~
F { a f ( x) + bg ( x)} = a f ( p) + b g ( p),
where a and b are constants. (Gorakhpur 2014)

Proof: We have
~ 1 ∞
F { f ( x)} = f ( p) =
√ (2π) ∫−∞ e ipx f ( x) dx

~ 1 ∞
and F { g ( x)} = g ( p) =
√ (2π) ∫−∞ e ipx g( x) dx.

∴ F { a f ( x) + b g ( x)}
1 ∞
= ∫
√ (2 π) − ∞
e ipx { af ( x) + bg ( x)} dx

a ∞ b ∞
=
√ (2 π) ∫−∞ e ipx f ( x) dx +
√ (2 π) ∫−∞ e ipx g ( x) dx

~ ~
= a f ( p) + b g ( p).

11 Change of Scale Property

~
Theorem 1: (For Complex Fourier Transform). If f ( p) is the complex Fourier
1 ~  p
transform of f ( x), the complex Fourier transform of f (ax) is f  ⋅
a  a
(Gorakhpur 2008, 10)
 C-189

Proof. We have
~ 1 ∞
f ( p) = F { f ( x)} =
√ (2π) ∫−∞ e ipx f ( x) dx. …(1)

1 ∞
Now F { f (ax)} =
√ (2π) ∫−∞ e ipx f (ax) dx

1 1 ∞
= ∫−∞ e ip( t / a ) f (t) dt,
a √ (2 π)
1
putting ax = t so that dx = dt
a
1 1 ∞
= ⋅ ∫−∞ e i( p / a ) t f (t) dt
a √ (2 π)
1 ~  p
= f   , from (1).
a  a
~
Theorem 2: (For Fourier Sine Transform). If f s ( p) is the Fourier sine transform of f ( x),
1 ~  p
then the Fourier sine transform of f (ax) is fs   ⋅
a  a
Proof: We have
~
f s ( p) = Fs { f ( x)}

 2 ∞
=  
 π ∫0 f ( x) sin px dx …(1)

 2 ∞
Now Fs { f (ax)} =  
 π ∫0 f (ax) sin px dx

1  2 ∞ p 
a  π  ∫0
= ⋅   f (t) . sin  t dt,
a 
putting ax = t so that dx = (1 / a) dt
1 ~  p
= f s   , from (1).
a  a
~
Theorem 3: (For Fourier Cosine Transform). If f c ( p) is the Fourier Cosine
1 ~  p
Transform of f ( x) , then the Fourier Cosine transform of f (ax) is f c   .
a  a
Proof: We have
~
f c ( p) = Fc { f ( x)}

 2 ∞
=  
 π ∫0 f ( x) cos px dx …(1)

 2 ∞
Now Fc { f (ax)} =  
 π ∫0 f (ax) cos px dx
 C-190

1  2 ∞ p 
= ⋅  
a  π ∫0 f (t) cos  t dt ,
a 
putting ax = t so that dx = (1 / a) dt
1 ~  p
= f c   , from (1).
a  a

12 Shifting Property
~
If f ( p) is the complex Fourier transform of f ( x), then the complex Fourier transform of f ( x − a)
~
is e ipa f ( p). (Gorakhpur 2006, 08; Purvanchal 14)

Proof: We have
~ 1 ∞
f ( p) = F { f ( x)} =
√ (2π) ∫−∞ e ipx f ( x) dx …(1)

1 ∞
Now F { f ( x − a)} =
√ (2π) ∫ −∞ e ipx f ( x − a) dx

1 ∞
= ∫ e ip( a + t ) f (t) dt,
√ (2 π) ∞−

putting x − a = t so that dx = dt
1 ∞
= e ipa ⋅
√ (2 π) ∫ − ∞
e ipt f (t) dt

~
= e ipa f ( p) , from (1).

13 Modulation Theorem
~
If f ( p) is the Complex Fourier transform of f ( x), then the Fourier transform of f ( x) cos ax is
1 ~ ~
[ f ( p − a) + f ( p + a)].
2 (Gorakhpur 2007, 08, 12)

Proof: We have
~ 1 ∞
f ( p) = F { f ( x)} =
√ (2π) ∫ −∞ e ipx f ( x) dx …(1)

Now F { f ( x) cos ax}

1 ∞ e iax + e − iax
=
√ (2 π) ∫ −∞ e ipx . f ( x) ⋅
2
dx

1 1 ∞ 1 ∞ 
=  ∫ −∞ e i( p + a ) x f ( x) dx + ∫ e i ( p − a ) x f ( x) dx
2  √ (2 π) √ (2 π) − ∞

1 ~ ~
= [ f ( p + a) + f ( p − a)].
2
 C-191

14 Important Results
~ ~
Theorem: If f s ( p) and f c ( p) are Fourier sine and cosine transforms of f ( x) respectively, then
1 ~ ~
(i) Fs { f ( x) cos ax} = [ f s ( p + a) + f s ( p − a)]
2
1 ~ ~
(ii) Fc { f ( x) sin ax} = [ f s ( p + a) − f s ( p − a)]
2
1 ~ ~
(iii) Fs { f ( x) sin ax} = [ f c ( p − a) − f c ( p + a)].
2
Proof: (i) We have
 2 ∞
Fs { f ( x) cos ax} =  
 π ∫0 f ( x) cos ax sin px dx

 2 1 ∞
=   ⋅
 π 2 ∫0 f ( x) ⋅ [sin ( p + a) x + sin ( p − a) x] dx

1   2 ∞
2   π ∫ 0
=    f ( x) ⋅ sin ( p + a) x dx

 2 ∞ 
+  
 π ∫0 f ( x) ⋅ sin ( p − a) x dx

1 ~ ~
= [ f s ( p + a) + f s ( p − a)].
2
(ii) We have
 2 ∞
Fc { f ( x) sin ax} =  
 π ∫0 f ( x) sin ax cos px dx

 2 1 ∞
=   ⋅ ∫ f ( x) [sin ( p + a) x − sin ( p − a) x] dx
 π 2 0

1   2 ∞
=   
2   π ∫0 f ( x) sin ( p + a) x dx

 2 ∞ 
−  
 π ∫0 f ( x) ⋅ sin ( p − a) x dx

1 ~ ~
= [ f s ( p + a) − f s ( p − a)].
2
(iii) We have
 2 ∞
Fs f { ( x) sin ax} =   ∫ f ( x) sin ax sin px dx
 π 0

 2 1 ∞
=   ⋅ ∫ f ( x) [cos ( p − a) x − cos ( p + a) x] dx
 π 2 0
 C-192

1   2 ∞
2   π  ∫0
=    f ( x) cos ( p − a) x dx

 2 ∞ 
−   ∫ f ( x) cos ( p + a) x dx
 π 0

1 ~ ~
= [ f c ( p − a) − f c ( p + a)].
2

15 Theorem
If φ ( p) is the Fourier sine transform of f ( x) for p > 0, then
Fs { f ( x)} = − φ (− p) for p < 0 .
Proof: We have
 2 ∞
Fs { f ( x)} =   ∫ F ( x) sin px dx
 π 0
= φ ( p), for p > 0. …(1)
For p < 0, let p = − s where s > 0.
 2 ∞
∴ Fs { f ( x)} =  
 π ∫0 f ( x) sin (− sx) dx

 2 ∞
=−  
 π ∫0 f ( x) sin sx dx = − φ (s)

= − φ (− p ), for p < 0.
Hence in general
 φ (| p |), p > 0
Fs { f ( x)} = 
 − φ (| p |), p < 0
or Fs { f ( x)} = φ (| p |). Sgn p,
 + 1, p > 0
where the symbol Sgn p = 
 − 1, p < 0 .

16 Multiple Fourier Transforms

Let f ( x, y) be a function of two variables x and y. Regarding f ( x, y), temporarily, as a
function of x, its Fourier transform is
~ 1 ∞
f ( p, y) = ∫
√ (2π) − ∞
f ( x, y) e ipx dx .

~
Now regarding f ( p, y) as a function of y, its Fourier transform is
~ 1 ∞ ~
F ( p, q) =
√ (2π) ∫ −∞ f ( p, y) e ipy dy
 C-193

~ 1 ∞ ∞
f ( x, y) e i ( px + qy )dx dy
2 π ∫ −∞ ∫ −∞
or F ( p, q) =

which is Fourier transform of f ( x, y).

Inversion formula: Using inversion formula for Fourier transforms, we have
1 ∞ ~
f ( x, y) = ∫ f ( p, y) e − ipx dp
√ (2π) − ∞
~ 1 ∞ ~
and f ( p, y) = ∫ −∞ F ( p, q) e − iqy dq.
√ (2π)
1 ∞ ~
Hence f ( x, y) = ∫ −∞ F ( p, q) e − i( px + qy )dp dq,
2π
which is the inversion formula for the Fourier transform of f ( x, y).

17 Convolution
The function
1 ∞
H ( x) = F * G =
√ (2π) ∫ −∞ F (u). G ( x − u) du

is called the convolution or Falting of two integrable functions F and G over the
interval (− ∞, ∞).
Note: Some authors also define
∞
F*G= ∫ −∞ F (u). G ( x − u) du.

18 The Convolution or Falting Theorem for Fourier

Transforms
If F { f ( x)} and F { g ( x)} are the Fourier transforms of the functions f ( x) and g ( x) respectively,
then the Fourier transform of the convolution of f ( x) and g ( x) is the product of their Fourier
transforms
i.e. F { f ( x) * g ( x)} = F { f ( x)} . F { g ( x)}.
Proof: We have F { f ( x) * g ( x)}
 1 ∞ 
= F
 √ (2 π)
∫ −∞ f (u) ⋅ g ( x − u) du

1 ∞  1 ∞ 
= ∫  ∫
√ (2 π) − ∞  √ (2 π) − ∞
f (u) g ( x − u) du e ipx dx

1 ∞  ∞ 
2π ∫ − ∞
= f (u)  ∫ g ( x − u) e ipx dx du
 − ∞ 
 C-194

1 ∞  ∞ 
= ∫ f (u)  ∫ g ( y) e ip ( u + y )dy du,
2π − ∞  −∞ 
putting x − u = y so that dx = dy,
1 ∞  ∞ ipu 
=
2π ∫ −∞ f (u)  ∫
 −∞
e g ( y) e ipy dy du

1 ∞  ∞ 
=
2π ∫ −∞ f (u)  e ipu ∫
 − ∞
g ( y) e ipy dy du

1 ∞  1 ∞ 
= ∫
√ (2 π) − ∞
f (u) e ipu

∫
√ (2 π) − ∞
g ( x) e ipx dx du

1 ∞
√ (2 π) ∫ − ∞
= f (u) [e ipu F { g ( x)}] du

 1 ∞ 
= ∫
 √ (2 π) − ∞
f (u) e ipu du F { g ( x)}

 1 ∞ 
=
 √ (2 π ) ∫ −∞
f ( x) e ipx dx F { g( x)}

= F { f ( x)} ⋅ F { g ( x)} .

19 Relationship Between Fourier and Laplace

Transforms
(Gorakhpur 2011)
Let us consider the function
 e − x t g(t), t > 0
f (t) =  …(1)
0 , t < 0.
∴ The Fourier transform of f (t) is given by
∞
F { f (t)} = ∫ −∞ e ipt . f (t) dt, (Taking non-symmetrical form of F. T.)

0 ∞
=∫ 0 ⋅ e ipt dt + ∫0 e − xt g(t) ⋅ e ipt dt
−∞
∞ ∞
= ∫0 e( ip− x ) t g (t) dt = ∫0 e − st g (t) dt , putting x − ip = s

= L {g (t)}.
Hence the Fourier transform of the function f (t) defined by (1) is the Laplace transform of the
function g (t).

20 Fourier Transform of The Derivatives of a Function

~ ~
(a) The Fourier transform of f ′ ( x) , the derivative of f ( x) is − ip f ( p), where f ( p) is the
Fourier transform of f ( x). (Gorakhpur 2015)
 C-195

Proof: By definition
1 ∞
F { f ′ ( x)} =
√ (2π) ∫ −∞ f ′ ( x) ⋅ e ipx dx

1 ∞ f ( x + h) − f ( x) ipx
=
√ (2 π) ∫ − ∞
lim
h→ 0 h
⋅ e dx

1 ∞ f ( x + h) ipx 1 ∞ f ( x) ipx
= lim
h→ 0 √ (2 π)
∫ − ∞ h
⋅ e dx − lim
h→ 0 √ (2 π)
∫ −∞ h
⋅ e dx

~
1 ∞ f ( x + h) ip( x + h ) − iph f ( p)
= lim
h→ 0 √ (2π)
∫ − ∞ h
⋅e e d ( x + h) − lim
h→ 0 h
~
e − iph ∞ 1 f ( p)
= lim ∫
h→ 0 √ (2 π) − ∞ h
⋅ f ( y) e ipy dy − lim
h→ 0 h

~ ~
e − iph f ( p) f ( p)
= lim − lim
h→ 0 h h→ 0 h
~ e − iph − 1 ~
= f ( p) ⋅ lim = (− ip) f ( p).
h→ 0 h

(b) The Fourier transform of f n( x), the nth derivative of f ( x) is (− ip)n times the Fourier
transform of f ( x) provided that the first (n − 1) derivatives of f ( x) vanish as x → ± ∞ .
Proof: By definition
1 ∞
F { f n ( x)} = ∫ −∞ f n ( x) . e ipx dx.
√ (2π)
Integrating by parts, we have
1 1 ∞
F { f n ( x)} = [ f n−1( x) ⋅ e ipx ] ∞
−∞ − ∫ f n−1( x) ipe ipx dx
√ (2 π) √ (2 π) ∞−

(− ip) ∞ lim
= ∫ −∞ f n−1 ( x) e ipx dx, since f n−1 ( x) = 0 .
√ (2 π) x→ ± ∞

Repeating the same process of integration by parts (n − 1) times more, we have

1 ∞
F { f n ( x)} = (− ip)n ∫ f ( x) e ipx dx
√ (2π) − ∞

~ ~
or f n ( p) = (− ip)n f ( p).
(c) The Fourier cosine and sine transforms of the derivatives of f ( x) are given by
~ n −1 ~
 2
f c2 n ( p) = −  
 π ∑
(− 1)r α2 n−2 r −1 p2 r + (−1)n p2 n f c ( p);
r=0
~ n ~
 2
f c2 n+1 ( p) = −   ∑ (− 1)r α2 n−2 r p2 r + (−1)n p2 n+1 f s ( p);
 π
r =1
 C-196

~ n ~
 2
f s2 n ( p) = −  
 π ∑ (− 1)r α2 n−2 r p2 r −1 + (−1)n+1 p2 n f s ( p);
r =1
~ n ~
 2
and f c2 n+1 ( p) = −   ∑ (− 1)r α2 n−2 r −1 p2 r −1 + (−1)n+1 p2 n+1 f c ( p);
 π
r =1

provided that first (n − 1) derivatives of f ( x) vanish as x → ∞ and

d n−1 f
→ α n−1 etc . as x → 0 .
dx n−1
Proof: By definition, we have
~ ∞
 2
f cn ( p) =   ∫0 f n ( x) cos px dx …(1)
 π
~ ∞
 2
and f cn ( p) =   ∫0 f n ( x) sin px dx . …(2)
 π
Integrating R.H.S. of (1) by parts, we have
~ ∞
 2 n−1  2
f cn ( p) =  [f ( x)cos px]0∞ + p   ∫0 f n−1( x)sin px dx
 π  π
~ ~
 2
or f cn ( p) = −   α n−1 + p f sn−1( p). …(3)
 π
Similarly integrating R.H.S. of (2), we have
~ ~
f sn ( p) = − p f cn−1( p). …(4)
From (3) and (4), we have
~ ~
 2
f cn ( p) = −   α n−1 − p2 f nn−2 ( p). …(5)
 π
~ ~
By repeated application of these rules f c n ( p) is obtained as a sum of α’s and f c ′ ( p) or
~
f c ( p).
~
It is clear that f c ′ ( p) will occur when n is odd and in that case it may be replaced by

 2 ~
−   α0 + p f s ( p).
 π
Thus, we have
~ n −1 ~
 2
∑
2n
fc ( p) = −   (− 1)r α2 n−2 r −1 p2 r + (−1)n p2 n f c ( p)
 π
r=0
~ n
2n + 1  2 ~
and fc ( p) = −  
 π ∑ (− 1)r α2 n−2 r p2 r + (−1)n p2 n+1 f s ( p)
r=0

Also from (3) and (4), we have

 C-197

~  ~ 
 2
f sn( p) = − p −   α n−2 + p f sn−2 ( p)
  π 
 
~ ~
 2
or f sn( p) = p   α n−2 − p2 f sn−2 ( p).
 π
~ ~
By repeated application of these results f s n ( p) is obtained as a sum of α’s and f s ′ ( p) or
~
f s ( p).
~ ~
It is clear that f s ′ ( p) will occur when n is odd and it may be replaced by − p f c ( p).

Thus, we have
~ n ~
 2
f s2 n( p) = −  
 π ∑ (− 1)r α2 n−2 r p2 r −1 + (−1)n+1 p2 n f s ( p)
r =1
~ n
 2
and f c2 n + 1( p) = −  
 π ∑ (− 1)r α2 n−2 r +1 p2 r −1
r =1
~
+ (− 1)n+1 p2 n+1 f c ( p).
Note. The infinite sine and cosine transforms can be applied when the range of the
variable selected for exclusion is 0 to ∞.

Example 4: Find the Fourier complex transform of f ( x), if

iωx
 e a< x< b
f ( x) = 
0 x < a, x > b. (Meerut 2013B; Avadh 14)

Solution: We have
1 ∞
F { f ( x)} =
√ (2π) ∫ −∞ e ipx f ( x) dx

1  a b ipx ∞ 
⋅ e iωx dx + ∫
√ (2 π)  ∫ − ∞
= 0 ⋅ e ipx dx + ∫a e 0 ⋅ e ipx dx
b 
1 b i ( p + ω)x
= ∫
√ (2 π) a
e dx

b
1  e i( p + ω ) x 
= ⋅ 
√ (2 π)  i ( p + ω) 
a

1  ei ( p + ω ) a − ei ( p + ω ) b 
=  ⋅
√ (2 π)  p+ω 
 C-198

Example 5: Find the Fourier transform of F ( x) defined by

 1, | x | < a
F ( x) = 
 0 ,| x | > a.
(Gorakhpur 2005, 07; Kanpur 08; Meerut 13; Bundelkhand 13; Avadh 13)
and hence evaluate
∞ sin pa cos px
(a) ∫ dp,
−∞ p (Gorakhpur 2009, 11)
∞ sin p
(b) ∫0 p
dp.
(Gorakhpur 2008)

Solution: We have
~ 1 ∞
F ( p) =
√ 2π) ∫ −∞ e ipx F ( x) dx

a
1 a 1  e ipx 
=
√ (2 π) ∫ −a e ipx dx =  
√ (2 π)  ip 
−a
 e ipa − ipa 
1 e
=  − 
√ (2 π)  ip ip 
2 i sin pa 2 sin pa
= = , p ≠ 0.
ip √ (2 π) p √ (2 π)
~
For p = 0 , F ( p) = 2 a / √ (2 π).

(a) We know that if

~ 1 ∞
∫ − ∞ F ( x) e
ipx
F ( p) = dx
√ (2π)
1 ∞ ~
then F ( x) = ∫ −∞ F ( p) e − ipx dp .
√ (2π)

1 ∞ 2 sin pa  1, | x | < a
∴ ∫ −∞ ⋅ e − ipx dp = 
√ (2 π) p √ (2 π) 0 , | x | > a.

1 ∞ sin pa cos px i ∞ sin pa sin px

But L.H.S. =
π ∫ −∞ p
dp −
π ∫ −∞ p
dp

1 ∞ sin pa cos px
=
π ∫ −∞ p
dp,

since the integrand in the other integral is an odd function of p.

∞ sin pa cos px  π, | x | < a

∴ ∫ −∞ p
dp = 
0 , | x | > a.
 C-199

(b) If x = 0 and a = 1 in (a), then

∞ sin p
∫ − ∞ p dp = π
∞ sin p ∞ sin p π
or 2 ∫0 p
dp = π or ∫0 p
dp =
2
⋅ …(1)

Note: Putting x = 0 in (a) and on simplification, we get

∞ sin ap π
∫0 p
dp =
2
⋅ …(2)

The results (1) and (2) can be used as standard formulae.

Example 6: Find Fourier sine and cosine transforms of e − x and using the inversion formulae
recover the original functions, in both the cases. (Gorakhpur 2006, 09, 13; Kanpur 09)

Solution: Let f ( x) = e − x .
~  2 ∞  2 ∞ −x
Then f s ( p) =  
 π ∫0 f ( x) sin px dx =  
 π ∫0 e sin px dx

∞
 2  e 
−x
=    (− sin px − p cos px)
 π  1 + p 2
0

p  2
=  
1+ p 2  π
~  2 ∞  2 ∞
and f c ( p) =   ∫0 f ( x) cos px dx =   ∫0 e − x cos px dx
 π  π
∞
 2  e 
−x
=    (− cos px + p sin px)
 π  1 + p 2
0

1  2
=   ⋅
1+ p 2  π

Applying inversion to the sine transform, we have

 2 ∞ ~
f ( x) =  
 π ∫0 f s ( p) ⋅ sin px dp

2 ∞ p sin px
=
π ∫0 1 + p2
dp …(1)

and applying inversion to the cosine transform, we have

 2 ∞ ~
f ( x) =  
 π ∫0 f c ( p) cos px dp

2 ∞ cos px
=
π ∫0 1 + p2
dp. …(2)

Now from Fourier integral theorem, we have

 C-200

1 ∞ ∞
f ( x) =
π ∫0 dp ∫ −∞ f (v) cos p ( x − v) dv

1 ∞ ∞
or f ( x) =
π ∫0 cos px dp ∫ −∞ f (v) cos pv dv

1 ∞ ∞
+
π ∫0 sin px dp ∫ −∞ f (v) sin pv dv. …(3)

Case I: Defining f ( x) in (− ∞, 0 ) such that f ( x) is an even function of x, from (3), we

have
2 ∞ ∞
f ( x) =
π ∫0 cos px dp ∫0 f (v) cos pv dv.

Taking f ( x) = e − x , we have
2 ∞ ∞
e− x = ∫0 cos px dp ∫0 e − v cos pv dv
π
∞
2 ∞  e− v 
= ∫ cos px  (− cos pv + p sin pv) dp
π 0 1 + p
2
0

2 ∞ cos px
π ∫ 0 1 + p2
= dp.

∞ cos px π −x
∴ ∫0 1 + p2
dp =
2
e .

2 π −x
∴ from (2) we have f ( x) = ⋅ e = e− x .
π 2
Case II: Again defining f ( x) in (− ∞, 0 ) such that f ( x) is an odd function of x, from
(2), we have
2 ∞ ∞
f ( x) =
π ∫0 sin px dp ∫0 f (v) sin pv dv.

Taking f ( x) = e − x and simplifying, we have

∞ p sin px π −x
∫0 1 + p2
dp =
2
e .
(Gorakhpur 2013)
2 π −x
∴ from (1), f ( x) = ⋅ e = e− x .
π 2
1
Example 7: Find Fourier cosine transform of f ( x) = and hence find Fourier sine
1 + x2
x
transform of F ( x) = ⋅
1 + x2 (Kanpur 2007, 10; Meerut 13; Rohilkhand 14)

Solution: We have
~  2 ∞
f c ( p) =  
 π ∫0 f ( x) cos px dx
 C-201

 2 ∞ cos px
=  
 π ∫0 2
1+ x
dx .

Differentiating both sides w.r.t. p, we have

d ~  2 ∞ x sin px
dp
f c ( p) = −  
 π ∫0 2
1+ x
dx

 2 ∞ ( x2 + 1 − 1) sin px
=−  
 π ∫0 2
x (1 + x )
dx

 2 ∞ sin px  2 ∞ sin px
=−  
 π ∫0 x
dx +  
 π ∫0 x (1 + x2 )
dx

 2 π  2 ∞ sin px  ∞ sin px π
=−   ⋅ +  
 π 2  π ∫0 2
x (1 + x )
dx . ∵

∫0 x
dx =
2 

Differentiating again w.r.t. p, we have

d2 ~  2 ∞ cos px ~

dp
f ( p) =  
2 c  π ∫ 0 2
1+ x
dx = f c ( p)

~
or ( D2 − 1) f c ( p) = 0
whose general solution is
~
f c ( p) = Ae p + Be − p . …(1)
Now when p = 0 ,
∞
~  2 ∞ dx  2  −1 
f c ( p) =  
 π ∫ 0 1 + x2 =  π   tan x 
0
π  2  π
=   =  
2  π  2

d ~  π
and f c ( p) = −   .
dp  2

 π
∴ from (1), we have   = A + B
 2

 π
and −   = A − B.
 2

 π
Solving, A = 0 , B =   .
 2
~  π
∴ from (1), we have f c ( p) =   e − p.
 2

Second part: We have

~  2 ∞ cos px  π − p
f c ( p) =  
 π ∫0 2
1+ x
dx =   ⋅e .
 2
 C-202

Now differentiating both sides w.r.t. p, we have

∞ x sin px π
− ∫ dx = − ⋅ e − p .
0 2
1+ x 2
~  2 ∞ x  π
∴ Fs ( p) =   ∫0 sin px dx =   e − p .
 π 1+ x 2  2

Example 8: Find the sine and cosine transforms of x ne − ax .

Solution: Let f ( x) = x ne − ax .
~  2 ∞
∴ f s ( p) =  
 π ∫0 f ( x) sin px dx

 2 ∞ n − ax
=  
 π ∫0 x e sin px dx …(1)

∞
∞ − ax
 e − ax 
We have ∫0 e sin px dx = 
2
 a + p
2
(− a sin px − p cos px)
0
p 1 1 1 
= =  − ⋅
2
a +p 2 2 i  a − ip a + ip

Differentiating both sides w.r.t. a, n times, we have

∞
(−1)n ∫0 x n e − ax sin px dx

1  dn −1 dn −1
=  n (a − ip) − n (a + ip) 
2 i  da da 
1
= (− 1)n (n !) [(a − ip)−( n+1) − (a + ip)−( n+1)]
2i
1
= (− 1)n (n !) [2 i r −( n+1) sin (n + 1) θ], putting a = r cos θ, p = r sin θ
2i
= (− 1)n n !(1 / r)n+1 sin (n + 1) θ.
∞
∴ ∫0 x n e − ax sin px dx

= (n !) . [1 / (a2 + p2 )( n+1)/2 ] sin {(n + 1) tan−1 ( p / a)}.

[ ∵ r = (a2 + p2 )1 /2 and θ = tan−1 ( p / a)]
Hence, from (1),
−1
~  2  n !sin {(n + 1) tan ( p / a)}
f s ( p) =   ⋅
 π (a2 + p2 )( n + 1) /2
~  2 ∞
Also f c ( p) =  
 π ∫0 f ( x) cos px dx

 2 ∞ n − ax
=  
 π ∫0 x e cos px dx . …(2)
 C-203

∞
∞ − ax
 e − ax 
We have, ∫0 e cos px dx = 
2
 a + p
2
(− a cos px + p sin px)
0
a
=
a + p2
2

1 1 1 
=  + ⋅
2  a − ip a + ip

Differentiating both sides w.r.t. a, n times, we have

∞
(− 1)n ∫ x n e − ax cos px dx
0

1
= (− 1)n (n !) [(a − ip)−( n+1) + (a + ip)−( n+1)]
2
= (− 1)n (n !) (1 / r)n+1 cos (n + 1) θ,

putting a = r cos θ, p = r sin θ and on simplification.

∞ cos { (n + 1) tan−1 ( p / a)}

∴ ∫0 x n e − ax cos px dx = (n !) . ⋅
(a2 + p2 )( n + 1)/2

Hence from (2), we have

−1
~  2  n !cos {(n + 1) tan ( p / a)}
f c ( p) =   ⋅ .
 π (a2 + p2 )( n + 1) /2

Alternative method to evaluate the integrals

∞ − ax ∞
∫0 e x n sin px dx and ∫0 e − ax x n cos px dx.

∞ ∞
We have ∫0 e − ax x n cos px dx + i ∫0 e − ax x n sin px dx

∞
=∫ e − ax x n (cos px + i sin px) dx
0
∞ ∞
= ∫0 e − ax x n e ipx dx = ∫0 e −( a − ip ) x x ( n + 1) −1 dx

Γ (n + 1)  ∞ Γ (n)
= ∵ ∫0 e − az z n − 1 dz = 
n+1  an 
(a − ip )
Γ (n + 1)
= , putting a = r cos θ and p = r sin θ
n+1
r (cos θ − i sin θ) n + 1

n!  1 
= (cos θ + i sin θ) n + 1 ∵ = cos θ + i sin θ
n +1
r  cos θ − i sin θ 
n!
= [cos (n + 1) θ + i sin (n + 1) θ]. ...(1)
r n +1
 C-204

Equating real and imaginary parts on both sides of (1), we have

∞ n!
∫0 e − ax x n cos px dx = cos (n + 1) θ
r n +1
∞ n!
and ∫0 e − ax x n sin px dx = sin (n + 1) θ ,
r n +1
where r2 = a2 + p2 i. e., r = (a2 + p2 )1 /2 and θ = tan−1 ( p / a).
∞ n!
Hence ∫0 e − ax x n cos px dx = cos { (n + 1) tan−1 ( p / a)}
(a + p2 )( n+1)/2
2

∞ n!
and ∫0 e − ax x n sin px dx = sin {(n + 1) tan−1 ( p / a)}.
2 2 ( n+1)/2
(a + p )

Example 9: Find the sine transform of

e ax + e − ax
⋅
e πx − e − πx (Meerut 2013B; Purvanchal 14; Kanpur 14)

e ax + e − ax
Solution: If f ( x) = πx , then we have
e − e − πx
~  2 ∞
f s ( p) =  
 π ∫0 f ( x) sin px dx

 2 ∞ e ax + e − ax
=  
 π ∫ 0 πx
e − πx
−e
sin px dx

 2 ∞ e ax + e − ax e ipx − e − ipx
=  
 π ∫0 e πx − e − πx
⋅
2i
dx

( a + ip) x
 2  1 ∞ e − e −( a + ip) x 1 ∞ e( a − ip) x − e −( a − ip) x 
=    ∫
2i ∫ 0
dx − dx
 π  2 i 0
 e π x − e − πx e π x − e− π x 

 2  1 1 a + ip 1 1 a − ip
=    ⋅ tan − ⋅ tan
 π  2 i 2 2 2i 2 2 
 ∞ e az − e − az 1 a
∵ From definite integrals, ∫ π z − π z
dz = tan 
 0 e −e 2 2 

 a + ip a − ip 
sin sin
 2  1 2 − 1 2 
=   
 π  4i a + ip 4 i a − ip 
 cos cos 
 2 2 
a + ip a − ip a − ip a + ip
sin cos − sin cos
 2 2 2 2 2
=  
 π a + ip a − ip
4 i cos cos
2 2
 C-205

 2  sin a + sin ip − (sin a − sin ip) sin ip

=   =
 π 2 ⋅ 2 i [cos ip + cos a] √ (2π) i [cos ip + cos a]
sinh p e p − e− p
= = ⋅
√ (2π) (cosh p + cos a) √ (2 π) (e p + e − p + 2 cos a)

Example 10: Find the Fourier sine transform of

1
f ( x) = ⋅
x (a2 + x2 )
Solution: We have
~  2 ∞ 1
f s ( p) =  
 π ∫0 x (a + x2 )
2
sin px dx . …(1)

∞ 1
Let I = ∫0 x (a + x2 )
2
sin px dx. …(2)

dI d ∞ sin px ∞∂ 
 sin px  
Then =
dp dp ∫0 x (a2 + x2 )
dx = ∫    dx
0  ∂p  x (a2 + x2 ) 
   
∞ cos px
= ∫0 a2 + x2
dx. …(3)

d2 I ∞ x sin px
∴
dp2
=− ∫0 a2 + x2
dx

∞ x2 sin px ∞ ( x2 + a2 ) − a2
=− ∫0 x (a2 + x2 )
dx = − ∫0 x (a2 + x2 )
sin px dx

∞ sin px ∞ sin px
=− ∫0 dx + a2 ∫0 dx
x x (a2 + x2 )
π  ∞ sin px π
=− + a2 I. ∵ ∫0 dx =
2  x 2 
2
d I π
∴ − a2 I = −
2 2
dp
π d
or ( D2 − a2 ) I = − , where D ≡ ⋅
2 dp
The solution of the above differential equation is
π
I = Ae − ap + Be ap + ⋅ …(4)
2 a2
dI
∴ = − Aae − ap + Bae ap. …(5)
dp

Now from (2), when p = 0, we have I = 0 and from (3), when p = 0, we have
 C-206

∞
dI ∞ 1 1  −1 x  π
dp
= ∫0 a2 + x2
dx =
a  tan =
a 0 2 a
⋅

So putting p = 0 in (4) and (5), we get

π
A+ B= − …(6)
2 a2
π π
and a (− A + B) = i. e., − A + B = ⋅ …(7)
2a 2 a2
π
Solving (6) and (7), we get B = 0 , A = − ⋅
2 a2
Putting the values of A and B in (4), we get
∞ sin px π π π
I = ∫0 dx = − e − ap + = (1 − e − ap).
2 2 2 2
x (a + x ) 2a 2a 2 a2

Now putting the value of I in (1), we get

~  2 π 1  π
− ap − ap
f s ( p) =   ⋅ 2 (1 − e ) = 2   ⋅ (1 − e ).
 π  2a a  2
2
Example 11: Find the Fourier cosine transform of e − x .

Solution: We have
~ 2  2 ∞ 2
Fc { e − x } =   ∫0 e − x cos px dx = I …(1)
 π

Differentiating w.r.t.‘p’ we have

dI  2 ∞ 2 1  2 ∞ 2
=−   ∫0 xe − x sin px dx =   ∫0 (− 2 xe − x ) ⋅ sin px dx
dp  π 2  π

1  2   − x2 ∞ − x2 
=   (e sin px)0∞ − p ∫ e cos px dx
2  π  0 
(Integrating by parts taking sin px as first function)
p dI p
=− I. ∴ =− dp.
2 I 2
Integrating, we have
p2 2
log I = − + log A or I = Ae − p /4 . …(2)
4
 2 ∞ − x2 1
But when p = 0, from (1) I =  
 π ∫0 e dx =
√2
⋅

∴ from (2), A = 1 / √ 2.
2 2
Hence I = Fc {e − x } = (1 / √ 2) e − p /4 .
 C-207

Comprehensive Exercise 2

1. (i) Find the Fourier transform of f ( x), if

 √ (2 π)
 , | x| ≤ ε
f ( x) =  2 ε
 0 , | x| > ε.

(ii) Find the Fourier transform of
 x, | x| ≤ a
f ( x) =  (Gorakhpur 2004; Kanpur 08)
 0 , | x| > a.
2 2
2. (i) Show that the Fourier transform of f ( x) = e − x /2 is e − p /2 . (Kanpur 2011)

(ii) Find the Fourier transform of the function

 x
1 + a , for − a < x < 0

 x
f ( x) = 1 − , for 0 < x < a
 a
 0 , other wise.

1 − x2 , | x | ≤ 1
3. (i) Find the Fourier transform of F ( x) = 
 0, | x | > 1
(Gorakhpur 2004, 12; Kanpur 10; Purvanchal 14)
∞  x cos x − sin x x
and hence evaluate ∫0 
 3
x
 cos dx.
 2
cos x, 0 < x < a
(ii) Find the cosine transform of the function f ( x) = 
 0, x > a.
sin x, 0 < x < a
(iii) Find the sine transform of the function f ( x) = 
 0, x > a.
(Gorakhpur 2013)
4. (i) Find the Fourier sine and cosine transform of f ( x), if
 x, 0 < x <1

f ( x) = 2 − x, 1 < x < 2 (Kanpur 2007, 11, 14; Gorakhpur 16)
0 , x > 2.

(ii) Find the Fourier sine and cosine transform of the function
f ( x) = x m − 1.
 ∞ ∞
Hint. ∫0 x m − 1 cos px dx − i ∫0 x m − 1 sin px dx

∞ Γ (m) Γ (m)  mπ mπ  
= ∫0 e ipx x m − 1 dx = =  cos − i sin 
(ip) m pm  2 2  
 C-208

5. (i) Find the Fourier sine transform of x /(1 + x2 ).

e ax + e − ax
(ii) Find the cosine transform of ⋅
e πx + e − πx (Purvanchal 2014; Kanpur 14)
1
6. Find the sine transform of and deduce that
e πx − e − πx
1
Fs (cosech πx) = tanh ( p / 2).
√ (2 π)
7. Find the Fourier sine transform of f ( x) , if
0 , 0 < x < a

f ( x) =  x, a ≤ x ≤ b
0 , x > b.

8. Find f ( x) if its sine transform is π / 2.
9. Find f ( x) if (i) its sine transform is e − ap, (Kanpur 2012)
− ap
(ii) its cosine transform is e .
~
10. Find the inverse Fourier transform of f ( p) = e −| p | y .
~ e − ap
11. Find f ( x) if f s ( p) = . Hence deduce Fs −1 {1/ p}.
p (Gorakhpur 2009)

12. Solve the integral equation

∞ (1 − λ ), for 0 ≤ λ ≤ 1
∫0 f ( x) cos λ x dx = 
 0 , for λ > 1.
(Gorakhpur 2007, 09, 11, 13)

A nswers 2
sin pε i  2
1. (i) (ii) − ⋅   (ap cos ap − sin ap)
pε 2
p  π

1 2
2. (ii) ⋅ [1 − cos pa]
2
ap π
 p cos p − sin p
3.
 2
(i) −2   ⋅  ; − 3π
 π  p3  16
 
1 sin (1 + p) a sin (1 − p) a 
(ii)  + 
√ (2 π)  1 + p 1− p 

1 sin (1 − p) a sin (1 + p) a 
(iii)  − 
√ (2 π)  1 − p 1+ p 
 C-209

 2  sin p  2  cos p
4. (i) 2   ⋅ (1 − cos p); 2   ⋅ (1 − cos p).
 π  p2  π p2
Γ(m)  2  mπ Γ(m)  2  mπ
(ii)   sin ; m   cos
pm  π  2 p  π  2
 2  cos (a / 2) ⋅ (e
p /2
+ e − p /2 )
5. (i) √ (π / 2) . e − p (ii)  
 π 2 cos a + e p + e − p
1 ep − 1
6.
2 √ (2 π) ep + 1

 2   − b cos pb + a cos pa sin pb − sin pa 

7.    + 
 π  p p2 

8. √ (π / 2).(1 / x)
 2 x  2 a y √2
9.   ⋅ 2 ;   ⋅ 2 10.
 π  a + x2  π  a + x2 √ π( y2 + x2 )
2 ∞ 1 − cos x
√ (2 / π) tan−1 ( x / a) ; √ (π / 2)
π ∫0
11. 12. cos λ x dx
x2

Objective Type Questions

Multiple Choice Questions

Indicate the correct answer for each question by writing the corresponding letter from (a),
(b), (c) and (d).
~
1. If f ( p) is the complex Fourier transform of f ( x), then the complex Fourier
transform of f (ax) is
1 ~  p ~  p
(a) f   (b) f  
a  a  a
1~
(c) f ( p) (d) None of these.
a
~ ~
2. If f s ( p) and f c ( p) are infinite Fourier sine and cosine transforms of f ( x)
respectively, then Fc { f ( x) sin ax} is
~ ~ 1 ~ ~
(a) [ f s ( p + a) − f s ( p − a)] (b) [ f s ( p + a) − f s ( p − a)]
2
~ ~ 1 ~ ~
(c) 2 [ f s ( p + a) − f s ( p − a)] (d) [ f s ( p − a) − f s ( p + a)]
2
3. The infinite Fourier sine transform of e − x is
1  2 p  2
(a)  
2  π
(b)  
2  π
1+ p 1+ p
1  2 1  2
(c)   (d)  
p  π  p2  π 
 C-210

e − ap
4. What is f ( x) if its infinite Fourier sine transform is ?
p
 2 −1 x  2 −1 a
(a)   tan (b)   tan
 π a  π x
 2 −1 2 x  2 −1 2 a
(c)   tan (d)   tan ⋅
 π a  x x

A nswers
1. (a) 2. (b) 3. (b) 4. (a)

¨
 C-211

1 Finite Fourier Sine Transforms

et f ( x) denote a function that is sectionally continuous over some finite interval
L (0 , l ) of the variable x. The finite Fourier sine transform of f ( x) on this interval is
defined as
~ l pπx
f s ( p) = ∫ f ( x) sin dx ,
0 l
where p is an integer.
By the proper choice of the origin and the unit of length, if the end points of the interval
become x = 0 and x = π, then
~ π
∫
f s ( p) =
0
f ( x) sin px dx.
~
Note that f s ( p) is always zero when p = 0.
~
The function f ( x) is called the inverse finite Fourier sine transform of f s ( p).
~
i. e., f ( x) = Fs −1 { f s( p)}.
 C-212

2 Inversion Formula For Sine Transform

~
If f s ( p) is the finite Fourier sine transform of f ( x) over the interval (0 , l ) then the inversion
formula for sine transform is given by
2 ∞ ~ pπx 2 ∞ ~
f ( x) = ∑ f s ( p)sin l
l p=1
or f ( x) = ∑ f s ( p)sin px
π p=1
~
if (0 , π) is the interval considered for f s ( p).

3 Finite Fourier Cosine Transforms

Let f ( x) denote a function that is sectionally continuous over some finite interval (0 , l)
of the variable x. The finite Fourier cosine transform of f ( x) on this interval is defined
as
~ l pπx
f c ( p) = ∫ f ( x) cos dx
0 l
~
where p is an integer. On the interval (0 , π), f c ( p) is defined as
~ π
f c ( p) = ∫0 f ( x) cos px dx.
~
The function f ( x) is called the inverse finite Fourier cosine transform of f c ( p) i. e.,
~
f ( x) = Fc −1 { f c ( p)}.

4 Inversion Formula for Cosine Transform

~
If f c ( p) is the finite Fourier cosine transform of f ( x) over the interval (0 , l ) then the inversion
formula for cosine transform is given by
1~ 2 ∞ ~ pπx
f ( x) = f c (0 ) +
l
∑
l p=1
f c ( p) cos
l
,

~ l
where f c (0 ) = ∫0 f ( x) dx .

If π is taken as the upper limit for the finite Fourier cosine transform then the inversion formula is
given by
1 ~ 2 ∞ ~
f ( x) =
π
f c (0 ) + ∑ f c ( p) cos px ,
π p=1
~ π
where f c (0 ) = ∫0 f ( x) dx .

Note: The upper limit for Fourier sine or cosine transforms will be taken as x = π, if not
given in the problem.
 C-213

5 Multiple Finite Fourier Transforms

Let f ( x, y) be a function of two variables x and y, defined in the square 0 ≤ x ≤ π and
0 ≤ y ≤ π.
Considering it, temporarily, as a function of x, the finite sine transform is given by
~ π
f s ( p, y) = ∫0 f ( x, y) sin px dx
~
and now the finite sine transform of f s ( p, y) which is a function of y is given by
~ π ~
Fs ( p, q) = ∫ f ( p, y) sin qy dy.
0 s
~ π π
∴ Fs ( p, q) = ∫0 ∫0 f ( x, y) sin px sin qy dx dy,

which is the double finite sine transform of f ( x, y) in the region 0 ≤ x ≤ π, 0 ≤ y ≤ π.

Similarly
~ π π
f c ( p, q) = ∫0 ∫0 f ( x, y) cos px cos qy dx dy,

which is the double finite cosine transform of f ( x, y) in the region 0 ≤ x ≤ π, 0 ≤ y ≤ π.

Using inversion formula, we have
~ ∞ ~
2
f s ( p, y) =
π q =1
∑ Fs ( p, q) sin qy

2 ∞ ~
and f ( x, y) = ∑ f s ( p, y) sin px.
π p=1
∞ ∞ ~
4
Hence f ( x, y) = ∑ ∑
π2 p = 1 q = 1
Fs ( p, q) sin px sin qy.

Similarly we can find the inversion formula for double finite cosine transform of
f ( x, y).

6 Operational Properties of Finite Fourier Sine Transform

Theorem 1: The finite Fourier sine transformation resolves the differential form f ′ ′ ( x) into a
~
linear algebraic form in the transform f s ( p) and the boundary values f (0 ) and f (π) in the
manner
~
Fs { f ′ ′ ( x)} = − p2 f s ( p) + p { f (0 ) − (− 1) p f (π)}
whenever f ( x) and f ′ ( x) are continuous and f ′ ′ ( x) is sectionally continuous, on the interval,
0 ≤ x ≤ π.
 C-214

Note: If f ′ ′ ( x) and f ′ ′ ′ ( x) are continuous and f (4 ) ( x) is sectionally continuous, on

the interval, 0 ≤ x ≤ π then replacing f ( x) by f ′ ′ ( x), in the result of the above theorem
we have
Fs { f 4 ( x)} = − p2 Fs { f ′ ′ ( x)} + p [ f ′ ′ (0 ) − (−1) p f ′ ′ (π)]
~
= − p2 [− p2 f s ( p) + p { f (0 ) − (−1) p f (π)}] + p { f ′ ′ (0 ) − (−1) p f ′ ′ (π)}
~
= p4 f s ( p) − p3 [ f (0 ) − (− 1) p f (π)] + p [ f ′ ′ (0 ) − (−1) p f ′ ′ (0 )].
In the similar way finite Fourier sine transforms of other derivatives of even order can
be found.
~
Theorem 2: If f s ( p) is the sine transform of a sectionally continuous function f ( x), 0 ≤ x ≤ π,
then
~ 
 f ( p) x π t x t
Fs−1  s  = ∫0 ∫0 f (r) dr dt − ∫0 ∫0 f (r) dr dt
 p  π
2
 
x π π
=
π ∫0 (π − r) f (r) dr − ∫0 ( x − r) f (r) dr.

7 Operational Properties of Finite Fourier Cosine

Transforms
Theorem 1: If f ( x) and f ′ ( x) are continuous and if f ′ ′ ( x) is sectionally continuous, the
finite Fourier cosine transformation resolves the differential form f ′ ′ ( x) into an algebraic form in
~
f c ( p) and the boundary values f ′ (0 ) and f ′ (π), in the manner
~
Fc { f ′ ′ ( x)} = − p2 f c ( p) − f ′ (0 ) + (− 1) p f ′ (π).
Note: If f ′ ′ ( x) and f ′ ′ ′ ( x) are continuous and if f (4 ) ( x) is sectionally continuous,
then replacing f ( x) by f ′ ′ ( x) in (1), we have
Fc { f 4 ( x)} = − p2 Fc { f ′ ′ ( x)} − f ′ ′ ′ (0 ) + (−1) p f ′ ′ ′ (π)
~
= − p2 [− p2 f c ( p) − f ′ (0 ) + (− 1) p f ′ (π)] − f ′ ′ ′ (0 ) + (− 1) p f ′ ′ ′ (π)
~
= p 4 f c ( p) + p2 [ f ′ (0 ) − (− 1) p f ′ (π)] − f ′ ′ ′ (0 ) + (−1) p f ′ ′ ′ (π).
Similarly finite Fourier cosine transforms of other derivatives of even order can be
found.
~
Theorem 2: If f c ( p) is the cosine transform of a sectionally continuous function f ( x), 0 ≤ x ≤ π,
then
~  ~
−1  f c ( p) x π f c (0 )
Fc 
 p 
2  ∫0= ∫t f (r) dr dt +
2π
( x − π)2 + A,
 
where A is an arbitrary constant.
 C-215

8 Combined Properties of Finite Fourier Sine and Cosine

Transforms
When f ( x) is continuous and f ′ ( x) is sectionally continuous then
(a) Fs { f ′ ( x)} = − pFc { f ( x)}, ( p = 1, 2, 3,...)
and (b) Fc { f ′ ( x)} = pFs { f ( x)} − f (0 ) + (−1) p f (π), ( p = 0 , 1, 2, ...).
Note: If H ( x) is any sectionally continuous function, the above properties may be
written as
 x 
Fs { H ( x)} = − p Fc  ∫ H (r) dr , ( p = 1, 2, ......)
 0 
 1 ~   π x ~ 
and Fc  H ( x) − Hc (0 ) = p Fs  ∫ H (r) dr − Hc (0 ) ,
 π   0 π 
( p = 0 , 1, 2, ......)

9 Convolution
Let F ( x) and G ( x) be two functions defined on the interval −2 π < x < 2 π, then the function
π
F ( x) * G ( x) = ∫−π F ( x − y) G ( y) dy

is called the convolution of F ( x) and G ( x) on the interval − π < x < π.

Example 1: Find the finite Fourier sine and cosine transforms of the function
f ( x) = 2 x, 0 < x < 4. (Agra 2003)

Solution: We have
~ l
f s ( p) = ∫0 f ( x) sin ( pπx / l ) dx]

4
= ∫ 2 x sin ( pπx / 4) dx, as l = 4
0 (Given)
4
 − 2 x cos ( pπx / 4) 4 cos ( pπx / 4)
=  + 2 ∫0 dx
 pπ / 4 0 pπ / 4
4
32 8 sin ( pπx / 4) 32
=− cos pπ +   =− cos pπ.
pπ pπ  pπ / 4 0 pπ
~ l
Also f c ( p) = ∫0 f ( x) cos ( pπx / l) dx

4
= ∫ 2 x cos ( pπx / 4) dx, as l = 4
0
 C-216

4
2 x sin ( pπx / 4) 4 sin ( pπx / 4)
=  − 2 ∫0 dx
 p π / 4 0 pπ / 4
4
8 cos ( pπx / 4) 32
=   = (cos pπ − 1) , if p > 0
pπ  pπ / 4 0 p2 π2
~ 4
and if p = 0, then f c ( p) = ∫0 2 x ⋅ 1 dx = 16.

Example 2: Find the finite Fourier cosine transform of f ( x) if

π x2
(i ) f ( x) = −x+ ⋅ (ii) f ( x) = sin nx.
3 2π
~ π
Solution: (i) We have f c ( p) = ∫0 f ( x) cos px dx .

~ π π x2 
∴ f c ( p) = ∫0  −x+
3

 cos px dx
2 π 
π
 π x2  1  1 π  x
p ∫0
=  − x +  sin px −
 p  −1 +  sin px dx
 3 2 π  0  π
π
1  x 1  1 π1
= − −  −1 +  cos px −
p  
π p 0
∫ cos px dx
p2 0 π
1 1 π 1
3 [
= − sin px]0 = , if p > 0
2
p p π p2
and when p = 0,
~ π π x2 
f c ( p) = ∫ 0  3
 −x+  dx = 0
2 π 
~ π
(ii) f c ( p) = ∫0 sin nx cos px dx

1 π
=
2 ∫0 [sin (n + p) x + sin (n − p) x] dx
π
1  cos (n + p) x cos (n − p) x 
= − −  , if p ≠ n
2 n+ p n− p 0
~ 1  cos (n + p) π cos (n − p) π 1 1 
∴ if p ≠ n, f c ( p) = − − + + ⋅
2 n+ p n− p n+ p n− p
If n − p is even, then n + p is also even and so
~ 1 1 1 1 1 
f c ( p) = − − + +  =0.
2  n+ p n− p n+ p n− p
If n − p is odd, then n + p is also odd and so
~ 1 2 2  2n
f c ( p) =  + = ⋅
2 n + p n− p n2 − p2
 C-217

If p = n, then
~ π 1 π
f c ( p) = ∫
0
sin nx cos nx dx =
2 0 ∫
sin 2 nx dx

π
1  cos 2 nx 
=  − =0.
2 2 n 0
~ 2n
∴ f c ( p) = 0 or according as n − p is even or odd.
n − p2
2

cos k (π − x)
Example 3: Find the finite cosine transform of f ( x) if f ( x) = − ⋅
k sin kπ

Solution: We have
~ π cos { k (π − x)}
f c ( p) = −
0 ∫ k sin k π
cos px dx

1 π
=−
2 k sin kπ ∫0 [cos {k (π − x) + px} + cos {k (π − x) − px}] dx

π
1 sin (kπ − kx + px) sin (kπ − kx − px)
=−  − 
2 k sin kπ  p− k p+ k 0
1 sin pπ sin (− pπ) sin kπ sin kπ 
=−  − − + 
2 k sin kπ  p − k p+ k p− k p+ k 

1  1 1  1
=  −  = , k ≠ 0 , 1, 2, 3, ......
2 k  p − k p + k  p2 − k 2

Example 4: Find f ( x) if
 pπ 
6 sin − cos pπ
~  2  2
f c ( p) = for p = 1, 2, 3,...... and for p = 0,
(2 p + 1) π π
where 0 < x < 4.
Solution: We have
1~ 2 ∞ ~ pπx
f ( x) =
l
f c (0 ) + ∑ f c ( p) cos l
l p=1

 pπ 
6 sin − cos pπ
1 2 2 ∞  2  pπx
= ⋅ + ⋅ ∑ cos
4 π 4 p=1 (2 p + 1) π 4

 pπ 
sin − cos pπ
1 3 ∞  2   pπx 
= + ∑
2π π p = 1 2p + 1
cos 
 4 
⋅

Example 5: Find f ( x) if its finite sine transform is given by

~ 1 − cos pπ
f s ( p) = , where 0 < x < π.
2 2
p π (Kanpur 2008)
 C-218

Solution: We have
2 ∞ ~
f ( x) = ∑ f s ( p) sin px
π p=1

2 ∞ 1 − cos pπ 
= ∑
π p = 1  p2 π2 
sin px

∞ 
2 1 − cos pπ 
=
π 3 ∑ 
 p2
 sin px.

p =1 

Comprehensive Exercise 1

1. (i) Find the finite Fourier sine and cosine transforms of f ( x) = 1.

(Gorakhpur 2014)
(ii) Find the finite Fourier sine and cosine transforms of f ( x) = x.
2. (i) Find the finite Fourier sine transforms of
 x x
1 −  and ⋅
 π 4π
(ii) Find the finite Fourier cosine transform of
 x x
1 −  and ⋅
 π 4π
3. Find the finite Fourier sine transform of f ( x) if
 x, 0 ≤ x≤ π/2
(i) f ( x) = 
 π − x, π / 2 ≤ x ≤ π .
 − x, x< c
(ii) f ( x) = 
 π − x, x > c , where 0 ≤ c ≤ π .
4. (i) Find the finite Fourier cosine transform of f ( x) if
 1, 0 < x < π / 2
f ( x) = 
 −1, π / 2 < x < π. (Kanpur 2012)
2
(ii) Find the finite cosine transform of (1 − x / π) . (Avadh 2013)
5. Find the finite Fourier sine transforms of
(i) x (π − x) (ii) x (π2 − x2 ).
6. Find the finite Fourier sine transform of f ( x) if
π x2
(i) f ( x) = sin nx (ii) f ( x) =
−x+ ⋅
3 2π
7. Find the finite sine transform of f ( x), if
(i) f ( x) = cos kx (ii) f ( x) = x3 (iii) f ( x) = e cx .

cosh { c (π − x)}
8. Find finite Fourier cosine transform of f ( x) if f ( x) = ⋅
sinh (πc )
 C-219

sin k (π − x)
9. Find finite Fourier sine transform of f ( x) , if f ( x) = ⋅
sin (kπ)
10. Find the finite Fourier sine and cosine transforms of
f ( x) = x2 , 0 < x < 4. (Gorakhpur 2008, 14, 15)
~ cos(2 pπ / 3)
11. Find f ( x) if f c ( p) = , if 0 < x < 1.
2
(2 p + 1)
12. When f ( x) = sin mx, where m is a positive integer, show that
~ ~
f s ( p) = 0 if p ≠ m and f s ( p) = π / 2 if p = m.

A nswers 1
1
1. (i) [1 − (− 1) p]; 0
p
π (−1) p + 1 (−1) p − 1 π2
(ii) ; , if p = 1, 2, 3,... and , if p = 0 ⋅
p p2 2
1 (−1) p + 1
2. (i) ;
p 4p
1 1
(ii) [1 − (−1) p] ; [(−1) p − 1]
2
πp 4 πp2
3. (i) (2 / p2 ) sin ( pπ / 2) , (ii) (π / p) cos pc .
4. (i) (2 / p) sin ( pπ / 2), p > 0 and 0, if p = 0.
2 π
(ii) , if p > 0 and , if p = 0
2
πp 3
2 6π
5. (i) p
[1 − (− 1) ], (ii) (−1) p + 1
3 3
p p
6. (i) 0, if p ≠ n , and π / 2, if p = n
π 1
(ii) {(− 1) p + 2} + {(− 1) p − 1}
3
6p πp
p 6 π2 
7. (i) [1 − (−1) p cos kπ], (ii) π (− 1) p  − ,
p2 − k 2  p3 p 

p
(iii) [1 − (−1) p e cπ ]
c + p2
2

c p
8. 9., k ≠ 0 , 1, 2,... .
2 2
c +p p − k2 2

64 128 128 64
10. − cos pπ + (cos pπ − 1) ; cos p π, if p > 0; , if p = 0
pπ 3
p π 3 2
p π 2 3
∞
cos (2 pπ / 3)
11. 1 + 2 ∑ (2 p + 1)2
cos pπx
p=1
 C-220

Objective Type Questions

Multiple Choice Questions

Indicate the correct answer for each question by writing the corresponding letter from (a),
(b), (c) and (d).
1. If f ( x) is continuous and f ′ ( x) is sectionally continuous then Fs { f ′ ( x)} is
(a) − Fc { f ( x)} (b) Fc { f ( x)}
(c) − p Fc { f ( x) } (d) − p2 Fc { f ( x) }
2. The finite Fourier sine transform of f ( x) = x is
π π
(a) (− 1) p + 1 (b) (− 1) p
p p
π π
(c) (− 1) p + 1 (d) (− 1) p
2
p p2
~ 2 π (− 1) p − 1
3. What is f ( x) if its finite sine transform f s ( p) is , where 0 < x < π ?
p3
2 ∞ (− 1) p − 1 ∞
(− 1) p
(a) ∑
π p=1 p3
sin px (b) 4 ∑ p3
sin px
p =1
∞
(− 1) p − 1 4 ∞ (− 1) p − 1
(c) 4 ∑ p3
sin px (d) ∑
π p=1 p3
sin px
p=1
x
4. The finite Fourier sine transform of is
π
(− 1) p (− 1) p + 1
(a) (b)
p2 p2
(− 1) p − 1 (− 1) p + 1
(c) (d)
p p

A nswers
1. (c) 2. (a) 3. (c) 4. (d)

¨
 C-221

1 Some Important Results

he following results are useful in Fourier Series :
T
(i) sin nπ = 0 , cos nπ = (−1)n,
 1  1
sin  n +  π = (− 1)n, cos  n +  π = 0 , where n ∈ I .
 2  2
(ii) ∫ uv = uv1 − u′ v3 + u′ ′ v2 − u′ ′ ′ v4 + … ,
du d2 u
where u′ =
dx
, u′ ′ = 2 , … and v1 =
dx ∫ v dx , v2 = ∫ v1 dx, … .

2π 2π
(iii) ∫0 sin nx dx = 0 . (iv) ∫0 cos nx dx = 0 .

2π 2π
(v) ∫0 sin2 nx dx = π. (vi) ∫0 cos2 nx dx = π.

2π 2π
(vii) ∫0 sin nx . sin mx dx = 0 . (viii) ∫
0
cos nx . cos mx dx = 0 .

2π 2π
(ix) ∫0 sin nx . cos mx dx = 0 . (x) ∫
0
sin nx . cos nx dx = 0 .
 C-222

2 Determination of Fourier Coefficients

(Euler’s Formulae)
The Fourier series is
a0
f ( x) = + a1 cos x + a2 cos 2 x + … + an cos nx + …
2
+ b1 sin x + b2 sin 2 x + … + bn sin nx + … …(1)
To find a0 . Integrating both sides of (1) from x = 0 to x = 2π, we have
2π a 2π 2π 2π
∫ 0 f ( x) dx = 20 ∫ 0 dx + a1 ∫ 0 cos x dx + a2 ∫ 0 cos 2 x dx + …
2π 2π
+ an ∫ cos nx dx + … + b1 ∫0 sin x dx
0
2π 2π
+ b2 ∫0 sin 2 x dx + … + bn ∫
0
sin nx dx + …

a0 2π
=
2 ∫0 dx, [Other integrals vanish]

2π a0
or ∫0 f ( x) dx =
2
2 π.

1 2π
∴ a0 =
π ∫0 f ( x) dx. .…(2)

To find an. Multiplying each side of (1) by cos nx and integrating from x = 0 to x = 2π,
we have
2π a 2π 2π
∫ 0 f ( x) cos nx dx = 20 ∫ 0 cos nx dx + a1 ∫ 0 cos x cos nx dx + …
2π 2π
+ an ∫ cos2 nx dx … + b1 ∫0 sin xnx dx
0
2π
+ b2 ∫0 sin 2 x cos nx dx + …

2π
= an ∫ cos2 nx dx [Other integrals vanish]
0

= an π.
1 2π
∴ an = ∫ f ( x) cos nx dx. …(3)
π 0
By taking n = 1, 2, … we get the values of a1, a2 , … .
To find bn. Multiplying each side of (1) by sin nx and integrating from x = 0 to x = 2π,
we have
2π a 2π 2π
∫ 0 f ( x) sin nx dx = 20 ∫ 0 sin nx dx + a1 ∫ 0 cos x sin nx dx + …
2π
+ an ∫ cos nx sin nx dx + …
0
2π 2π
+ b1 ∫0 sin x sin nx dx + … + bn ∫
0
sin2 nx dx + ...
 C-223
2π
= bn ∫ sin2 nx dx [Other integrals vanish]
0

= bn π.
1 2π
∴ bn =
π ∫0 f ( x) sin nx dx. …(4)

By taking n = 1, 2 , … we get the values of b1, b2 , …… .

1
Note: To get similar formula of a0 , has been written with a0 in Fourier series.
2
∞
If we write the Fourier series as f ( x) = a0 + Σ [an cos nx + bn sin nx],
n =1
then the values of constants is given as :
1 2π 1 2π
a0 =
2π ∫0 f ( x) dx ; an =
π ∫0 f ( x) cos nx dx

1 2π
and bn =
π ∫0 f ( x) sin nx dx .

3 Fourier Series Expansion in the Interval α < x < α + 2 π

The Fourier series is
a0 ∞ ∞
f ( x) = + Σ an cos nx + Σ bn sin nx. …(1)
2 n =1 n =1
To find a0 . Integrating both sides of (1) from x = α to x = α + 2 π, we have
2π + α a0 2π + α 2π + α  ∞ 
∫α f ( x) dx =
2 ∫α dx + ∫
α
 Σ an cos nx dx
 n = 1 
2π + α  ∞ 
+ ∫α  Σ bn sin nx dx
 n =1 
a0
= ⋅ 2π + 0 + 0.
2
1 2π + α
∴ a0 =
π ∫α f ( x) dx .

To find an. Multiplying both sides of (1) by cos nx and then integrating from x = α to
x = α + 2 π , we have
2π + α α 2π + α
∫α f ( x) cos nx dx = 0 ∫
2 α
cos nx dx

2π + α  ∞ 
+ ∫α  Σ an cos nx cos nx dx
 n =1 
2π + α  ∞ 
+ ∫α  Σ bn sin nx cos nx dx
 n =1 
= 0 + π an + 0 .
1 2π + α
∴ an =
π ∫α f ( x) cos nx dx .
 C-224

1 2π + α
Similarly bn =
π ∫α f ( x) sin nx dx.

Note 1. Putting α = 0, the interval becomes 0 < x < 2 π and

1 2π 1 2π 1 2π
a0 = ∫ f ( x) dx, an = ∫ f ( x) cos nx dx and bn = ∫0 f ( x) sin nx dx.
π 0 π 0 π
Note 2. Putting α = − π, the interval becomes − π < x < π and
1 π 1 π 1 π
a0 = ∫ f ( x) dx, an = ∫ f ( x) cos nx dx and bn = ∫−π f ( x) sin nx dx.
π −π π −π π

4 Fourier Series for Discontinuous Functions

 f1 ( x), α < x < x0
Let the function f ( x) be defined by f ( x) = 
 f2 ( x), x0 < x < α + 2 π,
in the interval (α, α + 2 π). Here x0 is the point of discontinuity.
The Fourier series for f ( x) in such cases is obtained in the usual way. The values of
a0 , an , bn are given by
α + 2π
a0 =  ∫ 0 f1 ( x) dx + ∫ f2 ( x) dx  ;
1 x
π α x0 
1  x0 α + 2π
f2 ( x) cos nx dx  ;
π ∫ α ∫ x0
an = f1 ( x) cos nx dx +

1  x0 α + 2π
f2 ( x) sin n x dx .
π ∫ α ∫ x0
bn = f1 ( x) sin nx dx +

At the point of discontinuity, x = α, the Fourier series gives the value of f ( x) as the
arithmetic mean of left and right limits.
1
∴ At x = α, f ( x) = [ f (α − 0 ) + f (α + 0 ) ].
2

Example 1: Find a Fourier series to represent x − x2 from x = − π to x = π .

1 1 1 1 π2
Deduce that − + − +…= ⋅
12 22 32 42 12
Solution: The Fourier series for f ( x) in (− π, π) is
∞ ∞
f ( x) = a0 + Σ an cos nx + Σ bn sin nx.
n =1 n =1
π
1 π 1  x2 x3  π2
Here a0 =
2π ∫− π ( x − x2 ) dx = 
2π 2
−
3
 =−
3
;
−π
1 π
an =
π ∫− π ( x − x2 ) cos nx dx
 C-225

π
1 sin nx  cos nx   sin nx  
= ( x − x2 ) − (1 − 2 x)  − 2 
+ (−2)  − 
π n  n  n3   −π

n
− 4 (−1)
= ; [ ∵ cos nπ = (−1)n ]
n2
1 π
and bn =
π ∫− π ( x − x2 ) sin nx dx

π
1  cos n x  sin nx   cos nx  
= ( x − x2 )  −  − (1 − 2 x)  −  + (−2)  3  
π  n   n 
2  n  −π

n
= − 2 (−1) / n.
∴ The required Fourier series is
π2 cos x cos 2 x cos 3 x cos 4 x 
x − x2 = − +4 2 − + − + …
3  1 22 32 42 
sin x sin 2 x sin 3 x sin 4 x 
+2 − + − + … ⋅ …(1)
 1 2 3 4 
Deduction: Putting x = 0 in (1), we get
π2 1 1 1 1 
0=− + 4  2 − 2 + 2 − 2 + …
3  1 2 3 4 
1 1 1 1 π2
or 2
− 2
+ 2
− 2
+…= ⋅
1 2 3 4 12

Example 2: Find the Fourier series expansion for f ( x), if

− π , − π < x < 0
f ( x) = 
 x, 0< x< π. (Gorakhpur 2006)

1 1 1 π2
Deduce that 2
+ 2
+ 2
+…= ⋅
1 3 5 8
Solution: The Fourier series of f ( x) in (− π, π) is
∞ ∞
f ( x) = a0 + Σ an cos nx + Σ bn sin nx.
n =1 n =1
π 1  0 π
x dx 
1
Then a0 =
2π ∫−π f ( x) dx =
2 π ∫ − π
(− π) dx + ∫0 
 π  1  2 
 x2 
=
1 − π ( x ) 0
+   =  −  π2 + π   = − π ;
π −π
 2  π   2   2
 0 
π 1 0 π
x cos nx dx
1
an =
π ∫−π f ( x) cos nx dx =
π ∫ − π
(− π) cos nx dx + ∫0 
 0 π 
1  sin nx   x sin nx cos nx 
=  −π  + +  
π   n −π  n n 0
2

 C-226

=
1 0 + 1 cos nπ − 1  = 1 (cos nπ − 1) ;
π  n2 n2  π n2
π 1 0 π
x sin nx dx
1
and bn =
π ∫−π f ( x) sin nx dx =
π ∫ − π
(− π) sin nx dx + ∫0 
  π cos nx  0 π
1  cos nx sin nx  
=    + − +  
 
π n −π  n n2  0 
1 π π
=  (0 − cos nπ) − cos nπ = (1 − 2 cos nπ).
1

π n n 
 n
∴ The required Fourier series is
π 2 cos 3 x cos 5 x 
f ( x) = − − cos x + + + …
4 π  32
52 
 2 sin 2 x 3 sin 3 x sin 4 x 
+ 3 sin x − + − + ... …(1)
 2 3 4 
Deduction: Putting x = 0 in (1), we get
π 2 1 1 
f (0 ) = − − 1 + 2 + 2 + … ∞ ⋅ …(2)
4 π  3 5 
But f ( x) is discontinuous at x = 0, and we have f (0 − 0 ) = − π and f (0 + 0 ) = 0 .
1
∴ f (0 ) = [ f (0 − 0 ) + f (0 + 0 )] = − (π / 2)
2 …(3)
Hence from (2) and (3), we have
π π 2 1 π2
= − −  2 + 2 + 2 + …
1 1 1 1 1
− or + + +…= ⋅
2 4 π 1 3 5  12 32 52 8

 π − x
Example 3: Obtain the Fourier series of f ( x) =   in the interval (0 , 2π) and hence deduce
 2 
π 1 1 1
= 1− + − + …
4 3 5 7
Solution: The Fourier series for f ( x) in (0 , 2π) is
∞ ∞
f ( x) = a0 + Σ an cos nx + Σ bn sin nx
n =1 n =1
1 2π 1 2π  π − x
Here a0 =
π ∫0 f ( x) dx =
π ∫0 
 2 
 dx
2π
1  x2  1
= πx −  = [2 π2 − 2 π2 ] = 0 ;
2π  2 0 2 π
1 2π 1 2π
an =
π ∫0 f ( x) cos nx dx =
2π ∫0 (π − x) cos nx dx

2π
1  sin nx  − cos nx 
= (π − x) n − (− 1)   = 0;
2π   n2  0
1 2π 1 2π
bn =
π ∫0 f ( x) sin nx dx =
2π ∫0 (π − x) sin nx dx
 C-227
2π
=
1 − (π − x) cos nx − sin nx  1
= ⋅
2π  n 2
n 0  n
π−x ∞ 1 1 1
∴ f ( x) =
=0 +0 + Σ sin nx = sin x + sin 2 x + sin 3 x + …
2 n =1 n 2 3
...(1)
π π 1 1 1
Deduction: Putting x = in (1), we get = 1 − + − + …
2 4 3 5 7

Comprehensive Exercise 1

1. Find a series of sines and cosines of multiples of x which will represent

π
e x in the interval − π < x < π.
2 sinh π
2. Find the Fourier series of the function defined as
 x+π ; 0≤ x≤ π
f ( x) =  and f ( x + 2π) = f ( x).
 − x − π ; − π≤ x<0

3. Obtain the Fourier series for f ( x) = e − x in the interval 0 < x < 2 π.

4. Find the Fourier series of the function
 − 1 for − π < x < − π / 2

f ( x) =  0 for − π / 2 < x < π / 2
 + 1 for π / 2 < x < π.

5. Find the Fourier series representing f ( x) = x, 0 < x < 2π. (Gorakhpur 2012)
2
6. Find the Fourier series expansion of function f ( x) = x + x , − π < x < π ;
π2 1 1 1
then prove that = 1+ 2 + 2 + 2 + ….
6 2 3 4 (Gorakhpur 2010, 13, 14)

A nswers 1
π 1 1 1 1 1 
1. e x = −  cos x − cos 2 x + cos 3 x − cos 4 x + ...
2 sinh π 2 2 2 10 17 
1 2 3 4 
+  sin x − sin 2 x + sin 3 x − sin 4 x +...
2 5 10 17 
π 4  cos x cos 3 x   sin x sin 3 x 
2. f ( x) = −  + +... + 4  + +. ..
2 π  12 32   1 3 
1 − e− 2 π  1  1 1 1 
3. e− x =  +  cos x + cos 2 x + cos 3 x +...
π 2 2 5 10 
1 2 3 
+  sin x + sin 2 x + sin 3 x + ... 
2 5 10 
 C-228

1
2 sin x − 2 sin 2 x + 2 sin 3 x + ….
4. f ( x) =

π 3 

x = 2 π − 2 sin x + sin 2 x + sin 3 x + ….

1 1
5.
 2 3 

π2
+ 4 − cos x + 2 cos 2 x − 2 cos 3 x + ...
1 1
6. x + x2 =
3  2 3 

−2 − sin x + sin 2 x − sin 3 x + ….

1 1
 2 3 

5 Change of Period
Generally in engineering problems, the period of the given function is not always 2π but is
T or 2c.To use the formula of an and bn this period must be converted to the length 2π.
The independent variable x is also to be changed proportionally.
Let the function f ( x) be defined in the interval (0 , 2c ).
∵ 2c is the interval for the variable x.
2 πx πx
∴ 2π is the interval for the variable = = ⋅
2c c
πx zc
Putting z = or x = , the function f ( x) of period 2c is transformed to the function
c π
 cz 
f   or F (z ) of period 2π.
 π
Now F (z ) can be expanded in the Fourier series as
 cz  a
F (z ) = f   = 0 + a1 cos z + a2 cos 2 z + … + b1 sin z + b2 sin 2 z + …
 π 2
1 2π 1 2π cz
where a0 =
π ∫0 F (z ) dz =
π ∫0 f   dz
π
1 2c π 1 2c Putting z = π x 
= ∫
π 0
f ( x) dx = ∫
c c 0
f ( x) dx ;  c 
1 2π
an =
π ∫0 F (z ) cos nz dz

1 2π cz
=
π ∫0 f   cos nx dz
π
1 2c nπx π
=
π ∫0 f ( x) cos
c c
dx

1 2π nπx Putting z = πx 
c ∫0
= f ( x) cos dx.
c  c 
1 2c nπx
Similarly, bn = ∫ f ( x) sin dx.
c 0 c
 C-229

 π x, 0 ≤ x ≤1
Example 4: Obtain Fourier Series for the function f ( x) = 
 π (2 − x), 1 ≤ x ≤ 2.
2
π 1 1 1
Deduce that = 2 + 2 + 2 + ….
8 1 3 5
Solution: Here the function is defined in the interval (0 , 2). Let
a ∞
f ( x) = 0 + Σ (an cos n π x + bn sin nπx) . …(1)
2 n =1
1 2c
c ∫0
Then, we have a0 = f ( x) dx , where c = 1

1 2 1 2
= ∫ f ( x) dx = ∫ πx dx + ∫ π (2 − x) dx
1 0 0 1
1 2
 x2   x2  π π
= π   + π 2 x −  = + = π;
 2 0  2
1
2 2
1 2c nπ x 1 2
an =
c ∫ 0
f ( x) cos
c
dx = ∫ f ( x) cos nπ x dx
1 0
1 2
= ∫0 π x cos nπ x dx + ∫1 π (2 − x) cos nπ x dx

1 2
 sin nπx cos nπx   sin nπx cos nπx 
= π x + 2  + π − (2 − x) − 
 nπ (nπ) 0  nπ (nπ)2 1

cos nπ 1   cos nπ  2 cos nπ

=  2 − 2 
1 1
=π 2
− 2
+ π − 2
+ 2 
 (nπ) (nπ)   (nπ) (nπ)  π  n n 

 − 4 , if n is odd

=  π n2
 0 , if n is even ;

1 2c nπ x 1 2
and bn = ∫
c 0
f ( x) sin
c
dx = ∫ f ( x) sin nπ x dx
1 0
1 2
=π ∫0 x sin nπ x dx + π ∫
0
(2 − x) sin nπ x dx
1 2
  − cos nπ x  sin nπx    cos nπx  sin nπx 
= π x   +  + π (2 − x)  −  − 
  nπ  2
(nπ) 0   nπ  (nπ)2 1
cos nπ cos nπ
=− + = 0.
n n
Putting these values of a0 , an , bn in (1), we get the required Fourier series as
π 4 cos π x cos 3 πx cos (5 πx) 
f ( x) = −  2 + + + ... …(2)
2 π 1 32 52 
 C-230

Deduction: Putting x = 0 in (2), we get

π 4 1
f (0 ) = −  2 + 2 + 2 + …
1 1
2 π 1 3 5 
π 4 1
+ ...
1 1
or 0= − + +
2 π 12 32 52 
π 4  1 + 1 + 1 + … π2 1 1 1
or = 12 32 52  or = 2 + 2 + 2 +…
2 π 8 1 3 5
Example 5: Find the Fourier series corresponding to the function f ( x) defined in (−2 , 2) as
follows
 2 , if − 2 ≤ x ≤ 0
f ( x) = 
 x, if 0 < x < 2 .
Solution: Here the interval is (−2, 2) and l = 2.
∞ nπ x nπ
a0  x
Let f ( x) = + Σ an cos l + bn sin l  ⋅
2 n =1

f ( x) dx = ∫ x dx 
1 l 1 0 2
Then a0 = ∫
l −l 2  −2
2 dx + ∫0 

1   x2   1
2
= (2 x)−2 +    = [4 + 2] = 3 ;
0
2  2 0 2
 
1 l  nπx 
an = ∫ ƒ( x) cos   dx
l −l  l 
1 0 nπx nπx
dx 
2
2 ∫ −2 l ∫0
= 2 cos dx + x cos
2 2 
 0  2
1 4  nπ x  2 nπx 4 nπx  
=  sin  +  x sin + 2 2 cos 

2 nπ  2   nπ 2 n π 2  
−2 0

1 4 4  2 n
=  2 2 cos nπ − 2 2  = 2 2 [(−1) − 1]
2 n π n π  n π
 − 4 , when n is odd

=  n2 π2
 0 , when n is even;

1 l nπx 1 0 nπx 1 2 nπx
l ∫− l ∫ −2 ∫0
bn = f ( x) sin dx = 2 sin dx + x sin dx
l 2 2 2 2
1   2 nπx  
0
1   2 nπx  4 nπx 
2
= 2  − nπ cos 2   + x − cos  + (1) 2 2 sin
2 −2 2   nπ 2  n π 2  0

= − 4 + 4 cos nπ + 1 − 4 cos nπ + 4 sin nπ

1
 nπ nπ
2  2  nπ n2 π2

 

= −
1 4 2
=− .
 
2  nπ  nπ
 C-231

∴ The required Fourier series is

a πx 2 πx 3 πx
f ( x) = 0 + a1 cos + a2 cos + a3 cos +…
2 l l l
πx 2 πx 3 πx
+ b1 sin + b2 sin + b3 sin +…
l l l
πx 3 πx
= − 2  2 cos + …
3 4 1 1
+ 2 cos
2 π 1 2 3 2 
2 1 πx 1 2 πx 1 3 πx
−  sin + sin + sin +... .
π 1 2 2 2 3 2 

Comprehensive Exercise 2

l l ∞1 2 nπx
1. Prove that − x = Σ sin , 0 < x < l.
2 π 1 n l
1 + 2 x , − 1 < x < 0
 l
2. Obtain Fourier series of the function F ( x) = 
2x
 1− , 0 < x < l.
 l
 x + 1 for − 1 < x < 0 ,
3. Given f ( x) = 
 x − 1 for 0 < x < 1.
Expand f into a Fourier series on (− 1, 1).
4. Obtain the Fourier series expansion of the periodic function of period 1
 1 + x, − 1 < x ≤ 0

f ( x) =  2 2
1 1
 − x, 0 < x < .
2 2
0 , 0 < x < c
5. Given f ( x) =  ; expand f ( x) in a Fourier series of period 2c.
1, c < x < 2 c

A nswers 2
8 1 πx 3 πx 5 πx
+ ...
1 1
2. F ( x) = 2  2
cos + 2 cos + 2 cos
π l l 3 l 5 l 

f ( x) = −  sin πx + sin 2 πx + sin 3 πx + …

2 1 1 1
3.
π 1 2 3 
1 2 cos 2 πx cos 6 πx cos 10 πx 
4. f ( x) = + + + + …
4 π2  12 32 52 
1 2 1 πx 1 3 πx 
5. f ( x) = −  sin + sin + …
2 π 1 c 3 c 
 C-232

6 Fourier Series Expansion of An Even or an Odd

Function in ( − π, π)
A function f ( x) is called an even (symmetric) function if f (− x) = f ( x).
The area under such a curve from − π to π is double the area from 0 to π.
π π
∴ ∫ −π f ( x) dx = 2 ∫
0
f ( x) dx .

A function f ( x) is called an odd (or skew

symmetric) function if f (− x) = − f ( x).
Here the area under the curve from − π to π is
zero.
π
∴ ∫ −π f ( x) dx = 0 .

Expansion Of An Even Function

In ( − π, π ) :
Here
1 π 2 π
a0 =
π ∫ −π f ( x) dx =
π ∫0 f ( x) dx .

As f ( x) and cos nx are both even functions,

therefore the product of f ( x) . cos nx is also
an even function.
1 π
∴ an = ∫ f ( x) cos nx dx
π −π
2 π
= ∫ f ( x) cos nx dx.
π 0
As sin nx is an odd function so f ( x) . sin nx is also an odd function. We need not to
calculate bn. It saves our labour a lot.
1 π
∴ bn = ∫ f ( x) sin nx dx = 0 .
π −π
Thus the series of the even function will contain only cos terms.
Expansion Of An Odd Function In ( − π, π ) :
1 π
Here a0 = ∫ f ( x) dx = 0 ;
π −π
1 π
an = ∫ f ( x) cos nx dx = 0 [ ∵ f ( x) . cos nx is an odd function]
π −π
1 π
and bn = ∫ f ( x) sin nx dx
π −π
2 π
= ∫ f ( x) sin nx dx [ ∵ f ( x) . sin nx is an even function]
π 0
Thus the series of the odd function will contain only sin terms.
 C-233

Example 6: Obtain Fourier’s series for the expansion of f ( x) = x sin x in the interval
− π < x < π. Hence deduce that
π −2 1 1 1
= − + − …⋅
4 1. 3 3 . 5 5 . 7 (Gorakhpur 2005, 07, 09, 11)

Solution: Here f ( x) = x sin x being an even function of x. The Fourier series is

∞
ƒ( x) = a0 + Σ an cos nx
n =1

1 π 2 ∞ π
or ƒ( x) =
π ∫0 ƒ(u) du +
π n =1
Σ cos n x ∫0 ƒ(u) cos nu du,

where ƒ(u) = u sin u.

π π
Now ∫0 u sin u du = [− u cos u + sin u]0 = π

π 1 π
and ∫0 u sin u cos nu du =
2 ∫0 u [ sin (n + 1)u − sin (n − 1)u] du

π
1  − u cos (n + 1)u sin (n + 1)u
=  + 
2  n +1 (n + 1)2  0
π
1  u cos (n − 1)u sin (n − 1)u
− − + 
2  n −1 (n − 1)2  0

π cos (n − 1)π cos (n + 1)π  π  cos nπ cos nπ 

=  −  = − + 
2 n −1 n +1  2  n −1 n +1 
π cos nπ
= , when n > 1.
1 − n2
When n = 1, we have
π 1 π
∫0 u sin u cos u du =
2 ∫0 u sin 2 u du

π
1  u cos 2 u sin 2 u π
= − + =− ⋅
2 2 4  0 4

2  π ∞ π cos nπ 
∴ x sin x = 1 +  − cos x + Σ cos nx
π 2
 4 n=2 1− n 
 1 1 1 1 
= 1 + 2 − cos x − cos 2 x + cos 3 x − cos 4 x + …
 4 1 . 3 2. 4 3 . 5 

1
Deduction: Putting x = π, we get
2
 C-234

π  1 1 1 
= 1+ 2  − + − …
2 1 . 3 3 ⋅ 5 5 . 7 
π 1 1 1 1 π −2 1 1 1
or = + − + − … or = − + − ….
4 2 1. 3 3 . 5 5 . 7 4 1. 3 3 . 5 5 . 7

 x, − π< x<0
Example 7: Obtain Fourier Series of the function f ( x) = 
 − x, 0 < x < π.

 x, − π< x<0  − x, π> x>0

Solution: We have f ( x) =  and f (− x) =  x,
 − x, 0< x< π  0 > x > − π.

Thus f ( x) is an even function in (− π , π) and so bn = 0.

The Fourier series in this case is
∞
f ( x) = a0 + Σ an cos n x.
n =1
π
1 π 1 π 1  x2  1  π2  π
Here a0 =
π ∫0 f ( x) dx =
π ∫0 − x dx = − 
π 2 
 =− 
π 2 
=− ;
2
0
2 π 2 π
an =
π ∫0 f ( x) cos nx dx =
π ∫0 − x cos nx dx

π
2  sin nx cos nx  2  (− 1)n 1 
=−  x + 2  = −  2 − 2 
π n n 0 π  n n 

2 0 , n is even
= [ 1 − (− 1)n ] = 
2 2
πn 4 / πn , n is odd ⋅
π 4 cos x cos 3 x cos 5 x
∴ The required series is f ( x) = − +  2 + 2
+ 2

+...
2 π 1 3 5 

Example 8: A periodic function of period 4 is defined as

f ( x) = | x |, − 2 < x < 2.
Obtain its Fourier series expansion.
Solution: We have
| x |, − 2 < x < 2

f ( x) =  x, 0< x<2
 − x, 2 < x < 0 .

Here f ( x) is an even function in the
interval (–2, 2) and so bn = 0.
The Fourier series in this case is
a0 ∞
f ( x) = + Σ an cos nx.
2 n =1
1 c 1 2 1 0
Here a0 =
c ∫ − c
f ( x) dx =
2 ∫0 x dx +
2 ∫ −2 (− x) dx
 C-235

2 0
1  x2  1  − x2  1 1
=   +   = (4 − 0 ) + (0 + 4) = 2 ;
2 2 2  2  4 4
0 −2

1 c nπ x
an = ∫ f ( x) cos dx
c − c c
1 2 nπ x 1 0 nπ x
= ∫ x cos dx + ∫ (− x) cos dx
2 0 2 2 −2 2
2
1  2 nπx   4 nπx  
=  x sin  − (1)  − 2 2 cos  
2   nπ 2   n π 2  
0
0
1  2 nπx   4  nπx 
+ (− x)  sin  − (− 1)  − 2 2  cos
2   nπ 2   n π  2  −2

=
1  0 + 4 (− 1)n − 4  + 1  0 − 4 + 4 (− 1)n 
2  n2 π2 n2 π2  2  n2 π2 n2 π2 
1 4 4
= 2 2
[(− 1)n − 1 − 1 + (− 1)n ] = 2 2 [ (− 1)n − 1 ]
2 n π n π
 − 8 , if n is odd

=  n2 π2
 0 , if n is even.
∴ The required Fourier series is
a πx 2 πx
f ( x) = 0 + a1 cos + a2 cos +…
2 c c
cos πx cos
3 πx
cos
5 πx 
8  2 +… .
= 1− 2  2 2 + 2 +

π  1 32 52 
 

Comprehensive Exercise 3

1. Obtain Fourier series for the function f ( x) given by

1 + (2 x / π), − π ≤ x ≤ 0
f ( x) =  .
1 − (2 x / π), 0 ≤ x ≤ π.
1 1 1 π2
Deduce that + + +…= ⋅
12 32 52 8
2. Find the Fourier series expansion of the periodic function of period 2 π
f ( x) = x2 , − π ≤ x ≤ π .
1 1 1 1
Hence, find the sum of the series 2
− 2
+ 2
− +…
1 2 3 42
3. Obtain a Fourier expression for f ( x) = x3 for − π < x < π.
 C-236

4. Find the Fourier Series expansion for the function

f ( x) = x cos x, − π < x < π.
5. Find the Fourier Series for f ( x) in the interval − π < x < π where
 π + x, − π < x < 0
f ( x) = 
 π − x, 0 < x < π. (Gorakhpur 2005, 08, 16)
6. Find the Fourier Series expansion of the periodic function
 − x + 1, − π < x < 0
F ( x) = 
 x + 1, 0 < x < π. (Gorakhpur 2012)

A nswers 3
8 cos x cos 3 x cos 5 x
1. f ( x) = + + +....
π2  12 32 52 
π2 ∞ cos nx π2
2. x2 = + 4 Σ (− 1)n ; ⋅
3 n =1 n2 12
  π2 6  π2 6  π2 6 
3. x3 = 2 −  − + 3  sin x +  − + 3  sin 2 x −  − + 3  sin 3 x …
  1 1   2 2   3 3  
1 4 sin 2 x 6 sin 3 x
4. x cos x = − sin x + 2 − 2 +…
2 2 −1 3 −1
π 4 cos x cos 3 x cos 5 x 
5. f ( x) = + + + + ...
2 π  12 32
52

π 4 cos x cos 3 x cos 5 x 
6. f ( x) =  + 1 −  2 + + + ...
 2  π  1 32 52 

7 Half Range Fourier Cosine and Sine Series

[Interval ( 0, T )]
For the interval (0 , T ) the Half Range Fourier Cosine Series is
a πx 2 πx nπ x
f ( x) = 0 + a1 cos + a2 cos + … + an cos +…
2 T T T
2 T 2 T nπx
T ∫0 T ∫0
where a0 = f ( x) dx, an = f ( x) cos dx.
T
Thus when f ( x) is an even function with period T , we have
a ∞ nπx
f ( x) = 0 + Σ an cos ,
2 n =1 T
2 T 2 T nπx
where a0 =
T ∫0 f ( x) dx, an =
T ∫0 f ( x) cos
T
dx .
 C-237

In this case bn = 0.
Again for the interval (0 , T ) the Half Range Fourier sine series is
πx 2 πx nπx
f ( x) = b1 sin + b2 sin + … + bn sin + …,
T T T
2 T nπx
T ∫0
where bn = f ( x) sin dx.
T
Thus when f ( x) is an odd function with period T , we have
∞ nπx 2 T nπx
f ( x) = Σ bn sin
n =1 T
, where bn =
T ∫ 0
f ( x) sin
T
dx.

In this case a0 = 0 = an .

Remark: While expanding a function in (0 , T ) as a series of sines or cosines, we only see

if it is an odd or even function of period 2T . It makes no matter if the function is odd or
even or neither.

Example 9: Find the Fourier half-range cosine series of the function

 2 t, 0 < t <1
f (t) = 
2 (2 − t), 1 < t < 2.
Solution: The Fourier half range cosine series in interval (0 , T ) is
a0 πt 2 πt 3 πt
ƒ(t) = + a1 cos + a2 cos + a3 cos +… …(1)
2 T T T
Here T = 2 . We have
2 T 2 1 2 2
a0 =
T ∫0 ƒ(t) dt =
2 ∫0 2 t dt +
2 ∫1 2(2 − t ) dt

1 2 2
    t2    2 
=  t2  + 2 2 t −   = 1 + 4 t − t 
  0   2  
1
 1

= 1 + (8 − 4 − 4 + 1) = 2 ,

2 T nπt
an =
T ∫0 ƒ(t) cos
T
dt

2 1 nπt 2 2 nπt
=
2 ∫0 2 t cos
2
dt +
2 ∫1 2(2 − t) cos
2
dt

1
 2 nπt   4 nπt  
= 2 t  sin  − (2)  − 2 2 cos 
  nπ 2   n π 2   0

 2 nπt   4 nπt  
2
+ (4 − 2 t)  sin  − (−2)  − 2 2 cos 
  nπ 2   n π 2  1
 C-238

4 nπ 8 nπ 8 
=  sin + 2 2 cos − 2 2
 nπ 2 n π 2 n π 
nπ nπ 
+ 0 − 2 2 cos nπ −
8 4 8
sin + 2 2 cos
 n π nπ 2 n π 2 
8 nπ 8 8 8 cos nπ
= 2 2 cos − 2 2 − 2 2 cos nπ + 2 2
n π 2 n π n π n π 2
nπ
= 2 2 2 cos
8
− 1 − cos nπ ⋅
n π 
 2 
∞ 1  nπ nπt
8 
∴ f ( x) = 1 + 2 Σ 2 2 cos − 1 − cos nπ cos ⋅
π n = 1 n  2  2

Example 10: (i) Express f ( x) = x as a half-range sine series in 0 < x < 2 .

(ii) Express f ( x) = x as a half-range cosine series in 0 < x < 2.
Solution: (i) The Fourier sine series for f ( x) in (0, 2) is
∞ nπx
f ( x) = Σ bn sin
n =1 2
2 2 nπx 2 nπx
where bn =
2 ∫0 f ( x) sin
2
dx = ∫0 x sin
2
dx

2
nπx nπx  4 (−1)n
= −
2x 4
cos + 2 2 sin = − ⋅
 nπ 2 n π 2  0 nπ

∴ b1 = 4 / π, b2 = − 4 / 2 π, b3 = 4 / 3 π, b4 = − 4 / 4 π , etc.
Hence, the required half-range Fourier sine series for f ( x) in (0, 2) is
4 πx 1 2 πx 1 3 πx 1 4 πx 
f ( x) = sin − sin + sin − sin + … .
π 2 2 2 3 2 4 2 
(ii) The Fourier cosine series for f ( x) in (0, 2) is
a ∞ nπx
f ( x) = 0 + Σ an cos
2 n =1 2
2 2 2
where a0 = ∫
2 0
f ( x) dx = ∫ x dx = 2
0
2 2 nπx 2 nπx
and an = ∫ f ( x) cos dx = ∫ x cos dx
2 0 2 0 2
2
2 x nπx 4 n π x 4
=  sin + 2 2 cos  `= 2 2 [(−1)n − 1]
 nπ 2 n π 2 0 n π

∴ a1 = − 8 / π2 , a2 = 0 , a3 = − 8 / 32 π2 , a4 = 0 , etc.
a4 = 0 , a5 = − 8 / 52 π2 , etc.
Hence, the required half-range Fourier series for f ( x) in (0, 2) is
8 cos πx /2 cos 3 πx /2 cos 5 πx /2
f ( x) = 1 − + + + … ⋅
π2  12 32 52 
 C-239

Comprehensive Exercise 4

1. Find Fourier half-range even expansion of the function,

f ( x) = (− x / l) + 1, 0 ≤ x ≤ l.
2. Obtain the half-range sine series for the function f ( x) = x2 in the interval
0 < x < 3.
3. Find a series of sines of multiples of x which will represent f ( x) in the interval
π / 3 0 < x< π /3

(0 , π) where f ( x) =  0 π / 3 < x < 2π / 3
− π / 3 2 π / 3 π < x < π.

Also represent this function by a series of cosine of multiples of x as well. Draw
graph of these series and find the sine and cosine series where
1 2
x = − π, − π, − π .
3 3
 x , for 0 ≤ x ≤ π / 2 ,
4. Given f ( x) = 
 π − x , for π / 2 ≤ x ≤ π.
Express this function by a sine series and also by a cosine series.
5. Develop sin ( πt / l ) in half range cosine series in the range 0 < t < l.
6. Expand f ( x) = e x in a cosine series over (0 , 1).
 1 − x, if 0 < x < 1 ,

7. Expand f ( x) =  4 2
3 1
 x − , if < x < 1.
 4 2
8. Represent the following function by a Fourier sine series:
 t, 0 ≤ t< π/2
F (t) = 
 π / 2, π / 2 ≤ t < π. (Gorakhpur 2014)

A nswers 4
1 4 1 πx 1 3 πx 1 5 πx 
1. f ( x) = + cos + 2 cos + 2 cos … ⋅
2 π2 12 l 3 l 5 l 
2  108  sin x sin 3 x sin 5 x 
2. f ( x) =  + + + …
3  π3  13 33
5 3 
27  sin x sin 3 x sin 5 x  18  sin 2 x sin 4 x 
+  + + + … −  + +...
π  1 3 5  π  2 4 
1 1 1 1 1
3. f ( x) = sin 2 x + sin 4 x + sin 8 x + sin 10 x + ...
2 2 2 8 10 
π 81
cos 2 x + 2 cos 6 x + … ⋅
1
4. f ( x) = −
4 π 22 6 
 C-240

2 4 1 2πt 1 4πt 1 6πt 

5. f (t) = − cos + cos + cos + … ⋅
π π 3 l 15 l 35 l 
− e − 1 e −1 − e −1 
6. ex = e − 1+ 2  2 cos πx + 2
cos 2 πx + 2
cos 3 πx + …
 π + 1 4 π +1 9 π +1 
1 4  1 4   1 4 
7. f ( x) =  − 2  sin π x +  +  sin 3 π x +  −  sin 5 π x + …
π π   3 π 32 π2   5 π 52 π2 
2  1  2 1
8. f ( x) =  + 1 sin t − sin 2 t +  − +  sin 3 t + …
π  2  9π 3

8 Parseval’s Formula
If the Fourier series for f ( x) converges uniformly in (−l, l ), then
l a 2 ∞ 
∫−l [ f ( x)]2 dx = l  0 + Σ (an2 + bn2 ) ⋅
 2 n = 1 
Corollary 1: If the Fourier series for f ( x) in the interval 0 < x < 2 l converges
2l a 2 ∞ 
uniformly, then ∫ [ f ( x)]2 dx = l  0 + Σ (an2 + bn2 ) ⋅
0
 2 n =1 
Corollary 2: If the half range Fourier cosine series for f ( x) in the interval 0 < x < l
l l a 2 ∞ 
converges uniformly, then ∫ [ f ( x)]2 dx =  0 + Σ an2  ⋅
−l 2 2 n =1 
Corollary 3: If the half range Fourier sine series for f ( x) in the interval 0 < x < l
l l ∞
converges uniformly, then ∫ [ f ( x)]2 dx = Σ bn2 .
0 2 n =1
Corollary 4: Root Mean Square Value. The root mean square value or the effective
value of the function f ( x) denoted by [ f ( x)]rms over an interval (a, b) is defined as
1 /2
 b [ f ( x)]2 dx 
∫a 
[ f ( x)]rms =  ⋅
 b − a 
 
The root mean square value of a periodic function is frequently used in electric circuit
theory and in the theory of mechanical vibrations.

Example 11: Prove that for 0 < x < π

π2 cos 2 x cos 4 x cos 6 x 
(a) x (π − x) = − 2 + + + … ⋅
6  1 22 32 
 C-241

8 sin x sin 3 x sin 5 x 

(b) x (π − x) =  12 + 32 + 52 + … ⋅
π
Deduce from (a) and (b) respectively that
∞ 1 π4 ∞ 1 π
(c) Σ 4
= . (d) Σ = ⋅
n =1 n 90 n =1 n6 945
Solution: Let f ( x) = x (π − x), 0 < x < π.
(a) The half range cosine series for f ( x) is
a ∞
f ( x) = 0 + Σ an cos nx.
2 n =1
π
2 π 2  πx2 x3  2  π3 π3  π2
Here a0 =
π ∫0 x (π − x) dx = 
π 2
−
3
 = 
π2
−
3
=
3
0
2 π
and an =
π ∫0 x (π − x) cos nx dx

π
2 sin nx  − cos nx   − sin nx  
= (πx − x2 ) − (π − 2 x)   + (− 2)  
π n  n2   n3   0

2  π (−1)n π 2 π
= 0 − + 0 − 2  =  2  [− (−1)n − 1]
π  n2
n  π  n 
 −4 / n2 , when n is even
=
 0 , when n is odd.
π2 cos 2 x cos 4 x + cos 6 x + … ⋅
∴ x (π − x) = −4 2 + 
6  2 42 62
By Parseval’s formula, we get
2 π a02 ∞

π ∫0 x2 (π − x)2 dx =
2
+ Σ an2
n =1

π 1  π4 
  + 16  4 + 4 + 4 + …
2 1 1 1
or
π ∫0 (π2 x2 − 2 πx3 + x4 ) dx =
2 9 2 4 6 
π
2  π2 x3 2 πx4 x5  π4  1
+  4 + 4 + 4 + …
1 1
or  − +  =
π 3 4 5 18 1 2 3 
0

2  x5 2 π5 π5  π4  1
+  4 + 4 + 4 + …
1 1
or  − + =
π 3 4 5  18 1 2 3 
π4 π4 ∞ 1 ∞ 1 π4
or = + Σ 4
or Σ 4
= ⋅
15 18 n =1 n n =1 n 90
∞
(b) The half range sine series for f ( x) is f ( x) = Σ bn sin nx .
n =1
2 π
Here bn =
π ∫0 x (π − x) sin nx dx
 C-242

π
2  − cos nx   − sin nx  cos nx 
= (πx − x2 )   − (π − 2 x)   + (− 2) 
π  n   n2  n3  0

2  (−1)n 2 4 n
= − 2 3 + 3  = 3 [− (−1) + 1]
π  n n  π n
 8
, when n is odd

= n3 π
 0 , when n is even.
8 sin x sin 3 x sin 5 x 
∴ x (π − x) =  13 + 33 + 53 + … ⋅
π
By Parseval’s formula, we get
2 π ∞

π ∫0 x2 (π − x)2 dx = Σ
n =1
bn2

π2 64  1 π4
= 2  6 + 6 + 6 + …
1 1 1 1 1
or or = + + ⋅
15 π 1 3 5  960 16 36 56
1 1 1 1
Let S = 6 + 6 + 6 + 6 +…
1 2 3 4
1 1 1   1 1 1 
=  6 + 6 + 6 + … +  6 + 6 + 6 + …
1 3 5  2 4 6 
π4  1 1 1 
= +  6 + 6 + 6 + …
960  2 4 6 
π4 π4
+ 6  6 + 6 + 6 + … =
1 1 1 1 S
= + ⋅
960 2 1 2 3  960 64
S π4 63 S π4
∴ S− = or = ⋅
64 960 64 960
π4 64 π4 ∞ 1 π4
or S= × = or Σ = ⋅
960 63 945 n = 1 n6 945

πx , 0 < x <1

Example 12: If f ( x) = 
 π (2 − x) , 1 < x < 2;
1 1 1 π4
using half range cosine series, show that + + +…= ⋅
14 34 54 96
Solution: The half range cosine series for f ( x) in (0 , c ) is
a ∞ nπx
f ( x) = 0 + Σ an cos ⋅
2 n =1 c
2 1
π (2 − x) dx
2 c 2
Here a0 =
c ∫0 f ( x) dx =
2 ∫ 0
πx dx + ∫1 
1 2
 x2   x2  π   1 
= π   + π 2 x −  = + π (4 − 2) − 2 −  = π;
 2 0  2 2  2  
1
 C-243

2 c nπx
an =
c ∫0
f ( x) cos
c
dx

2 1 nπx 2 nπx 
= ∫ πx cos dx + ∫1 π (2 − x) cos dx
2 0 2 2 
1 2
 nπx  nπx    nπx  nπx  
 x sin  cos   sin  cos 
=π 2 − − 2   + π (2 − x) 2 − (−1)  − 2 
 nπ  2 2
n π   nπ  2 2
n π 
     
2  4  0  2  4   1
nπ nπ
= π  sin − 2 2 
2 4 4
+ 2 2 cos
 nπ 2 n π 2 n π 
nπ nπ 
+ π 0 − 2 2 cos nπ −
4 2 4
sin + 2 2 cos
 n π nπ 2 n π 2 
nπ
= π  2 2 cos − 2 2 − 2 2 cos nπx
8 4 4
n π 2 n π n π 
nπ
= 2 2 cos
4
− 1 − cos nπ ⋅
n π  2 
Putting n = 1, 2 , 3, … , we get
−4 −4
a1 = 0 , a2 = , a3 = 0 , a4 = 0 , a5 = 0 , a6 = …
π 9π
By Parseval’s formula, we get
c c  a02 
∫0 [ f ( x)]2 dx = 
2 2
+ a12 + a22 + a32 + …

1 2 2  π2 16 16 
or ∫0 (πx)2 dx + ∫1 π2 (2 − x)2 dx =
2

2
+ 2 +
π 81π 2
+ …

1 2
 x3  (2 − x)3  π2 16 16
or π2   − π2   = + 2 + +…
3
 0  3  1 2 π 81 π2

π2  1 π2 16  1
or − π2  0 −  = + 2 1+ + …
3  3 2 π  81 
2 π2 π2 16 1 + 1 + 1 + …
or − = 2  34 54 
3 2 π
π2 16 1 + 1 + 1 + … π4 1 1
or = 2  34 54  or = 1+ 4 + 4 + ….
6 π 96 3 5

9 Complex Form of Fourier Series

The Fourier series expansion of a periodic function f ( x) of period 2T is given
∞ nπx nπx 
a 
by f ( x) = 0 + Σ  an cos + bn  …(1)
2 n =1  T T 
 C-244

1 iθ 1
We have cos θ = (e + e − i θ ) and sin θ = (e i θ − e − i θ ),therefore from (1) we get
2 2i
inπx /T
a0 ∞   e + e − inπx /T   e inπx /T − e − inπx /T  
f ( x) = + Σ  an   + bn   
2 n =1   2   2i  
∞
= c0 + Σ { c n e inπx /T + c − n e − inπx /T }
n =1 …(2)
a0 1 1
where c0 = , c n = (an − ibn), c − n = (an + ibn).
2 2 2
1 2T
2T ∫ 0
Thus c0 = f ( x) dx ,

1  2T nπx 2T nπx 
2T  ∫ 0
cn =  f ( x) cos dx − i ∫ f ( x) sin dx
T 0 T 
1 2T  nπx nπx 
2T ∫ 0
= f ( x) cos − i sin  dx
 T T 
1 2T
f ( x) e − inπx /T dx
2T ∫ 0
=

1 2T  nπx nπx  1 2T
f ( x) e inπx /T dx.
2T ∫ 0 2T ∫ 0
and c −n = f ( x) cos + i sin  dx =
 T T 
Combining these, we get
1 2T
f ( x) e − inπx /T dx .
2T ∫ 0
cn = …(3)

(n = 0 , ± 1, ± 2 , ……)
Hence, the series (2) can be compactly written as
∞
f ( x) = Σ c n e inπx /T . …(4)
n= ∞

This is the complex form of Fourier series and its coefficients are given by (3).
If the function is defined in the interval (− T , T ), then the coefficients are given by
1 T
f ( x) e − i n πx /T dx.
2T ∫ −T
cn = (n = 0 , ± 1, ± 2,...)

The complex form of a Fourier series is especially useful in problems on electrical

circuits having periodic voltage.

Example 13: Obtain the complex form of the Fourier series of the function
0, − π≤ x≤0
f ( x) = 
 1, 0 ≤ x ≤ π.
Solution: The complex form of Fourier series for f ( x) is
 C-245
∞
f ( x) = Σ c n e inπ x /T
.
n= − ∞
1 T 1 π
Here cn = ∫ −T f ( x) e − inπ x /T
dx = ∫−π f ( x) e − inx dx
2T 2π
1  0 π
0 . e − inx dx + 1 . e − inx dx
2 π ∫ − π ∫0
=

π
1 π 1  e − inx  1
= ∫0 e − inx dx =   =− [e − inπ − 1]
2π 2 π  − in  2 nπ i
0

 1 , n is odd
1 1 
=− [cos nπ − 1] = − [(−1)n − 1] =  inπ
2 nπ i 2 nπ i  0 , n is even

1 T 1 π 1 π 1 1
and c0 =
2T ∫ −T f ( x) dx =
2π ∫ −π f ( x) dx =
2π ∫0 dx =
2π
[ x ]0π = 2 ⋅
1 1  e ix e 3 ix e 5 ix  1  e − ix e − 3 ix e − 5 ix 
∴ f ( x) = +  + + + … +  + + + …
2 iπ  1 3 5  i π  − 1 − 3 − 5 
1 1  ix
) + (e3 ix − e − 3 ix ) + (e5 ix − e −5 ix ) + … ⋅
− ix 1 1
= − (e − e
2 iπ 3 5 

Comprehensive Exercise 5

1. By using the sine series for f ( x) = 1 in 0 < x < π show that

π2 1 1 1
= 1+ 2 + 2 + 2 + …
8 3 5 7
2. Find the Fourier series for f ( x) = x2 in − π < x < π. Using the two values of f ( x),
1 1 π4 1 1
show that 4
+ 4
+ ⋅4
+ 4
+…+ =
1 2 3 4 90
3. Obtain the complex form of the Fourier series of
f ( x) = e − x in − 1 ≤ x ≤ 1.

A nswers 5
4 4 4 4
1. 1= sin x + sin 3 x + sin 5 x + sin 7 x + …
π 3π 5π 7π
n2 ∞ 4 ∞ (− 1)n (1 − in π)
2. f ( x) = + Σ 2 (− 1)n cos nx 3. e − x = Σ sinh 1 ⋅ e inπx .
3 n =1 n n= − ∞ 1 + n2 π2
 C-246

O bjective T ype Q uestions

Multiple Choice Questions

Indicate the correct answer for each question by writing the corresponding letter from (a),
(b), (c) and (d).
1. Fourier series for the function f ( x) in (− l, l ) is
a ∞ nπx
(a) f ( x) = 0 + Σ an cos ; f ( x) is even in (− l, l )
2 n =1 l
∞ nπx
(b) f ( x) = Σ bn sin ; f ( x) is odd in (− l, l )
n =1 l
a0 ∞ cos nπx ∞ nπx
(c) f ( x) = + Σ an + Σ bn sin
2 n =1 l n =1 l
(d) All of these

2. If f ( x) = x4 in (− 1 ≤ x ≤ 1), then the Fourier coefficient bn is

(a) 0 (b) 1
(c) 2 (d) None of these
3. The Fourier half range sine series for f ( x) = x in (0 , 2), then bn is
4 (− 1)n 4 (− 1)n
(a) (b) −
nπ nπ
4 (−1)n − 4 (− 1)n
(c) (d)
π π
1 1 1
4. The sum of the series + + + … is equal to
12 32 52
π2 π2
(a) (b)
12 8
π2 π2
(c) (d)
6 2
5. For Fourier expansion of an odd function in (− π, π)
(a) a0 is zero (b) an is zero
(c) bn is zero (d) both a0 and an are zero.

A nswers
1. (d) 2. (a) 3. (b) 4. (b)
5. (d)

¨
 1 Introduction
he calculus of variations began in 1696 AD with the brachistochrone problem.
T When Euler (1707-1783) discovered the basic differential equation known as
Euler’s equation for a minimizing curve, it matured into an independent
mathematical discipline after 1744 AD. It has been one of the major branches of
analysis for more than two centuries. It can be applied to a wide variety of problems in
pure mathematics. It is a tool of great power that can also be used to express the basic
principles of mathematical physics in forms of the utmost simplicity and elegance.
Most of the problems in physics and differential geometry can be formulated in
variational forms in which a curve x = x (t), y = y (t), a ≤ t ≤ b is sought. The integral
b
I = ∫ a F [ x (t), y (t), x ′ (t), y ′ (t)] dt
is a maximum or a minimum from amongst a class of curves. Here the integral I instead
of depending on independent variables varying over a given range, depends on the
entire course of functions x (t), y (t) drawn from a certain class of functions. These are
called admissible functions.
 C-248

2 Functionals
Functionals: A functional I is said to be defined on the class M of functions y ( x) if to
each function y ( x) ∈ M there is associated, by source law, a definite number I. It is
denoted as
I = I [ y ( x)] .
I [ y( x)] is said to be a functional with argument function y( x). The class M is called the
set of admissible functions of the domain of the functional.
(Gorakhpur 2007, 10, 12)
A quantity whose values are determined by one or several functions is called a
functional. The examples of functionals are :
(i) A simple example of a functional is the length l between two given points ( x1, y1) and
( x2 , y2 ) on a curve y = y ( x). Here the length is given by
1 /2
x2   dy  2 
l [ y ( x)] = ∫ 1 +    dx.
x1
  dx  
(ii) Another example of a functional is the area S of a surface bounded by a given curve C
because this area is determined by choice of the surface z = z ( x, y) as
1 /2
 2
 ∂z  
2
 ∂z 
S [z ( x, y)] = ∫∫ 1 +   +    dx dy,
D  ∂x   ∂y  
 
where D is the projection of the area bounded by the curve C on the xy-plane.

3 Variation of Argument Function

The difference between two functions y ( x) and y1 ( x) belonging to the chosen class M
of functions is called the increment or variation δy of the argument function y ( x) of the
functional I [ y ( x)].
δy = y( x) − y1( x).

4 Notion of Continuity of Functionals

A functional I [ y ( x)] is said to be continuous if a small change in y ( x) results in a small
change in I [ y ( x)]. Geometrically we say that if the curves y = y ( x) and y = y1 ( x) are
closed to each other, then the numbers I [ y ( x)] and I [ y1 ( x)] should be closed to each
other. The closeness of y ( x) and y1( x) is specified such that the absolute value of their
difference given by | y ( x) − y1 ( x)| is small for all x for which y ( x) and y1 ( x) are
defined. In this case, we say that y ( x) is close to y1 ( x) in the sense of zero-order
proximity. (Lucknow 2010)
 C-249

Alternatively, the two curves are close if not only the ordinate difference| y ( x) − y1 ( x)|
be small, the difference of slopes | y ′ ( x) − y ′1 ( x)| is also small. In this case the two
curves are said to be close in the sense of first-order proximity. (Lucknow 2010)

Proximity of Curves: If for the curves y = y ( x) and y = y1 ( x), the values of

(n) (n)
| y ( x) − y1 ( x)|,| y ′ ( x) − y ′1 ( x)|, … , | y ( x) − y1 ( x)|

are small for values of x for which these functions are defined, then the curves are said to
be close in the sense of nth order proximity. (Lucknow 2009)

If the two curves are close in the sense of nth order proximity, then they are certainly
close in the sense of any lower order proximity.
Y Y

B
B

A A
O X O X

Curves close in the Curves close in the

sense of zero-order sense of first order
proximity but not in proximity
the sense of first
order proximity

5 Continuity of a Functional
(Gorakhpur 2007, 10, 12)
The functional I [ y ( x)] is said to be continuous at y = y0 ( x), in the sense of nth order
proximity, if for a given positive number ε, there exists a δ > 0 such that
| I [ y ( x) ] − I [ y0 ( x)]| < ε
(n) (n)
for | y ( x) − y0 ( x)| < δ,| y ′ ( x) − y ′0 ( x)| < δ, … ,| y ( x) − y0 ( x)|< δ.

6 Linear Functional
(Gorakhpur 2006, 08, 11, 13, 14)
The functional I [ y ( x)] defined in the class of functions M is said to be a linear
functional if
(i) I [α y ( x)] = α I [ y ( x)] , α is a constant.
(ii) I [ y1 ( x) + y2 ( x)] = I [ y1 ( x)] + I [ y2 ( x)], where y1 ( x), y2 ( x) ∈ M.
Illustration : The functional
b
I1 [ y ( x)] = ∫ a [ y ′ ( x) + y ( x)] dx
 C-250

is linear, whereas the functional

b
∫ a [ p ( x) y ′
2
I1 [ y ( x)] = ( x) + q ( x) y2 ( x)] dx
is non-linear.

7 Increment of a Functional
Let I [ y ( x)] be a given functional on M. Also let δ ( y) = y ( x) − y1 ( x) be an increment
in the argument function y ( x). Then the increment of I [ y ( x)] corresponding to the
increment δy in the argument is given by the expression
∆I = I [ y ( x) + δ( y)] − I [ y ( x)] .

8 Variation of a Functional
Let L [ y ( x), δy] be a linear functional with respect to the argument δy and
β [ y ( x), δy] → 0 as the maximum value of δy (given by max |δy|) → 0. Also let the
increment ∆ I in the functional I [ y ( x)] be represented as
∆ I = L [ y ( x), δ ( y)] + β [ y ( x), δy] max | δy | ,
Then L [ y ( x), δy] is called the variation of the functional I [ y ( x)]or the first variation
of I and is denoted by δI.
In this case the functional I [ y ( x)] is said to be differentiable at the point y ( x).
Note : Here the variation δ I plays the same role for a functional I as the differential dy
does for a function y ( x).
Lemma: The variation of a differentiable functional I [ y ( x)] is given by
 ∂I [ y ( x) + α δy]
δI =   .
 ∂α α = 0
Proof: The increment ∆I of the differentiable functional I is
∆ I = L [ y ( x), δy] + β [ y ( x), δy] max | δy | ,
where β → 0 as | δy | → 0.
Replacing δy by α δy, we have
I [ y + α δy] − I [ y] = δI = L [ y, α δy] + β [ y ( x), α δy]| α | max | δy |
where β → 0 as α → 0.
By linearity we have
L [ y ( x), α δy] = α L [ y ( x), δy]

∆I I [ y + α δy] − I [ y]
∴ =
α α
β [ y, α δy]| α | max | δy |
= L [ y, δ y] +
α
 C-251

∆I ∆ I ∂ I [ y + α δ y]
Since lim = lim =
α→0 α α→0 ∆α ∂α
β [ y, α δy]| α | max | δy |
and → 0 as α → 0,
α
we have the variation L [ y, δy] of a functional I [ y ( x)] as

 ∂I ( y + α δy)
L [ y, δ y] =   .
 ∂α α = 0

9 Extremals of a Functional
A functional I [ y ( x)] is said to attain a maximum on a curve y = y0 ( x), if the value of I
on any curve close to y = y0 ( x) does not exceed I [ y0 ( x)], i. e.,
∆ I = I [ y ( x)] − I [ y0 ( x)] ≤ 0.
If ∆I ≤ 0 and ∆ I = 0 only on y = y0 ( x), then a strict maximum is said to be attained on
y = y0 ( x).
If ∆I ≥ 0 for all curves closed to y = y0 ( x), then the functional I is said to attain a
minimum on the curve y = y0 ( x). A strict minimum can be defined in the same way.
Theorem: If functional I [ y ( x)] attains a maximum or minimum on y = y0 ( x), where
the domain of definition belongs to certain class of functions M, then at y = y0 ( x),
δI = 0.
Proof: For a fixed y0 ( x) ∈ M and δy, we have
ψ (α) = I [ y0 ( x) + α δy]
a function of α, which reaches a maximum or minimum for α = 0.
∴ ψ ′ (0 ) = 0
 ∂ 
⇒  ∂α I [ y0 ( x) + α δy] =0
 α = 0

i. e., δI = 0.
Hence the theorem is proved.

10 Extremals with fixed Boundaries. The Fundamental

Lemma of the Calculus of Variation
If for every continuous function η ( x), we have (Lucknow 2007, 08, 10, 11)
b
I = ∫ a φ ( x) η ( x) dx = 0, …(1)

where φ ( x) is a continuous function on the interval [a, b], then φ ( x) ≡ 0 on [a, b] .

Proof: Assume that φ ( x) ≠ 0 be positive at a point x = x in a ≤ x ≤ b . As φ ( x) is
 C-252

continuous on [a, b], it follows that if φ ( x) ≠ 0 then φ ( x) maintains its positive sign in a
certain neighbourhood x0 ≤ x ≤ x1 of the point x since η ( x) is an arbitrary continuous
function, it may be so chosen that it remains positive in x0 ≤ x ≤ x1 but it vanishes
outside this interval.
Y

O a x0 x x1 b X
b x1
∴ ∫ a φ ( x) η ( x) dx = ∫ x0 φ ( x) η ( x) dx > 0 , …(2)

since the product φ ( x) η ( x) remains positive in [a, b] and vanishes outside this interval.
The contradiction between (1) and (2) shows that our assumption φ ( x) ≠ 0 at a point x
must be wrong and so we have φ ( x) ≡ 0 on [a, b] .

11 Euler’s Equation (Necessary Condition for Existence of

Extremal
(Meerut 2008; Lucknow 08, 10)
Y
Consider the functional
b
I [ y ( x)] = ∫ a F ( x, y( x), y ′ ( x)) dx …(1) B

A
For an extreme value the boundary points of the admissible
curves being fixed, we have y1 y2

y (a) = y1 …(2)
O a b X
and y (b) = y2 . …(3)
Y
We assume that the given function F ( x, y ( x), y ′ ( x)) has
continuous partial derivatives of second order with respect B
to any combination of its arguments. Let the curve y = y ( x) δy
y=y(x)
extremizes the functional (1) such that y ( x) is twice A

differentiable and satisfies the above boundary conditions

y1 y2
(2) and (3). Let y = y ( x) be an admissible curve close to
y = y ( x) such that both y ( x) and y ( x) can be included in a O a x b X

one parameter family of curves, then

y ( x, α) = y ( x) + α [ y ( x) − y ( x)]. …(4)
From (4) we get the curves y = y ( x) and y = y ( x) for α = 0 and α = 1 respectively.
The variation δy of the function y ( x) is
δy = y ( x) − y ( x). …(5)
 C-253

Therefore, from (4), we have

y ( x, α) = y ( x) + α δy …(6)
Since the variation δy is a function of x, from (5) we have
(δy)′ = y ′ ( x) − y ′ ( x) = δy ′. …(7)
From (7) it is clear that the derivative of the variation is equal to the variation of the
derivative.
We have all the family of curves (6), the functional (1) reduces to a function of α, say
ψ (α), such that
b
ψ (α) = ∫ a F ( x, y ( x, α), y ′ ( x, α)) dx …(8)

From calculus, we know that the necessary condition for the extremum of the function
ψ (α) for α = 0 is that its derivative for α = 0 must vanish, i. e.,
 dψ  = 0. …(9)
 dα 
α =0
From (8), by using Leibnitz's rule of differentiation under the integration sign, we have
dψ b ∂
=∫ F ( x, y ( x, α), y ′ ( x, α)) dx …(10)
dα a ∂α
Using chain rule of differentiation, we have
∂ ∂F ∂x ∂F ∂y ( x, α) ∂F ∂y ′ ( x, α)
F ( x, y ( x, α), y ′ ( x, α)) = + +
∂α ∂x ∂α ∂y ∂α ∂y ′ ∂α
…(11)
From (6), we have
y ′ ( x, α) = y ′ ( x) + α (δy)′ = y ′ ( x) + α δy ′
∂y ( x, α) ∂y ′ ( x, α)
= δy and = δy ′.
∂α ∂α
∂x
Using these results and noting that = 0, from (11) we have
∂α
∂
F ( x, y ( x, α), y ′ ( x, α)) = F y ( x, y ( x, α), y ′ ( x, y)) δy
∂α
+ F y ′ ( x, y ( x, α), y ′ ( x, α)) δy ′ …(12)
where a subscript denotes partial derivative with respect to the indicated variable.
Using (12), from (10) we have
dφ b
= ∫ [ F y ( x, y ( x, α), y ′ ( x, α)) δy + F y′ ( x, y ( x, α), y ′ ( x, α)) δy ′ ] dx
dα a

At α = 0,
 dφ  b
= ∫ [ F y ( x, y ( x), y ′ ( x)) δy + F y′ ( x, y ( x), y ′ ( x)) δy ′ ] dx
 dα  a
α =0
…(13)
= δl, the variation of the functional.
The necessary condition for the extremum of the functional is that its variation
vanishes i. e., δl = 0. Thus for the extremum of the functional (1), we have
 C-254

b
∫ a ( F y δy + F y′ δy ′ ) dx = 0
b b
or ∫a F y δy dx + ∫ a F y′ (δy)′ dx = 0 [∵ δy ′ = (δy)′ ]

b b d F y′
or ∫a F y δy dx + [ F y′ δy]ba − ∫a dx
δy dx = 0. …(14)

[integrating by parts the second term]

As both y = y ( x) and y = y ( x) pass through the points A (a, y1) and B (b, y2 ), we
have
y (a) = y (a) and y (b) = y (b). …(15)
From (5) and (15), we have
[δy] x = a = y (a) − y (a) = 0
and [δy] x = b = y (b) − y (b) = 0.
Thus from (14), we get
b  F − d F  δy dx = 0
∫a  y
 dx
y′  …(16)

d
Here the first factor F y − F y′ on the extremizing curve y = y ( x) is a given
dx
continuous function. Because of the arbitrary choice of the comparison curve y = y ( x),
the second factor δy is an arbitrary function subject to the vanishing of δy at x = a and
x = b.
Thus equation (16) and the fundamental lemma of the calculus of variation imply that
a necessary condition for the functional (1) to have an extreme value is that the
extremizing function y = y ( x) satisfies the differential equation
d
Fy − F y′ = 0
dx
∂F d  ∂F 
or −   = 0 , (a ≤ x ≤ b). …(17)
∂y dx  ∂y ′ 
The equation (17) is called the Euler’s equation for the functional (1). Some authors
also call it as Euler-Lagrange equation. (Gorakhpur 2007, 10)

Second Form of Euler’s Equation: We have F is a function of x, y, y ′, therefore

dF ∂F ∂F dy ∂F dy ′ ∂F ∂F ∂F
= + + ⋅ = + y ′+ y′ ′ …(18)
dx ∂x ∂y dx ∂y ′ dx ∂x ∂y ∂y ′
d  ∂F  d  ∂F  ∂F
Also  y′  = y′   + y′ ′ …(19)
dx  ∂y ′  dx  ∂y ′  ∂y ′
Subtracting (19) from (18), we get
dF d  ∂F  ∂F ∂F d  ∂F 
−  y′  = + y ′ − y′  
dx dx  ∂y ′  ∂x ∂y dx  ∂y ′ 
 C-255

d  ∂F  ∂F  ∂F d  ∂F  
or  F − y′  − = y′  −    = y ′⋅0, [Using (17)]
dx  ∂y ′  ∂x  ∂y dx  ∂y ′  

d  ∂F  ∂F
Hence  F − y′  − = 0. …(20)
dx  ∂y ′  ∂x
Equation (20) is called the alternative second form of Euler’s equation (17).

Third form of Euler’s Equation: We have F is a function of x, y, y ′, therefore

d  ∂F  ∂  ∂F  ∂  ∂F  dy ∂  ∂F  dy ′
  =   +   +  
dx  ∂y ′  ∂x  ∂y ′  ∂y  ∂y ′  dx ∂y ′  ∂y ′  dx

d  ∂F  ∂2 F ∂2 F ∂2 F
or   = + y′ + y′ ′ …(21)
dx  ∂y ′  ∂x ∂y ′ ∂y ∂y ′ ∂y ′2
Using (21), equation (17) gives
∂F ∂2 F ∂2 F ∂2 F
− − y′ − y′ ′ =0 …(22)
∂y ∂x ∂y ′ ∂x dy ′ ∂y ′2
Equation (22) is called the alternative third form of Euler’s equations.

b  y3 
Example 1: (i) Determine the extremals of the functional ∫ a  y +  dx .
3 
3
(ii) Determine the extremal of the functional ∫1 (3 x − y) y dx satisfying the boundary

9
conditions y (1) = 1, y (3) = ⋅
2
1
(iii) Determine the extremals of the functional ∫0 (e y + xy ′ ) dx satisfying the boundary

conditions y (0 ) = 0 , y (1) = 0 .
b  y3 
Solution: (i) Let I [ y ( x)] = ∫a y+


 dx
3 
…(1)

b
Comparing (1) with ∫a F ( x, y, y ′ ) dx, we have

y3
F ( x, y, y ′ ) = y + .
3
∂F d  ∂F 
In this case the Euler’s equation −   = 0 becomes
∂y dx  ∂y ′ 
d
1 + y2 − (0 ) = 0 or y = ± i, which is impossible.
dx
Hence, the given functional (1) has no extremal.
 C-256

3
(ii) Let I [ y ( x)] = ∫1 (3 x − y) y dx …(1)

3
Comparing (1) with ∫1 F ( x, y, y ′ ) dx, we have

F ( x, y, y ′ ) = (3 x − y) y.
∂F d  ∂F 
In this case Euler’s equation −   =0
∂y dx  ∂y ′ 
d
becomes 3x − 2 y (0 ) = 0 or 3x − 2 y = 0
dx
or 1 = 0, [Using the boundary condition y (1) = 1]
which is absurd.
Hence, the functional (1) has no extremal satisfying the given boundary conditions.
1
(iii) Let I [ y ( x)] = ∫0 (e y + xy ′ ) dx …(1)

1
Comparing (1) with ∫0 F ( x, y, y ′ ) dx, we have

F ( x, y, y ′ ) = e y + xy ′.
∂F d  ∂F 
In this case Euler’s equation −   =0
∂y dx  ∂y ′ 
d
becomes ey− (0 ) = 0 or e y =1
dx
or y = 0, …(2)
which satisfies the given boundary conditions.
Hence y = 0 is the only extremal of the functional (1) satisfying the given boundary
conditions.

π
Example 2: Determine the extremals of the functional ∫0 (4 y cos x + y ′2 − y2 ) dx which

satisfy the given boundary conditions y (0 ) = y (π) = 0 . (Kanpur 2013)

π
Solution: Let I [ y ( x)] = ∫0 (4 y cos x + y ′2 − y2 ) dx …(1)

π
Comparing (1) with ∫0 F ( x, y, y ′ ) dx, we have

F ( x, y, y ′ ) = 4 y cos x + y ′2 − y2 …(2)
∂F d  ∂F 
The Euler’s equation is −   =0 …(3)
∂y dx  ∂y ′ 
From (2), we have
∂F ∂F d  ∂F  d  dy 
= 4 cos x − 2 y, = 2 y′ ,   = 2  = 2 y ′ ′
∂y ∂y ′ dx  ∂y ′  dx  dx 
Putting these values in (3), we get
 C-257

d
4 cos x − 2 y − 2 y ′ ′ = 0 or ( D2 + 1) y = 2 cos x, D ≡ …(4)
dx
The auxiliary equation is
m2 + 1 = 0 or m = ± i .
∴ C.F. = c1 cos x + c2 sin x,
where c1 and c2 are arbitrary constants
1 x
and P.I. = 2 (2 cos x) = 2 sin x = x sin x
D +1 2

∴ the solution of (4) is

y = c1 cos x + c2 sin x + x sin x …(5)
Using the given boundary conditions, from (5) we have c1 = 0 . Putting this value of c1 in
(5), we get
y = c2 sin x + x sin x or y = sin x (c2 + x),
which is the required family of extremals.

2
Example 3: For the functional ∫ y ′ (1 + x2 y ′ ) dx, find the extremals passing through (0 , 3)
1

and (4, 11) such that its length between the given points is shortest.
Solution: Let A (0 , 3) and B (4, 11) be any two given points on the xy-plane. Let P ( x, y)
be any point on the curve joining A and B such that arc AB = s.
4 4
We have arc AB = ∫0 ds = ∫0 (1 + y ′2 )1 /2 dx …(1)

Let the length of the curve between the given points be minimum on the curve
y = f ( x).
4
Comparing (1) with ∫0 F ( x, y, y ′ ) dx, we get

F ( x, y, y ′ ) = (1 + y ′2 )1 /2 . …(2)
∂F d  ∂F 
Here the Euler’s equation −   = 0 gives
∂y dx  ∂y ′ 
d y′ d y′
0− =0 or =0
dx (1 + y ′2 )1 /2 dx (1 + y ′2 )1 /2
y′
or 2 1 /2
= c1 or y ′2 = c1 (1 + y ′2 )
(1 + y ′ )
dy c1
or = = c2 say or dy = c2 dx
dx (1 − c12 )1 /2
or y = c2 x + c3 , …(3)
where c2 and c3 are arbitrary constants. Equation (3) gives the extremals of (1).
Since the extremals of (1) must pass through the points (0, 3) and (4, 11), therefore
from (3) we get
 C-258

c3 = 3, and 11 = 4 c2 + c3
⇒ c2 = 2 and c3 = 3.
Thus from (3) we have
y = 2 x + 3,
which represents a straight line passing through the points (0, 3) and (4, 11).
Hence the required curve along which the distance between the points A and B is the
shortest is the straight line y = 2 x + 30.

1
Example 4: Determine the curves on which the functional ∫0 ( y ′2 + 12 xy) dx can be

extremized, y (0 ) = 0 , y (1) = 1. (Kanpur 2009, 14)

1
Solution: Let I [ y ( x)] = ∫0 ( y ′2 + 12 xy) dx. …(1)

1
Comparing (1) with ∫0 F ( x, y, y ′ ) dx, we have

F ( x, y, y ′ ) = y ′2 + 12 xy. …(2)
∂F d  ∂F 
The Euler’s equation is −   = 0. …(3)
∂y dx  ∂y ′ 
From (2), we have
∂F ∂F d  ∂F  d (2 y ′ )
= 12 x, = 2 y′ and   = = 2 y ′ ′.
∂y ∂y ′ dx  ∂y ′  dx
Putting these values in equation (3), we get
12 x − 2 y ′ ′ = 0 or y′ ′ = 6 x
or y ′ = 3 x2 + c1, where c1 is an arbitrary constant
or y = x3 + c1 x + c2 , …(4)
where c2 is an arbitrary constant.
Using the given boundary conditions y (0 ) = 0 , y (1) = 1 , we get c1 = c2 = 0 .
Hence, an extremum can be achieved only on y = x3 .

12 (a) Case I. When F is independent of y′ i. e., ∂F = 0.

∂y ′

∂F
In this case, the Euler’s equation reduces to =0 …(1)
∂y
which is a finite equation and not a differential equation. The solution of (1) does not
contain any arbitrary constant and thus it is not possible to find y satisfying the
boundary conditions y (a) = y and y (b) = y. Hence, this variational problem does not
in general admit a solution.
 C-259

Example 5: Test for an extremum the functional

b
I [ y ( x)] = ∫a y2 dx, y (a1) = y1, y (b) = y2 .
b
Solution: Comparing the given functional with ∫a F ( x, y, y ′ ) dx, we have

F ( x, y, y ′ ) = y2 . …(1)
The Euler’s equation is
∂F d  ∂F 
−   = 0. …(2)
∂y dx  ∂y ′ 
∂F ∂F
From (1), we have = 2 y and = 0.
∂y ∂y ′
Thus, Euler’s equation reduces to y = 0. The extremal y = 0 passes through the points
only for y1 = 0 and y2 = 0 . If y1 = 0 and y2 = 0 then the function y = 0 minimizes the
functional I [ y ( x)], since I [ y ( x)] ≥ 0. If at least one of the y1 and y2 is not zero, then
the functional I is not minimized on continuous functions.

12 (b) Case II. When F is linearly dependent on y′ such that

F ( x, y, y′ ) = M ( x, y ) + N ( x, y ) y′

∂F ∂M ∂N ∂F d  ∂F  d
We have = + y ′, = N ( x, y),   = N ( x, y). …(1)
∂y ∂y ∂y ∂y ′ dx  ∂y ′  dx
In this case the Euler’s equation
∂F d  ∂F  ∂M ∂N d
−   = 0 becomes + y′ − N ( x, y) = 0
∂y dx  ∂y ′  ∂y ∂y dx

∂M ∂N  ∂N ∂N dy 
or + y′ −  +  =0
∂y ∂y  ∂x ∂y dx 
∂M ∂N
so that − =0 …(2)
∂y ∂x
which is a finite equation and not a differential equation. The solution given by (2) does
not in general satisfy the given boundary conditions y (a) = y1 and y (b) = y2 . Hence
this variational problem does not in general admit a solution in the class of continuous
functions.
When equation (2) holds, M dx + N dy is an exact differential.
In this case the functional
b b
I [ y ( x)] = ∫a F ( x, y, y ′ ) dx = ∫a ( M dx + N dy)
 C-260

is independent of the path of integration. The value of the functional I [ y ( x)] is

constant on admissible curves and therefore the variational problem becomes
meaningless.

1
Example 6: Test for an extremum the functional ∫0 ( y2 + x2 y ′ ) dx, y (0 ) = 0 , y (1) = c.
1
Solution: Comparing the given functional with ∫0 F ( x, y, y ′ ) dx, we have

F ( x, y, y ′ ) = y2 + x2 y ′ …(1)
The Euler’s equation is
∂F d  ∂F 
−   =0 …(2)
∂y dx  ∂y ′ 
∂F ∂F d  ∂F  d 2
From (1) we have = 2 y, = x2 and   = ( x ) = 2 x. Thus, Euler’s
∂y ∂y ′ dx  ∂y ′  dx
equation reduces to
2 y − 2x = 0 or y=x …(3)
Equation (3) satisfies the boundary condition y (0 ) = 0 but the second boundary
condition y (1) = c is satisfied by (3) only if c = 1. However, if c ≠ 1 , then there is no
extremal satisfying the given boundary conditions.

12 (c) Caste III. When F is independent of x and y

In this case F is dependent only on y ′⋅ i. e., F = F ( y ′ ).
Third form of Euler’s equation is
∂F ∂2 F ∂2 F ∂2 F
− − y′ − y ′ ′ ′2 = 0 …(1)
∂y ∂x ∂y ′ ∂y dy ′ ∂y
∂F ∂F
We have F = F ( y ′ ) ⇒ = 0 and = 0.
∂y ∂x
∂2 F ∂2 F
∴ =0 and = 0.
∂x ∂y ′ ∂y ∂y ′
∂2 F
Thus from (1), we have y ′ ′ =0
∂y ′2
∂2 F
⇒ either, y ′ ′ = 0 or =0
∂y ′2
If y ′ ′ = 0, integrating twice, we get
y = c1 x + c2 , …(2)
which represents a two parameter family of straight lines.
 C-261

∂2 F
If = 0 has one or several real roots y ′ = k n, we get
∂y ′2
y = kn x + c, …(3)
which represents a one parameter family of straight lines contained in the two
parameter family given by (2). In this case extremals are all possible straight lines.

x2
Example 7: Determine the extremals ∫x1 (1 + y ′2 )1 /2 dx and hence show that the shortest
distance between two points in a plane is a straight line.
(Meerut 2008, 09; Gorakhpur 09, 13)
Solution: Let A ( x1, y1) and B ( x2 , y2 ) be two given points
Y
in the Euclidean xy-plane. Also let P ( x, y) be any point on B(x2, y2)
the curve joining A and B and Q be any point in the
neighbourhood of P such that arc AP = s and arc PQ = ds. ds Q
We have P(x, y)
x2 x2 s
arc AB = ∫x1 ds = ∫x1 (1 + y ′2 )1 /2 dx
A(x1, y1)

…(1) O X

where y ( x1) = y1 and y ( x2 ) = y2 . …(2)

x2
Comparing (1) with ∫ x1 F ( x, y, y ′ ) dx, we have

F ( x, y, y ′ ) = (1 + y ′2 )1 /2 . …(3)
The Euler’s equation is
∂F ∂2 F ∂2 F ∂2 F
− − y′ − y′ ′ =0 …(4)
∂y ∂x dy ′ ∂y ∂y ′ ∂y ′2
From (3), we get
∂F ∂F ∂2 F ∂2 F
= 0, =0 , = 0 and =0
∂y ∂x ∂x ∂y ′ ∂y ∂y ′
∴ from (4), we get
∂2 F
− y′ ′ =0 …(5)
∂y ′2
 ∂2 F 
or y′ ′ = 0 ∵ 2
≠ 0 
 ∂y ′ 
or y ′ = c1
or y = c1 x + c2 , …(6)
where c1 and c2 are arbitrary constants.
Here (6) is a straight line passing through A ( x1, y1) and B ( x2 , y2 ).
 C-262

∴ y1 = c1 x1 + c2 …(7)
and y2 = c1 x2 + c2 . …(8)
From (6), (7) and (8), we get
y − y1 = c1 ( x − x1) …(9)
and y2 − y1 = c1 ( x2 − x1). …(10)
Dividing (9) by (10), we get
y − y1 x − x1
=
y2 − y1 x2 − x1
y2 − y1
or y − y1 = ( x − x1) …(11)
x2 − x1
which is the equation of straight line joining the given points A and B.
Hence the shortest distance between two points A and B in a plane is a straight line.

12 (d) Case IV. When F is independent of y i. e. , F = F ( x, y′ )

∂F
In this case = 0.
∂y
The Euler’s equation
∂F d  ∂F  d  ∂F 
−   = 0 reduces to   =0
∂y dx  ∂y ′  dx  ∂y ′ 
Integrating, we get
∂F
= c, where c is a constant.
∂y ′
This being independent of y, can be solved for y ′ as a function of x.
Again integrating and using the given boundary conditions y (a) = y1 and y (b) = y2 ,
we get a solution involving two arbitrary constants.
∂F
In some cases = c can be solved by introducing a properly chosen parameter.
∂y ′

2 y ′2
Example 8: Prove that the extremal of ∫0 x
dx satisfying y (0 ) = 0 and y (2) = 1 is a

parabola.
2 y ′2
Solution: Let I [ y ( x)] = ∫0 x
dx …(1)

2
Comparing (1) with ∫0 F ( x, y, y ′ ) dx, we have
 C-263

y ′2
F ( x, y, y ′ ) = ⋅ …(2)
x
∂F d  ∂F 
The Euler’s equation is −   = 0. …(3)
∂y dx  ∂y ′ 
From (2), we get
∂F ∂F 2 y ′
=0 and = ⋅ …(4)
∂y ∂y ′ x
Putting these values in (3), we get
d  ∂F  d  ∂F 
0−   =0 or   =0
dx  ∂y ′  dx  ∂y ′ 
∂F
or = c1, where c1 is an arbitrary constant
∂y ′
2 y′
or = c1 [From (3)]
x
dy 1 1
or = c1 x or dy = c1 x dx
dx 2 2
1
or y = c1 x2 + c2 , …(5)
4
where c2 is an arbitrary constant.
Using the given boundary conditions y (0 ) = 0 and y (2) = 1 , (5) gives
c1 = 1 and c2 = 0 .
Hence from (5), it is clear that an extremum can be attained only on the curve
x2
y= or x2 = 4 y, which is a parabola.
4

Example 9: Show that the curve through (1, 0) and (2, 1) which minimizes
1 /2
2  (1 + y ′2 )
∫1  x2  dx is a circle. (Gorakhpur 2006, 08, 11, 12, 15, 16; Lucknow 06, 10)
 
1 /2
2  (1 + y ′2 )
Solution: Let I [ y ( x)] = ∫1  x2  dx …(1)
 
2
Comparing (1) with ∫1 F ( x, y, y ′ ) dx, we have
1 /2
1 + y ′2 
F ( x, y, y ′ ) =  2 
…(2)
 x 
The Euler’s equation is
∂F d  ∂F 
−   =0 …(3)
∂y dx  ∂y ′ 
From (2), we get
 C-264

∂F ∂F y ′ (1 + y ′2 )−1 /2
= 0 and = . …(4)
∂y ∂y ′ x
Putting these values in (3), we get
d  ∂F  d  ∂F 
0−   =0 or   =0
dx  ∂y ′  dx  ∂y ′ 
∂F
or = c1, where c1 is an arbitrary constant
∂y ′
y ′ (1 + y ′2 )−1 /2
or = c1 [From (4)]
x
or y ′ = c1 x (1 + y ′2 )1 /2 . …(5)
To solve (5) we introduce a new parameter t such that
y ′ = tan t. …(6)
∴ from (5), we get
tan t = c1 x (1 + tan2 t)1 /2 or tan t = c1 x sec t
1
or x = c2 sin t, where c2 = …(7)
c1
From (6), we get
dy = tan t dx = c2 tan t cos t dt [∵ from (7), dx = c2 cos t dt]
= c2 sin t dt.
Integrating, we get
y = − c2 cos t + c3 , where c3 is an arbitrary constant
∴ x = c2 sin t and y − c3 = − c2 cos t. …(8)
From (8), we get
x2 + ( y − c3 )2 = c22 …(9)
which represents a family of circles with centre on y-axis.
Using the boundary conditions x = 1, y = 0 and x = 2, y = 1,
Equation (9) gives
1 + c32 = c22 and 4 + (1 − c3 )2 = c22
∴ c2 = 5 , c3 = 2.
Hence from (9) the required curve is the circle x2 + ( y − 2)2 = 5.

12 (e) Case V. When F is dependent on y and y′ only i. e.,

F = F ( y, y′ ) i. e., F is independent of x.
(Gorakhpur 2006, 14; Meerut 06, 10; Lucknow 08)
∂F
In this case = 0.
∂x
 C-265

The Euler’s equations

d  ∂F  ∂F
 F − y′  − =0
dx  ∂y ′  ∂x
(alternative second form of Euler’s equation)
d  ∂F 
reduces to  F − y′  = 0.
dx  ∂y ′ 
Integrating, we get
∂F
F − y′ = c , where c is a constant.
∂y ′
This is a first order equation which does not contain x explicitly, which can be solved for
y ′ by integrating and separating the variables or by introducing a parameter.

Note: Sometimes Euler’s equation in its first form can also be used.

π/2
Example 10: Test for an extremal the functional I [ y ( x)] = ∫0 ( y ′2 − y2 ) dx with

π
y (0 ) = 0 , y   = 1 . (Lucknow 2010, 11)
 2
π/2
Solution: We have I [ y ( x)] = ∫0 ( y ′2 − y2 ) dx. …(1)

π /2
Comparing (1) with ∫0 F ( x, y, y ′ ) dx, we have

F ( x, y, y ′ ) = y ′2 − y2 …(2)
∂F d  ∂F 
The Euler’s equation is −   = 0. …(3)
∂y dx  ∂y ′ 
From (2), we get
∂F ∂F d  ∂F 
= − 2 y, = 2 y ′ and   = 2 y ′ ′.
∂y ∂y ′ dx  ∂y ′ 
Putting these values in equation (3), we get
d
− 2 y − 2 y′ ′ = 0 or ( D2 + 1) y = 0 , D ≡ ⋅ …(4)
dx
The auxiliary equation is
m2 + 1 = 0 or m = ± i.
Thus the general solution of (4) is
y = c1 cos x + c2 sin x. …(5)
Using the given boundary conditions y (0 ) = 0 and y π / 2 = 1 , we get
c1 = 0 and c2 = 1
Hence from (5), an extremum of (1) can be attained only on the curve y = sin x.
 C-266

Example 11: Find the curve passing through the points ( x1, y1) and ( x2 , y2 ) which when
rotated about the x-axis gives a minimum surface area. (Gorakhpur 2007, 09, 11, 15;
Meerut 10; Kanpur 10; Lucknow 08, 11)

Or
x2  x + b
Prove that the extremal of ∫ x1 y (1 + y ′2 )1 /2 dx is the catenary y = a cosh 
 a 
.
Y
Solution: The area of the surface of revolution formed by
P Q
revolving a curve joining A ( x1, y1) and B ( x2 , y2 ) about B (x2, y2)
A (x1, y1)
the x-axis is given by
x2 y ds
S [ y ( x)] = 2 π ∫ y (1 + y ′2 )1 /2 dx …(1) O X
x1
x2
Comparing (1) with ∫ x1 F ( x, y, y ′ ) dx, we have

F ( x, y, y ′ ) = y (1 + y ′2 )1 /2 , omitting the irrelevant factor 2π.

Here F ( x, y, y ′ ) is a function of y and y ′ only, therefore a first integral of Euler’s
equation is given by
∂F
F − y′ = constant = c
∂y ′
1
or y (1 + y ′2 )1 /2 − y ′ y (1 + y ′2 )−1 /2 2 y ′ = c
2
or y (1 + y ′2 ) − yy ′2 = c (1 + y ′2 )1 /2 or y = c (1 + y ′2 )1 /2
y2 − c 2 dy ( y2 − c 2 )1 /2
or y ′2 = or =
c2 dx c
dy dx
or 2 2 1 /2
= . [separating variables]
(y − c ) c

Integrating, we get
y x b
cosh −1 = +
c c c
 x + b
or y = c cosh  , …(2)
 c 
which is a two parameter family of catenaries having two arbitrary constants b and c.
These arbitrary constants can be obtained from the given boundary conditions. There
may exist one, two or no solution.

Example 12: Find the curve connecting the given points A and B which is traversed by a particle
sliding from A to B in the shortest time (friction and resistance of the medium are ignored).
(Brachistochrone Problem) (Lucknow 2006, 07; Meerut 06, 07, 08;
Kanpur 10; Gorakhpur 07, 11)
Solution: Let the origin be fixed at A with x-axis horizontal and y-axis vertically
downwards. Let the coordinates of B be ( x1, y1).
 C-267

Also let P ( x, y) be the position of the particle at any time t such that arc AP = s. By the
principle of conservation of energy, we have A(0, 0) M X
kinetic energy at P − kinetic energy at A = work done
90°
in moving the particle from A to P.
s
1 2
or mv − 0 = may
2
2 P (x
 ds  = 2 g y , y)
B(x1, y1)
or  
 dt 
ds Y
or = 2 g y. …(1)
dt
Thus the time taken by the particle in moving from A (0 , 0 ) to B ( x1, y1) is
B dx x (1 + y ′2 )1 /2
t [ y ( x)] = ∫A (2 g y)1 /2
=∫ 1
0 (2 g y)1 /2
dx …(2)

x1
Comparing (1) with ∫0 F ( x, y, y ′ ) dx, we have

(1 + y ′2 )1 /2
F ( x, y, y ′ ) = , …(3)
y1 /2
1
omitting the irrelevant factor ⋅
2g
Here F ( x, y, y ′ ) is independent of x, therefore a first integral of Euler’s equation is
∂F
F − y′ = constant = c1
∂y ′
1 y′
or 1 /2
(1 + y ′2 )1 /2 − (1 + y ′2 )−1 /2 2 y ′ = c1
y 2 y1 /2
or 1 + ( y ′ )2 − ( y ′ )2 = c1 y1 /2 (1 + y ′2 )1 /2
1
or y (1 + y ′2 ) = c2 , where c2 = . …(4)
c12
To solve equation (4) we introduce a new parameter t such that
dy
y′ = = cot t. …(5)
dx
From (4) and (5), we have
c2 1
y= 2
= c2 sin2 t = c2 (1 − cos 2 t)
1 + cot t 2
dy 2 c2 sin t cos t
and dx = = dt = 2 c2 sin2 t dt
y′ cot t
= c2 (1 − cos 2 t) dt.
Integrating, we get
 sin 2 t  c2
x = c2  t −  + c3 = (2 t − sin 2 t) + c3 .
 2  2
 C-268

Hence the parametric equations of the given curve are

c c
x − c3 = 2 (2 t − sin 2 t), y = 2 (1 − cos 2 t)
2 2
or x − c3 = a (φ − sin φ) , y = a (1 − cos φ) …(6)
c2
[Taking = a and 2t = φ]
2
At A (0 , 0 ), we have from (6),
φ = 0 and c2 = 0 .
Hence, the equation of a family of cycloids in standard parametric form is
x = a (φ − sin φ), y = a (1 − cos φ)
where a is the radius of a rolling circle determined by the fact that the cycloid passes
through B ( x1, y1).

13 Geodesics
A geodesic on a surface is a curve along which the distance between any two points of
the surface is a minimum. The problem of finding the geodesics on a surface is a
variational problem involving the conditional extremum. This problem was first
studied by Jacob Bernoulli in 1698 and its general method of solution was given by
Euler.
Remark 1: Geodesics on a plane are straight lines.
Remark 2 : Geodesics on a sphere of fixed radius are its great circles.
Remark 3 : Geodesics on a circular cylinder of fixed radius are circular helix.

Example 13: Find the geodesics on a sphere of radius a ?

Solution: The element of arc on the surface of a sphere of radius a is
ds2 = dr2 + (rdθ)2 + (r sin θ d φ)2 [In spherical coordinates]
2 2 2
= 0 + a dθ + (a sin dφ) [∵ r = a ⇒ dr = 0]

ds2 dφ 2  dφ 2 
= a2 + a2 sin2 θ   = a 1 + sin2 θ    ⋅
ds
or or
dθ2  dθ  dθ   dθ  

Therefore, arc length of any line on the sphere between θ1 and θ2 is
θ2
s = a∫ (1 + sin2 θ . φ′2 ) dθ.
θ1

Here F (θ, φ, φ′ ) = a (1 + sin2 θ . φ ′ 2 ).

The Euler’s equation reduces to

 C-269

d  ∂F 
  = 0, here F is a function of θ and φ′.
dθ  ∂φ′ 
 ∂F 
Integrating,   = constant.
 ∂φ′ 

∂F sin2 θ . φ′
∴ = = c1 (say)
∂φ′ √ (1 + sin2 θ . φ′2 )

or sin2 θ . φ′ = c1 √ (1 + sin2 θ . φ′2 )

or sin4 θ . φ′2 = c12 (1 + sin2 θ . φ′2 )

or sin2 θ (sin2 θ − c12 ) φ′2 = c12

dφ c1 c1 cos ec2 θ
or = =
2
dθ sin θ √ (sin θ − c1 ) √ (1 − c12 cos ec2 θ)
2

c1 cos ec2 θ
Integrating, φ= ∫ √ (1 − c12 cos ec2 θ)
dθ + c2

c1 cos ec2 θ dθ
= ∫ √ {(1 − c12 ) − (c1 cot θ)2 }
+ c2

 c cot θ   c cot θ 
φ = − sin−1  1 2 
+ c2 or sin−1  1  = c2 − φ
 √ (1 − c1 )  √ (1 − c12 )
c1 cot θ
or = sin (c2 − φ) = sin c2 cos φ − cos c2 sin φ
√ (1 − c12 )

or cot θ = A cos φ + B sin φ

or cos θ = A sin θ cos φ + B sin θ sin φ
or a cos θ = A a sin θ cos φ + B a sin θ sin φ
or z = Ax + By, in cartesian coordinates.
This represents a plane through the center (0,0,0) of the sphere which cuts the sphere
along a great circle. Hence the required geodesics are the arcs of the great circle.

Comprehensive Exercise 1

1
1. Find the stationary function of the functional ∫0 ( y2 + y ′2 ) dx that satisfies

the boundary conditions y (0 ) = 0 , y (1) = 1.

1
2. Determine the extremals and extremum value of the functional ∫1 /2 x2 y ′2 dx

satisfying y (1 / 2 ) = 1, y (1) = 2. (Gorakhpur 2007, 10, 12)

 C-270
π /2
3. Determine the extremal of the functional ∫0 ( y2 − y ′2 −8 y cosh x) dx that
π
satisfies the boundary conditions y (0 ) = 2 and y (π / 2) = 2 cosh   .
 2
π /2
4. On what curve the functional ∫ ( y ′2 − y2 + 2 xy) dx with y (0 ) = 0 and
0
π
y   =0 can be extremized.
 2
π /2
5. Find the extremal for the functional I [ y ( x)] = ∫0 ( y2 − y ′2 −2 y sin x) dx

π
that satisfies the boundary conditions y (0 ) = 0 , y   = 1 .
 2
x2
6. Test for an extremum of the functional I [ y ( x)] = ∫ x1 ( y + xy ′ ) dx, y ( x1) = y1

and y ( x2 ) = y2 .
1
7. Test for an extremum of the functional I [ y ( x)] = ∫0 ( y ′2 + y ′+1 ) dx,

y (0 ) = 1, y (1) = 2.
2  x3 
8. Determine the extremal of the functional ∫   dx, y (1) = 0 , y (2) = 3.
1  y ′2 

(Lucknow 2008, 11; Kanpur 11, 14)

2 1 /2
x (1 + y ′ )
∫x1
2
9. Find the extremals of the functional I [ y ( x)] = dx.
y
(Lucknow 2008, 11; Kanpur 10, 12)
10. Find the Euler’s equation for the extremals of the functional
x2
∫ x1 ( y2 − yy ′ + y ′2 ) dx.

b
11. Determine the extremal of the functional I [ y ( x)] = ∫a ( y2 + y ′ 2 + 2 ye x ) dx.
(Kanpur 2009)
12. (i) Find the geodesics on a right circular cylinder of radius a.
(ii) Find the geodesics on a right circular cone.
13. Show that the shortest line between any two points on a cylinder is a helix.

A nswers 1
sinh x 1
1. y= 2. y=− + 3; 1
sinh 1 x
π
3. y = 2 cosh x 4. y = x − sin x
2
x
5. y = sin x − cos x
2
 C-271

6. The integral is independent of the path of integration, hence the given

variational problem is meaningless.
7. y = x +1 8. y = x2 − 1
9. ( x − c2 )2 + y2 = c12 10. y′ ′ − y = 0
x x
11. y = c1 e x + c2 e − x + e
2
12. (i) z = c2 θ + c3 (ii) r sin α = c1 sec (φ sin α + c )

14 Variational Problems with Subsidiary Conditions

Various applications of the calculus of variations involve problems which not only have
boundary conditions, but also have conditions of quite a different type, known as
subsidiary conditions (side conditions or constraints).

15 Isoperimetric Problems
The word isoperimetric means ‘with the same parameter’. The ancient Greeks
purposed the problem of finding a closed curve of given length having maximum area.
They called this problem the isoperimetric problem. Further the term isoperimetric
problem is extended to include the general case of finding extremals of one integral
subject to any constraint requiring a second integral to take on a prescribed value.

16 Working Rule for Solving Isoperimetric Problems

Such problems are generally solved by the method of Lagrange multipliers.
Suppose we are to find a curve y = y( x) that gives extreme value of the functional
x2
I [ y( x)] = ∫ x1 f ( x, y, y ′ )dx. ...(1)

While keeping another integral

x2
J[ y( x)] = ∫ x1 g( x, y, y ′ )dx = Constant = c, say. ...(2)

Step 1: Let y = y( x) satisfy the boundary conditions

y( x1) = y1 and y( x2 ) = y2 . ...(3)
Step 2: Let F = f ( x, y, y ′ ) + λ g( x, y, y ′ )
where λ is called the Lagrange multiplier.
Step 3: The required extreme value of y( x) must satisfy Euler’s equation
∂F d  ∂F 
−   =0 ...(4)
∂y dx  ∂y ′ 
 C-272

d
or Fy − Fy ′ = 0
dx

fy−
d
f y′ + λ  g − d g  = 0.
or  y y′  ...(5)
dx  dx
Step 4: The extremals of our problem i. e, the solutions of equation (4) involve three
undetermined parameters, two constants of integrations and the Lagrange
multiplier. Definite values of these parameters can be found if an extremizing
function exists, by applying the two boundary conditions (3) and the value of the
integral prescribed in (2).

17 Finite Subsidiary Conditions

In the isoperimetric problem, the subsidiary conditions which must be satisfied are
specified by functionals. Now consider a different problem:
Find the functions yi( x) for which the functional has an extremum, where the
admissible functions satisfy the boundary conditions
yi(a) = Ai, yi(b) = Bi (i = 1,..., n)
and k finite subsidiary conditions (k < n)
g j( x, y1,..., yn) = 0 ( j = 1,... k )
In a certain sense, we can consider a variational problem with a finite subsidiary
condition to be a limiting case of a isoperimetric problem. The function λ ( x) appearing
in the problem with a finite subsidiary condition can be interpreted as a Lagrange
multiplier for each point x.
Remark: In elementary analysis for finding an extremum of a function of n variables
subject to k constraints (k < n), we can use the constraints to express k variables in terms
of the other n − k variables. Thus the problem is reduced to that of finding an
unconstrained extremum of a function of n − k variables i. e., an extremum subject to no
subsidiary conditions. In the calculus of variations we have the same situation. The
problem of finding geodesics on a given surface can be regarded as a problem subject to
a constraint. If we express the coordinates x, y and z as functions of two parameters, we
can reduce the problem to that of finding an unconstrained extremum.

2 2
Example 14: Find the extremal of the functional ∫ y ′2 dx under the constraint ∫ ydx = 1
0 0
given y(0 ) = 0 and y(2) = 1.
2
Solution: Let I = ∫0 y ′2 dx . ..(1)
2
and J= ∫0 ydx = 1. ...(2)
 C-273

The given boundary conditions are

y (0 ) = 0 and y (2) = 1. ...(3)
2
Let F ( x, y, y ′ ) = y ′ + λ y ...(4)
where λ is Lagrange multiplier.
The required extremal satisfies the Euler’s equation,
∂F d  ∂F 
−   =0
∂y dx  ∂y ′ 
d
or λ− (2 y ′ ) = 0 [Using (4)]
dx
λ d2 y d2 y λ
or − 2 =0 or = ⋅ ...(5)
2 dx dx2 2
Integrating (5) twice, we get
λ 2
y= x + c1 x + c2 , ...(6)
4
where c1 and c2 are arbitrary constants.
From (3) and (4), we have
(1 − λ )
c2 = 0 and 1 = λ + 2 c1 or c1 = ⋅
2
Putting these values in (6), we have
λ 2 1
y= x + (1 − λ ) x. ...(7)
4 2
Substituting this value of y in (2), we have
2
 λx2
2 x  λx3 x2 
∫0  4 + (1 − λ ) 2 dx = 1 or 
 12
+ (1 − λ )  = 1
4 
  0
2λ
or + 1− λ = 1 or λ = 0.
3
1
∴ c1 = ⋅
2
x
Hence from (6), the required extremal is y = .
2
π
Example 15: Find the extremal of the functional ∫0 ( y ′2 − y2 ) dx under the conditions
π
y(0 ) = 0 , y(π) = 1 and subject to the constraint ∫0 y dx = 1.
π
Solution: Let I = ∫0 ( y ′2 − y2 ) dx ...(1)

π
and J= ∫0 y dx = 1. ...(2)

The given boundary conditions are

y (0 ) = 0 and y (π) = 1. ...(3)
2 2
Let F ( x, y, y ′ ) = y ′ − y + λy ...(4)
where λ is the Lagrange multiplier.
 C-274

The required extremal satisfies the Euler’s equation

∂F d  ∂F  d
−   =0 or −2y + λ − (2 y ′ ) = 0
∂y dx  ∂y ′  dx
λ
or y′′ + y = [Using (4)]
2
λ d
or ( D2 + 1) y = , where D ≡ . ...(5)
2 dx
Auxiliary equation is
D2 + 1 = 0 or D = ± i.
∴ C.F. = c1 cos x + c2 sin x
where c1 and c2 are arbitrary constants.
1 λ λ λ
P.I. = 2 = (1 + D2 )−1 . 1 = (1 − D2 + 1...) . 1
D +1 2 2 2
λ
= ⋅
2
∴ the general solution of (5) is
λ
y = c1 cos x + c2 sin x + ⋅ ...(6)
2
From (3) and (6), we have
λ λ
0 = c1 + and 1 = − c1 + ⋅
2 2
1
∴ λ = 1 and c1 = − ⋅
2
Hence from (6), we get
1 1
y = − cos x + c2 sin x + ⋅ ...(7)
2 2
Substituting this value of y in (2), we have
π  1 1
∫0  − 2 cos x + c2 sin x + 2 dx = 1
π
or − 1 sin x − c cos x + x  = 1.
 2 2
2  0
π (2 − π)
∴ 2 c2 + =1 or c2 = ⋅
2 4
Hence, from (7), the required extremal is
1 (2 − π) 1
y = − cos x + sin x +
2 4 2
1 (2 − π)
or y = (1 − cos x) + sin x.
2 4

Example 16: Determine the curve of length l which passes through the points (0 , 0 ) and (1, 0 )
and for which the area between the curve and the x-axis is a maximum.
 C-275

Solution: Given that the plane curve passing

Y
through two given points O (0 , 0 ) and A(1, 0 ) has P Q
fixed perimeter l. Let S be the area enclosed by
the given plane curve and x-axis. Then, we are to
maximize
1
S= ∫0 y dx ...(1) O (0,0) ds A (1,0) X
subject to the constraint
1
∫0 (1 + y ′2 )1 /2 dx = l ...(2)

and the boundary conditions

y (0 ) = 0 and y (1) = 0 . ...(3)
2 1 /2
Let F ( x, y, y ′ ) = y + λ (1 + y ′ ) ...(4)
where λ is the Lagrange multiplies.
The required extremal satisfies the Euler’s equation
∂F d  ∂F 
−   =0
∂y dx  ∂y ′ 

d  λy ′ 
or 1−  2 1 /2 
= 0. [Using (4)] ...(5)
dx  (1 + y ′ ) 
Integrating (5), we have
λy ′ λ2 y ′2
x− = c1 or ( x − c1)2 =
(1 + y ′2 )1 /2 1 + y ′2
1 + y ′2 λ2 ( x − c1)2
or 2
= 2
or y ′2 = 2
y′ ( x − c1) λ − ( x − c1)2
dy x − c1 ( x − c1) dx
or =± 2 or dy = ± ⋅
dx { λ − ( x − c1)2 }1 /2 { λ − ( x − c1)2 }1 /2
2

Again integrating, we have

y = c2 ± { λ2 − ( x − c1)2 }1 /2 or ( y − c2 )2 = λ2 − ( x − c1)2
or ( x − c1)2 + ( y − c2 )2 = λ2 ,
which is the equation of a circle with radius λ.

Aliter: From (5), we get

1 − λ{ y ′ ′ (1 + y ′2 )1 /2 − y ′ (1 + y ′2 )− 1 /2 y ′ y ′ ′ }
=0
1 + y ′2
1 y′ ′ y′ ′ 1
or − =0 or = ,
λ (1 + y ′2 )3 /2 (1 + y ′ )2 3 /2 λ
1
showing that the curvature of the required curve is equal to a constant. Hence,
λ
required curve is an arc of a circle of radius λ.
 C-276

Example 17: Show that the curve C of given length l which minimizes the curved surface area of
solid generated by the revolution of C about x-axis is a catenary.
Y
Solution: Consider the arc PP′ of the curve to rotate about
the x-axis. Let l and S be the length of arc PP′ and surface of
the solid generated by arc PP′ about the x-axis. Let ( x1, y1) Q
l
and ( x2 , y2 ) be the coordinates of P1 and P2 respectively. P
Now we are to minimize
x2 O
S= ∫x1 2 πy(1 + y ′2 )1 /2 dx ...(1) X

subject to the constraint P'

x2 Q'
∫x1 (1 + y ′2 )1 /2 dx = constant = l ...(2)

and the boundary conditions y ( x1) = y1 and y ( x2 ) = y2 . …(3)

2 1 /2 2 1 /2
Let F ( x, y, y ′ ) = 2 πy (1 + y ′ ) + λ (1 + y ′ ) ,
where λ is the Lagrange multiplier.
The required extremal satisfies the Euler’s equation. (Note : Here F is independent
of x.)
 ∂F 
∴ F − y′   = constant = 2 πc1, where c1 is constant
 ∂y ′ 
or (2 πy + λ ) (1 + y ′2 )1 /2 − y ′ (2 πy + λ ){ y ′ /(1 + y ′2 )1 /2 } = 2 πc1. …(4)
Let us choose λ = − 2 πµ , where µ is new Lagrange multiplier. Then (4) may be
re-written as
y ′2 c1 1 c1
(1 + y ′2 )1 /2 − 2 1 /2
= or 2 1 /2
=
(1 + y ′ ) y −µ (1 + y ′ ) y −µ
2
 dy  (y − µ )2
or 1+   =
 dx  c12
dy {( y − µ )2 − c12 }1 /2
or = ⋅
dx c1
Separating the variables and integrating, we have
dx dy
∫ c1 = ∫ {( y − µ )2 − c12 }1 /2
x y −µ
or + c2 = cosh −1
c1 c1
x 
or y = µ + c1 cosh  + c2 
 c1 
x − b
or y = µ + a cosh  ,
Taking c = a, c = − b 
 a 
 a  1 2

which are the required catenaries with base lines parallel to the x-axis.
 C-277

Example 18: Among all curves of length l(> 2 a) in the upper half plane passing through the
points (−a, 0 ) and (a, 0 ), find the one which together with the segment − a ≤ x ≤ a enclose maximum
area. (Gorakhpur 2009, 11, 12, 13, 14)

Solution: Given l is the fixed perimeter Y

of a plane curve passing through two P Q
given points A (− a, 0 ) and B (a, 0 ).
Let S be the area enclosed by that plane
curve and x-axis.
Now, we are to maximize X ' A (–a,0) O R ds S B (a,0) X
a
S= ∫−a y dx …(1)

subject to the constraint

a
∫−a (1 + y ′2 )1 /2 dx = l …(2)

and the boundary condition

y (− a) = y (a) = 0 . …(3)
2 1 /2
Let F ( x, y, y ′ ) = y + λ (1 + y ′ ) …(4)
where λ is the Lagrange multiplier.
The required extremal satisfies the Euler’s equation
∂F d  ∂F 
−   =0
∂y dx  ∂y ′ 

d  λy ′ 
or 1−  2 1 /2 
= 0. [Using (4)] …(5)
dx  (1 + y ′ ) 
Integrating (5), we have
λy ′ λ2 y ′2
= x + c1 or ( x + c1)2 = …(6)
(1 + y ′2 )1 /2 1 + y ′2
dy x + c1 ( x + c1)dx
or =± 2 or dy = ± ⋅
dx { λ − ( x + c1)2 }1 /2 { λ2 − ( x + c1)2 }1 /2
Again integrating, we have
y + c2 = ±{ λ2 − ( x + c1)2 }1 /2
or ( x + c1)2 ( y + c2 )2 = λ2 . …(7)
Using the given boundary condition (3), (7) becomes
(− a + c1)2 + c22 = λ2 or a2 − 2 ac1 + c12 + c22 = λ2 …(8)
2 2 2
and (a + c1) + c22 =λ or a + 2 ac1 + c12 + c22 =λ . …(9)
Subtracting (8) from (9), we have
4 ac1 = 0 or c1 = 0 . …(10)
Putting this value of c1 in (8), we have
 C-278

a2 + c22 = λ2 or c2 = (λ2 − a2 )1 /2 . …(11)

Putting the value of c1 and c2 from (10) and (11) in (7), we have
x2 + { y + (λ2 − a2 )1 /2 }2 = λ2 . …(12)
Now from (6) and (10), we have
x2
y ′2 = ⋅
(λ − x2 )
2

∴ from (2), we have

1 /2
 x2  x a
= 2 λ sin−1 
a a dx
I = ∫−a 1 + λ2 − x2  dx = 2 λ ∫
0 2
(λ − x ) 2 1 /2  λ 0
 

l = 2 λ sin−1   sin   = ⋅
a l a
or or …(13)
λ  2λ  λ
Let λ = λ 0 be a solution of the transcendental equation (13).
∴ From (12), we have
x2 + { y + (λ 02 − a2 )1 /2 }2 = λ 02 . …(14)

Thus the required arc is an arc of the circle Y

 (λ 2 − a2 )1 /2 
(14) whose centre is c 0 − 0  0
 2  λ
and radius λ 0 as shown in the figure.
X ' A (–a,0) O B (a,0) X
Particular Case: C
Take l = πa so that l > 2 a.
Then, from (13), we have Y
π a  a
sin    =
2  λ   λ
giving λ = λ 0 = a.
Hence (14) reduces to circle
x2 + y2 = a2 as shown in the figure. X ' A (–a,0) O B (a,0) X
Y'

18 Variational Problems for Functionals Involving Several

Dependent Variables i. e. of the form
b
I = ∫ a F ( x, y1 ( x ), y 2 ( x ), … y n ( x ), y′ 1 ( x ), y′ 2 ( x ),... y′ n ( x )) dx
Consider the functional
b
I [ y1, y2 , …, yn] = ∫ f ( x, y1, y2 , …, yn, y ′1 , y ′2 , …, y ′ n ) dx …(1)
a

To find the necessary conditions for the extremum of the functional I, the following
boundary conditions are considered
 C-279

y1(a) = a1, y2 (a) = a2 , …… , yn (a) = an …(2)

y1(b) = b1, y2 (b) = b2 , …… , yn(b) = bn …(3)
where a1, a2 , … b1, b2 , … are constants. Let us vary only one of the functions
yi ( x), (i = 1, 2, ..... , n)
while keeping the rest of the functions unchanged.
Then the functional I reduces to a functional dependent only on one of the functions
yi ( x). Thus the extremizing function yi ( x) having continuous derivative must satisfy
Euler’s equation
d  ∂f 
F yi −   = 0,
dx  ∂y ′ i 
where the boundary conditions on yi ( x) at x = a and x = b are given by (2) and (3).
As this argument is applicable to any function yi ( x), (i = 1, 2, … , n), we obtain a system
of second order differential equations
d  ∂F 
Fy −   = 0, (i = 1, 2, … , n) …(4)
i dx  ∂y ′ i 
The equations (4) define, in general, a 2n-parameter family of curves in space
x, y1, y2 , … , yn. This is the required family of extremals for the given variational
problem.

Example 19: Determine the differential equations of the lines of propagation of light in an
optically non-homogeneous medium with the speed of light v = v ( x, y, z ).
Solution: According to Fermat’s principle, light propagates from one point
A ( x1, y1, z1) to another point B ( x2 , y2 , z2 ) along a curve for which T , the time of
passage of light is least. If the path of the light ray is given by y = y( x) and z = z ( x), we
have
x2 ds x (1 + y ′2 + z ′2 )1 /2
T = ∫ x1 v ( x, y, z )
=∫ 2
x1 v
dx …(1)

where ds denotes a line element of the path.

(1 + y ′2 + z ′2 )1 /2
Here F ( x, y, z , y ′ , z ′ ) = . …(2)
v
The system of the Euler’s equations is given by
d d
Fy − F y′ = 0 and Fz − Fz ′ = 0
dx dx
∂v (1 + y ′2 + z ′2 )1 /2 d y′
∴ 2
+ =0 …(3)
∂y v dx v (1 + y ′ + z ′2 )1 /2
2
 C-280

∂v (1 + y ′2 + z ′2 )1 /2 d z′
and 2
+ = 0. …(4)
∂z v dx v (1 + y ′ + z ′2 )1 /2
2

The differential equation (3) and (4) determine required lines of light propagation.

Example 20: Find the extremals of the functional

π /2
I [ y ( x), z ( x)] = ∫0 ( y ′2 + z ′2 +2 yz ) dx, satisfying

the conditions y (0 ) = 0 , y ( π / 2) = 1, z (0 ) = 0 , z (π / 2) = −1 . (Lucknow 2006)

Solution: The given functional is

π /2
I [ y ( x), z ( x)] = ∫0 ( y ′2 + z ′2 + 2 yz ) dx. …(1)

∴ F ( x, y, z , y ′ , z ′ ) = y ′2 + z ′2 + 2 yz . …(2)
The system of Euler’s equations is given by
d d
Fy − F y′ = 0 and Fz − Fz ′ = 0
dx dx
d d
or 2z − (2 y ′ ) = 0 and 2y− (2 z ′ ) = 0
dx dx
or z − y′ ′ = 0 and y − z′ ′ = 0
d2 y d2 z
or =z and = y. …(3)
dx2 dx2
Differentiating the first equation of (3), twice w.r.t. x, we get
d4 y d2 z d4 y
= or = y [using (3)]
dx4 dx2 dx2

( D4 − 1) y = 0 ,  D ≡ 
d
or …(4)
 dx 
The auxiliary equation is
D4 − 1 = 0 or m4 − 1 = 0
or (m2 + 1) (m2 − 1) = 0 .
∴ m = 1, − 1, i, − i.
Thus the general solution of (4) is
y = c1e x + c2 e − x + c3 cos x + c4 sin x …(5)
where c1, c2 , c3 and c4 are arbitrary constants.
Differentiating (5), we get
dy
= c1e x − c2 e − x − c3 sin x + c4 cos x.
dx
d2 y
and 2
= c1e x + c2 e − x − c3 cos x − c4 sin x. …(6)
dx
Thus from (3), and (6), we get
z = c1e x + c2 e − x − c3 cos x − c4 sin x. …(7)
 C-281

Initially at x = 0, we have y = 0 and z = 0.

Therefore from (5) and (7), we have
c1 + c2 + c3 = 0 and c1 + c2 − c3 = 0
⇒ c3 = 0 …(8)
and c1 + c2 = 0 . …(9)
π
Also when x = , we have y = 1 and z = − 1.
2
Therefore from (5) and (7), we have
c1e π / 2 + c2 e − π /2 + c4 = 1
and c1 e π /2 + c2 e − π /2 − c4 = − 1
⇒ c4 = 1 …(10)
π /2 − π /2
and c1e + c2 e = 0. …(11)
Solving (9) and (11), we get c1 = c2 = 0 .
Putting the values of c1, c2 , c3 and c4 in (5) and (7), the required curves are
y = sin x and z = − sin z .

19 Functionals Dependent on Higher Order Derivatives

(Lucknow 2006, 09, 10; Kanpur 09)
Consider the functional
b
I [ y ( x)] = ∫a F [( x), y ( x), y ′ ( x), … , y(n) ( x)] dx …(1)

such that F is assumed to be differentiable (n + 2) times w.r.t. all its arguments and
boundary conditions are assumed to be given the values not only of the function but
also its derivatives upto the (n − 1) order.
y (a) = y1, y ′ (a) = y ′1 , ………, y(n − 1)(a) = y1(n − 1) …(2)
y (b) = y2 , y ′ (b) = y ′2 , ………, y(n − 1)(b) = y2(n − 1) …(3)
It is also assumed that an extremum is attained on the curve y = y( x), which is 2n times
differentiable and any admissible comparison curve y = y ( x) is also 2n times
differentiable.
Consider a one parameter family of functions
y ( x, α) = y ( x) + α { y ( x) − y ( x)}
or y ( x, α) = y ( x) + α δy
We have y ( x, α) = y ( x), for α = 0
and y ( x, α) = y ( x), for α = 1.
If the value of the functional I is considered only on the curves of the family
y = y ( x, α), then the functional given by (1) reduces to a function of parameter α,
which is extremized for α = 0.
 C-282

 dI [ y ( x, α)]
∴   = 0. …(4)
dα α =0

The derivative given in equation (4) is called the variation of the functional I and is
denoted by δI. Thus we have
 d b 
δI = 
 dα
∫a F ( x, y ( x, α), y ′ ( x, α), … , y(n) ( x, α)) dα
α = 0
b
or δI = ∫a ( F y δy + F y′ δy ′ + … + F ( n) δy(n)) dx
y
…(5)

Integrating by parts the terms on the R.H.S. of (5), we get

b b  d 
∫ a F y′ δy ′ dx = [ F y′ δy]a − ∫ a  dx F y′ δy dx
b

b
 δy + b  d F  δy dx
2
b  d
∫ a y′′ ∫
b
F δ y ′ ′ dx = [ F y′′ δ y ′ ] −  F 
 dx y′′  
a
 a a  dx2 y′′ 

… … … …
… … … …
b
b  d  
∫a F δy(n) dx = [ F δy(n − 1)]ba −  F ( n)  δy(n − 2) + …
y( n) y( n)
 dy y   a
b  dn 
+(− 1)n ∫  F ( n)  δy dx.
a  dx n y 

Using the boundary conditions, we get for x = a and x = b, the variations

δy = δy ′ = δy ′ ′ = ……= δy(n − 1) = 0.
Thus from (5), we get
b d d2 dn
δI = ∫a (F y −
dx
F y′ + 2 F y′′ + … + (− 1)n n F ( n) ) δydx.0
dx dx y

On the extremizing curve we have δI = 0, therefore for an arbitrary choice of the

function δy, we have
b  d d2 dn 
∫a Fy −
 dx
F y′ + 2 F y′′ + … + (− 1)n n F y(n) δy dx = 0.
dx dx 
The first factor under the integral sign is a continuous function of x on the curve
y = y ( x), therefore by virtue of the fundamental lemma of the calculus of variations,
the first factor is identically equal to zero.
d d2 dn
∴ Fy − F y′ + 2 F y′′ − … + (− 1)n n F ( n) = 0 , …(6)
dx dx dx y

Equation (6) is a differential equation of order 2n and is called Euler’s-Poisson equation.

The integral curves of this differential equation are called extremals of the variation
problems under consideration. The solution of (6) involves two arbitrary constants
which can be determined by using the given 2n boundary conditions (2) and (3).
 C-283

Remark: When the functional I is of the form

b
I [ y ( x), z ( x)] = ∫a F ( x, y, y ′ , y(n), z , z ′ , … , z (n)) dx …(7)

then it can be easily proved by varying only y( x) and assuming z ( x) to be fixed that the
extremizing functions y ( x) and z ( x) must satisfy the Euler-Poisson equation.
d d2 dn
Fy − F y ′ + 2 F y ′′ − … + (− 1)n 2 F ( n) = 0 …(8)
dx dx dx y

By varying z ( x) and assuming y ( x) to be fixed, the very same functions satisfying

the Euler-Poisson equation
d d2 dn
Fz − Fz ′ + 2 Fz ′′ + … + (− 1)n n F ( n) = 0 . …(9)
dx dx dx z
The system of equations (8) and (9) are satisfied by the functions y ( x) and z ( x).
In general the basic necessary condition for an extremum of a functional dependent on
any number of functions
b
I [ y1, y2 , …, ym] = ∫a F ( x, y1, y ′1 , … , y1(n1), y2 , y ′2 , … , y2(n), … ,

ym, y ′ m , .... , ym(nm)) dx

is obtained by varying some one function yi( x) and holding the others fixed, in the form
d d ni
F yi − F y′ + … + (− 1)ni n F ( ni ) = 0, i = 1, 2, …, m.
dx i dx i yi

Example 21: Find the extremal for the functional

π /2
I [ y ( x)] = ∫0 [ y ′ ′2 − y2 + x2 ] dx.

Given y (0 ) = 1 , y ′ (0 ) = 0 , y (π / 2) = 0 ,, y ′ (π / 2) = − 1 . (Lucknow 2006; Kanpur 13)

Solution: We have
π /2
I [ y ( x)] = ∫0 [ y ′ ′2 − y2 + x2 ] dx. …(1)

π /2
Comparing (1) with ∫0 F ( x, y, y ′ , y ′ ′ ) dx, we get

F ( x, y, y ′ , y ′ ′ ) = y ′ ′2 − y2 + x2 . …(2)
The Euler-Poisson equation
d d2
Fy − F y ′ + 2 F y ′′ = 0. …(3)
dx dx
From (1), we have
F y = − 2 y, F y ′ = 0 , F y′′ = 2 y ′ ′.
 C-284

Putting these values in (2), we get

d2 d4 y
−2y + 2
(2 y ′ ′ ) = 0 or − 2 y + 2 =0
dx dx4

( D4 − 1) y = 0 , D ≡ d 
or   …(4)
 dx 
The auxiliary equation is m4 − 1 = 0 or (m2 + 1) (m2 − 1) = 0 .
∴ m = 1, − 1, i, − i.
Thus the general solution of (4) is
y = c1e x + c2 e − x + c3 cos x + c4 sin x …(5)
where c1, c2 , c3 , c4 are arbitrary constants.
Differentiating (5), we have
y ′ = c1e x − c2 e − x − c3 sin x + c4 cos x. …(6)
Using y (0 ) = 1 and y (π / 2) = 0 , from (5) we have
c1 + c2 + c3 = 0 …(7)
π /2 − π /2
and c1 e + c2 e + c4 = 0 . …(8)
Using y ′ (0 ) and y′ (π / 2) = −1, from (6) we have
c1 − c2 + c4 = 0 …(9)
c1e π /2 − c2 e − π /2 − c3 = − 1 . …(10)
Adding (7) and (10), we get
c1 (1 + e π / 2 ) + c2 (1 − e − π /2 ) = 0 . …(11)
Subtracting (8) from (9), we get
c1 (1 − e π /2 ) − c2 (1 + e − π /2 ) = 0 . …(12)
The determinant of coefficients of (11) and (12) is
(1 + e π / 2 ) (1 − e − π /2 )
D=
(1 − e π /2 ) − (1 + e − π /2 )

= − (1 + e π /2 ) (1 + e − π /2 ) − (1 − e π /2 ) (1 − e − π /2 )
= − 4 ≠ 0.
Since D ≠ 0 the system of homogeneous equations (11) and (12) has only the trivial
solution i. e., c1 = 0 and c2 = 0 . Putting these values in (7) and (9), we get
c3 = 1 and c4 = 0 .
Hence, from (5), the required extremum is attained only on the curve y = cos x.

Example 22: Determine the extremal for the functional

I [ y ( x)] = ∫  µ y ′ ′2 + ρy dx
l 1
−l  2 
satisfying the boundary conditions
y (− l ) = 0 , y ′ (− l) = 0 , y ( l ) = 0 , y ′ ( l ) = 0. (Lucknow 2007, 08, 11)
 C-285

Solution: We have
l  1 µy ′ ′2 + ρy dx.
I [ y ( x)] = ∫ −l 
2


…(1)

l
Comparing (1) with ∫ −l F ( x, y, y ′ , y ′ ′ ) dx, we get

1
F ( x, y, y ′ , y ′ ′ ) = µy ′ ′ + ρy. …(2)
2
The Euler-Poisson equation
d d2
Fy − F y′ + 2 F y′′ = 0. …(3)
dx dx
From (1), we have
F y = ρ, F y ′ = 0, F y ′′ = µy ′ ′.
Putting these values in (2), we get
d2 d4 y ρ
ρ+ ( µy ′ ′ ) = 0 or =−
dx2 dx4 µ

D4 y = −( ρ / µ ),  D ≡ 
d
or …(4)
 dx 
The auxiliary equation is m4 = 0 or m = 0 , 0 , 0 , 0 .
Here C.F. = c1 x3 + c2 x2 + c3 x + c4 , …(5)
where c1, c2 , c3 , c4 are arbitrary constants.
1  ρ ρ 1 ρ 1
And P.I. =
D
−
4  µ 
=−
µ D3 ∫ dx = −
µ D3
x

ρ 1 ρ 1  x2 
=−
µ D2 ∫ x dx = −  
µ D2  2 

ρ 1 x2 ρ 1 x3
=−
µ D ∫ 2
dx = −
µ D 6
ρ x3 ρx4
=−
µ ∫ 6
dx = −
24 µ
.

Thus the general solution of (1) is

 ρ  4
y = C.F. + P.I. = −   x + c1 x3 + c2 x2 + c3 x + c4 . …(6)
 24 µ 
Differentiating (6), we get
dy  ρ
= −   x3 + 3 c1 x2 + 2 c2 x + c3 . …(7)
dx  6µ 
Using y (− l ) = 0 and y ( l ) = 0, from (6), we have
 −ρ 4
  l − c1l 3 + c2 l 2 − c3 l + c4 = 0 …(8)
 24 µ 
 C-286

 −ρ 4
and   l + c1l 3 + c2 l 2 + c3 l + c4 = 0 . …(9)
 24 µ 
Using y ′ (− l ) = 0 and y ′ ( l ) = 0, from (7) we have
 ρ 3 2
  l + 3 c1l − 2 c2 l + c3 = 0 …(10)
 6µ 
 ρ
and −   l3 + 3 c1 l2 + 2 c2 l + c3 = 0 . …(11)
 6µ 
Subtracting (9) from (8), we get
2 c1l3 + 2 c3 l = 0 . …(12)
Adding (10) and (11), we get
6 c1l2 + 2 c3 = 0 . …(13)
The determinant of coefficients of (12) and (13) is
2 l3 2 l
D= = 4 l3 − 12 l3 = − 8 l3 ≠ 0 .
6 l2 2
Since D ≠ 0, the system of homogeneous equations (10) and (11) has only the trivial
solution i. e., c1 = 0 and c2 = 0 . Putting these values in (6) and (8), we get
ρl2 ρl4
c2 = and c4 = − ⋅
12µ 24 µ

Hence from (6), the required extremum is attained only on the curve
 ρ  4  ρl2  2  ρ  4
y=−  x +  x −  l
 24µ  12µ   24µ 
 ρ  4 2 2 4  ρ  2 22
or y = −  ( x − 2l x + l ) or y = −  (x − l ) .
 24µ   24µ 

20 Functionals Dependent on Functions of More Than One

Independent Variable (Euler Ostrogradsky Equation)
(Lucknow 2007)
Consider the functional
I [z ( x, y)] = ∫∫D F ( x, y, z , p, q) dx dy, …(1)

∂z ∂z
where p= and q = ⋅
∂x ∂y
The values of the function z ( x, y) are given on the boundary C of the domain D, i. e., a
spatial path C′ is given, through which all permissible surfaces have to pass. We assume
F to be thrice differentiable. Consider an admissible surface z = z ( x, y) close to
z = z ( x, y). Let these two surfaces be included in a one parameter family of surfaces
 C-287

z ( x, y, α) = z ( x, y) + α δz …(2)
where δz, the variation of the function z ( x, y) is δz = z ( x, y) − z ( x, y). For α = 0, we
get the surface z = z ( x, y) and for α = 1, we get z = z ( x, y).
The functional I reduces to a function of α, on the functions of the family z ( x, y, α),
which has to have an extremum for α = 0.
 d I [z ( x, y, α)]
Therefore,   = 0. …(3)
dα α =0

The variation δl of the function is the derivative of I [z ( x, y, α)] with respect to α, for
α = 0. Thus, we have
 ∂ 
δI = 
 ∂ α ∫∫D
F ( x, y, z ( x, y, α), p ( x, y, α), q ( x, y, α)) dx dy
α = 0

or I = ∫∫D ( Fz δz + Fp δp + Fq δq) dx dy, …(4)

where z ( x, y, α) = z ( x, y) + α δz
∂z ( x, y, α)
p ( x, y, α) = = p ( x, y) + α δp
∂x
∂z ( x, y, α)
and q ( x, y, α) = = q ( x, y) + α δq.
∂y
∂( Fq δz ) ∂Fp ∂( Fp δz ) ∂Fp δz
We have = δz + Fp δp or Fpδp = − …(5)
∂y ∂x ∂x ∂x
∂( Fp δz ) ∂Fq ∂( Fq δz ) ∂Fq
and = δz + Fq δp or Fqδq = − δz …(6)
∂y ∂y ∂y ∂y
From (5) and (6), we have
∂ ∂ 
∫∫D ( Fpδp + Fq δq) dx dy = ∫∫D  ( Fpδz ) +
 ∂ x ∂ y
( Fq δz ) dx dy

 ∂Fp ∂Fq 
− ∫∫D 
 ∂x
+  δz dx dy,
∂y 
…(7)

∂Fp
where is the total partial derivative with respect to x. While computing it, we
∂x
assume y to be fixed, whereas z , p and q are considered to be dependent on x. Therefore,
we get
∂Fp ∂z ∂p ∂q
= Fpx + Fpz + Fpp + Fpq ⋅ …(8)
∂x ∂x ∂x ∂x
Similarly, we have
∂Fq ∂z ∂p ∂q
= Fqy + Fqz + Fqp + Fqq ⋅ …(9)
∂y ∂x ∂y ∂y
Using Green’s Theorem, we have
∂ ∂ 
∫∫D  ∂x ( Fp δz ) + ∂y ( Fq δz ) dx dy = ∫C ( Fp dy − Fq dx) δz = 0. …(10)
 C-288

Since all permissible surfaces pass through one and the same spatial contour C′, the
variation δz = 0 on the contour C and so the integral in (10) is equal to zero.
Thus from (7) we get
∂ ∂ 
∫∫D ( Fpδp + Fq δq) dx dy = − ∫∫
D
 Fp +
 ∂x
Fq  δz dx dy.
∂y 
…(11)

Using (11), (2) reduces to

∂ ∂ 
δI = ∫∫D Fzδz dx dy − ∫∫D  Fp +
 ∂x
Fq  δz dx dy.
∂y 
…(12)

Thus the necessary condition for an extremum of the functional (1), δI = 0, takes the
form
 ∂ ∂ 
∫∫D  Fz − ∂x Fp − ∂y Fq  δz dx dy = 0. …(13)

The variational δz being arbitrary implies that only restrictions of a general nature are
imposed on δz that have to do with continuity and differentiability, vanishing on the
contour C.
Here, as the first factor is continuous, it follows from the fundamental lemma of the
calculus of variation that on the extremizing surface z = z ( x, y), we must have
∂ ∂ ∂F ∂  ∂F  ∂  ∂F 
Fz − Fp − Fq = 0 or −   −   = 0, …(14)
∂x ∂y ∂z ∂x  ∂p  ∂y  ∂q 
which is the Euler’s equation for an extremal of functional (1).
We know that a stationary function if it exists is an extremal satisfying the given
boundary conditions. Hence, the required extremizing function z ( x, y) is the solution
of the second order partial differential equation (14), also called Euler-Ostrogradsky
equation.
Remark 1: If the functional is of the form
I [z ( x1, x2 , ..., xn)] = ∫ ∫ … ∫D F ( x1, x2 , … xn, z , p1, p2 , … , pn ) dx1 dx2 ... dxn

∂z
where pi = , then from the basic necessary condition for an extremal δI = 0, the
∂x i
function z = z ( x1, x2 , … , xn) extremizing the functional I must satisfy
Euler-Ostrogradsky equation
n ∂
Fz − Σ Fpi = 0.
i = 1 ∂x i

Remark 2: If the functional I is such that its integrand depends on derivatives of higher
order, then by applying the transformations used in deriving the Euler-Ostrogradsky
equation several times, we find that the necessary condition for an extremum is that the
extremizing function must satisfy an equation similar to the Euler-Poisson equation.
Consider the functional
I [z ( x, y)] = ∫∫D F ( x, y, z , p, q, r, s, t) dx dy …(1)
 C-289

∂ ∂ ∂2 ∂2 ∂2
∴ Fz − Fp − Fq + 2 Fr + Fs + 2 Ft = 0, …(2)
∂x ∂y ∂x ∂x ∂y ∂y
∂z ∂z ∂2 z ∂2 z ∂2 z
where p= ,q = , r = 2 ,s = ,t= 2 .
∂x ∂y ∂x ∂x ∂y ∂y
Here the function extremizing the functional I must satisfy the fourth order partial
differential equation (2).

Example 23: Determine the Euler-Ostrogradsky equation for

 ∂z 2  ∂z  2 
  
I [z ( x, y)] = ∫∫    +   + 2 z f ( x, y) dx dy,
D  ∂x   ∂y 
 
given that the values of z are prescribed on the boundary C of the domain D.
Solution: The given functional is
I [z ( x, y)] = ∫∫D { p2 + q2 + 2 zf ( x, y)} dx dy, …(1)

∂z ∂z
where = p and = q.
∂x ∂y
Comparing with ∫∫D F ( x, y, z , p, q) dx dy, we get

F ( x, y, z , p, q) = p2 + q2 + 2 z f ( x, y). …(2)
The Euler-Ostrogradsky equation for the extremal of (1) is given by
∂F ∂  ∂F  ∂  ∂F 
−   −   = 0. …(3)
∂z ∂x  ∂p  ∂y  ∂q 
From (2), we have
∂F ∂F  ∂z  ∂F  ∂z 
= 2 f ( x, y), = 2 p = 2  , = 2q = 2   .
∂z ∂p  ∂x  ∂q  ∂y 
Using these, the equation (3) can be rewritten as
∂  ∂z  ∂  ∂z 
2 f ( x, y) − 2  − 2  = 0
∂x  ∂x  ∂y  ∂y 
∂2 z ∂2 z
or + = f ( x, y). …(4)
∂x2 ∂y2
Equation (4) is the Poisson’s equation and we are to find its solution continuous in D
that takes on prescribed values on the boundary C of the domain D.

Example 24: Determine the Euler-Ostrogradsky equation for

 ∂u 2  ∂u  2 2
   ∂u  
I [u ( x, y, z )] = ∫∫∫    +   +    dx dy dz
D  ∂x   ∂y   ∂z  
 
given that the values of u are prescribed on the boundary C of the domain D.
 C-290

Solution: The given functional is

2 2 2
I [u ( x, y, z )] = ∫∫∫D ( p1 + p2 + p3 ) dx dy dz …(1)

∂u ∂u ∂u
where p1 = , p2 = and p3 = .
∂x ∂y ∂z
Comparing with ∫∫∫D F ( x, y, z , u, p1, p2 , p3 ) dx dy dz , we get
2 2 2
F ( x, y, z , u, p1, p2 , p3 ) = p1 + p2 + p3 . …(2)

The Euler-Ostrogradsky equation (refer remark 1 of article 22) for extremal of (1) is
∂F ∂  ∂F  ∂  ∂F  ∂  ∂F 
− − − = 0.
∂u ∂x  ∂p1  ∂y  ∂p2  ∂z  ∂p3 

From (2), we have

∂ ∂ ∂
0− (2 p1) − (2 p2 ) − (2 p3 ) = 0
∂x ∂y ∂z
∂  ∂u  ∂  ∂u  ∂  ∂u  ∂2 u ∂2 u ∂2 u
or   +   +   =0 or + + = 0.
∂x  ∂x  ∂y  ∂y  ∂z  ∂z  ∂x2 ∂y2 ∂z 2

Example 25: Find the surface of a minimum area, stretched over a given closed space curve C,
enclosing the domain D in the xy-plane.
Solution: In this problem we are to find the extremal of the functional
1 /2
 2
 ∂z  
2
  ∂z 
S [z ( x, y)] = ∫∫ 1 +   +    dx dy
  ∂x   ∂y  
D

or S [z ( x, y)] = ∫∫D (1 + p2 + q2 )1 /2 dx dy …(1)

∂z ∂z
where p= and q = .
∂x ∂y
Comparing (1) with ∫∫D F ( x, y, z , p, q) dx dy, we get

F ( x, y, z , p, q) = (1 + p2 + q2 )1 /2 . …(2)
The Euler-Ostrogradsky equation for the extremal of (1) is given by

∂F ∂  ∂F  ∂  ∂F 
−   −   = 0. …(3)
∂z ∂x  ∂p  ∂y  ∂q 
From (2), we have

∂F ∂F ∂F
= 0, = p (1 + p2 + q2 )−1 /2 , = q (1 + p2 + q2 )−1 /2 .
∂z ∂p ∂q
Using these, the equation (3) can be rewritten as

∂ p ∂ q
− 2 2 1 / 2
− =0
∂x (1 + p + q ) ∂y (1 + p + q2 )1 /2
2
 C-291

∂ zx ∂ zy
or 2 2 1 / 2
+ = 0,
∂x (1 + z x + z y) ∂y (1 + z x + z 2y)1 /2
2

∂z ∂z
where z x = and z y =
∂x ∂y

z xx (1 + z x2 + z y2 )−1 /2 −   z x (1 + z x2 + z y2 )−3 /2 . 2 (z x z xx + z y z yx )
1
or
 2

+ z yy (1 + z x2 + z y2 )−1 /2 −   z y (1 + z x2 + z y2 )−3 /2 ⋅ 2 (z x z xy + z yz yy) = 0

1
 2
 ∂2 z ∂2 z ∂2 z ∂2 z 
where z xx = 2 , z yy = 2 , z xy = and z yx = 
 ∂x ∂y ∂x ∂y ∂y ∂x 

or z xx (1 + z x2 + z y2 ) − z x (z x z xx + z y z xy) + z yy (1 + z x2 + z y2 )

− z y (z x z xy + z y z yy) = 0
[∵ z xy = z yx ]
2 2
or z xx (1 + z y ) + z yy (1 + z x ) − 2 z x z y z xy = 0

∂2 z 
2
  ∂z   2 ∂z ∂z ∂ z
2 2 2
 ∂z   ∂ z
or 1 +   + 2 1 +
   − = 0. …(4)
∂x2   ∂y   ∂y  ∂x   ∂x ∂y ∂x dy
  
The solution of (4) gives the required minimal surface.

21 Variational Problems with Natural Boundary

Conditions
If one or both the boundary conditions are not prescribed in a variational problem,
then the added condition is called natural boundary condition.
Case I: For extremizing the functional
x2
I [ y( x)] = ∫x1 F ( x, y, y ′ )dx …(1)

Euler’s equation is
∂F d  ∂F 
−   =0
∂y dx  ∂y ′ 
and the natural boundary conditions are
 ∂F 
  = 0, if y( x1) is not prescribed
 ∂y ′  x1
 ∂F 
and   = 0, if y( x2 ) is not prescribed.
 ∂y ′  x2
 C-292

Case II: For extremizing functional

x2
I [ y( x), z ( x)] = ∫x1 F ( x, y, y ′ , z ′ )dx

Euler’s equations are

∂F d  ∂F  ∂F d  ∂F 
−   = 0, −   =0
∂y dx  ∂y ′  ∂z dx  ∂z ′ 

and the natural boundary conditions are

∂F
= 0, if y( x) is not prescribed at an end point
∂y ′
∂F
and = 0, if z ( x) is not prescribed at end point.
∂z ′
Case III: For extremizing the functional
x2
I [ y( x)] = ∫x1 F ( x, y, y ′ , y ′ ′ )dx

Euler’s equation is
∂F d  ∂F  d2  ∂F 
−   + 2  =0
∂y dx  ∂y ′  dx  ∂y ′ ′ 
and the natural boundary conditions are
∂F d  ∂F 
−   = 0, if y( x) is not prescribed at an end point
∂y dx  ∂y ′ 
∂F
and = 0, if y ′ ( x) is not prescribed at an end point.
∂y ′ ′

1
Example 26: Find the extremals of the functional I [ y ( x)] = ∫0 ( y ′2 − y2 ) dx subject to the

boundary conditions
(i) y (0 ) = 0 , y (1) = 1 (ii) y (0 ) = 0
(iii) y (1) = 1 (iv) neither y (0 ) nor y (1) is specified.
2 2
Solution: We have F = y ′ − y . …(1)
From Euler’s equation, we have
∂F d  ∂F  d
−   =0 or −2 y − (2 y ′ ) = 0 , [From equation (1)]
∂y dx  ∂y ′  dx
d
or y′′ + y =0 or ( D2 + 1) y = 0 , where =1
dx
or D2 = − 1 or D=±i
or y = c1 cos x + c2 sin x. …(2)
 C-293

(i) Given y (0 ) = 0 , y (1) = 1.

∴ from (2), we have
1
c1 = 0 , c2 sin 1 = 1 or c1 = 0 , c2 = ⋅
sin 1
Hence from (2), we get
sin x
y= , which is the required extremal.
sin1
(ii) Given y(0 ) = 0 and y( x2 ) is not prescribed
∴ from (2), we have
c1 = 0 .
 ∂F 
We take   =0 or 2 y ′ ( x2 ) = 0 [From (2)]
 ∂y ′  x2

or y ′ ( x2 ) = 0 or y′ (1) = 0 .
From (2), we get
y ′ = − c1 sin x + c2 cos x …(3)
and y′ (1) = 0.
∴ (− c1 sin x + c2 cos x) x = 1 = 0 [∵ x2 = 1]
or c2 cos 1 = 0 [∵ c1 = 0 ]
or c2 = 0 . [∵ cos 1 ≠ 0 ]
Putting these values in equation (2), we get
y ( x) = 0 ,
which is the required extremal.
(iii) Given y (1) = 1 and y ( x1) is not prescribed
∴ From (2), we have
c1 cos 1 + c2 sin1 = 1 …(4)
 ∂F 
and we take   =0 or y ′ ( x1) = 0
 ∂y ′  x1

or y′ (0 ) = 0 [∵ x1 = 0 ]
or (− c1 sin x + c2 cos x) x =0 = 0 [From (3)]
or c2 = 0 .
1
From equation (4), we have c1 = ⋅
cos 1
Putting these values in equation (2), we get
cos x
y( x) = , which is the required extremal.
cos 1
(iv) Here neither y( x1) nor y ′ ( x2 ) is specified.

 ∂F   ∂F 
We take   = 0 and   = 0
 ∂y ′  x1  ∂y ′  x2
 C-294

or y ′ ( x1) = 0 and y ′ ( x2 ) = 0
or y′ (0 ) = 0 and y′ (1) = 0
or c2 = 0 and − c1 sin 1 = 0
or c2 = 0 and c1 = 0 . [∵ sin 1 ≠ 0 ]
Putting these values in equation (2), we get
y( x) = 0, which is the required extremal.

22 Invariance of Euler’s Equation Under Co-ordinate

Transformation
(Gorakhpur 2006, 09, 13; Lucknow 10)
Let a given functional
b
I [ y ( x)] = ∫a F ( x, y, y ′ ) dx …(1)

be transformed by replacing the independent variable or by a simultaneous

replacement of the desired function and the independent variable. Then the extremals
of the functional (1) can be found from Euler’s equation for the transformed integrand.
This principle is known as the “principle of invariance of Euler’s equation”.
Let x and y be replaced by new variables u and v using the transformation
x = x (u, v), y = y (u, v) …(2)
Here the Jacobian of the transformation is
∂( x, y)
J= ≠0
∂(u, v)
xu xv
= ≠0 …(3)
yu yv
∂x ∂x ∂y ∂y
where xu = , xv = , yu = and yv = ⋅
∂u ∂v ∂u ∂v
dv
Also let v ′ = , then we have
du
 ∂x   ∂x 
dx =   du +   dv = xu du + xv dv
 ∂u   ∂v 
 ∂y   ∂y 
and dy =   du +   dv = yu du + yv dv
 ∂u   ∂v 
 dv 
i. e., dx =  xu + xv    du = ( xu + xv v ′ ) du …(4)
  du 
 dv 
and dy =  yu + yv    du = ( yu + yv v ′ ) du …(5)
  du 
dy ( yu + yvv ′ )
∴ y′ = = …(6)
dx xu + xvv ′
 C-295

From (2), (4), (5) and (6), we get

b u2 y + yv v ′
∫a F ( x, y, y ′ ) dx = ∫u1 F { x (u, v) , y (u, v), u
xu + xv v ′
} ( xu + xv v ′ ) du

…(7)
assuming that u = u1 when x = a and u = u2 when x = b.
Therefore from (7), we get
b u2
∫a F ( x, y, y ′ ) dx = ∫u1 G (u, v, v ′ ) du. …(8)

Thus the functional (1) takes the form

u2
I [v (u)] = ∫u1 G (u, v, v ′ ) du. …(9)

Hence, the extremals of the functional (1) are determined from Euler’s equation
for (9), i. e.,
∂G d  ∂G 
−   = 0. …(10)
∂v du  ∂v ′ 
Working Rule: For solving the problems we make the change of variables directly in
the integral representing the given functional.
Now solve the Euler’s equation for the new functional containing new variables.
Finally the new variables in the extremals obtained are replaced by the original
variables to obtain the required extremals.

θ2
Example 27: Determine the extremals for the functional ∫ (r2 + r ′2 )1 /2 dθ, given r = r(θ)
θ1
dr
and r ′ = ⋅
dθ
θ2
Solution: Let I [r (θ)] = ∫ θ1 (r2 + r ′2 )1 /2 dθ. …(1)

We have x = r cos θ and y = r sin θ. …(2)

dx dy
∴ = − r sin θ + r ′ cos θ, = r cos θ + r ′ sin θ
dθ dθ
2 2
 dx  +  dy  = r2 + r ′2 .
and     …(3)
 dθ   dθ 
When θ = θ1, we have x = x1 and when θ = θ2 , we have x = x2 . …(4)
  dx 2 2 1 /2
θ2 θ2  dy  
∴ ∫ θ1 (r2 + r ′2 )1 /2 dθ = ∫ θ1    +  
 dθ dθ


dθ, [Using (3)]

x2 x2
= ∫ x1 {(dx)2 + (dy)2 }1 /2 = ∫ x1 (1 + y ′2 )1 /2 dx. [Using (4)]
 C-296

If the transformed functional be denoted by T , then

x2
T [ y ( x)] = ∫ x1 (1 + y ′2 )1 /2 dx. …(5)

x2
Comparing the transformed functional with ∫ x1 F ( x, y, y ′ ) dx, we have

F ( x, y, y ′ ) = (1 + y ′2 )1 /2 .
∂F d  ∂F 
Thus the Euler’s equation −   = 0, becomes
∂y dx  ∂y ′ 

d  y′ 
0−  2 1 /2 
=0 or y ′ /(1 + y ′2 )1 /2 = c1
dx  (1 + y ′ ) 
or y ′ = c1 (1 + y ′2 )1 /2 or y ′2 = c12 (1 + y ′2 )
c12 c1
or y ′2 = or y′= = c2 (say)
1− c12 (1 − c12 )1 /2
dy
or = c2 or y = c2 x + c3 . …(6)
dx
From (2) and (6), we get
r sin θ = c1 r cos θ + c2 ,
which are the required extremals for the given functional (1).

Example 28: Determine the extremals for the functional

ln2
I [ y ( x)] = ∫0 (e − x y ′2 − e x y2 ) dx, where ln2 = loge2.
(Lucknow 2009)
π /2 −x
∫0
2 x 2
Solution: We have I [ y ( x)] = (e y ′ − e y ) dx …(1)

Let x and y be replaced by new variables u and v using the transformation

x = ln u = log e u and y = v. …(2)
∂x 1 ∂x ∂y ∂y
Here xu = = , xv = = 0, yu = = 0 and yv = = 1.
∂u u ∂v ∂u ∂v
y + yvv ′
∴ y′= u = uv ′ …(3)
xu + xvv ′
1
and dx = ( xu + xvv ′ ) du = du
u
dv
where v′ = ⋅ …(4)
du
We have x = ln 2 ⇒ log e u = log e 2 ⇒ u = 2
and x = 0 ⇒ log e u = 0 ⇒ u = e0 = 1 .
2 − log e u du
Thus I [ y ( x)] = ∫1 { e (uv ′ )2 − e log e u v2 }
u
2
∫1 (v ′
2
= − v 2) du = T [v (u)] …(5)

where T [v (u)] denotes the transformed functional .

 C-297

2
Comparing T [v (u)] = ∫1 G (u, v, v′ ) du with (5), we get

G (u, v, v ′ ) = v ′2 − v2 . …(6)
The Euler’s equation for (6) is
∂G d  ∂G  d
−   =0 or −v− (2 v ′ ) = 0
∂v du  ∂v ′  du
or v′ ′ + v = 0
d
or ( D2 + 1) v = 0 , D ≡ ⋅ …(7)
dv
The general solution of (7) is
v = c1 cos u + c2 sin u, where c1 and c2 are arbitrary constants
x x
or y = c1 cos e + c2 sin e , …(8)
which are the required extremals of (1). [∵ From (1), u = e x ]

Comprehensive Exercise 2

1
1. (i) Find the extremal of the functional ∫0 y ′2 dx, y (0 ) = 1, y (1) = 6
1
subject to the condition ∫0 ydx = 3.
x2
(ii) Find the extremal of the functional ∫x1 y ′2 dx

x2
subject to the condition ∫x1 dx = c , a constant.

1
2. (i) Find a function y( x), for which ∫0 ( x2 + y ′2 )dx is stationary given
1
that ∫0 y2 dx = 2; y (0 ) = 0 , y (1) = 0 .

(ii) Show that the sphere is the solid figure of revolution which, for a given
surface, has maximum volume.
3. (i) Determine the extremals of the functional
π /2
I [ y ( x), z ( x)] = ∫ ( y ′2 + z ′2 + 2 yz ) dx
0
that satisfy the boundary conditions y (0 ) = 0 , y (π / 2) = −1, z (0 ) = 0 and
z (π / 2) = 1.
(ii) Obtain the extremals of the functional
1
I [ y ( x), z ( x)] = ∫0 ( y ′2 + z ′2 ) dx

that satisfy the boundary conditions y (0 ) = 0 , y (1) = 1, z (0 ) = 0 and

z (1) = 2.
 C-298

4. (i) Determine the extremals of the functional

π
I [ y ( x), z ( x)] = ∫0 (2 yz − 2 y2 + y ′2 − z ′2 ) dx

that satisfy the boundary conditions y (0 ) = 0 , y (π) = 1, z (0 ) = 0 and

z (π) = 1.
1
(ii) Determine the extremals of the functional ∫0 (1 + y ′2 + z ′2 )1 /2 dx

that satisfy the boundary conditions y (0 ) = 0 , y (1) = 2, z (0 ) = 0 , and

z (1) = 4.
1  dx 2  dy  
2

5. (i) Show that the functional ∫ 2 x +   +    dt
0   dt   dt  
 
such that x (0 ) = 1, y (0 ) = 1, x (1) = 1 ⋅ 5 and y (1) = 1 is stationary for
(2 + t2 )
x= and y = 1.
2 (Kanpur 2013)
1
(ii) Find the extremal of the functional I [ y ( x)] = ∫0 (1 + y ′ ′2 ) dx.

Given y (0 ) = 0 , y ′ (0 ) = 1, y (1) = 1 and y ′ (1) = 1.

6. (i) Obtain the extremal of the functional
0
I [ y ( x)] = ∫ −1 (480 y − y ′ ′ ′2 ) dx, y = 0, y′ (0 ) = 0, y′ ′ (0 ) = 0.

(ii) Obtain the extremal of the functional

π /4
I [ y ( x)] = ∫0 ( y ′ ′2 − y2 + x2 ) dx.
1 1
Given y (0 ) = 0 , y ′ (0 ) = 1, y (π / 4) = , y ′ (π / 4) = ⋅
2 2
7. Find the Euler-Ostrogradsky equation for
 ∂z 2  ∂z  2 
  
∫∫D   ∂x  +  ∂y   dx dy,
I [z ( x, y)] =
 
where the values of z are prescribed on the boundary C of the domain D.
(Lucknow 2009)
8. Obtain the fourth order partial differential equation satisfied by the function
extremizing the functional
 2  2  2  2  ∂2 z  
2
 ∂ z ∂ z
I [z ( x, y)] = ∫∫   2  +  2  + 2    dx dy,
D  ∂x 
  ∂y   ∂x ∂y  

where the values of z are prescribed on the boundary C of the domain D.
9. Prove that the Euler-Ostrogradsky equation of the functional
 ∂u 2  ∂u  2 2 
   ∂u 
I [u ( x, y, z )] = ∫∫∫   +   +   + 2 uf  dx dy dz
D   ∂x   ∂y   ∂z  
 
2 2 2
∂ u ∂ u ∂ u
is the Poisson equation 2 + 2 + 2 = f ( x, y, z ).
∂x ∂y ∂z (Lucknow 2008)
 C-299

A nswers 2
1. (i) y = 3 x2 + 2 x + 1
3 x2
(ii) y= {2 c − c1 ( x22 − x12 ) − 2 c2 ( x2 − x1)} + c1 x + c2
2 ( x23 − x13 )
2. (i) when λ = 0, no solution; when λ = µ 2 , y = 0
and when λ = − µ 2 , y = ± 2 sin mπx
3. (i) y = − sin x, z = sin x (ii) y = x, z = 2 x
x 2
4. (i) y = c3 sin x − cos x, z = c3 sin x + (sin x − x cos x)
π π
(ii) y = 2 x, z = 4 x
5. (ii) y=x
x6 c c c
6. (i) y=− + 1 x 5 + 2 x 4 + 3 x 3 (ii) y = sin x
3 120 24 6
∂2 z ∂2 y
7. + =0
∂x2 ∂ y2
∂4 z ∂4 z ∂4 z
8. + 2 + =0
∂x4 ∂x2 ∂y2 ∂y4

O bjective T ype Q uestions

Multiple Choice Questions

Indicate the correct answer for each question by writing the corresponding letter from (a),
(b), (c) and (d).
b
1. The Euler’s equation for a functional of the form ∫ F ( x, y) dx is
a
(a) F y = c1 (b) F y′ = c1
(c) F y − y ′ F y′ = c1 (d) none of these
1
2. The extremum of the functional I [ y ( x)] = ∫0 ( y ′2 + 12 xy) dx satisfying the

conditions y (0 ) = 0 and y (1) = 1 is attained on the curve

πx
(a) y = x3 (b) y = sin2  
2
1 πx πx
(c) y = ( x3 + sin ) (d) y = sin
2 2 2
 C-300

2 y ′2
3. The functional I [ y( x)] = ∫0 x
dx, y (0 ) = 0 , y (2) = 1 can be extremized on

the curve
(a) y = x3 − 6 x (b) y = x
3
(c) 2 y = x (d) y = x2 / 4

b  y3 
4. The extremal of the functional I [ y ( x)] = ∫a 
 y +  dx is
3 

(a) y = 1 (b) y = 0
(c) y = − 1 (d) has no extremal
5. The Euler’s equation for the extremals of the functional
x2
∫x1 ( y2 − yy ′ + y ′2 ) dx is

(a) y ′ − y = 0 (b) y ′ ′ − y = 0
(c) y ′ ′ − y ′ = 0 (d) None of these

A nswers
1. (d) 2. (a) 3. (d) 4. (d)
5. (b)