Calculus of Variation and Integral

Equations
DEMTH655
Edited by:
Dr. Kulwinder Singh
Calculus of variation and
Integral Equations
Edited By
Dr. Kulwinder Singh
 CONTENTS

Unit 1: Introduction to Integral Equations 1

Pratiksha, Lovely Professional University
Unit 2: Types of Kernels, Eigenvalues and Eigenfunctions 14
Pratiksha, Lovely Professional University
Unit 3: Differential Equation to Integral Equation 22
Pratiksha, Lovely Professional University
Unit 4: Fredholm Integral Equations of the Second Kind withSeparable Kernels 28
Pratiksha, Lovely Professional University
Unit 5: Fredholm Alternative Theorem and Resolvent Kernel 38
Pratiksha, Lovely Professional University
Unit 6: Method of Successive Substitution and Approximation 49
Pratiksha, Lovely Professional University
Unit 7: Iterated and Resolvent Kernels 58
Pratiksha, Lovely Professional University
Unit 8: Green’s Functions 68
Pratiksha, Lovely Professional University
Unit 9: Introduction to Calculus of Variation 81
Pratiksha, Lovely Professional University
Unit 10: Euler’s Equation 90
Pratiksha, Lovely Professional University
Unit 11: Problem of Brachistochrone and Minimum Surface ofRevolution 105
Pratiksha, Lovely Professional University
Unit 12: Variational Problems in Parametric Form 116
Pratiksha, Lovely Professional University
Unit 13: Geodesics 128
Pratiksha, Lovely Professional University
Unit 14: Conditional Extremum under Constraints 137
Pratiksha, Lovely Professional University
 Notes

Pratiksha, Lovely Professional University Unit 01: Introduction to Integral Equations

Unit 01: Introduction to Integral Equations

CONTENTS
Objectives
Introduction
1.1 Abel's Problem
1.2 Integral equation
1.3 Linear and non-linear integral equations
1.4 Volterra and Fredholm integral equations
1.5 Solution of an integral equation
1.6 Recap of definite &indefinite integrals
1.7 Integration by Substitution
1.8 Integration by using Partial Fractions
Summary
Key words
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to

 recognize the integral equations

 differentiate between Fredholm and Volterra equations
 check whether a function is a solution of the integral equation.

Introduction
In real world, the solution of physical problems can be found by using the theory of ordinary or
partial differential equations or by the theory of integral equations. The solution of integral
equations is much easier than the original boundary or initial value problems.
Several situations of science lead to integral equations e.g., neutron diffusion problem, radiation
transfer problem etc.For example, while searching for the representation formula for the solution of
linear differential equation in such a manner to include boundary conditions or initial conditions
explicitly, an integral equation is obtained.

Lovely Professional University 1

Notes

Calculus of Variation and Integral Equation

1.1 Abel's Problem

We propose to give an example of a situation which leads to an integral equation.Niels Henrick
Abel was a Norwegian mathematician who proved the impossibility of solving algebraically the
general equation of the fifth degree.
In Abel's third paper, Solutions of some problems by means of definite integrals he gave the first solution
of an integral equation.

The Abel Prize is intended to give the mathematicians their own equivalent of a Nobel Prize.Today,
every mathematics undergraduate encounter Abel's name in connection with commutative groups,
which are more commonly known as "abelian groups."
Consider the following problem of mechanics.
Consider a smooth curve in a vertical plane and suppose a material point starts from rest at any
point P under the influence of gravity along the curve. Let T be the time taken by the particle from
P to the lowest point O.Treat O as the origin of coordinate axes; y-axis being horizontal and x-axis
being vertically upwards.
Let the coordinates of points P and Q be (x, y) and , respectively.Let the arc OQ= s. Then, at
the particle at Q, we have

1
= ℎ
2
=2 ℎ
=± 2 ℎ
=± 2 ( − )
We consider
= − 2 ( − )

=−
2 ( − )

= … (1)
2 ( − )

If the shape of the curve is given, then s can be expressed in terms of , and = ( )

(1) => =
( )
… (2)
2 ( − )

Abel treated this problem in modified form by finding the curve for which the time of descent T is a
given function of , say ( ). Thus, we are led to a problem of finding the unknown function
from the equation,
( )
( )= … (3)
2 ( − )

2 Lovely Professional University

 Notes

Unit 01: Introduction to Integral Equations

which is called the Abel's integral equation

1.2 Integral equation

An integral equation is an equation in which an unknown function appears under one or more
integral signs.
e.g. For ≤ ≤ , ≤ ≤ , ∫ ( , ) ( ) = ( )

( )− ( , ) ( ) = ( )

And ( ) = ∫ ( , )[ ( )]
Where ( ) is an unknown function, ( ) is a known function ( , ) is a known functionand
, , are the constants.

The above mentioned functions may be complex valued functions of the real variables and .

1.3 Linear and non-linear integral equations

An integral equation is called linear if only linear operations are performed in it upon the unknown
function. An integral equation which is not linear is known as a non-linear integral equation.

General Criterion for a linear operator:

{ ( )+ ( )} = { ( )} + { ( )}
The most general type of the linear integral equation is of the form:

( ) ( )= ( )+ ( , ) ( )

Where is non zero real or complex parameter.

For the integral equation

( , ) ( ) = ( )

Let ( ) = ∫ ( , ) ( )
And { ( )+ ( )} = { ( )} + { ( )}

Remarks:
1. can be incorporated into the kernel ( , ).
2. If ( ) is non zero then the equation ( ) ( ) = ( ) + ∫ ( , ) ( ) is known as linear

integral equation of third kind.

3. When ( ) is zero, then ( ) ( ) = ( ) + ∫ ( , ) ( ) is known as linear integral

equation of the first kind:

( )+ ( , ) ( ) =0

Lovely Professional University 3

Notes

Calculus of Variation and Integral Equation

4. When ( ) is unity, then ( ) ( ) = ( ) + ∫ ( , ) ( ) is known as linear integral

equation of the second kind:

( )= ( )+ ( , ) ( )

1.4 Volterra and Fredholm integral equations

V. Volterra, theItalian Mathematician and E. Fredholm, the Swedish mathematician are credited for
their work on integral equations.

The Fredholm integral equation

A linear integral equation of the form

( ) ( )= ( )+ ( , ) ( )

where and are both constants, ( ), ( )a ( , ) are known functions,

( ) is unknown function, is a non-zero real or complex parameter,function ( , ) is known as
the kernel of the integral equation,
is called Fredholm integral equation of third kind.

Special Cases

1. Fredholm Integral equation of the first kind

( ( ) = 0)
2. Fredholm Integral equation of the second kind
( ( ) = 1)

3. Homogeneous Fredholm Integral equation of the second kind

( ( ) = 1, ( ) = 0)
The Volterra integral equation
A linear integral equation of the form

( ) ( )= ( )+ ( , ) ( )

where is a constant, ( ), ( )a ( , ) are known functions,

( ) is unknown function, is a non zero real or complex parameter,
function ( , ) is known as the kernel of the integral equation,
is called Volterra integral equation of third kind.
Special Cases

1. Volterra Integral equation of the first kind

( ( ) = 0)
2. Volterra Integral equation of the second kind
( ( ) = 1)
3. Homogeneous Volterra Integral equation of the second kind
( ( ) = 1, ( ) = 0)

4 Lovely Professional University

 Notes

Unit 01: Introduction to Integral Equations

1.5 Solution of an integral equation

A solution of the integral equation is a function ( ) which when substituted into the equation,
reduces it into an identity.

Let us check if the function ( ) = (1 + ) is a solution of the equation

1
( )= − ( )
1+ 1+
Consider the right-hand side of the integral equation and substitute the solution function and
simplify the integral. You will get ( ) which is same as the left-hand side. Thus, we got an
identity. Therefore, the given function is a solution of the integral equation.
In the same way it can be shown that ( ) = 2 is a solution of the integral equation
( )= + 3 ∫ ( , ) ( ) where ( , ) = sin cos when 0 ≤ ≤ and ( , ) = cos sin
when ≤ ≤ . Using trigonometric identities, the obtained integrals can be solved easily, to arrive
at the identity.

1.6 Recap of definite &indefinite integrals

There are two types of integral
(i) Indefinite, in which we consider integration as anti differentiation and aren't given the
limits of integration, i.e. x=a to x=b, so we just calculate a generic, all purpose solution,
and
(ii) Definite, in which we are told a and b and so we can calculate an explicit value for an
area.

Indefinite integrals
If the differential of x3 is 3x2, then

 3x .dx = x
2 3

But 3x2 is also the differential of x3- 1 and x3+ 8, etc. so that this reversal is not unique - we've 'lost'
the constant! So in general, 3x2 is the differential of (x3+ k) where k is any constant – this is known as

the 'constant of integration'.We write this as: 3x .dx =x + k

2 3

(If we’re given more information, we can work out the value for k).
Since integration is the reverse of differentiation, for any polynomial y(x) = xn, we can simply
reverse the differentiation procedure, so that the integral is given by

(except for n = -1)

x n1
x dx =
n
(n  1)
+k

Some examples are given below:

 x .dx = x .dx
½ =
( / )
/
=
√
+k

abx .dx =
3 
ab x3.dx= ab +k

x 2
 x13 .dx = x .dx
-3 =
2
= -
1
2x 2
+k

Lovely Professional University 5

Notes

Calculus of Variation and Integral Equation

 2m2.dm =
2m 3
3
+k

10 3
5  .d = +k
3
Definite Integrals
We now know how to integrate simple polynomials, but if we want to use this technique to
calculate areas, we need to know the limits of integration. If we specify the limits x = ax = b, we
call the integral a definite integral.
To solve a definite integral, we first integrate the function as before (i.e. find its indefinite integral),
then feed in the 2 values of the limits. Subtracting one from the other gives the area.
What is the area under the curve y(x) = 2x2 between x=1 and x=3?

Area =∫ 2 .

x 3
2x3 
=  3  k
  x 1

= (18 + k) - (2/3 + k)

= 18 -2/3

= 171/3 sq. units

Note: The k's cancel. So, when we evaluate a definite integral, we can ignore the constant of
integration.

What is the area under the curve y(x) = 2x3- 6x between x = -1 and x = 0?

A=∫ 2 −6 . y
4

3
area
0
 x4 2
 2  3x 
2
=
  1 1

0 x
= (0 - 0) - (½ - 3) -1

-2
= 2½ sq.units
-3

-4
-2 -1 0 1 2

2
What is the area under the curve y(x) = between x = 1 and x = ?
x2
[If you draw the graph, you’ll see that although x goes to infinity, the curve is getting closer and
closer to the y axis and so the area is getting smaller. So, in this case, it is possible to calculate a
finite area, even though we are integrating to infinity].

6 Lovely Professional University

 Notes

Unit 01: Introduction to Integral Equations

∞
A=∫ .
∞
=∫ 2 .

  2
=
1 =  x 
[-2x-1]
 1

= (0) - (-2)

= 2 sq.units.

Lovely Professional University 7

Notes

Calculus of Variation and Integral Equation

Just as in differentiation, a function can by integrated term-by-term, and we only need one constant
of integration.

1.7 Integration by Substitution

Suppose we needed to compute the following integral:

Our rules of integration are of no help here. We note that the integrand is of the form
where and

Since we are looking for an antiderivative of and we know that we can re-write our
integral as

In practice, we use the following substitution scheme to verify that we can integrate in this way:

1. Let
2. Differentiate both sides so
3. Change the original integral in to an integral in :
4. Integrate with respect to :

where and

5. Change the answer back to :

While this method of substitution is a very powerful method for solving a variety of problems, we
will find that we sometimes will need to modify the method slightly to address problems, as in the
following example.
Compute the following indefinite integral:

Solution:

We note that the derivative of is ; hence, the current problem is not of the form
But we notice that the derivative is off only by a constant of and we
know that constants are easy to deal with when differentiating and integrating. Hence

Let

8 Lovely Professional University

 Notes

Unit 01: Introduction to Integral Equations

Then

Then and we are ready to change the original integral from to an integral in and
integrate:

Changing back to , we have

We can also use this substitution method to evaluate definite integrals. If we attach limits of
integration to our first example, we could have a problem such as

The method still works. However, we have a choice to make once we are ready to use the
Fundamental Theorem to evaluate the integral.

Recall that we found that for the indefinite integral. At this

point, we could evaluate the integral by changing the answer back to or we could evaluate the
integral in But we need to be careful. Since the original limits of integration were in , we need to
change the limits of integration for the equivalent integral in Hence,

where

Integrating Products of Functions

We are not able to state a rule for integrating products of functions, but we can get
a relationship that is almost as effective. Recall how we differentiated a product of functions:

So by integrating both sides we get

or

1.8 Integration by using Partial Fractions

Decomposes a rational function into simpler rational functions that are easier to integrate.
Essentially undoes the process of finding a common denominator of fractions.
PartialFractionsProcess
1. Check to make sure the degree of the numerator is less than the degree of the
denominator. If not, need to divide by long division.
2. Factor the denominator into linear or quadratic factors of the form
Linear: ( + ) Quadratic: ( + + )

Lovely Professional University 9

Notes

Calculus of Variation and Integral Equation

3. For linear functions:

( )
= + + + ⋯+ + where
( ) ( ) ( ) ( ) ( )

, , ,…, are real numbers.

4. For Quadratic Factors:

( )
= + + +⋯+
( ) ( ) ( ) ( )

where , , ,…, and , , ,…, .

Integrating Functions with Linear Factors Using Partial Fractions

1. Substitute the roots of the distinct linear factors of the denominator into the basic equation (the
equation obtained after eliminating the fractions on both sides of the equation) and find the
resulting constants.
2. For repeated linear factors, use the coefficients found in step 1 and substitute other
convenient values of x to find the other coefficients.
3. Integrate each term.
Integrating terms using Partial Fractions with Irreducible Quadratic Terms + + (quadratic
terms that cannot be factored) in the Denominator.

1. Expand the basic equation and combine the like terms of x.

2. Equate the coefficients of like powers and solve the resulting system of equations.
3. Integrate.
Having recapitulated the basic integral calculus, we can now proceed to solve the integral
equations in the next units.

Summary
 An integral equation is an equation in which an unknown function appears under one or
more integral signs.
 An integral equation is called linear if only linear operations are performed in it upon the
unknown function. An integral equation which is not linear is known as a non-linear
integral equation.
 A linear integral equation of the form

( ) ( )= ( )+ ( , ) ( )

where and are both constants, ( ), ( )a ( , ) are known functions,

( ) is unknown function, is a non-zero real or complex parameter,function ( , ) is known as
the kernel of the integral equation,is called Fredholm integral equation of third kind.

 A linear integral equation of the form

( ) ( )= ( )+ ( , ) ( )

where is a constant, ( ), ( )a ( , ) are known functions,

10 Lovely Professional University

 Notes

Unit 01: Introduction to Integral Equations

( ) is unknown function, is a non zero real or complex parameter,

function ( , ) is known as the kernel of the integral equation,is called Volterra integral equation
of third kind.

 A solution of the integral equation is a function ( ) which when substituted into the
equation, reduces it into an identity.

Key words
Fredholm integral equation, Volterra integral equation

Self Assessment
1. Which of the following statement is correct?
A. A solution of the integral equation is a function ( ) which when substituted into the
equation, reduces it into an identity.
B. A solution of the integral equation is a function ( ) which when substituted into the
equation, reduces it into another equation of same type but reduced order.
C. A solution of the integral equation is a function ( ) which when substituted into the
equation, reduces it into a constant.
D. None of these
2. Which of the following is true for linear functions?
( )
A. = − + +⋯+ +
( ) ( ) ( ) ( ) ( )
( )
B. = + + + ⋯+ +
( ) ( ) ( ) ( ) ( )
( )
C. = + + +⋯+ +
( ) ( ) ( ) ( ) ( )
( )
D. = + + + ⋯+ +
( ) ( ) ( ) ( ) ( )

3. In ( ) ( ) = ( ) + ∫ ( , ) ( ) , K(x,t) is called

A. Factor
B. Kernel
C. Kernal
D. Spectrum

4. ( )+ ∫ ( , ) ( ) = 0 is

A. Fredholm equation of first kind

B. Volterra equation of third kind
C. Volterra equation of second kind
D. Volterra equation of first kind

5. ( )+ ∫ ( , ) ( ) = 0 is

A. Fredholm equation of first kind

B. Volterra equation of third kind
C. Fredholm equation of second kind

Lovely Professional University 11

Notes

Calculus of Variation and Integral Equation

D. Volterra equation of first kind

6. General Criterion for a linear operator is

A. ( )+ ( ) = { ( )} . { ( )}

B. ( )+ ( ) = { ( )}/ { ( )}

C. ( )+ ( ) = { ( )} − { ( )}

D. ( )+ ( ) = { ( )} + { ( )}
7. Which of the following is a solution of the integral equation
1
( )= − ( )
1+ 1+

A. ( ) = (1 + )

B. ( )= 1+

C. ( ) = (1 + )

D. ( ) = (1 + )

8. A linear integral equation of the form

( ) ( )= ( )+ ∫ ( , ) ( ) with ( ) = 1, ( ) = 0 is

A. Homogeneous Volterra Integral equation of the third kind

B. Homogeneous Volterra Integral equation of the second kind
C. Homogeneous Volterra Integral equation of the first kind
D. Homogeneous Fredholm Integral equation of the second kind

9. A linear integral equation of the form

( ) ( )= ( )+ ( , ) ( )

where is a constant,which of the following are known functions?

A. ( )
B. ( )a ( , )
C. ( )a ( , )
D. ( ), ( ) a ( , )

2
10. What is the area under the curve y(x) = between x = 1 and x = ?
x2
A. 1
B. 3
C. 2
D. 4

12 Lovely Professional University

 Notes

Unit 01: Introduction to Integral Equations

Answers for Self Assessment

1. A 2. C 3. B 4. D 5. A

6. D 7. A 8. B 9. D 10. C

Review Questions
1. What is meant by Fredholm integral equation?
2. What is meant by Volterra integral equation?
3. What is meant by integral equation?
4. State the rules of integration of various elementary functions and derive if possible.

5. Show that ( ) = 2 is a solution of the integral equation

( )= + 3∫ ( , ) ( ) where ( , ) = sin cos when 0 ≤ ≤ and
( , ) = cos sin when ≤ ≤ .

Further Readings
1. I. M. Gelfand,S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

Web Links

1. https://www.wolfram.com/language/11/symbolic-and-numeric-calculus/solve-a-
volterra-integral-equation.html
2. https://mathworld.wolfram.com/FredholmIntegralEquationoftheSecondKind.html

Lovely Professional University 13

 Notes

Unit 02: Types of Kernels, Eigen values and Eigen functions

Pratiksha, Lovely Professional University

Unit 02: Types of Kernels, Eigenvalues and Eigenfunctions

CONTENTS
Objectives
Introduction
2.1 Singular Integral Equation
2.2 Square integrable kernel or - kernel
2.3 Special kinds of kernels of linear Integral Equations
2.4 Eigen values and Eigen functions
2.5 Leibnitz’s rule of Differentiation under the Integral Sign
Summary
Keywords
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to

 learn various types of kernels

 define Eigen values and Eigenfunctions in the context of integral equations
 differentiate under the integral sign

Introduction
In this chapter, we will learn about the singular and non singular integral equations, regularity
conditions, square integrable function and special kind of kernels. Moreover, we will learn about
the eigenvalues and eigenfunctions in the context of integral equations.

2.1 Singular Integral Equation

When one or both limits of integration in an integral equation becomes infinite or when the kernel
of an integral equation becomes infinite at one or more points within the range of integration, then
the integral equation is called as singular integral equation.
An integral equation which is not singular is known as non- singular integral equation.

Regularity conditions and square integrable function

A regularity condition is essentially just a requirement that whatever structure you are studying
isn't too poorly behaved.
While studying integral equations, we shall be mainly concerned with functions which are either
continuous or either integrable or square-integrable.
We are already familiar with continuous and integrable functions.
According to the definition introduced by Cauchy, and developed by Weierstrass,continuous
functions are functions that take nearby values at nearby points.
Let f : A → R, where A ⊂ R, and suppose that c ∈ A. Then f is continuous at c if for every ϵ> 0 there
exists a δ > 0 such that |x − c| <δ and x ∈ A implies that |f(x) − f(c)| <ϵ.

14 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

A function f : A → R is continuous on a set B ⊂ A if it is continuous at every

point in B, and continuous if it is continuous at every point of its domain A.
When mathematicians say that a function is integrable, they mean only that the integral is well
defined.
Let f be a bounded function from an interval [a, b] to R. We say that f is integrable on [a, b] if there
is a number V such that for every sequence of partitions{ } on [a, b] such that { ( )} → 0 and
every sequence { } where is a sample for
{∑( , , )} →

If f is integrable on [a, b] then the number V is denoted by ∫ and is called “the integral from a to
b of f.”
In practical terms, integrability depends on continuity. If a function is continuous on a given
interval, it’s integrable on that interval. Moreover, if a function has only a finite number of some
kinds of discontinuities on an interval, it’s also integrable on that interval.
Many functions — such as those with discontinuities, sharp turns, and vertical slopes — are non
differentiable. Discontinuous functions are also non differentiable. However, functions with sharp
turns and vertical slopes are integrable.
For example, the function y = |x| contains a sharp point at x = 0, so the function is non
differentiable at this point. However, the same function is integrable for all values of x.
So, the set of differentiable functions is actually a subset of the set of integrable functions.
The function y(x) will be a square integrablefunction, if it is measurable (i.e. finite) in the interval
[a, b] and ∫ | ( )| < ∞ , where the integral being taken in the sense of Lebesgue. The above
conditions are called as the 'regularity conditions’.

The function ( ) satisfying regularity conditions is called as 'square integrable function’ or simply
an -function.

2.2 Square integrable kernel or - kernel

The regularity conditions on the kernel ( , ) as a function of two variables are similar.
Consider a function ( , ) satisfying the conditions:
(i) ( , ) is a measurable function of x and t in the region (rectangle)

≤ ≤ , ≤ ≤ such that

( , ) <∞

(ii) ( , ) is a measurable function of t for each value of x such that

| ( , )| <∞

(iii) ( , ) is a measurable function of x for each value of t such that

| ( , )| <∞

A function ( , ) satisfying the above three regularity conditions is called as square integrable
kernel’ or an ' – kernel’.

Lovely Professional University 15

 Notes

Unit 02: Types of Kernels, Eigen values and Eigen functions

While dealing with -kernels, we shall often have occasion to use Fubini's theorem and its
extension by Tonelli and Hobson in order to justify changes in the order of integration in multiple
integrals.

G. Fubini (1879-1943) gave an important theorem related to changes in the order of integration in
multiple integrals, known as Fubini’s theorem which is stated as follows:

If ∫ ∫ ( , ) exists as a Lebesgue integral, then we have the following results:

(i) The integral ∫ ( , ) exists for almost all x and is an integrable function of x such that
∫∫ ( , ) =∫ ∫ ( , )

(ii) The integral ∫ ( , ) exists for almost all t and is an integrable function of t such that
∫∫ ( , ) =∫ ∫ ( , )
Tonelli and Hobson extended the facts given by Fubini's theorem, known as Tonelli and Hobson’s
theorem which is stated as follows:

If ( , ) is a measurable function of x and t and any one of the three integrals

∫ ∫ | ( , )| , ∫ ∫ | ( , )| , ∫ ∫ | ( , )| exists, then the integrals
∫∫ ( , ) , ∫ ∫ ( , ) , ∫ ∫ ( , ) exist and are equal to one another.

If ( , ) is an -kernel and ( ) is an -function, then it follows at once from

| ( , )| <∞

and from Schwarz's inequality ∫ ( ) ( ) ≤ ∫ | ( )|| ( )| that the function

( ) = ∫ ( , ) ( ) is defined for all x in the interval ( , ).

2.3 Special kinds of kernels of linear Integral Equations

Separable or Degenerate or Pincherle–Goursat or PG-Kernel:
A kernel ( , ) of an integral equation is called as 'separable’ or 'degenerate’ or ‘Pincherle –
Goursat’ or 'PG-kernel’ if it can be expressed in the following form:

( , )= ( ) ( )+ ( ) ( )+ ⋯+ ( ) ( )= ( ) ( )

The functions ( ) can be regarded as linearly independent, otherwise the number of terms in
above can be further reduced. Recall that the set of functions ( ) is said to be 'linearly
independent’ if
( )+ ( ) +⋯+ ( ) = 0, where , ,…, are arbitrary constants gives = =
⋯= =0
For example, ( , ) = +

Difference Kernel:

16 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

A kernel ( , ) of an integral equation, if dependent solely on the difference( − ) i.e. ( , ) =

( − ) where ( , ) is a certain function, then ( , ) is called the difference kernel of the given
integral equation.
For example, ( , ) = e

Convolution integral:
The following non-singular linear integral equations
( )= ( )+ ∫ ( − ) ( ) and

( )= ( )+ ( − ) ( )

are called integral equations of the convolution or Faltung type (closed cycle type), where ( − )
is the difference kernel.
Let ( ) and ( ) be two continuous functions defined for ≥ 0. Then, the convolution or
Faltung of and is denoted and defined by

∗ = ( − ) ( ) = ( ) ( − )

and this kind of integrals are called convolution integrals.

Transposed Kernel:
A kernel ( , ) is said to be a transposed kernel if ( , )= ( , ) where ( , ) denotes the
transpose of ( , ).

Hermitian (Complex Symmetric) Kernel:

A complex kernel ( , ) of an integral equation is said to be Hermitian or complex symmetric
kernel if ( , ) = ( , ) , where the denotes the complex conjugate of If a complex kernel of
an integral equation is not Hermitian, then it is said to be non- Hermitian kernel.

Iterated Kernels or Iterated Functions

Consider the following Fredholm integral equation:

( )= ( )+ ( , ) ( )

Then in this case, the iterated kernels ( , ); = 1,2,3, …are defined as follows:
( , )= ( , )

( , )= ( , ) ( , ) , = 2,3, …

Consider the following Volterra integral equation:

( )= ( )+ ∫ ( , ) ( )

Then in this case, the iterated kernels ( , ); = 1,2,3, … are

defined as follows:
( , )= ( , )
.
( , )= ( , ) ( , ) , = 2,3, …

Resolvent Kernel or Reciprocal Kernel

Consider the following non-singular linear integral equations:

Lovely Professional University 17

 Notes

Unit 02: Types of Kernels, Eigen values and Eigen functions

( )= ( )+ ∫ ( , ) ( ) and

( )= ( )+ ( , ) ( )

Let the solutions of these integral equations be

( )= ( )+ ∫ ( , ; ) ( ) and

( ) = ( ) + ∫ Γ( , ; ) ( )

Then, ( , ; ) and
Γ( , ; ) are respectively called the resolvent kernel or reciprocal kernel of the given integral equations.

2.4 Eigen values and Eigenfunctions

Geometrically, an eigenvector, corresponding to a real nonzero eigenvalue, points in a direction in
which it is stretched by the transformation and the eigenvalue is the factor by which it is stretched.
If the eigenvalue is negative, the direction is reversed.
Consider the following homogeneous Fredholm integral equation:

( )= ∫ ( , ) ( ) --- (1)

This equation has the obvious solution y(x) = 0, which is called as zero solution or trivial solution .

The values of the parameter for which (1) has a non-zero solution i.e. ( ) ≠ 0 , are called as the
eigenvalues of (1) or the eigenvalues of the kernel K (x, t) and each non-zero solution of (1) is called
an eigenfunction corresponding to the eigenvalue .

In other words, If ( ) is a continuous function such that ( ) ≠ 0, in the interval [a, b] and if
( )= ∫ ( , ) ( )

Then, the function ( ) is known as an eigenfunction of the integral equation (1) corresponding to
the eigenvalue .
Eigenvalues are also called as characteristic values or characteristic numbers or proper values or
fundamental values. Consequently, eigenfunctions are termed as characteristic functions or proper
functions or fundamental functions.
The following remarks can be made regarding the eigenvalues and eigenfunctions:
Remarks:

1. The number = 0 is not an eigenvalue of (1), since for = 0 it follows from (1) that ( ) = 0 ,
which is a trivial solution.

2. If ( ) is an eigenfunction of (1), then ( ), where c is an arbitraryconstant, is also an

eigenfunction of (1), which corresponds to the same eigenvalue .

2.5 Leibnitz’s rule of Differentiation under the Integral Sign

Let ( , ) and be continuous functions of x and t and also let the first derivatives of the two
functions G(x) and H(x) w.r.t. x be continuous. Then, the formula given below is known as the
Leibnitz’s rule of differentiation under the integral sign:
( ) ( )
( , ) = + , ( ) − , ( )
( ) ( )

If G and H are absolute constants, then Leibnitz's rule takes the following form :

( , ) =
x

Formula for converting a multiple integral into a single ordinary integral

18 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

The n fold integral can be written as a single integral as follows:

( − )
( ) = ( )
( − 1)!
This formula had brought a paradigm shift in mathematics. The fractional derivative got its
definition from this formula only. Though we are not going towards, fractional calculus, yet you
will find its much usage while conversion of a boundary value problem or initial value problem
into the integral equations. The literature says that its easy to solve the integral equations in
comparison to the BVP or IVP. So we convert to the integral equations and then solve the boundary
value or initial value problem.

Summary
 When one or both limits of integration in an integral equation becomes infinite or when
the kernel of an integral equation becomes infinite at one or more points within the range
of integration, then the integral equation is called as singular integral equation.
 An integral equation which is not singular is known as non- singular integral equation.
 Let f be a bounded function from an interval [a, b] to R. We say that f is integrable on [a, b]
if there is a number V such that for every sequence of partitions { } on [a, b] such that
{ ( )} → 0 and every sequence { } where is a sample for
{∑( , , )} →
 If f is integrable on [a, b] then the number V is denoted by ∫ and is called “the integral
from a to b of f.”

 In practical terms, integrability depends on continuity. If a function is continuous on a

given interval, it’s integrable on that interval.
 If ∫ ∫ ( , ) exists as a Lebesgue integral, then we have the following results:

I. The integral ∫ ( , ) exists for almost all x and is an integrable function of x such that
∫∫ ( , ) =∫ ∫ ( , )
II. The integral ∫ ( , ) exists for almost all t and is an integrable function of t such that
∫∫ ( , ) =∫ ∫ ( , )
 Consider the following homogeneous Fredholm integral equation:
( )= ∫ ( , ) ( ) --- (1)

Lovely Professional University 19

 Notes

Unit 02: Types of Kernels, Eigen values and Eigen functions

This equation has the obvious solution y(x) = 0, which is called as zero solution or trivial solution .
The values of the parameter for which (1) has a non-zero solution i.e. ( ) ≠ 0 , are called as the
eigenvalues of (1) or the eigenvalues of the kernel K (x, t) and each non-zero solution of (1) is called
an eigenfunction corresponding to the eigenvalue .

Keywords
Singular integral equation, L2 kernel, integrable function, Fubini’s theorem, reciprocal kernel,
eigenvalues of an integral equation

Self Assessment
1. Which of the following statement is correct about the following equation:
( )= ( )+ ( − ) ( )
A. It is a linear integral equation.
B. It is a non singular integral equation
C. It is a convolution type integral
D. All of these

2. ( , )= ( ) ( )+ ( ) ( ) + ⋯+ ( ) ( )=∑ ( ) ( ) is called
A. PG- kernel
B. Separable kernel
C. Degenerate kernel
D. All the above

3. A kernel ( , ) is said to be a transposed kernel if

A. ( , )= ( , )
B. ( , )= ( , )
C. ( , )= ( , )
D. ( , )= ( , )

4. ( , )=e is a
A. Difference kernel
B. Degenerate kernel
C. Both (a) and (b)
D. None of these

5. Which of the following is true?

I. The number = 0 is an eigenvalue of ( ) = ∫ ( , ) ( ) .

II. If ( ) is an eigenfunction of ( ) = ∫ ( , ) ( ) , then ( ), where c is an

arbitrary constant, is also its eigenfunction.

A. Only I is true
B. Only II is true
C. Both I and II are true
D. None is true

6. Which of the following is a correct expression of iterated kernel?

.
A. ( , )=∫ ( , ) ( , x) , = 2,3, …

20 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

.
B. ( , )=∫ ( , ) ( , ) , = 2,3, …
.
C. ( , )=∫ ( , ) ( , ) , = 2,3, …
.
D. ( , )=∫ ( , ) ( , ) , = 2,3, …

7. If ( , ) = ( , ) , then the kernel is

A. Symmetric
B. Non Hermitian
C. Hermitian
D. Non symmetric

Answers for Self Assessment

1. D 2. D 3. A 4. C 5. B

6. B 7. C

Review Questions
1. What is meant by a non singular integral equation?
2. Define kernel and elaborate.
3. Mention the various types of kernels of linear integral equations.
4. What is meant by eigenvalue in general and in the context of integral equations?
5. State Fubini’s theorem and contemplate on its role in solving the integral equations.

Further Readings
1. M. Gelfand, S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

Web Links
https://web.ma.utexas.edu/users/m408m/Display15-2-3.shtml

Lovely Professional University 21

 Notes

Pratiksha, Lovely Professional University Unit 03: Differential Equation to Integral Equation

Unit 03: Differential Equation to Integral Equation

CONTENTS
Objectives
Introduction
3.1 Initial Value Problem
3.2 Conversion of initial value problem into Volterra integral equation
3.3 Boundary Value Problem
3.4 Conversion of boundary value problem integral equation
Summary
Keywords
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to

 learn about initial value problems

 learn about boundary value problem
 convert an initial value problem into Volterra integral equation
 convert a boundary value problem into Fredholm integral equation

Introduction
While searching for the representation formula for the solution of an ordinary linear differential
equation in such a manner so as to include the boundary or initial conditions explicitly, we always
arrive at an integral equation.Thus, an ordinary differential equation may be converted to an
integral equation. Similarly, a partial differential equation may also be converted to an integral
equation.
Here, we shall describe in what ways the integral equations can arise from ordinary differential
equations. For the purpose, we shall start from initial and boundary value problems.
After converting an initial value problem or a boundary value problem to an integral equation, it
can be solved by shorter methods of solving integral equations.

3.1 Initial Value Problem

When an ordinary differential equation is to be solved under conditions involving dependent
variable and its derivatives at the same value of the independent variable, then the problem under
consideration is said to be an initial value problem.

For example,

+ = ; (1) = 2, (1) = 2

22 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

( )−3 ( )+2 ( )=4 ; (0) = 1, (0) = −2

3.2 Conversion of initial value problem into Volterra integral equation

Example1. Reduce the following IVP into an integral equation:

+ = 0; (0) = 0, ′(0) = 0.

Solution. We can write the differential equation as ( )=− ( )

Integrating both sides w.r.t. x from 0 to x, we get

( )− (0) = − ( )

Or
( )= − ( )

Again integrating the same way, we get

( ) − (0) = − ( )

( )=− ( − ) ( )

which is an integral equation and more specifically a Volterra integral equation.

Example2. Reduce the following IVP into an integral equation:

( ) − 3 ( ) + 2 ( ) = 4 sin ; (0) = 1, (0) = −2.

Also derive the original differential equation with initial conditions from the obtained integral
equation.
Solution. The given differential equation can be written as
( )=3 ( )−2 ( )+4
Integrating both sides w.r.t. x from 0 to x, we get

( )− (0) = 3 ( ) − (0) − 2 ( ) − 4 (cos − cos 0)

Applying the given initial conditions,

( ) = 3 ( )−2 ( ) − 4 cos − 1

Integrating again in the same manner, we get

( ) − (0) = 3 ( ) −2 ( − ) ( ) − 4 sin −

Or ( ) = 1 − − 4 sin + ∫ 3 − 2( − ) ( )

Which is a Volterra integral equation.

Now to convert the integral equation to differential equation we need to differentiate the integral
equation twice. And we then obtain

( ) = −1 − 4 cos + −2 ( ) +3 ( )

( ) = 4 sin − 2 ( ) + 3 ′( )

Lovely Professional University 23

 Notes

Unit 03: Differential Equation to Integral Equation

( )=4 − 2 ( ) + 3 ′( )
Or
( )− 3 ( )+2 ( )=4

And it can be observed from ( ) = 1 − − 4 sin + ∫ 3 − 2( − ) ( )

that (0) = 1 and from ( ) = −1 − 4 cos + ∫ −2 ( ) + 3 ( ) , it can be seen that (0) = −2.

Thus we got the initial conditions also and obtained a complete initial value problem.

3.3 Boundary Value Problem

When an ordinary differential equation is to be solved under conditions involving dependent
variables at the boundary values, then the problem under consideration is said to be a boundary
value problem. For example,

+ = 0; ( ) = , ( )=

is a boundary value problem. Here, it must be noted that the two different values x = a and x = b of
the independent variable x are involved.

3.4 Conversion of boundary value problem integral equation

Example1. Reduce the following BVP into an integral equation:

+ = 0; (0) = 0, ( ) = 0.

Solution. The given differential equation can be written as

( )=− ( )
Integrating both sides w.r.t. x from 0 to x, we get

( )− (0) = − ( )

Let (0) =

( )= − ( )

Integrate w.r.t. x from 0 to x

( ) − (0) = − ( )

( )= − ( − ) ( )

We know that ( ) = 0

()= − ( − ) ( )

= ( − ) ( )

( )= . ( − ) ( ) − ( − ) ( )

= ( − ) ( ) + ( − ) ( ) − ( − ) ( )

( − ) ( − )
= ( ) + ( )

24 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

= ( , ) ( )

( ) ( )
where ( , ) = ,0 < < and ( , )= , < <

Example2. Reduce the following BVP into an integral equation:

+ = 1; (0) = 0, (1) = 1.

Also recover the boundary value problem from the obtained integral equation.
Solution. We can write the given problem as
( )=1− ( ); (0) = 0, (1) = 1
Integrate both sides w.r.t. x

( )− (0) = − ( )

( )= + − ( )

where (0) =
On integration in the usual manner, we can write

( )= + − ( − ) ( )
2
( )
(Using∫ ( ) =∫ ( ) ))
( )!

Put =1
1
(1) = 1 = + − (1 − ) ( )
2
1
= + (1 − ) ( )
2

So ( ) = + ∫ (1 − ) ( ) + −∫ ( − ) ( )

+
( )= + (1 − ) ( ) + (1 − ) ( )
2

which can be written in a compact form as

+
( )= + ( , ) ( )
2
where, ( , ) = (1 − ), 0 < < and ( , ) = (1 − ), < <1
Thus, we got the desired integral equation.
Now to get back the differential equation, we must differentiate w.r.t. x
1+2
( )= − ( )
2
Differentiating w.r.t. x again
( )=1− ( )− ( )+ ( )
Thus we get
( )+ ( ) = 1, (0) = 0, (1) = 1
which is the required boundary value problem.

Lovely Professional University 25

 Notes

Unit 03: Differential Equation to Integral Equation

Summary
 An initial value problem when converted to integral equation, takes the Volterra form.
 A boundary value problem when converted to integral equation, takes the Fredholm form.

Keywords
IVP, BVP, Volterra integral equation, Fredholm integral equation

Self Assessment
1. The equation ( )+ ( ) = 1, (0) = 0, (1) = 1 is
A. an IVP
B. a BVP
C. Both (a) and (b)
D. None of these

2. A boundary value problem when converted to integral equation, takes the ________ form.

A. Volterra
B. Cauchy
C. Fredholm
D. Gaussian

3. The problem ( ) − 3 ( ) + 2 ( ) = 4 sin ; (0) = 1, (0) = −2 gets converted

to
A. Volterra integral equation
B. Fredholm integral equation
C. It is not possible to convert the given equation in any integral equation
D. None of these

4. Which of the following is correct?

( )
A. ∫ ( ) =∫ ( )!
( )

( )
B. ∫ ( ) =∫ ( )!
( )
( )
C. ∫ ( ) =∫ ( )
( )
( )
D. ∫ ( ) =∫ ( )!
( )

5. To convert the differential equation + = 0; (0) = 0, ( ) = 0 into integral

equation, it needs to be differentiated

A. once
B. twice
C. thrice
D. four times

26 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

Answers for Self Assessment

1. B 2. C 3. A 4. D 5. B

Review Questions
1. What is meant by IVP?
2. Define a BVP.
3. Differentiate in IVP and BVP.
4. Can we say that IVP and BVP are equivalent to the integral equations? Explain in
detail.
5. Write an initial value problem and convert it into the corresponding integral equation.
6. Write any boundary value problem and convert it into corresponding integral
equation.

Further Readings
1. I. M. Gelfand,S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

Web Links

1. https://www.youtube.com/watch?v=Z_i5ibKjSAg
2. https://www.youtube.com/watch?v=0VWaBJz3t1s
3. https://www.youtube.com/watch?v=WPlBrzjI1KI
4. https://projecteuclid.org/journals/differential-and-integral-equations

Lovely Professional University 27

 Notes

Unit 04: Fredholm Integral Equations of the Second Kind with Separable Kernels
Pratiksha, Lovely Professional University

Unit 04: Fredholm Integral Equations of the Second Kind with

Separable Kernels
CONTENTS
Objectives
Introduction
4.1 Characteristic functions
4.2 Orthogonality of Two Functions
4.3 Homogeneous Fredholm integral Equations of the Second kind with Separable kernels
4.4 Non-Homogeneous Fredholm Integral Equations of the second kind with Separable Kernels
Summary
Keywords
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to

 define characteristic function for integral equations

 check the orthogonality of two functions
 solve the homogeneous Fredholm integral equations of second kind with separable
kernels
 solve the non-homogeneous Fredholm integral equations of second kind with separable
kernels

Introduction
In this chapter, we will learn to solve the Fredholm integral equations of second kind in both
homogeneous and non-homogeneous forms. Prior to that, we need to learn about the orthogonality
of two functions.

4.1 Characteristic functions

Consider the following homogeneous Fredholm integral equation:

( )= ∫ ( , ) ( ) --- (1)

This equation has the obvious solution y(x) = 0, which is called as zero solution ortrivial solution .
The values of the parameter for which (1) has a non-zero solution i.e.( ) ≠ 0 , are called as the
eigenvalues or characteristic values of (1) or the eigenvalues or characteristic values of the kernel K
(x, t).

If ( ) is a continuous function such that ( ) ≠ 0, in the interval [a, b] and if

( )= ∫ ( , ) ( )

Then, the function ( ) is known as ancharacteristic function of the integral equation (1)
corresponding to the eigenvalue .

28 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

There are some important observations to be noted:

The number = 0 is not an eigenvalue of (1), since for = 0 it follows from (1) that ( ) = 0 ,
which is a trivial solution.

If ( ) is an eigenfunction of (1), then ( ), where c is an arbitrary constant, is also an

eigenfunction of (1), which corresponds to the same eigenvalue .

If the kernel ( , ) is continuous in the rectangle : ≤ ≤ , ≤ ≤ and the numbers

and are finite, then to every eigenvalue , there exists a finite number of linearly independent
eigenfunctions. The number of such functions is known as the index of the eigenvalue. Different
eigenvalues have different indices.
A homogeneous Fredholm integral equation may generally have no eigenvalues and no
eigenfunctions or it may not have any real eigenvalues and Eigenfunctions.

4.2 Orthogonality of Two Functions

Two functions ( ) and ( ), both continuous in the interval [a, b] are said to be orthogonal to
each other over the interval [a, b] if ∫ ( ) ( ) = 0.

Theorem 01: The eigenfunctions of a symmetric kernel of a homogeneous Fredholm integral

equation corresponding to two distinct eigenvalues are orthogonal.
Proof : Let ( ) and ( )be the eigenfunctions corresponding to two distinct eigenvalues and
respectively of the following homogeneous Fredholm integral equation of the second kind:

( )= ( , ) ( ) − − − (1)

We can also write ( )= ∫ ( , ) ( ) − − − (2)

and ( )= ∫ ( , ) ( ) − − − (3)

Again, since we are interested in symmetric kernel, therefore, the kernel can be written as follows :
( , )= ( , ) − − − (4)

Multiplying both sides of (2) by ( ), we get

( ) ( )= ( )∫ , ( )

Integrating it w.r.t. 'x' from a to b, we get

∫ ( ) ( ) = ∫ ( )∫ , ( )

Changing the order of integration in R.H.S. integral, we get

∫ ( ) ( ) = ∫ ( )∫ , ( )

Since the kernel is symmetric,

∫ ( ) ( ) = ∫ ( )∫ ( , ) ( ) − −(5)

Now, from equation (3), we may write

( )= ( , ) ( )

from which, we note that

( )= ( , ) ( )

Lovely Professional University 29

 Notes

Unit 04: Fredholm Integral Equations of the Second Kind with Separable Kernels

which can also be written in the following form:

( )= ∫ ( , ) ( ) − − − (6)

Putting this integral of (6) in (5)

( )
( ) ( ) = ( ) = ( ) ( )

Or

( ) ( ) = ( ) ( )

Or

∫ ( ) ( ) = ∫ ( ) ( )

Or

( − ) ( ) ( ) =0

Therefore, we should have ∫ ( ) ( ) = 0 because ≠

Hence, the Eigenfunctions ( ) and ( ) corresponding to two distinct eigenvalues and are
orthogonal to each other over the interval [ , ].
Theorem 02: The eigenvalues of a symmetric kernel of a homogeneous Fredholm integral equation
are real.
Consider the following homogeneous Fredholm integral equation of the second kind:

( )= ( , ) ( ) − − − (1)

Now, let us suppose that equation (1) has an eigenvalue , which is not real, therefore, we may
assume

= + − − − (2)

Also let

( )= + − − − (3)

be the complex eigenfunction corresponding to the eigenvalue . Then, the complex conjugate
of would necessarily be an eigenvalue corresponding to the eigenfunction ( ) which is
complex conjugate of ( ). Thus, we have

= − and ( )= −

Now, since ( ) and ( ) are the eigenfunctions corresponding to the eigenvalues and
respectively, they must satisfy equation (1). Thus, we have

( )= ( , ) ( ) − − − (4)

And

30 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

( )= ( , ) ( ) − − − (5)

Again, since we are interested in symmetric kernel, therefore, ( , ) = ( , )

Multiplying both sides of (4) by ( ) and then integrating w.r.t. x over the interval [a, b], we get

( ) ( ) = ( ) ( , ) ( )

Changing the order of integration in R.H.S. integral, we get

( ) ( ) = ( ) ( , ) ( )

Or

( ) ( ) = ( ) ( , ) ( ) − − − (6)

Now, from (5), we can write ( )= ∫ ( , ) ( )

from which, we can write ( )= ∫ ( , ) ( )

Or ( )= ∫ ( , ) ( )

( )
Therefore, =∫ ( , ) ( )

Putting this value of ∫ ( , ) ( ) in (6), we get

( )
( ) ( ) = ( )

Or

( ) ( ) = ( ) ( )

Or

− ( ) ( ) =0

Or

2 ∫ ( + ) =0

Since ( ) is an eigenfunction, therefore ( ) ≠ 0,

which implies that ( ) ≠ 0 and therefore ( ) ( )≠0
i.e. + ≠ 0,

Lovely Professional University 31

 Notes

Unit 04: Fredholm Integral Equations of the Second Kind with Separable Kernels

So, ∫ ( + ) ≠0

Thus, =0
Hence, the eigenvalues of a symmetric kernel of a homogeneous Fredholm integral equation are
real.

4.3 Homogeneous Fredholm integral Equations of the Second kind

with Separable kernels
Consider a homogenous Fredholm integral equation:

( )= ( , ) ( )

where ( , ) is a separable kernel and can be written as

( , )= ( ) ( )

Putting ( , ) in ( ), we get

( )= ( ) ( ) ( )

Interchanging the order of sum and integral, we have

( )= ( ) ( ) ( )

Now let us define,

( ) ( ) = ; = 1,2, … ,

Then, the integral equation

( )= ( ) ( ) ( )

becomes ( ) = ∑ ( ); = 1,2, … ,
where, constants are to be determined to find the complete solution of the given integral equation.
Multiply both sides of
( )= ∑ ( )∫ ( ) ( ) successively by ( ), ( )… ( ) and integrating over [ , ],
and define

= ( ) ( ) ; , = 1,2, … ,

We can write

= ; = 1,2, … ,

or(1 − ) − − ⋯− =0

Similarly, we get
(1 − ) − − ⋯− =0
(− ) + (1 − ) − ⋯− =0
………………………………………………………….
(− ) + (− ) − ⋯ + (1 − ) =0

32 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

Thus, we get a homogeneous linear equations which may be used to find the values of , and its
determinant can be written as:
1− − … −
− 1− … −
( )=
… … … …
− − … 1−
If this determinant is non zero, then the system of equations from which it is made up of, has only trivial
solution given by = = ⋯ = 0. Thus the integral equation also has a trivial solution i.e. a zero solution.
If this determinant is zero, then at least one of the can be assigned in an arbitrary manner and the other
′ can be obtained accordingly. Therefore infinitely many solutions exist for the integral equation in the
second case.
The eigenvalues of the given integral equation can be obtained from
( )=0
or
1− − … −
− 1− … −
=0
… … … …
− − … 1−
The degree of this equation in is less than or equal to n. So, if the integral equation has separable kernel,
then the integral equation has at most ‘n’ eigenvalues and corresponding eigenfunctions will be its
solutions.

4.4 Non-Homogeneous Fredholm Integral Equations of the second

kind with Separable Kernels

Consider a non-homogenous Fredholm integral equation of second kind:

( )= ( )+ ( , ) ( )

where ( , ) is a separable kernel and can be written as

( , )= ( ) ( )

Putting ( , ) in ( ), we get

( )= ( )+ ( ) ( ) ( )

Interchanging the order of sum and integral, we have

( )= ( )+ ( ) ( ) ( )

Now let us define,

( ) ( ) = ; = 1,2, … ,

Then, the integral equation

( )= ( )+ ( ) ( ) ( )

becomes ( ) = ( ) + ∑ ( ); = 1,2, … ,
where, constants are to be determined to find the complete solution of the given integral equation.

Lovely Professional University 33

 Notes

Unit 04: Fredholm Integral Equations of the Second Kind with Separable Kernels

Multiply both sides of

( )= ∑ ( )∫ ( ) ( ) successively by ( ), ( )… ( ) and integrating over [ , ],
and define

= ( ) ( ) ; , = 1,2, … ,

And =∫ ( ) ( ) ; = 1,2, … ,

We can write

= + ; = 1,2, … ,

= + + +⋯+

or (1 − ) − − ⋯− =
Similarly, we can simplify the equations further to
get
(1 − ) − − ⋯− =
(− ) + (1 − ) − ⋯− =

………………………………………………………….

(− ) + (− ) − ⋯ + (1 − ) =

Thus, we get a homogeneous linear equations which may be used to find the values of , and its
determinant can be written as:
1− − … −
− 1− … −
( )=
… … … …
− − … 1−
If this determinant is non zero, then the system of equations from which it is made up of, has only trivial
solution given by = = ⋯ = 0. Thus the integral equation also has a trivial solution i.e. a zero solution.
If this determinant is zero, then at least one of the can be assigned in an arbitrary manner and the other
′ can be obtained accordingly. Therefore infinitely many solutions exist for the integral equation in the
second case.
The eigenvalues of the given integral equation can be obtained from
( )=0
or
1− − … −
− 1− … −
=0
… … … …
− − … 1−
The degree of this equation in is less than or equal to n.
Now three cases arise:
Case 1: When at least one is non zero
If ( ) is non zero, then a unique non zero solution of the system of equations exists and hence the integral
equation has a unique non zero solution.
If ( ) is zero, then either infinitely many solutions or no solution of the system of equations exists and
hence the integral equation has infinitely many solutions or no solution.

Case 2: The function ( ) = 0 in right hand side of given integral equation

We see that all the =0∀

34 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

So the system of equations reduces to a system of homogeneous linear equations.

If ( ) ≠ 0, then a unique zero solution exists.
If ( ) = 0, then infinitely many non zero solutions exist.

Case 3: ( ) ≠ 0 and ( ) is orthogonal to functions ( )

The system of equations reduces to a system of homogeneous linear equations.

If ( ) ≠ 0, then a unique non-zero solution exists.

If ( ) = 0, then infinitely many non zero solutions exist.

Summary
 The equation ( ) = ∫ ( , ) ( ) has the obvious solution y(x) = 0, which is called as

zero solution or trivial solution.The values of the parameter for which the equation has a
non-zero solution i.e.( ) ≠ 0 , are called as the eigenvalues or characteristic valuesof the
integral equation or the eigenvalues or characteristic values of the kernel K (x, t).

 The number = 0 is not an eigenvalue of (1), since for = 0 it follows from (1) that
( ) = 0 , which is a trivial solution.
 If ( ) is an eigenfunction of (1), then ( ), where c is an arbitrary constant, is also an
eigenfunction of (1), which corresponds to the same eigenvalue .
 If the kernel ( , ) is continuous in the rectangle : ≤ ≤ , ≤ ≤ and the
numbers and are finite, then to every eigenvalue , there exists a finite number of
linearly independent eigenfunctions. The number of such functions is known as the index
of the eigenvalue. Different eigenvalues have different indices.
 Two functions ( ) and ( ), both continuous in the interval [a, b] are said to be

orthogonal to each other over the interval [a, b] if ∫ ( ) ( ) = 0.

 The eigenfunctions of a symmetric kernel of a homogeneous Fredholm integral equation

corresponding to two distinct eigenvalues are orthogonal.
 The eigenvalues of a symmetric kernel of a homogeneous Fredholm integral equation are
real.
 In case of homogeneous Fredholm integral equation with separable kernel, there can be
zero solution or infinitely many solutions.
 In case of non-homogeneous Fredholm integral equation with separable kernel, there can
be zero solution, unique non-zero solution, no solution or infinitely many solutions.

Keywords
Orthogonal functions, characteristic functions,Fredholm integral equation of second kind,
homogeneous integral equation, non-homogeneous integral equation

Self Assessment
1. The equation ( ) = ∫ ( , ) ( ) has the obvious solution y(x) = 0, which is called

A. Zero solution
B. Trivial solution

Lovely Professional University 35

 Notes

Unit 04: Fredholm Integral Equations of the Second Kind with Separable Kernels

C. Both (a) and (b)

D. None of these

2. The values of the parameter for which the equation has a zero solution i.e. ( ) = 0 , are
called as the eigenvalues or characteristic values
A. True
B. False

3. The number = 0 is not an eigenvalue of ( ) = ∫ ( , ) ( )

A. True

B. False

4. Two functions ( ) and ( ), both continuous in the interval [a, b] are said to be
orthogonal to each other over the interval [a, b] if

A. ∫ ( ) ( ) ≠0

B. ∫ ( ) ( ) =0

C. ∫ ( ) ( ) =0

D. ∫ ( ) ( ) =0

5. Which of the following statements is true?

A. The eigenfunctions of a symmetric kernel of a homogeneous Fredholm integral

equation corresponding to two distinct eigenvalues are orthogonal.
B. The eigenfunctions of any kernel of a homogeneous Fredholm integral equation
corresponding to two distinct eigenvalues are orthogonal.
C. The eigenfunctions of a symmetric kernel of a non-homogeneous Fredholm integral
equation corresponding to two distinct eigenvalues are orthogonal.
D. The eigenfunctions of a symmetric kernel of a homogeneous Fredholm integral
equation corresponding to two repeated eigenvalues are orthogonal.

6. The eigenvalues of a symmetric kernel of a homogeneous Fredholm integral equation are

A. Imaginary
B. Real
C. Integral
D. Natural

Answers for Self Assessment

1. C 2. B 3. A 4. D 5. A

6. B

36 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

Review Questions
1. Find the characteristic values and characteristic functions of the homogeneous
integral equation:
( )= ( )
2. Show that ( ) = ∫ (3 − 2) ( ) has no characteristic values and characteristic
functions.
3. Solve ( ) = ∫ sin( + ) ( )
4. Solve ( ) = ( ) + ∫ ( )
5. Solve ( ) = + ∫ ( + ) ( )

Further Readings
1. I. M. Gelfand,S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

Web Links

1. https://www.youtube.com/watch?v=rCWzF1yvZlQ
2. https://www.youtube.com/watch?v=K0t53t7RLWY

Lovely Professional University 37

 Notes

Unit 05: Fredholm Alternative Theorem and Resolvent Kernel

Pratiksha, Lovely Professional University

Unit 05: Fredholm Alternative Theorem and Resolvent Kernel

CONTENTS
Objectives
Introduction
5.1 Fredholm Alternative Theorem
5.2 Proof of Fredholm Alternative Theorem
5.3 Fredholm Theorem
5.4 Iterated Kernels of Non-homogeneous Fredholm Integral Equation
5.5 Resolvent Kernel or Reciprocal Kernel of Non-homogeneous Fredholm Integral Equation
5.6 Iterated Kernels of Non-homogeneous Volterra Integral Equation
5.7 Resolvent Kernel or Reciprocal Kernel of Non-homogeneous Volterra Integral Equation
Summary
Keywords
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to
 assure that the solution of integral equation exists
 define the iterated and resolvent kernel
 solve the non-homogeneous integral equations

Introduction
Separable kernel in an integral equation implies integral equation of second kind becomes the
algebraic system of equations
For more general kernels, what can be the consequence? To answer this we must know that any
well behaved kernel can be expressed as infinite series of separable kernels. When an integral
equation cannot be solved in closed form,we use approximation methods.
For any approximation method, existence of solution should be known in advance and Fredholm
theorem provides the assurance that the solution of integral equation exists.

5.1 Fredholm Alternative Theorem

Either the integral equation
( )= ( )+ ∫ ( , ) ( ) …(i)
with fixed possesses one and only one solution ( ) for arbitrary functions ( ) and ( , ), in
particular, the solution = 0 for = 0; or the homogeneous equation
( )= ∫ ( , ) ( ) …(ii)
possesses a finite number of linearly independent solutions , = 1, 2, … , .

38 Lovely Professional University

Notes

Calculus of Variation and Integral Calculus

In the first case, the transposed inhomogeneous equation

( )= ( )+ ∫ ( , ) ( ) …(iii)
also possesses a unique solution.
In the second case, the transposed homogeneous equation
( )= ∫ ( , ) ( ) …(iv)
also has linearly independent solutions , = 1, 2, … , ;
The inhomogeneous integral equation (i) has a solution if and only if the given function ( )
satisfies the conditions
∫ ( ) ( ) = 0, = 1, 2, … , …(v)
In this case the solution of (i) is determined only up to an additive linear combination ∑

5.2 Proof of Fredholm Alternative Theorem

Consider a non-homogenous Fredholm integral equation of second kind:

( )= ( )+ ( , ) ( )

where ( , ) is a separable kernel and can be written as

( , )= ( ) ( )

Putting ( , ) in ( ), we get

( )= ( )+ ( ) ( ) ( )

Interchanging the order of sum and integral, we have

( )= ( )+ ( ) ( ) ( )

Now let us define,

( ) ( ) = ; = 1,2, … ,

Then, the integralequation

( )= ( )+ ( ) ( ) ( )

becomes ( ) = ( ) + ∑ ( ); = 1,2, … ,
where, constants are to be determined to find the complete solution of the given integral equation.
Multiply both sides of
( )= ∑ ( )∫ ( ) ( ) successively by ( ), ( )… ( ) and integrating over [ , ],
and define

= ( ) ( ) ; , = 1,2, … ,

And =∫ ( ) ( ) ; = 1,2, … ,

We can write

= + ; = 1,2, … ,

= + + +⋯+

or (1 − ) − − ⋯− =
Similarly, we can simplify the equations further to

Lovely Professional University 39

 Notes

Unit 05: Fredholm Alternative Theorem and Resolvent Kernel

get
(1 − ) − − ⋯− =
(− ) + (1 − ) − ⋯− =

………………………………………………………….

(− ) + (− ) − ⋯ + (1 − ) =

Thus, we get a homogeneous linear equations which may be used to find the values of , and its
determinant can be written as:
1− − … −
− 1− … −
( )=
… … … …
− − … 1−
The required solution of (1) depends on the determinant given by (9).
Two cases arise:

Case I. If ( ) ≠ 0, then the system of equations (8) has only one solution given by Cramer's rule,

⋯
= ( )
, = 1, 2, … , …(10)

where, denotes the cofactor of the ( , ) element of the determinant ( ).

Substituting the value from (10) in (5), the unique solution of (1) is given by
⋯
( )= ( )+ ∑ ( )…(11)
( )

Consider the following determinant of the (n+1)th order

0 ( ) ( ) … ( )
( ) 1− − … −
( , : )= ( ) − 1− … −
… … … … …
( ) − − … 1−

Now by developing this determinant by the elements of the first row and the corresponding minor
of the first column, we have

40 Lovely Professional University

Notes

Calculus of Variation and Integral Calculus

{ ( )+ ( )+ ⋯+ ( )} ( ) = ( , : )

Therefore (13) can be written as

( , : )
( )= ( )+ ( )
( )

Or

( )= ( )+ ∫ ( , : ) ( ) …(14)

where

( , : )
( , : )= is known as the resolvent or reciprocal kernel of the given integral equation.
( )

The only possible singular points of ( , : ) in the plane are the roots of the equation ( ) = 0.

5.3 Fredholm Theorem

The inhomogeneous Fredholm integral equation (1) with a separable kernel has a unique solution,
given by

( )= ( )+ ∫ ( , : ) ( ) .
, :
The resolvent kernel ( , : ) is given by the quotient of two polynomials ( )
.

Case II. If ( ) = 0, then (8) has no solution in general, because an algebraic system with ( ) = 0
can be solved only for some particular values of , , … . For this situation, we re write the
system of equations (8) in matrix form as follows:

…
( − ) = , where = … … , = … ,
…

= …

And is the identity matrix of order . Since ( ) = 0, for each non trivial solution of the
homogeneous system ( − ) = 0, there corresponds a non-trivial solution (eigenfunction) of the
homogeneous integral equation (12).

To each eigenvalue for which the determinant ( ) = | − | has the rank , 1 ≤ ≤

, then there are = ( − ) linearly independent solutions of the algebraic system(8)
Similarly, the homogeneous equation (12), there will be linearly independent solutions say
( ), ( ), … ( ) and let us assume that they are normalized.

Then to each eigenvalue of index , there corresponds a solution ( ) of (3) of the form

Lovely Professional University 41

 Notes

Unit 05: Fredholm Alternative Theorem and Resolvent Kernel

( )= ( )

where are arbitrary constants.

Let be the multiplicity of the eigenvalue. From linear algebra, using elementary transformations
on the determinant | − |, we shall get at most + 1 identical rows and this is obtained when A
is symmetric.
⇒ ≥ −
⇒ = − ≤

where the equality holds if =

Therefore, if = is a root of multiplicity ≥ 1 of the equation ( ) = 0, then the homogeneous

equation (12) has linearly independent solutions, where is the index of eigenvalue such that 1 ≤ ≤ .

= geometric multiplicity

= algebraic multiplicity

For the situation when for non-homogeneous Fredholm integral equation ( ) = 0 we define the
transpose of equation (1) as

( )= ( )+ ∫ ( , ) ( ) …(15)

Equation (1) and (15) are transpose of each other and the relation between them is symmetric.
We can write its kernel as
( , )=∑ ( ) ( ) …(16)
And proceeding on this , we get the algebraic system

( − ) = …(17)

where =∫ ( ) ( ) and =∫ ( ) ( )

The determinant ( ) is the same function, except that there is an interchange of the rows and
columns. Hence, the eigenvalues of (15) are same as that of the original integral equation (1). So (15)
possesses a unique solution, whenever (1) does.

The number of LI eigenfunctions will be for the transposed homogeneous integral equation . Let
these LI solutions be denoted by , ,…, and let us assume that they have been
normalized. Then any solution ( ) of the transposed integral equation

( )= ∫ ( , ) ( ) …(18)

corresponding to the eigenvalue is of the form

( )=∑ ( ) where are arbitrary constants.

We now prove that the eigenfunctions ( ) and ( ) corresponding to the distinct eigenvalues
and , are orthogonal.
By definition,

( )= ∫ ( , ) ( ) …(19)

and

42 Lovely Professional University

Notes

Calculus of Variation and Integral Calculus

( )= ∫ ( , ) ( ) …(20)

Multiplying both sides of (19) by ( ) and then integrating w.r.t. over the interval ( , ), we get

( ) ( ) = ( ) ( , ) ( )

Or

∫ ( ) ( ) = ∫ ( )∫ ( , ) ( ) …(21)

Again multiplying both sides of (20) by ( ) and then integrating w.r.t. over the interval
( , ), we get

∫ ( ) ( ) = ∫ ( )∫ ( , ) ( ) …(22)

From (21) and (22), we get

( ) ( ) = ( ) ( )

And since these eigenvalues are distinct, so ∫ ( ) ( ) =0

⇒ ( ) and ( ) are orthogonal.

The solution of (1) for the case ( ) = 0: The necessary and sufficient condition for (1) to possess a
solution for = , a root of ( ) = , is that ( ) be orthogonal to the eigenfunctions of the
transposed equation (15).

Proof: The condition is necessary.

Suppose that (1) for = admits a certain solution ( ). Then we have

( )= ( )+ ( , ) ( )

Or

( )= ( )− ( , ) ( )

Multiplying both sides of above equation by ( ) and then integrating w.r.t. x over the interval (a,
b)

( ) ( ) = ( ) ( ) − ( ) ( , ) ( ) =0

The condition is sufficient.

Lovely Professional University 43

 Notes

Unit 05: Fredholm Alternative Theorem and Resolvent Kernel

The system (1 − ) = reduces to only − independent equations

⇒ the rank of the matrix (1 − ) is = −

⇒ the system (8)is solvable.
Substituting this solution in (5), we get the solution of (1).

Finally we know that the difference of any two solutions of (1) is a solution of the homogeneous equation
(12) . Hence the most general solution of (1) has the form
( )= ( )+ ( )+ ( ) + ⋯+ ( )

Where ( ) is a suitable linear combination of the functions ( ). ( ), … ( ).

Thus, we have proved that if = is a root of multiplicity ≥ 1 of the equation ( ) = 0, then the
inhomogeneous equation (1) has a solution if and only if ( ) is orthogonal to all the eigenfunctions of the
transposed equation.

5.4 Iterated Kernels of Non-homogeneous Fredholm Integral Equation

Consider the following non-homogeneous Fredholm integral equation:

( )= ( )+ ( , ) ( )

Then, the iterated kernels or functions ( , ); = 1, 2, 3, … , are defined as follows:

( , )= ( , )

and ( , )=∫ ( , ) ( , ) , = 2, 3, … ,

or ( , )=∫ ( , ) ( , ) , = 2, 3, … ,

The iterated kernels are also called as the approximate kernels or approximate functions.

5.5 Resolvent Kernel or Reciprocal Kernel of Non-homogeneous

Fredholm Integral Equation
Let the solution of non-homogeneous Fredholm integral equation

( )= ( )+ ( , ) ( )

takes the form

( )= ( )+ ∫ ( , ; ) ( )

where

( , ; )= ( , )

where ( , ) is the iterated kernel and ( , ; ) is the resolvent kernel of the non-
homogeneous Fredholm integral equation.

5.6 Iterated Kernels of Non-homogeneous Volterra Integral Equation

Consider the following non-homogeneous Volterra integral equation:

44 Lovely Professional University

Notes

Calculus of Variation and Integral Calculus

( )= ( )+ ( , ) ( )

Then, the iterated kernels or functions ( , ); = 1, 2, 3, … , are defined as follows:

( , )= ( , )

and ( , )=∫ ( , ) ( , ) , = 2, 3, … ,

or ( , )=∫ ( , ) ( , ) , = 2, 3, … ,

The iterated kernels are also called as the approximate kernels or approximate functions.

5.7 Resolvent Kernel or Reciprocal Kernel of Non-homogeneous

Volterra Integral Equation
Let the solution of non-homogeneous Volterra integral equation

( )= ( )+ ( , ) ( )

takes the form

( )= ( )+ ( , ; ) ( )

where

( , ; )= ( , )

where ( , ) is the iterated kernel and ( , ; ) is the resolvent kernel of the non-
homogeneous Volterra integral equation.

Summary
Either the integral equation

( )= ( )+ ∫ ( , ) ( ) …(i)
with fixed possesses one and only one solution ( ) for arbitrary functions ( ) and ( , ), in
particular, the solution = 0 for = 0; or the homogeneous equation
( )= ∫ ( , ) ( ) …(ii)
possesses a finite number of linearly independent solutions , = 1, 2, … , .
In the first case, the transposed inhomogeneous equation
( )= ( )+ ∫ ( , ) ( ) …(iii)
also possesses a unique solution.
In the second case, the transposed homogeneous equation
( )= ∫ ( , ) ( ) …(iv)
also has linearly independent solutions , = 1, 2, … , ;
The inhomogeneous integral equation (i) has a solution if and only if the given function ( )
satisfies the conditions
∫ ( ) ( ) = 0, = 1, 2, … , …(v)
In this case the solution of (i) is determined only up to an additive linear combination ∑

Lovely Professional University 45

 Notes

Unit 05: Fredholm Alternative Theorem and Resolvent Kernel

 The iterated kernels or functions ( , ); = 1, 2, 3, … , are defined as follows:

( , )= ( , )

and ( , )=∫ ( , ) ( , ) , = 2, 3, … ,

 Let the solution of non-homogeneous Volterra integral equation

( )= ( )+ ( , ) ( )

takes the form

( )= ( )+ ( , ; ) ( )

where

( , ; )= ( , )

where ( , ) is the iterated kernel and ( , ; ) is the resolvent kernel of the non-
homogeneous Volterra integral equation.

Keywords
 Fredholm Alternative Theorem,
 Iterated kernel,
 Resolvent kernel,
 Non-homogeneous integral equations

Self Assessment
1. Which of the following statement is correct?

A. The iterated kernels are also called as the approximate kernels.

B. The iterated kernels are also called as approximate functions.
C. Both (a) and (b)
D. None of these

2. The only possible singular points of ( , : ) in the plane are

A. roots of the equation ( − 1) = 0
B. roots of the equation ( + 1) = 0
C. roots of the equation ( ) = 0
D. roots of the equation ( ) = 0

3. Which of the following is true for the resolvent kernel?

A. ( , ; )=∑ ( , )
B. ( , ; )=∑ ( , )

46 Lovely Professional University

Notes

Calculus of Variation and Integral Calculus

C. ( , ; )=∑ ( , )
D. ( , ; )=∑ ( , )

4. If∫ ( ) ( ) = 0, then ( ) and ( ) are

A. diagonal
B. symmetric
C. Hermitian
D. Orthogonal

5. If = is a root of multiplicity ≥ 1 of the equation ( ) = 0, then the corresponding

homogeneous integral equation has linearly independent solutions, where is the index of
eigenvalue such that
A. 1 ≤ ≤
B. −1 ≤ ≤−
C. −1 ≤ ≤
D. m ≤ ≤1

Answers for Self Assessment

1. A 2. C 3. D 4. D 5. A

Review Questions
1. Find the iterated kernel for the following:

(i) ( , ) = sin( − 2 ) , 0 ≤ ≤2 , 0≤ ≤2

(ii) ( , ) = of Volterra integral equation

2. State the Fredholm Alternative Theorem.

3. Prove the Fredholm Alternative Theorem.
4. Find the resolvent kernel for the following:

( , ) = sin( − 2 ) , 0 ≤ ≤2 , 0≤ ≤2

5. Find the resolvent kernel for the following:

(i) ( , )= of Volterra integral equation
(ii) ( , )= of Volterra integral equation

Further Readings
1. I. M. Gelfand,S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

Web Links
1. https://mathworld.wolfram.com/FredholmAlternative.html
2. https://terrytao.wordpress.com/2011/04/10/a-proof-of-the-fredholm-alternative/

Lovely Professional University 47

 Notes

Unit 05: Fredholm Alternative Theorem and Resolvent Kernel

48 Lovely Professional University

 Notes

Unit 06: Method of Successive Substitution and Approximation

Pratiksha, Lovely Professional University

Unit 06: Method of Successive Substitution and Approximation

CONTENTS
Objectives
Introduction
6.1 Solution of Fredholm integral of second kind by successive substitution
6.2 The solution of a non-homogeneous Volterra integral equation by the method of successive
substitutions
6.3 Solution of non-homogenous Fredholm Integral Equation by Successive
Approximations/Iterative Method/ Iterative scheme/ Neumann's Series
6.4 Solution of non-homogenous Volterra Integral Equation by Successive Approximations/
Iterative Method/ Iterative Scheme/ Neumann's Series
Summary
Key words
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to

 Solve the Fredholm integral equation of second kind by successive substitution

 Solve the Volterra integral equation of second kind by successive substitution
 Solve the Fredholm and Volterra integral equation of second kind by successive
approximation

Introduction
In this chapter we will mainly learn to solve the Fredholm and Volterra integral equation of second
kind by the method of successive substitution and successive approximation. We establish the
methods using the theorems and then the problems can be solved using those methods.

6.1 Solution of Fredholm integral of second kind by successive

substitution
The solution of a non-homogenous Fredholm integral equation by the method of successive
substitutions is explained with the help of following theorem.
Consider the following non-homogeneous Fredholm integral equation:

( )= ( )+ ∫ ( , ) ( ) ---------- (1)

Also consider that

(i) ( , ) ≠ 0 is real and continuous in the rectangle R for which ∈ [ , ], ∈ [ , ] and

| ( , )| ≤ in R ------------(2)
(ii) ( ) ≠ 0 is real and continuous in the interval I for which ∈ [ , ]. Also | ( )| ≤
in I ---------------(3)

Lovely Professional University 49

Notes

Calculus of Variation and Integral Equation

(iii) is a constant such that | | ≤ ( )

-------------(4)
Then, the integral equation (1) has a unique continuous solution in I, which is given by the
following uniformly and absolutely convergent series:

( )= ( )+ ∫ ( , ) ( ) + ∫ ( , )∫ ( , ) ( ) + ⋯ --------------(5)

Proof: The given integral equation can be written as

( )= ( )+ ∫ ( , ) ( ) ---------- (6)

From which, we note that

( ) = ( )+ ∫ ( , ) ( ) -----------(7)

Substituting this value of ( ) in (1), we get

( )= ( )+ ∫ ( , ) ( ) + ∫ ( , )∫ ( , ) ( ) -----------(8)

Moreover, (7) can be written as

( ) = ( )+ ∫ ( , ) ( ) --------------(9)

From which we see that

( )= ( )+ ∫ ( . ) ( ) ---------------(10)

Substituting this value of ( ) in (8), we get

( )= ( )+ ( , ) ( ) + ( , ) ( , )[ ( ) + ( . ) ( ) ]

Or
( )=
( )+
∫ ( , ) ( ) +
∫ ( , )∫ ( , ) ( ) + ∫ ( , )∫ ( , )∫ ( , ) ( ) -------------- (11)

Proceeding in the same manner, we get

( )=
( )+
∫ ( , ) ( ) +
∫ ( , )∫ ( , ) ( ) +
…+ ∫ ( , )∫ ( , )…∫ ( , ) ( ) … + ( ) -------------- (12)

Where, ( )= ∫ ( , )∫ ( , ) ( , )…∫ ( , ) ( ) … -----(13)

Now consider the following infinite series:

( )= ( )+ ∫ ( , ) ( ) + ∫ ( , )∫ ( , ) ( ) + ⋯ -----------(14)

Because of (i) and (ii), each term of the series (14) is continuous in I. Therefore the series (14) is also
continuous in I, if it converges uniformly in I.
Let ( ) be the general term of the series (14), then we can write

( )= ∫ ( , )∫ ( , )… ∫ ( , ) ( ) … … -------------(15)

This implies that

| ( )| = | ∫ ( , )∫ ( , )…∫ ( , ) ( ) … |------------(16)

And
| ( )| ≤ | | ( − )
Therefore the series (14) converges absolutely and uniformly only when
1
| |<
( − )

50 Lovely Professional University

 Notes

Unit 06: Method of Successive Substitution and Approximation

which is the assumption (ii).
So the infinite series (14) converges absolutely and uniformly when condition (4) holds good.
Hence the non-homogeneous Fredholm integral equation (1) has a continuous solution given by
(12). If ( ) is continuous in I, then |y(x)| must have a maximum value, where, | ( )| < .
Now, from equation (13), we have
| ( )| ≤ | | ( − ) ---------- (17)
Now using (4), we get
lim → | ( )| = 0 ----------- (18)
Therefore, the function ( ) satisfying (12) is the continuous function given by the series (14).
Hence proved what was stated.

6.2 The solution of a non-homogeneous Volterra integral equation by

the method of successive substitutions
The solution of a non-homogenous Fredholm integral equation by the method of successive
substitutions is explained with the help of following theorem.
Consider the following non-homogeneous Volterra integral equation:
( )= ( )+ ∫ ( , ) ( ) ---------- (1)

Also consider that

(i) ( , ) ≠ 0 is real and continuous in the rectangle R for which ∈ [ , ], ∈ [ , ] and

| ( , )| ≤ in R ------------(2)
(ii) ( ) ≠ 0 is real and continuous in the interval I for which ∈ [ , ]. Also | ( )| ≤
in I ---------------(3)
(iii) is a constant such that | | ≤ ( )
-------------(4)
Then, the integral equation (1) has a unique continuous solution in I, which is given by the
following uniformly and absolutely convergent series:

( )= ( )+ ∫ ( , ) ( ) + ∫ ( , )∫ ( , ) ( ) + ⋯ --------------(5)

Proof: The given integral equation can be written as

( )= ( )+ ∫ ( , ) ( ) ---------- (6)

From which, we note that

( ) = ( )+ ∫ ( , ) ( ) -----------(7)

Substituting this value of ( ) in (1), we get

( )= ( )+ ∫ ( , ) ( ) + ∫ ( , )∫ ( , ) ( ) -----------(8)

Moreover, (7) can be written as

( ) = ( )+ ∫ ( , ) ( ) --------------(9)

From which we see that

( )= ( )+ ∫ ( . ) ( ) ---------------(10)

Substituting this value of ( ) in (8), we get

( )= ( )+ ( , ) ( ) + ( , ) ( , )[ ( ) + ( . ) ( ) ]

Or
( )=
( )+

Lovely Professional University 51

Notes

Calculus of Variation and Integral Equation

∫ ( , ) ( ) +
∫ ( , )∫ ( , ) ( ) + ∫ ( , )∫ ( , )∫ ( , ) ( ) -------------- (11)

Proceeding in the same manner, we get

( )=
( )+
∫ ( , ) ( ) +
∫ ( , )∫ ( , ) ( ) +
…+ ∫ ( , )∫ ( , )…∫ ( , ) ( ) … + ( ) -------------- (12)

Where, ( )= ∫ ( , )∫ ( , ) ( , )…∫ ( , ) ( ) … -----(13)

Now consider the following infinite series:

( )= ( )+ ∫ ( , ) ( ) + ∫ ( , )∫ ( , ) ( ) + ⋯ -----------(14)

Let ( ) be the general term of the series (14), then we can write

( )= ∫ ( , )∫ ( , )… ∫ ( , ) ( ) … … -------------(15)

This implies that

| ( )| = | ∫ ( , )∫ ( , )…∫ ( , ) ( ) … |------------(16)

And
| | ( )
| ( )| ≤ ; ≤ ≤ -------------(17)
!

which is the assumption (iii).

| | ( )
The series for which the positive constant Is the general expression of the nth term is
!
convergent for all values of , , ( − ).
So the infinite series (14) converges absolutely and uniformly.
Hence the non-homogeneous Volterra integral equation (1) has a continuous solution given by (12).
If ( ) is continuous in I, then |y(x)| must have a maximum value, where, | ( )| < .
Now, from equation (13), we have
| | ( )
| ( )| ≤
( )!
---------- (17)

Now using (4), we get

lim → | ( )| = 0 ----------- (18)
Therefore, the function ( ) satisfying (12) is the continuous function given by the series (14).
Hence proved what was stated.

6.3 Solution of non-homogenous Fredholm Integral Equation by

Successive Approximations/Iterative Method/ Iterative scheme/
Neumann's Series
Consider the following non-homogeneous Fredholm integral equation:

( )= ( )+ ∫ ( , ) ( ) ----------(1)

Consider ( ) = ( )--------(2)
as a zero order approximation for the solution of (1). Further, if ( ) and ( ) are respectively
the nth order and (n-1)th order approximations of y(x), then

( )= ( )+ ∫ ( , ) ( ) ---------(3)

We know that the iterated kernels ( , ), = 1,2, … are defined by

( , ) = ( , ) ------------(4a)

52 Lovely Professional University

 Notes

Unit 06: Method of Successive Substitution and Approximation

And ( , )=∫ ( , ) ( , ) ---------------(4b)

Put = 1 in (3), the first order approximation is given as

( )= ( )+ ∫ ( , ) ( ) --------- (5)

(2) implies that ( ) = ( ) ----------------(6)

Substituting this value of ( ) in (5), we get

( )= ( )+ ∫ ( , ) ( ) -------------(7)

Putting n=2 in (3), we get

( )= ( )+ ∫ ( , ) ( ) or

( )= ( )+ ∫ ( , ) ( ) -----------(8)

Replacing x by s in (7)

( ) = ( )+ ∫ ( , ) ( ) -----------(9)

Substituting this value from (9) in (8), we get

( )= ( )+ ∫ ( , ) ( ) + ∫ ( , )[∫ ( , ) ( ) ] ----------(10)

Or

( ) = ( ) + ∑{ } ∫ ( , ) ( ) -------------(11)

And it can be generalized to

( ) = ( ) + ∑{ } ∫ ( , ) ( ) -------------(12)

In the limiting case, we get Neumann series as:

( ) = lim → ( ) = ( ) + ∑{ } ∫ ( , ) ( ) -------------(13)

BY changing the order of integration in third term on RHS of (10), we can write

( )= ( )+ ( , ) ( ) + ( , ) ( )

We now determine the resolvent kernelor reciprocal kernel in terms of iterated kernels ( , ). For
this purpose, by changing the order of integration and summation in the Neumann series, we get

( ) = ( ) + ∫ [∑ ( , )] ( ) ------------(14)

Comparing (14) with ( ) = ( ) + ∫ ( , ; ) ( ) --------- (15)

We get ( , ; ) = ∑ ( , )]--------------(16)

6.4 Solution of non-homogenous Volterra Integral Equation by

Successive Approximations/ Iterative Method/ Iterative Scheme/
Neumann's Series
Consider the following non-homogeneous Volterra integral equation:
( )= ( )+ ∫ ( , ) ( ) ----------(1)

Consider ( ) = ( )--------(2)
as a zero order approximation for the solution of (1). Further, if ( ) and ( ) are respectively
the nth order and (n-1)th order approximations of y(x), then
( )= ( )+ ∫ ( , ) ( ) ---------(3)

We know that the iterated kernels ( , ), = 1,2, … are defined by

( , ) = ( , ) ------------(4a)

Lovely Professional University 53

Notes

Calculus of Variation and Integral Equation

And ( , )=∫ ( , ) ( , ) ---------------(4b)

Put = 1 in (3), the first order approximation is given as

( )= ( )+ ∫ ( , ) ( ) --------- (5)

(2) implies that ( ) = ( ) ----------------(6)

Substituting this value of ( ) in (5), we get
( )= ( )+ ∫ ( , ) ( ) -------------(7)

Putting n=2 in (3), we get

( )= ( )+ ∫ ( , ) ( ) or

( )= ( )+ ∫ ( , ) ( ) -----------(8)

Replacing x by z in (7)
( )= ( )+ ∫ ( , ) ( ) -----------(9)

Substituting this value from (9) in (8), we get

( )= ( )+ ∫ ( , ) ( ) + ∫ ( , )[∫ ( , ) ( ) ] ----------(10)

On changing the order of integration, we can write

( )= ( )+ ( , ) ( ) + ( )[ ( , ) ( , ) ]

By using 4a and 4b in it, we get

( )= ( )+∑ ∫ ( , ) ( ) ------------- (11)

And it can be generalized to

( ) = ( ) + ∑{ } ∫ ( , ) ( ) -------------(12)

In the limiting case, we get Neumann series as:

( ) = lim → ( ) = ( ) + ∑{ } ∫ ( , ) ( ) -------------(13)

BY changing the order of integration in third term on RHS of (10), we can write

( )= ( )+ ( , ) ( ) + ( , ) ( )

We now determine the resolvent kernel or reciprocal kernel in terms of iterated kernels ( , ).
For this purpose, by changing the order of integration and summation in the Neumann series, we
get

( ) = ( ) + ∫ [∑ ( , )] ( ) ------------(14)

Comparing (14) with ( ) = ( ) + ∫ ( , ; ) ( ) --------- (15)

We get ( , ; ) = ∑ ( , )]--------------(16)
The series converges absolutely and uniformly when K(x,t) is continuous inR.

Summary
 Consider the following non-homogeneous Fredholm integral equation:

( )= ( )+ ∫ ( , ) ( ) ---------- (1)

Also consider that

(i) ( , ) ≠ 0 is real and continuous in the rectangle R for which ∈ [ , ], ∈ [ , ] and

| ( , )| ≤ in R

54 Lovely Professional University

 Notes

Unit 06: Method of Successive Substitution and Approximation

(ii) ( ) ≠ 0 is real and continuous in the interval I for which ∈ [ , ]. Also | ( )| ≤

in I
(iii) is a constant such that | | ≤ ( )

Then, the integral equation (1) has a unique continuous solution in I, which is given by the
following uniformly and absolutely convergent series:

( )= ( )+ ( , ) ( ) + ( , ) ( , ) ( ) +⋯

 Consider the following non-homogeneous Volterra integral equation:

( )= ( )+ ∫ ( , ) ( ) ---------- (2)

Also consider that

(i) ( , ) ≠ 0 is real and continuous in the rectangle R for which ∈ [ , ], ∈ [ , ] and

| ( , )| ≤ in R
(ii) ( ) ≠ 0 is real and continuous in the interval I for which ∈ [ , ]. Also | ( )| ≤
in I
(iii) is a constant such that | | ≤ ( )

Then, the integral equation (2) has a unique continuous solution in I, which is given by the
following uniformly and absolutely convergent series:

( )= ( )+ ( , ) ( ) + ( , ) ( , ) ( ) +⋯

 Consider the following non-homogeneous Fredholm integral equation:

( )= ( )+ ∫ ( , ) ( ) ----------(3)

Let ( ) = ( )
as a zero order approximation for the solution of (3). Further, if ( ) and ( ) are respectively
the nth order and (n-1)th order approximations of y(x), then

( )= ( )+ ( , ) ( )

 Consider the following non-homogeneous Volterra integral equation:

( )= ( )+ ∫ ( , ) ( ) ----------(4)

Let ( ) = ( )
as a zero order approximation for the solution of (4). Further, if ( ) and ( ) are respectively
the nth order and (n-1)th order approximations of y(x), then

( )= ( )+ ( , ) ( )

Key words
 Fredholm integral equation,
 Volterra Integral equation,
 Integral equation of second kind,
 successive substitution, successive approximation

Self Assessment
1. Which of the following statement is correct in context of integral equations?

Lovely Professional University 55

Notes

Calculus of Variation and Integral Equation

A. ( , ) ≠ 0 is real and continuous in the rectangle R for which ∈ [ , ], ∈ [ , ] and
| ( , )| ≤ in R
B. ( ) ≠ 0 is real and continuous in the interval I for which ∈ [ , ]. Also | ( )| ≤ in I

C. is a constant such that | | ≤ ( )

D. All of the above

2. The nth order approximations of y(x) is given as

A. ( )= ( )+ ∫ ( , ) ( )

B. ( )= ( )+ ∫ ( , ) ( )

C. ( )= ( )+ ∫ ( , ) ( )

D. ( )= ( )+ ∫ ( , ) ( )

3. Which of the following is true for resolvent kernel?

A. ( , ; )= ∑ ( , )]
B. ( , ; )= ∑ ( , )]
C. ( , ; )= ∑ ( , )]
D. ( , ; )= ∑ ( , )]

4. The series ( ) = ( ) + ∫ ( , ) ( ) + ∫ ( , )∫ ( , ) ( ) +⋯

A. Converges absolutely
B. Converges uniformly
C. Converges absolutely and uniformly
D. Does not converge
| | ( )
5. term in context of the integral equation ( )= ( )+ ∫ ( , ) ( )
!

where M and N are such that

(i) ( , ) ≠ 0 is real and continuous in the rectangle R for which ∈ [ , ], ∈ [ , ] and
| ( , )| ≤ in set of real numbers

(ii) ( ) ≠ 0 is real and continuous in the interval I for which ∈ [ , ]. Also | ( )| ≤

in I
,is a/an

A. Constant
B. Positive constant
C. Negative constant
D. Arbitrary variable

Answers for Self Assessment

1. D 2. C 3. A 4. C 5. B

Review Questions
1. What is the method of successive substitution to solve integral equations?
2. State the Fredholm integral equation of second kind.
3. What is the method of successive approximation to solve the integral equations?
4. State the Volterra integral equation of second kind.

56 Lovely Professional University

 Notes

Unit 06: Method of Successive Substitution and Approximation

5. Solve ( ) = sin + 2 ∫ ( )

6. Solve ( ) = sin + 2 ∫ ( ) )

7. Reduce the boundary value problem ( ) + ( ) = . ∈ (0, ), (0) = , ( ) = −

to the Fredholm integral equation and hence solve it.

Further Readings
1. I. M. Gelfand,S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc.
(1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley
(1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press
(1947)

Web Links

1. https://www.youtube.com/watch?v=Qo9riz9NZ64
2. https://www.youtube.com/watch?v=u4yhu8QMC2M

Lovely Professional University 57

 Notes

Pratiksha, Lovely Professional University Unit 07: Iterated and Resolvent Kernels

Unit 07: Iterated and Resolvent Kernels

CONTENTS
Objectives
Introduction
7.1 Some important theorems on iterated and resolvent kernels
7.2 Volterra integral equation of the second kind
7.3 Volterra integral equation of the first kind
Summary
Key words
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to

 identify some properties of the Resolvent kernels

 solve the integral equations approximately
 solve the integral equations when the kernel takes a particular polynomial form

Introduction
In this chapter we will learn about some theorems on iterated kernels and resolvent kernels. We
will solve some integral equations also. We will also learn about the solution when the kernel takes
two specific polynomial forms.

7.1 Some important theorems on iterated and resolvent kernels

1. Let ( , ; )be the resolvent kernel of the following non-homogeneous Fredholm
integral equation:
( )= ( )+ ( , ) ( )

Then, this resolvent kernel satisfies the following relation:

( , ; )= ( , )+ ( , ) ( , ; )

2. Let ( , ; )be the resolvent kernel of the following non homogeneous Volterra integral equation:

( )= ( )+ ( , ) ( )

Then, this resolvent kernel satisfies the following relation:

( , ; )= ( , )+ ( , ) ( , ; )

3. The resolvent kernel satisfies the integro-differential equation

58 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

( , ; )
= ( , ; ) ( , ; )

Example:Solve by the method of successive approximation:

5 1
( )= + ( )
6 2
Here ( , ) =
( , )=

( , )= ( , ) ( , )

Therefore,

( , )= . =
3

( , )= . =
3 3
.
.
.

( , )=
3
Therefore, ( , ; ) = ∑ ( , )

=
3

=
6

6
=
5
The solution of given equation is
5 1 6 5
( )= +
6 2 5 6
6
= =
6

Example:Solve ( ) = 1 + ∫ ( )

Here ( ) = 1, = 1, ( , ) = 1 = ( , )

( , )= ( , ) ( , )

= 1 = −

( , )= ( , ) ( , )

( − )
= ( − ) = |
2

Lovely Professional University 59

 Notes

Unit 07: Iterated and Resolvent Kernels

( − )
=
2

( , )= ( , ) ( , )

( − )
=
2
( − )
=
3!
⋮
( − )
( , )=
( − 1)!

( , ; )= ( , )

( − )
= =
( − 1)!

The solution is,

( )=1+

=1+

= 1 + (− )( − 1)
=1−1+
=

Example:Solve ( ) = 1 + ∫ ( + ) ( )

Here
( , )= + = ( , )

( , )= ( + )( + )

+ 1
= + +
2 3
1
( , )= ( + ) + + +
2 3 2
7 1
= ( + )+ +
12 3
7 7 1
( , )= ( + ) + + +
12 12 3
15 13 13
= ( + )+ +
24 12 36

And proceeding like this the nth iterated kernel can be calculated and hence the resolvent kernel
too can be calculated.
When the resolvent kernel cannot be obtained in the closed form, we use the method
of successive approximations to find solution up to third order.
Let the given Fredholm integral equation of second kind be

( )= ( )+ ∫ ( , ) ( ) ---------- (1)

60 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

Let ( ) = ( ) -------- (2)
be the zero order approximation.
If ( ) be the nth order approximation, then

( )= ( )+ ∫ ( , ) ( ) ---------------(3)

From (2) and (3), we get ( ), ( ), ( ).

Example:Solve ( ) = +∫ ( − ) ( ) by the method of successive approximation till the

second approximation.
Here, = 1, = − , ( )= , ( )=

( )= + ( − )

= −
6

( )= + ( − ) −
6

= − +
3! 5!
7.2 Volterra integral equation of the second kind
Consider the equation,
( )= ( )+ ∫ ( , ) ( ) -------(1)

Where ( , ) is

I. a polynomial of degree (n-1) in x of the form:

( )( − ) ( − )
( , )= ( )+ ( )( − ) + +⋯+
2! ( − 1)!

II. a polynomial of degree (n-1) in t of the form:

( )( − ) ( − )
( , )= ( )+ ( )( − ) + + ⋯+ ( )
2! ( − 1)!

( , ; )
For Case I. ( , ; ) =

where ( , ; ) is the solution of the differential equation,

( ) ( )
− + +⋯+ ( ) =0

satisfying the conditions

= = =⋯= = 0 and = 1 when x=t

The required solution of (1) is

( )= ( )+ ( , ; ) ( )

( , ; )
For Case II. ( , ; ) = −

Lovely Professional University 61

 Notes

Unit 07: Iterated and Resolvent Kernels

where ℎ( , ; ) is the solution of the differential equation,

ℎ ( ) ℎ ( )
− + + ⋯+ ( )ℎ = 0

satisfying the conditions

ℎ= = =⋯= = 0 and = 1 when t=x

The required solution of (1) is

( )= ( )+ ( , ; ) ( )

Example:Solve ( ) = 29 + 6 + ∫ {5 − 6( − )} ( )

Comparewith ( ) = ( ) + ∫ ( , ) ( )

( ) = 29 + 6
=1
( , ) = 5 − 6( − ) = ( )+ ( )( − )
( , ; )
So = 5, = −6 Then ( , ; ) =

Or ( , ; 1) =

where ( , ; 1) satisfies the differential equation

( )
− [ + ( ) ] =0

Or − +6 =0

Or ( − 5 + 6) = 0 such that = 0 when =

and = 1 when = .

The auxiliary equation gives = 2,3

Therefore = +
Using the conditions we can find = , =− .
( ) ( )
So = ( , ; 1) = −

( ) ( )
=3 −2

( ) ( )
=9 −4
( ) ( )
So ( , ; 1) = 9 −4
Therefore the solution of the integral equation is

( )= ( )+ ( , ; ) ( )

( ) = 29 + 6 , = 1, ( , ; 1) = 9 ( ) ( )
−4

( ) ( )
∴ ( ) = 29 + 6 + [9 −4 ] [29 + 6 ]

= 93 − 64

7.3 Volterra integral equation of the first kind

The Volterra’s integral equation of first kind is ∫ ( , ) ( ) = ( ) and it can be converted to

62 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

( )= ( )+ ( , ) ( )

which is the Volterra’s integral equation of second kind.

Example:Solve ( ) = ∫ ( ) , (0) = 0

∫ ( ) = ( ) ------(1)

and (0) = 0 -----(2)

Kernel ( , ) = and are continuous.

Condition (2) is necessary for (1) to possess a continuous solution.

Because is continuous, the same must be true for ( ).

Differentiating (1) w.r.t.

( ) + ( ) = ′( )

Implies ( ) = ( )−∫ ( ) -------(3)

which is Volterra’s integral equation of second kind.

Compare (3) with

( )= ( )+ ∫ ( , ) ( ) ----- (4)

where ( ) = ( )

= −1

( , )=

Let be the mth iterated kernel, then

( , )= ( , )=

( , )= ( , ) ( , )

( , )= = |

=( − )

( − )
( , )=
2!

( − )
( , )=
3!

( − )
( , )=
( − 1)!

Lovely Professional University 63

 Notes

Unit 07: Iterated and Resolvent Kernels

( , ; )= ( , )

= (−1) ( , )

= ( , )− ( , )+ ( , )−⋯

− ( − )
= {1− + − ⋯}
1! 2!
( )
= . =1

Therefore the solution is given by

( )= ( )+ ( , ; ) ( )

= ( )− ( )

= ( ) − ( )|

= ( ) − ( ) + (0)

= ( )− ( )

Summary
 Let ( , ; )be the resolvent kernel of the following non-homogeneous Fredholm
integral equation:

( )= ( )+ ( , ) ( )

Then, this resolvent kernel satisfies the following relation:

( , ; )= ( , )+ ( , ) ( , ; )

 Let ( , ; )be the resolvent kernel of the following non homogeneous Volterra integral equation:

( )= ( )+ ( , ) ( )

Then, this resolvent kernel satisfies the following relation:

( , ; )= ( , )+ ( , ) ( , ; )

 The resolvent kernel satisfies the integro-differential equation

( , ; )
= ( , ; ) ( , ; )

 When the resolvent kernel cannot be obtained in the closed form, we use the method of
successive approximations to find solution up to third order.

 For the integral equation ( ) = ( ) + ∫ ( , ) ( ) ------ (1)

where ( , ) is

I. a polynomial of degree (n-1) in x of the form:

( )( − ) ( − )
( , )= ( )+ ( )( − ) + +⋯+
2! ( − 1)!

64 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

II. a polynomial of degree (n-1) in t of the form:

( )( − ) ( − )
( , )= ( )+ ( )( − ) + + ⋯+ ( )
2! ( − 1)!

( , ; )
For Case I. ( , ; ) =

where ( , ; ) is the solution of the differential equation,

( ) ( )
− + +⋯+ ( ) =0

satisfying the conditions

= = =⋯= = 0 and = 1 when x=t

The required solution of (1) is

( )= ( )+ ( , ; ) ( )

( , ; )
For Case II. ( , ; ) = −

where ℎ( , ; ) is the solution of the differential equation,

ℎ ( ) ℎ ( )
− + + ⋯+ ( )ℎ = 0

satisfying the conditions

ℎ= = =⋯= = 0 and = 1 when t=x

The required solution of (1) is

( )= ( )+ ( , ; ) ( )

Key words
 Iterated kernel,
 Resolvent kernel,
 Kernel in polynomial form

Self Assessment
1. Which of the following statement is correct?
A. The resolvent kernel satisfies the integro-differential equation
( , ; )
= ( , ; ) ( , ; )

B. The resolvent kernel satisfies the integro-differential equation

( , ; )
= ( , ; ) ( , ; )

C. The resolvent kernel satisfies the integro-differential equation

( , ; )
= ( , ; ) ( , ; )

Lovely Professional University 65

 Notes

Unit 07: Iterated and Resolvent Kernels

D. The resolvent kernel satisfies the integro-differential equation
( , ; )
= ( , ; ) ( , ; )

2. For integral equation:

( )= ( )+ ( , ) ( )

the resolvent kernel satisfies the following relation:

A. ( , ; ) = ( , )+ ∫ ( , ) ( , ; )

B. ( , ; ) = ( , )+ ∫ ( , ) ( , ; )

C. ( , ; ) = ( , )+ ∫ ( , ) ( , ; )

D. ( , ; ) = ( , )+ ∫ ( , ) ( , ; )

3. By the method of successive approximation for the equation,

5 1
( )= + ( )
6 2
the second iterated kernel is given by

A.

B.

C.

D.
( )
4. For = −1 and ( , )= , the resolvent kernel is given as
( )!

A. 1
B. 2
C.
D.
5. The equation ( ) = ( ) + ∫ ( , ) ( ) is

A. Volterra integral equation of first kind

B. Volterra integral equation of second kind
C. Volterra integral equation of third kind
D. Volterra integral equation of fourth kind

Answers for Self Assessment

1. A 2. B 3. C 4. A 5. B

Review Questions
1. What is the method of successive approximation?

2. Solve ( ) = +∫ ( + ) ( )

3. Solve ∫ ( ) = by all the methods that you know.

4. Solve ∫ ( ) = .

66 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

5. Solve ( ) = ∫ ( ) , (0) = 0

Further Readings
1. I. M. Gelfand,S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

Web Links

1.https://www.youtube.com/watch?v=Qo9riz9NZ64
2.https://www.youtube.com/watch?v=u4yhu8QMC2M

Lovely Professional University 67

 Notes

Pratiksha, Lovely Professional University Unit 08: Green’s Functions

Unit 08: Green’s Functions

CONTENTS
Objectives
Introduction
8.1 Self-Adjoint equation
8.2 Green’s formula
8.3 Dirac delta function
8.4 Derivatives of Dirac delta function
8.5 Green’s Function
8.6 Solution of a BVP using Green’s function
Summary
Key words
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to
 define the Green’s function
 solve the integral equations using Green’s function
 find the solution of boundary value problem using Green’s function

Introduction
In this chapter, we will learn some fundamental functions and then the Green’s function that is
used to solve the integral equations. Let us begin with the adjoint of a differential equation.

8.1 Self-Adjoint equation

Definition1: Adjoint equation of 2nd order linear differential equation is

( ) + ( ) + ( ) =0 (1)

Here the coefficient ( ), ( ) and ( ) are continuous and ( ) > 0 on ≤ ≤ .

Let be a differential operator defined by

≡ ( ) + ( ) + ( ) (2)

Then (1) can be rewritten as

( ) = 0, ≤ ≤ (3)
And the Adjoint operator of is defined as

( )= { ( ) ( )} − { ( ) ( )} (4)

68 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

Also
( )=0

i.e. { ( ) ( )} − { ( ) ( )} + ( ) ( )=0 (5)

is known as Adjoint of (1)

Definition 2: Self adjoint equationif the adjoint equation of any linear homogeneous equation is
identical with the equation itself, then the given equation is known as self adjoint equation.

Theorem1: The necessary and sufficient condition that the second order homogeneous linear
differential equation + + = 0 as defined in (1) on ≤ ≤ to be self adjoint is that
( ) = ( ) on ≤ ≤ .

Proof: By definition, the adjoint equation of

( ) + ( ) + ( ) =0 (1)

is

( ( ) )− ( ( ) )+ ( ) =0 (2a)

i.e.

( ) + {2 ( )− ( )} +{ ( )− ( )+ ( )} = 0 (2b)

Necessary condition: Let (1) be a self adjoint equation. Then (2b) must be identical with (1)
∴2 ( )− ( )= ( ) (3)
And
( )− ( )+ ( )= ( )
⇒ ( )= ( )
Or
( )= ( )+

(3) ⇒ 2( ( ) + ) − ( )= ( )
⇒ =0
Hence
( )= ( )

Sufficient condition: Let ( )= ( ) for (1)

Then
2 ( )− ( )=2 ( )− ( )= ( )
And
( )− ( )+ ( )= ( )− ( )+ ( )
⇒ ( )− ( )+ ( )= ( )
Then (2b)⇒

( ) + ( ) + ( ) =0

Which is same as (1)

∴ the condition is sufficient.

Lovely Professional University 69

 Notes

Unit 08: Green’s Functions

Corollary: If (1) is self adjoint, then it can also be written as

( ) + ( ) =0

Theorem 2: If the coefficients ( ), ( ) and ( ) in (1) are continuous on ≤ ≤ and ( )≠

0, then it can be transformed into the equivalent self adjoint equation

( ) + ( ) = 0,
( ) ( )
∫ ( ) ∫
where = ( ) and = ( )
( ) .

Theorem 3: If is the adjoint operator of operator , then is the adjoint of

Example: Find the adjoint operator of

( ) + (2 + 1) ( ) + ( ) = 0 (1)
Here ( )= , ( )=2 + 1 and ( )=1

We know that the adjoint equation of (1) is

( )− ( )+ =0

⇒ ( ) − + + =0

Or

( +2 )− − + =0

Or
+ 4 − (2 + 1) + (3 − 6 ) =0 (2)

8.2 Green’s formula

If is a differential operator such that

( )= ( ) + ( ) + ( ) ( ), < < (1)

Then its adjoint operator is defined as

( )= { ( )} − { ( )} + ( ) ( ), < < (2)

From (1) and (2)

− = { ( − )} + { ( − ′)}

∴ ( − ) = { ( − )+ ( − )}

70 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

=[ ( − )+ ( − )] (3)

∴(3) is called Green’s formula or operator

8.3 Dirac delta function

It represents –point mass, point charge, an impulse, a pulse produced by sudden electrical
discharge.
The role - solution of IVP (initial-value problems) and BVP (boundary value problems).
Definition: Consider a function ( ) such that
1
| |<
( )= 2 ,
0, | |>

Then ∫ ( ) =∫ =1

lim → ( ) = ( )=Dirac delta function

∴ ( ) can be defined as,

∞, =
( )=
0, ≠

And ∫ ( − ) =1

1, ∈( , )
Also ∫ ( − ) =
0, ∉( , )

8.4 Derivatives of Dirac delta function

Let ′( ) exists and both ( ) and ′( ) can be treated as ordinary functions in the rule for
integration by parts.
Then

( ) ( ) = ( ) ( )| − ( ) ( )

= − ′(0)

Using shifting property, ∫ ( ) ( − ) = ′( )

∴ ( ) ( ) = − ′(0)

Repeating this process,

( ) ( )( ) ( ) (0)
= (−1)

And
( )=− ( )

8.5 Green’s Function

IVP, BVP can be converted to integral equations by many ways. Here we learn to convert them to
integral equations with the help of Green's function.

Lovely Professional University 71

 Notes

Unit 08: Green’s Functions

 George Green, English mathematician who was first to attempt to devise a theory of
electricity and magnetism, gave this function.

Green’s Function
Consider a linear homogeneous differential equation of order :
( ) = 0, …(1)
where is the differential operator

≡ ( ) + ( ) + ⋯+ ( )…(2)

The functions ( ), ( ), … , ( ) are continuous on [ , ] and ( ) ≠ 0 on [ , ].

The boundary conditions are ( ) = 0, = 1, 2, … …(3)

where
( ) ( ) ( )( ( ) ( ) ( )(
( )≡ ( )+ ( )+ ⋯+ )+ ( )+ ( ) + ⋯+ ) (4)

( )( ( )(
where, the linear forms , ,…, in ( ), ( ), … , ), ( ), ( ), … , )are linearly
independent.
Suppose that the homogeneous BVP (1) to (4) has only a trivial solution ( ) = 0. Then the Green's
function of this BVP is the function ( , ) constructed for any point , < < and which has the
following properties:
(i) In each of the intervals [ , ), ( , ] the function ( , ) considered as a function of is a solution
of (1), that is

( )=0 (5)

(ii) ( , ) is continuous and has continuous derivatives w r t up to order ( − 2) for ≤ ≤

(iii) ( − 1) derivative of ( , ) with respect to at the point = has discontinuity of the first
kind , the jump being equal to − ( ) , i.e.

72 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

− =− ( )
… (6)

(iv) ( , ) satisfies the boundary conditions (3), that is

( ) = 0, = 1, 2, … , … (7)

Theorem: If the boundary value problem (1) to (4), has only a trivial solution ( ) ≡ 0, then the
operator has a unique Green's function ( , ).
Proof :Let ( ) , ( ), … , ( ) be linearly independent solutions of the equation (1).
Then, property (i) of ( , ) ⇒ ( , ) must have the following representation
( )+ ( ) + ⋯+ ( ), ≤ <
( , )=
( )+ ( ) + ⋯+ ( ), < ≤
(8)
where , ,…, , , ,…, are some functions of .
Property (ii) of ( , ) ⇒ at point =
[ ( ) + ⋯+ ( )] − [ ( ) +⋯+ ( )] = 0 ( )
[ ( ) + ⋯+ ( )] − [ ( ) +⋯+ ( )] = 0 ( )
⋮
[ ( ) + ⋯+ ( )] − [ ( ) + ⋯+ ( )] = 0 ( )
Property (iii) of ( , ) ⇒

[ ( ) + ⋯+ ( )] − [ ( ) + ⋯+ ( )] = − ( )
( )

= ( )− ( ), where = 1,2,3, … ,
∴( ), ( ), … , ( )⇒
( )+ ⋯+ ( )=0 ( )
( )+ ⋯+ ( )=0 ( )
⋮

( )+ ⋯+ ( )=−
( )
( )

The determinant of the systems ( ), ( ), … , ( ) is

( ) ( ) ⋯ ( )
( ) ( ) ⋯ ( )
=
⋮ ⋮ ⋮ ⋮
( ) ( ) ⋯ ( )
= ( , ,…, ) (10)
∵ , ,…, are linearly independent.
∴ = ( , ,…, )≠0 (11)
⇒ The system ( ,( ), … , ( ) possess a unique solution for , = 1,2,3, … ,
Now to find ( ), ( ), we use the property (iv) of ( , ).
Let ( )= ( ) + ( ) (12)
where
( ) ( ) ( )
( )= ( )+ ( ) +⋯+ ( )
( ) ( ) ( )
( )= ( )+ ( ) +⋯+ ( )
Property (iv) ⇒
( )= ( ) +⋯+ ( )+ ( ) + ⋯+ ( ), where = 1,2,3, … , (15)

Lovely Professional University 73

 Notes

Unit 08: Green’s Functions

Equation(9) [ = ( )− ( )] ⇒
= − , = 1,2,3, … , (16)
(15) and (16)⇒
( − ) ( )+ ⋯+( − ) ( )+ ( ) + ⋯+ ( )=0 (17)
Or
[ ( )+ ( )] + ⋯ + [ ( )+ ( )] = ( )+ ⋯+ ( )
Or
( ) +⋯+ ( )= ( )+ ⋯+ ( ), = 1,2, … , (18)

Here (18) is a linear system of equations for determining quantities , ,…, .

∵ , ,…, are linearly independent.
∴ determinant of system (18) is non zero.
( ) ( ) ⋯ ( )
( ) ( ) ⋯ ( )
i.e. ≠0 (19)
⋮ ⋮ ⋮ ⋮
( ) ( ) ⋯ ( ))

∴ (19)⇒(18) possesses a unique solution in ( ), ( ), … , ( ). Already ( ), = 1,2,3, … , are

unique.
∴ (16)⇒ ( ), ( ), … , ( ) are unique.
∴ the existence and uniqueness of the Green’s function ( , ) has been established.

Note: 1. If the BVP (1) to (4)is self adjoint then Green’s function is symmetric, i.e. ( , ) = ( , ).
The converse is also true.
2. If at one of the extremities of an interval [ , ], the coefficient of the highest derivative vanishes,
then the natural boundary condition for boundedness of the solution at = is imposed for
( ) = 0 and at the other extremity the ordinary boundary condition is specified.

8.6 Solution of a BVP using Green’s function

First of all we convert the BVP to a Fredholm integral equation
We shall use the following notations :

≡ ( ) + ( ) +⋯+ ( )

( ) ( ) ( )( ( ) ( ) ( )(
( )≡ ( )+ ( )+ ⋯+ )+ ( )+ ( ) + ⋯+ )

Suppose ( , ) is Green's function of the boundary value problem

( ) = 0, … (1)
( ) = 0, = 1, 2, … , … (2)

involving homogeneous boundary conditions (2) at the end points = and = of the interval
[ , ].

74 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

Example: Consider the BVP

( )+ ( )=0 (3)
( ) = 0, = 1,2, … , (4)
involving the same homogeneous boundary conditions as in (2)
Then, solution of BVP (3)-(4) is given as

( )=∫ ( , ) ( ) (5)

Example: Consider the BVP

( )+ ( )=0 (6)
( ) = 0, = 1,2, … , (7)
involving the same homogeneous boundary conditions as in (2)
The function ( ) can be written as
( )= , ( ) (8)

Then, the BVP (6)-(7) can be reduced to

( )=∫ ( , ) ( , ( )) (9)

Particular case: Let ( ) = ( ) ( ) − ( ),

Then the BVP
( )+ ( ) ( )= ( ) (10)
( ) = 0, = 1,2, … , (11)
reduces to the following

( )= ∫ ( , ) ( ) ( ) −∫ ( , ) ( ) (12)

Where ( , ) is the relevant Green’s function.

In (12), ( , ) ( ) is not symmetric, unless ( ) is a constant.

Example: When the prescribed end conditions are not homogeneous we’d use the modified
method as follows:
Let ( , ) be the Green’s function corresponding to the associated homogenous end conditions.
We look for a function ( ) such that the relation

( )= ( )+∫ ( , ) ( ) (13)

is equivalent to
( )+ ( )=0 (14)
together with the prescribed non-homogeneous end conditions

∫ ( , ) ( ) =− ( ) (15)

The requirement that (13)⇒ (14) leads to

Lovely Professional University 75

 Notes

Unit 08: Green’s Functions

( ) =0 (16)

Since the second term in (13) satisfies the associated homogeneous end conditions.
⇒ ( ) in (13) must be a solution of (16) which satisfies the prescribed non-homogeneous end
conditions.
When ( , ) exists, ( ) exists.

Special Case:
Consider a linear homogeneous differential equation of order two:
( ) + ( ) = 0, ≤ ≤ -------(1)
where L is a self adjoint operator

= + = + + -----------(2)

together with the homogeneous boundary conditions

( )+ ( )=0
( )+ ( ) = 0 --------------(3)
such that at least one of , and one of , is non zero.
Here , & are given real valued continuous functions defined on [a,b] such that ( ) ≠ 0 on [a,
b].
By definition, a function ( , ) is called a Green’s function for ( ) = 0 if for a given t
( , ), ≤ <
( , )={ ----------------(4)
( , ), < ≤
where and are such that

(i) =0 ℎ ≤ <
And =0 ℎ < ≤ ---------------(5)
(ii) and satisfythe equations
( )+ ( )=0
( )+ ( )=0

(iii) The function ( , ) is continuous at x=t i.e. ( , )= ( , ) -------------(6)

(iv) The derivative of ( , ) has a discontinuity of magnitude − ( )
at = ---------(7)
Now we will construct the Green’s function for the boundary value problem (2)-(3).
Let = ( ) is a non trivial solution of the associated equation = 0 ------(8)
Let = ( ) and = ( ) satisfy the two homogeneous conditions of (3) respectively. It follows
that the properties (i) and (ii) are satisfied if we take ( ) = ( ) and ( ) = ( ) and ,
are constants.
( ), ≤ <
We have ( , ) = {
( ), < ≤

Property (iii) implies that ( )− ( ) = 0 --------------(10)

Property (iv) implies that ( )− ( )=−

( )
---------- (11)

The determinant D of the system of equations (10) and (11) is

( ) − ( )
= ( ( ), ( ))
′( ) − ′( )

76 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

Let ( )& ( ) be two linearly independent solutions of (8)

Then = ≠ 0 ----------(13)
Implies that (10) and (11) possess a unique solution.
1
= =
− − − +
( ) ( )

Theorem. Let the Green’s function for ( ) = 0 be given by

1
( ) ( ), − ≤ <
( , )=={
1
− ( ) ( ), < ≤

where A is a constant independent of x and t; then ( ) is a solution of the boundary value problem
given by (1) and (3) iff

( )= ( , ) ( )

Summary
 Adjoint equation of 2nd order linear differential equation

( ) + ( ) + ( ) =0 is

{ ( ) ( )} − { ( ) ( )} + ( ) ( )=0

 ∫ ( − ) =∫ { ( − )+ ( − )}

=[ ( − )+ ( − )]
is called Green’s formula or operator

 Consider a linear homogeneous differential equation of order :

( ) = 0, …(1)
where is the differential operator

≡ ( ) + ( ) + ⋯+ ( )…(2)

The functions ( ), ( ), … , ( ) are continuous on [ , ] and ( ) ≠ 0 on [ , ].

The boundary conditions are ( ) = 0, = 1, 2, … …(3)

where
( ) ( ) ( )( ( ) ( ) ( )(
( )≡ ( )+ ( )+ ⋯+ )+ ( )+ ( ) + ⋯+ ) (4)

( )( ( )(
where , the linear forms , ,…, in ( ), ( ), … , ), ( ), ( ), … , )are
linearly independent.
Suppose that the homogeneous BVP (1) to (4) has only a trivial solution ( ) = 0. Then the Green's
function of this BVP is the function ( , ) constructed for any point , < < and which has the
following properties:
(i) In each of the intervals [ , ), ( , ] the function ( , ) considered as a function of is a solution
of (1), that is

Lovely Professional University 77

 Notes

Unit 08: Green’s Functions

( )=0 …(5)

(ii) ( , ) is continuous and has continuous derivatives w r t up to order ( − 2) for ≤ ≤

(iii) ( − 1) derivative of ( , ) with respect to at the point = has discontinuity of the first
kind , the jump being equal to − ( ) , i.e.

− =− ( )
… (6)

(iv) ( , ) satisfies the boundary conditions (3), that is

( ) = 0, = 1, 2, … , … (7)

 If the boundary value problem (1) to (4), has only a trivial solution ( ) ≡ 0, then the
operator has a unique Green's function ( , ).

Key words
 Adjoint equation,
 Dirac delta function,
 Green’s function,
 Green’s theorem

Self Assessment
1. Which of the following statement is correct?
A. The necessary and sufficient condition that the second order homogeneous linear
differential equation + + = 0 as defined in (1) on ≤ ≤ to be self
adjoint is that ( ) = ( ) on ≤ ≤ .
B. The necessary and sufficient condition that the second order homogeneous linear
differential equation + + = 0 as defined in (1) on ≤ ≤ to be self
adjoint is that ( ) = ( ) on ≤ ≤ .
C. The necessary and sufficient condition that the second order homogeneous linear
differential equation + + = 0 as defined in (1) on ≤ ≤ to be self
adjoint is that ( ) = ( ) on ≤ ≤ .
D. The necessary and sufficient condition that the second order homogeneous linear
differential equation + + = 0 as defined in (1) on ≤ ≤ to be self
adjoint is that ( ) = ( ) on ≤ ≤ .

2. If the BVP (1) to (4) as in summary, is self adjoint then Green’s function is
A. symmetric
B. Hermitian
C. anti-symmetric
D. skew symmetric

, | |<
3. ( )=
0, | | >

and ∫ ( ) =∫ = 1, then

78 Lovely Professional University

Notes

Calculus of Variation and Integral Equation

A. lim → ( ) = ′( ) is the Dirac delta function

B. ( ) = ( ) is the Dirac delta function
C. lim → ( ) = ( ) is the Dirac delta function
D. → ( ) = ( ) is the Dirac delta function

4. The expression − =− ( )
means

A. ( − 1) derivative of ( , ) with respect to at the point = has discontinuity of

the first kind
B. ( − 1) derivative of ( , ) with respect to at the point = has discontinuity of
the second kind
C. ( − 1) derivative of ( , ) with respect to at the point = has discontinuity of
the third kind
D. ( − 1) derivative of ( , ) with respect to at the point = has no discontinuity

5. Which of the following functions represents a pulse produced by sudden electrical

discharge?
A. Green’s function
B. Self adjoint function
C. Gauss function
D. Dirac delta function

Answers for Self Assessment

1. B 2. A 3. C 4. A 5. D

Review Questions
1. What are the conditions needed for construction of Green’s function?

2. Construct a Green’s function for the homogeneous boundary value problem:

= 0, (0) = (0) = (1) = (1) = 0

3. Using Green’s function solve + = , (0) = =0

4. Reduce the BVP + = , (0) = 0, (1) = 0 to a Fredholm integral equation.

5. Discuss the Green’s function for a linear homogeneous differential equation of second
order.

Further Readings
1. I. M. Gelfand,S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc.
(1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley
(1964)

Lovely Professional University 79

 Notes

Unit 08: Green’s Functions

3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press

(1947)

Web Links
1. https://www.youtube.com/watch?v=C2ZLHf1B_Jc
2. https://www.youtube.com/watch?v=3GJDXCgVM98

80 Lovely Professional University

 Notes

Pratiksha, Lovely Professional University Unit 09: Introduction to Calculus of Variation

Unit 09: Introduction to Calculus of Variation

CONTENTS
Objectives
Introduction
9.1 Functionals
9.2 Localization property of Functionals
9.3 Function Spaces
9.4 Motivating problems of calculus of variations
9.5 Continuity of a Functional
9.6 The variation of a functional. A Necessary Condition for an Extremum
9.7 Variation of a functional (Differential of a functional)
Summary
Key words
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to
 understand the nature of problems of calculus of variation
 define the function space
 derive the properties of the differentiable functional

Introduction
The calculus of variations was created as a tool to calculate minimizers (or maximizers) of certain
mathematical quantities taking values in the set of real numbers.
In the 18th century mathematicians and physicists, such as Maupertuis, d’Alembert, Euler, and
Lagrange, postulated variational principles stating that nature acts with minimal expenditure
(“principle of least action”).
The notion “calculus of variations” goes back to the year 1744 and was introduced by Euler: For
externals of a function, the derivative vanishes, and in the 18 th century that derivative was called
the “first variation.”
The “mathematical quantities” involved in this topic are not ranges of mappings from into , i.e.
the domain of definition of the mappings is infinite-dimensional. And such mappings are called
“functional”.

9.1 Functionals
Variable quantities called functionals play an important role in many problems arising in analysis,
mechanics, geometry, etc.

Lovely Professional University 81

Notes

Calculus of Variation and Integral Equation

By a functional, we mean a correspondence which assigns a definite (real) number to each function
(or curve) belonging to some class.
Thus, a functional is a kind of function, where the independent variable is itself a function (or
curve).

Example: Consider the set of all rectifiable plane curves. A definite number is associated with
each such curve, namely, its length. Thus, the length of a curve is a functional defined on the set of
rectifiable curves.

Example: Suppose that each rectifiable plane curve is regarded as being made out of some
homogeneous material. Then if we associate with each such curve the ordinate of its center of mass,
we again obtain a functional.

Example: Consider all possible paths joining two given points A and B in the plane. Suppose
that a particle can move along any of these paths, and let the particle have a definite velocity ( , )
at the point ( , ). Then we obtain a functional by associating with each path, the time the particle
takes to traverse the path.

Example: Let ( ) be an arbitrary continuously differentiable function, defined on the interval

[ , ]. Then the formula

[ ]= ( )

Defines a functional on the set of all such functions ( ).

Example: As a more general example, let ( , , ) be a continuous function of three variables.

Then the expression

[ ]=∫ [ , ( ), ( )] (1)

Where ( ) ranges over the set of all continuously differentiable functions defined on the interval
[ , ], defines a functional.
By choosing different functions ( , , ), we obtain different functionals.

9.2 Localization property of Functionals

Functionals of type ∫ ( , , ) have a localization property consisting of the fact that if we
divide the curve = ( ) into parts and calculate the value of the functional for each part, the sum
of the values of the functional for the separate parts equals the value of the functional for the whole
curve.
It is just these functionals which are usually considered in the calculus of variations.

e.g. Consider (i) [ ]=∫ (1 + )

∫
(ii) [ ]=
∫

82 Lovely Professional University

 Notes

Unit 09: Introduction to Calculus of Variation

9.3 Function Spaces

In the study of functions of n variables, it is convenient to use geometric language, by regarding a
set of n numbers ( , , … , ) as a point in an - dimensional space. In just the same way,
geometric language is useful when studying functional. Thus, we shall regard each function ( )
belonging tosome class as a point in some space, and spaces whose elements are functions will be
called function spaces.
The concept of continuity plays an important role for functional, just as it does for the ordinary
functions considered in classical analysis.
In order to formulate this concept for functional, we must introduce a concept of " closeness" for
elements in a function space. This is most conveniently done by introducing the concept of the
norm of a function, analogous to the concept of the distance between a point in Euclidean space and
the origin of coordinates.
Although in what follows we shall always be concerned with function spaces, it will be most
convenient to introduce the concept of a norm in a more general and abstract form, by introducing
the concept of a normed linear space.
The elements of a normed linear space can be objects of any kind, e.g., numbers, vectors (directed
line segments), matrices, functions, etc.
The following normed linear spaces are important for our subsequent purposes:
The space ( , ), consists of all continuous functions ( ) defined on a (closed) interval[ , ]. By
addition of elements of ( , ) and multiplication of elements of ( , ) by numbers, we mean
ordinary addition of functions and multiplication of functions by numbers, while the norm is
defined as the maximum of the absolute value, i.e.
| | = | ( )|

Thus, in the space ( , ), the distance between the function ∗ ( ) and the function ( ) does not
exceed if the graph of the function ∗ ( ) lies inside a strip of width 2 (in the vertical direction)
“bordering" the graph of the function ( ).

The space ( , ) consists of all functions ( ) defined on an interval [ , ] which are continuous
and have continuous first derivatives. The operations of addition and multiplication by numbers
are the same as in ( , ) , but the norm is defined by the formula
| | = | ( )| + | ′( )|

The space ( , ) consists of all functions ( ) defined on an interval [ , ] which are continuous
and have continuous derivatives up to order, where is a fixed integer. The operations of
addition and multiplication by numbers are the same as in ( , ) , but the norm is defined by the
formula

| | ( )
= | ( )|

Lovely Professional University 83

Notes

Calculus of Variation and Integral Equation

9.4 Motivating problems of calculus of variations
Some historic examples make us curious and lead to a set of problems which are known as
problems of calculus of variations. Now the theory has developed on these problems and there is a
set procedure, though historically these problems were seemingly solved by common sense.

1. The isoperimetric problem of queen Dido (9th century BC)

When the Phoenician Princess Dido had fled to North Africa, she asked the Berber king for a bit of
land. She was offered as much land as could be encompassed by an oxhide.
Dido cut the oxhide into fine stripes so that she got a long enough rope to encircle an entire hill
named Byrsa, the center of Carthage.
What shape was the region that she encircled so that area was maximal?

2. What is the shortest distance between two points?

This question can be answered only if an admitted connection is defined. It must be a continuous
curve that has a well-defined length. Moreover, it must be determined where that curve runs: in a
plane, in a space, on a sphere, on a manifold?
The variables of this problem are admitted curves that connect two fixed points, and the real
quantities to be minimized are their lengths.
Admitted curves cannot be determined by finitely many real variables, but they form a subset of an
infinite dimensional function space.

3. The brachistochrone problem

It was posed by Johann Bernoulli (1667–1748). In June 1696, he introduced the following
problem in the “Acta Eruditorum” (a Journal for Science published in Leipzig, Germany):
Given two points A and B in a vertical plane, what is the curve traced out by a point mass
M acted on only by gravity, which starts at A and reaches B in the shortest time?

84 Lovely Professional University

 Notes

Unit 09: Introduction to Calculus of Variation

9.5 Continuity of a Functional

After a norm has been introduced in the linear space (which may be a function space), we can
discuss about the continuity of functionals defined on .
The functional [ ] is said to be continuous at the point ∈ if for any > 0, there is a > 0 such
that | [ ] − [ ]| < provided that ‖ − ‖ < .
Note: In many variational problems, we have to deal with functional defined on sets of functions
which do not form linear spaces.
In fact, the set of functions (or curves) satisfying the constraints of a given variational problem,
called the admissible functions (or admissible curves), is in general not a linear space.

9.6 The variation of a functional. A Necessary Condition for an

Extremum
The concept of the variation (or differential) of a functional is analogous to the concept of the
differential of a function of n variables. This concept is used to find extrema of functional.
Linear functional
Given a normed linear space let each element ℎ ∈ be assigned a number [ℎ], i.e., let [ℎ] be a
functional defined on . Then [ℎ] is said to be a (continuous) linear functional if
1. [ ℎ] = [ℎ] for any ℎ ∈ and any real number ;
2. [ℎ + ℎ ] = [ℎ ] + [ℎ ] for any ℎ , ℎ ∈ ;
3. [ℎ] is continuous ∀ ℎ ∈ .

Example: The integral [ℎ] = ∫ ℎ( ) defines a linear functional on ( , ).

The integral [ℎ] = ∫ ( )ℎ ) defines a linear functional on ( , ) where ( ) is a fixed

function in ( , ).

The integral [ℎ] = ∫ { ( )ℎ( ) + ( )ℎ ( ) + ⋯ + ( )ℎ( ) ( )} defines a linear functional

on ( , ) where ( ) are a fixed function in ( , ).

Lemma1: If ( )is continuous in [ , ] and if [ℎ] = ∫ ( )ℎ( ) = 0 for every function ℎ( ) ∈

( , ) such that ℎ( ) = ℎ( ) = 0, then ( ) = 0 for all ∈ [ , ].
Let ( ) > 0 at some point in [ , ].
Then ( ) > 0 at some point [ , ] ⊂ [ , ].
( − )( − ), ∈ [ , ]
Set ℎ( ) = and ℎ( ) satisfies the conditions of the lemma.
0, ℎ
However,

( ) ℎ( ) = ( )( − )( − ) >0

Lovely Professional University 85

Notes

Calculus of Variation and Integral Equation

∵ ( )( − )( − ) > 0 except at and
Which is a contradiction.
∴ ( )=0∀ ∈[ , ]

Lemma2: If ( ) is continuous in [ , ] and if [ℎ] = ∫ ( )ℎ( ) = 0 for every function ℎ( ) ∈

( , ) such that ℎ( ) = ℎ( ) = 0, then ( ) = for all ∈ [ , ], where is a constant.

Let be defined as in ∫ [ ( ) − ] =0

And let ℎ( ) = ∫ [ ( ) − ]

Thus ℎ( ) ∈ ( , ) and satisfies ℎ( ) = ℎ( ) = 0.

Then ∫ [ ( ) − ] ℎ ( ) =∫ ( )ℎ ( ) − [ℎ( ) − ℎ( )] = 0

Also

[ ( ) − ]ℎ ( ) = [ ( )− ]

⇒ ( )− =0
⇒ ( )= ∀ ∈[ , ]

Lemma3: If ( )is continuous in [ , ] and if [ℎ] = ∫ ( )ℎ′( ) = 0 for every function

ℎ( ) ∈ ( , ) such that ℎ( ) = ℎ( ) = 0 and ℎ′( ) = ℎ′( ) = 0 then ( ) = + for all
∈ [ , ], where , is a constants.

Lemma 4: If ( ) and ( ) are continuous in [ , ] and if [ℎ] = ∫ { ( )ℎ( ) + ( )ℎ′( )} = 0

for every function ℎ( ) ∈ ( , ) such that ℎ( ) = ℎ( ) = 0 then ( ) is differentiable and
( ) = ( ) for all ∈ [ , ].

9.7 Variation of a functional (Differential of a functional)

Let [ ]be a functional defined on some normed linear space and letΔ [ℎ] = [ + ℎ] − [ ] be its
increment, corresponding to the increment ℎ = ℎ( ) of the independent variable = ( ).
If is fixed, Δ [ℎ] is a functional of ℎ, in general a non linear one.
Let Δ [ℎ] = [ℎ] + |ℎ| , where [ℎ] is a linear functional and → 0 as |ℎ| → 0. Then, the
functional [ ] is said to be differentiable and the principal linear part of the increment Δ [ℎ]
i.e. [ℎ] is called the variation of [ ] and is denoted as [ℎ].

Theorem 1: The differential of a differentiable functional is unique.

[ ]
We know that if [ℎ] is a linear functional and if | |
→ 0 as |ℎ| → 0

Then [ℎ] ≡ 0∀ ℎ
[ ]
Suppose [ℎ ] ≠ 0 for some ℎ ≠ 0. Then setting ℎ = , = | |
.

We see |ℎ | → 0 as →∞
But
[ℎ ] [ℎ ]
lim = lim = ≠0
→ |ℎ | → |ℎ |

86 Lovely Professional University

 Notes

Unit 09: Introduction to Calculus of Variation

Which is contradiction.
Now let [ ] is not uniquely defined, then
Δ [ℎ] = [ℎ] + |ℎ|

Δ [ℎ] = [ℎ] + |ℎ|

Where [ℎ] and [ℎ] are linear functional and , → 0 as ‖ℎ‖ → 0

⇒ [ℎ] − [ℎ] = ( − )‖ℎ‖ = ‖ℎ‖ (say)
Where [ℎ] − [ℎ] is an infinitesimal of order higher than one relative to ‖ℎ‖.

But ∵ [ℎ] − [ℎ] is a linear functional

∴ [ℎ] − [ℎ] ≡ 0
[ℎ] = [ℎ]

∴ [ ] is uniquely defined.

Theorem 2: A necessary condition for the differentiable functional [ ] to have an extremum for
= i.e. [ℎ] = 0 for = and all admissible ℎ.
Let [ℎ] has a minimum for = ,
By definition of variation,
Δ [ℎ] = [ℎ] + ‖ℎ‖ (1)
Where → 0 as ‖ℎ‖ → 0
∴ for sufficiently small ‖ℎ‖
The sign of Δ [ℎ] and of [ℎ] will be same.

Let [ℎ ] ≠ 0 for some ℎ , then for any > 0, however small ,, we have
[− ℎ ] = − [ ℎ ]
(1) can be made to have either sign for arbitrarily small ‖ℎ‖.
But this is impossible ∵ [ ] is supposed to have a minimum for =

i.e.
Δ [ℎ] = [ + ℎ] − [ ] ≥ 0
For all sufficiently small ‖ℎ‖, which is a contradiction.
Hence the theorem gets proved.
Having understood the definition of functional, we can now proceed to solve the problems on
calculus of variation.

Summary
 The calculus of variations was created as a tool to calculate minimizers (or maximizers) of
certain mathematical quantities taking values in the set of real numbers.
 Functionals of type ∫ ( , , ) have a localization property consisting of the fact that
if we divide the curve = ( ) into parts and calculate the value of the functional for each

Lovely Professional University 87

Notes

Calculus of Variation and Integral Equation

part, the sum of the values of the functional for the separate parts equals the value of the
functional for the whole curve.
 The functional [ ] is said to be continuous at the point ∈ if for any > 0, there is a
> 0 such that | [ ] − [ ]| < provided that ‖ − ‖< .
 Let Δ [ℎ] = [ℎ] + |ℎ| , where [ℎ] is a linear functional and → 0 as |ℎ| → 0. Then, the
functional [ ] is said to be differentiable and the principal linear part of the increment
Δ [ℎ] i.e. [ℎ] is called the variation of [ ] and is denoted as [ℎ].
 The differential of a differentiable functional is unique.
 A necessary condition for the differentiable functional [ ] to have an extremum for =
i.e. [ℎ] = 0 for = and all admissible ℎ.

Key words
Dido’s problem, function space, functional, differentiable functional

Self Assessment
1. Which of the following statement is correct?

A. By a functional, we mean a correspondence which assigns a definite number to each

function (or curve) belonging to some class.
B. By a functional, we mean a correspondence which assigns a real number to each
function (or curve) belonging to some class.
C. Both (a) and (b)
D. None of these

2. The integral [ℎ] = ∫ ℎ( ) defines a

A. linear functional on ( , )
B. non-linear functional on ( , )
C. linear functional on ( , )
D. nonlinear functional on ( , )

3. A necessary condition for the differentiable functional [ ] to have an extremum for =

is
A. [ℎ] = 0
B. [ℎ] = 0 for =
C. [ℎ] = 0 for = and all non-admissible ℎ
D. [ℎ] = 0 for = and all admissible ℎ

4. Which of the following is correct?

I. The differential of a differentiable functional is unique.
II. A necessary condition for the differentiable functional [ ] to have an extremum
for = i.e. [ℎ] = 0 for = and all admissible ℎ.
A. Only I
B. Only II
C. Both I and II
D. None of these

88 Lovely Professional University

 Notes

Unit 09: Introduction to Calculus of Variation

5. If ( ) is continuous in [ , ] and if [ℎ] = ∫ ( )ℎ( ) =0 for every function

ℎ( ) ∈ ( , ) such that ℎ( ) = ℎ( ) = 0, then
A. ( )= for all ∈ [ , ], where is a constant
B. ( ) = 0 for all ∈[ , ]
C. ( ) ≠ 0 for all ∈[ , ]
D. None of these

Answers for Self Assessment

1. C 2. A 3. D 4. C 5. A

Review Questions
1. What are the motivating problems of calculus of variations?

2. Explain Dido’s problem. How would you have solved that?

3. Prove, “If ( )is continuous in [ , ] and if [ℎ] = ∫ ( )ℎ′( ) =0 for every

function ℎ( ) ∈ ( , ) such that ℎ( ) = ℎ( ) = 0 and ℎ′( ) = ℎ′( ) = 0 then

( )= + for all ∈ [ , ], where , is a constants.”

4. Prove, “If ( ) and ( ) are continuous in [ , ] and if [ℎ] = ∫ { ( )ℎ( ) +

( )ℎ′( ) =0 for every function ℎ( )∈ 1( , ) such that ℎ( )=ℎ( )=0 then ( ) is

differentiable and ( ) = ( ) for all ∈ [ , ].”

5. Write a note on variation of a functional along with its properties.

Further Readings
1. I. M. Gelfand,S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

Web Links
https://mathoverflow.net/questions/109717/functional-variational-derivative-and-the-
leibniz-rule
https://mathworld.wolfram.com/CalculusofVariations.html

Lovely Professional University 89

 Notes
Pratiksha, Lovely Professional University Unit 10: Euler’s Equation

Unit 10: Euler’s Equation

Objectives
Students will be able to

 analyze the necessary and sufficient condition for the extremum of a functional
 remember the Euler’s equation for various forms of functional
 create the Euler’s equation to several variables
 understand the problem of shortest distance
 analyze the invariance of the Euler’s equations

Introduction
In this chapter, we will give necessary conditions for an extremum of a function of the type
[ ] = ∫ ( , , ) with various types of boundary conditions.

The necessary condition is in the form of adifferentialequation that the extremal curve should
satisfy, and this differential equation is called the Euler-Lagrange equation.
Consider the case where the curves are allowed to varybetweentwo fixed points

12.1 Euler's equation for one dependent function

The simplest variational problem
Let ( , , ) be a function with continuous first and second (partial) derivatives with respect to all
its arguments. Then, among all functions ( ) which are continuously differentiable for ≤ ≤
and satisfy the boundary conditions ( ) = , ( ) = , find the function for which the functional
[ ] = ∫ ( , , ) has a weak extremum.

The functional [ ] has a (relative) extremum for = if [ ] − [ ] does not change its sign in
some neighbourhood of the curve = ( ).
Subsequently, we shall be concerned with functionals defined on some set of continuously
differentiable functions, and the functions themselves can be regarded either as elements of the
space or elements of the space . Corresponding to these two possibilities, we can define two
kinds of extrema:

The functional [ ] has a weak extremum for = if there exists an > 0 such that [ ] − [ ] has
the same sign for all y in the domain of definition of the functional which satisfy the condition
‖ − ‖ < where‖ ‖ is the norm in the space .
And
The functional [ ] has a weak extremum for = if there exists an > 0 such that [ ] − [ ] has
the same sign for all y in the domain of definition of the functional which satisfy the condition
‖ − ‖ < where‖ ‖ is the norm in the space .

Clearly every strong extremum is simultaneously a weak extremum.

As a rule, finding a weak extremum is simpler than finding a strong extremum as the functionals
usually considered in the calculus of variations are continuous in the norm of the space .

90 Lovely Professional University

Notes
Calculus of Variation and Integral Equation
Let ( , , ) be a function with continuous first and second (partial) derivatives with respect to all
its arguments. Then, among all functions ( ) which are continuously differentiable for ≤ ≤
and satisfy the boundary conditions ( ) = , ( ) = , find the function for which the functional
[ ]=∫ ( , , ) has a weak extremum.

Suppose we give an increment of ℎ( ) to ( ), such that ℎ( ) = ℎ( ) = 0.

The corresponding increment in the functional [ ] = ∫ ( , , ) is given as

= [ + ℎ] − [ ] = ( , + ℎ, +ℎ ) − ( , , )

= [ ( , + ℎ, + ℎ′ ) − ( , , )]

Using the Taylor's theorem

=∫ [ ( , , )ℎ + ( , , ) ℎ′ ] + ⋯(Terms of order higher than relative to ℎ and ℎ′)

The variation of [ ] is

= [ ( , , )ℎ + ( , , ) ℎ′ ]

We know that a necessary condition for [ ] to have an extremum for = ( ) is that

= [ ( , , )ℎ + ( , , ) ℎ′ ] =0

for all admissible ℎ.

And Lemma4 implies that =

Or

− =0

i.e. if = ( , , ′), then − =0

which is known as Euler's equation.

Therefore, we have proved

Theorem 1. Let [ ] be a functional of the form∫ ( , , ) , 〗defined on the set of functions ( )

which have the continuous first derivatives in[ , ] and satisfy the boundary conditions ( ) = ,
( ) = . Then a necessary condition for [ ] to have an extremum for a given function ( ) is that
( ) satisfy Euler's equation − =0

The integral curves of Euler's equation are called extremals. Since Euler's equation is a second-order
differential equation, its solution will in general depend on two arbitrary constants, which are
determined from the boundary conditions ( ) = , ( ) = .

Lovely Professional University 91

 Notes
Unit 10: Euler’s Equation
Special cases:

1. When the integrand does not depend on i.e. the functional has the form ∫ ( , )

Euler’s equation becomes

=0

Or

= ⇒ = ( , )
And then can be calculated.

2. When the integrand does not depend on i.e. if the functional is of the form ∫ ( , ′)

then

− = − +

Or

− = − − ′′ (1)

Multiply (1) by ′

′ − = − − ′′ ′

= ( − )

Using Euler’s equation

− =0

⇒ − =

3. When does not depend on , Euler’s equation takes the form

=0

which is a finite equation, whose solution consists of one or more curves = ( ).

Let [ ] = ∫ , (1) = 0 and (2) = 1.

Here integrand =

i.e.

( , )= (1)

Euler’s equation becomes

92 Lovely Professional University

Notes
Calculus of Variation and Integral Equation
1 2
⇒ × =
2 1+
Or

=
1+
Squaring both sides

= (1 + )

− =

Or
1
− =

Or

=
1−
Or

=
√1 −
Or

=
√1 −
Separating the variables and integrating

=
√1 −
= − √1 − + (*)

Or
1
− =− 1−

Squaring both sides

1−
( − ) =

Or

+( − ) = (**)

Using condition

(1) = 0 ⇒ 1 + = (2)

Also

(2) = 1 ⇒ 4 + (1 − ) = (3)

Subtracting (2) and (3)

1 − 4 = (1 − ) − =1+ −2 −
Or
−4 = −2 ⇒ =2

Lovely Professional University 93

 Notes
Unit 10: Euler’s Equation
Also from eq.(2)
1 1
1+4= ⇒ =
5
∴ from eq.(**)
+ ( − 2) = 5
Is the required curve

12.2 Generalization of Euler's equation to several variables

For the functional ( , , ′) the Euler’s equation is

− =0

depends upon functions of one variable i.e. on curves.

Let ( , , , , ) be a function with continuous first and second partial derivatives w.r.t all its
arguments and consider the functional of the form

[ ]= , , , ,

Where is some closed region and

= =

= =

Are the partial derivatives of = ( , )

The Euler’s equation takes the form

− − =0

Or

− − =0

[ ]=∬ ( + ) , here = +

Euler’s equation:

− − =0

0− (2 ) − (2 ) = 0

⇒2 +2 =0

Or

+ =0

Which is Laplace equation.

94 Lovely Professional University

Notes
Calculus of Variation and Integral Equation
 For = ( , ,…, ) and = , ,…, , , ,…,
Euler Ostrogradsky equation is

− =0

Where = .

Let [ ] = ∫ ( , , be a functional defined on functions which are differentiable four

, ′′)
times and satisfy the boundary conditions ( ) = , ( ) = , ( ) = , ( ) =

If the functional [ ] has an extremum on the function ( ), then ( ) satisfies the differential
equation

− + =0

Which is Euler Poisson equation.

[ ] = ∫ (16 − + ) , Here = 16 − +

Euler Poisson equation,

− + =0

⇒ 32 − (0) + (−2 )=0

Or
32 − 2 =0 (1)
Or
− 16 = 0

− 16 = 0, where =

Auxiliary equation,
− 16 = 0
⇒ = ±2, ±2
∴ = + + sin 2 + cos 2

 Necessary condition of the functional of the form

[ ]= ( , , ,…, , , ,…, )

to have extremum is that

− = 0, = 1,2, … ,

Lovely Professional University 95

 Notes
Unit 10: Euler’s Equation

Show that ∫ 2 + + such that (0) = 0, (0) = 0, = −1 and =1

is stationary for = − sin and = sin

Here ( , , , , )=2 + + (1)

The Euler’s equations are

− = 0 and − =0

⇒2 − (2 ) = 0 and 2 − (2 ) = 0

Or
− ′′=0 and − =0
Or
= ⇒( ) =
Symbolically,

( − 1) = 0, here =

Auxiliary equation,
−1=0
Or
= ±1, ±
∴ = + + cos + sin
=− + − sin + cos
And
= + − cos − sin
Using (0) = 0 ⇒ (0) = 0 ⇒ + − =0 (2)
(0) = 0 ⇒ + + =0 (3)

= −1 ⇒ = −1 ⇒ + − = −1 (4)

=1⇒ + + =1 (5)

On solving (2), (3), (4) and (5), we get

= 0, = 0, = 0 and =1
∴ = − sin and = sin

12.3 Problem of shortest distance

Given two points in a plane, what is the shortest path between them?
We certainly know the answer: Straight line. Let’s prove it using the variation method.
Consider an arbitrary path ( ).
Let be the distance function and elementary length
be .

then,

96 Lovely Professional University

Notes
Calculus of Variation and Integral Equation

= ( ) +( ) = 1+

= 1+

Total path length = ∫ =∫ 1+

Necessary condition for the integral to be stationary (extremum) is

− =0 (1)

Here ( , , )= 1+
∴ from (1)
1
0− (1 + ) .2 =0
2
Or

=
1+
⇒ = (1 + )
Or
(1 − )=

⇒ = = (say)

Or

= √ = (say)
Or

= ⇒ =

Integrating both sides,

= +
Which is the equation of straight line.
Given two points on sphere, what is the shortest path between them? You will be able to answer
this obvious question soon enough. But let us solve some problems of finding the extremal using
Euler Lagrange equation.

Find the extremals of ∫ .

Here =

By Euler’s equation

− =0

Or
2
= ⇒ =

Or

Lovely Professional University 97

 Notes
Unit 10: Euler’s Equation

=
2
Or

=
2

⇒ =
2
Integrating both sides, we get

= . +
2 4
Or

= + , where =

Find the extremals of ∫ ( + +2 )

Here ( , , )= + +2

Using Euler’s equation

− =0

⇒2 +2 − (2 ) = 0

Or
2 +2 −2 =0
Or
− − =0
Or
− =
Symbolic form is

( − 1) = , here ≡

Auxiliary equation,
−1=0
⇒ = ±1
∴ . .= +
For particular Integral, P.I.

P.I.= = ^

=
2
∴ = . .+ . .

= + +
2

98 Lovely Professional University

Notes
Calculus of Variation and Integral Equation
12.4 Variational Derivative
The concept of the variational (or functional) derivative, plays the same role for functionals as the
concept of the partial derivative plays for functions of n variables.
We begin by considering functionals of the type

[ ]= ( , , ^′ ) , ( )= , ( )=

corresponding to the simplest variational problem. Our approach is to first go from the variational
problem to an n-dimensional problem, and then pass to the limit → ∞ .
Thus, we divide the interval [ , ] into + 1 equal subintervals by introducing the points
= , ,… , = ,( − = Δ ),
and we replace the smooth function ( ) by the polygonal line with vertices
( , ), ( , ), … , ( , ), ( , ),
Where = ( )

Then

[ ]=∫ ( , , ) , ( )= , ( )= ,

can be approximated by the sum

( , ,…, )=∑ , , Δ ,

Which is a function of variables.

Next, we calculate the partial derivatives and we consider what happens to these derivatives as the
number of points of subdivision increases without limit.

= , , Δ + , , Δ − , , Δ .

Or

= , , Δ + , , Δ − , , .

We note that the expression appearing in the denominator on the left has a direct geometric
meaning, and is in fact just the area of the region lying between the solid and the dashed curves in
Figure

As → 0 the expression above converges to the limit

≡ ( , , )− ( , , ′),

called the variational derivative of the functional.

We see that the variational derivative is just the left-hand side of Euler's equation and hence the
meaning of Euler's equation is just that the variational derivative of the functional under
consideration should vanish at every point.
This is the analogue of the situation encountered in elementary analysis, where a necessary
condition for a function of n variables to have an extremum is that all its partial derivatives vanish.

Lovely Professional University 99

 Notes
Unit 10: Euler’s Equation
In the general case, the variational derivative is defined as follows : Let [ ] be a functional
depending on the function ( ), and suppose we give ( ) an increment ℎ( ) which is different
from zero only in the neighbourhood of a point . Dividing the corresponding increment [ +
ℎ] − [ ] of the functional by the area lying between the curve = ℎ( ) and the − , we
obtain the ratio
[ + ℎ] − [ ]
Δ

Next, we let the area go to zero in such a way that both max |ℎ( )| and the length of the interval
in which ℎ( ) is non vanishing go to zero. Then,
if the ratio
[ + ℎ] − [ ]
Δ
converges to a limit as → 0, this limit is called the variational derivative of the functional [ ] at
the point [for the curve = ( )], and is denoted by

It can be shown that the analogues of all the familiar rules obeyed by ordinary derivatives (e.g., the
formulas for differentiating sums and products of functions, composite functions, etc.) are valid for
variational derivatives.

12.5 Invariance of Euler's Equation

Suppose that instead of the rectangular plane coordinates and , we introduce curvilinear
coordinates and , where

= ( , ), = ( , ), ≠ 0.

Then the curve given by the equation = ( ) in the -plane corresponds to the curve given by
some equation = ( ) in the - plane. When we make the change of variables, the functional

[ ]= ( , , )

goes into the functional

+
[ ]= ( , ), ( , ), ( + )
+

=∫ ( , , ) ,

Where
+
( , , )= ( , ), ( , ), ( + ).
+

100 Lovely Professional University

Notes
Calculus of Variation and Integral Equation
We now show that = ( )satisfies the Euler equation

− =0

corresponding to the original functional [ ], then = ( ) satisfies the Euler equation

− =0

corresponding to the new functional.

Let denote the area bounded by the curves = ( ) and = ( ) + ℎ( ), and let denote
the area bounded by the corresponding curves = ( ) and = ( ) + ( ) in the -plane. By
the standard formula for the transformation of areas, the limit as → 0, → 0 of the ratio
approaches the non zero Jacobian

which by hypothesis is nonzero. Thus, if

[ + ℎ] − [ ]
lim = 0,
→ Δ
then
[ + ]− [ ]
lim =0
→ Δ

It follows that ( ) satisfies

− =0

If ( ) satisifies

− =0

In other words, whether a curve is an extremal is a property which is independent of the choice of
the coordinate system.
In solving Euler's equation, changes of variables can often be used to advantage. Because of the
invariance property just proved, the change of variables can be made directly in the integral
representing the functional rather than in Euler's equation, and we can then write Euler's equation
for the new integral.

Summary
 Every strong extremum is simultaneously a weak extremum.
 For the functional = ( , , ′), the necessary condition for it to have an extremum is

− = 0.

 When the integrand does not depend on i.e. the functional has the form ∫ ( , )

Euler’s equation becomes

= ( , )
 When the integrand does not depend on i.e. if the functional is of the form ∫ ( , ′)

then − =

 When does not depend on , Euler’s equation takes the form

=0

 For several variables the Euler’s equation takes the form

Lovely Professional University 101

 Notes
Unit 10: Euler’s Equation
 − − =0

 The Euler Poisson equation is

− + =0

 Euler Ostrogradsky equation is

− =0

Key words
Functional, extremal, Euler Lagrange equation, Euler Poisson equation, Euler Ostrogradsky
equation

Self Assessment
1. Which of the following statement is correct?
(a) Every strong extremum is simultaneously a weak extremum.
(b) Every weak extremum is simultaneously a strong extremum.
(c) Strong extremum can never be weak.
(d) None of these
2. For the integrand of the functional = ( , , ′), the necessary condition for it to have an
extremum is

(a) − =0

(b) + =0

(c) − =0

(d) − =0

3. When the integrand of the functional does not depend on , the Euler’s equation takes the
form

(a) − =0

(b) = ( , )

(c) − =0

(d) − =0

4. When the integrand of the functional does not depend on , Euler’s equation takes the
form
(a) = ( , )
(b) = ( , )
(c) = ( , )
(d) of a finite equation

5. A curve is an extremal is a property which

(a) is independent of the choice of the coordinate system.
(b) is dependent of the choice of the coordinate system.
(c) Both statements can be correct
(d) None of the statements is correct
6. The Euler Poisson equation is

102 Lovely Professional University

Notes
Calculus of Variation and Integral Equation
(a) − − =0

(b) −∑ =0

(c) − + =0

(d) None of these

7. The Euler Ostrogradsky equation is

(a) − − =0

(b) −∑ =0

(c) − + =0

(d) None of these

8. The extremals of ∫

(a) = +
(b) = +
(c) = +
(d) = +

9. The concept of the ______________ plays the same role for functionals as the concept of the
partial derivative plays for functions of n variables.
(a) variation
(b) variational derivative
(c) extremal
(d) strong extremum
10. The Euler equation for the extremals of ∫ ( + +2 ) results in the differential

equation:

(a) + =
(b) − =
(c) + =
(d) − =
Answers:

a
1
c
2
b
3
d
4
a
5
c
6
b
7
a
8
b
9
d
10

Lovely Professional University 103

 Notes
Unit 10: Euler’s Equation

Review Questions
1. What is the necessary condition for the extremal of a functional?
2. Find the curve that gives the minimum distance between two points on a plane
surface?
3. Discuss the weak extremum and the strong extremum.
4. What are the various forms of Euler’s equation when the functional takes different
forms?
5. Discuss the invariance of Euler’s equation.
6. Define the variational derivative.

Further Readings
1. I. M. Gelfand, S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

Web Links
https://mathworld.wolfram.com/FunctionalDerivative.html

http://liberzon.csl.illinois.edu/teaching/cvoc/node26.html

https://www.ucl.ac.uk/~ucahmto/latex_html/chapter2_latex2html/node5.html

104 Lovely Professional University

 Notes

Unit 11: Problem of Brachistochrone and Minimum Surface of Revolution

Pratiksha, Lovely Professional University

Unit 11: Problem of Brachistochrone and Minimum Surface of

Revolution
CONTENTS
Objectives
Introduction
11.1 Brachistochrone Problem
11.2 Problem of Minimum Surface of Revolution
Summary
Key words
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to
 define the brachistochrone
 solve the brachistochrone problem
 visualize the surfaces of revolution
 solve the problem of minimum surface of revolution

Introduction
Many problems involve finding a function that maximizes or minimizes an integral expression.
One example is finding the curve giving the shortest distance between two points — a straight line,
in Cartesian geometry and a difficult one would be if the two points lie on a curved surface. The
mathematical techniques developed to solve this type of problem are collectively known as the
calculus of variations. In this unit we will learn about two famous problems of calculus of
variations, namely, the brachistochrone problem and the problem of minimum surface of
revolution.

11.1 Brachistochrone Problem

The brachistochrone problem was posed by Johann Bernoulli in Acta Eruditorumin June 1696. He
introduced the problem as follows:
I, Johann Bernoulli, address the most brilliant mathematicians in the world. Nothing is more
attractive to intelligent people than an honest, challenging problem, whose possible solution will
bestow fame and remain as a lasting monument. Following the example set by Pascal, Fermat etc., I
hope to gain the gratitude of the whole scientific community by placing before the finest
mathematicians of our time a problem which will test their methods and the strength of their
intellect. If someone communicates to me the solution of the proposed problem, I shall publicly
declare him worthy of praise.
The problem he posed was the following:
Given two points A and B in a vertical plane, what is the curve traced out by a point acted on only
by gravity, which starts at A and reaches B in the shortest time.

Lovely Professional University 105

Notes

Calculus of Variation and Integral Equation

Let A and B be two fixed points. Then the time it takes a particle to slide under the influence of
gravity along some path joining A and B depends on the choice of the path (curve), and hence is
functional.
The curve such that the particle takes the least time to go from A to B is called the brachistochrone.
The word brachistochrone is from the Greek for "shortest time".The problem was solved by John
Bernoulli, James Bernoulli, Newton, and L'Hospital.
The brachistochrone turns out to be a cycloid, lying in the vertical plane and passing through A and
B. (The cycloid is the locus of a point on the rim of a circle of radius 'a' rolling along a straight line.)

We'll choose a coordinate system with the origin at point A and the y axis directed downwards.
Points A and B are in this plane as shown.

Let be the velocity of the particle when it has reached point P. Let be the mass of the particle.

The potential energy at the point P is given as = ( )

The kinetic energy at the point P is given = −0

Thus, the speed of the particle is = 2 ( )

An element of distance traversed is:

= ( ) +( )

= 1+

= 1+
Thus, the element of time to traverse ds is

106 Lovely Professional University

 Notes

Unit 11: Problem of Brachistochrone and Minimum Surface of Revolution

1+
= =
2 ( )
and the total time of descent is

1+
=
2 ( )

1 1+
=
2 ( )

1
= 1+ ( )
2
Different functions ( ) will typically yield different values for T; we call T a "functional" of .
Our goal is to find the minimum of this functional with respect to possible functions .
Here

= (1)
√

Using Euler’s equation

− = (2)

∴ from (1) and (2)

1+ 1 2
− × =
2 1+

1+
− =
1+
Or
1+ −
=
1+
1
⇒ =
1+
Squaring both sides
1
=
(1 + )
Or
1
(1 + )= = ( )

⇒1+ =

Or

Lovely Professional University 107

Notes

Calculus of Variation and Integral Equation

−
=

Or

−
=

Separating the variable and integrating

=
−

Or

= +
−

Let = sin ⇒ = 2 sin cos

sin
= 2 sin cos +
− sin

Or
= 2 ∫ tanθ sinθ cos +

=2 sin θ +

1 − cos 2
=2
2

or
sin 2
= − +
2
and

= [1 − cos 2 ]
2
which are the parametric equations ofa cycloid. Thus, the brachistochrone is a cycloid.

11.2 Problem of Minimum Surface of Revolution

Symmetries
Azimuthally or cylindrical symmetry is symmetry around a straight line. Spherical
symmetry is symmetry around a point.

108 Lovely Professional University

 Notes

Unit 11: Problem of Brachistochrone and Minimum Surface of Revolution

Surface of Revolution
A surface of revolution is a surface generated by rotating a 2-D curve about an axis.
The resulting surface therefore always has azimuthal symmetry.

Example:Among all the curves joining the two given points ( , ) and ( , ), find the one
which generates the surface of minimum area when rotated about the y-axis.

Consider a strip at point

We know that surface area of cylinder = 2 ℎ

Here = and ℎ =

∴ Surface area of the strip at =2

Total surface area = ∫ 2

(∵ = + = 1+ )

Hence = 1+

=2 ∫ 1+ (1)

Here integrand = 1+

Equation (1) will be minimum if = 0 for which − =0

should be satisfied
1
×2 =
2 1+
Or

= 1+
Squaring both sides

Lovely Professional University 109

Notes

Calculus of Variation and Integral Equation

= (1 + )
Or
( − ) =
Or

=
−

Or

=
√ −

⇒ =
√ −
Separating the variables and integrating

=
√ −
⇒ = ∫√ + (2)

= ln + − +

Or

= cosh +

Or

− = cosh

Or
−
= cosh

Example:Among all the curves joining the two given points ( , ) and ( , ), find the one
which generates the surface of minimum area when rotated about the -axis.
The surface area of the strip at A = 2

=2 1+

Total surface area = = ∫ 2 1+ (1)

110 Lovely Professional University

 Notes

Unit 11: Problem of Brachistochrone and Minimum Surface of Revolution

Equation (1) should be minimum

⇒ =0

If − =0 (2)

(Euler Lagrange’s equation)

Here = 1+ = ( , ′)

Equation (2) reduces to

− =

2
⇒ 1+ − × =
2 1+
Or

(1 + )− = 1+
Hence

= 1+
or
= (1 + )
Or

−1 =

−
⇒ =

Or

−
=

Separating the variables and integrating both sides, we have

= +
−

⇒ cosh = +

Or
+
= cosh

Or

= cosh , where =

which is catenary.

A note on catenary
The catenary is the equilibrium shape assumed by a chain suspended from two points. One speaks
of a chain rather than any kind of string since a chain with very small links is fully flexible and un
stretchable, as the idealized physical model assumes.

Lovely Professional University 111

Notes

Calculus of Variation and Integral Equation

Catena in Latin means “chain.” A chain is very strong, so it does not stretch very much, and
because of the links, it does not support any bending force beyond a single link.

Stretching a soap film between two parallel circular wire loops generates a catenoidal minimal
surface of revolution.

Thus the solution to the problem of minimum surface of revolution is a catenary.

Summary
 The brachistochrone is a cycloid.
 The solution to the problem of minimum surface of revolution is a catenary.

Key words
Brachistochrone, brachistochrone problem, cycloid, surface of revolution, minimum surface,
catenary

112 Lovely Professional University

 Notes

Unit 11: Problem of Brachistochrone and Minimum Surface of Revolution

Self Assessment
1. If a particle in the absence of friction will slide from one point to another in the shortest time
under the action of gravity, then the path is
A. a right circular cone
B. a cone
C. a cylinder
D. a cycloid

2. A necessary condition that the functional will be stationary is

A. its last variation vanishes

B. its variation never vanishes
C. its second variation vanishes
D. its first variation vanishes

3. The brachistochrone problem was posed by

A. Johann Bernoulli
B. James Bernoulli
C. Issac Newton
D. Leonhard Euler

4. Stretching a soap film between two parallel circular wire loops generates a

A. catenary
B. cycloid
C. catenoid
D. circle

5. = − + and = [1 − cos 2 ] are the parametric equations of

A. catenary
B. helix
C. cycloid
D. great circle

6. Azimuthal or cylindrical symmetry is symmetry around a

A. straight line
B. point
C. corner
D. curve

7. A surface of revolution always has

A. spherical symmetry
B. cylindrical symmetry

Lovely Professional University 113

Notes

Calculus of Variation and Integral Equation

C. rectangular symmetry
D. line symmetry

8. Which of the following is the equilibrium shape assumed by a chain suspended from
two points?

A. Parabola
B. Catenary
C. Hyperbola
D. Cycloid

9. The locus of a point on the rim of a circle rolling along a straight line is

A. paraboloid
B. hyperboloid
C. cycloid
D. catenary

10. The equation = cosh represents

A. paraboloid
B. hyperboloid
C. cycloid
D. catenary

Answers for Self Assessment

1. D 2. D 3. A 4. C 5. C

6. A 7. B 8. B 9. C 10. D

Review Questions
1. State the brachistochrone problem.
2. Solve the brachistochrone problem.
3. State the problem of minimum surface of revolution.
4. Among all the curves joining the two given points ( , ) and ( , ), find the one which
generates the surface of minimum area when rotated about the -axis.
5. Among all the curves joining the two given points ( , ) and ( , ), find the one which
generates the surface of minimum area when rotated about the -axis.

Further Readings
1. I. M. Gelfand,S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

114 Lovely Professional University

 Notes

Unit 11: Problem of Brachistochrone and Minimum Surface of Revolution

Web Links
https://mathworld.wolfram.com/BrachistochroneProblem.html
https://mathworld.wolfram.com/MinimalSurfaceofRevolution.html

Lovely Professional University 115

 Notes

Unit 12: Variational Problems in Parametric Form

Pratiksha, Lovely Professional University

Unit 12: Variational Problems in Parametric Form

CONTENTS
Objectives
Introduction
12.1 Variational problem in parametric form
12.2 Isoperimetric Problem
Summary
Keywords
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to

 define the conditional extremum

 distinguish in geometric and integral constraints
 find the conditional extremum for geometric constraints
 find the conditional extremum for integral constraints

Introduction
So far, we have considered functionals of curves given by explicit equations,
e.g. = ( ) ---(1)
in the two-dimensional case.
However, it is often more convenient to consider functional of curves given in parametric form.
Moreover, in problems involving closed curves, it is usually impossible to get along without
representing the curves in parametric form.

12.1 Variational problem in parametric form

Suppose that in the functional

∫ ( , , ) ---(2)

we wish to regard the argument as a curve which is given in parametric form, rather than in the
form (1). Then (2) can be written as

̇( )
∫ ( ), ( ),
̇( )
̇( ) = ∫ Φ( , , ̇ , ̇ ) (3)

̇( )
Here ̇ ( ) = , ̇( ) = and = = = ̇( )

116 Lovely Professional University

Notes

Calculus of Variation and Integral Equations

(where the over dot denotes differentiation with respect to t) i.e. as functional depending on two
unknown functions ( ) and ( ). The function Φ appearing in the right-hand-side of (3) does not
involve explicitly, and is positive homogenous of degree 1 in ̇ ( ) and ̇ ( ), which means that
Φ( , , ̇, ̇ ) ≡ Φ( , , ̇ , ̇ ) , ∀ > 0.

Conversely, let

Φ( , , ̇ , ̇ )

be a functional whose integrand Φ does not involve explicitly and is positive homogenous of
degree 1 in ̇ and ̇ . We now show that the value of such a functional depends on the curve in the
- plane defined by the parametric equations = ( ), = ( ), and not on the functions ( ),
( ) themselves, i.e. that if we go from to some new parameter by setting
= ( ),

Where > 0 and the interval [ , ] goes into [ , ], then

∫ Φ , , , = ∫ Φ( , , ̇ , ̇ ) .

In fact, since Φ is positive-homogeneous of degree 1 in ̇ and ̇ , it follows that

Φ , , , = Φ , , ̇ , ̇

= Φ( , , ̇ , ̇ ) = Φ( , , ̇ , ̇ )

As asserted. Thus, we have proved the following.

Theorem. A necessary and sufficient condition for the functional

Φ( , , ̇ , ̇ )

to depend only on the curve in the -plane defined by the parametric equations = ( ), = ( )
and not on the choice of the parametric representation of the curve, is that the integrand Φ should
not involve explicitly and should be a positive- homogenous function of degree 1 in ̇ and ̇ .

Now, suppose some parameterization of the curve = ( ) reduces the functional (2) of the form
̇
∫ , , ̇
̇ = ∫ Φ( , , ̇ , ̇ ) . (4)

The variational problem for the right-hand side of (4) leads to the pair of Euler equations

− ̇
= 0, − ̇
=0 (5)

which must be equivalent to the single Euler equation

− = 0, here =

corresponding to the variational problem for the original functional (1).

Hence, the equations (5) cannot be independent, and they are connected by the identity

̇ Φ − Φ ̇ + ̇ Φ − Φ ̇ =0

Lovely Professional University 117

 Notes

Unit 12: Variational Problems in Parametric Form

∂Φ ∂Φ
Here Φ ̇ = ̇
,Φ ̇ = ̇
,Φ =
∂
and Φ = ∂

We can deduce an important theorem due to Noether, concerninginvariant variational problems",

for the case of many independent variable.
Suppose we have a functional,

[ ]=∫ ( , ,∇ ) (6)

And a transformation
∗
= Φ ( , , ∇ ),
∗
= Ψ( , , ∇ ) (7)

( = 1,2, … , ) carrying the surface ∗

with equation = ( ) into the surface with the equation
∗ = ∗ ( ∗ ),

Definition. The functional (6) is said to be invariant under the transformation (7) if [ ∗] = [ ], i.e. if

( ∗, ∗ , ∇∗ ∗ ) ∗
= ( , ,∇ )
∗

Example: The functional

[ ]=∫ ∫ + (8)

Is invariant under the rotation

∗
= cos − sin ,
∗ = sin + cos ,
= ∗ (9)
Where is an arbitrary constant . In fact, since the inverse of the transformation (9) is
∗ cos ∗ sin
= +
∗ sin ∗ cos
=− + ,
∗
=
∗
It follows that, given a surface with equation = ( , ), the “transformed surface ” has the
equation
∗ ∗ cos ∗ sin ∗ sin ∗ cos ∗( ∗, ∗ ).
= ( + ,− + )=
Consequently, we have
∗ ∗
[ ∗] ∗ ∗
= ∗
+ ∗
∗

∗ ∗
= cos − sin + sin + cos
∗

( ∗, ∗)
= + = +
∗ ( , ) ∗

118 Lovely Professional University

Notes

Calculus of Variation and Integral Equations

∗
∵ ∗
= ∗
+ ∗
= cos + (− sin )
∗ ∗, ∗
And = + = sin + cos and ∗ ∗ =
∗ ∗ ∗ ,
∗ ∗

( ∗, ∗)
cos − sin
And = ∗ ∗ = =1
( , ) sin cos

Theorem (Noether): If the functional

[ ]= ( , ,∇ )

Is invariant under the family of transformations

∗
= Φ ( , ,∇ ; ) ~ + ( , , ∇ ),
∗ = Ψ( , , ∇ ; ) ~ + ( , ,∇ )
( = 1,2, … , ) for an arbitrary region , then

+ =0

On each extremal surface of [ ], where

= −

12.2 Isoperimetric Problem

The isoperimetric problem of queen Dido (9th century BC)

When the Phoenician princess Dido had fled to North Africa, she asked the Berber king for a bit of
land. She was offered as much land as could be encompassed by an oxhide.
Dido cut the oxhide into fine stripes so that she got a long enough rope to encircle an entire hill
named Byrsa, the center of Carthage.

Lovely Professional University 119

 Notes

Unit 12: Variational Problems in Parametric Form

What shape was the region that she encircled so that area was maximal?

Thisoperimetric problem is the problem to find the curve = ( ) for which the functional

[ ]=∫ ( , , ) (1)

has an extremum, where the admissible curves satisfy the boundary conditions
( )= , ( )= ,
And are such that,another functional

[ ]=∫ ( , , ) (2)

takes a fixed value .

Theorem 1: Given the functional

[ ]= ( , , ′) ,

Let the admissible curves satisfy the conditions

( )= , ( )= , [ ]=∫ ( , , ′) = , (3)

Where [ ] is another functional, let [ ] have an extremum for = ( ). Then, if = ( ) is not

an extremal of [ ], there exists a constant such that = ( ) is an extremal of the functional

( + ) ,

i.e. = ( ) satisfies the differential equation

− + − =0 (4)

Proof: Let [ ] have an extremum for the curve = ( ), subject to the conditions (3). We choose
two points and satisfies a condition to be stated belowbut is otherwise arbitrary. Then we give
( ) an increment ( )+ ( ) is nonzero only in a neighborhood of .
Using variational derivatives, we can write the corresponding increment Δ of the functional in the
form

Δ = + Δ + + Δ , (5)

where

Δ =∫ ( ) ,Δ =∫ ( ) ,

and , → 0 as Δ , Δ → 0.

We now require that the “varied” curve

= ∗( )= ( )+ ( )+ ( )
satisfy the condition
[ ∗] = [ ].
Writing Δ is in a form similar to (5), we obtain

Δ = [ ∗] − [ ] = + Δ + + Δ =0 (6)

where , → 0 as Δ , Δ → 0. Next, we choose to be a part for which

120 Lovely Professional University

Notes

Calculus of Variation and Integral Equations

≠ 0.

Such a point exists, since by hypothesis = ( ) is not an extremal of the functional . With this
choice of , we can write the condition (6) in the form

Δ =− + ′ Δ (7)

Where → 0 as Δ → 0. Setting

=− ,

And substituting (7) into the formula (5) for Δ , we obtain

Δ = + Δ + Δ , (8)

where → 0 as Δ → 0.This expression for Δ explicitly involves variational derivatives only at

= , and the increment ℎ( ) is now just ( ), since the “compensating increment” ( ) has
been taken into account automatically by using the condition Δ = 0. Thus, the first term in the
right hand side of (8) is the principal linear part of Δ , i.e. the variation of the functional at the
point is

δ = + Δ .

Since a necessary condition for an extremum is that = 0, and since Δ is non zero while is
arbitrary, we finally have

+ = + − + = 0,

= − + − =0

which is precisely equation (4). This completes the proof of the theorem.

Remarks:
∵ Let = + , where =Lagrangian multiplier
The using Euler-Lagrange equation

− =0

⇒ ( + )− ( + ) =0

Or

− + − =0

Example: Find the plane curve of the fixed perimeter and maximum area.

We know that

Lovely Professional University 121

 Notes

Unit 12: Variational Problems in Parametric Form

Area =∫

Perimeter=∫ =∫ 1+ =

Here = and = 1+

Define = + 1+ (1)
To have an extremum,
Euler-Lagrange equation is

− =

1
+ 1+ − 2 =
2 1+
Or

1+ + (1 + )− =
Or

1+ + = ⇒ 1+ = −
Squaring both sides
(1 + )=( − )
Or
( − )
1+ =

( − )
⇒ = −1

Or

( − ) −
=

Or

= (2)
( )

Put ( − ) − = ⇒ −2 =

(2)⇒ − =
√

Integrating both sides

= −2
√

122 Lovely Professional University

Notes

Calculus of Variation and Integral Equations

2√ = −2 +
Or

√ =− + , =

Squaring both sides,

=( − )
Or
( − ) − =( − )
⇒( − ) + =( − )

which is a circle.
So now you can answer, what shape was the region that Queen Dido encircled so that area was
maximal!

Example: Prove that the extremal of

( )=∫ , (1) = 3, (4) = 24 (1)

such that

∫ = 36, (2)

is a parabola.

Here = , =
Let = + , where = Lagrangian multiplier
= +
For (1) to have extremum, given (2)

− =0

Or here,

− =

Or
+ − [2 ] =
⇒ − =
− =
Or

= −
Or

=
−
Integrating both sides

=
−

Lovely Professional University 123

 Notes

Unit 12: Variational Problems in Parametric Form

2
⇒ − = +

Squaring both sides

4
( − )=( + )

Or

4 = ( + ) + (3)

which is a parabola and more specifically an upward parabola.

Since we are given (1) = 3 and (4) = 24 , we can find the values of arbitrary constants ,
also.
4
∵ (1) = 3 ⇒ 12 = (1 + ) +
4
And
4
(4) = 24 ⇒ 96 = (4 + ) +
4

Hence

∫ = 36 along with
4
(1 + ) + = 12
4
and
4
(4 + ) + = 96
4
the exact equation of the parabola can be obtained.

Summary
 Suppose that in the functional

∫ ( , , ) ---(2)

we wish to regard the argument as a curve which is given in parametric form, rather than in the
form (1). Then (2) can be written as

̇( )
∫ ( ), ( ),
̇( )
̇( ) = ∫ Φ( , , ̇ , ̇ ) (3)

The function Φ appearing in the right-hand-side of (3) does not involve explicitly, and is positive
homogenous of degree 1 in ̇ ( ) and ̇ ( ), which means that
Φ( , , ̇, ̇ ) ≡ Φ( , , ̇ , ̇ ) , ∀ > 0.

Conversely, let

Φ( , , ̇ , ̇ )

be a functional whose integrand Φ does not involve explicitly and is positive homogenous of
degree 1 in ̇ and ̇ . The value of such a functional depends on the curve in the - plane defined
by the parametric equations = ( ), = ( ), and not on the functions ( ), ( ) themselves.

124 Lovely Professional University

Notes

Calculus of Variation and Integral Equations

 A necessary and sufficient condition for the functional

Φ( , , ̇ , ̇ )

to depend only on the curve in the -plane defined by the parametric equations = ( ), = ( )
and not on the choice of the parametric representation of the curve, is that the integrand Φ should
not involve explicitly and should be a positive- homogenous function of degree 1 in ̇ and ̇ .

 If the functional

[ ]= ( , ,∇ )

Is invariant under the family of transformations

∗
= Φ ( , ,∇ ; ) ~ + ( , , ∇ ),
∗
= Ψ( , , ∇ ; ) ~ + ( , ,∇ )
( = 1,2, … , ) for an arbitrary region , then

+ =0

On each extremal surface of [ ], where

= −

Keywords
Variational problem, isoperimetric problem, Dido’s problem, Noether theorem

Self Assessment
̇( )
1. For the functional ∫ ( ), ( ),
̇( )
̇( ) = ∫ Φ( , , ̇ , ̇ ) , the function Φ is

A. positive homogenous of degree 2 in ̇ ( ) and ̇ ( )

B. positive homogenous of degree 1 in ̇ ( ) and ̇ ( )
C. negative homogenous of degree 1 in ̇ ( ) and ̇ ( )
D. negative homogenous of degree 2 in ̇ ( ) and ̇ ( )

2. A necessary and sufficient condition for the functional

Φ( , , ̇ , ̇ )

to depend only on the curve in the -plane defined by the parametric equations =
( ), = ( )and not on the choice of the parametric representation of the curve, is that
I. the integrand Φ should not involve explicitly
II. the integrand should be a positive- homogenous function of degree 1 in ̇ and ̇ .
A. Only I is true
B. Only II is true
C. Both I and II are true

Lovely Professional University 125

 Notes

Unit 12: Variational Problems in Parametric Form

D. None is true

3. The plane curve of the fixed perimeter and maximum area is a

A. Great circle
B. Straight line
C. Helix
D. Circle

4. The isoperimetric problem of queen Didodates back to

A. 5th century BC
B. 6th century BC
C. 8th century BC
D. 9th century BC

5. For the functional at the point ,the necessary condition for an extremum is that
A. =0
B. ≠0
C. >0
D. <0

Answers for Self Assessment

1. B 2. C 3. D 4. D 5. A

Review Questions
1. Define an isoperimetric problem with example.
2. Explain the solution of Dido’s problem.
3. State the Noether’s theorem.
4. Find the necessary and sufficient condition for the functional∫ Φ( , , ̇ , ̇ ) to depend

only on the curve in the -plane defined by the parametric equations = ( ), = ( )

and not on the choice of the parametric representation of the curve.

5. Find the extremal for the functional

( )=∫ , (1) = 3, (4) = 24such that ∫ = 36.

Further Readings
1. I. M. Gelfand,S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

Web Links
https://www.youtube.com/watch?v=EaPyCNUbu40

126 Lovely Professional University

Notes

Calculus of Variation and Integral Equations

Lovely Professional University 127

 Notes
Unit 13: Geodesics
Pratiksha, Lovely Professional University

Unit 13: Geodesics

CONTENTS
Objectives
Introduction
13.1 The problem to calculate a Geodesic
13.2 Geodesics on Various Surfaces
Summary
Key words
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to

 define a geodesic
 distinguish between geodesics of various manifolds
 find the geodesics for a plane, sphere, cylinder and cone

Introduction

The problem of finding geodesics, is the problem of finding shortest curves joining two points of
some manifold.
The same kind of problem arises in geometric optics, in finding the paths along which light rays
propagate in an inhomogeneous medium. In fact, according to Fermat's principle, light goes from a
point to a point along the path for which the transit time is the smallest.

128 Lovely Professional University

Notes
Calculus of Variations and Integral Equations

Geodesics are curves of shortest distance on a given surface. Apart from their intrinsic interest, they
are of practical importance in the transport of goods and passengers at minimal expense of time
and energy. They are also of paramount importance as escape routes during flights.
Finding geodesics can be accomplished using the methods of differential geometry as well as the
calculus of variation.
As an application, consider the calculation of great circle routes on a sphere, in particular, for any
two cities on the Earth. The airlines have long discovered this, with flight paths between cities as
sections of planes through the Earth’s center.

13.1 The problem to calculate a Geodesic

The fundamental equation in the calculus of variation is the Euler-Lagrange equation:

− = 0, ! ≤ ≤ (1)

In differential calculus, we are looking for those values of which give some function ( ) its
maximum or minimum values. In the calculus of variation, we are seeking the function itself that
makes some integral of , satisfying certain conditions, a maximum or minimum. In the geodesic
problem, we wish to find the curve = ( , , ′) that joins the two points on a given surface such
that the distance between them is minimized. Thus, the problem is to find that integrand which
minimizes the integral of the arc length.:

=∫ =∫ 1+( ) , = 1+( ) (2)

Here, = ( , , ′),

And = + + .

Hence, (1) is equivalent to :

− − =0 (3)

There are first integrals of (1) based on two simplifying assumptions: (i) if is independent of , i.e.,
= 0. Then

=0⇒ = (4)

A first - order degree 1 differential equation.

(ii) If is independent of :

− = − − =− − − , (5)

The last term = 0, ⇒

− = (5’)

= ( , ), = ( , ), = ( , ) (6)

The arclength is:

( ) =( ) +( ) +( ) = +2 + (7)

With = + + , = + + , = + + .

And the integral (2) to minimize is:

Lovely Professional University 129

 Notes
Unit 13: Geodesics

= ∫ [ +2 + ] (8)

13.2 Geodesics on Various Surfaces

1. A Plane:
For any two points on a plane,
the geodesic is a straight line.
We have proved this in the topic of
problem of shortest distance.

2. A Sphere
To find the geodesic for any two points on a
Sphere, the ideal way will be to work with
spherical coordinate system.
The Cartesian and spherical coordinates are
Related as follows:

= sin cos ,
= sin sin ,
= cos
The respective variations are
= cos cos + sin (− sin ) ;
= cos sin + sin cos ;
= (− sin )
The elementary arc length is given as,

( ) =( ) +( ) +( ) = [( ) + sin ( ) ]= ( ) 1 + sin

= 1 + sin

= = 1 + sin

=∫ (let = ′)

Here

= 1 + sin ′ (1)

− =0

=
′
sin 2 ′
∴ (1) ⇒ =
2 1 + sin ′

130 Lovely Professional University

Notes
Calculus of Variations and Integral Equations

Or
sin ′
=
1 + sin ′

sin = 1 + sin
Or
[sin − C sin ]=

∴ =
sin (sin − )
Or
csc
=
(sin − )
csc
=
(1 − csc )

Or
csc
=
(1 − (1 + cot ))
Hence
csc
=
1− (1 + cot )

Put = , ⇒ − csc =

∴ =−
1− (1 + )
Or

= −∫
√1 − −

= − sin +
√1 −

− = − sin
√1 −
Or

sin( − )= =
√1 − √1 −
cos
=
√1 − sin
cos
sin sin( − )=
√1 −
Or
cos
sin sin( − )=
√1 −
cos
sin (sin cos − cos sin ) =
√1 −
Or

Lovely Professional University 131

 Notes
Unit 13: Geodesics

sin − cos =
√1 −

Which is the equation of a plane passing through the origin, known as the Great Circle.

Thus, the geodesics for a sphere are the great circles as shown.

3. Right Circular Cylinder

To find the geodesic for any two points on a
cylinder, the cylindrical coordinates work the
best. The relations in Cartesian and cylindrical
Coordinates are given as
= cos
= sin
=
The elementary arc length is given as,
( ) =( ) +( ) +( )

= ( ) +( ) +( )

= (− sin ) + ( cos ) +( )

= ( ) +( )

= ( ) 1+

Or

= √1 + , where = ′( )

The total length, =∫ = ∫ √1 +

Here,

( , , ) = √1 + (1)

− =0

⇒ =
′

(1) ⇒ =
√1 +
= 1+
On squaring both sides, we get
= (1 + )
Or
( − )=
Or

=
−

=
√ −

132 Lovely Professional University

Notes
Calculus of Variations and Integral Equations

Or

= , where =
√

⇒ = +

which is the equation of helix. Therefore the geodesic for two points on the cylinder
is a helix.

4. Right Circular Cone

For any two points on the cone, we use the spherical
coordinates system. The Cartesian and spherical coordinate
systems are related as,
= sin cos ,
= sin sin ,
= cos

The elementary length of the arc is given as,

( ) =( ) +( ) +( )

= ( ) +( ) +( )

And
= sin (− sin ) + sin cos ;
= sin cos + sin sin ;
= cos
∴( ) = + ( sin )
Now
=∫

=∫ + ( sin )
Or

= ∫ 1 + ( sin )

= ∫ 1 + ( sin ) , where =

Lovely Professional University 133

 Notes
Unit 13: Geodesics

Here

( , , )= 1 + ( sin ) (*)

Using Euler-Lagrange’s equation

− =0

⇒ =

Or
sin ′
=
1+ sin
Or

sin = 1 + ( sin )
Squaring both sides, we get
sin = (1 + ( sin ) )
Or
[ sin − sin ]=
Or

=
sin ( sin − )
Or

=
sin √ sin −
Or

=
sin √ sin −
Integrating both sides

= ∫ (using∫ = sec + )
√

sin
= × sec +
sin

1 sin
= sec +
sin

which is the equation of a helix. Thus, the geodesic for two points on a cone
is a helix.

134 Lovely Professional University

Notes
Calculus of Variations and Integral Equations

Summary
 The problem of finding geodesics, is the problem of finding shortest curves joining two
points of some manifold.
 Geodesic on a plane is a straight line.
 Geodesic on a sphere is a great circle.
 Geodesic on a cylinder and cone are the helix.

Key words
Geodesic, manifold, straight line, helix, great circle

Self Assessment
1. The geodesic for two points in a plane is
A. Helix
B. Straight line
C. Great circle
D. Cycloid

2. The geodesic for two points on a sphere is

A. Helix
B. Straight line
C. Great circle
D. Cycloid

3. The geodesic for two points on a right circular cylinder is

A. Helix
B. Straight line
C. Great circle
D. Cycloid

4. The geodesic for two points on a right circular cone is

A. Helix
B. Straight line
C. Great circle
D. Cycloid

Lovely Professional University 135

 Notes
Unit 13: Geodesics

5. The problem of finding geodesics, is the problem of finding

A. shortest curves joining two points of some manifold.
B. longest curves joining two points of some manifold.
C. optimum curves joining two points of some manifold.
D. None of these

Answers for Self Assessment

1. B 2. C 3. A 4. A 5. A

Review Questions
1. Derive the geodesic for two points on a plane.
2. Define geodesic with example.
3. Derive the geodesic for two points on a sphere.
4. Derive the geodesic for two points on a right circular cylinder.
5. Derive the geodesic for two points on a right circular cone.

Further Readings
1. I. M. Gelfand,S. V. Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)
3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

Web Links
https://www.youtube.com/watch?v=NfqrCdAjiks

136 Lovely Professional University

 Notes

Pratiksha, Lovely Professional University Unit 14: Conditional Extremum under Constraints

Unit 14:Conditional Extremum under Constraints

CONTENTS
Objectives
Introduction
14.1 Conditional Extremum under Geometric Constraints
14.2 Conditional Extremum under Integral Constraints
14.3 Isoperimetric Problem
Summary
Key words
Self Assessment
Answers for Self Assessment
Review Questions
Further Readings

Objectives
Students will be able to

 define the conditional extremum

 distinguish in geometric and integral constraints
 find the conditional extremum for geometric constraints
 find the conditional extremum for integral constraints

Introduction
Variational problems involving a conditional extremum are problems in which we find an
extremum of a functional v such that certain constraints are imposed on the functions on which the
functional v is dependent.

For example, find the extremum of the functional

[ , ,…, ]= ( , , ,…, , , ,…, )

with the conditions

[ , , ,…, ]=0 ( = 1,2, … ; < )

Recall the solution of a similar problem dealing with the investigation of the function =
( , , … , ) for an extremum of the constraints
[ , ,…, ] = 0 ( = 1, 2, … , ; < )
The natural way is to solve the system

[ , ,…, ] = 0 ( = 1, 2, … , )

Lovely Professional University 137

Notes

Calculus of Variations and Integral Equations

The equations of which are independent to some kind of m variables, say , ,…, :
Without loss of generality, we can solve these equations for , ,…,

= ( , ,…, );
= ( , ,…, );
⋮
= ( , ,…, );

and w.r.t. the substitution of , ,…, into ( , , … , ). Then the functions

( , , … , )become a function ( , , … , ) only of the − variables , ,…,
which are already independent and so the problem has reduced to investigating the function for
an unconditional extremum. This approach can also be used to solve the above variational problem.
Solving the system
( , , ,…, ) = 0, = 1,2, … ,

for , ,…, and substituting their expressions into [ , , … , ], we get the functional
[ , ,…, ] which depends only on n−m arguments that are already independent.

However, both for the functions and functionals, we may apply a more convenient method known
as method of undetermined coefficients.

14.1 Conditional Extremum under Geometric Constraints

If we investigate the extremum of the function = ( , ,…, ). with the given constraints.
( , ,…, ) = 0, = 1,2, … ,

Then using the method of undetermined coefficients, we construct a new auxiliary function

̅= +

are the constant factors and the function ̅is now investigated for an unconditional extremum i.e.
we form a system of equations
̅
= 0, = 1, 2, … ,

supplemented by the constraint equations = 0 ( = 1,2, … , )

from which all the + unknowns , ,…, and , ,…, are determined.
Also the problem involving a conditional extremum for functional is solved in similar fashion,
namely if

= ( , , ,…, , , ,…, )

with the conditions

[ , , ,…, ] = 0, = 1,2, …
Then the functional ̅ =∫ ( +∑ ( ) )

138 Lovely Professional University

 Notes

Unit 14: Conditional Extremum under Constraints

or ̅=∫ is constructed, where = +∑ ( ) which is now investigated for an
unconditional extremum, i.e. we solve the system of Euler's equations

− =0 = 1, 2, … , (1)

with the constraint equations

( , , ,…, ) = 0, = 1,2, … ,
The number of equations + is sufficient to determine the + unknown functions , , … ,
and , , … , and the boundary conditions ( ) = and ( ) = ; = 1,2, … which
must not contradict the constraint equations and will help to determine the 2n arbitrary constants
in the general solution of the system of Euler's equations.
The curves thus found on which a minimum or a maximum of the functional ̅ is achieved will also
be solutions of the original variational problem.

In fact the functions ( ), = 1,2, … , and , = 1,2, … found from the system (1), all the
= 0 and hence ̅ = and if for = ( ), = 1,2, … determined from the system (1) there is
achieved an unconditional extremum of the functional v ̅ i.e. an extremum relative to all close
lying curves (both which satisfy the constraint equations and those who do not), then, in particular
an extremum is also achieved w.r.t. a subclass of curves that satisfy the constraint equations.

Theorem: Given the conditions

[ , , ,…, ]=0 ( = 1,2, … ; < )

the functions , ,…, that extremize the functional

= ( , , ,…, , , ,…, )

satisfy the Euler's equations formed for the functional

̅= ( + ( ) ) =

for an appropriate choice of factors ( ), = 1,2, … ,

The functions and are determined from the Euler's equations

− =0, = 1, 2, … ,

and
( , , ,…, ) = 0, = 1,2, … ,

The constraint equations

( , , , … , ) = 0, = 1,2, … , are assumed to be independent i.e. one of the Jacobians of

order is different from zero,

Lovely Professional University 139

Notes

Calculus of Variations and Integral Equations

( , ,…, )
≠0
( , ,…, )

Proof : In this case, the condition for extremum = 0 takes the form

( + ) =0

On integration by parts and denoting = and = 0, = 0, we get

( − ) =0

Since the functions are subject to independent constraints

( , , ,…, ) = 0, = 1,2, … ,

It follows that the variations are not arbitrary and the fundamental lemma cannot be applied in
the present form. The variations must satisfy the following conditions obtained by means of
varying the constraint equations =0

= 0, = 1,2, … ,

And hence only − of the variations may be considered arbitrary.

Multiplying each of these equations term by term by ( ) and integrating from to , we get

( ) = 0, = 1,2, … ,

Adding term wise all these m equations, which are satisfied by the permissible variations , with
the equation

( − ) =0

we will have

+ ( ) − ) =0

Let us assume = +∑ ( )
We get

140 Lovely Professional University

 Notes

Unit 14: Conditional Extremum under Constraints

( − ) =0

Now choose factors ( ), ( ), … , ( ), so that they should satisfy the equations

( − ) = 0, = 1,2, … ,

or

+ ( ) − = 0, = 1,2, … ,

These equations form a system that is linear in with a non-zero determinant

( , ,…, )
≠0
( , ,…, )
Hence this system has the solution ( ), ( ), … , ( ). Given this choice of ( ), ( ), … , ( ),
the basic necessary condition for an extremum

( − ) =0

takes the form

( − ) =0

Now for the extremizing functions , , … , of the functional , this functional equation reduces
to an identity for an arbitrary choice of , = + 1, … , it follows that the fundamental lemma is
now applicable
We get

( − ) =0

Now choose factors

( ), ( ), … , ( ), so that they should satisfy the equations

( − ) = 0, = 1,2, … ,

or

+ ( ) − = 0, = 1,2, … ,

These equations form a system that is linear in with a non-zero determinant

Lovely Professional University 141

Notes

Calculus of Variations and Integral Equations

( , ,…, )
≠0
( , ,…, )
Putting all the equal to zero in turn, except one, and applying the lemma, we get

( − ) = 0, = + 1, … ,

Taking into account the above obtained equations

( − ) = 0, = 1,2, … ,

And finally we find the functions which achieve a conditional extremum of the functional v and the
factors ( ) must satisfy the system of equations

( − ) = 0, = 1,2, … ,

( , , ,…, ) = 0, = 1,2, … ,

Find the shortest distance between the two points ( , , ) and ( , , ) on the surface
( , , ) = 0.

= + +

= + +

= 1+ +

= 1+ +

We’ll find the minimum of provided ( , , ) = 0.

Consider the auxiliary functional
̅ =∫ ( 1+ + + ( , , )) ---- (1)

By Euler-Lagrange’s equation,

∫ ( , . ) then − =0

And if
↔ ( , , , , ′)
then

− = 0 and − =0

∴ from (1), we have

− = 0---- (2)

− = 0---- (3)

And
( , , ) = 0---- (4)
(2), (3) and (4) ⇒ = ( ) and = ( ) on which the conditional minimum of the functional ( )
can be achieved as well as all the ’s.

142 Lovely Professional University

 Notes

Unit 14: Conditional Extremum under Constraints

14.2 Conditional Extremum under Integral Constraints

In the preceding section we examined the problem of investigating the functional for an extremum:

= ( , , ,…, , , ,…, )

( )= , ( )= , = 1,2, … …
given the finite constraints

( , , ,…, ) = 0, = 1,2, … ,
Now if the constraint equations are the differential equations

( , , ,…, , , ,…, ) = 0, = 1,2, … , ; <

This type of constraints is called non holonomic. In this case we will prove a conditional extremum
of a functional is achieved on the same curves on which is achieved an unconditional extremum
of the functional.

̅= + ( ) =

where

= + ( )

Suppose that one of the functional determinants of order m is different from zero say

( , ,…, )
≠0
( , ,…, )

This guarantees independence of the constraints.

Solving the equation

( , , ,…, , , ,…, ) = 0 for , ,…, which is possible since

( , ,…, )
≠0
( , ,…, )

we get = ( , , ,…, , , , … , ), i=1,2,…,m. If we consider , ,…, arbitrarily

specified functions, then , … , are determined from this system of differential equations.

Thus, , ,…, are arbitrarily differentiable functions with fixed boundary values and
hence, their variations are arbitrary in the same sense.

Lovely Professional University 143

Notes

Calculus of Variations and Integral Equations

Let , ,…, be an arbitrary permissible system of functions that satisfies the constraint
equations = 0, = 1,2, … , .
Vary the constraint equations

+ = 0, = 1,2, … ,

Multiply term by term each of the equations obtained by the undetermined factor ( ) and
integrate from to ; this gives

( ) + ( ) = 0, = 1,2, … ,

Integrating each term of the second integral by parts and taking into consideration that =
and = = 0, we will have

∫ ∑ ( ) − ( ) =0 (1)

From the basic necessary condition for an extremum

=∫ ∑ − =0 (2)

Since =∫ ∑ + =∫ ∑ −

Adding term wise all the equations (1) and equation (2) and introducing the notation
= +∑ ( ) , we have

∫ ∑ ( − ) =0 (3)

Since the variations , = 1,2, … , are not arbitrary, we cannot yet use the fundamental lemma.
Choose factors ( ), ( ), … , ( ) so that they satisfy the equations

− = 0, = 1,2, … ,

These equations form a system of linear differential equations in ( ) and , = 1,2, … ,

which has the solution ( ), ( ), … , ( ) , (3) is reduced in the form

( − ) =0

where the variations , = + 1, … , are now arbitrary and hence assuming all variations =
0, except one, and applying the fundamental lemma we obtain

− = 0, = + 1, … ,

Thus, the functions ( ), … , ( ) that render the functional a conditional extremum, and the
factors ( ), ( ) … , ( ) must satisfy the system of + equations

144 Lovely Professional University

 Notes

Unit 14: Conditional Extremum under Constraints

− = 0, = 1,2, … ,

and

= 0, = 1,2, … ,

i.e. they must satisfy Euler equations of the auxiliary functional ̅ , which is regarded as a functional
dependent on the n+m functions , , … , , , … , .

14.3 Isoperimetric Problem

To obtain the necessary condition in an isoperimetric problem involving finding an extremum of a
functional = ∫ given the constraints ∫ = , = 1,2, … , it is necessary to form the
auxiliary functional

∗ = +

where are constants and write the Euler equations for it. The arbitrary constants , , … in
the general solution of a system of Euler's equations and the constants ,…, are determined
from the boundary conditions
( )= , ( )= , = 1,2, … ,

and from the isoperimetric conditions

= , = 1,2, … ,

∗ does ∗
The system of Euler's equations for the functional not vary if is multiplied by some
constant factor and hence is given in the form

∗
=∫ ∑ where = and = , = 1,2, … , .

Now all the functions enter symmetrically and therefore the extremals in the original variational
problem and in the problem involving finding an extremum of the functional ∫ given the
isoperimetric conditions

= , = 1,2, … − 1, + 1, … ,

coincide with any choice of , = 0,1, … , . This property is called the reciprocity principle.

NOTE:
• The finite constraints are
( , , ,…, ) = 0, = 1,2, … ,

• The constraint equations are the differential equations

Lovely Professional University 145

Notes

Calculus of Variations and Integral Equations

( , , ,…, , ′, ′, … , ′) = 0, = 1,2, … , ; <

The constraints∫ ( , , ,…, ) = , = 1,2, … ,

We can assume a new variable

=∫ ( , , ,…, ) , ( ) = 0, ( )= = 1,2, … ,

This implies that = ( , , ,…, )

Or − ( , , ,…, )=0

Or ( , , ,…, , , ,…, )=0

which is same as the above problem.

Prove that among all the triangles with fixed area, equilateral triangle has the smallest
perimeter.
We will use the variables , and ℎ as indicated in the figure below:
With the variables , anf ℎ the aarea and the perimeter are given as follow:

1
= ( + )ℎ
2

= + + +ℎ + +ℎ

Applying the method of Lagrange multipliers, we get the equations

( + )ℎ = 2

1+ = ℎ (1)
√

1+ = ℎ (2)

ℎ ℎ
+ = ( + )
√ +ℎ +ℎ

Equations (1) and (2) yield = . So, the above four equations reduce to the following three
equations
ℎ= (3)

146 Lovely Professional University

 Notes

Unit 14: Conditional Extremum under Constraints

1+ = ℎ (4)
√

= (5)
√

Multiplying (4) by and (5) by ℎ, we obtain

ℎ
+ = ℎ =
√ +ℎ √ +ℎ
Or equivalently,
ℎ −
=
√ +ℎ
Squaring the last equation and simplifying we get
( + ℎ ) = (ℎ − )
+ℎ = ℎ − 2ℎ +
Now we use (3) and further simplify,
= ℎ −2
Or
3 =ℎ
√
ℎ = √3√ and = =
√

Now we calculate the sides of the triangle:

2
= + = √
√3

1 1 3 2
= +ℎ = + √3 = + = √
√3 √3 √3 √3

1 1 3 2
= +ℎ = + √3 = + = √
√3 √3 √3 √3

The triangle obtained by the Lagrange method is equilateral.

The Lagrange method produced only one possible extreme value for the perimeter and since for a
small ℎ the corresponding is large, we conclude that we obtained a triangle with the minimal
perimeter.

Summary
 Given the conditions

[ , , ,…, ]=0 ( = 1,2, … ; < )

the functions , ,…, that extremize the functional

= ( , , ,…, , , ,…, )

satisfy the Euler's equations formed for the functional

̅= ( + ( ) ) =

for an appropriate choice of factors ( ), = 1,2, … ,

The functions and are determined from the Euler's equations

Lovely Professional University 147

Notes

Calculus of Variations and Integral Equations

− =0, = 1, 2, … ,

and
( , , ,…, ) = 0, = 1,2, … ,

The constraint equations

( , , , … , ) = 0, = 1,2, … , are assumed to be independent i.e., one of the

Jacobians of order is different from zero,
( , ,…, )
≠0
( , ,…, )

 Among all the triangles with fixed area, equilateral triangle has the smallest perimeter.

Key words
 Integral constraints,
 geometric constraints,
 isoperimetric problem,
 Lagrange method of undetermined multipliers

Self Assessment
1. Among all the triangles with fixed area, which triangle has the smallest perimeter
A. Right angled
B. Scalane
C. Isosceles
D. Equilateral

2. Which of the following is an auxiliary functional for the perimetric problem?

A. ∫ ( + ∑ )
B. ∫ (∑ )
C. ∫ ( )

D. None of these

3. The externals in the original variational problem and in the problem involving finding an
extremum of the functional ∫ given the isoperimetric conditions

= , = 1,2, … − 1, + 1, … ,

coincide with any choice of , = 0,1, … , . This property is called the

A. Conditional principle
B. Reciprocity principle
C. Extremal principle

148 Lovely Professional University

 Notes

Unit 14: Conditional Extremum under Constraints

D. None of these

4. One of the functional determinants of order m is different from zero say

( , ,…, )
( )
≠ 0 guarantees the
, ,…,
A. Independence of constraints
B. Dependence of constraints
C. Independence of functional
D. Dependence of functional

5. In the system of Euler's equations

− =0 = 1, 2, … ,

with the constraint equations ( , , ,…, ) = 0, = 1,2, … , , how many equations are
needed
to determine the unknown functions?

A. m
B. n
C. m+n
D. m-n

Answers for Self Assessment

1. D 2. A 3. B 4. A 5. C

Review Questions
1. Isoperimetric problem is an example of (a) integral constraints or (b) geometric constraints?

2. Prove that among all the triangles with fixed area, equilateral triangle has the smallest
perimeter.
3. To obtain the necessary condition in an isoperimetric problem involving finding an

extremum of a functional =∫ given the constraints ∫ = , =

1,2,… , it is necessary to form

(a) An integral or
(b) An auxiliary functional

4. Variational problems involving a ______ extremum are problems in which we find an

extremum of a functional v such that certain constraints are imposed on the functions
on which the functional v is dependent.
5. Explain the method of Lagrange method of undetermined coefficients.
6. Find the shortest distance between the two points ( , , ) and ( , , ) on the
surface ( , , ) = 0.

Further Readings
1. I. M. Gelfand,S. V.Fomin, “Calculus of Variations”, Prentice-Hall Inc. (1963)
2. G.F. Hadley, "Nonlinear and dynamic programming", Addison-Wesley (1964)

Lovely Professional University 149

Notes

Calculus of Variations and Integral Equations

3. G.A. Bliss, "Lectures on the calculus of variations", Chicago Univ. Press (1947)

Web Links
https://www.ucl.ac.uk/~ucahmto/latex_html/pandoc_chapter2.html

150 Lovely Professional University

LOVELY PROFESSIONAL UNIVERSITY
Jalandhar-Delhi G.T. Road (NH-1)
Phagwara, Punjab (India)-144411
For Enquiry: +91-1824-521360
Fax.: +91-1824-506111
Email: [email protected]