linear integral equations and its applicationsijaema.com/gallery/261-august-2292.pdflinear integral...
TRANSCRIPT
LINEAR INTEGRAL EQUATIONS AND ITS APPLICATIONS
Christy Tom Mathews, Bijesh P Biju Department of Mathematics, Deva Matha College Kuravilangad, Kerala, India
Department of Basic Science,Muslim Association College of Engineering,Venjaramoodu,Kerala,India
Email:[email protected]
Abstract
The theory of integral equations was first introduced by J. Fourier. In this study ,we go through
some basic definitions and classifications of integral equations. We will be concentrating more on
Volterra and Fredholm types of Linear integral equations, where we will see numerical methods of
solving them. Finally the study ends with some applications of integral equations to real world
problems.
Keywords: Integral equations, Linear integral equations ,Volterra integral equations ,Fredholm integral
equations , Numerical solutions
1. Introduction
Fourier (1768-1830) is the initiator of the theory of integral equations. The term integral
equation was first suggested by Du-Bois Reymond in 1888. The pioneering systematic investigations
goes back to late 19th and early 20th century works of Volterra, Fredholm & Hilbert. In 1887, Volterra
published a series of famous papers in which he singled out the notion of a functional and pioneered in
the development of a theory of functionals in the theory of linear integral equations of special types.
Fredholm presented the fundamentals of the Fredholm integral equation theory in a paper published in
1903 in the Acta Mathematica.
The Study here mainly points towards linear integral equations especially on Fredholm and
Volterra type of integral equations. One of the main relevance of this study arise from the idea of
conversion of differential equations in to integral equations and vice versa. It also includes some
numerical methods to solve these equations. Finally, the study establishes some real-world applications
of Linear integral equations.
2. Integral equation
Any equation in which the unknown function appears under the integral sign is known as an
integral equation. A typical form of an integral equation is
𝑢(𝑥) = 𝑓(𝑥) + 𝜆 ∫ 𝑘(𝑥, 𝑡)𝑢(𝑡)𝑑𝑡𝛽(𝑥)
𝛼(𝑥) (1)
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2162
Where, 𝑘(𝑥, 𝑡) is called kernel of the integral equation (1), 𝛼(𝑥), 𝛽(𝑥) are the limits of integration, 𝜆 is
a constant parameter. It can be easily observed that the unknown function 𝑢(𝑥) appears under integral
sign. It is to be noted that kernel 𝑘(𝑥, 𝑡) and the function 𝑓(𝑥) in the equation (1) are given functions.
Our prime objective is to find unknown function 𝑢(𝑥) that will satisfy equation (1) using a number of
solution techniques.
Eg: 𝑢(𝑥) = 𝑥 +1
2∫ 𝑥𝑡 𝑢(𝑡)𝑑𝑡
1
0
2.1. Classification of Integral Equations
An integral equation can be classified in to different categories as we have seen in ordinary and
partial differential equations. We can classify integral equations as Linear or Nonlinear,
Homogeneous or Nonhomogeneous. The most frequently used integral equations fall under two major
classes, namely Fredholm and Volterra integral equations.
An integral equation is called Linear if only linear operations are performed in it upon the
unknown function. For a linear integral equation, the unknown function 𝑢(𝑥) appearing under integral
sign is given in the functional form 𝐹(𝑢(𝑥)) such that the power of 𝑢(𝑥) is unity. The most general
type of linear integral equation is of the form,
ℎ(𝑥)𝑔(𝑥) = 𝑓(𝑥) + 𝜆 ∫ 𝑘(𝑥, 𝑡)𝑔(𝑡)𝑑𝑡𝑎
(2)
where, the upper limit may be either variable or fixed. The functions ℎ(𝑥), 𝑓(𝑥), 𝑘(𝑥, 𝑡) are known
functions while 𝑔(𝑥) is to be determined, 𝜆 is the non-zero real or complex parameter.
Example 1: 𝑢(𝑥) = 𝑓(𝑥) + ∫ 𝑘(𝑥, 𝑡)𝑢(𝑡)𝑑𝑡𝑥
0
If the unknown function 𝑢(𝑥) appearing under integral sign is given in the functional form
𝐹(𝑢(𝑥)) such that the power of 𝑢(𝑥) is no longer unity, then such integral equations can be classified
as Nonlinear integral equations.(That is, 𝐹(𝑢(𝑥)) = 𝑢𝑛(𝑥); 𝑛 ≠ 1 𝑜𝑟 𝑠𝑖𝑛 𝑢(𝑥) 𝑒𝑡𝑐. )
Example 2: 𝑢(𝑥) = 𝑓(𝑥) + ∫ 𝑘(𝑥, 𝑡)𝑢2(𝑡)𝑑𝑡𝑥
0
Throughout this study we mainly focus on Linear Integral Equations.
2.1.1 Fredholm integral equations
The most standard form of Fredholm linear integral equations is given by the form
(𝑥)𝑢(𝑥) = 𝑓(𝑥) + ∫ k(𝑥, t)u(t)dtb
a (3)
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2163
where the limits of integration a and b are constants and the unknown function 𝑢(𝑥) appears linearly
under the integral sign. K(𝑥,t) is the kernel of the equation
If (𝑥) 1, then (3) becomes
u(𝑥) = 𝑓(𝑥) + ∫ k(x, t)u(t)dtb
a (4)
and this equation is called Fredholm integral equation of second kind.
If (𝑥) 0, then (3) yields
𝑓(𝑥) + ∫ k(𝑥, t)u(t)dt = 0b
a (5)
which is called Fredholm integral equation of the first kind.
2.1.2 Volterra integral equations.
The most standard form of Volterra linear integral equations is of the form.
(𝑥)𝑢(𝑥) = 𝑓(𝑥) + ∫ k(𝑥, t)u(t)dt𝑥
a (6)
where the limits of integration are functions of 𝑥 and the unknown function 𝑢(𝑥) appears linearly under
the integral sign.
If the function (𝑥) 1, then (6) becomes
𝑢(𝑥) = 𝑓(𝑥) + ∫ k(𝑥, t)u(t)dt𝑥
a (7)
and this equation is known as the Volterra integral equation of the second kind.
If (𝑥) 0, then (6) becomes
𝑓(𝑥) + ∫ k(𝑥, t)u(t)dt = 0𝑥
a (8)
which is the Volterra integral equation of first kind.
2.1.3 Singular and Non – singular integral equations
A Singular integral equation is defined as an integral with the infinite limits when the Kernel
k(𝑥, t) of the integral becomes unbounded at a certain point in the interval.
Example 3: 𝑢(𝑥) = 𝑓(𝑥) + ∫ u(t)dt
−
If the Kernel k(𝑥, t)is bounded and continuous then the integral equation is said to be non – singular.
Remark;
If we set f(x)=0, in Volterra and Fredhlom integral equations then the resulting equation is
called a homogenous integral equation, otherwise it is called non-homogenous.
3. Relation Between Differential and Integral Equations
Integral and differential equations have a fundamental importance in functional analysis and
practice problems. But in many cases the resolution of differential equations with constant coefficients
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2164
is easy but the resolution of these equations with variable coefficients is practically difficult or
impossible in more part of the cases. Here we present an analytical method which transforms a
differential equation into an integral equation. Also, it presents methods to convert an integral equation
into a differential equation.
3.1 Transformation of Differential Equations into Integral Equations.
A boundary value problem of the form 𝑦′′ + 𝑦′f1(𝑥) + 𝑦𝑓2(𝑥) = 𝑓3(𝑥) with 𝑦 (𝑎) =
𝑦0 & 𝑦 (𝑏) = 𝑦1 can be transformed in to a Fredholm equation. An initial value problem of the form
𝑦′′ + 𝑦′𝑓1(𝑥) + 𝑦𝑓2(𝑥) = 𝑓3(𝑥) with 𝑦 (𝑎) = 𝑦0 & 𝑦′ (𝑎) = 𝑦1 can be transformed in to a Volterra
equation.
Example 4:
Construct the linear integral equation corresponding to the differential equation 𝑦′′ + 𝑥𝑦 = 1
with initial conditions 𝑦(0) = 𝑦′(0) = 0
Solution
Given 𝑦′′ = 1 − 𝑥𝑦
Integrate w.r.t x from 0 to x
∫ 𝑦′′(𝑥)𝑑𝑥 = ∫ (1 − 𝑥𝑦)𝑑𝑥𝑥
0
𝑥
0
𝑦′(𝑥)]0𝑥 = 𝑥]0
𝑥 - ∫ 𝑥𝑦(𝑥)𝑑𝑥𝑥
0
Integrate w.r.t x from 0 to x
∫ [𝑦′(𝑥) − 𝑦′(0)]𝑑𝑥 = ∫ (𝑥 − 0)𝑑𝑥 − ∫ [∫ 𝑥𝑦(𝑥)𝑑𝑥]𝑑𝑥𝑥
0
𝑥
0
𝑥
0
𝑥
0
𝑦(𝑥)]0𝑥 = [
𝑥2
2]
0
𝑥
- 1
(2−1)!∫ (𝑥 − 𝑡)2−1𝑥
0ty(t)dt
y(x) - y (0) = 𝑥2
2 -∫ (𝑥 − 𝑡)𝑡𝑦(𝑡)𝑑𝑡
𝑥
0
y(x) = 𝑥2
2 -∫ (𝑥 − 𝑡)𝑡𝑦(𝑡)𝑑𝑡
𝑥
0
which is the required linear Volterra integral equation.
Example 5:
Consider the differential equation 𝑦′′(𝑥) = 𝑓(𝑥) with boundary conditions
𝑦(0) = 0, 𝑦(1) = 0
Solution
Given 𝑦′′(𝑥) = 𝑓(𝑥)
Integrate w.r.t 𝑥, ∫ 𝑦′′(𝑥)𝑥
0𝑑𝑥 = ∫ 𝑓(𝑥)
𝑥
0𝑑𝑥
⟹ 𝑦′(𝑥)]0𝑥 = ∫ 𝑓(𝑥)
𝑥
0𝑑𝑥
Again integrating, ∫ [𝑦′(𝑥)𝑥
0− 𝑦′(0)]𝑑𝑥 = ∫ ∫ 𝑓(𝑥)𝑑𝑥𝑑𝑥
𝑥
0
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2165
=1
(2−1)!∫ ((𝑥 − 𝑡)2−1
𝑓(𝑡)𝑑𝑡𝑥
0
⟹ 𝑦(𝑥)]0𝑥 − 𝑥𝑦′(0)]0
𝑥 = ∫ (𝑥 − 𝑡)𝑥
0𝑓(𝑡)𝑑𝑡
⟹ 𝑦(𝑥) − 𝑦(0) − 𝑥𝑦′(0) = ∫ (𝑥 − 𝑡)𝑥
0𝑓(𝑡)𝑑𝑡
⟹ 𝑦(𝑥) = 𝑥𝑦′(0) + 𝑦(0) + ∫ (𝑥 − 𝑡)𝑥
0𝑓(𝑡)𝑑𝑡
Given that 𝑦(1) = 0
𝑦(1) = 𝑦′(0) + 𝑦(0) + ∫ (1 − 𝑡)𝑓(𝑡)𝑑𝑡1
0
⟹ 𝑦′(0) = 𝑦(1) − 𝑦(0) − ∫ (1 − 𝑡)𝑓(𝑡)𝑑𝑡1
0
⟹ 𝑦′(0) = 0 − 0 − ∫ (1 − 𝑡)𝑓(𝑡)𝑑𝑡1
0
⟹ 𝑦(𝑥) = −𝑥 ∫ (1 − 𝑡)𝑓(𝑡)𝑑𝑡1
0+ ∫ (𝑥 − 𝑡)𝑓(𝑡)𝑑𝑡
𝑥
0
= ∫ (𝑥 − 𝑡)𝑓(𝑡)𝑑𝑡 − 𝑥 [∫ (1 − 𝑡)𝑓(𝑡)𝑑𝑡 + ∫ (1 − 𝑡)𝑓(𝑡)𝑑𝑡1
𝑥
𝑥
0
]
𝑥
0
= ∫ (𝑥 − 𝑡 − 𝑥 + 𝑥𝑡)𝑓(𝑡)𝑑𝑡 − 𝑥 ∫ (1 − 𝑡)𝑓(𝑡)𝑑𝑡1
𝑥
𝑥
0
= ∫ 𝑡(𝑥 − 1)𝑓(𝑡)𝑑𝑡 + ∫ 𝑥(𝑡 − 1)𝑓(𝑡)𝑑𝑡1
𝑥
𝑥
0
= ∫ [𝑡(𝑥 − 1) + (𝑡 − 1)𝑥]𝑓(𝑡)𝑑𝑡1
0
= ∫ 𝑘(𝑥, 𝑡)𝑓(𝑡)𝑑𝑡1
0
; 𝑘(𝑥, 𝑡) = 𝑓(𝑥) = {(𝑥 − 𝑡)𝑡; 𝑡 < 𝑥(𝑡 − 1)𝑥; 𝑥 < 𝑡
3.2 Transformation of Integral Equations into Differential Equations
If the upper limit is 𝑥 & Kernel doesn’t depend on 𝑥, then Volterra equation is
equivalent to an ordinary differential equation.
(𝑥) = 𝑓(𝑥) + ∫ 𝑘(𝑥, 𝑡)𝑥
0(𝑡)𝑑𝑡
⟹ ′(𝑥) = 𝑓′(𝑥) + 𝑘(𝑥)(𝑥)
If the upper limit is 𝑥 and kernel depends on 𝑥 then it is a possible to reduce the integral
equation to an ordinary differential equation by differentiating several times w.r.t. 𝑥
Example 6:
Convert 𝑦(𝑥) = − ∫ (𝑥 − 𝑡)𝑦(𝑡)𝑑𝑡𝑥
0 into an initial value problem.
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2166
Solution
𝑦(𝑥) = − ∫ (𝑥 − 𝑡)𝑦(𝑡)𝑑𝑡𝑥
0 (9)
Differentiating equation (9) w.r.t 𝑥,
𝑦′(𝑥) = −𝑑
𝑑𝑥∫ (𝑥 − 𝑡)𝑦(𝑡)𝑑𝑡
𝑥
0
𝑦′(𝑥) = − ∫ 𝑦(𝑡)𝑑𝑡𝑥
0 (10)
Again, differentiating equation (10) w.r.t 𝑥 will give
𝑦′′(𝑥) = −𝑑
𝑑𝑥∫ 𝑦(𝑡)𝑑𝑡
𝑥
0
𝑦′′(𝑥) = −𝑦(𝑥)
𝑦′′(𝑥) + 𝑦(𝑥) = 0 (11)
Putting the lower limit 𝑥 = 0 (𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑣𝑎𝑙𝑢𝑒) the equations (9) & (10) will give respectively the
following.
𝑦(0) = − ∫ (0 − 𝑡)0
0𝑦(𝑡)𝑑𝑡
𝑦(0) = 0 (12)
𝑦′(0) = − ∫ 𝑦0
0(𝑡)𝑑𝑡
𝑦′(0) = 0 (13)
(11), (12) & (13) form the ordinary differential form of given integral equation.
4. Numerical Solutions of Integral Equations
Now we discuss about the different methods to solve integral equations. First, we shall discuss
the various methods of solutions of the Fredholm integral equation. Then we’ll study the different
methods to solve Volterra integral equations.
4.1 Numerical Methods for Fredholm Equations
There exist several methods for the numerical solutions of Fredholm equations like the method of
degenerate Kernels, methods of successive approximations, method of successive substitutions etc.
4.1.1 Method of Degenerate Kernels
We consider the integral equation
𝑓(𝑥) − ∫ 𝑘(𝑥, 𝑡)𝑏
𝑎𝑓(𝑡)𝑑𝑡 = (𝑥) (14)
A Kernel 𝑘(𝑥, 𝑡) is said to be degenerate it can be expressed in the form.
𝑘(𝑥, 𝑡) = ∑ ui(x)vi(t)ni=1 (15)
Substituting this in (9) we obtain
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2167
𝑓(𝑥) − ∑ ∫ ui(x)vi(t)𝑓(𝑡)b
a𝑑𝑡 = (𝑥)𝑛
𝑖=1 (16)
Setting ∫ 𝑣𝑖(𝑡)𝑓(𝑡)𝑑𝑡 =𝑏
𝑎𝐴𝑖 (17)
equation (11) gives
𝑓(𝑥) = ∑ 𝐴𝑖 𝑢𝑖 (𝑥) + (𝑥)𝑛𝑖=1 (18)
The constants 𝐴𝑖 are still to be determined, but substituting from (18) in (17) we get
∫ 𝑣𝑖(𝑡)[∑ 𝐴𝑗 𝑢𝑗 (𝑡) + (𝑡)𝑛𝑗=1 ]𝑑𝑡 =
𝑏
𝑎𝐴𝑖
Or
∑ 𝐴𝑗 ∫ 𝑣𝑖(𝑡)𝑏
𝑎𝑢𝑗 (𝑡)𝑑𝑡 +𝑛
𝑗=1 ∫ 𝑣𝑖(𝑡)𝑏
𝑎(𝑡)𝑑𝑡 = 𝐴𝑖
which represent a system of n equations in the n unknowns 𝐴1,𝐴2 … … … 𝐴𝑛. When the 𝐴𝑖 𝑠 are
determined, equation (18) then gives 𝑓(𝑥).
Example 7:
We consider the equation 𝑓(𝑥) − ∫ sin 𝑥 𝑐𝑜𝑠𝑡𝜋
20
𝑓(𝑡)𝑑𝑡 = sin 𝑥
Setting ∫ 𝑐𝑜𝑠𝑡𝜋
20
𝑓(𝑡)𝑑𝑡 = 𝐴, the integral equation becomes
𝑓(𝑥) = 𝐴 𝑠𝑖𝑛 𝑥 + sin 𝑥 = (𝐴 + 1) sin 𝑥
Substituting this in above equation we get
∫ 𝑐𝑜𝑠𝑡(𝐴 + 1)𝑠𝑖𝑛𝑡 𝑑𝑡 = 𝐴𝜋/2
0
⟹ ∫ 𝑢𝑑𝑢 =𝐴
𝐴+1
1
0 ; 𝑢 = 𝑠𝑖𝑛𝑡
⟹ 𝑢2
2]
0
1
= 𝐴
𝐴+1
⟹ 1
2 =
𝐴
𝐴+1
⟹ 𝐴 = 1
2−
Hence 𝑓(𝑥) = (
2−+ 1) sin 𝑥
Therefore, the solution of the integral equation is
𝑓(𝑥) =2
2−sin 𝑥 ( ≠ 2)
4.1.2 Method of Successive Substitutions
Let 𝑢 (𝑥) = 𝑓(𝑥) + ∫ 𝑘(𝑥, 𝑡)𝑢 (𝑡)𝑑𝑡𝑏
𝑎 (19)
In equation (19) as usual k(𝑥, 𝑡) ≠ 0 , is real & continuous in the rectangle R, for which 𝑎 ≤ 𝑥 ≤ 𝑏,
𝑎 ≤ 𝑡 ≤ 𝑏; 𝑓(𝑥) ≠ 0 is real and continuous in the interval I for which 𝑎 ≤ 𝑥 ≤ 𝑏 𝑎𝑛𝑑 , a constant
parameter. Substituting in the 2nd member of (19), in place of 𝑢 (𝑡), its value as given by the equation
itself yields
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2168
u(x) = 𝑓(𝑥) + ∫ k(x, t)𝑏
𝑎
𝑓(𝑡)𝑑𝑡 + 2 ∫ k(x, t)
𝑏
𝑎
∫ k(t, 𝑡1)u(t1)𝑏
𝑎
𝑑𝑡1𝑑𝑡
Here, again we substitute for u(𝑡1) its value as given in equation (19). Thus, we get,
u(x) = 𝑓(𝑥) + ∫ k(x, t)𝑏
𝑎
𝑓(𝑡)𝑑𝑡 + 2 ∫ k(x, t)
𝑏
𝑎
∫ k(t, 𝑡1)𝑏
𝑎
𝑓(𝑡1)𝑑𝑡1𝑑𝑡 +
3 ∫ k(x, t)
𝑏
𝑎
∫ k(t, 𝑡1) ∫ k(𝑡1, 𝑡2)u𝑏
𝑎
𝑏
𝑎
(𝑡2)𝑑𝑡2𝑑𝑡1𝑑𝑡
Proceeding like this we get
u(x) = 𝑓(𝑥) + ∫ k(x, t)𝑏
𝑎
𝑓(𝑡)𝑑𝑡 + 2 ∫ k(x, t)
𝑏
𝑎
∫ k(t, 𝑡1)𝑏
𝑎
𝑓(𝑡1)𝑑𝑡1𝑑𝑡 +
3
∫ k(x, t)𝑏
𝑎 ∫ k(t, 𝑡1) ∫ k(𝑡1, 𝑡2)𝑏
𝑎
𝑏
𝑎𝑓(𝑡2)𝑑𝑡2𝑑𝑡1𝑑𝑡 + ⋯ (20)
The series solution given in (20) converges uniformly in the interval [a, b] if
𝑀(b − a) < 1 where |k (x, t) |≤ 𝑀
Example 8:
Consider the Fredholm equation 𝑢(𝑥) = cos 𝑥 +1
2∫ sin 𝑥
𝜋
20
u(t)dt
Here =1
2, f(x) = cos 𝑥 & k(x, t) = sin 𝑥 and substituting these values in equation (20) yields.
𝑢(𝑥) = cos 𝑥 +1
2∫ sin 𝑥
𝜋
2
0
costdt
+1
4∫ sin 𝑥
𝜋
2
0
∫ sin 𝑡 cos 𝑡1𝑑𝑡1𝑑𝑡
𝜋
2
0
+ 1
8∫ sin 𝑥
𝜋
2
0
∫ sin 𝑡 ∫ sin 𝑡1 cos 𝑡2𝑑𝑡2𝑑𝑡1𝑑𝑡 +
𝜋
2
0
…
𝜋
2
0
= cos 𝑥 +1
2𝑠𝑖𝑛 𝑥 +
1
4𝑠𝑖𝑛𝑥 +
1
8𝑠𝑖𝑛𝑥 + ⋯
= cos 𝑥 + sin 𝑥
4.2 Numerical Method for Volterra Equations
Here we discuss three methods: series solutions method, differentiation method and method of
resolvant kernels.
4.2.1 Series Solutions of Volterra Equations
Consider a simple Volterra equation (𝑥) = 𝑥 + ∫ 𝑥
0(𝑡)𝑑𝑡. For small ,
0
(𝑥) = 𝑥 is a first approximation. Insert this in the integral term to get a better approximation.
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2169
1
(𝑥) = 𝑥 + ∫ 𝑡𝑑𝑡 = 𝑥 +𝑥2
2
𝑥
0
Again, put this in to the integral to get
2
(𝑥) = 𝑥 + ∫ (𝑡 +1
2𝑡2)𝑑𝑡
𝑥
0
= 𝑥 + [t2
2]
0
x
+
2
2[
𝑡3
3]
0
𝑥
= 𝑥 +𝑥2
2+
2𝑥3
6
Continuing like this we get,
𝑛
(𝑥) = 𝑥 +𝑥2
2+ ⋯
1
𝑛!
𝑛−1𝑥𝑛
We let 𝑛 → , the series converges to 𝑒𝑥−1
. Substituting into the equation verifies that this is
the correct solution. So, for this Volterra equation the technique of expansion in series gives a result
which is convergent for all .
We now show that this works for all Volterra equations, subject to some fairly general
conditions. Suppose that we look for a solution with 𝑥 in some finite interval [0,1]& and that on this
interval f(x) is bounded with [𝑎, 𝑏]x[𝑎, 𝑏]; 𝑘[𝑥, 𝑠] < 𝑚
Then |0(𝑥)| < |𝑓(𝑥)| < 𝑀
|1(𝑥)| = |𝑓(𝑥) + ∫ 𝑘(𝑥, 𝑠)𝑓(𝑠)𝑑𝑠
𝑥
0
|
< 𝑚 + ||𝑚𝑀(𝑥 − 𝑎)
|2(𝑥)| < |𝑓(𝑥) + ∫ 𝑘(𝑥, 𝑠)
1(𝑥)𝑑𝑠
𝑥
0
|
< 𝑚 + ∫ 𝑀(𝑚 + 𝑚𝑀(𝑠 − 𝑎)𝑥
0
𝑑𝑠
= 𝑚(1 + ||𝑀(𝑥 − 𝑎) + |2|𝑀2 (𝑥−𝑎)2
2)
Carrying on like this we get
|𝑛(𝑥)| < 𝑚(1 + ||)𝑀(𝑥 − 𝑎) + |2|𝑀2 (𝑥−𝑎)2
2+ ⋯
1
𝑛!| |𝑛𝑀𝑛(𝑥 − 𝑎)𝑛)
Since this series here is of exponential type, we get convergence for all values of
Example 9:
(𝑥) = ∫ 1
0(𝑠)𝑑𝑠
Now we get 0
(𝑥) = 𝑥
1
(𝑥) = 𝑥 + ∫ 1
0(𝑠)𝑑𝑠
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2170
= 𝑥 + ∫ 𝑠𝑑𝑠1
0
= 𝑥 + [𝑠2
2]
0
1
= 𝑥 +
2
2
(𝑥) = 𝑥 + ∫ (𝑠 +
2)𝑑𝑠
1
0
= 𝑥 + [𝑠2
2+s
2]
0
1
= 𝑥 + (1
2+
2)
= 𝑥 +
2+2
2
Continuing this way, 𝑛
(𝑥) = 𝑥 +1
2(+
2 + ⋯ + 𝑛)
4.2.2 Method of Differentiation
A Volterra equation with a simple separable Kernel can be solved by reducing it to a differential
equation.
Example 10:
Consider (𝑥) = 𝑥5 + ∫ 𝑥𝑠2𝑥
0(𝑠)𝑑𝑠
We can divide this through by 𝑥, so that the integral term does not depend on 𝑥, getting
(𝑥)
𝑥= 𝑥4 + ∫
𝑥
0
(𝑠)𝑠2𝑑𝑠
Differentiating w.r.t x gives
𝑑
𝑑𝑥((𝑥)
𝑥) = 4𝑥3 + 𝑥2(𝑥)
= 4𝑥3 + 𝑥3((𝑥)
𝑥)
which is a simple linear differential equation. We get
𝑑
𝑑𝑥((𝑥)
𝑥) 𝑒−
1
4𝑥4
= 4𝑥3𝑒−1
4𝑥4
And so (𝑥) = −4𝑥 + 𝑐𝑥𝑒−1
4𝑥4
This involves an arbitrary constant, whereas a Volterra integral equation has an unique solution. We
can evaluate the constant by going back to the integral equation. The condition that (𝑥) = 0 when 𝑥 =
0 tells us nothing, but if we use the fact that ∅(𝑥)
𝑥→ as 𝑥 → 0 we see that the solution is
(𝑥) = −4𝑥 + 4𝑥𝑒−1
4𝑥4
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2171
5. Applications of Integral Equations
The study of integral equations has a significant role in both pure and applied field of mathematics.
The development of science has led to the formation of many physical laws, which when restated in
mathematical form, often appear as differential equations. Engineering problems can be mathematically
described by differential equations and thus differential equations play very important roles in the
solution of practical problems. For example, Newton’s law, stating that the rate of change of the
momentum of a particle is equal to the force acting on it, can be translated into mathematical language
as a differential equation. Similarly, problems arising in electric circuits, chemical kinetics and transfer
of heat in a medium can all be represented mathematically as differential equations. These differential
equations can be transformed to the equivalent integral equations of Volterra and Fredholm types. There
are many physical problems that are governed by integral equations and these equations can be easily
transformed to the differential equation.
Applications of some special integral equations are tested below.
5.1 Abel’s integral equation (𝑓(𝑥) = ∫ k(𝑥, 𝑡)(𝑡)𝑑𝑡𝑥
0)
This equation arises in the problem of finding the path of a particle which is constrained to
move under gravity in a vertical plane.
5.2 Vandrey’s equation (𝑉(𝑠) = (𝑠) −
𝜋∫ k(𝑠, 𝜎)𝑣(𝜎)𝑑𝜎; 0 ≤ 𝑠 ≤ 𝐿
𝐿
0)
This equation occurs in fluid dynamics while calculating the pressure distribution on the surface of the
body of revolution moving in a fluid.
5.3 Seismic Response of Dams
In order to analyze the safety and stability of an earth dam during an earthquake we need to
know the response of the dam to earthquake ground motion so that the inertia forces that will be
generated in the dam by the earthquake can be derived. Once the inertia forces are known, the safety
and the stability of the structure can be determined.
The inertia forces generated during an earthquake will depend on the geometry of the dam, the material
properties, the earthquake time history.
The reality of the problem is that an earth dam is a three dimensional structure. The material
properties are non- linear inelastic and the earthquake time history is a time varying phenomenon. The
problem is therefore very complex and a proper solution requires the use of a finite element program,
which can deal with nonlinear inelastic material properties. The earthquake is the travelling wave
phenomenon, which arrives at the base of the dam through the foundation rock. Since the foundation is
not rigid, part of the energy, which vibrates the dam is lost through the foundation causing radiation
damping.
The analytical solution to such a problem is not at all possible. However, it is often necessary
to have approximate solutions that can be used to understand the behaviour of the dam during
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2172
earthquakes. In order to make the problem amenable to analytical solutions, some approximations are
made to create a mathematical model. Such a model in this case is known as the shear beam model (SB)
of earth dams. Here we formulate a differential equation using the ideas from physics such as shear
modulus, strain, inertia, acceleration, momentum and so on. And we convert this differential equation
to an integral equation to get solution in an easier way.
5.5 Transverse Oscillations of a Homogenous Elastic Bar
Consider a homogeneous elastic bar with linear mass density d. Its axis coincides
with the segment (0, 𝑙) of the s axis when the bar is in its state of rest. It is clamped at
the end 𝑠 = 0, free at the end 𝑠 = 𝑙, and is forced to perform simple harmonic
oscillations with period 2𝜋𝜔⁄ , 𝜔 is the angular frequency.
s
𝑦(𝑠)
L
y
Figure 1: Transverse oscillations of a homogenous elastic bar
The problem is to find the deflection 𝑦(𝑠) that is parallel to the y axis and satisfies the system
of equations
𝑑4𝑦
𝑑𝑠4− 𝑘4𝑦 = 0, 𝑘4 =
𝜔2𝑑
𝐸𝐼
𝑦(0) = 𝑦 ,(0) = 0, 𝑦 ,,(𝑙) = 𝑦 ,,,(𝑙) = 0
Where 𝐸𝐼 is the bending rigidity of the bar
The above differential equation with initial conditions can be reduced to 𝑦 ,,(0) = 𝑐2 𝑦 ,,,(0) = 𝑐3, then
we obtain the required integral equation as
𝑔(𝑠) = 𝑘4 (𝑠2
2!𝑐2 +
𝑠3
3!𝑐3) + 𝑘4 ∫
(𝑠 − 𝑡)3
3!
𝑠
0
𝑔(𝑡)𝑑𝑡
Where, 𝑔(𝑠) =𝑑4𝑦
𝑑𝑠4
Then the solution 𝑦(𝑠) of the above differential equation is given by
𝑦(𝑠) = ∫(𝑠 − 𝑡)3
3!
𝑠
0
𝑔(𝑡)𝑑𝑡 +𝑠3
3!𝑐3 +
𝑠2
2!𝑐2
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2173
Here we can find the constants 𝑐2 and 𝑐3 from given conditions. Hence we can transform given
differential equation with initial conditions in to an integral equation and obtain the solution easily.
6. Conclusion
This study briefly explains the different concepts associated with an integral equation,
especially linear integral equation. It mainly points towards linear integral equations such as Fredholm
and Volterra integral equations. There are a number of methods to solve the two types of integral
equations. We have discussed a few among them in this study. One obvious reason for using integral
equations rather than differential equations is that all of the conditions specifying the initial value
problem or boundary value problem for a differential equation can often be condensed into a single
integral equation.
Linear integral equations have undoubtedly a wide range of applications in all fields of science.
In this work, we have discussed its importance in the study of seismic response of dams and transverse
oscillations of a homogeneous elastic bar. Although the idea of linear integral equations can be extended
to a wide range of real-world problems such as to study interaction of ocean waves, Flow of heat in a
metal bar etc.
References;
1. Rahman M., Integral equation and their Applications, WIT press Southampton, Boston, 2007
2. Ram P. Kanwal, Linear Integral Equations Theory and Technique, Academic press New
York,1971.
3. James S. Scarborough, Numerical Mathematical Analysis, 6th edition.
4. Kreyzig Erwin, Introductory functional analysis with applications, John Wiley & Sons, 1978.
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2174
The International journal of analytical and experimental modal analysis
Volume XI, Issue VIII, August/2019
ISSN NO: 0886-9367
Page No:2175