Advances in Difference Equations

, 2013:123

Numerical solution of linear integral equations system using the Bernstein collocation method

  • Ahmad Jafarian
  • Safa A Measoomy Nia
  • Alireza K Golmankhaneh
  • Dumitru Baleanu
Open Access
Research Article

DOI: 10.1186/1687-1847-2013-123

Cite this article as:
Jafarian, A., Measoomy Nia, S.A., Golmankhaneh, A.K. et al. Adv Differ Equ (2013) 2013: 123. doi:10.1186/1687-1847-2013-123
Part of the following topical collections:
  1. Progress in Functional Differential and Difference Equations

Abstract

Since in some application mathematical problems finding the analytical solution is too complicated, in recent years a lot of attention has been devoted by researchers to find the numerical solution of this equations. In this paper, an application of the Bernstein polynomials expansion method is applied to solve linear second kind Fredholm and Volterra integral equations systems. This work reduces the integral equations system to a linear system in generalized case such that the solution of the resulting system yields the unknown Bernstein coefficients of the solutions. Illustrative examples are provided to demonstrate the preciseness and effectiveness of the proposed technique. The results are compared with the exact solution by using computer simulations.

1 Introduction

As a matter of fact, it might be said that many phenomena of almost all practical engineering and applied science problems like physical applications, potential theory and electrostatics are reduced to solving integral equations. Since these equations usually cannot be solved explicitly, so it is required to obtain approximate solutions. There are numerous numerical methods which have been focusing on the solution of integral equations. For example, Tricomi in his book [1], introduced the classical method of successive approximations for integral equations. Variational iteration method [2] and Adomian decomposition method [3] were effective and convenient for solving integral equations. Also, the Homotopy analysis method (HAM) was proposed by Liao [4] and then has been applied in [5]. Taylor expansion approach was presented for solving integral equations by Kanwal and Liu in [6] and then has been extended in [7, 8]. In addition, Babolian et al. [9] solved some integral equations systems by using the orthogonal triangular basis functions. Jafari et al. [10] applied Legendre wavelets method to find numerical solution system of linear integral equations. Moreover, some different valid methods for solving this kind of equations have been developed. First time, the Bernstein polynomials have been used for the solution of some linear and nonlinear differential equations in [11, 12, 13, 14]. Mandal and Bhattacharya [15] obtained approximate numerical solutions of some classes of integral equations by using Bernstein polynomials. Also, they used these polynomials to approximate solution of linear Volterra integral equations [16]. In addition, Maleknejad et al. [17] has applied the polynomials for solving Volterra integral equations of the second kind. Furthermore, in [18] an architecture of artificial neural networks (ANNs) was suggested to approximate solution of linear integral equations systems. For this aim, first the truncations of the Taylor expansions for unknown functions were substituted in the original system. Then the proposed neural network has been applied for adjusting the real coefficients of the given expansions in the resulting system.

In this paper, we are going to propose a new numerical approach to approximate the solutions of linear Fredholm and Volterra integral equations systems of the second kind. This method converts the given systems with unique solutions, into a system of linear algebraic equations in generalized case. To do this, first the Bernstein polynomials of certain degree n of unknown functions are substituted in the given integral equations system. Suppose that the given closed interval [ a , b ] Open image in new window is partitioned into uniform spacing ( b a ) / n Open image in new window and nodes t i = a + i h Open image in new window (for i = 0 , , n Open image in new window). If we put t = t i Open image in new window (for i = 0 , , n Open image in new window), the given system of integral equations, yields a linear algebraic system. The solution of the resulting system yields the unknown Bernstein coefficients of the solution functions.

Here is an outline of the paper. Section 2 describes how to find approximate solutions of the given linear integral equations systems by using the proposed approach. In Section 3, the convergence of the method is established for each class of integral equations systems. Finally in Section 4, two numerical examples are provided and the results are compared with the analytical solutions to demonstrate the validity and applicability of the method. Also, a comparison is made with other numerical approaches that were proposed recently for solving the given systems.

2 The general method

The basic definition of integral equation is given in [15, 17, 19]. In this section, we intend to use the Bernstein polynomials to get a new numerical method for solving the linear Fredholm and Volterra integral equations systems of the second kind. In other words, it will be described how to apply these polynomials for approximating solutions of the unknowns in the systems.

2.1 System of the Fredholm integral equations

In this subdivision, we want to obtain a numerical solution of the linear Fredholm integral equations system of the second kind in the form
j = 1 m ( A i j ( t ) F j ( t ) ) = f i ( t ) + j = 1 m ( λ i j a t k i j ( s , t ) F j ( s ) d s ) , i = 1 , , m Open image in new window
(1)
or
{ j = 1 m ( A 1 j ( t ) F j ( t ) ) = f 1 ( t ) + j = 1 m ( λ 1 j a b k 1 j ( s , t ) F j ( s ) d s ) , j = 1 m ( A i j ( t ) F j ( t ) ) = f i ( t ) + j = 1 m ( λ i j a b k i j ( s , t ) F j ( s ) d s ) , j = 1 m ( A m j ( t ) F j ( t ) ) = f m ( t ) + j = 1 m ( λ m j a b k m j ( s , t ) F j ( s ) d s ) . Open image in new window
(2)
Let us consider
F j , n ( t ) = p = 0 n a j , p B n , p ( t ) , Open image in new window
(3)
where
a j , p = F j ( a + p h ) and h = ( b a ) / n , Open image in new window
which are the Bernstein expansions of degree n for the unknown functions F j ( t ) Open image in new window for j = 1 , , m Open image in new window. After substituting these polynomials instead of the unknowns in the system (2), we have:
{ j = 1 m ( A 1 , j ( t ) p = 0 n a j , p B p , n ( t ) ) = f 1 ( t ) + j = 1 m ( λ 1 , j a b k 1 , j ( s , t ) p = 0 n a j , p B p , n ( s ) d s ) , j = 1 m ( A i , j ( t ) p = 0 n a j , p B p , n ( t ) ) = f i ( t ) + j = 1 m ( λ i , j a b k i , j ( s , t ) p = 0 n a j , p B p , n ( s ) d s ) , j = 1 m ( A m , j ( t ) p = 0 n a j , p B p , n ( t ) ) = f m ( t ) + j = 1 m ( λ m , j a b k m , j ( s , t ) p = 0 n a j , p B p , n ( s ) d s ) . Open image in new window
(4)
In order to find a j , n Open image in new window in Eq. (3), the system (4) is converted to an algebraic system of linear equations by replacing t with t p = a + p ( b a ) / n Open image in new window (for p = 0 , , n Open image in new window). Notice that in this way we can skip the singularity problem. After this work, the system is transformed to the following form:
j = 1 m p = 0 n ( A i , j ( t q ) B p , n ( t q ) ) a j , p = f i ( t q ) + j = 1 m p = 0 n ( λ i , j a b k i , j ( s , t q ) B p , n ( s ) d s ) a j , p , i = 1 , , m ; q = 0 , , n . Open image in new window
(5)
For brevity, we define below symbols as:
T q , p ( i , j ) = A i , j ( t q ) B p , n ( t q ) , R q , p ( i , j ) = λ i j a b k i j ( s , t q ) B p , n ( s ) d s . Open image in new window
Now we can write the system (5) in the form
j = 1 m p = 0 n T q , p ( i , j ) a j , p = f i ( t q ) + j = 1 m p = 0 n R q , p ( i , j ) a j , p , i = 1 , , m ; q = 0 , , n . Open image in new window
(6)
Consequently, the expression (6) can be summarized in a matrix form as follows:
W Y = E + V Y , Open image in new window
(7)
where
Y = [ a 1 , 0 , , a 1 , n , , a i , 0 , , a i , n , , a m , 0 , , a m , n ] T , E = [ f 1 ( t 0 ) , , f 1 ( t n ) , , f i ( t 0 ) , , f i ( t n ) , , f m ( t 0 ) , , f m ( t n ) ] T , W = [ W ( 1 , 1 ) W ( 1 , m ) W ( m , 1 ) W ( m , m ) ] and V = [ V ( 1 , 1 ) V ( 1 , m ) V ( m , 1 ) V ( m , m ) ] . Open image in new window
Parochial matrices W ( i , j ) Open image in new window, V ( i , j ) Open image in new window for ( i , j = 1 , , m Open image in new window) are defined with following elements:
W ( i , j ) = [ W 0 , 0 ( i , j ) W 0 , n ( i , j ) W n , 0 ( i , j ) W n , n ( i , j ) ] , V ( i , j ) = [ V 0 , 0 ( i , j ) V 0 , n ( i , j ) V n , 0 ( i , j ) V n , n ( i , j ) ] , Open image in new window
where
W q , p ( i , j ) = T q , p ( i , j ) and V q , p ( i , j ) = R q , p ( i , j ) . Open image in new window

The resulting generalized linear system can be solved for a j , p Open image in new window for j = 1 , , m Open image in new window; p = 0 , , n Open image in new window by a standard method, and hence F j ( t ) Open image in new window is obtained.

2.2 System of Volterra integral equations

At first, consider again the system of linear Volterra integral equations (1). For numerical solving of the present system, the unknown function F j ( t ) Open image in new window is approximated by its Bernstein approximation (3). Now we have the following system:
{ j = 1 m p = 0 n ( A 1 , j ( t ) B p , n ( t ) ) a j , p = f 1 ( t ) + j = 1 m p = 0 n ( λ 1 , j a t k 1 , j ( s , t ) B p , n ( s ) d s ) a j , p , j = 1 m p = 0 n ( A i , j ( t ) B p , n ( t ) ) a j , p = f i ( t ) + j = 1 m p = 0 n ( λ i , j a t k i , j ( s , t ) B p , n ( s ) d s ) a j , p , j = 1 m p = 0 n ( A m , j ( t ) B p , n ( t ) ) a j , p = f m ( t ) + j = 1 m p = 0 n ( λ m , j a t k m , j ( s , t ) B p , n ( s ) d s ) a j , p . Open image in new window
(8)
Similarly, by replacing the variable t with t p = a + p ( b a ) / n Open image in new window for p = 0 , , n Open image in new window, we obtain the generalized linear system
j = 1 m p = 0 n T q , p ( i , j ) a j , p = f i ( t q ) + j = 1 m p = 0 n R q , p ( i , j ) a j , p , i = 1 , , m ; q = 0 , , n , Open image in new window
(9)
where
T q , p ( i , j ) = A i , j ( t q ) B p , n ( t q ) , R q , p ( i , j ) = λ i j a t k i j ( s , t q ) B p , n ( s ) d s . Open image in new window
Consequently, the expression (6) can be summarized in a matrix form as follows:
[ W ( 1 , 1 ) W ( 1 , m ) W ( m , 1 ) W ( m , m ) ] [ Y 1 Y m ] = [ E 1 E m ] + [ V ( 1 , 1 ) V ( 1 , m ) V ( m , 1 ) V ( m , m ) ] [ Y 1 Y m ] , Open image in new window
where
Y i = [ a i , 0 a i , n ] , E i = [ f i ( t 0 ) f i ( t n ) ] , W ( i , j ) = [ W 0 , 0 ( i , j ) W 0 , n ( i , j ) W n , 0 ( i , j ) W n , n ( i , j ) ] Open image in new window
(10)
and
V ( i , j ) = [ V 0 , 0 ( i , j ) V 0 , n ( i , j ) V n , 0 ( i , j ) V n , n ( i , j ) ] . Open image in new window
In the above generalized linear system, the symbols are defined as follows:
W q , p ( i , j ) = T q , p ( i , j ) and V q , p ( i , j ) = R q , p ( i , j ) . Open image in new window

3 Convergence analysis

In this section, we prove that the present numerical method converges to the exact solutions of the systems (2) and (1).

Theorem 1Let F j , n ( t ) Open image in new windowfor j = 1 , , m Open image in new windowbe the Bernstein polynomials of degreensuch that their coefficients have been produced by solving the generalized linear system (7). Then the given polynomials converge to the exact solution of the Fredholm integral equations system (2), when n + Open image in new window.

Proof Consider the system (2). Since the series (3) converge to F j ( t ) Open image in new window for j = 1 , , m Open image in new window, respectively, then we conclude that:
j = 1 m ( A i j ( t ) F j , n ( t ) ) = f i ( t ) + j = 1 m ( λ i j a b k i j ( s , t ) F j , n ( s ) d s ) , i = 1 , , m , Open image in new window
(11)
and it holds that
F j ( t ) = lim n F j , n ( t ) . Open image in new window
We defined the error function e n ( t ) Open image in new window by subtracting Eqs. (2) and (11) as follows:
e n ( t ) = i = 1 m e i , n ( t ) , Open image in new window
(12)
where
e i , n ( t ) = j = 1 m A i , j ( t ) ( F j ( t ) F j , n ( t , r ) ) j = 1 m λ i j ( a b k i , j ( s , t ) ( F j ( s ) F j , n ( s ) ) d s ) . Open image in new window
We must prove that when n + Open image in new window, the error function e n ( t ) Open image in new window tends to zero. Hence, we proceed as follows:
e n i = 1 m e i , n i = 1 m j = 1 m ( A i , j ( t ) ( F i ( t ) F i , n ( t ) ) ) + i = 1 m j = 1 m ( | λ i , j | a b k i , j F j ( s ) F j , n ( s ) d s ) . Open image in new window

Since k i , j Open image in new window and A j Open image in new window are bounded, therefore, F j ( s ) F j , n ( s ) 0 Open image in new window implies that e n 0 Open image in new window and the proof is completed. □

Theorem 2Suppose that F j , n ( t ) Open image in new windowfor j = 1 , , m Open image in new windoware the Bernstein polynomials of degreensuch that their coefficients have been produced by solving the generalized linear system (10). Then the given polynomials converge to the exact solution of the Volterra integral equations system (1), when n + Open image in new window.

Proof Consider the system (1). Using a similar procedure as an outline in mentioned Theorem 1, we have the following corollary in which we are refrained from going through proof details. Now the error of the approximation method can be written as
e n ( t ) = i = 1 m e i , n ( t ) , Open image in new window
(13)
where
e i , n ( t ) = j = 1 m A i , j ( t ) ( F j ( t ) F j , n ( t , r ) ) j = 1 m λ i j ( a t k i , j ( s , t ) ( F j ( s ) F j , n ( s ) ) d s ) . Open image in new window
Due to Theorem 1, the error function e n ( t ) Open image in new window must tend to zero, when n + Open image in new window. Hence, we proceed as follows:
e n i = 1 m e i , n i = 1 m j = 1 m ( ( F i ( t ) F i , n ( t ) ) ) + i = 1 m j = 1 m ( | λ i , j | a t k i , j ( F j ( s ) F j , n ( s ) ) d s ) . Open image in new window

Since k i , j Open image in new window and A j Open image in new window are bounded, therefore, ( F j ( s ) F j , n ( s ) ) 0 Open image in new window implies that e n 0 Open image in new window and the proof follows immediately. The stability and the convergence of Bernstein polynomials is studied in [12, 13]. □

4 Numerical examples

In this section, in order to investigate the accuracy of the proposed method, we have chosen three examples of linear integral equations systems of the second kind. Also, to show the efficiency of the present method for our problem, results will be compared with the exact solution. Moreover, the present method is compared with artificial neural network (ANN) method and the trapezoidal quadrature rule (TQR) [18].

Example 1 Consider the system of linear Fredholm integral equations
{ 2 F 1 ( t ) + 3 F 2 ( t ) = f 1 ( t ) + 0 1 ( t + s ) F 1 ( s ) d s + 0 1 t s F 2 ( s ) d s , 3 F 1 ( t ) 4 F 2 ( t ) = f 2 ( t ) + 0 1 ( 2 t s 1 ) F 1 ( s ) d s + 0 1 ( t s ) F 2 ( s ) d s Open image in new window
with
f 1 ( t ) = 2 e t 3 t 2 t ( e 1 ) t ( ln ( 4 ) 1 ) 1 , f 2 ( t ) = 2 t + ln ( 4 ) + 3 e t + 4 t 2 t ln ( 2 ) 2 + e , Open image in new window
where the exact solution is F 1 ( t ) = e t Open image in new window and F 2 ( t ) = 1 2 t Open image in new window. In this example, we illustrate the use of the present technique to approximate solution of this integral equations system. Using Eq. (7), the coefficients matrices W, V and E are calculated for n = 3 Open image in new window as following:
W = [ W ( 1 , 1 ) W ( 1 , 2 ) W ( 2 , 1 ) W ( 2 , 2 ) ] , V = [ V ( 1 , 1 ) V ( 1 , 2 ) V ( 2 , 1 ) V ( 2 , 2 ) ] , E = [ E 1 E 2 ] , Open image in new window
where
W ( 1 , 1 ) = [ 2 0 0 0 29 , 630 50 , 000 8 , 889 10 , 000 11 , 110 25 , 000 2 , 120 28 , 610 212 2 , 861 1 , 111 2 , 500 8 , 889 10 , 000 2 , 963 5 , 000 0 0 0 2 ] , W ( 1 , 2 ) = [ 3 0 0 0 8 , 889 10 , 000 13 , 333 10 , 000 6 , 667 10 , 000 1 , 111 10 , 000 1 , 111 10 , 000 6 , 667 10 , 000 13 , 333 10 , 000 8 , 889 10 , 000 0 0 0 3 ] , W ( 2 , 1 ) = [ 3 0 0 0 8 , 889 10 , 000 13 , 333 10 , 000 6 , 667 10 , 000 1 , 111 10 , 000 1 , 111 10 , 000 6 , 667 10 , 000 13 , 333 10 , 000 8 , 889 10 , 000 0 0 0 3 ] , W ( 2 , 2 ) = [ 4 0 0 0 296 250 8 , 889 5 , 000 8 , 889 10 , 000 456 3 , 079 456 3 , 079 8 , 889 10 , 000 8 , 889 5 , 000 2 , 963 2 , 500 0 0 0 4 ] , V 1 , 1 = [ 21 20 1 10 3 20 1 5 2 15 11 60 7 30 17 60 13 60 4 15 19 60 11 30 3 10 7 20 2 5 9 20 ] , V 1 , 2 = [ 0 0 0 0 1 60 1 30 1 20 1 15 1 30 1 15 1 10 2 15 1 20 1 15 3 20 1 5 ] , E 1 = [ 5 2 3 , 563 1 , 233 2 , 339 625 1 , 333 250 ] , V 2 , 1 = [ 1 4 1 4 1 4 1 4 13 60 11 60 3 20 7 60 11 60 7 60 1 20 1 60 3 20 1 20 1 20 3 20 ] , V 2 , 2 = [ 1 20 1 10 3 20 1 5 1 30 1 60 1 15 7 60 7 60 1 15 1 60 1 30 1 5 3 20 1 10 1 20 ] , E 2 = [ 7 , 420 2 , 390 1 , 901 635 1 , 097 348 1 , 291 362 ] . Open image in new window
Now by using the above matrices, the vector solution of the generalized linear system (7) is obtained as follows:
Y = [ a 1 , 0 a 1 , 1 a 1 , 2 a 1 , 3 a 2 , 0 a 2 , 1 a 2 , 2 a 2 , 3 ] = [ 0.9999 1.3380 1.8181 2.7186 0.5001 0.5918 0.6835 1.0001 ] . Open image in new window
Consequently, the approximate functions F 1 , 3 ( t ) Open image in new window and F 2 , 3 ( t ) Open image in new window can be written as follows:
F 1 , 3 = 2 , 353 8 , 445 t 3 + 272 639 t 2 + 2 , 185 2 , 154 t + 11 , 171 11 , 172 , F 2 , 3 = 9 40 t 3 1 345 , 468 , 948 , 216 , 863 t 2 + 2 , 747 9 , 989 t + 1 , 846 3 , 691 . Open image in new window
Figures 1 and 2 show the accuracy of the solution functions F 1 , 3 ( t ) Open image in new window and F 2 , 3 ( t ) Open image in new window, respectively. As shown, the difference between the exact solutionand the computed solution is dispensable. Numerical results can be seen in Table 1 and also Table 2 illustrates the absolute values of the errors obtained here and the absolute errors of [18] for this example.
Figure 1

The comparison between F 1 ( t ) Open image in new windowand F 1 , 3 ( t ) Open image in new windowfor Example 1. Show the accuracy of the solution functions F 1 ( t ) Open image in new window and F 1 , 3 ( t ) Open image in new window, respectively with exact solutions. As shown, the difference between the exact solution and the computed solution is dispensable.

Figure 2

The comparison between F 2 ( t ) Open image in new windowand F 2 , 3 ( t ) Open image in new windowfor Example 1. The comparison between solution of Example 1.

Table 1

Numerical results with error analysis for Example 1

t = 1 2 r Open image in new window

Exact solution

Approximate solution

Error

F 1 ( t ) Open image in new window

F 2 ( t ) Open image in new window

F 1 , 3 ( t ) Open image in new window

F 2 , 3 ( t ) Open image in new window

e 1 , 3 ( t ) Open image in new window

e 2 , 3 ( t ) Open image in new window

r = 1

1.6487

0.6667

1.6484

0.6658

0.0004

0.0009

r = 2

1.2840

0.5714

1.2845

0.5724

0.0004

0.0010

r = 3

1.1331

0.5333

1.1339

0.5350

0.0008

0.0016

r = 4

1.0645

0.5161

1.0650

0.5174

0.0005

0.0012

r = 5

1.0317

0.5079

1.0320

0.5087

0.0003

0.0008

r = 6

1.0157

0.5039

1.0159

0.5044

0.0001

0.0005

r = 7

1.0078

0.5020

1.0079

0.5023

0.0001

0.0003

r = 8

1.0039

0.5010

1.0039

0.5012

0.0000

0.0002

r = 9

1.0020

0.5005

1.0019

0.5007

0.0001

0.0002

Table 2

Absolute errors for Example 1

t = 1 2 r Open image in new window

Presented method

ANN method

TQR

e 1 , 4 ( t ) Open image in new window

e 2 , 4 ( t ) Open image in new window

e 1 , 4 ( t ) Open image in new window

e 2 , 4 ( t ) Open image in new window

e 1 , 4 ( t ) Open image in new window

e 2 , 4 ( t ) Open image in new window

r = 1

3.50 × 10−6

1.17 × 10−5

7.32 × 10−4

3.84 × 10−4

4.11 × 10−4

7.35 × 10−4

r = 2

2.90 × 10−6

1.06 × 10−5

1.15 × 10−4

5.62 × 10−4

3.81 × 10−4

5.14 × 10−4

r = 3

4.78 × 10−5

2.50 × 10−4

5.34 × 10−4

6.11 × 10−4

1.29 × 10−2

8.41 × 10−4

r = 4

5.03 × 10−5

2.50 × 10−4

6.71 × 10−4

6.30 × 10−4

4.17 × 10−3

3.28 × 10−3

r = 5

3.71 × 10−5

1.63 × 10−4

9.80 × 10−4

3.27 × 10−4

2.08 × 10−3

8.20 × 10−4

r = 6

2.53 × 10−5

8.90 × 10−5

1.17 × 10−4

4.83 × 10−4

3.27 × 10−3

3.53 × 10−3

r = 7

1.79 × 10−5

4.30 × 10−5

4.15 × 10−4

4.88 × 10−4

1.08 × 10−2

2.71 × 10−2

r = 8

1.37 × 10−5

1.77 × 10−5

3.12 × 10−4

4.87 × 10−4

6.31 × 10−3

3.84 × 10−4

r = 9

1.16 × 10−5

4.30 × 10−6

1.16 × 10−4

4.82 × 10−4

2.10 × 10−4

3.84 × 10−3

Trapezoidal rule Moreover, this example is going to show the difference between the present method and the trapezoidal quadrature rule.Consider again Example 1, let the region of integration be subdivided into 4 equal intervals of width h = 0.25 Open image in new window, 5 integration nodes t i = 0.25 i Open image in new window for i = 0 , , 4 Open image in new window. Table 2 illustrates the absolute values of the errors obtained here and the absolute errors of [18] for this example. It should be noted that the Lagrange basis functions have been used for finding the 4th degree collocation polynomials through the points ( t i , F i , j ) Open image in new window for i = 0 , , 4 Open image in new window; j = 1 , 2 Open image in new window.

Example 2 Let the system of linear Volterra integral equations
{ ( 3 t 8 ) F 1 ( t ) + ( 2 t + 5 ) F 2 ( t ) = f 1 ( t ) + 0 t ( t + s ) F 1 ( s ) d s + 0 t t s F 2 ( s ) d s , 4 t F 1 ( t ) + ( t 5 ) F 2 ( t ) = f 2 ( t ) + 0 t ( 2 t s ) F 1 ( s ) d s + 0 t ( s + s t ) F 2 ( s ) d s , 0 t 2 , Open image in new window
with
f 1 ( t ) = 5 e t 2 t 9 sin ( t ) t 2 e t + 2 t cos ( t ) t e t + 3 t sin ( t ) , f 2 ( t ) = 3 t 4 e t + sin ( t ) t 2 e t + t cos ( t ) + t e t + 4 t sin ( t ) 1 . Open image in new window
The exact solution of the present problem is, F 1 ( t ) = sin ( t ) Open image in new window and F 2 ( t ) = e t Open image in new window. Similarly, the present method is applied to approximate solution of the integral equations system. We calculate the coefficients matrices W, V and E by using Eq. (10) for n = 4 Open image in new window as following:
W = [ W ( 1 , 1 ) W ( 1 , 2 ) W ( 2 , 1 ) W ( 2 , 2 ) ] , V = [ V ( 1 , 1 ) V ( 1 , 2 ) V ( 2 , 1 ) V ( 2 , 2 ) ] , E = [ E 1 E 2 ] , Open image in new window
where
W ( 1 , 1 ) = [ 8 0 0 0 0 10 , 283 5 , 000 1 , 670 609 835 609 188 617 127 5 , 000 5 16 5 4 15 8 5 4 5 16 137 10 , 000 341 2 , 078 773 1 , 047 1 , 546 1 , 047 1 , 691 1 , 527 0 0 0 0 2 ] , E 1 = [ 5 1 , 286 391 3 , 492 1 , 597 2 , 093 1 , 250 1 , 878 119 ] , W ( 1 , 2 ) = [ 5 0 0 0 0 791 625 27 16 524 621 3 16 39 2 , 500 3 16 3 4 9 8 3 4 3 16 39 5 , 000 469 5 , 000 262 621 524 621 791 1 , 250 0 0 0 0 1 ] , W ( 2 , 1 ) = [ 0 0 0 0 0 791 1 , 250 524 621 262 621 469 5 , 000 39 5 , 000 1 4 1 3 2 1 1 4 117 5 , 000 528 1 , 877 791 625 1 , 577 623 2 , 373 1 , 250 0 0 0 0 8 ] , W ( 2 , 2 ) = [ 5 0 0 0 0 682 479 2 , 373 1 , 250 1 , 009 1 , 063 1 , 099 5 , 211 11 625 1 4 1 3 2 1 1 4 137 10 , 000 341 2 , 078 773 1 , 047 1 , 546 1 , 047 1 , 691 1 , 527 0 0 0 0 3 ] , E 2 = [ 5 4 , 131 607 2 , 501 247 3 , 999 203 15 , 526 353 ] , V 1 , 1 = [ 0 0 0 0 0 737 3 , 433 165 1 , 391 183 5 , 120 43 7 , 680 11 30 , 720 81 160 1 2 27 80 2 15 11 480 932 1 , 273 821 969 891 1 , 024 1701 2 , 560 343 1 , 314 14 15 16 15 6 5 4 3 22 15 ] , V 1 , 2 = [ 0 0 0 0 0 141 4 , 538 132 5 , 843 77 10 , 240 19 15 , 360 1 12 , 288 19 160 7 40 11 80 7 120 1 96 515 2 , 587 343 891 1 , 651 3 , 313 343 803 729 4 , 096 4 15 8 15 4 5 16 15 4 3 ] , V 2 , 1 = [ 0 0 0 0 0 834 3 , 433 781 7 , 680 27 1 , 024 29 7 , 680 7 30 , 720 21 32 19 40 21 80 11 120 7 480 3 , 596 3 , 373 650 703 548 737 243 512 1 , 701 10 , 240 22 15 4 3 6 5 16 15 14 15 ] , V 2 , 2 = [ 0 0 0 0 0 423 4 , 538 347 5 , 120 231 10 , 240 19 5 , 120 1 4 , 096 19 80 7 20 11 40 7 60 1 48 1 , 359 4 , 096 657 1 , 024 1 , 701 2 , 048 729 1 , 024 361 1 , 217 2 5 4 5 6 5 8 5 2 ] . Open image in new window
Now by using the above matrices, the vector solution of the generalized linear system (10) is obtained as follows:
Y = [ a 1 , 0 a 1 , 1 a 1 , 2 a 1 , 3 a 1 , 1 a 2 , 0 a 2 , 1 a 2 , 2 a 2 , 3 a 2 , 4 ] = [ 0 0.4976 1.0113 1.1247 0.9062 1.0000 1.4883 2.3751 3.7268 7.3813 ] . Open image in new window
Consequently, the approximate functions F 1 , 4 ( t ) Open image in new window and F 2 , 4 ( t ) Open image in new window can be written as follows:
F 1 , 4 = 108 3 , 565 t 4 319 1 , 532 t 3 + 142 5 , 855 t 2 + 613 616 t , F 2 , 4 = 197 1 , 779 t 4 + 352 10 , 621 t 3 + 489 818 t 2 + 1 , 793 1 , 836 t + 1 . Open image in new window
Similarly, Figures 3 and 4 show the accuracy of the solution functions F 1 , 4 ( t ) Open image in new window and F 2 , 4 ( t ) Open image in new window, respectively. As shown, the difference between the exact solution and the computed solution is dispensable. Similarly, numerical results can be seen in Table 3.
Figure 3

The comparison between F 1 ( t ) Open image in new windowand F 1 , 4 ( t ) Open image in new windowfor Example 2. Show the accuracy of the solution functions F 1 ( t ) Open image in new window and F 1 , 4 ( t ) Open image in new window, respectively with exact solutions. As shown, the difference between the exact solution and the computed solution is dispensable.

Figure 4

The comparison between F 2 ( t ) Open image in new windowand F 2 , 4 ( t ) Open image in new windowfor Example 2. The comparison between solution of Example 2.

Table 3

Numerical results with error analysis for Example 2

t = 1 2 r Open image in new window

Exact solution

Approximate solution

Error

F 1 ( t ) Open image in new window

F 2 ( t ) Open image in new window

F 1 , 4 ( t ) Open image in new window

F 2 , 4 ( t ) Open image in new window

e 1 , 4 ( t ) Open image in new window

e 2 , 4 ( t ) Open image in new window

r = 1

0.4794

1.6487

0.4795

1.6488

0.0001

0.0001

r = 2

0.2474

1.2840

0.2472

1.2825

0.0002

0.0016

r = 3

0.1247

1.1331

0.1244

1.1315

0.0003

0.0016

r = 4

0.0625

1.0645

0.0622

1.0634

0.0002

0.0011

r = 5

0.0312

1.0317

0.0311

1.0311

0.0001

0.0006

r = 6

0.0156

1.0157

0.0156

1.0154

0.0001

0.0003

r = 7

0.0078

1.0078

0.0078

1.0077

0.0001

0.0002

r = 8

0.0039

1.0039

0.0039

1.0038

0.0001

0.0001

r = 9

0.0020

1.0020

0.0019

1.0019

0.0001

0.0001

Trapezoidal rule Suppose that the region of integration is subdivided into 5 equal intervals of width h = 0.4 Open image in new window, 6 integration nodes t i = 0.4 i Open image in new window for i = 0 , , 5 Open image in new window. Table 4 illustrates the absolute values of the errors obtained here and the absolute errors of [18] for this example. Furthermore, the Lagrange interpolation method has been used to design the interpolation polynomials.
Table 4

Absolute errors for Example 2

t = 1 2 r Open image in new window

Presented method

ANN method

TQR

e 1 , 5 ( t ) Open image in new window

e 2 , 5 ( t ) Open image in new window

e 1 , 5 ( t ) Open image in new window

e 2 , 5 ( t ) Open image in new window

e 1 , 5 ( t ) Open image in new window

e 2 , 5 ( t ) Open image in new window

r = 1

2.66 × 10−5

4.92 × 10−5

1.16 × 10−2

9.28 × 10−4

8.32 × 10−3

8.07 × 10−3

r = 2

5.85 × 10−5

1.37 × 10−5

4.30 × 10−3

1.48 × 10−3

2.14 × 10−2

4.81 × 10−3

r = 3

8.55 × 10−5

2.04 × 10−5

2.04 × 10−3

4.43 × 10−3

3.01 × 10−2

3.71 × 10−2

r = 4

6.14 × 10−5

4.11 × 10−5

1.73 × 10−3

5.91 × 10−3

3.91 × 10−2

3.16 × 10−2

r = 5

3.89 × 10−5

9.44 × 10−5

8.01 × 10−4

7.94 × 10−3

1.12 × 10−1

4.24 × 10−1

r = 6

2.12 × 10−5

5.17 × 10−5

6.23 × 10−4

1.94 × 10−2

3.62 × 10−1

2.80 × 10−1

r = 7

1.11 × 10−5

2.71 × 10−5

5.37 × 10−4

4.13 × 10−2

8.54 × 10−2

1.78 × 10−1

r = 8

5.70 × 10−6

1.38 × 10−5

4.95 × 10−4

6.89 × 10−2

6.37 × 10−2

7.07 × 10−2

r = 9

2.90 × 10−6

7.01 × 10−6

4.75 × 10−4

9.40 × 10−2

8.64 × 10−2

9.04 × 10−3

Example 3 Consider
{ ( 2 t 2 + 3 ) F 1 ( t ) = f 1 ( t ) + 0 t ( t 2 2 s ) F 1 ( s ) d s + 0 t ( s 2 t ) F 2 ( s ) d s ( 2 t 2 + 3 ) F 1 ( t ) = + 0 t 2 s F 3 ( s ) d s , ( 1 3 t 2 ) F 2 ( t ) = f 2 ( t ) + 0 t s ( t + 1 ) F 1 ( s ) d s + 0 t s t ( t 2 + 1 ) F 2 ( s ) d s ( 1 3 t 2 ) F 2 ( t ) = + 0 t ( 2 s 2 + t 3 ) F 3 ( s ) d s , ( 3 t 2 + 6 ) F 3 ( t ) = f 3 ( t ) + 0 t ( s t ) F 1 ( s ) d s + 0 t ( s 2 t 3 ) F 2 ( s ) d s ( 3 t 2 + 6 ) F 3 ( t ) = + 0 t ( 2 t s + s 2 ) F 3 ( s ) d s , Open image in new window
where
f 1 ( t ) = 1 15 ( 6 t 5 + 35 t 4 95 t 3 270 t 2 225 t 360 ) , f 2 ( t ) = 1 30 ( 15 t 7 + 10 t 6 33 t 5 + 275 t 4 + 115 t 3 390 t 2 + 150 ) , f 3 ( t ) = 1 60 ( 40 t 6 66 t 5 175 t 4 + 250 t 3 + 780 t 2 + 360 t + 360 ) , Open image in new window
with the exact solution, F 1 ( t ) = 5 t + 8 Open image in new window, F 2 ( t ) = 2 t 2 5 Open image in new window and F 3 ( t ) = t 2 + t + 1 Open image in new window. Again, we solved this example by this method and the results are given in Table 5. Table 6 illustrates the absolute errors of ANN method and TQR for this example.
Table 5

Numerical results with error analysis for Example 3

t = 1 2 r Open image in new window

Exact solution

Approximate solution

Error

F 1 ( t ) Open image in new window

F 2 ( t ) Open image in new window

F 3 ( t ) Open image in new window

F 1 , 3 ( t ) Open image in new window

F 2 , 3 ( t ) Open image in new window

F 3 , 3 ( t ) Open image in new window

e 1 , 3 ( t ) Open image in new window, i = 1 , , 3 Open image in new window

r = 1

10.500

−4.5000

1.7500

10.500

−4.5000

1.7500

0.0000

r = 2

9.2500

−4.8750

1.3125

9.2500

−4.8750

1.3125

0.0000

r = 3

8.6250

−4.9688

1.1406

8.6250

−4.9688

1.1406

0.0000

r = 4

8.3125

−4.9922

1.0664

8.3125

−4.9922

1.0664

0.0000

r = 5

8.1563

−4.9980

1.0322

8.1563

−4.9980

1.0322

0.0000

r = 6

8.0781

−4.9995

1.0159

8.0781

−4.9995

1.0159

0.0000

r = 7

8.0391

−4.9999

1.0079

8.0391

−4.9999

1.0079

0.0000

r = 8

8.0195

−5.0000

1.0039

8.0195

−5.0000

1.0039

0.0000

r = 9

8.0098

−5.0000

1.0020

8.0098

−5.0000

1.0020

0.0000

Table 6

Absolute errors for Example 3

t = 1 2 r Open image in new window

ANN method

Trapezoidal quadrature rule

e 1 , 3 ( t ) Open image in new window

e 2 , 3 ( t ) Open image in new window

e 3 , 3 ( t ) Open image in new window

e 1 , 3 ( t ) Open image in new window

e 2 , 3 ( t ) Open image in new window

e 3 , 3 ( t ) Open image in new window

r = 1

2.4 × 10−2

7.2 × 10−3

7.8 × 10−4

6.2 × 10−3

5.8 × 10−3

8.1 × 10−3

r = 2

6.1 × 10−3

4.9 × 10−3

5.7 × 10−4

2.0 × 10−2

2.0 × 10−2

2.1 × 10−2

r = 3

1.3 × 10−2

4.4 × 10−3

5.1 × 10−4

3.7 × 10−2

2.1 × 10−2

2.4 × 10−2

r = 4

1.4 × 10−2

4.2 × 10−3

5.1 × 10−4

2.8 × 10−2

1.5 × 10−2

3.2 × 10−2

r = 5

1.5 × 10−2

4.3 × 10−3

5.1 × 10−4

2.5 × 10−3

4.8 × 10−3

5.7 × 10−3

r = 6

1.6 × 10−2

4.2 × 10−3

5.1 × 10−4

2.1 × 10−2

5.3 × 10−2

6.7 × 10−3

r = 7

1.6 × 10−2

4.3 × 10−3

5.0 × 10−4

4.7 × 10−2

6.9 × 10−3

7.3 × 10−3

r = 8

1.6 × 10−2

4.3 × 10−3

5.0 × 10−4

4.3 × 10−2

3.4 × 10−2

2.2 × 10−2

r = 9

1.6 × 10−2

4.3 × 10−3

5.1 × 10−4

2.4 × 10−2

3.3 × 10−4

2.6 × 10−2

As we can see this method will be useful when the exact solution is a polynomial. In other word, the proposed method give the analytical solution for the system, if the exact solution be polynomials of degree n or less than n.

5 Conclusions

In some cases, an analytical solution cannot be found for integral equations system, therefore, numerical methods have been applied. In this study, we have worked out a computational method to approximate solution of the Fredholm and Volterra integral equations systems of the second kind. The present course is a method for approximating unknown functions in terms of truncated sequences including Bernstein polynomials. It is clear that to get the best approximating solutions of the given systems, the truncation degree n must be chosen large enough. An interesting feature of this method is finding the analytical solution for given system, if the exact solution be polynomials of degree n or less than n. Additionally, the proposed method has been compared with ANN method [18] and TQR. The analyzed examples illustrated the ability and reliability of the present method. The obtained solutions, in comparison with exact solutions admit a remarkable accuracy. Extensions to the case of more general systems of integral equations are left for future studies.

Acknowledgements

We would like to thank the referees for their comments and remarks.

Supplementary material

13662_2013_535_MOESM1_ESM.gif (9 kb)
Authors’ original file for figure 1
13662_2013_535_MOESM2_ESM.gif (9 kb)
Authors’ original file for figure 2
13662_2013_535_MOESM3_ESM.gif (9 kb)
Authors’ original file for figure 3
13662_2013_535_MOESM4_ESM.gif (9 kb)
Authors’ original file for figure 4

Copyright information

© Jafarian et al.; licensee Springer 2013

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Authors and Affiliations

  • Ahmad Jafarian
    • 1
  • Safa A Measoomy Nia
    • 1
  • Alireza K Golmankhaneh
    • 2
  • Dumitru Baleanu
    • 3
    • 4
    • 5
  1. 1.Department of Mathematics, Urmia BranchIslamic Azad UniversityUrmiaIran
  2. 2.Department of Physics, Urmia BranchIslamic Azad UniversityUrmiaIran
  3. 3.Department of Mathematics and Computer Sciences, Faculty of Art and SciencesÇankaya UniversityBalgatTurkey
  4. 4.Department of Chemical and Materials Engineering, Faculty of EngineeringKing Abdulaziz UniversityJeddahSaudi Arabia
  5. 5.Institute of Space SciencesMagureleRomania

Personalised recommendations