Some Algorthim For Solving System of Linear Volterra Integral Equation of Second Kind by Using MATLAB 7
Some Algorthim For Solving System of Linear Volterra Integral Equation of Second Kind by Using MATLAB 7
Some Algorthim For Solving System of Linear Volterra Integral Equation of Second Kind by Using MATLAB 7
……
! " # "
+ . " $% & # ' ( )
( 3/1 . " / 0 ) ! " ,
56 7, 8 ! 9 + :" ! " , . " .
7, 8 2009 < (version 7 ) ( MATLAB7 ) ; " 5
/ 0 . " :" + % . "
Abstract
In this paper it is study and modify some Quadrature method to solve system
of Volterra integral equation of order 1 and using some Algorithm to appear
steps of solve using two kind of Quadrature method (Trapezium Simpson
1/3) and comparing between to method.
The leas square error and running time to find the result t by using modify
program (MATLAB) version 7, product 2009. the Simpson's rule, very good
method all less the leas square error running time
347
……………2012 .……
1.1 Introduction:
A numerical quadrature (numerical integration) rules is a primary tool used
by engineers and scientists to obtain approximate answers for definite
integrals that can not be solved analytically [5].
Also, it is a basis of every numerical method for solution of integral
equation [3].
By used quadratre methods to solve second kind integral equations of
convolution type.
We first present the most familiar of numerical integration the quadrature
rules (Trapezoidal rule, Simpson'sl/3 rule. And used to treat system of linear
VIE's.
n
Q[ f ] = ∑ wk f (xk ) = w0 f (x0 ) + w1 f (x1 ) + ... + wn f (xn )
k =0
∫ f (x )dx = Q[ f ] + E[ f ]
a
∫ u(x )dx =
h
[u (x0 ) + u (x1 )] − h u n (z ) (1.1)
x0
2 12
Where z ∈ (x0, x1)
when the length of interval [a, b] is not sufficiently small, the trapezoidal
rule (1.1) is not of much use. For such an interval, we break it into a slim of
integrals over small subinterval, and then apply (1.1) to each of these smaller
integrals, we call it the composite trapezoidal rule.
348
……………2012 .……
Let h = b-a/n, n ≥ 1 and xi = a + i * h(i = 0,1, …, n); then it can be written as:
Using Trapezoidal Rule
Using repeated Trapezoidal rule in equation, at the first stage x = xl, we
obtain the following set of approximating equations:
u11 (x ) = f1 (x ) + K11 (x1 , t 0 )u10 + k11 (x1 , t1 )u11 + ... + k m (x1 , t0 )um 0 + k1m (x1 , t1 )u m1
h h h h
2 2 2 2
u21 (x ) = f 2 + (x1 ) + k 21 (x1 , t0 )u10 + k21 (x1 , t1 )u11 + ... + k 2 m (x1 , t 0 )um 0 + k 2 m (x1 , t1 )u m1
h h h h
2 2 2 2
um (x ) = fm (x1 ) + km1(x1, t 0 )u10 + km1(x1, t1 )u11 + ... + kmm(x11, t 0 )um0 + kmm(x1, t1 )um1
h h h h
2 2 2 2
where uir ≅ ui (xr), for all r = 1, 2, …, n. on simplifying the above we obtain:
h
1 − 2 k11 (x1 , t1 ) − k12 (x1 , t1 ) Λ − k1m (x1 , t1 )
h h
2 2 u11
h
− k21 (x1 , t1 ) 1 − k22 (x1 , t1 ) Λ − k2 m (x1 , t1 ) u21
h h
2 2 2 Μ =
Μ Μ Μ
h u
− 2 km1 (x1 , t1 ) − km 2 (x1 , t1 ) Λ 1 − kmm (x1 , t1 ) m1
h h
2 2
f1 ( x1 ) +
h
k11 ( x1 , t 0 )u10 +
h
k12 (x1 , t0 )u20 + Λ
h
k1m (x1 , t0 )um 0
+
2 2 2
f 2 (x1 ) + k21 (x1 , t0 )u10 + k22 (x1 , t0 )u20 + Λ + k2 m (x1 , t0 )um 0
h h h
2 2 2
Μ Μ Μ Μ
f (x ) + h k (x , t )u + km 2 (x1 , t0 )u20 +
h
Λ + kmm (x1 , t0 )um 0
h
m 1 2 m1 1 0 10 2 2
If follows that the rth stage, uir; I = 1, 2, …, m, r = 1, 2, …, n. are obtained by
solving the system.
h
1 − 2 k11 (xr , t r ) − k12 (xr , t r ) − k1m (xr , t r ) u
h h
Λ
2 2 1r
h
− k 21 (xr , t r ) 1 − k 22 (xr , t r ) Λ − k 2 m ( xr , t r ) u 2 r
h h
2 2 2 Μ
Μ Μ Μ Μ
− h k (x , t ) − k m 2 ( xr , t r )
h
Λ 1 − k mm (xr , t r ) mr
h u
2 m1 r r 2 2
349
……………2012 .……
r −1 r −1
k11 (xr t j )u1 j + k1m (xr , t0 )um 0 + h∑ k1m (xr , t j )urr
f1 ( xr ) +
h
k11 ( xr , t 0 )u10 + h ∑ Λ
h
+
2 2
j =1
r −1
j =1
r −1
f (x ) + h k (x , t )u + h k (x t )u + Λ + k2 m (xr , t0 )um 0 + h∑ k2 m (xr , t j )ur
h
2 r 2 21 r 0 10 ∑j =1
21 r j 1 j
2 j =1
(1.2)
Μ r −1
Μ Μ
r −1
f (x ) + h k (x , t )u + h∑ k (x t )u + Λ + kmm (xr , t0 )um 0 + h∑ kmm (xr , t j )u
h
m r 2 m1 r 0 10 j =1
m1 r j 1 j
2 j =1
Finally, the Gauss-Elimination procedure is used to solve the system (1. 2)
for each value of r, r = 1,2, ... ,n to find uj(xr); I = 1,2, ... ,m.
Now, the Simpson's 1/3 rule is used to approximate each integral term in
equation (1.1), at first stage x = x1 approximated each integral term in
equation (1.1) by the Trapezoidal rule by using equ. (1.1) to get Ujl,
i = 1, 2, ..., m.
If follows that at the rth stage if the number of subintervals are even we apply
the Simpson's 1/3 rule.
Therefore, Uir, 1, 2, ..., m, are obtained by solving the system:
350
……………2012 .……
h
1 − 3 k11 (xr , t r ) − k12 (xr , tt ) − k1m (xr , t r )
h h
Λ
3 3 u1r
h
− k21 (xr , t r ) 1 − k22 (xr , t r ) Λ − k2 m (xr , t r ) u2 r
h h
3 3 3 Μ
(1.3)
Μ Μ Μ Μ
− h k (x , t ) − k m 2 (xr , t r )
h
Λ 1 − kmm (xr , t r ) mr
h u
3 m1 r r 3 3
f1 (xr ) + 3 k11 (xr , t0 )u10 + 3 k11 (xr , t1 )u11 + k11 (xr , tr −1 )u1r −1
h 4h 4h
Λ +
3
+ Λ + k1m (xr , t0 )um 0 + k1m (xr , t1 )um1 + k1m (xr , tr −1 )umr −1
h 4h 4h
Λ +
3 3 3
f (x ) + h k (x , t )u + 4h k (x , t )u + Λ +
4h
k21 (xr , t r −1 )u1r −1
2 r 3 21 r 0 10 3 21 r 1 11 3
=
+ Λ + k2 m (xr , t0 )um 0 + k2 m (xr , t1 )um1 + k2 m (xr , tr −1 )umr −1
h 4h 4h
Λ +
3 3 3
Μ Μ Μ Μ Μ
f (x ) + k (x , t )u + km1 (xr , tr −1 )u1r −1
m r 3 m1 r 0 10 3 km1 (xr , t1 )u11 +
h 4 h 4h
Λ +
3
+ Λ + kmm (xr , t0 )um 0 + kmm (xr , t1 )um1 + k1m (xr , tr −1 )umr −1
h 4h 4h
Λ +
3 3 3
Then, the Gauss-Elimination procedure is used to solve the system (1. 3) for
each value of r = 2, 4,6, ... , n, to find Ui (xr), i = 1,2, ... , m.
Also, if the number of subintervals are odd, we get a combination between
Trapezoidal and Simpson's 1/3 rules.
A the rth stage Uir, i = 1, 2, ..., m, are obtained by solving the system:
h
1 − 3 k11 (xr , t r ) − k12 (xr , tt ) − k1m (xr , t r )
h h
Λ
3 3 u1r
h
− k21 (xr , tr ) 1 − k22 (xr , tr ) Λ − k2 m (xr , tr ) u2 r
h h
3 3 3 Μ
Μ Μ Μ Μ
− h k (x , t ) − km 2 (xr , tr )
h
Λ 1 − kmm (xr , tr ) mr
h u
3 m1 r r 3 3
351
……………2012 .……
f1 (x r ) + 2 k11 (x r , t 0 )u10 + 6 k11 (x r , t 1 )u11 + Λ + 3 k11 (x r , t 2 )u12
h 5h 4h
k11 (x r , t r −1 )u1r −1 + Λ + k1m (x r , t 0 )u m 0 + k1m (x r , t1)u m1
4h h 5h
+ Λ +
3 2 6
+ 3h k (x , t )u + Λ + 4h k (x , t )u
r r −1 mr −1
3 1m r 2 m2
3
1m
f 2 (x r ) + k 21 (x r , t 0 )u10 + k 21 (x r , t1 )u11 + Λ + k 21 (x r , t 2 )u12
h 5h 4h
2 6 3
+ Λ + 4h k 21 (x r , t r −1 )u1r −1 + Λ + h k 2 m (x r , t 0 )u m 0 + 5h k 2 m (x r , t1)u m1
3 2 6
4h
+
3 k 2 m (x r , t 2 )u m 2 + Λ +
4h
k 2 m (x r , t r −1 )u mr −1
3
Μ Μ Μ Μ
f m (x r ) + k m1 (x r , t 0 )u10 + k m1 (x r , t 1 )u11 + Λ +
h 5 h 4 h
k m1 (x r , t 2 )u12
2 6 3
+ Λ + 3 k m1 (x r , t r −1 )u1r −1 + Λ + 2 k mm (x r , t 0 )u m 0 + 6 k mm (x r , t1)u m1
4h h 5h
4h
k mm (x r , t 2 )u m 2 + Λ + k mm (x r , t r −1 )u mr −1
4h
+
3 3
…… (1-4)
Finally, the Gauss-Elimination procedure is used to solve the system (1. 4)
for each value of r, r = 3, 5, 7, 9, ..., n to find ui (xr); I= 1, 2, ..., m.
352
……………2012 .……
353
……………2012 .……
Example 1
Consider the problem:
x x
8x 3
u1 (x ) = x + 1 − x −
2
+ ∫ xu1 (t )dt + ∫ (x + t )u 2 (t )dt
6 0 0
x x
x 2 x3
u 2 (x ) = x − − + tu1 (t )dt + ∫ (x − t )u 2 (t )dt
2 2 ∫0 0
which is a system of two linear VIE's, with u1 (x) = x + 1, u2 (x) = x.
Table (1-1) and (1-2) present a comparison between the exact and quadrature
method for u1(x) and u2(x) respectively depending on least square error and
running time with h = 0.1.
Table (1-1)
Table (1-2)
354
……………2012 .……
Example (2):
Consider the problem:
x
x4
u1 (x ) = e − + xu 2 (t )dt
3 ∫0
x
x
u 2 (x ) = x − xe + e − 1 + ∫ tu1 (t )dt
2 x x
0
Which is a system of two linear VIE's, with u1(x) = ex, u2(x) = x2.
Table (1.3) and (1.4) present a comparison between the exact and numerical
solution of five types of quadrature methods for u1(x) and u2(x) respectively
depending on least square error and running time with h = 0.1.
Table (1-3)
355
……………2012 .……
Table (1-4)
1.4 Discussion:
In this paper , we have introduced quadrature methods for solving a system
of linear VIE's: For each method a computer program was written and an
example was solved.
For easier comparison the solutions were represented in tabular forms. We
conclude the following remarks:
• The number of subintervals n is restricted to be even for Simpson's 1/3
• Through the solution of system of linear VIE's, we see that
Trapezoidal and Simpson's 1/3 gives the best results when it is
compared with other quadrature rules. [See example (1) , (2) ].
References:
[1] ABD-AL-Hammeed, F.T.; "Numerical Solution of Fredholm Integro-
differential Equation Using Spline Function"; M.Sc. thesis, University of
Technology, 2002.
356
……………2012 .……
357