Journal of Computer Science & Computational Mathematics, Volume 4, Issue 3, December 2014

33

A Wavelet Operational Matrix Approach for

Solving a Nonlinear Mixed Type Fractional IntegroDifferential Equation
Habibollah Saeedi1*, Mahmoud Mohseni Moghadam1,2
1Department

of Applied Mathematics, Faculty of Mathematics and Computer, Shahid Bahonar University of Kerman, Kerman, Iran, 7616914111.
2Department of Mathematics, Islamic Azad University of Kerman, Kerman, Iran
*Corresponding author email: saeedi@uk.ac.ir.

Abstract: In this paper, an effective operational method will be used

to determine the numerical solution of a specific Nonlinear
Fractional Volterra-Fredholm integro-differential (NFVFID)
equation. The method is based on CAS wavelets and Block Pulse
Functions (BPFs) and their operational matrices. The main
characteristic of this approach is to reduce a NFVFID equation to a
system of algebraic equations, which greatly simplifies the
problem. Also some numerical examples are provided to illustrate
the accuracy and computational efficiency of the method.
Keywords: Fractional Calculus; CAS Wavelets; Block Pulse
Functions; Operational Matrix.

1. Introduction
The aim of this work is to present a numerical method for
approximating the solution of the following Nonlinear
Fractional Volterra-Fredholm Integro-Differential (NFVFID)
equation:
1

D f ( x) 1 k1 ( x, t ) F1 (t , f (t ))dt
0

2 k 2 ( x, t ) F2 (t , f (t ))dt = g ( x),

(1)

with supplementary conditions:

f (i ) (0) = f i , i = 0,1,, r 1, r 1 < r, r N,

(2)
where

D is the Caputo fractional differentiation operator,

1 ,
2
and
k1 ( x, t ), k2 ( x, t ), F1 (t , f (t )), F2 (t , f (t ))

parameters

given functions in

functions
and g (x) are

L [0,1) and L [0,1) , respectively, and

f (x) is the unknown function. Here, we assume that

q

F1 (t , f (t )) = f 1 (t ) and F2 (t , f (t )) = f
q1 and q2 are positive integers.

q2

(t ) , where

In recent years, it has turned out that many phenomena in

signal
processing,
control
engineering
[1,
2],
electromagnetism [3], biosciences [4], fluid mechanics [5],
electrochemistry [6], diffusion processes [7], dynamic of
viscoelastic materials [8], continuum and statistical
mechanics [9] and propagation of spherical flames [10] can

be successfully modelled by the use of fractional derivatives

and integrals. It is well known that the fractional order
differential and integral operators are non-local operators.
This is one reason why fractional differential operators
provide an excellent instrument for description of memory
and hereditary properties of various physical processes.
Motivated by increasing number of applications of fractional
differential equations, considerable attention has been given
to provide efficient methods for exact and numerical
solutions of fractional differential equations.
In general, most of the fractional integro-differential
equations do not have exact solutions. Particularly, there is
no known method for solving NFVFID equations exactly.
Therefore the numerical solution of these problems is very
important. However, there are a few methods for solving
these equations, numerically, and most of these methods
have been applied to the linear and non-fractional problems,
see [11,12,13,14,15,16,17].
In this paper, we will introduce a new operational method,
based on CAS wavelets and Block Pulse Functions (BPFs),
to solve NFVFID equations. The method is based on
reducing Eq. (1) to a system of algebraic equations by
expanding the solution via CAS wavelets with unknown
coefficients. The main characteristic of an operational
method is to convert a differential equation into an algebraic
one. This method does not only simplify the problem but
also speeds up the computation. The sparse structure of the
operational matrices will simplify the problem and reduce
the solution procedure. It is considerable that, our method
can be easily applied for solving Eq. (1), when N .
The plan of this paper is as follows. In section 2, fractional
integral and differentiation operators, CAS wavelets, Block
Pulse Functions and some of their properties are introduced.
Section 3 introduces function approximations via CAS
wavelets and BPFs in the matrix forms. In Section 4, the
CAS wavelet operational matrix of fractional integration is
derived. In Section 5, the approximation of CAS wavelets
and BPFs and their operational matrices are used to reduce
Eq. (1) to a system of algebraic equations. The error analysis
of the method is discussed in section 6. In section 7 some
numerical examples are presented to show the convergence,
accuracy and advantages of the proposed method with a
comparison to other methods. There is a conclusion of the

34

Journal of Computer Science & Computational Mathematics, Volume 4, Issue 3, December 2014

whole work in section 8. In appendices A and B, some

details of calculations related to sections 4 and 5, are given.

n
n 1
k/2
2 CAS m (n 2 k x), if k x < k ;
n , m ( x) =
2
2

otherwise.
0

2. Preliminaries
In this section, we present some notations, definitions, and
preliminaries that will be used in the rest of the paper.
2.1 Fractional Calculus
There are several definitions of a fractional derivative of
order > 0 . The two most commonly used definitions are
the Riemann- Liouville and Caputo fractional derivative.
Each definition uses Riemann- Liouville fractional
integration and derivatives of whole order. The RiemannLiouville fractional integration of order is defined as:

I f ( x) =

1 x
( x t ) 1 f ( x)dt , x > 0, I 0 f ( x) = f ( x),

0
( )

and the Caputo fractional derivative of order

as

Definition 2.1 An m- set of Block Pulse Functions (BPFs)

over the interval [0,T ) is defined as [19]:

1, if
bi ( x) =
0

b ( x), i = j;
bi ( x)b j ( x) = i
0, i j.

D* f ( x) = I m D m f ( x) , where D m is the ordinary

1/m, i = j;
i j.

b ( x)b ( x)dx = 0,
0

xk
I D* f ( x) = f ( x) f ( k ) (0 ) , x > 0.
k!
k =0

(5)

Completene ss :
For every f L2 ([0,1)) , the sequence {bi } is complete if

b f = 0 results in
i

f = 0 , almost everywhere.

Because of the completeness of {bi ( x)} , Parsevals

identity holds, i.e. we have

m 1

(4)

Orthogonality :

D* I f ( x) = f ( x) , and:

i = 0,1,2,, m 1,

with a positive integer value for m . In this paper, it is

assumed that T = 1 , so BPFs are defined over [0,1) .
Now we explain some applicable properties of BPFs:
Disjointne ss :

is defined

integer differential operator of order m and I

is the
Riemann- Liouville integral operator of order m with
m 1 < m .
The relationship between the Riemann- Liouville operator
and the Caputo operator is given by the following lemma
[18]:
Lemma 2.1 If m 1 < m, m N , then:

iT
(i 1)T
x<
;
m
m
otherwise,

f
0

( x)dx = i =0 f i 2Pbi ( x)P2 ,

for every real bounded function f ( x) L2 ([0,1)) , and:

1

2.2 CAS Wavelets and Block Pulse Functions(BPFs)

Wavelets are special kinds of oscillatory functions with
compact support that are constructed by using dilation and
translation of a single function, called the mother wavelet,
denoted by (x) and must satisfy in certain requirements.
If the dilation parameter is a and translation parameter is b
, then we have the following family of wavelets:

a ,b ( x) =| a |

1/2

x b
(
), a, b R, a 0.
a

(3)

f i = m bi ( x) f ( x)dx.
BPFs have compact supports, i.e.:
i i 1
Supp(bi ( x)) = [ ,
].
m m

3. Function Approximation
The set of CAS wavelets forms an orthonormal basis for
L2 ([0, 1)) . This implies that any function f (x) defined over
[0,1) , can be expanded as:

The CAS wavelets employed in this paper, are defined as:

f ( x) = cn,m n,m ( x)

n
n 1
k/2
2 CAS m (2k x n), if k x < k ;
n ,m ( x ) =
2
2
0
otherwise,
where:

(6)

n = 0 mZ

2k 1

c
n =0 m= M

n ,m ( x )

n ,m

= cT ( x),

CAS m ( x) = cos (2mx) sin(2mx),

and n = 0,1,...,2 1, k N {0}, m Z.

Let us introduce the following useful
corresponding to CAS wavelets:
k

where cn,m =< f ( x), n,m ( x) >=

f ( x)
0

n,m

< f , g > is the inner product of the functions

notation,

and are 2 (2M 1) 1 -vectors given by:

k

( x)dx, and

f and g , c

35

Journal of Computer Science & Computational Mathematics, Volume 4, Issue 3, December 2014

c = [c0, M , c0, M 1 , , c0,M , c1, M , , c1,M , , c

2k 1, M

( x) = [ 0, M , 0, M 1 ,, 0,M , 1, M ,, 1,M ,,

, , c

2k 1, M

2k 1,M

,,

1 1 2
0 1
1

0 0 1
1
1

F =

m ( 2)
0 0

0 0 0

]T ,

2k 1,M

]T .

Also from the orthogonality property of BPFs, it is

possible to expand functions into their Block- Pulse series
[19]. This means that for every f ( x) L2 ([0,1)) , we have:

3 m1
2 m 2
1 m3

,
1

1
0

m 1

f ( x) f i bi ( x) = f T B m ( x),

(7)

i =0

Remark 4.1 For = 1 , F is the BPFs operational

matrix of integration.

where:

f = [ f 0 , f1 ,..., f m1 ]T ,

Let:

Bm ( x) = [b0 ( x), b1 ( x),..., bm1 ( x)]T ,

such that

( I m )( x) Pmm m ( x),

f i s, i = 0,1,..., m 1 , are obtained by Eq. (6).

(8)

and:
(8)

where a is an m -vector and ~

a = diag (a) .
Moreover, we can clearly conclud that for every m m
matrix A :
~
(9)
BTm ( x) AB m ( x) = AT B m ( x),

where A is an m -vector with elements equal to the

diagonal entries of matrix A .
Notation . From now on, we define m = 2k (2M 1) ,
where k , M N {0} .

4. Operational
Integration

Matrix

of

Fractional

There is a relationship between BPFs and CAS wavelets.

Providing some calculations outlined in appendix A, we have
found:
(10)
( x) = mmB m ( x).
The Block Pulse operational matrix of fractional
integration F is given as follows [20]:
(11)
( I B m )( x) F B m ( x),
where:

where matrix Pmm is called the CAS wavelet operational

matrix of fractional integration. Using Eqs. (11) and (12), we
have:
( I m )( x) ( I mmBm )( x) = mm ( I Bm )( x) mmF Bm ( x).

(13)
By Eqs. (13) and (14), we get:
Pmm m ( x) mmF B m ( x).
Therefore, the CAS wavelet operational matrix of fractional
integration Pmm is given by:

B Tm ( x)B m ( x) = 1,

B m ( x)B Tm ( x)a = ~
a B m ( x),

(12)

From the above representation and disjointness property, it

follows that:
0

0
b0 ( x)
0
b
(
x
)

T
B m ( x )B m ( x ) =

0 ,

0
b
m 1 ( x)

and k = (k 1) 1 2k 1 (k 1) 1.

Pmm = mmF m1m .

5. Analysis of the Method

Consider

Eq.

(1),
two
variables
functions
,
,
can
be
approximated
as:
kr ( x, t ) L ([0,1)) r = 1,2

k r ( x, t )

2 k 1 M

2 k 1

kr
n = 0 l1 = M m = 0 l2 = M

i, j

n ,l1

( x) m,l (t ),
2

for i = n(2M 1) l1 M 1, j = m(2M 1) l2 M 1,

or in the matrix form:
(14)
kr ( x, t ) T ( x)K r (t ),
where K r = [kri , j ] and kri , j =< n,l ( x), < kr ( x, t ), m,l (t ) >> ,
1

for r = 1,2 . The right hand side of Eq. (1) can also be written
as:
(15)
g ( x) gT ( x).
Remark 5.1 CAS wavelets cannot be used directly for
solving
differential
equations
because of
their
discontinuities. There are two possible ways to overcome
this problem. The first possibility is converting the
underlying differential equation into an equivalent integral
equation and approximating the solution by truncated
orthogonal CAS series and using operational matrices of
integration to eliminate the integral operators. The second
possibility is to expand the highest derivative term,
appearing in the differential equation, into the CAS wavelet
series.

36

Journal of Computer Science & Computational Mathematics, Volume 4, Issue 3, December 2014

So, let:

D* f ( x) c ( x).
T

(16)

By using lemma 2.1, Eqs. (13) and (17), we have:

m 1

f ( x) cT Pmm ( x) f ( k ) (0 )
k =0

xk
.
k!

(17)

Hence, substituting the supplementary conditions (2) in

(17) and approximating it via CAS walelets, we get:
T
(18)
f ( x) (cT Pmm c1 )( x),
where c1 is an

m -vector. According to Eq. (11):

f ( x) (cT Pmm c1 )mmB m ( x).

T

(19)

T
a = [a0 , a1 ,..., am1 ] = (cT Pmm c1 )mm . (20)
By some calculations which are mentioned in details in
appendix B, we will have:

1
NBm ( x),
m

(22)

~
q
( x, t )[ f (t )] 2 dt A T B m ( x),

(23)

q1

dt

1
NB m ( x)
m
~
2 A T B m ( x) g T mm B m ( x) |

Rm ( x) =| c T mm B m ( x) 1

0.
r
Set x = xi , our aim is to have Rm ( xi ) 10 i , where

ri is

Define:

k ( x, t )[ f (t )]

We can easily check the accuracy of the method. Since the

truncated CAS wavelet series is an approximate solution of
Eq. (1), the resulting equation, (24), must be satisfied
approximately in the main equation, that is for x [0,1) :

any positive integer. If we prescribe Max{ri } = 10 , then we

increase m as long as the following inequality holds at each
point xr :

Rm ( xi ) 10r ,
in other words, by increasing m the error function Rm ( xi )
approaches to zero.
(21)

and:
x

k
0

(22)

where:

N = (Tmm K 1 mm ~
a q1 ) T ,

~
a q = [a0q , a1q ,..., a mq 1 ],
~
and A is an m -vector whose elements are equal to the
diagonal entries of the following matrix:

A = Tmm K 2 mm diag (~
a q 2 )F 1 .
Now by substituting approximations (16), (17), (22) and
(23) into Eq. (1), we obtain:
1
~
NB m ( x) 2 A T B m ( x) g T mm B m ( x).
m
(23)
According to orthogonality of BPFs, we have the
following nonlinear system of algebraic equations:
c T mm B m ( x) 1

c T mm 1

1
~
N 2 A T g T mm .
m

(24)

By solving this system, with respect to the unknown

T

vector c , the approximate solution of Eq. (1) will be

obtained according to Eq. (17).

6. Error Analysis
Theorem 6.1 [21] A function f ( x) L2 [0,1] , with bounded
second derivative, say f '' ( x) N , can be expanded as an
infinite sum of CAS wavelets, and the corresponding series
converges uniformly to f (x) , that is:

f ( x) = cn,m n,m ( x)
n = 0 mZ

Figure 1. The comparison between approximate and exact

solutions of Exmple 7.1 for some k and M , with = 1 .

7. Numerical Examples
To illustrate the effectiveness of the proposed method,
several test examples are carried out in this section. Note
that:

em ( x)

= ( em2 ( x)dx)1/2 (
0

1
N

2
m

( xi ))1/2 ,

i =0

where em ( xi ) = f ( xi ) f m ( xi ) , i = 0,1,..., N , f (x) is the

exact solution and f m (x) is the approximate solution which
is obtained by Eq. (17).
The computations associated with the examples were
performed using Matlab on a Personal Computer.

37

Journal of Computer Science & Computational Mathematics, Volume 4, Issue 3, December 2014

Table 1. The comparison between the approximate solutions

and the exact solution (Example 7.3)

Figure 2. The comparison between approximate solutions of

Exmple 7.1 for some 0.5 1 , with k = 5 and M = 1 .
Example 7.1 Error! Reference source not found.
Consider a NFVFID equation, as follows:
1 1
1 x
D* f ( x) tf 3 (t )dt xf 2 (t )dt = g ( x), (25)
0
4
2 0
6
where g ( x) = x 2 x 1 , with the initial condition
10
10
f (0) = 0 , and the exact solution f ( x) = x 2 , in the case = 1
. A comparison between the CAS wavelets solutions with
some k and M , beside the approximate solutions and the
exact solutions, are shown in Figure 1. Also, Figure 2 shows
the numerical results for k = 4, M = 1 and various 0 < 1 .
The comparisons show that as 1, the approximate
solutions tend to f ( x) = x 2 , which is the exact solution of the
equation in the case = 1 . Table 2 presents the 2-norm of
the absolute error in the case of = 1 .
Example 7.2 [22-23] For the following nonlinear integrodifferential equation:
x

D f ( x) 3cos ( x t ) f 2 (t )dt = sin(2 x),

(26)

with the initial condition f (0) = 1, and in the case = 1 the

exact solution is f ( x) = cos ( x) . Table 2 presents the
comparison of the numerical solutions with the presented
solution and figure 3 gives a comparison between the
approximate solutions and the exact solution with some k
and M .

Exact
solution

0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0

1.000000
0.995004
0.980067
0.955336
0.921061
0.877583
0.825336
0.764842
0.696707
0.621610
0.540302

BPFs
Presented Presented
Adomian's
method
method
method
method [23]
[22]
(k=4,M=1) (k=5,M=1)
(m=16)
(m=16)
0.999210 0.999803 1.000000 1.000000
0.994216 0.995507 0.995141 0.994951
0.982054 0.981213 0.975784 0.980303
0.952894 0.953794 0.960386 0.955685
0.925488 0.919364 0.918443 0.921165
0.861828 0.869898 0.862193 0.877048
0.819314 0.827871 0.828963 0.822596
0.770616 0.768348 0.752929 0.755333
0.690310 0.692816 0.710418 0.667739
0.630059 0.618108 0.617232 0.547241
0.549002 0.542915 0.566917 0.364798

Figure 4. The comparison between approximate and exact

solutions of Exmple 7.3 for some k and M .
Example 7.3 Consider the following nonlinear integrodifferential equation:
1

D 2 f ( x) xt[ f (t )]4 dt [ f (t )]2 = g ( x),

(27)

1 8 3
x
x 3 (6 x 2 15 x 10) ,
( x 2 x)

1
1260
30
( ) 3
2
with the initial condition f (0) = 0 , and the exact solution

where:

g ( x) =

f ( x) = x 2 x . Table 1 and figure 4 show the comparison

between the approximate solutions and the exact solution.

Table 2. The 2-norm of the absolute error.

Examples E 2 , ( k 3, M 1)

Figure 3. The comparison between approximate and exact

solutions of Exmple 7.2 for some k and M , with = 1 .

E 2 , ( k 4, M 1)

E 2 , ( k 5, M 1)

Example
7.1
( = 1)

1.693401e-005

2.689175e-006

9.968712e-008

Example
7.2
( = 1)

1.001033e-005

2.498805e-006

6.243634e-007

1.814125e-004

9.614030e-005

4.942423e-005

1.184921e-004

6.636244e-005

3.507248e-005

Example
7.3
Example
7.4

38

Journal of Computer Science & Computational Mathematics, Volume 4, Issue 3, December 2014

Figure 5. The comparison between approximate solutions of

Exmple 7.3 for some 0.25 0.75 .

some examples. In these examples the approximate solutions

are briefly compared with the exact and approximate
solutions obtained by the other methods.
Increasing the number of CAS wavelets over [0,1) ,
decreases the error of the solution rapidly. To show its
convergence and stability, the current method can be run
with increasing m , until the computed results have
appropriate accuracy. The advantage of this method is the
low cost of setting up the equations without using any
projection or collocation method and integration. The most
interesting ability of this method is that, it can be simply
applied to the cases where is integer which is observed in
the numerical examples.
Appendix A: Expanding CAS Wavelets Via BPFs

Example 7.4 For the following nonlinear integrodifferential equation:

1

f (4) ( x) ( x t ) 2 f 3 (t )dt x 2tf 2 (t )dt = g ( x),

(28)

where:
g ( x) =

n
n 1
k/2
2 CAS m (n 2 k x), if k x < k ;
2
2

otherwise.
0,

~ n,m ( x) =

x10 3x 9 5 x 8 x 7 15 x 6 x 5 x 4
x
1

, Eq. (7) implies that CAS wavelets can be also expanded into
8
7
8
2
64
16 128 5632 11264
an m -term BPF as:

1
3
f (0) = , f (0) = ,
8
4
1
and
the
exact
solution
, f (0) = 3, f (0) = 6
f ( x) = ( x ) 3 ,
2

with

First, let us introduce the following useful notation,

corresponding to CAS wavelets:

the

initial

conditions

the comparison between the CAS wavelets method, with

different values of k and M , beside the exact solutions, are
shown in table 3 and figure 4.

m1

nm ( x) f i bi ( x).
i =0

By using the properties of CAS wavelets and Eq. (6), we

get:

f i = m

( i 1)/m

i/m

nm ( x)dx

m
i
i 1
{~ nm ( ) ~ nm (
)},
m
m
2 k 1 m

for:
i = n(2M 1),, n(2M 1) (M 1), n(2M 1) (M 1),, (n 1)(2M 1) 1.

Note that for i = n(2M 1) M , we have f i = 2k/2 and

otherwise f i = 0 . Therefor, we get:
nm ( x)

m
i
i 1
i 2M ~ i 2M 1
[0,0,
,0
,~ nm ( ) ~ nm ( ), ,~ nm (
) nm (
),

m
m
m
m
2 k 1 m n(2
M 1)
0,,0]B m ( x),

Figure 6. The comparison between approximate and exact

solutions of Exmple 7.4 for some k and M .

8. Conclusion
A numerical scheme, based on operational matrices of
integration for CAS wavelets and BPFs, transforms a
fractional nonlinear Volterra-Fredholm integro-differential
equation to a set of algebraic equations without applying any
projection method. Solving this system by an iterative
method gives an approximate solution which is a linear
combination of m = 2k (2M 1) CAS wavelets. The
applicability and accuracy of the method are checked by

where i = n(2M 1) , n = 0,1,,2k 1 and m = M ,, M .

Hence:
(29)
( x) = mmBm ( x),
such that mm = Diag (0 , 1 ,,

2k 1

), and n for

n = 0,1,,2k 1 is a (2M 1) (2M 1) matrix which is

introduced as:

~ i
~ i 1
~ i 2M ~ i 1 ~ i 2
~ i 2M 1
) n (
) n (
) n (
) n (
) ,
n = n ( ) n (

m
m
m
m
m
m

here:

39

Journal of Computer Science & Computational Mathematics, Volume 4, Issue 3, December 2014

M
1
m M
= k 1
2
1

1
1

1
M
1
1

1
M ,

1
1

1
M

23k/21
m
23k/21
m

23k/21
m

1
1

M 1
1
1
1

M 1
1

1
1

M 1
1

k ( x, t )[ f (t )]
0

q1

dt

(34)

or:
1

k ( x, t )[ f (t )]
0

q1

1 T
B m ( x)Tmm K 1 mm ~
a q1
m

dt
=

1
(Tmm K 1 mm ~
a q1 )T B m ( x)
m

1
NB m ( x).
m

~
~
~
~
n ( x) = n, M ( x) n, M 1 ( x) n,M ( x) .

1 T
( x)K 1 mm ~
a T q1 ,
m

(35)

Also the second integral part of Eq. (1), by the following

process, can be written via CAS wavelets. Using Eqs. (11),
(15) and (31), we will have:

Appendix B: Expanding the Integral Part of the Main

Equation Via CAS Wavelets
According to Eqs. (19) and (20), we have
f ( x) a B m ( x). From the disjoint property of the BPFs, we
get:
[ f ( x)]2 [a B m ( x)]2

k
0

x
q
( x, t )[ f (t )] 2 dt = T ( x)K 2 (t )B Tm (t ) ~
a T q2 dt
0

x
= T ( x)K 2 mm B m (t )B Tm (t ) ~
a T q2 dt
0

= [a0b0 ( x) a1b1 ( x) am1bm1 ( x)]2

= a02b0 ( x) a12b1 ( x) am2 1bm1 ( x)

=~
a2 B m ( x),

x
= T ( x)K 2 mm B m (t )B Tm (t ) ~
a T q2 dt.

Using Eq. (8), we simplify the integral part of (36) as:

where ~
a2 = [a02 , a12 ,, am2 1 ] .
It is easy to show by induction that:
x

[ f ( x)] [a , a ,, a
q

q
0

q
1

(33)

q
m1

]B m ( x) = ~
a q B m ( x),

(30)

B
0

x
(t )B Tm (t ) ~
a T q2 dt = diag (~
a q2 )B m (t )dt
0

where:

~
a q = [a0q , a1q ,, amq 1 ],
Using Eqs. (11), (15) and (31), we will have:
1

k ( x, t )[ f (t )]
0

q1

(31)

x
= diag (~
a q2 ) B m (t )dt
0

= diag (~
a q2 )F1B m ( x).

dt = T ( x)K 1 (t )B Tm (t ) ~
a T q1 dt

Thus in (36), we get:

1
= T ( x)K 1 mm B m (t )B Tm (t ) ~
a T q1 dt

1
= T ( x)K 1 mm B m (t )B Tm (t ) ~
a T q1 dt.

q
( x, t )[ f (t )] 2 dt T ( x)K 2 mm diag (~
a q2 )F1B m ( x)

(33)

Applying properties (4) and (5), we simplify the integral

part of (33) as:

B
0

(t )B Tm (t ) ~
aT

q1

b0 (t )

1
dt = {
0

a 01
q1
b1 (t )
a1
}dt

bm1 (t ) a mq11

= [a b (t ), a b (t ), , a
0

q1
0 0

1 q1 q1
q
[a0 , a1 ,, am11 ]T

1 ~T
a q1 .
m
Thus in (33), we have:
=

q1
1 1

(34)

q1
m1 m1

~
= BTm ( x)Tmm K 2 mm diag (~
a q2 )F1B m ( x) = AT B m ( x),

(t )] dt

where A is an m -vector with elements equal to the

diagonal entries of the following matrix:
A = Tmm K 2 mm diag (~
a q 2 )F 1 .

40

Journal of Computer Science & Computational Mathematics, Volume 4, Issue 3, December 2014

Reference
[1] O. P. Agrawal, A general formulation and solution
scheme for fractional optimal control problems,
Nonlinear Dyn., vol. 38 , pp. 323337, 2004.
[2] F. B. M. Duarte, J. A. Tenreiro Machado, Chaotic
phenomena and fractional-order dynamics in the
trajectory control of redundant manipulators,
Nonlinear Dynamic,. vol. 29, pp. 342362, 2002.
[3] N. Engheta, On fractional calculus and fractional
multipoles in electromagnetism, IEEE Transactions on
Antennas and Propagation, vol. 44, no. 4, pp. 554566,
1996.
[4] R. L. Magin, Fractional calculus models of complex
dynamics in biological tissues, Comput. Math. Appl.,
vol. 59, pp. 15861593, 2010.
[5] V. V. Kulish, J. L. Lage, Application of fractional
calculus to fluid mechanics, J. Fluids Eng, vol. 124,
no. 3, pp. 803-806, 2002.
[6] K. B. Oldham, Fractional differential equations in
electrochemistry, Advances in Engineering Software,
vol. 41, no.1, pp. 912, 2010.
[7] V. Gafiychuk, B. Datsko, V. Meleshko, Mathematical
modeling of time fractional reaction diffusion systems,
Computational and Applied Mathematics, vol. 220, no.
1, pp. 215225, 2008.
[8] C. Lederman, J-M Roquejoffre, N. Wolanski,
Mathematical
justification
of
a
nonlinear
integrodifferential equation for the propagation of
spherical flames, Ann. di Mate., vol. 183, pp. 173239,
2004.
[9] F. Mainardi, Fractional calculus: some basic problems
in continuum and statistical mechanics, in: A.
Carpinteri, F. Mainardi (Eds.), Fractals and Fractional
Calculus in Continuum Mechanics, Springer-Verlag,
New York, pp. 291348, 1997.
[10] F. C. Meral, T. J. Royston, R. Magin, Fractional
calculus in viscoelasticity: an experimental study,
Communications in Nonlinear Science and Numerical
Simulation, vol. 15, no. 4, pp. 939945, 2010.
[11] M. A. Abdou, Integral equation of mixed type and
integrals of orthogonal polynomials, J. Comput. Appl.
Math., vol. 138, no. 2, pp. 273285, 2002.
[12] M. A. Abdou, On asymptotic methods for Fredholm
Volterra integral equation of the second kind in contact
problems, J. Comput. Appl. Math., vol. 154, no. 2, pp.
431446, 2003.

[13] E. Babolian, Z. Masouri, S. Hatamzadeh-Varmazyar,

Numerical solution of nonlinear VolterraFredholm
integro-differential equations via direct method using
triangular functions, Comput. Math. Appl., vol. 58, no.
2, pp. 239247, 2009.
[14] C. Cattani, Shannon wavelets for the solution of
integrodifferential equations, mathematical problems
in engineering, pp. 122, 2010.
[15] C. H. Hsiao, Hybrid function method for solving
Fredholm and Volterra integral equations of the second
kind, Journal of Computational and Applied
Mathematics, vol. 230, no. 1, pp. 5968, 2009.
[16] K. Maleknejad, M.Tavassoli Kajani, Solving linear
integro-differential equation system by Galerkin
methods with hybrid functions, Appl. Math. Comput.,
vol. 159, no. 3, pp. 603612, 2004.
[17] A. M. Wazwaz, The combined Laplace transformAdomian decomposition method for handling nonlinear
Volterra integro-differential equations, Applied
Mathematics and Computation, vol. 216, no. 4, pp.
13041309, 2010.
[18] I. Podlubny, Fractional Differential Equations,
Academic Press, New York, 1999.
[19] C. R. Rao, Piecewise orthogonal functions and their
applications on system and control, Springer, Berlin,
1983.
[20] A. Kilicman, Z. Al Zhour Kronecker operational
matrices for fractional calculus and some applications.
Applied Mathematics and Computation, vol 187, no. 1,
pp. 250265, 2007.
[21] H. Adibi, P. Assari, Using CAS wavelets for numerical
solution of Volterra integral equations of the second
kind, Dynamics of Continuous, Discrete and Impulsive
Systems Series A: Mathematical Analysis, vol. 16, no.
5, pp. 673685, 2009.
[22] E. Babolian, Z. Masouri, S. Hatamzadeh-Varmazyar,
New direct method to solve nonlinear VolterraFredholm integral and integro differential equation
using operational matrix with Block-pulse functions,
Progress In Electromagnetics Research B, pp. 59-76,
2008.
[23] K. Maleknejad, M. Hadizadeh, The numerical analysis
of Adomians decomposition method for nonlinear
Volterra integral and integro-differential equations,
International Journal of Engineering Science, Iran
University of Science and Technology, vol. 8, no. 2a,
pp. 3348, 1997.