Anda di halaman 1dari 7

Commun. Theor. Phys. 56 (2011) 10091015 Vol. 56, No.

6, December 15, 2011


Application of the Generalized Dierential Quadrature Method in Solving Burgers
Equations
R. Mokhtari,
1,
A. Samadi Toodar,
2
and N.G. Chegini
2
1
Department of Mathematical Sciences, Isfahan University of Technology, Isfahan 84156-83111, Iran
2
Department of Mathematics, Tafresh University, Tafresh 39518-79611, Iran
(Received May 26, 2011; revised manuscript received August 22, 2011)
Abstract The aim of this paper is to obtain numerical solutions of the one-dimensional, two-dimensional and coupled
Burgers equations through the generalized dierential quadrature method (GDQM). The polynomial-based dierential
quadrature (PDQ) method is employed and the obtained system of ordinary dierential equations is solved via the total
variation diminishing RungeKutta (TVD-RK) method. The numerical solutions are satisfactorily coincident with the
exact solutions. The method can compete against the methods applied in the literature.
PACS numbers: 02.60.Cb, 02.30.Jr
Key words: generalized dierential quadrature method (GDQM), total variation diminishing RungeKutta
(TVD-RK) method, Burgers equations
1 Introduction
Exact and numerical solutions of the nonlinear par-
tial dierential equations and systems of nonlinear partial
dierential equations play an important role in the phys-
ical sciences as well as in the engineering elds, see e.g.
Refs. [15] and their bibliographies.
Burgers equation is one of the very few nonlinear par-
tial dierential equations that can be solved exactly using
the HopfCole transformation. Nevertheless, numerical
methods for solving such equations have practical signi-
cance and drawn the attention of many scientists.
Burgers equation was formulated by Bateman in
1915,
[6]
and later treated by Burgers.
[7]
This equation
is also called the nonlinear advection-diusion equation,
and can be regarded as a qualitative approximation of
the NavierStocks equations. It retains the nonlinear as-
pects of the governing equation in many practical trans-
port problems such as aggregation interface growth, the
formation of large-scale structures in the adhesion model
for cosmology, turbulence transport, shock wave theory,
wave processes in thermo-elastic medium, transport and
dispersion of pollutants in rivers and sediment transport.
One-dimensional Burgers equation is as follows
[8]
u
t
+u
u
x

2
u
x
2
= 0 , 0 < x < 1 , t > 0 , (1)
with the initial condition
u(x, 0) = f(x) , 0 x 1 ,
where = 1/Re in which Re is the Reynolds number.
Some exact/numerical solutions of the one-dimensional
Burgers equation are obtained by an explicit Backlund
transformation,
[9]
tanh-coth method,
[10]
dierential trans-
formation method,
[11]
variational iteration method,
[12]
homotopy analysis method,
[13]
modied local Crank
Nicolson method,
[14]
element-free characteristic Galerkin
method,
[15]
least-squares quadratic B-spline nite ele-
ment method,
[16]
a sixth-order compact nite dierence
scheme,
[17]
cubic B-spline quasi-interpolation method,
[18]
nite element method,
[19]
reproducing kernel method
[20]
and so on.
Two-dimensional Burgers equation is as follows
[14]
u
t
+u
u
x
+u
u
y
=

2
u
x
2
+

2
u
y
2
,
0 < x, y < 1 , t > 0 , (2)
with the initial condition
u(x, y, 0) = f(x, y) , 0 x, y 1 .
Some exact/numerical solutions of two-dimensional Burg-
ers equation reported in the literature are: mod-
ied local CrankNicolson method,
[14]
Adomian de-
composition method,
[21]
generalized (G

/G)-expansion
method,
[22]
lattice Boltzmann method,
[23]
Eulerian
Lagrangian method,
[24]
articial boundary method
[25]
and
etc.
The coupled Burgers equation which we consider is as
follows
[2627]
u
t


2
u
x
2
+u
u
x
+
(uv)
x
= 0 ,
0 < x < 1 , t > 0 ,
v
t


2
v
x
2
+v
v
x
+
(uv)
x
= 0 ,
0 < x < 1 , t > 0 , (3)

Corresponding author, E-mail: mokhtari@cc.iut.ac.ir


c 2011 Chinese Physical Society and IOP Publishing Ltd
http://www.iop.org/EJ/journal/ctp http://ctp.itp.ac.cn
1010 Communications in Theoretical Physics Vol. 56
with the initial conditions
u(x, 0) = f(x) , v(x, 0) = g(x) , 0 x 1 ,
where is a real constant, and are arbitrary constants
depending on the system parameters such as Peclet num-
ber, Stokes velocity of particles due to gravity and Brown-
ian diusivity. Some exact/numerical solutions of coupled
Burgers equations are obtained via dierential transfor-
mation method,
[11]
variational iteration method,
[12]
and
direct method.
[28]
The dierential quadrature method was rst proposed
by Bellman and coworkers
[29]
and then generalized by Shu
and Richards
[30]
and since then it called GDQM. ABCs
of the GDQM and its development have been explained
in the Shus book.
[31]
Some recent applications of the
method can be found in Refs. [5,3235] and references
cited therein.
In this paper, after applying PDQ method to discretize
the space variable, we use TVD RungeKutta method for
discretizing the time variable to obtain a fully discretized
scheme to solve one-dimensional, two-dimensional, and
coupled Burgers equations. The paper is followed by il-
lustrating some numerical examples and concluded by a
brief summary.
2 Analysis of the Method
In this section, we deal with implementing the method
of PDQ and proceed to employ the TVD RungeKutta
method.
2.1 PDQ Method
For one dimensional case, if a function U(x) is suf-
ciently smooth over the interval [x
L
, x
R
], its rst and
second order derivatives at the grid points x
i
for i =
1, 2, . . . , N are approximated as follows
U
x
(x
i
) =
N

j=1
a
ij
U(x
j
) , (4)
U
xx
(x
i
) =
N

j=1
b
ij
U(x
j
) , (5)
where a
ij
and b
ij
represent weighting coecients of the
rst and second order derivative approximations, respec-
tively. Following the key idea of PDQ,
[31]
the weighting
coecients a
ij
and b
ij
for i, j = 1, 2, . . . , N are determined
as follows
a
ij
=
M
(1)
(x
i
)
(x
i
x
j
)M
(1)
(x
j
)
, i = j , (6)
a
ii
=
N

j=1, j=i
a
ij
, (7)
b
ij
= 2a
ij
_
a
ii

1
x
i
x
j
_
, i = j , (8)
b
ii
=
N

j=1, j=i
b
ij
, (9)
where
M
(1)
(x
i
) =
N

k=1,k=i
(x
i
x
k
) .
Using derivative approximations (4) and (5), Eq. (1)
leads to
u
t
(x
i
, t) = u(x
i
, t)
N

j=1
a
ij
u(x
j
, t)+
N

j=1
b
ij
u(x
j
, t), (10)
in which i = 1, 2, . . . , N and weighting coecients a
ij
and
b
ij
are obtained from Eqs. (6)(9). The system of ordi-
nary dierential equations (10) is solved by using third
order TVD RungeKutta method.
Furthermore, using derivative approximations (4) and
(5), Eq. (3) leads to
U
t
(x
i
, t) =
N

j=1
b
ij
U(x
j
, t) + 2U(x
i
, t)
N

j=1
a
ij
U(x
j
, t)
_
V (x
i
, t)
N

j=1
a
ij
U(x
j
, t) +U(x
i
, t)
N

j=1
a
ij
V (x
j
, t)
_
,
V
t
(x
i
, t) =
N

j=1
b
ij
V (x
j
, t) + 2V (x
i
, t)
N

j=1
a
ij
V (x
j
, t)
_
V (x
i
, t)
N

j=1
a
ij
U(x
j
, t) +U(x
i
, t)
N

j=1
a
ij
V (x
j
, t)
_
, (11)
in which i = 1, 2, . . . , N and weighting coecients a
ij
and
b
ij
are obtained from Eqs. (6)(9). The system of ordi-
nary dierential equations (11) is solved by using third
order TVD RungeKutta method.
For two-dimensional case, if U(x, y) is a two-
dimensional function dened on a rectangular domain, its
rst and second order partial derivatives at the grid points
(x
i
, y
j
) for i = 1, 2, . . . , N and j = 1, 2, . . . , M are approx-
imated as follows
U
x
(x
i
, y
j
) =
N

k=1
a
x
kj
(x
k
, y
j
) , (12)
U
y
(x
i
, y
j
) =
M

k=1
a
y
ik
(x
i
, y
k
) , (13)
U
xx
(x
i
, y
j
) =
N

k=1
b
x
kj
(x
k
, y
j
) , (14)
No. 6 Communications in Theoretical Physics 1011
U
yy
(x
i
, y
j
) =
M

k=1
b
y
ik
(x
i
, y
k
) , (15)
where a
x
ij
and a
y
ij
represent weighting coecients of the
rst order derivative approximations with respect to x and
y, respectively and b
x
ij
and b
y
ij
represent weighting coe-
cients of the second order derivative approximations with
respect to x and y, respectively. Following the key idea
of PDQ,
[31]
the weighting coecients a
x
ij
, a
y
ij
, b
x
ij
, and b
y
ij
are determined as follows
a
x
ij
=
M
(1)
(x
i
)
(x
i
x
j
)M
(1)
(x
j
)
, i = j ,
a
x
ii
=
N

j=1,j=i
a
x
ij
,
_

_
i, j = 1, 2, . . . , N , (16)
a
y
ij
=
P
(1)
(y
i
)
(y
i
y
j
)P
(1)
(y
j
)
, i = j ,
a
y
ii
=
M

j=1,j=i
a
y
ij
,
_

_
i, j = 1, 2, . . . , M , (17)
b
x
ij
= 2a
x
ij
_
a
x
ii

1
x
i
x
j
_
, i = j ,
b
x
ii
=
N

j=1,j=i
b
x
ij
,
_

_
i, j = 1, 2, . . . , N , (18)
b
y
ij
= 2a
y
ij
_
a
y
ii

1
y
i
y
j
_
, i = j ,
b
y
ii
=
M

j=1,j=i
b
y
ij
,
_

_
i, j = 1, 2, . . . , M , (19)
where
M
(1)
(x
i
) =
N

k=1,k=i
(x
i
x
k
) , P
(1)
(y
i
) =
M

k=1,k=i
(y
i
y
k
) .
Using derivative approximations (12)(15), Eq. (2)
leads to
u
t
(x
i
, y
j
, t) = u(x
i
, y
j
, t)
N

k=1
a
x
ij
u(x
k
, y
j
, t)
u(x
i
, y
j
, t)
M

k=1
a
y
ij
u(x
i
, y
k
, t)
+
N

k=1
b
x
ij
u(x
k
, y
j
, t)
+
M

k=1
b
y
ij
u(x
i
, y
k
, t) , (20)
in which i = 1, 2, . . . , N, j = 1, 2, . . . , M, and the weight-
ing coecients a
x
ij
, a
y
ij
, b
x
ij
, and b
y
ij
are obtained from
Eqs. (16)(19). The system of ordinary dierential equa-
tions (20) is solved by using third order TVD Runge
Kutta method.
2.2 TVD RungeKutta Method
The TVD RungeKutta method is used to solve a sys-
tem of ordinary dierential equations such as
u
t
= L(u) , (21)
resulting from a method of lines approximation of the hy-
perbolic conservation law u
t
= f(u)
x
, where the spatial
derivative f(u)
x
is approximated by a TVD nite dier-
ence or nite element approximation denoted by L(u),
which has the property that the total variation of the nu-
merical solution TV (u) =

j
|u
j+1
u
j
|, does not in-
crease, i.e. TV (u
n+1
) TV (u
n
).
Proposition 1 (Ref. [36]) The optimal third order TVD
RungeKutta method for solving Eq. (21) is given by
u
(1)
= u
n
+ tL(u
n
) ,
u
(2)
=
3
4
u
n
+
1
4
u
(1)
+
1
4
tL(u
(1)
) ,
u
n+1
=
1
3
u
n
+
2
3
u
(2)
+
2
3
tL(u
(2)
) , (22)
with a CourantFriedrichsLevy (CFL) coecient c = 1.
3 Numerical Examples
In this section we apply the polynomial based dier-
ential quadrature method (PDQ) to dierent examples.
Accuracy of the method is measured by using the maxi-
mum and relative error norms which are dened by
E

= max
0jN
{|(u(x
j
, t
n
) U(x
j
, t
n
))|} ,
E =

N
j=1
(u(x
j
, t
n
) U(x
j
, t
n
))
2

N
j=1
(U(x
j
, t
n
))
2
,
where u(x
j
, t
n
) and U(x
j
, t
n
) are the exact and numerical
solutions at space x
j
and time t
n
, respectively.
1012 Communications in Theoretical Physics Vol. 56
3.1 One-Dimensional Burgers Equation
Example 1 Consider the one-dimensional Burgers equa-
tion (1), with the solitary wave solution
[3738]
u(x, t) = c/ + (2/) tanh(x ct) ,
where and c are arbitrary constants. Initial and bound-
ary conditions are established from the exact solution. In
Table 1, numerical results are compared with the method
of Chebyshev spectral collocation (CSC).
[38]
Table 1 Comparison of numerical solutions of Exam-
ple 1 by maximum and relative error norms for various
values of and at time t = 0.25 and for x [0, 1].
Method ||E|| ||E||
0.01 7.6410
5
5.4910
4
PDQ 1 0.001 7.6810
7
5.9310
6
0.0001 7.6910
9
5.9810
8
0.01 7.6210
5
5.0410
4
CSC
[38]
0.001 7.8210
7
5.8510
6
0.0001 8.9410
8
4.3110
7
0.01 7.6410
4
5.4910
4
PDQ 0.1 0.001 7.6810
6
5.9310
6
0.0001 7.6910
8
5.9810
8
0.01 7.6210
4
5.0410
4
CSC
[38]
0.001 7.9910
6
5.8810
6
0.0001 1.6710
6
4.3110
7
It must be pointed out that in Ref. [39], GDQM has
been applied to discretize a similar problem. Unfortu-
nately, established numerical results do not possess high
accuracy because of using the explicit Euler method in
solving the resulting system of dierential equations. We
solve this system through TVD-RK to obtain more accu-
rate numerical results.
Example 2 Consider the one-dimensional Burgers equa-
tion (1), with the initial condition
u(x, 0) = sin(x) , 0 x 1 ,
and the boundary conditions u(0, t) = u(1, t) = 0. The
exact solution was found in terms of the innite series by
Cole
[40]
as
u(x, t) = 2

j=1
ja
j
sin(jx) exp(j
2

2
t)
a
0
+ 2

j=1
a
j
cos(jx) exp(j
2

2
t)
,
where
a
0
=
_
1
0
exp[(2)
1
(1 cos(x))] dx,
a
j
= 2
_
1
0
exp[(2)
1
(1 cos(x))] cos(jx)dx.
In Table 2, numerical results at time t = 0.1 with
= 1 and parameters x = 0.1 and t = 0.001
are compared with the methods of cubic B-spline quasi-
interpolation (CBQI),
[41]
cubic B-spline (CB),
[42]
implicit
nite dierence (IFdM),
[43]
boundary element (BEM),
[44]
modied Adomian decomposition (ADM),
[45]
nite dif-
ference (DFDM),
[46]
least-squares quadratic B-spline -
nite element (BFEM),
[16]
and local discontinuous Galerkin
(LDG).
[47]
In Table 3, numerical results of PDQ method
at time t = 1 and = 0.1 with parameters x = 0.05
and t = 0.001 are compared with the methods of cubic
B-spline quasi-interpolation (CBQI),
[41]
implicit nite dif-
ference (IFDM),
[43]
boundary element (BEM),
[44]
Hon and
Maos scheme (HM),
[48]
multiquadric quasi-interpolation
(MQQI)
[49]
and local discontinuous Galerkin (LDG).
[47]
In Fig. 1, we display the numerical results with t = 0.1
for = 1, 0.1. And also the numerical results of the PDQ
method, for t = 0.0, 0.2, 0.4, 0.6, 0.8, 1.0 together with the
initial data, are illustrated in Fig. 2, which correspond to
= 0.1, 0.01, and 0.0001.
Fig. 1 (Color online) Numerical results at t = 0.1 for = 1 (a) and = 0.1 (b).
No. 6 Communications in Theoretical Physics 1013
Fig. 2 (Color online) Numerical results at t = 1, h = 0.05 for = 0.1 (a), = 0.01 (b), and h = 0.03 and
= 0.0001 (c).
Table 2 Comparison of numerical results of Example 2 at time t = 0.1 for = 1.
Method x = 0.1 x = 0.3 x = 0.5 x = 0.7 x = 0.9
PDQ (present) 0.109 53 0.291 89 0.371 57 0.309 90 0.120 68
CBQI
[41]
0.109 51 0.291 82 0.371 47 0.309 81 0.120 65
CB
[42]
0.109 52 0.291 84 0.371 49 0.309 83 0.120 65
IFDM
[43]
0.110 09 0.293 35 0.373 42 0.311 44 0.121 28
BEM
[44]
0.109 31 0.291 24 0.370 70 0.309 11 0.120 31
ADM
[45]
0.109 81 0.292 62 0.372 49 0.310 66 0.120 98
BFEM
[16]
0.109 78 0.292 38 0.372 12 0.310 44 0.120 97
DFDM
[46]
0.109 47 0.291 70 0.371 33 0.309 70 0.120 61
LDG (k = 1)
[47]
0.109 54 0.291 89 0.371 57 0.309 90 0.120 68
LDG (k = 2)
[47]
0.109 54 0.291 89 0.371 57 0.309 90 0.120 69
Exact 0.109 54 0.291 90 0.371 58 0.309 91 0.120 69
Table 3 Comparison of numerical results of Example 2 at time t = 1 for = 0.1.
Method x = 0.1 x = 0.3 x = 0.5 x = 0.7 x = 0.9
PDQ (present) 0.066 31 0.192 78 0.291 91 0.308 09 0.146 06
CBQI
[41]
0.066 28 0.192 69 0.291 75 0.307 91 0.145 83
HM
[48]
0.0664 0.1928 0.2919 0.3079 0.1459
IFDM
[43]
0.066 89 0.194 45 0.294 48 0.311 07 0.147 69
BEM
[44]
0.066 44 0.192 63 0.291 39 0.307 11 0.145 07
MQQI (c = 7.2 10
3
)
[49]
0.071 24 0.193 39 0.285 17 0.292 88 0.1354 2
LDG (k = 1)
[47]
0.066 31 0.192 78 0.291 91 0.308 08 0.146 06
LDG (k = 2)
[47]
0.066 32 0.192 78 0.291 91 0.308 09 0.146 06
Exact 0.066 32 0.192 79 0.291 91 0.308 10 0.146 06
1014 Communications in Theoretical Physics Vol. 56
3.2 Two-Dimensional Burgers Equation
Example 3 Consider the two-dimensional Burgers equa-
tion (2), with the following exact solution
[50]
u(x, y, t) =
1
1 + exp((x +y t)/2)
.
The initial and boundary conditions are taken from the
exact solution. In Table 4, numerical results at times
t = 0.05 and t = 0.25 with = 0.1, 0.01 and parame-
ters N = M = 10 and t = 0.001 are compared with the
methods of Chebyshev spectral collocation (CSC)
[38]
and
lattice Boltzmann (LBM).
[23]
Table 4 Comparison of numerical results of Example 3
at dierent times.
Method N M t t ||E||
10 10 0.05 0.005 0.1 2.7010
5
PDQ
10 10 0.05 0.0005 0.01 2.7310
7
10 10 0.25 0.005 0.1 1.2010
4
10 10 0.25 0.0005 0.01 1.2110
6
10 10 0.05 0.005 0.1 1.2810
6
CSC
[38]
30 30 0.05 0.0005 0.01 4.1410
5
30 30 0.25 0.005 0.1 4.3210
3
10 10 0.25 0.1 1.0610
2
LBM
[23]
20 20 0.25 0.1 3.0710
3
80 80 0.25 0.01 5.8910
2
3.3 Coupled Burgers Equations
Example 4 Consider the coupled Burgers equations (3)
for = 2 and dierent values of and at times t = 0.5
and t = 1.0. The exact solution of the equation is given
by Ref. [37] as
u(x, t) = a
0
(1 tanh(A(x 2At))) ,
v(x, t) = a
0
_
2 1
2 1
tanh(A(x 2At))
_
, (23)
where a
0
= 0.05 and A = a
0
(4 1)/(4 2). The
initial and boundary conditions are taken from the ex-
act solutions. In Tables 5 and 6, numerical results are
compared with the methods of Chebyshev spectral collo-
cation (CSC),
[38]
Fourier pseudo-spectral (FPM)
[51]
and
cubic B-spline collocation (CBC).
[52]
In Fig. 3, we display
the numerical and the exact solutions for u and v values
when N = 10, = 0.1, t = 1, = 1, = 2, and a = 0.1.
Table 5 Comparisons of errors at dierent times for
u(x, t) of Example 4.
Method t ||E|| ||E||
0.5 0.1 0.3 1.0010
4
2.0210
3
PDQ
0.3 0.03 2.5210
4
5.0710
3
1 0.1 0.3 2.0110
4
4.0310
3
0.3 0.03 5.0410
4
1.0010
2
FPM
[51]
0.5 0.1 0.3 9.61910
4
3.24510
5
0.3 0.03 4.31010
4
2.73310
5
1 0.1 0.3 1.15310
3
2.40510
5
0.3 0.03 1.26810
3
2.83210
5
0.5 0.1 0.3 4.3810
5
1.4410
3
CSC
[38]
0.3 0.03 4.5810
5
6.6810
4
1 0.1 0.3 8.6610
5
1.2710
3
0.3 0.03 9.1610
5
1.3010
3
0.5 0.1 0.3 4.1610
5
6.7310
4
CBC
[52]
0.3 0.03 4.5910
5
7.3210
4
1 0.1 0.3 8.2510
5
1.3210
3
0.3 0.03 9.1810
5
1.4510
3
Table 6 Comparisons of errors at dierent times for
v(x, t) of Example 4.
Method t ||E|| ||E||
0.5 0.1 0.3 3.8010
5
1.5610
3
PDQ
0.3 0.03 1.8510
4
1.5910
3
1 0.1 0.3 7.5810
5
3.1010
3
0.3 0.03 3.6710
4
3.1510
3
0.5 0.1 0.3 3.3310
4
2.7410
5
FPM
[51]
0.3 0.03 1.1410
3
2.4510
4
1 0.1 0.3 1.1610
3
3.7410
5
0.3 0.03 1.6310
3
4.5210
4
0.5 0.1 0.3 4.9910
5
5.4210
4
CSC
[38]
0.3 0.03 1.8110
4
1.2010
3
1 0.1 0.3 9.9210
5
1.2910
3
0.3 0.03 3.6210
4
2.3510
3
0.5 0.1 0.3 1.4810
4
9.0410
4
CBC
[52]
0.3 0.03 5.7210
4
1.5910
3
1 0.1 0.3 4.7710
5
1.2510
3
0.3 0.03 3.6110
4
2.2510
3
Fig. 3 (Color online) Numerical and exact solutions of Example 4 for N = 10, = 0.1, t = 1, = 1, = 2, and a = 0.1.
No. 6 Communications in Theoretical Physics 1015
4 Summary
In this paper, the polynomial-based generalized dif-
ferential quadrature method is applied to the Burgers
and coupled Burgers equations and the obtained system
of ordinary dierential equations is solved via the TVD
RungeKutta method. In comparison with methods ap-
plied in the literature such as spectral methods, this ap-
proach is conservative and produce reasonable numerical
results. As mentioned in Ref. [53], not only the applied
method achieves the high accuracy and spectacular con-
vergence rates of spectral methods but also its particular
advantage lies in its ease of implementation and its abil-
ity to use more general approximating polynomials, i.e.
the standard orthogonal polynomials of spectral methods
need not be used.
References
[1] C.Q. Dai, Q. Yang, and Y.Y. Wang, Commun. Theor.
Phys. 55 (2011) 622.
[2] R. Mokhtari, Commun. Theor. Phys. 55 (2011) 204.
[3] R. Mokhtari and M. Mohammadi, Int. J. Nonlinear Sci.
Numer. Simul. 10 (2009) 779.
[4] R. Mokhtari and M. Mohammadi, Comput. Phys. Com-
mum. 181 (2010) 1266.
[5] R. Mokhtari, A. Samadi Toodar, and N.G. Chegini, Chin.
Phys. Lett. 28 (2011) 020202.
[6] H. Bateman, Mon. Weather Rev. 43 (1915) 163.
[7] J.M. Burgers, Adv. Appl. Mech. 1 (1948) 171.
[8] J.M. Burgers, The Nonlinear Diusion Equation, Reiedl,
Dordtrecht (1974).
[9] Z.S. L

U, Commun. Theor. Phys. 44 (2005) 987.


[10] A.M. Wazwaz, Appl. Math. Comput. 190 (2007)
11981206.
[11] R. Abazari and A. Borhanifar, Appl. Math. Comput. 59
(2010) 2711.
[12] M.A. Abdou and A.A. Soliman, Appl. Math. Comput.
181 (2005) 245251.
[13] M.M. Rashidi, G. Domairry, and S. Dinarvand, Appl.
Math. Comput. 14 (2009) 708717.
[14] P. Huang and A. Abduwali, Appl. Math. Comput. 59
(2010) 2452.
[15] X.H. Zhang, J. Ouyang, and L. Zhang, Eng. Anal. Bound-
ary Elem. 33 (2009) 356.
[16] S. Kutluay, A. Esen, and I. Dag, Appl. Math. Comput.
167 (2004) 21.
[17] M. Sari and G. Gurarslan, Appl. Math. Comput. 208
(2009) 475.
[18] C.G. Zhu and R.H. Wang, Appl. Math. Comput. 208
(2009) 260.
[19] E.N. Aksan, Appl. Math. Comput. 170 (2005) 895.
[20] S.S. Xie, S. Heo, S. Kim, G. Woo, and S. Yi, J. Comput.
Appl. Math. 214 (2008) 417.
[21] H. Zhu, H. Shu, and M. Ding, Appl. Math. Comput. 60
(2010) 840.
[22] L.L. Zhang and X.Q. Liu, Commun. Theor. Phys. 52
(2009) 784.
[23] Y. Duan and R. Liu, Appl. Math. Comput. 206 (2007)
432.
[24] D.L. Young, C.M. Fan, S.P. Hu, and S.N. Atluri, Eng.
Anal. Boundary Elem. 32 (2008) 395.
[25] X. Wu and J. Zhang, Appl. Math. Comput. 56 (2008)
242.
[26] S.E. Esipov, Phys. Rev. E 52 (1995) 3711.
[27] C.J. Bing, G.X. Guo, and Q.Z. Jun, Chin. Phys. 19 (2010)
090403.
[28] L. Huang, Commun. Theor. Phys. 45 (2006) 781.
[29] R. Bellman, B.G. Kashef, and J. Casti, J. Comput. Phys.
10 (1972) 40.
[30] C. Shu and B. Richards, Int. J. Numer. Methods Fluids
15 (1992) 791.
[31] C. Shu, Dierential Quadrature and Its Application in
Engineering, Springer-Verlag London Limited (2000).
[32] G. Meral and M. Tezer-Sezgin, Int. J. Numer. Meth.
Biomedical Eng. 27 (2011) 485.
[33] G. Gurarslan and M. Sari, Int. J. Numer. Meth. Biomed-
ical Eng. 27 (2011) 69.
[34] Y. Zhao, Z. Zong, and Z.R. Li, Appl. Math Mech.-
ENGLISH EDITION 32 (2011) 349.
[35] Q.A. Shen, Int. J. Comput. Math. 88) 285.
[36] S. Gottlieb and C.W. Shu, Math. Comput. 67 (1998) 73.
[37] A.A. Soliman, Phys. Lett. A 361 (2006) 394.
[38] A.H. Khater, R.S. Temsah, and M.M. Hassan, Appl.
Math. Comput. 222 (2008) 333.
[39] M. Mestrovic, Generalized Dierential Quadrature
Method for Burgers Equation, Paper No. PVP2003-1905
pp. 211-214, doi:10.1115/PVP2003-1905 ASME 2003 Pre-
ssure Vessels and Piping Conference (PVP2003) July 20-
24, 2003, Cleveland, Ohio, USA.
[40] J.D. Cole, Quart. Appl. Math. 9 (1951) 225.
[41] C.G. Zhu and R.H. Wang, Appl. Math. Comput. 208
(2009) 260.
[42] I. Dag, D. Irk, and B. Saka, Appl. Math. Comput. 163
(2005) 199.
[43] A.R. Bahadir, Int. J. Appl. Math. 8 (1999) 897.
[44] A.R. Bahadir and M. Saglam, Appl. Math. Comput. 160
(2005) 663.
[45] S. Abbasbandy and M.T. Darvishi, Appl. Math. Comput.
163 (2005) 1265.
[46] K. Pandey, L. Verma, and A.K. Verma, Appl. Math. Com-
put. 215 (2009) 2206.
[47] Z.G. Zhong, Y.X. Jun, and W. Di, Appl. Math. Comput.
216 (2010) 3671.
[48] Y.C. Hon and X.Z. Mao, Appl. Math. Comput. 95 (1998)
37.
[49] R. Chen and Z. Wu, Appl. Math. Comput. 172 (2006)
472.
[50] O. Goyon, Int. J. Numer. Methods Fluids 22 (1996) 937.
[51] A. Rashid and A.I.B.M. Ismail, Comput. Methods Appl.
Math. 9 (2009) 412420.
[52] R.C. Mittal and G. Arora, Commun Nonlinear Sci. Nu-
mer. Simul. 16 (2011) 1304.
[53] G. Mansell, W. Merryeld, B. Shizgal, and U. Weinert,
Comput. Meth. Appl. Mech. Eng. 104 (1993) 295.

Anda mungkin juga menyukai