Available online at [Link].

com

Chaos, Solitons and Fractals 40 (2009) 1520–1529

[Link]/locate/chaos

Variational iteration method for solving partial

differential equations with variable coefficients
a,* b
A.H.A. Ali , K.R. Raslan
a
Mathematics Department, Faculty of Science, Menoufia University, Shebein El-Koom, Egypt
b
Department of Mathematics, Faculty of Science, Al-Azhar University, Nasr-City, Cairo, Egypt

Accepted 8 August 2007

Communicated by Prof. Ji-Huan He

Abstract

An extremely simple and elementary but rigorous derivation of exact solutions of partial differential equations in
different dimensions with variable coefficients is given using the variational iteration method. The efficiency of the con-
sidered method is illustrated by some examples. The results show that the proposed iteration technique, without line-
arization or small perturbation, is very effective and convenient.
Ó 2007 Elsevier Ltd. All rights reserved.

1. Introduction

A physically meaningful mathematical derivation of exact solutions of partial differential equations in different
dimensions with variable coefficients seemed to be something which should come at the end of physical understanding
of almost everything. This plausible attitude may have contributed to the fact that until very recently no a universal
approach existed, thus the study of numerical methods for the solutions of various partial differential equations (PDEs)
has enjoyed an intense period of activity over the last 40 years from both theoretical and practical points of view.
Improvements in numerical techniques, together with the rapid advance in computer technology, have meant that many
of the PDEs arising in engineering and scientific applications, which were previously intractable, can now be routinely
solved [1]. The Adomian decomposition method, which was developed by Adomian [2], depends only on the initial con-
ditions and results in a solution in series which converges to the exact solution of the problem. In recent years, other
ansatz methods have been developed, such as the exp-function method [3], variational approaches [4,5], the tanh
method [6–8], extended tanh function method [9,10], the modified extended tanh function method [11,12], the first inte-
gral method [13–17], and others [18]. Ansatz methods cannot applied to solve the partial differential equations in dif-
ferent dimensions with variable coefficients in x, y, and z, also the variational iteration method based on the initial
conditions, but the ansatz does not depend on the initial conditions.

*
Corresponding author. Tel.: +20 (02)22871081; fax: +20 (02)22744159.
E-mail addresses: ahaali_49@[Link] (A.H.A. Ali), kamal_raslan@[Link] (K.R. Raslan).

0960-0779/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/[Link].2007.09.031
 A.H.A. Ali, K.R. Raslan / Chaos, Solitons and Fractals 40 (2009) 1520–1529 1521

The variational iteration method (VIM) [18–22] gives rapid convergent successive approximations of the exact solu-
tion if such a solution exists without any restrictive assumption or transformation that may change the physical behav-
ior of the problem. The VIM gives several successive approximations through an iteration formulation of a correction
functional. Moreover, the VIM reduces the size of calculation, while the tedious calculation is required for Adomian
polynomials; hence the iteration is direct and straightforward.
In this paper, we use the VIM to find exact solutions of the PDEs in one, two and three dimensions with var-
iable coefficients which will be useful in numerical studies. We hope that we can simplify the derivation procedure
and reduce the mathematics to a bare minimum which as the reader will see, does not exceed a few line of elemen-
tary arithmetic.

2. The variational iteration method [18–22]

To illustrate the basic idea of the VIM we consider the following general PDE:
Lt u þ Lx u þ Ly u þ Lz u þ Nu ¼ gðx; y; z; tÞ: ð1Þ
where Lt, Lx, Ly and Lz are linear operators of t, x, y, and z, respectively, and N is a nonlinear operator. According to
the VIM, we can express the following correction functional in t-, x-, y- and z-directions, respectively, as follows:
Z t
unþ1 ðx; y; z; tÞ ¼ un ðx; y; z; tÞ þ k1 fLs un þ ðLx þ Ly þ Lz þ NÞ~
un  ggds; ð2aÞ
Z0 x
unþ1 ðx; y; z; tÞ ¼ un ðx; y; z; tÞ þ k2 fLs un þ ðLt þ Ly þ Lz þ NÞ~
un  ggds; ð2bÞ
Z0 y
unþ1 ðx; y; z; tÞ ¼ un ðx; y; z; tÞ þ k3 fLs un þ ðLx þ Lt þ Lz þ NÞ~
un  ggds; ð2cÞ
Z0 z
unþ1 ðx; y; z; tÞ ¼ un ðx; y; z; tÞ þ k4 fLs un þ ðLx þ Ly þ Lt þ NÞ~
un  ggds: ð2dÞ
0

where k1, k2, k3 and k4 are general Lagrange multipliers [18], which can be identified optimally via the variational
theory [5,18], and ~un is a restricted variation which means d~ un ¼ 0. By this method, we determine first the La-
grange multipliers ki (i = 1, 2, 3, 4) which will be identified optimally. The successive approximations un+1, n P 0,
of the solution u will be readily obtained by suitable choice of trial function u0. Consequently, the solution is given
as
uðx; y; z; tÞ ¼ lim un ðx; y; z; tÞ: ð3Þ
n!1

Applications of the VIM to various nonlinear problems can be found in details in Refs. [23–33]. In this paper, we will
apply the VIM to the following problems to illustrate the strength of the method.

2.1. Parabolic-like equations

In this section, we consider the parabolic like equation in three dimensions which can be written in the form:
ut þ f1 ðx; y; zÞuxx þ f2 ðx; y; zÞuyy þ f3 ðx; y; zÞuzz ¼ 0; ð4Þ

with the initial condition

uðx; y; z; 0Þ ¼ f4 ðx; y; zÞ: ð5Þ
According to the VIM, we construct a correction functional for Eq. (4) in the form
Z t
unþ1 ðx; y; z; tÞ ¼ un ðx; y; z; tÞ þ kfðun Þs þ f1 ðx; y; zÞð~
un Þxx þ f2 ðx; y; zÞð~
un Þyy þ f3 ðx; y; zÞð~
un Þzz gds; ð6Þ
0

where n P 0 and u0(x, y, z, t) = u(x, y, z, 0).

The VIM solution u(x, y, z, t) can be expressed as
uðx; y; z; tÞ ¼ lim un ðx; y; z; tÞ; ð7Þ
n!1

where un(x, y, z, t) will be determined in a recursive manner.

1522 A.H.A. Ali, K.R. Raslan / Chaos, Solitons and Fractals 40 (2009) 1520–1529

2.2. Hyperbolic-like equations

We consider here the three-dimensional hyperbolic-like equations of the form

utt þ g1 ðx; y; zÞuxx þ g2 ðx; y; zÞuyy þ g3 ðx; y; zÞuzz ¼ 0; ð8Þ

and the initial conditions are

uðx; y; z; 0Þ ¼ g4 ðx; y; zÞ; ut ðx; y; z; 0Þ ¼ g5 ðx; y; zÞ: ð9Þ

The variational iteration scheme of Eq. (8) takes the form

Z t
unþ1 ðx; y; z; tÞ ¼ un ðx; y; z; tÞ þ kfðun Þss þ g1 ðx; y; zÞð~
un Þxx þ g2 ðx; y; zÞð~
un Þyy þ g3 ðx; y; zÞð~
un Þzz gds ð10Þ
0

where n P 0 and u0(x, y, z, t) = u(x, y, z, 0) + tut(x, y, z, 0). The VIM solution u(x, y, z, t) takes the form
uðx; y; z; tÞ ¼ lim un ðx; y; z; tÞ; ð11Þ
n!1

where un(x, y, z, t) will be determined in a recursive manner.

Example 1. Consider the one-dimensional heat equation with variable coefficients in the form [34]

x2
ut ðx; tÞ  uxx ðx; tÞ ¼ 0; ð12Þ
2
and the initial condition u(x, 0) = x2. The variational iteration scheme of Eq. (12) has the form
Z t  
x2
unþ1 ðx; tÞ ¼ un ðx; tÞ þ k ðun ðx; sÞÞs  ð~un ðx; sÞÞxx ds; ð13Þ
0 2

where n P 0 and u0(x, t) = x2. This yields the stationary conditions

1 þ kjs¼t ¼ 0; k0 ðsÞ ¼ 0: ð14Þ

Hence, the Lagrange multiplier is

k ¼ 1: ð15Þ
Substituting this value of the Lagrange multiplier into the functional (13) gives the iteration formula
Z t 
x2
unþ1 ðx; tÞ ¼ un ðx; tÞ  ðun ðx; sÞÞs  ðun ðx; sÞÞxx ds: ð16Þ
0 2
We start with an initial approximation: u0(x, t) = u(x, 0) = x2, and using the iteration formula (16), we obtain the fol-
lowing successive approximations
u1 ðx; tÞ ¼ ð1 þ tÞx2 ;
 
t2 2
u2 ðx; tÞ ¼ 1 þ t þ x;
2!
 
t2 t3 2
u3 ðx; tÞ ¼ 1 þ t þ þ x; ð17Þ
2! 3!
..
.
 
t2 t3 tn 2
un ðx; tÞ ¼ 1 þ t þ þ þ    þ x:
2! 3! n!
Then, the exact solution is given as
 
t2 tn 2
uðx; tÞ ¼ lim 1 þ t þ þ    þ x ¼ et x2 : ð18Þ
n!1 2! n!
A comparison between the VIM and the ADM is given in Table 1.
 A.H.A. Ali, K.R. Raslan / Chaos, Solitons and Fractals 40 (2009) 1520–1529 1523

Table 1
A comparison between the VIM and the ADM
VIM ADM [34]
2
u0(x, t) = x u0(x, t) = x2
u1(x, t) = x2(1 + t) u1(x, t) = x2t
 
t2 2
u2 ðx; tÞ ¼ x2 1 þ t þ 2! u2 ðx; tÞ ¼ x2 2!
t
 
t2 t3 3
u3 ðx; tÞ ¼ x2 1 þ t þ 2! þ 3! u3 ðx; tÞ ¼ x2 3!
t
 
t2 t3 t4 4
u4 ðx; tÞ ¼ x2 1 þ t þ 2! þ 3! þ 4! u4 ðx; tÞ ¼ x2 4!
t

.. ..
. .
P1
uðx; tÞ ¼ limn!1 un ðx; tÞ uðx; tÞ ¼ n¼0 un ðx; tÞ

Example 2. Consider the two-dimensional heat equation with variable coefficients as [34]
y2 x2
ut ðx; y; tÞ  uxx ðx; y; tÞ  uyy ðx; y; tÞ ¼ 0; ð19Þ
2 2
where the initial condition is u(x, y, 0) = y2. The correction functional for Eq. (19) can be written as
Z t  
y2 x2
unþ1 ðx; y; tÞ ¼ un ðx; y; tÞ þ k ðun ðx; y; sÞÞs  ð~un ðx; y; sÞÞxx  ð~un ðx; y; sÞÞyy ds; ð20Þ
0 2 2
where n P 0 and u0(x, y, t) = u(x, y, 0) = y2. The stationary conditions are
1 þ kjs¼t ; k0 ðsÞ ¼ 0: ð21Þ
This in turn gives
k ¼ 1: ð22Þ
Substituting this value of the Lagrange multiplier into the functional (20) gives the iteration formula
Z t 
y2 x2
unþ1 ðx; y; tÞ ¼ un ðx; y; tÞ  ðun ðx; y; sÞÞs  ðun ðx; y; sÞÞxx  ðun ðx; y; sÞÞyy ds: ð23Þ
0 2 2
We begin with an initial arbitrary approximation: u0(x, y, t) = u(x, y, 0) = y2, and using the iteration formula (23), we
obtain the following successive approximations
u1 ðx; y; tÞ ¼ tx2 þ y 2 ;
 
t2 2
u2 ðx; y; tÞ ¼ tx2 þ 1 þ y ;
2!
   
t3 2 t2 2
u3 ðx; y; tÞ ¼ t þ x þ 1þ y ; ð24Þ
3! 2!
..
.
   
t3 t2nþ1 2 t2 t2n
un ðx; y; tÞ ¼ t þ þ    þ x þ 1 þ þ  þ y2:
3! ð2n þ 1Þ! 2! ð2nÞ!
Then, the exact solution is given by
   
t3 t2nþ1 t2 t2n
uðx; y; tÞ ¼ lim t þ þ    þ x2 þ 1 þ þ    þ y 2 ¼ x2 sinh t þ y 2 cosh t: ð25Þ
n!1 3! ð2n þ 1Þ! 2! ð2nÞ!

Example 3. Consider the three-dimensional heat equation with variable coefficients as [34]

1 2
ut  ðxyzÞ4  ðx uxx þ y 2 uyy þ z2 uzz Þ ¼ 0; ð26Þ
36
and the initial condition u(x, y, z, 0) = 0. The correction functional for Eq. (26) is given as
1524 A.H.A. Ali, K.R. Raslan / Chaos, Solitons and Fractals 40 (2009) 1520–1529
Z t
unþ1 ðx; y; z; tÞ ¼ un ðx; y; z; tÞ þ kfðun ðx; y; z; sÞÞs  ðxyzÞ4  Fe gds; ð27Þ
0

where n P 0, Fe ¼ 361 ðx2 ð~un ðx; y; z; sÞÞxx þ y 2 ð~un ðx; y; z; sÞÞyy þ z2 ð~

un ðx; y; z; sÞÞzz Þ, and u0(x, y, z, t) = 0. Then, the stationary
conditions are given by

1 þ kjs¼t ; k0 ðsÞ ¼ 0: ð28Þ

This again yields

k ¼ 1: ð29Þ
Substituting this value of the Lagrange multiplier into the functional (27) gives the iteration formula
Z t
unþ1 ðx; y; z; tÞ ¼ un ðx; y; z; tÞ  kfðun ðx; y; z; sÞÞs  ðxyzÞ4  F gds; ð30Þ
0

where F ¼ 361 ðx2 ðun ðx; y; z; sÞÞxx þ y 2 ðun ðx; y; z; sÞÞyy þ z2 ðun ðx; y; z; sÞÞzz Þ.
Starting with an initial approximation: u0(x, y, z, t) = u(x, y, z, 0) = 0, and using the iteration formula (30), we obtain
the following successive approximations

u1 ðx; y; z; tÞ ¼ tx4 y 4 z4 ;
 
t2 4 4 4
u2 ðx; y; z; tÞ ¼ t þ xy z;
2
 
t2 t3 4 4 4
u3 ðx; y; z; tÞ ¼ t þ þ xy z; ð31Þ
2! 3!
..
.
 
t2 t3 tn 4 4 4
un ðx; y; z; tÞ ¼ t þ þ þ    þ xy z:
2! 3! n!

Then the exact solution is given by

 
t2 tn 4 4 4
uðx; y; z; tÞ ¼ lim t þ þ    þ x y z ¼ x4 y 4 z4 ðet  1Þ: ð32Þ
n!1 2 n!

Example 4. Consider the one-dimensional wave equation with variable coefficients as [34]

x2
utt ðx; tÞ  uxx ðx; tÞ ¼ 0; ð33Þ
2
with the initial conditions u(x, 0) = x, ut(x, 0) = x2. The correction functional for the Eq. (33) is given as
Z t  
x2
unþ1 ðx; tÞ ¼ un ðx; tÞ þ k ðun ðx; sÞÞss  ð~un ðx; sÞÞxx ds; ð34Þ
0 2

where n P 0 and u0(x, t) = x + t x2. Then, the stationary conditions are

1  k0 js¼t ¼ 0; kjs¼t ¼ 0; k00 ðsÞ ¼ 0: ð35Þ

This in turn gives

k ¼ s  t: ð36Þ

Substituting this value of the Lagrange multiplier into the functional (34) gives the iteration formula
Z t  
x2
unþ1 ðx; tÞ ¼ un ðx; tÞ þ ðs  tÞ ðun ðx; sÞÞss  ðun ðx; sÞÞxx ds: ð37Þ
0 2

Selecting the initial approximation: u0(x, t) = x + tx2, with the iteration formula (37), we obtain the following successive
approximations
 A.H.A. Ali, K.R. Raslan / Chaos, Solitons and Fractals 40 (2009) 1520–1529 1525
 
t3 2
u1 ðx; tÞ ¼ x þ t þ x;
3!
 
t3 t5 2
u2 ðx; tÞ ¼ x þ t þ þ x;
3! 5!
 
t3 t5 t7 2 ð38Þ
u3 ðx; tÞ ¼ x þ t þ þ þ x;
3! 5! 7!
..
.
 
t3 t5 t7 t2nþ1
un ðx; tÞ ¼ x þ t þ þ þ þ    þ x2 :
3! 5! 7! ð2n þ 1Þ!
Thus, the exact solution is given in the closed form as
 
t3 t2nþ1
uðx; tÞ ¼ lim x þ t þ þ . . . þ x2 ¼ x þ x2 sinh t: ð39Þ
n!1 3! ð2n þ 1Þ!

Example 5. Consider the two-dimensional wave equation with variable coefficients as [34]

x2 y2
utt ðx; y; tÞ  uxx ðx; y; tÞ  uyy ðx; y; tÞ ¼ 0; ð40Þ
12 12
with the initial conditions u(x, y, 0) = y2, ut(x, y, 0) = y4. The correction functional for Eq. (40) is given as
Z t  
x2 y2
unþ1 ðx; y; tÞ ¼ un ðx; y; tÞ þ k ðun ðx; y; sÞÞss  ð~un ðx; y; sÞÞxx  ð~un ðx; y; sÞÞyy ds ð41Þ
0 12 12
where n P 0 and u0(x, y, t) = x4 + ty4. Then, the stationary conditions are given by
1  k0 js¼t ¼ 0; kjs¼t ¼ 0; k00 ðsÞ ¼ 0: ð42Þ
This in turn gives
k ¼ s  t: ð43Þ
Substituting this value of the Lagrange multiplier into the functional (41) gives the iteration formula
Z t  
x2 y2
unþ1 ðx; y; tÞ ¼ un ðx; y; tÞ þ ðs  tÞ ðun ðx; y; sÞÞss  ðun ðx; y; sÞÞxx  ðun ðx; y; sÞÞyy ds: ð44Þ
0 12 12
Beginning with an arbitrary initial approximation: u0(x, y, t) = x4 + ty4, and using the iteration formula (44), we obtain
the following successive approximations
 
t2 t3
u1 ðx; y; tÞ ¼ x4 1 þ þ y 4 ðt þ Þ;
2! 3!
 2 4
  
t t t3 t5
u2 ðx; y; tÞ ¼ x4 1 þ þ þ y4 t þ þ ;
2! 4! 3! 5! ð45Þ
..
.
   
t 2 t4 t2n t3 t5 t2nþ1
un ðx; y; tÞ ¼ x4 1 þ þ þ    þ þ y4 t þ þ þ    þ :
2! 4! ð2nÞ! 3! 5! ð2n þ 1Þ!
Hence, the exact solution is
   
t2 t2n t3 t2nþ1
uðx; y; tÞ ¼ lim x4 1 þ þ    þ þ y4 t þ þ    þ ¼ x4 cosh t þ y 4 sinh t: ð46Þ
n!1 2 ð2nÞ! 6 ð2n þ 1Þ!

Example 6. Consider the three-dimensional wave equation with variable coefficients [34]

1
utt  ðx2 þ y 2 þ z2 Þ  ðx2 uxx þ y 2 uyy þ z2 uzz Þ ¼ 0; ð47Þ
2
with the initial conditions u(x, y, z, 0) = 0, ut(x, y, z, 0) = x2 + y2  z2. The correction functional for Eq. (47) is given as
1526 A.H.A. Ali, K.R. Raslan / Chaos, Solitons and Fractals 40 (2009) 1520–1529
Z t
unþ1 ðx; y; z; tÞ ¼ un ðx; y; z; tÞ þ kfðun ðx; y; z; sÞÞss  ðx2 þ y 2 þ z2 Þ  F~ gds; ð48Þ
0

where n P 0, Fe ¼ 12 ðx2 ð~un ðx; y; z; sÞÞxx þ y 2 ð~un ðx; y; z; sÞÞyy þ z2 ð~

un ðx; y; z; sÞÞzz Þ, and u0(x, y, z, t) = t(x2 + y2  z2). Then,
the stationary conditions are given by
1  k0 js¼t ¼ 0; kjs¼t ¼ 0; k00 ðsÞ ¼ 0: ð49Þ

This in turn gives

k ¼ s  t: ð50Þ

Substituting this value of the Lagrange multiplier into the functional (48) gives the iteration formula
Z t
unþ1 ðx; y; z; tÞ ¼ un ðx; y; z; tÞ þ ðs  tÞfðun ðx; y; z; sÞÞss  ðx2 þ y 2 þ z2 Þ  F gds; ð51Þ
0

where F ¼ 12 ðx2 ðun ðx; y; z; sÞÞxx þ y 2 ðun ðx; y; z; sÞÞyy þ z2 ðun ðx; y; z; sÞÞzz Þ.
Then, the exact solution is given in the closed form by
   
t2 tn t2 tn
uðx; y; z; tÞ ¼ lim t þ þ    þ ðx2 þ y 2 Þ þ z2 t þ     þ
n!1 2 n! 2 n!
¼ ðx2 þ y 2 Þet þ z2 et  ðx2 þ y 2 þ z2 Þ: ð52Þ

Example 7. Consider the linear Klein–Gordon equation in the form

utt ðx; tÞ  uxx ðx; tÞ  uðx; tÞ ¼ 0; ð53Þ

with the initial conditions u(x, 0) = 1 + sin x, ut(x, 0) = 0. The correction functional for Eq. (53) is given as
Z t
unþ1 ðx; tÞ ¼ un ðx; tÞ þ kfðun ðx; sÞÞss  ð~un ðx; sÞÞxx  uðx; sÞgds; ð54Þ
0

where n P 0 and u0(x, t) = u(x, 0) = 1 + sin x. Then, the stationary conditions are
k00 ðsÞ  kðsÞ ¼ 0; kjs¼t ¼ 0; k0 js¼t ¼ 1: ð55Þ
This in turn gives
est  ets
k¼ : ð56Þ
2
Substituting this value of the Lagrange multiplier into the functional (54) gives the iteration formula
Z t  st 
e  ets
unþ1 ðx; tÞ ¼ un ðx; tÞ þ fðun ðx; sÞÞss  ðun ðx; sÞÞxx  uðx; sÞgds: ð57Þ
0 2
Selecting an arbitrary initial approximation: u0(x, t) = 1 + sin x, and using the iteration formula (57), we obtain the fol-
lowing successive approximations
et ð1 þ e2t Þ
u1 ðx; tÞ ¼ þ sin x;
2
t 2t
e ð1 þ e Þ
u2 ðx; tÞ ¼ þ sin x;
2
et ð1 þ e2t Þ ð58Þ
u3 ðx; tÞ ¼ þ sin x;
2
..
.
et ð1 þ e2t Þ
un ðx; tÞ ¼ þ sin x:
2
Then, the exact solution is given by
uðx; tÞ ¼ lim un ¼ sin x þ cosh t: ð59Þ
n!1
 A.H.A. Ali, K.R. Raslan / Chaos, Solitons and Fractals 40 (2009) 1520–1529 1527

Example 8. Consider the nonlinear partial differential equation [2]:

1 1
ut ðx; tÞ þ ux ðx; tÞ þ uðx; tÞ2 ¼ 0; where uðx; 0Þ ¼ ; uð0; tÞ ¼ : ð60Þ
2x t
Case I: We solve Eq. (60) for Lt(u), and then the equation can be written as:
o
Lt ðuÞ þ ux þ u2 ¼ 0; where Lt ðÞ ¼ ðÞ: ð61Þ
ot
Its t-direction correction functional can be constructed as follows:
Z t
unþ1 ðx; tÞ ¼ un ðx; tÞ þ kfðun ðx; sÞÞs þ ðun ðx; sÞÞx þ ðun ðx; sÞÞ2 gds; ð62Þ
0

where u0 ðx; tÞ ¼ 2x1 , n P 0. This yields the following stationary conditions

1 þ kjs¼t ; k0 ðsÞ ¼ 0: ð63Þ
Hence, the Lagrange multiplier is
k ¼ 1: ð64Þ
Substituting this value of the Lagrange multiplier into the functional (62) gives the iteration formula
Z t
unþ1 ðx; tÞ ¼ un ðx; tÞ  fðun ðx; sÞÞs þ ðun ðx; sÞÞx þ ðun ðx; sÞÞ2 gds: ð65Þ
0

We start with an initial approximation: u0 ðx; tÞ ¼ 2x1 , and using the iteration formula (65), we obtain the following suc-
cessive approximations:
1
u0 ðx; tÞ ¼ ;
2x
1 t
u1 ðx; tÞ ¼ 1þ ;
2x  2x 
1 t t2
u2 ðx; tÞ ¼ 1 þ þ 2  small term ;
2x 2x 4x
 
1 t t2 t3
u3 ðx; tÞ ¼ 1 þ þ 2 þ 3  small terms ; ð66Þ
2x 2x 4x 8x
 2

1 t t t3 t4
u4 ðx; tÞ ¼ 1þ þ 2þ 3þ  small terms ;
2x 2x 4x 8x 16x4
..
.
 
1 t t2 t3 t4 tn
un ðx; tÞ ¼ 1þ þ 2þ 3þ þ  þ  small terms
2x 2x 4x 8x 16x4 ð2xÞn
Then, the exact solution is given by
1
uðx; tÞ ¼ lim un ðx; tÞ ¼ : ð67Þ
n!1 2x  t
Case II: If we solve Eq. (60) for Lx(u):
o
Lx ðuÞ þ ut þ u2 ¼ 0; where Lx ðÞ ¼ ðÞ: ð68Þ
ot
Its x-direction correction functional can be expressed as follows:
Z x
unþ1 ðx; tÞ ¼ un ðx; tÞ þ kfðun ðs; tÞÞt þ ðun ðs; tÞÞs þ ðun ðs; tÞÞ2 gds; ð69Þ
0

where u0 ðx; tÞ ¼ 1
t
, n P 0.
The nonlinear term in the correction functional (69) is considered as a restricted variation, therefore its stationary
conditions are
1 þ kjs¼t ; k0 ðsÞ ¼ 0: ð70Þ
1528 A.H.A. Ali, K.R. Raslan / Chaos, Solitons and Fractals 40 (2009) 1520–1529

Hence, the Lagrange multiplier is

k ¼ 1: ð71Þ
Substituting this value of the Lagrange multiplier into the functional (69) gives the iteration formula
Z x
unþ1 ðx; tÞ ¼ un ðx; tÞ  fðun ðs; tÞÞt þ ðun ðs; tÞÞs þ ðun ðs; tÞÞ2 gds: ð72Þ
0

We begin with an initial approximation: u0 ðx; tÞ ¼ 1

t
, and using the iteration formula (72), we obtain the following suc-
cessive approximations
1
u0 ðx; tÞ ¼ ;
t  
1 2x
u1 ðx; tÞ ¼  1þ ;
t t
 
1 2x 4x2
u2 ðx; tÞ ¼  1 þ þ 2 þ small term ;
t t t
 
1 2x 4x2 8x3
u3 ðx; tÞ ¼  1 þ þ 2 þ 3 þ small terms ; ð73Þ
t t t t
 
1 2x 4x2 8x3 16x4
u4 ðx; tÞ ¼  1 þ þ 2 þ 3 þ 4 þ small terms ;
t t t t t
..
.
 
1 2x 4x2 8x3 ð2xÞn
un ðx; tÞ ¼  1 þ þ 2 þ 3 þ    þ n þ small terms :
t t t t t
Then, the exact solution is given by
1
uðx; tÞ ¼ lim un ðx; tÞ ¼ : ð74Þ
n!1 2x  t
Both operator equations yield distinct sequences which converge to the same solution.

3. Conclusion

The VIM has been successfully applied for solving linear and nonlinear PDEs in one, two and three dimensions with
variable coefficients. Eight examples are solved exactly. The VIM reduces the volume of calculations, while tedious cal-
culation is required in Adomian method; hence the iteration is direct and straightforward. For nonlinear equations, the
VIM facilitates the computational work and gives the solution rapidly compared with Adomian method. The reliability
of the VIM and the reduction in the size of computational domain give this method a wider applicability. Thus, we
conclude that the proposed method can be extended to solve many nonlinear PDEs with variable coefficients which
arise in physical and engineering applications.

Acknowledgements

The authors would like to thank Prof. Ji-Huan He for reading the manuscript and fruitful comments.

References

[1] Mitchell AR, Griffiths DF. The finite difference method in partial equations. John Wiley & Sons; 1980.
[2] Adomian G. Solving frontier problem of physics: the decomposition method. Boston, MA: Kluwer Academic; 1994.
[3] He JH, Wu XH. Exp-function method for nonlinear wave equations. Chaos, Solitons & Fractals 2006;30:700–8.
[4] He JH. Variational approach for nonlinear oscillators. Chaos, Solitons & Fractals 2007;34:1430–9.
[5] He JH. Variational principles for some nonlinear partial differential equations with variable coefficients. Chaos, Solitons &
Fractals 2004;19:847–51.
 A.H.A. Ali, K.R. Raslan / Chaos, Solitons and Fractals 40 (2009) 1520–1529 1529

[6] Parkes EJ, Duffy BR. An automated tanh-function method for finding solitary wave solutions to nonlinear evolution equations.
Comput Phys Commun 1998;98:288–300.
[7] Khater AH, Malfiet W, Callebaut DK, Kamel ES. The tanh method, a simple transformation and exact analytical solutions for
nonlinear reaction–diffusion equations. Chaos, Solitons & Fractals 2002;14:513–22.
[8] Evans DJ, Raslan KR. The tanh-function method for solving some important nonlinear partial differential equation. Int J
Comput Math 2005;82:897–905.
[9] Fan E. Extended tanh-function method and its applications to nonlinear equations. Phys Lett A 2000;277:212–8.
[10] Fan E. Traveling wave solutions for generalized Hirota–Satsuma coupled KdV systems. Z Naturforsch A 2001;56:312–8.
[11] Elwakil SA, El-Labany SK, Zahran MA, Sabry R. Modified extended tanh-function method for solving nonlinear partial
differential equations. Phys Lett A 2002;299:179–88.
[12] Ali AHA. The modified extended tanh-function method for solving coupled MKdV and coupled Hirota–Satsuma coupled KdV
equations. Phys Lett A 2007;363:420–5.
[13] Feng ZS. On explicit exact solutions to the compound Burgers–KdV equation. Phys Lett A 2002;293:57–66.
[14] Feng ZS. The first integral method to study the Burgers–Korteweg-de Vries equation. Phys Lett A: Math Gen 2002;35:343–9.
[15] Feng ZS. Exact solution to an approximate Sine–Gordon equation in (n + 1)-dimensional space. Phys Lett A 2002;302:64–76.
[16] Feng ZS, Wang XH. The first integral method to the two-dimensional Burgers–Korteweg-de Vries equation. Phys Lett A
2003;308:173–8.
[17] Raslan KR. The first integral method for solving some important nonlinear partial differential equations. J Nonlinear Dynam
2007. doi:10.1007/s11071-007-9262-x.
[18] He JH. Some asymptotic methods for strongly nonlinear equations. Int J Mod Phys B 2006;20:1141–99.
[19] He JH, Wu XH. Construction of solitary solution and compact on-like solution by variational iteration method. Chaos, Solitons
& Fractals 2006;29:108–13.
[20] He JH. Variational iteration method for autonomous ordinary differential systems. Appl Math Comput 2000;114:115–23.
[21] He JH. Variational iteration method – a kind of non-linear analytical technique: some examples. Int J Nonlinear Mech
1999;34:699–708.
[22] He JH. Approximate analytical solution for seepage flow with fractional derivatives in porous media. Comput Method Appl Mech
Eng 1998;167:57–68.
[23] Odibat ZM, Momani S. Application of variational iteration method to nonlinear differential equations of fractional order. Int J
Nonlinear Sci Numer Simul 2006;7:27–34.
[24] Bildik N, Konuralp A. The use of variational iteration method, differential transform method and adomian decomposition
method for solving different types of nonlinear partial differential equations. Int J Nonlinear Sci Numer Simul 2006;7:65–70.
[25] Yusufoglu E. Variational iteration method for construction of some compact and noncompact structures of Klein–Gordon
equations. Int J Nonlinear Sci Numer Simul 2007;8(2):152–8.
[26] Sweilam NH, Khader MM. Variational iteration method for one-dimensional nonlinear thermoelasticity. Chaos, Solitons &
Fractals 2007;32:145–9.
[27] Tari H, Ganji DD, Rostamian M. Approximate solutions of K(2,2), KdV and modified KdV equations by variational iteration
method, homotopy perturbation method and homotopy analysis method. Int J Nonlinear Sci Numer Simul 2007;8(2):203–10.
[28] Abdou MA, Soliman AA. Variational iteration method for solving Burgers’ and coupled Burgers’ equation. J Comput Appl Math
2005;181:245–51.
[29] Soliman AA, Abdou MA. Numerical solutions of nonlinear evolution equations using variational iteration method. J Comput
Appl Math 2007;207(1):111–20.
[30] Soliman AA. Numerical simulation of the generalized regularized long wave equation by He’s variational iteration method. Math
Comput Simul 2005;70:119–24.
[31] Abdou MA, Soliman AA. New applications of variational iteration method. Physica D 2005;211:1–8.
[32] Wazwaz A-M. A comparison between the variational iteration method and adomian decomposition method. J Comput Appl
Math 2006. doi:10.1016/[Link].2006.07.018.
[33] Wazwaz A-M. The variational iteration method for rational solutions for KdV, K(2,2), Burgers, and cubic Boussinesq equations.
J Comput Appl Math 2006. doi:10.1016/[Link].2006.07.010.
[34] Wazwaz A-M, Gorguis A. Exact solutions for heat-like and wave-like equations with variable coefficients. Appl Math Comput
2004;149:15–29.