1 July 2002

Physics Letters A 299 (2002) 179188

www.elsevier.com/locate/pla
Modied extended tanh-function method for solving nonlinear
partial differential equations
S.A. Elwakil
a
, S.K. El-labany
b
, M.A. Zahran
a,
, R. Sabry
b
a
Theoretical Physics Group, Physics Department, Faculty of Science, Mansoura University, Mansoura, Egypt
b
Theoretical Physics Group, Physics Department, Faculty of Science, Mansoura University, New Damietta 34517,
Damietta, Egypt
Received 20 February 2002; received in revised form 2 May 2002; accepted 7 May 2002
Communicated by A.R. Bishop
Abstract
Based on an extended tanh-function method, a general method is suggested to obtain multiple travelling wave solutions for
nonlinear partial differential equations (PDEs). The validity and reliability of the method is tested by its application to some
nonlinear PDEs. The obtained results are compared with that of an extended tanh-function method and hyperbolic-function
method. New exact solutions are found. 2002 Elsevier Science B.V. All rights reserved.
Keywords: Nonlinear evolution equation; Traveling wave solution; Symbolic computation
1. Introduction
Nonlinear PDEs are widely used to describe complex phenomena in various elds of sciences, especially in
physics. Various methods for obtaining explicit travelling solitary wave solutions to nonlinear evolution equations
have been proposed. Among these are Hirotas bilinear methods [1], the inverse scattering transform [1], Painlev
expansions [2], the real exponential method [3], homogeneous balance method [4,5], rank analytical method [6]
and several ansatz methods (see, for example, Ref. [7]).
Recently, tanh-function method [813] has been proposed to construct exact solutions to PDEs. Later, Fan
[14,15] has proposed an extended tanh-function method and obtained new travelling wave solutions that cannot be
obtained by tanh-function method [813].
Now, let us simply describe the modied extended tanh-function. Consider a given PDE, say in two variables
(1) H(u, u
x
, u
t
, u
xx
, . . .) =0.
*
Corresponding author.
E-mail addresses: [email protected] (M.A. Zahran), [email protected] (R. Sabry).
0375-9601/02/$ see front matter 2002 Elsevier Science B.V. All rights reserved.
PII: S0375- 9601( 02) 00669- 2
180 S.A. Elwakil et al. / Physics Letters A 299 (2002) 179188
We rst consider its travelling solutions u(x, t ) =u( ), =x +t , then Eq. (1) becomes an ordinary differential
equation. The next crucial step is that the solution we are looking for is expressed in the form
(2) u( ) =
m

i=0
a
i

i
+
m

i=0
b
i

i
and
(3)

=b +
2
,
where b is a parameter to be determined, = ( ),

= d/d . The parameter m can be found by balancing

the highest-order linear term with the nonlinear terms [9,14]. Substituting (2) and (3) into the relevant ordinary
differential equation will yield a system of algebraic equations with respect to a
i
, b
i
, b, (where i = 0, . . . , m)
because all the coefcients of
i
have to vanish. With the aid of Mathematica, one can determine a
i
, b
i
, b and .
The Riccati equation (3) has the general solutions
(4) =
_

b tanh

b, with b < 0,

b coth

b, with b < 0,
(5) =
1

, with b =0,
and
(6) =
_

b tan

b, with b > 0,

b cot

b, with b > 0.
Since coth- and cot-type solutions appear in pairs with tanh- and tan-type solutions, respectively, they are
omitted in this Letter.
In the next section, we study some nonlinear equations to illustrate this method.
2. Examples
Example 1. Consider the nonlinear evolution equation
(7) u
t t
+u
xx
+u + u
3
=0,
where , , and are constants. Eq. (7) contains some particular important equations such as Dufng, Klein
Gordon, LandauGinsburgHiggs, and
4
equation [16]. To look for the travelling wave solution of Eq. (7), we
use the transformation u(x, t ) =u( ), =x +t . Then Eq. (7) is reduced to the following ordinary differential
equation:
(8)
_

2
+
_
u

+u + u
3
=0.
Balancing u

with u
3
yields m=1. Therefore, we have
(9) u =a
0
+a
1
+b
0
+b
1

1
.
Substituting Eq. (9) into Eq. (8) and making use of Eq. (3), with the help of Mathematica we get a system of
algebraic equations for a
0
, a
1
, b
0
, b
1
, b, and :
(a
0
+b
0
)
_
+
_
(a
0
+b
0
)
2
+6a
1
b
1
__
=0,
2ba
1
+a
1
+2b
2
a
1
+3 a
2
0
a
1
+6 a
0
a
1
b
0
+3 a
1
b
2
0
+3 a
2
1
b
1
=0,
S.A. Elwakil et al. / Physics Letters A 299 (2002) 179188 181
a
0
a
2
1
+ a
2
1
b
0
=0,
2a
1
+2
2
a
1
+ a
3
1
=0,
2bb
1
+b
1
+2b
2
b
1
+3 a
2
0
b
1
+6 a
0
b
0
b
1
+3 b
2
0
b
1
+3 a
1
b
2
1
=0,
a
0
b
2
1
+ b
0
b
2
1
=0,
2b
2
b
1
+2b
2

2
b
1
+ b
3
1
=0,
from which we nd
(10) b =

2( +
2
)
, a
0
=b
0
, a
1
=0, b
1
=b
_
2( +
2
)

,
(11) b =

2( +
2
)
, a
0
=b
0
, a
1
=
_
2( +
2
)

, b
1
=0,
(12) b =

8( +
2
)
, a
0
=b
0
, a
1
=
_
2( +
2
)

, b
1
=ba,
(13) b =

4( +
2
)
, a
0
=b
0
, a
1
=
_
2( +
2
)

, b
1
=ba
1
.
According to Eq. (10), it is clear that ( +
2
)/ < 0 for b
1
to be real. Then, for b > 0 the solution to Eq. (7)
reads
(14) u(x, t ) =
_

cot
__

2( +
2
)
(x +t )
_
,
where ( +
2
)/ < 0 . While for b < 0 the solution to Eq. (7) is
(15) u(x, t ) =
_

coth
__

2( +
2
)
(x +t )
_
,
where ( +
2
)/ < 0.
Due to Eq. (11), it is clear that ( +
2
)/ < 0 for a
1
to be real. Then, for b > 0 the solution to Eq. (7) reads
(16) u(x, t ) =
_

tan
_
_

2( +
2
)
(x +t )
_
,
where ( +
2
)/ < 0. While for b < 0 it is
(17) u(x, t ) =
_

tanh
_
_

2( +
2
)
(x +t )
_
,
where ( +
2
)/ < 0.
From Eq. (12), it is clear that ( +
2
) < 0 for a
1
and b
1
to be real. Then, for b > 0 we get
(18) u(x, t ) =
_

4
_
tan
__

8( +
2
)
(x +t )
_
cot
__

8( +
2
)
(x +t )
__
,
182 S.A. Elwakil et al. / Physics Letters A 299 (2002) 179188
where ( +
2
) < 0. Also, the solution of Eq. (7) due to Eq. (12) for the case b < 0 is
(19) u(x, t ) =
_

4
_
tanh
_
_

8( +
2
)
(x +t )
_
+coth
_
_

8( +
2
)
(x +t )
__
,
where ( +
2
) < 0.
Eq. (13) indicates that ( +
2
) < 0 for a
1
and b
1
to be real. Then, for b < 0,
(20) u(x, t ) =
_

2
csch
__

4( +
2
)
(x +t )
_
sech
__

4( +
2
)
(x +t )
_
,
while for b > 0,
(21) u(x, t ) =
_

csc
_
_

( +
2
)
(x +t )
_
.
We have recovered some new exact solutions that are not obtained by the hyperbolic-function method [16].
Example 2. Consider the two-dimensional KdVBurgers equation
(22) (u
t
+uu
x
+pu
xxx
qu
xx
)
x
+ru
yy
=0,
where p, q are real constants and r =1 [13,14]. On using u(x, t ) =u( ), =x +dy +ct , Eq. (22) reduces to
(23) (cu

+uu

+pu

qu

+rd
2
u

=0.
Balancing u

with uu

gives m=2, hence

(24) u =a
0
+a
1
+a
2

2
+b
0
+b
1

1
+b
2

2
.
Substitute Eq. (24) into Eq. (23) and make use of Eq. (3). With the help of Mathematica, we get a system of
algebraic equations, for a
0
, a
1
, a
2
, b
0
, b
1
, b
2
, b, c, and d:
2b
2
qa
1
+b
2
a
2
1
+2b
2
a
2
_
c +8bp +d
2
r +a
0
+b
0
_
+b
1
(2bq +b
1
) +2b
2
_
c +8bp +d
2
r +a
0
+b
0
_
=0,
bca
1
+8b
2
pa
1
+bd
2
ra
1
+ba
0
a
1
8b
2
qa
2
+3b
2
a
1
a
2
+ba
1
b
0
+ba
2
b
1
=0,
8bqa
1
+4ba
2
1
+8bca
2
+136b
2
pa
2
+8bd
2
ra
2
+8ba
0
a
2
+6b
2
a
2
2
+8ba
2
b
0
=0,
2ca
1
+40bpa
1
+2d
2
ra
1
+2a
0
a
1
40b
2
qa
2
+18ba
1
a
2
+2a
1
b
0
+2a
2
b
1
=0,
6qa
1
+3a
2
1
+6ca
2
+240bpa
2
+6d
2
ra
2
+6a
0
a
2
+16ba
2
2
+6a
2
b
0
=0,
2pa
1
2qa
2
+a
1
a
2
=0,
12pa
2
+a
2
2
=0,
bcb
1
+8b
2
pb
1
+bd
2
rb
1
+ba
0
b
1
+bb
0
b
1
+8bqb
2
+ba
1
b
2
+3b
1
b
2
=0,
8b
2
qb
1
+4bb
2
1
+8bcb
2
+136b
2
pb
2
+8bd
2
rb
2
+8ba
0
b
2
+8bb
0
b
2
+6b
2
2
=0,
2b
2
cb
1
+40b
3
pb
3
1
+2b
2
d
2
rb
1
+2b
2
a
0
b
1
+2b
2
b
0
b
1
+40b
2
qb
2
+2b
2
a
1
b
2
+18bb
1
b
2
=0,
6b
3
qb
1
+3b
2
b
2
1
+6b
2
cb
2
+240b
3
pb
2
+6b
2
d
2
rb
2
+6b
2
a
0
b
2
+6b
2
b
0
b
2
+16bb
2
2
=0,
2b
4
pb
1
+2b
3
qb
2
+b
2
b
1
b
2
=0,
12b
4
pb
2
+b
2
b
2
2
=0.
S.A. Elwakil et al. / Physics Letters A 299 (2002) 179188 183
Solving the above system using Mathematica, we get
a
0
=
_
c +8bp
q
2
25p
+d
2
r +b
0
_
, a
1
=
12q
5
, a
2
=12p,
(25) b
1
=0, b
2
=0, b =
q
2
100p
2
,
a
0
=
_
c +8bp
q
2
25p
+d
2
r +b
0
_
, a
1
=0, a
2
=0,
(26) b
1
=
12bq
5
, b
2
=12b
2
p, b =
q
2
100p
2
, where b =0,
a
0
=
_
c +8bp
q
2
25p
+d
2
r +b
0
_
, a
1
=
12q
5
, a
2
=12p,
(27) b
1
=
12bq
5
, b
2
=12b
2
p, b =
q
2
400p
2
, where b =0.
Due to Eq. (25), it is clear that b 0 and we have
(28) u(x, t ) =c d
2
r
12p

2
, b =0 and q =0,
(29) u(x, t ) =
3q
2
25p

_
c +d
2
r
_

6q
2
25p
tanh
_
q
10p
_

3q
2
25p
tanh
2
_
q
10p
_
, =x +dy +ct .
In Eq. (26) b < 0, thus we get
(30) u(x, t ) =
3q
2
25p

_
c +d
2
r
_

6q
2
25p
coth
_
q
10p
_

3q
2
25p
coth
2
_
q
10p
_
, =x +dy +ct .
Finally, as a result of Eq. (27), one gets
(31) u(x, t ) =
_
c +d
2
r
_

3q
2
25p
csch
2
_
q
10p
__
1 +sinh
_
q
5p
__
, b < 0, =x +dy +ct .
In fact, the solution obtained by Fan [14] which is
u
1
=c +
q
25p
d
2
r
12q
5z

12p
z
2
, z =x +dy +ct ,
is not correct and holds only under the condition that b and q must be equal to zero. The correct solution is
given above by Eq. (28). Solution in (28) is a rational-type solution. Solution (29) is identical to that obtained
by Fan [14]. As shown in [14], since c and d are left arbitrary, we could make the transformation d (5p/q)d,
c +d
2
r c, then solution (29) is exactly the same with that in [17]. Solution (30) is guaranteed by replacing
the tanh-type solution of the Riccati equation by coth-type in (29). Solution (31) is a new one. We have recovered
all the solutions that obtained by an extended tanh-function method and new solutions are obtained.
Example 3. Consider the system
(32) u
t
+
x
+uu
x
+pu
xxt
=0,
184 S.A. Elwakil et al. / Physics Letters A 299 (2002) 179188
(33)
t
+(u)
x
+qu
xxx
=0,
which is the variant Boussinesq equation [14,18]. Using u(x, t ) = U( ), (x, t ) = V ( ), = x + ct , Eqs. (32)
and (33) become
(34) cU

+V

+UU

+cpU

=0,
(35) cV

+(UV )

+qU

=0.
Balancing U

with UU

, and U

with (UV)

leads to the following ansatz:

(36) U =
2

i=0
a
i

i
+
2

i=0
b
i

i
, V =
2

i=0

i
+
2

i=0

i
,
where satises Eq. (3). Substituting Eq. (36) into Eqs. (34) and (35), and making use of Eq. (3) gives
ba
2
1
+2bca
2
+16b
2
cpa
2
+2ba
0
a
2
+2ba
2
b
0
+2b
2
=0,
ca
1
+8bcpa
1
+a
0
a
1
+3ba
1
a
2
+a
1
b
0
+a
2
b
1
+
1
=0,
a
2
1
+2ca
2
+40bcpa
2
+2a
0
a
2
+2ba
2
2
+2a
2
b
0
+2
2
=0,
2cpa
1
+a
1
a
2
=0,
12cpa
2
+a
2
2
=0,
b
2
1
+2cb
2
+16bcpb
2
+2a
0
b
2
+2b
0
b
2
+2
2
=0,
bcb
1
+8b
2
cpb
1
+ba
0
b
1
+bb
0
b
1
+ba
1
b
2
+3b
1
b
2
+b
1
=0,
bb
2
1
+2bcb
2
+40b
2
cpb
2
+2ba
0
b
2
+2bb
0
b
2
+2b
2
2
+2b
2
=0,
2b
3
cpb
1
+bb
1
b
2
=0,
12b
3
cpb
2
+bb
2
2
=0,
(c +2bcp +a
0
ba
2
+b
0
)b
1
+b
1

1
+a
1
_
b(c +2bcp +a
0
+b
0
) b
2
_
=0,
8b
2
qa
2
+ba
2

0
+ba
1

1
+bc
2
+ba
0

2
+bb
0

2
+ba
2

0
=0,
8bqa
1
+a
1

0
+c
1
+a
0

1
+3ba
2

1
+b
0

1
+3ba
1

2
+b
1

2
+a
1

0
+a
2

1
=0,
20bqa
2
+a
2

0
+a
1

1
+c
2
+a
0

2
+2ba
2

2
+b
0

2
+a
2

0
=0,
2qa
1
+a
2

1
+a
1

2
=0,
6qa
2
+a
2

2
=0,
8bqb
2
+b
2

0
+b
2

0
+b
1

1
+c
2
+a
0

2
+b
0

2
=0,
8b
2
qb
1
+bb
1

0
+bb
2

1
+bb
1

0
+bc
1
+ba
0

1
+bb
0

1
+3b
2

1
+ba
1

2
+3b
1

2
=0,
20b
2
qb
2
+bb
2

0
+bb
2

0
+bb
1

1
+bc
2
+ba
0

2
+bb
0

2
+2b
2

2
=0,
2b
3
qb
1
+bb
2

1
+bb
1

2
=0,
6b
3
qb
2
+bb
2

2
=0,
_
b(c +a
0
+b
0
) b
2
_

1
(2bp +
0
b
2
+
0
)b
1
(c +a
0
ba
2
+b
0
)
1
+
_
b(2bp +
0
+
0
)
2
_
a
1
=0,
which with the help of Mathematica yields three different cases.
S.A. Elwakil et al. / Physics Letters A 299 (2002) 179188 185
Case 1.
a
0
=c 8bcp
q
2cp
b
0
, a
1
=0, a
2
=12cp,
b
1
=0, b
2
=0,
0
=4bq +
q
2
4c
2
p
2

0
,
1
=0,
2
=6q,
(37)
1
=0,
2
=0,
where c =0 and b are left arbitrary.
Case 2.
a
0
=c 8bcp
q
2cp
b
0
, a
1
=0, a
2
=0,
b
1
=0, b
2
=12b
2
cp,
0
=4bq +
q
2
4c
2
p
2

0
,
1
=0,
2
=0,
(38)
1
=0,
2
=6b
2
q,
where b =0 and c =0.
Case 3.
a
0
=c 8bcp
q
2cp
b
0
, a
1
=0, a
2
=12cp,
b
1
=0, b
2
=12b
2
cp,
0
=4bq +
q
2
4c
2
p
2

0
,
1
=0,
2
=6q,
(39)
1
=0,
2
=6b
2
q,
where b =0 and c =0.
According to case 1, we have three different types of travelling wave solutions for u and :
Type 1: for b =0,
(40) u =c
q
2cp

12cp
(x +ct )
2
, =
q
2
4c
2
p
2

6q
(x +ct )
2
.
Type 2: for b < 0,
u =c 8bcp
q
2cp
+12bcptanh
2
_
b(x +ct )
_
,
(41) =4bq +
q
2
4c
2
p
2
+6bq tanh
2
_
b(x +ct )
_
.
Type 3: for b > 0,
u =c 8bcp
q
2cp
12bcptan
2
_
b(x +ct )
_
,
(42) =4bq +
q
2
4c
2
p
2
6bq tan
2
_
b(x +ct )
_
.
The solutions for u and in Eqs. (40)(42) are identical to those obtained by Fan [14].
The solutions obtained from case 2 can be obtained directly from Eqs. (37) and (38) by replacing tanh- and
tan-functions by coth- and cot-functions, respectively, so there is no need to list them here.
Finally, case 3 admits the following two types:
186 S.A. Elwakil et al. / Physics Letters A 299 (2002) 179188
Type 1: for b < 0,
u =c 8bcp
q
2cp
+12bcpcoth
2
_
b(x +ct )
_
+12bcptanh
2
_
b(x +ct )
_
,
(43) =4bq +
q
2
4c
2
p
2
+6bq coth
2
_
b(x +ct )
_
+6bq tanh
2
_
b(x +ct )
_
.
Type 2: for b > 0,
u =c 8bcp
q
2cp
12bcpcot
2
_
b(x +ct )
_
12bcptan
2
_
b(x +ct )
_
,
(44) =4bq +
q
2
4c
2
p
2
6bq cot
2
_
b(x +ct )
_
6bq tan
2
_
b(x +ct )
_
.
The solutions for u and in Eqs. (43) and (44) cannot be obtained by extended tanh-function method [14].
Example 4. Consider the nonlinear equation
(45) u
t

_
u
2
_
xx
pu +qu
2
=0,
where p, q =0 are constants. Eq. (45) is known as the nonlinear reactiondiffusion equation [8,14]. Applying the
transformation u(x, t ) =U( ), =x +ct to Eq. (45) will result in
(46) cU

_
U
2
_

pU +qU
2
=0.
Balancing U

with (U
2
)

yields m=1, which is not a positive integer. Following the idea of [8,9,14], we can
introduce the transformation U =V
1
. Thus we get
(47) pV
3
6(V

)
2
+(q cV

)V
2
+2VV

=0.
Balancing V
2
V

with V V

yields the required balancing number, m=1.Therefore, V becomes

(48) V =(a
0
+b
0
) +a
1
+b
1

1
.
After substitution of Eqs. (48) and (3) in Eq. (47), we get
4ba
0
a
1
+2qa
0
a
1
3pa
2
0
a
1
2bca
0
a
2
1
+4ba
1
b
0
+2qa
1
b
0
6pa
0
a
1
b
0
2bca
2
1
b
0
3pa
1
b
2
0
3pa
2
1
b
1
=0,
ca
2
0
a
1
8ba
2
1
+qa
2
1
3pa
0
a
2
1
bca
3
1
2ca
0
a
1
b
1
3pa
2
1
b
1
ca
1
b
2
1
+16a
1
b
1
ca
2
1
b
1
=0,
4a
0
a
1
2ca
0
a
3
1
pa
3
1
+4a
1
b
0
2ca
2
1
b
0
=0,
2a
2
1
ca
3
1
=0,
4ba
0
b
1
+2qa
0
b
1
3pa
2
0
b
1
+4bb
0
b
1
+2qb
0
b
1
6pa
0
b
0
b
1
3pb
2
0
b
1
+2ca
0
b
2
1
3pa
1
b
2
1
+2cb
0
b
2
1
=0,
bca
2
0
b
1
+16b
2
a
1
b
1
+2bca
0
b
0
b
1
+bcb
2
0
b
1
8bb
2
1
+qb
2
1
3pa
0
b
2
1
+bca
1
b
2
1
3pb
0
b
2
1
+cb
3
1
=0,
4b
2
a
0
b
1
+4b
2
b
0
b
1
+2bca
0
b
2
1
+2bcb
0
b
2
1
pb
3
1
=0,
2b
2
b
2
1
+bcb
3
1
=0,
pa
3
0
+b
2
0
(q cb
0
) +cb
2
0
b
1
6b
2
1
ba
2
1
(6b +cb
1
) +a
2
0
(q bca
1
3pb
0
+cb
1
)
+a
1
_
bcb
2
0
+2b
1
(16b +q 3pb
0
) +cb
2
1
_
+a
0
_
3pb
2
0
6pa
1
b
1
+2b
0
(q bca
1
+cb
1
)
_
=0,
which yields three different cases:
S.A. Elwakil et al. / Physics Letters A 299 (2002) 179188 187
Case 1.
(49) b =
p
2
16c
2
, c =
p

q
, a
0
=
p
2c
2
b
0
, a
1
=
2
c
, b
1
=0.
Case 2.
(50) b =
p
2
16c
2
, c =
p

q
, a
0
=
p
2c
2
b
0
, a
1
=0, b
1
=
p
2
8c
3
.
Case 3.
(51) b =
q
64
, c =
p

q
, a
0
=
p
2c
2
b
0
, a
1
=
2
c
, b
1
=
b
2c
.
According to case 1, b < 0 and we have
(52) u =
2p
q
_
1 tanh
_
q
4
_
x
p

q
t
__
_
1
.
In case 3, b < 0 and hence
(53) u =
p
q
_
1 +e
(

q/2)(x(p/

q)t )
_
.
The solutions in Eq. (52) are identical to those obtained in [14]. The solutions obtained from case 2 can be
obtained directly from Eq. (52) by replacing tanh-function by coth-function, respectively. So, there is no need
to list them. The solutions given by Eq. (53) appear to be new. It has been shown that Eq. (45) have explicit
(nonpropagating) similarity solutions that are explosively unstable, i.e., the instabilities tend to grow to innite
amplitude in a limited period of time. This typical behaviour is shown by solution (52) which is singular and blows
up for x [8]. There is much current interest in the formation of so-called hot spots or blow-up of
solutions [14,1921].
3. Conclusion
We have developed a modication of an extended tanh-function method, which is named modied extended
tanh-function method. The validity and reliability of the method is tested by applying it to some nonlinear PDEs.
We have successfully recovered the previously known solutions that had been found by an extended tanh-function
method. We also found new exact solutions that are not obtained by both an extended tanh- and a hyperbolic-
function methods. The travelling wave solutions derived in this Letter include soliton, periodical, rational and
singular solutions. Soliton and periodical solutions are physically clear. The particle system describing the motion
of pole of rational solutions of KdV and Boussinesq equation was analyzed in [14,18] which conrms the
applicability of our obtained rational solutions.
Acknowledgement
We would like to express our sincere thanks to referees for helpful advice and suggestions.
188 S.A. Elwakil et al. / Physics Letters A 299 (2002) 179188
References
[1] P.G. Drazin, R.S. Johnson, Solitons: An Introduction, Cambridge University Press, Cambridge, 1989.
[2] F. Cariello, M. Tabor, Physica D 39 (1989) 77.
[3] W. Hereman, M. Takaoka, J. Phys. A 23 (1990) 4805.
[4] M. Wang, Phys. Lett. A 213 (1996) 279.
[5] E. Fan, Phys. Lett. A 265 (2000) 353.
[6] X. Feng, Int. J. Theor. Phys. 39 (1) (2000) 207.
[7] B.Q. Lu, B.Z. Xiu, Z.L. Pang, X.F. Jiang, Phys. Lett. A 175 (1993) 113.
[8] W. Maliet, Am. J. Phys. 60 (1992) 650.
[9] E.J. Parkes, B.R. Duffy, Comput. Phys. Commun. 98 (1996) 288.
[10] B.R. Duffy, E.J. Parkes, Phys. Lett. A 214 (1996) 271.
[11] E.J. Parkes, B.R. Duffy, Phys. Lett. A 229 (1997) 217.
[12] Y.T. Gao, B. Tian, Comput. Math. Appl. 33 (1997) 115.
[13] V.V. Gudkov, J. Math. Phys. 38 (1997) 4794.
[14] E. Fan, Phys. Lett. A 277 (2000) 212.
[15] E. Fan, Z. Naturforsch. 56A (2001) 312.
[16] C. Bai, Phys. Lett. A 288 (2001) 191.
[17] E.J. Parkes, J. Phys. A: Math. Gen. 27 (1994) L497.
[18] R.L. Sachs, Physica D 30 (1988) 1.
[19] N.A. Kudryashov, D. Zargaryan, J. Phys. A: Math. Gen. 29 (1996) 8067.
[20] N.F. Smyth, J. Austral. Math. Soc. Ser. B 33 (1992) 403.
[21] P.A. Clarkson, E.L. Manseld, Physica D 70 (1993) 250.