Numerical simulation for general Rosenau-RLW equation: an average linearized conservative scheme

Xintian Pan1 Luming Zhang2 ,

1

School of Mathematics and Information Science, Weifang University, Weifang, Shandong, 261061, P.R. China.

Department of Mathematics, Nanjing University of Aeronautics and Astronautics, Nanjing, Jiangsu, 210016, P.R.China.

Abstract In this paper, numerical solutions for the general Rosenau-RLW equation are considered and an energy conservative linearized nite dierence scheme is proposed. Existence of the solutions for the dierence scheme has been shown. Stability, convergence and priori error estimate of the scheme are proved using energy method. Numerical results demonstrate that the scheme is ecient and reliable.

Keywords: General Rosenau-RLW equation; Conservative scheme; Solvability; Convergence.

Introduction
ut + uxxxxt uxxt + ux + (up )x = 0, x [xl , xr ], t [0, T ], (1.1)

In this paper, we examine the use of the nite dierence method for the general Rosenau-RLW equation

with an initial condition u(x, 0) = u0 (x), and boundary conditions u(xl , t) = u(xr , t) = 0, uxx (xl , t) = uxx (xr , t) = 0, t [0, T ]. (1.3) x [xl , xr ], (1.2)

where p 2 is a integer and u0 (x) is a known smooth function. When p = 2, the equation (1.1) is called usual Rosenau-RLW equation.When p = 3, (1.1) is called modied Rosenau-RLW equation. It can be proved easily that the problem (1.1)-(1.3) possesses the following conservative laws:
xr xr

Q(t) =
xl

u(x, t) dx =
xl

u0 (x, t) dx = Q(0),

(1.4) (1.5)

E(t) = ||u||2 2 + ||ux ||2 2 + ||uxx ||2 2 = E(0). L L L

Corresponding

author. E-mail address: panxintian@yahoo.com.cn(X. Pan).

As already pointed out by Zhang and Vzquez [2], the non-conservative dierence schemes may a easily show nonlinear blow-up, and the conservative dierence schemes perform better than the nonconservative ones. In [35,717], some conservative nite dierence schemes were used for Sine-Gordon equation, Cahn-Hilliard equation, Klein-Gordon equation, a system of Schrdinger equation, Zakharov o equations, Rosenau equation, GRLW equation, Klein-Gordon-Schrdinger equation, respectively. Nuo merical results of all the schemes are very good. As far as computational studies are concerned, Zuo et al. [1] have proposed a Crank-Nicolson difference scheme for the Rosenau-RLW equation. The dierence scheme in [1] is nonlinear implicit, so it requires heavy iterative calculations and is not suitable for parallel computation. In a recent work [16], we have made some preliminary computation by proposing a conservative linearized dierence scheme for GRLW equation which is unconditionally stable and reduces the computational work, and the numerical results are encouraging. In this paper, we continue our work and propose a conservative linearized dierence scheme for the general Rosenau-RLW equation which is unconditionally stable and second-order convergent and simulates conservative laws (1.4)-(1.5) at the same time. The remainder of this paper is organized as follows. In section 2, an energy conservative linearized dierence scheme for the general Rosenau-RLW equation is described and the discrete conservative laws of the dierence scheme are discussed. In section 3, we show that the scheme is uniquely solvable. In section 4, convergence and stability of the scheme are proved. In section 5, numerical experiments are reported.

An average linearized conservative scheme and its discrete conservative law

In this section, we describe a new conservative dierence scheme for the problems of (1.1)-(1.3). Let h and be the uniform step size in the spatial and temporal direction, respectively. Denote xj = jh (0 0 j J), tn = n (0 n N ), un u(xj , tn ) and Zh = {u = (uj )|u0 = uj = 0, j = 0, 1, 2, , J}. j Dene un un j+1 j , h un+1 un j j (un )t = , j (un )x = j
J1

un un un un j j1 j+1 j1 , (un )x = , j h 2h un+1 un1 un+1 + un1 j j j j (un )t = , un = j , j 2 2 (un )x = j ||un ||2 = (un , un ), ||un || =
1jJ1

(un , v n ) = h
j=1

n u n vj , j

max

|un |, j

and in the paper, C denotes a general positive constant which may have dierent values in dierent occurrences. p Notice that (up )x = [up1 ux + (up )x ]. We consider the following three-level average linearized p+1 conservative scheme for the IBV problems (1.1)-(1.3).

(un )t + (un )xxxt (un )xt + j j j x x +

p {(un )p1 (n )x + [(un )p1 (n )]x } = 0, uj uj j j p+1 un J = 0, 1 j J, (un )x 0 x = (un )x J x = 0,

1 n+1 (uj + un1 )x + (un )x j j 2 1 j J 1, 1 n N 1, (2.1) (2.2) 0 n N, (2.3)

u0 = u0 (xj ), j un 0 =

where 0 1 is a real constant. The scheme (2.1)-(2.3) is three-level and linear-implicit, so it can be easily implemented. It should be pointed out that we need another suitable two-level scheme (such as C-N scheme) to compute u1 . For convenience, the last term of (2.1) is dened by (un , un ) = p {(un )p1 (n )x + [(un )p1 (j )]x }. uj un j j p+1 (2.4)

0 Lemma 1 [6]. For any two mesh functions: u, v Zh , we have

((uj )x , vj ) = (uj , (vj )x ), and

(vj , (uj )x ) = ((vj )x , (uj )x ), x

(uj , (uj )x ) = ((uj )x , (uj )x ) = ||ux ||2 . x Furthermore, if (un )x = (un )x = 0, then 0 x J x (uj , (uj )xxx ) = ||uxx ||2 . x
2 Theorem 1. Suppose u0 H0 [xl , xr ] and u(x, t) C 5,3 . Then the scheme (2.1)-(2.3) admits the following invariant

Qn =
n

h 2

J1 j=1

(un+1 + un ) = Qn1 = = Q0 . j j
J1

(2.5) (un )x un+1 j j

j=1

1 1 1 E = (||un+1 ||2 + ||un ||2 ) + (||un+1 ||2 + ||un ||2 ) + (||un+1 ||2 + ||un ||2 ) + h xx xx x x 2 2 2 =E
n1

Proof. Multiplying (2.1) with h, according to the boundary conditions (2.3), then summing up for j from 1 to J 1, we obtain J1 h (un+1 un1 ) = 0. (2.7) j 2 j=1 j Let Qn = h 2
J1

= = E .

(2.6)

(un+1 + un ). j j
j=1

(2.8)

Then we obtain (2.5) from (2.7). Taking the inner product of (2.1) with 2n , according to Lemma 2.1, we have u 1 1 1 (||un+1 ||2 ||un1 ||2 ) + (||un+1 ||2 ||un1 ||2 ) + (||un+1 ||2 ||un1 ||2 ) + h (un )x un+1 xx xx x x j j 2 2 2 j=1
J1 J1

(un1 )x un + ((un , un ), 2n ) = 0. uj j j
j=1

(2.9)

Now, computing the last term of the left-hand side in (2.9), we have ((un , un ), 2n ) = uj = 2p h {(un )p1 (n )x + [(un )p1 un ]x }n uj j uj j j p + 1 j=1 p p+1 p p+1 = 0. Substitute (2.10) into (2.9), and we let En = 1 1 1 (||un+1 ||2 + ||un ||2 ) + (||un+1 ||2 + ||un ||2 ) + (||un+1 ||2 + ||un ||2 ) + h xx xx x x 2 2 2
J1 J1 j=1 J1 j=1 J1

[(un )p1 (n un ) + (un )p1 un (un )p1 un ]n uj+1 j1 j+1 j1 uj j j+1 j1 [(un )p1 un un (un )p1 un un ] j+1 j j+1 j j j+1 [(un )p1 un un (un )p1 un un ] j j1 j j1 j1 j (2.10)

p p+1

J1 j=1

(un )x un+1 . j j
j=1

By the denition of E n , (2.6) holds.

Solvability
Theorem 2. The dierence scheme (2.1) is uniquely solvable.

In this section, we shall prove the solvability of the dierence scheme (2.1).

Proof. By the mathematical induction. It is obvious that u0 is uniquely determined by (2.2). We can choose a second order method to compute u1 (such as C-N scheme [1]). Assuming that u1 , , un are uniquely solvable, consider un+1 in (2.1) which satises 1 n 1 n 1 n 1 n+1 u + (u )xxx (u )x + (uj )x x x 2 j 2 j 2 j 2 p + {(un )p1 (un+1 )x + [(un )p1 un+1 ]x } = 0. j j j j 2(p + 1) Taking the inner product of (3.1) with un+1 , we obtain 1 1 1 ||un+1 ||2 + ||un+1 ||2 + ||un+1 ||2 + ((un , un+1 ), un+1 ) = 0, 2 2 xx 2 x 4 (3.2)

(3.1)

where (un , un+1 ) = Notice that ((u , u

n n+1

p n p1 n+1 (uj )x 2(p+1) (uj ) J1 j=1 J1 j=1

+ [(un )p1 un+1 ]x . j j

), u

n+1

ph )= 2(p + 1) = p 4(p + 1)

{(un )p1 (un+1 )x + [(un )p1 un+1 ]x }un+1 . j j j j j [(un )p1 (un+1 un+1 ) + (un )p1 un+1 (un )p1 un+1 ]un+1 j j+1 j1 j+1 j1 j+1 j1 j (3.3)

= 0. It follows from (3.2) that 1 1 1 ||un+1 ||2 + ||un+1 ||2 + ||un+1 ||2 = 0. 2 2 xx 2 x

That is, there uniquely exists trivial solution satisng equation (3.1). Hence, un+1 in (2.1) is uniquely j solvable. This completes the proof of Theorem 2. Remark 1. All results above in this paper are correct for IBV problem of the general RosenauRLW equation with nite or innite boundary.

Convergence and stability of nite dierence scheme

n First we shall consider the truncation error of the dierence scheme of (2.1)-(2.3). Denote vj = u(xj , tn ). We dene the truncation error as follows 1 n+1 n1 n n n n n Erj = (vj )t + (vj )xxxt (vj )xt + (vj + vj )x + (vj )x x x 2 p n + {(v n )p1 (j )x + [(vj )p1 (j )]x }. vn vn (4.1) p+1 j n Using Taylor expansion, we obtain that Erj = O( 2 + h2 ) holds if , h 0. This is that

Lemma 2. Assume u(x, t) is smooth enough, then the local truncation error of dierence scheme (2.1)-(2.3) is O( 2 + h2 ). Next, we shall discuss the convergence and stability of nite dierence scheme (2.1)-(2.3). The following two lemmas are introduced. Lemma 3(Discrete Sobolevs inequality [18]). There exist two constants C1 and C2 such that ||un || C1 ||un || + C2 ||un ||. x Lemma 4(Discrete Gronwall inequality [18]). Suppose w(k), (k) are nonnegative mesh functions and (k) is nondecreasing. If C > 0 and
k1

w(k) (k) + C

w(l)),
l=0

k, 5

then w(k) (k)eC k , k.

2 Lemma 5. Suppose u0 H0 [xl , xr ], then the solution un of (2.1) satises: ||un || C, ||un || C, x which yield ||un || C (n = 1, 2, , N ).

Proof. It follows from (2.6) that 1 1 1 (||un+1 ||2 + ||un ||2 ) + (||un+1 ||2 + ||un ||2 ) + (||un+1 ||2 + ||un ||2 ) xx xx x x 2 2 2
J1

= C h Thus

1 (un )x un+1 C + (||un+1 ||2 + ||un ||2 ). j j x 2 j=1

(4.2)

1 1 1 [(1 )||un+1 ||2 + ||un ||2 ] + (||un+1 ||2 + ||un ||2 ) + [(1 )||un ||2 + ||un+1 ||2 ] C. (4.3) xx xx x x 2 2 2 This implies for small which satises 1 > 0, we get ||un || C, ||un || C. x

Using Lemma 3, we obtain ||un || C. Remark 2. Lemma 5 implies that scheme (2.1)-(2.3) is unconditionally stable.
2 Theorem 3. Assume that u0 H0 [xl , xr ] and u(x, t) C 5,3 . Then the solution un of the scheme (2.1)-(2.3) converges to the solution of problem (1.1)-(1.3) and the rate of convergence is O( 2 + h2 ) by the || || norm. n Proof. Subtracting (4.1) from (2.1) and letting en = vj un , we have j j n Erj = (en )t + (en )xxxt (en )xt + (en )x + j j j j x x

1 n+1 (ej + en1 )x + (en )x j j 2 p p n + {(v n )p1 (j )x + [(vj )p1 vj ]x } vn n {(un )p1 (n )x + [(un )p1 un ]x }. (4.4) uj j j j p+1 j p+1

Taking the inner product in (4.4) with 2n , we obtain e

n (Erj , 2n ) = e

1 1 1 (||en+1 ||2 ||en1 ||2 ) + (||en+1 ||2 ||en1 ||2 ) + (||en+1 ||2 ||en1 ||2 ) xx xx x x 2 2 2
J1

+h
j=1

(en )x (en+1 + en1 ) + (I + II, 2n ), e j j j

(4.5)

where p [(v n )p1 (j )x (un )p1 (n )x ], vn uj j p+1 j p n II = {[(vj )p1 (j )]x [(un )p1 un ]x }. vn j j p+1 I= According to lemma 5, the fth term of right-hand side 0f (4.5) is estimated as follows: (I, 2n ) = e 2p h [(v n )p1 (j )x (un )p1 (n )x ]n vn uj e j p + 1 j=1 j 2p 2p n h (vj )p1 (n )x en + ej h [(v n )p1 (un )p1 ](n )x en uj j p + 1 j=1 p + 1 j=1 j 2p 2p h (v n )p1 ((n )x en + ej h [en p + 1 j=1 j p + 1 j=1 j C(||n ||2 + ex C(||en+1 ||2 x ||e || + || || ) e
n 2 n 2 J1 J1 p2 n (vj )p2k (un )k ](n )x en uj j k=0 J1 J1 J1

+ ||en1 ||2 + ||en+1 ||2 + ||en ||2 + ||en1 ||2 ), x

(4.6)

and similarly we can prove (II, 2n ) C(||en+1 ||2 + ||en1 ||2 + ||en+1 ||2 + ||en ||2 + ||en1 ||2 ). e x x In addition, it is obvious that 1 n (Erj , 2n ) ||Ern ||2 + (||en+1 ||2 + ||en1 ||2 ), e 2
J1

(4.7)

(4.8) (4.9)

1 (en )x (en+1 + en1 ) ||en ||2 + (||en+1 ||2 + ||en1 ||2 ). j j x j 2 j=1

Substituting (4.6)-(4.9) into (4.5), we get 1 1 1 (||en+1 ||2 ||en1 ||2 ) + (||en+1 ||2 ||en1 ||2 ) + (||en+1 ||2 ||en1 ||2 ) xx xx x x 2 2 2 1 1 ||Ern ||2 + (||en+1 ||2 + ||en1 ||2 ) + [||en ||2 + (||en+1 ||2 + ||en1 ||2 )] x 2 2 n+1 2 n1 2 n 2 n+1 2 +C(||e || + ||e || + ||e || + ||ex || + ||en ||2 + ||en1 ||2 ). x x

(4.10)

Let B n = 1 (||en+1 ||2 + ||en ||2 ) + 1 (||en+1 ||2 + ||en ||2 ) + 1 (||en+1 ||2 + ||en ||2 ), then (4.10) can be written xx xx x x 2 2 2 as follows: B n B n1 ||Ern ||2 + C (B n + B n1 ). 7

Thus (1 C )(B n B n1 ) 2C B n1 + ||Ern ||2 . Hence, for suciently small, such that 1 C > 0, we obtain B n B n1 C B n1 + C ||Ern ||2 . Summing up (4.11) from 1 to n yields
n n

(4.11)

B n B 0 + C

l=1

||Erl ||2 + C

Bl.
l=1

(4.12)

Choose a second order method to compute u1 (such as C-N scheme) and notice that
n

l=1

||Erl ||2 n max ||Erl ||2 T O( 2 + h2 )2 .

1ln

From the discrete initial conditions, we know that e0 is of second-order accuracy, then B 0 = O( 2 + h2 )2 . Then we obtain
n1

B n O( 2 + h2 )2 + C

Bl.
l=0

An application of Lemma 4 yields B n O( 2 + h2 )2 . Thus ||en || O( 2 + h2 ), It follows from Lemma 3 that ||en || O( 2 + h2 ). This completes the proof of Theorem 3. Similarly, we can prove stability of the dierence solution. Theorem 4. Under the conditions of Theorem 3, the solution of the scheme (3.1)-(2.3) is unconditionally stable by the || || norm. 8 ||en || O( 2 + h2 ), x ||en || O( 2 + h2 ). xx

Table 1: The errors of numerical solutions at t = 10 with p = 2 and = 0.1. h 0.25 0.125 0.0625 0.03125 ||v n un || 1.456039e-4 3.657043e-5 9.084201e-6 2.202821e-6 ||v n un || 1.967455e-4 5.032358e-5 1.257422e-5 3.052964e-6 3.981465 4.025718 4.123894
||v 4 u 4 || ||v n un ||
n n n

3.909609 4.002124 4.118692

||v 4 u 4 || ||v n un ||

Numerical experiments

In this section, we conduct some numerical experiments to verify our theoretical results obtained in the previous sections. Consider the general Rosenau-RLW equation ut + uxxxxt uxxt + ux + (up )x = 0, with an initial condition u(x, 0) = u0 (x), and boundary conditions u(xl , t) = u(xr , t) = 0, uxx (xl , t) = uxx (xr , t) = 0, t [0, T ]. (5.3) x [xl , xr ], (5.2) x [xl , xr ], t [0, T ], (5.1)

The exact solution of the system (5.1)-(5.2) has the following form: u(x, t) = eln {(p+3)(3p+1)(p+1)/[2(p
4 2

+3)(p2 +4p+7)]}/(p1)

sech4/(p1) [

4p2

p1

+ 8p + 20

(x ct)], (5.4)

where p 2 is a integer and c = p4 +4p3 +14p2 +20p+25 . p +4p +10p +12p+21 It follows from (5.4) that the initial-boundary value problem (5.1)-(5.3) is consistent to the initial value problem (5.1)-(5.2) for xl 0, xr 0. In the numerical experiments, we take xl = xr = 30, T = 10, and consider three cases p = 2, 4, 8, respectively. The errors in the sense of L -norm and L2 -norm of the numerical solutions are listed on Tables 1-3 for three cases p = 2, 4, 8 with = 1. Table 1-3 verify the second-order convergence and good stability of the numerical solutions. We have shown in Theorem 1 that the numerical solution un of the scheme (2.1) satises the conh J1 n+1 servation of discrete mass and energy, respectively. In Tables 4-6, the values of (u + un ) j 2 j=1 j 1 1 n+1 2 1 n+1 2 J1 and (||un+1 ||2 + ||un ||2 |) + (|uxx || + ||un ||2 ) + (|ux || + ||un ||2 ) + h j=1 (un )x un+1 for the xx x j j 2 2 2 scheme (2.1) are presented for three cases p = 2, 4, 8 under steps h = = 0.1 with = 0, 0.5 and 1, respectively. It is easy to see from Tables 4-6 that the scheme (2.1) preserves the discrete mass and discrete energy very well, thus it can be used to computing for a long time. We make a comparision between C-N scheme [1] and our scheme with = 0, 0.5, 1 under the meshes h = = 0.1 in Figures 1 and 2 when p = 4. It is obvious from Figures 1 and 2 that our scheme 9

Table 2: The errors of numerical solutions at t = 10 with p = 4 and = 0.1. h 0.25 0.125 0.0625 0.03125 ||v n un || 2.447510e-4 6.146847e-5 1.526817e-5 3.702240e-6 ||v n un || 3.299748e-4 8.525290e-5 2.119579e-5 5.159101e-6 3.981732 4.025923 4.124036
||v 4 u 4 || ||v n un ||
n n n

3.870541 4.022161 4.108427

||v 4 u 4 || ||v n un ||

Table 3: The errors of numerical solutions at t = 10 with p = 8 and = 0.1. h 0.25 0.125 0.0625 0.03125 ||v n un || 2.854206e-4 7.170514e-5 1.781268e-5 4.319352e-6 ||v n un || 3.856020e-4 9.871671e-5 2.465926e-5 5.988124e-6 3.980476 4.025511 4.123924
||v 4 u 4 || ||v n un ||
n n n

3.906148 4.003231 4.118027

||v 4 u 4 || ||v n un ||

Table 4: Discrete mass and energy of scheme (2.1) for h = = 0.1 and p = 2. =0 = 0.5 t Qn En Qn 2 3.7953131 1.0663504 3.7953130 4 3.7953035 1.0663504 3.7953024 6 3.7952587 1.0663504 3.7952598 8 3.7950800 1.0663504 3.7950814 10 3.7943581 1.0663504 3.7943658

a few of values at dierent time t with =1 Qn 3.7953129 3.7953024 3.7952591 3.7950829 3.7943715

En 1.0661275 1.0661275 1.0661275 1.0661275 1.0661275

En 1.0659045 1.0659045 1.0659045 1.0659045 1.0659045

Table 5: Discrete mass and energy of scheme (2.1) for h = = 0.1 and p = 4. =0 = 0.5 n n t Q E Qn 2 6.2655606 2.8676742 6.2655603 4 6.2653440 2.8676742 6.2653437 6 6.2648079 2.8676742 6.2648083 8 6.2635545 2.8676742 6.2635570 10 6.2606441 2.8676742 6.2606529

a few of values at dierent time t with =1 Qn 6.2655600 6.2653435 6.2648086 6.2635595 6.2606616

E 2.8670879 2.8670879 2.8670879 2.8670879 2.8670879

En 2.8665016 2.8665016 2.8665016 2.8665016 2.8665016

10

Table 6: Discrete mass and energy of scheme (2.1) for h = = 0.1 and p = 8. =0 = 0.5 t Qn En Qn 2 9.7202265 4.7351269 9.7202105 4 9.7140074 4.7351270 9.7139777 6 9.7024346 4.7351270 9.7023976 8 9.6834925 4.7351270 9.6834566 10 9.6536097 4.7351270 9.6535850

a few of values at dierent time t with =1 Qn 9.7201954 9.7139486 9.7023610 9.6834209 9.6535602

En 4.7345960 4.7345960 4.7345960 4.7345960 4.7345960

En 4.7340650 4.7340650 4.7340650 4.7340650 4.7340650

1.6

x 10

3.5
CN scheme[1] =0 =0.5 =1

x 10

1.4

CN scheme[1] =0 =0.5 =1

1.2

2.5
1

2
||en|| 0.8

||en||2 1.5 1 0.5 0 1

0.6

0.4

0.2

6 t

10

11

6 t

10

11

Figure 1: Errors in the sense of ||en || computedFigure 2: Errors in the sense of ||en ||2 computed by by the scheme (2.1) when h = = 0.1 and p = 4. the scheme (2.1) when h = = 0.1 and p = 4.

0.4 t=0 t=5 t=10

0.7 t=0 t=5 t=10

0.35

0.6

0.3

0.5

0.25

0.4
0.2

0.3
0.15

0.2
0.1

0.1
0.05

0.05

100

200

300

400

500

600

700

0.1

100

200

300

400

500

600

700

Figure 3: Exact solutions of u(x, t) at t = 0 andFigure 4: Exact solutions of u(x, t) at t = 0 and numerical solutions computed by the scheme (2.1)numerical solutions computed by the scheme (2.1) at t = 5, 10 with = 0 for p = 2. at t = 5, 10 with = 1 for p = 4.

11

performs better than C-N scheme [1] in the numerical precision when = 0.5 and 1. Figures 1-2 also show that numerical precision of the scheme (2.1) depends on the choice of parameter . The curves of the solitary waves with time computed by the scheme (2.1) with = 0 for p = 2 and = 1 for p = 4 under mesh sizes of h = = 0.1 are given in Figures 3 and 4, respectively; the waves at t = 5, 10 agree with the ones at t = 0 quite well, which also demonstrate the accuracy of the scheme in present paper. From the numerical results, the scheme of this paper is accurate and ecient.

Acknowledgements
This work is supported by the Youth Research Foundation of WFU (No. 2011Z17). The authors would like to thank the editor and the reviewers for their valuable comments and suggestions.

References
[1] J.M. Zuo, Y.M. Zhang, T.D. Zhang et al., A new conservative dierence scheme for the General Rosenau-RLW equation, Boundary Value Problems, vol. 2010, Article ID 516260, 13 pages, 2010. [2] Zhang Fei, Vctor M. Prez-Garca, Luis Vzquez, Numerical simulation of nonlinear Schrdinger o system: a new conservative scheme, Appl. Math. Comput. 71 (1995) 165-177. [3] Zhang Fei, Luis Vzquez, Two energy conserving numerical schemes for the Sine-Gordon equation, Appl. Math. Comput. 45 (1991) 17-30. [4] S. M. Choo, S. K. Chung, Conservative nonlinear dierence scheme for the Cahn-Hilliard equation, Comput. Math. Appl. 36 (7) (1998)31-39. [5] S. M. Choo, S. K. Chung, K. I. Kim, Conservative nonlinear dierence scheme for the CahnHilliard equation II, Comput. Math. Appl. 39 (1-2) (2000) 229-243. [6] B. Hu, Y. Xu, J. Hu, Crank-Nicolson nite dierence scheme for the Rosenau-Burges equation, Appl. Math. Comput. 204 (2008) 311-316. [7] Y. Wang, Q. Chang, L. Gong, An initial-boundary value problem of a nonlinear Klein-Gordon equation, Appl. Math. Comput. 84 (1997) 77-93. [8] T.-C. Wang, L. Zhang, Analysis of some new conservative schemes for nonlinear Schrdinger o equation with wave operator, Appl. Math. Comput. 182 (2006) 1780-1794. [9] T.-C. Wang, B. Guo, L. Zhang, New conservative dierence schemes for a coupled nonlinear Schrdinger system, Appl. Math. Comput. 217 (2010) 1604-1619. o [10] Q. Chang, E. Jia, W. Sun, Dierence schemes for solving the generalized nonlinear Schrdinger o equation, J. Comput. Phys. 148 (1999) 397-415. [11] Q. Chang, B. Guo, H. Jiang, Finite dierence method for generalized Zakharov equations, Math. Comput. 64 (1995) 537-553. [12] Q. Chang, H. Jiang, A conservative scheme for the Zakharov equations, J. Comput. Phys. 113 (1994) 309-319. 12

[13] R.T. Glassey, Convergence of an energy-preserving scheme for the Zakharov equations in one space dimension, Math. Comput. 58 (1992)83-102. [14] J. S. Hu, K. L. Zheng, Two conservative dierence schemes for the generalized Rosenau equation, Boundary Value Problems, vol. 2010, Article ID 543503, 18 pages, 2010. [15] K. Omrani, F. Abidi, and T. Achouri et al., A new conservative nite dierence scheme for the Rosenau equation, Appl. Math. Comput. 201 (2008) 35-43. [16] L. Zhang, A nite dierence scheme for the generalized regularized long-wave equation, Appl. Math. Comput. 168 (2005) 962-972. [17] L. Zhang, Convergence of a conservative dierence scheme for a class of Klein-Gordon-Schrdinger o equations in one space dimension, Appl. Math. Comput. 163 (2005)343-355. [18] Y. Zhou, Application of Discrete Functional Analysis to the Finite Dierence Method, Inter. Acad. Publishers, Beijing, 1990.

13