APP下载

广义Rosenau-KdV-RLW方程的一个新的高精度守恒差分格式

2024-05-15胡俊林刘哲含胡劲松

关键词:收敛性稳定性

胡俊林 刘哲含 胡劲松

DOI:10.16783/j.cnki.nwnuz.2024.03.014

收稿日期:2023-10-20;修改稿收到日期:2023-11-25

基金项目:四川省应用基础研究资助项目(2019YJ0387);国家自然科学青年基金资助项目(11701481)

作者简介:胡俊林(1998—),男,四川会理人,硕士研究生.主要研究方向为微分方程数值解.

E-mail:1072450812@qq.com

*通信联系人,男,四川射洪人,教授,博士.主要研究方向为微分方程数值解.

E-mail:hjs888hjs@163.com

摘要:对一类广义Rosenau-KdV-RLW方程的初边值问题提出一个新的高精度守恒差分算法.利用Taylor展式,在空间层做部分外推处理,直接从整体上抵消空间截断误差的二阶部分,在时间层采用Crank-Nicolson格式,从而在时间方向和空间方向分别达到了二阶精度和四阶精度;合理模拟了问题本身的一个守恒量,并利用离散Sobolev嵌入不等式和离散泛函分析方法,证明了格式的收敛性和稳定性;最后,数值算例验证了该方法的有效性.

关键词:广义Rosenau-KdV-RLW方程;高精度守恒差分格式;收敛性;稳定性

中图分类号:O 241.82    文献标志码:A    文章编号:1001-988Ⅹ(2024)03-0127-06

A new high-accuracy conservative difference scheme

for the generalized Rosenau-KdV-RLW equation

HU Jun-lin,LIU Zhe-han,HU Jin-song

(School of Science,Xihua University,Chengdu 610039,Sichuan,China)

Abstract:A new conservative difference algorithm with the high-accuracy is proposed for the initial boundary value problem of the generalized Rosenau-KdV-RLW equation.For the space direction in this scheme,the Taylor expansion and the partial extrapolation are used to make the second-order term of the truncation error be removed directly,then it can achieve the fourth-order accuracy.And for the time direction,the Crank-Nicolson method is performed so that it has second-order accuracy in time.This algorithm can simulate reasonably a conservative property of the original problem.By using discrete Sobolev embedding inequality and the discrete functional analysis method,the convergence and stability of this algorithm are proved,respectively.Finally,the numerical examples show that this algorithm is reliable.

Key words:generalized Rosenau-KdV-RLW equation;high-accuracy conservative difference scheme;convergence;stability

考虑广义Rosenau-KdV-RLW方程[1,2]初边值问题

ut-3ux2t+5ux4t+

ux+3ux3+upx=0,

x∈(xL,xR),t∈(0,T],(1)

u|t=0=u0(x), x∈[xR,xL],(2)

u|x=xR=u|x=xL=0,

2ux2x=xR=2ux2x=xL=0,

t∈(0,T],(3)

其中u0(x)是一個已知的光滑函数,p≥2为整数.问题(1)~(3)满足守恒律[1,2]:

Q(t)=∫xRxLu(x,t)dx=

∫xRxLu0(x)dx=Q(0),(4)

其中Q(0)为仅与初始条件有关的常数.方程(1)是描述非线性波动行为的重要数学模型,在许多领域都有着广泛的应用,著名的KdV方程[3]、RLW方程[4]和Rosenau方程[5]等均可视为其特殊情形,其数值求解方法研究也备受关注[6-16].由于不能用离散分部求和公式[17]在齐次边界条件下推导出如下结论[16]:

(n)p,2n=0, p≥2,

故文献[16]不能从理论上严格保证其方法的有效性.本文利用Tayor展开,在空间层做部分外推离散,直接从整体上抵消空间截断误差的二阶部分,在时间层采用Crank-Nicolson格式,从而对问题(1)~(3)提出一个理论精度为O(τ2+h4)的两层非线性差分格式,并合理模拟了守恒量(4),并利用离散Sobolev嵌入不等式[17]和离散泛函分析方法给出了其收敛性和稳定性的理论证明,最后进行数值验证.

1  差分格式及其守恒律

对区域[xL,xR]×[0,T]作等距网格剖分,设时间步长为τ,令

tn=nτ, 0≤n≤N,N=Tτ;

xj=xL+jh, 0≤j≤J,h=xR-xLJ

为空间步长;记unj=u(xj,tn),Unj≈u(xj,tn),约定C是一般常数,且C>0.并定义

(Unj)x=Unj+1-Unjh, (Unj)=Unj-Unj-1h,

(Unj)=Unj+1-Unj-12h, (Unj)t=Un+1j-Unjτ,

Un+1/2j=Un+1j+Unj2, Un,Vn=h∑J-1j=1UnjVnj,

Un2=Un,Un, Un∞=max1≤j≤J-1Unj,

Z0h={U=(Uj):U-2=U-1=U0=UJ=UJ+1=

UJ+2=0, j=-2,-1,…,J+1,J+2}.

若函數u(x,t)足够光滑,当τ,h→0时,通过Taylor展开可得

(unj)x=2ux2nj+112h24ux4nj+O(h4),(5)

(unj)=2ux2nj+13h24ux4nj+O(h4),(6)

(unj)x=3ux3nj+14h25ux5nj+O(h4),(7)

(unj)=3ux3nj+12h25ux5nj+O(h4).(8)

由(5)~(8)式,有

23(unj)x+13(unj)=2ux2nj+

16h24ux4nj+O(h4),(9)

43(unj)x-13(unj)=3ux3nj+

16h25ux5nj+O(h4),(10)

(unj)t=utn+1/2j+O(τ2),(11)

(unj)=uxnj+16h23ux3nj+O(h4),(12)

[(unj)p]=upxnj+16h23upx3nj+O(h4),(13)

(unj)xx=4ux4nj+16h26ux6nj+O(h4).(14)

将方程(1)对x两次求偏导,则有

3ux2t-5ux4t+7ux6t+

3ux3+5ux5+3upx3=0,(15)

再将方程(1)在点(xj,tn+1/2)处进行差分离散,并结合(9)~(14)式有

(unj)t-23(unj)xt+13(unj)t-

16h25ux4tn+1/2j+

(unj)xxt-16h27ux6tn+1/2j+

(un+1/2j)-16h23ux3n+1/2j+

43(un+1/2j)x-13(un+1/2j)-

16h25ux5n+1/2j+

[(un+1/2j)p]-16h23upx3n+1/2j=

O(τ2+h4).

结合(15)式,整理有

(unj)t-23(unj)xt+13(unj)t+

(unj)xxt+(un+1/2j)+

43(un+1/2j)x-13(un+1/2j)+

[(un+1/2j)p]+16h23ux2tnj=

O(τ2+h4).

再由(5)式可得

(unj)t-23(unj)xt+13(unj)t+

(unj)xxt+(un+1/2j)+

43(un+1/2j)x-13(un+1/2j)+

[(un+1/2j)p]+16h2(unj)xt=

O(τ2+h4).(16)

于是,对问题(1)~(3)构造如下差分格式:

(Unj)t-23(Unj)xt+13(Unj)t+

(Unj)xxt+(Un+1/2j)+

43(Un+1/2j)x-13(Un+1/2j)+

[(Un+1/2j)p]+16h2(Unj)xt=0,

j=1,2,…,J-1,

n=0,1,…,N-1;(17)

Unj=u0(xj), j=0,1,…,J;(18)

Un∈Z0h, n=0,1,…,N.(19)

定理1  差分格式(17)~(19)关于离散能量

Qn=h∑J-1j=1Unj=Qn-1=…=Q0(20)

守恒,其中,n=1,2,…,N.

证明  将(17)式两端乘以h后,对j从1到J-1求和,由边界条件(19)和分部求和公式[17]有

h∑J=1j=1(Unj)t=0,

由Qn的定义,关于n递推即可得(20)式.  】

2  收敛性和稳定性

定义差分格式(17)~(19)的截断误差为:

rnj=(unj)t-23(unj)xt+13(unj)t+

(unj)xxt+(un+1/2j)+

43(un+1/2j)x-13(un+1/2j)+

[(un+1/2j)p]+16h2(unj)xt,

j=1,2,…,J-1,

n=0,1,…,N-1;(21)

u0j=u0(xj), j=0,1,…,J;(22)

un∈Z0h, n=0,1,…,N.(23)

由(16)式可知,当h,τ→0时,

|rnj|=O(τ2+h4).(24)

引理1[8]  设u0足够光滑,则初边值问题(1)~(3)的解满足:

uL∞≤C, uxL∞≤C.

引理2[15]  对U∈Z0h有U2≤Ux2.

定理2  设u0足够光滑,若时间步长τ和空间步长h充分小,则差分格式(17)~(19)的解Un以·∞收敛到初边值问题(1)~(3)的解,且收敛阶为O(τ2+h4).

证明  用数学归纳法.记enj=unj-Unj,由(21)~(23)式减去(17)~(19)式,有

rnj=(enj)t-23(enj)xt+13(enj)t+

(enj)xxt+(en+1/2j)+

43(en+1/2j)x-13(en+1/2j)+

[(un+1/2j)p]-[(Un+1/2j)p]+h26(enj)xt,

j=1,2,…,J-1,

n=0,1,…,N-1;(25)

e0j=0, j=0,1,…,J;(26)

en∈Z0h, n=0,1,…,N.(27)

由引理1以及(24)式知,存在与τ和h无关的常数Cu和Cr,使得

un∞≤Cu, rn∞≤Cr(τ2+h4),

n=1,2,…,N.(28)

再由(26)式以及(18)式可得以下估计:

e0=0, U0≤Cu.(29)

现在假设当n≤N-1时,有

el+elx+elxx≤Cl(τ2+h4),(30)

其中Cl(l=1,2,…,n)为与τ和h无关的常数,则由离散的Sobolev不等式[17]和Cauchy-Schwarz不等式,有

el∞≤C0elelx+el≤

12C0(2el+elx)≤

32C0Cl(τ2+h4),(31)

Ul∞≤ul∞+el∞≤

Cu+32C0Cl(τ2+h4),

l=1,2,…,n.(32)

将(25)式两端与en+1/2取内积,由边界条件(27)式和分部求和公式[17],并注意到

en+1/2,en+1/2=0,

en+1/2x,en+1/2=0,

en+1/2,en+1/2=0,

整理得

12en2t+13-h212enx2t+

16en2t+12enxx2t=

rn,en+1/2-[(un+1/2)p-

(Un+1/2)p],en+1/2.(33)

再取h和τ充分小,使得

32C0·max0≤l≤nCl(τ2+h4)≤1,(34)

則由(32),(34)式和分部求和公式[17]、引理2以及Cauchy-Schwarz不等式可得

-[(un+1/2)p-(Un+1/2)p],en+1/2=

[(un+1/2)p-(Un+1/2)p],en+1/2=

h∑J-1j=1en+1/2j∑p-1i=1(un+1/2j)p-1-i(Un+1/2j)i)×

(en+1/2j)≤

∑p-1i=1(Cu)p-1-i(Cu+1)ih×

∑J-1j=1en+1/2j·(en+1/2j)≤

p(Cu+1)p-1h∑J-1j=1en+1/2j·(en+1/2j)≤

p2(Cu+1)p-1en+1/22+en+1/2x2≤

p4(Cu+1)p-1(en+12+en2+

en+1x2+enx2),(35)

rn,en+1/2≤12rn2+

14(en+12+en2).(36)

将(35)和(36)式代入(33)式,整理得

(en+12-en2)+23-h26×

(en+1x2-enx2)+

13(en+12-en2)+

(en+1xx2-enxx2)≤

τrn2+12τ(en+12+en2)+

τp2(Cn+1)p-1×

(en+12+en2+en+1x2+enx2)≤

τrn2+τp(Cu+1)p-1(en+12+

en2+en+1x2+enx2).(37)

An=en2+enx2+enxx2,

Bn=en2+23-h26enx2+

13en2+enxx2,

将(37)式从1到n递推求和,并整理有

Bn+1≤B1+τ∑nk=1rk2+

τ∑n+1k=12p(Cn+1)p-1×

(ek2+ekx2).(38)

由(28)式和(30)式有

τ∑nk=1rk2≤nτmax1≤k≤nrk2≤T(Cr)2(τ2+h4)2,

B1≤(C1)2(τ2+h4)2,

取h充分小使得2/3-h2/6>1/3,则(38)式变为

An+1≤3Bn+1≤3(T(Cr)2+(C1)2)(τ2+h4)2+

τ∑n+1k=16p(Cu+1)p-1Ak.

利用离散Gronwall不等式[17],取

τ<112p(Cn+1)p-1,

An+1≤(Cn+1)2(τ2+h4)2, n=1,2,…,N-1,

其中Cn+1=(3TCr+3C1)eT[6p(Cu+1)p-1].显然常数Cn+1与时间层n无关.从而由归纳假设有

en≤O(τ2+h4), enx≤O(τ2+h4),

enxx≤O(τ2+h4), n=1,2,…,N.

最后由离散的Sobolev不等式[17],有

en∞≤O(τ2+h4), n=1,2,…,N.  】

定理3  在定理2的条件下,差分格式(17)~(19)的解满足:

Un∞≤0, n=1,2,…,N,

其中0是与τ和h无关的常数.

证明  对于充分小的τ和h,由定理2有

Un∞≤un∞+en∞≤0.  】

注1  定理3表明差分格式(17)~(19)的解Un以·∞关于初值稳定.

3  数值实验

考虑p=3和p=5两种情形进行数值实验.当p=3时,方程(1)的孤波解[1]为

u(x,t)=15(57-5)/(45(57-5)-8)×

sech218257-10

x-142(557+33)t;

当p=5时,方程(1)的孤波解[1]为

u(x,t)=3-3/42243-10×

(9/80(43-5)-72))1/4×

sech16443-20×

x-191(59+1043)t.

在数值实验中,取初值函数u0(x)=u(x,0).固定xL=-40,xR=120,T=40.就τ和h的不同取值,数值解在一些不同时刻的误差见表1~2,对差分格式(17)~(19)的理论精度检验见表3~4,对守恒量(4)的数值模拟部分数据见表5.

从数值实验结果可以看出,本文对问题(1)~(3)提出的差分格式(17)~(19)是可靠的.

参考文献:

[1]  RAZBOROVA P,AHMED B,BISWAS A.Solitons,shock waves and conservation laws of Rosenau-KdV-RLW equation with power law nonlinearity[J].Appl Math Inf Sci,2014,8(2):485.

[2]  SANCHEZ P,EBADI G,MOJAVER A,et al.Solitons and other solutions to perturbed Rosenau-KdV-RLW equation with power law nonlinearity[J].Acta Phys Polon A,2015,127(6):1577.

[3]  KORTEWAG D J,DEVRIES G.On the change of form of long waves advancing in a rectangular canal,and on a new type of long stationary waves[J].Philos Mag,1985,39(5):422.

[4]  BENJAMIN T B,BONA J L,MAHONY J J.Model equations for long waves in nonlinear dispersive systems[J].Philos T R Soc B,1972,272:47.

[5]  ROSENAU P.A quasi-continuous description of a nonlinear transmission line[J].Phys Scripta,1986,34:827.

[6]  FOROUTAN M,EBADIAN A,Chebyshev rational approximations for the Rosenau-KdV-RLW equation on the whole line[J].Int J Anal Appl,2018,16(1):1.

[7]  SIBEL .An effective numerical technique for the Rosenau-KdV-RLW equation[J].Balkesir niv Fen Bilim Enst derg,2018,20(3):1.

[8]  卓茹,李佳佳,胡勁松.求解广义Rosenau-KdV-RLW方程的一个非线性守恒差分格式[J].四川大学学报(自然科学版),2017,54(4):703.

[9]  WONGSAIJAI B,POOCHINAPAN K.A three-level average implicit finite difference scheme to solve equation obtained by coupling the Rosenau-KdV equation and the Rosenau-RLW equation[J].Appl Math Comput,2014,245:289.

[10]  PAN X,WANG Y,ZHANG L.Numerical analysis of a pseudo-compact C-N conservative scheme for the Rosenau-KdV equation coupling with the Rosenau-RLW equation[J].Bound Value Probl,2015:65.

[11]  SHALLU,KUMAR K V.An efficient collocation algorithm with SSP-RK43 scheme to solve Rosenau-KdV-RLW equation[J].Int J Ap Mat Com-Pol,2021,7(161):1.

[12]  AHMAT M,QIU J X.SSP IMEX Runge-Kutta WENO scheme for generalized Rosenau-KdV-RLW equation[J].J Math Study,2022,55(1):1.

[13]  ANKUR,JIWARI R,KUMAR N.Analysis and simulation of Korteweg-de Vries-Rosenau-Regularised long-wave model via Galerkin finite element method[J].Comput Math Appl,2023,135:134.

[14]  VERMA A K,RAWANI M K.Numerical solutions of generalized Rosenau-KdV-RLW equation by using haar wavelet collocation approach coupled with nonstandard finite difference scheme and quasilinearization[J].Numer Meth Part D E,2022,39(2):1085.

[15]  GHILOUFI A,OMRANI K.New conservative difference schemes with fourth-order accuracy for some model equation for nonlinear dispersive waves[J].Numer Meth Part D E,2018,34(2):451.

[16]  WANG X,DAI W.A new conservative finite difference scheme for the generalized Rosenau-KdV-RLW Equation[J].Comput Math Appl,2020,39(3):1.

[17]  ZHOU Y.Application of Discrete Functional Analysis to the Finite Difference Method[M].Beijing:International Academic Publishers,1991.

(責任编辑  马宇鸿)

猜你喜欢

收敛性稳定性
一类k-Hessian方程解的存在性和渐近稳定性
SBR改性沥青的稳定性评价
行间AANA随机变量阵列加权和的完全矩收敛性
WOD随机变量序列的完全收敛性和矩完全收敛性
END随机变量序列Sung型加权和的矩完全收敛性
非线性中立型变延迟微分方程的长时间稳定性
END随机变量序列Sung型加权和的矩完全收敛性
半动力系统中闭集的稳定性和极限集映射的连续性
行为ND随机变量阵列加权和的完全收敛性
松弛型二级多分裂法的上松弛收敛性