偏微分方程保结构算法构造及分析
[Abstract]:With the rapid development of science, more and more physical, chemical and biological processes can be described by nonlinear evolution equations or electromagnetic field equations. In many cases, these systems are conservative, so how to design efficient and conservative algorithms for these systems has been a hot topic in computational science. This dissertation is devoted to the study and numerical analysis of locally conserved algorithms for nonlinear evolution equations (multi-symplectic algorithm, local energy, momentum conservation algorithm) and three-dimensional Maxwell equation efficient structure-conserving algorithm. The main research results are as follows: 1) in dealing with partial differential equations, the symplectic algorithm and the traditional conserved structure algorithms such as global conserving energy and momentum must consider not only whether the equation is a conservative system, but also whether the boundary conditions are appropriate. These algorithms can only be used under proper boundary conditions. In order to increase the applicability of structure-preserving algorithm, we construct a series of local structure-preserving algorithms for coupled nonlinear Schrodinger system, Boussinesq system and Klein-Gordon-Schrodinger system, including multi-symplectic algorithm, local energy, and local energy. Momentum conservation algorithm. These locally structure-preserving algorithms can maintain discrete local conservation laws in the space domain at any time. When the boundary conditions are appropriate, these local structure preserving algorithms are naturally global preserving algorithms, otherwise they are not. In addition, we also analyze the nonlinear stability and convergence of some locally conserved algorithms. Numerical experiments show that the locally conserved structure algorithm can not only obtain a good numerical solution, but also preserve the local conservation law and global conservation law of the system. The advantages of the proposed algorithm are demonstrated by comparing with the existing numerical algorithms in the literature. 2) the three-dimensional Maxwell equation has a double Hamiltonian structure. The Hamiltonian function and Hamiltonian operator are discretized by spectral method, and then the finite dimensional Hamiltonian system is quadrature by the mean vector field method. Thus, we obtain two schemes for solving three dimensional Maxwell equations (AVF (2) and AVF (4). AVF (2) and AVF (4) automatically hold the two Hamiltonian systems. We prove that AVF (2) and AVF (4) maintain discrete energy, momentum and divergence. Numerical dispersion analysis shows that they are unconditionally stable and nondissipative. The strict error analysis shows that AVF (2) and AVF (4) have the second order and fourth order convergence respectively in the time direction and the spectral accuracy in the space direction. The results of theoretical analysis are well confirmed by numerical experiments. 3) AVF (2) and AVF (4) are obtained by direct discrete Maxwell equations. In order to design a more efficient energy preserving algorithm, we obtain time second order and fourth order splitting methods for approximating Maxwell equations by using the techniques of exponential operator splitting and combination, respectively. Each subproblem of these splitting models is a Hamiltonian system and has the same Hamiltonian function as the original problem. For each subproblem, the Hamiltonian function and Hamiltonian operator are discretized by spectral method, and then the finite dimensional Hamiltonian system is obtained by means of the average vector field method. The second and fourth order splitting schemes are obtained (S-AVF (2) and S-AVF (4). We prove that S-AVF (2) and S-AVF (4) can maintain four discrete energies at the same time and are unconditionally stable. In addition, using discrete Fourier transform, we can write the obtained scheme into explicit form. By means of energy analysis, we obtain the error estimates of S-AVF (2) and S-AVF (4). The results of theoretical analysis are confirmed by numerical experiments.
【学位授予单位】:南京师范大学
【学位级别】:博士
【学位授予年份】:2015
【分类号】:O175.2
【相似文献】
相关期刊论文 前10条
1 斯仁道尔吉;;一个非线性发展方程的相似约化[J];内蒙古师范大学学报(自然科学汉文版);2010年05期
2 陆启韶 ,蒋正新;一类非线性发展方程的分叉问题[J];北京航空学院学报;1985年02期
3 赵家兵;;拟非线性发展方程的B解[J];应用数学;1993年04期
4 谭绍滨,,韩永前;一般非线性发展方程解的长时间行为[J];数学年刊A辑(中文版);1995年02期
5 荔炜;一类非线性发展方程解的存在性[J];宁夏大学学报(自然科学版);1996年01期
6 蒋志民;一类非线性发展方程的几何对称[J];黄淮学刊(自然科学版);1996年S2期
7 罗党;一类非线性发展方程的整体解[J];天中学刊;2000年02期
8 江成顺,孙同军,崔国忠;一非线性发展方程的反问题(英文)[J];数学研究与评论;2001年01期
9 吕蓬,吴耀红,彭武安,张辉;一类非线性发展方程的计算稳定性[J];现代电力;2001年04期
10 谢福鼎,闫振亚,张鸿庆;源于Fermi-Pasta-Ulam问题的非线性发展方程的相似约化[J];应用数学和力学;2002年04期
相关会议论文 前2条
1 史小卫;梁昌洪;;非线性弧立子理论[A];1989年全国微波会议论文集(上)[C];1989年
2 朝鲁;;求解非线性发展方程精确解的一个新方法[A];数学·力学·物理学·高新技术交叉研究进展——2010(13)卷[C];2010年
相关博士学位论文 前10条
1 张利;几类非线性发展方程解的研究[D];清华大学;2014年
2 李仕明;某些非线性发展方程的适定性与渐近性态[D];华南理工大学;2015年
3 蔡加祥;偏微分方程保结构算法构造及分析[D];南京师范大学;2015年
4 宋海涛;两类非线性发展方程的动力学行为研究[D];兰州大学;2008年
5 江杰;非线性发展方程组的整体解及渐近性态[D];复旦大学;2009年
6 吴昊;非线性发展方程及方程组整体解的渐近性态[D];复旦大学;2007年
7 尚婵妤;非线性发展方程(组)整体解及其渐近性态[D];复旦大学;2009年
8 郝江浩;若干非线性发展方程解的性质的研究[D];山西大学;2006年
9 任玉杰;非线性发展方程求解法的研究与数学机械化实现[D];大连理工大学;2007年
10 康静;非线性发展方程的势对称及线性化[D];西北大学;2008年
相关硕士学位论文 前10条
1 贾婷婷;基于符号计算的非线性发展方程的精确解和孤子运动的研究[D];太原理工大学;2016年
2 乔丽静;两类非线性发展方程的显式精确行波解[D];桂林电子科技大学;2016年
3 李琦;求解几种非线性发展方程的两类保结构方法[D];兰州大学;2016年
4 李拔萃;非线性发展方程求解中的几种构造方法[D];辽宁师范大学;2008年
5 曾嵘;关于三个非线性发展方程解的爆破和衰减性[D];重庆大学;2009年
6 张欢;基于双线性方法的非线性发展方程的求解[D];北京邮电大学;2009年
7 耿涛;若干非线性发展方程的求解研究[D];北京邮电大学;2009年
8 刘若辰;1+2-维非线性发展方程的一维最优系统及其对称约化[D];西北大学;2001年
9 胡伟;一类混合类型耗散非线性发展方程解的局部存在性[D];华中师范大学;2008年
10 包福祥;一些非线性发展方程的研究[D];内蒙古师范大学;2008年
本文编号:2410040
本文链接:https://www.wllwen.com/kejilunwen/yysx/2410040.html