项目名称: 求解非线性方程的加速迭代算法
项目编号: No.11471092
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 数理科学和化学
项目作者: 韩丹夫
作者单位: 杭州师范大学
项目金额: 65万元
中文摘要: 非线性迭代法是求解非线性偏微分方程的基本数值方法, 特别在多物理耦合的问题中,迭代算法的求解效率和收敛特性变得十分关键。本项目主要研究建立在传统非线性迭代法之上的二重空间加速方案及其在实际计算中的应用。通过结合经典非线性迭代(如Newton迭代)法,构造具有高收敛的二重空间加速方案,用于数值求解非线性偏微分方程或多物理的非线性耦合问题。本项目研究两种二重空间的构建方法和理论分析:一是基于同网格上的不同函数空间的二重空间方案,二是基于不同网格的二重空间方案,并针对非线性椭圆型方程和Navier-Stokes方程研究二重空间构建加速算法并给出算法的收敛性和误差估计,为二重空间方法提供算法构建和理论分析基础。本项目还将研究二重空间方法应用于多物理的耦合问题:NS/Darcy耦合流体问题和流固问题,给出基于有限元离散的二重空间加速方案、收敛性分析及算法实现等。
中文关键词: 迭代算法;二重网格;收敛性分析;Navier-Stokes方程;耦合问题
英文摘要: Nonlinear iteration is a basic strategy for numerically solving nonlinear partial differential equations. When the multi-physical problems are considered, the efficiency and the convergence rate are also critical features of the iterative schemes. Based on the classical nonlinear iterative methods, for e.g., the Newton iterations, the corresponding two-level acceleration is found to be more efficient than usual, where higher order convergence rate are archived. The current project is supposed to establish the framework of two-level methods. Two different strategies are considered, which are based on mesh refinement and adaptive polynomials' order respectively. The convergence rate and error analysis are preformed for nonlinear elliptic equation and Navier-Stokes equation. Besides, the proposed two-level schemes are applied to multi-physical coupled problems, for e.g. NS/Darcy problem and the Fluid Structure Interaction problem.
英文关键词: Iterative algorithms;Two-grid mesh;Convergence analysis;Navier-Stokes equations;nonlinear coupled problems