您的位置:首页 > 论文页面
粒子输运方程的预条件迭代快速求解
发表时间:2018-04-13 浏览量:1521 下载量:505
全部作者: | 张慧慧,曹艳华 |
作者单位: | 华北电力大学数理学院信息与计算研究所 |
摘 要: | 通过源迭代和Krylov子空间方法求解输运方程。由于Krylov子空间方法的收敛速度依赖于系数矩阵的谱半径,为此引入三种预条件矩阵对系数矩阵进行预处理,并将预处理后的广义极小张量(generalized minimal residual,GMRES)算法、稳定双共轭梯度(bi-conjugate gradient stabilized,BiCGSTAB)算法与源迭代方法进行比较。数值实验结果表明,预条件的Krylov子空间方法收敛所需的迭代次数与CPU时间均比源迭代方法少很多,同时预条件的BiCGSTAB算法收敛所需的迭代次数与CPU时间比预条件的GMRES算法稍少一些。此外,对于同一种算法而言,近似逆预条件方法的预处理效果略优于其余两种预条件方法。 |
关 键 词: | 计算数学;粒子输运;源迭代;Krylov子空间;预条件 |
Title: | Preconditioned iterative fast solution for particle transport equation |
Author: | ZHANG Huihui, CAO Yanhua |
Organization: | Institute of Information and Computation, Mathematical and Physical Science School, North China Electric Power University |
Abstract: | In this paper, the transport equation is solved by source iteration and Krylov subspace methods. Because the convergence rate of the Krylov subspace method depends on the spectral radius of the coefficient matrix, three preconditioned matrices are introduced to preprocess the coefficient matrix. Moreover, the pretreated generalized minimal residual (GMRES) algorithm and bi-conjugate gradient stabilized (BiCGSTAB) algorithm are compared with the source iteration method. Numerical experiment shows that the preconditioned Krylov subspace method requires less iterations and CPU time for convergence than the source iteration method. At the same time, iterations and CPU time of the preconditioned BiCGSTAB algorithm required for convergence are slightly less than the preconditioned GMRES algorithm. Besides, the pretreatment effect of approximate inverse preconditioning method is better than other two kinds of preconditioning methods for the same algorithm. |
Key words: | computational mathematics; particle transport; source iteration; Krylov subspace; precondition |
发表期数: | 2018年4月第7期 |
引用格式: | 张慧慧,曹艳华. 粒子输运方程的预条件迭代快速求解[J]. 中国科技论文在线精品论文,2018,11(7):689-697. |

请您登录
暂无评论