求解二维Bratu问题的有效并行算法研究
首发时间:2018-04-20
摘要:在科学计算领域,适用于大规模科学计算的并行算法的探索是当前研究的一个热点和难点.一个良好的并行算法,必须具有鲁棒性,可扩张性,能实现长时间大规模的模拟计算.而区域分解算法,恰恰具备这些特点,在本文中我们研究了求解Bratu问题的并行的Newton-Krylov-Schwarz(NKS)算法.此算法包括非精确牛顿法,克里洛夫子空间法及施瓦兹预处理技术.基于该算法,完成了二维Bratu问题的计算,计算结果表明了此算法是可行且有效的.
For information in English, please click here
AN EFFICIENT PARALLEL ALGORITHM FOR TWO-DIMENSIONAL BRATU PROBLEM
Abstract:\justifying In the field of scientific computing, the exploration of parallel algorithms for large scale scientific computing is a hot and difficulty topic in current research.An excellent parallel algorithm often have the characteristic of robustness and scalability. It also can achieve large scale simulation for a long time. The domain composition method has these characteristic. In this paper, we study the parallel Newton-Krylov-Schwarz(NKS) algorithm for Bratu problem, the algorithm contains the inexact Newton method, the Krylov subspace method and the Schwarz preconditioning technique. Based on this algorithm, the computational results of the 2-D Bratu problem are given. The results show that the algorithm is feasible and effective.
Keywords: parallel computing domain decomposition preconditioner inexact Newton method
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
求解二维Bratu问题的有效并行算法研究
评论
全部评论0/1000