摘要
文中使用范数极小技术 ,提出一种构造稀疏矩阵并行近似逆预条件子的方法 ,所构造的稀疏矩阵近似逆的稀疏结构和系数矩阵的转置矩阵相同 ,计算量和存储量小 ,其求解过程易于并行 ,且并行计算不影响其收敛效果 .通过试算表明 ,该方法对很多应用问题的求解具有明显的加速效果 .文中给出了该方法的并行算法 ,并提出了一种自适应分配算法来解决负载平衡问题 .
Based on norm minimization technique, a parallel sparse approximate inverse preconditioning method is presented for the unsymmetric sparse linear systems. The sparsity pattern of the approximate inverse is the same as that of the transpose of the coefficient matrix. This keeps the amount of work and the need of storage small. The computation of the preconditioner is inherently parallel. Some numerical experiments show that PPAT preconditioners can accelerate the convergence. This paper gives the parallel algorithm of the method, and presents a self adaptive distributed method to keep the load balance.
出处
《计算机学报》
EI
CSCD
北大核心
2000年第3期255-260,共6页
Chinese Journal of Computers
基金
计算物理实验室预研基金
关键词
线性方程组
稀疏近似逆
预条件子
并行计算
unsymmetric sparse linear system, sparse approximate inverse, preconditioner
