一种求解线性对称变换方程的并行算法

摘要

A parallel algorithm with SYMMLQ method for solving the equation of symmetrical linear transformation was studied.The SYMMLQ method for solving linear equations is extended to solve the equation of symmetrical linear transformation.The degree of reduce operator is decreased from twice to once in the parallel process so as to improve the parallelism of the algorithm and thus reduce the computing time.The Poisson equation and elliptic partial differential equation were tested with the proposed algorithm and the original one,and the results were compared and analyzed.It is shown that the proposed SYMMLQ algorithm is superior to the original SYMMLQ algorithm.%研究求解线性对称变换方程的SYMMLQ并行算法.将求解线性方程组的SYMMLQ算法推广应用到求解线性对称变换方程,将并行过程中的两次全归约减少到一次,并对该算法进行改进,以提高并行性,减少计算时间.利用改进后的SYMMLQ算法在并行机上对Poisson方程与椭圆偏微分方程进行效果测试,并与未改进的SYMMLQ算法进行比较和分析.结果表明,改进的SYMMLQ算法的并行效率明显优于未改进的SYMMLQ算法.