| 三对角线性方程组的分布式并行算法 |
| |
| 引用本文： | 迟利华,李晓梅.三对角线性方程组的分布式并行算法[J].计算机研究与发展,1998,35(11):1004-1007. |
| |
| 作者姓名： | 迟利华 李晓梅 |
| |
| 作者单位： | 国防科学技术大学计算机学院并行与分布处理国家重点实 |
| |
| 基金项目： | “九五”国防预研基金,应用物理与计算数学研究所计算物理实验室预研基金 |
| |
| 摘 要： | 文中回顾了Ｍｉｃｈｉｅｌｓｅ＆Ｖｏｒｓｔ算法，分析了影响Ｍ＆Ｖ．算法并行效率的主要因素，根据分布治之思想提出了一种求解三对角方程组的并行算法，新算法通信建立次数的Ｍ＆Ｖ算法的５０％，数据传输量为其３３％，最后的工作站网络环境下实现了新算法，就并行效率与Ｍ＆Ｖ算法进行了比较，结果表明在由６台工作站的组成的网络中新算法必能提高可达到４０％。
|
| 关 键 词： | 线性方程组 并行算法 工作站机群 |
A PARALLEL ALGORITHM FOR TRIDIAGONAL LINEAR EQUATIONS FOR DISTRIBUTED MEMORY SYSTEMS |
| |
| Chi Lihua and Li Xiaomei.A PARALLEL ALGORITHM FOR TRIDIAGONAL LINEAR EQUATIONS FOR DISTRIBUTED MEMORY SYSTEMS[J].Journal of Computer Research and Development,1998,35(11):1004-1007. |
| |
| Authors: | Chi Lihua and Li Xiaomei |
| |
| Abstract: | The method given by Michielse and Vorst is reviewed in the paper.The major factors are analyzed,which affect the efficiency of the M & V method.Then a new method for the solution of tridiagonal systems is given based on the divide and conquer computation model.The communication number of the method is half that of the M & V method,and the total data transmissions of the method is reduced to 33% that of the M & V method.The method is implemented on 6 nodes workstation clusters.The results show that compared with the M & V method,the improvement of the new method is up to 40%. |
| |
| Keywords: | tridiagonal linear systems parallel algorithm overlapping of communication and computation workstation cluster Class number TP301 |
| 本文献已被 CNKI 维普 等数据库收录！ |
