论文部分内容阅读
本文通过对目前流行的两类关于大型稀疏矩阵优化压缩设计技巧的分析,提出一种在矩阵加、乘算法中统一使用的压缩格式。文中提出的算法,无论是矩阵的输入或者运算结果的输出均采用统一的压缩格式,所以能方便地应用于矩阵链相加或相乘的运算,而且所用的存储量较少,也大大加快了运算速度。用FORTRAN77编制的两算法的标准子程序已经在DPS7机上调试成功。
Based on the analysis of two popular types of optimization design techniques for large-scale sparse matrices, this paper presents a compression scheme that is used uniformly in matrix sum and multiplication algorithms. The algorithm proposed in this paper can be conveniently applied to the operation of adding or multiplying matrix chains by using a uniform compression format regardless of the input of the matrix or the output of the operation result, and the memory used is less and greatly accelerated calculating speed. The standard subroutine of the two algorithms compiled with FORTRAN77 has been successfully debugged on the DPS7 machine.