SPARSEM: a subroutine package to operate with large sparse matrices

Abstract We present here a subroutine package to operate with sparse matrices stored in compact form. For each row (or column) of the matrix, just the non-null elements are stored. The package contains routines to add and multiply matrices, convert from and to normal format, etc., and is thought to be called by programs performing more complex works. We also present the results of some tests which measure the routines' performance in comparison with standard procedures. Title of program: SPARSEM Catalogue Id: ACNQ_v1_0 Nature of problem Sparse matrices appear in a number of physical problems, such as discrete systems with many degrees of freedom; exact diagonalization of quantum clusters; etc. Versions of this program held in the CPC repository in Mendeley Data ACNQ_v1_0; SPARSEM; 10.1016/0010-4655(93)90039-F This program has been imported from the CPC Program Library held at Queen's University Belfast (1969-2019)摘要 本文提出一款用于操作压缩存储格式稀疏矩阵（sparse matrices）的子程序包（subroutine package）。该程序包仅存储矩阵各行（或各列）的非零元素（non-null elements），内置矩阵加法、乘法以及格式互转等例程（routines），可供执行更复杂任务的程序调用。本文同时给出了该子程序包与标准流程（standard procedures）的性能对比测试结果。 程序名称：SPARSEM 目录编号：ACNQ_v1_0 问题背景 稀疏矩阵广泛存在于诸多物理问题中，例如多自由度（degrees of freedom）离散系统、量子团簇（quantum clusters）的精确对角化（exact diagonalization）等场景。 Mendeley数据中CPC仓库内的该程序存档版本： ACNQ_v1_0; SPARSEM; 10.1016/0010-4655(93)90039-F 本程序源自贝尔法斯特女王大学馆藏的CPC程序库（1969-2019年）。