A new release of SPARSPAK: the Waterloo sparse matrix package

Abstract

Large sparse matrix problems arise frequently in many scientific and engineering computations. In order to solve these problems efficiently, both in terms of execution time and storage requirement, complicated storage management and data structures are usually required. However, if solving a sparse matrix problem is the prime objective, it is undesirable to have to understand the sophisticated techniques employed before the problem can be solved. Thus it is important to have software packages for solving large sparse matrix problems that provide a set of interface subroutines which insulate the users from the complicated internal structures.