Том: 14388 , Год издания: 2023

Издатель: Springer International Publishing , Место издания: Berlin

In this work, we present the software package USPARS, developed by us for solving sparse linear systems of equations using the Gaussian elimination method. The package is conceptually similar to existing packages, but there are also differences. It includes its own component for matrix reordering to reduce the fill-in of triangular factors, as well as Python interfaces that extend the package's capabilities using Python libraries. A component for solving symmetric eigenvalue problems was developed based on the linear system solver. Through careful performance optimization, the package is competitive with popular solvers, as confirmed by numerical examples presented in the article.