A domain decomposing parallel sparse linear system solver

TL;DR

This paper introduces a hybrid parallel solver for large sparse linear systems that combines direct and iterative methods, improving scalability and robustness for distributed memory systems.

Contribution

A novel parallel hybrid sparse linear system solver that balances direct and iterative approaches for better scalability and robustness.

Findings

Achieves better scalability than pure direct solvers.

Provides more robustness than classical preconditioned iterative solvers.

Demonstrates improved performance on parallel architectures.

Abstract

The solution of large sparse linear systems is often the most time-consuming part of many science and engineering applications. Computational fluid dynamics, circuit simulation, power network analysis, and material science are just a few examples of the application areas in which large sparse linear systems need to be solved effectively. In this paper we introduce a new parallel hybrid sparse linear system solver for distributed memory architectures that contains both direct and iterative components. We show that by using our solver one can alleviate the drawbacks of direct and iterative solvers, achieving better scalability than with direct solvers and more robustness than with classical preconditioned iterative solvers. Comparisons to well-known direct and iterative solvers on a parallel architecture are provided.