A domain-decomposing parallel sparse linear system solver

Creative Commons License

Manguoglu M.

JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS, vol.236, no.3, pp.319-325, 2011 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 236 Issue: 3
  • Publication Date: 2011
  • Doi Number: 10.1016/j.cam.2011.07.017
  • Journal Name: JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.319-325
  • Keywords: Sparse linear systems, Parallel solvers, Direct solvers, Iterative solvers, IRREGULAR GRAPHS, INDEFINITE, ALGORITHM, SCHEME, SPIKE
  • Middle East Technical University Affiliated: Yes

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. (C) 2011 Elsevier B.V. All rights reserved.