The PRISM project: infrastructure and algorithms for parallel eigensolvers

Author

Bischof, Christian ; Huss-Lederman, Steven ; Sun, Xiaobai ; Tsao, Anna

Author_Institution

Argonne Nat. Lab., IL, USA

fYear

1993

fDate

6-8 Oct 1993

Firstpage

123

Lastpage

131

Abstract

The goal of the PRISM project is the development of infrastructure and algorithms for the parallel solution of eigenvalue problems. We are currently investigating a complete eigensolver based on the invariant Subspace Decomposition Algorithm for dense symmetric matrices (SYISDA). After briefly reviewing the SYISDA approach, we discuss the algorithmic highlights of a distributed-memory implementation of an eigensolver based on this approach. These include a fast matrix-matrix multiplication algorithm, a new approach to parallel band reduction and tridiagonalization, and a harness for coordinating the divide-and-conquer parallelism in the problem. We also present performance results of these kernels as well as the overall SYISDA implementation on the Intel Touchstone Delta prototype and the IBM SP/1

Keywords

distributed memory systems; eigenvalues and eigenfunctions; mathematics computing; matrix algebra; parallel algorithms; IBM SP/1; Intel Touchstone Delta prototype; PRISM project; SYISDA approach; distributed-memory implementation; divide-and-conquer parallelism; eigenvalue problems; invariant Subspace Decomposition Algorithm for dense symmetric matrices; matrix-matrix multiplication algorithm; parallel band reduction; parallel eigensolvers algorithms; tridiagonalization; Concurrent computing; Contracts; Eigenvalues and eigenfunctions; Kernel; Laboratories; Matrix decomposition; Parallel algorithms; Polynomials; Software algorithms; Symmetric matrices;

fLanguage

English

Publisher

ieee

Conference_Titel

Scalable Parallel Libraries Conference, 1993., Proceedings of the

Conference_Location

Mississippi State, MS

Print_ISBN

0-8186-4980-1

Type

conf

DOI

10.1109/SPLC.1993.365575

Filename

365575