A Parallel Solver for Extreme Eigenpairs

Leonardo Borges and Suely Oliveira

Abstract
In this paper a parallel algorithm for finding a group of extremeeigenvalues is presented.The algorithm is based on the well known Davidson method for findingone eigenvalue of a matrix.Here we incorporate knowledge about the structure ofthe subspace through the use of an arrowhead solver which allows moreparallelization in both the original Davidson and the new algorithM.In our numerical results various precondioners (diagonal, multigrid andADI) are compared.The performance results presented are for theParagon but our implementation is portable to machines which provideMPI and BLAS.
Contact
Suely Oliveira
Texas A&M University,Department of Computer Science,College Station, TX 77843-3112,USA,,betweem Jan.98 through July 98 I will be visiting:,THe University of Iowa,Department of Mathematics,Iowa City, Iowa 52242,USA.
suely@cs.tamu.edu