32. Domain Decomposition Techniques for Contour Integration Eigenvalue Solvers
Authors: Vassilis Kalantzis (University of Minnesota)Yousef Saad (University of Minnesota)James Kestyn (University of Massachusetts)Eric Polizzi (University of Massachusetts)
Abstract: This poster discusses techniques for computing a few selected eigenvalue-eigenvector pairs of large and sparse symmetric matrices. A recently developed powerful class of techniques to solve this type of problems is based on integrating the matrix resolvent operator along a complex contour that encloses the interval containing the eigenvalues of interest. This poster considers such contour integration techniques from a domain decomposition viewpoint, and extends the concept of domain decomposition linear system solvers in the framework of contour integration methods for eigenvalue problems, such as FEAST. We describe a multi-parallel implementation of FEAST using domain decomposition, discuss how the different levels of parallelism can be exploited, and report results on distributed computing environments reported. These results show that domain decomposition approaches can lead to reduced runtimes and improved scalability.
Two-page extended abstract: pdf