A parallel algorithm for computing eigenvalues of very large real symmetric matrices on message passing architectures

The response of an airplane to air turbulence and the response of a power s ystem network to a fault in the network are two examples of important types of analyses which require large scale eigenvalue and eigenvector computati ons. In this paper we present message passing parallel algorithms for compu ting eigenvalues of very large real symmetric matrices. These algorithms ar e based upon a simple real symmetric Lanczos recursion, can achieve signifi cant speedups, have very small memory requirements, and can be used to comp ute a few or many eigenvalues. (C) 1999 Elsevier Science B.V. and IMACS. Al l rights reserved.