An approximation method for eigenvectors of very large matrices

Document Type

Article

Publication Date

9-1991

Department

Department of Physics; Department of Mathematical Sciences

Abstract

A Monte-Carlo approach for solving huge, dense matrices for eigenvalues and eigenvectors is proposed. The matrix must satisfy certain conditions including a smooth density of diagonal elements curve and relatively constant off-diagonal elements. The approach simply involves randomly choosing a finite order (as large as computationally possible) subset matrix from the original matrix and then diagonalizing the subset. The results are crude, but often informative.

Publisher's Statement

© 1991 Plenum Publishing Corporation. Publisher’s version of record: https://doi.org/10.1007/BF01062812

Publication Title

Journal of Scientific Computing

Share

COinS