Research reports

A stochastically generated preconditioner for stable matrices

by F. M. Buchmann and W. P. Petersen

(Report number 2002-24)

Abstract
In this paper we simulate the Ornstein-Uhlenbeck process (OUP) to generate an approximate inverse of any real valued stable matrix. Matrix A being stable means that analytically the OUP will converge to a stationary Gaussian process in n dimensions with a covariance (2 A)-1. If the eigenvalues of A are widely separated in absolute value, however, the stiffness of the simulated linear stochastic differential equations must be considered. Hence, we consider a splitting scheme to permit large step sizes but keep convergence. Methods are described for both symmetric and non-symmetric matrices. Our preconditioner is also tested in the symmetric positive definite case by its effect on the convergence of conjugate gradient iterations.

Keywords:

BibTeX
@Techreport{BP02_310,
  author = {F. M. Buchmann and W. P. Petersen},
  title = {A stochastically generated preconditioner for stable matrices},
  institution = {Seminar for Applied Mathematics, ETH Z{\"u}rich},
  number = {2002-24},
  address = {Switzerland},
  url = {https://www.sam.math.ethz.ch/sam_reports/reports_final/reports2002/2002-24.pdf },
  year = {2002}
}

Disclaimer
© Copyright for documents on this server remains with the authors. Copies of these documents made by electronic or mechanical means including information storage and retrieval systems, may only be employed for personal use. The administrators respectfully request that authors inform them when any paper is published to avoid copyright infringement. Note that unauthorised copying of copyright material is illegal and may lead to prosecution. Neither the administrators nor the Seminar for Applied Mathematics (SAM) accept any liability in this respect. The most recent version of a SAM report may differ in formatting and style from published journal version. Do reference the published version if possible (see SAM Publications).

JavaScript has been disabled in your browser