| Home > Publications > Reports > Numerical Analysis and Applied Mathematics (TW) |
TW 256
J. Janssen and S. Vandewalle
Convolution-based Chebyshev acceleration of waveform relaxation methods
Abstract
Waveform relaxation is a numerical method for solving large-scale systems of ordinary differential equations. In this paper, it is investigated whether the convergence of waveform relaxation can be accelerated by Chebyshev acceleration techniques. The Chebyshev method was originally developed for solving linear systems of equations, and can be generalised in a straightforward manner to the waveform case. It is shown that superior convergence results can be obtained from a related, convolution-based acceleration approach. A general convergence study of the convolution-based Chebyshev waveform relaxation method is presented, together with a discussion of several specific variants, some model problem analyses and numerical experiments.
report.pdf / mailto: J. Janssen