Convergence Analysis and Acceleration of the Schwartz Alternating Method,
Abstract
The convergence rate of the Schwarz Alternating Method(SAM) is studied for applications involving the solution of elliptic equations on composite grids. Such problems arise when solvers which can only be used on special domains, such as rectangles, are used for more general region; and in the disection of problems for parallel processing. It is shown that the convergence rate is a function of the overlap, number and shape of the subregions into which the problem domain is divided. The convergence rates for SAM are slow and an accelerated method based on overrelaxation techniques is developed. The SAM analysis is extended to predict the performance of the accelerated method and optimal relaxation parameters. Finally, we study the effects of changing the iteration order for the SAM and accelerated SAM methods.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 26, 1986
- Accession Number
- ADA173961
Entities
People
- Joseph Oliger
- Wei-pai Tang
- William Skamarock
Organizations
- Stanford University