An Overdetermined Schwarz Alternating Method

Developing modern extensions of the Schwarz alternating method (SAM) has been a primary focus in the research of domain decomposition during the past 10 years. Among the various research efforts, the generalized Schwarz alternating method (GSAM) was an attempt to use the Robin condition\[ \omega u +...

Full description

Saved in:
Bibliographic Details
Published in:SIAM journal on scientific computing 1996-07, Vol.17 (4), p.884-905
Main Authors: Sun, Huosheng, Tang, Wei-Pai
Format: Article
Language:English
Subjects:
Algorithms
Decomposition
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Developing modern extensions of the Schwarz alternating method (SAM) has been a primary focus in the research of domain decomposition during the past 10 years. Among the various research efforts, the generalized Schwarz alternating method (GSAM) was an attempt to use the Robin condition\[ \omega u + (1 - \omega )\frac{{\partial u}} {{\partial n}} \]on those artificial boundaries to improve the performance of the SAM. Its convergence rate is much faster than the classical SAM (with a large overlap), yet only a minimum overlap is required. Unfortunately, sensitivity of the convergence of the GSAM to the parameter w has limited its practical applications. In this paper, a new kind of coupling is proposed which possesses the same benefits as the GSAM. The advantage of our new algorithm over the GSAM is that the optimal convergence rate is achieved on a wider range of the parameter. That is, selection of the optimal parameter is not crucial to the new algorithm's performance. Numerical tests have been carried out for a variety of difficult problems including nonsymmetric and indefinite problems.
ISSN:1064-8275
1095-7197
DOI:10.1137/0917057