Towards physics-oriented smoothing in algebraic multigrid for systems of partial differential equations arising in multi-ion transport and reaction models

Algebraic multigrid (AMG) methods offer an efficient solution technique for many industrial applications. However, the employment of AMG for physically involved systems of partial differential equations comprising convection, diffusion, migration (drift) and reaction terms is rather new. For such ap...

Full description

Saved in:
Bibliographic Details
Published inNumerical linear algebra with applications Vol. 17; no. 2-3; pp. 253 - 271
Main Authors Thum, Peter, Clees, Tanja
Format Journal Article
LanguageEnglish
Published Chichester, UK John Wiley & Sons, Ltd 01.04.2010
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:Algebraic multigrid (AMG) methods offer an efficient solution technique for many industrial applications. However, the employment of AMG for physically involved systems of partial differential equations comprising convection, diffusion, migration (drift) and reaction terms is rather new. For such applications, point‐based AMG (PAMG) methods are promising preconditioners. This paper focuses on developing a reordering framework for PAMG's smoothing, which is oriented on physical properties of systems arising in an electrochemical multi‐ion transport and reaction model. Level‐dependent, alternating smoothing is used. Numerical results are presented for a range of electrochemical test cases with scientific and industrial relevance demonstrating the robustness of the new framework. Overall, the results clearly indicate the usefulness and potentials of a physics‐oriented smoothing framework. Copyright © 2010 John Wiley & Sons, Ltd.
Bibliography:ArticleID:NLA706
istex:2F1FD6F2F9E63802409EE4B868E9BF23AAC8FF54
FhG Internal Programs - No. MAVO 817759
ark:/67375/WNG-TTFVX7RL-1
ObjectType-Article-2
SourceType-Scholarly Journals-1
ObjectType-Feature-1
content type line 23
ISSN:1070-5325
1099-1506
DOI:10.1002/nla.706