skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Analysis of Krylov subspace recycling for sequences of linear systems.

Abstract

Many problems from engineering and the sciences require the solution of sequences of linear systems where the matrix and right-hand side change from one system to the next, and the linear systems are not available simultaneously. We review a class of Krylov subspace methods for sequences of linear systems, which can significantly reduce the cost of solving the next system in the sequence by 'recycling' subspace information from previous systems. These methods have been successfully applied to sequences of linear systems arising from several different application areas. We analyze a particular method, GCRO-DR, that recycles approximate invariant subspaces, and establish residual bounds that suggest a convergence rate similar to one obtained by removing select eigenvector components from the initial residual. We review implications of this analysis, which suggests problem classes where we expect this technique to be particularly effective. From this analysis and related numerical experiments we also demonstrate that recycling the invariant subspace corresponding to the eigenvalues of smallest absolute magnitude is often not the best choice, especially for nonsymmetric problems, and that GCRO-DR will, in practice, select better subspaces. These results suggest possibilities for improvement in the subspace selection process.

Authors:
 [1];
  1. (University of Illinois at Urbana-Champaign)
Publication Date:
Research Org.:
Sandia National Laboratories
Sponsoring Org.:
USDOE
OSTI Identifier:
970200
Report Number(s):
SAND2005-2794C
TRN: US201003%%428
DOE Contract Number:  
AC04-94AL85000
Resource Type:
Conference
Resource Relation:
Conference: Proposed for presentation at the Seventh IMACS International Symposium on Iterative Methods in Scientific Computing held May 5-8, 2005 in Toronto, Canada.
Country of Publication:
United States
Language:
English
Subject:
99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE; CONVERGENCE; EIGENVALUES; EIGENVECTORS; ITERATIVE METHODS; RECYCLING

Citation Formats

de Sturler, Eric, and Parks, Michael L. Analysis of Krylov subspace recycling for sequences of linear systems.. United States: N. p., 2005. Web.
de Sturler, Eric, & Parks, Michael L. Analysis of Krylov subspace recycling for sequences of linear systems.. United States.
de Sturler, Eric, and Parks, Michael L. Sun . "Analysis of Krylov subspace recycling for sequences of linear systems.". United States.
@article{osti_970200,
title = {Analysis of Krylov subspace recycling for sequences of linear systems.},
author = {de Sturler, Eric and Parks, Michael L.},
abstractNote = {Many problems from engineering and the sciences require the solution of sequences of linear systems where the matrix and right-hand side change from one system to the next, and the linear systems are not available simultaneously. We review a class of Krylov subspace methods for sequences of linear systems, which can significantly reduce the cost of solving the next system in the sequence by 'recycling' subspace information from previous systems. These methods have been successfully applied to sequences of linear systems arising from several different application areas. We analyze a particular method, GCRO-DR, that recycles approximate invariant subspaces, and establish residual bounds that suggest a convergence rate similar to one obtained by removing select eigenvector components from the initial residual. We review implications of this analysis, which suggests problem classes where we expect this technique to be particularly effective. From this analysis and related numerical experiments we also demonstrate that recycling the invariant subspace corresponding to the eigenvalues of smallest absolute magnitude is often not the best choice, especially for nonsymmetric problems, and that GCRO-DR will, in practice, select better subspaces. These results suggest possibilities for improvement in the subspace selection process.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = {2005},
month = {5}
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share: