scientific article; zbMATH DE number 108458

MaRDI QIDQ4023918zbMATH OpenFDO

Authors Marlis Hochbruck

Publication date 23 January 1993

zbMATH Keywords

\(p\)-cyclic matricesconvergenceKrylov subspace iterationlook-ahead Lanczos methodnon-Hermitian matricesorthogonal polynomialspreconditioningquasi-minimal residual methodssingular systemsSylvester matrix equations

Mathematics Subject Classification ID

Matrix equations and identities (15A24) Iterative numerical methods for linear systems (65F10) Numerical computation of eigenvalues and eigenvectors of matrices (65F15)

Review text

One of the key points in this thesis is a compact and unified description of the nonsymmetric look-ahead Lanczos method based on formally orthogonal polynomials. This representation allows elegant proofs of \textit{M. H. Gutknecht}'s theory [SIAM J. Matrix Anal. Appl. 13, No. 2, 594-639 (1992; Zbl 0760.65039)] of these algorithms without refering to the Padé literature. In addition, a new \(s\)-step variant of the look-ahead Lanczos method as proposed by \textit{R. W. Freund}, \textit{M. H. Gutknecht}, and \textit{N. M. Nachtigal} [An implementation of the look-ahead Lanczos algorithm for non-Hermitian matrices. SIAM J. Sci. Stat. Comput. (to appear)] is developed, which is better suited for implementation on a parallel computer. The different look-ahead Lanczos algorithms are then combined with quasi- minimal residual methods, namely with \textit{R. W. Freund} and \textit{N. M. Nachtigal}'s QMR method [Numer. Math. 60, No. 3, 315-339 (1991; Zbl 0754.65034)] and with \textit{R. W. Freund}'s transpose-free variant \(TFQMR\) [A transpose-free quasi-minimal residual algorithm for non-Hermitian linear systems. Techn. Rep. 91, 18, RIACS, NASA Ames Res. Cent. (1991)]. Here, the application of quasi-minimal residual methods, which originally have been proposed for nonsingular linear systems, is extended to the singular case and to Sylvester matrix equations. A complete theory of Krylov subspace methods for singular systems is given; in particular the known convergence results for \(QMR\) and \(TFQMR\) are generalized to the singular case. Moreover, the application of look- ahead algorithms to \(p\)-cyclic systems is discussed. For Sylvester matrix equations, not only the application of Krylov subspace methods is considered, but also several approaches of preconditioning (e.g. \(SSOR\) and \(ADI\) preconditioning) are analyzed.

Recommendations

• scientific article; zbMATH DE number 53017
• The look-ahead Lanczos process for nonsymmetric matrices and its applications
• scientific article; zbMATH DE number 437442
• Krylov solvers for linear algebraic systems.
• QMR: A quasi-minimal residual method for non-Hermitian linear systems

Cited in

(11)

• Sharp error bounds of some Krylov subspace methods for non-Hermitian linear systems
• A look-back-type restart for the restarted Krylov subspace methods for solving non-Hermitian linear systems
• A theoretical overview of Krylov subspace methods
• Krylov solvers for linear algebraic systems.
• Feast eigensolver for non-Hermitian problems
• scientific article; zbMATH DE number 53017 (Why is no real title available?)
• scientific article; zbMATH DE number 2152273 (Why is no real title available?)
• ABLE: An Adaptive Block Lanczos Method for Non-Hermitian Eigenvalue Problems
• The look-ahead Lanczos process for nonsymmetric matrices and its applications
• An Implementation of the Look-Ahead Lanczos Algorithm for Non-Hermitian Matrices
• On Convergence Property of the Lanczos Method for Solving a Complex Shifted Hermitian Linear System

This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4023918)