Shift for nonsymmetric generalised eigenvalue problems

J Berns-Muller, A Spence

Research output: Contribution to journalArticle

  • 14 Citations

Abstract

In this paper we analyze inexact inverse iteration for the nonsymmetric generalized eigenvalue problem Ax = λMx, where M is symmetric positive definite and the problem is diagonalizable. Our analysis is designed to apply to the case when A and M are large and sparse and preconditioned iterative methods are used to solve shifted linear systems with coefficient matrix A − σM. We prove a convergence result for the variable shift case (for example, where the shift is the Rayleigh quotient) which extends current results for the case of a fixed shift. Additionally, we consider the approach from [V. Simoncini and L. Elden, BIT, 42 (2002), pp. 159–182] to modify the right-hand side when using preconditioned solves. Several numerical experiments are presented thatillustrate the theory and provide a basis for the discussion of practical issues.
LanguageEnglish
Pages1069-1082
Number of pages14
JournalSIAM Journal On Matrix Analysis and Applications (SIMAX)
Volume28
Issue number4
DOIs
StatusPublished - Dec 2006

Fingerprint

Generalized Eigenvalue Problem
Inverse Iteration
Preconditioned Iterative Methods
Rayleigh quotient
Positive definite
Convergence Results
Linear Systems
Numerical Experiment
Coefficient

Keywords

  • Eigenvalue approximation
  • Inverse iteration
  • Iterative methods

Cite this

Shift for nonsymmetric generalised eigenvalue problems. / Berns-Muller, J; Spence, A.

In: SIAM Journal On Matrix Analysis and Applications (SIMAX), Vol. 28, No. 4, 12.2006, p. 1069-1082.

Research output: Contribution to journalArticle

@article{2816007200874be1a5553f79b9f6b1c4,
title = "Shift for nonsymmetric generalised eigenvalue problems",
abstract = "In this paper we analyze inexact inverse iteration for the nonsymmetric generalized eigenvalue problem Ax = λMx, where M is symmetric positive definite and the problem is diagonalizable. Our analysis is designed to apply to the case when A and M are large and sparse and preconditioned iterative methods are used to solve shifted linear systems with coefficient matrix A − σM. We prove a convergence result for the variable shift case (for example, where the shift is the Rayleigh quotient) which extends current results for the case of a fixed shift. Additionally, we consider the approach from [V. Simoncini and L. Elden, BIT, 42 (2002), pp. 159–182] to modify the right-hand side when using preconditioned solves. Several numerical experiments are presented thatillustrate the theory and provide a basis for the discussion of practical issues.",
keywords = "Eigenvalue approximation, Inverse iteration, Iterative methods",
author = "J Berns-Muller and A Spence",
note = "This is the author's final, peer-reviewed version. {\circledC}Society for Industrial and Applied Mathematics",
year = "2006",
month = "12",
doi = "10.1137/050623255",
language = "English",
volume = "28",
pages = "1069--1082",
journal = "SIAM Journal On Matrix Analysis and Applications (SIMAX)",
issn = "0895-4798",
publisher = "Society for Industrial and Applied Mathematics Publications",
number = "4",

}

TY - JOUR

T1 - Shift for nonsymmetric generalised eigenvalue problems

AU - Berns-Muller,J

AU - Spence,A

N1 - This is the author's final, peer-reviewed version. ©Society for Industrial and Applied Mathematics

PY - 2006/12

Y1 - 2006/12

N2 - In this paper we analyze inexact inverse iteration for the nonsymmetric generalized eigenvalue problem Ax = λMx, where M is symmetric positive definite and the problem is diagonalizable. Our analysis is designed to apply to the case when A and M are large and sparse and preconditioned iterative methods are used to solve shifted linear systems with coefficient matrix A − σM. We prove a convergence result for the variable shift case (for example, where the shift is the Rayleigh quotient) which extends current results for the case of a fixed shift. Additionally, we consider the approach from [V. Simoncini and L. Elden, BIT, 42 (2002), pp. 159–182] to modify the right-hand side when using preconditioned solves. Several numerical experiments are presented thatillustrate the theory and provide a basis for the discussion of practical issues.

AB - In this paper we analyze inexact inverse iteration for the nonsymmetric generalized eigenvalue problem Ax = λMx, where M is symmetric positive definite and the problem is diagonalizable. Our analysis is designed to apply to the case when A and M are large and sparse and preconditioned iterative methods are used to solve shifted linear systems with coefficient matrix A − σM. We prove a convergence result for the variable shift case (for example, where the shift is the Rayleigh quotient) which extends current results for the case of a fixed shift. Additionally, we consider the approach from [V. Simoncini and L. Elden, BIT, 42 (2002), pp. 159–182] to modify the right-hand side when using preconditioned solves. Several numerical experiments are presented thatillustrate the theory and provide a basis for the discussion of practical issues.

KW - Eigenvalue approximation

KW - Inverse iteration

KW - Iterative methods

UR - http://scitation.aip.org/journals/doc/SIAMDL-home/jrnls/top.jsp?key=SJMAEL

U2 - 10.1137/050623255

DO - 10.1137/050623255

M3 - Article

VL - 28

SP - 1069

EP - 1082

JO - SIAM Journal On Matrix Analysis and Applications (SIMAX)

T2 - SIAM Journal On Matrix Analysis and Applications (SIMAX)

JF - SIAM Journal On Matrix Analysis and Applications (SIMAX)

SN - 0895-4798

IS - 4

ER -