Numerical implementation of the QMR algorithm by using discrete stochastic arithmetic

In each step of the quasi-minimal residual (QMR) method which uses a look-ahead variant of the nonsymmetric Lanczos process to generate basis vectors for the Krylov subspaces induced byA, it is necessary to decide whether to construct the Lanczos vectorsvn+1 andwn+1 as regular or inner vectors. For...

Full description

Saved in:
Bibliographic Details
Published in:Journal of applied mathematics & computing Vol. 17; no. 1-2; pp. 457 - 473
Main Authors: Toutounian, Faezeh, Salkuyeh, Davod Khojasteh, Asadi, Bahram
Format: Journal Article
Language:English
Published: Dordrecht Springer Nature B.V 01.03.2005
Subjects:
Algorithms
Applied mathematics
Arithmetic
Computation
Floating point arithmetic
Mathematical analysis
Mathematical models
Stochastic models
Stochasticity
Studies
Subspaces
Tolerances
Vectors (mathematics)
ISSN:1598-5865, 1865-2085
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In each step of the quasi-minimal residual (QMR) method which uses a look-ahead variant of the nonsymmetric Lanczos process to generate basis vectors for the Krylov subspaces induced byA, it is necessary to decide whether to construct the Lanczos vectorsvn+1 andwn+1 as regular or inner vectors. For a regular step it is necessary thatDk=WkTVk is nonsingular. Therefore, in the floating-point arithmetic, the smallest singular value of matrix Dk,σmin(Dk), is computed and an inner step is performed ifσmin(Dk)<∈, where ∈ is a suitably chosen tolerance. In practice it is absolutely impossible to choose correctly the value of the tolerance ∈. The subject of this paper is to show how discrete stochastic arithmetic remedies the problem of this tolerance, as well as the problem of the other tolerances which are needed in the other checks of the QMR method with the estimation of the accuracy of some intermediate results. Numerical examples are used to show the good numerical properties.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ObjectType-Feature-1
content type line 23
ObjectType-Article-2
ISSN:1598-5865
1865-2085
DOI:10.1007/BF02936068