Iterative refinement techniques for solving block linear systems of equations

Agata Smoktunowicz; Alicja Smoktunowicz

doi:10.1016/j.apnum.2011.11.004

Iterative refinement techniques for solving block linear systems of equations

Agata Smoktunowicz, Alicja Smoktunowicz

School of Mathematics

Research output: Contribution to journal › Article › peer-review

Abstract

We study the numerical properties of classical iterative refinement (IR) and k-fold iterative refinement (RIR) for computing the solution of a nonsingular linear system of equations Ax = b with A partitioned into blocks using floating point arithmetic. We assume that all computations are performed in the working (fixed) precision. We prove that the numerical quality of RIR is superior to that of IR.

Original language	English
Pages (from-to)	220-229
Number of pages	10
Journal	Applied Numerical Mathematics
Volume	67
DOIs	https://doi.org/10.1016/j.apnum.2011.11.004
Publication status	Published - May 2013

Access to Document

10.1016/j.apnum.2011.11.004

Iterative refinement techniques for solving block linear systems of equationsAccepted author manuscript, 597 KB

Link to publication in Scopus

1 Finished

Nil algebras, algebraic algebras and algebras with finite Gelfand-Kirillov dimension
Smoktunowicz, A.
EPSRC
1/08/06 → 31/07/11
Project: Research

Cite this

@article{48a0b506f789429fb27b9710eef9e16f,

title = "Iterative refinement techniques for solving block linear systems of equations",

abstract = "We study the numerical properties of classical iterative refinement (IR) and k-fold iterative refinement (RIR) for computing the solution of a nonsingular linear system of equations Ax = b with A partitioned into blocks using floating point arithmetic. We assume that all computations are performed in the working (fixed) precision. We prove that the numerical quality of RIR is superior to that of IR.",

author = "Agata Smoktunowicz and Alicja Smoktunowicz",

year = "2013",

month = may,

doi = "10.1016/j.apnum.2011.11.004",

language = "English",

volume = "67",

pages = "220--229",

journal = "Applied Numerical Mathematics",

issn = "0168-9274",

publisher = "Elsevier",

}

TY - JOUR

T1 - Iterative refinement techniques for solving block linear systems of equations

AU - Smoktunowicz, Agata

AU - Smoktunowicz, Alicja

PY - 2013/5

Y1 - 2013/5

N2 - We study the numerical properties of classical iterative refinement (IR) and k-fold iterative refinement (RIR) for computing the solution of a nonsingular linear system of equations Ax = b with A partitioned into blocks using floating point arithmetic. We assume that all computations are performed in the working (fixed) precision. We prove that the numerical quality of RIR is superior to that of IR.

AB - We study the numerical properties of classical iterative refinement (IR) and k-fold iterative refinement (RIR) for computing the solution of a nonsingular linear system of equations Ax = b with A partitioned into blocks using floating point arithmetic. We assume that all computations are performed in the working (fixed) precision. We prove that the numerical quality of RIR is superior to that of IR.

UR - http://www.scopus.com/inward/record.url?scp=84876133919&partnerID=8YFLogxK

U2 - 10.1016/j.apnum.2011.11.004

DO - 10.1016/j.apnum.2011.11.004

M3 - Article

VL - 67

SP - 220

EP - 229

JO - Applied Numerical Mathematics

JF - Applied Numerical Mathematics

SN - 0168-9274

ER -

Iterative refinement techniques for solving block linear systems of equations

Abstract

Access to Document

Nil algebras, algebraic algebras and algebras with finite Gelfand-Kirillov dimension

Cite this