IBLU decompositions based on Padé approximants

A. Buzdin; D. Logashenko; G. Wittum

doi:10.1002/nla.595

IBLU decompositions based on Padé approximants

A. Buzdin, D. Logashenko, G. Wittum^*

^*Corresponding author for this work

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

1 Scopus citations

Abstract

In this paper, we introduce a new class of frequency-filtering IBLU decompositions that use continued-fraction approximation for the diagonal blocks. This technique allows us to construct efficient frequencyfiltering preconditioners for discretizations of elliptic partial differential equations on domains with non-trivial geometries. We prove theoretically for a class of model problems that the application of the proposed preconditioners leads to a convergence rate of up to 1O(h^1/4) of the CG iteration.

Original language	English (US)
Pages (from-to)	717-746
Number of pages	30
Journal	Numerical Linear Algebra with Applications
Volume	15
Issue number	8
DOIs	https://doi.org/10.1002/nla.595
State	Published - Oct 2008
Externally published	Yes

Keywords

Block decomposition
Iterative method
Preconditioner for the PCG method
System of linear equations

ASJC Scopus subject areas

Algebra and Number Theory
Applied Mathematics

Access to Document

10.1002/nla.595

Cite this

@article{af0c3bea13954a67a89153cfb18b7dc5,

title = "IBLU decompositions based on Pad{\'e} approximants",

abstract = "In this paper, we introduce a new class of frequency-filtering IBLU decompositions that use continued-fraction approximation for the diagonal blocks. This technique allows us to construct efficient frequencyfiltering preconditioners for discretizations of elliptic partial differential equations on domains with non-trivial geometries. We prove theoretically for a class of model problems that the application of the proposed preconditioners leads to a convergence rate of up to 1O(h1/4) of the CG iteration.",

keywords = "Block decomposition, Iterative method, Preconditioner for the PCG method, System of linear equations",

author = "A. Buzdin and D. Logashenko and G. Wittum",

year = "2008",

month = oct,

doi = "10.1002/nla.595",

language = "English (US)",

volume = "15",

pages = "717--746",

journal = "Numerical Linear Algebra with Applications",

issn = "1070-5325",

publisher = "Wiley",

number = "8",

}

TY - JOUR

T1 - IBLU decompositions based on Padé approximants

AU - Buzdin, A.

AU - Logashenko, D.

AU - Wittum, G.

PY - 2008/10

Y1 - 2008/10

N2 - In this paper, we introduce a new class of frequency-filtering IBLU decompositions that use continued-fraction approximation for the diagonal blocks. This technique allows us to construct efficient frequencyfiltering preconditioners for discretizations of elliptic partial differential equations on domains with non-trivial geometries. We prove theoretically for a class of model problems that the application of the proposed preconditioners leads to a convergence rate of up to 1O(h1/4) of the CG iteration.

AB - In this paper, we introduce a new class of frequency-filtering IBLU decompositions that use continued-fraction approximation for the diagonal blocks. This technique allows us to construct efficient frequencyfiltering preconditioners for discretizations of elliptic partial differential equations on domains with non-trivial geometries. We prove theoretically for a class of model problems that the application of the proposed preconditioners leads to a convergence rate of up to 1O(h1/4) of the CG iteration.

KW - Block decomposition

KW - Iterative method

KW - Preconditioner for the PCG method

KW - System of linear equations

UR - http://www.scopus.com/inward/record.url?scp=55249121383&partnerID=8YFLogxK

U2 - 10.1002/nla.595

DO - 10.1002/nla.595

M3 - Article

AN - SCOPUS:55249121383

SN - 1070-5325

VL - 15

SP - 717

EP - 746

JO - Numerical Linear Algebra with Applications

JF - Numerical Linear Algebra with Applications

IS - 8

ER -

IBLU decompositions based on Padé approximants

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this