A blocked QR-decomposition for the parallel symmetric eigenvalue problem

T. Auckenthaler; T. Huckle; R. Wittmann

doi:10.1016/j.parco.2014.03.010

A blocked QR-decomposition for the parallel symmetric eigenvalue problem

T. Auckenthaler
, T. Huckle
, R. Wittmann

Informatics 5 - Chair of Scientific Computing

Technical University of Munich

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

9 Scopus citations

Abstract

In this paper we present a new stable algorithm for the parallel QR-decomposition of "tall and skinny" matrices. The algorithm has been developed for the dense symmetric eigensolver ELPA, where the QR-decomposition of tall and skinny matrices represents an important substep. Our new approach is based on the fast but unstable CholeskyQR algorithm (Stathopoulos and Wu, 2002) [1]. We show the stability of our new algorithm and provide promising results of our MPI-based implementation on a BlueGene/P and a Power6 system.

Original language	English
Pages (from-to)	186-194
Number of pages	9
Journal	Parallel Computing
Volume	40
Issue number	7
DOIs	https://doi.org/10.1016/j.parco.2014.03.010
State	Published - Jul 2014

Keywords

Eigenvalue and eigenvector computation
Parallelization
QR-decomposition
Two-step tridiagonalization

Access to Document

10.1016/j.parco.2014.03.010

Cite this

@article{1045a482a7604d0fbde92460eea07f53,

title = "A blocked QR-decomposition for the parallel symmetric eigenvalue problem",

abstract = "In this paper we present a new stable algorithm for the parallel QR-decomposition of {"}tall and skinny{"} matrices. The algorithm has been developed for the dense symmetric eigensolver ELPA, where the QR-decomposition of tall and skinny matrices represents an important substep. Our new approach is based on the fast but unstable CholeskyQR algorithm (Stathopoulos and Wu, 2002) [1]. We show the stability of our new algorithm and provide promising results of our MPI-based implementation on a BlueGene/P and a Power6 system.",

keywords = "Eigenvalue and eigenvector computation, Parallelization, QR-decomposition, Two-step tridiagonalization",

author = "T. Auckenthaler and T. Huckle and R. Wittmann",

year = "2014",

month = jul,

doi = "10.1016/j.parco.2014.03.010",

language = "English",

volume = "40",

pages = "186--194",

journal = "Parallel Computing",

issn = "0167-8191",

publisher = "Elsevier B.V.",

number = "7",

}

TY - JOUR

T1 - A blocked QR-decomposition for the parallel symmetric eigenvalue problem

AU - Auckenthaler, T.

AU - Huckle, T.

AU - Wittmann, R.

PY - 2014/7

Y1 - 2014/7

N2 - In this paper we present a new stable algorithm for the parallel QR-decomposition of "tall and skinny" matrices. The algorithm has been developed for the dense symmetric eigensolver ELPA, where the QR-decomposition of tall and skinny matrices represents an important substep. Our new approach is based on the fast but unstable CholeskyQR algorithm (Stathopoulos and Wu, 2002) [1]. We show the stability of our new algorithm and provide promising results of our MPI-based implementation on a BlueGene/P and a Power6 system.

AB - In this paper we present a new stable algorithm for the parallel QR-decomposition of "tall and skinny" matrices. The algorithm has been developed for the dense symmetric eigensolver ELPA, where the QR-decomposition of tall and skinny matrices represents an important substep. Our new approach is based on the fast but unstable CholeskyQR algorithm (Stathopoulos and Wu, 2002) [1]. We show the stability of our new algorithm and provide promising results of our MPI-based implementation on a BlueGene/P and a Power6 system.

KW - Eigenvalue and eigenvector computation

KW - Parallelization

KW - QR-decomposition

KW - Two-step tridiagonalization

UR - https://www.scopus.com/pages/publications/84903375514

U2 - 10.1016/j.parco.2014.03.010

DO - 10.1016/j.parco.2014.03.010

M3 - Article

AN - SCOPUS:84903375514

SN - 0167-8191

VL - 40

SP - 186

EP - 194

JO - Parallel Computing

JF - Parallel Computing

IS - 7

ER -

A blocked QR-decomposition for the parallel symmetric eigenvalue problem

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this