Incremental condition estimator for parallel sparse matrix factorization

Jesse Louis Barlow, U. B. Vemulapati

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

There is often a trade-off between preserving sparsity and numerical stability in sparse matrix factorizations. In applications like the direct solution of Equality Constrained Least Squares problem, the accurate detection of the rank of a large and sparse constraint matrix is a key issue. Column pivoting is not suitable for distributed memory machines because it forces the program into a lock-step mode, preventing any overlapping of computations. So factorization algorithms on such machines need to use a reliable, yet inexpensive incremental condition estimator to decide on which columns to be included. We describe an incremental condition estimator that can be used during a sparse QR factorization. We show that it is quite reliable and is well suited for use on parallel machines. We supply experimental results to support its effectiveness as well as suitability for parallel architectures.

Original languageEnglish (US)
Title of host publicationApplications
EditorsDavid W. Walker, Quentin F. Stout
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages322-327
Number of pages6
ISBN (Electronic)0818621133, 9780818621130
DOIs
StatePublished - Jan 1 1990
Event5th Distributed Memory Computing Conference, DMCC 1990 - Charleston, United States
Duration: Apr 8 1990Apr 12 1990

Publication series

NameProceedings of the 5th Distributed Memory Computing Conference, DMCC 1990
Volume1

Conference

Conference5th Distributed Memory Computing Conference, DMCC 1990
CountryUnited States
CityCharleston
Period4/8/904/12/90

Fingerprint

Factorization
Parallel architectures
Convergence of numerical methods
Data storage equipment
Incremental
Estimator
Matrix factorization
Overlapping
Trade-offs
Parallel machines
Least squares
Equality

All Science Journal Classification (ASJC) codes

  • Hardware and Architecture
  • Information Systems and Management
  • Computer Networks and Communications

Cite this

Barlow, J. L., & Vemulapati, U. B. (1990). Incremental condition estimator for parallel sparse matrix factorization. In D. W. Walker, & Q. F. Stout (Eds.), Applications (pp. 322-327). [555401] (Proceedings of the 5th Distributed Memory Computing Conference, DMCC 1990; Vol. 1). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/DMCC.1990.555401
Barlow, Jesse Louis ; Vemulapati, U. B. / Incremental condition estimator for parallel sparse matrix factorization. Applications. editor / David W. Walker ; Quentin F. Stout. Institute of Electrical and Electronics Engineers Inc., 1990. pp. 322-327 (Proceedings of the 5th Distributed Memory Computing Conference, DMCC 1990).
@inproceedings{44cd17431e2b49c18f2a95f7c1c082df,
title = "Incremental condition estimator for parallel sparse matrix factorization",
abstract = "There is often a trade-off between preserving sparsity and numerical stability in sparse matrix factorizations. In applications like the direct solution of Equality Constrained Least Squares problem, the accurate detection of the rank of a large and sparse constraint matrix is a key issue. Column pivoting is not suitable for distributed memory machines because it forces the program into a lock-step mode, preventing any overlapping of computations. So factorization algorithms on such machines need to use a reliable, yet inexpensive incremental condition estimator to decide on which columns to be included. We describe an incremental condition estimator that can be used during a sparse QR factorization. We show that it is quite reliable and is well suited for use on parallel machines. We supply experimental results to support its effectiveness as well as suitability for parallel architectures.",
author = "Barlow, {Jesse Louis} and Vemulapati, {U. B.}",
year = "1990",
month = "1",
day = "1",
doi = "10.1109/DMCC.1990.555401",
language = "English (US)",
series = "Proceedings of the 5th Distributed Memory Computing Conference, DMCC 1990",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "322--327",
editor = "Walker, {David W.} and Stout, {Quentin F.}",
booktitle = "Applications",
address = "United States",

}

Barlow, JL & Vemulapati, UB 1990, Incremental condition estimator for parallel sparse matrix factorization. in DW Walker & QF Stout (eds), Applications., 555401, Proceedings of the 5th Distributed Memory Computing Conference, DMCC 1990, vol. 1, Institute of Electrical and Electronics Engineers Inc., pp. 322-327, 5th Distributed Memory Computing Conference, DMCC 1990, Charleston, United States, 4/8/90. https://doi.org/10.1109/DMCC.1990.555401

Incremental condition estimator for parallel sparse matrix factorization. / Barlow, Jesse Louis; Vemulapati, U. B.

Applications. ed. / David W. Walker; Quentin F. Stout. Institute of Electrical and Electronics Engineers Inc., 1990. p. 322-327 555401 (Proceedings of the 5th Distributed Memory Computing Conference, DMCC 1990; Vol. 1).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Incremental condition estimator for parallel sparse matrix factorization

AU - Barlow, Jesse Louis

AU - Vemulapati, U. B.

PY - 1990/1/1

Y1 - 1990/1/1

N2 - There is often a trade-off between preserving sparsity and numerical stability in sparse matrix factorizations. In applications like the direct solution of Equality Constrained Least Squares problem, the accurate detection of the rank of a large and sparse constraint matrix is a key issue. Column pivoting is not suitable for distributed memory machines because it forces the program into a lock-step mode, preventing any overlapping of computations. So factorization algorithms on such machines need to use a reliable, yet inexpensive incremental condition estimator to decide on which columns to be included. We describe an incremental condition estimator that can be used during a sparse QR factorization. We show that it is quite reliable and is well suited for use on parallel machines. We supply experimental results to support its effectiveness as well as suitability for parallel architectures.

AB - There is often a trade-off between preserving sparsity and numerical stability in sparse matrix factorizations. In applications like the direct solution of Equality Constrained Least Squares problem, the accurate detection of the rank of a large and sparse constraint matrix is a key issue. Column pivoting is not suitable for distributed memory machines because it forces the program into a lock-step mode, preventing any overlapping of computations. So factorization algorithms on such machines need to use a reliable, yet inexpensive incremental condition estimator to decide on which columns to be included. We describe an incremental condition estimator that can be used during a sparse QR factorization. We show that it is quite reliable and is well suited for use on parallel machines. We supply experimental results to support its effectiveness as well as suitability for parallel architectures.

UR - http://www.scopus.com/inward/record.url?scp=85067895714&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85067895714&partnerID=8YFLogxK

U2 - 10.1109/DMCC.1990.555401

DO - 10.1109/DMCC.1990.555401

M3 - Conference contribution

AN - SCOPUS:85067895714

T3 - Proceedings of the 5th Distributed Memory Computing Conference, DMCC 1990

SP - 322

EP - 327

BT - Applications

A2 - Walker, David W.

A2 - Stout, Quentin F.

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Barlow JL, Vemulapati UB. Incremental condition estimator for parallel sparse matrix factorization. In Walker DW, Stout QF, editors, Applications. Institute of Electrical and Electronics Engineers Inc. 1990. p. 322-327. 555401. (Proceedings of the 5th Distributed Memory Computing Conference, DMCC 1990). https://doi.org/10.1109/DMCC.1990.555401