Communication-avoiding symmetric-indefinite factorization
Abstract
We describe and analyze a novel symmetric triangular factorization algorithm. The algorithm is essentially a block version of Aasen's triangular tridiagonalization. It factors a dense symmetric matrix A as the product A=PLTLTPT where P is a permutation matrix, L is lower triangular, and T is block tridiagonal and banded. The algorithm is the first symmetric-indefinite communication-avoiding factorization: it performs an asymptotically optimal amount of communication in a two-level memory hierarchy for almost any cache-line size. Adaptations of the algorithm to parallel computers are likely to be communication efficient as well; one such adaptation has been recently published. As a result, the current paper describes the algorithm, proves that it is numerically stable, and proves that it is communication optimal.
- Authors:
-
- Sandia National Lab. (SNL-CA), Livermore, CA (United States)
- Univ. of Tennessee, Knoxville, TN (United States)
- Univ. of California, Berkeley, CA (United States)
- Univ. of Tennessee, Knoxville, TN (United States); Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States); Univ. of Manchester (United Kingdom)
- Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United States)
- Tel Aviv Univ., Tel Aviv (Israel)
- Publication Date:
- Research Org.:
- Sandia National Lab. (SNL-CA), Livermore, CA (United States)
- Sponsoring Org.:
- USDOE National Nuclear Security Administration (NNSA)
- OSTI Identifier:
- 1237467
- Report Number(s):
- SAND-2015-1851J
Journal ID: ISSN 0895-4798; 579666
- Grant/Contract Number:
- AC04-94AL85000
- Resource Type:
- Accepted Manuscript
- Journal Name:
- SIAM Journal on Matrix Analysis and Applications
- Additional Journal Information:
- Journal Volume: 35; Journal Issue: 4; Journal ID: ISSN 0895-4798
- Publisher:
- SIAM
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 97 MATHEMATICS AND COMPUTING; symmetric-indefinite matrices; communication-avoiding algorithms; Aasen's factorization
Citation Formats
Ballard, Grey Malone, Becker, Dulcenia, Demmel, James, Dongarra, Jack, Druinsky, Alex, Peled, Inon, Schwartz, Oded, Toledo, Sivan, and Yamazaki, Ichitaro. Communication-avoiding symmetric-indefinite factorization. United States: N. p., 2014.
Web. doi:10.1137/130929060.
Ballard, Grey Malone, Becker, Dulcenia, Demmel, James, Dongarra, Jack, Druinsky, Alex, Peled, Inon, Schwartz, Oded, Toledo, Sivan, & Yamazaki, Ichitaro. Communication-avoiding symmetric-indefinite factorization. United States. https://doi.org/10.1137/130929060
Ballard, Grey Malone, Becker, Dulcenia, Demmel, James, Dongarra, Jack, Druinsky, Alex, Peled, Inon, Schwartz, Oded, Toledo, Sivan, and Yamazaki, Ichitaro. Thu .
"Communication-avoiding symmetric-indefinite factorization". United States. https://doi.org/10.1137/130929060. https://www.osti.gov/servlets/purl/1237467.
@article{osti_1237467,
title = {Communication-avoiding symmetric-indefinite factorization},
author = {Ballard, Grey Malone and Becker, Dulcenia and Demmel, James and Dongarra, Jack and Druinsky, Alex and Peled, Inon and Schwartz, Oded and Toledo, Sivan and Yamazaki, Ichitaro},
abstractNote = {We describe and analyze a novel symmetric triangular factorization algorithm. The algorithm is essentially a block version of Aasen's triangular tridiagonalization. It factors a dense symmetric matrix A as the product A=PLTLTPT where P is a permutation matrix, L is lower triangular, and T is block tridiagonal and banded. The algorithm is the first symmetric-indefinite communication-avoiding factorization: it performs an asymptotically optimal amount of communication in a two-level memory hierarchy for almost any cache-line size. Adaptations of the algorithm to parallel computers are likely to be communication efficient as well; one such adaptation has been recently published. As a result, the current paper describes the algorithm, proves that it is numerically stable, and proves that it is communication optimal.},
doi = {10.1137/130929060},
journal = {SIAM Journal on Matrix Analysis and Applications},
number = 4,
volume = 35,
place = {United States},
year = {Thu Nov 13 00:00:00 EST 2014},
month = {Thu Nov 13 00:00:00 EST 2014}
}
Web of Science
Works referenced in this record:
On the reduction of a symmetric matrix to tridiagonal form
journal, September 1971
- Aasen, Jan Ole
- BIT, Vol. 11, Issue 3
Minimizing Communication in Numerical Linear Algebra
journal, July 2011
- Ballard, Grey; Demmel, James; Holtz, Olga
- SIAM Journal on Matrix Analysis and Applications, Vol. 32, Issue 3
Some stable methods for calculating inertia and solving symmetric linear systems
journal, January 1977
- Bunch, James R.; Kaufman, Linda
- Mathematics of Computation, Vol. 31, Issue 137
LogP: a practical model of parallel computation
journal, November 1996
- Culler, David E.; Karp, Richard M.; Patterson, David
- Communications of the ACM, Vol. 39, Issue 11
A set of level 3 basic linear algebra subprograms
journal, March 1990
- Dongarra, J. J.; Du Croz, Jeremy; Hammarling, Sven
- ACM Transactions on Mathematical Software, Vol. 16, Issue 1
CALU: A Communication Optimal LU Factorization Algorithm
journal, October 2011
- Grigori, Laura; Demmel, James W.; Xiang, Hua
- SIAM Journal on Matrix Analysis and Applications, Vol. 32, Issue 4
Recursion leads to automatic variable blocking for dense linear-algebra algorithms
journal, November 1997
- Gustavson, F. G.
- IBM Journal of Research and Development, Vol. 41, Issue 6
Stability of the Diagonal Pivoting Method with Partial Pivoting
journal, January 1997
- Higham, Nicholas J.
- SIAM Journal on Matrix Analysis and Applications, Vol. 18, Issue 1
The Snap-Back Pivoting Method for Symmetric Banded Indefinite Matrices
journal, January 2006
- Irony, Dror; Toledo, Sivan
- SIAM Journal on Matrix Analysis and Applications, Vol. 28, Issue 2
Communication lower bounds for distributed-memory matrix multiplication
journal, September 2004
- Irony, Dror; Toledo, Sivan; Tiskin, Alexander
- Journal of Parallel and Distributed Computing, Vol. 64, Issue 9
The retraction algorithm for factoring banded symmetric matrices
journal, January 2007
- Kaufman, Linda
- Numerical Linear Algebra with Applications, Vol. 14, Issue 3
LU Factorization with Panel Rank Revealing Pivoting and Its Communication Avoiding Version
journal, January 2013
- Khabou, Amal; Demmel, James W.; Grigori, Laura
- SIAM Journal on Matrix Analysis and Applications, Vol. 34, Issue 3
Reduction of the symmetric eigenproblemAx=λBx and related problems to standard form
journal, February 1968
- Martin, R. S.; Wilkinson, J. H.
- Numerische Mathematik, Vol. 11, Issue 2
Elemental: A New Framework for Distributed Memory Dense Matrix Computations
journal, February 2013
- Poulson, Jack; Marker, Bryan; van de Geijn, Robert A.
- ACM Transactions on Mathematical Software, Vol. 39, Issue 2
Partial factorization of a dense symmetric indefinite matrix
journal, December 2011
- Reid, John K.; Scott, Jennifer A.
- ACM Transactions on Mathematical Software, Vol. 38, Issue 2
Partitioned Triangular Tridiagonalization
journal, February 2011
- Rozložník, Miroslav; Shklarski, Gil; Toledo, Sivan
- ACM Transactions on Mathematical Software, Vol. 37, Issue 4
Smoothed Analysis of the Condition Numbers and Growth Factors of Matrices
journal, January 2006
- Sankar, Arvind; Spielman, Daniel A.; Teng, Shang-Hua
- SIAM Journal on Matrix Analysis and Applications, Vol. 28, Issue 2
Locality of Reference in LU Decomposition with Partial Pivoting
journal, October 1997
- Toledo, Sivan
- SIAM Journal on Matrix Analysis and Applications, Vol. 18, Issue 4
Average-Case Stability of Gaussian Elimination
journal, July 1990
- Trefethen, Lloyd N.; Schreiber, Robert S.
- SIAM Journal on Matrix Analysis and Applications, Vol. 11, Issue 3
A bridging model for parallel computation
journal, August 1990
- Valiant, Leslie G.
- Communications of the ACM, Vol. 33, Issue 8
Algorithms for parallel memory, I: Two-level memories
journal, September 1994
- Vitter, J. S.; Shriver, E. A. M.
- Algorithmica, Vol. 12, Issue 2-3
Works referencing / citing this record:
Non-GPU-resident symmetric indefinite factorization: Non-GPU-resident Dense Symmetric Indefinite Factorization
journal, November 2016
- Yamazaki, Ichitaro; Tomov, Stanimire; Dongarra, Jack
- Concurrency and Computation: Practice and Experience, Vol. 29, Issue 5