The DOI matches its Crossref record; 42 of the 42 references it cites were found in a registry or an index.
Minimizing Communication in Numerical Linear Algebra
Open accessAbstract
In 1981 Hong and Kung proved a lower bound on the amount of communication (amount of data moved between a small, fast memory and large, slow memory) needed to perform dense, [Formula: see text]-by-[Formula: see text] matrix multiplication using the conventional [Formula: see text] algorithm, where the input matrices were too large to fit in the small, fast memory. In 2004 Irony, Toledo, and Tiskin gave a new proof of this result and extended it to the parallel case (where communication means the amount of data moved between processors). In both cases the lower bound may be expressed as [Formula: see text], where [Formula: see text] is the size of the fast memory (or local memory in the parallel case). Here we generalize these results to a much wider variety of algorithms, including LU factorization, Cholesky factorization, [Formula: see text] factorization, QR factorization, the Gram–Schmidt algorithm, and algorithms for eigenvalues and singular values, i.e., essentially all direct methods of linear algebra. The proof works for dense or sparse matrices and for sequential or parallel algorithms. In addition to lower bounds on the amount of data moved (bandwidth cost), we get lower bounds on the number of messages required to move it (latency cost). We extend our lower bound technique to compositions of linear algebra operations (like computing powers of a matrix) to decide whether it is enough to call a sequence of simpler optimal algorithms (like matrix multiplication) to minimize communication, or whether we can do better. We give examples of both. We also show how to extend our lower bounds to certain graph-theoretic problems. We point out recently designed algorithms that attain many of these lower bounds.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Golub, G. (2012) · doi:10.56021/9781421407944 · Cited by 34,505
Rayward-Smith, V. J. (1991) · Journal of the Operational Research Society · doi:10.1057/jors.1991.155 · Cited by 16,821
Iterative Methods for Sparse Linear Systems
Saad, Y. (2003) · Society for Industrial and Applied Mathematics eBooks · doi:10.1137/1.9780898718003 · Cited by 13,790
Applied Numerical Linear Algebra
Demmel, J. W. (1997) · Society for Industrial and Applied Mathematics eBooks · doi:10.1137/1.9781611971446 · Cited by 2,299
Self-citationApplied numerical linear algebra
(1998) · Choice Reviews Online · doi:10.5860/choice.35-3919 · Cited by 2,170
Introduction to algorithms [2nd ed.]
Cormen, T. H. (2001) · Cited by 1,911
Nested Dissection of a Regular Finite Element Mesh
George, A. (1973) · SIAM Journal on Numerical Analysis · doi:10.1137/0710032 · Cited by 1,064
An updated set of basic linear algebra subprograms (BLAS)
(2002) · ACM Transactions on Mathematical Software · doi:10.1145/567806.567807 · Cited by 834
Frigo, M., Leiserson, C. E., Prokop, H., & Ramachandran, S. (2003) · 40th Annual Symposium on Foundations of Computer Science (Cat. No.99CB37039) · doi:10.1109/sffcs.1999.814600 · Cited by 803
A class of parallel tiled linear algebra algorithms for multicore architectures
Buttari, A., Langou, J., Kurzak, J., & Dongarra, J. (2009) · Parallel Computing · doi:10.1016/j.parco.2008.10.002 · Cited by 529
OSKI: A library of automatically tuned sparse matrix kernels
Vuduc, R., Demmel, J. W., & Yelick, K. A. (2005) · Journal of Physics: Conference Series · doi:10.1088/1742-6596/16/1/071 · Cited by 490
Self-citationI/O complexity: The red-blue pebble game
Hong, J., & Kung, H. T. (1981) · ACM Symposium on Theory of Computing (STOC) · doi:10.1145/800076.802486 · Cited by 474
Communication-optimal Parallel and Sequential QR and LU Factorizations
Demmel, J., Grigori, L., Hoemmen, M., & Langou, J. (2012) · SIAM Journal on Scientific Computing · doi:10.1137/080731992 · Cited by 407
Self-citationA cellular computer to implement the kalman filter algorithm
Cannon, L. E. (1969) · Montana State University ScholarWorks (Montana State University) · Cited by 394
A Storage-Efficient $WY$ Representation for Products of Householder Transformations
Schreiber, R., & Van Loan, C. (1989) · SIAM Journal on Scientific and Statistical Computing · doi:10.1137/0910005 · Cited by 313
The WY representation for products of householder matrices
Bischof, C., & Van Loan, C. (1987) · SIAM Journal on Scientific and Statistical Computing · doi:10.1137/0908009 · Cited by 295
Fast sparse matrix multiplication
Yuster, R., & Zwick, U. (2005) · ACM Transactions on Algorithms · doi:10.1145/1077464.1077466 · Cited by 251
Communication lower bounds for distributed-memory matrix multiplication
Irony, D., Toledo, S., & Tiskin, A. (2004) · Journal of Parallel and Distributed Computing · doi:10.1016/j.jpdc.2004.03.021 · Cited by 250
Gunnels, J. A., Gustavson, F. G., Henry, G., & Geijn, R. A. (2001) · ACM Transactions on Mathematical Software · doi:10.1145/504210.504213 · Cited by 235
Recursion leads to automatic variable blocking for dense linear-algebra algorithms
Gustavson, F. G. (1997) · IBM Journal of Research and Development · doi:10.1147/rd.416.0737 · Cited by 229
Demmel, J., Dumitriu, I., & Holtz, O. (2007) · Numerische Mathematik · doi:10.1007/s00211-007-0114-x · Cited by 169
Self-citationLocality of Reference in LU Decomposition with Partial Pivoting
Toledo, S. (1997) · SIAM Journal on Matrix Analysis and Applications · doi:10.1137/s0895479896297744 · Cited by 163
Recursive Blocked Algorithms and Hybrid Data Structures for Dense Matrix Library Software
Elmroth, E., Gustavson, F., Jonsson, I., & Kågström, B. (2004) · SIAM Review · doi:10.1137/s0036144503428693 · Cited by 161
A three-dimensional approach to parallel matrix multiplication
Agarwal, R. C., Balle, S. M., Gustavson, F. G., Joshi, M., & Palkar, P. (1995) · IBM Journal of Research and Development · doi:10.1147/rd.395.0575 · Cited by 158
Demmel, J. W., Dumitriu, I., & Holtz, O. (2007) · Cited by 143
Self-citationThe analysis of a nested dissection algorithm
Gilbert, J. R., & Tarjan, R. E. (1986) · Numerische Mathematik · doi:10.1007/bf01396660 · Cited by 134
Orthogonal Eigenvectors and Relative Gaps
Dhillon, I. S., & Parlett, B. Ν. (2003) · SIAM Journal on Matrix Analysis and Applications · doi:10.1137/s0895479800370111 · Cited by 121
Applying recursion to serial and parallel QR factorization leads to better performance
Elmroth, E., & Gustavson, F. G. (2000) · IBM Journal of Research and Development · doi:10.1147/rd.444.0605 · Cited by 114
Parallel out-of-core computation and updating of the QR factorization
Gunter, B. C., & Geijn, R. A. (2005) · ACM Transactions on Mathematical Software · doi:10.1145/1055531.1055534 · Cited by 99
The Multishift QR Algorithm. Part I: Maintaining Well-Focused Shifts and Level 3 Performance
Braman, K. S., Byers, R., & Mathias, R. (2002) · SIAM Journal on Matrix Analysis and Applications · doi:10.1137/s0895479801384573 · Cited by 96
Algorithm 807: The SBR Toolbox—software for successive band reduction
Bischof, C. H., Lang, B., & Sun, X. (2000) · ACM Transactions on Mathematical Software · doi:10.1145/365723.365736 · Cited by 92
Memory-Efficient Matrix Multiplication in the BSP Model
McColl, W. F., & Tiskin, A. (1999) · Algorithmica · doi:10.1007/pl00008264 · Cited by 92
Graph expansion and communication costs of fast matrix multiplication
Ballard, G., Demmel, J. W., Holtz, O., & Schwartz, O. (2012) · Journal of the ACM · doi:10.1145/2395116.2395121 · Cited by 90
Self-citationCALU: A Communication Optimal LU Factorization Algorithm
Grigori, L., Demmel, J. W., & Xiang, H. (2011) · SIAM Journal on Matrix Analysis and Applications · doi:10.1137/100788926 · Cited by 89
Self-citationA framework for symmetric band reduction
Bischof, C. H., Lang, B., & Sun, X. (2000) · ACM Transactions on Mathematical Software · doi:10.1145/365723.365735 · Cited by 78
A recursive formulation of Cholesky factorization of a matrix in packed storage
Andersen, B. S., Waśniewski, J., & Gustavson, F. G. (2001) · ACM Transactions on Mathematical Software · doi:10.1145/383738.383741 · Cited by 64
Communication-optimal Parallel and Sequential Cholesky Decomposition
Ballard, G., Demmel, J., Holtz, O., & Schwartz, O. (2010) · SIAM Journal on Scientific Computing · doi:10.1137/090760969 · Cited by 57
Self-citationNew serial and parallel recursive QR factorization algorithms for SMP systems
Elmroth, E., & Gustavson, F. G. (1998) · Lecture notes in computer science · doi:10.1007/bfb0095328 · Cited by 50
Modification of the Householder Method Based on the Compact WY Representation
Puglisi, C. (1992) · SIAM Journal on Scientific and Statistical Computing · doi:10.1137/0913042 · Cited by 50
QR factorization with Morton-ordered quadtree matrices for memory re-use and parallelism
Frens, J. D., & Wise, D. S. (2003) · ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming · doi:10.1145/781498.781525 · Cited by 36
Optimal sparse matrix dense vector multiplication in the I/O-model
Bender, M. A., Brodal, G. S., Fagerberg, R., Jacob, R., & Vicari, E. (2007) · ACM Symposium on Parallelism in Algorithms and Architectures (SPAA) · doi:10.1145/1248377.1248391 · Cited by 31
Kumar, P. (2003) · Cited by 27
Cited by
Towards dense linear algebra for hybrid GPU accelerated manycore systems
Tomov, S., Dongarra, J., & Baboulin, M. (2010) · Parallel Computing · doi:10.1016/j.parco.2009.12.005 · Cited by 451
Communication-optimal Parallel and Sequential QR and LU Factorizations
Demmel, J., Grigori, L., Hoemmen, M., & Langou, J. (2012) · SIAM Journal on Scientific Computing · doi:10.1137/080731992 · Cited by 407
Self-citation
Metadata
- Authors
- Ballard, G., Demmel, J., Holtz, O., & Schwartz, O.
- Venue
- SIAM Journal on Matrix Analysis and Applications
- Published
- 2011-07
- Volume
- 32
- Issue
- 3
- Pages
- 866-901
- ISSN
- 0895-4798 (ISSN-L)
- 1095-7162
- OpenAlex
- W2104373803
- Cited by
- 276
- References
- 42
- FWCI
- 28.70
- Type
- journal article
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.