Summary
The solution to many scientific and engineering problems requires the determination of the numerical rank of matrices. We present new parallel algorithms for computing rank-revealing QR (RRQR) factorizations of dense matrices on multicomputers, based on a serial approach developed by C. H. Bischof and G. Quintana-Ortí. The parallel implementations include the usual QR factorization with column pivoting, and a new faster approach that consists of two stages: a QR factorization with local column pivoting and a reliable rank-revealing algorithm appropriate for triangular matrices. Our parallel implementations include the BLAS-2 and BLAS-3 QR factorizations without pivoting since they are a good reference point, though they are not appropriate for rank-revealing purposes.
Experimental comparison shows considerable performance improvements of our new approach over classical rank-revealing algorithms on the platforms we used: an IBM SP2 platform and a cluster of SGI workstations.
We study the effect of the computer communication network and the processor computational power on the performance of the algorithms. In this case, as well as in many other parallel and distributed applications, the latency and bandwidth of the network are much more important than the processor computational power and, thus, these are the key factors impacting performance.
All authors were partially supported by the Spanish CICYT Project Grant TIC96-1062-C03-03.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Bibliography
E. Anderson, Z. Bai, C. H. Bischof, J. Demmel, J. Dongarra, J. DuCroz, A. Greenbaum, S. Hammarling, A. McKenney, S. Ostrouchov, and D. Sorensen, LAPACK User’s Guide Release 2.0, SIAM, Philadelphia, 1994.
C. H. Bischof, A block QR factorization algorithm using restricted pivoting in Proceedings SUPERCOMPUTING ‘89, Baltimore, Md., 1989, ACM Press, pp. 248–256.
C. H. Bischof, A parallel QR factorization algorithm with controlled local pivoting, SIAM J. on Scientific and Statistical Computing, 12 (1991), pp. 36–57.
C. H. Bischof and P. C. Hansen, Structure-Preserving and Rank’Revealing QR Factorizations, SIAM J. on Scientific and Statistical Computing, 12 (1989), pp. 1332–1350.
C. H. Bischof and G. Shroff, On updating signal subspaces, IEEE Trans, on Signal Processing, 40 (1992), pp. 96–105.
C. H. Bischof and P. C. Hansen, A Block Algorithm for Computing Rank- Revealing QR Factorizations, Numerical Algorithms, 2 (1992), pp. 371–392.
C. H. Bischof and P. T. P. Tang, A robust incremental condition scheme, Argonne Preprint MCS-P225–0391, Mathematics and Computer Science Division, Argonne National Laboratory, 1991.
C. H. Bischof and G. Quintana-Orti Computing rank-revealing QR factorizations of dense matrices, Argonne Preprint MCS-P559–0196, Mathematics and Computer Science Division, Argonne National Laboratory, 1996.
C. H. Bischof and G. Quintana-Orti, Codes for rank-revealing QR factorizations of dense matrices, Argonne Preprint MCS-P560–0196, Mathematics and Computer Science Division, Argonne National Laboratory, 1996.
S. F. Hsieh, J. R. Liu, and K. Yao, Comparisons of Truncated QR and SVD methods for AR spectral estimations, in Proceedings SVD and Signal Processing II, 1991, Elsevier Science Publishers, pp. 403–418.
T. F. Chan, Rank-Revealing QR Factorizations, Linear Algebra & Appl., 88 /89 (1987), pp. 67–82.
S. Chandrasekaran and I. Ipsen, On rank-revealing QR factorizations, SIAM J. on Matrix Analysis and Applications, 15 (1994), pp. 592–622.
M. Cosnard, J. M. Muller, and Y. Robert, Parallel QR decomposition of a rectangular matrix, Numerische Mathematik, 48 (1986), pp. 239–250.
J. J. Dongarra, J. R. Bunch, C. B. Moler, and G. W. Stewart, LIN- PACK Users’ Guide, SIAM Press, Philadelphia, 1979.
J. J. Dongarra, A. Sameh, and D. Sorensen, Implementation of some concurrent algorithms for matrix factorization, Parallel Computing, 3 (1986), pp. 25–34.
J. J. Dongarra, R. van de Geijn, and R. Whaley, Two dimensional basic linear algebra communication subprograms. Computer Science Dept. Technical Report CS-91–138, University of Tennessee, 1991 (LAPACK Working Note #37).
J. J. dongarra and d. Walker, The design of linear algebra libraries for high performance computers. Computer Science Dept. Technical Report CS-93–188, University of Tennessee, 1993 (LAPACK Working Note #58).
L. Elden and R. Schreiber, An application of systolic arrays to linear discrete ill-posed problems, SIAM J. on Scientific and Statistical Computing, 7 (1986), pp. 892–903.
L. Eldén, A parallel QR decomposition algorithm, Report LiTh Mat R 1988–02, Dept. of Math., Linköping University, Sweden, 1988.
G. H. Golub and C. F. van Loan, Matrix Computations, The Johns Hopkins University Press, Baltimore, 2nd ed., 1989.
G. H. Golub, Numerical methods for solving linear least squares problems, Numerische Mathematik, 7 (1965), pp. 206–216.
G. H. Golub, P. Manneback, and P. L. Toint, A comparison between some direct and iterative methods for certain large scale geodetic least-squares problem, SIAM J. on Scientific and Statistical Computing, 7 (1986), pp. 799–816.
T. A. Grandine, An iterative method for computing multivariate C 1 piecewise polynomial interpolants, Computer Aided Geometric Design, 4 (1987), pp. 307–319.
T. A. Grandine, Rank deficient interpolation and optimal design: An example, Tech. Report SCA-TR-113, Boeing Computer Services, Engineering and Scientific Services Division, February 1989.
W. Gropp, E. Lusk, and A. Skjellum, Using MPI: Portable parallel programming with the message-passing interface, The MIT Press, Cambridge — Massachusetts, 1994.
M. Gu and S. Eisenstat, An efficient algorithm for computing a strong rank-revealing factorization, Tech. Report YALEU/DCS/RR-967, Yale University, Department of Computer Science, 1994.
P. C. Hansen, Truncated SVD solutions to discrete ill-posed problems with ill-determined numerical rank, SIAM J. on Matrix Analysis and Applications, 11 (1990), pp. 503–518.
P. C. Hansen, S. Takishi, and S. Hiromoto, The Modified Truncated SVD- Method for Regularization in General Form, SIAM J. on Scientific and Statistical Computing, 13 (1991), pp. 1142–1150.
M. T. Heath and C. H. Romine. Parallel solution of triangular systems on distributed-memory multiprocessors. SIAM J. Scientific & Statistical Computing, 9, pp. 558–588, 1988.
G. Henry and R. van de Geijn. Parallelizing the QR algorithm for the unsymmetric algebraic eigenvalue problem: myths and reality. Lapack Working note #79, 1994.
I. C. F. Ipsen, Y. Saad and M. Schultz. Dense linear systems on a ring of processors. Linear Algebra and Its Appl., 77, pp. 205–239, 1986.
J. More, The Levenberg-Marquardt algorithm: Implementation and theory, in Proceedings of the Dundee Conference on Numerical Analysis, G. A. Watson (ed.), Berlin, 1978, Springer-Verlag.
D. P. O’Leary and G. W. Stewart. Assignment and scheduling in parallel matrix factorization. Linear Algebra and Its Appl., 77, pp. 275–300, 1986.
C.-T. Pan and P. T. P. Tang, Bounds on singular values revealed by QR factorization, Argonne Preprint MCS-P332–1092, Mathematics and Computer Science Division, Argonne National Laboratory, 1992.
A. Chtchelkanova, C. Edwards, J. Gunnels, G. Morrow, J. Overfelt, R. van de Geijn Towards Usable and Lean Parallel Linear Algebra Libraries PLAPACK Working Note #5, TR-96–09, Department of Computer Sciences, University of Texas, May 1996.
G. Quintana-Ortí, Algoritmos Secuenciales, por Bloques y Paralelos para el Cdlculo del Rango Numérico Matricial, Ph. D. Thesis, Universidad Politécnica de Valencia, 1995.
G. Quintana-Ortí, X. Sun, and C. H. Bischof, A BLAS-3 version of the QR factorization with column pivoting, Argonne Preprint MCS-P551–1295, Mathematics and Computer Science Division, Argonne National Laboratory, 1995.
G. Quintana-Ortí and E. S. Quintana-Ortí, Guaranteeing termination of Chandrasekaran & Ipsen’s algorithm for computing rank-revealing QR factorizations, Argonne Preprint MCS-P564–0196, Mathematics and Computer Science Division, Argonne National Laboratory, 1996.
J. Choi, J. J. Dongarra, D. W. Walker and R. C. Whaley, ScaLAPACK Reference Manual. Parallel Factorizations Routines (LU, QR, and Cholesky) and Parallel Reduction Routines (HRD, BRD, and TRD), Technical Memorandum ORNL/TM-12470, Oak Ridge National Laboratory, 1994, USA.
R. Schreiber and C. F. van Loan, A Storage Efficient WY Representation for Products of Householder Transformations, Scientific and Statistical Computing, 10 (1989), pp. 53–57.
G. W. Stewart, Introduction to Matrix Computations, Academic Press, New York, 1973.
B. Waldén, Using a Fast Signal Processor to Solve the Inverse Kinematic Problem with Special Emphasis on the Singularity Problem, Ph.D. Thesis, Linköping University, Dept. of Mathematics, 1991.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1997 Springer-Verlag London
About this paper
Cite this paper
Quintana-Ortí, G., Quintana-Ortí, E.S. (1997). Parallel Algorithms for Computing Rank-Revealing QR Factorizations. In: Cooperman, G., Michler, G., Vinck, H. (eds) Workshop on High Performance Computing and Gigabit Local Area Networks. Lecture Notes in Control and Information Sciences, vol 226. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3540761691_9
Download citation
DOI: https://doi.org/10.1007/3540761691_9
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-76169-3
Online ISBN: 978-3-540-40937-3
eBook Packages: Springer Book Archive