Abstract
This paper concerns accuracy-guaranteed numerical computations for linear systems. Due to the rapid progress of supercomputers, the treatable problem size is getting larger. The larger the problem size, the more rounding errors in floating-point arithmetic can accumulate in general, and the more inaccurate numerical solutions are obtained. Therefore, it is important to verify the accuracy of numerical solutions. Verified numerical computations are used to produce error bounds on numerical solutions. We report the implementation of a verification method for large-scale linear systems and some numerical results using the RIKEN K computer and the Fujitsu PRIMEHPC FX100, which show the high performance of the verified numerical computations.
Similar content being viewed by others
References
L. S. Blackford, J. Choi, A. Cleary, E. DAzevedo, J. Demmel, I. Dhillon, J. Dongarra, S. Hammarling, G. Henry, A. Petitet, K. Stanley, D. Walker, R. C. Whaley: ScaLAPACK — Scalable Linear Algebra PACKage. Available at http://www.netlib.org/scalapack/ (2019).
A. M. Castaldo, R. C. Whaley, A. T. Chronopoulos: Reducing floating point error in dot product using the superblock family of algorithms. SIAM J. Sci. Comput. 31 (2008), 1156–1174.
FUJITSU: FUJITSU Supercomputer PRIMEHPC FX100. Available at https://www.fujitsu.com/global/products/computing/servers/supercomputer/primehpc-fx100/ (2020).
N. J. Higham: Accuracy and Stability of Numerical Algorithms. Society for Industrial and Applied Mathematics, Philadelphia, 2002.
N. J. Higham, T. Mary: A new approach to probabilistic rounding error analysis. SIAM J. Sci. Comput. 41 (2019), A2815–A2835.
IEEE Computer Society: IEEE Standard for Floating-Point Arithmetic: IEEE Std 754–2008. IEEE, NewYork, 2008.
C.-P. Jeannerod, S. M. Rump: Improved error bounds for inner products in floating-point arithmetic. SIAM J. Matrix Anal. Appl. 34 (2013), 338–344.
M. Kolberg, G. Bohlender, L. G. Fernandes: An efficient approach to solve very large dense linear systems with verified computing on clusters. Numer. Linear Algebra Appl. 22 (2015), 299–316.
X. Li, J. Demmel, D. Bailey, Y. Hida, J. Iskandar, A. Kapur, M. Martin, B. Thompson, T. Tung, D. Yoo: XBLAS — Extra Precise Basic Linear Algebra Subroutines. Available at https://www.netlib.org/xblas/ (2008).
A. Minamihata, K. Sekine, T. Ogita, S. M. Rump, S. Oishi: Improved error bounds for linear systems with H-matrices. Nonlinear Theory Appl., IEICE 6 (2015), 377–382.
Y. Morikura, K. Ozaki, S. Oishi: Verification methods for linear systems using ufp estimation with rounding-to-nearest. Nonlinear Theory Appl., IEICE 4 (2013), 12–22.
M. Nakata: The MPACK: Multiple Precision Arithmetic BLAS (MBLAS) and LAPACK (MLAPACK). Available at http://mplapack.sourceforge.net/ (2011).
A. Neumaier: A simple derivation of the Hansen-Bliek-Rohn-Ning-Kearfott enclosure for linear interval equations. Reliab. Comput. 5 (1999), 131–136.
T. Ogita, S. Oishi: Fast verification for large-scale systems of linear equations. IPSJ Trans. 46 (2005), 10–18. (In Japanese.)
T. Ogita, S. Oishi, Y. Ushiro: Computation of sharp rigorous componentwise error bounds for the approximate solutions of systems of linear equations. Reliab. Comput. 9 (2003), 229–239.
T. Ogita, S. M. Rump, S. Oishi: Accurate sum and dot product. SIAM J. Sci. Comput. 26 (2005), 1955–1988.
T. Ogita, S. M. Rump, S. Oishi: Verified Solution of Linear Systems Without Directed Rounding: Technical Report No. 2005–04. Advanced Research Institute for Science and Engineering, Waseda University, Tokyo, 2005.
S. Oishi, S. M. Rump: Fast verification of solutions of matrix equations. Numer. Math. 90 (2002), 755–773.
K. Ozaki, T. Ogita: Generation of linear systems with specified solutions for numerical experiments. Reliab. Comput. 25 (2017), 148–167.
K. Ozaki, T. Ogita, S. Miyajima, S. Oishi, S. M. Rump: A method of obtaining verified solutions for linear systems suited for Java. J. Comput. Appl. Math. 199 (2007), 337–344.
K. Ozaki, T. Ogita, S. Oishi: An algorithm for automatically selecting a suitable verification method for linear systems. Numer. Algorithms 56 (2011), 363–382.
A. Petitet: PBLAS — Parallel Basic Linear Algebra Subprograms. Available at http://www.netlib.org/scalapack/pblas_qref.html.
RIKEN Center for Computational Science: What is K? Available at https://www.r-ccs.riken.jp/en/k-computer/about/ (2019).
S. M. Rump: Kleine Fehlerschranken bei Matrixproblemen: PhD Thesis. Universität Karlsruhe, Karlsruhe, 1980. (In German.)
S. M. Rump: Accurate solution of dense linear systems I: Algorithms in rounding to nearest. J. Comput. Appl. Math. 242 (2013), 157–184.
S. M. Rump: Accurate solution of dense linear systems II: Algorithms using directed rounding. J. Comput. Appl. Math. 242 (2013), 185–212.
R. D. Skeel: Iterative refinement implies numerical stability for Gaussian elimination. Math. Comput. 35 (1980), 817–832.
V. Strassen: Gaussian elimination is not optimal. Numer. Math. 13 (1969), 354–356.
N. Yamanaka, T. Ogita, S. M. Rump, S. Oishi: A parallel algorithm for accurate dot product. Parallel Comput. 34 (2008), 392–410.
Acknowledgments
The authors wish to thank the anonymous reviewers for comments on earlier version of this paper and suggestions for future work. We sincerely express our thank to Fujitsu Limited for developing the function of switches of rounding modes and giving us the fruitful information of BLAS functions. Thanks to Mr. Ryota Ochiai, Mr. Atsushi Sakamoto, and Mr. Ryota Kobayashi, former students in Shibaura Institute of Technology for the assistance of coding and numerical tests.
Author information
Authors and Affiliations
Corresponding author
Additional information
This research was supported by MEXT as “Exploratory Challenge on Post-K computer” (Development of verified numerical computations and super high-performance computing environment for extreme researches) using computational resources of the K computer at RIKEN R-CCS and Fujitsu FX100 at Nagoya University through the HPCI System Research Project (Project ID: hp190192).
Rights and permissions
About this article
Cite this article
Ozaki, K., Terao, T., Ogita, T. et al. Verified Numerical Computations for Large-Scale Linear Systems. Appl Math 66, 269–285 (2021). https://doi.org/10.21136/AM.2021.0318-19
Received:
Published:
Issue Date:
DOI: https://doi.org/10.21136/AM.2021.0318-19
Keywords
- verified numerical computation
- floating-point arithmetic
- high-performance computing
- large-scale linear system