Abstract
The article considers the problem of low-rank column and cross (\(CGR\), \(CUR\)) approximation of matrices in the Frobenius norm up to a fixed factor \(1 + \varepsilon \). It is proved that, for random matrices, in average, an estimate of the form \(1 + \varepsilon \leqslant \tfrac{{m + 1}}{{m - r + 1}}\tfrac{{n + 1}}{{n - r + 1}}\) holds, where \(m\) and \(n\) are the number of rows and columns of the cross approximation. Thus, matrices for which the maximum volume principle cannot guarantee high accuracy are quite rare. A connection of the estimates obtained with the methods for finding the submatrix of the maximum volume and the maximum projective volume is also considered. Numerical experiments show the closeness of theoretical estimates and practical results of fast cross approximation.
Similar content being viewed by others
REFERENCES
N. Halko, P. Martinsson, and J. Tropp, “Finding structure with randomness: Probabilistic algorithms for constructing approximate matrix decompositions,” SIAM Rev. 53, 217–288 (2011).
A. Civril and M. Magdon-Ismail, “Column subset selection via sparse approximation of SVD,” Theor. Comput. Sci. 412, 1–14 (2012).
C. Boutsidis and D. P. Woodruff, “Optimal CUR matrix decompositions,” SIAM J. Comput. 46, 543–589 (2017).
S. A. Goreinov and E. E. Tyrtyshnikov, “The maximal-volume concept in approximation by low-rank matrices,” Contemp. Math. 268, 47–51 (2001).
A. I. Osinsky and N. L. Zamarashkin, “Pseudo-skeleton approximations with better accuracy estimates,” Linear Algebra Appl. 537, 221–249 (2018).
A. Deshpande and S. Vempala, “Adaptive sampling and fast low-rank matrix approximation,” Lect. Notes Comput. Sci. 1, 292–303 (2006).
V. Guruswami and A. K. Sinop, “Optimal column-based low-rank matrix reconstruction,” (2012). arXiv:1104.1732.
N. L. Zamarashkin and A. I. Osinsky, “On the existence of a nearly optimal skeleton approximation of a matrix in the Frobenius norm,” Dokl. Math. 97 (2), 164–166 (2018).
A. Deshpande and L. Rademacher, “Efficient volume sampling for row/column subset selection,” IEEE 51st Annual Symposium on Foundations of Computer Science (2010), pp. 329–338.
S. A. Goreinov and E. E. Tyrtyshnikov, “Quasioptimality of skeleton approximation of a matrix in the Chebyshev norm,” Dokl. Math. 83 (3), 374–375 (2011).
S. A. Goreinov, E. E. Tyrtyshnikov, and N. L. Zamarashkin, “A theory of pseudo-skeleton approximations,” Linear Algebra Appl. 261, 1–21 (1997).
A. Y. Michalev and I. V. Oseledets, “Rectangular maximum-volume submatrices and their applications,” Linear Algebra Appl. 538, 187–211 (2018).
C. T. Pan, “On the existence and computation of rank revealing LU factorizations,” Linear Algebra Appl. 316, 199–222 (2000).
M. Gu and S. C. Eisenstat, “Efficient algorithms for computing a strong rank-revealing QR factorization,” SIAM J. Sci. Comput. 17 (4), 848–869 (1996).
S. A. Goreinov, I. V. Oseledets, D. V. Savostyanov, E. E. Tyrtyshnikov, and N. L. Zamarashkin, “How to find a good submatrix,” in Matrix Methods: Theory, Algorithms, Applications, Ed. by V. Olshevsky and E. Tyrtyshnikov (World Scientific, Singapore, 2010), pp. 247–256.
A. I. Osinsky, “Rectangular maximum volume and projective volume search algorithms,” (2018). arXiv:1809.02334.
Funding
This work was supported by the Department of the Moscow Center for Fundamental and Applied Mathematics at the Institute of Numerical Mathematics, Russian Academy of Sciences (agreement no. 075-15-2019-1624 with the Ministry of Education and Science of the Russian Federation).
Author information
Authors and Affiliations
Corresponding authors
Additional information
Translated by E. Chernokozhin
Rights and permissions
About this article
Cite this article
Zamarashkin, N.L., Osinsky, A.I. On the Accuracy of Cross and Column Low-Rank Maxvol Approximations in Average. Comput. Math. and Math. Phys. 61, 786–798 (2021). https://doi.org/10.1134/S0965542521050171
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1134/S0965542521050171