Abstract
A new decomposition of a matrix triplet (A, B, C) corresponding to the singular value decomposition of the matrix productABC is developed in this paper, which will be termed theProduct-Product Singular Value Decomposition (PPSVD). An orthogonal variant of the decomposition which is more suitable for the purpose of numerical computation is also proposed. Some geometric and algebraic issues of the PPSVD, such as the variational and geometric interpretations, and uniqueness properties are discussed. A numerical algorithm for stably computing the PPSVD is given based on the implicit Kogbetliantz technique. A numerical example is outlined to demonstrate the accuracy of the proposed algorithm.
Similar content being viewed by others
References
A. Bojanczyk, L. M. Ewerbring, F. T. Luk, P. van Dooren,An algorithm for the singular value decomposition of a matrix product. Proc. SPIE Vol. 1348,Advanced Signal Processing Algorithms, Architectures and Implementations, (1990), 382–393.
J. de Leeuw,Fixed rank matrix approximation with singular weights matrices, CQS-Computational Statistics Quarterly 1 (1984), 3–12.
B. de Moor,On the structure and geometry of the product singular value decomposition, Numerical Analysis Project, Manuscript NA-90-05, Computer Science Dept. Stanford Univ.
B. de Moor, H. Zha,A tree of generalizations of the ordinary singular value decomposition, Lin. Alg. Appli. Vol. 147, (1991) 469–500.
L. M. Ewerbring,A new generalization of the singular value decomposition: algorithms and applications, Ph.D. dissertation, School of Electrical Engineering, Cornell University, Ithaca, New York, 1989.
L. M. Ewerbring, F. T. Luk,Canonical correlations and generalized SVD, applications and new algorithms, J. Comput. Appl. Math. 27 (1989) 37–52.
K. V. Fernando and S. Hammarling:A product induced singular value decomposition (PSVD) for two matrices and balanced realization. Proceedings of the conference on linear algebra in signals, systems and control, SIAM Philadelphia (1988) 128–140.
G. H. Golub, C. Reinsch,Singular value decomposition and least squares solutions, Numer. Math. 14 (1970) 403–420.
G. H. Golub, C. F. van Loan,Matrix Computation. The Johns Hopkins University Press, Baltimore, USA (1983).
M. T. Heath, A. J. Laub, C. C. Paige, R. C. Ward,Computing the singular value decomposition of a product of two matrices. SIAM J. Sci. Stat. Comput. 7 (1986) 1147–1159.
C. C. Paige,Computing the generalized singular value decomposition. SIAM J. Sci. Sta. Comput. 7 (1986) 1126–1146.
C. C. Paige,Some aspects of generalized QR factorizations, inReliable Numerical Computation, Proceedings of the meeting at NPL Teddington, 8–10 July 1987, dedicated to the late James Hardy Wilkinson, FRS. (eds M. Cox and S. Hammarling). Oxford University Press.
C. C. Paige, M. A. Saunders,Towards a generalized singular value decomposition. SIAM J. Numer. Anal. 18 (1981) 398–405.
H. Zha,The restricted singular value decomposition for matrix triplets, SIAM J. Matrix Anal. Appli. Vol. 12, No. 1, 172–194, 1991.
H. Zha,A numerical algorithm for computing the restricted singular value decomposition for matrix triplets, Scientific report ZIB, No. 89-01, Berlin. (Accepted for publication in Lin. Alg. Appl.)
H. Zha,Analysis and numerical solution of the fixed rank matrix approximation problem with general weighting matrices (in preparation).
Author information
Authors and Affiliations
Additional information
The work was partially supported by NSF grant DCR-8412314.
Rights and permissions
About this article
Cite this article
Zha, H. The product-product singular value decomposition of matrix triplets. BIT 31, 711–726 (1991). https://doi.org/10.1007/BF01933183
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF01933183