A Singular Value Decomposition Updating Algorithm for Subspace Tracking
- 1 October 1992
- journal article
- Published by Society for Industrial & Applied Mathematics (SIAM) in SIAM Journal on Matrix Analysis and Applications
- Vol. 13 (4) , 1015-1038
- https://doi.org/10.1137/0613061
Abstract
In this paper, the well-known QR updating scheme is extended to a similar but more versatile and generally applicable scheme for updating the singular value decomposition (SVD). This is done by supplementing the QR updating with a Jacobi-type SVD procedure, where apparently only a few SVD steps after each QR update suffice in order to restore an acceptable approximation for the SVD. This then results in a reduced computational cost, comparable to the cost for merely QR updating.The usefulness of such an approximate updating scheme when applied to subspace tracking is examined. It is shown how an $\mathcal{O}(n^2 )$ SVD updating algorithm can restore an acceptable approximation at every stage, with a fairly small tracking error of approximately the time variation in $\mathcal{O}(n)$ time steps.Finally, an error analysis is performed, proving that the algorithm is stable, when supplemented with a Jacobi-type reorthogonalization procedure, which can easily be incorporated into the updating scheme. In this paper, the well-known QR updating scheme is extended to a similar but more versatile and generally applicable scheme for updating the singular value decomposition (SVD). This is done by supplementing the QR updating with a Jacobi-type SVD procedure, where apparently only a few SVD steps after each QR update suffice in order to restore an acceptable approximation for the SVD. This then results in a reduced computational cost, comparable to the cost for merely QR updating.The usefulness of such an approximate updating scheme when applied to subspace tracking is examined. It is shown how an $\mathcal{O}(n^2 )$ SVD updating algorithm can restore an acceptable approximation at every stage, with a fairly small tracking error of approximately the time variation in $\mathcal{O}(n)$ time steps.Finally, an error analysis is performed, proving that the algorithm is stable, when supplemented with a Jacobi-type reorthogonalization procedure, which can easily be incorporated into the updating scheme.
Keywords
This publication has 21 references indexed in Scilit:
- Tracking a few extreme singular values and vectors in signal processingProceedings of the IEEE, 1990
- Linear convergence of the row cyclic Jacobi and Kogbetliantz methodsNumerische Mathematik, 1989
- On Kogbetliantz's SVD Algorithm in the Presence of ClustersLinear Algebra and its Applications, 1987
- On Jacobi Methods for Singular Value DecompositionsSIAM Journal on Scientific and Statistical Computing, 1987
- Updating the singular value decompositionNumerische Mathematik, 1978
- Rank-one modification of the symmetric eigenproblemNumerische Mathematik, 1978
- Error analysis of QR decompositions by Givens transformationsLinear Algebra and its Applications, 1975
- Methods for modifying matrix factorizationsMathematics of Computation, 1974
- Some Modified Matrix Eigenvalue ProblemsSIAM Review, 1973
- The cyclic Jacobi method for computing the principal values of a complex matrixTransactions of the American Mathematical Society, 1960