Updating the singular value decomposition

作者:

Highlights:

摘要

The spectral decomposition of a symmetric matrix A with small off-diagonal and distinct diagonal elements can be approximated using a direct scheme of R. Davies and Modi (Linear Algebra Appl. 77 (1986) 61). In this paper a generalization of this method for computing the singular value decomposition of close-to-diagonal A∈Rm×n is presented. When A has repeated or “close” singular values it is possible to apply the direct method to split the problem in two with one part containing the well-separated singular values and one requiring the computation of the “close” singular values.

论文关键词:Singular value decomposition,Skew-symmetric matrices,Hadamard products,Kogbetliantz's algorithm,Jacobi's algorithm

论文评审过程:Received 12 September 2002, Revised 20 November 2003, Available online 26 February 2004.

论文官网地址:https://doi.org/10.1016/j.cam.2003.12.039