Parallel Computation of the SVD of a Matrix Product

In this paper we study a parallel algorithm for computing the singular value decomposition (SVD) of a product of two matrices on message passing multiprocessors. This algorithm is related to the classical Golub-Kahan method for computing the SVD of a single matrix and the recent work carried out by Golub et al. for computing the SVD of a general matrix product/quotient. The experimental results of our parallel algorithm, obtained on a network of PCs and a SUN Enterprise 4000, show high performances and scalability for large order matrices.

• Type: Conference
• Year: 1999
• Where: PVM
• Authors: José M. Claver, Manuel Mollar, Vicente Hernández
• Keywords: Distributed And Parallel Computing, Message Passing Multiprocessors, Parallel Algorithm, PVM 1999, Singular Value Decomposition