B.B. Zhou, Comput. Sci. Lab., Australian Nat. Univ., Canberra, ACT, Australia
R.P. Brent, Comput. Sci. Lab., Australian Nat. Univ., Canberra, ACT, Australia
In this paper we give evidence to show that in one-sided Jacobi SVD computation the sorting of column norms in each sweep is very important. Two parallel Jacobi orderings are described. These orderings can generate n(n-1)/2 different index pairs and sort column norms at the same time. The one-sided Jacobi SVD algorithm using these parallel orderings converges in about the same number of sweeps as the sequential cyclic Jacobi algorithm. Some experimental results on a Fujitsu AP1000 are presented. The issue of equivalence of orderings is also discussed.
Index Terms:
Jacobian matrices; parallel algorithms; singular value decomposition; parallel implementation; one-sided Jacobi algorithm; singular value decompositions; sorting; column norms; parallel Jacobi orderings; parallel orderings; Fujitsu AP1000; equivalence
Citation:
B.B. Zhou, R.P. Brent, "On parallel implementation of the one-sided Jacobi algorithm for singular value decompositions," pdp, pp.401, 3rd Euromicro Workshop on Parallel and Distributed Processing, 1995