Rank-one update of Cholesky factorization
Loading...
Authors
Ntlatlapa, N
Issue Date
1995
Type
Language
en
Keywords
Alternative Title
Abstract
Symmetric positive definite matrices appear in most methods for
Unconstrained Optimization. The method widely adopted for factoring
these matrices is Cholesky Factorization. Furthermore, in Quassi-Newton
methods for unconstrained optimization these matrices are
continually updated and factorized.
Here we consider factoring an n x n symmetric positive definite
matrix of the form:
A' = A + CXZZT ,
where A is symmetric positive definite, a is a scalar and z is a vector of
length n. We assume that A has already been factorized by Cholesky
factorization. The adopted methods are due to Gill et. al. [GGS75, GM72].
Description
Citation
Ntlatlapa, N. (1995) Rank-one update of Cholesky factorization. Papers Delivered at the SAICSIT 95 Research and Development Symposium (South African Institute for Computer Scientists and Information Technologists), Film Auditorium, University of South Africa, Pretoria, 25-26 May1995, edited by A.L. Steenkamp (UNISA) (ISBN 0-86981-909-7)