Vectorization of conjugate-gradient methods for large-scale minimization

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

3 Scopus citations

Abstract

Vectorization techniques are applied to the nonlinear conjugate-gradient method for large-scale unconstrained minimization. Computational results are presented for a robust limited-memory quasi-Newton-like conjugate-gradient algorithm applied to meteorological problems. The vectorization results in speedups up to a factor of 21 compared to the performance of the scalar code, when nonlinear functions of 104-105 variables are minimized. A sizable reduction in the CPU time required for the minimization of large-scale nonlinear functions is obtained, showing the advantages of the approach.

Original languageEnglish
Title of host publicationProc Supercomputing 88
PublisherPubl by IEEE
Pages410-418
Number of pages9
ISBN (Print)081860882X
StatePublished - 1988

Publication series

NameProc Supercomputing 88

Fingerprint

Dive into the research topics of 'Vectorization of conjugate-gradient methods for large-scale minimization'. Together they form a unique fingerprint.

Cite this