A Comparative Study of Three New Conjugate Gradient Methods with Exact Line Search

Conjugate Gradient methods play an important role in solving unconstrained optimization, especially for large scale problems. In this paper, we compared the performance profile of the classical conjugate gradient coefficients FR, PRP with three new beta(k). These three new beta(k) possess global con...

Full description

Bibliographic Details
Main Authors: Abshar, A (Author), Hamoda, M (Author), Mamat, M (Author), Rivaie, M (Author)
Format: Article
Language:English
Published: 2015
Subjects:
Online Access:View Fulltext in Publisher
Description
Summary:Conjugate Gradient methods play an important role in solving unconstrained optimization, especially for large scale problems. In this paper, we compared the performance profile of the classical conjugate gradient coefficients FR, PRP with three new beta(k). These three new beta(k) possess global convergence properties using the exact line search. Preliminary numerical results show that the three new beta(k) are very promising and efficient when compared to CG coefficients IA.PRP.
DOI:10.1063/1.4932439