Brief review on the performance of RMIL Conjugate Gradient Methods

• Published in: AIP Conference Proceedings
• Main Author: Idalisa N.; Rivaie M.; Fadhilah N.H.; Nasir M.A.S.
• Format: Conference paper
• Language: English
• Published: American Institute of Physics Inc. 2022
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85160346926&doi=10.1063%2f5.0078462&partnerID=40&md5=49861fe93469b8f0b28530d9f1404f3d

This paper focuses on the Rivaie, Mohamad, Ismail and Leong (RMIL) type conjugate gradient (CG) method to solve problem of unconstrained optimization. This method always yields a descent search direction and possesses good global convergence properties. However, the issue of RMIL method practical performance motivates various research and development of numerous variants of RMIL. These variations cause a problem to decide the most efficient method for the purpose of future study and potential opportunity to be extended in other research area. Thus, the objective of this paper is to determine the most superior RMIL variant based on their numerical performance for solving 67 standard test problems. The performance of selected RMIL variant are evaluated using the strong Wolfe Line Search (LS) through a performance profile tool. Within the scope of this study, result shows that the three-term approach of RMIL are superior than other CG approaches based on the number of iterations and CPU time metric. © 2022 American Institute of Physics Inc.. All rights reserved.