A new modified conjugate gradient method under the strong Wolfe line search for solving unconstrained optimization problems

Abstract

Conjugate gradient (CG) method is well-known due to efficiency to solve the problems of unconstrained optimization because of its convergence properties and low computation cost. Nowadays, the method is widely developed to compete with existing methods in term of their efficiency. In this paper, a modification of CG method will be proposed under strong Wolfe line search. A new CG coefficient is presented based on the idea of make use some parts of the previous existing CG methods to retain the advantages. The proposed method guarantees that the sufficient descent condition holds and globally convergent under inexact line search. Numerical testing provides strong indication that the proposed method has better capability when solving unconstrained optimization compared to the other methods under inexact line search specifically strong Wolfe–Powell line search.