Some three-term conjugate gradient methods with the new direction structure

Abstract

In this paper, a three-term conjugate gradient method with the new direction structure is proposed for solving large-scale unconstrained optimization problems, which generates a sufficient descent direction in per-iteration by the aid of some inexact line search conditions. Under suitable assumptions, the proposed method is globally convergent for nonconvex smooth problems. We further generalize the new direction structure to other traditional methods and obtain some algorithms with the same structure as the proposed method. Preliminary numerical comparisons show that the proposed methods are effective and promising for the test problems.