A new modification of nonlinear conjugate gradient formula

Abstract

Nonlinear conjugate gradient (CG) methods are widely used for solving large-scale unconstrained optimization problems. Many works have tried to improve this method. It requires simplicity and low memory in numerical computation. The exact line search is used for analyzing and implementing CG methods. In this paper, a class of conjugate gradient methods possessing global convergence properties is presented. The global convergence and sufficient descent property are established using exact line searches. Numerical result demonstrated that the new formula of conjugate gradient method is superior and robust as compared to other CG coefficients

Research paper insights: A new modification of nonlinear conjugate gradient formula. Available from: https://www.researchgate.net/publication/292336885_A_new_modification_of_nonlinear_conjugate_gradient_formula [accessed Apr 10, 2017]