Acceleration of the convergence of the gradient method by using the conjugate gradient

Abstract

The conjugate gradient method is considered one of the most important methods used to speed up the gradient algorithm, for this purpose, several related algorithms have been developed. We will present a new method that accelerates the convergence of the gradient method (the higher slope method) using a new version of the conjugate gradient and a powerful non-exact linear Wolf search. It will be shown that this algorithm generates descent trends and converges globally.