Abstract
Conjugate gradient methods play an important role in unconstrained optimization. Numerous studies and modifications have been devoted recently to improve this method. In this paper we propose a new conjugate gradient coefficient (β k ) by modifying the already proven Hestenes-Steifel formula. In this new β k we introduce a new formula for the denominator and retain the numerator of the Hestenes-Steifel formula. Numerical results have shown that the new formula for β k performs far better than the original Hestenes-Steifel, but still possesses global convergence properties. This new method also outperforms the other conjugate gradient methods.
Keywords and phrases: