A class of efficient new descent methods. (Chinese. English summary) Zbl 1125.65053

Summary: In this paper, we propose a class of new descent methods and, also, we give two hybrid methods based on Hestenes-Stiefel and our new methods. We prove their global convergence with Wolfe line search without descent condition. Numerical experiments shows that our methods are very efficient, especially, for large scale problems.


65K05 Numerical mathematical programming methods
90C30 Nonlinear programming