Mathematica Eterna

Mathematica Eterna
Open Access

ISSN: 1314-3344


A new Non-monotone Spectral Conjugate Gradient method for Unconstrained Optimization

Ruyue Hou, Ke Su and Zixing Rong

In this paper, we combine spectral Conjugate Gradient method which combines the conjugate gradient direction with spectral step-length effectively with a non-monotone line search technique and we obtain a new algorithm. besides, in this spectral Conjugate Gradient method the spectral step-length is applied to the entire Conjugate Gradient direction rather than the negative gradient direction, The global convergent properties of the algorithm are proved under some appropriate conditions