A New Non-monotone Adaptive Retros-pective Trust Region Method for Uncon-strained Optimization Problems
Yajing Wang
College of Mathematics and Information Science, Hebei University, Baoding, 071002, China
Abstract: In this paper, we propose and analyze a new nonmonotone adaptive retrospective trust region me-thod for unconstrained optimization problems. Actually, we incorporate a new proposed nonmonotone tech-nology with the adaptive retrospective trust region method. Some properties of the new algorithm are ana-lyzed. Theoretical analysis shows that the new proposed method has a global convergence under some mild conditions.
Keywords: Nonmonotone technology; Unconstrained optimization; Retrospective trust region; Global convergence