A class of nonmonotone trust region algorithm for solving unconstrained nonlinear optimization problems

Abstract

Based on the nonmonotone line search technique proposed by Gu and Mo a nonmonotone trust region algorithm is proposed for solving unconstrained nonlinear optimization problems. The new algorithm is resets the ratio ρ k for evaluating whether the trial step d k is acceptable. The global and superlinear convergence of the algorithm are proved under suitable conditions. Numerical results show that the new algorithm is effective.