A new nonmonotone conic trust-region method for unconstrained optimization problems

A new nonmonotone trust-region algorithm of conic model for unconstrained optimization problems has been provided in this paper. The new method combines a conic model of trust-region subproblem with a nonmonotone strategy for solving unconstrained optimization problems. The local and global convergence properties have been proved under some reasonable assumptions. Numerical experiments show the effectiveness of the new method in comparison with the other nonmonotone techniques that are recently proposed in the literature