Combining nonmonotone conic trust region and line search techniques for unconstrained optimization (Q629507)

The authors propose a trust region method for solving a general unconstrained optimization problem. After outlining the necessary background and an overview of the literature in the first section, they proceed to describe a new trust region algorithm which can be regarded as a combination of the conic model, non-monotone and line-search techniques. The third and fourth sections study the convergence properties of the proposed algorithm, whereas the last section presents the results of numerical experimentation using the proposed algorithm.