非单调信赖域方法求解无约束非光滑优化问题

Non-monotone trust region methods for solving unconstrained nonsmooth optimization problems

提出了非单调信赖域算法求解无约束非光滑优化问题,并和经典的信赖域方法作比较分析。同时,设定了一些条件,在这些假设条件下证明了该算法是整体收敛的。数值实验结果表明,非单调策略对无约束非光滑优化问题的求解是行之有效的,拓展了非单调信赖域算法的应用领域。

It proposes a solution for non-smooth unconstrained optimization problems of non-monotone trust region methods, and compares the approach with the classical trust region methods. At the same time, the algorithm sets some conditions, under the conditions of these assumptions, the algorithm is provea global convergence. Numerical results show that non-monotonic strategy not only in solving non-smooth unconstrained optimization problems is effective, but also expands the non-monotonic trust region algorithm applications.