基于信赖域技术的非单调超记忆梯度算法

A Non-monotone Super-memory Gradient Method Based on Trust Region Technique

基于信赖域技术和修正拟牛顿方程,结合Zhang H.C.非单调策略,设计了新的求解无约束最优化问题的非单调超记忆梯度算法,分析了算法的收敛性和收敛速度.数值实验表明算法是有效的,适于求解大规模问题.

Based on trust region technique and quasi-Newton equation, by combining with Zhang H.C. non-monotone strategy, we present a new non-monotone super-memory gra- dient method for unconstrained optimization problem. The global convergence properties and convergence rate of the new method are proved. The numerical results show that the new meth- ods are effective and attractive for large-scale optimization problems.