用信赖域半径收敛到0的信赖域方法求解无约束优化问题,基于函数值平均权重的非单调技术减少了算法的计算量。证明了算法的全局收敛性以及超线性收敛性。
In this paper,we present a new trust region method for unconstrained optimization with the trust region converging to zero.The aids of nonmonotonous technical reduce the algorithm computation.The proposed algorithm is shown to be globally convergent and superlinearly rate.