无约束优化问题的锥模型回溯过滤信赖域算法

A retrospective conic trust-region filter method for unconstrained optimization

基于锥模型信赖域框架,结合多维滤子集技巧,提出一个求解无约束优化问题的回溯过滤信赖域算法,锥模型比二次模型更一般,其信赖域模型是它的一个特例.而且对比于一般的二次模型,更多地利用了每一个迭代点的信息.本文在通常的假设条件下,分析了算法的全局收敛性.

The conic model method is a new type of method with more information available at each iteration than standard quadratic based method.Recently,a filter technique is presented for unconstrained optimization.Based on these work,we present a retrospective conic trust-region filter method for unconstrained optimization.The new algorithm is shown to be globally convergent under standard conditions.