摘要：本文主要介绍了求解无约束最优化问题的三种比较典型的非单调信赖域算法，并利用国际上广泛采用的优化测试函数包（文献[1]）对这三种方法进行大量的数值试验研究，结果表明算法2（朱德通提出的算法）（文献[18]）效率最高，算法1（孙文瑜提出的算法）（文献[9]）效率最低。从以上的数值计算结果和分析来看，几种非单调信赖域方法的数值表现均与其中的控制非单调性的参数选择有较大的关系。在实际应用当中我们要恰当的选择算法中的参数来获得好的计算效果。
关键词: 无约束最优化; 信赖域方法; 非单调; 数值结果
Abstract:This paper describes the unconstrained optimization problem solving three typical non-monotone trust region algorithm, and is widely used by international optimize the test function package ([1]) this Three methods for a large number of numerical tests, the results show that the algorithm 2 (Zhu Detong's algorithm) is the highest efficient, the algorithm 1 (Sun Wenyu the algorithm) is the least efficient. From the above numerical results and analysis, several non-monotone trust region methods are the numerical performance of the control with which the parameters of non-monotonic relationship between greater. In practice, we have to choose the appropriate algorithm in the calculation of parameters to obtain good results.
Key words: unconstrained optimization; trust region; non-monotone; numerical result
