一类极小值复合向量函数Clarke广义Jacobi的有效算法

An effective algorithm of Clarke generalized Jacobi for a class of minimum complex vector function

在非光滑优化的数值方法中,Clarke广义Jacobi的计算是必要子算法.由于一般类Lipschitz函数的广义微分不容易计算,因此对一类极小值复合向量函数Clarke广义Jacobi计算方法进行研究.借助于中值函数和KKT最优条件,将混合互补问题转化为两种不同的等价形式,然后以这两种形式为应用背景,给出了求解一类极小值复合向量函数广义Jacobi的有效算法.

The calculation of Clarke generalized Jacobi,which is not easyfora Lipschitz function,is necessary in nonsmoothoptimization.We devoted toa class of minimum complex vector function in this paper.Firstly,we transformed the mixed complementarity problem into two different equivalent forms by using the median function and KKT optimal condition,based on which,an effective algorithm for solving the class of minimum-value composite vector function generalized Jacobi is presented in the final.