极大值复合函数Clarke广义梯度计算的一个新方法

A NEW APPROACH TO COMPUTING A CLARKE GENERALIZED GRADIENT FOR A SMOOTH COMPOSITION OF MAX-TYPE FUNCTIONS

对于一类非光滑函数一光滑函数的有限次极大值复合函数，给出了计算它在一点处Ｃｌａｒｋｅ广义梯度中一个元素的新方法．与以往方法比较，本文的方法不需判别线性不等式组的相容性，因而易于实现．

This paper refers to computing Clarke generalized gradient for a smooth composition of max-type functions. A new approach to computing an element of Clarke generalized gradient for this class of functions at a point, is proposed. In contrast to previous publications, the present approach does not employ the procedure of determining consistencies of system of strictly linear inequalities. Hence, it is easy to be implemented.