摘要
本文对于一类形如F(x)=g(x,maxΦ_(ij)(x),…,maxΦ_(mj)(x))+h(x)的拟可微函数(在Demyanov和Rubinov意义下)给出了一种优化算法,其中g,Φ_(ij)分别为R^(m+n)和R^n上的连续可微函数,且g(x,y_1,…,y_m)关于每一个y_i都是非增的,h(x)为R^n上的凸函数。
An optimization algorithm for a certain class of qua-sidifferentiable functions (in the sense of Demyanov and Rubinov)F(x)=g(x, maxΦ_(1j)(x), …, max j∈Jm Φ_(mj)(x))+h(x) is given. Where, gand Φ_(ij) are continuously differentiable functions defined onR^(m+n) and R^n respectively, and g(x, y_1, …, y_m) is nonincreasingwith respect to each y_i, and h(x) is a convex function on R^n.
关键词
拟可微函数
不可微优化
函数
nondifferentiable optimization
quasidifferentiable function
