摘要
本文对一类形如F(x)=g(x, (?)(x),…,(?)(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,并证明了算法的收敛性。这里g,φ_(ij)分别为R^(m+n)和R^n上的连续可微函数,h(x)为R^n上的凸函数。
An algorithm for minimizing a class of quasidifferentiable functions (in the sense of Demyanov and Rubinov) F(x)=g(x, max φ_ (ij)(x),……, max φ_(mj)(x))+h(x) is given, and the convergence therom of the algorithm is proved, where g and φ_(ij) are continuously differentiable functions on R^(m+n) and R^n respectively,
关键词
不可微优化
拟可微函数
下降算法
nondifferentiable optimization
quasidifferentiable function
desent algorithm
