A Stochastic Algorithm for Minimax Problems
Authors: Ermoliev YM, Gaivoronski AA
Publication Year: 1982
Reference: IIASA Collaborative Paper CP-82-088
Abstract
This paper deals with minimax problems in which the "inner" problem of maximization is not concave. A procedure based on the approximation of the inner problem by a stochastic set of elements which can contain only two elements at each iteration is shown to converge with probability 1.