Stochastic Quasigradient Methods and their Implications
Authors: Ermoliev YM, Gaivoronski AA
Publication Year: 1984
Reference: IIASA Working Paper WP-84-055
Abstract
A number of stochastic quasigradient methods are discussed from the point of view of implementation. The discussion revolves around the interactive package of stochastic optimization routines (STO) recently developed by the Adaptation and Optimization group at IIASA. (This package is based on the stochastic and nondifferentiable optimization package (NDO) developed at the V. Glushkov Institute of Cybernetics in Kiev.) The IIASA implementation is described and its use illustrated by application to three problems which have arisen in various IIASA projects.