Algorithmic models of human behavior and stochastic optimization

Abstract

The article explores the parallelization of computations in solving stochastic optimization
problems; The application of the results obtained here to the search for an equilibrium distribution
of flows along paths is considered; The dependence of the rate of convergence of optimal algorithms
in the problems of stochastic, gradient optimization is investigated, depending on the number of
calls to the oracle behind the implementation of the function at each iteration. A distinctive feature
of this article is the demonstration of the results obtained with illustrative examples.
Key words: Stochastic mirror descent, gradient methods, the search for equilibrium in