![Theory of Reinforcement Learning_hi-res logo](/sites/default/files/styles/workshop_banner_sm_1x/public/2023-05/Theory%20of%20Reinforcement%20Learning_hi-res.png.jpg?itok=LJoCeC_d)
Abstract
In this talk we discuss computational approaches based on Monte Carlo sampling techniques to solving multistage stochastic programming problems. In some applications the considered programs have a periodical behavior. We demonstrate that in such cases it is possible to drastically reduce the number of stages by introducing a periodical analog of the so-called Bellman equations, used in Markov Decision Processes and Stochastic Optimal Control. Furthermore, we describe a primal - dual variant of
the Stochastic Dual Dynamic Programming algorithm, applied to the constructed periodical Bellman equations. We consider risk neutral and risk averse settings.