On Incorporating Forecast Information into MDPs
Abstract
In an increasing number of sequential decision-making applications, one has forecast information that can be used to enhance the decision process. In particular, in developing more efficient energy systems, one has the ability to use weather forecasts to improve the control of the associated system. In this talk, we will discuss ongoing work at Stanford that relates to the operation of Stanford’s district heating and cooling system, as well as a mathematical MDP formulation for linear control systems that takes into account forecast information in a principled way. By principled, we mean that the forecasts are required to be mathematically consistent with the underlying state dynamics, in the sense that the forecasts are required to satisfy a certain martingale property. This work is joint with Jacques de Chalendar.
Bio
Peter W. Glynn is the Thomas Ford Professor in the Department of Management Science and Engineering (MS&E) at Stanford University and also holds a courtesy appointment in the Department of Electrical Engineering. He was Director of Stanford's Institute for Computational and Mathematical Engineering from 2006 until 2010 and served as Chair of MS&E from 2011 through 2015. He is a Fellow of INFORMS and a Fellow of the Institute of Mathematical Statistics. His research interests lie in simulation, computational probability, queueing theory, statistical inference for stochastic processes, and stochastic modeling.
