Details
Abstract: We embed modeling features from Markov decision processes (MDPs) in a class of multi-stage stochastic programs. This class captures a type of MDP with continuous states and actions and also with a particular structure. We extend Markov-chain based policy graphs from stochastic programming to include a form of statistical learning with Bayesian updates and to include decision-dependent one-step transition probabilities. Much of the talk aims to illustrate the expressiveness of our modeling approach. To computationally handle these new features, we extend the growing class of stochastic dual dynamic programming algorithms. This is joint work with Nazlican Arslan, Oscar Dowson, and Bernardo Pagnoncelli.
Bio: Dave Morton is the Walter P. Murphy Professor in Industrial Engineering and Management Sciences (IEMS) at Northwestern University, where he previously served as department chair. He worked as a Fulbright Research Scholar at Charles University in Prague, received the Presidential Early Career Award for Scientists and Engineers, and is an INFORMS fellow. He advances methods in optimization for decision making under uncertainty. He has applied research interests in energy, security, and public health.