Date/Time
Date(s) - April 12, 2019
11:45 am - 12:35 pm
Location
234 Weil Hall
Categories
David Morton, Ph.D.
David A. & Karen Richards Sachs Professor & Department Chair
Department of Industrial Engineering & Management Sciences, Northwestern University
Abstract: Models and Algorithms for Multi-stage Stochastic Programming
We consider two classes of multi-stage stochastic linear programs (MSLPs) that lend themselves to solution by stochastic dual dynamic programming (SDDP). First, we consider a distributionally robust MSLP. Here, the specific realizations in each stage are fixed, and distributional robustness is with respect to the probability mass function governing those realizations. Second, we consider a class of partially observable MSLPs. In both cases, we describe a computationally tractable variant of SDDP to solve the model. This is joint work with Oscar Dowson, Daniel Duque, and Bernardo Pagnoncelli.