Seminar Series | Compromise Policies: Extensions to Multi-Stage and Discrete Stochastic Optimization

All dates for this event occur in the past.

270 Journalism Building
242 W 18th Ave
Columbus, OH 43210
United States

Speaker: Suvrajeet Sen, University of Southern California

Title: Compromise Policies:  Extensions to Multi-Stage and Discrete Stochastic Optimization 

 

Abstract:   Compromise Decisions for Stochastic Programming were first proposed for Two-stage Stochastic LPs (SLP), and have been shown to provide very good decisions even in problems with very high variability.  In this lecture we will discuss extensions of Compromise Decisions of SLP to the case of Multi-stage Stochastic LPs as well as Stochastic MIPs (with binary variables).  We will show that despite the differences in the mathematical structures arising in continuous and discrete optimization problems, the same notion of compromise policies provides a common umbrella for variance reduced decisions (or policies).  This talk is based on joint work with a former Ph.D. student Dr. Jiajun Xu. 

Bio: Suvrajeet Sen is a Professor in the Epstein Department of Industrial and Systems Engineering at University of Southern California (USC).  Prior to his latest stint, he served as a Professor at Ohio State where he set up the 3D Laboratory.  Indeed, prior to that, Optimization labs were placed in lower dimensions.  In fact, before joining Ohio State he was at the University of Arizona, where he held a faculty line (1D?) since before the Internet and Latex days!  Over the years, he has had the good fortune of working on many higher dimensional Optimization models and algorithms, and has partnered with many Hokies, Lions, Cats, Buckeyes, and Trojans.  These mascots have kept him feeling younger than he might have otherwise.  He is looking forward to being in Buckeyeland!

Category: Seminars