Management Science Seminar - Modeling Discrete-Event System Dynamics as Solutions to Optimization Models
Wednesday 24 September 2008, 12:00
LT5, Management School
Professor Lee W. Schruben
UC Berkeley
and
Professor Wai Kin Chan
Rensselaer Polytechnic Inst.
Abstract: Discrete-event dynamic models are used to study a wide variety of stochastic systems in production, healthcare, supply chains, transportation, security, communication, etc. Such systems are typically studied in using simulations. This talk will introduce a new approach for modeling discrete-event system dynamics as the solutions to optimization models. This allows use of the rich theory and algorithms from mathematical programming in analyzing discrete-event systems. Several applications and theoretical results are presented using this new methodology including response gradient estimation using shadow prices, scheduling (where the concept of virtual resources is introduced), new proofs and types of reversibility for tandem queues, and sample-path duality for queueing systems. The motivation of this paper is to develop and demonstrate explicit synergies between the major branches in Operations Research of queueing, simulation, and mathematical programming.
This talk is based on a paper that will appear in the next issue of Operations Research. Stephan has the copy of the paper and will distribute it before the seminar.
