Some Ideas on Input Modeling and Input Uncertainty in Stochastic Optimization and Simulation

In this talk we focus on the interface between the probabilistic input and the stochastic optimization/simulation models. We first discuss using Gaussian mixture models (GMM) to conduct input modeling in chance constrained programs (CCP). We show how to solve several classes of CCP with GMM, using gradient based approaches and branch-and-bound procedures. We next discuss input uncertainty when simulating stochastic systems. We model the uncertainty using likelihood ratio and develop a robust simulation (RS) approach, which aims to quantify the worst-case performance when uncertainty exists. We show that the RS approach is computationally tractable and the corresponding results reveal important information of the stochastic systems.

HU Zhaolin is a professor of School of Economics and Management at Tongji University. His research interests include simulation, stochastic optimization, statistical learning, and risk management. He has published on journals such as Management Science, Operations Research, and INFORMS Journal on Computing. He serves as an associate editor of Journal of Management Science and Engineering.