[NOVA Math] Seminar of Operations Research

Abstract: Two-stage distributionally robust optimization (DRO) is a recent optimization technique for handling uncertainty. It is less conservative than robust optimization and more flexible than stochastic programming. In DRO, the probability distribution of uncertain parameters is assumed to be unknown but to belong to a prescribed ambiguity set. The size of the ambiguity set is frequently governed by a single parameter that controls the level of conservatism of the resulting optimization problem. Determining appropriate values for this parameter is therefore a key research challenge. In this talk, we first introduce a DRO model for the berth allocation problem with uncertain handling times, which is one of the most important problems in port terminals. Then, we present methods for identifying suitable values of the ambiguity-set control parameter used in the DRO model. Those values are expected to enable us to obtain all relevant first-stage solutions worth considering.

Short Bio: PhD in applied mathematics in the field of operations research by the University of Aveiro in 2019. Filipe Rodrigues is currently an assistant professor at Lisbon School of Economics and Management (ISEG), University of Lisbon. He is also a researcher in CEMAPRE – Centro de Matemática Aplicada à Previsão e Decisão Económica. His research mainly focuses on two topics: optimization problems in port management and logistics, and optimization under uncertainty, covering different methodological areas such as robust optimization, stochastic programming, and distributionally robust optimization.

March 12 | 15h00 | Room 115 - Building II