L. F. Escudero Bueno, A. Alonso-Ayuso, J. F. Monge, D. Morales, L. Pardo
Distributionally robust optimization (DRO) is motivated as a counterpart of the usually unknown underlying probability distribution followed by the uncertainty in dynamic problems. A MILP modeling paradigm is presented for problem solving in multi-horizon stochastic environments (DRO-MHSE), where the uncertainty is represented in a finite set of scenarios for the realization of the uncertain parameters in the related strategic and operational ambiguity sets defined by considering the Wasserstein distance. The aim is to maximize the overall expected DRO solution value in the scenarios, subject to the constraint system for each ambiguity set member and the related one for the stochastic dominance risk averse functional. SFR3, a constructive matheuristic algorithm, is specialized for dealing with DRO-MHSE. A supply network design planning is considered as a pilot case to validate the proposal. A broad computational experience is reported.
Keywords: Multi-horizon stochastic MILP, embedded two-stage distributionally robust optimization, Wasserstein distance; stochastic dominance risk averse, matheuristics, supply network design planning
Scheduled
Integer Optimisation and Combinatorics
November 8, 2023 5:20 PM
CC2: Conference Room