Groupe d’études et de recherche en analyse des décisions

New Trends and Results in Multiobjective Programming for Stochastic or Robust Optimization and Decision-Making under Uncertainty

Alexander Engau University of Colorado Denver, États-Unis

One of the most challenging problems in operations research is how to handle and effectively deal with both data and model uncertainties. In addition to the classical approaches from probabilistic or stochastic programming and the still more recent paradigm of robust optimization, there also been an increased interest in using the theory and methodology of multiple criteria optimization for decision making under uncertainty and robustness in particular. In this presentation, we plan to give a brief overview of some of these new trends including contributions by both the speaker as well as several other colleagues in the broader optimization community. Specifically, we intend to describe new results and approaches to achieve Pareto optimality in robust optimization, to relate robust and stochastic formulations through the concept of scalarization, and to provide a comprehensive analysis of trade-offs and risks based on extensions of the theory of proper (Pareto) efficiency. The talk will conclude with a summary of criticisms that have challenged the applicability of multicriteria methods especially in the context of robustness, and that suggest an interesting and likely controversial further discussion.