Modeling Fairness in Facility Location and Routing, a seminar by Professor Ivana Ljubić (ESSEC Business School)

Modeling Fairness in Facility Location and Routing

By Professor Ivana Ljubić from the ESSEC Business School

Abstract

Most combinatorial optimization problems assess solution quality by aggregating individual performance measures into a single metric. In facility location, for example, allocation costs from clients to their nearest open facility are typically evaluated through the average allocation cost. Similarly, in routing problems, optimization usually aims to minimize the total cost across all routes, often overlooking the quality of individual routes from the perspective of each driver. Even in machine learning applications such as supervised classification, the objective is commonly to minimize the average misclassification error over all observations.

In this talk, we investigate alternative quality measures that explicitly account for fairness and equity—such as the min-max criterion, range, Gini deviation, and Hurwicz criterion—as well as robustness measures including conditional value-at-risk and k-sum objectives. To model these objectives, we employ the discrete ordered operator, which provides a unified optimization framework capable of capturing all these measures and many others. This contrasts with much of the existing literature, where dedicated models and solution methods are typically developed for each individual criterion.

We present a generic mixed-integer programming (MIP) framework for modeling and solving these diverse objectives within a common optimization paradigm. For facility location problems, we demonstrate how discrete ordered objectives can be efficiently addressed through Benders decomposition, leading to substantial improvements over state-of-the-art MIP formulations. For routing problems, we exploit connections with bilevel optimization to incorporate fairness considerations directly into route planning.

The talk is based on joint work with M. Pozo, J. Puerto Albandoz, and A. Torrejón.

About the speaker

Ivana Ljubić is Professor of Operations Research at ESSEC Business School in Paris. She received her habilitation in Operations Research from the University of Vienna in 2013 and earned a PhD in Computer Science from the Vienna University of Technology in 2004. Until 2015, she held a faculty position at the University of Vienna. She has also been a visiting researcher at several international institutions, including Sapienza University of Rome, the University of Maryland, TU Berlin, TU Dortmund, and Paris Dauphine University. Since September 2015, she has been a faculty member at ESSEC Business School.

Professor Ljubić serves as Area Editor for Transportation Science and Associate Editor for Operations Research and Networks. She has also served as guest editor for the European Journal of Operational Research and the Annals of Operations Research, and is a former Chair of the INFORMS Telecommunications Section.

Her research interests include network design, combinatorial optimization, optimization under uncertainty, and bilevel optimization. Her work focuses on the development of mixed-integer linear and nonlinear optimization methods for applications in telecommunications, transportation, logistics, data and distribution network design, and social networks.