Leuven Optimization Talks, May 13th, 2025
| Posted by: | Hande Yaman Paternotte |
| Date: | 2025-04-07 |
| Contact: | [email protected] |
Dear all,
We would like to invite you to a day of four tutorials/talks on May13th, 2025 in Leuven.
The talks will take place at BETH 00.08 (aula wolfspoort, Huis Bethlehem, Schapenstraat 34, 3000 Leuven).
Here is the schedule:
10:00 -11:15
Martine Labbé (Computer Science Department, Université Libre de Bruxelles/ Inria Lille)
On linear bilevel optimization problems
11:15 - 12:30
Marius Roland (Inria Lille)
Relaxation Strengthening Techniques for Chance-Constrained Stochastic Programs with Finite Support
12:30 - 14:00 Lunch
14:00 – 15:15
Bernardo K. Pagnoncelli (SKEMA Business School, Lille)
Sequential decision making under uncertainty: DADS and CACS
15:15 – 16:30
Merve Bodur (School of Mathematics, The University of Edinburgh)
Recent Advances in Solving Multistage Stochastic Mixed-integer Programs
The abstracts are at the end of this email.
If you would like to attend, please fill in the form below until April 22nd:
Please forward to people who may be interested.
Best regards,
Hande Yaman
Speaker: Martine Labbé (Computer Science Department, Université Libre de Bruxelles/ Inria Lille)
Title: On linear bilevel optimization problems
Abstract: A bilevel optimization problem is a hierarchical optimization framework where some constraints require a subset of variables to be optimal solutions to another embedded optimization problem. This structure naturally models competitive interactions between two decision-makers—a leader and a follower—who act sequentially. In this talk, I will focus on the simplest class of bilevel problems: linear bilevel optimization. I will introduce their fundamental properties, key characteristics, and the main algorithms used to solve them. Additionally, I will discuss both theoretical insights and computational challenges, emphasizing the inherent complexity of these problems.
Speaker: Marius Roland (Inria Lille)
Title: Relaxation Strengthening Techniques for Chance-Constrained Stochastic Programs with Finite Support
Abstract: This presentation considers Chance Constrained Stochastic Programs (CCSPs) with Finite Support. It offers an accessible tutorial for researchers with little or no prior knowledge of the topic. CCSPs are a class of optimization problems used in fields such as finance and power systems, where decisions must satisfy uncertain constraints with high probability. We begin by explaining the fundamental structure of CCSPs and their key challenges, including their combinatorial nature and the non-convexity of the solution space. The presentation then provides an overview of widely-used solution techniques, to set the stage for the main topic: relaxation strengthening techniques for CCSPs. Two categories are explored. First, we discuss big-M tightening techniques. Second, we introduce valid inequalities, including both established families and new families. Finally, we evaluate the discussed approaches using numerical results from the literature, focusing on their applicability to different types of CCSPs. This evaluation offers practical guidance for selecting appropriate methods based on problem-specific characteristics, such as constraint structure and size of the scenario set.
Speaker: Bernardo K. Pagnoncelli (SKEMA Business School, Lille)
Title: Sequential decision making under uncertainty: DADS and CACS
Abstract: In this tutorial, I will explore two fundamental paradigms in sequential decision-making under uncertainty: Markov decision processes (MDPs) and multi-stage stochastic programming (MSSP).
While MDPs are well-suited for discrete-action, discrete-state space (DADS) problems, MSSP has been successfully applied to continuous-action, continuous-state space (CACS) settings.
My focus will be on CACS, demonstrating how different formulations—such as infinite-horizon and risk-averse problems—can be expressed without relying on scenario trees.
I will conclude by discussing a scalable algorithm capable of solving large-scale instances of CACS.
Speaker: Merve Bodur (School of Mathematics, The University of Edinburgh)
Title: Recent Advances in Solving Multistage Stochastic Mixed-integer Programs
Abstract: Multistage stochastic mixed-integer programs (MSMIPs) can model complex sequential decision-making problems under uncertainty and appear in many applications. However, due to both the stochastic and integer components, their inherent computational challenges require sophisticated solution methodologies. In this talk, we will review recent advances in solving MSMIPs, in particular for the scenario-tree-based approaches. We will discuss exact methods, bounding ideas, partially extended formulations, and various policy classes, including aggregation-based policies, decision-rule-based policies, and interpretable policies.
