Pre-Conference Workshop: „Smarter Modeling with Gurobipy“

Speaker: Robert Luce

Time: Tuesday, September 1, 14:45 – 16:00

Room: TBA

Abstract: This hands-on workshop explores the powerful capabilities of gurobipy, Gurobi’s Python interface, with a focus on best practices and comparative analysis of different modeling paradigms. Participants will gain practical insights into different modeling flavors:

• Traditional Term-Based Modeling: The foundational approach using object-oriented constructs with variables, constraints, and expressions built iteratively.
• Matrix-Friendly API: Introduced in Gurobi 9.0 and enhanced through version 13.0, this paradigm leverages NumPy-style vectorization, broadcasting, and multi-dimensional operations.
• gurobipy-pandas: A convenience layer that allows mathematical model building directly out of DataFrames. The workshop is suitable for both optimization practitioners seeking to enhance their gurobipy skills and students who want to get started with modelling in gurobipy.

Pre-Conference Workshop: „Modeling and Solving Routing Problems with Hexaly Studio“

Speaker: Léa Blaise

Time: Tuesday, September 1, 13:15 – 14:30

Room: TBA

Abstract: Hexaly Optimizer is a next-generation mathematical optimization solver that extends the traditional Mixed-Integer Programming (MIP) paradigm with a richer modeling framework. Alongside classical algebraic formulations, it supports native combinatorial constructs, nonlinear expressions, and black-box optimization, enabling users to express complex optimization problems in a compact and natural way. By preserving high-level problem structure, Hexaly automatically leverages advanced algorithmic techniques, including decomposition methods, column generation, branch-price-and-cut, and decision diagrams. As a result, it delivers state-of-the-art performance across challenging optimization domains such as routing, scheduling, packing, simulation-based optimization, machine-learning-driven decision problems, and more.
In this hands-on workshop, we will focus on the application of Hexaly to routing problems, showcasing how the solver’s set-based modeling capabilities naturally align with the structure of these problems. Participants will be guided through building and solving real-world routing problems using Hexaly Studio, a web-based integrated development environment designed specifically for optimization modeling. Hexaly Studio features intuitive dashboards, interactive widgets, and graphical solution visualizations, making it easy to both formulate models and interpret results. Through guided examples and interactive problem-solving, participants will gain hands-on experience with Hexaly and will be equipped to explore it further in their own research or applied projects.

Keynote: „Gurobi’s Primal-Dual Hybrid Gradient (PDHG) algorithm: practice and performance“

Speaker: Robert Luce

Time: Wednesday, September 2, 15:00 – 16:00

Room: TBA

Abstract: Gurobi’s Primal-Dual Hybrid Gradient (PDHG) algorithm has evolved rapidly since its introduction, with both the CPU and GPU backends having grown into production grade Gurobi features. In this session, we will share key insights from this exciting period of algorithmic innovation, including benchmarking results, customer success stories, and an overview of the technical capabilities of Gurobi’s PDHG implementation. We will also discuss where PDHG fits within the broader optimization landscape and the types of problems where it can deliver the greatest value.

Keynote: „Hexaly: the Next-Generation MIP Solver“

Speaker: Léa Blaise

Time: Thursday, September 3, 16:00 – 17:00

Room: TBA

Abstract: Mixed-Integer Linear Programming (MIP) has been the dominant optimization framework for decades. While highly successful, many real-world problems remain difficult to express naturally within a purely linear paradigm. Large-scale routing, scheduling, packing, simulation-based optimization, and machine-learning-driven decision problems often require extensive reformulations that can obscure problem structure and limit performance.
Hexaly extends the traditional MIP approach with a richer modeling framework designed to capture optimization problems closer to their original form. In addition to classical algebraic modeling, it introduces native combinatorial constructs such as sets, lists, and intervals, together with nonlinear and black-box optimization capabilities. This allows practitioners to model complex systems, simulations, and predictive models directly, without relying on large reformulations. At the algorithmic level, Hexaly automatically exploits advanced optimization techniques, including decomposition methods, column generation, branch-price-and-cut, and decision diagrams, making sophisticated solution approaches available out of the box.
In this talk, I will present the principles behind this next-generation MIP approach, illustrate its modeling capabilities on routing, scheduling, and packing applications, and provide an overview of the solver’s current performance on selected benchmarks.