From August 24–28, the Research Campus hosted the block course Computational Integer Optimization as part of its MODAL Academy program. More than 30 students from TU Berlin, FU Berlin, and partner universities in France and Italy participated, together with participants from several MODAL industry partners.
Over five intensive days, the course covered the key algorithmic components of modern mixed-integer programming solvers, including branching, primal heuristics, presolving, cutting planes, symmetry handling, numerics, machine learning, and GPU-based optimization. Lectures were complemented by hands-on coding sessions focusing on how these techniques are implemented in practice.
A particular highlight was the course project: working in ten teams, the participants developed their own MIP solvers for multi-knapsack problems, gradually adding techniques discussed during the week and putting their implementations to the test at the end of the course.