Computational Optimization

IE 586 is about what to do when an optimization model is too big to throw at a solver as-is: how you reformulate it, break it apart, or attack it piece by piece so the structure actually works for you instead of against you. You'll spend the semester implementing decomposition machinery (Dantzig-Wolfe, Benders, Lagrangian) and cutting-plane and column-generation methods on problems from production planning, network design, and logistics, with two homeworks, a midterm, a final, and a project where you pick a real large-scale model and crack it open. It builds on a solid linear/integer programming background and is where the gap between "I can write the LP" and "I can actually solve the LP that matters" gets closed - essential if you're heading into operations research, stochastic programming, or any industrial optimization research.