What it is about
You will learn methods of global optimization for nonconvex and possibly nonsmooth objective functions and how to apply them. Beginning with (stochastic) gradient methods and simulated annealing, you walk toward multi‑particle methods such as Particle Swarm Optimization (PSO) and Consensus‑Based Optimization (CBO) and understand their global convergence properties.
What you will be able to do
- Understand mathematical foundations of global optimization
- Implement SGD, Simulated Annealing, PSO and CBO
- Analyze and prove global convergence for suitable classes of functions
- Understand the relationships between PSO, CBO and gradient methods
What the module consists of
- LecturesConveying the theoretical results on methods and convergence
- Exercises in lectureProblems are posed during the lecture (no separate exercise sessions)
- Online/In-Person MixLectures alternate between in-person and Zoom; all sessions are streamed and recorded
Teaching method
- Theory LecturePresentation and proof of the mathematical results
- Problems during the lectureDeepening through exercises in the sessions
- Live stream and recordingsEnable participation online and later review