Decision Making in Large Scale Systems
MIT's graduate course on large-scale dynamic programming covers Markov decision processes, dynamic programming algorithms, simulation-based methods, value function approximation, and policy search techniques. Lectures and problem sets work through applications in dynamic resource allocation, finance, and queueing networks, along with game-theoretic extensions to standard decision models. The course targets students who already know probability and optimization and want to see how those tools scale to systems with enormous state spaces, where exact computation becomes infeasible and approximation methods take over. Materials on MIT OpenCourseWare include lecture notes and assignments, distributed under a Creative Commons license, free to access with no certificate offered. It suits engineers and researchers working on control, operations research, or machine learning who need a rigorous grounding in how to make sequential decisions under uncertainty at scale.