Dynamic Programming and Reinforcement Learning

4-None

This graduate course provides a comprehensive treatment of dynamic programming (DP) as the fundamental framework for staged optimization problems under uncertainty. Core topics include the principle of optimality, Bellman equations, value and policy iteration, deterministic and stochastic dynamic programming, and Markov decision processes. Applications span linear-quadratic optimal control, inventory control, asset management, hypothesis testing, the Viterbi algorithm, among others. The course concludes with modern reinforcement learning topics including Q-learning, temporal difference learning, and value function approximation, demonstrating how classical DP principles form the theoretical foundation for contemporary AI algorithms.