Reinforcement Learning for Finite Horizon

In financial mathematics, mean field games have been used for modeling large population interactions. Specifically, mean field games have previously been explored through reinforcement Q-learning in infinite horizon. We aim to extend this topic into finite horizon, implementing rewritten algorithms in Python. In this project, we implement the algorithms for two Mean Field Example problems, Capital Accumulation and Price Impact Model. We then further considered the problems in a idealized deterministic system allowing for faster algorithms and a solution for the Mean Field Control of the system.