(资料图)

CliffWalkingEnv.py

PolicyIteration.py

ValueIteration.py

TruncatedPolicyIteration.py

dynamic_programming.py

Results of policy iteratoin, value iteration and truncated policy iteration with iteration of 1, 10 and 100 are respectively shown below.

The above codes mainly refer to Chapter 4 of Hands-on Reinforcement Learning, but some changes have been made based on David Silver's lecture and Shiyu Zhao's Mathematical Foundation of Reinforcement Learning.

[1] https://hrl.boyuai.com/

[2] https://www.davidsilver.uk/teaching/

[3] https://github.com/MathFoundationRL/Book-Mathmatical-Foundation-of-Reinforcement-Learning

标签