hanoi tower problem in Q learning (parallel version)
Base code credit to : https://github.com/khpeek/Q-learning-Hanoi/blob/master/hanoi.py
Parallel version inspired by:
Kretchmar, R. Matthew. "Parallel reinforcement learning." The 6th World Conference on Systemics, Cybernetics, and Informatics. 2002.
The implemention is a dumb version, time complexity is somehow huge.
Python 3.6 since we are using asyncio
The right part is the original version, you can see the left desends 10 times faster when we have 10 agents working at the same time.