Matheus Portela

This will most likely speed up Q-learning performance.

Something like this would be enough: ```python from setuptools import setup, find_packages setup( name='multiagentrl', version='0.1.1', description='Multiagent Reinforcement Learning framework', classifiers=[ 'Development Status :: 3 - Alpha', 'License :: OSI Approved...

The class diagram must describe the system and its components after refactoring, including detailed descriptions of modules, classes, methods, attributes, and interfaces. This will guide the refactoring efforts.

Now, controller is one single-thread process which has to handle action selection for all agents. Probably, multithreading will make it be faster (although Python's GIL may reduce gains).