Multi-Agent-War-Simulation-Using-Reinforcement-Learning

This code simulates a virtual war between n agents all of whom use RL. This is a level-1 reasoning type where each agent assumes that their neighbors or opponents use the same reasoning as them (imagine rock-paper-scissors!). Please see the project proposal for domain design and details. The code is completely written in Python and asks for relevant user input. Please reach out to me @[email protected] for any questions. Currently, the code works without any bottlenecks if the number of agents is <= 6. My future work is to scale it 10+ agents.

Instructions:

1. Read the design document
2. Update the settings file as per your liking
3. Run the Run.py file
4. Input the main settings as user input (just for fun :D)
5. Enjoy the war

WIP as of 05/15/2024:

1. Working on visualizing the agent health during both training, and simulation
2. Working on profiling the code to eliminate any potential bottlenecks