This repository contains the supplementary code for a paper and a master's thesis. The code includes simulations of agents, machine learning components, and tools for evaluating the results using Jupyter notebooks.
This product was developed and tested under Ubuntu 23.04. It utilizes Unity as well as ML-Agents
This project uses Unity version 2021.3.16f1 (check ./ProjectSettings/ProjectVersion.txt). There exists defenders and attackers for Gray Wolf Optimizer (GWO) and Slime Mould Algorithm (SMA) in ./Assets/GWO and ./Assets/SlimeMould. The defender that defend GWO and SMA in combination can be found under ./Assets/MergeTrainer.
For analyzing the produced data by the Unity part Python is used. The Python code utilizes a standard pip-sync setup. The Python Version is specified in ./python_part/.python-version. The code can be found in the ./python_part` folder. Data evaluation is done in ConvexHullAnalysis.ipynb. As the Unity part produces JSONL (JSON line seperated) files there is jsonl2sqlite.py to convert them into SQLite databases. This way more tooling can work with the data directly.
This project serves as supplementary material for a paper and a master's thesis. Currently, there are no plans for external contributions or collaborations on this project.