This is a research repository for neurally guided Monte Carlo tree search (AlphaGo-esque algorithms) for single-player problems (although the focus is on the job shop problem).
It allows for the configuration of a wide variety of algorithmic variation in the different MCTS phases and more. For details, please consult my PhD Thesis. Further information can also be found in my systematic literature review on neural MCTS as well as my paper on neural MCTS for the JSP.
All code is written to be executed from the main directory.
To start an experiment, simply execute
python main.py exp_name
where exp_name is the name of an experiment configuration file in data/config/experiments without the file ending.
For instance: python main.py exp_001