Probabilistic Tic-Tac-Toe is an online game developed by Cameron Sun. An optimal solution exists, and was discovered by Louis Abraham. ProbLog is a probabilistic logic programming language developed at KU Leuven. In this project, we model Probabilistic Tic-Tac-Toe in ProbLog, and we implement strategies that we test against the optimal one or a random agent.
Some code is borrowed from Louis Abraham (see his blogpost here) - this is made explicit wherever relevant.
python3 probtictactoe.py; use -h for help.
Requires problog (which might also require setuptools) and multiprocess. You might want to do this in a virtual environment. Please use Python 3.12.
You need to install the Python problog package from here. Since our ProbLog program is very slow, we tried to speed it up via multi-threading - you therefore also need the package multiprocess. Note that it is still quite slow (though at least usable) as simulating even only 30 games takes around 15 minutes (on an M1 MacBook).
On our (Apple Silicon) MacBooks, we had to create it in a virtual environment and run our code in it. This is how we did it:
python3 -m venv ~/.venvin the root directorysource ~/.venv/bin/activate- actually run the program, with
python3 game.py- we also redirected the output tooutput.txt deactivateto exit the virtual environment
Please use Python 3.12.
We are all running Apple Silicon MacBooks (M1 and M2), so maybe this is only relevant to people using similar machines. On an M1 MacBook, we ran into a ModuleNotFoundError: No module named 'distutils' whenever we tried to run the program more than once - to solve this, we found out that it (mostly) worked to uninstall the package setuptools, and then reinstall it again.
This is definitely not an ideal solution, but we did not dig deeper into this as we could still run the program easily on another machine.