Dark Hex is the imperfect information version of the game Hex. This game has a really scarce work on it. In this repository I am including an implementation of the game along with some algorithms implemented specifically for DarkHex. Also I included the results of some of the experiments.

For details on DarkHex and any of the algorithms used as well as the experiments please check my thesis -link-.

• What is Hex
• Dark-Hex Open Spiel Implementation
• Sample Game

• pONE
• Vanilla CFR
• FSI-CFR
• Backward Induction Best Response
• AlphaZero Approximate Best Response
• CFR+
• MCCFR
• NFSP
• Deep CFR

The library is uploaded on Pypi and can be installed using pip. Before pip called make sure you have the following packages installed:

• Python 3.6
• PyGObject - bindings for GTK3
• Graphviz
• GCC
• Libcairo

They are mostly required for graph drawing and setup reasons. Please make sure you have these packages installed before installing the library.

sudo apt-get install libgirepository1.0-dev gcc libcairo2-dev pkg-config gir1.2-gtk-3.0 graphviz

brew install gobject-introspection graphviz cairo pkg-config gtk+3 gcc

After installing the packages you can install the library using pypi.

pip install darkhex

Installing from source would be helpful if you wanted a custom version of the library / helping with development. Make sure you have the packages above installed.

We recommend using a virtual environment for the development.

E.g. for Ubuntu or Debian:

sudo apt-get install virtualenv python3-virtualenv
virtualenv venv
source venv/bin/activate

Rest of the installation is just installing python requirements.

E.g. for Ubuntu or Debian:

# Install the libraries listed above
sudo apt-get install libgirepository1.0-dev gcc libcairo2-dev pkg-config gir1.2-gtk-3.0 graphviz

# Install the python packages
pip install -r requirements.txt

Update PYTHONPATH on your .bashrc or .zshrc file (or .venv/bin/activate)