MinCompSpin (mcmpy): a package for modeling discrete datasets with minimally complex models

MinCompSpin is a C++ library with Python bindings for analyzing discrete datasets using Minimally Complex Models (MCMs).
Once compiled, it creates the Python package mcmpy.

The package was developed for the paper Modeling discrete data with high-order vector Potts models (arXiv:2606.03429). More information about the method and the algorithms can be found in the paper.

The package can be used to analyze datasets with up to 128 discrete variables and for variables that can take up to q=10 different states (though it was extensively tested only up to q=5).

All information about the package (installation, available objects/functions, and how to use them) can be found in the documentation.

The documentation also includes step-by-step examples, with Jupyter notebooks available in the folder docs/examples (in this repository).