pyotc

A python implementations of optimal transport coupling algorithms.

Documentation

Find sphinx documentation here.

Installation

We expect pyotc to be pip-installable across all platforms.

1. Install from pypi (Recommended)

pip install pyotc

Note: pyotc requires Python 3.10 or above.

2. Install from github

pip install https://github.com/pyotc/pyotc.git

3. Install for Development

We test in venvs provided by uv via nox. It’s helpful, but not strictly necessary to do the same.

git clone https://github.com/pyotc/pyotc.git
cd pyotc
pip install -e .

`uv` workflow

Install the uv tool. Then

git clone https://github.com/pyotc/pyotc.git
cd pyotc
uv sync
uv pip install -e .

To verify your installation, run

uv run pytest

Run Tests

With a uv setup one can simply

uv run pytest

Otherwise, in pip installed context with deps met, pytest should behave as expected.

Contributing

Guidelines for contribution to pyotc are provided in CONTRIBUTING.md.

Changelog

A summary of changes and guide to versioning are recoreded in CHANGELOG.md.

Citing this Repository

If you wish to cite our work, please use the following BibTeX code:

@article{yi2025alignment,
  title={Alignment and comparison of directed networks via transition couplings of random walks},
  author={Yi, Bongsoo and O'Connor, Kevin and McGoff, Kevin and Nobel, Andrew B},
  journal={Journal of the Royal Statistical Society Series B: Statistical Methodology},
  pages={qkae085},
  year={2024},
  doi = {10.1093/jrsssb/qkae085}
}