Welcome to MMOT2D’s documentation!

This repository contains an implementation of the Multi-Marginal Optimal Transport (MMOT) algorithm described in

2. Zhou and M. Parno (2022) Efficient and Exact Multimarginal Optimal Transport with Pairwise Costs

It is built on top of some of the core functionality (c-transforms and measure transformatiosn) provided by Flavien Leger in the BFM package .

Installation

First install the cairo and jupyterlab packages using conda:

conda install -c conda-forge cairo jupyterlab

Then you can install mmot by cloning this repository and install the package with pip:

git clone git@github.com:simda-muri/mmot.git
pip install -e mmot

If you plan on building the sphinx documentation, you will also need to install sphinx and the pydata-sphinx-theme:

conda install -c conda-forge sphinx pydata-sphinx-theme nbsphinx pandoc ipython ipykernel

Then you can build the documentation using

sphinx-build -b html docs/source/ docs/build/html

Typical Usage

Contents

Examples

• Introduction to MMOT with Pairwise Costs
• Simple Barycenter Computation
• Barycenter of MNIST Digits
• Push Forward Comparison

API Documentation:

• Solver API
• BFM Wrapper API
• Graph Utilities API
  • ConvertToTree()

Indices and tables

• Index

• Module Index

• Search Page