# PythonOT/POT

POT : Python Optimal Transport

Repository: https://github.com/PythonOT/POT
Canonical: https://ross.abutalabs.com/products/pot
Homepage: https://PythonOT.github.io/
Language: Python
License: MIT
License Family: permissive
Topics: optimal-transport, numerical-optimization, machine-learning, emd, ot-mapping-estimation, wasserstein-barycenter, ot-solver, python, wasserstein, wasserstein-discriminant-analysis, gromov-wasserstein, wasserstein-barycenters, sinkhorn-divergences, sinkhorn-knopp, pot, domain-adaptation
Last push: 2026-08-26T10:48:50+00:00

## Health v2 (maintenance only)
Score: 89/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 71, longevity 100
- inputs: {"age_days": 3604, "days_push": 7, "days_rel": 35, "gap_med": 154.0, "n_releases_24m": 5}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2837, forks 556 (observed 2026-08-28T04:07:24.626202+00:00)

## What it is
POT is an open-source Python library providing a large set of differentiable solvers for optimal transport problems, including exact and regularized OT, Sinkhorn divergences, Gromov-Wasserstein distances, and Wasserstein barycenters. It supports multiple array backends (NumPy, PyTorch, JAX, TensorFlow, CuPy) and includes machine-learning solvers such as domain adaptation and OT mapping estimation.

## Use cases
- compute Wasserstein distance between distributions in python
- solve earth movers distance between histograms
- compute sinkhorn divergence with pytorch autograd
- estimate gromov-wasserstein distance between graphs
- compute wasserstein barycenter of distributions
- perform domain adaptation with optimal transport
- differentiable optimal transport loss for deep learning

## When to choose
- you need optimal transport solvers in Python with differentiability for ML pipelines
- you want Gromov-Wasserstein, unbalanced, or partial OT in one library
- you need OT computations across NumPy, PyTorch, JAX, TensorFlow, or CuPy backends

## When to avoid
- you need a non-Python or standalone C++/GPU-only OT implementation
- your problem is unrelated to distribution comparison or transport
- you need a lightweight dependency-free solution

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, math, data-science
- domain: machine-learning, data-science, image-processing
- platform: python, cross-platform
- tags: optimal-transport, wasserstein, sinkhorn, gromov-wasserstein, emd, domain-adaptation, barycenter, differentiable, optimization, algorithms, gpu

## Member repositories
- PythonOT/POT (main) score 89

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:24.626202+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T07:37:29.295601+00:00, confidence not recorded.
  - readme: https://github.com/PythonOT/POT (fetched 2026-08-28T04:07:24.626202+00:00, sha 8f7db8137fd5)
  - homepage: https://PythonOT.github.io/ (fetched 2026-08-29T09:53:04.584743+00:00, sha 0a44a5ac4c1d)
- Data as of 2026-08-30T08:39:29.467469+00:00.