POT
Python Optimal Transport Library
Decision gist · record as of 2026-08-14
Yes. POT is actively maintained, production-stable (Development Status 5), has no known vulnerabilities, and offers a comprehensive suite of optimal transport algorithms backed by research. The medium install friction is manageable with prebuilt wheels. Install if you need optimal transport solvers for machine learning, domain adaptation, or distribution comparison; skip if you have no use for these mathematical tools.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires numpy and scipy.
- Compiled wheels available for Python 3.7–3.13; source build requires C/C++ compiler and Cython.
- Medium install friction due to compiled components (C, C++, Cython).
License · maintenance · safety
MIT (permissive) — MIT license (permissive) allows commercial and private use with minimal restrictions—include a copy of the license and copyright notice.
last release 2026-07-29 (16 days) · last repo commit 2026-07-29 · 2,837 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 587,786 downloads/mo, #5,872 on PyPI
Alternatives
Verify before relying
import pot
import numpy as np
# Compute Wasserstein distance between two distributions
a = np.array([0.5, 0.5])
b = np.array([0.3, 0.7])
M = np.array([[0, 1], [1, 0]]) # cost matrix
wasserstein_dist = pot.emd(a, b, M)- Whether all advertised solvers (e.g., Graph Neural Network layers, all backend integrations) are production-ready or still experimental.
- Performance characteristics and scalability limits for large-scale problems mentioned in the description.
What it is and what it does
POT is a Python library for solving optimal transport problems—a mathematical framework for measuring and computing distances between distributions. It provides a large collection of differentiable solvers covering exact linear OT, entropic and quadratic regularized variants, Gromov-Wasserstein distances, unbalanced and partial OT, and specialized algorithms for 1D, circular, and Gaussian mixture model transport. The library also includes machine learning applications such as domain adaptation, transport mapping, and subspace learning.
The package depends on numpy and scipy for numerical computation and offers multiple backend support (PyTorch, JAX, TensorFlow, CuPy) to work with different array types. It is actively maintained, production-stable, and widely used in research and applied settings where distribution comparison, alignment, or interpolation is needed.
Use it for
- Compute Wasserstein distances and optimal transport plans between point clouds or distributions for metric learning.
- Perform domain adaptation by aligning source and target distributions in machine learning pipelines.
- Calculate Gromov-Wasserstein distances to compare structured data like graphs or point clouds with different geometries.
- Compute Wasserstein barycenters to find representative distributions or aggregate multiple datasets.
- Solve unbalanced or partial optimal transport problems when mass conservation is relaxed.
- Integrate OT solvers into neural networks via differentiable layers for end-to-end learning.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes.
POT is actively maintained, production-stable (Development Status 5), has no known vulnerabilities, and offers a comprehensive suite of optimal transport algorithms backed by research. The medium install friction is manageable with prebuilt wheels. Install if you need optimal transport solvers for machine learning, domain adaptation, or distribution comparison; skip if you have no use for these mathematical tools.
Install
pot on PyPI
Before you install
Medium install friction due to compiled components (C, C++, Cython). Prebuilt wheels available for Python 3.10–3.13 on macOS, Linux, and Windows. Active maintenance with recent release (16 days old) and 2837 GitHub stars.
Requires numpy and scipy. Compiled wheels available for Python 3.7–3.13; source build requires C/C++ compiler and Cython.
License in practice
MIT license (permissive) allows commercial and private use with minimal restrictions—include a copy of the license and copyright notice.
Quickstart
import pot
import numpy as np
# Compute Wasserstein distance between two distributions
a = np.array([0.5, 0.5])
b = np.array([0.3, 0.7])
M = np.array([[0, 1], [1, 0]]) # cost matrix
wasserstein_dist = pot.emd(a, b, M)Verify before relying
- Whether all advertised solvers (e.g., Graph Neural Network layers, all backend integrations) are production-ready or still experimental.
- Performance characteristics and scalability limits for large-scale problems mentioned in the description.
Package facts
| License | MIT permissive |
| Python support | Supports the current Python release >=3.7 |
| Install friction | Medium. Platform-specific wheel |
| Runtime dependencies | 2 packagesnumpyscipy |
| Maintenance | Actively maintained 16 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 587,786 / month, #5,872 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | Development Status :: 5 - Production/StableEnvironment :: ConsoleIntended Audience :: DevelopersIntended Audience :: EducationIntended Audience :: Science/ResearchLicense :: OSI Approved :: MIT LicenseOperating System :: MacOSOperating System :: Microsoft :: WindowsOperating System :: OS IndependentOperating System :: POSIXOperating System :: POSIX :: LinuxProgramming Language :: CProgramming Language :: C++Programming Language :: CythonProgramming Language :: PythonProgramming Language :: Python :: 3Programming Language :: Python :: 3.10Programming Language :: Python :: 3.11Programming Language :: Python :: 3.12Programming Language :: Python :: 3.7Programming Language :: Python :: 3.8Programming Language :: Python :: 3.9Topic :: Scientific/Engineering :: Artificial IntelligenceTopic :: Scientific/Engineering :: Information AnalysisTopic :: Scientific/Engineering :: MathematicsTopic :: Utilities |
Evidence: pot-0.9.7.post1-cp310-cp310-macosx_10_9_universal2.whl; pot-0.9.7.post1-cp310-cp310-macosx_10_9_x86_64.whl; pot-0.9.7.post1-cp310-cp310-macosx_11_0_arm64.whl; pot-0.9.7.post1-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl; pot-0.9.7.post1-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; pot-0.9.7.post1-cp310-cp310-win_amd64.whl; pot-0.9.7.post1-cp311-cp311-macosx_10_9_universal2.whl; pot-0.9.7.post1-cp311-cp311-macosx_10_9_x86_64.whl; pot-0.9.7.post1-cp311-cp311-macosx_11_0_arm64.whl; pot-0.9.7.post1-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl; pot-0.9.7.post1-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; pot-0.9.7.post1-cp311-cp311-win_amd64.whl; pot-0.9.7.post1-cp312-cp312-macosx_10_13_universal2.whl; pot-0.9.7.post1-cp312-cp312-macosx_10_13_x86_64.whl; pot-0.9.7.post1-cp312-cp312-macosx_11_0_arm64.whl; pot-0.9.7.post1-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl; pot-0.9.7.post1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; pot-0.9.7.post1-cp312-cp312-win_amd64.whl; pot-0.9.7.post1-cp313-cp313-macosx_10_13_universal2.whl; pot-0.9.7.post1-cp313-cp313-macosx_10_13_x86_64.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “optimal transport solver”
- POTPOT provides solvers for optimal transport problems, including…
- lapxSolves linear assignment problems using Jonker-Volgenant and related…
- munkresImplements the Munkres algorithm (Hungarian algorithm) to solve the…
Give your agent the search over MCP, or paste the wish link into any chat.
More Utilities packages
Converts domain names between Unicode and ASCII-compatible encoding (Punycode) according to IDNA 2008 and Unicode Technical Standard 46, with security validation and broader script coverage than the standard library.
Install it if you work with internationalized domain names, need to validate domains, or use HTTP clients that depend on it transitively.
Detects and normalizes text encoding from unknown or ambiguous sources, supporting all IANA character sets that Python's core library provides codecs for, with the ability to register custom codecs.
Setuptools is a Python build backend and package management tool that handles building, distributing, and installing Python packages, including support for C/C++ extension modules.
Pluggy provides a plugin system that lets you define hook specifications and register implementations to be called in sequence, enabling extensible Python applications without tight coupling.
Install it if you're building an extensible application or framework.
Pygments is a syntax highlighter that colorizes source code and text in over 500 languages and formats, outputting to HTML, LaTeX, RTF, SVG, images, or ANSI terminal sequences.
Install it if you need to display or transform source code.
Six provides utility functions to write Python code that runs on both Python 2.7 and Python 3.3+, smoothing over language differences between the two versions.
See also ott-jax · geomloss · ortools · ropwr · pyamg · pyroots · torchdiffeq · Mosek · quadprog · munkres