EigenPy — Versatile and efficient Python bindings between Numpy and Eigen

EigenPy is an open-source framework that allows the binding of the famous Eigen C++ library in Python via Boost.Python.

EigenPy provides:

• full memory sharing between Numpy and Eigen, avoiding memory allocation
• full support Eigen::Ref avoiding memory allocation
• full support of the Eigen::Tensor module
• exposition of the Geometry module of Eigen for easy code prototyping
• standard matrix decomposion routines of Eigen such as the Cholesky, SVD and QR decompositions
• full support of SWIG objects
• full support of runtime declaration of Numpy scalar types
• extended API to expose several STL types and some of their Boost equivalents: optional types, std::pair, maps, variants…
• full support of vectorization between C++ and Python (all the hold objects are properly aligned in memory)

Installation

The installation of EigenPy on your computer is made easy for Linux/BSD, Mac OS X, and Windows environments.

Conda

You simply need this simple line:

conda install eigenpy -c conda-forge

Ubuntu

You can easily install EigenPy from binaries.

Add robotpkg apt repository

1. Register the authentication certificate of robotpkg:

curl http://robotpkg.openrobots.org/packages/debian/robotpkg.asc | sudo tee /etc/apt/keyrings/robotpkg.asc

1. Add robotpkg as source repository to apt:

sudo tee /etc/apt/sources.list.d/robotpkg.list <<EOF
deb [arch=amd64 signed-by=/etc/apt/keyrings/robotpkg.asc] http://robotpkg.openrobots.org/packages/debian/pub $(lsb_release -cs) robotpkg
EOF

1. You need to run at least one apt update to fetch the package descriptions:

sudo apt update

Install EigenPy

1. The installation of EigenPy and its dependencies is made through the line:

sudo apt install robotpkg-py35-eigenpy

where 35 should be replaced by the Python 3, you want to work this (e.g., robotpkg-py36-eigenpy to work with Python 3.6).

Mac OS X

The installation of EigenPy on Mac OS X is made via HomeBrew. You just need to register the tap of the software repository.

brew tap gepetto/homebrew-gepetto

and then install EigenPy for Python 3.x with:

brew install eigenpy

Docker

docker run --rm -it ghcr.io/stack-of-tasks/eigenpy:devel

Build

Build instruction can be found here

Credits

The following people have been involved in the development of EigenPy:

• Justin Carpentier (Inria): main developer and manager of the project
• Nicolas Mansard (LAAS-CNRS): initial project instructor
• Wolfgang Merkt (University of Edinburgh): ROS integration and support
• Sean Yen (Microsoft): Windows integration
• Loïc Estève (Inria): Conda integration

File truncated at 100 lines see the full file

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog.

[Unreleased]

Changed

• Remove accelerate.hpp header that clash with Accelerate.hpp in non case sensitive OS (#593 We don’t consider it an API break since this header was rarely used.

Added

• Support for Python slice, tuple and list indexing for std::vector bindings (#592)
• ROS: jrl_cmakmodules dependency + kilted CI (#602)

Fixed

• Fix partly the support of the change of API of GeneralizedEigenSolver in Eigen 5+ (#594)
• Fix Eigen decompositions and solvers for Eigen 5 (#596)

[3.12.0] - 2025-08-12

Added

• Add additional decompositions and solvers from Eigen (#571)
• Docker images ghcr.io/stack-of-tasks/eigenpy (#575)

Changed

• Change the default branch to devel (#547)

[3.11.0] - 2025-04-25

Added

• Add user-defined literal ““_a for bp::arg (#545)

Fixed

• Fix handling of non sorted sparse matrix (#538)

Changed

• Update clang-format standard to C++11, reformat code

[3.10.3] - 2025-02-11

Added

• Add BUILDING_ROS2_PACKAGE to generate ament specific file (#530)

Changed

• Modernize doxygen documentation (#533)

[3.10.2] - 2025-01-13

Fixed

• Fix Python library linkage for Debug build on Windows (#514)
• Fix np.ones when dtype is a custom user type (#525)

[3.10.1] - 2024-10-30

Added

• Add Pixi support (#444)

Fixed

• Don’t use C++14 feature (#510)
• Add inline to deprecationTypeToPyObj definition to avoid linking error (#512)

[3.10.0] - 2024-09-26

Added

• GenericMapPythonVisitor/StdMapPythonVisitor can now take an extra visitor argument in the expose() method, similar to StdVectorPythonVisitor

Changed

• Move GenericMapPythonVisitor to its own header eigenpy/map.hpp
• Rename overload_base_get_item_for_std_map to overload_base_get_item_for_map, move out of eigenpy::details namespace
• Move EmptyPythonVisitor to new header eigenpy/utils/empty-visitor.hpp

[3.9.1] - 2024-09-19

Added

• Add test returning reference of std::pair (#503)
• Add more general visitor GenericMapPythonVisitor for map types test boost::unordered_map<std::string, int> (#504)
• Support for non-default-contructible types in map types (#504)
• Add type_info helpers (#502)
• Add NumPy 2 support (#496)

Changed

• Move StdMapPythonVisitor out of eigenpy::python namespace, which was a mistake (#504)

[3.9.0] - 2024-08-31

Changed

File truncated at 100 lines see the full file