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 decomposition (SVD and QR decompositions can be added)
• 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.

The Conda approach

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. Add robotpkg as source repository to apt:

sudo sh -c "echo 'deb [arch=amd64] http://robotpkg.openrobots.org/packages/debian/pub $(lsb_release -cs) robotpkg' >> /etc/apt/sources.list.d/robotpkg.list"

1. Register the authentication certificate of robotpkg:

curl http://robotpkg.openrobots.org/packages/debian/robotpkg.key | sudo apt-key add -

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

sudo apt-get 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

Contributing

Standard matrix decomposion routines of Eigen such as the SVD and QR decompositions can be readily added to EigenPy following the example of the Cholesky decomposition that is already implemented. Feel free to open a PR if you wrap them for your use case.

Build/install from source with Pixi

To build EigenPy from source the easiest way is to use Pixi.

Pixi is a cross-platform package management tool for developers that will install all required dependencies in .pixi directory. It’s used by our CI agent so you have the guarantee to get the right dependencies.

Run the following command to install dependencies, configure, build and test the project:

pixi run test

The project will be built in the build directory. You can run pixi shell and build the project with cmake and ninja manually.

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
• Wilson Jallet (Inria/LAAS-CNRS): core developer
• Joris Vaillant (Inria): core developer and manager of the project

If you have taken part in the development of EigenPy, feel free to add your name and contribution here.

Acknowledgments

The development of EigenPy is supported by the Gepetto team @LAAS-CNRS and the Willow team @INRIA.

Changelog

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

The format is based on Keep a Changelog.

Unreleased

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

• The exposeStdVectorEigenSpecificType() template function now takes the vector allocator as a template parameter (#500)

Added

• Add bp::dist to std::map converter (#499)

3.8.2 - 2024-08-26

Fixed

• Fix function signature on Windows (#494)

3.8.1 - 2024-08-25

Fixed

• Fix compatibility issue with NumPy 2.x on Windows (#492)

3.8.0 - 2024-08-14

Added

• Add compatibility with jrl-cmakemodules workspace (#485)
• Remove support of Python 3.7 (#490)

Fixed

• Remove CMake CMP0167 warnings (#487)
• Fix compilation error on armhf (#488)

3.7.0 - 2024-06-11

Added

• Added id() helper to retrieve unique object identifier in Python (#477)
• Expose QR solvers (#478)

3.6.0 - 2024-06-05

Added

• Added a deprecation call policy shortcut (#466)

Fixed

• Fix register_symbolic_link_to_registered_type() for multiple successive registrations (#471)

3.5.1 - 2024-04-25

Fixed

• Allow EigenToPy/EigenFromPy specialization with CL compiler (#462)
• Fix missing include for boost >= 1.85 (#464)

3.5.0 - 2024-04-14

Added

• Allow use of installed JRL-cmakemodule (#446
• Support of Numpy 2.0.0b1 (#448)
• Support new primitive type (char, int8_t, uint8_t, int16_t, uint16_t, uint32_t, uint64_t) (#455https://github.com/stack-of-tasks/eigenpy/pull/455)
• Support conversion between signed <-> unsigned integers (#455)
• Support conversion between complex numbers (#455)

Fixed

• Fix unit test build in C++11 (#442)
• Fix unit test function signature #443)
• Fix CMake export (#446
• Fix int management on Windows (#455)
• Fix long long management on Mac (#455)
• Allow to run test in the build directory on Windows (#457)

Removed

• Remove casting when converting from Eigen scalar to Numpy scalar. This should not remove any functionality since Numpy array are created from the Eigen scalar type (#455)

3.4.0 - 2024-02-26

Added

• Support for Eigen::SparseMatrix types (#426)
• Support for boost::variant types with VariantConverter (#430)
• Support for std::variant types with VariantConverter (#431)
• Support for std::unique_ptr as a return types with StdUniquePtrCallPolicies and boost::python::default_call_policies (#433)
• Support for std::unique_ptr as an internal reference with ReturnInternalStdUniquePtr (#433)
• Support for Eigen::Simplicial{LLT,LDLT} and Eigen::Cholmod{Simplicial,Supernodal}{LLT,LDLT} Cholesky de compositions (#438)
• Switch to ruff for lints, format, and import sort (#441)

Fixed

• Fix the issue of missing exposition of Eigen types with __int64 scalar type (#426)
• Fix namespace use in unittest/std_pair.cpp (#429)
• Fix case of zero-size sparse matrices (#437)

3.3.0 - 2024-01-23

Fixed

• Fix potential memory leak when returning a list from an std::vector or an std::array (423)

3.2.0 - 2023-12-12

Added

• Support for C++11 std::array types (#412)
• Support for std::pair types (#417)

3.1.4 - 2023-11-27

Added

• Add new helper functions to check Tensor support

Fixed

• Fix stub generation on Windows

3.1.3 - 2023-11-09

Fixed

• Install include/eigenpy/registration_class.hpp

3.1.2 - 2023-11-09

Added

• Support Python 3.12 (#391)

Fixed

• Add method to std::vector<MatrixXX> binding even if another library had registered it (#393)

Changed

• CMake minimal version is now 3.10 (#388)

3.1.1 - 2023-07-31

What’s Changed

• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/374
• Fix ROS CI by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/384
• Extended copyable visitor by @cmastalli in https://github.com/stack-of-tasks/eigenpy/pull/383
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/382

3.1.0 - 2023-06-01

What’s Changed

• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/362
• Sync submodule cmake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/363
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/366
• WIP: Expose boost::none_t type (and std::nullopt_t when available) by @ManifoldFR in https://github.com/stack-of-tasks/eigenpy/pull/367
• optional: check registration of none type and optional type before exposing converter by @ManifoldFR in https://github.com/stack-of-tasks/eigenpy/pull/368
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/371
• Sync submodule CMake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/373

3.0.0 - 2023-04-22

What’s Changed

• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/345
• Clean use of namespace bp:: by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/346
• Add full support of Eigen::Tensor by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/348
• Enable scalar template specialization by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/349
• Tests: add user-struct test by @ManifoldFR in https://github.com/stack-of-tasks/eigenpy/pull/350
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/351
• CMake: have python stubs target depend on pywrap target by @ManifoldFR in https://github.com/stack-of-tasks/eigenpy/pull/352
• Sync submodule cmake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/354
• fix INSTALL_RPATH on ROS & OSX by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/355
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/356
• Add util to expose optional types by @ManifoldFR in https://github.com/stack-of-tasks/eigenpy/pull/357
• Fix CI + sync submodule cmake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/358
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/359
• Add example showing how to bind virtual classes, passing to overrides… by @ManifoldFR in https://github.com/stack-of-tasks/eigenpy/pull/360
• Remove support of numpy.matrix class by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/361

2.9.2 - 2023-02-01

What’s Changed

• Fix for Python 3.6 by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/344

2.9.1 - 2023-01-31

What’s Changed

• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/340
• test python 2/3 ubuntu 18/20/22 by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/341
• Add and expose printEigenVersion by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/342
• Fix issue with Boost.Python < 1.71 by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/343

2.9.0 - 2023-01-09

What’s Changed

• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/333
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/334
• Move and update license by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/335
• Add full support of vectorization by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/336
• Sync submodule cmake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/338
• Simplify alignment procedure by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/337
• Remove deprecated EIGENPY_DEFINE_STRUCT_ALLOCATOR_SPECIALIZATION by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/339

2.8.1 - 2022-12-07

What’s Changed

• Fix handling of Numpy blocks on vector types by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/332

2.8.0 - 2022-12-05

What’s Changed

• Modify stride assertion in numpy-map.hpp to be valid for empty vector by @acmiyaguchi in https://github.com/stack-of-tasks/eigenpy/pull/321
• Sync submodule cmake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/323
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/326
• Fix potential issues related to Boost.Python >= 1.80 by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/327
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/328
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/329
• Copy std-vector and std-map from Pinocchio by @ManifoldFR in https://github.com/stack-of-tasks/eigenpy/pull/325

New Contributors

• @acmiyaguchi made their first contribution in https://github.com/stack-of-tasks/eigenpy/pull/321

2.7.14 - 2022-09-11

What’s Changed

• Fix Boost.Python export by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/319

2.7.13 - 2022-09-08

What’s Changed

• Fix packaging by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/311
• add relocatable test by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/313
• [CI] Always run PRERELEASE tests by @wxmerkt in https://github.com/stack-of-tasks/eigenpy/pull/307
• Test https://github.com/jrl-umi3218/jrl-cmakemodules/pull/547 by @wxmerkt in https://github.com/stack-of-tasks/eigenpy/pull/314
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/315
• Sync submodule cmake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/317
• Fix flake8 errors by @wxmerkt in https://github.com/stack-of-tasks/eigenpy/pull/318

2.7.12 - 2022-08-24

What’s Changed

• Sync submodule CMake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/306
• define HAVE_SNPRINTF for windows, fix #309 by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/310
• export SEARCH_FOR_BOOST_PYTHON by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/308

2.7.11 - 2022-08-11

What’s Changed

• CMake: typo by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/301
• FIx issue related to Python 3.11 by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/304
• Sync submodule Cmake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/305

2.7.10 - 2022-07-27

What’s Changed

• pre-commit / cmake-format by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/300
• ci: test packaging by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/298

2.7.9 - 2022-07-27

What’s Changed

• Enhance CI by testing hpp-fcl by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/296
• Sync submodule cmake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/297

2.7.8 - 2022-07-24

What’s Changed

• Fix cross compilation by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/293
• Fix python numpy linking error by @wxmerkt in https://github.com/stack-of-tasks/eigenpy/pull/295

2.7.7 - 2022-07-19

What’s Changed

• Sync submodule cmake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/287
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/288
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/289
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/290
• cmake: relocatable package for recent CMake versions by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/291
• ROS2/Colcon integration for AMENT_PREFIX_PATH and PYTHONPATH by @wxmerkt in https://github.com/stack-of-tasks/eigenpy/pull/292

2.7.6 - 2022-05-22

What’s Changed

• Allow template specialization of getitem by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/286

2.7.5 - 2022-05-20

What’s Changed

• Fix for Refs to dynamic (1, N) blocks by @ManifoldFR in https://github.com/stack-of-tasks/eigenpy/pull/284
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/285

2.7.4 - 2022-05-06

This release fixes a major bug related to Eigen::Ref when using Row Major matrices.

What’s Changed

• unit tests: fix super() for python 2 by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/281
• test/eigen_ref: test using Eigen::Ref as data member by @ManifoldFR in https://github.com/stack-of-tasks/eigenpy/pull/282
• Fix RowMajor case for Eigen::Ref by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/283

2.7.3 - 2022-05-02

What’s Changed

• numpy: avoid deprecated header by @ManifoldFR in https://github.com/stack-of-tasks/eigenpy/pull/280

2.7.2 - 2022-04-22

What’s Changed

• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/276
• ci: autoupdate devel instead of master by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/277
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in https://github.com/stack-of-tasks/eigenpy/pull/278
• Test: modify a matrix block through Python subclass by @ManifoldFR in https://github.com/stack-of-tasks/eigenpy/pull/279

New Contributors

• @pre-commit-ci made their first contribution in https://github.com/stack-of-tasks/eigenpy/pull/276
• @ManifoldFR made their first contribution in https://github.com/stack-of-tasks/eigenpy/pull/279

2.7.1 - 2022-04-09

What’s Changed

• Expose const Eigen::Ref<const … + update pre-commit + correctly export the project version by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/275

2.7.0 - 2022-04-02

What’s Changed

• Export name types by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/273
• Apply pre-commit on all the project by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/274

2.6.11 - 2022-02-25

What’s Changed

• Fix quaternion initialisation segfault under -march=native by @wxmerkt in https://github.com/stack-of-tasks/eigenpy/pull/267
• Fix Quaternion constructor by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/269
• Fix memory issue with Quaternion::normalized by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/271
• Remove useless std::cout by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/272

2.6.10 - 2022-02-02

This new release enhances portability for ROS2 and fixes issues with Eigen::RowMajor matrix types.

What’s Changed

• Prepare ROS2 release by @wxmerkt in https://github.com/stack-of-tasks/eigenpy/pull/264
• Fix RowMajor issues by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/266

2.6.9 - 2021-10-29

Mostly a maintenance release that enhances packaging support and provides fixes for Windows.

What’s Changed

• Fix stubgen by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/260
• Remove useless setup by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/261
• FetchContent on missing submodule by @nim65s in https://github.com/stack-of-tasks/eigenpy/pull/262
• Sync submodule cmake by @jcarpent in https://github.com/stack-of-tasks/eigenpy/pull/263

2.6.8 - 2021-09-05

This new release:

• enhances the compatibility with Boost >= 1.77
• fixes Python doc
• adds the support of Python stubs

2.6.7 - 2021-08-19

This new release provides additional support in user-type for extracting class objects exposed through boost python.

2.6.6 - 2021-08-13

This new release provides:

• extended support of Custom types with Numpy
• extended support of Eigen::{LLT,LDLT} for matrix solution
• MINRES solver support

2.6.5 - 2021-07-30

This new release extends the support of custom types registration within NumPy arrays.

2.6.4 - 2021-05-25

This new release fixes some bugs when compiling with -march=native. It also fixes a bug of include orders.

2.6.3 - 2021-04-16

This new release enhances the support of EigenPy with Numpy >= 1.2.0

2.6.2 - 2021-03-28

Add support of boolean matrices.

2.6.1 - 2021-01-20

This new release fixes a bug related to Quaternion initialization in Python.

2.6.0 - 2021-01-04

This new release provides extended support for arbitrary scalar types.

2.5.0 - 2020-08-25

This new release provides a support of Eigen::Ref to Python with shared memory.

2.4.4 - 2020-08-18

This new release fixes a bug encountered when trying to expose non-square matrix. It also improves the packaging support on Conda.

2.4.3 - 2020-07-20

This new release fixes some packaging issues and removes some CMake warnings.

2.4.2 - 2020-07-17

This new release provides extended support to NumPy 1.19 and more.

2.4.1 - 2020-06-09

This new release improves the packaging of the project by removing any dependency to pkg-config when searching for dependencies. In addition, the Windows compatibility is enhanced.

2.4.0 - 2020-05-25

This new release enables:

• the exposition of user types inside NumPy
• a enhance compatibility with Windows

2.3.2 - 2020-04-23

This new release enforces the compatibility with:

• ROS and Python3
• Ubuntu 20.04
• Eigen >= 3.3.90

2.3.1 - 2020-04-08

This new release fixes some packaging issues for OS X systems.

2.3.0 - 2020-04-03

This new release comes with new features:

• the Geometry module has now full shared memory with Python

and comes with additional bug fixes:

• the project can still be used with former CMake style
• the compilations issues with Boost 1.71 has been fixed
• on OS X systems, the project can be relocated

2.2.2 - 2020-03-30

This release fixes a packaging with former versions of ROS.

2.2.1 - 2020-03-27

This new release of EigenPy:

• removes boring compilation warnings in C++11
• fixes a bug in the Python documentation of the Geometry module
• fixes the CMake export of the project with respect to ROS

2.2.0 - 2020-03-18

This new release fully introduces the support of sharing of memory from Eigen to Numpy. This feature can be disabled thanks to “eigenpy.sharedMemory(False)” to recover the previous behavior.

In addition, eigenpy::Refhas been removed, as Eigen::Ref is now fully supported.

2.1.2 - 2020-02-28

This release fixes a packaging bug with older version of CMake.

2.1.1 - 2020-02-27

This new release fixes the support with Windows systems and fixes some bugs with the CMake packaging of the project.

2.1.0 - 2020-02-25

This new release adds the full support for:

• sharing the memory between Eigen and Numpy
• provides a complete way of exposing Eigen::Ref (with allocation only when needed)
• improves the whole efficiency of the code

2.0.3 - 2020-02-20

This new release fixes:

• improve the search of Python
• fix a potential memory leak when using Eigen::MatrixBase objects

2.0.2 - 2020-02-06

This new release provides some additional supports on Eigen::Matrix to np.array conversion. It also reduces the memory overload when compiling EigenPy.

2.0.1 - 2020-01-31

This new release improves the compatibility of EigenPy with Win32 systems.

2.0.0 - 2020-01-30

This is the new release of EigenPy, which some new features:

• full support of np.array which becomes the default conversion format. The end-user is still able to switch to numpy.matrix by calling eigenpy.switchToNumpyMatrix()
• we provide a full exposition of the Eigen decompositions (LLT, LDLT, EigenSolvers, etc.)
• fixes have been done that were appearing in some very particular cases

1.6.13 - 2020-01-10

This new release uniformizes the build of unitary tests and also provides a new function to set the random seed of the std random generator.

1.6.12 - 2019-12-10

This release fixes the compatibility on Win32 systems and provides a fix for a bug introduced in 1.6.10.

1.6.11 - 2019-12-09

This new release fixes an important bug introduced in 1.6.10.

1.6.10 - 2019-12-09

This new release:

• makes the project fully compatible with CMake package policy
• improves the documentation of the Python bindings
• fixes the convertibility of Eigen base classes

1.6.9 - 2019-11-25

Missing update of the package.xml file.

1.6.8 - 2019-11-25

This new release improves the compatibility version between numpy.array and Eigen::Matrix conversions. It also fixes some compilations issues on Win32 systems.

1.6.7 - 2019-11-15

This new release fixes some compilations of EigenPy on the ROS build farm. Thanks to [@wxmerkt](https://github.com/wxmerkt] for the fix.

1.6.6 - 2019-11-13

This new patched release improves the packaging of the project with respect to external forges.

1.6.5 - 2019-11-08

This new release fixes the export of the Eigen project for CMake targets relying on EigenPy. It also provides some fixes aroung Geometry classes.

1.6.4 - 2019-11-07

This new release improves the packaging of the project with respect to ROS and also fixes some constructors issues with respect to the Quaternion class.

1.6.3 - 2019-10-29

This release fix issues introduced recently in the packaging which was preventing the project to correctly export the main library.

1.6.2 - 2019-10-24

This is a maintenance release that improves the packaging for Windows and ROS. It also fixes some bugs in the bindings of Quaternion.

1.6.1 - 2019-10-16

This new release provides default exposition of Eigen vector and matrix which are common. On the packaging side, the project is now fully compatible with the new CMake rules for defining and calling the PROJECT master function.

1.6.0 - 2019-09-19

This new release makes a step towards removing the conversion of Eigen::Matrix objects into numpy.matrix objects. This new release now warms by default, saying that you need to make an explicit choice between numpy.matrix and numpy.array.

Future major releases will enforce the default case with numpy.array.

1.5.8 - 2019-09-09

This new release mostly removes a useless print when loading a converter.

1.5.7 - 2019-07-19

This new release adds the support of Windows OS (thanks to @seanyen. It also provides some fixes with respect to recent versions of Boost.Python (>= 1.70.0).

1.5.6 - 2019-07-16

This release fixes some compatibility issues with anaconda.

1.5.5 - 2019-07-15

This new release improves the packaging with respect to Boost >= 1.70.0. This is needed for complete integration inside Anaconda.

1.5.4 - 2019-07-13

This new release mostly improves the packaging of the project, also with ROS.

1.5.3 - 2019-06-28

This new release fixes a serious bug with duplicates symbols.

1.5.2 - 2019-06-26

This new release fixes two important bugs:

• a type already registered by another library will have a symbolic link in the current scope
• when using the array convention, the Eigen vector objects become flatten numpy.array

1.5.1 - 2019-04-16

This new release fixes a serious bug related to Python3 which was occurring when exiting the interpreter. This new release makes also official the new BSD license.

1.5.0 - 2018-10-29

This new release allows now to support the conversion from C++ to Python either as numpy.matrix or as numpy.array.

The use can switch between both features by calling in Python

eigenpy.switchToNumpyArray()

to convert Eigen::Matrix to numpy.array. Or

eigenpy.switchToNumpyMatrix()

to select conversion from Eigen::Matrix to numpy.matrix.

Thanks to [@jviereck](https://github.com/jviereck] for raising the lost of performances induced by numpy.matrix.

1.4.5 - 2018-08-29

Fix compatibility with Eigen 3.3.5

1.4.4 - 2018-07-19

This is a minor release. It allows to register Eigen::MatrixBase<MatType> from Python to Eigen converters.

1.4.3 - 2018-05-12

This release fixes several compatibility issues at the Cmake level for both Python 3 and recent versions of Boost >= 1.67.0.

1.4.2 - 2018-05-02

This new release allows the binding of Eigen::Ref, facilitates the conversion between compatible scalars (eigenpy won’t warn if the numpy matrix has integer scalar type). This new release also improves the compatibility with Python 3.x.

1.4.1 - 2018-02-26

This is mostly a maintenance release where some fixes have been done with respect to BSD systems. We also fixed an issue concerning the catching of exceptions: they are now verbose.

1.4.0 - 2018-01-14

This new release introduces an independent eigenpy library that exposes Eigen solvers in Python. It also allows to use directly NumPy memory through famous Eigen::Ref that are just memory mapping.

1.3.3 - 2017-06-09

It is mostly a maintenance release with suppression of warnings and use of correct types.

1.3.2 - 2016-11-21

Summary

This new release allows the check of registration of any class. This allows to not define twice the same symbols. Thanks to the macro defined in memory.hpp, one can now relies on aligned vector and matrices. The unaligned equivalent type is no more required. This release also improves the exposed API of Quaternions and AngleAxis Eigen classes. The install of documentation can be avoided by setting cmake option INSTALL_DOCUMENTATION to OFF.

Bug Fix

• the Python function _import_array() must be called first before any class to PyArray functions.

1.3.1 - 2016-09-23

Summary

This new release fixes several bugs encounter in the linkage of EigenPy with other libs integrating Python interpreter. It also improve the packaging of the module by removing the dependencies to the Python lib and making Boost Python defined with undefined symbols (useful for OS X - the library is no more static on OS X).

Bug Fix

• UnalignedEquivalent struct take now an optional Scalar template
• PyMatrixType is now a singleton, avoiding bad initialisation during dynamic loading of shared lib.

1.3.0 - 2016-02-03

Summary

Alignment of Eigen and Numpy objects is now properly handled. One can now easily expose C++ struct containing Eigen objects in Python avoiding any unnecessary conversion and data are aligned in memory.

Bug Fix

• Fix issue in the creation of row and column vectors.
• The library has to be static under OS X to properly expose symbols.

1.2.0 - 2014-11-13

1.1.0 - 2014-09-16

1.0.1 - 2014-07-18

1.0.0 - 2014-07-18