Skip to content
A pythonic wrapper around the NFFT library
Branch: master
Clone or download
ghisvail Merge pull request #58 from aplavin/patch-1
Fix compilation error with NFFT 3.3
Latest commit e2da0af Aug 22, 2017


PyNFFT - Pythonic bindings around the NFFT library

"The NFFT is a C subroutine library for computing the nonequispaced discrete Fourier transform (NDFT) in one or more dimensions, of arbitrary input size, and of complex data."

The NFFT library is licensed under GPLv2.

This wrapper provides a somewhat Pythonic access to some of the core NFFT library functionalities and is largely inspired from the pyFFTW project developped by Henry Gomersall (

The documentation is hosted on pythonhosted, the source code is available on github and the Python package index page is here.


There has been a major API change from version 1.2.x to 1.3 in prevision of version 2.0. Please check the documentation for more information.

The only officially supported platform for version 1.3 is Linux. Support for more platforms is scheduled for the 2.x development tree.


See the tutorial section of the documentation.


Support for pip/easy_install has been added via the Python Package Index. The pyNFFT package can be installed with:

pip install pynfft

Installation will fail if the NFFT library is not installed in a system-aware location. A workaround is to first call pip with:

pip install --no-install pynfft

cd to where pip downloaded the package, then build with

python build_ext -I <path_to_include> -L <path_to_lib>
-R <path_to_lib>

and do a final call to pip:

pip install --no-download pynfft


The pyNFFT package can be built from the cloned git repository by calling:

python build

and then installed with:

python install

The build process requires Cython in order to generate the cythonized c-files:

python build_ext --inplace

Build info

The NFFT library has to be compiled with the --enable-openmp flag to allow for the generation of the threaded version of the library. Without it, any attempt to building the project will fail.


  • Python 2.7 or greater
  • Numpy 1.6 or greater
  • NFFT library 3.2 or greater, compiled with openMP support
  • (optional) Cython 0.12 or greater


See the CONTRIBUTING file.


The pyNFFT project is licensed under the GPLv3. See the bundled COPYING file for more details.

You can’t perform that action at this time.