# Overview

Welcome to PyPy's fork of Numpy, NumPyPy. In order to install, first install PyPy, hints are here http://pypy.org/download.html. Note this is a binary install, no lengthy translation or compilation necessary. Once you have pypy working and feel comfortable using it, you can install our version of the numpy module into a virtual environment in a separate directory:

virtualenv -p /path/to/pypy/bin/pypy /directory/to/try/pypy-numpy
git clone https://bitbucket.org/pypy/numpy.git;
cd numpy; /director/to/try/pypy-numpy/bin/pypy setup.py install


or without a git checkout:

virtualenv -p /path/to/pypy/bin/pypy /directory/to/try/pypy-numpy
/directory/to/try/pypy-numpy/bin/pip install git+https://bitbucket.org/pypy/numpy.git


If you are using PyPy 4.0.1 (and not a nightly build), you must checkout the pypy-4.0.1 tagged revision, so replace the git line above with:

git clone https://bitbucket.org/pypy/numpy.git; git checkout pypy-4.0.1


If you get a message about missing Python.h you must install the pypy-dev package for your system

If you installed to a system directory, you may need to run:

sudo pypy -c 'import numpy'


once to initialize the cffi cached shared objects as root

For now, NumPyPy does not work with PyPy3*, and is not complete. You may get warnings or NotImplemented errors. Please let us know if you get crashes or wrong results.

If you do not have lapack/blas runtimes, it may take over 10 minutes to install, since it needs to build a lapack compatability library. However, you may later install upstream compatible runtimes, and NumPyPy should pick them up automatically the next time you run PyPy.

Also note that the latest version of NumPyPy will probably not run in an older PyPy. Specifically, we require cffi 1.0 or later. Since cffi is baked into PyPy, you cannot update cffi in any version of PyPy (true as of Nov 2015) so there is no recourse but to update PyPy.

NumPy is the fundamental package needed for scientific computing with Python. This package contains:

• a powerful N-dimensional array object
• tools for integrating C/C++ and Fortran code
• useful linear algebra, Fourier transform, and random number capabilities.

It derives from the old Numeric code base and can be used as a replacement for Numeric. It also adds the features introduced by numarray and can be used to replace numarray.