Contents
Official source and binary releases
Official releases include source code, and binaries for Mac OS X and Microsoft Windows. For Linux users, third-party party binaries are available (see below).
Note that the Mac OS X binaries work with the Python from python.org, not with the Python provided by Apple.
NumPy
Official releases are on SourceForge download site for numpy.
SciPy
Official releases are on SourceForge download site for scipy.
Unofficial Add-ons
SciKits
Please visit the SciKits portal for additional toolkits, such as delaunay, umfpack, audiolab, learn and many others.
SciPy for .NET
SciPy for .NET can be downloaded from the Enthought repository.
Bleeding-edge repository access
(See also the Developer Zone.)
NumPy
git clone git://github.com/numpy/numpy.git numpy
SciPy
git clone git://github.com/scipy/scipy.git scipy
Also read the building instructions.
Windows 64 bit unofficial releases
Christoph Gohlke has put together an impressive collection of Windows binary installers of recent versions of scientific packages for Python, including amd64 versions of Numpy and Scipy compiled against Intel's MKL, and SciKits:
http://www.lfd.uci.edu/~gohlke/pythonlibs
Linux unofficial releases
As with a lot of open-source software, the best way to fully exploit and contribute to Scipy is to compile it from source. This will guarantee you the latest stable releases and a better support from mailing-lists. However, this can be challenging, and the second best way to run Scipy is to use binaries.
- Many recent Linux distributions -- including Ubuntu, Fedora, and Debian as of Oct 2009 -- ship their own pre-built binary packages for both Numpy and Scipy. Note however that those distributions tend to only update packages during their pre-release cycles, which mean you probably won't have access to latest stable Scipy/Numpy releases this way unless your Linux distribution release was very recent. This is usually acceptable for end-users, although mailing-list questions and bug reports relating to latest releases are highly preferred.
There are also several user-contributed package repositories available for some specific versions or specific distributions, see Installing_SciPy/Linux. This includes many tips and tricks worth reading. The point of the user-contributed packages is to provide support for more Linux distributions (e.g. RedHat, Suse, Ubuntu LTS) or to work around some subtle inconveniences of the Linux-vendor repositories. Therefore, they are almost always preferable over vendor-shipped equivalent releases.
Numpy and Scipy are also included in certain distributions of scientific Python packages, such as the Enthought Python distribution or Python(x,y) (experimental Linux support). Enthought also provides commercial support to Scipy.
Note that binary packages for the mathematical libraries Scipy and Numpy depend on, shipped by Linux distributions, have in some cases showed to be subtly broken. Running Numpy and Scipy test suites with numpy.test() and scipy.test() is recommended, as a first step to confirm that your installation functions properly. If it doesn't, you may want to try another set of binaries if available, or buy some above-mentioned commercial packages.
Mac OS X 64-bit unofficial releases
Mac OS X binaries installed from GitHub by Chris Fonnesbeck: http://fonnesbeck.github.com/ScipySuperpack/
- Make sure you are using OSX 10.7 Lion's preinstalled Python 2.7.1. Note: The Superpack's version detection may fail with other Python distributions (e.g., fink, Darwin Ports), and it will refuse to install.
Download and run the SciPy Superpack for Python 2.7 (64-bit) installation script.
NumPy is included in the Superpack. For best compatibility, make sure you use the version in the Superpack.
- Note that the Chris Fonnesbeck's Superpacks are based on recent Github code, and not the latest official release.
Marketing material
- The Scipy logo
