Recommended installation options¶
Summary of options¶
The table below lists other recommended options for installing scientific Python. These options have different features and different strengths, but all of them are known to be well-supported and expected to work well.
This list includes both root-level package managers like RPM or MacPorts (for which Python is just a small part of their content) along with a number of good distributions that essentially just include Python and associated Python packages.
| Distribution | Mac | Linux | Windows | Notes |
|---|---|---|---|---|
| Anaconda | Y | Y | Y | [1] |
| MacPorts | Y | – | – | [2] |
| Homebrew | Y | – | – | [3] |
| RPM, APT | – | Y | – | [4] |
| ActiveState CE | Y | Y | Y | [5] |
| Enthought Canopy | Y | Y | Y | [6], [7] |
| STSci_Python | Y | [8] | Y | [9] |
| yt Project | Y | Y | – | [10] |
| Ureka | Y | Y | – | [11] |
Notes
| [1] | Anaconda is one of the easiest and fastest ways to install a full scientific Python stack. See the section Install core Python for details. |
| [2] | MacPorts has the best built-in support for Python and is generally stable after Mac OS system or security updates. MacPorts has the drawback of being slow to install so that it can take several hours to build a working Python distribution. |
| [3] | Homebrew is a simpler and faster solution for Mac but does not include Python packages and uses the MacOS libraries instead of building them separately. It has been reported that MacOS system updates can break homebrew packages, but many people successfully use this system. |
| [4] | Linux package managers: For recent versions of linux distributions like Ubuntu, the installed Python and supporting packages available through the package manager will be sufficiently current to support science analysis. For a linux distribution like CentOS-5 this is not the case. This option requires root privilege. |
| [5] | ActivePython: The ActiveState Community Python Distribution, has no license requirements and is freely available. It features a package manager tool which installs from the ActiveState PyPM repository. It handles the “difficult” packages like PyQt and SciPy with support for dependency resolution, and also includes most packages from PyPI. |
| [6] | Enthought Canopy: The Enthought Canopy environment is a bundled binary distribution of Python with a large set of useful packages built in. There is also a bundled editor and GUI package manager. The full version of Canopy requires a paid subscription, but it is free for use by students or employees of a degree-granting institution (see license details). |
| [7] | Enthought Canopy Express is the free version and it has a useful subset of packages including NumPy, SciPy, Matplotlib, IPython, Traits, and Chaco. It is available in 32-bit for Mac and Windows and 64-bit for Linux. Qt and PyQt are not available. These are GUI toolkits which are used by a number of useful applications, in particular the IPython Qt console. |
| [8] | Available only as a source install on Linux. |
| [9] | The STSci_Python distribution provides PyRAF, various analysis packages, and the core NumPy, SciPy, and Matplotlib packages. Qt and PyQt are not included. |
| [10] | Provides NumPy, Matplotlib, HDF5, and yt (astrophysical simulation analysis). See Installing yt. |
| [11] | Ureka is a collection of useful astronomy software that is generally centered around Python and IRAF. The software provides everything you need to run the data reduction packages provided by STScI and Gemini. See http://ssb.stsci.edu/ureka/. |