Python and its scientific computing packages

Python & its scientific computing packages

Python is a relatively new language that has become popular and well supported by a rich set of high level scientific and visualization libraries. The combination of this and the fact that it is an interactive interpreted language means that one can relatively quickly develop useful applications.

This matches the character of scientific software development, especially when in the service of research: highly exploratory, with a need to minimize distractions when trying to articulate complex concepts. Tweaking code for performance and focusing on robust production code gain attention later in the scientific software development cycle.

Rapid development and exploration are why Python was chosen for the course, which is concerned with techniques for building research tools.

This year we will use Python version 2.6, though you will see mention of the most cutting-edge versions (2.7 and 3.0) around.

Tutorials and documentation are available at the Python site. The book Learning Python by Lutz and Ascher (Fourth Edition, O'Reilly Media, 2009) is a good place to get started and learn the language.

Here is the set of scientific computation support packages that we will use.

iPython: iPython is a very convenient editing and development environment for Python.
SciPy and NumPy: SciPy and NumPy provide a fairly complete set of computational tools and basic numerical and linear algebra libraries. Tutorials and documentation are here.
Matplotlib: Matplotlib is a MATLAB-inspired 2D plotting package for Python. A Users Guide, a cookbook, and a PDF users manual are available.
PyGlet: PyGlet is a cross-platform windowing and multimedia library for python. It provides an object-oriented programming interface for visual and sound programming.
MayaVi: MayaVi is a scientific data visualization toolkit for 3D graphics.
Python Imaging Library: Python Imaging Library (PIL) adds image processing to Python and supports a range of image file formats, along with graphics capabilities.

These are the packages you will need to have installed.

In past years, the class has used

Scientific Python: Scientific Python is set of higher-level computational methods than SciPy.
Visual Python: VPython or Visual Python is a 3D graphics environment. The visual package offers real-time 3D output and is easily usable by novice programmers.
Gnuplot: An alternative package is Gnuplot which has been around for some time and provides an extensive set of 2D and some 3D plotting routines. The glue that integrates Gnuplot into Python is in this package.
PyGame: PyGame provides a fast, though primitive, interface for 2D graphics.
ImageMagick: ImageMagick provides a complementary set of utilities for manipulating images.

So you might see code from past class projects or the occasional example which use these. Generally, though we are moving away from these packages, in favor of the above set.

Installation

This year I recommend the Enthought Python Distribution. It is free for academic and educational use and, in a single downloadable installer, it has all of the above packages, and much more.

Importantly, there are installers for Windows (XP, Vista, and 7), Macintosh OS X, and Linux (Ubuntu and RedHat).

Here are some comments on the installation for these platforms.

Macintosh OS X, go here.
Windows, go here.
Linux, go here.