Scientific Computing in Python

Mike Hadmack - August 4th 2006

Here is a collection of references to help set up a working Python system with the necessary scientific libraries

On any UNIX based system Python should be installed by default. You can easily check the version number by typing 'python' at any shell prompt. It is best to have an up to date version of Python, the latest is 2.4.3. To exit python and return to the shell type 'CTRL-D'.

Setting up Python

Mac Users

The default python installation is 2.3 and you must upgrade to version 2.4.3. There is a good article at the BMI Students Blog that describes every step in detail. This article was intended for users of PowerPC Macs. For Intel Mac users you may need to download a special Universal Binary version of python 2.4.3.

Linux Users

You most likely have a relatively new version of Python installed with your system by default. Give the command 'python' a try at the command line.

Windows Users

There is a complete win32 distribution of python available at http://www.python.org/download/. Just download and install. Refer to the included instructions for how to run your new Python installation.

All Users

Now that you have Python 2.4.3 installed. You must now install the scientific packages:

• numpy The default arrays in Python are designed for flexibility in programming not for the performance demanded by scientific computing. That's where numpy comes in providing powerful arrays implemented in C for faster performance. Download at numpy.scipy.org/. Unpack and run 'python setup.py install'.
• scipy. Adds a wide spectrum of scientific utilities to Python such as numerical integration, interpolation, fast fourier transforms, special functions. Scipy also offers a package called 'weave' which allows the integration of C code into your Python code for better performance when you really need it. Refer to the scipy installation guide.
• pylab (MatPlotLib): A Matlab-like plotting environment for python. Uses the same plotting syntax as Matlab so that combined with the tools above provides a relatively complete alternative to Matlab and its brethren. Go to the matplotlib page for installation files and instructions

Code Samples

• hello.py
• temps.py
• simple_plot.py

Other Resources

• The Python Programming Language web site
• An introduction to Python, with application to scientific computing Thanks to Chris Myers' web page, Cornell Theory Center
• Python Imaging Library Allows for advanced image processing and drawing in Python
• Python 2.4 Quick Reference A good introduction to Python for experienced programmers
• Numeric Documentation Documentation to numpy's predecessor Numeric which is very similar in operation. Numpy documentation is presently fee based to cover development costs.
• How to Think Like a Computer Scientist: Learning with Python A good introduction to Python and computer programming in general
• SciPy Tutorial An introduction to working with Scipy
• SciPy Cookbook Many examples of applying scipy to specific scientific programming problems
• pylab examples A variety of examples of plotting in pylab
• More of my Python links at del.icio.us