PyEntropy is a Python library to implement calculation of entropy and information quantities using a range of bias correction methods. This contains the core information routines I use throughout my work, and was produced for eventual implementation as an online service in the CARMEN project. It also includes pyentropy.maxent - a module for computing maximum entropy solutions subject to marginal constraints over finite alphabet probability spaces using the information geometric methods of Amari. It is an open source project hosted at Google code. See Ince et al. (2009) for details. The documentation, including a brief primer on entropy and information theory, is available here.
This module for computing finite alphabet maximum entropy solutions subject to marginal contraints is now distributed as part of the PyEntropy package (pyentropy.maxent).
The Python programming language, together with the numerical libraries NumPy and SciPy, provides a compelling alternative to MATLAB for computational science. Here are some simple scripts to help copying and pasting from MATLAB to Python and from Python to MATLAB.
I have also contributed to pymex, which embeds a
Python interpreter in a MATLAB mex extension and provides an elegent interface
for interacting with Python from the MATLAB shell.
My win64 branch supports 64
bit windows builds with a Microsoft toolchain.
The excellent modern Fortran
interface
to the MATLAB mex API makes extending MATLAB with compiled Fortran extensions
much easier.
My contributions consist of
modifications to support gfortran on Linux as well as the -largeArrayDims 64 bit
interface on all platforms.
Other snippets and thoughts can be found on my blog.