Written by Walter
I'm looking into python more and more and found a possible matrix implementation alternative to boost (currently we use boost::numeric). Problem with boost is it's efficiency, my own implementation is already faster but I'm coping with a lot of work still to be done to get all features up and running (slices and efficient linear algebra routines seem harder to implement properly than expected).
So maybe calling python code might save me a bunch of work.
Anyway here are some interesting links on the topic :
1. Matplotlib (some nice plotting functions)
2. Numpy (n-dim array with slices in python are waiting to be explored here :) )
3. Linear algebra vs LinAlg.py (lapack alternatives and/or implementations in python)
4. IPython a development environment similar to MPL based on python, maybe I'll find some inspiration here too...
5. A manual describing how to implement matlab code in python numpy. This was very interesting to read : MATLAB synonymous commands in numerical Python (NumPy). One idea that pops in my head when reading this is that it would be fairly easy to use mpl's parser to convert matlab code to equivalent numpy code (hardest part is that indexing in numpy is from 0 instead of from 1. I will look at some benchmarks of numpy vs matlab first to see if it's feasible to try it. Also I will have to look at how to extend numpy to use our new types like Rational and MpIeee should it be possible it is definitely worth the effort!
More interesting stuff can be found on www.scipy.org