[cclib-devel] improved Overlap Population code

[Adam T. <a-t...@st...>]

I just rewrote the OPA code. Previously, any overlap population on  
sizable molecules (say more than 60 basis functions) took FOREVER. I  
removed the loop over molecular orbitals and took advantage of a few  
Numeric features that allow addition and multiplication of entire  
arrays at once. OPA of a molecule with four fragments and 500+ basis  
functions now takes about 7 seconds, which is approximately 6 times  
faster than my old PyMOlyze c-module and possibly 50-100 times faster  
than the old OPA code (maybe a slight exaggeration, but I remember it  
being painfully slow compared to the old PyMOlyze).

Although I haven't extensively tested it, the handful of comparisons  
I've made against older versions give the same results for a few  
molecules. I recommend including it in the 0.6 release, unless you  
think there should be more robust testing.

Adam