Re: [cclib-devel] contribute to cclib

[Noel O'B. <noe...@ma...>]

On Tue, 2006-08-29 at 13:26 +0200, Mehdi Bounouar wrote:
> > 
> > We do intend to extract the cartesian displacements for vibrational
> > frequencies - I didn't realise you were talking about this. See:
> > http://sourceforge.net/mailarchive/forum.php?thread_id=29367282&forum_id=48080 for a possible name for this attribute. We intend to add this to the next release, if possible. Are you suggesting something above and beyond this, though?
> 
> Well IMHO I would rather prefer to an option to extract the Force
> constant matrix (FCM), because if I would like to do displacements along a
> vector, I have a problem to trust the outputed vectors (sometimes
> truncated, sometimes even wrong). Also I think having a "unified" parsed 
> FCM, but unified I mean that one does not have to check if it is mass weighted, 
> units etc. 
> Lets say for instance I want to use Molpro because it has nice and quick method 
> to generate to generate a FCM, I could then parse it and use it as a guess for 
> doing a Geom-Opt in GAMESS because I trust the Optimiser Algorithm.
> 
So you're saying that it's possible to calculate the cartesian
displacements from the FCM. How is this done exactly? It's possible that
we will extract both, as many output files will only contain the
cartesian displacements. However, we will provide (separately) a method
to calculate the cartesians from the FCM, if this is possible.

Before you write some specific code, it would be great also if you could
look at the output of the other comp chem packages. Do they already
contain this information? If not, we will have to create further example
input files that *do* contain it. I am about to finish a postdoc in two
days time, after which I will have reduced access to comp chem packages.
Do you have access to any others besides Molpro?

On a different question: looking at the web, I gather you do a lot of
work on vibrational frequencies. Are you familiar with so-called scaled
quantum mechanical (SQM) methods? e.g.
http://dx.doi.org/10.1016/j.theochem.2006.03.025
It would be nice to extract the information necessary for such a study,
and then to implement the method as an algorithm. I remember trying to
do this during my PhD but I didn't have access to any software to do
this, nor did I have the technical background to be able to do this
myself. If the FCM were required for this, I think it would certainly
justify extracting it. Do you know what else is needed?

Regards,
  Noel