Re: [cclib-devel] pseudopotentials

[Noel O'B. <bao...@gm...>]

> > Can you suggest an example case that we could probably run for all
> > programs that would contain one or two heavy atoms with
> > pseudopotentials plus one/two light atoms? What's a very commonly
> > available pseudopotential?
>
> I'm going to suggest [MoOCl4]2-, which is a singlet molecule. I'm
> most familiar with the CEP and LanL2 psuedopotentials, and associated
> basis sets (CEP-121G and LanL2DZ) which I use for Mo; there's lots of
> info on the Gaussian website about this stuff. ADF has it's own set
> of basis functions. I don't know about GAMESS or Jaguar.

I've used LanL2DZ and LanL2MB...the latter is a minimal basis.

> Do we only want single point files to test?
I think so, and I don't think there's any need for the overlap matrix;
or is there? If so, it might be better to reduce the basis to a
minimal basis (e.g. LanL2MB instead of LanL2DZ).

> And do these files go in
> the basic* data subdirectories?
I think so. We are going to be running the basic unit tests on them,
e.g. testpseudo.py. And a good rule of thumb is that anything that's
unit tested goes into basic*, and everything else onto the webserver.

Noel