Re: [cclib-users] Any Gaussian users familiar with PBC?

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

On 14 October 2011 15:56, Manuel Melle Franco <man...@gm...> wrote:
> HI Karol,
>
> Crystal uses K points, I also have my own software that does graphitic
> materials in reciprocal state.
>
> Kpoints are points in reciprocal space typically defined by four
> numbers  kx, ky,kz, wk
> where wk is the relative weight.
>
> the DOS are basically a list of the eigenvalues for each kpoint
> broadened with a gaussian and weighted (in "intensity") by the
> relative weight of each kpoint.
>
> The projected dos, if I remember right, are based on the contribution
> of an atom to the DOS (list of eigenvalues) based on the relative
> mulliken population for each eigenvalue for that atom.
>
> Does it make sense?
>
> I have used gaussian with PBC's but since I know how to use crystal I
> prefer it,otherwise I would have already scripts that do that,
> sorry,:D.

This is very useful. I didn't realise about the relative weight - I'd
better check for that in the Gaussian output.

Is there anything special about k-point 0 0 0? Gaussian seems to treat
that one specially by printing out the full eigenvalues and AO
contributions to the MOs.

My other question is probably dumb. Are the corresponding orbitals at
each k-point identical in terms of composition of the AOs? I realise
that they have different energies, and so I guess the answer is no.
(But if it's no, it means that I can't do the PDOS as Gaussian only
prints out one set of AO contributions, the point at k-point 0 0 0)

> To test you may follow this:
> http://scuseria.rice.edu/gau/gdv_pbc/pbc_guide.html#dos_f2

I have already seen this. It seems to be by the guys who wrote the PBC
code in Gaussian. However, they seem to be using a patched version of
Gaussian in order to get output for the PDOS :-/

- Noel