The ABINIT Users Mailing List ( CLOSED )

[Xavier Gonze <gonze@pcpm.ucl.ac.be>]

Dear Silvio,

Silvio a Beccara wrote:
> Hello everybody,
>
> I'm trying to do some simple response function phonon calculation on Ge, but 
> the tutorial is not clear about some important points:
>
> -- one (from the Tutorial, lesson 5): "One must distinguish two cases. Either 
> the q wavevector connects k points that belong to the same grid, or the 
> wavevector q is general." What should one do in the two different cases? 
> Should I calculate the difference of all the k points I'm using, and see 
> whether they correspond to a q vector? Isn't there a way to distinguish the 
> cases automatically?

The tutorial explains the view point of "abinis" : one supposes that
the user wants the computation of one well-defined q point (for phonon
perturbation), using one well-defined k point grid (for electronic 
wavefunctions). Then, there are indeed two different cases,
depending on that the points k+q belongs to the set of k points or not.
If they belong to it, then, the wavefunctions at k+q are already
available to the response-functional part of ABINIT. If not,
the wavefunctions at k+q must be provided by the user.
The code distinguishes the two cases automatically, and complains
in the case the k+q grid should be given by the user, and is not...
But the code will not automatically compute the wavefunctions in that
case...

> -- two: when is the density (command prtden) needed for a phonon calculation?

Well, I think it is never needed, in the present status of the code !

Still, it is very often present in the input files of the tests cases. 
This is because in a previous version, one had to do
additional non-self-consistent computations. Now, it is confusing ...

OK, it will be corrected in v3.4.3 .

> -- three: how should the dynamical matrices be read? What do the keywords 
> "j1, j2, dir, pert" mean? 

They make reference to idir and ipert for the 1st perturbation
and the second perturbation. Mmm, this might also be explained
in more detail in the ~Infos/respfn_help.htm file.

> Can the code also calculate, and give out, phonon 
> eigenvectors (needed for vibrational dynamics calculation)?

The IFC code does this. Please, read the associated doc :
http://www.abinit.org/ABINIT/Infos_v3.4/ifc_help.htm#eivec

> --is there a way to automatically give in a list of wanted q points (needed 
> for vibrational Brillouin zone integration)?

This is clearly a weak point of the present setting.
Either one establishes the set of q point beforehand,
or one uses IFC repeatively, by filling one q at a time.

I will look whether it is possible for IFC to give
the full set of needed q points, instead of one at a time.
After all, this should not be complicated ...

Good continuation,
Xavier