The ABINIT Users Mailing List ( CLOSED )

[Matthieu Verstraete <mjv500@york.ac.uk>]

Hi Joe,

> I am computing the phonon spectrum of a cubic solid (with a large unit 
> cell). I am a little puzzled about the input variable "ngqpt" for 
> anaddb. I computed the phonons using kptrlatt 2 2 -2 -2 2 -2 -2 2 2,
This should be fine, but it has been pointed out that to be really 
consistent (esp. to avoid small numerical errors) the qpoints you use 
should always be differences of kpoints you actually calculate on your 
initial grid. In principle, though, the formalism works with any q, and 
abinit allows this, so you should be fine.

> and the set of qpts obtained from this as in the tutorial.
which are these?

> There were six such q-points, including gamma. What is the proper input 
> for anaddb? ngqpt 2 2 2, or something else?
Depends which qpoints you used, as above. An easy way to get the 
irred qpoints you actually need to calculate is to do an abinit run with 
(say you want ngQpt 4 4 4)

ngKpt 4 4 4
nshiftk 1
shiftk 0 0 0
kptopt 1

stop the calculation after the headers get printed and the kpoints for 
this run are the qpoints you need.

Be careful in all cases (k and qpoints) to use grids which are fully 
consistent with the symmetries of your system. for FCC this gives the 4 
shifts:
1/2 1/2 1/2
1/2 0 0
0 1/2 0
0 0 1/2

and so on for BCC, etc... See the variable definition for nshiftk. 
Otherwise, abinit runs, but by applying a symop you get a kpoint which is 
not on the grid. Ugly. The kgrid you mention above looks like it's for an 
FCC system, right?

> I am afraid I made a mistake with my initial choice of kptrlatt, and it 
> is not really compatible with what anaddb can do. Any advice is welcome.
as mentioned as well, any qpoints should be calculable based on any 
kgrid, provided you also have all the k+q wavefunctions.

Ciao

Matthieu

--
================================================================
Dr. Matthieu Verstraete                 mailto:mjv500@york.ac.uk
Dept. of Physics, University of York,     tel: +44 1904 43 22 08
Heslington, YO10 5DD York, United Kingdom fax: +44 1904 43 22 14