[Q-e-developers] Fwd: Phonon matdyn problem1

[Paolo Giannozzi]

On Fri, Nov 18, 2016 at 8:44 AM, Antoine JAY <antoine.jay at polytechnique.edu>
wrote:

I had trouble performing phonon DOS on a nq=5^3 grid:
> frc_blk','wrong total_weight'
>

I confirm that this is true ...

But the total_weight seems to be calculated through parameters defined in
> the q_gen subroutine (matdyn.f90). In this subroutine, nr1 n2 and nr3 are
> fixed:
> INTEGER, PARAMETER:: nr1=4, nr2=4, nr3=4
> which give in my example, a tota_weight that is 124.1, far from 125 (5^3)
>
> When I changed in q_gen the 4^3 to 5^3 values, I obtained the good weight
> of 125 and was able to run properly the phonon DOS.
>

... and that this is not true. I made the same change and the error is
still there.


> I think the nr1, nr2 and nr3 that are use in q_gen must be the same as the
> one ask by the user and not fixed parameters.
>

I don't think so. There is no relation between the nr1,nr2,nr3 parameters
that define the wavevector grid and those in the q_gen subroutine that
define the largest supercell for the "mass approximation" (that you did not
use anyway).

The problem of occasional "wrong total_weight" errors is well known. Like
for several other known phonon problems, no solution is currently available.

Paolo

-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20161118/d123e3f7/attachment.html>