[Q-e-developers] Problem with PAW, 4 sites

[Kucukbenli Emine]

Hi Davide,
Could you link the mentioned pseudos or report the differences other than one being an ultrasoft and the other being paw dataset?
cheers
emine
________________________________________
From: q-e-developers-bounces at qe-forge.org [q-e-developers-bounces at qe-forge.org] on behalf of Davide Ceresoli [davide.ceresoli at istm.cnr.it]
Sent: Wednesday, February 05, 2014 2:39 PM
To: q-e-developers at qe-forge.org
Subject: [Q-e-developers] Problem with PAW, 4 sites

Dear all,
     I don't understand if this I'm doing a big mistake or if it is
a problem of PW. I've a metallorganic compound with 4 Co ions per cell.

The cell is monoclinic but I use ibrav=14. The Co ions are in special
positions: (0,0,0), (1/2,1/2,0), (1/2,0,1/2), (0,1/2,1/2). Their chemical
environment leaves two inequivalent Co-pairs (i.e. Co1 equiv. to Co2,
Co3 equiv. to Co4). In a ferromagnetic calculation, I expected only two
values of magnetic moment per site.

Using PAW pseudopotentials, I've got 4 different values, and the SCF
calculation will not converge (SCF accuracy ~ 1e-4, never decreasing).
However, if I use USPP pseudopotentials, the calculation evolves as
expected. In the attached pictures, I've plotted the magnetic moment per
site, as a function of SCF step (in USPP.png, the values are almost
perfectly overlapping). In both calculations I've disabled symmetry
(nosym = .true.). I can provide input/output files.

Am I missing some thing that I should take into account when using
PAW pseudos with low-symmetry structures?

Best regards,
     Davide