[gpaw-users] Ni in a box doesn't convergence and spin-polarized set-up

[Nichols A. Romero]

I think we tried at 0.10 and it didn't work.

Alvaro,

Can you play around with this script and see if you can get it to converge?

Nichols A. Romero, Ph.D.
Argonne Leadership Computing Facility
Argonne National Laboratory
Building 240 Room 2-127
9700 South Cass Avenue
Argonne, IL 60490
(630) 252-3441


----- Original Message -----
From: "Ask Hjorth Larsen" <askhl at fysik.dtu.dk>
To: "Alvaro Vazquez Mayagoitia" <fray.gory at gmail.com>
Cc: "Nichols A. Romero" <naromero at alcf.anl.gov>, "gpaw-users" <gpaw-users at listserv.fysik.dtu.dk>
Sent: Tuesday, December 13, 2011 6:53:01 AM
Subject: Re: [gpaw-users] Ni in a box doesn't convergence and spin-polarized set-up

Hi

On Tue, 13 Dec 2011, Alvaro Vazquez Mayagoitia wrote:

> Hi Ask,
>
> yes, Ni is a weird/tricky example, even for gaussian basis set codes
>
> I tried adding
>  fixmom=True
> and
>  occupations=FermiDirac(x)
> (with 0<x<1)
>
> and no success,
>
> Could you provide me an example, please?
>
>
> Thank you in advance,
> -Alvaro

Note that it isn't entirely trivial whether the calculation makes *sense* 
when the occupations are fractional - but I think it's the most correct 
way (given multiple "incorrect" alternatives) to deal with the problem.

This works:

from ase import Atoms
from gpaw import GPAW, FermiDirac, PoissonSolver

system = Atoms('Ni')
system.center(vacuum=5.0)
calc = GPAW(occupations=FermiDirac(0.3), hund=True,
             poissonsolver=PoissonSolver(relax='GS', eps=1e-8))

system.set_calculator(calc)
system.get_potential_energy()



(the Poisson solver just makes it take less time)

You can probably reduce the Fermi smearing (just tested that it works down 
to 0.1).

Regards
Ask