[gpaw-users] Electronic convergence

[Jens Jørgen Mortensen]

On Wed, 2010-04-14 at 14:19 -0500, Daniel Torres wrote:
> Hi
> These were old values I got with dacapo. The RPBE cell parameter I got 
> from GPAW for Au GPAW (7x7x7 Kpoints h=0.15) is 4.29. 4.20 was the value 
> obtained with the same parameter using GPAW command line.

Well, you should sort that out: 4.29 or 4.20 Å ?  A quick calculation
gave me 4.22 Å:

  gpaw -x fcc Au --kpts=8,8,8 -a 4.18 --xc=RPBE -i rpbe.8.16 --h=0.16
--fit

but I didn't check for convergence with respect to number of k-points
and grid points.

Maybe you want the Ni layer to have its bulk lattice constant - I don't
know.

In you calculation you use 15 empty bands.  That is too few.  If you
don't specify the number of bands, you will get a more reasonable number
- it may be too high, but at least your SCF loop should converge.  I
also recommend that you use the default eigensolver (RMM-DIIS), which is
the most efficient.  And finally, less aggressive density mixing will
also help.  Something like this should work:

calc = GPAW(xc='RPBE',
            mixer=MixerSum(0.04, 5,  100.0),
            width=0.1,
            kpts=(1, 1, 1),
            h=?,
            txt='slab.txt')

And again, check whatever property you are interested in calculating for
convergence with respect to h and kpts.

Another tip:

When you set up your slab, you should use the fcc111() function from
ASE.  Something like this (not tested):

from ase.lattice.surface import *
a = 4.2
slab = fcc111('Ni', size=(6, 6, 1), a=a / 6 * 5)
slab.positions[:, 2] -= ???
slab.set_initial_magnetic_moments([0.8] * 36)
slab += fcc111('Au', size=(5, 5, 2), a=a)
slab.center(vacuum=?, axis=2)

See: https://wiki.fysik.dtu.dk/ase/ase/surface.html

JJ

> Thanks in advance and best regards
> Dani
> Jens Jørgen Mortensen wrote:
> > On Tue, 2010-04-13 at 15:04 -0500, Daniel Torres wrote:
> >   
> >> Hi,
> >> this other script attached does not converge electronically. Any suggestion 
> >> would be helpful.
> >>     
> >
> > A quick question:  Your Au-Au and Ni-Ni distances are 3.096 and 2.568 Å
> > which corresponds to lattice constants of 4.38 and 3.63 Å (experimental
> > numbers are 4.08 and 3.52 Å).  Did you calculate Au and Ni lattice
> > constants for RPBE and what did you get?
> >
> > JJ
> >
> >   
> >> Thanks and best regards
> >> Dani
> >> _______________________________________________
> >> gpaw-users mailing list
> >> gpaw-users at listserv.fysik.dtu.dk
> >> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
> >>     
> >
> >
> >   
> 
> _______________________________________________
> gpaw-users mailing list
> gpaw-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users