The ABINIT Users Mailing List ( CLOSED )

[Josef Zwanziger <jzwanzig@gmail.com>]

My experience is: start by checking convergence with respect to ecut,
using a modest kpoint grid and experimental acell. A kptrlatt
generated by kptrlen = 40 is where I would start. The total energy
will converge (variationally) as a function of ecut but is not as
sensitive as some other observables, you should be checking
convergence if possible against the observable you actually want to
compute (stress, force, etc, maybe just total energy). Once you have a
good ecut for your pseudo's, repeat the same same cycle checking kpt
choices (you can let abinit do the work by just computing increasing
values of kptrlen). Once you have a kpt set that gives reliable
converged answers for the observable you want, you can optimize acell.
Acell is not a convergence parameter, it's just
the shape of the cell. You shouldn't vary this to get the answer you
want, you should optimize it to zero stress or some target stress. If
it's a really simple cell you can do this in one shot, using ionmov 2
and optcell 2 (moves the ions and the cell parameters). Or you can do
it in two passes: ionmov 2, optcell 0, to optimize the ion locations,
and then a second pass with ionmov 2 optcell 2.


Josef W. Zwanziger
Professor of Chemistry
Canada Research Chair in NMR Studies of Materials
Director, Nuclear Magnetic Resonance Research Resource (NMR-3)
Dalhousie University
Halifax, NS B3H 4J3 Canada
tel: +1 (902) 494-1960
fax: +1 (902) 494-1867
web: http://jwz.chem.dal.ca
jzwanzig@gmail.com, jzwanzig@dal.ca



On Tue, Feb 16, 2010 at 11:02 AM,  <asadidomakani@yahoo.com> wrote:
> Dear Abiniters,
>
> I would like to study convergence with ecut,kpoint and acll .
> for reach the best result,what are the order of them ? and what are the 
> startin
> g values for them .I want for Ga2O3.
> thank you for your attention
>