The ABINIT Users Mailing List ( CLOSED )

[Paul Tangney <tangney@civet.berkeley.edu>]

Hello again,
Thanks to those who replied to my mail yesterday about the core-corrections.

I *think* I have found the problem - and I think that it is sufficiently
important to tell others
about it :

When ABINIT reads the core charge from the FHI pseudopotential file it converts
it from a logarithmic radial grid to a uniform grid. The uniform grid is
defined from
zero to the radius "rchrg". ( Note: rchrg is *not* the core cutoff that is
used in generating
the pseudopotential) . The number of points on this uniform grid is a parameter
"n1xccc"
and by default it is set to be 2501.  If the value of rchrg is large (it
should be at a radius
where the core charge has decayed to zero) then this number is not big enough
and this can lead 
to problems for the steep gradients near the core. 

For example, I had used rchrg=10.0 and the error caused a discrepancy of
3 GPa in the stress. I found that both  for large values of rchrg and for
too-small values (where
the core charge has not yet gone to zero) the stress is quite sensitive to
this value.
It is very difficult to find the in-between region where everything works
well - so in practice,
(for gallium, at least) the stress (and energy) varies continuously with
rchrg.

The problem is cured by increasing the value of "n1xccc_default" which can
be found in
the file  "Src_8seqpar/iofn2.f"

Please let me know if I'm wrong about any of this. Increasing this parameter
seems to help.

Regards,



Paul




 
> > -----Message d'origine-----
> > De : Paul Tangney [mailto:tangney@civet.berkeley.edu]
> > Envoyé : 27 mars, 2003 04:17
> > À : forum@abinit.org
> > Objet : [abinit-forum] Nonlinear-core corrections
> >
> >
> > Dear ABINITioners,
> >
> > Has anybody encountered problems with nonlinear core corrections
> > in ABINIT ?
> >
> > I am using pseudopotentials that I generated with the Fritz-Haber
> > code FHI98PP. The total energy that I get disagrees slightly with
> > the results using the PARATEC plane wave code. Differences are of
> > the order of 1 or 2 mRyd. (and about 4 GPa in the stress).
> > The system is GaAs - I haven't checked other systems, but if
> > I make PPs without core-corrections I can get both codes to agree to a
> > high precision.
> >
> > I am a new user of ABINIT (and also PARATEC) and so the problem
> > might be something very simple, but before I get my hands dirty
> > by delving into the source I thought I should ask to try to
> > figure out whether the problem is with me, PARATEC, or ABINIT.
> > In particular, I'm a little confused about the input variables
> > "rchrg","fchrg" and "qchrg" that are on the 4th line of the
> > PP files. I have set rchrg to 10.0 where the core charge is zero
> > and I have values of 1.0 for both fchrg and qchrg but I don't
> > know if this is correct.
> >
> > I would be very grateful for any help you could give me.
> >
> > Best regards,
> >
> >
> > Paul
> >
> >
> >
> > -- 
> > Paul Tangney
> > OoOo-------------------------------------------------------------
> > University of California at Berkeley   /    Phone : (510) 642-2635
> > Physics Department #7300       /       Home Phone : (510) 883-1259
> > 366 Le Conte Hall             /               Fax : (510) 643-9473
> > Berkeley, CA 94720       /      Email : tangney@civet.berkeley.edu
> > Home page : http://civet.berkeley.edu/tangney/
> > -----------------------------------------------------------------
> >
> >
> >
> >