The ABINIT Users Mailing List ( CLOSED )

[Josef Zwanziger <jzwanzig@dal.ca>]

Oh, I see the issue. nsspol requires ixc = 0, 1, 7, or 11 only, while the 
default LDA output for uspp2abinit gives ixc = 2. The default LDA for atompaw 
is ixc = 7, that's why it always works with atompaw. I guess I would have to 
explore uspp2abinit more thoroughly to investigate changing this, but I don't 
think I'll take the time now because I am using atompaw all the time. In the 
past though I have used uspp together with "forcing" ixc to 1 and it worked 
fine.

Joe


On Tuesday 27 May 2008 10:27, Bernard Amadon wrote:
> In fact, as atompaw generates LDA (PW) or GGA (PBE) atomic datasets, I
> think that at least
> ixc=7 (PW) and ixc=11 (PBE) have worked since the beginning. Maybe the
> documentation was not sufficiently
> explicit on this point.
> But to my mind, any value of ixc should work as far as it is compatible
> with nspden=2. If you
> have an example of any problem with a specific value if ixc, i would be
> interested to know !
>
> Best regards
> Bernard
>
> Josef Zwanziger a écrit :
> > Did the ixc issue change? At least in some versions, the ixc variable had
> > to be set to a fairly limited set of possibilities for LDA+U
> > calculations, has that restriction been lifted?
> >
> > Joe
> >
> > On Tuesday 27 May 2008 09:32, Bernard Amadon wrote:
> >> Hello,
> >>
> >> To complete the answer, you can find related information in variable
> >> usepawu
> >> (http://www.abinit.org/Infos_v5.5/input_variables/varpaw.html#usepawu)
> >> and in the file  doc/users/HowtodoLDA+U.txt.
> >> I would also say, that in principle ixc should not be given in the input
> >> as it is supplied in the (PAW) atomic data.
> >> In the current implementation in Abinit, both LSDA+U and GGA+U
> >> calculations can be carried out:
> >> you just have to use LDA or GGA atomic data.
> >>
> >> Best
> >> Bernard Amadon
> >>
> >> Josef Zwanziger a écrit :
> >>> Hi,
> >>> I have found the following combination to work quite well:
> >>> 1) Use PAW, and in particular use atompaw to make your PAW data sets (I
> >>> had been using USPP for a long time, but I switched--atompaw is much
> >>> easier and gives very good results).
> >>> 2) In your input, use:
> >>>
> >>> usepawu 1
> >>> lpawu ... (here -1 for the atom types without U+J, and L for the
> >>> angular momentum of the atom types to be treated with U and J)
> >>> upawu : value of U for each atom type (irrelevant for the atoms in
> >>> lpawu that are not being treated)
> >>> jpawu: value of J for each atom type (also irrelevant for atoms not
> >>> being treated)
> >>> nsppol 1 or 2 depending on your system
> >>> nspden 2 (required for LDA+U in abinit)
> >>> ixc 1 (fairly limited set of ixc's available for LDA+U calcs, see
> >>> documentation)
> >>> spinat ... (use this to break spin symmetry if desired)
> >>>
> >>> good luck--
> >>>
> >>> On Tuesday 27 May 2008 06:52, Anurag Chaudhry wrote:
> >>>> Hello everyone,
> >>>>
> >>>> Few days back I saw a couple of mails about LDA+U calculations with
> >>>> Abinit. I would like to do LSDA+U calculations.  can anyone suggest
> >>>> how to setup such a calculation and what pseudopotentials to use ?
> >>>>
> >>>> thanks and regards,
> >>>> Anurag
> >>>>
> >>>>
> >>>>
> >>>>       __________________________________________________________
> >>>> Sent from Yahoo! Mail.
> >>>> A Smarter Email http://uk.docs.yahoo.com/nowyoucan.html
> >>>> This email was Anti Virus checked by Astaro Security Gateway.
> >>>> http://www.astaro.com
> >>
> >> This email was Anti Virus checked by Astaro Security Gateway.
> >> http://www.astaro.com
>
> This email was Anti Virus checked by Astaro Security Gateway.
> http://www.astaro.com

-- 
Josef W. Zwanziger
Canada Research Chair in NMR Studies of Materials
Director, Atlantic Region Magnetic Resonance Centre
Dept. of Chemistry and Inst. for Research in Materials
Dalhousie University
Halifax B3H 4J3 Canada
+1 (902) 494-1960
jzwanzig@dal.ca
http://jwz.chem.dal.ca

On sabbatical January-July 2008
Max Planck Institute for Polymer Research
Mainz, Germany