[Pw_forum] LDA+U and occupations matrix problems / call for NC PP Cu & Co
Stefano de Gironcoli
degironc at sissa.it
Mon Apr 27 08:49:42 CEST 2009
Dear Ricardo Faccio,
The problem of normalization of atomic wavefunctions in LDA+U should
have been solved about one year ago in the cvs version and incorporated
in recent distributions. I include below the CVS log message of that
correction that should clarify the change.
It should be possible to repeat your calculations without the
norm-atomic flag on and atomic wavefunctions should be renormalized on
the atomic grid, that is in a configuration independent way, hence
forces and stress should be ok).
Please report any problem that you may experience,
> date: 2008/07/14 21:50:33; author: degironc; state: Exp; lines: +84 -0
> Normalization of atomic wavefuncitons is checked (only for those with
> non-negative occupation) after pseudopotential reading and if it found to
> be different from 1 by more than 1.d-6 the wavefunction is renormalized.
> Should make no difference in all usual cases since atomic wfcs are only
> used in order to generate the initial set of trial wfcs and normalization
> is not important there...
> It is VERY important for LDA+U calculations using atomic wfc in the
> (the default case) since proper normalization is assumed and lack of
> it leads
> to disasters.
> Hopefully this will solve many of the difficulties people have encounterd
> when using LDA+U with older RRKJ pseudopotentials, where normalization
> was not imposed properly (and nobody realizes except when doing LDA+U) or
> pseudopotentials of unknown origin.
Ricardo Faccio wrote:
> Dear Pwscf users
> I'm a newbie using pwscf, and I’m trying to reproduce some full-potential
> all-electrons results from a previous work using WIEN2k. My system is a
> cuprate/colbatite YBaCuCoO5. I succeed running the lda+u calculations, but I
> found the previous discussed problem regarding the lack of normalization in
> some PP (non norm-conserving). I worked with the UPF's files from the QE web
> page: Cu.pbe-d-rrkjus.UPF and Co.pbe-nd-rrkjus.UPF. The problem of
> occupations higher than 1 was resolved with the flag: U_projection_type =
> 'norm-atomic'. But I am interested in checking the stresses and forces, in
> order to verify my previous equilibrium solution and these are not
> implemented for such correction ('norm-atomic').
> For these reason I would like to ask you for a PBE-Norm-conserving PP for Cu
> and Co, if anyone has it available. I think this is the fattest solution in
> order to avoid the U dependency upon normalization (through
> U_projection_type) and for checking my previous results. But, if I have no
> answers, I will try to generate these files from myself (but it is a little
> bit more complicated that ATOMS implemented in SIESTA).
> Thanks in advance and sorry for the inconvenience.
> Ricardo Faccio
> Dr. Ricardo Faccio
> Prof. Adjunto de Física
> Mail: Cryssmat-Lab., Cátedra de Física, DETEMA
> Facultad de Química, Universidad de la República
> Av. Gral. Flores 2124, C.C. 1157
> C.P. 11800, Montevideo, Uruguay.
> E-mail: rfaccio at fq.edu.uy
> Phone: 598 2 924 98 59
> 598 2 929 06 48
> Fax: 598 2 9241906
> Web: http://cryssmat.fq.edu.uy/ricardo/ricardo.htm
> Pw_forum mailing list
> Pw_forum at pwscf.org
More information about the users