[Wien] to Dr. Cottenier

[Stefaan Cottenier]

>By the sentence " I did whatever was said in the userguide! " I did not
mean that the
>userguide is not complete but I cannot understand it. Please excuse me for
that.

That's no shame, and it's perfectly legitimate to ask such a question on the
mailing list. But your original question was too general. Now you have sent
a detailed procedure, hence we can comment on it :

>First I ran the scf

I guess you verified that it converged well. In order to apply LDA+U to
lanthanides in a meaningful way, you should use spin-orbit coupling first
(see also your later question). Your specific case of Gd might be
exceptional, in the sense that it has a fully filled up-band, and an
completely empty dn-band, which should be described rather well without
LDA+U (I'm not sure on the spin-orbit, see below).

>I generated the two files Gd.indm and Gd.inorb in the corresponding
directory, and then
>changed the data as follows:

> Gd.indm
> -9.    Emin cutoff energy
> 1      number of atoms for which density matrix is calculated
> 1 1 3  index of 1st atom, number of L's, L1
> 0 0    r-index, (l,s)index

OK, you calculate the density matrix for the Gd f-states.

> Gd.inorb
> 1 1 0        nmod, natorb, ipr
> BROYD 0.2    BROYD/PRATT, mixing
> 1 1 3        iatom nlorb, lorb
> 1            nsic 0..AFM, 1..SIC, 2..HFM
> 0.7 0.05     U J (Ry)

1) The UG tells *not* to change the mixing line, which is by default at
PRATT 1.0
2) In the more recent versions of wien2k, the advice is to keep J=0, and
change only U (see UG)
(and 3) : there has been a bug report recently that affects LDA+U. Be sure
you use only the latest version of wien2k)

>and the program just stopped in first cycle of scf giving empty lapwdm. I
also ran x
>lapwup/dn. I cannot find where I'm wrong.

At which stage did it stop, and what was the error message (if any)?

> Two more questions:
>
> 1. Is it O.K. we say that since in Gd L=0 then because of "orbital angular
momentum
> quenching" the spin-orbit coupling does not play a major role in
calculations?

See above: Gd might be the exceptional case, but if your computer can afford
it, don't hesitate to use spin-orbit for lanthanides (I never tested for
myself Gd without spin-orbit, only for Gd without LDA+U I am sure that it is
virtually identical to LDA+spin-orbit -- many examples of that can be found
in the (somewhat older) literature as well).

> 2. When including SO it asks for the case.struture to be modified. How
should this be done?

Follow the instructions on the screen, it should go more or less
automatically. Specify a direction in case.inso the orientation of the
magnetic moments, and add a Relativistic Local Orbital if you want (for
details, see UG). Don't forget to test convergence as a function of Emax
(start with Emax=3.0, which can be already fine).

Stefaan