[Wien] two mBJ questions

[tran at theochem.tuwien.ac.at]

Hi,

It seems that the LDA energy with the mBJ orbitals leads
to the correct stability ordering. This sounds interesting, but
is it for the right reason? In other words, is it because the shape of
the mBJ orbitals is more correct than with LDA/PBE/hybrid?
It's probably difficult to answer to this question. Beside this, in my
opinion, there is nothing wrong (technically or theoretically) in using
the LDA energies calculated with mBJ orbitals.

It's possible to use the PBE energy instead of LDA for the energy.
If you are using a recent version of WIEN2k which uses keywords
in case.in0 to specify the functional, then you need to specify
this for PBE energy with mBJLDA for the potential:
EX_PBE EC_PBE VX_MBJ VC_LDA

If your WIEN2k is too old to have keywords, then you have two possibilities:

1) Modify vxclm2.f in SRC_lapw0 such that the PBE energy is used
instead of the LDA energy. In principle the changes in the block
"(IGRAD .EQ. 28)" should be only this:
Replace
       exu=2.d0*(exupls + eclsd)
       exd=2.d0*(exdnls + eclsd)
by
       exu=2.d0*(exuppb + ecpbe)
       exd=2.d0*(exdnpb + ecpbe)

2) Restore your calculation (restore -f my_calculation), replace TOT by
KXC in case.in0, and then just execute "x lapw0" with LDA (ixc=5 in case.in0)
and then with PBE (ixc=13 in case.in0). In case.scf0, the LDA and PBE
xc-energy are printed (:EXC). Then subtract :EXC for LDA from your
previously calculated total LDA energy and add :EXC for PBE.

The mBJ parameters were calibrated such that the mBJLDA potential
reproduces at best the experimental band gaps. The energy was not
used at all in this procedure.

If my explanations are too complicated, then I will try to make
more clear.

F. Tran

On Tue, 24 Nov 2015, Pavel Ondračka wrote:

> Dear Wien2k mailing list,
>
> I have two questions regarding the mBJ potential.
>
> I'm working on some TiO2 based ternary oxide solid solutions. The
> supercells were relaxed using PBE and than I'm using the mBJ to
> calculate band gaps and optical properties. The known problem for TiO2
> with PBE (but also with LDA and maybe even some hybrids) is that the
> anatase is predicted as the most stable structure instead of rutile.
> However I have noticed, that mBJ actually gets this right. So I was
> wondering if this would be a good enough reason to use the mBJ total
> energies in discussion of for example structural stability and similar
> stuff for the TiO2 based solid solutions instead of the PBE ones. 
>
> I am aware that this has been brought up multiple times on the mailing
> list, however there seems to be some disagreement, with some post
> implying the mBJ total energies (and total energy dependent quantities)
> should not be used at all: http://www.mail-archive.com/wien%40zeus.theo
> chem.tuwien.ac.at/msg08020.html 
> and some other posts suggest that maybe in some special cases it can be
> used:
> http://www.mail-archive.com/wien%40zeus.theochem.tuwien.ac.at/msg08522.
> html
> http://www.mail-archive.com/wien%40zeus.theochem.tuwien.ac.at/msg08523.
> html
>
> What would be your opinion in this case?
>
> Another question is regarding the mBJ-GGA. It has been mentioned
> several times, that it is possible to use GGA instead of LDA for energy
> functional in mBJ. So far I've always used the default mBJLDA and
> applied it to structures relaxed using the PBE (as recommended in the
> original mBJ article).
>
> However I have been asked, if it maybe would be better to use mBJ with
> PBE as energy functional, when I have structures relaxed with PBE? My
> current understanding is that since the mBJ parameters were calibrated
> using LDA as energy functional, to use it with some else I would need
> to recalibrate (refit) the parameters. 
>
> Is this understanding correct?
>
> Best regards
> Pavel Ondračka
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:  http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html