Converse approach for the calculation of magnetic moments

[leonardo]

Dear Colleagues,

When I was running a spin-orbit calculation for a small silicon-hydrogen cluster, I have accidentally discovered that VASP seems to be able to use the converse approach for the calculation of magnetic moments. However, I could not find any document explaining how to use this approach in VASP. I have no idea which keywords in the INCAR file trigger this calculation in a controlled manner. I also do not understand the output in OUTCAR; e.g. I understand that this calculation has to be done for each atom, but in the output file only the atom numbers “0” and “0” show up:

MAGNETIC MOMENT (mu_B/1e6)

ATOM X Y Z

-----------------------------------------------------------------------------------

0 0.090667 0.152010 0.104430

0 0.076881 0.132337 0.076091

Many thanks for any hint!

Leo

PS: Here's my INCAR file:

# Note that VASP uses the FIRST occurence of a keyword
SYSTEM = DFT NONCOLLINEAR after mag=2 w MP2
PREC = Accurate
ALGO = FAST
EDIFF = 1E-7

NCORE=40
LPLANE=.TRUE.

LNONCOLLINEAR=.True.
LORBIT = 11
LSORBIT = .TRUE.
ORBITALMAG=T

ENCUT = 325.000
IBRION = 2
NSW = 5000
ISIF = 2
NELM = 60
NELMIN = 2
EDIFFG = -0.02
VOSKOWN = 1
NBLOCK = 1
ISPIN = 2
INIWAV = 1
ISTART = 0
ICHARG = 2
LWAVE = .FALSE.
LCHARG = .FALSE.
ADDGRID = .TRUE.
LVTOT = .FALSE.
LELF = .FALSE.
ISMEAR = 0
SIGMA = 0.2
LREAL = AUTO

[jonathan_lahnsteiner2]

Dear leonardo,

I am sorry to tell you that this feature is not supported for the user community of vasp.
This is because the current implementation is restricted to norm conserving
PAW. The usually used PAWs in vasp do not belong to this group.
Therefore there is no documentation on it and it should not be used.

All the best Jonathan

[leonardo]

Many thanks for this clarification Jonathan!

That's really a pity... Do you have any idea when it will be available?
All my best!

Leonardo

[jonathan_lahnsteiner2]

Dear Leonardo,

I am sorry to tell but currently there
are no plans about further development for
the converse approach.

All the best Jonathan