[Wien] the sequence of doing +U and SOC in a GGA+U+SOC calculation

 

 

Bin Shao Thu, 21 Jan 2016 00:30:34 -0800

Dear all,

I need to do GGA+U+SOC calculation with 3d and 4/5d atoms. I found that the
results with different sequence of doing +U and SOC were different, i.e.,
if you first do +U calculation and then +SOC, the result is different from
that of first doing SOC and then +U. Shouldn't that be the same? What is
the right way to do GGA+U+SOC calculation?

Best,

Bin Shao

Answers:

tran Thu, 21 Jan 2016 02:07:14 -0800




With solids containing open d-shell, it is usually possible
to stabilize several solutions corresponding to different
occupations of the d-orbitals.  Which solution one finds, may
depends on the electron density used at the beginning of the
SCF procedure. A good criteria is to choose the solution that
has the most negative total energy (:ENE in case.scf).

F. Tran








Peter Blaha Thu, 21 Jan 2016 02:24:19 -0800




In addition to what Fabien was saying:


If you are using 14.2 there was a fix for the off-diagonal matrix-elements and when doing LDA+U AND spin-orbit, one should probably change


runsp -orb -so      to

runsp -orb -so -orbdu


This is a suggestion from Jan Kunes, personally I have no experience with it, but it may be of importance for certain cases.





Regards




Reference: https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg13673.html