[Wien] spin-orbit and forces

[Peter Blaha]

a) Be sure to use SO WITHOUT RLOs (EFGs are wrong with RLOs)
b) The core and Hellmann-Feinman contribution to the force just come from 
the nonspherical V_1x(r) potential. So it is not too surprising that they 
remain similar with and without SO (indicating that this non-spherical 
component of rho (and V) does not change too much in this case).
c) Yes, I think forces with SO can really go "wild", in particular in your 
case, where the IBS valence correction is also in the non-SO case very 
big.

  
> In a calculation of a heavy impurity (Pb) in a heavy host lattice (hcp-Re),
> I noticed the following behaviour which I don't understand:
> 
> First I relax the position of several neighbour shells around the impurity,
> without spin-orbit coupling (it doesn't make sense to do this with
> spin-orbit, as the forces by wien2k are incorrect for spin-orbit anyway).
> Then I rerun the relaxed case with spin-orbit (adding spin-orbit has a
> significant effect on e.g. the EFG of all the atoms). In this spin-orbit
> calculation, the forces without the FOR switch are somewhat different from
> the ones without spin-orbit, but with the FOR switch they are crazy :
> 
> grepline :FGL013 '*.scf' 3
> 
> no-SO.scf::FGL013:  13.ATOM                       221.045        190.184
> 0.081
> no-SO.scf::FGL013:  13.ATOM                       221.067        190.208
> 0.195
> no-SO.scf::FGL013:  13.ATOM                         1.120
>  0.036         -0.079
> 
> with-SO.scf::FGL013:  13.ATOM                       214.047        185.388
> 0.151
> with-SO.scf::FGL013:  13.ATOM                       214.040        185.393
> 0.095
> with-SO.scf::FGL013:
> OM                      -440.734      -3270.016       -582.134
> 
> (every time in the last iteration FOR is switched on)
> 
> I know one that the forces with spin-orbit are meaningless numbers in
> wien2k, but I cannot remember to have met such high values before, and this
> worries me a bit (indication of an error somewhere?). Can somebody confirm
> that forces-with-SO can really go wild, or can somebody point me to the
> likely error if this would not happen to be the case?
> 
> Basis set, k-mesh and Emax are well-converged, in1new is used.
> 
> Thanks,
> Stefaan
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> 


                                      P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------