[Pw_forum] SOC

Bhagawan Sahu brsahu at physics.utexas.edu
Wed May 17 18:59:50 CEST 2006


Andrea,

  Is the PAW feature in 'ld1 atomic' implemented with spin-orbit (ie 
rel=2). Is there a write-up explaining the implementation of PAW in RRKJ 
ultra-soft scheme? I guess there is not yet any publication. If any, pl. 
point it to me. 

Sahu

On Tue, 16 May 2006, Andrea Dal Corso wrote:

> On Mon, 2006-05-15 at 14:54 -0500, Bhagawan Sahu wrote:
> > Andrea,
> > 
> >  In the relativistic PP generation, How the the orbitals, to be pseudozied 
> > , are filled in in case of Pt?
> > 
> > I am trying to generate the Bi PP using the example of Pt given with the 
> > distribution.
> > 
> > for example, in pt.in 
> > 
> > 7
> > 5D  3  2  4.00  0.00  2.10  2.40  1.50
> > 5D  3  2  0.00 -0.20  2.10  2.40  1.50
> > 5D  3  2  4.00  0.00  2.10  2.40  2.50
> > 5D  3  2  0.00 -0.20  2.10  2.40  2.50
> > 6P  2  1  0.00 -0.00  3.30  3.30  0.50
> > 6P  2  1  0.00 -0.00  3.40  3.40  1.50
> > 6S  1  0  2.00  0.00  2.60  2.60  0.50
> > 
> > 6s is local so is placed at the end.
> > 
> > four 5d states---> for j=3/2, there are 4 states so max occupation in 
> > these states is 4.0 (for up-spin) so down-spin occupation is zero. 
> > These counts for first two 5d orbitals (one for up and one for down, 
> > j=3/2).
> 
> There are not up or down states. The states are eigenfunctions of J^2
> and J_z (total angular momentum). They are not eigenfunctions of the
> spin operator. When j=3/2, j_z can be -3/2, -1/2, 1/2, -3/2 so there are
> four states that can contain a maximum of 4 electrons. The other 4
> electrons go in j=5/2 states. 
> 
> > 
> > for j=5/2 there are 6 states, so maximum occupation in these states is 6.0 
> > but we have 4 remaining electrons so next two 5d states are for j=5/2 
> > (one up and one down j=5/2).
> > 
> > What I do not understand is the 6P states. for l=1 we have j=1/2 or 3/2 
> > 
> > for j=1/2, we have two states with maximum occupation 2.0 
> > for j=3/2, we have four states with maxmum occupation 4.0 
> > 
> > For Bi, there are 3 p-electrons  so 
> > the p config is
> > 
> > 6P  2  1  2.00 -0.00  3.30  3.30  0.50
> > 6P  2  1  0.00 -0.00  3.40  3.40  0.50
> > 6P  2  1  1.00 -0.00  3.30  3.30  1.50
> > 6P  2  1  0.00 -0.00  3.40  3.40  1.50
> > 
> 
> I think here the configuration is OK. However in US there are two states
> for each channel (these are not spin-up and spin-down but two different
> energies). Here you set all energies to 0.00 (that means: take the
> eigenvalue) and this is not correct. The second energy should be
> different from zero:
> 
> 6P  2  1  2.00 -0.00  3.30  3.30  0.50
> 6P  2  1  0.00 Energy  3.40  3.40  0.50
> 6P  2  1  1.00 -0.00  3.30  3.30  1.50
> 6P  2  1  0.00 Energy  3.40  3.40  1.50
> 
> Which energy I do not know. It depends on the atom and is a matter of
> cooking. Often an energy slightly higher than the eigenvalue works. In
> other cases you need to look at the logarithmic derivative to guess a
> reasonable value.
> 
> 
> > Is this true?
> > 
> > the first two are for j=1/2 and last two are for j=3/2
> > 
> > Also, since pseudotype =3 (ultrasoft), are the entries for rcut(meant for 
> > norm conserving PP's) in Pt.in meaningful? I guess they are ignored in the 
> > ultrasoft generation.
> > 
> 
> They are not ignored. The ultrasoft pseudization is done starting from
> the norm conserving wavefunctions.
> 
> 
> > Sahu
> > 
> > 
> > four 5d states--> why there are four of these?
> > 
> >  
> 
> two values of j and two energies for each j.
> 
> 
> > 
> > On Thu, 13 Apr 2006, Andrea Dal Corso wrote:
> > 
> > > Yes the SO term is only in the pseudo-potential but valence states
> > > interact with a fully relativistic pseudo-potential at each scf cycle. 
> > > We make an approximation because we neglect the spin-orbit coupling
> > > outside the core radius where however the effect is expected to be
> > > small.
> > > 
> > > Andrea
> > > 
> > > 
> > > On Wed, 2006-04-12 at 12:43 -0500, Bhagawan Sahu wrote:
> > > > Dear pwscf users,
> > > > 
> > > >   Is it true that
> > > > 
> > > > the  bulk spin-orbit coupling (SOC) calculation is done using a 
> > > > relativistic 
> > > > PP for the atom/ion (RRKJ scheme and Andrea Dal Carso's PRB paper) and the 
> > > > SOC effect is not included self-consistently on the valence states during 
> > > > the scf cycle?
> > > > 
> > > > Sahu 
> > > > 
> > > > _______________________________________________
> > > > Pw_forum mailing list
> > > > Pw_forum at pwscf.org
> > > > http://www.democritos.it/mailman/listinfo/pw_forum
> > > 
> > 
> > _______________________________________________
> > Pw_forum mailing list
> > Pw_forum at pwscf.org
> > http://www.democritos.it/mailman/listinfo/pw_forum
> 




More information about the users mailing list