Treatment of Core vs Valence electrons in ELK

Last edit: Jack Whaley-Baldwin 2020-05-22

Jack Whaley-Baldwin - 2020-05-31

Bumping this up!

Dear Jack,
here's a late (partial) answer to some of your questions.

The core states are treated with the radial Dirac equation. They are not frozen, but solved self-consistently. The valence electrons are solved with the Schrödinger equation instead, as usual. spinorb adds spin-orbit coupling in the second-variational step as a perturbation, which is treated self-consistently.

For the meaning of the local orbitals: elk uses effectively a full-potential APW+lo+LO method. The lo's and LO's have different meaning in terms of how many functions are used to describe these types of local orbitals. For details, see this excellent article: https://journals.aps.org/prb/abstract/10.1103/PhysRevB.64.195134

Hope this helps.
Cheers,
Markus

Dear all,

May I ask how precisely core and valence electrons are treated differently in ELK? I'm still not entirely sure after reading the code manual and looking at the species file sticky at the top of this forum.

From the code manual, I can find (if an orbital's spcore value is set to .true. in the species file):

"if state is in the core and therefore treated with the Dirac equation in the spherical part of the muffin-tin Kohn-Sham potential"

So this seems to imply that the difference between core and valence electrons in ELK is that 'core' electrons are treated with the Dirac equation for r < R_MT, whereas non-core electrons are not (?). I.e. any electrons flagged as core are assumed to have a high kinetic energy => Must be treated with the full Dirac Equation (regardless of what spinorb is set to)? Is this the only difference?

Coming from a pseudopotential background, I am used to the core/valence partition deciding which electrons to explicitly include in the pseudopotential definition (i.e. if an electron is not in the valence, it is frozen in the core from the start and cannot partipate in bonding).

I guess I'm confused as to what exactly the core/valence partition actually means for an all-electron code. Are core electrons still present in the calculation, and just treated differently? Or are they explicitly frozen out of the calculation and assumed to be part of a frozen core?

There's an (old) post from a user here: https://sourceforge.net/p/elk/discussion/897820/thread/198f0233/ also asking about treatment of core Vs valence electrons, in which John Kay Dewhurst states:

"The valence states are calculated with the scalar relativistic Schrodinger equation, and the cores states with the Dirac equation"

Again, is this the full Dirac equation then for the core electrons, even when spin_orb is set to .false.?

There's also a reference in that post to a frozencr boolean variable which is no longer used in the code, which supposedly fixed the occpuancy of states flagged as 'core' to those of the free atom (which makes sense). Does this mean that ELK no longer has a frozen core?

Furthermore, I would like to ask what the 'local orbitals' in the species files are actually doing, e.g. for sulfur:

   3                                        : nlorb
   0   2                                    : lorbl, lorbord
    0.1500   0  F                           : lorbe0, lorbdm, lorbve
    0.1500   1  F
   1   2                                    : lorbl, lorbord
    0.1500   0  F                           : lorbe0, lorbdm, lorbve
    0.1500   1  F
   0   2                                    : lorbl, lorbord
    0.1500   0  F                           : lorbe0, lorbdm, lorbve
   -0.6346   0  T

Must the nxlo keyword be expliticly set > 0 in order to include these local orbitals? This feature, and the nxlo keyword don't appear to be well-documented in the code manual.

I understand that 'local orbitals' in a FLAPW context are extra (k-independent, atom-specific) basis functions that can be added to increase variational flexibility.

https://sourceforge.net/p/elk/discussion/897820/thread/e459f0d112/