This is José M. Soler answer to this question:
Dear Nichols:
The basis set and the pseudopotentials are completely
independent things:
- If you have no l=d pseudo projector, but you use l=d basis
functions, they will feel only the local part of the
pseudopotential, exactly like in a PW program (the SIESTA
local part of the pseudopotential is rather non-standard, but
this is a separate issue and not very relevant in practice).
- If you have no l=d basis, but you use a l=d pseudo projector,
it will still act on the l=s,p functions of neighbor atoms.
If these were very rich, they might 'in principle' represent
the l=d part of the wavefunction around the atom.
In summary: there is no difference on how the pseudopotentials
act in SIESTA and PW codes. If the wavefunctions are the same
(though expanded in different basis) the results are the same.
Best wishes,
Jose M. Soler
Hi,
This is a conceptual question about basis sets (pw vs. localized
basis) and pseudopotentials as implemented in SIESTA. The best way
to ask my question
is by a very simple example.
Consider Si where L = d states are important. In SIESTA,
contributions from L= d states require L=d basis functions _plus_
the d-pseudopotential. As opposed to a code like ABINIT, where the
PW have this variational freedom builtin. One would get
contributions from L=d states, even without the d-pseudopotential.
(One can do a PDOS per angular momentum and see this.)
So in SIESTA, unless you actually have L=d basis functions
including the d-pseudopotential would _not_ do anything. Or to put
it another way, does the
d-pseudopotential act on the basis functions of different symmetry
in the localized basis set?
Thanks,
--
Nichols A. Romero, Ph.D.
1613 Denise Dr. Apt. D
Forest Hill, MD 21050
443-567-8328 (C)
410-306-0709 (O)
