Dear Alberto,

Many thanks for the guidance. I was looking for that section (%block
PAO.PolarizationScheme) in the manual v. 4.1-b4 but couldn't find any.

However, I entered as you wrote and it worked, but it goes only for one CG
steps and ends.

Then I tried with "PAO.oldstylepolorbs to be false and it also worked but with
all given CG steps (for example in my case it was 20).  Here is the difference
in the result output file:

1st CASE:  Using %block PAO.PolarizationScheme
 Mg non-perturbative
%endblock PAO.PolarizationScheme ---->   it only make non-perturbative
polarization for Mg but not for O as obviously.

2nd CASE: , using PAO.OldStylePolOrbs     F -----> it makes perturbative
polarization orbitals for both Mg and O.

In both case Cut-off radius for the neutral-atom potential was   8.935841

For the 2nd case I found this line "cgvc: WARNING: CG file not found" before
starting the next CG move.
For the 1st case it was "iocg: Reading CG continuation file"  and there was
another line "cgvc: No target stress found, assuming hydrostatic
MD.TargetPressure." before writing the band energy which was absent for 2nd

Why the 1st case ends after one CG move? BTW, there was difference in LUMO
band structure and also Fermi energy about 0.02 eV. So, use of
"PAO.OldStylePolOrbs   F" would be ok?

Best regards,

P.S:  I had another question to all of you. For example MgO where we have
Mg(2+) and (o2-) state. Should I generate pseudopotential with these oxidation
state (Mg- 1s2 2s2 2p6 and O - 1s2 2s2 2p6) or should I generate as neutral
atom (Mg- 1s2 2s2 2p6 3s2 and O - 1s2 2s2 2p4)? It has to be neutral atom for
PP generation always?

SIESTA is supported by the Spanish Research Agency (AEI) and by the European 
H2020 MaX Centre of Excellence (

Responder a