[CP2K:5630] Pseudopotentials and DFT functionals

[mpol]

Dear CP2k developers and users,

Apparently, the reason to introduce NLCC pseudopotentials is the associated 
error in modeling of spin-polarized states (
http://dx.doi.org/10.1063/1.4793260). Indeed, in the case of isolated atoms 
the error may be very significant. Therefore the following questions/notes 
come to mind:

1. Is there any estimation of this NC-PP-related error in the case of 
molecular or periodic systems?

2. The error should be less pronounced when norm conserving 
pseudopotentials including semicore states are used, isn't it?

3. As I know, GGA functionals may be a bad choice for prediction of atomic 
properties. It is therefore would be interesting to estimate the error 
introduced by norm-conserving pseudopotentials *plus* the GGA-related error.

4. So this error introduced by NC-PPs persists in the calculations of 
energies of spin-polarized molecules or periodic structures with meta-GGA 
or hybrid functionals, isn't it?

With best regards,
MP
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140904/fe797f64/attachment.htm>