[Pw_forum] Serious problem in total energy calculation ?

[Paolo Giannozzi]

On Jun 14, 2007, at 3:08 , Eun-Cheol Lee wrote:

> I didn't know this kind of problem in  norm-conserving
> pseudopotential, where sufficient kinetic energy
> cutoffs are used and negative charge does not exist.

the charge density with norm-conserving pseudopotentials is
positive by definition. Unfortunately ultrasoft pseudopotentials
may have a small amount of negative charge, typically because
the cutoff is not sufficiently high, but also because the charge
density is not guaranteed to be positive (although, at least in
the atomic configuration used to produce the pseudopotential,
it should).

Anyway: with a LDA ultrasoft pseudopotential for O, also having
some negative charge, the difference between PW and atomic
calculations is much smaller than with the PBE pseudopotential,
so it is more likely that differences in the numerical implementation
of gradient-corrected functionals are to be blamed for the difference.

Anyway: I don't think such differences have any effect on the
quality of the results. If you have any evidence of the opposite,
please let us know

Paolo
---
Paolo Giannozzi, Democritos and University of Udine, Italy