[CP2K:4894] Re: Non-variational SCF energies

Ari Paavo Seitsonen ari.p.s... at gmail.com
Tue Jan 14 17:05:41 UTC 2014

Hi Chris,

  Adding this time to Jürg's answer, rigorously the total energy _is_ only
defined at self-consistency, and then the Hohenberg-Kohn theorem holds and
you should have a the lowest possible energy. The mixing causes the
"estimates" of the total energy (what is printed out at each SCF step) to
be a bit more "confused" because the density and orbitals are even less
consistent. The problem is usually worst in metallic systems where the
density oscillations occur the easiest (smallest "excitation" energies for
mixing up the electronic structure).

  Another sources of small errors would be from the fractional occupation
numbers (again, only in metallic systems) or due to numerical inaccuracies
(that is why I suggested checking those parametres like the cut-off energy).



2014/1/14 Chris E <christoph... at gmail.com>

> Marcella and Ari,
> Thank you for the information. I have already tried tightening the EPS_SCF
> cutoff but not the EPS_DEFAULT. I will try that as well.
> Maybe I should review my DFT books to find this, but its worth asking: if
> tightened cutoffs produce the same results, is the higher energy calculated
> earlier in the SCF run a false energy due to mixing? Or does the mixing
> simply allow the energy to be non-variational?
> Thanks again,
> Chris
>>  --
> You received this message because you are subscribed to the Google Groups
> "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to cp2k+uns... at googlegroups.com.
> To post to this group, send email to cp... at googlegroups.com.
> Visit this group at http://groups.google.com/group/cp2k.
> For more options, visit https://groups.google.com/groups/opt_out.

  Ari P Seitsonen / Ari.P.S... at iki.fi / http://www.iki.fi/~apsi/
  Physikalisch-Chemisches Institut der Universität Zürich
  Tel: +41 44 63 55 44 97  /  Mobile: +41 79 71 90 935
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140114/c9f40023/attachment.htm>

More information about the CP2K-user mailing list