Is it possible to study/converge magnetic surfaces? repost..

marci marc... at
Fri Oct 28 11:16:49 UTC 2011

Dear Valerio,

If you re speaking of the  '# Total charge and spin" given at the end
of the population analysis, it should not give always integer numbers
if the smearing in activated and the calculation is spin polarized.
If this is not the case, it might be a problem of the version of the
code. The correct smearing with LSD was implemented early this year.

The rules to choose the parameters are the same as with other codes
used with metallic systems.
Consider that the default electronic temperature in cp2k is quite low
(300K) if compared to typical values used with other codes.
But you can always increase it from input, and this should help in
speeding the convergence up.


On Oct 28, 12:53 pm, Valerio Bellini <valerio... at>
> Il 28/10/11 12.42, Dr. Roman Leitsmann ha scritto:
> > Dear Marcella,
> > I have some (may be stupid) questions about your calculation:
> > (1) You are using additional MOs and a certain smearing, right?
> > However, in the case of fractional charges the computation of the spin
> > moment is not yet proper working in cp2k (WARNING: S** computation
> > does not yet treat fractional occupied orbitals). So my question is:
> > How have you calculated the spin moment.
> > (2) How have you calculated the difference between spin up and spin
> > down channel. In my version of cp2k only integer numbers of electrons
> > are given for each spin channel.
> > best
> > Roman
> Dear Roman and Marcella,
> I had the same (maybe stupid) questions myself! ;-)
> in the line '# Total charge and spin" there is always an integer total
> spin moment.
> maybe that is not where one should look, is it?
> Secondly .. 140 iterations.. WOW!!
> are there some flag that one should put in the .inp file we should be
> aware of?
> And more importantly, are the some general rule (apart from experience) one
> can rely on, in the choice of the mixing/number of additional MOs/Fermi
> Dirac temperature?
> Thanks again for the preciuos help,
> Valerio

More information about the CP2K-user mailing list