[Pw_forum] [Fwd: paramagnetic structure]

[Malgorzata Wierzbowska]

On Fri, 9 Feb 2007, Davide Ceresoli wrote:

   Hi Subhradip Ghosh and Davide,

   Take for instance Niobium, 5s^2 4d^{3} -- it looks like unpaired...
   but it is paramagnetic (except some spin-fluctuations which have no
   force to change much in the nonsuperconducting state).
   So my calculations for that are nspin=1. I believe that if I
   did nspin=2 I would get the same but my k-grid would be twice
   bigger and maybe convergence would take a few more iterations because
   there would be more degrees of freedom.

   For metals, the band occupations can be fractional.
   With an open shell singlet (which is the first excited state of
   oxygen) I have no experience with espresso. For such cases I used
   localized basis set programs.

   By the way, this is an interesting question. Does anybody know
   whether it could be calculated with espresso?
   Is any fractional/weighted piece of code there? An implementation of
   fractional DFT by EKU Gross and Oliveira or Agnes Nagy and Theophilou?
   Gosia



> Subhradip Ghosh wrote:
>> Dear Davide,
>> 
>> Thanks for the reply.
>> I guess I understand that a calculation can be only 'non-spin-pol' or
>> 'spin-pol'(not PM,FM,AFM etc.) . But, don't we use phrases like these
>> during discussions? Atleast, I have been in discussions where people say
>> 'PM calcs or FM calcs' which nobody interprets in the literal sense!
> Dear Subhradip,
>    still I don't understand the difference between a "paramagnetic
> calculation" or a "ferromagnetic calculation". Do you mean that in one
> case the total spin magnetization is zero and in the other case it
> is different from zero?
> Take for instance the oxygen molecule. Would you perform a "paramagnetic"
> calculation or a "ferromagnetic" one? and for bulk iron instead?
>
>> Nevertheless, let me be more specific.
>> How does espresso simulate a paramagnetic state  for a system having
>> unpaired spins ( a system which does not have closed shells).
> Take again the simplest case of the oxygen molecule. We all now that
> the ground state is a triplet (a di-radical). So the calculation will
> be spin-polarized:
>
>  occupations = 'fixed'
>  nspin = 2
>  multiplicity = 3   (multiplicity = 2S+1)
>
> Since there are 12 valence electrons, this will place 7 electrons in
> the spin up channel and 5 electrons in the spin down channel.
>
> Davide
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>