PBE0 MULTIPLICITY 1 calculation

[marci]

Dear Carlo,

since you used the smearing of the occupation numbers, the multiplicity is 
not constrained. 
The occupation of the eigenstates is assigned according to their energy, 
irrespective of alpha and beta.
The Fermi energy is then determined consistently for the full electronic 
system, and it is -0.14381945698278 a.u. in your case.

If this is really meaningful for your system, I do not know. 
Is this really a metallic system, as it apparently results from this 
calculation?
Is PBE0 appropriate for a metallic system?

ciao
Marcella

Il giorno martedì 14 gennaio 2014 12:08:05 UTC+1, c.pignedoli ha scritto:
>
> Dear all,
>  I am doing a PBE0 calculation for a periodic quasi 1D system (a graphene 
> nanoribbon)
>
> I did restart the calculation from a converged antiferromagnetic PBE 
> calculation
>
> The PBE0 calculation terminated with a unequal number of alpha and beta 
> orbitals thus violating (I guess) the MULTIPLICITY 1 of the input and of 
> the restart.
> (
> 355 alpha and 349 beta while it should be  352 alpha and 352 beta
> this can be seen checking   the added mos
> declared at the end of the file: should have been 100 and 100 instead it 
> is 97 102
> and form the file with occupation numbers of the states (which I do not 
> include)
> ) 
>
> I enclose the input file and the output, what is scaring is a sudden jump 
> of the energy at 
> the beginning of the SCF (from -980 to -1200 at step 24 of the scf) that 
> could be related to problems with the coulomb cutoff.
>
>
>
> Does anybody have suggestions about what could be wrong?
> Could it be that the cell size is too small for this particular system 
> (however it is big enough
> to have a fair PBE description)?
> (Basically I have 12 unit cells of a zig-zag graphene nanoribbon with at 
> least 20A of vacuum in the non periodic directions, the PBE UKS 
> calculationn produces the correct edge states but of course these states 
> are nearly metallic and have, in PBE a small gap)
>
> Kind regards
>
>
> Carlo
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140114/0c0f5132/attachment.htm>