[gpaw-users] Ni in a box doesn't convergence and spin-polarized set-up

[Marcin Dulak]

Ask Hjorth Larsen wrote:
> Hi
>
> On Wed, 14 Dec 2011, Marcin Dulak wrote:
>
>   
>> Hi,
>>
>> better avoid fractional occupancies.
>>     
>
> But does it make sense to avoid fractional occupations here?  It seems to 
> be a more fundamental issue with Kohn-Sham DFT.  It's not supposed to 
> converge at all with integer occupations (unless one artificially fixes 
> them), so I think a calculation which manages to converge it must be wrong 
> in some other sense.
>   
i think it would simply be nice to be able to relate to what we learn 
about atomic configurations and Hund rule during the studies.
DFT has a problem as d states are not degenerate 
http://dx.doi.org/10.1063/1.2723118 . I'm not sure if this is something 
that makes convergence impossible,
but apparently this introduces a risk of getting different states 
depending of tiny implementation details.
The question with allowing smearing is which smearing to choose (it's 
true this can be answered by checking the dependence of
the results on smearing and relating this to the expected error in the 
calculated property inherent to DFT),
but if one manages to converge with smearing of zero and had a method to 
analyse the result (https://trac.fysik.dtu.dk/projects/gpaw/ticket/217),
then explicit stating the of the sd-configuration may be possible.
More literature about the problem:
http://dx.doi.org/10.1016/S0009-2614(96)01449-2

Marcin

> Regards
> Ask
>
>   


-- 
***********************************
 
Marcin Dulak
Technical University of Denmark
Department of Physics
Building 307, Room 229
DK-2800 Kongens Lyngby
Denmark
Tel.: (+45) 4525 3157
Fax.: (+45) 4593 2399
email: Marcin.Dulak at fysik.dtu.dk

***********************************