Broadening technique

[Cyrille]

Hello

I have a question regarding the broadening technique in a standard scf calculation within ATK.
From the vnl interface it seems that I have only access to Fermi-Dirac broadening.
Is it possible to use Marzari-Vanderbilt or Methfessel-Paxton broadening as well?

thanks in advance

Cyrille


--------------------------------------------------------------------------
Cyrille Barreteau
CEA Saclay, IRAMIS, SPEC Bat. 771
91191 Gif sur Yvette Cedex, FRANCE
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
DTU Nanotech
Ørsteds Plads, building 345E
DK-2800 Kgs. Lyngby,  DENMARK
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+33 1 69 08 29 51 / +33  6 47 53 66 52  (mobile)   (Fr)
+45    45 25 63 12/  +45     28 72 55 18  (mobile)  (Dk)
email:     cyrille.barreteau@cea.fr  /  cyrbar@nanotech.dtu.dk
Web:     http://iramis.cea.fr/Pisp/cyrille.barreteau/
 -------------------------------------------------------------------------------

[Jess Wellendorff]

Hi Cyrille,
FD broadening is the only option with ATK. Is there a particular reason you would prefer MV or MP broadening instead?

[Anders Blom]

We'll probably add the other options to ATK 2016, for bulk materials. But the question is interesting - are they really much better? What is your experience, Cyrille?

[Cyrille]

Hi Anders

Thanks for your answer.
I would say it is mainly for convergence (and total energies) issues. With MV and MP it is possible to use larger broadening (temperature T) than Fermi Dirac (FD) while keeping a good estimate of the total energy. Indeed if you plot E(T) it is very flat over a very large range for MV and MP while it increases rapidly with FD. This usually facilitates the convergence keeping a behaviour similar to T=0..

Cyrille

[Jess Wellendorff]

Great, thanks for the input! We will keep this in mind for a future release of ATK.

[asanchez]

Nicola Marzari's PhD thesis has some interesting information regarding this whole issue if anyone's interested;

http://theossrv1.epfl.ch/Main/Theses

[Jess Wellendorff]

Thanks very much for the reference. Yes, I am aware that the PWscf code offers several different options for electron occupation smearing, and that there may be good reasons to prefer the cold smearing scheme. Very interesting.

[Cyrille]

MP and MV have rather similar behaviours. The only  problem with MP is that you can sometimes get negative charges which is not the case of MV.
Another very interesting discussion (with many additional references) on the same website:
http://theossrv1.epfl.ch/Main/ElectronicTemperature

Cyrille