[Pw_forum] ensemble DFT vs "PWSCF" DFT in metals

Stefano de Gironcoli degironc at sissa.it
Fri Apr 17 10:24:13 CEST 2009

The functional is the same but the approach used to reach the solution 
is different.
PWscf uses a self-consistent cycle technique while ensemble DFT recast 
it in a global-minimization problem.

Ensemble DFT is guaranteed (if I remember correctly) to be a scheme 
where the total energy decreases at each iteration while for the 
self-consistent solution sometime the energy does not monotonically 
decrease with the number of iteration before converging.
I think that the relative efficiency of the two methods may be different 
for different systems and only experience can tell you wich one is more 
efficient for your system.
Tricky systems are typically very anisotropic  metallic slabs subjects 
to charge sloshing (dealt with mixing_mode='local-TF' in PWscf) and/or 
magnetic systems where many alternative solutions are close in energy 
(larger smearing usually helps in stabilizing the system but may wash 
out the physical effect... very many kpoints and a small smearing may be 
needed) .
In these cases the always decreasing nature of  ensembe-DFT may be useful.
At variance with CP, PWscf exploits symmetry so for "crystals" PWscf may 
be preferable.
I don't know how the many-k-points case, often needed for metals, is 
dealt with in the ensemble DFT method... in principle all partially 
occupied electronic states (nbnd*nks) should be treated together. But I 
don't know how this is done in CP.
Methfessel and Paxton broadening may leads  to non positive-definite 
occupation numbers. This is usually not a big problem for PWscf , when 
decent k-point sampling is used, but I think it is incompatible with the 
ensemble DFT  scheme. Luckily other efficient smearing techniques such 
as the Marzari-Vanderbilt scheme  are available that are positive-definite.

Stefano de Gironcoli - SISSA and DEMOCRITOS

Eduardo Ariel Menendez Proupin wrote:
> Hi,
> I am curious about the difference between the ensemble DFT method 
> implemented in CP and the smeared DFT used in PWSCF with 
> Methfessel-Paxton smearing. ¿What is better for what system?
> -- 
> Eduardo Menendez
> Departamento de Fisica
> Facultad de Ciencias
> Universidad de Chile
> Phone: (56)(2)9787439
> URL: http://fisica.ciencias.uchile.cl/~emenendez 
> <http://fisica.ciencias.uchile.cl/%7Eemenendez>
> ------------------------------------------------------------------------
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum

More information about the users mailing list