[Pw_forum] occupation problem when system changes from insulator to metal

sbraccia carlo sbraccia at sissa.it
Tue Dec 14 08:03:38 CET 2004

Dear W. YU,

the smearing technique can be used also with insulators.

Indeed this is the correct thing to do whenever you are not sure that 
your system has a gap. For example, in the case of a structural 
relaxation, even if you know that the converged structure is an 
insulator, you generally ignore what is the electronic structure of the 
starting and of all the intermediate configurations. In this case the 
safest thing to do is to use a small smearing: if the system results to 
be an insulator the smearing will be in the gap and it will not affect 
the total energy (provided that degauss is much smaller than the band gap).

The same idea applies to any kind of dynamics (standard molecular 
dynamics and variable cell dynamics included).

The drawback of the smearing technique is that the number of bands is 
20% larger than in the case of fixed occupation but, in general, this 
overload is worth.


W. YU wrote:
> Dear all pw users,
> In pwscf, if one wants to do a calculation for metal,
> one has to add the following lines:
> occupations = 'smearing',
> degauss = 0.02,
> smearing = 'methfessel-paxton'
> or something similar. This is not neccessary for
> semiconductor or insulator. Now my quesiton is: What
> should I do when I am dealing with a system which is
> expected to undergo a transformation from insulator to
> metal under high pressure, but I don't know when it
> will happens. Of course, I will start without the
> above lines, but ... then, when should I add them, or
> don't add them at all even if I am sure the state
> under investigation is metallic now.  
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around 
> http://mail.yahoo.com 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum

More information about the users mailing list