[Pw_forum] cp with metals

[Nicola Marzari]

Ciao Carlo,

eDFT is a fully variational formulation - that is, it
guarantees that at every step the free energy functional
always goes down, so the code is formally bound to converge
under any circumstance. This is particularly important as
the size of your system grows. Mixing schemes can be very good
at brinign you to selfconsistency, but they are not self-consistently
variational (e.g. they can minimze the energy at fixed hamiltonian,
but when you update the charge density, and the hamiltonian, you
are not guaranteed that your energy has decreased. Again, this becomes
much more of an issue in metals, especially large systems, or in
small molecules with a lot of level crossings - i.e. when the fraction
of states that change their occupation is a significant contribution to
the total charge density).

To answer Nicholas question - everything is in 3.0: example29
of the current public distribution. Keep in mind that, being this the
first espresso release, it still needs a lot of use and testing to iron
out the code.

But the algorithm has been used extensively (admittedly, by 2 or 3
people) for many years in a different code, and it is exceedingly
robust (again, if implemented correctly).

Best,

			nicola



carlo sbraccia wrote:

> Dear Nicola,
> 
>> In the recently developed version 3.0 of Espresso, the CP code
>> can also do Born-Oppenheimer moelcular dynamics with variable
>> occupations (we have a 1997 PRL on that).
>>
>> It probably needs a few more weeks of testing to make sure
>> everything is ironed out, but that would be the first choice to
>> deal with system that are really metallic (the electronic thermostat
>> helps only if the loss of adiabiticity is small).
> 
> 
> what is the advantage of this approach with respect to the standard 
> scf+smearing that is implemented in PWscf ?
> 
> carlo
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum

-- 
---------------------------------------------------------------------
Prof Nicola Marzari   Department of Materials Science and Engineering
13-5066   MIT   77 Massachusetts Avenue   Cambridge MA 02139-4307 USA
tel 617.4522758 fax 2586534 marzari at mit.edu http://quasiamore.mit.edu