[QE-users] "Re2: Convergence Issues?"

[Stefano Baroni]

We do not necessarily disagree. The internal energy is harder to converge than the free energy, because the first is not variational, whereas the latter is so. Also, level crossing is obviously much reduced (and eventually eliminated) for large enough smearing. S.

> On 25 Jan 2021, at 16:59, Nicola Marzari <nicola.marzari at gmail.com> wrote:
> 
> 
> 
> Dear Stefano, Lucas,
> this is one of the few areas where I disagree with Stefano - I think we should use the free energy (forces and stresses are the total derivative of the free energy, not the internal energy), and that level-crossing instabilities are exceedinfly difficult to cure with iterative algorithms (my own opinions are here: http://theossrv1.epfl.ch/Main/ElectronicTemperature <http://theossrv1.epfl.ch/Main/ElectronicTemperature>),
> 
> 
>             nicola
> 
> 
> On 25/01/2021 16:33, Stefano Baroni wrote:
>> I beg to slight differ. It does not make much sense to assign a “default value” to the Gaussian smearing, unless one makes sure that the computations are well converged with respect to the number of k points, for that specific value of the smearing. The rational procedure to follow is: 1) start with whatever value of the smearing one wishes, and make sure that the (internal, *not* “fake free”) energy is well converged with respect to the number of k points. If the number of k points is sufficient, it is likely that no nasty oscillation would arise; 2) decrease the smearing and vary the number of k points accordingly. HTH — Stefano B
>> 
>> 
>> —
>> Stefano Baroni -  SISSA, Trieste - http://stefano.baroni.me <http://stefano.baroni.me/>, stefanobaroni (skype)
>> 
>> I do not mind if you think slowly, but I do object when you publish more quickly than you think [W.E. Pauli to an unknown colleague, as reported by N. Kemmer]
>> 
>>> On 25 Jan 2021, at 15:08, Lucas Nicolás Lodeiro Moraga <lucas.lodeiro at ug.uchile.cl <mailto:lucas.lodeiro at ug.uchile.cl>> wrote:
>>> 
>>> Hi, 
>>> It appears a metallic oscillation, where a couple or more states change their relative energy, and change their energetical order, then the occupation changes abruptly (non continuous). If this is the problem, it is better to increase the degauss value and/or change the smearing method.
>>> Typically I use methfessel-paxton with 0.015 Ry for degauss, which mimics the VASP default values. When you get a converged system, you can try to decrease (slowly) the degauss.
>>> 
>>> Regards - Lucas Lodeiro
>>> _______________________________________________
>>> Quantum ESPRESSO is supported by MaX (www.max-centre.eu <http://www.max-centre.eu/>)
>>> users mailing list users at lists.quantum-espresso.org <mailto:users at lists.quantum-espresso.org>
>>> https://lists.quantum-espresso.org/mailman/listinfo/users <https://lists.quantum-espresso.org/mailman/listinfo/users>
>> 
>> 
>> _______________________________________________
>> Quantum ESPRESSO is supported by MaX (www.max-centre.eu <http://www.max-centre.eu/>)
>> users mailing list users at lists.quantum-espresso.org <mailto:users at lists.quantum-espresso.org>
>> https://lists.quantum-espresso.org/mailman/listinfo/users <https://lists.quantum-espresso.org/mailman/listinfo/users>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20210125/d70ba809/attachment.html>