[QE-users] relaxation of a charged system

[Paolo Giannozzi]

The energy of charged cells depends upon a rather arbitrary choice of the
G=0 potential. If you compare the energies from CP with those from PW, you
will notice that they are quite the same (once the factor 2 between Ry and
Ha is taken into account!) for neutral cells, while they differ for charged
cells: CP uses a different convention for G=0. In principle, forces and
stresses might also depend upon the choice of the G=0. I have never felt or
heard that this effect is in practice relevant for structural optimization.
I am less sure about variable-cell optimization, though

Paolo

On Fri, Apr 30, 2021 at 4:11 PM Hanghui Chen <chenhanghuipwscf at gmail.com>
wrote:

> Dear QE developers,
>      I have been using Quantum Espresso for more than a decade and
> recently a paper (PRB 91, 024107 (2015)) caught my attention, which shows
> that the total energy, pressure and cell relaxation are ill-defined in
> charged systems (even if a compensating charge background is present). The
> paper shows that the total energy and pressure of a charged system depend
> on the average electrostatic potential convention, and thus is
> code-dependent (QE, ABINIT and VASP use different conventions).
>     I think the paper is theoretically correct. However in practice, we
> have recently tested doped ferroelectric materials using both QE and VASP.
> In this charged system, we find both QE and VASP yield qualitatively
> consistent structural properties.
>     Since relaxation of a charged system is very common in literature
> using QE or VASP (e.g. PRL 109, 247601 (2012) and many others), I am
> wondering whether the relaxed structure of a charged system from QE is
> reliable. According to PRB 91, 024107, it is not legitimate to do cell
> relaxation of a charged system. However both QE and VASP do allow to
> calculate total energy and perform cell relaxation in a charged system. And
> from literature, it seems that the relaxed structure of a charge system
> from QE or VASP calculations is "not too wrong".
>     So what is your insight on this issue? Shall we prohibit ourselves
> from doing cell relaxation of a charged system in QE, or can we still use
> the standard procedure but with the caution that the optimized structure of
> a charged system from QE might not be accurate (due to the QE convention of
> average electrostatic potential)?
>      Thank you very much.
>
> Dr. Hanghui Chen
> Department of Physics
> NYU Shanghai and New York University
>
> _______________________________________________
> 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



-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 206, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20210502/68d2e263/attachment.html>