[Pw_forum] Bug (or not) with epsil = .false. in PH (trunk version)

[Andrea Dal Corso]

On Sat, 2015-06-13 at 16:38 +0100, Samuel Poncé wrote:
> Dear all,
> 
> I found out that it was impossible de make a phonon calculation without
> calculating the Born effective charge in a semi-conductor even if we set
> epsil = .false.
> 
> This is due to the routine prepare_q.f90 that is called inside do_phonon.f90
> In the routine there is
>      IF ( lgamma ) THEN
>         !
>         IF ( .NOT. lgauss ) THEN
>            !
>            ! ... in the case of an insulator at q=0 one has to calculate
>            ! ... the dielectric constant and the Born eff. charges
>            ! ... the other flags depend on input
>            !
>            epsil = .TRUE.
>            zeu = .TRUE.
>            zue = .TRUE.
> 
> 
> This means that if we compute q=Gamma and if we do not have
> gaussian-smearing (i.e. an semi-cond or insulator), then epsil is
> automatically set to TRUE.
> 
> I know that physically one should have such LO/TO splitting but the user
> should be able to choose.
> 
> Maybe this forcing could be reported in the input variable ? or simply put
> the default value to .TRUE. instead of false and not enforcing that rule?
> 
> What do you think?

This is true only when ldisp=.TRUE.. If you do a phonon calculation at
gamma with ldisp=.FALSE. you can set/unset the dielectric constant
calculation with the flag epsil.

It has to be like this, otherwise the dispersion would show a jump at
gamma and the interpolation would not work.

HTH,

Andrea


> 
> Best,
> 
> Samuel Ponce,
> Department of Materials, University of Oxford
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum

-- 
Andrea Dal Corso                    Tel. 0039-040-3787428
SISSA, Via Bonomea 265              Fax. 0039-040-3787249
I-34136 Trieste (Italy)             e-mail: dalcorso at sissa.it