Re: [Phonopy-users] Problem with setting nac_params with Phono3py API

[Thomas H. <tha...@gm...>]

Are there examples of how the choice of NAC direction leads to erroneous
results? I know that in the Gonze paper, the authors discussed how the LO
eigenvector can change with the NAC term, which may depend on the
q-direction. Also, when using the Phonopy API to get phonons in mesh-mode,
e.g.
*phonopy =* *phonopy.load(vasp_inputs, nac_params)*
*phonopy.run_mesh()*
automatically sets the nac_q_direction to None during the operation of the
code. So, is the only way to get the phonons with LO-TO splitting from
phonopy using the *phonopy.run_qpoints()* method?

Thanks for all the help by the way!

On Wed, Apr 28, 2021 at 8:14 PM Atsushi Togo <atz...@gm...> wrote:

> Hi,
>
> It's implied by the formula. There is no standard direction.
>
> Togo
>
> On Thu, Apr 29, 2021 at 11:42 AM Thomas Harrelson <tha...@gm...>
> wrote:
> >
> > I see, so by default, setting the nac_params does not turn on the
> non-analytical term correction when calculating the phonons in mesh mode?
> Is there a standard direction that is typically used, or does the direction
> change the convergence rates, and thus any direction can be used?
> >
> > On Tue, Apr 27, 2021 at 8:06 PM Atsushi Togo <atz...@gm...> wrote:
> >>
> >> Maybe also,
> https://github.com/phonopy/phono3py/blob/5c66ad0a47ec349aa117f3f75b841242c6171d58/phono3py/api_phono3py.py#L926
> >>
> >> Togo
> >>
> >> On Wed, Apr 28, 2021 at 12:04 PM Atsushi Togo <atz...@gm...>
> wrote:
> >> >
> >> > Hi,
> >> >
> >> > See
> https://phonopy.github.io/phono3py/command-options.html#q-direction-q-direction/
> >> >
> >> > Togo
> >> >
> >> > On Wed, Apr 28, 2021 at 11:44 AM Thomas Harrelson <tha...@gm...>
> wrote:
> >> > >
> >> > > I am attempting to use the Phono3py API to calculate the imaginary
> self energy, phonon frequency, and phonon eigenvector arrays for further
> analysis within a code. I'm using this for materials with optically active
> phonons, so the NAC terms need to be present. At the moment, I'm following
> the code in Si.py script in the examples folder:
> >> > >
> >> > > primitive = phono3py.get_phonon_primitive()
> >> > > nac_params = parse_BORN(primitive, filename="BORN")
> >> > > nac_params['factor'] = Hartree * Bohr
> >> > > phono3py.set_phph_interaction(nac_params=nac_params)
> >> > > data = phono3py.get_phonon_data()
> >> > >
> >> > > Since the set_phph_interaction is deprecated, I also tried:
> >> > > phono3py.nac_params = nac_params
> >> > > phono3py.init_phph_interaction()
> >> > >
> >> > > In both of the above cases, the program works for GaAs, but there
> is no apparent LO-TO splitting for the optical phonons at gamma (all
> optical frequencies are the same), indicating that no NAC method was used.
> The phono3py._interaction object has the nac_params correctly loaded within
> it. Also, it appears that phono3py._interaction._dm has all the nac_params
> as well. Then why is there no LO-TO splitting? Am I setting up the code
> incorrectly, or what am I missing?
> >> > > _______________________________________________
> >> > > Phonopy-users mailing list
> >> > > Pho...@li...
> >> > > https://lists.sourceforge.net/lists/listinfo/phonopy-users
> >> >
> >> >
> >> >
> >> > --
> >> > Atsushi Togo
> >>
> >>
> >>
> >> --
> >> Atsushi Togo
> >>
> >>
> >> _______________________________________________
> >> Phonopy-users mailing list
> >> Pho...@li...
> >> https://lists.sourceforge.net/lists/listinfo/phonopy-users
> >
> > _______________________________________________
> > Phonopy-users mailing list
> > Pho...@li...
> > https://lists.sourceforge.net/lists/listinfo/phonopy-users
>
>
>
> --
> Atsushi Togo
>
>
> _______________________________________________
> Phonopy-users mailing list
> Pho...@li...
> https://lists.sourceforge.net/lists/listinfo/phonopy-users
>