[Pw_forum] LO-TO splitting in dynmat.x

[Stefano de Gironcoli]

the electristatic interaction at the origin of LO-TO splitting is always 
present...

for any q<>0 (even very small) it is included in the calculation.

in the limit of q->0 in non-metallic systems it gives origin to a non 
analytic behavior that must be calculate separately.

If you want to Fourier interpolate the phonon dispersions calculated on 
a regular grid of q-points
you are in trouble because non analyticity of the phonon dispersion 
implyes long-range (1/R^3) interatomic force constants and so you need to

1) evaluate Z* and epsilon_infty in the limit of q->0 that determine the 
non-analyticity
2) remove from dynamical matrix in every q point in you grid an 
electrostatic model that gives the correct non-analyticity for q->0 and 
is smooth elsewhere
3) Fourier interpolate  the modified (hopefully short-range) dynamical 
matrices
4) add back the model in any q-point you want to study.

This is what the sequence ph.x -> q2r.x -> matdyn.x does (in example06 
for instance)

thete is some discussion of these issues in Review of Modern Physics 73, 
515 (2001) and in Phys Rev 43, 7231 (1991)

stefano

On 04/29/2011 05:05 AM, xirainbow wrote:
> Dear Eyvaz:
> Thank you very much;)
>
>> Is there LO-TO splitting far away from Gamma point?
>> No.
>>
> Does LO-TO splitting must disappear at the boundary of Brillouin zone?
>
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.democritos.it/pipermail/pw_forum/attachments/20110429/093d10e5/attachment.htm