[Pw_forum] about Phonon results

Stefano de Gironcoli degironc at sissa.it
Tue Jun 17 10:03:51 CEST 2008


Eduardo Ariel Menendez Proupin wrote:
> This happens because
> a) 0 frecuencies at Gamma occurs because the energy is is invariant 
> with respect to arbitrary translations of the whole crystal (the same 
> translation for all atoms). From this assumption the accoustic sum 
> rule (ASR) is obtained theoretically.
> b) The plane wave DFT calculations do not have this invariance 
> exactly, because of the plane wave cutoff. The energy is invariant not 
> for arbitrary translations, but for translations in the FFT grid.
> c) I guess that the absolute translation invariance is lost in the 
> back and forth FFTs of the exchange-correlation potential, which is 
> non linear, and probably in some part related with the ultrasoft 
> pseudopotentials. Someone correct me if I am wrong, please.
everything correct, except for the clarification of a detail.
The problem is actually not in the finite kinetic energy cutoff but in 
the fact that the exchange and correlation energy is computed in real 
space on a discrete grid and hence the total energy is invariant (as you 
said) only for translation in the FFT grid.
Increasing the charge density cutoff increases the grid density thus 
making the integral more exact thus reducing the problem.... 
unfortunately rather slowly...
This problem is usually more severe for GGA  than with LDA because the 
GGA functionals have functional forms that vary more strongly with  the 
position; particularly so for isolated molecules or system with 
significant portions of "vaccum" because in the exponential tail of the
charge density a) the finite cutoff  (hence there is an effect due to 
cutoff) induces oscillations in rho and b) the reduced gradient is diverging

stefano
 
> You may increase the density cutoff (ecutrho) and see if the first 
> three frequencies get closer to zero. First do it with one of the 
> ESPRESSO examples, that run in minutes. I guess that I made that test 
> some years ago.
> Using larger cutoffs increase a lot the calculation time. That is why 
> you have the tricky option to impose the ASR in postprocessing (matdyn.x).
>
> Practical way. If your "wrong" frecuencies are smaller than 50 cm^-1, 
> this is normal. Apply the ASR in the input for matdyn. Check that the 
> other frequencies do not change more than 1 or 2 cm^-1. Then, if  this 
> makes you happy, go to write your report
>
> >I have another question about the output: What is the >difference 
> between the total magnetic moment and the >absolute magnetic moment.
> This has been asked in the forum before. Please, google for this. If I 
> remember well, the absolute magnetic moment is the integral of the 
> absolute value of the magnetization.
>
> Regards
> Eduardo Menendez
>
> On Sat, Jun 14, 2008 at 2:57 AM, <xwzhang at uestc.edu.cn 
> <mailto:xwzhang at uestc.edu.cn>> wrote:
>
>     Dear my pwscf friends,
>
>     Recently, I am trying to do phonon calculations in Heusler. As a
>     first step, I tried to reproduce the results of a paper by
>     Zayak(PRB, 72, 054113, 2005). I choose Ni2MnGe as an example. To
>     my surprise, the acoustic branches at Gamma point gives a
>     frequency of about 1 THz, which should be zero. Other branches
>     agree with the published data.
>
>     I wonder what might be the reason? I attached my scf, and phonon
>     input files, and the phonon out put files.
>
>     I have another question about the output: What is the difference
>     between the total magnetic moment and the absolute magnetic moment.
>
>     Thank you and wish you a happy weekend!
>
>     Wenxu Zhang
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>   



More information about the Pw_forum mailing list