[Pw_forum] the sum of the born effective charge in phonon calculation

[Nicola Marzari]

Dear Cheng,


your question provides again a chance for you (and everyone else)
to try out a few things, as exercise - these will, I assure you,
allow you to understand better what is going on.

My understanding is that the greatest error in Born effective charges
comes from BZ sampling - so a good exercise would be to plot
the effective charge on either atom as a function of k-point sampling -
say for a 2,2,2, then 3,3,3, then 4,4,4, ... k-point mesh. You could
plot two curves, one for an unshifted Monkhorst-Pack mesh (i.e. n n n 0 
0 0) and one for a shifted one (i.e. n n n 1 1 1).

See how these value converge; also, since you know that the sum of the
effective charges must be zero, see how the semidifference of the two 
converges (this is equivalent, for 2 atoms, to applying the acoustic sum 
rule, i.e. forcing the sum of the eff charges to be zero by subtracting
to each of the N charges 1/N of the their sum).

Then try to see if increasing or decreasing the cutoff changes the 
values you have. See also if changing the thresholds changes the results
(I tend to be overzealous on thresholds, since it doesn't cost that
much to increase accuracy with a few more iterations. For phonon 
calculations it is typically very useful to converge really well the 
ground state and the phonons themselves).

I'm not sure on the definition of the thresholds - would also be good if 
someone (e.g. from my group ?) could comment on typical safe thresholds 
for ground-state calculations, or for phonon calculations. Especially 
the latter, almost a FAQ, and requiring great care.


				nicola



程迎春 wrote:
> Dear Prof. Eyvaz lsaev,
>         Thank you for your suggestion.  I have another problem 
> concerning this question.  The keyword "conv_thr" is the convergence 
> threshold for scf, and the "th2_ph" is the threshold for selfconsistency 
> in phonon calculation.  What is the difference between the two 
> keywords?   Previously I thought that the two keywords are both for 
> wavefunction optimization and they should have the same meaning. 
> 

-- 
---------------------------------------------------------------------
Prof Nicola Marzari   Department of Materials Science and Engineering
13-5066   MIT   77 Massachusetts Avenue   Cambridge MA 02139-4307 USA
tel 617.4522758 fax 2586534 marzari at mit.edu http://quasiamore.mit.edu