[Pw_forum] ecutrho and rrkj psp

[Stefano de Gironcoli]

there are two reasons why ecutrho > 4*ecutwfc may be needed
-USPP: rho is intrinsically  harder than 4*ecutwfc
-XC functionals: while hartree needs the same cutoff as the density, the 
xc functional may require higher an cutoff. this is particularly so for 
gradient corrected functionals while LDA are usually not so hard.

BEWARE the pseudopotentaila naming: _rrkj are NCPP but _rrkjus are USPP !

which pesudopotentials are you using ?

stefano





On 01/31/2011 12:41 PM, Davide Sangalli wrote:
> Dear forum,
> I'm doing calculations with rrkj pseudo-potentials.
>
> As far as I understand rrkj are norm-conserving psps and only ecut should be
> checked, while ecutrho=4*ecut should be fine.
>   From this I guess that the ecutrho paramater, if increased, should not
> influence the convergence.
>
> I did some test and I saw that this is not true. Increasing ecutrho (at fixed
> ecut) the results changes significantly. For example the elements of the stress
> tensor changes more than 15 Kbar (in some case even more) moving ecutrho from
> 140 to 300 Ry (with ecut=35 Ry). Here I was testing FeO, CoO and NiO, with
> pbe/rrkj psps for all thee elements.
> *Question 1: Why ?
> *(I saw that in the psps header Rcut and Rcut US are different ... ?)
>
> I did these checks even because I would like to perform calculations using
> vanderbilt psp (for some elements) together with rrkj psp (for other element?).
> *Question 2: Is there any counter indication in doing that ?
> *
> Thanks and best regards,
> Davide Sangalli
>
>
> _______________________________________________
> 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://lists.quantum-espresso.org/pipermail/users/attachments/20110131/fa2fbb84/attachment.html>