I could manage it by playing with ecutwfc and ecutfock.
Regards
Bhamu
On Mon, Aug 19, 2019 at 1:32 PM Dr. K. C. Bhamu wrote:
> Dear Experts,
> I am running GAUPBE calculations for a FCC binary scalar relativistic case
> with latest version of QE with mkl-2015 on a cluster with pseudo-dojo PPs
> and SG15 ONCV.
> K_POINTS {automatic}
> X X X 0 0 0(X=12, 15).
> taking converged value of ecutwfc (4meV). Beyond this criteria I am facing
> memory issue.
>
> I am getting the dexx is negative error:
> Error in routine electrons (1):
> dexx is negative! Check that exxdiv_treatment is appropriate for the
> system, or ecutfock may be too low
>
> I would like to mention two case where I encounter this error:
> 1.
> k_POINTS
> 12 12 12 0 0 0
> nqx=1,2,3,6 finished well but error was occurred with nqx=4.
> 2.
> k_POINTS
> 15 15 15 0 0 0
> nqx=1,5 finished well but error was occurred with nqx=3.
>
> I have looked over all the available threads on this forum and trying to
> overcome this error.
>
> To dig into the available options I would like to ask about input
> variables of gaupbe.
>
> 1. exxdiv_treatment: as per pw.x input descriptions it should be none but
> the error says "exxdiv_treatment" ?
> Can we use other approach to be used for treating the Coulomb potential
> divergencies? I am afraid of using option other than none as the other
> options are suggested for supercells.
>
> 2. By default ecutvcut is 0.0.
> Can we vary this ecutvcut with gaupbe? if yes then what is the approximate
> range?
>
>
> regards
> Bhamu
>
>
>
___
Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
users mailing list users@lists.quantum-espresso.org
https://lists.quantum-espresso.org/mailman/listinfo/users
