[Pw_forum] pw.x seg. faults with EXX

Omololu Akin-Ojo prayerz.omo at gmail.com
Thu Aug 4 13:04:15 CEST 2011


Ciao ancora Tutti,

It seems I diagnosed the problem: with the pbe0 hybrid, it seems you
cannot set ecutrho to more than 4 times ecutwfc. I don't know why. The
default seems to be ecutrho = 4 X ecutwfc so I just removed my
specification for ecutrho and it runs without segfault. So far it is
still running.

Grazie.

o.

On Thu, Aug 4, 2011 at 5:50 AM, Omololu Akin-Ojo <prayerz.omo at gmail.com> wrote:
> Ciao Tutti,
>
> I am trying to do a calculation with the PBE0 functional. I compiled
> pw.x using -DEXX. However, I got the following message when I tried to
> run. (If I replace input_dft='pbe0' with input_dft='pbe', I don't have
> this error but this is obviously not what I want.
> Thanks in advance for any help.
>
>     XC functional enforced from input :
>     Exchange-correlation      = PBE0 (6484)
>     EXX-fraction              =        0.25
>     !!! Any further DFT definition will be discarded
>     !!! Please, verify this is what you really want !
>
>     file O.pbe-mt.UPF: wavefunction(s)  0S 0P 0D renormalized
>
>     Subspace diagonalization in iterative solution of the eigenvalue problem:
>     a serial algorithm will be used
>
> [compute-1-7:09041] *** Process received signal ***
> [compute-1-7:09041] Signal: Segmentation fault (11)
> [compute-1-7:09041] Signal code: Address not mapped (1)
> [compute-1-7:09041] Failing at address: 0x5
>
>
> o.
>
> Postdoctoral Fellow,
> ICTP, Trieste, ITALY
>
> --
> ***************** Seek GOD! *************
>



-- 
***************** Seek GOD! *************



More information about the users mailing list