[Pw_forum] VC-Relax taking too long on a 4-atom system

[stefano de gironcoli]

Dear Elliot,

  i would say that electron_maxstep = 1000000 is not a great idea....  
if the scf does not converge in 50-100 iterations there is something 
weird in the electronic configuration and it might go on forever.... 
better stop and try to understand what it's going on.
  smearing    ='fermi-dirac' (why FD ? usually one choose MV or MP)
  degauss      =0.003 this is a very small degauss that will require a 
lot of k-points to converge. A larger degauss should improve scf stability
   nbnd=24 is a bit tight for 4 Ni ferromagnetic atoms. i would allow a 
little margin like 30 or so
   etot_conv_thr = 1.0D-7, forc_conv_thr = 1.0D-6, are unrealistically 
small... especially forc_conv_thr. starts with the default and then 
change from that.

   stefano

On 10/22/2014 04:04 PM, Elliot Menkah wrote:
> Hi Everyone,
> I'm trying to determine a lattice parameter in a study and I'm running a
> vc-relax calculation.
> It's just a 4-atom bulk ferromagnetic nickel system yet my calculation
> has been running for over 5hrs already on a 16 core resource.
>
> It's presently on ' iteration #*** '
>
> Is there anything wrong with the setup of the job?
>
> Shouldn’t the job have completed by now?
>
> Could you please peruse and vet my input file for any possibles reasons
> why it's so if it shouldn’t be?
>
> What could be affecting the convergence?
>
>
> Thank you
>
>
>
>
> Kind Regards,
>
> Elliot
>
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20141022/bc994ead/attachment.html>