[Pw_forum] scf_must_converge

Davide Ceresoli davide.ceresoli at istm.cnr.it
Fri Nov 2 09:27:50 CET 2012


Dear Florence,
     which kind of relaxation are you performing? ions, cell, ions+cell?
are you using hybrid functionals? the logic of PW/src/electrons.f90 is
quite complicated and there are several points in which one can check
if (last_step .and. .not. scf_must_converge). If you send me a fragment
of the output, I can have a look.

Best wishes,
Davide


On 11/02/2012 08:06 AM, florence liu wrote:
> Dear all,
> i am trying to do some optimizations with PWSCF v.5.0 and I have the following
> problem.
> I want to let a optimization to continue, even if the scf has not fully
> converged. As I have under stood the QE documentation, one need to set the tag
> scf_must converge, as i have done in my input file:
> (...)
>   &electrons
>   electron_maxstep=100,
>   conv_thr=7.35D-8,
>   scf_must_converge=.false.
> /
> (...)
> however, the optimization calcualtion still terminated with the message:
> "convergence NOT achieved after 100 iterations: stopping" after one scf cycle,
> in whihch the convergence critereriva have not been reached.
> Can anybody tell me, whether I am missing some settings to let the optimization
> not terminate, when a scf cycle has not converged? or does anybody see which
> mistake i have done?
> best wishes,
> florence
> TU Munich
>

-- 
+--------------------------------------------------------------+
   Davide Ceresoli
   CNR Institute of Molecular Science and Technology (CNR-ISTM)
   c/o University of Milan, via Golgi 19, 20133 Milan, Italy
   Email: davide.ceresoli at istm.cnr.it
   Phone: +39-02-50314276, +39-347-1001570 (mobile)
   Skype: dceresoli
+--------------------------------------------------------------+



More information about the users mailing list