[QE-users] about nstep in thermo pw

Sathya Perumal sathyasrrperumal at gmail.com
Sat Sep 29 12:39:59 CEST 2018


Dear Andrea,
                     Thanks for the help. I also noticed that, for the
first deformation geometry optimization is done(for frozen_ions=.false. in
thermo_control)  with the confirmation message  "End of BFGS Geometry
Optimization", for the specified nstep or lesser than nstep according to
the conv. criteria.. For subsequent strained configurations the BFGS
optimisation is done only for ONE nstep, and then the code leaves for the
next deformation. BFGS optimization is not done for subsequent
deformations. Could you please comment on this?. Thanks

 egrep "BFGS|energy   new|and now|nstep" aelas.out
     I am the master and now I do geometry    3 /   24
     nstep                     =           20
     BFGS Geometry Optimization
     energy   new            =    -617.8695318083 Ry
     energy   new            =    -617.8696044767 Ry
     End of BFGS Geometry Optimization
     I am the master and now I do geometry    6 /   24
     nstep                     =           20
     BFGS Geometry Optimization
     energy   new            =    -617.8694741420 Ry
     I am the master and now I do geometry    8 /   24
     nstep                     =           20
     BFGS Geometry Optimization
     energy   new            =    -617.8691134743 Ry
     I am the master and now I do geometry   12 /   24

the above run was done in parallel, thus deformation geometries are not
sequential, but similar message in out_?_0 files.

Thanks and regards
Sathya S R R Perumal

On Wed, 26 Sep 2018 at 17:45, Andrea Dal Corso <dalcorso at sissa.it> wrote:

> You can increase nstep in src/set_thermo_work_todo.f90 and recompile. It
> should be an input parameter but it is not.
>
> Thank you for reporting,
>
> Andrea
>
>
> On Wed, 2018-09-26 at 15:40 +0530, Sathya Perumal wrote:
> > Dear Andrea,
> The total energy at the each deformation is not converging within 20 step.
> I thought of changing nstep and compile again, is it a recommended value to
> stick to. Or should I use very loose convergence for force and energy?,
> what are the typical value for convergence for such run.
>
> I'm using scf_elastic with advanced algorithm. the system is  molecular
> crystal.
> --
> Andrea Dal Corso                    Tel. 0039-040-3787428
> SISSA, Via Bonomea 265              Fax. 0039-040-3787249
> I-34136 Trieste (Italy)             e-mail: dalcorso at sissa.it
>
>

-- 
S S R R Perumal PhD
TUE-CMS HPC Data Center
Solid State and Structural Chemistry Unit,
IISc, Bangalore.
sathyasrrperumal at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20180929/061a7809/attachment.html>


More information about the users mailing list