[Octopus-devel] sternheimer

[David Strubbe]

Miguel,

Since the Van der Waals calculation consists of electric-response
runs, I tried to find a simple example with a convergence problem for
just electric response.  I ran calculations of the H2O test for
electric response, using 02-h2o_pol_lr.01_h2o_gs.inp and
02-h2o_pol_lr.02_h2o_dyn_pol.inp.  There is no problem with
convergence using these files as written, but the newer versions
converge the ground state to a tighter tolerance than 3.0.1 due to a
change in the meaning of the criterion.  If the trunk version is run
with ConvRelDens = 1.0e-2 instead of 1.0e-4, then the ground state is
converged to the same tolerance as in 3.0.1.  In this case, the
Sternheimer SCF iterations diverge in direction 2 for frequency zero,
but converges in the other parts of the calculation.  I determined
that this problem does not exist for revisions 4211, 4233, 4244, 4250;
it does exist for revisions 4255, 4275, 4298, 4385.

Namely, the error is introduced between r4250 and r4255, the point
where you introduced smearing into the Sternheimer equation.  Despite
various fixes to these routines by r4275, the problem persists.  So it
seems that there is still a problem in the implementation of
semiconducting smearing for linear response (i.e. I didn't break it,
you did :-p).  Perhaps with this information, and being more familiar
with these parts you wrote, you can find the problem.

This also suggests a work-around, which is to converge the ground
state more highly if there is a problem in converging Sternheimer.

David

2008/7/2 Miguel Marques <Miguel.Marques at tddft.org>:
>  Hi all,
>
>  I found that the Sterheimer equation, for the case of the calculation
> of vdw coefficients, is having enormous difficulties converging now, and
> this did not happen before. Any of you guys has any idea of what is
> going on? I am kind of busy now with benasque, so I do not think I will
> have time to look into it...
>
>  cheers,
>  Miguel
>
> --
> Dr. Miguel A. L. Marques
> marques at tddft.org
>
> Laboratoire de Physique de la Matière Condensée et Nanostructures
> (LPMCN) - Université Lyon I
> Bâtiment Brillouin, Domaine scientifique de la DOUA
> 69622 Villeurbanne Cedex
> Tel +33 (0)4 72448187
> Fax +33 (0)4 72432648
> _______________________________________________
> Octopus-devel mailing list
> Octopus-devel at tddft.org
> http://www.tddft.org/mailman/listinfo/octopus-devel
>