[Pw_forum] Fixing starting starting_ns_eigenvalue

[Giovani Faccin]

Hi Stefano,

Thanks for the reply!

I tried your patch and it did not fix the eigenvalues. So I changed it a
little bit:

Original:
       *IF ( first .AND. istep == 0 .AND. &
                starting_pot == 'atomic' ) CALL ns_adj() ;  rhoin%ns =
rho%ns*

New one:
       *IF ( first .AND. istep == 0 .AND. &
       starting_pot == 'atomic' ) CALL ns_adj();

       IF ( first .AND. istep == 0 .AND. &
       starting_pot == 'atomic' ) rhoin%ns = rho%ns;*

I tested both cases on the same setup (using the intel compiler), and the
second one works out. Maybe it should be updated on cvs?

Regards,

Giovani Faccin
IFGW/Unicamp - Brazil



2011/3/29 Stefano de Gironcoli <degironc at sissa.it>

>  I think this is a bug ...
>
> The initial value of ns should be overwritten by the input request...
> A simple way to obtain this should be to replace, in electrons.f90, the
> line
>
>             IF ( first .AND. istep == 0 .AND. &
>                 starting_pot == 'atomic' ) CALL ns_adj()
> by
>
>           IF ( first .AND. istep == 0 .AND. &
>                 starting_pot == 'atomic' ) CALL ns_adj() ;  rhoin%ns =
> rho%ns
>
> I updated the cvs version.
>
> stefano
>
>
>
> On 03/29/2011 06:55 PM, Giovani Faccin wrote:
>
> Dear Gabriele,
>
> Yes, I tried mixing_fixed_ns. The problem is that this command fixes the
> initial occupation during all the run, not the occupation imposed on step 2
> with starting_ns eigenvalues. I'm attaching an example input and output file
> that shows this happening.
>
> As you can see on the output.txt file, after iteraction 1 the program
> inserts the eigenvalues I wanted on atom 1 spin2 and atom 2 spin2. However,
> just after this it prints:
>
>
> RESET ns to initial values (iter <= mixing_fixed_ns)
>
> on every new iteration. Looking at the end of the calculation, you can see
> that the final states are exactly the default states from step 0. In other
> words, it overwrites the eigenvalues defined by starting_ns eigenvalues,
> placing the default initial occupations instead. Perhaps an "if" clause
> could be added to the code to manage situations when both mixing_fixed_ns
> and starting_ns eigenvalues are defined by the user? Or maybe I'm doing this
> all wrong?
>
> Regards,
>
> Giovani Faccin
>
> IFGW/Unicamp-Brazil
>
>
>
>
>
> 2011/3/29 Gabriele Sclauzero <sclauzer at sissa.it> <sclauzer at sissa.it>
>
>  Dear Giovani,
>
>     have you tried with mixing_fixed_ns?
>
>     Could you please provide your affiliation?
>
>
> Regards,
>
>
> GS
>
> mixing_fixed_ns
> Il giorno 29/mar/2011, alle ore 18.16, Giovani Faccin ha scritto:
>
> Hi all!
>
> Is it possible, in a LDA+U calculation, to fix the starting_ns eigenvalues
> for the entire calculation run, instead of just the second interaction?
>
> Specifically, I'd like to be able to track how a certain electronic
> configuration's associated energy varies when other parameters (ex: atomic
> distances) change. But in order to do that, I need to be able to impose
> specific electronic configurations for the system, even if those
> configurations are not the global energy mininum.
>
> Can that be done in QE?
>
> Thanks!
>
> --
> Giovani
>
>
> _______________________________________________
> Pw_forum mailing listPw_forum at pwscf.orghttp://www.democritos.it/mailman/listinfo/pw_forum
>
>
>
> § Gabriele Sclauzero, EPFL SB ITP CSEA
> *   PH H2 462, Station 3, CH-1015 Lausanne*
>
>
> _______________________________________________
> Pw_forum mailing listPw_forum at pwscf.orghttp://www.democritos.it/mailman/listinfo/pw_forum
>
>
> _______________________________________________
> Pw_forum mailing listPw_forum at pwscf.orghttp://www.democritos.it/mailman/listinfo/pw_forum
>
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>
>


-- 
Giovani
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.democritos.it/pipermail/pw_forum/attachments/20110402/9b6cf427/attachment.htm