[Pw_forum] Hubbard U

Pang Rui pang.r at sustc.edu.cn
Fri May 16 12:00:01 CEST 2014


Dear Tommaso:
I think maybe there are some misuderstanding. The following is what I do. 
1:calculate a scf and save all the output(the key is the *.save folder in
fact)
2:Bulid several folders, for example, 10.
3:Copy the scf file into the 10 folders
4:Put the perturbative input files which have different hubbard alpha into
the 10 folders and do the calculation. Of course, some additional input
must be added as Mr.Cococcioni suggested in the other mail of you. 
5:Grep "Tr" in every output files. The second one is the n0, the last one
is the n.Record the data of the one you take perturbation
6:Fitting two lines of n and n0(x) vs hubbard_alpha(y), the difference of
the slopes is the Hubbard U.
This is the process of calculating single U in a cell.
Hope this can help you.
On Fri, 16 May 2014 10:40:54 +0200, Tommaso Francese
<neutrinofrancese at gmail.com> wrote:
> Dear all,
>  as you suggested, i try to follow your tutorial for the Hubbard U
>  coefficients determination, but i have some practical difficulties.
> 
> Summarizing, i proceded as follow:
> 
> step 1)—> i have done a simple SCF with very low coefficients for
> Hubbard_U1 and Hubbard_U2, enabling the wavefunction separate collection
> (wf_collect) ;
> step 2)—> i have performed some SCF, restarting from the charge file
> stored previously,  introducing some fluctuation values in Hubbard_alpha
> (?), is it correct? 
> 
> the problem is at this point: do i need to rename every scf-file in
which
> i performed a different perturbation? How can i, in the next step,
“recall”
> the different perturbation that i performed previously? 
> 
> At this point is obvious that i don’t know also how to calculate the
> response functions and so, obtaining the Hubbard U coefficient.
> 
> Can someone help me? 
> Thanks a lot,
> Tommaso,
> Università Cà Foscari of Venice
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum



More information about the users mailing list