[Wien] Doubt in calculation with spin-orbit interaction

[Marcos Veríssimo Alves]

Hi Xavier and Stefaan,

Thanks for the responses. However, what I am doing is a convergence study in
the same spirit as one would do for calculating a DOS, or optical
properties:

1) The scf convergence on the k-point mesh is monitored with respect to some
quantity (say, the magnetic moment, or/and value of the Fermi level as is
customarily done for electronic transport calculations). For a hypothetical
system, let's say that the Monkhorst-Pack mesh which is used for a very well
converged density matrix is 30x30x30;

2) but then, for the calculation of the DOS/PDOS/optical properties, one
uses a denser k-mesh, calculating non-self-consistently the eigenvalues
using the density matrix converged in step (1). For this hypothetical
system, the k-mesh would be varied from 30x30x30 till say 80x80x80, for
which the DOS doesn't change any more.

My doubt was if, besides running lapw1, if I would also have to run lapwso
in a system with spin-orbit coupling (non-self-consistently, always with a
well-converged density matrix), for all those different values for the
denser k-mesh. From Stefaan's answer, I see that this should indeed be done.
I'm still "stepping on eggs" with W2k, especially because soon I'll have to
do some really serious calculations to wrap some work up, but I think I'm
starting to get the grip :)

Thanks a lot,

Marcos


On Thu, Jul 22, 2010 at 9:50 PM, Stefaan Cottenier <
Stefaan.Cottenier at ugent.be> wrote:

>
> As you mention 'x lapw1' and 'x lapwso' (and not 'run_lapw -so'): is your
> goal to make a k-mesh test by inspecting convergence of the
> non-selfconsistent eigenvalues of the first scf-iteration as a function of
> k-mesh density? (this is indeed a reasonable alternative for a k-mesh test
> using converged scf results -- and much faster) In that case, you can
> monitor the eigenvalues with and without spin-orbit separately. I guess both
> should be converged in order to decide you have a good k-mesh.
>
> Hence:
>
> x kgen (first input)
> x lapw1
> x lapwso
> inspect case.output1 and case.outputso
>
> x kgen (second input)
> x lapw1
> x lapwso
> inspect case.output1 and case.outputso
>
> etc.
>
> Stefaan
>
>
>
> Marcos Veríssimo Alves wrote:
>
>> Dear all,
>>
>> I have finished the scf cycle for Bi2Te3, non-spin-polarized and with
>> spin-orbit interaction included, and now I have a doubt in how to perform
>> the calculation of the eigenvalues with denser and denser k-meshes.
>> Basically, I want to perform a convergence study on the number of k-points
>> in the IBZ wedge to try and reproduce the results of the PRB article of
>> Scheidemantel et al, PRB 68, 125210 (2003) and Madsen and Singh, CPC 175
>> (2006) 67-71.
>>
>> Normally (please correct me if I'm wrong) if there weren't spin-orbit, it
>> would simply be a matter of generating denser k-meshes using kgen, and for
>> each of them issuing "x lapw1". However, now that I have spin-orbit, should
>> I issue "x lapwso"  right after "x lapw1" ? I have issued "x lapw1" but the
>> only file that changes is Bi2Te3.energy. It seems logical to me that I
>> should issue x lapwso as well but I am still a bit unsure, and I want to be
>> certain of the correct procedure to be followed without the interface, since
>> I guess it'll be more straightforward to run it in the IBM machine we have
>> at our disposal by inserting the correct commands in the script file...
>>
>> Thanks in advance,
>>
>> Marcos
>>
>>
>> ------------------------------------------------------------------------
>>
>>
>> _______________________________________________
>> Wien mailing list
>> Wien at zeus.theochem.tuwien.ac.at
>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>>
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20100723/7e9f8b09/attachment.htm>