Dear yoshi,
Thank you very much!
2010-05-10
Guangping Zhang
发件人: yo...@df.ufscar.br
发送时间: 2010-05-10 11:16
主 题: Re: [SIESTA-L] why transiesta can not get convergence
收件人: siesta-l@uam.es
Hi,
I don't understand in details the variable OccupationFunction too, but
there is reference in the manual. In page 55 of pdf siesta-3.0-b:
"OccupationFunction (string): String variable to select the function that
determines the
occupation of the electronic states. Two options are available:
?FD: The usual Fermi-Dirac occupation function is used.
?MP: The occupation function proposed by Methfessel and Paxton
(Phys. Rev. B, 40,
3616 (1989)), is used.
The smearing of the electronic occupations is done, in both cases,
using an energy width...."
In my calculations, the convergence problem occurs and I thought that the
charge at interface between scattering and the electrode regions was
suffering variation. Because this, I expand the scattering region with
more planes of electrodes. Thus, the electrodes remain far from
low-dimensional structures or impurities that are in the scattering
region, avoiding the charge variation at interface. I don't know if this
is correct, I hope that someone confirm this thinking, or not.
>From my calculations, I tested transiesta in nanoribbons with a vacancy,
there isn't convergence problem. In wires of carbon, the problems occurs.
In the last one I used FD in OccupationFunction and I solved the problem
with the inclusion of 3 electrode structures in scattering region. From
Topsakal's files, the MP was used and the convergence is great! You should
test this options. There is one more option that could help you:
TS.UpdateDMCROnly
search this at siesta user's guide.
regards
Dear yoshi,
I can not understand OccupationFunction MP and FD,this option is not in the
manual,is it useful to use it in the transiesta *.fdf input?
Bye the way I have a Au(111) electrode-3*3*4.
It is not enought?
Thanks !
2010-05-09
Guangping Zhang
发件人: yo...@df.ufscar.br
发送时间: 2010-05-09 21:42
主 题: Re: [SIESTA-L] why transiesta can not get convergence
收件人: siesta-l@uam.es
About this problem,
You can include more electrode structures in scattering region. This
should increase the convergence.
There are two way to perform scf:
OccupationFunction MP
or the default
OccupationFunction FD
if you use the MP, you can use high temperatures, like 600-700. But beware
about the temperature at FD option.
regards.
> If you are using DZP, you may need to increase the tempereture to 600-700
> K.
> SZP usually converges at 300K.
>
>
>
> On Sun, May 9, 2010 at 1:28 PM, Kemal Bagci <vmkba...@gmail.com> wrote:
>
>> Hello everyone,
>>
>> I have encountered the same problem. My convergence numbers were much
>> worse
>> though.
>>
>> I believed that the problem had to do with reading the .TSHS file
>> obtained
>> from the electrode run, but I think it is deeper then that.
>>
>> If anyone had encountered a similar problem, let's exchange information.
>> I
>> can also send the output file I have for the failed run.
>>
>> Cheers,
>>
>> Kemal
>>
>
>
