Re: [Pw_forum] kpoints for transmission calculations and complex band structures?

[Yue-Wen Fang]

Dear Dr. Filipe,

Thank you for your kind assistance! It seems that I cannot get empirical
parameters here, thus performing many tests for k grid convergence as you
said could be the only way to settle the matter.

Bests
Fang



2015-09-11 11:34 GMT+08:00 Filipe Camargo Dalmatti Alves Lima <
fl...@if.usp.br>:

> Dear Fang,
>
> Unfornately I don't have any experience with the systems you are
> investigating, but I'll add my comments that might provide you some
> insights.
>
> The answer for your question is simple but will require some effort to
> accomplish it: you must test your system and make sure the properties are
> converged.
>
> Convergence means the following: As you increase a given parameter that
> you are tracking, the results are not changing anymore.
>
> In order to achieve that for your systems, I would recommend two things:
>
> 1) Look for the literature and check the parameters they used the same
> systems.
>
> 2) Do a series of tests increasing the parameters at a point that the
> properties you are investigating are not changing anymore. We all know that
> for some systems these tests might be prohibitive (this is why the step 1
> is so important). In that case, go for the best grid you can simulate and
> honestly note that on the reports. Just to remember you, if the DOS and
> total energy are converged, your tests for band structure can be only NSCF.
> But please MAKE A BACKUP of your ".save"  folder (if you are using the
> wf_collect=.true.) because the NSCF messes up with the restart.
>
>
> I hope you find at least part of these insights useful.
>
> Best regards,
>
> Filipe
>
>
>
>
>
>
> On Sun, Sep 6, 2015 at 8:31 AM, Yue-Wen Fang 
> wrote:
>
>> Dear all,
>>
>> For a  junction, e.g. Fe/MgO/Fe MTJ with a=b=2.8, c = 34 angstrom, a
>> general DOS  calculation usually require a kgrid as 11*11*1, however, could
>> anyone tell me how many kpoints do I need to set for it in transmission
>> calculation and complex band structure calculation, respectively?
>>
>> Thank you in advance!
>>
>> Bests
>> Fang
>>
>> --
>>
>> 
>> Yue-Wen FANG, PhD candidate
>> Key Laboratory of Polar Materials and Devices, Ministry of Education
>> 
>> East China Normal University  
>>
>>
>>
>>
>> ___
>> Pw_forum mailing list
>> Pw_forum@pwscf.org
>> http://pwscf.org/mailman/listinfo/pw_forum
>>
>
>
>
> --
> _
> Dr. Filipe Camargo Dalmatti Alves Lima
> Pos-Doc at Materials Physics Department.
> University of São Paulo, Physics Institute, Brazil.
> Phones:+55 (11) 3091-6881  (USP)
> +55 (11) 97408-2755 (Cellphone - Vivo)
>
> ___
> Pw_forum mailing list
> Pw_forum@pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>



-- 

Yue-Wen FANG, PhD candidate
Key Laboratory of Polar Materials and Devices, Ministry of Education

East China Normal University  
___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum

Re: [Pw_forum] kpoints for transmission calculations and complex band structures?

[Filipe Camargo Dalmatti Alves Lima]

Dear Fang,

Unfornately I don't have any experience with the systems you are
investigating, but I'll add my comments that might provide you some
insights.

The answer for your question is simple but will require some effort to
accomplish it: you must test your system and make sure the properties are
converged.

Convergence means the following: As you increase a given parameter that you
are tracking, the results are not changing anymore.

In order to achieve that for your systems, I would recommend two things:

1) Look for the literature and check the parameters they used the same
systems.

2) Do a series of tests increasing the parameters at a point that the
properties you are investigating are not changing anymore. We all know that
for some systems these tests might be prohibitive (this is why the step 1
is so important). In that case, go for the best grid you can simulate and
honestly note that on the reports. Just to remember you, if the DOS and
total energy are converged, your tests for band structure can be only NSCF.
But please MAKE A BACKUP of your ".save"  folder (if you are using the
wf_collect=.true.) because the NSCF messes up with the restart.


I hope you find at least part of these insights useful.

Best regards,

Filipe






On Sun, Sep 6, 2015 at 8:31 AM, Yue-Wen Fang  wrote:

> Dear all,
>
> For a  junction, e.g. Fe/MgO/Fe MTJ with a=b=2.8, c = 34 angstrom, a
> general DOS  calculation usually require a kgrid as 11*11*1, however, could
> anyone tell me how many kpoints do I need to set for it in transmission
> calculation and complex band structure calculation, respectively?
>
> Thank you in advance!
>
> Bests
> Fang
>
> --
>
> 
> Yue-Wen FANG, PhD candidate
> Key Laboratory of Polar Materials and Devices, Ministry of Education
> 
> East China Normal University  
>
>
>
>
> ___
> Pw_forum mailing list
> Pw_forum@pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>



-- 
_
Dr. Filipe Camargo Dalmatti Alves Lima
Pos-Doc at Materials Physics Department.
University of São Paulo, Physics Institute, Brazil.
Phones:+55 (11) 3091-6881  (USP)
+55 (11) 97408-2755 (Cellphone - Vivo)
___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum

[Pw_forum] kpoints for transmission calculations and complex band structures?

[Yue-Wen Fang]

Dear all,

For a  junction, e.g. Fe/MgO/Fe MTJ with a=b=2.8, c = 34 angstrom, a
general DOS  calculation usually require a kgrid as 11*11*1, however, could
anyone tell me how many kpoints do I need to set for it in transmission
calculation and complex band structure calculation, respectively?

Thank you in advance!

Bests
Fang

-- 

Yue-Wen FANG, PhD candidate
Key Laboratory of Polar Materials and Devices, Ministry of Education

East China Normal University  
___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum