This thread is locked.Only browsing is available.
Top Page > Browsing
Problem with unfolding the band structure of 9x9 supercell
Date: 2022/02/08 09:20
Name: Artem Tarasov   <artem.tarasov@spbu.ru>

Hello.

I would like to find out what could be the reason for my problem with the band structure unfolding for a unit cell of 9x9 size. The results of the SCF converging seem to me quite reasonable. However, when the calculation reaches the "Unfolding of bands" block, it freezes for several days without giving a result for any k-point, and after that it crashes completely. At the same time, I do not receive any messages and I have not noticed any problems with RAM overflow either (I use a large amount of 512 Gb). I believe that the problem lies in the search for "the origin of the reference unit cell" for unfolding. However, attempts to set this point manually end with error messages.

Best regards,
Artem Tarasov
メンテ
Page: [1]

Re: Problem with unfolding the band structure of 9x9 supercell ( No.1 )
Date: 2022/02/08 12:22
Name: Naoya Yamaguchi

Hi,

Although I don't know your system, you can try a smaller system such as 2x2 first, and as you increase the supercell size, you can guess what causes your issue.

Regards,
Naoya Yamaguchi
メンテ
Re: Problem with unfolding the band structure of 9x9 supercell ( No.2 )
Date: 2022/02/08 20:54
Name: Artem Tarasov  <artem.tarasov@spbu.ru>

Hello,

I can bring my system here. This is Graphene/Au/Co(0001). I tried to calculate and unfold the band structure of the 2x2 version of this system by turning graphene by 30 degrees relative to the substrate and everything went well. Maybe the problem is somehow related with using a k-grid of one Г point and the "Сluster" calculation mode for the 9x9 cell?

https://drive.google.com/file/d/1tjWE8nvM6D9VWUDAXQR5OFjB6Rn9EYb7/view?usp=sharing

Best regards,
Artem Tarasov
メンテ
Re: Problem with unfolding the band structure of 9x9 supercell ( No.3 )
Date: 2022/02/14 18:38
Name: Naoya Yamaguchi

Dear Artem,

Your issue may be caused by lack of the RAM.
I tried your input, and it was possible to finish when I used a smaller number of the MPI processes.
In the case of too many MPI processes, in the unfolding routine, some processes looked freezing, and in this case, the usage of the RAM was almost full.
I felt that the required amount is much higher than the SCF calculation part.
Therefore, a workaround is to decrease the MPI processes.

Regards,
Naoya Yamaguchi
メンテ

Page: [1]