Accelerated GW

Concerns issues with computing quasiparticle corrections to the DFT eigenvalues - i.e., the self-energy within the GW approximation (-g n), or considering the Hartree-Fock exchange only (-x)

Moderators: Davide Sangalli, andrea.ferretti, myrta gruning, andrea marini, Daniele Varsano

Post Reply
rishabhsrsw7
Posts: 10
Joined: Thu Sep 15, 2022 6:17 am

Accelerated GW

Post by rishabhsrsw7 » Thu Feb 01, 2024 5:21 pm

Hi,

I have been trying to understand the accelerated GW for 2D h-BN. I have followed the GW tutorial, but I have this doubt.

As mentioned in the tutorial, the calculated GW gap at 6x6x1, performing the W-av method, should be a little less than 6.5 eV. https://www.yambo-code.eu/wiki/images/d ... w_conv.png . I am unable to converge to this value, my calculations always end up at 6.7 eV. I want to know why that is and what changes can I do to make it converge it to 6.5. I am attaching the input file here which I am using.
gw.txt
Regards
You do not have the required permissions to view the files attached to this post.
Rishabh Saraswat
PhD student
Indian Institute of Information Technology-Allahabad
India
Institute: http://www.iiita.ac.in/

Online
User avatar
Daniele Varsano
Posts: 3975
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Accelerated GW

Post by Daniele Varsano » Thu Feb 01, 2024 5:53 pm

Dear Rishabh,

besides the k point convergence, other parameters need to be converged, as the energy cutoff for the screening and the summation of empty bands. As you can see in the original publication from where the figure is extracted:
Guandalini et al. npj Computational Materials volume 9, Article number: 44 (2023)
https://www.nature.com/articles/s41524-023-00989-7

The calculations of that plot are not converged wrt bands and cutoff as the figure is meant to compare to "standard" methods that need very large grids. You can have a look to the used parameters and the converged one in the sections of the computational details. Note that the converged value reported in Tab. 1 is 6.82 eV.

Having said that, looking at your input, you set:

Code: Select all

rimw_type= "graphene"
What's the reason for that? This is meant for systems presenting a Dirac cone, while you are studying hBN so the keyword "semiconductor" need to be used.

Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

rishabhsrsw7
Posts: 10
Joined: Thu Sep 15, 2022 6:17 am

Re: Accelerated GW

Post by rishabhsrsw7 » Fri Feb 02, 2024 7:22 am

Greetings Daniele,

Thank you for your response. Regarding the rimw_type, I assumed that hBN is a 2D material, so I had to use graphene. Thank you for pointing that out too.

Regards
Rishabh Saraswat
PhD student
Indian Institute of Information Technology-Allahabad
India
Institute: http://www.iiita.ac.in/

rishabhsrsw7
Posts: 10
Joined: Thu Sep 15, 2022 6:17 am

Re: Accelerated GW

Post by rishabhsrsw7 » Tue Apr 09, 2024 12:57 pm

Daniele Varsano wrote: Thu Feb 01, 2024 5:53 pm Dear Rishabh,

besides the k point convergence, other parameters need to be converged, as the energy cutoff for the screening and the summation of empty bands. As you can see in the original publication from where the figure is extracted:
Guandalini et al. npj Computational Materials volume 9, Article number: 44 (2023)
https://www.nature.com/articles/s41524-023-00989-7

The calculations of that plot are not converged wrt bands and cutoff as the figure is meant to compare to "standard" methods that need very large grids. You can have a look to the used parameters and the converged one in the sections of the computational details. Note that the converged value reported in Tab. 1 is 6.82 eV.

Having said that, looking at your input, you set:

Code: Select all

rimw_type= "graphene"
What's the reason for that? This is meant for systems presenting a Dirac cone, while you are studying hBN so the keyword "semiconductor" need to be used.

Best,
Daniele
Hi Daniele,
I have run into an issue while trying out the accelerating of GW in 2D systems. I am trying to do it for monolayer MoS2. When I run this input at NGsBlkXp = 3 Ry, it gives me an output (although that doesn't capture the spin orbit splitting), but if I tweak this value of response block size to ANY other value, I get this error:
P60-cn068: [ERROR] STOP signal received while in[05] Coloumb potential CutOffslab
P60-cn068: [ERROR] Reading File ./output//ndb.cutoff; Variable NOT DEFINED; NetCDF: HDF error


I am attaching the input file and the r_output file.
gw.in.txt
r_output.txt
Will be looking forward for your reply.
Thank you.
You do not have the required permissions to view the files attached to this post.
Rishabh Saraswat
PhD student
Indian Institute of Information Technology-Allahabad
India
Institute: http://www.iiita.ac.in/

Online
User avatar
Daniele Varsano
Posts: 3975
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Accelerated GW

Post by Daniele Varsano » Tue Apr 09, 2024 1:39 pm

Dear Rishabh,

It seems you have a previous ndb.cutoff stored in your directory. Try to remove it and rerun the calculation.
Please note rimw_type is "semiconductor" and not "semiconductors".

Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

rishabhsrsw7
Posts: 10
Joined: Thu Sep 15, 2022 6:17 am

Re: Accelerated GW

Post by rishabhsrsw7 » Tue Apr 09, 2024 3:43 pm

Thank you for your reply Daniele,

I did rectify that typo of "semiconductors", but I have been running these gw calculations from the scratch, the output directory is generated fresh each time, I do not use the output directory of any previously run calculation, and I still get the same error again and again.

Thank you.
Rishabh Saraswat
PhD student
Indian Institute of Information Technology-Allahabad
India
Institute: http://www.iiita.ac.in/

rishabhsrsw7
Posts: 10
Joined: Thu Sep 15, 2022 6:17 am

Re: Accelerated GW

Post by rishabhsrsw7 » Wed Apr 10, 2024 10:59 am

Daniele Varsano wrote: Tue Apr 09, 2024 1:39 pm Dear Rishabh,

It seems you have a previous ndb.cutoff stored in your directory. Try to remove it and rerun the calculation.
Please note rimw_type is "semiconductor" and not "semiconductors".

Best,
Daniele
Hi Daniele, I was able to rectify the previous error, but one doubt of mine still remains, normal gw calculations capture the spin orbit splitting of MoS2, but when I employ the RIM, my GW bandstructure does not capture the spin orbit splitting. Attaching the files here.
gw.in.txt
r_output.txt
Will be looking forward for your response.
Thank you
You do not have the required permissions to view the files attached to this post.
Rishabh Saraswat
PhD student
Indian Institute of Information Technology-Allahabad
India
Institute: http://www.iiita.ac.in/

Online
User avatar
Daniele Varsano
Posts: 3975
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Accelerated GW

Post by Daniele Varsano » Thu Apr 11, 2024 7:40 am

Dear Rishabh,

I think I'm missing something: One of the SO effect in MoS2 is the splitting of the valence band around the K point. In your input/output I can see, you are calculating the GW correction for the last valence band only, so how can you see such a splitting? But probably you are referring to something else, so please elaborate your problem with more details.

Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

Post Reply