Dear developers:
In my calculation, the GW correction reduces the bandgap. I’m not sure if this result is reasonable or if it is due to an issue with my input files.
I have attached my GW input file and results. Thank you in advance
GW calculations reduce the band gap
Moderators: Davide Sangalli, andrea.ferretti, myrta gruning, andrea marini, Daniele Varsano
-
- Posts: 27
- Joined: Tue Jun 06, 2023 2:55 am
GW calculations reduce the band gap
You do not have the required permissions to view the files attached to this post.
Jingda Guo
Beijing Institute of Technology
Beijing Institute of Technology
- Daniele Varsano
- Posts: 3838
- Joined: Tue Mar 17, 2009 2:23 pm
- Contact:
Re: GW calculations reduce the band gap
Dear Jingda Guo,
this is indeed rather unusual. I would anyway carefully check the convergences.
Here goes some suggestions.
1. VXCRLvcs possibly should be kept to the maximum number of G vectors (cutoff of the density)
2. EXXRLvcs, check the convergence, this is not time and memory intensive and can be raised
3. Number of bands in X and G, even if you used a terminator technique, check if it is at convergence
4. if your pseudopotential have non-linear core correction, you need to activate them to take into account with the keyword: UseNLCC
5. If your system is a 2D material, you want to use the coulomb cutoff technique in order to speed up convergence with repsect the amount of vacuum needed to isolate the monolyaer.
Not related with the result:
In X and SE parallelization, avoid parallelization over "q" as it could bring unbalances. In general, push as you can the parallelization over bands "c" and "v" in X and "b" in SE.
Best,
Daniele
this is indeed rather unusual. I would anyway carefully check the convergences.
Here goes some suggestions.
1. VXCRLvcs possibly should be kept to the maximum number of G vectors (cutoff of the density)
2. EXXRLvcs, check the convergence, this is not time and memory intensive and can be raised
3. Number of bands in X and G, even if you used a terminator technique, check if it is at convergence
4. if your pseudopotential have non-linear core correction, you need to activate them to take into account with the keyword: UseNLCC
5. If your system is a 2D material, you want to use the coulomb cutoff technique in order to speed up convergence with repsect the amount of vacuum needed to isolate the monolyaer.
Not related with the result:
In X and SE parallelization, avoid parallelization over "q" as it could bring unbalances. In general, push as you can the parallelization over bands "c" and "v" in X and "b" in SE.
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/
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/