ERROR: occupations (ifmax) inconsistent between WFN and WFNq files.

202 views
Skip to first unread message

matineh hoseini

unread,
Aug 26, 2024, 3:57:44 PM8/26/24
to BerkeleyGW Help
Dear all,
I received this error while I was running the epsilon.cplx.x : 
"ERROR: occupations (ifmax) inconsistent between WFN and WFNq files.  Remember that you should NOT use WFNq for metals and graphene."
I am doing it for GaN which is a noncentro-symmetric material (without inversion symmetry), so I chose real_or_complex=2 in the pw2bgw.in file and received the WFN and WFNq. but here for running the epsilon, I receive this error. What should I do?
Thank you in advance for your help.

Zhenglu Li

unread,
Aug 26, 2024, 4:33:05 PM8/26/24
to matineh hoseini, BerkeleyGW Help
Dear Matineh,

You might have used smearing for your GaN calculations?  If it is a semiconductor/insulator, you can use fixed occupation scheme in quantum espresso.

Best,
Zhenglu


From: matineh hoseini <matinehh...@gmail.com>
Sent: Monday, August 26, 2024 12:57 PM
To: BerkeleyGW Help <he...@berkeleygw.org>
Subject: ERROR: occupations (ifmax) inconsistent between WFN and WFNq files.
 
--
You received this message because you are subscribed to the Google Groups "BerkeleyGW Help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to help+uns...@berkeleygw.org.
To view this discussion on the web visit https://groups.google.com/a/berkeleygw.org/d/msgid/help/1d35d8df-77a4-4e0e-ac91-13e3884dd5fdn%40berkeleygw.org.

matineh hoseini

unread,
Aug 26, 2024, 5:17:00 PM8/26/24
to BerkeleyGW Help, zhen...@usc.edu
Thank you for your reply. I used " fixed " as the occupation type, but still have the same error. What do you think?

Zhenglu Li

unread,
Aug 26, 2024, 5:20:01 PM8/26/24
to matineh hoseini, BerkeleyGW Help
It could be some incompatibility with version updates.  Could you try the following:  In your WFN and WFNq step, use smearing, but set the broadening to be very small, e.g., 10^-8 Ry.  This should fix your issue.  You can (and probably should) still use fixed occupation for your SCF step.

Best,
Zhenglu


From: matineh hoseini <matinehh...@gmail.com>
Sent: Monday, August 26, 2024 2:17 PM
To: BerkeleyGW Help <he...@berkeleygw.org>
Cc: Zhenglu Li <zhen...@usc.edu>
Subject: Re: ERROR: occupations (ifmax) inconsistent between WFN and WFNq files.
 
Message has been deleted

matineh hoseini

unread,
Aug 30, 2024, 11:22:55 AM8/30/24
to BerkeleyGW Help, zhen...@usc.edu
Dear Zhenglu,
I am facing this error after running inteqp.cplx.x for GaN which is without inversion symmetry:" Missing bands in file eqp_co.dat"
*
here is my inteqp input file:
number_val_bands_coarse 18
number_val_bands_fine 18
number_cond_bands_coarse 6
number_cond_bands_fine 6

use_symmetries_coarse_grid
no_symmetries_fine_grid

* and here is my sigma input file (with 216 k-points)
number_bands 18

band_index_min 1
band_index_max 18

screening_semiconductor

begin kpoints
  0.000000000  0.000000000  0.000000000   1.0
  0.000000000  0.000000000  0.166666667   1.0
  0.000000000  0.000000000  0.333333333   1.0
  0.000000000  0.000000000  0.500000000   1.0
  0.000000000  0.000000000  0.666666667   1.0
  0.000000000  0.000000000  0.833333333   1.0
  0.000000000  0.166666667  0.000000000   1.0
  0.000000000  0.166666667  0.166666667   1.0
  0.000000000  0.166666667  0.333333333   1.0
  0.000000000  0.166666667  0.500000000   1.0
  0.000000000  0.166666667  0.666666667   1.0
  0.000000000  0.166666667  0.833333333   1.0
  0.000000000  0.333333333  0.000000000   1.0
  0.000000000  0.333333333  0.166666667   1.0
  0.000000000  0.333333333  0.333333333   1.0
  0.000000000  0.333333333  0.500000000   1.0
  0.000000000  0.333333333  0.666666667   1.0
  0.000000000  0.333333333  0.833333333   1.0
  0.000000000  0.500000000  0.000000000   1.0
  0.000000000  0.500000000  0.166666667   1.0
  0.000000000  0.500000000  0.333333333   1.0
  0.000000000  0.500000000  0.500000000   1.0
  0.000000000  0.500000000  0.666666667   1.0
  0.000000000  0.500000000  0.833333333   1.0
  0.000000000  0.666666667  0.000000000   1.0
  0.000000000  0.666666667  0.166666667   1.0
  0.000000000  0.666666667  0.333333333   1.0
  0.000000000  0.666666667  0.500000000   1.0
  0.000000000  0.666666667  0.666666667   1.0
  0.000000000  0.666666667  0.833333333   1.0
  0.000000000  0.833333333  0.000000000   1.0
  0.000000000  0.833333333  0.166666667   1.0
  0.000000000  0.833333333  0.333333333   1.0
  0.000000000  0.833333333  0.500000000   1.0
  0.000000000  0.833333333  0.666666667   1.0
  0.000000000  0.833333333  0.833333333   1.0
  0.166666667  0.000000000  0.000000000   1.0
  0.166666667  0.000000000  0.166666667   1.0
  0.166666667  0.000000000  0.333333333   1.0
  0.166666667  0.000000000  0.500000000   1.0
  0.166666667  0.000000000  0.666666667   1.0
  0.166666667  0.000000000  0.833333333   1.0
  0.166666667  0.166666667  0.000000000   1.0
  0.166666667  0.166666667  0.166666667   1.0
  0.166666667  0.166666667  0.333333333   1.0
  0.166666667  0.166666667  0.500000000   1.0
  0.166666667  0.166666667  0.666666667   1.0
  0.166666667  0.166666667  0.833333333   1.0
  0.166666667  0.333333333  0.000000000   1.0
  0.166666667  0.333333333  0.166666667   1.0
  0.166666667  0.333333333  0.333333333   1.0
  0.166666667  0.333333333  0.500000000   1.0
  0.166666667  0.333333333  0.666666667   1.0
  0.166666667  0.333333333  0.833333333   1.0
  0.166666667  0.500000000  0.000000000   1.0
  0.166666667  0.500000000  0.166666667   1.0
  0.166666667  0.500000000  0.333333333   1.0
  0.166666667  0.500000000  0.500000000   1.0
  0.166666667  0.500000000  0.666666667   1.0
  0.166666667  0.500000000  0.833333333   1.0
  0.166666667  0.666666667  0.000000000   1.0
  0.166666667  0.666666667  0.166666667   1.0
  0.166666667  0.666666667  0.333333333   1.0
  0.166666667  0.666666667  0.500000000   1.0
  0.166666667  0.666666667  0.666666667   1.0
  0.166666667  0.666666667  0.833333333   1.0
  0.166666667  0.833333333  0.000000000   1.0
  0.166666667  0.833333333  0.166666667   1.0
  0.166666667  0.833333333  0.333333333   1.0
  0.166666667  0.833333333  0.500000000   1.0
  0.166666667  0.833333333  0.666666667   1.0
  0.166666667  0.833333333  0.833333333   1.0
  0.333333333  0.000000000  0.000000000   1.0
  0.333333333  0.000000000  0.166666667   1.0
  0.333333333  0.000000000  0.333333333   1.0
  0.333333333  0.000000000  0.500000000   1.0
  0.333333333  0.000000000  0.666666667   1.0
  0.333333333  0.000000000  0.833333333   1.0
  0.333333333  0.166666667  0.000000000   1.0
  0.333333333  0.166666667  0.166666667   1.0
  0.333333333  0.166666667  0.333333333   1.0
  0.333333333  0.166666667  0.500000000   1.0
  0.333333333  0.166666667  0.666666667   1.0
  0.333333333  0.166666667  0.833333333   1.0
  0.333333333  0.333333333  0.000000000   1.0
  0.333333333  0.333333333  0.166666667   1.0
  0.333333333  0.333333333  0.333333333   1.0
  0.333333333  0.333333333  0.500000000   1.0
  0.333333333  0.333333333  0.666666667   1.0
  0.333333333  0.333333333  0.833333333   1.0
  0.333333333  0.500000000  0.000000000   1.0
  0.333333333  0.500000000  0.166666667   1.0
  0.333333333  0.500000000  0.333333333   1.0
  0.333333333  0.500000000  0.500000000   1.0
  0.333333333  0.500000000  0.666666667   1.0
  0.333333333  0.500000000  0.833333333   1.0
  0.333333333  0.666666667  0.000000000   1.0
  0.333333333  0.666666667  0.166666667   1.0
  0.333333333  0.666666667  0.333333333   1.0
  0.333333333  0.666666667  0.500000000   1.0
  0.333333333  0.666666667  0.666666667   1.0
  0.333333333  0.666666667  0.833333333   1.0
  0.333333333  0.833333333  0.000000000   1.0
  0.333333333  0.833333333  0.166666667   1.0
  0.333333333  0.833333333  0.333333333   1.0
  0.333333333  0.833333333  0.500000000   1.0
  0.333333333  0.833333333  0.666666667   1.0
  0.333333333  0.833333333  0.833333333   1.0
  0.500000000  0.000000000  0.000000000   1.0
  0.500000000  0.000000000  0.166666667   1.0
  0.500000000  0.000000000  0.333333333   1.0
  0.500000000  0.000000000  0.500000000   1.0
  0.500000000  0.000000000  0.666666667   1.0
  0.500000000  0.000000000  0.833333333   1.0
  0.500000000  0.166666667  0.000000000   1.0
  0.500000000  0.166666667  0.166666667   1.0
  0.500000000  0.166666667  0.333333333   1.0
  0.500000000  0.166666667  0.500000000   1.0
  0.500000000  0.166666667  0.666666667   1.0
  0.500000000  0.166666667  0.833333333   1.0
  0.500000000  0.333333333  0.000000000   1.0
  0.500000000  0.333333333  0.166666667   1.0
  0.500000000  0.333333333  0.333333333   1.0
  0.500000000  0.333333333  0.500000000   1.0
  0.500000000  0.333333333  0.666666667   1.0
  0.500000000  0.333333333  0.833333333   1.0
  0.500000000  0.500000000  0.000000000   1.0
  0.500000000  0.500000000  0.166666667   1.0
  0.500000000  0.500000000  0.333333333   1.0
  0.500000000  0.500000000  0.500000000   1.0
  0.500000000  0.500000000  0.666666667   1.0
  0.500000000  0.500000000  0.833333333   1.0
  0.500000000  0.666666667  0.000000000   1.0
  0.500000000  0.666666667  0.166666667   1.0
  0.500000000  0.666666667  0.333333333   1.0
  0.500000000  0.666666667  0.500000000   1.0
  0.500000000  0.666666667  0.666666667   1.0
  0.500000000  0.666666667  0.833333333   1.0
  0.500000000  0.833333333  0.000000000   1.0
  0.500000000  0.833333333  0.166666667   1.0
  0.500000000  0.833333333  0.333333333   1.0
  0.500000000  0.833333333  0.500000000   1.0
  0.500000000  0.833333333  0.666666667   1.0
  0.500000000  0.833333333  0.833333333   1.0
  0.666666667  0.000000000  0.000000000   1.0
  0.666666667  0.000000000  0.166666667   1.0
  0.666666667  0.000000000  0.333333333   1.0
  0.666666667  0.000000000  0.500000000   1.0
  0.666666667  0.000000000  0.666666667   1.0
  0.666666667  0.000000000  0.833333333   1.0
  0.666666667  0.166666667  0.000000000   1.0
  0.666666667  0.166666667  0.166666667   1.0
  0.666666667  0.166666667  0.333333333   1.0
  0.666666667  0.166666667  0.500000000   1.0
  0.666666667  0.166666667  0.666666667   1.0
  0.666666667  0.166666667  0.833333333   1.0
  0.666666667  0.333333333  0.000000000   1.0
  0.666666667  0.333333333  0.166666667   1.0
  0.666666667  0.333333333  0.333333333   1.0
  0.666666667  0.333333333  0.500000000   1.0
  0.666666667  0.333333333  0.666666667   1.0
  0.666666667  0.333333333  0.833333333   1.0
  0.666666667  0.500000000  0.000000000   1.0
  0.666666667  0.500000000  0.166666667   1.0
  0.666666667  0.500000000  0.333333333   1.0
  0.666666667  0.500000000  0.500000000   1.0
  0.666666667  0.500000000  0.666666667   1.0
  0.666666667  0.500000000  0.833333333   1.0
  0.666666667  0.666666667  0.000000000   1.0
  0.666666667  0.666666667  0.166666667   1.0
  0.666666667  0.666666667  0.333333333   1.0
  0.666666667  0.666666667  0.500000000   1.0
  0.666666667  0.666666667  0.666666667   1.0
  0.666666667  0.666666667  0.833333333   1.0
  0.666666667  0.833333333  0.000000000   1.0
  0.666666667  0.833333333  0.166666667   1.0
  0.666666667  0.833333333  0.333333333   1.0
  0.666666667  0.833333333  0.500000000   1.0
  0.666666667  0.833333333  0.666666667   1.0
  0.666666667  0.833333333  0.833333333   1.0
  0.833333333  0.000000000  0.000000000   1.0
  0.833333333  0.000000000  0.166666667   1.0
  0.833333333  0.000000000  0.333333333   1.0
  0.833333333  0.000000000  0.500000000   1.0
  0.833333333  0.000000000  0.666666667   1.0
  0.833333333  0.000000000  0.833333333   1.0
  0.833333333  0.166666667  0.000000000   1.0
  0.833333333  0.166666667  0.166666667   1.0
  0.833333333  0.166666667  0.333333333   1.0
  0.833333333  0.166666667  0.500000000   1.0
  0.833333333  0.166666667  0.666666667   1.0
  0.833333333  0.166666667  0.833333333   1.0
  0.833333333  0.333333333  0.000000000   1.0
  0.833333333  0.333333333  0.166666667   1.0
  0.833333333  0.333333333  0.333333333   1.0
  0.833333333  0.333333333  0.500000000   1.0
  0.833333333  0.333333333  0.666666667   1.0
  0.833333333  0.333333333  0.833333333   1.0
  0.833333333  0.500000000  0.000000000   1.0
  0.833333333  0.500000000  0.166666667   1.0
  0.833333333  0.500000000  0.333333333   1.0
  0.833333333  0.500000000  0.500000000   1.0
  0.833333333  0.500000000  0.666666667   1.0
  0.833333333  0.500000000  0.833333333   1.0
  0.833333333  0.666666667  0.000000000   1.0
  0.833333333  0.666666667  0.166666667   1.0
  0.833333333  0.666666667  0.333333333   1.0
  0.833333333  0.666666667  0.500000000   1.0
  0.833333333  0.666666667  0.666666667   1.0
  0.833333333  0.666666667  0.833333333   1.0
  0.833333333  0.833333333  0.000000000   1.0
  0.833333333  0.833333333  0.166666667   1.0
  0.833333333  0.833333333  0.333333333   1.0
  0.833333333  0.833333333  0.500000000   1.0
  0.833333333  0.833333333  0.666666667   1.0
  0.833333333  0.833333333  0.833333333   1.0
end
*
I chose 18 and 6 in the input files after checking the degeneracy from "degeneracy_check.x WFN". I don't know why I received this error. I have made sure that eqp_co.dat have been  linked to eqp1.dat (which is the output of sigma). Any hints will be appreciated.
Thank you,
Matineh

Zhenglu Li

unread,
Aug 30, 2024, 11:39:16 AM8/30/24
to matineh hoseini, BerkeleyGW Help
Dear Matineh,

You could check in your Sigma step calculation (output eqp_co.dat), if you have included more than 18 bands.

Best,
Zhenglu


From: matineh hoseini <matinehh...@gmail.com>
Sent: Friday, August 30, 2024 8:22 AM

To: BerkeleyGW Help <he...@berkeleygw.org>
Cc: Zhenglu Li <zhen...@usc.edu>
Subject: Re: ERROR: Missing bands in file eqp_co.dat.
 

matineh hoseini

unread,
Aug 30, 2024, 11:59:01 AM8/30/24
to BerkeleyGW Help, zhen...@usc.edu, matineh hoseini
Thank you for your reply. Output file (eqp_co.dat) has been calculated for 18 bands in each kpoint. I checked several times and couldn't find any mismatch. This is my eqp_co.dat file for your reference:
*
  0.000000000  0.000000000  0.000000000      18
       1       1   -5.761305007  -15.554492050
       1       2   -5.040694172  -12.763906778
       1       3   -3.769771762   -8.512890530
       1       4   -3.761613416   -8.491882032
       1       5   -3.761612976   -8.491881591
       1       6   -3.708486576   -8.375496289
       1       7   -3.708486206   -8.375495919
       1       8   -3.535654658   -7.976425206
       1       9   -3.535654432   -7.976424979
       1      10   -3.444498945   -7.850028724
       1      11   -3.444498704   -7.850028483
       1      12   -2.013826793   -8.254305511
       1      13    2.899443509   -2.645695628
       1      14    8.832949276    5.252393407
       1      15    8.832951664    5.252395795
       1      16    9.806433279    6.900339214
       1      17    9.839440611    6.869682860
       1      18    9.839442223    6.869684473
  0.000000000  0.000000000  0.166666667      18
       1       1   -5.685689300  -15.415615391
       1       2   -5.056489367  -12.920734039
       1       3   -3.760477578   -8.487614963
......
*
I am looking forward to hearing back from you.
Thank you.
Matineh

Zhenglu Li

unread,
Aug 30, 2024, 6:55:38 PM8/30/24
to matineh hoseini, BerkeleyGW Help
Dear Matineh,

Sorry I had a typo in my previous email.  In the input, you asked for 18 valence bands, and 6 conduction bands.  So you need to have all of them in your eqp.dat, which is at least 18 + 6 = 24 bands (if your band #1 is the bottom of the valence band).

Best,
Zhenglu


From: matineh hoseini <matinehh...@gmail.com>
Sent: Friday, August 30, 2024 8:59 AM
To: BerkeleyGW Help <he...@berkeleygw.org>
Cc: Zhenglu Li <zhen...@usc.edu>; matineh hoseini <matinehh...@gmail.com>

Zhenglu Li

unread,
Sep 2, 2024, 12:45:08 AM9/2/24
to matineh hoseini, BerkeleyGW Help
Dear Matineh,

If so, it looks like something else is wrong that triggered the error.  You may compare step by step with a working semiconductor example.

For GaAs, yes, it has a broken inversion symmetry, and one has to use complex version of BerkleyGW.  For DFT, the wavefunctions typically are always complex (except for Gamma-only version, sometimes), and the complex/real flavor is turned on only for BerkeleyGW, hence starting at the pw2bgw step.

Best,
Zhenglu


From: matineh hoseini <matinehh...@gmail.com>
Sent: Wednesday, August 28, 2024 8:23 AM

To: BerkeleyGW Help <he...@berkeleygw.org>
Cc: Zhenglu Li <zhen...@usc.edu>
Subject: Re: ERROR: occupations (ifmax) inconsistent between WFN and WFNq files.
 
Dear Zhenglu,
I tried the way that you mentioned, but it didn't work and I faced the same error in each case: first, I changed the occupation to 'fixed' in scf, WFN, WFNq folders. second, I kept the scf occupation as 'fixed' and changed the WFN and WFNq occupation to smearing with degauss=1.0d-8, and lastly, I changed the scf also to smearing and again ran, but it didn't work. 
I was also looking into the GaAs example in github, and realized that for this material which is also without inversion symmetry you used "real_or_complex=2" in pw2bgw input file, but didn't mention the occupation in scf, WFN and WFNq input files and it seems that you received the output. What is the reason for that? 
I don't know why exactly I am facing this error while I am running the epsilon.in file. I am thinking to just run it with "real_or_complex=1" and see what will happen. I also attached the folder for your reference. I will appreciate if you could take a look and let me know your thoughts.
Thank you in advance for your time and help.
Matineh

On Monday, August 26, 2024 at 5:20:01 PM UTC-4 zhen...@usc.edu wrote:

matineh hoseini

unread,
Sep 3, 2024, 7:09:13 AM9/3/24
to BerkeleyGW Help, zhen...@usc.edu, matineh hoseini
Dear Zhenglu,
In sigma.in file, when I define the # of bands as 24 it shows me an error regarding the degeneracy and when I checked the degeneracy with "degeneracy_check.x WFN_inner" the maximum allowable # of bands is 18 and the number of valence bands to be used are 6,12,18 which I chose 16 for the number of valence bands and 2 for the number of conduction bands in inteqp.in file this time. But again, I received the same error: "missing bands in file eqp_co.dat". I don't know why. now I have 16 # VB and 2 # CB and total bands number is 18 :(16+2=18) so what should be the problem? I am thinking maybe it is because of the WFN_co and WFN_fi that I am using, because I found them by considering the number of bands=24. Do you think if I again run them with 18 # total bands will it be corrected? Or do you think solving again by considering smearing will solve the problem? 
here is the input of sigma:
number_bands 18
band_index_min 1
band_index_max 18
screening_semiconductor
begin kpoints
  0.000000000  0.000000000  0.000000000   1.0
  0.000000000  0.000000000  0.166666667   1.0
.......
*
here is the input of inteqp:
number_val_bands_coarse 16
number_val_bands_fine 16
number_cond_bands_coarse 2
number_cond_bands_fine 2
use_symmetries_coarse_grid
no_symmetries_fine_grid
*
here is the input of epsilon:
epsilon_cutoff 40.0
number_bands 18
begin qpoints
  0.000000000  0.000000000  0.001000000   1.0 1
  0.000000000  0.000000000  0.166666667   1.0 0
  0.000000000  0.000000000  0.333333333   1.0 0
.......
*
here is the input of bands:
&control
   prefix = 'GaN'
   calculation = 'bands'
   restart_mode = 'from_scratch'
   wf_collect = .true.
   tstress = .false.
   tprnfor = .false.
   outdir = './'
   wfcdir = './'
   pseudo_dir = './'
   verbosity = 'high'
/
&system
   ibrav = 4, celldm(1)=6.0779067998, celldm(3)=1.6291938848,
   nat = 4
   ntyp = 2
   ecutwfc = 80
   nbnd = 24
   occupations='fixed'
   degauss = 0.D0
   smearing = 'gaussian'
/
&electrons
   electron_maxstep = 100
   conv_thr = 1.0d-10
   mixing_mode = 'plain'
   mixing_beta = 0.7
   mixing_ndim = 8
   diagonalization = 'david'
   diago_david_ndim = 4
   diago_full_acc = .true.
/
ATOMIC_SPECIES
Ga  69.723  Ga.upf
N   14.01   N.upf
ATOMIC_POSITIONS crystal
Ga            0.3333333430        0.6666666870        0.9989902533
Ga            0.6666666270        0.3333333130        0.4989902533
N             0.3333333430        0.6666666870        0.3760097467
N             0.6666666270        0.3333333130        0.8760097467
K_POINTS crystal
  216

  0.000000000  0.000000000  0.000000000   1.0
  0.000000000  0.000000000  0.166666667   1.0
............
*
Please let me know your thoughts. I am looking forward hearing back from you.
Thank you,
Matineh

Zhenglu Li

unread,
Sep 3, 2024, 12:53:57 PM9/3/24
to matineh hoseini, BerkeleyGW Help
Dear Matineh,

If you only have a total of 18 bands in Sigma, then you will be unable to go beyond that in BSE.  The solution would be you include more states for eqp.dat via band_index_min and band_index_max.

I see another issue in your calculations.  The number_bands for both epsilon and sigma steps (used for internal summation, independent of band_index_min and band_index_max) seems small for GaN, which is a semiconductor.  A typical value is few hundreds to few thousands, upon a reasonable convergence test.  One of the reference for the convergence of GW calculations is included in the Workshop resources: https://drive.google.com/file/d/1XoNeCk0tK1XOeh13jMGWbMmIAJl8NZVz/view

Best,
Zhenglu


From: matineh hoseini <matinehh...@gmail.com>
Sent: Tuesday, September 3, 2024 4:09 AM
Message has been deleted

matineh hoseini

unread,
Sep 18, 2024, 10:08:23 AM9/18/24
to BerkeleyGW Help, zhen...@usc.edu, matineh hoseini
Dear Zhenglu,
Thanks to your help, I calculated the band diagram of GaN, and currently I am testing the convergence with respect to the number of bands in epsilon & sigma and the screened cut off. I have done testing the convergence with respect to the number of bands used in CH summation and to the screened cutoff of the dielectric matrix. I attached the resulted pictures for you to make sure that my calculation is correct. I am not sure about the result that I found after testing the convergence with respect to the screened cutoff, because at first it has an increasing behavior and then decreasing, I am wondering what the reason for that could be, is it normal or is it a sign of having an error. Due to the memory issue of my cluster, I've done it on a 4*4*4 kgrid and 100 bands. 
Any opinion and thoughts will be greatly appreciated.
I am looking forward to hearing back from you.
Thank you,
Matineh
CH_bands_conv.pngScreened_cut_conv.pngbandstructure.png

Zhenglu Li

unread,
Sep 19, 2024, 12:47:53 PM9/19/24
to matineh hoseini, BerkeleyGW Help
Dear Matineh,

The strange behavior you observed may come from a mismatch in convergence of bands and epsilon cutoff, which are interlocked and dependent convergence.  You may check this set of tutorial slides https://drive.google.com/file/d/1_WeFMA5F5n6bCtTCCYhxmv7-wGkmEwaG/view and the references therein.  In general, your epsilon cutoff for GaN should fall in 20 - 40 Ry range, and the number of bands certainly needs to go much higher, possibly somewhere 500 - 5000, which need to be tested.  The k/q grid might need to be tested as well.

Another excellent convergence reference is Brad D Malone and Marvin L Cohen 2013 J. Phys.: Condens. Matter 25 105503.  Hope these help!

Best,
Zhenglu


From: matineh hoseini <matinehh...@gmail.com>
Sent: Wednesday, September 18, 2024 7:08 AM

matineh hoseini

unread,
Sep 19, 2024, 4:01:13 PM9/19/24
to BerkeleyGW Help, zhen...@usc.edu, matineh hoseini
Thank you for your reply. I will try to increase the number of bands to see what will happen.
Also, I have a question about the eqp0.dat and eqp1.dat, which are the outputs of sigma step. I looked into these outputs and compared them with sigma.out  and couldn't figure out what is the third column in eqp0.dat and eqp1.dat file related to? I found out that the last column in these files is shown in sigma.out , but what about the third column? what is it telling us?
Another question is that GW will not give us any information about the frequency? I am asking because after finding the matrix element with EPW, it will give us both the omega and matrix element in each point, but here I couldn't find the frequency of the points.
I attached the picture of eqp0.dat and sigma.out for your reference.
Thank you,
Matineh

1.1.jpg
1.2.jpg

matineh hoseini

unread,
Nov 19, 2024, 9:10:58 AM11/19/24
to BerkeleyGW Help, matineh hoseini, zhen...@usc.edu
Dear Zhenglu,
Do you know what might be the possible reason that I couldn't get the correct energy gap of GaN after doing the calculation? All parts of GW are working well, and I have tested different number of bands in the range of 500-5000 and varied the kgrid, but couldn't find the exact value of the energy gap of GaN and the best value that I have received till now is almost 3.2eV. I have used the LDA pseudopotential files from Pseudo Dojo. I will appreciate if you could let me know your thoughts.
Thank you,
Matineh

matineh hoseini

unread,
Dec 19, 2024, 2:26:22 PM12/19/24
to BerkeleyGW Help, matineh hoseini, zhen...@usc.edu
Dear Zhenglu,
I am sending this email as a kind reminder. I'll appreciate any advice or help from you.
Thanks,
Matineh

Zhenglu Li

unread,
Dec 19, 2024, 2:31:04 PM12/19/24
to matineh hoseini, BerkeleyGW Help
Dear Matineh,

It seems that the experimental gap of GaN is 3.4 eV.  Your value of 3.2 eV is pretty good.  Typically, the intrinsic error within the GW approximation is around 0.1 eV, so your results are pretty reasonable already.  If you want to improve further, you may consider further increase your convergence, e.g., bands, k/q point mesh density.  Some self-consistency in G, e.g. at the eigenvalue level, might help as well.  But overall, I think 3.2 vs 3.4 is already decent.

Best,
Zhenglu


From: matineh hoseini <matinehh...@gmail.com>
Sent: Thursday, December 19, 2024 11:26 AM
To: BerkeleyGW Help <he...@berkeleygw.org>
Cc: matineh hoseini <matinehh...@gmail.com>; Zhenglu Li <zhen...@usc.edu>

Subject: Re: ERROR: Missing bands in file eqp_co.dat.
Reply all
Reply to author
Forward
0 new messages