Absorption errors

[Hong Tang]

Dear BGW experts,

In a GW+BSE run, 

in 08-sigma, I calculated 18 valence and 18 conducting bands of quasiparticle energies. in eqp1.dat 

In 09-kernel, I calculated 20 valence and 20 conducting bands of BSE kernel on the 04-wfn_co q grid.

In 10-absorption, I set "

number_val_bands_coarse 18
number_val_bands_fine 6
number_cond_bands_coarse 18
number_cond_bands_fine 6

" 

But the absorption run got errors"

...

...

...

Assembling spin-singlet kernel.
 Calculating Coulomb potential v(q) for 256 q-points.
 Performing kernel interpolation.
 ERROR: incompatible values found in file "bsemat.h5".
 Quantity: # of val. bands
 Expected: 18
 Got: 20

ERROR: Incompatible values in file "bsemat.h5".


Program received signal SIGABRT: Process abort signal.

Backtrace for this error:
#0  0x7f9606ab5862 in ???
#1  0x7f9606ab48f5 in ???
#2  0x7f96066428ff in ???

...

...

..."

Does this mean I have to re-do 09-kernel with 18 v and 18 c bands? 

Is there any way that I can still use the results of "bsemat.h5" with 20 v ans 20 c bands? is there any setting for this? or how? 

Appreciate your help very much!

Best,

Hong

[Rafael Rodrigues Del Grande]

Hi Hong,

In principle, you would need to rerun the kernel calculation with 18 cond and 18 val bands. Although I do have a script that makes a copy of bsemat.h5 with a smaller number of bands. You can find it here https://github.com/rafaeldgrande/utilities/blob/main/BGW/kernel/create_kernel_subset.py 

Bests,

Rafael

[Hong Tang]

Hi Rafael,

That script worked for running. However, it won't work for the following 10-absorption calculation. The format or some data in the generated file may still have some issues.   Anyway, thanks for your response.

Best,
Hong

[Rafael Rodrigues Del Grande]

Sorry to hear that! What error did you get? I used this script to do the same task you are doing now. 

Best regards,

Rafael

--

Rafael Rodrigues Del Grande

 

Postdoctoral Scholar at University of California, Merced

SAIL group http://sail.ucmerced.edu

https://rafaeldgrande.github.io/ 

[Hong Tang]

Dear Rafael and Zhenglu,

Thanks so much for helping!

With your code generated 18 c and v bands.  kernel  "bsemat.h5" file, the 10-absorption run got the error 

"

...

...

...

 Epsilon matrix for q/=0 read from epsmat.h5:
 - Number of q-points: 89
 - Number of q-points in the full BZ: 168
 - Dielectric cutoff (Ry): 20.00
 - Number of G-vectors for interpolation of epsinv_00(q+G): 1279609
 - Maximum value of imaginary part of epsinv: .000000
 - Dielectric constant: 1.319413
 - Max. |q+G|^2 to interp. epsinv (Ry): 1.20
 - Number of q+G points to interp. epsinv: 3767

 Assembling spin-singlet kernel.
 Calculating Coulomb potential v(q) for 256 q-points.
 Performing kernel interpolation.

HDF5-DIAG: Error detected in HDF5 (1.12.2) MPI-process 20:
  #000: ../../src/H5F.c line 620 in H5Fopen(): unable to open file
    major: File accessibility
    minor: Unable to open file
  #001: ../../src/H5VLcallback.c line 3501 in H5VL_file_open(): failed to iterate over available VOL connector plugins
    major: Virtual Object Layer
    minor: Iteration failed
  #002: ../../src/H5PLpath.c line 578 in H5PL__path_table_iterate(): can't iterate over plugins in plugin path '(null)'
    major: Plugin for dynamically loaded library
    minor: Iteration failed
  #003: ../../src/H5PLpath.c line 620 in H5PL__path_table_iterate_process_path(): can't open directory: /usr/local/hdf5/lib/plugin
    major: Plugin for dynamically loaded library
    minor: Can't open directory or file
  #004: ../../src/H5VLcallback.c line 3351 in H5VL__file_open(): open failed
    major: Virtual Object Layer
    minor: Can't open object
  #005: ../../src/H5VLnative_file.c line 97 in H5VL__native_file_open(): unable to open file
    major: File accessibility

    minor: Unable to open file

...

...

...

". 

anyway, I already re-calculated kernel with 18 c and v bands, and the following 10- calculation is OK now. 

Best,

Hong

On Mon, May 25, 2026 at 3:31 PM Zhenglu Li <zhen...@usc.edu> wrote:

Dear Hong,

Sorry for missing your email earlier.  In this case, I would just suggest redoing absorption or kernel, while making the number of bands parameters consistent.

Best,

Zhenglu

---

From: Hong Tang <tangh...@gmail.com>
Sent: Monday, May 11, 2026 6:15 PM
To: Zhenglu Li <zhen...@usc.edu>
Subject: Fwd: Absorption errors

 

Dear Zhenglu, I am forwarding my question on BerkeleyGW forum, please help me for this question. Thank you very much! Best, Hong ---------- Forwarded message --------- From: Hong Tang <Unknown> Date: Monday, May 11, 2026 at 8: 11: 31 PM

Dear Zhenglu,

I am forwarding my question on BerkeleyGW forum, please help me for this question.

Thank you very much!

Best,

Hong