"Impossible to build the kernel"

[subhayan]

Dear developer,

I am trying to run a diagonalization-solver for the BSE but my calculation stops with the message:

"[WARNING]Bethe Salpter section skipped. Impossible to build the kernel."

I am attaching the output file for the kernel-construction, which seems to finish running without issues (output=r-3D_BSE_optics_bse_dipoles_em1s.txt) and the input and the output files for diagonalization (input=03_3D_QP_BSE_in.txt, output=r-3D_QP_BSE_optics_bse_bss_dipoles_01.txt). Could you please have a look and let me know what you think I might have done wrong? Also, isn't the construction of the BS kernel taking place when when the output "r-3D_BSE_optics_bse_dipoles_em1s.txt" is generated? Why, then, is such a warning message shown in the output "r-3D_QP_BSE_optics_bse_bss_dipoles_01.txt", which is for solving the BSE?

Please let me know if you need any other file.

Best regards,
Subhayan

[Daniele Varsano]

Dear Subhayan,
there is some incompatibility between the database you previously calculated and the input file:
from: r-3D_QP_BSE_optics_bse_bss_dipoles_01.txt

Code: Select all

  BSK|Identifier              :  9770
  *ERR*    |Dimension               :  756
      |Bands                   :   8   60

while the BS dimension from the actual input file is:

[BSK] Size (resonant) 1019

The different dimension is due to the fact that in r-3D_QP_BSE_optics_bse_bss_dipoles_01.txt, after applying the QP correction the system is seen as metallic:

Code: Select all

 [WARNING][K] Metallic system

The discrepancy arise from the fact you have considered the QP correction (KfnQPdb) in one of the input and not in the others. Note that the diagonal part of the BSE is added after the kernel is built.

Check carefully your input file, and in order to avoid conflicts you can add the diagonalization runlevel in the same input file of the kernel construction.

Best,
Daniele

[subhayan]

Hi Daniele,

Thank you for this detailed reply. This explains and solves the issue. Sorry for bombarding you with questions but now my ypp calculation produces lots of negative eigenvalues for excitation. I can notice an error saying "*ERR* X band range limits : 16 16" in the report files for kernel building and diagonalization but I don't know what this means. The attached tar file contains all the input and the report files. I am also attaching the file containing the energy-sorted eigenvalues. Could you kindly take a look and let me know what I might have done wrong? Please let me know if you need anything else.
Even though this is not a thread for ypp, I posted this problem here since this is actually a continuation of the same calculation. If you advise, I can repost it in the right thread.

Best regards,
Subhayan

[Daniele Varsano]

Dear Subhayan,
I cannot see anything wrong, beside a previous BS database left in the 3D_BSE directory not compatible with the input file. It seems to me that Yambo badly calculates the Fermi level after applying QP correction in this spin-polarized case. We need to reproduce your error and see what is going wrong. Can you please post the pseudopotentials you used for the ground state calculation?

Thanks for reporting,
Daniele

[subhayan]

Dear Daniele,

Thanks for looking into this. It would be really helpful if you could take the time to reproduce the error and investigate this further. The attached tar file contains all the input files, including the scf input file and the pseudopotentials (there are three). Please note that the GW input file is named yambo.in.
Please let me know what you think. Once again, thanks a million for doing this.

Best regards,
Subhayan

[Daniele Varsano]

Dear Subhayan,
I've reproduced your error,
I hope to spot the problem shortly.

Take care,
Daniele

[subhayan]

Hi Daniele,

I'll be waiting to hear back from you. Thanks a ton for doing this.

Take care and stay safe,
Subhayan

[Daniele Varsano]

Dear Subhayan,
indeed there were some problems, that we have now fixed (thanks to Andrea Marini!).
The fixing is currently in the developer version of the code, we need to do some extensive testing and next, we will export it in the GPL release.
I hope to come to you with good news shortly.

Best,
Daniele

[subhayan]

Hi Daniele,

This is good news indeed! I shall be eagerly waiting to run my calculations with the revised code.
Many thanks for taking care of this

Best regards,
Subhayan

[Daniele Varsano]

Dear Subhayan,
please find attached the patch for the 4.5 version.
This should solve your problem. If problems persist please let us know.
Thank you for your patience,

Best,
Daniele