Eigenvalue self-consistency problems

[Kostas Fykouras]

Dear group,

I want to perform GW calculation with self consistency using the tag eqp_corrections in the sigma.cplx.x executable. I run to the following problem:

ERROR: Missing bands in file eqp.dat

At the moment I am trying to understand this problem using silicon because the calculations are fast. I first perform a regular sigma calculation with the following sigma.inp:

#######

 number_bands 30
 
 degeneracy_check_override
 
 
 band_index_min 2
 band_index_max 23
 
 
 screening_semiconductor
 
 
 begin kpoints
   0.000000000  0.000000000  0.000000000  1.0
   0.000000000  0.000000000  0.200000000  1.0
   0.000000000  0.000000000  0.400000000  1.0
   0.000000000  0.200000000  0.200000000  1.0
   0.000000000  0.200000000  0.400000000  1.0
   0.000000000  0.200000000  0.600000000  1.0
   0.000000000  0.200000000  0.800000000  1.0
   0.000000000  0.400000000  0.400000000  1.0
   0.000000000  0.400000000  0.600000000  1.0
   0.200000000  0.400000000  0.600000000  1.0
 end

#####

After that, I copy the eqp1.dat to eqp.dat file and I add the additional tag eqp_corrections.

In the first calculation without the eqp.file, I sum over 30 bands and in the eqp1.dat, bands 2 to 23 are printed. Given that when I use eqp_corrections energies will be read from eqp.dat how should I modify the parameters number_bands, band_index_min and band_index_max?

Thanks in advance for your time.

All the best,

Kostas

[Zhenglu Li]

Dear Kostas,

To perform eigenvalue self-consistent calculations, you will need GW energies for all bands in your summation (indicated by number_bands).  The simplest solution in your test now is you need to compute GW corrections for all 1 - 30 bands.

Note that 30 bands will not give converged results.  In practice, hundreds to thousands bands are often needed, and it becomes unrealistic to compute GW energies for all these bands.  Most often, one can do the so-called "scissors" operation, that is to generate GW band energies of the higher/lower states based on the small subset of computed GW band energies.  One can either to rigid shift, or more complicated fitting processes.

Best,

Zhenglu

---

From: Kostas Fykouras <kostasf...@gmail.com>
Sent: Friday, December 1, 2023 7:40 AM
To: BerkeleyGW Help <he...@berkeleygw.org>
Subject: Eigenvalue self-consistency problems

 

--
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/c5aa9008-35d7-4394-9167-a9e599867844n%40berkeleygw.org.

[Kostas Fykouras]

Hi Zhenglu,

thank for your answer. I saw in the documentation that it is also possible to use the eqp_corrections tag in an epsilon calculation. It seems that epsilon requires two files, eqp.dat (eqp1.dat from sigma)

 and eqp_q.dat. How would one generate eqp_q.dat ?

Thanks in advance for your time.

All the best,

Kostas

[Zhenglu Li]

Dear Kostas,

The quickest solution may be using the inteqp.x utility in BerkeleyGW to compute the energies for the shifted grid.

Best,

Zhenglu

---

From: Kostas Fykouras <kostasf...@gmail.com>
Sent: Tuesday, December 5, 2023 12:52 AM
To: BerkeleyGW Help <he...@berkeleygw.org>
Cc: Zhenglu Li <zhen...@usc.edu>
Subject: Re: Eigenvalue self-consistency problems