ABINIT Discussion Forums

Hi all,

I'm trying to run Bethe-Salpeter calculations of the macroscopic dielectric function with version 6.2.1. But I'm confused about the following:

My first goal, of course, is to get the Bethe-Salpeter macroscopic dielectric function without having to include any scissors operator shift by hand. It seems that for that one has to use bs_calctype=3. Is this correct? Note that the description of calctype in the GW variables web page only mentions calctype 0, 1, and 2, but this is probably just a typo error. Doing a run with bs_calctyp=0 crashes the program.

The value bs_calctype=3 is used in t91.in. But then the result depends on the value of soenergy. If one uses soenergy=0, the program crashes! If one uses a non-zero value for soenergy, the dielectric function (imaginary part) has a blue shift corresponding to the value of soenergy. But if bs_calctype=3 corresponds to the Bethe-Salpeter calculation using the GW eigenvalues, one shouldn't need to give any soenergy value, right?

Also, the description of the bs_calctyp variable indicates it is a two digit variable. But the default value is said to be calctyp=1, and t91.in uses only one digit too (?)

Any clarification on the above will be greatly appreciated.

Best regards,

Rolando Saniz
Departement Fysica
Universiteit Antwerpen

Hi All,
I also have the similar question as Saniz mentioned. I'm using abinit-6.10.3 and performing some tests on BSE.

I note that if I set bs_calctype=2 to use the _GW file from one-shot GW calculation as the Input and set soenergy=0 eV, the output spectrum has a red shift compared with the case using bs_calctype=1. if setting a non-zero soenergy, the result turn to be OK. So it seems as if the value of bs_calctype doesn't make any difference. How can I set these variables to work properly?

Thanks in advance!

Hi all,

It seems that I made some mistake in my previous calculation. When running BSE with bs_calctype=2, the QP energies are obtained from a __in.gw__ file and the k-points in BSE and GW calculation must be consistent, otherwise the QP energies can not be used correctly.

Agnxy wrote:Hi all,

It seems that I made some mistake in my previous calculation. When running BSE with bs_calctype=2, the QP energies are obtained from a __in.gw__ file and the k-points in BSE and GW calculation must be consistent, otherwise the QP energies can not be used correctly.

Concernig the bs_calctype=2 and qps energies,
"__in.gw__ " Is this file created by optdriver=4 with gwcalctype=10? It is a little bit confusing, because I can't get such file after GW calculation!
I would apprreciate, If someone could help.

Hello,

This __in.gw__ is the _GW file you get after any GW calculation (optdriver 4). Just rename the _GW file to __in.gw__ and your BSE calculation should run.

David

Dear All
Refer to Agnzy, when I used kpoints in BSE consist with GW, I got the energies using bs_calctype=2 with _GW file from one-shot shifted down than bs_calctype=1 with soenergy. I don't know how to do it. I test with CuBO_2.
Could any one help me ?

Abdulmutta Thatribud
Department of Physics, Prince of Songkla University,
Thailand

Dear All,
I'm using abinit-6.12 and performing some tests on BSE in CuAlO2.
If I set bs_calctype=2 to use the _GW file and rename __in.gw__ from G0W0 calculation as the Input and set soenergy=0 eV, the output spectrum has a same shift compared with the case using bs_calctype=1 and also BSE spectrum has non-shift corrected with __in.gw__ energies. When we set a soenergy with non-zero, the result is OK. So it seems as if the value of bs_calctype doesn't make any difference. How can I set these variables to work properly?

Thank you
Mutta