Page 1 of 1
EMPTY WF index
Posted: Tue Jul 20, 2021 11:34 pm
by burkzdemir
Dear developers,
I am trying to perform GW+BSE calculation, however I am getting the following error. What should I do? Input and report files attached.
[05.04.03] Kernel loop
======================
[ERROR] STOP signal received while in :[05.04.03] Kernel loop
[ERROR]EMPTY WF index @ b:40 k:1 spin:1 ID:0
Re: EMPTY WF index
Posted: Wed Jul 21, 2021 9:15 am
by Daniele Varsano
Dear Burak,
this happens because after the QP correction the system is seen as metallic:
Code: Select all
K States summary : Full Metallic Empty
0001-0037 0038-0043 0044-0200
So you need to include to the BSE at least one occupied and one empty state e.g.
Code: Select all
% BSEBands
37 | 43 | # [BSK] Bands range
%
Anyway check if what you obtained in the GW calculation is reasonable. If not wither you need to calculate QP corrections for more bands, otherwise as it seems
you are interested in including only two bands in the BSE you can include the QP correction using a scissor operator (% KfnQP_E) extracting the parameter from your GW calculation.
Best,
Daniele
Re: EMPTY WF index
Posted: Wed Jul 21, 2021 11:30 pm
by burkzdemir
Dear Daniele,
In the r_setup file (attached), the system seems to be a semiconductor. Moreover, I find about 1 eV band gap in the GW calculation (o.qp) is attached. However, as you say, in the BSE level it is complaining that it is metallic. Also, in the DFT level, system is metallic. What is the problem here?
Burak
Re: EMPTY WF index
Posted: Thu Jul 22, 2021 6:32 am
by Daniele Varsano
Dear Burak,
actually I do not know what is happening here.
I just spotted you are using a quite old version of the code (4.4).
I suggest you to update to the new version of the code, many improvements happened since there:
Code: Select all
git clone git@github.com:yambo-code/yambo.git yambo
repeat all the step, including the p2y conversion.
If the problem persists, post here again attaching the report files and any other info can help to spot the problem.
If there is a problem and we need to dig into the code, it is easier and most useful for us to do that in the newer version.
Best,
Daniele