My Community

Dear friends:
When calculating BSE, I have some doubts about the identification of occupied bands，I konw we could set the NBANDSO and the NBANDSV to control the bands which included in the BSE calculation.However,I encountered some problems when calculating GW+BSE with vasp5.4.4.
I calculated GW+SOC, and found that the highest band occupied is the 30th band, with the number of occupancy being 1, so the 30th band should be the VBM(valence band max), and the 31st band should be CBM(conduct band min). However, when I calculated BSE, I considered the situation of NBADSO=4 ,NBADSV=4 and NBADSO=2 ,NBADSV=2 respectively，then VB (max)=31 and CB (min)=30 were found in OUTCAR of BSE，which is so strange.
Because my system considers SOC and has a very small gap of about tens of mev, I suspect that there is an error in recognition
Please give me some help.
Thanks
Quxiao

Dear Quxiao,

Please include if possible the files you used during this run (INCAR, POSCAR, POTCAR, KPOINTS), as well as the output files created by VASP (OUTCAR, vasprun.xml, vaspout.hdf5). Otherwise it will be very hard to reproduce the issue and give you some feedback from our side.

Kind regards,
Pedro Melo

I face the same issue and as a result unphysical excitation spectrum.
Bands included in the BSE spin= 1
VB(min)= 125 VB(max)= 144
CB(min)= 145 CB(max)= 164

Bands included in the BSE spin= 2
VB(min)= 123 VB(max)= 142
CB(min)= 142 CB(max)= 161
I have checked the GW OUTCAR. For all k-points the band 142 is occupied and the band 143 is unoccupied.
Is it possible to change CB(min) by hand?
Thank you,
Sergey