Getting a KSS file from a large supercell

[ivasan]

Dear all,

I am trying to generate a KSS file in a 219 atom supercell to perform subsequent GW calculations.

For the GS calculation I used these variables (apart from others related to the atoms, paralelization ...):

toldff 5.0d-5
ecut 15.0
nband 496
kptopt 1
ngkpt 2 2 2
nshiftk 1
shiftk 0.0 0.0 0.0

Due to the size of my system, I will like to include as few empty states as possible as F. Bruneval and X. Gonze did in Phys. Rev. B 78, 085125 (2008), but I didn't find in the abinit website how to do it. For this purpose I thought the following.

First, to perform a new calculation with the following variables in the input file:

symmorphi 0
iscf -2
getden -1
tolwfk 1d-12
ecut 15.0
kptopt 1
nband 496
ngkpt 2 2 2
nshiftk 1
shiftk 0.0 0.0 0.0
nbandkss 538 #corresponds to 100 empty bands
istwfk *1
kssform 1

Then, I will repeat this calculation for different values of nbandkss corresponding to different values of empty bands in order to get the convergence.

Nevertheless, I have some questions:

1.- Is the previous procedure for finding the lowest number of occupied bands right?
2.- Being nband=496 in the GS calculation, can I use nbandkss > nband to get the KSS?

Thanks in advance for your time,

Kind regards,

Iván