To converge optical spectra

[ams_nanolab]

What's the rule of thumb to converge optical spectra w.r.t. the no. of bands to be included above and below the Fermi level? If I look for spectra in the range of 0-5 eV, so it should include all the bands within -2.5 eV to +2.5 eV ? But I have noticed if there are say 100 bands in the energy range and I go on increasing even more bands (say 200-300) the spectra still changes, although so many bands may not be even present in the said energy range. What am I missing? 

[Ulrik G. Vej-Hansen]

It depends somewhat on your system, but we have seen cases where it is necessary to go to more than 30 eV above the Fermi level. There is not a 1-to-1 correspondence between the energy of your photons and the energy range which is necessary to investigate.

[ams_nanolab]

Yes, that seems to be the case. Thanks