[Wien] c-Si optics sum rule too low

[Pavel Ondračka]

Dear WIEN2k mailing list,

I've been trying to compute the effective number of valence electrons of
crystalline silicon from the optical sum rule and I'm getting value of
7.4 (as calculated by the joint program) instead of the expected value
around 8 (4 valence electrons per atom).

I already have well converged case with 10000 k-point, Rmt*Kmax = 9 and
I've also properly increased the EMAX in case.in1, case.inop and
case.injoint that any further increase have no effect on the sum rule.
I'm using the mBJ exchange potential, however even with GGA or LDA the
result doesn't change much.

Any ideas what further steps might be needed to get the expected value?

BTW is there an easy way to also compute the dielectric function of core
electron transitions and hence see if at least the total sum rule checks
out?

Best regards

Pavel Ondračka