[Wien] In and spin orbit interaction

[Peter Blaha]

> The other question was not directly related to the In problem (I only suspected that something was going wrong in that direction), 
> I observe more often qtl-b errors when doing SO calculations compared to non-SO calculations, even if I do not make the copy-mistake.

I have seen this for "localized" semicore states with a fairly large 
spin-orbit splitting, so that the E-l is not good anymore (or cannot 
equally well describe both, the eg. d3/2 AND 5/2 states, since
they are 0.5 Ry apart.

Eventually experiment with "hand-modified E_l" and/or local orbitals.

For p-states one could try to set the relativistic LO energy to the p1/2 
state and the E_p to the 3/2 Energy ?