[Wien] large E-cut is necessary in SO?

[Stefaan Cottenier]

>     I am doing a calculation with PbTe spin orbit coupling. I calculated 
> their energy gap.I used 3 methods as follows:
> 1. set p-1/2 zero, E-cut is 3.0
> 2. set p-1/2 zero, E-cut is 10.0
> 3.add p-1/2 LO for Pb. E-cut is 3.0
> 
>  in my opinion, the 2 and 3 should be better and they are almost the 
> same.However,their results are only a half of the 1, which are close to 
> the experimental value.but I  think the 1 is just a coincidence.Which 
> result should i believe?why?Thanks for your suggestions.
> btw. The DOS of 1 and 3 is different, a spliting of d-DOS for Pb is 
> noticed.

Regardless whether a result is physically meaningfull or not, it first 
should be numerically stable (=converged). Convince yourself that you 
can go from result 1 to result 2 by varying Emax from 3.0 to 13.0 in 
steps of 2. Does this converge to always the same numbers at some point 
between 3 and 10? If so, then you can safely ignore result 1 as being 
unconverged.

Stefaan