[Wien] What is p-1/2 LOs ?

[Hua Wu]

On Sunday 13 March 2005 16:17, Stefaan Cottenier wrote:

>
> In a relativistic treatment, p-electrons are indexed not by
> l=1 and their spin s, but by the quantum number j, which is
> either l-s=1/2 or l+s=3/2. In a density of states, you see
> this as a split of the p-peaks in two parts. This splitting is
> reproduced by LDA+spin-orbit poorly, unless you go to very
> high Emax (>10). In order to get the correct splitting using
> smaller Emax (3), you can 'help' the basis set a bit by adding
> LO's that have the proper shape for these p-1/2 orbitals.
>
> This is useful only for heavy atoms (5p, 6p). If this is not
> your case, then do not add p-1/2 LO's.
>
> Stefaan

Does it mean that for 3d and 4d transition metals or oxides,
one does not need to add the P1/2 LO ?  If so, how to deal with
the case.inso file to include the spin-orbital coupling which 
in some cases is important. 

thanks a lot --- H. Wu