[Pw_forum] question about pseudopotential generation by ld1.x

[Paolo Giannozzi]

On Friday 20 August 2010 04:29, jia chen wrote:

> I don't know why, for example, 5D orbitals need four lines? 

two D states to be pseudized, a bound state with 4.00 electrons
and an empty unbound state, with j=3/2 (1.5) and j=5/2 (2.50), 
due to spin-orbit nteractions (notice rel=2 in the namelist). The
file INPUT_LD1.* contains a rather detailed description of the input.

> Secondly, after plotting logarithmic derivatives, I found some
> 'sparks', both in atomic and pseudo wave functions. I would like to
> know if this is a sign of trouble?  For atomic wave functions, I have
> a 'spark' in deep place (negative energy), and for pseudo wave
> function, I have a 'spark' at energy 0.

spikes are normal features of the logarithmic derivatives. What is
not normal is to have different spikes for the pseudopotential 
and for the all-electron potential, in the region of energies that
is relevant for valence states.

Paolo
-- 
Paolo Giannozzi, Democritos and Udine University