The ABINIT Users Mailing List ( CLOSED )

[PGanesh <pganesh@ciw.edu>]

Hi,

1) I am studying a AF magnetic structure of PbCrO3 (using 
norm-conserving psps.) and wanted to get the LDOS projected on the 
spherical harmonics to get s,p,d,f contributions and the local magnetic 
moment on each atom.  Ideally, the difference in the integrated LDOS 
between the up and douwn spins, integrated upto Ef, should give me the 
magnetic moment on that particular atom.  But it doesn't (I use prtdos 
=3 and chose a reasonable sphere radius, 'ratsph' for each 'iatsph'.  
Similar input file to study AF FCC Fe gives correct result as in the 
tutorial ).  If I use the python script with one of the tutorials (which 
reads the charge density .DEN file) then I get a reasonable value for 
the moment on the two Cr atoms, and they have equal and opposite value 
as expected.  

It appears that for complicated structures with many atoms the prtdos=3 
is not working right.  I am using v5.4.2.

2) How to get the true ang. momentum resolved LDOS when I use PAW psps. 
?  The description of 'prtdos' variable talks about adding a spherical 
part.  Is there a detail description somewhere how to perform this 
operation since the WFK files are unformatted.

Thank you very much.

-Ganesh

Matthieu Verstraete wrote:
> Hello Shinekin,
>
> I'm not sure exactly what you are referring to, but in principle 
> you're right, the integrated DOS is just the integral, and there 
> shouldn't be any particular difference. The DOS is built, in the 
> standard version, from a sum of dirac peaks which are smeared with the 
> electronic temperature. If you use a very small smearing the peaks 
> will be huge to get the correct integrated weight. The integrated DOS 
> will have a jump but is usually much smoother (what with being 
> integrated and all).
>
> Use a larger smearing or, (even better), a denser kpoint set, or the 
> tetrahedron method, which gives a continuous curve. See prtdos 1 2 3.
>
> Matthieu