The ABINIT Users Mailing List ( CLOSED )

[Takeshi Nishimatsu <takeshi@physics.rutgers.edu>]

Dear Ganesh and all,

> Ah! Now the value output is summed over m: it only depends on l. Someone 
> working with David Vanderbilt's group has been thinking of generalizing 
> the code, but I don't know if there is any progress yet.

I'm working on it, I'm working on it. I'm just as frustrated as you are.
It turns out to be a non-trivial problem.  (Cited from:
http://kerneltrap.org/Linux/Compiler_Optimization_Bugs_and_World_Domination 
;-)

This week, I will concentrate on this problem. Or you may help me.


Following is my current understanding.

Process of PDOS calculation in ABINIT is
src/21drive/outscfcv.F90
  src/14occeig/partial_dos_fractions.F90
     calculate partial DOS fractions per k-point (dos_fractions) to feed to 
the tetrahedron method
  src/13paw/partial_dos_fractions_paw.F90
     that PAW
  src/14occeig/tetrahedron.F90
     dos_fractions -> partial_dos with tetrahedron method
     src/14occeig/get_dos_1band.F90
         Weight the PDOS with weight of each k-point (*)
     Write PDOS to a file.

Problem is at (*). src/14occeig/get_dos_1band.F90 does not care about
symmetries and starts of k-points. Just using weight.

The m-decomposed PDOS per k-point is depends on the k-point even among its
stars. So non-self-consistent calculations WITHOUT symmetry (spgroup 1)
may be required to get correct m-decomposed PDOS. This way is what they
do with VASP.

Sincerely,
--
Takeshi Nishimatsu
http://loto.sourceforge.net/feram/
http://loto.sourceforge.net/