Re: polarized Feff

[Alex Ankudinov <alex@leonardo.phys.washington.edu>]

Hi Matt,

I will address your questions below, but since the last part
sounds almost like bug report, please, send me your feff.inp
file, so I can troubleshoot the problem myself and provide 
better answeres.

Alex
On Mon, 16 Apr 2001, Matt Newville wrote:

> Alex, John,
> 
> I know I've asked this same question many years ago, but I'm
> still confused, so maybe  others are confused too:
> 
>   How do I get a polarization average from Feff?
> 
> I'd like to average the polarization of an L edge over the ab
> plane in a crystal.  A simple cos^2 is not good enough for L
> edge, and I really want a polarization average.  Circular
> polarization seems to be one approach.
> 
> =Question 0:  Should circular polarization along z give a
>               polarized average of the ab plane?
> 
> Assuming it will, the doc says
>    ELLIPITICY 1. 0. 0. -2.
> 
> would have circular polarization.
> 
> =Question 1:  Why -2?
  Any nonzero value would work. I just wante to emphasize in the manual
that one need not normalize the vector in the direction of x-ray
propagation. "-" sign does not matter for most of polarization
dependence calculation, with notarious exception of XNCD, where
effect changes sign with the change in light direction.
> 
> =Question 2:  Do I also need to set
>    POLARIZATION 1. 0. 0.
>    (or anywhere else in the ab plane?)
> 
> My recollection is that this is needed.  The doc is somewhat
> ambiguous.  The weird thing is that when I use

 Yes, you need to specify the polarization, even though for 100%
circular polarized light it seems to be unnecessary information.

> 
>    POLARIZATION 1. 0. 0.
>    ELLIPITICIY  1. 0. 0. 1.
> 
> I seem to get only paths with a finite Z component, and none in
> the ab-plane.

  This sounds like a bug and I need feff.inp file for troubleshooting.
In general, for L2,3 edges I would not expect the paths with finite
z-components be neglibible, while like for K edge their importance
should be smaller. Also code rotates frame of reference to have
simpler form or polarization tensor (see comments in subroutine mkptz).
In your case the new z axis will be along polarization vector
(i.e. x-direction in old frame of reference).
> 
> Commenting out the ELLIPICITY card gives paths entirely within
> the ab-plane, though it is not an average in the ab-plane.
This also sound like a bug. 
> 
> Commenting out the POLARIZATION card gives paths entirely within
> the ab-plane, but it appears that this gives no polarization
> dependence at all.
> 
> 
> Is there any way to turn degeneracy checking off?
>
Of course there is . You can write out all paths with degeneracy 1.
But, why do you want to do that? If paths are degenerate they do not need
to be calculated many times. If you suspect the bug, it needs to be fixed. 
> 
> PS: I'm using Feff824.  I'm assuming this is the latest, but
>     have no idea if this is actually the case.

We (I and John) almost finished testing feff8.20x5 and once we are
satisfied with the tests, we'll give the code to other developers
for further testing. 

Best wishes
Alex
>