[Ifeffit] Spline range and FT range

[Bruce Ravel]

Hi Kug-Seung,

I was at the XAFS conference last week, as were a bunch of the other
folks who might answer a question like this.  Guanghui pointed you at
the Athena user guide, which will help to some extent with your
questions.

On 08/25/2015 08:20 AM, 이국승(에너지환경소재연구팀) wrote:
> While I was processing my data, I found that the FT features are
> changing as I change the spline range.
>
> I think the change is severe as my data contain a lot of noises at
> high-k regions (above 12k).
>
> The change is remarkable even when FT range is fixed and even when the
> spline range is far from the low and high limits of FT range.
>
> When data are imported into athena, the defalut values for spline ranges
> from 0 to the end of measured data and the high limit of FT range is set
> to  a value of the high limit of spline range minus 2.
>
> There must be a relationship.
>
> Could anyone explain these things and general rules for setting the
> spline range?

Everything in Athena is based on Fourier analysis, including
background removal.  The algorithm for background removal is explained
in http://dx.doi.org/10.1103/PhysRevB.47.14126

The basic idea is that the background spline is found by minimizing
the Fourier components below Rbkg.  Thus, the details of how the FT is
done affects the functional form or chi(k) that results.  If your data
are very noisy, then the spline function will be sensitive to the
noise.  That's just how it works.

The defaults in Athena are not chosen because they are "right" --
certainly not!  When I give lectures about Athena, I explain that the
defaults are chosen because they are "non-stupid".  That is, they are
values that often work reasonably well.  If your data are fairly clean
and well-behaved, then the defaults can sometimes even be used
unchanged.

When you have very noisy data, it is incumbent upon you, the user, to
examine the background removal and normalization parameters to verify
that they are chosen sensibly and that you can understand the
resulting data processing.

As for setting the parameters of the spline, it may help to play with
Rbkg.  It may help to make the k-weighting for the background removal
smaller (remember that the spline is found by doing a FT, if that
k-weighting is large, the noise gets weighted and the spline is yet
more sensitive to the noise).  It may help to limit the range to
include only the cleaner part of the data.  It may help to loosen the
high end spline clamp (which you can read about in the Athena manual).

Since you did not show us what your data looks like, it is hard to say
something more specific than that.

Regards,
B



-- 
  Bruce Ravel  ------------------------------------ bravel at bnl.gov

  National Institute of Standards and Technology
  Synchrotron Science Group at NSLS-II
  Building 535A
  Upton NY, 11973

  Homepage:    http://bruceravel.github.io/home/
  Software:    https://github.com/bruceravel
  Demeter:     http://bruceravel.github.io/demeter/