Re: fitting...

[Matt Newville <newville@cars.uchicago.edu>]

Hi Patrick,

> Now I have started comparing fitting with Artemis and EXAFSPAK and
> noticed some differences again, so I decided to do a very simple test.
> When I create a single EXAFS wave with FEFF8 and use this exact wave
> to refit the data, I get the exact correct answer with EXAFSPAK.
> With artemis in R-space or K-space fitting mode, I can't get the same
> nice fit.  The problem looks like it's in the phase shift.
> I'm not sure  what's going on, but my discussion last time ended by
> looking at the 4psigma2/Reff term in FEFFIT's Chi(k) expression.

I remember there were differences in amplitude due to the
sigma2/Reff correction in feffit, and possibly a noticeable
difference due to using k and p in the cumulant expansion, and
that all the changes went away with sigma2=0.  Is that right?

I don't know why EXAFSPAK and FEFFIT give different resuls.
I'll admit that it's been a while since I checked any of this
stuff:  I tried to use the feff.dat files correctly, and apply
the cumulant expansion correctly (the use of sigma2/Reff and
the use of p instead of k *is* related to that).  I would not
want to guarantee it is right, but I think it is.

On the other hand, I have no idea what EXAFSPAK does because
the source code is not available.

It sure would be nice if Feff provided code to read its own
output files.  Unfortunately, Feff does not come with any
utility routines, makefiles, or scripts to deal with any of the
feff data files.  I cannot explain why.

--Matt