[Ifeffit] High-R data
Matt Newville
newville at cars.uchicago.edu
Thu Aug 26 10:01:05 CDT 2004
Bruce, Peter,
> > Athena and Artemis do not plot chi beyond 20 Å-1 or the
> > Fourier transform beyond 10 Å. Is this a limitation of
> > Athena/Artemis or of Ifeffit? Can it be changed?
>
> This only seems to be half true. The fe.060 file that comes
> with the Athena examples has data out to 22 invAng and gets
> plotted correctly by both programs.
The 'chi(k) only goes to k=20' issue might be due to the fact that
the Feff calculations don't normally go past k=20. That makes it
hard to do a fit that far out! I think this could be changed, but
it would take a little time. Are you using Feff6l or Feff8?
> On the other hand, you are correct that data only gets plotted
> in R up to 10 A. That was not a conscious decision by me and
> seems to be some kind of default in Ifeffit. Matt, how does one
> change that default?
You can set the maximum R for chi(R) from fftf() with
fftf(data.chi, kmin=2,kmax=19, rmax_out=15)
> > The 10.5 Å peak raises another issue. I understand that
> > Ifeffit by default assumes that the high-R region is pure
> > noise in order to make its estimates of uncertainty. (I know
> > the actual R-range that is used was mentioned recently but I
> > can't find it in the list archives or any of my manuals.) Can
> > this range be changed? Does anyone have any advice on how to
> > deal with this issue?
Just to add to Bruce's notes:
The prevailing wisdom is that the automated estimate of epsilon_k
is generally too low -- that makes chi_square too large. Having
'real data' out there might actually help!! But simply setting
epsilon_k to an appropriate value is probably the best bet.
Remember that epsilon_k is the uncertainty in unweighted chi(k),
and so is fairly easy to interpret (ie, no units). Artemis reports
this as 'Measurement uncertainty (k)' in the Results window after
a fit but you can overwrite it with the Epsilon value on the page
with the data file and Fit ranges. A reasonable number for most
good EXAFS would be epsilon_k = 0.001.
--Matt
More information about the Ifeffit
mailing list