[Ifeffit] Athena: k-weight=3 in background removal problem

[Matt Newville]

Hi Joseph,

The Fourier transform parameters for the background subtraction do not
need to be the same for the analysis of chi(k).  In fact, they really
shouldn't be the same -- the goals for these FTs are very different.,
as the FT for background subtraction is only to determine a good
background function and to identify the oscillations in chi(k), not to
analyze them.        Setting the k-weighting to 3 for the background
subtraction effectively tells the program to not care about the
background at low k.  This can cause wild oscillations in the
background at both low- and high-k, especially when using clamps of
any kind.

For background subtraction, the k-weight should be 1, though using 2
and sometimes 0 also work well in many situations.  Furthermore kmin
should be very close to 0 (1 Ang^-1, at the most, though there may be
special situations), which is usually too low for most analysis of
chi(k).

Hope that helps,

--Matt