[Ifeffit] Re: ifeffit window

[Matt Newville]

Hi Frank,

I Hope it's OK that I CC: this to the ifeffit mailing list.

On Thu, 24 Apr 2003, Frank Haass wrote:
> Hi Matt!
> 
> I have worked on my data with the IFEFFIT for a few days now, and I think it
> is quite a very good improvement from the autobk and feffit-Versions I used
> bevor. Great! But, I have a question concerning the windows for FT which can
> be used in IFEFFIT.
> 
> In the "old" FEFFIT, the keyword ikwindo could be used to change the window
> function from Hanning to Gauß etc.
> 
> Using the keyword kwindow (kwindow = 1 etc.) in IFEFFIT, it was not possible
> to do this. Have I taken the wrong keyword? Are the numbers uses for the
> separate window-functions in IFEFFIT the same like in FEFFIT?

The 'ikwindo' flags in Feffit were too hard for most of to
remember, so Ifeffit uses character strings for the name of
the window function as with
  fftf(data.chi, kmin=2,kmax=12,dk=1, kwindow='hanning')

The feffit() and window() command use this keyword too, or it
can set with the string $kwindow:
  set $kwindow = 'hanning'

I'm not sure that's easier than it was in Feffit, but it
doesn't seem worse.  For back-transforms, use 'rwindow' in
place of 'kwindow'.  The supported window types are:
   hanning   fhanning    kaiser      parzen
   welch     sine        gaussian

Only the first 3 letters of these namew are really needed.  
The default is 'hanning', but 'kaiser' is often a good choice.  
Of course, the GUIs have drop-down lists for this.

I see that this information is not in the on-line docs.  
Eeek!  I would have sworn it was....  I'll be sure to add this
when I post the next version.  The good news is that I am
actually working on this these days, and even though I keep
finding more to add and fix, I'm optimistic that 1.0077 will
be ready next week.

--Matt