[Ifeffit] minimize function

Bruce Ravel ravel at phys.washington.edu
Tue Jan 28 16:48:19 CST 2003


 > On Tue, 28 Jan 2003, "Kelly, Shelly D." <SKelly at anl.gov> said:

 SDK> Hi, I would like to use the minimize function to fit a line to
 SDK> some data points.  I could not find documentation on the
 SDK> r-square factor that is referred to as the confidence level for
 SDK> the fit.

I was just browsing through `iff_minimize.f'.  I am sure that Matt
will correct me if I am wrong, but it seems as though no R-factor is
evaluated by a call to minimize() in ifeffit.  Chi-square is
essentially the sum of squares of the function that you minimize
(which, in your case, presumably, is the data minus the line).  It and
a reduced chi-square (normalized by the number of points in the
minimization function minus the number of variables) are evaluated and
stored as program variables, but there appears to be no R-factor that
is stored.

I just did a minimize and that seems to be the case, despite what it
says in chapter 7 of the Ifeffit tutorial on Matt's website.

If you were to make an R-factor that is the real-array equivalent of
Ifeffit computes when fitting (imaginary) R-space EXAFS data, it would
be something like:

       Npts     [function(i) - data(i)]^2
   R =   sum  ----------------------------
       i=i             data(i)^2

That would be a sort of positive-definate, percentage misfit and would
be trivial to compute using ifeffit's array manipulation.

HTH,
B


-- 
 Bruce Ravel  ----------------------------------- ravel at phys.washington.edu
 Code 6134, Building 3, Room 222
 Naval Research Laboratory                          phone: (1) 202 767 5947
 Washington DC 20375, USA                             fax: (1) 202 767 1697

 NRL Synchrotron Radiation Consortium (NRL-SRC)
 Beamlines X11a, X11b, X23b, X24c, U4b
 National Synchrotron Light Source
 Brookhaven National Laboratory, Upton, NY 11973

 My homepage:    http://feff.phys.washington.edu/~ravel 
 EXAFS software: http://feff.phys.washington.edu/~ravel/software/exafs/



More information about the Ifeffit mailing list