[Ifeffit] minimize function

Matt Newville newville at cars.uchicago.edu
Tue Jan 28 17:06:53 CST 2003


Hi Shelly,

Right now, minimize() doesn't report an r-factor, (does the doc
say somewhere that an R-factor is reported?). minimize() does
generate and report values for chi-square and reduced chi-square.

  chi-square = sum-of-squares < residual / uncertainty >

where 'residual' is the array to minimize and 'uncertainty' is an
array of uncertainties, one value for each point in the residual.
If no uncertainty is given, the uncertainties are set to 1.

reduced chi-square is simply
     reduced chi-square =  chi-square / (npts(residual) - n_variables)

minimize() is flexible enough that you could include the
uncertainties in the definition of the function to minimize.
It's also flexible enough to let you cheat, and scale
chi-square anyway you want, for good or bad....

The reported uncertainties in the parameters follows the
'normal' rule of increasing chi-square by reduced-chi-square.

An R-factor might be useful, I suppose ....

--Matt

On Tue, 28 Jan 2003, Kelly, Shelly D. wrote:

> Hi,
>
> I would like to use the minimize function to fit a line to some
> data points.  I could not find documentation on the r-square factor
> that is referred to as the confidence level for the fit.
>
> Thanks
> Shelly
>
> _______________________________________________
> Ifeffit mailing list
> Ifeffit at millenia.cars.aps.anl.gov
> http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
>






More information about the Ifeffit mailing list