[Ifeffit] statistical parameters

[Bruce Ravel]

On Wednesday 14 December 2005 10:21, Stefano Ciurli wrote:
> Here is an example. We find with IFEFFIT (we use Dk= 10 (from 2 to 
> 12), DR=5 (from 1 to 6).

> chi^2 = 87.1
> reduced chi^2 = 3.7
> parameters: E0, three distances and three sigma^2
> we set S0^2 fixed to 1
> R-factor: 1.15%
>
> The ratio between chi^2 and reduced chi^2 is 23.54, which is not an
> integer number as we would expect. Plus, this number does not
> correspond to Ninp -Nvar (31.83 - 5 = 26.83)

According to what you said above, Nvar = 7, so nu=24.83, which is much
closer to 23.54.  That still seems bigger than rounding errors, though.

Here is my understanding of Ifeffit, but I am sure Matt can elaborate.
Background removal uses the integer part of dk*dr*2/pi+1 to determine
the number of knots.  Fitting uses dk*dr*2/pi as a real number to
determine Nidp.  In any case, that's what Artemis reports when you ask
it how many independent points in a project.  My memory is that I
checked it against Ifeffit's report of Nidp.

B

-- 
 Bruce Ravel  ---------------------------------------------- bravel at anl.gov

 Molecular Environmental Science Group, Building 203, Room E-165
 MRCAT, Sector 10, Advance Photon Source, Building 433, Room B007

 Argonne National Laboratory         phone and voice mail: (1) 630 252 5033
 Argonne IL 60439, USA                                fax: (1) 630 252 9793

 My homepage:    http://cars9.uchicago.edu/~ravel 
 EXAFS software: http://cars9.uchicago.edu/~ravel/software/