On Mon, 19 Apr 2004, Rene Brun wrote: > Hi Gero, > > What you observe is not connected with the use of TF1::RejectPoint, but rather > with the likelihood option. When the "l" option is specified, > TF1::GetChisquare returns an equivalent chisquare value using the parameters > computed from the likelihood fit. This value is different from the value > of the objective function using the likelihood method. > Hi Rene, I don't use the likelihood option 'l', but integral option 'I'. If I reject a point, I always return 0 in the function, and the chi^2 I get is too high by roughly the sum of the two bins that I exclude by RejectPoint. Because it's a histogram without weights this makes me quite sure that RecejtPoint is not taken into account in calculating chi^2. Of course the fit result itself is not affected as it's a constant offset of chi^2. Do you need a small script to demonstrate the problem? Thanks Gero > Rene Brun > > Gero Flucke wrote: > > > > Hi, > > fitting histograms with a non-contigous function via TF1::RejectPoint > > works rather well, but for me it seems that the chi^2 that is shown in the > > statistic box ignores the rejection. I tried to get the chi^2 directly > > after the fit via > > > > TF1* fitFunc = ... // my TF1 defining the function > > TF1* histToFit = ... // the hist that I fit > > histToFit->Fit(fitFunc, "EMBRI"); > > TVirtualFitter::Fitter(histToFit) > > ->Chisquare(fitFunc->GetNpar(), fitFunc->GetParameters()) > > > > but that gives the same result which is also printed on screen as FCN > > while fitting: > > > > FCN=761.11 FROM MINOS STATUS=SUCCESSFUL 804 CALLS <snip> > > > > How can I get the 'real' chi^2? > > > > Cheers > > > > Gero > > > > PS: 3.10_02 on linux > > > > ----------------------------------------------------------------------- > > Gero Flucke > > 1a / 603 > > DESY > > Notkestr. 85 > > 22607 Hamburg > > +49-(0)40-8998-2454 > > > > Privat / private: > > ----------------- > > Von-Hutten-Str. 20 > > 22761 Hamburg > ----------------------------------------------------------------------- Gero Flucke 1a / 603 DESY Notkestr. 85 22607 Hamburg +49-(0)40-8998-2454 Privat / private: ----------------- Von-Hutten-Str. 20 22761 Hamburg
This archive was generated by hypermail 2b29 : Sun Jan 02 2005 - 05:50:07 MET