Chi2/ndf issue

Newbie
1

I have an issue with Chi2/ndf in fitting data on Root. For example o have fit function of sqrt(ax+b) , and data. The visual of graph is ok, i think, but the value of chi2/ndf espesialy value of chi2 is very very big. I cant understand where is the problem.
Here how its looks like.

image
image1016×682 67.1 KB

Minimizer is Minuit2 / Migrad
Chi2 = 46335
NDf = 8
Edm = 3.0815e-06
NCalls = 561
p0 = 0.391581 +/- 0.00229985 -0.0023068 +0.00230125 (Minos)
p1 = 0.00284131 +/- 5.91676e-06 -5.91951e-06 +5.93552e-06 (Minos)
Reduced chi2= 5791.87
P.s. previosly i have fitted plutonium spectra with counts/channel data. I done the energy calibration and thw chi2/ndf value was closer to 1. Then i have done this Fwhm/energy calibration but have issue with chi2/ndf.

2

In my very naive understanding, it’s caused by the points that away from the fitted function (with error weighted). You can try:

  1. Set a better initial value so that minimizer can avoid the local minimum. ROOT::Math::Minimizer::SetVariable() or TF1::SetParameter()
  2. Try another function that fits the data better, say log(ax+b)