Strange issue while fitting

Newbie
1

Hi I am trying to fit

TF1 *func = new TF1("fitf","[3] + [2]*x + [0]/TMath::Sqrt(x + [1])", 50, 1400);
func->SetParNames("a","b","c", "d");

to a set of four (x,y) points
The fit results shows

***************************************
Minimizer is Minuit / Migrad
Chi2                      =      152.672
NDf                       =            0
Edm                       =  1.27128e-07
NCalls                    =          272
a                         =      -667829   +/-   3978.14     
b                         =      2501.96   +/-   27.8618     
c                         =     -1.19202   +/-   0.0117273   
d                         =      13044.7   +/-   11.1118

Showing the degrees of freedom to be 0 can someone please help me what I am messing here?
Further, if I change the initial parameters very slightly the fit seems to get upset as evident from

Warning in <Fit>: Abnormal termination of minimization.

****************************************
         Invalid FitResult  (status = 4 )
****************************************
Minimizer is Minuit / Migrad
Chi2                      =      154.198
NDf                       =            0
Edm                       =      0.61487
NCalls                    =         1490
p0                        =      -636027   +/-   14397.5     
p1                        =      2435.92   +/-   32.2021     
p2                        =     -1.16013   +/-   0.0250483   
p3                        =      12578.4   +/-   231.218

Thanks a lot

Edit 1: I guess ndf =0 since the number of data points are equal as fit parameters but is there any way around to fit these points ?

2

Hi,

It does not look like a problem of ROOT. Without a reproducer, it’s hard to say what the message is hinting to.

Best,
Danilo

3

Hi,
Thanks for your reply
Here is the minimalistic macro to reproduce the issue if it helps.
fit1.C (664 Bytes)

4

Hi,

Thanks. Indeed this post seems to have addressed a very similar inquiry: Is fit validity or minimizer status more important? - #2 by moneta

Best,
D

5

Thanks a lot for sharing the relevant post. I’ll go through it and get back if any queries.