Fit data point with error on x and y

Roofit and RooStats
1

Hi all,
I have a quetion about the fit of data point with errors.

I have 5 values and errors of y and 5 values and errors of x.

These point follow the function y=a*x - b/x and I want to find the values of a and b.

In the forum I found another topic with a link to:

root.cern.ch/root/htmldoc/TGraph.html#TGraph:Fit

I created a TGraphErrors and I fitted with the my function, but I obtain, for a and b, the values of the initial condition.

This is the code:

[code] {

TCanvas *c1 = new TCanvas(“c1”,“CLT”,200,10,700,500);

Int_t n=5;

Double_t y[n]={-4.04,-2.74,-1.15,1.49,6.873};
Double_t error_y[n]={0.50,0.25,0.08,0.09,1.90};
Double_t x[n]={22000.0,22930.0,23880.0,25130.0,26390.0};
Double_t error_x[n]={440.0,470.0,500.0,530.0,540.0};

TGraphErrors *gr1 = new TGraphErrors(n,x,y,error_x,error_y);

TF1 *fit = new TF1(“fit”,"([0]*x) - ([1]/x)",21000,27000);

fit->SetParName(0,“a1”);
fit->SetParName(1,“a2”);
fit->SetParameter(0,1);
fit->SetParameter(1,1);

gr1->Fit(“fit”);
gr1->Draw(“AP”);

c1->Update();
}
[/code]

Where is my error?
Thank you
Ciccio

2

Hello,

your fit fails because you are using initial values of parameters too far way from the minimum. I have tried with

   fit->SetParameter(0,1/24000);
   fit->SetParameter(1,24000);

and it works fine with me

Lorenzo

3

Hi,
thank you very much, now works!!!
Thank you
Ciccio