Integral error with a TF1 functoin not from fitting

[Y.S.Zhang]

Dear experts

I have a problem about how to get the integral error of the TF1 function not from fitting?

The case is like I have a histogram contain both signal and background distribution. And I have a TF1 function consisting of signal part (Gaussian) and background part (3rd order polynomial). Now I fit the histogram with the function. I would like to integral the signal part to get the signal yield. The way I choose to get the yield is to set the parameters of the background to be zeroes and then to integral the function. But how could I get the integral error in this condition that I do not know what the covariance matrix exaclyt is.

Many thanks in advance.

Yousen

[Wile_E_Coyote]

Instead of “gaus” try “gausn” (normalized gaussian), then the “Constant” parameter will be the “peak area” (and you immediately get its error, too), not the “peak height”:

gROOT->GetFunction("gaus")->Print();
gROOT->GetFunction("gausn")->Print();

[Y.S.Zhang]

Thank you, Wile. It is a good try. If the signal function is not gaussian, how could I do to get the integral error? I think it would be a general problem I would meet in the future frequently for me.

[Wile_E_Coyote]

See this old post (in the old “TH1F variable bin width” thread) and / or that old post (in the old “How to find the area error using peakfind.C" thread).

[Y.S.Zhang]

Thank you, Wile! Good post!