Problem with TF1::IntegralError in ROOT 5.22.0

ROOT
1

Hi,

I am using TF1::IntegralError() function to calculate the error of the integral. It was working fine in root5.19.04 and hence gives me finite value of the error. Now, I am running root5.22.0 and the same code with same data gives me “nan” for error.

Do I have to take care something else in root 5.22.0?

Thanks,
Manojg

2

Could you post a shorts script reproducing your problem? Lorenzo will investigate next week.

Rene

3

Hi,

Thanks you very much. I have attached the macro and a root file with histogram.

Thanks,
testFit.C (2.49 KB)
test.root (3.75 KB)

4

Actually, the fit function h->Fit(func, “RME”) behaves differently in root5.19 and root5.22 versions. Because of this, it give different value of Integral also.

5

Hello,

there was a bug when dealing with the fixed parameter in the covariance matrix. It is now fixed in the ROOT trunk version when calculating the integral error

Thank you for reporting this problem

Best Regards

Lorenzo