I’m using Integral and IntegralError withmy function
TF1 *fit = new TF1(“fit”,finalFit,0.31914,1.13914,6);
but I think that to compute the error on the integral it is used the symmetric error from the fit
and not the asymmetric errors from MINOS.
Is there any way to use Minos errors in my Integral Error computation?
Thanks a lot
TF1::IntegralError uses the covariance matrix from the fit to compute in the approximation of gaussian errors
(standard error propagation) the error on the integral.
The Minos errors are equivalent to those obtained from the covariance matrix when this approximation is valid.
When is not, in my opinion, the best way to obtain the integral error is to fit directly for the integral, by re-parametring your fitting function and have it as a fit parameter. In this way you can perform a Minos analysis directly on the integral.