Hello root users i’ m beginner root user i see that Up=1 for chi square and Up=0.5 for likelihood ﬁts so i think that the value of Up depend on the distribution that follow the data, it’s right ? for example for the normal distribution i use the value Up=2.3 for 1sigma. thank you for your replay in advance
No, the Up values needs to be used if you are fitting by passing directly the likelihood function or the least square function to Minuit or to compute the parameter contours in multi-dimensions.
By default ROOT uses the correct value when fitting an histogram or graph with TH1::Fit and the value does not need to be different when using different model functions (e.g. normal or exponential)
Thank you very much Lorenzo for your reply usually before calling MIGRAD and MINOS I define the up as up=1 then when I want to plot the joint confidence contours for two parameters (2D) and fixing the others I give the up=2.3 it’s right what i’m doing ?
This is correct to set the value at 2.3 when you want to plot the 68% confidence contour for the two parameters.
The value you should set is given by ROOT::Math::chisquared_quantile(prob,2) for two parameters contours
and prob is the desired probability content of the contour