Continuing the discussion from ROOT::Fit::UnBinData. example is not working:
so far, I have been using
-Unbinned likelihood ( ROOT::Fit , link above). It requires the fit function to be normalized in the fit range, and in order to avoid convergence problems, one of the free parameters has to be obtained from the normalization condition.
-Simultaneous fit of several binned data sets: each histogram has its own fit function, and its own ROOT::Fit::PoissonLLFunction; the FCN function is the sum of individual PoissonLLFunction. All the parameters are shared between all the fit functions.
So now, I would like to use unbinned likelihood for fitting simultaneously several data sets. My question is: how the fit functions must be normalized? ( should I normalize them individually, normalize the sum…)
Thank you for your time.
ROOT Version: 6.10
Compiler: gcc 4.8.5