Unconstrained parameter in limit setting


I am new to RooStat and I am trying to use it in order to set expected limits (for detector sensitivity studies). I found in the documentations how to Gaussian constrain background parameters. My question is if it is possible to “float” certain backgrounds in such away that it is used as a free parameter during the fit while constraining some others.

Many thanks,


You can leave the background parameter free, without any additional constraint. If you are using a frequentist tool, RooStats will then fit them to their best value on the observed data, while if you are using a Bayesian tool the likelihood will be integrated in these parameter, thus it will be marginalized. If no prior pdf is supplied in these parameters, a uniform prior in the background parameter range will be used in the integration.
If you have further question, please let us know,

Best Regards