I’m trying to compute a CLs limit using the hybrid calculator with a code similar to hep.by/root/tutorials/roosta … emo.C.html
My background PDF is a RooHistPDF. I would like to add nuisance parameters corresponding to sytematic uncertainties on this PDF (varying each bin according to a gaussian, for example) but I don’t find a way to do that. Is that possible ?
The RooHistPdf has as parameter only a normalisation factor, so you cannot vary each bin independently. You should then build your model as a product of individual Poisson term for each bin, multiplied by a Gaussian constraint. You can also use the HistFactory which build the model starting from histogram and you can include several systematic variations.
See cds.cern.ch/record/1456844/file … 12-016.pdf