Problem with RooFFTConvPdf

Dear experts,

I’m trying to convolute a breit-wigner with a crystal-ball for estimating a signal pdf, both the width of the Breit-Wigner and the sigma of the Crystal-Ball are mass-depedenent. I’m using RooFFTConvPdf for this. The pdf shape is sane but when trying to using it to compute an upper-limit

We have realised that when evaluating the function at different mass points I get values not taking into account the number of bins I had to use for the creation of the RooFFTConvPdf.

Then I’m using this to compute the upper-limits on the number of signal events using RooStats::MCMCCalculator I was thinking the code should be smart enough to take into account the fact that this pdf is binned.

Any help is much appreciated.

I attach a dummy macro that reads the workspace (and the workspace) and prints some values to demonstrate what I’ve just stated so you can also reproduce it. I run it simply ./ (2.11 KB)
dimuon_13TeV_decJamb2015_BWCB_workspace_5000.root (103 KB)