Upper limit using StandardHypoTestInvDemo.C with different bin-size pdf and dataset

Dear RooStat expert,

I am trying to use “StandardHypoTestInvDemo.C” to calculate the upper limit. In the workspace, I have one binned data(asimov dataset), and four pdf. Three of the pdf are histogram pdf and the last one is an analytical function(second-order polynomial function).

To my understanding, since it is a binned dataset, so when calculating the upper limit, it will use a binned likelihood fit. Now what if the bin size for the data and the histogram pdfs are different? For example, the data has 50 bins, and hist-pdf1 has 30 bins, hist-pdf2 has 40 bins, hist-pdf3 has 13 bins. How will the code build the likelihood? I guess it still uses the same binning for data(50 bins)? And how will it calculate the bin content for each histogram pdf for these 50 bins?

And for the analytical function, is it using the integral, maximum, minimum, or other algorithm to estimate the bin content?(I guess it is using integral, but I just want to double check :slightly_smiling_face:)



@moneta is the author of this tutorial (StandardHypoTestInvDemo.C) he might help you.

The used binning will be the one of the data. The pdf will be evaluated at the data bin center and an interpolated value (or not depending on the RooHistPdf option) will be used. Same will happen when using a pdf based on a functional form


1 Like

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.