Passing codependencies between parameters of a fit in RooFit

Roofit and RooStats
1

I wanted to ask the experts here a question regarding letting RooFit know the correlations between normalization parameters in a fit. I am currently performing a template profile likelihood fit and I extract normalizations for three parameters that should sum to unity in the SM. I am using the fitter to extract the fractions of a model from data.
Currently, in RooFit, one has a possibility of passing a parameter in RooStats::HistFactory::Measurement under the AddPreprocessFunction method. Unfortunately for me, this function only takes one argument. Now, since for me, x+y+z = 1, could someone help me on how to actually implement this in the code? I basically need the fitter to know the correlations between each other.
Thanks

2

Hi,
Since you cannot put direct correlation between parameter in the minimisation when fitting, you should just redefine on of the parameter as a function of the others. In this way you are will have the constraints that the sum of them is equal to 1.

Lorenzo

3

Hi Lorenzo,
Thanks for the reply. The only way I can redefine the parameters is by saying:
z = 1-x-y and passing it into the AddPreprocessFunction method. The problem with this is, AddPreprocessFunction can only take in 1 argument extra, basically either 1-x or 1-y, but not both. This is what I hoped to fix.
Thanks

4

HI,

I understand now the issue. Then one needs to fix the AddPreprocessFunction. If you could do this and submit a PR will be great

Lorenzo

5

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