# Combination in RooStats

Hello,

I use HybridCalculatorOriginal to compute a limit and I would like to combine my results with another. I tried with a roosimultaneous PDF and a roocategory but the results don’t make any sense.
Does anyone have an example of this kind of combination?
thanks,

Mathieu

Hi,

I recommend that you post RooStats question in the Math and Stats forum.

Thanks,
Philippe.

Hi,

For doing a combination in RooStats you need to build a combined model. The best way to do it is to use the RooSimultaneus class from RooFit. You have some examples to show how to use the class in various RooFit tutorials. Once you have the combined model as a workspace you can use the RooStats classes.
One more comment, I reccomend you to use the new HybridCalculator and HypoTestInverter to compute the limit, instead of HybridCalculatorOriginal. The macro to use it is StandardHypoTestInvDemo.C

Best Regards

Lorenzo

P.S. The post has been moved to the “Stat and Math Tools” forum

Thanks a lot for the input.
I have some question with the RooSimutaneous PDF and the combination of the data.
From various tutorials I ended up with something like:

[code]RooCategory analyses(“analyses”,“analyses”) ;
analyses.defineType(“LHCb”) ;
analyses.defineType(“CMSb”) ;
analyses.defineType(“CMSe”) ;

RooSimultaneous CombSigBkgPdf(“CombSigBkgPdf”,“CombSigBkgPdf”,analyses) ;

RooSimultaneous CombBkgPdf(“CombBkgPdf”,“CombBkgPdf”,analyses) ;

//combined data
RooDataHist combData1 = RooDataHist(Data, “DataLHCb”);
RooDataHist combData2(DataCMSb, “DataCMSb”);
RooDataHist combData3(DataCMSe, “DataCMSe”);
combData1.Print();
combData2.Print();
combData3.Print();[/code]

I am trying to combine results obtained in 3 experiments binned on different variables (1bin+1bin+72bins), as the “prints” give me:
RooDataHist::DataLHCb[bin] = 72 bins (1880 weights)
RooDataHist::DataCMSb[CMSbbin] = 1 bins (2 weights)
RooDataHist::DataCMSe[CMSebin] = 1 bins (4 weights)

and when I print out the combined data I get:
RooDataHist::CombData[bin,CMSbbin,CMSebin,analyses] = 216 bins (1886 weights)

Why 216 bins?? I was expecting 72+1+1!

Mathieu