I have a question regarding simultaneous fits using RooFIt.
I have 3 data samples, A, B and C built from 3 histograms. These are 1-dimensional samples each one spanned in a different variable varA, varB and varC, with different ranges, binning and the histograms with different total weight, ie number of entries. I want to perform a simultaneous fit on this 3 samples since it share some parameters, then:
RooRealVar varA, varB, varC;
RooDataHist A(“A”, “A”, varA, histoA);
RooDataHist B(“B”, “B”, varB, histoB);
RooDataHist C(“C”, “C”, varC, histoC);
//to later assign the right pdf to each sample it is necessary to define a category
mapping[ “sampleA” ] = A;
mapping[ “sampleB” ] = B;
mapping[ “sampleC” ] = C;
//then I proceed to make the combined dataset
RooDataHist dataset( “dataset” ,“dataset” , RooArgSet(varA,varB,varC), cat, mapping );
This creates a 3D binned dataset!!! (why?). If each of the initial histograms have for instance 100bins, then it creates a dataset with 100100100*3(the 3 comes from the 3 types defined in the category) bins… but I’m dealing with ~300bins each histo!!! and it saves each bin in memory causing a large CPU overhead. It cannot be possible that from 3 simple histos I fill the full cpu of my laptop, thus I’m dong something wrong but I cannot find a different way of doing this.
Do you know which is the correct way of doing this??
thanks in advance!