Hi all,
i am trying to performe a significance study using RooMCStudy and the RooDLLSignificanceMCSModule, but i get an error when the MCStudy attempts to merge the data from the significance module:
[#0] ERROR:InputArguments -- RooDataSet::merge(fitParData_model_distribution) ERROR: datasets have different size
I include a minimum reproduction code, which gives the output
[#0] ERROR:InputArguments -- RooDataSet::merge(fitParData_model_distribution) ERROR: datasets have different size
RooDataSet::fitParData_model_distribution[#mean_{gauss},#sigma_{gauss},nbg,nsig,power,shape,NLL,ngen,#mean_{gauss}err,#mean_{gauss}pull,#sigma_{gauss}err,#sigma_{gauss}pull,nbgerr,nbgpull,nsigerr,nsigpull,powererr,powerpull,shapeerr,shapepull] = 85 entries
So the fitParData does not contain the full 100 entries (as expected because some will not converge) but more importantly does not contain the RooDLLSignificanceMCSModule variables
nll_nullhypo_nsig, dll_nullhypo_nsig, significance_nullhypo_nsig
Edit: i believe what is happening is that for MCStudy the fitParData is only added if the fit converges, whereas for the module there is no fitResult->status() == 0 check.
Can someone confirm this and find a workaround?
min_mcsign_error.cpp (2.2 KB)