search

HEP-KBFI / hh-bbww

repository for HH->bbWW analysis
1 stars 3 forks source link

Revisit reweighting in non-resonant analysis #36

Closed ktht closed 3 years ago

ktht commented 3 years ago

I have a feeling that I didn't update this repository in parallel with the multilepton repository when migrating to the latest HH reweighting scheme. At least lines like these: https://github.com/HEP-KBFI/hh-bbww/blob/10361290adc2d11543707e52fcf538ec458156c0/bin/analyze2_hh_bb1l.cc#L596-L600 raise eyebrows because we shouldn't reweight to LO and then to NLO. Instead, we should straight up use NLO samples and reweight from NLO-to-NLO.

The second issue reported by @saswatinandan is that the BDT Ntuples are not filled with the reweighting weights. The two problems might be connected.

ktht commented 3 years ago

I misread line 598 as HHWeightLO_calc. There's no need for two instances of HHWeightInterfaceNLO because the class provides necessary methods to retrieve relative reweighting weights. Also, setting apply_coupling_fix_CMS argument to true makes no sense when reweightig NLO samples. Undid commits d6d2541e3e30998a2b4a9249b4a6d4cb7cacbb9a and 63a4e386b010a4ec5260bbf4b7a8edf5e3bee1b9.

The real reason why the BDT Ntuples was missing the reweighting weights is because the branches were booked only if HHWeightNames was initialized: https://github.com/HEP-KBFI/hh-bbww/blob/10361290adc2d11543707e52fcf538ec458156c0/bin/analyze2_hh_bb1l.cc#L1036-L1040 whichi was true only for LO reweighting: https://github.com/HEP-KBFI/hh-bbww/blob/10361290adc2d11543707e52fcf538ec458156c0/bin/analyze2_hh_bb1l.cc#L589-L594 @siddhesh86 spotted the problem in 2lss and 3l analyses in HH-multilepton.