tttt fake rate Nick Amin June 14, 2018
Overview • Reminder • Last set of slides here • Re-tuned multi-isolation WPs for 2017 to lower fake rate • Fixed bug in process — muon and electron (ptratio,ptrel) cuts were swapped in measurement region • Muon FR was probably higher than it should have been • Electron FR lower than should have been • Want to now revisit closure with updated WPs and bugfix • Reference links: • old FR 1D/2D maps • old FR closure • new FR 1D/2D maps, new EWK fits • new FR closure 2
FR vs p T • From "old" to "new", QCD FR drops by ~40% as we OLD expected • Data still in disagreement with QCD at 15<p T <40 NEW 3
Closure (HT) • Previously, total closure was within a percent (luck?), but pred/obs high for muons and low for electrons • Now, closure for electrons spot on (<0.1% — luck), but muons are a factor of 2 low e+ 𝜈 𝜈 e OLD NEW 4
Closure (type) • Also compare match type (unmatched/light vs b vs c) • Similar trends in both — under/over predict b fakes for 𝜈 /e e+ 𝜈 𝜈 e OLD NEW 5
Next steps • Strange that muons are off by so much, but now I think we’ve reduced two problems (muons, electrons) to one (muons) • Doing some synch exercises between LeptonBabyMaker and FT/ SS babymaker to make sure IDs/ptrel/ptratio are consistent • Found some inconsistencies at per mille level for computations involving 4-momenta • Found per mille issue with trigger emulation for electrons • Fixing the above didn’t change anything 6
Recommend
More recommend