This page documents the boosted jet performance studies.
Physics studies
Look at performance of chose tagger(s) for sensitivity in prototype analysis.
Background Rejection studies
Jet mass responses
Try smaller R jets ?
MVA performance
Try and revive our
TaggerMVA code once 'single tagger' studies completed. Just quickly to see if BG rejection can be improved upon with multivariate.
Technical details
Synchronising with UGeneva
Compare latest bg rejectoin numbers with bug fixes. If not matching use their mass windows etc. directly (very sensitive to those values).
MC event weights
- QCD
- event weight are calculated by xsec * filter_efficiency * mc_event_weight / NEVENTS. Note that the NEVENTS are the number of events processed in the grid. This is not essentially the same as the generated events, as you might have 0.1% unfinished grid jobs.
- In the Edinburgh ntuple, this corresponds to evt_scale1fb * mc_event_weight[0]
- Signal MC
- Need to apply jet pt reweighting to match to the QCD jet spectrum, this is done by the leading ungroomed CA 1.2 truth jet
pt Binning
- Ungroomed CA 1.2 truth jets |eta|<1.2, and present studies in different bins of this jet pT
Truth matching
- applying a W parton matching ∆R < 0.75R and is matched to truth groomed jet (∆R < 0.75R, where R is the jet radius parameter)
68% mass window
- How is this being determined?
documentations
Talks
--
YanyanGao - 2015-03-18