260 likes | 404 Views
Physics Week May – DPG-PH JetMET Robert Schoefbeck (Hephy Vienna) May, 17th. Four topics in one talk. MET-bisector studies (Jared Sturdy) Decompose MET along the jet-bisector axis in di -jet events Disentangle cont. to res. from Calo -response/noise and un/clustered energy
E N D
Physics Week May – DPG-PH JetMETRobert Schoefbeck (Hephy Vienna)May, 17th
Four topics in one talk • MET-bisector studies (Jared Sturdy) • Decompose MET along the jet-bisector axis in di-jet events • Disentangle cont. to res. from Calo-response/noise and un/clustered energy • MET in Multi-Jet Events (>2) This is not a QCD study! • low statistics; still ~ factor 1000 away from 1pb-1 • Sarah’s suggestion: Suppose that MET-mismeasurement depends on Sum-ET: • consider bins in Sum-Et, Overlay MET-shapes with 2,3,4 Jets. If true, shapes will vary with the Sum-Et bin but less/little/not with n-Jets • The point: “How (well) does MET behave in >2-Jet events” • works nicely for the data I ran over (May 5th) • 3. B-Tag / MET in heavy flavour events This should not be a B-Tagging study! • In a Diject selection, cut on e.g. SSV-BTag(Impactparameter significance) • needs more statistics, will be good for 1pb-1 • enrich sample and check with MC; don’t try to commission B-Tagger • sync WP with B-Taggers • The point: Understand Neutrino-contribution; Don’t expect big effect for SSV! • 4. Pile-Up • We had ~0.2% PU before April 24th • Pile-up is best studied in early low-PU runs where the non-PU sample is pure • PU has/will increase a lot
Technicalities CMSSW_3_6_0 on top of 3_5_7 rereco Monte Carlo: /MinBias/Spring10-START3X_V26A_357ReReco-v3/GEN-SIM-RECO MC with pile-Up: /MinBias/Spring10-START3X_V25B_156BxLumiPileUp-v1/GEN-SIM-RECO Data: /MinimumBias/Commissioning10-Apr20Skim_GOODCOLL-v1/RAW-RECOofficially certified runs as of May 5th • Event selections : – BSC Technical Trigger (TT) bits 40 OR 41 (At least one hit on each BSC side) – Veto BSC BeamHalo TT bits 36, 37, 38, 39 (BSC hit timing) – BPTX TT bit 0 (2 beam coincidence) – Physics Declared bit set (tracker and pixel HV ON)
Technicalities con’t. • “Scraping events” removal: – Require the fraction of “high-purity” tracks in events with more than 10 tracks is greater than 25%• Primary Vertex Selection – Require that: vertex is NOT fake, NDF>4, |Z|<=15cm, d0<2cm• ECAL spike cleaning in DATA, HCAL spike cleaning wrt. to recipy (V1) for JES: threshold at 20GeV, for both typeI-Met and L2L3 - Jets according to new Spring10 JEC in CMSSW_3_6_0 An event is removed if any Jet does not pass loose Jet-ID (pT-L2L3>20, EMF>.01 if abs(eta)<2.6, fHPD>.98, n90Hits>1) DiJet-selection, used througout: pT-L2L3>20 abs(eta)<3 abs(abs(delta-phi)-Pi)<1.0
2. MET in >2-Jet Events (1/4) Understand what we can expect in multi-jet events (upper left) w/o doing QCD-study. Jet multiplicities for >2 hard jets are modelled quite well in Pythia (upper right). too little statistics for Sum-ET-discrepancy in DiJet Events (lower right) but probably it’s there.
2. calo-MET in >2-Jet Events: MET-shapes in Sum-ET-bins (2/4) Bin in sum-ET; look at calo-MET-shapes with a definite number of Jets (>=2). Point: See, if MET-mismeasurement is driven by Sum-ET or by how Sum-ET splits among the jets. MET-shapes in Sum-ET-Bins inclusive (blue) and with a fixed number of jets Expect Sum-ET-dependence of (inclusive) MET-shapes, but don’t expect a dep. on n-jet
2. MET in >2-Jet Events (3/4) Met-shapes in Sum-Et bins Same, normalized MET-shapes dependend on Sum-ET
Normalized Met-Shapes in Bins of Sum-ET for different n-Jet (4/4) Preliminary result for 2- and 3-Jet Bin is promising. MET depends on SUM-ET, but not on how SUM-ET is distributed among jets.
3. MET in pile-Up (1/3) we have ~ 0.2% PU, consistent with ~100Hz and 50% efficiency. PU-MC agrees quite well; MET and, even more so, sum-ET are higher in PU-events.
3. Normalized Distributions (z-Seperation 1cm) (2/3) Data/Data comparison for events with 1 and 2 Vertices; Normalized; In PU: Higher jet-multiplicites (upper right), higher Sum-ET (lower right),higher MET (column 1&2) Very small dependence on z-Separation parameter (here: 1cm, default = 10cm ). See Backup. We will have to move to tighter z-Sep. and maybe start to subtract PU when it increases.
3. Understand MET in PU: Re-weight PU-MET w.r.t. Sum-ET (3/3) The point here is: Can the larger MET be attributed to sum-ET or nJet? MET in non-PU (blue) ~~ MET in PU, re-weighted , such that the Sum-ET shape agrees with the non-PU-shape. Good Agreement! No additional source of MET in PU. Source of MET in PU seems to be understood. Re-binning wrt. n-jets clearly isn’t appropriate.
4. MET and Heavy Flavour / MC Flavour 8.6+/-016 7.36+/-0.91 8.78+/-0.04 9.71+/-0.36 13.85+/-0.25 13.53+/-1.47 14.22+/-0.05 16.00+/-0.53 don’t expect big contribution from non-lepton B-taggers Need to move on from MinimumBias –MC (in the making)
BACKUP (Flip through the slides for visual effect of z-Separation!!)
1. type-I corrections to MET in Inclusive and Di-Jet selections