190 likes | 310 Views
+. Dan Damiani, Wolfgang Ehrenfeld, Helen Hayward, Osamu Jinnouchi, Andy Kuhl, Jovan Mitrevski, Jason Nielsen, Bruce Schumm, Peter Vankov, Martin Wildt, Helenka Przysiezniak Frey, Francisco Alonso, Tere Dova. Inclusive search for SUSY with photons and EtMiss final states.
E N D
+ Dan Damiani, Wolfgang Ehrenfeld, Helen Hayward, Osamu Jinnouchi, Andy Kuhl, Jovan Mitrevski, Jason Nielsen, Bruce Schumm, Peter Vankov, Martin Wildt, Helenka Przysiezniak Frey,Francisco Alonso, Tere Dova Inclusive search for SUSY with photons and EtMiss final states Helen Hayward on behalf of the SUSY-Pho group
+ 2 Introduction • GMSB model: • LSP is a gravitino • minimal GMSB, should now be competitive Tevatron limits set on SPS8 plane • General Gauge Mediation (GGM) • Universal Extra Dimensions • Single UED , R~1TeV-1 • Lightest Kaluza-Klein particle γ* • γ*->γG • We look for γγ+X+MET • Search for diphoton events with large MET • Main Backgrounds • QCD with instrumental MET • W/ttbar events where electrons fake photons SUSY 15/07/2011
+ 3 2010 : SUBMITTED !!! SUSY 15/07/2011
+ 4 Selection • Require two selected photons in the event. • Currently implemented cuts: • EF_2g20_loose • pT > 25 GeV – change comes from the trigger. • PhotonTightAR • |η| < 1.81 • Pass the bad cluster OQ bits • Etcone20_corrected < 5 GeV • Pt and NPV corrected • official photon cleaning cuts • We use 10 ns cut for all photons (and electrons) • Updated 2011 egamma energy rescaling SUSY 15/07/2011
+ 5 Other Selections • Overlap removal: • Unchanged from last year. • Selected photons kill electrons if dR of the clusters < 0.01. • Photon-jet overlap: like SUSY standard for electrons. • Etmiss: • MET_LocHadTopo + muon corrections > 125 GeV. • Unchanged from previous analysis. • Photon terms in MET_RefFinal are working better, but we still see problems with most converted photons (get treated as jets) – hopefully to be fixed in future reprocessing. • All other objects using the default SUSY selections. • Lar hole • We are currently using SUSY nominal veto for jets with pT>40GeV • inefficient for our signal • We will implement the “smart veto” • Technical question of how to weight MC for different periods. • (if any unforeseen problems we will stick to nominal veto) • GRL: • data11_7TeV.periodAllYear_DetStatus-v18-pro08-04_CoolRunQuery-00-03-98_Susy_ph_met.xml • (will update to -05) • Requires good photon and electron triggers • Requires good photons, leptons, jets,MET SUSY 15/07/2011
+ 6 Cut Optimization – Photon eta • 2010 used photon eta 1.81 • Egamma recommendation due to MC/data discrepancy for forward regions. • Optimisation study demonstrates cut is still good for good for 2011 • Optimise using best s2/b =>η<1.81 SUSY 15/07/2011
+ 7 Data Driven BG estimate I • Understanding the QCD control sample normalization issues • Understanding the QCD control sample shape issue (two-photon versus one-photon control sample) • Estimate background from • dijet,γ-jet, γγ • Model with QCD control sample: • Define anti tight photon as a loose isEM photon which fails some of the tight isEM requirements. • This control sample models the diphoton MET distribution very well at low MET • Development expected (~ 1-week): • Small study to understand if any W contamination in this control sample and how to remove if present • Small study to improve understanding of QCD control sample SUSY 15/07/2011
+ 8 Data Driven BG estimate II • To estimate BG from genuine MET sources Electron-Photon control sample • Scale by electron->photon fake rate • Subtract QCD/Z->ee contamination by using QCD template • (the blue histogram in figure) • Difference is estimate of W and Ttbar contributions • MC templates are scaled to data for 40<MET<125 GeV • Need to add Wgamma MC samples to this plot • ~ few days 2010:Cone20/Et<0.1 2011:EtCone20_cor<5GeV SUSY 15/07/2011
+ 9 BG results • DiJet,PhoJet,PhoPho BG : Predicted background above 125 GeV = 2.73 • +/- 0.86 (stat) • +/-0.59(template shape) • +/- 0.088(template scale) • GenuineMet BG : Predicted background above 125 GeV = 2.87 • +/- 1.16(stat) • +/- 0.28(MC extrapolation) • +/- 0.017(template for Z/QCD) • +/- 0.29(error on scale factor) • QCD+EWK = 5.6 +/- 1.3(stat) +/- 0.7(sys) • Zgg=0.12+/-0.014(stat) • Wgg=0.019+/-0.009(stat) • Total = 5.8+/-1.3(stat)+/-0.7(sys) • QCD template is normalised to the diphoton distribution for MET< 20 GeV • W/ttbar entries in 1st bin are artifact of this scaling. SUSY 15/07/2011
+ 10 Non collision BG • Cosmics: • Use L1_EM3_EMPTY triggers to determine the rate of tight diphotons per unit time per bunch in cosmic events. • Assumed to pass the primary vertex cut by being overlaid with a minbias event. • Background demonstrated to negligible in 2010 • Estimate for 2011 due < 1 week • Beam background • Use unpaired bunches • <2 weeks. SUSY 15/07/2011
+ 11 Systematic errors on signal MC 2010 *2010 values still used. Updates expected within a week SUSY 15/07/2011
+ 12 List of Egamma Systematics I • Trigger • Now have new systematics based on the bootstrap method. An uncertainty of 0.6% is assigned based on the difference in the efficiency measured in data and MC • The efficiency measured in data is 100+0-0.38%. • Because fakes also get triggered, an uncertainty of 0.6% is assigned based on the 99.4% efficiency calculated for JF17 and JF35 samples that have offline photons. • To be repeated on our MC samples. • Isolation • use egamma prescribed systematics for the isolation: apply shifts separately to pt and nPV corrections and treat the two uncertainties as uncorrelated. ETA: end of the week. SUSY 15/07/2011
List of Egamma Systematics II • Identification • Use official egamma fudge factors. These have two uncertainties: those associated with the method itself, and those associated with imperfections in measuring the FFs, i.e. the background contamination. For the former, we reweigh the parametrized uncertainty that was given in the SM γγ note. For the background contamination we vary the FF sets. ETA: end of the week. Lastly, we might try extrapolation again, but that was negligible last year • Material • Use same method as used for GGM in 2010 analysis: reweigh the parametrized uncertainty that was given in the SM γγ note. ETA: end of the week • Pile up • Expected systematic to be smaller than 2010 due to reweighting technique • Use method prescribed in H->γγ note • “Efficiency versus μ. The effect of the pile-up is estimated looking at the variation of the isEM tight efficiency as a function of μ (the average interactions per beam crossing) and a ±2% uncertainty is assigned per photon.”
+ 14 Preliminary PDF and scale errors • values in Scale/PDF/SUM are relative value against the NLO sigma. • e.g. at Mgl=1000GeV, the error is 38.9%. • Scale is estimated by changing the Factorization/Normalization scales by x1/2 and x2, • PDF errors (90% CL) are translated to 68% by dividing them by 1.645. • Scale/PDF are the average of the two calculations (i.e. a half and twice in scales for Scale, +/- error PDF sets for PDF.) The SUM column is the error used in the limit calculation • Large k-factors due to poorly modelled LO numbers. SUSY 15/07/2011
+ 15 Provisional 2011 Limits • LIMITS USING LAST WEEKS NUMBERS • WE NOW AGREE ON 5 OBSERVED EVENTS • Using: • Observed events = 4 • BG= 4.79 +/- 1.38(stat) +/- 0.68(sys) 2010 SUSY 15/07/2011
+ 16 Signal Interpretation • GGM-2D • Preliminary results presented • GGM-3D • Points exist, but may not make it into immediate paper/note if time short. • UED: • MC ready, preliminary 2011 exclusion limits ~1 week. • SPS8 • Not competitive with Tevatron in 2010, should be competitive in 2011. • Herwig++ default of 500 GeV used for Graviton in error • Currently no useable MC samples • Bugfix SLHA files now available • Discussing with SUSY MC experts about priorites and timescale for new SPS8 full simulation samples. SUSY 15/07/2011
+ 17 Conclusions • We are mainly repeating 2010 analysis with some refinements • Had a first look at the 2011 data . • Candidate events are present in the signal . • Data appears to agree well with our background model. • We are aiming for a conference note for lepton-photon and to develop our signal interpretation for a paper for the end of the year. SUSY 15/07/2011
+ 18 Backup SUSY 15/07/2011
+ 19 Cut Optimization – Jet cut for LAr FEb cleaning • (use photon eta > 1.81) • Remove event if jet with pT> X GeV is pointing to Missing LAr Febs. • What should we use for 2011? SUSY 15/07/2011