1. 1 RA1 Analysis Exclusive n-jet + MET search Overview and Status Report Tanja Rommerskirchen on behalf of the RA1 group Universitaet Zuerich SUSY Workshop; 13.05.09

2. 2 Overview RA1: Exclusive n-jet + MET search A search for an excess in the n-jet channel based on kinematic variables. Object definition and pre-selection selection Background control distributions Background estimation Systematic studies Most of the work has been carried out so far by the combined effort of: Imperial College of London (Markus Stoye, Tom Whyntie, Robert Bainbridge, Jad Marrouche) University of Zurich (Tanja Rommerskirchen) CERN (Henning Flacher) More groups start to join the RA1 effort. Interested groups are heartily invited to express their interest!

3. 3 Pre-selection and object definition Pre-selection in a nutshell: >=2 (good) jets + no leptons Object definition: good jet: iterative cone 5 calo jets & Fem 50 GeV Pre-selection: Trigger: (see Taylan’s talk) Jet Trigger, MET, MHT, HT trigger,... No (PAT)-electron or global muon with pT > 10 GeV |η| first jet < 2 pT of first and second jet > 100 GeV

4. 4 Datasets datasets are listed at: https://twiki.cern.ch/twiki/bin/view/CMS/SusyPATLayer1 https://twiki.cern.ch/twiki/bin/view/CMS/SusyPATLayer1 16 M PYTHIA QCD Pt > 80 GeV ( Thanks to Arnd, Aachen) 20 M Madgraph QCD events HT > 100 GeV (Thanks to Dietrich, Vienna) 10 M W->lν (madgraph) 1 M Z-> νν (madgraph) 1.4 M Z->jj (madgraph) 1 M tt (madgraph) ~100k SUSY LM0, LM1, LM2, LM3, LM4, LM5 sftsht-PYTHIA After pre-selection further background reduction by: HT > 350 GeV

5. 5 Selection: α T α T extension to n-jets as proposed in AN 2008/114 Two pseudo jets are formed which balance each other as good as possible in HT. Where “pseudo-jets” HT = Sum(ET), with ET being the transverse energy of the jets in the “pseudo-jet”. The constructed pseudo dijet system showed to have similar properties than a real dijet system.,where For a perfect balanced system α T = 0.5

6. 6 Selection: Closure of jet kinematics Several jets with pT below 50 GeV can contain a considerable amount of the energy in the event. MHT ratio is > 1 if the jets with pT between 30 GeV and 50 GeV point in the opposite direction to MHT(selected jets). Can protect from events in which neglected low energy jet lead to considerable MHT Redundant if jet threshold can be lowered MHT ratio after pre-selection and α T cut

7. 7 Cutflow and Synchronisation Selection cutQCDZ nunuW lnuttZ+jetsLM1LM0 pre-selection2.2x10^7690247935472656303011 HT > 350 GeV5.2x10^630596026601006052757 αT > 0.5510.912.815.410.30.3169335 MHT ratio < 1.252.412.815.49.10.3168321 This numbers are the result from synchronisatin effort between the: Imperial College group (Markus Stoye, Tom Whyntie, Robert Bainbridge, Jad Marrouche) University of Zurich (Tanja Rommerskirchen) CERN (Henning Flacher) Further synchronisation is ongoing between the RA1 and the university of IOWA (Taylan Yetkin, Sercan Sen, Kerem Cankocak, Ferhat Ozok) Everybody who is interested is welcome to join the synchronisation effort! # jets = 2...6

8. 8 Control Distributions: (Markus Stoye) Several control distributions should help to identify the remaining events after the final selection Comparison of track based MET (MPT) and jet based MET (MHT) “biased DeltaPhi” QCD events at small values At least for MC remaining QCD events show up at expected regions!

9. 9 Control Distributions: (Markus Stoye) 2dim plots give valuable information by studying correlations For real missing energy events discriminating variables are correlated Again remaining QCD events are in “typical” region

10. 10 Background Estimation: Matrix Method (Tanja Rommerskirchen) The matrix (ABCD) method requires two uncorrelated variables LM1 + bkgd background only Signal region: first jet |η| < 2 Control region: 2 < first jet |η| < 3 The α T cut efficiency seems to be uncorrelated with respect to |eta| of the leading jet This allows to apply the matrix method

11. 11 Background Estimation: Matrix Method (Tanja Rommerskirchen) Plots illustrate results for control region: 2.0 < |η| < 3.0 Background closure test:In case of LM1 signal: data sampleestimated # background eventssimulated # total events background (QCD PYTHIA) 42 +- 17 +- (9 MC error)40 +- 6 +- (4 MC error) background (QCD MADGRAPH) 40 +- 16 +- (9 MC error)38 +- 6 +- (3 MC error) LM0 + background (QCD PYTHIA)153 +- 32 +- (3 MC error)361 +- 19 +- ( 7 MC error) LM1 + background (QCD PYTHIA)82 +- 23 +- (9 MC error)208 +- 14 +- (4 MC error) Closure test in MC passed! Significant excess could be found!

12. 12 Background Estimation: Matrix Method Validation plan with real data In case of no SUSY In case of LM0 signal For lower HT bins signal contribution is negligible ⇒ ratio remains approximately flat in eta ⇒ signal should show up for higher HT bins (heavy SUSY particles) Elementary check needs to be carried out once data is available!

13. 13 Background Estimation: W (Jad Marrouche) Select W to estimate Z to invisible. -> invert muon veto (require isolated global muon with pT > 10 GeV) -> use W→lν event topology to reduce SUSY contamination After full selection with 1 global muon Variables do discriminate between SUSY and W, tt W and tt very similar ⇒ overestimation of Z->νν

14. 14 Systematic uncertainties Robustness Several tests have been made to estimate the robustness of the selection QCDZ nunuW lnuttZ+jetsLM1LM0 10% pT smearing2.317.518.211.70.3173349 0.1 rad Φ smearing2.613.216.29.30169323 + 5% energy scaling4.618.717.412.30.3174358 - 5% energy scaling2.412.114.98.90.3168316 +3% energy in FWD4.514.514.19.60.3169326 -3% energy in FWD2.412.315.39.20.3168321 Overall the signal to background ratio remains stable under the varying conditions! Also the Matrix Method has been tested for all this variations and found to remain valid

15. 15 Conclusions In a combined effort a plan for a fully hadronic search has been developped This search is not based on calorimetric MET but uses jet-kinematics and event- shape variables instead A significant signal should be observable with 100pb-1 for LM0 and LM1 Several control distributions in place to characterize remaining events after the final selection Data-driven background estimation with Matrix Method passes closure test done with MC events; also after systematic variations Complementary background estimations are possible: gamma+jet for Z to invisible (Approved Note CMS PAS SUS-08-002) W for Z to invisible background (Approved Note CMS PAS SUS-08-002) W, tt for W->τν background. (Mariarosaria)

16. 16 BACKUP

17. 17 CMS mSuGra points CMS PTDR II

18. 18

19. 19 Control distribution: Meff Instead of a pure cut and count approach looking at Meff gives valuable insight in the nature of the remaining events

20. 20 Mono-Jet Final States (CMS) Predicted by ADD models (introduces δ flat extra dimensions; fundamental mass scale MD from which the Planck scale in four dimensions is derived) signature: high transverse momentum jet > 300-400 GeV in central region + large ETmiss (same order of magnitude as jet) ~back to back with jet main background: Z→νν; W+jets, QCD, t, tt trigger: HT (scalar sum of jet pT); MHT (vectorial sum of jet pT) selection: jet electromagnet fraction < 0.9 (against electrons) track isolation veto < 0.1 (reduces tt and W(lν)+jets) leading jet pT > 350 GeV reject events with more than two jets ΔΦ(j2,MET) > 0.5 & ΔΦ(j1,MET) > 2.8 exclusion limit discovery reach 100 pb -1

21. 21 Background estimation Z + neutrinos + jets Several methods investigated by CMS and ATLAS: from Z->ll + jets (replace method): (ATLAS) from photons : (CMS) from W+jets (CMS) Good to have several methods! Crosschecks! Different syst. uncertainties! Different possible signal contamination same kinematic problem Br(Z->ll) nunu) --> statistical uncertainties

22. 22 Background estimation: Replace method Z + neutrinos + jets Method: - select clean sample of kinematic similar control background - define “pseudo E T miss ” in control sample - correct for differences between Z→νν and control sample i.e.: different cross-section, different efficiencies, different kinematics Z->ll + jets: lepton p T ≡ “pseudo E T miss ” ; similar process ⇒ small systematic uncertainties few events left after selection ⇒ large statistical uncertainties photon + jets: photon p T + E T miss ≡ “pseudo E T miss ” ; small statistical uncertainties large corrections needed (MC input) ⇒ systematic uncertainties W+jets: lepton p T + E T miss ≡ “pseudo E T miss ” Good to have several methods! Different uncertainties and SUSY contaminations ATLAS: CERN-Open-2008-020 CMS PAS SUS-08-002 E T miss [GeV]