1. Search for Heavy Stable Charge Particles in pp collision at √s=7TeV Loïc Quertenmont on behalf of the HSCP group: T. Adams, G. Bruno, C. Carrillo, J. Chen, S. Cooper, C. Delaere, A. Giammanco, Y. Kubota, A. Meneguzzo, C. Nuttens, L. Quertenmont, A. Rizzi, P. Traczyk Université catholique de Louvain & FNRS Center for Particle Physics and Phenomenology (CP3) 20 July 2010

2. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch2 Outline Introduction Trigger and data samples Offline reconstruction Ionization energy loss Mass measurement Offline selection Background estimation Results Systematic errors NOTE IN SUPPORT: CMS AN 2010/053 (MC) CMS AN 2010/194 (Data)

3. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch3 Analysis overview Heavy Stable Charged Particles (HSCP) are predicted in a number of models HSCP Signature High momentum High ionization High time of flight (not used at this stage) Muon-like but R-Hadrons do nuclear interactions and may change their charge Analyses Muon Id + dE/dx in silicon strip tracker Lepton-like and R-hadrons without charge suppression dE/dx in silicon strip tracker R-hadrons with charge suppression

4. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch4 Trigger and datasets The analysis was updated from 67.9/nb (Approval) to 198/nb Figure of the presentation are with updated statistics Data JetMetTau + Mu Secondary Datasets (SD) (only unprescaled paths) Details at: https://twiki.cern.ch/twiki/bin/viewauth/CMS/HscpAnalysisICHEP https://twiki.cern.ch/twiki/bin/viewauth/CMS/HscpAnalysisICHEP Certified integrated luminosity: 198 nb -1 June14th ReReco + Prompt Reconstruction Main Trigger paths: Jet50U, Met45 Mu3, DoubleMu0 MC Minimum bias + QCDPt30 + QCDPt80 Same triggers as for Data.

5. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch5 Pt Stability Plot (Not in the PAS) We observed a jump in ratio for the last few runs…  few of our trigger go precaled! We had to repeat the whole analysis this morning removing the prescale trigger path After correction, everything is rather stable…

6. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch6 Ias Stability Plot (Not in the PAS)

7. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch7 Benchmark models used for exclusion Lepton-like MSSM staus on SPS Line 7 [100 - 300] GeV PYTHIA R-Hadrons Direct pair-production of stops PYTHIA and MadGraph K-factors from PROSPINO (NLO) Direct pair-production of gluinos PYTHIA K-factors from PROSPINO (NLO+NLL) Cross sections: [10 -3, 10 3 ] pb Hadronization performed by PYTHIA For gluinos : gluino-ball fraction = 10% R-Hadron interaction with matter simulated by Geant4 R.Mackeprang and A.Rizzi, hep-ph/0612161, Eur.Phys.J.C50(2007) p.353

8. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch8 Ionization energy loss dE/dx MPV estimator Harmonic-2 estimator: Well known model (BB)  Mass Reconstruction (details later) dE/dx discriminators Full use of charge information Tail prob. depends on the path-length ADC cut-off Optimal discrimination  candidate selection Test statistic f(P h ) Ph = Probability for a MIP to release as much or less charge than observed Modified Smirnov-Cramer-vonMises: Short pathlength (~0.3 mm) Long pathlength (~0.6 mm) Muons (5 GeV) Normalized Charge (ADC/mm)

9. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch9 Mass Reconstruction Approximate Bethe-Bloch Formula before minimum (0.2<β<0.9), few % agreement Extract parameters from a known mass line (Protons) Reverse the relation to compute the mass of any track from dE/dx Estimator and P. All documented in TRK-10-001 (plots already approved) PAS UNCHANGED UNCHANGED

10. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch1013/07/2010 - HSCP Meeting - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch10 Mass Reconstruction II There is a bias at high mass due to the ADC cut-off ADC Range is limited to [0,253] counts 254 indicates a charge in [254,1023] 255 indicates a charge above 1023 Second peak at lower mass also due to this effect… (>1 strip saturating / cluster) This effect has no impact on this analysis (counting experiment) This effect has no impact on this analysis (counting experiment) PAS

11. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch11 Cluster Cleaning Candidate Mass:162.13GeV P=91.12GeV Pt=38.44GeV NOH=6 NOM=4 dEdx=10.74 eta=1.51 Phi-0.10 Candidate Mass:136.74 GeV P=92.42GeV Pt=24.54GeV NOH7 NOM=4 dEdx=8.21 eta=2.00 Phi=3.14 charge[0]:61 charge[1]:254 charge[2]:254 charge[3]:254 charge[4]:254 charge[5]:254 charge[6]:238 charge[7]:40 charge[0]:12 charge[1]:29 charge[2]:114 charge[3]:254 charge[4]:254 charge[5]:254 charge[6]:254 charge[7]:254 charge[8]:104 charge[9]:18 charge[0]:19 charge[1]:215 charge[2]:164 charge[3]:64 charge[0]:14 charge[1]:93 charge[2]:33 charge[3]:81 charge[4]:12 charge[0]:22 charge[1]:178 charge[2]:214 charge[3]:109 charge[4]:254 charge[5]:58 charge[0]:111 charge[1]:15 charge[2]:10 charge[3]:107 More than 2 consecutive strips with comparable charge (cross-talk induced charge on neighboring strips ~10% ) Multiple Maximum Cluster Multiple Tracks in the same cluster

12. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch12 Cluster Cleaning II Cluster cleaning: discard clusters likely to be produced by overlapping tracks (e.g.multiple maxima) from the dE/dx computation dE/dx tail (data) highly reduced. No significant modification of the signal dE/dx distribution DATA Gluino200 PAS

13. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch13 Candidate pre-selection Track type: Track type: Tracker+Muon analysis: Inner track from « Global muons » and « Tracker muons » collections No inner track sharing allowed Tracker-Only analysis: Tracks from « general tracks » collection Track Quality: Track Quality: « High Purity » Pt>7.5GeV (Pt>5GeV in the frozen PAS) PtErr/Pt<15% |dZ|<2cm, |d0|<0.25cm (|dZ|<10cm,|d0|<1cm in the frozen PAS) #dE/dx measurements > 3

14. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch14 HSCP Selection Relatively good agreement in shape Data – MC MC normalized to data Efficient discrimination of HSCP (against MIPs) with cuts on Pt and dE/dx Signal for tracker-only and tracker+muon plots are different! Tracker+Muon Tracker+Muon Tracker-Only Tracker-Only PAS Bias in shape apearing because of few MC-MB events with very high weight… Integral are compatible…

15. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch15 Improved selection I Tails of pt and dE/dx distribution for MIPs highly dependent of #hits of tracks and moderately on the η region dE/dx vs #Hits dE/dx vs Eta for 15Hit tracks Tracker+Muon Tracker+Muon Tracker-Only Tracker-Only PAS Change in slope is observed in all η region… Is due to the loose of discrimination power when MIPs have saturating strip, that is why the slope is lower after some point.

16. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch16 Improved selection II Sensitivity to tracks with few hits desirable Neutral HSCP at production HSCP with charge suppression Split candidates into sub-samples (#dE/dx hits and η regions) 0.0<|η|<0.5; 0.5<|η|<1.0; 1.0<|η|<1.5; 1.5<|η|<2.0; 2.0<|η|<2.5 Impose cuts that provide a fixed selection efficiency in each subsample Improvements Signal-to-noise ratio a factor 2 improvement is observed in the present study (67.9/nb) a factor 10 improvement was seen in MC at 10/pb (AN2010/053) Reliability of background prediction (discussed later)

17. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch17 Data-driven background estimation dE/dx discriminator distribution for pre-selected tracks Control (7.5 20GeV) No significant correlation between dE/dx and pt PAS

18. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch18 Data-driven background estimation II Use « ABCD » method to estimate background in the signal region # entries in signal region D = (B*C)/A Can also predict shape of mass distribution Move to p – dE/dx estimator plane Procedure is applied in every #Hit/η sub-samples and results are combined. A A Pre-Selected track failing the Pt and I cuts. B B Pre-Selected track failing the Pt cut but passing the I cut. C C Pre-Selected track failing the I cuts but passing the Pt cut. D D Pre-Selected track passing the Pt and I cut. PtCut ICut

19. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch19 Systematic error on expected background Main source is residual correlation between momentum and ionization energy loss measurements Low momentum (pt>7.5GeV) particles are mainly used to model the ionization energy loss distribution Relativistic rise in Bethe-Bloch is not accounted for in background prediction Expected background in signal region is corrected by comparing the expected background and the observation in the control region for all (loose) selections with a background efficiency lower than 10 -3 and at least 10 entries. Get distribution of correction factors for tested selections Mean of the distribution is the correction factor applied to the expected background Tracker+muon: 1.10 Tracker-only: 1.25 Tracker+muon: 1.10 Tracker-only: 1.25 2 x RMS assigned as systematic uncertainty on expected background in signal region 2 is a safety factor for the extrapolation of the correction to the signal region Tracker+muon: 2x11%Tracker-only: 2x14% Tracker+muon: 2x11%Tracker-only: 2x14%

20. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch20 Observed mass spectrum for loose selection Loose Selection ε Pt =10 -2.0 ε I =10 -2.0 Loose Selection ε Pt =10 -1.0 ε I =10 -1.5 PAS

21. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch21 dE/dx vs P for Tracks in LOOSE selection High mass (M>300) candidate have a relatively small ionization and a large impulsion, not a strong candidate All points with Ih>5MeV/cm are small tracks (<5hits) at high eta, with generally few of their SiStrip clusters having at least one saturating strip. None of them are real candidates, but well expected background.

22. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch22 Search Strategy Define a mass region for signal search: [75,1200] GeV Choose optimal selection from data-driven background prediction (~0.05 events) and simulated signal samples Count events in signal region If compatible with expected background, set 95% C.L. upper limit on cross section for benchmark signals Statistical methods Full Bayesian method (code courtesy of G.Landsberg) with lognormal prior for integration over nuisance parameters

23. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch23 Variation of expected signal with selection Signal not strongly affected by the chosen selection Signal not strongly affected by the chosen selection Chosen selection: ~0.05 expected bckg events Chosen selection: ~0.05 expected bckg events PAS

24. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch24 Observed mass spectrum Signal region: ~0.05 events expected for both analyses. No events are observed for chosen selections.

25. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch25 Tracker+Muon analysis results Gluino masses < 284GeV/c² are excluded (under 15% TH-uncertainty Hypothesis) Systematic errors already incorporated in Cross-Section limits PAS

26. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch26 Tracker-Only analysis results PAS Gluino masses < 271GeV/c² are excluded (under 15% TH-uncertainty Hypothesis) Systematic errors already incorporated in Cross-Section limits

27. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch27 Systematic errors Signal Acceptance

28. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch28 Future evolution of the analysis 1pb -1 1pb -1 Increase of integrated luminosity by 1-2 orders of magnitude Can keep the expected background unchanged (~0.05 Events) Signal efficiency will only be reduced by ~35% (for lightest HSCPs) 10pb -1 10pb -1 A simulation study showed that tracker-only analysis is at the limit of moving to peak search for the lowest masses There was no cluster cleaning 100pb -1 100pb -1 Most likely can go on with counting experiment up to 100pb -1 even for Tk-only analysis.

29. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch29 Conclusions Improved selection in #Hits/η sub-samples used. Cluster cleaning algorithm has been deployed and used Rescaled data driven background prediction in good agreement with observed data. Cross section upper limits for several models set with a counting experiment technique using two different analysis scheme: tracker-only and tracker+muon Tracker-only analysis excluded Gluino masses below 284 GeV/c² under the 15% theoretical uncertainty hypothesis Tracker-muon analysis excluded Gluino masses below 271 GeV/c² under the 15% theoretical uncertainty hypothesis Tevatron results: Gluino: No limit set: but could have excluded up to 350GeV/c² Stop: masses below 250GeV/c² and XSection > ~0.1/Pb are excluded by CDF We thank our ARC (Tulika Bose, Franco Ligabue, Luc Pape, Shahram Rahatlou) and other contributors for all the positive suggestions we’ve got.

30. Backups Slides of last Friday Approval

31. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch31 Analysis overview Heavy Stable Charged Particles (HSCP) are predicted in a number of models HSCP Signature High momentum High ionization High time of flight (not used at this stage) Muon-like but R-Hadrons do nuclear interactions and may change their charge Analyses Muon Id + dE/dx in silicon strip tracker Lepton-like and R-hadrons without charge suppression dE/dx in silicon strip tracker R-hadrons with charge suppression

32. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch32 Trigger and datasets The analysis was updated from 35.9/nb (frozen PAS) to 67.9/nb Figure of the presentation are with updated statistics Data JetMetTau + Mu Secondary Datasets (SD) (only unprescaled paths) Details at: https://twiki.cern.ch/twiki/bin/viewauth/CMS/HscpAnalysisICHEP https://twiki.cern.ch/twiki/bin/viewauth/CMS/HscpAnalysisICHEP Certified integrated luminosity: 67.9 nb -1 June14th ReReco + Prompt Reconstruction Main Trigger paths: Jet30U, Met45 Mu3, DoubleMu0 MC Minimum bias + QCDPt30 + QCDPt80 Same triggers as for Data.

33. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch33 Benchmark models used for exclusion Lepton-like MSSM staus on SPS Line 7 [100 - 300] GeV PYTHIA R-Hadrons Direct pair-production of stops PYTHIA and MadGraph K-factors from PROSPINO (NLO) Direct pair-production of gluinos PYTHIA K-factors from PROSPINO (NLO+NLL) Cross sections: [10 -3, 10 3 ] pb Hadronization performed by PYTHIA For gluinos : gluino-ball fraction = 10% R-Hadron interaction with matter simulated by Geant4 R.Mackeprang and A.Rizzi, hep-ph/0612161, Eur.Phys.J.C50(2007) p.353

34. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch34 Ionization energy loss dE/dx MPV estimator Harmonic-2 estimator: Well known model (BB)  Mass Reconstruction (details later) dE/dx discriminators Full use of charge information Tail prob. depends on the path-length ADC cut-off Optimal discrimination  candidate selection Test statistic f(P h ) Ph = Probability for a MIP to release as much or less charge than observed Modified Smirnov-Cramer-vonMises: Short pathlength (~0.3 mm) Long pathlength (~0.6 mm) Muons (5 GeV) Normalized Charge (ADC/mm)

35. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch35 Mass Reconstruction Approximate Bethe-Bloch Formula before minimum (0.2<β<0.9), few % agreement Extract parameters from a known mass line (Protons) Reverse the relation to compute the mass of any track from dE/dx Estimator and P. All documented in TRK-10-001 (plots already approved) PAS

36. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch3613/07/2010 - HSCP Meeting - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch36 Mass Reconstruction II There is a bias at high mass due to the ADC cut-off ADC Range is limited to [0,253] counts 254 indicates a charge in [254,1023] 255 indicates a charge above 1023 This effect has no impact on this analysis (counting experiment) This effect has no impact on this analysis (counting experiment) PAS

37. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch37 Cluster Cleaning Candidate Mass:162.13GeV P=91.12GeV Pt=38.44GeV NOH=6 NOM=4 dEdx=10.74 eta=1.51 Phi-0.10 Candidate Mass:136.74 GeV P=92.42GeV Pt=24.54GeV NOH7 NOM=4 dEdx=8.21 eta=2.00 Phi=3.14 charge[0]:61 charge[1]:254 charge[2]:254 charge[3]:254 charge[4]:254 charge[5]:254 charge[6]:238 charge[7]:40 charge[0]:12 charge[1]:29 charge[2]:114 charge[3]:254 charge[4]:254 charge[5]:254 charge[6]:254 charge[7]:254 charge[8]:104 charge[9]:18 charge[0]:19 charge[1]:215 charge[2]:164 charge[3]:64 charge[0]:14 charge[1]:93 charge[2]:33 charge[3]:81 charge[4]:12 charge[0]:22 charge[1]:178 charge[2]:214 charge[3]:109 charge[4]:254 charge[5]:58 charge[0]:111 charge[1]:15 charge[2]:10 charge[3]:107 More than 2 consecutive strips with comparable charge (cross-talk induced charge on neighboring strips ~10% ) Multiple Maximum Cluster

38. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch38 Cluster Cleaning II Cluster cleaning: discard clusters likely to be produced by overlapping tracks (e.g.multiple maxima) from the dE/dx computation dE/dx tail (data) highly reduced. No significant modification of the signal dE/dx distribution DATA Gluino200 PAS

39. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch39 Candidate pre-selection Track type: Track type: Tracker+Muon analysis: Inner track from « Global muons » and « Tracker muons » collections No inner track sharing allowed Tracker-Only analysis: Tracks from « general tracks » collection Track Quality: Track Quality: « High Purity » Pt>7.5GeV (Pt>5GeV in the frozen PAS) PtErr/Pt<15% |dZ|<2cm, |d0|<0.25cm (|dZ|<10cm,|d0|<1cm in the frozen PAS) #dE/dx measurements > 3

40. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch40 HSCP Selection Relatively good agreement in shape Data – MC MC normalized to data Efficient discrimination of HSCP (against MIPs) with cuts on Pt and dE/dx Signal for tracker-only and tracker+muon plots are different! Tracker+MuonTracker+Muon Tracker-OnlyTracker-Only PAS

41. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch41 Improved selection I Tails of pt and dE/dx distribution for MIPs highly dependent of #hits of tracks and moderately on the η region dE/dx vs #Hits dE/dx vs Eta for 15Hit tracks Tracker+Muon Tracker+MuonTracker-Only Tracker-Only PAS

42. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch42 Improved selection II Sensitivity to tracks with few hits desirable Neutral HSCP at production HSCP with charge suppression Split candidates into sub-samples (#dE/dx hits and η regions) 0.0<|η|<0.5; 0.5<|η|<1.0; 1.0<|η|<1.5; 1.5<|η|<2.0; 2.0<|η|<2.5 Impose cuts that provide a fixed selection efficiency in each subsample Improvements Signal-to-noise ratio a factor 2 improvement is observed in the present study (67.9/nb) a factor 10 improvement was seen in MC at 10/pb (AN2010/053) Reliability of background prediction (discussed later)

43. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch43 Data-driven background estimation dE/dx discriminator distribution for pre-selected tracks Control (7.5 20GeV) No significant correlation between dE/dx and pt PAS

44. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch44 Data-driven background estimation II Use « ABCD » method to estimate background in the signal region # entries in signal region D = (B*C)/A Can also predict shape of mass distribution Move to p – dE/dx estimator plane Procedure is applied in every #Hit/η sub-samples and results are combined. A A Pre-Selected track failing the Pt and I cuts. B B Pre-Selected track failing the Pt cut but passing the I cut. C C Pre-Selected track failing the I cuts but passing the Pt cut. D D Pre-Selected track passing the Pt and I cut. PtCut ICut

45. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch45 Systematic error on expected background Main source is residual correlation between momentum and ionization energy loss measurements Low momentum (pt>7.5GeV) particles are mainly used to model the ionization energy loss distribution Relativistic rise in Bethe-Bloch is not accounted for in background prediction Expected background in signal region is corrected by comparing the expected background and the observation in the control region for all (loose) selections with a background efficiency lower than 10 -3 and at least 10 entries. Get distribution of correction factors for tested selections Mean of the distribution is the correction factor applied to the expected background Tracker+muon: 1.10 Tracker-only: 1.25 Tracker+muon: 1.10 Tracker-only: 1.25 2 x RMS assigned as systematic uncertainty on expected background in signal region 2 is a safety factor for the extrapolation of the correction to the signal region Tracker+muon: 2x11%Tracker-only: 2x14% Tracker+muon: 2x11%Tracker-only: 2x14%

46. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch46 Observed mass spectrum for loose selection Loose Selection ε Pt =10 -2.0 ε I =10 -2.0 Loose Selection ε Pt =10 -1.0 ε I =10 -1.5 PAS

47. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch47 Search Strategy Define a mass region for signal search: [75,1200] GeV Choose optimal selection from data-driven background prediction (~0.05 events) and simulated signal samples Count events in signal region If compatible with expected background, set 95% C.L. upper limit on cross section for benchmark signals Statistical methods Full Bayesian method (code courtesy of G.Landsberg) with lognormal prior for integration over nuisance parameters

48. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch48 Variation of expected signal with selection Signal not strongly affected by the chosen selection Signal not strongly affected by the chosen selection Chosen selection: ~0.05 expected bckg events Chosen selection: ~0.05 expected bckg events PAS

49. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch49 Choice of selection - Tk+Muon Analysis X Axis: Selection efficiency obtained on pre-selected tracks with the Pt cut alone X Axis: Selection efficiency obtained on pre-selected tracks with the Pt cut alone Axis: Selection efficiency obtained on preselected tracks with the dE/dx cut alone Y Axis: Selection efficiency obtained on preselected tracks with the dE/dx cut alone Left: Z Axis: Predicted number of background entries in the signal region. Right: Z Axis: Expected (MC) number of stau100 in the signal region. 0.1-0.01 events ? PAS ?

50. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch50 Choice of selection – Tk-Only Analysis X Axis: Selection efficiency obtained on pre-selected tracks with the Pt cut alone X Axis: Selection efficiency obtained on pre-selected tracks with the Pt cut alone Axis: Selection efficiency obtained on preselected tracks with the dE/dx cut alone Y Axis: Selection efficiency obtained on preselected tracks with the dE/dx cut alone Left: Z Axis: Predicted number of background entries in the signal region. Right: Z Axis: Expected (MC) number of stau100 in the signal region. 0.1-0.01 events ? PAS ?

51. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch51 Observed mass spectrum Signal region: ~0.05 events expected for both analyses. No events are observed for chosen selections.

52. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch52 Tracker+Muon analysis results Gluino masses < 255GeV/c² are excluded Systematic errors already incorporated in Cross-Section limits Theoretical x-section uncertainty not added yet (working on it) Expected to be less than 15%  Gluino masses < 246GeV/c² excluded PAS

53. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch53 Tracker-Only analysis results Gluino masses < 247GeV/c² are excluded Systematic errors already incorporated in Cross-Section limits Theoretical x-section uncertainty not added yet (working on it) Expected to be less than 15%  Gluino masses < 236GeV/c² excluded PAS

54. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch54 Systematic errors Signal Acceptance

55. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch55 Future evolution of the analysis 1pb -1 1pb -1 Increase of integrated luminosity by 1-2 orders of magnitude Can keep the expected background unchanged (~0.05 Events) Signal efficiency will only be reduced by ~35% (for lightest HSCPs) 10pb -1 10pb -1 A simulation study showed that tracker-only analysis is at the limit of moving to peak search for the lowest masses There was no cluster cleaning 100pb -1 100pb -1 Most likely can go on with counting experiment up to 100pb -1 even for Tk-only analysis.

56. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch56 Conclusions Improved selection in #Hits/η sub-samples used. Cluster cleaning algorithm has been deployed and used Rescaled data driven background prediction in good agreement with observed data. Cross section upper limits for several models set with a counting experiment technique using two different analysis scheme: tracker-only and tracker+muon Tracker-only analysis excluded Gluino masses below 247 (236) GeV/c² if no (15%) theoretical uncertainty is considered Tracker-muon analysis excluded Gluino masses below 255 (246) GeV/c² if no (15%) theoretical uncertainty is considered Tevatron results: Gluino: No limit set Stop: masses below 250GeV/c² and XSection > ~0.1/Pb are excluded by CDF We thank our ARC (Tulika Bose, Franco Ligabue, Luc Pape, Shahram Rahatlou) and other contributors for all the positive suggestions we’ve got.

57. Backups

58. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch58 PLOTS FROM THE FROZEN PAS Double counting in the MC Sample between QCD30 and Minbias PTHat>30GeV Corrected in the updated results

59. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch59 PLOTS FROM THE FROZEN PAS

60. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch60 PLOTS FROM THE FROZEN PAS Pt plot replaced by de/dx plot in eta regions

61. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch61 PLOTS FROM THE FROZEN PAS

62. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch62 PLOTS FROM THE FROZEN PAS

63. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch63 PLOTS FROM THE FROZEN PAS

64. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch64 PLOTS FROM THE FROZEN PAS

65. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch65 PLOTS FROM THE FROZEN PAS

66. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch66 PLOTS FROM THE FROZEN PAS

67. Backups

68. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch68 Acceptance uncertainty Trigger Efficiency: JetMET dominated by JES (10%) Moving the Trigger Tresholds by 10%  acceptance on signal of max 10% Muon: disagrement between data and MC is observed (+10% in MC) Uncertainty on slow particle  assuming 15% drop in muon trigger, results in a 5% drop in the total trigger efficiency Considered a 15% uncertainty on trigger Offline tracks (<5%) and muon (~5%) reconstruction efficiency Momentum scale (<5%), assuming 5%  <5% to the acceptance dE/dx scale (<5%), assuming 5%  ~3% to the acceptance (8% for stau100)

69. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch69 Selection Tracker-Only Loose DATA SELECTION DETAILS -------------------- #Events weighted (unweighted) = 1.34E+06 (1.34E+06) #Triggered Events weighted (unweighted) = 1.34E+06 (1.34E+06) Eff=1.000E+00 (1.000E+00) #Tracks weighted (unweighted) = 2.32E+06 (2.32E+06) #Tracks passing Hits cuts weighted (unweighted) = 2.18E+06 (2.18E+06) Eff=9.398E-01 (9.398E-01) #Tracks passing Qual cuts weighted (unweighted) = 2.15E+06 (2.15E+06) Eff=9.857E-01 (9.857E-01) #Tracks passing Chi2 cuts weighted (unweighted) = 2.15E+06 (2.15E+06) Eff=1.000E+00 (1.000E+00) #Tracks passing PtErr cuts weighted (unweighted) = 2.14E+06 (2.14E+06) Eff=9.947E-01 (9.947E-01) #Tracks passing dZ cuts weighted (unweighted) = 1.66E+06 (1.66E+06) Eff=7.739E-01 (7.739E-01) #Tracks passing dXY cuts weighted (unweighted) = 1.60E+06 (1.60E+06) Eff=9.676E-01 (9.676E-01) #Tracks passing Min Pt cuts weighted (unweighted) = 1.60E+06 (1.60E+06) Eff=1.000E+00 (1.000E+00) #Tracks passing Min I cuts weighted (unweighted) = 1.60E+06 (1.60E+06) Eff=1.000E+00 (1.000E+00) #Tracks passing Basic cuts weighted (unweighted) = 1.60E+06 (1.60E+06) Eff=6.900E-01 (6.900E-01) #Tracks passing Pt cuts weighted (unweighted) = 1.59E+04 (1.59E+04) Eff=9.912E-03 (9.912E-03) #Tracks passing I cuts weighted (unweighted) = 2.15E+02 (2.15E+02) Eff=1.354E-02 (1.354E-02) #Tracks passing selection weighted (unweighted) = 2.15E+02 (2.15E+02) Eff=9.262E-05 (9.262E-05) MC TRUTH SELECTION DETAILS -------------------- #Events weighted (unweighted) = 1.25E+06 (1.95E+06) #Triggered Events weighted (unweighted) = 1.25E+06 (1.95E+06) Eff=1.000E+00 (1.000E+00) #Tracks weighted (unweighted) = 2.21E+06 (6.13E+06) #Tracks passing Hits cuts weighted (unweighted) = 2.06E+06 (5.89E+06) Eff=9.334E-01 (9.606E-01) #Tracks passing Qual cuts weighted (unweighted) = 2.02E+06 (5.84E+06) Eff=9.803E-01 (9.906E-01) #Tracks passing Chi2 cuts weighted (unweighted) = 2.02E+06 (5.84E+06) Eff=1.000E+00 (1.000E+00) #Tracks passing PtErr cuts weighted (unweighted) = 2.01E+06 (5.80E+06) Eff=9.947E-01 (9.942E-01) #Tracks passing dZ cuts weighted (unweighted) = 1.96E+06 (5.63E+06) Eff=9.750E-01 (9.693E-01) #Tracks passing dXY cuts weighted (unweighted) = 1.90E+06 (5.45E+06) Eff=9.711E-01 (9.693E-01) #Tracks passing Min Pt cuts weighted (unweighted) = 1.90E+06 (5.45E+06) Eff=1.000E+00 (1.000E+00) #Tracks passing Min I cuts weighted (unweighted) = 1.90E+06 (5.45E+06) Eff=1.000E+00 (1.000E+00) #Tracks passing Basic cuts weighted (unweighted) = 1.90E+06 (5.45E+06) Eff=8.618E-01 (8.889E-01) #Tracks passing Pt cuts weighted (unweighted) = 1.78E+04 (1.53E+05) Eff=9.350E-03 (2.803E-02) #Tracks passing I cuts weighted (unweighted) = 3.22E+02 (3.23E+03) Eff=1.808E-02 (2.114E-02) #Tracks passing selection weighted (unweighted) = 3.22E+02 (3.23E+03) Eff=1.457E-04 (1.464E-03)

70. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch70 Selection Tracker+Muon Loose DATA SELECTION DETAILS -------------------- #Events weighted (unweighted) = 1.34E+06 (1.34E+06) #Triggered Events weighted (unweighted) = 1.34E+06 (1.34E+06) Eff=1.000E+00 (1.000E+00) #Tracks weighted (unweighted) = 1.37E+05 (1.37E+05) #Tracks passing Hits cuts weighted (unweighted) = 1.34E+05 (1.34E+05) Eff=9.791E-01 (9.791E-01) #Tracks passing Qual cuts weighted (unweighted) = 1.34E+05 (1.34E+05) Eff=9.958E-01 (9.958E-01) #Tracks passing Chi2 cuts weighted (unweighted) = 1.34E+05 (1.34E+05) Eff=1.000E+00 (1.000E+00) #Tracks passing PtErr cuts weighted (unweighted) = 1.33E+05 (1.33E+05) Eff=9.974E-01 (9.974E-01) #Tracks passing dZ cuts weighted (unweighted) = 1.04E+05 (1.04E+05) Eff=7.775E-01 (7.775E-01) #Tracks passing dXY cuts weighted (unweighted) = 1.02E+05 (1.02E+05) Eff=9.848E-01 (9.848E-01) #Tracks passing Min Pt cuts weighted (unweighted) = 1.02E+05 (1.02E+05) Eff=1.000E+00 (1.000E+00) #Tracks passing Min I cuts weighted (unweighted) = 1.02E+05 (1.02E+05) Eff=1.000E+00 (1.000E+00) #Tracks passing Basic cuts weighted (unweighted) = 1.02E+05 (1.02E+05) Eff=7.445E-01 (7.445E-01) #Tracks passing Pt cuts weighted (unweighted) = 1.01E+04 (1.01E+04) Eff=9.881E-02 (9.881E-02) #Tracks passing I cuts weighted (unweighted) = 3.20E+02 (3.20E+02) Eff=3.174E-02 (3.174E-02) #Tracks passing selection weighted (unweighted) = 3.20E+02 (3.20E+02) Eff=2.335E-03 (2.335E-03) MC TRUTH SELECTION DETAILS -------------------- #Events weighted (unweighted) = 1.25E+06 (1.95E+06) #Triggered Events weighted (unweighted) = 1.25E+06 (1.95E+06) Eff=1.000E+00 (1.000E+00) #Tracks weighted (unweighted) = 1.38E+05 (1.31E+05) #Tracks passing Hits cuts weighted (unweighted) = 1.36E+05 (1.28E+05) Eff=9.816E-01 (9.761E-01) #Tracks passing Qual cuts weighted (unweighted) = 1.35E+05 (1.27E+05) Eff=9.945E-01 (9.946E-01) #Tracks passing Chi2 cuts weighted (unweighted) = 1.35E+05 (1.27E+05) Eff=1.000E+00 (1.000E+00) #Tracks passing PtErr cuts weighted (unweighted) = 1.35E+05 (1.26E+05) Eff=9.984E-01 (9.929E-01) #Tracks passing dZ cuts weighted (unweighted) = 1.33E+05 (1.23E+05) Eff=9.859E-01 (9.731E-01) #Tracks passing dXY cuts weighted (unweighted) = 1.32E+05 (1.20E+05) Eff=9.909E-01 (9.799E-01) #Tracks passing Min Pt cuts weighted (unweighted) = 1.32E+05 (1.20E+05) Eff=1.000E+00 (1.000E+00) #Tracks passing Min I cuts weighted (unweighted) = 1.32E+05 (1.20E+05) Eff=1.000E+00 (1.000E+00) #Tracks passing Basic cuts weighted (unweighted) = 1.32E+05 (1.20E+05) Eff=9.521E-01 (9.191E-01) #Tracks passing Pt cuts weighted (unweighted) = 9.85E+03 (3.58E+04) Eff=7.484E-02 (2.975E-01) #Tracks passing I cuts weighted (unweighted) = 3.84E+02 (1.93E+03) Eff=3.892E-02 (5.394E-02) #Tracks passing selection weighted (unweighted) = 3.84E+02 (1.93E+03) Eff=2.773E-03 (1.398E-02)

71. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch71 Selection: #dEdx Hits

72. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch72 Selection: Quality

73. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch73 Selection: PtErr

74. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch74 Selection: dZ

75. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch75 Selection: dXY

76. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch76 Selection: Pt

77. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch77 Selection: I

78. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch78 Observed mass spectrum for loose selection Loose Selection ε Pt =10 -2.0 ε I =10 -2.0 Loose Selection ε Pt =10 -1.0 ε I =10 -1.5 PAS M>150GeV  Exp=12+-3.36 Obs=10 (<1Sigma) M>150GeV  Exp=42+-9.2 Obs=58 (1.7Sigma)

79. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch7913/07/2010 - HSCP Meeting - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch79 Trigger Efficiencies

80. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch8013/07/2010 - HSCP Meeting - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch80 Trigger Efficiencies

81. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch8113/07/2010 - HSCP Meeting - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch81 Trigger Efficiencies

82. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch8213/07/2010 - HSCP Meeting - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch82 Trigger Efficiency vs Beta STOP 130

83. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch8313/07/2010 - HSCP Meeting - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch83 Trigger Efficiency vs Beta STAU 100

84. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch8413/07/2010 - HSCP Meeting - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch84 Rescale prediction (tracker+muon)

85. 20/07/2010 - EXO-10-004 Update - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch8513/07/2010 - HSCP Meeting - UCL - CP3Loïc Quertenmont - loic.quertenmont@cern.ch85 Rescale prediction (tracker-only)