Lepton ID at the LHC S Rajagopalan February

  • Slides: 25
Download presentation
Lepton ID at the LHC S. Rajagopalan February 5, 2005 TEV 4 LHC workshop

Lepton ID at the LHC S. Rajagopalan February 5, 2005 TEV 4 LHC workshop

LHC turn on T Summer 2007 q Colliding beams in machine q L ~

LHC turn on T Summer 2007 q Colliding beams in machine q L ~ Can expect 1031 during the early months q 3 month shutdown (Fall 07) followed by ~ 7 month of physics run q Can expect L to steadily increase to 2 x 1033 T Can expect 1 to 10 fb-1 per experiment in the first year q Though a lot of uncertainties in schedule and luminosity T Commissioning Phase: (starting summer 2005) q Sub-System Calibration + Cosmic Ray Commissioning q April 2007 : single beam in machine z S. Rajagopalan Beam Gas, Beam Halo TEV 4 LHC Workshop, Feb 5, 2005 2

Preparing for Day 1 T Will the detectors be operational? q Understand calibrate the

Preparing for Day 1 T Will the detectors be operational? q Understand calibrate the detector response q Validate SM signatures (Z ℓℓ, W ℓn, …) q Which will allow us to prepare the groundwork for discovery physics T As with any previous experiments, can expect to spend the first few months q understanding the detector response q optimization of reconstruction algorithms, calibration/alignment T It is here we can benefit from the Tevatron experience q Understand what chaos we can expect on Day 1 q Help us ensure that we go in prepared. S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 3

Initial layout Impact on physics visible but acceptable S. Rajagopalan TEV 4 LHC Workshop,

Initial layout Impact on physics visible but acceptable S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 4

Muon Performance ATLAS Physics TDR (1999) CMS: N. Neumeister CHEP 04 Inner tracker standalone

Muon Performance ATLAS Physics TDR (1999) CMS: N. Neumeister CHEP 04 Inner tracker standalone Muon spectrom. standalone Muon Spectrometer resolution dominates for PT > 100 Ge. V/c Resolution fairly constant over whole eta range • Coverage | | < 2. 7 S. Rajagopalan Silicon Tracker resolution dominates for all PT Excellent p. T resolution in barrel region, worse in endcap • Coverage | | < 2. 4 TEV 4 LHC Workshop, Feb 5, 2005 5

H ZZ* 4 m From “Physics at LHC 2004”, Vienna, July 2004 ATLAS CMS

H ZZ* 4 m From “Physics at LHC 2004”, Vienna, July 2004 ATLAS CMS Signal(m. H=130 Ge. V) ZZ 4 m ZZ 2 m 2 t ttbar Zbb S. Rajagopalan Mass resolution for MH = 130 Ge. V = 1. 6 Ge. V (ATLAS) = 1. 3 Ge. V (CMS) TEV 4 LHC Workshop, Feb 5, 2005 6

EM Performance Requirement H gg used as benchmark to assess EM Calorimeter performance ~

EM Performance Requirement H gg used as benchmark to assess EM Calorimeter performance ~ 1% mass resolution to observe signal over gg continuum Constant term in energy resolution < 0. 7% CMS, m. H = 130 Ge. V, 100 fb-1 Mass resolution ~ 700 Me. V (at high luminosity) 5 s significance with L = 30 fb-1 S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 7

Calorimeter performance Resolution (CMS) Scintillating Crystal EM calorimeter, 75000 Pb. W 04 crystals (0.

Calorimeter performance Resolution (CMS) Scintillating Crystal EM calorimeter, 75000 Pb. W 04 crystals (0. 0175 x 0. 0175) (s/E)2 = (2. 7%/√E)2 + (0. 55%/E)2 + (0. 155)2 @ =0 (s/E)2 = (5. 7%/√E)2 + (0. 55%/E)2 + (0. 205)2 @ =2 S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 8

Inter Calibration of CMS crystals To achieve ~ 0. 5% constant term, Long-term :

Inter Calibration of CMS crystals To achieve ~ 0. 5% constant term, Long-term : Must rely on E/p from W en uncertainty in tracking material (particularly endcaps) studies ongoing to deal with brem effects Early day running: Intercalibration with min bias events or di-jet triggers (phi uniformity) ~ 2 – 3% precision in few hours + Intercalibration of eta-rings with Z ee ~ 1% precision in 1 day running. S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 9

ATLAS EM Calorimeter T Pb-LAr Accordion sampling calorimeter q Completed, Barrel calorimeter in the

ATLAS EM Calorimeter T Pb-LAr Accordion sampling calorimeter q Completed, Barrel calorimeter in the pit, Endcap under tests q Extensive tests of modules at test-beam q Commissioning in pit to commence mid-2005 4 Longitudinal Samplings: ( < 3. 2) PS (0. 025 x 0. 1); Strips (0. 003 x 0. 1) Middle(. 025 x. 025), Back(. 05 x. 025) S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 10

Linearity (Erec/Ebeam) ATLAS EM Resolution/Linearity (test-beam data) EMB Data ( =0. 48) a =

Linearity (Erec/Ebeam) ATLAS EM Resolution/Linearity (test-beam data) EMB Data ( =0. 48) a = 8. 95% c = 0. 33% ± 0. 1% E(Ge. V) /E (%) ( =1. 9) • Data o MC a = 10. 35% c = 0. 27% Beam energy (Ge. V) EMEC E(Ge. V) S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 11

Intercalibration with Z ee T Constant term c = c. LR < 0. 7%

Intercalibration with Z ee T Constant term c = c. LR < 0. 7% (goal) T c. L = Local contribution to constant term < 0. 5% q (variation in D x Df = 0. 2 x 0. 4, measured in TB) T c. LR = Long range variations corrected with Z ee q ~250 electrons in each unit of D x Df = 0. 2 x 0. 4 q 105 Z ee events (few days at 1 Hz) q to achieve c. LR < 0. 4% T Pessimistic scenario : constant term ~ 2% q H gg significance at m. H = 115 Ge. V degraded by 25% q Need 50% more L to recover the significance. S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 12

Examples of ongoing studies (to understand detector effects) Effect of dead channels Effect of

Examples of ongoing studies (to understand detector effects) Effect of dead channels Effect of variation in lead thickness 1% variation 0. 6% drop in response Measured variation s = 9 mm Translates to < 0. 2% effect on constant term Number of dead channels in calorimeter < 0. 1 % S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 13

High PT electron id : efficiency/rejection Several multi-variate analysis being explored Simple Cuts, Likelihood

High PT electron id : efficiency/rejection Several multi-variate analysis being explored Simple Cuts, Likelihood methods, neural nets, … Table below shows studies based on simple cuts Electron Efficiency ~ 70% with stringent cuts, Jet Rejection Factors ~ 105 (large systematic errors, sensitive to fragmentation models) S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 14

SOFT ELECTRONS # TR hits electrons pions Soft e (p. T>2 Ge. V) Eff=90%

SOFT ELECTRONS # TR hits electrons pions Soft e (p. T>2 Ge. V) Eff=90% Rej=211 ± 20 Et(calo)/pt fraction of E in 3 rd sampling diff between shower and impact position shower isolation log( ПPDF(e)/ПPDF(h)) S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 15

Pointing with Photons CMS relies on other charged tracks in event or converted photons

Pointing with Photons CMS relies on other charged tracks in event or converted photons ATLAS : Relies on extrapolation using first (D = 0. 003) and second sampling (D = 0. 025) DQ from χ0 Ĝg for ct = 1. 1 km 180 will decay in tracker volume for 10 fb-1 82% efficiency for DQ > 5 s Allows to set limit of ct = 100 km with 30 fb-1 (If no non-pointing photons are found) S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 16

Impact on H 4ℓ Mass resolution at 1033 = 1. 54 Ge. V Resolution

Impact on H 4ℓ Mass resolution at 1033 = 1. 54 Ge. V Resolution at 1034 = 1. 87 Ge. V Acceptance in ± 2 s window ~ 85% Efficiency of four electron id ~ 69% Average Eff per electron ~ 91% S. Rajagopalan 5 s discovery potential for m. H = 200 Ge. V Requires good quality E, p measurement in ECAL and Tracker to limit tails and improve s/m ~ 1% TEV 4 LHC Workshop, Feb 5, 2005 17

Tau Identification T Three primary cuts are used to identify tau’s: q REM :

Tau Identification T Three primary cuts are used to identify tau’s: q REM : Radius computed using EM cells q DET 12 : ET in 0. 1 < DR < 0. 2 q NTR : # of tracks with PT > 2 Ge. V in DR < 0. 3 Identification Efficiency and background rejection: S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 18

Tau Identification (2) Jet Rejection vs tau Efficiency Dependence on PT mainly via REM

Tau Identification (2) Jet Rejection vs tau Efficiency Dependence on PT mainly via REM cut Minimal change at high luminosity Reconstructed Z tt jet vt ℓ vℓ vt mass spectrum 1300 events with 10 fb-1 (B = 6%) Select high PT Z (1. 8<DF<2. 7, 3. 6<DF<4. 5) to use Missing ET for reconstruction S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 19

Tau Identification (3) Other approaches being considered for tau identification Work begun on Likelihood

Tau Identification (3) Other approaches being considered for tau identification Work begun on Likelihood approach : Seems to give factor of 2 to 5 improvement in rejection over simple cuts Caveat: Different sample comparison, more input variables (impact parameter), large errors (30%) (Other approaches: Neural network, track based approach + energy flow being investigated) S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 20

Tau Identification (4) impact on VBF H tt TLikelihood approach: q L > 1

Tau Identification (4) impact on VBF H tt TLikelihood approach: q L > 1 give single tau efficiency of 70% with background of ~ x 100 q 5 s significance for VBF H tt expected with 30 fb-1 (fast simulation) q Studies with full simulation foretell smaller significance z z z S. Rajagopalan realistic tau efficiencies and Missing ET resolutions Cuts need to be reoptimized to reflect realism Possibility to recover 30% more events which have unphysical solultion FAST SIM FULL SIM TEV 4 LHC Workshop, Feb 5, 2005 21

Conclusion T Lepton-ID studies and impact on physics being studied. q Geant-4 based simulation

Conclusion T Lepton-ID studies and impact on physics being studied. q Geant-4 based simulation + new reconstruction software T Experience being gained through q Test-beam exercises q Data Challenges q Commissioning phase to begin soon. T Must prepare ourselves for Day 1 scenario q How to deal with limited understanding of detector sub-systems? q Experience from Tevatron can play a crucial role q To bring us up to speed quickly and be in a position for real physics S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 22

What are the elements of reconstruction Digitized Output to Reconstructed Energy Compute cell energy

What are the elements of reconstruction Digitized Output to Reconstructed Energy Compute cell energy from digitized waveform E = F * ∑ ai S i Si = digitized samples (digitized every 25 nsec) F = ADC DAC m. A Ge. V*SF ai = Optimal Filtering Coefficients Measured using physics pulse shape (gi) Need to derive from calibration signals TB data has shown we can predict physics pulse shape from calibration signals to better than 0. 2%. But need to understand underlying event effects Need to verify during real data taking from a sample of isolated electrons S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 23

Clustering T Measure Cell energy to EM scale q Correcting for HV, dead channels,

Clustering T Measure Cell energy to EM scale q Correcting for HV, dead channels, … T Calorimeter Clustering: (Cone, Nearest Neighbour) q Two clustering algorithms being studied : Cone and Nearest Neighbour q Correct for position biases, energy modulation, upstream material, cracks, containment, etc. S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 24

Combined reconstruction T Combined Reconstruction q Identify EM clusters and correct for calorimeter effects

Combined reconstruction T Combined Reconstruction q Identify EM clusters and correct for calorimeter effects q Look for well matched track (+ TRT confirmation) q Handle conversions and brem recovery q Corrections for electrons vs photons q candidates for input to physics analysis. T Soft electrons q Start with track as seed, extrapolate back to calorimeter and cluster. S. Rajagopalan TEV 4 LHC Workshop, Feb 5, 2005 25