18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 1 Tracking and Physics Studies in LHCb Jeroen van Tilburg NIKHEF Jaarvergadering Outline: LHCb light OT simulation Track finding/fitting Trigger Physics analysis

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 2 LHCb classic setupLHCb light setup ~6  5 m 2 21 stations R and φ sensors VELO ~1.4  1.2 m 2

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 3 LHCb in a new light Why this re-optimization: Material budget was too high in a realistic setup. Trigger performance dropped with new tuning of Pythia. Classic setup (Techical Proposal): 11 tracking stations. Tracking through magnet. Magnetic shield for RICH1. Realistic design was too heavy: 60% X 0 in front of RICH2. Light setup (now): 3 tracking stations plus Trigger Tracker. Also VELO, beampipe & RICH1 lighter. Shield removed: B-field in RICH1. New: all silicon TT stations. Much lighter: 40% X 0 in front of RICH2. Tracking & reconstruction improved: Less interactions in detector. Better trigger performance: add p T information using TT.

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 4 LHCb Software Transition from Fortran  C++ Geant3  Geant 4 in progress. Now completely C++: Digitization Reconstruction Physics analysis Data production last summer 3.6 M events generated at various centers First physics studies done with all new software in place: Similar performance compared to Technical Proposal. Next data challenge Feb-April 2003 with 10x more statistics needed for the Trigger and Light TDR (submission Sept 2003). More realistic simulation Material budget based on TDR’s. Detailed detector responses; tuned to test beam data. No use of MC Truth: Full pattern recognition (realistic ghost rate).

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 5 OT Software Geant3 LHCb event display OT Software divided in three steps: Detector Simulation: Currently in Geant3 (Fortran). Stores entry- and exitpoints of sensitive layers. Digitization: Finds which OT wires are hit. Detailed simulation of the detector response. T 0 -calibration: Correct for the time-of-flight. Input to the tracking. Outer Tracker simulated in LHCb software

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 6 OT Digitization OT geometry: 3 stations with 4 double layers of straws per station These 4 layers arranged in 0°,+5°,-5°,0°. OT double layer cross section Find closest distance to wire. Smear with 200 μm Gaussian. Calculate drift time. Maximum drift time 32.5 ns. Tuned to test beam results pitch 5.25 mm 5mm cells Track e-e- e-e- e-e- Track less efficient near edge of straw

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 7 Time spectrum of single bunch for interaction at t=0. But LHC bunch spacing is 25 ns. So time spectrum of previous and next bunches overlap  spillover.... Time spectrum with spillover: Two previous spills and next spill. Dead time of 50 ns included. Cross talk of 5% included. But electronics read-out window is only 50 ns.... Time spectrum as it is measured with read-out gating. But particles need time to travel to OT stations  Time of flight correction.... Time spectrum after time of flight correction: For each hit assume that the particle traveled in a straight line from IP with speed of light. OT Digitization

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 8 Efficiency = 98 % (at least one hit in a double layer for p>2GeV; inefficiency due to dead regions and dead time) On average 20 % of the hits are due to cross-talk and spillover. Average occupancy in OT ~ 4 % (hottest region ~ 7 %) OT Performance Core resolution = 200 micron Long Tails: Time of flight correction not a good approximation for particles with p < 2 GeV. No problem for reconstruction of “physics” tracks.

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 9 Track finding algorithms Many track types, many algorithms Velo tracks:used to find primary vertex. Forward tracks:used for most physics studies: B decay products. Seed tracks:improve RICH2 performance. Matched tracks:additional to Forward tracks. Upstream (T  TT):enhance K S finding. Velo  TT (VTT):improve RICH1 performance, slow pions, kaon tagging.

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 10 Efficiency for tracks p>5 GeV ~ 95% Track finding Ghost rate for tracks p>5 GeV ~ 10%

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 11 The Kalman Fit properties: Adds measurements recursively: easy to use in the pattern recognition. Mathematically equivalent to least χ 2 method. But faster. Needs as input initial track estimate. Easy to include multiple scattering and energy loss. Track Fit The tracks are fitted using the Kalman Filter. The Kalman Fit: 1.The prediction step. 2.The filter step. Adds measurements one-by-one. 3.The smoother step. direction of the filter track prediction filtered track

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 12 Kalman fit provides an excellent momentum estimate at the vertex. N.B. Fitted with single Gaussian in each bin. Track Fit Δp/p LHCb light LHCb classic (TP) Momentum resolution core σ = 0.35 % 2nd σ = 1.0% (fraction 0.1)

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 13 Trigger performance Magnetic shield removed  B-field between Velo and TT. Trigger (Level 1) reconstructs tracks in Velo to cut on large impact parameters. Add TT clusters to these tracks for rough estimate on momentum (~30%). Trigger performance improved by combining impact parameter and p T information. Minimum bias Feasibility study to include OT information in Level-1. Light Classic

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 14 This autumn first (large) physics study after TP. With new software. Many decay modes are studied. Large contribution from NIKHEF: B s  D s - (  K + K - π - )π + and c.c.B s oscillations, measures Δm S B s  D s - (  K + K - π - )K + and c.c.measures sin(γ-2δγ) B s  J/ψ (  l + l - ) φ (  K + K - )measures sin(2δγ) Physics studies LHC provides huge statistics for B-physics: bb pairs produced per year σ bb /σ inel = 1/160

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 15 B s  D s π and B s  D s K B s  D s π:Annual yield = 50 k B/S < 90% CL B s  D s K:Annual yield = 6.7 k B/S < 90% CL Estimates after trigger and offline selection (before tagging): After applying selection cuts on 1 M bb background event 0 events are selected.

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 16 Δm S sensitivity After one year: >5σ measurement of Δm S up to 48 ps -1 95% CL excl. of Δm S up to 58 ps -1 Good time resolution = precise measurement of Δm S A few months statistics. (Tagging not yet tuned: use tag from MC.) B s  D s K: σ γ =[6º,13º] per year

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 17 B s  J/ψ (  μ + μ - ) φ(  KK) B S mass σ = 13.9 ± 0.6 MeV/c 2 Time resolution σ = 37.2 ± 1.5 fs Annual yield = 80 k B/S < 90% CL σ δγ ~ 0.6° per year Estimates after trigger and offline selection (before tagging): After applying selection cuts on 1 M bb background event 0 events are selected.

18 december 2002, NIKHEF Jamboree Tracking and Physics Studies, Jeroen van Tilburg 18 Conclusions Light status Tracking: High efficiency for “physics” tracks. Good resolutions. Trigger: Much improved. More robustness. Fine tuning for Trigger TDR with coming MC production. Overall reconstruction & Physics analysis: Final decay channel efficiencies similar to TP. Current estimates on background are inconclusive. Winter production will generate much more statistics. Last year a lot of studies done with LHCb-light. Also much more realism all parts of software.