CMS High Level Trigger Selection Giuseppe Bagliesi INFN-Pisa On behalf of the CMS collaboration EPS-HEP 2003 Aachen, Germany

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ Outline LHC Environment High Level Trigger strategy Object selection e/  Jet  , b HLT rates and efficiencies Conclusions

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ p-p collisions at LHC Crossing rate 40 MHz Event Rates: ~10 9 Hz Max LV1 Trigger 100 kHz Event size ~1 Mbyte Readout network 1 Terabit/s Filter Farm ~10 6 Si95 Trigger levels 2 Online rejection % (100 Hz from 50 MHz) System dead time ~ % Event Selection: ~1/10 13 Crossing rate 40 MHz Event Rates: ~10 9 Hz Max LV1 Trigger 100 kHz Event size ~1 Mbyte Readout network 1 Terabit/s Filter Farm ~10 6 Si95 Trigger levels 2 Online rejection % (100 Hz from 50 MHz) System dead time ~ % Event Selection: ~1/10 13 Event rate “Discovery” rate Luminosity Low 2x10 33 cm -2 s -1 High cm -2 s -1

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ Level-1 (~µs) 40 MHz High-Level ( ms-sec) 100 kHz Event Size ~ 10 6 Bytes Level-1 (~µs) 40 MHz High-Level ( ms-sec) 100 kHz Event Size ~ 10 6 Bytes Trigger environment 40 MHz Clock driven Custom processors 100 kHz Event driven PC network Totally software 100 Hz To mass storage two trigger levels two trigger levels All charged tracks with pt > 2 GeV Reconstructed tracks with pt > 25 GeV Operating conditions: one “good” event (e.g Higgs in 4 muons ) + ~20 minimum bias events)

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ High Level Trigger requirements and operation HLT: Reconstruction and selection of electrons, photons, muons, jets, missing E T, and b and  tagging. HLT has access to full event data (full granularity and resolution) maximum flexibility Main requirements: Satisfy CMS physics program with high efficiency Inclusive selection (we like to see also unexpected physics!) Must not require precise knowledge of calibration/run conditions Efficiency must be measurable from data alone The HLT code/algorithms must be as close as possible to the offline reconstruction Limitations: CPU time Output selection rate (~10 2 Hz) Precision of calibration constants

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ Regional Reconstruction Regional process (e.g. DIGI to RHITs) each detector on a "need" basis link detectors as one goes along physics objects: same Global process (e.g. DIGI to RHITs) each detector fully then link detectors then make physics objects

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ e/  selection Level-2 Level-3 Level-1 Level-2.5 Photons Threshold cut Electrons Track reconstruction E/p, matching (  ) cut ECAL reconstruction Threshold cut Pixel matching In addition : Isolation cuts (ECAL, pixel, track) Had/EM isolation  0 rejection

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ HLT Electron selection (I) Level-2 electron: 1-tower margin around 4x4 area found by Lvl-1 trigger Apply “clustering” Accept clusters if E HCAL /E ECAL <0.05 Select highest E T cluster Brem recovery: Seed cluster with E T >E Tmin Road in  around seed Collect all clusters in road  “supercluster”and add all energy in road

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ HLT Electron selection (II) Level-2.5 selection: add pixel information Very fast, high rejection high efficiency (  =95%), high background rejection (14) Pre-bremsstrahlung: Matching hits given by most electrons and by few photons Require at least 2 hits (3 pixel hits available almost always)

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ HLT Electron  selection (III) Level-3 electron: Build tracks from pixel seeds found in pixel- matching step Very loose track requirements for high efficiency for radiating tracks: 3-hit layers Allow 2 consecutive missing layers Track selection: Barrel: E/p and  (track-cluster) Endcap: E/p Also (non-track): H/E With tight cuts is always possible to select almost no-radiating electron with very high purity

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ HLT muon track reconstruction Inclusion of Tracker Hits: “Level-3” Define a region of interest through tracker based on L2 track with parameters at vertex Find pixel seeds, and propagate from innermost layers out, including muon Standalone Muon Reconstruction: “Level-2” Seeded by Level-1 muons Kalman filtering technique applied to DT/CSC/RPC track segments GEANE used for propagation through iron Trajectory building works from inside out Track fitting works from outside in Fit track with beam constraint Single muons 10<P t <100 GeV/c Level-3 Algorithmic efficiency 

L2 & L3 muon p T resolution and efficiency  =0.11  =0.013 L2 L3 P T resolution barrel 10 GeV threshold 30 GeV Efficiency vs P T threshold L1 L2 L3 (1/p T rec -1/p T gen ) /(1/p T gen )

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ Isolation and physics content after muon Level-3 Before isolationAfter isolation Muons from b,c,K,  decays are greatly suppressed by isolation Isolation is based on transverse energy (E T ) or momentum (P T ) measurements in cones around the muon Calorimeter isolation - E T from calorimeter towers in a cone around the muon Pixel isolation - P T of 3-hit tracks in the pixel detector in cone around the muon - Requires that contributing tracks come from same primary vertex as the Level-3 muon (to reduce pile-up contamination) Tracker isolation - P T of tracks in the Tracker (regional reconstruction around L3 muon)

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ HLT efficiencies on H  WW  2  2 low lumi: M H =120 GeV: single mu 74%, di-mu exclusive 14%, combined: 87 % M H =160 GeV: single mu 87%, di-mu exclusive 5%, combined: 92 % L3 threshold L3 muon thresholds at low luminosity: Single  19 GeV Double  7 GeV

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ Jet rates and thresholds Low luminosity: 1 kHz at Level-1: 177 GeV (1 jet), 85 GeV (3 jet), 70 GeV (4 jet) 1 Hz at HLT: 657 GeV (1 jet), 247 GeV (3 jet), 149 GeV (4 jet) High luminosity: 1 kHz at Level-1: 248 GeV (1 jet), 112 GeV (3 jet), 95 GeV (4 jet) 1 Hz at HLT: 860 GeV (1 jet), 326 GeV (3 jet), 199 GeV (4 jet) Very high rates and thresholds! HLT triggers need some other condition to have acceptably low threshold MET, leptons, isolation, vertices…

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ MET Rates Calorimeter coverage: |  |<5 Generator level: real neutrinos -> E T miss >60 GeV E T miss <60 GeV mostly due to limited coverage Much higher E T miss at HLT than at generator level “E T miss ” objects selection is done in association with other requirements, like a energetic jet

17 Partial track reconstruction strategy at HLT At HLT ultimate resolution is not needed Good track parameter resolution is obtained already with 4 or more hits The time for track reconstruction increases linearly with the number of hits Momentum resolution Full tracker Impact parameter resolution Full tracker Reconstruct only a ROI (Region Of Interest) from LVL1 candidate objects (regional tracking) Use a reduced number of hits (conditional tracking)

18 HLT: tau tagging Regional Tracking : Look only in Jet-track matching cone Loose Primary Vertex association Conditional Tracking : Stop track as soon as Pixel seed found (PXL) / 6 hits found (Trk) If Pt<1 GeV with high C.L. Reject event if no “leading track” found Regional Tracking : Look only inside isolation cone Loose Primary Vertex association Conditional Tracking : Stop track as soon as Pixel seed found (PXL) / 6 hits found (Trk) If Pt<1 GeV with high C.L. Reject event as soon as additional track found Regional seeding: look for seeds in a specific region Essential at High Luminosity activity well advanced TEST CHANNELS A 0 /H 0 (200, 500 GeV) ->  -jet  -jet,  -jet lepton H + (200, 400 GeV) ->  -jet Efficiencies ~ 40-50%, Background rej. after LVL1 ~10 3

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ Inclusive b tagging at HLT Inclusive b tag at HLT possible, provided alignment under control Use tracks to define Jet axis (if rely on L1 Calo Jet ~ randomize signed IP) Performance of simple signed IP “track counting” tags ~ same as after full track reconstruction Regional Tracking: Look only in Jet-track matching cone Loose Primary Vertex association Conditional Tracking: Stop track as soon as Pixel seed found (PXL) / 6 hits found (Trk) If Pt<1 GeV with high C.L. ~300 ms low lumi ~1 s high lumi

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ HLT table: LHC start… Level-1 rate “DAQ staging”: 50 KHz Total Rate: 105 Hz Average HLT CPU: 300ms*1GHz Improvements are possible ChannelEfficiency (for fiducial objects) H(115 GeV)  77% H(160 GeV)  WW*  2  92% H  ZZ  4  92% A/H(200 GeV)  2  45% SUSY (~0.5 TeV sparticles)~60% With R P -violation~20% W  e 67% (fid: 60%) WW 69% (fid: 50%) Top  X 72% HLT performances: Priority to discovery channels TriggerThreshold (  =90-95%) (GeV) Indiv. Rate (Hz) Cumul rate(Hz) 1e, 2e29, , 2  80, (40*25)943 1 , 2  19, , 2  86, Jet * Miss-E T 180 * jet, 3-jet, 4-jet657, 247, e * jet19 * Inclusive b-jets Calibration/other10105

G. Bagliesi – EPS 2003 – Aachen 17-24/7/ HLT: CPU usage All numbers for a 1 GHz, Intel Pentium-III CPU TriggerCPU (ms) Rate (kHz) Total (s) 1e/ , 2e/  , 2  , 2  Jets, Jet * Miss-E T e * jet B-jets Total: 4092 s for 15.1 kHz  271 ms/event Time completely dominated by slow GEANE extrapolation in muons – will improve! Consider ~50% uncertainty! Today: ~300 ms/event on a 1GHz Pentium-III CPU Physics start-up (50 kHz LVL1 output): need 15,000 CPUs Moore’s Law: 2x2x2 faster CPUs in 2007 ~ 40 ms in 2007, ~2,000 CPUs ~1,000 dual-CPU boxes in Filter Farm

G. Bagliesi – EPS 2003 – Aachen 17-24/7/2003 Summary The regional/conditional reconstruction is very useful to reduce CPU time and very effective in the HLT selection Tracker at HLT: Essential for muons, electron and tau selection inclusive/esclusive b-trigger is possible Standard Model physics: “just do it” at lower initial luminosity (“dedicated” triggers could be implemented) Pre-scale or lower thresholds when luminosity drops through fill Conclusions Start-up system 50kHz (Level-1) and 105 Hz (HLT) satisfy basic “discovery menu” The HLT design based on a purely software selection will work: Maximum flexibility and scalability Possibility to use “off-line” reconstruction/algorithms