LHCb trigger algorithms and performance Introduction The LHCb
LHCb trigger: algorithms and performance § Introduction § The LHCb trigger: § § § Level 0 Level 1 HLT § Performance Hugo Ruiz on behalf of the LHCb collaboration 9/9/2020 Hugo Ruiz 1
LHCb environment § LHC: § § 40 MHz crossing rate 30 MHz with bunches from both directions § Luminosity: 2· 1032 cm-2 s-1 § § 10 to 50 times lower than @ ATLAS, CMS (well under machine design!) Reason: single interaction preferred to identify secondary vertices from B mesons § Relevant rates: (for visible events at least 2 tracks in acceptance) § Total rate (minimum bias): 10 MHz § bb: 100 KHz § Whole decay of one B in acceptance: 15 KHz § 9/9/2020 cc: 600 KHz HERA-LHC workshop. Hugo Ruiz 2
Detector overview VELO: primary vertex impact parameter displaced vertex RICHES: Muon System PID: K, separation Pile. Up System Interaction point Trigger Tracker: p for trigger 9/9/2020 Tracking Stations: Calorimeters: PID: e, , 0 p of charged particles HERA-LHC workshop. Hugo Ruiz 3
MC generation § PYTHIA 6. 2 used § § § Minimum bias: hard QCD, single / double diffraction, elastic scattering Signal: forcing B-mesons in a minimum bias event to decay into specific final state Charged particle distributions tuned to data for s < 1. 8 Te. V Predicted cross-sections: sinel = 79. 2 mb, sbb=633 mb Pileup (multiple interactions in single bunch crossing) simulated § GEANT 3 for full simulation of all events (minimum bias, signal) § Moving to GEANT 4 in this year’s Data Challenge § Additional backgrounds: off-beam muons, low-energy background at muon chambers § Spillover simulated in detector response (from two preceding and one following bunch crossings). 9/9/2020 HERA-LHC workshop. Hugo Ruiz 4
Trigger overview 10 MHz L 0: hight p. T + not too busy § Fully synchr. (40 MHz), 4 ms latency § On custom boards Pileup system 1 MHz L 1: IP + high p. T § Ave. latency: 1 ms (max 50 ms) § Buffer: 58254 events VELO + Trigger tracker Calorimeters + Muon system 40 KHz HLT + reconstruction § Full detector: ~ 40 kb / evt Single PC farm ~1800 CPUs 9/9/2020 HERA-LHC workshop. Hugo Ruiz ~ 200 Hz 5
Level 0 § Fast search for ‘high’ p. T particles (calorimeters, muon syst) § § § Charged hadrons: HCAL (~ 3 Ge. V) Electrons, photons, 0: ECAL (~ 3 Ge. V) Muons: muon system (~ 1 Ge. V) § Cut on global variables: § Require minimum total ET in HCAL (calorimeters) § Reduces background from halo-muons § Rejection of multi-PV and busy events (Pileup system, SPD) : § fake B signatures (IP) § Busy events spend trigger resources without being more signal-like § 9/9/2020 Better throw them early and use bandwidth to relax other cuts HERA-LHC workshop. Hugo Ruiz 6
Level 0: calorimeter trigger Scintillator Pad Detector (SPD) § The LHCb calorimeter: § § ECAL: 6000 cells, 8 x 8 to 24 x 24 cm 2 HCAL: 1500 cells, 26 x 26, 52 x 52 cm 2 ECAL HCAL § Trigger strategy: look for high ET candidates: § § Pre-Shower Detector In regions of 2 x 2 cells Particle identification from § ECAL / HCAL energy § PS and SPD information § ECAL SPD-Pre. Shower FE ET threshold ~ 3 Ge. V Validation cards § Sent to L 0 decision unit: § § HCAL Highest ET candidate each type Global variables: § Total calorimeter energy § SPD multiplicity 9/9/2020 Selection crates SPD mult e± HERA-LHC workshop. Hugo Ruiz 0 ETtot hadr 7
Level 0: muon trigger § The LHCb muon system: § § § 5 stations Variable segmentation Projective geometry § Trigger strategy: § § Straight line search in M 2 -M 5 Look for compatible hits in M 1 § Momentum measurement § Sent to L 0 decision unit: 2 highest p. T candidates per quadrant threshold 9/9/2020 HERA-LHC workshop. Hugo Ruiz 8
Level 0: Pile-up system § Pileup system: § § § 2 silicon planes Measure R coordinate backwards from interaction point no tracks from signal B Interaction region § Trigger strategy: veto multi-PV evts § From hits on two planes produce a histogram of z on beam axis § Sent to L 0 Decision Unit: height of two highest peaks + multiplicity 9/9/2020 HERA-LHC workshop. Hugo Ruiz 9
Level 0: Decision § L 0 decision unit: § § OR of high ET candidates Applies cuts on global properties § Thresholds and partial rates: (Trigger TDR, Sept 2003) Type Thresh (Ge. V) Rate (k. Hz) Global Variable Hadron 3. 6 705 Electron 2. 8 103 Tracks in 2 nd vertex 3 Photon 2. 6 126 p 0 local 4. 5 110 Pile-Up multiplicity 112 hits p 0 global 4. 0 145 SPD multiplicity 280 hits Muon 1. 1 110 Total ET 5 Ge. V 1. 3 145 Di-muon Sp. T § Composition: 9/9/2020 m bb % Cut cc % Generated 1. 1 5. 6 After L 0 3. 0 10. 6 HERA-LHC workshop. Hugo Ruiz 10
L 1 -HLT infrastructure § L 1 & HLT share infrastructure: § § § Ethernet network Sub-farm controllers Computing nodes Front-end Electronics FE 126 links 44 k. Hz 5. 5 GB/s FE FE FE Switch FE FE Switch § HLT & reconstruction run in background FE SFC CPU CPU § CPU share: ~ 55% L 1, 25% HLT, 20% reconstruction HERA-LHC workshop. Hugo Ruiz CPU CPU SFC … SFC 94 Links 7. 1 GB/s 94 SFCs ~1800 CPUs Switch CPU CPU FE TRM L 1 Decision Sorter Readout Network Switch § L 1 task has top priority FE Switch Multiplexing Layer 64 Links § Provides flexibility, scalability 9/9/2020 FE CPU Farm CPU Gb Ethernet Level-1 Traffic HLT Traffic Mixed Traffic 11
Level 1 § Trigger strategy: § § The LHCb VELO: § § § 21 stations (~ 100 cm) Alternated R-f sensors 40 μm to 100 μm pitch sensor § find high IP tracks (tracking in VELO) Confirm track / Estimate p. T from TT Special treatment for clear L 0 signatures R sensor § Interaction region 0 10 m c § Busy environment: § § ~ 70 tracks/event after L 0 but low occupancy in VELO (~0. 5%) 9/9/2020 HERA-LHC workshop. Hugo Ruiz 12
Level 1: IP at VELO § Fast-tracking strategy: § § § First in R-Z view (only R sensors) Primary vertex σZ ~ 60 mm Select tracks with IP in (0. 15, 3) mm § about 8. 5 / event § 3 D tracking for those tracks § p. T measurement using TT § § § Silicon, 2 layers, 200 mm pitch Only 0. 15 T. m between VELO and TT Dp. T / p. T ~ 20 -40% Rejects most low momentum tracks, which can fake high IP 9/9/2020 HERA-LHC workshop. Hugo Ruiz 13
Level 1 decision § Based on IP and pt of the two highest pt tracks § Decision modified on clear L 0 signatures matching L 1 tracks § § High p. T electrons & photons High mass di-muons: § If mass lies within ± 500 Me. V of J/y or B, accept automatically § Else, ‘bonus’ proportional to di-mass Minimum bias § Composition: bb % 9/9/2020 Signal events cc % Generated 1. 1 5. 6 After L 0 3. 0 10. 6 After L 1 9. 7 14. 2 HERA-LHC workshop. Hugo Ruiz 14
Possible HLT flow diagram 40 KHz (9. 7% bb, 14. 2% cc) Re-reconstruct L 1 tracks (now using all tracking stations) Leptonlike evts “Lepton highway” Rest HLT no Confirm L 1 decision (p)/p ~ 0. 6 %! Reject uds, e > 95% 20 KHz (14. 0% bb, 14. 7% cc) Reconstruct whole event HLT no Specific: Generic algorithm Exclusive (ex: B Dsh) Inclusive (ex: DX) Long-lived b sample CP channels Open charm (systs, backgrounds) with e ~100% J/y-like + Tagging -leptonenriched Full reconstruction / Storage 9/9/2020 HERA-LHC workshop. Hugo Ruiz 15
Performance: L 0 x L 1 § Results from Trigger TDR (Sept 2003) § Efficiencies computed on offline selected events § Overall L 0 x. L 1 efficiency: § § 30% for § hadronic channels § e/γ/π0 channels 60 -70% for di-muons § Software and hardware prototyped and working, within time budget § see Trigger TDR, Sept 2003 9/9/2020 HERA-LHC workshop. Hugo Ruiz L 0 efficiency L 1 efficiency L 0 L 1 efficiency 16
Performance: HLT § A complete implementation of HLT is being prepared for the Computing TDR (due summer 2005) § The reconstruction part has already been implemented and tested in terms of performance and time consumption, in particular: § § Tracking L 1 confirmation § First results on exclusive selections show that individual b physics channels give rates of ~ 10 Hz with § § § 500 Me. V side bands for b mass Affordable time consumptions > 95% efficiencies 9/9/2020 HERA-LHC workshop. Hugo Ruiz 17
Expected event yield § Taking into account efficiency from: § § L 0 x. L 1 Offline selection 9/9/2020 HERA-LHC workshop. Hugo Ruiz 18
- Slides: 18