1 / 32

Triggering the LHCb experiment

Triggering the LHCb experiment. The LHCb detector LHCb trigger system High Level Trigger with muons: Specific selections Generic selections Hugo Ruiz. 4a Trobada de Nadal de Física Teòrica a la Universitat de Barcelona. Detector overview. Muon System. RICHES: PID: K,  separation.

libba
Download Presentation

Triggering the LHCb experiment

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Triggering the LHCb experiment The LHCb detector LHCb trigger system High Level Trigger with muons: Specific selections Generic selections Hugo Ruiz 4a Trobada de Nadal de Física Teòrica a la Universitat de Barcelona

  2. Detector overview Muon System RICHES: PID: K, separation VELO: primary vertex impact parameter displaced vertex PileUp System Interaction point Calorimeters: PID: e,, 0 Trigger Tracker: p for trigger Tracking Stations: p of charged particles

  3. LHCb environment • LHC: • 40 MHz crossing rate • 30 MHz with bunches from both directions at LHCb IP • 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 (60 mb out of 80 mb) • bb:100KHz • Whole decay of one B in acceptance: 15KHz • cc: 600KHz

  4. 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 TeV • Predicted cross-sections: sinel = 79.2 mb, sbb=633 mb • Pileup (multiple interactions in single bunch crossing) simulated • GEANT4 for full simulation of all events (minimum bias, signal) • Additional backgrounds: • off-beam muons • low-energy background at muon chambers • Spilloversimulated in detector response • from two preceding and one following bunch crossings

  5. Trigger overview Pileup system VELO + Trigger tracker Calorimeters + Muon system new proposal: ~ 2 KHz 10 MHz On custom boards: LHCb trigger TDR September ‘03 L0:hight pT + not too busy • Synchronous 40 MHz, latency: 4ms PC farm ~1800 CPUs 1 MHz L1:IP + high pT • <latency>: 1 ms (max 50 ms) • Buffer: ~ 59k events 40 KHz HLT + reconstruction • Full detector: ~ 40 kb / evt 200 Hz

  6. Level 0 • Fast search for ‘high’ pT particles(calorimeters, muon syst) • Charged hadrons: HCAL (~ 3 GeV) • Electrons, photons, p0: ECAL (~ 3 GeV) • Muons: muon system (~ 1 GeV) • 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 (lots of tracks with high IP) • Busy events spend trigger resources without being more signal-like • Better throw them early and use bandwidth to relax other cuts

  7. Level 0: calorimeter trigger ECAL HCAL SPD-PreShower FE Scintillator Pad Detector (SPD) • The LHCb calorimeter: • ECAL: 6000 cells, 8x8 to 24x24 cm2 • HCAL: 1500 cells, 26x26, 52x52 cm2 • Trigger strategy: look for high ET candidates: • In regions of 2x2 cells • Particle identification from • ECAL / HCAL energy • PS and SPD information • ET threshold ~ 3 GeV • Sent to L0 decision unit: • Highest ET candidate each type • Global variables: • Total calorimeter energy • SPD multiplicity ECAL HCAL Pre-Shower Detector Validation cards Selection crates g ETtot e± p0 hadr SPD mult

  8. Level 0: muon trigger • The LHCb muon system: • 5 stations • Variable segmentation • Projective geometry • Trigger strategy: • Straight line search in M2-M5 • Look for compatible hits in M1 • Momentum measurement • Sent to L0 decision unit: 2 highest pT candidates per quadrant threshold

  9. Level 0: Pile-up system Interaction region • Pileup system: • 2 silicon planes • Measure R coordinate • Backwards from interaction point  no tracks from signal B • Trigger strategy: veto multi-PV evts • From hits on two planes  produce a histogram of z on beam axis • Sent to L0 Decision Unit: height of two highest peaks + multiplicity

  10. Level 0: Decision • L0 decision unit: • OR of high ET candidates • Apply cuts on global properties • Thresholds and partial rates: (Trigger TDR, Sept 2003) • Composition:

  11. L1-HLT infrastructure Front-end Electronics FE FE FE FE FE FE FE FE FE FE FE FE TRM 126 links 44 kHz 5.5 GB/s Multiplexing Layer Switch Switch Switch Switch Switch 64 Links L1-Decision Readout Network Sorter 94 Links 7.1 GB/s 94 SFCs SFC SFC SFC SFC CPUFarm … ~1800 CPUs Switch Switch Switch Switch CPU CPU CPU CPU CPU CPU CPU CPU CPU CPU CPU CPU Gb Ethernet Level-1 Traffic HLT Traffic Mixed Traffic • L1 & HLT share infrastructure: • Ethernet network • Sub-farm controllers • Computing nodes • Provides flexibility, scalability • HLT & reconstruction run in background • L1 task has top priority • CPU share: ~ 55% L1, 25% HLT, 20% reconstruction

  12. Level 1  sensor R sensor 100 cm • Trigger strategy: • Find high IP tracks (tracking in VELO) • Confirm track / estimate pT from TT • Special treatment treatment for calorimeter and muon-matched objects • The LHCb VELO: • 21 stations (~ 100 cm) • Alternated R-f sensors • 40 μm to 100 μm pitch • Busy environment: • ~ 70 tracks/event after L0 • but low occupancy in VELO (~0.5%) Interaction region

  13. 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 • 3D tracking for those tracks • pT measurement using TT • Silicon, 2 layers, 200 mm pitch • Only 0.15 T.m between VELO and TT DpT / pT ~ 20-40% • Rejects most low momentum tracks, which can fake high IP

  14. Level 1 decision • ‘OR’ of 5 different streams: • Generic two high pt tracks with IP>0.15 mm • Electron or photon with high pT, together with two high PT & IP tracks • Single muon with high PT & IP • Dimuons • High mass, flight significance • Mass ~ mJ/y, no flight bias • Composition of triggered sample: Dimuon masses at L1

  15. Performance: L0 x L1 • Results from Trigger TDR (Sept 2003) • Efficiencies computed on offline selected events • Overall L0xL1 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 L0 efficiency L1 efficiency L0L1 efficiency

  16. HLT flow diagram D* ~300 Hz Flight-unbiased J/ys ~ 500 Hz B-generic (single m) ~1KHz Storage (more complicated access due to distribution on grid) Recent additions, effect on computing model 40 KHz (15% bb, 18% cc) Re-reconstruct L1-firing tracks (now using all tracking stations) and confirm calo/m objects Good m or calo object Rest Confirm generic L1 decision (p)/p ~ 0.6 % Apply loose pre-selection HLT no 10 KHz (45% bb, 20% cc) Reconstruct all tracks (RICH info?) HLT selection algorithms HLT no CP channels, large e, ~200 Hz Complete reco Hot !

  17. Status of HLT • Two milestones on summer 2005: • Trigger-DAQ Challenge: one complete sub-farm running continuously on MC data • Computing TDR • Status of HLT reconstruction: • All pieces are there • They ensure use of full detector at enough speed • Calo reconstruction, m id, tracking, L1 confirmation • Studying feasibility of using pId from RICH at HLT! • Effort is put in improving performance of tracking: • Efficiency: current losses of 3 to 15% per track depending channel, compared to offline reconstruction • reverts in > 8% selection inefficiencies! • Computation of errors in track parameters • Use of Kalman filter takes too long • Alternative method needs to be used, as such errors needed to compute c2 of vertices, significance of displacements from PV next 2 slides

  18. Status of HLT Tracking • Obtain track errors from a parameterization on 1/pT: 1. Fit s(IPx), s(IPx) vs 1/pT 2. Assume cylindrical errors and insert in covariance matrix sx sy observed 10 GeV 125 MeV 50 mm • Use the matrix to compute IP significances, vertex c2, flight significances 1 p0 45.2 ± 0.7 2 p1 12.2 ± 0.6 3 p2 1.8 ± 0.1

  19. Status of HLT Tracking • Advantage of the parameterization: • Fast • Direct control of track retention • Trigger can adjust rate by itself, no need of understanding errors of Kalman filter • Performance proven to be ~ the same • (Near) future improvement: • Make use of the fact that main component of the error is multiple scattering at VELO’s RF foil Flight significance computed using parameterized matrix: Dimuons from offline-selected B  J/y f All pairs of pions FS Hugo Ruiz

  20. Status of HLT Specific selections Reconstruct all tracks (RICH info?) HLT selection algorithms HLT no D* ~300 Hz Flight-unbiased J/ys ~ 500 Hz B-generic (single m) ~1KHz CP channels, large e, ~200 Hz Complete reco Storage Hot ! • Chosen 10 benchmark channels representing the spectrum of interesting channels • Proved that they can be selected with enough sidebands with a rate of some tenths of Hz • Caveats: • Currently some inefficiency due to tracking • Some other tenths of channels to be included • Example: channels with di-muons: • Bs  J/(+-) • B  J/(+-) KS • Bs +- • B K*+- Bd->Jpsi(mm)K* Bs->Jpsi(mm)eta(gg) Bs->Jpsi(mm)phi Bs->Jpsi(mm)eta'(rhog) Bs->Jpsi(mm)eta(pipipi) Bs->Jpsi(mm)eta'(etapipi) Bc->Jpsi(mm)pi Bu->Jpsi(mm)K+ Bu->K+mumu Bu->smumu Bs->phimumu D0->mumu Benchmark Hugo Ruiz

  21. Statatus of HLT Specific selections with ms • BUT: • Expected ~ 100% (modulus tracking inefficiency), via J/y line • Ideally, the HOT dimuon trigger would be: • Based only on muon tracks (robustness) • As inclusive as possible • ex: all J/y channels at once  J/y with displacement • avoids explosion of # of selections • Performance: • HOT: • 90Hz with efficiency of ~ 85% on all J/y channels • 50 Hz of true J/y ! • To get ~ 85% on B K*+-, use of K* is needed Reconstruct all tracks (RICH info?) HLT selection algorithms HLT no D* ~300 Hz Flight-unbiased J/ys ~ 500 Hz B-generic (single m) ~1KHz CP channels, large e, ~200 Hz Complete reco Storage Hot ! Hugo Ruiz

  22. Status of HLT The high rate flows • Inclusive J/  (+ higher mass ): ~ 600 Hz • 150–200 Hz of J/  signal • Detailed calibration of tracking resolution • Unbiased b  J/ X (lifetimes) • Inclusive D*: ~ 300 Hz • Charm physics • PID calibration: D*+ D0( K-p+) p+,d(mD*-mD0) = 146 MeV (dmD* = 1865 MeV) • Inclusive bX: ~ 2 kHz • ~800 Hz of unbiased b-hadrons with good tagging • Data mining: effective number of generic B’s / year ~ B factories 2014! • Study acceptance and trigger biases Reconstruct all tracks (RICH info?) HLT selection algorithms HLT no D* ~300 Hz Flight-unbiased J/ys ~ 500 Hz B-generic (single m) ~1KHz CP channels, large e, ~200 Hz Complete reco Storage Hot ! m (always there!) tagging B X signal B BR(B X) ~ 10% Hugo Ruiz

  23. Inclusive J/y: CDF experience • Dimuon trigger: pTm > 1.5 GeV • ~ 2 million J/y • 80% prompt • 20% from B • Used to set absolute mass scale better than 1 MeV: • Needed for spectroscopy of B, D and quarkonium states • Understanding of trigger acceptance: • Compare acceptance vs pT, IP between MC and high statistics of J/y data

  24. Inclusive J/y: LHCb • After 0.5 seconds of LHC running (hopefully): -50 MeV +50 MeV sm = 36 MeV Offline 9 MeV • ~ 1.5·109 J/ys per year

  25. Inclusive J/y • As a function of flight significance:  Total rate True Jpsi * True J/y & bb

  26. Inclusive b  X For cut in pT of 1,2,3,4 GeV 50 % 100 % • Purity of the triggered sample

  27. Inclusive b  X • Comparison with specific selection (ex: Bd p+p-, using offline selected events) ex(1-2w)2 Very robust against online reco inefficiencies!

  28. Conclusions • L0&L1 of LHCb trigger are mature and ready • Of course, expect proposals for upgrades! • Lot of activity in developing HLT • Reconstruction still need some work • Most of specific selection algorithms there • High-rate flows in HLT are promising • Open doors to new physics • Allow study systematics from data instead of MC • Safer • And… negligible economic impact, as less MC needs to be produced! Hugo Ruiz

  29. BACKUP SLIDES

  30. L1

  31. Expected event yield • Taking into account efficiency from: • L0xL1 • Offline selection

  32. Close-up comparison of effective # evts Hugo Ruiz

More Related