1. BEAUTY 2006, 28/09/06Julie Kirk1 B-triggers at ATLAS & CMS Julie Kirk Rutherford Appleton Laboratory On behalf of ATLAS and CMS

2. BEAUTY 2006, 28/09/06Julie Kirk2 Outline Introduction –B physics at LHC –B-triggering at LHC ATLAS –Overview of trigger and B-strategy –Specific triggers Di-muon trigger RoI guided triggers CMS –Overview of trigger and B-strategy –Examples of specific channels – B s →J/ ψ  Plans for 900 GeV running Summary

3. BEAUTY 2006, 28/09/06Julie Kirk3 B physics at LHC LHC: proton-proton collisions at √s = 14 TeV bunch crossing rate 40kHz High bb production cross section: ~500 µb (~ 1 in 100 p-p collisions → bb pair). Must select those of interest. Current luminosity plans: –Pilot-run in 2007, 900 GeV, ~10 29 cm -2 s -1 –low-luminosity: up to 2x10 33 cm -2 s -1 (~10 fb -1 per year) –high-luminosity: 10 34 cm -2 s -1 (~100 fb -1 per year) B-physics programme - plan to study: (covered by other talks) –CP violation (e.g. B→J/ ψ(X), B→  ) –B s oscillations (e.g. B s →D s π, B s →D s a 1 ) –Rare decays (e.g. B→  (X), B→K* γ )

4. BEAUTY 2006, 28/09/06Julie Kirk4 B trigger strategies Limited bandwidth for B-triggers (ATLAS & CMS emphasis on high-p T physics) – need to be efficient and selective. Factor ~2 drop in luminosity during a fill ( use some of spare capacity for B-physics?) Many b-decays contain J/ ψ (useful for calibration/understanding detector as well as B-physics) B-trigger is based on single and di-muons –BR ~ 10 % but clean signature at early level in trigger and give flavour tag (needed in many analyses) Different strategies in different lumi regimes : High lumi (>2x10 33 cm -2 s -1 ) LVL1 di-muon trigger → events with 2 muons rare decays (B→  (X)) & J/ ψ →  Lower lumi (< 2x10 33 cm -2 s -1 ) Continue di-muon trigger Add triggers using LVL1 single muon trigger. High Level Trigger (HLT) reconstruction in secondary Regions of Interest identified by LVL1. Broad programme of B-physics in initial low luminosity period. Continue with rare-decay searches at high luminosity

5. BEAUTY 2006, 28/09/06Julie Kirk5 ➸ 46m Long, 22m Diameter, 7'000 Ton Detector ATLAS ( see talk by H.Burckhart )

6. BEAUTY 2006, 28/09/06Julie Kirk6 Overview of ATLAS trigger <2.5  s ~10 ms ~1 s HLT ~1-2 kHz out ~100 Hz out LEVEL 1 TRIGGER Hardware based (FPGAs ASICs) Uses coarse granularity calorimeter and muon information Identifies Regions Of Interest for further processing LEVEL 2 TRIGGER Full granularity Confirm LVL1 trigger Combine info from different detectors in RoIs around LVL1 EVENT FILTER Refines LVL2 selection using “offline-like” algorithms Better alignment and calibration data available HLT: software based B physics allowed ~5-10% of total trigger resources => it must be fast, efficient and selective.

7. BEAUTY 2006, 28/09/06Julie Kirk7 ATLAS B-trigger High luminosity – di-muon trigger (p T > 6 GeV) –B → J/ ψ(  ) X –Rare decays with di-muon, e.g. B→ , B→K 0*  Low luminosity – add single muon trigger with additional JET/EM ROI information from LVL1. At LVL2 have 2 possible approaches: –Full reconstruction inside inner detector (time costly) –Use LVL1 Regions of Interest (RoI) to seed LVL2 reconstruction: Jet RoI for hadronic final states (e.g. B s →D s (  π ) π ) EM RoI for e/  final states (e.g. J/ ψ →ee, K* γ,  γ ) Muon RoI to recover di-muon final-states in which second muon was missed at LVL1. –New comparison of the 2 approaches using Jet RoI for B s →D s (  π ) π

8. BEAUTY 2006, 28/09/06Julie Kirk8 ATLAS LVL1 muon trigger LVL1 muon trigger requires hits in different layers within coincidence window (3/4 hits for low p T muon). Efficiency ~85% (inefficiency mostly due to geometrical reasons (feet and supports)) Inner Detector Muon Trigger Chambers (RPC) Muon Precision Chambers (MDT) Muon Trigger Chambers (TGC) RPC: Restive Plate Chambers TGC: Thin Gap Chambers MDT: Monitored Drift Tubes pT>6 GeV pT>20 GeV

9. BEAUTY 2006, 28/09/06Julie Kirk9 LVL1 muon trigger p T of muons from different processes Predict LVL1 rate from convolution of efficiency with predicted cross-section as a function of p T Rate ~ 21kHz (p T >6GeV) ~15% due to b events Main background from π/K Reduce rate at LVL2 by: use precision muon chambers extrapolate and match to inner detector tracks

10. BEAUTY 2006, 28/09/06Julie Kirk10 Di-muon trigger for rare decays LVL2:  Confirm each  RoI  In precision muon chambers  Combine  with Inner Detector track  Mass cut (>2GeV) EF: Refit ID tracks in Level-2 RoI Decay vertex reconstruction Transverse Decay length cut: L xy > 500mm Angular Distribution cut Efficiency estimation after EF: –70% of B   +  - –60% of B  K *  +  - –Output rate < 10 Hz bb   +  - both  p T >6 GeV Di-  Mass, (MeV) B  K *  +  - B   +  - Not normalized Found mass cut very sensitive for asymmetry study (B→  K*) Re-analysis underway investigating vertex quality cuts Also used for J/ ψ →  LVL1: 2  RoI p T (  ) > 6GeV (~500 Hz @ L=10 33 cm -2 s -1 )

11. BEAUTY 2006, 28/09/06Julie Kirk11 LVL2 ROI guided approach Limits on cpu and bandwidth for B-triggers => need to be fast, efficient and selective. Retrieve information for smaller region of detector => faster execution times Speed depends on RoI size and mean LVL1 RoI multiplicity per event RoI multiplicity required to be about 1-2 to keep resource needs reasonable => determines thresholds chosen Higher Jet ET threshold → lower rate but reduced efficiency Jet RoI EM RoI RoI multiplicity bb→  (6)X

12. BEAUTY 2006, 28/09/06Julie Kirk12 B s →D s (φπ)π – RoI guided Efficiency LVL1: 1 μ (> 6GeV) + ≥1 Jet RoI Efficiency for B to be within RoI is 78% (p T (B)>10 GeV, p T (K,K, π)> 1.5GeV). LVL2: confirm muon Reconstruct tracks in area around Jet RoI : ΔηxΔ φ=1.5x1.5 Search for pairs of opposite sign tracks with |M(KK) –M( φ )| < 3 σ Add additional tracks to form D s and apply mass cuts |M(KK π ) –M(D s )| < 3 σ M(KK) (GeV) M(KK π ) (GeV) True KK(π) combinations Mean = 1.020 GeV σ = 5 MeV Mean = 1.968 GeV σ = 18 MeV p T (B s ) (GeV) Efficiency for B to be within LVL1 RoI

13. BEAUTY 2006, 28/09/06Julie Kirk13 B s →D s (  π)π : RoI vs Full Scan For fullscan method: –LVL1 muon confirmed at LVL2 –reconstruct tracks in whole inner detector and combine to form  and D s  Lose less low pT B’s LVL2 efficiencies : (p T (B)>10 GeV, p T (K,K,π)>1.5GeV) –RoI: 60% –Fullscan: 68% Background rate (bb→  X) ~175Hz (L=1x10 33 ) Efficiency to select D s after LVL2 - RoI guided - Fullscan

14. BEAUTY 2006, 28/09/06Julie Kirk14 Timings for track reconstruction  RoI guided approach ~4 x faster than full scan.  Also EM and muon RoIs which will increase RoI guided time.  Fullscan looks possible for early running or if use a higher muon pT threshold (pT>8GeV => ~2x reduction in rate) (RoI based) = 23ms = 44 ms 100 ms (Fullscan) = 160ms 400 ms LVL2 muon confirmation reduces rate by factor ~4 (21 → 5kHz) Time available for track reconstruction x4 (10ms → 40ms) Includes RoI multiplicity

15. BEAUTY 2006, 28/09/06Julie Kirk15 Trigger performance predicted using offline code, for 30fb -1 : 15000 B d  K* 0 γ 4800 B s   γ Now repeating analysis using trigger reconstruction code. ATLAS EM RoI J/ ψ ->e + e - LVL1: 1 μ (> 6GeV) + ≥1 EM RoI LVL2: confirm LVL1 muon and EM cluster reconstruct tracks in enlarged area around EM RoI electron ID using calorimeter & TRT combine pairs of tracks to form J/ ψ, mass cuts Efficiency after LVL2 ~ 68% (both e+/e- p T >5GeV) background ~170 Hz Rare radiative decays (e.g. B d →K* γ, B s →  γ) Rate (EF): 0.6 Hz Bd  K* 0 γ 0.5 Hz Bs   γ EM RoI e+e+ e-e- EM Calorimeter TRT Δη ( γ -  daughter ) Δ φ ( γ -  daughter ) Need large RoI (ΔηxΔ  ~ 2x2) B d  J /  (e + e - )  K s (  +  - )

16. BEAUTY 2006, 28/09/06Julie Kirk16 J/ ψ→  +  - at low lumi (lowering 2 nd muon threshold) Add single muon at LVL1 (22kHz) LVL2: –confirm muon (first in muon detector and then combined with ID) => rate ~ 5 kHz –Open region around muon RoI and search for J/ ψ in inner detector. –Mass cut (M(  +  - )>2.8GeV) –Extrapolate tracks to muon system. EF : Refit tracks in RoI –Vertex reconstruction LVL2:Efficiency 68-77% (1 st  6Gev,2 nd  3GeV) Background 260-380 Hz J/       efficiency vs. background rate 66 68 70 72 74 76 78 0.20.250.30.350.40.45 Background rate, (kHz) J/psi Efficiency (%) Efficiency vs. Δη (RoI half-width) 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 0.10.20.30.40.50.60.70.8 Delta eta Efficiency Use similar method to improve efficiency for other di-muon channels at low lumi

17. BEAUTY 2006, 28/09/06Julie Kirk17 MUON BARREL CALORIMETERS Silicon Microstrips Pixels ECAL Scintillating PbWO 4 Crystals Cathode Strip Chambers (CSC) Resistive Plate Chambers (RPC) Drift Tube Chambers (DT) Resistive Plate Chambers (RPC) SUPERCONDUCTING COIL IRON YOKE TRACKERs MUON ENDCAPS Total weight : 12,500 t Overall diameter : 15 m Overall length : 21.6 m Magnetic field : 4 Tesla HCAL Plastic scintillator brass sandwich ➸ Tracking up to η  2.5 CMS detector (see talk by O. Buchmuller)

18. BEAUTY 2006, 28/09/06Julie Kirk18 Overview CMS trigger Two level trigger: Level 1 based on muons and calorimeters (40 MHz → ~100 kHz) High Level trigger (HLT) uses similar reconstruction to offline (100 kHz → ~150 Hz) For B-physics: Level 1 : single or di- muon trigger –single muon pT>14 GeV (~3.2 kHz) –di-muon pT> 3GeV (~900 Hz) HLT : –Inclusive b, c trigger through b-tagging: ~ 5 Hz (L1: high E T jet). Doesn’t trigger decays of interest for B-physics measurements. –Exclusive b trigger: reconstruct b-decays using partial reconstruction in region of interest around Level1 muon.

19. BEAUTY 2006, 28/09/06Julie Kirk19 CMS trigger for B-physics Benchmark channels: –B→  LVL1: di-muon HLT : reconstruct tracks around muons, vertex fit and mass cuts –B s →D s π →KK ππ LVL1: single muon (p T >14GeV) or lower p T muon and low p T jet trigger HLT: Partial track reconstruction, mass cuts and vertex fit, decay length cut –B s →J/ ψ  →  KK LVL1: di-muon HLT step 1: partial track reconstruction around muons, J/ ψ search HLT step 2:  and B s search TDR results (2002) Recent update

20. BEAUTY 2006, 28/09/06Julie Kirk20 Partial track reconstruction Track parameter resolution asymptotic after only 5/6 hits. Partial Reconstruction – stop track reconstruction once enough information is available to answer a specific question, e.g. momentum resolution is good enough. Resolutions as a function of the number of hits used: (b-jets, 2.5<pT<5, |η|<0.9) (“0 hits” indicates full track reconstruction!) pT resolution Transverse impact parameter resolution

21. BEAUTY 2006, 28/09/06Julie Kirk21 B s →J/ ψ ( μ + μ - )  (K + K - ): HLT step 1 HLT split into 2 steps. First step – reconstruct J/ ψ : Regional, partial track reconstruction in cones around L1 muon candidates Partial reconstruction up to 5 hits maximum p T μ > 2.5 GeV/c (|η| 2 GeV/c, p T J/ψ > 4 GeV/c Track pairs with opposite charge: |M(μμ) - M(J/ψ)| <150 MeV/c2 Vertex Fit of track pairs: χ2<20 Transverse decay length significance > 3 Cosine of angle (momentum/decay length) > 0.9 J/ψ  mass distribution (HLT) Mean = 3.098 GeV/c 2,  = 51 MeV/c 2 Signal *10 3 Inclusive b→J/ψ X Prompt J/ψ Transverse decay length significance

22. BEAUTY 2006, 28/09/06Julie Kirk22 B s →J/ ψ ( μ + μ - )  (K + K - ): HLT step 2 Level 3: Further reduction through full reconstruction –  and B s search Regional, partial track reconstruction in cones around J/ψcandidates ➣ Partial reconstruction up to 5 hits maximum ➣ p T K > 0.7 GeV/c, p T  > 1.0 GeV/c, p T Bs > 5.0 GeV/c ➣ |M(KK) - M(  )| < 20 MeV/c 2 ➣ |M(  KK)- M(B s )| < 200 MeV/c 2 Vertex Fit on 4 tracks, similar requirements (Cosine of angle > 0.95) B s mass distribution Mean = 5.372 GeV/c 2,  = 65.4 MeV/c 2  mass distribution Mean = 1.019 GeV/c 2,  = 4.5 MeV/c 2

23. BEAUTY 2006, 28/09/06Julie Kirk23 B s →J/ ψ ( μ + μ - )  (K + K - ) LVL2: Accept rate reduced to ~ 15 Hz 80% of J/ψ are from B decays After LVL3: Events/10fb-1: ~150'000 HLT accept rate < 0.1 Hz (L=2x10 33 ) Efficiency w.r.t. generated sample (p T (  )>3(2)GeV, p T (K)>0.8GeV)

24. BEAUTY 2006, 28/09/06Julie Kirk24 Plans for 900 GeV running Cross-section for bb much lower w.r.t σ tot => not much b-physics ( O (1- 10) bb→J/ ψX) For prompt J/ ψ and  expect ~100 events (after 30 days @ 10 29 ) Use these to test mass reconstruction, etc. Run loose triggers with single-  or minimum-bias at LVL1. HLT in pass- through mode – i.e. record information about HLT reconstruction but don’t reject events on this information. 900 GeV

25. BEAUTY 2006, 28/09/06Julie Kirk25 Summary Demonstrated flexible B-physics trigger strategies Various approaches for coping with increased luminosity (RoI guided track reconstruction/ partial reconstruction, eventually running only di-muon trigger at design luminosity) Broad programme of B-physics during early running and rare decay searches continue at high luminosity. Now preparing trigger menus to maximise B-physics potential of experiments. Look forward to first data.

26. BEAUTY 2006, 28/09/06Julie Kirk26 Backup slides

27. BEAUTY 2006, 28/09/06Julie Kirk27 6 GeVBarrelEndcapCombined π/K9.35.915.2 b1.61.83.4 c0.91.01.9 W0.003 negligible 0.001 total11.88.720.5 ATLAS LVL1 muon rates

28. BEAUTY 2006, 28/09/06Julie Kirk28 rates30d = 10 6 s jet p T >15GeV |  <2.5 24 10 -1 Hz 2 400 000 Min bias   X |   |<2.5 1400 10 -4 Hz 140 000 b-jet p T >15GeV |  <2.5 370 10 -4 Hz 37 000 jet p T >50 GeV |  <2.5 45 10 -4 Hz 4 500 bb   X |   |<2.5 60 10 -4 Hz 6 000 bb   X |   |<2.5 2 10 -4 Hz 200 pp   |   |<2.5 1.7 10 -4 Hz 170 pp  J/  X |   |<2.5 1 10 -4 Hz 100 b-jet p T >50 GeV |  <2.5 0.63 10 -4 Hz 63 bb  J/  X |   |<2.5 0.1 10 -4 Hz 10 900 GeV 10 29 cm -2 s -1 rates, statistics

29. BEAUTY 2006, 28/09/06Julie Kirk29 Level-1: Calorimeter Calorimeter Trigger looking for e/  + Jets +  objects Using trigger towers of Hadronic and Electromagnetic calorimeters The requirement for a trigger object: The RoI cluster – local maximum The most energetic cluster > E T Total E T in EM isolation < EM Isolation Threshold Total E T in Hadron < Hadronic isolation threshold