1. ATLAS Beauty 2002 June 17 - 21 Santiago de Compostela ATLAS B-TRIGGER John Baines Rutherford Appleton Laboratory, UK RAL On behalf of the ATLAS Collaboration

2. ATLAS B-Trigger John Baines RAL Outline Introduction: LHC & ATLAS The ATLAS detector Trigger Architecture B-Physics Programme History & Recent Developments ATLAS B-TRIGGER UPDATE

3. ATLAS B-Trigger John Baines RAL LHC: Switch on: 2007 Peak Luminosity: 2x10 33 cm -2 s -1 10 34 cm -2 s -1 4.6 23 interactions per bunch crossing Coast (fill) lasts ~10 hours  Factor ~2 drop in L during coast ~1 per 100 interactions produce  bb B-physics programme includes: CP violation measurements in B-decays Flavour Oscillations in B s Searches and measurements of very rare decays Precision Measurements Production Measurements Requires a Highly Selective and Flexible Trigger pp 14 TeV Introduction – LHC & ATLAS LHC Ring

4. ATLAS B-Trigger John Baines RAL The ATLAS Detector 11m 23m

5. ATLAS B-Trigger John Baines RAL CoM Energy14 TeV Design Luminosity10 34 cm -2 s -1 Interactions per x-ing  23 Bunch spacing25 ns High rate (40 MHz) High granularity  large event size (1-2 MBytes) Trigger/DAQ at the LHC

6. ATLAS B-Trigger John Baines RAL ATLAS Trigger StrategyHLT

7. ATLAS B-Trigger John Baines RAL ATLAS Trigger/DAQ – Overview > Latency: 2.5s (max) > Hardware based (FPGA, ASIC) > Calo/Muon (coarse granularity) > Latency: ~10 ms (average) > Software (specialised algs) > Uses LVL1 Regions of Interest > All sub-dets, full granularity > Emphasis on early rejection > Latency: few sec (average) > Offline-type algorithms > Full calibration/alignment info > Access to full event possible LVL1 LVL2 EF

8. ATLAS B-Trigger John Baines RAL Event Selection relies on: –Processing in Steps Alternate steps of feature extraction / hypothesis testing Events can be rejected at any step if features do not fulfil certain criteria (signatures) –Reconstruction in Regions of Interest (RoIs) RoI size/position derived from previous step(s) HLT Strategy Emphasis on early event rejection Emphasis on minimising a. Processing time b. Network traffic

9. ATLAS B-Trigger John Baines RAL Region of Interest (RoI) mechanism LVL1 finds an EM cluster in the calorimeter or a muon track in the external muon spectrometer LVL2 uses LVL1 info to define a region LVL2 accesses data for that region (a few percent of the total)

10. ATLAS B-Trigger John Baines RAL HLT Strategy – Example Iso– lation pt> 30GeV Cluster shape track finding Iso– lation pt> 30GeV Cluster shape track finding EM20i + e30i + e30 + e e + ecand + t i m e Signature  Level1 seed  STEP 1 STEP 4 STEP 3 STEP 2 Strategy at HLT: > Validate step-by-step > Check intermediate signatures > Reject as early as possible Sequential/modular approach facilitates tuning for early rejection LVL1 claims two isolated e/m clusters with pT>20GeV (possible signature: Z–>ee)

11. ATLAS B-Trigger John Baines RAL HLTSSW – Design HLTSSW Steering ROBData Collector Data Manager HLT Algorithms Processing Application Event DataModel Processing Application Event Filter HLT Core Software HLT Algorithms Level2 HLT Selection Software HLT DataFlow Software Interface Dependency Package HLTSSW runs on the L2/EF processors HLTSSW runs on the L2/EF processors Several external dependencies Several external dependencies (Monitoring Svc, MetaData Svc, offline…)

12. ATLAS B-Trigger John Baines RAL Core Components Steering –Controls the order in which HLT algorithms should run, given the result of the previous triggering step All possible signatures form the Trigger Menu All possible sequences form the Sequence Table RegionSelector – select detectors in RoI DataStore – stores data produced by each processing step Convertors – convert raw data to input objects on demand – “Lazy unpacking”

13. ATLAS B-Trigger John Baines RAL HLT Select. Software: Components HLTSSW Steering ROBData Collector Data Manager HLT Algorithms Processing Application Event DataModel Processing Application Event Filter HLT Core Software HLT Algorithms Level2 HLT Selection Software HLT DataFlow Software Interface Dependency Package

14. ATLAS B-Trigger John Baines RAL Data access by an HLT algo Algorithm Region Selector HLT Algorithm Region Selector Trans. Event Store Data Access Byte Stream Converter Data source organized by ROB Transient EventStore region list DetElem IDs ROB ID raw event data Data arranged by DetElems list DetElem IDs Data arranged by DetElems DetElems: Detector Elements (e.g. Pixel wafers) IDs: Identifiers – Allow access to Geometry and mapping to ROBs For the Event Filter: data already in the TES

15. ATLAS B-Trigger John Baines RAL Event Data and Algs Closely coupled to offline software –Common class definitions (Track, Cluster etc) Facilitate code migration between LVL2/EF/Offline Saves effort in development and maintenance Makes comparisons and performance studies easier –Same arguments for data access mechanism However, special online requirements (esp. LVL2) –Timing –Multi-threaded running

16. ATLAS B-Trigger John Baines RAL Exercising the HLTSSW 1.LVL1 seeding Get list of Read-out Buffers (ROBs) for RoI, initially identified by LVL1 2.Data access Network retrieval of raw data from ROBs, on demand. 3.Data preparation Unpacking of raw data into objects convenient for the reconstruction algorithms, again done on demand. Calibration of the data objects. 4.Algorithm Perform feature extraction and then hypothesis validation. For example, cluster finding and identification of the cluster. 5.Take Trigger Decision

17. ATLAS B-Trigger John Baines RAL ByteStream converters Bytestream conversion process: –Raw data (as received from the detector electronics)  Data objects (convenient for the HLT algorithms) –Example of data objects: Calorimeter cells with energy, position, etc… Example of offline code used in online. –The interface to the converter is the ATLAS offline software Transient Data Store. Creation on demand. –When the converter is called objects are only created from raw data as needed. Caching of data. –The conversion will only happen if the data objects are not in memory already.

18. ATLAS B-Trigger John Baines RAL ATLAS Trigger Architecture Implementation Higher Level Trigger Hardware (ASIC/FPGA) General Purpose Processors : offline type algorithms General Purpose Processors optimised algorithms 10 8 10 9 Hz 2 x < 2.5  s ~ 10 ms ~ few sec Decision times FPGA = Field Programmable Gate Array ASIC = Application Specific Integrated Circuit

19. ATLAS B-Trigger John Baines RAL ATLAS B-Physics Programme CP Violation: Measurement of Asymmetry in: B d J/  (ee) K 0 (  ) B d J/  (  ) K 0 (  ) Control channels: B + J/  (  )K + B d J/  (  )K 0 *(K +   ) & equiv. (ee) Measurement of Asymmetry in: B d     + other hadron final states Analysis of B s J/  (  )  (KK) B s,d J/  (  )  (  ) Rare decays of the type : B d,s  (X) Branching fraction for B d,s     Branching fractions for: B d  0  and B d K *0  F-B Asymmetry in  d    Precision measurements, eg. B c measurements : B c J/  , B c J/    b polarisation  b J/  (  )  o (p  ) The following have been evaluated for possible inclusion in the ATLAS B-physics programme: sin2  CP-viol. ampl. a,b (sin2   s  s A // A  2 -  1  Measurement of B s oscillations: B s D s  and B s D s a 1 with D s  o  K  K    msms |V td | / |V ts | Note: The B-physics programme will be discussed in detail in the ATLAS Physics Overview talk. Searches for: B K + K -  -

20. ATLAS B-Trigger John Baines RAL History & Recent Developments l Target for LHC startup luminosity is now 2 x 10 33 cm -2 s -1, although: Actual start-up luminosity uncertain Luminosity may vary fill to fill During a coast luminosity is expected to fall by a factor ~2  Re-evaluate trigger thresholds (single muon p T threshold and ID reconstruction thresholds)  Assess impact of removing triggers requiring the most resources (e.g. J/  (ee))  Develop flexible trigger strategies - possibility to include more B-triggers as luminosity falls, e.g. di-muon trigger only at L = 2x10 33 cm -2 s -1 add other triggers as L decreases (e.g. B(  ), D s (  ) based on ID full-scan or low ET RoI) Financial constraints  Investigate new possibilities for reducing resource requirements e.g. Low E T Level-1 calorimeter RoI used to guide reconstruction at Level-2 Level-2 RoI used to limit region for reconstruction at the Event Filter Possibility of reduced detector at start-up: TRT only at |  | < 2 (c.f. full TRT |  | < 2.5) Only 2 of the 3 Pixel layers (inner “B-layer” maintained)  Need flexibility to cope with evolution of detector ATLAS B-trigger strategy outlined in the DAQ and High Level Trigger Technical proposal in 2000 Since then, LHC delayed, start-up now expected in 2007  Main c.p.u purchase delayed to 2006 (cheaper/faster c.p.u – expect 180 SpecInt95 = 4.5 GHz PC) ~ ~ RoI = Region of Interest

21. ATLAS B-Trigger John Baines RAL Outline Trigger Strategy : Final states including two muons Hadron Final States Muon Trigger Rejection of  from  /K decays Based on ID full-scan: Inner Detector (ID) ID Full-Scan B d (  ) trigger D s (  (KK),  ) trigger Alternative based on Calorimeter RoI Final states with electrons and muons TRT Full-scan J/  (ee) trigger Alternative using calorimeter EM RoI ATLAS B-TRIGGER UPDATE RoI = Region of Interest

22. ATLAS B-Trigger John Baines RAL Level-1 Level-2 & EF Confirmation of Muons in: Precision Muon Chambers Inner Detector Trigger for Di-Muon Final States At Least 2 Muons: Minimum thresholds: Muon Barrel: p T > 5 GeV Muon End-Cap: p T > 3 GeV Actual thresholds used will be determined by rate limitations ~ ~ Level-1 Di-Muon triggers mainly due to: muons from heavy flavour decays single muons giving double trigger in end-cap trigger chambers Rate ~600 Hz (p T > 6 GeV threshold, L = 2 x 10 33 cm -2 s -1 ) EF Selections: Refit ID tracks in Level-2 RoI Decay vertex reconstruction Select J/  (  ) B(  ) etc. using mass & decay length cuts Search for hadrons from B K 0 *(h,h) , etc. Select using mass cuts B d J/  (  )(K/K*) B s J/  (  )  B  B K 0 * , etc.  b  0 J/  (  ) B c J/  (  )  e.g. Total Rate ~ 20 Hz ( L = 2 x 10 33 cm -2 s -1 )

23. ATLAS B-Trigger John Baines RAL Hadron Final States Full-Scan RoI-Guided Refit ID tracks in Level-2 RoI Decay vertex reconstruction Mass & Decay length cuts. Level-1 Level-2 & EF At Least 1 Muon: p T > 6 - 8 GeV Confirmation of Muon in: Precision Muon Chambers Inner Detector Full-Scan of Inner Detector Mass cuts Refit ID tracks in Level-2 RoI Decay vertex reconstruction Mass & Decay length cuts Confirmation of Muon in: Precision Muon Chambers Inner Detector Confirmation of Jet in calorimeter Scan of ID in Jet RoI Mass Cuts Options B d     B s D s  B s D s a 1 D s  o ,  o K  K  >1 Muon: p T > 6 - 8 GeV Plus: >1 Jet cluster E T > 5 GeV ~ e.g.

24. ATLAS B-Trigger John Baines RAL The Muon Trigger Level-1 trigger from Muon Trigger Chambers Muon confirmed at Level-2 using Precision Muon Chamber Data 20 points per , resolution ~ 80  m Better track measurement allows tighter threshold. Muon confirmed in Inner Detector: Extrapolate Muon track to ID, Search for ID track. Combine parameters & apply matching cuts. Inner Detector Muon Trigger Chambers (RPC) Muon Precision Chambers (MDT) RPC: Restive Plate Chambers TGC: Thin Gap Chambers MDT: Monitored Drift Tubes Muon Trigger Chambers (TGC)

25. ATLAS B-Trigger John Baines RAL Level-1 Single Muon Trigger: Rate: ~20kHz p T > ~6 GeV @ L=10 33 cm -2 s -1 –Most are  from  /K decay with true p T < 6 GeV Level-2 Muon Confirmation: Using Precision Muon Detector info.: –Better track measurement allows tighter threshold. Rate: ~9 kHz 35% b  and c , 65%  /K  Using Combined Muon & ID info: Single Muon Trigger Rate: ~5 kHz @ L=10 33 cm -2 s -1 ~50%  /K  Rejection of  /K  decays LVL2 muon standalone LVL2 muon + ID Efficiency (%) 100 80 60 40 20 0 100 80 60 40 20 0 0 2 4 6 8 10 12 14 muon p T (GeV) Raising threshold by 2 GeV  ~ factor 2.5 rate reduction e.g. p T > ~8 GeV (  8) Rate ~ 2 kHz @ L=10 33 cm -2 s -1

26. ATLAS B-Trigger John Baines RAL An HLT Algorithm: T2Calo LVL2 clustering algorithm for electromagnetic (EM) showers, seeded by LVL1 EM RoI positions. Main variables built: (1) Energy of EM clusters (2) Associated Hadronic Energy

27. ATLAS B-Trigger John Baines RAL An HLT Algorithm: T2Calo (cont) (3) E3x7/E7X7 in Layer 2 background (4) (E1-E2)/(E1+E2) in Layer 1 signa l

28. ATLAS B-Trigger John Baines RAL System performance Conditions of the measurements: –RoI of (  x  = (0.3 x 0.3). –Dijet events at low luminosity with pileup. –Machine: CPU 2.4 GHz Xeon with 1 GByte of memory. Measure minimum time that the data converter function would take using no offline-inherited code. Remember: Average LVL2 processing budget is ~10 ms. Largest contribution is from Data Preparation Algorithm is the smallest contribution New offline-compatible version incorporating these and other improvements approaches performance requirements (now 3-4 ms for data preparation)

29. ATLAS B-Trigger John Baines RAL SemiConductor Tracker (SCT) Si micro-strip detector: 6.4 cm x 12.8 cm. 80  m r/o pitch. Barrel: 4 cylinders; End-cap: 9 Wheels each with 2 stereo layers:  + u or v (40 mRad) 8 hits along track i.e. 4 space-points. ATLAS Inner Detector See Inner Detector Talk for more details 3.5m  =2.5 1.2m Transition Radiation Tracker (TRT) : Straws 40 cm - 70 cm long filled with Xe/CO 2 /CH 4. Single sense wire per straw. ~36 measurements along track. Two readout thresholds - Electron ID via higher threshold Transition Radiation hits Pixel Detector : Si wafer: 2.1 cm x 6.5 cm with 50  m x 300/400  m pixel r/o. 2(3) measurements along track. Inner layer at R = 5.05 cm

30. ATLAS B-Trigger John Baines RAL Start with the Pixels and SCT: Less affected by material interactions 3-D measurements Full |  | coverage from start-up IDscan - Inner Detector full-scan SCT Pixels Hit Filter: Forms groups of hits - group contains the hits from a track; may also contain some extra nearby hits Use muon track to define Z of vertex of primary i.p. Form 2D    histogram of SCT & Pixel hits Select hits in bins with >3 layers hit Group hits from neighbouring bins Group Cleaner: Select the hits from a group forming a track candidate Determine  0 and  for hit triplets Fill 2D histogram in  0,1/p T Bin with hits from >4 layers => track candidate Track Fit: Determine track parameters IDscan Algorithm

31. ATLAS B-Trigger John Baines RAL No. Pixel Layers 3 2 Efficiency for B  events with pile-up ( L = 10 33 cm -2 s -1 ): B  events, p T ( ,  ) > 4 GeV B  events selected offline Background B  X events Level-2 rate ~20 Hz for 2 kHz  8 EF selection reduces rate to ~3 Hz Event Filter Selection: Tighter mass cut Vertex fit cuts :  2 / N d.o.f. < 8, L xy > 100  m,  xy < 5 B h + h - Trigger Level-2 selection: Tracks separated from trigger  by  R  > 0.2 p T > 3.9 GeV o Two opposite sign tracks with: p T + p T > 10 GeV  z 0 < 2 cm 4.3 < M(  ) < 6.3 GeV + - 10 15 20 25 30 35 40 45 50 B d p T spectrum both  p T >4 GeV 100 80 60 40 20 0 Level-2 Efficiency (%) 0 2 4 6 8 M(      (GeV) B d  Events + min. bias No. Events  xy Decay Vertex y x L xy 78% 80% 89% 93% 1.1% p T of B d (GeV)

32. ATLAS B-Trigger John Baines RAL No. Pixel Layers 2 3 69% 68% 78% 79% 3.5% 3.8% D s  Trigger o Level-2 Selection: Tracks: p T > 1.4 GeV  R  > 0.2 w.r.t. trigger  Two opp. sign tracks satisfying: M(KK) - M(  ) < 17 MeV Third track with: M(KK  ) - M(D s ) < 74 MeV K+K+ EF Event Selection: Tracks: p T > 1.5 GeV Mass cuts : M(KK) - M(  ) < 14 MeV M(KK  ) - M(D s ) < 56 MeV  vertex fit cuts:  2 prob. >0.5%, L xy >200  m D s vertex fit :  2 prob>0.5%, L xy > 200  m Level-2 Efficiencies for B s D s (  (KK))  events with pile-up ( L = 10 33 cm -2 s -1 ): Signal events with p T ( ,  and K) > 1.5 GeV Signal events selected offline Background B  X events Level-2 Trigger rate ~60 Hz for 2 kHz  8 EF selection reduces rate to ~9 Hz Signal Events + min. bias Events / 2.5 MeV Events / 20 MeV 0.95 1 1.1 1.2 M(KK) (GeV) Signal Events + min. bias 1.25 1.5 1.75 2 2.25 M(KK  ) (GeV)

33. ATLAS B-Trigger John Baines RAL Alternative Using Level-1 Jet RoI to guide B-physics Triggers LVL2 reconstruction inside RoI corresponding to ~10% of ID acceptance  potential to save ~factor 10 execution time c.f. full-scan  but with lower efficiency Preliminary studies of an alternative to the full- scan using, instead, low E T Level-1 RoI to define regions to search ID at LVL2 Studied using fast simulation + parameterisation of calorimeter Jet RoI (0.8 x 0.8 cluster) E T > 5 : Mean Multiplicity = 1.7 (B  X events,  p T > 6 GeV) Reconstruct tracks at Level2 inside regions e.g. for B(  ) and D s (  ) Jet RoI Multiplicity (E T > 5 GeV) Jet RoI Multiplicity 0 1 2 3 4 5 6 7 8 9 No. Events ~

34. ATLAS B-Trigger John Baines RAL Using Level-1 Jet RoI to guide B-physics Triggers B   p T   4 GeV RoI E T  5 GeV B hadron p T (GeV) 1 0.8 0.6 0.4 0.2 0 Efficiency 0 5 10 15 20 25 30 Actual efficiencies and c.p.u. savings depend on thresholds & multiplicities => to be studied using full simulation Efficiency B  D s  p T Ds,   1 GeV RoI E T  5 GeV B hadron p T (GeV) 0 5 10 15 20 25 30 1 0.8 0.6 0.4 0.2 0 Efficiency for Jet RoI within |  | < 0.4, |  | < 0.4 of B hadron Based on fast simulation + calorimeter parameterisation

35. ATLAS B-Trigger John Baines RAL RoI Guided Reconstruction at the Event Filter Following Level-2 B-trigger selection: Use Level-2 to guide reconstruction at the Event Filter Level-2 defines a region which contains all tracks forming D s (  ), B(  ) candidates Region corresponds to ~10% of the Inner Detector acceptance  Factor ~10 saving in resources compared to full reconstruction B d  +  -   0 0.5 1 1.5 2 2.5 3 2 1.5 1 0.5 0  D s  (K + K - )   2 1.5 1 0.5 0 0 0.5 1 1.5 2 2.5 3 

36. ATLAS B-Trigger John Baines RAL Electron & Muon-Electron Final States  bb  X B d (J/  (ee)K o )  bb eX B d (J/  (  )K 0 ) e.g. Full-Scan RoI-Guided Options Refit ID tracks in Level-2 RoI Decay vertex reconstruction Mass & Decay length cuts Level-1 Level-2 & EF At Least 1 Muon: p T > 6 - 8 GeV Confirmation of Muon in: Precision Muon Chambers Inner Detector Full-Scan of Inner Detector (SCT, Pixels & TRT) Mass cuts Refit ID tracks in Level-2 RoI Decay vertex reconstruction Mass & Decay length cuts >1 Muon: p T > 6 - 8 GeV Plus: >1 EM cluster E T > 2 GeV Confirmation of Muon in: Precision Muon Chambers Inner Detector Confirmation of electron in EM RoI using: Calorimeter Inner Detector Possible search for second electron

37. ATLAS B-Trigger John Baines RAL TRT-SCAN Histogram for a single muon  1/p T Each hit straw populates a set of bins forming a line in the histogram in transform space (  0,1/p T ) or (  0,1/p L ). Maxima in the histogram correspond to track candidates. Set of trajectories through a straw Track candidates are examined and can be split or merged if required. A track fit is performed to improve the track parameter resolution. Drift time information can be included at this stage. Track search using a Histogramming method based on a Hough Transform. Track trajectories are described by: (  0,p T ) - barrel and (  0,p L ) - endcap. Sets of trajectories are defined with discrete steps in  0 and 1/p T or 1/p L. Each trajectory corresponds to a histogram bin. No. of hits along trajectory (  0, 1/p T ) 30 20 10 0 Execution time scales linearly with: inverse p T (p L ) threshold no. hits in event

38. ATLAS B-Trigger John Baines RAL Event Filter Selection: Tighter mass cuts Vertex fit cuts :  2 / N d.o.f. < 8, L xy > 220  m,  xy < 40 Two opposite-sign e tracks with: p T + p T > 4 GeV |  | < 1.4, |  z 0 | < 2 cm cos(  ee ) > 0.2 2 < M(ee) < 3.5 GeV + - J/  e  e  Trigger Level-2 Selection: Tracks: p T > 0.5 - 1.5 GeV Identified as electrons by TRT Level-2 Efficiencies for B d J/  (ee)K s events with pile-up (L = 10 33 cm -2 s -1 ) Recon. tracks p T > 0.8 GeV: 40% : Signal events with p T (e,e) > 1 GeV 53% : Events selected offline 2% : Background B  X events Level-2 Trigger rate ~40 Hz for 2 kHz  8 EF selection reduces rate to ~4 Hz o Signal Events + Min. Bias 0 1 2 3 4 5 M(ee) (GeV)

39. ATLAS B-Trigger John Baines RAL RoI E T > 2 GeV 90% Efficiency for e from B e e + e - separation  0 1 2 3 No. Events  separation J/  e+e- e p T > 0.5 GeV 0 0.5 1 1.5 2 e + e - separation  No. Events  separation J/  e+e- e p T > 0.5 GeV electron p T (GeV) 0 2 4 6 8 10 12 14 1 0.8 0.6 0.4 0.2 0 Efficiency RoI Multiplicity EM RoI E T > 2 GeV RoI Multiplicity 0 1 2 3 4 5 6 7 8 9 No. Events If only one e found at Level-1, could search larger region for 2 nd e  Level-2 Reconstruction in ~10% of ID acceptance If both electrons found at Level-1: confirmation at Level-2 inside small region about each electron Alternative Using Level-1 EM RoI Preliminary study of the possibility of using calorimeter to provide RoI to search for low p T electrons at level-2 for J/  (ee) and  -e EM RoI E T >2 : Mean Multiplicity = 1.1 (B - >  X,  p T > 6 GeV) Effic. to tag both e in J/  (e,e) : 80% (e p T >3 GeV) Fast Simulation + Calorimeter Parameterisation

40. ATLAS B-Trigger John Baines RAL Execution Times Execution times measured on a different platforms Determine scaling with occupancy Times scaled to a 4 GHz PC (assumed 160 SpecInt 95) Used to estimate resources needed for B-trigger L=10 33 cm -2 s -1 Time (ms) muon0.2 SCT in Muon RoI0.3 IDscan11 Event Filter6 Execution Times Scaled to 4GHz PC 0 20 40 Execution Time (ms) No. Events Execution Time (ms) 30 20 10 0 0 2000 4000 6000 8000 10000 No. space-points in Event IDscan Linear scaling with occupancy

41. ATLAS B-Trigger John Baines RAL Resource Estimates Study several options for B-physics triggers: Chosen to represent a broad range of possibilities Do not necessarily reflect final choices di-muon trigger when L > 2x10 33 cm -2 s -1 Add B(hh) and D s (  ) triggers for L < 2x10 33 cm -2 s -1 based on ID full-scan for events with muon p T > 8 GeV di-muon triggers only Minimal additional resources Requires some additional resources B-trigger based on level-1 RoI : di-muon trigger B(  ), D s (  ), J/  (ee) etc. based on Level-1 EM Jet & EM RoI Requires additional resources, but less than (2) B-trigger based on level-1 RoI: di-muon trigger when L > 2x10 33 cm -2 s -1 Add B(hh), D s (  ), J/  (ee) etc. triggers for L < 2x10 33 based on Level-1 EM Jet and EM RoI Modest additional resources Resources additional to those needed for full menu of high p T triggers (1) (2) (3a) (3b) ~ ~ ~ ~

42. ATLAS B-Trigger John Baines RAL Summary The ATLAS B-trigger strategy was outlined in the DAQ and High Level Trigger Technical proposal in 2000 Since then, B-trigger strategy has been re-assessed in the light of: LHC luminosity target for start-up doubled to 2 x 10 33 cm -2 s -1 Detector changes, including possibility of reduced detector at start-up Need to minimise trigger resources in the light of financial constraints  Develop flexible B-trigger strategies to: Cope with evolution of detector Provide possibility of adding more B-triggers as luminosity falls  Investigate new possibilities for reducing resource requirements e.g. Low E T Level-1 calorimeter RoI used to guide reconstruction at Level-2 Level-2 RoI used to limit region for reconstruction at the Event Filter As a result ATLAS hopes to pursue a full programme of B-physics from LHC start-up Next review stage is the Technical Design Report (2003) – some architectural choices must be made based on physics simulation studies, prototyping and full system modelling. Ready for first Physics in 2007