The trigger system of the ATLAS experiment

Transcription

1 TU Dresden, IKTP Institutsseminar, The trigger system of the ATLAS experiment Johannes Haller (Universität Hamburg)

2 Outline Important questions in particle physics The Large Hadron Collider at CERN The ATLAS experiment Installation, Commissioning, Status Answers to the questions? The ATLAS Trigger system Constraints for the design Technical implementation LVL1 HLT Expected performance Status, current activities Summary Johannes Haller ATLAS Trigger 2

3 Important questions in particle physics Completion of the Standard Model : SM describes all processes in particle physics SU(3)xSU(2)xU(1) gauge theory Precise prediction in low and high energy regime Extremely successful for 2-3 decades Irreplaceable component is the Higgs Boson Higgs Boson mass expected ~ 100 GeV Extension of the Standard Model. We know that SM is not the full story Gravity is missing Hierarchy problem Experimentally: CDM in the universe Need SM extensions (e.g. SUSY, ED, etc.) Expected below ~1 TeV Most important questions: Search for the Higgs boson Search for New Physics, new particles Johannes Haller ATLAS Trigger 3

4 The Large Hadron Collider For studies/searches of Higgs and New Physics we nee colliders with highest energies LHC ILC highest energy particle collider ever built pp collisions at s = 14 TeV, L = cm -2 s -1 7 x energy, 100 x luminosity of Tevatron LHC = the discovery machine First collisions scheduled for Mid 2008 covers energy regime up to 2-3 TeV. Johannes Haller ATLAS Trigger 4

5 Important Event Signatures at the LHC Production of Higgs Bosons at LHC Various production modes: E.g. gluon-gluon-fusion, VBF, etc. Gold-Plated decay channels: 4-Lepton-Final state 2-photon final state Production of supersymmetric particles Pair Production of squarks and gluinos via strong interaction Decay via long decay chains into LSP: Jets (squarks, gluinos) Missing transverse energy (LSP) Leptons p H µ + µẕ µ + µ - q q Important experimental signatures: muons, photons, electrons, jets, missing E T X Z g q p p q L χ 0 2 p H γ l R l γ no interaction with detector E T,miss p χ 0 1 l Johannes Haller ATLAS Trigger 5

6 The ATLAS Detector 22 m 40 m largest collider detector ever built Characteristic features: Muon spectrometer with sophisticated toroidal magnets (H 4 μ) Highly segmented LAr electromagnetic calorimeter (H 4l, H 2γ) Johannes Haller ATLAS Trigger 6

7 Building the ATLAS detector Feb 2003 Sep 2005 Sep 2005 Sep 2005 Johannes Haller ATLAS Trigger 7

8 Building the ATLAS detector today today Construction and assembly at the surface has really come to an end Installation in the cavern is also proceeding and nearing completion well along the planning. Johannes Haller ATLAS Trigger 8

9 Higgs and SUSY: example performance of ATLAS Higgs Boson: Gold-Plated Channels: 4-Lepton-Final state 2-photon final state Similar situation for searches for New Physics: E.g. SUSY Signals clearly visible in distributions Combined significances show that the Higgs Boson can be discovered with ATLAS after ~3-4years of data taking Almost all scenarios with new physics below 2-3 TeV can be discovered. Johannes Haller ATLAS Trigger 9

10 Event Rates and Multiplicities cross section of p-p collisions s tot (14 TeV) 100 mb s inel (14 TeV) 70 mb LHC cm energy (GeV) R = event rate L = luminosity = cm -2-2 s -1-1 σ inel inel = inel. Cross section = 70 mb N = interactions // bunch crossing Δt = bunch crossing interval = 25 ns R = L x σ inel inel = cm -2-2 s -1-1 x 70mb = Hz Hz N = R // Δt Δt = s -1-1 x s = 17.5 = 17.5 x 3564 // 2808 (not all all bunches filled) = interactions // bunch crossing (pileup) With every bunch crossing 23 Minimum Bias events with ~1700 particles produced n ch ch = charged particles // interaction N ch ch = charged particles // BC N tot tot = all particles // BC n ch ch 50 N ch ch = n ch ch x 23 = ~ 1150 N to to = N ch ch x 1.5 = ~ 1725 Johannes Haller ATLAS Trigger 10

11 Looking for Interesting Events No problem in the beginning when luminosity is low, but detector (and trigger) must be designed to cope with this Higgs ZZ 2e+2μ 23 min bias events Johannes Haller ATLAS Trigger 11

12 another Constraint: ATLAS Event Size pile-up, adequate precision need small granularity detectors Detector Pixels SCT TRT TGC 4.4*10 5 Atlas event size: 1.5 MB (140( 140 million channels) at 40 MHz: 1 PB/sec affordable mass storage: < 300 MB/sec storage rate: < 200 Hz 3 PB/year for offline analysis LAr Tile MDT CSC RPC LVL1 Channels 1.4* * * * * * *10 5 Fragment size [KB] Johannes Haller ATLAS Trigger 12

13 The Trigger Challenge σ total interaction rate rate IA IA rate:~ 1 GHz; BC rate: MHz; storage:~ 200 Hz Hz online rejection: % (!) (!) crucial for for physics goals(!) storage rate discoveries E T powerful and flexible trigger system needed: enormous rate reduction retaining the rare events in in the very tough LHC environment remember: must be shared: physics triggers -high p T physics T (un-pre-scaled) -low p T physics T (pre-scaled, excl.) technical triggers: --monitor triggers --calibration triggers - - Johannes Haller ATLAS Trigger 13

14 Technical Implementation Johannes Haller ATLAS Trigger 14

15 ATLAS Trigger: Overview 3-Level Trigger System: software hardware 2.5 μs ~ 10 ms ~ sec. 1) 1) LVL1 decision based on on data data from fromcalorimeters and and muon muontrigger chambers; synchronous at at MHz; MHz; bunch crossing identification 2) 2) LVL2 uses usesregions of of Interest (identified by by LVL1) data data (ca. (ca. 2%) 2%) with with full full granularity from from all all detectors, asynchronous 3) 3) Event Filter (L3) (L3) has has access to to full full event event and and can can perform more more refined event event reconstruction Johannes Haller ATLAS Trigger 15

16 LVL1 Trigger Overview Calorimeter trigger Jet / Energy-sum Processor Pre-Processor (analogue E T ) multiplicities of e/γ, τ/h, jet for 8 p T thresholds each; flags for ΣE T, ΣE T j, E T miss over thresholds Cluster Processor (e/g, t/h) Muon Barrel Trigger (RPC) Central Trigger Processor (CTP) Muon-CTP Interface (MuCTPI) Muon trigger multiplicities of μ for 6 p T thresholds L1A signal Muon End-cap Trigger (TGC) LVL1 latency: 2.5 μs = 100 BC TTC TTC TTC TTC TTC Johannes Haller ATLAS Trigger 16

17 LVL1 Calorimeter Trigger electronic components (installed in counting room outside the cavern; heavily FPGA based): available example: thresholds: e/γ algorithm: EM (e/gamma): goal: good 8 discrimination - 16 Tau/ hadron: 0 e/γ - 8 jets Jets: identify 8 2x2 RoI with local fwd. Jets: 8 E T maximum E sum cluster/ isolation cuts on T, E sum T (jets), E miss T : 4 (each) various E T sums PPM crate output: at 40 MHz: multiplicities for e/γ, jets, τ/had and flags for energy sums to Central Trigger (CTP) accepted events: position of objects (RoIs) to LVL2 and additional information to DAQ 7 JEMs 6 CPMs Johannes Haller ATLAS Trigger 17

18 LVL1 Muon Trigger algorithm: dedicated muon chambers with good timing resolution for for trigger: Barrel η <1.0 :: Resistive Plate Chambers (RPCs) End-caps 1.0< η <2.4 :: Thin Gap Chambers (TGCs) local track finding for for LVL1 done ondetector (ASICs) looking for for coincidences in in chamber layers programmable widths of of 6 coincidence windows determines p T threshold Johannes Haller ATLAS Trigger 18

19 LVL1 Trigger Decision in Central Trigger Processor CTP: (one 9U VME64x crate, FPGA based) central part of LVL1 trigger system CTP in USA15: signals from LVL1 systems: Multiplicities of 8-16 EM, 0-8 TAU 8 JET, 8 FWDJET 4 XE, 4 JE, 4 TE, 6 Muon calculation of trigger decision for up to 256 trigger items: e.g. XE70+JET70 raw trigger bits other external signals e.g. MB scintillator, internal signals: 2 random rates 2 pre-scaled clocks 8 bunch groups application of veto/ dead time application of pre-scale factors actual trigger bits L1A Central Trigger Processor Johannes Haller ATLAS Trigger 19

20 Interface to HLT: RoI Mechanism LVL1 triggers on (high) p T objects L1Calo and L1Muon send Regions of Interest (RoI) to LVL2 for e/γ/τ-jet-μ candidates above thresholds to LVL2 via optical fibers. LVL2 uses Regions of Interest as seed for reconstruction (full granularity) only data in RoI are used advantage: total amount of transfered data is is small ~2% of the total event data can be dealt with at 75 khz Event Filter (ie. Level-3) runs after event building, full access to event Johannes Haller ATLAS Trigger 20

21 HLT Selection Strategy fundamental principles: Example: Dielectron Trigger 1) 1) step-wise processing and decision inexpensive (data, time) algorithms first, complicated algorithms last. early reject 2) 2) seeded reconstruction algorithms use results from previous steps initial seeds for LVL2 are LVL1 RoIs LVL2 confirms & refines LVL1 EF confirms & refines LVL2 note: EF tags accepted events according to to physics selection ( streams, offline analysis!) ATLAS trigger terminology: Trigger chain Trigger signature (called item in LVL1) Trigger element Johannes Haller ATLAS Trigger 21

22 ATLAS Trigger & DAQ Architecture Johannes Haller ATLAS Trigger 22

23 RODs Front-end LVL2 SFI (s) EF EF EF EF SFOs Technical example: Trigger Configuration LVL1/CTP Event Builder L1Result Streams TriggerDB TriggerDB DbProxy Server/ rack System s functionality: Configuration of full trigger chain for online data taking Integrated system for LVL1 and HLT Distribution of trigger settings to online clients Online operation of the trigger (shift crew, experts) Configuration of Trigger emulation in MC production Archival of trigger settings Storage of trigger data (trigger bits) in events System s components: Trigger Run Control, called TriggerTool TriggerDB: relational DB Software for fast online distribution of settings for storage and interpretation of trigger results in event data. Trigger Run Control: Johannes Haller ATLAS Trigger 23

24 Expected Physics Performance Derived from simulation Concentrating on electrons and muons Johannes Haller ATLAS Trigger 24

25 Electron Performance LVL1 Efficiency Trigger Rate /(KHz) Rates for 2x10 33 cm -2 s -1 No isolation Isolated Electron p T (GeV) e/γ p T trigger threshold(gev) rather steep turn-on curve up up to to ~ 100% noise (pile-up) slightly decreases performance isolation important to to reduce rate general trigger problem: cover phase space while keeping the triggers rates low. Johannes Haller ATLAS Trigger 25

26 Electron Performance LVL2 E 3x7 /E 7x7 (E 1 -E 2 )/(E 1 +E 2 ) E T,EM E T, HAD LVL2 gets full granularity data in EM RoI: 3 samplings in EM with varying granularity various estimators can be used (mostly 2 nd sampling, cuts optimised) After cuts: strip towers of sampling 1 used to discriminate γ/π (important for H γγ) π 0 γ Johannes Haller ATLAS Trigger 26

27 Electron Performance HLT Rejection factor for jets Cut optimisation for L2Calo L2Calo Efficiency for Electrons Combined electron performance for HLT L1 L2Calo L2IDCalo EFCalo EFID EFID& Calo Efficiency 98.8% 96.8% 91.7% 90.4% 89.7% 86.3% Rate 5 khz 1.1KHz 192Hz 180Hz 170Hz 45.5Hz Varying the cuts to find optimum Compromise between rate and efficiency Rate reduces during stepwise processing New methods under study (MVA, Neural Networks) eff. wrt MC truth Johannes Haller ATLAS Trigger 27

28 Important: Determination of Trigger efficiencies from data Preparing to use data (not MC simulation!) to understand the efficiency of the trigger Task: use only events that are recorded to find out which event are not recorded! tag & probe method for electron trigger: Trigger on a single electron ( tag ) Offline: select events with 2 good electrons with M inv = M Z ±20 GeV Determine trigger efficiency of second electron ( probe electron ) Z-Signal L= 100 pb -1 Bkgd from jets Trigger efficiency L1 L1+L2 M inv L1+L2+EF Johannes Haller ATLAS Trigger p (GeV) T 28

29 Muon Performance LVL1 Efficiency Eta Muon p T (GeV) Phi (rad) turn-on curve for muons reaches ~ 80 % reason for ineffi. are holes in muon system due to support structures Johannes Haller ATLAS Trigger 29

30 Muon Performance LVL2 Purpose: confirm LVL1 with full granularity data (MDT) in Muon RoI within 10 ms 1. select clusters of MDT tubes in a muon road around LVL1 trajectory. 2. fit straight line to each MDT chamber (using MDT timing info) super Point 3. estimate sagitta from three muon stations and use LUT for P T determination Approximated Approximated Muon Muon trajectory trajectory Performance: ΔP T /P T =5.5% for 6 GeV ΔP T /P T =4.0% for 20 GeV Johannes Haller ATLAS Trigger 30

31 Muon Performance LVL2 steep turn-on curves combination with inner detector possible on on LVL2: keeps high efficiency Steeper turn-on curve Johannes Haller ATLAS Trigger 31

32 Overall Trigger Strategy general trigger problem: cover as much as possible of the kinematic phase space for physics low trigger thresholds keep the trigger rate low high trigger thresholds trigger menu is a compromise menu for L=10 33 cm -2 s -1 ATLAS strategy: inclusive selection not to miss the unexpected reduce selection bias large uncertainties on predicted rates prescaled triggers not listed, monitor triggers not listed all in all 100 trigger signatures expected open and efficient for new physics preserving good rejection against background and SM processes with large cross section Johannes Haller ATLAS Trigger 32

33 Quick Status Report: LVL1 Calorimeter Trigger Big part of modules installed Full chain available Analog cables installed (big effort! ) Internal cabelling ongoing Muon Trigger: All chambers installed, some without power supplies Electronics installation ongoing Central trigger: Fully installed in USA15 Two spare system produced and functional in test lab Johannes Haller ATLAS Trigger 33

34 Quick Status Report: HLT Current status of HLT/DAQ farms: ~10% of final system available EF: 4x32 boxes (8core) (design: ~1500) L2: 2x32 boxes (4core) (design: ~500) Algorithms: Detailed optimisation studies ongoing Effort to optimize the full trigger menu. Due to financial constraints parts of the online farm are staged Consequences for physics: In 2008: maximum LVL1 rate: ~40 khz (instead of design 75/100 khz) Storage rate: ~80 Hz (instead of design 200 Hz) Johannes Haller ATLAS Trigger 34

35 Current activities at CERN: Integration and commissioning Big effort to test the combined data-taking (ie. all subdetectors) MC test data injected early in the system chain Real Data from cosmic muons. Dates Systems Integration Detector configuration Operations Cosmic run Training ATLAS Control Room M1: 11-19/ DAQ R/O Barrel Lar & Tile CTP Barrel calorimeters Achieve combined run 2 days Tile cosmic trigger N/A Initial setup: 5 desks Central DCS M2 28/2 to 13/ DAQ/EB DAQ V. 1.7 Muon barrel (S. 13) Monitoring/DQ Barrel calorimeters Barrel Muon Combined runs Mixed runs 2 x weekends Tile cosmic trigger + RPC cosmic trigger Periodic cosmic runs after M2 After M2 week Increase to 7 desks M3 4/6 to 18/ Barrel SCT Barrel TRT Muon EC (MDT, TGC) Offline Barrel and End Cap calorimeters Barrel muon (5&6) EC muon MDT Barrel SCT, TRT EC muon TGC 1st week focus on operations, checklist management, coordination between desks 1 week Tile + Muon cosmic trigger (side A) 4/6 to 11/6 Towards final layout: 13 desks M4 23/8 to 3/ day setup 2 week ends Level-1 Calo HLT DAQ 1.8 Offline 13 Barrel & EC calos Barrel & EC muon Barrel TRT SCT R/O Level-1 Mu, Calo ATLAS-like operations Use of DQ assessment 1 week Try also calorimeter trigger Whole week Final setup M5 22/10 to 5/ ID EC (TRT) Pixel (R/O only) SCT quadrant M4 + Pixel (R/O only, no detector) Week 1 system assessment Week 2 ATLASlike operation 1 week 1 week M6 February SCT and Pixel detectors ATLAS-like Operations Global cosmic run Johannes Haller ATLAS Trigger 35

36 ATLAS Control room Johannes Haller ATLAS Trigger 36

37 Just passed: M5 milestone week Combined running: all but one subsystem (Muon CSC) Highlights: L1 calorimeter trigger: integrated into L1 trigger Principle cosmic trigger from muon system. HLT included in the standard trigger Routine runs of > 100 kevents. especially under test: Trigger and DAQ. ATLAS trigger seems well prepared for first collision data in summer. Example event of a cosmic muon: Enormous effort is going into the commissioning and integration of the system. Many problems and solutions found. No show stoppers uncovered. On track Johannes Haller ATLAS Trigger 37

38 Summary With the LHC, particle physics enters a new exciting era. complete study of energy range up to 1 TeV possible fantastic perspectives LHC: discovery machine, (ILC needed for precision measurements) First data from LHC expected mid 2008 ATLAS experiment is on schedule Surface installations finished Some cabeling work ongoing in the caverne Test of data-taking with cosmics, DAQ/HLT test runs with all subdetectors. ATLAS trigger seems well prepared for first collisions Very high LHC interaction rate trigger is crucial for the physics results ATLAS: 3-level system with Region-of-Interest-mechanism. Inclusive trigger strategy based on high p T objects Looking forward to an exciting period for ATLAS and the entire HEP community with first LHC collisions scheduled for next year! Johannes Haller ATLAS Trigger 38