1. CMS101 Introduction to the CMS Trigger Darin Acosta University of Florida

2. CMS101 - Trigger, 13 Sep 2004Darin Acosta2 CMS Trigger & Data Acquisition LHC beam crossing rate is 40 MHz  1 GHz collisions CMS has a multi-tiered system to handle this: è Level-1 trigger reduces rate from 40 MHz to 100 kHz (max) p Custom electronic boards and ASICs process calorimeter and muon data to select objects è High-Level triggers reduce rate from 100 kHz to O(100 Hz) p Filter farm runs online programs to select physics channels p No custom Level-2 hardware 40 TB/s 100 MB/s Large switching network (~Tbit/s) O(1000) node PC cluster

3. CMS101 - Trigger, 13 Sep 2004Darin Acosta3 Trigger/DAQ Evolution CMS DAQ is a number of functionally identical, parallel, small DAQ systems è Build up 512  512 switch from 8 64  64 switches (1/8 th DAQ slice) Turn-on in 2007 expected to have 4 such slices (US contributes 1), yielding 50 kHz maximum L1A input rate to HLT è L1 fully scoped to deliver up to 100 kHz rate, and front-ends able to absorb 100 kHz rate

4. CMS101 - Trigger, 13 Sep 2004Darin Acosta4 The CMS Level-1 Trigger Reduces data rate from 40 MHz to 100 kHz è Suppress background 400X while keeping high P T physics è Requires custom electronic hardware, some of which must be radiation hard Only the muon and calorimeter systems participate è Select muons, electrons, photons, jets, and MET è Silicon tracker data is unavailable until the High-Level Trigger p Significant handicap compared to the Tevatron expts. Hardware and simulation results described in Level-1 Technical Design Report: è CERN/LHCC 2000-038 è Simulations based on GEANT and trigger emulation code

5. CMS101 - Trigger, 13 Sep 2004Darin Acosta5 Level-1 Trigger Scheme è Algorithms are pipelined at 40 MHz for deadtime-free operation è Total decision latency: 3.2  s electrons, photons, jets, MET muons 3<|  |<5 |  |<3 |  |<3 |  |<2.1 0.9<|  |<2.4 |  |<1.2

6. CMS101 - Trigger, 13 Sep 2004Darin Acosta6 Level-1 Trigger Hardware (U.S.) BSCAN ASICs PHASE ASICs MLUs SORT ASICs EISO Sort ASICs BSCAN ASICs Phase ASIC RCT Receiver cardRCT Jet/Summary card RCT Electron isolation card Custom ASICs Large FPGAs SRAM Gbit/s Optical links Dense boards Optical links SRAM FPGA CSC Track-Finder

7. CMS101 - Trigger, 13 Sep 2004Darin Acosta7 Level-1 Calorimeter Trigger Scheme

8. CMS101 - Trigger, 13 Sep 2004Darin Acosta8 Calorimeter Trigger Geometry EB, EE, HB, HE map to 18 RCT crates Provide e/  and jet,  E T triggers Trigger towers:  =  = 0.087

9. CMS101 - Trigger, 13 Sep 2004Darin Acosta9 Level-1 Calorimeter Trigger Algorithms Electron (Hit Tower + Max) è 2-tower  E T + Hit tower H/E è Hit tower 2x5-crystal strips >90% of E T in 5x5 (Fine Grain) Isolated Electron (3x3 Tower) è Quiet neighbors: all towers pass Fine Grain & H/E è One group of 5 EM E T < Thr. Jet or  E T è 12x12 trig. tower  E T sliding in 4x4 steps w/central 4x4 E T > others  : isolated narrow energy deposits  Energy spread outside  veto pattern sets veto  Jet   if all 9 4x4 region  vetoes off

10. CMS101 - Trigger, 13 Sep 2004Darin Acosta10 Level-1 Global Calorimeter Trigger Implements sliding window jet cluster algorithm Sorts electron and jet objects Computes energy and missing E T sums Output to Global Trigger: è 4 non-isolated e/  è 4 isolated e/  è 4 central jets è 4 forward jets è 4  objects è Total E T è MET è MET  angle

11. CMS101 - Trigger, 13 Sep 2004Darin Acosta11 Electron/photon Level-1 Efficiencies Response to electrons

12. CMS101 - Trigger, 13 Sep 2004Darin Acosta12 Single Electron/photon Level-1 Rates Rate from jet background Low lumi High lumi

13. CMS101 - Trigger, 13 Sep 2004Darin Acosta13 Level-1 Jet Trigger Rate Low lumi High lumi

14. CMS101 - Trigger, 13 Sep 2004Darin Acosta14 Level-1 Muon Trigger Scheme

15. CMS101 - Trigger, 13 Sep 2004Darin Acosta15 Muon Trigger Geometry

16. CMS101 - Trigger, 13 Sep 2004Darin Acosta16 DT and CSC Local Triggers Bunch & Track Identifier (BTI) uses shift registers to search for patterns in drift tubes (  and  ) and to assign correct BX Cathode strips Local Charged Track (LCT) logic identifies track stubs in CSCs (in  and  ) Both are multi-layer detectors

17. CMS101 - Trigger, 13 Sep 2004Darin Acosta17 CSC and DT sectors align for overlap region Track-Finding   è DT and CSC Track-Finders link local track segments into distinct tracks p 2-D tracks for DT, 3-D tracks for CSC è RPC Pattern Comparator Trigger applies coincidence logic along roads in  and  with  ×  = 0.1 × 0.005 è Standalone momentum measurement using B-field in yoke p Require < 25% P T resolution for sufficient rate reduction è Highest quality candidates sent to Global Muon Trigger

18. CMS101 - Trigger, 13 Sep 2004Darin Acosta18 Global Muon Trigger Combines/matches muons from all 3 systems: è Maximize efficiency è Minimize rate è Cancel duplicates è Apply calorimeter isolation or MIP è Programmable P T thresholds from 1 to 140 GeV/c

19. CMS101 - Trigger, 13 Sep 2004Darin Acosta19 Level-1 Muon Efficiency Di-muon trigger coverage extends to |  |<2.4  PTPT |  |<0.8 0.8<|  |<1.2 1.2<|  |<2.1

20. CMS101 - Trigger, 13 Sep 2004Darin Acosta20 Single Muon Level-1 Rate Rate comes from real muons! Limited P T resolution flattens rate curve for high thresholds Low lumi High lumi

21. CMS101 - Trigger, 13 Sep 2004Darin Acosta21 Level-1 Global Trigger A maximum of 128 trigger lines can be implemented è e.g. 1e, 2 , 1  +1e, etc. Topological cuts can be defined è e.g. 2 jets not back-to-back in  Only place where thresholds are applied Level-1 decision is transmitted to the Trigger Throttle System, which in turn transmits a Level-1 Accept via the Trigger Timing and Control system to the detector front-end read-out electronics

22. CMS101 - Trigger, 13 Sep 2004Darin Acosta22 Example Level-1 Trigger Table (L=2×10 33 )  3 safety factor  50 kHz (expected start-up DAQ bandwidth) Only muon trigger has low enough threshold for B-physics

23. The High-Level Triggers CMS does not have a custom Level-2 trigger in hardware Everything beyond Level-1 performed in the Filter Farm HLT does partial event reconstruction “on demand” using full detector resolution Historically, Level-2 uses only calorimeter + muon and Level-3 uses tracker (90% of data volume) Documented in DAQ & HLT TDR CERN/LHCC 2002-26

24. CMS101 - Trigger, 13 Sep 2004Darin Acosta24 The CMS High-Level Triggers Reduces rate from 100 kHz to O(100 Hz) è Final rate will depend on data bandwidth, storage capability, and background rejection capability è Deployed as software filters running in an online computer farm (~1000 PCs) p Software is in principle the same as used offline Starts with a data sample already enriched in physics! è Level-1 already applied a factor 400 background rejection What can be done: è Electrons: require high-P T track match to veto  0 fakes, recover bremsstrahlung è Photons: veto tracks è Muons: require high-P T track match to improve momentum resolution è Jets: run standard jet algorithms è Tracks: improve measurement of impact parameter, p T and charge è Apply isolation criteria to all leptons è Apply topology and invariant mass cuts

25. CMS101 - Trigger, 13 Sep 2004Darin Acosta25 HLT selection: electrons and photons Level-2 Level-3 Level-1 Level-2.5 Photons Threshold cut Isolation Electrons Track reconstruction E/p, matching (  ) cut ECAL reconstruction Threshold cut Pixel matching è Issue is electron reconstruction and rejection è Higher E T threshold on photons è Electron reconstruction p key is recovery of radiated energy è Electron rejection p key tool is pixel detector

26. CMS101 - Trigger, 13 Sep 2004Darin Acosta26 Electron selection: Level-2 “Level-2” electron: è Search for match to Level-1 trigger p Use 1-tower margin around 4x4-tower trigger region è Bremsstrahlung recovery “super-clustering” è Select highest E T cluster Bremsstrahlung recovery:  Road along  — in narrow  -window around seed è Collect all sub-clusters in road  “super-cluster” basic cluster super-cluster

27. CMS101 - Trigger, 13 Sep 2004Darin Acosta27 Electron selection: Level-2.5 “Level-2.5” selection: use pixel information  Very fast, large rejection with high efficiency (>15 for  =95%) p Before most material  before most bremsstrahlung, and before most conversions p Number of potential hits is 3: demanding  2 hits quite efficient Full pixel system Staged option

28. CMS101 - Trigger, 13 Sep 2004Darin Acosta28 Electron selection: Level-3 “Level-3” selection è Full tracking, loose track- finding (to maintain high efficiency) è Cut on E/p everywhere, plus  Matching in  (barrel) p h/e (endcap) è Isolation (used for photons) 2x10 33 cm -2 s -1

29. CMS101 - Trigger, 13 Sep 2004Darin Acosta29 Muon HLT Selection Standalone Muon Reconstruction: “Level-2” è Seeded by Level-1 muons è Local reconstruction exploiting full detector resolution (100  m) è Kalman filtering technique applied to DT/CSC/RPC track segments è “GEANE”-like algorithm used for propagation through iron è Trajectory building works from inside out è Track fitting works from outside in è Fit track with beam constraint è Isolation based on  E T from calorimeter towers in cone around  Inclusion of Tracker Hits: “Level-3” è Define a region of interest through tracker based on Level-2 track with parameters at vertex è Find pixel seeds, and propagate from innermost layers out, including muon è Isolation based on  P T from pixel/tracker tracks in cone around 

30. CMS101 - Trigger, 13 Sep 2004Darin Acosta30 Muon HLT Efficiency vs. P T Threshold Level-1, Level-2, Level-3 Single muons with |  |<2.1 Thresholds sharpen with improved P T resolution (15% without tracker, 1.5% with tracker)

31. CMS101 - Trigger, 13 Sep 2004Darin Acosta31 L1, L2, L3 Trigger Rates @ High Lumi è Better resolution offers superior rate reduction è Isolation only yields minor gains p Low P T muons and W decays already isolated

32. CMS101 - Trigger, 13 Sep 2004Darin Acosta32 Muon HLT Results 30 Hz output rate Efficiencies L=2  10 33 s -1 cm -2 H  ZZ *     98% for M=150 GeV H  WW *     92% for M=160 GeV p T >20

33. CMS101 - Trigger, 13 Sep 2004Darin Acosta33 Jet Selection: Level-2 Cone algorithm, R = 0.5 Single-jet E T thresholds expected to be very high

34. CMS101 - Trigger, 13 Sep 2004Darin Acosta34 Jets + MET Triggers è Reconstructed MET rate below 100 GeV mainly from calorimeter coverage and energy resolution

35. CMS101 - Trigger, 13 Sep 2004Darin Acosta35 Inclusive SUSY Trigger Exercise è Consider several points in the m 0  m 1/2 plane near the Tevatron reach (most difficult for LHC) Possible triggers at Level-2: è 1 jet E T >180 GeV & MET>120 è 4 jets E T >110 GeV Overall efficiency to pass both Level-1 and Level-2: è  =0.63, 0.63, 0.37 è Background rate of ~12Hz @ L = 2  10 33 dominated by QCD jets è Trigger becomes more efficient at high luminosity since one expects to explore higher masses è More exclusive triggers can further improve efficiency 456

36. CMS101 - Trigger, 13 Sep 2004Darin Acosta36 Example HLT Trigger Menu (L=2×10 33 ) è 1e: P T > 30 GeV è 2e: P T > 15 GeV è 1  : P T > 80 GeV è 2  : P T > 40, 25 GeV è 1  : P T > 20 GeV è 2  : P T > 7 GeV è 1  : P T > 85 GeV è 2  : P T > 60 GeV è 1 jet: E T > 660 GeV è 1 jet + MET: E T > 180, 120 GeV è 3 jets: E T > 250 GeV è 4 jets: E T > 110 GeV Estimated rate to tape ~100 Hz

37. CMS101 - Trigger, 13 Sep 2004Darin Acosta37 Summary Trigger system applies an overall factor of 10 6 filtering while maintaining good efficiency Level-1: è First factor of 1000 è Hadronic tau trigger implemented è Sliding window jet triggers è Isolated and non-isolated lepton triggers (without central tracking) è 128 trigger lines available HLT: è Second factor of 1000 è Access to full event information è Partial reconstruction based on the calorimeter and muon systems initially (verify and improve Level-1 decision), followed by pixel + tracker information for final rejection è Lots of flexibility Your analysis may require yet another factor 10 6 rejection!