CMS Trigger Simulation

Transcription

1 CMS Trigger Simulation Sridhara Dasu University of Wisconsin Outline: Calorimeter Level- Trigger Algorithms Rates and Efficiencies Physics Performance Muon Level- Trigger Algorithms Rates and Efficiencies Global Trigger Physics Performance DoE/NSF Review, May, 998

2 CMS/LHC Trigger Physics Standard model Higgs (high luminosity) H (8 GeV) γ γ H (2 GeV) Z Z* (4 leptons) H (>5 GeV) leptons ( + ν's) H (< 2M W Associated t or W or Z) b b (lepton + X) SUSY Higgs (low luminosity) (standard model Higgs like channels) h, H, A τ τ (lepton + X) or µµ A Z h ; h bb (lepton + X) p p t t X; t H + b; H + τ ν; t lepton + X; τ X SUSY sparticle searches (low luminosity) MSSM sparticle LSP (Missing E t ) + n jets MSSM sparticle Same sign dileptons + X Other new particles Z' dileptons Leptoquarks: dileptons Top physics (low luminosity) t lepton + X t multijets Bottom physics (low luminosity) b lepton + X b ψk s (leptons + X) QCD Low luminosity GeV jets High luminosity 2 GeV jets fi Trigger candidate requirements: High luminosity: lepton/γ (3 GeV), dileptons/γγ (5 GeV) missing E t ( GeV), jets (2 GeV) Low luminosity: lepton/γ (5 GeV), dileptons/γγ ( GeV) missing E t (5 GeV), jets ( GeV) DoE/NSF Review, May, 998 2

3 DoE/NSF Review, May, 998 Calorimeter requirements Input ECAL trigger towers,.87φ x.87η Matching HCAL towers Data every 25ns - including any corrections for time development of calorimeter signal 8 bit transverse energy bit finegrain charecterization of energy deposit Data presynchronized across all channels, ECAL and HCAL Output Top 4 nonisolated electrons/photons (E t and location) Top 4 isolated electrons/photons (E t and location) Top 4 jets (E t and location) Top 4 isolated hadrons (E t and location) Total and missing transverse energy (E t, E x, E y ) Minimimum ionization ID and isolation bits for use with muon trigger Outut rate 5 khz maximum for calorimeter trigger Simulations should indicate about a factor of 3 safety margin - i.e., ~5 khz Efficiency Trigger should contribute no more than a few percent inefficiency for any physics channel compared to other offline analysis cuts. Trigger efficiencies should be measurable 3

4 Cal Regional Trigger Input ECAL Crystals in Trigger Tower.75 η η.75 φ φ.87 φ Worsening factor Transverse Energy in Trigger Tower A factor to account for worsening of limited resolution trigger energy compared to full energy resolution n c = 7 n c = 6 5%.5 n c = 8.5% E s T (GeV).87 η 8 bit transverse energy output from both ECAL and HCAL trigger towers E t Shower Profile in Trigger Tower Counts 4 Counts 3 CMSIM FASTSIM 3 CMSIM FASTSIM Fine-grain parameter Fine-grain parameter Single bit to indicate electron like shower profile within the trigger tower - Comparison of the maximum strip pair energy sum to the total energy sum in the trigger tower. DoE/NSF Review, May, 998 4

5 φ η Electron/photon algorithm.75 η.75 φ Sliding window centered on all ECAL/HCAL trigger tower pairs Tower count = 72φ x 54η x 2 = 7776 Had Shower Profile Cuts: Fine-grain feature Compare max E t η-strip pair out of 4 pairs versus total E t in trigger tower, e.g., require 9% energy in a pair. HAC Veto Compare HCAL versus ECAL E t in Memory Lookup to veto non-em deposits, e.g., H/E<5% when E is significant. DoE/NSF Review, May, 998 Hit Max EM.87 η Isolation Cuts:.87 φ Neighbor HAC Veto HAC Veto passes on all eight neighbors also. New E t isolation: Quiet Neighborhood At least one of four corners has all five quiet towers, i.e, (E t <.5 GeV) Candidate Energy: Max Hit Summary: Max E t of 4 Neighbors Hit + Max E t > Threshold Regional Pick highest energy candidate in 4x4 trigger tower region. Global Sort to find top-4 isolated and non-isolated candidates separately. 5

6 Jet, Missing E t algorithms Trigger Tower 4x4 Jet Region 4x4 centers used for E x, E y computation HCAL ECAL η, φ =.348 Jet E t is given by the sum of ECAL and HCAL trigger tower E t in a non-overlapping 4x4 region Jet candidates are sorted to find highest energy jets Jet trigger is caused by core of the physical jet. This allows for jet counting without the problems of dealing with multiple jets overlapping in large (.ηx.φ) regions E x and E y are obtained by a memory lookup using 4x4 E t Signed E x and E y sums over the entire calorimeter are made to calculate missing E t DoE/NSF Review, May, 998 6

7 Trigger/Calorimeter Map 3. η φ π π 8 7 Regional Trigger Crate φ.35.7 η 2.5 Electron Identification card (two 4x4 regions) Hit Center Max Edge Corner DoE/NSF Review, May, 998 7

8 DoE/NSF Review, May, 998 Simulation programs FSTSIM - Fast simulation of event response Simplified CMS geometry, uniform tracking medium, decays of mesons, and parameterized calorimeter showers are implemented. CMSIM - Version CMS standard GEANT based detector simulation Detailed geometry for calorimeter, average response for tracker but no preshower. PYTHIA - common for FSTSIM/CMSIM QCD background events are used for rate studies. High P t signal events, e.g., top, Higgs and SUSY particle decays, are used for efficiency studies. Noise hits are superposed with high P t events. Minimum bias included - FSTSIM minbias is added for both CMSIM and FSTSIM high P t events. Trigger simulation - common for FSTSIM/CMSIM Various digital scales with limited resolution and dynamic range involved in the trigger system are fully implemented. Algorithms are performed in integer arithmetic using memory lookup tables when needed. 8

9 Electron/photon Efficiency - CMSIM vs FSTSIM Efficiency for triggering top to electron decay events is plotted versus the P t of the electron for various cuts. Identical values for the four cut parameters yield similar efficiencies - custom tuning was not necessary. All efficiencies are over 9%. DoE/NSF Review, May, 998 9

10 EM Rates - CMSIM vs FSTSIM Integrated rate above E t cut is plotted versus E t cut. All four, i.e., finegrain, HAC veto, neighbor HAC veto and quiet neighborhood, cuts are included. For 25 GeV E t cut, CMSIM rate is 9 khz versus to 4 khz in FASTSIM. DoE/NSF Review, May, 998

11 EM Rate Breakdown Rates after finegrain and quiet neighborhood cuts match but those after HAC veto and neighbor HAC veto cuts do not. DoE/NSF Review, May, 998

12 CMSIM vs FASTSIM Differences Single pion events Counts 3 2 Counts Energy within 5x5 2 3 Total energy in calorimeter Counts 3 2 Counts 3 2 CMSIM FASTSIM.5 Energy outside 5x5 2 3 H/E ratio within 5x5 Energy loss and transverse spread are higher in CMSIM. Therefore, longitudinal profile, i.e., H/E ratio is also different. This leads to the difference in rates. Should examine this further. DoE/NSF Review, May, 998 2

13 Rate (khz) 2 Jet rates - CMSIM vs FSTSIM Jet trigger rates UW-Madison CMS PbWO 4 Calorimeter CMSIM Vs Fast Simulation L = 34 cm -2 s - Single Jet - FASTSIM Single Jet - CMSIM Double Jet - FASTSIM Double Jet - CMSIM Triple Jet - FASTSIM Triple Jet - CMSIM Quad Jet - FASTSIM Quad Jet - CMSIM Trigger E t Cutoff (GeV) Integrated rates above trigger E t cutoff are plotted versus the E t cutoff for single, double, triple and quad jet events. Rates are in reasonable agreement More statistics are needed to get the correct spectrum at high E t. DoE/NSF Review, May, 998 3

14 Jet efficiency - CMSIM vs FASTSIM Efficiency Efficiency.8 QCD jet efficiency - 4x4 algorithm UW-Madison CMS PbWO 4 Calorimeter Level- trigger simulation L = 34 cm -2 s - Single jet E t > GeV Double jet E t > 6 GeV Triple jet E t > 3 GeV Quadruple jet E t > 2 GeV Reconstructed Jet P t (GeV) QCD jet efficiency - 4x4 algorithm Efficiency for triggering on single, double, triple and quad jet events is plotted versus the reconstructed jet Pt of the lowest energy jet. N-jet efficiency is cumulative of all jet cuts -N. CMSIM efficiency turn-on is somewhat slower than FSTSIM UW-Madison CMS PbWO 4 Calorimeter Level- trigger simulation L = 34 cm -2 s - Single jet E t > GeV Double jet E t > 6 GeV Triple jet E t > 3 GeV Quadruple jet E t > 2 GeV Reconstructed Jet P t (GeV) This can be explained by the lower energy deposition for the hadrons in the events. DoE/NSF Review, May, 998 4

15 Missing and total E t Rates Integrated rates above missing and total E t trigger cutoffs are plotted versus the missing and total E t cutoff respectively. The agreement between CMSIM and FSTSIM is quite good. Efficiency studies are underway. DoE/NSF Review, May, 998 5

16 Missing E t Efficiency Missing E T Trigger at L = 34 cm -2 s - Efficiency.8.6 UW-Madison CMS PbWO 4 Calorimeter Level- trigger simulation L = 34 cm -2 s - CMSIM FASTSIM Missing E T Efficiency is plotted versus the ISAJET hadron level missing P T for SUSY sparticle production..4.2 Trigger Cutoff Hadron Level Missing P t (GeV) Both CMSIM and FASTSIM efficiency turn-on is somewhat slower than desirable. Efficiency.8.6 Missing E T Trigger at L = 33 cm -2 s - UW-Madison CMS PbWO 4 Calorimeter Level- trigger simulation L = 33 cm -2 s - CMSIM FASTSIM Trigger factors are responsible for only half the worsening of resolution compared to the ISAJET hadron level values..4.2 Trigger Cutoff Hadron Level Missing P t (GeV) However, SUSY events efficiency is supplemented by the total and multijet triggers. DoE/NSF Review, May, 998 6

17 Cal Physics at high lumi Trigger Type Trigger Et Cutoff (GeV) DoE/NSF Review, May, 998 CMSIM Rate (khz) FASTSIM Individual Incremental Individual Incremental Sum Et Missing Et Electron DiElectron Single jet Dijet Trijet Quadjet Jet+Elctrn 5 & Cumulative Rate Process Efficiency (%) CMS-TN-95/83 FASTSIM CMSIM H (8 GeV) γ γ H (2 GeV) Z Z e e µ µ 76* 76* 74* H (2 GeV) Z Z e e j j p p t t e X p p t t H+ X t X * Inclusion of muon trigger provides full efficiency Luminosity = 34 cm -2 s - QCD Background The sum and missing E t cutoffs are chosen to provide a 2 khz rate. Electron/photon triggers are emphasized, with about 8 khz rate out of total available 5 khz. The remaining 5 khz is available for jet triggers. Signal Efficiency High efficiency for all channels with electrons and photons. The difficult-to-trigger top decay events have high efficiency, enabling studies of associated Higgs production. Unfortunately, no eta cut on trigger particle in new FAST and CMSIM studies! 7

18 Cal Physics at low lumi Trigger Type Trigger Et Cutoff (GeV) CMSIM Rate (khz) FASTSIM Individual Incremental Individual Incremental Sum Et Missing Et Electron DiElectron Single jet Dijet Trijet Quad jet Jet+Elctrn 5 & Luminosity = 33 cm -2 s - QCD Background CMSIM and FASTSIM rates are compared for the low luminosity E t cutoffs selected in CMS-TN-95/83 Electron trigger rate is twice as high in CMSIM results Cumulative Rate Process Efficiency (%) CMS-TN-95/83 FASTSIM CMSIM p p t t e X p p t t H+ X t X p p b b (hadronize), B e X.2 (But 4Hz) - - SUSY CMS TP Scenario A (MLSP = 45, Mspart ~ 3 GeV) SUSY Neutral Higgs (Range of tan b and MH values) Signal efficiency High efficiency is realized for the benchmark processes involving top decays and SUSY sparticles. A dedicated tau trigger is under study to improve efficiency for low mass range of SUSY Higgs. There is also high rate of B signal in level- sample. DoE/NSF Review, May,

19 Muon Trigger Requirements MUON IDENTIFICATION At least 6 λ of material is present up to η =2.4 with no acceptance losses. STANDALONE MOMENTUM RESOLUTION From 8-5% dp T /p T at GeV and 2-4% at TeV. GLOBAL MOMENTUM RESOLUTION After matching with the Central Tracker: from.-.5% at GeV, and from 6-7% at TeV. Momentum-dependent spatial position matching at TeV less than mm in the bending plane and less than mm in the non-bending plane. CHARGE ASSIGNMENT Correct to 99% conf. up to the kinematic limit of 7 TeV. MUON TRIGGER The combination of precise muon chambers and fast dedicated trigger detectors provide unambiguous beam crossing identification and trigger on single and multimuon events with well defined p T thresholds from a few GeV to GeV up to η =2.. DoE/NSF Review, May, 998 9

20 Muon Trigger Number of events Solenoidal fields of CMS bends tracks in Rϕ plane. Thus tracks keep almost constant η. ϕ B dl η Even for the softest forward tracks η< < η <2.3 p t = GeV 5 5 p t = 2.5 GeV η 4 -η For η <2.5 the B dl is enough to provide a p t cut. Strips of ϕ=/3 are narrow enough. Number of events < η <2.3 p t = 2.5, 5,, 5 GeV φ 2 -φ [/3 deg strips] DoE/NSF Review, May, 998 2

21 Muon Chamber Trigger Logic Drift Tubes 4 2 µ 3 CSC wires Track Finder 2 x extrapolation threshold muon station 4 track segment muon station 3 3 Meantimers recognize tracks and form vector / quartet. 4 2 threshold Q + Q2 + Q3 + strips + muon station 2 muon station φ2 - φ Comparators give /2 strip resol. 3 track segment pairs are combined to one track string Correlator combines them into one vector / station. Hit strips of 6 layers form a vector. combines vectors, forms a track, assigns p t value.

22 Cathode Strip Chambers wires strips TIME: OR of 6 planes of wires first pulse gives t but 4 planes out of 6 must be fired t max drift = 4 ns σ(t ) < 3 ns POSITION: cluster center in every plane of strips x = /2 strip width ANGLE: local charged track (LCT) determined by pattern of fired halfstrips in a gi ven station MOMENTUM: combination of LCT s from all stations DoE/NSF Review, May,

23 Single Muon Rates Rate [Hz] η <2.4 µ from anything µ from all minimum bias events µ from W µ from DRELL-YAN µ from Z µ from top µ from J/ψ L= 34 cm -2 s - 2 DoE/NSF Review, May,

24 Muon Trigger Efficiency efficiency efficiency efficiency RPC η=.9.8 = p t (GeV) RPC η=.9.8 = p t (GeV) RPC η= p t (GeV) efficiency efficiency efficiency DT η=.9.8 = p t (GeV) DT/CSC η= p t (GeV) CSC η= p t (GeV) DoE/NSF Review, May,

25 Cal/Muon Combined Rates trigger type threshold [GeV] L = 33 cm -2 s - L = 34 cm -2 s - rate [khz] cumulative rate [khz] threshold [GeV] rate [khz] cumulative rate [khz] µ µ µ µ e/γ 2-4, , µ e b 2-4, µ j 2-4, , miss µ E t 2-4, , µσe t 2-4, , threshold = 2-4 GeV means: 4 GeV in the barrel, 2 GeV in the endcaps muon threshold = transverse momentum threshold calorimeter threshold = transverse energy threshold e/γ electron/photon trigger, e b trigger on electron from b-quark decay DoE/NSF Review, May,