VI symposium on LHC Physics and Detectors Fermilab, Batavia, Il, U.S., may, 2003 CMS: Muon System and Physics performance Stefano Lacaprara, I.N.F.N. and Padova University on behalf of the CMS collaboration S. Lacaprara: CMS: Muon System and Physics performance 0
Outline Introduction, Muon detectors, Muon reconstruction, High Level Trigger, Local reconstruction, Global reconstruction, Muon isolation, Physics performance, Summary S. Lacaprara: CMS: Muon System and Physics performance 1
Introduction Results presented for two LHC instantaneous luminosities: Low luminosity (LL) first year of data taking: High luminosity (HL) after: All results presented with full CMS simulation (GEANT3) Pile up included for both luminosities, OO-reconstruction ORCA (Object-oriented Reconstruction for CMS Analysis) S. Lacaprara: CMS: Muon System and Physics performance 2
Muon Detectors in CMS S. Lacaprara: CMS: Muon System and Physics performance 3
Muon Detectors (2) Three type of gaseous detectors: Drift Tubes in barrel region, Cathod Strips Chambers in endcap regions, DT s and CSC s provide precise position measurements, ( Resistive Plate Chambers in both barrel and endcaps. RPC provide precise bunch crossing (bx) id. all the sub-sistem contribute to the L1-trigger. ), S. Lacaprara: CMS: Muon System and Physics performance 4
#" $ # Muon Detectors (3) R (cm) 800 700 600 500 400 300 200 100 0 DT eta=0.8 RPC 1.04 1.2 CSC 0 200 400 600 800 1000 1200 Z (cm) $ Barrel Endcaps or station of RPC s up to L1 trigger up to! Start-up staged detector: no ME4 and RPC 2.1 2.4! stations of DT s, interleaved with the iron of magnet yoke, self triggering and bx identification, stations of CSC s with same capabilities, interleaved with iron disk yoke, up to! #", #", S. Lacaprara: CMS: Muon System and Physics performance 5
Drift Tube chambers %Made of "in 10 5 7 7 < 7 5 " - / ' % # " + the &(' station) separated by honeycomb spacer SuperLayers, ) and in & ' *(not in %Each SL made layers of of staggered -,wide cells,.-field shaped to give maximum linearity, %Gas mixture: & ' 2435 %Single point resolution: %Chamber resolution: pos %Total -, dir 8 8 9 7 8chambers, 8=wires. " 8 8 9 -, ;:&, 6, S. Lacaprara: CMS: Muon System and Physics performance 6
@ 7 " Cathode Strip Chambers %Arranged disks, in %Inner ring have 8?>, outer $covering 3CSCs each covering 8 > %Made $Layers, of trapezoidal shape, %Radial strips measure precise bending coordinate by interpolation of induced charge, -, 8 8 A " 9 8 7 5 " 8 C 10 0B 5 / @ 7 %Orthogonal anode wires readout -,, ;points, %Gas mixture: & ' 2 ' 8' 8 ' 6, %Total 8chambers, 5 8#channels. S. Lacaprara: CMS: Muon System and Physics performance 7
$ 7 $ 9 % 7 @ 7 Resistive Plate Chambers %Double gap, single readout (strips), %Dedicated trigger detectors, %Fast timing response, precise identification of bx %Coarse ED, -,position resolution, assignment via Pattern Comparator Trigger, %Help in identification of tracks, %Total "chambers, 8=channels. S. Lacaprara: CMS: Muon System and Physics performance 8
Muon Reconstruction S. Lacaprara: CMS: Muon System and Physics performance 9
" 8 8 " Muon Reconstruction In CMS the High Level Trigger is performed on a commercial processor farm (Filter Unit) running software algorithm on raw data, Software used will be as much as possible the same used in offline reconstruction Offline reconstruction will make full use of complete calibration, alignment, etc... Robust, high quality reconstruction software %Use of a common framework %Object-oriented Reconstruction for CMS Analisys: ORCA Basic principle: regional reconstruction %Reconstruction performed in region only, %Need seed to start: for HLT is Level-1 trigger object Working prototype of full HLT reconstruction and selection already developed: DAQ & HLT TDR, CERN/LHCC 2002/26, 5december S. Lacaprara: CMS: Muon System and Physics performance 10
9 P*SL (where present): initial R 2 FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG P H FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG N N Local Pattern Recognition DT s and CSC s are multi-layers detectors: first step of muon reconstruction is local, i.e. inside chambers Barrel R Z SL R φ SL honeycomb spacer R φ SL % LKJ I M depends on %Reconstruct ) SL hits (time-space conversion), hit error first approx.: %Build segments Hresolution, Ofield and impact angle O@ center of wire, Q" linear fit, %Solve L/R ambiguity by best 8, criterion, %Apply impact angle correction to hits and refit, P, %Build segment in %Associate two projections (if present) to build a Qsegment, & ' assuming from I.P., S. Lacaprara: CMS: Muon System and Physics performance 11 & '
8 8 @ @ - N N N Local Pattern Recognition: Barrel (2) %Apply further correction for wire ( Ovaries!) and do final refit. O, using knowledge of hit position along the %Position and direction of segment and corresponding error matrices from linear fit, %Hits error (used in the fit) depends on %Resolution: SUT 7 Oand 9 -for bending coordinate, P, apply proper corrections, L JK 7 ;:&. 2 χ / ndf 600.9 / 7 2 χ / ndf 1262 / 7 Constant 5232 Constant 5750 5000 4000 Mean 0.0003419 Sigma 0.01182 6000 5000 Mean 2.463e-05 Sigma 0.001075 4000 3000 3000 2000 2000 1000 1000 0-0.15-0.1-0.05 0 0.05 0.1 0.15 x rec -x sim (cm) 0-0.015-0.01-0.005 0 0.005 0.01 0.015 rec sim dx/dz -dx/dz S. Lacaprara: CMS: Muon System and Physics performance 12
@ Local Pattern Recognition (2) Endcap %Reconstruct Qhits, %Fit ( Gatti function) charge distribution on nearby strips to get cluster centroid, %Use discriminated wire-group signal to get nonbending coordinate, %Associate two projection by time coincidence, %Fit Qsegment using %Resolution: SUT 7 Qhits, linear fit, -for bending 8 8 A 5 " 9 8 coordinate, depending on chamber, S. Lacaprara: CMS: Muon System and Physics performance 13
Local Pattern Recognition: Endcap(2) Residuals ME1/1 2 χ / ndf 1173 / 97 Residuals ME2/1,ME3/1,ME4/1 2 χ / ndf 324.2 / 97 1600 1400 1200 1000 Constant 1302 Mean 0.0009831 Sigma 0.01096 1600 1400 1200 1000 Constant 1494 Mean 0.0002587 Sigma 0.02013 800 800 600 600 400 400 200 200 0-0.04-0.03-0.02-0.01 0 0.01 0.02 0.03 0.04 rec sim x -x (cm) φ φ 0-0.04-0.03-0.02-0.01 0 0.01 0.02 0.03 0.04 rec sim x -x (cm) φ φ 500 Residuals ME1/2,3 2 χ / ndf 175.6 / 97 Constant 489.5 Mean 0.0001817 Sigma 0.01843 Residuals ME2/2,ME3/2,ME4/2 1200 2 χ / ndf 210.7 / 97 Constant 1149 Mean 0.0001227 Sigma 0.02356 400 1000 300 800 600 200 400 100 200 0-0.04-0.03-0.02-0.01 0 0.01 0.02 0.03 0.04 rec sim x -x (cm) φ φ 0-0.04-0.03-0.02-0.01 0 0.01 0.02 0.03 0.04 rec sim x -x (cm) φ φ S. Lacaprara: CMS: Muon System and Physics performance 14
Standalone Muon Reconstruction (1) R (cm) 800 700 gen. muon 600 500 400 300 200 100 0 0 200 400 600 800 1000 1200 Z (cm) S. Lacaprara: CMS: Muon System and Physics performance 15
Standalone Muon Reconstruction (2) gen. muon L1 Based only on Muon detectors: DT, CSC and RPC, Seed generation: L1-output for L2 trigger or From track segments from local reconstruction, S. Lacaprara: CMS: Muon System and Physics performance 16
Standalone Muon Reconstruction (2) Search of detectors compatible with seed, S. Lacaprara: CMS: Muon System and Physics performance 16
Standalone Muon Reconstruction (2) Search of detectors compatible with seed, Local pattern reconstruction performed only in those detectors: regional reconstruction on demand, Use DT 3D segments, CSC 3D hits constituents of segments, and RPC hits, S. Lacaprara: CMS: Muon System and Physics performance 16
R 2 Standalone Muon Reconstruction (2) rec. muon Trajectory building inside-out to collect hits, Kalman filter technique to grow trajectory, State propagation across iron with non-const CPU time consuming! Track fitting works outside-in, Extrapolation to nominal I.P. and refit cuts to reject bad hits, Ofield: S. Lacaprara: CMS: Muon System and Physics performance 16
6 8 6 6 5 7 7 bc bc f g g h d ] `[ ] ] \[ $ 7 bc jgi d j `[ Standalone Muon Reconstruction (6) Resolution at L2: VXWZY VXWa^ VXW_^ a) Barrel ( bedf h) b) Overlap ( b d jgi) c) Endcap ( b d ig) reso reso reso 1200 1000 (a) 400 350 (b) 800 700 (c) 800 300 250 600 500 600 200 400 400 150 300 200 100 50 200 100 0-1 -0.5 0 0.5 1 0-1 -0.5 0 0.5 1 0-1 -0.5 0 0.5 1 S. Lacaprara: CMS: Muon System and Physics performance 17
Global Muon Reconstruction (1) Inclusion of inner Tracker hits: Use L2 reconstructed muons as seed for tracker reconstruction, Get muon state at outermost tracker surface and at interaction point, S. Lacaprara: CMS: Muon System and Physics performance 18
Global Muon Reconstruction (1) Inclusion of inner Tracker hits: Use L2 reconstructed muons as seed for tracker reconstruction, Get muon state at outermost tracker surface and at interaction point, Define region of interest in the tracker, S. Lacaprara: CMS: Muon System and Physics performance 18
Global Muon Reconstruction (1) Inclusion of inner Tracker hits: Use L2 reconstructed muons as seed for tracker reconstruction, Get muon state at outermost tracker surface and at interaction point, Define region of interest in the tracker, Create one or more seeds from pairs of reconstructed hits in different tracker layers, S. Lacaprara: CMS: Muon System and Physics performance 18
R 2 Global Muon Reconstruction (1) For a given seed build trajectory: Building inside-out, Kalman filter applied, Uniform O, few material: propagation is much simpler and faster than in the muon system, Trajectory cleaning and final fit: Solve ambiguities, Ghosts suppression (# hits and ), Use tracker and muon hits for final fit, Reject bad tracks. S. Lacaprara: CMS: Muon System and Physics performance 18
# 8 6 7 # 6 7 # l 6 7 bc bc bc f g g h d g i g i k d j ] `[ ] ] \[ `[ Global Muon Reconstruction (2) Resolution with Tracker: V WZY VXWa^ VXW_^ a) Barrel ( bedf h) b) Overlap ( b d k) c) Endcap ( b d ig) reso reso reso 1000 (a) 400 350 (b) 800 700 (c) 800 300 250 600 500 600 200 400 400 150 300 200 100 50 200 100 0-0.1-0.05 0 0.05 0.1 0-0.1-0.05 0 0.05 0.1 0-0.1-0.05 0 0.05 0.1 S. Lacaprara: CMS: Muon System and Physics performance 19
mn m Global Muon Reconstruction (3) L1, L2, L3 efficiency vs Eta coverage limit 2.1 ME 1/1a does not participate to L1 trigger Zero suppressed! S. Lacaprara: CMS: Muon System and Physics performance 20
q m Global Muon Reconstruction (4) L1, L2, L3 efficiency vs mo p Suppressed zero! S. Lacaprara: CMS: Muon System and Physics performance 21
. 9. u v t Muon Isolation (1) Based on or in cones around reconstructed muons, Cones sizes and thresholds are optimized to get maximum rejection for minimum bias events above trigger on reference signal ( Flat by construction, Calorimeter isolation, @ L2 9sr) threshold for a fixed efficiency from ECAL & HCAL towers: sensitive to PileUp, Pixel isolation, @ L2.5 Need from 'hit tracks in the pixel detector, layers, no with pixel staging, problem if pixel inefficiencies, Take tracks from the same vertex as Tracker isolation, @ L3, reduce PU, from tracker tracks, regional reconstruction around muon, S. Lacaprara: CMS: Muon System and Physics performance 22
q } } ~ w w w w { Muon Isolation (2) yxvs o z for yxvs Efficiency MB 1 0.9 0.8 0.7 0.6 0.5 Calorimetry isolation Pixel isolation Tracker isolation High lumi eta <2.4 Eff(W)>97% 1.5 2 Rejection MB Efficiency MB 0.5 0.4 0.3 High lumi Pt(gen)>22 GeV eta <2.4 Calorimeter isolation Pixel isolation Tracker isolation 0.4 2.5 0.2 0.3 3 0.2 4 5 0.1 0.1 0 10 100 10 20 40 Pt gen. 0 0.75 0.8 0.85 0.9 0.95 1 Efficiency W S. Lacaprara: CMS: Muon System and Physics performance 23
Physics Performance to here! From here... S. Lacaprara: CMS: Muon System and Physics performance 24
Physics Performance (1) Single muon rate Low (a) and High (b) lumi: Rate [Hz] 10 4 10 3 generator L1 L2 L2 + isolation (calo) L3 L3 + isolation (calo + tracker) Rate [Hz] 10 5 generator L1 10 4 L2 L2 + isolation (calo) L3 L3 + isolation (calo + tracke 10 3 10 2 10 2 10 (a) 10 (b) 1 10 20 30 40 50 60 threshold [GeV/c] m p T 1 10 20 30 40 50 60 threshold [GeV/c] m p T S. Lacaprara: CMS: Muon System and Physics performance 25
Physics Performance (2) Contribution to single rate High Lumi before and after isolation: 1 pi/k/b/c suppressed! S. Lacaprara: CMS: Muon System and Physics performance 26
o Physics Performance (3) Di-muon rate (symmetric thr.) Low (a) and High (b) lumi: Rate [Hz] 10 4 generator 10 3 L1 L2 L2 + isolation (calo) Rate [Hz] L3 10 4 L3 + isolation (calo + tracker) generator 10 5 L1 L2 L2 + isolation (calo) L3 L3 + isolation (calo + tracker 10 2 10 3 10 1 (a) 0 5 10 15 20 25 30 threshold [GeV/c] m p T 10 2 10 (b) 1 0 5 10 15 20 25 30 threshold [GeV/c] m p T S. Lacaprara: CMS: Muon System and Physics performance 27
Di-muon p T threshold [GeV/c] 20 18 16 14 12 10 8 6 Physics Performance (4) Combined signal and di-muon rate Low (a) and High (b) lumi: 60 Hz 40 Hz 30 Hz 20 Hz (a) Di-muon p T threshold [GeV/c] 25 20 15 10 60 Hz 40 Hz 30 Hz 20 Hz (b) 4 80 Hz 2 14 15 16 17 18 19 20 21 22 23 24 Single muon p T threshold [GeV/c] 5 80 Hz 25 30 35 40 45 50 Single muon p T threshold [GeV/c] S. Lacaprara: CMS: Muon System and Physics performance 28
Physics Performance (5) Efficiency for ƒ and : Efficiency 1 0.9 0.8 0.7 L1 L2 L3 L3 + isolation Efficiency 1 0.9 0.8 0.7 L1 L2 L3 L3 + isolation 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.1 (a) 0.2 0.1 (b) 0 0 10 20 30 40 50 60 70 80 threshold [GeV/c] m p T 0 0 10 20 30 40 50 60 70 80 threshold [GeV/c] m p T S. Lacaprara: CMS: Muon System and Physics performance 29
Physics Performance (6) Efficiency for ƒ and : S. Lacaprara: CMS: Muon System and Physics performance 30
Physics Performance (7) Efficiency for vs single and di- thr.: L3 Singleµ after Iso L3 isolated Diµ 0.9 0.8 0.7 0.6 0.5 M Higgs M Higgs M Higgs = 200 = 160 = 120 0.9 M Higgs = 200 M Higgs = 160 = 120 0.8 0.7 0.6 0.5 M Higgs 0.4 0.4 0.3 0.3 0.2 0.1 (a) 0.2 0.1 (b) 0 0 20 40 60 80 100 µ T P threshold 0 0 10 20 30 40 50 60 70 µ T P threshold S. Lacaprara: CMS: Muon System and Physics performance 31
O t 5 7 p ) œ- 7 7 5 5 p Ž ( 8 / ( 7 8 v / } š Possible Working Point at Low and High Lumi Lumi thr. di ˆthr. total rate ( / Š / /Œ) L1-HLT GeV L1-HLT GeV frac) LL 14-19 3-7 ( / ) / HL 20-31 5-10 / / ) Different threshold for L1 and HLT, room for more exclusive HLT selection at lower threshold (correlation, topological trigger,- selection,... ), physics contents is relevant at low lumi, and can be enhanced with dedicated trigger below HLT threshold, e.g. mass reconstruction ( : trigger at L1 di- 9 šregional tracker reco ž, žwith full reco) can give 6for signal 8 Ÿ ev/yr u " total rate is dominated (especially at High Lumi) by, to lower the threshold must reject part of these events, without losing efficiency, S. Lacaprara: CMS: Muon System and Physics performance 32
» within D Œ Œ p p p p } p µ µ¹ º ² Œ Œ ³ p ± } } } } «qq ª Œ p p Signal Efficiency at Nominal Threshold Signal efficiency relative events with at least one muon in Higgs efficiency relative to events with within! #". #", #", and all S. Lacaprara: CMS: Muon System and Physics performance 33
¼ Summary CMS muon system has been designed with high redundancy for selection and reconstruction, High Level Trigger reconstruction already developed, will be very close to the offline reco, Fully exploit muon and tracker detectors to give excellent performance in term of resolution and efficiency, Still room for further optimization and development, Physics performance very promising, (eagerly) Waiting for first collision!! S. Lacaprara: CMS: Muon System and Physics performance 34
Backup Slide S. Lacaprara: CMS: Muon System and Physics performance 35
Ä Å Å Ç Å Á Å Å Á Ã Â À Muon Alignment Issue ½Rate of trigger): usable for alignment (o ½With perfect knowledge of luminosity to reach ½But with ½to achieve resolution) years!! /sector q ¾ ½must have external alignment for field, m precision, GeV from L1 need several months, Ædays even at Low m precision (i.e. better than chamber m precision ½web of laser and optical link to align muon detectors also with inner tracker will be present in CMS ½in the endcaps might use halo muons, need special trigger S. Lacaprara: CMS: Muon System and Physics performance 36
8 8 "Í "Í È È { Ì Ë { Ë Muon Reconstruction Timing mean CPU time (ms/event) mean CPU time (ms/event) Low Lumi ]ÊÉ ÌHigh Lumi algorithm total excluding GEANE total excluding GEANE ]ÊÉ L2 640 100 580 100 Calo iso 100 25 90 40 L3 420 200 590 420 Pixel iso 65 65 320 320 Tk iso 190 190 370 370 Total 710 125 660 150 Average CPU time (on INTEL PIII previous selection, expect a factor GHz) for most of time spent in propagation in iron with GEANE, ""from Moore s law in event passing work in progress to replace it with a fully customizable and optimized propagator, expected major improvement. l, S. Lacaprara: CMS: Muon System and Physics performance 37