How to Measure Top Quark Mass with CMS Detector??? Ijaz Ahmed Comsats Institute of Information Technology, Islamabad
Outlines o o o o o o o High Pt top basic idea Methods for jets selection Top quark mass reconstruction from jets Jets clustering and clusters method for M top clus Underlying Event (UE clus ) estimation and subtraction Systematics errors Summary
Large Hadron Collider (LHC) Experiment 34 1 9 R = σ L = 80mb 10 cm s 10 / L Event Rate γfk b N P 34 = F = 10 cm s * 4πε n β Large distance collisions o Soft scattering Short distance collisions o Hard scattering (rare events) 1 Integrated luminosity=10fb -1 =10 33 s -1 cm - s
The CMS Detector Muon System Solenoid Magnet Electromagnetic Calorimeter (Ecal) Hadronic Calorimeter (Hcal) Silicon Tracker m long & 15 m in diameter More than 1 Million Geometrical volumes
Top quark properties and Decays Heaviest particle (spin ½,, charge /3) Origin Origin of mass, EWSB Short Short life time No No bound state 90% 10% need to reconstruct and identify ~44% muons b-jets light jets (u,d,c,s) missing E T (neutrinos) ~4.9% ~9% NLO Cross-section for tt~ production = σ(tt)~830 ± 100 pb -1
Boosted Top Quark Analysis Highly boosted top quarks : Decay back-to to-back Higher top boost : Small opening angle of W-boson W and b-quarkb High Pt top quarks : Large probability of jets overlaping in space. Invariant mass of the objects (jets/clusters) in larger cone around the top quark flight direction : Correlation with the real top quark mass. Top quark needs to have a larger boost : Pt > 00 GeV. WHY? Reduces the combinatorial background. The systematic effects due to jet energy calibration and gluon effects Potential to reduce the systematic errors
Kinematical variables Invariant mass Transverse momentum m T μ = p p μ = E p = p + x p p y Transverse mass m T = m + p T = E p z Transverse energy Pseudo-rapidity Rapidity Jet cone radius E T = η = y = Δ R 1 = ln E ln tan sin E E Δ ϑ + ϑ p p z z = η + Δ ϕ E + ln m T p z Missing Transverse Energy E T miss
Event Simulation: Tools and Methods 1. Event generation (PYTHIA, TOPREX, CMKIN). Simulation of the interaction of the generated particles with the detector (OSCAR, FAMOS, CMSSW : GEANT4) 3. Simulation of digitized phase (FAMOS, CMSSW, ORCA) Level-1 1 trigger (100 KHz) High Level Trigger (100 Hz) 4. Local and global event reconstruction (FAMOS, CMSSW, ORCA) 5. Physics Analysis tools (PAW,ROOT)
Even Selection at Partonic Level tt bw + bw bbqqμν μ P top t > 00 GeV, η < 3.0 P anti-top t top > 00 GeV, η < 3.0 + tt bw bw bqqblν ( l = μ) P tμ > 30 GeV, η <.0 P tq > 0 GeV, η <.5 Fast simulation based samples 165 Top mass point = 0K events 175 Top mass point = 50K events 185 Top mass point = 0K events Pile-up events are included Cross-section approximately 1% of the total tt cross-section tt + bw bw bqqblν ( l = μ) No. of events With pile-up Int. luminosity fb -1 X-section pb 49535 7.3 6.85
Distributions at Decay Vertex (1) P t top MC η top ΔR(q,qbar) P t W
Distributions at Decay Vertex () ΔR(top,b-par) ΔR(top,W) ΔR(top, min W-quarks) ΔR(top, max W-quarks)
Reconstruction MET-> Missing Transverse Energy MET > 30 GeV At least 1 iso. muon, P t >0 GeV, η <.0 leptonic W reco mass
Muon reconstruction and isolation Isolation Criteria Most likely muon tracks
Jets Reconstruction and Identification combined b-tag discriminator combined b-tag disc. > 0 (60% b-tag efficiency based on the secondary vertex, a vertiex which is displaced from the primary vertex. Leading light jets P t P t jets > 0 GeV Leading b-jets P t P t b jets > 0 GeV
Jet-Parton Matching light jets corresponds to quarks from W boson Four possible jet combinations Take best combination which gives correctly matching (J1,j), (q1, q) worst of quarks matching angles Correctly matched if ΔR < 0.4 *************** (J1,q1), (j1,q) (j, q1), (j,q) ΔR(j1,q1) ΔR(j1,q) ΔR(j,q1) ΔR(j,q) I1=Max (ΔR(j1,q1),ΔR(j1,q)) I=Max (ΔR(j,q1),ΔR(j,q)) Min(I1, I)<0.4
Various Approaches for Jets Selection Steps to measure top quark direction Leading jets > = b-tagged b jets, > = non b-tagged b jets Exactly 4 jets, = b-tagged b jets, = non b-tagged b jets > leading b-jets, b light jets with m jj closest to W mass
Top Quark Selection: Leading Jets Topology Kinematical cuts Selection efficiency % No. of events Before selection 100 49535 no of iso. muons 93.6 46370 1 iso muon P t > 30 GeV 1 reco light jets P t > 0 GeV 9.7 4590 91.1 45117 1 quark matched = 4.7% quarks matched = 18.17% reco light jets η <.5 1 b-jet b P t > 0 GeV 73.6 36484 55.6 7543 b-jets b η <.5 18.6 914 m jj m nom W < 0 GeV 8.5 435 m nom W = 65.4 (gaussian fitted correctly jet-parton matching) b-jet with biggest angle wr.t muon called Hadronic b-jets
Top Quark Selection: Four Jets Topology Hadronic top selection Four highest Pt jets selection b-jets identification with b-tagging Two light jets invariant mass reconstruction Hadronic b-jet requires for away from isolated muon with maximum distance 0.4 or closests to light jets 1 quark matched = 0.98% quarks matched = 43.6% Kinematical cuts Selection efficiency % No. of events Before selection 100 49535 no of iso muons 93.6 46370 1 iso muon P t > 30 GeV 1 reco light jets P t > 0 GeV Exectly 4 jets η <.5 9.7 45916 9.7 45915 1.3 10551 Exectly light jets 8.0 3941 Exectly b-jetsb 8.0 3941 m jj m W < 0 GeV 3.9 1937
Top Quark Selection: JJ W Kinematical cuts Selection efficiency % No. of events Before selection 100 49535 1 quark matched = 0.76% quarks matched = 40.6% no of iso muons, P t > 30 GeV,, η <.0 jj W, P t > 0 GeV, η <.5 b-jets b P t > 0 GeV, η <.5 9.7 4590 73.6 36484 18.6 914 m jj m W < 0 GeV 11.9 5917
W Mass from Three Approaches Nominal mass fitted mass ~ 65 GeV Same m W nominal used in all selections (JPM)
Comments on M jjb Study based on shape of distributions for top direction determination. Expalored three types of selection criteria for hadronic top mass reconstruction Four jets selection results low efficiency with higher W purity Jets with invariant mass close to W have higher efficiency with intermediate purity of W Leading jets selection gives sharp and narrow dist. shape with less long tail behaviour and reasonable selection efficiency
Top Quark Selection: Leading Jets Topology First peak from the wrong jet combination Exchanging the leptonic b-jet into hadronic b-jet One of the 4 leading jets could be coming from the gluon radiation Soft QCD events Second peak corresponds to the correct combinations At preselection level we demand high Pt jets Wrong Right (GeV/c) T Top P 300 Mean 0.974 Meany 146.8 RMS 0.4 RMSy 77.97 50 00 M jjb (P top T >00 GeV) Efficiency ~ % 150 100 0.5 0.6 0.7 0.8 0.9 1 1.1 1. 1.3 1.4 ΔR(jets-jets)
Calibrated Top Quark Mass Peaks are shifted towards the nominal Top mass
Calo Clusters Reco Method Invariant mass of all calorimeters clusters in Δη Δϕ around top direction m Calorimetric Clusters Reconstruction Method clusters nδr nδr ( ΔR) = ( E P ) = ( E i) ( P i i= 0.7 i= 0.7 ) E i represents total energy of the ith cluster ndr runs over all clusters within selected cone size P i its 3-momenta 3 vector Known: E,η, ϕ about clusters Assumptions: considering particles to be mass-less m 0 E P Px = E sinϑ cosϕ Py = E sinϑ sinϕ Pz = E cosϑ
E Reco clusters pseudo-rapidy Th clus >1 MeV Calorimeters identifications R = X + Y ECAL ( Z <350 cm, R < 170 cm) HCAL ( Z <350 cm, R < 300 cm)
E T clus clus Deposition
M clus top Determination Jets decay back back Reduce intrinsic complexities of effects due to energy leakage outside a narrow cone Reduce system errors arising due to jet calibration Clusters lie close to the top quark flight direction
UE Estimation Method It is not only minimum bias event The underlying event is everything except the two outgoing hard scattered jets In a hard scattering process, the underlying event has a hard component (initial+final( state radiation and particles from the outgoing hard scattered partons) ) and a soft component (beam-beam remnants) Jet Isolation variable Jet Isolation ΔR = 0.7 66.89 38 ΔR = 0.8 63.07 33 ΔR = 0.9 618.47 9 ΔR = 1.1 614.85 ΔR = 1.5 599.83 8 8 56 Electromagnetic Calorimeter <E t > / cluster (MeV) < no of clusters > / high P t event η <0.7 η < 1.4 η <.1 η < 3.0 η > 3.0 η < 5.0 509.19 88 503.17 80 496.66 73 485.4 459.49 445.77 134 439.69 15 433.11 116 40.8 394.11 Hadronic Calorimeter 363.00 9 356.43 18 349.36 180 335.58 306.64 136 36.77 18 36.76 181 36.69 330 36.35 37.03 169 36.78 35 31.39 33 315.67 08 304.6 83.05 53 Min ΔR(jets,clus)>0.7
Top Quark Mass M clus Before UE Subtraction top clus and UE clus and UE clus Subtraction After UE Subtraction A correlation with a slop about 0.7866 is observed, which implies that error of 0.9 GeV in the mean of peak translates to an statistical uncertainity of 0.9/0.786 = 1.1456 GeV/c in M jjb and 1.1 1.6 GeV/c in M top clus 50K events corresponds to 7. fb -1, statistical uncertainty about δm=1-1.5 1.5 GeV on top mass.
Systematic Source of uncertainty Re-calibration Electronic noise ISR on/off FSR on/off B-quark fragmentation UE estimate (+-10%) Cluster mis calibration:+-1(5)% Calorimeter: e/h=1.5 (1.63) Δm top (GeV/c ) 0.9 1. 0.14 0.07 0.3 1.34 0.7(1.3) 0.8(0.3)
Summary An alternate method for top mass reconstruction in CMS is presented, which strongly depends on CMS Calorimeters. A new method for Underlying Event (UE) estimation, subtraction and calibration is developed. This analysis is performed with both Full and Fast Simulations techniques. Statistical error on top mass M jjb (1-1.5 1.5 GeV) and 1.1 1.6 GeV/c in M clus top clus is is estimated.