$% CDF quark 7 3 5 ( "#!
Tevatron Run II Started Spring 1. proton-antiproton collider with (Run I :. antiproton recycler commissioning electron cooling operational by Summer 5. increase in luminosity. 36 bunches 36 bunches. crossing time. Long term luminosity projection : Base goal : Design : by end FY9.
Tevatron Performance Total Luminosity (pb -1 Year Month 9 8 7 6 5 4 3 3 4 5 1 4 7 1 1 4 7 1 1 4 7 1 1 Delivered To tape 1 15 5 3 35 4 Store Number Record :. of collisions delivered in Run II. Tevatron delivered more than CDF already collected on tape. in 4.
CDF Run II Detector General purpose detectors with : tracking in magnetic field precision tracking with silicon calorimeter muon chamber Already collected on tape. This analysis uses.
Top Quark Physics - motivation - Most recently discovered quark, and the least well understood. Discovered by CDF and D in 1995. Intimate relationship with EWSB. Top mass is close to the EWSB scale. Strongly coupled to Higgs boson : Top may be sensitive to new physics. Studying tests electroweak theory.. We can look for non-sm production or decay of : Background for new particle searches. Study on is only possible at Tevatron until LHC.
Top Mass Measurement - Motivation Top mass is a fundamental parameter of SM. Related to Higgs mass through radiative correction. Top and mass measurements constrain Higgs mass. data (LEP direct search (LEP/Tevatron Run I Top mass measurement in Run I : Mass of the Top Quark Measurement M [GeV/c ] CDF di-l 167.4 ± 11.4 D di-l 168.4 ± 1.8 CDF l+j 176.1 ± 7.3 D l+j 18.1 ± 5.3 CDF all-j 186. ± 11.5 15 175 M [GeV/c ] χ / dof =.6 / 4 TEVATRON Run-I 178. ± 4.3
"! ( * * Production and Decay of Top Quark at Tevatron Production Predominantly pair production. Theoretical total cross section :. events in an hour at. q q t t + g g t t Decay Decays before it hadronizes. #%$ q q t t b W + W - b q q v l Mode Br. ( dilepton 5 Clean but few signal. Two lepton+jets 3 One in final state. all jets 44 Challenging background. in final state. + X ID and backgrounds challenging.
Particle Identification and Event Selection One isolated high : cluster and track. lepton (., shower shape, matching between calorimeter :, energy deposit in calorimeter, matching between muon detector hit and track. Missing to ensure there was a in the final state. Jets are reconstructed using JETCLU algorithm with cone size.4. 1 and tag channels : More than three jets with The fourth jet with tag channel : Four jets with When jets are found in a event, we only consider the leading four jets as the products of decay. We use -tagging algorithms SECVTX and Jet Probability....
" /. @ 1 $ #? 4 H H M U O N P Q Event Top Mass Reconstruction in Each fit "! $% #, -, / ( *. -,, * ( + + + + 1> <= %; 5768:9 13 H EGF J I < EGF I B, EGF D C AB, Fluctuate particle momenta around measured values. Constrain reconstructed masses. K L K Assume. as a free parameter. Minimization with solutions -tagged jets O U 1 jet-parton permutations. -jets., and 4 1 1 4 jets from RTS solutions. combinatorial background. Wrong reconstruction Choose solution with smallest. reconstructed mass in each event.
< Combinatorial Background -taggers : SECVTX - displacement of secondary vertex. Jet Probability - impact parameter of tracks. Combinatorial Background in Monte Carlo ( tag 1 tag tag Subdivide the sample into, 1, and tag samples. H EGF J Tightened tagging condition reduces combinatorial background. of events are dilepton or w/ gluon jet.
? of Candidates and Background Estimate estimated number of background events non- process tag 1 tag tag / +LF + HF / single (s single (t total Candidates 4 17 11 193 16 16 J tag Bkg template arbitrary Background template for tag channel 14 1 1 8 6 4 1 15 5 3 35 Reconstructed Top Mass (GeV/c 1 tag Bkg template arbitrary Background template for 1 tag channel.8.7.6.5.4.3..1 non-w + Mistags W + HF 1-, s-chan 1-, t-chan 1 15 5 3 35 Reconstructed Top Mass (GeV/c tag Bkg template arbitrary Background template for tag channel.45.4.35.3.5..15.1.5 non-w + Mistags W + HF 1-, s-chan 1-, t-chan 1 15 5 3 35 Reconstructed Top Mass (GeV/c
Candidate Event with SECVTX Tagged Jets in CDF.8 Jet1 8.9 GeV CDF II Preliminary Secondary Vertex HT = 9 GeV.6 L xy =.1 mm Y (cm.4. I.P. L xy =.7 mm Jet 65.6 GeV MET 4. GeV Electron 5.7 GeV Jet3 35.4 GeV -.6 -.4 -. -. X (cm
C J? / Mass Template Fitting Function We will describe the signal / background distribution shapes with the following functions. mass template : : reconstructed mass in the event : true mass of quark K K D C where dependence K include (to K M M M background template : J D is fixed to 1. For 1 and tag channel
Template Fitting ( tag sample tag Signal Template (Gamma + Gauss HERWIG arbitrary scale samples with 1 different Signal function parametrization. HERWIG tt Monte Carlo.4.35.3.5..15.1.5.3.5..15.1.5 1 15 5 3 35 Entries 116 Mean 14.1 RMS 5.11 = 13 GeV/c m 1 15 5 3 35 Entries 1871 Mean 16.6 RMS 5.59 = 16 GeV/c m.4.35.3.5...15.1.5 1 15 5 3 35.3.5..15.1.5 Entries 139 Mean 149. RMS 5.3 = 14 GeV/c m 1 15 5 3 35 Entries 13 Mean 168.7 RMS 6.88 = 17 GeV/c m.35.3.5..15.1.5 1 15 5 3 35.5..15.1.5 Entries 1713 Mean 154.8 RMS 4.49 = 15 GeV/c m 1 15 5 3 35 Entries 3696 Mean 17.5 RMS 5.74 = 175 GeV/c m tag Bkg Template (Gamma arbitrary scale Background for Tag Channel Mean 16.1 RMS 35.11..15.1.5..15.1.5 Entries 148 Mean 177. RMS 7.15 = 18 GeV/c m.5..15.1.5 Entries 13 Mean 184. RMS 7.75 = 19 GeV/c m.5..15.1.5 Entries 47 Mean 19.5 RMS 7.79 = GeV/c m.5 1 15 5 3 35 Reconstructed Top Mass (GeV/c 1 15 5 3 35 1 15 5 3 35 1 15 5 3 35...18.16.14.1.1.8.6.4. 1 15 5 3 35 Entries 84 Mean 199 RMS 3.96 = 1 GeV/c m..18.16.14.1.1.8.6.4. 1 15 5 3 35 Entries 377 Mean 4.9 RMS 31.15 = GeV/c m..18.16.14.1.1.8.6.4. 1 15 5 3 35 Reconstructed Top Mass (GeV/c Entries 34 Mean 13.5 RMS 3. = 3 GeV/c m
. Obtained Template Functions - tag template arbitrary scale Template Functions for Tag Channel.3 m.5..15.1.5 = 145 GeV/c m = 155 GeV/c m = 165 GeV/c m m = 175 GeV/c = 185 GeV/c m = 195 GeV/c m = 5 GeV/c background 1 15 5 3 35 Reconstructed Top Mass (GeV/c 1 tag template Template Functions for 1 Tag Channel in template fitting - tag 1tag tag signal 178.4 / 878 93.7 / 86 774.5 / 697 background 8.9 / 34 36.7 / 3 8.9 / 34 tag template Template Functions for Tag Channel arbitrary scale.3 m.5..15.1.5 = 145 GeV/c m = 155 GeV/c m = 165 GeV/c m m = 175 GeV/c = 185 GeV/c m = 195 GeV/c m = 5 GeV/c background 1 15 5 3 35 Reconstructed Top Mass (GeV/c arbitrary scale.4 m = 145 GeV/c.35.3.5..15.1.5 m = 155 GeV/c m = 165 GeV/c m m = 175 GeV/c = 185 GeV/c m = 195 GeV/c m = 5 GeV/c background 1 15 5 3 35 Reconstructed Top Mass (GeV/c Shape of reconstructed mass distribution is parametrized as a function of K
/ ; Measurement of the Top Mass - Likelihood Definition / M / / M L / K? = M / / /, Look for distribution. Constrain " and! the reconstructed mass for each event in the sample to be fitted. true mass for each event in data sample. number of candidate events in the sample. number of signal events. number of background events. Template function parametrization. free parameters in the fit. K and background fraction that best reproduce the data of background around the estimated number. Likelihood Definition for Combined Fitting : /M L K
D? C J D? D Likelihood Fit of Candidates: Likelihood fit with tag sample Events/(1 GeV/c 4 3.5 3.5 1.5 1.5 +6.1 M = 18.9-5.8 (stat. GeV/c Data (11 evts Signal + Bkgd Bkgd only sample Reconstructed Top Mass (GeV/c of Candidates tag 16 11 1 tag 16 17 tag 193 4 and background frac as free parameter. K 1 1 14 16 18 Top mass (GeV/c 8 1 1 14 16 18 4 6 8 Reconstructed Top Mass (GeV/c Results : tag sample combined -ln(l/l max 8 7 6 5 4 3 1 Likelihood vs mass L = 16 pb -1, 1 and tag samples combined -ln(l/l max Likelihood vs mass 18 16 14 1 1 8 6 4 M = 177. +4.8-4.6 L = 16 pb L = 193 pb GeV/c (1 tag 1 14 16 18 mass (GeV/c -1-1 (tag
Systematic Uncertainty and Near Future Prospects We are updating the measurements with of data in CDF (double stat.. source combined tag (GeV/ (GeV/ ISR.61.69 FSR 1. 1. PDF.19.15 b-tagging..7 bkgd shape.68.17 jet resolution.71.54 jet energy scale 6.37 5.56 generator.11.3 total 6.56 5.77 Current systematic uncertainty dominated by Jet Energy Scale (JES. Lots of work done to reduce JES in CDF. Expect JES to scale down to level this Spring.
Top Mass Measurements in Run II - Summary CDF Run II results 1 CDF Run Preliminary Dilepton: φ of ν 17. ± ± 7.4-1 (L= 193pb Dilepton: P tt 176.5 ± ± 6.9-1 (L= 193pb Dilepton: ν weighting 168.1 ± ± 8.6-1 (L= pb Lepton+Jets: Multivariate 179.6 ± ± 6.8-1 (L= 16pb Lepton+Jets: M 177. ± ± 6.6-1 (L= 16pb Lepton+Jets: DLM 177.8 ± ± 6. -1 (L= 16pb Run 1 CDF Lepton+Jets 176.1 ± ± 5.3 (Run I only Run 1 D Lepton+Jets 18.1 ± ± 3.9 (Run I only.7.7 Run 1 World Average 178. ± ± 3.3 (Run I only z reco 16.6 16.6 17. 16. 11. 9.8 6.4 6.3 4.9 4.7 4.5 5. 5.1 5.1 3.6 3.6 15 16 17 18 19 Top mass (GeV/c Dilepton Underconstraint system with Assumption on distribution of kinematic variables,. Template fit. Lepton + Jets, Template Fit. Minimum assumpti MultiVariate DLM Template fit w/ kinematic variabl in likelihood. Matrix element.. D Run II results L+Jets ( Dilepton ( : :
D? C J D? D Summary Tevatron and CDF are operating well. We measured the mass with the lepton+jets channel. Following of candidates : sample Results of mass measurement : of Candidates tag 16 11 1 tag 16 17 tag 193 4 tag sample : combined : Updating our measurement ( Using few months. double stat.. Systematic uncertainty largely scaled down. Current total error in combined measurement
BACKUP
# SECVTX Algorithm main tagging algorithm in CDF ν b-jet light flavor jet p B hadron B hadron p light flavor jet Reconstruct the secondary vertex using tracks in the jet. Require cut on the decay length significance :. lepton b-jet hadrons decay after running a distance.
Jet Probability Algorithm (1 Track 3 Y Jet Track 1 # tracks tracks in b-jets Secondary Vertex φ D 1 1 Primary Vertex φ D φ 3 D 3 Track X tracks in light flavor jets (Track 1 : D is positive signed 1 (Track : D is negative signed (Track 3 : D is positive signed 3 x D / σd Assign sign ( direction. to the impact parameter of each track based on its
Jet Probability Algorithm ( Combine significance of all the tracks in the jet and calculate the probability of the jet originating in the primary vertex (Jet Probability. Prompt Jet Charm Jet Bottom Jet (Jet Probability in MC Events We can cut at arbitrary Jet Probability value for the -tagging. This enables us to loosen the -tagging condition easily.
Acceptance for HERWIG Events ( - for cut = - cut tag ( kinematic 7.19 tag 1.41 chi 1.9 mw 1 tag (.4 7.19..75..9 errors are from MC statistics. If we put. tag (.4 7.19.3 1.77. 1.14.97, we can expect SECVTX (.4.19..79..53..4.1.1 events in tag sample at We double the acceptance for tag channel by the use of jet probability.
Pseudo Experiment (PE tag channel, Jet Probability Cut=.5,, signal template histogram background template histogram Entries/(5 GeV/c Reconstructed Top Mass 5 4 3 massdbltagonly_mwcut Entries 3696 Mean 17.5 RMS 5.74 Entries/(5 GeV/c.5..15 Reconstructed Top Mass massdbltagonly_mwcut Entries 156 Mean 16. RMS 34.37 1 1 15 5 3 35 (GeV/c M generate pseudo signal masses Events/(1 GeV/c 3.5 1.5 1.1.5 1 15 5 3 35 (GeV/c M generate pseudo background masses pseudo data sample perform likelihood fit Reconstructed Top Mass (GeV/c MC events (11 evts Signal + Bkgd Bkgd only -ln(l/l max 8 Likelihood vs mass 7 6 5 4 3 1 1 1 14 16 18 Top mass (GeV/c PEs are performed for : sensitivity study validation of the likelihood fit method estimation of the systematic uncertainty.5 1 15 5 3 +6.6 = 175.7 (stat. GeV/c M -6.6
Sensitivity Study - expected statistical errors at - - Stat. Sensitivity (GeV/c Sensitivity on Top Mass 11 1 9 8 7 tag 1tag tag,1 tag comb. >=1 tag comb Stat. Sensitivity (GeV/c Sensitivity on Top Mass 4.4,1 tag comb. >=1 tag comb 4.35 4.3 4.5 6 5 4. 4.15 4 1-4 1-3 1-1 -1 1 Cut on Jet Probability The far left point is for SECVTXonly analysis. We use JP cut =.5 for now, because this is the best point among the calibrated. Pink point is for tag and SECVTX tag channels combined. 4.1 1-4 1-3 1-1 -1 1 Cut on Jet Probability Cut expected stat. error / K tag 4.355 SECV 4.71.1 4.186.5 4.17. 4.154 (GeV (.4 -.5..4 3.9.4 4..5 4.6
Sanity Check tag: output vs input output vs input 1 19 18 17 16 output mass (GeV/c Pseudo Exp. combined: output vs input output vs input 1 19 18 17 16 output mass (GeV/c Pseudo Exp. 15 output = input 15 16 17 18 19 1 input mass (GeV/c tag: (output-input vs input (output-input vs input 3 1-1 - (output - input (GeV/c Pseudo Exp. output = input -3 15 16 17 18 19 1 input mass (GeV/c 15 output = input 15 16 17 18 19 1 input mass (GeV/c combined: (output-input vs input (output-input vs input 3 1-1 - (output - input (GeV/c Pseudo Exp. output = input -3 15 16 17 18 19 1 input mass (GeV/c
Pull Study tag: pull mean vs input mean of pull. pull mean.15.1.5 - -.5 -.1 Pseudo Exp. -.15 Pull Mean = -. 15 16 17 18 19 1 input mass (GeV/c tag: pull width vs input combined: pull mean vs input mean of pull. pull mean.15.1.5 - -.5 -.1 Pseudo Exp. -.15 Pull Mean = -. 15 16 17 18 19 1 input mass (GeV/c combined: pull width vs input width of pull 1.3 1. 1.1 pull width width of pull 1.3 1. 1.1 pull width 1 1.9.8.7 15 16 17 18 19 1 input mass (GeV/c.9 Pseudo Exp. Pseudo Exp..8 Pull Width = 1 Pull Width = 1.7 15 16 17 18 19 1 input mass (GeV/c Pull width tends to be larger than 1. We will revisit this later.
Estimation of Systematic Uncertainties example of Monte Carlo generator dependence of tag channel signal templates results of pseudo experiments Template Histograms Results of Pseudo Experiments scaled by area.1.1.8 HERWIG, m t =175 PYTHIA, m t =175 Entries / ( GeV/c 1 1 8 median = 174.833 median = 174.8 HERWIG, m t =175 PYTHIA, m t =175.6 6.4 4. 1 15 5 3 35 (GeV/c We assign M t 13 14 15 16 17 18 19 1 (GeV/c as the systematic uncertainty coming from this source. m t
Checking of Result with Pseudo Experiments Pseudo experiments assume Red arrows indicate results from fit to data. tag channel # of experiments / (.5 GeV/c Statistical errors on mass 5 Pseudo Exp. L = 16 pb -1 Data 15 1 5-15 -1-5 5 1 15 statistical errors on mass (GeV/c combined fit # of experiments / (.5 GeV/c Statistical errors on mass 4 35 3 5 15 1 5 L = 16 pb -1 (1 and tag L = 193 pb -1 ( tag Pseudo Exp. Data -15-1 -5 5 1 15 statistical errors on mass (GeV/c
D? D? D Scale factor for the statistical error Our pull width is a few larger than 1. We have to apply a scale factor to correct the statistical error. Suppose an arbitrary scale factor for the statistical error. Measure the percentage of the PEs which has the input stat. error of the center value from the fit (Using K K Decide the scale factor so that the above percentage comes to tag: scale factor scan scale factor for stat. error Percentage of Pseudo Exp. (% 74 7 7 68 66 64 6 scale factor =1.43 6.85.9.95 1 1.5 1.1 1.15 1. scale factor scale factor = 1.43 within the combined: scale factor scan scale factor for stat. error Percentage of Pseudo Exp. (% 74 7 7 68 66 64 6 scale factor =1.19 6.85.9.95 1 1.5 1.1 1.15 1. scale factor scale factor = 1.19 sample..