Measurement of the Z production cross section with ATLAS Frank Seifert supervised by Arno Straessner, Wolfgang Mader on behalf of the ATLAS Z ττ analysis working group 1
Outline 1. Motivation 2. Tau identification 3. Cut flow and W suppression cuts 4. Results and combination 5. Summary 2
1. Motivation Measurement of the Z cross section in ATLAS is challenging, because: h One hadronically decaying tau needs to be identified. Huge Multijet cross section at LHC Several background processes with similar signature Multijets and W+jets can produce leptons and fake τ Z 1 1 2 2 (e,µ) 3 Low trigger thresholds for the leptons are needed Unprescaled trigger thresholds become higher with more pile up 3
2. Tau identification τ leptons have a a mean lifetime of 2.9 10-13 s. decays ~2/5 into e±/μ± and ~3/5 into hadrons. Multijets and hadronically decaying τ have similar signature and can be distinguished on a statistical basis. 4
2. Tau identification τ leptons have a a mean lifetime of 2.9 10-13 s. decays ~2/5 into e±/μ± and ~3/5 into hadrons. Multijets and hadronically decaying τ have similar signature and can be distinguished on a statistical basis. Variables with discriminating power are e.g.: Number of tracks Width pf the decay producs Calorimeter and track radius Invariant cluster and track mass Transverse flight path significance 5
2. Tau identification Variables with significant discriminating power are put into a BDT. A cut on the score with an optimum between signal efficiency and background rejection is chosen. 6
2. Tau identification Variables with significant discriminating power are put into a BDT. A cut on the score with an optimum between signal efficiency and background rejection is chosen. Boosted decision tree (BDT): multivariate method The cut on the BDT score depends on the pt of the tau candidate (reach flat signal efficiency). Working point with ~60% signal efficiency [1] is chosen (BDT medium) as optimal one for the analysis. [1] The ATLAS Collaboration, Performance of the Reconstruction and Identification of Hadronic Tau Decays with ATLAS, ATLAS CONF 2011 152, Nov, 2011. 7
3. Cut flow and W suppression cuts Dilepton veto Against Z ll background Opposite charge of lepton and h Cuts against W l + jets background h : Nprong= 1 or 3, charge = 1 8
3. Cut flow and W suppression cuts Dilepton veto Against Z ll background Opposite charge of lepton and h Cuts against W l + jets background (mu τh channel): mt = sqrt( 2pTlep MET (1 cos[δϕ(lep, MET)]) ) Cut values: ΣcosΔϕ = cos[δϕ(lep, MET)] + cos[δϕ(tau, MET)] Σ cosδϕ > 0.15 mt < 50 GeV 9
Cuts against W l + jets background (el τ h channel): mt = sqrt( 2pTlep MET (1 cos[δϕ(lep, MET)]) ) ΣcosΔϕ = cos[δϕ(lep, MET)] + cos[δϕ(tau, MET)] 10
Visible mass distributions e τ had channel μ τ had channel The electron τhad fake rate is much higher than the muon τhad fake rate! For the final measurement, the events within the mass window of 35GeV < mvis < 75GeV are taken to further reduce Z ee or Z μμ background. 11
4. Results and combination Summary of systematic uncertainties: Most dominant systematics contributions. 12
4. Results and combination Combination with the BLUE (best linear unbiased estimate) method [1] Combining correlated estimates of a single physical quantity In correlated case: error matrix E BLUE method looks for an estimate y, which is a linear, unbiased combination with the minimal variance σ2. Constant weighting factors αi of the individual measurements. y = Σαiyi Σαi = 1 σ2 = αte α = ΣiΣj E αiαj ij Finding the αi which minimize σ2. Method is very sensitive to large uncertainties with correlations between the measurements. Therefore the biggest correlated systematics have been excluded from E for the combination and added on the combined result by normal error propagation. Gives a safe combination value with a slightly higher final uncertainty. [1] L. Lyons, D. Gibaut, and P. Clifford, How to combine correlated estimates of a single physical 390 quantity, Nucl. Instrum. Meth. A270 (1988) 110. 13
4. Results and combination The Z ττ cross section is calculated as: Ndata Nbackground σ(z ττ) = AZ x CZ x Luminosity x BR with BR = branching ratio ( ττ μ-τh,el τh, el μ ) μ τh channel el τh channel el μ channel Ndata 5184 2600 1035 Nbackground 794 450 43 CZ 0.1417 0.0016 0.0955 0.0015 0.1348 0.0058 AZ 0.0976 0.0002 0.0687 0.0002 0.0784 0.0003 fiducial σ(z ττ)xbr [pb] total σ(z ττ) [pb] 20.0 0.4 2.0 0.7 912 18.6 94.0 34 15.9 0.5 2.0 0.6 998 29.4 130.8 37 4.7 0.3 0.4 0.2 964 51.1 81.0 36 Uncertainties are stat. syst. lumi. (Summary of systematics in backup.) Combination of these three channels leads to: 19.01.2012 σ = 921 23 79 34 pb Ztautau analysis 14
5. Summary Systematics dominated result in agreement with the theory prediction and to the measured Z ee/ cross section. 19.01.2012 Ztautau analysis 15
Backup 16
Correlations between the systematic uncertainties 17