Measurement of the Z to tau tau cross section with the ALAS detector On behalf of the ALAS Collaboration Charles University in Prague, Faculty of Mathematics and Physics, Institute of Particle and Nuclear Physics E-mail: jana.novakova@cern.ch he Z to tau tau cross section has been measured using data at s = 7 ev collected with the ALAS detector ahe LHC corresponding to the integrated luminosity of 1.34 1.55 fb 1. he analysis is performed in three different final states determined by decay modes of the tau leptons. Cross sections are measured separately in each final state in a fiducial kinematic phase space and extrapolated to the full phase space in the invariant mass region 66 116 GeV. he individual cross sections are combined together and the product of the total Z boson production cross section and the Z ττ branching fraction is measured to be.92±.2 (stat) ±.8 (syst) ±.3 (lumi) nb, in agreement with the NNLO theoretical expectations. PoS(ICHEP212)87 36th International Conference on High Energy Physics, July 41, 212 Melbourne, Australia Speaker. c Copyright owned by the author(s) under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike Licence. http://pos.sissa.it/
1. Introduction Processes with tau leptons play an important role in searches for new physics phenomena at the LHC. he decay of the Z boson to two tau leptons forms the main background to some of the searches, e.g. H ττ. herefore, it is importano measure its production cross section precisely. Furthermore, the cross section measurement of Z ττ process is a complementary analysis to the Z ee and Z µµ precision measurements. Finally, the well known Standard Model processes involving tau leptons are important for the commissioning and validation of τ identification techniques [1] used ahe ALAS experiment [2]. his paper presents the measurement of the Z ττ cross section performed in proton-proton collisions at centre of mass energy of 7 ev. he analysis using dataset from 211 corresponding to the integrated luminosity of 1.34 1.55 fb 1 is documented in Ref. [3]. hree different final states determined by the τ lepton decay modes: Z ττ µ+ hadrons+3ν (denoted τ µ τ had in the following), Z ττ e+hadrons+3ν (denoted τ e τ had ) and Z ττ µ+ e+4ν (denoted τ e τ µ ) are used. he branching fractions given by the PDG are (22.5±.9)%, (23.13±.9)% and (6.2±.2)% respectively [4]. Furthermore, the high τ-purity dataset obtained in the τ µ τ had channel is used for studies of the distributions of variables relevant for the hadronic τ identification. 2. Event selection After selecting good collision candidate events, data are required to pass dedicated triggers. he muon trigger with a transverse momentum threshold at 15 GeV and an isolation requirement is used in the τ µ τ had and τ e τ µ channels, while a combined electron + hadronic tau trigger (with transverse energy cut at 15 GeV on electron + 16 GeV on hadronic tau) in the τ e τ had channel. Next, exactly one muon candidate (or electron) and a hadronic τ is required in the τ µ τ had (or τ e τ had ) channel. Similarly, one muon and one electron with opposite charges are looked for in the τ e τ µ channel. he transverse momentum threshold is seo 17 GeV for muons and electrons and to 2 (or 25 GeV) for hadronic τ in τ µ τ had (or τ e τ had ) channel to avoid the turn-on region of the trigger efficiency. Special cuts reducing the W +jets background are applied. he variables used are motivated by a different evenopology in the signal and background events. he variables as defined in the τ µ τ had channel are shown bellow (the definitions in the other channels are analogous): cos( φ)=cos(φ(µ) φ(e miss ))+cos(φ(τ had ) φ(e miss )) (2.1) m = 2p (µ) E miss [1 cos φ(µ,e miss )] (2.2) he first variable is used in all three channels and the optimal cut is found to be cos( φ)>.15. he transverse mass defined in Equation (2.2) is required to be smaller than 5 GeV in the τ µ τ had and τ e τ had channels. he distributions of these variables are shown in Figure 1. he t t background contributes significantly in the τ e τ µ channel. Contrary to Z ττ, t t events in the dilepton channel can be characterised by multiple high-p jets and leptons and large E miss. herefore, events in the τ e τ µ channel are required to have Σ < 14 GeV where Σ is defined as Σ = p (µ)+e (e)+e (jets)+e miss. (2.3) PoS(ICHEP212)87 2
Events/.5 7 6 5 4 3 Ldt = 1.34 fb, s = 7 ev Data 211 γ*/z ττ W e ν γ*/z ee 7 6 5 4 3 Ldt = 1.55 fb, s = 7 ev Data 211 2 2 1-2.5 -.5.5 1 1.5 2 (a) τ e τ had channel cos φ 1 2 4 6 8 1 12 14 (b) τ µ τ had channel m (µ, E miss ) [GeV] Figure 1: he distributions of (a) cos( φ) in the τ e τ had channel and (b) m in the τ µ τ had channel [3]. All selection criteria excephe cuts on these two variables are applied. he red line indicates the cut value used. he distribution of Σ is shown in Figure 2 (a). Finally, the Z ll background, wherelstands for e or µ, is reduced by requiring the invariant mass of the visible parts of the decay (i.e. not considering the neutrinos) to be within the mass window 35 GeV to 75 GeV. While the Z ll events tend to accumulate in the region around 9 GeV, the Z ττ signal events peak at around 6 GeV due to the missing energy of the neutrinos. he visible mass distributions for all three channels are shown in Figures 2 (b) and 3 (a), (b). Events/1 GeV 22 2 18 16 14 12 1 8 6 4 2 Ldt = 1.55 fb, s = 7 ev Data 211 γ */Z ττ W lν γ*/z ll tt 1 2 3 4 5 6 7 8 9 1 [GeV] 12 1 8 6 4 2 ALAS Preliminary Ldt = 1.55 fb, s = 7 ev Data 211 γ */Z ττ W lν γ*/z ll tt 2 4 6 8 1 12 14 m vis (e,µ) [GeV] PoS(ICHEP212)87 (a) τ e τ µ channel (b) τ e τ µ channel Figure 2: he distributions of (a) and (b) the visible mass in the τ e τ µ channel [3]. All selection criteria excephe cut on the plotted variable itself are applied. he red line in (a) indicates the cut value used. 3. Expected number of background events Contributions of the non-dominant backgrounds (t t and dibosons in all three channels and W, Z in the τ e τ µ channel) are obtained from Monte Carlo simulations. All other backgrounds (multijet in all three channels and W, Z backgrounds in τ µ τ had and τ e τ had channels) are derived by partially or fully data-driven methods. Normalisation factors are derived in W/Z-enriched control 3
35 3 25 2 15 Ldt = 1.34 fb, s = 7 ev Data 211 γ*/z ττ W e ν γ*/z ee Ldt = 1.55 fb, s = 7 ev 6 Data 211 5 4 3 1 2 5 2 4 6 8 1 12 14 (a) τ e τ had channel m vis (e, τ h ) [GeV] 1 2 4 6 8 1 12 14 (b) τ µ τ had channel m vis (µ, τ ) [GeV] h Figure 3: he distributions of the visible mass in the (a) τ e τ had channel and (b) τ µ τ had channel [3]. he contribution of the Z ll background is larger in the τ e τ had due to the higher fake rate for electrons. All selection criteria excephe cut on the visible mass are applied. regions to correctly scale the Monte Carlo predictions. he multijet background contribution is estimated from the multijet-enriched region with inverse isolation requirement on the lepton and is extrapolated to the signal region. More details abouhe methods used are given in Ref. [3]. he expected number of background events and the number of observed events are summarised in able 1. able 1: he expected number of background events and the number of observed events after the full selection [3]. he quoted uncertainties are statistical only. he notation l stands for e and µ only. τ µ τ had (1.55 fb 1 ) τ e τ had (1.34 fb 1 ) τ e τ µ (1.55 fb 1 ) γ /Z ll 81±7 64±4 23±4 W lν 186±13 45±5 <.5 49±5 18±2 <.5 t t 31±1 17±1 2±1 15±2 6±1 18±2 432±3 3±21 13±7 otal background 793 ± 34 449 ± 22 56 ± 8 γ /Z ττ 4544±49 229±25 981±26 N obs 5184 26 135 PoS(ICHEP212)87 4. Cross section measurement and results he measurement of the cross section is done separately in each channel and the obtained values are then combined. he calculation is performed using the formula σ(z ττ,m ττ [66,116] GeV) B= N obs N bkg A Z C Z L where N obs is the number of observed events, N bkg is the number of estimated background events, B is the branching fraction for the channel considered and L denotes the integrated luminosity for (4.1) 4
the final state of interest. A Z is the geometrical and kinematic acceptance factor for events with the ττ invariant mass between 66 GeV and 116 GeV. C Z stands for the experimental correction factor which accounts for the efficiency of triggering, reconstructing and identifying the Z ττ events. Several sources of systematic effects have been studied. he dominant systematic uncertainties contributing in all three channels are the energy scale, luminosity and A Z uncertainty. Furthermore, the tau identification leads to a large uncertainty in the τ µ τ had and τ e τ had channels, the tau trigger efficiency and electron identification efficiency in the τ e τ had channel. he measured cross sections with their statistical and systematic uncertainties are summarised in able 2. able 2: he production cross section times branching fraction for the Z ττ processes measured in each final state [3]. Final State σ(z ττ,m ττ [66,116] GeV) τ µ τ had τ e τ had τ e τ µ.91±.1(stat)±.9(syst)±.3(lumi) nb 1.±.2(stat)±.13(syst)±.4(lumi) nb.96±.3(stat)±.9(syst)±.4(lumi) nb he individual results are combined by means of the BLUE (Best Linear Unbiased Estimate) method [5, 6] leading to the value of σ(z ττ) =.92 ±.2(stat) ±.8(syst) ±.3(lumi) nb. he individual cross sections together with the combined result are shown in Figure 4. he theoretical expectation [7, 8, 9] of.96±.5 nb is also shown. Z τ µ 1.55fb Z τ e 1.34fb Z τ e 1.55fb 33-36pb τ h τ h τ µ Z ee/µµ combined Z ττ combined 1.34.55fb ALAS Preliminary Stat Syst Stat Syst Stat Lumi heory (NNLO).7.8.9 1 1.1 1.2 σ(z ττ, 66<m <116 [GeV]) [nb] inv PoS(ICHEP212)87 Figure 4: he Z ττ individual cross section measurements and the combined result [3]. he Z ll combined cross section measured by ALAS [1] is shown for comparison. he grey band indicates the uncertainty on the NNLO cross section prediction. 5. Variables used in the hadronic τ identification he sample obtained after the full selection in the τ µ τ had has a relatively high purity in hadronic τ decays allowing a study of the variables used by the τ identification for signal-like τ candidates. As an example, a calorimeter cluster mass (defined in Ref. [1]) is shown in Figure 5 (a) and the distribution of the Boosted Decision ree (BD) score [11, 12, 1] before any identification require- 5
ments on the τ candidate have been applied in Figure 5 (b). More distributions are compared in Ref. [3] and in general, a good agreement between data and Monte Carlo is observed. candidates/.2 τ 7 6 5 4 3 2 1 Ldt = 1.55 fb, s = 7 ev Data 211 1 2 3 4 5 6 7 8 (a) τ µ τ had channel [GeV] m clusters candidates/.2 τ 3 25 2 15 1 5 Ldt = 1.55 fb, s = 7 ev Data 211.1.2.3.4.5.6.7.8.9 1 (b) τ µ τ had channel BD Score Figure 5: Variables relevant for the hadronic τ identification: (a) calorimeter cluster mass and (b) distribution of the BD score [3]. 6. Summary he measurement of the Z ττ cross section with the ALAS detector using collision data collected ahe centre-of-mass energy of 7 ev corresponding to the integrated luminosity of 1.34 1.55 fb 1 has been presented. he measurements are performed in three channels (τ µ τ had, τ e τ had and τ e τ µ ) and the individual results are combined together resulting in the total cross section of.92±.2(stat)±.8(syst)±.3(lumi) nb which is in a good agreement with the theoretical predictions and previous measurements. he obtained sample after all selections has a high purity of τ leptons and is used to study a number of variables relevano the identification of hadronic τ decays. A good description of these variables by the Monte Carlo simulation is observed. PoS(ICHEP212)87 References [1] ALAS Collaboration, ALAS-CONF-21152, http://cdsweb.cern.ch/record/1398195. [2] ALAS Collaboration, 28 JINS 3 S83. [3] ALAS Collaboration, ALAS-CONF-212-6, http://cdsweb.cern.ch/record/1426991. [4] Particle Data Group Collaboration, J. Phys. G37 (21) 7521. [5] L. Lyons, D. Gibaut, and P. Clifford, Nucl. Instrum. Meth. A27 (1988) 11. [6] A. Valassi, Nucl. Instrum. Meth. A5 (23) 391 45. [7] K. Melnikov and F. Petriello, Phys. Rev. D74 (26) 11417. [8] R. Gavin, Y. Li, F. Petriello, S. Quackenbush, Comput. Phys. Commun. 182 (211) 2388 243. [9] S. Catani, L. Cieri, G. Ferrera, D. de Florian and M. Grazzini, Phys. Rev. Lett. 13 (29) 821. [1] ALAS Collaboration, Phys. Rev. D85 (212) 724. [11] L. Breiman, J. Friedman, C. Stone, and R. Olshen, Chapman & Hall, 1984. [12] Y. Freund and R. Shapire, Proceedings 13th International Conference on Machine Learning, 1996. 6