Performance of tau-lepton reconstruction and identification in CMS

TL;DR

The paper assesses tau_h reconstruction and identification performance in CMS using 7 TeV data (36 pb^-1), focusing on two algorithms, HPS and TaNC, implemented with PF objects. It measures efficiencies and misidentification rates via Z → ττ events, derives data-to-MC scale factors, and characterizes tau_h decay-mode reconstruction and energy scale. The results show good data-MC agreement, with modest energy-scale uncertainties and misidentification rates around 1% for jets, enabling reliable ττ-based analyses such as Z and Higgs studies. These measurements provide essential inputs for precision tau physics in early LHC data and for calibrating tau-related analyses at CMS.

Abstract

The performance of tau-lepton reconstruction and identification algorithms is studied using a data sample of proton-proton collisions at sqrt(s)=7 TeV, corresponding to an integrated luminosity of 36 inverse picobarns collected with the CMS detector at the LHC. The tau leptons that decay into one or three charged hadrons, zero or more short-lived neutral hadrons, and a neutrino are identified using final-state particles reconstructed in the CMS tracker and electromagnetic calorimeter. The reconstruction efficiency of the algorithms is measured using tau leptons produced in Z-boson decays. The tau-lepton misidentification rates for jets and electrons are determined.

Figures (6)

• Figure 1: Invariant mass distribution of the $\mu$-jet system for preselected events which pass (left) and fail (right) the HPS "loose" $\tau_{\rm{h}}$ identification requirements (solid symbols) compared to predictions of the MC simulation (histograms).
• Figure 2: The expected efficiency of the $\tau_{\rm{h}}$ algorithms as a function of generated $p_{\mathrm{T}}\xspace^{\tau_{\rm{h}}}$, estimated using a sample of simulated $\mathrm{Z} \to \tau \tau$ events for the HPS (left) and TaNC (right) algorithms, for the "loose", "medium", and "tight" working points.
• Figure 3: (left) The fraction of generated $\tau_{\rm{h}}$ decays of a given type reconstructed in a certain decay mode for the HPS "loose" working point from simulated $\mathrm{Z}\to\tau\tau$ events. (right) The relative yield of $\tau_{\rm{h}}$ reconstructed in different decay modes in the $\mathrm{Z}\to\tau\tau\to\mu\tau_{\rm{h}}$ data sample compared to the MC predictions. The MC simulation is a mixture of the signal and background samples based on the corresponding cross sections, as shown by the histograms.
• Figure 4: The reconstructed invariant mass of $\tau_{\rm{h}}$ decaying into one charged and one neutral pion (left) and into three charged pions (right) from data, compared to predictions of the simulation. The solid lines represent results of the best fit described in the text and the dashed lines represent the predictions with the tau energy scale, TauES, varied up and down by 3% with respect to the best fit value.
• Figure 5: Misidentification probabilities for jets to pass "loose" working points of the HPS (left) and TaNC (right) algorithms as a function of jet $p_{\mathrm{T}}\xspace$ for QCD, $\mu$-enriched QCD, and $\mathrm{W}$ type events. The misidentification rates measured in data are shown by solid symbols and compared to MC prediction, displayed with open symbols.