A New Mass Reconstruction Technique for Resonances Decaying to di-tau

TL;DR

The paper tackles the difficulty of reconstructing the invariant mass of resonances decaying to tau pairs, hampered by multiple neutrinos. It introduces the Missing Mass Calculator (MMC), a likelihood-based method that scans possible invisible neutrino configurations guided by tau decay kinematics and MET, accommodating both hadronic and leptonic tau decays. In simulated ideal and realistic detector conditions, MMC yields a tau-tau mass peak near the true mass with high efficiency and improved resolution, outperforming the collinear approximation and avoiding its long-tail misreconstructions. Data from CDF and MC studies demonstrate MMC’s superior mass reconstruction performance and its potential to substantially boost Higgs searches in tau tau final states at the LHC and Tevatron.

Abstract

Accurate reconstruction of the mass of a resonance decaying to a pair of $τ$ leptons is challenging because of the presence of multiple neutrinos from $τ$ decays. The existing methods rely on either a partially reconstructed mass, which has a broad spectrum that reduces sensitivity, or the collinear approximation, which is applicable only to the relatively small fraction of events. We describe a new technique, which provides an accurate mass reconstruction of the original resonance and does not suffer from the limitations of the collinear approximation. The major improvement comes from replacing assumptions of the collinear approximation by a requirement that mutual orientations of the neutrinos and other decay products are consistent with the mass and decay kinematics of a $τ$ lepton. This is achieved by minimizing a likelihood function defined in the kinematically allowed phase space region. In this paper we describe the technique and illustrate its performance using $Z/γ^{*}\toττ$ and $H\toττ$ events simulated with the realistic detector resolution. The method is also tested on a clean sample of data $Z/γ^{*}\toττ$ events collected by the CDF experiment at the Tevatron. We expect that this new technique will allow for a major improvement in searches for the Higgs boson at both the LHC and the Tevatron.

Figures (9)

• Figure 1: Example of the transverse mass (left plot) defined as an invariant mass of $\hbox{${E\!\!\!\!/_{\rm T}}$}$ and visible $\tau$ decay products, and the fully reconstructed mass (right plot) using the collinear approximation for three event samples: inclusive $Z/\gamma^{*}\to\tau\tau$ and $gg\to H\to\tau\tau$ with $M_H$=115 and 130 GeV/c$^2$. Results are obtained for the fully hadronic $\tau\tau$ decay mode. Events are simulated with the realistic detector resolution (discussed in Sec. \ref{['sec:MMCdetector']}). All distributions are normalized to the unit area.
• Figure 2: Example of the probability distribution functions ${\cal P}(\Delta R,p_{\tau})$ for a particular value of the original $\tau$ lepton momentum ($p_{\tau}$). These functions are used in the calculation of the likelihood $\cal L$ for three cases: 1-prong $\tau$ (left plot), 3-prong $\tau$ (middle plot), and leptonic decays (right plot) of $\tau$ leptons. These distributions depend only on the decay type and initial momentum of the $\tau$ lepton.
• Figure 3: Example of the $M_{\tau \tau}$ distribution filled for all grid points in one of the $H\to\tau\tau$ events for each of the three decay modes. An entry for each point is weighted by its $\cal L$. Plot on the left shows $M_{\tau \tau}$ for the case of the ideal detector resolution and plot on the right demonstrates $M_{\tau \tau}$ for the same three events in the case of the realistic detector resolution. All distributions are normalized to a unit area.
• Figure 4: Left plot demonstrates the reconstructed $M_{\tau\tau}$ in $H\to\tau\tau$ events with $M_H=$115 GeV/c$^2$ for each of the three decay categories: both $\tau$'s decay hadronically (solid line); one $\tau$ decays leptonically and the other one hadronically (dashed line); and both $\tau$'s decay leptonically (dashed-dotted line). Right plot shows the reconstructed mass in $Z/\gamma^{*}\to\tau\tau$ and $H\to\tau\tau$ events with $M_H=$115 and 130 GeV/c$^{2}$ in the fully hadronic decay mode. Results are obtained in the assumption of the ideal detector resolution. Each distribution is normalized to a unit area.
• Figure 5: Left plot illustrates the reconstructed $M_{\tau\tau}$ mass in $Z\to\tau\tau$ (solid line) and $H\to\tau\tau$ events with $M_H=$115 (dashed line) and 130 GeV/c$^{2}$ (dashed-dotted line) in the fully hadronic decay mode. Right plot demonstrates a comparison of the reconstructed mass in $H\to\tau\tau$ events with $M_H=$115 GeV/c$^{2}$ when both $\tau$'s decay hadronically (solid line) and when one $\tau$ decays leptonically and the other one hadronically (dashed line). All results are obtained by using the MMC technique in events simulated with the realistic detector resolution. Each distribution is normalized to a unit area.