Higgs look-alikes at the LHC

TL;DR

The paper assesses how well a Standard Model Higgs can be distinguished from Higgs look-alikes at the LHC using the golden channel H -> ZZ* -> 4l. It develops a general formalism for Higgs couplings to ZZ, derives the full set of angular distributions for various J^PC hypotheses, and employs detector-acceptance modeling and sPlot background subtraction to perform likelihood-based discrimination near discovery. The results quantify discrimination power across masses (145–350 GeV) and show that, with tens of events, one can separate SM Higgs from many look-alikes, including CP and compositeness scenarios, while also outlining the limitations and the need for incorporating production information in a complete analysis. The work provides a rigorous framework for exploiting multi-angle decay information to characterize a newly discovered boson at the LHC.

Abstract

The discovery of a Higgs particle is possible in a variety of search channels at the LHC. However the true identity of any putative Higgs boson will at first remain ambiguous, until one has experimentally excluded other possible assignments of quantum numbers and couplings. We quantify to what degree one can discriminate a Standard Model Higgs boson from "look-alikes" at, or close to, the moment of discovery at the LHC. We focus on the fully-reconstructible "golden" decay mode to a pair of Z bosons and a four-lepton final state, simulating sPlot-weighted samples of signal and background events. Considering both on-shell and off-shell Z's, we show how to utilize the full decay information from the events, including the distributions and correlations of the five relevant angular variables. We demonstrate how the finite phase space acceptance of any LHC detector sculpts the decay distributions, a feature neglected in previous studies. We use likelihood ratios to discriminate a Standard Model Higgs from look-alikes with other spins or nonstandard parity, CP, or form factors. For a benchmark resonance mass of 200 GeV/c^2, we achieve a median expected discrimination significance of 3 sigma with as few as 19 events, and even better discrimination for the off-shell decays of a 145 GeV/c^2 resonance.

Figures (55)

• Figure 1: The Cabibbo-Maksymowicz angles CMvariables in the $H \to ZZ$ decays.
• Figure 2: 2D $p_T$-$\eta$pdf of a ${0^+}$ HLL resonance (left) and a $1^-$ one (right) for $\sqrt{s}$$=$$10$ TeV collisions, obtained using PYTHIA and the CTEQL5 parton density functions and for $m_H$$=$145, 200, 350 GeV/c$^2$ (top, middle and bottom).
• Figure 3: Distribution of the dimuon invariant mass for a sample of signal $H \to ZZ$ events, generated using our very-fast muon simulation. The parameters of the superimposed fit are extracted from candle.
• Figure 4: Reconstruction resolution for the angular variables of $\vec{X}$ shown here for a resonance with mass 145 GeV/$c^{2}$. The cos$\,\theta_2$ and cos$\,\theta_1$ distributions are very similar in this case. Only events surviving the signal selection are included. All distributions are normalized to unit integral.
• Figure 5: Distributions showing systematic biases for a subset of the reconstructed variables $\vec{X}$ for a resonance with mass 145 GeV/$c^{2}$. Only events that survive the signal selection are included. All biases are negligible.