Discrimination of $H\rightarrow Zγ\rightarrow \ell^{+}\ell^{-}γ$ from $Z/γ^*$ Processes Using Kinematically Correlated Observables

TL;DR

This work addresses the challenge of distinguishing the rare Higgs decay $H \rightarrow Z\gamma \rightarrow \ell^{+}\ell^{-}\gamma$ from the dominant $Z/\gamma^{*}$ background at the LHC by exploiting physics-motivated correlations in the 2D plane spanned by $P_{\mathrm{Higgs}}$ and $\theta_{Z\gamma}$. It combines a baseline set of kinematic observables with newly introduced correlated variables, notably $\log(\theta_{Z\gamma} \times P_{\mathrm{Higgs}})$, in an XGBoost classifier, achieving measurable improvements in AUC (about +0.011) and substantial background rejection in the $P_{\mathrm{H}}$-\theta plane. A momentum-dependent background rejection strategy, operating via asymmetric $n\sigma$ bands fitted to the background trend, yields a meaningful enhancement of signal purity near the Higgs mass (up to +3.4% in the muon channel) while maintaining high signal efficiency. The approach demonstrates robust performance against over-training and offers a flexible framework that can be extended to rare Higgs decays and beyond the Standard Model searches in resonant analyses.

Abstract

At LHC energies, the Drell--Yan ($Z/γ^{*}$) processes have a substantially large cross section. Their di-lepton ($\ell^+\ell^-$) final state contributes significantly to many resonant signal regions, making them one of the dominant backgrounds in numerous physics analyses. The study focuses on improving the discrimination and suppression of the $Z/γ^{*} \rightarrow \ell^{+}\ell^{-}$ background from the $H \rightarrow Zγ\rightarrow \ell^{+}\ell^{-}γ$ signal at $\sqrt{s}=13~\text{TeV}$ by leveraging Monte Carlo simulated data. The analysis introduces physics-motivated correlated observables derived from the two-dimensional $(P_{\mathrm{Higgs}}, θ_{Zγ})$ plane. These observables encode differences in angular and momentum information to enhance signal--background separation while maintaining high signal efficiency. We present a multivariate analysis (MVA) employing a Boosted Decision Tree (XGBoost) classifier. By incorporating additional physics-motivated correlated observables, the classifier achieves measurable improvements in performance. A significant increase in the area under the ROC curve (AUC) is observed in both the electron and muon channels, demonstrating the effectiveness of the expanded feature set. Further, optimised background rejection using $(P_{\mathrm{Higgs}}, θ_{Zγ})$ plane increases the signal-to-background ratio to 2.1\% and 3.4\% for the electron and muon channel respectively near the Higgs mass. This work demonstrates that combining kinematic correlations with interpretable multivariate techniques leads to improved sensitivity and robust background rejection. The approach is flexible and can be readily applied to a wide range of analyses, including rare Higgs decays, resonant searches, and studies beyond the Standard Model.

Figures (11)

• Figure 1: Analytic two-body decay model showing the decrease of the opening angle $\theta_{\mathrm{open}}$ between decay products with increasing parent particle momentum $P_M$ due to Lorentz boost effects. The curves represent decay configurations relevant to $H\to Z\gamma$ and $Z/\gamma^\ast\to \ell^+\ell^-$ processes, considering only electron and muon final states.
• Figure 2: Kinematic distributions in the $(\theta_{Z\gamma}, P_{\mathrm{Higgs}})$ plane for the electron (left) and muon (right) channels, obtained from Monte Carlo samples. Signal events (blue) exhibit a characteristic correlated structure, while background events (red) populate a broader region without any kinematic constraints.
• Figure 3: Overlaid signal and background distributions in the case of electrons for all-purpose variables. The Kolmogorov--Smirnov statistic $D_{\mathrm{KS}}$ is calculated to quantify the shape differences between the signal and background distributions.
• Figure 4: Overlaid signal and background distributions in case of muons for the all-purpose variables. The Kolmogorov--Smirnov statistic $D_{\mathrm{KS}}$ is calculated to quantify the shape difference between the signal and background distributions.
• Figure 5: ROC curves (left) and the corresponding signal and background efficiencies (right) for the electron (top) and muon (bottom) channels to evaluate the performance of the BDT classifiers.