Probing the dimuon channel of a Z' boson at the HL-LHC using multivariate analysis

TL;DR

This paper explores the potential of multivariate analysis techniques to detect a Z' boson decaying into dimuons at the HL-LHC, aiming to enhance sensitivity to new physics beyond the Standard Model.

Contribution

It applies and compares advanced multivariate classifiers like BDT, DNN, and likelihood estimators for Z' boson detection in the dimuon channel at the HL-LHC.

Findings

BDT, DNN, and likelihood estimators improve signal-background discrimination.

Deep Neural Networks show promising performance in identifying Z' signals.

The analysis demonstrates the potential for enhanced sensitivity at the HL-LHC.

Abstract

The upcoming upgrade of the existing LHC facility at CERN is known as the High-Luminosity Large Hadron Collider (HL-LHC). It is designed to extend its physics reach by substantially increasing the integrated luminosity. It will enable more precise measurements of the Standard Model (SM) and improve sensitivity to rare events and possible new physics signatures. This study adopts a multivariate analysis (MVA) approach to effectively discriminate the dark Higgs (DH) signal against the dominant SM background. The analysis targets the leptonic decay mode of the $Z'$ boson, focusing on the dimuon final state at $\sqrt{s} = 14,\text{TeV}$ and $3000,\text{fb}^{-1}$ integrated luminosity corresponding to the HL-LHC. The DH signal is examined using the Toolkit for Multivariate Analysis (TMVA), employing and comparing the performance of three classifiers: Boosted Decision Trees (BDT), Deep Neural Networks (DNN), and Likelihood estimators.