Machine Learning in the Hunt for Heavy Charged Higgs Bosons at Gamma-Gamma Colliders in the Type III Two Higgs Doublet Model

Abstract

We conduct a detailed exploration of charged Higgs boson masses $M_{H^{\pm}}$ within the range of $100-190~GeV$. This investigation is grounded in the benchmark points that comply with experimental constraints, allowing us to systematically account for uncertainties inherent in the analysis. Our results indicate significant production prospects for the process $H^{+}H^{-} \rightarrow τν_τ τν_τ$, which could provide essential insights into the properties of $H^{\pm}$ bosons. By examining these decay channels, we aim to illuminate the interplay between the charged Higgs boson and the established Standard Model. The research uses machine learning methods like Boosted Decision Trees (BDT) and Multilayer Perceptrons (MLP), as well as Likelihood and LikelihoodD, to improve the identification of heavy charged Higgs bosons compared to Standard Model backgrounds at a 3.0 TeV $γγ$ collider with an integrated luminosity of $\mathcal{L}_{int}=3000~fb^{-1}$.