Probing dark matter through charged Higgs pair production at future multi-TeV muon colliders: A machine-learning analysis

Abstract

Probing dark matter (DM) via charged Higgs pair production at future multi-TeV muon colliders is investigated within the Inert Doublet Model (IDM). The viable parameter space of the IDM is first updated by incorporating theoretical constraints and current experimental data. Based on the allowed parameter space, we evaluate DM relic density and compare the results with the latest constraints from direct DM detection experiments. The resulting parameter points consistent with all DM constraints are subsequently employed to study charged Higgs pair production at future multi-TeV muon colliders, including the subsequent decays of the charged Higgs bosons into Standard Model (SM) particles in association with DM candidate. In particular, we study the following production processes: $μ^- μ^+ \to ν_μ \barν_μ H^{\pm} H^{\mp} \to \ell^+ \ell^- + ν_μ \barν_μ ν_{\ell} \barν_{\ell} HH$, $μ^- μ^+ \to ν_μ \barν_μ H^{\pm} H^{\mp} \to \ell^\pm + 2\,\text{jets} + ν_μ \barν_μ ν_{\ell} HH$ and $μ^- μ^+ \to ν_μ \barν_μ H^{\pm} H^{\mp} \to 4\,\text{jets} + ν_μ \barν_μ HH$ for $\ell =e, μ$. The signal significance is evaluated against the corresponding SM backgrounds using both cut-based and machine-learning (ML) approaches. We find that ML framework substantially enhances the sensitivity to the signal processes compared with the conventional cut-based analysis. Furthermore, our results indicate that the DM signals through charged Higgs pair production can be indirectly probed with a statistical significance exceeding $5σ$ for several viable benchmark points at future multi-TeV muon colliders.