Unraveling Dirac Magnetic Monopoles with Muon Beams at TeV Energies Using Machine Learning

Abstract

The focus of this paper is the production of magnetic monopoles Drell-Yan and the Photon-Fusion mechanisms to generate velocity-dependent scalar, fermionic, and vector monopoles of spin angular momentum $0,\frac{1}{2},1$ respectively at a future muon collider. A computational study compares the monopole pair-production cross-sections for both methods at various center-of-mass energies ($\sqrt{s}$) with different magnetic dipole moments. The comparison of kinematic distributions of monopoles at the generator and reconstructed level is demonstrated for both DY and PF mechanisms. We demonstrate the observability of magnetic monopoles against the most relevant Standard Model background using multivariate analysis techniques. Specifically, we apply three different classifiers based on neural networks, e.g., Boosted Decision Trees, Multilayer Perceptrons, and Likelihood methods, to evaluate their effectiveness. Our results highlight the efficiency and robustness of these approaches in distinguishing magnetic monopole signals from background noise.