Unsupervised hadronic SUEP at the LHC

Confining dark sectors with pseudo-conformal dynamics produce SUEP, or Soft Unclustered Energy Patterns, at colliders: isotropic hadrons soft and democratic energies. We target the experimental nightmare scenario, SUEPs in exotic Higgs decays, where all decay promptly to SM hadrons. First, we identify three promising observables, charged particle multiplicity, event ring isotropy, matrix of geometric distances between tracks. Their patterns can be exploited through a cut-and-count search, supervised machine learning, an unsupervised autoencoder. find that HL-LHC will probe branching ratios per-cent level, even without detailed knowledge signal features. Our techniques applied other SUEP searches, especially strategy, which is independent overly specific model assumptions corresponding precision simulations.