Abstract:
The jet trimming procedure has been demonstrated to greatly improve event reconstruction in hadron collisions by mitigating contamination due initial state radiation, multiple interactions, and event pileup. Meanwhile, Qjets—a nondeterministic approach to tree-based jet substructure—has been shown to be a powerful technique in decreasing random statistical fluctuations, yielding significant effective luminosity improvements. This manifests through an improvement in the significance S/δ B, relative to conventional methods. Qjets also provides novel observables in many cases, like mass-volatility, that could be used to further discriminate between signal and background events. The statistical robustness and volatility observables, for tagging, are obtained simultaneously. We explore here a combination of the two techniques, and demonstrate that significant enhancements in discovery potentials may be obtained in nontrivial ways. We will illustrate this by considering a diboson resonance analysis as a case study, enabling us to interpolate between scenarios where the gains are purely due to statistical robustness and scenarios where the gains are also reinforced by volatility variable discriminants. The former, for instance, is applicable to digluon/diquark resonances, while the latter will be of relevance to di − W ± / di − Z 0 resonances, where the boosted vector bosons are decaying hadronically and have an intrinsic mass scale attached to them. We argue that one can enhance signal significance and discovery potentials markedly through stochastic grooming, and help augment studies at the Large Hadron Collider and future hadron colliders.