Running bumps from stealth bosons

Abstract

For the 'stealth bosons' S, light boosted particles with a decay S → A A → qqqq into four quarks and reconstructed as a single fat jet, the groomed jet mass has a strong correlation with groomed jet substructure variables. Consequently, the jet mass distribution is strongly affected by the jet substructure selection cuts when applied on the groomed jet. We illustrate this fact by recasting a CMS search for low-mass dijet resonances and show a few representative examples. The mass distributions exhibit narrow and wide bumps at several locations in the 100-300 GeV range, between the masses of the daughter particles A and the parent particle S, depending on the jet substructure selection. This striking observation introduces several caveats when interpreting and comparing experimental results, for the case of non-standard signatures. The possibility that a single boosted particle decaying hadronically produces multiple bumps, at quite different jet masses, and depending on the event selection, brings the anomaly chasing game to the next level.

The minimal stealth boson: models and benchmarks

“…(Several other grooming algorithms and parameters were explored in ref. [17] with similar results.) This is clearly observed in figure 5, where the maximum significance is for the stealth boson signal is near 2 TeV while the Z mass is 2.2 TeV.…”

“…Although the resonance is relatively narrow, the detector resolution effects yield a wider distribution, especially for the decays into scalars which produce four-pronged jets. As it has previously been shown [17,28], standard grooming algorithms are not adequate for multi-pronged jets, shifting jet masses and momenta from their original values. The effect can clearly be seen in the signal profile for the dijet analysis, which is much wider in model 2, where more than half of the dijet events are actually Z → H 3 H 4 .…”

Collider tests of the Renormalizable Coloron Model

“…The situation is dramatically different when M G M φ I > M Θ . The signal arising from the asymmetric decay of a coloron produced in the s-channel, pp → G → Θφ I , is then a "dijet" resonance, with one of the jets (J Θ ) having a 2-prong substructure, and the other jet (of mass M φ I and labelled by J jjΘ ) having a 4-prong substructure (similar boosted objects are discussed in [34,35]). In this case, the effective dijet branching fraction is closer to 100% (provided that each multi-prong jet fits inside a ∆R = 0.4 cone), so that the constraint from dijet resonance searches approaches the dotted line in Figure 7.…”

Extending the search for new resonances with machine learning

“…Both ATLAS and CMS 1 have conducted extensive searches for resonances decaying into jets originating from generic quarks and gluons [5][6][7], from boosted W [8,9], Z [10], Z 0 [11 -13] or Higgs bosons [14-16], from b-quarks [17], as well as from boosted top quarks [18,19]. There is some overlapping sensitivity in these searches, but in general the sensitivity is greatly diminished away from the target process (see e.g., [20][21][22] for examples). It is not feasible to perform a dedicated analysis for every possible topology and so some signals may be missed.…”