On the Phenomenology of Hidden Valleys with Heavy Flavor

TL;DR

This paper explores Hidden Valley models with heavy flavor, highlighting challenges in identifying signals due to complex decay patterns and proposing alternative detection strategies involving vertex counting, muons, and event-shape variables.

Contribution

It introduces a new Monte Carlo simulation for heavy-flavor Hidden Valley models and suggests novel analysis methods beyond jet counting for signal detection.

Findings

Jet-parton correspondence is significantly broken in these models.

Counting vertices and displaced tracks can improve signal isolation.

Single-jet invariant mass is a promising observable for detection.

Abstract

A preliminary investigation of a large class of Hidden Valley models is presented. These models are more challenging than those considered in arXiv:0712.2041 [hep-ph]; although they produce a new light resonance which decays to heavy standard model fermions, they exhibit no light dilepton resonance. A heavy $Z'$ decaying to v-hadrons, which in turn decay mainly to bottom quarks and tau leptons, is considered; six case studies are investigated, using a new Monte Carlo simulation package. It is found that the one-to-one correspondence of jets and partons is badly broken, and the high-multiplicity heavy-flavor signal probably cannot be isolated by counting jets, with or without heavy-flavor tags. Instead, other measures, such as counting and correlating vertices or displaced tracks, and possibly counting of (non-isolated) muons and use of event-shape variables, should be combined with scalar transverse energy and/or missing transverse energy to reduce backgrounds. Within the resulting sample, searches for the v-pion mass resonance in both di-jet and single-jet invariant mass can help confirm a signal. The best observable in a perfect calorimeter seems to be single-jet invariant mass for jets of larger radius ($R$=0.7), although this needs further study in a realistic setting. A more detailed signal-to-background study is needed as a next step, but will face the difficulty of estimating the various high-multiplicity backgrounds.