Combining subjet algorithms to enhance ZH detection at the LHC

TL;DR

This paper explores combining different subjet algorithms—filtering, pruning, and trimming—to improve the detection of heavy resonances like the Higgs boson at the LHC, demonstrating enhanced statistical power in signal identification.

Contribution

It introduces a method to combine multiple subjet algorithms using a likelihood approach, showing improved detection capabilities for boosted heavy particles.

Findings

Combining methods increases statistical power for signal detection.

The approach improves identification of boosted top quarks.

Enhanced discrimination of Higgs-Z signals from background.

Abstract

The signal for a highly boosted heavy resonance competing against a background of light parton jets at the LHC can be enhanced by analyzing subjets in the "fat" jet that possibly contains the heavy resonance. Three methods for doing this are known as filtering, pruning, and trimming. We study the possibility of combining these methods using a relative likelihood approach. We find that, because the methods are not the same, one achieves an enhanced statistical power by combining them. We illustrate the possibilities first with a simple problem of combining trimming and pruning to enhance the signal for finding a boosted top quark. We then study the more difficult problem of disentangling from the background the signal for the production of a Higgs boson in association with a Z-boson. For this problem, we combine filtering, trimming, and pruning.