Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Quark and Gluon Tagging at the LHC

Jason Gallicchio, Matthew D. Schwartz

TL;DR

An exhaustive search of existing and novel jet substructure observables is found that a multivariate approach can filter out over 95% of the gluon jets while keeping more than half of the light-quark jets.

Abstract

Being able to distinguish light-quark jets from gluon jets on an event-by-event basis could significantly enhance the reach for many new physics searches at the Large Hadron Collider. Through an exhaustive search of existing and novel jet substructure observables, we find that a multivariate approach can filter out over 95% of the gluon jets while keeping more than half of the light-quark jets. Moreover, a combination of two simple variables, the charge track multiplicity and the $p_T$-weighted linear radial moment (girth), can achieve similar results. While this pair appears very promising, our study is only Monte Carlo based, and other discriminants may work better with real data in a realistic experimental environment. To that end, we explore many other observables constructed using different jet sizes and parameters, and highlight those that deserve further theoretical and experimental scrutiny. Additional information, including distributions of around 10,000 variables, can be found on this website http://jets.physics.harvard.edu/qvg .

Quark and Gluon Tagging at the LHC

TL;DR

An exhaustive search of existing and novel jet substructure observables is found that a multivariate approach can filter out over 95% of the gluon jets while keeping more than half of the light-quark jets.

Abstract

Being able to distinguish light-quark jets from gluon jets on an event-by-event basis could significantly enhance the reach for many new physics searches at the Large Hadron Collider. Through an exhaustive search of existing and novel jet substructure observables, we find that a multivariate approach can filter out over 95% of the gluon jets while keeping more than half of the light-quark jets. Moreover, a combination of two simple variables, the charge track multiplicity and the -weighted linear radial moment (girth), can achieve similar results. While this pair appears very promising, our study is only Monte Carlo based, and other discriminants may work better with real data in a realistic experimental environment. To that end, we explore many other observables constructed using different jet sizes and parameters, and highlight those that deserve further theoretical and experimental scrutiny. Additional information, including distributions of around 10,000 variables, can be found on this website http://jets.physics.harvard.edu/qvg .

Paper Structure

This paper contains 3 equations, 3 figures.

Figures (3)

  • Figure 1: Data on the integrated jet shape $\Psi(r)$ is usually published only when averaged over all events. Here we show the distribution of $\Psi(0.1)$, for quarks (blue, solid) and gluons (red, hollow). The event-by-event distributions of $\Psi(r)$ and other observables are much more important for gluon tagging than average values.
  • Figure 2: 2D Histograms of the two best observables, along with the likelihood formed by combining them bin-by-bin.
  • Figure 3: Gluon rejection curves for several observables as a function of Quark Jet Acceptance. The results for 200 GeV Jets are shown, but other samples give similar results. The best pair of observables is charged track multiplicity and linear radial moment (girth). The best group of five also includes jet mass for the hardest subjet of size R=0.2, the average $k_T$ of all R$_\mathrm{sub}$=0.1 subjets, and the 3rd such small subjet's $p_T$ fraction.