Pure Samples of Quark and Gluon Jets at the LHC

TL;DR

<3-5 sentence high-level summary>This paper investigates how to obtain relatively pure samples of quark or gluon jets at the LHC using simple, LO-driven event samples and kinematic cuts. It systematically evaluates multiple final states (e.g., γ+jets, W/Z+jets, 3jets, b+2jets) to maximize flavor purity while preserving usable cross sections, employing both multivariate analyses and single-variable discriminants. The authors find that γ+2jets with appropriate cuts can yield ~95–99% quark purity at cross sections from a few pb down to the sub-pb level for jet pT around 200 GeV, while gluon purification is more challenging; the best gluon samples come from 3-jet or b-tagged events, though strong b-tagging is required for very high purities. They also discuss a rigorous theoretical framework for defining quark and gluon jets beyond leading order, reinforcing the utility of these pure samples for testing QCD predictions and calibrating simulations in jet substructure studies.

Abstract

Having pure samples of quark and gluon jets would greatly facilitate the study of jet properties and substructure, with many potential standard model and new physics applications. To this end, we consider multijet and jets+X samples, to determine the purity that can be achieved by simple kinematic cuts leaving reasonable production cross sections. We find, for example, that at the 7 TeV LHC, the pp {\to} γ+2jets sample can provide 98% pure quark jets with 200 GeV of transverse momentum and a cross section of 5 pb. To get 10 pb of 200 GeV jets with 90% gluon purity, the pp {\to} 3jets sample can be used. b+2jets is also useful for gluons, but only if the b-tagging is very efficient.

Figures (16)

• Figure 1: Leading order cross sections, including kinematic cuts and branching ratios for $Z/W$ decay to include an electron or muon. The $x$-axis indicates the $p_T$ cut applied to all light quarks and gluons, but not $b$-quarks. The constraint on the $p_T$ for $b$'s, photons, and charged leptons or neutrinos from $Z/W$ (though not the $Z/W$ itself) is fixed at 20 GeV. Note that the 3-jet cross section falls below $b$+2jets due to the harder cuts on the non-$b$ jets. The $t \bar{t}$ cross section refers to the semi-leptonic sample, and, in contrast to all the other samples, the $p_T$ cut is applied to only one of the two light-quark jets. Since its cross section is so low, it will not be considered further.
• Figure 2: Fraction of $X$+1jet events where the jet is $uds$ quark (bottom and blue in each plot) as compared to gluon (top and red). The horizontal axis is a $p_T$ cut on the jet, which in these events translates into an identical $p_T$ cut on the other object.
• Figure 3: Fraction of $X$+2jet events where the jets are both light quark 'QQ' (bottom blue) vs one light quark one gluon 'QG' (middle purple) vs both gluon 'GG' (top red). Notice $\gamma+GG$ almost never happens, nor does $b+QQ$. These are starting points for quark and gluon purification. The horizontal axis is a $p_T$ cut on all jets, while the other objects ($b$, $\gamma$, and leptons from $Z/W$) have $p_T>20$ GeV.
• Figure 4: Division of the multijet (dominantly QCD) sample. The horizontal axis is a $p_T$ cut on all jets. Notice that all three jets are almost never all quark, and in the 4-jet sample, there are almost always at least two gluons. The 3-jet sample will be a staring point for gluon purification.
• Figure 5: The chance that a given jet is a light quark jet rather than a gluon jet. (This ratio does not include bottom or charm.) The $W$ and $Z$ were nearly identical and combined on this plot, but they are slightly different from the photon, mostly due to the $\gamma$ and lepton cuts.