Subjet Multiplicity of Gluon and Quark Jets Reconstructed with the $k_{\perp}$ Algorithm in $p\bar{p}$ Collisions

TL;DR

This paper investigates jet substructure in hadron collisions using the kT jet algorithm, comparing to cone jets to validate momentum calibration and jet properties. By exploiting gluon- and quark-enriched samples at two center-of-mass energies, the authors extract the mean subjet multiplicities for gluon and quark jets and form the ratio r = (⟨M_g⟩−1)/(⟨M_q⟩−1). The results show gluon jets have more subjets than quark jets, in line with HERWIG predictions and QCD resummation, and significantly larger than naive color-factor expectations. This work demonstrates the viability of using jet substructure as a discriminator between gluon and quark jets at hadron colliders and validates the kT algorithm for detailed jet studies in complex environments.

Abstract

The D0 Collaboration has studied for the first time the properties of hadron-collider jets reconstructed with a successive-combination algorithm based on relative transverse momenta ($k_{\perp}$) of energy clusters. Using the standard value D = 1.0 of the jet-separation parameter in the $k_{\perp}$ algorithm, we find that the $p_T$ of such jets is higher than the $E_T$ of matched jets reconstructed with cones of radius R = 0.7, by about 5 (8) GeV at $p_T \approx 90$ (240) GeV. To examine internal jet structure, the $k_{\perp}$ algorithm is applied within D = 0.5 jets to resolve any subjets. The multiplicity of subjets in jet samples at $\sqrt{s} = 1800$ GeV and 630 GeV is extracted separately for gluons ($M_g$) and quarks ($M_q$), and the ratio of average subjet multiplicities in gluon and quark jets is measured as $(M_{g} - 1) / (M_{q} - 1) = 1.84 \pm 0.15 (stat.) \pm ^{0.22}_{0.18} (sys.)$. This ratio is in agreement with the expectations from the HERWIG Monte Carlo event generator and a resummation calculation, and with observations in $e^+e^-$ annihilations, and is close to the naive prediction for the ratio of color charges of $C_A/C_F = 9/4 = 2.25$.

Figures (31)

• Figure 1: One quadrant of the DØ calorimeter and drift chambers, projected in the $r - z$ plane. Radial lines illustrate the detector pseudorapidity and the pseudo-projective geometry of the calorimeter towers. Each tower has size $\Delta \eta \times \Delta \phi = 0.1 \times 0.1$.
• Figure 2: Illustration of the pedestal energy distribution in a calorimeter cell (solid line), stemming from uranium noise. The mean value is defined to be zero, and the peak occurs at negative values. Removal of the portion between the vertical dashed lines (a symmetric window about the mean) yields a positive mean for the remaining distribution.
• Figure 3: Schematic of signal voltage in a calorimeter cell as a function of time. The solid line represents the contribution for a given event (the "current" $p\bar{p}$ bunch crossing). In the absence of previous bunch crossings, the cell is sampled correctly at $t_b$, just before a crossing, to establish a base voltage, and at $t_p$, to establish a peak voltage. The voltage difference $\Delta V = V(t_p) - V(t_b)$ is proportional to the initial energy deposited in the cell. The dashed lines show example contributions from a previous bunch crossing containing three different numbers of $p\bar{p}$ interactions. The observed signal is the sum of the signals from the current and previous crossings. (The figure is not to scale.)
• Figure 4: A simplified example of the final state of a collision between two hadrons. (a) The particles in the event (represented by arrows) comprise a list of objects. (b-f) Solid arrows represent the final jets reconstructed by the $k_{\perp}$ algorithm, and open arrows represent objects not yet assigned to jets. The five diagrams show successive iterations of the algorithm. In each diagram, a jet is either defined (when it is well-separated from all other objects), or two objects are merged (when they have small relative $k_{\perp}$). The asterisk labels the relevant object(s) at each step.
• Figure 5: Mean energies in calorimeter cells for a sample of minimum-bias events. The contribution from instrumental effects is included, which occasionally leads to negative energy readings. For each cell, the energy distribution illustrated in Fig. \ref{['fig:u_decay']} is fitted to a Gaussian. Before readout, the zero-suppression circuit in each cell's electronics sets to zero energy the channels in a symmetric window about the mean pedestal. These channels are not read out, causing the dip observed near zero.