Search for the production of $ZW$ and $ZZ$ boson pairs decaying into charged leptons and jets in $p\overline{p}$ collisions at $\sqrt{s}\mathbf{=}1.96\mathrm{TeV}$

We present a measurement of the production cross section for $ZW$ and $ZZ$ boson pairs in final states with a pair of charged leptons, from the decay of a $Z$ boson, and at least two jets, from the decay of a $W$ or $Z$ boson, using the full sample of proton-antiproton collisions recorded with the CDF II detector at the Tevatron, corresponding to $8.9{\mathrm{fb}}^{-1}$ of integrated luminosity. We increase the sensitivity to vector-boson decays into pairs of quarks using a neural-network discriminant that exploits the differences between the spatial spread of energy depositions and charged-particle momenta contained within the jet of particles originating from quarks and gluons. Additionally, we employ new jet energy corrections to Monte Carlo simulations that account for differences in the observed energy scales for quark and gluon jets. The number of signal events is extracted through a simultaneous fit to the dijet mass spectrum in three classes of events: events likely to contain jets with a heavy-quark decay, events likely to contain jets originating from light quarks, and events that fail these identification criteria. We determine the production cross section to be ${\sigma }_{ZW+ZZ}={2.5}_{-1.0}^{+2.0}\mathrm{pb}$ ($<6.1\mathrm{pb}$ at the 95% confidence level), consistent with the standard model prediction of 5.1 pb.

Figure 1

Balancing distributions, (a) $K_Z$ and (b) $K_{\gamma }$, in data and MC simulation as a function of $E_T{}^{\mathrm{jet}}$. The uncertainties include solely the contribution from the fluctuations in the mean of a Gaussian fit to the balancing distributions in bins of $E_T{}^{\mathrm{jet}}$.

Figure 2

Derived balancing variable for (a) quark jets, $K_q$, and (b) gluon jets, $K_g$, in data and MC simulation as a function of $E_T{}^{\mathrm{jet}}$. The uncertainties on each point are from the uncertainties from the mean of the Gaussian fit and the uncertainties on the quark fractions, added in quadrature.

Figure 3

Derived correction for simulated quark jets and gluon jets as a function of $E_T{}^{\mathrm{jet}}$. The open triangles represent corrections derived using both $\gamma$-jet and $Z$-jet balancing samples, while the filled triangles represent the assumed uniform correction for quarks and the corresponding correction for gluons calculated from the $Z$-jet balancing sample alone. The uncertainties shown are statistical only. The short dashed lines are the fits of the correction to a constant across jet $E_T$ (as opposed to the corrections in each bin of $E_T{}^{\mathrm{jet}}$), and the long dashed lines represent the total systematic uncertainty bands on that constant correction, further described in Sec. 4c.

Figure 4

Typical distributions of energy (momentum) content of jets as a function of the $\Delta R$ (a) between pairs of towers and (b) between pairs of tracks in light-flavor quark and gluon jets in simulation.

Figure 5

Distributions of the outputs of the NNs processing (a) tower information and (b) track information in light-flavor quark and gluon jets in simulation. Higher NN scores indicate jets that are more quark-like.

Figure 6

Distribution of the output of the final NN for light-flavor quark and gluon jets in simulation. Higher NN scores indicate jets that are more quark-like.

Figure 7

Distribution of (a) the maximum and (b) minimum jet QG values of the two jets in the $W+2$ jet sample.

Figure 8

Distribution of (a) the maximum and (b) minimum jet QG values of the two jets in the $t\overline{t}$ sample. The distinction between $q$ and $b$ jets refers to the lower two $b\mathrm{ness}$ jets: events where both jets are matched to non-$b$ quark jets are labeled $q$ jets, while if one of the jets is matched to a $b$ jet, the event is labeled $b$ jets.

Figure 9

Invariant dijet mass distributions with fit results overlaid for the $ZW+ZZ$ process in the $\mathrm{\text{dilepton}}+\mathrm{\text{dijet}}$ selection in the heavy-flavor–tagged channel (left panels), light-flavor-tagged channel (center panels), and untagged channel (right panels). The top row shows the output from the fit compared to the data, while the bottom row shows the background subtracted from data, compared to the expected (dashed line) and fitted (solid line, with uncertainties in bands) signal contributions.

Figure 10

Confidence bands showing the expected range of observed cross sections as a function of the true cross section, with 68% C.L. (black dashed region) and 95% C.L. (solid gray region).