Measurements of WW and WZ production in W + jets final states in p-pbar collisions

TL;DR

This study measures WW and WZ production in WV final states using 4.3 fb^-1 of p-pbar collisions at 1.96 TeV collected by the D0 detector. A Random Forest classifier is employed to separate diboson signals from W+jets backgrounds, with b-tagging used to disentangle WZ (Z→bb) from WW. The WV cross section is extracted via likelihood fits that incorporate numerous systematic uncertainties as Gaussian priors, yielding σ(WV)=19.6 pb with high significance, and a WW/WZ breakdown of σ(WW)=15.9 pb and σ(WZ)=3.3 pb in agreement with SM predictions. The analysis demonstrates a robust approach for detecting hadronically decaying diboson final states and provides essential cross-checks relevant to low-mass Higgs searches.

Abstract

We study WW and WZ production with $lνq{q}$ ($l=e,μ$) final states using data collected by the D0 detector at the Fermilab Tevatron Collider corresponding to 4.3 fb^-1 of integrated luminosity from p-pbar collisions at sqrt{s}=1.96 TeV. Assuming the ratio between the production cross sections $σ(WW)$ and $σ(WZ)$ as predicted by the standard model, we measure the total WV (V=W,Z) cross section to be $σ(WV)= 19.6^{+3.2}_{-3.0}$ pb, and reject the background-only hypothesis at a level of 7.9 standard deviations. We also use b-jet discrimination to separate the WZ component from the dominant WW component. Simultaneously fitting WW and WZ contributions, we measure $σ(WW) = 15.9^{+3.7}_{-3.2}$ pb and $σ(WZ) = 3.3^{+4.1}_{-3.3}$ pb, which is consistent with the standard model predictions.

Figures (8)

• Figure 1: (color online) A comparison of the measured $WV$ signal (filled histogram) to background-subtracted data (points) in the RF output distribution (summed over electron and muon channels, and 0, 1, and 2-tag sub-channels), after the combined fit to data using the RF output distributions. Also shown is the $\pm$1 standard deviation uncertainty on the background prediction. The $\chi^{2}$ fit probability, P$(\chi^{2})$, is based on the residuals using data and MC statistical uncertainties.
• Figure 2: (color online) Results from the simultaneous fit of $\sigma(WW)$ and $\sigma(WZ)$ using the RF output distributions. The plot shows the best fit value with 68% and 95% confidence level (CL) regions and the NLO SM prediction.
• Figure 3: (color online) A comparison of the signal+background prediction to data in the RF output distribution (summed over electron and muon channels) for 0, 1, and 2-tag sub-channels after the combined fit to data using the RF output distribution (LP denotes light partons such as $u$, $d$, $s$ or gluon, and HF denotes heavy-flavor such as $c\bar{c}$ or $b\bar{b}$). The systematic uncertainty band is evaluated after the fit of the total $WV$ cross section in the RF output distribution.
• Figure 4: (color online) A comparison of the measured $WW$ and $WZ$ signals (filled histograms) to background-subtracted data (points) in the dijet mass distribution (summed over electron and muon channels) for 0, 1, and 2-tag sub-channels after the combined fit to data using the dijet mass distribution. Also shown is the $\pm$1 standard deviation uncertainty on the background prediction. The $\chi^{2}$ fit probability, P$(\chi^{2})$, is based on the residuals using data and MC statistical uncertainties.
• Figure 5: (color online) Distributions of the $b$-jet identification variables used as inputs to the RF classifier (first two of fifteen) for electron and muon channels combined, with logarithmic ((a) and (b)) and linear ((c) and (d)) scales. To better show the $WW$ and $WZ$ signals the lowest bin is cut off in the distributions with a linear scale. The signal and background predictions and the systematic uncertainty band are evaluated after the fit of the total $WV$ cross section in the RF output distribution. Definitions for each variable are provided in the text (LP denotes light partons such as $u$, $d$, $s$ or gluon, and HF denotes heavy-flavor such as $c\bar{c}$ or $b\bar{b}$).