• ATLAS
• Electroweak
• High energy physics
• LHC
• W
• Z

In this thesis, the measurement of the associated production of WW/WZ pairs in proton-proton collisions at √s=8 TeV with the ATLAS detector is presented. The boson pairs are searched through their semileptonic decay where a W boson decays leptonically, in electron-neutrino or muon-neutrino, and one boson decays hadronically. The hadronic mass resolution does not allow to separate the W and Z boson mass peaks, therefore the final state consists of pairs of WW and WZ bosons.
Measuring the cross section of this process provides a powerful probe of the Electroweak sector of Standard Model and an estimation of the irreducible backgrounds in important Higgs boson analysis channels. In addition, precision measurements of WW/WZ kinematics can be used to search for new physics through the test of the coupling between gauge bosons (anomalous Triple Gauge Couplings or aTGC).
The semileptonic WW/WZ decay channel is particularly challenging because of the low signal to background ratio (at the level of percent). The production of a W associated with jets and of top-antitop are the processes that give the major contributions to the background.
A specific event selection has been designed, to optimize the signal over background ratio. It requires one well reconstructed electron or muon of high transverse momentum and large missing transverse energy, with the aim of selecting events containing W bosons decaying leptonically.
The hadronic decay of the second boson is then reconstructed as a system of jets. Two analysis selections have been defined, depending on the W/Z hadronic decay topology.
- The ‘resolved’ channel, in which the W/Z hadronic decay is reconstructed as a pair of well separated jets. This is the more conventional event signature, also adopted in analogous analysis at √s=7 TeV.
- The ‘boosted’ channel, in which the full decay is reconstructed as a single jet of large radius. This method is meant to detect W/Z bosons with large Lorentz boost, increasing the analysis efficiency at high transverse momentum.
In the resolved channel, two jets of transverse momentum larger than 25 GeV are required, while boosted events must contain a single large jet of at least 200 GeV transverse momentum. Additional kinematic cuts are then applied to further reduce background contribution.
After applying the complete event selection, the signal to background ratio is about 5-10%. To provide a measurement of the number of signal events, a method based on Monte Carlo simulation has been used. The defined event selection has been applied to Monte Carlo simulated samples, providing a model for each background and signal contribution after event selection.
The resulting samples have been used as input to a binned Maximum Likelihood fit to data, which provides a measurement of the number of signal events. The adopted fit variable is the invariant mass of the jet-jet system in the case of the resolved channel and the mass of the single large jet in the case of the boosted.
Prior to performing the fit, the agreement between data and simulation is tested in dedicated control regions. The agreement is found to be not always optimal, hence data-driven correction factors for the two main sources of background (W+jets and top production) are derived.
The number of signal events obtained from the fit is then divided by the integrated luminosity to provide a measurement of the WW/WZ production cross section.
In this way, the cross section is measured within the phase space region accessed by the applied event selection, called ‘fiducial phase space’. The fiducial phase space is defined using truth-level Monte Carlo objects, by applying a truth-level selection as close as possible to the one applied to select reconstructed events.
The fiducial phase space definition and the actual event selection are however not identical, because of e.g. particle reconstruction inefficiencies. The measured number of signal events is then corrected to account for this effect.
Systematic uncertainties arise both from uncertainties in the reconstruction of physical objects and from Monte Carlo simulation. They are accounted for in the fit, the maximum impact on the cross section measurement is due to uncertainties associated to the choice of Monte Carlo generator.
The measured cross section agrees with what expected from Standard Model in both resolved and boosted channel. In the resolved case a clear signal peak is visible and the signal is observed with a significance of 4 standard deviations. In the boosted channel, on the other hand, the available statistics is much smaller and the significance amounts to less than 2 standard deviations.
This measurement method has been cross-checked in the resolved channel, by fitting the background with an empirical function on signal sideband rather than using a Monte Carlo-based Maximum Likelihood fit. This procedure is however complicated by the need of subtracting the contribution from top background, which contribute to the observed mess peak when a real WW pair is produced by top-antitop decay. The resulting cross section is in agreement both with Standard Model expectations and with the Maximum Likelihood measurement, confirming the goodness of the measurement.
Finally, the selected data sample has been used to search for anomalous Triple Gauge Couplings. The WW/WZ production rate is sensitive to the coupling between three gauge bosons. This coupling might deviate from its Standard Model value because of new particles of high mass contributing through loop corrections. The expected effect of such a contribution would be an increase of the event rate at high energy.
A fit to the transverse momentum of the jet-jet system (single wider jet) is used to set limits on this new physics contributions in the resolved (boosted) channel. The boosted channel provides a better sensitivity to anomalous coupling than the resolved, since it accesses a region of higher transverse momentum. No evidence of new physics is observed. However, limits from the boosted channel reach the same order of precision as currently best published aTGC limits.