Search for $VH\rightarrow$ leptons + $b\bar{b}$ with the ATLAS experiment at the LHC

The search for a Higgs boson decaying to a $b\bar{b}$ pair is one of the key analyses ongoing at the ATLAS experiment. Despite being the largest branching ratio decay for a Standard Model Higgs boson, a large dataset is necessary to perform this analysis because of the very large backgrounds affecting the measurement. To discriminate the electroweak $H\rightarrow b\bar{b}$ signal from the large QCD backgrounds, the associated production of the Higgs with a $W$ or a $Z$ boson decaying leptonically is used. Different techniques have been proposed to enhance the signal over background ratio in the $VH(b\bar{b})$ channel, from dedicated kinematic cuts, to a single large radius jet to identify the two collimated $b$'s in the Higgs high transverse momentum regime, to multivariate techniques. The high-$p_{\mathrm{T}}$ approach, using a large radius jet to identify the $b$'s coming from the Higgs decay, has been tested against an analysis based on kinematic cuts for a dataset of $4.7$ fb$^{-1}$ luminosity at $\sqrt{s}=7$ TeV, and compatible results were found for the same transverse momentum range. Using a kinematic cut based approach the $VH(b\bar{b})$ signal search has been performed for the full LHC Run 1 dataset: $4.7$ fb$^{-1}$ at $\sqrt{s}=7$ TeV and $20.7$ fb$^{-1}$ at $\sqrt{s}=8$ TeV. Several backgrounds to this analysis, such as $Wb\bar{b}$ have not been measured in data yet, and an accurate study of the theoretical description has been performed, comparing the predictions of various Monte Carlo generators at different orders. The complexity of the analysis requires a profile likelihood fit with several categories and almost 200 parameters, taking into account all the systematics coming from experimental or modelling limitations, to extract the result. To validate the fit model, a test of the ability to extract the signal is performed on the resonant $VZ(b\bar{b})$ background. A $4.8\sigma$ excess compatible with the Standard Model rate expectation has been measured, with a best fit value $\mu_{VZ}=0.93^{+0.22}_{-0.21}$. The full LHC Run1 dataset result for the $VH(b\bar{b})$ process is a limit of $(1.3)1.4 \times$ SM (expected) observed, with a best fit value of $0.2\pm0.5\mathrm{(stat)}\pm0.4\mathrm{(sys)}$ for a Higgs boson of $125$ GeV mass.