Search for the Higgs boson decaying into two muons with the CMS experiment

The Higgs boson was discovered at the CERN LHC by both the ATLAS and CMS collaborations with a mass near 125 GeV in 2012. The measurement of the Higgs boson properties has been one of the main targets of LHC experiments since the discovery. In order to probe the mechanism giving mass to fermions and bosons, the Higgs boson couplings have to be measured. The Standard Model Higgs couplings can be extracted by measuring Higgs boson production and decay rates. Therefore a lot of effort has been done to observe every single experimentally accessible decay in various production modes in the last few years. The bosonic decay modes of the Higgs in gamma gamma, ZZ, and WW have been established using data collected until 2012 (LHC Run 1) while the data collected in 2016 and 2017 allowed to measure the decays to the third generation fermions. The decay of the Higgs boson into tautau has been observed using 2016 data and it confirmed the coupling of the Higgs with leptons. The decay of the Higgs boson into bb has been observed in 2018, hence establishing the coupling with down-type quark. The ttH production mode has also been observed, thus probing directly the coupling to up-type quarks. So far, the Higgs boson couplings have been observed only with the third generation of fermions, thus, the next step would be the search for decays into the second generation. Among those, the Higgs decay into c-quark pairs is the one with the higher event rate. Probing the Standard Model expected event rate is notwithstanding impossible at the LHC because of the large amount of irreducible background and because of the low mass resolution in the dijet final state. Instead, the Higgs boson decay into muons has smaller rates but it also has a clean final state that can be reconstructed with high efficiency and excellent mass resolution, thanks to the CMS tracking and muon systems. However, since the branching fraction B(H→mu mu) is very small (2x104) and the LHC produces abundant backgrounds, an observation of this decay has not been reached yet and upper limits only have been set. My work took part in the most recent search for the Higgs boson decaying to two muons performed by the CMS collaboration. The analysis uses the full proton-proton collision data collected by the CMS experiment at \sqrt{s}=13 TeV during Run 2, corresponding to an integrated luminosity of 137.4 fb-1. The search of the Higgs boson to muon decay targets each of the four main Higgs production channels at the LHC. The four searches use orthogonal data and their results are combined in a single measurement. My work has been dedicated to the improvement of the search in the vector boson fusion (VBF) production mode. In this work, we demonstrate that the VBF channel is the most sensitive among the four H→mu mu searches. The VBF dedicated analysis is based on the fit of a multivariate discriminator, in contrast with the other three dedicated Higgs searches where the fit is performed on the dimuon mass distribution. Moreover, the background estimation is completely based on Monte-Carlo simulations. These two aspects make the strategy to be very similar to the one adopted by two important CMS measurements: the measurement of electroweak production of two jets in association with a Z boson, and the measurement of the Higgs boson decaying into b-quark pairs. Machine learning is a fundamental tool to improve the sensitivity of this search. Two different machine learning techniques are tested for the H→ mu mu search: boosted decision tree (BDT) and deep neural network (DNN). The performance of the two algorithms are compared and the results are extracted from the algorithm that separates better the signal from the background on simulated data. The Higgs boson signal strength is extracted among several regions with a simultaneous maximum likelihood fit and the results are computed for a Standard Model Higgs boson with a mass of 125.38 GeV. In the VBF channel, an excess of events is observed with a strength of 1.36+0.69-0.61, defined as the ratio between the expected and the observed event rate sigmaobs(pp→H→mu mu)/sigmaSM(pp→H→mu mu). The observed excess of events corresponds to a signal significance of 2.4 standard deviations above the background-only hypothesis. The excess of events observed from the combination of the four dedicated searches corresponds to a signal significance of 3.0 standard deviations and it constitutes the evidence of the Higgs boson decay into muons. The results obtained with the data collected at 13 TeV are combined with the ones obtained during Run 1 at the center of mass energies of 7 and 8 TeV. The final combination improves the results of about 1%, resulting in a signal strength of 1.19+0.40-0.30(stat)+0.15-0.14(syst). The thesis is structured as follows. The Standard Model and the Higgs mechanism are introduced in chapter 1. Moreover, the most important measurements of the Higgs boson properties are summarized in that chapter. In chapter 2 the experimental apparatus is described. Firstly, the LHC machine is introduced, then the CMS experiment and the various subdetectors are presented. The algorithms used by CMS to reconstruct physics objects are introduced in chapter 3. In particular, lepton reconstruction and identification are described in the first half of the chapter, while the second half is dedicated to jets. Chapter 4 introduces the search for H$\rightarrow\mu\mu$ with CMS Run 2 data. An overview of the entire analysis is given and the dedicated searches in the ggH, VH, and ttH channels are summarized. The VBF analysis that I have carried out is described in detail in chapter 5. Finally, chapter 6 presents the results of the VBF dedicated search and the results obtained from the combination with the other three dedicated searches and with the data collected during Run 1.