Development for multi-quarks tagger and search for resonances decaying to triple W-boson

The Standard Model (SM) is an extremely successful theory describing the elementary particles in the nature and three of the four well-known fundamental forces (the electromagnetic, weak, and strong force and not including the gravitational force). In the theory, fundamental particles are spin- $\frac{1}{2}$ fermions with interact between them through exchange of spin-1 gauge bosons. The last piece of SM, Higgs, was found by both CMS and ATLAS experiments at Large Hadron Collider (LHC) in 2012. However, the SM has shortcomings. The obvious one is that it fails to include the gravitational force at high energies. In addition, there is a hierarchy problem between the electroweak force and Planck scale, the latter one is 10$^{16}$ times larger than the former one. All these problems motivate the presence of Beyond the Standard Model (BSM) at TeV scale, which leads to rich and novel phenomenology. Searching for new particles, which are introduced by the BSM, is an important topic at LHC. In recent years, probing these heavy resonances through two-body decays into diboson final states have so far found no evidence of new physics. In recent years, many new BSM, i.e., warped extra dimensional model, motivate a triboson resonance indicating discovery potential within LHC reach, while this has not been performed before. This thesis presents the first search for heavy resonances decaying in triple massive W bosons, through W$_{KK}$ → RW → WWW, where W$_{KK}$ is a Kaluza-Klein excited massive gauge boson and R is a scalar radion. The final state topology analyzed consists of one isolated lepton, missing transverse momentum ($p$$_{T}^{miss}$), and one or two massive large radius jets. In most cases, it originates from a W decaying to isolated leptons while the other two Ws decaying hardonically into either two merged AK8 W-jet or even merged to a single merged WW-jet depending on the relative masses of W$_{KK}$ and R bosons. I also consider the case where one of the two merged W bosons decay leptioniclly with a nonisolated lepton inside the jet. So the hadronic final states can be quite different, including resolved R into two merged W-jets, one merged radion R-jet, or partly merged R-jet with particles leaked outside the AK8-jet. Thus the jet mass spectrum can be very complicated, and the tagger to be used can be different in separate mass regions. The analysis is based on proton-proton collision data at √s = 13 TeV collected by the CMS experiment at the CERN LHC during 2016, 2017 and 2018, corresponding to an integrated luminosity of 137fb$^{-1}$. Both regimes with two W-bosons merged in a single WW-jet (merged R) and with two separated W-jets (resolved R) are simultaneously probed. In this search DeepAK8 tagger implementing Deep Neural Networks (DNN) are used to further increase signal-background discrimination, deep-W are used to tag the resolved R, while for the merged R reconstructed as 3- or 4- prong jets, deep-WH is defined as the discriminator. Since there is no 4-prong object with large statistics in the SM, fully merged top together with an extra quark or gluon forms the proxy as a mimic. A novel method, which is refered as "Matix Method", is developed to calibrate the discriminator. The development and calibration of multi-quarks tagger offer possibilities for many analysis, like the measurement of Higgs→ WW → 4q. And the background contributions after signal selection are estimated using a data-driven technique in dedicated control regions. Finally, a wide area of the WKK-R mass parameter space(ranging in M$_{WKK}$ :1.5-5TeV, M$_{R}$:0.18TeV-M$_{WKK}$) is probed for the first time. The observed data are in agrement with the estimated background simulation and WKK scenarios are excluded up to masses of 3.5 TeV and 3.3 TeV for merged and resolved R cases respectively. Another work mentioned in this thesis is the Monte Carlo (MC) contact work for 2 of CMS object or analysis groups (JME and B2G) since 2019. The CMS experiment provides the analyzers with official simulation of Monte Carlo samples, which can be used for analyses. MC contact should creat and validate requests for dedicated processes required by the analyzers and prepare tickets for submission in the online system. Multi-gridpack technique and smarter automatic scripts were implemented to reduce repeating work and help to improve the efficiency.