Study of rare decays of B hadrons and development of a data viewer application of the detector control system in the atlas detector

Abstract

The first part of the thesis is a study of the sensitivity of the ATLAS detector of the di-onia resonances. In this analysis, which is a first study of di-onia resonances within ATLAS, the desired final measurement is the Br(χb → J/ψJ/ψ). For the needs of this study, specific simulation data of the signal (statistics of 100k events) are being produced and studied. The initial pre- selection of the data provide a first reasonable skim of the events while all ATLAS recommended data quality criteria are fulfilled. Dedicated studies show that the best choice of trigger for this analysis is the EF_2mu4_Jpsimumu, which requires two muons of 4 GeV/c each, to have an invariant mass compatible with the mass of the J/ψ particle. A discrimination between the background and the signal is performed with the use of 6 separation variables, t0, χ2xyz, χ2ndf, α2D, pT, IΔR≤0,7. It follows the optimisation step with a simultaneous scan of all the discriminating variables which provides the optimum corresponding cut values. Before the performance of the sensitivity studies using LHC 2011 data, studies using the LHC 2010 data are performed. With the statistics of 2010, the theoretically expected number of χ is 2010Ntheory = 0.02. The 2010 limit which is calculated with the Rolke method at 95% CL is Br(χb → J/ψJ/ψ) < 4.6 × 10^-3. The branching ratio limit is more than one order of magnitude higher than the theoretically expected value but it is not surprising due to the extremely low statistics of LHC 2010 data. Despite the big discrepancy, this is still a valuable measurement since it is the first measurement of this decay mode of the χb state in hadron colliders. Τhe measured effective cross section σ' of the signal (σ' = N/Lint), is σ'bgr = (0.58 ± 0.01) fb while the theoretical value for the effective cross section of the signal based on [82] lie in the range σ'bgr = [0.23 − 1.47] fb. We conclude that the effective cross section of the measured background is comparable with the effective cross section of the expected signal, which indicates that the analysis has obtained the required sensitivity and that the expected statistics of the LHC 2012 data (even a fraction of them) should lead to a first observation. The second part of this thesis, describes the measurement of the Br(B0s → μ−μ+) using 2.4 fb−1 of the LHC ATLAS data. While the author has extensively contributed to some important steps of the study, the final measurement is a result of a work of many people (indicated in the main document accordingly). The analysis performed is blind and the Br(B0s → μ−μ+) is measured with respect to a well reconstructed reference channel B± → J/ψK±, which minimise systematic uncertainties in the evaluation of relative efficiencies and acceptances, provides small statistical uncertainties and its branching ratio value is taken from PDG. The background is mainly composed of a di–muon non-resonant part (B → μ+μ −X) which is estimated from side-bands interpolation, and a double fake resonant part which is estimated from simulations studies. A big list of 27 discriminating variables has been studied while a method to extract the minimum number of discriminating variables without loss of separation power is developed in order to be achieved a better control of the optimisation step and smaller systematic errors. The final list of discriminating variables is α2D, ΔR, Lxy, t0 significance, χ2xy, χ2z, IΔR≤0,7, a multivariate analysis Boosted Decision Tree classifier, specifically tuned for the needs of this study. A special treatment of the MC–data discrepancies reduce the systematic error of accep- tance and efficiency by tuning the simulation with generator level and data driven per event weights. Nevertheless, systematic uncertainties affect both the number of background events in the signal region (NBGR) and the Single Event Sensitivity (SES). Instead of the NBGR, in the analysis is used the τ (ratio between the background in the signal region and the background in the side-bands region) and its relative systematic uncertainty is 4%. In the case of SES, the main sources of uncertainties are the uncertainty of hte branching ratio of the reference chan- nel and the difference in the b quark fragmentation probabilities, and the tatal δ(S ES )/S ES is 11.8%. Since it is proven that the invariant mass resolution is different for the different η regions, the yield of the reference channel is evaluated from carefully studied fit models in the three mass resolution categories accordingly. Finally, the number of observed Bs events in the signal region is 2,1,0 corresponding to the three mass resolution categories accordingly too. A proper statistical treatment, based on the CLs method, is performed and a limit of the Br(B0s → μ−μ+) < 2.2 × 10−8 at 95% CL is extracted. The third part of the thesis described a data viewing application for ATLAS DCS data. The tool is called Dcs Data Viewer and it is the official ATLAS software that covers the needs of offline detector data visualisation. It is based in a server–client architecture where many different modules execute specific tasks. The server accepts requests of the client, communicates with the DB and reports back the results while the client is the main interface of the project which is composed of a selection module that permits navigation through the DCS information and an output module that offers different ways of visualisation of the data. It is a platform and browser independent project with reasonable application start–up and small response time of typical queries. It offers multiple output formats and its configuration can be saved in XML format. The project is in production since February 2011 with a downtime at zero levels. Due to itʼs intuitive and user friendly features, like the DCS metadata search engine, many detector users that are not DCS experts are using the tool. Furthermore, a CERN authentication mechanism in a combination with database protection algorithms within the DDV, allow exposure to hundreds of users in the world wide web. The proper organisation of the software of the tool in RPM packages with the combination of its simple configuration, make it integrable in other CERN experiments that follow the central DB archiving recommendations, for instance, the COMPASS experiment is already using the DDV. Currently 150 active users are using the tool from many countries all around the world.