Lepton Flavour Universality and Analysis Frameworks

This thesis investigates Lepton Flavour Universality (LFU), the property of the Standard model of particle physics that the electroweak force couples to all three generations of the charged leptons with the same strength. For this purpose, it discusses the measurement of the occurrence of strange B meson decays to tau leptons in the final states at the Large Hadron Collider Beauty experiment (LHCb) experiment at CERN. Building and operating the LHCb experiment requires the effort of more than 1400 physicists and engineers, a colossal detector and massive amounts of computing resources to process and analyse the recorded data. This data and all research outcomes are available to all participants, and should be usable by future physicists, as stated in the convention for the establishment of CERN. Preserving this research and all the associated data is a difficult task, but it is crucial and has a high priority for CERN and for agencies funding particle physics all over the world, who insist on the application of the FAIR principles for data management (the data should Findable, Accessible, Interoperable and Reusable) . This work summarises the methods and tools available or already applied by the LHCb experiment to preserve its data and the derived analyses, in order to identify gaps in the software and methods used. Investigations point to the need to improve the traceability of data analysis byproducts, making sure that the link between recorded data and final results is preserved. For this purpose, this thesis proposes new tools that allow analysts to enrich the metadata associated with LHCb datasets. This metadata can subsequently be used to identify and select datasets, thus simplifying the analysts work. Furthermore, in order to further track the various stages ofLHCb data analyses, these new tools integrate with the popular Snakemake workflow engine, and allow the use of testing and continuous integration systems available at CERN. Using these new tools, this work therefore continues on previous efforts to measure strange B meson decays at LHCb: using simulated data, it refines the categorisation of decay candidates in order to be able to model different types of background. A decision tree is trained to identify signal decays from double charm background which is used both to filter the data and as an input to a three-dimensional fit (alongside the square of the momentum transferred to the tau/tau neutrino system and tau proper decay time). This fit is run on toy simulations, as well as on the existing simulation data, showing that it allows to reliably extract the signal yield within the available samples. This is a major step in the measurement that needs to be followed by further validation of the model and evaluation of systematic uncertainties.