A software in the reinterpretation of model independent search results and constraining theories beyond the Standard Model

One of the main tasks of the CMS experiment at the Large Hadron Collider is the general search for new phenomena not described by the Standard Model, and the experimental examination of specific postulated models and theories that extend the Standard Model (called “BSM theories”). The standard approach for those tasks is the development of many different dedicated analyses, each optimized for one specific signature or model. The “Model Unspecific Search in CMS” (MUSiC) project was developed as an alternative approach to search for new phenomena: An automated scan for deviations between the measured data and the Standard Model simulation is performed without optimizing with respect to properties of the new phenomena. This approach addresses the problem of missing a potential new signature hidden in the data because it is not being looked for. This could happen since the large amount of possible extensions to the Standard Model make it impossible to perform a dedicated search for all of them. Additionally, it could be possible that a new phenomenon exists in the data for which no theory has been developed, yet. The project described in this thesis is an extension to the MUSiC analysis. The goal is to provide information which can be used to restrain the allowed parameter space of many of the previously mentioned theories for which no dedicated analyses have been performed. For this purpose, model independent exclusion limits are calculated on the number of signal events on top of the Standard Model expectation. The limits depend on kinematic properties of the final state in many different event classes which are defined by the particle content of the final state. The event and object reconstruction as well as the Standard Model description and the event classification used for this are shared with the main MUSiC analysis [1] [2]. Once the model independent exclusion limits have been calculated, they can be reinterpreted in terms of a new theory: For this purpose, the total signal selection efficiency, depending on the same kinematic properties and event classes used to calculate the limits, has to be calculated and applied to the model independent limits in order to translate them into model specific limits. As part of the project described in this thesis, a software framework called “MUSiC Embedded Tool for Ascertaining Limits” (METAL) was developed which can be used to calculate the model independent exclusion limits. It also provides a tool that performs the reinterpretation in a fast and easy to use manner. The limit setting procedure as well as the reinterpretation have been validated and a full set of limits has been calculated for the proton-proton collision data recorded with the CMS detector in the year 2012 at a center of mass energy of 8 TeV. The thesis is divided into three part: In the first one, a short overview over the current state of the Standard Model and the CMS detector as well as an introduction to the statistical methods used in this analysis are given. The second part contains a general description of the MUSiC data preparation and event classification as well as the simulation of the Standard Model expectation which is also used by the METAL project. Additional details on that can be found in [1]. The third part contains the discussion of the exclusion limit calculation and the reinterpretation method. Validation results for both aspects are given and example limits are calculated for two BSM models which are then compared to the results of dedicated analyses for the same models. A short user’s guide to the METAL reinterpretation tool and instructions on how to download it can be found in the Appendix B.