Selection techniques for optimal meta-analysis of beyond standard model physics

Yellen, Jamie (2025) Selection techniques for optimal meta-analysis of beyond standard model physics. PhD thesis, University of Glasgow.

Abstract

This thesis addresses the development of selection techniques tailored for optimising meta-analyses in Beyond Standard Model (BSM) physics, focusing on three key objectives: (i) identifying a minimally overlapping set of results, (ii) implementing these selections for model exclusion, and (iii) extending them to anomaly detection applications. Central to this study are the hdfs and whdfs algorithms—graph-based methods that systematically address the combinatorial challenge of selecting optimal result combinations.

In the context of model exclusion, the thesis applies the whdfs algorithm within the taco project to optimise combinations of analyses by estimating overlaps in signal regions (SRs). Using simplified model spectra looking at susy-like processes, the project demonstrates a measurable increase in exclusion. The proto-models project, an extension to the taco project and previous work by the SModelS collaboration, focused on anomaly detection, adapting the whdfs algorithm to construct a test statistic for identifying significant deviations from the Standard Model (SM) hypothesis. Through iterative improvements in the algorithms’ weighting mechanisms, the study presents a self-regulating test statistic for the measure of significance.

The findings highlight the dual utility of the hdfs and whdfs algorithms across domains, from collider-based physics applications to machine learning contexts. This work thus contributes a computationally robust framework that enhances reinterpretation capacity in particle physics and supports further integration with reinterpretation tools like SModelS, MadAnalysis 5 Rivet and Contur. The research underscores the increasing importance of efficient, adaptable algorithms for data-intensive BSM analyses. It lays the groundwork for future reinterpretation methodologies necessary for maximising data utility in HL-LHC and related high-energy physics experiments.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Additional Information: Supported by funding from the Scottish Data-Intensive Science Triangle (ScotDIST) funding program.
• Subjects: Q Science > QB Astronomy; Q Science > QC Physics
• Colleges/Schools: College of Science and Engineering > School of Physics and Astronomy
• Funder's Name: Scottish Data-Intensive Science Triangle (ScotDIST)
• Supervisor's Name: Buckley, Professor Andy
• Date of Award: 2025
• Depositing User: Theses Team
• Unique ID: glathesis:2025-85009
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 08 Apr 2025 10:45
• Last Modified: 08 Apr 2025 10:48
• Thesis DOI: 10.5525/gla.thesis.85009
• URI: https://theses.gla.ac.uk/id/eprint/85009