Investigating jet physics through machine learning-enhanced representation

Abstract

The exploration of fundamental particles and their interactions has been central to particle physics for centuries, evolving from ancient philosophical concepts to rigorous scientific inquiry. The Standard Model (SM), formulated in the 1960s, represents a monumental achievement in describing three of the four fundamental forces and predicting the properties of elementary particles. Despite its successes, the SM leaves several critical questions unanswered, including issues related to quantum gravity, the hierarchy problem, mass generation, and the cosmological constant.High-energy particle collisions at facilities such as the Large Hadron Collider (LHC) have been instrumental in advancing our understanding of the subatomic world. These collisions generate vast amounts of data, necessitating sophisticated analysis techniques, including clustering and anomaly detection algorithms, to unravel the complexities of particle interactions. The integration of artificial intelligence (AI) and machine learning (ML) into high-energy physics represents a transformative shift in data analysis. By leveraging advanced algorithms, AI enhances the speed and precision of data processing, offering new pathways for discovery and optimisation. This thesis addresses two primary objectives: first, the development of a novel jet clustering algorithm using machine learning that adheres to physical constraints, providing transparency and interpretability compared to traditional deep learning approaches; and second, the application of innovative methods to mitigate performance bias in synthetic data through the exploitation of inherent data symmetries. The structure of this thesis encompasses a comprehensive exploration of the theoretical underpinnings of jets, the impact of machine learning on high-energy physics, the application of spectral clustering techniques, the influence of physical variables on advanced ML models, and strategies for reducing bias in synthetic datasets. Through these contributions, this work aims to enhance the analytical capabilities in particle physics and further our understanding of the universe’s fundamental nature.

More information

Identifiers

ORCID for Stefano Moretti: orcid.org/0000-0002-8601-7246

ORCID for Srinandan Dasmahapatra: orcid.org/0000-0002-9757-5315

Catalogue record

Export record