Research
Infinite Deconvolution, Lawrence Berkeley National Laboratory
Fall 2023 - Summer 2024
Infinite Deconvolution extends the methodology of Moment Unfolding to encompass the full distribution phase space of jet substructure variables through the application of renormalization techniques. This allows for the analysis of the entire phase space without resorting to the binned or iterative methods common in deep learning. The project highlights the capability of generative models to provide insights into the analysis of particle physics data, through a precise, flexible, and efficient technique.
Moment Unfolding, Lawrence Berkeley National Laboratory
Fall 2021 - Summer 2023
Developed an innovative Generative Adversarial Network (GAN) setup using TensorFlow and Keras to deconvolute truth-level moments from high-energy physics detector data. This method represents a significant advancement by enabling the analysis of complex data distributions without relying on traditional binned and iterative unfolding methods. Thisw is novel, efficient solution to a longstanding challenge in high-energy physics data analysis. This project streamlines the extraction of critical information from detector data and sets a new standard for precision and efficiency. Paper: NeurIPS. (GitHub).
Symmetry Discovery through Deep Learning, Lawrence Berkeley National Laboratory
Fall 2020 – Summer 2021
Symmetry discovery using deep learning, applying machine learning for high energy physics data. Developing a novel, flexible and fully differntiable Generative Adversarial Network (GAN) using TensorFlow and Keras, analysing customised loss functions computationally and analytically to discover symmetries in data sets. Framing a rigorous statistical definition of ‘symmetries’ for physical data sets. Magnifying the power of data sets by reducing their effective dimension. Papers: NeurIPS2021, Physical Review D. (GitHub).
Padé Approximants and the Anharmonic Oscillator, Yale College
Spring 2020
Studying the mathematical formalism of Padé approximants and applying them to prove theorems about the Borel resummability of energy eigenvalues of the quartic anharmonic oscillator (paper).
Supersymmetric Conformal Field Theory, Yale College
Fall 2019
Programming in Mathematica to extract meaningful information from conformal bootstrap data to study new physical theories at the frontier of modern physics. Suggesting explanations for observed rational values for the central charge of the theory at fixed points.
Closed Geodesics on Flat Surfaces, Yale College
Summer 2019
Research on the topology of moduli spaces of Abelian and quadratic differentials on d = 2 Riemmanian manifolds and the identification and classification of points on such surfaces with no closed geodesics through them. Construction of a novel approach to produce such connected surfaces in every genus ≥ 3 with arbitrarily many such points on them. (Journal of Geometry)
Digital Noise Source Development, Yale College
Summer 2018
Programming software defined radio signals to extract phase data for the Canadian Hydrogen Intensity Mapping Experiment in C, python and GNURadio for cutting-edge, frontier research.