{"id":{"repo_id":"stellenbosch","oai_identifier":"oai:scholar.sun.ac.za:10019.1/135890"},"canonical_url":"https://search.dev.ndltd.org/etd/stellenbosch/oai:scholar.sun.ac.za:10019.1/135890","repository":{"repo_id":"stellenbosch","name":"Stellenbosch University","base_url":"https://scholar.sun.ac.za/server/oai/request"},"display":{"title":"Machine Learning for Physics Applications: Quantum Algorithm Design, Chaotic Dynamics, and Mass Spectrometry","abstract":"This thesis examines the application of machine learning as a framework for discovery and modelling across diverse areas of modern physics. The first chapter introduces computational optimisation as a tool for automated quantum algorithm design, where evolutionary search with a domain-specific language enables the synthesis of quantum algorithms that automatically scale to any problem size. By rediscovering known protocols such as the quantum Fourier transform, Grover’s search, and the Deutsch–Jozsa algorithm, this work demonstrates how machine-learning-inspired search strategies can provide a new framework for quantum algorithm design. The results highlight how a domainspecific language combined with neural architecture search can efficiently navigate the combinatorial spaces inherent to gate-based quantum circuits. The thesis extends this machine-learning-for-physics paradigm to two further domains: mass spectrometry imaging and non-linear chaotic dynamics. In the second chapter, supervised and interpretable machine learning models are applied to mass spectrometry imaging data to identify chemical biomarkers and quantify spatial and spectral structure in biological samples. The results illustrate some of the methodological strengths and weaknesses of applying machine learning as a tool for biomarker discovery, and support the idea that machine learning can complement, but not replace, expert-driven analysis. In the third chapter, neural networks are trained to reproduce the dynamics of the chaotic tent map, which inspired a novel bias initialisation scheme. The proposed initialisation scheme has the advantages that the biases are evenly distributed in the domain, it avoids the dying ReLU problem in the first layer, and training is accelerated since the network starts in a favourable state. Collectively, these studies demonstrate how machine learning serves as a tool for quantum algorithm design and mass spectrometry image data analysis, and conversely, how models of physical systems can be a tool for improving machine learning methods.","abstract_html":"This thesis examines the application of machine learning as a framework for discovery and modelling across diverse areas of modern physics. The first chapter introduces computational optimisation as a tool for automated quantum algorithm design, where evolutionary search with a domain-specific language enables the synthesis of quantum algorithms that automatically scale to any problem size. By rediscovering known protocols such as the quantum Fourier transform, Grover’s search, and the Deutsch–Jozsa algorithm, this work demonstrates how machine-learning-inspired search strategies can provide a new framework for quantum algorithm design. The results highlight how a domainspecific language combined with neural architecture search can efficiently navigate the combinatorial spaces inherent to gate-based quantum circuits. The thesis extends this machine-learning-for-physics paradigm to two further domains: mass spectrometry imaging and non-linear chaotic dynamics. In the second chapter, supervised and interpretable machine learning models are applied to mass spectrometry imaging data to identify chemical biomarkers and quantify spatial and spectral structure in biological samples. The results illustrate some of the methodological strengths and weaknesses of applying machine learning as a tool for biomarker discovery, and support the idea that machine learning can complement, but not replace, expert-driven analysis. In the third chapter, neural networks are trained to reproduce the dynamics of the chaotic tent map, which inspired a novel bias initialisation scheme. The proposed initialisation scheme has the advantages that the biases are evenly distributed in the domain, it avoids the dying ReLU problem in the first layer, and training is accelerated since the network starts in a favourable state. Collectively, these studies demonstrate how machine learning serves as a tool for quantum algorithm design and mass spectrometry image data analysis, and conversely, how models of physical systems can be a tool for improving machine learning methods.","abstract_has_math":false,"creators":["Rouillard, Amy Shirley"],"institution":"Stellenbosch : Stellenbosch University","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Petruccione, Francesco","Sinayskiy, Ilya"],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026-03","date_published":"2026-03","updated_at":"2026-07-24T04:40:12Z","subjects":[],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://scholar.sun.ac.za/handle/10019.1/135890","outbound_label":"Repository record","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Petruccione, Francesco","Sinayskiy, Ilya"]},{"key":"dc:contributor.other","label":"Dc Contributor Other","values":["Stellenbosch University. Faculty of Science. Dept. of Physics."]},{"key":"dc:creator","label":"Author","values":["Rouillard, Amy Shirley"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-04-14T09:51:46Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2026-04-14T09:51:46Z"]},{"key":"dc:date.issued","label":"Date","values":["2026-03"]},{"key":"dc:publisher","label":"Institution","values":["Stellenbosch : Stellenbosch University"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://scholar.sun.ac.za/handle/10019.1/135890"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Thesis (PhD)--Stellenbosch University, 2026.","Rouillard, A. S. 2026. Machine Learning for Physics Applications: Quantum Algorithm Design, Chaotic Dynamics, and Mass Spectrometry. Unpublished doctoral dissertation. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/bece7683-89f2-4e28-b506-b6abe7b8ef8e"]},{"key":"dc:description.abstract","label":"Abstract","values":["This thesis examines the application of machine learning as a framework for discovery and modelling across diverse areas of modern physics. The first chapter introduces computational optimisation as a tool for automated quantum algorithm design, where evolutionary search with a domain-specific language enables the synthesis of quantum algorithms that automatically scale to any problem size. By rediscovering known protocols such as the quantum Fourier transform, Grover’s search, and the Deutsch–Jozsa algorithm, this work demonstrates how machine-learning-inspired search strategies can provide a new framework for quantum algorithm design. The results highlight how a domainspecific language combined with neural architecture search can efficiently navigate the combinatorial spaces inherent to gate-based quantum circuits. The thesis extends this machine-learning-for-physics paradigm to two further domains: mass spectrometry imaging and non-linear chaotic dynamics. In the second chapter, supervised and interpretable machine learning models are applied to mass spectrometry imaging data to identify chemical biomarkers and quantify spatial and spectral structure in biological samples. The results illustrate some of the methodological strengths and weaknesses of applying machine learning as a tool for biomarker discovery, and support the idea that machine learning can complement, but not replace, expert-driven analysis. In the third chapter, neural networks are trained to reproduce the dynamics of the chaotic tent map, which inspired a novel bias initialisation scheme. The proposed initialisation scheme has the advantages that the biases are evenly distributed in the domain, it avoids the dying ReLU problem in the first layer, and training is accelerated since the network starts in a favourable state. Collectively, these studies demonstrate how machine learning serves as a tool for quantum algorithm design and mass spectrometry image data analysis, and conversely, how models of physical systems can be a tool for improving machine learning methods."]},{"key":"dc:title","label":"Title","values":["Machine Learning for Physics Applications: Quantum Algorithm Design, Chaotic Dynamics, and Mass Spectrometry"]}]}],"canonical_facts":{"dc:contributor.advisor":["Petruccione, Francesco","Sinayskiy, Ilya"],"dc:contributor.other":["Stellenbosch University. Faculty of Science. Dept. of Physics."],"dc:creator":["Rouillard, Amy Shirley"],"dc:date.accessioned":["2026-04-14T09:51:46Z"],"dc:date.available":["2026-04-14T09:51:46Z"],"dc:date.issued":["2026-03"],"dc:description":["Thesis (PhD)--Stellenbosch University, 2026.","Rouillard, A. S. 2026. Machine Learning for Physics Applications: Quantum Algorithm Design, Chaotic Dynamics, and Mass Spectrometry. Unpublished doctoral dissertation. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/bece7683-89f2-4e28-b506-b6abe7b8ef8e"],"dc:description.abstract":["This thesis examines the application of machine learning as a framework for discovery and modelling across diverse areas of modern physics. The first chapter introduces computational optimisation as a tool for automated quantum algorithm design, where evolutionary search with a domain-specific language enables the synthesis of quantum algorithms that automatically scale to any problem size. By rediscovering known protocols such as the quantum Fourier transform, Grover’s search, and the Deutsch–Jozsa algorithm, this work demonstrates how machine-learning-inspired search strategies can provide a new framework for quantum algorithm design. The results highlight how a domainspecific language combined with neural architecture search can efficiently navigate the combinatorial spaces inherent to gate-based quantum circuits. The thesis extends this machine-learning-for-physics paradigm to two further domains: mass spectrometry imaging and non-linear chaotic dynamics. In the second chapter, supervised and interpretable machine learning models are applied to mass spectrometry imaging data to identify chemical biomarkers and quantify spatial and spectral structure in biological samples. The results illustrate some of the methodological strengths and weaknesses of applying machine learning as a tool for biomarker discovery, and support the idea that machine learning can complement, but not replace, expert-driven analysis. In the third chapter, neural networks are trained to reproduce the dynamics of the chaotic tent map, which inspired a novel bias initialisation scheme. The proposed initialisation scheme has the advantages that the biases are evenly distributed in the domain, it avoids the dying ReLU problem in the first layer, and training is accelerated since the network starts in a favourable state. Collectively, these studies demonstrate how machine learning serves as a tool for quantum algorithm design and mass spectrometry image data analysis, and conversely, how models of physical systems can be a tool for improving machine learning methods."],"dc:identifier.uri":["https://scholar.sun.ac.za/handle/10019.1/135890"],"dc:language.iso":["en"],"dc:publisher":["Stellenbosch : Stellenbosch University"],"dc:title":["Machine Learning for Physics Applications: Quantum Algorithm Design, Chaotic Dynamics, and Mass Spectrometry"],"dc:type":["Thesis"]},"updated_at":"2026-07-24T04:40:12Z"}