QSpace

Queen's Scholarship & Digital Collections

QSpace is an open access repository for scholarship and research produced at Queen's University. QSpace offers faculty, students, staff, and researchers a free and secure home to preserve and present their scholarship.

Communities in DSpace

Select a community to browse its collections.

Now showing 1 - 5 of 5

Digital Collections 5940
This community includes digital collections produced by members of the Queen’s community, as well as digital special collections made available via W.D. Jordan Rare Books & Special Collections.
Exams & Syllabi 0
This community provides access for staff and students at Queen’s University to degree examination papers and syllabi. To access Exams & Syllabi off campus please login using your Queen's NetID and password.
Graduate Theses, Dissertations and Projects 10097
This community includes graduate theses, dissertations and projects produced by students at Queen’s University.
Research Data 93
This community includes research data produced by faculty and staff at Queen’s University.
Scholarly Contributions 14365
This community includes Queen’s peer-reviewed research publications, including journal articles, book chapters, conference proceedings, and more.

Recent Submissions

Access status: Embargo ,
Conic Sections of (anti)-de Sitter and Weyl Spacetimes
(2026-03-12) Rochfort, Dominic Maxim; Physics, Engineering Physics and Astronomy; Lake, Kayll
This thesis investigates exact stationary electrovacuum solutions to Einstein’s field equations through the development and application of coordinate systems related to conic sections in both (anti)-de Sitter and Weyl spacetimes. This work addresses a longstanding and much underappreciated limitation of oblate-spheroidal coordinates that has persisted since Carter (1968), specifically on the focal ring radius, a, when the cosmological constant Λ is negative. Consequently, the spin parameter of a rotating black hole is bounded above when Λ < 0. We derive Boyer–Lindquist coordinates in (anti)-de Sitter space from first principles, introducing a new coordinate system that resolves three coordinate pathologies present in Boyer–Lindquist coordinates as a approaches the cosmological length scale l, resulting in a new metric valid for all values of a. We utilize these well-behaved coordinates to analyse for the first time the ‘critically spinning’ and ‘overspinning’ Kerr–anti-de Sitter spacetime. We determine these spacetimes contain no curvature singularity, the causal structure is such that the roots of ∆r do not locate event horizons, and they cannot contain a Killing vector field that is asymptotically static for all observers at spatial infinity. We calculate the variables appearing in the first law of black hole thermodynamics, and evaluate them for the first time using coordinates that behave well for all values of a, and conclude a < l is a true physical limitation and not a coordinate pathology; the energy and angular momentum genuinely diverge. Weyl spacetimes possess an analogue to a Newtonian potential ψ that satisfies the Laplace equation. By restricting to only those Weyl spacetimes that can be written in orthogonal coordinates (t, u, v, ϕ), such that ψ = ψ(u), we obtain nine one-parameter families of spacetimes – including Curzon–Chazy and Zipoy–Voorhees spacetimes; each one associated with a conic section and a Newtonian source. Following Abdelqader and Lake (2012), we establish a coordinate invariant comparison between these spacetimes and their Newtonian analogues by examining the topology of the gradient fields of their tidal invariants. We demonstrate exactly 26 special cases which are Petrov type D, all of which are either Schwarzschild, Taub, or Minkowski.
Access status: Open Access ,
Making Sense of Decolonization Through Artificial Intelligence, Digital & Conventional Art Creation
(2026-03-12) Milburn, Carleigh Candice Mignonne; Cultural Studies; Morcom, Lindsay
Making Sense of Decolonization Through Artificial Intelligence, Digital and Conventional Art Creation is a research-creation project that uses interactive art installations, collaborative research practices, and dialogue to explore the intersections of Indigenous perspectives, artificial intelligence (AI), and decolonization. This project asks: How might innovative data art collection practices in Indigenous contexts facilitate understanding of decolonization and promote meaningful relationships toward reconciliation? To what extent can AI in art be considered decolonial? This research-creation examines the strength of both Indigenous and non-Indigenous art, alongside AI, as storytelling tools. It fosters inclusive conversations about decolonization through both technological and traditional artistic approaches. Decolonization is the process of dismantling colonial control and restoring Indigenous rights. It emphasizes self-governance, land stewardship, and the continual questioning of colonial systems and power structures (King, et al. 2024). This project emphasizes the importance of building nurturing systems for future generations. Fundamental to the inquiry is the question: What does decolonization look like? The project engages with the Anishinaabemowin concept of Biskaabiiyang, and the role of the self, by exploring how technology, particularly AI, and traditional collaborative art practices can help visualize decolonization. Findings suggest that collaborative and digital art practices, especially those guided by Indigenous protocols, can foster deeper understandings of decolonization and relationality. Through workshops, participant co-creation, and AI-generated visual analysis, the research highlights how respect, reciprocity, and relational accountability are enacted through artistic engagement. Current AI systems are not inherently decolonial, they hold potential when shaped by Indigenous-led development and ethical frameworks. The use of digital media in this context enhances accessibility and expands possibilities for inclusive and future-oriented artistic practices.
Access status: Embargo ,
On Land, Life, and Law: Mediations Between Canadian Legal Regimes and Kanyen'kehá:ka Expressions of Law and Polity
(3/5/2026) Collis, Susan; Geography and Planning; Mullings, Beverley; Pasternak, Shiri; Cohen, Dan
This dissertation investigates the role of Canadian law in producing and managing colonial relations of power. It centres the meaning and practice of law as multivalent, malleable, and always in motion in any number of both expected and less predictable directions. Framed by the period spanning the last quarter of the twenty-first century to the present day, I examine a rewriting of the legal regime structuring settler-colonialism in Canada and the related development of a state project of reconciliation. Noting the gaps as much as the content of Canada’s settler-colonial legality, I examine where, when, how, and why the settler-colonial legal regime shows up in full force or is absent or altered by interaction and interrelation with Kanyen’kehá:ka (Mohawk) enunciations of law and polity. I foreground how Indigenous lands are not governed by a singular, coherent legal regime, but rather by disparate sources of law and power, exploring some of the many ways that Kanyen’kehá:ka people transgress state visions for reconciliation and the related legal frameworks demarcating First Nations governance and jurisdiction.
Access status: Embargo ,
Sex, Metabolic Profiles, and Alcohol: Evaluating the Prognostic Significance of Steatotic Liver Disease Classifications in the Canadian Longitudinal Study on Aging (CLSA)
(3/5/2026) Rapino, Carmela; Public Health Sciences; Saeed, Sahar and Sebastiani, Giada
Background: The subclassifications of Steatotic Liver Disease (SLD) were intended to provide a clinically meaningful framework for assessing prognosis and guiding intervention that encompasses a spectrum of mixed aetiologies combining metabolic dysfunction and increasing levels of alcohol consumption. However, little consideration is given to the disease's dynamic nature over time. We aimed to evaluate the association between increasing levels of alcohol consumption and all-cause mortality in those with and without steatosis. Methodology: We analyzed data from the Canadian Longitudinal Study on Aging, a prospective cohort study following adults aged 45-85 years between baseline (2012-2015), 3 years later for follow-up 1 (FUP1), and until 2021(follow-up 2). Hepatic steatosis was identified using the NAFLD Ridge Score, with cardiometabolic factors determined from laboratory and medication data. Alcohol consumption was quantified using self-reported weekly intake and used to define SLD subclasses (MASLD, MetALD and ALD). Time-varying, and Marginal Structural Model Cox proportional hazards models were used to estimate the association between increasing levels of alcohol consumption and mortality among those with and without steatosis, overall and sex-stratified, adjusting for age, income, physical activity, diet, smoking, binge drinking, sex, and the cumulative number of cardiometabolic factors. Results: Between 2012 and 2021, 16,560 participants (52.1% female, median age 58 (IQR: 51-66)) were included. At baseline, MASLD was the most prevalent subtype at baseline (n = 8430 | 33.9%). Between baseline to FUP1, 35.4% of people with ALD, 23.7% MetALD, and 22.6% MASLD transitioned to different SLD subtypes. Over a median follow-up of 7.8 years, 1041 deaths occurred. Overall, steatosis increased the risk of death and after accounting for time-varying exposures and confounders, ALD had the highest risk of all-cause mortality (adjusted hazard ratio (aHR) 6.11 | 95% CI: 1.76-21.19). Conclusion: SLD is dynamic, with evolving mortality risks that cannot be captured by baseline classifications alone. Incorporating time-varying covariates yields a more accurate understanding of prognosis.
Access status: Embargo ,
Analog training of neuromorphic hardware on silicon photonics
(3/5/2026) Guo, Zhimu; Physics, Engineering Physics, and Astronomy; Bhavin J. Shastri
Conventional computers are organized around a centralized processing model known as the von Neumann architecture, however, recent advancements in artificial intelligence have exposed its deficiencies for computational neural network models that are distributed, massively parallel and adaptive. The hardware bottleneck has led to renewed interest in researching analog photonic computing and neuromorphic hardware platforms. While neuromorphic engineering implements elements of machine learning algorithms directly in hardware, photonic circuits can provide high-performance implementations of neural networks thanks to their interconnectivity and linear operations. Combining their advantages, neuromorphic photonics have shown success in various machine learning applications implementing a diverse range of network architectures. However, there is a significant gap when it comes to the training these neuromorphic photonic hardware platforms despite their impressive performance during inference. Previous attempts such as backpropagation have shown limitations such as low precision and insufficient scalability. Further studies on analog-friendly algorithms aim to avoid the analog hardware limitations while maintaining similar level of performance. Nevertheless, digital computations such as numerical matrix multiplications still require higher precision to achieve better training performance. Recent developments have seen a strong momentum towards analog training methods that are specifically designed for analog hardware. Among these proposals, multiplexed gradient descent (MGD) proves to be particularly suited for analog neuromorphic photonic hardware. MGD algorithm provides a fully analog training pipeline that minimizes the requirements for digital computations, while offering excellent scalability for training large-scale analog networks. In this dissertation, we explored MGD algorithm with different analog neuromorphic photonic hardware platforms and demonstrated its performance and versatility in a variety of real-world applications. This signifies a meaningful step towards a fully-autonomous analog end-to-end training system that requires minimal intervention from external digital or human inputs.