Theses

Permanent URI for this collectionhttps://uwspace.uwaterloo.ca/handle/10012/6

The theses in UWSpace are publicly accessible unless restricted due to publication or patent pending.

This collection includes a subset of theses submitted by graduates of the University of Waterloo as a partial requirement of a degree program at the Master's or PhD level. It includes all electronically submitted theses. (Electronic submission was optional from 1996 through 2006. Electronic submission became the default submission format in October 2006.)

This collection also includes a subset of UW theses that were scanned through the Theses Canada program. (The subset includes UW PhD theses from 1998 - 2002.)

Browse

Now showing 1 - 20 of 16785

Advancing Proteomic Analyses with Graph-Based Deep Learning: Protein Inference and DIA De Novo Peptide Sequencing
(University of Waterloo, 2025-08-12) Ma, Zheng
Proteomic analysis plays a central role in unraveling the complex molecular underpinnings of biological systems. However, traditional approaches to protein inference and peptide sequencing have been hampered by challenges such as data complexity, label scarcity, and spectral noise. In this thesis, we leverage advanced deep learning techniques to address these challenges, thereby expanding the efficacy of proteomic analyses. Our work is organized around three major contributions. First, we introduce GraphPI, a novel protein inference framework that redefines the inference problem as a node classification task within a tripartite graph structure. In GraphPI, proteins, peptides, and peptide-spectrum matches (PSMs) are modeled as interconnected nodes, while edges incorporate features such as peptide identification scores and a specialized peptide-sharing attribute. By harnessing a tailored graph neural network (GNN) architecture inspired by GraphSAGE, our approach effectively aggregates and propagates information across heterogeneous node types. Critically, GraphPI is trained in a semi-supervised manner using pseudo-labels generated from established protein inference methods, combined with hard negative decoy information. This training process not only circumvents the typical bottleneck of limited labeled data but also yields protein scores that generalize across diverse datasets, all while substantially reducing computational overhead relative to Bayesian network–based approaches. Experimental evaluations on multiple benchmark datasets demonstrate that GraphPI delivers competitive accuracy with significant speed improvements, thus paving the way for real-time applications in large-scale proteomic studies. Second, we present DIANovo, an innovative deep learning method designed to tackle the inherent complexities of Data-Independent Acquisition (DIA) data for de novo peptide sequencing. Unlike conventional de novo approaches that often struggle with the multiplexed nature of DIA spectra, DIANovo incorporates a suite of strategies to manage coelution and spectral noise. Our approach begins by constructing a spectrum graph that captures the mass differences between peaks. Next, a Transformer-based encoder, enhanced with Rotary Positional Embeddings (RoPE), processes the graph by encoding these mass differences along its edges, effectively treating the spectrum graph as fully connected. Furthermore, DIANovo introduces a coelution-aware pretraining stage, where the model is first optimized to predict ion types from coeluting peptides. This pretraining step equips the network with a nuanced understanding of spectral interferences, thereby improving the fidelity of subsequent peptide sequence predictions. In addition, a two-stage decoding strategy is employed: the first stage identifies an optimal path through the spectrum graph, while the second refines this path to generate a final amino acid sequence by filling in mass tags. Comparative analyses against state-of-the-art methods reveal that DIANovo achieves significant improvements in both amino acid and peptide recall, especially when applied to high-quality narrow-window DIA data obtained from next-generation instruments such as the Orbitrap Astral. Moreover, we investigate whether DIA identifies more peptides than DDA in de novo sequencing by comparing their performance on the same biological sample under varying acquisition modes and parameters. Our results demonstrate that DIA only outperforms DDA when employing narrower isolation windows. The third component of this thesis presents a comprehensive theoretical analysis that sheds light on the performance limits of peptide identification methods. By linking the signal-to-noise profile to peptide identification accuracy, our study elucidates the inherent trade-offs between Data-Dependent Acquisition (DDA) and DIA strategies. We derive quantitative metrics to predict peptide identification performance under a range of experimental conditions, and these predictions are validated against empirical data. This framework not only explains why Astral DIA data can provide superior peptide coverage in certain scenarios but also delineates the conditions under which peptide identification is most favorable. These insights are crucial for guiding the design of future mass spectrometry experiments and for optimizing computational pipelines in proteomic research. Collectively, the three contributions of this thesis demonstrate the transformative potential of integrating deep learning with advanced computational frameworks in proteomics. GraphPI and DIANovo both showcase how novel neural network architectures can overcome longstanding challenges in protein inference and de novo peptide sequencing, while the theoretical analysis provides a foundation for understanding and further refining these methodologies. The experimental results across multiple datasets underscore the robustness, efficiency, and generalizability of our approaches, suggesting that deep learning–based strategies will play an increasingly central role in the future of proteomic analysis. In conclusion, this work not only advances the state-of-the-art in protein and peptide identification but also offers practical solutions for handling large-scale, complex proteomic data. By bridging the gap between theoretical insights and practical implementations, our integrated framework lays the groundwork for enhanced biomarker discovery, more accurate disease diagnosis, and a deeper understanding of biological systems at the molecular level.
Planar Microcavity-enhanced Lasing from Semiconducting Carbon Nanotubes
(University of Waterloo, 2025-08-12) Yang, HeeBong
Single-walled carbon nanotubes (SWCNTs) exhibit exceptional optical properties, including strong oscillator strength, size- and chirality-dependent tunable bandgap energies, and near-infrared (NIR) photoluminescence (PL) at room temperature. These features make them highly attractive and promising candidates for a wide range of optoelectronic and quantum optical applications. However, their practical deployment has been significantly hindered by several challenges such as PL quenching due to nanotube bundling, structural degradation from high-temperature annealing, or internal loss mechanism of exciton-exciton annihilation (EEA) under high excitation power. To overcome these challenges and enhance the radiative efficiency of SWCNTs, we design and fabricate wafer-scale dielectric distributed Bragg reflectors (DBR) to construct high-quality Fabry-Perot (FP) planar microcavities, where thin films of chirality-sorted, well-dispersed SWCNTs. These cavities serve to promote strong light-matter coupling between confined cavity photons and excitons in the SWCNT films. We perform optical characterization of the SWCNT-embedded microcavity devices both in weak and strong light-matter coupling regimes. First, in the strong coupling regime, we observe the SWCNT exciton-polariton dispersions with vacuum Rabi splitting (VRS) strength around 210 meV. Unfortunately, the polariton condensation was not achieved, primarily due to internal material loss mechanisms such as EEA process or strong reabsorption. On the other hand, we observe nonlinear lasing behavior in weak light-matter coupling regime, evidenced by a pronounced spectral linewidth narrowing signature with increasing excitation power. This would represent the first demonstration of lasing from SWCNT thin films integrated into planar dielectric microcavities, marking a significant milestone in carbon-based nanophotonics. While the results are promising, there remains considerable scope for enhancing the coherence and quality of the emitting light from these SWCNT-embedded microcavity devices, an essential step toward realizing scalable quantum nanophotonic platforms.
Identifying Latent Profiles of Family-Wide Dynamics: Associations with Child and Caregiver Mental Health
(University of Waterloo, 2025-08-12) Castelino, Chantelle
Family functioning and the mental health and wellbeing of individual members are intricately connected and intertwined. However, the conceptualization of mental health concerns based on relational patterns remains underutilized within clinical practice. Moreover, most family research focuses on dyadic processes, often overlooking broader family-wide phenomena and within-family differences. To address these limitations, the present study identified latent profiles of family functioning, based on indicators relating to family subsystems and contextual factors. Subsequently, latent profiles were used to predict child and caregiver mental health. Participants came from two samples with harmonized measurement: one general population cohort (n = 549 families) and one cohort from a family-based psychological clinic (n =124 families). After identifying latent profiles, child and caregiver mental health outcomes at a later time (i.e., 18 and 12 months, respectively) were examined as a function of profile. The distribution of covariates was also examined across profiles. Results of this study support the presence of heterogeneity in family dynamics in two populations. Four profiles emerged in the general population sample: Higher Functioning, Moderate, Couple Distress, and High Conflict. Three profiles emerged in the treatment-seeking sample: Higher Functioning, Couple Distress, and Child/Sibling Tension. In both samples, profile membership predicted later child and caregiver outcomes. This study’s findings demonstrate the importance of studying family processes across multiple relational subsystems. Moreover, results support the utility of person-centered approaches and their applications towards clinical conceptualizations and tailored interventions.
CO2 conversion to light hydrocarbons over K/Fe2O3-Al2O3 synthesized via the reverse microemulsion method
(University of Waterloo, 2025-08-12) Lin, Zixuan
The increasing concentration of atmospheric carbon dioxide (CO2), primarily driven by human activity, has intensified global concerns over climate change. One promising strategy to address this issue is the catalytic conversion of CO2 into valuable hydrocarbons, offering a sustainable route for emission reduction and fuel production. Potassium-promoted iron-based catalysts were investigated for CO2 hydrogenation via a modified Fischer–Tropsch (FT) process. High specific surface area catalysts were synthesized using the reverse microemulsion method, enabling controlled particle size and dispersion. The effects of potassium (K) loading (0-11.3 wt%), active phase, support, H2:CO2 feed ratio (1-4), reaction temperature (300-500 °C), pressure (4-12 bar), and GHSV (750-4000 mL/(gcat∙h)) were examined. Catalytic performance was evaluated by CO2 conversion, C2+ hydrocarbon selectivity, and space time yield (STY). Fresh and spent catalysts were characterized using XRD, TPR, BET, TEM, TGA-FTIR, and ICP techniques. The 7.8%K/Fe2O3-Al2O3 catalyst exhibited the highest activity, achieving 50% CO2 conversion, 53% C2+ selectivity, and a STY of 7.72 mmol/(gcat∙h) at 11 bar, 1000 mL/(gcat∙h), and 400 °C. In contrast, the catalyst without potassium showed significantly lower performance, with 24% conversion, 12% selectivity, and a STY of 0.87 mmol/(gcat∙h). The enhanced activity is attributed to the formation of active χ-Fe5C2 and Fe3O4 phases under reaction conditions, facilitated by the uniform nanoscale morphology of the catalysts synthesized via the reverse microemulsion method.
Quantum Error Correction and Quantum Metrology with Non-Markovian Noise
(University of Waterloo, 2025-08-12) Mann, Zachary
Quantum technologies have the potential to solve many important problems across science and industry. An important example is quantum computation. Quantum simulators promise to better model chemistry. Further, Shor’s factoring algorithm solves a problem in exponentially less time than what it would take now on our classical computers. This has led many to believe that quantum computers could bring exponential speedups to other difficult, real-world problems, such as optimization. Another example is quantum sensing, where quantum mechanical effects can be leveraged to increase measurement precision beyond the classical state of the art. This has many applications in both fundamental science, such as the LIGO experiment, and in industry, such as Nitrogen vacancy magnetometers. For these quantum technologies to reach their full potential, however, the barrier of noise must be overcome. Quantum effects usually live at very small system sizes or very cold temperatures, making them extra sensitive to thermal noise or small perturbations of the environment. A proposed solution to this problem, for both computation and sensing, is to use quantum error correction. Quantum error correction encodes a few quantum degrees of freedom into many physical degrees of freedom, building in redundancy. This redundancy allows for the detection and correction of unwanted errors in our protocol. Most of the literature on quantum error correction focuses on Markovian noise models, i.e., models where the noise is not temporally correlated. The temporally correlated, or non-Markovian, regime remains relatively unexplored. In this thesis, we explore quantum error correction for non-Markovian noise models. We first present a few of the many definitions and models for quantum non-Markovian phenomena present in the literature. We then generalize the Knill-Laflamme quantum error conditions to the hidden Markov model, an experimentally motivated model of non-Markovian noise. These conditions allow one to guarantee that a quantum error-correcting code will still do its job for more realistic noise models. Finally, we apply our notion of non-Markovian error correction to quantum sensing. We generalize previous Markovian results and derive conditions for guaranteeing Heisenberg limited precision scaling in the presence of temporally correlated noise using quantum error correction. The Heisenberg limit is the fundamental precision limit allowed by quantum mechanics for parameter estimation in a physical system. We also study the next-best achievable precision scaling when the Heisenberg limit is unattainable.
Design, Fabrication, Packaging, and Characterization of Nano Inertial and FET Sensors
(University of Waterloo, 2025-08-11) Gulsaran, Ahmet
Miniaturization is necessary and unavoidable for the advancement of sensor technology. This thesis explores the effect of miniaturization on performance and proposes solutions to addressed challenges through the development of a novel packaging solution and engineering device interfaces. First, a theoretical model is developed to understand the impact of miniaturization on sensor properties such as sensitivity and signal-to-noise ratio for nano and micro electro mechanical systems-based (N/MEMS-based) gas sensing applications. Then, these theoretical predictions are experimentally verified through the experimental comparison of NEMS and MEMS-based humidity sensors. The experiments confirm that miniaturization enhances the sensitivity of inertial gas sensors. However, it also introduces challenges such as lower signal-to-noise ratio (SNR) due to attenuated output signals that are more susceptible to noise, thereby motivating the development of improved packaging and interface solutions. To minimize the noise sources, a novel packaging technique, called built-in packaging, is proposed and performance of MEMS resonators and metal-insulator-metal (MIM) diodes are compared with the conventional techniques such as wire bonding and probing. Experimental results show that built-in packaging methods enhance output signal levels by 12\% while preserving overall performance. Finally, to implement novel materials into sensing technologies, interface engineering is studied for development of colloidal quantum dot based field effect transistors (CQD-based FETs). These devices can quantify the electrical properties of novel CQDs and highlights the importance of interface engineering. The quantified results accelerates the development of improved CQDs, leading to better photodetector performance in infrared (IR) detection. Overall, this thesis presents a detailed analysis of the trade-offs in sensor miniaturization and provides critical insight for the advancement of compact and high-performance sensors.
Distracted Desire: Extending the Study of Non-Erotic Thoughts to the South Asian Context
(University of Waterloo, 2025-08-11) Padda, Taranjot
Purity culture is characterized by sexual norms emphasizing premarital abstinence, modesty, and positioning women as sexual gatekeepers. To date, purity culture has primarily been studied within Evangelical Christian communities. However, similar sexual values may operate across other cultural contexts, including South Asian communities where sexuality is often framed as procreative, within marriage, and as a source of familial honour or shame. This thesis investigated whether purity culture provides a useful framework for understanding the sexual experiences of South Asian diasporic women, focusing specifically on non-erotic thoughts (NETs)—intrusive cognitions during sexual activity that interfere with sexual enjoyment. NETs were selected as the focus because they are viewed as a central mechanism in several influential theories of sexuality, including Masters and Johnson's concept of spectatoring, Barlow's cognitive-affective model, and Brotto's mindfulness framework. Despite widespread recognition that sexuality is culturally embedded, no research has examined whether NETs differ across cultural groups. Through two studies, this thesis explored cultural differences in NETs and their underlying mechanisms among South Asian and White women in North America and the United Kingdom. Study 1 (N = 301) compared the frequency and anxiety associated with NETs between second-generation South Asian women and White women. Results revealed significant cultural differences: South Asian women endorsed greater thoughts and anxiety related to moral concerns, shame or guilt, pain, pregnancy and sexual health, and their partners' emotions, while White women endorsed greater concerns about body appearance. Study 2 (n = 284) replicated these findings and tested a series of mediational models examining factors theorized to influence the occurrence of NETs and their influence on sexual outcomes. Results identified shame as a significant antecedent for both South Asian and White women and internal conflict over one’s identity as an antecedent to NETs for South Asian women. In these mediational models, NETs subsequently predicted lower sexual satisfaction and higher sexual distress. Contrary to expectations, purity culture endorsement did not directly predict NETs, suggesting that cultural influences on sexuality may operate through more proximal psychological mechanisms. These findings demonstrate meaningful cultural variation in sexual cognitions, highlighting the need for culturally sensitive models of sexuality and clinical interventions that address the intersection of cultural identity and sexual well-being.
Local Theories and Efficient Partial Quantifier Elimination
(University of Waterloo, 2025-08-11) Getachew, Estifanos Sisay
Quantifier elimination is used in various automated reasoning tasks, including quantified SMT solving, exists/forall solving, program synthesis, model checking, and constrained Horn clause (CHC) solving. Complete quantifier elimination, however, is computationally intractable for many theories. The recent algorithm QEL shows a promising approach to approximate quantifier elimination, which has resulted in improvements in solver performance. QEL performs partial quantifier elimination with a completeness guarantee that depends on a certain semantic property of the given formula. In this thesis, we study local theories, focusing on their proof theoretic and semantic characterization. We identify a subclass of local theories in which partial quantifier elimination can be performed efficiently. By considerably generalizing the previous approach, we present T-QEL, a parametrized polynomial time algorithm that is relatively complete for this class of theories. The algorithm utilizes the proof theoretic characterization of the theories, which is based on restricted derivations. Finally, we prove for T-QEL, soundness in general, and relative completeness with respect to the identified class of theories.
Are We on the Same Track? Using Lived Experiences to Understand the Complex Impacts of New Transportation Investment
(University of Waterloo, 2025-08-11) McDougall, Emma
North American planning is now emphasizing more integrated approaches that centre new transportation infrastructure in larger urban redevelopment approaches, such as transit-oriented development. Research demonstrates numerous measurable land-use and economic benefits associated with planning that prioritizes density, vibrancy, and accessibility through public transit and active transit infrastructure. But planners and policymakers lack insight into the mobility benefits and potential consequences for the original community and particularly, deeply marginalized residents. Importantly, these planning approaches are using transportation infrastructure as much more than a tool for improved transportation, and thus the implications also extend beyond travel. As a response, this dissertation challenges the inherent benefit of large-scale transportation investment by providing a series of counternarratives to current understanding of the impact of new transportation on a community and its residents. This is done through lived experience data, which this dissertation argues is necessary for acquiring a comprehensive understanding of how these planning projects bring about change. Currently, implications of new transportation on individual mobility and on community transformation are underexplored. Using a mobility justice lens, this research interrogates unique dimensions of the impact and divide for residents and stakeholders navigating new active transportation, such as bike lanes, and new transit, such as light rail. Using the Region of Waterloo as a study area, this dissertation provides three unique case studies that centre lived experiences and perspectives from different stakeholders, with a particular emphasis on deeply marginalized residents. Three sets of semi-structured interviews were conducted: the first with 22 key stakeholders (planners, politicians, etc.) who were involved in the development of the region’s ION light rail transit (LRT) project and have in-depth knowledge about its planning and political economy. Second, 22 realtors and developers working in the region were interviewed, as they have a strong sense of the regional market and associated trends. Finally, 20 deeply marginalized Region of Waterloo residents were interviewed, as they have experienced this change firsthand and are the most impacted by it. Waterloo Region presents as a particularly salient case study, as it is the smallest North American region to operate an LRT. Further, the arrival of the LRT came with a fundamental reorganization of the previously strong bus transit network. This means all regional ridership has been impacted by the LRT, whether residents use the LRT or not. Findings are presented through three empirical manuscripts. The first manuscript demonstrates the divide between perceived project goals and ridership experience from community members who are navigating the integration of a new transit system. These findings highlight conflict between expert opinions and community needs, arguing that while appearances have improved, the ridership experience has actually declined for some. The second manuscript finds that opinions of transportation can be influenced by positionality, both physically and professionally, as suburban residents and realtors struggle to see themselves, or their needs, in new active and public transit. The final manuscript provides a holistic vantage of the interconnectedness of transportation and other social, political, and economic constraints. Through interviews with deeply marginalized residents, this manuscript finds that transportation investment has cascading impacts that disproportionately affect equity-deserving groups in visible and invisible ways. Collectively, these findings demonstrate the inherent bias that individuals and groups bring to conversations of transportation. Residents, stakeholders, and experts bring their own lived experiences that can prevent them from fully acknowledging or understanding oppositional opinions from individuals with vastly different experiences than their own. Essentially, this work stresses the importance of acknowledging, incorporating, and collecting diverse experiences and perspectives from as many groups as possible in the planning of new transportation. These perspectives can then be used to inform transportation systems that are beneficial for residents and the community at large. This research is valuable for theory as it contributes to the further theorization of mobility justice in planning research while addressing a growing gap in the way planners consider transportation’s complex impacts. It further provides numerous recommendations for planning practice, including integrating more diverse perspectives into all stages of the planning process, partnering with community organizations to ensure equity-deserving groups have access to the planning process, and reflecting on the role that transportation is meant to play within a community.
Unregulated drug use in Sudbury, Ontario: A rapid ethnography examining risks and harm reduction service access in the North
(University of Waterloo, 2025-08-11) Tucker, Lucas
Background: Unregulated drug use in Ontario, Canada, has led to high rates of HIV, Hepatitis C, and drug overdose mortality. This crisis disproportionately impacts people who use drugs (PWUD) in Northern, rural, and smaller urban Canadian communities due to barriers such as limited resources and heightened stigma. For example, Sudbury, Ontario, has over double the provincial overdose rate. Supervised consumption sites (SCS) are intended to address these individual and community harms. However, the only SCS in Sudbury, The Spot, closed as of March 2024. The current thesis had the following objectives: i) Understand any perceived impacts of the local SCS closure, ii) Explore how necropolitics and structural violence create harm towards PWUD, iii) Qualitatively visualize areas of drug use, harm reduction sites, and participants’ living locations, iv) Provide recommendations to local organizations to address harm reduction gaps, and v) Provide research that addresses the literature gap regarding the health and well-being of Northern PWUD. Methods: We conducted a community-based rapid ethnography, which included naturalistic observations, semi-structured interviews (n=27), and spatial mapping. The data were analyzed thematically using a participatory analytic approach. Results: In Chapter Four, participants identified four key ways the SCS closure impacted them. They reported using substances unsupervised, which increased risks associated with isolated drug use. This led some to begin using drugs in public as a harm reduction strategy, hoping for intervention in case of an overdose; however, public use triggered fears of police intervention and social stigma if they were observed. Furthermore, losing an SCS resulted in social isolation and barriers to accessing harm reduction supplies, as the stigma associated with public health sites caused participants to avoid them altogether. In Chapter Five, our geospatial analysis of data derived from interviews, ethnographic observations, and peer knowledge allowed for a visualization of drug use, harm reduction access points, and living locations among participants. Our maps illustrate different daily geographies between those who used The Spot and those who did not. Overall, clients of The Spot had activity zones more centralized to downtown Sudbury than those who didn’t use services at The Spot, as the latter had more geographically sparse and distant activity zones. Conclusion: Northern and rural communities face unique barriers to accessing harm reduction services amid Canada’s current toxic, unregulated drug supply crisis. The closure of the only SCS in Sudbury, Ontario, with no replacement service planned in the foreseeable future, has resulted in participants increasingly consuming unregulated drugs in dangerous ways, leading to an increase of stigma, social isolation, and additional barriers to accessing health services. Furthermore, an apparent geospatial mismatch has been identified between the preferred areas for health services among participants and the locations of those services. Overall, we find that Sudbury’s layout of unregulated drug use is unique, fluid, and influences Sudbury’s high unregulated drug mortality rate.
Resistance is Our Heritage: An Archive of Survival and Efforts to Resist Gentrification in Little Jamaica.
(University of Waterloo, 2025-08-11) Pereira, Marcus Thomas
Rooted in my experience as a community member and organizer, this thesis documents the history and resistance of Little Jamaica in Toronto as it continues to face the impacts of gentrification and displacement. Using Black archival practice as my methodology, I draw from oral histories, protest materials, community reports, and digital media to center the voices and experiences of residents, business owners, and activists. Grounded in the framework of racial capitalism, this research understands gentrification as part of a longer history of displacement, extraction, and state neglect targeting Black communities. It traces the development of Little Jamaica through Caribbean migration and examines how planning interventions—particularly the Metrolinx Light Rail Transit construction—have intensified economic pressure, disrupted local business, and contributed to cultural erasure. By amplifying community narratives and mobilizing knowledge for advocacy, this thesis not only documents the fight to preserve Little Jamaica’s cultural identity, but also contributes to broader discussions on gentrification, displacement, and resistance. It highlights the everyday strategies, care networks, and collective organizing that continue to sustain the neighborhood. Ultimately, it seeks to equip residents with knowledge and tools for organizing while challenging dominant narratives of progress and revitalization. At its core, this work affirms that the fight for Little Jamaica is ongoing—and that resistance has always been part of its story.
Turbulence Closure Modeling Using Kolmogorov-Arnold Networks and Bayesian Optimization
(University of Waterloo, 2025-08-11) Kalia, Nikhila
This thesis presents two complementary approaches to improving Reynolds-averaged Navier–Stokes (RANS) turbulence modeling through machine learning and optimization. First, we introduce a realizability-informed framework for training stable and physically consistent machine learning-based anisotropy closures within the tensor basis neural network (TBNN) paradigm. We develop a physics-based loss function that penalizes non-realizable predictions during training, enhancing model stability and generalization. To reduce model complexity and improve interpretability, we replace conventional multilayer perceptrons (MLPs) with Kolmogorov–Arnold Networks (KANs), forming the Tensor Basis KAN (TBKAN) architecture. The TBKAN framework is evaluated across three canonical flows, flat plate, periodic hills, and square duct, demonstrating improved prediction fidelity, stability, and realizability compared to baseline TBNNs and traditional eddy-viscosity models. Second, we explore the application of Bayesian optimization for data-driven calibration of RANS model coefficients, focusing on the generalized k–omega (GEKO) model. The proposed turbo-RANS framework is applied to a converging-diverging channel flow case characterized by adverse pressure gradients and separation. Optimized coefficients yield improved predictions of wall-bounded quantities and streamwise velocity profiles when compared against direct numerical simulation (DNS) and large eddy simulation (LES) references. Together, these contributions address key limitations in existing data-driven turbulence modeling approaches, namely, lack of physical realizability, interpretability, and predictive robustness, while providing practical tools for improved RANS performance in engineering flows. All code and models developed in this work are made publicly available to encourage further research and adoption.
Efficient User Interaction for High-Recall Retrieval: Model Priming
(University of Waterloo, 2025-08-11) Manaam, Abdul
High Recall Information Retrieval (HRIR) tasks, including legal e-discovery, information retrieval test collection construction and systematic review, require finding all, or nearly all relevant documents with the least amount of effort. Past research has shown that Technology Assisted Review (TAR) generally outperforms traditional e-discovery tools, and established that Continuous Active Learning (CAL) performs better than other commonly used TAR tools like Simple Active Learning and Simple Passive Learning. Prior research has also shown that adding search in a CAL-based HRIR tool can slow users down, and restricting access to full documents can speed up the document review process. Our goal was to design a system that provides more autonomy to users without affecting performance. Specifically, we wanted to investigate ways in which search can speed up the document review process. Systems like CAL often go through an initial training phase. We hypothesized that this training phase can be significantly shortened if search is used to seed the model. Moreover, we also created a novel interface that combines search with CAL on a single page. To test our hypothesis and the newly created user-interface, we conducted a user study with 40 participants to investigate five different configurations of an information retrieval system. We found that the addition of search, when preceded with proper user training, can significantly improve precision, performance, user experience and perceived effectiveness of the system. We also found that the newly designed interface, "Integrated CAL" performs comparably to the traditional interface, while providing a more familiar search based interface for users to interact with. Our findings reinforce the importance of hybrid High Recall Information Retrieval systems built on both search and CAL, that provide maximum control to users.
Population trends and behaviours of bats in Maritime Canada
(University of Waterloo, 2025-08-11) Golestaneh, Sepidar
In a rapidly changing world with many drivers of extinction, monitoring wild populations has become of increasing priority to researchers. For many bats (order Chiroptera) in North America, population assessments have become critical in monitoring population trends following disease disturbance, specifically White Nose Syndrome. White Nose Syndrome is a fungal disease that has resulted in mass mortality of many hibernating bat species in eastern North America since its detection in 2006. Given the time elapsed for many areas since White Nose Syndrome introduction, many affected populations have reached an established state where disease-associated mass mortality is not currently observed. As such, the motivation for population assessments has shifted more towards describing patterns of recovery in White Nose-affected populations. However, population assessments for bats, like for other taxa, may be biased and consequently impact the inferences made by researchers. This thesis aims to describe population trends of species affected by White Nose Syndrome in Maritime Canada and assess the impacts of research techniques on study findings. The specific objectives of this thesis were to (1) assess population trends of resident hibernating bats with respect to White Nose Syndrome detection in Maritime Canada, and (2) assess the potential biases of capture surveys on bats, a standard research method used to conduct population assessments and study bat activity.
Fundamental studies for small molecule aptamer selection using capture-SELEX
(University of Waterloo, 2025-08-11) Ding, Yuzhe
DNA aptamers for small molecules hold transformative promise in biosensing, diagnostics, and therapeutics, yet their in vitro evolution has been hampered by incomplete knowledge of the parameters that drive efficient enrichment. In recent years, the development of library-immobilization based method, so called capture-SELEX, has generated over 100 high-quality DNA aptamers for various types of small molecules. Importantly, capture-SELEX allows systematic investigation of fundamental problems in the selection of aptamers. This thesis studies the capture-SELEX platform by dissecting thermodynamic, kinetic, and methodological variables to accelerate the discovery of high-affinity DNA aptamers. Using adenosine/ATP as targets for selection has repeatedly produced the same guanine-rich aptamer motif that was first reported by the Szostak group in 1995. This aptamer has been considered as the adenosine/ATP aptamer by the field. First, by gradually increasing the selection stringency on classical targets (adenosine and ATP), we selected two new aptamers with Kd ≈ 230 nM, 35-fold tighter than that of the classical aptamer sequence. This was achieved through gradual reduction of target concentration from 5 mM to the low-micromolar range. The evolution of the sequence abundance cross different rounds was traced by deep sequencing, and the reason for the previous repeated report of the classical sequence was attributed to its short 12-nucleotide conserved binding regions, whereas the two new aptamers have approximately 16 conserved nucleotides. This study highlights the importance of using low target concentration in order to enrich high affinity aptamers. During aptamer selection, using a lower target concentration tends to favor the enrichment of higher affinity binders, raising the question of whether a practical lower limit exists. Next, we performed three capture-SELEX campaigns using 5 µM, 500 nM, and 50 nM guanine as the target, respectively, to investigate it. Both the 5 µM and 500 nM selections successfully enriched the same guanine aptamer-requiring eight rounds at 5 µM guanine versus 17 rounds at 500 nM guanine. However, the 50 nM selection failed to yield any aptamers. The highest affinity and most enriched aptamer from these selections displayed a Kd of 200 nM, indicating that if the target concentration is much lower than Kd can lead to failed selections. Mutation analysis further revealed a critical cytosine in the guanine binding pocket: substituting this cytosine with a thymine switched selectivity from guanine to adenine. A similar specificity switching was previously seen in the natural guanine riboswitches. These findings define a lower limit for target concentration in capture-SELEX and offer a practical guidance for selecting target levels to isolate high-affinity aptamers. Selection of high-affinity aptamers underpins all downstream applications, yet most protocols emphasize thermodynamic factors-such as target concentration-while overlooking binding kinetics. Third, we performed a library-immobilization selection against ampicillin to dissect these influences. Under typical gravity-flow conditions (1-2 min interaction), a low-affinity aptamer (Kd = 12.7 µM) dominated the enriched pool. In contrast, extending the incubation time to 10 min enriched a higher affinity sequence (Kd = 1.8 µM), differing by only three nucleotides from the weaker Kd aptamer. Systematic comparison of library immobilization efficiency, release fraction, and release kinetics confirmed that dissociation rate from the capture duplex was the primary determinant of the selection outcome. We observed the same kinetic bias in parallel adenosine selections, demonstrating the generality of this effect. Based on these findings, we recommend combining low target concentrations with extended incubation time to favor the enrichment of high-affinity aptamers. This study not only yields a robust, high affinity and selective ampicillin aptamer but also highlights a critical interplay between thermodynamics and kinetics during in vitro aptamer selection. Since 1990, numerous aptamer-selection techniques have been developed, yet quantitative comparisons of their enrichment efficiencies remain scarce. Finally, we evaluated three library‐immobilization SELEX methods, capture‐SELEX, GO‐SELEX, and gold‐SELEX, using a spiked library containing DNA aptamers with varying affinities for adenosine. Using 100 µM adenosine as target, all three methods showed that <1 % of the library was released by adenosine as revealed by qPCR, with gold‐SELEX showing virtually no DNA elution. Deep sequencing of three model aptamers (Ade1301, Ade1304, and the classical adenosine aptamer) revealed 30-50‐fold enrichment in capture-SELEX, whereas GO‐SELEX and gold‐SELEX both yielded enrichment factors below 1, indicating a lack of aptamer enrichment. Blocking the primer‐binding regions improved GO‐SELEX enrichment to ~14 % but still fell far short of capture‐SELEX’s performance. Finally, we compared nonspecific versus target‐induced release and elucidated why capture‐SELEX’s structural-switching mechanism offers superior aptamer enrichment. Overall, capture‐SELEX is a markedly more efficient strategy for isolating high‐affinity aptamers. Collectively, this work establishes a quantitative framework for capture-SELEX-balancing target concentration, kinetic control, and partitioning strategy-to reliably isolate nanomolar-class DNA aptamers for small molecules.
Reverse Logistics Network Design for Additive Remanufacturing
(University of Waterloo, 2025-08-08) Afros, Sheila
The current push for businesses to adopt sustainable supply chain practices contributes to a circular economy, a systems-focused approach designed to allow resources to be used and reused for as long as possible. The Canadian Net-Zero Emissions Accountability Act and the 2030 Emissions Reduction Plan developed by the Canadian federal government aim to regulate the environmental footprint by reducing greenhouse gas (GHG) emissions. Some of these goals can be achieved by designing reverse logistics networks. Reverse logistics is the design of a network for the purpose of collecting end-of-life/end-of-use products and reusing, repairing, refurbishing, remanufacturing, and/or recycling. This thesis proposes four models to design a reverse logistics network: deterministic, multi-period, stochastic, and a game theory model. The models are formulated as bi-objective mixed-integer linear programs. The design of the network is optimized with respect to economic and environmental objectives. The balancing of costs and environmental objectives during the design of the reverse logistics network provides the decision-maker with additional information on how environmental goals can be met. The models aim to determine the optimal locations for remanufacturing facilities and the optimal flow of parts to and from these facilities. The bi-objective models are solved using the weighted sum method, which allows for Pareto-optimal solutions to be analyzed. The models are solved using Gurobi in Python. Deterministic, stochastic, and game theory models are applied to a case study for the remanufacturing of front lower control arms in the automotive industry in Ontario, Canada. The stochastic model is motivated by the uncertainty in the supply of end-of-life vehicles. The stochastic model is solved using the deterministic equivalent. The game theory model complements the aforementioned approach. It facilitates the triangulation of the results. The models select the optimal locations for the remanufacturing facilities. Significantly, all three models applied to the case study produce similar results. The quantitative results demonstrate that the optimal solution based on the case study data is when multiple facilities are located, one facility should be opened in Ottawa and at least one other in the GTA. When only one facility is located, it should be placed in the GTA (either Mississauga or Brampton). Overall, the results reveal that small investments can lead to significant reductions in greenhouse emissions released during transportation. The results can be scaled to design a reverse logistics network for Canada and inform environmental policies. The results and findings of this study may be used to inform policies on the reduction of transportation emissions. The contributions that the thesis makes to the field are: (i) it incorporates greenhouse gas emissions into the models; (ii) it allows decision-makers to compare the results of the three models (deterministic, stochastic, and game theory) applied to the case study; and (iii) it applies the models to real-world data.
Dynamic Modeling, Analysis, and Control of Integrated Electricity and District Heating Systems
(University of Waterloo, 2025-08-08) Muhammad, Muhammad Abuelhamd Mahmoud
Some of the main challenges with more efficient and cleaner energy systems include the development of Integrated Electricity and Heating Systems (IEHSs). Thus, the proposed research explores IEHSs in the context of Microgrids (MGs) and Bulk Power System (BPS), including the integration of Small Modular Reactors (SMRs) to provide both electricity and heat. In this context, the main goal is to study dynamics and control of IEHSs, with an emphasis on District Heating Networks (DHNs) for efficient energy use. Thus, the proposed work aims to support a cleaner energy transition in both remote communities and urban areas through the use of IEHSs, focusing on two key objectives, namely, developing a comprehensive dynamic models of a combined DHN and Electric Power Network (EPN) to study energy exchanges between these systems, and the provision of ancillary services for power grids for both BPSs and MGs. A dynamic DHN and EPN model is first developed for MGs, considering all relevant heating and electricity system details, but especially soil limitations, extreme low temperatures, and piping insulation to minimize heat loss. The accurate sizing of the Heat Pumps (HPs) based on thermal load requirements, weather conditions, and consumer profiles is also discussed, proposing a demand management control to enhance MG primary frequency regulation, which facilitates the integration of variable Renewable Energy Sources (RESs) in power grids. The proposed dynamic models are applied, tested, and validated in a realistic community MG based on a remote community EPN located at Kasabonika Lake First Nation (KLFN) in Northern Ontario. It is shown that the DHN facilitates the proper integration of RESs in isolated MGs through the development of novel EPN control systems. The presented research also studies IEHSs to serve a broader geographical region beyond MGs. Thus, it proposes to investigate the impacts of IEHSs on the dynamic operation of the BPS in collaboration with Ontario’s Independent Electricity System Operator (IESO), using SMRs as the primary source for both electricity and thermal energy demand. For this purpose, a comprehensive dynamic model of a boiling water SMR power plant, incorporating all essential components, including the reactor, steam turbine, and steam flow control system, is first developed to evaluate its practical application in power grids. Unlike previous studies that focus on individual elements, this work proposes the first full system model with direct steam expansion, capturing key thermal-hydraulic dynamics of the reactor, such as pressure variations and two-phase drift velocity, to improve simulation accuracy. The incorporation of a DHN is then studied, demonstrating that the reactor efficiency can be improved by enabling combined heat and power generation and the provision of ancillary services, which collectively would increase the overall performance and efficiency of the BPS. The proposed SMR dynamic models are integrated into a widely used power system analysis tool to assess their dynamic performance and impact on Ontario’s BPS. The results highlight the operational flexibility and frequency regulation potential of the BWRX-300, demonstrating its feasibility for practical deployment in modern power grids while also discussing the potential challenges of integrating it within large-scale power networks. Furthermore, the results of the SMR-based DHN models highlight the strong potential of SMRs for cogeneration applications, demonstrating their operational flexibility in practical settings and illustrating a significant reduction in energy consumption compared to heating systems based solely on HPs, which is the current trend.
Development of Lora Battery-Free Water Leak Detection
(University of Waterloo, 2025-08-07) Brown, Brandon
This study explores the development of a battery-free water leak detection system that leverages energy harvesting sensors and LoRa (Long Range) communication to overcome the limitations of Bluetooth Low Energy (BLE) technology. While BLE offers low power consumption, its range is limited to approximately 50 meters, making it unsuitable for wide area monitoring. In contrast, LoRa provides transmission distances of up to 10 kilometres and excellent indoor wall penetration, enabling broader scalability for smart infrastructure applications. The goal of this research is to integrate LoRa modules into a self-powered sensing platform, eliminating the need for conventional batteries or wired power sources. However, the transition to a battery-free system introduces a key challenge: the LoRa module requires high peak currents, often exceeding 100 mA, for successful activation and data transmission, which electrochemical energy harvesting devices cannot directly supply. Experimental testing showed that single energy harvesting cells failed to activate the LoRa chip due to insufficient current, despite producing sufficient open-circuit voltage. Attempts to connect sensors in series or parallel offered partial improvements but still lacked the required stability and power density to sustain beaconing. These issues were further compounded by mismatched performance between cells and high internal resistance, limiting energy transfer efficiency. To overcome power limitations, the system integrated a DC-DC boost converter and a 100 mF supercapacitor to store and deliver energy in short bursts for LoRa transmission. Enhancements in sensor design and enclosure durability improved activation speed and energy reliability. Electrical tests confirmed efficient energy use and successful LoRa beaconing, demonstrating the feasibility of battery-free, long-range leak detection for scalable, low-maintenance monitoring.
Autoregressive Generative Models for Many-body Physics
(University of Waterloo, 2025-08-07) Teoh, Yi Hong
Many body physics, the study of emergent behavior from the microscopic interactions between countless degrees of freedom, is fundamental to our understanding of the universe. Understanding these systems enable the design and development of everything from materials, pharmaceuticals and even machine learning algorithms. However, to our knowledge, classical simulation of these systems are naturally insufficient and or inefficient. Recent developments in quantum processors herald a new age in the study of many body systems. The ability to process, extract and generalize the information from such devices is pertinent to new discoveries in the field. In recent years, autoregressive generative models have been proven to have remarkable capabilities in a wide range of applications, from machine translation, text summarization to image generation. These models not only allow exact evaluation of the likelihood but also enable independent and identically distributed samples to be drawn from their encapsulated complex joint probability distribution of many degrees of freedom. Additionally, these models have demonstrated emergence, where the model exhibits complex behaviors allowing exceptional performance in a wide range of scenarios. These algorithms are prime candidates for the extraction of information from quantum processors and many-body systems. In current times, data from quantum processors is still rare and expensive, as such we desire the most efficient method to extract information from such limited data. Generative models reveal themselves to be powerful tools in this scenario, achieving higher accuracies with little data. Rich many-body systems typically inhabit higher dimensional spaces compared to the 1-dimensional sequence of autoregressive models. Consequently, a choice is required regarding the traversal of these higher dimensional systems. We explore the effects linked to this choice of traversal in the training of such models. Furthermore, we systematically probe the generalization abilities of autoregressive generative models in a variety of axes, such as in size and parameters, for typical discrete and continuous many-body systems. Finally, Inspired by the recent establishment of correspondence between machine learning and many-body physics, we integrate real-space renormalization group into the model architecture. This integration of a many-body physics technique with machine learning demonstrates considerable promise for the future development of powerful architectures capable of achieving generality.
They decried the bio-politics of illicit substance use: the perspectives of people who use drugs on access to opioid agonist treatment in rural and remote communities of British Columbia
(University of Waterloo, 2025-08-07) Jayathilake, Amreetha
Overdose toxicity claims the lives of 21 people per day on average in Canada, making the overdose crisis a leading public health concern. British Columbia (BC) has one of the highest rates of illicit drug-related overdoses in the country, with rural and remote communities being disproportionately affected by 30-50% increased risk of overdose mortality. Despite this, most research related to illicit drug use focuses on large urban centres and lacks research for the impetus defining increased overdose risk in rural and remote communities. Many people who use drugs (PWUD) are comorbidly diagnosed with opioid use disorder (OUD) for which the current gold-standard of treatment is opioid agonist treatment (OAT). OAT is prescribed by a medical professional and includes medications such as methadone or buprenorphine/naloxone and requires the observation of dose intake by a clinician, pharmacist, or nurse. OAT is said to decrease mortality risk and improve treatment outcomes for PWUD with OUD, albeit, its use and availability is lacking in rural and remote communities, owing to a multitude of intrinsic, extrinsic, and environmental factors, stemming from prohibitionary laws on drugs. Given the unique barriers faced by PWUD drugs in rural and remote settings, the resulting policy and program recommendations of urban centres cannot directly be translated to smaller settings. Accordingly, the primary goal of this thesis work is to understand the experiences of PWUD and their access to OAT in rural and remote communities of the qathet region in BC. This work is complemented with three sub-goals: firstly, to understand the perspective of using a virtual OAT application program to aid in OAT medication adherence; secondly, to acknowledge the individual participant stories in such a way that aids in fighting-back against drug policy; and thirdly, to illustrate the importance of selecting appropriate analytical frameworks that work to establish agency for the people who the research is about. These efforts are demonstrated within the manuscripts included in chapters 4 and 5. Chapter 4 utilizes a content analysis to describe participant experience with access to OAT in the qathet, with a focus on data related to the virtual application, Sonara Health, that was presented as a potential solution to problems identified in data from year-1 interviews. Chapter 5 uses longitudinal data from participant experience with access to OAT in the qathet to describe and exemplar the use of feminist relational discourse analysis (FRDA), an analytic tool rooted in Foucauldian discourse and a feminist epistemology, with a focus on individual and contrapuntal voices. These works utilized a qualitative research design involving a rapid ethnography taking place in 2024 and 2025 and is embedded within a larger longitudinal research project that began in 2022. These works involve a total of 32 PWUD and Indigenous-PWUD who had previous experience accessing OAT in the qathet. Semi-structured interviews were completed with participants in 2023 and 2024 years of the research study. These works utilized a community based participatory research (CBPR) design partnering with community organizations (qathet Community Action Team, Tla’min Health Centre) and involve the principles of ownership, control, access, and possession (OCAP) principles as it relates to data involving Indigenous participants. Taken together, these works establish the overall aim of the thesis: to aid in understanding the rural specific access barriers and facilitators to OAT for PWUD in rural and remote communities of BC, whilst simultaneously strived to use the research to empower PWUD by giving voice to their lived experiences. Accordingly, this thesis adds to the limited body of research regarding experiences of PWUD on OAT in rural and remote communities and presented a novel solution through gaining perspectives on the Sonara app. Additionally, this thesis involved the demonstration of FRDA as a promising analytic approach to showcase research with PWUD, which to our knowledge is the first of its kind. This work highlights the clear need to implement novel approaches to OAT service utilization in rural and remote communities whilst demonstrating the importance of focusing on the participants and their voices as primary to the research.

Browse

Recent Submissions