University of Illinois Urbana-Champaign
Illinois Digital Environment for Access to Learning and Scholarship RepositoryNot a member yet
123813 research outputs found
Sort by
QStore: quantization-aware compressed model storage
Get PDFSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo termsThe student, Raunak Shah, accepted the attached license on 2025-05-01 at 01:17.The student, Raunak Shah, submitted this Thesis for approval on 2025-05-01 at 01:30.This Thesis was approved for publication on 2025-05-06 at 11:52.DSpace SAF Submission Ingestion Package generated from Vireo submission #22138 on 2025-10-19 at 18:11:27Modern applications commonly leverage large, multi-modal foundation models. Such use cases often feature complex workflows that demand the storage and usage of models in multiple precisions. However, to make model development accessible to the average user, model providers opt to maintain separate files for each precision (e.g., INT8, BF16); Hence, naively handling model usage in multi-precision workflows (e.g., downloading and storing each precision separately) can incur prohibitive storage costs. We present QStore, a unified, lossless compression format for simultaneously storing a model in two (high and low) precisions. QStore’s encoding scheme stores a pair of different-precision models with even less storage cost versus storing the high-precision model alone: it compresses the low-precision model, then stores a novel representation of the ’conditional information’ present in the high-precision model but not in the low precision model. Then, for model usage, QStore allows direct access to the low precision model via decoding; if the high precision model is required, QStore recovers it losslessly by applying the stored conditional information onto the decoded low-precision model. We evaluate QStore for compressing multiple precisions of popular foundation models, and show that QStore reduces the overall storage footprint by up to 2.2× (45% of the original size) while enabling up to 1.7× and 1.8× faster model saving and loading versus existing approaches
Innovative method to dietary education for individuals undergoing hemodialysis: Immersive virtual reality low-sodium education and grocery shopping
Get PDFSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo termsThe student, Kaitlyn Pawelczyk, accepted the attached license on 2025-05-01 at 14:33.The student, Kaitlyn Pawelczyk, submitted this Thesis for approval on 2025-05-01 at 16:26.This Thesis was approved for publication on 2025-05-05 at 12:44.DSpace SAF Submission Ingestion Package generated from Vireo submission #22159 on 2025-10-19 at 18:11:30Dietary sodium intake plays a critical role in chronic volume overload (VO) and interdialytic weight gain (IDWG), both of which are associated with adverse cardiovascular outcomes in hemodialysis (HD) patients. Effective management of end-stage kidney disease (ESKD) requires adherence to dietary recommendations; however, the renal diet’s complexity presents significant challenges in patient education, comprehension, and implementation, often resulting in noncompliance. To address this, a virtual reality (VR) educational tool was developed to help individuals on HD in overcoming behavioral barriers that impact sodium intake such as food selection and grocery shopping. This approach enhances nutrition education, focusing on critical components of the renal diet, specifically sodium, due to its impact on IDWG. Through this innovative way of learning, individuals on HD can actively engage with the material while fully immersed in the interactive educational environments. Within these virtual environments, individuals on HD can explore interactive lessons on sodium content, perimeter shopping, and label reading. This program seeks to increase understanding and retention of key dietary concepts, ultimately promoting better compliance with low-sodium dietary recommendations among individuals on HD through this technology-enhanced learning experience
Graph-based microarchitecture exploration: Automated optimization on RISC-V hardware
Get PDFSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo termsThe student, Chentai Yuan, accepted the attached license on 2025-05-02 at 10:05.The student, Chentai Yuan, submitted this Thesis for approval on 2025-05-02 at 10:24.This Thesis was approved for publication on 2025-05-08 at 15:24.DSpace SAF Submission Ingestion Package generated from Vireo submission #22174 on 2025-10-19 at 18:11:34As the demand for computing power continues to grow in the modern world, the chip industry has been driven to develop increasingly advanced processors to meet these growing needs. The microarchitecture of modern processors has become very complex, making it more challenging to achieve sustained improvements in performance and efficiency with each new generation. We explore two novel approaches to optimize the efficiency of modern processors: ArchGen, a design-space exploration framework that models processor microarchitecture using graphs, and MESA, a dataflow mapper that performs dynamic binary translation for reconfigurable accelerators. ArchGen introduces a graph representation of hardware architecture to allow rapid design estimations and iterative optimization. MESA, on the other hand, introduces an on-chip hardware controller that monitors running programs on the CPU and offloads them to an accelerator when suitable. This thesis delivers two major contributions: the development of a composable RISC-V vector processor as a base RTL library for ArchGen and the verification of an AXI4 port within MESA. We demonstrate that ArchGen can use our custom RISC-V processor as a baseline for iterative optimization and hardware synthesis. The addition of an AXI4-compliant port to MESA allows it to easily interface with other devices in a system-on-chip. Both methods are evaluated to test their effectiveness and correctness through hardware simulation, synthesis, and performance modeling
Reduced-order modeling of a cooling system for a variable pole induction motor
Get PDFSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo termsThe student, Holton Miller, accepted the attached license on 2025-05-07 at 20:59.The student, Holton Miller, submitted this Thesis for approval on 2025-05-07 at 21:07.This Thesis was approved for publication on 2025-05-08 at 15:44.DSpace SAF Submission Ingestion Package generated from Vireo submission #22196 on 2025-10-19 at 18:12:21Electrothermal co-design is an approach that has enabled e-machine designers to improve the capabilities of power electronics to meet high performance demands. The United State Department of Energy has established the goal of achieving an 88% reduction in motor and electronics volume in electronic traction drive systems, while also minimizing or eliminating the use of permanent magnets. Induction motors have been identified as viable alternatives to permanent magnet machines that provide the benefits of low cost, high reliability, and ruggedness. Adding pole count as an additional degree of freedom in induction motor design expands the speed and torque range of the motor while also improving power density and efficiency. In this work, the electrothermal codesign for a toroidally wound variable pole induction machine and cooling system is presented. Toroidal windings offer thermal benefits by increasing the available surface area for heat transfer. COMSOL Multiphysics and a Foster thermal equivalent circuit reduced order model are used in thermal design. The reduced order model is used to rapidly predict the maximum temperature rise in the stator windings when subjected to core and winding losses. Steady state operating conditions and transient drive cycles are simulated to assess the performance of cooling jacket designs. It is found the reduced order model is accurate to within 1 °C for both steady state and transient simulations compared to COMSOL. Using these techniques, the cooling jacket is optimized to minimize pumping losses while enabling the IM to safely operate with a large speed and torque range within temperature limits. The final cooling jacket design enables the IM to deliver 110 kW of continuous power and 200 kW of power for 30 seconds. Peak temperatures are also found to be below 115 °C during representative drive cycles. Simulation results are presented for the final, optimized cooling system design
LEU+ to HALEU nuclear fuel cycle transitions and dynamic reactor models
Get PDFSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo termsThe student, Nathan Ryan, accepted the attached license on 2025-05-05 at 13:12.The student, Nathan Ryan, submitted this Thesis for approval on 2025-05-05 at 13:50.This Thesis was approved for publication on 2025-05-07 at 14:47.DSpace SAF Submission Ingestion Package generated from Vireo submission #22210 on 2025-10-19 at 18:12:45Understanding the nuclear fuel cycle is crucial when designing sustainable and efficient nuclear energy systems. This thesis studies timely transition scenarios for fleets of Micro Modular Reactors (MMRs), X-Energy Xe-100s (Xe-100s), and AP1000s where low-enriched uranium plus (LEU+) fuel delays the demand for high-assay low-enriched uranium (HALEU) for the TRi-structural ISOtropic (TRISO) fueled reactors through a greedy, random, and initially random then greedy deployment scheme to meet energy demand growths from the U.S. Department of Energy (DOE) and U.S. Energy Information Administration (EIA). Using the open-source code Cyclus to model fuel cycles and the Monte Carlo code Serpent to perform neutronics calculations for the Xe-100 and USNC MMR, the results show that the reactor deployment scheme impacts the separative work units (SWU) required to meet energy demand. The greedy scheme, which prefers the highest capacity reactors, leads to the most significant increase in SWU for AP1000 low-enriched uranium (LEU), while the random and initially random then greedy schemes result in more consistent increases across fuel types. By evaluating the masses of fresh and used fuel, SWU, the number of reactors, and how well each simulation meets the projected energy demand, this thesis provides a comprehensive understanding of the impact of reactor deployment schemes on the nuclear fuel cycle. Additionally, this thesis examines the computational complexity of reactor fuel trading and removes assumptions about reactor power. The Trading On-Demand (TOD) reactor reduces the number of instructions in a simulation by trading fuel only when needed, while the Dynamic Power Reactor (DPR) allows for flexible power output to mirror historical or projected capacity factors. The results show that improving reactor models and simulating fuel cycle transitions leads to more efficient reactor deployment and fuel cycle design
Black girls radically hope: an exploration of how Black girls practice radical hope in their everyday lives
Get PDFSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo termsThe student, Whitney Clarke, accepted the attached license on 2025-05-09 at 10:47.The student, Whitney Clarke, submitted this Thesis for approval on 2025-05-09 at 11:09.This Thesis was approved for publication on 2025-05-09 at 12:40.DSpace SAF Submission Ingestion Package generated from Vireo submission #22284 on 2025-10-19 at 18:14:01Black girls’ experiences of oppression are uniquely shaped by the intersection of racism and sexism, often leading to emotional distress and barriers to envisioning or pursuing future possibilities. Research shows that the middle school years are a critical developmental period in which identity is shaped by family, peer, and school messages (Mims & Williams, 2020; Umana-Taylor et al., 2014). While previous scholarship has highlighted the protective benefits of affirming Black girls’ identities and the positive impacts of hope on health outcomes (Butler-Barnes, 2017; Chang et al., 2019), little is known about how Black girls specifically enact radical hope—the steadfast belief in the collective power to create liberatory futures in the face of oppression (Mosley et al., 2020). This study addresses this gap by exploring the central questions: a) How do Black girls conceptualize radical hope? b) How do Black girls practice radical hope in their everyday lives? I used reflexive thematic analysis (RTA) alongside an abductive analytic approach to gather and interpret the data. RTA, as outlined by Braun and Clarke (2006, 2019), is an iterative, flexible method that requires the researcher to continually reflect on their assumptions, values, and positionality throughout the research process. As a Black woman and former Black girl, my lived experience shaped how I engaged with and interpreted the data, particularly as I was welcomed into the participants’ school community. To complement RTA, I employed abductive analysis (Timmermans & Tavory, 2012; Thomas, 2006), which blends inductive openness with deductive structure—allowing me to identify emergent patterns while remaining anchored in conceptual frameworks. This dual approach ensured that both participant meaning and theoretical insight were honored in the analysis. Findings revealed that Black girls’ conceptualizations of radical hope are multidimensional, deeply relational, and grounded in community, identity, and resistance. They defined three dimensions of hope: myself, us, and my people. Their practices of radical hope emerged through three central themes: engaging in safe spaces, helping others, and navigating barriers to hope. Hope, as defined and embodied by the girls, is not just a mindset but an active, political, and cultural expression—rooted in the self, in solidarity with others, and in defiance of oppressive structures. By building sisterhood, resisting dehumanization, and envisioning futures beyond societal limitations, the girls constructed a new pathway toward hope that centers connection and collective empowerment. This study extends existing frameworks of hope (Bernardo, 2010; Callina et al., 2015; Ginwright, 2010; Snyder, 1995) and contributes to calls for youth-centered, asset-based research (Agger et al., 2024; Young, 2021). It also offers developmental insight into the radical hope framework—previously applied only to adults—and informs educational practices, future therapeutic practices, intervention development, and community programming for Black girls. Ultimately, this work positions Black girls not only as knowledge holders, but as architects of liberatory futures
Nucleation and evolution of graphitic particles in atmospheric pressure microwave plasma
Get PDFSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo termsThe student, E. Hays, accepted the attached license on 2025-05-09 at 12:15.The student, E. Hays, submitted this Thesis for approval on 2025-05-09 at 12:20.This Thesis was approved for publication on 2025-05-09 at 15:11.DSpace SAF Submission Ingestion Package generated from Vireo submission #22286 on 2025-10-19 at 18:14:02Graphene’s exceptional electrical, thermal, and mechanical properties have driven two decades of research toward scalable, high-quality synthesis methods for industrial applications. Traditional techniques such as chemical vapor deposition (CVD) offer good material control but are often limited by substrate constraints and low production rates. This study investigates an alternative approach to synthesis using atmospheric pressure microwave plasmas to grow graphene-like carbons without the need for a substrate, thereby enabling volumetric, gas-phase growth and higher production rates. Key reactor parameters, including tube geometry, methane flow rate, and input power, were systematically varied to study their effect on particle formation and material quality, allowing for reactor optimization towards maximizing production rate while yielding good quality carbon. Brunauer-Emmett-Teller (BET) surface area analysis and Raman spectroscopy revealed that straight tube geometries in the place of taper or step tube geometries and higher plasma power favor the formation of more crystalline, lower-defect graphene structures, while increased methane flow generally led to higher defect densities and larger particle sizes. Spatially resolved sampling was achieved for the carbon grown in the reactor via probe arms, and showed that mean particle diameter increased with distance from the plasma core, supporting a nucleation-in-the-bulk, growth-at-the-boundary model consistent with existing literature. These results highlight the sensitivity of plasma-grown carbon to reactor conditions and support the continued development of atmospheric plasma systems for scalable, tunable graphene synthesis. Future work may explore the incorporation of moving substrates or low-pressure variants to further refine the balance between quality and throughput for next-generation graphene-based technologie
Bridging the gap: Understanding SQL learning challenges through quantitative analyses and qualitative insights
Get PDFSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo termsThe student, Sophia Yang, accepted the attached license on 2025-02-04 at 19:51.The student, Sophia Yang, submitted this Dissertation for approval on 2025-02-04 at 20:13.This Dissertation was approved for publication on 2025-02-07 at 11:54.DSpace SAF Submission Ingestion Package generated from Vireo submission #21633 on 2025-10-19 at 18:17:21As the demand for SQL experts continues to rise, teaching SQL effectively has become increasingly important. Previous studies on SQL education have identified learning challenges, analyzed student errors, and developed learning tools. Building on this foundation, my research provided additional directions to enhance educators’ understanding of SQL learning challenges and explored tools to teach more effectively. Leveraging large-scale data collected from a public institution’s database course, I conducted a series of quantitative studies with students’ data. I developed use cases for pattern-mining techniques to track students’ submission attempts for their SQL assignments. With advanced pattern-mining techniques, such as Levenshtein Edit Distance, hierarchical clustering, and sequence alignment, I examined and captured students’ problem-solving behaviors under various conditions. Furthermore, previous studies on general education suggested several factors that can affect students’ learning ability, including study time, course modality, collaborative environments, etc. Thus, I conducted several studies to explore factors influencing students’ ability to learn SQL, including the sequence in which query languages are introduced, study time, assignment types, error feedback, and collaborative environments. Lastly, since students often struggle to solve SQL problems due to a lack of error feedback, I also investigated the use of generative AI tools to provide more effective and personalized feedback. To complement these quantitative findings, I designed and conducted a qualitative think-aloud study to gain deeper insights into students’ SQL challenges. Based on the combined findings, I provided actionable recommendations to improve teaching strategies and curriculum design. Instructors should adopt pattern mining tools to identify students needing extra support, emphasize non-timed SQL problems, provide syntax drills, integrate generative AI for error feedback, and explore an online flipped-classroom model with in-person office hours. Additionally, studying the inclusion of Parsons problems could further scaffold student learning in SQL
The representation of biracial occupation children
Get PDFSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo termsThe student, Felix Ayanbode, accepted the attached license on 2025-04-14 at 12:56.The student, Felix Ayanbode, submitted this Dissertation for approval on 2025-04-14 at 12:59.This Dissertation was approved for publication on 2025-04-17 at 16:19.DSpace SAF Submission Ingestion Package generated from Vireo submission #21762 on 2025-10-19 at 18:18:14The First World War marked a turning point for several European nations that participated in the war. Two of those nations are France and Germany. The events that took place during and after the war were critical to a small community of mixed-race children who were born as a result of relationships between black French soldiers, who were stationed in Germany after Germany lost the First World War, and white German women. These children were perceived as a shame and humiliation to the German nation. While research has been done on the racialized experiences of these children as well as the inhumane treatment they endured in the Third Reich, little research has been done on how these children have been represented in works of literature and how they represented themselves through oral and written histories. My dissertation investigates how these biracial occupation children are represented by others and how they decided to represent themselves to reclaim their agency. My dissertation, The Representation of Biracial Occupation Children, examines how these individuals grappled with matters of representation, nationality, identity, and belonging during the Third Reich. After Germany lost World War I, the Rhineland was occupied by French troops, including Black soldiers from France’s African colonies, primarily from West and North Africa. These soldiers were part of the “tirailleurs sénégalais,” a corps recruited from Senegal and other French colonial territories. Some of these soldiers engaged in relationships with white German women. Children born from these relationships were called “Rhineland Bastards” by the Nazis and suffered immense hardships at their hands, ranging from exclusion from public spaces to forced sterilization. In order not to perpetuate Nazi terminology and ideology, I choose to call them “biracial occupation children.” In four chapters, my dissertation delves into how these biracial children were portrayed by Nazi discourse and how they navigated fluid spaces and embraced multiple identities. Through three texts by and two interviews with surviving biracial occupation children, written in French, German, and English, I examine the relationship between race, space, and place in the Third Reich: Hans Massaquoi’s Destined to Witness (1999), Didier Daeninckx’s Galadio (2010), Esi Edugyan’s Half-Blood Blues (2011), and two interviews, one of Hans Hauck and Fasia Jansen. While Destined to Witness is an autobiographical work, Galadio and Half-Blood Blues are novels. Both types of works try to give a voice to these excluded voices. My project is part of the broader effort to decenter not only German but European studies more broadly. I aim to recenter the experiences of black individuals – those who were biracial occupation children and those who were perceived and treated as such like Hans Massaquoi—by exploring questions of nationality, citizenship, and belonging relevant not only to the Nazi period in Germany but also to the time when and the place where these texts were written, including the US, Canada, France, and Germany. Drawing on postcolonial theory, critical race theory, and memory studies, this dissertation argues that biracial occupation children operated in realms of fluid identities due to the racialization of space and the spatialization of race. This dissertation explores the triangulated and transatlantic nature of their experiences
Spectrality in the works of Christoph Ransmayr
Get PDFSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo termsThe student, Zachary Hader, accepted the attached license on 2025-04-16 at 14:58.The student, Zachary Hader, submitted this Dissertation for approval on 2025-04-16 at 15:05.This Dissertation was approved for publication on 2025-04-17 at 16:22.DSpace SAF Submission Ingestion Package generated from Vireo submission #21800 on 2025-10-19 at 18:18:21My dissertation examines novels by contemporary Austrian author Christoph Ransmayr (1954– ) to demonstrate that the literary-cultural figure of the “specter” is more multifaceted than previously thought, and is still helpful for understanding the ways in which the past and future are mediated in the present, perhaps especially in a hypermedial, information-flooded era. My readings are informed by the “spectral turn” in critical theory, which began with the publication of Jacques Derrida’s Spectres of Marx (1992). Scholars in myriad fields have since used a concept of spectrality based at least loosely on Derrida’s work to better understand textual relationships between past, present, and future; see e.g., Julian Wolfreys, Meera Atikinson, and Grace M. Cho. I modify current generalized notions of spectrality and argue that there are at least three types of specters in literature: 1. the ghost, i.e., the living embodiment of a dead person/thing; 2. the meta-specter, who appears through mediated means, e.g. through writing; 3. the mytho-specter, who exists in myth or in other narrations of collective cultural memory. These spectral types all appear in Christoph Ransmayr’s novels and help us better understand his postmodern negotiation of the relation between pasts and presents. Scholars of the specter should find these distinctions useful as it can provide a way for them to distinguish between various types of specters as well as show the ways in which they may be similar. My reading of Atlas of an Anxious Man (2012) demonstrates how specters enable a better understanding of the text’s relationship with history, mourning, and myth. Turning to The Flying Mountain (2006), I examine linguistic spectrality and its relation to present re-representations of past trauma. I then examine Ransmayr’s historical fiction, including The Dog King (1997), to show how the specter influences and is influenced by history. Finally, I examine his latest novel, The Lockmaster (2021), to illuminate the relation between spectrality and the Gothic. These readings all aim to further the understanding not just of the cultural figure of the specter, but also of Ransmayr’s works. Scholars of Ransmayr as well as of spectrality should find this project useful in showing the various ways in which Ransmayr utilizes the figure of the specter and the ways in which messages are passed throughout differing temporalities in his works