33760 research outputs found
Sort by
Optimization and Trajectory Analysis of Morphing Waveriders
No full textA waverider is a flight vehicle that creates a shock wave attached to the leading edge, which leads to high lift-to-drag ratio at a design Mach number. A morphing waverider changes the lower surface in order to keep the shock wave attached to the leading edge at a range of Mach numbers. The classic waverider design method creates the vehicle shape with sharp leading edges from a known flow field. This is unrealistic for vehicle design, first because real-world leading edges are rounded, and second because this method limits the design space of possible geometries to a small subset that generate flow fields that can be calculated analytically.
This objective of this fundamental research was to bypass the current limitations facing round-leading-edge waverider optimization by including a reduced-order blunt-leading-edge model and a Computational Fluid Dynamics (CFD) algorithm in the optimization process. A new morphing methodology was created that could apply additional off-design features to the leading edge and lower surface of planar shock-derived waveriders. An analytical model for off-design waveriders was developed and used to optimize morphing waverider geometries alongside the CHAMPS+ CFD tool. The impact of morphing on waverider range was evaluated using a direct collocation trajectory optimization algorithm.
While the analytical model and CFD tool predicted similar performance trends, the CFD based method was better suited to capture the complex features experienced by round-leading edge morphing waveriders. For both methods, the optimized morphing waverider geometries were predicted to give lift-to-drag ratio and range increases over non-morphing waveriders. To provide additional insights into the effects of morphing on round-leading-edge waverider performance, further morphing optimization iterations, as well as the addition of constraints which take into account real-world low-cost access to space considerations, are recommended
Building Reliable AI under Distribution Shifts
No full textMachine learning models are increasingly deployed in real-world settings where distribution shifts—differences between training and deployment data—can significantly impact their reliability. These shifts affect models in multiple ways, leading to degraded generalization, fairness collapse, loss of robustness, and new safety vulnerabilities. This dissertation investigates how to build reliable AI under distribution shifts, providing theoretical insights and practical solutions across diverse applications.
We begin by studying generalization under distribution shifts, exploring how model invariance affects performance. We introduce a theoretical framework that quantifies the role of data transformations in shaping generalization, providing insights into selecting transformations that improve model robustness in shifted environments. This foundation also extends to fairness, where we examine how pre-trained fair models fail when deployed in new distributions and propose a method to transfer fairness reliably under distribution shifts.
Next, we focus on robust perception and AI-generated content under shifting distributions. We investigate how models interpret visual information, showing that contextual reasoning can help mitigate spurious correlations and improve robustness under domain shifts. We also assess the reliability of AI-generated content, revealing how image watermarks, designed for provenance tracking, often fail when subjected to real-world distortions and adversarial attacks. To address this, we introduce a comprehensive benchmark for evaluating watermark robustness, providing a framework for improving their reliability.
Finally, we turn to safety in large language models (LLMs) and investigate how distribution shifts in training and deployment introduce new vulnerabilities. We analyze false refusals in safety-aligned LLMs, demonstrating that misaligned decision boundaries lead to excessive conservatism at test time. We also explore retrieval-augmented generation (RAG) models, showing that despite their promise, they can introduce new safety risks when deployed in settings for which they were not originally trained. Our findings highlight critical gaps in existing AI safety evaluations and emphasize the need for new methods tailored to evolving AI architectures.
By addressing generalization, robustness, and safety under distribution shifts, this dissertation contributes to a deeper understanding of these challenges and provides practical strategies for improving AI reliability in real-world deployment
CODE SMOOTHING, UNIFORM DISTRIBUTIONS, AND APPLICATIONS
No full textThis dissertation investigates the problem of approximating uniform distributions over Hamming spaces, with applications to secure communication, randomness extraction, and computational hardness reductions. The central technique used in this study is code smoothing, which involves applying additive noise to a uniformly random codeword, producing an output distribution that closely approximates uniformity.
We characterize the code rate thresholds required to achieve strong uniformity guarantees, quantified using R{\'e}nyi divergence.
Our results show that random linear codes, as well as structured families like Reed-Muller and LDPC codes, are highly effective at smoothing.
Building on these code families, we develop practical and effective coding schemes for use in wiretap channel settings.
We also study uniformity guarantees achievable through hash functions, using techniques closely related to code smoothing.In particular, we show that certain classes of linear and -universal hash families provide strong uniformity guarantees, quantified by R{\'e}nyi divergence.
This enables the extraction of uniform distributions to meet modern cryptographic requirements that impose stringent security guarantees.
We further apply smoothing techniques to investigate cryptographic reductions, with a focus on the computational hardness of the Learning Parity with Noise (LPN) problem---a foundational problem in lightweight cryptography. While prior work has explored reductions from the decoding problem to LPN to establish its hardness, our work sharpens this connection by delineating the parameter regimes where such reductions are viable, thus clarifying both the possibilities and the limitations of this approach.
Together, these contributions demonstrate how smoothing serves as a unifying analytical tool across coding theory, cryptography and computer science enabling both practical constructions and theoretical insights
Crystalline Topological Invariants in Invertible Fermionic States and Fractional Chern Insulators
No full textTopological phases of matter have long been a central theme in condensed matter physics. When subject to certain symmetries, these phases are classified by topological invariants that often take fractional, quantized values. Despite significant theoretical advancements, major open questions remain regarding the computation of these invariants in Chern insulators or fractional Chern insulators with crystalline symmetry.
The first part of this dissertation establishes the necessary theoretical background, defining symmetry operators, geometrical measures, and topological actions used to determine crystalline topological invariants.
The second part focuses on extracting these invariants numerically for invertible fermionic states in multiple ways. Specifically, we calculate the discrete shift , the electric polarization , and the invariants . We discuss the properties of these invariants, such as how and determines the universal charge response of crystalline defects and boundaries, and how they depend on an origin in real space. Concrete numerical methods are demonstrated using the Hofstadter model, which exhibits non-zero Chern number and magnetic field. These invariants, together with several established invariants in the literature, gives a complete classification of ground states and reveal novel colorings of the Hofstadter butterfly.
The third part of this dissertation extends the calculation of crystalline topological invariants to fractional quantum Hall states, such as the 1/2-Laughlin state constructed by projecting parton states onto the same position. We show that the numerically extracted values of these invariants align with theoretical predictions from conformal field theory and -crossed braided tensor categories
RACIAL DISPARITIES IN CESAREAN BIRTH AND POSTPARTUM HEMORRHAGE: THE ROLE OF US HEALTH SYSTEM POLICIES AND PRACTICES
No full textThough lifesaving, cesarean birth imposes unnecessary risks when not indicated. Cesarean rates rose sharply from 1996 to 2009 alongside maternal morbidity and mortality. In response, states implemented policies to reduce unnecessary interventions. While overall cesarean rates stabilized, the Black-White disparities in cesarean birth and maternal morbidity widened. Some research implicated insurance reform in growing these disparities.
This dissertation aims to:1) describe trends in Black-White disparities in postpartum hemorrhage (PPH) corresponding with cesarean birth,
2) deconstruct the factors contributing to the disparity in PPH, and
3) examine the effects of Medicaid policy reform on trends in cesarean birth and PPH.
Aim 1 examines racial disparities in PPH management and severity over time. After analyzing 7.6 million hospital delivery admissions (2002–2021), results show growing cesarean disparities since the early 2010s. Black individuals experience PPH with blood transfusion or severe outcomes, especially with cesarean more often than White individuals, while being less likely to receive conservative interventions. Despite advances in care, severe PPH morbidity has not declined.
Aim 2 uses 2.2 million delivery admissions (2016–2021) and applies Oaxaca-Blinder decomposition to quantify the contribution of disparate risk factors (e.g., cesarean birth) to the Black-White disparity in severe PPH. The results showed that cesarean birth accounts for up to 16% of the disparity, while anemia and delivery in predominantly Black-serving hospitals contribute up to 50%.
Aim 3 uses difference-in-differences analyses of 6 million birth certificates (2009–2019) to examine the effect of Medicaid payment reform policies (e.g., nonpayment; pay-for-performance) on cesarean birth and a proxy for PPH. Despite heterogeneity across states, cesarean births increased by 0.9 percentage points (ppt) among multiparous Black individuals (95% Confidence Interval (CI): 0.4-1.4) and 1.6 ppt for first births (CI: 0.6-2.6) in nonpayment versus control states. Results for PPH were inconsistent and unreliable.
This dissertation finds that insurance policies likely contributed to the growing racial disparities in cesarean birth, and the disparity in cesarean explains a substantial portion of the PPH disparity. Findings underscore the role of structural racism in obstetric outcomes, emphasizing the need for evidence-based policies to dismantle systemic barriers and foster maternal health equity
DISTRIBUTED COMPUTING AND UNSUPERVISED DEEP LEARNING FOR ANALYZING HUMAN TRAVEL BEHAVIORS USING BIG TRAJECTORY DATA
No full textHuman travel behaviors, which refer to the movement of either individuals or groups across space and time, has been a research focus for the understanding of not only the behaviors themselves, but also the interaction between humans and their environment. In recent years, with the development of information and communication technologies, passively collected trajectory data (i.e., sequences of sampled geographic coordinates with timestamps collected by GPS sensors in devices such as mobile phones and vehicles) has become more available. This new data type has provided unprecedented opportunities for research in human travel behaviors offering advantages of broad spatial scale, large data volume, and fine granularity compared with traditional actively collected data, such as census data and survey data. This dissertation focused on exploring how advanced research methods, including distributed computing and unsupervised deep learning, can be extended and applied to examine human travel behaviors using passively collected trajectory data. In this dissertation, three studies are described that examine three different topic that contribute to our understanding of vehicular travel behaviors. The first study involved developing a distributed research framework based on Apache Spark and Sedona to estimate traffic speed and speeding across a state-wide road network and multiple road types using passively collected mobile device data. The study analyzes spatio-temporal patterns in traffic speed and speeding behaviors in the state of California and examines differences in patterns from different road types such as freeways and residential roads. The second study developed a research framework that combines a rule-based distributed parking extracting module and an LSTM-autoencoder model to extract parking trajectories and automatically classify them into different categories (e.g., direct parking and cruising for parking) using in-vehicle trajectory data. Using datasets collected in the Washington, DC metropolitan area, this study further investigates the driving patterns and spatial distribution of cruising trips, identifying areas where parking demand may exceed supply. The third study in the dissertation proposed a research framework that is comprised of a distributed contour plot-building module (designed to capture speed changes compared with free-flow and historical averages), a W-net-based semantic segmentation model to segment the non-recurrent congestion (NRC) impact areas in these plots, and a postprocessing module to ensure that the identified propagation patterns follow the law of shockwaves using passively collected in-vehicle trajectory data. Based on detected NRC impact areas, propagation of NRC over actual road networks in the Washington, DC metropolitan area was examined to identify how these types of congestion move through these road networks. The contributions of this dissertation include providing cutting-edge distributed computing and deep learning methods to make full use of passively collected trajectory data for human travel behavior analyses and provide feasible means to overcome the challenges brought by the large data volumes, broad spatio-temporal ranges, and lack of ground truth labels. The integration of passively collected trajectory data and new research methods have provided new insights into human travel behaviors over large spatial scales and fine granularities, and lay the foundation for future research on human mobility
THE EFFECT OF DIGITAL LIBRARIES IN RURAL MALAWIAN SECONDARY SCHOOLS: IMPLICATIONS FOR THE FUTURE OF AFRICA
No full textThis exploratory case study investigates the impact of offline digital libraries on teaching and learning in rural Malawian Community Day Secondary Schools (CDSSs). Grounded in Afro-futurism and self-efficacy theory, the research examines the implementation of an offline digital library as an educational resource in five CDSSs. The study addresses digital libraries’ impact on attitudes toward teaching and learning, the effects of access to resources through offline digital libraries, and the that factors influence commitment to implementing digital resources. Through semi-structured interviews with students and teachers, the thematic analysis revealed two main themes: 1) an impact on the participant’s attitudes toward teaching and learning (primarily mind/perspective shifts in school and community culture and behavior changes in participants in their ability to access to new tools and aptitudes, and 2) uncovers the many visions of personal and national life and motivations that influence commitment to a digital future for Malawi. The study highlights the transformative potential of offline digital libraries in bridging the educational gap in resource-limited settings and underscores the importance of teacher training, community engagement, and localized content to enhance the effectiveness of digital education tools. By leveraging existing technologies, rural Malawian schools can empower students and teachers to become future-oriented, self-reliant individuals capable of contributing to local and global development. This research contributes to the growing literature on digital education in sub-Saharan Africa and offers practical recommendations for policymakers and educators aiming to improve educational outcomes in underserved rural communities
IN SEARCH OF OUR GRANDMOTHERS’ GARDENS: ROOTED IN BLACKNESS, DANCING THROUGH ERASURE, AND BLOOMING BEYOND RESISTANCE
No full textThis paper presents multifaceted research conducted over the span of three years, pertaining to my personal endeavor as a Black dance artist, choreographer, and educator. Honoring generational customs and culture while embracing new understandings, I embark on a journey of evolving, developing, and learning within the world of Black dance making and storytelling
MOVING MOUNTAINS: A CASE STUDY OF TEACHER ENTRY AND RETENTION IN THE REMOTE HIGHLANDS OF UPPER SVANETI, GEORGIA
No full textMountain ecosystems and mountain residents play a critical role in environmental sustainability, in both mountainous regions and in the lowland communities that are reliant upon their resources. However, mountainous regions experience significant social, economic, and environmental challenges, and mountain residents frequently experience marginalization due to the increased poverty and decreased resources and opportunities common in these communities, as well as their physical distance and isolation from urban centers of economic and political power. Among these challenges is the provision of quality education in mountainous areas, including difficulties recruiting and retaining effective and qualified teachers. Thus, this qualitative case study explores the challenges teachers face in mountainous regions and the factors that impact teacher entry and retention from the perspectives of educators living and working in communities throughout the historical-geographic high mountainous region of Upper Svaneti, Georgia. This study was informed by teacher recruitment and retention theories, geographical perspectives and spatial theories, the sociological concepts of Critical Rural Theory and urbanormativity, and social-ecological conceptualizations of mountain communities. It employed semi-structured interviews and focus groups with 14 teachers and school leaders representing six communities and schools throughout Upper Svaneti and used a combination of inductive and deductive approaches to data analysis.
The findings reveal myriad challenges experienced by teachers in mountainous communities, illuminate the ways teachers perceive living and working in mountainous areas to differ from teaching in urban or lowland areas, and identify key factors that contribute to teacher entry, retention, and attrition decisions in rural, mountainous areas. Additionally, the use of geographical concepts, Critical Rural Theory, and social-ecological conceptualizations of mountain regions highlights the importance of attending to spatial dynamics in educational research and policymaking, providing deeper insight into the complex realities that impact teachers’ paths in rural and remote mountainous areas
SCENIC DESIGN FOR THE OPERA - THE MERRY WIVES OF WINDSOR
No full textThis thesis serves as a documentation of the scenic design process for Otto Nicolai's Die Lustigen Weiber von Windsor (Nicolai 1849), which was produced by the School of Music at the University of Maryland—College Park.The production process and scenic design documentation for this performance are included in this thesis. The basis for this scenic design is these documents.
Every component served as a means of conveying technical requirements and design concepts to the director, other designers, and artisans working on the project. A final reflection of the scenic design and production process is included, along with research photos, set sketches and renderings, photographs of ¼" scale models, drafting plates, paint elevations, a props book, and a props list that lists furniture and hand props