33760 research outputs found
Sort by
College Park Parkrun
No full textFinal report for HNUH269T: Building Community: Showing Up for Social Change (Spring 2025). University of Maryland, College ParkThis project explores the vibrant community formed around the College Park Parkrun, a free, weekly 5K event that fosters health, inclusion, and social connection. As part of the University of Maryland’s HNUH269T course, students engaged directly with the Parkrun community to understand the elements that make it successful and welcoming.College Park Parkru
Codes for Data Retrieval and Node Repair in Graphical Networks
No full textThis dissertation investigates the problem of efficient data recovery in distributed storage systemsrelying on erasure codes, when the connections between individual nodes of the system are constrained by
a connected graph. In this model, when a node fails, i.e., the data stored in it becomes lost or unavailable,
the other surviving nodes in the network send information, which is a function of their local stored data,
along the edges of the graph to repair the failed node. We show that savings in communication complexity
can be attained if the intermediate vertices along the path process the information rather than simply relay
it toward the failed node.
We derive information-theoretic bounds on the amount of information communicated between thenodes in the course of the repair. Moreover, we show that the lower bound on the information exchange
is achievable by modifying codes from the class of Minimum Storage Regenerating (MSR) codes to per-
form intermediate processing. Our analysis extends to both deterministic connected graphs and random
graphs, where we derive conditions on the system parameters that support recovery of the failed node with
complexity lower than relaying.
In the second part of the thesis, we extend our study to general regenerating codes. We derive alower bound on the repair bandwidth and formulate repair procedures with intermediate processing for
several algebraic families of regenerating codes. We also address the problem of data retrieval in the
communication-constrained setting, deriving lower bounds and optimal protocols.
In the final part, we consider regenerating codes with nonuniform contribution for node repair ongraphs. We begin with deriving information-theoretic lower bounds for communication complexity of
repair and propose code constructions and repair schemes that attain these bounds. As the main conclusion,
we show that a combination of nonuniform contributions and intermediate processing can further reduce
the communication complexity. Additionally, for repair on graphs in the presence of adversarial nodes that
can introduce errors during repair, we construct codes that support simultaneous intermediate processing
and error correction
EDUCATION WITHOUT SCHOOLS: WITNESSING BLACK LIBERATION THROUGH ALTERNATIVE EDUCATION SPACES AND PRACTICES
No full textThis dissertation explores how Black families in the United States are reclaiming educational sovereignty by opting out of traditional school systems that often function as sites of antiblackness and dehumanization. Through three qualitative studies, I examine homeschooling, unschooling, and self-directed education as acts of resistance and care, rooted in Black radical traditions and guided by the pursuit of Black freedom. Across the studies, I engage theoretical frameworks including motherwork, marronage, and BlackCrit, while grounding the work in an understanding of antiblackness as both structural and ontological.The first study draws on interviews with Black homeschooling mothers, highlighting the multidimensional labor of motherwork in resisting state violence and nurturing Black children's liberation. The second study analyzes a podcast by a Black unschooling mother, positioning unschooling as a contemporary form of marronage—an intentional disengagement from the state in pursuit of autonomy, self-determination, and joy. The final case study examines a Black self-directed education community in Maryland, uncovering the tensions, joys, and healing practices involved in sustaining educational spaces centered on Black sovereignty.
Together, these papers argue that traditional schools are incapable of fully honoring the possibilities inherent in Black children. Instead, Black families are creating alternative education spaces where their children can thrive. This work contributes to the growing body of research on Black educational resistance and raises critical questions about what it means to raise free people in a world not currently built for Black freedom
Social Media and the Epistemic Environment: How Individuals Navigate the Impact
No full textThis dissertation investigates three questions related to how individuals can and should navigate the current epistemic environment in the age of social media.
In Chapter 1, “Epistemic Decentralization,” I explore what is philosophically and ethically unique about the current epistemic environment. I propose that with the advent of social media, we are experiencing an epistemic phenomenon called epistemic decentralization. I argue that there are several decentralized epistemic resources that profoundly affect our capacity as a knower. Moreover, I argue that epistemic decentralization brings both significant costs and benefits.
In Chapter 2, “Fake News and the Duty to Reserve Reliance,” I investigate the duty individual users have in the face of the challenges arising from the advent of social media, specifically the problem of fake news. I argue that individuals have at least a pro tanto duty to reserve reliance on their beliefs concerning important social news they encounter on social media. More specifically, this duty entails that individuals have one of the two following duties when they share information online. If users can verify the truthfulness, they have the duty of verification. If they cannot, they have the duty of disambiguation.
In Chapter 3, “Social Media and Open-Mindedness,” I argue that while it is difficult to identify experts and fake news is rampant on social media platforms, open-mindedness should still be regarded as an important virtue for social media platform users for two reasons. First, genuine open-mindedness not only requires an outward-looking willingness to be open and impartial when evaluating new arguments and evidence; it also requires an inward-looking awareness of the arenas within which one is capable of responding to reasons. Moreover, as genuine open-mindedness requires the inward-looking component, it can help them develop other epistemic and non-epistemic virtues, such as intellectual humility and empathy, which can foster a better epistemic environment
Integrated Planning of Residential and Commercial Electric Vehicle Charging Infrastructure: A Strategic Bi-Level Optimization and Queuing Framework Approach
No full textElectric vehicles (EVs) have gained significant popularity, becoming an attractive option for cleaner transportation systems. The market adoption of different types of EVs in the USA has grown significantly over the years. A critical challenge for this expanding market is the limited charging infrastructure available in both residential and commercial spaces. This research aims to study the optimal number, placement, and management of charging stations required to accommodate EVs at both residential and commercial locations, considering varying market penetration rates, household adoption of charging infrastructure, and the constraints posed by power network capacities.The developed model is applied to the Baltimore Metropolitan Statistical Area (MSA) to determine the optimal number and distribution of charging stations. This study integrates real-world Origin-Destination (OD) trip data with census data to investigate the relationship between residential and commercial EV charging infrastructures. Furthermore, advanced queuing theory models are employed to analyze and manage congestion at commercial charging stations, evaluating system performance indicators such as waiting times, blocking probabilities, and station utilization under realistic, stochastic demand conditions.
Scenario analysis illustrates the shifting landscape of charging infrastructure demands for both current (0.09%, 0.75%, and 1.61%) and future market penetration rates (5%, 7.5%, and 10%), reflecting Maryland's current adoption levels and future growth projections driven by policy initiatives and technological advancements in EV infrastructure. The findings reveal a clear inverse relationship between the availability of residential charging facilities and the necessity for commercial chargers. Utilizing a bi-level optimization model, this research balances investor interests with EV user satisfaction by maximizing profits and minimizing user charging costs. Additionally, the queuing simulation highlights critical operational insights, illustrating how variations in arrival rates, service times, and charger availability impact overall system efficiency and user experience.
These combined insights emphasize the importance of integrating residential and commercial charging infrastructure planning with congestion management strategies. Policymakers, utility providers, and infrastructure investors can utilize these findings to formulate sustainable EV charging strategies, considering market penetration rates, household charger adoption, optimal State of Charge (SoC) management, and operational efficiency through queuing management. This comprehensive framework provides robust guidance for stakeholders aiming to develop resilient, sustainable, and efficient EV charging networks, promoting greater EV adoption and contributing to environmental sustainability
COMPARISON OF EFFECTS OF PROCESS TECHNOLOGY-DERIVED INPUT PARAMETERS OF DIE-LEVEL FAILURE MODELS
No full textIn integrated circuits, time-dependent dielectric breakdown, hot carrier injection, andelectromigration are primary die-level wear-out failure mechanisms. Failure models for these
mechanisms can be used to assess and compare the parts based on their ability to withstand these
failure mechanisms. The failure models require electrical input parameters: gate voltage, drain
current, and current through interconnect, as well as dimensional input parameters: gate oxide
thickness, gate length, and die metallization dimensions. These input parameters are unavailable
in traditional documentation, such as datasheets and application notes. As a result, part users face
difficulty using the failure model for part assessment. This thesis presents methodologies to obtain
die-level electrical and dimensional input parameters for individual parts. The approach developed
to find the input parameters uses the process technology information of a part and literature on
process technology. The electrical input parameters for the time-dependent dielectric breakdown
and hot carrier injection failure model are determined from transistor-level voltage-current
characteristic curves provided in the literature on process technologies. The methodologies to
determine the electrical input parameters are developed by utilizing transistor circuit information
and the associated characteristic curves.
Part manufacturers use different technologies and design rules, leading to differences in inputparameters such as die-level dimensions and electrical and environmental loads. These variations
affect the ability of parts to withstand die-level failure mechanisms. Therefore, a die-level
comparative assessment should be performed to compare and select the parts. Comparative
assessment refers to quantifying and comparing the influence of die-level input parameters on
time-to-failure of parts for individual die-level failure mechanisms using simulation-based design
of experiments. Identifying the parameters that affect the part’s time-to-failure using a simulation-based design of experiments supports decision-making for derating considerations, acceptable
manufacturing variations, and part selection. This thesis provides guidelines for extracting input
parameters for die-level failure mechanisms and a methodology to perform comparative part
assessment based on the application load condition of the system
PENALIZED STATISTICAL MODELS FOR PATHWAY-BASED TWAS AND HIGH-DIMENSIONAL MEDIATION ANALYSIS
No full textHigh–throughput genomics and neuroimaging now generate thousands of correlated molecular and imaging features for each participant, presenting unprecedented opportunities—and methodological challenges—for causal and genetic discovery. This dissertation develops two complementary statistical frameworks that address key limitations of classical tools when confronted with such high-dimensional data. In Chapter 1, I introduce that genome-wide association studies (GWAS) have pinpointed numerous SNPs linked to human diseases and traits, yet many of these SNPs are in non-coding regions and hard to interpret. Transcriptome-wide association studies (TWAS) integrate GWAS and expression reference panels to identify the associations at gene level with tissue specificity, potentially improving the interpretability. However, the list of individual genes identified from univariate TWAS contains little unifying biological theme so the underlying mechanisms remain largely elusive. These limitations motivate a unified framework that not only identifies trait-associated genes through multivariate TWAS, but also traces how their effects propagate through intermediate brain features to influence clinical outcomes—thus naturally extending gene-level association analysis into high-dimensional mediation analysis. Mmediation analysis is a fundamental tool for elucidating causal mechanisms in complexsystems. However, the emergence of high-throughput biological and neuroimaging technologies has introduced multivariate exposures and mediators of increasing dimensionality, rendering classical approaches inadequate.
In the Chapter 2 we propose a novel multivariate TWAS method that Incorporates Pathway or gene Set information, namely TIPS, to identify genes and pathways most associated with complex polygenic traits. We jointly modeled the imputation and association steps in TWAS, incorporated a sparse group lasso penalty in the model to induce selection at both gene and pathway levels and developed an expectation-maximization algorithm to estimate the parameters for the penalized likelihood. We applied our method to three different complex traits: systolic and diastolic blood pressure, as well as a brain aging biomarker white matter brain age gap in UK Biobank and identified critical biologically relevant pathways and genes associated with these traits. These pathways cannot be detected by traditional univariate TWAS + pathway enrichment analysis approach, showing the power of our model. We also conducted comprehensive simulations with varying heritability levels and genetic architectures and showed our method outperformed other established TWAS methods in feature selection, statistical power and prediction. The R package that implements TIPS is available at \url{https://github.com/nwang123/TIPS}.
In Chapter 3, we propose a novel aggregation-based mediation framework that simultaneously models and selects high-dimensional multivariate exposures and mediators. Our method identifies sparse linear combinations of variables in each domain that jointly maximize the mediated effect, defined as the product of exposure–mediator and mediator–outcome effects. To estimate these low-dimensional aggregators, we formulate a bi-convex objective function integrating residual sum-of-squares penalties from standard mediation submodels, a structured mediation-enhancing term, and -penalties that induce sparsity. The resulting optimization problem is solved efficiently using an alternating direction method of multipliers (ADMM) algorithm with block coordinate updates.
Through extensive simulations, we demonstrate that our method achieves superior performance in recovering true mediators and estimating mediation proportions across a wide range of signal strengths, noise levels, and correlation structures. Compared to existing methods—including MMP, Pathway Lasso, sparse PCA mediation, and the Directions of Mediation framework—our approach exhibits higher selection accuracy and reduced bias, particularly in challenging high-correlation regimes. We further illustrate the utility of the method in a real data application involving the mediation of smoking behavior through neuroimaging features, revealing biologically meaningful pathways linking gene expression in the nucleus accumbens to structural and functional brain indices implicated in addiction. These findings highlight the potential of our framework for integrative mediation analysis in high-dimensional biomedical studies
EXPLORING THE INTERPLAY OF FOOD INSECURITY AND CORRESPONDING COPING BEHAVIORS AMONG COLLEGE STUDENTS: IMPLICATIONS FOR INTERVENTION THROUGH RESOURCE MANAGEMENT BEHAVIORS
No full textFood insecurity affects 20 to 50% of US college students, which is significantly higher than the national average of 12.8%, with significant impacts on students' physical health, mental wellbeing, and academic performance. Despite its prevalence, key research gaps persist. Current USDA food security survey modules may inadequately capture college students’ unique experiences, despite their widespread use in this population. Additionally, there is limited understanding of long-term effects among students, and a scarcity of evidence-based solutions and targeted policies to address campus food insecurity. This study aims to address key gaps in literature by examining the coping strategies adopted by food-insecure college students, particularly how these strategies vary by food insecurity severity, and by exploring how resource management behaviors including food planning and shopping routines, food literacy and financial behavior, are associated with food security levels. Using a cross-sectional design, data were collected from 373 college students through an online self-administered questionnaire hosted on the Qualtrics platform. Participants were eligible if they were undergraduate students enrolled in a four-year college in Maryland. Descriptive statistics, Kruskal-Wallis and ANOVA tests, and ordinal logistic regression were used to analyze the data.
Results indicated that as food insecurity worsened, students not only increased the frequency of their coping strategies but also progressed to more extreme measures. Initially, food insecure groups relied on strategies such as asking friends and family for food or money to buy food, buying the cheapest food available, avoiding expensive foods such as fruit and vegetables, eating at places in which you pay what you can and eating as much as possible when food is available. However, as food insecurity worsens, students reported reliance on extreme coping strategies such as choosing between food and essential expenses (rent, utilities, medicine), implementing stricter food shopping budgets resulting in limited diet variety, selling personal possessions to buy food, and stretching food to last longer. Additionally, frequent use of these coping strategies significantly predicted very low food security. The present finding suggests that certain coping strategies employed specifically by very low food secure students could serve as more sensitive indicators for identifying students in urgent need, potentially offering greater precision than current USDA food insecurity assessment modules.
Also, financial behavior encompassing day-to-day money management and financial planning emerged as a significant predictor of food insecurity. Paradoxically, students who demonstrated stronger financial behaviors were more likely to experience higher levels of food insecurity. Our analysis also revealed that although not significant predictors in the ordinal logistic model, food insecure students were more likely to discard food based on date label expiration and demonstrated lower food literacy compared to their food secure counterparts.
Overall, this study fills an important research gap by mapping how coping strategies evolve across varying levels of food insecurity, offering insights for developing context-specific tools andtargeted interventions for college students. To our knowledge, this is also the first study to examine key aspects of the utilization dimension of food insecurity among college students concerning resource management—specifically food literacy, planning, shopping routines, and financial behaviors—to identify potential areas for intervention. Findings highlight the urgent need to address food insecurity in this population, with resource management emerging as a promising intervention point
LATTICE-MATCHED GROWTH OF TOPOLOGICAL SEMIMETAL CADMIUM ARSENIDE ON VIRTUAL TERNARY III-V SUBSTRATES
No full textTopological materials are a recently discovered electronic class which exhibits electronic band structures that deviate from the behavior of conventional electronic materials. These topological materials exhibit conical bands and host unique band elements such as Dirac and Weyl nodes through which phenomena such as the quantum Hall effect, quantum anomalous Hall effect, and the quantum spin Hall effect may be realized. These effects are protected against the destabilizing effects of thermal fluctuations through crystal symmetry-derived protection. The topological semimetal Cd3As2 is an excellent candidate for the realization of novel devices through magnetic alloying because of its demonstrated high electron mobility (>10,000 cm2/V•s). This dissertation details the molecular beam epitaxy growth and characterization of topological semimetal Cd3As2 thin films that are lattice-matched to ternary III-V alloy virtual substrates and optimized for high crystalline quality. This dissertation will first review the deposition of Cd3As2 and the relevant calculations made to determine the suitability of virtual substrates for its lattice matched-growth. Therefore, the stability of ternary III-V alloy buffers and their interfaces with Cd3As2 are discussed. Thermal and spinodal decomposition ranges for the ternary III-V alloys are also determined to fully understand them as virtual substrates for Cd3As2. Cd3As2 thin films are grown with thicknesses targeting 30-100 nm. Three ternary III-V alloy systems GaInSb, AlInSb, and InAsSb are grown on both GaAs and AlAs buffers, using GaAs (001) as a substrate. The structural properties of these high-quality, epitaxially grown single crystals are characterized using in situ pyrometry, reflection high-energy electron diffraction, high-resolution X-ray diffraction, X-ray reflectivity, atomic force microscopy, cross-sectional scanning electron microscopy, and tunneling electron microscopy. The optimal processing parameters for producing these Cd3As2 thin films and their virtual substrates are described in detail for easy application in future molecular beam epitaxy. Assessment of the electronic properties of Cd3As2 are conducted using Van der Pauw geometry Hall measurements at 2K with external magnetic fields ranging up to 14 T. The electronic attributes of resistivity, carrier density and mobility, and the presence of the quantum Hall effect are probed and reported. The potential epitaxy of different magnetic Cd3As2 alloys and proximal magnetic films are explored through calculations of Gibbs free energy. Sm is chosen as the candidate for future work seeking the quantum anomalous Hall effect in magnetic Cd3As2
ESSAYS ON CONSUMER DEMAND, POLICY, AND MARKET STRUCTURE
No full textThis dissertation explores how consumers respond to changes in policy, product offerings, and firm behavior in the U.S. food and beverage industry. Through three empirical essays, it examines how consumer choices adapt to new taxes, product introductions, and distribution changes—and how these adjustments vary across demographic groups and market environments.
The second chapter analyzes the impact of sugar-sweetened beverage taxes in five U.S. cities using NielsenIQ Retail Scanner and Consumer Panel data. A staggered Difference-in-Differences (DID) analysis finds that taxed soda prices increased by 15%, while sales declined by 17%. However, city-level effects vary: Philadelphia saw the largest decline (30%), while other cities showed smaller or insignificant effects. Pass-through rates ranged from 20% to 80%, with higher rates linked to larger sales declines. Effects are heterogeneous across retail channels and package sizes. Food stores saw the steepest reductions, while drug stores exhibited smaller or even positive effects. Larger package sizes experienced the sharpest declines, while single-serving products showed small changes. Beyond direct reductions, consumers adjusted through cross-border shopping and substitution. Taxed soda purchases in bordering cities increased in Philadelphia and Oakland, and cross-price elasticity estimates suggest that diet soda, water, and snacks are substitutes. Household data confirms these shifts, showing increased purchases from discount stores. These findings indicate how consumer responses shape soda tax effectiveness, moderating its overall impact.
The third chapter quantifies the welfare gains from the introduction of Greek yogurt using a random coefficients discrete choice model. Drawing on data from 2007 to 2019, the analysis shows that Greek yogurt initially appealed to higher-income, college-educated, and younger consumers, with significant heterogeneity in preferences in 2007. These demographic differences diminished over time as Greek yogurt became mainstream. Counterfactual analysis of removing Greek yogurt from consumer choice sets in 2011 and 2015 suggests sizable welfare losses. While losses are similar across income quartiles, they vary substantially by education: non-college-educated households experience much larger reductions in consumer surplus, indicating that they had come to rely more heavily on Greek yogurt despite not being the earliest adopters.
The fourth chapter examines the impact of mergers between large beverage firms and small brands on product distribution. Using brand-level sales data and cross-market variation, the analysis finds that acquisitions significantly expand distribution: acquired brands become 23 percentage points more likely to be nationally distributed, and the number of markets in which they are sold increases by 233% on average. Effects vary by pre-merger brand characteristics and similarity to the acquirer’s portfolio