http://www.rssboard.org/rss-specification 720 https://escholarship.org/uc/ucsc_etd/rss Recent eScholarship items from UC Santa Cruz Electronic Theses and Dissertations Fri, 15 May 2026 11:10:50 +0000 https://escholarship.org/uc/item/3rw5x7nk As multimodal artificial intelligence systems become increasingly embedded in everyday technology, there is a growing need to design human-centered AI agents, supporting and amplifying human capabilities rather than replace them. This dissertation investigates how to build human-centered multimodal AI agents, framing human-centeredness as an agent-level objective that requires both accessible, assistive interaction and reliable, trustworthy behavior across physical and digital environments. This dissertation explores two complementary dimensions of human-centered agent design. The first focuses on enhancing accessibility through conversational and interactive agents that assist users in everyday tasks. We study both embodied and digital settings in which agents reduce physical and cognitive burdens via natural language interaction, including hands-free drone control, navigation assistance in unfamiliar environments, and interactive access to complex graphical user interfaces.... https://escholarship.org/uc/item/3rw5x7nk Wed, 6 May 2026 00:00:00 +0000 Fan, Yue https://orcid.org/0000-0002-2357-6809 https://escholarship.org/uc/item/33s115hj The investigation of hyperthermals, or short-lived periods of extreme warming, over the course of Earth’s history is crucial to understanding the effects of climate change on the planet’s global carbon cycle, climate, ecosystems, and biogeochemical cycling. Geologic sedimentary archives record a number of such hyperthermals in the Cenozoic era. The most extreme hyperthermal on record is the Paleocene-Eocene Thermal Maximum (PETM) approximately 55 million years ago. The PETM is characterized by a negative carbon isotopic excursion (CIE) and a negative shift in oxygen isotopic composition indicative of global warming of 5℃. The warming triggered an abrupt shift in deep ocean circulation and deep-water formation recorded by the carbon isotope composition of benthic foraminifera and by seafloor carbonate dissolution patterns. This evidence suggests the pattern of globally overturning circulation underwent a major reorganization eventually returning to its original state as the planet... https://escholarship.org/uc/item/33s115hj Thu, 30 Apr 2026 00:00:00 +0000 Holo, Johanna https://escholarship.org/uc/item/97z569b3 Game balance is a term widely used among players, researchers, and designers of games. It is a concept that feels vitally important to how we make and play games, but when we try to define it or implement it, we seldom get the same definition twice. Balance appears differently to whoever is judging it, but as researchers and designers, we still must translate this element of game design into technical practice. It is also an expensive and time-consuming subject, one that requires a constant loop of playtesting and design iteration through nearly the entirety of the game development process. This work seeks to focus our understanding of balance while offering procedural methods to increase speed or improve quality when performing balancing tasks in game design and research. It accomplishes this by offering a taxonomy of balance along with a generic design framework that can be used to apply balancing strategies in any game context. In addition, it provides a catalog of balancing... https://escholarship.org/uc/item/97z569b3 Fri, 24 Apr 2026 00:00:00 +0000 Shields, Samuel Michael https://orcid.org/0000-0001-6372-5656 https://escholarship.org/uc/item/74r4d4c5 Generative AI services often gate access behind per-user subscriptions, API keys, and manual endpoint configuration. These are barriers that often scale with the number of users and applications on a network. Saturn, the central contribution of my thesis, provisions AI the way printers and speakers already provision themselves: through Multicast DNS (mDNS) and DNS-based Service Discovery (DNS-SD), network protocols that ship on every major operating system. With Saturn, AI endpoints register under the service type saturn. tcp.local. and every device on the network discovers them without accounts, credentials, or configuration files. This thesis contributes the Saturn protocol specification, six reference implementations across four languages, and an on-device deployment on a consumer-grade wireless router. https://escholarship.org/uc/item/74r4d4c5 Fri, 24 Apr 2026 00:00:00 +0000 Perrello, Joseph Lorenzo https://escholarship.org/uc/item/9k2121t4 Expression of late G1 phase cyclins is the critical molecular event that marks commitment to enter the cell cycle. In budding yeast, late G1 phase cyclins initiate and sustain growth of a new daughter bud, so their expression also marks the start of a new growth phase during the cell cycle. Expression of late G1 phase cyclins is influenced by nutrient availability – cells growing in poor nutrients progress through late G1 phase with lower levels of late G1 phase cyclins. However, little is known about how or why nutrients modulate expression of late G1 phase cyclins. Here, we investigated the signals that control expression of the late G1 phase cyclin Cln2. We discovered that nutrients modulate expression of Cln2 primarily via post-transcriptional mechanisms that influence Cln2 phosphorylation and turnover. Nutrient modulation of Cln2 protein expression requires a TORC2-Pkc1-MAP kinase signaling axis. Furthermore, expression of Cln2 is closely correlated with bud growth and required... https://escholarship.org/uc/item/9k2121t4 Thu, 23 Apr 2026 00:00:00 +0000 Zebarjadi, Navid https://escholarship.org/uc/item/9fp6x23s The pursuit of strategies to modulate challenging and undruggable protein targets has fueled interest in therapeutic modalities that are larger and more complex than traditional small molecules. Macrocyclic peptides possess structural complexity allowing effective modulation to protein targets where small molecules often fail. Synthetic accessibility of macrocyclic peptides enable more tunable modification than biological modalities to improve the pharmacological properties including membrane permeability. Combinatorial and encoding power of DNA-encoded libraries allows easy construction of a big library with an unprecedented number of members and effective identification of ligands by affinity-based selection. The integration of DEL with peptide chemistry accelerates the discovery campaign of macrocyclic peptide drugs. In this study, we aimed to apply DEL technology to facilitate the development of membrane permeable peptide therapeutics. To achieve this goal, we are adopting... https://escholarship.org/uc/item/9fp6x23s Thu, 23 Apr 2026 00:00:00 +0000 Zhang, Panpan https://orcid.org/0000-0003-0081-7320 https://escholarship.org/uc/item/994983j7 Anonymous communication systems hide who is talking to whom, not just what is said. However, existing systems are either vulnerable to traffic analysis attacks---attacks where adversaries observe and correlate the network traffic of users---or rely on unrealistic and unenforceable assumptions about how users behave. Worse, existing theory cannot model traffic analysis attacks, and consequently cannot distinguish between systems secure and insecure against traffic analysis nor inform the design of traffic analysis resistant systems. We make several contributions toward our goal of practical anonymity systems that resist traffic analysis. First, we develop the only formal framework for describing the security of systems against traffic analysis attacks, allowing us to quantitatively describe and compare the security of all existing works. Second, leveraging this framework, we identify a property, input/output independence, that distinguishes between systems that are and are... https://escholarship.org/uc/item/994983j7 Thu, 23 Apr 2026 00:00:00 +0000 Fredrickson, Kyle https://orcid.org/0000-0002-0947-2659 https://escholarship.org/uc/item/98h1m637 Advanced driver-assistance systems (ADAS) are commonly organized as modular pipelines that transform raw sensor measurements into low-level actuation commands through perception, planning, and control. While learning-based methods have achieved state-of-the-art performance in perception and environment modeling, the planning layer remains a key bottleneck for reliable autonomy. Highway driving in particular requires long-horizon reasoning and socially aware interaction with multiple actors, while also producing smooth and dynamically feasible motion that can be tracked by classical controllers. This thesis focuses on decision-making and planning for highway autonomy using simulation ground truth at both the object and sensor levels. We study the problem through two complementary simulation environments: the high-fidelity CARLA simulator for motion planning and continuous trajectory generation under realistic vehicle dynamics and road geometry, and the lightweight HighwayEnv... https://escholarship.org/uc/item/98h1m637 Thu, 23 Apr 2026 00:00:00 +0000 Moghadam, Majid https://orcid.org/0000-0002-6186-9645 https://escholarship.org/uc/item/8qx9r4wv This work documents the implementation of a Hysteresis-Based Reinforcement Learning (HyRL) algorithm for obstacle avoidance with quadrotors subject to adversarial noise on the position estimate. To the author’s knowledge, this is the first time this algorithm has been applied on a real vehicle. Additionally, we will present a distributed control architecture that switches seamlessly between software-in-the-loop (SITL) and hardware-in-the-loop (HITL) configurations. The capability to switch between simulated and physical environments allows for rapid development in software, followed by verification and validation on real hardware. The architecture presented in this paper supports parallel execution of mixed software and hardware experiments, with the potential to share data on a federated, high-throughput messaging bus. Our initial experiment presents a scenario where a quadrotor must traverse a space with a virtual object blocking its path. The quadrotor control architecture... https://escholarship.org/uc/item/8qx9r4wv Thu, 23 Apr 2026 00:00:00 +0000 Rodriguez, Ryan Andrew https://escholarship.org/uc/item/8p57d0zv The Vera Rubin Observatory will soon begin its ten year survey, the Legacy Survey of Space and Time (LSST). Rarities and true novelties can be expected in the coming deluge of data. As a worked example, this thesis focuses on the potential of the Rubin Observatory to detect one such novelty: gravitational microlensing by a cosmic superstring.Cosmic superstrings produce a distinctive gravitational lensing signature; a transient doubling of the flux. As part of this thesis, a detection algorithm is developed and applied to $\sim6.6$ million real light curves from the NOIRLab Source Catalog (NSC). Twenty-five candidate events are found. These are shown to be due to contamination by neighboring sources in crowded regions and additional filtering criteria are introduced. The algorithm's efficiency is measured by injecting simulated events onto real light curves. The final algorithm achieves an efficiency of approximately 75%.In addition, a tool is developed to quantify the sensitivity... https://escholarship.org/uc/item/8p57d0zv Thu, 23 Apr 2026 00:00:00 +0000 Shestakov, Adrian https://escholarship.org/uc/item/8nh81926 Vision encoders have demonstrated significant performance gains in visual understanding, generation and multimodal reasoning. These improvements are primarily attributed to the scaling of data, model capacity, and compute. However, this progress is becoming less accessible due to a lack of transparency in data curation and training recipes. In combination with the high compute requirements of large-scale pre-training, these factors hinder independent reproducibility. In this dissertation, we democratize large-scale visual encoder training by developing compute-efficient, reproducible training recipes for video encoders, vision-language models (VLMs), and multimodal large language models (MLLMs). First, we challenge the common belief that scaling necessarily requires proportionally more resources. Specifically, we show that decoupled pre-training separates key factors such as space/time and token length, and learns strong priors first. This design yields dramatic efficiency... https://escholarship.org/uc/item/8nh81926 Thu, 23 Apr 2026 00:00:00 +0000 Li, Xianhang https://orcid.org/0009-0001-9536-1161 https://escholarship.org/uc/item/7rm083zr Acts of physical force against others are generally considered to be wrong. However, many people use, accept, and even celebrate force throughout childhood and into adulthood. For instance, people use physical force during contact sports, rough-and-tumble play, or while giving a friend a high five. A critical developmental task for young children is to learn how to distinguish between “good” and “bad” acts of force. The present research addressed an understudied question about how preschoolers learn about the permissibility of physical force. Using a naturalistic observational design, the overall goal was to chart preschoolers’ acts of force, and how those acts were responded to by their teachers and peers. Findings suggest that acts of physical force are common in preschoolers’ lives and elicit systematically positive and negative responses from peers and teachers. Positive and negative responses of peers and adults may provide a mechanism by which children slowly learn to distinguish... https://escholarship.org/uc/item/7rm083zr Thu, 23 Apr 2026 00:00:00 +0000 Baxley, Charles Paul https://escholarship.org/uc/item/748357tn Alzheimer’s disease (AD) neuroimaging presents three recurring methodological challenges: making reliable predictions when some modalities are missing, generating 3D brain MRI under clinically meaningful conditions, and learning MRI representations that separate disease-related signals from age-related and acquisition-related variation. This dissertation develops one method for each of these challenges in the context of AD-focused imaging analysis. First, we introduce a compositional, graph-based framework for multimodal AD detection. The framework represents datasets as nodes and learnable models as directed edges in a computational graph, allowing end-to-end predictors to be assembled from modular components when different combinations of MRI, PET, cognitive, genetic, and derived imaging features are available. Second, we propose a clinically conditioned 3D VAE-GAN for brain MRI synthesis, paired with diffusion-driven sampling in clinical feature space to generate realistic... https://escholarship.org/uc/item/748357tn Thu, 23 Apr 2026 00:00:00 +0000 Mashhadi, Najmeh https://escholarship.org/uc/item/73w5m5h2 This dissertation is about the relation between perception and cognition. Broadly speaking, I am interested in pursuing the question “how is seeing related to thinking?” In order to understand this relation, I will be focusing on three interrelated questions: 1. How should we understand the nature of perception and its contents? 2. How does perception result in and justify perceptual belief? And 3. Does cognition effect perception?Some philosophers claim that there are certain laws that restrict what kinds of things we can perceptually represent. Those laws do not apply, however, to beliefs. To be a representationalist is to hold that there is a similarity between perception and belief. If this is the case, why do the laws apply to one kind of mental state, but not the other? In Chapter 1, I will make the case that the puzzle is not a puzzle for representationalists in general, but only for some forms of representationalism that hold excessive analogies between perception and... https://escholarship.org/uc/item/73w5m5h2 Thu, 23 Apr 2026 00:00:00 +0000 Groth, Philip Douglas https://orcid.org/0009-0004-3269-5932 https://escholarship.org/uc/item/5g35v26x Modern edge computing increasingly relies on heterogeneous System-on-Chip (SoC) architectures. These chips tightly integrate general-purpose CPUs with various specialized accelerators, including GPUs, FPGAs, and AI accelerators, all under a shared memory architecture. Although these shared-memory SoCs enable more efficient communication and data sharing between different processing units, they are notoriously difficult to program and tune due to architectural diversity across vendors and asymmetric compute capabilities within each SoC.This dissertation introduces Redwood and BetterTogether, two frameworks that rethink CPU-accelerator collaboration on heterogeneous SoCs. Redwood targets a class of algorithms termed traverse–compute, that combine irregular tree traversals with dense leaf-level computation, e.g., Nearest-Neighbor Search and Barnes–Hut algorithm. It addresses the efficient mapping of these algorithms onto heterogeneous systems by exploiting the architectural strengths... https://escholarship.org/uc/item/5g35v26x Thu, 23 Apr 2026 00:00:00 +0000 Xu, Yanwen https://orcid.org/0009-0001-6750-9798 https://escholarship.org/uc/item/56x610qj Accumulating conservation stressors, including increasing environmental variability and human-wildlife conflict, pose significant threats to marine predator populations and the ecological, social, and economic systems they support. Addressing these multifaceted conservation challenges requires interdisciplinary and collaborative approaches that holistically integrate insights across multiple fields and perspectives from broad interest-holder groups. Ultimately establishing a more comprehensive understanding of the stressors facing marine predators, and how we can design effective solutions. This dissertation connects insights across scales and disciplines by combining physiology, biogeography, and targeted conservation to develop new perspectives on conservation threats and solutions of marine predators. In Chapter 1, I investigated post-dive recovery timelines in bottlenose dolphins (Tursiops truncatus) and beluga whales (Delphinapterus leucas), demonstrating that changes... https://escholarship.org/uc/item/56x610qj Thu, 23 Apr 2026 00:00:00 +0000 Nazario, Emily Corinne https://orcid.org/0000-0003-2372-8742 https://escholarship.org/uc/item/4xn7r46m Neurodegenerative diseases are a class of incurable conditions that cause irreversible neuronal loss and cognitive decline through the pathological actions of intrinsically disordered proteins (IDPs). Alzheimer’s disease (AD) is the most common form of dementia and the sixth leading cause of death in the United States. The prevailing amyloid cascade hypothesis implicates the formation of extracellular amyloid-beta (Aβ) plaques and subsequent intracellular neurofibrillary-tau tangle accumulation as primary drivers of AD pathogenesis. Aβ is a 38-42 amino acid long intrinsically disordered, aggregation-prone peptide that forms soluble β-sheet rich oligomers that grow to mature insoluble plaques. Over the last decade, the cellular prion protein (PrPC) has been shown as the highest affinity binding receptor and a critical mediator of Aβ oligomer toxicity in neurons. The interaction between PrPC and Aβ has been implicated in amyloid accumulation, suppression of long-term potentiation,... https://escholarship.org/uc/item/4xn7r46m Thu, 23 Apr 2026 00:00:00 +0000 Smart, Amanda https://orcid.org/0000-0002-2138-2680 https://escholarship.org/uc/item/4n30b0tw This dissertation investigates comparatives and superlatives in Japanese from a crosslinguistic perspective. The central question concerns how natural languages encode comparison. Drawing on new empirical evidence from Japanese, this dissertation argues that the variation observed in comparatives and superlatives can be derived from a unified semantic core. Specifically, comparison in natural language is built on a single primitive ordering relation, ER, which constitutes the core meaning of both comparatives and superlatives. Apparent variation across these constructions arises from two independent sources: (i) the elements with which ER combines, and (ii) differences in syntactic derivation.Within comparatives, this dissertation shows that the surface differences between phrasal and clausal comparatives arise from distinct syntax, while both constructions are based on the same primitive ER. Phrasal comparatives are argued to involve ellipsis within the standard phrase, although... https://escholarship.org/uc/item/4n30b0tw Thu, 23 Apr 2026 00:00:00 +0000 Tamura, Jun https://escholarship.org/uc/item/4mb1q7b6 Understanding reproductive energy requirements is essential for predicting individual fitness and population viability. The beluga whale (Delphinapterus leucas), with its polar distribution, aquatic lifestyle, and challenges associated with studying the species in the wild, has made these requirements difficult to quantify. Here we assessed reproductive energetic costs of individual belugas using long-term husbandry records, which provided unique insights into the energetic demands of different life-history stages. We compiled caloric intake data from ten female beluga whales housed at two zoological institutions to compare daily energy requirements across reproductive stages. Early lactation was the most energetically intensive period, resulting in a 2.7-fold increase in energetic demand compared to nonpregnant periods. Translating this into prey demands in the wild using the Beluga Fish Calculator (John et al., 2024) demonstrated that meeting daily prey requirements by lactating... https://escholarship.org/uc/item/4mb1q7b6 Thu, 23 Apr 2026 00:00:00 +0000 Reynolds, Amelia Hope https://orcid.org/0000-0003-2511-3530 https://escholarship.org/uc/item/4cb928zh Fruit wetness duration (FWD) is an important measurement that can help farmers with disease forecasting, as it is a determining factor for pathogen infection and pesticide use. While there is already commercially available electronic wetness sensors implemented in crop management systems, they are unable to mimic the wide range of surface textures that influence the wetness duration for different crops. To investigate how we can measure more accurate FWD measurements, my team and I developed a bio-mimetic sensor that closely replicates the surface properties and hydrophilicity of strawberries using flexible electrode fabrication techniques and a molding process involving various silicones. Measurements and qualitative analysis of the wetting behavior of the sensor in comparison to a real fruit and commercial sensor suggests that the bio-mimetic sensor can better represent the drying time of a real strawberry. https://escholarship.org/uc/item/4cb928zh Thu, 23 Apr 2026 00:00:00 +0000 Feng-Liu, Willie https://orcid.org/0009-0004-5057-9126 https://escholarship.org/uc/item/40b5686x Trace metals play essential roles in marine ecosystems as micronutrients and as tracers of ocean circulation and biogeochemical transformation. Their vertical distributions reflect complex interactions among biological uptake, particle transport, remineralization, and chemical partitioning between dissolved and particulate phases. Although trace metal cycling research has expanded significantly in recent decades, processes occurring within the mesopelagic ocean remain under-constrained despite the region’s central role in connecting surface production to deep ocean reservoirs. This dissertation investigates trace metal cycling in the mesopelagic and examines the observational and collaborative structures that enable its study. First, analyses of compiled and newly generated particulate cadmium and phosphorus data evaluate anomalous subsurface cadmium cycling observed across ocean basins. Variable regeneration of cadmium relative to phosphorus reproduces part of the observed... https://escholarship.org/uc/item/40b5686x Thu, 23 Apr 2026 00:00:00 +0000 Laubach, Allison https://orcid.org/0009-0006-2924-8802 https://escholarship.org/uc/item/3dx2469z Sepsis is a life-threatening condition characterized by a dysregulated host response to infection, often leading to widespread inflammation, tissue damage, and organ failure. Despite advances in supportive care, sepsis remains a major clinical challenge, with the Centers for Disease Control and Prevention reporting that one in three patients who die in hospitals are affected by it. Survivors frequently experience long-term cognitive and physical impairments driven by immune dysregulation, including lymphocyte apoptosis and cellular reprogramming of innate immune cells, underscoring the urgent need for deeper mechanistic insights into the pathways that govern inflammation and immune dysfunction. Macrophages are central mediators of the innate immune response and play a critical role in orchestrating inflammation during bacterial infection. Upon sensing microbial components such as lipopolysaccharide (LPS), a component of the outer membrane of gram-negative bacteria, macrophages... https://escholarship.org/uc/item/3dx2469z Thu, 23 Apr 2026 00:00:00 +0000 Escalona, Diana https://escholarship.org/uc/item/2rb8b3x9 This dissertation examines how financial market structure shapes macroeconomic resilience and long-run growth, combining historical evidence from the Great Depression with contemporary demographic dynamics. The first two chapters exploit variation in intrastate branch banking regulation across U.S. states between 1925 and 1941 to identify the causal effects of branch banking deregulation on bank behavior and local economic outcomes. Using novel historical datasets linking bank-level balance sheets to county-level retail sales, I show that liberalizing branching restrictions strengthened bank capitalization, shifted portfolios toward active credit intermediation, and generated sustained gains in per capita income and retail activity. Border discontinuity and difference-in-differences designs confirm that even modest statutory reforms were sufficient to substantially attenuate the Depression’s propagation through the bank-lending channel. The third chapter examines how population... https://escholarship.org/uc/item/2rb8b3x9 Thu, 23 Apr 2026 00:00:00 +0000 Bessa Ribeiro, Mariana https://escholarship.org/uc/item/28q3h646 End-to-end neural generation models have largely displaced the modular architectures that once gave dialogue system designers explicit control over what is said and how it is said. While these models produce fluent text, they collapse content planning, sentence planning, and surface realization into a single undifferentiated decoding step, sacrificing the controllable structure that earlier systems provided. This dissertation investigates how that structure can be recovered through the over-generate-and-rank (OGR) paradigm: generating multiple candidate outputs and selecting among them using learned or prompt-based ranking functions that jointly optimize semantic fidelity, stylistic appropriateness, and conversational coherence. We instantiate OGR at three levels of natural language generation for dialogue: utterance-level stylistic control, cross-domain semantic evaluation, and dialogue-level response selection.First, we show that explicit conditioning mechanisms, specifically... https://escholarship.org/uc/item/28q3h646 Thu, 23 Apr 2026 00:00:00 +0000 Harrison, Vrindavan https://orcid.org/0000-0002-2615-9170 https://escholarship.org/uc/item/26c7b6b8 We develop new uncertainty propagation methods for feed-forward neural network architectures with leaky ReLU activation functions subject to random perturbations in the input vectors and the neural network parameters. In particular, we derive analytical expressions for the probability density function (PDF) of the neural network output and its statistical moments as a function of the input uncertainty and the parameters of the network, i.e., weights and biases. A key finding is that an appropriate linearization of the leaky ReLU activation function yields accurate statistical results for both perturbations in the parameters of the network, and even for large perturbations in the input vector. This can be attributed to the way information propagates through the network. We also propose new analytically tractable Gaussian copula surrogate models to approximate the full joint PDF of the neural network output. To validate our theoretical results, we conduct Monte Carlo simulations... https://escholarship.org/uc/item/26c7b6b8 Thu, 23 Apr 2026 00:00:00 +0000 Diamzon-larot, Jeremy https://orcid.org/0009-0005-6989-4625 https://escholarship.org/uc/item/267443sp Distributed quantum computing (DQC) promises to overcome the scalability limitations of individual quantum processors by interconnecting multiple quantum processing units (QPUs) through quantum networks. However, realizing this vision requires addressing fundamental challenges across the entire computing stack, from physical-layer entanglement generation to system-level scheduling and circuit compilation.This thesis presents a series of contributions spanning the distributed quantum computing stack. We first address the simulation infrastructure gap by developing two complementary tools: A2Tango, a physical-layer simulator for atom-atom entanglement generation using atomic ensembles, and QuCloudSim, a discrete-event system-level simulator for quantum cloud environments. Building on this foundation, we design Q2R, a QoS-aware quantum network routing framework that jointly optimizes latency, fidelity, and application-level goodput beyond conventional throughput metrics. At the data... https://escholarship.org/uc/item/267443sp Thu, 23 Apr 2026 00:00:00 +0000 Zhou, Ruilin https://escholarship.org/uc/item/1kh0749p Modern AI and cloud services demand databases that deliver low latency and high throughput under tight memory budgets across disaggregated storage, memory, and network tiers, yet prevailing designs suffer from large DRAM footprints for indexes and RDMA paths that either burn memory-node CPUs or add extra round trips. This dissertation develops indexing and lookup techniques that make such disaggregated databases fast, predictable, and memory-efficient. First, I decouple lookup from directory maintenance unit and introduce a compact, minimal-state directory structure that preserves flexible data placement without the conventional DRAM overhead of hash-based directories. Second, I redesign RDMA-based key-value store indexing to retain one-round-trip reads while shifting compute off memory nodes, easing the computation burden in remote CPU side. Across multi-node evaluations, these techniques consistently reduce index memory footprint, improve throughput, and lower tail latency,... https://escholarship.org/uc/item/1kh0749p Thu, 23 Apr 2026 00:00:00 +0000 Liu, Yi https://escholarship.org/uc/item/0zp0t0tf While helping is usually considered a good thing, young children do not always help at home, and parents might not always want their children’s help. This study examined how young children and parents evaluate and discuss various opportunities for a child to help at home. U.S. 3- to 6-year-olds (N = 107; 55% girls, 45% boys; 65% Asian American, 30% White or European American, 15% Latinx, 3% Native American, 1% African American or Black, and 1% Hawaiian/Pacific Islander) and their parent discussed whether and why the protagonist child in a storybook should help their parent at home. Children then evaluated whether a protagonist child in four hypothetical vignettes should help in interviews, while their parent shared about their beliefs around children’s help in surveys. Children in interviews more often judged that a child should help if they had regularly helped with a given task (H1), and this effect was mediated by their perceptions of the protagonist parent’s interest in the... https://escholarship.org/uc/item/0zp0t0tf Thu, 23 Apr 2026 00:00:00 +0000 Martinez-Mora, Marie Grace https://escholarship.org/uc/item/0mn7h7t1 This dissertation contains three essays focusing on how incentives and institutional design shape pollution, firm behavior, and worker welfare in developing countries.In Chapter 1, I evaluate the impact of subsidizing early-maturity variety rice seeds (EMV) and providing technical training on residue management, one of the major causes of air pollution in India. I run a randomized controlled trial with rice farmers in Punjab, India. I find that relative to the control group, providing subsidies alone reduces residue burning by 1.1 acres (18% reduction compared to the control group). When combined with training, the burning reduces by 2.1 acres (a 34% reduction relative to the control mean). Training enhances the impact of subsidies by promoting greater EMV adoption and addressing knowledge gaps in sustainable residue management. A cost-benefit calculation suggests that each $1 spent generates $32.28 of social benefits in the Seed Subsidy Only group and $28.35 in the Subsidy and... https://escholarship.org/uc/item/0mn7h7t1 Thu, 23 Apr 2026 00:00:00 +0000 Gandhi, Piyush https://escholarship.org/uc/item/0g88h0sz Understanding how biological neural networks learn is both a fundamental scientific question and an increasingly urgent practical one. Despite decades of work demonstrating that in vitro neural cultures can be modified through electrical stimulation, the field lacks both a shared experimental framework for biological learning, as well as results mirroring those of artificial neural networks. This dissertation presents the tools, experiments, and scientific findings developed while pursuing biological learning in a dish. The work begins with a hybrid soft-rigid robotic system where model-free reinforcement learning outperforms classical control, establishing a first learning environment framework and motivating the extension of reinforcement learning from artificial to biological substrates. Contributions to closed-loop optogenetic modulation of epileptiform activity in human brain slices and a cloud-connected platform for organoid-based neuroscience education provided foundational... https://escholarship.org/uc/item/0g88h0sz Thu, 23 Apr 2026 00:00:00 +0000 Robbins, Ash https://orcid.org/0000-0002-5556-8547 https://escholarship.org/uc/item/9dn76024 This thesis presents the design, implementation, and evaluation of WildBerryEye, an open-source, low-cost ecological monitoring system based on Raspberry Pi single-board computers and the Sony IMX500 AI vision sensor. The system supports two runtime-selectable detection modes: on-sensor object detection and motion detection via frame differencing. The software architecture combines a modular Flask backend with a lightweight, browser-accessible front-end for live image streaming, REST API control, and batch data handling. Testing was conducted under idle and continuous detection workloads, focusing on battery life, CPU usage, and thermal behavior. Results indicate power constraints, with battery life under load limited to just over three hours, prompting consideration of alternative power solutions such as 18650 lithium-ion cells. In addition to controlled testing, the system was validated outdoors using real hummingbird activity. This field evaluation revealed substantial... https://escholarship.org/uc/item/9dn76024 Mon, 2 Feb 2026 00:00:00 +0000 Espinosa Ramirez, Carlos Isaac https://escholarship.org/uc/item/90921034 Located on occupied Mescalero Apache land in south-central New Mexico, the architecture of El Camino Real International Heritage Center rises out of an expanse of the desert off Interstate 25, the mast-like metal portion of its façade gleaming in the sun. The modern Interstate 25 closely parallels the historic Camino Real de Tierra Adentro, or Royal Road of the Interior, a series of shifting routes along which settlers and supplies moved between the New Spanish/Mexican capital of Mexico City and the region known today as New Mexico between approximately 1598 and 1880. This dissertation draws on archival sources to document and contextualize a revival of public interest in this “first highway of the Americas” in the final decades of the twentieth century, demonstrating that this interest culminated in the creation of El Camino Real International Heritage Center, a multi-million-dollar museum that was open to the public between 2005 and 2016. Applying insights from critical heritage... https://escholarship.org/uc/item/90921034 Mon, 2 Feb 2026 00:00:00 +0000 Bonner, Rachel Elizabeth https://escholarship.org/uc/item/1168417s Predation risk has driven the evolution of antipredator behaviors, such as alarm calling, which provide social information that may benefit a variety of individuals in the caller’s community. Although production specificity —the ability to produce distinct vocalizations in response to different predator types—has been documented in many species, it remains poorly understood in the context of Neotropical mixed-species flocks (MSFs). I investigated whether nuclear species in Amazonian understory MSFs produce predator-specific alarm calls. Field experiments conducted in Madre de Dios, Peru introduced Bicolored hawk (Accipiter bicolor) models as aerial predator threats, ocelot (Leopardus pardalis) models and humans as terrestrial threats, and agouti (Dasyprocta variegata) models and a remote-controlled car as terrestrial controls. We recorded and analyzed the vocalizations of six nuclear species in these flocks. We found that Dusky-throated Antshrikes (Thamnomanes ardesiacus) was... https://escholarship.org/uc/item/1168417s Mon, 2 Feb 2026 00:00:00 +0000 Henderson, Allene https://escholarship.org/uc/item/0v50262n Flow models generate high quality images, but they often miss long range dependencies because past steps are compressed into a single noisy state. We present ARFlow, which integrates autoregressive conditioning into the flow process. During training, we build causally ordered sequences by sampling images from the same class at multiple noise levels, with higher noise treated as predecessors to lower noise. This encourages the model to capture broader category level variation while respecting the causal path of denoising. At inference, the model conditions on previously generated images from earlier steps, producing a coherent generation trajectory. We further design a hybrid linear attention mechanism that improves memory and speed while matching full attention accuracy. On ImageNet $256 \times 256$, ARFlow achieves 6.63 FID without classifier-free guidance (CFG) and 1.96 FID with CFG, outperforming the SiT at 2.06 FID. Ablations validate the modeling strategy and chunk wise attention. https://escholarship.org/uc/item/0v50262n Mon, 2 Feb 2026 00:00:00 +0000 Hui, Mude https://escholarship.org/uc/item/00v4v86g In the field of game studies, the ability to share specific moments of games is critical. Preserving historical software and games in a holistic manner requires the capturing of not only static code or visual data, but also the dynamic user experience. While emulation can provide a path to interactivity, many emulators do not support capturing software interactions. Current replay systems rely on the assumption of purely deterministic emulators, which may not reproduce software performances with complete accuracy in practice. Video recordings are an obvious choice for accurate replayability, but they do not capture the software's interactivity. I introduce Clickies, a novel medium for recording and replaying software interaction performances. Clickies implement a three-stream format for recording, consisting of a textual stimulus log of user inputs, a high-temporal video response, and a compact, losslessly compressed Savestream containing periodic emulator states. For performance... https://escholarship.org/uc/item/00v4v86g Mon, 2 Feb 2026 00:00:00 +0000 Wong, Thanyared Amanda https://orcid.org/0009-0006-8783-0120 https://escholarship.org/uc/item/24034254 Wide-field spectroscopic surveys have transformed cosmology by mapping millions of galaxies and constraining the nature of dark energy. However, the escalating cost-per-spectrum and scaling limitations of conventional spectrographs present a critical bottleneck for next-generation facilities. This dissertation explores astrophotonics as a transformative solution, leveraging mature photonic technologies to enable compact, multiplexed, and cost-effective instrumentation. I present the design, implementation, and on-sky validation of two novel instruments at Lick Observatory: the Parallel Lantern Injection Unit (PLIU) and the Astrophotonic Advancement at Lick Observatory (APALO). The PLIU successfully demonstrated simultaneous dual-fiber injection and building on this foundation, APALO was developed as a modular, facility-class platform, enabling a more precise fiber injection platform, improved point-spread function quality, and flexible integration of diverse photonic devices,... https://escholarship.org/uc/item/24034254 Sun, 1 Feb 2026 00:00:00 +0000 DeMartino, Matthew https://orcid.org/0009-0003-2280-9820 https://escholarship.org/uc/item/9r16v0bg In this work, we develop numerical methods for conservation laws that explore statistical, structure-preserving, and machine-learning-based approaches, each built on top of traditional numerical solvers. First, we develop a general Gaussian-process-based “recipe’’ for constructing high-order linear operators such as interpolation, reconstruction, and derivative approximations. Building on this recipe, we derive a kernel-agnostic convergence theory for GP-based operators that interprets them as generalized finite-difference schemes, defines an effective order-of-accuracy proxy that captures non-ideal truncation-error structure, and uses this metric to select stencil geometries and kernel hyperparameters analytically. We then introduce a new second-order kernel, Discontinuous Arcsin (DAS), that is stationary and prevents oscillations. DAS is integrated into a shock-capturing framework called the Multidimensional Optimal Order Detection (MOOD) method and shows an increase in efficiency... https://escholarship.org/uc/item/9r16v0bg Sat, 31 Jan 2026 00:00:00 +0000 DeGrendele, Christopher Joseph https://orcid.org/0000-0002-7815-1496 https://escholarship.org/uc/item/9cs4b328 Dimensionality reduction is crucial to data analysis and visualization, as data is often recorded with more than three values. This make the data impossible to visualize as it is stored. UMAP is a dimensionality reduction algorithm that offers a topological method of reduction, focusing on maintaining the metric geometric and topological features of the data. The current implementation of UMAP uses the 1-skeleton, but we seek to establish a method for incorporating 2-simplices, as the 1-skeleton occasionally fails to accurately capture the geometry of the data. We use the 1-skeleton calculated by the current UMAP algorithm to search for potential triangles, and use weighted edges to keep the three vertices ’stuck’ together. This easy-to-use contribution furthers the cases in which UMAP is a useful dimensionality reduction algorithm. https://escholarship.org/uc/item/9cs4b328 Sat, 31 Jan 2026 00:00:00 +0000 McBride, Nicholas https://orcid.org/0009-0007-8653-9873 https://escholarship.org/uc/item/96m4c3vx Certificate functions, such as barrier functions and Lyapunov functions, are commonly used to verify control system properties. The construction of these certificates, however, is often difficult, typically requiring significant trial and error. Once a certificate function is found, modifications to the controller are hindered because each change requires the construction of a new certificate function. This problem is addressed in this dissertation by the design of uniting feedback strategies that allow uncertified controllers to be safely used by exploiting a controller with a known certificate as a backup. In uniting feedback, an automatic supervisor switches between two controllers. The result is a hybrid control strategy that switches between certified and uncertified controllers while preserving the safety or asymptotic property that is guaranteed for the certified controller. By using a certified controller as a backup, these uniting feedback strategies allow for exploiting... https://escholarship.org/uc/item/96m4c3vx Sat, 31 Jan 2026 00:00:00 +0000 Wintz, Paul Kenna https://orcid.org/0000-0003-0070-6663 https://escholarship.org/uc/item/7d89k0th This thesis explores the formalization of Weyl’s Law and its implementation in the Lean 4 proof assistant, bridging classical mathematical analysis with modern computational verification techniques. Under the guidance of Professor Pedro Morales-Almazan at UCSC, this research investigates the asymptotic growth behavior of eigenvalues in the Laplace eigenvalue problem across various bounded domains including intervals, rectangles, and disks.The project systematically develops the theoretical foundations of the Laplace operator in one, two, and three dimensions, addressing the eigenvalue problem −∆u = λu across different coordinate systems and geometries. In this work, we consider the Laplacian with Dirichlet boundary conditions, which ensure that the eigenvalue problem is well defined and admits a discrete spectrum. A significant contribution lies in the translation of classical mathematical proofs into formally verified code using Lean 4, supported by contemporary large language... https://escholarship.org/uc/item/7d89k0th Sat, 31 Jan 2026 00:00:00 +0000 Huang, ZhiBin https://escholarship.org/uc/item/0pj2b9nq Advances in next-generation sequencing technologies have enabled the recovery of genetic data from minimal, contaminated, and highly degraded samples, overcoming long-standing barriers in forensic analysis. However, standard PCR-based short tandem repeat (STR) typing can fail when poor-quality DNA prevents complete genotype calls at the sites used for comparison, as is often the case with small bone fragments or single, rootless hairs. Meanwhile, DNA identification methods relying on single-nucleotide polymorphic (SNP) sites are limited by the lack of tools that can incorporate low-coverage sequence data from trace samples to provide statistical evidence for identity testing. This dissertation addresses these challenges through the development of computational methods for reliable identity analysis of forensic samples. First, I present IBDGem, a fast and robust computational procedure for detecting identity-by-descent (IBD) regions by comparing low-coverage shotgun sequence... https://escholarship.org/uc/item/0pj2b9nq Sat, 31 Jan 2026 00:00:00 +0000 Nguyen, Remy Nghi Hoang https://orcid.org/0000-0002-2115-9059 https://escholarship.org/uc/item/002726r0 This dissertation, entitled “Masking Women: History as Performance in the Indian Ocean, 1844-2020,” investigates masking practices that emerge in nineteenth-century French colonial visual culture and find a resonance in the present. The study focuses on practices enacted by local practitioners in and around Reunion Island and the Comoros archipelago—islands located in the western Indian Ocean that became French colonies in the nineteenth century, and continue to share an enduring entanglement with France.Predicated on an open and capacious understanding of “masking” as referring to a variety of practices that involve forms of literal or metaphorical covering to achieve a political transformation—whether personal or collective—this dissertation examines a composite corpus of lithographs, photographs, and performances ranging from the historical to the contemporary, yet positioned within colonial visual culture, with the aim of interrogating the decolonial valences of masking. Does... https://escholarship.org/uc/item/002726r0 Sat, 31 Jan 2026 00:00:00 +0000 Toussaint, Axelle https://orcid.org/0009-0004-9659-5199 https://escholarship.org/uc/item/9z60g8v0 During the Summer of 2019, thousands of people in Puerto Rico protested in the streets to demand the resignation of the governor, Ricardo Rosselló. Rosselló was caught up in a public scandal, after the leak of a chatroom conversation on the Telegram App with his executive team. This dissertation studies how activists, and organizers intervened both the physical space, and the virtual space to force the governor to resign. In this research, I analyze the uses of social media (specially Twitter, and Instagram) by activists to amplify the protests in the streets. I performed a participant hybrid observation, where I place myself both in the streets, and scrolling through social media to document the activities from activists, organizers, and artists. During the observations I coded and decoded: protests in the streets, memes, illustrations, the role of reggaeton music, practices of citizen journalism, among other tactics that were used. I analyze how this two-week event shaped the... https://escholarship.org/uc/item/9z60g8v0 Thu, 29 Jan 2026 00:00:00 +0000 Davila Santiago, Juan Carlos https://escholarship.org/uc/item/9t62b960 In eukaryotes, chromatin is a highly organized condensed structure made up of nucleosome core particles (NCPs). NCPs are DNA-protein complexes composed of histone octamers that wrap around DNA. The ends of eukaryotic chromosomes consist of repetitive elements known as telomeres which need to be protected from the activation of unwarranted DNA damage response. In vertebrates, a six-protein complex named shelterin regulates telomere end protection. The shelterin proteins TRF1 and TRF2 mediate interaction with the double-stranded telomere DNA, while POT1 binds to the G-rich single-stranded telomere DNA overhang.Shelterin-nucleosome interactions have been characterized in some detail; however, the precise mechanism for how shelterin binds, and remodels, nucleosomes at telomeres is not completely understood.RNA Structure mediated splicing regulation and antisense therapeutic designAccurate pre-mRNA splicing is essential for human gene expression, yet is frequently disrupted in disease.... https://escholarship.org/uc/item/9t62b960 Thu, 29 Jan 2026 00:00:00 +0000 Chacaltana, Guillermo A https://escholarship.org/uc/item/96m824hx Assembly of the spliceosome onto introns during pre-mRNA processing is critical for proper mRNA production however, the spliceosome assembly mechanism is poorly understood. While the 5’ and 3’ splice sites are well conserved in human introns, the branchpoint sequence is a loose consensus. Proper recognition of the branchpoint sequence dictates success of the first step of splicing and 3’ss choice. Spliceosome mutations associated with cancer and neurodevelopmental disorders largely localize to factors involved in branchpoint sequence recognition including the spliceosome subunit U2 snRNP. The non-coding U2 snRNA, the central component of the U2 snRNP, forms a stem loop structure called the branchpoint-interacting stem loop (BSL) that is used for initial recognition of the branchpoint sequence. This stem must also unwind to form the branch helix for first step chemistry. The mechanism by which the BSL recognizes the branchpoint sequence, especially in human introns, and how it... https://escholarship.org/uc/item/96m824hx Thu, 29 Jan 2026 00:00:00 +0000 Stevers, Meredith Barbara https://orcid.org/0000-0002-4477-226X https://escholarship.org/uc/item/8xj4r3z0 The central dogma of molecular biology describes the process by which genetic information is expressed, wherein DNA is transcribed into mRNA and mRNA is translated into protein. It was once believed that each mRNA corresponded to a single gene and produced a single protein, with RNA molecules playing only a minor regulatory role in this process. However, advances in transcriptome sequencing uncovered a vast repertoire of noncoding RNAs and revealed the extensive diversity and complexity of the transcriptome. Since then, studies have shown that noncoding RNAs are expressed in a highly context and cell-type specific manner, suggesting their important regulatory roles in numerous biological processes. Long noncoding RNAs (lncRNAs) have emerged as key regulators of a variety of biological processes, influencing diverse aspects of cellular differentiation, proliferation, and immune signalling. The innate immune system is the first line of defense against foreign pathogens. Monocytes... https://escholarship.org/uc/item/8xj4r3z0 Thu, 29 Jan 2026 00:00:00 +0000 Montano, Christy https://orcid.org/0009-0009-7027-8458 https://escholarship.org/uc/item/8sz831cz In response to the security challenges inherent in the Internet of Things (IoT), in 2023, National Institute of Standards and Technology (NIST) has endorsed ASCON, a cipher suite designed to secure the communications between resource-constrained IoT devices. Thread, a popular wireless mesh protocol designed for smart homes and smart buildings, does not currently support ASCON. This thesis describes how OpenThread, the open source implementation of Thread, can be modified such that it is capable of using the encryption algorithms defined in the ASCON cipher suite. Compared to the original version of OpenThread, this thesis shows that the network performance and battery lifetime of smart home devices are not negatively impacted when OpenThread is secured by ASCON encryption. To the best of my knowledge, this thesis is the first to investigate the potential of ASCON in securing the communications of smart home devices operating under OpenThread. https://escholarship.org/uc/item/8sz831cz Thu, 29 Jan 2026 00:00:00 +0000 Tran, Simeon Alexander https://orcid.org/0009-0006-7661-6612 https://escholarship.org/uc/item/8r47f5r2 The cosmopolitan, chain-forming diatom Pseudo-nitzschia is a significant cause of harmful algal blooms (HABs) on the west coast of the United States. Toxigenic species produce the neurotoxin, domoic acid, which bioaccumulates in marine food webs and seafood products. Impacts from Pseudo-nitzschia HABs include closures of commercial fisheries, risks to sustainable aquaculture, and toxin exposure for marine animals from benthos to surface. Solid-phase adsorption toxin tracking (SPATT) is a passive sampling method for monitoring dissolved toxins and other compounds. It has gained widespread popularity for the ability to adsorb a diverse range of dissolved molecules in single sample. The research contained in this dissertation leverages historical ocean observing at the Santa Cruz Municipal Wharf (SCW) with novel applications for SPATT. Findings highlight seasonal and interannual patterns of dissolved domoic acid in relation to unique physical dynamics of Monterey Bay. New methodology... https://escholarship.org/uc/item/8r47f5r2 Thu, 29 Jan 2026 00:00:00 +0000 Trapp, Aubrey J https://orcid.org/0000-0001-9826-8811 https://escholarship.org/uc/item/7pb521m3 Hematite (α-Fe2O3) is an abundant, chemically robust oxide widely explored for photoelectrochemical and other energy-conversion applications, yet its conductivity remains severely limited by small-polarons and low carrier mobility and concentration. This thesis uses extended X-ray absorption fine structure (EXAFS), combined with constrained structural models and complementary thermodynamic calculations, to build an element-specific picture of donor doping in hematite across a broad concentration range. At low to moderate bulk loadings (« 1%at), tetravalent donors bind electron polarons and, cooperatively, one another to form electrostatic multipoles; EXAFS measurements on Sn-doped hematite directly resolve dopant-dopant correlations that signal such clusters and show how they deplete the population of mobile carriers without macroscopic phase separation. At higher concentrations, above a dopant-specific critical value Ccrit, additional donors increasingly occupy secondary oxide... https://escholarship.org/uc/item/7pb521m3 Thu, 29 Jan 2026 00:00:00 +0000 Mayford, Kiley https://escholarship.org/uc/item/6rz5918g This thesis explores the viability for two classes of specialized datacenter hardware, SmartNICs and memory-semantic-supporting interconnects, to address the overhead and variability of remote procedure call (RPC) libraries in microservices applications. This exploration takes two directions; the offloading of an RPC library and networking stack to a SmartNIC, and the bypassing of the network stack in favor of communication over a shared memory substrate supported by a CXL interconnect.We offload the RPC stack by splitting an industry standard C++ RPC library into two modules; "infrastructure'' which handles the network stack and the initial RPC processing, and "business logic'' which executes user implemented functions. Our evaluation demonstrates that offloading is a viable strategy to reduce RPC variability due to the isolation benefits of modularization, yet introduces a higher base cost for communication due to the overheads of moving data between address spaces.We explore... https://escholarship.org/uc/item/6rz5918g Thu, 29 Jan 2026 00:00:00 +0000 Ramos, Esteban https://orcid.org/0009-0006-4687-1057 https://escholarship.org/uc/item/62x4x1mv A common form of Wolbachia-induced manipulation of host reproduction is Cytoplasmic Incompatibility (CI). In CI, Wolbachia modification of sperm results in pronounced defects in paternal chromosome condensation, replication, and segregation during the first mitotic division. Recent studies in D. simulans have demonstrated that CI also induces independent and distinct later developmental defects, resulting in high rates of mitotic errors during the mid-blastula transition and larval lethality. Here, I show that in D. melanogaster, embryos derived from CI crosses that developed normally through to cellularization experienced significant mitotic defects during gastrulation and increased larval lethality, both of which were eliminated in the progeny of Rescue crosses (both sexes infected). Examination of 13 wild-caught lines from the Drosophila Genetic Reference Panel (DGRP) revealed extensive uncorrelated variation in the strength of the CI-induced early pre-hatching and late larval... https://escholarship.org/uc/item/62x4x1mv Thu, 29 Jan 2026 00:00:00 +0000 Perez, Claire Nicole https://escholarship.org/uc/item/5wj8k243 Genetic information flow throughout the cell is an essential component of life. The central dogma of biology best describes this genetic flow of information: that DNA is transcribed into messenger RNAs (mRNAs) and then translated into functional proteins. This process of taking information held in DNA and turning it into functional molecules (RNA/proteins) must be tightly regulated with quality control pathways to ensure cellular homeostasis.Among the most common deleterious mutations life can face is the Premature Termination Codon (PTC). PTCs are among the most harmful mutations a gene can acquire as they prematurely terminate the flow of genetic information, causing the cell to lose all information downstream of the mutation. The partial protein products created are potentially harmful to the cell as well. Nonsense-Mediated Decay (NMD) is a quality control pathway that recognizes and degrades mRNA transcripts containing premature termination codons. Despite its importance in... https://escholarship.org/uc/item/5wj8k243 Thu, 29 Jan 2026 00:00:00 +0000 Modena, Matthew Stephen https://escholarship.org/uc/item/5t11w855 Understanding how animal reproduction will change in a changing world requires integrating environmental, social, and behavioral perspectives. My dissertation provides new insights into how rising temperatures affect nesting, mating, and parenting behaviors in species with complex mating systems. I used the ocellated wrasse (Symphodus ocellatus), a Mediterranean fish, as a model organism because it exhibits alternative male reproductive tactics, complex social interactions, and obligate male-only care. This species inhabits shallow coastal waters and experiences strong seasonal and interannual variation in water temperature, providing a natural setting to examine how warming affects reproductive systems under realistic environmental and social conditions. In Chapter 1, I compiled a long-term dataset containing daily observations of all nests found at our study site, during approximately one month of the annual reproductive season. I analyzed data from 12 seasons to investigate... https://escholarship.org/uc/item/5t11w855 Thu, 29 Jan 2026 00:00:00 +0000 De Morais, Louise Alissa https://orcid.org/0000-0002-5673-2775 https://escholarship.org/uc/item/5qx8c2jq This research presents the first archaeological study of marronage as a form of resistance in Saint-Domingue (nowadays Haiti). In 1791, Saint-Domingue, then the most profitable plantation colony of the eighteenth-century Atlantic world, began a massive uprising, which became the most successful slave revolt that culminated in the creation of the first independent Black nation. While scholars have debated the significance of marronage in this process, interpretations remain divided. Two schools of thought emerged. One argues that maroons played a pivotal role in that success across the colony, whereas the other claims that the impact of marronage was limited, framing the Haitian Revolution primarily as a by-product of the French Revolution. Both perspectives, however, lack evidence from maroon settlements. Their discourse was supported by archival documents and oral history. This research addresses the gap by introducing archaeological data from the first maroon site excavated... https://escholarship.org/uc/item/5qx8c2jq Thu, 29 Jan 2026 00:00:00 +0000 Louis, Camille https://escholarship.org/uc/item/5783c462 This thesis presents the design, fabrication, and validation of a miniature five-axis capacitive force/torque (F/T) sensor for robotic tweezer manipulation. Conventional robotic grippers typically sense forces at the wrist or jaw, which obscures the true tool–object interaction. To address this limitation, a low-cost, tool-tip sensor was developed using a printed circuit board electrode array and a compliant silicone dielectric, integrated with an the Analog Devices AD7147 capacitance-to-digital converter (CDC) and ESP32 microcontroller. Finite-element simulations in ANSYS Electronics guided the electrode geometry and gap sizing, while prototypes were manufactured through PCB stacking, silicone casting, and low-cost assembly methods. The sensor was calibrated against a commercial ATI Industrial Automation F/T sensor using least-squares regression with higher-order polynomial features. Experimental testing with a robotic tweezer gripper demonstrated reliable detection of slip,... https://escholarship.org/uc/item/5783c462 Thu, 29 Jan 2026 00:00:00 +0000 King, Leo https://escholarship.org/uc/item/4wq5060g Sound, Selfhood, and the Social: Aurality and Narrative in the English Long Eighteenth Century considers the role of sound in the literature of the English long eighteenth century, and is especially focused on the development of the novel form. I argue that sound and failure attend one another, and that this failure is inscribed into the history of the early English novel in a way that forces scholars to revise their understanding of that history as one that celebrates epistemological mastery and conceives of the self as bounded and individual. Instead, the difficulty of apprehending sound—both aurally and in writing—frustrates the individual and their ways of knowing and being in their social and natural environment, and occasions a reckoning with failure that orients characters, writers, and readers toward a more fundamentally entangled understanding of themselves and their social worlds. I make this scholarly revision through close readings of the works of Daniel Defoe, Eliza... https://escholarship.org/uc/item/4wq5060g Thu, 29 Jan 2026 00:00:00 +0000 Armada, Spencer Gregory https://escholarship.org/uc/item/4937m5s2 Locally Managed Marine Areas (LMMAs) are widely viewed as a solution to overexploited nearshore fisheries in low-governance countries. In this dissertation, I use a socio-ecological system framework to evaluate the impacts of LMMAs on reducing poverty and conserving ecosystems in some of Madagascar’s more than 150 LMMAs. First, I assess the effects of LMMAs on poverty reduction using a difference-in-differences approach. I show that Madagascar LMMAs have mixed effects on poverty reduction, depending on whether they are established within Marine Protected Areas (MPAs). Only LMMAs located within MPAs experienced poverty reduction; second, LMMAs not associated with MPAs showed a decrease in perceived fish abundance and perceived catch reliability. Then, I use a remote sensing tool, the Modular Mangrove Recognition Index, combined with a performance analysis to examine the impacts of LMMAs’ establishment on mangrove cover change. I show that LMMAs significantly increased mangrove... https://escholarship.org/uc/item/4937m5s2 Thu, 29 Jan 2026 00:00:00 +0000 Rabearisoa, Ando Landisoa https://orcid.org/0000-0001-5371-7695 https://escholarship.org/uc/item/3qf0s989 Large Language Models (LLMs) have shown that a model pre-trained on general knowledge can perform well on specific tasks. However, LLMs natively perform poorly when it comes to demonstrating an understanding of rules, such as applying them, interacting with them, generating or modifying them, or evaluating them. Fine-tuning LLMs on a specific set of rules can significantly improve this performance. Yet, doing so undermines one of the main advantages of using a pre-trained model, which is its ability to generalize to rulesets outside its training distribution. This ability is critical for using LLMs as a tool in the game development process to give feedback or suggest rule modifications. In this paper, we introduce a framework for generating datasets to benchmark and train LLMs on their understanding of rules. We use Solitaire card games as our testbed for generating these datasets, as they have simple rules but offer a large space of possible variants, each played completely differently.... https://escholarship.org/uc/item/3qf0s989 Thu, 29 Jan 2026 00:00:00 +0000 Bateni, Bahar https://orcid.org/0000-0002-0701-0311 https://escholarship.org/uc/item/39v9q7fj The purpose of this dissertation is to bring to light the experiences of women who were barred from official authority in religious spaces, but managed to carve out for themselves an identity as exemplary saints. Through the memorial literature and memoirs, it is clear that suffering could be an empowering experience. The women included throughout embraced the language of suffering and operated within the limitations of their gender to shape the narrative of their experience. In so doing, they proved their piety, accumulated unofficial authority as lay leaders, and left an enduring legacy. https://escholarship.org/uc/item/39v9q7fj Thu, 29 Jan 2026 00:00:00 +0000 Kuehn, Danielle Catherine https://escholarship.org/uc/item/3761x207 How do coast redwood methods of existence—their sensibility and receptivity, durational temporalities, and capacity for transmutation—offer strategies for developing artistic research methods that honor both human and arboreal ways of being while fostering conditions for multispecies flourishing? Contributing to the emerging fields of arboreal humanities and environmental media studies, I theorize coast redwoods (sequoia sempervirens) as mediators, materials, and models of ethical environmental relationality. Through sustained engagement with redwood methods—their sensibility and receptivity, phenomenological responsiveness, durational temporalities, rhizomatic placemaking, and capacity for transmutation through fire—I develop "arboreality": a method of being that foregrounds relational care, continuance, and reciprocity. I then transform these redwood qualities into arts-based research methodologies, making the process of apprehending how redwoods persist, adapt, and flourish... https://escholarship.org/uc/item/3761x207 Thu, 29 Jan 2026 00:00:00 +0000 Bird, Sarah https://escholarship.org/uc/item/34m7b4r0 AI-generated audio descriptions (AI ADs) offer scalable solutions for making visual media accessible to blind and low-vision (BLV) audiences. Nevertheless, little is known about how BLV users experience and evaluate these descriptions across emerging platforms. In this qualitative study, we conducted semi-structured interviews with ten (N=10) BLV participants, recruited based on divergences in prior survey ratings, to explore their perceptions of both human- and AI-generated ADs in contexts ranging from traditional film to short-form video and livestreams. Thematic analysis revealed four key themes: (1) information prioritization and genre-sensitive details, (2) the social dynamics of shared viewing, (3) the “better-than-nothing” consensus tempered by emotional contextual gaps in new media accessibility deserts, and (4) the artistry-precision dilemma. Our findings highlight the need for adaptive, transparent, and user-informed AD systems that balance narrative resonance with efficiency.... https://escholarship.org/uc/item/34m7b4r0 Thu, 29 Jan 2026 00:00:00 +0000 Yee, Diana https://orcid.org/0009-0008-7484-8377 https://escholarship.org/uc/item/2k71q7ss A bridge is a diffusion process that connects two given points in a finite dimensional vector space that are pinned at two given times. A Schrödinger bridge is a diffusion process that lifts this concept to the infinite dimensional manifold of probability measures. This dissertation contributes to the rapidly growing research area of Schrödinger bridge that is undergoing an explosion of mathematical and algorithmic developments across several disciplines: stochastic control, non-equilibrium statistical mechanics, and generative AI. This burgeoning interest is partly due to the fact that a Schrödinger bridge comes with a large deviation guarantee: it is the most-likely measure-valued path connecting the given endpoint measures. In this sense, the Schrödinger bridge is the most parsimonious model consistent with observed snapshots at two times. Part of its modern popularity is also because both the theory and algorithm for Schrödinger bridge are nonparametric, i.e., dispense the... https://escholarship.org/uc/item/2k71q7ss Thu, 29 Jan 2026 00:00:00 +0000 Teter, Alexis Mikayla Horne https://orcid.org/0009-0002-9925-5187 https://escholarship.org/uc/item/2j32b6qd I aim in this essay to provide theoretical, practical, and philosophical frameworks for my musical practice. The original works of mention are written with flexible instrumentation of two “melody” and two “tanpura” parts with drum set. I expand on rhythmic and melodic strategies related to my practice of Hindustani music to encompass their application from soloist centric Hindustani music to an ensemble of multiple centers. I link rhythmic and melodic to concepts of time and intonation from South Asian and Afro-Diasporic cosmologies to propose new philosophical engagements with memory. Alternative scoring techniques for ensemble writing in staff notation and digital audio workstations are also discussed herein. https://escholarship.org/uc/item/2j32b6qd Thu, 29 Jan 2026 00:00:00 +0000 Batish, Keshav https://escholarship.org/uc/item/2bx5b5jp Modern applications, such as data center workloads, have become increasingly complex. These applications primarily operate on massive datasets, which involve large memory footprints, irregular access patterns, and complex control and data flows. The processor-memory speed gap, combined with these complexities, can lead to unexpected performance inefficiencies in these applications, preventing them from achieving optimal performance. Considering the complexity and size of data center applications, manually identifying and resolving performance issues is often impractical or impossible. Instead, developing new compiler optimization techniques can be a more effective and scalable solution to boost both performance and energy efficiency.In this dissertation, we focus on identifying the root causes that limit the performance of data center workloads. We analyze the limitations of current profile-guided compiler optimization techniques for addressing these performance gaps. Finally,... https://escholarship.org/uc/item/2bx5b5jp Thu, 29 Jan 2026 00:00:00 +0000 Jamilan, Saba https://escholarship.org/uc/item/29m4v7sm Terahertz (THz) frequency accelerating structures could provide the accelerating gradients needed for compact next generation particle accelerators. However, accelerator applications are limited by losses during transport and coupling of THz radiation to the acceleration structure. Application of the near-field of a tilted pulse front THz source on electron bunches mitigates transport and coupling losses. Moreover, this technique increases the THz interaction length offering a simplified and efficient route toward THz-driven streaking. One of the most promising THz generation techniques for accelerator applications is optical rectification in LiNbO3 (LN) using the tilted pulse front method. The spatiotemporal characterization of THz pulses from these sources is currently limited to far-field methods. While simulations of tilted pulse front THz generation have been published, little work has been done to measure the near-field properties of the THz source. Measuring and understanding... https://escholarship.org/uc/item/29m4v7sm Thu, 29 Jan 2026 00:00:00 +0000 Gabriel, Annika https://escholarship.org/uc/item/27d575w8 In this thesis, I will define the word ‘community’ as a group of artists from the same shared cultural and social background, bonding by collaborating on a shared goal of putting on a production that uplifts underrepresented voices. With this definition guiding me, I took a script that was originally written as a one-woman show and cast an ensemble instead. With the Panza ensemble, I argue that I was able to foster an environment during rehearsals and performances where collaboration was encouraged and vulnerability was celebrated.The Panza Monologues, written by Virginia Grise and Irma Mayorga, is a series of monologues that center on reflections on the panza (belly) as a site of embodied knowledge on various topics. My goal was to stage these monologues with the intention of creating a space for Latina actors to find community through storytelling. As this year’s Barnstorm Theater Company’s Managing Director, I had the resources to create new opportunities for Latina actors... https://escholarship.org/uc/item/27d575w8 Thu, 29 Jan 2026 00:00:00 +0000 Farias, Maddie Jeul https://escholarship.org/uc/item/1p65w4zt A central question in evolutionary biology is whether evolution is predictable – driven by similar selective pressures leading to repeated outcomes – or contingent on historical and stochastic processes. While convergent traits are widespread across the tree of life, the underlying genetic mechanisms often vary, raising questions about how different sources of variation can influence evolutionary trajectories. In this dissertation, I use the Poecilia mexicana species complex, which includes multiple sulfide-adapted and non-sulfide-adapted populations across four river drainages, as a model for repeated adaptation to extreme environments. In Chapter One, I used targeted capture sequencing to test for selection in 250 candidate genes, identifying shared signatures of selection in key sulfidic detoxification genes across independent sulfidic lineages. In Chapter Two, I generated new genome assemblies for six Poecilia and Gambusia species, assessed synteny and demographic history,... https://escholarship.org/uc/item/1p65w4zt Thu, 29 Jan 2026 00:00:00 +0000 Ryan, Kara https://orcid.org/0000-0002-5352-4663 https://escholarship.org/uc/item/16s5x7rc As climate change intensifies drought frequency and severity, California’s endemic blue oak (Quercus douglasii) faces increasing stress, dieback, and regeneration challenges. Understanding the mechanisms that enable drought survival is critical for forecasting persistence and informing climate-adaptive conservation strategies. My dissertation investigated the mechanisms underlying blue oak drought resilience by examining ecotypic variation, soil biotic interactions, and spring leaf phenological plasticity across climatic gradients. Using greenhouse and field common garden experiments, I tested whether populations differ in drought coping ability, how soil microbial communities influence drought tolerance, and whether ecotypes vary in spring phenology and herbivore interactions when translocated. In Chapter One, I used a greenhouse common garden drought experiment with seedlings from five populations to test for ecotypic differences in drought coping ability. I found that seedlings... https://escholarship.org/uc/item/16s5x7rc Thu, 29 Jan 2026 00:00:00 +0000 Poudel, Sushmita https://orcid.org/0000-0002-2261-8835 https://escholarship.org/uc/item/0s3791fq Manual berry harvesting is a labor-intensive and time-sensitive task that often suffers from inconsistency and susceptibility to human error. The increasing demand for efficiency and reliability in agricultural practices has created interest in robotic solutions, particularly those leveraging machine vision for fruit detection. However, the inherent variability in berry ripeness, as well as their delicate structure, presents significant challenges to automation. Ripeness is commonly assessed through visual indicators such as color and size, but these cues are often subjective, vary significantly between species, and can be unreliable under different lighting conditions. Moreover, the harvesting process requires careful manipulation to prevent bruising or damaging the fruit, necessitating a high degree of precision and control. This research presents a fully integrated robotic harvesting system that combines machine learning, computer vision, and tactile sensing. The system autonomously... https://escholarship.org/uc/item/0s3791fq Thu, 29 Jan 2026 00:00:00 +0000 Spott, Elizabeth India https://orcid.org/0009-0002-8608-8148 https://escholarship.org/uc/item/0984c6dh Hardware design flows have become increasingly complex as modern chips integrate billions of transistors and rely on aggressive synthesis optimizations to meet performance, area, and power targets. While these transformations improve circuit efficiency, they also erase the correspondence between gate-level netlists and their originating HDL source lines. The loss of traceability makes post-synthesis debugging, timing back-annotation, and root-cause analysis extremely difficult. Existing solutions depend on tool-specific metadata or preserved signal names, which are often lost after flattening, retiming, or logic restructuring.To address this long-standing problem, this thesis presents SynAlign, a structural alignment framework that restores the mapping between optimized netlists and source code without relying on synthesis metadata. SynAlign treats both the reference RTL and synthesized designs as graphs and iteratively aligns them using shared structural cues—such as sequential... https://escholarship.org/uc/item/0984c6dh Thu, 29 Jan 2026 00:00:00 +0000 Garg, Sakshi https://orcid.org/0000-0003-3781-8159 https://escholarship.org/uc/item/06x8z4xn Wildfire smoke is becoming a fast-growing challenge for California's agricultural regions, affecting both crop production and the communities who sustain it. Although wildfire impacts on ecosystems and public health have received growing attention, far less is known about how smoke influences agricultural productivity or how existing data systems shape our understanding of vulnerability in agricultural regions in California. This dissertation addresses these gaps through three connected studies that examine crop productivity and the social dimensions of wildfire risk in California’s agricultural sector.First, I quantify the historical relationship between PM2.5 and wine grape yields in Northern California using county level regression models that incorporate climate, air quality, and crop data. I find that wildfire smoke negatively affects wine grape yields in two of the three study sites, shifting crop yield modeling from being primarily climate-driven to models that incorporate... https://escholarship.org/uc/item/06x8z4xn Thu, 29 Jan 2026 00:00:00 +0000 Hoefler, Astrid https://orcid.org/0009-0006-6078-0477 https://escholarship.org/uc/item/2h71s4qt The theoretical goal of this dissertation is to sharpen and refine our understanding of the A-A-bar divide (Postal, 1971; Chomsky, 1977): a distinction which governs the ways that steps of movement and their associated landing sites interact with the wider systems of the syntax and its interfaces. The traditional view takes this divide to reflect a strictly binary cut: positions and steps of attraction are exhaustively classified as either “A” or “A-bar,” and these two labels are picked up by a range of systems across the grammar. But a growing literature recognizes that the division is not so clear-cut: the alignment between A- and A-bar-properties is often imperfect and incomplete, such that apparent steps of A- and A-bar-movements often show properties drawn from the opposite set. The task of this dissertation is thus to address the following questions: (1) Empirically: what are the ways that the classical A- and A-bar-properties can and cannot combine?(2) Theoretically: what... https://escholarship.org/uc/item/2h71s4qt Wed, 5 Nov 2025 00:00:00 +0000 Brodkin, Dan https://orcid.org/0000-0001-5663-1301 https://escholarship.org/uc/item/2xv4d53z This thesis investigates the potential of ultra-wideband (UWB) radar as a sensing tool for remote agricultural sensing. With its portability and low-power consumption, UWB radar addresses the challenges posed by traditional sensor networks that are often lacking in agricultural environments. UWB’s broad bandwidth allows it to penetrate various media, making it particularly suited for sensing in high-attenuation media such as soil. This thesis presents two projects that contribute to the field of UWB-based sensing. The first project, WaDAR, integrates a UWB radar with backscatter technology for soil moisture sensing capabilities. Building on prior work, this project introduces dual backscatter tags that facilitate accurate, low-cost, low-power, point soil moisture readings at depths of up to 30 cm, overcoming the limitations of previous systems. The second project explores using ultrawideband (UWB) radar to estimate soil bulk density, a primary indicator of compaction. This work... https://escholarship.org/uc/item/2xv4d53z Tue, 21 Oct 2025 00:00:00 +0000 Vetha, Eric David https://orcid.org/0000-0001-7155-7058 https://escholarship.org/uc/item/3zh6g116 This dissertation investigates the grammar and processing of the Telugu anaphor tanu, which, unlike standard reflexives, resists binding by local co-arguments and instead prefers non-local, perspectival antecedents. I argue that tanu is a logophoric reflexive, licensed by a covert perspectival operator in the left periphery of its clause. Across self-paced reading and eye-tracking studies, I demonstrate that Telugu comprehenders initially retrieve the local, unlicensed co-argument before reanalyzing it to the licensed, non-local one. These findings support the local search hypothesis, which posits that anaphor resolution begins with the retrieval of the most local antecedents before more structurally distant ones. I suggest that this is part of a multi-stage resolution process (cf. Bonding and Binding; Sanford and Garrod, 1989), where an early heuristic link is formed to the nearest antecedent by local search, followed by a later stage of (potential) reanalysis and integration... https://escholarship.org/uc/item/3zh6g116 Wed, 15 Oct 2025 00:00:00 +0000 Sunil Arvindam, Vishal https://escholarship.org/uc/item/50r7q2g0 Large language models (LLMs) demonstrate strong performance on complex reasoning, yet their reliability and efficiency in real-world deployments remain limited, particularly in specialized domains with sparse coverage, in scalable post-training for reasoning, and in the underexplored dynamics of distilling long chain-of-thought (CoT) behaviors. This dissertation investigates these challenges through the lens of reasoning-intensive tasks, focusing on claim verification and mathematical reasoning.First, we introduce SYNTHVERIFY, a step-by-step synthetic data generation framework that integrates structured domain specification, evidence synthesis, and aspect-conditioned claim creation. Without relying on external corpora or costly human annotation, SYNTHVERIFY yields diverse training data and substantially improves zero-shot transfer claim verification across multiple specialized domains. Second, we propose a three-stage training recipe that develops smaller reasoning models capable... https://escholarship.org/uc/item/50r7q2g0 Wed, 8 Oct 2025 00:00:00 +0000 Zhao, Rongwen https://escholarship.org/uc/item/9f36k7sb Alpine ecosystems are among the most spatially heterogeneous systems, shaped by extreme climate and rugged topography, making their endemic species especially vulnerable to environmental change. The Brown-capped Rosy-Finch (Leucosticte australis), a cliff-nesting passerine nearly endemic to the Colorado alpine, lacks comprehensive ecological data needed to inform conservation. This dissertation characterizes the species’ realized niche and evaluates how environmental conditions, habitat heterogeneity, and research practices influence its ecology across the annual cycle.In Chapter 1, fine-scale habitat surveys and behavioral observations during the breeding season revealed strong associations with cliff proximity, snow patches, and rocky substrates, underscoring the spatial constraints of central-place foraging. Chapter 2 used multi-year line-transect distance sampling across 57 alpine basins to produce the first statewide population estimates (116,421–148,546 individuals) and... https://escholarship.org/uc/item/9f36k7sb Tue, 7 Oct 2025 00:00:00 +0000 Bernier, Kathryn M. https://orcid.org/0009-0005-6276-6719 https://escholarship.org/uc/item/1d79b03f The Ensemble Kalman Filter (EnKF) is a widely used method for combining model forecasts with observations to improve state estimation in dynamical systems. However, its standard formulation assumes the availability of time-resolved observations and a known observation operator—conditions that are often impractical when direct measurements are sparse, costly, or unavailable. This work proposes a data-driven alternative in which the role of observations is replaced by statistical information derived from the system’s invariant measure. A neural network is trained on simulated trajectories to jointly learn the invariant measure and a nonlinear mapping from model state space to an “observation” space consistent with that measure, while also estimating associated uncertainty statistics. Embedding this learned mapping within the EnKF enables updates that maintain dynamical consistency with the invariant measure, suppress spurious noise, and preserve long-term stability even without... https://escholarship.org/uc/item/1d79b03f Tue, 7 Oct 2025 00:00:00 +0000 Silberstein, Arnold James https://orcid.org/0009-0006-6282-0014 https://escholarship.org/uc/item/5fm8d1cw Digital technologies have made large-scale online interaction a central part of how people communicate, connect, and work together. Yet scaling often comes at the cost of depth, and interactions can become superficial and chaotic, drifting away from the richer interactional contexts of small-scale or in-person settings that support trust and meaningful exchange, and that make it possible for participants to respond to and build constructively on one another’s ideas. Although recent advances such as large language models have opened new possibilities for shaping online interaction, there has been relatively little exploration of how to design interaction mechanisms that take advantage of large-scale engagement while fostering interactions that are engaged, authentic, connected, and generative.In this dissertation, I explore how large-scale online systems can be designed to support engaged and meaningful interaction at scale from three distinct angles: 1) creating few- to-many conversation... https://escholarship.org/uc/item/5fm8d1cw Mon, 6 Oct 2025 00:00:00 +0000 Lei, Kehua https://orcid.org/0000-0002-1446-9008 https://escholarship.org/uc/item/3mx4j1xp Live-cell fluorescence microscopy enables high-resolution, non-invasive imaging and optical stimulation of biological processes, yet existing systems are often prohibitively expensive, mechanically complex, and poorly suited for wide field-of-view, high-speed, multi-channel experiments—particularly those involving optogenetics. To address these limitations, I developed a low-cost, modular fluorescence microscope constructed from 3D-printed mechanical components and off-the-shelf optics. I also contributed to the engineering of the incubator to support long-term live-cell imaging. The platform accommodates multi-channel fluorescence imaging with simultaneous optogenetic stimulation, features configurable optical paths for different fluorophores, offers motorized positioning via a 3-axis manipulator, enables dual-camera alignment for concurrent imaging, and incorporates integrated brightfield capability.The system’s performance was demonstrated by imaging mScarlet and GCaMP6f fluorescence... https://escholarship.org/uc/item/3mx4j1xp Mon, 6 Oct 2025 00:00:00 +0000 Hawthorne, Nicholas Ambrose https://orcid.org/0000-0002-6933-0513 https://escholarship.org/uc/item/3jh4v4mr In observing the solar magnetic field, possibly the most prominent features visible on the surface are sunspots, which emerge at di!erent latitudes as the solar cycle progresses. Sunspot pairs are believed to be formed by concentrated bundles of mainly toroidal magnetic field (flux tubes) looping through the surface. These regions exhibit surprisingly ordered patterns of behavior such as the Hale’s Polarity Law, Joy’s Law, and the Solar Hemispherical Helicity Rule (SHHR). The latter states that emerging flux in the Northern hemisphere generally has left-handed current helicity, whereas the Southern hemisphere has right handed. While there is magnetic field at all scales on the sun, the origins of these active regions and the connection between large-scale dynamo generated fields and active region scales is a long standing and difficult question. The flux tubes that form sunspots most likely originate from magnetic buoyancy instabilities that occur in the solar tachocline and then... https://escholarship.org/uc/item/3jh4v4mr Mon, 6 Oct 2025 00:00:00 +0000 Lydon, Sean Robert https://escholarship.org/uc/item/270147t1 Accurate and efficient subgrid–scale (SGS) parameterization remains a long-standing challenge in geophysical fluid simulation. Cutting-edge climate models often operate at ∼ 25 km resolution, which is too coarse to capture subgrid–scale dynamics, such as mesoscale eddies in turbulent flow, necessitating SGS closures (Hallberg, 2013). Ross et al. (2023) showed that data-driven surrogates trained on filtered/coarse-grained high-resolution simulations have promise, yet systematic benchmarks in idealized quasi-geostrophic (QG) turbulence report poor generalization across regimes. Recent work by Darman et al. (2025) showed that transfer learning can improve out-of-distribution (OOD) performance by adapting a model’s effective spectral filters to the target regime.Building from these insights, we propose an Autoregressive (AR) Model using a Fourier Neural Operator (AR–FNO) as described by Li et al. (2020); as a foundation model for SGS forcing: a lagged timeseries FNO encoder is pre-trained... https://escholarship.org/uc/item/270147t1 Mon, 6 Oct 2025 00:00:00 +0000 Kinsinger-Dang, Lumina Ahon https://orcid.org/0009-0007-1912-7123 https://escholarship.org/uc/item/1kw6m09n Infection is a complex interaction between host organisms and infectious agents. Circadian rhythms, a key aspect of host physiology, influence infection outcomes, but how they modulate — or are modulated by — infection is not entirely understood. It is unclear if infection can induce disruptions to the core circadian clock. Clinical evidence suggests bacterial meningitis influences outputs of circadian rhythms such as the sleep cycles and wakefulness of patients. This suggests infection of tissues surrounding the brain may affect circadian rhythms differently than infection in the rest of the body. To investigate this, we used zebrafish larvae (Danio rerio) as a model, leveraging their genetic tractability and optical transparency to determine how infection with Streptococcus pneumoniae proceeds and influences per3 expression. Per3 expression was monitored by infecting Tg(per3:luciferase) larvae via the hindbrain ventricle or caudal vein. This revealed that infection elevated... https://escholarship.org/uc/item/1kw6m09n Mon, 6 Oct 2025 00:00:00 +0000 McEachen, John Colin Peter https://escholarship.org/uc/item/90n2t54j My dissertation uses que(e)rying as a methodological and theoretical framework to analyze how early-modern criminal records from Spain and Colombia produced narratives about race, religion, gender, sexuality, and, by extension, intimacy, in the lives of enslaved people of African descent. In response, I put the cases into conversation with literary and visual texts—such as a diagram of the slave ship Marie-Seraphique, Autobiografia de un esclavo by Juan Francisco Manzano (along with some of his poems), Chango, el gran putas by Manuel Zapata Olivella, and Insurrection: Holding History by Robert O’Hara—to create counter-narratives to the ones suggested by the cases. As a result, my dissertation proposes that by que(e)rying these documents, we see how queer intimacies and an intimacy of knowing were cultivated in the lives of enslaved black people. https://escholarship.org/uc/item/90n2t54j Sun, 5 Oct 2025 00:00:00 +0000 Booth, Bree https://escholarship.org/uc/item/2fk1x5s9 Neural networks are widely used to model physical systems, but robust uncertainty quantification (UQ) is essential for reliable scientific applications. We present a two-part study. First, we introduce a simple framework combining multi-layer perceptrons with finite difference schemes to model autonomous dynamical systems. Applied to the Lorenz63 and Kuramoto–Sivashinsky equations, this approach achieves accuracy comparable to auto-regressive baselines. Second, we benchmark three UQ methods — Deep Ensembles, Monte Carlo Dropout, and Variational Bayesian Last Layers — on multivariate regression tasks with randomized components. Evaluated using RMSE, Negative Log Likelihood, and Calibration Error, we find that Deep Ensembles provide the most consistent and calibrated performance, while VBLL, though computationally efficient and theoretically grounded, exhibits drift and weaker calibration. Repurposing these UQ methods as generative models further highlights the strength of ensembles.... https://escholarship.org/uc/item/2fk1x5s9 Sun, 5 Oct 2025 00:00:00 +0000 Adams, Janice Lorrain https://escholarship.org/uc/item/9kr7926q Time resolved coherent X-ray scattering allows for a wide range of systems to be measured across many timescales and length scales. This dissertation work discusses recent results in time-resolved coherent magnetic spin texture systems. This work is comprised of two main projects, the first of which reports on the discovery of two nematic phase as found in amorphous-Fe51Ge49 and combines spatio-temporal X-ray scattering correlative studies to probe the order in the system using both synchrotron X-rays and X-ray Free Electron Lasers. Time resolved studies highlight temporal fluctuations onset in the helimagnetic system on the timescale of both sub-nanoseconds and minutes near phase boundaries. Additionally, computational micromagnetic simulations highlight that this is likely commensurate with a reorientation of the magnetic helices and suggest that this is the reason for the presence of temporal dynamics. The second project continues to expand some of the computational methods... https://escholarship.org/uc/item/9kr7926q Thu, 25 Sep 2025 00:00:00 +0000 Tumbleson, Zoey Hazel https://orcid.org/0000-0002-6193-112X https://escholarship.org/uc/item/9324s81m This thesis addresses critical challenges in decentralized federated learning (DFL) related to network efficiency, practical deployment, security, and scalability to advanced models. We introduce near-optimal overlay networks using expander graphs for enhanced DFL performance and robustness. Furthermore, we present FedLay, the first decentralized overlay network with protocols for construction, maintenance, and efficient model aggregation in real-world DFL scenarios. To enhance security, we propose BDFL, a blockchain-based DFL framework incorporating model verification, incentives, and a reputation system to mitigate poisoning attacks. Finally, we extend DFL principles to the demanding task of serving large language models with PlanetServe, a decentralized overlay demonstrating significant latency reduction. This research collectively advances the state-of-the-art in DFL, paving the way for more efficient, practical, secure, and scalable decentralized machine learning systems. https://escholarship.org/uc/item/9324s81m Thu, 25 Sep 2025 00:00:00 +0000 Hua, Yifan https://escholarship.org/uc/item/8z06m09n Despite a wide range of interpretations and seemingly unanswered questions, it is not a mystery why entropy increases in closed, classical systems - trajectories in phase space tend toward higher entropy macrostates with high probability. However, it remains a mystery why, if there is truly a second law, the entropy was so low in the past. Such a low entropy in the past amounts to an incredibly unlikely assumption. This is called the past hypothesis, and it is widely accepted among physicists and philosophers of physics. Regardless, the past hypothesis comes with important conceptual issues. This dissertation will investigate solutions to two of these issues: a) there is a physical mystery, called the initial state problem, as to how total entropy could have been low right after the big bang since we know matter had maximum entropy; b) derivations of the second law from the past hypothesis do not condition on all reliable information, as obligated by Bayesian axioms. In particular,... https://escholarship.org/uc/item/8z06m09n Thu, 25 Sep 2025 00:00:00 +0000 Scharnhorst, Jordan https://escholarship.org/uc/item/6rs6k5vr Disorder is pervasive in condensed matter systems, significantly influencing electronic transport, thermodynamics, and phase transitions. This thesis presents theoretical explorations into how disorder impacts transport properties and thermodynamic behavior in nodal-line semimetals, Weyl semimetals, and geometrically frustrated magnets. In nodal-line semimetals, quenched disorder induces substantial renormalizations of quasiparticle properties, manifesting in observable singularities such as divergences in the density of states and conductivity near critical disorder strengths. Remarkably, disorder-driven instabilities share deep theoretical connections with Cooper pairing and excitonic condensation phenomena observed in interacting two-dimensional metals. An explicit duality mapping further bridges disorder-induced effects in nodal-line semimetals and interaction-driven transitions in lower-dimensional electron systems, thus providing novel opportunities for experimental verification. Weyl... https://escholarship.org/uc/item/6rs6k5vr Thu, 25 Sep 2025 00:00:00 +0000 Zhu, Siyu https://orcid.org/0000-0003-2935-5422 https://escholarship.org/uc/item/5h42v75q This thesis investigates the structure and scaling of Yang-Lee zeros in classical and quantum many-body systems through a combination of analytical and numerical techniques. We first derive analytically that the logarithm of Yang-Lee zeros of the nearest-neighbor Ising model scales as √ k at high temperatures for arbitrary regular lattices, and we obtain sum rules for the expansion coefficients of these logarithms. These constraints are verified for multiple lattice geometries using exact enumeration. To study the thermodynamic limit of antiferromagnetic systems, we implement a numerical linked cluster expansion (NLCE) on the distribution of zeros for finite clusters. This approach reveals root curve structures and critical fields for square and triangular lattices, with NLCE providing more robust estimates than exact enumeration.We then explore the mean-field Ising model with both ferromagnetic and antiferromagnetic interactions. Using the mean-field free energy, we numerically... https://escholarship.org/uc/item/5h42v75q Thu, 25 Sep 2025 00:00:00 +0000 Sedik, Muhammad https://orcid.org/0009-0006-9546-0932 https://escholarship.org/uc/item/483369r5 Salt marshes are unique hydrologic systems in which tidal exchange, terrestrial groundwater, and precipitation interact to control salinity, moisture, and ecological zonation. While tidal elevation is a well-known driver of plant species distributions in Mediterranean-climate marshes, the potential for incoming terrestrial groundwater to ameliorate stressful conditions has only recently been recognized. Sediment addition is a common restoration practice used to restore lost elevation in marshes that have subsided, yet the influence of terrestrially-sourced groundwater and moisture retention in revegetation on the restored marsh platform have not been identified. Here we combine shallow sediment cores, weekly hillslope–marsh groundwater transect monitoring and neutron-probe vadose-zone moisture measurements to test whether hydrodynamic drivers of salinity and moisture mediate revegetation of Salicornia pacifica following sediment addition at Hester Marsh, central California. We... https://escholarship.org/uc/item/483369r5 Thu, 25 Sep 2025 00:00:00 +0000 Robbins, Levi Dailey https://orcid.org/0009-0006-7971-8083 https://escholarship.org/uc/item/3r44z0fz Accurate assessment of wound healing progress is critical for optimizing patient care and preventing complications, yet clinicians currently lack precise tools to determine where a wound stands in the healing timeline. Wound healing progresses through overlapping stages of inflammation, proliferation, and maturation, each marked by characteristic shifts in gene expression that are difficult to interpret without robust computational methods. This paper proposes to classify wound healing stages from transcriptomic data using support vector machines combined with biologically informed clustering to serve as features for the hierarchical SVM classifiers. This approach is applied to two distinct wound types: excisional wounds in pigs (21-day timeline) and burn wounds in mice (42-day timeline), enabling comparison of classification performance across different injury mechanisms. The models achieved high overall accuracy, with the burn model performing better at the classification of... https://escholarship.org/uc/item/3r44z0fz Thu, 25 Sep 2025 00:00:00 +0000 Moreland, Zoe Nicole https://escholarship.org/uc/item/3g66k49n Laryngeal consonants within and across languages often showcase notable asymmetries in phonological and phonetic patterning. These asymmetries regularly require specific, special treatment of the laryngeal consonants, resulting in a large amount of inconsistent and contradictory behavior and explanations across patterns. The explanatory space for asymmetries involving laryngeals is varied, ranging from formal phonological explanations to phonetic diachronic-based explanations. This dissertation investigates asymmetries involving laryngeal consonants in three separate case studies, working through the explanatory space to identify and argue for the most plausible explanation for the asymmetric behavior in each case study. The first case study involves an instance of patterned exceptionality in Chamorro, where glottals appear to license a wider range of vocalic contrast in preceding vowels compared to non-glottals in the language. The second case study concerns an asymmetry within... https://escholarship.org/uc/item/3g66k49n Thu, 25 Sep 2025 00:00:00 +0000 Bibbs, Richard Lee https://escholarship.org/uc/item/9r64q24x High sensitivity detection of mass loading would greatly expand applications of surface acoustic wave (SAW) biosensors for high-throughput and label-free viral diagnosis. However, the use of SAW biosensors has been limited, primarily because their low measurement reliability in the presence of unfavorable environmental fluctuations (e.g., temperature, humidity, pressure, strain, and so on. Here, we present a novel interferometer-based SAW sensor that uses two delay lines: i) sensing delay line and ii) reflection delay line allowing interference-based detection. Each delay line provides a complementary signal (peak amplitude/resonant frequency) response enabling double verification of the accumulated biomass signature, drastically improving detection reliability and quantification capability. As an example, virus size dielectric nanoparticles (each particle has a diameter of 200 nm and mass of 4.44 fg) are dynamically monitored as they are deposited on the sensor channel. We show... https://escholarship.org/uc/item/9r64q24x Wed, 24 Sep 2025 00:00:00 +0000 Atwater, Jackson Harry https://orcid.org/AJAAX https://escholarship.org/uc/item/9jd5f5vm Misophonia, a condition marked by decreased tolerance to everyday sounds such as chewing or slurping, can cause severe impairment in social and private settings. The present study examined whether the temporal binding window (TBW) differs between individuals with misophonia and controls for simple flash-beep, complex trigger, and complex neutral stimuli. We measured participants’ TBWs using a simultaneity judgment task(SJ) with simple, trigger, and neutral audiovisual stimuli. Our findings showed that the overall width of TBW varied as a function of stimulus complexity in both groups, with trigger stimuli having the widest TBW, followed by neutral stimuli. Our results also showed that the right TBW was significantly narrower for the trigger PAVS stimuli in the misophonia group, suggesting more temporal precision for misophonics when the video preceded the trigger sound. We also report a novel finding that the benefit of PAVS stimuli is maximal when they are synchronized compared... https://escholarship.org/uc/item/9jd5f5vm Wed, 24 Sep 2025 00:00:00 +0000 Mahzouni, Ghazaleh https://orcid.org/0009-0004-2619-4269 https://escholarship.org/uc/item/8td74454 My dissertation, Re-Pairing Passion: Milton’s Poetics of Embodied Education, explores how Milton’s depiction of passion establishes an innovative early modern phenomenology of feeling that prefigures our modern understanding of emotional experience. In the 17th century, at the peak of the debate about the passions and how one should relate to them, Milton wrote Paradise Lost, which presents the complex affective lives of Adam and Eve as archetypal humans. Re-Pairing Passion proposes reading Paradise Lost as a pedagogical tool that offers readers complex models for passionate living in each of the poem’s characters. Milton’s portrait of passion reflects his belief in monism, the early modern idea that body and mind are indivisibly united, unlike modern dualism, which treats them as separate. I argue that Paradise Lost is a pedagogical tool intended to educate by portraying passion that provokes readerly emotion.The five chapters of the dissertation trace the connections between... https://escholarship.org/uc/item/8td74454 Wed, 24 Sep 2025 00:00:00 +0000 Multer, Monica https://orcid.org/0009-0005-5684-9497 https://escholarship.org/uc/item/8fv8667t Urbanization is a major driver of biodiversity change worldwide. Although patterns of change have been documented across many taxa, species interactions driving those patterns have received less attention. Mixed-species flocks are a common type of social grouping within bird communities, known to provide foraging and/or anti-predation benefits. We evaluated the structure and organization of mixed flocks and vegetative characteristics along an urban-natural gradient in Southern California. We hypothesized that flock species composition, species richness, size, and foraging guild composition would be driven by differences in vegetation between urban and natural areas, and predicted that urban flocks would be larger and more diverse due to clustered food resources produced by exotic vegetation. NMDS analyses showed that flock species composition differed significantly between urban and natural site types, as did composition of the entire bird community. However, flock size, species... https://escholarship.org/uc/item/8fv8667t Wed, 24 Sep 2025 00:00:00 +0000 Arthur, Noah https://escholarship.org/uc/item/8f16n786 The classic restoration ecology model of ecosystem recovery predicts that restoring the initial conditions of a formerly degraded site will facilitate its recovery and convergence with a reference site. Few restoration studies have long-term, multi-site restored and reference forest data to evaluate habitat recovery. My dissertation focused on multi-decadal forest recovery along 100-km of the Sacramento River, California. First, I used repeat surveys to evaluate if the trajectory of vegetation structure and community composition in 11 restored forests converged with 8 reference forests and whether the trajectory differed for forest overstories and understories 9–16 years after the prior surveys. My results suggested that this system has converged with reference forest in overstory structure, based on tree and shrub stem sizes and densities, but not yet in overstory or understory composition, in part due to the invasive, shade-tolerant shrub Rubus armeniacus expanding in restored... https://escholarship.org/uc/item/8f16n786 Wed, 24 Sep 2025 00:00:00 +0000 Constantz, Brook M https://orcid.org/0000-0002-6155-5769 https://escholarship.org/uc/item/8cx9n98k My dissertation examines the role that urban Black female youth culture played in the formation of the sixties girl group phenomenon, a style of music popular in the United States from the late 1950s to the mid-1960s, which focused on the voices and emotional concerns of teenage girls. Through oral histories, interviews, and correspondence with over a dozen girl group vocalists, my dissertation locates the roots of girl groups and girl group music within urban Black neighborhoods and communities of the post-war era. In these environments, cooperation, harmony, and collectivity were central strategies young Black women and girls developed in response to the changing social and economic realities of urban renewal, suburbanization, and Cold War politics. My research shows how girl groups and girl group music emerged from the new technology, youth-oriented consumer culture, and shifting gender and sexual roles of the post-war era. As members of the first generation of African American... https://escholarship.org/uc/item/8cx9n98k Wed, 24 Sep 2025 00:00:00 +0000 Berney, Apryl Kathryn https://orcid.org/0000-0002-8049-5642 https://escholarship.org/uc/item/87f776mz This thesis investigates the use of deep learning for classifying wound infections from photographic images, using colony-forming unit (CFU) counts as a quantitative labeling standard. Leveraging the visual information in wound photographs and the clinical relevance of bacterial burden, the study implements a multi-task U-Net architecture for both image reconstruction and binary classification in a shared-encoder framework. Three experimental conditions were explored: one using original images with positive class weighting, one incorporating data augmentation to enhance visual diversity, and one employing 5-fold cross-validation with augmentation to improve validation reliability. The non-augmented model achieved 91.7% accuracy at a threshold of 0.8, correctly identifying 4 of 5 infected cases, while Experiment 2 achieved 87.5% accuracy at a moderate threshold of 0.5 but became more conservative at higher thresholds. The third experiment reached 79.6% accuracy at a threshold... https://escholarship.org/uc/item/87f776mz Wed, 24 Sep 2025 00:00:00 +0000 Osorio, Sebastian https://escholarship.org/uc/item/7tf1c99q Reptile and amphibian (herpetofauna) populations face a substantial risk of decline, with potentially cascading consequences for ecosystem function. Effective conservation requires an understanding of their adaptive capacity, and examining responses to past environmental change can provide valuable insight. Stable isotope analysis is a widely used method for reconstructing ecological and physiological patterns in both modern and ancient organisms, yet it remains underutilized for modern herpetofauna and is exceedingly rare in fossil herpetofauna studies.The overarching goal of this thesis is to advance stable isotope research of reptiles and amphibians. To achieve this, I (1) reviewed the existing literature to identify trends, gaps, and high-priority research directions; (2) generated baseline functional knowledge to improve interpretation of stable isotope results; and (3) contributed original data on the foraging ecology of modern and ancient lizards and niche partitioning... https://escholarship.org/uc/item/7tf1c99q Wed, 24 Sep 2025 00:00:00 +0000 Ricker, Adrienne Cherie