http://www.rssboard.org/rss-specification 720 https://escholarship.org/uc/ucb_etd/rss Recent eScholarship items from UC Berkeley Electronic Theses and Dissertations Fri, 15 May 2026 08:39:32 +0000 https://escholarship.org/uc/item/6np4w53h Modern edge, client, and robotics platforms increasingly rely on domain-specific systems-on-chip (SoCs) that combine general-purpose CPUs with many accelerators for machine learning and perception workloads. At the same time, the slowdown of technology scaling and the rapid evolution of models and software stacks make it difficult to design accelerator-rich SoCs that remain effective as applications and deployment scenarios change. This dissertation develops system approaches for scalable specialization in heterogeneous SoCs and shows how to make specialized accelerators deliver high performance and continue to be effective as workloads, resource budgets, and system configurations evolve. The dissertation argues that scalable specialization is fundamentally a full-system problem. Specialized accelerators are most effective when architects can co-explore accelerator designs and system effects with full-system pre-silicon evaluation, when accelerators are integrated and virtualized... https://escholarship.org/uc/item/6np4w53h Wed, 11 Feb 2026 00:00:00 +0000 Kim, Seah https://escholarship.org/uc/item/4df2w4xm Mycobacterium tuberculosis (Mtb) is an intracellular bacterial pathogen and causative agent of the disease tuberculosis. Mtb establishes infection in the lungs via inhalation of aerosol droplets containing the bacteria. Approximately a quarter of the world's population is latently infected with Mtb. Control of Mtb infection is dependent on the adaptive immune response to Mtb infection, however which cell types and immune pathways are necessary for controlling Mtb infection remains unclear. The scientific community has identified type 1 CD4+ T helper cells (Th1) and their key cytokine interferon gamma (IFN-γ) as important for controlling infection, yet infected humans will eventually lose control of Mtb infection and develop active tuberculosis disease while Th1s are present in the lungs. There is growing evidence in human studies, non-human primate infection models, and murine tuberculosis vaccine models that type 17 CD4+ T helper cells (Th17s) can promote control of Mtb infection,... https://escholarship.org/uc/item/4df2w4xm Wed, 11 Feb 2026 00:00:00 +0000 Zilinskas, Alex Henry https://escholarship.org/uc/item/1jc855s5 As fusion energy is being developed, a fundamental mechanistic understanding of how materials will behave when exposed to the intense reactor environment is imperative. F82H is one of the most mature candidate materials for the structural vessel of the pioneering ITER experiment. The alloy is a tempered martensitic steel developed to withstand the fusion reactor environment however much is still not known about it. Particularly, the mechanical effects of carbide nucleation in the steel has not been deeply studied across the microstructure. To investigate this, quenched and tempered F82H needs to be mechanically investigated using small scale mechanical testing to probe individual boundaries. Small scale mechanical testing probes micron-scale volumes of material in order to understand specific material defect interactions. Using such resolved techniques will allow for conclusions to be formed about carbide precipitation and the effects on mechanical properties within different... https://escholarship.org/uc/item/1jc855s5 Wed, 11 Feb 2026 00:00:00 +0000 Lam, Sebastian https://escholarship.org/uc/item/28b4b3gt This dissertation addresses the challenge of scalable post-training for large language models along two axes: data generation and algorithmic robustness. On the data side, we demonstrate that AI-generated preference data can achieve state-of-the-art alignment results through the Nectar dataset and Starling reward models, then advance to fully self-generated supervision via Meta-Rewarding, where models act as actor, judge, and meta-judge simultaneously. This progression---from expensive human annotation to external AI teachers to autonomous self-play---eliminates traditional data bottlenecks while maintaining or exceeding performance. The key insight is that comparative feedback (K-wise rankings, pairwise judgments) is more reliable than absolute scoring. On the algorithm side, we identify and resolve fundamental inconsistencies in standard RLHF through P3O (Pairwise Proximal Policy Optimization), which performs comparative reinforcement learning rather than optimizing... https://escholarship.org/uc/item/28b4b3gt Tue, 10 Feb 2026 00:00:00 +0000 Wu, Tianhao https://escholarship.org/uc/item/9rn2w1gd Many algorithmic systems rely on information about people’s preferences and needs, from personalization to policy design. A common assumption is that more individual-level data will yield better outcomes. We challenge this by asking how much we need to learn about individuals, and in what ways, to improve a group objective. In the first part, we study settings where individuals bear the cost of communicating their preferences and steering the system. The challenge here is not limited data, but the difficulty of expressing what matters to users. We show that overly simplistic preference models cause alignment methods to fail when preferences deviate from their assumptions: they aggregate heterogeneous preferences in undesirable ways and misinterpret revealed preferences from individuals with inconsistent preferences. This motivates additional channels for preference expression, such as annotator information during training or costly signaling during inference. In the... https://escholarship.org/uc/item/9rn2w1gd Mon, 9 Feb 2026 00:00:00 +0000 Shirali, Ali https://escholarship.org/uc/item/86f530x6 Understanding brain function requires modeling a high-dimensional, nonlinear dynamical system continuously driven by sensory input, shaped by internal state and context, and expressed through behavior. Machine learning can be used to extract predictive structure from large-scale neural recordings, but its scientific value hinges on whether fitted parameters admit interpretation as testable hypotheses about neural computation. This thesis develops and applies interpretable predictive models—primarily linearized encoding models that map stimulus or behavioral features onto neural responses—to investigate (i) how human cortex represents abstract concepts during naturalistic language comprehension and (ii) how deep brain stimulation (DBS) modulates movement-related neural dynamics in Parkinson's disease. The first study addresses a central limitation of encoding models built on dense word embeddings or neural network activations: superposition. When latent semantic factors... https://escholarship.org/uc/item/86f530x6 Mon, 9 Feb 2026 00:00:00 +0000 Zeng, Alicia https://escholarship.org/uc/item/8528621n This thesis presents progress on a demonstration experiment of two-color ionization injection (TCII) in laser driven plasma accelerators (LPAs). The method has been theorized to potentially produce ultra-low, tens of nm rad, emittance beams, but has yet to be demonstrated in the laboratory. Were LPAs to reliably produce ultra-low emittance beams, they would become compact sources for a number of applications that require high brightness beams enabled by low emittance.Circularly polarized drive pulses were used for the first time to turn off ionization injection and prepare a dark-current-free wake for further injection. A regime was found in which wakes generated by circularly polarized pulses were proven to be of equivalent amplitude as those generated by linearly polarized pulses, but unlike the linearly polarized case, ionized electrons were not injected into the trapped region of the wake.High intensity, broadband, ultrashort third harmonic pulses were generated for ionization... https://escholarship.org/uc/item/8528621n Mon, 9 Feb 2026 00:00:00 +0000 Fan-Chiang, Liona https://escholarship.org/uc/item/7wz6c2nd This dissertation studies the job market at varying levels of granularity using three distinct yet complementary approaches: behavioral modeling, causal inference, and generative AI. Chapter 1 develops a behavioral job search model to understand the aggregate distribution of wage growth for job switchers in the labor market. Loss aversion predicts excess mass and sharp asymmetries in the density of wage growth around zero, which is empirically tested using administrative data from South Korea. Estimated loss aversion implies that the marginal value of additional pay is 12% higher for pay cuts than pay raises, and ignoring loss aversion would overstate the pass-through of hiring subsidies to worker salaries by 18%. Chapter 2 applies causal inference to study how family-friendly benefits offered by firms shape career decisions made by workers with children. The analysis exploits quasi-experimental variation from the staggered rollout of onsite childcare and mandated... https://escholarship.org/uc/item/7wz6c2nd Mon, 9 Feb 2026 00:00:00 +0000 Chu, Ross https://escholarship.org/uc/item/7vf7f1xh What happens when you squeeze a 2D semiconductor beyond its comfort zone? In this thesis, I demonstrate that mechanical rippling induced by controlled compression can carve out flat valence bands in 2D group-VI TMD monolayers (MoS2, WS2, MoSe2, and WSe2), transforming them into self-assembled lateral heterostructures with nothing more than a simple squeeze.With first-principles calculations in hand, I follow the cascade that links mechanical rippling to electronic flat bands. Under compression, the monolayer’s buckling instabilities spawn ripples that reconfigure the band structure, shifting EK downward and EΓ upward (the latter sitting between a nearly flat segment along ΓX and the precisely flat line ΓY). As the ripple amplitude grows further, a critical point is reached where EΓ overtakes EK and the valence band maximum relocates to Γ, making the gap indirect and signaling the birth of massive holes and a new electronic identity.This mechanical squeezing accomplishes what typically... https://escholarship.org/uc/item/7vf7f1xh Mon, 9 Feb 2026 00:00:00 +0000 Alawein, Meshal https://escholarship.org/uc/item/77f039gj Global fossil fuel dependence has led to a dramatic increase in CO2 emissions over recent decades. Reducing this dependence rests on widespread adoption of intermittent energy sources and electric vehicles. However, state-of-the-art energy storage technologies like rechargeable lithium-ion batteries are insufficient to meet the necessary energy density and reliability requirements. One promising approach is to replace graphite anodes in conventional lithium-ion batteries with pure lithium metal to achieve a ten-fold increase in energy density. Lithium metal anodes, however, are incompatible with traditional organic liquid electrolytes due to high chemical reactivity and high propensity for unstable lithium deposition. New classes of solid polymer electrolytes have emerged as conductive, mechanically stable alternatives to liquid electrolytes that aim to promote stable lithium deposition and suppress dendrite growth. Balancing the mechanical strength and ionic conductivity in these... https://escholarship.org/uc/item/77f039gj Mon, 9 Feb 2026 00:00:00 +0000 Patel, Vivaan https://escholarship.org/uc/item/6qq038tx T cell exhaustion is a dysfunctional state that arises during chronic infections and cancer, characterized by impaired effector functions and sustained expression of inhibitory receptors. While transcriptional, epigenetic, and metabolic rewiring have been well documented in exhausted T cells, a comprehensive understanding of how translation is regulated in this state remains incomplete. To address this gap, we performed ribosome profiling and RNA sequencing on exhausted human CD8⁺ T cells to globally assess translational control. Our analyses reveal a marked repression of 5’ terminal oligopyrimidine (TOP) mRNAs during exhaustion. Unexpectedly, we demonstrate that this translational repression occurs despite evidence of elevated mTOR activity in exhausted T cells. These findings uncover a previously unknown layer of translational control in exhausted T cells. https://escholarship.org/uc/item/6qq038tx Mon, 9 Feb 2026 00:00:00 +0000 Pledger, Erica https://escholarship.org/uc/item/5cm6x0qj The four serotypes of dengue virus (DENV1-4) cause the most prevalent human mosquito-borne viral disease. Dengue is a self-limiting febrile disease that can progress to life-threatening diseases known as dengue hemorrhagic fever and dengue shock syndrome. Secondary heterotypic DENV infection and pre-existing intermediate levels of cross-reactive DENV binding antibodies are known risk factors for dengue and severe dengue. Recently, non-immunological host factors, specifically obesity, have been associated with severe dengue. However, the association between obesity and the risk of DENV infection, the development of non-severe dengue, and the quantity, quality, and durability of antibody responses to DENV remain undefined. Given that obesity and DENV are both on the rise in children worldwide, this dissertation aims to fill these gaps in the scientific literature. To do so, I used data and banked blood samples from the Pediatric Dengue Cohort Study (PDCS), a longitudinal prospective... https://escholarship.org/uc/item/5cm6x0qj Mon, 9 Feb 2026 00:00:00 +0000 Mercado Hernandez, Reinaldo Javier https://escholarship.org/uc/item/4wc954jp Before there may be democracy, or group decision-making, there must be a demos, or a decision-making group. Central to any theory of democracy, then, is an answer to the question of how to constitute the demos, or how to decide who the decision-making group will be. There are three principles that serve as answers to the question of how to constitute the demos: an All Standing Citizens Principle, an All Coerced Subjects Principle, and an All Affected Interests Principle. This dissertation carries out a defense of neither the All Standing Citizens Principle nor the All Coerced Subjects Principle, each of which has many champions, but the All Affected Interests Principle, which has many critics by comparison. The defense consists of the claim that, in contrast to either the All Standing Citizens Principle or the All Coerced Subjects Principle, the All Affected Interests Principle can satisfy both a Substantive Condition for Democratic Legitimacy, concerning the ends for which the... https://escholarship.org/uc/item/4wc954jp Mon, 9 Feb 2026 00:00:00 +0000 Zuhone, Kristin Lauren https://escholarship.org/uc/item/4np8g18k The ability of cells to power energy-demanding processes depends on maintaining the ATP hydrolysis reaction a billion-fold away from equilibrium. Cells respond to changes in energy state by sensing changes in ATP, ADP, AMP, and inorganic phosphate. A key barrier to a better understanding of the maintenance of energy homeostasis is a lack of tools for direct manipulation of energy state in living cells. Here, we report the development of ATPGobble a genetically encoded tool for controlling cellular ATP hydrolysis rate. We validated ATPGobble by showing that it doubles the energy demand, decreases [ATP]/[ADP] and [ATP]/[AMP] ratios, and activates AMPK activity in human cells. We then used ATPGobble to systematically characterize the proteome and phosphoproteome changes caused by direct manipulation of the energy state. Our results establish ATPGobble as a powerful approach for dissecting the regulatory roles of energy state in human cells, opening new opportunities to study how... https://escholarship.org/uc/item/4np8g18k Mon, 9 Feb 2026 00:00:00 +0000 Ekvik, Alex E. https://escholarship.org/uc/item/3pz9n64q This dissertation investigates how international human rights law reshapes the practice of solitary confinement when supported by diverse state actors. While existing scholarship demonstrates that human rights law can be effective when bolstered by robust advocacy, it rarely examines whether and how the support of state actors shapes the reform process. Drawing on over two years of fieldwork and 149 in-depth interviews across three penal regimes, I identify three mechanisms—alienation, co-optation, and polarization—through which state support facilitates substantive human rights reforms but also generates unintended outcomes that constrain their scope and depth. Each mechanism is exemplified by a respective case: in Taiwan, strong backing from high-level policymakers enabled sweeping statutory reform, yet its top-down nature alienated correctional bureaucrats, who, feeling exploited and stripped of professional dignity, engaged in passive resistance that weakened implementation.... https://escholarship.org/uc/item/3pz9n64q Mon, 9 Feb 2026 00:00:00 +0000 Lin, Yen-Tung https://escholarship.org/uc/item/2cm0c15n Large language model (LLM) inference is increasingly deployed in latency-critical and cost-sensitive settings, yet remains fundamentally constrained by the sequential nature of autoregressive decoding. Speculative decoding has emerged as a promising technique to mitigate this bottleneck by introducing intra-request parallelism, allowing multiple tokens to be verified in parallel by a target model. However, despite its growing adoption, speculative decoding exhibits fragile and highly variable performance in real-world systems, with effectiveness depending sensitively on workload characteristics, batch sizes, model configurations, and system conditions.This dissertation studies speculative decoding from a full-stack systems perspective, spanning algorithmic design, empirical characterization, and production-grade control mechanisms. First, we introduce Online Speculative Decoding (OSD), a framework that continuously adapts draft models to the evolving query distribution during... https://escholarship.org/uc/item/2cm0c15n Mon, 9 Feb 2026 00:00:00 +0000 Liu, Xiaoxuan https://escholarship.org/uc/item/2886k59z Robots remain far behind humans in their ability to learn broadly from the world. While large language and vision models have achieved remarkable generalization by training on the collective artifacts of human activity, robot learning continues to depend on narrowly curated, robot-collected datasets and hand-designed objectives. This dissertation explores an alternative paradigm in which robots learn directly from human experience, using the physical interactions, perceptions, and decisions of people in everyday life as a scalable source of supervision.The dissertation develops this idea across three intertwined themes. First, it reconsiders the objective of learning itself. Rather than relying on manually specified rewards, robots can align their internal motivations with signals that reflect human-like structure in behavior. This begins with intrinsic reward matching, which enables autonomous skill discovery through the comparison and alignment of intrinsic and extrinsic objectives.... https://escholarship.org/uc/item/2886k59z Mon, 9 Feb 2026 00:00:00 +0000 Adeniji, Ademiloluwa Andrew https://escholarship.org/uc/item/1pn3g5ps This dissertation is a portfolio of compositions exploring syncretic instrumentality: a practice that treats hybrid instrumental configurations as sites where questions of agency, cultural positioning, and embodied knowledge are tested under real-time conditions. Three notated compositions—beyond the barrens (2023), shaped and already gone (2023), and outpouring beneath my skin (2024)—trace a trajectory from stratified temporal strata through Afro-Brazilian kinetic morphologies toward distributed resonance networks. The portfolio culminates in entrecorpo, a solo performance environment for expanded guitar and string stations with live electronics.entrecorpo exists as a multi-modal composition: album, performer-facing score materials, technical dossier, and instrumental preparation guide. This format enacts the dissertation’s arguments about notation as guided embodiment, studio-to-stage cycles as compositional method, and distributed agency across human and technological actants.... https://escholarship.org/uc/item/1pn3g5ps Mon, 9 Feb 2026 00:00:00 +0000 Ribeiro, Danniel https://escholarship.org/uc/item/1f40q9sj The mammalian retina contains neural circuits that extract salient features from the visual scene before relaying this information to the brain. For example, there is a dedicated neural circuit—the direction-selective circuit—that extracts the signals on the retina due to self-motion and motion within the visual field. Direction-selective ganglion cells (DSGCs) fire more robustly to motion in the preferred direction, than to motion in the opposite, or null, direction. It is well understood the direction-selective response is mainly driven by asymmetrical release of g-aminobutyric acid (GABA) from presynaptic interneuron starburst amacrine cells (SACs) during null direction motion, where each SAC process preferentially forms more synapses with the DSGC subtype whose preferred direction is anti-parallel to the orientation of that process. How these circuits are wired up during development remains an open question. In many parts of the brain, circuit wiring is influenced by sensory... https://escholarship.org/uc/item/1f40q9sj Mon, 9 Feb 2026 00:00:00 +0000 Bistrong, Karina https://escholarship.org/uc/item/9vr7m463 The use of colloidal particles as surfactants that spontaneously assemble at fluid–fluid interfaces has led to the discovery of many functional materials, including Pickering emulsions, bijels, colloidosomes, and biliquid foams. Such materials are found in both biological and synthetic contexts and have applications in catalysis, sensing, and numerous everyday products. Recently, nanoparticles co-assembled with polymeric surfactants at fluid–fluid interfaces have been used to create so-called structured liquids: liquid droplets with solid-like properties. These materials display shape controllability and structural integrity like solid materials while retaining their all-fluid character. In addition to their mechanical stability, these materials inherit the unique attributes of their nanoparticle building blocks, enabling physics exotic to disordered systems, such as ferromagnetism and conductivity.The origins of these solid-like properties are believed to be rooted in the dynamically... https://escholarship.org/uc/item/9vr7m463 Sun, 8 Feb 2026 00:00:00 +0000 Bordia, Gautam https://escholarship.org/uc/item/7b60k9bz Post-irradiation examination (PIE) of nuclear fuels and structural materials remains one of the most time and resource intensive components of nuclear materials research, limiting the pace of innovation for next-generation fission and fusion systems. Traditional destructive techniques introduce substantial contamination, require hot cell infrastructure, and provide limited throughput, resulting in long delays in obtaining critical microstructural, isotopic, and thermophysical data. This dissertation presents the development of a femtosecond laser-based characterization platform designed to dramatically reduce the burdens associated with PIE. By leveraging the ultrashort pulse ablation regime, which minimizes thermal damage and produces chemically representative debris, the system enables rapid, minimally destructive sampling suitable for downstream analysis. The platform integrates precision sample handling, aerosol-tight contamination, and real time debris transport to mass spectrometry,... https://escholarship.org/uc/item/7b60k9bz Sun, 8 Feb 2026 00:00:00 +0000 peddeti, Chaitanya https://escholarship.org/uc/item/74f365g1 This dissertation comprises two parts. In the first part, we study log-Sobolev and Poincaré inequalities. We prove a bound for the log-Sobolev constant of convolution measures and use this to show that the log-Sobolev constant is monotone in the central limit theorem. We also prove results for quantitative convergence of the log-Sobolev and Poincaré constants in the central limit theorem. The second part covers transportation-information inequalities for multi-marginal optimal transportation problems. We prove a Gaussian saturation result for transportation-information inequalities with quadratic costs. In the case of Wasserstein barycentres, we compute the sharp constant and relate this to the functional Blaschke-Santaló inequality. We also extend some of our results to a more general setting that encompasses the forward-reverse Brascamp-Lieb inequality. https://escholarship.org/uc/item/74f365g1 Sun, 8 Feb 2026 00:00:00 +0000 Wang, Edric https://escholarship.org/uc/item/5wq9z5nc This dissertation presents two frameworks that provide solutions to the problems that arise when contention occurs that better align with the incentives of the underlying system than the state of the art solution. Contention is an important part of large-scale systems. In large-scale systems it becomes untenable to provision for the peak usage due to the sheer scale of resource needs. As a result these systems must handle periods of overuse by allocating the available resources based on the overarching goals of the system. In both of these frameworks we: discuss the nature of the contention that happens in the problem area, outline the desirable properties of an ideal resource allocation scheme, find that existing allocation schemes fall short with respect to those properties, and propose an alternative solution that is better aligned with those properties. In the first framework, RCS, we focus on the wide-area Internet and propose a new set of guidelines for bandwidth allocations... https://escholarship.org/uc/item/5wq9z5nc Sun, 8 Feb 2026 00:00:00 +0000 Brown, Lloyd Anthony https://escholarship.org/uc/item/29r617xb Persistent Cold Air Pools, or PCAPs, are multiday “temperature inversion” events where wintertime high-pressure systems trap cold air at the bottom of valley basins or mountainous regions, lowering the height of the atmospheric mixed layer and preventing the transport of polluted air out of the area. The inversion also concentrates emissions during these events which, when combined with the increased residence time of air in the basin, promotes the formation of secondary (i.e., not directly emitted) air pollution like ozone and PM2.5. In Utah’s Salt Lake Valley, four counties have been designated as in “severe” non-attainment for the PM2.5 U.S. National Ambient Air Quality Standards (NAAQS) because of frequent PCAP events. Measurements show that PM2.5 during these PCAPs is mostly composed of secondary species, especially ammonium and nitrate. In the Salt Lake Valley, ammonium nitrate concentrations during PCAPs can reach up to 70% of total PM2.5. Because ammonium nitrate is a... https://escholarship.org/uc/item/29r617xb Sun, 8 Feb 2026 00:00:00 +0000 Phelan, Cam https://escholarship.org/uc/item/11g9z5nw Large language models have achieved remarkable performance on standardized benchmarks. However, real-world applications demand more than answering questions or executing actions within fully-defined environments—they require models capable of interacting with humans who bring their own context, goals, and preferences. This thesis addresses a central question: how do we design models that can effectively collaborate with humans? I approach this challenge through three complementary perspectives: evaluation, architecture, and learning objectives. First, I introduce DialOp, a benchmark for assessing how language model agents and humans make complex everyday decisions together, providing a rigorous framework for measuring collaborative capabilities. Second, I demonstrate how augmenting language model architectures with memory layers enables efficient knowledge updates, establishing a foundation for continual learning across interactions. Third, I extend these ideas beyond text to... https://escholarship.org/uc/item/11g9z5nw Sun, 8 Feb 2026 00:00:00 +0000 Lin, Jessy https://escholarship.org/uc/item/0gj0q1cz As we move deeper into the "post-Moore" era, two trends have emerged as defining forces in computer architecture. At the architectural level, modern RISC-like ISAs have become the dominant foundation of programmable compute. At the microarchitectural level, continued improvements in performance and efficiency increasingly rely on specialization, rather than general-purpose capability. This thesis argues that RISC-like architectures provide an effective basis for exploiting specialized microarchitectures. First, I develop a representative VLIW architecture and its associated microarchitecture tailored for specialized cores. Through an empirical comparison with a similarly targeted superscalar RISC design, I show that superscalar RISC-style microarchitectures impose a negligible overhead over VLIW designs for specialized cores. I next show that complex vector extensions, often perceived as antithetical to RISC principles, actually follow the modern RISC design philosophy... https://escholarship.org/uc/item/0gj0q1cz Sun, 8 Feb 2026 00:00:00 +0000 Zhao, Jerry https://escholarship.org/uc/item/0bw1z1xb This dissertation investigates multi-scale surface deformation across Northern California using advanced InSAR techniques, focusing on ALOS-2 ScanSAR data to constrain regional-scale velocity fields, characterize surface creep rates along the Maacama and Rodgers Creek faults, and examine triggered accelerations of slow-moving landslides. In Chapter 2, we utilize L-band InSAR data from the ALOS-2 satellite to capture surface deformation across Northern California from 2015 to 2024. Our analysis spans spatial scales, from slow-moving landslides to tectonic processes around the Mendocino Triple Junction. By validating InSAR observations with GNSS velocities and time series, we establish robust regional velocity fields and address the challenges associated with detecting long-wavelength deformation. Our new vertical velocity field encompasses the region north of San Francisco to the Oregon border, nearly spanning the width of the state. Our observations reveal significant aquifer... https://escholarship.org/uc/item/0bw1z1xb Sun, 8 Feb 2026 00:00:00 +0000 Lindsay, Danielle Marie Pilaar https://escholarship.org/uc/item/9x06t63x Ribosomes are macromolecular machines that catalyze cellular protein synthesis. While ribosome structure is well understood in model organisms, we lack a holistic understanding of how ribosome structure and adaptation has changed across divergent organisms. This thesis investigates structural changes in ribosomes and ribosome-binding proteins across prokaryotic species. Chapter 1 reviews our current understanding of ribosome structure, function, and adaption and their relevance in ribosome engineering.Chapter 2 examines rare rRNA sequence variation in the A loop, the binding site for the A-site tRNA in the LSU, found in a subset of hyperthermophilic archaea. In organisms from the archaeal phylum Thermoproteota, two highly conserved uridines in the A loop are instead cytosines. Substitution of these two positions with cytosines in the E. coli ribosome globally increased the stability of the LSU compared to wild type (WT), without affecting the activity of E. coli ribosomes. This... https://escholarship.org/uc/item/9x06t63x Sat, 7 Feb 2026 00:00:00 +0000 Nissley, Amos James https://escholarship.org/uc/item/9sz800bc Over the past two decades, a robust body of scholarship has emerged focusing on the experiences of undocumented college students (Abrego & Negrón-Gonzales, 2020; Bjorklund, 2018; Gleeson & Gonzales, 2012; Gonzales, 2011, 2015; Kantamneni et al., 2016; Muñoz & Vigil, 2018; Negrón-Gonzales, 2015; Pérez Huber & Malagon, 2007; Suárez-Orozco et al., 2015; Teranishi et al., 2015). Within this body of scholarship, there has been a growing interest in students’ experiences leading up to and immediately following their undergraduate graduation (Enriquez et al., 2021; Gonzales, 2015; Morales Hernandez & Enriquez, 2021; Ortiz & Hinojosa, 2010; Pérez Huber, 2015; Salazar et al., 2024; Zamacona, 2022). More recently, there has been a growing interest in the experiences of graduate students (Bedolla, Montiel, & Chen, 2020; Escudero et al., 2019; Freeman & Valdivia, 2021; Kennedy, 2014; Kuczewski & Brubaker, 2014; Lara, 2014; Lara & Nava, 2018; Montiel,... https://escholarship.org/uc/item/9sz800bc Sat, 7 Feb 2026 00:00:00 +0000 Ortega Mendoza, Martha https://escholarship.org/uc/item/9rq9v234 Photosynthetic biohybrids offer a promising platform for selective CO2 reduction, a potential pathway for CO2 utilization necessary in the changing climate. Photosynthetic biohybrids combine the enzymatic machinery of acetogenic bacteria with efficient light harvesters, realizing a highly selective and efficient method of artificial photosynthesis that doesn’t require high overpotentials. However, much is still unknown about the electron transfer pathways of these biotic-abiotic interfaces, and studying the mechanisms from the semiconductor through the cell proves challenging. These unknowns hamper progress within the field, as bottlenecks in the system remain unclear, and without deep mechanistic understanding, system designs cannot target the fastest and most efficient pathways. Microbiological studies have shed light on the enzymes found in the cytosol and membrane of different acetogenic species, and proteomic studies have investigated which proteins are upregulated... https://escholarship.org/uc/item/9rq9v234 Sat, 7 Feb 2026 00:00:00 +0000 Lineberry, Elizabeth Catherine https://escholarship.org/uc/item/9q8557nk Human intelligence is characterized by the ability to set and pursue self-defined goals, yet standard reinforcement learning models assume predetermined objectives. This Dissertation opens with an opinion piece, which highlights the centrality of goals to human learning. The remainder of the Dissertation investigates how internally generated goals influence learning and decision-making through three interconnected projects combining behavioral experiments and computational modeling. The first project demonstrates that incorporating goal achievement signals into value computation significantly improves models of human decision-making through the case study of context-sensitive valuation. The second project proposes a framework for goal-dependent learning that entails a dynamic exchange of information across cognitive processes supporting reinforcement learning, and validates it through a series of experiments and computational modeling. The third project examines how humans construct... https://escholarship.org/uc/item/9q8557nk Sat, 7 Feb 2026 00:00:00 +0000 Molinaro, Gaia https://escholarship.org/uc/item/9mg145cz Large language models (LLMs) have emerged as transformative technology capable of human-level or superhuman performance across diverse tasks, from writing complex software systems to discovering novel algorithms and processing multimodal data. Despite these remarkable capabilities, deploying LLMs at scale presents significant challenges due to their enormous computational and memory requirements. State-of-the-art models contain trillions of parameters and perform tens of thousands of generation steps, executed across large GPU clusters, often under strict latency constraints. These challenges are further compounded by the rapidly evolving model architectures and the growing diversity of hardware accelerators. To address these challenges, this thesis presents the design and implementation of vLLM, an efficient and flexible open-source LLM inference engine. We first introduce PagedAttention, vLLM's core memory management algorithm that enables high-throughput LLM inference.... https://escholarship.org/uc/item/9mg145cz Sat, 7 Feb 2026 00:00:00 +0000 Kwon, Woosuk https://escholarship.org/uc/item/9h03r98f Glaciers are known to vary in erosion rate by many orders of magnitude, eroding almost ten cm/year in some places and a tenth of a millimeter per year in others. Likely a key contributor to this variation in magnitude is quarrying, the process by which glaciers remove up to meter-scale blocks of bedrock and grind them across the bed downstream to cause further erosion as abrasive tools carried by ice and drainage conduits. Conceptually, plucking barely-connected smaller clasts of bedrock will be faster than trying to grind them down with other rocks. Yet ice has a lower intact strength than rock, so if the rocks are too strong and too massive, the ice will fail to break bedrock off and erosion from quarrying will be essentially zero. In fact, quarrying is the most suited subglacial erosive process to create roughness and longitudinal topographic jumps (knickpoints) – roches moutonnées and waterfalls like Lembert Dome and Nevada Falls in Yosemite – because abrasion tends to... https://escholarship.org/uc/item/9h03r98f Sat, 7 Feb 2026 00:00:00 +0000 Theiss, Christopher Roland https://escholarship.org/uc/item/9g54v0s3 Fire whirls have long been a fascination in the field of fire science, but few studies have been conducted with the intention of creating fire whirls at the field scale, and even fewer with associated measurements of emissions. Fire whirls are known to increase the flame length and burning rate for flames over liquid fuel pools, with recent research indicating that they reduce the amount of particulate soot emitted by the flame. These properties make it desirable for the process of in-situ burning, where air pollution and burning duration both must be minimized. In this study, the sensitivity of fire whirl behavior and burning characteristics to the dimensions of a fixed-frame fire whirl generator were first investigated at the laboratory scale. Fire whirls were found to enhance burning rates and reduce emissions when burning slicks of crude oil over water when compared to pool fires, regardless of the thickness of the oil slick. Further increases in slick thickness improved the... https://escholarship.org/uc/item/9g54v0s3 Sat, 7 Feb 2026 00:00:00 +0000 Dowling, Joseph Lee https://escholarship.org/uc/item/99x5g653 This dissertation develops a general anisotropic damage–plasticity framework for modeling the nonlinear, inelastic behavior of concrete structures with special interest in the inelastic, cyclic response of reinforced concrete members. The formulation is built around thermodynamically consistent stress decomposition operators which separate tensile and compressive damage and are capable of describing the unilateral effect. These operators preserve the secant stiffness symmetry and prevent spurious energy dissipation, enabling the damage and plasticity components to be expressed in a classical plasticity format. A compliance-based damage formulation is adopted, allowing physically motivated evolution rules and unifying a wide range of existing damage laws into the proposed framework. The dissertation develops a concrete constitutive model by combining a concrete plasticity formulation with isotropic compressive damage and an anisotropic tensile damage rule. Special attention... https://escholarship.org/uc/item/99x5g653 Sat, 7 Feb 2026 00:00:00 +0000 Velkov, Nikolay https://escholarship.org/uc/item/97m0z6mf The particle nature of dark matter remains one of the largest unanswered questions in the field. Outside of its abundance, most of its properties are unknown and broadly constrained. As such, a wide variety of models have been proposed and explored. Other open questions in the field, like the Strong CP problem, motivate the existence of new particle content that can behave as dark matter. The first half of this dissertation explores the experimental program for the QCD axion as a dark matter candidate. The second half concentrates on a dark QCD-like dark matter model with freeze out via semi-annihilation. We emphasize parameter spaces of interest for each model that are consistent with direct and indirect detection signatures. https://escholarship.org/uc/item/97m0z6mf Sat, 7 Feb 2026 00:00:00 +0000 Moldovsky, Serah https://escholarship.org/uc/item/9143b4q4 Studying modern Artificial Intelligence systems requires new forms of measurement beyond traditional metrics like test-set accuracy. This thesis approaches measurement in AI by focusing on two key categories of metrics that move beyond this standard input-output analysis. First, we take a bottom-up point of view, focusing on the internal representations and mechanisms that underlie model behavior. We introduce a statistical framework to assess interpretability methods, which we use to audit representation similarity metrics and model visualizations. We also show how measurements of neural network internals can help us better understand in-context learning in large language models. Second, we take a more zoomed-out perspective, and construct a novel measurement to track the long-term trajectory of AI capabilities. By stitching together disparate, quickly saturating benchmarks onto a unified scale, we create a long time series of model capabilities that makes it possible... https://escholarship.org/uc/item/9143b4q4 Sat, 7 Feb 2026 00:00:00 +0000 Denain, Jean-Stanislas https://escholarship.org/uc/item/8zv3p1mm The past several decades have seen a paradigm shift in our understanding of the role played by the immune system in cancer. Perhaps the most impactful discoveries have related to basic mechanisms by which T cells drive antitumor immune responses, which in turn have led to an explosion of interest in immunotherapies that seek to promote antitumor T cell responses. This effort has produced several massive successes—most notably the pioneering of checkpoint blockade—but nevertheless most patients’ cancers remain recalcitrant to immunotherapy. This suggests that a deeper understanding of the mechanisms that regulate antitumor T cell responses will further increase the efficacy of cancer immunotherapy. Contributing to this effort was the major driving aim of my thesis work, which focused on the role played by upstream regulators of T cells, namely antigen presenting cells, in driving productive antitumor immune responses. My thesis work encompassed two broad but related approaches.... https://escholarship.org/uc/item/8zv3p1mm Sat, 7 Feb 2026 00:00:00 +0000 Campbell, Timothy https://escholarship.org/uc/item/8zd6k5m8 Foundation models in vision and language have demonstrated strong generalization capabilities by leveraging internet-scale data with scalable training recipes. In contrast, robot policies still face fundamental challenges in robustness, reliability, and generalization when deployed in the unseen real-world scenarios. A key limiting factor is the scarcity of large-scale, diverse, and high-quality robot data. Unlike vision-language datasets, physical robot data is expensive and slow to collect and more difficult to parallelize. These challenges create a fundamental gap between the data and training regimes that enable success in vision-language foundation models and those currently available for robot learning.This dissertation investigates scalable and generalizable approaches toward robot foundation models by improving how robotic learning systems acquire, represent, and utilize data. It argues that progress toward robust robot foundation models requires advances along three complementary... https://escholarship.org/uc/item/8zd6k5m8 Sat, 7 Feb 2026 00:00:00 +0000 Liu, Fangchen https://escholarship.org/uc/item/8sj406f6 Intracellular bacterial pathogens must withstand the host metabolic and immune pressures that threaten their survival, including exposure to the reactive aldehyde methylglyoxal (MG). MG is produced by both bacterial and host metabolism and damages DNA, proteins, and essential metabolites, and therefore needs to be detoxified. The glyoxalase system, which is highly conserved across all domains of life, is the primary pathway that detoxifies this toxic metabolite. Work presented in this thesis demonstrated that macrophages increase MG production upon infection as a consequence of increased glycolytic flux, and that this represented a significant antimicrobial stress encountered by bacterial pathogens in vivo. Increase in MG production was driven by inflammatory cues including TLR-mediated activation and interferon-gamma signaling, that reprogram macrophage metabolism toward aerobic glycolysis. These findings highlight that MG is not merely a byproduct of host metabolism but a regulated... https://escholarship.org/uc/item/8sj406f6 Sat, 7 Feb 2026 00:00:00 +0000 Anaya Sanchez, Andrea https://escholarship.org/uc/item/8s32x9x7 This dissertation develops a unified framework for binding large language models (LLMs) to coherent virtual personas through narrative backstories, enabling scalable, and valid simulation of human attitudes and behaviors. The central idea is that backstories—synthetic life narratives created by LLMs, which encode demographic information, psychological context, and human beliefs, values, and perspectives, both implicitly and explicitly—can serve as conditioning contexts that stabilize and differentiate LLM behavior. Through this lens, the work investigates how backstory conditioning improves representativeness, consistency, and behavioral realism in simulated populations. A key assumption underlying this framework is the use of pretrained base models, whose heterogeneous “mixture of voices” enable backstories to bind naturally through prefix conditioning. This reliance on pretrained models distinguishes the approach from much of the related work on LLM conditioning, which often... https://escholarship.org/uc/item/8s32x9x7 Sat, 7 Feb 2026 00:00:00 +0000 Moon, Suhong https://escholarship.org/uc/item/8qr9c4hm This comprehensive dissertation consists of three interrelated sections: an assessment of pneumococcal carriage risk factors in adults, an examination of Doxycycline Post-Exposure Prophylaxis (doxyPEP) as an intervention for bacterial sexually transmitted infection (STI) transmission, and an evaluation of the role of quarterly STI screenings for HIV pre-exposure prophylaxis (PrEP) users. Collectively, these sections aim to advance our understanding of the transmission and natural history of bacterial pathogens of significant public health importance, with a specific focus on diseases not preventable by current vaccination strategies.1. Pneumococcal Carriage: The bacterial pathogen Streptococcus pneumoniae (pneumococcus) is a prominent cause of severe invasive infections as well as mucosal conditions such as non-bacteremic pneumonia among both children and adults. The burden of pneumococcal disease disproportionately impacts populations with low socioeconomic status, racial minorities,... https://escholarship.org/uc/item/8qr9c4hm Sat, 7 Feb 2026 00:00:00 +0000 Parker, Anna Marie https://escholarship.org/uc/item/8pk5j709 Regulatory T cells (Tregs) are critical guardians within the immune system that are essential for preserving immune tolerance. Accordingly, Treg dysfunction is a common feature of many autoimmune disorders, potentially contributing to both the development and progression of disease. Efforts to overcome this breakdown in peripheral tolerance have been prioritized in the development of novel therapies for these diseases, with a focus on adoptive transfer of engineered Treg populations. However, the cell intrinsic mechanisms that maintain Treg identity and function, especially in response to the stresses of inflammatory tissue environments, remain incompletely understood. Filling these gaps in our knowledge of Treg biology may better inform future advancements of therapeutic intervention for autoimmune and inflammatory conditions. With that goal in mind, my thesis research broadly focuses on the identification of factors that regulate Treg stability and suppressive function, especially... https://escholarship.org/uc/item/8pk5j709 Sat, 7 Feb 2026 00:00:00 +0000 Silveria-Davies, Stephanie https://escholarship.org/uc/item/8bg616k6 Transmission electron microscopy is one of the most powerful characterization techniques available for studying nanomaterials systems at the atomic scale. Recent machine learning advances offer to revolutionize our characterization workflows with unparalleled quantitative analysis and information extraction abilities. Such advanced capabilities could vastly increase the scope and throughput of electron microscopy experiments, but due to the limited quantity of experimental microscopy data which is suitable for training machine learning models, developing new machine learning capabilities for electron microscopy applications can be difficult or practically infeasible. This thesis develops methodologies for training machine learning models for electron microscopy applications entirely via simulation using random structure generation and multislice simulation techniques, focusing on semantic segmentation of high-resolution transmission electron microscopy (HRTEM) images of nanoparticles... https://escholarship.org/uc/item/8bg616k6 Sat, 7 Feb 2026 00:00:00 +0000 Rangel DaCosta, Luis https://escholarship.org/uc/item/89k2x29k The interiors of Uranus, Neptune, and ice giant–like exoplanets remain poorly constrained, particularly regarding the high-pressure chemical processes that govern their structure, evolution, and energy balance. To address this gap, we conducted high-pressure experiments on a range “Synthetic Uranus” (water, ammonium hydroxide, and isopropanol) compositions formulated to approximate cosmic carbon, hydrogen, oxygen, and nitrogen ratios relevant to ice-giant mantles. In particular, we studied carbon-poor compositions with <10% carbon and correspondingly larger abundances of oxygen or nitrogen.Using laser-heated diamond anvil cells with in situ X-ray diffraction and Raman spectroscopy, we observed diamond precipitation between 14 and 55 GPa at temperatures as low as 1,500 K, and for compositions with as little as ~2% carbon. These pressures, temperatures, and carbon abundances are significantly lower than reported for simpler hydrocarbon mixtures. Our results indicate that oxygen-... https://escholarship.org/uc/item/89k2x29k Sat, 7 Feb 2026 00:00:00 +0000 Spears, Daniel Patrick https://escholarship.org/uc/item/86q8h1tp Optical neural networks (ONNs) have emerged as a promising computing paradigm capable of overcoming the energy and latency bottlenecks inherent to electronic deep neural networks. While linear optical operations—such as matrix multiplication and convolution—are naturally energy-efficient and highly parallel, the absence of scalable optical nonlinearities has long impeded the realization of large, deep, and trainable ONN architectures. This dissertation addresses this long-standing challenge by developing the Nonlinear Optical Microdevice Array (NOMA), a hybrid electro–optical platform that provides femtojoule-level optical ReLU activation across more than half a million pixels, and by demonstrating its system-level integration into multilayer optical neural networks.NOMA integrates silicon photodiodes with liquid-crystal electro-optic modulators at the single-pixel level, forming a massive array of self-modulating optical neurons. Detailed characterization of NOMA reveals three... https://escholarship.org/uc/item/86q8h1tp Sat, 7 Feb 2026 00:00:00 +0000 Feng, Qixin https://escholarship.org/uc/item/853442rj Large Language Models (LLMs) have achieved remarkable capabilities across diverse and complex tasks, yet a fundamental decoupling persists between frontier models and the thousands of downstream AI applications with domain-specific needs. This division limits both performance and efficiency: models are trained and served without knowledge of downstream applications. In this dissertation, we design systems and algorithms that couple the model and application layers together—advancing application-aware infrastructure and models.We first present three systems for application-aware infrastructure that optimize cost and performance. Agentix is a serving engine that introduces application-aware scheduling for AI agents. By tracking dependencies between LLM calls and leveraging application-level statistics, Agentix improves end-to-end response-times by 4-15× over state-of-the-art systems like vLLM, which reduces costs. Next, Stylus is a scalable model router that retrieves and composes... https://escholarship.org/uc/item/853442rj Sat, 7 Feb 2026 00:00:00 +0000 Luo, Michael Zhi Yu https://escholarship.org/uc/item/7ph7p499 Information asymmetries, where some agents have information that others do not have, are ubiquitous in social interactions. This poses a challenge to social planners, whose role is to implement outcomes that are beneficial to the constituents they represent. However, planners require information about the state of the world, agents' preferences, and how they make decisions. Hence to bridge this informational gap they must leverage learning. I study three approaches to learning, which include data-driven learning, communication and Bayesian learning, and the design of incentives. The key methodological tools lie in leveraging optimization theory to characterize optimal solutions, game theory to model the role of incentives and strategic behaviour, and market design to formulate desirable properties of solutions and how to implement them. These approaches are applied to different environments, including the development of robust large language models, economic interactions between... https://escholarship.org/uc/item/7ph7p499 Sat, 7 Feb 2026 00:00:00 +0000 Siththaranjan, Anand https://escholarship.org/uc/item/7cp0c37g Air quality regulation in the United States relies fundamentally on a sparse network of monitors that serve a dual purpose: providing information about ambient pollution and triggering mandatory abatement policies under the Clean Air Act. This dissertation investigates the economic efficiency and institutional integrity of this monitoring infrastructure, reframing the monitor network not merely as a passive measurement tool, but as a strategic policy instrument. The analysis proceeds in three parts, moving from a theoretical optimization framework for monitor siting, to an empirical welfare analysis of regulatory incentives, and finally to an assessment of data integrity through novel third-party data sources. The first half of the dissertation addresses the optimal design of the monitoring network. Theoretically, I model a Bayesian regulator facing uncertainty about pollution diffusion, deriving a formal siting rule where the optimal location maximizes the Expected Value... https://escholarship.org/uc/item/7cp0c37g Sat, 7 Feb 2026 00:00:00 +0000 Watt, Aaron https://escholarship.org/uc/item/7793f9mm The advancement of Artificial Intelligence (AI) has shifted the research focus from recognizing static patterns to mastering dynamic environments. This transition relies on two fundamental pillars: time series analysis, which seeks to model and forecast evolving data, and intelligent decision-making, which aims to execute optimal actions under uncertainty. However, applying modern AI to these domains presents unique challenges. Time series data lacks the universal semantic consistency of language and requires robust handling of continuous, high-dimensional inputs. Furthermore, real-world deployment in systems like cloud computing and recommendation requires decision-making models to respect strict latency constraints and incorporate structured human knowledge, which are often neglected in purely end-to-end learning approaches. This dissertation introduces contributions across two themes: (1) Time Series Analysis, where we propose a Lipschitz-regularized GAN for stable continuous... https://escholarship.org/uc/item/7793f9mm Sat, 7 Feb 2026 00:00:00 +0000 Zhang, Yunkai https://escholarship.org/uc/item/72c6h4bq Each major American political party has a history of conflict from within its own ranks.Waged mostly by party factions—groups attempting to influence the party’s activity, leadership structure, or ideological direction—intraparty conflict has appeared to increase in recent decades in the Republican Party. While extant political science research is already unclear about what precipitates intraparty conflict, even less is known about intraparty conflict at the sub-national level. This dissertation seeks to fill these gaps by employing a comparative case study of eight state Republican parties, focused on on three periods of heightened Republican intraparty conflict over the past few decades: the Christian Right era, the Tea Party movement, and the emergence of Donald Trump and Trump Republicanism in more recent years. I hypothesize that the linkages between political groups in states, the design of the state party itself, and status of two-party competition within the state give... https://escholarship.org/uc/item/72c6h4bq Sat, 7 Feb 2026 00:00:00 +0000 Ste Claire, Casey https://escholarship.org/uc/item/727670rb The majority of the world’s languages are categorized as understudied and underserved in Natural Language Processing research. Collectively, these languages are termed “low-resource languages.” As the languages themselves are diverse, so are the cultural, political, and social contexts in which they are spoken. Most modern language technologies—e.g., machine translation, speech recognition—either exclude low-resourced languages entirely or perform poorly with low-resourced language data. In this thesis, I argue that language technologies are parts of complex socio-technical systems, not isolated tools. I lay out my thesis in two parts: Part I presents work that evaluates existing language technologies in the social contexts in which they are deployed. Findings from Part I illustrate how the diverse social contexts in which languages are spoken affect how well our tools perform. Part II presents work that integrates the social contexts of low-resourced languages in how we... https://escholarship.org/uc/item/727670rb Sat, 7 Feb 2026 00:00:00 +0000 Nigatu, Hellina Hailu https://escholarship.org/uc/item/71k6q9m6 The synthetic chemistry of the C2-symmetric hexahydrodicyclopenta[8]annulene (hdcCOT) ligand with tri- and tetravalent rare earth and actinide elements is detailed. Owing to its high degree of alkyl substitution and conformational rigidity, this ligand framework was hypothesized to enhance crystallinity and to improve stabilization of higher oxidation states through increased electron donation. Here we report the preparation and full characterization of a family of homoleptic, neutral complexes of the form M(hdcCOT)2 (M = Th, U, Np, Pu), establishing synthetic routes that are general across the tetravalent actinides. This chemistry is further optimized using cerium as a surrogate system, enabling the isolation of Ce(hdcCOT)2 from as little as 0.5 mg of Ce(3+), thereby demonstrating the compatibility of the method with scarce isotopes. These advances facilitate the synthesis and structural authentication of Bk(hdcCOT)2, providing rare access to a well-defined berkelocene derivative.... https://escholarship.org/uc/item/71k6q9m6 Sat, 7 Feb 2026 00:00:00 +0000 Russo, Dominic https://escholarship.org/uc/item/6kg3g8n6 This work describes the structure of molten mixtures of lithium fluoride and beryllium fluoride for their relevance to fission reactors and fusion machines, as well as the speciation of chromium in molten fluoride salts more broadly. In an operating molten salt system, a multitude of possible chemical species exist including solvent ions, oxidants from moisture, corrosion products, fission products, and activation products. Describing speciation in the salt entails identification of the possible species (their oxidation states, molecular structures of solvation and corresponding coordination numbers) in which an ion or molecule might exist in the melt. Understanding both solvation structure (first-nearest neighbors, and intermediate-range ordering in the melt) and speciation (oxidation states) is relevant to describing the interaction between the melt and its solutes. In addressing both the structure of LiF-BeF2 solvents (in Chapter 2) and chromium solute speciation (in Chapter... https://escholarship.org/uc/item/6kg3g8n6 Sat, 7 Feb 2026 00:00:00 +0000 Williams, Haley https://escholarship.org/uc/item/6hc4q8dr Do Tell argues that the rise of self-consciously queer, feminist and trans literature since the 1960s corresponded with a rejection of opaque modernist aesthetics in favor of the explicit. As restrictions on sexual representation eased, formerly taboo subjects moved from subtext to text and were accompanied by a shift toward more direct, even middlebrow, modes of expression. By “the explicit”, I refer not only to frank literary depictions of sex but also to the direct citation of contemporary sexual politics and subcultures within the text itself. I understand the explicit not merely as a synonym for the pornographic but as a complex representational challenge that has preoccupied contemporary art and literature. Yet despite this turn toward the explicit in contemporary fiction, queer literary criticism continues to index the representation of marginalized sexualities and identities to oblique, avant-garde forms. If much of the literature about sexuality produced since the 1960s... https://escholarship.org/uc/item/6hc4q8dr Sat, 7 Feb 2026 00:00:00 +0000 Sutton, Emily https://escholarship.org/uc/item/6bz095w6 Drug discovery is concerned with designing, modifying, or repurposing biomolecules that can prevent or treat disease. The application of AI for biology is gaining mainstream recognition, notably with the awarding of the 2024 Nobel Prize in Chemistry for protein design and structure prediction. As artificial intelligence (AI) for drug discovery moves beyond proof-of-concept towards the real-world, subtler questions emerge around interpretability, efficiency, and the need for fine grained generation specification to satisfy the complex constraints of biology. This thesis aims to examine these questions, with emphasis on controllability, interpretability, and deployability. First, we examine the compressibility of protein folding model latent spaces. This improves the reusability of pretrained model weights, and helps elucidate the representation information content necessary to solve different tasks. Next, we introduce PLAID, an all-atom protein diffusion model which samples from... https://escholarship.org/uc/item/6bz095w6 Sat, 7 Feb 2026 00:00:00 +0000 Lu, Amy X. https://escholarship.org/uc/item/6944d82d This study aims to understand the implementation and leadership decisions that have gone into the development of Full Service Community Schools in the Bayside Unified School District and to lift up parent voices in the interpretation of whether those efforts are having positive impacts on students and their families. The study was guided by the following two research questions: RQ1 - What are the stated goals of the FSCS initiative in BUSD and what leadership actions have been undertaken by central leadership, principals, and CSDs in the development of the initiative? RQ2 - Based on guardians’ perceptions, in which ways, if any, are these leadership actions supporting the improvement of lived and educational experiences of students, and their families in BUSD? In this case study I conducted five staff interviews including one central office leader and two staff members from each of two focal schools. I also conducted two parent focus groups at each school, totalling 17 parents.... https://escholarship.org/uc/item/6944d82d Sat, 7 Feb 2026 00:00:00 +0000 Fleischman, Guthrie Samuel https://escholarship.org/uc/item/6644g65v A central discourse in the field of superconducting quantum computing surrounds the question of whether the transmon qubit is the correct unit cell for scaling to fault-tolerant quantum computation. Since the transmon was first introduced, there has been a rapid development of more complex qubit designs that promise better noise protection and higher fidelity control. Despite these innovations, all of the largest superconducting qubit processors that exist today rely on the transmon qubit as the fundamental unit cell, and many use the flux-tunable variant to manage many of the difficult scaling constraints. However, the simplicity of the fixed-frequency, single-junction transmon remains highly appealing for building robust, large-scale systems. In this thesis, we address several key scaling challenges that have historically limited fixed-frequency transmon architectures. First, we present an analysis of the frequency-crowding problem for fixed-frequency architectures. We model... https://escholarship.org/uc/item/6644g65v Sat, 7 Feb 2026 00:00:00 +0000 Chen, Larry https://escholarship.org/uc/item/612002q0 Mating systems are central to animal evolutionary biology. The selective pressures associated with mate choice and mating frequency shapes diverse behavioral, morphological, and physiological adaptations designed to maximize fitness and reproductive output. Mammals provide an interesting opportunity to explore how differences in mating systems influence phenotypic diversity. For most mammals, polygamy is the most evolutionary advantageous strategy given that males can maximize their reproductive success by mating with multiple females, while females are limited by fecundity. However, a small percentage of mammals are both socially and genetically monogamous and will only reproduce with one pair-bonded partner. We can uncover key differences between polygynous and monogamous mammals by studying their mating systems comparatively. From these comparative studies we can begin to uncover important differences in mating strategies, sexual selection intensity and reproductive success.Peromyscus... https://escholarship.org/uc/item/612002q0 Sat, 7 Feb 2026 00:00:00 +0000 Horr, Daisy Mae https://escholarship.org/uc/item/60z6566v This dissertation investigates the electrochemical behavior of porous electrodes, specifically on those used in fuel cells. These technologies can enable a transition towards sustainable and renewable energy sources, offering significant potential in reducing Carbon emissions and advancing energy efficiency. However, despite their promise, a number of critical challenges remain in optimizing the performance of these devices, particularly with regard to the materials and design of the electrodes themselves. Central to these challenges is the transport of ions, electrons, and gases within the porous structure, which directly impacts the efficiency, stability, and longevity of the electrochemical systems. The performance of porous electrodes is inherently tied to their microstructural design, which governs not only their electrochemical behavior but also their ability to facilitate efficient transport and minimize losses during operation. This dissertation explores the role of porous... https://escholarship.org/uc/item/60z6566v Sat, 7 Feb 2026 00:00:00 +0000 Srivastav, Harsh https://escholarship.org/uc/item/5wm601n8 The technique of laser cooling, and its application in the generation of ultracold gases of atoms, has reshaped AMO physics and related fields such as quantum many-body physics, precision metrology and quantum information science. However, only a fraction of the atoms in the periodic table have been laser cooled, and by realizing ultracold gases of different classes of elements, new phenomena can be explored. This work describes the laser cooling of neutral titanium (Ti) atoms, beginning with the theoretical proposal and ending with the implementation of the first magneto-optical trap (MOT) of Ti atoms.After describing how laser cooling proceeds in previously laser cooled atoms, the atomic structure of neutral Ti is examined to determine how laser cooling can be achieved with minimal experimental complexity. The proposed Ti laser cooling scheme is extended to twelve other transition metal elements that had not previously been considered as candidates for laser cooling, building... https://escholarship.org/uc/item/5wm601n8 Sat, 7 Feb 2026 00:00:00 +0000 Eustice, Scott Ethan https://escholarship.org/uc/item/5vt4t2fj This dissertation includes three chapters, all of which focus on different aspects of environmental economics. In the first chapter I model the growth of the petrochemical industry to quantify the impacts that the shale gas boom and various proposed energy policies had or would have on aggregate petrochemical capacity and emissions in the United States. I estimate the relationship between input and output prices and firm capacity investment decisions in the petrochemical industry using a competitive, dynamic model of plant capacity and location choice. Using my estimated parameters and a series of counterfactual simulations, I find that the shale gas boom led to an increase in the size of the industry of between 1% and 12% per year over the years 2012 to 2024 and that subsidy-based energy policy and aggressive LNG export policy would lead to average annual aggregate capacity changes between 0.18% and 3.6% and -0.9% and -0.24%, respectively. I also find that a subsidy-based environmental... https://escholarship.org/uc/item/5vt4t2fj Sat, 7 Feb 2026 00:00:00 +0000 Marcoux, Kendra https://escholarship.org/uc/item/5t55p5kx Accelerated fuel qualification for advanced nuclear systems requires mechanistic understanding coupled with rigorous uncertainty quantification to compress traditional timelines. This dissertation investigates whether mechanistic models grounded in first-principles physics, systematically validated against experimental data across multiple operating regimes, and calibrated through Bayesian inference can substantially reduce epistemic uncertainty in fuel performance predictions while establishing defensible operational boundaries. The work applies this hypothesis to uranium mononitride (UN) fuel for micro-reactor deployment, addressing the challenge that historical UN irradiation data exhibit significant scatter and gaps at high burnup and elevated temperatures relevant to modern micro-reactor designs.A mechanistic model for fission gas release and swelling model was investigated in the BISON fuel performance code. The model explicitly captures temperature-dependent diffusion mechanisms... https://escholarship.org/uc/item/5t55p5kx Sat, 7 Feb 2026 00:00:00 +0000 Miller, Zachary Aaron https://escholarship.org/uc/item/5sr662vz Zero-Knowledge Virtual Machines (zkVMs) enable scalable blockchain computation by allowing off-chain execution with on-chain verification through succinct proofs. A zkVM consists of a frontend, which compiles programs into constraint systems, and a backend, which generates cryptographic proofs that these constraints are satisfied. However, zkVMs remain prohibitively expensive: proof generation can be one million times slower than native execution. Since backend proving time scales linearly with constraint count, efficient frontends that minimize constraints are essential.This dissertation identifies key bottlenecks in zkVM frontends and develops three techniques for constraint reduction, organized into two themes.Constraint-Friendly Cryptography. Cryptographic primitives, especially hash functions, dominate constraint counts in real zkVM workloads. We develop two complementary strategies. Substitution applies when the application permits flexibility, such as zkVM memory consistency... https://escholarship.org/uc/item/5sr662vz Sat, 7 Feb 2026 00:00:00 +0000 ZHANG, YINUO https://escholarship.org/uc/item/5n29t422 Although originating in statistical physics, the study of Gibbs measures has found a second home in modern probability theory, an interaction that has resulted in a host of new problems and techniques. This thesis aims to illustrate and showcase some of these perspectives through four examples in different subdomains around this broad theme.The problems under consideration include (a) large deviations for subgraph counts in sparse random graphs in Chapter 2, (b) analysis of the number of independent sets in random subgraphs of the hypercube in Chapter 3, a model supporting infinitely many phase transitions, (c) universality of the Gibbs measure arising from exponentiating log-correlated fields, referred to as the Gaussian multiplicative chaos measure (in the Gaussian case) in Chapter 4, and finally (d) a spectral approach to understanding the ground state in the Edwards-Anderson spin glass in Chapter 5.Along the way, we will develop and discuss a host of general-purpose probabilistic... https://escholarship.org/uc/item/5n29t422 Sat, 7 Feb 2026 00:00:00 +0000 Basu Roy Chowdhury, Mriganka https://escholarship.org/uc/item/5fx5f4vh High-effort logic synthesis is a new paradigm in logic synthesis. It is motivated by recent design trends, including AI accelerators and edge computing devices, where more resources, compared to the traditional design flows, can be allocated for optimization of some blocks due to the high reusability and repeatability of these blocks. With more runtime available, we revisit a concept of redundancy analysis, which is a formal approach to remove redundancies in logic circuits, but has been avoided due to its computational complexity. On the other hand, redundancy analysis can also be applied to redundancy addition. Iteratively performing redundancy addition and removal can open up an enormous search space, leading to improved quality compared to other conventional methods, given sufficient exploration time. Moreover, it can derive more unique structures that potentially improve some indirect metrics, such as verification runtime, as the reasoning process is influenced by structural... https://escholarship.org/uc/item/5fx5f4vh Sat, 7 Feb 2026 00:00:00 +0000 Miyasaka, Yukio https://escholarship.org/uc/item/5cq0q1tv This study examines how shareholder votes influence executive disclosures, leveraging the mandatory adoption of Say-on-Pay (SoP) in the United States, which requires regular shareholder votes on executive compensation. My identification strategy relies on a difference-in-differences design that exploits intra-firm variations in SoP exposure among executives participating in earnings conference calls. I find that executives subject to SoP provide abnormally optimistic disclosures to potentially influence shareholders’ perceptions of their performance and voting decisions. This tone inflation is associated with more favorable SoP voting results but subsequent declines in firm value, suggesting it is strategic. Further analysis indicates that the documented tone inflation is driven by executives’ heightened career concerns and compensation more closely tied to stock performance after SoP adoption. Strong internal and external monitoring partially mitigates the executives’ strategic... https://escholarship.org/uc/item/5cq0q1tv Sat, 7 Feb 2026 00:00:00 +0000 Zhao, Zhufang (Summer) https://escholarship.org/uc/item/5bv4m03h This dissertation explores the work of four contemporary Latinx artists—Firelei Báez (b. 1981), Enrique Chagoya (b. 1953), Sandy Rodriguez (b. 1975), and Juan Sánchez (b. 1954)— who interrogate the co-imbrication of territory, culture, and identity. These artists engage with contested spaces they know intimately, including the U.S.-Mexico border, the boundary between Haiti and the Dominican Republic, and Puerto Rico. By reappropriating historical artifacts and contemporary documents such as maps, books, newspaper articles, and state surveillance files, their works illuminate how current territorial conflicts and identity formations are linked to discovery narratives and early modern cartography.While their practices examine how racialized, gendered bodies become sites for these spatial conflicts, they also deconstruct Eurocentric representations of land, subverting their discursive power. This project situates these artists within a complex web of local and global histories, spanning... https://escholarship.org/uc/item/5bv4m03h Sat, 7 Feb 2026 00:00:00 +0000 Pastorelli-Sosa, Angela Giovanna https://escholarship.org/uc/item/56w664td This work presents first experimental and theoretical steps towards establishing a novel architecture for quantum information processing (QIP) using trapped electrons in Paul traps. Electrons are attractive for quantum computing because they are extremely light and are natural two-level spin systems (qubit) with a large enough magnetic moment to be manipulated with well-established microwave technology and thermal reservoirs.To that end, we investigate the feasibility for all-electric quantum computing with trapped electron spin qubits. We identify the necessary experimental steps to realize a trapped electron quantum processor through a concrete design proposal, including trapping, cooling, electronic detection, spin readout, and single- and multiqubit gate operations. We numerically study the electron dynamics in linear Paul traps to find the threshold temperature to form two-electron Wigner crystals. We also derive the gate infidelities due to motional errors for trapped electrons... https://escholarship.org/uc/item/56w664td Sat, 7 Feb 2026 00:00:00 +0000 Yu, Qian https://escholarship.org/uc/item/4n24k4nw All organisms possess defense systems ranging from constitutive barriers to complex recognition/response networks coined immune systems. This thesis investigates immune-like systems in two organisms lacking canonical immune components or defined innate immune systems: duckweed (Lemnaceae) and the filamentous fungus Neurospora crassa. In duckweed, which has lost key plant immune components including most NLRs and the EDS1/PAD4 signaling hub, we identified lineage-specific helper Pseudomonas from the native microbiome of Landoltia punctata that protects against Pseudomonas syringae DC3000. This mutualistic novel Pseudomonas was not observed to protect Arabidopsis thaliana. For fungi, we provide an extensive synthesis of literature connecting putative innate immune-like responses to diverse microbial pressures, revealing significant gaps in mechanistic understanding despite rich comparative hypotheses. We then develop further N. crassa as a model for studying fungal responses to... https://escholarship.org/uc/item/4n24k4nw Sat, 7 Feb 2026 00:00:00 +0000 Stark, Frances Grace https://escholarship.org/uc/item/4mz2h65n Embodied agents must perceive the world reliably even as their raw observations change constantly. While the visual appearance of an object fluctuates wildly due to perspective shifts, lighting conditions, and occlusions, its underlying identity remains the same. Formally, this principle is captured by the notion of symmetry: transformations that alter the appearance of an object but preserve what it is.Historically, computer vision has addressed symmetries through several approaches: encoding rigid invariances directly into network architectures, using data augmentation to teach invariance by brute force, or scaling models with massive datasets in the hope that invariances will emerge implicitly. However, architectural approaches are too rigid for the messy transformations that embodied agents encounter, augmentation is data-inefficient and fails to generalize to rare classes, and scaling remains brittle to out-of-distribution examples.This dissertation argues that robust embodied... https://escholarship.org/uc/item/4mz2h65n Sat, 7 Feb 2026 00:00:00 +0000 Singhal, Utkarsh https://escholarship.org/uc/item/4j99k3st Pebble Bed Reactors (PBRs) are a class of advanced nuclear reactors that utilize spherical graphite fuel elements, known as pebbles, typically ranging from 4 to 6 cm in diameter. Each pebble contains thousands of micrometre-scale tristructural isotropic (TRISO) fuel particles, designed with multiple protective layers to retain fission products and withstand high temperatures and burnups. Unlike conventional static-core reactors, PBRs operate with continuous fuel motion: pebbles are inserted at one end of the core, pass through the active region, and are discharged at the other end. Upon discharge, each pebble is inspected to assess burnup and mechanical integrity, after which it may be reinserted into the core or replaced by a fresh pebble, typically based on a burnup threshold. The random nature of reinsertion positions leads to a broad range of pebble trajectories and, consequently, diverse irradiation histories. This reactor design offers several advantages that align... https://escholarship.org/uc/item/4j99k3st Sat, 7 Feb 2026 00:00:00 +0000 Jantzen, Ludovic https://escholarship.org/uc/item/4h53v4zj Geysers, hot springs that have periodic steam-driven eruptions, are a natural window into fluid flow within the crust and provide insight into common processes at volcanoes. Their eruptions also capture public interest, and millions of tourists visit geyser fields each year. Geysers are ephemeral features and even those with reputations for being regular performers are susceptible to perturbations from the hydrologic cycle, climate change, and earthquakes. Study of geysers and their associated hydrothermal systems is hampered by a lack of long-term monitoring, making it difficult to assess the factors that control their activity on timescales of days to decades. I take advantage of data available from an unusually well-monitored geyser basin in Yellowstone, United States to understand if, why, and how certain factors affect geysers. Steamboat Geyser is currently the tallest active geyser on Earth with a water jet that can reach a height of 137 m. It is located in the dynamic... https://escholarship.org/uc/item/4h53v4zj Sat, 7 Feb 2026 00:00:00 +0000 Reed, Mara https://escholarship.org/uc/item/4g61f3x5 Medicaid is the nation’s healthcare safety net, providing healthcare coverage to some of the most vulnerable individuals and families in the United States. It was once a publicly administered program where states directly contracted with providers to deliver care. However, Medicaid has transformed into a system largely operated by private, for-profit managed care organizations (MCOs). Today, those MCOs hold immense influence over how care is delivered to millions of low-income Americans. There is a need to better understand how this shift affects patient access, quality of care, and engagement with the Medicaid program. This dissertation spans from a national overview of Medicaid managed care to an in-depth analysis of a specific case management program. The first paper is a multi-state analysis of the current Medicaid managed care landscape, characterizing the different managed care organizations, where they operate across the United States, and the populations they serve. The... https://escholarship.org/uc/item/4g61f3x5 Sat, 7 Feb 2026 00:00:00 +0000 Lo, Christine https://escholarship.org/uc/item/47r5v0z3 The suppression of Vodou was waged heavily during the U.S. occupation of Haiti, from 1915 to 1934. Amidst its prohibition, a substantial flow of Vodou drums and other sacred objects circulated within and out of the country, and ultimately into museums abroad. Bringing together anthropologies of collecting, theoretical investigations of the ways objects move, and interdisciplinary Vodou scholarship, I trace the itineraries of these materials out of Haiti, from museums and archives in the United States and in Europe, tracking how, where, and why they circulated along the way. Part I investigates the adjacent forces of U.S. military occupation and U.S-American anthropology, as well as the enduring presence of Vodou, during the early- and mid-century period of collecting. Part II brings together object biography and itinerary methods to trace the circulation of two sacred classes of objects within and out of Haiti. I argue that, following Paul Mocombe’s concept of “the Vodou ethic,”... https://escholarship.org/uc/item/47r5v0z3 Sat, 7 Feb 2026 00:00:00 +0000 Boucicaut, Pascale Danielle https://escholarship.org/uc/item/43n2v550 Flowering plants are adapted to the physical conditions of the environments they live in, including temperature, precipitation, atmospheric CO2 concentration, and light level. In cooler and drier habitats, leaves within the plant communities tend to be smaller with condensed venation and toothed margins. While under warmer and wetter conditions, leaves tend to be larger, have ‘drip tips’, and entire margins. Atmospheric CO2 concentration and light regulate leaf size, stomatal patterns, epidermal cell size and shape in developing leaves. These so-called leaf traits are preserved in the plant fossil record—for instance in cuticles, the decay-resistant waxy layer that envelops leaves—and can be used to infer local conditions at the time the plants were alive. This kind of information can be used to better understand Earth’s history, the evolution of ecosystems, and to help predict the effects of global warming. Paleobotanists rely on preserved characteristics in fossil leaves to... https://escholarship.org/uc/item/43n2v550 Sat, 7 Feb 2026 00:00:00 +0000 Wagner, Jennifer Dawn https://escholarship.org/uc/item/41p5j7r9 This work seeks to address the underlying question of how does caloric restriction (CR) drive lifespan extension. In doing so, this work identifies a class of proteins called vitellogenins as a major mediator of CR lifespan extension in C. elegans. Secondly, this work introduces a toolkit to study ribosome and translation control of longevity and characterizes their differential effects on development and aging rates. In the first chapter, I will introduce aging as a field of study. I will introduce key components needed to understand the following chapters including CR, vitellogenins, and translation as they pertain to aging. In the second chapter, I describe uncovering a major role for vitellogenins in mediating the lifespan extension effects during CR. By taking a label-free microscopy approach, we generated a dataset comprising of whole-body SRS images of C. elegans over age along with changes in diet and genetics. One of the most dramatic phenotypes we observed... https://escholarship.org/uc/item/41p5j7r9 Sat, 7 Feb 2026 00:00:00 +0000 Yang, Bowen https://escholarship.org/uc/item/3qv301pk Biology offers a sustainable alternative to chemical methods across industries, but cost remains prohibitive for most applications. Protein optimization is a key bottleneck: enzymes operating in non-native environments often exhibit orders of magnitude lower activity than theoretical maxima. Rational protein engineering methods require cumbersome structure prediction tools and expert knowledge. Directed evolution requires high-throughput screening capabilities unavailable for most industrially relevant targets. Chapter 2 presents a case study in biofuel production where we engineered Pseudomonas putida to consume hydrogen as a cheap secondary feedstock. We show with flux balance analysis a theoretical 94% yield improvement, and upon construction of the strain saw a 12% increase in biomass yield. This indicates that the pathway is viable but the activity is too low. The experimental constraints - requiring hours of gas chromatography analysis per variant - exemplify why many critical... https://escholarship.org/uc/item/3qv301pk Sat, 7 Feb 2026 00:00:00 +0000 Roberts, Jacob Buswell https://escholarship.org/uc/item/3pm8k2kv Generative AI systems enable creation---of text, images, code, and more---through humanlike language interfaces, promising to transform how we design and create. Yet these interfaces often lead users to apply intuitions about understanding and reasoning that then lead them astray. In this dissertation, I examine these misaligned intuitions and how they can get in the way of better leveraging unique Generative AI capabilities, such as proposing large sets of diverse alternatives. Through case studies in chatbot authoring and interactive programming, I illustrate (1) how humans approach instructing LLM behavior by drawing on human-human instructional experiences---and how those approaches can fail to serve users' goals; (2) how designing for generative AI systems is uniquely challenging; and (3) ways that computer program design tools can leverage AI's strengths while addressing these tensions, through explicit structures and proactive exploration. https://escholarship.org/uc/item/3pm8k2kv Sat, 7 Feb 2026 00:00:00 +0000 Zamfirescu-Pereira, J.D. https://escholarship.org/uc/item/3pd780rz Ghosts in the Machine is a portfolio of three electroacoustic compositions and performance designs developed throughout my doctoral studies at UC Berkeley’s Center for New Music and Audio Technologies (CNMAT). These works explore the creative possibilities and philosophical implications of working with technology as a creative and collaborative force. In these works, technology is used to enhance and augment live instruments in the traditional concert setting. Technology therefore becomes an active collaborator, transforming not only the final output but also the creative process itself.This triptych is divided into three movements, each for a different instrumental force. Nested Catastrophes is scored for percussion, electronics, lights, and two auxiliary instrumentalists. This movement opens with an installation component, wherein the two auxiliary musicians perform scraping gestures inside of amplified terrariums filled with small stones. The sound of these terrariums are combined... https://escholarship.org/uc/item/3pd780rz Sat, 7 Feb 2026 00:00:00 +0000 Harlan, Andrew J https://escholarship.org/uc/item/3p0686rx The study of topological quantum field theories lies at the center of a rich interplay between algebraic geometry, algebraic topology, and mathematical physics. The Verlinde formula is a fundamental example: it computes the indices of certain line bundles on the moduli stack of principal G-bundles over a Riemann surface Σg. These indices can be interpreted as partition functions of 3d Chern–Simons theory on Σg × S 1 , and were reformulated by Freed, Hopkins, and Teleman in purely topological terms as pushforwards in twisted equivariant K-theory. Teleman and Woodward generalized the Verlinde formula to compute indices for a class of generators in the K-theory of the moduli of G-bundles on Σg.This thesis generalizes the Teleman–Woodward index formula from fixed Riemann surfaces to families of curves. Focusing on the abelian case G = GL(1), we explicitly compute the resulting index, which can be interpreted as the propagator of a 2d gauged Gromov–Witten theory. Our computation combines... https://escholarship.org/uc/item/3p0686rx Sat, 7 Feb 2026 00:00:00 +0000 Lee, Catherine https://escholarship.org/uc/item/3kr7754d The increasing frequency of drought stress due to global climate change necessitates novel, sustainable strategies to enhance crop resilience, with the rhizosphere microbiome representing a critical asset. This body of work utilizes the drought-tolerant crop Sorghum bicolor as a model to mechanistically dissect the assembly and function of beneficial, stress-adapted microbial communities, culminating in the rational design of effective Synthetic Microbial Communities (SynComs). We first establish a link between metabolites and microbial community structuring. Through integrated metabolomic and microbial profiling, we demonstrate that drought enrichment correlates with a rapid shift in root chemistry (Chapter 2), notably the significant accumulation of the stress-responsive metabolite pipecolic acid, a known precursor to the defense hormone N-hydroxy-pipecolic acid. Exogenous application of pipecolic acid confirmed its potential to selectively impact Actinobacterial taxa, suggesting... https://escholarship.org/uc/item/3kr7754d Sat, 7 Feb 2026 00:00:00 +0000 Pettinga, Dean https://escholarship.org/uc/item/3k31z71v Globally, approximately 730,000 individuals die by suicide each year. Most of these suicide deaths occur in low-and middle-income countries. The United Nations Sustainable Development goals include a target to reduce the global suicide rate by one-third by 2030. However, progress towards this goal remains slow due to pervasive stigma that limits access to mental health care and the continued criminalization of suicide in nearly 20 countries worldwide. However, in Kenya, there has been an 18% reduction since 2020. The current age-standardized suicide rate is 4.58 per 100,000 population. While the declining trend is notable, each suicide affects an estimated 65 to 135 individuals who have either been directly exposed or indirectly affected by the loss. These suicide bereaved individuals usually refrain from seeking mental health care and social support due to stigma, increasing their vulnerability to depression and suicidality. The primary objectives of this study were to:... https://escholarship.org/uc/item/3k31z71v Sat, 7 Feb 2026 00:00:00 +0000 Chatterjee, Purba https://escholarship.org/uc/item/3283h23w How do patterns of gene expression give rise to the structural and functional organization of the human brain? A central challenge in neuroscience is to bridge levels of analysis, from genes to cells, circuits, large-scale networks, and even behavior so that we can better understand how molecular mechanisms contribute to the architecture of the brain. Transcriptomics paired with neuroimaging information offers a powerful tool for addressing this challenge by linking variation in gene expression to both anatomical and functional features. The uniting theme across this thesis is the identification of transcriptomic principles that underlie brain organization across regions and scales.In Chapter 1, I examine how transcriptomic differences support a modern cytoarchitectonic parcellation of the human cerebral cortex, revealing genetic correlates of cortical thickness, myelination, and cell density.In Chapter 2, I extend this framework to the human cerebellum, testing whether gene expression... https://escholarship.org/uc/item/3283h23w Sat, 7 Feb 2026 00:00:00 +0000 King, Leana https://escholarship.org/uc/item/2z98c35v This dissertation demonstrates that the rise of state-led water management, or “hydrocracy,” underpinned new forms of political authority and technocratic governance in the United States and India from the late nineteenth through the late twentieth centuries. By tracing the evolution of ambitious hydrological infrastructures and the bureaucracies that constructed and governed them, this dissertation argues that dam building was an essential part of state-building, legitimation, and resistance in both the Western and postcolonial world. Hydrocracies in both the United States and India grew out of and actively shaped the political and ideological currents of their respective eras. In the American West, agencies like the Bureau of Reclamation emerged as central agents in transforming arid landscapes and extending federal sovereignty. These agencies also often dismissed Indigenous claims to water to prioritized scientific expertise and national progress. In India, postcolonial... https://escholarship.org/uc/item/2z98c35v Sat, 7 Feb 2026 00:00:00 +0000 Venkatasubramanian, Varsha https://escholarship.org/uc/item/2z20c4w6 This dissertation presents a theoretically informed study of A'ingae (or Cofán, ISO 639-3: con), an Amazonian language isolate spoken in Ecuador and Colombia. The first part of the dissertation is descriptive. In a chapter on phonology, I touch on the language's phonotactics, laryngeal agreement, diphthongal processes, recent sound changes, nasal spreading, as well as prosody and glottalization. In a chapter on morphosyntax, I describe the structure of A'ingae sentences, including matrix and (co)subordinate clauses, auxiliary verbs, serial verb constructions, WH-movement, and second-position clitics, as well as the morphology of A'ingae verbs, including the meanings of verbal morphemes and their co-occurrence restrictions. The second part of the dissertation is theoretical, and focuses on the verbal morphophonology of A'ingae metrical stress and glottal stops. A detailed study of the operations that these features trigger (and undergo) reveals an interaction of two morphophonological... https://escholarship.org/uc/item/2z20c4w6 Sat, 7 Feb 2026 00:00:00 +0000 Dąbkowski, Maksymilian Michał https://escholarship.org/uc/item/2vj4p4nw Philosophers tend to think that art is good and important. And they think it’s good and important because it has transformative potential: it can disrupt norms, reshape perception, and generate meaning. But this hasn’t always been the case. Historically, art was often subordinate to external ends (say, religious or moral) or even viewed as dangerous. Plato, for instance, argued that art is deceptive, a mere play of appearances that detracts from knowledge and truth. So, what changed?My dissertation, The Play of Appearances, answers this question. Specifically, I argue that a new appreciation of aesthetic appearances emerges after Kant, reclaiming them not as illusion or deception, but as the very condition for aesthetic meaning and transformation. Art, though still pictured as a play of appearances, is now valued in a way that turns Plato on his head. It works “honestly” with appearances, allowing them to appear as appearances, rendering them productively available. Rarely examined... https://escholarship.org/uc/item/2vj4p4nw Sat, 7 Feb 2026 00:00:00 +0000 Cowan, Scott https://escholarship.org/uc/item/2dq7342r Germline mutations are the primary source of genetic variation, playing a key role in evolution across species. Studying germline mutations allows for insight into heritable disease variation, calibrating molecular clocks, and dissecting examples of natural selection. My thesis aims to implement an approach inferring germline mutation rates across species while reliably estimating error, apply this method to outlier species to understand how biological factors impact mutation rate, and finally characterize a mutation that has been strongly selected for in humans. First, I executed a reliable, well-calibrated pipeline to infer de novo germline mutations (DNMs) using whole genome sequencing of pedigrees. Many studies have estimated mutation rates across species, but each study implements their own unique pipeline and fails to account for both false discovery and false negative rates. Thus it is impossible to disentangle the biological effects on mutation rate from technical bias.... https://escholarship.org/uc/item/2dq7342r Sat, 7 Feb 2026 00:00:00 +0000 Marohn, Meaghan https://escholarship.org/uc/item/17277964 High-accuracy electronic structure methods such as variational Monte Carlo (VMC) and Aufbau suppressed coupled cluster (ASCC) theory can surpass the reliability of traditional quantum chemistry approaches when modeling complex excited states such as charge transfer, doubly-excited states, and Rydberg states. However, their steep computational costs limit their practical application to small systems sizes. This thesis develops and tests new strategies that substantially reduce these costs while preserving their accuracy, thereby enabling simulations of larger and more complex chemical systems. First, we introduce an approach for augmenting Gaussian atomic orbitals with correct nuclear cusps that, when applied to VMC calculations of a small benchmark set, significantly improves the statistical efficiency and subsequent cost. Next, we investigate ways to reduce the per-sample cost of electron move propagation in VMC by exploiting the locality of electron correlation in insulating... https://escholarship.org/uc/item/17277964 Sat, 7 Feb 2026 00:00:00 +0000 Bumann, Sonja Katherine https://escholarship.org/uc/item/14f2h2rf What accounts for subnational variation in the implementation of crisis-related policies? While prior scholarship has emphasized the centralization of government responses during economic downturns and national security threats, recent analyses of pandemic governance have highlighted the role of subnational governments—particularly in federal systems and contexts of weak central authority. These studies have pointed to factors such as political polarization, party alignment, policy diffusion, and state health capacity. Yet the role of local economic structures in shaping crisis response remains under-examined.This dissertation explores how subnational governments respond to emergencies at the municipal level. Drawing on Brazil’s experience with the COVID-19 pandemic, I investigate two central questions. How do economic profiles influence the implementation of public health policies? What role do businesses play in checking the power of the executive during a state of emergency?... https://escholarship.org/uc/item/14f2h2rf Sat, 7 Feb 2026 00:00:00 +0000 Martinez, Adan Steve https://escholarship.org/uc/item/1413m1b2 The deployment of mobile robots in unstructured, real-world environments demands robust solutions for both perception and control. Conventional approaches often rely on precise, expensive sensors and rigid control structures, which can be brittle under noise and constraints. This dissertation explores a shift toward learning-based frameworks to enable robust, efficient, and self-reliant autonomy across aerial and legged platforms. To address the challenge of aerial navigation in global positioning system-denied environments, we first introduce a real-time, image-based geo-localization framework. By utilizing a database of rendered synthetic views from satellite imagery and topography, this two-stage pipeline performs global and local descriptor matching to recover the 6-degree-of-freedom vehicle pose. The system achieves onboard inference at approximately 1.1Hz with a position root-mean-square error of 2.8m, demonstrating robustness to camera pitch variations, lens flare,... https://escholarship.org/uc/item/1413m1b2 Sat, 7 Feb 2026 00:00:00 +0000 Chen, Shuxiao https://escholarship.org/uc/item/0v92688b Aphasia sometimes spares singing when speech is impaired, yet the neural basis of this dissociation is unclear. This dissertation investigates how speech and song draw on overlapping but partially dissociable circuits that differ in hemispheric involvement, motor planning demands, and reliance on feedforward vs. feedback control. Chapter 1 frames the historical observations of singing and speaking. Chapter 2 uses a single-case study of post-stroke apraxia of speech to demonstrate selective disruption of speech motor planning. Chapter 3 characterizes the speech–language network in Broca’s aphasia via lesion mapping, emphasizing network disruptions to speech and language, using a large cohort of post-stroke individuals with aphasia. Chapter 4 tests a post-stroke cohort with a sentence production paradigm isolating motor speech, syntax, and lexical retrieval; benefits of singing emerge for motor planning and syntax but not naming. Chapter 5 analyzes intracranial EEG during speaking... https://escholarship.org/uc/item/0v92688b Sat, 7 Feb 2026 00:00:00 +0000 Pracar, Alexis Lucille https://escholarship.org/uc/item/0p6775pq Although information visualizations are widely used to communicate data-driven insights, the role of text in these visualizations remains understudied. Titles, captions, annotations, and other text components are pervasive and influential in real-world visualization designs. Despite the prevalence of text, the visualization field has limited empirical evidence, theoretical structure, and design guidance for understanding how text and visual elements work together. This dissertation addresses these gaps by investigating the role of text in visualization from both reader and designer perspectives. The first part of the dissertation presents three empirical studies examining how text shapes reader experience and interpretation. These studies show that readers prefer text-rich visualizations and that text influences some interpretations (e.g., takeaways and perceptions of bias) but not others (e.g., predictions). The second part investigates how visualization designers use and reason... https://escholarship.org/uc/item/0p6775pq Sat, 7 Feb 2026 00:00:00 +0000 Stokes, Chase https://escholarship.org/uc/item/0hz5j1vn This dissertation investigates Indigenous foodways and land-use practices in the coastal highlands of the Santa Cruz Mountains through a collaborative eco-archaeological approach. Focusing on the upper San Vicente Creek watershed, the study integrates low-impact excavation and flotation analysis at two mid-elevation habitation sites (CA-SCR-67 and CA-SCR-362), along with landscape survey of 182 acres, and comparison with published lowland datasets. Radiocarbon dating of short-lived botanicals and fauna places CA-SCR-67 in the Middle Period (~1000 BCE–AD 1050 CE, with terminal Early Period use) and CA-SCR-362 in the Late Period/protohistoric–historic era (~AD 1630–1900+). Standardized densities from water flotation indicate distinct montane foodways: Middle Period CA-SCR-67 preserves low-density grass-seed assemblages (0.92 n/l, 36 % Poaceae) comparable with grass seed densities of lowland sites. There is a change during the Late Period at CA-SCR-362, exhibiting nutshell... https://escholarship.org/uc/item/0hz5j1vn Sat, 7 Feb 2026 00:00:00 +0000 Apodaca, Alec Jazz-Rio https://escholarship.org/uc/item/0bw7c187 Modern real-world evidence generation increasingly requires statistical methods that can integrate heterogeneous data sources and efficiently learn from partially observed data. This dissertation develops new targeted learning methodologies that advance these goals in two commonly arising settings: the integration of randomized controlled trial (RCT) data with external real-world data (RWD) and two-phase sampling designs. Across three projects, we leverage targeted maximum likelihood estimation (TMLE) to construct robust, efficient, and practically deployable estimators for causal inference in modern clinical and observational studies.In the first project, we consider the problem of estimating the average treatment effect (ATE) when both randomized controlled trial (RCT) data and external real-world data (RWD) are available. We decompose the ATE estimand as the difference between a pooled-ATE estimand that integrates RCT and RWD and a bias estimand that captures the conditional... https://escholarship.org/uc/item/0bw7c187 Sat, 7 Feb 2026 00:00:00 +0000 Qiu, Tian https://escholarship.org/uc/item/0bf6d01b Predicting biological properties directly from molecular data remains a central challenge in computational biology. This thesis addresses this problem across molecular and phenotypic scales, beginning with proteome-wide prediction of protein function. Using the Gene Ontology as a structured representation of biological roles, I evaluate how different data sources, model architectures, and deep-learning pre-training strategies influence sequence-to-function inference. I also develop approaches that move beyond binary functional labels, including similarity-based predictions and residue-level estimates of functional relevance that highlight motifs, domains, and critical sites. These methods enable richer and more fine-grained characterization of proteins than conventional yes/no annotation schemes. Building on these molecular-level insights, I examine how genetically encoded bacterial traits give rise to community-level behavior. I develop agent-based models that couple reaction–diffusion... https://escholarship.org/uc/item/0bf6d01b Sat, 7 Feb 2026 00:00:00 +0000 Dickson, Andrew Macdonald https://escholarship.org/uc/item/08r7t6kt Root microbiomes profoundly influence plant nutrient acquisition, stress tolerance, and overall plant health, Unfortunately, the ecological and genetic processes that structure these communities are poorly resolved. Traditional sequencing methods identify taxa but cannot distinguish closely related strains or quantify the dynamics of microbial colonization, competition, and coexistence. In this dissertation, I develop and apply DNA barcoding strategies to overcome these limitations in support of a contemporary framework for studying plant–microbe interactions at ecological and genetic resolution. First, I synthesize the ecological literature to classify the interaction types that shape microbial community assembly, with an emphasis on priority effects. By reanalyzing published datasets, I show that many interactions occur at the strain level, reinforcing the need for tools beyond 16S-based profiling to understand these interactions. Next, I develop BarTn7, a chromosomal barcoding... https://escholarship.org/uc/item/08r7t6kt Sat, 7 Feb 2026 00:00:00 +0000 Herbert, Robin Avery https://escholarship.org/uc/item/02f3z33c The integration of autonomous vehicles into transportation infrastructure presents a fundamental challenge: these systems must operate within environments designed by and for human drivers while sharing road spaces with human-operated vehicles. This dissertation addresses the central question of how autonomous vehicles can function as cooperative agents in mixed-autonomy traffic systems, where behavioral coordination between automated and human-driven vehicles determines overall system performance.The research develops a four-dimensional framework for analyzing this interaction paradigm: coexistence in shared road spaces, coordination through hierarchical control architectures, comprehension of human behavioral patterns for risk assessment, and certification protocols enabling third-party evaluation without proprietary access. Each dimension represents a distinct facet of the operational challenges facing autonomous vehicle deployment.This work progresses from theoretical foundations... https://escholarship.org/uc/item/02f3z33c Sat, 7 Feb 2026 00:00:00 +0000 Wang, Han https://escholarship.org/uc/item/00k285b1 Time-dependent motion (dynamics) of nuclei in molecules often occurs on ultrafast (picosecond (10-12 second) or femtosecond (10-15 second)) timescales. Experimentally, it is a question how to launch and detect these ultrafast nuclear dynamics in molecules in real time? In this dissertation, I studied two representative systems with the technique Extreme Ultraviolet (XUV)-Attosecond Transient Absorption Spectroscopy (ATAS). One is the coherent vibrational dynamics in CBr4 in the electronic ground state of the fully symmetric stretch mode, with a period of 125 fs. The results on CBr4 show that XUV-ATAS can resolve tiny bond-length displacements down to the order of 10-4 Å with 26 fs temporal resolution, with the knowledge of calculated core-excited potential energy surfaces. Such spatial and temporal resolution of XUV-ATAS is exceptional. It provides roughly 1,000 times better spatial resolution and 5 times better temporal resolution compared to the current MeV-UED at SLAC. ... https://escholarship.org/uc/item/00k285b1 Sat, 7 Feb 2026 00:00:00 +0000 Ou, Jen-Hao https://escholarship.org/uc/item/75h2g05p This dissertation examines representations of social revolution and its agents in literary and cultural texts written, published, and, in the case of the dramatic text, staged during the three-year conflict known as the Spanish Civil War (1936-1939). More specifically, this dissertation discusses what these texts postulate about the relationship of theory and practice in revolutionary politics. The body is central in this dissertation since, in many regards, it is a place where theory meets praxis. In order to examine this relationship, we focus on the revolutionary process — the Spanish Revolution of 1936 — that began alongside the civil war, both of which ensued from the military coup d´état against the democratically-elected government of the Frente Popular (Popular Front) during the Second Republic in July of 1936. We analyze how the texts respond to the Spanish Revolution of 1936 and, in some instances, the civil war, by examining their depictions of bodies, their compositions,... https://escholarship.org/uc/item/75h2g05p Tue, 7 Oct 2025 00:00:00 +0000 Blanco, Yesenia