UCLA

This qualitative case study investigates the effects of governance configuration and organizational design on financial decision-making at a leading public research university on the west coast of the United States. Rather than conceptualizing financial management as a neutral administrative function, the study frames financial authority as constructed through sequence, thresholds, and exposure. Interviewing academic and financial leaders and analyzing institutional documents, the research investigates how reporting hierarchies, revenue structures, and escalation pathways shape the timing, allocation, and compression of decision authority under fiscal constraints. Results reveal that financial expertise enters strategy at structurally determined points, that autonomy expands and contracts within institutional risk thresholds, and that decentralization creates uneven fiscal capacities across units.Governance configuration emerges as the moderating structure of institutional resilience: when the sequence of authority shifts under fiscal pressure, it does not happen randomly, but follows a pattern. By identifying structural sequencing as the primary mechanism of financial authority, this case study contributes to the literature on governance and argues that resilience in public higher education is determined less by sources and volumes of revenue and more by the design of authority through which decisions are made, escalated, and absorbed.

Facts about us—what we are motivated by and what we are able to do—can change our responsibilities and obligations. But how and why? This dissertation takes up these questions.In Chapter I, I consider the suggestion that, because we cannot simply choose what to see as reason-giving, the reasons for which an agent acts cannot be relevant to the permissibility of the agent’s conduct. I argue this is false. I reach this conclusion by first considering the agency we have over acting-for-certain-reasons. I argue that, even if we cannot simply choose what to see as reasons, we can often simply choose whether to act-for-certain-reasons. That is to say, we can often choose to act-for-certain-reasons just as readily as we can choose to act in a way that can be identified independently of the reasons for which it is done. It follows from this acknowledgement that facts about the reasons for which an agent acts can be relevant to the permissibility of her conduct in just the same way other facts about that conduct can be relevant.In Chapter II, I take up the claim that a contractualist in particular should not think that the permissibility of an agent’s conduct can turn on the reasons for which she acts. This is mistaken. The contractualist has a straightforward explanation of why the reasons for which we act can change whether we act permissibly: because we have interests not only in what happens to us, but also, at least at times, in why they happen, we can be constrained not only in how we act but also in the reasons for which we act.In Chapter III, I turn to the relationship between what we can do and what we are obligated to do. According to a familiar proposal, considerations of fairness settle this issue clearly: it would be unfair to require agents to act in ways they cannot. I first show that this proposal has a costly implication: that whether agents facing vicious or demeaning conduct can demand to be treated better depends on the abilities of those who mistreat them. I then offer an alternative way of thinking about how considerations of fairness bear on our obligations—one that avoids this implication while still explaining what seems right about the familiar proposal.

In the wake of the Supreme Court’s 2023 decision in Students for Fair Admissions (SFF A) v. Harvard, the college admissions essay has been transformed from a standard application component into a primary legal and sociopolitical battlefield for racial identity. This qualitative study investigates how eight Black and Latinx high school seniors navigate this heightened systemic pressure while completing the University of California’s Personal Insight Questions (PIQs). Grounded in Yosso’s (2005) Community Cultural Wealth (CCW) framework, this research challenges dominant deficit-centered educational narratives by examining how marginalized youth treat their lived experiences not as structural burdens, but as legitimate sites of institutional expertise and asset cultivation.Data collection involved utilizing textual analysis of 16 personal admissions essays alongside semi-structured interviews. The cross-case synthesis revealed that participants actively reject the institutional expectation to perform or trauma dump for admissions gatekeepers. Instead, students practice sophisticated narrative gatekeeping and strategic silence to preserve their personal dignity and maintain ownership over their family histories.The analysis identified four distinct archetypal modes of narrative self-authorship: The Reformers, The Sustainers, The Alchemists, and The Visionaries. Together, these archetypes illustrate a vibrant spectrum of how Black and Latinx youth leverage linguistic, navigational, familial, and resistant capital to claim space within higher education. Ultimately, this study argues that the college essay functions as a critical site of Resistant Capital. Rather than hyper-focusing on the mere mechanics of application access, this research offers concrete structural frameworks including a Class Asset Dashboard and a Bridge Dialogue series to shift institutional accountability. The findings demand that post-secondary institutions transition from evaluating student deficits to actively preparing classrooms to receive the sophisticated, pre-existing brilliance of incoming minority students.

At lower RHIC energies, hadronization in the quark–gluon-plasma (QGP) fireball is significantly influenced by transported quarks carried in from the colliding nuclei. Within the quark coalescence picture, these transported quarks, which experience the full evolution of the system’s pressure gradients, imprint distinct flavor and dynamical signatures on final-state hadrons. This thesis presents three complementary measurements from the STAR Beam Energy Scan Phase II (BES-II) program in Au+Au collisions.The influence of transported quarks is directly reflected in the elliptic flow (v2) of charged pions. In Au+Au collisions at √ sNN = 7.7–27 GeV, π − carries systematically larger v2 than π + across centralities and energies, consistent with the higher availability of transported d quarks relative to u quarks in the neutron-rich Au nucleus. Using a coalescence sum rule with the antiproton as a proxy for the produced-quark flow, the transported quark ratio Ntr d /Ntr u is extracted and found to be consistent with the Au stoichiometric value 315/276 ≈ 1.14, providing a direct link between the quark content of the colliding nuclei, via nuclear geometry, and the final collective expansion of the QGP drop.Electromagnetic fields generated by the spectator protons in peripheral collisions offer a complementary window into the quark-level response at hadronization. The directed flow (v1) splitting between Λ 0 and Λ¯ 0 is measured over the same BES-II energy range; the slope difference ∆(dv1/dy) becomes negative in peripheral collisions, mirroring the behavior previously observed for charged hadrons and attributed to the combined Faraday and Coulomb effects. A first-order coalescence test using p–p¯, supplemented by a second-order correction based on K+–K−, supports the coalescence picture and indicates that the electromagnetic response acts at the level of the constituent quarks and is not set by electric charge alone.These flow-based analyses probe the transported-quark coalescence picture through anisotropic momentum signatures. A more direct handle on the transported-baryon population is provided by the Ω − hyperon: at √ sNN = 7.7–27 GeV, the Ω − yield carries a substantial excess over Ω¯ + from baryon transport. We establish the Combinatorial Background Subtraction (CBS) correlation method, which uses associated kaons to decompose the Ω − population into its transported and pair-produced components and to probe whether their hadronization signatures differ. At √ sNN = 19.6 GeV, the transported Ω − shows a modest excess of associated kaons over a pair-produced Ω, 1.10±0.74 per Ω, in the direction expected if its strangeness is balanced by kaons rather than by co-produced anti-baryons.

Extensive research has documented how culturally responsive or sustaining pedagogies benefit minoritized youth. Yet, critical scholars assert that prevalent professional development models for in-service teachers overemphasize generic best practices or data-driven school improvement and subsequently proliferate problematic deficit paradigms, rather than asset-oriented pedagogies. While understudied, existing research on equity-driven and asset-oriented professional learning suggests critical reflection, collaborative inquiry, and relational trust are key components. Building on such literature, this qualitative case study explored how a public secondary school’s Instructional Leadership Team (ILT) piloted teacher-led and asset-oriented Professional Learning Communities (PLCs). As the study’s lead researcher and the school site’s instructional coach, I utilized Reciprocal Learning Partnerships (RLP) as an equity-focused framework for designing the study’s activities, developing its research questions, and analyzing the resultant data. Through observations, interviews, and document analysis, this study detailed the process for piloting a bottom-up model of context-specific and asset-oriented professional learning. It further uncovered that critical reflection, collaborative inquiry, and relational trust developed as a result of facilitators cultivating a sense of authentic, purposeful work and modeling dispositions of vulnerability and care. To more deeply understand the intricacies of relational trust, this case study also explored the extent to which its multiracial participants expressed vulnerable reflections on lived experiences and surfaced their personal identities and positionalities in professional learning spaces. The resultant findings suggest that the emergence of identities and positionalities often remained internal and that relational trust may not have been cultivated evenly amongst educators encompassing a broad array of racial, ethnic, gender and other identities. Taken together, the insights of this case study stress the importance of adult-facing equity work in diverse public schools—teachers must first experience their own learning spaces as asset-oriented and culturally sustaining to reproduce the same humanizing conditions in classrooms with their students.

Multimodality has elevated the utility of modern machine learning models. Through data-driven deep learning, multimodal neural networks learn generalizable and robust features, providing increased performance and capabilities while promising greater resilience to changing environments. Despite such capabilities, the increased complexity of these networks renders them susceptible to other variations that can manifest at inference time - ranging from dynamics during a system's deployment to its active runtime. These weaknesses of multimodal networks have not been adequately addressed in existing work, resulting in severe inefficiencies or subpar performance when subject to inference-time variations in real-world environments. This dissertation examines the weaknesses of multimodal networks across a variety of inference-time variations and diverse datasets, demonstrating that the efficient adaptation of internal weights constitutes a powerful and highly generalizable solution. First, we examine the challenge of deployment-time sensor perspective shift in distributed multimodal localization systems. Existing multimodal deep neural networks do not tolerate novel sensor perspectives arising during deployment, suffering from large accuracy degradations. We introduce FlexLoc, which combats sensor perspective shift by conditioning a portion of network weights on the sensors' pose information. FlexLoc improves zero-shot localization under unseen configurations of sensor pose by almost 50% in comparison to baselines. Second, we propose ADMN, which further explores inference-time adaptation by addressing variations that manifest during runtime in real-world multimodal systems, which may necessitate per-sample adaptation. Across a wide range of downstream tasks (e.g., audio-visual classification, gesture detection), we showcase how intelligent allocation of resources across modalities is critical for addressing runtime variations, ensuring optimal usage of limited computational resources. Third, we introduce SWAN, which extends the innovations of ADMN into the complex realm of multi-object detection in autonomous driving. We jointly explore adaptation to three forms of runtime variations -- changes in modality quality, platform resources, and complexity of the input sample itself. Moreover, we demonstrate how such a system can be practically realized on a real-world edge device. Finally, moving away from dedicated, task-specific models, we introduce CRAFT, an efficient method of adapting a general purpose multimodal foundation model at deployment time to an arbitrary downstream task. During this, we highlight the unique challenges associated with this class of neural networks. Ultimately, by leveraging the shared paradigm of context-aware model adaptation, these works collectively demonstrate that adjusting neural network parameters according to the state of the world is a highly effective strategy for mitigating inference-time variations.

Dust storms and chronic dust loadings are a growing but heterogeneously characterized publichealth concern in California, where a warming climate, more frequent heat and drought, and expanding disturbed land surfaces are increasing dust activity. This dissertation builds, in sequence, the evidence a dust warning system requires, organized around the four components of a people-centered early warning system: risk knowledge, monitoring and forecasting, communication, and response. Aim 1 systematically reviews global dust storm early warning systems. Of 22 included studies, 19 addressed monitoring and forecasting while only three addressed risk knowledge, communication, or response, a gap that has persisted for more than two decades and that frames the dissertation. Aim 2 develops and externally validates a daily, statewide surface of the dust-attributable fraction of PM (DAF-PM ) at population-weighted ZIP-code centroids for 2006 through ₁₀ ₁₀ 2019, combining a machine-learning ensemble with a transparent rule-based comparator. Evaluated against independent composition-based dust, the ensemble preserved strong rank- order agreement (pooled Spearman rho 0.65) and discriminated high-dust days well (AUC up to 0.90), localizing chronic burden to the southern San Joaquin and Imperial Valleys.Aim 3 links this exposure surface to roughly 24 million case-crossover strata of cardiovascular and respiratory encounters, 2006 through 2018. A one-interquartile-range increase in DAF-PM₁₀ raised the odds of a respiratory encounter by about 1% on the same day, peaking one day later and accumulating to roughly 1.4% over a week, and raised cardiovascular odds by about 0.2%. The respiratory response was nonlinear, concentrated in asthma and chronic obstructive pulmonary disease, amplified specifically by peak daytime heat, and positive across most populated air basins, with the notable exception of the chronically dust-saturated Salton Sea basin, where the case-crossover design returned a null estimate; cardiovascular effects concentrated in hypertensive disease and coastal basins. Aim 4 translates this evidence into a risk communication and community engagement plan, an evidence-based infographic, and a short educational video, and shows how the exposure metric and the dust-by-heat findings can inform alert thresholds and timing. Together, the four aims form an exposure-to-action pathway that makes a climate-sensitive hazard more measurable, more clearly linked to health, and more actionable.

Phylodynamic inference exploits pathogen genomic data to estimate epidemiological quantities of direct public-health relevance, including time-varying transmission and recovery rates, effective population size trajectories, and migration rates between regions. Bayesian implementations on coalescent and birth–death tree priors are standard in outbreak surveillance, but the underlying inference machinery scales poorly with the size of contemporary genomic datasets and with the parameter dimensionality of modern episodic and structured models. This dissertation develops scalable algorithms, both gradient-based and parallel, that remove these bottlenecks for two model families central to applied phylodynamics. The first contribution targets the episodic birth–death–sampling model. The likelihood is reformulated to admit a closed-form, linear-time analytic gradient with respect to all epoch-specific birth, death, and sampling parameters, enabling Hamiltonian Monte Carlo sampling of the full epoch-rate vector. Combined with regularized shrinkage priors over the high-dimensional rate vectors, the resulting sampler increases the minimum effective sample size per unit time 10- to 200-fold over univariate Metropolis–Hastings, while recovering smooth effective-reproductive-number trajectories. The methodology is demonstrated on HIV-1 subtype A sequences from Odesa, Ukraine, a New York State seasonal influenza A/H3N2 hemagglutinin alignment, and the 2014–2016 West African Ebola virus epidemic. The second contribution addresses the structured coalescent approximation that underlies modern phylogeographic inference. Reformulating its peeling recursion to expose fine-grained parallelism across both lineages and demes accelerates the likelihood by 10–26-fold on multi-core central processing units and up to 60-fold on graphics processing units, placing previously intractable analyses within practical wall-clock budgets. Reverse-mode differentiation of the same recursion then returns the exact gradient for all migration and population-size parameters at the cost of a few likelihood evaluations rather than one per parameter, making Hamiltonian Monte Carlo practical for high-dimensional phylogeographic posteriors. The methods are validated on a dengue virus phylogeographic analysis and applied to a continental-scale reconstruction of highly pathogenic avian influenza A(H5N1). A concluding chapter sketches extensions: a log-linear parameterization of migration rates for hypothesis-driven analysis of pathogen spread, and both parametric and non-parametric models for time-varying effective population sizes. All methods are released as open-source extensions to the BEAST X phylogenetic inference platform and the BEAGLE library.

Fluid flows are high-dimensional, nonlinear dynamical systems governed by partial differential equations, making their direct simulation computationally expensive for many practical applications. Nevertheless, many flows evolve on low-dimensional manifolds, motivating reduced-order models that capture the essential dynamics while significantly reducing computational complexity. Recent advances in machine learning provide powerful nonlinear dimensionality reduction techniques capable of learning such representations directly from data. This thesis investigates data-driven latent representations for reduced-order modeling in fluid mechanics, with an emphasis on aerodynamic applications. Rather than treating latent spaces solely as compact embeddings optimized for reconstruction, this work explores representations that incorporate physical and mathematical structure to improve interpretability, robustness, and predictive capability. First, an autoencoder-based framework is developed for aerodynamic design optimization of industrial automobile geometries. The learned latent representation efficiently captures geometric variations, enabling accelerated design exploration while achieving an 11% reduction in drag coefficient. The framework predicts aerodynamic performance within 2% of experimentally validated large-eddy simulations, demonstrating sufficient fidelity for practical design optimization.Second, a physically structured latent space based on optimal transport distances is introduced for flow control analysis. Applied to separated flow over a NACA 0012 airfoil with leading-edge thermal actuation, the approach identifies distinct control regimes associated with changes in separation bubble size and the onset of partial and global laminarization, providing an interpretable representation of the underlying flow dynamics.Finally, a probabilistic state estimation framework combining latent diffusion models with particle filtering is developed for reconstructing and predicting transient aerodynamic states from limited observations. Applied to a high-incidence airfoil subjected to gust disturbances, the framework enables uncertainty-aware estimation of complex unsteady flows.Collectively, these studies demonstrate that physically informed latent representations provide efficient and interpretable reduced-order models for aerodynamic design optimization, flow control assessment, and probabilistic state estimation. The results highlight the importance of incorporating physical structure into representation learning, illustrating the potential of hybrid machine learning and physics-based approaches for reliable modeling and analysis of complex fluid systems.

Intrapersonal communication examines how individuals interact with themselves from a communication perspective. We engage in self-talk, imagine interactions, generate internalized media such as visualizations, and so on, yet a unified study of these phenomena is missing. This dissertation aims to revive the study of intrapersonal communication by describing what it is through a literature review of related subjects and presenting frameworks for how it can be studied. Empirical work is then presented demonstrating a multidimensional approach to studying one example of intrapersonal communication – overt linguistic streams-of-consciousness. In a series of three studies, participants provide their typed or spoken streams-of-consciousness, aiming to report their thoughts continuously and spontaneously. The first study prompts participants to think about life before or during the COVID-19 pandemic to see how their thoughts about the future may shift through psychological momentum. From analyzing the language in the future-directed thoughts alone, it is possible to predict the temporal priming of prior thoughts. The second study then explores how typing versus speaking thoughts may impact how thoughts unfold. Typing enhanced lexical diversity as well as proportions of emotional language usage. When Large Language Models generated comparable thoughts, emotion language was reduced relative to humans. Finally, the third study manipulates the environment under which spoken thoughts are generated to see how darkness or the presence of a mirror may shift language and the delivery of our voices. Darkness and a mirror led to exploration of different language spaces, and speaking in the dark attenuated vocal intensity while boosting roughness. Together these studies show that the conditions under which we deliver our thoughts to ourselves can shape what we think about and how.