UC Santa Barbara

In quantum materials, phenomena occurring at the subatomic level can manifest as properties on a macroscopic scale. The exploration of quantum effects in materials with nontrivial band topology and strongly correlated electrons holds great promise for technological advancement in fields such as quantum computing. This thesis examines properties of three such quantum materials using computational methods.The primary material system investigated is strontium titanate, an incipient ferroelectric that gives rise to an unconventional superconducting state at exceptionally low doping levels. The polar phase can be stabilized through strain or chemical substitution. Remarkably, superconductivity is enhanced within the polar phase, suggesting that the polar instability plays a pivotal role in the superconducting pairing mechanism. We develop a simplified free energy model combined with statistical mechanics methods to assess the character of the polar transition, which we find to be neither order-disorder nor displacive. We explore the effects of doping on the structural phase transitions and find that, in agreement with experiment, the polar distortion and formation of polar nanodomains are suppressed in the presence of free carriers, while antiferrodistortive order remains essentially unchanged. The single-domain nature and insensitivity to doping suggest that the antiferrodistortive order does not play an important role in Cooper pairing. By calculating electronic properties in the polar phase, we analyze parameters that are relevant to superconductivity such as the density of states at the Fermi level, the Rashba splitting of the energy bands, and the Migdal ratio. We explore the chemical phase space of the naturally occurring minerals herbertsmithite [ZnCu3(OH)6Cl2] and Zn-substituted barlowite [ZnCu3(OH)6BrF], which both feature perfect kagome layers of spin-1/2 copper ions and display experimental signatures consistent with a quantum spin liquid state at low temperatures. To identify other possible candidates within this material family, we perform a systematic first-principles combinatorial exploration of structurally related compounds [ACu3(OH)6B2 and ACu3(OH)6BC] by substituting nonmagnetic divalent cations (A) and halide anions (B, C). We select several promising candidate materials that we believe deserve further attention. Finally, we examine CsV3Sb5, a member of the AV3Sb5 (A = K,Rb,Cs) family of kagome metals, whose low-energy physics is dominated by an unusual charge density wave phase. We elucidate the nature of the charge density wave order parameter using first-principles density functional theory calculations which support the findings of experimental coherent phonon spectroscopy measurements. Through our study of the structural phase space of CsV3Sb5, we find that the charge density wave can be described as tri-hexagonal ordering with interlayer modulation along the c-axis.

First-principles techniques are often limited by their inability to incorporate the effects of temperature and disorder. Here, we augment first-principles density functional calculations using statistical mechanics methods such as the Metropolis Monte Carlo algorithm and Langevin dynamics to incorporate temperature effects on large, disordered supercells to simulate the thermal phase space of strontium titanate. The chemical phase of the herbertsmithite material family is systematically explored through high-throughput first-principles pseudo-convex hull calculations and an assessment of defect formation energy. We use frozen phonon calculations to investigate the structural phase space of CsV3Sb5 and find that the charge density wave order expands beyond the previously studied 2x2x1 construction. Our techniques can be applied more broadly to other material systems to expand the capabilities of computational methods to accurately capture thermal effects and structural disorder in quantum materials.

Climate change and its associated environmental impacts pose immense challenges that require innovative approaches to address. This dissertation presents three distinct studies that showcase the application of advanced modeling and machine learning methods to investigate critical issues at the intersection of changing human and natural systems.In Chapter 1, I employ a novel modeling framework to analyze the complex relationship between vegetation dynamics and hydroclimate variability across East Africa. Empirical dynamic modeling is a data-driven approach for studying state-dependent dynamics and interactions within complex systems, enabling the identification of key driving variables and the prediction of future system behaviors. Adopting this method, the study provides insights into how the stability and vulnerability of ecosystems vary with environmental conditions, land cover type, and seasonality. In Chapter 2, I explore how various factors contribute to human displacement, focusing on the environmental drivers and mechanisms of migration in Somalia. Gravity models are a class of spatial interaction models that estimate the flow or movement between locations based on the attractiveness of the destinations and the impedance between the origin and destination. I use these models to examine the connections between climate, socio-economic, and political factors influencing population movements. Notably, I find that livelihood is an important differentiating factor in determining whether the climate strongly impacts individuals' migration patterns. In Chapter 3, I implement advanced natural language processing techniques to develop an automated system for classifying global multi-hazard disaster events from humanitarian news articles and reports. Large language models are a form of artificial intelligence and deep learning that can process, understand, and generate human language by learning from vast amounts of textual data, enabling them to perform a wide range of natural language processing tasks. By employing these models, the study demonstrates the potential of emerging technologies in improving the efficiency of disaster information retrieval and response. With a geographical framework that unifies perspectives from environmental, social, and computer science, these chapters collectively contribute to developing data-driven solutions for understanding environmental stressors.

Hispanic Serving Institutions (HSIs) play a leading role in supporting educational equity, enrolling two-thirds of all U.S. Latino students despite representing less than a fifth of all American higher education institutions. These colleges and universities manage to provide substantial support and prioritize student-centered organizational outcomes with limited resources, receiving nearly a third less federal funding per student compared to the national average. This makes the study of their leadership practices not only relevant but essential for broadening our understanding of effective educational leadership under constraints. This qualitative multiple case study explores the organizational behavior of university presidents at Hispanic Serving Institutions (HSIs) by focusing on their navigation of the job, which is shaped by shared governance, diverse student needs, and external pressures. By examining decision-making processes and leadership approaches, this study aims to enhance the effectiveness of colleges and universities in supporting their diverse student bodies. This work not only contributes to academic knowledge but also provides a wide range of perspectives on managing and leading schools in an increasingly diverse educational landscape and offers practical insights for improving leadership across all types of higher education institutions. Using a thematic analytical approach, this research examined two rounds of semi-structured interviews with 12 California HSI presidents from various backgrounds and institutional contexts. The Critical Incident Technique (CIT) was employed to investigate 38 critical incidents by the leaders’ utilization of five dimensions of organizational behavior: collegial, bureaucratic, political, systemic, and symbolic. The results indicate that leadership at HSIs typically consists of a multidimensional approach, where presidents often leverage layered strategies to resolve institutional issues. Additionally, the adaptability of leadership approaches in response to various critical incidents and factors—such as the type of institution, source of the incident, and the president’s level of experience—points to the importance of context-dependent practices and decision-making.

Fish move through the ocean unaware of the socially-constructed spatial boundaries imposed on them by legal and administrative systems. The vast majority of marine species move beyond a single national jurisdiction and these species play a pivotal role in global food security. However, movement across boundaries increases competitive incentives among nations, leading to overfishing. One way to slow this ‘race to fish’ is by forming agreements, where countries come together to manage how marine resources are shared. Because agreements are self-enforcing, agreement success depends on the strategic manipulation of incentives. Here I measure how effective an existing agreement has been at preventing overfishing, how a different agreement that has failed to curb illegal fishing can use competition from aquaculture to reduce poaching, and how fish movement can incentivize a marine reserve agreement. First, I use an econometric approach to measure how effective a tuna agreement management measure has been at reducing fishing mortality for highly mobile, boundary-spanning tuna and billfishes. Next, I use a bioeconomic model to find new solutions for an international wildlife trade agreement, specifically examining how competitive responses to aquaculture can disrupt a lucrative illegal trade in Mexico. Finally, I create a dynamic and spatial game to examine how fish movement can incentivize the development of a bilateral transboundary marine reserve agreement, which can help countries meet their commitment to protect 30% of the ocean by 2030. These results demonstrate how successful agreements can resolve externalities generated from the movement of fish and fish products across boundaries, and how effective agreements can be measured and designed.

Atmospheric acoustic signals produced by crashing ocean waves may be highly variable from sub-hourly to seasonal time scales. Specifically, low-frequency acoustics (below 20 Hz) known as infrasound have been reported near coastlines under variable ocean conditions such as significant wave height, wave period, swell direction, locations of crashing surf, and style of breakers (e.g., Garcés et al. [2003], Aucan et al. [2006], Le Pichon et al. [2004]). Here, we use infrasound and seismic data collected at Coal Oil Point Reserve in Santa Barbara, California, to investigate how the ambient acoustic wave-field is perturbed by variable surf and ocean conditions. Data collection involved one principal infrasound sensor (September 2022–July 2023), complemented by four temporary infrasound arrays of varying geometries: 1) January 11, 2023; 2) January 12–19, 2023; 3) July 10, 2023; 4) October 17–23, 2023. Each array included video recordings, and the latter two arrays involved broadband seismometers as well. We find that distinct groups of surf infrasound signals are produced throughout 10 months of data, and that the amplitudes of these signals are correlated with offshore significant wave height and local wind speed. We estimate source locations of surf infrasound using array-based and network-based processing methods and find a correspondence with expected locations of breaking waves as seen in video footage. We highlight the opportunity to monitor the littoral sea surface state using infrasound and seismic records.

How does the U.S. military make use of the foreign cultures into which it has inserted itself? Based on twenty-two months of ethnographic fieldwork on and around U.S. military facilities in Japan (host to over 100,000 American military personnel and family members), this dissertation argues that military policy generates narratives for American troops’ self-identification and aspiration vis-à-vis the local culture that are ultimately detrimental to the U.S.-Japan Alliance. Specifically, it identifies four such narratives and their consequences: First, American troops are told they are ambassadors in a bid to get them to spend more time outside of the bases. I conclude that this is a means of shifting responsibility for troops’ mental well-being and morale onto the Japanese communities surrounding the bases and that it makes some Japanese feel they are being forced into complicity with American militarism. Second, military narratives divide troops into “good neighbors” and “bad apples” in a move that both isolates the “bad apples” from cultural and historical patterns of behavior (and thus absolves the military of responsibility for those patterns) and rewards personnel for good intentions and the appearance of good deeds regardless of the often-problematic consequences of their altruistic efforts. Third, American troops adopt the mantle of samurai as a means of replacing the aspirational fantasy at domestic bases of being “super-citizens” (Lutz 2001, 236), naturalizing U.S. military deployment in Japan in a way that encourages the widespread dismissal of all forms of Japanese masculinity. Finally, Okinawans—residents of the prefecture most impacted by the military—are painted as adversaries to the U.S. military’s goals and operations, sorting them into binaries of good/pro-base/Japanese and bad/anti-base/Okinawans that deny the complexities of their relationships with troops, bases, the United States, and Japan. This is the first ethnographic study of how forward-deployed military bases navigate and utilize local culture and contributes to scholarship on the constitutive interplay between interpersonal and international relations, highlighting how imperial and Orientalist legacies inform the everyday functions and expressions of alliance.

This dissertation presents four distinct applications in Population and Health Geography. While varied, in each of these applications I center geographic constructs (space, time, exposure and scale), and implement a specific study design (often with novel methods and/or measures) to elucidate these themes in context. In particular, I trace the challenges of measuring exposure, effects, and uncertainty through each application. In the first application I use historical wildfire evacuation zones to identify wildfire exposed populations and investigate the relationship between wildfire exposure and cardiovascular events. For the second application, I describe and test a new time use data collection method in the context of existing tools and simulation models. In the third application I introduce a statistical model into wildfire vulnerability index construction and discuss implications for spatial variability. For the final application I construct a novel indirect measure of school climate and combine multiple data sources to explore mental health variation across school districts in New York City. Together these applications demonstrate the natural partnership of Geography and Statistics in social science.

With the end of Moore’s Law and the rise of compute- and data-intensive deep learning (DL) applications, the focus on arduous new processor design has shifted towards a more effective and agile approach: Intelligent Software to maximize the performance gains of DL hardware like GPUs. There are several highlights of such intelligent software design. First, it would maximize the execution efficiency of existing and emerging DL algorithms on powerful platforms like GPUs. Second, it would promote the adaptiveness of systems to handle a diverse range of inputs. Third, it would maintain sufficient portability and scalability across a diverse range of platforms, such as mobile devices and high-performance clusters.

In this thesis, I will first highlight the importance of software innovation to bridge the gap between the increasingly diverse DL applications and the existing powerful DL hardware platforms. The second part of my thesis will recap my research work on DL system software innovation, focusing on 1) Precision Mismatch between DL applications and high-performance GPU units like Tensor Cores (e.g., QGTC [PPoPP ’22] and APNN-TC [SC ’21]), to improve the efficiency of quantized deep learning on powerful GPU platforms, and 2) Computing Pattern Mismatch between the sparse and irregular DL applications, such as Graph Neural Networks, and the dense and regular tailored GPU computing paradigm (e.g., GNNAdvisor [OSDI ’21] and MGG [OSDI ’23]), to highlight system adaptability and scalability. Finally, I will conclude this thesis with my vision and future work for building efficient, scalable, and secure DL systems.

Ambitious presumptivism says that all our testimony based beliefs are on-balance immediately and defeasibly warranted. The rational deception objection says that ambitious presumptivism is not true because it is sometimes rational for a speaker to assert lies rather than truths. One logically possible reply is to argue that it is never rational for a speaker to assert lies rather than truths. In this essay, I develop such a non-conciliatory response to the rational deception objection.

In chapter 1, I explain ambitious presumptivism and the rational deception objection. I identify Kant's prohibition against lying as a historical predecessor to the non-conciliatory response to the rational deception objection. I then identify Burge as the heir apparent to a neo-Kantian non-conciliatory response to the rational deception objection. In chapter 2, I explain my interpretation of how Burge is heir apparent to a neo-Kantian non-conciliatory response. I call Burge's response the ``functional unity'' argument.

In chapter 3, I defend my attribution to Burge of the functional unity argument. In chapters 4 and 5, I defend the functional unity argument itself from the most influential objections raised against it. in chapter 6, I defend the functional unity argument from the classic murderer-at-the-door objection that dogged Kant's prohibition against lying.

Studying microbiomes presents an ever evolving and complex set of analytical challenges (Berges and Franklin, 2001). Microbiomes, regardless of ecosystem origin, typically comprise hundreds of bacterial species (Flemming et al., 2016). This complexity makes it difficult to simultaneously analyze one species' physiology in detail while understanding its broader context and co-occurrence with other microbiome members (Galloway-Peña and Hanson, 2020). Consequently, researchers face two main methodological strategies: isolate one bacterial member in a controlled lab setting or analyze the network of microbes at a coarse level (Galloway-Peña and Hanson, 2020). The first strategy, species isolation and culturing, offers a high-resolution understanding of a specific species' functions. However, this approach loses context for how that bacterium interacts within the broader microbiome (Sarhan et al., 2019). Additionally, less than 1% of all bacterial species on Earth have been cultured (Lewis et al., 2021) because many bacteria have complex interactions with other microbiome members requiring metabolite exchange, which researchers are currently unaware of (Flemming et al., 2016). The second strategy, quantifying compositional changes of microbiome members, allows researchers to understand the context of co-occurrences and taxa proportionality in each member of the microbiome (Berg et al., 2020; Galloway-Peña and Hanson, 2020; Weinroth et al., 2022). However, this approach lacks detailed information about any individual species' function (Weinroth et al., 2022). This analytical weakness is problematic because it obscures each bacterium's specific contributions to the microbiome's functionality and formation processes. Fortunately, next-generation sequencing is now cost-effective enough to sequence deeply and construct metagenome-assembled genomes (MAGs), bridging the gap between these two experimental strategies (Van Nimwegen et al., 2016). MAGs enable the assessment of compositional changes in microbiomes using read information while also predicting the metabolic capabilities of each MAG representing a bacterium (Forbes et al., 2017; Laudadio et al., 2019). Although not empirically validated with wet-lab tests, the predictive power of MAGs has been immensely beneficial in guiding wet-lab research efforts efficiently (Laudadio et al., 2019). For example, studying MAGs has led to the discovery of natural product resistance mechanisms in pathogens, allowing scientists to develop new clinical strategies for treating pathogenic infections (Ma et al., 2023). For my Ph.D., I set out to use metagenomics to analyze microbiomes exposed to multiple environmental conditions simultaneously. Specifically, I aimed to understand how microbiomes develop under differing environmental conditions. To achieve this, I selected study systems where microbiomes of the same host genotype were exposed to different environmental conditions, reducing the variation of host genotype on microbiome composition (Wagner et al., 2016; Sanders-Smith et al., 2020; Chaudhry et al., 2021). I settled on the giant kelp forest, which allowed me to assess surface and subsurface microbiomes in juvenile and mature samples (Chapter 1) and characterize microbiome member carbohydrate metabolic capabilities (Chapter 2). Simultaneously, I collaborated with other members of the Wilbanks lab on the pink berry system, where geographically isolated bacterial aggregates are exposed to marshes with different environmental conditions. This collaboration provided additional insights into the influence of environmental factors on microbial biosynthetic gene cluster diversity and strain-level variation (Chapter 3).