36214 research outputs found
Sort by
Extreme Wave Localization in Nonlinear Mechanical Metamaterials
No full textThesis (Ph.D.)--University of Washington, 2025Mechanical metamaterials, engineered structures with tailored mechanical properties, have emerged as a powerful platform for manipulating waves, spurring extensive research in both linear and nonlinear wave dynamics. Wave localization—the concentration of energy within specific regions—is central to controlling waves. In the linear regime, topological insulators offer robust wave-guiding through protected boundary states; however, existing designs often exhibit limitations in tunability and energy management.On the other hand, nonlinear wave localization manifests as distinct phenomena such as solitons and rogue waves. Solitons, stable and localized wave packets, hold promise for efficient energy transport and have been extensively studied in optical systems. Conversely, rogue waves, characterized by extreme amplitudes and transient appearances, provide insights into abrupt wave phenomena that typically pose significant challenges for prediction and control in diverse physical systems ranging from oceans to optical fibers. While these nonlinear wave phenomena are well-studied theoretically and numerically, experimental investigations in mechanical metamaterials remain limited, hindering a comprehensive understanding of their behavior and subsequent application in mechanical systems.
In this work, we address these limitations by investigating wave localization in mechanical metamaterials across both linear and nonlinear regimes, focusing on manipulating waves and experimentally realizing extreme wave events. For linear waves, we explore topological wave localization using a one-dimensional (1D) dimer lattice with a coupled degree of freedom. Leveraging its inherent axial-rotation coupling, we demonstrate an in situ tunability of the dispersion relationship, enabling a controlled transfer of topologically protected edge states. This coupling-driven energy control is then extended to the nonlinear regime, where we propose a method for manipulating solitary and rogue waves.
Experimentally, we first conduct a series of experiments using single-component 1D lattices, quantifying the phase shift from head-on rarefaction soliton collisions to be compared with analytical and numerical estimates. Finally, we experimentally demonstrate rogue wave formation in a mechanical metamaterial using a low-dissipation setup. Gaussian initial profiles are assigned to the lattice to form rogue waves, and the influence of varying initial energy landscapes is examined. We believe that our findings shall provide deeper insight into wave localizations in mechanical metamaterials, paving the way for applications in vibration control, energy harvesting, and extreme event mitigation
Using genomic technology to transform how genetics is used to diagnose and treat disease
No full textThesis (Ph.D.)--University of Washington, 2025Interpreting the clinical significance of rare genetic sequence variants is challenging due limited evidence, and as a result, most newly identified missense variants are interpreted as variants of uncertain significance (VUS). Multiplexed assays of variant effet (MAVEs), where hundreds to thousands of variant effects are measured in a single experiment have significantly accelerated the rate at which functional data are generated. Since functional data can be applied when interpreting variants, MAVEs have the potential to revolutionize clinical genetics by providing functional data at scale to resolve VUS. We systematically evaluated the clinical utility of MAVEs by integrating published MAVE data with clinical interpretations and resolved 49% of VUS for BRCA1, 69% for TP53, and 15% for PTEN. Although we demonstrated the potential for MAVEs to resolve uncertainty in genetic testing, MAVE technologies were limited to genes with phenotypes in utilitarian cancer derived cell lines. We addressed this limitation by developing iPSC-SGE, where variants are edited into iPSCs, enabling phenotyping in differentiated cells. We introduced 498 SNVs into POLG and 496 variants into MYBPC3. POLG variant effects were measured with a growth assay in iPSCs in the context of different background alleles and MYBPC3 variant effects were measured by variant abundance in cardiomyocytes. iPSC-SGE data was validated with known pathogenic and benign variants and is poised to generate functional data for genes previously inaccessible with MAVEs. Finally, we explored the use of variant effect predictors for variant interpretation, a major factor contributing to the VUS problem. We found that current calibration methods lead to inappropriate evidence for up to 75% of variants and offer a new solution for calibration via clustering VEP data for protein domains on similarity of score distributions. This method enables more accurate evidence strength thresholding while maintaining robust sets of calibration varants. Taken together, cell context specific functional data and variant specific VEP calibration will result in significant reduction to VUS while providing rich phenotypic insight for advancing precision medicine
Essays on Risk-Return Relation and Asset Pricing
No full textThesis (Ph.D.)--University of Washington, 2025In this dissertation, I empirically examine stock market valuations, with a particular focus on the risk-return relation and the respective roles of cash flow and discount rate news.In Chapter 1, I empirically investigate the risk-return relation under the investors' subjective volatility expectations, which deviate from the rational expectations. I first derive the objective risk premium under the slow-moving subjective volatility expectations based on the theoretical model of Lochstoer and Muir (2022) and show that the slow-moving volatility expectation generates a lead-lag specification in the objective risk-return relation. Then, I develop and estimate an empirical model by employing the log-linear present value framework. The empirical results using U.S. monthly excess stock returns suggest that the slow-moving feature of volatility expectations and the lead-lag structure in the objective risk-return relation are both significantly identified from the data. The parameter estimates suggest that while the objective risk-return relation can be negative, the subjective risk-return relation remains strongly positive, aligning with the key prediction and assumption in Lochstoer and Muir (2022). Moreover, I find that incorporating subjective expectations that deviate from rational expectations helps explain the variation in the Sharpe ratio.
In Chapter 2, I examine the subjective risk-return relation from the observed stock return data in the presence of information rigidity in investors' volatility expectations. Based on the present-value approach of Campbell and Shiller (1988), I develop an empirical model for excess stock returns by introducing the sticky information model of Mankiw and Reis (2002) into aggregate subjective volatility expectations, while using realized volatility to capture time-varying risk. The estimation results based on U.S. monthly excess stock returns and realized volatility suggest that a significant information rigidity component and a positive and statistically significant subjective risk-return relation are identified from the observed stock return data. Meanwhile, the restriction among parameters implied by the present-value approach is rejected, indicating that other factors may influence stock return variations. I suggest that investors' overextrapolative belief may help explain the rejection of the restriction. I also find a state-dependent information rigidity: it increases during a period with lower macroeconomic volatility. Consistent with the findings of Coibion and Gorodnichenko (2015), the estimation results indicate that the degree of information rigidity increased during the Great Moderation period.
Chapter 3 explores the regime dependency and time variation of the relative importance of cash flow news and discount rate news in explaining excess stock return variance. To this end, I apply the variance decomposition method of Campbell and Ammer (1993) to the threshold VAR (TVAR) and the time-varying parameter VAR with stochastic volatilities (TVP-VAR-SV). To identify the regimes in the stock market, I use the Chicago Fed's financial condition index and investor sentiment index constructed by Baker and Wurgler (2006). The variance decomposition results using TVAR suggest that the contribution of discount rate news increases during tight financial conditions or high investor sentiment regimes. The result of TVP-SV-VAR indicates that cash flow news has become more important than discount rate news after the 1990s. I propose possible explanations for the results. First, the regime-dependent relative importance may be associated with the change in attention allocations of investors and the asymmetric stock return predictability across the regimes. Second, the reversal of the relative importance after the 1990s may be attributed to the less volatile discount rate news caused by increased information rigidity and changes in the return-earnings relationship after the onset of the Great Moderation
Smart Micropantry: Enhancing Food Accessibility through Embedded Sensing and Automation
Get PDFThesis (Master's)--University of Washington, 2025Community micro-pantries provide critical access to food for individuals experiencing food insecurity, but their distributed and volunteer-managed nature makes it difficult to monitor usage, food safety, and inventory. In this work, we present Smart Micropantry, an embedded sensing system designed to enhance accessibility and reliability of community pantries. Our system integrates weight, environmental, and access sensors with a low-power microcontroller, enabling continuous monitoring while remaining energy-efficient. The design supports wireless data transmission for real-time updates while maintaining robustness in outdoor environments. We implemented and deployed the Smart Micropantry system in a real-world setting, demonstrating its ability to track pantry usage patterns, detect changes in inventory, and monitor environmental conditions relevant to food safety. Our evaluation shows that the system can provide accurate, reliable measurements at low power cost, making it sustainable for long-term operation. Insights from deployment highlight both technical performance and opportunities for community engagement, as the system can inform restocking schedules, identify potential food safety risks, and improve equitable access. This work demonstrates how embedded sensing and ubiquitous computing can extend beyond traditional laboratory settings into grassroots community infrastructure. Smart Micropantry offers a scalable model for integrating low-cost, low-power sensing into public resources, ultimately contributing to improved food accessibility and community well-being
Dynamical Quantum Phase Transitions, Scrambling and Quantum Simulations of Many-Body Neutrino Systems
Get PDFThesis (Ph.D.)--University of Washington, 2025This dissertation covers three main topics: dynamical quantum phase transitions, the far-from-equilibrium phenomenon of quantum information scrambling, and quantum simulations on both trapped ion and superconducting qubit devices, all in the context of many-body neutrino systems.For the dynamical quantum phase transitions study in the first chapter, the analysis of Loschmidt echos within dense neutrino systems yields insight into a system's initial state requirements needed to achieve a dynamical quantum phase transition (DQPT). Focus is paid to Loschmidt echo crossing distributions, which confirm the presence of two distinct classes of DQPTs in two-flavor neutrino systems. Further analysis reveals a nontrivial dependence on the coupling angle distributions chosen for the two-body interaction term. The results establish two distinct classes of DQPT's in two flavor neutrino systems, verfied via robust statistics.
Scrambling as diagnosed by the Out-of-Time-Ordered Correlator (OTOC) has been demonstrated in the Sachdev-Ye-Kitaev model, Transverse Field Ising Model, the transverse axial next nearest neighbor Ising model among many others. However they have yet to be characterized in many-body neutrino systems. Such systems are often modeled as all-to-all connected random-Heisenberg spin chains.
In the second chapter, this work demonstrates numerical evidence for scrambling's occurrence in two-flavor many-body neutrino systems. The results demonstrate dynamical quantum phase transitions (DQPTs) potential role as a witness for scrambling in many- body neutrino systems. We see what appears to be discreet modes of scrambling times corresponding to a system's first DQPT occurrence. We attempt to formulate an analytical argument resting on the concept of weak measurement schemes to explain how the DQPTs can serve as a witness for OTOCs in families of random-coupled two-flavor many-body neutrino systems in the forward scattering limit.
In the last chapter, quantum circuits for three flavor many body neutrino systems are constructed, for both qubit and qutrit devices. The qubit-based circuits are run on super- conducting qubit devices with heavy-hex connectivity and trapped ion devices with all-to-all connectivity, demonstrating the one of the first quantum simulations of three flavor neutrino
systems on two level devices. This work demonstrates a proof of principle for simulating three-flavor neutrino systems on two-level devices, the performance of the qubit circuits on each device, and calculation of physical observables off of the device
The Pacific sand dollar Dendraster excentricus: A New Model to Explore Novelty in Neural Circuits
Get PDFDecoding how neural circuitry functions and evolved is no small task, but studying how disparate nervous systems produce similar behaviors may offer unexpected insights. Cephalopods possess complex hierarchical nervous systems with a centralized brain adjacent to their decentralized nerve ring, cords, and ganglia of their arms, whereas echinoderms lack a centralized brain but have an independently evolved nerve ring with ganglia and radial nerves. Both groups possess numerous specialized appendages on “multi-arm” axes that serve analogous locomotor and sensory functions. This study focuses on how Dendraster excentricus, the Pacific sand dollar, can be used as a novel research model to investigate neural circuit evolution, given its unique secondary bilateral symmetry that is superimposed on the ancestral pentaradial structure observed in sea urchins. We used time-lapse videography in lab and field settings to generate behavioral ethograms of D. excentricus and initiated work using deep learning tools to analyze locomotion. These approaches, along with microCT and confocal imaging, will enable us to compare body movements and coordination of tube feet and spines across individuals. Initial findings are reported here. Specifically, our findings revealed distinct locomotor behaviors in D. excentricus that suggest directional control and spatial awareness, despite its decentralized neural anatomy. This research contributes to the understanding of how morphological and ecological divergence shape neural circuit functionality and provides a comparative framework for studying the evolution and function of nervous systems in marine invertebrates
Quasi-Background-Free Neutrinoless Double-Beta Decay Searches with LEGEND: Statistical Methods and Cryogenic SiPM Characterization
Get PDFThesis (Ph.D.)--University of Washington, 2025As an electrically neutral and massive fermion, the neutrino is the only Standard Model particle whose Lagrangian mass term could include a Majorana term. The nature of the neutrino mass is exciting because it is inherently Beyond the Standard Model: neutrino oscillations show that the neutrino has a mass and that it violates conservation of individual lepton numbers, in direct contradiction to predictions of the Standard Model. The most sensitive probe of the Majorana nature of the neutrino is the search for a hypothetical second-order weak decay called neutrinoless double-beta decay. Observation of this decay would prove that conservation of total baryon minus lepton number is violated and that the neutrino has a Majorana mass term. The Large Enriched Germanium Experiment for Neutrinoless-double beta Decay (LEGEND) collaboration is searching for the neutrinoless double-beta decay of 76Ge by deploying an array of highly enriched germanium detectors inside of a liquid argon cryostat. In order to fully cover the allowed parameter space for inverted ordered light Majorana neutrinos, the LEGEND collaboration is pursuing a phased approach. The already-built LEGEND-200 experiment aims for a half-life discovery sensitivity of 1E+27 years using 200 kg of enriched detectors, and the proposed tonne-scale follow-on experiment LEGEND-1000 aims for a discovery sensitivity of 1E+28 years. The success of these experiments rests in their great ability to reduce external background through techniques such as pulse shape discrimination and liquid argon scintillation
anti-coincidence vetoing. The LEGEND-200 experiment has completed its first year of searching for neutrinoless double-beta decay with an accumulated exposure of 61 kg·yr. This thesis reports the first frequentist statistical analysis performed on data from the LEGEND experiment. A new Python-based framework, freqfit, is introduced to robustly and rapidly produce frequentist statistical inference on unbinned data. Using this framework, no evidence of neutrinoless double-beta decay is found (p = 0.1), and a lower limit on the half-life is placed at T_1/2 > 5 × 1E+25 years (90% confidence level). The background index for the group of detectors mainly comprised of the geometry to be used in LEGEND-1000 is computed to be 5^+3_-2 × 1E−4 counts/keV/kg/yr. A frequentist joint analysis incorporating LEGEND-200 data and the data from two recent 76Ge experiments, the MAJORANA DEMONSTRATOR (MJD) and Germanium Detector Array (GERDA), is also presented. The combined analysis also observes no evidence for a signal (p = 0.29) and results in the strongest lower limit, and highest half-life sensitivity, on the half-life in 76 Ge to-date: T_1/2 > 1.9 × 1E+26 years (90% confidence level). This half-life limit can be converted into a limit of the effective Majorana mass of the neutrino, assuming mediation by a light Majorana neutrino and a range of phenomenological nuclear matrix elements, and yields mββ < 75 − 200 meV (90% confidence level). With fewer than one expected count due to accidental radioactive background near the Q-value of neutrinoless double-beta decay, LEGEND-1000 will be a quasi-background-free experiment. The behavior of the frequentist profile likelihood ratio treatment for statistical inference on unbinned data from quasi-background-free experiments is investigated. It is found that test statistic distributions in this regime deviate from the asymptotic form predicted by Wilks’ theorem — it is required to generate pseudo-experiments for these statistical analyses. The coverage is computed for ensembles of pseudo-experiments generated with a known Poisson-distributed background index. We show that the profile likelihood ratio treatment guarantees coverage very close to the nominal value for these ensembles. In order for LEGEND-1000 to reach its nominal background index goal of 1E−5 counts/keV/kg/yr, it is necessary that its liquid argon detector collects as much scintillation light in its silicon photomultiplier (SiPM) readout as possible. The detection efficiency of the liquid argon system is directly proportional to the photon detection efficiency (PDE) of the SiPMs. The PDE is a well-known quantity reported by the manufacturer, at room temperature; however, LEGEND operates its SiPMs at cryogenic temperature. This work describes the operation and results of a cryogenic SiPM characterization test stand built at the University of Washington. We report that two SiPMs exhibit a nearly 20% drop in their PDE at liquid nitrogen temperature relative to their room temperature values. This drop was measured for the green wavelengths (562 nm) of light that match the optical emission spectra of light collection technology for LEGEND-1000. This drop in the PDE represents an important input for forecasting the background index for LEGEND-1000 in order to guarantee that it reaches its discovery sensitivity goal
Beyond Grief: Dirge Writing in the Han China (206 BCE-220 CE)
No full textThesis (Master's)--University of Washington, 2025The dirge (lei 誄) was a funerary text composed to commemorate sociocultural elites in Han funerary practices. However, it became obsolete after the Tang dynasty (618-907 CE) and receives limited scholarly attention today, leaving a significant aspect of Han memorial culture understudied. After an overview of the current scholarship on the dirge, this study aims to provide a cultural history of the dirge during the Han. Instead of answering “what is a dirge” with a static definition, this study questions the conventional understanding of genre and examines the interaction between the function and textual structure of the dirge. By analyzing the writer-mourner-deceased power dynamics, this study will pay specific attention to the role of the writer in dirge production. The first two chapters will examine the structure and function of the dirge. In addition to providing a standardized paradigm of dirge writing during the Han, Chapter 1 plans to elucidate the orality and materiality that destabilized the text. I will challenge the conventional idea of genre by offering a function-centered perspective. I will analyze the composite texture of the dirge to further explain its textual instability. Chapter 2 will examine the function of dirge by comparing the dirge with functionally and textually similar texts. I try to answer whether the function of the dirge lies in conferring the posthumous name. By elucidating the changing role of scholar-officials as dirge writers in late Eastern Han, this study will delve into the power dynamics between the writer, the mourners, and the deceased in the production of the dirge. Chapter 3 will focus on the shifting communal identity of scholars and its reflection in their archetypical reconstruction of the deceased. Chapter 4 will elaborate the interplay between funerary practice and personal expression in Han-Wei period dirges. I will examine Cao Zhi’s 曹植 (192-232 CE) “Dirge for Wang Zhongxuan” in terms of the interaction between his identity and the text, and how this demonstrates the distinctive state-scholar relationship in that sociocultural context.
This study aims to dismiss the teleological narrative of the dirge and situate the dirge as a textual product of the funerary practice within its sociocultural context. This study will also enrich scholarly understanding of the relationship between text, writer, and context in dirge production. Through the lens of the dirge, this study will contribute to the understanding of genre, authorship, and identity against the backdrop of Han memorial culture
Methods for time series network analysis
Get PDFThesis (Ph.D.)--University of Washington, 2025Statistical networks can encode arbitrary relationships between variables in a system. Due to this flexibility, scientific hypotheses about interactions between variables can typically be formulated as a statistical network analysis. In addition to analyzing static networks, studying how statistical networks change in response to experimental or environmental conditions is often of scientific interest. A network is typically defined as a set of vertices and edges. Specifically a network or graph, G, can be written as G = (V, E) where V = {1,...,k} are the vertices or variables and E is the edge set that encodes the relationship between variables. A common example of a statistical network is the correlation matrix, where an edge represents the correlation between variables. While analysis of networks and their changes are ubiquitous across many domains, our work is motivated specifically by applications in which networks are derived from time series data. In contrast to independent data, statistical analysis of time series data is complicated by the inherent serial correlation. In practice, the degree of this correlation is unknown and network analysis methods that can flexibly handle varying degrees of dependence are needed. We approach this problem from two angles. The first angle, used in the first two portions of this thesis, focuses on developing methods with minimal assumptions on temporal dependence. In the third portion of this thesis we approach this problem from the second angle which attempts to leverage the flexibility of deep learning methods to analyze statistical networks. In the first chapter, we propose a novel order selection method in vector autoregressive (VAR) models. Order selection is an essential step in fitting VAR models and while many order selection methods exist, all come with weaknesses. Our proposed order selection method is based on the observation that the expected squared error loss is flat once the fitted order reaches or exceeds the true order. We show that under mild assumptions on the underlying process our new order selection method consistently estimates the true order. Motivated by applications in neuroscience, the second chapter of this thesis develops a novel estimation and inference procedure for a difference in the inverse spectral densities. In neuroscience, it is often of interest to study how brain networks change in response to electrical stimulation with the hopes of developing stimulation-based treatments for neurodegenerative diseases. Furthermore, it is essential to study networks in the frequency domain as higher frequencies contain key brain connectivity information. With this in mind, we develop methods to directly estimate and perform statistical inference on a difference in inverse spectral densities. Crucially, our method relies on minimal assumptions and can flexibly handle a large range of data dependence. The last chapter of this thesis proposes a new deep learning-based change-point detection framework. The core idea behind this method is a continuous approximation of the indicator function. With this approximation, change-points can be specified as parameters of a deep learning model. Thus, change-points and model parameters can be jointly learned using stochastic optimization techniques. The proposed framework is general and can be applied to both independent and dependent data, such as time series data. Furthermore, the framework is model-agnostic and thus can be used to encode networks and study their changes
Data-Aware Complexity Analysis and Program Optimization
Get PDFThesis (Ph.D.)--University of Washington, 2025This dissertation explores the problem of analyzing and optimizing data-dependent programs from a theoretical and practical perspective. The performance of these programs depends in a complex manner on the distribution of the input data, and they arise in many contexts, e.g. databases, sparse tensor programming, and graph analytics. By definition, these programs cannot be optimized by considering the code alone, so an optimizer for them must consider information about the data distribution. This dissertation presents two new theoretical approaches for analyzing data-dependent programs by bounding the size of their intermediate results: the degree sequence bound and partition constraints. It then describes two practical systems for producing these bounds: SafeBound and COLOR. Lastly, we present a state-of-the-art optimizer for sparse tensor programs, Galley, that demonstrates the value of data-aware optimization