12023 research outputs found
Sort by
Investigating Drivers of Repeated Behaviors in Field Data
Get PDFThis dissertation investigates the influences on frequently repeated human behaviors (e.g. eating, exercising, washing hands) using empirical tests on field data. While some of the phenomena discussed have been studied in lab settings (e.g., self-regulation failures, insensitivity to reward devaluation), these studies present some of the first tests of these behavioral phenomena in the field. This dissertation also assembles a number of methodologies which can be used to study individual-level field data, informed by an interdisciplinary perspective on social and decision science research.
The first chapter uses field data to study spillovers across behavioral domains, namely exercise and food choice. This work joins a small group of papers which document field evidence related to domain spillovers and failures of self-regulation. Most of the existing research on self-regulation has been conducted in controlled laboratory settings, where participants are either asked to imagine making hypothetical restrained choices or exert effort on a laboratory task as a proxy for making a restrained choice. As is the critique of many lab studies without direct field equivalents however, it is debatable whether the self-regulation behaviors observed in survey and laboratory settings necessarily generalize to the field. We fill this gap by looking at how natural (rather than incentivized) changes in exercise systematically affect food choice, thus empirically identifying spillovers across two behavioral domains in field data. We find that, even after controlling for individual fixed effects, there is a robust effect of morning exercise on the healthiness of a lunch choice. We complement the analysis of field data with surveys to better understand the mechanism driving this result.
The second chapter presents a novel methodology for identifying behaviors that are highly and predictably context-sensitive, and thus candidates for being habitual. While there is a large body of laboratory research documenting the mechanisms underlying well-developed habits in animals and humans, there is much less field research on how human habits naturally develop over time. Using two large datasets on gym attendance and handwashing behavior, we use machine learning to statistically classify when choices are predicted by an identifiable set of context variables. This technique generates a person-specific measure of behavioral predictability, which can then be used to study individual differences in predictability and speed of habit formation. This allows us to establish two important discoveries. First, the sets of context cues that are predictive of individual-level behavior are different for different people. Specifically, while historical behavior is an important universal predictor, other context variables such as day of the week or month of the year have more heterogeneous effects. Second, contrary to common wisdom, there is no "magic number" for how long it takes to form a habit. Instead, the speed of habit formation appears to vary significantly, both between behavioral domains and between individuals within domains.
The third chapter uses a novel methodology to run a field experiment testing the effect of a price promotion on consumer behavior. The goal of this "pilot study" is to credibly dissociate predictions made by brand loyalty/habit formation from reference-dependence theories. A customizable vending machine serves as a "mini-retailer," allowing for full control of price promotion details in an ecologically valid setting. The vending machine allows controlling for stockpiling behavior, an important concern for empirical work analyzing price promotions in the marketing literature. Analysis of the data collected from this pilot study suggests that price promotions increase the sales of both discounted and non-discounted items, as well as the total number of unique customers making purchases. Furthermore, in line with the loss leader hypothesis, more items are purchased during the sale period overall.</p
Towards Learning Robotic Dynamics: Application to Multirotor Takeoff and Landing
Get PDFMultirotors have become widespread but their usage is still limited. Ensuring safety during take-off and landing is still an open problem. Towards this goal this thesis proposes two different solutions to address this problem. The two approaches complement each other and they are tested on hardware.
The first approach is to design a vehicle that is stable during take-off, despite hardware failures or unsteady take-off platforms. A solution is to use a ballistic launch to impose a deterministic path, preventing collisions with its environment. Following this approach led to the development of several SQUID (Streamlined Quick Unfolding Investigation Drone) vehicles. The main challenges are the ballistic initial flight, large accelerations during launch, and limited volume. A first prototype was developed, which is able to transition mid-flight from stable ballistic flight to a fully controllable multirotor. The system has been fabricated and field tested from a moving vehicle up to 50mph to successfully demonstrate the feasibility of the concept and experimentally validate the design's aerodynamic stability and deployment reliability. A second prototype expanded the first one's capabilities incorporating fully-autonomous vision-based navigation, while keeping the ballistic passive stability and stable transition abilities. The new design includes a more reliable plate-based structure and more effective folding fins.
The second approach focuses on designing controllers that are safe regardless of the platform. For that purpose, a Model Predictive Control (MPC) is used to ensure state and input constraints. Given the highly non-linear dynamics platforms and fast dynamics that require a quick controller evaluation, the work in this thesis is built using Koopman Operator theory, which allows tools from linear analysis to be applied to systems with inherently non-linear dynamics. One of the main contributions is a novel method to find Koopman Eigenfunctions directly from data. Another key contribution is an episodic approach to model non-linear actuation dynamics. The proposed method is first tested on simulation and it outperforms comparable approaches. The method is also demonstrated on-board a multirotor for a fast landing application, where the nonlinear ground effect is learned and used to improve landing speed and quality. An additional extension considers model uncertainty in the MPC architecture, where an Ensemble Kalman Sampler is used to learn the uncertainty distribution.</p
Dielectric Metasurfaces for Integrated Imaging Devices and Active Optical Elements
Get PDFOptical dielectric metasurfaces have shown great advances in the last two decades and become promising candidates for next-generation free-space optical elements. In addition to their compatibility with scalable semiconductor fabrication technology, metasurfaces have provided new and efficient ways to manipulate diverse characteristics of light. In this thesis, we demonstrate the potential of dielectric metastructures in the realization of compact imaging devices, reconfigurable optical elements, and multi-layer inverse-designed metasurfaces. With the metasurfaces’ extreme capability to simultaneously control phase and polarization, we first showcase their potential toward optical field imaging applications. In this regard, we demonstrate a system of dielectric metasurfaces and designed random metasurfaces for single-shot phase gradient microscopes and computational complex field imaging system, respectively. Then, we propose nano-electromechanically tunable resonant dielectric metasurfaces as a general platform for active metasurfaces. For example, we demonstrate two different types of the phase and amplitude modulators. While one utilizes resonant eigenmodes in the lattice such as leaky guided mode resonances and bound-states in the continuum modes, the other is based on the high-Q Mie resonances in the dielectric nanostructures where symmetry is broken. In addition to the modulation of the phase and amplitude, we also show tuning of strong chiroptical responses in dielectric chiral metasurfaces. Next, we experimentally demonstrate inverse-designed multi-layer metasurfaces. Not only do they provide increased degree of freedom in the design space, but also overcome limits of conventional design methods of the metasurfaces. Finally, we summarize the presented works and conclude this thesis with a brief outlook on what aspects of the metasurfaces can be important for their real-world applications in the future and what challenges and opportunities remain.</p
Analogues of Amenability
Get PDFIn this thesis I study multiple notions related to and inspired by amenability coming from the points of the view of random walks on groups, dynamical systems, Borel equivalence relations, and descriptive linear algebra. In particular I study notions related to harmonic functions, invariant measures, hyperfiniteness, and dichotomy theorems
Learning to Optimize: from Theory to Practice
Get PDFOptimization is at the heart of everyday applications, from finding the fastest route for navigation to designing efficient drugs for diseases. The study of optimization algorithms has focused on developing general approaches that do not adapt to specific problem instances. While they enjoy wide applicability, they forgo the potentially useful information embedded in the structure of an instance. Furthermore, as new optimization problems appear, the algorithm development process relies heavily on domain expertise to identify special properties and design methods to exploit them. Such design philosophy is labor-intensive and difficult to deploy efficiently to a broad range of domain-specific optimization problems, which are becoming ubiquitous in the pursuit of ever more personalized applications.
In this dissertation, we consider different hybrid versions of classical optimization algorithms with data-driven techniques. We aim to equip classical algorithms with the ability to adapt their behaviors on the fly based on specific problem instances. A common theme in our approaches is to train the data-driven components on a pre-collected batch of representative problem instances to optimize some performance metrics, e.g., wall-clock time. Varying the integration details, we present several approaches to learning data-driven optimization modules for combinatorial optimization problems and study the corresponding fundamental research questions on policy learning. We provide multiple practical experimental results to showcase the practicality of our methods which lead to state-of-the-art performance on some classes of problems.</p
Investigating the Catalytic Mechanisms of Bio-degrading Copper Proteins: Multi-copper Oxidases (MCOs) and Lytic Polysaccharide Monooxygenases (LPMOs)
Get PDFLignin and cellulose comprise a large portion of the renewable biomass on Earth. However, substantially due to laborious course of processing, the conversion efficiency of these biomaterials to accessible biofuel is very low. Therefore, effective depolymerization and utilization of these biopolymers are requirements for environmentally friendly and sustainable energy development. In the hope of finding solutions to these biomass utilization challenges, there have been growing interests in using biodegrading metalloenzymes as active biocatalysts. However, there still remain many questions regarding mechanistic details of enzyme catalysis and effective application of these enzymes. This thesis focuses on investigating the redox chemistry involved in the catalytic mechanisms of two main lignin- and cellulose- degrading copper enzymes: multicopper oxidases (MCOs) and lytic polysaccharide monooxygenases (LPMOs).
MCOs are capable of aerobic oxidation of lignin as their primary function, but the nature of their substrate variability also allows the oxidation of not only diverse high potential organic and inorganic complexes, but also earth abundant divalent metal ions such as manganese. LPMOs, on the other hand, enable the cleavage of glycosidic bonds in recalcitrant insoluble cellulosic substances, which are not degradable by other hydrolytic enzymes such as endoglucanases and cellulobiohydrolases.
It is remarkable that nature has created such versatile enzymes with specific active site metals and redox-active amino acids involved in electron transfer, which contribute to substrate oxidation as well as enzyme survival against oxidative damage during catalysis. By gaining a deeper understanding of how these enzymes work, we could greatly enhance current usage efficiencies and develop more energy-efficient biocatalysts.
Chapter I gives an introduction to biological coppers, two groups of bio-degrading copper enzymes: multicopper oxidases (MCOs) and lytic polysaccharide monooxygenases (LPMOs), and the role of redox-active amino acids in electron transfer and enzyme catalysis. For the MCO work, a thermophilic laccase (Tth-lac) from Thermus thermophilus HB27 and a CotA laccase (CotA-lac) from Bacillus Subtilis were studied. For the LPMO work, two cellulose active LPMOs (ScLPMO10B and ScLPMO10C) and a chitin active LPMO (BlLPMO10A) were studied.
Chapter II describes thermodynamic aspects of Tth-lac catalysis. The temperature dependence of the formal potential of type I copper (CuT1) in Tth-lac is reported, and the interplay between many competing dynamic and thermodynamic factors which results in thermostability and activity of Tth-lac is discussed.
Chapter III reports the electron transfer (ET) kinetics data obtained with Tth-lac using the transient absorption spectroscopy. The results of photochemical electron/hole transfer studies indicate that the chains of Trp and Tyr can participate in electron transfer through Tth-lac, which could potentially have a role in enzyme catalysis as well.
Chapter IV discusses the protective role of a Trp/Tyr pair positioned close to the trinuclear copper cluster (TNC) in Tth-lac. It is indeed remarkable that laccases are capable of utilizing the power of oxygen to catalyze the oxidation of diverse high-potential substrates. But, as a tradeoff, the utilization of dioxygen can make the enzyme highly susceptible to oxidative damage. Chapter IV provides supporting evidence that led us to conclude that the TNC-proximal Trp/Tyr pair functions as an internal antioxidant for prolonging the enzyme lifetime.
Chapter V describes investigations on the factors that affect MCO catalysis, which include the potentials of the active site coppers, possible reactive intermediates, and common structural motifs. Based on the structural homology between Tth-lac and CotA-lac, some preliminary work done on CotA-lac is also reported.
Chapter VI outlines the work on LPMOs. After the successful expression and purification of ScLPMO10B, ScLPMO10B and BlLPMO10A, standard activity assays were done with insoluble cellulose and chitin substrates to confirm the enzyme activity. The results are compared with that from the photo-degradation experiments to investigate if the photochemically generated Cu(III) species are active intermediates in LPMO catalysis.
Chapter VII reports the results on bioinformatics analysis on the distribution of vicinal amino acids in different enzyme classes. This study was to examine the biological significance of amino acid pairs and clusters existing in many different enzyme classes, with vicinal surface tyrosines in CotA-lac as an underlying motivation behind the work.
This thesis demonstrates that MCOs and LPMOs are truly versatile enzymes which can oxidize such diverse refractory substrates, and there could be multiple pathways that the enzymes achieve this task. As shown so far, not only the active site metals but also the chain of redox-active amino acids as well as metal coordinating residues can contribute to enzyme catalysis.</p
Identification of Structural Damage, Ground Motion Response, and the Benefits of Dense Seismic Instrumentation
Get PDFThis study explores the problems of identifying structural damage in steel frame buildings, through the use of dense instrumentation over the height of the building, and of characterizing the ground motion response in urban Los Angeles following the 2019 Ridgecrest earthquakes, through the use of dense instrumentation from available seismic networks, including the very dense Community Seismic Network.
First we explore the possibility of tracing possible nonlinear behavior of a structure by updating an equivalent linear system model in short time segments of the earthquake-induced excitation and response time histories, using a moving time window approach. The stiffness and damping related parameters of the equivalent linear model are estimated by minimizing a measure of fit between the measured and model predicted response time histories for each time window. We explore the effectiveness of the methodology for two example applications, a single-story and a six-story steel moment frame building. For the single-story building, the methodology is shown to be very effective in tracing the nonlinearities, while the six-story building is designed to also reveal the limitations of the methodology, mainly arising from the different types of model errors manifested in the formulation.
Next, we investigate the problem of structural damage identification through the use of sparse Bayesian learning (SBL) techniques. This is based on the premise that damage in a structure appears only in a limited number of locations. SBL methods that had been previously applied for structural damage identification used measurements related to modal properties and were thus limited to linear models. Here we present a methodology that allows for the application of SBL in non-linear models, using time history measurements recorded from a dense network of sensors installed along the building height. We develop a two-step optimization algorithm in which the most probable values of the structural model parameters and the hyper-parameters are iteratively obtained. An equivalent single-objective minimization problem that results in the most probable model parameter values is also derived. We consider the example problem of identifying damage in the form of weld fractures in a 15-story moment resisting steel frame building, using a nonlinear finite element model and simulated acceleration data. Fiber elements and a bilinear material model are used in order to account for the change of local stiffness when cracks at the welds are subjected to tension and the model parameters characterize the loss of stiffness as the crack opens under tension. The damage identification results demonstrate the effectiveness and robustness of the proposed methodology in identifying the existence, location, and severity of damage for a variety of different damage scenarios, and degrees of model and measurement errors. The results show the great promise of the SBL methodology for damage identification by integrating nonlinear finite element models and response time history measurements.
The final part of the thesis involves studying the ground motion response in urban Los Angeles during the two largest events (M7.1 and M6.4) of the 2019 Ridgecrest earthquake sequence using recordings from multiple regional seismic networks as well as a subset of 350 stations from the much denser Community Seismic Network. The response spectral (pseudo) accelerations for a selection of periods of engineering significance are calculated. Significant spectral acceleration amplification is present and reproducible between the two events. For the longer periods, coherent spectral acceleration patterns are visible throughout the Los Angeles Basin, while for the shorter periods, the motions are less spatially coherent. The dense Community Seismic Network instrumentation allows us to observe smaller-scale coherence even for these shorter periods. Examining possible correlations of the computed response spectral accelerations with basement depth and Vs30, we find the correlations to be stronger for the longer periods. Furthermore, we study the performance of two state-of-the-art methods for estimating ground motions for the largest event of the Ridgecrest earthquake sequence, namely 3D finite difference simulations and ground motion prediction equations. For the simulations, we are interested in the performance of the two Southern California Earthquake Center 3D Community Velocity Models (CVM-S and CVM-H). For the ground motion prediction equations, we consider four of the 2014 Next Generation Attenuation-West2 Project equations. For some cases, the methods match the observations reasonably well; however, neither approach is able to reproduce the specific locations of the maximum response spectral accelerations, or match the details of the observed amplification patterns.</p
Improving Reduced Order Models of Soil-Structure Interaction Using an Ensemble Kalman Inversion Finite Element Model Updating Framework
Get PDFIn civil engineering, almost all structures are somehow in contact with soil - i.e., have foundations or support elements that either rest on or are embedded in soil. Thus, their seismic response is governed by the interaction between the structure, the non-structural components, the foundation, and the surrounding soil. Predicting such interaction becomes increasingly complex when uncertainties of soil and structural material, ground motion variability, and dissipation mechanisms are considered. The accuracy of numerical models to predict the linear or nonlinear responses of structures depends not only on how well the uncertainties in the material properties and input motion are estimated, but also on how well the various sources of energy dissipation and their interaction are modeled. Therefore, high-fidelity simulation of soil-structure interaction (SSI) problems require advanced models that can capture the nonlinear behavior of soils and structures, and parallel computing capabilities to optimize the cost associated with large scale problems. In spite of this fact, SSI in practice is widely accounted for using fixed-base building and reduced-order-models (ROM) which usually trade accurate solution for fast ones. Unfortunately, if SSI effects are neglected or poorly estimated, then critical response measures of a structure can be over- or under-estimated, which in turn can lead to unsafe or overly
conservative designs.
Motivated by the previous challenge, in this thesis work we present a robust and efficient framework for finite element model (FEM) updating based on ensemble-Kalman inversion (EnKI). The EnKI-FEM updating framework is used to obtain suitable parameters to inform a ROM from data generated using high-fidelity FEM simulations. Since high-fidelity SSI simulations call for accurate and computationally efficient capabilities, as a part of this work, we developed Seismo-VLAB, a simple, fast, and extendable C++ finite element software to optimize large-scale simulations of dynamic and nonlinear SSI problems. The EnKI-FEM updating framework is thus integrated in Seismo-VLAB allowing to identify any parameter of the ROM without compromising accuracy. The so-generated ROM are finally employed to propose a new dimensionless frequency mapping to estimate the soil impedance for time domain analysis and to investigate soil-structure-interaction effects at a regional-scale. The presented methodology is general enough and it can be extended to more complex structural and/or geotechnical systems, allowing to construct highly-accurate ROM in a simple manner.</p
Tensor Network Methods for Nonequilibrium Statistical Mechanics
Get PDFLarge deviation theory has emerged as a powerful mathematical scaffolding for studying nonequilibrium statistical mechanics, particularly for characterizing the macroscopic effects of microscopic fluctuations. While the large deviation approach is firmly established, it relates the effects of fluctuations to the likelihood of exponentially rare events, which naively requires exponentially large simulation costs. This, in turn, necessitates the development of appropriate numerical simulation techniques. While the standard Monte Carlo toolkit has expanded to incorporate methods towards making rare events typical, in this thesis I propose and evaluate a powerful unorthodox approach adopted from quantum simulation, namely tensor network algorithms, which can work in concert with standard methods to deepen our understanding of nonequilibrium phenomena.
As a testbed for this novel approach, I consider the dynamical phase behavior of several versions of the simple exclusion process, a paradigmatic model of classical driven diffusion. Using a matrix product state, a one-dimensional tensor network ansatz, and the density matrix renormalization group algorithm, a corresponding optimization routine, I characterize the dynamical phase transition between a jammed and maximal current phase in both the one-dimensional and multi-lane simple exclusion processes. The matrix product state is found to be an efficient representation of the nonequilibrium steady-state biased to arbitrarily rare currents via large deviation theory. Because the one-dimensional ansatz is limited to finite-width systems, I extend this success to study the fully two-dimensional simple exclusion process. There, the projected entangled pair state, a two-dimensional tensor network ansatz, is used with the time evolution via block decimation algorithm to demonstrate that the phase transition observed in one-dimension persists in the fully two-dimensional system.
Towards the goal of making tensor network methods adaptable for a broad range of physically important systems, both classical and quantum, I also present progress towards studying systems in the continuum limit with interacting particles in two dimensions. This builds upon previous work proposing tensor network representations of quantum operators with long-range interactions in two dimensions by evaluating three operator representations in practice and finding two competitive and viable approaches.</p
C. elegans Models of ASD-Associated Missense Variants
Get PDFThe evolving next-generation sequencing technology accelerates the identification of disease-associated genetic variants. However, interpretation of these variants remains challenging, especially variants with subtle effects such as missense variants. Missense variants account for a large proportion of genetic variants in human diseases, including autism spectrum disorder (ASD). The causal relationship of most missense variants in the pathogenesis of ASD has not yet been demonstrated, and an experimental method systematically prioritizing missense alleles can gain crucial insight into the molecular basis for disease pathology. Therefore, I developed an in vivo multi-cellular system using Caenorhabditis elegans to systematically evaluate the functional consequences of disease-associated missense variants. I identified highly conserved human ASD-associated missense variants in their C. elegans orthologs, used a CRISPR/Cas9-mediated homology-directed knock-in strategy to generate missense mutants, and analyzed their impact on behaviors and development via several broad-spectrum assays. Overall, I generated 60 ASD-associated missense variants and characterized these missense mutant strains using a fecundity assay, an automated locomotor tracking system, and a chemotaxis assay. I found that 19% of the human disease-associated alleles have conserved loci in their C. elegans orthologs. Among the genes I tested, 64-70% of the missense variants predicted to perturb protein function showed detectable phenotypic changes in morphology, locomotion, or fecundity. Our results also revealed that missense mutants in different gene networks displayed distinct phenotypic profiles. Moreover, I focused on studying the genetic properties of missense variants on two ASD risk genes. I discovered the developmental defects in the ALDH1A3 C174Y missense mutation involved in the retinoic acid signaling pathway. I also identified a conserved missense residue lin-45(K565N), orthologous to human BRAF(K499N), which displayed a hypersensitive non-dominant phenotype in the diacetyl chemotaxis assay that was capable of being inhibited by RNAi. The finding suggests a potential gain-of-function allele in BRAF, especially in the sensory function. To sum up, I established a working pipeline to systematically identify and generate evolutionarily conserved ASD-associated missense mutants in C. elegans. This approach will help assess the impact of a single missense mutation in the whole organism and prioritize consequential missense variants for further intensive analysis in vertebrate models and human cells.</p