12023 research outputs found
Sort by
Theoretical and Computational Analysis of Cell Migration in Complex Tissue Environments
Get PDFCells sense and respond in spatially structured environments, including soils and tissue. My Ph.D. projects centered on developing new theoretical models and computational methods to understand how cells migrate in complex environments.
The first project is more theoretical in nature, leveraging information theory to study how the spatial organization of cell signaling pathways are adapted to the cell's natural environment. In tissue and soil, cells must localize to their targets by navigating distributions of extracellular ligands that are spatially discontinuous, consisting of local concentration peaks, due to binding a non-uniform network of ECM fibers. It is unclear how cells navigate patchy environments while not getting trapped in local concentration peaks. To answer this question, we framed navigation as a problem of maximizing mutual information in space and developed a computational algorithm for computing signaling pathway architectures that maximize mutual information in simulated natural environments. We found that for cells in tissues and soils, dynamic localization of membrane receptors dramatically boosts sensing precision and enables cells to navigate to chemical sources 30 times faster, but this receptor localization strategy is relatively inconsequential for cells in purely diffusive environments. Further, we found that anisotropic receptor dynamics previously observed in immune cells and growth cones are nearly optimal as predicted by our model.
The second project is more computational in nature, leveraging multiplexed tissue imaging to understand T-cell migration in tumor microenvironments. Immunotherapies can halt or slow down cancer progression by activating either endogenous or engineered T-cells to detect and kill cancer cells. T-cells must infiltrate the tumor core for immunotherapies to be effective. However, many solid tumors resist T-cell infiltration, challenging the efficacy of current therapies. In collaboration with clinician scientists at Cedars-Sinai Medical Center, we developed an integrated deep learning framework, Morpheus, that takes large-scale spatial omics profiles of patient tumors, and combines a formulation of T-cell infiltration prediction as a self-supervised machine learning problem with a counterfactual optimization strategy to generate minimal tumor perturbations predicted to boost T-cell infiltration. We applied Morpheus to 368 metastatic melanoma and colorectal cancer samples assayed using 40-plex imaging mass cytometry, discovering cohort-dependent, combinatorial perturbations, involving CXCL9, CXCL10, CCL22 and CCL18 for melanoma and CXCR4, PD-1, PD-L1 and CYR61 for colorectal cancer, predicted to support T-cell infiltration across large patient cohorts. Using only raw image data, Morpheus also identified distinct therapeutic strategies for different patient strata such as cancer stage or fatty liver presence. Our work presents a paradigm for counterfactual-based prediction and design of cancer therapeutics using spatial omics data.</p
Fault Zone Structure and Rupture Behavior with Fiber-Optic Sensing and Second Moments
Get PDFThe structure of fault zones and the behavior of ruptures are indivisible. Fault structure is molded by the continued overprinting of slip events, and rupture propagation is highly sensitive to fault zone parameters. Observational constraints on both fault zone characteristics and the behavioral response of ruptures to fault variability are thus needed to understand earthquakes. Fault zones are narrow structures that are difficult to image in detail, particularly at depth. This means that fault structure is often oversimplified in rupture models and inversions. Earthquake source descriptions are frequently high dimensional. Fault slip distributions are often complicated and nonunique and seismicity catalogs can contain hundreds of thousands of events. This complexity can be difficult to reduce for the purpose of making clear conclusions on earthquake phenomenology. In this sense, observations of fault structure may benefit from a dimensionality expansion and observations of earthquakes may benefit from a dimensionality reduction. In Chapters 2-5 of this thesis I address the former problem. I show how an emergent technology, distributed acoustic sensing (DAS), that transforms fiber optic cables into dense arrays of strainmeters can be used to resolve fault zone characteristics with astonishing detail. This applies to small and large faults at shallow and deep depths. I define a framework for partitioning the seismic wavefield and show that scattered phases in earthquake wavefields encode information on the location and dimensions of small faults. I then investigate the Garlock Fault with an intersecting DAS array and uncover structural variability across the fault at shallow and seismogenic depths. I also use this array to investigate Moho topography, and find that the Garlock Fault offsets and, by extension, penetrates the mantle over a narrow width. In Chapters 6-8 of this thesis I address the latter problem. I show that second moments, both of source and seismicity distributions, can improve the clarity of observations and make drawing meaningful conclusions about rupture behavior possible. I first develop a framework to probabilistically invert for the second moments of source distributions and use it to investigate all large strike-slip events of the past three decades. These solutions show several patterns between faults and rupture behavior. I also create a seismicity catalog for the Ridgecrest earthquake sequence and use a second moment measure to constrain the evolution of fault orientation and the stress state
Safe and Scalable Learning-Based Control: Theory and Application in Sustainable Energy Systems
Get PDFFrom intelligent transportation systems to the smart grid, the next generation of cyber-physical systems (CPS) will substantially transform our society. It is vital that these systems are scalable and robust to uncertainties, with contextual awareness and fast adaptation. This dissertation presents progress towards addressing key challenges arising in the control of large-scale CPS, with a special focus on applications in sustainable energy systems.
Large-scale CPS such as the smart grid often consist of numerous interconnected and heterogeneous subsystems that must coordinate to achieve global objectives by exchanging information over a communication network.
Therefore, the first part of this thesis focuses on developing control algorithms that handle crucial design requirements emerging from scalability and communication constraints, such as disturbance localization, communication delay conformation, and distributed implementation.
Sustainable energy systems are crucial for reducing greenhouse gas emissions and mitigating climate change. However, the inherent unpredictability and large uncertainties associated with renewable generation pose significant challenges for maintaining system stability and safety. Traditional control approaches, while robust and effective for known system models, often fall short when faced with the dynamic and uncertain nature of modern power systems. In the second part of the thesis, we address this challenge by integrating machine learning techniques with model-based control methods using uncertainty sets constructed from real-time data. In particular, we will introduce and provide convergence guarantees for a classic uncertainty set estimation method. Building on these uncertainty sets, we combine learning and control techniques to tackle core CPS control problems, such as adversarial stability certification for linear time-varying systems as well as networked systems under communication constraints where the system models are unknown.
The final part of this thesis applies the developed methodologies to address the voltage control problem in power distribution networks with unknown grid topologies. We will combine online learning techniques and a robust predictive controller to achieve provably finite-time convergence to safe voltage limits, despite uncertainties in network topology and load variations. Our case study on a Southern California Edison 56-bus distribution system demonstrates the effectiveness of this approach in nonlinear, partial observation, and partial control settings.</p
Numerical Stability and Reduced Order Chemistry Modeling in Detonation Simulations
Get PDFThe coupling between shocks and chemistry in detonations poses a challenge for simulations. In this thesis, a simulation framework is developed to address key components of detonation modeling: numerical stability of shocks and discontinuities, and computational efficiency in chemistry modeling.
To ensure numerical stability in the vicinity of shocks, a variety of methods have been used, including shock-capturing schemes such as weighted essentially non-oscillatory (WENO) schemes, as well as the addition of artificial diffusivities to the governing equations. In this work, all necessary viscous/diffusion terms are derived from first principles, and the performance of these analytical terms is demonstrated within a centered differencing framework. The physical Euler equations are spatially-filtered with a Gaussian-like filter. Sub-filter scale (SFS) terms arise in the momentum and energy equations. Analytical closure is provided for each of them by leveraging the jump conditions for a shock. No SFS terms are present in the continuity or species equations. For contact discontinuities, the analytical SFS terms are identically zero. However, numerically, the transport of a contact discontinuity may result in artificial oscillations due to dispersive errors. To treat contact discontinuities, a WENO-like correction term is applied to the enthalpy transport. Implemented within a centered difference code, this filtered framework performs well for a range of shock-dominated flows without introducing excessive diffusion. In addition to providing new insight into the placement and form of required diffusion terms in the governing equations, this framework is general and may be used with any numerical scheme.
Chemistry modeling in detonations typically relies on two broad approaches: simplified models with one- or two-step chemistry, and detailed chemistry. These approaches require choosing between computational efficiency or physical accuracy. In detailed chemistry simulations, there are physical constraints that must be met when transporting species mass fractions; nonlinear transport schemes such as WENO do not satisfy these constraints automatically. A new method is presented to ensure that the sum of mass fractions equals 1, without penalizing inert species. The approach is better able to capture the physical instability expected for detonations. To reduce the cost of chemistry while maintaining accurate physics, tabulated chemistry has been used extensively for flames/deflagrations in the low Mach number framework. In the simplest tabulated chemistry model for premixed flames, a progress variable, describing the progress of reactions in the system, is transported in the simulation. This progress variable is then used to look up all other species, transport properties, and thermodynamic variables from a pre-computed table. Unfortunately, there is no existing tabulation approach designed specifically for detonations. As such, this work extends the tabulated chemistry method to detonations. To describe the enthalpy and specific heat capacity, the temperature is selected as a second table coordinate. The two table coordinates are able to capture virtually all variations in the progress variable source term. The Zel'dovich-von Neumann-Döring (ZND) model is found to be the most appropriate one-dimensional problem for generation of the table. The ZND tabulation approach is validated for both one-dimensional stable and pulsating and two-dimensional regular and irregular detonations in various hydrogen-oxygen mixtures. The tabulated chemistry simulations are able to reproduce the detailed chemistry results in terms of propagation speed, cellular structures, and source term statistics at a reduced computational cost, demonstrating the benefits of this approach for predictive modeling of detonations.</p
Physics and Applications of Compact Optical Frequency Comb
Get PDFOptical frequency combs (OFC) have been vastly developing and were awarded half of the Nobel Prize in 2005. OFCs are series of optical signals with distinct and equally spaced frequencies. One reason why OFCs are essential for modern optics and photonics engineering is that OFCs serve as a bridge between optical frequencies (hundreds of THz) and frequencies within the electronic bandwidth (from MHz to GHz, which is the distance between adjacent comb teeth). In this thesis, I first introduce some physical principles of optical resonators, which are critical components for confining optical energy and generating OFCs. Then, in the main body of this thesis, I study the physics and applications of two types of compact OFCs: soliton microcombs and electro-optical frequency combs.
Microcombs are OFCs generated on integrated photonics devices. Here, I first develop a methodology to experimentally characterize two important physical properties (material absorption loss and optical nonlinearity) of integrated photonic materials. Next, I focus on a novel method to generate mode-locked soliton microcombs on ultra-low-loss Si3N4 material. It was considered challenging to support bright solitons due to its normal dispersion. This novel method involves two resonators that are partially coupled together, which can modify the dispersion through mode hybridization and feature symmetry breaking. Following this, I investigate two characteristics closely related to the symmetry breaking of this coupled-ring device: the observation of Kelly sidebands and multicolor bright soliton generation. Finally, I demonstrate bright soliton generation in Al0.2Ga0.8As resonators, which feature high nonlinearity but were considered difficult to support bright solitons at room temperature due to its high material loss. Here, we mitigate the effect of material loss by pulse-pumping operation.
Electro-optical frequency combs are OFCs generated by modulating a continuous wave laser using an external radio-frequency source. Taking advantage of low-noise radio frequency and stable continuous-wave laser frequency, this OFC can serve as a frequency reference for astronomical observation. In this thesis, I first introduce the physics and operating principle of electro-optical frequency combs in Chapter 1, then discuss developing and deploying the near-infrared laser frequency comb at the W.M. Keck Observatory in Chapter 7.
In summary, the thesis discusses the physics and applications of mode-locked bright soliton microcombs, which can generate radio frequencies by taking the beat note of this OFC. I also discuss the physics and applications of electro-optical frequency combs, which are stable OFCs used for astronomical frequency references generated by radio-frequency modulation of continuous wave lasers. The critical role of OFCs as a bridge between optical frequencies and frequencies within the electronic bandwidth (MHz to GHz) is demonstrated, and their potential to revolutionize various fields, including high-precision metrology, telecommunications, and astrophysics, is highlighted.</p
Classical and Quantum Simulation of Chemical and Physical Systems
Get PDFVarious quantum mechanics effects have been found and widely studied in different microscopic systems, such as quantum nuclear effects and electron correlation in molecular systems, electron-phonon coupling in crystal systems, and quantum Zeno effects in open quantum systems. However, exact numerical simulations require exponentially scaled classical resources. In this thesis, we study these quantum systems by a series of classical or quantum methods, which include semiclassical, ab initio, machine learning, and quantum computing approaches.
In Chapter 2, we develop the molecular-orbital-based machine learning (MOB-ML) method as a general-purpose method to learn molecular electronic structure properties. By preserving physical constraints, including invariance conditions and size consistency, MOB-ML is shown to be able to capture both weak and strong interactions. Furthermore, the Gaussian Process framework is extended for learning both scalar properties such as energies, and linear-response properties like dipole moments with the rotationally equivariant derivative kernel. With these improvements, MOB-ML shows not only significantly higher learning rates for organic molecules, non-covalent interactions, and transition states but also excellent transferability from small systems to large systems.
In Chapter 3, we develop a generalized class of integrators in the thermostatted ring-polymer molecular dynamics (T-RPMD) method, which is a semi-classical quantum dynamics method to capture various types of molecular nuclear quantum effects, including zero point energy, quantum tunneling, and kinetic isotopic effects. Such generalized integrators are carefully designed to be strong stable and dimension-free, which are essential for robust numerical computations. In particular, a so-called "BCOCB" integrator is proved to be superior in terms of accuracy and efficiency in the harmonic limit. Such superiority is further verified in strongly anharmonic systems featured by liquid water.
In Chapter 4, we develop an ab initio-based semi-analytical model of electron-phonon scattering to describe the transport and noise behavior in GaAs, which is a widely-used semiconductor. Such a semi-analytical model lifts a few approximations in the standard ab initio calculation of intervalley scatterings, which were believed to be the origin of the failure to capture the nonmonotonic noise phenomena. We find qualitatively unchanged transport and noise properties and agreements on the scattering rates between the photoluminescence experiments. These results indicate the most probable origin of the nonmonotonic noise behavior is the formation of space-charge domains rather than the intervalley scattering.
In Chapter 5, we simulate the challenging measurement-induced phase transitions (MIPT) behavior in quantum many-body systems on a superconducting quantum processor. Due to the intrinsic exponential scaling of the quantum state tomography and post-selection process, traditional simulations of MIPT were limited to a few qubits. With the recently introduced linear cross-entropy benchmarking, such exponential overhead is eliminated, and the correct critical behavior of MIPT is observed on a 22-qubit system. Our work paves the way for the studies of open quantum systems on large-scale near-term quantum devices.</p
Essays on Sequential Sampling in Value-Based Choice
Get PDFThis dissertation comprises three chapters related to the fields of psychology, computational neuroscience, and experimental economics. Chapters 1 and 2 use experimental and computational methods to study the role of attention in simple, value-based choices. Chapter 3 examines risky choices from experience and tests some of the underlying assumptions of sequential sampling models.
A growing body of research has shown that simple choices involve the construction and comparison of values at the time of decision. These processes are modulated by attention in a way that leaves decision makers susceptible to attentional biases. In Chapter 1, co-authored with Stephanie Dolbier and Antonio Rangel, we studied the role of peripheral visual information on the choice process and on attentional choice biases. We used an eye-tracking experiment in which participants (N = 50 adults) made binary choices between food items that were displayed in marked screen ``shelves'' in two conditions: (a) where both items were displayed, and (b) where items were displayed only when participants fixated within their shelves. We found that removing the nonfixated option approximately doubled the size of the attentional biases. The results show that peripheral visual information is crucial in facilitating good decisions and suggest that individuals might be influenceable by settings in which only one item is shown at a time, such as e-commerce.
In Chapter 2, co-authored with Stephen Gonzalez and Antonio Rangel, we studied the role of attention in aversive risky choices where all outcomes were unpleasant. We used two eye-tracking experiments in which participants made binary choices between two lotteries in two conditions: (a) a gain condition where outcomes for lotteries were weakly positive, and (b) a loss condition where outcomes were weakly negative. Contrary to the predictions of the standard aDDM, we found that attentional choice biases in the loss condition were identical to those found in the gain condition, suggesting that attention nudges choices towards the attended option even in losses. To explain these results, we propose a variation of the Attentional Drift-Diffusion-Model (called the Hybrid aDDM) that incorporates (a) both a value-dependent and a value-independent effect of attention on the choice process and (b) reference-dependent value signals. We show that the observed attentional choice biases and other behavioral signatures in the loss condition can only be explained by the Hybrid aDDM with a reference-point rule that sets the reference-point at or below the minimum possible outcome in a given context.
In Chapter 3, co-authored with Antonio Rangel, we establish that sequential sampling models apply to risky decisions from experience and test some of the underlying assumptions of these models. We ran an online study in which participants chose to Play or Skip a slot machine, based on a stream of samples drawn from its outcome distribution. We found evidence for leakage, collapsing decision boundaries, and a delay in sample integration. We also found evidence of non-linear sample weighting depending on when the sample occurred during the trial. As a bonus, we established a link between the fixed decision boundaries in a Drift-Diffusion-Model and a Modified Probit model, allowing for estimation of decision boundaries in cumulative sample space without the need to fit a computational model.</p
A Glitch and the Matrix: Advances in Gravitational-Wave Glitch Mitigation and Acceleration of Pulsar Timing Analyses
No full textSince the first detection of gravitational-waves in 2015, the field of gravitational-wave astronomy has developed rapidly. Today, there are more than 300 transient gravitational-wave event candidates from stellar-mass sources and we have found evidence for a stochastic background of supermassive black-holes. In this thesis I present work addressing two significant challenges on analyzing these data. The first: mitigating transient, non-Gaussian noise in gravitational-wave detectors, or ``glitches'', that can bias our estimates of physical properties of compact objects. The second: introducing a faster method to analyze pulsar-timing data containing a stochastic background of supermassive black-hole sources. Gravitational-wave astronomy is a data-rich field, and is only becoming more so with upgraded detectors, additional detectors, and longer observing time; we need robust, fast, and unbiased techniques to analyze that data
Strain Sensing in Thin Composite Laminates with Embedded Fiber Bragg Grating Sensors
Get PDFDeployable structures are popular for space applications as they enable large, complex spacecraft structures to overcome the size constraints of launch vehicle fairings. Such structures are increasingly manufactured out of thin (< 200 μm thick) composite laminates as they have a high stiffness-to-weight ratio, the ability to withstand high curvatures during stowage, and the potential for self-deployment using stored strain energy. To ensure the reliability of these thin composite spacecraft structures in operation, it is of interest to be able to continuously monitor their internal strain state to detect potential changes or damage that may compromise their integrity.
Although there are a number of potential sensors that could be used for this, fiber Bragg grating (FBG) sensors are especially well suited for this task and have a track record of successfully monitoring both composite materials and large aerospace structures. However standard size FBG sensors, which have a cladding diameter of 125 μm, are too large to be integrated into the thin composite structures of interest. To overcome this, we worked with several suppliers to develop and manufacture ultra-thin FBG sensors (< 30 μm cladding diameter) for this work that are able to be successfully embedded into thin composite laminates.
The primary objective of this thesis was to investigate the suitability of ultra-thin FBG sensors for the monitoring of strain changes in thin composite spacecraft structures. To this end, the work in this thesis first investigated how to best embed ultra-thin FBG sensors to be able to measure the internal strain changes of interest while minimizing their disruptions to the surrounding laminates. Second, mechanical testing was performed to assess the effect that the embedded ultra-thin FBG sensors have on the mechanical properties of thin laminates. Third, the ability of these sensors to detect and monitor for strain changes in thin composite laminates was assessed through further mechanical testing. Finally, the effects of temperature on ultra-thin FBG sensors were studied experimentally.
Through this work, which was done at the coupon level, we sought to demonstrate the ability of these ultra-thin FBG sensors to monitor for strain changes in thin composite laminates and their potential for the health monitoring of thin composite spacecraft structures. It is our hope that our findings in this thesis help lay the groundwork for the future implementation of these sensors in not only thin composite spacecraft structures, but to many other composite materials and aerospace structures as well.</p
Essays in Matching Theory
Get PDFThis dissertation consists of three essays on matching theory. The first two essays examine provide new cooperative solutions for two problems arising within matching markets in practice. The third contributes a theoretical analysis of the causes and effects of a market failure within the medical residency match.
Chapter 1 analyzes a matching market in which some agents have made prior commitments to each other. Typically, matching market models ignore prior commitments. I analyze two-sided matching markets with pre-existing binding agreements between market participants. In this model, a pair of participants bound to each other by a pre-existing agreement must agree to any action they take. To analyze their behavior, I propose a new solution concept, the agreeable core, consisting of the matches which cannot be renegotiated without violating the binding agreements. My main contribution is an algorithm that constructs such a match by a novel combination of the Deferred Acceptance and Top Trading Cycles algorithms. The algorithm is robust to various manipulations and has applications to numerous markets including the resident-to-hospital match, college admissions, school choice, and labor markets.
In Chapter 2, I turn to the problem of increasing the efficiency of student assignments in school choice subject to constraints imposed by policymakers. In school choice, policymakers consolidate a district’s objectives for a school into a priority ordering over students. They then face a trade-off between respecting these priorities and assigning students to more-preferred schools. However, because priorities are the amalgamation of multiple policy goals, some may be more flexible than others. This paper introduces a model that distinguishes between two types of priority: a between-group priority that ranks groups of students and must be respected, and a within-group priority for efficiently allocating seats within each group. The solution I introduce, the unified core, integrates both types. I provide a two-stage algorithm, the DA-TTC, that implements the unified core and generalizes both the Deferred Acceptance and Top Trading Cycles algorithms. This approach provides a method for improving efficiency in school choice while honoring policymakers’ objectives.
Chapter 3 introduces a a behavioral model of early matching within the context of the National Resident Matching Program, the system by which graduating medical students are matched to hospital residency programs. In my model, two hospitals compete to match to a continuum of doctors. Each hospital can make early offers or wait until the match is produced through the matching program. Some doctors have a behavioral preference to match early while others do not. I show that the less-desirable hospital benefits from the option to make early offers. My results provide a theoretical foundation for behavior widely documented within the medical ethics and graduate medical education literature and confirm beliefs commonly held by residency program directors.</p