80264 research outputs found
Sort by
all things flow
Get PDFall things flow is a compilation of texts exploring artistic practices centered in the interplay of fragmentation, circulation, and ephemerality. Active from the 1940s through the 1990s, the artist Ray Johnson described his work as miniatures cataloguing free-form fragmentations. His practice of mail art, the making of collages, relief surfaces, and correspondence objects demonstrates how fragments can be constantly rearranged and reassembled. Engaging with Daoist philosophy, his work reflects the fluid nature of social relationships.
Through experimental autoethnographical writings about Johnson, this thesis critiques the limitations of institutional categorization and the ways modern and contemporary art are defined, created, sold, and understood through International Art English. Johnson’s work resists classification within completed/enclosed systems. By engaging with his methods of creating counterparts and paradoxes, this project invites readers to move away from normative modes of speculation, and explore alternative interpretations of existing categories in art historical and archival practices
First-Principles Study of Electron Transport Properties of Low-Dimensional Materials
No full textElectron transport in low-dimensional materials is crucial for the development of
advanced electronic devices and innovative applications in nanotechnology. This dissertation
presents a first-principles calculation study of electron transport phenomena
in three di↵erent nanoscale systems, each contributing to the frontier of nanoelectronics.
DNA-inspired programmable self-assembly has been a paradigm for achieving ever
more complex graphene-based nanostructures for novel electronic devices and applications.
However, it is still unknown how hydrogen-bonded junctions inherent in
such devices will perform as electron transport media. In the first part, we examine
the electron transport properties of graphene nanoribbons (GNRs) connected by DNA nucleobases, utilizing first-principles Density Functional Theory (DFT) with the Non-Equilibrium Green’s Function (NEGF) method. Pronounced rectifying behavior and negative differential resistance are found, as well as high conductance in certain structures. The identified sensitivity of the current response to atomic details of the interfaces offers initial hints and guidance for experimental realization.
The current response is found to be tunable by electrostatic doping, indicating the potential application of the junction as a nanoscale switch.
The second part focuses on electron transport across lateral interfaces between graphene and borophene. By employing DFT+NEGF methods, we analyze the electron transmission and contact resistance in different borophene-graphene lateral junctions.
We also present a rigorous method for extracting the contact resistance between bulk materials from quantum transport in the nanoscale. Borophene exhibits low contact resistance with graphene compared to other metals, due to the similarity in conducting orbitals. This indicate that borophene could be a good conductor in graphene based devices.
In the third part, we discuss the atomistic processes responsible for the non-volatile resistance switching in a monolayer-thin memristor, also known as an “atomristor”. We demonstrate how the write, read, and retention performances of the atomristor
depend on control parameters such as applied bias, choice of electrode materials, and device geometry. We map out the underlying energy landscape that governs the translocation kinetics of a single atom responsible for the atomristor operation. This allows us to prescribe a suitable metal-electrode material and provides a clear picture of the underlying processes in memristors at the 2D limit.
Overall, this dissertation provides valuable insights into electron transport phenomena in low-dimensional materials, paving the way for their integration into nextgeneration electronic devices
Templating Metal Clusters via Resorcinarene- and Cyclen- Based Ligand Platforms
No full textPolynuclear metal sites, found in both metalloenzymes and heterogeneous catalysts, play a critical role in small molecule activation by enabling multi-electron redox transformations through cooperative reactivity. One of the most compelling examples is biological nitrogen fixation, mediated by the nitrogenase enzyme, which offers a blueprint for sustainable alternatives to the energy-intensive Haber-Bosch process. Additionally, the Nitrous Oxide reductase presents a Cuz/Cuz* site that facilitates the reduction of Nitrous Oxide to environmentally benign forms. At the heart of metalloenzyme function are complex polynuclear clusters capable of storing and delivering multiple electrons while supporting diverse substrate binding modes across its metal centers. To bridge the knowledge gap between structure and reactivity as well as what intermediates ensue after substrate activation, this thesis employs polynucleating ligand scaffolds to template the synthesis of a series of copper and iron-based clusters and investigates their (multi)electron reactivity. In Chapter I introduce the role multimetallic centers in metalloenymes and heterogenous catalysis play in substrate activation and conversion. I also cover the current method of developing templating ligands to drive the synthesis of clusters and their reactivity. In Chapter 2, I discuss the synthesis of a tetraamine scaffold RL(NH2)4, where L is a rigidified resorcin[4]arene ligand and R = Methyl, Phenyl and Isobutyl groups. We found that treatment of this ligand scaffold with a Cu4Mes4(py)2 leads to the formation of C4
ii
symmetric RL(NH)4Cu4 clusters with Cu−Cu distances indicative of metal-metal interactions. The one-electron oxidized [Cu4]+ species was probed by variable temperature X-band continuous wave-electron paramagnetic resonance (CW-EPR), which displays a multiline spectrum at room temperature as well as DFT calculations. In Chapter 3, I demonstrate that, by ligand modification, we can tune the cavity size of these clusters. While the local coordination environment and electronic structure at the [Cu4] core remains largely the same, the imparted cavity size engenders reversible host guest interactions with acetonitrile (MeCN) and molecular oxygen (O2), with O2 binding corroborated by variable-temperature NMR, resonance Raman spectroscopy, and DFT calculations. In Chapter 4, I introduce a new 1,4,7,10-tetraazacyclododecane (cyclen)-derived octaamine ligand LH4. To demonstrate the versatility of this ligand platform, we isolate a family of divalent first-row dinuclear transition metal complexes—LCr2, LMn2, LFe2, LCo2, and LZn2. Crystallographic and DFT analysis reveals varying degree of metal-metal interactions from weak to strong quadrupole bonding. Lastly, in Chapter 5, LH4 is used as a starting point in the synthesis of hexanuclear iron cluster, LFe6Mes4, Reaction with organic and inorganic azides yields a terminal μ4-nitride cluster, LFe6Mes4(μ4-N), via a net three-electron oxidation to a mixed-valent [FeII3FeIII3] core. Comparative analysis reveals a significant structural reorganization in the [Fe4] unit upon nitride incorporation. Cyclic voltammetry demonstrates that LFe6Mes4 is capable of traversing at least eight distinct oxidation levels across a 2.5 V window—approaching the electron stoichiometry of enzymatic nitrogen fixation. These findings highlight the potential of this ligand scaffold to stabilize redox-active, high-nuclearity iron clusters relevant to small molecule activation and multi-electron redox transformations
Solving Hybrid Boolean SAT by Continuous Optimization
No full textThe Boolean SATisfiability problem (SAT) is of central importance in computer science. Despite the NP-completeness of SAT, progress on the engineering side—especially that of Conflict-Driven Clause Learning (CDCL) and Local Search SAT solvers—has been remarkable. Yet, while SAT solvers, aimed at solving industrial-scale benchmarks in Conjunctive Normal Form (CNF), have become quite mature, other types of non-CNF, i.e., hybrid constraints (e.g., cardinality/pseudo-Boolean constraints and XOR}s play important roles in critical applications such as cryptography, task scheduling, and quantum computing. Nevertheless, SAT solvers that are effective on hybrid constraints are less well studied; a general approach for handling hybrid constraints is still lacking.
The thesis focuses on the applications and approaches for hybrid SAT solving. It starts with a motivating application of hybrid SAT solving in quantum computing. A smart hybrid SAT encoding is proposed to solve the problem with quantum interest. Then the thesis introduces efforts for addressing critical limitations of a continuous-optimization-based framework for hybrid SAT solving, named FourierSAT. In FourierSAT, Boolean constraints are converted into polynomials via Walsh-Hadamard-Fourier Transform and constrained continuous optimizers are then applied to search for a solution. FourierSAT suffers from slow gradient computation, incapability of handling pseudo-Boolean (PB) constraints, and the restriction to constrained optimizers, which together limit its potential. The thesis presents to use binary decision diagrams (BDDs) to replace polynomials, which significantly accelerates the gradient computation both in theory and practice, meanwhile handling PB constraints. The thesis then proposes an approach for removing the requirement of supporting constraints from the optimizers while still maintaining soundness, making FourierSAT compatible to several machine-learning-inspired optimizers. Empirical results demonstrate that the improved FourierSAT achieves promising performance on certain classes of applications and qualifies to be a useful complement to the SAT solver portfolio. This thesis further opens a solid research direction by bridging analysis of Boolean functions, continuous optimization and SAT solving. FourierSAT will continue benefiting from future advancements in multiple fields
GHz Gravitational Wave and THz Dark Matter Detectors: Analysis and Simulation
No full textGravitational waves have been directly detected at sub-kHz frequencies. Detecting high-frequency gravitational waves (HFGWs) encounters challenges due to smaller interferometers and weaker astrophysical sources. A promising homodyne detector utilizes the inverse Gertsenshtein effect, where HFGWs convert to photons in a magnetic field. A physical analysis highlights intrinsic noise and other complications.
Dark matter (DM) axions can similarly be detected through the inverse Primakoff effect. Although not directly detected yet, axions are anticipated to produce stronger coherent signals and would additionally solve the strong charge-parity problem in QCD. Current detection mechanisms struggle at THz frequencies due to complex mechanical tuning and a lack of low-loss THz materials. The proposed THz-Radiometer for AXions (T-RAx) leverages recent advancements in semiconductor quantum structures, with electromagnetically tunable resonance frequencies, to target the THz (meV) axion. COMSOL simulations show significant enhancements of the axion-induced photons, optimistically reaching the QCD axion parameter space
Characterizing the Extracellular Matrix of Migratory Neural Crest Cells in the Enteric Nervous System Development of Zebrafish
No full textThe enteric nervous system (ENS) is a neuronal and glial cellular network that regulates intestinal function and is derived from a multipotent and highly migratory population of cells known as neural crest cells (NCCs). NCCs give rise to a mature ENS through exposure to various intrinsic and extrinsic cues as they colonize the gut. Our understanding of the role of the extracellular matrix (ECM) in ENS formation is limited, with patterns in ECM gene expression and protein remodeling during development remaining unknown. An in silico analysis of ECM genes in zebrafish, an optimal model organism for ENS development, showed upregulation of fibronectins, laminins, tenascins, integrins, collagens, and syndecans during ENS development. In situ hybridization and immunofluorescence coupled with confocal microscopy show temporal dynamics in ECM gene expression and protein organization patterns. Interestingly, CRISPR-Cas9 mutagenesis of tenascin-C shows aberrations in the number of cells colonizing the developing ENS
Curving Beam Reflections: Model and Experimental Validation
No full textCurving beams are a promising new method for bypassing obstacles in future millimeter-wave to sub-Terahertz (sub-THz) networks but lack a general predictive model for their reflections from arbitrary surfaces. We show that, unfortunately, attempting to "mirror" the incident beam trajectory across the normal of the reflector, as in ray optics, fails in general. Thus, we introduce the first geometric framework capable of modeling the reflections of arbitrary convex sub-THz curving beams from general reflectors with experimental verification. Rather than "mirroring" the trajectory, we decompose the beam into a family of tangents and demonstrate that this process is equivalent to the Legendre transform. This approach allows us to accurately account for reflectors of any shape, size, and position while preserving the underlying physics of wave propagation. Our model is validated through finite element method simulations and over-the-air experiments using terahertz time-domain spectroscopy, demonstrating millimeter-scale accuracy in predicting reflections. Our model provides a foundation for creating reflected curved-beam paths with applications to both communications and sensing
Metrics and Methods for Fair and Interpretable Machine Learning
No full textUnderstanding the impact of features in machine learning models is critical for both interpretability and fairness. As these models are increasingly deployed in high-stakes domains, it becomes essential to develop tools that can explain model behavior and detect sources of unfairness. Interpretability helps build trust, diagnose errors, and support accountability, while fairness ensures that predictions do not systematically disadvantage certain groups. However, existing methods often address these goals in isolation, making it difficult to understand whether a model’s behavior is both accurate and equitable. In this thesis, we introduce novel methods that enhance transparency, mitigate bias, and improve the interpretability of model explanations.
We begin by developing FairTreeFIS, the first fairness-aware feature importance score for decision trees, ensembles, and global surrogate models. FairTreeFIS captures each feature’s contribution to unfairness in model predictions by extending impurity-based scores to incorporate group fairness metrics. To extend this beyond tree-based models, we next propose model-agnostic metrics based on feature occlusion and permutation, which quantify how individual features affect fairness across a wide range of model classes. Motivated by the need for fairness-aware methods not only in interpretation but also in learning, we then introduce Fair MP-Boost, a novel stochastic boosting algorithm that jointly optimizes fairness and accuracy through adaptive sampling over both features and observations. By integrating FairTreeFIS and traditional tree-based importance scores into the training process, Fair MP-Boost yields interpretable models while maintaining competitive performance and reducing unfairness.
While the previous methods focus primarily on fairness-aware interpretation and model training, interpretability more broadly encompasses understanding complex model behavior, including how features interact. Many real-world prediction tasks rely on such interactions, which may or may not be directly tied to fairness concerns but are nonetheless crucial for understanding and trusting model outputs. To address this, we introduce iLOCO, a distribution-free, model-agnostic method for quantifying feature interactions, which enables a deeper understanding of how features jointly influence predictions without making assumptions about the data distribution. We establish distribution-free inference procedures for interaction effects, allowing us to compute valid confidence intervals for detected interactions. This contribution broadens the interpretability toolkit to include methods for understanding and validating the role of feature interactions in model predictions.
Finally, to address the limitations of existing benchmarks for interpretability, specifically those focused on feature interaction, which are often too small, lack structure, or require substantial domain-specific knowledge, we introduce a benchmark dataset constructed from cocktail recipes and expert and community-generated ratings. This dataset offers a structured and interpretable setting for evaluating how both individual features and their interactions contribute to model predictions. By introducing new metrics for interpretability and fairness-aware interpretation, as well as empirical tools for model training, feature interaction analysis, and benchmarking, this thesis contributes practical methods for responsible artificial intelligence and lays the groundwork for more transparent and equitable decision-making systems
Collective quantum phenomena in strongly correlated multi-level systems
No full textIn this dissertation, I theoretically investigate the equilibrium properties of three strongly correlated multi-level quantum platforms -- synthetic dimension using ultracold Rydberg atoms, multi-flavor Fermi-Hubbard simulators with ultracold atoms, and magnon-spin coupled rare-earth orthoferrites -- using various numerical and analytical methods.
After the broad research motivation in the first chapter, Chapter 2
introduces the general principles of quantum simulation using synthetic dimension platforms. Chapter 3 discusses the first realization with Rydberg atom platform to simulate the topological Su-Schrieffer-Heeger (SSH) model, and how environmental noise affects it.
Chapter 4 shows the mean-field phase diagram of dipole-coupled ultracold Rydberg atom arrays exhibiting quantum strings and other ordered states.
The connection of the thermal phase transition to that of a classical extended Potts model is also discussed from a renormalization (RG) and generalized duality perspective.
These examples illustrate the versatility of synthetic dimension in simulating and engineering quantum phenomena.
Chapters 5-6 explore the properties of multi-flavor Fermi-Hubbard models on square optical lattices using numerical linked-cluster expansions (NLCE). First, I show universal thermodynamics of SU(N) Fermi-Hubbard models above superexchange temperatures. Then, I discuss the agreement between numerical simulated equations of state and correlation functions to optical lattice experiments using 173Yb [SU(N)-symmetric interactions] and three spin-flavors of 6Li [spin-dependent interactions] at intermediate temperatures. Thermometry performed with this method agrees well with the model-independent thermometry. Lastly, I propose a method of computing observables at low temperatures in bipartite ordered states by adding a suitable boundary field to the NLCE.
NLCE agreeing to quantum simulations at temperatures above the superexchange temperatures and a possibility to make accurate predictions at low temperatures for certain states make it a powerful algorithm to be used side-by-side with quantum simulations.
Chapter 7 is based on the observation of magnonic superradiance in ErFeO3.
The spin Hamiltonian reveals a simultaneous kink and mode-softening of two of the lower polaritons, strongly suggesting a superradiant phase transition.
Its mapping to an extended Dicke model is also discussed.
This is the first evidence towards a macroscopically squeezed spin state.
Finally, I conclude with a broad discussion and an outlook in Chapter 8