A memristor is a device with the property that its electrical resistance depends on the history of applied voltage, and changes with the amount of current that passes through it. This atypical property of history-dependent resistance is analogous to how memory functions. As a result, memristors have been used in applications ranging from non-volatile analog memory to synaptic analogs and neuromorphic computing. However, there is a lack of in-depth understanding of the atomic-scale behavior that causes this property. My project aims to understand how the dominant dynamics, electromigration and ion transport, result in the unique behavior of the resistance. This understanding is a necessary precursor to solving the two main challenges in the current field of memristors: large parameter variability and cycling endurance degradation. I will study the relationship between dynamical defect structures and 2D device behavior by using Scanning Tunneling Microscopy (STM) to directly image ion dynamics and measure […]
As processing power available to high performance computers has increased, the overall performance has bottlenecked due to the limits of data transfer and storage. For small and medium sized computational fluid dynamics (CFD) researchers, the size of generated data is becoming problematic. In particular, CFD data faces a large resistance to general compression algorithms such as Lempel-Ziv encoding found in .zip files. Applications in CFD including wind-turbine design and aeronautics demand accurate capture of fluid dynamics phenomena on complex geometries such as turbine blades or airplane wings. For these complex geometries, grid-like straight-edged meshes prove ineffective. Instead meshes which are both curved (referred to as “high-order”) and non-gridlike (referred to as “unstructured”) are gaining popularity. As such, research into a novel compression algorithm of CFD data on high-order unstructured meshes will help small scale CFD researchers efficiently store their data.
Understanding literary narratives requires a strong grasp of the underlying language; unlike text drawn from Wikipedia, social media, or the news, literary prose is ripe with embellished descriptions and figurative language. Plot points are revealed in the form of nuanced events rather than explicit recollections: we read about a character’s ‘final breath’ rather than their death. The success of attention-based neural networks on an assortment of semantic inference and interpretation tasks suggests that they provide an effective model of human language, but a noticeable drop in performance when applied to literary data has inspired a call for domain-specific systems. This work draws on equal parts natural language processing research and literary theory to develop such ‘born-literary’ NLP models, designed specifically for the task of narrative understanding. If successful in predicting plot outcomes, the representations learned by these models may capture some of the latent narrative structures that encode storytelling devices […]
The human body experiences various mechanical loads which cells respond to correspondingly to maintain homeostasis. However, drastic stimuli changes might lead to various diseases and degeneration. To enable regenerative medical strategies, we must first understand the connection between cell and organ mechanics. With this motivation, my mentor is developing a deformable microfluidic chip that can apply these mechanical loads to a single layer of cells. Currently, the chip is tailored for loads in the annulus fibrosus of the spine, but other systems with similar loading could be considered. The current design is limited to testing one tissue sample and one stretching condition per cycle. We will create a new design that improves the chip’s ability to produce data by testing multiple stretching conditions and samples simultaneously. Specifically, I will employ finite element modeling and create a genetic algorithm to manipulate and optimize the chip design to match physiologically relevant, experimental […]
Topological insulators, electronic materials that behave like insulators in their interior but can conduct current on their edges or surfaces, hold a lot of potential for improving the function and energy efficiency of various devices. These materials have unique electrical properties resulting from their conducting edge and surface states. Specifically, these states can transport charge with very low resistivity, leading to efficient conduction. Topological insulator materials are heavily influenced by the symmetry of their crystal structure─the pattern in which their atoms are arranged. For this reason, my project will investigate the effects of structural disorder─which breaks spatial symmetries─on the properties of a relatively new topological insulator with the chemical formula MnBi2Te4 (Manganese, Bismuth, and Tellurium) through computational quantum mechanical modeling methods such as Density Functional Theory. Namely, I will look at the effect of slightly distorting the crystal structure on the magnetism, electronic structure, and other properties of the material.
In our present world that faces an unprecedented amount of language contact, code-switching has become an increasingly common sociolinguistic phenomenon. In particular, code-switching between tonal and atonal languages poses interesting questions that are yet to be studied in tone research. In Mandarin Chinese, the negator “bu” is documented to change from a falling tone to a rising tone before another falling tone (Chao, 1965). Considering Mandarin syntax and the structure of Mandarin syllables, I propose the hypothesis that “bu” has a rising tone before monosyllabic English words and a falling tone before polysyllabic English words. A pilot study (2020) featuring four Mandarin-English bilinguals found results that didn’t perfectly match, but nonetheless support, the hypothesis. The goal of this project is to build on the pilot study, refining the experiment design, collecting additional data, and performing more rigorous data analysis. The results of this study could shed light on how tonal […]
From infancy, people are told stories. The patterns, morals, and relationships in these stories help children form schemas to interpret the world. While cultures have prominent stories that are passed down to most children (e.g. Romeo and Juliet), people spontaneously invent narratives for purposes of entertainment, distraction, or teaching. My research aims to understand how androcentric bias (the bias that centers the experiences of men over the consideration of other genders) functions at an interpersonal, individual level during storytelling; more specifically, how a character’s agency or communality in a story influences the likelihood that someone will tell a child a story with a male protagonist. These conditions are chosen because agency, an essential function of being human, has become specifically associated with masculinity, while community is associated with femininity. With an awareness of how androcentric bias functions in storytelling, people can consciously shift from androcentric descriptions to more gender-neutral language […]
Being able to understand and predict the behavior of bovine intervertebral discs under different mechanical stresses is important because bovine discs are often used as a stand-in for human discs in disc biomechanics studies. Current computational methods of modeling stress and strain mechanics in the disc are very inconsistent because they are unable to reliably and accurately depict fiber mechanics, due to the fact that they treat a non-homogenous, fibrous part of the disc (the annulus fibrosis) as a homogenous structure. This newly developed computational model is better able to predict the behavior of the disc as it more accurately models the annulus fibrosis as a structure made up of separate fiber bundles. With this improved model, the scientific community will be able to more accurately predict the mechanical stability of a disc with simulated repair or degeneration. However, this model has yet to be experimentally validated. My project is […]
Quantum computing is a new paradigm that could revolutionize how we process information. For these applications, it is critical that the systems can maintain quantum coherence for long periods of time. In this project, we will investigate new acoustic designs to extend quantum coherence in superconducting quantum qubits and nanomechanical resonators. At low temperatures where quantum devices operate, the dominant energy loss in acoustic and superconducting resonators occurs due to coupling to two-level-system defects. The goal of our research will be to engineer a platform that can be used to study these defects. Because proper shielding is critical to keep a quantum system isolated from its environment, a challenge will be to design a shielding system with a bandgap that blocks acoustic waves from escaping into the environment at our frequencies of interest. We will use an acoustic metamaterial-based design, which will be a periodic arrangement of unit cells engineered […]
When mosquitoes bite a host, they extract blood as a source of nutrition for their eggs. The DNA contained within mosquito blood meals can be extracted and identified using a method called DNA metabarcoding. Millions of DNA sequences of known organisms are contained within bioinformatic databases and extracted DNA sequences can be matched up to these known sequences. Using DNA metabarcoding of mosquito blood meals, I will map out mosquito predation from 5 locations across the state of California to test the hypothesis that mosquito predation will vary across locations, due to the different ecosystem types in these different regions.