80264 research outputs found
Sort by
Investigating the Role of Transfer Learning in Enhancing CNN-Based Subgrid-Scale Models for Geophysical Turbulence
No full textTransfer learning (TL) is a powerful tool for enhancing the performance of neural networks (NNs) in applications such as weather and climate prediction and turbulence modeling. TL enables models to generalize to out-of-distribution data with minimal data input. In this study, we employed a 9-layer convolutional NN to predict subgrid forcing in quasi-geostrophic systems and examined which metrics best describe its performance and generalizability. Fourier analysis of the NNs' kernels reveals that they learn low-pass, band-pass, and high-pass filters, regardless of their training dataset's isotropic or anisotropic nature. By analyzing the activation spectra, we also identified the reasons behind NNs' failure to generalize and how TL can overcome these limitations. The main reason is that learned weights and biases on one dataset underestimate the out-of-distribution sample spectra as they pass through NN, leading to an underestimation of output spectra. By only re-training one layer with new data from the target system, this underestimation is fixed and results in NN producing predictions matching target system dataset spectra. These findings are broadly applicable to data-driven parameterization of high-dimensional dynamical systems
Center for Teaching Excellence Spring 2025 Newsletter
No full textInside this issue: Symposium -- University Awards Ceremony -- Reading Group -- Teaching Research & Partnerships -- AI Updates -- Featured Resources -- Graduate Workshops -- Faculty Fellow
Angle-Resolved Photoemission Spectroscopy Study on Quantum Materials
No full textUnderstanding and engineering electronic states in quantum materials have long been an important research aspect that is essential for both fundamental condensed matter physics and modern applications. In this work, we use angle-resolved photoemission spectroscopy to study the electronic band structure in three different materials, covering topics in ferroaxial orders, van der Waals ferromagnetic semiconductors, and topological insulators. Each one provides a unique perspective on identifying exotic electronic states in solids and how to manipulate them.
In the rare earth tritellurides, additional to the well-known charge density wave (CDW) ordering in the low temperature regime, we discover new spectroscopic evidence for an electronic ferroaxial order, which spontaneously breaks all reflection symmetries. Combined with the density functional theory calculations, we find that in the CDW state, the degeneracy between the Te in-plane p orbitals is lifted. We conclude that the combination of the p orbital asymmetry and the uniaxial CDW order breaks all vertical mirrors, leading to a composite ferroaxial order. Through temperature dependence measurement, we observe the mirror symmetry breaking disappears above the CDW ordering temperature, which connects the electronic ferroaxial order to the CDW order. Our study provides a new orbital-based perspective to electronic ferroaxial systems.
In a magnetic semiconductor, Cr2Ge2Te6, we realize the tuning of a Van Hove singularity in the electronic structure by in-situ surface alkali metal dosing. Subsequently, we observe a shadow intensity in the spectra, accompanied by a renormalization effect of the conduction band, originating from a doping-dependent electron-plasmon coupling. Besides, as a function of the doping level, there is a bandwidth evolution from the Cr d orbitals that could be related to the magnetism in this material. In this study, we demonstrate that tuning the charge carrier density can be an effective way to modify electronic correlations and magnetic couplings in low-dimensional semiconductors.
In a prototypical topological insulator, Bi2Se3, by active hydrogen etching, we demonstrate that the surface Se atoms are gradually removed, resulting in the formation of Bi-Bi2Se3 heterostructures. This causes a charge transfer between different layers and directly modifies the band structure of the system, represented by a shifting of the topological surface state in both real and momentum spaces. We present a precisely controlled methodology of surface etching, allowing the fine tuning of the robust topological surface state, which significantly advances the nanoengineering of topological insulators
Familial Multigenerational Housing: A Resurgent Typology
No full textMajority of housing models in the United States do not address the financial crisis and care responsibilities that burden many contemporary families. While many cultures practice multigenerational living as a means to ease some of these burdens, the homes in which this lifestyle is practiced are not designed for this purpose. As the population of Houston continues to grow, its urban landscape will need to adapt and ultimately densify as it cannot continue to expand its city limits indefinitely. Through the development of middle scale familial multigenerational housing complexes, residents of the city can transition into denser housing models while also easing some of their current financial and care responsibilities
Living with Trash
No full text“Living with Trash” explores the possibility of a new waste treatment system that integrates trash infrastructure into Seoul’s residential and commercial districts, strengthening the urban fabric as an interconnected ecosystem. This new typology aims to enhance the city's resilience to waste crises by decentralizing waste management into a more intricate network, restoring shared responsibility between the city and its residents
6.1 Establishing the Global Network for Organisms and Multi-disciplinary Exchange (GNOME)
No full textThis entreaty was created as part of The Spirit of Asilomar and the Future of Biotechnology summit (February 23-26, 2025) in Pacific Grove, CA.This entreaty focuses on the necessary elements to broaden and diversify engagement beyond the Spirit of Asilomar conference community through the creation of a Global Network for Organisms and Multi-disciplinary Exchange (GNOME). The network would build a broader coalition of expertise, experience, perspectives and cultural knowledge to decide on how to use biotechnology to address the biodiversity crisis, and foster collaborative decision-making on the ethical, legal, cultural, and social implications of emerging biotechnologies, especially engaging South-South and South-North stakeholders in connections. GNOME has three key goals - the first is to build and strengthen trust — the most important takeaway from the Spirit of Asilomar meeting — between communities, biotechnologists, conservationists, ecologists, ethicists, and other stakeholders, particularly with the Global South. The second goal is to foster the exchange of technical, cultural values, and local knowledge and perspectives to inform/evolve existing decision-making and operational frameworks for the use of biotech in conservation and the environment. Our final goal is to build responsible and accountable biotech innovation reaffirming local community values, priority needs, the right to withhold consent, self-determination, and the rights of nature. We invite you to join our efforts to design the network, its structure, governance and activities
Advancing Polycyclic Aromatic Hydrocarbon Bioremediation using Genetic Bioaugmentation in Soil Microbial Communities
No full textPolycyclic aromatic hydrocarbons (PAHs) are introduced into the environment through forest fires, fossil fuel combustion, and crude oil spills, posing significant health and ecological risks. These compounds are carcinogenic and disrupt soil microbial processes essential for ecosystem functions. Bioremediation, which uses microorganisms to degrade pollutants, can be applied to ameliorate contaminated environments. However, biodegradative functions are often limited because exogenous bacteria cannot compete with native microbes.
Genetic bioaugmentation offers a promising solution by equipping native microbes with biodegradative capabilities encoded and delivered via mobile genetic elements such as plasmids. This approach leverages the adaptability and ecology of native microbial communities. I hypothesize that delivering catabolic genes on plasmids to native microbes enhances PAH removal by engaging a diverse, well-adapted bacterial community rather than relying on a single species. Previous research on genetic bioaugmentation has inadequately addressed the fitness impacts of plasmids on recipient bacteria, the range of plasmid recipients, and their effect on biodegradation rates.
This thesis investigates these factors by engineering plasmids with the bphC dioxygenase gene and conjugating them to soil bacteria. Results revealed that plasmid fitness effects significantly influenced conjugation rates, community structure, and PAH biotransformation. Moreover, plasmid transfer rates were strongly associated with recipient bacterial abundance in synthetic communities. To track plasmid persistence, the pKJK5 plasmid was modified with a genetic memory biosensor. This plasmid persisted in soil microbial communities for 10 days without selective pressure and showed enhanced stability and biodegradation efficiency in the presence of a model PAH.
These findings highlight the critical role of plasmid fitness effects in shaping microbial community dynamics and biodegradation efficiency. By addressing the activity and longevity of biodegradative functions at the community level, this research advances the design of effective genetic bioaugmentation strategies for PAH-contaminated environments
Identifying hosts of bacteriophage transduction in microbial communities with RNA-addressable modification
Get PDFBacteriophages (phages) are the most abundant life form on earth. They facilitate important ecological interactions in all microbial ecosystems by killing their host organism or contributing to the rapid evolution of bacteria via horizontal gene transfer (HGT). Because of these traits, phages are increasingly being applied to manipulate microbial communities by killing specific hosts or delivering functional genes. Central to understanding a phage’s role in a microbial ecosystem, as well as the efficacy of their application, is knowledge of which bacterial species in a microbial community a phage can infect. Either through killing or HGT, a phage’s host is the component of microbial ecosystems that phages act on to impart their changes.
The objective of this thesis is to develop and deploy a novel technology to characterize phage hosts in high throughput and in natural microbial communities. Prior techniques to connect phage to host, such as the plaque assay, are low-throughput, difficult to apply in environmental contexts, and/or require a culturable host organism. RNA-addressable modification (RAM) is a ribozyme that can be encoded into mobile genetic elements (MGE) to track their hosts. Upon infection of a host by a RAM encoded MGE, RAM will splice an RNA barcode onto the host’s 16S rRNA. Sequencing of barcoded 16S rRNA can then reveal host information for a given MGE. RAM has previously been demonstrated in plasmids delivered to a community via conjugation.
For the first time we demonstrate deployment of RAM in phages and start by outlining a blueprint for incorporating RAM into various phages by adding it to the genome of the well-characterized phage P1 and P1’s associated phagemids. Addition of the P1 constructs to a synthetic microbial community of human pathogens verified RAMs ability to identify phage hosts in a community context and identified a novel host of P1 phagemids in Salmonella enterica. Further exposure of the P1 constructs to a wastewater community confirmed known hosts of P1 and captured a completely new order of P1 hosts in Aeromonadales. We finally showed that RAM can investigate host-specific phage factors and captured significant differences in host range imparted by P1’s two unique tail fibers. This thesis outlines an approach for implementing RAM in phages and lays the groundwork for the application of RAM as a foundational tool for studying phage-host ecology
Effect of composite elements on the mechanics of biopolymer networks
Get PDFNetworks of stiff biopolymers, such as collagen type I, are abundant in the human body and provide structural integrity to the skin and internal organs. These networks are known to exhibit mechanical properties that differ significantly from those of synthetic polymers, making them a subject of great interest in research. While pure collagen networks have been extensively studied, collagen in its native state in the
extracellular matrix is often found in conjunction with external elements such as polysaccharides, lipids, biological cells, and other fibrous elements and proteins.
In this thesis, we investigate the effects of these added components on biopolymer networks using coarse-grained simulations of fiber networks. First, we analyze the impact of the incompressible properties of hyaluronic acid, a polysaccharide embedded in collagen networks by studying how it contributes to a synergistic effect in stiffness of such tissues. This study demonstrates the interplay between diverse mechanical elements and focuses on their rheology. Next, we examine the mechanics of collagen
networks embedded with magnetic nanoparticles, focusing on the role of the external magnetic field and the influence of volume fraction of these magnetic particles. Our findings indicate that the combined effects of magnetic fields and external forces resemble stabilizing factors such as bending rigidity and applied temperature and explore how the collective behavior of magnetic particles play an important role in the rheology of such systems. Finally, we explore the effects of thermal fluctuations on three dimensional collagen networks. These fluctuations can stabilize a floppy isostatic network and exhibit non-mean field critical behavior in the mechanical response of networks. While two dimensional networks have been widely studied, real-world biological systems are inherently three-dimensional. Through 3D packing derived networks, we confirm these theoretical predictions for 3D collagen networks