62588 research outputs found
Sort by
High-Performance Grain Growth Simulations and their Machine Learning Applications
No full textDue to the difficulty in observing grain growth in real samples, grain growth simulations have become commonplace for analyzing and predicting grain behavior in manufacturing processes. In this work, we describe the theoretical basis for these simulations and their mechanisms, as well as how they are commonly implemented. We provide a novel transformed boundary propagation mechanism as well as novel simulation optimizations, such as using regional octrees for tracking subsystem activity, tracking grain boundaries sparsely in real-time, and simulating grain growth via parallel subregion simulation. We also provide information on how to properly store and visualize simulation outputs, as well as how to use those outputs for predictive machine learning tasks.</p
Predicting Kinetics of Flow-Induced Polymer Reactions in Varying Flow Conditions and Geometries
No full textNumerous biological processes are regulated by flow-induced polymer (i.e., multimer) reaction. One such process is blood clotting, jumpstarted by the activation of the mammalian glycoprotein von Willebrand Factor (vWF), which itself is driven by the surrounding environmental conditions. In this dissertation, via coarse-grained Brownian dynamics (BD) simulation, we advance the understanding of mechanistic behavior and reaction kinetics of vWF multimer populations while subjected to varying flow conditions and under different flow geometries. We cover background information about the blood protein vWF, as well as details about simulation methodology used for this research.In relatively low strain rate extensional flow conditions, unraveling kinetics of collapsed multimers is explored by computing the unraveling probability versus strain rate over varying exposure time to flow. Statistical modeling is performed on the distribution of time of exposure to flow prior to multimer unraveling. Using data for strain rates at which sufficient unraveling events were observed to construct statistical models, allowed for evaluation of unraveling kinetics for flow conditions in which unraveling events were not observed in BD simulations. Results indicate a highly non-linear influence of strain rate on the energy profile associated with the unraveling transition and this is related to the long length polymeric protrusions required to drive unraveling in such flow conditions. Under pulsatile shear flow conditions and in tethered geometries, collapsed multimers are experimentally observed to follow a nonintuitive behavior of decreasing in length in the unraveled conformations while there is an increase in peak shear rate. BD simulations were used to conduct a short study and try to reveal the mechanism of the unexpected experimental observations. Deposition of multimers onto a tethering surface prior to the flow experiments does not eliminate the possibility of polymers binding to the surface at multiple spots along the multimer chain. We then revealed that so-called multi-point tethering strongly affects the unraveled length due to physical constraint imposed by multiple binding sites. However, in simulations, multi-point tethering does not meaningfully reduce unraveled length with increasing shear rates. A number of other possible mechanisms were explored in simulations, but none of them reproduced the experimental observation. We thus conclude by discussing an unexplored mechanism that could be explored in future work as a possible cause of such behavior.The next study transitions to that of continuous shear flow conditions and in tethered geometries. At fixed ̇γ, force F along a polymer increases linearly with N as previously predicted; however, contrary to existing theory, the F(N) slope increases for N above a transition length that exhibits minimal dependence on ̇γ. Force profiles are used in a stochastic model of a force-mediated reaction to compute the time for x percent of a polymer population to experience a reaction, tx. Observations are insensitive to the selected value of x in that tx data for varying N and ̇γ can be consistently collapsed onto a single curve viaappropriate scaling, with one master curve for systems below the transition N (small N) and another for those above (large N). Different force scaling for small and large N results in orders of magnitude difference in force-mediated reaction kinetics as represented by the population response time. Data presented illustrate the possibility of designing mechano-reactive polymer populations with highly controlled response to flow across a range in ̇γ.Overall, models presented here advance our team\u27s ability to make predictions about reactivity of functional polymers in varying flow conditions. The parameters used in simulations here are built upon experimental data for vWF, and they are used to further the understanding of vWF behavior in realistic human blood flow conditions. But a goal of research presented herein is to consider the breadth of impact beyond the blood protein. On a wider perspective, the findings of this dissertation are generalized to consider generic multimer kinetics and behavior. They can thus can be used to inspire studies of other multimers, including artificially created ones for specific applications.</p
Quantum Computing and Optimization Methods
No full textQuantum computing has recently emerged as a groundbreaking paradigm capable of solving certain mathematical problems more efficiently than classical computing. Optimization problems, known for their computational intensity and broad applicability, have been a particular focus of efforts to develop efficient quantum algorithms for large-scale problems. This thesis explores the application of quantum linear system algorithms (QLSAs) in quantum interior point methods (QIPMs) to speed up the solution of conic optimization problems.QLSAs have the potential to expedite QIPMs; however, their efficiency is currently limited when applied to the ill-conditioned linear systems that typically arise within QIPMs. Furthermore, the requisite extraction of a classical description of the quantum solution via quantum tomography algorithms introduces significant error and noise into the calculations. This research addresses these challenges by developing robust QIPMs that incorporate classical techniques such as iterative refinement methods (IRMs) and preconditioning to improve both accuracy and efficiency.
The thesis initially analyzes an infeasible-inexact QIPM, where errors in linear system solutions and resulting infeasibility are shown to lead to suboptimal computational complexity. To enhance QIPM complexity, we propose two types of inexact feasible QIPMs by reformulating Newton systems to manage errors while maintaining feasible iterates. We also explore the application of IRMs to improve the precision of both quantum tomography and QIPMs.
Moreover, to tackle issues related to the condition number of the Newton systems, the thesis employs IRMs alongside preconditioning techniques specifically tailored for QIPMs. The computational complexity of these enhanced methods is rigorously analyzed, with a focus on their application to linear, semidefinite, and second-order conic optimization problems. This study also discusses the implementation of these approaches in machine learning contexts, such as solving least squares and support vector machine problems. Finally, the thesis presents numerical experiments conducted using QISKIT to simulate the proposed methods on quantum hardware, demonstrating their efficacy and potential for broader application.</p
Low Temperature Sintering Silver Conductive Ink Doped with Indium
No full textConductive inks are functional designed materials that allow for the full potential of Printed Flexible Electronics (PFE) technologies. The various applications of conductive ink printed structures are of significant interest due to their thermally and electrically conductive properties, light weight, flexibility, cost-effectiveness, and high production speed capabilities. Both the conductive ink materials formulation and manufacturing of printed electronics are growing fields accompanied by increasing demands for robustness, reliability, and functional integration with flexible printed electronic devices. The major drawback which hinders the full potential of conductive inks use is the requirement for a consistently high level of performance. This highlights the need to formulate metal conductive materials with properties that provide optimal functionality during extended service time and under challenging environmental and mechanical conditions. In addition, low-temperature processing profiles are desired because of their potential time production energy savings and the ability to print on more diverse, lightweight and flexible types of substrates.The following work is dedicated to the formulation development of an Ag nanoparticle (Ag NP) conductive ink that demonstrated excellent performance with the addition of indium nanoparticles (In NP). Indium (In) is a well-known material with a relatively low melting point that has been extensively studied for its use in the development of electronics. In this research work, the use of In NP in the experimental conductive ink aided the Ag nanoparticles sintering at a temperature of approximately 130°C which is lower than the 175-200°C sintering temperature range required for Ag NP alone. Based on the differences in the melting points and metal densities of Ag NP and In NP, the addition of In NP reduced the stress of shrinkage during high-temperature high-humidity (HTHH) storage conditions of 85°C/85% RH which resulted in consistently excellent adhesion and electrical conductivity performance. In addition, the use of a high flexibility thermoset polymer system in the formulation development of the conductive ink resulted in outstanding flexibility and adhesion on a polyimide (Kapton) substrate. The developed conductive ink has a low viscosity of 7-9 mPa.s which allows it to be dispensed using both aerosol jet printing and inkjet printing methods. Furthermore, printed lines made with the ink exhibited an electrical volume resistivity of 7.04x10-6 Ohms.cm even after the HTHH storage test for 1,000 hours. Based on the high metal load and metal NP size, the conductive ink provides high coverage for both single- and multi-layer printing tasks while resulting in minimal shrinkage and good adhesion to other common flexible substrates including glass and PC. The use of a bismaleimide polymer resin system enhanced the conductive ink flexibility and high temperature performance. This allowed electronic components to be printed on flexible materials which made it possible to manufacture durable devices. Overall, conductive inks are expensive materials. However, printing technology market studies have demonstrated a greater cost efficiency when compared to traditional manufacturing methods as they have the freedom of automated design and the ability to reduce material waste and energy consumption.</p
Preparing Pre-Service and In-Service Special Educators to Write High-Quality Individualized Education Programs through Online Professional Development
No full textIndividualized Education Programs (IEPs) are the legal documents guiding the education of students with disabilities served under IDEA. Despite the importance of these documents, noncompliance and lack of quality remain in the present levels of academic and functional performance, annual goals and objectives, and supplementary aids and services sections. Using a one-group pre-post design, this study examined the effectiveness of online professional development for 29 pre-service and in-service special educators to increase IEP quality in these three sections. Further, this study investigated how online professional development impacts special educators\u27 self-efficacy. Participants received a freely available online module on high-quality IEPs with one online synchronous group coaching session on an IEP area of need. Findings revealed an increase in scores from pre- to post-test following the IEP-focused professional development. Implications and future directions are discussed. </p
Beyond Gender Stereotypes: A Feminist Analysis of Sarah J. Maas\u27 Throne of Glass
No full textThis thesis conducts a feminist analysis of Sarah J. Maas\u27 Throne of Glass series, establishing it as a feminist narrative that offers a nuanced representation of women within the fantasy genre. An exploration of Aelin\u27s character reveals how the series dismantles patriarchal stereotypes. Using the work of feminist scholars, it becomes evident that Maas has crafted a female protagonist who disrupts negative portrayals of female characters. Divided into four subsections—\u27Maas and Female Archetypes in Literature,\u27 \u27Aelin as Cinderella: A Retelling,\u27 \u27Transforming Classic Romantic Tropes,\u27 and \u27Female Competition vs. Friendships\u27—this examination demonstrates that Throne of Glass not only challenges existing biases but also provides an empowering representation of women in literature
Integration of AI-Powered Speech Analysis Tools in Speech Therapy
No full textIntegrating AI-powered tools in diagnostics and treatment of Speech Sound Disorders, improving accessibility and reducing Speech-Language Pathologist workload.I utilized AI-powered speech analysis tools into speech therapy to enhance the diagnosis of Speech Sound Disorders (SSD). Given the limited access to Speech-Language Pathologists (SLPs) and the large number of people with the need for Speech Therapy in the United States, there is a large caseload for SLP\u27s in America. My work aims to bridge this gap by utilizing AI technologies such as automatic speech recognition and large language models. These tools can streamline diagnostics, create individualized therapy materials, and reduce SLP workloads, thereby addressing burnout and improving accessibility to care. Focusing on preschool-aged children due to its relevance in SSD disorder diagnosis, my approach involves reviewing existing diagnostic tools, collaborating with AI and programming experts, developing a coding framework, and planning for validation and replication of the AI system.</p
Build Your Own Sci-Fi Writing Bot - Project Summary
No full textAs a means to "demystify" generative models, I will develop a course module that begins with basic (first generation) language models to create simple autocomplete processes, and then move on to more complex implementations, such as RNNs, LSTMs, and transformers. This module will use R and Python, as well as open-source tools like Ollama.While GUIs for generative models and integration into existing apps is proliferating, many
implementations obfuscate the underlying mechanics of generative technologies. As a result, students generally do not understand how these models work, and are not able to anticipate when and how they are likely to fail. By getting hands-on experience with each advancement in language modeling and demonstrating how such advancements solved problems revealed by previous models, students will get a better understanding of how contemporary generative models operate. Students will also train these various models on a corpus of English science fiction, and thus build bots that write science fiction
A Machine Learning Approach to Discovering Exoplanets Transiting Old Stars - Project Summary
No full textThis random forest algorithm categorizes old stars as exoplanet hosts based on transits found in Transiting Exoplanet Survey Satellite photometry.Most stars have planets orbiting them, and some cross in front of the host star causing a temporary decrease in brightness known as a transit. These transits have measurable depth, period and duration. Only ~6% of known transiting planets orbit older host stars, and the rest orbit stars with ages similar to our sun1. As a result of stellar evolution, transits of old stars are sometimes missed by transit searches of young stars. The goal of this large-scale search is to quickly discover enough exoplanet candidates orbiting old stars to assist population level analysis. A manual search for exoplanet transits creates an unreasonable timeline. We run Box Least Squares (BLS) on 90,000 old stars using photometry from the Transiting Exoplanet Survey Satellite (TESS). BLS is a period search algorithm that fits parameters such as orbital period, depth, duration and signal to noise of planet transits whether or not there is a real planet. TESS is a near all-sky space-based photometric survey with a primary mission of discovering transiting exoplanets. Using BLS from a set of injected planet transits and a set of inverted lightcurves, we are able to train a random forest algorithm to classify old stars into hosts or non-hosts. This allows us to reduce vetting time greatly. We identified 17 previously known exoplanet candidates in our set, which shows that our methods are successful. We then discovered 32 new exoplanet hosts, which we are following up with ground based observations currently