Physical Sciences and Mathematics Commons^™

Evaluating Large Language Model Performance On Haskell, Andrew Chen

Undergraduate Honors Theses

I introduce HaskellEval, a Haskell evaluation benchmark for Large Language Models. HaskellEval’s curation leverages a novel synthetic generation framework, streamlining the process of dataset curation by minimizing manual intervention. The core of this research is an extensive analysis of the trustworthiness of synthetic generations, ensuring accuracy, realism, and diversity. Additional, I provide a comprehensive evaluation of existing open-source models on HaskellEval.

Security And Interpretability In Large Language Models, Lydia Danas

Undergraduate Honors Theses

Large Language Models (LLMs) have the capability to model long-term dependencies in sequences of tokens, and are consequently often utilized to generate text through language modeling. These capabilities are increasingly being used for code generation tasks; however, LLM-powered code generation tools such as GitHub's Copilot have been generating insecure code and thus pose a cybersecurity risk. To generate secure code we must first understand why LLMs are generating insecure code. This non-trivial task can be realized through interpretability methods, which investigate the hidden state of a neural network to explain model outputs. A new interpretability method is rationales, which obtains …

Modeling The Neutral Densities Of Sparc Using A Python Version Of Kn1d, Gwendolyn R. Galleher

Undergraduate Honors Theses

Currently, neutral recycling is a crucial contributor to fueling the plasma within tokamaks. However, Commonwealth Fusion System’s SPARC Tokamak is expected to be more opaque to neutrals. Thus, we anticipate that the role of neutral recycling in fueling will decrease. Since SPARC is predicted to have a groundbreaking fusion power gain ratio of Q ≈ 10, we must have a concrete understanding of the opacity
and whether or not alternative fueling practices must be included. To develop said understanding, we produced neutral density profiles via KN1DPy, a 1D kinetic neutral transport code for atomic and molecular hydrogen in an ionizing …

Identifying Transitions In Plasma With Topological Data Analysis Of Noisy Turbulence, Julius Kiewel

Undergraduate Honors Theses

Cross-field transport and heat loss in a magnetically confined plasma is determined by turbulence driven by perpendicular (to the magnetic field) pressure gradients. The heat losses from turbulence can make it difficult to maintain the energy density required to reach and maintain the conditions necessary for fusion. Self-organization of turbulence into intermediate scale so-called zonal flows can reduce the radial heat losses, however identifying when the transition occurs and any precursors to the transition is still a challenge. Topological Data Analysis (TDA) is a mathematical method which is used to extract topological features from point cloud and digital data to …

Mathematical Modeling And Examination Into Existing And Emerging Parkinson’S Disease Treatments: Levodopa And Ketamine, Gabrielle Riddlemoser

Undergraduate Honors Theses

Parkinson’s disease (PD) is the second most common neurodegenerative disease across the world, affecting over 6 million people worldwide. This disorder is characterized by the progressive loss of dopaminergic neurons within the substantia nigra pars compacta (SNpc) due to the aggregation of α-synuclein within the brain. Patients with PD develop motor symptoms such as tremors, bradykinesia, and postural instability, as well as a host of non-motor symptoms such as behavioral changes, sleep difficulties, and fatigue. The reduction of dopamine within the brain is the primary cause of these symptoms. The main form of treatment for PD is levodopa, a precursor …

Use Of Molecular Logic Gates For The Tuning Of Chemosensor Dynamic Range, Orhan Acikgoz

Undergraduate Honors Theses

The first molecular logic gates were created in the 1990s; integrating such logic gates into fluorescent chemosensors allowed for the detection of different types of ions in solution. In this study, we have developed a new use of molecular logic gates by having two of the same type of binding site. The two binding sites on a fluorophore that both detect Na⁺ ions led to an increase in the detection limit compared with the chemosensor with a single binding site. Since the two sodium binding sites create an AND logic gate, two sodium ions are needed to generate a …

Modeling Group 3 Medulloblastoma: Describing The Interconnected Pathway Of The Most Common Pediatric Brain Cancer, Amber Cantú

Undergraduate Honors Theses

Group 3 medulloblastoma is one of the most common pediatric brain cancers. Affecting infants and children, this cancer has the worst prognosis of the medulloblastoma group. Current treatments use surgical resection, radiation, and chemotherapy to afflict the cancer, however no cure has been found. This project aims to model one of the many pathways being investigated in Group 3 medulloblastoma which may be used to synthesize future treatments. Specifically, showing the interconnections between various precursors of BCL-xL, an antiapoptotic protein, and how these factors influence the progression of the disease. Scientific databases were used to find previous research articles which …

Harnessing The Power Of Virtual Reality For Organic Chemistry Education, Jungmin Shin

Undergraduate Honors Theses

Understanding organic chemistry concepts heavily relies on visualization of the geometry of molecules and spatial arrangement of molecules during mechanisms. 2D textbook depictions have their limitations in visualizing the three-dimensionality of organic chemistry. Student learning outcomes could be greatly improved from 3D visualizations of these topics. This project explores the potential of an emerging technology, Virtual Reality (VR), being incorporated as a teaching resource for organic chemistry.

This paper discusses two trials for evaluating the potential of VR as a teaching resource for organic chemistry in select topics of the Diels-Alder reaction and R/S configurations and stereoisomers. The Diels-Alder reaction …

Roads And Corresponding Travel Time To Markets: Assessing Climate Vulnerability In Nepal, Kaitlyn Crowley

Undergraduate Honors Theses

Roads exist as a physical and theoretical connection between people and places around the globe. In addition to providing a route from one point to another, roads are also an indicator of access to markets and of poverty. However, current road datasets, particularly the Global Roads Open Access Data Set, are out of date or incomplete, necessitating new sources of data for analyses involving road networks. This study explores the relationship between climate change and access to markets in Nepal. We seek to identify isolated communities that are likely to experience detrimental outcomes associated with environmental threats, such as increasing …

Process Modeling The Neuroprotective Effects Of A Plant-Based Diet On Parkinson's Disease, Julia Mitchell

Undergraduate Honors Theses

Parkinson's disease (PD) is a progressive neurodegenerative disorder characterized by motor symptoms such as tremors, bradykinesia, rigidity, and postural instability. Recent research suggests an avenue for potential neuroprotection through dietary intervention, specifically the adoption of a plant-based diet. A plant-based diet predominantly comprises foods derived from plants, emphasizing fruits, vegetables, grains, legumes, and nuts while minimizing or excluding animal products. This thesis aims to explore the biochemical pathways implicated in PD progression and the potential impact of dietary choices on these pathways. The investigation focuses on several key pathways: alpha-synuclein aggregation, the blood-brain barrier crossing of levodopa, oxidative stress, ferroptosis, …

Improving The Scalability Of Neural Network Surface Code Decoders, Kevin Wu

Undergraduate Honors Theses

Quantum computers have recently gained significant recognition due to their ability to solve problems intractable to classical computers. However, due to difficulties in building actual quantum computers, they have large error rates. Thus, advancements in quantum error correction are urgently needed to improve both their reliability and scalability. Here, we first present a type of topological quantum error correction code called the surface code, and we discuss recent developments and challenges of creating neural network decoders for surface codes. In particular, the amount of training data needed to reach the performance of algorithmic decoders grows exponentially with the size of …

Code Syntax Understanding In Large Language Models, Cole Granger

Undergraduate Honors Theses

In recent years, tasks for automated software engineering have been achieved using Large Language Models trained on source code, such as Seq2Seq, LSTM, GPT, T5, BART and BERT. The inherent textual nature of source code allows it to be represented as a sequence of sub-words (or tokens), drawing parallels to prior work in NLP. Although these models have shown promising results according to established metrics (e.g., BLEU, CODEBLEU), there remains a deeper question about the extent of syntax knowledge they truly grasp when trained and fine-tuned for specific tasks.

To address this question, this thesis introduces a taxonomy of syntax …

Dimensionlessly Comparing Hydrogen And Helium Plasmas At Lapd, Lela Creamer

Undergraduate Honors Theses

This project compares the hydrogen and helium gas puff plasmas created at the Large Plasma Device (LAPD) using dimensionless numbers to determine the extent to which the turbulence pattern can be explained by plasma physics. Since turbu- lence tends to dissipate energy and particles in a plasma, it can cause problems for fusion reactors by reducing their efficiency. With a better understanding of turbu- lence’s causes and behavior, some of this energy loss could potentially be avoided. In recent experiments at LAPD, an unexpectedly high amount of turbulence was de- tected when helium was used to create the plasma, which …

Ecological Monitoring Program At Vims Esl: Annual Report 2023, Paige G. Ross, Richard A. Snyder

Reports

An Ecological Monitoring Program (EMP) has been established at the Virginia Institute of Marine Science Eastern Shore Laboratory (VIMS ESL) for the coastal environment near the Wachapreague lab. The goals of the initiative are to 1) provide status and trends information to scientists who study and regulators who manage Virginia’s marine resources, 2) provide a scientific context for short-term research and grant proposals 3) provide pedagogical enrichment for educators to use in their classes, and 4) build capacity in staff expertise and training of interns and students at VIMS ESL.

The program formalizes and standardizes data collection for a long-term …

Ecology And Conservation Of Diamondback Terrapins In Virginia, Cypress Ambrose

Undergraduate Honors Theses

The diamondback terrapin (Malaclemys terrapin) is the only turtle species native to North America with specific morphological and physiological adaptations to estuarine environments. Along with many other pressures contributing to population declines, terrapins frequently become trapped and drown as bycatch in crab pots used in the commercial and recreational blue crab (Callinectes sapidus) fishery. A wealth of evidence supports the use of inexpensive bycatch reduction devices (BRDs) that can be attached to the entrances of these traps, which leads to a marked decrease in terrapin bycatch while not reducing crab catch dramatically. Virginia is the only …

Transcriptional Dynamics During Rhodococcus Erythropolis Infection With Phage Wc1, Dana Willner, Sudip Paudel, Andrew D. Halleran, Grace E. Solini, Veronica Gray, Margaret Saha

Arts & Sciences Articles

Background

Belonging to the Actinobacteria phylum, members of the Rhodococcus genus thrive in soil, water, and even intracellularly. While most species are non-pathogenic, several cause respiratory disease in animals and, more rarely, in humans. Over 100 phages that infect Rhodococcus species have been isolated but despite their importance for Rhodococcus ecology and biotechnology applications, little is known regarding the molecular genetic interactions between phage and host during infection. To address this need, we report RNA-Seq analysis of a novel Rhodococcus erythopolis phage, WC1, analyzing both the phage and host transcriptome at various stages throughout the infection process.

Results

By five …

Artificial Intelligence For The Electron Ion Collider (Ai4eic), C. Allaire, ..., Cristiano Fanelli, James Giroux, Joey Niestroy, Justin R. Stevens, Patrick Stone, L. Suarez, K. Suresh, Eric Walter, Et Al.

Arts & Sciences Articles

The Electron-Ion Collider (EIC), a state-of-the-art facility for studying the strong force, is expected to begin commissioning its first experiments in 2028. This is an opportune time for artificial intelligence (AI) to be included from the start at this facility and in all phases that lead up to the experiments. The second annual workshop organized by the AI4EIC working group, which recently took place, centered on exploring all current and prospective application areas of AI for the EIC. This workshop is not only beneficial for the EIC, but also provides valuable insights for the newly established ePIC collaboration at EIC. …

Monitoring The Abundance Of American Shad And River Herring In Virginia's Rivers: 2023 Annual Report, Eric J. Hilton, Patrick E. Mcgrath, Ashleigh Magee, Timothy Hoyt

Reports

This report describes the results of the twenty-sixth year of a continuing study to estimate the relative abundance and assess the status of American shad (Alosa sapidissima) stocks in Virginia by monitoring the spawning runs in the James, York and Rappahannock rivers in spring 2023, evaluating hatchery programs, and contributing to coast-wide assessments (ASMFC 2007, ASMFC 2020). We also report on two fisheryindependent monitoring programs using anchor gillnets in the Rappahannock River (year 6) and a major tributary of the James River, the Chickahominy River (year 9), to determine relative abundance and stock structure for the adult spawning run of …

Eluquant: Event-Level Uncertainty Quantification In Deep Inelastic Scattering, Cristiano Fanelli, James Giroux

Arts & Sciences Articles

We introduce a physics-informed Bayesian neural network with flow-approximated posteriors using multiplicative normalizing flows for detailed uncertainty quantification (UQ) at the physics event-level. Our method is capable of identifying both heteroskedastic aleatoric and epistemic uncertainties, providing granular physical insights. Applied to deep inelastic scattering (DIS) events, our model effectively extracts the kinematic variables x, Q2, and y, matching the performance of recent deep learning regression techniques but with the critical enhancement of event-level UQ. This detailed description of the underlying uncertainty proves invaluable for decision-making, especially in tasks like event filtering. It also allows for the reduction of true inaccuracies …

Mid-21st Century Chesapeake Bay Hypoxia: The Role Of Climate Change At The Atmospheric, Terrestrial, And Oceanic Boundaries, Colin Hawes

Dissertations, Theses, and Masters Projects

Climate change is already increasing the volume and duration of coastal hypoxia and threatening living resources, particularly in eutrophic ecosystems. Atmospheric warming exacerbates deoxygenation by decreasing gas solubility and enhancing respiration and remineralization. Changes in terrestrial runoff and sea level influence hypoxia via nutrient availability and altered water temperature, respectively. However, the impacts of other future climate changes, including winds, shortwave and longwave radiation, non-runoff precipitation, and ocean water conditions, are still unknown. In this study, the impacts of such future climate changes on hypoxia in Chesapeake Bay were examined using a 3-D coupled estuarine hydrodynamic–biogeochemical model linked to a …

Policy Recommendations For Tire Additive 6ppd And Its Derivative 6ppd-Q, Ashley E. King

Dissertations, Theses, and Masters Projects

Around 3.1 billion tires are produced around the world annually1. The antioxidant additive, 6PPD (i.e., N-(1,3-dimethylbutyl)-N’-phenyl-p-phenylenediamine) is widely employed in passenger and commercial vehicle tires at 0.4-2% by mass to impede tire degradation2. Antioxidants are intended to migrate to tire surfaces and form protective films to prevent rubber oxidation. 6PPD is designed to react with oxidant species like ozone, intentionally forming chemical transformation products that can then escape from the tire and into the environment. 6PPD-Q (i.e., 2-anilino-5-[(4-methylpentan-2-yl)amino]cyclohexa-2,5-diene-1,4-dione) is one such transformation product.

After release and disbursement in the environment, 6PPD-Q is bioavailable to aquatic animals and mammals and acute …

Projected Impacts Of Climate Change And Watershed Management On Carbonate Chemistry And Oyster Growth In A Coastal Plain Estuary, Catherine Czajka

Dissertations, Theses, and Masters Projects

Coastal acidification, warming, and nutrient management actions all alter water quality conditionsthat marine species experience, with potential impacts to their physiological processes. Decreases in calcite saturation state (ΩCa) and food availability, combined with warming water temperatures, pose a threat to calcifying organisms; however, the magnitude of future changes in estuarine systems is challenging to predict and not well known. This study aims to determine how and where oysters will be affected by future acidification, warming, and nutrient reductions, and the relative effects of these stressors. To address these goals, an oyster bioenergetics model for Eastern oysters (Crassostrea virginica) was embedded …

Learning-Based Sensing For Solving Health-Related Problems, Minglong Sun

Dissertations, Theses, and Masters Projects

In the field of ubiquitous computing, health-related problem analysis has gained increasing attention. Collaborations between domain doctors and computing researchers have been established to recognize and address health-related issues. However, accurate detection and recognition of health-related problems remain a major challenge that attracts extensive research efforts. Among all the research works, wearable sensors-based health-related problem recognition is promising as it is low cost, low power, and easy to carry. This dissertation focuses on utilizing wearable sensors to study health-related problems. The first project introduced in this dissertation is TremorSense, a PD tremor detection system designed to classify Parkinson's Disease hand …

Artificial Illumination Of Trawl Gear Components To Reduce Pacific Halibut (Hippoglossus Stenolepis) Bycatch In The U.S. West Coast Groundfish Bottom Trawl Fishery, Derek Jackson

Dissertations, Theses, and Masters Projects

Pacific halibut (Hippoglossus stenolepis) is a prohibited species for the U.S. West Coast Bottom Trawl Fishery and in the last decade, there has been a concentrated interest in the use of artificial illumination serving as a potential bycatch reduction device. Previous studies conducted off the coast of Oregon have found that the addition of green light-emitting diodes to the bridles of low-rise, cutback trawls greatly reduced the number of Pacific halibut caught. However, recent regulation changes now permit high-rise trawls, a gear configuration that fishes a very different volume of water than the previously permissible gear profile, in areas where …

Experimental Studies Of Neutral Particles And The Isotope Effect In The Edge Of Tokamak Plasmas, Ryan Chaban

Dissertations, Theses, and Masters Projects

The H-mode plasma edge is a region of steep gradients in density and temperature known as the “pedestal” which greatly increases energy confinement. The complex links between neutral-plasma interactions and both diffusive and convective transport in the pedestal must be understood to model, predict, and achieve the high performance required for a fusion power plant. This dissertation explores the effects of different hydrogenic isotope neutral particles and plasma transport from the edge pedestal region into the Scrape-Off Layer. Current experiments typically use deuterium (H with amu=2 or D), however future fusion power plants may startup with hydrogen (H), and eventually …

Low-Rank Matrix And Tensor Models For Data Science Applications, Jeremy Moulton Myers

Dissertations, Theses, and Masters Projects

Low-rank approximations play an important role in data science analysis and applications. When the model is linear and the data is represented as a matrix, the optimal rank-r approximation is given by the r dominant factors of the singular value decomposition (SVD). When the model is multilinear and the data is represented as a multi-way array called a tensor, the optimal rank-r approximation may not even exist. Nonetheless, the canonical polyadic decomposition (CP) provides a useful low-rank tensor approximation for interpretability and analysis in a way similar to the matrix SVD but across multiple axes simultaneously. In this dissertation, we …

Environmental Controls On Carbon And Nitrogen Cycling In Alaskan Arctic Coastal Lagoons, Brian Seung Taek Kim

Dissertations, Theses, and Masters Projects

At the intersection of land, sea, and atmosphere, coastal lagoons act as bioreactors, processing both terrestrial and autochthonous organic matter production before export to the coastal ocean. Approximately half of the Beaufort Sea coast is outlined by barrier island chains that enclose shallow lagoons and sounds. These lagoons are subject to extreme seasonal variations in ice cover, temperature, and salinity, yet are home to a diverse and productive food web. Bound between the arctic tundra and Beaufort Sea, these shallow systems receive and process resources from both sea and particularly land in the form of coastal erosion and riverine export. …

Exploring Transient Execution Vulnerabilities, Side-Channel Attacks, And Defenses, Tao Zhang

Dissertations, Theses, and Masters Projects

Modern microprocessors utilize branch prediction and speculative execution to enhance instruction throughput. Instead of stalling the pipeline and waiting for branch targets to be computed, the CPU consults branch predictors for a possible destination and performs speculative execution. These microarchitectural techniques improve the efficiency of instruction pipelining and out-of-order execution, enabling higher performance and better resource utilization. Despite their widespread adoption, the potential security implications of branch misprediction and transient execution have not drawn much attention until recently. Around early 2018, the discovery of Spectre attacks exposed critical vulnerabilities in CPUs, undermining both software and hardware isolation and confidentiality. These …