Mason Journals (George Mason Univ.)
Not a member yet
3256 research outputs found
Sort by
Zonal Wind Speeds in the Venusian Atmosphere from the Akatsuki and Venus Express Temperature Fields
No full textBetween 50 and 100 kilometers, superrotating zonal winds characterize the Venusian atmosphere. As a result of the superrotation, cyclostrophic balance can approximate the zonal winds in the Venusian mesosphere, where a pressure gradient counteracts the equatorward centrifugal force. Direct wind measurements are limited; the current available data on winds in the Venusian atmosphere use single-altitude cloud-tracked features. The existing calculations for zonal wind velocities using the cyclostrophic approximation do not utilize temperature fields from the Akatsuki Radio Science instrument or the Venus Express Solar Occultation in the Infrared (SOIR) instrument. Using the available information about cloud-tracked winds as boundary conditions, the assumption of cyclostrophic balance is applied up to 100 kilometers to calculate zonal wind velocities through upward and downward integrations. Above 100 kilometers, cyclostrophic balance breaks down due to a subsolar-to-antisolar circulation. Results demonstrate high wind velocities that peak between -100 m/s and -170 m/s, depending on the boundary condition and the temperature field used. Across all calculations, there is no evidence of a strong mid-latitudinal jet between 40 and 50 degrees latitude in either hemisphere, but there are slight increases in wind speed at latitudes around 30 degrees. Another finding that previous works on the zonal wind speed in the Venusian atmosphere do not account for is an increase in zonal winds with altitude at certain latitudes. A significant dependence of the zonal winds on the lower boundary and temperature field used is clear; however, all calculations agree on a superrotating atmosphere with significantly reduced wind speeds at high latitudes. The characterization of the zonal wind field will aid in future studies, like the characterization of gravity waves in the Venusian mesospher
The Effect of Rubric Language on Essay Scoring Accuracy in LLMs
No full textStudents and educators alike are increasingly turning to large language models (LLMs), a type of artificial intelligence, to assist with essay scoring and feedback. Unlike human scorers, however, LLMs are sensitive to how rubrics are worded. Variations in rubric design and phrasing can influence grading outcomes, even when the intended meaning remains unchanged. Despite a growing body of research on AI-assisted and automated essay scoring, no existing studies have experimentally tested whether changes in rubric phrasing alone, without altering semantic meaning, can affect LLM grading outcomes. To address this shortcoming, we rescored ten expert-graded AP U.S. History essays, obtained from publicly released College Board samples with included scoring rubrics, using two commonly used generative AI tools (Gemini 2.5 Flash and GPT-4o). Each essay was evaluated by the AI models using the original College Board rubric (verbatim neutral) and two reworded variations: a positively framed (affirming) version and a negatively framed (deficit-oriented) version. Rubric edits were screened with SBERT (cosine similarity ≥ 0.95) to confirm that wording shifts changed tone while preserving semantic content. Neutral rubric framing produced the highest overall exact-match accuracy with human scores (≈74%) and the most stable scoring across both models. Positive framing slightly inflated average scores (+0.2 on a 0–6 scale), while negative framing caused an average score decrease (−1.2 on a 0-6 scale) compared to human scorers. Grading variability also depends on the model: under a negatively framed rubric, Gemini 2.5 Flash lowered scores on mid-range essays (originally scored 3–4) 299% more often than GPT‑4o. The conclusion from our experiments is that LLMs are vulnerable to consistent, directional scoring shifts from rubric changes
The Effect of Different Lineup Orders on Average Runs Scored Per Softball Game
No full textSoftball, a sport often referred to as the “female variant” of baseball, is gradually gaining recognition and increasing participation. A key part of offensive play is determining a lineup order that would increase the chances of scoring runs. There are gaps in establishing an ideal batting order due to the controversies between coaches’ opinions and preferences. There is a lack of research with supporting evidence determining the most effective lineup. As this is an understudied field, investigating how different softball batting orders affect the average number of runs scored per game would help determine whether a specific lineup benefits a team’s offensive performance. We hypothesized that creating an order starting with a player possessing a high on-base percentage, followed by three stronger power hitters, would help teams reach the most runs possible. This experiment was conducted by programming a softball game simulator using Java on jGRASP. Thirty trials were conducted, one game representing one trial, for three different lineup types. At the end of each 30-trial experiment, the final mean runs scored were calculated to compare the effectiveness of each lineup type. The batting order with the greatest average runs scored was found to be the most ideal order. Future studies could further explore other aspects of the softball game that support the chances of winning
Digital Twin Framework for Real-Time Computing Infrastructure Monitoring
No full textSpatiotemporal studies demand significant computing power and infrastructure. The recent rise of artificial intelligencehas further amplified these computational requirements and introduced new cybersecurity risks. Existing data centermanagement tools diagnose system errors slowly and lack predictive capabilities. To address these challenges, wedeveloped a Computing Infrastructure Digital Twin (DT) of a 600 -machine data center (DC) to enable real-timemonitoring, autonomous detection of system issues, and efficient resource management. Metrics from the physical DCare collected using Prometheus, enabling real-time insights and an alert system based on pre-defined rules. Additionally,further information such as system log files are retrieved and stored in a PostgreSQL database for downstream tasks vialarge language models (LLMs) such as summarization, information extraction, and anomaly detection. The 3D virtualreplica, modeled in Autodesk Fusion and Onshape and visualized through NVIDIA Omniverse, reflects the real-time statusof the infrastructure, allowing users to detect and explore system errors through an interactive interface. Initial resultssuggest that this DT has implications for developing more efficient and secure data center management systems. Futureresearch will incorporate the use of artificial intelligence and machine learning (AI/ML) to predict potential anomalies,system errors, and security threats
Initial Evidence for Personal Development Support for College Students with Intellectual and Developmental Disabilities: Results from a Satisfaction Survey
No full textWith the increase in college students with intellectual and developmental disabilities (IDD) enrolled in inclusive postsecondary education (IPSE) programs, there is a growing call to understand additional support necessary for their mental health and wellness. Our IPSE program developed the personal development (PD) domain to meet the unique mental health and wellness needs of college students with IDD. Our study presents the findings from an annual satisfaction survey of 24 college students with IDD served within our PD domain. Data collected from the survey were analyzed through a convergent mixed methods design. Survey findings suggest promise for the PD domain to support student mental health and wellness, as well as individualized support for the development of independent living skills. Future investigation is needed to document the effectiveness of the PD domain, as well as health and mental health initiatives for college students with IDD in general
Laser-induced Graphene for Emerging Quantum and Energy Applications
No full textGraphene, an atomically thin honeycomb lattice of carbon, has been the most widely studied nanomaterial in the field. Despite its superior electrical, thermal, mechanical, and optical properties, employing graphene as a critical component of practical devices and systems requires a novel, cost-effective manufacturing process tailed to a specific application domain. In this work, we developed a simple process flow to fabricate a centimeter-scale graphene by illuminating a laser on a polyimide (PI) substrate. This laser-induced graphene (LIG) features a very high surface area due to a porous 3D structure while providing a unique platform to tune the physical properties of a metallic layer that is placed below the PI substrate. We performed electrical (sheet resistance), microscopic (scanning electron microscope), and spectroscopic (Raman) characterizations on LIG to ensure they possess ideal properties for emerging quantum and energy applications. With ongoing efforts in integrating LIG as an interdigitated electrode for supercapacitor, we expect to increase its energy density significantly. Also, our preliminary results indicate that LIG can easily diffuse into the underlying metal during the manufacturing process, thus forming a novel form of graphene-metal nanocomposites. This has the potential to develop a high Tc superconducting material that forms the foundation of state-of-the-art quantum computing hardware.  
Characterizing Developmental Changes in Long-Term Depression Through Chimeric NMDA Receptors
No full textSynaptic plasticity in the hippocampus, a key mechanism for learning and memory, involves long-term depression (LTD) mediated by NMDA receptor (NMDAR) activation. LTD, induced by low-frequency stimulation results in weaking of excitatory synaptic efficacy and shrinking of synapse size. Dysregulation in NMDAR signaling is linked to various neurological disorders, including depression, autism, schizophrenia, and Alzheimer’s disease. Although research has demonstrated that NMDARs operate through ionotropic and non-ionotropic mechanisms, the specific roles of individual NMDAR subunit functional domains in LTD induction, particularly those within GluN2A and GluN2B subunits, remain poorly understood. A developmental shift from GluN2B to GluN2A subunits at hippocampal excitatory synapses influences the ability to induce LTD, suggesting differing contributions of GluN2A and GluN2B in ionotropic or non-ionotropic signaling. This study aims to fill this gap by analyzing electrophysiological data from hippocampal slices of wild-type and transgenic mice with swapped GluN2A and GluN2B intracellular C-terminal domains, which separates ionotropic and non-ionotropic signaling. It is hypothesized that the changes in the signaling properties of GluN2A and GluN2B domains due to the switch in subunit composition will reduce the threshold for LTD induction. Field excitatory postsynaptic potentials (fEPSPs) were recorded in response to induced LTD from a 1 Hz low-frequency stimulation and assessed for input-output relationships and paired-pulse responses. Data analysis will utilize two-way ANOVA to determine the effects of age and genotype, with post-hoc tests to evaluate sex differences. This research is expected to advance understanding of the molecular mechanisms underlying LTD and inform the development of targeted therapies for cognitive and neurodegenerative disorders
The Slow Negative Effect of Drought on Denitrifying Bacteria Bioreactors’ Rate of Removal
No full textHigh nitrate levels in water can cause health issues and eutrophication. While most focus on treating nitrates in stormwater runoff, base-flow from groundwater can also contain nitrates. Bioreactors using denitrifying bacteria could remove nitrates perpetually rather than only after rainfall, and one bioreactor treating base-flow could be a substitute for dozens of bioreactors treating stormwater (Easton et al.). However, drought can decrease bioreactor efficiency as the bacteria die off. This study was conducted to determine how drought may impair bioreactor efficiency. In this experiment, two bioreactors were made in November with nitrate-laced distilled water, rocks, and woodchips. By creating the same conditions in both tanks, the rate of removal in each could be compared to check accuracy. Nitrate concentration readings were taken using nitrate probes and HACH TNT 835 tests, the latter being most accurate. The November average rate of removal was -0.02754 mg/L per hour. The tanks were then left alone without any additional nitrates, lowering the bacteria population to simulate drought. In December, the rate was tested again, and it had barely decreased to -0.02745 mg/L per hour. Then, in July, the bioreactors were drained and refilled with nitrate-laced water, causing the now smaller population to begin denitrification again. The new average rate, found using ANCOVA analysis, dropped to -0.00617 mg/L per hour. As such, drought can decrease bioreactor efficiency drastically, but the decrease in rate of removal occurs slowly enough for bioreactors to function during short droughts, increasing the potential applications of base-flow bioreactors in drier regions
Comparing Categorical Fire Risk Maps through 2001-2023
No full textThe recent increase in wildfires in California calls for studies aimed at understanding the impacts ofclimate change on fire risks in this region. In this study we used monthly Terra MODIS 0.05-degreeClimate-Modeling-Grid datasets for NDVI and LST to compute pre-fire-season (June) Fire-Risk Indexmaps over the California region for the years 2001, 2021, and 2022. The computation of the Fire-RiskIndex used a formula established in past studies. In order to understand the change in fire risk over thelast two decades we compared the 2021 and 2022 pre-fire-season Fire Risk maps to the corresponding2001 Fire Risk map. The method involved using 50th, 75th, 90th, and 95th percentiles from the June2001 Fire Risk data distribution to create Fire Risk category maps for all three years with the following riskclasses: Very-Low, Low, Medium, High, and Very-High. Analysis showed a 74.4% and 27.2% increase inarea under the Very-High Risk category, 96.7% and 44.1% increase in area under the High Risk category,and 39% and 5.7% increase in area under the Medium Risk category along with corresponding decreasesin area under Low and Very-Low Risk categories for the years 2021 and 2022 respectively whencompared to 2001. Our results suggest a phenomenon of conversion of lower fire risk areas to higher firerisk areas over the last two decades. However, more data-intensive studies are required to conclusivelyestablish any such phenomenon and link it to climate change
Evaluating the Impact of Governance and Funding Models on Open Source Software Success: A Quantitative and Qualitative Analysis
No full textOpen Source Software (OSS) projects, offering public accessibility and modifiability, have burgeoned with the advent of blockchain and other emerging technologies. Despite this growth, understanding the factors influencing their success remains complex. This study seeks to determine whether OSS projects with different funding and governance models exhibit varying GitHub development activities. The methodology involved a two-part approach. First, a broad quantitative analysis of OSS activity levels was conducted using data from 660 OSS projects hosted on GitHub. This analysis, performed with Google Cloud’s BigQuery API, included repository metadata such as repository names, dates, actor IDs, actor logins, and 28 distinct events, including total activities, distinct commits, and forks, covering the period from 2013 to 2023. Second, a qualitative deep dive into project backgrounds was performed to assess the influence of governance and funding models on project success. Preliminary results indicate varied activity levels across different project types, suggesting that specific governance and funding models may significantly impact OSS success, although the data is not yet fully realized. The study aims to provide a nuanced understanding of these correlations, ultimately guiding future developers and stakeholders in optimizing project design and funding approaches to enhance OSS sustainability and growth.
This abstract is part of a collection in which the overarching large project under Dr. Jiasun Li was subdivided into discrete critical tasks that were carried out by multiple individuals or smaller teams. Abstracts in this collection read similarly given the shared project goals, but represent distinct tasks completed by the abstract authors towards finalizing the described analysis