79297 research outputs found
Sort by
ChplX: The HPX Solution for Bridging Chapel and C++
No full textHistorically, domain scientists faced steep learning curves due to low-level programming models and fragmented tooling. Between 2003 and 2008, Cray, now part of HPE, introduced the Chapel language as part of DARPA’s High Productivity Computing Systems (HPCS) program. Today, Chapel remains under active development and is used across research and production projects. In parallel, the STE||AR Group has advanced C++-based parallel programming through HPX, a standards-conforming runtime that provides lightweight tasking, futures, and distributed execution while abstracting much of the algorithmic “heavy lifting.” Yet for many domain scientists, C++ presents a steeper learning curve than Chapel. To close this complexity introduced and to integrate with the existing C++ tooling and infrastructure, we introduce ChplX - a line-by-line source-to-source translator for Chapel to C++. This approach lets Chapel programs leverage mature HPX/C++ tooling, including APEX for profiling and Traveler for timeline (Gantt-style) visualization of execution. Additionally, this gives Chapel the benefit of integrating with existing C++ projects using CMake
Examining Private Landowner And Stakeholder Attitudes, Perceptions, and Knowledge Of Prescribed Fire In Louisiana
No full textPeriodic fire was a primary driver of ecosystems across the Southeastern United States for millennia; however, fire suppression policies have become pervasive in the United States and led to increased fuel loads in forests as well as reduced biodiversity and reduced disease resilience. Prescribed fire presents a risk mitigation strategy and land management tool whereby practitioners can apply fire to the landscape in a controlled manner, thus reaping the benefits of fire while minimizing the risks associated with fire. Land in Louisiana is primarily privately managed and owned, so state-level conservation goals cannot be met without private landowner participation. As such, it is imperative that we, as natural resource managers, understand private landowner attitudes towards and knowledge of prescribed fire. I prepared a ~45 question survey to identify private landowner attitudes towards and knowledge of prescribed fire and wildfire and distributed it to private landowners in grasslands, timberlands, and wetlands in Louisiana. I extended this survey to private landowners engaged in extension activities to identify differences in attitudes across private landowner groups. I also surveyed community leaders as they are likely to have influence over private landowners in their area. I analyzed this data in broad survey analysis groups (e.g., community leaders, private landowners, and private landowners engaged in extension activities), as well as, survey groups (e.g. grassland mailed private landowner surveys, wetland community leaders, and Florida parishes participants engaged in extension programming). I also modeled prescribed fire use by private landowners and found that land ownership time, education level, and focal area (e.g., grasslands, timberlands, and wetlands) were the most important factors in determining someone’s probability of using prescribed fire. Results of my study suggest that geographically targeted extension activities aimed at recruiting new private landowners could prove very effective in increasing the use of prescribed fire on private lands in Louisiana
Precipitation and binder effectiveness in the electrochemical reduction of CO2 at Cu electrocatalysts in zero-gap MEA cells
No full textMembrane Electrode Assembly (MEA) electrolyzers equipped with Gas Diffusion Electrodes (GDEs) present significant potential for industrial-scale CO₂ reduction (CO2RR) due to their capability to operate at high current densities with high Faradaic efficiencies (FEs). Despite this advantage, several key challenges persist, including the long-term stability of MEAs, enhanced selectivity for CO2RR, and the effective production of multi-carbon products. This study addresses critical factors leading to the failure of Cu-based CO₂ electrolyzers, specifically suppression of the hydrogen evolution reaction, the loss of electrocatalytic activity in the Gas Diffusion Layer (GDL) caused by electrocatalyst degradation or agglomeration, and the formation of carbonate salts, all within the context of various electrocatalyst binders (or ionomers when functional groups are employed) combinations including Cu-PTFE, Cu-PVDF, and Cu-Nafion. Energy-dispersive X-ray spectroscopy (EDX) and Cu L-edge X-ray absorption spectroscopy (XAS) demonstrate that PTFE maintains the highest loading of electrocatalysts on the gas diffusion layer, indicating sustained active sites for CO2RR. Complementary carbon K-edge and potassium L-edge XAS reveal the presence of carbonates in K2CO3, KHCO3, and malachite forms. Combining the electrochemical results with soft X-ray absorption spectroscopy, we hypothesize that carbonate plays a dual role in CO2RR. First, the dissolution of Cu under reaction conditions encourages the formation of copper carbonate hydroxide, which enhances C2 product selectivity. Second, the precipitation of K2CO3 and KHCO3 salts fills the pores in GDE, promoting the hydrogen evolution reaction
Bridging Modalities: Enhancing Multimodal Sentiment Analysis for Social Media Networks
No full textSocial media platforms like X (formerly Twitter) serve as rich sources of textual and visual information, making multimodal sentiment analysis essential for understanding complex human emotions. This dissertation aims to advance multimodal sentiment analysis by improving the semantic alignment and fusion of textual and visual features, thereby enabling more accurate and context-aware sentiment interpretation of social media content.
To address challenges in multimodal integration, this work proposes two complementary MSA approaches. The first approach introduces a similarity-based multi-layer attention neural network (SiMANN) that enhances modality integration through cosine-based similarity fusion and modality-specific attention to emphasize salient features in text and images. Although effective, this approach revealed limitations related to weak fine-grained alignment, heterogeneous feature embeddings, and limited cross-modal interaction modeling. To overcome these constraints, the dissertation presents SentiGAT, a graph attention network-based framework that employs a unified CLIP-based feature extractor to encode modalities into a shared semantic space. It further incorporates a graph attention network (GAT)-based word–object alignment to strengthen fine-grained semantic correspondence, and a GAT-based fusion module that learns content-aware inter-modal dependencies for sentiment prediction.
The proposed methods were evaluated on the MVSA-Single and MVSA-Multiple benchmark datasets. SentiGAT demonstrated superior performance compared to state-of-the-art models and large language model-based approaches, yielding notable improvements in both accuracy and F1 score. Consistent gains across 10-fold cross-validation confirmed the robustness and generalizability of the approach. Further analysis showed that SentiGAT effectively captures nuanced cross-modal relationships, enhances semantic grounding through word–object alignment, and improves interpretability by highlighting modality-level contributions via attention mechanisms.
This dissertation presents a progressive advancement in multimodal sentiment analysis, starting from real-world observations on social media and leading to a graph-based framework for improved alignment and fusion. By addressing key limitations in how textual and visual features interact, the proposed methods enable more accurate and context-aware sentiment interpretation. The findings provide a strong basis for future research in multimodal learning, affective computing, and social media analytics
Structural and magnetic properties of CoTeMo O6
No full textWe have conducted a comprehensive investigation into the magnetic properties of the chiral multiferroic material CoTeMoO6. In contrast with the previous claim of canted antiferromagnetic order with ferromagnetic components [Y. Doi, J. Solid State Chem. 182, 3232 (2009)10.1016/j.jssc.2009.09.008], our investigation reveals an antiferromagnetic ground state with compensated moments, providing an interesting platform for exploring exotic material properties. Through careful measurements of magnetization under a series of applied fields, we demonstrate that there exist two sequential field-induced magnetic transitions in CoTeMoO6, with one occurring at Hc1=460 Oe along the a axis, and the other at Hc2=1.16 T with the field along the b axis. The values of Hc1 and Hc2 exhibit strong angular dependence and diverge with different rates as the applied field is rotated 90 ° within the ab plane. This reflects the distinct nature of these transitions, which is further supported by the different critical behavior of Hc1 and Hc2, characterized by the values of γ, in the function of Hc=H0(1-T/Tc)γ. Furthermore, we have demonstrated that there exist structural and magnetic twin domains in CoTeMoO6 that strongly affect the experimental measurement of their macroscopic properties. Intriguingly, these twin domains can be related to the orthorhombicity/chirality of the crystal structure with the space group P21212. We further explored the magnetic and structural domains with uniaxial pressure and polarized light microscopy. Our results suggest that CoTeMoO6 could be used as a unique platform for investigating the intriguing physics involving intertwined degrees of freedom. The tunability of the underlying domain distribution and its strong anisotropy could also be useful for developing functional devices and applications
Trans-ARG: Predicting Antibiotic Resistance Genes with a Transformer-Based Model and Pretrained Protein Language Model
No full textAntibiotic resistance presents an emerging challenge to global health by reducing the efficacy of antibiotics administered to treat bacterial infections. Traditional approaches for detecting antibiotic resistance genes (ARGs) involve alignment-based sequence similarity techniques, which are time-consuming and resource-intensive. This signifies the necessity for developing advanced computational methods to detect ARGs early. To address this challenge, we introduce Trans-ARG, a novel multi-head attention transformer-based model designed to predict potential ARGs. Our approach leverages the pretrained ESMFold model to extract embeddings from protein sequences, capturing intricate structural and functional information. Protein sequences labelled as ARG or non-ARG were used as the dataset. We formulate our problem as a binary classification task, where the extracted embeddings serve as inputs to our transformer model with output results predicting ARGs. The transformer network is excellent at handling sequential data and capturing long-range dependencies, and the use of multi-head attention within the network improves he model’s capacity to comprehend relationships within the data from various perspectives. Additionally, we implemented a five-fold cross-validation strategy to ensure robust performance during training. Our experimental results demonstrate that Trans-ARG significantly outperforms standard existing baseline methods, presenting an accuracy of 90.96% and an AUC score of 97.08% on the test dataset. The high-efficiency performance of Trans-ARG is attributed to integrating embedding features obtained from the pretrained ESMFold model and effectively utilizing the transformer’s architecture. This integration allows Trans-ARG to generalize well across diverse protein sequences, making it a valuable tool for ARG prediction. Future research may explore applying this approach to predict antibiotic resistance categories, further enhancing our understanding of antibiotic resistance
Exact Inference for Transformed Large-Scale Varying Coefficient Models with Applications
No full textStudying migration patterns driven by extreme environmental events is crucial for building a sustainable society and stable economy. Motivated by a real dataset about human migrations, this paper develops a transformed varying coefficient model for origin and destination (OD) regression to elucidate the complex associations of migration patterns with spatio-temporal dependencies and socioeconomic factors. Existing studies often overlook the dynamic effects of these factors in OD regression. Furthermore, with the increasing ease of collecting OD data, the scale of current OD regression data is typically large, necessitating the development of methods for efficiently fitting large-scale migration data. We address the challenge by proposing a new Bayesian interpretation for the proposed OD models, leveraging sufficient statistics for efficient big data computation. Our method, inspired by migration studies, promises broad applicability across various fields, contributing to refined statistical analysis techniques. Extensive numerical studies are provided, and insights from real data analysis are shared
Protected Area Co-Management in the Context of State Conservation and Development: Comparing Cases in Costa Rica and Colombia
No full textParque Nacional Cahuita and Parque Nacional Natural Uramba Bahía Málaga are the !rst national parks in Costa Rica and Colombia, respectively, that are co-managed by the state and the community. Both co-management arrangements are between state governments and communities of primarily African descent, populations marginalized by enduring colonial logics within both states. Comparing the stories of these two communities and how they achieved park co-management o ers insights into the potential and limitations of co-manage-ment as a way for Afro-Latin Americans to assert their environmental rights. It also illustrates how di erent contexts of struggle result in distinct co-management experiences, including di erences in co-management precarity, local engagement, and perceived e ectiveness. Draw-ing on historical documents and qualitative research in both Costa Rica and Colombia, this paper o ers a critical analysis of how co-management intersects with state conservation prac-tices and impacts local communities and environments in complex and o#en incoherent ways.
Resumen
El Parque Nacional Cahuita y el Parque Nacional Natural Uramba Bahía Málaga son los prim-eros parques nacionales en Costa Rica y Colombia, respectivamente, que son comanejados por el estado y la comunidad. Ambos acuerdos de comanejo son entre gobiernos estatales y comunidades de predominante ascendencia africana, poblaciones marginadas por lógicas coloniales persistentes en ambos estados. Una comparación entre las historias de estas dos comunidades y cómo lograron el comanejo del parque ofrece perspectivas sobre los potencia-les y limitaciones del comanejo como una forma para que los afro-latinoamericanos a!rmen sus derechos ambientales. También ilustra cómo diferentes contextos de lucha resultan en experiencias distintas de comanejo, incluyendo diferencias en la precariedad del comanejo, la participación local y la efectividad percibida. Basándose en documentos históricos y meses de investigación cualitativa tanto en Costa Rica como en Colombia, este artículo ofrece un análisis crítico de cómo el comanejo se cruza con las prácticas de conservación del estado y afecta a las comunidades y entornos locales de maneras complejas y a menudo incoherentes
Real-time electron spectrometer utilizing a permanent magnet and diode detector array
No full textBackground: Magnetic spectrometers have been previously described for measuring energy spectra of therapeutic electron beams. However, challenges for clinical utilization have been their size, weight, and limited real-time capabilities. Development of a compact, lightweight, and inexpensive device with real-time readout will make an electron spectrometer a practical clinical tool. Purpose: This work integrates a commercial diode detector array with a permanent dipole magnet to create a practical real-time energy spectrometer for therapeutic electron beams. Methods: A 4 kg, 0.55 T permanent dipole magnet was coupled to two PC boards of a Sun Nuclear SRS MapCHECK device, which provided two interlaced diode arrays that sampled a radiation field with an effective spacing of 0.175 cm. These components were rigidly attached to a copper insert in an Elekta 14 × 14 cm2 electron applicator. The insert\u27s 0.6 cm diameter aperture on central axis selected the electron beam entering the magnet. The Lorentz force spatially dispersed the electron beam onto the diode arrays, which measured the spectrometer response, (Formula presented.), at diode location (Formula presented.) along beam central axis. Background X-ray dose to the diode detectors and its integrated circuits (IC), shielded by a 5.75 cm thick Cerrobend block attached to the copper insert, was measured. Corrections were made to (Formula presented.) for individual diode sensitivity, the 0.5 cm separation of the two diode array planes, and use of only four (three) diode columns from the proximal (distal) arrays. Modeling the magnet to have constant primary and fringe fields, while correlating peak positions of the energy spectra with Ep,0 from central-axis dose versus depth curves in water, produced an energy calibration curve (E vs. z). (Formula presented.) measurements were evaluated for 15 and 1 s (real-time) intervals. Monte Carlo-calculated, monoenergetic detector response functions, DRF(E, z), were used to extract the incident energy spectrum, (Formula presented.), from the corrected (Formula presented.). Results: Background dose to IC electronics was 11.6 ± 1.7 µGy/MU and 42.8 ± 4.4 µGy/MU for 9 and 20 MeV beams, respectively, allowing \u3e100 h of use at 400 MU/min before receiving 100 Gy. Background to diode detectors was less than 10% of peak signal for 7–20 MeV beams. Individual diode sensitivities varied ±6%, each varying insignificantly with energy. Mapping distal and proximal diode array readings with 0.175 cm spacing provided (Formula presented.) curves with 45 data points. Energy spectra from 1 and 15 s measurement times were identical, demonstrating real-time measurement rates ≥1 Hz. Energy spectra measured for six matched Elekta accelerators showed notable differences in shape (peak location and FWHM), demonstrating the spectrometer\u27s potential for beam tuning and quality assurance. Conclusions: This study showed that an SRS MapCHECK diode array coupled with a 0.55 T permanent magnet can be used to construct a real-time electron energy spectrometer that should be compact, lightweight, inexpensive, and practical. It offers accelerator engineers a new tool to improve efficiency and effectiveness of electron beam tuning during commissioning and maintenance. Also, it offers medical physicists a potentially efficient and effective paradigm for machine quality assurance
A Comparative Analysis Of Effectiveness Of Safety Training Modalities In Confined Spaces
No full textThis study presents a comparative analysis of the effectiveness of four safety training modalities for confined-space safety within the construction industry. The research evaluates traditional in-person training (IPO), online e-learning training (OLO), in-person training supplemented with virtual reality (IPVR), and online training augmented with virtual reality (OLVR). A true experimental, randomized pre-test/post-test design was employed using Analysis of Covariance (ANCOVA) to measure differences in learning outcomes using eighty participants (N = 80), assigned to IPO (n = 19), OLO (n = 23), IPVR (n = 20), and OLVR (n = 18). Each participant completed a modality-specific confined space safety course, and their knowledge was evaluated before and after training to assess retention and instructional effectiveness.
Based on adult learning and experiential learning theories, the study examined how instructional design and technology integration influence learner engagement and knowledge acquisition. Statistical analysis revealed significant differences among the four modalities. With a maximum score of 26 on the post-test assessment, the IPVR group achieved the highest adjusted post-test mean score (M = 24.30), followed closely by IPO (M = 24.21). In contrast, OLO (M = 22.43) and OLVR (M = 22.39) produced lower results. These findings indicate that combining direct instructor engagement with immersive simulation, such as virtual reality, makes the most effective learning outcomes for confined space safety training.
The results reinforce Kolb’s (2014) experiential learning model and Knowles’ (1971) andragogical principles, demonstrating that adult learners benefit most when instruction is interactive, relevant, and experience-based. The study provides evidence-based guidance for safety professionals, trainers, and industry employers in selecting the most effective training methods for confined space safety. Ultimately, the goal is to enhance workers\u27 safety knowledge, mitigate job site hazards, and reduce accidents and injuries in hazardous confined space environments