University of Tennessee Institute of Agriculture
University of Tennessee, Knoxville: TraceNot a member yet
47048 research outputs found
Sort by
Responsible Recklessness: Why High Performance Computing Must Embrace Randomized Numerical Linear Algebra
No full textRandomized Numerical Linear Algebra (RandNLA) is a relatively young branch of the field of Numerical Linear Algebra (NLA). It leverages randomization as the main approach to develop algorithms for computing solutions to classical linear algebra problems with performance superior to the deterministic NLA schemes. Despite that, RandNLA methods have yet to gain broad adoption within mainstream NLA software libraries, and, by extension, within the High-Performance Computing (HPC) toolkit aimed at combating the challenges posed by the end of Moore’s law. This dissertation argues that the HPC community is missing a powerful opportunity by not embracing RandNLA methods. As part of this work, I propose and analyze three RandNLA methods for the two NLA problem classes: full matrix factorizations through QR decomposition with column pivoting (QRCP), and low-rank factorization via Singular Value Decomposition (SVD). The core contribution of the projects I present lies in my development of high-performance implementations for each method, integrated into the open-source RandLAPACK library. First, I present a novel QRCP algorithm for tall matrices. I provide a comprehensive analysis of the method in exact arithmetic and give a broad characterization of its finite-precision arithmetic analysis. Experiments on the algorithm’s implementation demonstrate order-of-magnitude speedups over LAPACK’s standard function for QRCP. Next, I present an algorithmic framework that enables the design of practical QRCP algorithms for general matrices through user-controlled choices of the core subroutines. I provide a comprehensive overview of how to navigate these choices on modern hardware platforms, offering detailed descriptions of alternative methods for both CPUs and GPUs. CPU experiments demonstrate that the proposed method achieves performance improvements of up to two orders of magnitude over LAPACK’s standard QRCP routine. GPU experiments suggest that my method attains approximately 65% of the performance of cuSOLVER’s unpivoted QR factorization. Finally, I discuss a high-performance implementation of a partial SVD algorithm based on the Krylov subspace method. The algorithm is tested against randomized SVD and a deterministic SVD based on the implicitly restarted Lanczos method. Results on a dense synthetic dataset and a real-world sparse model reduction problem show the superiority of the algorithm I implemented
Counselor Education and Inclusive Postsecondary Education: A Partnership Model
No full textResearchers in the counseling field have alerted practitioners and counselor education professionals to the need for enhanced training related to supporting individuals with disabilities, namely intellectual and developmental disabilities (IDD; Feather & Carlson, 2019). As more individuals with IDD attend college, these emerging adults are specifically in need of counselors with skills and competencies to support their mental health. This paper describes a collaborative, mutually beneficial initiative between a counselor education program and an inclusive postsecondary education program to address the scarcity of personnel prepared to work with individuals with IDD
Relationships Win: Human Connection Outperforms Data in Collegiate Athletics
No full textAdvanced analytics and artificial intelligence (AI) occupy an increasing position and function within the institutional setting of collegiate athletics. At both the departmental and programmatic levels, collegiate athletic stakeholders have consistently increased their reliance on AI and data within decision-making processes and strategic development. Given its importance to competitive success, AI interfaces such as machine-based learning, predictive analysis, and natural language processing has been increasingly integrated within the recruiting process of prospective and transfer athletes. However, while such reliance on data-driven strategies and predictive analytics hold competitive value within the talent identification and evaluation process, the relational role of the recruitment of prospective athletes remains foundational to successful recruiting in collegiate athletics. This manuscript operationalizes the formation of recruiting within this setting and discusses the limited performative value of AI within the traditional college athlete recruiting process
From Enhancement to Substitution: A Strategic Provocation on Simulation-Based Sport
No full textAdvances in artificial intelligence, large-scale machine learning, and simulation technologies are rapidly transforming how sport is played, analyzed, and consumed. To date, most scholarly and industry discussions frame these technologies as tools that enhance embodied sport by improving performance, officiating, media production, and fan engagement. This paper extends that conversation by posing a more provocative strategic question: under what conditions might simulation move from enhancement to substitution? Focusing explicitly on sport as a business and entertainment enterprise, we argue that many of sport’s core sources of cultural and economic value—uncertainty of outcome, narrative continuity, legitimacy, and collective meaning—are structurally produced and not inherently dependent on human physical performance
Low-Cost Two-Color Pyrometry for Laser-Powder Bed Fusion
No full textLaser-Powder Bed Fusion (L-PBF) additive manufacturing relies on precise control of melting and solidification, making accurate thermal measurement essential for maintaining build quality and detecting process instabilities. This work presents a novel method for real-time, pixel-wise temperature measurement in L-PBF using a low-cost two-color pyrometry system built from synchronized Near-Infrared (NIR) cameras. The system uses two high-resolution USB 3.0 Basler acA5472 cameras—one monochrome and one color—outfitted with narrowband filters centered at 900 nm and 750 nm, respectively, along with neutral density filtering to prevent sensor saturation. Spatial and temporal alignment is achieved through a custom calibration routine using a 9×11 circular grid pattern, enabling precise perspective correction between the two camera images. Temperature mapping is performed using a ratio-of-radiance approach based on Wien’s approximation, with calibration constants derived from a tungsten filament reference source. The resulting output yields high-resolution, repeatable thermal maps of the melt pool that correlate well with expected trends and reference data from a FLIR A50 radiometric camera. The system demonstrates a scalable, cost-effective solution for in-situ thermal monitoring in metal additive manufacturing and offers a strong foundation for future investigations into melt pool dynamics, process stability and closed-loop control
Growth Condition for Leptospira: Assessing Metabolomic Regulation in Variable Media and Temperature
No full textLeptospirosis is a global zoonotic disease which is significant yet quite underrecognized. It is caused by Leptospira, which is transpired to humans from animals. There is no definitive identification of Leptospira infection in human. The objective of this study was to investigate the metabolites of Leptospira in vitro to create a metabolomic profile of the pathogenic bacteria. The metabolomic profile from the pellet, supernatant, and the control media allowed the development of an intracellular and extracellular metabolomic signature. The unique metabolomic signatures will further enable searches for biomarkers that will be significant for identification of leptospirosis in human. These results reveal distinct temperature-media interactions affecting both cellular composition and secreted factors, with significant differences observed in growth yield and metabolite secretion profiles. These findings provide new insights into the environmental adaptability of leptospira and offer guidance for optimizing laboratory cultivation and downstream analyses
Mixed Methods Analysis of User Engagement, Linguistic Indicators, and Psychological Associations in Online Suicidal Behavior
No full textThis research examined suicidal ideation and behavior in Sanctioned Suicide, an uncensored online forum where users openly discuss suicide. We analyzed 10,646 posts from 4,087 users. First, linguistic analysis revealed that positive tone predicted increased community responses while negative tone decreased responses. Second, receiving responses from others predicted continued posting only for users making venting posts, not method or goodbye posts. Third, “Goodbye” posts displayed significantly higher positive tone and future orientation compared to venting or method posts, consistent with paradoxical positivity shifts observed in suicide notes. Fourth, content analysis revealed that unbearability showed significant association with suicide behaviors while unlovability and unsolvability did not, providing partial ecological validation of these clinical constructs in naturalistic discourse. Fifth, specific stressor categories did not predict suicide behaviors in cross-sectional posts, suggesting temporal dynamics not captured within single communications.
Findings reveal distinct social, cognitive, and linguistic dynamics in online suicide discourse. Positive emotional expression increases engagement but may signal acute risk when appearing in farewell posts. Unbearability emerged as the most proximal cognitive marker of behavioral progression. The absence of cross-sectional stressor-behavior associations underscores limitations of single-post analysis and highlights the need for longitudinal approaches. These insights inform digital risk detection strategies and suicide prevention efforts tailored to online community dynamics
A UNIFIED REAL-TIME BENCHMARK TO EVALUATE THE PRACTICAL IMPACT OF WIRELESS TECHNOLOGY ON THE EDGE
No full textThis thesis presents the design, implementation, and evaluation of a portable, real-time edge computing and streaming system deployed over modern wireless network technologies, including Wi-Fi 6, 4G LTE, and 5G. Built around a high-demand critical service use case, the system benchmarks the practical performance limits of compact, off-the-shelf hardware in realistic operating conditions. By varying processing platforms, communication channels, and application parameters, this study monitors both application-level performance and the state of the underlying network in real time.
The primary contribution of this work is the development of a benchmarking framework that quantifies how well consumer-grade edge devices meet the stringent requirements of critical infrastructure applications such as those in automation, manufacturing, and security. The system captures live sensor data, applies task-specific processing, streams the resulting media across a network link, and logs performance metrics including bitrate, latency, and frame rate. Simultaneously, it observes key link-layer indicators such as signal strength and noise levels.
A secondary but significant contribution is the characterization of how network conditions directly influence the performance of delay- and bandwidth-sensitive applications. For example, experiments with a 5G mmWave channel revealed that even minor increases in signal attenuation can cause substantial degradation in streaming bitrate and delay. These findings provide actionable insight into the design constraints and expectations for wireless deployment of critical services.
Additional contributions include improvements to system performance through the refinement of streaming protocols and compression techniques. These optimizations further enhance the system’s ability to reflect real-world usage scenarios and make the findings more generalizable.
In summary, this thesis offers a practical and reproducible approach to benchmarking edge computing systems under real-world conditions, providing valuable guidance for engineers and researchers developing next-generation networked applications for critical infrastructure
Beyond Interdisciplinary: Igniting Transdisciplinary Thinking in Tomorrow’s Music Technology Graduates
No full textOutwardly, this presentation is an inquest into the structural quandaries of curricular design in music technology—a practical extension to the seminal discussion about music technology curriculum launched by Carola Boehm (2007). I recently pioneered a new BSc degree program in Digital Music Technology at my institution, obtaining the support of four departments across campus to manifest an interdisciplinary implementation, yet the curricular structure governing our degree programs was generally incompatible with Boehm’s foremost recommendations. This led me down the very curricular path she rejected as “modernist”: derived from 18th-century concepts of the university in which interdisciplinarity was realized through the acute study of each constituent, “pure” field. According to Boehm, too much is hefted upon the shoulders of the learner under this model—the pupil is often left alone with the task of actualizing interdisciplinary understanding between the different domains. Boehm instead advocates a move toward “post-modernist” curricula, which favor nuanced courses that expressly highlight interdisciplinarity. From this pronouncement I developed a temporary conviction that I had inadvertently piloted my curriculum in the wrong direction.
At its core, this presentation motivates a philosophical edict for music technology and its pedagogy, deduced from my crusade to transcend the perceived shortcomings of my unavoidably modernist curricular structure. A study mapping the employability of Australian music-technology graduates according to curricular attributes of the nation’s degree programs posed the idea of music technology as transdisciplinary (Klein & Lewandowski-Cox, 2019). In this I detected a novel perspective about the field’s multidisciplinary makeup that not only maintains compatibility with established interdisciplinary aspirations but also encourages curricular designers to regard core coursework in music technology as an optimal vehicle for pushing students toward fresh thinking through collective assimilation of individual disciplinary perspectives.
References:
Boehm, C. (2007). The discipline that never was: current developments in music technology in higher education in Britain. Journal of Music, Technology and Education 1(1), 7–21. https://doi.org/10.1386/jmte.1.1.7/1
Klein, E., and Lewandowski-Cox, J. (2019). Music technology and Future Work Skills 2020: An employability mapping of Australian undergraduate music technology curriculum. International Journal of Music Education, 37(4), 636-653. https://doi.org/10.1177/025576141986144
Hearing the Wind Again
No full textI love listening to the wind blow through the forests and mountains—hearing its pitch and timbre change, its approaching and passing, and its power. This piece was inspired by the sound of the wind, and my mother’s tears at hearing the wind. Before getting hearing aids, she hadn’t heard it for years