137091 research outputs found
Sort by
Developing And Validating a Framework for Measuring Quiet Quitting Intentions
No full textThis dissertation developed a new framework for measuring Quiet Quitting Intention (QQI), defined as the internal decision to withhold discretionary effort while still meeting formal job requirements. Unlike prior definitions that focus only on behavior, this research uses the Theory of Planned Behavior (TPB) to capture the motivational basis of quiet quitting. The construct is divided into two distinct forms: Extrinsic QQI (EQQI), based on external standards like role expectations, and Intrinsic QQI (IQQI), based on personal boundaries or values. Across four studies, the project uses expert interviews, AI-supported item generation, and survey-based validation to develop and refine the measure. Confirmatory factor analyses and cross-cultural tests are used to evaluate dimensionality, while a nomological network of related constructs—including burnout, engagement, and turnover intention—is used to assess construct validity. This work clarifies the psychological foundations of quiet quitting and offers a tool for future research and organizational assessment
Constraining Regional and Global Atmospheric Emissions and Chemistry Using Models and Observations
No full textThis thesis presents three interconnected studies advancing our understanding of atmospheric chemistry through investigations of biogenic emissions, ozone pollution control, and interhemispheric transport. In the first work, we developed the Speciated Isoprene Emission Model with the MEGAN Algorithm for China (SieMAC), a comprehensive biogenic isoprene emissions model specifically designed for China's diverse ecosystems. By integrating extensive local emission factor measurements, high-resolution vegetation distributions, and plant functional type-specific leaf area indices, SieMAC significantly improves upon existing MEGAN versions. The model incorporates optional water stress effects through vapor pressure deficit calculations and modified temperature response algorithms for boreal grasslands. Evaluation against ground-based observations and satellite formaldehyde data demonstrates superior performance compared to MEGAN v2.1 and v3.1, particularly in northern China where previous models showed systematic underestimation. The higher estimate of isoprene emissions in SieMAC highlights the significant role of biogenic emissions in China's ozone pollution. In the second work, through analysis of ~1,400 monitoring sites across China and comparison with historical data from the United States and Europe, we identified two key mechanisms explaining China's limited ozone response to substantial NOx emission reductions since 2013. First, China's high-ozone regions exhibit weak or negative correlations between odd oxygen (Ox = O3 + NO2) and NO2, indicating NOx-saturated chemical regimes resistant to initial emission cuts. A critical NO2 threshold is identified, below which ozone begins responding positively to further NOx reductions. Second, using modeling, we found that the emission redistribution process creates a "redistribution penalty" that undermines reduction benefits. In the third work, we evaluated and optimized the transport processes in the 12-box AGAGE atmospheric model, which is widely used for inversion of greenhouse gas emissions. Using SF6 observations, we optimized the model transport parameters to improve representation of large-scale atmospheric circulation. The tuned transport settings provide a more realistic foundation for future simulations of long-lived species distribution, including CH4 source apportionment studies.Ph.D.Earth and Atmospheric Science
Modeling Prescribed Burning Smoke at Regional and Local Scales Using Fire Behavior and Chemical Transport Models
No full textPrescribed burning is an essential land management tool for reducing hazardous fuel loads, maintaining ecosystem health, and mitigating wildfire risks. However, like wildfires, prescribed burns emit fine particulate matter (PM2.5), ozone precursors, and other pollutants that can degrade air quality and affect public health. This thesis addresses a critical need in fire and air quality management: improving the accuracy and applicability of smoke modeling frameworks for both regional and local scales, with a focus on addressing uncertainties in emissions estimation, meteorological inputs, and plume dynamics.
At the regional scale, the BlueSky pipeline and the chemical transport model (CTM), CMAQ, are applied to estimate emissions and simulate air quality impacts from prescribed fires. Burned areas from prescribed fires are identified from a remote sensing product and calibrated against state burn permit records. These improved burned area estimates are processed through BlueSky to produce hourly emissions for CMAQ simulations, enabling assessments of prescribed fire health impacts in the southeastern U.S. BlueSky and CMAQ are also used to simulate factual and counterfactual scenarios to evaluate air quality trade-offs between wildfires and prescribed fires. A case study of the 2016 Gatlinburg wildfire reveals that prescribed fire management can reduce overall smoke exposure. To address biases in regional CTM simulations, a generalized data fusion method integrates observations and CTM simulations, improving regional PM2.5 estimates and enabling prescribed fire-specific air quality impact assessments.
At the local scale, the fire behavior model WRF-SFIRE and the CTM CMAQ are effective modeling frameworks for simulating smoke from prescribed burns. The thesis investigates the impacts of biased wind simulations on smoke concentration simulations. Wind bias reduction methods improve meteorological simulations but cannot eliminate wind biases, prompting the development of smoke model evaluation methods to quantify concentration uncertainty from meteorology. Model intercomparison between WRF-SFIRE and BlueSky-CMAQ highlights their complementary strengths: WRF-SFIRE captures fire-atmosphere interactions and plume dynamics more realistically, whereas BlueSky-CMAQ simulates atmospheric chemistry and secondary pollutants but oversimplifies plume rise. To combine these capabilities, a generalizable offline coupling algorithm is developed to integrate WRF-SFIRE and CMAQ, improving PM2.5 and ozone predictions with reasonable computational cost.
Overall, the thesis demonstrates that refining emissions inventories, reducing meteorological bias, applying data fusion, and coupling fire behavior models with CTMs can meaningfully improve smoke exposure estimates both in regional and local scales. The developed modeling methods and findings can be used to plan prescribed fires that minimize air quality impacts, facilitate smoke forecasting for public communication, and support policies that reduce wildfire risk through prescribed burning while protecting air quality
Understanding Evolutionary Constraints of a Lunar Base From an Operational Perspective
No full textPresented at AIAA SciTech Forum 2026The United States along with many other spacefaring countries have set the goal of establishing sustained human presence on the lunar surface. To build up to that, certain requirements must be met to fulfill the objectives that have been determined as part of the human lunar campaign. Due to the absence of existing lunar infrastructure, there is an opportunity to establish a systematic approach that will account for the growth of the lunar base as it achieves key milestones over time. Within this systematic approach, measuring the operational performance is crucial to understand the feasibility of the architecture. In this paper, we simulate the operations of the lunar base to ensure that the requirements and objectives are satisfied in each phase of the evolutionary process. The aim is to understand the limitations of architectural phases and identify potential bottlenecks to better inform development timelines. We discuss the methodology used to sample several points in the architectural design space and simulate their operations. By analyzing the sensitivities of objective fulfillment to operational parameters, we identify an appropriate path for lunar base evolution over time
Machine Learning and Biomechanical Sensing Toward Real-Time In-The-Loop Gait and Joint Health Optimization
No full textDespite advances in wearable sensing and assistive devices, current systems often rely on indirect or delayed signals that limit their ability to capture subcutaneous physiological dynamics in real-time. This challenge hinders progress across domains ranging from exoskeleton control to clinical monitoring in populations with movement or inflammatory disorders. This work aimed to expand corresponding biomechanical sensing capabilities by developing a novel sensing framework capable of extracting under-the-skin biomechanical signals from muscle and tendon structures using ultrasound and active acoustics. To achieve this, we (Aim 1) developed a machine learning pipeline for real-time estimation of muscle fascicle lengths from B-mode ultrasound images to enable “muscle-in-the-loop” feedback systems; (Aim 2) introduced and benchmarked an active acoustics sensor capable of measuring Achilles tendon loading in real-time with low latency across a wide range of locomotion tasks; and (Aim 3) applied our acoustics sensing approach in a pediatric arthritis cohort to quantify how inflammation-related physiological alterations affect machine learning task classification performance, highlighting its potential as a non-invasive biomarker for disease presence and severity. Collectively, these studies establish a new direction for non-invasive, task-relevant muscle-tendon sensing that could inform next-generation systems for rehabilitation, augmentation, and clinical assessment
Software-Hardware Optimizations for Efficient Collective Communications in Distributed Machine Learning Platforms
No full textFoundation machine learning (ML) models have emerged as one of the most prominent applications in modern computing, exemplified by mixture-of-experts–based large language models. The immense resource demands of these models have driven the development of large-scale, high-performance computing platforms tailored for artificial intelligence workloads. In such distributed platforms, both model parameters and data are partitioned and processed across numerous neural processing units, requiring frequent synchronization of activations and gradients through collective communication operations. As collective communication constitutes a primary bottleneck in distributed ML, optimizing its efficiency remains a critical research challenge. This dissertation explores software-hardware optimizations for collective communications to better understand the tightly coupled design space of networking in distributed ML platforms. First, it introduces ASTRA-sim2.0, an end-to-end simulation and modeling framework that enables comprehensive design space exploration of distributed ML platforms with arbitrary parallelization strategies and multi-dimensional networks. Second, it presents LIBRA, which enhances the bandwidth utilization of hierarchical collective communication algorithms by optimizing multi-dimensional network topologies via analytical modeling. Finally, the dissertation proposes two collective communication algorithm synthesizers, TACOS and PCCL, which automatically generate optimized collective communication algorithms for arbitrary network topologies through algorithmic approaches. Together, the dissertation underscores the significance of judicious software-hardware approaches in achieving efficient collective communication for large-scale distributed ML platforms
Aerodynamic Flow Control on Axisymmetric Cylindrical Model at High Incidence
No full textThis research focuses on the evolution and instabilities of forebody vortices over slender
axisymmetric cylinders at high angles of incidence, with an emphasis on the control and
manipulation of the resulting aerodynamic loads. The investigations build on prior studies of
vortex asymmetry and stability, hypothesizing that fluidic actuation can alter vortex dynamics and
aerodynamic forces to achieve controlled flow symmetry and stability under varying conditions.
The experiments were conducted on cylindrical models integrated with synthetic jet actuators
and a forebody sectional bleed mechanism, allowing for systematic control of azimuthal actuation.
The flow evolution and aerodynamic loads were measured using wind tunnel testing,
complemented by planar particle image velocimetry and load cell measurements.
The investigations explored the receptivity of the forebody vortices to synthetic jet actuation,
demonstrating that azimuthally varying actuation could control asymmetries and stabilize the
aerodynamic loads at low and high incidence. Another approach employing forebody sectional
bleed was developed to investigate the potential for manipulating vortex asymmetry by leveraging
azimuthally distributed bleed driven by local pressure variations, with the aim of achieving
deliberate control over vortex configurations and associated side loads. An internally rotating
bleed mechanism was integrated into the forebody, enabling bi-directional control of aerodynamic
forces and vortex dynamics. These findings reveal the efficacy of both synthetic jet actuation and
aerodynamic bleed in manipulating forebody vortices and regulating associated loads under varied
flow conditionPh.D.Mechanical Engineerin
Tumor-Localized Control of CAR T Cells to Potentiate Solid Tumor Immunotherapy
No full textChimeric antigen receptor (CAR) T cell therapy has revolutionized the treatment of several hematological malignancies, achieving durable remissions in patients with B cell cancers. However, its efficacy against solid tumors remains limited due to challenges such as antigen heterogeneity, an immunosuppressive tumor microenvironment (TME), and the scarcity of tumor-specific antigens (TSAs). These obstacles not only impede T cell infiltration, activation, and persistence within tumors but also increase the risk of off-tumor toxicity due to the expression of tumor-associated antigens (TAAs) by healthy tissues. Addressing these barriers is crucial to unlock the full potential of CAR T cell therapy for solid tumor treatment. This thesis focuses on developing tumor-localized strategies to potentiate CAR T cell-mediated immunity against solid tumors.
I first develop thermal gene switches that enable remote control of gene expression in primary human and murine T cells in response to mild hyperthermia. By integratingPh.D.Biomedical Engineerin
AI-Driven Design of Chemically Recyclable Polymers to Replace Commodity Plastics
No full textAddressing the global plastic waste crisis requires a new paradigm of polymeric material that can be depolymerized back to monomer, enabling true chemical recycling. Polymers synthesized via Ring Opening Polymerization (ROP) have shown promise in the fact that they tend to have the necessary thermodynamics to be depolymerizable but lack the mechanical and thermal robustness needed for commercial adoption. This challenge provides an ideal opportunity for AI-driven design to develop such sustainable materials. Herein multiple machine learning (ML) models for relevant polymer properties work in tandem with generative algorithms to optimize across various necessary objectives for creating industry relevant and sustainable polymers. One crucial property in determining the depolymerizability tendencies of polymers is the change in enthalpy (∆H) of polymerization. To handle this property, a ML algorithm to predict ∆H, that utilizes both experimental and ab initio data for enhanced accuracy, has successfully been developed, and continues to be improved so that polymers can efficiently be screened for the potential to be depolymerizable. In addition, current mechanical and thermal ML polymer property predictors have also been retrained and improved to better account for the ROP chemical space. Moving forward, this work identifies robust screening criteria to identify recyclable polymers with the potential to replace conventional food packaging plastics such as polyethylene terephthalate (PET), high-density polyethylene (HDPE), polystyrene (PS), and polypropylene (PP). These criteria are then be put to action using two generative algorithms, Virtual Forward Synthesis (VFS) and a Genetic Algorithm (GA) to screen through millions of hypothetical polymers synthesized via ring opening polymerization (ROP polymers). VFS screens commercially available monomers to discover promising ROP polymers that can be synthesized today, while the GA looks to the future to discover new potential polymers, pushing the boundaries of truly recyclable plastics. Close collaborations with experimentalists to create the most promising polymers from this work have been in place and it is the true goal that recommended polymers from this work result in tangible progress in the creation of sustainable plastics for a circular economy
Control of Haptic Motors for Intuitive Touch Sensation
No full textHaptic signals in existing technologies are limited by profile designs and by the lack of testing for user interpretation. To develop new profile designs and a baseline for standardizing user testing, this dissertation provides the development of: wristband apparatuses, an accurate linear resonant actuator (LRA) system model that captures the electromagnetic effects, and command-shaping control implementation in profile designs. In addition, several user tests are conducted to understand interpretation of profile designs. Finally, a machine-learning-based haptic profile design analysis tool is developed and trained on profile features and the results of the user studies to provide insight into profile interpretation. The outcomes of this dissertation work provide a basis for other haptics researchers to develop intuitive profile designs for touch sensation