Environmental and Occupational Health Sciences Institute
Rutgers University Community RepositoryNot a member yet
58345 research outputs found
Sort by
Context-aware entity grounding using open vocabulary 3d scene graphs
No full textWe present an Open-Vocabulary 3D Scene Graph (OVSG), a formal framework for grounding a variety of entities, such as object instances, agents, and regions, with free-form text-based queries. Unlike conventional semantic-based object localization approaches, our system facilitates context-aware entity localization, allowing for queries such as "pick up a cup on a kitchen table" or "navigate to a sofa on which someone is sitting". In contrast to existing research on 3D scene graphs, OVSG supports free-form text input and open-vocabulary querying. Through a series of comparative experiments using the ScanNet dataset and a self-collected dataset, we demonstrate that our proposed approach significantly surpasses the performance of previous semantic-based localization techniques. Moreover, we highlight the practical application of OVSG in real-world robot navigation and manipulation experiments. The code and dataset used for evaluation can be found at https://github.com/changhaonan/OVSG.M.S.Includes bibliographical reference
Atomistically informed void nucleation in ductile fracture modeling
No full textDuctile structural metals fail through nucleation, growth, and coalescence of voids.While the basic mechanisms of this process have been understood for many decades, a quantitative theory for predicting the void nucleation rate is lacking.
In this thesis, molecular dynamics (MD) simulations of spherical particle debonding within an aluminum matrix are used to identify the microscale mechanisms of nucleation, develop governing equations for the nucleation rate, and then construct a nucleation constitutive model suitable for implementation in the finite element method. First, simulations were conducted to examine the early stages of void formation and nanocrack growth at the particle-matrix interface across a range of temperature and hydrostatic stress values. Crack growth without dislocation activity was identified to obey lattice-trapped fracture dynamics, obeying a thermally activated rate law with an Arrhenius activation energy of and an activation area of .
%In constant, crack growth in the presence of dislocation activity was significantly faster than lattice trapped fracture.
Next, a comprehensive set of several hundred MD simulations exploring stress states spanning a range of stress triaxialities and Lode parameter values was used to determine the critical conditions for void nucleation.
%Simulation results were compiled and mined in order to reveal the governing physics for void nucleation.
Void nucleation is observed when the local normal stress at the matrix-particle interface reaches a critical value of 8.14 GPa.
When plastic deformation occurs in the matrix before nucleation, additional plastic stress concentrations are observed due to plastic strain accumulation around the particle.
The findings reveal that this plastic stress concentration factor increases roughly linearly with equivalent plastic strain, independent of stress triaxiality or Lode parameter.
These results contrast with existing literature, showing a much stronger influence of plastic strain than predicted by continuum plasticity theory.
The discrepancy is attributed to nucleation being driven by localized stress hotspots resulting from heterogeneous plastic strain, whereas continuum theory is based on homogenized plastic strain fields. Using the results from these studies, a new nucleation model is developed that is suitable for ductile fracture modeling.%The well-established Gurson-Tvergaard-Needleman model is employed with a custom nucleation model that captures the influence of stress state, plastic strain, and particle size on the nucleation rate.
Unlike traditional nucleation models, our model takes as input the size distribution of the particle population in the material and material properties for the crack nucleation rate at the particle interfaces.
The behavior of the model is explored in a MATLAB script and a preliminary implementation in the ABAQUS finite element code.
Overall, the study concludes that void nucleation is predominantly stress-driven and significantly influenced by plastic strain.Ph.D.Includes bibliographical reference
Essays on semiparametric binary models with endogeneity applied in health economics
No full textThis thesis develops an innovative control function estimator for a binary response model that includes a continuous endogenous regressor in the absence of exclusion restrictions. The estimator is designed to handle empirical challenges associated with a binary outcome variable and a continuous endogenous regressor when exclusion restrictions are not feasible. The application of this estimator is to study the causal link between education and health. The empirical results highlight the heterogeneous effect of education on health and health-related behaviors. Chapter 1 addresses the endogeneity issue theoretically in a binary model without feasible exclusion restrictions. This chapter introduces parametric and semiparametric control function estimators. The main idea of this approach is to utilize heteroscedasticity to construct a control variable to address the endogeneity issue. This chapter establishes the large sample properties of the proposed estimator. In Monte-Carlo simulations, it performs quite well in finite samples. Chapter 2 applies the model proposed in Chapter 1 to solve the empirical challenge posed by the undesirability of using the commonly employed instrumental variable, compulsory schooling laws, in studying the educational effect on health. Using the Health and Retirement Study (HRS) data, the semiparametric estimation method is employed to estimate the heterogeneous effect of education on health and health behaviors without introducing an instrumental variable for the endogenous variable, education. The empirical results indicate that the effect of education on health varies based on the initial health status of individuals, age, and the level of education attained. These findings provide insights into the mechanisms through which education influences health outcomes and inform targeted interventions to enhance population health.Ph.D.Includes bibliographical reference
Exploring the effect of protostellar feedback on star formation and molecular cloud dynamics
No full textStar formation is a fundamental building block of astrophysics and plays an important role in processes ranging from galaxy formation and evolution to the formation of proto-planetary disks. A variety of cloud-scale physical mechanisms contribute to the process of star formation, including: collapse due to self-gravity, supersonic turbulence, magnetic fields, and stellar feedback. In this dissertation, I present my work using 3D magnetohydrodynamical simulations of star-forming regions and analytic models to study star formation. I first present my work using a suite of turbulent box simulations to study the shape and evolution of the density probability distribution function (PDF; Appel et al. 2022). We explore how the shape of the PDF evolves with the inclusion of different physical processes and as a function of time. In agreement with the model proposed in Burkhart (2018), we find that the inclusion of self-gravity produces a power-law tail at the high-density end of the PDF, while supersonic turbulence produces a lognormal shape near the mean density. We also find that the inclusion of protostellar jets produces non-lognormal, time-varying features at the low-density end of the PDF. We find that the amount of mass in the power-law tail portion of the PDF is constant in time, suggesting that, as stars form, the dense gas is rapidly replenished from the collapsing diffuse gas. I then present my work further exploring the gas dynamics in star-forming regions using the expansion and compression rates of the gas (Appel et al. 2023). We explore how the gas dynamics varies with the inclusion of different physical processes. We find that the inclusion of supersonic turbulence increases both the compression and expansion rates, resulting in an approximate equilibrium between compressing and expanding gas near the mean density. We find that self-gravity dominates the gas dynamics at the highest densities and that the inclusion of protostellar jets produces rapidly expanding and compressing gas at low densities. We also compare the gas dynamics to the star formation rate (SFR) using the net gas mass flux. We find that the net gas mass flux matches the SFR at densities above where the PDF begins to form a power-law distribution but below the sink formation threshold. This suggests that the gas dynamics at this density plays a role in setting the SFR. Finally, I present my work with Torch (Appel et al. in prep.), a numerical framework optimized for simulating the formation and evolution of stellar clusters. Previous work with Torch has explored a variety of questions related to star formation and includes the impact of several modes of stellar feedback, including stellar winds, radiative feedback, and supernovae. However, previous work with Torch has not included the impact of protostellar jets, which has been demonstrated to play an important role in star formation. We have now implemented a new module within the Torch framework for including the impact of protostellar jets. We present this new module as well as preliminary results from the first Torch runs to include jets. We find that the inclusion of protostellar jets slows star formation even in star-forming regions up to 2 X 10^4 M_sun and that our lower mass clouds (5 X 10^3 M_sun) are sensitive to the star formation prescription and chosen jet parameters.Ph.D.Includes bibliographical reference
Reconstructing humans and animals from visual data
No full textGiven the large number of applications that require a 3D model as input like augmented and virtual reality (AR/VR), film production, the ability to obtain precise 3D reconstructions for humans and animals is more relevant than ever. Over the past decade, significant advancements have been made in reconstructing humans in 3D from images and videos, largely due to the availability of accurate statistical parametric models and extensive training datasets with 3D and 2D annotations. However, despite these successes, current 3D human reconstruction systems still lag behind traditional 2D tasks, such as segmentation and pose estimation, in terms of accuracy and robustness. Existing approaches for reconstructing animals in 3D are not as accurate and successful compared to methods that reconstruct humans. Although humans and animals are both articulated object categories and, in theory, similar approaches could be employed for their reconstruction, the lack of extensive and accurate 3D datasets hampers progress in this field. It is impractical to bring a large number of animals into a lab environment for scanning, and thus existing approaches account for this limitation by creating systems that can be trained with with weaker forms of supervision. The goal of this thesis is to present our contributions towards reconstructing 3D humans and animals in 3D from images and videos. We propose distinct solutions to address the unique limitations of existing methods for each category. First, we will explore how detected features, such as 2D keypoints from other systems, can be leveraged to enhance 3D human reconstruction. Next, recognizing the limited availability of annotated data for animals, we will introduce an approach that effectively utilizes unlabeled images from the web to improve 3D animal reconstruction.Ph.D.Includes bibliographical reference
Phenol hydrodeoxygenation over ru(0001), effect of oxygen coverage on direct deoxygenation and water splitting
No full textLignin derived bio-oil produced through pyrolytic processes contains high levels of oxygenated fractions, necessitating treatment to achieve high-quality fuel. This treatment typically involves hydrotreatment over transition metal catalysts such as Pt, Pd, and Ru. Ruthenium has high oxophilicity, which enhances deoxygenation and influences product selectivity. Phenolics, a significant fraction of bio-oil, pose complex reaction mechanisms due to varied carbon numbers and functional groups. Hydrodeoxygenation (HDO) of phenolics over Ru catalysts shows varied product selectivity based on controlling factors, including solvent effects, catalyst supports, dispersion, and the incorporation of metallic and non-metallic components. Phenol serves as a model compound for studying C-O bond cleavage mechanisms. Using density functional theory (DFT) calculations, we investigate the activation energy barriers involved in deoxygenation and hydrogenation reactions over Ru(0001) surface.Three pathways including direct deoxygenation, tautomerization, and ring hydrogenation are analyzed to determine the most energetically favorable route for converting phenol to hydrocarbons and oxygenates, including benzene, cyclohexene, cyclohexane, cyclohexanone, and cyclohexanol. Tautomerization to cyclohexanone shows the lowest overall barrier, with microkinetic modeling predicting benzene formation via direct deoxygenation as the dominant pathway, with cyclohexanone as a secondary product. A high oxygen coverage is also predicted for the phenol, water and hydrogen system with bare surface energetics, limiting active site availability and phenol consumption.
Since the surface oxygen is removed as water, the equilibrium oxygen coverage over the Ru(0001) surface under reducing environments is investigated using water as an oxygen source. Using a large 4×4 slab model, the surface interactions and reaction energetics indicate that Ru can support up to 1/4 monolayer (ML) of oxygen, with an equilibrium coverage of ~0.2 ML at 500 °C. Stronger surface interactions beyond this coverage create a new regime of adsorption and reaction energetics. A kinetic Monte-Carlo model for water dissociation also predicts a limiting oxygen coverage of 1/4 ML.
Microkinetic models for phenol HDO based on variable oxygen coverage energetics, focusing on the direct deoxygenation pathway, show increasing production up to 3/16 ML beyond which a sharp decline is observed due to limiting reactant adsorption and reaction kinetics. Additionally, hydroxyl-mediated pathways offer a low barrier pathway for phenoxy removal from the surface, limiting benzene production. This study uses computational approaches to understand the energetic changes induced by spectator oxygen atoms, focusing on both adsorption characteristics as well as reaction barriers.Ph.D.Includes bibliographical reference
Enhancing Safety and Mobility with Predictive Modeling and Control in Uncertain Traffic Environments
No full textThis dissertation advances the field of autonomous vehicle navigation by developing sophisticated predictive modeling and control methods that address the inherent uncertainties encountered by Connected and Automated Vehicles (CAVs) operating in mixed traffic conditions and on unmarked roads. These uncertainties include the variability in driver behaviors and emotions, the unpredictability of human drivers' actions, and the challenges posed by poor or missing road markings. With the rise of autonomous technologies, effectively managing these uncertainties is critical for enhancing traffic safety and ensuring efficient mobility in complex urban environments.The core of this research is built around three innovative models, each designed to tackle specific challenges associated with the uncertain elements of autonomous driving. The first model, the Enhanced Safe Route Mapping (ESRM), leverages a combination of real-time and historical data to dynamically predict traffic risks and generate actionable insights for route optimization. This model uses advanced machine learning techniques, including Light Gradient Boosting Machines (LightGBM) and Fuzzy Logic, to refine its predictions and adjust to changing traffic patterns. The second model, the Cooperative Hybrid Automaton Model (CHAM), is designed to manage the state transitions of automated vehicles in complex road scenarios. It addresses the unpredictable interactions between CAVs and legacy vehicles that do not communicate their intentions. By integrating Model Predictive Control (MPC) with a Bayesian Long Short-Term Memory (BLSTM) network, CHAM enhances the CAVs' abilities to anticipate and adapt to the driving behaviors of human-operated vehicles, thereby reducing the risk of accidents and improving overall traffic flow. The third model introduces a novel diffusion-based motion predictor within an Active Inference Framework (AIF). This model is specifically designed for navigating environments without clear road markings, using probabilistic methods to predict a range of possible vehicle actions and determine the most advantageous maneuvers under various scenarios. The integration of this model with active inference principles allows for continuous adaptation to new environmental cues, thereby minimizing navigational errors and enhancing the decision-making process under uncertainty.
Extensive simulations and real-world tests have shown that these models improve the safety and operational efficiency of CAVs in mixed-traffic environments and on poorly marked roads. The results demonstrate that these advanced models provide practical and reliable navigation solutions. Future research will aim to refine these models for better adaptability across different regions and environmental conditions. This ongoing work seeks to reduce the operational uncertainty of autonomous vehicles, supporting their broader adoption and contributing to safer, more efficient urban mobility.Ph.D.Includes bibliographical referencesIncludes vit
Genetic variation affecting fitness in the Eastern oyster (Crassostrea Virginica)
No full textGenetic variation or genetic diversity plays a crucial role in the adaptability and survival of oyster populations. Comprehending its influence on oyster fitness helps us understand oyster adaptability and contributes to genetic improvement of oysters for aquaculture. This dissertation, comprised of five chapters, delves into the significance and practical applications of genetic variation in the context of oyster fitness and adaptability.Based on the foundation of genetic variation's crucial role in oyster adaptability and survival, this dissertation explores the phenomenon of heterosis and its implications for genetic diversity within oyster populations. Chapter 1 delves into the performance of inbred and hybrid oysters from the D larvae to adult stage, providing concrete evidence of heterosis in terms of oyster survival and growth. The utilization of the recently developed high density single nucleotide polymorphism (SNP) array provides insight into the importance of heterosis and genetic diversity in enhancing oyster performance including survival and growth, contributing to our understanding of heterosis in oysters. Chapter 2 addresses genetic load in the eastern oyster by analyzing genetic markers showing segregation distortion (SD) during the larval stage. This analysis reveals why oysters are sensitive to inbreeding and suffer from inbreeding depression. The high proportion of markers with SD suggests that balancing selection is the main contributing factor to the genetic load and genetic diversity. Furthermore, the discovery of high levels of null alleles in the eastern oyster genome offers insights into other forms of genetic variation in oysters. In Chapter 3, the focus shifts to understanding genetic mechanisms of growth and heterosis through transcriptome analysis. Through gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, distinctions in growth-related physiological strategies among different size groups are illuminated. The identification of candidate genes associated with shell formation and stress response suggest a relationship with growth. Comparing hybrid and inbred transcriptome data suggests hybrids have a more sophisticated regulatory network for optimized stress response. This chapter argues that the observed heterosis in oyster growth results from a combination of additive and non-additive genetic effects. Chapter 4 identifies candidate loci and genes associated with field survival by studying the genetic changes after field mortalities. The findings suggest that field survival is a polygenic trait influenced by many genes related to disease resistance, growth, and thermal tolerance. This study provides a rich source of genetic data for possible genetic improvement of oysters. Finally, Chapter 5 conducts genomic selection (GS) to improve dermo resistance by using genome-wide genotype data. The performance of the F1 generation proves the effectiveness of GS in improving dermo resistance. And it is the first report on the successful application of high-density SNP array and GS in improving dermo resistance in the eastern oyster.
In conclusion, this dissertation provides a comprehensive examination of the critical role of genetic variation in the adaptability and survival of eastern oysters. Through detailed investigations spanning from inbred and hybrid performance analyses to high genetic load assessments, to the exploration of genetic mechanisms underlying growth and heterosis, this research highlights the significance of maintaining and promoting genetic diversity within oyster populations. The utilization of high-density SNP arrays and transcriptome data has not only shed light on the phenomenon of heterosis but also contributed to genetic improvement practices in oyster aquaculture. By identifying candidate loci and genes associated with key traits such as field survival and growth, this study provides valuable insights into genetic architecture of production traits, potentially useful for the genetic improvement of oysters. This research demonstrates that GS is effective for improving polygenic traits such as dermo resistance. These findings emphasize that genetic variation is crucial for the adaptability of oysters, and understanding this diversity is critical for oyster aquaculture, breeding and conservation.Ph.D.Includes bibliographical reference
Structures of feeling in Protestant allegory, 1590-1679
No full textThis dissertation explores early modern allegory’s cultural pragmatics by bridging allegory studies with thecultural history of emotion. I argue that Protestant allegory composed between 1590 and 1679 is characterized by an
applied or practical orientation towards the relationship of Reformed hermeneutics and lived experience. This
orientation develops through structural identification between the affective life of an allegorical agent and the moral
life of the reader, which facilitates Protestant allegory’s translation of theological doctrine into ethical discipline.
While Romanticist aesthetic theory had identified allegory with personification, the historical roots of this
identification can be traced in late sixteenth and seventeenth century theories of the allegorical epic. In showing how
these theories involved Reformed concepts of allegorical reading to buttress the epic’s pretensions to the conceptual
truth of natural history, I demonstrate how Protestant allegory’s structural identification of reader and allegorical
agent – what I call “autoreferentiality” – emerges as a characteristic of poetic expression conditioned by the
religious beliefs, print technologies, and structures of feeling of Puritan reading and interpretive culture, with its
disciplinary orientation towards the relationship of hermeneutics and lived experience. Drawing from Marxist
aesthetics, I extend Raymond Williams’s concept of the “structure of feeling” to inform a project methodology
which brings current critical theories of affectivity, sophisticated philological and hermeneutic techniques, as well as
rigorous scholarly and archival research, to consideration of the relationship between personification and the reader
in Protestant allegorical poetics. By doing so, I hope to understand how Reformed Christianity influenced the
development of novel forms of literary expression in the period under consideration, and in their turn, the ways in
which early modern Protestant allegorists such as Edmund Spenser, John Milton, Lucy Hutchinson, and John
Bunyan wielded poetic forms of their own making to refigure the prevailing theological orthodoxies of their time.Ph.D.Includes bibliographical reference
Elucidating early events in nitrogen mustard-induced pulmonary toxicity using precision cut lung slices
No full textNitrogen mustard (NM; mechlorethamine) is a cytotoxic vesicant known to cause acute lung injury which can progress to chronic disease. Due to the complex nature of NM injury, it has been difficult to analyze early responses of resident lung cells that initiate and contribute to disease progression. To investigate this, we developed a model of acute NM toxicity using murine precision cut lung slices (PCLS) in Aim 1. PCLS were exposed to NM (1-100 μM) for 0.5-3 h and analyzed 1 and 3 d later. NM caused a dose-dependent increase in cytotoxicity and a reduction in metabolic activity, as measured by LDH release and WST-1 activity, respectively. Optimal responses were observed with 50 μM NM after 1 h and these conditions were used in further experiments. Analysis of PCLS bioenergetics using an Agilent Seahorse showed that NM impaired both glycolytic activity and mitochondrial respiration. This was associated with injury to the bronchial epithelium and a reduction in methacholine-induced airway contraction. NM was also found to cause DNA damage in bronchial epithelial cells, as measured by expression of ɣ-H2AX, and to induce oxidative stress, as evidenced by declines in glutathione, increased presence of peroxynitrite, and upregulation of the antioxidant catalase. Cleaved caspase-3 was also detected in airway smooth muscle cells indicating apoptotic cell death. In Aims 2 and 3, we focused on discerning the contribution of oxidative stress and inflammatory mediator production by resident lung cells to NM toxicity. To assess the impact of oxidative stress, PCLS were pretreated with the glutathione precursor, N-acetylcysteine (NAC). Inhibiting oxidative stress with NAC mitigated functional and metabolic impairments induced by NM but had minimal effects on cytotoxicity and structural alterations. We speculate that the insensitivity of these changes to NAC was due to S-nitrosylation of proteins required for mitochondrial respiration. In Aim 3, the production of inflammatory proteins by resident cells following NM exposure was evaluated. Expression of cyclooxygenase-2 (COX-2) and toll like receptor 4 (TLR), a pattern recognition receptor, was upregulated in bronchial and alveolar epithelium in response to NM-induced injury suggesting a potential role of these mediators in toxicity. To model pre-existing chronic inflammation, PCLS were generated from surfactant protein D (SP-D) knockout (KO) mice, which exhibit an emphysematous-like phenotype. Loss of SP-D had no effect on markers of lung injury. These findings suggest that continuous infiltration of inflammatory cells is key to injury progression in this model. Characterizing early events in NM toxicity is key in identifying therapeutic targets for the development of efficacious countermeasures.Ph.D.Includes bibliographical reference