50107 research outputs found
Sort by
3D-Scanning Inspection Techniques and Analysis of Naturally Corroded Steel Bridge Beam Ends
No full textBridge inspection and maintenance is an ever present necessity to ensure the reliability of the structure and safety of the public. New England is home to more than 9000 steel bridges which are susceptible to deterioration in the form of corrosion. The ends of the girders are the most common areas of the superstructure that experience corrosion due to water and ice melt chemicals leaking from the deck through faulty expansion joints.
This deterioration has lead to the need for extensive repairs, bridge closure and replacement, and even bridge failures which generate a hazard to the public and extensive costs. This necessitates consistent and rigorous inspection to be conducted on bridges, with particular emphasis on those that exhibit significant deterioration. In the case of measuring deterioration due to corrosion, bridge inspectors often depend on handheld measurement tools and visual assessments to evaluate the degradation and the residual capacity of corroded girder ends. Issues like data collection, accessibility, and time constraints pose challenges for inspectors to conduct these measurements. In general, a limited number of point measurements (ranging from one to a few points) are used to represent the remaining thickness of a corroded girder web, potentially resulting in significant overestimation or underestimation of the remaining capacity at the corroded girder ends.
This dissertation seeks to develop a more streamlined process from start to finish for evaluating corrosion, the consequent degradation, and the remaining strength of corroded steel girder ends by employing advanced 3D scanning techniques. Inspection data was meticulously collected and analyzed from reports across all New England States. After an in-depth review of these reports, corroded girder specimens were chosen for documentation and load testing. The use of 3D scanning advanced existing procedures and led to the development of a new protocol for documenting corroded girders, along with a novel ideal contour selection method. Multiple scanners were utilized to establish this protocol, and the specimens used in the load tests covered various states, beam types, and deterioration-related damages. Ultimately, the integration of scanning data and load testing outcomes was employed to verify the ideal contour selection method through finite element analysis, as well as to confirm the current capacity estimation equations developed by Tzortzinis [85,90,92] . The final chapter of this dissertation aims to mitigate uncertainty generated by surface delamination in the inspection process. To accurately record the actual extent of damage caused by corrosion and delamination on beam ends, it is essential to clean these areas. To achieve precise documentation, 3D scanning was utilized both before and after the cleaning of corroded girder ends. At present, no existing model can predict remaining structural steel using data from uncleaned corroded girders. In this study, data from both uncleaned and cleaned corroded girder ends were compared to identify corrosion patterns and develop a predictive model.New England Transportation Consortium (NETC)
Massachusetts Department of Transportation (MassDOT)
Federal Highway Administration (FHWA)Doctor of Philosophy (Ph.D.
Behavioral flexibility of a declining songbird, the Wood Thrush (Hylocichla mustelina), in suburban forests
No full textAs urbanization increases, so does the need to understand the mechanisms behind urban biodiversity loss and alternatively, population persistence, as wildlife adapt to novel ecosystems. Wildlife conservation has historically focused on relatively pristine natural areas and large tracts of contiguous habitat, and while those areas are certainly valuable, they are not alone in supporting native wildlife. Smaller, more fragmented, and disturbed natural habitats still support many native wildlife species, including birds. Globally, birds and their common prey, insects, are in decline. My dissertation sought to assess whether suburban forest fragments can support forest nesting species, focusing on the Wood Thrush (Hylocichla mustelina), a declining Neotropical migrant songbird species of conservation concern. First, I investigated the invertebrate communities in forest leaf litter between suburban forest fragments and a larger contiguous forest (hereafter referred to as rural), as Wood Thrush are primarily ground-foraging insectivores. Using mixed effects models and canonical correspondence analysis, I found that suburban forest fragments have higher diversity and biomass of invertebrates than rural forests, likely due to the prevalence of non-native decomposers. In my second research chapter, I examined whether parental behavior at the nest, specifically provisioning, differed between suburban and rural forest, and found that suburban nesting Wood Thrush are provisioning their young at higher rates than their rural counterparts. In my final research chapter, I compared nestling condition between forest types using the scaled mass index method, as well as ectoparasite loads of hematophagous bird blowflies, genus Protocalliphora. Neither ectoparasitism nor urbanization influenced nestling condition. My research indicates that suburban forest fragments are indeed supporting breeding Wood Thrush, although adults nesting in suburban fragments are working harder to raise young of similar condition. Future research on the potential costs of increased provisioning on the adults themselves is necessary, as increased seasonal parental investment can impact lifetime reproductive success. In addition, dietary comparisons between forest types would further assess Wood Thrush behavioral flexibility in suburban habitats, as suburban parents may be provisioning with invertebrates of lesser nutritional quality.The National Science Foundation Graduate Research Fellowship Program, The University of Massachusetts Graduate School Dissertation Research Grant, The Organismic and Evolutionary Biology Research Grant, The Bradford G. Blodget Scholarship Fund for Ornithological StudiesDoctor of Philosophy (Ph.D.
Combinatorial problems in spaces of matrices
No full textThis thesis is comprised of four projects. The first project concerns q-analogues of the classical rook and hit numbers. These q-rook and q-hit numbers are defined in terms of the number of matrices with fixed rank and support contained in a fixed set over a finite field of size q. We show that their residues modulo small powers of q are polynomial in q. We find a formula for these polynomials and use it to prove a positivity result for the (q − 1) coefficient of the q-hit number.
The second project concerns the signs of principal minors of a real symmetric matrix. These sign patterns are equivalent to uniform oriented Lagrangian matroids. We study their structure, symmetries, and asymptotics, proving that almost all of them are not representable by real symmetric matrices. We offer several conjectures and experimental results concerning representable sign patterns and the topology of their representation spaces.
The third project concerns chromatic symmetric functions related to the Stanley-Stembridge conjecture. We show that the allowable coloring weights for indifference graphs of Dyck paths are the lattice points of a permutahedron P_λ, and we give a formula for the dominant weight λ. Furthermore, we prove that chromatic symmetric functions for abelian Dyck paths are Lorenzian. We extend our results on the Newton polytope to incomparability graphs of (3+1)-free posets, and give a number of conjectures and results stemming from our work, including results on the complexity of computing the coefficients and relations with the ζ map from diagonal harmonics.
The fourth project concerns the regular and nilpotent Hessenberg variety of shape λ associated to the path graph. We give a combinatorial model in terms of rook walks
for the cells in an affine paving of these varieties which is a special case of Tymoczko’s filling rule. We note a mysterious phenomenon in the Poincaré polynomial of the
nilpotent Hessenberg variety of shape (n, n), and we present several questions with related experimental data.Doctor of Philosophy (Ph.D.
Establishing the Mechanical and Biological Influences of Microorganism Adhesion to Biomaterials
No full textThe use of polymer-based medical devices has reduced material costs, improved quality of care, and increased device biocompatibility. However, polymer devices are susceptible to fouling by foreign bacteria, leading to fatal hospital-acquired infections. Antibiotic-resistant bacterial strains have created the need to focus on decreasing initial adhesion of bacteria to the polymer surface. Previous research has shown that both material stiffness and chemistry impact adhesion of structurally different bacteria to polymeric surfaces. Gram-negative Escherichia coli and Gram-positive Staphylococcus aureus demonstrated greater adhesion on stiffer, hydrophilic poly(ethylene glycol) dimethacrylate (PEGDMA). These findings established a core trifecta: mechanical properties, material chemistry and biological components impact bacterial adhesion to biomaterials.
I will report how the attachment of E. coli and S. aureus are impacted by the material stiffness of hydrophobic polydimethylsiloxane (PDMS) gels. More bacteria adhered to softer PDMS gels, which was opposite of the observed trend on PEGDMA hydrogels. Next, by spin-coating thin PDMS gels, I investigated if their stiffness at different thicknesses (10 µm, 35 µm, 100 µm) impacted adhesion. It was demonstrated that as thickness decreased, bacterial adhesion increased, same as the trend observed on PEGDMA hydrogels. Next, biomaterials were designed to mimic usage in clinical settings, symbolizing PEG as an antifouling coating atop a PDMS catheter. These “layered” gels, featured PEGDMA hydrogels (15-60 µm thick) of different stiffnesses deposited onto spin-coated PDMS gels of various stiffnesses and a consistent thickness (100 µm). The attachment of E. coli and S. aureus were assessed on these layered gels to further elucidate the role of surface chemistry.
Finally, a collaboration used genetic analysis and CRISPRi tools to identify and repress important genetic targets involved in E. coli adhesion, and adhesion assays demonstrated that their adhesion phenotypes can be controlled. The results of this dissertation have elucidated important findings regarding how surface mechanics, surface chemistry and biology each are important pillars of understanding how bacteria interact with surfaces, and how E. coli adhesion-associated genes have been identified and repressed with CRISPRi systems. These results hold potential to guide the design of antifouling polymeric biomaterial devices and CRISPRi treatments to prevent hospital acquired infections.NSF Grant DMR-1904901
UMass Spaulding-Smith FellowshipDoctor of Philosophy (Ph.D.)2026-09-0
kDNA Catastrophe: A Morphological and Transcriptional Response to Mitochondrial DNA Loss in Trypanosoma brucei
No full textAfrican sleeping sickness is a fatal neglected tropical disease if left untreated. Trypanosoma brucei, the causative agent, contains unique mitochondrial DNA that is a network of catenated minicircle and maxicircle DNA molecules called kinetoplast DNA (kDNA). Replication and maintenance of the kDNA network is essential to parasite survival and lifecycle completion; thus identifying kDNA replication proteins as potential drug targets. Coordination of three independently essential DNA polymerases (POLIB, POLIC, and POLID) is required for network replication. Simultaneous knockdown of POLIB and POLID leads to rapid loss of the kDNA, as well as a unique growth defect where T. brucei experiences a growth arrest. During dual polymerase knockdown some parasites differentiatied from procyclic to epimastigote forms, based on basal body positioning seen via immunofluorescence microscopy.
This thesis investigates the question: does rapid kDNA loss cause procyclic form T. brucei to lose its identity and differentiate into the epimastigote life cycle stage? To enable proteomic analyses, we used a reporter system designed to express eGFP in tandem with the epimastigote surface protein: BARP. When this reporter system did not express eGFP during rapid kDNA loss, transcriptomal and morphological changes were also evaluated. Transcript abundance of BARP and other regulatory life cycle proteins changed according to early life cycle progression in the tsetse fly proventriculus. As did morphological markers like length and basal body positioning. However the parasites did not complete a transition from the procyclic to epimastigote form.Donald P. Reed Legacy FundMaster of Science (M.S.
DEEP LEARNING FOR DISCRETE EVENT SIMULATIONS
No full textDiscrete event simulation is a widely used technique for modeling and analyzing complex stochastic systems, especially in engineering, logistics, and healthcare. Adoption of deep learning in discrete event simulation has been slow, however, despite recent advances. This thesis explores how deep learning can solve existing problems and uncover new applications for simulation.
We first introduce Neural Input Modeling (NIM), a generative-neural-network framework that automates simulation input modeling. NIM addresses the challenge of fitting stochastic input-process models to data, especially for non-experts. Its core architecture, NIM-VL, uses a variational-autoencoder (VAE) with Long Short-Term Memory (LSTM) components to learn and reproduce input stochastic-process data while capturing temporal dependencies. NIM can model multivariate processes, handle nonstationary sequences, and perform conditional simulations through conditional NIM (CNIM), e.g., generating stochastic sequences of emergency calls given specified weather conditions. Experimental results show that NIM and CNIM effectively automate input modeling, reducing barriers for non-experts and enabling more accurate simulations.
We then develop Graphical Metamodels (GMMs), a new class of simulation metamodels that approximate complex simulation behavior via simple mathematical functions for faster analysis and optimization. Our approach incorporates graph neural networks (GrNNs) to account for the graph structure of simulation models, treating it as an input parameter alongside traditional numerical inputs. We introduce generative graph metamodels (GGMMs) that combine GrNNs with generative neural networks to produce a range of summary statistics and sequences of samples that mimic dynamic outputs. This innovation supports flexible simulation-based prediction and optimization under uncertainty, especially in settings needing quick results. We also propose HiLo, a novel metamodel-training method that reduces computational costs without compromising accuracy. Empirical results demonstrate that these methods can accurately model a wide range of simulation scenarios, paving the way for more efficient and flexible simulation studies. Finally, we provide theoretical results linking GMMs to classical Gaussian-process-based metamodels, which allow us to formally compare HiLo's performance against traditional metamodeling methods.
Finally, we propose approaches to simulation optimization problems involving GMMs. GMMs lead to hybrid optimization problems where both continuous numerical parameters, like work rates, and the graph structure, such as warehouse layouts, need to be jointly optimized. We consider an online optimization setting, where, after training a GrNN, we optimize numerical and structural inputs to identify optimal configurations in real-time. We develop a scalable heuristic optimization method based on Monte Carlo Tree Search. We then derive an exact mixed-integer linear programming (MILP) formulation for the online optimization, enabling provably optimal solutions, and accelerate its solution with a custom parallel affine-arithmetic branch-and-bound solver. Finally, we extend traditional offline optimization to hybrid discrete–continuous spaces by using a Bayesian optimization framework augmented with the Neural Tangent Kernel of a GMM.Doctor of Philosophy (Ph.D.
From Boundary Spanning to Role Blurring: Reconfiguring Professional Practice in IPOs
No full textProfessional roles in initial public offerings (IPOs) are formally distinct, but the preparation process brings together auditors, company executives, investment bankers, and attorneys into a single collaborative team. In this setting, professionals routinely cross formal role boundaries: auditors weigh in on messaging strategy, lawyers and consultants support financial verification, and CFOs participate in book-building. These behaviors are not exceptional; they are expected, repeated, and justified as necessary to the success of the transaction. Drawing on interviews with 19 experienced team members, this study finds that the IPO team functions as a structurally unified group, a configuration that weakens internal checks and reshapes accountability. Normalized cross-boundary behaviors give rise to role blurring — a cumulative and internalized redefinition of professional responsibility. This process produces two patterned consequences: expert overreach and performative accountability. The study develops a process model that clarifies the antecedents, behavioral dimensions, and governance implications of role blurring that occur during the IPO process. In doing so, it contributes to research on disclosure oversight and the evolving dynamics of professional work in capital markets.Doctor of Philosophy (Ph.D.
Evaluating BASICS as an Effective Intervention for Collegiate Cannabis and Prescription Misuse: A Scoping Review
No full textBackground: Prescription and cannabis misuse is a prevalent public health concern afflicting college students, though research on effective interventions remains limited. BASICS (Brief Alcohol Screening and Intervention for College Students) has been widely evaluated as an effective alcohol intervention, though its application to cannabis and prescription misuse remains under-studied.
Objective: To explore the current and evolving landscape of interventions for collegiate cannabis and prescription misuse by focusing on the core components of BASICS: motivational interviewing, personalized feedback, harm reduction, and psychoeducation.
Methods: A systematic review of three databases: Pubmed, Web of Science, and PsycINFO. Applicable studies were included based on inclusion and exclusion criteria.
Results: Findings highlight a growing interest in adapting BASICS to cannabis use, with more limited application to prescription misuse. Several studies utilizing motivational interviewing and harm reduction approaches showed promise in reducing risky use and enhancing student knowledge. However, gaps remain in research specific to prescription sedative and stimulant misuse
Conclusions: BASICS and its components represent a promising foundation for addressing cannabis and prescription drug misuse on college campuses. Further research is needed to evaluate their effectiveness across substance types and to inform the development of targeted, evidence-based interventions for this populationMaster of Public Health (MPH
Investigating the role of let-7/miR-17/92 axis in CD8 T lymphocyte differentiation
No full textCD8 T lymphocytes are essential for adaptive immunity, differentiating into either short-lived effector cells or long-lived memory cells that mediate durable protection. Post-transcriptional regulation by microRNAs plays a pivotal role in these fate decisions. The let-7 family and the miR-17/92 cluster have each been implicated in T cell differentiation, but their regulatory interplay and upstream control remain incompletely defined.
In this study, we generated a novel miR-17/92ZsG reporter mouse to directly visualize endogenous pri-miR-17/92 transcription at the single-cell level. This model revealed dynamic miR-17/92 expression during thymocyte maturation, selective enrichment in regulatory T cells, and broad heterogeneity among conventional and unconventional T cell subsets. Importantly, we identified the NFAT/AP-1 complex as a major inducer of miR-17/92 in activated CD8 T cells.
Using retroviral manipulation and in vitro activation assays, we further defined the functional relationship between let-7 and miR-17/92 in determining CD8 T cell fate. Let-7 overexpression repressed miR-17/92, whereas let-7 deficiency increased miR-17/92 levels during early activation. Deletion of miR-17/92 restored memory precursor formation in let-7–deficient cells, while enforced miR-17/92 expression abrogated the memory-promoting effects of let-7. Mechanistically, miR-17 directly targeted and suppressed the transcription factor Bcl11b, a key regulator of memory T cell development.
Collectively, these findings identify the let-7/miR-17/92/Bcl11b axis as a critical regulatory pathway in CD8 T cell differentiation. The miR-17/92ZsG reporter mouse provides a powerful genetic tool for dissecting the transcriptional control of miR-17/92 cluster across immune contexts. This work advances our understanding of microRNA-mediated regulation of T cell fate and offers new avenues for modulating immune memory.Doctor of Philosophy (Ph.D.)2026-09-0
HARDWARE-SOFTWARE CO-DESIGN OF MEMRISTIVE SYSTEMS FOR EDGE INTELLIGENCE
No full textAdvancements in artificial intelligence (AI) have driven remarkable performance across a wide range of applications. However, the increasing scale and complexity of AI algorithms have introduced significant challenges related to size, weight, power, and cost (SWaP+C) in conventional computing hardware, especially for edge intelligence applications where computing resources are limited. Analog in-memory computing (AIMC) hardware based on memristor devices offers a promising solution by enabling energy-efficient and parallel vector-matrix multiplication (VMM), a core operation in most AI models. As memristive technology evolves, from individual devices and crossbar arrays to integrated chips (IC) and multicore system-on-a-chip (SoC), a critical gap persists in bridging innovations across devices, peripheral circuits, architectures, algorithms, and applications to build energy-efficient and low-latency computing systems for edge intelligence. Addressing this challenge requires a hardware-software co-design approach: developing hardware that meet algorithm- and application-level requirements, while simultaneously optimizing algorithms to exploit the inherent characteristics of memristive hardware.
In this dissertation, we adopt a hardware-software co-design approach to address challenges spanning peripheral circuit development, algorithm designs, and edge intelligence applications. We begin by developing a multichannel time-encoding testing system that integrates peripheral circuits and software modules for efficient characterization of memristive crossbar arrays with varying device stacks and array dimensions. This system enables time-domain VMM and supports rapid prototyping of neural network algorithms, thereby facilitating the development and evaluation of memristive crossbar arrays, which are subsequently integrated into a memristive SoC.
Building on this foundation, we implement a hardware-algorithm co-design framework for hyperdimensional computing (HDC) using AIMC within the memristive SoC for language classification tasks. The co-design approach exploits the intrinsic randomness of memristor devices for data encoding and utilizes single-step analog VMM in memristive crossbar arrays to implement a single-layer perceptron (SLP). The memristive HDC achieves state-of-the-art performance in both simulation and hardware and demonstrates the energy efficiency of memristive hardware for edge intelligence. Additionally, we present optimization techniques that generalize the system for broader applications.
Extending these co-design techniques, we implement a radiofrequency (RF) signal processing system on the memristive SoC. This system comprises signal processing components and neural networks for RF transmitter identification and anomaly detection. The system distributes RF signal spectrum analysis, demodulation, and multilayer perceptron inference across multiple memristive computing cores, enabling low-latency and energy-efficient RF signal processing. This advancement brings intelligence to the edge of wireless communication and lays the foundation for intelligent radio receivers.
Finally, we further extend the memristive system to implement a cognitive radio receiver (CRR) by replacing conventional signal processing components with convolutional and fully connected neural networks. This CRR leverages the energy-efficient neural network models, accelerated by parallel analog VMMs implemented using memristive crossbar arrays within the SoC, to mitigate channel distortion and directly detect digital symbols from analog RF signals. This adaptive and energy-efficient signal processing solution, capable of direct data decoding on edge devices, marks a significant step toward intelligent and low-power wireless communication systems enabled by memristive hardware.
The hardware-software co-design of memristive systems, including mixed-signal hardware development, signal processing and neural network algorithm optimization, and applications in language classification and wireless communications, paves the way to fully unlock the potential of AIMC hardware based on memristor devices for next-generation edge intelligence.Doctor of Philosophy (Ph.D.)2026-09-0