116964 research outputs found
Sort by
Development of Monte Carlo models of a gamma-ray nondestructive assay system for nuclear material control and accountability
Get PDFNuclear material control and accountability (NMCA) is vital to safeguarding special nuclear materials (SNM) against theft and diversion. Nondestructive assay (NDA) via gamma-ray spectrometry is one of the most common methods employed by NMCA programs to determine the amount or composition of SNM. However, accurate gamma-ray NDA of SNM can be difficult in certain scenarios due to the complex distributions and compositions samples can exhibit. Monte Carlo models of a high-purity germanium (HPGe) gamma-ray detector used for assaying plutonium were developed and benchmarked to simulate two difficult measurement scenarios often faced at Los Alamos National Laboratory (LANL): electrorefining salt (ER) residues and holdup in gloveboxes. For holdup measurements, accuracy is inherently limited because the exact distribution of material is unknown, which can lead to high uncertainties in reported masses. While the uncertainty in holdup measurements is assumed to be substantial, a thorough investigation of what an adequate value for gloveboxes at LANL should be has not been completed. Thus, measurements were simulated with various distributions of plutonium within a glovebox, and the resulting spectra were analyzed to see how the spatial distribution of material affected the measured mass. In doing so, a more accurate uncertainty value was determined. For the ER salts, NDA is hindered by samples’ complex compositions which arise from the fresh separation of elemental components and the possible presence of heterogeneities. Gamma-ray measurements of both homogenous and heterogenous samples were modelled with varying amounts of plutonium present. The simulation results from these Monte Carlo models were then used to determine the minimum detectable amount (MDA) of ²³⁵U for different plutonium masses and to develop an improved parameter set used for determining the isotopic compositions of the salts.Mechanical Engineerin
Structural insights of antibody-mediated immunity against viral glycoproteins of Crimean-Congo hemorrhagic fever virus
Get PDFCrimean-Congo hemorrhagic fever virus (CCHFV) is the causative agent of the most widespread tick-borne viral illness, CCHF. Sub-clinical cases are common yet for severely symptomatic patients the fatality rate can range from 10-40%. Expansion of endemic zones and high case fatality rates has established CCHFV as an urgent public health threat as there are no licensed vaccines or therapeutics. CCHFV is a negative-sense RNA enveloped virus with structural and secreted glycoproteins that elicit neutralizing or protective antibodies making them promising targets for medical interventions. However, the modalities by which these antibodies neutralize and protect must be delineated.
The first part of this thesis presents the characterization of a panel of antibodies that target across four distinct antigenic sites on viral membrane fusion glycoprotein Gc, the sole target of neutralizing antibodies. Binding studies with recombinant Gc of diverse CCHFV isolates reveal the Gc-specific antibody panel to be broadly cross-reactive. Together with crystal and cryo-EM structures of Gc bound with antigen binding fragments (Fabs) and authentic virus neutralization assays, we reveal a structural basis for neutralization provided by Gc-specific antibodies. These data provide candidates that may be suitable for Gc-specific antibody therapeutics and support the use of monomeric Gc in vaccine antigen designs to elicit a broadly reactive and potently neutralizing antibody response.
GP38 is an essential viral factor during virion formation yet GP38 is a target of non-neutralizing, protective antibodies leading us to investigate if GP38 plays additional role(s) in pathogenesis. The second part of this thesis presents the isolation and characterization of 188 human antibodies targeting GP38 from three convalescent donors. Using competition studies, we mapped the non-neutralizing GP38-specific antibodies to target five distinct antigenic sites encompassing eleven overlapping regions on GP38. Structural studies of GP38 bound with Fabs reveal the antigenic landscape on GP38. Lastly, protection studies in two stringent rodent models of infection revealed that several antibodies display similar or improved protective efficacies compared to the murine antibody 13G8. Together, these data expand our understanding of antibody-mediated immunity against CCHFV glycoproteins and inform the development of broadly effective medical interventions against CCHFV.Biochemistr
Lidar at the margins of the Mediterranean : evaluating lidar-based archaeological survey in Mediterranean and cycladic landscapes
Get PDFLidar-based research has a long history in Mediterranean archaeology, yet interpretive and methodological achievements of this research often remain sequestered in regional frameworks. Given well-established connections of research interests and landscape types across sections of the Mediterranean Basin, this thesis analyzes lidar-based research as a Mediterranean-wide phenomenon, highlighting both shared and divergent methodological approaches and research interests across the region. Drawing on recent discussions about the importance of accuracy assessment in Mesoamerican lidar archaeology, this paper particularly focuses on accuracy assessment methods used across the Mediterranean and proposes the strength of area-based field verification methods combined with environmental spatial statistical analyses to better understand the strengths and weaknesses of lidar-based archaeological survey. A case study for this approach is presented using lidar-derived feature classifications from the Small Cycladic Islands Project (SCIP), an intensive archaeological survey of small, uninhabited islets of the Cycladic Islands, Greece, where results show the high accuracy for linear features and the impact of low maquis vegetation and geological features. Accuracy results are analyzed by feature and as a function of data quality at each pixel using a logistic regression model that finds association between low data quality and erroneous classifications. Beyond project-specific accuracy results, this paper proposes methods that can flexibly adapt to projects across the Mediterranean, improving data transparency and describing environmental impacts on lidar-based research to build a robust conception of Mediterranean-wide lidar-based archaeology.Geography and the Environmen
Making a market for security : private security entrepreneurship in Mexico City
Get PDFPrivate Security is one of the fastest-growing industries around the world. In the Global South, private security has become an integral part of life in the city. In Mexico City, private security has situated itself as a prerequisite of residential, commercial, and industrial activity. But how has private security positioned itself as an elementary element of urban life? Most approaches see the emergence of security in Latin America as the result of increased urban insecurity on the continent, or the introduction of structural changes in security provision. By contrast, by employing ethnographic methods in the booming security sector of Mexico City, this dissertation highlights the significant role played by entrepreneurial actors in shaping the security market and continuously expanding its scope. Security entrepreneurs have not only forged a security market, but introduced new methods of risk assessment, security protocols, and aesthetic images that imbue more dimensions of life in the city with their logic. My research follows the development of the Mexican private security market, the contemporary production of security, and the dynamics shaping the field. Furthermore, although often framed as an alternative to the incapacity of the state to eradicate urban insecurity, this dissertation shows that security entrepreneurs have a complex relationship with the state, which on the one hand, laid the social groundwork for the security market, but is simultaneously an ally and a rival of the entrepreneurs. I argue that rather than being a mere outcome of Mexico's evolving security governance, security entrepreneurs have found a dynamic and contested social arena, shaped by skilled actors who challenge the state and compete with one another over legitimate security expertise. Through their work, security entrepreneurs have created a new social model of private security provision in Mexico, defining and contesting new meanings of security expertise, and altering the social relations needed for security production. By positioning themselves as the new security experts of the city, entrepreneurs in Mexico epitomize the spirit behind one of the largest and most profitable industries around the world.Sociolog
Behavior and design of U-shaped girder bridges supporting heavy railings and sound walls
Get PDFBridge railings and sound walls are essential for vehicle safety and noise mitigation but can impose significant demands on bridge superstructures. Current design practices, which commonly assume a uniform railing load distribution, may be inadequate for heavy or tall railings, potentially leading to underestimation of girder and overhang forces.
This dissertation investigates the structural behavior of prestressed concrete (PSC) U-girder and steel tub girder bridges supporting heavy railing systems. Refined three-dimensional finite element FE models were developed to examine the load transfer mechanisms and identify key parameters influencing the Railing Load Distribution Factors (RLDFs), including span length, girder spacing, overhang width, and bearing stiffness. Extensive parametric analyses were conducted, and predictive RLDF equations were formulated using symbolic regression. Correction factors for skew and curvature were also proposed to enhance design accuracy.
In addition, the study evaluated TxDOT deck overhang details under both AASHTO (2020) and NCHRP 12-119 (Steelman et al., 2023) design frameworks. The results indicated that the 8.5-in. overhang detail is inadequate for heavy railings, whereas the 12-in. detail satisfies AASHTO requirements and remains recommended based on prior TL-6 crash testing.
Overall, this research advances the understanding of railing load behavior in slab-girder bridges and provides rational, conservative design recommendations. The proposed RLDF equations and design guidelines enable more accurate, efficient, and safer bridge designs capable of accommodating heavy railing and sound wall systems.Civil, Architectural, and Environmental Engineerin
Distributed fault tolerant distance computation in graphs : algorithms and lower bounds
Get PDFDistributed systems are fundamental to modern computing at scale, with a wide variety of applications, including computer networks, cloud computing, big data computing and blockchains. Fault tolerance is essential in distributed systems to manage failures gracefully and ensure reliability. Routing messages is an important problem in distributed networks, and efficient fault-tolerant routing requires methods to maintain shortest distances between nodes when faults occur.
The goal of this thesis is to design and analyze efficient distributed algorithms for these fault-tolerant, distance-related computations, and to develop bounds on the limits to the efficiency of such algorithms by proving unconditional lower bounds.
In this thesis, we develop distributed algorithms and lower bounds for the Replacement Paths problem, which addresses a shortest path between a pair of specific vertices in the event of an edge failure on the original path, and the Distance Sensitivity Oracle problem, which pertains to shortest paths between any pair of vertices when any single edge fails. Despite their importance in the distributed setting and extensive literature in the sequential setting, prior to this thesis, little was known about the distributed round complexity of these problems. In many cases, we obtain near optimal bounds for our problems.
The sequential fine-grained complexity of these fault-tolerant path problems is closely related to other fundamental graph problems, such as the Minimum Weight Cycle (MWC). We extend our investigation to these cycle problems, studying their exact and approximate distributed round complexity, including a novel sublinear 2-approximation algorithm for directed MWC and an algorithm that closes a longstanding gap for 2-approximation of undirected unweighted MWC (girth).
Together, our results display a rich landscape of distributed round complexity for exact and approximate versions of these problems in different graph types: directed or undirected, weighted or unweighted.
Finally, in addition to the distributed setting, we study algorithms for fault-tolerant distance problems in other models of parallel computation, such as PRAM (Parallel RAM).Computer Scienc
Geomechanical Property Evaluation of the Pennsylvanian Strawn-Canyon Group Using Machine-Learning Methods: Insights from the Salt Creek Field, Midland Basin, Kent County, Texas
Get PDFGeomechanical properties of rock are critical in determining fracture shapes and designing hydraulic fracturing procedures. These properties are usually obtained from static laboratory measurements but are limited to
borehole locations. However, in the absence of static properties, dynamically derived properties from surface seismic data can be used. By integrating 3D poststack seismic and wireline-log data, and using machine-learning
procedures, we generated volumes of three dynamic properties: namely, Young’s modulus (E), Mu-Rho (MR), and brittleness (BRI) to characterize the Salt Creek carbonate reservoir located in the Midland Basin, Kent
County, Texas. By generating crossplots, we show that (1) E versus MR, and E versus BRI exhibit strong linear relationship with R2 of 0.96, and 0.71 respectively. (2) On time slices, porous zones can easily be identified,
characterized by lower Gpa values compared to the surrounding non-porous rocks. (3) Because each of these properties can be correlated with photoelectric-factor (PEF) logs, and because PEF log values can indicate calcite
richness and porosity within a mixed carbonate-siliciclastic system, any of the three geomechanical properties, particularly E and MR, can be used to identify rigid rock layers within the reservoir and to deduce porous zones
and possible lithologies.Bureau of Economic Geolog
The critical element of protein-protein interactions from polyketide assembly lines
Get PDFSeveral pathways within the cell require proteins to dock to one another for proper biochemical reactions to occur. These minute protein-protein interactions are critical for polyketide biosynthesis by their modular megadalton assembly lines termed Polyketide Synthases (PKSs). Through carbon-carbon bond formation and selective reduction, dehydration, and hydrogenation, these complex pharmacologically active natural products can be created from simple α-carboxyacyl chains. From acyl carrier protein docking to enzymatic domains, to the joining of megapeptides to form complete PKS assembly lines, protein-protein interactions play a vital role in the synthesis of these diverse secondary metabolites. The Lateral INteracting Ketosynthase Sequence (LINKS) of ketosynthase domains within trans-acyltransferase PKSs generates organelle sized assembly lines to create a highly organized enzymatic network (Chapter 2). The critical extension reaction performed by ketosynthase domains is selective for the stereochemistry of the substrate from upstream processing domains and contains specific docking motifs for the intramodular acyl carrier protein (Chapter 3). The installation of methyl substituents on the growing polyketide are from S-adenosyl-methionine dependent methyltransferases, and the docking of the acyl carrier protein adds an additional level of avidity of the catalytic domain for its β-ketoacyl substrate (Chapter 4). Through engineering PKS docking motifs onto fluorescent proteins, new complexes can be formed to generate a reportable signal (Chapter 5). Investigations of PKS proteinprotein interactions presented in this dissertation yield additional understanding on how these massive molecular factories generate pharmacologically active compoundsBiochemistr
Nano-enabled microwave-harnessing technique as an alternative disinfection approach for the control of Legionella pneumophila and the free-living amoebae that harbor them
Get PDFLegionella pneumophila (Lp) is the most notable among the opportunistic pathogens, which are responsible for the increasing public health concerns originating from engineered water systems. The critical survival niche for Lp within these systems is their ability to infect free-living amoebae (FLA) and use the host cell for protection and permissive replication. This parasitic relationship allows Lp to evade conventional means of pathogen control (i.e., chlorination), making alternative disinfection strategies appealing and increasingly important. Notable examples in the literature of FLA and intracellular Lp vulnerabilities to thermal treatment prompted our development of a microwave (MW)-enabled treatment strategy, incorporating the high heat-conductance of silver to augment this low-energy, widely used form of radiation. Synergistic mechanisms between MW and silver were hypothesized to produce enhanced inactivation relative to the sum of the components. This dissertation reports the application of MW and silver (nanoparticulate and ionic) combined treatments towards various Lp forms of increasing resistivity, i.e., (i) free Lp cultured to the exponential growth phase, (ii) hyperinfectious, free Lp released from FLA after an intracellular growth cycle, and finally (iii) intracellular Lp, which is co-cultured and harbored within FLA. At each stage, rapid and effective inactivation was observed for MW coupled with silver nanoparticles (AgNPs) and ions (Ag⁺). A comprehensive study of Lp infectivity revealed that FLA not only protect, but also condition, Lp to express enhanced treatment resistance and infectiousness upon egress. Combined treatment involving MW exposure and Ag⁺ was the most effective treatment scenario reported for intracellular Lp. Complete inactivation (>5 log reduction) of intracellular Lp co-cultured with A. castellanii was observed with 6 min of MW exposure and Ag⁺ (0.1 mg L⁻¹), remarkably overcoming the increased resistance provided by host amoebae. Overall, these initial results emphasize the potential for this promising alternative disinfection strategy while simultaneously raising awareness of the inextricable link between Lp and FLA that must be considered in the design of control strategies.Civil, Architectural, and Environmental Engineerin
Scheduling and resource allocation in modern distributed systems : applications to federated learning, blockchains, and large language models
Get PDFDistributed systems underpin many of today’s most important applications, from financial to large-scale AI services. Their defining feature is that computation and communication are spread across networked nodes, making system performance critically dependent on how efficiently resources are allocated and scheduled for the network. In this dissertation, we develop new algorithms with rigorous performance guarantees for resource allocation and scheduling in three modern distributed applications: federated learning, blockchains, and large language model (LLM) inference. First, in federated learning, the aggregation of updates from clients to the server is a critical performance bottleneck. We propose two approaches to alleviate this. The first is an algorithm, Network Adaptive Compression (NAC-FL), that dynamically adjusts compression levels for client updates based on network congestion and achieves provably near-optimal theoretical performance, as well as provides improvements over other compression algorithms in several scenarios in practice. The second uses overlay aggregation, where clients aggregate each other’s updates en route to the server. Here, we design two policies: OptSched which schedules updates across aggregation trees to minimize aggregation time, and OptAgg, which introduces replication for robustness against failures. Second, in blockchains, it is well known that blocks need to be propagated quickly through the network to maintain security. We introduce the metric of Ordered Completion Time (OCT) to quantify the speed of in-order block dissemination. We show that optimal block dissemination is impossible under arbitrary arrival patterns, even in simple networks, and develop an age-based block-relaying protocol that achieves near-optimal OCT in tree overlays. Third, in LLM inference, designing algorithms for efficient request routing and scheduling is a key engineering challenge. We present one of the first theoretical frameworks for this setting, modeling GPU computation times and designing the Resource-Aware Dynamic (RAD) scheduler. RAD, combined with a uniformly random request router, is provably throughput-optimal and provides insight into the design trade-offs of current state-of-the-art inference systems. Together, these contributions advance the theory and practice of distributed systems, offering algorithmic principles that improve performance in the domains of federated learning, blockchains, and LLM inference.Electrical and Computer Engineerin