137091 research outputs found
Sort by
Extracting Black Hole Phenomena from Gravitational Waves
No full textOn the 14th of September 2015, the Laser Interferometer Gravitational-Wave Observatory
(LIGO) made the first detection of a gravitational wave (GW) signal from the merger of two
black holes, marking the start of a new era in our quest to understand gravity, the universe,
and the nature of reality. In the decade since, the LIGO/Virgo/KAGRA Collaboration has
detected over 200 gravitational wave events, primarily from the mergers of binary black hole
(BBH) systems. Measurements of a GW signal through Bayesian inference yield statistical
information on the parameters of their originating system, offering insight into the properties,
characteristics, and dynamics of the progenitor black holes. Herein, methods for extracting
information from GW signals related to BBH system phenomena are discussed over the
course of three original projects of my design.
The first project analyzes the precession of the binary’s orbital plane, which occurs when
the black hole spins are misaligned relative to the orbital angular momentum. An implementation of different ways to parameterize the precession in RIFT - an iterative parameter
estimation algorithm for analyzing GW data - is presented. It is shown that the interpretation of the inferred precession depends strongly on its parameterization if both spins are
misaligned, and methods are developed to leverage this dependence in the operation of RIFT.
The second project implements the first comprehensive parameter estimation infrastructure
for measuring the source properties from systems of initially unbound black holes that make
close hyperbolic encounters. Such systems exhibit diverse waveform morphology, and can
either scatter, dynamically capture after multiple flybys, or directly plunge to merger. It is
shown that with this implementation, RIFT can accurately recover the source parameters
of these systems.
The final project investigates how the geometry of the horizon that forms when two black
holes merge may be encoded within the time-frequency representation of signals from BBH
systems. First, methods are developed for visualizing time-frequency structure using the
continuous wavelet transform (CWT), utilizing both sine-Gaussian wavelets and ‘chirplets’
- sine-Gaussian wavelets that evolve in frequency. Second, the CWT is used to analyze
the post-merger signal from BBH systems with asymmetric mass ratio under a variety of
scenarios, including both aligned and precessing spin. This study shows that correlation
between time-frequency features and the horizon dynamics is plausible, and a route towards
‘imaging’ black holes is discussed.Ph.D.Physic
Digging the Details
No full textInterview portion of Lost in the Stacks, episode 670. Features interview with Vanesa Evers, the Institutional Repository Librarian at the Georgia Tech Library, discussing how the repository works and what it's like adjusting to a new librarian role.Interview portion of Lost in the Stacks, episode 670. Features interview with Vanesa Evers, the Institutional Repository Librarian at the Georgia Tech Library, discussing how the repository works and what it's like adjusting to a new librarian role
Development of a Multi-Disciplinary, Parametric Lunar Power Beaming Satellite Model Using Dyreqt
No full textExploring the lunar surface outside of a select few regions at the poles poses the obstacle of overcoming the two-week long lunar night. Traditional methods for exploration assets to survive the lunar night involve using a radioisotope heater for continuous power or switching to an extremely low-powered survival mode during night periods to minimize energy storage requirements. This can greatly increase mission cost or significantly limit the amount of science that assets can perform. One emerging solution to this problem is to utilize Orbital Power Beaming to enable limited global lunar power distribution. Orbital Power Beaming consists of using a satellite in low lunar orbit to wirelessly transfer power to lunar surface assets such that they have at least enough energy to sustain themselves until the beamcraft's next orbital pass. The beamcraft's concept of operations is defined using three phases. The first phase is a direct minimum energy transfer trajectory from Earth to a low lunar orbit, developed using STK. The second phase details how the beamcraft beams power to a surface asset while in orbit. The third phase explains beamcraft's disposal process at its end-of-life. A parametric model for sizing the beamcraft to perform this mission profile and satisfy the power requirements of the surface asset was developed using the Dyreqt framework. The beamcraft model is composed of high-fidelity subsystem component models that use a resource flow and allocation approach to perform mass, power, thermal, and cost sizing. This model is demonstrated by performing notional trade studies of different beamcraft designs and mission alternatives
Translating Roadway Alerts into Pedestrian Impedance Factors in Real-Time
No full textThe ITS4US Georgia project has established a navigation system for ADA mobility mode users that employs a complete paths network in OpenStreetMap (OSM) for use with OpenTripPlanner's (OTP’s) shortest path navigation system. Impedance factors that affect each ADA mode are assigned to each link in the complete paths network so that preferred paths can be found for each mode. For example, if the sidewalk is closed on one side due to maintenance, user may choose the other operational sidewalk which will be considered the preferable path for travel. Such alerts are typically not gathered or made available to improve accessible travel within a pedestrian network. Pedestrians usually encounter these closure events only when they are already navigating the affected path. This study uses crowd-sourced roadway alerts to associate the impact of roadway related events over pedestrian networks. For this research these roadway alerts are integrated with in the OSM network. The OSM-based ADA network used for this study was developed by the Georgia Tech ITS4US Spatial Analytics Team. The rectified OSM pedestrian network is represented in the ITS4US space-time-memory (STM) system in a graph database architecture (Amazon Neptune). This network assigns link-by-link impedance based on physical attributes (e.g. sidewalk width, surface quality, cross-slope, obstruction presence, etc.), asset condition (e.g. surface roughness, cracking) and traffic/roadway related events and alerts.
As mentioned, the overall route accessibility is also dependent on how the link performs during the roadway related events and alerts nearby, which is also considered the fundamental element of shortest-path routing, by adding impedance to such affected network links. For the ITS4US project, one source of roadway alerts information is Waze data, which can be spatially tied to the OSM network. Waze data is chosen based on its data coverage and attribute information for all road types. By analyzing dynamic data from Waze alerts, this research will identify and assesses sidewalk segments that appear to be affected by incidents, influencing routing decisions through impedance calculations based on alert severity (for example, the presence of an ambulance or fire engine that may impede the walking path). The research also explored the potential of Waze traffic data for more accessible travel, extending the analysis to adjacent roadways to optimize routing in real-time and mitigate hazards. The objective is to enhance the overall accessibility of the routes for vulnerable road users. Accessibility is determined by whether the alert affects the link and how it is affected (if the link is fully or partially obstructed by the alert). This research proposed a framework to associate the impact using spatiotemporal buffers conducted to identify the optimal spatial coverage of the Waze alert on the sidewalks. This algorithm models the nature of the Waze subtypes and their integration across space and time to detect link affected by alerts over time.
Challenges such as report reliability, incomplete descriptions, and disparities between actual incident locations and user-reported ones, underscore the need for cautious interpretation. Based on the results from two case studies, the scenario-based study concludes that for most cases the alert impedances are mostly effective to analyze new routes which are more accessible. While the case study considering the overall impact of Waze alerts in the study area concludes that the alternative routes are longer but offer greater unobstructed accessibility compared to the original route. These outcomes were attributed to the complete reliance on alert-based impedance factors. Since these factors are derived from roadway alerts, the impedance values were set higher to prioritize accessibility, accounting for sudden obstructions caused by the magnitude and nature of the roadway alerts. These final results are also subject to certain limitations such as links assigned to current attribute-based algorithm, impedance derivation, alerts impact coverage and alert accuracy determined for analysis. The limitations of the shortest path analysis included the use of ArcGIS Pro and its network, which typically identifies more desirable routes, but there are potential drawbacks to this approach. Despite employing impedance values for the analysis, if the network has not been adequately segmented into links for routing assignments, ArcGIS Pro may overlook such configurations and route based on the existing network structure. Nevertheless, by integrating real-time traffic alert information with spatial analysis, this research aims to associate the impact over the pedestrian network and improve the overall accessibility for vulnerable road users.M.S.Civil Engineering/City and Regional Plannin
Data for Observations of the Gas Phase Composition of the 2024 BioLab Industrial Plume in the Atlanta Metropolitan Area
No full textThis dataset contains gas-phase observations of chemical species measure in Midtown Atlanta following the BioLab chemical manufacturing plant fire in Conyers, GA on September 30, 2024. The fire at the pool chemical manufacturing facility, located approximately 21 miles east of the Georgia Institute of Technology campus, resulted in a significant chemical plume that prompted shelter-in-place and evacuation orders for 17,000 residents.
Data Collection: Measurements were conducted from September 30 to October 9, 2024, using a Chemical Ionization Mass Spectrometer (CIMS) deployed in Midtown Atlanta. Post-collection calibrations were performed to quantify concentrations of 18 target compounds, with an additional 9 species detected but not calibrated.
Data Contents:
· Calibrated concentrations for 18 identified chemical species
· Signal intensities for 9 additional uncalibrated species
· Peroxyacetyl nitrates (PANs) concentration measurements
· metadata
File Structure: The Excel workbook contains multiple sheets:
· Sheet 1: Metadata and methodology
· Sheet 2: Calibrated compound concentrations
· Sheet 3: Uncalibrated compound signals
· Sheet 4: Peroxyacetyl nitrates (PANs) data
Research Significance: This dataset provides critical atmospheric chemistry data for understanding the plume composition from the decomposition of spa and pool chemicals.A fire at a pool chemical manufacturing facility in Conyers, Georgia on September 29, 2024, released a persistent chemical plume that impacted the Atlanta metropolitan area for over two weeks, leading to evacuation of more than 17,000 people. We used high-resolution time-of-flight chemical ionization mass spectrometry at our laboratory in Midtown Atlanta (21 miles from the facility) and deployed a quadrupole CIMS in Conyers to characterize plume composition. We observed unexpectedly high concentrations of Br₂ (up to 1.4 x 10² ppb) dominating early plumes, along with elevated HNCO (31 ppb) and numerous other compounds. Twenty-six species were identified, including reactive nitrogen-containing compounds (HNCO, cyanoacetic acid, cyanamide) and oxygenated volatile organic compounds (acetaldehyde). Br₂ concentrations in Midtown exceeded EPA one-hour AEGL-1 thresholds by a factor of four. Later measurements in Conyers showed Cl₂ reaching 3.7 x 10² ppb during the second week. Given that Midtown observations were 21 miles downwind, concentrations near the source in Conyers initially were likely 1-2 orders of magnitude higher. This study provides the first comprehensive chemical characterization of a plume from pool chemical decomposition, revealing complexity far beyond simple halogen release and highlighting enhanced health risks from simultaneous exposure to multiple respiratory irritants.NSF #250933
A Systems-of-Systems Framework to Support Decision-Making for Lunar Communication Network Design Under Stochastic Disruptions
No full textAs space agencies and commercial actors prepare for sustained human presence on the Moon, the need for high-performance lunar communication infrastructure is becoming critical. Future missions will require networks that support high data rates, low latency, and robust, uninterrupted links between the lunar surface and Earth. Designing such networks poses significant challenges, including trade-offs in architecture and technology selection, a vast number of possible configurations, and the computational cost of evaluating them. This paper presents a system-of-systems framework for the rapid evaluation and comparison of lunar communication network alternatives defined by constellations of Earth-orbiting and cislunar satellites. A high-bandwidth use case: real-time 16K video streaming of the Artemis III landing, is employed to benchmark network performance, as it exceeds the requirements of foreseeable lunar missions. Network architectures are analyzed using high-fidelity simulations in Ansys Systems Tool Kit (STK), combined with Python-based post-processing that integrates graph theory, Monte Carlo simulations of stochastic disruption events, and surrogate modeling for accelerated analysis. Evaluation metrics include network complexity, latency, robustness, and link continuity under disruptions such as satellite failures and adverse ground station weather. The proposed decision-making framework enables rapid and comprehensive exploration of the design space, providing critical insights for the development of scalable lunar communication networks
Nonlinear Opinion Dynamics on the Sphere for Distributed Multi-Agent Systems
No full textMulti-agent systems have been widely studied with various potential real-world applications. Much of the inspiration for this field comes from natural collective behaviors, such as fish schooling, ant colonies, flashing fireflies, etc. When a large swarm of individuals are performing complicated tasks, group-level decision-making is critical to successful task completion based on individual-level information collection and communication. During this process, individuals form various opinions, exchange opinions with their neighbors, and gradually reach a consensus or dissensus, which leads to a group-level decision. Thus, studying the opinion dynamics of multi-agent systems is crucial for understanding the decision-making process from both group and individual perspectives.
The objective of this dissertation is to explore and develop models of opinion dynamics for distributed multi-agent systems, focusing on the diverse behaviors of consensus and dissensus under various interaction rules in unsigned graphs. A novel aspect of this dissertation is the modeling of opinion states as unit-length vectors on a sphere, representing unique measures of expressed opinions. The evolution of these state vectors illustrates the change in opinions of each agent, influenced by neighboring opinions.
The dissertation's contributions are classified into theoretical work and practical applications. The theoretical contributions establish foundational models for understanding rich opinion behaviors, including consensus and various forms of dissensus. These models are significant not only for describing individual and group behaviors in social networks, but also for explaining different communication methods when agents interact and exchange information.
On the application front, the models are applied to multi-robot task allocation. In these contexts, opinion dynamics facilitate upper-level decision-making processes. The research thus bridges theoretical insights and practical implementations, enhancing the understanding and utility of opinion dynamics in complex systems.
This work bridges the gap between theoretical models and applications of opinion dynamics in distributed systems. By developing and applying innovative models, the research contributes to a deeper understanding of how opinions evolve and influence decision-making processes in complex, multi-agent environments. This work not only advances the field of opinion dynamics but also provides valuable insights and tools for practical implementations in various technological and social domains.Ph.D.Electrical and Computer Engineerin
“Listening to Shakira is an important part of the process”: On the importance of “non-productive” labor in the university library
No full textBook chapter: Library Jackassery: The silly ideas that work and embracing working silly (2026
Magma-Assisted Extension and Crustal Deformation in the Kivu Rift: Geodetic Constraints on the 2021 Nyiragongo Dike Intrusion
No full textThe East African Rift System (EARS) is one of the most tectonically and magmatically active continental rifts on Earth, offering critical insight into the mechanisms governing the transition from continental extension to seafloor spreading at divergent plate boundaries. Within its western branch, the Kivu Rift is a volcanically and structurally dynamic segment where repeated dike intrusions and eruptions at Nyiragongo and Nyamuragira volcanoes accommodate regional strain. The May 2021 Nyiragongo dike intrusion and eruption provided a unique opportunity to quantify how a transient magmatic event perturbs the regional stress field, reorganizes crustal strain accommodation, and advances long-term rift evolution within this portion of the EARS.
In this work, we integrate Interferometric Synthetic Aperture Radar (InSAR) and Global Navigation Satellite System (GNSS) observations to constrain the geometry, kinematics, and spatio-temporal evolution of deformation associated with the 2021 intrusion. A joint elastic dislocation model identifies a near-vertical dike between 3–6 km depth, with a maximum tensile opening of nearly 10 m beneath Goma and a total length of 25 km southward of Nyiragongo. The intruded volume (~1.8 × 10⁸ m³) exceeds the erupted lava volume by nearly an order of magnitude, confirming that the event was predominantly intrusive. The geodetic moment (~5.5 × 10¹⁸ N·m) accounts for more than 95 % of the total deformation energy, indicating that strain was accommodated almost entirely aseismically. The close alignment between the modeled dike, mapped faults, and co-eruptive seismicity indicates a structurally guided intrusion, consistent with strain localization along pre-existing fault corridors observed in other active rift zones such as Afar and Iceland.
Sentinel-1A/B InSAR time-series analysis (2019–2025), integrated with continuous GNSS measurements, captures the three-dimensional and temporal evolution of deformation across the Kivu Rift. Interferograms processed using the InSAR Scientific Computing Environment (ISCE) and its Dolphin module employed robust techniques to mitigate decorrelation and atmospheric noise in the mountainous Virunga region. The derived displacement fields reveal uplift and westward motion on the western flank and subsidence with eastward motion on the eastern flank, consistent with dike opening parallel to the regional fault architecture and principal extensional axis. Time-series results show that the 2021 intrusion accelerated rift extension by nearly a decade of tectonic strain within days, while persistent post-intrusion deformation indicates a long-lived modification of the regional stress field and shallow crustal rheology. Collectively, these results demonstrate that the 2021 event was a structurally controlled, dominantly aseismic dike intrusion that localized extension and efficiently transferred magma through the brittle crust, underscoring the importance of sustained geodetic monitoring for understanding magma–tectonic coupling and improving hazard preparedness in the Goma–Gisenyi region
Interval Finite Element Approach for Stability Analysis of Concrete Gravity Dams under Uncertainties
No full textThe stability of concrete gravity dams is a critical safety concern, particularly under the influence of uncertain loading and strength parameters. In current engineering practice, the crack-based Limit Equilibrium Method (LEM) relies on simplified assumptions and frequently yields conservative results with excess safety margins. In particular, LEM relies on the linear Mohr–Coulomb (MC) shear strength envelope and crack initiation criteria based only on the location of the resultant force. While the Mohr–Coulomb (MC) shear strength model is suitable for foundation where no asperities exist, most concrete gravity dams are founded on rock foundation, where asperities are present along the concrete–foundation rock interface. Due to these asperities, the Mohr–Coulomb (MC) model underestimates the shear strength of the sliding interface. Probabilistic methods, such as Monte Carlo Simulation (MCS), provide valuable insight but require extensive site-specific data and probability distributions that are rarely available for existing dams.
This research develops a novel Interval Finite Element Method (IFEM) framework to overcome these limitations by explicitly incorporating uncertainties in dam stability analysis, utilizing the nonlinear Barton–Choubey (BC) shear strength model, which accounts for the presence of asperities along the dam foundation. All sources of uncertainty are expressed in interval form and incorporated within the IFEM framework to evaluate the structural response of concrete gravity dams under parameter uncertainty. The proposed framework integrates interval arithmetic with the nonlinear Barton–Choubey shear strength model, an element-by-element assembly strategy to reduce dependency effects, and a new crack initiation and propagation criterion based on both tensile and shear stresses that iteratively updates uplift pressure and boundary conditions as the crack propagates. Together, these innovations enable realistic modeling of crack–uplift interaction and provide mathematically guaranteed enclosures of structural response.
The methodology was validated through a concrete gravity dam case study, where uncertain parameters, including Young’s modulus, concrete density, water unit weight, joint roughness coefficient (JRC), joint wall compressive strength (JCS), and residual friction angle, were expressed in interval form using laboratory test data, historical NOAA temperature records, and construction photographs. Results demonstrated that the dam remains stable, with interval factors of safety above unity and crack lengths consistent with MCS predictions, while avoiding the excessive safety margins associated with LEM.
By providing reliable, computationally feasible, and data-efficient safety bounds, the proposed IFEM framework offers dam owners and regulators a powerful tool for evaluating stability under uncertainty. This approach not only improves confidence in safety assessments but also has the potential to reduce unnecessary remediation measures, leading to safer and more cost-effective management of aging dam infrastructure