62588 research outputs found
Sort by
Femtosecond Laser Fabrication of Optically Active Single Crystals for Quantum Networking Applications
No full textUsage of a femtosecond laser to modify solids in 3D is of particular interest to an array of integrated photonic applications. Optical devices fabricated this way are not limited by an inability to cross over other optical elements, making themhighly desirable. Tuning this technique to fabricate crystals in glass allows for functionalities beyond waveguiding, especially in a material like LiNbO3. Further expanding this technique, introduction of a rare earth dopant is desirable for light emitting devices or for optical quantum memory. Of particular interest for telecom and quantum communications networks is erbium. Its 4I15/2 →4I13/2 transition is around 1.5µm, matching the wavelength band for the vast fiber-optic network spanning the globe. As such, erbium doped crystals in bulk form have been of particular interest for optical quantum memories. In this work, fs laser processing of glass to form crystalline waveguides is expanded, allowing for fabrication of long single crystals of Er:LiNbO3 in lithium niobosilicate glass. The erbium in these fs laser written single crystals in glass is characterized by a variety of spectroscopic techniques, to determine the nature of its incorporation into
the crystal lattice. The results of these measurements are further compared against bulk grown stoichiometric and congruent LiNbO3. Usage of spectral hole burning has allowed for probing of coherence lifetime through homogeneous linewidths. In the aim of expanding fs laser crystallization to fabricate new crystal phases in glass, desirable for applications such as optical quantum memory, a unique glass-making technique is used. Through this, we have established an ability to create glass that forms yttrium aluminum garnet during fs laser processing. Expanding this technique to fabricate glass of Y2SiO5 and Y2Si2O7 is also explored.</p
Characterizing Performance Enhancements in Vacuum Membrane Distillation using Computational Methods
No full textMembrane distillation (MD) is a separation technology driven by low-cost thermal energy and is effective on highly concentrated aqueous solutions. This process has been studied extensively in lab-scale environments. The key indicators of performances losses in the lab, namely radial temperature polarization, have been characterized and there are many published works mitigating temperature polarization with thermo-fluidic optimizations. However, when transitioning to prototype scale there is a major disconnect in the performance from lab experiments, showing the knowledge from existing modules hasn\u27t transitioned into high performance prototypes, and, that prototype modules are different and more complicated systems than what many researchers currently consider. Therefore, the goal of this work is characterize the changes in up-scaling membranes from lab to prototype-scale using well understood computational models to conduct high-fidelity studies. Next, simulations on prototype-scale modules will suggest mitigation methods for the performance losses considering the principles of fluid-flow and heat transfer optimizations that have been successfully utilized in the lab. Novel wiggly channel designs will be studied to improve thermal performance while reducing pressure drop using high-fidelity computational simulations. Embedded stiffeners will also be considered to provide manufacturing support to the membrane and promote additional vortex activity within the channel. Next, the model will be simplified with a reduced-order 2D assumption to study a wide range of prototype systems at low computational expense, a major limiting factor in existing literature. That work will characterize the discrepancies that occur between lab and prototype scale experiments, the additional complexities of performance losses in prototype modules, and methods to improve prototype performance. Finally, studies will also be conducted buoyancy driven convection in membrane distillation in the free, mixed, and forced convective regime. Many researchers operate in the forced convective regime based on the operational parameters, and this work found an easy method to transition into the mixed convective regime by changing the channel height, which promoted natural convection cells and provided a performance enhancement versus cases without natural convection. Additionally, these enhancements proved effective in long and short modules. Further simulations were also conducted to study novel operational conditions for vacuum membrane distillation including pulsed flow, cross flow, and multiple inlets on improving the thermal characteristics and therefore separation performance of membrane distillation.</p
Hydrogen Energy Storage: From Predictive Interaction with Metallic Materials to Thermodynamic Analysis of System Integration
No full textHydrogen is a versatile element with the potential to revolutionize various industries,including energy, transportation, and material science. However, its behavior
and interactions at the atomic level are complex, making it challenging to accurately
predict its properties and behavior in different environments. These items are important
to effectively use hydrogen in different societal applications. In the realm
of energy storage, hydrogen\u27s role is particularly significant. Hydrogen energy storage
offers a promising avenue for addressing intermittency issues in renewable energy
sources such as wind and solar power. Research was performed to investigate
hydrogen-based energy storage and its integration with a coal-fired plant. Additionally,
metal hydride-based hydrogen storage system modeling was also investigated
and shows promise for application in natural gas combined cycle (NGCC) plants.
These investigations employed multi-scale modeling, a computational approach that
integrates different levels of detail, ranging from quantum mechanics to classical
molecular dynamics, to system-level process modeling of the diverse phenomena
occurring in hydrogen systems. Machine learning techniques, on the other hand,
have gained prominence in recent years as powerful tools for data-driven discovery
and prediction. Multiscale modeling and machine learning techniques can enhance
our understanding of hydrogen and its interactions with other materials. This integrated
approach was also applied to facilitate the discovery of new materials with
tailored hydrogen-related properties, reducing the reliance on more expensive and
time-consuming approaches.</p
Exploring and Enhancing Quantum-Inspired Hamiltonian Monte Carlo for Data Science Applications and Optimization
No full textThis dissertation explores advancements and applications in machine learning problems through the development and implementation of Quantum-Inspired Hamiltonian Monte Carlo (QHMC) method, which is an efficient way to sample from a broad class of distributions by allowing a particle to have a random mass. Particularly, we focus on three projects, including missing data imputation, soft-constrained Gaussian process~(GP) regression and parameter estimation. In the first project, we introduce a hybrid technique combining Bayesian inference and QHMC method for imputation of missing datasets. This data imputation method uses stochastic gradient optimization in QHMC to avoid calculating the full gradient on the entire dataset when evolving the Hamiltonian system. We integrate stochastic gradient QHMC with first-order Langevin dynamics to generate samples that converge to the true posterior distribution.
In the second project, we design a GP regression framework for enforcing the physical constraints in a probabilistic way. Specifically, we consider inequality and monotonicity constraints under the GP regression framework. The GP model is trained via QHMC with an adaptive learning strategy to effectively identify constraint locations and reduce uncertainty.
The third project involves a direct application of QHMC to enhance the performance of predicting physical properties of organic molecules used in redox flow batteries. Similar to the second project, a QHMC-based machine learning model is designed under a GP framework.
All aforementioned algorithms are evaluated across multiple examples considering the accuracy, robustness and efficiency criteria. It is reported that our methods demonstrate superior performance compared to the state-of-the-art algorithms by improving the efficiency while maintaining the accuracy.</p
How Does Generative AI Usage Affect the Coding Performance of Developers? - Project Summary
No full textThis project examines how Generative AI tools impact developers\u27 coding performance, using empirical analysis and proprietary data from an IT organization.Generative AI (GenAI) tools, such as GitHub Copilot, have emerged as transformative technologies in software development, offering the potential to enhance coding efficiency through real-time suggestions. However, the impact of GenAI on developers\u27 coding performance remains underexplored. This study investigates the effects of GenAI usage on coding productivity by leveraging a difference-in-differences (DID) analysis of 27 weeks of proprietary data from a mid-sized global information technology organization. Results reveal a significant increase in the number of user stories completed per week among developers using GenAI tools. Additionally, we examine the moderating role of developers\u27 working experience, identifying nuanced mechanisms by which GenAI tools influence performance. These findings contribute to theoretical advancements in understanding GenAI\u27s role in software development and offer practical insights for optimizing AI tool adoption in professional settings. The paper, sharing the same title as this project, will be presented at the Hawaii International Conference on System Sciences (HICSS) in January 2025, a premier academic research conference on information systems
A Machine Learning Approach to Discovering Exoplanets Transiting Old Stars
No full textThis random forest algorithm categorizes old stars as exoplanet hosts based on transits found in Transiting Exoplanet Survey Satellite photometry.Most stars have planets orbiting them, and some cross in front of the host star causing a temporary decrease in brightness known as a transit. These transits have measurable depth, period and duration. Only ~6% of known transiting planets orbit older host stars, and the rest orbit stars with ages similar to our sun1. As a result of stellar evolution, transits of old stars are sometimes missed by transit searches of young stars. The goal of this large-scale search is to quickly discover enough exoplanet candidates orbiting old stars to assist population level analysis. A manual search for exoplanet transits creates an unreasonable timeline. We run Box Least Squares (BLS) on 90,000 old stars using photometry from the Transiting Exoplanet Survey Satellite (TESS). BLS is a period search algorithm that fits parameters such as orbital period, depth, duration and signal to noise of planet transits whether or not there is a real planet. TESS is a near all-sky space-based photometric survey with a primary mission of discovering transiting exoplanets. Using BLS from a set of injected planet transits and a set of inverted lightcurves, we are able to train a random forest algorithm to classify old stars into hosts or non-hosts. This allows us to reduce vetting time greatly. We identified 17 previously known exoplanet candidates in our set, which shows that our methods are successful. We then discovered 32 new exoplanet hosts, which we are following up with ground based observations currently.</p
Exploring the Integration of AI into Journalistic Pre-Professional Workflows and Its Impact on Co-Intelligence
No full textThis project explores how pre-professional journalists integrate AI into workflows, assessing its impact on ethics, creativity, and productivity.This research examines the integration of artificial intelligence (AI) into the workflows of pre-professional journalists and explores its implications for journalistic ethics. With a focus on human-AI employment, the study seeks to understand how AI tools are perceived, adopted, and used by student journalists. Participants from a collegiate student newspaper will engage in an survey to first better understand the newsroom attitudes and culture around AI knowledge and engagement. The study will analyze how varying levels of familiarity with AI and general technology savviness influence the integration of AI into journalistic workflows, and how these factors affect the perceived quality and ethics of the resulting work. Following that, a small subsection of students will participate in a one-hour observational think-aloud session. During these sessions, participants will verbalize their cognitive processes and decision-making while using AI tools to assist with writing tasks. This approach will capture not only the practical application of AI but also the ethical considerations that arise as journalists interact with these technologies. By comparing AI-assisted writing to traditional writing methods, the research will assess the cognitive impact of AI and explore its potential as an augmentation tool, rather than a replacement for human journalists. Ultimately, the research aims to provide insights into the sustainable, ethical implementation of AI in newsrooms, offering guidelines that preserve journalistic integrity while enhancing productivity and creativity. These findings will contribute to the broader conversation on the evolving relationship between AI and journalism, helping to shape future best practices for ethical AI adoption in the field.</p
Competition in Remanufacturing with Asymmetric Demand Information
No full textThis paper examines remanufacturing decisions in the context of outsourcing, which have important implications for environmental and economic sustainability. Specifically, we model the competition between an experienced Original Equipment Manufacturer (OEM) and an emerging Independent Remanufacturer (IR). The OEM can decide the manufacturing quantities of a brand-new product, and the IR can collect the OEM’s used products and remanufacture them for resale. The information structure is asymmetric, as only the OEM knows the market size. We identify the equilibrium quantities of both firms, which are shown to be strongly influenced by the IR’s cost efficiency and the consumers’ willingness to pay for the IR’s products. Asymmetric information also plays an important role. Is it always better to hide information? Interestingly, the OEM makes the most profit when the IR has full information on the market size. We find that when the market size is high, the OEM’s and IR’s production and encroachment decisions are the same as when both parties have equal information. The OEM also does not benefit from hiding market information from the IR when the market size is low. Indeed, if the IR’s cost efficiency is moderate and the market size is low, the OEM’s profits are actually hurt by hiding market information. Here, the diminished profits from hiding market information arises from the OEM’s substantially reduced production quantity to prevent IR encroachment. The OEM’s production quantity is higher if the OEM shares market information and the IR encroaches on the market. Thus, by sharing information, the OEM’s benefit gained from increased production quantity outweighs the cost of losing its monopoly. Additionally, consumer surplus increases when the IR engages in remanufacturing, while social surplus increases only when either the OEM’s or IR’s product is strongly favored. Even if the IR does not engage in remanufacturing, the resulting OEM monopoly can still lead to a higher environmental impact under certain market conditions. This arises when the OEM lowers production quantities when the IR encroaches on the market, thereby improving the overall environmental impact. Therefore, policymakers seeking to improve environmental and economic sustainability by encouraging IRs must consider these complex competition dynamics and consumer preferences, as they indirectly influence OEMs’ production decisions.</jats:p
Hyperoctahedral group characters and a type-BC analog of graph coloring
No full textWe state combinatorial formulas for hyperoctahedral group ()
character evaluations of the form , where is a type-BC Kazhdan-Lusztig basis element, with
corresponding to simultaneously smooth type-B and C Schubert varieties. We also
extend the definition of symmetric group codominance to elements of and show that for each element above, there exists a
BC-codominant element satisfying for all -characters . Combinatorial structures and maps appearing in these
formulas are type-BC extensions of planar networks, unit interval orders,
indifference graphs, poset tableaux, and colorings. Using the ring of type-BC
symmetric functions, we introduce natural generating functions for the above
evaluations. These provide a new type-BC analog of Stanley\u27s chromatic
symmetric functions [Adv. Math. 111 (1995) pp. 166-194]
Peridynamic Neural Operators: A Data-Driven Nonlocal Constitutive Model for Complex Material Responses
No full textNeural operators, which can act as implicit solution operators of hidden
governing equations, have recently become popular tools for learning the
responses of complex real-world physical systems. Nevertheless, most neural
operator applications have thus far been data-driven and neglect the intrinsic
preservation of fundamental physical laws in data. In this work, we introduce a
novel integral neural operator architecture called the Peridynamic Neural
Operator (PNO) that learns a nonlocal constitutive law from data. This neural
operator provides a forward model in the form of state-based peridynamics, with
objectivity and momentum balance laws automatically guaranteed. As
applications, we demonstrate the expressivity and efficacy of our model in
learning complex material behaviors from both synthetic and experimental data
sets. We show that, owing to its ability to capture complex responses, our
learned neural operator achieves improved accuracy and efficiency compared to
baseline models that use predefined constitutive laws. Moreover, by preserving
the essential physical laws within the neural network architecture, the PNO is
robust in treating noisy data. The method shows generalizability to different
domain configurations, external loadings, and discretizations