Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “data discovery”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,315 records · Page 73

Large language model-driven database for thermoelectric materials

Thermoelectric materials have the ability to convert waste heat into electricity, offering a valuable solution for energy harvesting. However, their widespread use is hindered by low conversion efficiency, the reliance on expensive rare earth elements, and the environmental and regulatory concerns associated with lead-based materials. A fast and cost-effective way to identify highly efficient thermoelectric materials is through data-driven methods. These approaches rely on robust and comprehensive datasets to train models. Although there are several databases on thermoelectric materials, there is still a need to collect and integrate experimental data from peer-reviewed research articles to capture diverse compositions and properties of materials. Here, in this work, we developed a comprehensive database of 7,123 thermoelectric compounds, containing key information such as chemical composition, structural detail, seebeck coefficient, electrical and thermal conductivity, power factor, and figure of merit (ZT). We used the GPTArticleExtractor workflow, powered by large language models (LLM), to extract and curate data automatically from the scientific literature published in Elsevier journals. This process enabled the creation of a structured database that addresses the challenges of manual data collection. The open access database could stimulate data-driven research and advance thermoelectric material analysis and discovery.

Database↗

Neural entropy-stable conservative flux form neural networks for learning hyperbolic conservation laws

We propose a neural entropy-stable conservative flux form neural network (NESCFN) for learning hyperbolic conservation laws and their associated entropy functions directly from solution trajectories, without requiring any predefined numerical discretization. While recent neural network architectures have successfully integrated classical numerical principles into learned models, most rely on prior knowledge of the governing equations or assume a fixed discretization. Our approach removes this dependency by embedding entropy-stable design principles into the learning process itself, enabling the discovery of physically consistent dynamics in a fully data-driven setting. By jointly learning both the flux function and a corresponding entropy, NESCFN promotes conservation and entropy dissipation, which is critical for long-term stability and fidelity in the system of hyperbolic conservation laws. Furthermore, numerical results demonstrate that the method achieves stability and conservation over extended time horizons and accurately captures shock propagation speeds, even without oracle access to future-time solution profiles in the training data.

Conservative flux form↗

Characterizing Defect Dynamics in Silicon Carbide Using Symmetry-Adapted Collective Variables and Machine Learning Interatomic Potentials

Silicon carbide (SiC) divacancies are attractive candidates for spin-defect qubits possessing long coherence times and optical addressability. The high activation barriers associated with SiC defect formation and motion pose challenges for their study by first-principles molecular dynamics. In this work, we develop and deploy machine learning interatomic potentials (MLIPs) to accelerate defect dynamics simulations while retaining ab initio accuracy. We employ an active learning strategy comprising symmetry-adapted collective variable discovery and enhanced sampling to compile configurationally diverse training data, calculation of energies and forces using density functional theory (DFT), and training of an E(3)-equivariant MLIP based on the Allegro model. Here, the trained MLIP reproduces DFT-level accuracy in defect transition activation free energy barriers, enables the efficient and stable simulation of multidefect 216-atom supercells, and permits an analysis of the temperature dependence of defect thermodynamic stability and formation/annihilation kinetics to propose an optimal annealing temperature to maximally stabilize VV divacancies.

Computer simulations↗

An informatics guided classification of miscible and immiscible binary alloy systems

The classification of miscible and immiscible systems of binary alloys plays a critical role in the design of multicomponent alloys. By mining data from hundreds of experimental phase diagrams, and thousands of thermodynamic data sets from experiments and high-throughput first-principles (HTFP) calculations, we have obtained a comprehensive classification of alloying behavior for 813 binary alloy systems consisting of transition and lanthanide metals. Among several physics-based descriptors, the slightly modified Pettifor chemical scale provides a unique two-dimensional map that divides the miscible and immiscible systems into distinctly clustered regions. Based on an artificial neural network algorithm and elemental similarity, the miscibility of the unknown systems is further predicted and a complete miscibility map is thus obtained. Impressively, the classification by the miscibility map yields a robust validation on the capability of the well-known Miedema’s theory (95% agreement) and shows good agreement with the HTFP method (90% agreement). Our results demonstrate that a state-of-the-art physics-guided data mining can provide an efficient pathway for knowledge discovery in the next generation of materials design.

36 MATERIALS SCIENCE↗

Uncertainty-aware mixed-variable machine learning for materials design

Abstract Data-driven design shows the promise of accelerating materials discovery but is challenging due to the prohibitive cost of searching the vast design space of chemistry, structure, and synthesis methods. Bayesian optimization (BO) employs uncertainty-aware machine learning models to select promising designs to evaluate, hence reducing the cost. However, BO with mixed numerical and categorical variables, which is of particular interest in materials design, has not been well studied. In this work, we survey frequentist and Bayesian approaches to uncertainty quantification of machine learning with mixed variables. We then conduct a systematic comparative study of their performances in BO using a popular representative model from each group, the random forest-based Lolo model (frequentist) and the latent variable Gaussian process model (Bayesian). We examine the efficacy of the two models in the optimization of mathematical functions, as well as properties of structural and functional materials, where we observe performance differences as related to problem dimensionality and complexity. By investigating the machine learning models’ predictive and uncertainty estimation capabilities, we provide interpretations of the observed performance differences. Our results provide practical guidance on choosing between frequentist and Bayesian uncertainty-aware machine learning models for mixed-variable BO in materials design.

36 MATERIALS SCIENCE↗

Roadmap for transforming heterogeneous catalysis with artificial intelligence

Artificial intelligence (AI) is poised to transform heterogeneous catalysis, opening avenues for catalytic materials discovery. By uncovering intricate patterns in high-dimensional data, AI has been reshaping our pursuit of sustainable catalytic processes across the energy, environmental and chemical sectors. This promise, however, hinges on overcoming fundamental barriers, including limitations in data availability and quality, challenges in the generalizability and interpretability of data-augmented decisions, and the persistent gap between in silico predictions and experiments. Furthermore, we outline a forward-looking roadmap for deeply integrating AI into heterogeneous catalysis with an AI-ready data ecosystem, multimodal foundation models, and ultimately autonomous laboratories to accelerate the development of next-generation catalytic technologies via AI-empowered human–machine collaboration.

Computational methods↗

Huntingtin structure is orchestrated by HAP40 and shows a polyglutamine expansion-specific interaction with exon 1

Huntington’s disease results from expansion of a glutamine-coding CAG tract in the huntingtin (HTT) gene, producing an aberrantly functioning form of HTT. Both wildtype and disease-state HTT form a hetero-dimer with HAP40 of unknown functional relevance. Here, we demonstrate in vivo and in cell models that HTT and HAP40 cellular abundance are coupled. Integrating data from a 2.6 Å cryo-electron microscopy structure, cross-linking mass spectrometry, small-angle X-ray scattering, and modeling, we provide a near-atomic-level view of HTT, its molecular interaction surfaces and compacted domain architecture, orchestrated by HAP40. Native mass spectrometry reveals a remarkably stable hetero-dimer, potentially explaining the cellular inter-dependence of HTT and HAP40. The exon 1 region of HTT is dynamic but shows greater conformational variety in the polyglutamine expanded mutant than wildtype exon 1. Our data provide a foundation for future functional and drug discovery studies targeting Huntington’s disease and illuminate the structural consequences of HTT polyglutamine expansion.

59 BASIC BIOLOGICAL SCIENCES↗

A Statistical Learning Framework for Materials Science: Application to Elastic Moduli of k -nary Inorganic Polycrystalline Compounds

Materials scientists increasingly employ machine or statistical learning (SL) techniques to accelerate materials discovery and design. Such pursuits benefit from pooling training data across, and thus being able to generalize predictions over, k -nary compounds of diverse chemistries and structures. This work presents a SL framework that addresses challenges in materials science applications, where datasets are diverse but of modest size, and extreme values are often of interest. Our advances include the application of power or Hölder means to construct descriptors that generalize over chemistry and crystal structure, and the incorporation of multivariate local regression within a gradient boosting framework. The approach is demonstrated by developing SL models to predict bulk and shear moduli (K and G, respectively) for polycrystalline inorganic compounds, using 1,940 compounds from a growing database of calculated elastic moduli for metals, semiconductors and insulators. The usefulness of the models is illustrated by screening for superhard materials.

36 MATERIALS SCIENCE↗

Relic density and future colliders: inverse problem(s)

Relic density calculations are often used to constrain particle physics models, and in particular supersymmetry. We will show that the presence of additional energy or entropy before the Big-Bang nucleosynthesis can however completely change the relic density constraints on the SUSY parameter space. Therefore one should be extremely careful when using the relic density to constrain supersymmetry as it could give misleading results, especially if combined with the future collider data. Alternatively, we will also show that combining the discoveries of the future colliders with relic density calculations can shed light on the inaccessible pre-BBN dark time physics. Finally we will present SuperIso Relic, a new relic density calculator code in Supersymmetry, which incorporates alternative cosmological models, and is publicly available.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

IS MAGNETIC RECONNECTION THE CAUSE OF SUPERSONIC UPFLOWS IN GRANULAR CELLS?

In a previous work, we reported on the discovery of supersonic magnetic upflows on granular cells in data from the SUNRISE/IMaX instrument. In the present work, we investigate the physical origin of these events employing data from the same instrument but with higher spectral sampling. By means of the inversion of Stokes profiles we are able to recover the physical parameters (temperature, magnetic field, line-of-sight velocity, etc.) present in the solar photosphere at the time of these events. The inversion is performed in a Monte-Carlo-like fashion, that is, repeating it many times with different initializations and retaining only the best result. We find that many of the events are characterized by a reversal in the polarity of the magnetic field along the vertical direction in the photosphere, accompanied by an enhancement in the temperature and by supersonic line-of-sight velocities. In about half of the studied events, large blueshifted and redshifted line-of-sight velocities coexist above/below each other. These features can be explained in terms of magnetic reconnection, where the energy stored in the magnetic field is released in the form of kinetic and thermal energy when magnetic field lines of opposite polarities coalesce. However, the agreement with magnetic reconnection is not perfect and, therefore, other possible physical mechanisms might also play a role.

79 ASTRONOMY AND ASTROPHYSICS↗

IMPROVED AND QUALITY-ASSESSED EMISSION AND ABSORPTION LINE MEASUREMENTS IN SLOAN DIGITAL SKY SURVEY GALAXIES

We present a new database of absorption and emission-line measurements based on the entire spectral atlas from the Sloan Digital Sky Survey (SDSS) 7th data release of galaxies within a redshift of 0.2. Our work makes use of the publicly available penalized pixel-fitting (pPXF) and gas and absorption line fitting (gandalf) IDL codes, aiming to improve the existing measurements for stellar kinematics, the strength of various absorption-line features, and the flux and width of the emissions from different species of ionized gas. Our fit to the stellar continuum uses both standard stellar population models and empirical templates obtained by combining a large number of stellar spectra in order to fit a subsample of high-quality SDSS spectra for quiescent galaxies. Furthermore, our fit to the nebular spectrum includes an exhaustive list of both recombination and forbidden lines. Foreground Galactic extinction is implicitly treated in our models, whereas reddening in the SDSS galaxies is included in the form of a simple dust screen component affecting the entire spectrum that is accompanied by a second reddening component affecting only the ionized gas emission. In order to check for systematic departures from the rather standard set of assumptions that enters our models, we provide a quality assessment for our fit to the SDSS spectra in our sample, for both the stellar continuum and the nebular emissions and across different wavelength regions. This quality assessment also allows the identification of objects with either problematic data or peculiar features. We hope to foster the discovery potential of our database; therefore, our spectral fit is available to the community. For example, based on the quality assessment around the H{alpha} and [N II] {lambda}6584 lines, approximately 1% of the SDSS spectra classified as 'galaxies' by the SDSS pipeline do in fact require additional broad lines to be matched, even though they do not show a strong continuum from an active nucleus, as do the SDSS objects classified as 'quasars'. Finally, we provide new spectral templates for galaxies of different Hubble types, obtained by combining the results of our spectral fit for a subsample of 452 morphologically selected objects.

79 ASTRONOMY AND ASTROPHYSICS↗

KEPLER MISSION STELLAR AND INSTRUMENT NOISE PROPERTIES

Kepler mission results are rapidly contributing to fundamentally new discoveries in both the exoplanet and asteroseismology fields. The data returned from Kepler are unique in terms of the number of stars observed, precision of photometry for time series observations, and the temporal extent of high duty cycle observations. As the first mission to provide extensive time series measurements on thousands of stars over months to years at a level hitherto possible only for the Sun, the results from Kepler will vastly increase our knowledge of stellar variability for quiet solar-type stars. Here, we report on the stellar noise inferred on the timescale of a few hours of most interest for detection of exoplanets via transits. By design the data from moderately bright Kepler stars are expected to have roughly comparable levels of noise intrinsic to the stars and arising from a combination of fundamental limitations such as Poisson statistics and any instrument noise. The noise levels attained by Kepler on-orbit exceed by some 50% the target levels for solar-type, quiet stars. We provide a decomposition of observed noise for an ensemble of 12th magnitude stars arising from fundamental terms (Poisson and readout noise), added noise due to the instrument and that intrinsic to the stars. The largest factor in the modestly higher than anticipated noise follows from intrinsic stellar noise. We show that using stellar parameters from galactic stellar synthesis models, and projections to stellar rotation, activity, and hence noise levels reproduce the primary intrinsic stellar noise features.

79 ASTRONOMY AND ASTROPHYSICS↗

ANDROMEDA XXVIII: A DWARF GALAXY MORE THAN 350 kpc FROM ANDROMEDA

We report the discovery of a new dwarf galaxy, Andromeda XXVIII, using data from the recently released Sloan Digital Sky Survey Data Release 8. The galaxy is a likely satellite of Andromeda, and, at a separation of 365{sup +17}{sub -1} kpc, would be one of the most distant of Andromeda's satellites. Its heliocentric distance is 650{sup +150}{sub -80} kpc, and analysis of its structure and luminosity shows that it has an absolute magnitude of M{sub V} = -8.5{sup +0.4}{sub -1.0} and half-light radius of r{sub h} = 210{sup +60}{sub -50} pc, similar to many other faint Local Group dwarfs. With presently available imaging we are unable to determine whether there is ongoing or recent star formation, which prevents us from classifying it as a dwarf spheroidal or a dwarf irregular.

79 ASTRONOMY AND ASTROPHYSICS↗

Unified feature association networks through integration of transcriptomic and proteomic data

High-throughput multi-omics studies and corresponding network analyses of multi-omic data have rapidly expanded their impact over the last 10 years. As biological features of different types (e.g. transcripts, proteins, metabolites) interact within cellular systems, the greatest amount of knowledge can be gained from networks that incorporate multiple types of -omic data. However, biological and technical sources of variation diminish the ability to detect cross-type associations, yielding networks dominated by communities comprised of nodes of the same type. We describe here network building methods that can maximize edges between nodes of different data types leading to integrated networks. We systematically rank several network inference methods and demonstrate that, in many cases, using a random forest method, GENIE3, produces the most integrated networks. This increase in integration does not come at the cost of accuracy as GENIE3 produces networks of approximately the same quality as the other network inference methods tested here. Using GENIE3, we also infer networks representing antibody-mediated Dengue virus cell invasion and receptor-mediated Dengue virus invasion. A number of functional pathways showed centrality differences between the two networks including genes responding to both GM-CSF and IL-4, which had a higher centrality value in an antibody-mediated vs. receptor-mediated Dengue network. Because a biological system involves the interplay of many different types of molecules, incorporating multiple data types into networks will improve their use as models of biological systems. The methods explored here are some of the first to specifically highlight and address the challenges associated with how such multi-omic networks can be assembled and how the greatest number of interactions can be inferred from different data types. The resulting networks can lead to the discovery of new host response patterns and interactions during viral infection, generate new hypotheses of pathogenic mechanisms and confirm disease associated mechanisms of disease.

59 BASIC BIOLOGICAL SCIENCES↗

Integrated Direct Air Capture and H₂-Free CO₂ Valorization

This project advances fundamental understanding of a novel integrated direct air capture (DAC) and CO₂ conversion process that valorizes atmospheric CO₂ without external H₂. The research encompasses four critical components: (1) design of task-specific ionic liquids for efficient CO₂ capture under ambient conditions, (2) development of H₂-free tandem catalytic systems using ethane as a reductant, (3) advanced operando characterization to elucidate capture and conversion mechanisms, and (4) data science-driven predictive computation to accelerate material discovery. Over the project period, we developed five high-performance DAC sorbent systems—including CaO/superbase ionic liquid composites, Ni-MOF/Ionic Liquid (IL) hybrids, fluorinated covalent organic frameworks with ion-pair functional groups, defect-engineered UiO-66, and a validated kinetic model for humid-condition operation, achieving CO₂ capacities up to 1.86 mmol/g at 400 ppm with excellent cycling stability. For H₂-free conversion, we constructed atomically synergistic Zn–O–Cr binuclear catalytic sites that achieve 100% ethylene selectivity, ~9.6% ethane conversion, and 99% CO₂ utilization in equimolar co-conversion of ethane and CO₂. We further demonstrated downstream valorization pathways converting CO and C₂H₄ into polyketones and C₃ chemicals. These advances strengthen the scientific foundation for producing value-added materials from ambient CO₂.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Shift Happens: Building Robust AI Models with Domain Adaptation

Artificial Intelligence (AI) is revolutionizing physics research—from probing the large-scale structure of the Universe to modeling subatomic interactions and fundamental forces. Yet, a major challenge persists: AI models trained on simulations or old experiment / astronomical survey often perform poorly when applied to new data—exposing issues of dataset (domain) shift, model robustness, and uncertainty in predictions. This summer school session will introduce students to common challenges in applying AI across domains and present solutions based on domain adaptation—a set of techniques designed to improve model generalization under domain shift. We will cover foundational ideas, practical strategies, and current research frontiers in this area. Through examples in astrophysics, we'll explore how domain adaptation can help bridge the gap between synthetic and real-world data, improve trust in model outputs, and advance scientific discovery. The concepts discussed are broadly applicable across physics and other scientific disciplines, making this a valuable topic for anyone interested in building robust, transferable AI models for science.

Ciprijanovic, A. [Fermilab] (ORCID:000000031281719↗

AI in Astrophysics: Tackling Domain Shift, Model Robustness and Uncertainty

Artificial Intelligence (AI) is revolutionizing physics research from probing the large-scale structure of the Universe to modeling subatomic interactions and fundamental forces. Yet, a major challenge persists: AI models trained on simulations or old experiment / astronomical survey often perform poorly when applied to new data exposing issues of dataset (domain) shift, model robustness, and uncertainty in predictions. This talk will introduce common challenges in applying AI across domains and present solutions based on domain adaptation a set of techniques designed to improve model generalization under domain shift. We will cover foundational ideas, practical strategies, and current research frontiers in this area. Through examples in astrophysics, we'll explore how domain adaptation can help bridge the gap between synthetic and real-world data, improve trust in model outputs, and advance scientific discovery.

Ciprijanvoic, Aleksandra [Fermilab] (ORCID:0000000↗

Interpreting Mobile and Handheld Air Sensor Readings in Relation to Air Quality Standards and Health Effect Reference Values: Tackling the Challenges

The US Environmental Protection Agency (EPA) and other federal agencies face a number of challenges in interpreting and reconciling short-duration (seconds to minutes) readings from mobile and handheld air sensors with the longer duration averages (hours to days) associated with the National Ambient Air Quality Standards (NAAQS) for the criteria pollutants-particulate matter (PM), ozone, carbon monoxide, lead, nitrogen oxides, and sulfur oxides. Similar issues are equally relevant to the hazardous air pollutants (HAPs) where chemical-specific health effect reference values are the best indicators of exposure limits; values which are often based on a lifetime of continuous exposure. A multi-agency, staff-level Air Sensors Health Group (ASHG) was convened in 2013. ASHG represents a multi-institutional collaboration of Federal agencies devoted to discovery and discussion of sensor technologies, interpretation of sensor data, defining the state of sensor-related science across each institution, and provides consultation on how sensors might effectively be used to meet a wide range of research and decision support needs. ASHG focuses on several fronts: improving the understanding of what hand-held sensor technologies may be able to deliver; communicating what hand-held sensor readings can provide to a number of audiences; the challenges of how to integrate data generated by multiple entities using new and unproven technologies; and defining best practices in communicating health-related messages to various audiences. This review summarizes the challenges, successes, and promising tools of those initial ASHG efforts and Federal agency progress on crafting similar products for use with other NAAQS pollutants and the HAPs.

54 ENVIRONMENTAL SCIENCES↗