Search NASA⌕ Search

SEARCH · Search NASA

Results for “embedding model”

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 649 records · Page 36
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

A Chemist’s Guide to Solid-State Qubits: Diamond Color Centers as Embedded Vacancy-Centered Molecules

Diamond point defects or color centers are at the crux of solid-state quantum technology applications, from computing to sensing. Modeling molecular qubits has advanced by leaps and bounds mainly due to the advanced quantum mechanical (QM) tools in the arsenal of quantum chemists for modeling (isolated) molecular systems, which should be easily translatable for analogous systems in the solid state. Here, we simulate a series of diamond defects: a carbon vacancy (V C ) and second row substitutions that are adjacent to a V C or D C V C , with D = B, N, and O, to evaluate their magnetic and optical properties. We applied molecular orbital principles, an embedded cluster method, and a multiconfigurational QM theory to computationally characterize the above-mentioned defects for a multitude of spin, charge, and excited states. Our computational approach delivered high-quality QM insights that compare well with experiments. Furthermore, with this framework, we confirm the spin-triplets B C V C – and O C V C 0 to be analogous to N C V C – , with calculated bright transitions that respectively correspond to UV (3.80 eV) and blue (2.65 eV) light absorption and emission lifetimes (τ) of 2.3 and 5.6 ns, which are higher in energy and more short-lived (and thus brighter) than that for N C V C – : green (2.26 eV), τ = 9.5 ns.

Martirez, John Mark P. [Princeton Plasma Physics L↗
NASA NTRS1986

Application of a computational model for vortex generators in subsonic internal flows

A model for the analysis of vortex generators in a fully viscous subsonic internal flow is evaluated. A vorticity source term is used in a modified form of the Parabolized Navier-Stokes equations to model the shed vortex. Computed results are compared with idealized flow vortex paths, and with experimental data for vortex generators embedded in a thick turbulent boundary layer. The analysis is also compared with experimental data for a separated diffusing S-duct and for a diffusing S-duct with vortex generators. Quantitative comparisons are shown for the latter three cases. Emphasis is placed on verifying the ability of the model to predict global distortions in the flow field.

Kunik, W. G.↗
NASA NTRS1986

Application of a computational model for vortex generators in subsonic internal flows

A model for the analysis of vortex generators in a fully viscous subsonic internal flow is evaluated. A vorticity source term is used in a modified form of the Parabolized Navier-Stokes equations to model the shed vortex. Computed results are compared with idealized flow vortex paths, and with experimental data for vortex generators embedded in a thick turbulent boundary layer. The analysis is also compared with experimental data for a separated diffusing S-duct and for a diffusing S-duct with vortex generators. Quantitative comparisons are shown for the latter three cases. Emphasis is placed on verifying the ability of the model to predict global distortions in the flow field.

Kunik, W. G.↗
NASA NTRS1983

Mathematical modeling of damage in unidirectional composites

Extending the work of Goree and Gross (1979), solutions are given for a two-dimensional region of unidirectional fibers embedded in an elastic matrix whose initial flaw may take the form of a transverse notch, a rectangular cutout, or a circular hole. Subsequent flaw-induced damage is generated by remote stresses acting parallel to the fibers. For the case of such ductile matrix composites as boron/aluminum, present results indicate that both longitudinal matrix yielding and transverse notch extension must be included in order for the model to agree with experimental results. Little difference is found for the three types of initial damage considered. In all cases, the presence of additional damage changes the nature of stress distribution through the unbroken fibers.

Goree, J. G.↗
NASA NTRS2010

Verify MesoNAM Performance

The AMU conducted an objective analysis of the MesoNAM forecasts compared to observed values from sensors at specified KSC/CCAFS wind towers by calculating the following statistics to verify the performance of the model: 1) Bias (mean difference), 2) Standard deviation of Bias, 3) Root Mean Square Error (RMSE), and 4) Hypothesis test for Bias = O. The 45 WS LWOs use the MesoNAM to support launch weather operations. However, the actual performance of the model at KSC and CCAFS had not been measured objectively. The analysis compared the MesoNAM forecast winds, temperature and dew point to the observed values from the sensors on wind towers. The data were stratified by tower sensor, month and onshore/offshore wind direction based on the orientation of the coastline to each tower's location. The model's performance statistics were then calculated for each wind tower based on sensor height and model initialization time. The period of record for the data used in this task was based on the operational start of the current MesoNAM in mid-August 2006 and so the task began with the first full month of data, September 2006, through May 2010. The analysis of model performance indicated: a) The accuracy decreased as the forecast valid time from the model initialization increased, b) There was a diurnal signal in T with a cool bias during the late night and a warm bias during the afternoon, c) There was a diurnal signal in Td with a low bias during the afternoon and a high bias during the late night, and d) The model parameters at each vertical level most closely matched the observed parameters at heights closest to those vertical levels. The AMU developed a GUI that consists of a multi-level drop-down menu written in JavaScript embedded within the HTML code. This tool allows the LWO to easily and efficiently navigate among the charts and spreadsheet files containing the model performance statistics. The objective statistics give the LWOs knowledge of the model's strengths and weaknesses and the GUI allows quick access to the data which will result in improved forecasts for operations.

Bauman, William H., III↗
DOE OSTI2024

E(n)-Equivariant cartesian tensor message passing interatomic potential

Machine learning potential (MLP) has been a popular topic in recent years for its capability to replace expensive first-principles calculations in some large systems. Meanwhile, message passing networks have gained significant attention due to their remarkable accuracy, and a wave of message passing networks based on Cartesian coordinates has emerged. However, the information of the node in these models is usually limited to scalars, and vectors. In this work, we propose High-order Tensor message Passing interatomic Potential (HotPP), an E(n) equivariant message passing neural network that extends the node embedding and message to an arbitrary order tensor. By performing some basic equivariant operations, high order tensors can be coupled very simply and thus the model can make direct predictions of high-order tensors such as dipole moments and polarizabilities without any modifications. The tests in several datasets show that HotPP not only achieves high accuracy in predicting target properties, but also successfully performs tasks such as calculating phonon spectra, infrared spectra, and Raman spectra, demonstrating its potential as a tool for future research.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Mitigating Algorithmic Bias in Cancer Site Classification Models

Purpose Integrating artificial intelligence in cancer diagnostics has improved tumor classification beyond rule-based systems. Despite these advancements, these models may still encode demographic biases. We conducted a large-scale, applied bias-probing study of a deep learning–based cancer site classifier to quantify race information encoded in document embeddings. We then evaluated how performance changes when race-correlated embedding dimensions are removed in a post-training sensitivity analysis. Methods The cancer site classifier was trained using 3.5 million electronic cancer pathology reports from six of the National Cancer Institute's SEER registries. We trained a hierarchical self-attention network to generate 400-dimensional document embeddings. These embeddings were used to train two downstream, gradient-boosted decision tree classifiers: one to classify the cancer sites and another to predict racial categories. We identified overlapping features by intersecting the top 50 feature-importance rankings from the site and race models and computed their cumulative feature importance in each model. As a post hoc sensitivity analysis, we progressively pruned these overlapping dimensions, retrained the site model, and compared overall macro-F1 and accuracy, race-stratified macro-F1, and group fairness metrics on the basis of demographic parity and equalized odds before and after pruning. Results The analysis revealed minimal feature overlap between the cancer site and race prediction models, and the cumulative importance scores indicated a negligible influence of racial information on clinical predictions. Post-training pruning of overlapping features did not compromise the models' diagnostic accuracy, with a 0.07% loss in accuracy. Conclusion Our findings demonstrate that HiSAN-generated embeddings from SEER data can be used effectively in cancer site classification without significant demographic bias influencing the outcomes. Post-training pruning therefore functions as a practical audit and sensitivity check.

Shivanna, Abhishek [ORNL] (ORCID:0009000665228593)↗
NASA NTRS1989

Development of a Navier-Stokes algorithm for parallel-processing supercomputers

An explicit flow solver, applicable to the hierarchy of model equations ranging from Euler to full Navier-Stokes, is combined with several techniques designed to reduce computational expense. The computational domain consists of local grid refinements embedded in a global coarse mesh, where the locations of these refinements are defined by the physics of the flow. Flow characteristics are also used to determine which set of model equations is appropriate for solution in each region, thereby reducing not only the number of grid points at which the solution must be obtained, but also the computational effort required to get that solution. Acceleration to steady-state is achieved by applying multigrid on each of the subgrids, regardless of the particular model equations being solved. Since each of these components is explicit, advantage can readily be taken of the vector- and parallel-processing capabilities of machines such as the Cray X-MP and Cray-2.

Swisshelm, Julie M.↗
NASA NTRS1988

Development of an aerodynamics algorithm for parallel-processing supercomputers

An explicit flow solver, applicable to the hierarchy of model equations ranging from Euler to full Navier-Stokes, is combined with several techniques designed to reduce computational expense. The computational domain consists of local grid refinements embedded in a global coarse mesh, where the locations of these refinements are defined by the physics of the flow. Flow characteristics are also used to determine which set of model equations is appropriate for solution in each region, thereby reducing not only the number of grid points at which the solution must be obtained, but also the computational effort required to get that solution. Acceleration to steady-state is achieved by applying multigrid on each of the subgrids, regardless of the particular model equations being solved. Since each of these components is explicit, advantage can readily be taken of the vector- and parallel-processing capabilities of machines such as the Cray X-MP and Cray-2.

Swisshelm, Julie M.↗
NASA NTRS2011

Episodes of Cross-Polar Transport in the Arctic Troposphere During July 2008 as Seen from Models, Satellite, and Aircraft Observations

During the POLARCAT summer campaign in 2008, two episodes (2 5 July and 7 10 July 2008) occurred where low-pressure systems traveled from Siberia across the Arctic Ocean towards the North Pole. The two cyclones had extensive smoke plumes from Siberian forest fires and anthropogenic sources in East Asia embedded in their associated air masses, creating an excellent opportunity to use satellite and aircraft observations to validate the performance of atmospheric transport models in the Arctic, which is a challenging model domain due to numerical and other complications. Here we compare transport simulations of carbon monoxide (CO) from the Lagrangian transport model FLEXPART and the Eulerian chemical transport model TOMCAT with retrievals of total column CO from the IASI passive infrared sensor onboard the MetOp-A satellite. The main aspect of the comparison is how realistic horizontal and vertical structures are represented in the model simulations. Analysis of CALIPSO lidar curtains and in situ aircraft measurements provide further independent reference points to assess how reliable the model simulations are and what the main limitations are. The horizontal structure of mid-latitude pollution plumes agrees well between the IASI total column CO and the model simulations. However, finer-scale structures are too quickly diffused in the Eulerian model. Applying the IASI averaging kernels to the model data is essential for a meaningful comparison. Using aircraft data as a reference suggests that the satellite data are biased high, while TOMCAT is biased low. FLEXPART fits the aircraft data rather well, but due to added background concentrations the simulation is not independent from observations. The multi-data, multi-model approach allows separating the influences of meteorological fields, model realisation, and grid type on the plume structure. In addition to the very good agreement between simulated and observed total column CO fields, the results also highlight the difficulty to identify a data set that most realistically represents the actual pollution state of the Arctic atmosphere.

Sodemann, H.↗
DOE OSTI2026

Semi-Analytical Hierarchical Bayesian Inference of Nonlinear Model Structure in Stochastic Dynamics: Applied to Compartmental Models of Infectious Diseases

A Bayesian computational framework for parsimonious inference in stochastic nonlinear dynamical systems is presented. This framework enables the concurrent estimation of system states, time-varying parameters, time-invariant parameters, and the optimal sparsity structure of the model parameters. Because differential equation-based models are often simplified mechanistic or phenomenological representations, robust inference from noisy measurement data requires explicit treatment of model error and uncertainty. Model error and time-varying parameters can be represented as random processes, enabling inference while making minimal assumptions about the underlying sources of discrepancy and variability. Adopting stochastic differential equation representations affords the model significant flexibility, but can also render it susceptible to overfitting during statistical inversion, where the inferred model may track noise rather than the underlying signal. To alleviate the effects of overfitting and to enable the discovery of the optimal sparse representation of the time-invariant parameters, a Bayesian sparse learning algorithm is embedded within the framework. This sparse learning framework adopts an approximate hierarchical Bayesian setting defined by a series of semi-analytical expressions. The model structure inference framework is validated using a stochastic compartmental model for tracking and forecasting active cases of an infectious disease. Compartmental models describe population-level infectious disease dynamics through interactions among population fractions grouped by disease state. Mathematically, such models consist of a system of coupled ordinary differential equations. This example adopts an expressive compartmental model that includes multiple possible interactions between disease states, motivated by early uncertainty surrounding COVID-19 reinfection dynamics and their implications for long-term epidemic forecasting. The sparse learning exercise permits the inference of a priori unknown epidemiological dynamics from simulated public health data, discovering the nested compartmental model that optimizes the trade-off between average data-fit and model complexity. It is shown that inducing sparsity among the model parameters eliminates redundant interactions between compartments, equivalently revealing the optimal coupling structure between differential equations.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1990

Embedded function methods for supersonic turbulent boundary layers

The development of embedded functions to represent the mean velocity and total enthalpy distributions in the wall layer of a supersonic turbulent boundary layer is considered. The asymptotic scaling laws (in the limit of large Reynolds number) for high speed compressible flows are obtained to facilitate eventual implementation of the embedded functions in a general prediction method. A self-consistent asymptotic structure is derived, as well as a compressible law of the wall in which the velocity and total enthalpy are logarithmic within the overlap zone, but in the Howarth-Dorodnitsyn variable. Simple outer region turbulence models are proposed (some of which are modifications of existing incompressible models) to reflect the effects of compressibility. As a test of the methodology and the new turbulence models, a set of self-similar outer region profiles is obtained for constant pressure flow; these are then coupled with embedded functions in the wall layer. The composite profiles thus obtained are compared directly with experimental data and good agreement is obtained for flows with Mach numbers up to 10.

He, J.↗
NASA NTRS1998

Estimation of 3-D Cloud Effects on TOMS Satellite Retrieval of Surface UV Irradiance

To improve surface UV irradiance retrieval from the Total Ozone Mapping Spectrometer (TOMS) we simulate errors of the TOMS cloud correction algorithm for summertime broken cloud conditions. Cloud scenes (50 km by 50 km) are modeled by a normal random (Gaussian) field with a fixed lower boundary and conservative scattering. The model relates stochastic field characteristics with the cloud amount, mean cloud diameter and aspect ratio. Clouds are embedded into Rayleigh atmosphere with standard ozone profile. Radiative transfer calculations of the radiance at the top of the atmosphere and irradiance at the surface were performed using 3-D Monte Carlo (MC) code. The results are averaged over the satellite field of view on the surface (50 km by 50 km) and compared with TOMS predicted surface irradiance for the same scene reflectance. The TOMS algorithm assumes horizontally homogeneous Cl-type cloud between 3 km and 5.5 km. The effective optical depth is determined by fitting observed (MC) radiance at 380 nm. Having the same radiance at the satellite the homogeneous and broken cloud models predict different average irradiances at the surface. This is due to the differences in Bidirectional Reflection Distribution Function (BRDF) for homogeneous and broken cloud scenes with the same hemispherical albedo. For typical TOMS observational geometry at mid-latitudes the simulated single pixels errors may be as large as +/- 20%. Qualitatively these errors are due to the dominance of the non-horizontal cloud surfaces, which are not accounted for in the homogeneous cloud model. However, due to high variability of the real cloud shapes and types it is unclear how these single pixel errors would affect TOMS time-integrated UV exposure over extended periods (weeks to months) for different regions.

Krotkov, Nickolay A.↗
DOE OSTI2025

Modeling and Optimization of a Rotating Packed Bed Contactor with a Tetraamine-Appended Metal−Organic Framework for CO 2 Capture

A potential contactor technology for sorbent-based CO 2 capture is the rotating packed bed that contains separate sections for continuous adsorption and desorption. A heat exchanger can be embedded to remove heat in the adsorption section and add heat in the desorption section. In this work, we develop a two-dimensional (2D) model of a rotating packed bed for use in CO 2 capture applications. Mass and energy balances for the model are developed based on a Ljungström-type air preheater, which accounts for the counter-current axial flow of gas phases in separate sections of the bed and the rotation of a solid sorbent, which cycles between adsorption and desorption sections. The sorbent used for this analysis is the tetraamine-appended metal−organic framework Mg 2 (dobpdc)(3−4− 3), chosen for its stability and affinity for CO 2 at low partial pressures, such as those from a natural gas power plant source. An optimization problem is solved that considers the trade-off between maximizing the productivity of the bed and minimizing energy consumption. Maximum productivity and minimum energy are found to be 8.53 kg/h/m 3 and 3.84 MJ/kg, respectively, when these objectives are optimized independently. It is observed that the flue gas pressure and bed rotational speed are the desired operating variables to vary for model-based design of experiments to reduce uncertainty in parameter estimation, as these two variables yielded the most information content based on the Fisher information matrix.

20 FOSSIL-FUELED POWER PLANTS↗
NASA NTRS1992

Comparison of measured and predicted performance of a SIS waveguide mixer at 345 GHz

The measured gain and noise of a SIS waveguide mixer at 345 GHz have been compared with theoretical values, calculated from the quantum mixer theory using a three port model. As a mixing element, we use a series array of two Nb-Al2O3-Nb SIS junctions. The area of each junction is 0.8 sq microns and the normal state resistance is 52 omega. The embedding impedance of the mixer has been determined from the pumped DC-IV curves of the junction and is compared to results from scale model measurements (105 x). Good agreement was obtained. The measured mixer gain, however, is a factor of 0.45 plus or minus 0.5 lower than the theoretical predicted gain. The measured mixer noise temperature is a factor of 4-5 higher than the calculated one. These discrepancies are independent on pump power and are valid for a broad range of tuning conditions.

Honingh, C. E.↗
NASA NTRS2014

Enabling Technologies for Entrepreneurial Opportunities in 3D printing of SmallSats

A consortium of innovative experts in additive manufacturing (AM) comprising Northrup Grumman Technical Services, University of Texas at El Paso (UTEP), Configurable Space Microsystems Innovations & Applications Center (COSMIAC), NASA Glenn Research Center (GRC), and Youngstown State University, have made significant breakthroughs in the goal of creating the first complete 3D printed small satellite. Since AM machines are relatively inexpensive, this should lead to many entrepreneurial opportunities for the small satellite community. Our technology advancements are focused on the challenges of embedding key components within the structure of the article. We have demonstrated, using advanced fused deposition modeling techniques, complex geometric shapes which optimize the spacecraft design. The UTEP Keck Center has developed a method that interrupts the printing process to insert components into specific cavities, resulting in a spacecraft that has minimal internal space allocated for what traditionally were functional purposes. This allows us to increase experiment and instrument capability by provided added volume in a confined small satellite space. Leveraging initial progress made on a NASA contract, the team investigated the potential of new materials that exploit the AM process, producing candidate compositions that exceed the capabilities of traditional materials. These "new materials" being produced and tested include some that have improved radiation shielding, increased permeability, enhanced thermal properties, better conductive properties, and increased structural performance. The team also investigated materials that were previously not possible to be made. Our testing included standard mechanical tests such as vibration, tensile, thermal cycling, and impact resistance as well as radiation and electromagnetic tests. The initial results of these products and their performance will be presented and compared with standard properties. The new materials with the highest probability to disrupt the future of small satellite systems by driving down costs will be highlighted, in conjunction with the electronic embedding process.

Communications↗
NASA NTRS2015

3D Printing Multi-Functionality: Embedded RF Antennas and Components

Significant research and press has recently focused on the fabrication freedom of Additive Manufacturing (AM) to create both conceptual models and final end-use products. This flexibility allows design modifications to be immediately reflected in 3D printed structures, creating new paradigms within the manufacturing process. 3D printed products will inevitably be fabricated locally, with unit-level customization, optimized to unique mission requirements. However, for the technology to be universally adopted, the processes must be enhanced to incorporate additional technologies; such as electronics, actuation, and electromagnetics. Recently, a novel 3D printing platform, Multi3D manufacturing, was funded by the presidential initiative for revitalizing manufacturing in the USA using 3D printing (America Makes - also known as the National Additive Manufacturing Innovation Institute). The Multi3D system specifically targets 3D printed electronics in arbitrary form; and building upon the potential of this system, this paper describes RF antennas and components fabricated through the integration of material extrusion 3D printing with embedded wire, mesh, and RF elements.

antennas↗
DOE OSTI2025

Redefining Design for Remanufacturing: A Practical Methodology for Prioritizing Remanufacturing Design Rules

Products are often discarded when they fail or no longer meet user needs. These outcomes are frequently shaped by early design decisions. While remanufacturing offers a sustainable alternative by restoring products to like‐new condition, its potential is often limited by designs that do not consider remanufacturing from the outset. This research addresses that challenge by introducing a structured Design for Remanufacturing (DfRem) methodology and a CAD‐integrated tool to support real‐time design decisions. The DfRem framework introduces a new primary design function focused on preserving product functionality across its life cycle. It is supported by a fault tree that identifies failure modes that limit remanufacturing potential and a hierarchy of design principles including Prevent, Minimize, Relocate, Restore, and others. Each principle is linked to actionable design rules that help engineers reduce the need for remanufacturing or improve its efficiency when necessary. To operationalize this framework, we developed CAD plugins for Autodesk Inventor and PTC Creo. These tools use a state machine model to present prioritized design rules based on selected failure modes and user input. By embedding DfRem logic directly into widely used CAD environments, the tool enables engineers to make sustainability‐informed decisions without disrupting existing workflows. Furthermore, this approach highlights the critical role of design in enabling circular and resource‐efficient product development, making remanufacturing a more practical and accessible strategy during the early stages of product design.

CAD↗