Search NASA⌕ Search

SEARCH · Search NASA

Results for “Model generalization”

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 505 records · Page 28
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Development and application of two-step uncertainty propagation and sensitivity analysis methodology for fast reactor safety analysis

Uncertainty quantification (UQ) in nuclear reactors for transients is directly linked with safety assessment through the cross-sections uncertainties, provided as a covariance matrix, which are propagated through the reactor system to output of interest pertaining to reactor safety, such as peak temperatures in fuel/clad/coolant. Using a two-step approach, uncertainties are first quantified and propagated from basic input variables (such as reaction cross-sections) to intermediate quantities (such as reactivity feedback coefficients) through lattice level calculations. Uncertainties of intermediate quantities (from the first step) are then propagated through the system transient calculations, in the second step, to obtain uncertainties on reactor safety output parameters of interest. The scope of this work consists of Uncertainty Quantification & Propagation of nuclear data uncertainties that are highly correlated through unprotected transient overpower and unprotected loss of flow to assess their impact on core safety parameters. This two-step approach in the presence of covariance renders the sensitivity analysis very challenging. In fact, usually the sensitivity analysis is restricted to each step, which limits its application since the sensitivities between the system output quantities and the basic input variables are difficult to obtain. Here, in this work, we address this issue by proposing a simple, general methodology to combine the sensitivity indices obtained in each step by assuming the model behavior being linear. For the first step Generalized Perturbation theory based indices are used while in the second step the recently studied Johnson indices. The uncertainty quantification and sensitivity methodologies discussed here are demonstrated on a generic LFR design which is based on the 500 MWth demonstration Lead-cooled fast reactor (DLFR) using oxide fuel, developed by Westinghouse Electric Company (WEC).

42 - ENGINEERING↗
DOE OSTI2025

Using Large Language Models to help customers monitor global threat data

Large Language Models have proven adept at answering general knowledge questions. To make these generative AI tools useful to our mission customers for monitoring global threats, the data sciences team at Sandia is utilizing retrieval augmented generation (RAG) techniques to customize these models with local data. The local data we use consists of data such as research articles and patent abstracts that we've collected over the last several years using automated pipelines.

Herzer, John Andrew [Sandia National Laboratories ↗
DOE OSTI2025

Large-scale deep learning for metastasis detection in pathology reports

Objectives No existing algorithm can reliably identify metastasis from pathology reports across multiple cancer types and the entire US population. In this study, we develop a deep learning model that automatically detects patients with metastatic cancer by using pathology reports from many laboratories and of multiple cancer types. Materials and Methods We use 60 471 unstructured pathology reports from 4 Surveillance, Epidemiology, and End Results (SEER) registries. The reports were coded into 1 of 3 labels: metastasis negative, metastases positive, or metastasis undetermined. We utilize a task-specific deep neural network trained from scratch and compare its performance with a widely used large language model (LLM). Results Our deep learning architecture trained on task-specific data outperforms a general-purpose LLM, with a recall of 0.894 compared to 0.824. We quantified model uncertainty and used it to defer reports for human review. We found that retaining 72.9% of reports increased recall from 0.894 to 0.969. Discussion A smaller deep learning architecture trained on task-specific data outperforms a general LLM. Equally critical to model performance is the incorporation of uncertainty quantification, achieved here through an abstention mechanism. Conclusions This study’s finding demonstrate the feasibility of developing algorithms to automatically identify metastatic cancer cases from unstructured pathology reports.

machine learning↗
DOE OSTI2025

Eight decades of research on the long-term health effects of radiation in atomic bomb survivors and their offspring

Abstract This year marks the 80th anniversary of the atomic bombings of Hiroshima and Nagasaki. Over the past eight decades, large-scale cohort studies of atomic bomb survivors and their offspring conducted by the Radiation Effects Research Foundation and its predecessor, the Atomic Bomb Casualty Commission, have provided critical insights into the long-term health effects of radiation exposure. Key findings include early identification of radiation-associated leukemia, as well as excess risks of all solid cancers combined, and most individual cancer sites. Observed radiation dose–response relationships have generally followed a linear-quadratic model for leukemia and a linear model for all solid cancers. Recent findings indicating possible upward curvature in the dose–response for all solid cancers may reflect underlying heterogeneity in factors related to individual cancer sites and should be explored further. Generally, younger age at exposure, lower attained age, and female sex appear to show greater radiation sensitivity for all solid cancers combined but results differ by individual cancer site. Recent studies have also identified potential radiation-related excesses for non-cancer diseases such as cataracts, various circulatory diseases, and others. Studies of heritable effects on the offspring of exposed atomic bomb survivors, in contrast, have shown no elevated risk to date in offspring from parental radiation exposure, either at the molecular or disease level. With the cooperation of the atomic bomb survivors and their families, Radiation Effects Research Foundation’s research will continue to play a crucial role in informing the health of survivors, their families, and global radiation protection in the decades to come.

Oncology↗
DOE OSTI2025

..delta..-Learning of High-Fidelity Electronic Structure Using Graph Neural Networks with Modified Node-Level Features

In this work, we present a ..delta..-learning approach for predicting the eigenvalues calculated with the hybrid functional HSE06 (..epsilon..nkHSE) for a set of metal and nitrogen doped graphene catalysts (MNCs) from Perdew-Burke-Ernzerhof (PBE) inputs. The model presented here incorporates electronic scalar features along with structural information in a graph neural network (GNN). In particular, the PBE eigenvalues for different bands and k-points and orbital-resolved projectors are combined with the applied potential as node-level features along with structural information within the Atomistic Line Graph Neural Network (ALIGNN) architecture. These features enable flexibility for systems with electrified interfaces, such as in electrocatalysts and achieves mean absolute error (MAE) of less than 0.1 eV. The machine learning model reported here achieves a strong generalization to left-out adsorbates (MAE = 0.074 eV) and leave-one-chemical-space-out (MAE = 0.08 eV) and completely left-out metals (MAE = 0.072 eV), confirming the robustness of the machine learning (ML) model in predicting ..epsilon..nkHSE.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Deep potential molecular dynamics simulations of low-temperature plasma-surface interactions

Machine learning approaches to potential generation for molecular dynamics (MD) simulations of low-temperature plasma-surface interactions could greatly extend the range of chemical systems that can be modeled. Empirical potentials are difficult to generalize to complex combinations of multiple elements with interactions that might include covalent, ionic, and metallic bonds. This work demonstrates that a specific machine learning approach, Deep Potential Molecular Dynamics (DeepMD), can generate potentials that provide a good model of plasma etching in the Si-Cl-Ar system. Comparisons are made between MD results using DeepMD models and empirical potentials, as well as experimental measurements. Pure Si properties predicted by the DeepMD model are in reasonable agreement with experimental results. Simulations of Si bombardment by Ar + ions demonstrate the ability of the DeepMD method to predict sputtering yields as well as the depth of the amorphous-crystalline interface. Etch yields as a function of flux ratio and ion energy for simultaneous Cl 2 and Ar + impacts are in good agreement with previous simulation results and experiment. Predictions of etch yields and etch products during plasma-assisted atomic layer etching of Si-Cl 2 -Ar are shown to be in good agreement with MD predictions using empirical potentials and with experiment. Finally, good agreement was also seen with measurements for the spontaneous etching of Si by Cl atoms at 300 K. Further, the demonstration that DeepMD can reproduce results from MD simulations using empirical potentials is a necessary condition to future efforts to extend the method to a much wider range of systems for which empirical potentials may be difficult or impossible to obtain.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

Kilonova Emissions from Neutron Star Merger Remnants: Implications for the Nuclear Equation of State

Multimessenger observations of binary neutron star mergers can provide valuable information on the nuclear equation of state (EOS). Here, we investigate the extent to which electromagnetic observations of the associated kilonovae allow us to place constraints on the EOS. For this, we use state-of-the-art three-dimensional general-relativistic magnetohydrodynamics simulations and detailed nucleosynthesis modeling to connect properties of observed light curves to properties of the accretion disk, and hence, the EOS. Using our general approach, we use multimessenger observations of GW170817/AT2017gfo to study the impact of various sources of uncertainty on inferences of the EOS. We constrain the radius of a 1.4M ⊙ neutron star to lie within 10.30 ≤ R 1.4 ≤ 13.0 km and the maximum mass to be M TOV ≤ 3.06M ⊙ .

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2024

Fuel Performance Calculations of Tagged UO 2 MiniFuel Disks

Isotopic taggants are being considered as a novel form of forensic identifiers for nuclear fuels. As part of this effort, eighteen tagged MiniFuel disks were irradiated in the High Flux Isotope Reactor at Oak Ridge National Laboratory. In this work, the BISON fuel performance code was used to recreate the pellets and predict their behavior and their post-irradiation material properties. These predictions are to serve as a baseline to help investigators identify results of interest during post-irradiation examination of actual MiniFuel disks. The simulation methodology and models are described herein, along with the individual disk predictions. Several general trends were identified in this analysis. Based on the models, it is predicted that the pellets’ internal temperatures will rise over time, fission gas release will remain constant at around 1%, regardless of the burnup, and no grain growth will occur in any of the disks. The reasons for these predictions are discussed below. Probable shortcomings in the BISON predictions are addressed, and future work is proposed.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2026

Tori, Klein bottles, and modulo 8 parity/time-reversal anomalies of 2+1d staggered fermions

We study the symmetries of lattice staggered fermions in 2+1d. Using the symmetries, we can place the system on any sheared torus or Klein bottle. These different backgrounds provide diagnostics of various ’t Hooft anomalies associated with the crystalline symmetries. We then compare the lattice model to its continuum limit. The symmetries of the lattice system are mapped in a nontrivial way to the symmetries of the continuum theories. Using this map, we match the ’t Hooft anomalies on the lattice and the continuum. Along the way, we develop a general formalism to study Hamiltonian lattice models on nontrivial, compact, flat spaces.

Algorithms and Theoretical Developments↗
DOE OSTI2024

Bridging Hydrological Ensemble Simulation and Learning Using Deep Neural Operators

Ensemble-based simulation and learning (ESnL) has long been used in hydrology for parameter inference, but computational demands of process-based ESnL can be quite high. To address this issue, we propose a deep neural operator learning approach. Neural operators are generic machine learning algorithms that can learn functional mappings between infinite-dimensional spaces, providing a highly flexible tool for scientific machine learning. Our approach is built upon DeepONet, a specific deep neural operator, and is designed to address several common problems in hydrology, namely, model parameter estimation, prediction at ungaged locations, and uncertainty quantification. Here we demonstrate the effectiveness of our DeepONet-based workflow using an existing large model ensemble created for an eastern U.S. watershed that is instrumented with 10 streamflow gages. Results suggest DeepONet achieves high efficiency in learning an ML surrogate model from the model ensemble, with the modified Kling-Gupta Efficiency exceeding 0.9 on holdout test sets. Parameter inference, carried out using the trained DeepONet surrogate model and genetic algorithm, also yields robust results. Additionally, we formulate and train a separate DeepONet model for physics-informed, seq-to-seq streamflow forecasting, which further reduces biases in the pre-trained DeepONet surrogate model. While this study focuses primarily on a single watershed, our approach is general and may be extended to enable learning from model ensembles across multiple basins or models. Thus, this research represents a significant contribution to the application of hybrid machine learning in hydrology.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Understanding plasma turbulence through exact coherent structures

Plasma turbulence is a key challenge in understanding transport phenomena in magnetically confined plasmas. This work presents a generalized framework to analyze plasma turbulence that utilizes periodic orbit theory. In periodic orbit theory, doubly periodic solutions (coherent structures) of the governing equation(s) serve as building blocks of the considered turbulent dynamics. To illustrate the concept and method, the particularly simple Kuramoto–Sivashinsky (referred to here as LMRT for the original authors: LaQuey, Mahajan, Rutherford, and Tang) trapped-ion mode toy model is used. By applying numerical optimization techniques to the LMRT equation, we extract coherent spacetime patterns that represent the library of allowable fundamental structures of the equation. These structures provide a framework to systematically describe turbulence as a composition of recurrent solutions, revealing an underlying order within chaotic plasma motion. Although illustrated here using the simplified LMRT model for clarity, this framework provides a general strategy that can be extended to more complex and realistic models of plasma turbulence, including gyrokinetic systems. This offers a new method for predicting and potentially controlling transport processes in fusion plasmas by providing a bridge between nonlinear dynamical systems theory and plasma physics in the form of a generalized framework with which to analyze and understand spatially extended nonlinear partial differential equations.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

Quantum Gravity and Laser Interferometry: Towards Observable Predictions

Understanding quantum gravity remains one of the deepest challenges in modern physics, as direct experimental access to Planck-scale effects is beyond current technological reach. However, recent theoretical advances indicate that quantum fluctuations of spacetime may produce measurable effects in precision experiments, particularly near causal horizons. This opens new avenues for testing quantum gravity phenomena through high-precision measurement techniques. This dissertation develops multiple theoretical models to characterize these effects and examines their potential observational signatures in future gravitational wave interferometers. We begin by investigating the role of quantum fluctuations in near-horizon geometries through the lens of the AdS/CFT correspondence, which provides a powerful framework for understanding the interplay between quantum field theory and general relativity via holographic principles. By modeling stochastic energy-momentum sources in Rindler-AdS spacetime, we demonstrate that vacuum fluctuations transform the Einstein equations into a Langevin-type stochastic differential equation, leading to potentially observable fluctuations in photon traversal times. Extending this approach to Minkowski spacetime, we establish a correspondence between gravitational shockwaves and fluid dynamics, showing that near-horizon perturbations satisfy an equation analogous to that governing incompressible fluids, thereby reinforcing the membrane paradigm and hydrodynamic analogies in the context of the fluid/gravity duality. Furthermore, we construct the covariant phase space of a spherically symmetric causal diamond in Minkowski spacetime, identifying two fundamental charges that govern its evolution. These results provide a foundation for quantizing causal horizons and understanding their microscopic degrees of freedom. Building upon these theoretical developments, we further examine a related stochastic phenomenon: the gravitational wave memory background arising from the cumulative memory steps produced by supermassive black hole mergers. After reviewing the standard stochastic gravitational wave background, gravitational memory effects, and BMS symmetries, we model the stochastic memory background using a Brownian motion framework. We show that while the cumulative memory background initially appears above the sensitivity curve of space-based interferometers like LISA, the realistic subtraction of individually resolvable merger events substantially suppresses the residual signal, making its detection more challenging. This highlights the critical importance of source subtraction when evaluating the detectability of gravitational memory effects. By bridging fundamental theory with experimental prospects, this dissertation contributes to the ongoing effort to uncover the quantum nature of spacetime through precision measurement techniques. Whether through detecting quantum spacetime fluctuations, gravitational memory backgrounds, or probing the symmetries of causal horizons, the pursuit of observable quantum gravity phenomena continues to expand the frontiers of both theory and experiment.

Zhang, Yiwen [Caltech] (ORCID:0000000323559416)↗
DOE OSTI2024

Dynamic Modeling and Simulation of a Subcritical Coal-Fired Power Plant under Load-Following Conditions

Dynamic models for power plants that capture realistic general process trends and effects of manipulated variables are needed to improve load-following, while minimizing carbon footprint. In this work, a dynamic modeling approach and simulation results for subcritical coal-fired power plant components are presented. These encompass simulation of the dynamics in the fireside, including the effects of fuel, air combustion, and the dynamics of the entire waterside and power generation sections. This model development enables the simulation and analysis of the important short and long-time scale dynamics of components such as heaters, evaporative loop, and power generation units. Furthermore, additional variables in the power generation section are introduced to improve model accuracy, extending the prediction capability of subcritical power plant models and opening new opportunities for research in operator training, optimization, and advanced model-based controller design that are based on these models. The change in process gain for different ramp rates associated with disturbance signals that affect process variables is also explored and a correlation developed. This provides opportunities to study disturbance rejection control implementation and adaptation for scenarios with such variations in ramp rates. The prediction capabilities of selected components are compared to data available in literature, with the obtained root mean squared error ranges that reflect the model performance and quality of predictions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Origin of Large Effective Phonon Magnetic Moments in Monolayer MoS 2

Recent helicity-resolved magneto-Raman spectroscopy measurement demonstrates large effective phonon magnetic moments of ~2.5 μ B in monolayer MoS 2 , highlighting resonant excitation of bright excitons as a feasible route to activate Γ-point circularly polarized phonons in transition metal dichalcogenides. However, a microscopic picture of this intriguing phenomenon remains lacking. In this work, we show that an orbital transition between the split conduction bands (Δ 0 = 4 meV) of MoS 2 couples to the doubly degenerate E" phonon mode (Ω 0 = 33 meV), forming two hybridized states. Our phononic and electronic Raman scattering measurements capture these two states: (i) one with predominantly phonon contribution in the helicity-switched channels and (ii) one with primarily orbital contribution in the helicity-conserved channels. An orbital-phonon coupling model successfully reproduces the large effective magnetic moments of the circularly polarized phonons and explains their thermodynamic properties. Strikingly, the Raman mode from the orbital transition is superimposed on a strong quasi-elastic scattering background, indicating the presence of spin fluctuations. As a result, the electrons excited to the conduction bands through the exciton exhibit paramagnetic behavior although MoS 2 is generally considered as a nonmagnetic material. By depositing nanometer-thickness nickel thin films on monolayer MoS 2 , we tune the electronic structure so that the A exciton perfectly overlaps with the 633 nm laser. The optimization of resonance excitation leads to pronounced tunability of the orbital-phonon hybridized states. Our results generalize the orbital-phonon coupling model of effective phonon magnetic moments to material systems beyond the paramagnets and magnets.

2D transition metal dichalcogenide↗
DOE OSTI2026

Beyond interpolation: Physics-inspired gating transformers for extrapolating irradiation conditions to novel nuclear fuels

The qualification of advanced nuclear fuels relies on irradiation experiments in test reactors that emulate commercial conditions. Designing these tests requires accurate prediction of key irradiation quantities, particularly heat generation rate and burnup, yet obtaining them typically involves computationally expensive multi-step simulation workflows. We propose a physics-inspired gating transformer (PIGT) that integrates an inverse-square, distance-based attenuation into the encoder representation to bias attention toward physically relevant spatial relationships while retaining data-driven flexibility. Using MiniFuel irradiation data from the High Flux Isotope Reactor at Oak Ridge National Laboratory, we benchmark against ensemble methods, feedforward and recurrent networks, convolutional models, and standard transformers. While baseline models perform well under interpolation, they exhibit a pronounced generalization gap when evaluated on fuels not included in the training set. The proposed model consistently improves extrapolative accuracy and stability, yielding the strongest performance on unseen fuel configurations. These results indicate that a lightweight physics structure embedded within attention mechanisms can substantially improve robustness, enabling more reliable surrogate predictions to accelerate the design of nuclear fuel irradiation experiments.

Fuel qualification↗
DOE OSTI2025

E3SM‐Arctic: Regionally Refined Coupled Model for Advanced Understanding of Arctic Systems Interactions

Earth system models are essential tools for climate projections, but coarse resolutions limit regional accuracy, especially in the Arctic. Regionally refined meshes (RRMs) enhance resolution in key areas while maintaining computational efficiency. This paper provides an overview of the United States (U.S.) Department of Energy's (DOE's) Energy Exascale Earth System Model version 2.1 with an Arctic RRM, hereafter referred to as E3SMv2.1-Arctic, for the atmosphere (25 km), land (25 km), and ocean/ice (10 km) components. We evaluate the atmospheric component and its interactions with land, ocean, and cryosphere by comparing the RRM (E3SM2.1-Arctic) historical simulations (1950–2014) with the uniform low-resolution (LR) counterpart, reanalysis products, and observational data sets. The RRM generally reduces biases in the LR model, improving simulations of Arctic large-scale mean fields, such as precipitation, atmospheric circulation, clouds, atmospheric river frequency, and sea ice thickness. However, it introduces a seasonally dependent surface air temperature bias, reducing the LR cold bias in summer but enhancing the LR warm bias in winter, which contributes to the underestimated winter sea ice area and volume. Radiative feedback analysis shows similar climate feedback strengths in both model configurations, with the RRM exhibiting a more positive surface albedo feedback and contributing to a stronger surface warming than LR. These findings underscore the importance of high-resolution modeling for advancing our understanding of Arctic climate changes and their broader global impacts, although some persistent biases appear to be independent of model resolution at 10–100 km scales.

Energy Exascale Earth System Model (E3SM)↗
DOE OSTI2024

Spectral gaps of two- and three-dimensional many-body quantum systems in the thermodynamic limit

We present an expression for the spectral gap, opening up new possibilities for performing and accelerating spectral calculations of quantum many-body systems. We develop and demonstrate one such possibility in the context of tensor network simulations. Our approach requires only minor modifications of the widely used simple update method and is computationally lightweight relative to other approaches. We validate it by computing spectral gaps of the 2D and 3D transverse-field Ising models and find strong agreement with previously reported perturbation theory results. Published by the American Physical Society 2024

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

Use of physics to improve solar forecast: Part III, impacts of different cloud types

Cloud-type impacts present a great challenge to solar forecasting due to diverse and complex cloud-radiation interactions. This third part of our paper sequence seeks to address this challenge by quantifying the forecast accuracies under eight cloud types: cumulus (Cu), stratified clouds (St), altocumulus (Ac), altostratus (As), cirrostratus/anvil (Cr), cirrus (Ci), congestus (Co), deep convective clouds (Dc) across four physics-informed persistence models reported in Part I. To generalize the cloud impacts, the eight cloud types are further grouped into three cloud categories based on their common features: weak convective clouds, stratiform clouds, and strong convective clouds. Here, the decade-long (2001 ~ 2014) collocated measurements of irradiances and cloud types at the U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) Program South Great Plain (SGP) Central Facility site are used for model evaluation. Results reveal a clear performance hierarchy for global horizontal irradiance (GHI) and direct normal irradiance (DNI): best for weak convective clouds and cirrus, intermediate for stratiform clouds, and worst for strong convective clouds. Performance for diffuse horizontal irradiance (DHI) is less influenced by cloud types. Cloud albedo dominates all three irradiances for Dc, while both cloud albedo and cloud fraction are influential for other cloud types. A 12 %~33 % improvement in accuracy at 6-hour lead time compared to the benchmark smart model confirms the effectiveness of incorporating physics into the models for various cloud types; further improvements are expected by directly integrating cloud type information into forecasting models by modifying the physical formulation of cloud-radiation interaction, and/or using more advanced machine learning models.

14 SOLAR ENERGY↗