SEARCH · Engineering Papers

Results for “array modeling”

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.

SourceResearch connectionFrom yearThrough year

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

At least 55 records · Page 3

All sources NASA NTRS DOE OSTI

DOE OSTI2024

Hyperparameter Studies for Vision Transformers Trained on High-Fidelity Simulations

This library is a collection of python modules that define, train, and analyze vision-transformer (ViT) machine learning models. The code implements, with mild modifications, ViT models that have been made publicly available through publication and GitHub code. The training data for these models is hydrodynamic simulation output in the form of numpy arrays. This library contains code to train these ViT models on the hydrodynamic simulation output with a variety of hyperparameters, and to compare the results of such models. Furthermore, the library contains definitions of simple convolutional neural network (CNN) machine learning architectures which can be trained on the same hydrodynamic simulation output. These are included as a reference point to compare the ViT models to. Additionally, the library includes trained ViT and CNN models and example input data for demonstration purposes. The code is based on the PyTorch python library.

Callis, Skylar↗

DOE OSTI2025

Vision Foundation Models in Remote Sensing: A survey

Artificial intelligence (AI) technologies have profoundly transformed the field of remote sensing (RS), revolutionizing data collection, processing, and analysis. Traditionally reliant on manual interpretation and task-specific models, RS research has been significantly enhanced by the advent of foundation models (FMs)—large-scale pretrained AI models capable of performing a wide array of tasks with unprecedented accuracy and efficiency. This article provides a comprehensive survey of FMs in the RS domain. We categorize these models based on their architectures, pretraining datasets, and methodologies. Through detailed performance comparisons, we highlight emerging trends and the significant advancements achieved by those FMs. Additionally, we discuss technical challenges, practical implications, and future research directions, addressing the need for high-quality data, computational resources, and improved model generalization. Our research also finds that pretraining methods, particularly self-supervised learning (SSL) techniques like contrastive learning (CL) and masked autoencoders (MAEs), remarkably enhance the performance and robustness of FMs. This survey aims to serve as a resource for researchers and practitioners by providing a panorama of advances and promising pathways for the continued development and application of FMs in RS.

data models↗

DOE OSTI2024

Bridging the gap between experiments and simulations using machine learning

The physics of inertial confinement fusion is rich and complex. Simulation codes that are used to design experiments are computationally expensive and lack the predictive capability required for extensive parameter exploration in search of a high-performing design for laser direct drive. In this work we use deep learning to build a fast emulator of experiments. To facilitate the development of the deep-learning model, an autoencoder is used to reduce the dimensionality of the input space. Two deep learning models are developed. One model is trained on a vast array of simulation data and is subsequently calibrated to expensive and limited experimental data using a technique known as “transfer learning.” The other model is trained on a statistical model and is subsequently calibrated using experimental data. A comparative study of the two predictive models is carried out. The models potentially reproduce key experimental observables with high accuracy and unprecedented inference times relative to those achieved with simulation codes. These models facilitate rapid exploration of a high dimensional input parameter space.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2024

Machine learning materials properties with accurate predictions, uncertainty estimates, domain guidance, and persistent online accessibility

One compelling vision of the future of materials discovery and design involves the use of machine learning (ML) models to predict materials properties and then rapidly find materials tailored for specific applications. However, realizing this vision requires both providing detailed uncertainty quantification (model prediction errors and domain of applicability) and making models readily usable. At present, it is common practice in the community to assess ML model performance only in terms of prediction accuracy (e.g. mean absolute error), while neglecting detailed uncertainty quantification and robust model accessibility and usability. Here, we demonstrate a practical method for realizing both uncertainty and accessibility features with a large set of models. We develop random forest ML models for 33 materials properties spanning an array of data sources (computational and experimental) and property types (electrical, mechanical, thermodynamic, etc). All models have calibrated ensemble error bars to quantify prediction uncertainty and domain of applicability guidance enabled by kernel-density-estimate-based feature distance measures. All data and models are publicly hosted on the Garden-AI infrastructure, which provides an easy-to-use, persistent interface for model dissemination that permits models to be invoked with only a few lines of Python code. We demonstrate the power of this approach by using our models to conduct a fully ML-based materials discovery exercise to search for new stable, highly active perovskite oxide catalyst materials.

domain of applicability↗

DOE OSTI2025

WFIP3

The Wind Forecasting Improvement Project 3 (WFIP-3) is the first offshore-based wind resource characterization project within the WFIP construct, funded by the U.S. Department of Energy. WFIP-3 will provide a unique field study that will deliver the comprehensive suite of data needed to inform a series of modeling efforts that will develop and evaluate parameterization schemes suited to offshore environments and improved industry-targeted applications. The field study has two goals: (1) detailed sampling of the vertical structure of the Marine Atmospheric Boundary Layer (MABL) at key observational areas, creating a rich dataset that will be used to refine and validate parameterization schemes, and (2) wide-area sampling of the MABL to create a multi-scale array of observations informing and guiding models of resource characterization. We will deploy a multi-platform array of measurements that span the MABL and create a multi-scale observational array stretching south from Marth’s Vineyard across the wind energy areas.

17 WIND ENERGY↗

DOE OSTI2025

Characterization of Infrasonic Signatures of Earth-Grazing Fireballs as Analogues to Hypersonic Vehicles (Final Report)

Accurate detection, discrimination, and characterization of high-altitude hypersonic events using infrasonic monitoring are critical to planetary defense and global strategic surveillance. This report synthesizes recent advances achieved through rigorous analysis of infrasonic signatures from natural meteoroids, emphasizing shallow entry-angle meteoroids as essentially proxies for artificial hypersonic systems. Meteoroids naturally encompass diverse velocities, trajectories, altitudes, and fragmentation behaviors, enabling systematic validation of empirical period–yield relationships, waveform morphology classifiers, and trajectory-induced back-azimuth deviation models. Integration of adaptive array-processing enhancements within Cardinal software further extends infrasonic detection sensitivity and signal classification reliability. Collectively these advances, based solely on infrasonic signatures or limited optical data, offer robust methodologies for distinguishing natural from artificial hypersonic sources, significantly reducing event geolocation uncertainties and refining source-function determination. The outcomes detailed herein lay foundational groundwork for improved global hypersonic event-surveillance frameworks, supporting improved security preparedness and informing strategic monitoring and defense policies.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2025

Models and Measurements Quantify Photon Recycling, Charge-Carrier Diffusion and Photon Scattering Contributions to Photoluminescence in InP Nanowire Arrays

Nanowire arrays present many unique advantages for solar-to-chemical energy conversion. One possible advantage is that photon recycling between neighboring nanowires has the potential to increase solar energy conversion efficiencies. Here, in this work, we explore three underlying mechanisms of optical and electronic coupling between neighboring nanowires─incident photon scattering, photon recycling, and charge-carrier transport from the photoexcited nanowire to the neighboring nanowire via the underlying substrate─using single nanowire-level microscopy and spectroscopy measurements. We present a comprehensive analysis of light absorption and emission of a single nanowire at open circuit, and subsequent re-absorption and re-emission by a neighboring nanowire. We developed a novel correlated single nanowire microspectroscopy and widefield imaging methodology to spatially resolve photon communication pathways between neighboring nanowires and selectively image re-emitted and reflected photons. We developed unique multiphysics models to couple wave optics and semiconductor photophysics to especially isolate contributions from photon recycling and electronic transport to photon emission from neighboring nanowires. By systematically varying the morphologies of the nanowires modeled, we identified pathways to maximize photon recycling between neighboring nanowires. We concluded that the measured photoluminescence is more strongly influenced by the diffusion of charge carriers as compared to photon recycling in materials with moderate-to-large charge-carrier mobilities (>10 cm 2 V –1 s –1 ), and that photon recycling dictates photoluminescence intensity only when the charge-carrier mobility is low (<1 cm 2 V –1 s –1 ). The experimental and simulation platforms developed herein for photon management strategies can be leveraged by the semiconductor photocatalysis community to enhance solar-to-chemical conversion efficiencies in semiconductor nanowire arrays.

25 ENERGY STORAGE↗

DOE OSTI2026

Intelligent Experiments Through Real-time AI: Fast Data Processing and Autonomous Detector Control for sPHENIX and Future EIC Detectors (Final Report)

The overall vision of this project was to integrate real-time artificial intelligence (AI) directly into the data acquisition and detector-control systems of nuclear physics experiments, including both fast online event selection and an autonomous detector-control feedback loop. The work carried out under the award focused on the fast online event-selection half of that vision: the efficient recording of low-momentum heavy-flavor (HF) hadron decays in proton-proton collisions at the sPHENIX experiment at the Relativistic Heavy Ion Collider (RHIC)—an observable that requires fast tracking and topological trigger selection not previously demonstrated at RHIC, and that is essential for QCD studies at future facilities such as the Electron-Ion Collider (EIC). The autonomous detector-control (GPU-based feedback) component named in the project title remained a design concept and was not implemented under this award. The Massachusetts Institute of Technology (MIT) group led the offline simulation and data processing needed to train the machine-learning (ML) models, the translation of trained models to Field-Programmable Gate Array (FPGA) firmware using the hls4ml framework, and the physics validation of heavy-flavor reconstruction. Over the award period, the team developed and hardware-tested the principal components of an AI-based heavy-flavor trigger on simulated and recorded sPHENIX tracker data: a software Bipartite Graph Attention Network (BiGAT) trigger model reaching > 95% signal efficiency at 99% background rejection; an FPGA-native hit clusterizer matching the offline clustering; smaller networks synthesized to FPGA within the required sub-10 µs latency; and an assembled decoder–clusterizer–inference firmware chain exercised on the FELIX readout board. A complete, fully integrated hardware demonstrator was not finished within the award period. This report documents the project goals, the MIT group’s contributions, the technical accomplishments, and the outlook toward applications at the future EIC ePIC detector.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2025

Architecture-Aware Models of AI Engines for High-Performance Matrix Matrix Multiplication

The AI Engine (AIE) architecture, available in systems from mobile SoCs to server-class FPGAs, aims to efficiently execute AI/ML tasks through a two-dimensional array of compute tiles. Previous work on AIEs has explored different approaches to mapping computation across spatial arrays, but the compute kernel running on each tile has not been the focus. Additionally, the AIE-ML architecture introduces memory tiles and omits programmable logic, requiring new approaches to staging and moving data throughout the array. In this work we update analytical models developed for CPUs to produce the design of high performance kernels while introducing new model considerations such as memory structure, throughput, and latency as required by the AIE hardware. We evaluate our models by developing AIE-ML kernels for matrix multiplication in low-precision data types showing performance up to 95% of compute peak for the kernel when data resides in local memory and above 90% of compute peak when data resides in main memory.

Binder, Elliott D. [Carnegie Mellon University, Pi↗

DOE OSTI2025

A neural master equation framework for multiscale modeling of molecular processes: application to atomic-scale plasma processes

Plasma-surface interactions (PSI) play a crucial role in microelectronics fabrication; however, their multiscale nature and array of complex, often unknown interactions make computational modeling of PSIs extremely difficult. To this end, we propose a general neural master equation (NME) framework that uses master equations to describe the dynamics of a molecular process, wherein neural networks learned from atomistic simulations represent unknown transitions between different system states. By leveraging the physics-based structure of master equations and data-driven state transitions, the NME framework promotes generalizability and physics interpretability, and can bridge disparate length and time scales. The framework is demonstrated for multiscale modeling of Si atomic layer etching and reactive ion etching, where the learned NME-based surface kinetic models exhibit good predictive and extrapolative capabilities for predicting experimentally relevant observables as a function of process parameters. The NME-based surface kinetic models obey physical constraints, which are violated in models based on neural ordinary differential equations. The proposed NME framework for multiscale modeling of molecular processes can pave the way for the discovery of new chemistries and materials in atomic-scale plasma processes.

Chemical engineering↗

DOE OSTI2025

Design and Analysis of a Mutual Inductance Level Sensor for Liquid Metals

Here, this article describes the design and analysis of an electromagnetic level sensor for use in high-temperature liquid metal systems. The mutual inductance level sensor (MILS) described in this work was fabricated using two single-conductor mineral insulated cables wrapped in a bifilar fashion around a stainless steel tube core and was housed in an isolating thimble that preserved the pressure boundary of the test vessel. Two sensor variations were fabricated that differ only in active length, 1016 and 1778 mm. Experimental data were collected using the 1016-mm sensor (MILS-MKII-040) that demonstrated a sensitivity of 9.2 μ V/mm in a room temperature testing stand that used solid aluminum as a surrogate for liquid metal. Experimental data were collected using the 1778-mm sensor (MILS-MKII-070) that demonstrated a sensitivity of 6.9 μ V/mm in the high-temperature (300 ° C) liquid sodium environment at the mechanisms engineering test loop (METL) of Argonne National Laboratory. The sensor performance was found to be repeatable over the course of several months, with roughly ±1% deviation from nominal output. Finite element models were developed in COMSOL Multiphysics that fully describe each test setup, and the models were validated using experimental data. The validated COMSOL models were used to perform an array of analyses that examined the performance of the sensor in differing environments. Maximizing the coil diameter inside the isolating thimble was found to maximize the signal and sensitivity of the sensor. An optimal operating frequency was found near 1000 Hz using both experimental data and COMSOL. The influence of a metallic thimble surrounding the sensor and a metallic sensor core was quantified and found to be negligible at the optimal operating frequency. The sensitivity of the sensor was quantified when monitoring the level of additional liquid metals. These include lead, lead-bismuth eutectic (LBE), sodium-potassium alloy (NaK), and lithium (in addition to sodium). The sensitivities were quantified using liquid metal properties at 350 ° C and 650 ° C. The geometry of the test stand model, all material properties used in the model, and the results are presented in a manner that allows the reader can replicate the model and perform additional analyses.

COMSOL↗

DOE OSTI2025

Probabilistic Error Bounds for Low-Rank Tensor Decompositions Used in Large-Scale Data Analysis Applications (LDRD Final Report)

This report documents a research project on analyzing low-rank tensor models for data analysis that took place at Sandia National Laboratories from October 2023–September 2025. The focus of this work was to extend theoretical frameworks from statistics and probability theory for use with models for scalar, vector, and matrix data to models with tensor, or general multi-dimensional array, data. Through this work, we have provided a new set of tools for bounding errors on low-rank tensor models of both complete and sampled data. The remainder of this report is organized as follows. In Section 1, we describe the proposed work at the start of the project. Section 2 describes the research advances made as part of the project. Other research contributions in the form of conference presentations and software development is provided in Section 3. Workforce development at Sandia and Florida Atlantic University (via a subcontract on this project) is provided in Section 4.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Assessment of the CTF subchannel code for modeling a large-break loss-of-coolant accident reflood transient

With increased industry interest in extending reactor operating cycles, the Nuclear Energy Advanced Modeling and Simulation (NEAMS) program has been investigating the behavior of high-burnup fuel during design basis accidents such as the large-break loss-of-coolant accident (LBLOCA) with consideration for risk of fuel fragmentation, relocation, and dispersal (FFRD). As part of that activity, the NEAMS subchannel thermal/ hydraulics (T/H) code, CTF, is being used for modeling of LBLOCA and to determine the impact of subchannel resolution on results. Although CTF includes a wide range of models for LBLOCA conditions, the code has not been used for this application while maintained at Oak Ridge National Laboratory (ORNL) until now. Therefore, here, in this work, a preliminary assessment of several of these models was performed using openly available reflood experimental data from the Flooding Experiments in Blocked Arrays (FEBA) tests. One coarse mesh and one fine mesh model were set up in CTF for high and low flooding rate tests performed in the unblocked FEBA facility. A coarse TRACE model was set up to be as consistent as possible with the coarse CTF model to allow for code-to-code benchmarking. The assessment shows a tendency of the codes to over-predict peak cladding temperature (PCT) near the top of the bundle and to quench early. Advanced spacer grid models were shown to improve upper bundle predictions in CTF. The resolved CTF model over-predicted PCT by a larger degree in the center channels in the low-flooding rate test, and it is believed that the radiative heat transfer model, which was not used in this study, may be needed to correct this over-prediction. Finally, this work demonstrates the importance of the droplet model in determining quench time and vapor temperature and PCT prediction, which necessitates a more in-depth validation of these models in the future.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

DOE OSTI2024

Toward Spatial Control of Reaction Selectivity on Photocatalysts Using Area-Selective Atomic Layer Deposition on the Model Dual Site Electrocatalyst Platform

Photocatalytic water splitting is a promising route to low-cost, green H 2 . However, this approach is currently limited in its solar-to-hydrogen conversion efficiency. One major source of efficiency loss is attributed to the high rates of undesired side and back reactions, which are exacerbated by the proximity of neighboring oxidation and reduction sites. Nanoscopic oxide coatings have previously been used to selectively block undesired reactants from reaching active sites; however, a coating encapsulating the entire photocatalyst particle limits activity as it cannot facilitate both half-reactions. In this work, area selective atomic layer deposition (AS-ALD) was used to selectively deposit semipermeable TiO 2 films onto model metallic cocatalysts for enhancing reaction selectivity while maintaining a high overall activity. Pt and Au were used as exemplary reduction and oxidation cocatalyst sites, respectively, where Au was deactivated toward ALD growth through self-assembled thiol monolayers while TiO 2 was coated onto Pt sites. Electroanalytical measurements of monometallic thin film electrodes showed that the TiO 2 -encapsulated Pt effectively suppressed undesired H 2 oxidation and Fe(II)/Fe(III) redox reactions while still permitting the desired hydrogen evolution reaction (HER). A planar model photocatalyst platform containing patterned interdigitated arrays of Au and Pt microelectrodes was further assessed using scanning electrochemical microscopy (SECM), demonstrating the successful use of AS-ALD to enable local reaction selectivity in a dual-reaction-site (photo)electrocatalytic system. Finally, interdigitated microelectrodes having independent potential control were used to show that selectively deposited TiO 2 coatings can suppress the rate of back reactions on neighboring active sites by an order of magnitude compared with uncoated control samples.

08 HYDROGEN↗

DOE OSTI2025

Observations and seismoacoustic simulations of earthquake-generated infrasound waves in non-epicentral regions

We analysed infrasound waves associated with the Gyeongju earthquake (ML 5.8) that occurred on 2016 September 12, in the southeastern Korean Peninsula. For infrasound wave detection, the Progressive Multichannel Correlation method was applied to the infrasound data set recorded at seven arrays operating in South Korea at epicentral distances ranging from 178 to 472 km. Based on the back-projection method constrained by array-dependent celerity and azimuth deviation models, the source regions were identified in both the epicentral and non-epicentral regions. Remarkably, the non-epicentral secondary sources of this earthquake were located in regions with shallow water depths: (i) the western coastal area in the Yellow Sea and (ii) the shallow ocean basin and bank in the East Sea. The location results obtained from the earthquake could be corroborated through its foreshock (ML 5.1), yielding location results consistent with those of the main shock. The generation of infrasound waves over shallow water depths was fortuitously validated by direct recordings of dominant single-frequency (~0.3 Hz) infrasound waves at close range via temporary sensors near the ocean basin and bank. We interpreted that low-frequency infrasound signals could be generated from interactions among the ocean floor, shallow seawater and atmosphere. We performed numerical simulations of seismoacoustic fields to predict ground motions on the seafloor and acoustic transmission efficiency between the water and air interface. The simulations quantified the energy transfer through different media and clarified our observational results. We found that because this solid Earth‒water‒atmosphere coupled air wave has a relatively low frequency (~0.3 Hz), it can survive propagation over long distances compared with high-frequency infrasound waves generated in inland and mountain regions. In this study, we extend our understanding of water‒atmosphere coupling and the monitoring framework for earthquake-associated non-epicentral infrasound waves, encompassing not only inland ground shaking but also shallow sea regions located far from the epicentre.

Earthquake monitoring and test-ban treaty verifica↗

DOE OSTI2024

cldera-tools

SAND2025-03848O CLDERA-Tools is a small library for performing online calculation of quantity of interest derived from variables in the host application. The library stores pointers to the arrays of variables of the host model and uses them at every timestep to compute desired quantity of interests (prescribed via yaml input files). Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

Watkins II, Jerry↗

DOE OSTI2026

Robustness of Kardar-Parisi-Zhang-like transport in long-range interacting quantum spin chains

Isotropic integrable spin chains such as the Heisenberg model feature superdiffusive spin transport belonging to an as-yet-unidentified dynamical universality class closely related to that of Kardar, Parisi, and Zhang (KPZ). To determine whether these results extend to more generic one-dimensional models, particularly those realizable in quantum simulators, we investigate spin and energy transport in non-integrable, long-range Heisenberg models using state-of-the-art tensor network methods. Despite the lack of integrability and the asymptotic expectation of diffusion, for power-law models (with exponent 2<α<∞) we observe long-lived z=3/2 superdiffusive spin transport and two-point correlators consistent with KPZ scaling functions, up to times t∼103/J. We conjecture that this KPZ-like transport is due to the proximity of such power-law-interacting models to the integrable family of Inozemtsev models, which we show to also exhibit KPZ-like spin transport across all interaction ranges. Finally, we consider anisotropic spin models naturally realized in Rydberg atom arrays and ultracold polar molecules, demonstrating that a wide range of long-lived, non-diffusive transport can be observed in experimental settings.

Anand↗

DOE OSTI2025

SQuaD: Smart Quantum Detection for Photon Recognition and Dark Count Elimination

Quantum detectors of single photons are an essential component for quantum information processing across computing, communication and networking. Today's quantum detection system, which consists of single photon detectors, timing electronics, control and data processing software, is primarily used for counting the number of single photon detection events. However, it is largely incapable of extracting other rich physical characteristics of the detected photons, such as their wavelengths, polarization states, photon numbers, or temporal waveforms. This work, for the first time, demonstrates a smart quantum detection system, SQuaD, which integrates a field programmable gate array (FPGA) with a neural network model, and is designed to recognize the features of photons and to eliminate detector dark-count. The SQuaD is a fully integrated quantum system with high timing-resolution data acquisition, onboard multi-scale data analysis, intelligent feature recognition and extraction, and feedback-driven system control. Our \name experimentally demonstrates 1) reliable photon counting on par with the state-of-the art commercial systems; 2) high-throughput data processing for each individual detection events; 3) efficient dark count recognition and elimination; 4) up to 100% accurate feature recognition of photon wavelength and polarization. Additionally, we deploy the SQuaD to an atomic (erbium ion) photon emitter source to realize noise-free control and readout of a spin qubit in the telecom band, enabling critical advances in quantum networks and distributed quantum information processing.

Linne, Karl C. [U. Chicago (main)] (ORCID:00090009↗