Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Data transfer”

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 685 records · Page 38
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Building a DFT+U machine learning interatomic potential for uranium dioxide

Despite uranium dioxide (UO 2 ) being a widely used nuclear fuel, fuel performance models rely extensively on empirical correlations of material behavior, leveraging the historical operating experience of UO 2 . Mechanistic models that consider an atomistic understanding of the processes governing fuel performance (such as fission gas release and creep) will enable a better description of fuel behavior under non-prototypical conditions such as in new reactor concepts or for modified UO 2 fuel compositions. To this end, molecular dynamics simulation is a powerful tool for rapidly predicting physical properties of proposed fuel candidates. However, the reliability of these simulations depends largely on the accuracy of the atomic forces. Traditionally, these forces are computed using either a classical force field (FF) or density functional theory (DFT). While DFT is relatively accurate, the computational cost is burdensome, especially for f-electron elements, such as actinides. By contrast, classical FFs are computationally efficient but are less accurate. For these reasons, we report a new accurate machine learning interatomic potential (MLIP) for UO 2 that provides high-fidelity reproduction of DFT forces at a similar low cost to classical FFs. We employ an active learning approach that autonomously augments the DFT training data set to iteratively refine the MLIP. To further improve the quality of our predictions, we utilize transfer learning to retrain our MLIP to higher-accuracy DFT+U data. We validate our MLIPs by comparing predicted physical properties (e.g., thermal expansion and elastic properties) with those from existing classical FFs and DFT/DFT+U calculations, as well as with experimental data when available.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2025

Advanced Interactive 3D Visualization Tool for Customizable Analyses of Tomography Datasets in Material Science

Current methods for visualizing and analyzing 3D tomography datasets in materials science often lack the interactivity and depth required for detailed structural insights. This limitation restricts a researchers' ability to accurately interpret complex data, which is critical for advancing material innovations and understanding structural properties. To address this issue, we have developed a novel, web-based interactive 3D visualization and analysis tool from the Trame framework that offers customizable features to enhance data interpretability. The tool allows users to adjust parameters such as visible range, slice planes, data rotation, and layering, providing a more detailed and dynamic view of complex structures. Its user-friendly web interface increases the accessibility and ease of use for both novice and experienced researchers, to visualize large volumetric datasets. The tool supports a diverse range of data formats, making it versatile for various research applications. Unique capabilities include real-time data manipulation, automated feature detection, context-sensitive feedback, and real-time volume calculations and distributions per sliced region or layer, alongside the ability to quickly generate high-quality screenshots and videos for presentations and reports. These advancements offer a comprehensive solution for enhanced 3D data exploration, significantly improving the analysis process and communication of results in materials science.

36 - MATERIALS SCIENCE↗
DOE OSTI2025

Machine learning for reactor power monitoring with limited labeled data

Real-time reactor power monitoring is critical for a variety of nuclear applications, spanning safety, security, operations, and maintenance. While machine learning methods have shown promise in monitoring reactor power levels, there is limited research on their efficacy in label-starved environments. The goal of this work is to assess the feasibility of classifying nuclear reactor power level using multisource data in scenarios with limited labels. Data were collected using low-resolution multisensors at four nuclear reactor facilities: two large research reactors and two TRIGA reactors. Within each pair, one reactor dataset served as the source and the other as the target in a transfer learning paradigm. Twenty-three supervised models were trained on labeled sequences of magnetic field and acceleration data from each of the target sites. Self-learning and transfer learning methods were applied to the top performing models to assess their classification performance with increasing amounts of labeled data. While reactor power level classification was achieved with a Matthews Correlation Coefficient of up to 0.739 ± 0.003 and 0.622 ± 0.009 with only 400 sequences per power state for the large research reactor and TRIGA target sites, respectively, self-learning and transfer learning leveraging source site data did not improve target classification performance. These findings suggest that alternative methods, such as higher sensitivity sensors, digital twins, or the use of physics-informed models, are required to enable high-performance classification in machine learning approaches to reactor monitoring with a dearth of target ground truth.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2025

HPDR: High-Performance Portable Scientific Data Reduction Framework

The rapid growth in scientific data generation is outpacing advancements in computing systems necessary for efficient storage, transfer, and analysis, particularly in the context of exascale computing. With the deployment of first-generation exascale computing systems and next-generation experimental facilities, this gap is widening and necessitates effective data reduction techniques to manage enormous data volumes. Over the past decade, various data reduction methods, including lossless compression, error-controlled lossy compression, and data refactoring, have been developed to accelerate I/O in scientific workflows. Despite significant reductions in data volume, these methods introduce considerable computational overhead, which can become the new bottleneck in data processing. To mitigate this, GPU-accelerated data reduction algorithms have been introduced. However, challenges remain in their integration into exascale workflows, including limited portability across different GPU architectures, substantial memory transfer overhead, and reduced scalability on dense multi-GPU systems. To address these challenges, we propose HPDR, a high-performance and portable data reduction framework. HPDR is designed to enable the execution of state-of-the-art reduction algorithms across diverse processor architectures while reducing memory transfer overhead to 2.3 % of the original, resulting in up to 3.5× faster throughput compared to existing solutions. It also achieves up to 96% of the theoretical speedup in multi-GPU settings. In addition, evaluations on accelerating I/O operations at scale up to 1,024 nodes of the Frontier supercomputer demonstrate that HPDR can achieve up to 103 TB/s reduction throughput, providing up to 4× acceleration in parallel I/O performance compared to existing data reduction routines. This work highlights the potential of HPDR to significantly enhance data reduction efficiency in exascale computing environments.

Chen, Jieyang [University of Oregon]↗
DOE OSTI2024

Proton removal from Br 73 , 75 to Se 72 , 74 at intermediate energies

Herein we report new experimental data for excited states of 72,74 Se obtained from proton removal from 73,75 Br secondary beams on a proton target. The experiments were performed with the Ursinus-NSCL Liquid Hydrogen Target and the combined GRETINA+S800 setup at the Coupled Cyclotron Facility of the National Superconducting Cyclotron Laboratory at Michigan State University. Within uncertainties, the inclusive cross sections for proton removal from 73,75 Br on a proton target are identical suggesting that the same single-particle orbitals contribute to the proton-removal reaction. In addition, details of the partial cross section fragmentation are discussed. The data might suggest that l = 1, 2, 3, and 4 angular momentum transfers are important to understand the population of excited states of 72,74 Se in proton removal. Available data for excited states of 74 Ge populated through the 75 As(d, 3 He) 74 Ge proton-removal reaction in normal kinematics suggest indeed that the fp and sd shell as well as the 1g 9/2 orbital contribute. A comparison to data available for odd-A nuclei supports that the bulk of the spectroscopic strengths could be found at lower energies in the even-even Se isotopes than in, for instance, the even-even Ge isotopes. In addition, the population of high-J states seems to indicate that multi-step processes contribute to proton-removal reactions at intermediate energies in these collective nuclei.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

Application of Sequential Design of Experiments (SDoE) to Large Pilot-Scale Solvent-Based CO2 Capture Process at Technology Centre Mongstad (TCM)

The United States Department of Energy’s Carbon Capture Simulation for Industry Impact (CCSI2) program has developed a framework for sequential design of experiments (SDoE) that aims to maximize knowledge gained from budget- and schedule-limited pilot scale testing. SDoE was applied to the planning and execution of campaigns for testing CO2 capture systems at pilot-scale in order to optimally allocate resources available for the testing. In this methodology, a stochastic process model is developed by quantifying the parametric uncertainty in submodels of interest; for a solvent-based CO2 capture system, these may include physical properties and equipment performance submodels (e.g., mass transfer, interfacial area). This uncertainty is propagated through the full process model, over variable operating conditions, for estimating the resulting uncertainty in key model outputs (e.g., percentage of CO2 capture, solvent regeneration energy requirement). In developing a data collection plan, the predicted output uncertainty is incorporated into an algorithm that seeks simultaneously to select process operating conditions for which the predicted uncertainty is relatively high and to ensure that the entire space of operation is well represented. This test plan is then used to guide operation of the pilot plant at varying steady-state conditions, with resulting process data incorporated into the existing model using Bayesian inference to refine parameter distributions. The updated stochastic model, with reduced parametric uncertainty from data collected, is then used to guide additional data collection, thus the sequential nature of the experimental design. The SDoE process was implemented at the pilot test unit (12 MWe in scale) at Norway’s Technology Centre Mongstad (TCM) in a summer 2018 test campaign with aqueous monoethanolamine (MEA). During the test campaign, the varied operating conditions included the flowrates of circulated solvent, flue gas, and reboiler steam and the CO2 concentration in the flue gas. The process data were used to update probability distributions of mass transfer and interfacial area parameters of a stochastic process model developed by the CCSI2 team. Two iterations of the SDoE process were executed, resulting in the uncertainty in model predicted CO2 capture percentage decreasing by an average of 58.0 ± 4.7% over the full input space of interest. This work demonstrates the potential of the SDoE process for model refinement through reduction in process model parametric uncertainty, and ultimately risk in scale-up, in CO2 capture technology performance.

carbon capture↗
DOE OSTI2022

Realistic galaxy image simulation via score-based generative models

ABSTRACT We show that a denoising diffusion probabilistic model (DDPM), a class of score-based generative model, can be used to produce realistic mock images that mimic observations of galaxies. Our method is tested with Dark Energy Spectroscopic Instrument (DESI) grz imaging of galaxies from the Photometry and Rotation curve OBservations from Extragalactic Surveys (PROBES) sample and galaxies selected from the Sloan Digital Sky Survey. Subjectively, the generated galaxies are highly realistic when compared with samples from the real data set. We quantify the similarity by borrowing from the deep generative learning literature, using the ‘Fréchet inception distance’ to test for subjective and morphological similarity. We also introduce the ‘synthetic galaxy distance’ metric to compare the emergent physical properties (such as total magnitude, colour, and half-light radius) of a ground truth parent and synthesized child data set. We argue that the DDPM approach produces sharper and more realistic images than other generative methods such as adversarial networks (with the downside of more costly inference), and could be used to produce large samples of synthetic observations tailored to a specific imaging survey. We demonstrate two potential uses of the DDPM: (1) accurate inpainting of occluded data, such as satellite trails, and (2) domain transfer, where new input images can be processed to mimic the properties of the DDPM training set. Here we ‘DESI-fy’ cartoon images as a proof of concept for domain transfer. Finally, we suggest potential applications for score-based approaches that could motivate further research on this topic within the astronomical community.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2021

Thermal Experiments for Fractured Rock Characterization: Theoretical Analysis and Inverse Modeling

Abstract Field‐scale properties of fractured rocks play a crucial role in many subsurface applications, yet methodologies for identification of the statistical parameters of a discrete fracture network (DFN) are scarce. We present an inversion technique to infer two such parameters, fracture density and fractal dimension, from cross‐borehole thermal experiments data. It is based on a particle‐based heat‐transfer model, whose evaluation is accelerated with a deep neural network (DNN) surrogate that is integrated into a grid search. The DNN is trained on a small number of the heat‐transfer model runs and predicts the cumulative density function of the thermal field. The latter is used to compute fine posterior distributions of the (to be estimated) parameters. Our synthetic experiments reveal that fracture density is well constrained by data, while fractal dimension is harder to determine. Adding nonuniform prior information related to the DFN connectivity improves the inference of this parameter.

Zhou, Zitong↗
DOE OSTI2020

First measurement of direct photoproduction of the a 2 (1320) 0 meson on the proton

We present the first measurement of the reaction γ p → a 2 ( 1320 ) 0 p in the photon energy range 3.5– 5.5 Ge V and four-momentum transfer squared 0.2 < - t < 2.0 Ge V 2 . Data were collected with the CEBAF Large Acceptance Spectrometer detector at the Thomas Jefferson National Accelerator Facility. The a 2 resonance was detected by measuring the reaction γ p → π 0 η p and reconstructing the π 0 η invariant mass. The most prominent feature of the differential cross section is a dip at -t ≈ 0.55 Ge V 2 . This can be described in the framework of Regge phenomenology, where the exchange degeneracy hypothesis predicts a zero in the reaction amplitude for this value of the four-momentum transfer.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2023

High–Resolution Maps of Near–Surface Permafrost for Three Watersheds on the Seward Peninsula, Alaska Derived From Machine Learning

Permafrost soils are a critical component of the global carbon cycle and are locally important because they regulate the hydrologic flux from uplands to rivers. Furthermore, degradation of permafrost soils causes land surface subsidence, damaging infrastructure that is crucial for local communities. Regional and hemispherical maps of permafrost are too coarse to resolve distributions at a scale relevant to assessments of infrastructure stability or to illuminate geomorphic impacts of permafrost thaw. Here we train machine learning models to generate meter–scale maps of near–surface permafrost for three watersheds in the discontinuous permafrost region. The models were trained using ground truth determinations of near–surface permafrost presence from measurements of soil temperature and electrical resistivity. We trained three classifiers: extremely randomized trees (ERTr), support vector machines (SVM), and an artificial neural network (ANN). Model uncertainty was determined using k–fold cross validation, and the modeled extents of near–surface permafrost were compared to the observed extents at each site. At–a–site near–surface permafrost distributions predicted by the ERTr produced the highest accuracy (70%–90%). However, the transferability of the ERTr to the sites outside of the training data set was poor, with accuracies ranging from 50% to 77%. The SVM and ANN models had lower accuracies for at–a–site prediction (70%–83%), yet they had greater accuracy when transferred to the non–training site (62%–78%). These models demonstrate the potential for integrating high–resolution spatial data and machine learning models to develop maps of near–surface permafrost extent at resolutions fine enough to assess infrastructure vulnerability and landscape morphology influenced by permafrost thaw.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

A geospatial risk analysis graphical user interface for identifying hazardous chemical emission sources

Background: Performing back trajectory and forward trajectory using the Hybrid Single-Particle Lagrangian Integrated Trajectory Model (HYSPLIT) is a reliable approach for assessing particle transport after release among mid-field atmospheric models. HYSPLIT has an externally facing online interface that allows non-expert users to run the model trajectories without requiring extensive training or programming. However, the existing HYSPLIT interface is limited if simulations have a large amount of meteorological data and timesteps that are not coincident. The objective of this study is to design and develop a more robust tool to rapidly evaluate hazard transport conditions and to perform risk analysis, while still maintaining an intuitive and user-friendly interface. Methods: HYSPLIT calculates forward and backward trajectories of particles based on wind speed, wind direction, and the corresponding location, timestamp, and Pasquill stability classes of the regions of the atmosphere in terms of the wind speed, the amount of solar radiation, and the fractional cloud cover. The computed particle transport trajectories, combined with the online Proton Transfer Reaction-Mass Spectrometry (PTR-MS) data (https://figshare.com/articles/dataset/ARL_Data_from_PROS_station_at_Hanford_site/19993964), can be used to identify and quantify the sources and affected area of the hazardous chemicals’ emission using the potential source distribution function (PSDF). PSDF is an improved statistical function based on the well-known potential source contribution function (PSCF) in establishing the air pollutant source and receptor relationship. Performing this analysis requires a range of meteorological and pollutant concentration measurements to be statistically meaningful. The existing HYSPLIT graphical user interface (GUI) does not easily permit computations of trajectories of a dataset of meteorological data in high temporal frequency. To improve the performance of HYSPLIT computations from a large dataset and enhance risk analysis of the accidental release of material at risk, a geospatial risk analysis tool (GRAT-GUI) is created to allow large data sets to be processed instantaneously and to provide ease of visualization. Results: The GRAT-GUI is a native desktop-based application and can be run in any Windows 10 system without any internet access requirements, thus providing a secure way to process large meteorological datasets even on a standalone computer. GRAT-GUI has features to import, integrate, and convert meteorological data with various formats for hazardous chemical emission source identification and risk analysis as a self-explanatory user interface. The tool is available at https://figshare.com/articles/software/GRAT/19426742.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Forest reorganisation effects on fuel moisture content can exceed changes due to climate warming in wet temperate forests

Abstract The distributions of vegetation and fire activity are changing rapidly in response to climate warming. In many regions, climate effects on dead fuel moisture content (FMC) are expected to increase future wildfire activity. However, forest FMC is largely driven by microclimate conditions, which are moderated from open weather by vegetation canopies. As shifts in vegetation increase under climate warming, the extent to which future fire activity will be driven by climate directly or associated vegetation shifts remains unresolved. Here, we present a study aimed at quantifying the relative magnitudes of (i) direct climate warming, and (ii) vegetation change, on FMC. Field sites to evaluate these effects were established in a natural laboratory of altered forest states to mature wet temperate forest in south‐eastern Australia. FMC was estimated using a process‐based model and 48 years of reconstructed climate data. Canopy effects on microclimate were captured by transferring inputs from climate to microclimate using models parameterised with field observations. To evaluate the relative magnitude of climate and vegetation effects, we calculated the maximum difference in mean annual FMC across annual climate replicates and compared this to FMC differences across reorganising forest sites. Our results show vegetation effects on FMC can exceed those related to expected climate change. Changes to forest structure and composition increased (+15.7%) and decreased (−12.3%) mean annual FMC, with a larger negative effect when forest cover was completely removed (−18.5%). In contrast, the largest climate effect on FMC was −6.6% across 48‐years of data. Our study demonstrates that the magnitude of vegetation effects on FMC can exceed expected climate change effects. Models of future fire activity that do not account for changing vegetation effects on microclimate are omitting a key biophysical control on FMC and therefore may not be accurately predicting future fire activity.

Brown, Tegan P.↗
DOE OSTI2021

Advanced Multi-Tube Mixer Combustion for 65% Efficiency (Final Report)

This project targeted advanced low NOx combustion for advanced gas turbines capable of 65%, or greater, efficiency in combined cycle application. This technology advancement has further potential to benefit gas turbines used in coal based IGCC applications with pre-combustion carbon capture and hydrogen as the resulting fuel. The program developed and synthesized the most advanced combustion system capable of achieving low NOX emissions up to turbine inlet temperatures of 3100F while also supporting the load-following needs of a modern grid. The combustion system contributes to the overall gas turbine efficiency goal by setting the maximum cycle temperature achievable for a given NOX level and by minimizing the through-combustor air flow pressure drop. Focus areas for this project targeted maximizing the turbine inlet temperature entitlement, as constrained by emissions considerations. The design also minimized parasitic air flow pressure drop by using advanced cooling techniques and performance materials selections and by minimizing hot surface area. These two technology objectives (maximum, emissions-compliant cycle temperature and minimum air flow pressure drop) were integrated into a prototype design. The primarily analytical project sought to identify the most promising technologies to meet these objectives. Additional critical “jugular” data were obtained from multi-tube mixer tests to realize the potential of leveraging “micro flames” for minimizing overall hot surface area. This data was used, in conjunction with an understanding of advanced material and cooling design technologies, to analytically develop multiple design concepts. Phase I focused on in-depth engineering analysis and design, with minimal supporting laboratory testing to enable a selection of the top three combustion architectures for achieving these overall objectives. Phase II of the program developed the selected design through a combination of sub-scale testing and analytical efforts. Early tests included a cold-flow cascade to establish aerodynamic performance characteristics and a sub-scale fired test at GE Global Research in Niskayuna, NY, to establish cooling and heat transfer characteristics in conjunction with combustion performance. The data from these tests validated the analytical models to ultimately design a full-scale, test article to evaluate at prototypical pressure and temperature conditions at GE Gas Power’s Gas Turbine Technology Laboratory in Greenville, SC. GE Gas Power also developed, tested, and recommended a suitable seal design to be applied to the unique features of the combustor. To assess the technology challenges from prospective future production of the combustor from a ceramic matrix composite material, screening tests of Environmental Barrier Coatings were completed.

20 FOSSIL-FUELED POWER PLANTS↗
DOE OSTI2020

Power Grid Contingency Analysis with Machine Learning: A Brief Survey and Prospects

We briefly review previous applications of machine learning (ML) in power grid analyses and introduce our ongoing effort toward developing a generative-adversarial (GA) model for fast and reliable grid contingency analyses. According to our review, the persisting limitation of traditional ML techniques in grid analyses is the need for an exhaustive amount of training data for model generalization and accurate predictions. GA models overcome this limitation by first learning true data distribution from a small training set, from which new samples assimilating true data are generated with some variations. Subsequently, GA models can transfer learn or super-generalize with increased accuracy, that is, accurately predict n - (k + 2) contingencies from a small n - k training set and generated n - (k + 1) data. The joint effort between Idaho National Lab and Florida State University strives to develop a zero-shot and deep learning-based contingency analysis tool, named Smart Contingency Analysis Neural Network (SCANN), by leveraging the aforementioned advantages of GA models. The basic architecture of SCANN stems from the Latent Encoding of Atypical Perturbations network combined with an adversarial network, and it is designed to generate imbalanced power flow data from learned true data distributions for prediction purposes. Here we also introduce the abstract concept of resilience-chaos plots, a new resilience characterization tool proposed to complement SCANN by aiding in the assessment of large amounts of high-order contingency predictions.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2020

Predicting Small Molecule Transfer Free Energies by Combining Molecular Dynamics Simulations and Deep Learning

Accurately predicting small molecule partitioning and hydrophobicity is critical in the drug discovery process. There are many heterogeneous chemical environments within a cell and entire human body. For example, drugs must be able to cross the hydrophobic cellular membrane to reach their intracellular targets, and hydrophobicity is an important driving force for drug–protein binding. Atomistic molecular dynamics (MD) simulations are routinely used to calculate free energies of small molecules binding to proteins, crossing lipid membranes, and solvation but are computationally expensive. Machine learning (ML) and empirical methods are also used throughout drug discovery but rely on experimental data, limiting the domain of applicability. We present atomistic MD simulations calculating 15,000 small molecule free energies of transfer from water to cyclohexane. This large data set is used to train ML models that predict the free energies of transfer. We show that a spatial graph neural network model achieves the highest accuracy, followed closely by a 3D-convolutional neural network, and shallow learning based on the chemical fingerprint is significantly less accurate. A mean absolute error of ~4 kJ/mol compared to the MD calculations was achieved for our best ML model. We also show that including data from the MD simulation improves the predictions, tests the transferability of each model to a diverse set of molecules, and show multitask learning improves the predictions. This work provides insight into the hydrophobicity of small molecules and ML cheminformatics modeling, and our data set will be useful for designing and testing future ML cheminformatics methods.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Resolving Heterogeneous Dynamics of Excess Protons in Aqueous Solution with Rate Theory

Rate theories have found great utility across the chemical sciences by providing a physically transparent way to analyze dynamical processes. Here we demonstrate the benefits of using transition state theory and Marcus theory to study the rate of proton transfer in HCl solutions. By using long ab initio molecular dynamics simulations we show that good agreement is obtained between these two different formulations of rate theory and how they can be used to study the pathways and life-time of proton transfer in aqueous solution. Since both rate theory formulations utilize identical sets of molecular data, this provides a self-consistent theoretical picture of the rates of aqueous phase proton transfer. Specifically, we isolate and quantify the rates of proton transfer, ion-pair dissociation, and solvent exchange in concentrated HCl solutions. Our analysis predicts a concentration dependence to both proton transfer and ion-pairing. Moreover, our estimate of the life-time for the Zundel species is 0.8 ps and 1.3 ps for 2M and 8M HCl, respectively. In this paper, we demonstrate that concentration effects, can indeed be quantified through the combination of state-of-the-art simulation and theory and provides a picture of the important correlations between the cation (hydronium) and the counter-ion in acid solutions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2021

Accelerating Multigrid-based Hierarchical Scientific Data Refactoring on GPUs

Rapid growth in scientific data and a widening gap between computational speed and I/O bandwidth make it increasingly infeasible to store and share all data produced by scientific simulations. Instead, we need methods for reducing data volumes: ideally, methods that can scale data volumes adaptively so as to enable negotiation of performance and fidelity tradeoffs in different situations. Multigrid-based hierarchical data representations hold promise as a solution to this problem, allowing for flexible conversion between different fidelities so that, for example, data can be created at high fidelity and then transferred or stored at lower fidelity via logically simple and mathematically sound operations. However, the effective use of such representations has been hindered until now by the relatively high costs of creating, accessing, reducing, and otherwise operating on such representations. We describe here highly optimized data refactoring kernels for GPU accelerators that enable efficient creation and manipulation of data in multigrid-based hierarchical forms. We demonstrate that our optimized design can achieve up to 250 TB/s aggregated data refactoring throughput—83% of theoretical peak—on 1024 nodes of the Summit supercomputer. We showcase our optimized design by applying it to a large-scale scientific visualization workflow and the MGARD lossy compression software.

Data refactoring↗