Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “random networks”

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 631 records · Page 35
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Comparative Assessment of U-Net-Based Deep Learning Models for Segmenting Microfractures and Pore Spaces in Digital Rocks

Segmentation of high-resolution X-ray microcomputed tomography (µCT) images is crucial in digital rock physics (DRP), affecting the characterization and analysis of microscale phenomena in the porous media. The complexity of geological structures and nonideal scanning conditions pose significant challenges to conventional image segmentation approaches. Motivated by the recent increasing popularity of deep learning (DL) techniques in image processing, this work undertakes a comparative study of DL models, specifically U-Net and its variants, for segmenting multiple targets with distinguished features in digital rocks, including discrete fracture networks (DFNs), pore spaces, and solid rock. Particularly, DFNs have a smaller volumetric fraction over others, bringing in a substantial challenge of imbalanced segmentation. The primary focus is to evaluate the architecture and feature enhancement strategies of various DL models, including U-Net, attention U-Net, residual U-Net, U-Net++, and residual U-Net++. The models were designed as 2.5D, utilizing a central 2D image and its two adjacent upper and lower 2D images as input to provide a pseudo-3D context. In addition, because the ground truth of segmentation was unknown for real-world digital rocks, we created a benchmark data set following the inverse operations of segmentation. The data synthesis started from the label images (i.e., solid rock, pore spaces, and DFNs), followed by simulating partial volume blurring, adding random background noise, and introducing ring artifacts to mimic real raw X-ray µCT images. The data set, which included various rock types (i.e., sandstone and artificial data), scanning resolution, and magnitudes of noise and artifacts, was divided into training and testing data sets with a 90% and 10% ratio, respectively. Moreover, in addition to the conventional pixel-wise evaluation metrics, the physics-based metric of the lattice-Boltzmann method (LBM) simulated permeability provided more comprehensive assessments. The results demonstrated that the residual connections, nested architectures, and redesigned skip connections contribute to the model performance and give the residual U-Net++ the highest accuracy. The improvements were mainly on the boundaries and small targets, especially the DFNs, which dominate the interconnectivity and therefore affect the permeability greatly. This study also rigorously evaluated the efficiency and generalization of each model, demonstrating that the sophisticated architectures achieved excellent practicability and maintained robust performance on completely unseen data, ensuring their suitability for diverse and challenging DRP applications.

58 GEOSCIENCES↗
DOE OSTI2023

Towards operational atmospheric correction of airborne hyperspectral imaging spectroscopy: Algorithm evaluation, key parameter analysis, and machine learning emulators

Atmospheric correction of airborne hyperspectral imaging spectroscopy (AHIS) to obtain high-quality surface reflectance is the prerequisite for remote sensing applications. Over the last decades, different atmospheric correction methods have been developed based on radiative transfer models (RTMs), however, the relative performances of different algorithms are unclear. Automated operational atmospheric correction methods to process large-volume AHIS data in a high-accurate and high-throughput manner are still lacking. Therefore, this study proposed an operational atmospheric correction pipeline for deriving surface reflectance from AHIS data. To ensure the accuracy and efficiency of the pipeline, we focused on three specific aspects: (1) selecting a suitable RTM for the development of atmospheric lookup tables (LUTs) by comparing the commercial MODerate resolution atmospheric TRANsmission (MODTRAN) and open-sourced Library for Radiative TRANsfer (LibRadTRAN) models, where the widely-used software, Atmospheric/Topographic Correction for Airborne Imagery (ATCOR), was used as benchmarks; (2) identifying key atmospheric correction parameters and determining suitable sources for parameter retrievals including AHIS, Moderate Resolution Imaging Spectroradiometer (MODIS), and AErosol RObotic NETwork (AERONET); and (3) testing the performance of using machine learning emulators to speed up the RTM-based atmospheric correction. Results indicate that (1) atmospheric correction based on MODTRAN LUTs can produce surface reflectance accurately with mean absolute errors < 0.05 and cosine similarities > 0.98 compared to field measurements, which is comparable to the software ATCOR and slightly outperforms the LibRadTRAN LUTs; (2) sobol global sensitivity analysis demonstrates that in the atmospheric correction, visibility and water vapor are two key parameters that can be accurately derived from AHIS in contrast to MODIS or AERONET data; and (3) Random Forest emulators can produce accurate estimations of surface reflectance with mean absolute errors < 0.03 and cosine similarities > 0.98 for higher processing efficiency and determine a suitable set of wavelengths for retrieving atmospheric visibility and water vapor. In conclusion, the proposed atmospheric correction pipeline also improved the four-stream radiative transfer theory for airborne applications by considering adjacent effects from airborne surrounding pixels and can also be applied for atmospheric correction of hyperspectral data from spaceborne missions.

47 OTHER INSTRUMENTATION↗
DOE OSTI2020

Neural MUSE Analysis

Researchers at Oak Ridge National Laboratory (ORNL) created data as part of the MUSE (Multi-Agency Urban Search Experiment Detector and Algorithm Test Bed) project simulating illicit nuclear materials located in various buildings along a road. In the simulation, a truck containing a radiation detector drives down the road gathering listmode data (counting the and energy of incident gamma radiation). Building materials, source shielding, driving speed, truck direction, truck location on the road, source type, and source placement are all varied between runs of the data set. This data was created using deterministic neutron transport and Monte Carlo methods through a combination of SCALE, MAVRIC, MCNP, and GADRAS. As part of a follow-on NA-22 project, two Kaggle competitions were created to determine the best algorithms for finding and identifying gamma sources in this simulated urban environment. The winning algorithm was neural network-based and had a test accuracy of 76.4% accuracy for source identification. This work seeks to build upon this work and improve the results through the application of novel machine learning techniques. As a first step, the data was classified by a simple Convolutional Neural Network (CNN) To accomplish this, the data was first preprocessed into “waterfall plots.” These plots are composed of energy vs count plots that are stacked vertically to show progression in time. The horizontal axis indicating the particle energy incorporated user defined bin spacing with options for in linear-, logarithmic-, square root-, and user-spaced bins. The z or color dimension showed the number of counts corresponding the energy-time combination. This data was then used to generate more data, by generating a local estimate of the mean of the distribution for a bin and then randomly re-sampling that bin from a Poisson distribution. Once all of this data was generated, it was fed into a well-known CNN architecture, ResNet50. The output layer of this model was removed and replaced with layers corresponding to the shape desired isotope outputs. The provided training data was used to train the classifier and the remaining testing data was used to evaluate the model. Results are soon to be forthcoming.

61 RADIATION PROTECTION AND DOSIMETRY↗
DOE OSTI2024

Evaluation of GlassNet for physics-informed machine learning of glass stability and glass-forming ability

Glassy materials form the basis of many modern applications, including nuclear waste immobilization, touch-screen displays, and optical fibers, and also hold great potential for future medical and environmental applications. However, their structural complexity and large composition space make design and optimization challenging for certain applications. Of particular importance for glass processing and design is an estimate of a given composition's glass-forming ability (GFA). However, there remain many open questions regarding the underlying physical mechanisms of glass formation, especially in oxide glasses. It is apparent that a proxy for GFA would be highly useful in glass processing and design, but identifying such a surrogate property has proven itself to be difficult. While glass stability (GS) parameters have historically been used as a GFA surrogate, recent research has demonstrated that most of these parameters are not accurate predictors of the GFA of oxide glasses. Here, in this work, we explore the application of an open-source pre-trained neural network model, GlassNet, that can predict the characteristic temperatures necessary to compute GS with reasonable performance and assess the feasibility of using these physics-informed machine learning (PIML)-predicted GS parameters to estimate GFA. In doing so, we track the uncertainties at each step of the computation—from the original ML prediction errors to the compounding of errors during GS estimation, and finally to the final estimation of GFA. While GlassNet exhibits reasonable accuracy on all individual properties, we observe a large compounding of error in the combination of these individual predictions for the PIML prediction of GS, finding that random forest models offer similar accuracy to GlassNet. We also break down the performance of GlassNet on different glass families and find that the error in GS prediction is correlated with the error in crystallization peak temperature prediction. Lastly, we utilize this finding to assess the relationship between top-performing GS parameters and GFA for two ternary glass systems: sodium borosilicate and sodium iron phosphate glasses. We conclude that to obtain true ML predictive capability of GFA, significantly more data needs to be collected.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Distributed deep reinforcement learning for simulation control

Abstract Several applications in the scientific simulation of physical systems can be formulated as control/optimization problems. The computational models for such systems generally contain hyperparameters, which control solution fidelity and computational expense. The tuning of these parameters is non-trivial and the general approach is to manually ‘spot-check’ for good combinations. This is because optimal hyperparameter configuration search becomes intractable when the parameter space is large and when they may vary dynamically. To address this issue, we present a framework based on deep reinforcement learning (RL) to train a deep neural network agent that controls a model solve by varying parameters dynamically. First, we validate our RL framework for the problem of controlling chaos in chaotic systems by dynamically changing the parameters of the system. Subsequently, we illustrate the capabilities of our framework for accelerating the convergence of a steady-state computational fluid dynamics solver by automatically adjusting the relaxation factors of the discretized Navier–Stokes equations during run-time. The results indicate that the run-time control of the relaxation factors by the learned policy leads to a significant reduction in the number of iterations for convergence compared to the random selection of the relaxation factors. Our results point to potential benefits for learning adaptive hyperparameter learning strategies across different geometries and boundary conditions with implications for reduced computational campaign expenses 4 4 Data and codes available at https://github.com/Romit-Maulik/PAR-RL . .

42 ENGINEERING↗
DOE OSTI2020

Collaborative Fault Tolerant Control of Non-Signalized Intersections for Connected and Autonomous Vehicles

With the potential of increased penetration of connected and autonomous vehicles (CAVs), intersectional signal control faces new challenges in terms of its operation and implementation. One possibility is to fully make use of the communication capabilities of CAVs so that intersectional signal control can be realized by CAVs alone – this leads to non-signalized intersectional operation for traffic networks in urban areas. In this paper, the state-of-the-art on collaborative fault tolerant control schemes for complex systems will be briefly described. This is then followed by the formulation of operational fault tolerant control that realizes the collaborative fault tolerance functionality at CAVs operational level in response to possible individual vehicle faults, where detailed modelling using vehicle movement dynamics will be described together with the construction of fast fault diagnosis and a collaborative fault tolerant control algorithm. A simple example will be given as well to demonstrate the proposed algorithm together with the discussions on other issues such as randomness of the system, communication errors and full energy consideration. These leads to several future directions of the research for the traffic flow control of non-signalized intersections with 100% penetration of CAVs.

Wang, Hong↗
DOE OSTI2021

ECNet is an evolutionary context-integrated deep learning framework for protein engineering

Abstract Machine learning has been increasingly used for protein engineering. However, because the general sequence contexts they capture are not specific to the protein being engineered, the accuracy of existing machine learning algorithms is rather limited. Here, we report ECNet (evolutionary context-integrated neural network), a deep-learning algorithm that exploits evolutionary contexts to predict functional fitness for protein engineering. This algorithm integrates local evolutionary context from homologous sequences that explicitly model residue-residue epistasis for the protein of interest with the global evolutionary context that encodes rich semantic and structural features from the enormous protein sequence universe. As such, it enables accurate mapping from sequence to function and provides generalization from low-order mutants to higher-order mutants. We show that ECNet predicts the sequence-function relationship more accurately as compared to existing machine learning algorithms by using ~50 deep mutational scanning and random mutagenesis datasets. Moreover, we used ECNet to guide the engineering of TEM-1 β-lactamase and identified variants with improved ampicillin resistance with high success rates.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2021

Fast tensor disentangling algorithm

Many recent tensor network algorithms apply unitary operators to parts of a tensor network in order to reduce entanglement. However, many of the previously used iterative algorithms to minimize entanglement can be slow. We introduce an approximate, fast, and simple algorithm to optimize disentangling unitary tensors. Our algorithm is asymptotically faster than previous iterative algorithms and often results in a residual entanglement entropy that is within 10 to 40% of the minimum. For certain input tensors, our algorithm returns an optimal solution. When disentangling order-4 tensors with equal bond dimensions, our algorithm achieves an entanglement spectrum where nearly half of the singular values are zero. We further validate our algorithm by showing that it can efficiently disentangle random 1D states of qubits.

Slagle, Kevin↗
DOE OSTI2025

2024 Results for Avian Monitoring at the Technical Area 36 Minie Site, Technical Area 39 Point 6, Technical Area 16 Burn Ground, and DARHT at Los Alamos National Laboratory

Los Alamos National Laboratory (LANL) biological subject matter experts in the Environmental Protection and Compliance Division initiated a multi-year program in 2013 to monitor avifauna (birds) at two open detonation sites and one open burn site on LANL property. Additional monitoring began in 2017 at a third firing site, the Dual-Axis Radiographic Hydrodynamic Test (DARHT) Facility. In this annual report, we compare monitoring results from these efforts among years to identify and evaluate firing and open burn site impacts on the local bird community. The objectives of this study are • to determine whether LANL operations impact bird abundance, species richness, or diversity; • to examine occupancy and nest success of secondary-cavity nesting birds that use nest boxes; and • to examine chemical concentrations (such as radionuclides, inorganic elements, and/or organic compounds) in nonviable eggs and deceased nestlings that are collected opportunistically with the upper-level bounds of background concentrations, when available. During May through July 2024, LANL biologists completed multiple avian point count surveys at each of the following treatment sites: • Technical Area (TA) 36 Minie Site, • TA-39 Point 6, • TA-16 Burn Ground, and • DARHT. We recorded a total of 1,088 birds that represented 65 species at the four treatment sites and compared these results with data from their associated control sites. In 2024, abundance and species richness at treatment and control sites continued to trend similarly from year to year, with minor random deviations expected from bird communities. Species richness at firing sites differed little from the previous year’s values. Two new bird species were observed at the firing sites—cedar waxwing (Bombycilla cedrorum) and pinyon jay (Gymnorhinus cyanocephalus). Shannon diversity values at TA-36 Minie Site, TA-39, and DARHT were statistically higher than one or more of their associated controls. Annual species diversity at treatment sites was high in 2024 across all firing sites relative to similar habitat control sites. We also monitored avian nest boxes to compare occupancy and nest success data from nest boxes at treatment sites with the overall avian nest box monitoring network and against a subset of relevant control sites. Nest box success has decreased at both treatment and control sites since monitoring began, suggesting that overlapping climatic factors are responsible for patterns of declining nest success. In 2024, nonviable avian eggs and one nestling were opportunistically collected at Bandelier National Monument, TA-16 Burn Ground, TA-36 Minie, TA-39 Point 6, and DARHT. All egg samples and the one nestling sample were evaluated for per- and polyfluoroalkyl substances, which were detected from all locations, including the control site at Bandelier National Monument. Overall results from 2024 continue to suggest that operations at the four treatment sites are not negatively impacting bird populations. This long-term project will continue to monitor for any changes over time.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Leveraging Artificial Intelligence to Predict Novel Eutectic Alloys

The goal of this project was to train an artificial neural network (ANN) to predict the fractional composition and melting point of eutectic alloys using fundamental atomic properties as inputs. The fundamental properties considered include atomic number, atomic weight, atomic radius, valence electron concentration, electronegativity, and electron affinity. The project involved several phases, starting with data preparation, where phase diagram data was harvested from the ASM International database. Approximately 1300 binary eutectics were collected and cleaned to ensure relevance and accuracy. A regression model was selected for training, utilizing a rectified linear unit as the activation function. Various model configurations were evaluated for predictive accuracy, with validation techniques employed to ensure robustness. The model demonstrated predictive capabilities above random guessing and was able to achieve up to 11% accuracy under certain conditions. An ablative test identified atomic radius and valence electron concentration as critical inputs for model performance. Incorporating the melting point of atomic constituents improved accuracy significantly, although ultimately the model’s predictive capability still fell short of the 80% target. This report details the methodology, results, and implications of the research, contributing to the understanding of employing artificial intelligence to predict the phase transition behavior of eutectic alloys.

36 MATERIALS SCIENCE↗
DOE OSTI2020

ROM-Based Surrogate Systems Modeling of EBR-II

We report the System Analysis Module (SAM), developed and maintained by Argonne National Laboratory, is designed to provide whole-plant transient safety analysis capabilities for a number of advanced non-light water reactors, including sodium-cooled fast reactor (SFR), lead-cooled fast reactor (LFR), and molten salt reactor (MSR)/fluoride-salt-cooled high-temperature reactor (FHR) designs. SAM is primarily constructed as a systems-level analysis tool, with the potential to incorporate reduced order models from three-dimensional computational fluid dynamics (CFD) simulations to improve characterization of complex, multidimensional physics. It is recognized that the computational expense associated with CFD can be intractable for various engineering analyses, such as uncertainty quantification, inference, and design optimization. This paper explores the reducibility of a SAM model using recent advances in randomized linear algebra techniques, which attempt to find recurring patterns in the various realizations generated by a model after randomly perturbing all its input parameters. The reduction is described in terms of fewer degrees of freedom (DOFs), referred to as the active DOFs, for the model variables such as input model parameters and model responses. The results indicate that there is significant room for additional reduction that may be leveraged for additional computational gains when employing SAM for engineering-intensive analyses that require repeated model executions. Different from physics-based reduction approaches, the proposed approach allows one to estimate upper bounds on the reduction errors, which are rigorously developed in this work. Finally, different methods for surrogate model construction, such as regression and neural network-based training, are employed to correlate the input and output active DOFs, which are related back to the original variables using matrix-based linear transformations.

42 ENGINEERING↗
DOE OSTI2020

Litter Production of Oak-Hickory Forest at Missouri Ozark (MOFLUX) Site: 2003-2022

This dataset contains measurements of litter production at the second-growth upland oak-hickory forest at the Missouri Ozark AmeriFlux (MOFLUX) site. The MOFLUX site is located in the University of Missouri Baskett Forest situated in the Ozark Border Region of central Missouri, USA, and is part of the AmeriFlux network (site ID: US-MOz). During 2003, 24 circular vegetation plots (each 0.08 hectares) were established in the MOFLUX forest. The plots were situated 50 meters apart on 5 linear transects radiating out from the flux tower base in southeast, south, southwest, west, and northwest directions—there were 5 plots per transect except for the northwest one, which had only 4 due to the presence of a small pond at the terminus (Fig. 1). Litter measurements were initiated in October 2003 and continue to the present. Data through 2022 are reported in this dataset. On each plot, two subplots were established for the collection of litter. Subplots were setup using a stratified random design to ensure the collectors were not immediately adjacent to one another. Litter collectors were usually emptied at monthly intervals, or less during the autumn leaf drop. Samples were dried to constant mass and then sorted before determining the dry mass of leaf, vegetative and woody litter. These litter data have been valuable in modeling efforts to examine the impacts of intensifying moisture extremes (i.e., moisture excess and deficit conditions) on soil carbon cycling. More recent data will be added periodically with no changes to previously released data or the format and structure of the data file. This dataset contains one data file in comma separate (.csv) format. Additional metadata are provided: one data dictionary and a file-level metadata file in comma separate (.csv) format and a user guide in PDF (*.pdf) format.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2022

Classification of Intensity Distributions of Transmission Eigenchannels of Disordered Nanophotonic Structures Using Machine Learning

Light-matter interaction optimization in complex nanophotonic structures is a critical step towards the tailored performance of photonic devices. The increasing complexity of such systems requires new optimization strategies beyond intuitive methods. For example, in disordered photonic structures, the spatial distribution of energy densities has large random fluctuations due to the interference of multiply scattered electromagnetic waves, even though the statistically averaged spatial profiles of the transmission eigenchannels are universal. Classification of these eigenchannels for a single configuration based on visualization of intensity distributions is difficult. However, successful classification could provide vital information about disordered nanophotonic structures. Emerging methods in machine learning have enabled new investigations into optimized photonic structures. In this work, we combine intensity distributions of the transmission eigenchannels and the transmitted speckle-like intensity patterns to classify the eigenchannels of a single configuration of disordered photonic structures using machine learning techniques. Specifically, we leverage supervised learning methods, such as decision trees and fully connected neural networks, to achieve classification of these transmission eigenchannels based on their intensity distributions with an accuracy greater than 99%, even with a dataset including photonic devices of various disorder strengths. Simultaneous classification of the transmission eigenchannels and the relative disorder strength of the nanophotonic structure is also possible. Our results open new directions for machine learning assisted speckle-based metrology and demonstrate a novel approach to classifying nanophotonic structures based on their electromagnetic field distributions. These insights can be of paramount importance for optimizing light-matter interactions at the nanoscale.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2023

Considerations for the temperature stratification in a pre-burn constant-volume combustion chamber

In recent years, the Engine Combustion Network (ECN) has developed as a worldwide reference for understanding and describing engine combustion processes, successfully bringing together experimental and numerical efforts. Since experiments and numerical simulations both target the same boundary conditions, an accurate characterization of the stratified environment that is inevitably present in experimental facilities is required. The difference between the core-, and pressure-derived bulk-temperature of pre-burn combustion vessels has been addressed in various previous publications. Additionally, thermocouple measurements have provided initial data on the boundary layer close to the injector nozzle, showing a transition to reduced ambient temperatures. The conditions at the start of fuel injection influence physicochemical properties of a fuel spray, including near nozzle mixing, heat release computations, and combustion parameters. To address the temperature stratification in more detail, thermocouple measurements at larger distances from the spray axis have been conducted. Both the temperature field prior to the pre-combustion event that preconditions the high-temperature, high-pressure ambient, as well as the stratification at the moment of fuel injection were studied. To reveal the cold boundary layer near the injector with a better spatial resolution, Rayleigh scattering experiments and thermocouple measurements at various distances close to the nozzle have been carried out. The impact of the boundary layers and temperature stratification are illustrated and quantified using numerical simulations at Spray A conditions. Next to a reference simulation with a uniform temperature field, six different stratified temperature distributions have been generated. These distributions were based on the mean experimental temperature superimposed by a randomized variance, again derived from the experiments. The results showed that an asymmetric flame structure arises in the computed results when the temperature stratification input is used. In these predictions, first-stage ignition is advanced by 24 $\mu$s , while second-stage ignition is delayed by 11 $\mu$s. At the same time a lift-off length difference between the top and the bottom of up to 1.1 mm is observed. Furthermore, the lift-off length is less stable over time. Given the shown dependency, the temperature data is made available along with the vessel geometry data as a recommended basis for future numerical simulations.

42 ENGINEERING↗
DOE OSTI2020

Toward particle-resolved accuracy in Euler–Lagrange simulations of multiphase flow using machine learning and pairwise interaction extended point-particle (PIEP) approximation

This study presents two different machine learning approaches for the modeling of hydrodynamic force on particles in a particle-laden multiphase flow. Results from particle-resolved direct numerical simulations (PR-DNS) of flow over a random array of stationary particles for eight combinations of particle Reynolds number (Re) and volume fraction (φ) are used in the development of the models. The first approach follows a two-step process. In the first flow prediction step, the perturbation flow due to a particle is obtained as an axisymmetric superposable wake using linear regression. In the second force prediction step, the force on a particle is evaluated in terms of the perturbation flow induced by all its neighbors using the generalized Faxén form of the force expression. In the second approach, the force data on all the particles from the PR-DNS simulations are used to develop an artificial neural network (ANN) model for direct prediction of force on a particle. Due to the unavoidable limitation on the number of fully resolved particles in the PR-DNS simulations, direct force prediction with the ANN model tends to over-fit the data and performs poorly in the prediction of test data. In contrast, due to the millions of grid points used in the PR-DNS simulations, accurate flow prediction is possible, which then allows accurate prediction of particle force. This hybridization of multiphase physics and machine learning is particularly important, since it blends the strength of each, and the resulting pairwise interaction extended point-particle model cannot be developed by either physics or machine learning alone.

42 ENGINEERING↗
DOE OSTI2023

DDStore: Distributed Data Store for Scalable Training of Graph Neural Networks on Large Atomistic Modeling Datasets

Graph neural networks (GNNs) are a class of Deep Learning models used in designing atomistic materials for effective screening of large chemical spaces. To ensure robust prediction, GNN models must be trained on large volumes of atomistic data on leadership class supercomputers. Even with the advent of modern architectures that consist of multiple storage layers that include node-local NVMe devices in addition to device memory for caching large datasets, extreme-scale model training faces I/O challenges at scale.We present DDStore, an in-memory distributed data store designed for GNN training on large-scale graph data. DDStore provides a hierarchical, distributed, data caching technique that combines data chunking, replication, low-latency random access, and high throughput communication. DDStore achieves near-linear scaling for training a GNN model using up to 1000 GPUs on the Summit and Perlmutter supercomputers, and reaches up to a 6.15x reduction in GNN training time compared to state-of-the-art methodologies.

Choi, Jong Youl↗
DOE OSTI2023

Operation Optimization Using Reinforcement Learning with Integrated Artificial Reasoning Framework

In large and complex systems, operational decision-making requires a systematic analysis with a vast amount of data from both process parameters and component status monitoring. In this paper, we present an integrated artificial reasoning approach for system state transition models that can help operational decision-making with explainable and traceable reasoning. The integrated artificial reasoning framework is a physics-based approach of defining the system structure in a Bayesian network, so we leveraged it in a Markov decision process (MDP) for finding optimal operational solutions. In our proposed framework, the MDP is implemented on a dynamic Bayesian network (DBN), which represents causalities in a system. The multilevel flow modeling was utilized in order to extract these causalities in a more efficient and objective manner. Since multilevel flow modeling is based on the fundamental energy and mass conservation laws, the target system is decomposed into several mass, energy, and information structures, which serve as the basis for a DBN. The MDP consists of the processes of finding a solution for the Bellman equation, which can be derived from the conditional probability equations of the constructed DBN. System operators can capture stochastic system dynamics as multiple subsystem state transitions based on their physical relations and uncertainties coming from the component degradation process or random failures. We analyzed a simplified example system to illustrate finding an optimal operational policy with this approach.

Kim, Junyung↗
DOE OSTI2021

Total absorption spectroscopy of the β decay of 101,102 Zr and 109 Tc

The β decay of 101,102 Zr and 109 Tc was studied using the technique of total absorption spectroscopy. The experiment was performed at the National Superconducting Cyclotron Laboratory using the Summing NaI(Tl) (SuN) detector in the first-ever application of total absorption spectroscopy with a fast beam produced via projectile fragmentation. The β -decay feeding intensity and Gamow-Teller transition strength distributions were extracted for these three decays. The extracted distributions were compared to three different quasiparticle random-phase approximation (QRPA) models based on different mean-field potentials. A comparison with calculations from one of the QRPA models was performed to learn about the ground-state shape of the parent nucleus. For 101 Zr and 102 Zr, calculations assuming a pure shape configuration (oblate or prolate) were not able to reproduce the extracted distributions. These results may indicate that some type of mixture between oblate and prolate shapes is necessary to reproduce the extracted distributions. For 109 Tc, a comparison of the extracted distributions with QRPA calculations suggests a dominant oblate configuration. The other two QRPA models are commonly used to provide β-decay properties in r-process network calculations. This work shows the importance of making comparisons between the experimental and theoretical β-decay distributions, rather than just half-lives and β-delayed neutron emission probabilities, as close to the r-process path as possible.

79 ASTRONOMY AND ASTROPHYSICS↗