Influence of business models on PV-battery dispatch decisions and market value: A pilot study of operating plants
Not Available
25 ENERGY STORAGE↗
SEARCH · Engineering Papers
Results for “empirical data”
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.
Microsimulation of Energy and Flow Effects from Optimal Automated Driving in Mixed Traffic
In this paper we study the energy and traffic impact of a proposed Anticipative Cruise Controller in a PTV VISSIM microsimulation environment. We dissect our controller into two parts: 1. the unconnected mode, active when following a human-driven vehicle, and 2. the connected mode, active when following another automated vehicle equipped with connectivity. Probabilistic constraints balance safety considerations with inter-vehicle compactness, and vehicle constraints for acceleration capabilities are expressed through the use of powertrain maps. Emergent highway traffic scenarios are then modeled using time headway distributions from empirical traffic data. To study the impact of automation over a range of demands of free-flow to stop-and-go, we vary vehicle flux from low to high and vary automated vehicle penetration from low to high. When examining all-human driving scenarios, network capacity failed to meet demand in high-volume scenarios, such as rush-hour traffic. We further find that with connected automated vehicles introduced, network capacity was improved to support the high-volume scenarios. Finally, we examine energy efficiencies of the fleet for conventional, electric, and hybrid vehicles. We find that automated vehicles perform at a 10%-20% higher energy efficiency over human drivers when considering conventional powertrains, and find that automated vehicles perform at a 3%-9% higher energy efficiency over human drivers when considering electric and hybrid powertrains. Due to secondary effects of smoothing traffic flow and reducing unnecessary braking, energy benefits also apply to human-driven vehicles that interact with automated ones. Such simulated humans were found to drive up to 10% more energy-efficiently than they did in the baseline all-human scenario.
32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
Multiscale Modeling of the Mechanical Response of Silicon Carbide Composite Within the Accelerated Fuel Qualification Framework
The accelerated fuel qualification (AFQ) framework has been used for the initial development of multiscale modeling of silicon carbide (SiC) fiber reinforced composite (SiC-SiC). The AFQ framework provides a methodology to leverage physics-informed multiscale modeling along with a reduced set of empirical test data to reduce the time and cost of licensing and qualification of new nuclear fuel systems while maintaining the overall nuclear power plant safety case. SiC-SiC is being proposed for in-core applications, most notably fuel cladding, for current and next-generation nuclear reactors because of its high temperature stability, irradiation tolerance, and ability to withstand many accident conditions. As these composites exhibit multiscale architectures and complex microstructure-based fracture mechanics, it is an appealing use case for the AFQ methodology. While the end goal of this work is a single multiscale model that can be used for predictive in-core performance, current focus is on the individual various length scale models. Four individual models have been initially developed from microscale to engineering system level to capture key physics-based effects across different length scales. These models include a microscale homogenized tow model, a mesoscale fast Fourier transform–based weave model that integrates the homogenized tow model, a mesoscale finite element–based weave model, and a system-level BISON fuel performance model. Results of these models have undergone an initial comparison with separate-effects test data showing a good match to experimental results. By using the AFQ framework during model development, several near-term benefits have been secured including a reduction in development time for the SiC-SiC cladding, more targeted irradiation testing, and a better understanding of uncertainty.
11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
Measuring H 0 using X-ray and SZ effect observations of dynamically relaxed galaxy clusters
ABSTRACT We use a sample of 14 massive, dynamically relaxed galaxy clusters to constrain the Hubble constant, H0, by combining X-ray and Sunyaev–Zel’dovich (SZ) effect signals measured with Chandra, Planck, and Bolocam. This is the first such analysis to marginalize over an empirical, data-driven prior on the overall accuracy of X-ray temperature measurements, while our restriction to the most relaxed, massive clusters also minimizes astrophysical systematics. For a cosmological-constant model with Ωm = 0.3 and ΩΛ = 0.7, we find $H_0 = 67.3^{+21.3}_{-13.3}\, \mathrm{km}\, \mathrm{s}^{-1}\, \mathrm{Mpc}^{-1}$, limited by the temperature calibration uncertainty (compared to the statistically limited constraint of $H_0 = 72.3^{+7.6}_{-7.6}\, \mathrm{km}\, \mathrm{s}^{-1}\, \mathrm{Mpc}^{-1}$). The intrinsic scatter in the X-ray/SZ pressure ratio is found to be 13 ± 4 per cent (10 ± 3 per cent when two clusters with significant galactic dust emission are removed from the sample), consistent with being primarily due to triaxiality and projection. We discuss the prospects for reducing the dominant systematic limitation to this analysis, with improved X-ray calibration and/or precise measurements of the relativistic SZ effect providing a plausible route to per cent level constraints on H0.
79 ASTRONOMY AND ASTROPHYSICS↗
A systematic analysis and data mining of opioid-related adverse events submitted to the FAERS database
The opioid epidemic has become a serious national crisis in the United States. An indepth systematic analysis of opioid-related adverse events (AEs) can clarify the risks presented by opioid exposure, as well as the individual risk profiles of specific opioid drugs and the potential relationships among the opioids. In this study, 92 opioids were identified from the list of all Food and Drug Administration (FDA)-approved drugs, annotated by RxNorm and were classified into 13 opioid groups: buprenorphine, codeine, dihydrocodeine, fentanyl, hydrocodone, hydromorphone, meperidine, methadone, morphine, oxycodone, oxymorphone, tapentadol, and tramadol. A total of 14,970,399 AE reports were retrieved and downloaded from the FDA Adverse Events Reporting System (FAERS) from 2004, Quarter 1 to 2020, Quarter 3. After data processing, Empirical Bayes Geometric Mean (EBGM) was then applied which identified 3317 pairs of potential risk signals within the 13 opioid groups. Based on these potential safety signals, a comparative analysis was pursued to provide a global overview of opioid-related AEs for all 13 groups of FDA-approved prescription opioids. The top 10 most reported AEs for each opioid class were then presented. Both network analysis and hierarchical clustering analysis were conducted to further explore the relationship between opioids. Results from the network analysis revealed a close association among fentanyl, oxycodone, hydrocodone, and hydromorphone, which shared more than 22 AEs. In addition, much less commonly reported AEs were shared among dihydrocodeine, meperidine, oxymorphone, and tapentadol. On the contrary, the hierarchical clustering analysis further categorized the 13 opioid classes into two groups by comparing the full profiles of presence/absence of AEs. The results of network analysis and hierarchical clustering analysis were not only consistent and cross-validated each other but also provided a better and deeper understanding of the associations and relationships between the 13 opioid groups with respect to their adverse effect profiles.
Research & Experimental Medicine↗
Boundary Effects in the Diffusion of New Products on Cartesian Networks
The Role of Boundaries in the Spreading of Solar Peer effects by neighbors play a key role in the spreading of residential solar. Thus, people are more likely to install a solar system on their roof if some of their neighbors have already done so. Because people who live near the municipality boundary have fewer neighbors, does this imply that they are less likely to adopt solar? In “Boundary Effects in the Diffusion of New Products on Cartesian Networks,” Fibich, Levin, and Gillingham analyze this problem analytically using the Bass model on two-dimensional networks and empirically using data on installations of solar systems. They show that boundaries have a significant impact on the adoption of residential units near the municipality boundary. Their effect on the aggregate adoption in the municipality, however, is negligible.
Business & Economics↗
Quantitative Characterization of Hyper-Local Atmospheric Greenhouse Gas Sources
Atmospheric greenhouse gas (GHG) emissions are often characterized using stationary, tower-based sensors. Ground based sensors reside in the turbulent boundary layer and are subject to intense concentration impulses from hyper-local (<100m) point sources of emissions. These high frequency spikes are often filtered out in broader emission flux studies, losing valuable information about how hyper-local sources influence receptors. In this study, we investigated how empirical atmospheric data can be used to locate and quantify a concurrently measured hyper-local point source in a dense urban setting. An eddy covariance style tower and a low-cost sensor tower were deployed in various locations around an urban, hyper-local CO2/CH4 emissions source (a continuously measured restaurant exhaust vent). A model using different processing and statistical techniques was built to examine the most effective procedures for source isolation, directional location, and emission quantification. Using excess concentrations above a minimum baseline, we identify the source using bivariate polar plots and quantify the relationship between source size, receptor distance, and statistical proxies. Furthermore, we find that varying statistical thresholds allows for identification of less influential sources which are drowned out by larger or closer sources. Finally, we show that large sources can be effectively characterized using low-cost sensors, a valuable outcome informing how networks for monitoring larger areas could be implemented. This work may provide a basis for source identification and monitoring protocols for networks that feature sensors influenced by hyper-local point sources, subject to site-specific assumptions.
54 ENVIRONMENTAL SCIENCES↗
Improving the Capabilities and Computational Efficiency of the RTE+RRTMGP Radiation Code (Final Report)
This report details progress on the RTE+RRTMGP radiation codes made during the period of performance. RTE+RRTMGP is a set of codes for computing radiative fluxes in planetary atmospheres. RRTMGP uses a k-distribution to provide an optical description (absorption and possibly Rayleigh optical depth) of the gaseous atmosphere, along with the relevant source functions, on a pre-determined spectral grid given temperatures, pressures, and gas concentration. RTE computes fluxes given spectrally-resolved optical descriptions and source functions. Spectrally-resolved fluxes are summarized (“reduced”) via a user extensible class. The initial release of the code and the design choices are described in Pincus et al. 2019; the codes are available on Github. Although RRTMGP was based on current (at the time) empirical spectroscopic data, RTE and RRTMGP were developed in large part to modernize software practices. The design focused on flexibility broadly interpreted: by separating code from data and allowing data to drive computation; in coupling to the host model (e.g. the coupling of clouds to radiative fluxes is user-controlled); with respect to programming languages (computational tasks are accessed via widely-compatible C interfaces); and with respect to hardware (the codes run on a range of CPU and GPU architectures). The code also puts an emphasis on modularity and clarity. RTE+RRTMGP v1.0 was released in September 20219. This award supported the evolution of the RTE+RRTMGP code base to support greater flexibility, accuracy, and efficiency.
54 ENVIRONMENTAL SCIENCES↗
FY25 Report on Water NSTF Testing: Parametric and Accident Testing with Lower Tank Inlet
The Natural Convection Shutdown Heat Removal Test Facility (NSTF) at Argonne National Laboratory has continued to generate empirical validation data on the performance of water-based reactor cavity cooling system (RCCS) for seven years. This data is actively being used to support the development of passive decay heat removal systems for advanced reactors. Distinguishing this facility are 1) the large, ½ scale of the facility and 2) its governance under an NQA-1 qualified program for producing data of the highest pedigree for advanced reactor designers and regulators. In addition to the experimental activities discussed in this report, a computational modeling program continues to support the experimental program and further accuracy and understanding of the computational models. Together, the experimental and computational work create a mutually beneficial relationship integral to the overall program objective of advancing the understanding of the RCCS technology.
21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
Least-cost Optimal Distribution Grid Expansion (LODGE): Utility Pilots
The Least-cost Optimal Distribution Grid Expansion (LODGE) model provides the optimal portfolio of distribution system upgrades—e.g., voltage regulators, feeder reconductoring, transformer upgrades and non-wires alternatives (NWA), such as strategic siting of storage and distributed generation—to interconnect distributed energy resources (DERs) and enable load growth. It can be used to assess grid infrastructure costs and explore policy and regulatory solutions for distribution planning and DER valuation.In 2025, Berkeley Lab conducted three pilot analyses to validate LODGE results with empirical utility data before the model’s first release in 2026. The pilots, done with utilities in Washington, Colorado, and New Mexico, provide examples that illustrate how the model works, what it can do, and the value of the analysis.
Heleno, Miguel↗
Comparing Outdoor to Indoor Performance for Bifacial Modules Affected by Polarization-Type Potential-Induced Degradation
Bifacial photovoltaic (PV) modules have the advantage of using light reflected off of the ground to contribute to power production. Predicting the energy gain is challenging and requires complex models to do so accurately. Often, module degradation over time is neglected in models for the sake of simplicity or is underestimated. Comparing outdoor and indoor current–voltage (I–V) performance for bifacial modules is more challenging than for monofacial modules, as there are additional variables to consider such as rear albedo non-uniformity, cell mismatch, and their effects on temperature. This challenge is compounded when heterogeneous degradation modes occur, such as polarization-type potential-induced degradation (PID-p). To examine the effects of PID-p on I–V predictions using an empirical data-driven approach, 16 bifacial PERC modules are installed outdoors on racks with different albedo conditions. A subset is exposed to high-voltage biases of −1500 V or +1500 V. Outdoor data are traced at irradiance ranges of 150–250 W/m 2 , 500–600 W/m 2 , and 900–1000 W/m 2 . These curves are corrected using control module temperature, wire resistivity, and module resistance measured indoors. We examine several methods to transform indoor I–V curves to accurately, and more simply than existing methods, approximate outdoor performance for bifacial modules without and with varying levels of PID-p degradation. This way, bifacial performance modeling can be more accessible and informed by fielded, degraded modules. Distributions of percent errors between indoor and outdoor performance parameters and Mean Absolute Percent Errors (MAPEs) are used to assess method quality. Results including low-irradiance data (150–250 W/m 2 ) are discussed but are filtered for quantifying method quality as these data introduce substantial errors. The method with the most optimal tradeoff between low MAPE and analysis simplicity involves measuring the front side of a module indoors at an irradiance equal to plane-of-array irradiance plus the product of module bifaciality and albedo irradiance. This method gives MAPE values of 1–6.5% for non-degraded and 1.6–5.9% for PID-p degraded module performance.
14 SOLAR ENERGY↗
Characterizing Signatures of Geothermal Exploration Data with Machine Learning Techniques: An Application to the Nevada Play Fairway Analysis
We are introducing machine learning methods to the play fairway analysis to generate geothermal potential maps to support the evaluation of geothermal resource potential and the exploration for undiscovered blind geothermal systems in the Nevada Great Basin region. Our project aims to identify new ways to combine the play fairway data and empirically organize relationships between feature weights and labels in an improved workflow. As a means of doing this, we introduce machine learning methods to evaluate the influence of certain geological and geophysical features/feature sets in predicting geothermal favorability. This report highlights promising approaches based on supervised and unsupervised learning methods. First, we demonstrate a filter method applied to supervised classification modeling. The supervised filter method is based on permutation analysis to evaluate every possible feature combination/drop out scenario and rank feature influence based on the performance variance of supervised classification models. Additionally, we present an unsupervised factor analysis based on principal component analysis coupled with a semi-supervised kmeans clustering algorithm. This analysis allows us to identify the optimal number of groups/clusters for training sites and structural settings to identify feature patterns including correlation, variance, and latent and dominant feature relationships. The results from these methods offer a promising avenue for identifying favorable sources of predictive information to identify the locations of blind geothermal systems and furthering our understanding of complex geothermal feature and label relationships in the Great Basin region and beyond.
15 GEOTHERMAL ENERGY↗
H2Fills™ [SWR-19-67]
H2FillS: Hydrogen Filling Simulation The Hydrogen Filling Simulation (H2FillS) software is a thermodynamic model designed to track and report on the transient change in hydrogen temperature, pressure, and mass flow when filling a fuel cell electric vehicle (FCEV). H2FillS simulates gas flow from the hydrogen station to the FCEV storage system. Using empirical fueling data sets, the model has been validated over a range of fueling conditions to match common light-duty FCEV fill profiles. Overall, it provides significant benefits to the light-duty fueling market and fill knowledge gaps of the interaction between a hydrogen station and an FCEV. Capabilities Use the comprehensive H2FillS model to: Safely design and operate a hydrogen fueling station Support code refinement with readily available data Develop system and operational improvements to reduce capital or operating costs at hydrogen stations or on-board FCEVs. How It Works H2FillS uses a "drag-and-drop" graphical user interface to simulate station and vehicle systems with preset parameters for common hydrogen station components as default values. When needed, users can define their own parameters to create their own station or vehicle components. Download H2FillS in Two Steps Read the terms of the end user software license agreement. Register to download the software. There are two versions of the model: A full-station model starts the simulation at high-pressure ground storage, runs through a dispenser, and ends at a vehicle storage system. A partial-station model starts at the dispenser breakaway and consists solely of the dispenser components and the vehicle storage system. H2FillS will automatically output fill performance data from the vehicle by tracking pressure and temperature throughout the fill. Users can input their own fill profiles into the model to run a variety of simulations.
Peters, Michael↗
Improving precision and accuracy of genetic mapping with genotyping‐by‐sequencing data in outcrossing species
Abstract Genotyping‐by‐sequencing (GBS) is a widely used strategy for obtaining large numbers of genetic markers in model and non‐model organisms. In crop plants, GBS‐derived marker datasets are frequently used to perform quantitative trait locus (QTL) mapping. In some plant species, however, high heterozygosity and complex genome structure mean that researchers must use care in handling GBS data to conduct QTL mapping most effectively. Such outbred crops include most of the perennial grass and tree species used for bioenergy. To identify strategies for increasing accuracy and precision of QTL mapping using GBS data in outbred crops, we conducted an empirical study of SNP‐calling and genetic map‐building pipeline parameters in a Miscanthus sinensis population, and a complementary simulation study to estimate the relationship between genome‐wide error rate, read depth, and marker number. The bioenergy grass Miscanthus is an obligate outcrossing species with a recent (diploidized) whole‐genome duplication. For the study of empirical M. sinensis data, we compared two SNP‐calling methods (one non‐reference‐based and one reference‐based), a series of depth filters (12×, 20×, 30×, and 40×) and two map‐construction methods (i.e., marker ordering: linkage‐only and order‐corrected based on a reference genome). We found that correcting the order of markers on a linkage map by using a high‐quality reference genome improved QTL precision (shorter confidence intervals). For typical GBS datasets of between 1000 and 5000 markers to build a genetic map for biparental populations, a depth filter set at 30× to 40× applied to outbred populations provided a genome‐wide genotype‐calling error rate of less than 1%, improved accuracy of QTL point estimates and minimized type I errors for identifying QTL. Based on these results, we recommend using a reference genome to correct the marker order of genetic maps and a robust genotype depth filter to improve QTL mapping for outbred crops.
59 BASIC BIOLOGICAL SCIENCES↗
Learning curves for drug response prediction in cancer cell lines
Motivated by the size and availability of cell line drug sensitivity data, researchers have been developing machine learning (ML) models for predicting drug response to advance cancer treatment. As drug sensitivity studies continue generating drug response data, a common question is whether the generalization performance of existing prediction models can be further improved with more training data. We utilize empirical learning curves for evaluating and comparing the data scaling properties of two neural networks (NNs) and two gradient boosting decision tree (GBDT) models trained on four cell line drug screening datasets. The learning curves are accurately fitted to a power law model, providing a framework for assessing the data scaling behavior of these models. The curves demonstrate that no single model dominates in terms of prediction performance across all datasets and training sizes, thus suggesting that the actual shape of these curves depends on the unique pair of an ML model and a dataset. The multi-input NN (mNN), in which gene expressions of cancer cells and molecular drug descriptors are input into separate subnetworks, outperforms a single-input NN (sNN), where the cell and drug features are concatenated for the input layer. In contrast, a GBDT with hyperparameter tuning exhibits superior performance as compared with both NNs at the lower range of training set sizes for two of the tested datasets, whereas the mNN consistently performs better at the higher range of training sizes. Moreover, the trajectory of the curves suggests that increasing the sample size is expected to further improve prediction scores of both NNs. These observations demonstrate the benefit of using learning curves to evaluate prediction models, providing a broader perspective on the overall data scaling characteristics. A fitted power law learning curve provides a forward-looking metric for analyzing prediction performance and can serve as a co-design tool to guide experimental biologists and computational scientists in the design of future experiments in prospective research studies.
60 APPLIED LIFE SCIENCES↗
Improving Residential Building Simulations Through Large-Scale Empirical Validation
Residential building energy simulations are increasingly used for energy-efficient building design, codes and standards analysis, home certifications and ratings, utility programs, and technology assessments. Various software tools exist to perform residential building simulations, and these tools often use different models, inputs, and assumptions. This leads to inconsistencies that can undermine confidence in the predicted results. Validation of these tools can increase confidence by ensuring their accuracy and consistency. One way to validate simulation tools is through empirical testing, which compares predicted energy usage to measured utility billing data. This paper describes the process of data collection, data standardization, and empirical validation, and illustrates its use with our residential EnergyPlus (R)-based software. The data and process can be extended to other simulation tools and contribute to improving residential building simulations more broadly.
empirical validation↗
An Agnostic Approach to Building Empirical Type Ia Supernova Light Curves: Evidence for Intrinsic Chromatic Flux Variation Using Nearby Supernova Factory Data
We present a new empirical Type Ia supernova (SN Ia) model with three chromatic flux variation templates: one phase dependent and two phase independent. No underlying dust extinction model or patterns of intrinsic variability are assumed. Implemented with Stan and trained using spectrally binned Nearby Supernova Factory spectrophotometry, we examine this model's 2D, phase-independent flux variation space using two motivated basis representations. In both, the first phase-independent template captures variation that appears dust-like, while the second captures a combination of effectively intrinsic variability and second-order dust-like effects. We find that ≈13% of the modeled phase-independent flux variance is not dust-like. Previous empirical SN Ia models either assume an effective dust extinction recipe in their architecture, or only allow for a single mode of phase-independent variation. The presented results demonstrate such an approach may be insufficient, because it could "leak" noticeable intrinsic variation into phase-independent templates.
79 ASTRONOMY AND ASTROPHYSICS↗
HydraGNN_Predictive_GFM_2026 - Ensemble of predictive graph foundation models for atomistic materials modeling
This release contains data and parameters of HydraGNN-based graph foundation models trained as a result of the work published in the pre-print "Exascale Multi-Task Graph Foundation Models for Imbalanced, Multi-Fidelity Atomistic Data" by M. Lupo Pasini et al. (https://arxiv.org/abs/2604.15380). We jointly train on 16 open first-principles datasets (544+ million structures covering 85+ elements) using a multi-task architecture with per-dataset heads and a scalable ADIOS2/DDStore data pipeline. On Frontier, we execute six large-scale DeepHyper hyperparameter optimization campaigns in FP64 and promote the top-performing message-passing models to sustained 2,048-node training, yielding a PaiNN-based lead model. The version of HydraGNN used to generate the outputs provided in this release is HydraGNN v5.0 (https://github.com/ORNL/HydraGNN/releases/tag/v5.0) The list of datasets used for the training of the graph foundation model is the following: 1) Alexandria [1] 2) ANI1x [2] 3) MPTrj [3] 4) Open Catalyst 2020 (OC20) [4] 5) Open Catalyst 2022 (OC22) [5] 6) Open Catalyst 2025 (OC25) [6] 7) Open Direct ir Capture 2023 (ODAC23) [7] 8) Open Materials 2024 (OMat24) [8] 9) Open Molecules 2025 (OMol25) [9] 10) OMol25-neutral (subset of OMol25 that contains only molecules with zero total charge) 11) OMol25-non-neutral (subset of OMol25 that contains only molecules with non-zero total charge) 12) Open Polymers 2026 (OPoly2026) [10] 13) Nabla2DFT [11] 14) QCML [12] 15) QM7X [reference 13] 16) transition1x [14] Dataset references: [1] J. Schmidt et al., “A dataset of 175k stable and metastable materials calculated with the PBEsol and SCAN functionals,” Scientific Data, vol. 9, p. 64, 2022. [2] J. S. Smith et al., “The ANI-1ccx and ANI-1x data sets, coupled-cluster and density functional theory properties for molecules,” Scientific Data, vol. 7, p. 134, 2020. [Online]. Available: https: //www.nature.com/articles/s41597-020-0473-z [3] A. Jain et al., “Commentary: The Materials Project: A materials genome approach to accelerating materials innovation,” APL Materials, vol. 1, no. 1, p. 011002, 07 2013. [Online]. Available: https://doi.org/10.1063/1.4812323 [4] L. Chanussot et al., “Open catalyst 2020 (oc20) dataset and community challenges,” ACS Catalysis, vol. 11, no. 10, pp. 6059–6072, 2021. [Online]. Available: https://doi.org/10.1021/acscatal.0c04525 [5] K. Tran et al., “Open catalyst 2022 (oc22) dataset and challenges for oxidation electrocatalysts,” ACS Catalysis, vol. 13, no. 5, pp. 3066–3084, 2023. [Online]. Available: https://doi.org/10.1021/acscatal.2c05426 [6] S. J. Sahoo et al., “The open catalyst 2025 (oc25) dataset and models for solid-liquid interfaces,” arXiv preprint arXiv:2509.17862, 2025. [Online]. Available: https://arxiv.org/abs/2509.17862 [7] A. Sriram et al., “The open DAC 2023 dataset and challenges for sorbent discovery in direct air capture,” ACS Central Science, vol. 10, no. 5, pp. 923–941, 2024. [8] L. Barroso-Luque et al., “Open materials 2024 (omat24) inorganic materials dataset and models,” 2024. [Online]. Available: https://arxiv.org/abs/2410.12771 [9] D. S. Levine et al., “The open molecules 2025 (OMol25) dataset, evaluations, and models,” 2025. [Online]. Available: https://arxiv.org/abs/2505.08762 [10] D. S. Levine et al., The open polymers 2026 (OPoly26) dataset and evaluations,” arXiv preprint arXiv:2512.23117, 2025. [Online]. Available: https://arxiv.org/abs/2512.23117 [11] K. Khrabrov et al., “Nabla2dft: A universal quantum chemistry dataset of drug-like molecules and a benchmark for neural network potentials,” in NeurIPS 2024 Datasets and Benchmarks Track, 2024. [Online]. Available: https://openreview.net/forum?id=ElUrNM9U8c [12] S. Ganscha et al., “The QCML dataset, quantum chemistry reference data from 33.5M DFT and 14.7B semi-empirical calculations,” Scientific Data, vol. 12, p. 406, 2025. [13] J. Hoja et al., “QM7-X, a comprehensive dataset of quantum-mechanical properties spanning the chemical space of small organic molecules,” Scientific Data, vol. 8, p. 43, 2021. [Online]. Available: https://www.nature.com/articles/s41597-021-00812-2 [14] M. Schreiner et al., “Transition1x - a dataset for building generalizable reactive machine learning potentials,” Scientific Data, vol. 9, p. 779, 2022. The folder "datasets_ADIOS2_format" contains the set of pre-processed datasets in Adaptable I/O System (ADIOS) format (https://www.exascaleproject.org/research-project/adios/) that have been used for the development and training of GFMs in this work. The "datasets_ADIOS2_format" directory contains 2 sub-directories, one for the version "v1" of the datasets and one for the version "v2" of the datasets. The version "v1" of the datasets provides values of the total energy as they are extracted from the original data as it was released by the respective institutions. The version "v2" of the datasets provides values of the energy that have been realigned. The realignment was performed by training a linear regression model that predicts the total energy as a function of the chemical composition of the atomistic structure, and then subtract such prediction from the original value of the total energy. Both folders "v1" and "v2" contain 16 sub-directories, each corresponding to an ADIOS2-formatted dataset The folder "DeepHyper-results" contains the configurational files and model's parameters for all the 186 HPO trials that were successfully completed by the scalable hyperparameter optimization (HPO) runs on Frontier. The content of the folder "DeepHyper-results" I structured as follows: 1) task-list.txt: list of mpnn name, jobid, and deephyper task id 2) gfm_${MPNN}_${JOBID}_0.${TASKID}: run directory with checkpoint files 3) gfm_${MPNN}: deephyper summary directory (*.csv) for each specific MPNN type 4) deephyper-experiment-${JOBID}: output and error logs for each job The file "deephyper-sorted.csv" contains the details of each HydraGNN model built and tested by HPO, obtained by merging the (*.csv) filed from each HPO run executed. Out of all the HPO trials, we selected 10 to continue the training of the respective HydraGNN models. Due to limited computational budget available in the LRN070 allocation we could not complete the training till convergence for all these 10 selected models. The folder "models" contains multiple sub-folders, one per each HydraGNN model trained. Each model sub-folder contains the parameters of each HydraGNN model, with multiple checkpoint-restarts. The list of sub-folders are as follows: 1) multidataset_hpo-BEST1-fp64 2) multidataset_hpo-BEST2-fp64 3) multidataset_hpo-BEST3-fp64 4) multidataset_hpo-BEST4-fp64 5) multidataset_hpo-BEST5-fp64 6) multidataset_hpo-BEST6-fp64 7) multidataset_hpo-BEST7-fp64 8) multidataset_hpo-BEST8-fp64 9) multidataset_hpo-BEST9-fp64 10) multidataset_hpo-BEST10-fp64 Within each one of these folders, additional auxiliary log files are provided with descriptions about how the training proceeded. The lead PaiNN-model is contained inside "multidataset_hpo-BEST6-fp64". The file "mlp_branch_weights" contains the parameters of the multi-layer perceptron (MLP) used to reconcile the predictions of the 16 output decoding heads of the HydragNN architectures. The MLP takes in input the chemical composition of the atomistic structure and predicts averaging weights to linearly mix the predictions of each output decoding head toward consolidating them into a single one. The folder "1.1billion-structure-inference" contains 1.1 billion atomistic structures randomly generated. Each structures is associated with energy and forces predicted with the lead-PaiNN model combined with the MLP model for reconciliation of the multi-branch predictions generated by the 16 output decoding heads. The folder "1.1billion-structure-inference" contains 9,300 (*.tar.gz) subdirectories, one per Frontier compute node used to execute the inference at exascale. Once uncompressed, each (*.tar.gz) subdirectory contains an ADIOS2 (*.bp) file container, where each atomistic structure is stored as a PyTorch-Geometric Data object. The file "export_dataset_environment_variables.sh" contains the environment variables that need to be set before running the HydraGNN code to reproduce the results provided in this dataset release. The code that can be used to load the ADIOS2 files, load HydraGNN models, and run inference is available at: https://github.com/ORNL/HydraGNN/releases/tag/v5.0
36 MATERIALS SCIENCE↗