Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “graph clustering”

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 379 records · Page 21

A CATALOG OF GLOBULAR CLUSTER SYSTEMS: WHAT DETERMINES THE SIZE OF A GALAXY'S GLOBULAR CLUSTER POPULATION?

We present a catalog of 422 galaxies with published measurements of their globular cluster (GC) populations. Of these, 248 are E galaxies, 93 are S0 galaxies, and 81 are spirals or irregulars. Among various correlations of the total number of GCs with other global galaxy properties, we find that N{sub GC} correlates well though nonlinearly with the dynamical mass of the galaxy bulge M{sub dyn}= 4{sigma}{sub e}{sup 2} R{sub e} /G, where {sigma}{sub e} is the central velocity dispersion and R{sub e} the effective radius of the galaxy light profile. We also present updated versions of the GC specific frequency S{sub N} and specific mass S{sub M} versus host galaxy luminosity and baryonic mass. These graphs exhibit the previously known U-shape: highest S{sub N} or S{sub M} values occur for either dwarfs or supergiants, but in the midrange of galaxy size (10{sup 9}-10{sup 10} L{sub Sun }) the GC numbers fall along a well-defined baseline value of S{sub N} {approx_equal} 1 or S{sub M} = 0.1, similar among all galaxy types. Along with other recent discussions, we suggest that this trend may represent the effects of feedback, which systematically inhibited early star formation at either very low or very high galaxy mass, but which had its minimum effect for intermediate masses. Our results strongly reinforce recent proposals that GC formation efficiency appears to be most nearly proportional to the galaxy halo mass M{sub halo}. The mean 'absolute' efficiency ratio for GC formation that we derive from the catalog data is M{sub GCS}/M{sub halo} = 6 Multiplication-Sign 10{sup -5}. We suggest that the galaxy-to-galaxy scatter around this mean value may arise in part because of differences in the relative timing of GC formation versus field-star formation. Finally, we find that an excellent empirical predictor of total GC population for galaxies of all luminosities is N{sub GC} {approx} (R{sub e} {sigma}{sub e}){sup 1.3}, a result consistent with fundamental plane scaling relations.

79 ASTRONOMY AND ASTROPHYSICS↗

Effect of realistic routing on the social burden metric

The distance people travel to reach critical services is a key input to the Social Burden metric used by Sandia’s Resilient Node Cluster Analysis Tool (ReNCAT) in the optimization’s objective function. By default, ReNCAT utilizes Euclidian distances between population blocks and critical facilities when calculating Social Burden. However, these straight-line distances do not reflect how most residents or goods would travel throughout the area. As distance is a vital input to the burden calculation, a more realistic distance calculation will yield more realistic burden values. This work uses real road networks and calculates the shortest distance path between population centers and critical facilities using a standard graph theory approach. These realistic route distances are then used to compute Social Burden for four areas of study. It was found that distances using real road routes are generally, but not always, longer than the Euclidean distance. The increased length increases the final Social Burden metric, however, the overall burden percent change ranged between 17% and 52%, which means the impact of realistic routes relies heavily upon the area’s road topology. It was found that rural locations within an area may have larger burden increases than urban areas as more dense road networks allow routes to more closely follow a straight-line path. Additionally, using the most straight forward routing algorithms requires high computational effort for areas with large road networks. While it is believed this process can be made more performant, that task is beyond this scope of work.

99 GENERAL AND MISCELLANEOUS↗

Harnessing distributed GPU computing for generalizable graph convolutional networks in power grid reliability assessments

Although machine learning (ML) has emerged as a powerful tool for rapidly assessing grid contingencies, prior studies have largely considered a static grid topology in their analyses. This limits their application, since they need to be re-trained for every new topology. Here, this paper explores the development of generalizable graph convolutional network (GCN) models by pre-training them across a range of grid topologies and contingency types. We found that a GCN model with auto-regressive moving average (ARMA) layers with a line graph representation of the grid offered the best predictive performance in predicting voltage magnitudes (VM) and voltage angles (VA). We introduced the concept of phantom nodes to consider disparate grid topologies with a varying number of nodes and lines. For pre-training the GCN ARMA model across a variety of topologies, distributed graphics processing unit (GPU) computing afforded us significant training scalability. The predictive performance of this model on grid topologies that were part of the training data is substantially better than the direct current (DC) approximation. Although direct application of the pre-trained model to topologies that are not part of the grid is not particularly satisfactory, fine-tuning with small amounts of data from a specific topology of interest significantly improves predictive performance. In general, this paper highlights the feasibility of training large-scale GNN models to assess the reliability of power grids by considering a wide variety of grid topologies and contingency types. With the advent of foundational models in ML and the exponential increase in GPU computing clusters, generalizable ML models will significantly enhance how utilities manage power systems and make decisions in real-time or near-real-time.

24 - POWER TRANSMISSION AND DISTRIBUTION↗

Identifying the Unique Properties of α-Bi 2 Mo 3 O 12 for the Activation of Propene

In order to understand the remarkable activity of α-Bi 2 Mo 3 O 12 for selective oxidation and ammoxidation of propene, the propene activation ability of four molybdenum-based mixed metal oxides - Bi 2 Mo 3 O 12 , PbMoO 4 , Bi 2 Pb 5 Mo 8 O 32 , and MoO 3 - was investigated using density functional theory. Propene activation is considered to occur via abstraction of a hydrogen atom from the methyl group of physisorbed propene by lattice oxygen. For each material, the apparent activation energy was estimated by summing the heat of adsorption of propene, the C-H bond dissociation energy, and the hydrogen attachment energy (HAE) for hydrogen addition to lattice oxygen; this sum provides a lower bound for the apparent activation energy. It was found that two structural features of oxide surfaces are essential to achieve low activation barriers: under-coordinated surface cation sites enable strong propene adsorption, and suitable 5- or 6-coordinate geometries at molybdenum result in favorable HAEs. The impact of molybdenum coordination on HAE was elucidated by carrying out a molecular orbital analysis using a cluster model of the molybdate unit. This effort revealed that, in 5- and 6-coordinate molybdates, oxygen donor atoms trans to molybdenyl oxo atoms destabilize the molybdate prior to H addition but stabilize the molybdate after H addition, thereby providing an HAE ~15 kcal/mol more favorable than that on 4-coordinate molybdate oxo atoms. Bi 3+ cations in Bi 2 Mo 3 O 12 thus promote catalytic activity by providing both strong adsorption sites for propene and forcing molybdate into 5-coordinate geometries that lead to particularly favorable values of the HAE. (Graph Presented).

36 MATERIALS SCIENCE↗

A scalable approach to solving dense linear algebra problems on hybrid CPU-GPU systems

Aiming to fully exploit the computing power of all CPUs and all graphics processing units (GPUs) on hybrid CPU-GPU systems to solve dense linear algebra problems, in this paper we design a class of heterogeneous tile algorithms to maximize the degree of parallelism, to minimize the communication volume, and to accommodate the heterogeneity between CPUs and GPUs. The new heterogeneous tile algorithms are executed upon our decentralized dynamic scheduling runtime system, which schedules a task graph dynamically and transfers data between compute nodes automatically. The runtime system uses a new distributed task assignment protocol to solve data dependencies between tasks without any coordination between processing units. By overlapping computation and communication through dynamic scheduling, we are able to attain scalable performance for the double-precision Cholesky factorization and QR factorization. Finally, our approach demonstrates a performance comparable to Intel MKL on shared-memory multicore systems and better performance than both vendor (e.g., Intel MKL) and open source libraries (e.g., StarPU) in the following three environments: heterogeneous clusters with GPUs, conventional clusters without GPUs, and shared-memory systems with multiple GPUs.

97 MATHEMATICS AND COMPUTING↗

Machine Learning Approaches for Nuclear Material Accounting Data from Irradiation and Reprocessing

We are currently exploring data analysis methods for their ability to strengthen the synthesis and evaluation of information generated within domestic and international safeguard regimes. Safeguard data typically includes rich heterogenous datasets amenable to advanced data analytics methods, such as machine learning. We have converted transactional data containing both numerical and categorical attributes from a domestic nuclear material control and accountability (NMC&A) system into a low-dimensional numerical structure via linear principal component analysis. This data representation allows for global structure discovery via cluster analysis, which can characterize the typical behavior of each of the primary types of transaction events. Furthermore, the structure of the top principal components captures the “typical behavior” of the data and is thus amenable to anomaly detection through statistical hypothesis testing. We explored this capability by generating erroneous permutations of the data and computing the Q-residual quantity increase associated with the information loss when these data are projected into the low-dimensional principal component analysis space representative of typical transactions.Future work will focus on identifying data transformations (e.g., graph networks) that more closely align with the inherent structure of these transactions to explain more salient information and disambiguate the underlying nuclear process—in this case, irradiation and reprocessing—from artifacts of NMC&A system transactional record keeping data.

Drescher, Adam↗

STOCHASTIC MODEL OF THE SPIN DISTRIBUTION OF DARK MATTER HALOS

We employ a stochastic approach to probing the origin of the log-normal distributions of halo spin in N-body simulations. After analyzing spin evolution in halo merging trees, it was found that a spin change can be characterized by a stochastic random walk of angular momentum. Also, spin distributions generated by random walks are fairly consistent with those directly obtained from N-body simulations. We derived a stochastic differential equation from a widely used spin definition and measured the probability distributions of the derived angular momentum change from a massive set of halo merging trees. The roles of major merging and accretion are also statistically analyzed in evolving spin distributions. Several factors (local environment, halo mass, merging mass ratio, and redshift) are found to influence the angular momentum change. The spin distributions generated in the mean-field or void regions tend to shift slightly to a higher spin value compared with simulated spin distributions, which seems to be caused by the correlated random walks. We verified the assumption of randomness in the angular momentum change observed in the N-body simulation and detected several degrees of correlation between walks, which may provide a clue for the discrepancies between the simulated and generated spin distributions in the voids. However, the generated spin distributions in the group and cluster regions successfully match the simulated spin distribution. We also demonstrated that the log-normality of the spin distribution is a natural consequence of the stochastic differential equation of the halo spin, which is well described by the Geometric Brownian Motion model.

79 ASTRONOMY AND ASTROPHYSICS↗

State-based network similarity visualization

We introduce an approach for the interactive visual analysis of weighted, dynamic networks. These networks arise in areas such as computational neuroscience, sociology, and biology. Network analysis remains challenging due to complex time-varying network behavior. For example, edges disappear/reappear, communities grow/vanish, or overall network topology changes. Our technique, TimeSum, detects the important topological changes in graph data to abstract the dynamic network and visualize one summary representation for each temporal phase, a state. We define a network state as a graph with similar topology over a specific time interval. To enable a holistic comparison of networks, we use a difference network to depict edge and community changes. Here, we present case studies to demonstrate that our methods are effective and useful for extracting and exploring complex dynamic behavior of networks.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

Navigating Large Chemical Spaces Using Graph Theory and Integer Programming

Navigating and analyzing large chemical spaces are necessary to accelerate the design and discovery of new molecules and chemical processes. In this work, we introduce a computational framework that integrates graph theory and integer programming to enable the efficient navigation of large chemical spaces. Our framework represents the chemical space as a graph, wherein nodes represent molecules and edges represent the degree of similarity or connectivity based on domain-specific information. Using the graph representation, we identify representative molecules by computing the so-called minimum dominating set (MDS), which in our context is the minimum set of molecules that is connected to all other molecules. We present a suite of solution strategies for the MDS problem including heuristic and rigorous integer programming (IP) approaches. We show that these approaches allow us to capture physicochemical properties and domain-specific logic and constraints, facilitating the identification of molecules with the target properties. We demonstrate the effectiveness of the proposed approach by navigating the chemical space of per- and polyfluoroalkyl substances (PFAS); this comprises approximately 15,000 molecular structures. We compare our framework against traditional dimensionality reduction and clustering methods such as t-SNE and K-means clustering.

Chemical structure↗

pGraph: Efficient Parallel Construction of Large-Scale Protein Sequence Homology Graphs

Detecting sequence homology between protein sequences is a fundamental problem in computational molecular biology, with a pervasive application in nearly all analyses that aim to structurally and functionally characterize protein molecules. While detecting the homology between two protein sequences is relatively inexpensive, detecting pairwise homology for a large number of protein sequences can become computationally prohibitive for modern inputs, often requiring millions of CPU hours. Yet, there is currently no robust support to parallelize this kernel. In this paper, we identify the key characteristics that make this problemparticularly hard to parallelize, and then propose a new parallel algorithm that is suited for detecting homology on large data sets using distributed memory parallel computers. Our method, called pGraph, is a novel hybrid between the hierarchical multiple-master/worker model and producer-consumer model, and is designed to break the irregularities imposed by alignment computation and work generation. Experimental results show that pGraph achieves linear scaling on a 2,048 processor distributed memory cluster for a wide range of inputs ranging from as small as 20,000 sequences to 2,560,000 sequences. In addition to demonstrating strong scaling, we present an extensive report on the performance of the various system components and related parametric studies.

Environmental Molecular Sciences Laboratory↗

Causal interaction in high frequency turbulence at the biosphere–atmosphere interface: Structure–function coupling

At the biosphere–atmosphere interface, nonlinear interdependencies among components of an ecohydrological complex system can be inferred using multivariate high frequency time series observations. Information flow among these interacting variables allows us to represent the causal dependencies in the form of a directed acyclic graph (DAG). Here, we use high frequency multivariate data at 10 Hz from an eddy covariance instrument located at 25 m above agricultural land in the Midwestern US to quantify the evolutionary dynamics of this complex system using a sequence of DAGs by examining the structural dependency of information flow and the associated functional response. We investigate whether functional differences correspond to structural differences or if there are no functional variations despite the structural differences. We base our analysis on the hypothesis that causal dependencies are instigated through information flow, and the resulting interactions sustain the dynamics and its functionality. To test our hypothesis, we build upon causal structure analysis in the companion paper to characterize the information flow in similarly clustered DAGs from 3-min non-overlapping contiguous windows in the observational data. We characterize functionality as the nature of interactions as discerned through redundant, unique, and synergistic components of information flow. Through this analysis, we find that in turbulence at the biosphere–atmosphere interface, the variables that control the dynamic character of the atmosphere as well as the thermodynamics are driven by non-local conditions, while the scalar transport associated with CO and H 2 O is mainly driven by short-term local conditions.

58 GEOSCIENCES↗

Performance optimization, modeling and analysis of sparse matrix-matrix products on multi-core and many-core processors

Sparse matrix-matrix multiplication (SpGEMM) is a computational primitive that is widely used in areas ranging from traditional numerical applications to recent big data analysis and machine learning. Although many SpGEMM algorithms have been proposed, hardware specific optimizations for multi- and many-core processors are lacking and a detailed analysis of their performance under various use cases and matrices is not available. In this work, we firstly identify and mitigate multiple bottlenecks with memory management and thread scheduling on Intel Xeon Phi (Knights Landing or KNL). Specifically targeting many-core processors, we develop a hash-table-based algorithm and optimize a heap-based shared-memory SpGEMM algorithm. We examine their performance together with other publicly available codes. Different from the literature, our evaluation also includes use cases that are representative of real graph algorithms, such as multi-source breadth-first search or triangle counting. Our hash-table and heap-based algorithms are showing significant speedups from libraries in the majority of the cases while different algorithms dominate the other scenarios with different matrix size, sparsity, compression factor and operation type. We wrap up in-depth evaluation results and make a recipe to give the best SpGEMM algorithm for target scenario. We build the performance model for hash-table and heap-based algorithms, which supports the recipe. A critical finding is that hash-table-based SpGEMM gets a significant performance boost if the nonzeros are not required to be sorted within each row of the output matrix. Finally, we integrate our implementations into a large-scale protein clustering code named HipMCL, accelerating its SpGEMM kernel by up to 10X and achieving an overall performance boost for the whole HipMCL application by 2.6X.

97 MATHEMATICS AND COMPUTING↗

Primitive numerical simulation of circular Couette flow - Carrousel wind tunnel nonturbulent solutions

The azimuthal-invariant, three-dimensional cylindrical, incompressible Navier-Stokes equations are solved to steady state for a finite-length, physically realistic model. The numerical method relies on an alternating-direction implicit scheme that is formally second-order accurate in space and first-order accurate in time. The equations are linearized and uncoupled by evaluating variable coefficients at the previous time iteration. Wall grid clustering is provided by a Roberts transformation in radial and axial directions. A vorticity-velocity formulation is found to be preferable to a vorticity-streamfunction approach. Subject to no-slip, Dirichlet boundary conditions, except for the inner cylinder rotation velocity (impulsive start-up) and zero-flow initial conditions, nonturbulent solutions are obtained for sub- and supercritical Reynolds numbers of 100 to 400 for a finite geometry where R(outer)/R(inner) = 1.5, H/R(inner) = 0.73, and H/Delta-R = 1.5. An axially-stretched model solution is shown to asymptotically approach the one-dimensional analytic Couette solution at the cylinder midheight. Flowfield change from laminar to Taylor-vortex flow is discussed as a function of Reynolds number. Three-dimensional velocities, vorticity, and streamfunction are presented via two-dimensional graphs and three-dimensional surface and contour plots.

Hasiuk, Jan↗

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↗

Lattice model theory of the equation of state covering the gas, liquid, and solid phases

The three stable states of matter and the corresponding phase transitions were obtained with a single model. Patterned after Lennard-Jones and Devonshires's theory, a simple cubic lattice model containing two fcc sublattices (alpha and beta) is adopted. The interatomic potential is taken to be the Lennard-Jones (6-12) potential. Employing the cluster variation method, the Weiss and the pair approximations on the lattice gas failed to give the correct phase diagrams. Hybrid approximations were devised to describe the lattice term in the free energy. A lattice vibration term corresponding to a free volume correction is included semi-phenomenologically. The combinations of the lattice part and the free volume part yield the three states and the proper phase diagrams. To determine the coexistence regions, the equalities of the pressure and Gibbs free energy per molecule of the coexisting phases were utilized. The ordered branch of the free energy gives rise to the solid phase while the disordered branch yields the gas and liquid phases. It is observed that the triple point and the critical point quantities, the phase diagrams and the coexistence regions plotted are in good agreement with the experimental values and graphs for argon.

Bonavito, N. L.↗

Thermochemical Data Fusion Using Graph Representation Learning

Large databases are required for “Big Data” applications in catalysis and materials science. Thermochemical databases can be created by combining data from various sources and by correcting low-fidelity datasets to higher accuracy with minimal computation. To achieve this “data fusion”, thermochemical quantities of interest, calculated at various levels of density functional theory (DFT), need to be mapped to the same, high levels of theory. In this work, a graph theoretical, statistical framework is proposed for such tasks. Subgraph frequencies are shown to provide a natural representation for learning these fusion maps. The maps are linear and are learnt with automated descriptor selection. Using a dataset of as few as ~1% from the QM9 database of 133,885 molecules, these models can predict multiple thermochemical quantities at a higher level of theory with an accuracy of 1 kcal/mol. Here, the method is explainable, generalizable, and provides a diagnostic tool for outlier identification

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Considerations on the Use of Custom Accelerators for Big Data Analytics

Accelerators, including Graphic Processing Units (GPUs) for gen- eral purpose computation, many-core designs with wide vector units (e.g., Intel Phi), have become a common component of many high performance clusters. The appearance of more stable and reliable tools tools that can automatically convert code written in high-level specifications with annotations (such as C or C++) to hardware de- scription languages (High-Level Synthesis - HLS), is also setting the stage for a broader use of reconfigurable devices (e.g., Field Pro- grammable Gate Arrays - FPGAs) in high performance system for the implementation of custom accelerators, helped by the fact that new processors include advanced cache-coherent interconnects for these components. In this chapter, we briefly survey the status of the use of accelerators in high performance systems targeted at big data analytics applications. We argue that, although the progress in the use of accelerators for this class of applications has been sig- nificant, differently from scientific simulations there still are gaps to close. This is particularly true for the ”irregular” behaviors exhibited by no-SQL, graph databases. We focus our attention on the limits of HLS tools for data analytics and graph methods, and discuss a new architectural template that better fits the requirement of this class of applications. We validate the new architectural templates by mod- ifying the Graph Engine for Multithreaded System (GEMS) frame- work to support accelerators generated with such a methodology, and testing with queries coming from the Lehigh University Benchmark (LUBM). The architectural template enables better supporting the task and memory level parallelism present in graph methods by sup- porting a new control model and a enhanced memory interface. We show that out solution allows generating parallel accelerators, pro- viding speed ups with respect to conventional HLS flows. We finally draw conclusions and present a perspective on the use of reconfig- urable devices and Design Automation tools for data analytics.

Data Analytics↗

Hierarchical Spatio-temporal Visual Analysis of Cluster Evolution in Electrocorticography Data

Here, we present ECoG ClusterFlow, a novel interactive visual analysis tool for the exploration of high-resolution Electrocorticography (ECoG) data. Our system detects and visualizes dynamic high-level structures, such as communities, using the time-varying spatial connectivity network derived from the high-resolution ECoG data. ECoG ClusterFlow provides a multi-scale visualization of the spatio-temporal patterns underlying the time-varying communities using two views: 1) an overview summarizing the evolution of clusters over time and 2) a hierarchical glyph-based technique that uses data aggregation and small multiples techniques to visualize the propagation of clusters in their spatial domain. ECoG ClusterFlow makes it possible 1) to compare the spatio-temporal evolution patterns across various time intervals, 2) to compare the temporal information at varying levels of granularity, and 3) to investigate the evolution of spatial patterns without occluding the spatial context information. Lastly, we present case studies done in collaboration with neuroscientists on our team for both simulated and real epileptic seizure data aimed at evaluating the effectiveness of our approach.

60 APPLIED LIFE SCIENCES↗