Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Computational graph”

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 541 records · Page 30
All sourcesNASA NTRSDOE OSTI
NASA NTRS1990

Predicting performance of parallel computations

An accurate and computationally efficient method for predicting the performance of a class of parallel computations running on concurrent systems is described. A parallel computation is modeled as a task system with precedence relationships expressed as a series-parallel directed acyclic graph. Resources in a concurrent system are modeled as service centers in a queuing network model. Using these two models as inputs, the method outputs predictions of expected execution time of the parallel computation and the concurrent system utilization. The method is validated against both detailed simulation and actual execution on a commercial multiprocessor. Using 100 test cases, the average error of the prediction when compared to simulation statistics is 1.7 percent, with a standard deviation of 1.5 percent; the maximum error is about 10 percent.

Mak, Victor W.↗
DOE OSTI2024

Development of message passing-based graph convolutional networks for classifying cancer pathology reports

Abstract Background Applying graph convolutional networks (GCN) to the classification of free-form natural language texts leveraged by graph-of-words features (TextGCN) was studied and confirmed to be an effective means of describing complex natural language texts. However, the text classification models based on the TextGCN possess weaknesses in terms of memory consumption and model dissemination and distribution. In this paper, we present a fast message passing network (FastMPN), implementing a GCN with message passing architecture that provides versatility and flexibility by allowing trainable node embedding and edge weights, helping the GCN model find the better solution. We applied the FastMPN model to the task of clinical information extraction from cancer pathology reports, extracting the following six properties: main site, subsite, laterality, histology, behavior, and grade. Results We evaluated the clinical task performance of the FastMPN models in terms of micro- and macro-averaged F1 scores. A comparison was performed with the multi-task convolutional neural network (MT-CNN) model. Results show that the FastMPN model is equivalent to or better than the MT-CNN. Conclusions Our implementation revealed that our FastMPN model, which is based on the PyTorch platform, can train a large corpus (667,290 training samples) with 202,373 unique words in less than 3 minutes per epoch using one NVIDIA V100 hardware accelerator. Our experiments demonstrated that using this implementation, the clinical task performance scores of information extraction related to tumors from cancer pathology reports were highly competitive.

59 BASIC BIOLOGICAL SCIENCES↗
NASA NTRS1997

Sequential Testing Algorithms for Multiple Fault Diagnosis

In this paper, we consider the problem of constructing optimal and near-optimal test sequencing algorithms for multiple fault diagnosis. The computational complexity of solving the optimal multiple-fault isolation problem is super-exponential, that is, it is much more difficult than the single-fault isolation problem, which, by itself, is NP-hard. By employing concepts from information theory and AND/OR graph search, we present several test sequencing algorithms for the multiple fault isolation problem. These algorithms provide a trade-off between the degree of suboptimality and computational complexity. Furthermore, we present novel diagnostic strategies that generate a diagnostic directed graph (digraph), instead of a diagnostic tree, for multiple fault diagnosis. Using this approach, the storage complexity of the overall diagnostic strategy reduces substantially. The algorithms developed herein have been successfully applied to several real-world systems. Computational results indicate that the size of a multiple fault strategy is strictly related to the structure of the system.

Shakeri, Mojdeh↗
DOE OSTI2022

Scalable training of graph convolutional neural networks for fast and accurate predictions of HOMO-LUMO gap in molecules

Abstract Graph Convolutional Neural Network (GCNN) is a popular class of deep learning (DL) models in material science to predict material properties from the graph representation of molecular structures. Training an accurate and comprehensive GCNN surrogate for molecular design requires large-scale graph datasets and is usually a time-consuming process. Recent advances in GPUs and distributed computing open a path to reduce the computational cost for GCNN training effectively. However, efficient utilization of high performance computing (HPC) resources for training requires simultaneously optimizing large-scale data management and scalable stochastic batched optimization techniques. In this work, we focus on building GCNN models on HPC systems to predict material properties of millions of molecules. We use HydraGNN, our in-house library for large-scale GCNN training, leveraging distributed data parallelism in PyTorch. We use ADIOS, a high-performance data management framework for efficient storage and reading of large molecular graph data. We perform parallel training on two open-source large-scale graph datasets to build a GCNN predictor for an important quantum property known as the HOMO-LUMO gap. We measure the scalability, accuracy, and convergence of our approach on two DOE supercomputers: the Summit supercomputer at the Oak Ridge Leadership Computing Facility (OLCF) and the Perlmutter system at the National Energy Research Scientific Computing Center (NERSC). We present our experimental results with HydraGNN showing (i) reduction of data loading time up to 4.2 times compared with a conventional method and (ii) linear scaling performance for training up to 1024 GPUs on both Summit and Perlmutter.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

A Data Processing Pipeline for Adversarial Socio-Technical Network Analysis

With the rapid adoption of emerging technologies, there is a need to catalog and model sociotechnical interdependencies that have been historically used to influence the operation of Critical Infrastructure networks including the impacts of mergers and acquisitions, hostile takeovers, and foreign investment. Our research intends to address this need with two primary contributions. First, we have developed a data curation and processing pipeline to generate sociotechnical networks extracted from a variety of data sources including SEC filings and infrastructure asset databases. The pipeline, implemented in Apache Airflow, extracts and normalizes the representation of entities and relations, specified within ontologies. Our intent is to provide an extensible, machine-actionable approach to quickly communicate such models, reproduce previous results, and adapt them to new, unanticipated situations. Second, networks produced by our pipeline enable the development of graph-theoretic metrics that consider the properties of network components in addition to its topology. Metadata associated with network components---whether semantic, temporal, or geospatial---affects the alignment of generated networks with assumptions underlying complexity metrics. Validation of generated networks relative to component types defined by an ontology, may allow the research community to adapt metrics to the semantics of the domains being studied. Generated networks may be processed as knowledge, dynamic, or spatial graphs and enables a variety of analyses including automated reasoning and measures of network complexity. Automated reasoning views extracted entities and relations as a knowledge graph; this enables application of inference rules that represent historically-attested adversarial business methods and applies that behavior to a specific geographic context. Measures of network complexity, including degree distribution, reachability analyses, temporal analysis, and community detection can be adapted to indicate adversarial organizational influence.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2006

Distributed Computing Architecture for Image-Based Wavefront Sensing and 2 D FFTs

Image-based wavefront sensing (WFS) provides significant advantages over interferometric-based wavefi-ont sensors such as optical design simplicity and stability. However, the image-based approach is computational intensive, and therefore, specialized high-performance computing architectures are required in applications utilizing the image-based approach. The development and testing of these high-performance computing architectures are essential to such missions as James Webb Space Telescope (JWST), Terrestial Planet Finder-Coronagraph (TPF-C and CorSpec), and Spherical Primary Optical Telescope (SPOT). The development of these specialized computing architectures require numerous two-dimensional Fourier Transforms, which necessitate an all-to-all communication when applied on a distributed computational architecture. Several solutions for distributed computing are presented with an emphasis on a 64 Node cluster of DSPs, multiple DSP FPGAs, and an application of low-diameter graph theory. Timing results and performance analysis will be presented. The solutions offered could be applied to other all-to-all communication and scientifically computationally complex problems.

Smith, Jeffrey S.↗
DOE OSTI2021

Three practical workflow schedulers for easy maximum parallelism

Runtime scheduling and workflow systems are an increasingly popular algorithmic component in HPC because they allow full system utilization with relaxed synchronization requirements. There are so many special-purpose tools for task scheduling, one might wonder why more are needed. Use cases seen on the Summit supercomputer needed better integration with MPI and greater flexibility in job launch configurations. Preparation, execution, and analysis of computational chemistry simulations at the scale of tens of thousands of processors revealed three distinct workflow patterns. A separate job scheduler was implemented for each one using extremely simple and robust designs: file-based, task-list based, and bulk-synchronous. Comparing to existing methods shows unique benefits of this work, including simplicity of design, suitability for HPC centers, short startup time, and well-understood per-task overhead. All three new tools have been shown to scale to full utilization of Summit, and have been made publicly available with tests and documentation. This work presents a complete characterization of the minimum effective task granularity for efficient scheduler usage scenarios. Here, these schedulers have the same bottlenecks, and hence similar task granularities as those reported for existing tools following comparable paradigms.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

HydraGNN_OPF_GFM_2026 - Ensemble of predictive graph foundation models for power grid applications

This dataset supports research on graph foundation models for optimal power flow (OPF) on electric grids using HydraGNN. It contains heterogeneous graph representations of PGLib-OPF cases spanning systems from 14 to 13,659 buses, together with packed HDF5 datasets for pretraining, feasibility classification, and N-1 contingency analysis. The release includes OPF solution data, downstream fine-tuning datasets, pretrained HeteroSAGE and HeteroHEAT model checkpoints, hyperparameter-optimization summaries across multiple heterogeneous GNN architectures, and aggregated fine-tuning results for sample-efficiency studies. The dataset is designed to enable scalable training, evaluation, and transfer-learning studies for OPF surrogate modeling, including node-level AC-OPF solution prediction, graph-level prediction, feasibility classification, operating-condition generalization, and contingency-response tasks.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2021

RivGraph: Automatic extraction and analysis of river and delta channel network topology

River networks sustain life and landscapes by carrying and distributing water, sediment, and nutrients throughout ecosystems and communities. At the largest scale, river networks drain continents through tree-like tributary networks. At typically smaller scales, river deltas and braided rivers form loopy, complex distributary river networks via avulsions and bifurcations.In order to model flows through these networks or analyze network structure, the topology, or connectivity, of the network must be resolved. Additionally, morphologic properties of each river channel as well as the direction of flow through the channel inform how fluxes travel through the network’s channels. Riv Graphis a Python package that automates the extraction and characterization of river channel networks from a user-provided binary image, or mask, of a channel network (Fig. 1). Masks may be derived from (typically remotely-sensed) imagery, simulations, or even hand-drawn. RivGraph will create explicit representations of the channel network by resolving river centerlines as links, and junctions as nodes. Flow directions are solved for each link of the network without using auxiliary data, e.g., a digital elevation model (DEM). Morphologic properties are computed as well, including link lengths, widths, sinuosities, branching angles,and braiding indices. If provided,RivGraph will preserve georeferencing information of the mask and will export results as ESRI shapefiles, GeoJSONs, and GeoTIFFs for easy import into GIS software.RivGraph can also return extracted networks as networkx objects for convenient interfacing with the full-featured networkx package (Hagberg et al., 2008). Finally, RivGraph offers a suite of topologic metrics that were specifically designed for river channel network analysis (Tejedor et al., 2015b).

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

Similarity Downselection: Finding the n Most Dissimilar Molecular Conformers for Reference-Free Metabolomics

Computational methods for creating in silico libraries of molecular descriptors (e.g., collision cross sections) are becoming increasingly prevalent due to the limited number of authentic reference materials available for traditional library building. These so-called “reference-free metabolomics” methods require sampling sets of molecular conformers in order to produce high accuracy property predictions. Due to the computational cost of the subsequent calculations for each conformer, there is a need to sample the most relevant subset and avoid repeating calculations on conformers that are nearly identical. The goal of this study is to introduce a heuristic method of finding the most dissimilar conformers from a larger population in order to help speed up reference-free calculation methods and maintain a high property prediction accuracy. Finding the set of the n items most dissimilar from each other out of a larger population becomes increasingly difficult and computationally expensive as either n or the population size grows large. Because there exists a pairwise relationship between each item and all other items in the population, finding the set of the n most dissimilar items is different than simply sorting an array of numbers. For instance, if you have a set of the most dissimilar n = 4 items, one or more of the items from n = 4 might not be in the set n = 5. An exact solution would have to search all possible combinations of size n in the population exhaustively. We present an open-source software called similarity downselection (SDS), written in Python and freely available on GitHub. SDS implements a heuristic algorithm for quickly finding the approximate set(s) of the n most dissimilar items. We benchmark SDS against a Monte Carlo method, which attempts to find the exact solution through repeated random sampling. We show that for SDS to find the set of n most dissimilar conformers, our method is not only orders of magnitude faster, but it is also more accurate than running Monte Carlo for 1,000,000 iterations, each searching for set sizes n = 3–7 out of a population of 50,000. We also benchmark SDS against the exact solution for example small populations, showing that SDS produces a solution close to the exact solution in these instances. Using theoretical approaches, we also demonstrate the constraints of the greedy algorithm and its efficacy as a ratio to the exact solution.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1977

Coherent wave induced particle precipitation into the upper atmosphere

A description is presented of the results of a computer simulation involving a study of the particle precipitation induced by coherent VLF waves in the magnetosphere. The results of a computation for a 10 picoweber/sq m wave amplitude are shown in a graph. The precipitated flux for three different energies is given in a table. For 1.5 keV the energy deposition rate is about 0.8 erg/sq cm-sec, almost as intense as a moderate aurora. It is concluded that significant energy is deposited by a wave of 10 picoweber/sq m intensity. Such a wave amplitude is representative of highly coherent VLF wave types that are found in the magnetosphere. On the basis of the considered results it appears that controlled VLF wave injection in the magnetosphere could be an important and useful tool to study the coupling processes between the atmosphere and magnetosphere.

Inan, U. S.↗
NASA NTRS1979

GEOS 3 STDN S band Doppler tracking investigation

GEOS 3S Doppler band and laser ranging data, acquired from August 1975 to March 1976 in the spacecraft altimeter calibration area, are examined. An evaluation of two-way and three-way Doppler data, for the positioning of Spaceflight Tracking and Data Network S band stations is presented, as well as the Goddard Space Flight Center laser system that is used to reference the exact position of the Doppler stations. The two-way and three-way Doppler tracking devices, situated at Rosman and Bermuda, have yielded data for the recovery of GEOS 3 arc height with an uncertainty of only 1 m. Attention is given to the effects of beacon signal frequency instability, controlled by a temperature sensitive auxiliary crystal oscillator on board the spacecraft, and to the one-way range rate tracking noise that was found to be within a range of 2 to 10 cm/s. 1- and 2-way passes and their different arc meters are graphed, showing the Doppler tracking interval. It was concluded that other accurate computations and recovery of station coordinates could be performed employing tracking data from S band stations.

Rosenbaum, B.↗
NASA NTRS1986

On the free vibrations of spinning paraboloids

The dynamic behavior of a spinning linear-elastic paraboloid subject to nonaxisymmetric deformation is investigated analytically, applying the Rayleigh-Ritz procedure described by Utku et al. (1983). Energy-density, strain-displacement, and velocity-displacement expressions are generated; expressions for the generalized strain and position vector are derived; and the discretized dynamic equations are obtained. Numerical results obtained with a computer-program implementation of the method are presented in extensive tables and graphs. The effects of spin rate and bending rigidity and results for the special case of a spinning disk are included.

Shoemaker, W. L.↗
NASA NTRS1987

Methods for design and evaluation of integrated hardware/software systems for concurrent computation

Two testbed programming environments to support the evaluation of a large range of parallel architectures have been implemented under the program Parallel Implementation of Scientific Computing Environments (PISCES). The PISCES 1 environment was applied to two areas of aerospace interest: a sparse matrix iterative equation solver and a dynamic scene analysis system. Currently, the NICE/SPAR testbed system for structural analysis is being modified for parallel operation under PISCES 2; the PISCES 1 applications are also being adapted for PISCES 2. A new formal model of concurrent computation has been developed, based on the mathematical system known as H graph semantics together with a timed Petri net model of the parallel aspects of a system.

Pratt, Terrence W.↗
NASA NTRS1988

Modeling and optimum time performance for concurrent processing

The development of a new graph theoretic model for describing the relation between a decomposed algorithm and its execution in a data flow environment is presented. Called ATAMM, the model consists of a set of Petri net marked graphs useful for representing decision-free algorithms having large-grained, computationally complex primitive operations. Performance time measures which determine computing speed and throughput capacity are defined, and the ATAMM model is used to develop lower bounds for these times. A concurrent processing operating strategy for achieving optimum time performance is presented and illustrated by example.

Mielke, Roland R.↗
NASA NTRS1989

Theoretical characterization of the minimum energy path for hydrogen atom addition to N2 - Implications for the unimolecular lifetime of HN2

Results are reported from CASSCF externally contracted CI ab initio computations of the minimum-energy path for the addition of H to N2. The theoretical basis and numerical implementation of the computations are outlined, and the results are presented in extensive tables and graphs and characterized in detail. The zero-point-corrected barrier for HN2 dissociation is estimated as 8.5 kcal/mol, and the lifetime of the lowest-lying quasi-bound vibrational state of HN2 is found to be between 88 psec and 5.8 nsec (making experimental observation of this species very difficult).

Walch, Stephen P.↗