Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Graph Processing”

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 577 records · Page 32

Meshkit

MeshKit is an open-source library for mesh generation and related algorithms. MeshKit is designed both for users that perform mesh generation and for developers that develop meshing algorithm/application(s). MeshKit includes new meshing algorithms implemented first in this library, as well as interfaces to meshing algorithms/packages developed in DOE Labs and elsewhere. MeshKit uses a directed graph-based approach for organizing meshing problems that supports both traditional BREP-driven meshing as well as more general meshing processes and tools. One such tool is the Reactor Geometry (and mesh) Generator (RGG); RGG generates hexagonal and rectangular reactor geometries and meshes from text-based input in serial and parallel. It has been used to support a variety of reactor simulation codes and reactor types. Although primarily a library, MeshKit is also supporting a collaboration with Kitware to develop a graphical/GUI interface to the library.

Jain, Rajeev↗

Continuous integration data-driven platform of industrial-scale subsurface storage for real-time analytics

This project helped address the growing need for efficient and scalable models to support geological carbon and energy storage, which are crucial for achieving net-zero emissions. Traditionally accurate high-fidelity numerical models have been used to simulate relevant storage processes under a handful of processes, however such models are computationally demanding, making uncertainty quantification impractical. Consequently, we first developed a machine learning framework, based on Graph Neural Operators (GNOs), to improving the accuracy of model predictions for a fixed computational budget. We then developed an Ensemble of Improved Neural Operators (ENO), which uses bagging and Monte Carlo dropout techniques, to further improve prediction accuracy. Lastly, we developed the way to explain progressive transfer learning methods to reduce the amount of training data and computational cost of training (i.e., reduce trainable parameters) when using our models for multiple storage sites. Our numerical investigation, which used real-world case studies, demonstrated that our framework can significantly improve the safety and efficiency of geological storage operations, with potential applications in other domains such as geothermal reservoirs and climate modeling.

54 ENVIRONMENTAL SCIENCES↗

Model-Based Fault Diagnosis: Performing Root Cause and Impact Analyses in Real Time

Generic, object-oriented fault models, built according to causal-directed graph theory, have been integrated into an overall software architecture dedicated to monitoring and predicting the health of mission- critical systems. Processing over the generic fault models is triggered by event detection logic that is defined according to the specific functional requirements of the system and its components. Once triggered, the fault models provide an automated way for performing both upstream root cause analysis (RCA), and for predicting downstream effects or impact analysis. The methodology has been applied to integrated system health management (ISHM) implementations at NASA SSC's Rocket Engine Test Stands (RETS).

Figueroa, Jorge F.↗

Combining protein sequences and structures with transformers and equivariant graph neural networks to predict protein function

Abstract Motivation Millions of protein sequences have been generated by numerous genome and transcriptome sequencing projects. However, experimentally determining the function of the proteins is still a time consuming, low-throughput, and expensive process, leading to a large protein sequence-function gap. Therefore, it is important to develop computational methods to accurately predict protein function to fill the gap. Even though many methods have been developed to use protein sequences as input to predict function, much fewer methods leverage protein structures in protein function prediction because there was lack of accurate protein structures for most proteins until recently. Results We developed TransFun—a method using a transformer-based protein language model and 3D-equivariant graph neural networks to distill information from both protein sequences and structures to predict protein function. It extracts feature embeddings from protein sequences using a pre-trained protein language model (ESM) via transfer learning and combines them with 3D structures of proteins predicted by AlphaFold2 through equivariant graph neural networks. Benchmarked on the CAFA3 test dataset and a new test dataset, TransFun outperforms several state-of-the-art methods, indicating that the language model and 3D-equivariant graph neural networks are effective methods to leverage protein sequences and structures to improve protein function prediction. Combining TransFun predictions and sequence similarity-based predictions can further increase prediction accuracy. Availability and implementation The source code of TransFun is available at https://github.com/jianlin-cheng/TransFun.

59 BASIC BIOLOGICAL SCIENCES↗

Quantifying VOC-Reaction Tracers, Ozone Production, and Continuing Aerosol Production Rates in Urban and Far-Downwind Atmospheres

We have found a surprisingly informative decomposition of the complex question of smoggy ozone production (basically, [HO2] in a more locally determined field of [NO]) in the process of linked investigations of modestly smoggy Eastern North America (by NASA aircraft, July 2004) and rather polluted Flushing, NYC (Queens College, July, 2001). In both rural and very polluted situations, we find that a simple contour graph parameterization of the local principal ozone production rate can be estimated using only the variables [NO] and j(sub rads) [HCHO]: Po(O3) = c (j(sub rads) [HCHO])(sup a) [HCHO](sup b). Here j(sub rads) is the photolysis of HCHO to radicals, presumably capturing many harder-UV photolytic processes and the principle ozone production is that due to HO2; mechanisms suggest that ozone production due to RO2 is closely correlated, often suggesting a limited range of different proportionality factors. The method immediately suggests a local interpretation for concepts of VOC limitation and NOx limitation. We believe that the product j(sub rads) [HCHO] guages the oxidation rate of observed VOC mixtures in a way that also provides [HO2] useful for the principle ozone production rate k [HO2] [NO], and indeed, all ozone chemical production. The success of the method suggests that dominant urban primary-HCHO sources may transition to secondary plume-HCHO sources in a convenient way. Are there other, simple, near-terminal oxidized VOC's which help guage ozone production and aerosol particle formation? Regarding particles, we report on, to the extent NASA Research resources allow, on appealing relationships between far-downwind (Atlantic PBL) HCHO and very fine aerosol (including sulfate. Since j(sub rads) [HCHO] provides a time-scale, we may understand distant-plume particle production in a more quantitative manner. Additionally we report on a statistical search in the nearer field for relationships between glyoxals (important near-terminal aromatic and isoprene reaction products) and aerosol production, looking for VOC's that might be most implicated. All three variables j(sub rads), [HCHO], and [NO] are relatively easily measured in widespread air pollution monitoring networks, and all are deducible form space-borne observations, though estimation of [NO] from [NO2] (the species observable from space) may require care. We report also on airborne and surface observations of HCHO, suggesting that concentrated (urban) and more diffuse (forest) sources may be distinguishable from space. The use of the 3.58 micron microwindow for HCHO remote sensing should allow much sharper resolution of HCHO than the UV. UV sensing requires large and expensive instruments, but even these seem justified since formaldehyde is so informative.

Chatfield, Robert↗

More Efficient Precipitation Rate Data Acquisition with the STORM Swath-Based Analysis Tool

A multitude of locations exist for acquiring satellite-derived precipitation rate data. Because the volume of data exceeds what can reasonably be analyzed manually, online tools have been developed to help researchers make more informed data acquisition decisions. Some interfaces provide sample images generated from prioritized variables to enable efficient visual inspection, while others include the ability to subset geographically to reduce the volume of the data to the area of interest. Building upon such value-added services is the STORM (the NASA Precipitation Processing System (PPS) data product ordering system) Swath-Based Analysis Tool (https://storm.pps.eosdis.nasa.gov/storm/Analysis.jsp). With it, users can compare overflights of multiple precipitation satellites simultaneously and from this make determinations about which swaths will best serve their research. They can preview these overflights with two distinct visualization tools. Finally, they can then download geographically subset swaths that have precipitation rate characteristics matching their research needs. A user starts by making up to six selections. They then select their date range of interest going back to 1997. Finally, they select a geographical region either by inputting latitude-longitude corners of a rectangle or drawing that rectangle on an interactive map. The system processes each file, determining aggregate statistics about the precipitation rates within the selected region as observed by each instrument. These statistical values are added to a time series graph as a "time remaining" counter is continuously updated. The chart is fully interactive, with detailed values displayed upon mouseover, and numerous options available upon clicking each point. Since the user is likely to need the relevant files for their own research, they can generate an order from their request. All files are geographically subset to the region of interest, and the user can use the statistical values to limit which files are added to the order. This flexibility helps to ensure researchers are only downloading the files they need for their work, making for a more efficient process both from a time-consumption and a storage perspective. This paper will introduce the Swath-Based Analysis Tool and explore its numerous features. It will discuss several use cases for precipitation scientists.

remote sensing↗

Knowledge engineering for temporal dependency networks as operations procedures

This paper presents a case study of the knowledge engineering process employed to support the Link Monitor and Control Operator Assistant (LMCOA). The LMCOA is a prototype system which automates the configuration, calibration, test, and operation (referred to as precalibration) of the communications, data processing, metric data, antenna, and other equipment used to support space-ground communications with deep space spacecraft in NASA's Deep Space Network (DSN). The primary knowledge base in the LMCOA is the Temporal Dependency Network (TDN), a directed graph which provides a procedural representation of the precalibration operation. The TDN incorporates precedence, temporal, and state constraints and uses several supporting knowledge bases and data bases. The paper provides a brief background on the DSN, and describes the evolution of the TDN and supporting knowledge bases, the process used for knowledge engineering, and an analysis of the successes and problems of the knowledge engineering effort.

Fayyad, Kristina E.↗

An Efficient Scheme for Updating Sparse Cholesky Factors

Raghavan had earlier developed the software package DCSPACK which can be used for solving sparse linear systems where the coefficient matrix is symmetric and positive definite (this project was not funded by NASA but by agencies such as NSF). DSCPACK-S is the serial code and DSCPACK-P is a parallel implementation suitable for multiprocessors or networks-of-workstations with message passing using MCI. The main algorithm used is the Cholesky factorization of a sparse symmetric positive positive definite matrix A = LL(T). The code can also compute the factorization A = LDL(T). The complexity of the software arises from several factors relating to the sparsity of the matrix A. A sparse N x N matrix A has typically less that cN nonzeroes where c is a small constant. If the matrix were dense, it would have O(N2) nonzeroes. The most complicated part of such sparse Cholesky factorization relates to fill-in, i.e., zeroes in the original matrix that become nonzeroes in the factor L. An efficient implementation depends to a large extent on complex data structures and on techniques from graph theory to reduce, identify, and manage fill. DSCPACK is based on an efficient multifrontal implementation with fill-managing algorithms and implementation arising from earlier research by Raghavan and others. Sparse Cholesky factorization is typically a four step process: (1) ordering to compute a fill-reducing numbering, (2) symbolic factorization to determine the nonzero structure of L, (3) numeric factorization to compute L, and, (4) triangular solution to solve L(T)x = y and Ly = b. The first two steps are symbolic and are performed using the graph of the matrix. The numeric factorization step is of dominant cost and there are several schemes for improving performance by exploiting the nested and dense structure of groups of columns in the factor. The latter are aimed at better utilization of the cache-memory hierarchy on modem processors to prevent cache-misses and provide execution rates (operations/second) that are close to the peak rates for dense matrix computations. Currently, EPISCOPACY is being used in an application at NASA directed by J. Newman and M. James. We propose the implementation of efficient schemes for updating the LL(T) or LDL(T) factors computed in DSCPACK-S to meet the computational requirements of their project. A brief description is provided in the next section.

Raghavan, Padma↗

Graph Generation with a Focusing Lexicon

Reference datasets are an essential part of developing effective data exploration tools. Many reference datasets are synthetic in nature, seeding real-world properties into generation routines. Others are laboriously curated from original source materials. This is a time-consuming process that requires significant domain expertise all along the process. This paper presents a third path: deriving a network from real-world source materials using automated tools. By properly grounding the tools in a small collection of curated documents, a larger corpus be automatically processed with greater confidence. The end system leverages human knowledge (in the initial document curation), but with much less time commitment enabled by automation. Our ultimate goal is to produce reference datasets, derived from real-world data that capture aspects salient to a specific activity in the context of all of the activities in the data-set. Our reference datasets take the form of graphs.

Chandra Shekar, Mayanka↗

Disease mapping based on stochastic SIR-SI model for Dengue and Chikungunya in Malaysia

This paper describes and demonstrates a method for relative risk estimation which is based on the stochastic SIR-SI vector-borne infectious disease transmission model specifically for Dengue and Chikungunya diseases in Malaysia. Firstly, the common compartmental model for vector-borne infectious disease transmission called the SIR-SI model (susceptible-infective-recovered for human populations; susceptible-infective for vector populations) is presented. This is followed by the explanations on the stochastic SIR-SI model which involve the Bayesian description. This stochastic model then is used in the relative risk formulation in order to obtain the posterior relative risk estimation. Then, this relative estimation model is demonstrated using Dengue and Chikungunya data of Malaysia. The viruses of these diseases are transmitted by the same type of female vector mosquito named Aedes Aegypti and Aedes Albopictus. Finally, the findings of the analysis of relative risk estimation for both Dengue and Chikungunya diseases are presented, compared and displayed in graphs and maps. The distribution from risk maps show the high and low risk area of Dengue and Chikungunya diseases occurrence. This map can be used as a tool for the prevention and control strategies for both diseases.

60 APPLIED LIFE SCIENCES↗

A one-dimensional stochastic approach to the study of cyclic voltammetry with adsorption effects

In this study, a one-dimensional stochastic model based on the random walk approach is used to simulate cyclic voltammetry. The model takes into account mass transport, kinetics of the redox reactions, adsorption effects and changes in the morphology of the electrode. The model is shown to display the expected behavior. Furthermore, the model shows consistent qualitative agreement with a finite difference solution. This approach allows for an understanding of phenomena on a microscopic level and may be useful for analyzing qualitative features observed in experimentally recorded signals.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Software Defined Architectures for Data Analytics

Data analytics applications increasingly are complex workflows composed of phases with very different program behaviors (e.g., graph algorithms and machine learning, algorithms operating on sparse an dense data structures, etc). To reach the levels of efficiency required to process these workflows in real time, upcoming architectures will need to leverage even more workload specialization. If, at one end, we may find even more heterogenous processors composed by a myriad of specialized processing elements, at the other end we may see novel reconfigurable architectures, composed of sets of functional units and memories interconnected with (re)configurable on-chip networks, able to adapt dynamically to adapt the workload characteristics. Field Programmable Gate Arrays are more and more used for accelerating various workloads and, in particular, inferencing in machine learning, providing higher efficiency than other solutions. However, their fine grained nature still leads to issues for the design software and still makes dynamic reconfiguration impractical. Future, more coarse-grained architectures could offer the features to execute diverse workloads at high efficiency while providing better reconfiguration mechanisms for dynamic adaptability. Nevertheless, we argue that the challenges for reconfigurable computing remain in the software. In this position paper, we describe a possible toolchain for reconfigurable architectures targeted at data analytics.

FPGA, Dataflow, Reconfigurable Architectures, Soft↗

Task Parallel Incomplete Cholesky Factorization using 2D Partitioned-Block Layout

We introduce a task-parallel algorithm for sparse incomplete Cholesky factorization that utilizes a 2D sparse partitioned-block layout of a matrix. Our factorization algorithm follows the idea of algorithms-by-blocks by using the block layout. The algorithm-byblocks approach induces a task graph for the factorization. These tasks are inter-related to each other through their data dependences in the factorization algorithm. To process the tasks on various manycore architectures in a portable manner, we also present a portable tasking API that incorporates different tasking backends and device-specific features using an open-source framework for manycore platforms i.e., Kokkos. A performance evaluation is presented on both Intel Sandybridge and Xeon Phi platforms for matrices from the University of Florida sparse matrix collection to illustrate merits of the proposed task-based factorization. Experimental results demonstrate that our task-parallel implementation delivers about 26.6x speedup (geometric mean) over single-threaded incomplete Choleskyby- blocks and 19.2x speedup over serial Cholesky performance which does not carry tasking overhead using 56 threads on the Intel Xeon Phi processor for sparse matrices arising from various application problems.

2D layout↗

Space Station Freedom science utilization: 2000 and beyond

The Office of Space Science and Applications (OSSA) Goals are: (1) to advance scientific knowledge of the Earth, the Solar System, and the Universe; (2) to understand the effects of the space environment on biological and physical processes; and (3) to expand the human presence beyond the Earth into the Solar System. The topics covered are presented in view graph form and include the following: (1) a strategic plan; (2) evolving U.S. space science capabilities; (3) life science goals; (4) OSSA Space Station Payload Traffic Model May 1991; and (5) microgravity science and applications goals.

Source record↗

Parametric design and gridding through relational geometry

Relational Geometric Synthesis (RGS) is a new logical framework for building up precise definitions of complex geometric models from points, curves, surfaces and solids. RGS achieves unprecedented design flexibility by supporting a rich variety of useful curve and surface entities. During the design process, many qualitative and quantitative relationships between elementary objects may be captured and retained in a data structure equivalent to a directed graph, such that they can be utilized for automatically updating the complete model geometry following changes in the shape or location of an underlying object. Capture of relationships enables many new possibilities for parametric variations and optimization. Examples are given of panelization applications for submarines, sailing yachts, offshore structures, and propellers.

Letcher, John S., Jr.↗

Support of Data Access for the IMP-8 UMD Experiment

This grant report provides information on data from the Interplanetary Monitoring Platform-8 (IMP-8). Topics covered include: (1) the science involved in the project; (2) the collection of data; (3) the processing of data; (4) the submission of data to other facilities; (5) the availability of data on the world wide web (WWW). Graphs are also included of data on the interstellar medium.

Ipavich, F. M.↗

Resilience Assessment Framework For Electric Distribution Systems Performance Under Extreme Conditions

The devastating impact of extreme weather-related events is increasingly evident on power grids, especially on distribution grids. The severity of their potential impact calls for 1) developing a suitable resilience assessment framework to capture the system performance and 2) assessing relevant mitigative strategies to lessen the impact of such events. This paper proposes a framework to identify grid vulnerabilities using the energy-at-risk concept to select, disconnect and isolate grid portions due to a resilience event. The proposed framework mainly consists of two steps; i) processing the utility's available infrastructure, i.e., a network model, possible switching combinations, and outage information for those combinations, as a graph-based database, and ii) implementing a novel optimal switching algorithm leveraging database and grid simulated metrics. These switching actions are generated to implement load curtailment in a rolling manner during anticipated grid scarcity conditions. In this study, a test case is created using two taxonomy feeders and is simulated against an extreme temperature event, e.g., a long, relatively cold, and prolonged freeze peak, thereby creating stress on the grid. It is demonstrated that the proposed framework allows utilities to predict the energy-at-risk during such resilience events and design suitable outage management strategies.

Poudel, Shiva↗

A spectral analysis of the domain decomposed Monte Carlo method for linear systems

The domain decomposed behavior of the adjoint Neumann-Ulam Monte Carlo method for solving linear systems is analyzed using the spectral properties of the linear operator. Relationships for the average length of the adjoint random walks, a measure of convergence speed and serial performance, are made with respect to the eigenvalues of the linear operator. In addition, relationships for the effective optical thickness of a domain in the decomposition are presented based on the spectral analysis and diffusion theory. Using the effective optical thickness, the Wigner rational approximation and the mean chord approximation are applied to estimate the leakage fraction of stochastic histories from a domain in the decomposition as a measure of parallel performance and potential communication costs. The one-speed, two-dimensional neutron diffusion equation is used as a model problem to test the models for symmetric operators. In general, the derived approximations show good agreement with measured computational results. (authors)

97 MATHEMATICS AND COMPUTING↗