Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Data Structures and Algorithms”

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 343 records · Page 19
All sourcesNASA NTRSDOE OSTI
NASA NTRS2006

Diurnal Variability of Vertical Structure from a TRMM Passive Microwave "Virtual Radar" Retrieval

Robust description of the diurnal cycle from TRMM observations is complicated by the limitations of Low Earth Orbit (LEO) sampling; from a 'climatological' perspective, sufficient sampling must exist to control for both spatial and seasonal variability, before tackling an additional diurnal component (e.g., with 8 additional 3-hourly or 24 1-hourly bins). For documentation of vertical structure, the narrow sample swath of the TRMM Precipitation Radar limits the resolution of any of these components. A neural-network based 'virtual radar" retrieval has been trained and internally validated, using multifrequency / multipolarization passive microwave(TM1) brightness temperatures and textures parameters and lightning (LIS) observations, as inputs, and PR volumetric reflectivity as targets (outputs). By training the algorithms (essentially highly multivariate, nonlinear regressions) on a very large sample of high-quality co-located data from the center of the TRMM swath, 3D radar reflectivity and derived parameters (VIL, IWC, Echo Tops, etc.) can be retrieved across the entire TMI swath, good to 8-9% over the dynamic range of parameters. As a step in the retrieval (and as an output of the process), each TMI multifrequency pixel (at 85 GHz resolution) is classified into one of the 25 archetypal radar profile vertical structure "types", previously identified using cluster analysis. The dynamic range of retrieved vertical structure appears to have higher fidelity than the current (Version 6) experimental GPROF hydrometeor vertical structure retrievals. This is attributable to correct representation of the prior probabilities of vertical structure variability in the neural network training data, unlike the GPROF cloud-resolving model training dataset used in the V6 algorithms. The LIS lightning inputs are supplementary inputs, and a separate offline neural network has been trained to impute (predict) LIS lightning from passive-microwave-only data. The virtual radar retrieval is thus, in principle, extensible to Aqua/AMSR-E and NPOESS/CMIS passive microwave instruments. The virtual radar approach yields a threefold increase in effective sampling from the mission, albeit of lower-quality "retrieved" data, reducing the variance of local estimates by one third (or the standard deviation by-0.57). In this talk, the variance reduction is leveraged to more finely resolve global diurnal variability in both space and time (local hour).

Boccippio, Dennis J.↗
NASA NTRS1986

Maximum-Likelihood Parameter-Estimation Algorithm

Efficient version of maximum-likelihood algorithm devised for calculating normal-mode frequencies and damping parameters of vibrating system from experimental data where both process noise and measurement noise present. Method applicable in vibration analysis of such complicated structures as vehicles, aircraft, and spacecraft. New algorithm simplification of existing maximum-likelihood formulation using Kalman filter that allows for both process and measurement noise.

Eldred, D. B.↗
NASA NTRS2009

Sequential Principal Component Analysis -An Optimal and Hardware-Implementable Transform for Image Compression

This paper presents the JPL-developed Sequential Principal Component Analysis (SPCA) algorithm for feature extraction / image compression, based on "dominant-term selection" unsupervised learning technique that requires an order-of-magnitude lesser computation and has simpler architecture compared to the state of the art gradient-descent techniques. This algorithm is inherently amenable to a compact, low power and high speed VLSI hardware embodiment. The paper compares the lossless image compression performance of the JPL's SPCA algorithm with the state of the art JPEG2000, widely used due to its simplified hardware implementability. JPEG2000 is not an optimal data compression technique because of its fixed transform characteristics, regardless of its data structure. On the other hand, conventional Principal Component Analysis based transform (PCA-transform) is a data-dependent-structure transform. However, it is not easy to implement the PCA in compact VLSI hardware, due to its highly computational and architectural complexity. In contrast, the JPL's "dominant-term selection" SPCA algorithm allows, for the first time, a compact, low-power hardware implementation of the powerful PCA algorithm. This paper presents a direct comparison of the JPL's SPCA versus JPEG2000, incorporating the Huffman and arithmetic coding for completeness of the data compression operation. The simulation results show that JPL's SPCA algorithm is superior as an optimal data-dependent-transform over the state of the art JPEG2000. When implemented in hardware, this technique is projected to be ideally suited to future NASA missions for autonomous on-board image data processing to improve the bandwidth of communication.

Duong, Tuan A.↗
NASA NTRS1981

Improved methods for fan sound field determination

Several methods for determining acoustic mode structure in aircraft turbofan engines using wall microphone data were studied. A method for reducing data was devised and implemented which makes the definition of discrete coherent sound fields measured in the presence of engine speed fluctuation more accurate. For the analytical methods, algorithms were developed to define the dominant circumferential modes from full and partial circumferential arrays of microphones. Axial arrays were explored to define mode structure as a function of cutoff ratio, and the use of data taken at several constant speeds was also evaluated in an attempt to reduce instrumentation requirements. Sensitivities of the various methods to microphone density, array size and measurement error were evaluated and results of these studies showed these new methods to be impractical. The data reduction method used to reduce the effects of engine speed variation consisted of an electronic circuit which windowed the data so that signal enhancement could occur only when the speed was within a narrow range.

Cicon, D. E.↗
DOE OSTI2021

FY21 Progress Report: SRNL Analysis of ICCWR LCM and WAMS data for Corrosion and Cracking

The development of algorithms for machine learning and data analysis for the 3013 Surveillance Program is a collaborative effort by the Savannah River National Laboratory (SRNL) and the University of South Carolina (USC). For corrosion detection, Laser Confocal Microscope (LCM) or Wide Area 3D Measurement System (WAMS) data is extracted from large binary files, with software written to convert the data to physical attributes (e.g., height, color and grayscale values; all as functions of a location in a plane projection). A user-friendly Matlab Graphical User Interface (GUI) that reads data from either LCM or WAMS files was developed to integrate input data with software developed for processing and evaluation. The GUI can selectively download binary data, interrogate data attributes, label data, flag significant features, execute Machine Learning (ML) algorithms, output parameters for trained ML algorithms, report ML model accuracy with respect to labeled data, and generate graphical representations for various analyses. Features can be called out by user-specified thresholds, manual labeling or machine learning algorithms when they have been completed. The ability to rapidly label data is important because of the volume of data required for training machine learning algorithms. The GUI has the flexibility to allow addition of improved ML algorithms, methods for data visualization, and statistical computations. Statistical analyses via the GUI include areas of pits within a defined range of pit depths, correlations between Red-Green-Blue (RGB) or grayscale intensity and relative surface height, covariances between values associated with features, and feature histograms. The development of supervised machine learning algorithms, however, has been hindered by a lack of training data. The machine learning algorithms for crack identification are being refined but require improvements to the true positive rate for crack detection. This shortcoming is an artifact of the limited training data currently available, perhaps more so than the structure of the neural networks. At present, the best results are had from a consensus over an ensemble of randomly generated Deep Neural Network (DNN) or Convolutional Neural Network (CNN) algorithms. Although the consensus accuracy method has yielded optimum true positive and true negative rates in excess of 80%, additional validation testing is necessary. In addition to the suite of LCM data that was initially used, and which represents the majority of the work presented in this report, WAMS image data was also reviewed at a preliminary level. The review included a comparison between image resolution and dynamic range for each method. WAMS (ZON file) image data was found to have a pixel pitch of 3.69μm compared to 1 μm for the LCM (vk4 file) data, which implies a lower resolution for the WAMS images. Conversely, the ratio of dynamic range of the WAMS data to the LCM data was approximately 41:20 for height data, suggesting that information from WAMS should more accurately determine the depth of pits. At present, the significance of the greater dynamic range of the WAMS data relative to the LCM data has not yet been evaluated.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2024

ESM data downscaling: a comparison of super-resolution deep learning models

Abstract Climate projections at fine spatial resolutions are required to conduct accurate risk assessment for critical infrastructure and design adaptation planning. Generating these projections using advanced Earth system models (ESM) requires significant computational resources. To address this issue, various statistical downscaling techniques have been introduced to generate fine-resolution data from coarse-resolution simulations. In this study, we evaluate and compare five deep learning-based downscaling techniques, namely, super-resolution convolutional neural networks, fast super-resolution convolutional neural network ESM, efficient sub-pixel convolutional neural network, enhanced deep residual network (EDRN), and super-resolution generative adversarial network (SRGAN). These techniques are applied to a dataset generated by the Energy Exascale Earth System Model (E3SM), focusing on key surface variables such as surface temperature, shortwave heat flux, and longwave heat flux. Models are trained and validated using paired fine-resolution (0.25 $$^{\circ }$$ ∘ ) and coarse-resolution (1 $$^{\circ }$$ ∘ ) monthly data obtained from a 9-year simulation. Next, blind testing is performed using monthly data obtained from two different years outside of the training and validation set. To evaluate the efficiency of each technique, different statistical metrics are used, including mean squared error (MSE), peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and learned perceptual image patch similarity (LPIPS). The results show that EDRN outperforms other algorithms in terms of PSNR, SSIM, and MSE, but struggles to capture fine-scale features in the data. In contrast, SRGAN, a generative model that uses perceptual loss, excels in capturing fine details at boundaries and internal structures, resulting in lower LPIPS than other methods.

Pawar, Nikhil M. (ORCID:0000000211613289)↗
DOE OSTI2022

Combining Physics and Machine Learning for the Next Generation of Molecular Simulation

Simulating molecules and atomic systems at quantum accuracy is a grand challenge for science in the 21 st century. Quantum-accurate simulations would enable the design of new medicines and the discovery of new materials. The defining problem in this challenge is that quantum calculations on large molecules, like proteins or DNA, are fundamentally impossible with current algorithms. In this work, we explore a range of different methods that aim to make large, quantum-accurate simulations possible. We show that using advanced classical models, we can accurately simulate ion channels, an important biomolecular system. We show how advanced classical models can be implemented in an exascale-ready software package. Lastly, we show how machine learning can learn the laws of quantum mechanics from data and enable quantum electronic structure calculations on thousands of atoms, a feat that is impossible for current algorithms. Altogether, this work shows that combining advances in physics models, computing, and machine learning, we are moving closer to the reality of accurately simulating our molecular world.

74 ATOMIC AND MOLECULAR PHYSICS↗
NASA NTRS2012

Rapid Process to Generate Beam Envelopes for Optical System Analysis

The task of evaluating obstructions in the optical throughput of an optical system requires the use of two disciplines, and hence, two models: optical models for the details of optical propagation, and mechanical models for determining the actual structure that exists in the optical system. Previous analysis methods for creating beam envelopes (or cones of light) for use in this obstruction analysis were found to be cumbersome to calculate and take significant time and resources to complete. A new process was developed that takes less time to complete beam envelope analysis, is more accurate and less dependent upon manual node tracking to create the beam envelopes, and eases the burden on the mechanical CAD (computer-aided design) designers to form the beam solids. This algorithm allows rapid generation of beam envelopes for optical system obstruction analysis. Ray trace information is taken from optical design software and used to generate CAD objects that represent the boundary of the beam envelopes for detailed analysis in mechanical CAD software. Matlab is used to call ray trace data from the optical model for all fields and entrance pupil points of interest. These are chosen to be the edge of each space, so that these rays produce the bounding volume for the beam. The x and y global coordinate data is collected on the surface planes of interest, typically an image of the field and entrance pupil internal of the optical system. This x and y coordinate data is then evaluated using a convex hull algorithm, which removes any internal points, which are unnecessary to produce the bounding volume of interest. At this point, tolerances can be applied to expand the size of either the field or aperture, depending on the allocations. Once this minimum set of coordinates on the pupil and field is obtained, a new set of rays is generated between the field plane and aperture plane (or vice-versa). These rays are then evaluated at planes between the aperture and field, at a desired number of steps perceived necessary to build up the bounding volume or cone shape. At each plane, the ray coordinates are again evaluated using the convex hull algorithm to reduce the data to a minimal set. When all of the coordinates of interest are obtained for every plane of the propagation, the data is formatted into an xyz file suitable for FRED optical analysis software to import and create a STEP file of the data. This results in a spiral-like structure that is easily imported by mechanical CAD users who can then use an automated algorithm to wrap a skin around it and create a solid that represents the beam.

Howard, Joseph↗
DOE OSTI2023

Representations of Materials for Machine Learning

High-throughput data generation methods and machine learning (ML) algorithms have given rise to a new era of computational materials science by learning the relations between composition, structure, and properties and by exploiting such relations for design. However, to build these connections, materials data must be translated into a numerical form, called a representation, that can be processed by an ML model. Data sets in materials science vary in format (ranging from images to spectra), size, and fidelity. Predictive models vary in scope and properties of interest. Here, we review context-dependent strategies for constructing representations that enable the use of materials as inputs or outputs for ML models. Furthermore, we discuss how modern ML techniques can learn representations from data and transfer chemical and physical information between tasks. Finally, we outline high-impact questions that have not been fully resolved and thus require further investigation.

36 MATERIALS SCIENCE↗
NASA NTRS2012

SureTrak Probability of Impact Display

The SureTrak Probability of Impact Display software was developed for use during rocket launch operations. The software displays probability of impact information for each ship near the hazardous area during the time immediately preceding the launch of an unguided vehicle. Wallops range safety officers need to be sure that the risk to humans is below a certain threshold during each use of the Wallops Flight Facility Launch Range. Under the variable conditions that can exist at launch time, the decision to launch must be made in a timely manner to ensure a successful mission while not exceeding those risk criteria. Range safety officers need a tool that can give them the needed probability of impact information quickly, and in a format that is clearly understandable. This application is meant to fill that need. The software is a reuse of part of software developed for an earlier project: Ship Surveillance Software System (S4). The S4 project was written in C++ using Microsoft Visual Studio 6. The data structures and dialog templates from it were copied into a new application that calls the implementation of the algorithms from S4 and displays the results as needed. In the S4 software, the list of ships in the area was received from one local radar interface and from operators who entered the ship information manually. The SureTrak Probability of Impact Display application receives ship data from two local radars as well as the SureTrak system, eliminating the need for manual data entry.

Elliott, John↗
NASA NTRS1993

Modal characterization of the ASCIE segmented optics testbed: New algorithms and experimental results

New frequency response measurement procedures, on-line modal tuning techniques, and off-line modal identification algorithms are developed and applied to the modal identification of the Advanced Structures/Controls Integrated Experiment (ASCIE), a generic segmented optics telescope test-bed representative of future complex space structures. The frequency response measurement procedure uses all the actuators simultaneously to excite the structure and all the sensors to measure the structural response so that all the transfer functions are measured simultaneously. Structural responses to sinusoidal excitations are measured and analyzed to calculate spectral responses. The spectral responses in turn are analyzed as the spectral data become available and, which is new, the results are used to maintain high quality measurements. Data acquisition, processing, and checking procedures are fully automated. As the acquisition of the frequency response progresses, an on-line algorithm keeps track of the actuator force distribution that maximizes the structural response to automatically tune to a structural mode when approaching a resonant frequency. This tuning is insensitive to delays, ill-conditioning, and nonproportional damping. Experimental results show that is useful for modal surveys even in high modal density regions. For thorough modeling, a constructive procedure is proposed to identify the dynamics of a complex system from its frequency response with the minimization of a least-squares cost function as a desirable objective. This procedure relies on off-line modal separation algorithms to extract modal information and on least-squares parameter subset optimization to combine the modal results and globally fit the modal parameters to the measured data. The modal separation algorithms resolved modal density of 5 modes/Hz in the ASCIE experiment. They promise to be useful in many challenging applications.

Carrier, Alain C.↗
DOE OSTI2024

Collinear limit of the four-point energy correlator in N = 4 supersymmetric Yang-Mills theory

We present a compact formula, expressed in terms of classical polylogarithms up to weight three, for the leading order four-point energy correlator in maximally supersymmetric Yang-Mills theory, in the limit where the four detectors are collinear. This formula is derived by combining a simplified, manifestly dual conformal invariant form of the 1 → 4 splitting function obtained from the square of the tree-level five-particle form factor of stress-tensor multiplet operators, with a novel integration-by-parts algorithm operating directly on Feynman parameter integrals. Our results provide valuable data for exploring the structure of physical observables in perturbation theory, and for calculations of jet substructure observables in quantum chromodynamics. Published by the American Physical Society 2024

Chicherin, Dmitry (ORCID:000000028985084X)↗
DOE OSTI2025

Desmearing two-dimensional small-angle neutron scattering data by central moment expansions

Resolution smearing is a critical challenge in the quantitative analysis of two-dimensional small-angle neutron scattering (SANS) data, particularly in studies of soft-matter flow and deformation using SANS. Here, we present a central moment expansion technique to address smearing in anisotropic scattering spectra, offering a model-free desmearing methodology. By accounting for directional variations in resolution smearing and enhancing computational efficiency, this approach reconstructs desmeared intensity distributions from smeared experimental data. Computational benchmarks using interacting hard-sphere fluids and Gaussian chain models validate the accuracy of the method, while simulated noise analyses confirm its robustness under experimental conditions. Experimental validation using rheological SANS data from shear-induced micellar structures demonstrates the practicality and effectiveness of the proposed algorithm. The desmearing technique provides a powerful tool for advancing the quantitative analysis of anisotropic scattering patterns, enabling precise insights into the interplay between material microstructure and macroscopic flow behavior.

anisotropic scattering spectra↗
DOE OSTI2025

Accelerating GNNs on GPU Sparse Tensor Cores through N:M Sparsity-Oriented Graph Reordering

Recent advancements in GPU hardware support have introduced the capability to leverage N:M sparse patterns for substantial performance gains. Graphs in Graph Neural Networks (GNNs) are typically sparse, but the sparsity is often irregular, not conforming to such sparse patterns. In this paper, we propose a novel graph reordering algorithm, the first of its kind, to reshape irregular graph data into the N:M structured sparse pattern at the tile level, allowing linear-algebra-based graph operations in GNNs to benefit from the N:M sparse hardware. The optimization is lossless, maintaining the accuracy of GNN. It can remove 98-100\% violations of the N:M sparse patterns at the vector level, and increase the proportion of conforming graphs in SuiteSparse collection from 5-9\% to 88.7-93.5\%. On A100 GPUs, the optimization accelerates Sparse Matrix Matrix (SpMM) by up to 43X (2.3X -- 7.5X on average) and speeds up the key graph operations in GNNs on real graphs by as much as 8.6X (3.5X on average).

artificial intelligence, graph neural networks↗
NASA NTRS1987

Least squares linear lags and limited memory filters

Pure autoregressive (AR) models which are linear in the short term, that is when a variable can be predicted by linear regression on a limited number of past observations, are discussed. When evenly spaced observations are available, a fixed set of AR coefficients can be calculated independent of the data. For filtering purposes, such a lag structure can be implemented recursively with an efficient algorithm. The method of computing variance recursively is also derived. A complete algorithm is presented in the appendix.

Discenza, Joseph H.↗
NASA NTRS1987

Adaptive finite element methods for high-speed compressible flows

An adaptive finite element algorithm for solving the unsteady Euler equations is described. The finite element algorithm is based on a Taylor/Galerkin formulation and uses a very fast and efficient data structure to refine and unrefine the grid in order to optimize the approximation. A general version of the method which can be applied to moving grids with sliding interfaces is given, and results for a transient supersonic calculation of rotor-stator interaction are presented.

Oden, J. T.↗
NASA NTRS1992

AMR on the CM-2

We describe the development of a structured adaptive mesh algorithm (AMR) for the Connection Machine-2 (CM-2). We develop a data layout scheme that preserves locality even for communication between fine and coarse grids. On 8K of a 32K machine we achieve performance slightly less than 1 CPU of the Cray Y-MP. We apply our algorithm to an inviscid compressible flow problem.

Berger, Marsha J.↗
NASA NTRS1994

Simulation of arthroscopic surgery using MRI data

With the availability of Magnetic Resonance Imaging (MRI) technology in the medical field and the development of powerful graphics engines in the computer world the possibility now exists for the simulation of surgery using data obtained from an actual patient. This paper describes a surgical simulation system which will allow a physician or a medical student to practice surgery on a patient without ever entering an operating room. This could substantially lower the cost of medial training by providing an alternative to the use of cadavers. This project involves the use of volume data acquired by MRI which are converted to polygonal form using a corrected marching cubes algorithm. The data are then colored and a simulation of surface response based on springy structures is performed in real time. Control for the system is obtained through the use of an attached analog-to-digital unit. A remote electronic device is described which simulates an imaginary tool having features in common with both arthroscope and laparoscope.

Heller, Geoffrey↗