SEARCH · Search NASA

Results for “task decomposition”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 181 records · Page 10

All sources NASA NTRS DOE OSTI

NASA NTRS2001

Real-Time Inhibitor Recession Measurements in the Space Shuttle Reusable Solid Rocket Motors

Real-time char line recession measurements were made on propellant inhibitors of the Space Shuttle Reusable Solid Rocket Motor (RSRM). The RSRM FSM-8 static test motor propellant inhibitors (composed of a rubber insulation material) were successfully instrumented with eroding potentiometers and thermocouples. The data was used to establish inhibitor recession versus time relationships. Normally, pre-fire and post-fire insulation thickness measurements establish the thermal performance of an ablating insulation material. However, post-fire inhibitor decomposition and recession measurements are complicated by the fact that most of the inhibitor is back during motor operation. It is therefore a difficult task to evaluate the thermal protection offered by the inhibitor material. Real-time measurements would help this task. The instrumentation program for this static test motor marks the first time that real-time inhibitors. This report presents that data for the center and aft field joint forward facing inhibitors. The data was primarily used to measure char line recession of the forward face of the inhibitors which provides inhibitor thickness reduction versus time data. The data was also used to estimate the inhibitor height versus time relationship during motor operation.

McWhorter, Bruce B.↗

NASA NTRS2001

Analysis and visualization of single-trial event-related potentials

In this study, a linear decomposition technique, independent component analysis (ICA), is applied to single-trial multichannel EEG data from event-related potential (ERP) experiments. Spatial filters derived by ICA blindly separate the input data into a sum of temporally independent and spatially fixed components arising from distinct or overlapping brain or extra-brain sources. Both the data and their decomposition are displayed using a new visualization tool, the "ERP image," that can clearly characterize single-trial variations in the amplitudes and latencies of evoked responses, particularly when sorted by a relevant behavioral or physiological variable. These tools were used to analyze data from a visual selective attention experiment on 28 control subjects plus 22 neurological patients whose EEG records were heavily contaminated with blink and other eye-movement artifacts. Results show that ICA can separate artifactual, stimulus-locked, response-locked, and non-event-related background EEG activities into separate components, a taxonomy not obtained from conventional signal averaging approaches. This method allows: (1) removal of pervasive artifacts of all types from single-trial EEG records, (2) identification and segregation of stimulus- and response-locked EEG components, (3) examination of differences in single-trial responses, and (4) separation of temporally distinct but spatially overlapping EEG oscillatory activities with distinct relationships to task events. The proposed methods also allow the interaction between ERPs and the ongoing EEG to be investigated directly. We studied the between-subject component stability of ICA decomposition of single-trial EEG epochs by clustering components with similar scalp maps and activation power spectra. Components accounting for blinks, eye movements, temporal muscle activity, event-related potentials, and event-modulated alpha activities were largely replicated across subjects. Applying ICA and ERP image visualization to the analysis of sets of single trials from event-related EEG (or MEG) experiments can increase the information available from ERP (or ERF) data. Copyright 2001 Wiley-Liss, Inc.

Non-NASA Center↗

DOE OSTI2017

Soda Pop: A Time-Series Clustering, Alarming and Disease Forecasting Application

To introduce Soda Pop, an R/Shiny application designed to be a disease agnostic time-series clustering, alarming, and forecasting tool to assist in disease surveillance “triage, analysis and reporting” workflows within the Biosurveillance Ecosystem (BSVE). In this poster, we highlight the new capabilities that are brought to the BSVE by Soda Pop with an emphasis on the impact of methodological decisions. The Biosurveillance Ecosystem (BSVE) is a biological and chemical threat surveillance system sponsored by the Defense Threat Reduction Agency (DTRA). BSVE is intended to be user-friendly, multi-agency, cooperative, modular and threat agnostic platform for biosurveillance. In BSVE, a web-based workbench presents the analyst with applications (apps) developed by various DTRA funded researchers, which are deployed on-demand in the cloud(e.g., Amazon Web Services). These apps aim to address emerging needs and refine capabilities to enable early warning of chemical and biological threats for multiple users across local, state, and federal agencies. Soda Pop is an app developed by Pacific Northwest National Laboratory (PNNL) to meet the current needs of the BSVE for early warning and detection of disease outbreaks. Aimed for use bya diverse set of analysts, the application is agnostic to data source and spatial scale enabling it to be generalizable across many diseases and locations. To achieve this, we placed a particular emphasis on clustering and alerting of disease signals within Soda Pop without strong prior assumptions on the nature of observed diseased counts. Although designed to be agnostic to the data source, Soda Pop was initially developed and tested on data summarizing Influenza-Like Illness in military hospitals from collaboration with the Armed Forces Health Surveillance Branch. Currently, the data incorporated also includes the CDC’s National Notifiable Diseases Surveillance System(NNDSS) tables and the WHO’s Influenza A/B Influenza Data(Flunet). These data sources are now present in BSVE’s Postgres data storage for direct access. Soda Pop is designed to automate time-series tasks of data summarization, exploration, clustering, alarming and forecasting. Built as an R/Shiny application, Soda Pop is founded on the powerful statistical tool R.. Where applicable, Soda Pop facilitates nonparametric seasonal decomposition of time-series; hierarchical agglomerative clustering across reporting areas and between diseases within reporting areas; and a variety of alarming techniques including Exponential Weighted Moving Average alarms and Early Aberration Detection. Soda Pop embeds these techniques within a user-interface designed to enhance an analyst’s understanding of emerging trends in their data and enables the inclusion of its graphical elements into their dossier for further tracking and reporting. The ultimate goal of this software is to facilitate the discovery of unknown disease signals along with increasing the speed of detection of unusual patterns within these signals. Soda Pop organizes common statistical disease surveillance tasks in a manner integrated with BSVE data source inputs and outputs. The app analyzes time-series disease data and supports a robust set of clustering and alarming routines that avoid strong assumptions on the nature of observed disease counts. This attribute allows for flexibility in the data source, spatial scale, and disease types making it useful to a wide range of analysts.

97 MATHEMATICS AND COMPUTING↗

NASA NTRS1991

The formal verification of generic interpreters

The task assignment 3 of the design and validation of digital flight control systems suitable for fly-by-wire applications is studied. Task 3 is associated with formal verification of embedded systems. In particular, results are presented that provide a methodological approach to microprocessor verification. A hierarchical decomposition strategy for specifying microprocessors is also presented. A theory of generic interpreters is presented that can be used to model microprocessor behavior. The generic interpreter theory abstracts away the details of instruction functionality, leaving a general model of what an interpreter does.

Windley, P.↗

NASA NTRS1990

Artificial intelligence and design: Opportunities, research problems and directions

The issues of industrial productivity and economic competitiveness are of major significance in the U.S. at present. By advancing the science of design, and by creating a broad computer-based methodology for automating the design of artifacts and of industrial processes, we can attain dramatic improvements in productivity. It is our thesis that developments in computer science, especially in Artificial Intelligence (AI) and in related areas of advanced computing, provide us with a unique opportunity to push beyond the present level of computer aided automation technology and to attain substantial advances in the understanding and mechanization of design processes. To attain these goals, we need to build on top of the present state of AI, and to accelerate research and development in areas that are especially relevant to design problems of realistic complexity. We propose an approach to the special challenges in this area, which combines 'core work' in AI with the development of systems for handling significant design tasks. We discuss the general nature of design problems, the scientific issues involved in studying them with the help of AI approaches, and the methodological/technical issues that one must face in developing AI systems for handling advanced design tasks. Looking at basic work in AI from the perspective of design automation, we identify a number of research problems that need special attention. These include finding solution methods for handling multiple interacting goals, formation problems, problem decompositions, and redesign problems; choosing representations for design problems with emphasis on the concept of a design record; and developing approaches for the acquisition and structuring of domain knowledge with emphasis on finding useful approximations to domain theories. Progress in handling these research problems will have major impact both on our understanding of design processes and their automation, and also on several fundamental questions that are of intrinsic concern to AI. We present examples of current AI work on specific design tasks, and discuss new directions of research, both as extensions of current work and in the context of new design tasks where domain knowledge is either intractable or incomplete. The domains discussed include Digital Circuit Design, Mechanical Design of Rotational Transmissions, Design of Computer Architectures, Marine Design, Aircraft Design, and Design of Chemical Processes and Materials. Work in these domains is significant on technical grounds, and it is also important for economic and policy reasons.

Amarel, Saul↗

DOE OSTI2019

O(N) Quantum Molecular Dynamics (QMD) Algorithms, Code Requirements, Release Notes, and Library Deployment

The Exascale Computing Project (ECP) Co-design Center for Particle Applications (CoPA) is addressing the challenges for particle applications to run on upcoming exascale computing architectures. This includes short- and long-range molecular dynamics (MD), particle-in-cell (PIC), and quantum molecular dynamics (QMD). This document addresses $\mathcal{O}(N)$ QMD. The goal of this CoPA team is to develop $\mathcal{O}(N)$ linear scaling methods for QMD optimized for new and next generation architectures. This will provide the capability to simulate larger systems and run for longer timeframes than is possible today. Focus is initially on the density matrix construction. This requires efficient matrix formats and linear algebra operations for distributed architectures. A significant reduction in the wallclock time for an MD timestep is necessary to achieve practical large scale QMD. Equally challenging is to develop scalable distributed implementations. This requires the exploration of efficient data formats (dense, sparse, etc.), data decomposition and distribution across compute nodes using block- and graph-based methods. Traditional programming models such as MPI/OpenMP, hybrid MPI/OpenMP/Accelerator, as well as asynchronous task-based models will be explored to achieve maximum parallelism. Parallelism at the level of a matrix operation, solver, and/or application needs to be explored to gain the most benefit on new and emerging architectures. This report serves as a “living document” of the additions and enhancements throughout the ECP CoPA project for $\mathcal{O}(N)$ QMD. The following sections address the applications we are working with, the quantum chemistry libraries (BML and PROGRESS), and our co-design exploration through the use of a proxy application.

36 MATERIALS SCIENCE↗

NASA NTRS1996

Fundamental Phenomena on Fuel Decomposition and Boundary-Layer Combustion Precesses with Applications to Hybrid Rocket Motors: Experimental Investigation - Part 1

This final report summarizes the major findings on the subject of 'Fundamental Phenomena on Fuel Decomposition and Boundary-Layer Combustion Processes with Applications to Hybrid Rocket Motors', performed from 1 April 1994 to 30 June 1996. Both experimental results from Task 1 and theoretical/numerical results from Task 2 are reported here in two parts. Part 1 covers the experimental work performed and describes the test facility setup, data reduction techniques employed, and results of the test firings, including effects of operating conditions and fuel additives on solid fuel regression rate and thermal profiles of the condensed phase. Part 2 concerns the theoretical/numerical work. It covers physical modeling of the combustion processes including gas/surface coupling, and radiation effect on regression rate. The numerical solution of the flowfield structure and condensed phase regression behavior are presented. Experimental data from the test firings were used for numerical model validation.

Kuo, Kenneth K.↗

DOE OSTI2023

Scaling Resolution of Gigapixel Whole Slide Images Using Spatial Decomposition on Convolutional Neural Networks

Gigapixel images are prevalent in scientific domains ranging from remote sensing, and satellite imagery to microscopy, etc. However, training a deep learning model at the natural resolution of those images has been a challenge in terms of both, overcoming the resource limit (e.g. HBM memory constraints), as well as scaling up to a large number of GPUs. In this paper, we trained Residual neural Networks (ResNet) on 22,528 x 22,528-pixel size images using a distributed spatial decomposition method on 2,304 GPUs on the Summit Supercomputer. We applied our method on a Whole Slide Imaging (WSI) dataset from The Cancer Genome Atlas (TCGA) database. WSI images can be in the size of 100,000 x 100,000 pixels or even larger, and in this work we studied the effect of image resolution on a classification task, while achieving state-of-the-art AUC scores. Moreover, our approach doesn't need pixel-level labels, since we're avoiding patching from the WSI images completely, while adding the capability of training arbitrary large-size images. This is achieved through a distributed spatial decomposition method, by leveraging the non-block fat-tree interconnect network of the Summit architecture, which enabled GPU-to-GPU direct communication. Finally, detailed performance analysis results are shown, as well as a comparison with a data-parallel approach when possible.

Tsaris, Aristeidis (aris)↗

DOE OSTI2019

International Collaboration Framework for the Calculation of Performance Loss Rates: Data Quality, Benchmarks, and Trends

Within the IEA PVPS Task 13, a group of experts representing several leading R&D centres, universities and industry companies, is developing a framework for the calculation of Performance Loss Rates (PLR) on a large number of commercial and research PV power plants and related weather data coming from various climatic zones. Various methodologies are applied for the calculation of PLR which are benchmarked in terms of uncertainties and 'true' values. The aim of the international collaboration is to show how to calculate the PLR on high quality (high time resolution, reliable data, irradiance, yield, etc.) and on low quality data (low time resolution, only energy data available). Various algorithms and models, along with different time averaging and filtering criteria, can be applied for the calculation of PLR each of which can have an impact on the results. The approach considers three pathways to ensure broad collaboration and increase the statistical relevance of the study: i) use of shared methodologies on shared time series, ii) use of confidential methodologies on shared time series, iii) use of shared methodologies on confidential time series. The data is used for benchmarking activities and to define which methodologies clusters around a 'true value' of PLR. The combination of metrics (PR or power based) and methodologies are benchmarked in terms of deviation from the average value and in terms of standard deviation. The combination of temperature corrected PR with the use of Year on Year (YOY) or Seasonal Trend decomposition using Loess (STL) performs well for all systems; other methodologies seems to be more depending on the filtering applied. Another set of data is represented by the IEA PVPS Task 13 'PV performance database' which includes more than 120 PV plants from different climates. The applied methodologies were STL and the YoY approach. STL results in an averaged PLR over all systems of -0.78%/a while YoY yields -0.63%/a. STL is better suited if the time series data are of higher resolution and high-quality weather data are available. One important outcome of this study is the setting of a framework for the contribution to a clear and structured quality classification for PV time series data and a guideline for PLR calculations in dependency of the data categorization.

14 SOLAR ENERGY↗

DOE OSTI2021

Multilevel Hierarchical Decomposition of Finite Element White Noise with Application to Multilevel Markov Chain Monte Carlo

In this work we develop a new hierarchical multilevel approach to generate Gaussian random field realizations in an algorithmically scalable manner that is well suited to incorporating into multilevel Markov chain Monte Carlo (MCMC) algorithms. This approach builds off of other partial differential equation (PDE) approaches for generating Gaussian random field realizations; in particular, a single field realization may be formed by solving a reaction-diffusion PDE with a spatial white noise source function as the right-hand side. While these approaches have been explored to accelerate forward uncertainty quantification tasks, e.g., multilevel Monte Carlo, the previous constructions are not directly applicable to multilevel MCMC frameworks which build fine-scale random fields in a hierarchical fashion from coarse-scale random fields. Our new hierarchical multilevel method relies on a hierarchical decomposition of the white noise source function in $L^2$ which allows us to form Gaussian random field realizations across multiple levels of discretization in a way that fits into multilevel MCMC algorithmic frameworks. After presenting our main theoretical results and numerical scaling results to showcase the utility of this new hierarchical PDE method for generating Gaussian random field realizations, this method is tested on a four-level MCMC algorithm to explore its feasibility.

algebraic multigrid↗

DOE OSTI2024

Random projection using random quantum circuits

The random sampling task performed by Google's Sycamore processor gave us a glimpse of the “quantum supremacy era.” This has definitely shed some light on the power of random quantum circuits in this abstract task of sampling outputs from the (pseudo)random circuits. In this paper, we explore a practical near-term use of local random quantum circuits in dimensional reduction of large low-rank data sets. We make use of the well-studied dimensionality reduction technique called the random projection method. This method has been extensively used in various applications such as image processing, logistic regression, entropy computation of low-rank matrices, etc. We prove that the matrix representations of local random quantum circuits with sufficiently shorter depths [ ∼ O ( n ) ] serve as good candidates for random projection. We demonstrate numerically that their projection abilities are not far off from the computationally expensive classical principal components analysis on MNIST and CIFAR-100 image datasets. We also benchmark the performance of quantum random projection against the commonly used classical random projection in the tasks of dimensionality reduction of image data sets and computing von Neumann entropies of large low-rank density matrices. And finally, using variational quantum singular value decomposition, we demonstrate a near-term implementation of extracting the singular vectors with dominant singular values after quantum random projecting a large low-rank matrix to lower dimensions. All such numerical experiments unequivocally demonstrate the ability of local random circuits to randomize a large Hilbert space at sufficiently shorter depths with robust retention of properties of large data sets in reduced dimensions. Published by the American Physical Society 2024

Kumaran, Keerthi (ORCID:0009000949125721)↗

NASA NTRS1991

Aerospace plane guidance using time-scale decomposition - A geometric approach

A method is proposed for developing the necessary guidance logic to steer single-stage vehicles into orbit. The minimum-fuel ascent problem is first considered to analyze the effects of dynamic pressure, acceleration, and heating constraints on guidance systems to thereby develop the guidance logic. The optimal solution consists of behavior with two time scales, and the control law is used to develop near-optimal guidance. The solution uses the slow manifold to delineate the control for minimum-fuel reduced-order trajectory and a separate control for tracking the optimal reduced-order trajectory. A family of fast manifolds is then employed to resolve the tracking problem via the feedback linearization methodology from nonlinear geometric control theory. The two-time-scale decomposition is found to produce a near-optimal ascent by tracking the applicable state-constraint boundary, as well as to simplify the control-design task.

Van Buren, Mark A.↗

NASA NTRS1984

On the identification of normal modes of oscillation from observations of the solar periphery

The decomposition of solar oscillations into their constituent normal modes requires a knowledge of both the spatial and temporal variation of the perturbation to the Sun's surface. The task is especially difficult when only limited spatial information is available. Observations of the limb darkening function, for example, are probably sensitive to too large a number of modes to permit most of the modes to be identified in a power spectrum of measurements at only a few points on the limb, unless the results are combined with other data. A procedure was considered by which the contributions from quite small groups of modes to spatially well resolved data obtained at any instant can be extracted from the remaining modes. Combining these results with frequency information then permits the modes to be identified, at least if their frequencies are low enough to ensure that modes of high degree do not contribute substantially to the signal.

Gough, D. D.↗

NASA NTRS1984

On the identification of normal modes of oscillation from observations of the solar periphery

The decomposition of solar oscillations into their constituent normal modes requires a knowledge of both the spatial and temporal variation of the perturbation to the sun's surface. The task can be especially difficult when only limited spatial information is available. Observations of the limb-darkening function, for example, are probably sensitive to too large a number of modes to permit most of the modes to be identified in a power spectrum of measurements at only a few points on the limb, unless the results are combined with other data. In this paper a procedure is considered by which the contributions from quite small groups of modes to spatially well resolved data obtained at any instant can be extracted from the remaining modes. Combining these results with frequency information then permits the modes to be identified, at least if their frequencies are low enough to ensure that modes of high degree do not contribute substantially to the signal.

Gough, D. O.↗

NASA NTRS2015

A Structural Model Decomposition Framework for Hybrid Systems Diagnosis

Nowadays, a large number of practical systems in aerospace and industrial environments are best represented as hybrid systems that consist of discrete modes of behavior, each defined by a set of continuous dynamics. These hybrid dynamics make the on-line fault diagnosis task very challenging. In this work, we present a new modeling and diagnosis framework for hybrid systems. Models are composed from sets of user-defined components using a compositional modeling approach. Submodels for residual generation are then generated for a given mode, and reconfigured efficiently when the mode changes. Efficient reconfiguration is established by exploiting causality information within the hybrid system models. The submodels can then be used for fault diagnosis based on residual generation and analysis. We demonstrate the efficient causality reassignment, submodel reconfiguration, and residual generation for fault diagnosis using an electrical circuit case study.

structural model decomposition↗

DOE OSTI2022

Towards Compact Neural Networks via End-to-End Training: A Bayesian Tensor Approach with Automatic Rank Determination

Post-training model compression can reduce the inference costs of deep neural networks, but uncompressed training still consumes enormous hardware resources and energy. To enable low-energy training on edge devices, it is highly desirable to directly train a compact neural network from scratch with a low memory cost. Low-rank tensor decomposition is an effective approach to reduce the memory and computing costs of large neural networks. However, directly training low-rank tensorized neural networks is a very challenging task because it is hard to determine a proper tensor rank a priori, and the tensor rank controls both model complexity and accuracy. Here, this paper presents a novel end-to-end framework for low-rank tensorized training. We first develop a Bayesian model that supports various low-rank tensor formats (e.g., CANDECOMP/PARAFAC, Tucker, tensor-train, and tensor-train matrix) and reduces neural network parameters with automatic rank determination during training. Then we develop a customized Bayesian solver to train large-scale tensorized neural networks. Our training methods shows orders-of-magnitude parameter reduction and little accuracy loss (or even better accuracy) in the experiments. On a very large deep learning recommendation system with over 4.2 ×10 9 model parameters, our method can reduce the parameter number to 1.6 ×10 5 automatically in the training process (i.e., by 2.6 ×10 4 times) while achieving almost the same accuracy. Code is available at https://github.com/colehawkins/bayesian-tensor-rank-determination.

compact neural networks↗

NASA NTRS1992

S3D: An interactive surface grid generation tool

S3D, an interactive software tool for surface grid generation, is described. S3D provides the means with which a geometry definition based either on a discretized curve set or a rectangular set can be quickly processed towards the generation of a surface grid for computational fluid dynamics (CFD) applications. This is made possible as a result of implementing commonly encountered surface gridding tasks in an environment with a highly efficient and user friendly graphical interface. Some of the more advanced features of S3D include surface-surface intersections, optimized surface domain decomposition and recomposition, and automated propagation of edge distributions to surrounding grids.

Luh, Raymond Ching-Chung↗

DOE OSTI2022

Quantum annealing algorithms for Boolean tensor networks

Abstract Quantum annealers manufactured by D-Wave Systems, Inc., are computational devices capable of finding high-quality heuristic solutions of NP-hard problems. In this contribution, we explore the potential and effectiveness of such quantum annealers for computing Boolean tensor networks. Tensors offer a natural way to model high-dimensional data commonplace in many scientific fields, and representing a binary tensor as a Boolean tensor network is the task of expressing a tensor containing categorical (i.e., $$\{0, 1\}$$ { 0 , 1 } ) values as a product of low dimensional binary tensors. A Boolean tensor network is computed by Boolean tensor decomposition, and it is usually not exact. The aim of such decomposition is to minimize the given distance measure between the high-dimensional input tensor and the product of lower-dimensional (usually three-dimensional) tensors and matrices representing the tensor network. In this paper, we introduce and analyze three general algorithms for Boolean tensor networks: Tucker, Tensor Train, and Hierarchical Tucker networks. The computation of a Boolean tensor network is reduced to a sequence of Boolean matrix factorizations, which we show can be expressed as a quadratic unconstrained binary optimization problem suitable for solving on a quantum annealer. By using a novel method we introduce called parallel quantum annealing, we demonstrate that Boolean tensor’s with up to millions of elements can be decomposed efficiently using a DWave 2000Q quantum annealer.

97 MATHEMATICS AND COMPUTING↗