Search NASA⌕ Search

SEARCH · Search NASA

Results for “sparse”

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 1,081 records · Page 60

A Method of Estimating Sparse and Doubly-Dispersive Channels

The large delay and Doppler spreads of skywave high-frequency (HF) channels complicates channel acquisition, especially in low-SNR conditions. Using pilot-symbol assisted modulation, we develop a robust method to acquire channel information by first estimating the power-delay profile (PDP) of the channel, then applying that estimate to obtain the channel coefficients only where the PDP is nonzero. We find the presented approach to be robust at low-SNRs, making it suitable for spread-spectrum and underlay waveforms. We show that this two-stage channel acquisition procedure has three advantages: i),the channel is estimated with fewer terms, reducing complexity; ii), the MSE of the channel estimate is lower than assuming a fixed channel length; and iii), each propagation mode is reliably recovered in high-Doppler conditions. We conclude this paper by presenting results of the developed techniques as applied to a filter-bank multicarrier spread-spectrum (FBMC-SS) waveform.

42 ENGINEERING↗

Building Alternative Indices of Socioeconomic Status for Population Modeling in Data-Sparse Contexts (Short Paper)

Population modeling requires clear definitions of socioeconomic status (SES) to ensure overall estimate accuracy and locate potentially underserved subpopulations. This presents a challenge as SES can be measured in myriad ways and for divergent purposes, and the data required to calculate these metrics may be lacking, particularly in low and middle income countries (LMICs). To support more refined SES measurement, we explore improvements upon the Demographic and Health Survey’s (DHS) Wealth Index (DHS-WI) using alternative characterizations of SES based on multiple correspondence analysis (MCA) and hierarchical clustering. We produce the MCA-derived metrics first on a full suite of household economic, demographic, and dwelling variables, then on a reduced set of occupant-only SES characteristics. We explore the utility of these metrics relative to DHS-WI based on their ability to 1) differentiate DHS household types and 2) identify mixtures of SES levels within DHS samples and mapped at high spatial resolution. We find that our full suite MCA yields more clearly defined SES segments and that our reduced MCA delineates occupant SES most clearly, suggesting potential pathways to improve upon the DHS-WI in future population modeling efforts for LMICs.

Cunningham, Angela↗

Application of adaptive hierarchical sparse grid collocation to the uncertainty quantification of nuclear reactor simulators

Recent efforts in the application of uncertainty quantification to nuclear systems have utilized methods based on generalized perturbation theory and stochastic sampling. While these methods have proven to be effective they both have major drawbacks that may impede further progress. A relatively new approach based on spectral elements for uncertainty quantification is applied in this paper to several problems in reactor simulation. Spectral methods based on collocation attempt to couple the approximation free nature of stochastic sampling methods with the determinism of generalized perturbation theory. The specific spectral method used in this paper employs both the Smolyak algorithm and adaptivity by using Newton-Cotes collocation points along with linear hat basis functions. Using this approach, a surrogate model for the outputs of a computer code is constructed hierarchically by adaptively refining the collocation grid until the interpolant is converged to a user-defined threshold. The method inherently fits into the framework of parallel computing and allows for the extraction of meaningful statistics and data that are not within reach of stochastic sampling and generalized perturbation theory. This paper aims to demonstrate the advantages of spectral methods-especially when compared to current methods used in reactor physics for uncertainty quantification-and to illustrate their full potential. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Amesos Solver Package

Amesos is the Direct Sparse Solver Package in Trilinos. The goal of Amesos is to make AX=S as easy as it sounds, at least for direct methods. Amesos provides interfaces to a number of third party sparse direct solvers, including SuperLU, SuperLU MPI, DSCPACK, UMFPACK and KLU. Amesos provides a common object oriented interface to the best sparse direct solvers in the world. A sparse direct solver solves for x in Ax = b. where A is a matrix and x and b are vectors (or multi-vectors). A sparse direct solver flrst factors A into trinagular matrices L and U such that A = LU via gaussian elimination and then solves LU x = b. Switching amongst solvers in Amesos roquires a change to a single parameter. Yet, no solver needs to be linked it, unless it is used. All conversions between the matrices provided by the user and the format required by the underlying solver is performed by Amesos. As new sparse direct solvers are created, they will be incorporated into Amesos, allowing the user to simpty link with the new solver, change a single parameter in the calling sequence, and use the new solver. Amesos allows users to specify whether the matrix has changed. Amesos can be used anywhere that any sparse direct solver is needed.

SciDAC↗

Reducing Communication in Graph Neural Network Training

Graph Neural Networks (GNNs) are powerful and flexible neural networks that use the naturally sparse connectivity information of the data. GNNs represent this connectivity as sparse matrices, which have lower arithmetic intensity and thus higher communication costs compared to dense matrices, making GNNs harder to scale to high concurrencies than convolutional or fully-connected neural networks. Here, we introduce a family of parallel algorithms for training GNNs and show that they can asymptotically reduce communication compared to previous parallel GNN training methods. We implement these algorithms, which are based on 1D, 1. 5D, 2D, and 3D sparse-dense matrix multiplication, using torch.distributed on GPU-equipped clusters. Our algorithms optimize communication across the full GNN training pipeline. We train GNNs on over a hundred GPUs on multiple datasets, including a protein network with over a billion edges.

97 MATHEMATICS AND COMPUTING↗

Towards adaptive spiking label propagation

Graph algorithms are a new class of applications for neuromorphic hardware. Rather than adapting deep learning and standard neural network approaches to a low-precision spiking environment, we use spiking neurons to analyze undirected graphs (e.g., the underlying modular structure). While fully connected spin glass implementations of spiking label propagation have shown promising results on graphs with dense communities, identifying sparse communities remains difficult. This work focuses on steps towards an adaptive spike-based implementations of label propagation, utilizing sparse embeddings and synaptic plasticity. Sparser embeddings reduce the number of inhibitory connections, and synaptic plasticity is used to simultaneously amplify spike responses between neurons in the same community, while impeding spike responses across different communities. We present results on identifying communities in sparse graphs with very small communities.

Hamilton, Kathleen↗

Dictionary Learning with Accumulator Neurons

The Locally Competitive Algorithm (LCA) uses local competition between non-spiking leaky integrator neurons to infer sparse representations, allowing for potentially real-time execution on massively parallel neuromorphic architectures such as Intel's Loihi processor. Here, we focus on the problem of inferring sparse representations from streaming video using dictionaries of spatiotemporal features optimized in an unsupervised manner for sparse reconstruction. Non-spiking LCA has previously been used to achieve unsupervised learning of spatiotemporal dictionaries composed of convolutional kernels from raw, unlabeled video. We demonstrate how unsupervised dictionary learning with spiking LCA (\hbox{S-LCA}) can be efficiently implemented using accumulator neurons, which combine a conventional leaky-integrate-and-fire (\hbox{LIF}) spike generator with an additional state variable that is used to minimize the difference between the integrated input and the spiking output. We demonstrate dictionary learning across a wide range of dynamical regimes, from graded to intermittent spiking, for inferring sparse representations of both static images drawn from the CIFAR database as well as video frames captured from a DVS camera. On a classification task that requires identification of the suite from a deck of cards being rapidly flipped through as viewed by a DVS camera, we find essentially no degradation in performance as the LCA model used to infer sparse spatiotemporal representations migrates from graded to spiking. We conclude that accumulator neurons are likely to provide a powerful enabling component of future neuromorphic hardware for implementing online unsupervised learning of spatiotemporal dictionaries optimized for sparse reconstruction of streaming video from event based DVS cameras.

artificial intelligence↗

Efficient Implementation of an Optimal Interpolator for Large Spatial Data Sets

Scattered data interpolation is a problem of interest in numerous areas such as electronic imaging, smooth surface modeling, and computational geometry. Our motivation arises from applications in geology and mining, which often involve large scattered data sets and a demand for high accuracy. The method of choice is ordinary kriging. This is because it is a best unbiased estimator. Unfortunately, this interpolant is computationally very expensive to compute exactly. For n scattered data points, computing the value of a single interpolant involves solving a dense linear system of size roughly n x n. This is infeasible for large n. In practice, kriging is solved approximately by local approaches that are based on considering only a relatively small'number of points that lie close to the query point. There are many problems with this local approach, however. The first is that determining the proper neighborhood size is tricky, and is usually solved by ad hoc methods such as selecting a fixed number of nearest neighbors or all the points lying within a fixed radius. Such fixed neighborhood sizes may not work well for all query points, depending on local density of the point distribution. Local methods also suffer from the problem that the resulting interpolant is not continuous. Meyer showed that while kriging produces smooth continues surfaces, it has zero order continuity along its borders. Thus, at interface boundaries where the neighborhood changes, the interpolant behaves discontinuously. Therefore, it is important to consider and solve the global system for each interpolant. However, solving such large dense systems for each query point is impractical. Recently a more principled approach to approximating kriging has been proposed based on a technique called covariance tapering. The problems arise from the fact that the covariance functions that are used in kriging have global support. Our implementations combine, utilize, and enhance a number of different approaches that have been introduced in literature for solving large linear systems for interpolation of scattered data points. For very large systems, exact methods such as Gaussian elimination are impractical since they require 0(n(exp 3)) time and 0(n(exp 2)) storage. As Billings et al. suggested, we use an iterative approach. In particular, we use the SYMMLQ method, for solving the large but sparse ordinary kriging systems that result from tapering. The main technical issue that need to be overcome in our algorithmic solution is that the points' covariance matrix for kriging should be symmetric positive definite. The goal of tapering is to obtain a sparse approximate representation of the covariance matrix while maintaining its positive definiteness. Furrer et al. used tapering to obtain a sparse linear system of the form Ax = b, where A is the tapered symmetric positive definite covariance matrix. Thus, Cholesky factorization could be used to solve their linear systems. They implemented an efficient sparse Cholesky decomposition method. They also showed if these tapers are used for a limited class of covariance models, the solution of the system converges to the solution of the original system. Matrix A in the ordinary kriging system, while symmetric, is not positive definite. Thus, their approach is not applicable to the ordinary kriging system. Therefore, we use tapering only to obtain a sparse linear system. Then, we use SYMMLQ to solve the ordinary kriging system. We show that solving large kriging systems becomes practical via tapering and iterative methods, and results in lower estimation errors compared to traditional local approaches, and significant memory savings compared to the original global system. We also developed a more efficient variant of the sparse SYMMLQ method for large ordinary kriging systems. This approach adaptively finds the correct local neighborhood for each query point in the interpolation process.

Memarsadeghi, Nargess↗

Group‐sparsity regularization for ill‐posed subsurface flow inverse problems

Abstract Sparse representations provide a flexible and parsimonious description of high‐dimensional model parameters for reconstructing subsurface flow property distributions from limited data. To further constrain ill‐posed inverse problems, group‐sparsity regularization can take advantage of possible relations among the entries of unknown sparse parameters when: (i) groups of sparse elements are either collectively active or inactive and (ii) only a small subset of the groups is needed to approximate the parameters of interest. Since subsurface properties exhibit strong spatial connectivity patterns they may lead to sparse descriptions that satisfy the above conditions. When these conditions are established, a group‐sparsity regularization can be invoked to facilitate the solution of the resulting inverse problem by promoting sparsity across the groups. The proposed regularization penalizes the number of groups that are active without promoting sparsity within each group. Two implementations are presented in this paper: one based on the multiresolution tree structure of Wavelet decomposition, without a need for explicit prior models, and another learned from explicit prior model realizations using sparse principal component analysis (SPCA). In each case, the approach first classifies the parameters of the inverse problem into groups with specific connectivity features, and then takes advantage of the grouped structure to recover the relevant patterns in the solution from the flow data. Several numerical experiments are presented to demonstrate the advantages of additional constraining power of group‐sparsity in solving ill‐posed subsurface model calibration problems.

Golmohammadi, Azarang↗

Fast and low-dose computed laminography using compressive sensing based technique

Computed laminography (CL) is well known for inspecting microstructures in the materials, weldments and soldering defects in high density packed components or multilayer printed circuit boards. The overload problem on x-ray tube and gross failure of the radio-sensitive electronics devices during a scan are among important issues in CL which needs to be addressed. The sparse-view CL can be one of the viable option to overcome such issues. In this work a numerical aluminum welding phantom was simulated to collect sparsely sampled projection data at only 40 views using a conventional CL scanning scheme i.e. oblique scan. A compressive-sensing inspired total-variation (TV) minimization algorithm was utilized to reconstruct the images. It is found that the images reconstructed using sparse view data are visually comparable with the images reconstructed using full scan data set i.e. at 360 views on regular interval. We have quantitatively confirmed that tiny structures such as copper and tungsten slags, and copper flakes in the reconstructed images from sparsely sampled data are comparable with the corresponding structure present in the fully sampled data case. A blurring effect can be seen near the edges of few pores at the bottom of the reconstructed images from sparsely sampled data, despite the overall image quality is reasonable for fast and low-dose NDT.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Particle hit clustering and identification using point set transformers in liquid argon time projection chambers

Liquid argon time projection chambers are often used in neutrino physics and dark-matter searches because of their high spatial resolution. The images generated by these detectors are extremely sparse, as the energy values detected by most of the detector are equal to 0, meaning that despite their high resolution, most of the detector is unused in a particular interaction. Instead of representing all of the empty detections, the interaction is usually stored as a sparse matrix, a list of detection locations paired with their energy values. Traditional machine learning methods that have been applied to particle reconstruction such as convolutional neural networks (CNNs), however, cannot operate over data stored in this way and therefore must have the matrix fully instantiated as a dense matrix. Operating on dense matrices requires a lot of memory and computation time, in contrast to directly operating on the sparse matrix. We propose a machine learning model using a point set neural network that operates over a sparse matrix, greatly improving both processing speed and accuracy over methods that instantiate the dense matrix, as well as over other methods that operate over sparse matrices. Compared to competing state-of-the-art methods, our method improves classification performance by 14%, segmentation performance by more than 22%, while taking 80% less time and using 66% less memory. Compared to state-of-the-art CNN methods, our method improves classification performance by more than 86%, segmentation performance by more than 71%, while reducing runtime by 91% and reducing memory usage by 61%.

calibration and fitting methods↗