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 1,369 records · Page 76

SU-E-I-41: Dictionary Learning Based Quantitative Reconstruction for Low-Dose Dual-Energy CT (DECT)

Purpose: DECT collects two sets of projection data under higher and lower energies. With appropriates composition methods on linear attenuation coefficients, quantitative information about the object, such as density, can be obtained. In reality, one of the important problems in DECT is the radiation dose due to doubled scans. This work is aimed at establishing a dictionary learning based reconstruction framework for DECT for improved image quality while reducing the imaging dose. Methods: In our method, two dictionaries were learned respectively from the high-energy and lowenergy image datasets of similar objects under normal dose in advance. The linear attenuation coefficient was decomposed into two basis components with material based composition method. An iterative reconstruction framework was employed. Two basis components were alternately updated with DECT datasets and dictionary learning based sparse constraints. After one updating step under the dataset fidelity constraints, both high-energy and low-energy images can be obtained from the two basis components. Sparse constraints based on the learned dictionaries were applied to the high- and low-energy images to update the two basis components. The iterative calculation continues until a pre-set number of iteration was reached. Results: We evaluated the proposed dictionary learning method with dual energy images collected using a DECT scanner. We re-projected the projection data with added Poisson noise to reflect the low-dose situation. The results obtained by the proposed method were compared with that obtained using FBP based method and TV based method. It was found that the proposed approach yield better results than other methods with higher resolution and less noise. Conclusion: The use of dictionary learned from DECT images under normal dose is valuable and leads to improved results with much lower imaging dose.

60 APPLIED LIFE SCIENCES↗

WE-AB-BRA-11: Development of Physiological-Based Virtual Thorax Phantoms for Evaluation of Deformable Image Registration

Purpose: To develop a methodology of constructing physiological-based virtual thorax phantom based on hyperpolarized (HP) gas tagging MRI for evaluating deformable image registration (DIR). Methods: Three healthy subjects were imaged at both the end-of-inhalation (EOI) and the end-of-exhalation (EOE) phases using a high-resolution (2.5mm isovoxel) 3D proton MRI, as well as a hybrid MRI which combines HP gas tagging MRI and a low-resolution (4.5mm isovoxel) proton MRI. A sparse tagging displacement vector field (tDVF) was derived from the HP gas tagging MRI by tracking the displacement of tagging grids between EOI and EOE. Using the tDVF and the high-resolution MR images, we determined the motion model of the entire thorax in the following two steps: 1) the DVF inside of lungs was estimated based on the sparse tDVF using a novel multi-step natural neighbor interpolation method; 2) the DVF outside of lungs was estimated from the DIR between the EOI and EOE images (Velocity AI). The derived motion model was then applied to the high-resolution EOI image to create a deformed EOE image, forming the virtual phantom where the motion model provides the ground truth of deformation. Five DIR methods were evaluated using the developed virtual phantom. Errors in DVF magnitude (Em) and angle (Ea) were determined and compared for each DIR method. Results: Among the five DIR methods, free form deformation produced DVF results that are most closely resembling the ground truth (Em=1.04mm, Ea=6.63°). The two DIR methods based on B-spline produced comparable results (Em=2.04mm, Ea=13.66°; and Em =2.62mm, Ea=17.67°), and the two optical-flow methods produced least accurate results (Em=7.8mm; Ea=53.04°; Em=4.45mm, Ea=31.02°). Conclusion: A methodology for constructing physiological-based virtual thorax phantom based on HP gas tagging MRI has been developed. Initial evaluation demonstrated its potential as an effective tool for robust evaluation of DIR in the lung.

60 APPLIED LIFE SCIENCES↗

Compressed-sensing-based content-driven hierarchical reconstruction: Theory and application to C-arm cone-beam tomography

Purpose: This paper addresses the reconstruction of x-ray cone-beam computed tomography (CBCT) for interventional C-arm systems. Subsampling of CBCT is a significant issue with C-arms due to their slow rotation and to the low frame rate of their flat panel x-ray detectors. The aim of this work is to propose a novel method able to handle the subsampling artifacts generally observed with analytical reconstruction, through a content-driven hierarchical reconstruction based on compressed sensing. Methods: The central idea is to proceed with a hierarchical method where the most salient features (high intensities or gradients) are reconstructed first to reduce the artifacts these features induce. These artifacts are addressed first because their presence contaminates less salient features. Several hierarchical schemes aiming at streak artifacts reduction are introduced for C-arm CBCT: the empirical orthogonal matching pursuit approach with the ℓ{sub 0} pseudonorm for reconstructing sparse vessels; a convex variant using homotopy with the ℓ{sub 1}-norm constraint of compressed sensing, for reconstructing sparse vessels over a nonsparse background; homotopy with total variation (TV); and a novel empirical extension to nonlinear diffusion (NLD). Such principles are implemented with penalized iterative filtered backprojection algorithms. For soft-tissue imaging, the authors compare the use of TV and NLD filters as sparsity constraints, both optimized with the alternating direction method of multipliers, using a threshold for TV and a nonlinear weighting for NLD. Results: The authors show on simulated data that their approach provides fast convergence to good approximations of the solution of the TV-constrained minimization problem introduced by the compressed sensing theory. Using C-arm CBCT clinical data, the authors show that both TV and NLD can deliver improved image quality by reducing streaks. Conclusions: A flexible compressed-sensing-based algorithmic approach is proposed that is able to accommodate for a wide range of constraints. It is successfully applied to C-arm CBCT images that may not be so well approximated by piecewise constant functions.

60 APPLIED LIFE SCIENCES↗

SU-G-TeP1-15: Toward a Novel GPU Accelerated Deterministic Solution to the Linear Boltzmann Transport Equation

Purpose: To develop a Graphic Processor Unit (GPU) accelerated deterministic solution to the Linear Boltzmann Transport Equation (LBTE) for accurate dose calculations in radiotherapy (RT). A deterministic solution yields the potential for major speed improvements due to the sparse matrix-vector and vector-vector multiplications and would thus be of benefit to RT. Methods: In order to leverage the massively parallel architecture of GPUs, the first order LBTE was reformulated as a second order self-adjoint equation using the Least Squares Finite Element Method (LSFEM). This produces a symmetric positive-definite matrix which is efficiently solved using a parallelized conjugate gradient (CG) solver. The LSFEM formalism is applied in space, discrete ordinates is applied in angle, and the Multigroup method is applied in energy. The final linear system of equations produced is tightly coupled in space and angle. Our code written in CUDA-C was benchmarked on an Nvidia GeForce TITAN-X GPU against an Intel i7-6700K CPU. A spatial mesh of 30,950 tetrahedral elements was used with an S4 angular approximation. Results: To avoid repeating a full computationally intensive finite element matrix assembly at each Multigroup energy, a novel mapping algorithm was developed which minimized the operations required at each energy. Additionally, a parallelized memory mapping for the kronecker product between the sparse spatial and angular matrices, including Dirichlet boundary conditions, was created. Atomicity is preserved by graph-coloring overlapping nodes into separate kernel launches. The one-time mapping calculations for matrix assembly, kronecker product, and boundary condition application took 452±1ms on GPU. Matrix assembly for 16 energy groups took 556±3s on CPU, and 358±2ms on GPU using the mappings developed. The CG solver took 93±1s on CPU, and 468±2ms on GPU. Conclusion: Three computationally intensive subroutines in deterministically solving the LBTE have been formulated on GPU, resulting in two orders of magnitude speedup. Funding support from Natural Sciences and Engineering Research Council and Alberta Innovates Health Solutions. Dr. Fallone is a co-founder and CEO of MagnetTx Oncology Solutions (under discussions to license Alberta bi-planar linac MR for commercialization).

60 APPLIED LIFE SCIENCES↗

MO-DE-207A-05: Dictionary Learning Based Reconstruction with Low-Rank Constraint for Low-Dose Spectral CT

Purpose: Spectral CT enabled by an energy-resolved photon-counting detector outperforms conventional CT in terms of material discrimination, contrast resolution, etc. One reconstruction method for spectral CT is to generate a color image from a reconstructed component in each energy channel. However, given the radiation dose, the number of photons in each channel is limited, which will result in strong noise in each channel and affect the final color reconstruction. Here we propose a novel dictionary learning method for spectral CT that combines dictionary-based sparse representation method and the patch based low-rank constraint to simultaneously improve the reconstruction in each channel and to address the inter-channel correlations to further improve the reconstruction. Methods: The proposed method has two important features: (1) guarantee of the patch based sparsity in each energy channel, which is the result of the dictionary based sparse representation constraint; (2) the explicit consideration of the correlations among different energy channels, which is realized by patch-by-patch nuclear norm-based low-rank constraint. For each channel, the dictionary consists of two sub-dictionaries. One is learned from the average of the images in all energy channels, and the other is learned from the average of the images in all energy channels except the current channel. With average operation to reduce noise, these two dictionaries can effectively preserve the structural details and get rid of artifacts caused by noise. Combining them together can express all structural information in current channel. Results: Dictionary learning based methods can obtain better results than FBP and the TV-based method. With low-rank constraint, the image quality can be further improved in the channel with more noise. The final color result by the proposed method has the best visual quality. Conclusion: The proposed method can effectively improve the image quality of low-dose spectral CT. This work is partially supported by the National Natural Science Foundation of China (No. 61302136), and the Natural Science Basic Research Plan in Shaanxi Province of China (No. 2014JQ8317).

60 APPLIED LIFE SCIENCES↗

Structural evidence for a dynamic metallocofactor during N 2 reduction by Mo-nitrogenase

The enzyme nitrogenase uses a suite of complex metallocofactors to reduce dinitrogen (N 2 ) to ammonia. Mechanistic details of this reaction remain sparse. We report a 1.83-angstrom crystal structure of the nitrogenase molybdenum-iron (MoFe) protein captured under physiological N 2 turnover conditions. This structure reveals asymmetric displacements of the cofactor belt sulfurs (S2B or S3A and S5A) with distinct dinitrogen species in the two αβ dimers of the protein. The sulfur-displaced sites are distinct in the ability of protein ligands to donate protons to the bound dinitrogen species, as well as the elongation of either the Mo–O5 (carboxyl) or Mo–O7 (hydroxyl) distance that switches the Mo-homocitrate ligation from bidentate to monodentate. These results highlight the dynamic nature of the cofactor during catalysis and provide evidence for participation of all belt-sulfur sites in this process.

Science & Technology - Other Topics↗

Fast and effective reordering of columns within supernodes using partition refinement

In this paper, we consider the problem of computing a triangular factorization of a sparse symmetric matrix using Gaussian elimination. We assume that the sparse matrix has been permuted using a fill-reducing ordering algorithm. When the matrix is symmetric positive definite, the sparsity structure of the triangular factor can be determined once the fill-reducing ordering has been computed. Thus, an efficient numerical factorization scheme can be designed so that only the nonzero entries are stored and operated on. On modern architectures, the positions of the nonzero entries in the triangular factor play a big role in determining the efficiency. It is desirable to have dense blocks in the factor so that the computation can be cast in terms of level-3 BLAS as much as possible. On architectures with GPUs, for example, it is also desirable for these dense blocks to be as large as possible in order to reduce the times to transfer data between the main CPU and the GPUs. We address the problem of locally refining the ordering so that the number of dense blocks is reduced and the sizes of these dense blocks are increased in the triangular factor.

Jacquelin, Mathias↗

Reducing Parallel Communication in Algebraic Multigrid through Sparsification

Algebraic multigrid (AMG) is an $\mathcal{O}(n)$ solution process for many large sparse linear systems. A hierarchy of progressively coarser grids which utilize complementary relaxation and interpolation operators is constructed. High-energy error is reduced by relaxation, while low-energy error is mapped to coarse-grid matrices and reduced there. However, large parallel communication costs often limit parallel scalability. Furthermore, as the multigrid hierarchy is formed, each coarse matrix is formed through a triple matrix product. The resulting coarse grids often have significantly more nonzeros per row than the original fine-grid operator, thereby generating high parallel communication costs associated with sparse matrix-vector multiplication (SpMV) on coarse levels. In this paper, we introduce a method that systematically removes entries in coarse-grid matrices after the hierarchy is formed, leading to improved communication costs. We sparsify by removing weakly connected or unimportant entries in the matrix, leading to improved solve time. The main trade-off is that if the heuristic identifying unimportant entries is used too aggressively, then AMG convergence can suffer. To counteract this, the original hierarchy is retained, allowing entries to be reintroduced into the solver hierarchy if convergence is too slow. This enables a balance between communication cost and convergence, as necessary. In this paper we present new algorithms for reducing communication and present a number of computational experiments in support.

97 MATHEMATICS AND COMPUTING↗

First- and Second-Order Methods for Online Convolutional Dictionary Learning

Convolutional sparse representations are a form of sparse representation with a structured, translation-invariant dictionary. Most convolutional dictionary learning algorithms to date operate in batch mode, requiring simultaneous access to all training images during the learning process, which results in very high memory usage and severely limits the training data size that can be used. Very recently, however, a number of authors have considered the design of online convolutional dictionary learning algorithms that offer far better scaling of memory and computational cost with training set size than batch methods. This study extends our prior work, improving a number of aspects of our previous algorithm; proposing an entirely new one, with better performance, that supports the inclusion of a spatial mask for learning from incomplete data; and providing a rigorous theoretical analysis of these methods.

97 MATHEMATICS AND COMPUTING↗

A Domain Decomposition Model Reduction Method for Linear Convection-Diffusion Equations with Random Coefficients

We developed a domain decomposition model reduction method for linear steady- state convection-diffusion equations with random coefficients. Of particular interest to this effort are the diffusion equation with random diffusivity and the convection-dominated transport equation with random velocity. We investigated two types of random fields, i.e., colored noises and discrete white noises, both of which can lead to high-dimensional parametric dependence in practice. The motivation is to exploit domain decomposition to reduce the parametric dimension of local problems in subdomains, such that an entire parametric map can be approximated with a small number of expensive partial differential equation (PDE) simulations. The new method combines domain decomposition with model reduction and sparse polynomial approximation, so as to simultaneously handle the high dimensionality and irregular behavior of the PDEs under consideration. Here, the advantages of our method lie in three aspects: (i) online-offline decomposition, i.e., the online cost is independent of the size of the triangle mesh; (ii) sparse approximation of operators involving nonaffine high-dimensional random fields; (iii) an effective strategy to capture irregular behaviors, e.g., sharp transitions of the PDE solutions. Two numerical examples are provided to demonstrate the advantageous performance of our method.

97 MATHEMATICS AND COMPUTING↗

Stochastic Gradients for Large-Scale Tensor Decomposition

Tensor decomposition is a well-known tool for multiway data analysis. This work proposes using stochastic gradients for efficient generalized canonical polyadic (GCP) tensor decomposition of large-scale tensors. GCP tensor decomposition is a recently proposed version of tensor decomposition that allows for a variety of loss functions such as Bernoulli loss for binary data or Huber loss for robust estimation. Here, the stochastic gradient is formed from randomly sampled elements of the tensor and is efficient because it can be computed using the sparse matricized-tensor times Khatri--Rao product tensor kernel. For dense tensors, we simply use uniform sampling. For sparse tensors, we propose two types of stratified sampling that give precedence to sampling nonzeros. Numerical results demonstrate the advantages of the proposed approach and its scalability to large-scale problems.

97 MATHEMATICS AND COMPUTING↗

Data-Driven Closures and Assimilation for Stiff Multiscale Random Dynamics

Here, we introduce a data-driven and physics-informed framework for propagating uncertainty in stiff, multiscale random ordinary differential equations (RODEs) driven by correlated (colored) noise. Unlike systems subjected to Gaussian white noise, a deterministic equation for the joint probability density function (PDF) of RODE state variables does not exist in closed form. Moreover, such an equation would require as many phase-space variables as there are states in the RODE system. To alleviate this curse of dimensionality, we instead derive exact, albeit unclosed, reduced-order PDF (RoPDF) equations for low-dimensional observables/quantities of interest. The unclosed terms take the form of state-dependent conditional expectations, which are directly estimated from data at sparse observation times. However, for systems exhibiting stiff, multiscale dynamics, data sparsity introduces regression discrepancies that compound during RoPDF evolution. This is overcome by introducing a kinetic-like defect term to the RoPDF equation, which is learned by assimilating in sparse, low-fidelity RoPDF estimates. Two assimilation methods are considered, namely nudging and deep neural networks, which are successfully tested against Monte Carlo simulations.

97 MATHEMATICS AND COMPUTING↗

A Distributed-Memory Package for Dense Hierarchically Semi-Separable Matrix Computations Using Randomization

In this paper, we present a distributed-memory library for computations with dense structured matrices. A matrix is considered structured if its off-diagonal blocks can be approximated by a rank-deficient matrix with low numerical rank. Here, we use Hierarchically Semi-Separable (HSS) representations. Such matrices appear in many applications, for example, finite-element methods, boundary element methods, and so on. Exploiting this structure allows for fast solution of linear systems and/or fast computation of matrix-vector products, which are the two main building blocks of matrix computations. The compression algorithm that we use, that computes the HSS form of an input dense matrix, relies on randomized sampling with a novel adaptive sampling mechanism. We discuss the parallelization of this algorithm and also present the parallelization of structured matrix-vector product, structured factorization, and solution routines. The efficiency of the approach is demonstrated on large problems from different academic and industrial applications, on up to 8,000 cores. Finally, this work is part of a more global effort, the STRUctured Matrices PACKage (STRUMPACK) software package for computations with sparse and dense structured matrices. Hence, although useful on their own right, the routines also represent a step in the direction of a distributed-memory sparse solver.

98 NUCLEAR DISARMAMENT, SAFEGUARDS, AND PHYSICAL P↗

A Parallel Solver for Graph Laplacians

Problems from graph drawing, spectral clustering, network flow and graph partitioning can all be expressed in terms of graph Laplacian matrices. There are a variety of practical approaches to solving these problems in serial. However, as problem sizes increase and single core speeds stagnate, parallelism is essential to solve such problems quickly. We present an unsmoothed aggregation multigrid method for solving graph Laplacians in a distributed memory setting. We introduce new parallel aggregation and low degree elimination algorithms targeted specifically at irregular degree graphs. These algorithms are expressed in terms of sparse matrix-vector products using generalized sum and product operations. This formulation is amenable to linear algebra using arbitrary distributions and allows us to operate on a 2D sparse matrix distribution, which is necessary for parallel scalability. Our solver outperforms the natural parallel extension of the current state of the art in an algorithmic comparison. We demonstrate scalability to 576 processes and graphs with up to 1.7 billion edges.

Konolige, Tristan↗

A supernodal all-pairs shortest path algorithm

We show how to exploit graph sparsity in the Floyd-Warshall algorithm for the all-pairs shortest path (Apsp) problem. Floyd-Warshall is an attractive choice for Apsp on high-performing systems due to its structural similarity to solving dense linear systems and matrix multiplication. However, if sparsity of the input graph is not properly exploited, Floyd-Warshall will perform unnecessary asymptotic work and thus may not be a suitable choice for many input graphs. To overcome this limitation, the key idea in our approach is to use the known algebraic relationship between Floyd-Warshall and Gaussian elimination, and import several algorithmic techniques from sparse Cholesky factorization, namely, fill-in reducing ordering, symbolic analysis, supernodal traversal, and elimination tree parallelism. When combined, these techniques reduce computation, improve locality and enhance parallelism. We implement these ideas in an efficient shared memory parallel prototype that is orders of magnitude faster than an efficient multi-threaded baseline Floyd-Warshall that does not exploit sparsity. Our experiments suggest that the Floyd-Warshall algorithm can compete with Dijkstra's algorithm (the algorithmic core of Johnson's algorithm) for several classes sparse graphs.

Sao, Piyush↗

Exploring the Use of Novel Spatial Accelerators in Scientific Applications

Driven by the need to find alternative accelerators which can viably replace GPUs in next-generation Supercomputing systems, this paper proposes a methodology to enable agile application/hardware co-design. The application-first methodology provides the ability to come up with design of accelerators while working with real-world workloads, available accelerators, and system software. The iterative design process targets a set of kernels in a workload for performance estimates that can prune the design space for later phases of detailed architectural evaluations. To this effect, in this paper, a novel data-parallel device model is introduced that simulates the latency of performance-sensitive operations in an accelerator including data transfers and kernel computation using multi-core CPUs. The use of off-the-shelf simulators, such as pre-RTL simulator Aladdin or multiple tools available for exploring the design of deep neural network accelerators (e.g., Timeloop) is demonstrated for evaluation of various accelerator designs using applications with realistic inputs. Examples of multiple device configurations that are instantiable in a system are explored to evaluate the performance benefit of deploying novel accelerators. The proposed device is integrated with a programming model and system software to potentially explore the impacts of high-level programming languages/compilers and low-level effects such as task scheduling on multiple accelerators. We analyze our methodology for a set of applications that represent high-performance computing (HPC) and graph analytics. The applications include a computational chemistry kernel realized using tensor contractions, triangle counting, GraphSAGE and Breadth-first Search. These applications include kernels such as dense matrix-dense matrix multiplication, sparse matrix-spare matrix multiplication, and sparse matrix-dense vector multiplication. Our results indicate potential performance benefits and insights for system design by including accelerators that realize these kernels along-side general purpose accelerators.

AI, codesign, Accelerated Computing, Modeling and ↗

SuperNeuro: A Fast and Scalable Simulator for Neuromorphic Computing

In many neuromorphic workflows, simulators play a vital role for important tasks such as training spiking neural networks, running neuroscience simulations, and designing, implementing, and testing neuromorphic algorithms. Currently available simulators cater to either neuroscience workflows (e.g., NEST and Brian2) or deep learning workflows (e.g., BindsNET). Problematically, the neuroscience-based simulators are slow and not very scalable, and the deep learning-based simulators do not support certain functionalities that are typical of neuromorphic workloads (e.g., synaptic delay). In this paper, we address this gap in the literature and present SuperNeuro, which is a fast and scalable simulator for neuromorphic computing capable of both homogeneous and heterogeneous simulations as well as GPU acceleration. We also present preliminary results that compare SuperNeuro to widely used neuromorphic simulators such as NEST, Brian2, and BindsNET in terms of computation times. We demonstrate that SuperNeuro can be approximately 10×--300× faster than some of the other simulators for small sparse networks. On large sparse and large dense networks, SuperNeuro can be approximately 2.2×--3.4× faster than the other simulators, respectively.

Date, Prasanna↗

Scalable Incremental Checkpointing using GPU-Accelerated De-Duplication

Writing large amounts of data concurrently to stable storage is a typical I/O pattern of many HPC workflows. This pattern introduces high I/O overheads and results in increased storage space utilization especially for workflows that need to capture the evolution of data structures with high frequency as checkpoints. In this context, many applications, such as graph pattern matching, perform sparse updates to large data structures between checkpoints. For these applications, incremental checkpointing techniques that save only the differences from one checkpoint to another can dramatically reduce the checkpoint sizes, I/O bottlenecks, and storage space utilization. However, such techniques are not without challenges: it is non-trivial to transparently determine what data has changed since a previous checkpoint and assemble the differences in a compact fashion that does not result in excessive metadata. State-of-art data reduction techniques (e.g., compression and de-duplication) have significant limitations when applied to modern HPC applications that leverage GPUs: slow at detecting the differences, generate a large amount of metadata to keep track of the differences, and ignore crucial spatiotemporal checkpoint data redundancy. This paper addresses these challenges by proposing a Merkle tree-based incremental checkpointing method to exploit GPUs' high memory bandwidth and massive parallelism. Experimental results at scale show a significant reduction of the I/O overhead and space utilization of checkpointing compared with state-of-the-art incremental checkpointing and compression techniques.

Tan, Nigel↗