Search NASA⌕ Search

SEARCH · Search NASA

Results for “Vectorized algorithm”

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 865 records · Page 48

Evaluation and Comparison of Machine Learning Techniques for Rapid QSTS Simulations

Rapid and accurate quasi-static time series (QSTS) analysis is becoming increasingly important for distribution system analysis as the complexity of the distribution system intensifies with the addition of new types, and quantities, of distributed energy resources (DER). The expanding need for hosting capacity analysis, control systems analysis, photovoltaic (PV) and DER impact analysis, and maintenance cost estimations are just a few reasons that QSTS is necessary. Historically, QSTS analysis has been prohibitively slow due to the number of computations required for a full-year analysis. Therefore, new techniques are required that allow QSTS analysis to rapidly be performed for many different use cases. This research demonstrates a novel approach to doing rapid QSTS analysis for analyzing the number of voltage regulator tap changes in a distribution system with PV components. A representative portion of a yearlong dataset is selected and QSTS analysis is performed to determine the number of tap changes, and this is used as training data for a machine learning algorithm. The machine learning algorithm is then used to predict the number of tap changes in the remaining portion of the year not analyzed directly with QSTS. The predictions from the machine learning algorithms are combined with the results of the partial year simulation for a final prediction for the entire year, with the goal of maintaining an error <10% on the full-year prediction. Five different machine learning techniques were evaluated and compared with each other; a neural network ensemble, a random forest decision tree ensemble, a boosted decision tree ensemble, support vector machines, and a convolutional neural network deep learning technique. A combination of the neural network ensemble together with the random forest produced the best results. Using 20% of the year as training data, analyzed with QSTS, the average performance of the technique resulted in ~2.5% error in the yearly tap changes, while maintaining a <10% 99.9 th percentile error bound on the results. This is a 5x speedup compared to a standard, full-length QSTS simulation. These results demonstrate the potential for applying machine learning techniques to facilitate modern distribution system analysis and further integration of distributed energy resources into the power grid.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Qubit lattice algorithm simulations of the scattering of a bounded two dimensional electromagnetic pulse from an infinite planar dielectric interface

Qubit lattice algorithm (QLA) simulations are performed for a two-dimensional spatially bounded pulse propagating onto a plane interface between two dielectric slabs. QLA is an initial value scheme that consists of a sequence of unitary collision and streaming operators, with appropriate potential operators, that recover Maxwell equations in inhomogeneous dielectric media to the second order in the lattice discreteness. For the case of total internal reflection, there is transient energy transfer into the second medium due to the evanescent fields as the Poynting unit vector of the pulse is rotated from its incident to reflected direction. Because of the finite spatial extent of the pulse, a self-consistent Goos–Hanchen-type displacement along the interface is found without imposing any explicit interface boundary conditions on the fields. For normal incidence, the standard Fresnel coefficients are recovered for appropriately averaged QLA fields. Energy is conserved at all times to seven significant figures.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

SU-E-J-91: Biomechanical Deformable Image Registration of Longitudinal Lung CT Images

Purpose: Spatial correlation of lung tissue across longitudinal images, as the patient responds to treatment, is a critical step in adaptive radiotherapy. The goal of this work is to expand a biomechanical model-based deformable registration algorithm (Morfeus) to achieve accurate registration in the presence of significant anatomical changes. Methods: Four lung cancer patients previously treated with conventionally fractionated radiotherapy that exhibited notable tumor shrinkage during treatment were retrospectively evaluated. Exhale breathhold CT scans were obtained at treatment planning (PCT) and following three weeks (W3CT) of treatment. For each patient, the PCT was registered to the W3CT using Morfeus, a biomechanical model-based deformable registration algorithm, consisting of boundary conditions on the lungs and incorporating a sliding interface between the lung and chest wall. To model the complex response of the lung, an extension to Morfeus has been developed: (i) The vessel tree was segmented by thresholding a vesselness image based on the Hessian matrix’s eigenvalues and the centerline was extracted; (ii) A 3D shape context method was used to find correspondences between the trees of the two images; (ii) Correspondences were used as additional boundary conditions (Morfeus+vBC). An expert independently identified corresponding landmarks well distributed in the lung to compute Target Registration Errors (TRE). Results: The TRE within 15mm of the tumor boundaries (on average 11 landmarks) is: 6.1±1.8, 4.6±1.1 and 3.8±2.3 mm after rigid registration, Morfeus and Morfeus+vBC, respectively. The TRE in the rest of the lung (on average 13 landmarks) is: 6.4±3.9, 4.7±2.2 and 3.6±1.9 mm, which is on the order of the 2mm isotropic dose grid vector (3.5mm). Conclusion: The addition of boundary conditions on the vessels improved the accuracy in modeling the response of the lung and tumor over the course of radiotherapy. Minimizing and modeling these geometrical uncertainties will enable future plan adaptation strategies. This work was funded in part by NIH 2P01CA059827-16.

60 APPLIED LIFE SCIENCES↗

A log-barrier Newton-CG method for bound constrained optimization with complexity guarantees

Abstract We describe an algorithm based on a logarithmic barrier function, Newton’s method and linear conjugate gradients that seeks an approximate minimizer of a smooth function over the non-negative orthant. We develop a bound on the complexity of the approach, stated in terms of the required accuracy and the cost of a single gradient evaluation of the objective function and/or a matrix-vector multiplication involving the Hessian of the objective. The approach can be implemented without explicit calculation or storage of the Hessian.

O’Neill, Michael↗

A support vector machine (SVM) for predicting preferred treatment position in radiotherapy of patients with breast cancer

Purpose: NYU 05-181 protocol compared the CT simulation in both supine and prone positions for 400 patients with breast cancer (200 left-breast and 200 right-breast) to identify which setup is better at sparing heart and lung involvement in the treatment process. The results demonstrated that all right-breast patients benefited from the prone treatment position, while for left-breast patients, 85% were better treated prone and 15% were better treated supine. Using the clinical data collected from this protocol, the authors aimed at developing an automated tool capable of identifying which of the left-breast cancer patients are better treated supine without obtaining a second CT scan in the supine position. Methods: Prone CT scans from 198 of the 200 left-breast cancer patients enrolled in NYU 05-181 protocol were deidentified and exported to a dedicated research planning workstation. Three-dimensional geometric features of the organs at risk and tumor bed were extracted. A two-stage classifier was used to classify patients into the prone class or the supine class. In the first stage, the authors use simple thresholding to divide the patients into two groups based on their in-field heart volume. For patients with in-field heart volume {<=}0.1 cc, the prone position was chosen as the preferred treatment position. Patients with in-field heart volume >0.1 cc will be further classified in the second stage by a weighted support vector machine (SVM). The weight parameters of the SVM were adjusted to maximize the specificity [true-supine/(true-supine+false-prone)] at the cost of lowering but still maintaining reasonable sensitivity [true-prone/(true-prone+false-supine)]. The authors used K-fold cross validations to test the performance of the SVM classifier. A feature selection algorithm was also used to identify features that give the best classification performance. Results: After the first stage, 49 of the 198 left-breast cancer patients were found to have >0.1 cc of in-field heart volume. The three geometric features of heart orientation, distance between heart and tumor, and in-field lung were selected by the feature selection algorithm in the second stage of the two-stage classifier to give the best predefined weighted accuracy. The overall sensitivity and specificity of the proposed method were found to be 90.4% and 99.3%, respectively. Using two-stage classification, the authors reduced the proportion of prone-treated patients that need a second supine CT scan down to 16.3/170 or 9.6%, as compared to 21/170 or 12.4% when the authors use only the first stage (thresholding) for classification. Conclusions: The authors' study showed that a feature-based classifier is feasible for predicting the preferred treatment position, based on features extracted from prone CT scans. The two-stage classifier achieved very high specificity at an acceptable expense of sensitivity.

60 APPLIED LIFE SCIENCES↗

Scalable Computation of Streamlines on Very Large Datasets

nderstanding vector fields resulting from large scientific simulations is an important and often difficult task. Streamlines, curves that are tangential to a ve ctor field at each point, are a powerful visualization method in this context. Application of streamline-based visualization to very large vector field data repr esents a significant challenge due to the non-local and data-dependent nature of streamline computation, and requires careful balancing of computational demands placed on I/O, memory, communication, and processors. In this paper we review two parallelization approaches based on established parallelization paradigms (stat ic decomposition and on-demand loading) and present a novel hybrid algorithm for computing streamlines. Our algorithm is aimed at good scalability and performanc e across the widely varying computational characteristics of streamline-based problems. We perform performance and scalability studies of all three algorithms on a number of prototypical application problems and demonstrate that our hybrid scheme is able to perform well in different settings.

Pugmire, Dave↗

Attention-Augmented Parametric Kernel Graph Neural Network (APKGNN) for Node Classification

We present a new graph neural network, the Attention-based Parametric-Kernel augmented Graph Neural Network (APKGNN), developed for node classification tasks. Despite extensive work on modeling multi-faceted relationships between connected nodes of a graph, the effect of attention on edge features mapped to relationships has not yet been analyzed through learning representation. This study derives such an attention vector by first calculating node features corresponding to endpoints of an edge and then aggregating these with extracted local intrinsic patches of a given graph to generate augmented local patch vectors. This process uses a parametric kernel based on Gaussian mixture models (GMMs) to embed local neighborhoods of the graph in local patches. The patch vectors then convolve with the above node features to produce an updated node representation. We show that this new learning representation (APKGNN) achieves higher node classification accuracy on tasks - both standard benchmarks (Cora, PubMed, Citeseer) and new experimental short text corpora where nodes correspond to text documents and words. This implementation of the GNN convolution layer outperforms state-of-the-art (SOTA) algorithms, achieving higher training, validation, and test accuracy by a significant margin on three standard benchmark data sets under both SOTA experimental settings and those for new testbeds.

Bose, Avishek↗

Artificial Diversity and Defense Security (ADDSec) Final Report

Critical infrastructure systems continue to foster predictable communication patterns and static configurations over extended periods of time. The static nature of these systems eases the process of gathering reconnaissance information that can be used to design, develop, and launch attacks by adversaries. In this research effort, the early phases of an attack vector will be disrupted by randomizing application port numbers, IP addresses, and communication paths dynamically through the use of overlay networks within Industrial Control Systems (ICS). These protective measures convert static systems into "moving targets," adding an additional layer of defense. Additionally, we have developed a framework that automatically detects and defends against threats within these systems using an ensemble of machine learning algorithms that classify and categorize abnormal behavior. Our proof-of-concept has been demonstrated within a representative ICS environment. Performance metrics of our proof-of-concept have been captured with latency impacts of less than a millisecond, on average.

97 MATHEMATICS AND COMPUTING↗

Spatio-Temporal Surrogates for Interaction of a Jet with High Explosives: Part I - Analysis with a Small Sample Size

Computer simulations, especially of complex phenomena, can be expensive, requiring high-performance computing resources. Often, to understand a phenomenon, multiple simulations are run, each with a different set of simulation input parameters. These data are then used to create an interpolant, or surrogate, relating the simulation outputs to the corresponding inputs. When the inputs and outputs are scalars, a simple machine learning model can suffice. However, when the simulation outputs are vector valued, available at locations in two or three spatial dimensions, often with a temporal component, creating a surrogate is more challenging. In this report, we use a two-dimensional problem of a jet interacting with high explosives to understand how we can build high-quality surrogates. The characteristics of our data set are unique - the vector-valued outputs from each simulation are available at over two million spatial locations; each simulation is run for a relatively small number of time steps; the size of the computational domain varies with each simulation; and resource constraints limit the number of simulations we can run. We show how we analyze these extremely large data-sets, set the parameters for the algorithms used in the analysis, and use simple ways to improve the accuracy of the spatio-temporal surrogates without substantially increasing the number of simulations required.

97 MATHEMATICS AND COMPUTING↗

A Hierarchical Multi-Agent Based Anomaly Detection for Wide-Area Protection in Smart Grid

Future smart grid capabilities provide assurance to expand the advanced information and communication technologies to evolve into densely interconnected cyber physical system. Remedial Action Scheme (RAS), widely used for wide-area protection, relies on the interconnected networks and data sharing devices, which are exposed to the multitude of vulnerabilities. This paper presents our proposed approach to developing multi-agent based RAS scheme against the system-aware stealthy cyber-attacks. Specifically, we propose the two-level hierarchical architecture which consists of distributed local RAS controllers (RAScs) as local agents, operating at different zones/ areas, which are constantly monitored by an overseer, the central agent. The local controllers receive local and randomly changing outside zonal measurements and cyclically forwards to the overseer. The overseer identifies the corrupted controller using the anomaly detection algorithm which processes the measurements coming from the local controllers, compute measurement errors using local and outside zonal measurements, perform validation checks, and finally detect anomalies based on the two-step verification. Next, as a proof of concept, we have implemented and validated the proposed methodology in cyber physical environment at Iowa State’s PowerCyber testbed. We have also implemented the coordinated attack vectors which involve corrupting the local controller and later performing stealthy attacks on the system’s generator. We have evaluated its performance during the online testing in terms of detection rate and Iatency. The experimental results show that it is efficient in detecting different classes of attacks, including ramp and pulse attacks.

24 POWER TRANSMISSION AND DISTRIBUTION↗

On the performance and energy efficiency of sparse linear algebra on GPUs

In this paper we unveil some performance and energy efficiency frontiers for sparse computations on GPU-based supercomputers. We compare the resource efficiency of different sparse matrix–vector products (SpMV) taken from libraries such as cuSPARSE and MAGMA for GPU and Intel’s MKL for multicore CPUs, and develop a GPU sparse matrix–matrix product (SpMM) implementation that handles the simultaneous multiplication of a sparse matrix with a set of vectors in block-wise fashion. While a typical sparse computation such as the SpMV reaches only a fraction of the peak of current GPUs, we show that the SpMM succeeds in exceeding the memory-bound limitations of the SpMV. We integrate this kernel into a GPU-accelerated Locally Optimal Block Preconditioned Conjugate Gradient (LOBPCG) eigensolver. LOBPCG is chosen as a benchmark algorithm for this study as it combines an interesting mix of sparse and dense linear algebra operations that is typical for complex simulation applications, and allows for hardware-aware optimizations. In a detailed analysis we compare the performance and energy efficiency against a multi-threaded CPU counterpart. The reported performance and energy efficiency results are indicative of sparse computations on supercomputers.

Anzt, Hartwig↗

SDAV Viz July Progress Update: LANL

SDAV Viz July Progress Update: (1) VPIC (Vector Particle in Cell) Kinetic Plasma Simulation Code - (a) Implemented first version of an in-situ adapter based on Paraview CoProcessing Library, (b) Three pipelines: vtkDataSetMapper, vtkContourFilter, vtkPistonContour, (c) Next, resolve issue at boundaries of processor domains; add more advanced viz/analysis pipelines; (2) Halo finding/merger trees - (a) Summer student Wathsala W. from University of Utah is working on data-parallel halo finder algorithm using PISTON, (b) Timo Bremer (LLNL), Valerio Pascucci (Utah), George Zagaris (Kitware), and LANL people are interested in using merger trees for tracking the evolution of halos in cosmo simulations; discussed possible overlap with work by Salman Habib and Katrin Heitmann (Argonne) during their visit to LANL 7/11; (3) PISTON integration in ParaView - Now available from ParaView github.

97 MATHEMATICS AND COMPUTING↗

Mapping Vegetation at Species Level with High-Resolution Multispectral and Lidar Data Over a Large Spatial Area: A Case Study with Kudzu

Mapping vegetation species is critical to facilitate related quantitative assessment, and mapping invasive plants is important to enhance monitoring and management activities. Integrating high-resolution multispectral remote-sensing (RS) images and lidar (light detection and ranging) point clouds can provide robust features for vegetation mapping. However, using multiple sources of high-resolution RS data for vegetation mapping on a large spatial scale can be both computationally and sampling intensive. Here, we designed a two-step classification workflow to potentially decrease computational cost and sampling effort and to increase classification accuracy by integrating multispectral and lidar data in order to derive spectral, textural, and structural features for mapping target vegetation species. We used this workflow to classify kudzu, an aggressive invasive vine, in the entire Knox County (1362 km2) of Tennessee (U.S.). Object-based image analysis was conducted in the workflow. The first-step classification used 320 kudzu samples and extensive, coarsely labeled samples (based on national land cover) to generate an overprediction map of kudzu using random forest (RF). For the second step, 350 samples were randomly extracted from the overpredicted kudzu and labeled manually for the final prediction using RF and support vector machine (SVM). Computationally intensive features were only used for the second-step classification. SVM had constantly better accuracy than RF, and the producer’s accuracy, user’s accuracy, and Kappa for the SVM model on kudzu were 0.94, 0.96, and 0.90, respectively. SVM predicted 1010 kudzu patches covering 1.29 km2 in Knox County. We found the sample size of kudzu used for algorithm training impacted the accuracy and number of kudzu predicted. The proposed workflow could also improve sampling efficiency and specificity. Our workflow had much higher accuracy than the traditional method conducted in this research, and could be easily implemented to map kudzu in other regions as well as map other vegetation species.

59 BASIC BIOLOGICAL SCIENCES↗

Intelligent Sun Tracking for a CPV Power Plant

The output of a solar panel is strongly dependent on the amount of perpendicular light flux falling on its surface, and a tracking system tries to parallel the vector area of the solar panel surface to the incident solar flux. We present a tracking technique based on a two-axis sun sensor which can be used to increase the power output from a number of CPV arrays connected together in a solar power plant. The outdoor testing procedure of the developed two-axis sun sensor is discussed. The detail of the algorithm used together with the related sun tracking equipment is also presented and discussed for the new two axes sun tracking system.

14 SOLAR ENERGY↗

A minimum spanning forest based classification method for dedicated breast CT images

Purpose: To develop and test an automated algorithm to classify different types of tissue in dedicated breast CT images. Methods: Images of a single breast of five different patients were acquired with a dedicated breast CT clinical prototype. The breast CT images were processed by a multiscale bilateral filter to reduce noise while keeping edge information and were corrected to overcome cupping artifacts. As skin and glandular tissue have similar CT values on breast CT images, morphologic processing is used to identify the skin based on its position information. A support vector machine (SVM) is trained and the resulting model used to create a pixelwise classification map of fat and glandular tissue. By combining the results of the skin mask with the SVM results, the breast tissue is classified as skin, fat, and glandular tissue. This map is then used to identify markers for a minimum spanning forest that is grown to segment the image using spatial and intensity information. To evaluate the authors’ classification method, they use DICE overlap ratios to compare the results of the automated classification to those obtained by manual segmentation on five patient images. Results: Comparison between the automatic and the manual segmentation shows that the minimum spanning forest based classification method was able to successfully classify dedicated breast CT image with average DICE ratios of 96.9%, 89.8%, and 89.5% for fat, glandular, and skin tissue, respectively. Conclusions: A 2D minimum spanning forest based classification method was proposed and evaluated for classifying the fat, skin, and glandular tissue in dedicated breast CT images. The classification method can be used for dense breast tissue quantification, radiation dose assessment, and other applications in breast imaging.

60 APPLIED LIFE SCIENCES↗

Efficient modeling of photonic crystals with local Hermite polynomials

Developing compact algorithms for accurate electrodynamic calculations with minimal computational cost is an active area of research given the increasing complexity in the design of electromagnetic composite structures such as photonic crystals, metamaterials, optical interconnects, and on-chip routing. We show that electric and magnetic (EM) fields can be calculated using scalar Hermite interpolation polynomials as the numerical basis functions without having to invoke edge-based vector finite elements to suppress spurious solutions or to satisfy boundary conditions. This approach offers several fundamental advantages as evidenced through band structure solutions for periodic systems and through waveguide analysis. Compared with reciprocal space (plane wave expansion) methods for periodic systems, advantages are shown in computational costs, the ability to capture spatial complexity in the dielectric distributions, the demonstration of numerical convergence with scaling, and variational eigenfunctions free of numerical artifacts that arise from mixed-order real space basis sets or the inherent aberrations from transforming reciprocal space solutions of finite expansions. The photonic band structure of a simple crystal is used as a benchmark comparison and the ability to capture the effects of spatially complex dielectric distributions is treated using a complex pattern with highly irregular features that would stress spatial transform limits. This general method is applicable to a broad class of physical systems, e.g., to semiconducting lasers which require simultaneous modeling of transitions in quantum wells or dots together with EM cavity calculations, to modeling plasmonic structures in the presence of EM field emissions, and to on-chip propagation within monolithic integrated circuits.

36 MATERIALS SCIENCE↗

MO-G-18C-03: Evaluation of Deformable Image Registration for Lung Motion Estimation Using Hyperpolarized Gas Tagging MRI

Purpose: Hyperpolarized gas (HP) tagging MRI is a novel imaging technique for direct measurement of lung motion during breathing. This study aims to quantitatively evaluate the accuracy of deformable image registration (DIR) in lung motion estimation using HP tagging MRI as references. Methods: Three healthy subjects were imaged using the HP MR tagging, as well as a high-resolution 3D proton MR sequence (TrueFISP) at the end-of-inhalation (EOI) and the end-of-exhalation (EOE). Ground truth of lung motion and corresponding displacement vector field (tDVF) was derived from HP tagging MRI by manually tracking the displacement of tagging grids between EOI and EOE. Seven different DIR methods were applied to the high-resolution TrueFISP MR images (EOI and EOE) to generate the DIR-based DVFs (dDVF). The DIR methods include Velocity (VEL), MIM, Mirada, multi-grid B-spline from Elastix (MGB) and 3 other algorithms from DIRART toolbox (Double Force Demons (DFD), Improved Lucas-Kanade (ILK), and Iterative Optical Flow (IOF)). All registrations were performed by independent experts. Target registration error (TRE) was calculated as tDVF – dDVF. Analysis was performed for the entire lungs, and separately for the upper and lower lungs. Results: Significant differences between tDVF and dDVF were observed. Besides the DFD and IOF algorithms, all other dDVFs showed similarity in deformation magnitude distribution but away from the ground truth. The average TRE for entire lung ranged 2.5−23.7mm (mean=8.8mm), depending on the DIR method and subject's breathing amplitude. Larger TRE (13.3–23.7mm) was found in subject with larger breathing amplitude of 45.6mm. TRE was greater in lower lung (2.5−33.9 mm, mean=12.4mm) than that in upper lung (2.5−11.9 mm, mean=5.8mm). Conclusion: Significant differences were observed in lung motion estimation between the HP gas tagging MRI method and the DIR methods, especially when lung motion is large. Large variation among different DIR methods was also observed.

60 APPLIED LIFE SCIENCES↗

Triggering on New Physics with the CMS Detector

The BU CMS group led by PI Tulika Bose has made several significant contributions to the CMS trigger and to the analysis of the data collected by the CMS experiment. Group members have played a leading role in the optimization of trigger algorithms, the development of trigger menus, and the online operation of the CMS High-Level Trigger. The group’s data analysis projects have concentrated on a broad spectrum of topics that take full advantage of their strengths in jets and calorimetry, trigger, lepton identification as well as their considerable experience in hadron collider physics. Their publications cover several searches for new heavy gauge bosons, vector-like quarks as well as diboson resonances.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗