Search NASA⌕ Search

SEARCH · Search NASA

Results for “distributed 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

A new calibration method for charm jet identification validated with proton-proton collision events at $\sqrt{s}$ =13 TeV

Many measurements at the LHC require efficient identification of heavy-flavour jets, i.e. jets originating from bottom (b) or charm (c) quarks. An overview of the algorithms used to identify c jets is described and a novel method to calibrate them is presented. This new method adjusts the entire distributions of the outputs obtained when the algorithms are applied to jets of different flavours. It is based on an iterative approach exploiting three distinct control regions that are enriched with either b jets, c jets, or light-flavour and gluon jets. Results are presented in the form of correction factors evaluated using proton-proton collision data with an integrated luminosity of 41.5 fb$^{-1}$ at √s = 13 TeV, collected by the CMS experiment in 2017. The closure of the method is tested by applying the measured correction factors on simulated data sets and checking the agreement between the adjusted simulation and collision data. Furthermore, a validation is performed by testing the method on pseudodata, which emulate various mismodelling conditions. The calibrated results enable the use of the full distributions of heavy-flavour identification algorithm outputs, e.g. as inputs to machine-learning models. Thus, they are expected to increase the sensitivity of future physics analyses.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Measuring the effects of heterogeneity on distributed systems

Distributed computer systems in daily use are becoming more and more heterogeneous. Currently, much of the design and analysis studies of such systems assume homogeneity. This assumption of homogeneity has been mainly driven by the resulting simplicity in modeling and analysis. A simulation study is presented which investigated the effects of heterogeneity on scheduling algorithms for hard real time distributed systems. In contrast to previous results which indicate that random scheduling may be as good as a more complex scheduler, this algorithm is shown to be consistently better than a random scheduler. This conclusion is more prevalent at high workloads as well as at high levels of heterogeneity.

El-Toweissy, Mohamed↗

Fast calculation of diffraction patterns from an ensemble of aligned molecules

We report an algorithm to calculate electron diffraction patterns for molecules with anisotropic angular distribution, which is significantly faster than existing methods. The algorithm uses a transform to convert the molecular orientation distribution, which is a function of three Euler angles, to the atom-pair distribution functions which depend on the polar and azimuthal angles. The diffraction signal can then be calculated from the atom-pair distributions. We demonstrate the computation method numerically by calculating electron diffraction patterns for a symmetric top molecule (trifluoroiodomethane) and an asymmetric top molecule (formaldehyde) and show that it reduces the calculation time by approximately two orders of magnitude compared to the standard brute-force method. Here, the method can also be applied to the calculation of x-ray diffraction patterns.

74 ATOMIC AND MOLECULAR PHYSICS↗

Generative Adversarial Network for Synthetic Time Series Data Generation in Smart Grids

The availability of fine grained time series data is a pre-requisite for research in smart-grids. While data for transmission systems is relatively easily obtainable, issues related to data collection, security and privacy hinder the widespread public availability/accessibility of such datasets at the distribution system level. This has prevented the larger research community from effectively applying sophisticated machine learning algorithms to significantly improve the distribution-level accuracy of predictions and increase the efficiency of grid operations. Synthetic dataset generation has proven to be a promising solution for addressing data availability issues in various domains such as computer vision, natural language processing and medicine. However, its exploration in the smart grid context remains unsatisfactory. Previous works have tried to generate synthetic datasets by modeling the underlying system dynamics: an approach which is difficult, time consuming, error prone and often times infeasible in many problems. In this work, we propose a novel data-driven approach to synthetic dataset generation by utilizing deep generative adversarial networks (GAN) to learn the conditional probability distribution of essential features in the real dataset and generate samples based on the learned distribution. To evaluate our synthetically generated dataset, we measure the maximum mean discrepancy (MMD) between real and synthetic datasets as probability distributions, and show that their sampling distance converges. To further validate our synthetic dataset, we perform common smart grid tasks such as k-means clustering and short-term prediction on both datasets. Experimental results show the efficacy of our synthetic dataset approach: the real and synthetic datasets are indistinguishable by solely examining the output of these tasks.

14 SOLAR ENERGY↗

An Implementation of a Texture Analysis System

From analysis of the crystalline texture (the distribution of crystalline orientations in a polycrystalline substance) of target samples, some critical material properties of those samples may be deduced. Texture typically results in a specimen from the conditions by which its substance is formed or engineered, and may be used to quantify its qualities. This project aimed to produce a full process for texture analysis of target samples. It was intended that this process be implemented for CCD or imaging plate diffraction data recorded in the Stanford Synchrotron Radiation Lightsource Beamline 11-3. A process was implemented for smoothly analyzing diffraction images and producing full pole figures. The MATLAB package MTEX formed a base for the implementation of this process. This process was applied to diffraction data from a rolled Cu specimen. The resulting raw-interpolated and completed pole figures did not sufficiently match to indicate a correct approximation of the orientation distribution function (ODF) for that specimen. A comparative analysis of MTEX and other pole figure inversion algorithms, as applied to this Cu data, may reveal that another method better approximates the ODF.

43 PARTICLE ACCELERATORS↗

TU-D-209-03: Alignment of the Patient Graphic Model Using Fluoroscopic Images for Skin Dose Mapping

Purpose: The Dose Tracking System (DTS) was developed to provide realtime feedback of skin dose and dose rate during interventional fluoroscopic procedures. A color map on a 3D graphic of the patient represents the cumulative dose distribution on the skin. Automated image correlation algorithms are described which use the fluoroscopic procedure images to align and scale the patient graphic for more accurate dose mapping. Methods: Currently, the DTS employs manual patient graphic selection and alignment. To improve the accuracy of dose mapping and automate the software, various methods are explored to extract information about the beam location and patient morphology from the procedure images. To match patient anatomy with a reference projection image, preprocessing is first used, including edge enhancement, edge detection, and contour detection. Template matching algorithms from OpenCV are then employed to find the location of the beam. Once a match is found, the reference graphic is scaled and rotated to fit the patient, using image registration correlation functions in Matlab. The algorithm runs correlation functions for all points and maps all correlation confidences to a surface map. The highest point of correlation is used for alignment and scaling. The transformation data is saved for later model scaling. Results: Anatomic recognition is used to find matching features between model and image and image registration correlation provides for alignment and scaling at any rotation angle with less than onesecond runtime, and at noise levels in excess of 150% of those found in normal procedures. Conclusion: The algorithm provides the necessary scaling and alignment tools to improve the accuracy of dose distribution mapping on the patient graphic with the DTS. Partial support from NIH Grant R01-EB002873 and Toshiba Medical Systems Corp.

60 APPLIED LIFE SCIENCES↗

What does fault tolerant Deep Learning need from MPI?

Deep Learning (DL) algorithms have become the {\em de facto} Machine Learning (ML) algorithm for large scale data analysis. DL algorithms are computationally expensive -- even distributed DL implementations which use MPI require days of training (model learning) time on commonly studied datasets. Long running DL applications become susceptible to faults -- requiring development of a fault tolerant system infrastructure, in addition to fault tolerant DL algorithms. This raises an important question: {\em What is needed from MPI for designing fault tolerant DL implementations?} In this paper, we address this problem for permanent faults. We motivate the need for a fault tolerant MPI specification by an in-depth consideration of recent innovations in DL algorithms and their properties, which drive the need for specific fault tolerance features. We present an in-depth discussion on the suitability of different parallelism types (model, data and hybrid); a need (or lack thereof) for check-pointing of any critical data structures; and most importantly, consideration for several fault tolerance proposals (user-level fault mitigation (ULFM), Reinit) in MPI and their applicability to fault tolerant DL implementations. We leverage a distributed memory implementation of Caffe, currently available under the Machine Learning Toolkit for Extreme Scale (MaTEx). We implement our approaches by extending MaTEx-Caffe for using ULFM-based implementation. Our evaluation using the ImageNet dataset and AlexNet neural network topology demonstrates the effectiveness of the proposed fault tolerant DL implementation using OpenMPI based ULFM.

Amatya, Vinay C.↗

Master State Distributed Estimator (masde)

The MSE technology provides a distributed state estimation scheme for power utilities by making it a real time dynamic system, versus the static snapshot method used today. This will allow utilities to verify power grid readings and identify false data on the communication network. The prevailing method of estimation takes all the data from the system into a single load flow equation at the utilities command center. This becomes a large algorithm that is very time consuming to solve, providing the engineers only snap shots of the system. The MFD distributes an algorithm to devises that are already installed at various locations on the power grid. This allows for much faster computer times because the algorithms contain magnitudes less data. When each distributed estimator completes its calculation, it sends the results back to the master device to check against all other distributed estimators.

Reen, DylanW.↗

A Comparison of Forecast Error Generators for Modeling Wind and Load Uncertainty

This paper presents four algorithms to generate random forecast error time series, including a truncated-normal distribution model, a state-space based Markov model, a seasonal autoregressive moving average (ARMA) model, and a stochastic-optimization based model. The error time series are used to create real-time (RT), hour-ahead (HA), and day-ahead (DA) wind and load forecast time series that statistically match historically observed forecasting data sets, used for variable generation integration studies. A comparison is made using historical DA load forecast and actual load values to generate new sets of DA forecasts with similar stoical forecast error characteristics. This paper discusses and compares the capabilities of each algorithm to preserve the characteristics of the historical forecast data sets.

Lu, Ning↗

A stochastic scan strategy for grain structure control in complex geometries using electron beam powder bed fusion

Spatial control of microstructure within a three-dimensional component has been a dream of materials scientists for centuries. However, limitations in traditional manufacturing processes prevent detailed control over the distribution of microstructures in a single part. Here, we demonstrate the ability to control grain structure and crystallographic texture during metal additive manufacturing for arbitrary cross-sections of a practical size, with profound implications for the design and optimization of next-generation products. The key to this advance is a new geometry agnostic scan path algorithm that manipulates the spatial distribution of solidification conditions. Utilizing a fundamental understanding of solidification dynamics and a model of the heat transfer during processing, we have designed this algorithm to manipulate the natural competition between epitaxial dendrite growth and grain nucleation. With this algorithm, we successfully controlled the grain structure of Ni-based superalloy IN718 in the shape of the Mona Lisa.

36 MATERIALS SCIENCE↗

Simulation of AMVs for OSSEs

An algorithm to simulate locations of atmospheric motion vectors (AMVs) for use in observing system simulation experiments (OSSEs) is described and demonstrated. It is intended to obviate likely deficiencies in nature run data [standard meteorological parameters and aerosol tracers] if used to produce images for feature tracking. The algorithm employs probabilistic functions that are tuned based on distributions of real observations and histograms of nature run fields. For distinct observation types, the algorithm produces geographical and vertical distributions, time - mean counts, and typical spacings of simulated locations that are, at least, qualitatively similar to those of real observations and are associated with nature run cloud and water vapor fields. It thus appears suitable for generating realistic atmospheric motion vectors for use in observing system simulation experiments.

Errico, R. M.↗

Keyless Infrastructure Security Solution (KISS) (Technology Landscape Analysis Report)

The following landscape analysis explores blockchain or distributed ledger technologies application to securing electricity infrastructure. The exploration of how distributed ledger technologies can be used to increase the trustworthiness and integrity of the grid’s edge is imperative to the economic, security and well-being of all modern societies that rely on electricity. Blockchain technology’s ability to secure, track and optimize complex data transactions provides an exciting value proposition to securing and optimizing the U.S. power grid. An energy delivery system (EDS) operating at the grid’s edge requires unprecedented levels of security and trustworthiness to verify integrity of data and manage complex transactive and distributed Energy Resources (DERs) exchanges. Moreover, grid-edge devices lack visibility, control and security to conduct real-time energy transactions with the required security, speed and scale. The U.S. power grid is a complex system of systems that requires secure, reliable and trustworthy EDSs. Grid modernization has increased the speed and size of data sets exchanged on these systems (Gordes and Mylrea 2014). Exasperating the challenge is these systems are increasingly distributed creating new data fidelity and interoperability challenges for grid operators struggling to balance and incorporate DERs. Blockchain technology provides an atomically verifiable cryptographic signature to help increase the trustworthiness of EDSs at the grid’s edge. This feature is especially important at distribution level since EDSs and field devices have increasing operational and security requirements that are often diametrically opposed. For example, as data, speed and analytic requirements increase, security and functionality requirements increase. Further, as the grid’s edge incorporate DERs and transacts in real time, availability is prioritized over the integrity and confidentially of that data. Blockchain presents the prospect of solving some of these complex challenges related to grid security and modernization. However, there are more questions about blockchain applicability than answers. There are number of theoretical blockchain applications to the energy sector, but few energy utilities have actually applied, implemented or even experimented with the underlying distributed ledger technology and consensus algorithms that enable blockchain to exchange data more efficiently and securely.

24 POWER TRANSMISSION AND DISTRIBUTION↗

TH-AB-209-03: Overcoming Resolution Limitations of Diffuse Optical Signals in X-Ray Induced Luminescence (XIL) Imaging Via Selective Plane Illumination and 2D Deconvolution

Purpose: To evaluate the resolution and sensitivity of XIL imaging using a surface radiance simulation based on optical diffusion and maximum likelihood expectation maximization (MLEM) image reconstruction. XIL imaging seeks to determine the distribution of luminescent nanophosphors, which could be used as nanodosimeters or radiosensitizers. Methods: The XIL simulation generated a homogeneous slab with optical properties similar to tissue. X-ray activated nanophosphors were placed at 1.0 cm depth in the tissue in concentrations of 10{sup −4} g/mL in two volumes of 10 mm{sup 3} with varying separations between each other. An analytical optical diffusion model determined the surface radiance from the photon distributions generated at depth in the tissue by the nanophosphors. The simulation then determined the detected luminescent signal collected with a f/1.0 aperture lens and back-illuminated EMCCD camera. The surface radiance was deconvolved using a MLEM algorithm to estimate the nanophosphors distribution and the resolution. To account for both Poisson and Gaussian noise, a shifted Poisson imaging model was used in the deconvolution. The deconvolved distributions were fitted to a Gaussian after radial averaging to measure the full width at half maximum (FWHM) and the peak to peak distance between distributions was measured to determine the resolving power. Results: Simulated surface radiances for doses from 1mGy to 100 cGy were computed. Each image was deconvolved using 1000 iterations. At 1mGy, deconvolution reduced the FWHM of the nanophosphors distribution by 65% and had a resolving power is 3.84 mm. Decreasing the dose from 100 cGy to 1 mGy increased the FWHM by 22% but allowed for a dose reduction of a factor of 1000. Conclusion: Deconvolving the detected surface radiance allows for dose reduction while maintaining the resolution of the nanophosphors. It proves to be a useful technique in overcoming the resolution limitations of diffuse optical imaging in tissue. C. S. acknowledges support from the NIH National Institute of General Medical Sciences (Award number R25GM109439, Project Title: University of Chicago Initiative for Maximizing Student Development, IMSD). B. Q. and P. L. acknowledge support from NIH grant R01EB017293.

60 APPLIED LIFE SCIENCES↗

Multiscale modeling of polyisoprene on graphite

The local dynamics and the conformational properties of polyisoprene next to a smooth graphite surface constructed by graphene layers are studied by a multiscale methodology. First, fully atomistic molecular dynamics simulations of oligomers next to the surface are performed. Subsequently, Monte Carlo simulations of a systematically derived coarse-grained model generate numerous uncorrelated structures for polymer systems. A new reverse backmapping strategy is presented that reintroduces atomistic detail. Finally, multiple extensive fully atomistic simulations with large systems of long macromolecules are employed to examine local dynamics in proximity to graphite. Polyisoprene repeat units arrange close to a parallel configuration with chains exhibiting a distribution of contact lengths. Efficient Monte Carlo algorithms with the coarse-grain model are capable of sampling these distributions for any molecular weight in quantitative agreement with predictions from atomistic models. Furthermore, molecular dynamics simulations with well-equilibrated systems at all length-scales support an increased dynamic heterogeneity that is emerging from both intermolecular interactions with the flat surface and intramolecular cooperativity. This study provides a detailed comprehensive picture of polyisoprene on a flat surface and consists of an effort to characterize such systems in atomistic detail.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

SU-E-T-517: Analytic Formalism to Compute in Real Time Dose Distributions Delivered by HDR Units

Purpose: Develop an analytical algorithm to compute the dose delivered by Ir-192 dwell positions with high accuracy using the 3-dimensional (3D) dose distribution of an HDR source. Using our analytical function, the dose delivered by an HDR unit as treatment progresses can be determined using the actual delivered temporal and positional data of each individual dwell. Consequently, true delivered dose can be computed when each catheter becomes active. We hypothesize that the knowledge of such analytical formulation will allow developing HDR systems with a real time treatment evaluation tool to avoid mistreatments. Methods: In our analytic formulation, the dose is computed by using the full anisotropic function data of the TG 43 formalism with 3D ellipsoidal function. The discrepancy between the planned dose and the delivered dose is computed using an analytic perturbation method over the initial dose distribution. This methodology speeds up the computation because only changes in dose discrepancies originated by spatial and temporal deviations are computed. A dose difference map at the point of interest is obtained from these functions and this difference can be shown during treatment in real time to examine the treatment accuracy. Results: We determine the analytical solution and a perturbation function for the 3 translational 3 rotational, and 1D temporal errors in source distributions. The analytic formulation is a sequence of simple equations that can be processed in any modern computer in few seconds. Because computations are based in an analytical solution, small deviations of the dose when sub-millimeter positional changes occur can be detected. Conclusions: We formulated an analytical method to compute 4D dose distributions and dose differences based on an analytical solution and perturbations to the original dose. This method is highly accurate and can be.

60 APPLIED LIFE SCIENCES↗

Single-Scattering Properties of Ellipsoidal Dust Aerosols Constrained By Measured Dust Shape Distributions

Most global aerosol models approximate dust as spherical particles, whereas most remote sensing retrieval algorithms approximate dust as spheroidal particles with a shape distribution that conflicts with measurements. These inconsistent and inaccurate shape assumptions generate biases in dust single-scattering properties. Here, we obtain dust single-scattering properties by approximating dust as triaxial ellipsoidal particles with observationally constrained shape distributions. We find that, relative to the ellipsoidal dust optics obtained here, the spherical dust optics used in most aerosol models underestimate dust single-scattering albedo, mass extinction efficiency, and asymmetry parameter for almost all dust sizes in both the shortwave and longwave spectra. We further find that the ellipsoidal dust optics are in substantially better agreement with observations of the scattering matrix and linear depolarization ratio than the spheroidal dust optics used in most retrieval algorithms. However, relative to observations, the ellipsoidal dust optics overestimate the lidar ratio by underestimating the backscattering intensity by a factor of ∼2. This occurs largely because the computational method used to simulate ellipsoidal dust optics (i.e., the improved geometric optics method) underestimates the backscattering intensity by a factor of ∼2 relative to other computational methods (e.g., the physical geometric optics method). We conclude that the ellipsoidal dust optics with observationally constrained shape distributions can help improve global aerosol models and possibly remote sensing retrieval algorithms that do not use the backscattering signal.

Dust↗

Optimal dynamic control of resources in a distributed system

The authors quantitatively formulate the problem of controlling resources in a distributed system so as to optimize a reward function and derive optimal control strategies using Markov decision theory. The control variables treated are quite general; they could be control decisions related to system configuration, repair, diagnostics, files, or data. Two algorithms for resource control in distributed systems are derived for time-invariant and periodic environments, respectively. A detailed example to demonstrate the power and usefulness of the approach is provided.

Shin, Kang G.↗

Optimization by decomposition

An algorithm is presented for solving the structural optimization problem as a set of smaller subproblems that correspond to levels of nested substructures. In all three of the algorithm variants presented, the matching was assisted by means of the behavior and optimum sensitivity derivatives. The algorithm is noted to be intrinsically germane to distributed computing, since the subproblems can be concurrently addressed; the algorithm can also be generalized to those multidisciplinary systems whose subsystems can be arranged into a hierarchy of substructure-like dependencies.

Sobieszczanski-Sobieski, Jaroslaw↗