Search NASA⌕ Search

SEARCH · Search NASA

Results for “distributed parallelization”

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 1,171 records · Page 65

STEREOSCOPIC RECONSTRUCTION FROM STEREO/EUV IMAGERS DATA OF THE THREE-DIMENSIONAL SHAPE AND EXPANSION OF AN ERUPTING PROMINENCE

On 2007 May 9, a prominence eruption was observed in the He II {lambda}304 filter by the two EUV Imagers (EUVI) telescopes aboard the STEREO A and B spacecrafts. The high spatial resolution ({approx}1.''5 pixel{sup -1}) EUVI images have been used to infer via triangulation the three-dimensional (3D) shape and orientation of the prominence {approx_equal}12 minutes after the beginning (13:40 UT) of the eruption. At this time, the prominence has the shape of a 'hook' with the base anchored at the Sun. The 'hook' prominence is highly inclined southward with respect to the radial direction, has an average thickness of 0.061 R {sub sun}, a length of 0.43 R {sub sun}, and lies in first approximation on a plane inclined by {approx}54.{sup 0}5 with respect to the line of sight. Thanks to the very high temporal cadence ({approx}37 s) of EUVI observations it has been possible also to infer the 3D early eruption trajectory. In the following {approx}20 minutes the prominence rotates westward, undergoing a strong latitudinal acceleration, {approx}3 times larger than the radial acceleration. In this time interval, the prominence expands in a direction mainly parallel to the plane of the sky; the total volume occupied by the plasma increases by a factor of {approx}8, while the prominence thickness increases only by {approx}12%. This is related to the fact that the early prominence expansion is anisotropic and occurs mainly on a plane parallel to the plane of the sky. Even if the small-scale spatial distribution of the erupting material observed in the He II EUVI images is quite complex, both the approximately planar shape and the successive planar expansion suggest that on larger spatial scales the prominence can be globally approximated as a two-dimensional 'ribbon-like' feature, instead of a 3D twisted flux tube.

99 GENERAL AND MISCELLANEOUS↗

Implementation of a practical Markov chain Monte Carlo sampling algorithm in PyBioNetFit

Abstract Summary Bayesian inference in biological modeling commonly relies on Markov chain Monte Carlo (MCMC) sampling of a multidimensional and non-Gaussian posterior distribution that is not analytically tractable. Here, we present the implementation of a practical MCMC method in the open-source software package PyBioNetFit (PyBNF), which is designed to support parameterization of mathematical models for biological systems. The new MCMC method, am, incorporates an adaptive move proposal distribution. For warm starts, sampling can be initiated at a specified location in parameter space and with a multivariate Gaussian proposal distribution defined initially by a specified covariance matrix. Multiple chains can be generated in parallel using a computer cluster. We demonstrate that am can be used to successfully solve real-world Bayesian inference problems, including forecasting of new Coronavirus Disease 2019 case detection with Bayesian quantification of forecast uncertainty. Availability and implementation PyBNF version 1.1.9, the first stable release with am, is available at PyPI and can be installed using the pip package-management system on platforms that have a working installation of Python 3. PyBNF relies on libRoadRunner and BioNetGen for simulations (e.g. numerical integration of ordinary differential equations defined in SBML or BNGL files) and Dask.Distributed for task scheduling on Linux computer clusters. The Python source code can be freely downloaded/cloned from GitHub and used and modified under terms of the BSD-3 license (https://github.com/lanl/pybnf). Online documentation covering installation/usage is available (https://pybnf.readthedocs.io/en/latest/). A tutorial video is available on YouTube (https://www.youtube.com/watch?v=2aRqpqFOiS4&t=63s). Supplementary information Supplementary data are available at Bioinformatics online.

59 BASIC BIOLOGICAL SCIENCES↗

Electron Trapping in Shear Alfven Waves that Power the Aurora

Results from 1D Vlasov drift-kinetic plasma simulations reveal how and where auroral electrons are accelerated along Earth's geomagnetic field. In the warm plasma sheet, electrons become trapped in shear Alfven waves, preventing immediate wave damping. As waves move to regions with larger v{sub Te}/v{sub A}, their parallel electric field decreases, and the trapped electrons escape their influence. The resulting electron distribution functions compare favorably with in situ observations, demonstrating for the first time a self-consistent link between Alfven waves and electrons that form aurora.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

PARVMEC: An Efficient, Scalable Implementation of the Variational Moments Equilibrium Code

The ability to sustain magnetically confined plasma in a state of stable equilibrium is crucial for optimal and cost-effective operations of fusion devices like tokamaks and stellarators. The Variational Moments Equilibrium Code (VMEC) is the de-facto serial application used by fusion scientists to compute magnetohydrodynamics (MHD) equilibria and study the physics of three dimensional plasmas in confined configurations. Modern fusion energy experiments have larger system scales with more interactive experimental workflows, both demanding faster analysis turnaround times on computational workloads that are stressing the capabilities of sequential VMEC. In this paper, we present PARVMEC, an efficient, parallel version of its sequential counterpart, capable of scaling to thousands of processors on distributed memory machines. PARVMEC is a non-linear code, with multiple numerical physics modules, each with its own computational complexity. A detailed speedup analysis supported by scaling results on 1,024 cores of a Cray XC30 supercomputer is presented. Depending on the mode of PARVMEC execution, speedup improvements of one to two orders of magnitude are reported. PARVMEC equips fusion scientists for the first time with a state-of-theart capability for rapid, high fidelity analyses of magnetically confined plasmas at unprecedented scales.

Seal, Sudip K↗

Seamless Transition of Critical Infrastructures using Droop Controlled Grid-forming Inverters

Seamless recovery of power to critical infrastructures, after grid failure, is a crucial need arising in scenarios that are increasingly becoming more frequent. Here, this article proposes a seamless transition strategy using a single and unified mode-dependent droop controlled grid-forming inverters. The control strategy achieves the following objectives: 1) regulates the output active and reactive power by the droop-controlled inverters to a desired value while operating in on-grid mode; 2) seamless transition and recovery of power injections into the load after grid failure by inverters that operates in grid-forming mode all the time; 3) requires only a single bit of information on the grid/network status for the mode transition. A framework for assessing the stability of the system and to guide the choice of parameters for controllers is developed using control-oriented modeling. A controller hardware-in-the-loop-based real-time simulation study on a test system based on the realistic electrical network of a commercial-scale medical center is conducted for initial prototyping of the control strategy. A hardware experiment is conducted with two 3 - $\phi$, 480 -V, 125 -kVA grid-forming inverters, a 3 - $\phi$, 480 -V, 270 -kVA grid simulator, a physical grid switch, and a physical load bank. The experimental data establishes the effectiveness of the always grid-forming operation and control of inverters in meeting power delivery objectives when on-grid and off-grid under various kinds of loads and scenarios while minimizing transients during transitions. Furthermore, performance comparison with existing strategies showcases the advantage of the proposed strategy.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Spin hall effect in semiconductor structures with spatially inhomogeneous spin relaxation

The extrinsic spin Hall effect in samples with coordinate-dependent relaxation time of the spin was studied. It was shown that the spin Hall effect in this case results in not only spatial separation of electrons with different spin projections, but also in the generation of a certain spin; therefore, the total spin's moment of a sample became nonzero under an electric current. A two-layer structure with different times of spin relaxation in layers and a homogeneous sample with linear behavior of the surface's spin relaxation were considered. An expression for the effective spin's relaxation time, which defines the spin polarization distribution in thin films, was derived. It was shown that the uniform spins polarization parallel to the film plane is possible, if the film thickness amounts to several diffusion lengths of the spin.

36 MATERIALS SCIENCE↗

Large-Scale Modeling of Epileptic Seizures: Scaling Properties of Two Parallel Neuronal Network Simulation Algorithms

Our limited understanding of the relationship between the behavior of individual neurons and large neuronal networks is an important limitation in current epilepsy research and may be one of the main causes of our inadequate ability to treat it. Addressing this problem directly via experiments is impossibly complex; thus, we have been developing and studying medium-large-scale simulations of detailed neuronal networks to guide us. Flexibility in the connection schemas and a complete description of the cortical tissue seem necessary for this purpose. In this paper we examine some of the basic issues encountered in these multiscale simulations. We have determined the detailed behavior of two such simulators on parallel computer systems. The observed memory and computation-time scaling behavior for a distributed memory implementation were very good over the range studied, both in terms of network sizes (2,000 to 400,000 neurons) and processor pool sizes (1 to 256 processors). Our simulations required between a few megabytes and about 150 gigabytes of RAM and lasted between a few minutes and about a week, well within the capability of most multinode clusters. Therefore, simulations of epileptic seizures on networks with millions of cells should be feasible on current supercomputers.

59 BASIC BIOLOGICAL SCIENCES↗

Livermore Big Artificial Neural Network Toolkit

LBANN is a toolkit that is designed to train artificial neural networks efficiently on high performance computing architectures. It is optimized to take advantages of key High Performance Computing features to accelerate neural network training. Specifically it is optimized for low-latency, high bandwidth interconnects, node-local NVRAM, node-local GPU accelerators, and high bandwidth parallel file systems. It is built on top of the open source Elemental distributed-memory dense and spars-direct linear algebra and optimization library that is released under the BSD license. The algorithms contained within LBANN are drawn from the academic literature and implemented to work within a distributed-memory framework.

Essen, BrianVan↗

KIVA-hpFE version 1.0.1

KIVA-hpFE is a computer software for solving the physics of multi-species turbulent reactive flow in complex geometries having immersed moving parts. The code is written in Fortran 90/95 and can be used on any computer platform with any popular complier. The code is in 2 versions, a serial version and a parallel version utilizing MPICH2 type Message Passing Interface (MPI or Intel MPI) for solving distributed domains.

Waters, Jiajia↗

A performance model for the communication in fast multipole methods on high-performance computing platforms

Exascale systems are predicted to have approximately 1 billion cores, assuming gigahertz cores. Limitations on affordable network topologies for distributed memory systems of such massive scale bring new challenges to the currently dominant parallel programing model. Currently, there are many efforts to evaluate the hardware and software bottlenecks of exascale designs. It is therefore of interest to model application performance and to understand what changes need to be made to ensure extrapolated scalability. The fast multipole method (FMM) was originally developed for accelerating N-body problems in astrophysics and molecular dynamics but has recently been extended to a wider range of problems. Its high arithmetic intensity combined with its linear complexity and asynchronous communication patterns make it a promising algorithm for exascale systems. In this paper, we discuss the challenges for FMM on current parallel computers and future exascale architectures, with a focus on internode communication. We focus on the communication part only; the efficiency of the computational kernels are beyond the scope of the present study. We develop a performance model that considers the communication patterns of the FMM and observe a good match between our model and the actual communication time on four high-performance computing (HPC) systems, when latency, bandwidth, network topology, and multicore penalties are all taken into account. To our knowledge, this is the first formal characterization of internode communication in FMM that validates the model against actual measurements of communication time. The ultimate communication model is predictive in an absolute sense; however, on complex systems, this objective is often out of reach or of a difficulty out of proportion to its benefit when there exists a simpler model that is inexpensive and sufficient to guide coding decisions leading to improved scaling. The current model provides such guidance.

Computer Science↗

Demonstration of Advanced Experimental and Theoretical Characterization of Hydrogen Dynamics and Associated Behavior in Advanced Reactors

Advanced materials development, manufacturing, and modeling capabilities for innovative reactor designs support nuclear security and mission-focused science through enhanced technology for safer and more efficient and secure production of nuclear energy. The research in this project has established: 1) a state-of-the-art neutron-based hydrogen mapping and cross-section measurement capability as well as detailed crystallographic characterization of hydrogen atoms at LANSCE, and 2) a multi-physics framework for simulating behavior of moderator materials and other material performance in advanced nuclear reactors. Through the course of this project, we successfully developed and demonstrated measurement techniques for hydrogen distribution and atomistic-scale behavior of hydrogen atoms using pulsed neutron techniques. In parallel, advanced multi-physics simulation tools to predict the behavior of hydrogen atoms, e.g. in a moderator for a nuclear reactor, through materials performance, neutron transport, and thermal mechanical behavior were enhanced. Multi-discipline areas across the laboratory were involved in the project as the integration of improved experimental capabilities with enhanced modeling and simulation through MST, NEN, SIGMA, and XCP division subject matter experts.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Role of Parallel Solenoidal Electric Field on Energy Conversion in 2.5D Decaying Turbulence with a Guide Magnetic Field

We perform 2.5D particle-in-cell simulations of decaying turbulence in the presence of a guide (out-of-plane) background magnetic field. The fluctuating magnetic field initially consists of Fourier modes at low wavenumbers (long wavelengths). With time, the electromagnetic energy is converted to plasma kinetic energy (bulk flow+thermal energy) at the rate per unit volume of J· E for current density J and electric field E. Such decaying turbulence is well known to evolve toward a state with strongly intermittent plasma current. Here we decompose the electric field into components that are irrotational, E ir , and solenoidal (divergence-free), E so . E ir is associated with charge separation, and J · E ir is a rate of energy transfer between ions and electrons with little net change in plasma kinetic energy. Therefore, the net rate of conversion of electromagnetic energy to plasma kinetic energy is strongly dominated by J · E so , and for a strong guide magnetic field, this mainly involves the component E so,∥ parallel to the total magnetic field B. We examine various indicators of the spatial distribution of the energy transfer rate J ∥ · E so,∥ , which relates to magnetic reconnection, the best of which are (1) the ratio of the out-of-plane electric field to the in-plane magnetic field, (2) the out-of-plane component of the nonideal electric field, and (3) the magnitude of the estimate of current helicity.

79 ASTRONOMY AND ASTROPHYSICS↗

Relativistic Alfvén Turbulence at Kinetic Scales

Abstract In a strongly magnetized, magnetically dominated relativistic plasma, Alfvénic turbulence can extend to scales much smaller than the particle inertial scales. It leads to an energy cascade somewhat analogous to inertial- or kinetic-Alfvén turbulent cascades existing in nonrelativistic space and astrophysical plasmas. Based on phenomenological modeling and particle-in-cell numerical simulations, we propose that the energy spectrum of such relativistic kinetic-scale Alfvénic turbulence is close to k −3 or slightly steeper than that due to intermittency corrections or Landau damping. We note the analogy of this spectrum with the Kraichnan spectrum corresponding to the enstrophy cascade in 2D incompressible fluid turbulence. Such turbulence strongly energizes particles in the direction parallel to the background magnetic field, leading to nearly one-dimensional particle momentum distributions. We find that these distributions have universal log-normal statistics.

79 ASTRONOMY AND ASTROPHYSICS↗

Parallel Heuristics for Scalable Community Detection

Community detection has become a fundamental operation in numerous graph-theoretic applications. It is used to reveal natural divisions that exist within real world networks without imposing prior size or cardinality constraints on the set of communities. Despite its potential for application, there is only limited support for community detection on large-scale parallel computers, largely owing to the irregular and inherently sequential nature of the underlying heuristics. In this paper, we present parallelization heuristics for fast community detection using the Louvain method as the serial template. The Louvain method is an iterative heuristic for modularity optimization. Originally developed by Blondel et al. in 2008, the method has become increasingly popular owing to its ability to detect high modularity community partitions in a fast and memory-efficient manner. However, the method is also inherently sequential, thereby limiting its scalability to problems that can be solved on desktops. Here, we observe certain key properties of this method that present challenges for its parallelization, and consequently propose multiple heuristics that are designed to break the sequential barrier. Our heuristics are agnostic to the underlying parallel architecture. For evaluation purposes, we implemented our heuristics on shared memory (OpenMP) and distributed memory (MapReduce-MPI) machines, and tested them over real world graphs derived from multiple application domains (internet, biological, natural language processing). Experimental results demonstrate the ability of our heuristics to converge to high modularity solutions comparable to those output by the serial algorithm in nearly the same number of iterations, while also drastically reducing time to solution.

clustering, graph algorithms↗

Livermore Big Artificial Neural Network Toolkit

LBANN is a toolkit that is designed to train artificial neural networks efficiently on high performance computing architectures. It is optimized to take advantages of key High Performance Computing features to accelerate neural network training. Specifically it is optimized for low-latency, high bandwidth interconnects, node-local NVRAM, node-local GPU accelerators, and high bandwidth parallel file systems. It is built on top of the open source Elemental distributed-memory dense and spars-direct linear algebra and optimization library that is released under the BSD license. The algorithms contained within LBANN are drawn from the academic literature and implemented to work within a distributed-memory framework.

Essen, Brian Van↗

SU-E-T-388: Evaluation of Electronic Brachytherapy Dose Distributions in Tissue Equivalent Materials

Purpose: To study the measured and calculated dose distributions for electronic brachytherapy (EBT) in various tissue equivalent homogenous materials. Methods: Calculated dose distributions in water were generated using published TG-43 parameters in Varian BrachyVision software for a 50 kVp, 50 cm Xoft source. Dose distributions were measured within a 3D-scanning tank using dosimeters including: PTW 0.125 cc, pin-point, and parallel-plate ion chambers, Sun Nuclear “Edge” diode and Gafchromic EBT3 film. Multi-channel film dosimetry was used in film analysis. EBT3 film curves were calibrated via radial dose comparison to both independently measured and published data. The resulting film calibration was utilized to measure dose distributions created by titanium filtered source utilized in clinical brachytherapy applications. Data was collected within homogenous PMMA, vinyl, polystyrene, paraffin, and water-equivalent plastic phantoms. Results: Ion-chamber data was corrected to effective points of measurement and normalized prior to comparison between calculated and measured dose distributions. Measurements made in water and water equivalent materials compared well with results from treatment planning software. The maximum percent differences (relative to water) observed between 1 cm and 3.5 cm depth from source for each of the phantom materials are as follows: PMMA 35%, polystyrene 41%, plastic-water 23%, vinyl 115%, and paraffin 46%. Conclusion: The increased probability of photoelectric interactions occurring within the patient during electronic brachytherapy will emphasize the radiological differences between varying human tissues in dose deposition. These differences can Result in clinically significant dose perturbations and it is therefore recommended to move to a model based dose calculation, as outlined in TG-186, to improve the dosimetric accuracy of low energy EBT.

60 APPLIED LIFE SCIENCES↗

Airborne Single Particle Mass Spectrometers (SPLAT II & miniSPLAT) and New Software for Data Visualization and Analysis in a Geo-Spatial Context

Understanding the effect of aerosols on climate requires knowledge of the size and chemical composition of individual aerosol particles - two fundamental properties that determine an aerosol’s optical properties and ability to serve as cloud condensation or ice nuclei. Here we present miniSPLAT, our new aircraft compatible single particle mass spectrometer, that measures in-situ and in real-time size and chemical composition of individual aerosol particles with extremely high sensitivity, temporal resolution, and sizing precision on the order of a monolayer. miniSPLAT operates in dual data acquisition mode to measure, in addition to single particle size and composition, particle number concentrations, size distributions, density, and asphericity with high temporal resolution. When compared to our previous instrument, SPLAT II, miniSPLAT has been significantly reduced in size, weight, and power consumption without loss in performance. We also present ND-Scope, our newly developed interactive visual analytics software package. ND-Scope is designed to explore and visualize the vast amount of complex, multidimensional data acquired by our single particle mass spectrometers, along with other aerosol and cloud characterization instruments on-board aircraft. We demonstrate that ND-Scope makes it possible to visualize the relationships between different observables and to view the data in a geo-spatial context, using the interactive and fully coupled Google Earth and Parallel Coordinates displays. Here we illustrate the utility of ND-Scope to visualize the spatial distribution of atmospheric particles of different compositions, and explore the relationship between individual particle composition and their activity as cloud condensation nuclei.

54 ENVIRONMENTAL SCIENCES↗

Analysis of LH Launcher Arrays (Like the ITER One) Using the TOPLHA Code

TOPLHA (Torino Polytechnic Lower Hybrid Antenna) code is an innovative tool for the 3D/1D simulation of Lower Hybrid (LH) antennas, i.e. accounting for realistic 3D waveguides geometry and for accurate 1D plasma models, and without restrictions on waveguide shape, including curvature. This tool provides a detailed performances prediction of any LH launcher, by computing the antenna scattering parameters, the current distribution, electric field maps and power spectra for any user-specified waveguide excitation. In addition, a fully parallelized and multi-cavity version of TOPLHA permits the analysis of large and complex waveguide arrays in a reasonable simulation time. A detailed analysis of the performances of the proposed ITER LH antenna geometry has been carried out, underlining the strong dependence of the antenna input parameters with respect to plasma conditions. A preliminary optimization of the antenna dimensions has also been accomplished. Electric current distribution on conductors, electric field distribution at the interface with plasma, and power spectra have been calculated as well. The analysis shows the strong capabilities of the TOPLHA code as a predictive tool and its usefulness to LH launcher arrays detailed design.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗