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 1,081 records · Page 60

Simulation of edge localized mode heat pulse using drift-kinetic ions and Boltzmann electrons

This article reports on 1D+2V heat pulse propagation studies using the COGENT guiding center kinetic code. The model uses magnetized kinetic ions and a simple Boltzmann electron model. Results agree with previous kinetic and fluid modeling benchmark studies that correspond to the parameters of edge localized modes (ELMs) observed on the JET tokamak. The plasma parameters for the edge pedestal and ensuing ELM dynamics are in the low collisionality regime. Hence, the dominant balance between the assumed Maxwellian ELM source and collisionless parallel advection causes the ion PDF to develop a significantly anisotropic velocity distribution. Adding nonlinear Coulomb ion-ion collisions to the model acts to smooth the sharp features of the ion distribution function, but the anisotropy remains robust due to the low collisionality.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

The non-random walk of chiral magnetic charge carriers in artificial spin ice

The flow of magnetic charge carriers (dubbed magnetic monopoles) through frustrated spin ice lattices, governed simply by Coulombic forces, represents a new direction in electromagnetism. Artificial spin ice nanoarrays realise this effect at room temperature, where the magnetic charge is carried by domain walls. Control of domain wall path is one important element of utilizing this new medium. By imaging the transit of domain walls across different connected 2D honeycomb structures we contribute an important aspect which will enable that control to be realized. Although apparently equivalent paths are presented to a domain wall as it approaches a Y-shaped vertex from a bar parallel to the field, we observe a stark non-random path distribution, which we attribute to the chirality of the magnetic charges. These observations are supported by detailed statistical modelling and micromagnetic simulations. The identification of chiral control to magnetic charge path selectivity invites analogy with spintronics.

36 MATERIALS SCIENCE↗

Transport induced by ion cyclotron range of frequencies waves

The Vlasov equation, which includes the effect of the ion cyclotron range of frequencies (ICRF) waves, can be written as the Fokker-Planck equation which describes the quasilinear transport in phase space by using the Lie-transform method. The radial transport fluxes of particle, energy and parallel momentum driven by ICRF waves in the slab geometry have been derived. The results show that the ICRF-induced radial redistributions of particle, energy and parallel momentum are driven by the inhomogeneity in energy of the equilibrium distribution function, and related to the correlation between the excursion in the real space and the excursion in energy. For the case with strong asymmetry of k{sub y} spectrum, the ICRF-induced radial transport driven by the energy inhomogeneity dominates the ICRF-induced radial transport driven by the spatial inhomogeneity.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Mean and extreme radio properties of quasars and the origin of radio emission

We investigate the evolution of both the radio-loud fraction (RLF) and (using stacking analysis) the mean radio loudness of quasars. We consider how these properties evolve as a function of redshift and luminosity, black hole (BH) mass and accretion rate, and parameters related to the dominance of a wind in the broad emission-line region. We match the FIRST source catalog to samples of luminous quasars (both spectroscopic and photometric), primarily from the Sloan Digital Sky Survey. After accounting for catastrophic errors in BH mass estimates at high redshift, we find that both the RLF and the mean radio luminosity increase for increasing BH mass and decreasing accretion rate. Similarly, both the RLF and mean radio loudness increase for quasars that are argued to have weaker radiation line driven wind components of the broad emission-line region. In agreement with past work, we find that the RLF increases with increasing optical/UV luminosity and decreasing redshift, while the mean radio loudness evolves in the exact opposite manner. This difference in behavior between the mean radio loudness and the RLF in L−z may indicate selection effects that bias our understanding of the evolution of the RLF; deeper surveys in the optical and radio are needed to resolve this discrepancy. Finally, we argue that radio-loud (RL) and radio-quiet (RQ) quasars may be parallel sequences, but where only RQ quasars at one extreme of the distribution are likely to become RL, possibly through slight differences in spin and/or merger history.

79 ASTRONOMY AND ASTROPHYSICS↗

Measurements and modeling of type-I and type-II ELMs heat flux to the DIII-D divertor

Type-I and type-II edge-localized-modes (ELMs) heat flux profiles measured at the DIII-D divertor feature a peak in the vicinity of the strike-point and a plateau in the scrape-off-layer (SOL), which extends to the first wall. The plateau is present in attached and detached divertors and it is found to originate with plasma bursts upstream in the SOL. The integrated ELM heat flux is distributed at ~65% in the peak and ~35% in this plateau. The parallel loss model, currently used at ITER to predict power loads to the walls, is benchmarked using these results in the primary and secondary divertors with unprecedented constraints using experimental input data for ELM size, radial velocity, energy, electron temperature and density, heat flux footprints and number of filaments. The model can reproduce the experimental near-SOL peak within ~20%, but cannot match the SOL plateau. Employing a two-component approach for the ELM radial velocity, as guided by intermittent data, the full radial heat flux profile can be well matched. The ELM-averaged radial velocity at the separatrix, which explains profile widening, increases from ~0.2 km s –1 in attached to ~0.8 km s –1 in detached scenarios, as the ELM filaments' path becomes electrically disconnected from the sheath at the target. The results presented here indicate filaments fragmentation as a possible mechanism for ELM transport to the far-SOL and provide evidence on the beneficial role of detachment to mitigate ELM flux in the divertor far-SOL. However, these findings imply that wall regions far from the strike points in future machines should be designed to withstand significant heat flux, even for small-ELM regimes.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

On how avalanches penetrate the SOL and broaden heat loads

Recent experiments reported a correlation between power law core temperature spectra and Dα emission, suggesting that heat avalanches penetrate the SOL. This paper derives a threshold criterion for avalanche penetration using a reduced model. Avalanches with ($\bigtriangledown$$\tilde{T}$) rms > $\bigtriangledown$$\tilde{T}$ crit at the separatrix are predicted to penetrate, and so broaden the SOL and heat load distribution. $\bigtriangledown$$\tilde{T}$ crit is ~ 1/$τ$ ∥ , where $τ$ ∥ is the parallel heat flow time through the SOL. Penetration occurs when avalanches are strong enough to steepen sufficiently to shock at the separatrix. A positive correlation is found between the nonlinear drive for steepening and the penetration depth. In particular, penetration depth exceeds that of the heuristic drift limit when shocks form. Implications for numerical and physical experiments are also discussed.

SOL heat load↗

HipMCL: a high-performance parallel implementation of the Markov clustering algorithm for large-scale networks

Biological networks capture structural or functional properties of relevant entities such as molecules, proteins or genes. Characteristic examples are gene expression networks or protein–protein interaction networks, which hold information about functional affinities or structural similarities. Such networks have been expanding in size due to increasing scale and abundance of biological data. While various clustering algorithms have been proposed to find highly connected regions, Markov Clustering (MCL) has been one of the most successful approaches to cluster sequence similarity or expression networks. Despite its popularity, MCL’s scalability to cluster large datasets still remains a bottleneck due to high running times and memory demands. In this paper, we present High-performance MCL (HipMCL), a parallel implementation of the original MCL algorithm that can run on distributed-memory computers. We show that HipMCL can efficiently utilize 2000 compute nodes and cluster a network of ~70 million nodes with ~68 billion edges in ~2.4 h. By exploiting distributed-memory environments, HipMCL clusters large-scale networks several orders of magnitude faster than MCL and enables clustering of even bigger networks. Finally, HipMCL is based on MPI and OpenMP and is freely available under a modified BSD license.

59 BASIC BIOLOGICAL SCIENCES↗

Microbunching Instability in a Chicane: Two-Dimensional Mean Field Treatment

We study the microbunching instability in a bunch compressor by a parallel code with some improved numerical algorithms. The two-dimensional charge/current distribution is represented by a Fourier series, with coefficients determined through Monte Carlo sampling over an ensemble of tracked points. This gives a globally smooth distribution with low noise. The field equations are solved accurately in the lab frame using retarded potentials and a novel choice of integration variables that eliminates singularities. We apply the scheme with parameters for the first bunch compressor system of FERMI{at}Elettra, with emphasis on the amplification of a perturbation at a particular wavelength. Gain curves agree with those of the linearized Vlasov model at long wavelengths, but show some deviation at the smallest wavelengths treated.

43 PARTICLE ACCELERATORS↗

Design and characterization of a new high-dose-rate brachytherapy Valencia applicator for larger skin lesions

Purpose: The aims of this study were (i) to design a new high-dose-rate (HDR) brachytherapy applicator for treating surface lesions with planning target volumes larger than 3 cm in diameter and up to 5 cm in size, using the microSelectron-HDR or Flexitron afterloader (Elekta Brachytherapy) with a {sup 192}Ir source; (ii) to calculate by means of the Monte Carlo (MC) method the dose distribution for the new applicator when it is placed against a water phantom; and (iii) to validate experimentally the dose distributions in water. Methods: The PENELOPE2008 MC code was used to optimize dwell positions and dwell times. Next, the dose distribution in a water phantom and the leakage dose distribution around the applicator were calculated. Finally, MC data were validated experimentally for a {sup 192}Ir mHDR-v2 source by measuring (i) dose distributions with radiochromic EBT3 films (ISP); (ii) percentage depth–dose (PDD) curve with the parallel-plate ionization chamber Advanced Markus (PTW); and (iii) absolute dose rate with EBT3 films and the PinPoint T31016 (PTW) ionization chamber. Results: The new applicator is made of tungsten alloy (Densimet) and consists of a set of interchangeable collimators. Three catheters are used to allocate the source at prefixed dwell positions with preset weights to produce a homogenous dose distribution at the typical prescription depth of 3 mm in water. The same plan is used for all available collimators. PDD, absolute dose rate per unit of air kerma strength, and off-axis profiles in a cylindrical water phantom are reported. These data can be used for treatment planning. Leakage around the applicator was also scored. The dose distributions, PDD, and absolute dose rate calculated agree within experimental uncertainties with the doses measured: differences of MC data with chamber measurements are up to 0.8% and with radiochromic films are up to 3.5%. Conclusions: The new applicator and the dosimetric data provided here will be a valuable tool in clinical practice, making treatment of large skin lesions simpler, faster, and safer. Also the dose to surrounding healthy tissues is minimal.

60 APPLIED LIFE SCIENCES↗

Scalable deep text comprehension for Cancer surveillance on high-performance computing

Background: Deep Learning (DL) has advanced the state-of-the-art capabilities in bioinformatics applications which has resulted in trends of increasingly sophisticated and computationally demanding models trained by larger and larger data sets. This vastly increased computational demand challenges the feasibility of conducting cutting-edge research. One solution is to distribute the vast computational workload across multiple computing cluster nodes with data parallelism algorithms. In this study, we used a High-Performance Computing environment and implemented the Downpour Stochastic Gradient Descent algorithm for data parallelism to train a Convolutional Neural Network (CNN) for the natural language processing task of information extraction from a massive dataset of cancer pathology reports. We evaluated the scalability improvements using data parallelism training and the Titan supercomputer at Oak Ridge Leadership Computing Facility. To evaluate scalability, we used different numbers of worker nodes and performed a set of experiments comparing the effects of different training batch sizes and optimizer functions.Results: We found that Adadelta would consistently converge at a lower validation loss, though requiring over twice as many training epochs as the fastest converging optimizer, RMSProp. The Adam optimizer consistently achieved a close 2nd place minimum validation loss significantly faster; using a batch size of 16 and 32 allowed the network to converge in only 4.5 training epochs.Conclusions: We demonstrated that the networked training process is scalable across multiple compute nodes communicating with message passing interface while achieving higher classification accuracy compared to a traditional machine learning algorithm.

Qiu, John X.↗

Asynchronous Iterative Solvers for Extreme-Scale Computing

The Asynchronous Iterative Solvers for Extreme-Scale Computing (AsyncIS) project aims to explore more efficient numerical algorithms by decreasing their overhead. AsyncIS does this by replacing the outer Krylov subspace solver with an asynchronous optimized Schwarz method, thereby removing the global synchronization and bulk synchronous operations typically used in numerical codes. AsyncIS—a U.S. Department of Energy (DOE)-funded collaboration between Georgia Tech, the University of Tennessee, Knoxville, Temple University, and Sandia National Laboratories—also focuses on the development and optimization of asynchronous preconditioners (i.e., preconditioners that are generated and/or applied in an asynchronous fashion). The novel preconditioning algorithms that provide fine-grained parallelism enable preconditioned Krylov solvers to run efficiently on large-scale distributed systems and manycore accelerators like GPUs.

97 MATHEMATICS AND COMPUTING↗

Semantics-based distributed I/O for mpiBLAST.

BLAST is a widely used software toolkit for genomic sequence search. mpiBLAST is a freely available, open-source parallelization of BLAST that uses database segmentation to allow different worker processes to search (in parallel) unique segments of the database. After searching, the workers write their output to a filesystem. While mpiBLAST has been shown to achieve high performance in clusters with fast local filesystems, its I/O processing remains a concern for scalability, especially in systems having limited I/O capabilities such as distributed filesystems spread across a wide-area network. Thus, we present ParaMEDIC - a novel environment that uses application-specific semantic information to compress I/O data and improve performance in distributed environments. Specifically, for mpiBLAST, ParaMEDIC partitions worker processes into compute and I/O workers. Compute workers, instead of directly writing the output to the filesystem, the workers process the output using semantic knowledge about the application to generate metadata and write the metadata to the filesystem. I/O workers, which physically reside closer to the actual storage, then process this metadata to re-create the actual output and write it to the filesystem. This approach allows ParaMEDIC to reduce I/O time, thus accelerating mpiBLAST by as much as 25-fold.

99 GENERAL AND MISCELLANEOUS↗

An Analytical Time–Domain Expression for the Net Ripple Produced by Parallel Interleaved Converters

We apply modular arithmetic and Fourier series to analyze the superposition of N interleaved triangular waveforms with identical amplitudes and duty-ratios. Here, interleaving refers to the condition when a collection of periodic waveforms with identical periods are each uniformly phase-shifted across one period. The main result is a time-domain expression which provides an exact representation of the summed and interleaved triangular waveforms, where the peak amplitude and parameters of the time-periodic component are all specified in closed-form. Analysis is general and can be used to study various applications in multi-converter systems. This model is unique not only in that it reveals a simple and intuitive expression for the net ripple, but its derivation via modular arithmetic and Fourier series is distinct from prior approaches. The analytical framework is experimentally validated with a system of three parallel converters under time-varying operating conditions.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Paxillin localisation in osteocytes-Is it determined by the direction of loading?

External mechanical loading of cells aligns cytoskeletal stress fibres in the direction of principle strains and localises paxillin to the mechanosensing region. If the osteocyte cell body can indeed directly sense matrix strains, then cytoskeletal alignment and distribution of paxillin in osteocytes in situ will bear alignment to the different mechanical loading patterns in fibulae and calvariae. We used confocal microscopy to visualise the immunofluorescence-labelled actin cytoskeleton in viable osteocytes and paxillin distribution in fixated osteocytes in situ. In fibular osteocyte cell bodies, actin cytoskeleton and nuclei were elongated and aligned parallel to the principal (longitudinal) mechanical loading direction. Paxillin was localised to the 'poles' of elongated osteocyte cell bodies. In calvarial osteocyte cell bodies, actin cytoskeleton and nuclei were relatively more round. Paxillin was distributed evenly in the osteocyte cell bodies. Thus in osteocyte cell bodies in situ, the external mechanical loading pattern likely determines the orientation of the actin cytoskeleton, and focal adhesions mediate direct mechanosensation of matrix strains.

60 APPLIED LIFE SCIENCES↗

Designing a parallel Feel-the-Way clustering algorithm on HPC systems

This paper introduces a new parallel clustering algorithm, named Feel-the-Way clustering algorithm, that provides better or equivalent convergence rate than the traditional clustering methods by optimizing the synchronization and communication costs. Our algorithm design centers on how to optimize three factors simultaneously: reduced synchronizations, improved convergence rate, and retained same or comparable optimization cost. To compare the optimization cost, we use the Sum of Square Error (SSE) cost as the metric, which is the sum of the square distance between each data point and its assigned clusters. Compared with the traditional MPI k-means algorithm, the new Feel-the-Way algorithm requires less communications among participating processes. As for the convergence rate, the new algorithm requires fewer number of iterations to converge. As for the optimization cost, it obtains the SSE costs that are close to the k-means algorithm. In the paper, we first design the full-step Feel-the-Way k-means clustering algorithm that can significantly reduce the number of iterations that are required by the original k-means clustering method. Next, we improve the performance of the full-step algorithm by adopting an optimized sampling-based approach, named reassignment-history-aware sampling. Our experimental results show that the optimized sampling-based Feel-the-Way method is significantly faster than the widely used k-means clustering method, and can provide comparable optimization costs. More extensive experiments with several synthetic datasets and real-world datasets (e.g., MNIST, CIFAR-10, ENRON, and PLACES-2) show that the new parallel algorithm can outperform the open source MPI k-means library by up to 110% on a high-performance computing system using 4,096 CPU cores. In addition, the new algorithm can take up to 51% fewer iterations to converge than the k-means clustering algorithm.

97 MATHEMATICS AND COMPUTING↗

Dissociative double-photoionization of butadiene in the 25-45 eV energy range using 3-D multi-coincidence ion momentum imaging spectrometry

Dissociative double-photoionization of butadiene in the 25-45 eV energy range has been studied with tunable synchrotron radiation using full three-dimensional ion momentum imaging. Using ab initio calculations, the electronic states of the molecular dication below 33 eV are identified. The results of the measurement and calculation show that double ionization from π orbitals selectively triggers twisting about the terminal or central C–C bonds. We show that this conformational rearrangement depends upon the dication electronic state, which effectively acts as a gateway for the dissociation reaction pathway. For photon energies above 33 eV, three-body dissociation channels where neutral H-atom evaporation precedes C–C charge-separation in the dication species appear in the correlation map. The fragment angular distributions support a model where the dication species is initially aligned with the molecular backbone parallel to the polarization vector of the light, indicating a high probability for double-ionization to the “gateway states” for molecules with this orientation.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Geodesic mode instability driven by electron and ion fluxes in tokamaks

The effect of the parallel electron current and plasma flux on Geodesic Acoustic Modes (GAM) in a tokamak is analyzed by kinetic theory taking into the account the ion Landau damping and diamagnetic drifts. It is shown that the electron current and plasma flow, modeled by shifted Maxwell distributions of electrons and ions, may overcome the ion Landau damping generating the GAM instability when the parallel electron current velocity is larger than the effective parallel GAM phase velocity of sidebands, Rqω. The instability is driven by the electron current and the parallel ion flux cross term. Possible applications to tokamak experiments are discussed. The existence of the geodesic ion sound mode due to plasma flow is shown.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Ultra-low emittance beam generation using two-color ionization injection in a CO 2 laser-driven plasma accelerator

Ultra-low emittance (tens of nm) beams can be generated in a plasma accelerator using ionization injection of electrons into a wakefield. An all-optical method of beam generation uses two laser pulses of different colors. A long-wavelength drive laser pulse (with a large ponderomotive force and small peak electric field) is used to excite a large wakefield without fully ionizing a gas, and a short-wavelength injection laser pulse (with a small ponderomotive force and large peak electric field), co-propagating and delayed with respect to the pump laser, to ionize a fraction of the remaining bound electrons at a trapped wake phase, generating an electron beam that is accelerated in the wake. The trapping condition, the ionized electron distribution, and the trapped bunch dynamics are discussed. Expressions for the beam transverse emittance, parallel and orthogonal to the ionization laser polarization, are presented. An example is shown using a 10-micron CO 2 laser to drive the wake and a frequency-doubled Ti:Al 2 O 3 laser for ionization injection.

43 PARTICLE ACCELERATORS↗