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 1,009 records · Page 56

Optimal design of 3D borehole seismic arrays for microearthquake monitoring in anisotropic media during stimulations in the EGS collab project

Multiple U.S. national laboratories, universities and industrial collaborators are conducting collaborative research under the EGS Collab project supported by the U.S. Department of Energy, to understand the fracture creation and imaging during fracturing in enhanced geothermal systems. Microearthquake hypocenter locations and focal mechanisms are used to monitor hydraulic fracturing growth and characterization at the EGS Collab experimental site at the Sanford Underground Research Facility using seismic receivers in multiple monitoring wells. In this study, we develop a methodology for optimal design a 3D borehole seismic array for cost-effective seismic monitoring in anisotropic media using not only the relationship between receiver distributions and standard deviation errors of microearthquake hypocenter locations, but also that between receiver distributions and focal-mechanism inversion errors. Our results indicate that microearthquake hypocenter locations and focal mechanisms can be reasonably well reconstructed for the EGS Collab Experiment I using six monitoring wells, including four fracture-parallel monitoring wells and two orthogonal wells. Lastly, eight seismic receivers evenly distributed in four parallel monitoring wells or twelve receivers in all six monitoring wells are required for hypocenter location, and twelve receivers evenly distributed in six wells or sixteen receivers in four wells are needed for focal-mechanism inversion.

58 GEOSCIENCES↗

Hierarchical power distribution in large scale computing systems

Techniques for managing power distribution amongst processors in a massively parallel computer architecture are disclosed. The techniques utilize a hierarchy that organizes the various processors of the massively parallel computer architecture. The hierarchy groups numbers of the processors at the lowest level. When processors complete tasks, the power assigned to those processors is distributed to other processors in the same group so that the performance of those processors can be increased. Hierarchical organization simplifies the calculations required for determining how and when to distribute power, because when tasks are complete and power is available for distribution, a relatively small number of processors are available for consideration to receive that power. The number of processors that are grouped together can be adjusted in real time based on performance factors to improve the trade-off between calculation speed and power distribution efficacy.

42 ENGINEERING↗

Chemical and Structural Alterations in the Amorphous Structure of Obsidian due to Nanolites

Obsidian is volcanic glass that results from the rapid cooling of silica-rich melt. Nanoscale crystallites precipitate out of the melt prior to solidification and remain embedded in the amorphous matrix. These crystallites provide information on the flow kinetics and composition of the melt. Due to the sparsity and size of nanolites, studies often focus on supramicron crystallites. This research takes advantage of the conchoidal fracture of obsidian by knapping samples with nanometer-thin edges for transmission electron microscopy characterization. Nanolites in the amorphous matrix are studied using energy-dispersive spectroscopy (EDS) and electron diffraction. Certain alkali and alkaline-earth cations exhibit patterns of depletion near Fe-oxide nanolites. EDS is used to identify nanolites and variations in the composition of the matrix. Parallel beam diffraction and radial distribution function analysis of nearest-neighbor distances determine average bond lengths in the matrix near nanolites, showing that nanolites influence the nearby short-range ordering and atomic character of the matrix. Analysis reveals decreased mean nearest-neighbor distances in the matrix adjacent to nanolites compared to the bulk. Our methods exhibit the required sensitivity to detect variations in the composition and structure near nanolites, and our findings indicate that obsidian nanolites contribute to quantifiable localized changes in the amorphous structure.

36 MATERIALS SCIENCE↗

Π4U: A high performance computing framework for Bayesian uncertainty quantification of complex models

We present Π4U,{sup 1} an extensible framework, for non-intrusive Bayesian Uncertainty Quantification and Propagation (UQ+P) of complex and computationally demanding physical models, that can exploit massively parallel computer architectures. The framework incorporates Laplace asymptotic approximations as well as stochastic algorithms, along with distributed numerical differentiation and task-based parallelism for heterogeneous clusters. Sampling is based on the Transitional Markov Chain Monte Carlo (TMCMC) algorithm and its variants. The optimization tasks associated with the asymptotic approximations are treated via the Covariance Matrix Adaptation Evolution Strategy (CMA-ES). A modified subset simulation method is used for posterior reliability measurements of rare events. The framework accommodates scheduling of multiple physical model evaluations based on an adaptive load balancing library and shows excellent scalability. In addition to the software framework, we also provide guidelines as to the applicability and efficiency of Bayesian tools when applied to computationally demanding physical models. Theoretical and computational developments are demonstrated with applications drawn from molecular dynamics, structural dynamics and granular flow.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Mapping Dense LU Factorization on Multicore Supercomputer Nodes

Dense LU factorization is a prominent benchmark used to rank the performance of supercomputers. Many implementations use block-cyclic distributions of matrix blocks onto a two-dimensional process grid. The process grid dimensions drive a trade-off between communication and computation and are architecture- and implementation-sensitive. The critical panel factorization steps can be made less communication-bound by overlapping asynchronous collectives for pivoting with the computation of rank-k updates. By shifting the computation-communication trade-off, a modified block-cyclic distribution can beneficially exploit more available parallelism on the critical path, and reduce panel factorization's memory hierarchy contention on now-ubiquitous multicore architectures. During active panel factorization, rank-1 updates stream through memory with minimal reuse. In a column-major process grid, the performance of this access pattern degrades as too many streaming processors contend for access to memory. A block-cyclic mapping in the row-major order does not encounter this problem, but consequently sacrifices node and network locality in the critical pivoting steps. We introduce 'striding' to vary between the two extremes of row- and column-major process grids. The maximum available parallelism in the critical path work (active panel factorization, triangular solves, and subsequent broadcasts) is bounded by the length or width of the process grid. Increasing one dimension of the process grid decreases the number of distinct processes and nodes in the other dimension. To increase the harnessed parallelism in both dimensions, we start with a tall process grid. We then apply periodic 'rotation' to this grid to restore exploited parallelism along the row to previous levels. As a test-bed for further mapping experiments, we describe a dense LU implementation that allows a block distribution to be defined as a general function of block to processor. Other mappings can be tested wit...

Computer Science↗

Dynamically Scheduled Region-Based Image Compositing

Algorithms for sort-last parallel volume rendering on large distributed memory machines usually divide a dataset equally across all nodes for rendering. Depending on the features that a user wants to see in a dataset, all the nodes will rarely finish rendering at the same time. Here, existing compositing algorithms do not often take this into consideration, which can lead to significant delays when nodes that are compositing wait for other nodes that are still rendering. In this paper, we present an image compositing algorithm that uses spatial and temporal awareness to dynamically schedule the exchange of regions in an image and progressively composite images as they become available. Running on the Edison supercomputer at NERSC, we show that a scheduler-based algorithm with awareness of the spatial contribution from each rendering node can outperform traditional image compositing algorithms.

97 MATHEMATICS AND COMPUTING↗

Parallel adaptive mesh refinement within the PUMAA3D Project

To enable the solution of large-scale applications on distributed memory architectures, we are designing and implementing parallel algorithms for the fundamental tasks of unstructured mesh computation. In this paper, we discuss efficient algorithms developed for two of these tasks: parallel adaptive mesh refinement and mesh partitioning. The algorithms are discussed in the context of two-dimensional finite element solution on triangular meshes, but are suitable for use with a variety of element types and with h- or p-refinement. Results demonstrating the scalability and efficiency of the refinement algorithm and the quality of the mesh partitioning are presented for several test problems on the Intel DELTA.

Freitag, Lori↗

SU-F-T-58: Dosimetric Evaluation of Breast Tissue Composition for Electronic Brachytherapy (BET) Source In High Dose Rate Accelerated Partial Breast (APBI) Irradiation

Purpose: To quantitatively evaluate the dosimetric impact of differing breast tissue compositions for electronic brachytherapy source for high dose rate accelerated partial breast irradiation. Methods: A series of Monte Carlo Simulation were created using the GEANT4 toolkit (version 10.0). The breast phantom was modeled as a semi-circle with a radius of 5.0 cm. A water balloon with a radius of 1.5 cm was located in the phantom with the Xoft AxxentTM EBT source placed at center as a point source. A mixed of two tissue types (adipose and glandular tissue) was assigned as the materials for the breast phantom with different weight ratios. The proportionality of glandular and adipose tissue was simulated in four different fashions, 80/20, 70/30, 50/50 and 30/70 respectively. The custom energy spectrum for the 50 kVp XOFT source was provided via the manufacturer and used to generate incident photons. The dose distributions were recorded using a parallel three dimensional mesh with a size of 30 × 30 × 30 cm3 with 1 × 1 × 1 mm3 voxels. The simulated doses absorbed along the transverse axis were normalized at the distance of 1 cm and then compared with the calculations using standard TG-43 formalism. Results: All simulations showed underestimation of dose beyond balloon surface compared to standard TG-43 calculations. The maximum percentage differences within 2 cm distance from balloon surface were found to be 18%, 11%, 10% and 8% for the fat breast (30/70), standard breast (50/50), dense breast (70/30 and 80/20), respectively. Conclusion: The accuracy of dose calculations for low energy EBT source was limited when considering tissue heterogeneous composition. The impact of atomic number on photo-electric effect for lower energy Brachytherapy source is not accounted for and resulting in significant errors in dose calculation.

60 APPLIED LIFE SCIENCES↗

Nongyrotropic Electrons in Guide Field Reconnection

We apply a scalar measure of nongyrotropy to the electron pressure tensor in a 2D particle-in-cell simulation of guide field reconnection and assess the corresponding electron distributions and the forces that account for the nongyrotropy. The scalar measure reveals that the nongyrotropy lies in bands that straddle the electron diffusion region and the separatrices, in the same regions where there are parallel electric fields. Analysis of electron distributions and fields shows that the nongyrotropy along the inflow and outflow separatrices emerges as a result of multiple populations of electrons influenced differently by large and small-scale parallel electric fields and by gradients in the electric field. The relevant parallel electric fields include large-scale potential ramps emanating from the x-line and sub-ion inertial scale bipolar electron holes. Gradients in the perpendicular electric field modify electrons differently depending on their phase, thus producing nongyrotropy. Magnetic flux violation occurs along portions of the separatrices that coincide with the parallel electric fields. An inductive electric field in the electron EB drift frame thus develops, which has the effect of enhancing nongyrotropies already produced by other mechanisms and under certain conditions producing their own nongyrotropy. Particle tracing of electrons from nongyrotropic populations along the inflows and outflows shows that the striated structure of nongyrotropy corresponds to electrons arriving from different source regions. We also show that the relevant parallel electric fields receive important contributions not only from the nongyrotropic portion of the electron pressure tensor but from electron spatial and temporal inertial terms as well.

Parallel Electric Fields↗

Ultrahigh-order Maxwell solver with extreme scalability for electromagnetic PIC simulations of plasmas

The advent of massively parallel supercomputers, with their distributed-memory technology using many processing units, has favored the development of highly-scalable local low-order solvers at the expense of harder-to-scale global very high-order spectral methods. Indeed, FFT-based methods, which were very popular on shared memory computers, have been largely replaced by finite-difference (FD) methods for the solution of many problems, including plasmas simulations with electromagnetic Particle-In-Cell methods. For some problems, such as the modeling of so-called “plasma mirrors” for the generation of high-energy particles and ultra-short radiations, we have shown that the inaccuracies of standard FD-based PIC methods prevent the modeling on present supercomputers at sufficient accuracy. Here, we demonstrate here that a new method, based on the use of local FFTs, enables ultrahigh-order accuracy with unprecedented scalability, and thus for the first time the accurate modeling of plasma mirrors in 3D.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

EVOLUTION OF THE WATER MASER EXPANDING SHELL IN W75N VLA 2

We present Very Long Baseline Interferometry (VLBI) observations of 22 GHz H{sub 2}O masers in the high-mass star-forming region of W75N, carried out with VLBI Exploration of Radio Astrometry (VERA) for three epochs in 2007 with an angular resolution of {approx}1 mas. We detected H{sub 2}O maser emission toward the radio jet in VLA 1 and the expanding shell-like structure in VLA 2. The spatial distribution of the H{sub 2}O masers detected with VERA and measured proper motions around VLA 1 and VLA 2 are similar to those found with previous VLBI observations in epochs 1999 and 2005, with the masers in VLA 1 mainly distributed along a linear structure parallel to the radio jet and, on the other hand, forming a shell-like structure around VLA 2. We have made elliptical fits to the VLA 2 H{sub 2}O maser shell-like structure observed in the different epochs (1999, 2005, and 2007), and found that the shell is still expanding eight years after its discovery. From the difference in the size of the semi-major axes of the fitted ellipses in epochs 1999 ({approx_equal}71 {+-} 1 mas), 2005 ({approx_equal}97 {+-} 3 mas), and 2007 ({approx_equal}111 {+-} 1 mas), we estimate an average expanding velocity of {approx}5 mas yr{sup -1}, similar to the proper motions measured in the individual H{sub 2}O maser features. A kinematic age of {approx}20 yr is derived for this structure. In addition, our VERA observations indicate an increase in the ellipticity of the expanding shell around VLA 2 from epochs 1999 to 2007. In fact, the elliptical fit of the VERA data shows a ratio of the minor and major axes of {approx}0.6, in contrast with an almost circular shape for the shell detected in 1999 and 2005 (b/a {approx} 0.9). This suggests that we are probably observing the formation of a jet-driven H{sub 2}O maser structure in VLA2, evolving from a non-collimated pulsed-outflow event during the first stages of evolution of a massive young stellar object (YSO). This may support predictions made earlier by other authors on this issue, consistent with recent magnetohydrodynamical simulations. We discuss possible implications of our results in the study of the first stages of evolution of massive YSOs.

79 ASTRONOMY AND ASTROPHYSICS↗

PETSc/TAO Users Manual (Rev. 3.19)

This manual describes the use of the Portable, Extensible Toolkit for Scientific Computation (PETSc) and the Toolkit for Advanced Optimization (TAO) for the numerical solution of partial differential equations and related problems on high-performance computers. PETSc/TAO is a suite of data structures and routines that provide the building blocks for the implementation of large-scale application codes on parallel (and serial) computers. PETSc uses the MPI standard for all distributed memory communication. PETSc/TAO includes a large suite of parallel linear solvers, nonlinear solvers, time integrators, and opti mization that may be used in application codes written in Fortran, C, C++, and Python (via petsc4py; see Getting Started). PETSc provides many of the mechanisms needed within parallel application codes, such as parallel matrix and vector assembly routines. The library is organized hierarchically, enabling users to employ the level of abstraction that is most appropriate for a particular problem. By using techniques of object-oriented programming, PETSc provides enormous flexibility for users. PETSc is a sophisticated set of software tools; as such, for some users it initially has a much steeper learning curve than packages such as MATLAB or a simple subroutine library. In particular, for individuals without some computer science background, experience programming in C, C++, python, or Fortran and experience using a debugger such as gdb or lldb, it may require a significant amount of time to take full advantage of the features that enable efficient software use. However, the power of the PETSc design and the algorithms it incorporates may make the efficient implementation of many application codes simpler than “rolling them” yourself. For many tasks a package such as MATLAB is often the best tool; PETSc is not intended for the classes of problems for which effective MATLAB code can be written. There are several packages, built on PETSc, that may satisfy your needs without requiring directly using PETSc. We recommend reviewing these packages functionality before starting to code directly with PETSc. PETSc can be used to provide a “MPI parallel linear solver” in an otherwise sequential, or OpenMP parallel code. This approach cannot provide extremely large improvements in the application time by utilizing large numbers of MPI processes but can still improve the performance. Certainly all parts of a previously sequential code need not be parallelized but the matrix generation portion must be parallelized to expect true scalability to large numbers of MPI processes. See PCMPI for details on how to utilize the PETSc MPI linear solver server. Since PETSc is under continued development, small changes in usage and calling sequences of routines will occur. PETSc has been supported for twenty-five years; see mailing list information on our website for information on contacting support.

97 MATHEMATICS AND COMPUTING↗

PETSc/TAO Users Manual V.3.21

This manual describes the use of the Portable, Extensible Toolkit for Scientific Computation (PETSc) and the Toolkit for Advanced Optimization (TAO) for the numerical solution of partial differential equations (PDEs) and related problems on high-performance computers. PETSc/TAO is a suite of data structures and routines that provide the building blocks for implementing large-scale application codes on parallel (and serial) computers. PETSc uses the MPI standard for all distributed memory communication. PETSc/TAO includes a large suite of parallel linear solvers, nonlinear solvers, time integrators, and optimizers that may be used in application codes written in Fortran, C, C++, and Python (via petsc4py; see Getting Started ). The library is organized hierarchically, enabling users to employ the abstraction level most appropriate for a particular problem. By using techniques of object-oriented programming, PETSc provides enormous flexibility for users.

97 MATHEMATICS AND COMPUTING↗

The OVIS analysis architecture.

This report summarizes the current statistical analysis capability of OVIS and how it works in conjunction with the OVIS data readers and interpolators. It also documents how to extend these capabilities. OVIS is a tool for parallel statistical analysis of sensor data to improve system reliability. Parallelism is achieved using a distributed data model: many sensors on similar components (metaphorically sheep) insert measurements into a series of databases on computers reserved for analyzing the measurements (metaphorically shepherds). Each shepherd node then processes the sheep data stored locally and the results are aggregated across all shepherds. OVIS uses the Visualization Tool Kit (VTK) statistics algorithm class hierarchy to perform analysis of each process's data but avoids VTK's model aggregation stage which uses the Message Passing Interface (MPI); this is because if a single process in an MPI job fails, the entire job will fail. Instead, OVIS uses asynchronous database replication to aggregate statistical models. OVIS has several additional features beyond those present in VTK that, first, accommodate its particular data format and, second, improve the memory and speed of the statistical analyses. First, because many statistical algorithms are multivariate in nature and sensor data is typically univariate, interpolation of data is required to provide simultaneous observations of metrics. Note that in this report, we will refer to a single value obtained from a sensor as a measurement while a collection of multiple sensor values simultaneously present in the system is an observation. A base class for interpolation is provided that abstracts the operation of converting multiple sensor measurements into simultaneous observations. A concrete implementation is provided that performs piecewise constant temporal interpolation of multiple metrics across a single component. Secondly, because calculations may summarize data too large to fit in memory OVIS analyses batches of observations at a time and aggregates these intermediate intra-process models as it goes before storing the final model for inter-process aggregation via database replication. This reduces the memory footprint of the analysis, interpolation, and the database client and server query processing. This also interleaves processing with the disk I/O required to fetch data from the database - also improving speed. This report documents how OVIS performs analyses and how to create additional analysis components that fetch measurements from the database, perform interpolation, or perform operations on streamed observations (such as model updates or assessments). The rest of this section outlines the OVIS analysis algorithm and is followed by sections specific to each subtask. Note that we are limiting our discussion for now to the creation of a model from a set of measurements, and not including the assessment of observations using a model. The same framework can be used for assessment but that use case is not detailed in this report.

97 MATHEMATICS AND COMPUTING↗

Electromagnetic cyclotron-loss-cone instability associated with weakly relativistic electrons

The amplification of fast extraordinary mode waves at frequencies very close to the electron cyclotron frequency, due to the presence of a population of energetic electrons with a loss-cone type distribution, is studied. Low-energy background electrons are included in the analysis. Two types of loss-cone distribution functions are considered, and it is found that the maximum growth rates for both distribution functions are of the same order of magnitude. When the thermal effects of the energetic electrons are included in the dispersion equation, the real frequencies of the waves are lower than those obtained by using the cold plasma approximation. This effect tends to enhance the growth rate. An idealized case including a parallel electric field such that the distribution function of the trapped energetic electrons is modified is also considered. It is assumed that the parallel electric field can remove the low-energy background electrons away from the source region of radiation. Both these effects increase the growth rate.

Wong, H. K.↗

PythonFOAM: In-situ data analyses with OpenFOAM and Python

Here, we outline the development of a general-purpose Python-based data analysis tool for OpenFOAM. Our implementation relies on the construction of OpenFOAM applications that have bindings to data analysis libraries in Python. Double precision data in OpenFOAM is cast to a NumPy array using the NumPy C-API and Python modules may then be used for arbitrary data analysis and manipulation on flow-field information. We highlight how the proposed wrapper may be used for an in-situ online singular value decomposition (SVD) implemented in Python and accessed from the OpenFOAM solver PimpleFOAM. Here, 'in-situ' refers to a programming paradigm that allows for a concurrent computation of the data analysis on the same computational resources utilized for the partial differential equation solver. In addition, to demonstrate parallel deployments, we deploy a distributed SVD, which collects snapshot data across the ranks of a distributed simulation to compute the global left singular vectors. Crucially, both OpenFOAM and Python share the same message passing interface (MPI) communicator for this deployment which allows Python objects and functions to exchange NumPy arrays across ranks. Subsequently, we provide scaling assessments of this distributed SVD on multiple nodes of Intel Broadwell and KNL architectures for canonical test cases such as the large eddy simulations of a backward facing step and a channel flow at friction Reynolds number of 395. Finally, we demonstrate the deployment of a deep neural network for compressing the flow-field information using an autoencoder to demonstrate an ability to use state-of-the-art machine learning tools in the Python ecosystem.

97 MATHEMATICS AND COMPUTING↗

Formation of Fe2Al5-xZnx intermetallic crystals at the Fe-Zn interface in hot-dip galvanizing

Highlights: • Interstitial-free steels were galvanized in Zn baths containing 0.14-0.28 wt% Al. • The Fe-Al layer formed in a 0.14 wt% Al bath is composed of equiaxed Fe{sub 2}Al{sub 5-x}Zn{sub x} crystals without preferred orientations. • The layers formed in baths containing 0.18 wt% Al or more are composed of orientated domains. • Each domain contains 4-12 Fe{sub 2}Al{sub 5-x}Zn{sub x} crystal variants of specific orientations. • A layer of a single variant crystal can grow epitaxially on the α-Fe grains as the Al content reaches 0.28 wt%. - Abstract: The formation of Fe{sub 2}Al{sub 5-x}Zn{sub x} inhibition layer at the Fe-Zn interface has been characterized by scanning electron microscopy/electron backscatter diffraction and transmission electron microscopy for interstitial-free steels galvanized in Zn baths containing 0.14–0.28 wt% Al. The inhibition layer formed in the 0.14 wt% Al bath is composed of fine, equiaxed Fe{sub 2}Al{sub 5-x}Zn{sub x} crystals without preferred orientations. The inhibition layers formed in the baths containing 0.18 wt% Al or more are composed of orientated domains and each domain contains 4–12 Fe{sub 2}Al{sub 5-x}Zn{sub x} crystal variants of specific orientations. All the variants follow the same Fe{sub 2}Al{sub 5-x}Zn{sub x}/Fe orientation relationship of [110]{sub Fe2Al5-xZnx}//[1 1-bar 1]{sub Fe} and (001){sub Fe2Al5-xZnx}//(110){sub Fe}. Therefore, each domain is formed on a single α-Fe grain and the number of variants in the domain is determined by the orientation of the α-Fe grain. The fraction of the area of epitaxial growth increases with increasing Al content in bath. When the Al content reaches 0.28 wt%, a layer of a single variant (SV) crystal can grow epitaxially on the α-Fe grains having orientations distributed from <110> to <113> parallel to the normal direction of the sheet surface. The formation of a dense SV inhibition layer is the major reason for the drastic reductions in both the Al-pickup rate and the inhibition layer thickness at high Al content baths.

36 MATERIALS SCIENCE↗