Search NASA⌕ Search

SEARCH · Search NASA

Results for “distributed clustering methods”

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 487 records · Page 27

Constraints on cosmological models from strong gravitational lensing systems

Strong lensing has developed into an important astrophysical tool for probing both cosmology and galaxies (their structure, formation, and evolution). Using the gravitational lensing theory and cluster mass distribution model, we try to collect a relatively complete observational data concerning the Hubble constant independent ratio between two angular diameter distances D{sub ds}/D{sub s} from various large systematic gravitational lens surveys and lensing by galaxy clusters combined with X-ray observations, and check the possibility to use it in the future as complementary to other cosmological probes. On one hand, strongly gravitationally lensed quasar-galaxy systems create such a new opportunity by combining stellar kinematics (central velocity dispersion measurements) with lensing geometry (Einstein radius determination from position of images). We apply such a method to a combined gravitational lens data set including 70 data points from Sloan Lens ACS (SLACS) and Lens Structure and Dynamics survey (LSD). On the other hand, a new sample of 10 lensing galaxy clusters with redshifts ranging from 0.1 to 0.6 carefully selected from strong gravitational lensing systems with both X-ray satellite observations and optical giant luminous arcs, is also used to constrain three dark energy models (ΛCDM, constant w and CPL) under a flat universe assumption. For the full sample (n = 80) and the restricted sample (n = 46) including 36 two-image lenses and 10 strong lensing arcs, we obtain relatively good fitting values of basic cosmological parameters, which generally agree with the results already known in the literature. This results encourages further development of this method and its use on larger samples obtained in the future.

79 ASTRONOMY AND ASTROPHYSICS↗

Monte Carlo Simulation of Gamma-Ray Response of BaF2 and CaF2

We have employed a Monte Carlo (MC) method to study intrinsic properties of two alkaline-earth halides, namely BaF2 and CaF2, relevant to their use as radiation detector materials. The MC method follows the fate of individual electron-hole (e-h) pairs and thus allows for a detailed description of the microscopic structure of ionization tracks created by incident γ-ray radiation. The properties of interest include the mean energy required to create an e-h pair, W, Fano factor, F, the maximum theoretical light yield, and the spatial distribution of e-h pairs resulting from γ-ray excitation. Although W and F vary with incident photon energy at low energies, they tend to constant values at energies higher than 1 keV. W is determined to be 18.9 and 19.8 eV for BaF2 and CaF2, respectively, in agreement with published data. The e-h pair spatial distributions exhibit a linear distribution along the fast electron tracks with high e-h pair densities at the end of the tracks. Most e-h pairs are created by interband transition and plasmon excitation in both scintillators, but the e-h pairs along fast electron tracks in BaF2 are slightly clustered, forming nanoscale domains and resulting in the higher e-h pair densities than in CaF2. Combining the maximum theoretical light yields calculated for BaF2 and CaF2 with those obtained for CsI and NaI shows that the theoretical light yield decreases linearly with increasing band gap energy.

detector materials↗

A Note on Exponential Decay in the Random Field Ising Model

For the two-dimensional random field Ising model (RFIM) with bimodal (i.e., two-valued) external field, we prove exponential decay of correlations either (i) when the temperature is larger than the critical temperature of the Ising model without external field and the magnetic field strength is small or (ii) at any temperature when the magnetic field strength is sufficiently large. Unlike previous work on exponential decay, our approach is not based on cluster expansions but rather on arguably simpler methods; these combine an analysis of the Kertész line and a coupling of Ising measures (and also their random cluster representations) with different boundary conditions. We also show similar but weaker results for the RFIM with a general field distribution and in any dimension.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Size-distribution evolution of ion-beam-synthesized nanoclusters in silica

A model to describe the growth of nanoclusters in silica via ion-beam synthesis is introduced. Kinetic Monte Carlo simulations indicate that nucleation, growth, coarsening, and fragmentation occur throughout implantation, leading to a steady-state size-distribution shape that agrees with experimental observations. A set of coupled rate equations are derived and solved within a self-consistent mean-field approximation. An intermediate asymptotic scaling analysis helps to identify the important experimentally accessible parameters that control ion-beam-synthesized nanocluster size distributions. The model predicts that the shape of the as-implanted size distribution depends only on a characteristic length governed by the effective diffusivity, effective ion solubility, and the volumetric flux while the average cluster size is determined by the solute/matrix interface energy.

77 NANOSCIENCE AND NANOTECHNOLOGY↗

A Scalable Distributed Louvain Algorithm for Large-Scale Graph Community Detection

We present a new distributed community detection algorithm for large graphs based on the Louvain method. We exploit a distributed delegate partitioning to ensure the workload and communication balancing among processors. In addition, we design a new heuristic strategy to carefully coordinate the community constitution in a distributed environment, and ensure the convergence of the distributed clustering algorithm. Our intensive experimental study has demonstrated the scalability and the correctness of our algorithm with various large-scale real-world and synthetic graph datasets using up to 32,768 processors.

Computer Science↗

Hierarchical Clock Synchronization in MPI

We present a new distributed community detection algorithm for large graphs based on the Louvain method. We exploit a distributed delegate partitioning to ensure the workload and communication balancing among processors. In addition, we design a new heuristic strategy to carefully coordinate the community constitution in a distributed environment, and ensure the convergence of the distributed clustering algorithm. Our intensive experimental study has demonstrated the scalability and the correctness of our algorithm with various large-scale real-world and synthetic graph datasets using up to 32,768 processors

Computer Science↗

Detecting Rain–Snow-Transition Elevations in Mountain Basins Using Wireless Sensor Networks

Here, to provide complementary information on the hydrologically important rain–snow-transition elevation in mountain basins, this study provides two estimation methods using ground measurements from basin-scale wireless sensor networks: one based on wet-bulb temperature T wet and the other based on snow-depth measurements of accumulation and ablation. With data from 17 spatially distributed clusters (178 nodes) from two networks, in the American and Feather River basins of California’s Sierra Nevada, we analyzed transition elevation during 76 storm events in 2014–18. A T wet threshold of 0.5°C best matched the transition elevation defined by snow depth. Transition elevations using T wet in upper elevations of the basins generally agreed with atmospheric snow level from radars located at lower elevations, while radar snow level was ~100 m higher due to snow-level lowering on windward mountainsides during orographic lifting. Diurnal patterns of the difference between transition elevation and radar snow level were observed in the American basin, related to diurnal ground-temperature variations. However, these patterns were not found in the Feather basin due to complex terrain and higher uncertainties in transition-elevation estimates. The American basin tends to exhibit 100-m-higher transition elevations than does the Feather basin, consistent with the Feather basin being about 1° latitude farther north. Transition elevation averaged 155 m higher in intense atmospheric river events than in other events; meanwhile, snow-level lowering was enhanced with a 90-m-larger difference between radar snow level and transition elevation. On-the-ground continuous observations from distributed sensor networks can complement radar data and provide important ground truth and spatially resolved information on transition elevations in mountain basins.

54 ENVIRONMENTAL SCIENCES↗

Contributions to the NUCLEI SciDAC-3 Project

This is the Final Report for Michigan State University for the NUCLEI SciDAC-3 project. The NUCLEI project, as defined by the scope of work, has developed, implemented and run codes for large-scale computations of many topics in low-energy nuclear physics. Physics studied included the properties of nuclei and nuclear decays, nuclear structure and reactions, and the properties of nuclear matter. The computational techniques used included Configuration Interaction, Coupled Cluster, and Density Functional methods. The research program emphasized areas of high interest to current and possible future DOE nuclear physics facilities, including ATLAS at ANL and FRIB at MSU (nuclear structure and reactions, and nuclear astrophysics), TJNAF (neutron distributions in nuclei, few body systems, and electroweak processes), NIF (thermonuclear reactions), MAJORANA and FNPB (neutrinoless double-beta decay and physics beyond the Standard Model), and LANSCE (fission studies).

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Computation of neutron fluxes in clusters of fuel pins arranged in hexagonal assemblies (2D and 3D)

For computations of fluxes, we have used Carvik's method of collision probabilities. This method requires tracking algorithms. An algorithm to compute tracks (in 2D and 3D) has been developed for seven hexagonal geometries with cluster of fuel pins. This has been implemented in the NXT module of the code DRAGON. The flux distribution in cluster of pins has been computed by using this code. For testing the results, they are compared when possible with the EXCELT module of the code DRAGON. Tracks are plotted in the NXT module by using MATLAB, these plots are also presented here. Results are presented with increasing number of lines to show the convergence of these results. We have numerically computed volumes, surface areas and the percentage errors in these computations. These results show that 2D results converge faster than 3D results. The accuracy on the computation of fluxes up to second decimal is achieved with fewer lines. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

THE AGES OF 55 GLOBULAR CLUSTERS AS DETERMINED USING AN IMPROVED ΔV{sup HB}{sub TO} METHOD ALONG WITH COLOR-MAGNITUDE DIAGRAM CONSTRAINTS, AND THEIR IMPLICATIONS FOR BROADER ISSUES

Ages have been derived for 55 globular clusters (GCs) for which Hubble Space Telescope Advanced Camera for Surveys photometry is publicly available. For most of them, the assumed distances are based on fits of theoretical zero-age horizontal-branch (ZAHB) loci to the lower bound of the observed distributions of HB stars, assuming reddenings from empirical dust maps and metallicities from the latest spectroscopic analyses. The age of the isochrone that provides the best fit to the stars in the vicinity of the turnoff (TO) is taken to be the best estimate of the cluster age. The morphology of isochrones between the TO and the beginning part of the subgiant branch (SGB) is shown to be nearly independent of age and chemical abundances. For well-defined color-magnitude diagrams (CMDs), the error bar arising just from the 'fitting' of ZAHBs and isochrones is ≈ ± 0.25 Gyr, while that associated with distance and chemical abundance uncertainties is ∼ ± 1.5-2 Gyr. The oldest GCs in our sample are predicted to have ages of ≈13.0 Gyr (subject to the aforementioned uncertainties). However, the main focus of this investigation is on relative GC ages. In conflict with recent findings based on the relative main-sequence fitting method, which have been studied in some detail and reconciled with our results, ages are found to vary from mean values of ≈12.5 Gyr at [Fe/H] ∼< – 1.7 to ≈11 Gyr at [Fe/H] ∼> –1. At intermediate metallicities, the age-metallicity relation (AMR) appears to be bifurcated: one branch apparently contains clusters with disk-like kinematics, whereas the other branch, which is displaced to lower [Fe/H] values by ≈0.6 dex at a fixed age, is populated by clusters with halo-type orbits. The dispersion in age about each component of the AMR is ∼ ± 0.5 Gyr. There is no apparent dependence of age on Galactocentric distance (R{sub G}) nor is there a clear correlation of HB type with age. As previously discovered in the case of M3 and M13, subtle variations have been found in the slope of the SGB in the CMDs of other metal-poor ([Fe/H] ∼< – 1.5) GCs. They have been tentatively attributed to cluster-to-cluster differences in the abundance of helium. Curiously, GCs that have relatively steep 'M13-like' SGBs tend to be massive systems, located at small R{sub G}, that show the strongest evidence of in situ formation of multiple stellar populations. The clusters in the other group are typically low-mass systems (with 2-3 exceptions, including M3) that, at the present time, should not be able to retain the matter lost by mass-losing stars due either to the development of GC winds or to ram-pressure stripping by the halo interstellar medium. The apparent separation of the two groups in terms of their present-day gas retention properties is difficult to understand if all GCs were initially ∼20 times their current masses. The lowest-mass systems, in particular, may have never been massive enough to retain enough gas to produce a significant population of second-generation stars. In this case, the observed light element abundance variations, which are characteristic of all GCs, were presumably present in the gas out of which the observed cluster stars formed.

79 ASTRONOMY AND ASTROPHYSICS↗

Inner-shell photoionization and core-hole decay of Xe and XeF{sub 2}

Photoionization cross sections and partial ion yields of Xe and XeF{sub 2} from Xe 3d{sub 5/2}, Xe 3d{sub 3/2}, and F 1s subshells in the 660–740 eV range are compared to explore effects of the F ligands. The Xe 3d-ϵf continuum shape resonances dominate the photoionization cross sections of both the atom and molecule, but prominent resonances appear in the XeF{sub 2} cross section due to nominal excitation of Xe 3d and F 1s electrons to the lowest unoccupied molecular orbital (LUMO), a delocalized anti-bonding MO. Comparisons of the ion products from the atom and molecule following Xe 3d photoionization show that the charge-state distribution of Xe ions is shifted to lower charge states in the molecule along with production of energetic F{sup +} and F{sup 2+} ions. This suggests that, in decay of a Xe 3d core hole, charge is redistributed to the F ligands and the system dissociates due to Coulomb repulsion. The ion products from excitation of the F 1s-LUMO resonance are different and show strong increases in the yields of Xe{sup +} and F{sup +} ions. The subshell ionization thresholds, the LUMO resonance energies, and their oscillator strengths are calculated by relativistic coupled-cluster methods and agree well with measurements.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Elastic scattering and total reaction cross sections for the {sup 8}B, {sup 7}Be, and {sup 6}Li +{sup 12}C systems

Angular distributions for the elastic scattering of {sup 8}B, {sup 7}Be, and {sup 6}Li on a {sup 12}C target have been measured at E{sub lab}=25.8, 18.8, and 12.3 MeV, respectively. The analyses of these angular distributions have been performed in terms of the optical model using Woods-Saxon and double-folding type potentials. The effect of breakup in the elastic scattering of {sup 8}B+{sup 12}C is investigated by performing coupled-channels calculations with the continuum discretized coupled-channel method and cluster-model folding potentials. Total reaction cross sections were deduced from the elastic-scattering analysis and compared with published data on elastic scattering of other weakly and tightly bound projectiles on {sup 12}C, as a function of energy. With the exception of {sup 4}He and {sup 16}O, the data can be described using a universal function for the reduced cross sections.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Investigating the Relation between Galaxy Properties and the Gaussianity of the Velocity Distribution of Groups and Clusters

We investigate the dependence of stellar population properties of galaxies on group dynamical stage for a subsample of the Yang catalog. We classify groups according to their galaxy velocity distribution into Gaussian (G) and Non-Gaussian (NG). Using two totally independent approaches, we have shown that our measurement of Gaussianity is robust and reliable. Our sample covers Yang’s groups in the redshift range 0.03 ≤ z ≤ 0.1, with mass ≥ 10{sup 14}M{sub ⊙}. The new method, called Hellinger Distance, to determine whether a group has a velocity distribution Gaussian or NG is very effective in distinguishing between the two families. NG groups present halo masses higher than the G ones, confirming previous findings. Examining the skewness and kurtosis of the velocity distribution of G and NG groups, we find that faint galaxies in NG groups are mainly infalling, for the first time, into the groups. We show that considering only faint galaxies in the outskirts; those in NG groups are older and more metal-rich than those in G groups. Also, examining the Projected Phase Space of cluster galaxies, we see that bright and faint galactic systems in G groups are in dynamical equilibrium—which does not seem to be the case in NG groups. These findings suggest that NG systems have a higher infall rate, assembling more galaxies that have experienced preprocessing before entering the group.

79 ASTRONOMY AND ASTROPHYSICS↗

Probabilistic Cosmological Mass Mapping from Weak Lensing Shear

We infer gravitational lensing shear and convergence fields from galaxy ellipticity catalogs under a spatial process prior for the lensing potential. We demonstrate the performance of our algorithm with simulated Gaussian-distributed cosmological lensing shear maps and a reconstruction of the mass distribution of the merging galaxy cluster Abell 781 using galaxy ellipticities measured with the Deep Lens Survey. Given interim posterior samples of lensing shear or convergence fields on the sky, we describe an algorithm to infer cosmological parameters via lens field marginalization. In the most general formulation of our algorithm we make no assumptions about weak shear or Gaussian-distributed shape noise or shears. Because we require solutions and matrix determinants of a linear system of dimension that scales with the number of galaxies, we expect our algorithm to require parallel high-performance computing resources for application to ongoing wide field lensing surveys.

79 ASTRONOMY AND ASTROPHYSICS↗

THE BINARY FRACTION OF THE YOUNG CLUSTER NGC 1818 IN THE LARGE MAGELLANIC CLOUD

We use high-resolution Hubble Space Telescope imaging observations of the young ({approx}15-25 Myr old) star cluster NGC 1818 in the Large Magellanic Cloud to derive an estimate for the binary fraction of F stars (1.3 < M{sub *}/M{sub sun} < 1.6). This study provides the strongest constraints yet on the binary fraction in a young star cluster in a low-metallicity environment ([Fe/H] {approx} -0.4 dex). Employing artificial-star tests, we develop a simple method that can efficiently measure the probabilities of stellar blends and superpositions from the observed stellar catalog. We create synthetic color-magnitude diagrams matching the fundamental parameters of NGC 1818, with different binary fractions and mass-ratio distributions. We find that this method is sensitive to binaries with mass ratios q {approx}> 0.4. For binaries with F-star primaries and mass ratios q > 0.4, the binary fraction is {approx}0.35. This suggests a total binary fraction for F stars of 0.55 to unity, depending on assumptions about the form of the mass-ratio distribution at low q.

79 ASTRONOMY AND ASTROPHYSICS↗

Particle Filtering for Model-Based Anomaly Detection in Sensor Networks

A novel technique has been developed for anomaly detection of rocket engine test stand (RETS) data. The objective was to develop a system that postprocesses a csv file containing the sensor readings and activities (time-series) from a rocket engine test, and detects any anomalies that might have occurred during the test. The output consists of the names of the sensors that show anomalous behavior, and the start and end time of each anomaly. In order to reduce the involvement of domain experts significantly, several data-driven approaches have been proposed where models are automatically acquired from the data, thus bypassing the cost and effort of building system models. Many supervised learning methods can efficiently learn operational and fault models, given large amounts of both nominal and fault data. However, for domains such as RETS data, the amount of anomalous data that is actually available is relatively small, making most supervised learning methods rather ineffective, and in general met with limited success in anomaly detection. The fundamental problem with existing approaches is that they assume that the data are iid, i.e., independent and identically distributed, which is violated in typical RETS data. None of these techniques naturally exploit the temporal information inherent in time series data from the sensor networks. There are correlations among the sensor readings, not only at the same time, but also across time. However, these approaches have not explicitly identified and exploited such correlations. Given these limitations of model-free methods, there has been renewed interest in model-based methods, specifically graphical methods that explicitly reason temporally. The Gaussian Mixture Model (GMM) in a Linear Dynamic System approach assumes that the multi-dimensional test data is a mixture of multi-variate Gaussians, and fits a given number of Gaussian clusters with the help of the wellknown Expectation Maximization (EM) algorithm. The parameters thus learned are used for calculating the joint distribution of the observations. However, this GMM assumption is essentially an approximation and signals the potential viability of non-parametric density estimators. This is the key idea underlying the new approach.

Solano, Wanda↗

A study of the threshold method utilizing raingage data

The threshold method for estimation of area-average rain rate relies on determination of the fractional area where rain rate exceeds a preset level of intensity. Previous studies have shown that the optimal threshold level depends on the climatological rain-rate distribution (RRD). It has also been noted, however, that the climatological RRD may be composed of an aggregate of distributions, one for each of several distinctly different synoptic conditions, each having its own optimal threshold. In this study, the impact of RRD variations on the threshold method is shown in an analysis of 1-min rainrate data from a network of tipping-bucket gauges in Darwin, Australia. Data are analyzed for two distinct regimes: the premonsoon environment, having isolated intense thunderstorms, and the active monsoon rains, having organized convective cell clusters that generate large areas of stratiform rain. It is found that a threshold of 10 mm/h results in the same threshold coefficient for both regimes, suggesting an alternative definition of optimal threshold as that which is least sensitive to distribution variations. The observed behavior of the threshold coefficient is well simulated by assumption of lognormal distributions with different scale parameters and same shape parameters.

Short, David A.↗

Performance and power modeling and prediction using MuMMI and 10 machine learning methods

Energy-efficient scientific applications require insight into how high performance computing system features impact the applications' power and performance. This insight can result from the development of performance and power models. Here, in this article, we use the modeling and prediction tool MuMMI (Multiple Metrics Modeling Infrastructure) and 10 machine learning methods to model and predict performance and power consumption and compare their prediction error rates. We use an algorithm-based fault-tolerant linear algebra code and a multilevel checkpointing fault-tolerant heat distribution code to conduct our modeling and prediction study on the Cray XC40 Theta and IBM BG/Q Mira at Argonne National Laboratory and the Intel Haswell cluster Shepard at Sandia National Laboratories. Our experimental results show that the prediction error rates in performance and power using MuMMI are less than 10% for most cases. By utilizing the models for runtime, node power, CPU power, and memory power, we identify the most significant performance counters for potential application optimizations, and we predict theoretical outcomes of the optimizations. Based on two collected datasets, we analyze and compare the prediction accuracy in performance and power consumption using MuMMI and 10 machine learning methods.

97 MATHEMATICS AND COMPUTING↗