Search NASA⌕ Search

SEARCH · Search NASA

Results for “random sampling”

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 181 records · Page 10

Compressing unstructured mesh data from simulations using machine learning

The amount of data output from a computer simulation has grown to terabytes and petabytes as increasingly complex simulations are being run on massively parallel systems. As we approach exaflop computing in the next decade, it is expected that the I/O subsystem will not be able to write out these large volumes of data. In this paper, we explore the use of machine learning to compress the data before it is written out. Despite the computational constraints that limit us to using very simple learning algorithms, our results show that machine learning is a viable option for compressing unstructured data. Furthermore, we demonstrate that by simply using a better sampling algorithm to generate the training set, we can obtain more accurate results compared to random sampling, but at no extra cost. Further, by carefully selecting and incorporating points with high prediction error, we can improve reconstruction accuracy without sacrificing the compression rate.

97 MATHEMATICS AND COMPUTING↗

Deployment of salt sample extraction system at an engineering-scale electrorefiner

The goal of the salt sampling program at Argonne is to develop and deploy automated molten salt sampling approaches for interfacing relevant unit operations with salt analysis to improve the timeliness of sampling-based accountancy measurements. Two technologies under development in support of this goal are a vacuum sampling loop module and a high-throughput pneumatic sample generator module. Compared to traditional point sampling approaches (i.e., dip probes), the vacuum sampling loop facilitates the collection of a larger cross-section of the bulk salt in order to collect more representative samples. The vacuum sampling approach also eliminates the risk of dross contamination of samples and avoids the use of moving parts in the salt. The pneumatic sample generator module is used to facilitate high-throughput sample analysis to improve the measurement precision of any given analytical technique by averaging out random sampling and measurement errors. In FY21, two methods for integrating these two modules were tested including direct fluidic coupling and coupling using a solid salt transfer mechanism. Solid salt transfer was ultimately selected over fluidic coupling, primarily to enable the transport of samples over longer distances to support automated at-line integration with high-precision techniques (such as microcalorimetry) that cannot withstand the conditions near an electrorefining process. To facilitate rapid solid salt coupling, new mechanisms were developed for rapidly charging and discharging salt sample tubes at the vacuum sampling loop and pneumatic sample generator modules, respectively. While the charging mechanism will be deployed in FY22, the discharge mechanism was tested in FY21 and is described here. The solid salt tube transfer method was deployed at one of Argonne’s engineering-scale electrorefiners to implement at-line high-throughput pneumatic micro-sample generation capabilities. The approach was used to generate precise uranium- and lanthanide-bearing electrorefiner micro-samples with the specific dimensions requested by researchers at Los Alamos National Laboratory for use in testing their novel microcalorimeter x-ray techniques. The solid salt transfer mechanism proved not only to be an effective means of integrating the precision sample generator with vacuum sampling, but also improved the performance of the sampler generator. To discharge salt from the sample tubes at the sampler generator, tube segments were inserted directly into the sample generator’s Helmholtz chamber and pressure pulse actuations were used to generate precision molten salt samples directly from the tube segments. The direct insertion of sample tubes into the sample generator enabled rapid loading of the salt and prevented salt from contacting most of the interior surfaces of the sample generator, which eliminated cross-contamination between runs. The vacuum sampling-loop tube charging mechanism will support high-throughput tube sampling operations by employing a dynamic vacuum filling process to fill short charge tubes that are configured to be rapidly connected and disconnected from the loop. The dynamic vacuum sampling operation will be automated, and sample tube handling can be executed with simple overhead actuation. Because the modular sampling approach described here eliminates the need for new high-radiation sample handling capabilities, salt-wetted seals, salt-wetted moving parts, and heated transfer lines outside the electrorefiner, it will address most of the remaining technical challenges for the at-line deployment of high-precision analytical techniques which will enable significant reductions in the time delay for sampling-based high-precision accountancy measurements.

42 ENGINEERING↗

Jimsphere wind and turbulence exceedance statistic

Exceedance statistics of winds and gusts observed over Cape Kennedy with Jimsphere balloon sensors are described. Gust profiles containing positive and negative departures, from smoothed profiles, in the wavelength ranges 100-2500, 100-1900, 100-860, and 100-460 meters were computed from 1578 profiles with four 41 weight digital high pass filters. Extreme values of the square root of gust speed are normally distributed. Monthly and annual exceedance probability distributions of normalized rms gust speeds in three altitude bands (2-7, 6-11, and 9-14 km) are log-normal. The rms gust speeds are largest in the 100-2500 wavelength band between 9 and 14 km in late winter and early spring. A study of monthly and annual exceedance probabilities and the number of occurrences per kilometer of level crossings with positive slope indicates significant variability with season, altitude, and filter configuration. A decile sampling scheme is tested and an optimum approach is suggested for drawing a relatively small random sample that represents the characteristic extreme wind speeds and shears of a large parent population of Jimsphere wind profiles.

Adelfang, S. I.↗

Bayesian Estimation of Earth’s Undiscovered Mineralogical Diversity Using Noninformative Priors

Recently, statistical distributions have been explored to provide estimates of the mineralogical diversity of Earth, and Earth-like planets. In this paper, a Bayesian approach is introduced to estimate Earth’s undiscovered mineralogical diversity. Samples are generated from a posterior distribution of the model parameters using Markov chain Monte Carlo simulations such that estimates and inference are directly obtained. It was previously shown that the mineral species frequency distribution conforms to a generalized inverse Gauss–Poisson (GIGP) large number of rare events model. Even though the model fit was good, the population size estimate obtained by using this model was found to be unreasonably low by mineralogists. In this paper, several zero-truncated, mixed Poisson distributions are fitted and compared, where the Poisson-lognormal distribution is found to provide the best fit. Subsequently, the population size estimates obtained by Bayesian methods are compared to the empirical Bayes estimates. Species accumulation curves are constructed and employed to estimate the population size as a function of sampling size. Finally, the relative abundances, and hence the occurrence probabilities of species in a random sample, are calculated numerically for all mineral species in Earth’s crust using the Poisson-lognormal distribution. These calculations are connected and compared to the calculations obtained in a previous paper using the GIGP model for which mineralogical criteria of an Earth-like planet were given.

Bayesian statistics↗

Errors of five-day mean surface wind and temperature conditions due to inadequate sampling

Surface meteorological reports of wind components, wind speed, air temperature, and sea-surface temperature from buoys located in equatorial and midlatitude regions are used in a simulation of random sampling to determine errors of the calculated means due to inadequate sampling. Subsampling the data with several different sample sizes leads to estimates of the accuracy of the subsampled means. The number N of random observations needed to compute mean winds with chosen accuracies of 0.5 (N sub 0.5) and 1.0 (N sub 1,0) m/s and mean air and sea surface temperatures with chosen accuracies of 0.1 (N sub 0.1) and 0.2 (N sub 0.2) C were calculated for each 5-day and 30-day period in the buoy datasets. Mean values of N for the various accuracies and datasets are given. A second-order polynomial relation is established between N and the variability of the data record. This relationship demonstrates that for the same accuracy, N increases as the variability of the data record increases. The relationship is also independent of the data source. Volunteer-observing ship data do not satisfy the recommended minimum number of observations for obtaining 0.5 m/s and 0.2 C accuracy for most locations. The effect of having remotely sensed data is discussed.

Legler, David M.↗

COBRA:COMPUTED-TOMOGRAPHY BASED RANDOM-FIELD APPROXIMATION

SF-25-115 COBRA (COmputed-tomography Based Random-field Approximation) is a Python application for generating statistically equivalent random fields from CT-scan imagery. It leverages Karhunen–Loève expansions to model microstructural variability, enabling users to: Preprocess CT scans (filtering and Gaussian transformation); Fit covariance kernels fromempirical data; Solve eigenproblems to obtain KL modes; Sample random fields onsistent with fitted statistics; Postprocess samples back into the physical domain.

Hu, Tianchen↗

Topology of Large-scale Structures of Galaxies in Two Dimensions—Systematic Effects

We study the two-dimensional topology of galactic distribution when projected onto two-dimensional spherical shells. Using the latest Horizon Run 4 simulation data, we construct the genus of the two-dimensional field and consider how this statistic is affected by late-time nonlinear effects—principally gravitational collapse and redshift space distortion (RSD). We also consider systematic and numerical artifacts, such as shot noise, galaxy bias, and finite pixel effects. We model the systematics using a Hermite polynomial expansion and perform a comprehensive analysis of known effects on the two-dimensional genus, with a view toward using the statistic for cosmological parameter estimation. We find that the finite pixel effect is dominated by an amplitude drop and can be made less than 1% by adopting pixels smaller than 1/3 of the angular smoothing length. Nonlinear gravitational evolution introduces time-dependent coefficients of the zeroth, first, and second Hermite polynomials, but the genus amplitude changes by less than 1% between z = 1 and z = 0 for smoothing scales R{sub G}>9 Mpc/h. Non-zero terms are measured up to third order in the Hermite polynomial expansion when studying RSD. Differences in the shapes of the genus curves in real and redshift space are small when we adopt thick redshift shells, but the amplitude change remains a significant ∼O(10%) effect. The combined effects of galaxy biasing and shot noise produce systematic effects up to the second Hermite polynomial. It is shown that, when sampling, the use of galaxy mass cuts significantly reduces the effect of shot noise relative to random sampling.

79 ASTRONOMY AND ASTROPHYSICS↗

Variability of reflectance measurements with sensor altitude and canopy type

Data were acquired on canopies of mature corn planted in 76 cm rows, mature soybeans planted in 96 cm rows with 71 percent soil cover, and mature soybeans planed in 76 cm rows with 100 percent soil cover. A LANDSAT band radiometer with a 15 degree field of view was used at ten altitudes ranging from 0.2 m to 10 m above the canopy. At each altitude, measurements were taken at 15 cm intervals also a 2.0 m transect perpendicular to the crop row direction. Reflectance data were plotted as a function of altitude and horizontal position to verify that the variance of measurements at low altitudes was attributable to row effects which disappear at higher altitudes where the sensor integrate across several rows. The coefficient of variation of reflectance decreased exponentially as the sensor was elevated. Systematic sampling (at odd multiples of 0.5 times the row spacing interval) required fewer measurements than simple random sampling over row crop canopies.

Daughtry, C. S. T.↗

Self-Supervised Anomaly Detection via Neural Autoregressive Flows with Active Learning

Many self-supervised methods have been proposed with the target of image anomaly detection. These methods often rely on the paradigm of data augmentation with predefined transformations such as flipping, cropping, and rotations. However, it is not straightforward to apply these techniques for non-image data, such as time series or tabular data, while the performance of the existing deep approaches has been under our expectation on tasks beyond images. In this work, we propose a novel active learning (AL) scheme that relied on neural autoregressive flows (NAF) for self-supervised anomaly detection, specifically on small-scale data. Unlike other generative models such as GANs or VAEs, flow-based models allow to explicitly learn the probability density and thus can assign accurate likelihoods to normal data which makes it usable to detect anomalies. The proposed NAF-AL method is achieved by efficiently generating random samples from latent space and transforming them into feature space along with likelihoods via invertible mapping. The samples with lower likelihoods are selected and further checked by outlier detection using Mahalanobis distance. The augmented samples incorporating with normal samples are used for training a better detector so as to approach decision boundaries. Compared with random transformations, NAF-AL can be interpreted as a likelihood-oriented data augmentation that is more efficient and robust. Extensive experiments show that our approach outperforms existing baselines on multiple time series and tabular datasets, and a real-world application in advanced manufacturing, with significant improvement on anomaly detection accuracy and robustness over the state-of-the-art.

Zhang, Jiaxin↗

Neutron and Gamma Correlations using CGM in MCNP 6.2.0

The general-purpose Monte Carlo radiation-transport code MCNP6{sup TM} came from the integration of MCNP5{sup TM} and MCNPX{sup TM}. The most recent release of MCNP, version 6.2.0, is scheduled for release in early 2017, which includes a variety of bug fixes and new features to the code. MCNP traditionally models nuclear reactions using Monte Carlo sampling techniques on measured and modeled cross-sectional data contained within ACE (A Compact Evaluated Nuclear Data File) libraries. In addition to reaction cross-sections, ACE libraries contain information relating to the production of secondary particles, such as secondary neutrons and photons. For example, an incident neutron on a nucleus can produce a nuclear reaction in the form (n, M{sub n}n'), where n is the incident neutron and M{sub n} is the number of secondary neutrons n' produced. Using this notation, different reactions can be described based on their neutron multiplicity, such as neutron capture (M{sub n} = 0), elastic and inelastic scattering (M{sub n} = 1), and reactions with multiple secondary neutrons (M{sub n} > 1). Accounting for secondary γ-rays emitted from the residual nuclei, this reaction can further be generalized as n, Mnn'M to describe a reaction producing M{sub n} neutrons and M -rays. These values M{sub n} and M are referred to as the neutron and multiplicities, respectively. ACE libraries contain the cross-sectional data for each reaction, and their corresponding statistically averaged multiplicities for neutrons and photons, M-bar{sub n} and M-bar{sub γ}. However, there are several limitations to reaction sampling using ACE libraries. First, the data contained within ACE libraries is limited by the accuracy of the model and/or experiment, which out of necessity can greatly simplify the true physics behind a reaction. For example, the value M{sub γ} in a (n, M{sub n}n'M{sub γ}γ) reaction is determined by the statistically averaged multiplicity M-bar{sub γ} instead of a distribution. Consequently, MCNP, unable to produce a continuous distribution, will utilizes a binary sampling around M-bar{sub γ}. While this sampling method provides a statistically average M{sub γ} over a large simulation, it is an inaccurate representation of an individual reaction and its corresponding secondary particles. Secondary neutrons are affected in a similar manner, such that different kinds of secondary neutrons can be sampled from the same interaction (eg., an inelastic neutron with an [n, 2n] neutron). Lastly, the random sampling technique for secondary particles remains completely independent of the sampled reaction, eliminating the ability to correlate secondary particles for a given reaction. To supplement these data libraries, additional codes and models describing certain nuclear reactions are employed. These codes can fill in the information missing from ACE, provide a better model to the true physics distribution, and can provide true correlated secondary particles, which has many useful applications in research and industry.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

SOARCA Peach Bottom Atomic Power Station Long-Term Station Blackout Uncertainty Analysis: Convergence of the Uncertainty Results

This paper describes the convergence of MELCOR Accident Consequence Code System, Version 2 (MACCS2) probabilistic results of offsite consequences for the uncertainty analysis of the State-of-the-Art Reactor Consequence Analyses (SOARCA) unmitigated long-term station blackout scenario at the Peach Bottom Atomic Power Station. The consequence metrics evaluated are individual latent-cancer fatality (LCF) risk and individual early fatality risk. Consequence results are presented as conditional risk (i.e., assuming the accident occurs, risk per event) to individuals of the public as a result of the accident. In order to verify convergence for this uncertainty analysis, as recommended by the Nuclear Regulatory Commission’s Advisory Committee on Reactor Safeguards, a ‘high’ source term from the original population of Monte Carlo runs has been selected to be used for: (1) a study of the distribution of consequence results stemming solely from epistemic uncertainty in the MACCS2 parameters (i.e., separating the effect from the source term uncertainty), and (2) a comparison between Simple Random Sampling (SRS) and Latin Hypercube Sampling (LHS) in order to validate the original results obtained with LHS. Three replicates (each using a different random seed) of size 1,000 each using LHS and another set of three replicates of size 1,000 using SRS are analyzed. The results show that the LCF risk results are well converged with either LHS or SRS sampling. The early fatality risk results are less well converged at radial distances beyond 2 miles, and this is expected due to the sparse data (predominance of “zero” results).

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Fast generation of sparse random kernel graphs

The development of kernel-based inhomogeneous random graphs has provided models that are flexible enough to capture many observed characteristics of real networks, and that are also mathematically tractable. We specify a class of inhomogeneous random graph models, called random kernel graphs, that produces sparse graphs with tunable graph properties, and we develop an efficient generation algorithm to sample random instances from this model. As real-world networks are usually large, it is essential that the run-time of generation algorithms scales better than quadratically in the number of vertices n. We show that for many practical kernels our algorithm runs in time at most ο(n(logn)²). As an example, we show how to generate samples of power-law degree distribution graphs with tunable assortativity.

97 MATHEMATICS AND COMPUTING↗

diamond-sampling

This software provides a means for computing only the largest few entries of the product of two matrices, both exactly and approximately (using randomized sampling techniques). The purpose of the code is to demonstrate both the time it takes to solve the problem as well as the accuracy of the approximate approach. It is also meant to serve as a foundation to test the applicability of the sampling technique to related problems in data mining, including maximum inner product search, nearest neighbor search, and maximum cosine similarity.

Ballard, Grey↗

DOES SIZE MATTER? THE UNDERLYING INTRINSIC SIZE DISTRIBUTION OF RADIO SOURCES AND IMPLICATIONS FOR UNIFICATION BY ORIENTATION

Unification by orientation is a ubiquitous concept in the study of active galactic nuclei. A gold standard of the orientation paradigm is the hypothesis that radio galaxies and radio-loud quasars are intrinsically the same, but are observed over different ranges of viewing angles. Historically, strong support for this model was provided by the projected sizes of radio structure in luminous radio galaxies, which were found to be significantly larger than those of quasars, as predicted due to simple geometric projection. Recently, this test of the simplest prediction of orientation-based models has been revisited with larger samples that cover wider ranges of fundamental properties-and no clear difference in projected sizes of radio structure is found. Cast solely in terms of viewing angle effects, these results provide convincing evidence that unification of these objects solely through orientation fails. However, it is possible that conflicting results regarding the role orientation plays in our view of radio sources simply result from insufficient sampling of their intrinsic size distribution. We test this possibility using Monte Carlo simulations constrained by real sample sizes and properties. We develop models for the real intrinsic size distribution of radio sources, simulate observations by randomly sampling intrinsic sizes and viewing angles, and analyze how likely each sample is to support or dispute unification by orientation. We find that, while it is possible to reconcile conflicting results purely within a simple, orientation-based framework, it is very unlikely. We analyze the effects that sample size, relative numbers of radio galaxies and quasars, the critical angle that separates the two subclasses, and the shape of the intrinsic size distribution have on this type of test.

79 ASTRONOMY AND ASTROPHYSICS↗

Rapid estimation of photosynthetic leaf traits of tropical plants in diverse environmental conditions using reflectance spectroscopy

Tropical forests are one of the main carbon sinks on Earth, but the magnitude of CO 2 absorbed by tropical vegetation remains uncertain. Terrestrial biosphere models (TBMs) are commonly used to estimate the CO 2 absorbed by forests, but their performance is highly sensitive to the parameterization of processes that control leaf-level CO 2 exchange. Direct measurements of leaf respiratory and photosynthetic traits that determine vegetation CO 2 fluxes are critical, but traditional approaches are time-consuming. Reflectance spectroscopy can be a viable alternative for the estimation of these traits and, because data collection is markedly quicker than traditional gas exchange, the approach can enable the rapid assembly of large datasets. However, the application of spectroscopy to estimate photosynthetic traits across a wide range of tropical species, leaf ages and light environments has not been extensively studied. Here, we used leaf reflectance spectroscopy together with partial least-squares regression (PLSR) modeling to estimate leaf respiration ( R dark25 ), the maximum rate of carboxylation by the enzyme Rubisco ( V cmax25 ), the maximum rate of electron transport ( J max25 ), and the triose phosphate utilization rate ( T p25 ), all normalized to 25°C. We collected data from three tropical forest sites and included leaves from fifty-three species sampled at different leaf phenological stages and different leaf light environments. Our resulting spectra-trait models validated on randomly sampled data showed good predictive performance for V cmax25 , J max25 , T p25 and R dark25 (RMSE of 13, 20, 1.5 and 0.3 μmol m -2 s -1 , and R 2 of 0.74, 0.73, 0.64 and 0.58, respectively). The models showed similar performance when applied to leaves of species not included in the training dataset, illustrating that the approach is robust for capturing the main axes of trait variation in tropical species. We discuss the utility of the spectra-trait and traditional gas exchange approaches for enhancing tropical plant trait studies and improving the parameterization of TBMs.

54 ENVIRONMENTAL SCIENCES↗

Analysis and Monte Carlo simulation of near-terminal aircraft flight paths

The flight paths of arriving and departing aircraft at an airport are stochastically represented. Radar data of the aircraft movements are used to decompose the flight paths into linear and curvilinear segments. Variables which describe the segments are derived, and the best fitting probability distributions of the variables, based on a sample of flight paths, are found. Conversely, given information on the probability distribution of the variables, generation of a random sample of flight paths in a Monte Carlo simulation is discussed. Actual flight paths at Dulles International Airport are analyzed and simulated.

Schiess, J. R.↗

Absolute proper motions to B approximately 22.5: Evidence for kimematical substructure in halo field stars

Radial velocities have been obtained for six of nine stars identified on the basis of similar distances and common, extreme transverse velocities in the proper motion survey of Majewski (1992) as a candidate halo moving group at the north Galactic pole. These radial velocities correspond to velocities perpendicular to the Galactic plane which span the range -48 +/- 21 to -128 +/- 9 km/sec (but a smaller range, -48 +/- 21 to -86 +/- 19 km/sec, when only our own measurements are considered), significantly different than the expected distribution, with mean 0 km/sec, for a random sample of either halo or thick disk stars. The probability of picking such a set of radial velocities at random is less than 1%. Thus the radial velocity data support the hypothesis that these stars constitute part of a halo moving group or star stream at a distance of approximately 4-5 kpc above the Galactic plane. If real, this moving group is evidence for halo phase space substructure which may be the fossil remains of a destroyed globular cluster, Galactic satellite, or Searle & Zinn (1978) 'fragment.'

Majewski, Steven R.↗

i- flow: High-dimensional integration and sampling with normalizing flows

In many fields of science, high-dimensional integration is required. Numerical methods have been developed to evaluate these complex integrals. We introduce the code i-flow, a python package that performs high-dimensional numerical integration utilizing normalizing flows. Normalizing flows are machine-learned, bijective mappings between two distributions. i-flow can also be used to sample random points according to complicated distributions in high dimensions. We compare i-flow to other algorithms for high-dimensional numerical integration and show that i-flow outperforms them for high dimensional correlated integrals. The i-flow code is publicly available on gitlab at https://gitlab.com/i-flow/i-flow.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗