Search NASA⌕ Search

SEARCH · Search NASA

Results for “Stochastic 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 703 records · Page 39

Efficient Bayesian inference with latent Hamiltonian neural networks in No-U-Turn Sampling

When sampling for Bayesian inference, one popular approach in the computational field is to use Hamiltonian Monte Carlo (HMC) and specifically the No-U-Turn Sampler (NUTS), which automatically decides the end time of the Hamiltonian trajectory. However, HMC and NUTS can require numerous numerical gradients of the target density and can prove slow in practice when relying on computationally expensive forward models. We propose Latent Hamiltonian neural networks (L-HNNs) with HMC and NUTS for solving Bayesian inference problems. Once trained, L-HNNs do not require numerical gradients of the target density during sampling, and hence numerous evaluations of the forward computational model. Moreover, L-HNNs satisfy important properties such as perfect time reversibility and Hamiltonian conservation, making them well-suited for use within HMC and NUTS because stationarity can be shown. We also propose the integration of L-HNNs in an online error monitoring scheme, in which numerical gradients of the target density are used for a few samples whenever the L-HNNs prediction errors are large. This online error monitor scheme prevents sample degeneracy in regions of low probability density and ensures robust uncertainty quantification. We demonstrate L-HNNs in NUTS with online error monitoring on several analytical examples involving complex, heavy-tailed, and high-local-curvature probability densities. We then demonstrate the applicability of L-HNNs in NUTS to two computational case studies, namely the Allen-Cahn stochastic partial differential equation and an elliptic partial differential equation with 25 and 50 inference parameters, respectively. Overall, the L-HNNs in NUTS with online error monitoring satisfactorily inferred these probability densities. In conclusion, compared to traditional NUTS, L-HNNs in NUTS with online error monitoring required 1–2 orders of magnitude fewer numerical gradients of the target density and improved the effective sample size (ESS) per gradient (which is a measure of both the sampling quality and the computational expense) by an order of magnitude.

97 MATHEMATICS AND COMPUTING↗

S-OPT: A Points Selection Algorithm for Hyper-Reduction in Reduced Order Models

While projection-based reduced order models can reduce the dimension of full order solutions, the resulting reduced models may still contain terms that scale with the full order dimension. Hyper-reduction techniques are sampling-based methods that further reduce this computational complexity by approximating such terms with a much smaller dimension. The goal of this work is to introduce the points selection algorithm developed by Shin and Xiu as a hyper-reduction method. The selection algorithm was originally proposed as a stochastic collocation method for uncertainty quantification. Since the algorithm aims at maximizing a quantity $\mathcal{S}$ that measures both the column orthogonality and the determinant, we refer to the algorithm as S-OPT. Numerical examples are provided to demonstrate the performance of S-OPT and to compare its performance with a gappy proper orthogonal decomposition (POD) algorithm. Here, we found that using the S-OPT algorithm is shown to predict the full order solutions with higher accuracy than gappy POD especially when the number of sampling points is small, although we note that S-OPT shows slow asymptotic convergence with respect to the number of samples for some applications, e.g., Lagrangian hydrodynamics.

97 MATHEMATICS AND COMPUTING↗

Disk-loss and disk-renewal phases in classical Be stars. II. Contrasting with stable and variable disks

Recent observational and theoretical studies of classical Be stars have established the utility of polarization color diagrams (PCDs) in helping to constrain the time-dependent mass decretion rates of these systems. We expand on our pilot observational study of this phenomenon, and report the detailed analysis of a long-term (1989-2004) spectropolarimetric survey of nine additional classical Be stars, including systems exhibiting evidence of partial disk-loss/disk-growth episodes as well as systems exhibiting long-term stable disks. After carefully characterizing and removing the interstellar polarization along the line of sight to each of these targets, we analyze their intrinsic polarization behavior. We find that many steady-state Be disks pause at the top of the PCD, as predicted by theory. We also observe sharp declines in the Balmer jump polarization for later spectral type, near edge-on steady-state disks, again as recently predicted by theory, likely caused when the base density of the disk is very high, and the outer region of the edge-on disk starts to self absorb a significant number of Balmer jump photons. The intrinsic V-band polarization and polarization position angle of γ Cas exhibits variations that seem to phase with the orbital period of a known one-armed density structure in this disk, similar to the theoretical predictions of Halonen and Jones. We also observe stochastic jumps in the intrinsic polarization across the Balmer jump of several known Be+sdO systems, and speculate that the thermal inflation of part of the outer region of these disks could be responsible for producing this observational phenomenon. Finally, we estimate the base densities of this sample of stars to be between ≈8 × 10{sup –11} and ≈4 × 10{sup –12} g cm{sup –3} during quasi steady state periods given there maximum observed polarization.

79 ASTRONOMY AND ASTROPHYSICS↗

Assembly of the Populus Microbiome Is Temporally Dynamic and Determined by Selective and Stochastic Factors

Recent work shows that the plant microbiome, particularly the initial assembly of this microbiome, influences plant health, survival, and fitness. Here, we characterize the initial assembly of the Populus microbiome across ten genotypes belonging to two poplar species in a common garden using 16S rRNA gene and ITS2 region amplicon sequencing of the leaf endosphere, leaf surface, root endosphere, and rhizosphere. We sampled these microbiomes three times throughout the first growing season and found that the composition of the microbiome changed dramatically over time across all plant-associated habitats and host genotypes. For archaea and bacteria, these changes were dominated by strong homogenizing selection (accounting for 29 to 62% of pairwise comparisons). However, fungal assembly was generally characterized by multiple ecological assembly processes (i.e., a mix of weak selective and dispersal processes). Interestingly, genotype, while a significant moderator of microbiome composition, generally explained less variation than sample date across plant-associated habitats. We defined a set of core genera that accounted for, on average, 36% of the microbiome. The relative abundance of this core community was consistent over time. Additionally, using source tracking modeling, we determined that new microbial taxa colonize from both aboveground and belowground sources, and combined with our ecological assembly null models, we found that both selective and dispersal processes explained the differences between exo- (i.e., leaf surface and rhizosphere) and endospheric microbiomes. Taken together, our results suggest that the initial assembly of the Populus microbiome is time-, genotype-, and habitat-dependent and is moderated by both selective and stochastic factors.

16S rRNA↗

Stochastic derivative-free optimization using a trust region framework

This study presents a trust region algorithm to minimize a function f when one has access only to noise-corrupted function values f¯. The model-based algorithm dynamically adjusts its step length, taking larger steps when the model and function agree and smaller steps when the model is less accurate. The method does not require the user to specify a fixed pattern of points used to build local models and does not repeatedly sample points. If f is sufficiently smooth and the noise is independent and identically distributed with mean zero and finite variance, we prove that our algorithm produces iterates such that the corresponding function gradients converge in probability to zero. As a result, we present a prototype of our algorithm that, while simplistic in its management of previously evaluated points, solves benchmark problems in fewer function evaluations than do existing stochastic approximation methods.

97 MATHEMATICS AND COMPUTING↗

Physics-Informed Sparse Gaussian Process for Probabilistic Stability Analysis of Large-Scale Power System with Dynamic PVs and Loads

This work proposes a physics-informed sparse Gaussian process (SGP) for probabilistic stability assessment of large-scale power systems in the presence of uncertain dynamic PVs and loads. The differential and algebraic equations considering uncertainties from dynamic PVs and loads are reformulated to a nonlinear mapping relationship that allows the application of SGP. Thanks to the nonparametric characteristic of Gaussian process, the proposed framework does not require distributions of uncertain inputs and this distinguishes it from existing approaches. As the original Gaussian process is not scalable to large-scale systems with high dimensional uncertain inputs, this paper develops the SGP with a stochastic variational inference technique. It leads to approximately two orders of complex reduction. A data pre-processing step is also introduced to tackle the coexistence of stable and unstable cases by sample clustering and constructing separate SGPs. The probabilistic transient stability index is analyzed to assess system stability under different uncertain dynamics loads and PVs. Comparisons are performed with the sampling-based, the polynomial chaos expansion-based, and traditional Gaussian process-based methods on the modified IEEE 118-bus and Texas 2000-bus systems under various scenarios, including different levels of uncertainties and the existence of nonlinear correlations among dynamic PVs. The impacts of data quality and quantity issues are also investigated. It is shown that the proposed SGP achieves significantly improved computational efficiency while maintaining high accuracy with a limited number of data.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Intrusive analysis for NEK5000: development of intrusive uncertainty quantification for high-dimensional, high-fidelity codes.

We study the use of lower-fidelity training data for uncertainty quantification of complex simulation models. In our approach, computationally expensive full-model outputs are approximated applying proper orthogonal-decomposition-based dimensionality reduction to the full model. A Gaussian-processes-based machine learning approach is then used to model the difference (or, rather, the correspondence) between the higher-fidelity and the lower-fidelity data. This stochastic model can be constructed by using few additional full code evaluations; it is used to calibrate arbitrarily many lower-quality outputs in order to create additional training data for sampling-based analysis. In effect, an adequate approximation of the simulation model’s response to uncertainty can be constructed at a modest computational cost. In fact, we aim at the number of full-model evaluations comparable to that required for a linear approximation (as opposed to numbers traditionally associated with sampling in high-dimensional spaces). In this report, we explain the basic algorithm, suggest some performance tuning options, and give a first characterization for the class of simulation models of nuclear engineering for which the suggested report is effective.

Roderick, O.↗

An information theory approach for evaluating earth radiation budget (ERB) measurements - Nonuniform sampling of diurnal longwave flux variations

Satellite measurements are subject to a wide range of uncertainties due to their temporal, spatial, and directional sampling characteristics. An information-theory approach is suggested to examine the nonuniform temporal sampling of ERB measurements. The information (i.e., its entropy or uncertainty) before and after the measurements is determined, and information gain (IG) is defined as a reduction in the uncertainties involved. A stochastic model for the diurnal outgoing flux variations that affect the ERB is developed. Using Gaussian distributions for the a priori and measured radiant exitance fields, the IG is obtained by computing the a posteriori covariance. The IG for the monthly outgoing flux measurements is examined for different orbital parameters and orbital tracks, using the Earth Observing System orbital parameters as specific examples. Variations in IG due to changes in the orbit's inclination angle and the initial ascending node local time are investigated.

Halyo, Nesim↗

Generalized correlation integral vectors: A distance concept for chaotic dynamical systems

Several concepts of fractal dimension have been developed to characterise properties of attractors of chaotic dynamical systems. Numerical approximations of them must be calculated by finite samples of simulated trajectories. In principle, the quantities should not depend on the choice of the trajectory, as long as it provides properly distributed samples of the underlying attractor. In practice, however, the trajectories are sensitive with respect to varying initial values, small changes of the model parameters, to the choice of a solver, numeric tolerances, etc. The purpose of this paper is to present a statistically sound approach to quantify this variability. We modify the concept of correlation integral to produce a vector that summarises the variability at all selected scales. The distribution of this stochastic vector can be estimated, and it provides a statistical distance concept between trajectories. Here, we demonstrate the use of the distance for the purpose of estimating model parameters of a chaotic dynamic model. The methodology is illustrated using computational examples for the Lorenz 63 and Lorenz 95 systems, together with a framework for Markov chain Monte Carlo sampling to produce posterior distributions of model parameters.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

An improved target velocity sampling algorithm for free gas elastic scattering

We present an improved algorithm for sampling the target velocity when simulating elastic scattering in a Monte Carlo neutron transport code that correctly accounts for the energy dependence of the scattering cross section. The algorithm samples the relative velocity directly, thereby avoiding a potentially inefficient rejection step based on the ratio of cross sections. Here, we have shown that this algorithm requires only one rejection step, whereas other methods of similar accuracy require two rejection steps. The method was verified against stochastic and deterministic reference results for upscattering percentages in 238 U. Simulations of a light water reactor pin cell problem demonstrate that using this algorithm results in a 3% or less penalty in performance when compared with an approximate method that is used in most production Monte Carlo codes

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

COSMOS: STOCHASTIC BIAS FROM MEASUREMENTS OF WEAK LENSING AND GALAXY CLUSTERING

In the theory of structure formation, galaxies are biased tracers of the underlying matter density field. The statistical relation between galaxy and matter density field is commonly referred to as galaxy bias. In this paper, we test the linear bias model with weak-lensing and galaxy clustering measurements in the 2 deg{sup 2} COSMOS field. We estimate the bias of galaxies between redshifts z = 0.2 and z = 1 and over correlation scales between R = 0.2 h{sup -1} Mpc and R = 15 h{sup -1} Mpc. We focus on three galaxy samples, selected in flux (simultaneous cuts I{sub 814W} < 26.5 and K{sub s} < 24) and in stellar mass (10{sup 9} < M{sub *} < 10{sup 10} h{sup -2} M{sub Sun} and 10{sup 10} < M{sub *} < 10{sup 11} h{sup -2} M{sub Sun }). At scales R > 2 h{sup -1} Mpc, our measurements support a model of bias increasing with redshift. The Tinker et al. fitting function provides a good fit to the data. We find the best-fit mass of the galaxy halos to be log (M{sub 200}/h{sup -1} M{sub Sun }) = 11.7{sup +0.6}{sub -1.3} and log (M{sub 200}/h{sup -1} M{sub Sun }) = 12.4{sup +0.2}{sub -2.9}, respectively, for the low and high stellar-mass samples. In the halo model framework, bias is scale dependent with a change of slope at the transition scale between the one and the two halo terms. We detect a scale dependence of bias with a turndown at scale R = 2.3 {+-} 1.5 h{sup -1} Mpc, in agreement with previous galaxy clustering studies. We find no significant amount of stochasticity, suggesting that a linear bias model is sufficient to describe our data. We use N-body simulations to quantify both the amount of cosmic variance and systematic errors in the measurement.

79 ASTRONOMY AND ASTROPHYSICS↗

Growth phase estimation for abundant bacterial populations sampled longitudinally from human stool metagenomes

Longitudinal sampling of the stool has yielded important insights into the ecological dynamics of the human gut microbiome. However, human stool samples are available approximately once per day, while commensal population doubling times are likely on the order of minutes-to-hours. Despite this mismatch in timescales, much of the prior work on human gut microbiome time series modeling has assumed that day-to-day fluctuations in taxon abundances are related to population growth or death rates, which is likely not the case. Here, we propose an alternative model of the human gut as a stationary system, where population dynamics occur internally and the bacterial population sizes measured in a bolus of stool represent a steady-state endpoint of these dynamics. We formalize this idea as stochastic logistic growth. We show how this model provides a path toward estimating the growth phases of gut bacterial populations in situ. We validate our model predictions using an in vitro Escherichia coli growth experiment. Finally, we show how this method can be applied to densely-sampled human stool metagenomic time series data. We discuss how these growth phase estimates may be used to better inform metabolic modeling in flow-through ecosystems, like animal guts or industrial bioreactors.

59 BASIC BIOLOGICAL SCIENCES↗

Mapping Stochastic Devices to Probabilistic Algorithms

Probabilistic and Bayesian neural networks have long been proposed as a method to incorporate uncertainty about the world (both in training data and operation) into artificial intelligence applications. One approach to making a neural network probabilistic is to leverage a Monte Carlo sampling approach that samples a trained network while incorporating noise. Such sampling approaches for neural networks have not been extensively studied due to the prohibitive requirement of many computationally expensive samples. While the development of future microelectronics platforms that make this sampling more efficient is an attractive option, it has not been immediately clear how to sample a neural network and what the quality of random number generation should be. This research aimed to start addressing these two fundamental questions by examining basic “off the shelf” neural networks can be sampled through a few different mechanisms (including synapse “dropout” and neuron “dropout”) and examine how these sampling approaches can be evaluated both in terms of evaluating algorithm effectiveness and the required quality of random numbers.

97 MATHEMATICS AND COMPUTING↗

Ultrasonic Attenuation Due to Grain Boundary Scattering in Pure Niobium

Experimental confirmation of various grain scattering theories exist but there are still outstanding questions concerning the characterization of microstructure using ultrasound. In this study, high purity niobium serves as a model material devoid of extrinsic scattering centers (e.g., voids, precipitates, second phase particles etc.) A range of microstructures were obtained by annealing in ultrahigh vacuum at different temperatures (600–800°C) after routine cold-rolling and recrystallization heat treatment. Ultrasonic attenuation is measured as a function of a frequency for each sample. For the samples with an intermediate grain size (typically ∼50 μm), attenuation follows a power law dependence on frequency with an exponent of n∼1.6, which is close to the prediction, n=2, of classical Stochastic scattering theory. However, quantitative comparison shows that the observed attenuation is higher than predicted by the classical theory. The Stanke-Kino unified grain scattering theory may provide an explanation for the lower frequency dependence than traditional theories predict, though it still under predicts the magnitude of the attenuation. In any event, the resulting empirical relations provide a useful approach for practical grain size measurement with an acceptable level of uncertainty. The effect of a layered microstructure typical of some sheets/plates is discussed.

36 MATERIALS SCIENCE↗

Machine-learning error models for approximate solutions to parameterized systems of nonlinear equations

Here, this work proposes a machine-learning framework for constructing statistical models of errors incurred by approximate solutions to parameterized systems of nonlinear equations. These approximate solutions may arise from early termination of an iterative method, a lower-fidelity model, or a projection-based reduced-order model, for example. The proposed statistical model comprises the sum of a deterministic regression-function model and a stochastic noise model. The method constructs the regression-function model by applying regression techniques from machine learning (e.g., support vector regression, artificial neural networks) to map features (i.e., error indicators such as sampled elements of the residual) to a prediction of the approximate-solution error. The method constructs the noise model as a mean-zero Gaussian random variable whose variance is computed as the sample variance of the approximate-solution error on a test set; this variance can be interpreted as the epistemic uncertainty introduced by the approximate solution. This work considers a wide range of feature-engineering methods, data-set-construction techniques, and regression techniques that aim to ensure that (1) the features are cheaply computable, (2) the noise model exhibits low variance (i.e., low epistemic uncertainty introduced), and (3) the regression model generalizes to independent test data. Finally, numerical experiments performed on several computational-mechanics problems and types of approximate solutions demonstrate the ability of the method to generate statistical models of the error that satisfy these criteria and significantly outperform more commonly adopted approaches for error modeling.

97 MATHEMATICS AND COMPUTING↗

Fatigue behavior analysis and multi-scale modelling of chopped carbon fiber chip-reinforced composites under tension-tension loading condition

Here, the fatigue damage behavior of chopped carbon fiber chip-reinforced composite has been experimentally and numerically investigated. A large scatter is exhibited in the S-N diagram due to the random distribution of carbon fiber chips, and therefore a new analysis procedure is proposed to link the local microstructure to the fatigue behavior of the bulk material. Interrupted fatigue tests are also performed so that microstructure characterization can be conducted on the tested samples to analyze the crack initiation and propagation. Inspired from the experimental findings, a multi-scale progressive damage fatigue model is proposed to predict the fatigue behavior. The model incorporates a new stochastic chip-packing algorithm for microstructure reconstruction along with continuum damage models into the representative volume element (RVE) model in ABAQUS/Explicit. Finally, the simulation results based on the proposed model are in good agreement with the experimental data in terms of cracking modes and predicted life.

36 MATERIALS SCIENCE↗

Nuclear data uncertainty propagation and modeling uncertainty impact evaluation in neutronics core simulation

Uncertainty analysis is a critical requirement in reactor simulation as it is used to quantify the reliability of best-estimate calculation. A comprehensive uncertainty analysis should characterize all sources of uncertainties in a computationally-feasible and scientifically-defendable manner. Here we employ a well-established reduced order modeling (ROM) based uncertainty quantification methodology to propagate uncertainties throughout neutronic calculations. ROM relies on recent advances in randomized data mining techniques applied to large data streams. In our proposed implementation, the nuclear data uncertainties are first propagated from multi-group level through lattice physics calculation to generate few-group parameter uncertainties, described using a vector of mean values and a covariance matrix. Employing an ROM-based compression of the covariance matrix, the few-group uncertainties are then propagated through downstream core simulation in a computationally efficient manner. This straightforward approach, albeit efficient as compared to brute force forward and/or adjoint-based methods, often employs a number of assumptions that have been unquestioned in the literature of neutronic uncertainty analysis. This manuscript argues that these assumptions could introduce another source of uncertainty referred to as modeling uncertainties, whose magnitude needs to be quantified in tandem with nuclear data uncertainties. Thus, our primary goal is to explore the interactions between these two uncertainty sources in order to assess whether modeling uncertainties have an impact on parameter uncertainties. To explore this endeavor, the impact of a number of modeling assumptions on core attributes uncertainties is quantified. The study employs a CANDU reactor model, with Serpent and NEWT as lattice physics solvers and NESTLE-C as core simulator. The modeling assumptions investigated include those related with the uncertainty propagation method employed, e.g., deterministic vs. stochastic, the few-group energy structure employed to represent the cross-sections, the resonance treatment in lattice physics calculation, the reference values for the cross-section, and the number of samples employed to render ROM compression. Results indicate that some of the modeling assumptions could have a non-negligible impact on the core responses propagated uncertainties, highlighting the need for a more comprehensive approach to combine parameter and modeling uncertainties.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Free energy landscape and molecular pathways of gas hydrate nucleation

Despite the significance of gas hydrates in diverse areas, a quantitative knowledge of hydrate formation at a molecular level is missing. The impediment to acquiring this understanding is primarily attributed to the stochastic nature and ultra-fine scales of nucleation events, posing a great challenge for both experiment and simulation to explore hydrate nucleation. Here we employ advanced molecular simulation methods, including forward flux sampling (FFS), p{sub B} histogram analysis, and backward flux sampling, to overcome the limit of direct molecular simulation for exploring both the free energy landscape and molecular pathways of hydrate nucleation. First we test the half-cage order parameter (H-COP) which we developed for driving FFS, through conducting the p{sub B} histogram analysis. Our results indeed show that H-COP describes well the reaction coordinates of hydrate nucleation. Through the verified order parameter, we then directly compute the free energy landscape for hydrate nucleation by combining both forward and backward flux sampling. The calculated stationary distribution density, which is obtained independently of nucleation theory, is found to fit well against the classical nucleation theory (CNT). Subsequent analysis of the obtained large ensemble of hydrate nucleation trajectories show that although on average, hydrate formation is facilitated by a two-step like mechanism involving a gradual transition from an amorphous to a crystalline structure, there also exist nucleation pathways where hydrate crystallizes directly, without going through the amorphous stage. The CNT-like free energy profile and the structural diversity suggest the existence of multiple active transition pathways for hydrate nucleation, and possibly also imply the near degeneracy in their free energy profiles among different pathways. Our results thus bring a new perspective to the long standing question of how hydrates crystallize.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗