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 235 records · Page 13

An adaptive Hessian approximated stochastic gradient MCMC method

Bayesian approaches have been successfully integrated into training deep neural networks. One popular family is stochastic gradient Markov chain Monte Carlo methods (SG-MCMC), which have gained increasing interest due to their ability to handle large datasets and the potential to avoid overfitting. Although standard SG-MCMC methods have shown great performance in a variety of problems, they may be inefficient when the random variables in the target posterior densities have scale differences or are highly correlated. Here, we present an adaptive Hessian approximated stochastic gradient MCMC method to incorporate local geometric information while sampling from the posterior. The idea is to apply stochastic approximation (SA) to sequentially update a preconditioning matrix at each iteration. The preconditioner possesses second-order information and can guide the random walk of a sampler efficiently. Instead of computing and saving the full Hessian of the log posterior, we use limited memory of the samples and their stochastic gradients to approximate the inverse Hessian-vector multiplication in the updating formula. Moreover, by smoothly optimizing the preconditioning matrix via SA, our proposed algorithm can asymptotically converge to the target distribution with a controllable bias under mild conditions. To reduce the training and testing computational burden, we adopt a magnitude-based weight pruning method to enforce the sparsity of the network. Our method is user-friendly and demonstrates better learning results compared to standard SG-MCMC updating rules. The approximation of inverse Hessian alleviates storage and computational complexities for large dimensional models. Numerical experiments are performed on several problems, including sampling from 2D correlated distribution, synthetic regression problems, and learning the numerical solutions of heterogeneous elliptic PDE. The numerical results demonstrate great improvement in both the convergence rate and accuracy.

97 MATHEMATICS AND COMPUTING↗

A heterogeneous stochastic FEM framework for elliptic PDEs

We introduce a new concept of sparsity for the stochastic elliptic operator −div(a(x,ω)∇(⋅)), which reflects the compactness of its inverse operator in the stochastic direction and allows for spatially heterogeneous stochastic structure. This new concept of sparsity motivates a heterogeneous stochastic finite element method (HSFEM) framework for linear elliptic equations, which discretizes the equations using the heterogeneous coupling of spatial basis with local stochastic basis to exploit the local stochastic structure of the solution space. We also provide a sampling method to construct the local stochastic basis for this framework using the randomized range finding techniques. The resulting HSFEM involves two stages and suits the multi-query setting: in the offline stage, the local stochastic structure of the solution space is identified; in the online stage, the equation can be efficiently solved for multiple forcing functions. An online error estimation and correction procedure through Monte Carlo sampling is given. Numerical results for several problems with high dimensional stochastic input are presented to demonstrate the efficiency of the HSFEM in the online stage.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

The range of variation of the mass of the most massive star in stellar clusters derived from 35 million Monte Carlo simulations

A growing fraction of simple stellar population models, in an aim to create more realistic simulations capable of including stochastic variation in their outputs, begin their simulations with a distribution of discrete stars following a power-law function of masses. Careful attention is needed to create a correctly sampled initial mass function (IMF), and here we provide a solid mathematical method, called MASSCLEAN IMF Sampling, for doing so. We use our method to perform 10 million MASSCLEAN Monte Carlo stellar cluster simulations to determine the most massive star in a mass distribution as a function of the total mass of the cluster. We find that a maximum mass range is predicted, not a single maximum mass. This range is (1) dependent on the total mass of the cluster and (2) independent of an upper stellar mass limit, M{sub limit} , for unsaturated clusters and emerges naturally from our IMF sampling method. We then turn our analysis around, starting with our new database of 25 million simulated clusters, to constrain the highest mass star from the observed integrated colors of a sample of 40 low-mass Large Magellanic Cloud stellar clusters of known age and mass. Finally, we present an analytical description of the maximum mass range of the most massive star as a function of the cluster's total mass and present a new M{sub max} -M{sub cluster} relation.

79 ASTRONOMY AND ASTROPHYSICS↗

Time Series, Stochastic Processes and Completeness of Quantum Theory

Most of physical experiments are usually described as repeated measurements of some random variables. Experimental data registered by on-line computers form time series of outcomes. The frequencies of different outcomes are compared with the probabilities provided by the algorithms of quantum theory (QT). In spite of statistical predictions of QT a claim was made that it provided the most complete description of the data and of the underlying physical phenomena. This claim could be easily rejected if some fine structures, averaged out in the standard descriptive statistical analysis, were found in time series of experimental data. To search for these structures one has to use more subtle statistical tools which were developed to study time series produced by various stochastic processes. In this talk we review some of these tools. As an example we show how the standard descriptive statistical analysis of the data is unable to reveal a fine structure in a simulated sample of AR (2) stochastic process. We emphasize once again that the violation of Bell inequalities gives no information on the completeness or the non locality of QT. The appropriate way to test the completeness of quantum theory is to search for fine structures in time series of the experimental data by means of the purity tests or by studying the autocorrelation and partial autocorrelation functions.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

STOCHASTIC OPTIMIZATION FOR LONG TERM CAPITAL STRUCTURES, SYSTEMS, AND COMPONENTS REFURBISHMENT AND REPLACEMENT

As commercial nuclear power plants (NPPs) pursue extended plant operations in the form of Second License Renewals (SLRs), opportunities exist for these plants to provide capital investments to ensure long-term, safe, and economic performance. Several utilities have already announced their intention to pursue extended operations for one or more of their NPPs via SLR2. The goal of this research is to develop a riskinformed approach to evaluate and prioritize plant capital investments made in preparation for, and during the period of, extended plant operations to support decisions in NPP operations. In order to prioritize project selection via a riskinformed approach we developed a single decision-making tool that integrates safety/reliability, cost, and stochastic optimization models to provide users with data analysis capabilities to more cost effectively manage plant assets. Both stochastic analysis methods—such as Monte Carlo-based sampling strategies—and multi-stage stochastic optimization strategies are employed to provide priority lists to decisionmakers in support of risk-informed decisions. We applied the proposed method to a trial application of projected replacement/refurbishment expenditures for plant capital assets (i.e., Structures, Systems, and Components [SSCs]). The objective is to optimize the SSC replacement/refurbishment schedule in terms of economic constraints, data uncertainties, and SSC reliability data, as well to generate a priority list for maximizing returns on investment.

42 ENGINEERING↗

Score-based deterministic density sampling

We propose a deterministic sampling framework using Score-Based Transport Modeling for sampling an unnormalized target density π given only its score ∇ log π. Our method approximates the Wasserstein gradient flow on KL($f_t$∥π) by learning the time-varying score ∇ log $f_t$ on the fly using score matching. While having the same marginal distribution as Langevin dynamics, our method produces smooth deterministic trajectories, resulting in monotone noise-free convergence. We prove that our method dissipates relative entropy at the same rate as the exact gradient flow, provided sufficient training. Numerical experiments validate our theoretical findings: our method converges at the optimal rate, has smooth trajectories, and is often more sample efficient than its stochastic counterpart. Experiments on high-dimensional image data show that our method produces high-quality generations in as few as 15 steps and exhibits natural exploratory behavior. The memory and runtime scale linearly in the sample size.

97 MATHEMATICS AND COMPUTING↗

Ensemble Grouping Strategies for Embedded Stochastic Collocation Methods Applied to Anisotropic Diffusion Problems

Previous work has demonstrated that propagating groups of samples, called ensembles, together through forward simulations can dramatically reduce the aggregate cost of sampling-based uncertainty propagation methods [E. Phipps, M. D'Elia, H. C. Edwards, M. Hoemmen, J. Hu, and S. Rajamanickam, SIAM J. Sci. Comput., 39 (2017), pp. C162--C193]. However, critical to the success of this approach when applied to challenging problems of scientific interest is the grouping of samples into ensembles to minimize the total computational work. For example, the total number of linear solver iterations for ensemble systems may be strongly influenced by which samples form the ensemble when applying iterative linear solvers to parameterized and stochastic linear systems. In this paper we explore sample grouping strategies for local adaptive stochastic collocation methods applied to PDEs with uncertain input data, in particular canonical anisotropic diffusion problems where the diffusion coefficient is modeled by truncated Karhunen--Loève expansions. Finally, we demonstrate that a measure of the total anisotropy of the diffusion coefficient is a good surrogate for the number of linear solver iterations for each sample and therefore provides a simple and effective metric for grouping samples.

97 MATHEMATICS AND COMPUTING↗

Partner with a Third-Party Delivery Service or Not? A Prediction-and-Decision Tool for Restaurants Facing Takeout Demand Surges During a Pandemic

Amidst the COVID-19 pandemic, restaurants become more reliant on no-contact pick-up or delivery ways for serving customers. As a result, they need to make tactical planning decisions such as whether to partner with online platforms, to form their own delivery team, or both. In this paper, we develop an integrated prediction-decision model to analyze the profit of combining the two approaches and to decide the needed number of drivers under stochastic demand. We first use the susceptible-infected-recovered (SIR) model to forecast future infected cases in a given region and then construct an autoregressive-moving-average (ARMA) regression model to predict food-ordering demand. Using predicted demand samples, we formulate a stochastic integer program to optimize food delivery plans. We conduct numerical studies using COVID-19 data and food-ordering demand data collected from local restaurants in Nuevo Leon, Mexico, from April to October 2020, to show results for helping restaurants build contingency plans under rapid market changes. Our method can be used under unexpected demand surges, various infection/vaccination status, and demand patterns. Here, our results show that a restaurant can benefit from partnering with third-party delivery platforms when (i) the subscription fee is low, (ii) customers can flexibly decide whether to order from platforms or from restaurants directly, (iii) customers require more efficient delivery, (iv) average delivery distance is long, or (v) demand variance is high.

97 MATHEMATICS AND COMPUTING↗

Multistart algorithm for identifying all optima of nonconvex stochastic functions

Here, we propose a multistart algorithm to identify all local minima of a constrained, nonconvex stochastic optimization problem. The algorithm uniformly samples points in the domain and then starts a local stochastic optimization run from any point that is the "probabilistically best" point in its neighborhood. Under certain conditions, our algorithm is shown to asymptotically identify all local optima with high probability; this holds even though our algorithm is shown to almost surely start only finitely many local stochastic optimization runs. We demonstrate the performance of an implementation of our algorithm on nonconvex stochastic optimization problems, including identifying optimal variational parameters for the quantum approximate optimization algorithm.

97 MATHEMATICS AND COMPUTING↗

RAVEN Theory Manual

RAVEN is a software framework able to perform parametric and stochastic analysis based on the response of complex system codes. The initial development was aimed at providing dynamic risk analysis capabilities to the thermohydraulic code RELAP-7, currently under development at Idaho National Laboratory (INL). Although the initial goal has been fully accomplished, RAVEN is now a multi-purpose stochastic and uncertainty quantification platform, capable of communicating with any system code. In fact, the provided Application Programming Interfaces (APIs) allow RAVEN to interact with any code as long as all the parameters that need to be perturbed are accessible by input files or via python interfaces. RAVEN is capable of investigating system response and explore input space using various sampling schemes such as Monte Carlo, grid, or Latin hypercube. However, RAVEN strength lies in its system feature discovery capabilities such as: constructing limit surfaces, separating regions of the input space leading to system failure, and using dynamic supervised learning techniques. The development of RAVEN started in 2012 when, within the Nuclear Energy Advanced Modeling and Simulation (NEAMS) program, the need to provide a modern risk evaluation framework arose. RAVEN’s principal assignment is to provide the necessary software and algorithms in order to employ the concepts developed by the Risk Informed Safety Margin Characterization (RISMC) program. RISMC is one of the pathways defined within the Light Water Reactor Sustainability (LWRS) program. In the RISMC approach, the goal is not just to identify the frequency of an event potentially leading to a system failure, but the proximity (or lack thereof) to key safety-related events. Hence, the approach is interested in identifying and increasing the safety margins related to those events. A safety margin is a numerical value quantifying the probability that a safety metric (e.g. peak pressure in a pipe) is exceeded under certain conditions. Most of the capabilities, implemented having RELAP-7 as a principal focus, are easily deployable to other system codes. For this reason, several side activates have been employed (e.g. RELAP5-3D, any MOOSE-based App, etc.) or are currently ongoing for coupling RAVEN with several different software. The aim of this document is to provide a set of commented examples that can help the user to become familiar with the RAVEN code usage.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Stochastic Inversion of 2D Magnetotelluric Data

The algorithm is developed to invert 2D magnetotelluric (MT) data based on sharp boundary parametrization using a Bayesian framework. Within the algorithm, we consider the locations and the resistivity of regions formed by the interfaces are as unknowns. We use a parallel, adaptive finite-element algorithm to forward simulate frequency-domain MT responses of 2D conductivity structure. Those unknown parameters are spatially correlated and are described by a geostatistical model. The joint posterior probability distribution function is explored by Markov Chain Monte Carlo (MCMC) sampling methods. The developed stochastic model is effective for estimating the interface locations and resistivity. Most importantly, it provides details uncertainty information on each unknown parameter. Hardware requirements: PC, Supercomputer, Multi-platform, Workstation; Software requirements C and Fortan; Operation Systems/version is Linux/Unix or Windows

Chen, Jinsong↗

A diffusive information preservation method for small Knudsen number flows

The direct simulation Monte Carlo (DSMC) method is a powerful particle-based method for modeling gas flows. It works well for relatively large Knudsen (Kn) numbers, typically larger than 0.01, but quickly becomes computationally intensive as Kn decreases due to its time step and cell size limitations. An alternative approach was proposed to relax or remove these limitations, based on replacing pairwise collisions with a stochastic model corresponding to the Fokker–Planck equation [J. Comput. Phys., 229, 1077 (2010); J. Fluid Mech., 680, 574 (2011)]. Similar to the DSMC method, the downside of that approach suffers from computationally statistical noise. To solve the problem, a diffusion-based information preservation (D-IP) method has been developed. The main idea is to track the motion of a simulated molecule from the diffusive standpoint, and obtain the flow velocity and temperature through sampling and averaging the IP quantities. To validate the idea and the corresponding model, several benchmark problems with Kn ∼ 10{sup −3}–10{sup −4} have been investigated. It is shown that the IP calculations are not only accurate, but also efficient because they make possible using a time step and cell size over an order of magnitude larger than the mean collision time and mean free path, respectively.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Learning the temporal evolution of multivariate densities via normalizing flows

In this work, we propose a method to learn multivariate probability distributions using sample path data from stochastic differential equations. Specifically, we consider temporally evolving probability distributions (e.g., those produced by integrating local or nonlocal Fokker–Planck equations). Here, we analyze this evolution through machine learning assisted construction of a time-dependent mapping that takes a reference distribution (say, a Gaussian) to each and every instance of our evolving distribution. If the reference distribution is the initial condition of a Fokker–Planck equation, what we learn is the time-T map of the corresponding solution. Specifically, the learned map is a multivariate normalizing flow that deforms the support of the reference density to the support of each and every density snapshot in time. We demonstrate that this approach can approximate probability density function evolutions in time from observed sampled data for systems driven by both Brownian and Lévy noise. We present examples with two- and three-dimensional, uni- and multimodal distributions to validate the method.

97 MATHEMATICS AND COMPUTING↗

Adaptive Sampling Trust Region Method for Bi-fidelity Simulation Optimization [SWR-25-166]

Adaptive Sampling Trust Region Method for Bi-fidelity Simulation Optimization aims to demonstrate the effect of adaptive sampling-based bi-fidelity stochastic trust region method (ASTRO-BFDF). ASTRO-BFDF, derived from a derivative-free adaptive sampling trust-region optimization (ASTRO-DF) (Shashaani et al. 2018, Ha and Shashaani 2023), intended to efficiently solve the bi-fidelity simulation optimization.

Mueller, Juliane [National Laboratory of the Rocki↗

SDSS-IV MaNGA: How Galaxy Interactions Influence Active Galactic Nuclei

We present a comparative study of active galactic nuclei (AGN) between galaxy pairs and isolated galaxies with the final data release of the MaNGA integral field spectroscopic survey. We build a sample of 391 kinematic galaxy pairs within the footprint of the survey and select AGN using the survey's spectra. We use the comoving volume densities of the AGN samples to quantify the effects that tidal interactions have on the triggering of nuclear accretion. Our hypothesis is that the pair sample contains AGN that are triggered by not only stochastic accretion but also tidally induced accretion and correlated accretion. With the level of stochastically triggered AGN fixed by the control sample, we model the strength of tidally induced accretion and correlated accretion as a function of projected separation (r p ) and compare the model expectations with the observed volume densities of dual AGN and offset AGN (single AGN in a pair). At r p ~ 10 kpc, we find that tidal interactions induce ~30% more AGN than stochastic fueling and cause ~12% of the offset AGN to become dual AGN because of correlations. The strength of both these effects decreases with increasing r p . We also find that the [O III ] luminosities of the AGN in galaxy pairs are consistent with those found in isolated galaxies, likely because stochastically fed AGN dominate even among close pairs. Our results illustrate that while we can detect tidally induced effects statistically, it is challenging to separate tidally induced AGN and stochastically triggered AGN in interacting galaxies.

79 ASTRONOMY AND ASTROPHYSICS↗

Design optimization of a scramjet under uncertainty using probabilistic learning on manifolds

Here, we demonstrate, on a scramjet combustion problem, a constrained probabilistic learning approach that augments physics-based datasets with realizations that adhere to underlying constraints and scatter. The constraints are captured and delineated through diffusion maps, while the scatter is captured and sampled through a projected stochastic differential equation. The objective function and constraints of the optimization problem are then efficiently framed as non-parametric conditional expectations. Different spatial resolutions of a large-eddy simulation filter are used to explore the robustness of the model to the training dataset and to gain insight into the significance of spatial resolution on optimal design.

42 ENGINEERING↗

Interplanetary dust particles collected from the stratosphere: Physical, chemical, and mineralogical properties and implications for their sources

The suggestion that significant quantities of interplanetary dust are produced by both main-belt asteroids and comets is based on the Infrared Astronomical Satellite (IRAS) detection of dust trails or bands associated with these objects. Gravitational focusing strongly biases all near-Earth collections of interplanetary dust in favor of particles with the lowest geocentric velocities, that is the dust from main-belt asteroids spiraling into the Sun under the influence of Poynting-Robertson radiation drag. The major dust bands in the main-belt appear to be associated with the catastrophic disruptions which produced the Eos, Themis and Koronis families of asteroids. If dust particles are produced in the catastrophic collision process, then Poynting-Robertson radiation drag is such an efficient transport mechanism from the main-belt to 1 AU that near-Earth collections of interplanetary dust should include, and perhaps be dominated by, this material. Interplanetary dust particles from 5 to 100 micrometers in diameter have been recovered from the stratosphere of the Earth by NASA sampling aircraft since the mid-1970s. The densities of a large fraction of these interplanetary dust particles are significantly lower than the densities of their constituent silicate mineral phases, indicating significant porosites. The majority of the particles are chemically and mineralogically similar to, but not identical to, the carbonaceous chondrite meteorites. Most stony interplanetary dust particles have carbon contents exceeding those of Allende, a carbonaceous chondrite meteorite having a low albedo. Higher albedo particles corresponding to S-type asteroids are underrepresented or absent from the stratospheric collections, and primitive carbonaceous particles seem to be overrepresented in the stratospheric collections compared to the fraction of main-belt asteroids classified as primitive. This suggests that much of the interplanetary dust may be generated by a stochastic process, probably preferentially sampling a few most recent collisional events.

Flynn, George J.↗