Search NASA⌕ Search

SEARCH · Search NASA

Results for “stochastic approximation”

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 649 records · Page 36

Large Eddy simulation of the Sandia Flame series (D-F) using the Eulerian stochastic field method

Three turbulent piloted methane jet flames with increasing levels of local extinction (Sandia Flames D-F) have been computed using Large Eddy simulation. The smallest unresolved scales of the flow, in which combustion occurs, are represented using the filtered probability density function method where the corresponding evolution equation is solved directly. A dynamic model for the sub-grid stresses together with a simple gradient diffusion approximation for the scalar fluxes is applied in conjunction with the linear mean square estimation closure for sub-filter scale mixing. An augmented reduced mechanism (ARM) derived from the full GRI 3.0 mechanism is incorporated to describe the chemical reaction. The results demonstrate the ability of the method in capturing quantitatively finite rate effects such as extinction and re-ignition in turbulent flames. (author)

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Quantum Vibronic Effects on the Electronic Properties of Molecular Crystals

We present a study of molecular crystals, focused on the effect of nuclear quantum motion and anharmonicity on their electronic properties. We consider a system composed of relatively rigid molecules, a diamondoid crystal, and one composed of floppier molecules, NAI-DMAC, a thermally activated delayed fluorescence compound. We compute fundamental electronic gaps at the density functional theory (DFT) level of theory, with the Perdew–Burke–Erzenhof (PBE) and strongly constrained and approximately normed (SCAN) functionals, by coupling first-principles molecular dynamics with a nuclear quantum thermostat. We find a sizable zero-point renormalization (ZPR) of the band gaps, which is much larger in the case of diamondoids (0.6 eV) than for NAI-DMAC (0.22 eV). We show that the frozen phonon (FP) approximation, which neglects intermolecular anharmonic effects, leads to a large error (~50%) in the calculation of the band gap ZPR. Instead, when using a stochastic method, we obtain results in good agreement with those of our quantum simulations for the diamondoid crystal. However, the agreement is worse for NAI-DMAC where intramolecular anharmonicities contribute to the ZPR. Our results highlight the importance of accurately including nuclear and anharmonic quantum effects to predict the electronic properties of molecular crystals.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

The ITER 3D Magnetic Diagnostic Response to Applied n=3 and n=4 RMP's

The ITER magnetic diagnostic response to applied n=3 and n=4 RMPs has been calculated for the 15MA scenario. The VMEC code was utilized to calculate free boundary 3D ideal MHD equilibria, where the non-stellarator symmetric terms were included in the calculation. This allows an assessment to be made of the possible boundary displacements due to RMP application in ITER. As the VMEC code assumes a continuous set of nested flux surface, the possibility of island and stochastic region formation is ignored. At the start of the current at-top (L-Mode) application of n = 4 RMP's indicates approximately 1 cm peak-to-peak displacements on the low field side of the plasma while later in the shot (H-mode) perturbations as large as 3 cm are present. Forward modeling of the ITER magnetic diagnostics indicates significant non-axisymmetric plasma response, exceeding 10% the axisymmetric signal in many of the flux loops. Magnetic field probes seem to indicate a greater robustness to 3D effects but still indicate large sensitivities to 3D effects in a number of sensors. Forward modeling of the diagnostics response to 3D equilibria allows assessment of diagnostics design and control scenarios.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

The stochastic spectator

We study the stochastic distribution of spectator fields predicted in different slow-roll inflation backgrounds. Spectator fields have a negligible energy density during inflation but may play an important dynamical role later, even giving rise to primordial density perturbations within our observational horizon today. During de-Sitter expansion there is an equilibrium solution for the spectator field which is often used to estimate the stochastic distribution during slow-roll inflation. However slow roll only requires that the Hubble rate varies slowly compared to the Hubble time, while the time taken for the stochastic distribution to evolve to the de-Sitter equilibrium solution can be much longer than a Hubble time. We study both chaotic (monomial) and plateau inflaton potentials, with quadratic, quartic and axionic spectator fields. We give an adiabaticity condition for the spectator field distribution to relax to the de-Sitter equilibrium, and find that the de-Sitter approximation is never a reliable estimate for the typical distribution at the end of inflation for a quadratic spectator during monomial inflation. The existence of an adiabatic regime at early times can erase the dependence on initial conditions of the final distribution of field values. In these cases, spectator fields acquire sub-Planckian expectation values. Otherwise spectator fields may acquire much larger field displacements than suggested by the de-Sitter equilibrium solution. We quantify the information about initial conditions that can be obtained from the final field distribution. Our results may have important consequences for the viability of spectator models for the origin of structure, such as the simplest curvaton models.

79 ASTRONOMY AND ASTROPHYSICS↗

Non-Markovian dynamics of a damped driven two-state system

We study a driven two-state system interacting with a generic structured environment. We outline the derivation of the time-local microscopic non-Markovian master equation, in the limit of weak coupling between the system and the reservoir, and we derive its analytic solution for general reservoir spectra in the regime of validity of the secular approximation. We also consider the non-Markovian master equation without the secular approximation and study the effect of nonsecular terms on the system dynamics for two classes of reservoir spectra: the Ohmic and the Lorentzian reservoir. Finally, we derive the analytic conditions for complete positivity of the dynamical map, with and without the secular approximation. Interestingly, the complete positivity conditions have a transparent physical interpretation in terms of the characteristic time scales of phase diffusion and relaxation processes.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

MEASURING GAS ACCRETION AND ANGULAR MOMENTUM NEAR SIMULATED SUPERMASSIVE BLACK HOLES

Using cosmological simulations with a dynamic range in excess of 10{sup 7}, we study the transport of gas mass and angular momentum through the circumnuclear region of a disk galaxy containing a supermassive black hole (SMBH). The simulations follow fueling over relatively quiescent phases of the galaxy's evolution (no mergers) and without feedback from active galactic nuclei (AGNs), as part of the first stage of using state-of-the-art, high-resolution cosmological simulations to model galaxy and black hole co-evolution. We present results from simulations at different redshifts (z = 6, 4, and 3) and three different black hole masses (3 x 10{sup 7}, 9 x 10{sup 7}, and 3 x 10{sup 8} M{sub sun}; at z = 4), as well as a simulation including a prescription that approximates optically thick cooling in the densest regions. The interior gas mass throughout the circumnuclear disk shows transient and chaotic behavior as a function of time. The Fourier transform of the interior gas mass follows a power law with slope -1 throughout the region, indicating that, in the absence of the effects of galaxy mergers and AGN feedback, mass fluctuations are stochastic with no preferred timescale for accretion over the duration of each simulation ({approx}1-2 Myr). The angular momentum of the gas disk changes direction relative to the disk on kiloparsec scales over timescales less than 1 Myr, reflecting the chaotic and transient gas dynamics of the circumnuclear region. Infalling clumps of gas, which are driven inward as a result of the dynamical state of the circumnuclear disk, may play an important role in determining the spin evolution of an SMBH, as has been suggested in stochastic accretion scenarios.

79 ASTRONOMY AND ASTROPHYSICS↗

Superfluid to Mott insulator transition of hardcore bosons in a superlattice

We study the superfluid to Mott-insulator transition of hardcore bosons in commensurate superlattices in two and three dimensions. We focus on the special case where the superlattice has period two and the system is at half-filling. We obtain numerical results by using the stochastic series expansion algorithm, and compute various properties of the system, such as the ground-state energy, the density of bosons in the zero-momentum mode, the superfluid density, and the compressibility. We employ finite-size scaling to extrapolate the thermodynamic limit, and find the critical points of the phase transition. We also explore the extent to which several approximate solutions such as mean-field theory, with and without spin-wave corrections, can help one gain analytical insight into the behavior of the system in the vicinity of the phase transition.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Some Considerations on Stochastic Neutron Populations

The neutron population in a multiplying body containing a weak random source may depart considerably from its average or expected value. The resulting behavior of the system is then unpredictable, and a fully stochastic description of the neutron population becomes necessary. Stochastic considerations are especially important when dealing with pulsed reactors or in the case of criticality excursions in the presence of a weak source. Using the theory of discrete-state continuous-time Markov processes, and subject to some physical approximations, Bell obtained approximate solutions for the neutron number probability distributions (pdf), with and without an intrinsic random neutron source, that were valid at late times and/ large neutron populations. In recent work, we obtained exact solutions for Bell's model problem, and in this paper we use these exact probability distributions to: (I) assess the accuracy of Bell's asymptotic solutions and show how the latter follow from the exact solutions, (2) rigorously examine the probability of obtaining a divergent chain reaction, and (3) demonstrate the existence of an abrupt transition from a stochastic to a deterministic phase with increasing source strength.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Stochastic estimation of conditional eddies in turbulent channel flow

Several long-standing issues regarding linear estimation and coherent structures are addressed, and some questions that were addressed partially, but never with the benefit of full, three-dimensional information are answered. The objectives were to: determine how well linear estimates approximate the field obtained by true conditional averaging, using events such as those in quadrant analysis; determine the extent to which the three-dimensional linearly estimated fields correspond to coherent structures, and the degree and manner in which they differ; evaluate the type and nature of the structural information gained by employing several different types of events; and learn more about the 3-D structure of important coherent motions that occur in wall turbulence. The results indicate that linear stochastic estimation can be used effectively in the study of numerical data bases consisting of three-dimensional vector fields, both velocity and vorticity. Two-point stochastic estimation yields more structural information and more detail than single-point estimation. The structures observed occurred repeatedly within the flow, but much can not be said about their dominance or the probability of their occurrence without further systematic studies of their frequency.

Adrian, R. J.↗

GFINNs: GENERIC formalism informed neural networks for deterministic and stochastic dynamical systems

Here we propose the GENERIC formalism informed neural networks (GFINNs) that obey the symmetric degeneracy conditions of the GENERIC formalism. GFINNs comprise two modules, each of which contains two components. We model each component using a neural network whose architecture is designed to satisfy the required conditions. The component-wise architecture design provides flexible ways of leveraging available physics information into neural networks. We prove theoretically that GFINNs are sufficiently expressive to learn the underlying equations, hence establishing the universal approximation theorem. We demonstrate the performance of GFINNs in three simulation problems: gas containers exchanging heat and volume, thermoelastic double pendulum and the Langevin dynamics. In all the examples, GFINNs outperform existing methods, hence demonstrating good accuracy in predictions for both deterministic and stochastic systems.

97 MATHEMATICS AND COMPUTING↗

Finite Element Aircraft Simulation of Turbulence

A turbulence model has been developed for realtime aircraft simulation that accommodates stochastic turbulence and distributed discrete gusts as a function of the terrain. This model is applicable to conventional aircraft, V/STOL aircraft, and disc rotor model helicopter simulations. Vehicle angular activity in response to turbulence is computed from geometrical and temporal relationships rather than by using the conventional continuum approximations that assume uniform gust immersion and low frequency responses. By using techniques similar to those recently developed for blade-element rotor models, the angular-rate filters of conventional turbulence models are not required. The model produces rotational rates as well as air mass translational velocities in response to both stochastic and deterministic disturbances, where the discrete gusts and turbulence magnitudes may be correlated with significant terrain features or ship models. Assuming isotropy, a two-dimensional vertical turbulence field is created. A novel Gaussian interpolation technique is used to distribute vertical turbulence on the wing span or lateral rotor disc, and this distribution is used to compute roll responses. Air mass velocities are applied at significant centers of pressure in the computation of the aircraft's pitch and roll responses.

McFarland, R. E.↗

A mixed-integer conic programming formulation for computing the flexibility index under multivariate gaussian uncertainty

We present a methodology for computing the flexibility index when uncertainty is characterized using multivariate Gaussian random variables. Our approach computes the flexibility index by solving a mixed-integer conic program (MICP). This methodology directly characterizes ellipsoidal sets to capture correlations in contrast to previous methodologies that employ approximations. We also show that, under a Gaussian representation, the flexibility index can be used to obtain a lower bound for the so-called stochastic flexibility index (i.e., the probability of having feasible operation). Our results also show that the methodology can be generalized to capture different types of uncertainty sets.

97 MATHEMATICS AND COMPUTING↗

Quinoa, Version 0.2

Quinoa is a set of computational tools that enables research and numerical analysis in fluid dynamics. At this time it remains a test-bed to experiment with various algorithms using fully asynchronous runtime systems. Currently, Quinoa consists of the following tools: (1) Walker, a numerical integrator for systems of stochastic differential equations in time. It is a mathematical tool to analyze and design the behavior of stochastic differential equations. It allows the estimation of arbitrary coupled statistics and probability density functions and is currently used for the design of statistical moment approximations for multiple mixing materials in variable-density turbulence. (2) Inciter, an overdecomposition-aware finite element field solver for partial differential equations using 3D unstructured grids. Inciter is used to research asynchronous mesh-based algorithms and to experiment with coupling asynchronous to bulk-synchronous parallel code. Two planned new features of Inciter, compared to the previous release (LA-CC-16-015), to be implemented in 2017, are (a) a simple Navier-Stokes solver for ideal single-material compressible gases, and (b) solution-adaptive mesh refinement (AMR), which enables dynamically concentrating compute resources to regions with interesting physics. Using the NS-AMR problem we plan to explore how to scale such high-load-imbalance simulations, representative of large production multiphysics codes, to very large problems on very large computers using an asynchronous runtime system. (3) RNGTest, a test harness to subject random number generators to stringent statistical tests enabling quantitative ranking with respect to their quality and computational cost. (4) UnitTest, a unit test harness, running hundreds of tests per second, capable of testing serial, synchronous, and asynchronous functions. (5) MeshConv, a mesh file converter that can be used to convert 3D tetrahedron meshes from and to either of the following formats: Gmsh, http://www.geuz.org/gmsh, Netgen, http://sourceforge.net/apps/mediawiki/netgen-mesher, ExodusII, http://sourceforge.net/projects/exodusii, HyperMesh, http://www.altairhyperworks.com/product/HyperMesh.

Bakosi, Jozsef↗

Summer term 2018 work with IMC radiative transport in binary stochastic media

In the quest for new and improved methods to solve thermal radiation transport in stochastic media, this summer project was to characterize radation transport in a deterministic lattice problem that represents a real, binary stochastic media. A representative geometry for a binary stochastic media consisting of a cube of a high volume fraction media containing constant-diameter non-overlappying spheres randomly distributed within was run previously. Previous thermal radiation wave problems in binary stochastic media displayed three characteristic thermal energy deposition wavefronts in the following order, front of the spheres, bulk material, and the backs of the spheres. Several physical aspects of the representative geometry for a stochastic binary media using an optically thin background with optically thick spheres were analyzed. This analysis includes wave speed propagation using a one dimensional diffusion approximation, energy fluence transmission, and material energy deposition. The approaches to analysis included representing one sphere of the high opacity material in a diffusive background of the low opacity material, as well as repeating spheres in a column con guration with reflective boundary conditions to approximate a lattice. Additionally, both atomic mix and Levermore-Pomraning homogenization schemes were compared against the representative cube geometry as well as the sphere column lattice geometry. The wave speed analysis with a one dimensional Marshak diffusion wave was largely ineffective at determining an effective opacity value to which a problem could be homogenized, likely due to low mean free path resolution within the problem. However, the sphere column lattice geometry showed promising results for approximating the energy fluence transmitted through the problem domain. One concern with the sphere column lattice geometry indicated the high forward-bias effect of fluence reaching the problem boundary due to streaming paths around the spheres in the case of small mean free path resolution of the background material.

36 MATERIALS SCIENCE↗

AEOLUS: Advances in Experimental Design, Optimal Control, and Learning for Uncertain Complex Systems

Sustained advances in the mathematics of modeling and simulation have resulted in the capability today for routine simulation of a number of large scale complex DOE-relevant systems. As remarkable as this capability for solving the so-called forward problem is, it is typically only the first step-an inner loop within an outer loop that explores the simulation model's parameter space and decision space to characterize uncertainty in the model's predictions, learn unknown model parameters from data, design the most informative experiments, determine optimal control strategies, and create optimal designs. Broadly, what unifies all of these outer loop problems is that they are, in one form or another, optimization problems over parameter/control/design space that are constrained by complex uncertain models. To fully realize the power of scientific simulation as a basis for scientific discovery, technological innovation, and rational decision-making, it is imperative to move beyond simulation to tackle the outer loop of optimization for learning from data, experimental design, and control with complex uncertain models. When the models under consideration are large-scale and complex, and when the optimization variable and uncertain parameter spaces are high (or infinite) dimensional, this constitutes a grand challenge of the highest order, and is intractable with conventional methods. To overcome these challenges, the AEOLUS Center was established to develop a unified mathematical, computational, and statistical framework for (1) Learning predictive models from complex data via Bayesian inference and optimization, and (2) Optimizing experiments, processes, and designs using the resulting uncertain models. These problems are intractable with conventional methods, for several reasons: (1) The simulation problems that govern the inner loops of the optimization problems are expensive to execute (due to severe nonlinearity, heterogeneity, multiphysics/multiscale coupling); (2) The optimization variable and uncertain parameter spaces are high dimensional, often stemming from discretizations of infinite dimensional fields such as initial conditions, sources, or material properties. We argue that the key to overcoming these challenges is to develop new mathematical, computational, and statistical methods that exploit the structure of the Bayesian inference and optimization problems mediated by their underlying complex uncertain models. This structure includes the regularity, sparsity, geometry, low intrinsic dimensionality, and multifidelity nature of the maps from uncertain parameter/optimization variable spaces to the specific objectives targeted: Bayesian inference, optimal experimental design, and optimal control design. Black box methods developed as generic tools are incapable of exploiting this structure. To be successful, we must create, integrate, and cross-fertilize ideas across multiple areas of applied math--including approximation theory, Bayesian inference, data science, experimental design, information theory, machine learning, model reduction, optimal control theory, parallel algorithms, PDE-constrained optimization, randomized algorithms, stochastic optimization, and uncertainty quantification--all while exploiting the structure of the problems at hand. With this goal in mind, we have marshaled a team of leading authorities in these areas. While the methods we develop will be broadly applicable across a wide spectrum of DOE problems in which experiments inform models and the systems those models describe must be optimized under uncertainty, we have chosen a specific area, advanced manufacturing and materials, to drive our work. AMM is characterized by complex models across multiple scales, and is a rich source of challenging problems in inference, experimental design, and optimal control, requiring multifaceted and integrated advances in applied mathematics. As such, AMM serves as an excellent vehicle to motivate and demonstrate the advances in applied mathematics developed by our center.

97 MATHEMATICS AND COMPUTING↗

An empirical polytrope law for solar wind thermal electrons between 0.45 and 4.76 AU: Voyager 2 and Mariner 10

Empirical evidence is presented that solar wind thermal electrons obey a polytrope law with polytrope index gamma = 1.175 plus or minus 0.03. The Voyager 2 and Mariner 10 data used as evidence are compared and discussed. The theoretical predictions that solar wind thermal electrons in the asymptotic solar wind should obey a polytrope law with polytrope index gamma = 1.16 plus or minus. The widespread impressions in the literature that solar wind electrons behave more like an isothermal than adiabatic gas, and the arguments that Coulomb collisions are the dominant stochastic process shaping observed electron distribution functions in the solar wind are reexamined, reviewed and evaluated. The assignment of the interplanetary potential as equal to approximately seven times the temperature of the thermal electrons is discussed.

Sittler, E. C., Jr.↗

Oscillator response to nonstationary excitation

Analytical solutions are presented regarding probability density distributions of various response parameters of a lightly damped oscillator. The oscillator is subjected to a broad-band stochastic excitation which possesses a time-variant power spectrum. The analytical solutions are derived by utilizing appropriate Fokker-Planck equations which govern Markovian approximations of the response parameters considered. The reliability of the approximate analytical solution is tested by using pertinent data generated by a digital Monte Carlo study.

Spanos, P.-T. D.↗

A Discrete Probability Function Method for the Equation of Radiative Transfer

A discrete probability function (DPF) method for the equation of radiative transfer is derived. The DPF is defined as the integral of the probability density function (PDF) over a discrete interval. The derivation allows the evaluation of the PDF of intensities leaving desired radiation paths including turbulence-radiation interactions without the use of computer intensive stochastic methods. The DPF method has a distinct advantage over conventional PDF methods since the creation of a partial differential equation from the equation of transfer is avoided. Further, convergence of all moments of intensity is guaranteed at the basic level of simulation unlike the stochastic method where the number of realizations for convergence of higher order moments increases rapidly. The DPF method is described for a representative path with approximately integral-length scale-sized spatial discretization. The results show good agreement with measurements in a propylene/air flame except for the effects of intermittency resulting from highly correlated realizations. The method can be extended to the treatment of spatial correlations as described in the Appendix. However, information regarding spatial correlations in turbulent flames is needed prior to the execution of this extension.

Sivathanu, Y. R.↗