Search NASA⌕ Search

SEARCH · Search NASA

Results for “Smoothing methods”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,045 records · Page 58

Determination of Refractive Indices of a Layered Medium under Pulsed Irradiation

Problems related to determination of refractive indices of layered strongly scattering biological media by means of an optical tomography are investigated. Methods of reconstruction of refractive indices of layers, which are based on the smoothness properties of the solution to the initial-boundary problem, are proposed within the framework of the nonstationary radiative-transfer model. The influence of the probe-pulse duration on the accuracy of recovery of the sought-for characteristics of the medium is investigated numerically.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Bias-Variance Trade-Off in Physics-Informed Neural Networks with Randomized Smoothing for High-Dimensional PDEs

Physics-Informed Neural Networks (PINNs) have triggered a paradigm shift in scientific computing, leveraging mesh-free properties and robust approximation capabilities. While proving effective for low-dimensional partial differential equations (PDEs), the computational cost of PINNs remains a hurdle in high-dimensional scenarios. This is particularly pronounced when computing high-order and high-dimensional derivatives in the physics-informed loss. Randomized Smoothing PINN (RS-PINN) introduces Gaussian noise for stochastic smoothing of the original neural net model, enabling the use of Monte Carlo methods for derivative approximation, which eliminates the need for costly automatic differentiation. Despite its computational efficiency, especially in the approximation of high-dimensional derivatives, RS-PINN introduces biases in both loss and gradients, negatively impacting convergence, especially when coupled with stochastic gradient descent (SGD) algorithms. We present a comprehensive analysis of biases in RS-PINN, attributing them to the nonlinearity of the Mean Squared Error (MSE) loss as well as the intrinsic nonlinearity of the PDE itself. We propose tailored bias correction techniques, delineating their application based on the order of PDE nonlinearity. The derivation of an unbiased RS-PINN allows for a detailed examination of its advantages and disadvantages compared to the biased version. Specifically, the biased version has a lower variance and runs faster than the unbiased version, but it is less accurate due to the bias. To optimize the bias-variance trade-off, we combine the two approaches in a hybrid method that balances the rapid convergence of the biased version with the high accuracy of the unbiased version. In addition to methodological contributions, we present an enhanced implementation of RS-PINN. Extensive experiments on diverse high-dimensional PDEs, including Fokker-Planck, Hamilton-Jacobi-Bellman (HJB), viscous Burgers’, Allen-Cahn, and Sine-Gordon equations, illustrate the bias-variance trade-off and highlight the effectiveness of the hybrid RS-PINN. Empirical guidelines are provided for selecting biased, unbiased, or hybrid versions, depending on the dimensionality and nonlinearity of the specific PDE problem.

97 MATHEMATICS AND COMPUTING↗

Neutron Electric Dipole Moments with Clover Fermions

We present preliminary results for the contributions to the neutron EDM arising from the QCD θ-term, the Weinberg three-gluon and the quark chromo-EDM operators from our ongoing lattice calculations using clover valence quarks on the MILC HISQ lattices. We use the gradient-flow technique to smooth the lattices and renormalize the gluonic operators, and use the Schwinger source method to incorporate the quark chromo-EDM interactions in the quark propagator. For the QCD θ-term and the Weinberg three-gluon operator, we report results in the gradient-flow scheme from 8 ensembles at four lattice spacings and three pion masses, including 2 physical pion mass ensembles. For the quark chromo-EDM, unrenormalized results are presented at two lattice spacings, a=0.12 and 0.09 fm, and two pion masses, M π =310 MeV and 220 MeV.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Prediction of sonic boom at a focus

The behavior of sonic boom at a focus has been reviewed for the purpose of extending present sonic boom computational methods to include focal zones. The geometry of a focal zone - whether a smooth caustic, a cusped caustic, or a perfect focus to a point - determines the character of focused signatures. The seeming contradiction of various experimental data can be resolved by noting these differences. A ray acoustic analysis has been developed for quantitative determination of caustic geometry. The only reliable theory presently available for signatures at a focus is for a smooth caustic. There has been some controversy between theoretical and experimental values of a constant in the scaling law for this case. It has been found that this discrepancy can be resolved by accounting for the finite thickness of real sonic boom shock waves. These findings have been incorporated into an existing sonic boom computer program.

Plotkin, K. J.↗

Design of slotted transonic wind tunnels for supersonic flow development

A simple analytical method to predict the supersonic flow development in a two-dimensional slotted transonic wind tunnel has been developed and validated. While doing that, an insight has been gained into the flow mechanism which causes overexpansion with some slot shapes. As a consequence of the success of the analysis method, a design method has been developed on similar lines, to obtain slot shapes for prescribed smooth supersonic flow development at a design Mach number. The performance of these slot shapes at off-design conditions have been studied to enable choice of slot shapes to be made, based on acceptable performance over the whole supersonic range of interest.

Ramaswamy, M. A.↗

Second- and third-order upwind difference schemes for hyperbolic conservation laws

Second- and third-order two time-level five-point explicit upwind-difference schemes are described for the numerical solution of hyperbolic systems of conservation laws and applied to the Euler equations of inviscid gas dynamics. Nonliner smoothing techniques are used to make the schemes total variation diminishing. In the method both hyperbolicity and conservation properties of the hyperbolic conservation laws are combined in a very natural way by introducing a normalized Jacobian matrix of the hyperbolic system. Entropy satisfying shock transition operators which are consistent with the upwind differencing are locally introduced when transonic shock transition is detected. Schemes thus constructed are suitable for shockcapturing calculations. The stability and the global order of accuracy of the proposed schemes are examined. Numerical experiments for the inviscid Burgers equation and the compressible Euler equations in one and two space dimensions involving various situations of aerodynamic interest are included and compared.

Yang, J. Y.↗

Color graphics techniques for shaded surface displays of aerodynamic flowfield parameters

A variety of algorithms are presented that are useful in the color graphics displays of large sets of three-dimensional aerodynamic data. These methods augment established techniques for rapidly and interactively reviewing an assortment of smooth-shaded, color-coded illustrations for experimental and computational distributions of both scalar and vector quantities.

Weston, Robert P.↗

Spectral element multigrid. Part 2: Theoretical justification

A multigrid algorithm is analyzed which is used for solving iteratively the algebraic system resulting from tha approximation of a second order problem by spectral or spectral element methods. The analysis, performed here in the one dimensional case, justifies the good smoothing properties of the Jacobi preconditioner that was presented in Part 1 of this paper.

Maday, Yvon↗

SAGE: A 2-D self-adaptive grid evolution code and its application in computational fluid dynamics

SAGE is a user-friendly, highly efficient, two-dimensional self-adaptive grid code based on Nakahashi and Deiwert's variational principles method. Grid points are redistributed into regions of high flowfield gradients while maintaining smoothness and orthogonality of the grid. Efficiency is obtained by splitting the adaption into 2 directions and applying one-sided torsion control, thus producing a 1-D elliptic system that can be solved as a set of tridiagonal equations.

Davies, Carol B.↗

Integration of Design, Thermal, Structural, and Optical Analysis, Including Thermal Animation

In many industries there has recently been a concerted movement toward 'quality management' and the issue of how to accomplish work more efficiently. Part of this effort is focused on concurrent engineering; the idea of integrating the design and analysis processes so that they are not separate, sequential processes (often involving design rework due to analytical findings) but instead form an integrated system with smooth transfers of information. Presented herein are several specific examples of concurrent engineering methods being carried out at Langley Research Center (LaRC): integration of thermal, structural and optical analyses to predict changes in optical performance based on thermal and structural effects; integration of the CAD design process with thermal and structural analyses; and integration of analysis and presentation by animating the thermal response of a system as an active color map -- a highly effective visual indication of heat flow.

Amundsen, Ruth M.↗

Single-Scattering Properties of Melting Precipitation for GPM Passive Microwave and Radar Remote Sensing Applications

Over the past two decades, detailed computational simulations of the intricate three dimensional structures of ice-phase crystals and aggregates of those crystals have been developed. The microwave single-scattering properties of these simulated ice particles have been computed and used to improve quantitative estimates of snow rates and to better deRne the vertical structure of snow water contents in deep convective systems, as derived from satellite-borne passive microwave and/or radar remote sensing measurements. The same icephase particles have more recently been used as the starting point for simulations of melting precipitation using computational melting methods. In the current study, a heuristic melting method, as well as a physically-based melting procedure based on smoothed-particle hydrodynamics, are applied to ice particle models to describe the full evolution of the particles from dry snow to liquid drops. The discrete dipole approximation is utilized to calculate the single-scattering properties of the mixed-phase particles throughout the melting process. Then, the properties of the particles are “mapped” into simpliRed microphysical simulations of particle size spectra in the melting layers of stratiform, precipitating clouds. The bulk single-scattering properties of the melting layers and the sensitivity of their properties to modeling assumptions are explored, and the implications for combined radar-radiometer precipitation remote sensing from GPM are discussed.

William S Olson↗

Carbon nanotube substrates and catalyzed hot stamp for polishing and patterning the substrates

The present invention is generally directed to catalyzed hot stamp methods for polishing and/or patterning carbon nanotube-containing substrates. In some embodiments, the substrate, as a carbon nanotube fiber end, is brought into contact with a hot stamp (typically at 200-800.degree. C.), and is kept in contact with the hot stamp until the morphology/patterns on the hot stamp have been transferred to the substrate. In some embodiments, the hot stamp is made of material comprising one or more transition metals (Fe, Ni, Co, Pt, Ag, Au, etc.), which can catalyze the etching reaction of carbon with H.sub.2, CO.sub.2, H.sub.2O, and/or O.sub.2. Such methods can (1) polish the carbon nanotube-containing substrate with a microscopically smooth finish, and/or (2) transfer pre-defined patterns from the hot stamp to the substrate. Such polished or patterned carbon nanotube substrates can find application as carbon nanotube electrodes, field emitters, and field emitter arrays for displays and electron sources.

36 MATERIALS SCIENCE↗

A minimally invasive, efficient method for propagation of full-field uncertainty in solid dynamics

In this work, we present a minimally invasive method for forward propagation of material property uncertainty to full-field quantities of interest in solid dynamics. Full-field uncertainty quantification enables the design of complex systems where quantities of interest, such as failure points, are not known a priori. The method, motivated by the well-known probability density function (PDF) propagation method of turbulence modeling, uses an ensemble of solutions to provide the joint PDF of desired quantities at every point in the domain. A small subset of the ensemble is computed exactly, and the remainder of the samples are computed with approximation of the evolution equations based on those exact solutions. Although the proposed method has commonalities with traditional interpolatory stochastic collocation methods applied directly to quantities of interest, it is distinct and exploits the parameter dependence and smoothness of the driving term of the evolution equations. The implementation is model independent, storage and communication efficient, and straightforward. We demonstrate its efficiency, accuracy, scaling with dimension of the parameter space, and convergence in distribution with two problems: a quasi-one-dimensional bar impact, and a two material notched plate impact. For the bar impact problem, we provide an analytical solution to PDF of the solution fields for method validation. With the notched plate problem, we also demonstrate good parallel efficiency and scaling of the method.

42 ENGINEERING↗

A spectral boundary integral equation method for the 2-D Helmholtz equation

In this paper, we present a new numerical formulation of solving the boundary integral equations reformulated from the Helmholtz equation. The boundaries of the problems are assumed to be smooth closed contours. The solution on the boundary is treated as a periodic function, which is in turn approximated by a truncated Fourier series. A Fourier collocation method is followed in which the boundary integral equation is transformed into a system of algebraic equations. It is shown that in order to achieve spectral accuracy for the numerical formulation, the nonsmoothness of the integral kernels, associated with the Helmholtz equation, must be carefully removed. The emphasis of the paper is on investigating the essential elements of removing the nonsmoothness of the integral kernels in the spectral implementation. The present method is robust for a general boundary contour. Aspects of efficient implementation of the method using FFT are also discussed. A numerical example of wave scattering is given in which the exponential accuracy of the present numerical method is demonstrated.

Hu, Fang Q.↗

STRUCTURE IN THE 3D GALAXY DISTRIBUTION. II. VOIDS AND WATERSHEDS OF LOCAL MAXIMA AND MINIMA

The major uncertainties in studies of the multi-scale structure of the universe arise not from observational errors but from the variety of legitimate definitions and detection methods for individual structures. To facilitate the study of these methodological dependencies, we have carried out 12 different analyses defining structures in various ways. This has been done in a purely geometrical way by utilizing the HOP algorithm as a unique parameter-free method of assigning groups of galaxies to local density maxima or minima. From three density estimation techniques (smoothing kernels, Bayesian blocks, and self-organizing maps) applied to three data sets (the Sloan Digital Sky Survey Data Release 7, the Millennium simulation, and randomly distributed points) we tabulate information that can be used to construct catalogs of structures connected to local density maxima and minima. We also introduce a void finder that utilizes a method to assemble Delaunay tetrahedra into connected structures and characterizes regions empty of galaxies in the source catalog.

79 ASTRONOMY AND ASTROPHYSICS↗

An Evaluation of Bang-Time Measurements from a Multichannel, Triaxial, nTOF Diagnostic for MagLIF Experiments at the Z facility

Neutron bang times for a series of MagLIF (Magnetic Liner Inertial Fusion) experiments with D 2 -filled targets have been measured at the Z facility. The emitted neutrons were detected as current-mode pulses in a multichannel, neutron time-of-flight (nTOF) diagnostic with conventional, scintillator-photomultiplier-tube (PMT) detectors. In these experiments, the detectors were fielded at known, fixed distances L (690-2510 cm) from the target, and on three, non-coplanar (but convergent) lines-of-sight (LOS). The primary goal of this diagnostic was to estimate a fiducial time (bang time) relative to an externally generated time-base for synchronizing all the diagnostics in an experiment. Recorded arrival times (A7) of the pulses were characterized experimentally by three numerical methods: a first-moment estimate (centroid) and two nodal measures — Savitzky-Golay (SG) smoothing and a single point peak estimate of the raw data. These times were corrected for internal detector time delays (transit and impulse-response function) — an adjustment that linked the recorded ATs to the corresponding arrival of uncollided neutrons at each detector. The bang time was then estimated by linearly regressing the arrival times against the associated distances to the source; t bang (on the system timescale) was taken as the temporal intercept of the regression equation at distance L = 0. This article reports the analysis for a representative shot #2584 for which (a) the recorded ATs — even without detector corrections — agreed by method in each channel to within 1-2 ns; (b) internal corrections were each ~3 — 5 ns; and (c) a 95% uncertainty (confidence) interval for t bang in this shot was estimated at ±3 ns with 4 degrees of freedom. A secondary goal for this diagnostic was to check that the bang time measurements corresponded to neutrons emitted by the D(d,n) 3 He reaction in a thermalized DD plasma. According to the theoretical studies by Brysk, such neutrons should be emitted with an isotropic Gaussian distribution of mean kinetic energy $ \overline{E}$ of 2.449 MeV; this energy translates to a mean neutron speed $ \overline{u}$ of 2.160 cm/ns [D. H. Munro, Nuclear Fusion, 56(3) 036001 (2016)]. In the MagLIF series of shots there was no evidence of spatial asymmetry in the time-distance regressions, and it was possible to extract the mean neutron speed from the slope of these fits. In shot 2584 $ \overline{u}$ was estimated at 2.152 cm/ns ± 0.010 cm/ns [95 % confidence, 4 dof] and the mean kinetic energy $ \overline{E}$ (with relativistic corrections) was 2.431 MeV ± 0.022 MeV [95 % confidence, 4 dof] — results supporting the assumption that D-D neutrons were, in fact, measured.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Variational projection methods for gradient crystal plasticity using Lie algebras

Summary A computational method is developed for evaluating the plastic strain gradient hardening term within a crystal plasticity formulation. While such gradient terms reproduce the size effects exhibited in experiments, incorporating derivatives of the plastic strain yields a nonlocal constitutive model. Rather than applying mixed methods, we propose an alternative method whereby the plastic deformation gradient is variationally projected from the elemental integration points onto a smoothed nodal field. Crucially, the projection utilizes the mapping between Lie groups and algebras in order to preserve essential physical properties, such as orthogonality of the plastic rotation tensor. Following the projection, the plastic strain field is directly differentiated to yield the Nye tensor. Additionally, an augmentation scheme is introduced within the global Newton iteration loop such that the computed Nye tensor field is fed back into the stress update procedure. Effectively, this method results in a fully implicit evolution of the constitutive model within a traditional displacement‐based formulation. An elemental projection method with explicit time integration of the plastic rotation tensor is compared as a reference. A series of numerical tests are performed for several element types in order to assess the robustness of the method, with emphasis placed upon polycrystalline domains and multi‐axis loading. Copyright © 2016 John Wiley & Sons, Ltd.

Truster, Timothy J.↗

Some new function spaces of variable smoothness

A new Besov space of variable smoothness is introduced on which the norm is defined in terms of difference relations. This space is shown to be the trace of a weighted Sobolev space with a weight in the corresponding Muckenhoupt class. Methods of nonlinear spline approximation are applied to derive an atomic decomposition theorem for functions in a Besov space of variable smoothness. A complete description of traces on the hyperplane of a Besov space of variable smoothness and of a weighted Besov space with a weight in the corresponding Muckenhoupt class is given. Bibliography: 27 titles.

97 MATHEMATICS AND COMPUTING↗