Search NASA⌕ Search

SEARCH · Search NASA

Results for “Discretization error”

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 73 records · Page 4

Non-singlet quark helicity $\mathrm{PDF}$s of the nucleon from pseudo-distributions

The non-singlet helicity quark parton distribution functions (PDFs) of the nucleon are determined from lattice QCD, by jointly leveraging pseudo-distributions and the distillation spatial smearing paradigm. A Lorentz decomposition of appropriately isolated space-like matrix elements reveals pseudo-distributions that contain information on the leading-twist helicity PDFs, as well as an invariant amplitude that induces an additional z 2 contamination of the leading-twist signal. An analysis of the short-distance behavior of the space-like matrix elements using matching coefficients computed to next-to-leading order (NLO) exposes the desired PDF up to this additional z 2 contamination. Due to the non-conservation of the axial current, we elect to isolate the helicity PDFs normalized by the nucleon axial charge at the same scale μ 2 . The leading-twist helicity PDFs as well as several sources of systematic error, including higher-twist effects, discretization errors, and the aforementioned z 2 contaminating amplitude are jointly determined by characterizing the computed pseudo-distribution in a basis of Jacobi polynomials. The Akaike Information Criterion is exploited to effectively average over distinct model parameterizations and cuts on the pseudo-distribution. Encouraging agreement is observed with recent global analyses of each non-singlet quark helicity PDF, notably a rather small non-singlet anti-quark helicity PDF for all quark momentum fractions.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Whitepaper prepared for DOE Workshop on Integrated Simulations for Magnetic Fusion Energy Sciences: Topic E: Beyond interpretive simulations

Predictive simulations of magnetic confinement fusion and burning plasmas will require the quantification of all uncertainties and errors relating to the simulation capabilities. These include: discretization error (temporal and spatial); incomplete convergence error (nonlinear, linear, etc.); uncertainties in input data (initial conditions, boundary conditions, coefficients, thermo-physical properties, source terms, etc.); and uncertainties in the component models (the specific form and parameter values).13 Typically, there is a focus on model parameter uncertainties but uncertainties about which model to use and about the bias or discrepancy of a particular model (sometimes called model form error) often dominate parameter uncertainty. This whitepaper addresses the challenges of performing uncertainty quantification (UQ) for expensive computational models.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A posteriori error estimators for the discrete ordinates approximation of the one-speed neutron transport equation

When calculating numerical solutions of the neutron transport equation it is important to have a measure of the accuracy of the solution. As the true solution is generally not known, a suitable estimation of the error must be made. The steady state transport equation possesses discretization errors in all its independent variables: angle, energy and space. In this work only spatial discretization errors are considered. An exact transport solution, in which the degree of regularity of the exact flux across the singular characteristic is controlled, is manufactured to determine the numerical solutions true discretization error. This solution is then projected onto a Legendre polynomial space in order to form an exact solution on the same basis space as the numerical solution, Discontinuous Galerkin Finite Element Method (DGFEM), to enable computation of the true error. Over a series of test problems the true error is compared to the error estimated by: Ragusa and Wang (RW), residual source (LER) and cell discontinuity estimators (JD). The validity and accuracy of the considered estimators are primarily assessed by considering the effectivity index and global L2 norm of the error. In general RW excels at approximating the true error distribution but usually under-estimates its magnitude; the LER estimator emulates the true error distribution but frequently over-estimates the magnitude of the true error; the JD estimator poorly captures the true error distribution and generally under-estimates the error about singular characteristics but over-estimates it elsewhere. (authors)

97 MATHEMATICS AND COMPUTING↗

Anisotropic mesh adaptation for solution of finite element problems using hierarchical edge-based error estimates

We present a new technology for generating meshes minimizing the interpolation and discretization errors or their gradients. The key element of this methodology is construction of a space metric from edge-based error estimates. For a mesh with N{sub h} triangles, the error is proportional to N{sub h}{sup -1} and the gradient of error is proportional to N{sub h}{sup -1/2} which are optimal asymptotics. The methodology is verified with numerical experiments.

99 GENERAL AND MISCELLANEOUS↗

A note on the reliability of goal-oriented error estimates for Galerkin finite element methods with nonlinear functionals

Here, we consider estimating the discretization error in a nonlinear functional J (u) in the setting of an abstract variational problem: find u ϵ $\mathscr{V}$ such that B (u, φ) = L (φ) ∀φ ϵ $\mathscr{V}$, as approximated by a Galerkin finite element method. Here, $\mathscr{V}$ is a Hilbert space, B (. , .) is a bilinear form, and L (∙) is a linear functional. We consider well-known error estimates η of the form J (u) - J (u h ) ≈ η = L (z) - B (u h , z), where u h denotes a finite element approximation to u, and z denotes the solution to an auxiliary adjoint variational problem. We show that there exist nonlinear functionals for which error estimates of this form are not reliable, even in the presence of an exact adjoint solution z. An estimate η is said to be reliable if there exists a constant C ϵ $\mathbb{R}$ >0 independent of u h such that |J (u) - J (u h )| ≤ C|η|. We present several example pairs of bilinear forms and nonlinear functionals where reliability of η is not achieved.

A posteriori↗

Neural networks for large eddy simulations of wall-bounded turbulence: numerical experiments and challenges

Here, we examine the application of neural network-based methods to improve the accuracy of large eddy simulations of incompressible turbulent flows. The networks are trained to learn a mapping between flow features and the subgrid scales, and applied locally and instantaneously—in the same way as traditional physics-based subgrid closures. Models that use only the local resolved strain rate are poorly correlated with the actual subgrid forces obtained from filtering direct numerical simulation data. We see that highly accurate models in a priori testing are inaccurate in forward calculations, owing to the preponderance of numerical errors in implicitly filtered large eddy simulations. A network that accounts for the discretization errors is trained and found to be unstable in a posteriori testing. We identify a number of challenges that the approach faces, including a distribution shift that affects networks that fail to account for numerical errors.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Fast Mix Table Construction for Material Discretization

An effective hybrid Monte Carlo--deterministic implementation typically requires the approximation of a continuous geometry description with a discretized piecewise-constant material field. The inherent geometry discretization error can be reduced somewhat by using material mixing, where multiple materials inside a discrete mesh voxel are homogenized. Material mixing requires the construction of a ``mix table,'' which stores the volume fractions in every mixture so that multiple voxels with similar compositions can reference the same mixture. Mix table construction is a potentially expensive serial operation for large problems with many materials and voxels. We formulate an efficient algorithm to construct a sparse mix table in $O(\text{number of voxels}\times \log \text{number of mixtures})$ time. The new algorithm is implemented in ADVANTG and used to discretize continuous geometries onto a structured Cartesian grid. When applied to an end-of-life MCNP model of the High Flux Isotope Reactor with 270 distinct materials, the new method improves the material mixing time by a factor of 100 compared to a naive mix table implementation.

Johnson, Seth R↗

Fast mix table construction for material discretization

An effective hybrid Monte Carlo-deterministic implementation typically requires the approximation of a continuous geometry description with a discretized piecewise-constant material field. The inherent geometry discretization error can be reduced somewhat by using material mixing, where multiple materials inside a discrete mesh voxel are homogenized. Material mixing requires the construction of a 'mix table,' which stores the volume fractions in every mixture so that multiple voxels with similar compositions can reference the same mixture. Mix table construction is a potentially expensive serial operation for large problems with many materials and voxels. We formulate an efficient algorithm to construct a sparse mix table in O(number of voxels x log number of mixtures) time. The new algorithm is implemented in ADVANTG and used to discretize continuous geometries onto a structured Cartesian grid. When applied to an end-of-life MCNP model of the High Flux Isotope Reactor with 270 distinct materials, the new method improves the material mixing time by a factor of 100 compared to a naive mix table implementation. (authors)

97 MATHEMATICS AND COMPUTING↗

Limited‐memory adaptive snapshot selection for proper orthogonal decomposition

Summary Reduced order models are useful for accelerating simulations in many‐query contexts, such as optimization, uncertainty quantification, and sensitivity analysis. However, offline training of reduced order models (ROMs) can have prohibitively expensive memory and floating‐point operation costs in high‐performance computing applications, where memory per core is limited. To overcome this limitation for proper orthogonal decomposition, we propose a novel adaptive selection method for snapshots in time that limits offline training costs by selecting snapshots according an error control mechanism similar to that found in adaptive time‐stepping ordinary differential equation solvers. The error estimator used in this work is related to theory bounding the approximation error in time of proper orthogonal decomposition‐based ROMs, and memory usage is minimized by computing the singular value decomposition using a single‐pass incremental algorithm. Results for a viscous Burgers' test problem demonstrate convergence in the limit as the algorithm error tolerances go to zero; in this limit, the full‐order model is recovered to within discretization error. A parallel version of the resulting method can be used on supercomputers to generate proper orthogonal decomposition‐based ROMs, or as a subroutine within hyperreduction algorithms that require taking snapshots in time, or within greedy algorithms for sampling parameter space. Copyright © 2016 John Wiley & Sons, Ltd.

Oxberry, Geoffrey M.↗

Limited-memory adaptive snapshot selection for proper orthogonal decomposition

Reduced order models are useful for accelerating simulations in many-query contexts, such as optimization, uncertainty quantification, and sensitivity analysis. However, offline training of reduced order models can have prohibitively expensive memory and floating-point operation costs in high-performance computing applications, where memory per core is limited. To overcome this limitation for proper orthogonal decomposition, we propose a novel adaptive selection method for snapshots in time that limits offline training costs by selecting snapshots according an error control mechanism similar to that found in adaptive time-stepping ordinary differential equation solvers. The error estimator used in this work is related to theory bounding the approximation error in time of proper orthogonal decomposition-based reduced order models, and memory usage is minimized by computing the singular value decomposition using a single-pass incremental algorithm. Results for a viscous Burgers’ test problem demonstrate convergence in the limit as the algorithm error tolerances go to zero; in this limit, the full order model is recovered to within discretization error. The resulting method can be used on supercomputers to generate proper orthogonal decomposition-based reduced order models, or as a subroutine within hyperreduction algorithms that require taking snapshots in time, or within greedy algorithms for sampling parameter space.

97 MATHEMATICS AND COMPUTING↗

Improved methods for Feynman path integral calculations and their application to calculate converged vibrational–rotational partition functions, free energies, enthalpies, entropies, and heat capacities for methane

We present an improved version of our “path-by-path” enhanced same path extrapolation scheme for Feynman path integral (FPI) calculations that permits rapid convergence with discretization errors ranging from O(P{sup −6}) to O(P{sup −12}), where P is the number of path discretization points. We also present two extensions of our importance sampling and stratified sampling schemes for calculating vibrational–rotational partition functions by the FPI method. The first is the use of importance functions for dihedral angles between sets of generalized Jacobi coordinate vectors. The second is an extension of our stratification scheme to allow some strata to be defined based only on coordinate information while other strata are defined based on both the geometry and the energy of the centroid of the Feynman path. These enhanced methods are applied to calculate converged partition functions by FPI methods, and these results are compared to ones obtained earlier by vibrational configuration interaction (VCI) calculations, both calculations being for the Jordan–Gilbert potential energy surface. The earlier VCI calculations are found to agree well (within ∼1.5%) with the new benchmarks. The FPI partition functions presented here are estimated to be converged to within a 2σ statistical uncertainty of between 0.04% and 0.07% for the given potential energy surface for temperatures in the range 300–3000 K and are the most accurately converged partition functions for a given potential energy surface for any molecule with five or more atoms. We also tabulate free energies, enthalpies, entropies, and heat capacities.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Blow-off momentum from melt and vapor in nuclear deflection scenarios

For Earth-impacting objects that are large in size or have short warning times nuclear explosives are an effective threat mitigation response. Nuclear-based deflection works by means of conservation of momentum: as material is heated by incoming photons and neutrons it is ejected from the body which imparts momentum to the remaining mass of the asteroid. Predicting the complete response of a particular object is difficult, since the ejecta size and velocity distributions rely heavily on the unknown, complicated internal structure of the body. However, lower bounds on the blow-off momentum can be estimated using the melted and vaporized surface material. In this paper, we model the response of a one-dimensional SiO2 surface to monoenergetic soft X-ray, hard X-ray and neutron sources using Arbitrary Lagrangian–Eulerian radiation/hydrodynamic simulations. Errors in the blow-off momentum due to our hydrodynamic mesh resolution are quantified and inform zone sizing that balances numerical discretization error with computational efficiency. We explore deposited energy densities ranging from 1.1 to 200 times the melt energy density for SiO2, and develop an approximate relation that gives the mesh resolution needed for a desired percent error in the blow-off momentum as a function of deposited energy density and melt depth. Using these mesh constraints, the response of our one-dimensional SiO2 surface to the energy sources is simulated, and lower bounds are placed on the melt/vapor blow-off momentum as a function of deposited energy density and source energy type.

79 ASTRONOMY AND ASTROPHYSICS↗

A Block-Structured Adaptive Mesh Framework to Solve Radiation Transfer Equation in Irregular Embedded Geometries

Radiation transport arises in various scientific, industrial, and medical fields, and understanding its effect in applications is needed to make accurate predictions, safety assessments and performance optimizations. Solving the Radiation Transport Equation (RTE) is challenging due to its integro-differential nature, which involves both differential and integral terms. The differential term describes the change in radiation intensity due to absorption and emission, while the integral term accounts for scattering. The accurate modeling of radiation is further complicated in many applications due to the complex, irregular geometries. Various methods exist for solving the RTE, including the zonal, Monte Carlo, spherical harmonics, discrete ordinates, and finite volume methods. Traditional mesh-based approaches, which rely on structured or unstructured meshes, struggle with irregular geometries due to: a) the difficulty of conforming structured grids to irregular domains, b) challenges in enforcing boundary conditions correctly, and c) the additional computational cost of unstructured mesh methods. This work presents a second-order accurate method for solving the RTE in irregular geometries. The radiation intensity is discretized using the finite-volume method in both spatial and angular directions on regular Cartesian grid blocks. Leveraging the block-structured adaptive mesh refinement (AMR) framework provided by AMReX, our method refines the grid locally to reduce spatial discretization error, ensuring a converged numerical solution while minimizing computational costs elsewhere. A two-stage deferred correction approach is employed: First, a first-order discretization on grid blocks is solved using an algebraic multigrid method in HYPRE. Second, a correction term is applied explicitly to achieve second-order accuracy. The correction term is calculated by approximating the radiation flux on cell faces using a Total Variation Diminishing (TVD) scheme. This approach ensures quick convergence of the multigrid method while preserving higher-order accuracy of the numerical solution. Irregular geometries are resolved as embedded boundaries (EB), resulting in both cut cells and regular cells. In cut cells, we modify the fluxes using face fractions and incorporate additional contributions from EB boundary conditions. To ensure higher-order convergence near the EB interface, the correction term is modified by interpolating the radiation intensity to fictitious ghost points. The implementation takes advantage of modern supercomputers by leveraging AMReX’sMPI/X parallelization strategy where X can be MPI or a GPU accelerator including CUDA, HIP and DPC++. We validate our solver using classical test cases, both with and without EB, demonstrating accuracy and efficiency. Additionally, we analyze the impact of adaptive mesh refinement on solution accuracy and computational cost, highlighting the advantages of our approach for high-resolution radiation transport simulations.

computational fluid dynamics (CFD)↗

Uncollided Flux Implementation for Discrete Ordinates Radiation Transport Solutions in Rattlesnake

Simulation of neutron streaming over long distances in optically thin narrow pipes sur- rounded by optically thicker materials that can absorb and/or scatter neutrons presents a simulation challenge to most neutron transport methods. Without sophisticated variance reduction techniques, this problem can be intractable for Monte Carlo techniques. Deterministic methods are often limited due to large angular discretization errors, this phenomenon is commonly known as ray effects. The uncollided component of the angular flux solution is the most anisotropic part and it can be difficult for discrete ordinate methods to accurately represent. The number of discrete angles required to overcome these ray effects is prohibitively large for the hodoscope configuration. Hence, other approaches must be taken to better simulate neutron streaming within the hodoscope collimator slots. In this report, we present an algorithm for semi-analytical calculation of the uncollided flux.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Robust verification of stochastic simulation codes

We introduce a robust verification tool for computational codes, which we call Stochastic Robust Extrapolation based Error Quantification (StREEQ). Unlike the prevalent Grid Convergence Index (GCI) [1] method, our approach is suitable for both stochastic and deterministic computational codes and is generalizable to any number of discretization variables. Building on ideas introduced in the Robust Verification [2] approach, we estimate the converged solution and orders of convergence with uncertainty using multiple fits of a discretization error model. In contrast to Robust Verification, we perform these fits to many bootstrap samples yielding a larger set of predictions with smoother statistics. Here, bootstrap resampling is performed on the lack-of-fit errors for deterministic code responses, and directly on the noisy data set for stochastic responses. This approach lends a degree of robustness to the overall results, capable of yielding precise verification results for sufficiently resolved data sets, and appropriately expanding the uncertainty when the data set does not support a precise result. For stochastic responses, a credibility assessment is also performed to give the analyst an indication of the trustworthiness of the results. Furthermore, this approach is suitable for both code and solution verification, and is particularly useful for solution verification of high-consequence simulations..

97 MATHEMATICS AND COMPUTING↗

Overlap valence on 2+1 flavor domain wall fermion configurations with deflation and low-mode substitution

The overlap fermion propagator is calculated on 2+1 flavor domain-wall fermion gauge configurations on 16{sup 3}x32, 24{sup 3}x64 and 32{sup 3}x64 lattices. With hyper-cubic (HYP) smearing and low eigenmode deflation, it is shown that the inversion of the overlap operator can be expedited by {approx}20 times for the 16{sup 3}x32 lattice and {approx}80 times for the 32{sup 3}x64 lattice. The overhead cost for calculating eigenmodes ranges from 4.5 to 7.9 propagators for the above lattices. Through the study of hyperfine splitting, we found that the O(m{sup 2}a{sup 2}) error is small and these dynamical fermion lattices can adequately accommodate quark mass up to the charm quark. A preliminary calculation of the low-energy constant {Delta}{sub mix} which characterizes the discretization error of the pion made up of a pair of sea and valence quarks in this mixed-action approach is carried out via the scalar correlator with periodic and antiperiodic boundary conditions. It is found to be small which shifts a 300 MeV pion mass by {approx}10 to 19 MeV on these sets of lattices. We have studied the signal-to-noise issue of the noise source for the meson and baryon. We introduce a new algorithm with Z{sub 3} grid source and low eigenmode substitution to study the many-to-all meson and baryon correlators. It is found to be efficient in reducing errors for the correlators of both mesons and baryons. With 64-point Z{sub 3} grid source and low-mode substitution, it can reduce the statistical errors of the light quark (m{sub {pi}{approx}2}00-300 MeV) meson and nucleon correlators by a factor of {approx}3-4 as compared to the point source. The Z{sub 3} grid source itself can reduce the errors of the charmonium correlators by a factor of {approx}3.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Consistent Feature Extraction From Vector Fields: Combinatorial Representations and Analysis Under Local Reference Frames

This dissertation presents research on addressing some of the contemporary challenges in the analysis of vector fields—an important type of scientific data useful for representing a multitude of physical phenomena, such as wind flow and ocean currents. In particular, new theories and computational frameworks to enable consistent feature extraction from vector fields are presented. One of the most fundamental challenges in the analysis of vector fields is that their features are defined with respect to reference frames. Unfortunately, there is no single “correct” reference frame for analysis, and an unsuitable frame may cause features of interest to remain undetected, thus creating serious physical consequences. This work develops new reference frames that enable extraction of localized features that other techniques and frames fail to detect. As a result, these reference frames objectify the notion of “correctness” of features for certain goals by revealing the phenomena of importance from the underlying data. An important consequence of using these local frames is that the analysis of unsteady (time-varying) vector fields can be reduced to the analysis of sequences of steady (timeindependent) vector fields, which can be performed using simpler and scalable techniques that allow better data management by accessing the data on a per-time-step basis. Nevertheless, the state-of-the-art analysis of steady vector fields is not robust, as most techniques are numerical in nature. The residing numerical errors can violate consistency with the underlying theory by breaching important fundamental laws, which may lead to serious physical consequences. This dissertation considers consistency as the most fundamental characteristic of computational analysis that must always be preserved, and presents a new discrete theory that uses combinatorial representations and algorithms to provide consistency guarantees during vector field analysis along with the uncertainty visualization of unavoidable discretization errors. Together, the two main contributions of this dissertation address two important concerns regarding feature extraction from scientific data: correctness and precision. The work presented here also opens new avenues for further research by exploring more-general reference frames and more-sophisticated domain discretizations.

97 MATHEMATICS AND COMPUTING↗

Quantitative assessment of eddy viscosity rans models for turbulent mixed convection in a differentially heated plane channel

Turbulent mixed convection between two vertical, infinite parallel plates at different temperatures is studied using various two-equation turbulence models. The numerical simulations are performed at a turbulent Reynolds number of Re τ = 150 and a Grashof number of Gr = 9.6 × 10 5 . Comparisons are made against the highly trusted DNS results. Consistent with the DNS approach, the current simulations are performed using constant properties and the Boussinesq approximation to predict the influence of buoyancy. Previous studies have provided assessments of two-equation turbulence models for various scenarios, but often rely on a qualitative “eye” test in order to determine the most appropriate model to predict a given flow. This study aims to provide a new form of quantitative assessment that accounts for both the physics captured by the turbulence model as well as the magnitude of the system response quantities (SRQ) using a modified symmetric mean absolute percent error (SMAPE) method. This method is designed to be approachable to researchers at any level and can be applied to system response quantities from multiple research fields. Uncertainty quantification is also performed to determine the discretization error for each turbulence model. Recommendations are made as to which turbulence models best capture the physics – hydrodynamically and thermally – using both local and global validation metrics. Lastly, a sensitivity analysis is performed on the damping functions used in the most accurate models. This underpins the potential of model developments and adjustments most worth pursuing for buoyant flows. Finally, this framework provides a more physics-based comparative analysis of the selected turbulence models.

42 ENGINEERING↗