Search NASA⌕ Search

SEARCH · Search NASA

Results for “numerical 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 1,621 records · Page 90

Simulated Half-Precision Implementation of Blocked QR Factorization and Graph Clustering Applications

We explored half-precision implementation of blocked QR algorithms with the following motivations: 1. New GPUs perform fast half-precision arithmetic (4 to 16 times as fast as doubleprecision). 2. QR factorization is a basic linear algebra tool useful for many physics and data analysis applications. 3. Communication-avoiding, parallelizable QR algorithms already exist for talland- skinny matrices. While the standard QR algorithms are highly unstable in half-precision, our numerical simulations show that the Tall-and-Skinny QR (TSQR) algorithm can improve the backward error of QR factorization. When using subspace iteration for graph clustering applications, half-precision accuracy in forming the eigenspace is sufficient for clustering with high precision and recall for some medium-scale benchmark problems. Note that all half-precision arithmetic were simulated with conversions to single-precision floats. Therefore, the results from this work are somewhat optimistic.

97 MATHEMATICS AND COMPUTING↗

Preliminary Theoretical Analysis of Mixed Precision Krylov Subspace Methods (Q3 Report)

The third quarter of the project was spent performing theoretical finite precision analysis of Krylov subspace method variants that use mixed precision. Our focus here is on the Conjugate Gradient (CG) method and the Lanczos method. We have performed an analysis of maximum attainable accuracy for the classical CG method in which 3 precisions are used: a working precision ε, a precision ε IP for the inner product computations, and a precision ε MV for the matrix-vector products. Our results show that performing inner product computations in lower precision does not affect the attainable accuracy. Further, we have performed a complete error analysis of the s-step Lanczos algorithm. In this case, we show that the numerical behavior of the algorithm can be significantly improved by using extra precision in a small part of the computation. We summarize the main theorems in the remainder of the document. Other activities include attending biweekly xSDK meetings. The subsequent quarter will be spent finalizing these results into technical reports and/or manuscripts for submission to journals, as well as identifying opportunities for future work.

97 MATHEMATICS AND COMPUTING↗

Near-Efficient and Non-Asymptotic Multiway Inference

We establish non-asymptotic efficiency guarantees for tensor decomposition–based inference in count data models. Under a Poisson framework, we consider two related goals: (i) parametric inference , the estimation of the full distributional parameter tensor, and (ii) multiway analysis , the recovery of its canonical polyadic (CP) decomposition factors. Our main result shows that in the rank-one setting, a rank-constrained maximum-likelihood estimator achieves multiway analysis with variance matching the Cramér–Rao Lower Bound (CRLB) up to absolute constants and logarithmic factors. This provides a general framework for studying “near-efficient” multiway estimators in finite-sample settings. For higher ranks, we illustrate that our multiway estimator may not attain the CRLB; nevertheless, CP-based parametric inference remains nearly minimax optimal, with error bounds that improve on prior work by offering more favorable dependence on the CP rank. Numerical experiments corroborate near-efficiency in the rank-one case and highlight the efficiency gap in higher-rank scenarios.

97 MATHEMATICS AND COMPUTING↗

Scaling Turbulent Combustion Fields in Explosions

We considered the topic of explosions from spherical high-explosive (HE) charges. We studied how the turbulent combustion fields scale. On the basis of theories of dimensional analysis by Bridgman and similarity theories of Sedov and Barenblatt, we found that all fields scaled with the explosion length scale r0. This included the blast wave, the mean and root mean squared (RMS) profiles of thermodynamic variables, combustion variables, velocities, vorticity, and turbulent Reynolds stresses. This was a consequence of the formulation of the problem and our numerical method, which both satisfied the similarity conditions of Sedov. We performed numerical simulations of 1 g charges and 1 kg charges; the solutions were identical (within roundoff error) when plotted in scaled variables. We also explored scaling laws related to three-phase pyrotechnic explosions. We show that although the scaling formally broke down, the fireball still essentially scaled with the explosion length scale r0. However, the discrete Lagrange particles (DLP) (phase 2) and the heterogeneous continuum model (HCM) of the DLP wakes (phase 3) did not scale with r0, and mean and RMS profiles could differ by a factor of 10 in some regions. This was because the DLP particles and wakes introduced an additional scale that broke the similarity conditions.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Optimum quantisers for a Gaussian input probability density and for the magnitude-error distortion measure

The parameters of non-uniform and uniform quantizers up to ten bits of quantization, optimum for a Gaussian input probability and for the magnitude-error distortion criterion are computed. Optimum quantizers must be understood as quantizers with minimum distortion. The numerical method used for the optimization converges relatively rapidly. The comparison between optimum non-uniform quantizers and optimum uniform quantizers is made.

Demaret, J. C.↗

A mathematical analysis of the theory of interplanetary scintillation in the weak scattering approximation

A simplified analytical technique is presented for modeling the interplanetary scintillation of radio sources of finite angular size with a power-law electron-density-fluctuation power spectrum. The simplification results from the representation of the scintillation spectrum in confluent hypergeometric functions. The approximations presented allow fast numerical evaluation of a spectrum for a weakly scattering but extended medium with less than 10% error over the entire spectrum. Parameters describing anisotropic electron irregularities as well as anisotropic source structure are included, and the dependence of the spectrum normalization on the scales of the medium is derived explicitly. The parametric description of the domains of convergence of the approximate expansions also provides a simple conceptualization of the relative contributions of the scattered radiation along the line of sight to the observed spectrum. This is particularly useful for sources of finite angular size. This technique is applied to previously published observations.

Mitchell, D. G.↗

Assessing the quality of curvilinear coordinate meshes by decomposing the Jacobian matrix

An algebraic decomposition of the Jacobian matrix which relates physical and computational variables is presented. This invertible decomposition parameterizes the mesh by the physically intuitive qualities of cell orientation, cell orthogonality, cell volume, and cell aspect ratio. The decomposition can be used to analyze numerically generated curvilinear coordinate meshes and to assess the contribution of the mesh to the truncation error for any specific differential operator and algorithm. This is worked out in detail for Laplace's equation in nonconservative and conservative forms. The analysis is applied to the solution of the full potential code TAIR, showing grid plots, carpet plots, and truncation error for a NACA 0012 airfoil.

Kerlick, G. D.↗

Approximate penetration factors for nuclear reactions of astrophysical interest

The ranges of validity of approximations of P(l), the penetration factor which appears in the parameterization of nuclear-reaction cross sections at low energies and is employed in the extrapolation of laboratory data to even lower energies of astrophysical interest, are investigated analytically. Consideration is given to the WKB approximation, P(l) at the energy of the total barrier, approximations derived from the asymptotic expansion of G(l) for large eta, approximations for small values of the parameter x, applications of P(l) to nuclear reactions, and the dependence of P(l) on channel radius. Numerical results are presented in tables and graphs, and parameter ranges where the danger of serious errors is high are identified.

Humblet, J.↗

Real-Time Keratometer

Optical/electronic keratometer system produces contour map of corneal surface illuminated by infrared light pulses. Prevents operator error and eliminates need to apply fluorescent liquids to cornea. Keratometer provides both video display and numerical recording for corneal-transplant and radial-keratotomy surgery and for contact-lens fitting. Optical system of keratometer, circular grating pattern transmitted to corneal surface reflected onto identical circular grating on surface of fiber plate, producing Moire pattern. Data-analysis-and-display mode, laser light pulsed, and Moire pattern converted to digital information in frame grabber. Digital picture information sent to computer for processing and display.

Frazer, Robert E.↗

Total Variation Diminishing (TVD) schemes of uniform accuracy

Explicit second-order accurate finite-difference schemes for the approximation of hyperbolic conservation laws are presented. These schemes are nonlinear even for the constant coefficient case. They are based on first-order upwind schemes. Their accuracy is enhanced by locally replacing the first-order one-sided differences with either second-order one-sided differences or central differences or a blend thereof. The appropriate local difference stencils are selected such that they give TVD schemes of uniform second-order accuracy in the scalar, or linear systems, case. Like conventional TVD schemes, the new schemes avoid a Gibbs phenomenon at discontinuities of the solution, but they do not switch back to first-order accuracy, in the sense of truncation error, at extrema of the solution. The performance of the new schemes is demonstrated in several numerical tests.

Hartwich, PETER-M.↗

Turbulent separated flow past a backward-facing step: A critical evaluation of two-equation turbulence models

The ability of two-equation models to accurately predict separated flows is analyzed from a combined theoretical and computational standpoint. Turbulent flow past a backward facing step is chosen as a test case in an effort to resolve the variety of conflicting results that were published during the past decade concerning the performance of two-equation models. It is found that the errors in the reported predictions of the k-epsilon model have two major origins: (1) numerical problems arising from inadequate resolution, and (2) inaccurate predictions for normal Reynolds stress differences arising from the use of an isotropic eddy viscosity. Inadequacies in near wall modelling play a substantially smaller role. Detailed calculations are presented which strongly indicate the standard k-epsilon model - when modified with an independently calibrated anisotropic eddy viscosity - can yield surprisingly good predictions for the backstep problem.

Thangam, S.↗

HZETRN: A heavy ion/nucleon transport code for space radiations

The galactic heavy ion transport code (GCRTRN) and the nucleon transport code (BRYNTRN) are integrated into a code package (HZETRN). The code package is computer efficient and capable of operating in an engineering design environment for manned deep space mission studies. The nuclear data set used by the code is discussed including current limitations. Although the heavy ion nuclear cross sections are assumed constant, the nucleon-nuclear cross sections of BRYNTRN with full energy dependence are used. The relation of the final code to the Boltzmann equation is discussed in the context of simplifying assumptions. Error generation and propagation is discussed, and comparison is made with simplified analytic solutions to test numerical accuracy of the final results. A brief discussion of biological issues and their impact on fundamental developments in shielding technology is given.

Wilson, John W.↗

Turbulent flow past a backward-facing step - A critical evaluation of two-equation models

The ability of two-equation models to accurately predict separated flows is analyzed from a combined theoretical and computational standpoint. Turbulent flow past a backward facing step is chosen as a test case in an effort to resolve the variety of conflicting results that were published during the past decade concerning the performance of two-equation models. It is found that the errors in the reported predictions of the k-epsilon model have two major origins: (1) numerical problems arising from inadequate resolution, and (2) inaccurate predictions for normal Reynolds stress differences arising from the use of an isotropic eddy viscosity. Inadequacies in near wall modelling play a substantially smaller role. Detailed calculations are presented which strongly indicate the standard k-epsilon model - when modified with an independently calibrated anisotropic eddy viscosity - can yield surprisingly good predictions for the backstep problem.

Thangam, S.↗

Airborne gravimetry, altimetry, and GPS navigation errors

Proper interpretation of airborne gravimetry and altimetry requires good knowledge of aircraft trajectory. Recent advances in precise navigation with differential GPS have made it possible to measure gravity from the air with accuracies of a few milligals, and to obtain altimeter profiles of terrain or sea surface correct to one decimeter. These developments are opening otherwise inaccessible regions to detailed geophysical mapping. Navigation with GPS presents some problems that grow worse with increasing distance from a fixed receiver: the effect of errors in tropospheric refraction correction, GPS ephemerides, and the coordinates of the fixed receivers. Ionospheric refraction and orbit error complicate ambiguity resolution. Optimal navigation should treat all error sources as unknowns, together with the instantaneous vehicle position. To do so, fast and reliable numerical techniques are needed: efficient and stable Kalman filter-smoother algorithms, together with data compression and, sometimes, the use of simplified dynamics.

Colombo, Oscar L.↗

High-spatial-resolution passive microwave sounding systems

The principal contributions of this combined theoretical and experimental effort were to advance and demonstrate new and more accurate techniques for sounding atmospheric temperature, humidity, and precipitation profiles at millimeter wavelengths, and to improve the scientific basis for such soundings. Some of these techniques are being incorporated in both research and operational systems. Specific results include: (1) development of the MIT Microwave Temperature Sounder (MTS), a 118-GHz eight-channel imaging spectrometer plus a switched-frequency spectrometer near 53 GHz, for use on the NASA ER-2 high-altitude aircraft, (2) conduct of ER-2 MTS missions in multiple seasons and locations in combination with other instruments, mapping with unprecedented approximately 2-km lateral resolution atmospheric temperature and precipitation profiles, atmospheric transmittances (at both zenith and nadir), frontal systems, and hurricanes, (3) ground based 118-GHz 3-D spectral images of wavelike structure within clouds passing overhead, (4) development and analysis of approaches to ground- and space-based 5-mm wavelength sounding of the upper stratosphere and mesosphere, which supported the planning of improvements to operational weather satellites, (5) development of improved multidimensional and adaptive retrieval methods for atmospheric temperature and humidity profiles, (6) development of combined nonlinear and statistical retrieval techniques for 183-GHz humidity profile retrievals, (7) development of nonlinear statistical retrieval techniques for precipitation cell-top altitudes, and (8) numerical analyses of the impact of remote sensing data on the accuracy of numerical weather predictions; a 68-km gridded model was used to study the spectral properties of error growth.

Staelin, D. H.↗

Numerical Analysis of a Radiant Heat Flux Calibration System

A radiant heat flux gage calibration system exists in the Flight Loads Laboratory at NASA's Dryden Flight Research Center. This calibration system must be well understood if the heat flux gages calibrated in it are to provide useful data during radiant heating ground tests or flight tests of high speed aerospace vehicles. A part of the calibration system characterization process is to develop a numerical model of the flat plate heater element and heat flux gage, which will help identify errors due to convection, heater element erosion, and other factors. A 2-dimensional mathematical model of the gage-plate system has been developed to simulate the combined problem involving convection, radiation and mass loss by chemical reaction. A fourth order finite difference scheme is used to solve the steady state governing equations and determine the temperature distribution in the gage and plate, incident heat flux on the gage face, and flat plate erosion. Initial gage heat flux predictions from the model are found to be within 17% of experimental results.

Jiang, Shanjuan↗

Re-Computation of Numerical Results Contained in NACA Report No. 496

An extensive examination of NACA Report No. 496 (NACA 496), "General Theory of Aerodynamic Instability and the Mechanism of Flutter," by Theodore Theodorsen, is described. The examination included checking equations and solution methods and re‐computing interim quantities and all numerical examples in NACA 496. The checks revealed that NACA 496 contains computational shortcuts (time‐ and effort‐saving devices for engineers of the time) and clever artifices (employed in its solution methods), but, unfortunately, also contains numerous tripping points (aspects of NACA 496 that have the potential to cause confusion) and some errors. The re‐computations were performed employing the methods and procedures described in NACA 496, but using modern computational tools. With some exceptions, the magnitudes and trends of the original results were in fair‐to‐very‐good agreement with the re‐computed results. The exceptions included what are speculated to be computational errors in the original in some instances and transcription errors in the original in others. Independent flutter calculations were performed and, in all cases, including those where the original and re‐computed results differed significantly, were in excellent agreement with the re‐computed results. Appendix A contains NACA 496; Appendix B contains a Matlab(Reistered) program that performs the re‐computation of results; Appendix C presents three alternate solution methods, with examples, for the two‐degree‐of-freedom solution method of NACA 496; Appendix D contains the three‐degree‐of‐freedom solution method (outlined in NACA 496 but never implemented), with examples.

Perry, Boyd, III↗