SEARCH · Search NASA

Results for “Computer Algebra”

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.

SourceResearch connectionFrom yearThrough year

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

At least 955 records · Page 53

All sources NASA NTRS DOE OSTI

DOE OSTI2018

Efficient Surrogate Model Development: Impact of Sample Size and Underlying Model Dimensions

This study compares eight surrogate-model construction approaches using computational experiments. Here, the construction approaches considered include: Artificial Neural Networks (ANNs), Automated Learning of Algebraic Models using Optimization (ALAMO), Radial Basis Networks (RBNs), Extreme Learning Machines (ELMs), Gaussian Progress Regression (GPR), Random Forests (RFs), Support Vector Regression (SVR), and Multivariate Adaptive Regression Splines. Each approach is used to construct surrogate models for predicting the outputs of thirty-four test functions. The data used to construct surrogate models were generated using Latin Hypercube Sampling (LHS), Halton and Sobol sampling methods. We studied the impact of sampling method, sample size, and challenge function characteristics on the accuracy of surrogate model predictions. Our results revealed that for smaller sample sizes, on average, surrogate models trained using the data points generated by Sobol sequences provided the best estimation for all surrogate model construction approaches. As the sample size increased, the impact of sampling methods diminished. The comparison of surrogate-modeling construction approaches at large sample sizes revealed that surrogate models trained using ANN, ALAMO, and ELM yielded smaller root mean squared error and higher adjusted R-squared values than the models trained using the rest of the approaches.

42 ENGINEERING↗

DOE OSTI2020

A conservative, consistent, and scalable meshfree mimetic method

Mimetic methods discretize divergence by restricting the Gauss theorem to mesh cells. Because point clouds lack such geometric entities, construction of a compatible meshfree divergence remains a challenge. In this work, we define an abstract Meshfree Mimetic Divergence (MMD) operator on point clouds by contraction of field and virtual face moments. This MMD satisfies a discrete divergence theorem, provides a discrete local conservation principle, and is first-order accurate. We consider two MMD instantiations. The first one assumes a background mesh and uses generalized moving least squares (GMLS) to obtain the necessary field and face moments. This MMD instance is appropriate for settings where a mesh is available but its quality is insufficient for a robust and accurate mesh-based discretization. The second MMD operator retains the GMLS field moments but defines virtual face moments using computationally efficient weighted graph-Laplacian equations. This MMD instance does not require a background grid and is appropriate for applications where mesh generation creates a computational bottleneck. It allows one to trade an expensive mesh generation problem for a scalable algebraic one, without sacrificing compatibility with the divergence operator. We demonstrate the approach by using the MMD operator to obtain a virtual finite-volume discretization of conservation laws on point clouds. Finally, numerical results in the paper confirm the mimetic properties of the method and show that it behaves similarly to standard finite volume methods.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2017

Compressed Continuous Computation v. 12/20/2016

A library for performing numerical computation with low-rank functions. The (C3) library enables performing continuous linear and multilinear algebra with multidimensional functions. Common tasks include taking "matrix" decompositions of vector- or matrix-valued functions, approximating multidimensional functions in low-rank format, adding or multiplying functions together, integrating multidimensional functions.

Gorodetsky, Alex↗

NASA NTRS1982

Pole placement and order reduction in two-time-scale control systems through Riccati iteration

A transformation of variables taken from singular perturbations may be applied to two-time-scale linear systems in state space form to reduce the system to block-diagonal form with slow and fast modes decoupled. The transformation is easily computed by applying the new Riccati iteration. The iteration yields a solution to the nonsymmetric algebraic Riccati equation obtained by partitioning the original system matrix A. The numerical procedure is initiated with the trivial iterate L(0) = 0, and is globally convergent to the desired unique time scale decoupling solution. After transformation, the decoupled system may be used in controller design to achieve exact closed loop pole placement in the slow subsystem without altering the poles of the fast subsystem. The decoupled form may also be used to reduce system order by wetting a small parameter to zero. Provided the fast subsystem is stable, the order reduction can be expected to yield a good approximation to the original system. These methods are demonstrated using the 16th order linear model of a turbofan engine.

Anderson, L. R.↗

NASA NTRS1989

A parametric study of three-dimensional separation at a wing/body junction for supersonic free-stream conditions

The problem of three-dimensional separation at a wing/body junction has been investigated numerically using a three-dimensional Navier-Stokes code which employs the MacCormack's time split finite volume technique. An algebraic grid generation technique is used for generating the grid at a wing/body junction. Specific computational results on velocity and pressure distribution in the separated flow region are compared with the experimental results. A parametric study of flow parameters such as Mach number and Reynolds number have been carried out to understand their effect in interaction flow field. The parametric study indicates a strong dependency of the number of vortices at the junction on Mach number and Reynolds number.

Lakshmanan, B.↗

NASA NTRS1991

An algebraic RNG-based turbulence model for three-dimensional turbomachinery flows

An algebraic eddy viscosity turbulence model based on Renormalization Group (RNG) theory for complex three-dimensional turbomachinery flows is presented. Modifications are made to the baseline RNG model for wakes and separated flows. The model has several advantages over popular algebraic models most notably its lack of empirically determined coefficients. The model is used to compute the mean flow in a low speed axial compressor rotor. The agreement with blade boundary layer and radial flow experimental data is very good and shows improvement over the Baldwin-Lomax model. The development of the tip leakage vortex is also well predicted. The computed wake decay also compares favorably with recent experimental data.

Kirtley, K. R.↗

NASA NTRS1999

Exploration of Uncertainty in Glacier Modelling

There are procedures and methods for verification of coding algebra and for validations of models and calculations that are in use in the aerospace computational fluid dynamics (CFD) community. These methods would be efficacious if used by the glacier dynamics modelling community. This paper is a presentation of some of those methods, and how they might be applied to uncertainty management supporting code verification and model validation for glacier dynamics. The similarities and differences between their use in CFD analysis and the proposed application of these methods to glacier modelling are discussed. After establishing sources of uncertainty and methods for code verification, the paper looks at a representative sampling of verification and validation efforts that are underway in the glacier modelling community, and establishes a context for these within overall solution quality assessment. Finally, an information architecture and interactive interface is introduced and advocated. This Integrated Cryospheric Exploration (ICE) Environment is proposed for exploring and managing sources of uncertainty in glacier modelling codes and methods, and for supporting scientific numerical exploration and verification. The details and functionality of this Environment are described based on modifications of a system already developed for CFD modelling and analysis.

Thompson, David E.↗

NASA NTRS2005

A Unifying Multibody Dynamics Algorithm Development Workbench

The development of new and efficient algorithms for multibody dynamics has been an important research area. These algorithms are used for modeling, simulation, and control of systems such as spacecraft, robotic systems, automotive applications, the human body, manufacturing operations, and micro-electromechanical systems (MEMS). At JPL's Dynamics and Real Time Simulation (DARTS) Laboratory we have developed software that serves as a computational workbench for these algorithms. This software utilizes the mathematical perspective of the spatial operator algebra, which allows the development of dynamics algorithms and new insights into multibody dynamics.

Ziegler, John L.↗

NASA NTRS1995

Numerical Simulations of High Enthalpy Pulse Facilities

Axisymmetric flows within shock tubes and expansion tubes are simulated including the effects of finite rate chemistry and both laminar and turbulent boundary layers. The simulations demonstrate the usefulness of computational fluid dynamics for characterizing the flows in high enthalpy pulse facilities. The modeling and numerical requirements necessary to simulate these flows accurately are also discussed. Although there is a large body of analysis which explains and quantifies the boundary layer growth between the shock and the interface in a shock tube, there is a need for more detailed solutions. Phenomena such as thermochemical nonequilibrium. or turbulent transition behind the shock are excluded in the assumptions of Mirels' analysis. Additionally there is inadequate capability to predict the influence of the boundary layer on the expanded gas behind the interface. Quantifying the gas in this region is particularly important in expansion tubes because it is the location of the test gas. Unsteady simulations of the viscous flow in shock tubes are computationally expensive because they must follow features such as a shock wave over the length of the facility and simultaneously resolve the small length scales within the boundary layer. As a result, efficient numerical algorithms are required. The numerical approach of the present work is to solve the axisymmetric gas dynamic equations using an finite-volume formulation where the inviscid fluxes are computed with a upwind TVD scheme. Multiple species equations are included in the formulation so that finite-rate chemistry can be modeled. The simulations cluster grid points at the shock and interface and translate this clustered grid with these features to minimize numerical errors. The solutions are advanced at a CFL number of less than one based on the inviscid gas dynamics. To avoid limitations on the time step due to the viscous terms, these terms are treated implicitly. This requires a block tri-diagonal matrix inversion along each line of cells normal to the wall. The cost of this inversion is more than offset by the larger allowable time step. The source terms representing the finite-rate chemical kinetics are also treated implicitly. An algebraic turbulence model for compressible flow is used. The flow in a low pressure shock tube is computed and the results are compared with Mirels'analysis. The driven gas is nitrogen at 70 Pa, and the incident shock speed is approximately 2.9 km/sec so that there is little dissociation. The simulations include a laminar boundary layer and are run until the limiting flow regime is achieved. At this limit, the shock and interface travel at the same velocity because the amount of driven gas between these two features remains the same: the mass flow across the shock is equal to the mass of gas being entrained at the interface by the boundary layer. Simulations with several grids are presented to establish the grid independence of the solution, Good agreement is achieved between Mirels' correlations and the computations. This is expected since the flow conditions are chosen to be consistent with the assumptions used in Mirels' analysis. This comparison adds credibility to the numerical approach and highlights some of the differences between the theory and the detailed simulations. In addition, simulations of the HYPULSE expansion tube are presented for two operating conditions and the computations are compared to experimental data. The operating gas for both cases is nitrogen. One test condition is at a total enthalpy of 15.2 MJ/Kg and a relatively low pressure of 2 kPa. This case is characterized by a laminar boundary layer and significant chemical nonequilibrium. in the acceleration gas. The second test condition is at a total enthalpy of 10.2 MJ/Kg and a pressure of 38 kPa and is characterized by a turbulent boundary layer. The simulations compare well with experiment and reveal that the nonuniformity in pressure observed during the test time is related to variations in the boundary layer displacement thickness.

Wilson, Gregory J.↗

NASA NTRS1992

Workshop on Engineering Turbulence Modeling

Discussed here is the future direction of various levels of engineering turbulence modeling related to computational fluid dynamics (CFD) computations for propulsion. For each level of computation, there are a few turbulence models which represent the state-of-the-art for that level. However, it is important to know their capabilities as well as their deficiencies in order to help engineers select and implement the appropriate models in their real world engineering calculations. This will also help turbulence modelers perceive the future directions for improving turbulence models. The focus is on one-point closure models (i.e., from algebraic models to higher order moment closure schemes and partial differential equation methods) which can be applied to CFD computations. However, other schemes helpful in developing one-point closure models, are also discussed.

Povinelli, Louis A.↗

DOE OSTI2019

Algorithms and algebraic solutions of decay chain differential equations for stable and unstable nuclide fractionation

Radioactive decay processes, such as alpha decay, produce decay chains where the mass numbers of nuclides decrease as larger nuclides expel energetic particles to form smaller nuclides. Under these conditions, the coefficient matrix that describes the differential rate expressions for radioactive decay can be made lower triangular. With this special structure, formulating an algebraic solution to the decay chains can be done by first formulating the eigenvectors that make up the coefficient matrix, which can then be solved using forward substitution for a lower triangular matrix. This work details the derivation of algebraic solutions for decay chains of any number of stable and unstable nuclides with any number of branching based on this eigenvector analysis. A prototype computational code was developed to validate and compare this methodology against a number of other methods for solving similar systems. A two-phase sorting algorithm yielding the lower triangular matrix structure was established to apply the developed algebraic solutions for decay chains involving beta-emitting radionuclides transformed into daughter nuclides without change in their mass number. In conclusion, the methodologies produced in this work provide an efficient way to estimate nuclide fractions from natural decay processes.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2017

Extreme-Scale Algorithms & Software Resilience (EASIR) Architecture-Aware Algorithms for Scalable Performance and Resilience on Heterogeneous Architectures

This project addresses both communication-avoiding algorithms, and reproducible floating-point computation. Communication, i.e. moving data, either between levels of memory or processors over a network, is much more expensive per operation than arithmetic (measured in time or energy), so we seek algorithms that greatly reduce communication. We developed many new algorithms for both dense and sparse, and both direct and iterative linear algebra, attaining new communication lower bounds, and getting large speedups in many cases. We also extended this work in several ways: (1) We minimize writes separately from reads, since writes may be much more expensive than reads on emerging memory technologies, like Flash, sometimes doing asymptotically fewer writes than reads. (2) We extend the lower bounds and optimal algorithms to arbitrary algorithms that may be expressed as perfectly nested loops accessing arrays, where the array subscripts may be arbitrary affine functions of the loop indices (eg A(i), B(i,j+k, k+3*m-7, …) etc.). (3) We extend our communication-avoiding approach to some machine learning algorithms, such as support vector machines. This work has won a number of awards. We also address reproducible floating-point computation. We define reproducibility to mean getting bitwise identical results from multiple runs of the same program, perhaps with different hardware resources or other changes that should ideally not change the answer. Many users depend on reproducibility for debugging or correctness. However, dynamic scheduling of parallel computing resources, combined with nonassociativity of floating point addition, makes attaining reproducibility a challenge even for simple operations like summing a vector of numbers, or more complicated operations like the Basic Linear Algebra Subprograms (BLAS). We describe an algorithm that computes a reproducible sum of floating point numbers, independent of the order of summation. The algorithm depends only on a subset of the IEEE Floating Point Standard 754-2008, uses just 6 words to represent a “reproducible accumulator,” and requires just one read-only pass over the data, or one reduction in parallel. New instructions based on this work are being considered for inclusion in the future IEEE 754-2018 floating-point standard, and new reproducible BLAS are being considered for the next version of the BLAS standard.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2017

Fast synchronization-free algorithms for parallel sparse triangular solves with multiple right-hand sides

The sparse triangular solve kernels, SpTRSV and SpTRSM, are important building blocks for a number of numerical linear algebra routines. Parallelizing SpTRSV and SpTRSM on today's many-core platforms, such as GPUs, is not an easy task since computing a component of the solution may depend on previously computed components, enforcing a degree of sequential processing. As a consequence, most existing work introduces a preprocessing stage to partition the components into a group of level-sets or colour-sets so that components within a set are independent and can be processed simultaneously during the subsequent solution stage. However, this class of methods requires a long preprocessing time as well as significant runtime synchronization over-heads between the sets. To address this, we

HPC↗

NASA NTRS1994

Direct numerical simulations and modeling of a spatially-evolving turbulent wake

Understanding of turbulent free shear flows (wakes, jets, and mixing layers) is important, not only for scientific interest, but also because of their appearance in numerous practical applications. Turbulent wakes, in particular, have recently received increased attention by researchers at NASA Langley. The turbulent wake generated by a two-dimensional airfoil has been selected as the test-case for detailed high-resolution particle image velocimetry (PIV) experiments. This same wake has also been chosen to enhance NASA's turbulence modeling efforts. Over the past year, the author has completed several wake computations, while visiting NASA through the 1993 and 1994 ASEE summer programs, and also while on sabbatical leave during the 1993-94 academic year. These calculations have included two-equation (K-omega and K-epsilon) models, algebraic stress models (ASM), full Reynolds stress closure models, and direct numerical simulations (DNS). Recently, there has been mutually beneficial collaboration of the experimental and computational efforts. In fact, these projects have been chosen for joint presentation at the NASA Turbulence Peer Review, scheduled for September 1994. DNS calculations are presently underway for a turbulent wake at Re(sub theta) = 1000 and at a Mach number of 0.20. (Theta is the momentum thickness, which remains constant in the wake of a two dimensional body.) These calculations utilize a compressible DNS code written by M. M. Rai of NASA Ames, and modified for the wake by J. Cimbala. The code employs fifth-order accurate upwind-biased finite differencing for the convective terms, fourth-order accurate central differencing for the viscous terms, and an iterative-implicit time-integration scheme. The computational domain for these calculations starts at x/theta = 10, and extends to x/theta = 610. Fully developed turbulent wake profiles, obtained from experimental data from several wake generators, are supplied at the computational inlet, along with appropriate noise. After some adjustment period, the flow downstream of the inlet develops into a fully three-dimensional turbulent wake. Of particular interest in the present study is the far wake spreading rate and the self-similar mean and turbulence profiles. At the time of this writing, grid resolution studies are underway, and a code is being written to calculate turbulence statistics from these wake calculations; the statistics will be compared to those from the ongoing PIV wake measurements, those of previous experiments, and those predicted by the various turbulence models. These calculations will lead to significant long-term benefits for the turbulence modeling effort. In particular, quantities such as the pressure-strain correlation and the dissipation rate tensor can be easily calculated from the DNS results, whereas these quantities are nearly impossible to measure experimentally. Improvements to existing turbulence models (and development of new models) require knowledge about flow quantities such as these. Present turbulence models do a very good job at prediction of the shape of the mean velocity and Reynolds stress profiles in a turbulent wake, but significantly underpredict the magnitude of the stresses and the spreading rate of the wake. Thus, the turbulent wake is an ideal flow for turbulence modeling research. By careful comparison and analysis of each term in the modeled Reynolds stress equations, the DNS data can show where deficiencies in the models exist; improvements to the models can then be attempted.

Cimbala, John M.↗

DOE OSTI2021

Electromagnetic Transient Simulation Algorithms for Evaluation of Large-Scale Extreme Fast Charging Systems (Distribution Grid Models)

The distribution and transmission grids are observing an increased penetration of power electronics in loads and generations. For example, there is increasing interest in integrating in extreme fast charging (XFC) systems for fast charging of electrical vehicles. As these systems are integrated, developing high-fidelity electromagnetic transient model of XFC systems in distribution grids and evaluating their interactions with the power grid would be of significant interest. This model will be utilized for design of XFC systems, to identify upgrades in distribution and/or transmission grids, for planning purposes by transmission planners or operators or owners, among others. It can also be utilized in operations for improved reliable performance of the grid and/or XFC station. The challenge with simulating these models is the high computational complexity introduced by the large number of states present in the system and the time-step needed to simulate the system. In this paper, advanced simulations algorithms are applied to reduce the computational complexity of simulating large-scale XFC systems. The algorithms include numerical stiffness-based segregation, time constant-based segregation, clustering and aggregation on differential algebraic equations (DAEs), and multi-order integration approaches. While the first three algorithms split the matrix that needs to be inverted from a large matrix to much smaller matrices, the final algorithm reduces the computational burden of applying higher-order integration approaches in the complete system. The comparison made in the previous sentence is with respect to use of homogeneous integration approaches used in conventional electromagnetic transient simulators like power systems computer aided design (PSCAD). The approaches mentioned here have resulted in speed-up of 36x in the simulation of a single distribution system with 15 XFCs.

Debnath, Suman↗

DOE OSTI2020

Approximate Inverse Chain Preconditioner: Iteration Count Case Study for Spectral Support Solvers

As the growing availability of computational power slows, there has been an increasing reliance on algorithmic advances. However, faster algorithms alone will not necessarily bridge the gap in allowing computational scientists to study problems at the edge of scientific discovery in the next several decades. Often, it is necessary to simplify or precondition solvers to accelerate the study of large systems of linear equations commonly seen in a number of scientific fields. Preconditioning a problem to increase efficiency is often seen as the best approach; yet, preconditioners which are fast, smart, and efficient do not always exist. Following the progress of [1], we present a new preconditioner for symmetric diagonally dominant (SDD) systems of linear equations. These systems are common in certain PDEs, network science, and supervised learning among others. Based on spectral support graph theory, this new preconditioner builds off of the work of [2], computing and applying a V-cycle chain of approximate inverse matrices. This preconditioner approach is both algebraic in nature as well as hierarchically-constrained depending on the condition number of the system to be solved. Due to its generation of an Approximate Inverse Chain of matrices, we refer to this as the AIC preconditioner. We further accelerate the AIC preconditioner by utilizing precomputations to simplify setup and multiplications in the con-text of an iterative Krylov-subspace solver. While these iterative solvers can greatly reduce solution time, the number of iterations can grow large quickly in the absence of good preconditioners. Initial results for the AIC preconditioner have shown a very large reduction in iteration counts for SDD systems as compared to standard preconditioners such as Incomplete Cholesky (ICC) and Multigrid (MG). We further show significant reduction in iteration counts against the more advanced Combinatorial Multigrid (CMG) preconditioner. We have further developed no-fill sparsification techniques to ensure that the computational cost of applying the AIC preconditioner does not grow prohibitively large as the depth of the V-cycle grows for systems with larger condition numbers. Our numerical results have shown that these sparsifiers maintain the sparsity structure of our system while also displaying significant reductions in iteration counts.1 2

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2020

Approximate Inverse Chain Preconditioner: Iteration Count Case Study for Spectral Support Solvers

97 MATHEMATICS AND COMPUTING↗

NASA NTRS1985

Comparison of Two Algebraic Methods for Curve/curve Intersection

Most geometric modeling systems use either polynomial or rational functions to represent geometry. In such systems most computational problems can be formulated as systems of polynomials in one or more variables. Classical elimination theory can be used to solve such systems. Here Cayley's method of elimination is summarized and it is shown how it can best be used to solve the curve/curve intersection problem. Cayley's method was found to be a more straightforward approach. Furthermore, it is computationally simpler, since the elements of the Cayley matrix are one variable instead of two variable polynomials. Researchers implemented and tested both methods and found Cayley's to be more efficient. Six pairs of curves, representing mixtures of lines, circles, and cubic arcs were used. Several examples had multiple intersection points. For all six cases Cayley's required less CPU time than the other method. The average time ratio of method 1 to method 2 was 3.13:1, the least difference was 2.33:1, and the most dramatic was 6.25:1. Both of the above methods can be extended to solve the surface/surface intersection problem.

Demontaudouin, Y.↗