Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Approximation 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 379 records · Page 21
All sourcesNASA NTRSDOE OSTI
DOE OSTI2016

The arbitrary order mixed mimetic finite difference method for the diffusion equation

Here, we propose an arbitrary-order accurate mimetic finite difference (MFD) method for the approximation of diffusion problems in mixed form on unstructured polygonal and polyhedral meshes. As usual in the mimetic numerical technology, the method satisfies local consistency and stability conditions, which determines the accuracy and the well-posedness of the resulting approximation. The method also requires the definition of a high-order discrete divergence operator that is the discrete analog of the divergence operator and is acting on the degrees of freedom. The new family of mimetic methods is proved theoretically to be convergent and optimal error estimates for flux and scalar variable are derived from the convergence analysis. A numerical experiment confirms the high-order accuracy of the method in solving diffusion problems with variable diffusion tensor. It is worth mentioning that the approximation of the scalar variable presents a superconvergence effect.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2018

Light Scattering by Small Multilayer Particles: A Generalized Separation of Variables Method

A Rayleigh approximation is constructed for light scattering by small multilayer axisymmetric particles, in which their polarizability is determined by the generalized separation of variables method (SVM). In this method, scalar potentials, the gradients of which yield the electric-field strengths, are represented as expansions in spherical harmonics of the Laplace equation. Unknown coefficients of expansions are determined from the boundary conditions, which are reduced to infinite systems of linear algebraic equations (ISLAEs), since the separation of variables is incomplete. The T matrix of the electrostatic problem, principal element T{sub 11} of which is proportional to the particle polarizability, is determined. The necessary condition for the ISLAEs solvability for the SVM coincides with the condition of the correct application of the extended boundary conditions method (EBCM). However, numerical calculations in which finite-dimensional (i.e., reduced) systems are solved, yield different results in ranges of variation of parameters that are close to the boundary of the range of applicability. An analysis of the numerical calculations of the scattering and absorption cross sections for two-layer confocal spheroids, an exact solution for which can be obtained using spheroidal harmonics, shows that the SVM is preferable to the EBCM. It turned out that the proposed method yields workable results in a wider range of variation of parameters. Even outside the range of applicability, in which it should be regarded as a certain approximate solution, its use in a number of cases is quite acceptable. Additional calculations for three-layer nonconfocal spheroids, as well as for three-layer similar pseudospheroids and Pascal’s snails, which can be obtained from spheroids as a result of the inversion with respect to the coordinate origin and one of the foci, respectively, confirm these inferences. We note that, for certain values of the parameters, the shapes of the latter particles are nonconvex.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2015

Ballistic-diffusive approximation for the thermal dynamics of metallic nanoparticles in nanocomposite materials

Based on ballistic-diffusive approximation, a method is presented to model heat transfer in nanocomposites containing metal nanoparticles. This method provides analytical expression for the temperature dynamics of metallic nanoparticles embedded in a dielectric medium. In this study, nanoparticles are considered as spherical shells, so that Boltzmann equation is solved using ballistic-diffusive approximation to calculate the electron and lattice thermal dynamics in gold nanoparticles, while thermal exchange between the particles is taken into account. The model was used to investigate the influence of particle size and metal concentration of the medium on the electron and lattice thermal dynamics. It is shown that these two parameters are crucial in determining the nanocomposite thermal behavior. Our results showed that the heat transfer rate from nanoparticles to the matrix decreases as the nanoparticle size increases. On the other hand, increasing the metal concentration of the medium can also decrease the heat transfer rate.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2019

Synthesis and Characterization of Tris-chelate Complexes for Understanding f -Orbital Bonding in Later Actinides

An isostructural family of f-element compounds (Ce, Nd, Sm, Gd; Am, Bk, Cf) of the redox-active dioxophenoxazine ligand (DOPOq; DOPO = 2,4,6,8-tetra-tert-butyl-1-oxo-1H-phenoxazin-9-olate) was prepared. This family, of the form M(DOPOq)3, represents the first nonaqueous isostructural series, including the later actinides berkelium and californium. The lanthanide derivatives were fully characterized using 1H NMR spectroscopy and SQUID magnetometry, while all species were structurally characterized by X-ray crystallography and electronic absorption spectroscopy. In order to probe the electronic structure of this new family, CASSCF calculations were performed and revealed these systems to be largely ionic in contrast to previous studies, where berkelium and californium typically have a small degree of covalent character. To validate the zeroth order regular approximation (ZORA) method, the same CASSCF analysis using experimental structures versus UDFT-ZORA optimized structures does not exhibit sizable changes in bonding patterns. This shows that UDFT-ZORA combined with CASSCF could be a useful first approximation to predict and investigate the structure and electronic properties of actinides and lanthanides that are difficult to synthesize or characterize.

Chemistry↗
DOE OSTI2025

Backpropagation-based learning with local derivative approximation and memory replay in biologically plausible neural systems

When learning, the brain modifies individual synaptic connections to reach a desired behavior. Animal and human brains have been shown to be incredibly capable of learning complex and varied functions across a wide variety of tasks. In recent years, artificial neural networks, inspired by human and animal brains, have shown great capabilities in learning a wide variety of difficult tasks. However, artificial neural networks primarily teach themselves through the use of backpropagation, a learning method which has no clear analogue within the brain. Additionally, Artificial Neural Networks primarily use continuous activation functions, which differ significantly from the spiking neuronal behavior present in the brain. In this paper, we discuss and demonstrate a biologically plausible learning method that approximates backpropagation through two techniques on Spiking Neural Networks. First, we show that the local temporal derivatives that are necessary for backpropagation can be approximately recovered through reconstruction using spike timings. Second, we show that through learning during a sleep phase, inspired by neuroscience research into memory replay, the localized parallel feedback path can learn to approximate the derivative through the forward path weight matrix, thus solving the weight transport problem. Lastly, we demonstrate that the combination of these two methods can approach or exceed the accuracy of backpropagation-based methods for a variety of neuromorphic vision tasks while maintaining biological plausibility.

42 ENGINEERING↗
DOE OSTI2025

Compactly‐Supported Nonstationary Kernels for Computing Exact Gaussian Processes on Big Data

The Gaussian process (GP) is a widely used method for analyzing large-scale data sets, including spatio-temporal measurements of nonlinear processes that are now commonplace in the environmental sciences. Traditional implementations of GPs involve stationary kernels (also termed covariance functions) that limit their flexibility, and exact methods for inference that prevent application to data sets with more than about 10,000 points. Modern approaches to address stationarity assumptions generally fail to accommodate large data sets, while all attempts to address scalability focus on approximating the Gaussian likelihood, which can involve subjectivity and lead to inaccuracies. In this work, we explicitly derive an alternative kernel that can discover and encode both sparsity and nonstationarity. We embed the kernel within a fully Bayesian GP model and leverage high-performance computing resources to enable the analysis of massive data sets. We demonstrate the favorable performance of our novel kernel relative to existing exact and approximate GP methods across a variety of synthetic data examples. Furthermore, we conduct space–time prediction based on more than 1 million measurements of daily maximum temperature and verify that our results outperform state-of-the-art methods in the Earth sciences. More broadly, having access to exact GPs that use ultra-scalable, sparsity-discovering, nonstationary kernels allows GP methods to truly compete with a wide variety of machine learning methods.

Gaussian processes↗
DOE OSTI2022

A new method for inferring city emissions and lifetimes of nitrogen oxides from high-resolution nitrogen dioxide observations: a model study

We present a new method to infer emissions and lifetimes of nitrogen oxides (NO x ) based on tropospheric nitrogen dioxide (NO 2 ) observations together with reanalysis wind fields for cities located in polluted backgrounds. Since the accuracy of the method is difficult to assess due to lack of “true values” that can be used as a benchmark, we apply the method to synthetic NO 2 observations derived from the NASA-Unified Weather Research and Forecasting (NU-WRF) model at a high horizontal spatial resolution of 4 km × 4 km for cities over the continental United States. We compare the inferred emissions and lifetimes with the values given by the NU-WRF model to evaluate the method. The method is applicable to 26 US cities. The derived results are generally in good agreement with the values given by the model, with the relative differences of 2 % ± 17 % (mean ± standard deviation) and 15 % ± 25 % for lifetimes and emissions, respectively. Our investigation suggests that the use of wind data prior to the satellite overpass time improves the performance of the method. The correlation coefficients between inferred and NU-WRF lifetimes increase from 0.56 to 0.79 and for emissions increase from 0.88 to 0.96 when comparing results based on wind fields sampled simultaneously with satellite observations and averaged over 9 h data prior to satellite observations, respectively. We estimate that uncertainties in NO x lifetime and emissions arising from the method are approximately 15 % and 20 %, respectively, for typical (US) cities. The total uncertainties reach up to 43% (lifetimes) and 45% (emissions) by considering the additional uncertainties associated with satellite NO 2 observations and wind data. We expect this new method to be applicable to NO 2 observations from the TROPOspheric Monitoring Instrument (TROPOMI) and geostationary satellites, such as Geostationary Environment Monitoring Spectrometer (GEMS) or the Tropospheric Emissions: Monitoring Pollution (TEMPO) instrument, to estimate urban NO x emissions and lifetimes globally.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Recent Developments in DFTB+, a Software Package for Efficient Atomistic Quantum Mechanical Simulations

DFTB+ is a flexible, open-source software package developed by its community, designed for fast and efficient atomistic quantum mechanical simulations. It employs various methods that approximate density functional theory (DFT), such as density functional-based tight binding (DFTB) and the extended tight binding (xTB) approach allowing simulations of large systems over extended time scales with reasonable accuracy, while being significantly faster than traditional ab initio methods. In recent years, several new extensions of the DFTB method have been developed and implemented in the DFTB+ program package in order to improve the accuracy and generality of the available simulation results. In this paper, we review those enhancements, show several use case examples and discuss the strengths and limitations of its features.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Second-order spectral line shift comparisons

The second-order spectral line width formulae from the projection operator and kinetic theory methods were recently compared. It was shown that a systematic expansion of the projection operator width expression including initial correlations formally agrees with the second-order kinetic theory result. It is now shown that the second-order dynamic shifts are also formally the same. The static shifts, however, differ due to an ad hoc treatment of electron-electron correlations in the projection operator method. The approximation is necessary in order to screen the radiator-electron interactions. The differences, however, are expected to be small. Finally, the results suggest using the rigorous and more compact second-order width and shift expressions from the kinetic theory method as the starting point for spectral line shape calculations. At line center, however, the projection operator second-order expression for the width and shift simplifies and reduces to the kinetic theory result.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2015

Preserving Lagrangian Structure in Nonlinear Model Reduction with Application to Structural Dynamics

Our work proposes a model-reduction methodology that preserves Lagrangian structure and achieves computational efficiency in the presence of high-order nonlinearities and arbitrary parameter dependence. As such, the resulting reduced-order model retains key properties such as energy conservation and symplectic time-evolution maps. We focus on parameterized simple mechanical systems subjected to Rayleigh damping and external forces, and consider an application to nonlinear structural dynamics. To preserve structure, the method first approximates the system's “Lagrangian ingredients''---the Riemannian metric, the potential-energy function, the dissipation function, and the external force---and subsequently derives reduced-order equations of motion by applying the (forced) Euler--Lagrange equation with these quantities. Moreover, from the algebraic perspective, key contributions include two efficient techniques for approximating parameterized reduced matrices while preserving symmetry and positive definiteness: matrix gappy proper orthogonal decomposition and reduced-basis sparsification. Our results for a parameterized truss-structure problem demonstrate the practical importance of preserving Lagrangian structure and illustrate the proposed method's merits: it reduces computation time while maintaining high accuracy and stability, in contrast to existing nonlinear model-reduction techniques that do not preserve structure.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2022

Loschmidt-echo approach to error estimation in Krylov-subspace approximation

The Krylov subspace method is a traditional approach to approximate quantum evolution, allowing us to treat systems with large Hilbert spaces. Despite its popularity, current bounds typically overestimate the error, which translates into more expensive simulation routines. Here, in this paper, we tackle this problem by realizing that the error can be understood as a Loschmidt echo in a one-dimensional (1D) noninteracting tight-binding Hamiltonian. We show that the different time regimes of the approximation can be understood using simple physical ideas. More importantly, we obtain computationally cheap error bounds that describe with high precision the actual error in the approximation.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

Active learning approach to simulations of strongly correlated matter with the ghost Gutzwiller approximation

Quantum embedding (QE) methods such as the ghost Gutzwiller approximation (gGA) offer a powerful approach to simulating strongly correlated systems, but come with the computational bottleneck of computing the ground state of an auxiliary embedding Hamiltonian (EH) iteratively. In this work, we introduce an active learning (AL) framework integrated within the gGA to address this challenge. The methodology is applied to the single-band Hubbard model and results in a significant reduction in the number of instances where the EH must be solved. Through a principal component analysis (PCA), we find that the EH parameters form a low-dimensional structure that is largely independent of the geometric specifics of the systems, especially in the strongly correlated regime. Our AL strategy enables us to discover this low-dimensionality structure on the fly, while leveraging it for reducing the computational cost of gGA, laying the groundwork for more efficient simulations of complex strongly correlated materials. Published by the American Physical Society 2024

36 MATERIALS SCIENCE↗
DOE OSTI2016

Regularized deconvolution method for turbulent combustion modeling

The modeling of filtered chemical source terms in large-eddy simulation (LES) of turbulent reacting flows remains a challenge. Deconvolution methods are an attractive technique for representing these unclosed terms. With this technique, filtered scalars are reconstructed through deconvolution. The chemical source terms that are computed directly from the deconvolved scalars are filtered explicitly to represent the turbulence-chemistry interaction. However, the approximate deconvolution method (ADM), frequently employed for non-reacting flows, exhibits shortcomings for reacting scalars. This is because ADM does not ensure essential conservation and boundedness conditions. So to address this issue, we propose a regularized deconvolution method (RDM) based on an optimization procedure. We conduct a priori and a posteriori analyses to examine RDM as a closure in LES. These investigations are performed in the context of explicit filtering. By showing that RDM is accurate and stable with respect to both the filter width and time, we conclude that the new deconvolution method shows promise in application to combustion LES.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2019

Non-Intrusive Inference Reduced Order Model for Fluids Using Deep Multistep Neural Network

In this effort we propose a data-driven learning framework for reduced order modeling of fluid dynamics. Designing accurate and efficient reduced order models for nonlinear fluid dynamic problems is challenging for many practical engineering applications. Classical projection-based model reduction methods generate reduced systems by projecting full-order differential operators into low-dimensional subspaces. However, these techniques usually lead to severe instabilities in the presence of highly nonlinear dynamics, which dramatically deteriorates the accuracy of the reduced-order models. In contrast, our new framework exploits linear multistep networks, based on implicit Adams–Moulton schemes, to construct the reduced system. The advantage is that the method optimally approximates the full order model in the low-dimensional space with a given supervised learning task. Moreover, our approach is non-intrusive, such that it can be applied to other complex nonlinear dynamical systems with sophisticated legacy codes. We demonstrate the performance of our method through the numerical simulation of a two-dimensional flow past a circular cylinder with Reynolds number Re = 100. The results reveal that the new data-driven model is significantly more accurate than standard projection-based approaches.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2019

Learning to predict the cosmological structure formation

Matter evolved under the influence of gravity from minuscule density fluctuations. Nonperturbative structure formed hierarchically over all scales and developed non-Gaussian features in the Universe, known as the cosmic web. To fully understand the structure formation of the Universe is one of the holy grails of modern astrophysics. Astrophysicists survey large volumes of the Universe and use a large ensemble of computer simulations to compare with the observed data to extract the full information of our own Universe. However, to evolve billions of particles over billions of years, even with the simplest physics, is a daunting task. We build a deep neural network, the Deep Density Displacement Model ( D 3 M ), which learns from a set of prerun numerical simulations, to predict the nonlinear large-scale structure of the Universe with the Zel’dovich Approximation (ZA), an analytical approximation based on perturbation theory, as the input. Our extensive analysis demonstrates that D 3 M outperforms the second-order perturbation theory (2LPT), the commonly used fast-approximate simulation method, in predicting cosmic structure in the nonlinear regime. We also show that D 3 M is able to accurately extrapolate far beyond its training data and predict structure formation for significantly different cosmological parameters. Our study proves that deep learning is a practical and accurate alternative to approximate 3D simulations of the gravitational structure formation of the Universe.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2014

Swiss-cheese models and the Dyer-Roeder approximation

In view of interpreting the cosmological observations precisely, especially when they involve narrow light beams, it is crucial to understand how light propagates in our statistically homogeneous, clumpy, Universe. Among the various approaches to tackle this issue, Swiss-cheese models propose an inhomogeneous spacetime geometry which is an exact solution of Einstein's equation, while the Dyer-Roeder approximation deals with inhomogeneity in an effective way. In this article, we demonstrate that the distance-redshift relation of a certain class of Swiss-cheese models is the same as the one predicted by the Dyer-Roeder approach, at a well-controlled level of approximation. Both methods are therefore equivalent when applied to the interpretation of, e.g., supernova obervations. The proof relies on completely analytical arguments, and is illustrated by numerical results.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2016

SU-G-TeP4-07: Automatic EPID-Based 2D Measurement of MLC Leaf Offset as a Quality Control Tool

Purpose: The MLC dosimetric leaf gap (DLG) and transmission are measured parameters which impact the dosimetric accuracy of IMRT and VMAT plans. This investigation aims to develop an efficient and accurate routine constancy check of the physical DLG in two dimensions. Methods: The manufacturer’s recommended DLG measurement method was modified by using 5 fields instead of 11 and by utilizing the Electronic Portal Imaging Device (EPID). Validations were accomplished using an ion chamber (IC) in solid water and a 2D IC array. EPID data was collected for 6 months on multiple TrueBeam linacs using both Millennium and HD MLCs at 5 different clinics in an international consortium. Matlab code was written to automatically analyze the images and calculate the 2D results. Sensitivity was investigated by introducing deliberate leaf position errors. MLC calibration and initialization history was recorded to allow quantification of their impact. Results were analyzed using statistical process control (SPC). Results: The EPID method took approximately 5 minutes. Due to detector response, the EPID measured DLG and transmission differed from the IC values but were reproducible and consistent with changes measured using the ICs. For the Millennium MLC, the EPID measured DLG and transmission were both consistently lower than IC results. The EPID method was implemented as leaf offset and transmission constancy tests (LOC and TC). Based on 6 months of measurements, the initial leaf-specific action thresholds for changes from baseline were set to 0.1 mm. Upper and lower control limits for variation were developed for each machine. Conclusion: Leaf offset and transmission constancy tests were implemented on Varian HD and Millennium MLCs using an EPID and found to be efficient and accurate. The test is effective for monitoring MLC performance using dynamic delivery and performing process control on the DLG in 2D, thus enhancing dosimetric accuracy. This work was supported by a grant from Varian Medical Systems.

61 RADIATION PROTECTION AND DOSIMETRY↗
DOE OSTI2020

Automation of Active Space Selection for Multireference Methods via Machine Learning on Chemical Bond Dissociation

Predicting and understanding the chemical bond is one of the major challenges of computational quantum chemistry. Kohn–Sham density functional theory (KS-DFT) is the most common method, but approximate density functionals may not be able to describe systems where multiple electronic configurations are equally important. Multiconfigurational wave functions, on the other hand, can provide a detailed understanding of the electronic structures and chemical bonds of such systems. In the complete active space self-consistent field (CASSCF) method, one performs a full configuration interaction calculation in an active space consisting of active electrons and active orbitals. However, CASSCF and its variants require the selection of these active spaces. This choice is not black box; it requires significant experience and testing by the user, and thus active space methods are not considered particularly user-friendly and are employed only by a minority of quantum chemists. Our goal is to popularize these methods by making it easier to make good active space choices. Here, we present a machine learning protocol that performs an automated selection of active spaces for chemical bond dissociation calculations of main group diatomic molecules. The protocol shows high prediction performance for a given target system as long as a properly correlated system is chosen for training. Good active spaces are correctly predicted with a considerably better success rate than random guess (larger than 80% precision for most systems studied). Our automated machine learning protocol shows that a “black-box” mode is possible for facilitating and accelerating the large-scale calculations on multireference systems where single-reference methods such as KS-DFT cannot be applied.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗