Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “nonlinear map”

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 145 records · Page 8

Fed-DeepONet: Stochastic Gradient-Based Federated Training of Deep Operator Networks

The Deep Operator Network (DeepONet) framework is a different class of neural network architecture that one trains to learn nonlinear operators, i.e., mappings between infinite-dimensional spaces. Traditionally, DeepONets are trained using a centralized strategy that requires transferring the training data to a centralized location. Such a strategy, however, limits our ability to secure data privacy or use high-performance distributed/parallel computing platforms. To alleviate such limitations, in this paper, we study the federated training of DeepONets for the first time. That is, we develop a framework, which we refer to as Fed-DeepONet, that allows multiple clients to train DeepONets collaboratively under the coordination of a centralized server. To achieve Fed-DeepONets, we propose an efficient stochastic gradient-based algorithm that enables the distributed optimization of the DeepONet parameters by averaging first-order estimates of the DeepONet loss gradient. Then, to accelerate the training convergence of Fed-DeepONets, we propose a moment-enhanced (i.e., adaptive) stochastic gradient-based strategy. Finally, we verify the performance of Fed-DeepONet by learning, for different configurations of the number of clients and fractions of available clients, (i) the solution operator of a gravity pendulum and (ii) the dynamic response of a parametric library of pendulums.

Moya, Christian↗

Fine Sun Sensor Field of View Calibration

The fine Sun sensor (FSS) used on many spacecraft consists of two independent single-axis sensors, nominally mounted perpendicularly, that detect Sun angle across a typical field of view of +/- 32 degrees. The nonlinear function that maps the measured counts into an observed angle is called the transfer function. The FSS transfer function provided by the manufacturer consists of nine parameters for each of the two sensitive axes. An improved transfer function has been previously reported that achieves a significant accuracy improvement across the entire field of view. This new function expands the parameter set to 12 coefficients per axis and includes cross terms combining counts from both axes. To make best use of the FSS for spacecraft attitude determination, it must be calibrated after launch. We are interested in simplifying the postlaunch calibration procedure for estimating improvements to the 24 parameters in the transfer function. This paper discusses how to recombine the terms of the transfer function to reduce their redundancy without decreasing its accuracy and then presents an attitude dependent procedure for estimating the parameters. The end result is a calibration algorithm that is easier to use and does not sacrifice accuracy. Results of calibration using on-orbit data are presented.

Sedlak, Joseph E.↗

Neural Networks as Surrogates of Nonlinear High-Dimensional Parameter-to-Prediction Maps.

We present a preliminary investigation of the use of Multi-Layer Perceptrons (MLP) and Recurrent Neural Networks (RNNs) as surrogates of parameter-to-prediction maps of com- putational expensive dynamical models. In particular, we target the approximation of Quan- tities of Interest (QoIs) derived from the solution of a Partial Differential Equations (PDEs) at different time instants. In order to limit the scope of our study while targeting a rele- vant application, we focus on the problem of computing variations in the ice sheets mass (our QoI), which is a proxy for global mean sea-level changes. We present a number of neural network formulations and compare their performance with that of Polynomial Chaos Expansions (PCE) constructed on the same data.

97 MATHEMATICS AND COMPUTING↗

Neural Networks as Surrogates of Nonlinear High-Dimensional Parameter-to-Prediction Maps

We present a preliminary investigation of the use of Multi-Layer Perceptrons (MLP) and Recurrent Neural Networks (RNNs) as surrogates of parameter-to-prediction maps of computational expensive dynamical models. In particular, we target the approximation of Quantities of Interest (QoIs) derived from the solution of a Partial Differential Equations (PDEs) at different time instants. In order to limit the scope of our study while targeting a relevant application, we focus on the problem of computing variations in the ice sheets mass (our QoI), which is a proxy for global mean sea-level changes. We present a number of neural network formulations and compare their performance with that of Polynomial Chaos Expansions (PCE) constructed on the same data.

97 MATHEMATICS AND COMPUTING↗

A unified materials approach to mitigating optical nonlinearities in optical fiber. III. Canonical examples and materials road map

Abstract This paper, Part III in the Trilogy (Ballato, Cavillon, Dragic, 2018; Dragic, Cavillon, Ballato, et al ., 2018a,b), provides a road map for the development of simple core/clad optical fibers whose enhanced performance—in particular, marked reductions in optical nonlinearities—is achieved materially and not through the more conventional present routes of geometrically complex fiber design. More specifically, the material properties that give rise to Brillouin, Raman and Rayleigh scattering, transverse mode instabilities ( TMI ), and n 2 ‐mediated nonlinear effects are compiled and results on a wide range of optical fibers are discussed with a focus on trends in performance with glass composition. Furthermore, optical power scaling estimations as well as binary and ternary property diagrams associated with Rayleigh scattering, the Brillouin gain coefficient ( BGC ) and the thermo‐optic coefficient (d n /d T ) are developed and employed to graphically represent general trends with composition along with compositional targets for a single intrinsically low nonlinearity, silica‐based optical fiber that can achieve the power scaling goals of future high energy fiber laser applications. A foundational finding of this work is that the high‐silica content optical fibers fabricated using conventional chemical vapor deposition methods will not suffice to meet the power scaling demands of future high‐power and high‐energy fiber lasers.

36 MATERIALS SCIENCE↗

Grassmannian Diffusion Maps--Based Dimension Reduction and Classification for High-Dimensional Data

This work introduces the Grassmannian diffusion maps (GDMaps), a novel nonlinear dimensionality reduction technique that defines the affinity between points through their representation as low-dimensional subspaces corresponding to points on the Grassmann manifold. Here, the method is designed for applications, such as image recognition and data-based classification of constrained high-dimensional data where each data point itself is a high-dimensional object (i.e., a large matrix) that can be compactly represented in a lower-dimensional subspace. The GDMaps is composed of two stages. The first is a pointwise linear dimensionality reduction wherein each high-dimensional object is mapped onto the Grassmann manifold representing the low-dimensional subspace on which it resides. The second stage is a multipoint nonlinear kernel-based dimension reduction using diffusion maps to identify the subspace structure of the points on the Grassmann manifold. To this end, an appropriate Grassmannian kernel is used to construct the transition matrix of a random walk on a graph connecting points on the Grassmann manifold. Spectral analysis of the transition matrix yields low-dimensional Grassmannian diffusion coordinates embedding the data into a low-dimensional reproducing kernel Hilbert space. Further, a novel data classification/recognition technique is developed based on the construction of an overcomplete dictionary of reduced dimension whose atoms are given by the Grassmannian diffusion coordinates. Three examples are considered. First, a "toy" example shows that the GDMaps can identify an appropriate parametrization of structured points on the unit sphere. The second example demonstrates the ability of the GDMaps to revealing the intrinsic subspace structure of high-dimensional random field data. In the last ex- ample, a face recognition problem is solved considering face images subject to varying illumination conditions, changes in face expressions, and occurrence of occlusions. The technique presented high recognition rates (i.e., 95% in the best case) using a fraction of the data required by conventional methods.

42 ENGINEERING↗

Residual Stresses and Plastic Deformation in Self-Pierce Riveting of Dissimilar Aluminum-to-Magnesium Alloys

In this study, the complex relationship between deformation history and residual stresses in a magnesium-to-aluminum self-pierce riveted (SPR) joint is elucidated using numerical and experimental approaches. Non-linear finite element (FE) simulations incorporating strain rate and temperature effects were performed to model the deformation in the SPR process. In order to accurately capture the deformation, a stress triaxiality-based damage material model was employed to capture the sheet piercing from the rivet. Strong visual comparison between the physical cross-section of the SPR joint and the simulation was achieved. To aid in understanding of the role of deformation in the riveting process and to validate the modeling approach, several experimental measurements were conducted. To quantify the plastic deformation from the piercing of the rivet, micro hardness mapping was performed on a cross-section of the SPR joint. The FE model showed very strong correlation to the experimental hardness mapping results suggesting the nonlinear model captured the plastic deformation with high accuracy. To measure the elastic residual stresses in the SPR joint, neutron and x-ray diffraction mapping techniques were conducted and in general, the FE model correlated well to the trends and magnitudes of the elastic stresses. While some error occurred in between the model and the neutron and x-ray diffraction results, the numerical approach developed in this study shows potential as a tool for understanding SPR behavior as well as optimizing the process parameters.

36 MATERIALS SCIENCE↗

Analysis of the nonlinear behavior of shear-Alfvén modes in tokamaks based on Hamiltonian mapping techniques

We present a series of numerical simulation experiments set up to illustrate the fundamental physics processes underlying the nonlinear dynamics of Alfvénic modes resonantly excited by energetic particles in tokamak plasmas and of the ensuing energetic particle transports. These phenomena are investigated by following the evolution of a test particle population in the electromagnetic fields computed in self-consistent MHD-particle simulation performed by the HMGC code. Hamiltonian mapping techniques are used to extract and illustrate several features of wave-particle dynamics. The universal structure of resonant particle phase space near an isolated resonance is recovered and analyzed, showing that bounded orbits and untrapped trajectories, divided by the instantaneous separatrix, form phase space zonal structures, whose characteristic non-adiabatic evolution time is the same as the nonlinear time of the underlying fluctuations. Bounded orbits correspond to a net outward resonant particle flux, which produces a flattening and/or gradient inversion of the fast ion density profile around the peak of the linear wave-particle resonance. The connection of this phenomenon to the mode saturation is analyzed with reference to two different cases: a Toroidal Alfvén eigenmode in a low shear magnetic equilibrium and a weakly unstable energetic particle mode for stronger magnetic shear. It is shown that, in the former case, saturation is reached because of radial decoupling (resonant particle redistribution matching the mode radial width) and is characterized by a weak dependence of the mode amplitude on the growth rate. In the latter case, saturation is due to resonance detuning (resonant particle redistribution matching the resonance width) with a stronger dependence of the mode amplitude on the growth rate.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Learning high-dimensional parametric maps via reduced basis adaptive residual networks

We propose a scalable framework for the learning of high-dimensional parametric maps via adaptively constructed residual network (ResNet) maps between reduced bases of the inputs and outputs. When just few training data are available, it is beneficial to have a compact parametrization in order to ameliorate the ill-posedness of the neural network training problem. By linearly restricting high-dimensional maps to informed reduced bases of the inputs, one can compress high-dimensional maps in a constructive way that can be used to detect appropriate basis ranks, equipped with rigorous error estimates. A scalable neural network learning framework is thus to learn the nonlinear compressed reduced basis mapping. Unlike the reduced basis construction, however, neural network constructions are not guaranteed to reduce errors by adding representation power, making it difficult to achieve good practical performance. Inspired by recent approximation theory that connects ResNets to sequential minimizing flows, we present an adaptive ResNet construction algorithm. This algorithm allows for depth-wise enrichment of the neural network approximation, in a manner that can achieve good practical performance by first training a shallow network and then adapting. We prove universal approximation of the associated neural network class for $L^2_v$ functions on compact sets. Our overall framework allows for constructive means to detect appropriate breadth and depth, and related compact parametrizations of neural networks, significantly reducing the need for architectural hyperparameter tuning. Numerical experiments for parametric PDE problems and a 3D CFD wing design optimization parametric map demonstrate that the proposed methodology can achieve remarkably high accuracy for limited training data, and outperformed other neural network strategies we compared against.

42 ENGINEERING↗

Auto calibration of a cone-beam-CT

Purpose: This paper introduces a novel autocalibration method for cone-beam-CTs (CBCT) or flat-panel CTs, assuming a perfect rotation. The method is based on ellipse-fitting. Autocalibration refers to accurate recovery of the geometric alignment of a CBCT device from projection images alone, without any manual measurements. Methods: The authors use test objects containing small arbitrarily positioned radio-opaque markers. No information regarding the relative positions of the markers is used. In practice, the authors use three to eight metal ball bearings (diameter of 1 mm), e.g., positioned roughly in a vertical line such that their projection image curves on the detector preferably form large ellipses over the circular orbit. From this ellipse-to-curve mapping and also from its inversion the authors derive an explicit formula. Nonlinear optimization based on this mapping enables them to determine the six relevant parameters of the system up to the device rotation angle, which is sufficient to define the geometry of a CBCT-machine assuming a perfect rotational movement. These parameters also include out-of-plane rotations. The authors evaluate their method by simulation based on data used in two similar approaches [L. Smekal, M. Kachelriess, S. E, and K. Wa, 'Geometric misalignment and calibration in cone-beam tomography,' Med. Phys. 31(12), 3242-3266 (2004); K. Yang, A. L. C. Kwan, D. F. Miller, and J. M. Boone, 'A geometric calibration method for cone beam CT systems,' Med. Phys. 33(6), 1695-1706 (2006)]. This allows a direct comparison of accuracy. Furthermore, the authors present real-world 3D reconstructions of a dry human spine segment and an electronic device. The reconstructions were computed from projections taken with a commercial dental CBCT device having two different focus-to-detector distances that were both calibrated with their method. The authors compare their reconstruction with a reconstruction computed by the manufacturer of the CBCT device to demonstrate the achievable spatial resolution of their calibration procedure. Results: Compared to the results published in the most closely related work [K. Yang, A. L. C. Kwan, D. F. Miller, and J. M. Boone, 'A geometric calibration method for cone beam CT systems,' Med. Phys. 33(6), 1695-1706 (2006)], the simulation proved the greater accuracy of their method, as well as a lower standard deviation of roughly 1 order of magnitude. When compared to another similar approach [L. Smekal, M. Kachelriess, S. E, and K. Wa, 'Geometric misalignment and calibration in cone-beam tomography,' Med. Phys. 31(12), 3242-3266 (2004)], their results were roughly of the same order of accuracy. Their analysis revealed that the method is capable of sufficiently calibrating out-of-plane angles in cases of larger cone angles when neglecting these angles negatively affects the reconstruction. Fine details in the 3D reconstruction of the spine segment and an electronic device indicate a high geometric calibration accuracy and the capability to produce state-of-the-art reconstructions. Conclusions: The method introduced here makes no requirements on the accuracy of the test object. In contrast to many previous autocalibration methods their approach also includes out-of-plane rotations of the detector. Although assuming a perfect rotation, the method seems to be sufficiently accurate for a commercial CBCT scanner. For devices which require higher dimensional geometry models, the method could be used as a initial calibration procedure.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Topological horseshoes in travelling waves of discretized nonlinear wave equations

Applying the concept of anti-integrable limit to coupled map lattices originated from space-time discretized nonlinear wave equations, we show that there exist topological horseshoes in the phase space formed by the initial states of travelling wave solutions. In particular, the coupled map lattices display spatio-temporal chaos on the horseshoes.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Improved high-resolution mapping of phase-matching conditions in second-order nonlinear crystals via imaging

Large-aperture nonlinear crystals for frequency conversion and parametric amplification exhibit spatial variations in phase-matching conditions, which affect the performance of high-power laser systems. We demonstrate significant improvements in speed and resolution to a technique that spatially maps these phase-matching conditions using imaging of the beam resulting from detuned second-harmonic generation. The phase-matching conditions at the pyramidal sector boundary of partially deuterated potassium dihydrogen phosphate (DKDP) are mapped with sub-0.1-mm resolution. We examine the effect of further increases in scan rate on the precision of the technique and demonstrate a scan rate of 0.8 cm 2 per minute at the aforementioned resolution.

Wampler, R. [Univ. of Rochester, NY (United States↗

Airborne synthetic aperture radar observations and simulations for waves in ice

The Canada Centre for Remote Sensing CV-580 aircraft collected C-band SAR data over the marginal ice zone off the east coast of Newfoundland during the Labrador Ice Margin Experiment (LIMEX) in March 1989. One component of the LIMEX '89 program was the study of ocean waves penetrating the marginal ice zone. We consider nearly coincidental observations of waves in ice by airborne SAR and wave-induced ice motion measurements. We explain the wave patterns observed in the SAR imagery, and the corresponding SAR image spectra, in terms of SAR wave imaging models. These include the well-known tilt cross-section modulation, linear, quasi-linear, and nonlinear velocity bunching forward mapping models (FMMs), and the assertion that the concept of coherence time limitation applies differently to the cases of waves in ice and open water. We modify the concept of the scene coherence time to include two parts: first, a decorrelation time deduced from the inherent azimuth cutoff in the nonlinear velocity bunching FMM; and second, the intrinsic scene coherence time which is a measure of the time scale over which an open water Bragg scattering patch retains its phase structure. Either of these coherence time scales could dominate the SAR image formation process, depending upon the environmental conditions (the wave spectrum and the wind speed, for example). Observed SAR image spectra and forward mapped ice motion package spectra are favorably compared.

Vachon, Paris W.↗

Parallel homotopy curve tracking on a hypercube

An investigation is conducted to find good parallel algorithms for solving systems of nonlinear equations using probability-one homotopy methods. Particular attention is paid to algorithms for the hypercube. Methods for one of the most computationally expensive steps of the homotopy approach, the computation of the kernel of the Jacobian matrix of the homotopy map, are studied. General nonlinear systems of equations with small and dense Jacobian matrices are considered, however, polynomial systems are not, since their structure leads to different strategies for parallelism. The mathematics behind the homotopy algorithm is summarized and the use of orthogonal factorizations is discussed. Parallel algorithms for orthogonal factorizations and triangular system solving are described. Computational results are presented and discussed.

Chakraborty, A.↗

Optimum receiver structures for phase-multiplexed modulations

A MAP (maximum a posteriori) estimation loop is derived for a QPSK signal and, by suitable approximation to the nonlinearity which arises as a consequence of the MAP theory, is reconfigured with practical realizations that are valid for high and low SNRs. In particular, it is shown that by approximating the hyperbolic tangent nonlinearity in the MAP estimation loop by the first two terms in its power series, an interesting practical realization of this loop results which applies at low SNRs. The error signal in this loop is formed by multiplying the error signal and lock detector output signal of a conventional biphase Costas loop. A generalized linear in-phase channel configuration is proposed which allows carrier reconstruction from an unbalanced QPSK signal at all ratios of data rates and powers in the two channels, even in the limit as these ratios simultaneously approach unity, i.e., balanced quadriphase.

Simon, M. K.↗

Influence of doping and thickness on domain avalanches in lead zirconate titanate thin films

In undoped lead zirconate titanate films of 1–2 μm thick, domain walls move in clusters with a correlation length of approximately 0.5–2 μm. Band excitation piezoresponse force microscopy mapping of the piezoelectric nonlinearity revealed that niobium (Nb) doping increases the average concentration or mobility of domain walls without changing the cluster area of correlated domain wall motion. In comparison, manganese (Mn) doping reduces the contribution of mobile domain walls to the dielectric and piezoelectric responses without changing the cluster area for correlated motion. In both Nb and Mn doped films, the cluster area increases and the cluster density drops as the film thickness increases from 250 to 1250 nm. This is evident in spatial maps generated from the analysis of irreversible to reversible ratios of the Rayleigh coefficients.

36 MATERIALS SCIENCE↗

Improved Fine Sun Sensor Field of View Calibration

The fine Sun sensor used on many spacecraft consists of two independent single-axis sensor heads, nominally mounted perpendicularly. These detect the Sun angle over a field of view typically of +32 deg. (There is a trade-off between accuracy and size of the field of view that allows for much latitude in any numbers quoted.) The nonlinear "transfer" function that maps the telemetered counts into observed angles consists of 9 adjustable parameters for each axis (1 8 total). An augmented transfer function has previously been reported that achieves a significant accuracy improvement across the entire field of view. That function expands the parameter set to 12 coefficients per axis (24 total) and includes cross terms combining counts from both axes. To make the best use of the Sun sensor for attitude determination, it must be calibrated after launch. However, the large number of parameters and the nonlinearity of the problem make this a challenging task. The purpose of this paper is to examine ways to improve convergence of the parameter search algorithm. In particular, experience has shown that the problem should be broken down into several steps, solving for a selected subset of the parameters at each step. This approach has now been incorporated as an option in the calibration utility.

Sedlak, J.↗