Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “gaussian”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

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

At least 73 records · Page 4

Conversion of a TEM10 beam into two nearly Gaussian beams

To achieve high output power and/or high efficiencies, laser systems are often designed to operate with either high-order or multiorder transverse modes. This comes at the expense of beam quality. Herein, a technique for converting the two lobes of a higher-order (Hermite–Gaussian TEM10) beam into two TEM00 beams is discussed. It is shown that nearly 94% of the power in a single lobe of the TEM10 beam spatially overlaps an appropriately chosen TEM00 beam. Beam quality (defined by the M2 beam quality factor) is increased by separating the two TEM10 lobes with an edge aperture. A single-mode optical fiber can be used as a spatial filter to eliminate residual higher-order mode content. This study suggests that more than 90% of the power in a TEM10 beam can be converted into two separate TEM00 beams.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Enhanced Data Efficiency Using Deep Neural Networks and Gaussian Processes for Aerodynamic Design Optimization

Adjoint-based optimization methods are attractive for aerodynamic shape design primarily due to their computational costs being independent of the dimensionality of the input space and their ability to generate high-fidelity gradients that can then be used in a gradient-based optimizer. This makes them very well suited for high-fidelity simulation based aerodynamic shape optimization of highly parametrized geometries such as aircraft wings. However, the development of adjoint-based solvers involve careful mathematical treatment and their implementation require detailed software development. Furthermore, they can become prohibitively expensive when multiple optimization problems are being solved, each requiring multiple restarts to circumvent local optima. In this work, we propose a machine learning enabled, surrogate-based framework that replaces the expensive adjoint solver, without compromising on predicting predictive accuracy. Specifically, we first train a deep neural network (DNN) from training data generated from evaluating the high-fidelity simulation model on a model-agnostic design of experiments on the geometry shape parameters. The optimum shape may then be computed by using a gradient-based optimizer coupled with the trained DNN. Subsequently, we also perform a gradient-free Bayesian optimization, where the trained DNN is used as the prior mean. We observe that the latter framework (DNN-BO) improves upon the DNN-only based optimization strategy for the same computational cost. Overall, this framework predicts the true optimum with very high accuracy, while requiring far fewer high-fidelity function calls compared to the adjoint-based method. Furthermore, we show that multiple optimization problems can be solved with the same machine learning model with high accuracy, to amortize the offline costs associated with constructing our models. Our methodology finds applications in the early stages of aerospace design. (C) 2021 Published by Elsevier Masson SAS.

Renganathan, S. Ashwin↗

Ab initio electronic structure of quasi-two-dimensional materials: A “native” Gaussian–plane wave approach

Ab initio electronic structure calculations of two-dimensional layered structures are typically performed using codes that were developed for three-dimensional structures, which are periodic in all three directions. The introduction of a periodicity in the third direction (perpendicular to the layer) is completely artificial and may lead in some cases to spurious results and to difficulties in treating the action of external fields. In this paper we develop a new approach, which is “native” to quasi-2D materials, making use of basis function that are periodic in the plane, but atomic-like in the perpendicular direction. We show how some of the basic tools of ab initio electronic structure theory — density functional theory, GW approximation and Bethe-Salpeter equation — are implemented in the new basis. We argue that the new approach will be preferable to the conventional one in treating the peculiarities of layered materials, including the long range of the unscreened Coulomb interaction in insulators, and the effects of strain, corrugations, and external fields.

Trevisanutto, Paolo E.↗

Building unbiased estimators from non-gaussian likelihoods with application to shear estimation

We develop a general framework for generating estimators of a given quantity which are unbiased to a given order in the difference between the true value of the underlying quantity and the fiducial position in theory space around which we expand the likelihood. We apply this formalism to rederive the optimal quadratic estimator and show how the replacement of the second derivative matrix with the Fisher matrix is a generic way of creating an unbiased estimator (assuming choice of the fiducial model is independent of data). Next we apply the approach to estimation of shear lensing, closely following the work of Bernstein and Armstrong (2014). Our first order estimator reduces to their estimator in the limit of zero shear, but it also naturally allows for the case of non-constant shear and the easy calculation of correlation functions or power spectra using standard methods. Both our first-order estimator and Bernstein and Armstrong’s estimator exhibit a bias which is quadratic in true shear. Our third-order estimator is, at least in the realm of the toy problem of Bernstein and Armstrong, unbiased to 0.1% in relative shear errors Δg/g for shears up to |g| = 0.2.

79 ASTRONOMY AND ASTROPHYSICS↗

Measurement-induced disturbances and nonclassical correlations of Gaussian states

We study quantum correlations beyond entanglement in two-mode Gaussian states of continuous-variable systems by means of the measurement-induced disturbance (MID) and its ameliorated version (AMID). In analogy with the recent studies of the Gaussian quantum discord, we define a Gaussian AMID by constraining the optimization to all bi-local Gaussian positive operator valued measurements. We solve the optimization explicitly for relevant families of states, including squeezed thermal states. Remarkably, we find that there is a finite subset of two-mode Gaussian states comprising pure states where non-Gaussian measurements such as photon counting are globally optimal for the AMID and realize a strictly smaller state disturbance compared to the best Gaussian measurements. However, for the majority of two-mode Gaussian states the unoptimized MID provides a loose overestimation of the actual content of quantum correlations, as evidenced by its comparison with Gaussian discord. This feature displays strong similarity with the case of two qubits. Upper and lower bounds for the Gaussian AMID at fixed Gaussian discord are identified. We further present a comparison between Gaussian AMID and Gaussian entanglement of formation, and classify families of two-mode states in terms of their Gaussian AMID, Gaussian discord, and Gaussian entanglement of formation. Our findings provide a further confirmation of the genuinely quantum nature of general Gaussian states, yet they reveal that non-Gaussian measurements can play a crucial role for the optimized extraction and potential exploitation of classical and nonclassical correlations in Gaussian states.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Searching for primordial non-Gaussianity in Planck CMB maps using a combined estimator

The extensive search for deviations from Gaussianity in cosmic microwave background radiation (CMB) data is very important due to the information about the very early moments of the universe encoded there. Recent analyses from Planck CMB data do not exclude the presence of non-Gaussianity of small amplitude, although they are consistent with the Gaussian hypothesis. The use of different techniques is essential to provide information about types and amplitudes of non-Gaussianities in the CMB data. In particular, we find interesting to construct an estimator based upon the combination of two powerful statistical tools that appears to be sensitive enough to detect tiny deviations from Gaussianity in CMB maps. This estimator combines the Minkowski functionals with a Neural Network, maximizing a tool widely used to study non-Gaussian signals with a reinforcement of another tool designed to identify patterns in a data set. We test our estimator by analyzing simulated CMB maps contaminated with different amounts of local primordial non-Gaussianity quantified by the dimensionless parameter f{sub NL}. We apply it to these sets of CMB maps and find ∼> 98% of chance of positive detection, even for small intensity local non-Gaussianity like f{sub NL} = 38±18, the current limit from Planck data for large angular scales. Additionally, we test the suitability to distinguish between primary and secondary non-Gaussianities: first we train the Neural Network with two sets, one of nearly Gaussian CMB maps (|f{sub NL}| ≤ 10) but contaminated with realistic inhomogeneous Planck noise (i.e., secondary non-Gaussianity) and the other of non-Gaussian CMB maps, that is, maps endowed with weak primordial non-Gaussianity (28 ≤ f{sub NL} ≤ 48); after that we test an ensemble composed of CMB maps either with one of these non-Gaussian contaminations, and find out that our method successfully classifies ∼ 95% of the tested maps as being CMB maps containing primordial or secondary non-Gaussianity. Furthermore, we analyze the foreground-cleaned Planck maps obtaining constraints for non-Gaussianity at large-angles that are in good agreement with recent constraints. Finally, we also test the robustness of our estimator including cut-sky masks and realistic noise maps measured by Planck, obtaining successful results as well.

79 ASTRONOMY AND ASTROPHYSICS↗

Non-Gaussianity in the foreground-reduced CMB maps

A detection or nondetection of primordial non-Gaussianity by using the cosmic microwave background radiation (CMB) data is crucial not only to discriminate inflationary models but also to test alternative scenarios. Non-Gaussianity offers, therefore, a powerful probe of the physics of the primordial Universe. The extraction of primordial non-Gaussianity is a difficult enterprise since several effects of a nonprimordial nature can produce non-Gaussianity. Given the far-reaching consequences of such a non-Gaussianity for our understanding of the physics of the early Universe, it is important to employ a range of different statistical tools to quantify and/or constrain its amount in order to have information that may be helpful for identifying its causes. Moreover, different indicators can in principle provide information about distinct forms of non-Gaussianity that can be present in CMB data. Most of the Gaussianity analyses of CMB data have been performed by using part-sky frequency, where the mask is used to deal with the galactic diffuse foreground emission. However, full-sky map seems to be potentially more appropriate to test for Gaussianity of the CMB data. On the other hand, masks can induce bias in some non-Gaussianity analyses. Here we use two recent large-angle non-Gaussianity indicators, based on skewness and kurtosis of large-angle patches of CMB maps, to examine the question of non-Gaussianity in the available full-sky five-year and seven-year Wilkinson Microwave Anisotropy Probe (WMAP) maps. We show that these full-sky foreground-reduced maps present a significant deviation from Gaussianity of different levels, which vary with the foreground-reducing procedures. We also make a Gaussianity analysis of the foreground-reduced five-year and seven-year WMAP maps with a KQ75 mask, and compare with the similar analysis performed with the corresponding full-sky foreground-reduced maps. This comparison shows a significant reduction in the levels of non-Gaussianity when the mask is employed, which provides indications on the suitability of the foreground-reduced maps as Gaussian reconstructions of the full-sky CMB.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

On the Response of a Nonlinear Structure to High Kurtosis Non-Gaussian Random Loadings

This paper is a follow-on to recent work by the authors in which the response and high-cycle fatigue of a nonlinear structure subject to non-Gaussian loadings was found to vary markedly depending on the nature of the loading. There it was found that a non-Gaussian loading having a steady rate of short-duration, high-excursion peaks produced essentially the same response as would have been incurred by a Gaussian loading. In contrast, a non-Gaussian loading having the same kurtosis, but with bursts of high-excursion peaks was found to elicit a much greater response. This work is meant to answer the question of when consideration of a loading probability distribution other than Gaussian is important. The approach entailed nonlinear numerical simulation of a beam structure under Gaussian and non-Gaussian random excitations. Whether the structure responded in a Gaussian or non-Gaussian manner was determined by adherence to, or violations of, the Central Limit Theorem. Over a practical range of damping, it was found that the linear response to a non-Gaussian loading was Gaussian when the period of the system impulse response is much greater than the rate of peaks in the loading. Lower damping reduced the kurtosis, but only when the linear response was non-Gaussian. In the nonlinear regime, the response was found to be non-Gaussian for all loadings. The effect of a spring-hardening type of nonlinearity was found to limit extreme values and thereby lower the kurtosis relative to the linear response regime. In this case, lower damping gave rise to greater nonlinearity, resulting in lower kurtosis than a higher level of damping.

Rizzi, Stephen A.↗

Nonlinear clustering in models with primordial non-Gaussianity: The halo model approach

We develop the halo model of large-scale structure as an accurate tool for probing primordial non-Gaussianity. In this study we focus on understanding the matter clustering at several redshifts in the context of primordial non-Gaussianity that is a quadratic correction to the local Gaussian potential, characterized by the parameter f{sub NL}. In our formulation of the halo model we pay special attention to the effect of halo exclusion and show that this can potentially solve the long-standing problem of excess power on large scales in this model. The halo model depends on the mass function, clustering of halo centers, and the density profiles. We test these ingredients using a large ensemble of high-resolution Gaussian and non-Gaussian numerical simulations, covering f{sub NL}={l_brace}0,+100,-100{r_brace}. In particular, we provide a first exploration of how halo density profiles change in the presence of primordial non-Gaussianity. We find that for f{sub NL} positive (negative) high-mass haloes have an increased (decreased) core density, so being more (less) concentrated than in the Gaussian case. We also examine the halo bias and show that, if the halo model is correct, then there is a small asymmetry in the scale dependence of the bias on very large scales, which arises because the Gaussian bias must be renormalized. We show that the matter power spectrum is modified by {approx}2.5% and {approx}3.5% on scales k{approx}1.0 h Mpc{sup -1} at z=0 and z=1, respectively. Our halo model calculation reproduces the absolute amplitude to within < or approx. 10% and the ratio of non-Gaussian to Gaussian spectra to within < or approx. 1%. We also measure the matter correlation function and find similarly good levels of agreement between the halo model and the data. We anticipate that this modeling will be useful for constraining f{sub NL} from measurements of the shear correlation function in future weak lensing surveys such as Euclid.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Foundations of Feature Selection and Classification for Non-Gaussian Distributed Targets

This report describes the author’s research toward creating algorithms for selecting useful statistical features in pattern/target classification problems in which the features are non-Gaussian distributed. In engineering practice, it is common to either (1) Not perform any feature selection procedure, or (2) Use a feature selection algorithm that assumes the features are Gaussian distributed. This process can be far from optimal if the data are actually non-Gaussian, as they frequently are in practice. This research has the goal of mitigating that problem by creating a useful algorithm that can be used in practice. The approach is to focus on the performance indices used in common feature selection algorithms, such as Branch and Bound algorithms and the Sequential Forward Selection Algorithm. Ordinarily, the performance index used to measure the class separation in feature space involves assuming the data are Gaussian and deriving tractable performance indices that use closed form equations that represent the probability density functions of the class data. The advantage of this approach is that it produces feature selection algorithms that have low computational complexity and do not require knowledge of the actual data densities. The disadvantage is that these algorithms may not perform reasonably when the data are non-Gaussian, as they commonly are in practice. This research examines the use of information-theoretic class separability measures that can deal with the non-Gaussian case. In particular, this work shows that the Hellinger Distance (a divergence measure), has very desirable mathematical properties and can be useful for feature selection. The work in this paper employs a multivariate kernel density estimator along with the Hellinger Distance and an optimal subset selection algorithm to do feature selection. This report (1) Defines the classification and feature selection problems, (2) Describes the continuous variable Hellinger distance in one dimension and discusses its mathematical properties in comparison with other distance measures, (3) Develops the discrete variable Hellinger distance expression for one and two dimensions, and shows that the discrete Hellinger distance for multiple dimensions is a straightforward extension, and (4) Presents experimental density estimation results using a new MATLAB implementation of a multivariate kernel density estimator. The draft of this report served as a starting point for Ensign Matthew Wilder’s MSEE thesis work at the Naval Postgraduate School (NPS) in 2010 and 2011. Ensign Wilder completed his thesis and graduated from NPS in June 2011 (M. J. Wilder, “Automatic Target Recognition: Statistical Feature Selection of Non-Gaussian Distributed Target Classes,” MSEE Thesis, Naval Postgraduate School, Thesis Advisor: Grace A. Clark, Second Reader: Monique P. Fargues, June 2011). In the thesis, Ensign Wilder created implementations of the Branch and Bound and Sequential Feature Selection algorithms, extended the multivariate kernel density estimator code, and evaluated the performance of Hellinger Distance-based feature selection algorithms using both simulated data and data from a real-world benchmark data set. The new algorithm was shown to be very effective for both Gaussian and non-Gaussian data sets. The advantages gained for non-Gaussian data come at the cost of increased computational complexity. Using a laptop computer for practical problems, we generally use initial feature vectors containing no more than about ten or twelve features. Proposed future research lies in creating faster kernel density estimator algorithms that would allow for the use of higher dimensional data.

42 ENGINEERING↗