Search NASA⌕ Search

SEARCH · Search NASA

Results for “Gaussian Process”

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

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

At least 1,441 records · Page 80

Performance of Compact Pulsed Thermal Imaging System for In-Service Applications. Pulsed thermal tomography nondestructive examination of additively manufactured reactor materials and components

Additive manufacturing (AM) is an emerging method for cost-efficient fabrication of complex topology nuclear reactor parts from high-strength corrosion resistance alloys, such as stainless steel and Inconel. AM of metallic structures for nuclear energy applications is currently based on laser powder bed fusion (LPBF) process, which has the capability of melting metallic powder and net shaping the structures with relatively high precision. Some of the challenges with using LPBF method for nuclear manufacturing include the possibility of introducing pores into metallic structures. Integrity of AM structures needs to be evaluated nondestructively because material flaws could lead to premature failures due to creep in high temperature nuclear reactor environment. Currently, there exist limited capabilities to evaluate actual AM structures nondestructively. Pulsed Thermography (PT) imaging provides a capability for non-destructive evaluation (NDE) of sub-surface defects in arbitrary size structures. The PT method is based on recording material surface temperature transients with infrared (IR) camera following thermal pulse delivered on material surface with flash light. The PT method has advantages for NDE of actual AM structures because the method involves one-sided non-contact measurements and fast processing of large sample areas captured in one image. The data cube of PT measurements consists of surface temperature taken at sequential time intervals T(x,y,t). Material defects can be detected either by analyzing the thermograms T(x,y,t) data cube, or by using thermal tomography (TT) algorithm to obtain 3D spatial reconstruction of thermal effusivity e(x,y,z). To reduce the cost and enable in-service NDE in spatially constrained environment, it is highly desirable to develop PT with compact and inexpensive IR camera. Following initial qualification of an AM component for deployment in a nuclear reactor, a compact PT system can also be used for in-service nondestructive evaluation (NDE) applications. However, data cube obtained with PT based on compact IR camera suffers from strong thermal noises and loss of features due to relatively low sampling rate. In this report we describe two unsupervised machine learning (ML) algorithms for enhancement of PT images obtained with compact IR camera. In one approach, we introduce Sparse Coding Discrete Cosine Transform (SC/DCT) algorithm to remove additive white Gaussian noise (AWGN) from spatial thermal effusivity reconstructions. In another approach we introduce a Spatial Temporal Denoised Thermal Source Separation (STDTSS) ML algorithm to process thermograms. The STDTSS algorithm consists of spatial and temporal denoising using Gaussian and Savitzky–Golay filtering, followed by the matrix decomposition using Principal Component Analysis (PCA), and Independent Component Analysis (ICA) to automatically detect flaws. In the work described in this report, we constructed a compact PT system using a relatively small and low-cost FLIR A65 camera, consisting on uncooled microbolometer detector. Performance of SC/DCT algorithm was demonstrated on enhancing TT images of Inconel 718 AM plate. Performance of the STDTSS methods was investigated using thermography data obtained from imaging stainless steel 316L specimens produced with LPBF method with imprinted calibrated porosity defects.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Models and error analyses in urban air quality estimation

Estimation theory has been applied to a wide range of aerospace problems. Application of this expertise outside the aerospace field has been extremely limited, however. This paper describes the use of covariance error analysis techniques in evaluating the accuracy of pollution estimates obtained from a variety of concentration measuring devices. It is shown how existing software developed for aerospace applications can be applied to the estimation of pollution through the processing of measurement types involving a range of spatial and temporal responses. The modeling of pollutant concentration by meandering Gaussian plumes is described in some detail. Time averaged measurements are associated with a model of the average plume, using some of the same state parameters and thus avoiding the problem of state memory. The covariance analysis has been implemented using existing batch estimation software. This usually involves problems in handling dynamic noise; however, the white dynamic noise has been replaced by a band-limited process which can be easily accommodated by the software.

Englar, T., Jr.↗

Bayesian sequential optimal experimental design for nonlinear models using policy gradient reinforcement learning

We present a mathematical framework and computational methods for optimally designing a finite sequence of experiments. This sequential optimal experimental design (sOED) problem is formulated as a finite-horizon partially observable Markov decision process (POMDP) under a Bayesian setting and with information-theoretic utilities. The formulation is general and may accommodate continuous random variables, non-Gaussian posteriors, and nonlinear forward models. The sOED design policy incorporates elements of feedback and lookahead simultaneously, and we show it to generalize the commonly-used batch and greedy design strategies. We solve for the sOED policy using the policy gradient (PG) method from reinforcement learning, and provide a derivation for the PG expression in the sOED context. Adopting an actor-critic approach, the policy and value functions are parameterized using deep neural networks and improved via PG estimates produced from simulated episodes of designs and observations. The new PG-sOED algorithm is first validated on a linear-Gaussian benchmark, and then compared against other design baselines on a sensor movement problem for contaminant source inversion in a convection-diffusion field. As a result, we provide explanation for the policy behaviors using knowledge of the underlying physical process.

97 MATHEMATICS AND COMPUTING↗

Effective Pore-Scale Dispersion Upscaling with the Correlated Continuous Time Random Walk Approach

We propose a general framework for upscaling dispersion in porous media. A key challenge of the upscaling procedure is to relate the temporal evolution of spreading to the small scale velocity field properties. The representation of the Lagrangian velocity transition process as a Markovian process in space provides a simple way to quantify complex correlation properties, i.e. non-Gaussian velocity distributions. The resulting effective transport model is a correlated CTRW. We use this framework to upscale pore scale dispersion for a periodic pore geometry. The correlated CTRW model is defined by the transit time distribution across one pore and the transition probability density quantifying the correlation between successive transit times. The latter is of central importance since it accounts for incomplete mixing at the pore throats. The predictions of the correlated CTRW model are in good agreement with the pore scale simulations over the pre-asymptotic and asymptotic regimes. We investigate the representation of this effective dispersion model in phase space (position, velocity) in a form similar to a Boltzmann transport equation.

multiscale modeling, dimension reduction, transpor↗

Non-Gaussian halo mass function and non-spherical halo collapse: theory vs. simulations

The mass distribution of dark matter halos is a sensitive probe of primordial non-Gaussianity (NG). We derive an analytical formula of the halo mass function by perturbatively computing excursion set path-integrals for a non-Gaussian density field with non-vanishing skewness, f{sub NL}. We assume a stochastic barrier model which captures the main features of the ellipsoidal collapse of halos. Contrary to previous results based on extensions of the Press-Schechter formalism to NG initial conditions, we find that the non-spherical collapse of halos directly alter the signature of primordial NG. This points toward a potential degeneracy between the effect of primordial non-Gaussianity and that of non-linear halo collapse. The inferred mass function is found to be in remarkable agreement with N-body simulations of NG local type. Deviations are well within numerical uncertainties for all values of f{sub NL}{sup loc} in the range of validity of the perturbative calculation (|f{sub nl}{sup loc}|∼<200). Moreover, the comparison with simulation results suggests that for |f{sub NL}|∼>30 the non-linear collapse of halos, as described by our barrier model, strongly deviates from that of Gaussian initial conditions. This is not surprising since the effect of non-linear gravitational processes may be altered by initially large NG. Hence, in the lack of prior theoretical knowledge, halo collapse model parameters should be included in statistical halo mass function data analysis which aim to constrain the signature of primordial NG.

79 ASTRONOMY AND ASTROPHYSICS↗

Angular Schmidt modes in spontaneous parametric down-conversion

We report a proof-of-principle experiment demonstrating that appropriately chosen set of Hermite-Gaussian modes constitutes a Schmidt decomposition for transverse momentum states of biphotons generated in the process of spontaneous parametric down-conversion. We experimentally realize projective measurements in the Schmidt basis and observe correlations between appropriate pairs of modes. We perform tomographical state reconstruction in the Schmidt basis, by direct measurement of single-photon density matrix eigenvalues.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Development of Thermal Tomography Imaging System for In-Service Nondestructive Evaluation of Additively Manufactured Components

Additive manufacturing (AM) is an emerging method for cost-efficient fabrication of nuclear reactor parts. AM of metallic structures for nuclear energy applications is currently based on laser powder bed fusion (LPBF) process, which can introduce internal material flaws, such as pores and anisotropy. Integrity of AM structures needs to be evaluated nondestructively because material flaws could lead to premature failures due to exposure to high temperature, radiation and corrosive environment in a nuclear reactor. Thermal tomography (TT) provides a capability for non-destructive evaluation of sub-surface defects in arbitrary size structures. We investigate TT of AM stainless steel 316L specimen with imprinted internal porosity defects using relatively low-cost, small form factor infrared (IR) camera based on uncooled micro bolometer detector. Sparse coding related K-means singular value decomposition (SVD) machine learning, image processing algorithms are developed to improve quality of TT images through removal of Additive white Gaussian noise without blurring the images. Following initial qualification of an AM component for deployment in a nuclear reactor, a compact TT can also be used for in-service nondestructive evaluation (NDE). With capability to perform in-service NDE of the AM component lifecycle, TT data can be used for development of a component digital twin. MIT research reactor (MITR) has been identified as a representative environment to perform investigation of feasibility of in-service NDE during shutdown. Preliminary study of radiation level in the MITR viewport, with geometric parameters potentially compatible with compact TT system, has been conducted.

36 MATERIALS SCIENCE↗

Pulsed Thermal Tomography Nondestructive Examination of Additively Manufactured Reactor Materials (Second Annual Progress Report)

Additive manufacturing (AM) is an emerging method for cost-efficient fabrication of nuclear reactor parts. AM of metallic structures for nuclear energy applications is currently based on laser powder bed fusion (LPBF) process, which can introduce internal material flaws, such as pores and anisotropy. Integrity of AM structures needs to be evaluated nondestructively because material flaws could lead to premature failures due to exposure to high temperature, radiation and corrosive environment in a nuclear reactor. Thermal tomography (TT) provides a capability for non-destructive evaluation of sub-surface defects in arbitrary size structures. Thermal tomography is a computational method for heat diffusion-based imaging of solids, which provides 3D visualization of data from flash thermography measurements. We investigate thermal tomography imaging and nondestructive evaluation of stainless steel and nickel super alloy metallic structures produced with laser powder bed fusion (LPBF) additive manufacturing (AM) process. Metallic structures produced with LPBF contain defects, and there are limited capabilities to evaluate these structures non-destructively. Thermal tomography reconstruction of 3D apparent spatial effusivity provides information about AM structure geometry and internal material flaws. We study performance of thermal tomography in imaging of metallic structures through COMSOL computer simulations of transient heat transfer, and through reconstruction of data obtained from experimental measurements. Reconstruction of internal defects is investigated using a stainless steel 316L specimen with flat bottom hole (FBH) indentations, and Inconel 718 plate produced with laser powder bed fusion (LPBF) method, which contains imprinted hemispherical shape low density regions containing non-sintered metallic powder. The FBH’s have the same sizes as the imprinted defects in the LPBF specimens, but offer better imaging contrast. Thermal tomography reconstructions provide visualizations of internal defects, and allow for estimation of their sizes and locations. Detection sensitivity of TT is limited by noises. We investigate separation of signal from noise in thermography images using several machine learning (ML) methods, including new spatio-temporal blind source separation (STBSS) and spatio-temporal sparse dictionary learning (STSDL) methods. Performance of the ML methods is benchmarked using thermography data obtained from imaging stainless steel 316L and Inconel 718 specimens produced LPBF method with imprinted calibrated porosity defects. The ML methods are ranked by F-score and execution runtime. Finally, we investigate TT of AM stainless steel 316L specimen with imprinted internal porosity defects using relatively low-cost, small form factor infrared (IR) camera based on uncooled micro bolometer detector. Sparse coding related K-means singular value decomposition (SVD) machine learning, image processing algorithms are developed to improve quality of TT images through removal of Additive white Gaussian noise without blurring the images. Following initial qualification of an AM component for deployment in a nuclear reactor, a compact TT can also be used for in-service nondestructive evaluation (NDE). With capability to perform in-service NDE of the AM component lifecycle, TT data can be used for development of a component digital twin.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Digital simulation of continuous error models with application to an instrument landing system error

A digital simulation of the continuous error of the localized beam of a conventional instrument landing system is discussed. The digital simulation was developed during the analysis of space shuttle navigation capabilities. A discrete mathematical model for use on a digital computer is described. The model generates an output random sequence which is equivalent, for simulation purposes, to the desired random process. The model is a system of difference equations driven by a zero-mean Gaussian random sequence.

Merrick, R. B.↗

A co-designed equalization, modulation, and coding scheme

The commercial impact and technical success of Trellis Coded Modulation seems to illustrate that, if Shannon's capacity is going to be neared, the modulation and coding of an analogue signal ought to be viewed as an integrated process. More recent work has focused on going beyond the gains obtained for Average White Gaussian Noise and has tried to combine the coding/modulation with adaptive equalization. The motive is to gain similar advances on less perfect or idealized channels.

Peile, Robert E.↗

X-ray light curves of active galactic nuclei are phase incoherent

We compute the Fourier phase spectra for the light curves of five low-luminosity active galactic nuclei observed by EXOSAT. There is no statistically significant phase coherence in any of them. This statement is equivalent, subject to a technical caveat, to a demonstration that their fluctuation statistics are Gaussian. Models in which the X-ray output is controlled wholly by a unitary process undergoing a nonlinear limit cycle are therefore ruled out, while models with either a large number of randomly excited independent oscillation modes or nonlinearly interacting spatially dependent oscillations are favored. We also demonstrate how the degree of phase coherence in light curve fluctuations influences the application of causality bounds on internal length scales.

Krolik, Julian↗

Galaxy Correlation Function and Local Density from Photometric Redshifts Using the Stochastic Order Redshift Technique (SORT)

The stochastic order redshift technique (SORT) is a simple, efficient, and robust method to improve cosmological redshift measurements. The method relies upon having a small (∼10 per cent) reference sample of high-quality redshifts. Within pencil- beam-like sub-volumes surrounding each galaxy, we use the precise dN/dz distribution of the reference sample to recover new redshifts and assign them one-to-one to galaxies such that the original rank order of redshifts is preserved. Preserving the rank order is motivated by the fact that random variables drawn from Gaussian probability density functions with different means but equal standard deviations satisfy stochastic ordering. This process is repeated for sub-volumes surrounding each galaxy in the survey. This results in every galaxy being assigned multiple ‘recovered’ redshifts from which a new redshift estimate is determined. An earlier paper applied SORT to a mock Sloan Digital Sky Survey at z 0.2 and accurately recovered the two-point correlation function (2PCF) on scales > 4 h−1Mpc. In this paper, we test the performance of SORT in surveys spanning the redshift range 0.75 < z < 2.25. We used two mock surveys extracted from the Small MultiDark–Planck and Bolshoi–Planck N-body simulations with dark matter haloes that were populated by the Santa Cruz semi-analytic model. We find that SORT overall improves redshift estimates, accurately recovers the redshift-space 2PCF ξ (s) on scales > 2.5 h−1Mpc, and provides improved local density estimates in regions of average or higher density, which may allow for improved understanding of how galaxy properties relate to their environments.

James Kakos↗

A new stationary gridline artifact suppression method based on the 2D discrete wavelet transform

Purpose: In digital x-ray radiography, an antiscatter grid is inserted between the patient and the image receptor to reduce scattered radiation. If the antiscatter grid is used in a stationary way, gridline artifacts will appear in the final image. In most of the gridline removal image processing methods, the useful information with spatial frequencies close to that of the gridline is usually lost or degraded. In this study, a new stationary gridline suppression method is designed to preserve more of the useful information. Methods: The method is as follows. The input image is first recursively decomposed into several smaller subimages using a multiscale 2D discrete wavelet transform. The decomposition process stops when the gridline signal is found to be greater than a threshold in one or several of these subimages using a gridline detection module. An automatic Gaussian band-stop filter is then applied to the detected subimages to remove the gridline signal. Finally, the restored image is achieved using the corresponding 2D inverse discrete wavelet transform. Results: The processed images show that the proposed method can remove the gridline signal efficiently while maintaining the image details. The spectra of a 1D Fourier transform of the processed images demonstrate that, compared with some existing gridline removal methods, the proposed method has better information preservation after the removal of the gridline artifacts. Additionally, the performance speed is relatively high. Conclusions: The experimental results demonstrate the efficiency of the proposed method. Compared with some existing gridline removal methods, the proposed method can preserve more information within an acceptable execution time.

60 APPLIED LIFE SCIENCES↗

Coupled wave-packets for non-adiabatic molecular dynamics: a generalization of Gaussian wave-packet dynamics to multiple potential energy surfaces

Accurate simulation of the non-adiabatic dynamics of molecules in excited electronic states is key to understanding molecular photo-physical processes. Here we present a novel method, based on a semiclassical approximation, that is as efficient as the commonly used mean field Ehrenfest or ad hoc surface hopping methods and properly accounts for interference and decoherence effects. This novel method is an extension of Heller's thawed Gaussian wave-packet dynamics that includes coupling between potential energy surfaces. By studying several standard test problems we demonstrate that the accuracy of the method can be systematically improved while maintaining high efficiency. The method is suitable for investigating the role of quantum coherence in the non-adiabatic dynamics of many-atom molecules.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Using support vector machines for anomalous change detonation

We cast anomalous change detection as a binary classification problem, and use a support vector machine (SVM) to build a detector that does not depend on assumptions about the underlying data distribution. To speed up the computation, our SVM is implemented, in part, on a graphical processing unit. Results on real and simulated anomalous changes are used to compare performance to algorithms which effectively assume a Gaussian distribution. In this paper, we investigate the use of support vector machines (SVMs) with radial basis kernels for finding anomalous changes. Compared to typical applications of SVMs, we are operating in a regime of very low false alarm rate. This means that even for relatively large training sets, the data are quite meager in the regime of operational interest. This drives us to use larger training sets, which in turn places more of a computational burden on the SVM. We initially considered three different approaches to to address the need to work in the very low false alarm rate regime. The first is a standard SVM which is trained at one threshold (where more reliable estimates of false alarm rates are possible) and then re-thresholded for the low false alarm rate regime. The second uses the same thresholding approach, but employs a so-called least squares SVM; here a quadratic (instead of a hinge-based) loss function is employed, and for this model, there are good theoretical arguments in favor of adjusting the threshold in a straightforward manner. The third approach employs a weighted support vector machine, where the weights for the two types of errors (false alarm and missed detection) are automatically adjusted to achieve the desired false alarm rate. We have found in previous experiments (not shown here) that the first two types can in some cases work well, while in other cases they do not. This renders both approaches unreliable for automated change detection. By contrast, the third approach reliably produces good results, but at the cost of larger computational requirements caused by the need to estimate very small false alarm rates. To address these computational requirements, we employ a recently developed in-house solver for SVMs that is significantly faster than freely available standard solvers. But these computational issues are secondary to the larger question: do kernelized solutions provide better performance, in terms of detection rates and false alarm rates, than more traditional methods for change detection that effectively assume Gaussian data distributions? To this end, we will compare ROC curves obtained from the SVM with those from chronochrome, covariance equalization, and hyperbolic anomalous change detection.

97 MATHEMATICS AND COMPUTING↗

Determination of Cross-Sectional Area of Focused Picosecond Gaussian Laser Beam

Measurement of the waist diameter of a focused Gaussian-beam at the 1/e(sup 2) intensity, also referred to as spot size, is key to determining the fluence in laser processing experiments. Spot size measurements are also helpful to calculate the threshold energy and threshold fluence of a given material. This work reports an application of a conventional method, by analyzing single laser ablated spots for different laser pulse energies, to determine the cross-sectional area of a focused Gaussian-beam, which has a nominal pulse width of approx. 10 ps. Polished tungsten was used as the target material, due to its low surface roughness and low ablation threshold, to measure the beam waist diameter. From the ablative spot measurements, the ablation threshold fluence of the tungsten substrate was also calculated.

Ledesma, Rodolfo↗

SUPERPOSITION OF POLYTROPES IN THE INNER HELIOSHEATH

This paper presents a possible generalization of the equation of state and Bernoulli's integral when a superposition of polytropic processes applies in space and astrophysical plasmas. The theory of polytropic thermodynamic processes for a fixed polytropic index is extended for a superposition of polytropic indices. In general, the superposition may be described by any distribution of polytropic indices, but emphasis is placed on a Gaussian distribution. The polytropic density–temperature relation has been used in numerous analyses of space plasma data. This linear relation on a log–log scale is now generalized to a concave-downward parabola that is able to describe the observations better. The model of the Gaussian superposition of polytropes is successfully applied in the proton plasma of the inner heliosheath. The estimated mean polytropic index is near zero, indicating the dominance of isobaric thermodynamic processes in the sheath, similar to other previously published analyses. By computing Bernoulli's integral and applying its conservation along the equator of the inner heliosheath, the magnetic field in the inner heliosheath is estimated, B ∼ 2.29 ± 0.16 μG. The constructed normalized histogram of the values of the magnetic field is similar to that derived from a different method that uses the concept of large-scale quantization, bringing incredible insights to this novel theory.

79 ASTRONOMY AND ASTROPHYSICS↗

Exploring the energy landscape of RBMs: reciprocal space insights into bosons, hierarchical learning and symmetry breaking

Deep generative models have become ubiquitous due to their ability to learn and sample from complex distributions. Despite the proliferation of various frameworks, the relationships among these models remain largely unexplored, a gap that hinders the development of a unified theory of AI learning. In this work, we address two central challenges: clarifying the connections between different deep generative models and deepening our understanding of their learning mechanisms. We focus on Restricted Boltzmann Machines (RBMs), a class of generative models known for their universal approximation capabilities for discrete distributions. By introducing a reciprocal space formulation for RBMs, we reveal a connection between these models, diffusion processes, and systems of coupled bosons. Our analysis shows that at initialization, the RBM operates at a saddle point, where the local curvature is determined by the singular values of the weight matrix, whose distribution follows the Marc̆enko-Pastur law and exhibits rotational symmetry. During training, this rotational symmetry is broken due to hierarchical learning, where different degrees of freedom progressively capture features at multiple levels of abstraction. This leads to a symmetry breaking in the energy landscape, reminiscent of Landau’s theory. This symmetry breaking in the energy landscape is characterized by the singular values and the weight matrix eigenvector matrix. We derive the corresponding free energy in a mean-field approximation. We show that in the limit of infinite size RBM, the reciprocal variables are Gaussian distributed. Our findings indicate that in this regime, there will be some modes for which the diffusion process will not converge to the Boltzmann distribution. To illustrate our results, we trained replicas of RBMs with different hidden layer sizes using the MNIST dataset. Our findings not only bridge the gap between disparate generative frameworks but also shed light on the fundamental processes underpinning learning in deep generative models.

97 MATHEMATICS AND COMPUTING↗