Search NASA⌕ Search

SEARCH · Search NASA

Results for “Gaussian Process Models”

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 469 records · Page 26

Applying Constrained Bayesian Optimization to the Design of Critical Experiments

Often when planning a criticality experiment, many design configurations are iteratively investigated with a Monte Carlo transport code. The goal is that the experiment will be optimal with respect to some variable, like the fraction of fissions occurring at a certain energy range, while simultaneously being critical. Unfortunately, the Monte Carlo transport simulations are expensive, which can ultimately limit the number of configurations that can be explored. In this work, we present how Gaussian processes (GPs) can be used as a reduced-order model in a constrained Bayesian optimization (CBO) algorithm to design a criticality experiment. The GPs replace the Monte Carlo transport simulations that explore the design space. The CBO algorithm efficiently identifies new points in the design space to run the Monte Carlo transport code while respecting the criticality constraint. It does so in a manner that both improves the accuracy of the GP and finds the approximate global optimum. We demonstrate the performance of CBO with the design of a Thermal Epithermal eXperiment (TEX) for the criticality safety validation of nuclear waste models of the Hanford Tank Farm.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Radiation dose assessment in a 320-detector-row CT scanner used in cardiac imaging

Purpose: In the present era of cone-beam CT scanners, the use of the standardized CTDI{sub 100} as a surrogate of the idealized CTDI is strongly discouraged and, consequently, so should be the use of the dose-length product (DLP) as an estimate of the total energy imparted to the patient. However, the DLP is still widely used as a reference quantity to normalize the effective dose for a given scan protocol mainly because the CTDI{sub 100} is an easy-to-measure quantity. The aim of this article is therefore to describe a method for radiation dose assessment in large cone-beam single axial scans, which leads to a straightforward estimation of the total energy imparted to the patient. The authors developed a method accessible to all medical physicists and easy to implement in clinical practice in an attempt to update the bridge between CT dosimetry and the estimation of the effective dose. Methods: The authors used commercially available material and a simple mathematical model. The method described herein is based on the dosimetry paradigm introduced by the AAPM Task Group 111. It consists of measuring the dose profiles at the center and the periphery of a long body phantom with a commercial solid-state detector. A weighted dose profile is then calculated from these measurements. To calculate the CT dosimetric quantities analytically, a Gaussian function was fitted to the dose profile data. Furthermore, the Gaussian model has the power to condense the z-axis information of the dose profile in two parameters: The single-scan central dose, f(0), and the width of the profile, {sigma}. To check the energy dependence of the solid-state detector, the authors compared the dose profiles to measurements made with a small volume ion chamber. To validate the overall method, the authors compared the CTDI{sub 100} calculated analytically to the measurement made with a 100 mm pencil ion chamber. Results: For the central and weighted dose profiles, the authors found a good agreement between the measured dose profile data and the fitted Gaussian functions. The solid-state detector had no energy dependence--within the energy range of interest--and the analytical model succeeded in reproducing the absolute dose values obtained with the pencil ion chamber. For the case of large cone-beam single axial scans, the quantity that better characterizes the total energy imparted to the patient is the weighted dose profile integral (DPI{sub w}). The DPI{sub w} can be easily determined from the two parameters that define the Gaussian functions: f(0) and {sigma}. The authors found that the DLP underestimated the total energy imparted to the patient by more than 20%. The authors also found that the calculated CT dosimetric quantities were higher than those displayed on the scanner console. Conclusions: The authors described and validated a method to assess radiation dose in large cone-beam single axial scans. This method offers a simple and more accurate estimation of the total energy imparted to the patient, thus offering the possibility to update the bridge between CT dosimetry and the estimation of the effective dose for cone-beam CT examinations in radiology, nuclear medicine, and radiation therapy.

61 RADIATION PROTECTION AND DOSIMETRY↗

A Gaussian process guide for signal regression in magnetic fusion

Extracting reliable information from diagnostic data in tokamaks is critical for understanding, analyzing, and controlling the behavior of fusion plasmas and validating models describing that behavior. Recent interest within the fusion community has focused on the use of principled statistical methods, such as Gaussian process regression (GPR), to attempt to develop sharper, more reliable, and more rigorous tools for examining the complex observed behavior in these systems. While GPR is an enormously powerful tool, there is also the danger of drawing fragile, or inconsistent conclusions from naive GPR fits that are not driven by principled treatments. Here we review the fundamental concepts underlying GPR in a way that may be useful for broad-ranging applications in fusion science. We also revisit how GPR is developed for profile fitting in tokamaks. We examine various extensions and targeted modifications applicable to experimental observations in the edge of the DIII-D tokamak. Finally, we discuss best practices for applying GPR to fusion data.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Comparing emulation methods for a high‐resolution storm surge model

Abstract Realistic simulations of complex systems are fundamental for climate and environmental studies. Large computer systems are often not sufficient to run sophisticated computational models for large numbers of different input settings. Statistical surrogate models, or emulators, are key tools enabling fast exploration of the simulator input space. Gaussian processes have become standard for computer simulator emulation. However, they require careful implementation to scale appropriately, motivating alternative methods more recently introduced. We present a comparison study of surrogates of the Sea, Lake, and Overland Surges from Hurricanes (SLOSH) simulator—the simulator of choice for government agencies—using four emulation approaches: BASS; BART; SEPIA; and RobustGaSP. SEPIA and RobustGaSP use Gaussian processes, BASS implements adaptive splines, and BART is based on ensembles of regression trees. We describe the four models and compare them in terms of computation time and predictive metrics. These surrogates use proven and distinct methodologies, are available through accessible software, and quantify prediction uncertainty. Our data cover millions of response values. We find that SEPIA and RobustGaSP provide exceptional predictive power, but cannot scale to emulate experiments as large as the one considered in this paper as effectively as BASS and BART.

Hutchings, Grant↗

Sensor selection and tool wear prediction with data‐driven models for precision machining

Abstract Estimation of tool wear in precision machining is vital in the traditional subtractive machining industry to reduce processing cost, improve manufacturing efficiency and product quality. In this vein, fusion of time and frequency‐domain features of commonly sensed signals can provide an early indication of tool wear and improve its prediction accuracy for prognostics and health management. This paper presents a data‐driven methodology and a complete tool chain for the inference of precision machining tool wear from fused machine measurements, such as cutting force, power, audio and vibration signals, and quantify the usefulness of each measurement. Indicators of tool wear are extracted from time‐domain signal statistics, frequency‐domain analysis, and time‐frequency domain analysis. Correlation coefficients between the extracted features (indicators) and the tool wear are used to select the most informative features. Principal Component Analysis and Partial Least‐Squares are used to reduce the dimensionality of the feature space. Regression models, including linear regression, support vector regression, Decision tree regression, neural network regression and Gaussian process regression, are used to predict the tool wear using data from a Haas milling machine performing spiral boss face milling. The performance of the regression models based on subsets of sensors validates the preliminary estimates about the saliency of the sensors. The experimental results show that the proposed methods can predict the machine tool wear precisely, with readily available sensor measurements. Neural network and Gaussian process regression were able to achieve good estimates of tool wear at different machine operating conditions. The most informative signal in predicting tool wear was shown to be the vibration signal. Time‐frequency domain features were the most informative features among the combination of features of three domains. In addition, using partial least squares components extracted from the original features of signals led to higher prediction accuracy.

Han, Seulki↗

Tempered fractional calculus

Fractional derivatives and integrals are convolutions with a power law. Multiplying by an exponential factor leads to tempered fractional derivatives and integrals. Tempered fractional diffusion equations, where the usual second derivative in space is replaced by a tempered fractional derivative, govern the limits of random walk models with an exponentially tempered power law jump distribution. The limiting tempered stable probability densities exhibit semi-heavy tails, which are commonly observed in finance. Tempered power law waiting times lead to tempered fractional time derivatives, which have proven useful in geophysics. The tempered fractional derivative or integral of a Brownian motion, called a tempered fractional Brownian motion, can exhibit semi-long range dependence. The increments of this process, called tempered fractional Gaussian noise, provide a useful new stochastic model for wind speed data. A tempered fractional difference forms the basis for numerical methods to solve tempered fractional diffusion equations, and it also provides a useful new correlation model in time series.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Active learning of constitutive relation from mesoscopic dynamics for macroscopic modeling of non-Newtonian flows

We simulate complex fluids by means of an on-the-fly coupling of the bulk rheology to the underlying microstructure dynamics. In particular, a continuum model of polymeric fluids is constructed without a pre-specified constitutive relation, but instead it is actively learned from mesoscopic simulations where the dynamics of polymer chains is explicitly computed. To couple the bulk rheology of polymeric fluids and the microscale dynamics of polymer chains, the continuum approach (based on the finite volume method) provides the transient flow field as inputs for the (mesoscopic) dissipative particle dynamics (DPD), and in turn DPD returns an effective constitutive relation to close the continuum equations. In this multiscale modeling procedure, we employ an active learning strategy based on Gaussian process regression (GPR) to minimize the number of expensive DPD simulations, where adaptively selected DPD simulations are performed only as necessary. Numerical experiments are carried out for flow past a circular cylinder of a non-Newtonian fluid, modeled at the mesoscopic level by bead-spring chains. The results show that only five DPD simulations are required to achieve an effective closure of the continuum equations at Reynolds number RE = 10. Furthermore, when Re is increased to 100, only one additional DPD simulation is required for constructing an extended GPR-informed model closure. Compared to traditional message-passing multiscale approaches, applying an active learning scheme to multiscale modeling of non-Newtonian fluids can significantly increase the computational efficiency. Although the method demonstrated here obtains only a local viscosity from the polymer dynamics, it can be extended to other multiscale models of complex fluids whose macro-rheology is unknown.

Computer Science↗

Applying Gaussian Process Machine Learning and Modern Probabilistic Programming to Satellite Data to Infer CO 2 Emissions

Satellite data provides essential insights into the spatiotemporal distribution of CO 2 concentrations. However, many atmospheric inverse models fail to adequately incorporate the spatial and temporal correlations inherent in satellite observations and often lack rigorous methods for estimating parameters like spatial length scales. We introduce an inference model that processes the spatiotemporal covariance in satellite data and estimates hyperparameters such as covariance length scales. Our approach uses the Gaussian process (GP) machine learning (ML) and modern probabilistic programming languages (PPLs) to perform atmospheric inversions of emissions from satellite data. We develop a GP ML inversion system based on modern PPLs and the GEOS-Chem chemical transport model, simulating atmospheric CO 2 concentrations corresponding to the Orbiting Carbon Observatory-2/3 (OCO-2/3) data for July 2020. In our supervised learning framework, we treat the GEOS-Chem simulated data set as the target, with predictors derived by scaling the target with sector-specific factors hidden from the GP machine. Our results show that the GP model, combined with GPU-enabled PPLs, effectively retrieves true emission scaling factors and infers noise levels concealed within the data. This suggests that our method could be applied over larger areas with more complex covariance structures, enabling comprehensive analysis of the spatiotemporal patterns observed in OCO-2/3 and similar satellite data sets.

54 ENVIRONMENTAL SCIENCES↗

Data-Driven Multi-agent Deep Reinforcement Learning for Distribution System Decentralized Voltage Control with High Penetration of PVs

This paper proposes a novel model-free/data-driven centralized training and decentralized execution multi-agent deep reinforcement learning (MADRL) framework for distribution system voltage control with high penetration of PVs. The proposed MADRL can coordinate both the real and reactive power control of PVs with existing static var compensators and battery storage systems. Unlike the existing DRL-based voltage control methods, our proposed method does not rely on a system model during both the training and execution stages. This is achieved by developing a new interaction scheme between the surrogate modeling of the original system and the multi-agent soft actor critic (MASAC) MADRL algorithm. In particular, the sparse pseudo-Gaussian process with a few-shots of measurements is utilized to construct the surrogate model of the original environment, i.e., power flow model. This is a data-driven process and no model parameters are needed. Furthermore, the MASAC enabled MADRL allows to achieve better scalability by dividing the original system into different voltage control regions with the aid of real and reactive power sensitivities to voltage, where each region is treated as an agent. This also serves as the foundation for the centralized training and decentralized execution, thus significantly reducing the communication requirements as only local measurements are required for control. Comparative results with other alternatives on the IEEE 123-nodes and 342-nodes systems demonstrate the superiority of the proposed method.

14 SOLAR ENERGY↗

Lattice QCD estimates of thermal photon production from the QGP

Thermal photons produced in heavy-ion collision experiments are an important observable for understanding quark-gluon plasma (QGP). The thermal photon rate from the QGP at a given temperature can be calculated from the spectral function of the vector current correlator. Extraction of the spectral function from the lattice correlator is known to be an ill-conditioned problem, as there is no unique solution for a spectral function for a given lattice correlator with statistical errors. The vector current correlator, on the other hand, receives a large ultraviolet contribution from the vacuum, which makes the extraction of the thermal photon rate difficult from this channel. We therefore consider the difference between the transverse and longitudinal part of the spectral function, only capturing the thermal contribution to the current correlator, simplifying the reconstruction significantly. The lattice correlator is calculated for light quarks in quenched QCD at T = 470 MeV ( ∼ 1.5 T c ), as well as in 2 + 1 flavor QCD at T = 220 MeV ( ∼ 1.2 T p c ) with m π = 320 MeV . In order to quantify the nonperturbative effects, the lattice correlator is compared with the corresponding NLO + LPM LO estimate of correlator. The reconstruction of the spectral function is performed in several different frameworks, ranging from physics-informed models of the spectral function to more general models in the Backus-Gilbert method and Gaussian process regression. We find that the resulting photon rates agree within errors. Published by the American Physical Society 2024

Astronomy & Astrophysics↗

gp_blendclass_singleband

The code used for the data preprocessing, image simulation, and model training and analysis reported in the paper "Gaussian Process Classification for Galaxy Blend Identification in LSST" (arXiv:2107.09246).

Buchanan, JamesJ.↗

Analysis of stimulated Raman scattering using diode data at the Pecos experimental platform

In order to increase neutron yield in fusion experiments on the Magnetized Linear Inertial Fusion (MagLIF) platform, it is important to maximize the energy coupled to the fuel during the laser-preheat stage. However, laser-energy coupling is limited by laser—plasma instabilities (LPI). In this regard, the Pecos facility at Sandia National Laboratories uses the Z-beamlet laser to test and study the effects of LPI on MagLIF-relevant targets. In particular, stimulated Raman scattering (SRS) is measured in Pecos by using two photo-diodes and a near-beam imager. The measurements from the photo-diodes are processed using a synthetic spectrum based on a Gaussian model. With this relatively simple model, the mean wavelength and intensity of backscattered light can be deduced. Our measurements show similar trends as those given by time-resolved spectrometer data. Hence, this model provides a simple way to approximate time-resolved spectra for scattered light by SRS.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A qualitative assessment of a random process proposed as an atmospheric turbulence model

A random process is formed by the product of two Gaussian processes and the sum of that product with a third Gaussian process. The resulting total random process is interpreted as the sum of an amplitude modulated process and a slowly varying, random mean value. The properties of the process are examined, including an interpretation of the process in terms of the physical structure of atmospheric motions. The inclusion of the mean value variation gives an improved representation of the properties of atmospheric motions, since the resulting process can account for the differences in the statistical properties of atmospheric velocity components and their gradients. The application of the process to atmospheric turbulence problems, including the response of aircraft dynamic systems, is examined. The effects of the mean value variation upon aircraft loads are small in most cases, but can be important in the measurement and interpretation of atmospheric turbulence data.

Sidwell, K.↗

Statistical Orbit Determination using the Particle Filter for Incorporating Non-Gaussian Uncertainties

The tracking of space objects requires frequent and accurate monitoring for collision avoidance. As even collision events with very low probability are important, accurate prediction of collisions require the representation of the full probability density function (PDF) of the random orbit state. Through representing the full PDF of the orbit state for orbit maintenance and collision avoidance, we can take advantage of the statistical information present in the heavy tailed distributions, more accurately representing the orbit states with low probability. The classical methods of orbit determination (i.e. Kalman Filter and its derivatives) provide state estimates based on only the second moments of the state and measurement errors that are captured by assuming a Gaussian distribution. Although the measurement errors can be accurately assumed to have a Gaussian distribution, errors with a non-Gaussian distribution could arise during propagation between observations. Moreover, unmodeled dynamics in the orbit model could introduce non-Gaussian errors into the process noise. A Particle Filter (PF) is proposed as a nonlinear filtering technique that is capable of propagating and estimating a more complete representation of the state distribution as an accurate approximation of a full PDF. The PF uses Monte Carlo runs to generate particles that approximate the full PDF representation. The PF is applied in the estimation and propagation of a highly eccentric orbit and the results are compared to the Extended Kalman Filter and Splitting Gaussian Mixture algorithms to demonstrate its proficiency.

Mashiku, Alinda↗

NEAR-INFRARED EMISSION SPECTRUM OF WASP-103B USING HUBBLE SPACE TELESCOPE/WIDE FIELD CAMERA 3

We present here our observations and analysis of the dayside emission spectrum of the hot Jupiter WASP-103b. We observed WASP-103b during secondary eclipse using two visits of the Hubble Space Telescope with the G141 grism on Wide Field Camera 3 in spatial scan mode. We generated secondary eclipse light curves of the planet in both blended white-light and spectrally binned wavechannels from 1.1 to 1.7 μm and corrected the light curves for flux contamination from a nearby companion star. We modeled the detector systematics and secondary eclipse spectrum using Gaussian process regression and found that the near-IR emission spectrum of WASP-103b is featureless across the observed near-IR region to down to a sensitivity of 175 ppm, and shows a shallow slope toward the red. The atmosphere has a single brightness temperature of T{sub B}=2890 K across this wavelength range. This region of the spectrum is indistinguishable from isothermal, but may not manifest from a physically isothermal system, i.e., pseudo-isothermal. A solar-metallicity profile with a thermal inversion layer at 10{sup −2} bar fits the spectrum of WASP-103b with high confidence, as do an isothermal profile with solar metallicity and a monotonically decreasing atmosphere with C/O > 1. The data rule out a monotonically decreasing atmospheric profile with solar composition, and we rule out a low-metallicity decreasing profile as unphysical for this system. The pseudo-isothermal profile could be explained by a thermal inversion layer just above the layer probed by our observations, or by clouds or haze in the upper atmosphere. Transmission spectra at optical wavelengths would allow us to better distinguish between potential atmospheric models.

79 ASTRONOMY AND ASTROPHYSICS↗

Detection Limits of Low-mass, Long-period Exoplanets Using Gaussian Processes Applied to HARPS-N Solar Radial Velocities

Radial velocity (RV) searches for Earth-mass exoplanets in the habitable zone around Sun-like stars are limited by the effects of stellar variability on the host star. In particular, suppression of convective blueshift and brightness inhomogeneities due to photospheric faculae/plage and starspots are the dominant contribution to the variability of such stellar RVs. Gaussian process (GP) regression is a powerful tool for statistically modeling these quasi-periodic variations. We investigate the limits of this technique using 800 days of RVs from the solar telescope on the High Accuracy Radial velocity Planet Searcher for the Northern hemisphere (HARPS-N) spectrograph. These data provide a well-sampled time series of stellar RV variations. Into this data set, we inject Keplerian signals with periods between 100 and 500 days and amplitudes between 0.6 and 2.4 m s{sup −1}. We use GP regression to fit the resulting RVs and determine the statistical significance of recovered periods and amplitudes. We then generate synthetic RVs with the same covariance properties as the solar data to determine a lower bound on the observational baseline necessary to detect low-mass planets in Venus-like orbits around a Sun-like star. Our simulations show that discovering planets with a larger mass (∼0.5 m s{sup −1}) using current-generation spectrographs and GP regression will require more than 12 yr of densely sampled RV observations. Furthermore, even with a perfect model of stellar variability, discovering a true exo-Venus (∼0.1 m s{sup −1}) with current instruments would take over 15 yr. Therefore, next-generation spectrographs and better models of stellar variability are required for detection of such planets.

47 OTHER INSTRUMENTATION↗

Predicting concrete compressive strength using hybrid ensembling of surrogate machine learning models

This study aims to implement a hybrid ensemble surrogate machine learning technique in predicting the compressive strength (CS) of concrete, an important parameter used for durability design and service life prediction of concrete structures in civil engineering projects. For this purpose, an experimental database consisting of 1030 records has been compiled from the machine learning repository of the University of California, Irvine. The database was used to train and validate four conventional machine learning (CML) models, namely Artificial Neural Network (ANN), Linear and Non-Linear Multivariate Adaptive Regression Splines (MARS-L and MARS-C), Gaussian Process Regression (GPR), and Minimax Probability Machine Regression (MPMR). Subsequently, the predicted outputs of CML models were combined and trained using ANN to construct the Hybrid Ensemble Model (HENSM). It is observed that the proposed HENSM produces higher predictive accuracy compared to the CML models used in the present study. The predictive performance of all models for CS prediction was compared using the testing dataset and it is found that the HENSM model attained the highest predictive accuracy in both phases. Based on the experimental results, the newly constructed HENSM model is very potential to be a new alternative in handling the overfitting issues of CML models and hence, can be used to predict the concrete CS, including the design of less polluting and more sustainable concrete constructions.

36 MATERIALS SCIENCE↗