Search NASA⌕ Search

SEARCH · Search NASA

Results for “Bayesian parameter estimation”

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 649 records · Page 36

Assimilation of satellite microwave observations over the rainbands of tropical cyclones

A novel Bayesian Monte Carlo integration (BMCI) technique was developed to retrieve geophysical variables from satellite microwave radiometer data in the presence of tropical cyclones. The BMCI technique includes three steps: generating a stochastic database, simulating satellite brightness temperatures using a radiative transfer model, and retrieving geophysical variables such as profiles of temperature, relative humidity, and cloud liquid and ice water content from real observations. The technique also provides uncertainty estimates for each retrieval and can output the error covariance matrix of selected parameters. The measurements from the Advanced Technology Microwave Sounder (ATMS) on board Suomi National Polar-Orbiting Partnership (Suomi NPP) and the Global Precipitation Measurement (GPM) Microwave Imager (GMI) were used as input. A new technique was developed to correct the ATMS and GMI observations for the beam-filling effect, which is due to small-scale variability of precipitation and clouds when compared with the instrument footprint and also the nonlinear relation between the brightness temperature and precipitation. In addition, the assimilation of the BMCI retrievals into the NASA GEOS model is discussed for Hurricane Maria. The results show that assimilating the BMCI retrievals can influence the dynamical features of the cyclone, including a stronger warm core, a symmetric eye, and vertically aligned wind columns. Two possible factors that may limit the impact of the BMCI retrievals include 1) the resolution of the model (about 25 km), which was too coarse to show the potential of the BMCI data in improving the representation of tropical storms in the model forecast, and 2) the data assimilation system not being able to consider vertically correlated observation errors.

Isaac Moradi↗

A POSSIBLE BINARY SYSTEM OF A STELLAR REMNANT IN THE HIGH-MAGNIFICATION GRAVITATIONAL MICROLENSING EVENT OGLE-2007-BLG-514

We report the extremely high-magnification (A > 1000) binary microlensing event OGLE-2007-BLG-514. We obtained good coverage around the double peak structure in the light curve via follow-up observations from different observatories. The binary lens model that includes the effects of parallax (known orbital motion of the Earth) and orbital motion of the lens yields a binary lens mass ratio of q = 0.321 {+-} 0.007 and a projected separation of s = 0.072 {+-} 0.001 in units of the Einstein radius. The parallax parameters allow us to determine the lens distance D{sub L} = 3.11 {+-} 0.39 kpc and total mass M{sub L} = 1.40 {+-} 0.18 M{sub Sun }; this leads to the primary and secondary components having masses of M{sub 1} = 1.06 {+-} 0.13 M{sub Sun} and M{sub 2} = 0.34 {+-} 0.04 M{sub Sun }, respectively. The parallax model indicates that the binary lens system is likely constructed by the main-sequence stars. On the other hand, we used a Bayesian analysis to estimate probability distributions by the model that includes the effects of xallarap (possible orbital motion of the source around a companion) and parallax (q = 0.270 {+-} 0.005, s = 0.083 {+-} 0.001). The primary component of the binary lens is relatively massive, with M{sub 1} = 0.9{sup +4.6}{sub -0.3} M{sub Sun} and it is at a distance of D{sub L} = 2.6{sup +3.8}{sub -0.9} kpc. Given the secure mass ratio measurement, the companion mass is therefore M{sub 2} = 0.2{sup +1.2}{sub -0.1} M{sub Sun }. The xallarap model implies that the primary lens is likely a stellar remnant, such as a white dwarf, a neutron star, or a black hole.

79 ASTRONOMY AND ASTROPHYSICS↗

Enhancing Gaussian Process Surrogates for Optimization and Posterior Approximation via Random Exploration

This paper proposes novel noise-free Bayesian optimization strategies that rely on a random exploration step to enhance the accuracy of Gaussian process surrogate models. The new algorithms retain the ease of implementation of the classical GP-UCB algorithm, but the additional random exploration step accelerates their convergence, nearly achieving the optimal convergence rate. Furthermore, to facilitate Bayesian inference with intractable likelihoods, we propose to utilize optimization iterates for maximum a posteriori estimation to build a Gaussian process surrogate model for the unnormalized log-posterior density. We provide bounds for the Hellinger distance between the true and the approximate posterior distributions in terms of the number of design points. We demonstrate the effectiveness of our Bayesian optimization algorithms in nonconvex benchmark objective functions, in a machine learning hyperparameter tuning problem, and in a black-box engineering design problem. The effectiveness of our posterior approximation approach is demonstrated in two Bayesian inference problems for parameters of dynamical systems.

Bayesian inference↗

Bayesian Design of Experiments for Logistic Regression to Evaluate Multiple Forensic Algorithms

When evaluating the performance of several forensic classification algorithms, it is desirable to construct a design that considers a variety of performance levels for each of the algorithms. We describe a strategy to use Bayesian design of experiments with multiple prior estimates to capture anticipated performance. Our goal is to characterize results from the different classification algorithms as a function of multiple explanatory variables and use this to choose a design about which units to test. Bayesian design of experiments has been successful for generalized linear models, including logistic regression models. Here, we develop methodology for the case where there are several potentially nonoverlapping priors for anticipated performance under consideration. The weighted priors method performs well for a broad range of true underlying model parameter choices and is more robust when compared to other candidate design choices. Additionally, we show how this can be applied in the multivariate input case and provide some useful summary measures. The shared information plot is used to evaluate design point allocation, and the D-value difference plot allows for the comparison of design performance across multiple potential parameter values in higher dimensions. Here, we illustrate the methods with several examples.

97 MATHEMATICS AND COMPUTING↗

Thoughts on Analyzing Neutron Multiplicity Data

Bayesian methods offer many advantages for analyzing neutron multiplicity data. Ideally one would make nonparametric estimates of the the posterior distributions, the probabilities - given the data - for unknowns to take on particular values. Once one has the posterior, uncertainty quantification is very natural. We describe how different ways of formulating the statistical theory of fission chains are mathematically equivalent. When confronted with actual data and the burden of finite statistics, different methods of expressing the theory offer advantages for understanding the underlying source parameters from an information theory standpoint.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Inferring Type II-P Supernova Progenitor Masses from Plateau Luminosities

Abstract Connecting observations of core-collapse supernova explosions to the properties of their massive star progenitors is a long-sought, and challenging, goal of supernova science. Recently, Barker et al. presented bolometric light curves for a landscape of progenitors from spherically symmetric neutrino-driven core-collapse supernova (CCSN) simulations using an effective model. They find a tight relationship between the plateau luminosity of the Type II-P CCSN light curve and the terminal iron-core mass of the progenitor. Remarkably, this allows us to constrain progenitor properties with photometry alone. We analyze a large observational sample of Type II-P CCSN light curves and estimate a distribution of iron-core masses using the relationship of Barker et al. The inferred distribution matches extremely well with the distribution of iron-core masses from stellar evolutionary models and namely, contains high-mass iron cores that suggest contributions from very massive progenitors in the observational data. We use this distribution of iron-core masses to infer minimum and maximum masses of progenitors in the observational data. Using Bayesian inference methods to locate optimal initial mass function parameters, we find M min = 9.8 − 0.27 + 0.37 and M max = 24.0 − 1.9 + 3.9 solar masses for the observational data.

79 ASTRONOMY AND ASTROPHYSICS↗

A computer program for estimation from incomplete multinomial data

Coding is given for maximum likelihood and Bayesian estimation of the vector p of multinomial cell probabilities from incomplete data. Also included is coding to calculate and approximate elements of the posterior mean and covariance matrices. The program is written in FORTRAN 4 language for the Control Data CYBER 170 series digital computer system with network operating system (NOS) 1.1. The program requires approximately 44000 octal locations of core storage. A typical case requires from 72 seconds to 92 seconds on CYBER 175 depending on the value of the prior parameter.

Credeur, K. R.↗

Taylor approximation variance reduction for approximation errors in PDE-constrained Bayesian inverse problems

In numerous applications, surrogate models are used as a replacement for accurate parameter-to-observable mappings when solving large-scale inverse problems governed by partial differential equations (PDEs). The surrogate model may be a computationally cheaper alternative to the accurate parameter-to-observable mappings and/or may ignore additional unknowns or sources of uncertainty. The Bayesian approximation error (BAE) approach provides a means to account for the induced uncertainties and approximation errors, i.e. the errors between the accurate parameter-to-observable mapping and the surrogate. The statistics of these errors are, however, in general unknown a priori, and are thus calculated using Monte Carlo sampling. Although the sampling is typically carried out offline, i.e. before considering the data, the process can still represent a computational bottleneck. In this work, we develop a scalable computational approach for reducing the costs associated with the sampling stage of the BAE approach. Specifically, we consider the Taylor expansion of the accurate and surrogate forward models with respect to the uncertain parameter fields either as a control variate for variance reduction or as a means to directly and efficiently approximate the mean and covariance of the approximation errors. We propose efficient methods for evaluating the expressions for the mean and covariance of the Taylor approximations based on linear(-ized) PDE solves. Furthermore, the proposed approach is independent of the dimension of the uncertain parameter, depending instead on the intrinsic dimension of the data, ensuring scalability to high-dimensional problems. The potential benefits of the proposed approach are demonstrated for two high-dimensional inverse problems governed by PDE examples, namely for the estimation of a distributed Robin boundary coefficient in a linear diffusion problem, and for a coefficient estimation problem governed by a nonlinear diffusion problem.

Bayesian approximation error↗

A Physical Model to Estimate Snowfall over Land using AMSU-B Observations

In this study, we present an improved physical model to retrieve snowfall rate over land using brightness temperature observations from the National Oceanic and Atmospheric Administration's (NOAA) Advanced Microwave Sounder Unit-B (AMSU-B) at 89 GHz, 150 GHz, 183.3 +/- 1 GHz, 183.3 +/- 3 GHz, and 183.3 +/- 7 GHz. The retrieval model is applied to the New England blizzard of March 5, 2001 which deposited about 75 cm of snow over much of Vermont, New Hampshire, and northern New York. In this improved physical model, prior retrieval assumptions about snowflake shape, particle size distributions, environmental conditions, and optimization methodology have been updated. Here, single scattering parameters for snow particles are calculated with the Discrete-Dipole Approximation (DDA) method instead of assuming spherical shapes. Five different snow particle models (hexagonal columns, hexagonal plates, and three different kinds of aggregates) are considered. Snow particle size distributions are assumed to vary with air temperature and to follow aircraft measurements described by previous studies. Brightness temperatures at AMSU-B frequencies for the New England blizzard are calculated using these DDA calculated single scattering parameters and particle size distributions. The vertical profiles of pressure, temperature, relative humidity and hydrometeors are provided by MM5 model simulations. These profiles are treated as the a priori data base in the Bayesian retrieval algorithm. In algorithm applications to the blizzard data, calculated brightness temperatures associated with selected database profiles agree with AMSU-B observations to within about +/- 5 K at all five frequencies. Retrieved snowfall rates compare favorably with the near-concurrent National Weather Service (NWS) radar reflectivity measurements. The relationships between the NWS radar measured reflectivities Z(sub e) and retrieved snowfall rate R for a given snow particle model are derived by a histogram matching technique. All of these Z(sub e)-R relationships fall in the range of previously established Z(sub e)-R relationships for snowfall. This suggests that the current physical model developed in this study can reliably estimate the snowfall rate over land using the AMSU-B measured brightness temperatures.

Kim, Min-Jeong↗

Sparse-grid, reduced-basis Bayesian inversion: Nonaffine-parametric nonlinear equations

We extend the reduced basis (RB) accelerated Bayesian inversion methods for affine-parametric, linear operator equations which are considered in [16,17] to non-affine, nonlinear parametric operator equations. We generalize the analysis of sparsity of parametric forward solution maps in [20] and of Bayesian inversion in [48,49] to the fully discrete setting, including Petrov–Galerkin high-fidelity (“HiFi”) discretization of the forward maps. We develop adaptive, stochastic collocation based reduction methods for the efficient computation of reduced bases on the parametric solution manifold. The nonaffinity and nonlinearity with respect to (w.r.t.) the distributed, uncertain parameters and the unknown solution is collocated; specifically, by the so-called Empirical Interpolation Method (EIM). For the corresponding Bayesian inversion problems, computational efficiency is enhanced in two ways: first, expectations w.r.t. the posterior are computed by adaptive quadratures with dimension-independent convergence rates proposed in [49]; the present work generalizes [49] to account for the impact of the PG discretization in the forward maps on the convergence rates of the Quantities of Interest (QoI for short). Second, we propose to perform the Bayesian estimation only w.r.t. a parsimonious, RB approximation of the posterior density. Based on the approximation results in [49], the infinite-dimensional parametric, deterministic forward map and operator admit N-term RB and EIM approximations which converge at rates which depend only on the sparsity of the parametric forward map. In several numerical experiments, the proposed algorithms exhibit dimension-independent convergence rates which equal, at least, the currently known rate estimates for N-term approximation. We propose to accelerate Bayesian estimation by first offline construction of reduced basis surrogates of the Bayesian posterior density. The parsimonious surrogates can then be employed for online data assimilation and for Bayesian estimation. They also open a perspective for optimal experimental design.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Towards optimal extraction of cosmological information from nonlinear data

One of the main unsolved problems of cosmology is how to maximize the extraction of information from nonlinear data. If the data are nonlinear the usual approach is to employ a sequence of statistics (N-point statistics, counting statistics of clusters, density peaks or voids etc.), along with the corresponding covariance matrices. However, this approach is computationally prohibitive and has not been shown to be exhaustive in terms of information content. Here we instead develop a hierarchical Bayesian approach, expanding the likelihood around the maximum posterior of linear modes, which we solve for using optimization methods. By integrating out the modes using perturbative expansion of the likelihood we construct an initial power spectrum estimator, which for a fixed forward model contains all the cosmological information if the initial modes are gaussian distributed. We develop a method to construct the window and covariance matrix such that the estimator is explicitly unbiased and nearly optimal. We then generalize the method to include the forward model parameters, including cosmological and nuisance parameters, and primordial non-gaussianity. We apply the method in the simplified context of nonlinear structure formation, using either simplified 2-LPT dynamics or N-body simulations as the nonlinear mapping between linear and nonlinear density, and 2-LPT dynamics in the optimization steps used to reconstruct the initial density modes. We demonstrate that the method gives an unbiased estimator of the initial power spectrum, providing among other a near optimal reconstruction of linear baryonic acoustic oscillations.

79 ASTRONOMY AND ASTROPHYSICS↗

Bayesian Spectral Moment Estimation and Uncertainty Quantification

We present a Bayesian spectral fitting method developed for spectroscopic data analysis, particularly (but not solely) in the context of fusion energy research. The presented techniques are particularly valuable to estimating moments and corresponding uncertainties whenever the spectra result from line-integrated measurements in nonuniform plasmas, for which the approximation of atomic line shapes being ideal Gaussians gives poor estimates. We decompose multiple, potentially overlapping spectral lines into a sum of Gauss-Hermite polynomials, whose properties allow efficient truncation and uncertainty quantification, often with only three free parameters per atomic emission line. Tests with both synthetic and experimental data demonstrate the effectiveness and robustness where more standard nonlinear fitting routines may experience difficulties. A parallelized version of our implementation is publicly released under an open source license.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Updated Orbital Monitoring and Dynamical Masses for Nearby M-dwarf Binaries

Young M-type binaries are particularly useful for precise isochronal dating by taking advantage of their extended pre-main sequence evolution. Orbital monitoring of these low-mass objects becomes essential in constraining their fundamental properties, as dynamical masses can be extracted from their Keplerian motion. Here, we present the combined efforts of the AstraLux Large Multiplicity Survey, together with a filler sub-programme from the SpHere INfrared Exoplanet (SHINE) project and previously unpublished data from the FastCam lucky imaging camera at the Nordical Optical Telescope (NOT) and the NaCo instrument at the Very Large Telescope (VLT). Building on previous work, we use archival and new astrometric data to constrain orbital parameters for 20 M-type binaries. We identify that eight of the binaries have strong Bayesian probabilities and belong to known young moving groups (YMGs). We provide a first attempt at constraining orbital parameters for 14 of the binaries in our sample, with the remaining six having previously fitted orbits for which we provide additional astrometric data and updated Gaia parallaxes. The substantial orbital information built up here for four of the binaries allows for direct comparison between individual dynamical masses and theoretical masses from stellar evolutionary model isochrones, with an additional three binary systems with tentative individual dynamical mass estimates likely to be improved in the near future. We attained an overall agreement between the dynamical masses and the theoretical masses from the isochrones based on the assumed YMG age of the respective binary pair. The two systems with the best orbital constrains for which we obtained individual dynamical masses, J0728 and J2317, display higher dynamical masses than predicted by evolutionary models.

Astrometry / binaries↗

INVERTING COLOR-MAGNITUDE DIAGRAMS TO ACCESS PRECISE STAR CLUSTER PARAMETERS: A NEW WHITE DWARF AGE FOR THE HYADES

We have extended our Bayesian modeling of stellar clusters-which uses main-sequence stellar evolution models, a mapping between initial masses and white dwarf (WD) masses, WD cooling models, and WD atmospheres-to include binary stars, field stars, and two additional main-sequence stellar evolution models. As a critical test of our Bayesian modeling technique, we apply it to Hyades UBV photometry, with membership priors based on proper motions and radial velocities, where available. Under the assumption of a particular set of WD cooling models and atmosphere models, we estimate the age of the Hyades based on cooling WDs to be 648 {+-} 45 Myr, consistent with the best prior analysis of the cluster main-sequence turnoff (MSTO) age by Perryman et al. Since the faintest WDs have most likely evaporated from the Hyades, prior work provided only a lower limit to the cluster's WD age. Our result demonstrates the power of the bright WD technique for deriving ages and further demonstrates complete age consistency between WD cooling and MSTO ages for seven out of seven clusters analyzed to date, ranging from 150 Myr to 4 Gyr.

99 GENERAL AND MISCELLANEOUS↗

OGLE-2017-BLG-1186: First Application of Asteroseismology and Gaussian Processes to microlensing

We present the analysis of the event OGLE-2017-BLG-1186 from the 2017 Spitzer microlensing campaign. This is a remarkable microlensing event because its source is photometrically bright and variable, which makes it possible to perform an asteroseismic analysis using ground-based data. We find that the source star is an oscillating red giant with average timescale of ∼9 d. The asteroseismic analysis also provides us source properties including the source angular size (∼27 μas) and distance (∼11.5 kpc), which are essential for inferring the properties of the lens. When fitting the light curve, we test the feasibility of Gaussian processes (GPs) in handling the correlated noise caused by the variable source. We find that the parameters from the GP model are generally more loosely constrained than those from the traditional χ(exp 2) minimization method. We note that this event is the first microlensing system for which asteroseismology and GPs have been used to account for the variable source. With both finite-source effect and microlens parallax measured, we find that the lens is likely a ∼0.045 Mʘ brown dwarf at distance ∼9.0 kpc, or a ∼0.073 Mʘ ultracool dwarf at distance ∼9.8 kpc. Combining the estimated lens properties with a Bayesian analysis using a Galactic model, we find a ∼ 35 per cent probability for the lens to be a bulge object and ∼ 65 per cent to be a background disc object.

S.-S. Li↗

Uncertainty quantification in (α,n) neutron source calculations for an oxide matrix

Here we present a methodology to propagate nuclear data covariance information in neutron source calculations from (α,n) reactions. The approach is applied to estimate the uncertainty in the neutron generation rates for uranium oxide fuel types due to uncertainties on 1) 17,18 O( α,n) reaction cross sections and 2) uranium and oxygen stopping power cross sections. The procedure to generate reaction cross section covariance information is based on the Bayesian fitting method implemented in the R-matrix SAMMY code. The evaluation methodology uses the Reich-Moore approximation to fit the 17,18 O(α,n) reaction cross-sections in order to derive a set of resonance parameters and a related covariance matrix that is then used to calculate the energydependent cross section covariance matrix. The stopping power cross sections and related covariance information for uranium and oxygen were obtained by the fit of stopping power data in the -energy range of 1 keV up to 12 MeV. Cross section perturbation factors based on the covariance information relative to the evaluated 17,18 O( α,n) reaction cross sections, as well as uranium and oxygen stopping power cross sections, were used to generate a varied set of nuclear data libraries used in SOURCES4C and ORIGEN for inventory and source term calculations. The set of randomly perturbed output (α,n) source responses, provide the mean values and standard deviations of the calculated responses reflecting the uncertainties in nuclear data used in the calculations. Lastly, the results and related uncertainties are compared with experiment thick target (α,n) yields for uranium oxide.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Vacuum phase transition solves the H 0 tension

Taking the Planck cosmic microwave background data and the more direct Hubble constant measurement data as unaffected by systematic offsets, the values of the Hubble constant H 0 interpreted within the ΛCDM cosmological constant and cold dark matter cosmological model are in ~3.3σ tension. We show that the Parker vacuum metamorphosis (VM) model, physically motivated by quantum gravitational effects and with the same number of parameters as ΛCDM, can remove the H 0 tension and can give an improved fit to data (up to a mean Δχ 2 = -7.5). It also ameliorates tensions with weak lensing data and the high redshift Lyman alpha forest data. Considering Bayesian evidence, we found in the case of the Planck data set alone positive evidence for a VM model against a cosmological constant both in the six- and nine-parameter framework. When the R16 data set is also considered, we found a strong evidence for the VM model against a cosmological constant in nine-parameter space. Here we separately consider a scale-dependent scaling of the gravitational lensing amplitude, such as provided by modified gravity, neutrino mass, or cold dark energy, motivated by the somewhat different cosmological parameter estimates for low and high CMB multipoles. We find that no such scale dependence is preferred.

79 ASTRONOMY AND ASTROPHYSICS↗