Search NASA⌕ Search

SEARCH · Search NASA

Results for “distributed inference”

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 271 records · Page 15

Dimensional Reduction for Sampled Priors and Application to Photometric Redshift Distributions

A typical Bayesian inference on the values of some parameters of interest q from some data D involves running a Markov Chain (MC) to sample from the posterior $p$($q$,$n$|$D$) $\propto$ $\mathcal{L}$($D$|$q$,$n$)$p$(q)$p$($n$), where n are some nuisance parameters with a separable prior. In some cases, the nuisance parameters are high-dimensional, and their prior p(n) is itself defined only by a set of samples that have been drawn from some other MC. The MC for the posterior will typically require evaluation of p(n) at arbitrary values of n, i.e., one needs to provide a density estimator over the full n space from the provided samples. But the high dimensionality of n hinders both the density estimation and the efficiency of the MC for the posterior. We describe a solution to this problem: a linear compression of the n space into a much lower-dimensional space u, which projects away directions in n space that cannot appreciably alter $\mathcal{L}$. The algorithm for doing so is a slight modification to principal components analysis, and is less restrictive on p(n) than other proposed solutions to this issue. We demonstrate this “mode projection” technique using the analysis of 2-point correlation functions of weak lensing fields and galaxy density in the Dark Energy Survey, where n is a binned representation of the redshift distribution n(z) of the galaxies.

79 ASTRONOMY AND ASTROPHYSICS↗

A Scalable Gaussian Process Approach to Shear Mapping with MuyGPs

Analysis of cosmic shear is an integral part of understanding structure growth across cosmic time, which in turn provides us with information about the nature of dark energy. Conventional methods generate shear maps from which we can infer the matter distribution in the universe. Current methods (e.g., Kaiser–Squires inversion) for generating these maps, however, are tricky to implement and can introduce bias. Recent alternatives construct a spatial process prior for the lensing potential, which allows for inference of the convergence and shear parameters given lensing shear measurements. Realizing these spatial processes, however, scales cubically in the number of observations—an unacceptable expense as near-term surveys expect billions of correlated measurements. Therefore, we present a linearly scaling shear map construction alternative using a scalable Gaussian process prior called MuyGPs. MuyGPs avoids cubic scaling by conditioning interpolation on only nearest neighbors and fits hyperparameters using batched leave-one-out cross-validation. This work is the first step toward a full, scalable mass mapping method. We work in a simplified regime where we validate our method by interpolating and analyzing maps given noisy point-estimate data from all three shear fields, taken from a suite of N -body ray-tracing simulations. We also show that we can perform these operations at the scale of billions of galaxies on high-performance computing platforms.

79 ASTRONOMY AND ASTROPHYSICS↗

The SINS/zC-SINF survey of z ∼ 2 galaxy kinematics: Evidence for gravitational quenching

As part of the SINS/zC-SINF surveys of high-z galaxy kinematics, we derive the radial distributions of Hα surface brightness, stellar mass surface density, and dynamical mass at ∼2 kpc resolution in 19 z ∼ 2 star-forming disks with deep SINFONI adaptive optics spectroscopy at the ESO Very Large Telescope. From these data we infer the radial distribution of the Toomre Q-parameter for these main-sequence star-forming galaxies (SFGs), covering almost two decades of stellar mass (10{sup 9.6}-10{sup 11.5} M {sub ☉}). In more than half of our SFGs, the Hα distributions cannot be fit by a centrally peaked distribution, such as an exponential, but are better described by a ring, or the combination of a ring and an exponential. At the same time the kinematic data indicate the presence of a mass distribution more centrally concentrated than a single exponential distribution for 5 of the 19 galaxies. The resulting Q-distributions are centrally peaked for all, and significantly exceed unity there for three-quarters of the SFGs. The occurrence of Hα rings and of large nuclear Q-values appears to be more common for the more massive SFGs. While our sample is small and biased to larger SFGs, and there remain uncertainties and caveats, our observations are consistent with a scenario in which cloud fragmentation and global star formation are secularly suppressed in gas-rich high-z disks from the inside out, as the central stellar mass density of the disks grows.

79 ASTRONOMY AND ASTROPHYSICS↗

Improving unfolding and systematic uncertainty estimation using generative diffusion networks (Final Technical Report)

This final technical report summarizes the key accomplishments on the unfolding using diffusion model project, a DOE award received by PI Pierre-Hugues Beauchemin at Tufts University. This project main goal was to investigate the potential of diffusion models for unfolding experimental High Energy Physics data from detector effects while controlling systematics uncertainties. The project accomplished its goals by completing the following objectives: 1) Performing an object-by-object, event-by-event unfolding of various kinematic distributions reconstructed from detector data in HEP in a way that keeps correlations between unfolded observables while demonstrating competitive performance compared to standard algorithms used in the field; 2) Address the generalization problem by developing an unfolding algorithm capable to correctly infer the underlying distributions of observables and processes never seen before, while controlling the dominant theoretical uncertainties affecting the process, therefore increasing the effectiveness, the precision, and the applicability of the developed algorithm; 3) Understand the theoretical foundations between the developed algorithm so to extend it to applications beyond experimental HEP, for broader benefits to the society. This report provides an overview of the accomplishments related to each of these key objectives.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Assessing correlated truncation errors in modern nucleon-nucleon potentials

We test the BUQEYE model of correlated effective field theory (EFT) truncation errors on Reinert, Krebs, and Epelbaum's semilocal momentum-space implementation of the chiral EFT (𝜒⁢EFT ) expansion of the nucleon-nucleon (NN) potential. This Bayesian model hypothesizes that dimensionless coefficient functions extracted from the order-by-order corrections to NN observables can be treated as draws from a Gaussian process (GP). We combine a variety of graphical and statistical diagnostics to assess when predicted observables have a 𝜒⁢EFT convergence pattern consistent with the hypothesized GP statistical model. Our conclusions are that, first, the BUQEYE model is generally applicable to the potential investigated here, which enables statistically principled estimates of the impact of higher EFT orders on observables. Second, parameters defining the extracted coefficients such as the expansion parameter 𝑄 must be well chosen for the coefficients to exhibit a regular convergence pattern—a property we exploit to obtain posterior distributions for such quantities. Third, the assumption of GP stationarity across lab energy and scattering angle is not generally met; this necessitates adjustments in future work. We provide a workflow and interpretive guide for our analysis framework, and show what can be inferred about probability distributions for 𝑄, the EFT breakdown scale Λ 𝑏 , the scale associated with soft physics in the 𝜒⁢EFT potential 𝑚 eff , and the GP hyperparameters. All our results can be reproduced using a publicly available Jupyter notebook, which can be straightforwardly modified to analyze other 𝜒⁢EFT NN potentials.

Bayesian methods↗

Vertical distribution of aerosol extinction cross section and inference of aerosol imaginary index in the troposphere by lidar technique

The paper reports on vertical profiles of aerosol extinction and backscatter in the troposphere which were obtained from multi zenith angle lidar measurements. It is reported that a direct slant path solution was found to be not possible due to horizontal inhomogeneity of the atmosphere. Attention is given to the use of a regression analysis with respect to zenith angle for a layer integration of the angle dependent lidar equation in order to determine the optical thickness and aerosol extinction-to-backscatter ratio for defined atmospheric layers and the subsequent evaluation of cross-section profiles.

Spinhirne, J. D.↗

Inference of high-latitude ionization and conductivity from AE-C measurements of auroral electron fluxes

Electron fluxes in the kilovolt energy range measured by the AE-C satellite have been used to infer the latitudinal distribution of ionization at high latitudes in the altitude range 90 to 200 km. These distributions are compared with simultaneous measurements of electron density by the Chatanika incoherent scatter radar. The study was based on more than 120 electron density profiles obtained on three different nights at magnetic local times between 0200 and 0400. The calculated distributions agree very well with the altitude and latitude distributions measured by the radar. The only obvious difference was the presence in the radar data of additional ionization produced by solar illumination. A comparison was also conducted of the height-integrated electrical conductivities computed from the measured profiles and those inferred from the satellite measurements. For Hall and Pedersen conductances greater than 5 mhos, the agreement is within 25 percent. The data were also used to deduce the altitude profile of the effective recombination coefficient. This profile agrees well with the results of previous studies.

Vondrak, R.↗

Aerosol size distributions obtained by inversion of spectral optical depth measurements

Columnar aerosol size distributions have been inferred by numerically inverting particulate optical depth measurements as a function of wavelength. An inversion formula which explicitly includes the magnitude of the measurement variances is derived and applied to optical depth measurements obtained in Tucson with a solar radiometer. It is found that the individual size distributions of the aerosol particles (assumed spherical), at least for radii greater than or approximately equal to 0.1 micron, fall into one of three distinctly different categories. Approximately 50% of all distributions examined thus far can best be represented as a composite of a Junge distribution plus a distribution of relatively monodispersed larger particles centered at a radius of about 0.5 micron. Scarcely 20% of the distributions yielded Junge size distributions, while 30% yielded relatively monodispersed distributions of the log-normal or gamma distribution types. A representative selection of each of these types will be presented and discussed. The sensitivity of spectral attenuation measurements to the radii limits and refractive index assumed in the numerical inversion will also be addressed.

King, M. D.↗

A CONSTANT LIMITING MASS SCALE FOR FLAT EARLY-TYPE GALAXIES FROM z {approx} 1 TO z = 0: DENSITY EVOLVES BUT SHAPES DO NOT

We measure the evolution in the intrinsic shape distribution of early-type galaxies from z {approx} 1 to z {approx} 0 by analyzing their projected axis-ratio distributions. We extract a low-redshift sample (0.04 < z < 0.08) of early-type galaxies with very low star formation rates from the Sloan Digital Sky Survey, based on a color-color selection scheme and verified through the absence of emission lines in the spectra. The inferred intrinsic shape distribution of these early-type galaxies is strongly mass dependent: the typical short-to-long intrinsic axis ratio of high-mass early-type galaxies (>10{sup 11} M{sub Sun }) is 2:3, whereas at masses below 10{sup 11} M{sub Sun} this ratio narrows to 1:3, or more flattened galaxies. In an entirely analogous manner, we select a high-redshift sample (0.6 < z < 0.8) from two deep-field surveys with multi-wavelength and Hubble Space Telescope/Advanced Camera for Surveys imaging: GEMS and COSMOS. We find a seemingly universal mass of {approx}10{sup 11} M{sub Sun} for highly flattened early-type systems at all redshifts. This implies that the process that grows an early-type galaxy above this ceiling mass, irrespective of cosmic epoch, involves forming round systems. Using both parametric and non-parametric tests, we find no evolution in the projected axis-ratio distribution for galaxies with masses >3 Multiplication-Sign 10{sup 10} M{sub Sun} with redshift. At the same time, our samples imply an increase of 2-3 Multiplication-Sign in comoving number density for early-type galaxies at masses >3 Multiplication-Sign 10{sup 10} M{sub Sun }, in agreement with previous studies. Given the direct connection between the axis-ratio distribution and the underlying bulge-to-disk ratio distribution, our findings imply that the number density evolution of early-type galaxies is not exclusively driven by the emergence of either bulge- or disk-dominated galaxies, but rather by a balanced mix that depends only on the stellar mass of the galaxy. The challenge for galaxy formation models is to reproduce this overall non-evolving ratio of flattened to round early-type galaxies in the context of a continually growing population.

79 ASTRONOMY AND ASTROPHYSICS↗

Numerical convergence and validation of the DIMP inverse particle transport model

The data integration with modeled predictions (DIMP) model is a promising inverse radiation transport method for solving the special nuclear material (SNM) holdup problem. Unlike previous methods, DIMP is a completely passive nondestructive assay technique that requires no initial assumptions regarding the source distribution or active measurement time. DIMP predicts the most probable source location and distribution through Bayesian inference and quasi-Newtonian optimization of predicted detector re-sponses (using the adjoint transport solution) with measured responses. DIMP performs well with for-ward hemispherical collimation and unshielded measurements, but several considerations are required when using narrow-view collimated detectors. DIMP converged well to the correct source distribution as the number of synthetic responses increased. DIMP also performed well for the first experimental validation exercise after applying a collimation factor, and sufficiently reducing the source search vol-ume's extent to prevent the optimizer from getting stuck in local minima. DIMP's simple point detector response function (DRF) is being improved to address coplanar false positive/negative responses, and an angular DRF is being considered for integration with the next version of DIMP to account for highly collimated responses. Overall, DIMP shows promise for solving the SNM holdup inverse problem, especially once an improved optimization algorithm is implemented.

42 ENGINEERING↗

Eddy covariance observations of surface leakage during shallow subsurface CO2 releases

We tested the ability of eddy covariance (EC) to detect, locate, and quantify surface CO{sub 2} flux leakage signals within a background ecosystem. For 10 days starting on 07/09/2007, and for seven days starting on 08/03/2007, 0.1 (Release 1) and 0.3 (Release 2) t CO{sub 2}d{sup -1}, respectively, were released from a horizontal well {approx}100 m in length and {approx}2.5 m in depth located in an agricultural field in Bozeman, MT. An EC station measured net CO{sub 2} flux (F{sub c}) from 06/08/2006 to 09/04/2006 (mean and standard deviation = -12.4 and 28.1 g m{sup -2} d{sup -1}, respectively) and from 05/28/2007 to 09/04/2007 (mean and standard deviation = -12.0 and 28.1 g m{sup -2} d{sup -1}, respectively). The Release 2 leakage signal was visible in the F{sub c} time series, whereas the Release 1 signal was difficult to detect within variability of ecosystem fluxes. To improve detection ability, we calculated residual fluxes (F{sub cr}) by subtracting fluxes corresponding to a model for net ecosystem exchange from F{sub c}. F{sub cr} had reduced variability and lacked the negative bias seen in corresponding F{sub c} distributions. Plotting the upper 90th percentile F{sub cr} versus time enhanced the Release 2 leakage signal. However, values measured during Release 1 fell within the variability assumed to be related to unmodeled natural processes. F{sub cr} measurements and corresponding footprint functions were inverted using a least-squares approach to infer the spatial distribution of surface CO{sub 2} fluxes during Release 2. When combined with flux source area evaluation, inversion results roughly located the CO{sub 2} leak, while resolution was insufficient to quantify leakage rate.

54 ENVIRONMENTAL SCIENCES↗

Comparative Genomics and Functional Analysis of Carbohydrate Utilization Networks in Unicyanobacterial Consortia Derived from Hypersaline Lake Microbial Mats

Hot Lake is a meromictic, hypersaline lake located in Washington state, USA. It harbors a phototrophic microbial mat community stable in dramatic seasonal environmental changes. The complexity of natural community hinders to analyze metabolic interactions between its members. To manage this obstacle, unicyanobacterial consortia (UCC) were sampled from the community and sequenced. UCC is a microbial community in which one cyanobacterium serves as the sole primary producer of carbon sources to several heterotrophic bacteria. Recently available metagenomes for two UCC revealed a shared set of 19 microbial operational taxonomic units (OTU), whose individual genomes were reconstructed by co‐assembly and abundance profile binning. We used this genomic data and a comparative genomics approach to infer carbohydrate utilization pathways and their transcription regulation. We scanned all UCC proteomes against the KEGG Orthology, Pfam, GO and SEED databases to reveal genes potentially involved in sugar utilization. Then we analyzed the genomic and functional contexts of gene loci encoding the obtained proteins and reconstructed the respective metabolic pathways and transcriptional regulons. As result, we observed catabolic pathways for 32 various carbohydrates and their derivatives. Distribution of the inferred catabolic pathways has highly mosaic structure across the UCC genomes. The largest number of pathways was observed in two representatives of the Halomonas genus (14–15 pathways) and one Rhodobacteriaceae sp. (15 pathways). In other UCC genomes, we found up to 10 carbohydrate utilization pathways per genome. However in 7 genomes we didn't observe any pathway. The most widely distributed pathways in UCC were the alpha‐glucosides and the DeLey‐Doudoroff galactose utilization pathways. Both were found in 7 genomes. Majority of pathways are distributed among 1–3 genomes. The reconstructed catabolic pathways were partially validated by phenotypic analyses of UCC community members that were isolated in a pure culture. In the Rhodobacteriaceae sp., we found a novel catabolic pathway that was proposed to be involved in mannoheptulose utilization. To our knowledge, the proposed mannoheptulose catabolic pathway is the first case of a pathway for heptose utilization and will require further experimental validation. The identified pathways are potentially regulated by local transcription factors (TFs) that are often co‐localized with the respective sugar catabolic gene loci. For 63 out of the 80 identified TFs we identified their cognate DNA binding motifs and reconstructed regulons using the bioinformatics approach. Support or Funding Information This research was supported by the Genomic Science Program (GSP), Office of Biological and Environmental Research (OBER), and U.S. Department of Energy (DOE) and is a contribution of the Pacific Northwest National Laboratory (PNNL) Foundational Scientific Focus Area.

Leyn, Semen A.↗

Diabatic forcing and intialization with assimilation of cloud water and rainwater in a forecast model

In this study, diabatic forcing, and liquid water assimilation techniques are tested in a semi-implicit hydrostatic regional forecast model containing explicit representations of grid-scale cloud water and rainwater. Diabatic forcing, in conjunction with diabatic contributions in the initialization, is found to help the forecast retain the diabatic signal found in the liquid water or heating rate data, consequently reducing the spinup time associated with grid-scale precipitation processes. Both observational Special Sensor Microwave/Imager (SSM/I) and model-generated data are used. A physical retrieval method incorporating SSM/I radiance data is utilized to estimate the 3D distribution of precipitating storms. In the retrieval method the relationship between precipitation distributions and upwelling microwave radiances is parameterized, based upon cloud ensemble-radiative model simulations. Regression formulae relating vertically integrated liquid and ice-phase precipitation amounts to latent heating rates are also derived from the cloud ensemble simulations. Thus, retrieved SSM/I precipitation structures can be used in conjunction with the regression-formulas to infer the 3D distribution of latent heating rates. These heating rates are used directly in the forecast model to help initiate Tropical Storm Emily (21 September 1987). The 14-h forecast of Emily's development yields atmospheric precipitation water contents that compare favorably with coincident SSM/I estimates.

Raymond, William H.↗

Geomorphic evidence for the distribution of ground ice on Mars

High-resolution Viking orbiter images show evidenced for quasi-viscous relaxation of topography. The relaxation is believed to be due to creep deformation of ice in near-surface materials. The global distribution of the inferred ground ice shows a pronounced latitudinal dependence. The equatorial regions of Mars appear to be ice-poor, while the heavily cratered terrain poleward of + or - 30 deg latitude appears to be ice-rich. The style of creep poleward of + or - 30 deg varies with latitude, possibly due to variations in ice rheology with temperature. The distribution suggests that ice at low latitudes, which is not in equilibrium with the present atmosphere, has been lost via sublimation and diffusion through the regolith, thereby causing a net poleward transport of ice over martian history.

Squyres, S. W.↗

Latitudinal variability of large-scale coronal temperature and its association with the density and the global magnetic field

In this paper we utilize the latitiude distribution of the coronal temperature during the period 1984-1992 that was derived in a paper by Guhathakurta et al, 1993, utilizing ground-based intensity observations of the green (5303 A Fe XIV) and red (6374 A Fe X) coronal forbidden lines from the National Solar Observatory at Sacramento Peak, and establish it association with the global magnetic field and the density distributions in the corona. A determination of plasma temperature, T, was estimated from the intensity ratio Fe X/Fe XIV (where T is inversely proportional to the ratio), since both emission lines come from ionized states of Fe, and the ratio is only weakly dependent on density. We observe that there is a large-scale organization of the inferred coronal temperature distribution that is associated with the large-scale, weak magnetic field structures and bright coronal features; this organization tends to persist through most of the magnetic activity cycle. These high-temperature structures exhibit time-space characteristics which are similar to those of the polar crown filaments. This distribution differs in spatial and temporal characterization from the traditional picture of sunspot and active region evolution over the range of the sunspot cycle, which are manifestations of the small-scale, strong magnetic field regions.

Guhathakurta, M.↗

Bayesian data assimilation for stochastic multiscale models of transport in porous media.

We investigate Bayesian techniques that can be used to reconstruct field variables from partial observations. In particular, we target fields that exhibit spatial structures with a large spectrum of lengthscales. Contemporary methods typically describe the field on a grid and estimate structures which can be resolved by it. In contrast, we address the reconstruction of grid-resolved structures as well as estimation of statistical summaries of subgrid structures, which are smaller than the grid resolution. We perform this in two different ways (a) via a physical (phenomenological), parameterized subgrid model that summarizes the impact of the unresolved scales at the coarse level and (b) via multiscale finite elements, where specially designed prolongation and restriction operators establish the interscale link between the same problem defined on a coarse and fine mesh. The estimation problem is posed as a Bayesian inverse problem. Dimensionality reduction is performed by projecting the field to be inferred on a suitable orthogonal basis set, viz. the Karhunen-Loeve expansion of a multiGaussian. We first demonstrate our techniques on the reconstruction of a binary medium consisting of a matrix with embedded inclusions, which are too small to be grid-resolved. The reconstruction is performed using an adaptive Markov chain Monte Carlo method. We find that the posterior distributions of the inferred parameters are approximately Gaussian. We exploit this finding to reconstruct a permeability field with long, but narrow embedded fractures (which are too fine to be grid-resolved) using scalable ensemble Kalman filters; this also allows us to address larger grids. Ensemble Kalman filtering is then used to estimate the values of hydraulic conductivity and specific yield in a model of the High Plains Aquifer in Kansas. Strong conditioning of the spatial structure of the parameters and the non-linear aspects of the water table aquifer create difficulty for the ensemble Kalman filter. We conclude with a demonstration of the use of multiscale stochastic finite elements to reconstruct permeability fields. This method, though computationally intensive, is general and can be used for multiscale inference in cases where a subgrid model cannot be constructed.

54 ENVIRONMENTAL SCIENCES↗

Bayesian Analysis of Savings from Retrofit Projects

Estimates of savings from retrofit projects depend on statistical models, but because of the complicated analysis required to determine the uncertainty of the estimates, savings uncertainty is not often considered. Numerous simplified methods have been proposed to determine savings uncertainty, but in all but the simplest cases, these methods provide approximate results only. The objective of this paper is to show that Bayesian inference provides a consistent framework for estimating savings and savings uncertainty in retrofit projects. We review the mathematical background of Bayesian inference and Bayesian regression, and present two examples of estimating savings and savings uncertainty in retrofit projects. The first is a simple case where both baseline and post-retrofit monthly natural gas use can be modeled as a linear function of monthly heating degree days. The Efficiency Valuation Organization (EVO 2007) defines two methods of determining savings in such cases: reporting period savings, which is an estimate of the savings during the post-retrofit period; and normalized savings, which is an estimate of the savings that would be obtained during a typical year at the project site. For reporting period savings, classical statistical analysis provides exact analytic results for both savings and savings uncertainty in this case. We use Bayesian analysis to calculate reporting period savings and savings uncertainty and show that the results are identical to the analytical results. For normalized savings, the literature contains no exact expression for the uncertainty of normalized savings; we use Bayesian inference to calculate this quantity for the first time, and compare it with the result of an approximate formula that has been proposed. The second example concerns a problem where the baseline data exhibit nonlinearity and serial autocorrelation, both of which are common in real-world retrofit projects. No analytical solutions exist to determine savings or savings uncertainty in this situation, but several simplified formulas have been proposed. We model the data using a 5-parameter model with first-order autoregressive errors, and use Bayesian inference to develop distributions for the model parameters and for the reporting period savings, which allows us to determine the savings uncertainty. We find the energy savings to be about 5% lower than the result obtained by ignoring the autocorrelation. In addition, the Bayesian analysis finds the savings uncertainty to be narrower than the approximate uncertainty calculated using the simplified formula. These results show that Bayesian inference can be used to determine savings and savings uncertainty for a wide variety of real-world problems.

Im, Piljae↗