SEARCH · Search NASA

Results for “inference methods”

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.

SourceResearch connectionFrom yearThrough year

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

At least 415 records · Page 23

All sources NASA NTRS DOE OSTI

NASA NTRSDate not supplied

Transition Modeling Based on the Dual N-factor Method for the CRM-NLF Wind Tunnel Configuration

The dual N-factor method is used to model the boundary-layer transition over the common research model with natural laminar flow (CRM-NLF) aircraft configuration. The flow conditions match selected test conditions from a wind tunnel experiment in the National Transonic Facility at the NASA Langley Research Center. The paper presents a systematic methodology for transition prediction in the presence of a dual shock system and extends the prior capability for iteratively coupled computational fluid dynamics (CFD) predictions to incorporate three-dimensional, transonic wings. The method employs stability computations based on the linear parabolized stability equations (PSE), along with a dual N-factor criterion. The iterative process begins with the fully turbulent Reynolds-averaged-Navier-Stokes (RANS) mean flow solution. For the first iteration, a mean flow solution is calculated with an imposed transition front that aligns with the shock front from the fully turbulent solution. Subsequently, stability computations are performed along a set of streamlines across the wing to calculate the amplification of planar Tollmien-Schlichting (TS) and stationary crossflow (CF) modes. The transition criterion based on the dual N-factor method is used to infer the updated transition front and the process is successively repeated until convergence of the solution. Within three iterations, the predicted fronts for angles of attack of 1.45, 1.98, 2.46 and 2.94 degrees and a mean-aerodynamic-chord Reynolds number equal to 15 million, approach visual convergence in most regions of the studied cases, and the resulting predictions are in good agreement with the transition fronts deduced from measurements of temperature-sensitive paint. Even though surface pressure measurements based on fully-turbulent flow agree well with the measured pressure coefficient distributions, strong viscous-inviscid interaction effects cause significant shifts in the shock locations based on the imposed transition front, underscoring the intrusive nature of static pressure measurements using surface mounted ports on the CRM-NLF configuration.

Boundary Layer Transition↗

NASA NTRS2024

Short–Period Variables in TESS Full–Frame Image Light Curves Identified via Convolutional Neural Networks

The Transiting Exoplanet Survey Satellite (TESS) mission measured light from stars in ∼85% of the sky throughout its 2 yr primary mission, resulting in millions of TESS 30-minute-cadence light curves to analyze in the search for transiting exoplanets. To search this vast data set, we aim to provide an approach that is computationally efficient, produces accurate predictions, and minimizes the required human search effort. We present a convolutional neural network that we train to identify short-period variables. To make a prediction for a given light curve, our network requires no prior target parameters identified using other methods. Our network performs inference on a TESS 30-minute-cadence light curve in ∼5 ms on a single GPU, enabling large-scale archival searches. We present a collection of 14,156 short-period variables identified by our network. The majority of our identified variables fall into two prominent populations, one of close-orbit main-sequence binaries and another of δ Scuti stars. Our neural network model and related code are additionally provided as open-source code for public use and extension.

Convolutional neural networks↗

DOE OSTI2017

Systematic parameter inference in stochastic mesoscopic modeling

We propose a method to efficiently determine the optimal coarse-grained force field in mesoscopic stochastic simulations of Newtonian fluid and polymer melt systems modeled by dissipative particle dynamics (DPD) and energy conserving dissipative particle dynamics (eDPD). The response surfaces of various target properties (viscosity, diffusivity, pressure, etc.) with respect to model parameters are constructed based on the generalized polynomial chaos (gPC) expansion using simulation results on sampling points (e.g., individual parameter sets). To alleviate the computational cost to evaluate the target properties, we employ the compressive sensing method to compute the coefficients of the dominant gPC terms given the prior knowledge that the coefficients are “sparse”. The proposed method shows comparable accuracy with the standard probabilistic collocation method (PCM) while it imposes a much weaker restriction on the number of the simulation samples especially for systems with high dimensional parametric space. Fully access to the response surfaces within the confidence range enables us to infer the optimal force parameters given the desirable values of target properties at the macroscopic scale. Moreover, it enables us to investigate the intrinsic relationship between the model parameters, identify possible degeneracies in the parameter space, and optimize the model by eliminating model redundancies. The proposed method provides an efficient alternative approach for constructing mesoscopic models by inferring model parameters to recover target properties of the physics systems (e.g., from experimental measurements), where those force field parameters and formulation cannot be derived from the microscopic level in a straight forward way.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2023

Uncertainty quantification in scientific machine learning: Methods, metrics, and comparisons

Neural networks (NNs) are currently changing the computational paradigm on how to combine data with mathematical laws in physics and engineering in a profound way, tackling challenging inverse and ill-posed problems not solvable with traditional methods. However, quantifying errors and uncertainties in NN-based inference is more complicated than in traditional methods. This is because in addition to aleatoric uncertainty associated with noisy data, there is also uncertainty due to limited data, but also due to NN hyperparameters, overparametrization, optimization and sampling errors as well as model misspecification. Although there are some recent works on uncertainty quantification (UQ) in NNs, there is no systematic investigation of suitable methods towards quantifying the total uncertainty effectively and efficiently even for function approximation, and there is even less work on solving partial differential equations and learning operator mappings between infinite-dimensional function spaces using NNs. In this work, we present a comprehensive framework that includes uncertainty modeling, new and existing solution methods, as well as evaluation metrics and post-hoc improvement approaches. Further, to demonstrate the applicability and reliability of our framework, we present an extensive comparative study in which various methods are tested on prototype problems, including problems with mixed input-output data, and stochastic problems in high dimensions. In the Appendix, we include a comprehensive description of all the UQ methods employed. Further, to help facilitate the deployment of UQ in Scientific Machine Learning research and practice, we present and develop in [1] an open-source Python library (github.com/Crunch-UQ4MI/neuraluq), termed NeuralUQ, that is accompanied by an educational tutorial and additional computational experiments.

11 physics-informed neural networks↗

DOE OSTI2026

Multidimensional scaling informed by F -statistic: Visualizing grouped microbiome data with inference

Multidimensional scaling (MDS) is a widely used dimensionality reduction technique in microbial ecology data analysis that captures the multivariate structure of the data while preserving pairwise distances between samples. While improvements in MDS have enhanced the ability to reveal group-specific data patterns, these MDS-based methods require prior assumptions for inference, limiting their application in general microbiome analysis. Here, in this study, we introduce a new MDS-based ordination method, “F-informed MDS,” which configures the data distribution based on the F-statistic, the ratio of dispersion between groups sharing common and different characteristics. Using semisynthetic datasets, we demonstrate that the proposed method is robust to hyperparameter selection while maintaining statistical significance throughout the ordination process. Various quality metrics for evaluating dimensionality reduction confirm that F-informed MDS is comparable to state-of-the-art methods in preserving both local and global data structures. Its application to a diatom-associated bacterial community suggests the role of this new method in interpreting the community’s response to the host. Our approach offers a well-founded refinement of MDS that aligns with statistical test results, which can be beneficial for broader multidimensional data analyses in microbiology and ecology. This new visualization tool can be incorporated into standard microbiome data analyses.

Biological and medical sciences↗

DOE OSTI2026

Active operator learning with predictive uncertainty quantification for partial differential equations

With the increased prevalence of neural operators being used to provide rapid solutions to partial differential equations (PDEs), understanding the accuracy of model predictions and the associated error levels is necessary for deploying reliable surrogate models in scientific applications. Existing uncertainty quantification (UQ) frameworks employ ensembles or Bayesian methods, which can incur substantial computational costs during both training and inference. Here, we propose a lightweight predictive UQ method tailored for Deep operator networks (DeepONets) that also generalizes to other operator networks. Numerical experiments on linear and nonlinear PDEs demonstrate that the framework’s uncertainty estimates are unbiased and provide accurate out-of-distribution uncertainty predictions with a sufficiently large training dataset. Our framework provides fast inference and uncertainty estimates that can efficiently drive outer-loop analyses that would be prohibitively expensive with conventional solvers. We demonstrate how predictive uncertainties can be used in the context of Bayesian optimization and active learning problems to yield improvements in accuracy and data-efficiency for outer-loop optimization procedures. In the active learning setup, we extend the framework to Fourier Neural Operators (FNO) and describe a generalized method for other operator networks. To enable real-time deployment, we introduce an inference strategy based on precomputed trunk outputs and a sparse placement matrix, reducing evaluation time by more than a factor of five. Our method provides a practical route to uncertainty-aware operator learning in time-sensitive settings.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

Simple and statistically sound recommendations for analysing physical theories

Physical theories that depend on many parameters or are tested against data from many different experiments pose unique challenges to statistical inference. Many models in particle physics, astrophysics and cosmology fall into one or both of these categories. These issues are often sidestepped with statistically unsound ad hoc methods, involving intersection of parameter intervals estimated by multiple experiments, and random or grid sampling of model parameters. Whilst these methods are easy to apply, they exhibit pathologies even in low-dimensional parameter spaces, and quickly become problematic to use and interpret in higher dimensions. In this article we give clear guidance for going beyond these procedures, suggesting where possible simple methods for performing statistically sound inference, and recommendations of readily-available software tools and standards that can assist in doing so. Our aim is to provide any physicists lacking comprehensive statistical training with recommendations for reaching correct scientific conclusions, with only a modest increase in analysis burden. Furthermore, our examples can be reproduced with the code publicly available at Zenodo.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2020

Turbulence model reduction by deep learning

A defining problem of turbulence theory is to produce a predictive model for turbulent fluxes. These have profound implications for virtually all aspects of the turbulence dynamics. In magnetic confinement devices, drift-wave turbulence produces anomalous fluxes via cross-correlations between fluctuations. In this work, we introduce an alternative, data-driven method for parametrizing these fluxes. The method uses deep supervised learning to infer a reduced mean-field model from a set of numerical simulations. We apply the method to a simple drift-wave turbulence system and find a significant new effect which couples the particle flux to the local gradient of vorticity. Notably, here, this effect is much stronger than the oft-invoked shear suppression effect. We also recover the result via a simple calculation. The vorticity gradient effect tends to modulate the density profile. In addition, our method recovers a model for spontaneous zonal flow generation by negative viscosity, stabilized by nonlinear and hyperviscous terms. We highlight the important role of symmetry to implementation of the new method.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

NASA NTRS2003

Wildfires Dynamics in Mid-Siberian Larch Dominated Communities

The longterm wildfire dynamics, including fire return interval (FRI), in Siberian larch communities were examined. A wildfire chronology encompassing the 15th through the 20th centuries was developed from analyzing tree stem fire scars. Two methods were used to calculate the time interval between fires: 1) direct counting of annual tree growth rings between stem fire scars and 2) the next earlier fire date was estimated from growth ring analysis and added to the first estimate. Average FRI determined from stem fire scar dating was 82 plus or minus 7 using Method I or 95 plus or minus 7 when age of the next earlier fire was inferred from observed larch regeneration structure (Method II). FRI was also found to be dependent on site topography. FRI on north-east facing slopes was 86 plus or minus 11 years (105 plus or minus 12). FRI on south-west facing slopes was significantly less at 61 plus or minus 8 (73 plus ot minus 8) years. Flat terrain showed little difference between methods 68 plus or minus 14 (70 plus or minus 13). This was also the case for bogs, but FRI was much longer; 139 plus or minus 17 (138 plus ot minus 18). The maximum number of annual fires occurred with periods of 36 and 82 years on average. The temporal trend in the FRI decreased from 101 years in the 19 th century to 65 years in the 20th century. The effect of post-fire forest recovery on depth to permafrost was also estimated. After initial melting from increased local temperatures permafrost depth decreased at a rate of 0.3 cm/yr on average as forest canopies developed.

Kharuk, V. I.↗

DOE OSTI2023

Inference of relative permeability curves in reservoir rocks with ensemble Kalman method

Not provided.

Chemistry↗

DOE OSTI2025

Accelerating Hamiltonian Monte Carlo for Bayesian inference in neural networks and neural operators

Hamiltonian Monte Carlo (HMC) is a powerful and accurate method to sample from the posterior distribution in Bayesian inference. However, HMC techniques are computationally demanding for Bayesian neural networks due to the high dimensionality of the network’s parameter space and the non-convexity of their posterior distributions. Therefore, various approximation techniques, such as variational inference (VI) or stochastic gradient MCMC, are often employed to infer the posterior distribution of the network parameters. Such approximations introduce inaccuracies in the inferred distributions, resulting in unreliable uncertainty estimates. In this work, we propose a hybrid approach that combines inexpensive VI and accurate HMC methods to efficiently and accurately quantify uncertainties in neural networks and neural operators. The proposed approach leverages an initial VI training on the full network. We examine the influence of individual parameters on the prediction uncertainty, which shows that a large proportion of the parameters do not contribute substantially to uncertainty in the network predictions. This information is then used to significantly reduce the dimension of the parameter space, and HMC is performed only for the subset of network parameters that strongly influence prediction uncertainties. This yields a framework for accelerating the full batch HMC for posterior inference in neural networks. We demonstrate the efficiency and accuracy of the proposed framework on deep neural networks and operator networks, showing that inference can be performed for large networks with tens to hundreds of thousands of parameters. Finally, we show that this method can effectively learn surrogates for complex physical systems by modeling the operator that maps from upstream conditions to wall-pressure data on a cone in hypersonic flow.

Bayesian inference↗

DOE OSTI2019

Parameter Estimation and Uncertainty Quantification for Systems Biology Models

Mathematical models can provide quantitative insights into immunoreceptor signaling, and other biological processes, but require parameterization and uncertainty quantification before reliable predictions become possible. We review currently available methods and software tools to address these problems. We consider gradient-based and gradient-free methods for point estimation of parameter values, and methods of profile likelihood, bootstrapping, and Bayesian inference for uncertainty quantification. We consider recent and potential future applications of these methods to systems-level modeling of immune-related phenomena.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2021

BraggNN : fast X-ray Bragg peak analysis using deep learning

X-ray diffraction based microscopy techniques such as high-energy diffraction microscopy (HEDM) rely on knowledge of the position of diffraction peaks with high precision. These positions are typically computed by fitting the observed intensities in detector data to a theoretical peak shape such as pseudo-Voigt. As experiments become more complex and detector technologies evolve, the computational cost of such peak-shape fitting becomes the biggest hurdle to the rapid analysis required for real-time feedback in experiments. To this end, we propose BraggNN, a deep-learning based method that can determine peak positions much more rapidly than conventional pseudo-Voigt peak fitting. When applied to a test dataset, peak center-of-mass positions obtained from BraggNN deviate less than 0.29 and 0.57 pixels for 75 and 95% of the peaks, respectively, from positions obtained using conventional pseudo-Voigt fitting (Euclidean distance). When applied to a real experimental dataset and using grain positions from near-field HEDM reconstruction as ground-truth, grain positions using BraggNN result in 15% smaller errors compared with those calculated using pseudo-Voigt. Recent advances in deep-learning method implementations and special-purpose model inference accelerators allow BraggNN to deliver enormous performance improvements relative to the conventional method, running, for example, more than 200 times faster on a consumer-class GPU card with out-of-the-box software.

36 MATERIALS SCIENCE↗

NASA NTRSDate not supplied

Consequences of Asteroid Characterization on the State of Knowledge about Inferred Physical Properties and Impact Risk

Physical characteristics of Near-Earth Objects (NEOs) are essential inputs to planetary defense assessments. The size, density, and strength of an NEO are critical inputs to modeling behavior during atmospheric entry as well as assessing the risk of impact. Similarly, knowledge of the physical characteristics of an object are necessary to evaluate the probable result of a mitigation mission. Usually, these attributes cannot be directly measured, but increasingly sophisticated methods have been developed to infer physical properties from related measurements of asteroids, meteors, and/or meteorites. Fortuitously, some of these measurements have been obtained for enough NEOs to elucidate the distribution of values across the sampled population. However, the situation becomes more challenging when considering a specific asteroid, since it is unlikely that all the relevant measurements have been made for any given object. We have developed a Bayesian network that can combine available information about a particular NEO with knowledge of the larger population to infer probabilistic values and uncertainties for physical characteristics of interest. Distributions of asteroid population albedos, taxonomic classes, and macroporosities, along with meteorite density distributions and associations between taxonomic classes and meteorite classes, provide the default distributions for the network’s parameter nodes. The inference network links parameters for each virtual asteroid either deterministically or probabilistically as appropriate, and eliminates any unphysical combinations of parameters. Within the context of planetary defense, our Bayesian network can be used to constrain the ranges of likely impactor properties, which can subsequently reduce the uncertainty in modelling of atmospheric entry, mitigation efficacy, and impact risk assessment. When additional measurements become available for a specific object, the network incorporates those measurements to generate virtual asteroids with property distributions that are consistent with the measurements. We will use the 2023 PDC scenario to demonstrate how the inference network can be combined with plausible characterization measurements to refine the state of knowledge about likely combinations of physical parameters and the resulting impact risk.

risk assessment↗

NASA NTRSDate not supplied

Consequences of Asteroid Characterization on the State of Knowledge about Inferred Physical Properties and Impact Risk

risk assessment↗

DOE OSTI2019

H β and H γ Absorption-line Profile Inconsistencies in Laboratory Experiments Performed at White Dwarf Photosphere Conditions

The spectroscopic method relies on hydrogen Balmer absorption lines to infer white dwarf (WD) masses. These masses depend on the choice of atmosphere model, hydrogen atomic line shape calculation, and which Balmer series members are included in the spectral fit. In addition to those variables, spectroscopic masses disagree with those derived using other methods. Here we present laboratory experiments aimed at investigating the main component of the spectroscopic method: hydrogen line shape calculations. These experiments use X-rays from Sandia National Laboratories’ Z-machine to create a uniform ~15 cm 3 hydrogen plasma and a ~4 eV backlighter that enables recording high-quality absorption spectra. The large plasma, volumetric X-ray heating that fosters plasma uniformity, and the ability to collect absorption spectra at WD photosphere conditions are improvements over past laboratory experiments. Analysis of the experimental absorption spectra reveals that electron density (n e ) values derived from the Hγ line are ~34% ± 7.3% lower than from Hβ. Two potential systematic errors that may contribute to this difference were investigated. A detailed evaluation of self-emission and plasma gradients shows that these phenomena are unlikely to produce any measurable Hβ–Hγ n e difference. WD masses inferred with the spectroscopic method are proportional to the photosphere density. Hence, the measured Hβ–Hγ n e difference is qualitatively consistent with the trend that WD masses inferred from their Hβ line are higher than that resulting from the analysis of Hβ and Hγ. In conclusion, this evidence may suggest that current hydrogen line shape calculations are not sufficiently accurate to capture the intricacies of the Balmer series.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2016

STRATEGIES IN SEISMIC INFERENCE OF SUPERGRANULAR FLOWS ON THE SUN

Observations of the solar surface reveal the presence of flows with length scales of around 35 Mm, commonly referred to as supergranules. Inferring the subsurface flow profile of supergranules from measurements of the surface and photospheric wavefield is an important challenge faced by helioseismology. Traditionally, the inverse problem has been approached by studying the linear response of seismic waves in a horizontally translationally invariant background to the presence of the supergranule; following an iterative approach that does not depend on horizontal translational invariance might perform better, since the misfit can be analyzed post iterations. In this work, we construct synthetic observations using a reference supergranule and invert for the flow profile using surface measurements of travel times of waves belonging to modal ridges f (surface gravity) and p {sub 1} through p {sub 7} (acoustic). We study the extent to which individual modes and their combinations contribute to infer the flow. We show that this method of nonlinear iterative inversion tends to underestimate the flow velocities, as well as inferring a shallower flow profile, with significant deviations from the reference supergranule near the surface. We carry out a similar analysis for a sound-speed perturbation and find that analogous near-surface deviations persist, although the iterations converge faster and more accurately. We conclude that a better approach to inversion would be to expand the supergranule profile in an appropriate basis, thereby reducing the number of parameters being inverted for and appropriately regularizing them.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS2002

A statistical model for interpreting computerized dynamic posturography data

Computerized dynamic posturography (CDP) is widely used for assessment of altered balance control. CDP trials are quantified using the equilibrium score (ES), which ranges from zero to 100, as a decreasing function of peak sway angle. The problem of how best to model and analyze ESs from a controlled study is considered. The ES often exhibits a skewed distribution in repeated trials, which can lead to incorrect inference when applying standard regression or analysis of variance models. Furthermore, CDP trials are terminated when a patient loses balance. In these situations, the ES is not observable, but is assigned the lowest possible score--zero. As a result, the response variable has a mixed discrete-continuous distribution, further compromising inference obtained by standard statistical methods. Here, we develop alternative methodology for analyzing ESs under a stochastic model extending the ES to a continuous latent random variable that always exists, but is unobserved in the event of a fall. Loss of balance occurs conditionally, with probability depending on the realized latent ES. After fitting the model by a form of quasi-maximum-likelihood, one may perform statistical inference to assess the effects of explanatory variables. An example is provided, using data from the NIH/NIA Baltimore Longitudinal Study on Aging.

NASA Discipline Neuroscience↗