Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “uncertainty calibration”

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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

Uncertainty guided online ensemble for non-stationary data streams in fusion science

Machine Learning (ML) is poised to play a pivotal role in the development and operation of next-generation fusion devices. Fusion data shows non-stationary behavior with distribution drifts, resulted by both experimental evolution and machine wear-and-tear. ML models assume stationary distribution and fail to maintain performance when encountered with such non-stationary data streams. Online learning techniques have been leveraged in other domains, however it has been largely unexplored for fusion applications. In this paper, we investigate online learning for continuous adaptation to drifting data streams in the prediction of Toroidal Field (TF) coils deflection at the DIII-D fusion facility. We further address the short-term performance degradation inherent to standard online learning, which arises because ground truth is unavailable at prediction time. To mitigate this issue, we propose an uncertainty-guided online ensemble framework. The method leverages the Deep Gaussian Process Approximation (DGPA) for calibrated uncertainty estimation and uses these uncertainty measures to guide a meta-algorithm that aggregates predictions from learners trained over different historical horizons. Our results show that online learning reduces prediction error by 80% compared to a static model. The online ensemble and the proposed uncertainty-guided ensemble further reduce error by approximately 6%, and 10% respectively, relative to standard single-model online learning, while also providing calibrated uncertainty estimates to support operational decision-making.

AI↗
DOE OSTI2023

Massively Parallel Bayesian Model Calibration and Uncertainty Quantification with Applications to Nuclear Fuels and Materials

The U.S. Department of Energy (DOE)’s Nuclear Energy Advanced Modeling and Simulation (NEAMS) program aims to develop predictive capabilities by applying computational methods to the analysis and design of advanced reactor and fuel cycle systems. This program has been providing engineering-scale support for the development of BISON, a high-fidelity and high-resolution fuel performance tool. Fuel behavior in a nuclear reactor is governed by a complex network of mechanisms interacting with various other physics aspects in the reactor system. Any model developed to represent the fuel behavior will likely be idealized resulting in uncertainties in their predictions compared to the observed data. As such, this report was motivated by the need to identify the sources of uncertainties and quantify and propagate them through the fuel model outputs. Such quantification of uncertainties will establish a level of model trustworthiness, identify approaches to improve the model trustworthiness, and even guide optimal experiment design for maximal information gain. To accomplish the uncertainty quantification for computational models, this report has relied on the Bayesian framework which provides probabilistic treatment of models their inputs and outputs. The current state-of-the-art on performing Bayesian Uncertainty Quantification (UQ) for nuclear engineering models using High Performance Computing (HPC) resources have been reviewed. Implementation of capabilities for massively parallel Bayesian UQ in Multiphysics Object-Oriented Simulation Environment (MOOSE) is discussed. Several verification cases are discussed to verify the accuracy of the quantified uncertainties using the developed computational capabilities in MOOSE. Then, the problem of quantifying the uncertainties in TRI-Structural isOtropic (TRISO) fuel silver release is addressed. For the first time, the uncertainties arising from the TRISO Fission Gas Release (FGR) model due to model inadequacy and experimental noise are quantified. Also, the Bayesian capabilities are applied to the calibration of the MATPRO creep model, a widely used model in several fuel assessment cases. The impact of the prediction uncertainties in the MATPRO model on the fuel cladding behavior as part of the TRIBULATION assessment case (which is an integral effects case) is investigated. This report concludes with a discussion on the future work for the UQ for computational models.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2023

BEYONDPLANCK X. Planck Low Frequency Instrument frequency maps with sample-based error propagation

We present Planck Low Frequency Instrument (LFI) frequency sky maps derived within the BEYONDPLANCK framework. This framework draws samples from a global posterior distribution that includes instrumental, astrophysical, and cosmological parameters, and the main product is an entire ensemble of frequency sky map samples, each of which corresponds to one possible realization of the various modeled instrumental systematic corrections, including correlated noise, time-variable gain, as well as far sidelobe and bandpass corrections. This ensemble allows for computationally convenient end-to-end propagation of low-level instrumental uncertainties into higher-level science products, including astrophysical component maps, angular power spectra, and cosmological parameters. We show that the two dominant sources of LFI instrumental systematic uncertainties are correlated noise and gain fluctuations, and the products presented here support – for the first time – full Bayesian error propagation for these effects at full angular resolution. We compared our posterior mean maps with traditional frequency maps delivered by the Planck Collaboration, and find generally good agreement. The most important quality improvement is due to significantly lower calibration uncertainties in the new processing, as we find a fractional absolute calibration uncertainty at 70 GHz of Δg 0 /g 0 = 5 x 10 -5 , which is nominally 40 times smaller than that reported by Planck 2018. However, we also note that the original Planck 2018 estimate has a nontrivial statistical interpretation, and this further illustrates the advantage of the new framework in terms of producing self-consistent and well-defined error estimates of all involved quantities without the need of ad hoc uncertainty contributions. We describe how low-resolution data products, including dense pixel-pixel covariance matrices, may be produced from the posterior samples directly, without the need for computationally expensive analytic calculations or simulations. We conclude that posterior-based frequency map sampling provides unique capabilities in terms of low-level systematics modeling and error propagation, and may play an important role for future Cosmic Microwave Background (CMB) B-mode experiments aiming at nanokelvin precision.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Frequency-Selectable Laser Source (FLS) Calibrator for CMB Bandpass Characterization

One of the biggest challenges for Cosmic Microwave Background (CMB) experiments comes from our detector bandpass calibration. Uncertainties in bandpass can severely limit our measurements by limiting foreground removal and spectral fitting, which is particularly important for high-$\ell$ observations like cluster science using the Sunyaev-Zeldovich (SZ) effect. Currently, CMB experiments typically use a Fourier Transform Spectrometer (FTS) to measure the detector bandpasses. However, the resolution of the FTS is dependent on the length of the interferometer arms, leading to a need for increasingly large FTS instruments as CMB experiments require tighter constraints on detector bandpasses. Additionally, systematic effects like shifts in bandpass shape from uneven illumination from the FTS further limit the calibration uncertainties. As a complement to the FTS, we have developed a Frequency-selectable Laser Source (FLS) calibrator, which uses a laser with adjustable frequency housed in a calibrator that allows for varying degrees of laser power attenuation. We present several tests used to characterize the first prototype design of the FLS calibrator, as well as the improvements to the calibrator design currently underway.

Saunders, Lauren [Fermilab]↗
DOE OSTI2025

Emulation With Uncertainty Quantification of Regional Sea‐Level Change Caused by the Antarctic Ice Sheet

Abstract Projecting regional sea‐level change under various climate‐change scenarios typically involves running forward simulations of the Earth's gravitational, rotational and deformational (GRD) response to ice‐mass change, which requires substantial computational cost if applied to probabilistic frameworks requiring thousands to millions of samples. Here we build emulators of regional sea‐level change at 27 coastal locations, due to the GRD effects associated with future Antarctic Ice Sheet mass change over the 21st century. The emulators are evaluated against a numerical sea‐level model applied to an ensemble of ice‐sheet model simulations of the Antarctic Ice Sheet through 2100. We build a physics‐based emulator using a recent sensitivity kernel approach and compare it to machine learning based emulators (neural network and conditional variational autoencoder methods). In order to quantify uncertainty, we derive well‐calibrated prediction intervals for regional sea‐level change via split‐conformal inference and linear regression, and show that Monte Carlo dropout does not yield well‐calibrated uncertainties in this instance. We also demonstrate substantial gains in computational efficiency using both the physics‐based emulator and neural networks in comparison to the numerical model for the complete regional sea‐level solution. Overall, we find the physics‐based emulator modestly outperforms the machine learning emulators for this problem.

58 GEOSCIENCES↗
DOE OSTI2021

A comparison of Gaussian processes and neural networks for computer model emulation and calibration

The Department of Energy relies on complex physics simulations for prediction in domains like cosmology, nuclear theory, and materials science. These simulations are often extremely computationally intensive, with some requiring days or weeks for a single simulation. In order to assure their accuracy, these models are calibrated against observational data in order to estimate inputs and systematic biases. Because of their great computational complexity, this process typically requires the construction of an emulator, a fast approximation to the simulation. In this paper, two emulator approaches are compared: Gaussian process regression and neural networks. Their emulation accuracy and calibration performance on three real problems of Department of Energy interest is considered. On these problems, the Gaussian process emulator tends to be more accurate with narrower, but still well-calibrated uncertainty estimates. The neural network emulator is accurate, but tends to have large uncertainty on its predictions. Finally, as a result, calibration with the Gaussian process emulator produces more constrained posteriors that still perform well in prediction.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Cosmology with the Roman Space Telescope – Synergies with CMB lensing

ABSTRACT We explore synergies between the Nancy Grace Roman Space Telescope and CMB lensing data to constrain dark energy and modified gravity scenarios. A simulated likelihood analysis of the galaxy clustering and weak lensing data from the Roman Space Telescope High Latitude Survey combined with CMB lensing data from the Simons Observatory is undertaken, marginalizing over important astrophysical effects and calibration uncertainties. Included in the modelling are the effects of baryons on small-scale clustering, scale-dependent growth suppression by neutrinos, as well as uncertainties in the galaxy clustering biases, in the intrinsic alignment contributions to the lensing signal, in the redshift distributions, and in the galaxy shape calibration. The addition of CMB lensing roughly doubles the dark energy figure-of-merit from Roman photometric survey data alone, varying from a factor of 1.7 to 2.4 improvement depending on the particular Roman survey configuration. Alternatively, the inclusion of CMB lensing information can compensate for uncertainties in the Roman galaxy shape calibration if it falls below the design goals. Furthermore, we report the first forecast of Roman constraints on a model-independent structure growth, parametrized by σ8(z), and on the Hu-Sawicki f(R) gravity as well as an improved forecast of the phenomenological (Σ0, μ0) model. We find that CMB lensing plays a crucial role in constraining σ8(z) at z > 2, with percent-level constraints forecasted out to z = 4. CMB lensing information does not improve constraints on the f(R) model substantially. It does, however, increase the (Σ0, μ0) figure-of-merit by a factor of about 1.5.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2022

Mapping variations of redshift distributions with probability integral transforms

We present a method for mapping variations between probability distribution functions and apply this method within the context of measuring galaxy redshift distributions from imaging survey data. This method, which we name PITPZ for the probability integral transformations it relies on, uses a difference in curves between distribution functions in an ensemble as a transformation to apply to another distribution function, thus transferring the variation in the ensemble to the latter distribution function. This procedure is broadly applicable to the problem of uncertainty propagation. In the context of redshift distributions, for example, the uncertainty contribution due to certain effects can be studied effectively only in simulations, thus necessitating a transfer of variation measured in simulations to the redshift distributions measured from data. We illustrate the use of PITPZ by using the method to propagate photometric calibration uncertainty to redshift distributions of the Dark Energy Survey Year 3 weak lensing source galaxies. For this test case, we find that PITPZ yields a lensing amplitude uncertainty estimate due to photometric calibration error within 1 per cent of the truth, compared to as much as a 30 per cent underestimate when using traditional methods.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2021

Gap-filling eddy covariance methane fluxes: Comparison of machine learning model predictions and uncertainties at FLUXNET-CH4 wetlands

Time series of methane fluxes measured by eddy-covariance require gap-filling to estimate annual emissions. Gap-filling methane fluxes is challenging because of high variability and complex responses to multiple drivers. To date, there is no widely established gap-filling standard for methane, with regards both to the best model algorithms and predictors. In this study, we address the need for standardization by synthesizing results of gap-filling methods applied at 17 wetland sites spanning boreal to tropical regions including all major wetlands classes and two rice paddies. We introduce new procedures for: 1) creating realistic artificial gap scenarios, 2) training and evaluating gap-filling models without overstating performance, and 3) predicting half-hourly methane fluxes and annual emissions with robust uncertainty estimates. We tested a conventional method (marginal distribution sampling) and four machine learning algorithms - penalized linear regression, artificial neural networks, random forests, and boosted decision trees - and four predictor sets, including temporal, meteorological, ecosystem carbon and energy flux, and soil predictors. We find that the conventional method can achieve similar median performance to the machine learning models but is worse than the best machine learning models and relatively insensitive to predictor choices. Of the machine learning models, decision tree algorithms performed the best in cross-validation experiments, even with a baseline predictor set, and artificial neural networks showed comparable performance when using all predictors. Soil temperature was frequently the most important predictor whilst water table depth was important at sites with substantial water table fluctuations, highlighting the value of data on soil conditions. Raw gap-filling uncertainties from the machine learning models were underestimated and we propose a method to calibrate uncertainties to observations. Finally, we gap-fill and provide summary evaluation metrics for all 81 sites in the FLUXNET-CH4 community dataset and publicly release the python code for model development, evaluation, and uncertainty estimation.

42 ENGINEERING↗
DOE OSTI2026

Benchmarking the performance of uncertainty quantification methods for neural network-based interatomic potentials

Machine-learned interatomic potentials (ML-IAPs) continue to gain popularity as accurate, computationally efficient replacements for traditional, physics-based interatomic potentials and expensive ab initio methods. Uncertainty quantification (UQ) of ML-IAPs is a growing area of research as UQ is critical in many applications of IAPs, such as developing curated datasets, active learning-based data augmentation, self-improving models, and estimating the uncertainty of molecular dynamics simulations. In this paper, we construct and benchmark a series of different neural network potentials (NNPs) with varying network architectures to determine the performance of these models with respect to both the mean and uncertainty calibration error. Each NNP method is specifically designed to predict either epistemic or aleatoric uncertainty with particular focus on the differences in behavior between the epistemic and aleatoric uncertainty estimates. We benchmark these methods using multiple datasets common in the ML-IAP literature. The results show that the aleatoric uncertainty from single-shot model architectures is a competitive alternative to ensemble-based epistemic uncertainty predictions in regions of sufficient data-density. However, in regions where the representative data is sparse, aleatoric uncertainty models tend to overpredict and epistemic methods tend to underpredict the actual model error. We conclude that the type of UQ is crucial when discussing performance of probabilistic model results as different methods have different performance characteristics depending on the regime in which they are evaluated. Therefore, the type of UQ method should be carefully evaluated against both the data characteristics and requirements for the intended application.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Stellar Population Properties in the Stellar Streams around SPRC047

Abstract We have investigated the properties (e.g., age, metallicity) of the stellar populations of a ringlike tidal stellar stream (or streams) around the edge-on galaxy SPRC047 (z= 0.031) using spectral energy distribution (SED) fits to integrated broadband aperture flux densities. We used visual images in six different bands and Spitzer/IRAC 3.6μm data. We have attempted to derive best-fit stellar population parameters (metallicity, age) in three noncontiguous segments of the stream. Due to the very low surface brightness of the stream, we have performed a deconvolution with a Richardson–Lucy–type algorithm of the low spatial resolution 3.6μm IRAC image, thereby reducing the effect of the point-spread function aliasedemissionfrom the bright edge-on central galaxy at the locations of our three stream segments. Our SED fits that used several different star formation (SF) history priors, from an exponentially decaying SF burst to continuous SF, indicate that the age–metallicity–dust degeneracy is not resolved, most likely because of inadequate wavelength coverage and low signal-to-noise ratios of the low surface brightness features. We also discuss how future deep visual–near-infrared observations, combined with absolute flux calibration uncertainties at or below the 1% level, complemented by equally well absolute flux-calibrated observations in ultraviolet and mid-infrared bands, would improve the accuracy of broadband SED fitting results for low surface brightness targets, such as stellar streams around nearby galaxies that are not resolved into stars.

Astronomy & Astrophysics↗
DOE OSTI2025

Uncertainty based Online Ensemble on Non-Stationary Data for Fusion Science

Machine Learning (ML) is poised to play a pivotal role in the development and operation of next-generation fusion devices. Fusion data shows non-stationary behavior due to drifts in the data. The drifts can arise from both experimental evolution and machine wear-and-tear. ML models assume stationary distribution and fail to maintain performance when encountered with non-stationary data streams.Online learning can be used to continuously adapt the models with new data as it is acquired. However, traditional online learning can suffer from short-term performance degradation, as ground truth are not available before making the prediction. To address this challenge, we propose uncertainty aware ensemble approach for online learning. We use Deep Gaussian Process Approximation (DGPA) technique for calibrated uncertainty estimation and use the uncertainty values to guide a meta-algorithm that produces predictions based on ensemble of learners. Moreover, DGPA also provides uncertainty estimation along with the predictions for decision makers. This paper demonstrates that the proposed method outperforms traditional online learning approach, and a naive ensemble without uncertainty guidance by about 7% and 6%, respectively, on B-coil deflection prediction at DIII-D Fusion Facility.

Rajput, Kishansingh [Thomas Jefferson National Acc↗
DOE OSTI2025

Uncertainty based Online Ensemble on Non-Stationary Data for Fusion Science

Machine Learning (ML) is poised to play a pivotal role in the development and operation of next-generation fusion devices. Fusion data shows non-stationary behavior due to drifts in the data. The drifts can arise from both experimental evolution and machine wear-and-tear. ML models assume stationary distribution and fail to maintain performance when encountered with non-stationary data streams.Online learning can be used to continuously adapt the models with new data as it is acquired. However, traditional online learning can suffer from short-term performance degradation, as ground truth are not available before making the prediction. To address this challenge, we propose uncertainty aware ensemble approach for online learning. We use Deep Gaussian Process Approximation (DGPA) technique for calibrated uncertainty estimation and use the uncertainty values to guide a meta-algorithm that produces predictions based on ensemble of learners. Moreover, DGPA also provides uncertainty estimation along with the predictions for decision makers. This paper demonstrates that the proposed method outperforms traditional online learning approach, and a naive ensemble without uncertainty guidance by about 7% and 6%, respectively, on B-coil deflection prediction at DIII-D Fusion Facility.

Rajput, Kishansingh [Thomas Jefferson National Acc↗
DOE OSTI2025

Bayesian Optimized Deep Ensemble for Uncertainty Quantification of Deep Neural Networks: a System Safety Case Study on Sodium Fast Reactor Thermal Stratification Modeling

Deep neural networks (DNNs) are increasingly important to scientific computing and engineering system simulations. Accurate uncertainty quantification (UQ) for DNNs is critical in safety-sensitive engineering domains. Traditional Deep Ensemble (DE) methods, while easy to implement, frequently suffer from poorly calibrated uncertainty estimates and limited predictive accuracy due to reliance on fixed architectures with varied weight initializations. To address these issues, we introduce a workflow that combines Bayesian Optimization (BO) and DE. The workflow is modular, scalable, and integrates parallel BO initialized with Sobol sequences to individually optimize the hyperparameters of each ensemble member. This method enhances ensemble diversity, improves predictive accuracy, and provides reliable uncertainty estimates. We evaluate the proposed BODE approach in a sodium fast reactor thermal stratification modeling case study, where we used a densely connected convolutional neural network to predict turbulent viscosity during the reactor transient with consideration of data noise. We benchmark its performance against several optimization approaches, including baseline deep ensemble, evolutionary algorithm-optimized ensemble, ensemble formed via random search combined with greedy selection, and a BO ensemble using random initialization. Here, our results demonstrate superior performance of the developed BODE approach. In noise-free scenarios, BODE notably reduces incorrect aleatoric uncertainty and significantly enhances predictive accuracy. Under conditions of 5% and 10% Gaussian noise, BODE adaptively quantifies uncertainty proportional to data noise, achieving up to an 80% reduction in root mean square error compared to baseline methods and producing well-calibrated prediction intervals.

Bayesian optimization↗
DOE OSTI2023

QUIJOTE scientific results – IV. A northern sky survey in intensity and polarization at 10–20 GHz with the multifrequency instrument

ABSTRACT We present QUIJOTE intensity and polarization maps in four frequency bands centred around 11, 13, 17, and 19 GHz, and covering approximately 29 000 deg2, including most of the northern sky region. These maps result from 9000 h of observations taken between May 2013 and June 2018 with the first QUIJOTE multifrequency instrument (MFI), and have angular resolutions of around 1°, and sensitivities in polarization within the range 35–40 µK per 1° beam, being a factor ∼2–4 worse in intensity. We discuss the data processing pipeline employed, and the basic characteristics of the maps in terms of real space statistics and angular power spectra. A number of validation tests have been applied to characterize the accuracy of the calibration and the residual level of systematic effects, finding a conservative overall calibration uncertainty of 5 per cent. We also discuss flux densities for four bright celestial sources (Tau A, Cas A, Cyg A, and 3C274), which are often used as calibrators at microwave frequencies. The polarization signal in our maps is dominated by synchrotron emission. The distribution of spectral index values between the 11 GHz and WMAP 23 GHz map peaks at β = −3.09 with a standard deviation of 0.14. The measured BB/EE ratio at scales of ℓ = 80 is 0.26 ± 0.07 for a Galactic cut |b| > 10°. We find a positive TE correlation for 11 GHz at large angular scales (ℓ ≲ 50), while the EB and TB signals are consistent with zero in the multipole range 30 ≲ ℓ ≲ 150. The maps discussed in this paper are publicly available.

Astronomy & Astrophysics↗