Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “parameter sampling”

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 1,261 records · Page 70
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

Wasserstein Normalized Autoencoder for Anomaly Detection in ProtoDUNE Vertical-Drift Detector

ProtoDUNE Vertical Drift needs a selective triggering algorithm. The detector sits on Earth's surface, so cosmic activity dominates its data. Our goal in this paper is to trigger on neutrino events more robustly than the current deployed Analog-to-Digital Converter Simple Window (ADCSW) model and, eventually, search for signals of Beyond Standard Model (BSM) physics at DUNE as our ultimate North Star objective. As a step towards this goal, we evaluate a Wasserstein Normalized Autoencoder (WNAE) on simulated collection-plane only windows of shape $1\times10\times10$ where Neutrinos act as our BSM-proxy and Cosmic-ray Muons serve as our learned background. The network parameters are fitted using only cosmic-ray muon events as background in order to maintain an unsupervised pipeline. Training uses finite-step Langevin $x^-$ samples, positive-sample reconstruction energy, and an empirical sliced $2$-Wasserstein objective to learn a normalized Boltzmann energy model. We then calibrate on a nominal $5\,\mathrm{Hz}$ operating threshold calculated from cosmic validation data. Both WNAE and ADCSW accept 311 of 194,083 held-out cosmic background events at this $5\,\mathrm{Hz}$ threshold. We found that WNAE accepts 9,677 of 34,634 neutrino-proxy events $(27.9\pm0.24)\%$, compared with 10,076 $(29.1\pm0.24)\%$ for ADCSW, an observed WNAE-minus-ADCSW difference of $-1.15\%$. At another nominal $2\,\mathrm{Hz}$ target threshold, the corresponding efficiencies are $(20.5\pm0.22)\%$ and $(22.6\pm0.22)\%$, respectively. Of the WNAE-selected neutrino proxies at $5\,\mathrm{Hz}$, $(20.8\pm0.4)\%$ of the classified neutrino-proxy events are unique to WNAE, where the uncertainty is an absolute binomial standard error of $0.4\%$.

Zheng, Jake [Chicago U.] (ORCID:0009000218901379)↗
DOE OSTI2021

Spatially inhomogeneous superconductivity in UTe 2

The newly discovered superconductor UTe 2 is a strong contender for a topological spin-triplet state wherein a multicomponent order parameter arises from two nearly degenerate superconducting states. A key issue is whether both of these states intrinsically exist at ambient pressure. Through thermal expansion and calorimetry, we show that UTe 2 at ambient conditions exhibits two detectable transitions only in some samples, and the size of the thermal expansion jump at each transition varies when the measurement is performed in different regions of the sample. This result indicates that the two transitions arise from two spatially separated regions that are inhomogeneously mixed throughout the volume of the sample, each with a discrete superconducting transition temperature (T c ). Notably, samples with higher T c only show a single transition at ambient pressure. Above 0.3 GPa, however, two transitions are invariably observed in ac calorimetry. Our results not only point to a nearly vertical line (constant pressure) in the pressure-temperature phase diagram but also provide a consistent scenario for the sample dependence of UTe 2 .

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2020

Low-mass Group Environments Have No Substantial Impact on the Circumgalactic Medium Metallicity

We explore how environment affects the metallicity of the circumgalactic medium (CGM) using 13 low-mass galaxy groups (two to five galaxies) at &lang;z{sub abs}&rang;=0.25 identified near background quasars. Using quasar spectra from the Hubble Space Telescope/COS and from Keck/HIRES or the Very Large Telescope/UVES, we measure column densities of or determine limits on CGM absorption lines. We use a Markov Chain Monte Carlo approach with Cloudy to estimate metallicities of cool (T ~ 10{sup 4} K) CGM gas within groups and compare them to CGM metallicities of 47 isolated galaxies. Both group and isolated CGM metallicities span a wide range (-2 < [Si/H] < 0), where the mean group (-0.54 ± 0.22) and isolated (-0.77 ± 0.14) CGM metallicities are similar. Group and isolated environments have similar distributions of H I column densities as a function of impact parameter. However, contrary to isolated galaxies, we do not find an anticorrelation between H I column density and the nearest group galaxy impact parameter. We additionally divided the groups by member luminosity ratios (i.e., galaxy–galaxy and galaxy–dwarf groups). While there was no significant difference in their mean metallicities, a modest increase in sample size should allow one to statistically identify a higher CGM metallicity in galaxy–dwarf groups compared to galaxy–galaxy groups. We conclude that either environmental effects have not played an important role in the metallicity of the CGM at this stage and expect that this may only occur when galaxies are strongly interacting or merging or that some isolated galaxies have higher CGM metallicities due to past interactions. Thus, environment does not seem to be the cause of the CGM metallicity bimodality.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Characterizing How Meteorological Forcing Selection and Parameter Uncertainty Influence Community Land Model Version 5 Hydrological Applications in the United States

Despite the increasing use of large-scale Land Surface Models (LSMs) in predicting hydrological responses in extreme conditions, there's a critical gap in understanding the uncertainties in these predictions. This study addresses this gap through a detailed diagnostic evaluation of the uncertainties arising from meteorological forcing selection and model parametrization in hydrological simulations of the Community Land Model version 5 (CLM5). CLM5 is configured at a spatial scale of about 12-km to simulate runoff processes for 464 headwater watersheds, selected from the Catchment Attributes for Large-Sample Studies (CAMELS) dataset to be representative of physiographic and climatic gradients across the conterminous United States. For each watershed, CLM5 is driven by five commonly used gridded forcing datasets in combination with a large ensemble (> 1200) of key CLM5 hydrologic parameters. Our results suggest that uncertainty in CLM5 runoff simulations resulting from both forcing and parametric sources is markedly higher in arid regions, e.g., Great Plains and Midwest regions. Uncertainty in low flow is dominated by parametric uncertainty, while the selection of meteorological forcing contributes more dominantly to high flow and seasonal flows during fall and spring. Our analysis also demonstrates that the selection of forcing datasets and the metrics used to calibrate CLM5 significantly impact the model’s predictive accuracy in extreme event severity for both floods and droughts. Overall, the results from this study highlight the need to understand and account for forcing and parametric uncertainties in CLM5 simulations, particularly for hazard and risk assessments addressing hydrologic extremes.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

Grain Boundary Plane Measurement Using Transmission Electron Microscopy Automated Crystallographic Orientation Mapping for Atom Probe Tomography Specimens

Abstract Grain boundaries are critical in determining the properties of materials, including mechanical stability, conductivity, and corrosion resistance. The specific properties of materials depend not only on the misorientation of the crystals, the three most commonly characterized parameters, but also on the angle of the grain boundary plane between the two crystals, the final two parameters in the five-parameter macroscopic description of the grain boundary. The method presented here allows for the direct measurement of all five parameters of the grain boundary in a transmission electron microscopy specimen of various morphologies. This is especially applicable to atom probe specimens, where only a single-tilt axis is generally available, allowing the crystallographic description to be matched to the detailed chemical data available in the atom probe tomography. This method provides a platform for efficient grain boundary analysis in unique samples, saving operator time and allowing for ease of acquisition and interpretation in comparison with traditional electron diffraction methods.

Materials Science↗
DOE OSTI2020

Nuclear data uncertainty propagation and modeling uncertainty impact evaluation in neutronics core simulation

Uncertainty analysis is a critical requirement in reactor simulation as it is used to quantify the reliability of best-estimate calculation. A comprehensive uncertainty analysis should characterize all sources of uncertainties in a computationally-feasible and scientifically-defendable manner. Here we employ a well-established reduced order modeling (ROM) based uncertainty quantification methodology to propagate uncertainties throughout neutronic calculations. ROM relies on recent advances in randomized data mining techniques applied to large data streams. In our proposed implementation, the nuclear data uncertainties are first propagated from multi-group level through lattice physics calculation to generate few-group parameter uncertainties, described using a vector of mean values and a covariance matrix. Employing an ROM-based compression of the covariance matrix, the few-group uncertainties are then propagated through downstream core simulation in a computationally efficient manner. This straightforward approach, albeit efficient as compared to brute force forward and/or adjoint-based methods, often employs a number of assumptions that have been unquestioned in the literature of neutronic uncertainty analysis. This manuscript argues that these assumptions could introduce another source of uncertainty referred to as modeling uncertainties, whose magnitude needs to be quantified in tandem with nuclear data uncertainties. Thus, our primary goal is to explore the interactions between these two uncertainty sources in order to assess whether modeling uncertainties have an impact on parameter uncertainties. To explore this endeavor, the impact of a number of modeling assumptions on core attributes uncertainties is quantified. The study employs a CANDU reactor model, with Serpent and NEWT as lattice physics solvers and NESTLE-C as core simulator. The modeling assumptions investigated include those related with the uncertainty propagation method employed, e.g., deterministic vs. stochastic, the few-group energy structure employed to represent the cross-sections, the resonance treatment in lattice physics calculation, the reference values for the cross-section, and the number of samples employed to render ROM compression. Results indicate that some of the modeling assumptions could have a non-negligible impact on the core responses propagated uncertainties, highlighting the need for a more comprehensive approach to combine parameter and modeling uncertainties.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2025

Reducing the Parameter Dependency of Phase-Picking Neural Networks with Dice Loss

Training a neural network for picking seismic phase arrivals has been commonly posed as a segmentation problem. It is a highly imbalanced segmentation problem in the sense that the background vastly dominates the foreground because we are trying to pick the optimal single sample point that represents the arrival of a seismic phase in a many seconds long time window. Here, we test the Dice loss, which is a preferred loss function for highly imbalanced image segmentation problems. We show that phase-picking neural networks trained on the Dice loss behave in a binary fashion for which the prediction output is almost always either nearly 1 or nearly 0. This feature removes the strong dependence of data processing workflows on the prediction score threshold, which is an otherwise critical parameter to determine when using neural networks trained on the cross-entropy loss. When strategically used, models trained on the Dice loss can reduce the parameter dependency of machine learning-based seismic monitoring.

58 GEOSCIENCES↗
DOE OSTI2026

The DESI Single Fiber Lens Search. I. Four Thousand Spectroscopically Selected Galaxy–Galaxy Gravitational Lens Candidates

We present 4110 strong gravitational lens candidates, 3887 of which are new discoveries, selected from a sample of 5,837,154 luminous red galaxies (LRGs) observed with the Dark Energy Spectroscopic Instrument (DESI). Candidates are identified via the presence of background ionized oxygen [O II ] nebular emission lines in the foreground LRG spectra, which may originate from the lensing of higher-redshift star-forming galaxies. Using the measured foreground redshift, background redshift, and integrated flux of the background [O II ] doublet, we integrate over impact parameters to compute the probability that each candidate is a lens. We expect 53% of candidates to be true lenses with Einstein radii ranging from 0$^{''}_.$1–4", which can be confirmed with high-resolution imaging. Confirmed strong lenses from this sample will form a valuable cosmological data set, as strong gravitational lensing is the only method to directly measure dark matter halo substructure at cosmological distances. We independently recover the host of the multiply imaged gravitationally lensed type Ia supernova iPTF16geu. Monitoring these lenses for future multiply lensed transients will enable (a) H 0 measurements via time-delay cosmography and (b) substructure measurements via flux ratios.

Karp, Juliana S. M. [Univ. of Washington, Seattle,↗
DOE OSTI2024

In situ study of microstructure evolution and α → ω phase transition in annealed and pre-deformed Zr under hydrostatic loading

The detailed study of the effect of the initial microstructure on its evolution under hydrostatic compression before, during, and after the irreversible α → ω phase transformation and during pressure release in Zr using in situ x-ray diffraction is presented. Two samples were studied: one is plastically pre-deformed Zr with saturated hardness and the other is annealed. Phase transformation α → ω initiates at lower pressure for a pre-deformed sample but for a volume fraction of ω Zr, c > 0.7, a larger volume fraction is observed for the annealed sample. This implies that the proportionality between the athermal resistance to the transformation and the yield strength in the continuum phase transformation theory is invalid; an advanced version of the theory is outlined. Phenomenological plasticity theory under hydrostatic loading is outlined in terms of microstructural parameters, and plastic strain is estimated. During transformation, the first rule is suggested, i.e., the average domain size, microstrain, and dislocation density in ω Zr for c < 0.8 are functions of the volume fraction, c of ω Zr only, which are independent of the plastic strain tensor prior to transformation and pressure. The microstructure is not inherited during phase transformation. Surprisingly, for the annealed sample, the final dislocation density and the average microstrain after pressure release in the ω phase are larger than for the severely pre-deformed sample. The results suggest that an extended experimental basis is required for the predictive models for the combined pressure-induced phase transformations and microstructure evolutions.

36 MATERIALS SCIENCE↗
DOE OSTI2021

On the Order Parameter of the Continuous Phase Transition in the Classical and Quantum Mechanical Limits

The mean field theory is revisited in the classical and quantum mechanical limits. Taking into account the boundary conditions at the phase transition and the third law of the thermodynamics the physical properties of the ordered and disordered phases were reported. The equation for the order parameter predicts the occurrence of a saturation of Ψ 2 = 1 near Θ S , the temperature below the quantum mechanical ground state is reached. The theoretical predictions are also compared with high resolution thermal expansion data of SrTiO 3 monocrystalline samples and other some previous results. An excellent agreement has been found suggesting a universal behavior of the theoretical model to describe continuous structural phase transitions.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2023

Cosmogenic 10 Be in pyroxene: laboratory progress, production rate systematics, and application of the 10 Be– 3 He nuclide pair in the Antarctic Dry Valleys

Here, we present cosmogenic- 10 Be and cosmogenic- 3 He data from Ferrar dolerite pyroxenes in surficial rock samples and a bedrock core from the McMurdo Dry Valleys, Antarctica, with the goal of refining the laboratory methods for extracting beryllium from pyroxene, further estimating the 10 Be production rate in pyroxene and demonstrating the applicability of 10 Be– 3 He in mafic rock. The ability to routinely measure cosmogenic 10 Be in pyroxene will open new opportunities for quantifying exposure durations and Earth surface processes in mafic rocks. We describe scalable laboratory methods for isolating beryllium from pyroxene, which include a simple hydrofluoric acid leaching procedure for removing meteoric 10 Be and the addition of a pH 8 precipitation step to reduce the cation load prior to ion exchange chromatography. 10 Be measurements in pyroxene from the surface samples have apparent 3 He exposure ages of 1–6 Myr. We estimate a spallation production rate for 10 Be in pyroxene, referenced to 3 He, of 3.6 ± 0.2 atoms g –1 yr –1 . 10 Be and 3 He measurements in the bedrock core yield initial estimates for parameters associated with 10 Be and 3 He production by negative-muon capture ($f_{10}^{*}$ = 0.00183 and $f_{3}^{*}f_{C}f_{D}$ = 0.00337). Next, we demonstrate that the 10 Be– 3 He pair in pyroxene can be used to simultaneously resolve erosion rates and exposure ages, finding that the measured cosmogenic-nuclide concentrations in our surface samples are best explained by 2–8 Myr of exposure at erosion rates of 0–35 cm Myr –1 . Finally, given the low 10 Be in our laboratory blanks (average of 5.7 × 10 3 atoms), the reported measurement precision, and our estimated production rate, it should be possible to measure 2 g samples with 10 Be concentrations of 6 × 10 4 and 1.5 × 10 4 atoms g –1 with 5 % and 15 % uncertainty, respectively. With this level of precision, Last Glacial Maximum to Late Holocene surfaces can now be dated with 10 Be in pyroxene. Application of 10 Be in pyroxene, alone or in combination with 3 He, will expand possibilities for investigating glacial histories and landscape change in mafic rock.

58 GEOSCIENCES↗
DOE OSTI2020

Carrier recombination parameters in diamond after surface boron implantation and annealing

In this study, an optical pump–probe technique was used to detect spatial distribution of carrier lifetimes across the thickness of a high-quality diamond device structure. Two samples with as-received and boron implanted surfaces were compared to assess the role of implantation and related processing on carrier recombination mechanisms. It was found that the two implanted surfaces show very different behaviors despite undergoing the same treatment. At one of the surfaces, carrier lifetimes remained relatively unchanged, indicating surface recombination rates in the 102–103 cm/s range. At the other surface, carrier lifetimes were almost a magnitude lower and correlated with the locally detected nitrogen vacancy defect that was attributed to the elevated concentration of residual nitrogen.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2022

Survey of Surveys: I. The largest compilation of radial velocities for the Galaxy

In the present-day panorama of large spectroscopic surveys, the amount, diversity, and complexity of the available data continuously increase. The overarching goal of studying the formation and evolution of our Galaxy is hampered by the heterogeneity of instruments, selection functions, analysis methods, and measured quantities. Here, we present a comprehensive catalogue, the Survey of Surveys (SoS), built by homogeneously merging the radial velocity (RV) determinations of the largest ground-based spectroscopic surveys to date, such as APOGEE, GALAH, Gaia-ESO, RAVE, and LAMOST, using Gaia as a reference. This pilot study serves to prove the concept and to test the methodology that we plan to apply in the future to the stellar parameters and abundance ratios as well. We have devised a multi-staged procedure that includes: (i) the cross match between Gaia and the spectroscopic surveys using the official Gaia cross-match algorithm, (ii) the normalisation of uncertainties using repeated measurements or the three-cornered hat method, (iii) the cross calibration of the RVs as a function of the main parameters on which depend (magnitude, effective temperature, surface gravity, metallicity, and signal-to-noise ratio) to remove trends and zero point offsets, and (iv) the comparison with external high-resolution samples, such as the Gaia RV standards and the Geneva-Copenhagen survey, to validate the homogenisation procedure and to calibrate the RV zero-point of the SoS catalogue. We provide the largest homogenised RV catalogue to date, containing almost 11 million stars, of which about half come exclusively from Gaia and half in combination with the ground-based surveys. We estimate the accuracy of the RV zero-point to be about 0.16–0.31 km s –1 and the RV precision to be in the range 0.05–1.50 km s –1 depending on the type of star and on its survey provenance. We validate the SoS RVs with open clusters from a high resolution homogeneous samples and provide the systemic velocity of 55 individual open clusters. Additionally, we provide median RVs for 532 clusters recently discovered by Gaia data. The SoS is publicly available and ready to be applied to various research projects, such as the study of star clusters, Galactic archaeology, stellar streams, or the characterisation of planet-hosting stars, to name a few. We also plan to include survey updates and more data sources in future versions of the SoS.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2023

Investigating the vertical distribution of the disk as a function of radial action

As heating processes can broaden the distributions of radial actions and the vertical distributions of the Galactic disks, we investigate the vertical distribution of the Galactic disks as a function of radial action based on Apache Point Observatory Galactic Evolution Experiment (APOGEE) and Gaia data in order to deepen our understanding of the formation and heating history of the Galactic disks. The vertical distributions of the thin and thick Galactic disks defined in the chemical plane were fitted with a simple exponential function with a free parameter of scale height in different radial action ranges. Therefore, we were able to analyze the scale height as a function of radial action for different disk populations. We find that the distributions of radial action for both the thin and thick disks can be approximately described by pseudo-isothermal distributions, which give a statistical measurement for the temperature of the disk as indicated by the mean radial action of the star sample. Estimations of the scale heights in different radial action ranges for these pseudo-isothermal distributions of the disks seem to show fixed relationships between radial action J R and scale height h. We describe these relationships with a two-parameter function of h = √(J R /a)+b, where a and b are free parameters. When testing with a three-parameter function of h = α √(J R /a)+b, we find that this two-parameter function describes the thin disk well, but we note the function should be used with care for the thick disk. When comparing the best-fit relationships between the inner and outer disk for both of the thin and thick disks, we find that the relationships are nearly the same for the thin disks but are different for the thick disks. The inner thick disk shows a nearly flattened relationship, while the outer thick disk presents a gradually increasing relationship. This work highlights an alternative way to unveil the heating history of the disks by investigating the relationship between scale height and radial action, as these relationships encode the formation and heating history of the Galactic disks.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2021

Probabilistic effects of porosity and chemical kinetics on the shock initiation of an octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX) based explosive

In this work, we investigate the effects of porosity and chemical reactivity on the shock initiation of an HMX (octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine)-based polymer bonded explosive due to thin flyer impact using two-dimensional mesoscale simulations. Microstructural samples are computationally generated, and the calculated shock response is compared to experiments. The sample dimensions are 0.1×1.05 mm 2 , and each sample accounts for grain scale heterogeneities, including randomized grain morphology and pore spacing. Simulations are performed using the Lawrence Livermore National Laboratory (LLNL) arbitrary Lagrangian–Eulerian multi-physics code ALE3D. Chemistry and equation of states are handled with LLNL's thermochemical code Cheetah. We employ a modified Arrhenius-based chemical kinetics model that scales the temperature exponent fitting parameter with flyer shock conditions. This new model is calibrated to the mean response of experimental shock initiation thresholds for three separate Kapton flyer thicknesses. A simplified shock initiation criterion is proposed, which classifies the macroscale initiation response based on cross-sectional temperature profiles of the shocked sample. This criterion is used to further investigate the probabilistic response of the sample to variations in the mean porosity and chemical kinetics rate, and we find an inverse correlation between the shock pressure and the confidence interval of the experimental threshold velocity. Furthermore, the methodology presented in this work for quantifying the probabilistic nature of shock initiation is material agnostic. These mesoscale studies may be used to inform continuum scale models and expedite the development of future, tailored energetic formulations.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Increasing the power of weak lensing data with multipole-based intrinsic alignment estimators

ABSTRACT It has long been known that galaxy shapes align coherently with the large-scale density field. Characterizing this effect is essential to interpreting measurements of weak gravitational lensing, the deflection of light from distant galaxies by matter overdensities along the line of sight, as it also produces coherent galaxy alignments that we wish to interpret in terms of a cosmological model. Existing direct measurements of intrinsic alignments using galaxy samples with high-quality shape and redshift measurements typically use well-understood but sub-optimal projected estimators, which do not make good use of the information in the data when comparing those estimators to theoretical models. We demonstrate a more optimal estimator, based on a multipole expansion of the correlation functions or power spectra, for direct measurements of galaxy intrinsic alignments. We show that even using the lowest order multipole alone increases the significance of inferred model parameters using simulated and real data, without any additional modelling complexity. We apply this estimator to measurements of parameters of the non-linear alignment model using data from the Sloan Digital Sky survey, demonstrating consistent results with a factor of ∼2 greater precision in parameter fits to intrinsic alignments models. This result is functionally equivalent to quadrupling the survey area, but without the attendant costs – thereby demonstrating the value in using this new estimator in current and future intrinsic alignments measurements using spectroscopic galaxy samples.

Astronomy & Astrophysics↗
DOE OSTI2022

Functional fiber-optic sensors embedded in stainless steel components using ultrasonic additive manufacturing for distributed temperature and strain measurements

The nuclear industry is progressing toward microreactors that can be factory assembled and deployed to remote regions for reliable, scalable power generation. However, the reduced power output requires enhanced monitoring capabilities to reduce staffing and eventually move toward autonomous control to improve the economics of microreactors. The ability to embed sensors such as fiber-optics, which can provide spatially distributed temperature and strain measurements, within microreactor components for real-time health monitoring would be particularly advantageous. Recent advances in ultrasonic additive manufacturing (UAM) have demonstrated the successful embedding of fiber-optics in soft materials such as Al and Cu by placing the sensors in machined cavities and ultrasonically welding over the top with thin foils. This article reports the first successful embedment of fiber-optic sensors and thermocouples within a common nuclear reactor material, SS304, via UAM. UAM parameters were first explored with simple plate geometries before moving to more complex geometries, such as pipes and other test articles for heat pipe–based microreactors. Select samples were sectioned for microscopy to evaluate the sensor/foil and foil/foil interfaces from samples fabricated by using 100% SS304 foils vs. foils plated with a Ni coating to improve UAM bonding. Furthermore, embedding metal-coated, low-bend loss fibers resulted in greatly reduced signal attenuation and adequate compressive residual strain in the embedded region to ensure successful strain coupling. Pipe specimen functional testing was performed by monitoring temperature and strain during transient and steady-state thermal testing.

36 MATERIALS SCIENCE↗
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↗