Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “data normalization”

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 145 records · Page 8
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

National Climate Database (NCDB)

The National Climate Database (NCDB) is a high resolution, bias-corrected climate dataset consisting of the three most widely used variables of solar radiation- global horizontal (GHI), direct normal (DNI), and diffuse horizontal irradiance (DHI)- as well as other meteorological data. The goal of the NCDB is to provide unbiased high temporal and spatial resolution climate data needed for renewable energy modeling. The NCDB is modeled using a statistical downscaling approach with Regional Climate Model (RCM)-based climate projections obtained from the North American Coordinated Regional Climate Downscaling Experiment (NA-CORDEX; linked below). Daily climate projections simulated by the Canadian Regional Climate Model 4 (CanRCM4) forced by the second-generation Canadian Earth System Model (CanESM2) for two Representative Concentration Pathways (RCP4.5 or moderate emissions scenario and RCP8.5 or highest baseline emission scenario) are selected as inputs to the statistical downscaling models. The National Solar Radiation Database (NSRDB) is used to build and calibrate statistical models.

Array↗
DOE OSTI2024

Utah FORGE: Direct Shear Test Data for Investigating Seismic Precursors to Shear Failure of Fractures

This dataset includes results of direct shear tests to investigate the mechanical and geophysical response of dry and saturated fractures in Indiana limestone and Sierra White granite. Direct shear tests were performed on tensile-induced fractures in Indiana limestone and Sierra White granite in a custom water-pressurized chamber. The provided Excel files include the representative seismic wave signals and the normalized wave amplitudes of ultrasonic wave transducers. A link to the published journal article presenting the data and describing the experiment in detail is provided as well.

15 GEOTHERMAL ENERGY↗
DOE OSTI2026

Systematic Uncertainties in the NOvA \nu_\mu-Disappearance Analysis

As the NOvA long-baseline neutrino oscillation experiment takes more data, systematic uncertainty begins to more heavily impact the oscillation analyses. The largest effects are calibration on calorimetric energy, detector modeling on muon energy scale, and near-detector pileup on normalization of the far-detector expectation. Cross-section systematics will be discussed in another poster. Detailed studies have justified previous systematics as too conservative and suggested smaller values. Focused study of detector modeling using dE/dx and external measurements constrain the calibration and energy scale uncertainties. A novel technique is used to assess the impact of activity from neutrino activity piling up over signal, where a simulated neutrino is overlaid into both data and simulation. The difference in efficiency of selecting that single neutrino among data and simulated spills then dominates the normalization systematic.

Alion, Tyler [Sussex U.]↗
DOE OSTI2024

Analysis of Superconducting Magnet Quench Antenna Data

Quenching poses a serious problem for superconducting magnets operating at high currents. It occurs when the material transitions from the superconducting to the normal state, which leads to heating and potential damage to the magnet. To understand and mitigate quenching, the Magnet Department at Fermilab is developing and testing superconducting magnet quench antenna arrays. This study delves into the anomalous events preceding the quench during magnet training by analyzing the collected data. With the moving average and Fast Fourier Transform techniques, we investigate the trends and frequency patterns of the data. Moreover, we introduce an unsupervised anomaly detection algorithm based on Principal Component Analysis and DBSCAN clustering. It can autonomously identify events within background noise, without relying on any predefined event features. Our analysis reveals that the spatio-temporal distribution of these anomalous events has little connection to the quench location, indicating that a majority of them bear no relation to the quenching process.

43 PARTICLE ACCELERATORS↗
DOE OSTI2025

Data for: Climatic Imprint on Interfacially-Controlled Platinum-Palladium Resources

Data package for manuscript "Climatic Imprint on Interfacially-Controlled Platinum-Palladium Resources" by Emily G. Wright, Ivey Wang, Yihang Fang, Elaine D. Flynn, and Jeffrey G. Catalano. This dataset contains adsorption results from experiments designed to investigate the effect of chloride on Pd(II) adsorption to goethite and Pt(II) adsorption to hematite and goethite, including lab experiments, X-ray absorption fine structure spectroscopy, and models of retention within a laterite. See the associated manuscript for full methods information. The file "Wright2025_PtAds_data.csv" contains the target starting Pt concentration (uM), final aqueous Pt and associated error (in uM), calculated adsorbed Pt and associated error (in umol/m2), target and measured aqueous chloride (mM), target aqueous nitrate (mM), final pH, and mineral concentration/loading (g/L). Associated mineral-free controls (mineral loading = 0 g/L) are included; the aqueous Pd error was not calculated and chloride was not measured in every sample. These data appear in Figures 1, S3, S4, S5, S20, and S22 in the associated manuscript. The file "Wright2025_PdAds_data.csv" contains the target starting Pd concentration (uM), final aqueous Pd and associated error (in uM), calculated adsorbed Pd and associated error (in umol/m2), target and measured aqueous chloride (mM), and mineral concentration/loading (g/L). Associated mineral-free controls (mineral loading = 0 g/L) are included; the aqueous Pd error was not calculated and chloride was not measured in every sample. These data appear in Figures 1, S3, S4, S5, and S20 in the associated manuscript. The file "Wright2025_MineralBatches_data.csv" contains the mineral identity and BET specific surface area (m2/g) for every mineral batch synthesized and used in experiments. The annealing time used is listed for hydrothermally annealed goethite. These data appear in Table S2 in the associated manuscript. The file "Wright2025_XRD_data.csv" contains the XRD patterns for every mineral batch synthesized as the counts as a function of two theta (in degrees). See "Wright2025_MineralBatches_data.csv" for more details on specific mineral batches. These data appear in Figure S2 in the associated manuscript. The file "Wright 2025_ZetaPotential_data.csv" contains the measured zeta potentials for samples of goethite (batch G2) at pH 4 the presence of varying amounts of sodium chloride. These data appear in Table S3 in the associated manuscript. The file "Wright2025_XAFSSamples_data.csv" contains the specific mineral batch, measured final aqueous Pd or Pt (uM), measured final aqueous chloride (mM), and estimated adsorbed Pd or Pt (umol/m2) of all XAFS samples. These data appear in Tables S4, S7, S8, and S10 in the associated manuscript. The files "Wright2025_PdXAFS_data.csv" and "Wright2025_PtXAFS_data.csv" contain the normalized spectra of Pd and Pt, respectively, adsorbed to minerals at varying chloride concentrations. See "Wright2025_XAFSSamples_data.csv" for a guide to sample names. Note that "05" in a sample name is equivalent to "0.5". These data appear in Figures 2, S6, S7, S8, S12, S13, and S14 in the associated manuscript. The file "Wright2025_LateriteProfileProfileModelParameters_data.csv" include the ratio of hematite to hematite and goethite in two synthetic, modeled profiles, as well as the modeled surface areas of goethite and hematite as a function of relative depth within the modeled weathering zone. These data were used, in conjunction with equations presented in the paper, to calculate the theoretical concentrations of Pd and Pt (and the resulting Pt/Pd ratio) within the profiles. These data appear in Figure 3 in the associated manuscript. The file "Wright2025_Imagery_data.zip" is a zipped folder containing the TEM and STEM images appear in Figures S18 and S19. Individual files are labeled as either STEM (Fig. S18) or TEM (Fig. S19) with a letter representing the part of the multipart figure.

58 GEOSCIENCES↗
DOE OSTI2026

High-Fidelity Simulation Aerodynamics Dataset of NACA 0012 and 0021 Airfoils

The repository contains time series data of pressure, viscous and moment forces for the NACA 0012 and 0021 airfoils at angles of attack (AOA), 5, 17, 30, 45, 60 and 90 degrees. The data was generated during the high-fidelity simulations of the airfoils using the open-source CFD code, Nalu-Wind (https://github.com/Exawind/nalu-wind). The research objective was to study the effects of mesh resolution and turbulence model on the three-dimensional deep stall aerodynamics of the airfoils, which has been published in the Journal of Turbulence (https://doi.org/10.1080/14685248.2023.2225141). Several meshes of the O-grid type were generated using the modified Pointwise Glyph script of Carrigan (https://github.com/pointwise/AirfoilMesh) by varying the wall normal and spanwise resolutions. Note that, the geometry was extruded by four chord lengths in the spanwise direction. In the paper, we showed that Improved Delayed Detached Eddy Simulation (IDDES) hybrid RANS/LES turbulence model with a minimum of 24 cells per chord length in the spanwise direction is necessary to correctly predict the loads in the deep stall regime. Under the NACA 0012 airfoil directory, the data for several combinations of wall-normal and spanwise resolutions are presented in each of the AOA subdirectories. The corresponding experimental data for the lift and drag polars can be found in the "exp_data" subdirectory. The directory for the NACA 0021 airfoil consists of CFD force data and airfoil polars for Reynolds numbers 2.7e5 and 2.0e6.

17 WIND ENERGY↗
DOE OSTI2024

ThermoPore: Predicting part porosity based on thermal images using deep learning

Part qualification is often a critical and labor-intensive process in additive manufacturing, particularly in the detection of defects such as porosity, which stands to benefit significantly from advancements in machine learning. We present a deep learning approach for quantifying and localizing ex-situ porosity within Laser Powder Bed Fusion fabricated samples utilizing in-situ thermal image monitoring data. Our goal is to build the real time porosity map of parts based on thermal images acquired during the build. The quantification task builds upon the established Convolutional Neural Network model architecture to predict pore count and the localization task leverages the spatial and temporal attention mechanisms of the novel Video Vision Transformer model to indicate areas of expected porosity. Our model for porosity quantification achieved a R 2 score of 0.57 and our model for porosity localization produced an average Intersection over Union (IoU) score of 0.32 and a maximum of 1.0. This work is setting the foundations of part porosity “Digital Twins” based on additive manufacturing monitoring data and can be applied downstream to reduce time-intensive post-inspection and testing activities during part qualification and certification. In addition, we seek to accelerate the acquisition of crucial insights normally only available through ex-situ part evaluation by means of machine learning analysis of in-situ process monitoring data.

Deep learning↗
DOE OSTI2025

Effect of likelihood misspecification in Gaussian process-driven autonomous experimentation

In recent years, several groups have designed Autonomous Experiment (AE) models with the aim of using them as an alternative method for neutron scattering scanning. In an AE, Gaussian processes (GPs) are most frequently used due to their interpretability, their non-parametric nature, their universal approximation, and their closed-form predictive distribution. GPs have two key components, namely, the model for the likelihood of a neutron count knowing the underlying dynamic structure factor and the acquisition function. In this paper, we investigate the impact, on the quality of an AE, of the likelihood and acquisition function choices, in energy scans and (Q, ω) ones, with respect to the signal-over-noise ratio. While we hypothesized that the quality of GP predictions would decrease when the normal to Poisson likelihood approximation breaks down at low count rates, we found that the use of the correct Poisson likelihood does not improve the quality of the data collected, as well as yields very poor results in (Q, ω) scans at low count rates. In fact, the best results are obtained with a combination of normal likelihood, including the observation noise, and the change in variance acquisition function. In addition, we find that the performance, or quality of the predictive distribution, is a misleading measure of efficiency, that is, of the quality of the data collected.

Perryman, David Elliott [Inst. Laue-Langevin (ILL)↗
DOE OSTI2024

Hawai‘i Supernova Flows: a peculiar velocity survey using over a Thousand Supernovae in the near-infrared

ABSTRACT We introduce the Hawai‘i Supernova Flows project and present summary statistics of the first 1217 astronomical transients observed, 668 of which are spectroscopically classified Type Ia Supernovae (SNe Ia). Our project is designed to obtain systematics-limited distances to SNe Ia while consuming minimal dedicated observational resources. To date, we have performed almost 5000 near-infrared (NIR) observations of astronomical transients and have obtained spectra for over 200 host galaxies lacking published spectroscopic redshifts. In this survey paper, we describe the methodology used to select targets, collect/reduce data, calculate distances, and perform quality cuts. We compare our methods to those used in similar studies, finding general agreement or mild improvement. Our summary statistics include various parametrizations of dispersion in the Hubble diagrams produced using fits to several commonly used SN Ia models. We find the lowest dispersions using the SNooPy package’s EBV_model2, with a root mean square deviation of 0.165 mag and a normalized median absolute deviation of 0.123 mag. The full utility of the Hawai‘i Supernova Flows data set far exceeds the analyses presented in this paper. Our photometry will provide a valuable test bed for models of SN Ia incorporating NIR data. Differential cosmological studies comparing optical samples and combined optical and NIR samples will have increased leverage for constraining chromatic effects like dust extinction. We invite the community to explore our data by making the light curves, fits, and host galaxy redshifts publicly accessible.

Do, Aaron (ORCID:0000000334297845)↗
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 [U. Chicago (main)] (ORCID:00090002189↗
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 OSTI2024

Light scattering by V 4 O 7 film across the metal–insulator transition

The experimental study of the angle-resolved hemispherical light scattering by V 4 O 7 film within a broad temperature range across metal–insulator transition reveals complex structural reorganization of the film deposited on the c-cut sapphire crystal. The bidirectional scattering distribution function and the surface autocorrelation function were obtained from scattering data to visualize statistics of the spatially resolved contributions of optical inhomogeneities in normal and lateral directions to the surface. The measurements reveal an anisotropic surface roughness distribution due to the twinned domain structure, with significant anisotropy changes across the phase transition. The V 4 O 7 film deposited on sapphire leads to a polydomain structure, minimizing elastic strain energy with distinct multiscale distributions of surface domains. Near T c , the material shows the lowest roughness but the highest lateral disorder of the surface.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Utah FORGE: Triaxial Direct Shear Results - February 2025

This dataset contains results from nine triaxial direct shear tests conducted by Los Alamos National Laboratory on samples from FORGE Well 16A(78)-32. The primary objectives of this work were to determine the shear strength in both intact and residual states, evaluate dilation against displacement, assess permeability in relation to displacement, time, and normal stress, understand the relationship between aperture and normal stress, and monitor the effluent chemistry as a function of time. The data includes time-series measurements of stress, displacement, permeability, and effluent chemistry, with and without experimental dilution corrections. Additional materials include profilometry data, photographic documentation of the experimental setups and apparatus, and test notes. The dataset is organized into folders corresponding to each test, containing hydromechanical data, effluent chemistry measurements, and images. The hydromechanical data consists of detailed time-series records capturing parameters such as shear force, confining pressure, permeability, and temperature. Effluent chemistry data tracks fluid composition changes over time. Also included are conference papers, presentation slides, and a summary document outlining the experiments.

15 GEOTHERMAL ENERGY↗
DOE OSTI2022

Internet of Things Data Characterization Process: Pattern of Life Behavioral Data Study

The HoneyBee™ TARDIS LDRD team completed a data scoping study that identified the initial processes and procedures to baseline the normal and expected behaviors during operability and interoperability of Internet of Things (IoT) device networks. This research is the initial step in developing a process (or methodology) to inform a much broader information framework incorporating machine learning to determine device pattern-of-life which enables the detection of abnormal IoT behaviors on an individual device, as well as in the context of a larger network.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

From models to reality: a systematic review on simulated and measured residential heat pump energy savings

High-performance HVAC solutions are central to residential energy management. A substantial share of these are electric, reversible-cycle systems, with heat pumps representing the largest portion of current and near-term adoption. This review synthesizes peer-reviewed and grey literature on residential space heating and cooling heat pumps. The academic literature is dominated by modeling (73.8%), with limited field measurement (13.1%). Grey literature from United States serve as a supplemental resource providing measured savings. Conversions from electric-resistance heating consistently show the largest site energy reductions, while oil/propane baselines yield moderate savings, and gas baseline scenario often deliver small and region-dependent savings. This study cross-checks the grey literature measured data with simulation data filtered from the ResStock dataset. The comparison indicates a discrepancy between simulations and measured data: simulated site EUIs are typically lower than measured EUIs, but percentage energy savings fall in similar ranges, implying simulations capture directional effects while underestimating energy use. Factors associated with variability and model–measurement differences include system characterization and control representation (e.g., backup heat engagement, thermostat/setpoint strategies, commissioning/installation quality), occupant behavior, weather normalization, metering scope, and envelope characterization. This paper also outlines the proposed methodology for comparing simulation and measured data for heat pumps. It emphasizes the metrics used for comparison and units harmonization, building characteristics matching, and compact metadata are needed for simulations to match measured data. The proposed methodology is expected to improve the credibility of simulated savings as measured evidence grows.

Yu, Lili↗
DOE OSTI2026

ZTF-SEDm Type Ia supernova sample for Twins Embedding spectrophotometric standardization

Aims. This paper has two aims: the first aim is to build a large homogeneous spectrophotometric sample of Type Ia supernovae (SNe Ia) from the second data release of the Zwicky Transient Facility (ZTF DR2). We used the spectrum sample from the low-resolution ( R ∼ 100) SEDmachine (SEDm) Integral Field Spectrograph (IFS) that gathered 3069 spectra. This is one of the largest samples of such collections that can attempt to reproduce the Twins Embedding (TE) spectrophotometric standardization method. This is our second objective. The method was developed based on high-quality spectra from 200 SNe Ia of the Nearby Supernova factory (SNfactory) and led to an exceptionally low value of 0.073 mag for the intrinsic scatter. Methods. As the SEDm is not designed as a spectrophotometric instrument, we first improved the flux-calibration accuracy of the SN Ia spectrum sample using the ZTF photometric data, which were calibrated at the percent level. We corrected the spectra for second-order polynomials, fitted by comparing the synthetic photometry in the ZTF g , r , i filters with the light-curve (LC) data. We then applied the three steps of the TE parameterization to a subset of 783 ZTF SN spectra near maximum light while comparing results from SNfactory and ZTF. We finally analyzed the standardization methods based on the TE parameters. Results. The precision of the phase-correction model, which is the first step of the TE, is estimated at 0.01 mag in g band based on ZTF data. Despite the challenge posed by the spectrum-extraction pipeline associated with the SEDm (flux calibration, leftover host signal, low signal-to-noise ratio, and low resolution), we applied a first standardization in color based on the second step of the TE, called read between the lines (RBTL), to the ZTF sample. We reached a Hubble residual scatter of 0.153 mag, all in normalized median absolute deviation, which is to be compared to the ∼0.11 mag obtained with the SNfactory data. The SALT color and stretch standardization reaches a scatter of 0.164 mag for the same ZTF SN Ia sample, and its host steps are ∼0.1 mag and zero for RBTL. When considering the scatter due to the redshift error and flux calibration error, we estimated a RBTL scatter of ∼0.129 mag for this ZTF sample as an upper limit because we identified an additional contribution from a systematic error in color. We tested the standardization based on the nonlinear TE parameters, and, as expected from the low spectrum quality, it did not improve the overall dispersion. Conclusions. We release 1897 flux calibrated spectra of 1607 SNe Ia with an estimated photometric accuracy of 0.07 mag. We further demonstrate that some amount of spectrophotometric SN Ia standardization can be done with limited-quality spectra. The RBTL standardization is more efficient than that of SALT with one parameter less, and the resulting host steps are consistent with zero. This makes it less prone to astrophysical bias. For future spectroscopic surveys, targeting the extraction pipeline for a thorough flux calibration and good signal-to-noise ratio would enable us to compute the full TE standardization, which would further reduce the scatter in the distance estimate.

Ganot, C↗
DOE OSTI2024

GAINN: The Galaxy Assembly and Interaction Neural Networks for High-redshift JWST Observations

We present the Galaxy Assembly and Interaction Neural Networks (Gainn), a series of artificial neural networks for predicting the redshift, stellar mass, halo mass, and mass-weighted age of simulated galaxies based on James Webb Space Telescope (JWST) photometry. Our goal is to determine the best neural network for predicting these variables at 11 < z < 15. The parameters of the optimal neural network can then be used to estimate these variables for real, observed galaxies. The inputs of the neural networks are JWST filter magnitudes of a subset of five broadband filters (F150W, F200W, F277W, F356W, and F444W) and two medium-band filters (F162M and F182M). We compare the performance of the neural networks using different combinations of these filters, as well as different activation functions and numbers of layers. The best neural network predicted redshift with a normalized rms error of $0.010^{+0.003}_{-0.001}$, stellar mass with rms = $0.089^{+0.044}_{-0.022}$, halo mass with a mean-squared error of $0.022^{+0.014}_{-0.008}$, and mass-weighted age with rms = $12.466^{+5.065}_{-2.408}$. We also test the performance of Gainn on real data from MACS0647JD, an object observed by JWST. Predictions from Gainn for the first projection of the object (JD1) have normalized bias $\langle$Δz$\rangle$ < 0.00228, which is significantly smaller than found with template-fitting methods. We find that the optimal filter combination is F277W, F356W, F162M, and F200W when considering both theoretical accuracy and observational resources from JWST.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Surrogate-driven design optimization with uncertainty constraints in Monte Carlo simulations

In multi-objective design tasks, the computational cost increases rapidly when high-fidelity simulations are used to evaluate objective functions. Surrogate models help mitigate this cost by approximating the simulation output, simplifying the design process. However, under high uncertainty, surrogate models trained on noisy data can produce inaccurate predictions, as their performance depends heavily on the quality of training data. This study investigates the impact of data uncertainty on two multi-objective design problems modelled using Monte Carlo transport simulations: a neutron moderator and an ion-to-neutron converter. For each, a grid search was performed using five different tally uncertainty levels to generate training data for neural network surrogate models. These models were then optimized using NSGA-III. The recovered Pareto-fronts were analyzed across uncertainty levels: in the moderator problem, normalized hypervolume dropped from 0.886 at 1.0% uncertainty to 0.748 at 10% uncertainty, while in the converter problem it remained near 0.50 for all cases. Average simulation times were also compared to evaluate the trade-off between accuracy and computational cost. Results show that the influence of simulation uncertainty is strongly problem-dependent. In the neutron moderator case, higher uncertainties led to exaggerated objective sensitivities and distorted Pareto-fronts, reducing normalized hypervolume. In contrast, the ion-to-neutron converter task was less affected—low-fidelity simulations produced results similar to those from high-fidelity data. These findings suggest that a fixed-fidelity approach is not optimal. Surrogate models can recover the Pareto-front under noisy conditions, and multi-fidelity studies help identify suitable uncertainty levels for each problem to balance efficiency and accuracy.

07 ISOTOPE AND RADIATION SOURCES↗