Search NASA⌕ Search

SEARCH · Search NASA

Results for “Generative modeling”

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 397 records · Page 22

Distributed Generation Market Demand Model (dGen): Documentation

The Distributed Generation Market Demand model (dGen) is a geospatially rich, bottom-up, market-penetration model that simulates the potential adoption of distributed energy resources (DERs) for residential, commercial, and industrial entities in the continental United States through 2050. The National Renewable Energy Laboratory (NREL) developed dGen to analyze the key factors that will affect future market demand for distributed solar, wind, storage, and other DER technologies in the United States. The new model builds off, extends, and replaces NREL's SolarDS model (Denholm et al. 2009a), which simulates the market penetration of distributed PV only. Unlike the SolarDS model, dGen can model various DER technologies under one platform--it currently can simulate the adoption of distributed solar (the dSolar module) and distributed wind (the dWind module) and link with the ReEDS capacity expansion model (Appendix C). The underlying algorithms and datasets in dGen, which improve the representation of customer decision making as well as the spatial resolution of analyses (Figure ES-1), also are improvements over SolarDS.

17 WIND ENERGY↗

Mesoscopic modeling of DNA denaturation rates: Sequence dependence and experimental comparison

Using rare event simulation techniques, we calculated DNA denaturation rate constants for a range of sequences and temperatures for the Peyrard-Bishop-Dauxois (PBD) model with two different parameter sets. We studied a larger variety of sequences compared to previous studies that only consider DNA homopolymers and DNA sequences containing an equal amount of weak AT- and strong GC-base pairs. Our results show that, contrary to previous findings, an even distribution of the strong GC-base pairs does not always result in the fastest possible denaturation. In addition, we applied an adaptation of the PBD model to study hairpin denaturation for which experimental data are available. This is the first quantitative study in which dynamical results from the mesoscopic PBD model have been compared with experiments. Our results show that present parameterized models, although giving good results regarding thermodynamic properties, overestimate denaturation rates by orders of magnitude. We believe that our dynamical approach is, therefore, an important tool for verifying DNA models and for developing next generation models that have higher predictive power than present ones.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Integrating AlphaFold and deep learning for atomistic interpretation of cryo-EM maps

Abstract Interpretation of cryo-electron microscopy (cryo-EM) maps requires building and fitting 3D atomic models of biological molecules. AlphaFold-predicted models generate initial 3D coordinates; however, model inaccuracy and conformational heterogeneity often necessitate labor-intensive manual model building and fitting into cryo-EM maps. In this work, we designed a protein model-building workflow, which combines a deep-learning cryo-EM map feature enhancement tool, CryoFEM (Cryo-EM Feature Enhancement Model) and AlphaFold. A benchmark test using 36 cryo-EM maps shows that CryoFEM achieves state-of-the-art performance in optimizing the Fourier Shell Correlations between the maps and the ground truth models. Furthermore, in a subset of 17 datasets where the initial AlphaFold predictions are less accurate, the workflow significantly improves their model accuracy. Our work demonstrates that the integration of modern deep learning image enhancement and AlphaFold may lead to automated model building and fitting for the atomistic interpretation of cryo-EM maps.

59 BASIC BIOLOGICAL SCIENCES↗

Deblending galaxies with variational autoencoders: A joint multiband, multi-instrument approach

ABSTRACT Blending of galaxies has a major contribution in the systematic error budget of weak-lensing studies, affecting photometric and shape measurements, particularly for ground-based, deep, photometric galaxy surveys, such as the Rubin Observatory Legacy Survey of Space and Time (LSST). Existing deblenders mostly rely on analytic modelling of galaxy profiles and suffer from the lack of flexible yet accurate models. We propose to use generative models based on deep neural networks, namely variational autoencoders (VAE), to learn probabilistic models directly from data. We train a VAE on images of centred, isolated galaxies, which we reuse, as a prior, in a second VAE-like neural network in charge of deblending galaxies. We train our networks on simulated images including six LSST bandpass filters and the visible and near-infrared bands of the Euclid satellite, as our method naturally generalizes to multiple bands and can incorporate data from multiple instruments. We obtain median reconstruction errors on ellipticities and r-band magnitude between ±0.01 and ±0.05, respectively, in most cases, and ellipticity multiplicative bias of 1.6 per cent for blended objects in the optimal configuration. We also study the impact of decentring and prove the method to be robust. This method only requires the approximate centre of each target galaxy, but no assumptions about the number of surrounding objects, pointing to an iterative detection/deblending procedure we leave for future work. Finally, we discuss future challenges about training on real data and obtain encouraging results when applying transfer learning.

Arcelin, Bastien↗

Measurement of muon plus proton final states in ν μ interactions on hydrocarbon at < E ν > = 4.2 GeV

A study of charged-current muon neutrino scattering on hydrocarbon in which the final state includes a muon, at least one proton, and no pions is presented. Although this signature has the topology of neutrino quasielastic scattering from neutrons, the event sample contains contributions from quasielastic and inelastic processes where pions are absorbed in the nucleus. The analysis accepts events with muon production angles up to 70° and proton kinetic energies greater than 110 MeV. The cross section, when based completely on hadronic kinematics, is well described by a relativistic Fermi gas nuclear model including the neutrino event generator modeling for inelastic processes and particle transportation through the nucleus. This is in contrast to the quasielastic cross section based on muon kinematics, which is best described by an extended model that incorporates multinucleon correlations. As a result, this measurement guides the formulation of a complete description of neutrino-nucleus interactions that encompasses the hadronic as well as the leptonic aspects of this process.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Evaluation of Black Carbon Estimations in Global Aerosol Models

We evaluate black carbon (BC) model predictions from the AeroCom model intercomparison project by considering the diversity among year 2000 model simulations and comparing model predictions with available measurements. These model-measurement intercomparisons include BC surface and aircraft concentrations, aerosol absorption optical depth (AAOD) from AERONET and OMI retrievals and BC column estimations based on AERONET. In regions other than Asia, most models are biased high compared to surface concentration measurements. However compared with (column) AAOD or BC burden retreivals, the models are generally biased low. The average ratio of model to retrieved AAOD is less than 0.7 in South American and 0.6 in African biomass burning regions; both of these regions lack surface concentration measurements. In Asia the average model to observed ratio is 0.6 for AAOD and 0.5 for BC surface concentrations. Compared with aircraft measurements over the Americas at latitudes between 0 and 50N, the average model is a factor of 10 larger than observed, and most models exceed the measured BC standard deviation in the mid to upper troposphere. At higher latitudes the average model to aircraft BC is 0.6 and underestimate the observed BC loading in the lower and middle troposphere associated with springtime Arctic haze. Low model bias for AAOD but overestimation of surface and upper atmospheric BC concentrations at lower latitudes suggests that most models are underestimating BC absorption and should improve estimates for refractive index, particle size, and optical effects of BC coating. Retrieval uncertainties and/or differences with model diagnostic treatment may also contribute to the model-measurement disparity. Largest AeroCom model diversity occurred in northern Eurasia and the remote Arctic, regions influenced by anthropogenic sources. Changing emissions, aging, removal, or optical properties within a single model generated a smaller change in model predictions than the range represented by the full set of AeroCom models. Upper tropospheric concentrations of BC mass from the aircraft measurements are suggested to provide a unique new benchmark to test scavenging and vertical dispersion of BC in global models.

09 BIOMASS FUELS↗

JMOCUP Physics Depletion Calculations for the As-Run AGR-5/6/7 TRISO Particle Experiment in ATR Northeast Flux Trap

This ECAR documents the as-run Jim Sterbentz’s MCNP ORIGEN Coupled Utility Program (JMOCUP) physics depletion calculation and the calculated results for the AGR-5/6/7 irradiation experiment in the Advanced Test Reactor (ATR). AGR 5/6/7 was irradiated for nine power cycles in the northeast flux trap. The depletion calculations were performed to provide input data for a variety of other engineering analyses supporting the AGR-5/6/7 experiment along with post-irradiation characterization of the tri-structural isotropic (TRISO) particle fuel compacts. Detailed full-core MCNP models and Oak Ridge Isotope Generation (ORIGEN2) radionuclide generation models were specifically developed as part of the JMOCUP Monte Carlo depletion calculations. The MCNP ORIGEN2 computer codes were coupled using the well-established JMOCUP utility modules to couple the two codes and run the depletion calculations. The physics calculations were performed in support of the Advanced Gas Reactor (AGR) program.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Principles of Bayesian Ensemble Creation: Theoretical and Empirical Observations

Bayesian methods are theoretically optimal given their underlying assumptions. Bayesian model averaging (BMA) is generally considered the standard model for creating ensembles of learners using Bayesian methods, but this technique is often out-performed by more ad hoc methods in empirical studies. It has been proposed that Bayesian model averaging struggles in practice because it accounts for uncertainty about which model is the data-generating model but still operates under the assumption that only one of them is. Thus, BMA is unable to enrich the hypothesis space as effectively as other ensemble creation methods, serving as a form of optimal model selection, not optimal model combination. This work provides empirical verification for this hypothesis using both synthetic and real-world data. We propose several different approaches for Bayesian model combination, and demonstrate that they overcome this problem with BMA. We empirically show that even the most simplistic of Bayesian model combination strategies can compete with the traditional ad hoc techniques of bagging, boosting and stacking, as well as significantly outperforming BMA over a wide variety of cases. We also explore the implications of this observation as it relates to the effectiveness of ensemble learning.

97 MATHEMATICS AND COMPUTING↗

Challenging Geostatistical Methods To Represent Heterogeneity in CO2 Reservoirs Under Residual Trapping

Abstract Geostatistical methods based on two-point spatial-bivariate statistics have been used to model heterogeneity within computational studies of the dispersion of contaminants in groundwater reservoirs and the trapping of CO2 in geosequestration reservoirs. The ability of these methods to represent fluvial architecture, commonly occurring in such reservoirs, has been questioned. We challenged a widely used two-point spatial-bivariate statistical method to represent fluvial heterogeneity in the context of representing how reservoir heterogeneity affects residual trapping of CO2 injected for geosequestration. A more rigorous model for fluvial architecture was used as the benchmark in these studies. Both the geostatistically generated model and the benchmark model were interrogated, and metrics for the connectivity of high-permeability preferential flow pathways were quantified. Computational simulations of CO2 injection were performed, and metrics for CO2 dynamics and trapping were quantified. All metrics were similar between the two models. The percentage of high-permeability cells in spanning connected clusters (percolating clusters) was similar because percolation is strongly dependent upon proportions, and the same proportion of higher permeability cross-strata was specified in generating both models. The CO2 plume dynamics and residual trapping metrics were similar because they are largely controlled by the occurrence of percolating clusters. The benchmark model represented more features of the fluvial architecture and, depending on context, representing those features may be quite important, but the simpler geostatistical model was able to adequately represent fluvial reservoir architecture within the context and within the scope of the parameters represented here.

defects, mechanical behavior, carbon sequestration↗

AI-assisted rapid crystal structure generation towards a target local environment

In material design, traditional crystal structure prediction approaches are expensive as they require extensive structural sampling through expensive energy minimization methods. Emerging artificial intelligence (AI) generative models have shown great promise in rapidly generating realistic crystals, but they typically handle only a few tens of atoms per unit cell. To overcome this limitation, we introduce a symmetry-informed approach, the Local Environment Geometry-Oriented Crystal Generator (LEGO-xtal). Our method generates initial structures using AI models trained on an augmented dataset, and then optimizes them using structure descriptors rather than energy-based optimization. We demonstrate its effectiveness by expanding from 25 known low-energy sp2 carbon allotropes to over 1700, all within 0.5 eV/atom of the ground-state energy of graphite. This framework offers a generalizable strategy for the targeted design of materials with modular building blocks, such as metal-organic frameworks and battery materials.

Ridwan, Osman Goni [University of North Carolina a↗

Symplectic machine learning model for fast simulation of space-charge effects

Symplectic simulation of space-charge effects is crucial for the design and operation of high-intensity particle accelerators. Traditional methods for simulating these effects are often computationally expensive, resulting in significant overhead. In this work, we introduce a generative model based on a U-Net architecture within a generative adversarial network framework to efficiently simulate space-charge effects. The model is trained to predict the transverse multiparticle space-charge Hamiltonian, which can be physically computed using a gridless spectral method. The one-step symplectic transverse transfer map for the particles is then obtained by differentiating the predicted Hamiltonian. Benchmarking results demonstrate that this generative model achieves an order of magnitude higher computational efficiency compared to the spectral method, providing a highly efficient alternative for simulating space-charge effects with a large number of particles. By maintaining symplecticity, the model effectively preserves the phase-space structure and mitigates nonphysical errors in long-term simulations. This model has been integrated into jutrack, a novel autodifferentiable accelerator modeling code developed in the julia programming language.

Beam code development & simulation techniques↗

CaloChallenge 2022: a community challenge for fast calorimeter simulation

Here, we present the results of the ‘Fast Calorimeter Simulation Challenge 2022’—the CaloChallenge. We study state-of-the-art generative models on four calorimeter shower datasets of increasing dimensionality, ranging from a few hundred voxels to a few tens of thousand voxels. The 31 individual submissions span a wide range of current popular generative architectures, including variational autoencoders (VAEs), generative adversarial networks (GANs), normalizing flows, diffusion models, and models based on conditional flow matching. We compare all submissions in terms of quality of generated calorimeter showers, as well as shower generation time and model size. To assess the quality we use a broad range of different metrics including differences in one-dimensional histograms of observables, KPD/FPD scores, AUCs of binary classifiers, and the log-posterior of a multiclass classifier. The results of the CaloChallenge provide the most complete and comprehensive survey of cutting-edge approaches to calorimeter fast simulation to date. In addition, our work provides a uniquely detailed perspective on the important problem of how to evaluate generative models. As such, the results presented here should be applicable for other domains that use generative AI and require fast and faithful generation of samples in a large phase space.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

The Sylvester Graphical Lasso (SyGlasso)

This paper introduces the Sylvester graphical lasso (SyGlasso) that captures multiway dependencies present in tensor-valued data. The model is based on the Sylvester equation that defines a generative model. The proposed model complements the tensor graphical lasso that imposes a Kronecker sum model for the inverse covariance matrix by providing an alternative Kronecker sum model that is generative and interpretable. A nodewise regression approach is adopted for estimating the conditional independence relationships among variables. The statistical convergence of the method is established, and empirical studies are provided to demonstrate the recovery of meaningful conditional dependency graphs. Here we apply the SyGlasso to an electroencephalography (EEG) study to compare the brain connectivity of alcoholic and nonalcoholic subjects. Here, we demonstrate that our model can simultaneously estimate both the brain connectivity and its temporal dependencies.

Wang, Yu↗

Synthetic Scientific Image Generation with VAE, GAN, and Diffusion Model Architectures

Generative AI (genAI) has emerged as a powerful tool for synthesizing diverse and complex image data, offering new possibilities for scientific imaging applications. This review presents a comprehensive comparative analysis of leading generative architectures, ranging from Variational Autoencoders (VAEs) to Generative Adversarial Networks (GANs) on through to Diffusion Models, in the context of scientific image synthesis. We examine each model's foundational principles, recent architectural advancements, and practical trade-offs. Our evaluation, conducted on domain-specific datasets including microCT scans of rocks and composite fibers, as well as high-resolution images of plant roots, integrates both quantitative metrics (SSIM, LPIPS, FID, CLIPScore) and expert-driven qualitative assessments. Results show that GANs, particularly StyleGAN, produce images with high perceptual quality and structural coherence. Diffusion-based models for inpainting and image variation, such as DALL-E 2, delivered high realism and semantic alignment but generally struggled in balancing visual fidelity with scientific accuracy. Importantly, our findings reveal limitations of standard quantitative metrics in capturing scientific relevance, underscoring the need for domain-expert validation. We conclude by discussing key challenges such as model interpretability, computational cost, and verification protocols, and discuss future directions where generative AI can drive innovation in data augmentation, simulation, and hypothesis generation in scientific research.

Generative Adversarial Networks↗

Distributed Generation Market Demand (dGen) model

The Distributed Generation Market Demand (dGen) model simulates customer adoption of distributed energy resources (DERs) for residential, commercial, and industrial entities in the United States or other countries through 2050. The dGen model can be used for identifying the sectors, locations, and customers for whom adopting DERs would have a high economic value, for generating forecasts as an input to estimate distribution hosting capacity analysis, integrated resource planning, and load forecasting, and for understanding the economic or policy conditions in which DER adoption becomes viable, and for illustrating sensitivity to market and policy changes such as retail electricity rate structures, net energy metering, and technology costs.

Array↗

Generative large language models for predictive maintenance planning

Maintenance planning and the generation of necessary components for tasks can prove time-consuming and complex. Automating the creation of recurring or similar tasks by leveraging previous planning packages and data, while uncovering insights to automate planning package generation, presents an opportunity to conserve valuable time and resources. This work aims to harness the textual and probabilistic capabilities of large language models (LLMs) to automate the generation of planning packages. Utilizing diverse data sources ranging from raw data to handwritten text, both singular and collaborative LLMs are trained and tested. Results demonstrate their capability to generate essential planning package components, effectively replicating the statistical patterns in the data. This demonstrates the use of these tools inside a digital asset for automated planning. This work outlines a methodology for constructing datasets, a training suite, and evaluation methods for LLM-based textual and conversational planning tools utilized in an asset digital twin. Results indicate that the fine-tuned models generate estimated planning information within the statistical ranges observed in real maintenance data. The models achieve high accuracy (>90%) in document question-answering and instruction generation tasks. Furthermore, the conversational retrieval-augmented generation (RAG) assistant system achieves 100% document retrieval accuracy, while conversational information capture exceeds 98% across the majority of work-package assistant modules.

97 MATHEMATICS AND COMPUTING↗

A Survey of Machine Learning-Based Physics Event Generation

Event generators in high-energy nuclear and particle physics play an important role in facilitating studies of particle reactions. We survey the state of the art of machine learning (ML) efforts at building physics event generators. We review ML generative models used in ML-based event generators and their specific challenges, and discuss various approaches of incorporating physics into the ML model designs to overcome these challenges. Finally, we explore some open questions related to super-resolution, fidelity, and extrapolation for physics event generation based on ML technology.

Alanazi, Yasir↗