Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Adaptation models”

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 1,297 records · Page 72
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

One hundred years into the study of ecotypes, new advances are being made through large-scale field experiments in perennial plant systems

A hundred years after Turesson first clearly described how locally adaptive variation is distributed within species, plant biologists are making major breakthroughs in our understanding of mechanisms underlying adaptation from local populations to the scale of continents. Although the genetics of local adaptation has typically been studied in smaller reciprocal transplant experiments, here it is being evaluated with whole genomes in large-scale networks of common garden experiments with perennial switchgrass and poplar trees. These studies support the hypothesis that a complex combination of loci, both with and without adaptive trade-offs, underlies local adaptation and that hybridization and adaptive introgression play a key role in the evolution of these species. Future studies incorporating high-throughput phenotyping, gene expression, and modeling will be used to predict responses of these species to climate change.

Trade-off↗
DOE OSTI2023

Limited potential of irrigation to prevent potato yield losses in Germany under climate change

Climate models project higher temperatures and a substantial net water deficit during the summer months over Germany until the end of the 21st century. The agricultural sector is particularly vulnerable to changing weather and climate conditions. In this study, we examine the role of irrigation for potato yields as a climate change adaptation measure in Northeast Lower Saxony (Germany). The region represents the largest irrigated area in Germany and is one of the main growing areas of potatoes. We develop a statistical multivariate regression model to investigate whether regional climate and irrigation data at the monthly and county level are suitable to describe potato yield variability from 1978 to 2018. Irrigation is estimated based on the climatic water balance and crop water demand. These estimates are calibrated against irrigation data from local agencies to account for realistic monthly water withdrawals. We analyze future yield changes based on climate model projections for the 21st century and different irrigation scenarios. We find that the potato yield will decrease significantly by 18% on average until 2050 with climate projection uncertainty of ±6.4% under the high emission scenario RCP 8.5 and without irrigation. Irrigation at current levels could compensate for about 54% of these yield losses. Doubling the amount of irrigation minimizes yield losses by as much as 80%. Under the low (middle) emission scenario RCP 2.6 (RCP 4.5), potato yield are projected to decrease moderately by 4.5% ± 3.4% (5.4% ± 2.8%) with irrigation at current levels. We show that the yield losses mainly originate from higher temperatures during the summer. Our analysis indicates that the capacity of irrigation to minimize yield losses is limited. Strengthening climate change mitigation efforts may be more effective in preventing yield losses. However, the effect of elevated CO 2 might buffer the high yield losses, especially in RCP 8.5. The studies supports farmers and decision makers in assessing future risks in potato yields. It can guide policymakers in allocating economic resources to cope with climate change.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2024

Automated Operational Forecasting of Monsoon Low Pressure Systems

Monsoon low pressure systems (LPSs) are the dominant rain-bearing weather system of South Asia, often producing extreme precipitation and hydrological disasters in a region inhabited by nearly two billion people. Despite the importance of these storms, no operational system has automatically identified and tracked LPS in real time in numerical weather prediction model output; many commonly used vortex-tracking algorithms are ill suited for monsoon LPS because of the weak winds and cold cores of these systems. Here, we describe a new system that uses optimized algorithms to identify monsoon LPS in short- to medium-range forecasts from the U.S. Global Ensemble Forecast System (GEFS) and a version of the deterministic Global Forecast System (GFS) adapted and used operationally by the Indian Institute of Tropical Meteorology (IITM). We also assess the historical performance of these models in forecasting South Asian monsoon LPS, comparing this with the performance of the Integrated Forecasting System of the ECMWF. We assess the accuracy of model predictions of LPS genesis, position, intensity, and precipitation rates for forecast lead times of 1–5 days, yielding quantitative information on model biases to guide operational forecasters and disaster managers. The system we introduce here could be extended to other low-latitude regions affected by dynamically weak, heavily precipitating atmospheric vortices that are often not included in tropical cyclone inventories.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Towards exascale for wind energy simulations

We examine large-eddy-simulation modeling approaches and computational performance of two open-source computational fluid dynamics codes for the simulation of atmospheric boundary layer flows that are of direct relevance to wind energy production. The first code, NekRS, is a high-order, unstructured-grid, spectral element code. The second code, AMR-Wind, is a second-order, block-structured, finite-volume code with adaptive mesh refinement capabilities. The objective of this study is to co-develop these codes in order to improve model fidelity and performance for each. These features will be critical for running ABL-based applications such as wind farm analysis on advanced computing architectures. To this end, we investigate the performance of NekRS and AMR-Wind on the Oak Ridge Leadership Facility supercomputers Summit, using 4 to 800 nodes (24 to 4,800 NVIDIA V100 GPUs), and Crusher, the testbed for the Frontier exascale system, using 18 to 384 Graphics Compute Dies on AMD MI250X GPUs. We compare strong- and weak-scaling capabilities, linear solver performance, and time to solution. We also identify leading inhibitors to parallel scaling.

17 WIND ENERGY↗
DOE OSTI2020

Integrated network modeling approach defines key metabolic responses of soil microbiomes to perturbations

The soil environment is constantly changing due to shifts in soil moisture, nutrient availability and other conditions. To contend with these changes, soil microorganisms have evolved a variety of ways to adapt to environmental perturbations, including regulation of gene expression. However, it is challenging to untangle the complex phenotypic response of the soil to environmental change, partly due to the absence of predictive modeling frameworks that can mechanistically link molecular-level changes in soil microorganisms to a community’s functional phenotypes (or metaphenome). Towards filling this gap, we performed a combined analysis of metabolic and gene co-expression networks to explore how the soil microbiome responded to changes in soil moisture and nutrient conditions and to determine which genes were expressed under a given condition. Our integrated modeling approach revealed previously unknown, but critically important aspects of the soil microbiomes’ response to environmental perturbations. Incorporation of metabolomic and transcriptomic data into metabolic reaction networks identified condition-specific signature genes that are uniquely associated with dry, wet, and glycine-amended conditions. A subsequent gene co-expression network analysis revealed that drought-associated genes occupied more central positions in a network model of the soil community, compared to the genes associated with wet, and glycine-amended conditions. These results indicate the occurrence of system-wide metabolic coordination when soil microbiomes cope with moisture or nutrient perturbations. Importantly, the approach that we demonstrate here to analyze large-scale multi-omics data from a natural soil environment is applicable to other microbiome systems for which multi-omics data are available.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Active Tuning of Optical Constants in the Visible–UV: Praseodymium-Doped Ceria—a Model Mixed Ionic–Electronic Conductor

Mixed ionic-electronic conductors offer chemical and electrical means for active tuning of their optical constants, e.g. with variations in oxygen non-stoichiometry in Pr 0.1 Ce 0.9 O 2-δ , enabling implementation of adaptive thin film optical devices. Here, we demonstrate in-situ chemo-tuning of the extinction coefficient in Pr 0.1 Ce 0.9 O 2-δ at elevated temperatures and provide a tuning model that treats the interdependence of mobile oxygen vacancies and small polarons coupled to variations in optically active praseodymium ions. Further, a new means for electro-tuning of the optical constants of mixed ionic-electronic conductors was demonstrated experimentally and modeled for Pr 0.1 Ce 0.9 O 2-δ thin films deposited on grid-like electrode structures. Modeling of non-steady state optical transmittance modulations in the latter allows for estimation of oxygen vacancy mobility that determines the switching speed of the device. Quenched-in values of nr and k to room temperature become nonvolatile, providing a modulation range in the extinction coefficient of Δk ~0.1 (change of ~800%) and in the refractive index of Δn r ~0.1 (relative to initial n r of ~2.35). Key figures of merit including, transmission optical modulation of ~0.04 per 1 mV nm -1 , switching energy per area of 670 pJ μm -2 and switching times of seconds, were demonstrated, with further improvements possible.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Poster Abstract: Leveraging Large Language Models to Reveal Interpretable Cooling Behaviors from Smart Thermostat Data

Frequent heatwaves and hot summers increasingly challenge occupant comfort, health, and energy grid stability. Addressing these challenges requires a detailed understanding of household cooling behaviors, such as thermostat adjustments and adaptive responses to extreme conditions. Traditional analyses often rely on aggregated numerical metrics that overlook subtle but important household-specific variations. In this study, we introduce a generalizable methodology that integrates large language models (LLMs) with vision capabilities to enable scalable and detailed analysis of residential thermostat data. Using Ecobee's Donate Your Data (DYD) dataset—which provides five-minute records of indoor temperatures, thermostat setpoints, and HVAC runtimes—we focus on two U.S. cities with contrasting summer climates : Austin (TX) and Phoenix (AZ). Because raw time-series data are not well suited for direct LLM analysis, we transform them into visual representations, such as daily indoor temperature trajectories and weekly runtime histograms, to better capture behavioral variations. Leveraging LLMs' visual interpretation, we extract descriptive behavioral features, including temperature preferences, time-of-day cooling orientation, anticipatory versus reactive heatwave responses, and behavioral consistency. These semantic features support unsupervised clustering to identify distinct occupant archetypes at scale, revealing differences—such as morning-centric anticipatory coolers versus households that shift toward warmer setpoints during heatwaves—that can inform demand response, resilience planning, and health-aware interventions. By converting raw numerical data into interpretable behavioral patterns, this methodology enables scalable and practical analysis of occupant behavior, supporting actionable insights for comfort, resilience, and energy management.

Nihar, Kopal↗
DOE OSTI2025

Sequence length scaling in vision transformers for scientific images on frontier

Vision Transformers (ViTs) are pivotal for foundational models in scientific imagery, including Earth science applications, due to their capability to process large sequence lengths. While transformers for text have inspired scaling sequence lengths in ViTs, adapting these for ViTs introduces unique challenges. We develop distributed sequence parallelism for ViTs, enabling them to handle up to 1M tokens. Our approach, leveraging DeepSpeed-Ulysses and Long-Sequence-Segmentation with model sharding, is the first to apply sequence parallelism in ViT training, achieving a 94% batch scaling efficiency on 2,048 AMD-MI250X GPUs. Evaluating sequence parallelism in ViTs, particularly in models up to 10B parameters, highlighted substantial bottlenecks. We countered these with hybrid sequence, pipeline, and flash attention strategies, to scale beyond single GPU memory limits. Our method significantly enhances climate modeling accuracy by 20% in temperature predictions, marking the first training of a vision transformer model to convergence with a sequence length of 188K tokens, using full self-attention.

Tsaris, Aristeidis (aris) [ORNL] (ORCID:0000000277↗
DOE OSTI2025

Computational Modeling to Advance Novel Medical Isotopes for Radiotheranostics: A DOE-NIH Joint Workshop Executive Summary

The DOE-NIH Joint Workshop on Computational Modeling to Advance Novel Medical Isotopes for Radiotheranostics, held on September 27, 2024, brought together experts from government, academia, and industry to address critical challenges in radionuclide production and clinical translation. Here, the workshop emphasized interdisciplinary collaboration, particularly between the Department of Energy (DOE) and the National Institutes of Health (NIH), to strengthen the domestic isotope supply, streamline regulatory pathways, and further integrate computational tools into radiopharmaceutical therapy (RPT). Key discussions explored the role of AI-driven modeling, machine learning, and digital twin technologies in optimizing dosimetry, dynamically personalizing treatments, and reducing time to clinical adoption. Advances in predictive computational modeling were highlighted as essential for improving radionuclide yield, purity, and synthesis efficiency. Regulatory considerations and equitable access were central themes, with participants advocating for harmonized global standards, adaptive trial designs, and expanded infrastructure for clinical implementation. DOE computational and production infrastructure was emphasized. Future priorities identified include increased investment in radionuclide production infrastructure, expanded workforce development in radiopharmaceutical sciences and computational modeling, and the creation of robust public-private partnerships. The workshop concluded that continued strategic collaboration and sustained resources will be vital for advancing next-generation radiotheranostics, ensuring safe and effective therapies accessible to all patients.

digital twins↗
DOE OSTI2026

Resource-Adaptive Federated Text Generation with Differential Privacy

In cross-silo federated learning (FL), sensitive text datasets remain confined to local organizations due to privacy regulations, making repeated training for each downstream task both communication-intensive and privacy-demanding. A promising alternative is to generate differentially private (DP) synthetic datasets that approximate the global distribution and can be reused across tasks. However, pretrained large language models (LLMs) often fail under domain shift, and federated finetuning is hindered by computational heterogeneity: only resource-rich clients can update the model, while weaker clients are excluded, amplifying data skew and the adverse effects of DP noise. We propose a flexible participation framework that adapts to client capacities. Strong clients perform DP federated finetuning, while weak clients contribute through a lightweight DP voting mechanism that refines synthetic text. To ensure the synthetic data mirrors the global dataset, we apply control codes (e.g., labels, topics, metadata) that represent each client’s data proportions and constrain voting to semantically coherent subsets. This two-phase approach requires only a single round of communication for weak clients and integrates contributions from all participants. Experiments show that our framework improves distribution alignment and downstream robustness under DP and heterogeneity.

Wang, Jiayi [ORNL]↗
DOE OSTI2024

Core-Shell Oxidative Aromatization Catalysts for Single Step Liquefaction of Distributed Shale Gas (Final Technical Report)

The objective of this project was to design and demonstrate a core-shell structured multifunctional catalyst to convert the light (dry) components of shale gas into liquid aromatic compounds (primarily benzene and toluene) in a single step. Operated in a modular oxidative aromatization system (OAS) under a cyclic redox scheme, the novel catalyst and process can significantly improve the value and transportability of distributed shale gas. Since the project started, each quarter addressed a different set of tasks related to the completion of the milestone detailed in the project award. The yearly summaries of these tasks are summarized below: Q1-Q4: • Conducted project planning and literature search. • Investigated a number of SHC redox catalysts using thermogravimetric analysis and fixed-bed reactor experiments. • Initiated process modeling towards generating two process models for the methane DHA base case and OAS process. • Developed DHA catalysts capable of producing >500 g/kg-cat-hr aromatics at 80% or greater aromatics selectivity at 700°C. Q5-Q8: • Developed alternative approaches with sequential bed configurations to enhance the aromatic yields based on OCM+DHA • Improved the zeolite synthesis efficiency by using the microwave-assisted technique and investigated the synthesis conditions on the zeolite yield, crystalline structure and morphology • Constructed a set of Aspen Plus process models with significant energy savings for OAS as compared to the base case non-oxidative DHA. • Adapted conventional hydrothermal method to be applicable to the microwave synthesizer unit for more efficient catalyst synthesis. • Studied the structure of the OCM catalyst and the dispersion of the carbonate in the redox reactions and in methane flow with Raman Spectroscopy. Q9-Q12: • Scaled up the catalyst synthesis with the microwave synthesis method. Based on its performance, procedural characterizations and catalytic performance testing were further conducted for the new microwave synthesized catalysts with the newly-developed product analysis procedure. • Developed the reaction system setup for the C2-DHA or OCM+DHA reaction product and achieved a better product collection-analysis method for the aromatic products with an improved carbon balance. The product from the OCM reaction exhibited complicated effects on the DHA catalyst. • Conducted additional OCM catalyst characterization using Near Ambient Pressure X-ray Photoelectron Spectroscopy and in situ Raman characterization • Validated the significant energy savings for OAS as compared to the base case non-oxidative DHA. Successfully set up the simulation model for the OCM+DHA+SHC reaction system based on the updated experimental results from NCSU. Q13-End of project: • Synthesized new zeolite catalysts by the microwave method, conducted characterizations (XRD, SEM, and TEM) and catalytic behavior testing. • Explored the “wet” C 2 H 6 and C 2 H 4 DHA reactions with using steam co-feed. A subsequent reduction as the regeneration step can regenerate the DHA catalyst and recover 99% activity of the fresh performance. • Achieved a 15.3% single-pass aromatic yield from methane by rationally combining the OCM and DHA at different temperatures. • Conducted a 105-hour stability test with an improved regeneration procedure, with an average aromatic yield of 13.8%. • Developed new catalyst and achieved a record-high 23.2% yield.

03 NATURAL GAS↗
DOE OSTI2024

Report on the AAPM grand challenge on deep generative modeling for learning medical image statistics

Abstract Background The findings of the 2023 AAPM Grand Challenge on Deep Generative Modeling for Learning Medical Image Statistics are reported in this Special Report. Purpose The goal of this challenge was to promote the development of deep generative models for medical imaging and to emphasize the need for their domain‐relevant assessments via the analysis of relevant image statistics. Methods As part of this Grand Challenge, a common training dataset and an evaluation procedure was developed for benchmarking deep generative models for medical image synthesis. To create the training dataset, an established 3D virtual breast phantom was adapted. The resulting dataset comprised about 108 000 images of size 512 512. For the evaluation of submissions to the Challenge, an ensemble of 10 000 DGM‐generated images from each submission was employed. The evaluation procedure consisted of two stages. In the first stage, a preliminary check for memorization and image quality (via the Fréchet Inception Distance [FID]) was performed. Submissions that passed the first stage were then evaluated for the reproducibility of image statistics corresponding to several feature families including texture, morphology, image moments, fractal statistics, and skeleton statistics. A summary measure in this feature space was employed to rank the submissions. Additional analyses of submissions was performed to assess DGM performance specific to individual feature families, the four classes in the training data, and also to identify various artifacts. Results Fifty‐eight submissions from 12 unique users were received for this Challenge. Out of these 12 submissions, 9 submissions passed the first stage of evaluation and were eligible for ranking. The top‐ranked submission employed a conditional latent diffusion model, whereas the joint runners‐up employed a generative adversarial network, followed by another network for image superresolution. In general, we observed that the overall ranking of the top 9 submissions according to our evaluation method (i) did not match the FID‐based ranking, and (ii) differed with respect to individual feature families. Another important finding from our additional analyses was that different DGMs demonstrated similar kinds of artifacts. Conclusions This Grand Challenge highlighted the need for domain‐specific evaluation to further DGM design as well as deployment. It also demonstrated that the specification of a DGM may differ depending on its intended use.

Radiology, Nuclear Medicine & Medical Imaging↗
DOE OSTI2022

Extended Modeling of DOE Sealed Canisters with Updated Chemistry Models

Road-ready and final disposition packaging configurations for the advanced test reactor (ATR) fuel currently specifies storage within helium backfilled DOE sealed standard canisters. The aluminum cladding of the ATR fuel contains an oxyhydroxide layer of boehmite/bayerite that generates hydrogen when subjected to irradiation. Understanding the effect of this hydrogen buildup over time to important for long term storage considerations. Previous modeling efforts have built a coupled CFD-chemical model to simulate the temperature gas phase concentrations within the DOE sealed standard canisters. These models have coupled the temperature conditions to both the gas phase radiolysis chemistry and the surface chemistry associated with the oxyhydroxide layer. The previous iteration of the model utilized constant G-values for the hydrogen generation a 50-year period. This new iteration of the model utilizes new experimental data to update the hydrogen generation rate, as well as increase the simulated time to a 200-year period. Given the half-life assumed for the primary Cs-137 isotope responsible for gamma radiation in the ATR spent fuel, the 200-year period decreases the decay heat and dose rate of the spent fuel by a factor of 100, which combined with the updated chemistry substantially decreases new hydrogen generation. Continued experimental work has identified trends for aluminum surrogate samples with oxyhydroxide layers for tests done at higher dose rates. At low initial doses a fast generation rate of hydrogen occurs which starts to roll over to a lower generation as the total dose applied increase. A small-scale chemical model was built to replicate as mini-canister surrogate system at SRNL as well as for the smaller capsule tests performed at INL. A variety of chemical models to capture this effect were tested, and a back reaction of H radical absorbing onto the surface, or inhibition of the reaction by significant H 2 cover gas were both able to fit both the mini-canister and the small capsule test data. Both kinetic fits were used to generate data from the new 200-year simulation. As additional experiments continue, the kinetic fits may be adjusted to adapt to new data. However, updated data shows that hydrogen atmosphere has little effect on actual generation data, so the model was reverted to use a star-stepped G-value for low-dose and high-dose regions. For the undried fuel case, the three models differ significantly with an end concentration of 1% for back reaction, 4.2% for inhibition, and 13.1% for constant G-value for the nominal scenario. For the dried fuel case, the nominal cases showed end concentrations of 0.23% for back reaction, 1.9% for inhibition, and 4.2% for const G-value for the nominal scenario. For the constant G-value case that is less conservative than the other two, the pressure for undried fuel increases to 1.94 atm over 200 years for the nominal case and 2.11 atm for the high decay heat case. The total hydrogen concentration after 200 years is 4.92% for the low decay heat case and 27.4% for the high decay heat case for undried fuel and is 1.4% and 9.6% for low and high decay heat fuel for the dried fuel case. If small amount of residual air is present, the potential for nitric acid formation of 595, 1568, and 2816 ppm for the lower, nominal, and upper fuel decay heat can occur. At long-term timeframes no significant shift in major species present occurs, so no appreciable amount of oxygen is present in the system. The continued rate of hydrogen generation in the canister occurs at an increase of 0.04% by mole over the final 10 years. Increasing the model range out to 1000 years continues to drop the hydrogen generation rate through decreasing dose rate.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2025

MPEX AI Digital Twins

All magnetically confined plasma fusion power plant concepts (Tokamak, Spherical Tokamak, Stellarator, Mirror, ...) must exhaust the heat and plasma from the core confinement region to the material walls. The primary channel for this exhaust is through a plasma divertor which directs plasma along open magnetic field lines to a material target. The Material Plasma Exposure eXperiment (MPEX) illustrated in Figure 1, is a high-power, steady-state linear plasma device designed to produce the plasma material interaction (PMI) conditions of the divertor of future magnetic confinement fusion power plants: energy flux 20MW/m 2 , ion fluence 1031/m 2 , pulse duration 106 sec. These goals of plasma exposure in MPEX are well beyond those achieved in magnetic fusion experimental devices. Successfully achieving these high power steady state conditions for long pulses requires operational control of the heating and particle sources and the plasma flux to the walls and target. The MPEX AI Hot Spot Controller, proposed in this project, will help achieve the operational milestones of MPEX. The MPEX device will begin commissioning at the end of FY26. A smaller proto-MPEX was operated for 14,666 plasma discharges and will resume operation in September of 2025 as proto-MPEX-lite, with reduced capability, to test a new window for the Helicon plasma source. The proto-MPEX data has undergone surrogate modeling with machine learning methods (R. Archibald, 2022 IEEE International Conference on Big Data). This proto-MPEX data will be used to begin development of the AI digital twins described in this white paper. The scientific mission of MPEX is to qualify materials of different composition for use in the high energy and plasma flux conditions of a fusion power plant. The materials exposed in MPEX will in some cases be exposed to high neutron fluxes at other ORNL facilities to measure the changes to their PMI properties. The targets exposed in MPEX will be transported under vacuum to a Surface Analysis Station (SAS). The SAS will be equipped with the following diagnostics: Focused Ion Beam (FIB) for trench milling, 100-400 angstrom resolution scanning electron microscope (SEM), surface mapping x-ray spectrometer, high resolution camera, and a future upgrade to a laser induced breakdown spectroscopy quadruple mass spectrometer (LIBS-QMS). The MPEX experiments will generate diverse pre- and post-exposure measurement data of detailed material properties down to the crystal grain level in 3D for post-exposure assessment of PMI damage (e.g. cracking, melting, erosion and redeposition of the material). Physics models for the PMI, and how the material composition and manufacturing impact its performance under high energy plasma exposure, need to be validated with MPEX data to guide the selection of new candidate materials. Our vision for the MPEX AI Digital Twins project is to supply experimental and physics model simulation data to train Artificial Intelligence (AI) models for data processing, analysis, operational control, PMI and materials simulation to maximize the scientific output of the MPEX device. Ultimately, an AI digital twin of MPEX material assessment metrics for tested and synthetic material types with simulated PMI will be trained by the AI Modeling Teams on the experimental and physics simulation data submitted to the American Science Cloud by this project. A purely empirical search for the best material is inefficient given the finite number of samples that can be tested on MPEX. In order to expand the material properties database for training the MPEX Material Assessment AI Digital Twin, and to gain physics understanding of the PMI processes, physics models of the material properties and PMI processes are required. The physics simulations provide detailed simulation data, like impact angles for plasma ions, sputtering yields, transport of the ionized sputtered target material in the plasma, and redeposition locations. This simulation data expands the measurement data for deeper physics understanding. The experimental data is essential to validate the PMI and material structure simulation models. The validated models can then be used to generate new simulation data of MPEX material assessments for synthetic material compositions that have not been exposed in MPEX. These predictive simulations, plus the whole experimental dataset, will be used to train the MPEX Material Assessment AI Digital Twin allowing a rapid generative AI search for new materials with reduced PMI damage by interpolating the domain of the training set. These new optimum materials can be simulated with the physics codes and/or tested in MPEX. The ability of AI neural networks to interpolate multi-dimensional parameter spaces and generate virtual data is exploited for a more efficient search for optimum materials. The advent of the Transformational AI Models Consortium (TAIMC) is an opportunity to engage with state of the art private and public AI developers to achieve the goals of the AI digital twins and AI accelerated physics models proposed in this project. Our partners at ORNL from the Advance Scientific Computing Research (ASCR) organization will collaborate in accelerating the integrated plasma material interaction simulation framework. This simulation framework will provide a platform for generating simulation data across a range of physical fidelities, including hybrid methods that produce multi-fidelity results. This data will be leveraged for AI model development, both for generation of surrogates and the automation of simulation campaigns. A part of the research below will include collaborative efforts with the TAIMC to (i) adapt data storage approaches to ensure AI-readiness, (ii) provide a protypical exemplar to inform and exercise constructed workflows, and (iii) generate and share data, using the TAIMC unified AI data standard, for foundational models that will be trained from multiple sources across the DOE complex. We will also collaborate with the TAIMC, as well as the planned AI modeling teams, to develop approaches for reducing the cost of data generation. These include tailored multi-fidelity approaches as well as fine-tuning strategies to augment general, large-scale foundational models.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

Enhancing Security and Resiliency in Operational Technology Environments Through Network Slicing and Federated Learning

The growing convergence of Information Technology (IT) and Operational Technology (OT) within Industry 4.0 environments has introduced new demands on industrial network infrastructure. As cyber-physical systems become increasingly interconnected, ensuring the secure, timely, and efficient exchange of critical data is essential. This thesis explores how network slicing, a method of creating isolated virtual network segments, can be applied within OT environments to address challenges such as latency, security, and resource allocation. The first research question addressed in this thesis is: How can OT networks take advantage of NFV and SDN technology to become cyber resilient? This study examines the operational, security, and architectural implications of introducing network slicing into traditionally static OT infrastructures such as Industrial Control Systems (ICS) and SCADA. Through simulated deployments and case studies, the research demonstrates how slicing enables better isolation between critical and non-critical services, thereby improving response time, throughput, and security in sensitive environments. The second question considers: How to dynamically implement network slicing and take advantage of network resources towards integrating decentralized machine learning? In response, this thesis proposes a framework that combines Software-Defined Networking (SDN), Network Function Virtualization (NFV), and Federated Learning (FL) to enable real-time analytics while maintaining data locality. The proposed approach reduces the burden on centralized infrastructure and minimizes privacy risks by supporting on-site training of models across distributed OT nodes, coordinated through dynamically allocated network slices. The third focus explores: How slicing helps to increase the resiliency of OT networks through the orchestration of a dynamic DMZ? To answer this, the thesis presents a method for creating and managing Dynamic Demilitarized Zones (DMZs) using network slicing. This enables flexible and automated isolation of sensitive subsystems during threat scenarios or high-risk operations. Coupled with intelligent orchestration and containerized security services, the dynamic DMZ significantly enhances the system's ability to respond to cyber incidents without halting production. Ultimately, this thesis contributes a comprehensive architecture that blends network slicing with machine learning, secure segmentation, and automation, paving the way for resilient, adaptive, and intelligent OT environments. Performance evaluations across multiple scenarios show improvements in system reliability, threat response time, model accuracy, and resource utilization, providing a strong foundation for future industrial automation systems.

Rodiles Delgado, Brian G↗
DOE OSTI2024

Low-depth Clifford circuits approximately solve MaxCut

We introduce a quantum-inspired approximation algorithm for MaxCut based on low-depth Clifford circuits. We start by showing that the solution unitaries found by the adaptive quantum approximation optimization algorithm (ADAPT-QAOA) for the MaxCut problem on weighted fully connected graphs are (almost) Clifford circuits. Motivated by this observation, we devise an approximation algorithm for MaxCut, ADAPT-Clifford, that searches through the Clifford manifold by combining a minimal set of generating elements of the Clifford group. Our algorithm finds an approximate solution of MaxCut on an N -vertex graph by building a depth O ( N ) Clifford circuit. The algorithm has runtime complexity O ( N 2 ) and O ( N 3 ) for sparse and dense graphs, respectively, and space complexity O ( N 2 ) , with improved solution quality achieved at the expense of more demanding runtimes. We implement ADAPT-Clifford and characterize its performance on graphs with positive and signed weights. The case of signed weights is illustrated with the paradigmatic Sherrington-Kirkpatrick model, for which our algorithm finds solutions with ground-state mean energy density corresponding to ∼ 94 % of the Parisi value in the thermodynamic limit. The case of positive weights is investigated by comparing the cut found by ADAPT-Clifford with the cut found with the Goemans-Williamson (GW) algorithm. For both sparse and dense instances we provide copious evidence that, up to hundreds of nodes, ADAPT-Clifford finds cuts of lower energy than GW. Published by the American Physical Society 2024

Muñoz-Arias, Manuel H. (ORCID:000000025711029X)↗
DOE OSTI2020

On Transfer Learning Of Neural Networks Using Bi-Fidelity Data For Uncertainty Propagation

Due to their high degree of expressiveness, neural networks have recently been used as surrogate models for mapping inputs of an engineering system to outputs of interest. Once trained, neural networks are computationally inexpensive to evaluate and remove the need for repeated evaluations of computationally expensive models in uncertainty quantification applications. However, given the highly parameterized construction of neural networks, especially deep neural networks, accurate training often requires large amounts of simulation data that may not be available in the case of computationally expensive systems. In this paper, to alleviate this issue for uncertainty propagation, we explore the application of transfer learning techniques using training data generated from both high- and low-fidelity models. We explore two strategies for coupling these two datasets during the training procedure, namely, the standard transfer learning and the bi-fidelity-weighted learning. In the former approach, a neural network model mapping the inputs to the outputs of interest is trained based on the low-fidelity data. The high-fidelity data are then used to adapt the parameters of the upper layer(s) of the low-fidelity network, or train a simpler neural network to map the output of the low-fidelity network to that of the high-fidelity model. In the latter approach, the entire low-fidelity network parameters are updated using data generated via a Gaussian process model trained with a small high-fidelity dataset. The parameter updates are performed via a variant of stochastic gradient descent with learning rates given by the Gaussian process model. Using three numerical examples, we illustrate the utility of these bi-fidelity transfer learning methods where we focus on accuracy improvement achieved by transfer learning over standard training approaches.

97 MATHEMATICS AND COMPUTING↗