Results for “Scalable Model” — Page 29

NASA NTRS2016

Cloud-Based Orchestration of a Model-Based Power and Data Analysis Toolchain

The proposed Europa Mission concept contains many engineering and scientific instruments that consume varying amounts of power and produce varying amounts of data throughout the mission. System-level power and data usage must be well understood and analyzed to verify design requirements. Numerous cross-disciplinary tools and analysis models are used to simulate the system-level spacecraft power and data behavior. This paper addresses the problem of orchestrating a consistent set of models, tools, and data in a unified analysis toolchain when ownership is distributed among numerous domain experts. An analysis and simulation environment was developed as a way to manage the complexity of the power and data analysis toolchain and to reduce the simulation turnaround time. A system model data repository is used as the trusted store of high-level inputs and results while other remote servers are used for archival of larger data sets and for analysis tool execution. Simulation data passes through numerous domain-specific analysis tools and end-to-end simulation execution is enabled through a web-based tool. The use of a cloud-based service facilitates coordination among distributed developers and enables scalable computation and storage needs, and ensures a consistent execution environment. Configuration management is emphasized to maintain traceability between current and historical simulation runs and their corresponding versions of models, tools and data.

Post, Ethan↗

DOE OSTI2026

A Unified Design Theory for Multi-Port Polyphase Transformers Enabling Scalable Power-Multiplexed EV Fleet Charging Systems

This paper presents a unified analytical design theory for multi-port polyphase transformers, targeting scalable and isolated high-power Electric Vehicle (EV) fleet charging systems with power multiplexing capability. As fleet electrification accelerates, conventional one-to-one charger architectures face significant challenges in infrastructure cost, peak power demand, and low utilization of installed power electronics. Power-multiplexed charging architectures, which dynamically distribute power from a shared pool of converter modules across multiple vehicles, have emerged as a promising solution. However, such architectures require scalable, isolated multi-port power interfaces capable of routing energy among multiple inputs and outputs, whose design remains complex and dependent on iterative modeling. To address this gap, the proposed theory provides closed-form expressions for self-inductance, leakage inductance, and mutual coupling terms for arbitrary multi-phase, multi-port transformer structures. The formulation enables direct synthesis of isolated multi-input and multi-output resonant converter systems without reliance on geometry-specific finite-element analysis or extensive parameter extraction. This capability is particularly critical for power-multiplexed systems, where modular converter structures must interface with multiple vehicles while maintaining galvanic isolation and flexible power allocation. The effectiveness of the proposed framework is demonstrated through the design of a 360 kW multi-phase system operating over a 700–900 VDC input and 400–1250 VDC output range. PLECS simulation results confirm accurate prediction of system behavior and validate the applicability of the approach to multi-port, power-multiplexed charging scenarios. The proposed method significantly reduces design complexity while enabling scalable, cost-effective, and fully utilized EV fleet charging infrastructure.

Asa, Erdem [ORNL] (ORCID:0000000190884812)↗

DOE OSTI2025

Hybrid learning techniques for scientific data reduction with performance guarantees

The research initiatives supported by the U.S. Department of Energy (DOE) Grant DE-SC0022265 are fundamentally aimed at pioneering advanced machine learning (ML) techniques for scientific data compression within high-performance computing (HPC) environments. This comprehensive body of work addresses the critical challenge posed by the exponential growth of data generated by scientific simulations in domains such as fusion energy, climate modeling, and computational fluid dynamics (CFD). A core objective is to develop compression algorithms that achieve substantial data reduction—often by orders of magnitude—while rigorously ensuring the fidelity of both the primary data (PD) and scientifically crucial derived quantities of interest (QoI). The methodologies deployed under this grant integrate sophisticated deep learning architectures, prominently featuring autoencoders, advanced generative models like conditional diffusion, and hybrid learning techniques. Key innovations include the development of Guaranteed Autoencoders (GAE) and the Guaranteed Conditional Diffusion with Tensor Correction (GCDTC) framework, which provide explicit, instance-level error bounds on reconstructed data. Furthermore, specialized strategies such as nonlinear constraint satisfaction are employed to preserve the integrity of QoI, a vital requirement for the trustworthiness of downstream scientific analyses. This research also focuses on the design and implementation of scalable, GPU-accelerated software pipelines that seamlessly integrate into existing HPC workflows, ensuring both computational efficiency and practical applicability. The CAESAR framework, for example, unifies foundation and generative models to create an adaptive and efficient compression solution for spatio-temporal scientific data. Collectively, these efforts represent a significant advancement in mitigating the scientific data deluge, enabling more effective data management, accelerated scientific discovery, and optimized utilization of HPC resources.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2025

Final report- UFL - RAPIDS2: A SciDAC Institute for Computer Science, Data, and Artificial Intelligence

The research initiatives supported by the U.S. Department of Energy (DOE) Grant DE-SC0022265 are fundamentally aimed at pioneering advanced machine learning (ML) techniques for scientific data compression within high-performance computing (HPC) environments. This comprehensive body of work addresses the critical challenge posed by the exponential growth of data generated by scientific simulations in domains such as fusion energy, climate modeling, and computational fluid dynamics (CFD). A core objective is to develop compression algorithms that achieve substantial data reduction—often by orders of magnitude—while rigorously ensuring the fidelity of both the primary data (PD) and scientifically crucial derived quantities of interest (QoI). The methodologies deployed under this grant integrate sophisticated deep learning architectures, prominently featuring autoencoders, advanced generative models like conditional diffusion, and hybrid learning techniques. Key innovations include the development of Guaranteed Autoencoders (GAE) and the Guaranteed Conditional Diffusion with Tensor Correction (GCDTC) framework, which provide explicit, instance-level error bounds on reconstructed data. Furthermore, specialized strategies such as nonlinear constraint satisfaction are employed to preserve the integrity of QoI, a vital requirement for the trustworthiness of downstream scientific analyses. This research also focuses on the design and implementation of scalable, GPU-accelerated software pipelines that seamlessly integrate into existing HPC workflows, ensuring both computational efficiency and practical applicability. The CAESAR framework, for example, unifies foundation and generative models to create an adaptive and efficient compression solution for spatio-temporal scientific data. Collectively, these efforts represent a significant advancement in mitigating the scientific data deluge, enabling more effective data management, accelerated scientific discovery, and optimized utilization of HPC resources.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2026

Data for FUN-PROSE: A Deep Learning Approach to Predict Condition-Specific Gene Expression in Fungi

mRNA levels of all genes in a genome is a critical piece of information defining the overall state of the cell in a given environmental condition. Being able to reconstruct such condition-specific expression in fungal genomes is particularly important to metabolically engineer these organisms to produce desired chemicals in industrially scalable conditions. Most previous deep learning approaches focused on predicting the average expression levels of a gene based on its promoter sequence, ignoring its variation across different conditions. Here we present FUN-PROSE—a deep learning model trained to predict differential expression of individual genes across various conditions using their promoter sequences and expression levels of all transcription factors. We train and test our model on three fungal species and get the correlation between predicted and observed condition-specific gene expression as high as 0.85. We then interpret our model to extract promoter sequence motifs responsible for variable expression of individual genes. We also carried out input feature importance analysis to connect individual transcription factors to their gene targets. A sizeable fraction of both sequence motifs and TF-gene interactions learned by our model agree with previously known biological information, while the rest corresponds to either novel biological facts or indirect correlations.

Genomics↗

NASA NTRSDate not supplied

AI-Enhanced Computational Tools for Entry Systems Modeling

To advance the understanding of complex atmospheric entry phenomena, NASA’s Entry Systems Modeling (ESM) team [1] has developed high-fidelity computational tools addressing multiscale challenges, from material microstructures to full-scale heatshield response. This abstract highlights a subset of ESM tools, focusing on AI integration to enhance workflows and predictive modeling. - PuMA [2] computes effective material properties from high-resolution micro-CT scans, supporting TPS analysis for NASA missions. - TomoSAM [3] automates 3D tomography dataset segmentation for PuMA using the Segment Anything Model, reducing manual effort and improving accuracy. - PATO [4] models porous reactive materials under extreme conditions, with advancements such as unified solvers, mechanical erosion, and TPS coatings for NASA missions. - arcjetCV [5] employs deep learning to analyze arc jet test footage, measuring recession rates, shape changes, and shock standoff distances, bridging simulations, and experiments to reveal TPS ablation behavior. - ARCHeS [6] simulates arc heater plasma flows, modeling turbulence, radiation, and electromagnetic interactions to optimize arc heater performance, validate TPS under extreme conditions, and serve as a foundation for developing digital twins of arc heater facilities. - SPARTA [7] simulates rarefied hypersonic flows and gas-surface interactions for planetary entry missions, leveraging GPU architectures for scalable and efficient aerothermal and ablation analyses. AI-driven solutions, such as deep learning segmentation, have streamlined workflows in ESM tools and still hold significant potential to further accelerate processes and enhance automation in entry systems modeling. [1] Haskins, J.B. (2023), [2] Ferguson, J.C. (2018), [3] Meurisse, J.B.E. (2018), [4] Semeraro, F. (2023), [5] Quintart, A. (2024) [6] Meurisse, J.B.E. (2022), [7] Plimpton, S.J. (2019)

Predictive Modeling↗

DOE OSTI2025

The global Sahel monsoon ocean-pressure index reconciles its regional and large-scale features

Monitoring Sahelian rainfall variability is increasingly critical as climate extremes intensify across the region. Here, we develop the Sahelian Monsoon Ocean-Pressure Index (SMOPI), a novel global synthetic indicator constructed from five dynamically coherent sea-level pressure regions statistically linked to June-September Sahel monsoon rainfall. SMOPI captures intra-seasonal and interannual variability, and crucially, reflects the influence of both regional processes and large-scale teleconnections on monsoon dynamics. It aligns with the dominant rainfall variability mode in reanalyses and 29 CMIP6 models. Strong/positive SMOPI phases coincide with wet years and are associated with enhanced convergence, favorable jet configurations, and robust Pacific, Atlantic, and Indian Ocean teleconnections. Conversely, weak/negative SMOPI phases correspond to drought conditions and divergent moisture fluxes. SMOPI exposes model failures in reproducing historical droughts and offers new physical insights into rainfall-driving mechanisms. It stands out as a scalable, potentially transferable diagnostic tool for monitoring/forecasting and evaluating Sahelian monsoon rainfall under global warming.

Tamoffo, Alain T. [Helmholtz-Zentrum Hereon GmbH, ↗

DOE OSTI2026

Industry-driven Training and Curriculum Development Process

The development of a sustainable, skilled fusion workforce requires coordinated strategy between all sectors of fusion industry. This paper outlines a framework to align training programs with evolving technical and professional demands of fusion, including enhancing existing curricula, the establishment of new programs at educational institutions, and the identification of workforce gaps informed through industry engagement. Effective curriculum development requires input from both educators and employers to ensure that academic content reflects real-world challenges and can prepare students for successful transitions into the field. Collaborative models, such as industry-led training programs, inter-institutional partnerships, and faculty development initiatives, are highlighted as mechanisms for scalable and inclusive workforce development. Continued program success and relevance will be dependent on continuous review processes, including feedback from employers, alumni, and advisory boards. The combination of these programs supports the formation of flexible, industry-informed training pathways. This approach aims to foster a competent workforce capable of advancing fusion energy research and commercialization.

Gehrig, Monica [ORNL] (ORCID:0000000341022612)↗

DOE OSTI2024

Privacy-Preserving Federated Learning for Science: Challenges and Research Directions

This paper discusses the key challenges and future research directions for privacy-preserving federated learning (PPFL), with a focus on its application to large-scale scientific AI models, in particular, foundation models~(FMs). PPFL enables collaborative model training across distributed datasets while preserving privacy-- an important collaborative approach for science. We discuss the need for efficient and scalable algorithms to address the increasing complexity of FMs, particularly when dealing with heterogeneous clients. In addition, we underscore the need for developing advance privacy-preserving techniques, such as differential privacy, to balance privacy and utility in large FMs emphasizing fairness and incentive mechanisms to ensure equitable participation among heterogeneous clients. Finally, we emphasize the need for a robust software stack supporting scalable and secure PPFL deployments across multiple high-performance computing facilities. We envision that PPFL would play a crucial role to advance scientific discovery and enable large-scale, privacy-aware collaborations across science domains.

Kim, Kibaek [Argonne National Laboratory (ANL)]↗

DOE OSTI2026

AI-driven neutrino diagnostics and radiation-hard beam instrumentation for next-generation neutrino experiments

The Long Baseline Neutrino Facility (LBNF) at Fermilab will deliver a high-intensity, multi-megawatt neutrino beam to the Deep Underground Neutrino Experiment (DUNE), enabling precision tests of the three-neutrino paradigm, CP violation searches, neutrino mass ordering determination, and supernova neutrino studies. To accelerate DUNE’s physics reach and ensure robust beam operations, we propose an integrated AI-driven framework with real-time diagnostics and radiation-hardened instrumentation. At its core is a Real-Time Beam Integrity Monitor using a physics-informed Digital Twin. By reconstructing pion phase space from muon profiles and exploiting magnetic horn optic linearity, it enables spill-by-spill beam correction and flux stabilization. By using this approach, flux-related systematics could be reduced from 5% to 1%, potentially accelerating the discovery of CP violations by four to six years. Complementing this, a US–Japan R&D effort will deploy a LAPPD-based muon monitor in the NuMI beamline. ToF measurements can be acquired with picosecond precision using this radiation-hard system, enhancing sensitivity to horn chromatic effects. Simulations confirm strong response to these effects. ML models predict beam quality and horn current to sub-percent accuracy from muon data, enhancing anomaly detection and stability. This scalable, AI-enabled strategy improves beam fidelity, reduces systematics, and sets a new standard for high-power accelerator operations.

Ganguly, Sudeshna [Fermilab] (ORCID:00000003163482↗

DOE OSTI2025

Demonstration of Utility Managed Smart Charging for Multiple Benefit Streams (Final Report)

In the summer of 2020, the U.S. Department of Energy (DOE) awarded funding to Exelon’s Maryland utilities—Baltimore Gas and Electric (BGE), Delmarva Power & Light (DPL), and Potomac Electric Power Company (Pepco)—to implement the Smart Charge Management (SCM) pilot. This initiative aimed to design and implement managed electric vehicle (EV) charging strategies, evaluate the grid impacts of EV charging, and assess the utilities' ability to control EV load based on real-time grid conditions. The SCM pilot explored four aspects for continued improvement: (1) cybersecurity and managed charging functionality testing of two vendor platforms—WeaveGrid (telematics-based) and Shell Recharge Solutions (network-based)—which pursued charge scheduling and optimization through distinct approaches; (2) an analysis by Argonne National Laboratory (ANL) modeling team of three potential SCM enrollment scenarios within BGE and Pepco service territories over the next decade to assess future scalability; (3) employing customer engagement strategies, including surveys and a responsive pricing approach; and (4) the launch and implementation of pilots in Exelon’s Maryland territories in collaboration with WeaveGrid.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2025

AI-driven neutrino diagnostics and radiation-hard beam instrumentation for next-generation neutrino experiments

The Long Baseline Neutrino Facility (LBNF) at Fermilab will deliver a high-intensity, multi-megawatt neutrino beam to the Deep Underground Neutrino Experiment (DUNE), enabling precision tests of the three-neutrino paradigm, CP violation searches, neutrino mass ordering determination, and supernova neutrino studies. To accelerate DUNE’s physics reach and ensure robust beam operations, we propose an integrated AI-driven framework with real-time diagnostics and radiation-hardened instrumentation. At its core is a Real-Time Beam Integrity Monitor using a physics-informed Digital Twin. By reconstructing pion phase space from muon profiles and exploiting magnetic horn optic linearity, it enables spill-by-spill beam correction and flux stabilization. By using this approach, flux-related systematics could be reduced from 5% to 1%, potentially accelerating the discovery of CP violations by four to six years. Complementing this, a US–Japan R&D effort will deploy a LAPPD-based muon monitor in the NuMI beamline. ToF measurements can be acquired with picosecond precision using this radiation-hard system, enhancing sensitivity to horn chromatic effects. Simulations confirm strong response to these effects. ML models predict beam quality and horn current to sub-percent accuracy from muon data, enhancing anomaly detection and stability. This scalable, AI-enabled strategy improves beam fidelity, reduces systematics, and sets a new standard for high-power accelerator operations.

Ganguly, Sudeshna [Fermilab] (ORCID:00000003163482↗

NASA NTRS2010

A Hardware Model Validation Tool for Use in Complex Space Systems

One of the many technological hurdles that must be overcome in future missions is the challenge of validating as-built systems against the models used for design. We propose a technique composed of intelligent parameter exploration in concert with automated failure analysis as a scalable method for the validation of complex space systems. The technique is impervious to discontinuities and linear dependencies in the data, and can handle dimensionalities consisting of hundreds of variables over tens of thousands of experiments.

Davies, Misty Dawn↗

NASA NTRS2018

Solution of Ordinary Differential Equations in Gradient-Based Multidisciplinary Design Optimization

A gradient-based approach to multidisciplinary design optimization enables efficient scalability to large numbers of design variables. However, the need for derivatives causes difficulties when integrating ordinary differential equations (ODEs) in models. To simplify this, we propose the use of the general linear methods framework, which unifies all Runge-Kutta and linear multistep methods. This approach enables rapid implementation of integration methods without the need to differentiate each one, even in a gradient-based optimization context. We also develop a new parallel time integration algorithm that enables vectorization across time steps. We present a set of benchmarking results using a stiff ODE, a non-stiff nonlinear ODE, and an orbital dynamics ODE, and compare integration methods. In a modular gradient-based multidisciplinary design optimization context, we find that the new parallel time integration algorithm with high-order implicit methods, especially Gauss-Legendre collocation, is the best choice for a broad range of problems.

Hwang, John T.↗

DOE OSTI2026

Strong Lensing Cosmology with Population-level Calibrated Neural Ratio Estimation

Strong gravitational lensing contains key information about cosmic acceleration. Modern and next-generation galaxy imaging surveys are expected to provide high-quality data on $\mathcal{O}(10^5)$ galaxy-galaxy lensing systems. The plethora and complexity of the data are likely to present computational challenges for parameter inference methods for fitting high-dimensional likelihoods, which are often analytically intractable. Neural Ratio Estimation (NRE) efficiently computes individual likelihood ratios that can be combined into population-level posteriors. We use simulations to study the capacity of NRE to jointly predict the dark energy equation-of-state parameter $w$ and the total matter density $Ω_{m}$ from lensing images and companion spectroscopic information. We also introduce a post hoc posterior coverage calibration procedure that mitigates the model overconfidence that is typically found in neural density estimation applications. Our experiments show that the errors on both parameters decrease with increasing inference population sizes. In particular, for 100 lenses in a standard $Λ$CDM Universe, our calibrated NRE model achieves median fractional uncertainty of $22.8\%$ in $w$ and $2.9\%$ in $Ω_{m}$. This proof of concept demonstrates a potentially scalable approach for efficient cosmological parameter inference with large populations of galaxy-scale lenses observed in future surveys.

Jarugula, Sreevani [Fermilab] (ORCID:0000000253867↗

NASA NTRS2007

Symbolic LTL Compilation for Model Checking: Extended Abstract

In Linear Temporal Logic (LTL) model checking, we check LTL formulas representing desired behaviors against a formal model of the system designed to exhibit these behaviors. To accomplish this task, the LTL formulas must be translated into automata [21]. We focus on LTL compilation by investigating LTL satisfiability checking via a reduction to model checking. Having shown that symbolic LTL compilation algorithms are superior to explicit automata construction algorithms for this task [16], we concentrate here on seeking a better symbolic algorithm.We present experimental data comparing algorithmic variations such as normal forms, encoding methods, and variable ordering and examine their effects on performance metrics including processing time and scalability. Safety critical systems, such as air traffic control, life support systems, hazardous environment controls, and automotive control systems, pervade our daily lives, yet testing and simulation alone cannot adequately verify their reliability [3]. Model checking is a promising approach to formal verification for safety critical systems which involves creating a formal mathematical model of the system and translating desired safety properties into a formal specification for this model. The complement of the specification is then checked against the system model. When the model does not satisfy the specification, model-checking tools accompany this negative answer with a counterexample, which points to an inconsistency between the system and the desired behaviors and aids debugging efforts.

Rozier, Kristin Y.↗

NASA NTRS2008

Distributed Aviation Concepts and Technologies

Aviation has experienced one hundred years of evolution, resulting in the current air transportation system dominated by commercial airliners in a hub and spoke infrastructure. While the first fifty years involved disruptive technologies that required frequent vehicle adaptation, the second fifty years produced a stable evolutionary optimization of decreasing costs with increasing safety. This optimization has resulted in traits favoring a centralized service model with high vehicle productivity and cost efficiency. However, it may also have resulted in a system that is not sufficiently robust to withstand significant system disturbances. Aviation is currently facing rapid change from issues such as environmental damage, terrorism threat, congestion and capacity limitations, and cost of energy. Currently, these issues are leading to a loss of service for weaker spoke markets. These catalysts and a lack of robustness could result in a loss of service for much larger portions of the aviation market. The impact of other competing transportation services may be equally important as casual factors of change. Highway system forecasts indicate a dramatic slow down as congestion reaches a point of non-linearly increasing delay. In the next twenty-five years, there is the potential for aviation to transform itself into a more robust, scalable, adaptive, secure, safe, affordable, convenient, efficient and environmentally friendly system. To achieve these characteristics, the new system will likely be based on a distributed model that enables more direct services. Short range travel is already demonstrating itself to be inefficient with a centralized model, providing opportunities for emergent distributed services through air-taxi models. Technologies from the on-demand revolution in computers and communications are now available as major drivers for aviation on-demand adaptation. Other technologies such as electric propulsion are currently transforming the automobile industry, and will also significantly alter the functionality of future distributed aviation concepts. Many hurdles exist, including technology, regulation, and perception. Aviation has an inherent governmental role not present in other recent on-demand transformations, which may pose a risk of curtailing aviation democratization .

Moore, Mark D.↗

DOE OSTI2024

Generating An Advanced Cross-section Library For HTGR Pebble Bed Depletion Calculations Using Reduced-Order Model Generation Techniques

For code development, Advanced Reactor Technologies - Gas Cooled Reactors Program (ART-GCR) rely on a collaboration with the Nuclear Energy Advanced Modeling and Simulation (NEAMS) program, but the cross sections generation and the methodology definition is part of this program area goals. Based on previous studies in FY23, the size of microscopic cross section libraries increases rapidly with the number of tabulations, requiring significant amount of memory and drastically slowing down the Griffin calculations when evaluating cross sections via the multivariate linear interpolation approach. Rising to these challenges, this work investigates constructing Reduced-order Models (ROMs) for the multi-group microscopic cross sections to accelerate the cross section evaluation in Griffin. A database of multigroup cross sections is first collected considering all possible parameters that a designer could change for optimization. Down-selection of the ROM techniques afterward shows Deep Neural Network (DNN) as the best candidate when jointly consider memory efficiency, predictive accuracy, computational cost, scalability, flexibility and ease of implementation of the algorithms in comparison to the multidimensional interpolation. This work develops a specific interface that enables the cross section predictions using pre-trained DNN models into Griffin leveraging the existing ROM capabilities. DNNs have been trained for all isotopes for use in Griffin. Preliminary Griffin testing shows that DNNs exhibit exceptional predictive accuracy and the use of DNNs provides orders of magnitude improvement in memory efficiency compared to conventional interpolation techniques. With such ROM techniques, it holds great promise to further increase the fidelity of the Pebble Bed Reactor (PBR) simulation by increasing the number of tabulations/state variables during cross section evaluation, while maintaining the computational cost affordable in Griffin.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗