Search NASA⌕ Search

SEARCH · Search NASA

Results for “model efficiency”

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 91 records · Page 5
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Scalable training of trustworthy and energy-efficient predictive graph foundation models for atomistic materials modeling: a case study with HydraGNN

We present our work on developing and training scalable, trustworthy, and energy-efficient predictive graph foundation models (GFMs) using HydraGNN, a multi-headed graph convolutional neural network architecture. HydraGNN expands the boundaries of graph neural network (GNN) computations in both training scale and data diversity. It abstracts over message passing algorithms, allowing both reproduction of and comparison across algorithmic innovations that define nearest-neighbor convolution in GNNs. This work discusses a series of optimizations that have allowed scaling up the GFMs training to tens of thousands of GPUs on datasets consisting of hundreds of millions of graphs. Our GFMs use multitask learning (MTL) to simultaneously learn graph-level and node-level properties of atomistic structures, such as energy and atomic forces. Using over 154 million atomistic structures for training, we illustrate the performance of our approach along with the lessons learned on two state-of-the-art US Department of Energy (US-DOE) supercomputers, namely the Perlmutter petascale system at the National Energy Research Scientific Computing Center and the Frontier exascale system at Oak Ridge Leadership Computing Facility. The HydraGNN architecture enables the GFM to achieve near-linear strong scaling performance using more than 2000 GPUs on Perlmutter and 16,000 GPUs on Frontier.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Intelligent Surrogate Model Development: Boosting Computational Efficiency for Autonomous Control of Advanced Reactors

Advanced reactors promise enhanced safety, greater efficiency, and waste reductions. To fully realize these benefits, it is crucial to address the need for autonomous or semi-autonomous control systems that require fewer operators. This research primarily supports the MARVEL autonomous control system, which requires real-time operation. However, the current RELAP5 reactor thermal hydraulic transient simulation is excessively time-consuming. Therefore, this study aims to leverage deep learning techniques to develop a surrogate model, providing a more efficient and accurate alternative for real-time performance. The model was trained using a combination of one-timestep prediction and scheduled sampling. It was then used for recursive prediction of the reactor state. This developed surrogate model significantly improves computational efficiency, achieving a 12 times acceleration.

46 - INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AN↗
DOE OSTI2026

Simulated moving bed-inspired method for continuous adsorptive denitrogenation of model fuel

Efficient, continuous routes for removing nitrogen-containing compounds from hydrothermal liquefaction-derived synthetic aviation fuel are needed to enable direct blending with conventional jet fuels. Here, we report a simulated moving bed-inspired process for adsorptive denitrogenation of a model fuel. Unlike conventional simulated moving bed systems, which are designed for sharp separations between similar solutes, this approach was run deliberately outside the classical separation region so that both pyridine and indole were removed together from the hydrocarbon stream. Alcohol solvents were used to regenerate the silica adsorbent, maintaining performance over extended operation and avoiding the downtime and energy demand associated with calcination. Under these conditions, the system demonstrates removal of more than 98% of nitrogen while cutting solvent use by 28% compared to batch operation. Classical modeling tools predicted column concentration profiles even in this nontraditional regime, suggesting a straightforward path to scaling. Together, these results motivate solvent-efficient, continuous denitrogenation strategies that could be integrated with biorefinery processes.

Adsorption↗
DOE OSTI2024

Investigating resource-efficient neutron/gamma classification ML models targeting eFPGAs

There has been considerable interest and resulting progress in implementing machine learning (ML) models in hardware over the last several years from the particle and nuclear physics communities. A big driver has been the release of the Python package, hls4ml, which has enabled porting models specified and trained using Python ML libraries to register transfer level (RTL) code. So far, the primary end targets have been commercial field-programmable gate arrays (FPGAs) or synthesized custom blocks on application specific integrated circuits (ASICs). However, recent developments in open-source embedded FPGA (eFPGA) frameworks now provide an alternate, more flexible pathway for implementing ML models in hardware. These customized eFPGA fabrics can be integrated as part of an overall chip design. In general, the decision between a fully custom, eFPGA, or commercial FPGA ML implementation will depend on the details of the end-use application. In this work, we explored the parameter space for eFPGA implementations of fully-connected neural network (fcNN) and boosted decision tree (BDT) models using the task of neutron/gamma classification with a specific focus on resource efficiency. We used data collected using an AmBe sealed source incident on Stilbene, which was optically coupled to an OnSemi J-series silicon photomultiplier (SiPM) to generate training and test data for this study. We investigated relevant input features and the effects of bit-resolution and sampling rate as well as trade-offs in hyperparameters for both ML architectures while tracking total resource usage. The performance metric used to track model performance was the calculated neutron efficiency at a gamma leakage of 10 -3 . The results of the study will be used to aid the specification of an eFPGA fabric, which will be integrated as part of a test chip.

47 OTHER INSTRUMENTATION↗
DOE OSTI2024

A Bayesian Multi-fidelity Neural Network to Predict Nonlinear Frequency Backbone Curves

The use of structural mechanics models during the design process often leads to the development of models of varying fidelity. Often low-fidelity models are efficient to simulate but lack accuracy, while the high-fidelity counterparts are accurate with less efficiency. Here, this paper presents a multi-fidelity surrogate modeling approach that combines the accuracy of a high-fidelity finite element model with the efficiency of a low-fidelity model to train an even faster surrogate model that parameterizes the design space of interest. The objective of these models is to predict the nonlinear frequency backbone curves of the Tribomechadynamics Research Challenge benchmark structure which exhibits simultaneous nonlinearities from frictional contact and geometric nonlinearity. The surrogate model consists of an ensemble of neural networks that learn the mapping between low and high-fidelity data through nonlinear transformations. Bayesian neural networks are used to assess the surrogate model's uncertainty. Once trained, the multi-fidelity neural network is used to perform sensitivity analysis to assess the influence of the design parameters on the predicted backbone curves. Additionally, Bayesian calibration is performed to update the input parameter distributions to correlate the model parameters to the collection of experimentally measured backbone curves.

42 ENGINEERING↗
DOE OSTI2025

Efficient Simulation of Cascading Outages Using an Energy Function-Embedded Quasi-Steady-State Model

Here, this paper proposed an energy function-embedded quasi-steady-state model for efficient simulation of cascading outages on a power grid while addressing transient stability concerns. Compared to quasi-steady-state models, the proposed model incorporates short-term dynamic simulation and an energy function method to efficiently evaluate the transient stability of a power grid together with outage propagation without transient stability simulation. Cascading outage simulation using the proposed model conducts three steps for each disturbance such as a line outage. First, it performs time-domain simulation for a short term to obtain a post-disturbance trajectory. Second, along the trajectory, the system state with the local maximum potential energy is found and used as the initial point to search for a relevant unstable equilibrium by Newton's method. Third, the transient energy margin is estimated based on this unstable equilibrium to predict an out-of-step condition with generators. The proposed energy function-embedded quasi-steady-state model is tested in terms of its accuracy and time performance on an NPCC 140-bus power system and compared to a quasi-steady-state model embedding transient stability simulation.

Guo, Zhenping [Univ. of Tennessee, Knoxville, TN (↗
DOE OSTI2025

Computationally Guided and Experimentally Validated Design of Custom Chelators for Critical Mineral Recovery

Selective, high throughput separation of target critical metals from complex environments such as fly ash leachates and mining process streams presents a significant challenge for economical production. Custom chelators and sorbents are an attractive technology for selective metal extraction, however it can be difficult to predict their performance, and significant experimental efforts are often required to develop chelating technologies. Here, we present a computational strategy focused on modelling chelator-metal binding interactions and benchmark these results versus experimental data. A computational pipeline combining forcefield, semiempirical, and meta-GGA methods with a thermodynamic framework optimized for error cancellation has been developed to predict binding energies of chelator complexes towards critical mineral recovery applications. This approach, originally validated on [2.2.2] cryptates binding mono- and divalent cations, demonstrated robust predictive capabilities with an R2 of 0.850 against experimental aqueous binding energies. The workflow includes metadynamics for exploring high-dimensional potential energy surfaces and a cluster-continuum model for accurate yet computationally efficient solvation modeling. Error cancellation between solvation energies of free and chelator-coordinated ions enables faster convergence, even with finite cluster sizes. Initial studies on the cryptates revealed consistent metal-ligand coordination patterns, with systematic variations influenced by ion size and charge, highlighting key structural features linked to binding selectivity. Further studies of a proprietary chelator have resulted in identification of previously unreported selectivity towards economically significant metals, which in-house experiments have confirmed, demonstrating the feasibility of this approach. By applying this methodology to new chelators targeting critical minerals such as lithium, cobalt, nickel and other strategic metals, we aim to accelerate the discovery of next-generation chelators for efficient recovery, recycling, and separation processes. This computational framework serves as the backbone of a high-throughput design pipeline tailored for sustainable resource utilization and may be applied to a wide range of systems to meet experimental needs.

computational materials↗
DOE OSTI2025

NEML2: An efficient and modular multiphysics constitutive modeling library for hybrid computing environments

This paper presents NEML2, an open-source, high-performance library developed for constitutive material modeling, designed to support the flexible and modular development of models for complex material behavior. Building on the foundational structure of its predecessor, NEML, the NEML2 library introduces significant improvements, including enhanced vectorization, automatic differentiation, and seamless integration with PyTorch, facilitating the application of machine learning techniques in material simulations. NEML2 provides a C++ backend with Python bindings, enabling users to create custom material models that can be executed efficiently on both CPU and GPU platforms. The library also supports coupling with Multiphysics simulation frameworks like MOOSE, making it suitable for realistic simulations involving coupled physical processes. Rigorous quality assurance through unit and regression testing ensures the reliability of results, while the extensible, user-friendly design encourages collaboration and reproducibility across the scientific community. This paper provides an overview of NEML2’s architecture, core features, and applications, highlighting its impact on accelerating material qualification and advancing computational methods in materials science.

GPU↗
DOE OSTI2024

The Case for Co-Designing Model Architectures with Hardware

While GPUs are responsible for training the vast majority of state-of-the-art deep learning models, the implications of their architecture are often overlooked when designing new deep learning (DL) models. As a consequence, modifying a DL model to be more amenable to the target hardware can significantly improve the runtime performance of DL training and inference. In this paper, we provide a set of guidelines for users to maximize the runtime performance of their transformer models. These guidelines have been created by carefully considering the impact of various model hyperparameters controlling model shape on the efficiency of the underlying computation kernels executed on the GPU. We find the throughput of models with “efficient” model shapes is up to 39% higher while preserving accuracy compared to models with a similar number of parameters but with unoptimized shapes.

Yin, Junqi↗
DOE OSTI2025

Ten questions on building stock modeling to inform energy efficiency and sustainability

To enhance economic competitiveness and ensure energy efficiency, resilience, and security, cities and governments are adopting technologies and strategies to improve their existing building stocks. This approach aims to reduce energy use, improve energy affordability, and ensure a reliable power supply while safeguarding occupants during extreme weather events that may disrupt energy services. The effectiveness of these solutions will depend on building stock characteristics, use patterns, weather conditions, evolving technologies and their markets, and a city’s socio-economic conditions. This paper presents ten questions and answers that highlight the most important issues regarding the use of building stock modeling as a powerful tool to provide insights for informing stakeholders’ actions and decision-making on energy efficiency, costs reduction, and resilience of buildings in cities. Building stock modeling should build upon the fit-for-purpose framework, balancing the use case accuracy requirements, level of complexity, and needed resources (expertise, compute). The advancements in Artificial Intelligence (AI), the increasingly available open dataset of building stock in cities, and the more affordable powerful computing will accelerate the adoption of building stock modeling across scales by researchers and practitioners to inform decision making on sustainability and efficiency.

AI↗
DOE OSTI2025

Concurrent two-way coupling of global and local models across internal boundaries with non-matching discretizations

Coupling local and global models enables efficient simulation of multiscale systems, where global models capture large-scale behavior and local models, with enhanced physics, resolve finer details over a smaller region. Here, this paper presents a mathematically consistent method for coupling physics-based models of varying fidelity across adjacent, non-overlapping subdomains, even when discretizations do not match at the immersed interdomain interfaces. Incompressible Navier-Stokes equations (NSE) constitute the global model while residual-based turbulence model serves as the local high-fidelity model. In addition, a scalar advection-diffusion equation that models the convection of an active scalar field is appended to the turbulence model in the local domain. This scalar field does not have its complement in the global model, giving rise to unequal number of equations at the immersed boundary between local and global models. Interdomain coupling terms are derived via the Variational Multiscale Discontinuous Galerkin (VMDG) method with new developments in scale representation and efficient fine-scale estimation. While transient laminar flows modeled with NSE in the global domain can be resolved with relatively coarse mesh, turbulent flow calculations in the local model require much finer spatial discretizations as well as smaller time-step for appropriately resolving the turbulent flow physics. The proposed framework also accommodates non-matching meshes at the immersed boundaries. Test problems in 2D and 3D numerically showcase the concurrent two-way coupling of unknown fields across the immersed boundaries. The 3D test presents a case with an unequal number of equations, where the scalar field represents the convection of contaminant concentration. This provides more detailed physics in the local region and highlights its application in climate modeling and atmospheric sciences.

Variational Multiscale Discontinuous Galerkin (VMD↗
DOE OSTI2025

A staged deep learning approach to spatial refinement in 3D temporal atmospheric transport

High-resolution spatiotemporal simulations effectively capture the complexities of atmospheric plume dispersion in complex terrain. However, their high computational cost makes them impractical for applications requiring rapid responses or iterative processes, such as optimization, uncertainty quantification, or inverse modeling. To address this challenge, this work introduces the Dual-Stage Temporal Three-dimensional UNet Super-resolution (DST3D-UNet-SR) model, a highly efficient deep learning model for plume dispersion predictions. DST3D-UNet-SR is composed of two sequential modules: the temporal module (TM), which predicts the transient evolution of a plume in complex terrain from low-resolution temporal data, and the spatial refinement module (SRM), which subsequently enhances the spatial resolution of the TM predictions. We train DST3D-UNet-SR using a comprehensive dataset derived from high-resolution large eddy simulations (LES) of plume transport. We propose the DST3D-UNet-SR model to significantly accelerate LES of three-dimensional (3D) plume dispersion by three orders of magnitude. Additionally, the model demonstrates the ability to dynamically adapt to evolving conditions through the incorporation of new observational data, substantially improving prediction accuracy in high-concentration regions near the source.

3D temporal sequences↗
DOE OSTI2024

Bayes_Opt-SWMM: A Gaussian process-based Bayesian optimization tool for real-time flood modeling with SWMM

Real-time flood model plays a pivotal role in averting urban flood damage, particularly when there is minimal lead time for preparatory measures. However, urban flood modeling in real-time often contends with inherent uncertainties arising from input data uncertainty and parameter ambiguities. Here this study introduces a real-time calibration (RTC) tool called Bayes_Opt-SWMM, specifically tailored for real-time urban flood modeling and uncertainty optimization. This tool leverages the Gaussian process-based Bayesian optimization algorithm and interfaces seamlessly with the Stormwater Management Model (SWMM). It integrates real-time model forcing data and flood monitoring collected through sensors and gauges which are strategically placed within critical locations of urban drainage systems. Our approach hinges on the Surrogate Model based Uncertainty Optimization (SMUO) concept, providing an avenue for enhancing real-time flood modeling. Bayes_Opt-SWMM runs the optimization process using a surrogate model called Gaussian Process emulator with two inference methods: (1) the Gaussian Process (GP) model and (2) Markov Chain Monte Carlo (MCMC) algorithm in GP model (GP_MCMC). Furthermore, three acquisition functions, namely Expected Improvement (EI), Maximum Probability of Improvement (MPI), and Lower Confidence Bound (LCB), facilitate optimal parameter fitting within the surrogate models. The efficiency of GP-based surrogate models in learning SWMM model parameters, leads to an improved uncertainty quantification and accelerated real-time flood modeling in urban areas. Overall, Bayes_Opt-SWMM emerges as a cost-effective and valuable tool for real-time flood modeling and monitoring, with significant potential for managing intelligent storm water systems in urban environments.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Statistically-driven Experimental Design to Improve Reference-free Quantification of Small Molecules by Liquid Chromatography-Mass Spectrometry

Non-targeted analysis of small molecules and metabolites in unknown, complex samples using liquid chromatography-tandem mass spectrometry remains challenging. One of the main bottlenecks is the extensive unannotated regions of metabolomics mass spectrometry data, resulting in knowledge gaps. Small molecule annotation in mass spectrometry data has conventionally relied on reference standards and libraries for compound identification and confirmation, which can constrain compound identification to those molecules already known, thus limiting the ability to discover new knowledge and new markers. Retention time prediction can facilitate and expedite unknown compound identification in non-targeted analysis of complex metabolomics samples. Additionally, accurate retention time predictions can also inform sample mixture design for LC-MS/MS analyses. However, current machine learning-based methods for retention time prediction are typically developed for specific chromatographic platforms and are not generalizable across scales. And while technologies and methods to improve reference-free metabolite identification for more comprehensive annotation of unknowns has received much attention, development of the same for quantitation without reference standards has been much more limited, despite its importance in toxicological, environmental, food safety, forensics, and clinical applications. We believe that a reference-free quantitation strategy that exploits mass spectrometry data already collected for reference-free identification can provide much more insight on unknowns, and move the metabolomics field for more complete unknowns characterization. As such, we pursue two efforts to improve upon current state-of-the-art methods in non-targeted analysis: (1) machine learning-based retention time prediction and (2) statistical design of experiments framework for reference-free quantitation. In this work, we develop and demonstrate (1) a generalizable retention time prediction capability across chromatographic conditions and scales, and (2) a statistical design-based framework for response factor contribution elucidation and reference-free quantitation. Evaluation of our retention time prediction model, PrediToR, showed approximately 24% improvement over current models, and we observed approximately 10X improvement in concentration estimation accuracy from our statistical design-based response factor model over a primarily ionization efficiency-based model. We expect that future efforts to improve upon these new capabilities will further advance non-targeted analysis of small molecules towards truly reference-free metabolomics.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2026

Multi-task Parallelism for Robust Pre-training of Graph Foundation Models on Multi-source, Multi-fidelity Atomistic Modeling Data

Graph foundation models using graph neural networks promise sustainable, efficient atomistic modeling. To tackle challenges of processing multi-source, multi-fidelity data during pre-training, recent studies employ multi-task learning, in which shared message passing layers initially process input atomistic structures regardless of source, then route them to multiple decoding heads that predict data-specific outputs. This approach stabilizes pre-training and enhances a model’s transferability to unexplored chemical regions. Preliminary results on approximately four million structures are encouraging, yet questions remain about generalizability to larger, more diverse datasets and scalability on supercomputers. We propose a multi-task parallelism method that distributes each head across computing resources with GPU acceleration. Implemented in the open-source HydraGNN architecture, our method was trained on over 24 million structures from five datasets and tested on the Perlmutter, Aurora, and Frontier supercomputers, demonstrating efficient scaling on all three highly heterogeneous super-computing architectures.

Lupo Pasini, Massimiliano [ORNL] (ORCID:0000000249↗
DOE OSTI2026

Offline Maximizing Minimally Invasive Proper Orthogonal Decomposition for Reduced-Order Modeling of S n Radiation Transport

Deterministic solutions to the Sn radiation transport equation can be computationally expensive to calculate. Reduced-order modeling enables efficient approximation of the full-order model (FOM) solution. We propose a novel method for constructing reduced-order models (ROMs) of the S n radiation transport equation, offline maximizing minimally invasive (OMMI) proper orthogonal decomposition (POD). POD uses the method of snapshots to create a reduced-order basis for constructing an ROM. Minimally invasive POD leverages the sweep infrastructure existing in deterministic transport codes to create a POD-based ROM, even when infeasible by traditional methods. Offline maximizing minimally invasive proper orthogonal decomposition (OMMI-POD) extends minimally invasive POD by performing sweeps offline, therefore maximizing the potential speedup. OMMI-POD does so by creating a library of reduced systems from a training set. This library of reduced systems is then interpolated to provide a rapid approximate solution of the S n radiation transport equation. The model is evaluated on a set of test problems, achieving a low error with a 466 times speedup over the FOM. Also presented is a study of the effect of sampling method on the performance of OMMI-POD, specifically comparing naive uniform sampling to the more accurate and computationally expensive greedy sampling.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Efficient general method for numerically modeling laser pulse propagation, overlap, and lifetime effects in amplifiers

An efficient numerical time-dependent general method is developed to address incoherent pulse overlap and lifetime effects in laser amplifiers. The alternating propagation-population laser energetics method (APPLE) has been validated against a semi-discrete coupled rate equation numerical method (SDRE) and analytic formalisms in bounding cases. APPLE is based on decoupled rates applied to a time-dependent framework where both space-time-dependent populations and pulse energetics are consistently updated in each time step. A significant advantage of APPLE lies in its conceptual simplicity, ease of implementation, and relatively small computational cost. SDRE tracks the populations through coupled rates and uses the method of lines to discretize the hyperbolic partial differential transport equations allowing for use of ordinary differential equation solvers. With reasonably sized mesh, we report both energetic and power pulse shape relative differences on the order of one percent between the models over a large range of initial conditions.

47 OTHER INSTRUMENTATION↗