SEARCH · Engineering Papers

Results for “machine learning for scientific computing”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 397 records · Page 22

All sources NASA NTRS DOE OSTI

DOE OSTI2025

Method to simultaneously facilitate all jet physics tasks

Machine learning has become an essential tool in jet physics. Due to their complex, high-dimensional nature, jets can be explored holistically by neural networks in ways that are not possible manually. However, innovations in all areas of jet physics are proceeding in parallel. We show that specially constructed machine learning models trained for a specific jet classification task can improve the accuracy, precision, or speed of all other jet physics tasks. This is demonstrated by training on a particular multiclass generation and classification task and then using the learned representation for different generation and classification tasks, for datasets with a different (full) detector simulation, for jets from a different collision system ($pp$ versus $ep$), for generative models, for likelihood ratio estimation, and for anomaly detection. We consider our omnilearn approach thus as a jet-physics foundation model. It is made publicly available for use in any area where state-of-the-art precision is required for analyses involving jets and their substructure.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2022

Green AI: Insights Into Deep Learning's Looming Energy Efficiency Crisis

As demands grow to integrate artificial intelligence into every aspect of industry, commerce, and life, deep learning's exploding energy cost has become a looming crisis, making AI systems a salient energy-efficiency challenge. One might expect that doubling a neural network's size would halve its error rate, or at least allow it to achieve greater performance given the same amount of time and energy. I will present clear and substantial scientific evidence which indicates that not only is this intuition wildly wrong, but that neural networks scale so poorly that to increase deep learning performance by only a small fraction can easily require an order of magnitude or more increase in computational resources and energy. Further, the marginal trade-off price of to increase model performance rapidly explodes as performance targets are increased. To address this challenge, I will provide a toolkit of techniques that can be applied today to mitigate the inefficiency of modern deep learning. And, I will conclude by illuminating a practical path forward towards efficient, Green AI.

artificial intelligence↗

DOE OSTI2025

Optimal Transport as a Tool for Scientific Discovery in Radiation Biology

This report summarizes findings from research conducted for the “Exploration of the Poten tial for Artificial Intelligence and Machine Learning to Advance Low-Dose Radiation Biology Re search” (RadBio-AI) program, supported by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, under Awards KP1601011/FWP CC121 and KP1601017/FWP CC121. The research reported here was undertaken in an effort to assess the potential of optimal measure transport methods as components within the larger scope of a com putational framework envisioned to support research in the radiation biology domain. Within this effort, our interest centered on enabling a unified generic framework where probabilistic modeling, inference, and statistical learning can be carried out for a wide range of data distributions. As described next in Section 1 (and in more detail in our original publication), optimal measure transport offers the possibility of such unified approach.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Portable, heterogeneous ensemble workflows at scale using libEnsemble

libEnsemble is a Python-based toolkit for running dynamic ensembles, developed as part of the DOE Exascale Computing Project. The toolkit utilizes a unique generator–simulator–allocator paradigm, where generators produce input for simulators, simulators evaluate those inputs, and allocators decide whether and when a simulator or generator should be called. The generator steers the ensemble based on simulation results. Generators may, for example, apply methods for numerical optimization, machine learning, or statistical calibration. libEnsemble communicates between a manager and workers. Flexibility is provided through multiple manager–worker communication substrates each of which has different benefits. These include Python’s multiprocessing, mpi4py, and TCP. Multisite ensembles are supported using Balsam or Globus Compute. We overview the unique characteristics of libEnsemble as well as current and potential interoperability with other packages in the workflow ecosystem. We highlight libEnsemble’s dynamic resource features: libEnsemble can detect system resources, such as available nodes, cores, and GPUs, and assign these in a portable way. These features allow users to specify the number of processors and GPUs required for each simulation; and resources will be automatically assigned on a wide range of systems, including Frontier, Aurora, and Perlmutter. Such ensembles can include multiple simulation types, some using GPUs and others using only CPUs, sharing nodes for maximum efficiency. We also describe the benefits of libEnsemble’s generator–simulator coupling, which easily exposes to the user the ability to cancel, and portably kill, running simulations based on models that are updated with intermediate simulation output. We demonstrate libEnsemble’s capabilities, scalability, and scientific impact via a Gaussian process surrogate training problem for the longitudinal density profile at the exit of a plasma accelerator stage. In conclusion, the study uses gpCAM for the surrogate model and employs either Wake-T or WarpX simulations, highlighting efficient use of resources that can easily extend to exascale.

Dynamic ensembles↗

DOE OSTI2022

Technical Characterization and Benefit Evaluation of 5G-Enabled Grid Data Transport and Applications

This report summarizes the Year 1 work of Pacific Northwest National Laboratory’s (PNNL’s) 5G Fabricated Resource and Asset Management Encompassment for energy infrastructure (Energy FRAME) project funded by the Department of Energy Office of Science’s Advanced Scientific Computing Research Program. 5G is a breakthrough technology that enables a fully mobile and connected society, and a 5G-enabled digital continuum will be one of the critical foundations for a clean energy economy and grid modernization. In collaboration with PNNL’s Advanced Wireless Communication team and Center for Advanced Technology Evaluation team, the project team has been evaluating the system performance of 5G testbeds in the PNNL 5G Innovation Studio, and has formulated a co-simulation test case of power system transmission, distribution, and communication (T&D&C) networks considering 5G technology and high penetration of distributed energy resources. The methodology developed in the 5G Energy FRAME project can be customized to fit different future grid scenarios to evaluate multiple (dynamic) configurations (computing, sensing, communication, environment) for different stakeholders. In summary, our main technical highlights in project Year 1 are as follows: 1) Technical characterization of 5G standalone architectures, 2) Formulation of co-simulation test case of T&D&C networks embedded with 5G, 3) Initial benefit evaluation of 5G communication platform for grid use cases, and 4) Additional extended discussions on edge computing, artificial intelligence and machine learning, and high-performance computing and cloud computing adoptions. In addition, a collection of system performance data is shared through the publicly available weblink, https://www.pnnl.gov/projects/5g-energy-frame/publications

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2021

End-to-End Differentiable Modeling and Management of the Environment

Focal Area: (2) Data acquisition and assimilation enabled by machine learning, AI, and advanced methods including experimental/network design/optimization. We emphasize the importance of leveraging optimization techniques from AI/machine learning (ML) to solve challenging problems in Earth system modeling. Science Challenge: Automatic differentiation has had a transformative effect on ML by allowing the calculation of gradients of arbitrary functions in an incredibly large class of models. We can potentially realize similar improvements in parameter estimation and control for Earth system models (ESMs) by reimplementing them in computational frameworks from ML. Practitioners working with large (>10 7 parameters) models in ML and AI can obtain good predictive performance in a range of spatiotemporal tasks by making use of optimization via stochastic gradient descent and incorporating prior knowledge at multiple levels. We propose writing ESMs in open-source computational frameworks such as Torch, Tensorflow, and JAX to greatly expand the scope of environmental forecasting and management challenges, which can be addressed by leveraging automatic differentiation and gradient descent-like algorithms. We do not call for a wholesale replacement of physical models with data-driven surrogates, but rather advocate for interleaving physical and empirical equations in a manner that is most faithful to the extent of our scientific knowledge and observational data. Central to this topic is the merging of differentiable physical simulations with differentiable optimization layers, which are now both beginning to come to the forefront.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2020

Opportunities and Challenges from Artificial Intelligence and Machine Learning for the Advancement of Science, Technology, and the Office of Science Missions

In February 2019, the President signed Executive Order 13859, Maintaining American Leadership in Artificial Intelligence. This order launched the American Artificial Intelligence Initiative, a concerted effort to promote and protect AI technology and innovation in the United States. The Initiative implements a government-wide strategy in collaboration and engagement with the private sector, academia, the public, and like-minded international partners. Among other actions, key directives in the Initiative called for Federal agencies to: Prioritize AI research and development investments, Enhance access to high-quality cyberinfrastructure and data, Ensure that the US maintains an international leadership role in the development of technical standards for AI, and Provide education and training opportunities to prepare the American workforce for the new era of AI. The mission of the Department of Energy (DOE) is to ensure America’s security and prosperity by addressing its energy, environmental, and nuclear challenges through transformative science and technology solutions. In terms of Science and Innovation, the DOE’s mission is to maintain a vibrant US effort in science and engineering as a cornerstone of our economic prosperity with clear leadership in strategic areas. From July to October in 2019, the Argonne, Oak Ridge, and Berkeley National Laboratories hosted a series of four AI for Science Town Hall meetings in Chicago, Oak Ridge, Berkeley, and Washington DC. The four meetings were attended by over 1300 scientists from the 17 DOE Labs, 39 companies, and over 90 universities. The goal of the Town Hall series was ‘to examine scientific opportunities in the areas of artificial intelligence, Big Data, and high-performance computing (HPC) in the next decade, and to capture the big ideas, grand challenges, and next steps to realizing these.’ The discussions at the meetings were captured in the final report of the AI for Science Town Hall meetings.

42 ENGINEERING↗

DOE OSTI2024

Spin-Controllable Dynamics in Defect-Engineered Carbon Nanotubes as Single Photon Emitters: Data-Driven Modeling and Computations

Quantum technologies, such as quantum computing and sensing, require efficient single-photon emission (SPE) sources that operate at room temperature in telecom wavelengths. While several materials can serve as SPE sources, no single platform meets all the criteria for efficiency, ambient operation, and scalability. Single-walled carbon nanotubes (SWCNTs) with covalently attached molecules offer a promising solution. Their SPE can be easily tuned via modifications of the SWCNT's diameter, chirality, and bonded molecules, enabling emission across near-IR to telecom wavelengths at ambient conditions. However, to fully realize the potential of SWCNTs and unlock their quantum capabilities, a deeper understanding of how structural defects from molecular adducts affect their emission and competing photoexcited processes is essential. To address this gap in our knowledge, this project combined quantum chemistry calculations with data-driven methods of cheminformatics (QSAR) and machine learning (ML). The developed computational approaches have provided several design strategies for covalent functionalization of SWCNTs to improve their optical response. The collaboration with Los Alamos National Lab (LANL) enabled direct comparison of computational and experimental data, facilitating method validation. This partnership was enhanced through access to LANL's Center for Integrated Nanotechnologies (CINT) utilizing User Facility Program and summer internships, which provided three NDSU graduate students with hands-on experience at LANL. The outcomes of this project included (1) Advancing the current stage of computational methods in accurate modeling of non-adiabatic spin-dependent photoexcited dynamics and its applicability to nanosystems consisting of thousands of atoms, realized as open-access codes linked to existing DFT-based software; (2) Establishing the relationship between the structure of adducts and SWCNTs and intrinsic excitonic and spin properties of defect states for guiding novel synthetic strategies and experimental probes of chemically functionalized SWCNTs as near-IR emitting materials; (3) Generating virtual libraries of hypothetical functionalized SWCNTs for virtual screening of their chemical structures and optical properties, leveraging new functionalities of SWCNTs; (4) Offering a unique experience for NDSU graduate students that prepared them for future scientific careers related to materials modeling and big data processing. These results were summarized in 12 published journal papers and 3 recently submitted papers. One of a key finding is that the position of defect sites on the SWCNT surface primarily drives the emission redshift (up to 100 meV), while the polarity of the defect-inducing molecules has a much smaller effect (~10 meV). However, the electron-donating or withdrawing properties of a molecule influence selecting reactivity of defect sites. These insights important for optimizing synthetic protocols for desired emissions in SWCNTs. We also revealed that the interaction between two defects at various positions on the SWCNT enhances the redshift and optical activity of states, favoring strong near-IR emission. This suggests that manipulations in defect concentrations is a promising strategy for controlling efficient emission. Mostly important, the defect position was found controllable by the spin states of photoexcited intermediates: Excited aromatic molecules form ortho defects with SWCNTs at their singlet states in the presence of oxygen, while oxygen-free conditions favor para defects via the triplet-state mechanism. Additionally, a heat-activated [2+2] cycloaddition reaction facilitates divalent defect formation with fewer bonding positions that narrows emission bands. These groundbreaking findings have been experimentally validated and significantly advance our understanding of defect chemistry in SWCNTs. Using a novel encoding technique and 3D-MoRSE descriptors, we developed highly accurate ML/QSAR models to predict both the 3D structure and optical properties of SWCNTs with chemical defects. This model enabled the creation of a virtual library of 125,556 structures, providing new insights into the relationship between SWCNT-defect structure and emission.

77 NANOSCIENCE AND NANOTECHNOLOGY↗

DOE OSTI2021

AI-based design of a nuclear reactor core

The authors developed an artificial intelligence (AI)-based algorithm for the design and optimization of a nuclear reactor core based on a flexible geometry and demonstrated a 3× improvement in the selected performance metric: temperature peaking factor. The rapid development of advanced, and specifically, additive manufacturing (3-D printing) and its introduction into advanced nuclear core design through the Transformational Challenge Reactor program have presented the opportunity to explore the arbitrary geometry design of nuclear-heated structures. The primary challenge is that the arbitrary geometry design space is vast and requires the computational evaluation of many candidate designs, and the multiphysics simulation of nuclear systems is very time-intensive. Therefore, the authors developed a machine learning-based multiphysics emulator and evaluated thousands of candidate geometries on Summit, Oak Ridge National Laboratory’s leadership class supercomputer. The results presented in this work demonstrate temperature distribution smoothing in a nuclear reactor core through the manipulation of the geometry, which is traditionally achieved in light water reactors through variable assembly loading in the axial direction and fuel shuffling during refueling in the radial direction. The conclusions discuss the future implications for nuclear systems design with arbitrary geometry and the potential for AI-based autonomous design algorithms.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

DOE OSTI2024

Journey over Destination: Dynamic Sensor Placement Enhances Generalization

Reconstructing complex, high-dimensional global fields from limited data points is a challenge across various scientific and industrial domains. This is particularly important for recovering spatio-temporal fields using sensor data from, for example, laboratory-based scientific experiments, weather forecasting, or drone surveys. Given the prohibitive costs of specialized sensors and the inaccessibility ofcertain regions of the domain, achieving full field coverage is typically not feasible. Therefore, the development of machine learning algorithms trained to reconstruct fields given a limited dataset is of critical importance. In this study, we introduce a generalapproach that employs moving sensors to enhance data exploitation during the training of an attention based neural network, thereby improving field reconstruction. The training of sensor locations is accomplished using an end-to-end workflow, ensuringdifferentiability in the interpolation of field values associated to the sensors, and is simple to implement using differentiable programming. Additionally, we have incorporated a correction mechanism to prevent sensors from entering invalid regions within the domain. We evaluated our method using two distinct datasets; the results show that our approach enhances learning, as evidenced by improved test scores.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2025

CLPNets: Coupled Lie–Poisson neural networks for multi-part Hamiltonian systems with symmetries

To accurately compute data-based prediction of Hamiltonian systems, it is essential to utilize methods that preserve the structure of the equations over time. We consider a particularly challenging case of systems with interacting parts that do not reduce to pure momentum evolution. Such systems are essential in scientific computations, such as discretization of a continuum elastic rod, which can be viewed as the group of rotations and translations $SE(3)$. The evolution involves not only the momenta but also the relative positions and orientations of the particles. The presence of Lie group-valued elements, such as relative positions and orientations, poses a problem for applying previously derived methods for data-based computing. We develop a novel method of data-based computation and complete phase space learning of such systems. We follow the original framework of SympNets (Jin et al., 2020) and LPNets (Eldred et al., 2024), building the neural network from phase space mappings that preserve the Lie–Poisson structure. We derive a novel system of mappings that are built into neural networks describing the evolution of such systems. We call such networks Coupled Lie–Poisson Neural Networks, or CLPNets. We consider increasingly complex examples for the applications of CLPNets, starting with the rotation of two rigid bodies about a common axis, progressing to the free rotation of two rigid bodies, and finally to the evolution of two connected and interacting $SE(3)$ components, describing the discretization of an elastic rod into two elements. Our method preserves all Casimir invariants to machine precision, preserves energy to high accuracy, and shows good resistance to the curse of dimensionality, requiring only a few thousand data points for all cases studied (three to eighteen dimensions). Additionally, the method is highly economical in memory requirements, requiring only about 200 parameters for the most complex case considered.

Data-based modeling↗

DOE OSTI2026

Report for the DOE Office of Science Workshop on Envisioning Frontiers in AI and Computing for Biological Research

Artificial intelligence (AI), machine learning (ML), and high-performance computing (HPC) are poised to transform biological research, spurring innovation in biotechnology and biosystems design. "is transformation will bring an explosion of new capabilities to control the expression of genomic information in living organisms and harness that information to invent new biobased technologies (Jinek et al. 2012; NASEM 2025).

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2024

PlasmoData.jl — A Julia framework for modeling and analyzing complex data as graphs

Datasets encountered in scientific and engineering applications appear in complex formats (e.g., images, multivariate time series, molecules, video, text strings, networks). Graph theory provides a unifying framework to model such datasets and enables the use of powerful tools that can help analyze, visualize, and extract value from data. In this work, we present PlasmoData.jl, an open-source, Julia framework that uses concepts of graph theory to facilitate the modeling and analysis of complex datasets. The core of our framework is a general data modeling abstraction, which we call a DataGraph. We show how the abstraction and software implementation can be used to represent diverse data objects as graphs and to enable the use of tools from topology, graph theory, and machine learning (e.g., graph neural networks) to conduct a variety of tasks. We illustrate the versatility of the framework by using real datasets: (i) an image classification problem using topological data analysis to extract features from the graph model to train machine learning models; (ii) a disease outbreak problem where we model multivariate time series as graphs to detect abnormal events; and (iii) a technology pathway analysis problem where we highlight how we can use graphs to navigate connectivity. Further, our discussion also highlights how PlasmoData.jl leverages native Julia capabilities to enable compact syntax, scalable computations, and interfaces with diverse packages. Overall, we show that the DataGraph abstraction and PlasmoData.jl Julia package are able to model data within graphs and enable useful analysis.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

DOE OSTI2021

Plant Single-Cell Solutions for Energy and the Environment (Second Workshop Report)

Plants are important sources of energy and materials, and they collectively represent a critical component of Earth’s ecosystem. With increasing environmental stresses due to climate change and intensive agricultural practices, the need for resilient plants is greater than ever before. To secure plant resources for bioenergy, biomaterials, food, and ecosystem adaptation, a deeper understanding of the fundamental biology of plants at a cellular level is urgently needed. Plants contain a multitude of specialized cell types that compose tissues and organs. Pathogens often target specific cell types within plants, and the response of one cell to a particular stimulus is likely to be distinct from its neighbor because of underlying molecular and contextual differences. Understanding how these responses are distributed among cells, the main goal of single-cell approaches, will substantially enhance our ability to use targeted engineering for improving plant productivity and resilience. Furthermore, single-cell approaches are necessary to understand the interactions between plants and other ecosystem members such as fungi, bacteria, and archaea. Unlocking these gene-response mechanisms at a cellular level can improve our ability to adapt plants to environmental stresses, increasing their utility as feedstocks for biomaterials and bioenergy. Recent advances in high-throughput sequencing, mass spectrometry, microfluidics and miniaturization, artificial intelligence and machine learning, and bioinformatics have greatly improved our ability to detect and understand processes at a cellular level. In mammalian systems, single-cell transcriptomics has already led to many advances, such as newly identified cell types and cell-targeted treatment of diseases, and mass spectrometry-based single-cell proteomics has recently been demonstrated as a promising emerging technology. However, plant single-cell omics has lagged behind mammalian approaches due to the high cost of the technologies relative to available resources and to the innate biological features of plants, including the complexity of the cell wall and polyploidy. To better understand how single-cell methods could enable plant science, Lawrence Berkeley National Laboratory (Berkeley Lab) hosted a workshop on April 29, 2021, that brought together a diverse group of leaders in plant and/or single-cell biology. Attendees represented federal research programs and domestic and international academic institutions. During the workshop, three presenters described the current state of research in both experimental and computational approaches. While the focus of the workshop was on factors preventing plant biology researchers from fully adopting single-cell methodologies, workshop participants agreed that most barriers could be overcome with focused, strategic investment and coordinated efforts among institutions leading to significant scientific discoveries that would be difficult to obtain using more conventional technologies.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2023

Learning continuous models for continuous physics

Abstract Dynamical systems that evolve continuously over time are ubiquitous throughout science and engineering. Machine learning (ML) provides data-driven approaches to model and predict the dynamics of such systems. A core issue with this approach is that ML models are typically trained on discrete data, using ML methodologies that are not aware of underlying continuity properties. This results in models that often do not capture any underlying continuous dynamics—either of the system of interest, or indeed of any related system. To address this challenge, we develop a convergence test based on numerical analysis theory. Our test verifies whether a model has learned a function that accurately approximates an underlying continuous dynamics. Models that fail this test fail to capture relevant dynamics, rendering them of limited utility for many scientific prediction tasks; while models that pass this test enable both better interpolation and better extrapolation in multiple ways. Our results illustrate how principled numerical analysis methods can be coupled with existing ML training/testing methodologies to validate models for science and engineering applications.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Combining data and theory for derivable scientific discovery with AI-Descartes

Abstract Scientists aim to discover meaningful formulae that accurately describe experimental data. Mathematical models of natural phenomena can be manually created from domain knowledge and fitted to data, or, in contrast, created automatically from large datasets with machine-learning algorithms. The problem of incorporating prior knowledge expressed as constraints on the functional form of a learned model has been studied before, while finding models that are consistent with prior knowledge expressed via general logical axioms is an open problem. We develop a method to enable principled derivations of models of natural phenomena from axiomatic knowledge and experimental data by combining logical reasoning with symbolic regression. We demonstrate these concepts for Kepler’s third law of planetary motion, Einstein’s relativistic time-dilation law, and Langmuir’s theory of adsorption. We show we can discover governing laws from few data points when logical reasoning is used to distinguish between candidate formulae having similar error on the data.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2023

A New Era of Discovery: The 2023 Long-Range Plan for Nuclear Science (V.1.2)

Nuclear science is the investigation of how protons and neutrons are formed from elementary particles and how the forces between those particles produce both nuclei and the vast variety of nuclear phenomena that occur in the universe. It has evolved into a broad field that addresses profound scientific questions: Where does the mass of visible matter come from? How do stars ignite, live, and die? How do nuclei illuminate the search for new laws of nature? This science points the way to using nuclei to build new technologies that benefit society. The 2015 Nobel Prize in physics was shared by nuclear physicists Art McDonald and Takaaki Kajita for the discovery of neutrino oscillations, which confirmed that neutrinos have mass. Our progress on big questions like this one since 2015 has been remarkable owing to new experimental tools, theoretical breakthroughs, powerful computational techniques, and the talented people who make these innovations possible. Focusing on these new tools, the Facility for Rare Isotope Beams (FRIB) at Michigan State University is already producing exciting results on decays of never-before-produced isotopes a year after it was completed on time and on budget. The energy upgrade of the Continuous Electron Beam Accelerator Facility (CEBAF) at the Thomas Jefferson National Accelerator Facility (Jefferson Lab) was also completed on schedule and on budget—new data from this facility are revealing the spectrum, structure, and dynamics of protons, neutrons, nuclei, and mesons. On the theory front, we can now calculate the distribution of quarks inside the proton from first principles. The implementation of artificial intelligence (AI) and machine learning (ML) techniques has led to improved data analysis and increased efficiency in running experiments and theoretical calculations. The impact of nuclear science goes beyond expanding the frontiers of knowledge about matter in the universe. We simultaneously develop a STEM work force that advances the security, technology, health, and wealth of our nation. Some connections are obvious. Expert scientists trained to work with radioactive nuclei are in demand in nuclear security arenas and are highly sought after by various government agencies and private industries. Graduate students and postdoctoral fellows (postdocs) obtain extensive computational, modeling, and data science skills that are similarly in high demand. Less obvious but equally important is the connection between these trained scientists and success in other professions, including medicine, energy, and entrepreneurial pursuits. The workforce that enables discovery in nuclear science also makes breakthroughs in technologies with tremendous impact on the nation’s economic advancement.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2025

Integrating and Characterizing HPC Task Runtime Systems for hybrid AI-HPC workloads

Scientific workflows increasingly involve both HPC and machine-learning tasks, combining MPI-based simulations, training, and inference in a single execution. Launchers such as Slurm’s srun constrain concurrency and throughput, making them unsuitable for dynamic and heterogeneous workloads. We present a performance study of RADICAL-Pilot (RP) integrated with Flux and Dragon, two complementary runtime systems that enable hierarchical resource management and high-throughput function execution. Using synthetic and production-scale workloads on Frontier, we characterize the task execution properties of RP across runtime configurations. RP+Flux sustains up to 930 tasks/s, and RP+Flux+Dragon exceeds 1,500 tasks/s with over 99.6% utilization. In contrast, srun peaks at 152 tasks/s and degrades with scale, with utilization below 50%. For IMPECCABLE.v2 drug discovery campaign, RP+Flux reduces makespan by 30–60% relative to srun/Slurm and increases throughput more than four times on up to 1,024. These results demonstrate hybrid runtime integration in RP as a scalable approach for hybrid AI-HPC workloads.

HPC-AI↗