SEARCH · Engineering Papers
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.
This code provides natural language annotation and labels for multiple activities in high performance. The purpose is to enable machine learning for natural language processing algorithms with high performance computing systems.
Atomistic simulations are essential in chemistry and materials science but remain challenging to run due to the expert knowledge required for the setup, execution, and validation stages of these calculations. We present ChemGraph, an agentic framework powered by artificial intelligence and state-of-the-art simulation tools to streamline and automate computational chemistry and materials science workflows. ChemGraph leverages graph neural network-based foundation models for accurate yet computationally efficient calculations and large language models (LLMs) for natural language understanding, task planning, and scientific reasoning to provide an intuitive and interactive interface. We evaluate ChemGraph across 13 benchmark tasks and demonstrate that smaller LLMs (GPT-4o-mini, Claude-3.5-haiku, Qwen-2.5-14B) perform well on simple workflows, while more complex tasks benefit from using larger models. Importantly, we show that decomposing complex tasks into smaller subtasks through a multi-agent framework enables GPT-4o to reach perfect accuracy and smaller LLMs to match or exceed single-agent GPT-4o's performance in these benchmarks.
Quantum Machine Learning (QML) offers a promising paradigm that leverages quantum computing principles to develop efficient and expressive models for learning from complex and structured data. Recent advances in natural language processing (NLP) and artificial intelligence (AI) have demonstrated capabilities in understanding, generating, and reasoning over linguistic and multimodal information. In this work, we present the Quantum Graph Transformer (QGT), a hybrid quantum–classical architecture that extends graph transformer capabilities through quantum self-attention. The QGT models variable-length sentences as token graphs, where both the embedding encoding and the self-attention mechanisms are implemented using parameterized quantum circuits (PQCs), enabling efficient contextual learning with significantly fewer trainable parameters. We train QGT using both fully connected and 𝑘 -nearest-neighbor graph structures and evaluate it on five benchmark sentiment-classification datasets. Experimental results show that QGT consistently achieves higher or comparable accuracy to existing quantum NLP models and outperforms a Classical Graph Transformer (CGT) baseline with identical architecture, achieving 29.4 × fewer parameters while requiring 3–5 × fewer samples to reach comparable performance. These findings highlight the potential of graph-based quantum models as scalable and data-efficient architectures for natural language understanding.
Given the emergence of data science and machine learning throughout all aspects of society, but particularly in the scientific domain, there is increased importance placed on obtaining data. Data in materials science are particularly heterogeneous, based on the significant range in materials classes that are explored and the variety of materials properties that are of interest. This leads to data that range many orders of magnitude, and these data may manifest as numerical text or image-based information, which requires quantitative interpretation. The ability to automatically consume and codify the scientific literature across domains - enabled by techniques adapted from the field of natural language processing - therefore has immense potential to unlock and generate the rich datasets necessary for data science and machine learning. This review focuses on the progress and practices of natural language processing and text mining of materials science literature and highlights opportunities for extracting additional information beyond text contained in figures and tables in articles. Here, we discuss and provide examples for several reasons for the pursuit of natural language processing for materials, including data compilation, hypothesis development, and understanding the trends within and across fields. Current and emerging natural language processing methods along with their applications to materials science are detailed. We, then, discuss natural language processing and data challenges within the materials science domain where future directions may prove valuable.
The Workflow Orchestration Language Framework (WOLF) is an agentic framework grounded in natural language with an architecture inspired by reinforcement learning (RL)—designed to orchestrate, scale, and accelerate complex workflows. The concept of WOLF was born out of the very successful ASC Tri-lab Multi-Agent Design Assistant (MADA) project, but extends beyond its domain-specific design agents to provide a more general and extensible architecture. WOLF capitalizes on the lessons learned from MADA and is fully aligned with Sutton’s The Bitter Lesson—that the most enduring progress in AI comes from general-purpose methods that scale with computation, rather than narrow techniques built on domain-specific human knowledge. In this spirit, WOLF enables agents to autonomously learn workflows, capture strategies as reusable playbooks, and build a growing corpus of interpretable, auditable “wisdom artifacts.” These artifacts, expressed in natural language, bridge human and machine understanding while preserving adaptability and scalability as computational power continues to expand.
Compositional generalization, the ability to infer and reason using novel combinations of previously encountered entities and structures, is a trait of great utility across a variety of deep learning tasks; it implies a certain ability to reason using consistent and therefore interpretable rules, to understand the constituent parts of its input at multiple levels of granularity, to be robust to spurious differences between semantically equivalent inputs, and more. We seek to understand the degree to which existing natural language models achieve or fall short of compositional generalization across a variety of tasks, whether there are distinct types of compositional generalization in practice, and to identify avenues of intervention through which we can improve models’ compositional generalization ability. After exploring these questions using an array of models, tasks, and potential interventions we see that large, pretrained language models have encountered sufficient training data to account for a variety of fine-grained compositional behavior but still struggle to reason at the level of phrases or larger language structures. Non-intrusive, data-based interventions in the form of augmenting individual sequences with compressed versions of themselves or deriving new examples from induced grammars prove insufficient to encourage greater levels of compositional reasoning, indicating that future work might benefit most from focusing on changes to a model’s inductive bias at the architectural or loss level, or by integrating compositionality-boosting data interventions into the large-scale pretraining process itself.
I/O performance monitoring tools such as Darshan and Recorder collect I/O-related metrics on production systems and help understand the applications' behavior. However, some gaps prevent end-users from seeing the whole picture when it comes to detecting and drilling down to the root causes of I/O performance slowdowns and where those problems originate. These gaps arise from limitations in the available metrics, their collection strategy, and the lack of translation to actionable items that could advise on optimizations. This paper highlights such gaps and proposes solutions to drill down to the source code level to pinpoint the root causes of I/O bottlenecks scientific applications face by relying on cross-layer analysis combining multiple performance metrics related to I/O software layers. We demonstrate with two real applications how metrics collected in high-level libraries (which are closer to the data models used by an application), enhanced by source-code insights and natural language translations, can help streamline the understanding of I/O behavior and provide guidance to end-users, developers, and supercomputing facilities on how to improve I/O performance. Using this cross-layer analysis and the heuristic recommendations, we attained up to 6.9× speedup from run-as-is executions.
Abstract Recent advances in computer vision (CV) and natural language processing have been driven by exploiting big data on practical applications. However, these research fields are still limited by the sheer volume, versatility, and diversity of the available datasets. CV tasks, such as image captioning, which has primarily been carried out on natural images, still struggle to produce accurate and meaningful captions on sketched images often included in scientific and technical documents. The advancement of other tasks such as 3D reconstruction from 2D images requires larger datasets with multiple viewpoints. We introduce DeepPatent2, a large-scale dataset, providing more than 2.7 million technical drawings with 132,890 object names and 22,394 viewpoints extracted from 14 years of US design patent documents. We demonstrate the usefulness of DeepPatent2 with conceptual captioning. We further provide the potential usefulness of our dataset to facilitate other research areas such as 3D image reconstruction and image retrieval.
Focus Area: Model predictability improvements (Primary); Data optimization (secondary); Data complexity insights (secondary). The Scientific Challenge: If we believe that a future under extreme conditions will look very differently from today, we can likely agree that ML/AI models trained on past and present datasets will not be adequate to make reliable predictions into the future. This is true for water cycling, as well as biogeochemistry and other Earth system components and behaviors. Additionally, ML/AI models are inherently non-physical. Despite the flourishing success of ML/AI in many applications, such as computer vision, natural language process, and gaming, even the most sophisticated AI models don’t understand the very basic physical laws. Therefore, a natural question is: Can we trust ML/AI based predictions of Earth system behaviors that are fundamentally driven by physical laws? So, are physics models with meticulous process representation a better choice? Not exactly. Physical models, when firmly rooted in first principles, work great at predicting behaviors of systems with a well-defined set of boundary conditions and variables. However, as a complex system, the number of parameters and the degree of complexity and dynamics in processes, coupling, and scale dependent emergent behaviors make the Earth system behaviors very challenging to predict with physical models composed of deterministic laws. In addition, due to the lack of fundamental understandings, physical models often implement empirical correlations derived from observations with biases from locality of data generation. Because correlation is not necessarily causation or comply with first principles, scaling of model predictions beyond locality is often invalid. Beyond the limitation of models, physics or AI, our knowledge of the Earth system is limited by the lack of observational technologies and resources. Insufficient data density, dimensionality and diversity only offer a sliced (or projected) view of the Earth system, e.g. Plato’s Cave analogy, limiting our capability to better understand and represent fundamental processes in models.
Understanding geometric properties of the latent spaces of natural language processing models allows the manipulation of these properties for improved performance on downstream tasks. One such property is the amount of data spread in a model’s latent space, or how fully the available latent space is being used. We demonstrate that the commonly used measures of data spread, average cosine similarity and a partition function min/max ratio I (V), do not provide reliable metrics to compare the use of latent space across data distributions. We propose and examine six alternative measures of data spread, all of which improve over these current metrics when applied to seven synthetic data distributions. Of our proposed measures, we recommend one principal component-based measure and one entropy-based measure that provide reliable, relative measures of spread and can be used to compare models of different sizes and dimensionalities.
Industry-wide operating experience is a critical source of raw data for reliability and risk model parameter estimations for nuclear power plants. A large portion of operating experience data are failure events stored as reports that contain unstructured data, such as narratives. In current practice, a failure report is usually reviewed and manually coded by analysts. The coding is based on extracting several event characteristics such as system name, component type, sub-part type, failure mode, and failure cause. Event narratives are mostly used to help understand events and extract their characteristics. In this line of research, we aim to maximize the usage of event narratives by leveraging natural language processing (NLP) methods to automatically convert an event narrative to a causal graph. This research has promise to improve physical understanding of failure initiation and propagation and to facilitate use of non-failure data (e.g., near-misses and degradations) to complement the limited data pool of failures. In our previous work, we developed an NLP tool and applied it to analyze a number of licensee event reports submitted by U.S. nuclear power plants to the Nuclear Regulatory Commission. In this paper, we will report our recent research progress in aggregating the results of multiple reports, developing network model(s), and drawing statistical insights.
Additive Manufacturing (AM), referred to as 3D printing, has emerged as a key pillar of Industry 4.0 enabling layer-by-layer fabrication of intricate geometries from CAD models. In parallel, Large Language Models (LLMs), deep learning models for natural language generation trained on vast text corpora, have demonstrated unprecedented capabilities in understanding and generating human-like text. The convergence of these trends opens new opportunities at the intersection of AM and AI/ML, where LLMs can assist engineers and researchers in design, manufacture planning, and knowledge discovery. Recent academic work has begun to explore LLM applications in AM and adjacent fields, such as material science, mechanical engineering, and design for additive manufacturing. This exploration ranges from intelligent process planning to domain-specific knowledge retrieval. This survey provides a comprehensive review of current developments, focusing on peer-reviewed literature contributions that apply, adapt, and advance LLMs in general and domain-specific domains. We analyze state-of-the-art (SOTA) techniques, such as fine-tuning foundational models for specific domains, retrieval-augmented generation (RAG) pipelines, knowledge graph integration, and delve into the architectures and evaluation methods employed. The goal of this survey is to inform researchers and practitioners of the current capabilities and limitations of LLMs in general and in domain-specific applications, and to outline how these models are being tailored to meet the requirements of these applications.
The construction of high-quality numerical features is critical to any quantitative data analysis. Feature engineering has been historically addressed by carefully hand-crafting data representations based on domain expertise. This work views the internal representations of modern deep neural networks (DNNs), called embeddings, as an implicit form of traditional feature engineering. For trained DNNs, we show that these embeddings can reveal interpretable, high-level concepts in unstructured sample data. We use these embeddings in natural language and computer vision tasks to uncover both inherent heterogeneity in the underlying data and human-understandable explanations for it. In particular, we find empirical evidence that there is inherent separability between real data and those generated from AI models.
The poster discusses data analysis of the WAVgraph database and applied machine learning methods for it. The database is a long-term project that seeks to be a comprehensive repository of information on cyber threats and is updated regularly. It was previously unanalyzed and unexplored. The goal was to learn more about it and its contents in order to have a better understanding and enable better use. The data analysis and discovery enabled further exploration through natural language processing, similarity, and clustering methods. The poster shows some of the insights from the analysis and explains the methods used for the machine learning applications.
Abstract Plants, and the biological systems around them, are key to the future health of the planet and its inhabitants. The Plant Science Decadal Vision 2020–2030 frames our ability to perform vital and far‐reaching research in plant systems sciences, essential to how we value participants and apply emerging technologies. We outline a comprehensive vision for addressing some of our most pressing global problems through discovery, practical applications, and education. The Decadal Vision was developed by the participants at the Plant Summit 2019, a community event organized by the Plant Science Research Network. The Decadal Vision describes a holistic vision for the next decade of plant science that blends recommendations for research, people, and technology. Going beyond discoveries and applications, we, the plant science community, must implement bold, innovative changes to research cultures and training paradigms in this era of automation, virtualization, and the looming shadow of climate change. Our vision and hopes for the next decade are encapsulated in the phrase reimagining the potential of plants for a healthy and sustainable future. The Decadal Vision recognizes the vital intersection of human and scientific elements and demands an integrated implementation of strategies for research (Goals 1–4), people (Goals 5 and 6), and technology (Goals 7 and 8). This report is intended to help inspire and guide the research community, scientific societies, federal funding agencies, private philanthropies, corporations, educators, entrepreneurs, and early career researchers over the next 10 years. The research encompass experimental and computational approaches to understanding and predicting ecosystem behavior; novel production systems for food, feed, and fiber with greater crop diversity, efficiency, productivity, and resilience that improve ecosystem health; approaches to realize the potential for advances in nutrition, discovery and engineering of plant‐based medicines, and "green infrastructure." Launching the Transparent Plant will use experimental and computational approaches to break down the phytobiome into a "parts store" that supports tinkering and supports query, prediction, and rapid‐response problem solving. Equity, diversity, and inclusion are indispensable cornerstones of realizing our vision. We make recommendations around funding and systems that support customized professional development. Plant systems are frequently taken for granted therefore we make recommendations to improve plant awareness and community science programs to increase understanding of scientific research. We prioritize emerging technologies, focusing on non‐invasive imaging, sensors, and plug‐and‐play portable lab technologies, coupled with enabling computational advances. Plant systems science will benefit from data management and future advances in automation, machine learning, natural language processing, and artificial intelligence‐assisted data integration, pattern identification, and decision making. Implementation of this vision will transform plant systems science and ripple outwards through society and across the globe. Beyond deepening our biological understanding, we envision entirely new applications. We further anticipate a wave of diversification of plant systems practitioners while stimulating community engagement, underpinning increasing entrepreneurship. This surge of engagement and knowledge will help satisfy and stoke people's natural curiosity about the future, and their desire to prepare for it, as they seek fuller information about food, health, climate and ecological systems.
We evaluate AI-assisted generative capabilities on fundamental numerical kernels in high-performance computing (HPC), including AXPY, GEMV, GEMM, SpMV, Jacobi Stencil, and CG. We test the generated kernel codes for a variety of language-supported programming models, including (1) C++ (e.g., OpenMP [including offload], OpenACC, Kokkos, SyCL, CUDA, and HIP), (2) Fortran (e.g., OpenMP [including offload] and OpenACC), (3) Python (e.g., numpy, Numba, cuPy, and pyCUDA), and (4) Julia (e.g., Threads, CUDA.jl, AMDGPU.jl, and KernelAbstractions.jl). We use the GitHub Copilot capabilities powered by the GPT-based OpenAI Codex available in Visual Studio Code as of April 2023 to generate a vast amount of implementations given simple + + prompt variants. To quantify and compare the results, we propose a proficiency metric around the initial 10 suggestions given for each prompt. Results suggest that the OpenAI Codex outputs for C++ correlate with the adoption and maturity of programming models. For example, OpenMP and CUDA score really high, whereas HIP is still lacking. We found that prompts from either a targeted language such as Fortran or the more general purpose Python can benefit from adding code keywords, while Julia prompts perform acceptably well for its mature programming models (e.g., Threads and CUDA.jl). We expect for these benchmarks to provide a point of reference for each programming model's community. Overall, understanding the convergence of large language models, AI, and HPC is crucial due to its rapidly evolving nature and how it is redefining human-computer interactions.
Understanding and diagnosing the state of a particle accelerator requires navigating high-dimensional control system data, often involving hundreds of interdependent parameters. We propose a novel multimodal embedding framework that jointly learns representations of machine states from both numerical control system readouts and natural language descriptions. This enables the translation of complex machine conditions into human-readable summaries while maintaining fidelity to the underlying physical system. The obtained embeddings are subsequently adapted to an open-weights large language model via cross-attention conditioning. We demonstrate a first implementation trained on European XFEL machine state data. This work covers the embedding model architecture, training methodology, and presents initial examples demonstrating the model's capabilities in action. Due to the general concept of machine state, the model can be easily adapted to other facilities and control system environments.
We explore how visualizations can help users understand what an AI agent is doing as it builds and runs queries over data. As part of the LinkQ system, a natural language interface for querying knowledge graphs with a large language model (LLM), we designed two complementary views: A State Diagram that shows where the agent is within a larger workflow, and a Live Action Display that gives real-time updates about the agent's current task. In a study with 14 practitioners, we found that these visuals helped participants build stronger mental models of the agent's behavior while also increasing their confidence in the system. However, we also observed that users sometimes trusted incorrect outputs simply because the agent appeared to be doing the "right" thing. Our findings point to both the value and risk of visualizing agent behavior in interactive AI systems.