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.
The CRADA objective is to 1) Generalize web tools for data analytics and machine learning, and 2) Develop and test a smart-alarm system at Jefferson Lab.
Artificial intelligence and machine learning (AI/ML) are becoming important tools for scientific modeling and simulation as in several other fields such as image analysis and natural language processing. ML techniques can leverage the computing power available in modern systems and reduce the human effort needed to configure experiments, interpret and visualize results, draw conclusions from huge quantities of raw data, and build surrogates for physics based models. Domain scientists in fields like fluid dynamics, microelectronics and chemistry can automate many of their most difficult and repetitive tasks or improve the design times by use of the faster ML-surrogates. However, modern ML and traditional scientific highperformance computing (HPC) tend to use completely different software ecosystems. While ML frameworks like PyTorch and TensorFlow provide Python APIs, most HPC applications and libraries are written in C++. Direct interoperability between the two languages is possible but is tedious and error-prone. In this work, we show that a compiler-based approach can bridge the gap between ML frameworks and scientific software with less developer effort and better efficiency. We use the MLIR (multi-level intermediate representation) ecosystem to compile a pre-trained convolutional neural network (CNN) in PyTorch to freestanding C++ source code in the Kokkos programming model. Kokkos is a programming model widely used in HPC to write portable, shared-memory parallel code that can natively target a variety of CPU and GPU architectures. Our compiler-generated source code can be directly integrated into any Kokkosbased application with no dependencies on Python or cross-language interfaces.
Surveillance of nuclear material storage containers is required to ensure that the container safety boundary is maintained during the service life of the container. As many of these container types are the first and only containment barrier to release protecting the worker, public, and environment, it is imperative to develop robust surveillance tools to characterize container degradation. Machine learning (ML) techniques have matured quickly in the past decade and are slowly becoming a routine application in data analysis. While more popular architectures have been developed with different applications in mind, these can be easily translated to container surveillance requirements. In this study, we developed a ML model based on the Detectron2 framework to identify slip lid containers and common exterior container defects, primarily dents. Commercial-off-the-shelf cameras are combined into a stationary camera array or are mounted to a robotic arm to provide more mobility for positioning needs. Results from the model evaluations on the images indicate that the stationary camera arrays outperform the accuracy of the mobile camera systems for both individual camera and composite image detections. Overall, however, the detection accuracies of the systems fell short of 50% and were less than satisfactory. Future efforts to improve the system involve focusing on a single camera deployable solution that combines a controllable light source that eliminates one of the biggest environmental factors that influence ML model detection performance. Additional training set collection is planned to build a more robust model that enables accurate detection from a wider range of imaging conditions.
Overview presentation given at the 2024 FECM / NETL Carbon Management Research Project Review Meeting on NETL's Bipartisan Infrastructure Law-funded Smart CO2 Transport-Route Planning Tool and associated geodatabase. This machine learning informed, data-driven public resource was designed to inform regulators, industry, and researchers plan and develop safe and efficient transport routes across the country.
From the lightest Hydrogen isotopes up to the recently synthesized Oganesson (Z = 118), it is estimated that as many as about 8,000 atomic nuclei could exist in nature. Most of these nuclei are too short-lived to be occurring on Earth, but they play an essential role in astrophysical events such as supernova explosions or neutron star mergers that are presumed to be at the origin of most heavy elements in the Universe. Understanding the structure, reactions, and decays of nuclei across the entire chart of nuclides is an enormous challenge because of the experimental difficulties in measuring properties of interest in such fleeting objects and the theoretical and computational issues of simulating strongly-interacting quantum many-body systems. Nuclear density functional theory (DFT) is a fully microscopic theoretical framework which has the potential of providing such a quantitatively accurate description of nuclear properties for every nucleus in the chart of nuclides. Thanks to high-performance computing facilities, it has already been successfully applied to predict nuclear masses, global patterns of radioactive decay like β or γ decay, and several aspects of the nuclear fission process such as, e.g., spontaneous fission half-lives. Yet, predictive simulations of nuclear spectroscopy—the low-lying excited states and transitions between them—or of nuclear fission, or the quantification of theoretical uncertainties and their propagation to basic or applied nuclear science applications, would require several orders of magnitude more calculations than currently possible. However, most of this computational effort would be spent into generating a suitable basis of DFT wavefunctions. Such a task could potentially be considerably accelerated by borrowing tools from the field of machine learning and artificial intelligence. In this paper, we review different approaches to applying supervised and unsupervised learning techniques to nuclear DFT.
The Interactive Corpus Analysis Tool (ICAT) is an interactive machine learning dashboard for unlabeled text/natural language processing datasets that allows a user to iteratively and visually define features, explore and label instances of their dataset, and simultaneously train a logistic regression model. ICAT was created to allow subject matter experts in a specific domain to directly train their own models for unlabeled datasets visually, without needing to be a machine learning expert or needing to know how to code the models themselves. This approach allows users to directly leverage the power of machine learning, but critically, also involves the user in the development of the machine learning model.
The interaction between defect and solute atoms is critical to the thermodynamic and kinetic behavior of metallic alloys under exposure to high-energy radiation, causing irradiation damage in materials. Radiation can generate non-equilibrium concentrations of point defects such as vacancies and interstitials. The excess point defects not only accelerate diffusional processes such as precipitation that cause radiation embrittlement, but also change the pathway of phase transformations, including nucleation processes. Understanding these defect behaviors is complicated by the challenge and complexity of addressing each possible local and discrete distribution of environments and chemical interactions around targeted defects-solute or solute-solute complexes. To resolve the challenge, machine learning regression techniques have emerged as powerful tools that can train and construct an energy model to accurately describe the chemical interactions of solutes and defects. In Fiscal Year 2022, the work focused on the workflow development and demonstration using machine learning regression, density functional theory, cluster expansion, and Monte Carlo simulation to predict the effects of ternary solute elements (e.g., aluminum and molybdenum) and point defects on the Cr-rich $\alpha^{\prime}$ precipitation in multicomponent FeCr model alloys. The computational outcomes include the prediction of the ternary phase diagram, vacancy formation energy for different compositions, and the effect of vacancies on the nucleation of Cr-rich clusters. The simulations predict a pronounced change of Cr solubility in bcc Fe by the addition of Al and the rejection of Al atoms from $\alpha^{\prime}$ precipitates. Additionally, the simulations show the formation of Cr-vacancy clusters as the initial nuclei for stable nucleation and growth of $\alpha^{\prime}$ particles. The results demonstrate important outcomes and applications of using machine learning pipeline to study model or commercial alloys with multicomponent solute species and point defects.
Next generation systems, such as edge devices, will need to provide efficient processing of machine learning (ML) algorithms along several metrics, including energy, performance, area, and latency. However, the quickly evolving field of ML makes it extremely difficult to generate accelerators able to support a wide variety of algorithms. At the same time, designing accelerators in hardware description languages (HDLs) by hand is hard and time consuming, and does not allow quick exploration of the design space. In this paper we present the Software Defined Accelerators From Learning Tools Environment (SODALITE), an automated open source high-level ML framework-to-verilog compiler targeting ML Application-Specific Integrated Circuits (ASICs) chiplets. The SODALITE approach will implement optimal designs by seamlessly combining custom components generated through high-level synthesis (HLS) with templated and fully tunable Intellectural Properties (IPs) and macros, integrated in an extendable resource library. Through a closed loop design space exploration engine, developers will be able to quickly explore their hardware designs along different dimensions.
The Nuclear and Chemical Sciences (NACS) Division furnishes the expertise in the scientific areas of chemical, nuclear and isotopic sciences that are foundational in the Laboratory’s national security missions. This expertise is maintained and advanced through identification, development and application of state-of-the-art theoretical, computational and experimental methods and tools. Recent developments in artificial intelligence and machine learning (AI/ML) techniques enabled by advances in computing capabilities and widespread availability of powerful software implementations have made use of these techniques ubiquitous across both science and industry. While the scope of AI/ML applications is incredibly large and evolves very rapidly, the topics most relevant to NACS missions fall into the general category of detecting, categorizing or identifying features in large, complex datasets using either supervised or unsupervised learning. This covers both basic scientific data analysis and the development of efficient surrogate models of real-life technological systems, experimental detectors, or theoretical models. To remain at the forefront of its core scientific disciplines, NACS must both cultivate ML expertise as well as continuously explore applying this expertise to new problems or utilizing new methods. This document identifies the key areas where this support is critical and provides a strategy for investing in them.
Abstract The general method of graph coarsening or graph reduction has been a remarkably useful and ubiquitous tool in scientific computing and it is now just starting to have a similar impact in machine learning. The goal of this paper is to take a broad look into coarsening techniques that have been successfully deployed in scientific computing and see how similar principles are finding their way in more recent applications related to machine learning. In scientific computing, coarsening plays a central role in algebraic multigrid methods as well as the related class of multilevel incomplete LU factorizations. In machine learning, graph coarsening goes under various names, e.g., graph downsampling or graph reduction. Its goal in most cases is to replace some original graph by one which has fewer nodes, but whose structure and characteristics are similar to those of the original graph. As will be seen, a common strategy in these methods is to rely on spectral properties to define the coarse graph.
Here, a review of current trends in scientific computing reveals a broad shift to open-source and higher-level programming languages such as Python and growing career opportunities over the next decade. Open-source modeling tools accelerate innovation in equation-based and data-driven applications. Significant resources have been deployed to develop data-driven tools (PyTorch, TensorFlow, Scikit-learn) from tech companies that rely on machine learning services to meet business needs while keeping the foundational tools open. Open-source equation-based tools such as Pyomo, CasADi, Gekko, and JuMP are also gaining momentum according to user community and development pace metrics. Integration of data-driven and principles-based tools is emerging. New compute hardware, productivity software, and training resources have the potential to radically accelerate progress. However, long-term support mechanisms are still necessary to sustain the momentum and maintenance of critical foundational packages.
Generalization bounds are a critical tool to assess the training data requirements of Quantum Machine Learning (QML). Recent work has established guarantees for in-distribution generalization of quantum neural networks (QNNs), where training and testing data are drawn from the same data distribution. However, there are currently no results on out-of-distribution generalization in QML, where we require a trained model to perform well even on data drawn from a different distribution to the training distribution. Here, we prove out-of-distribution generalization for the task of learning an unknown unitary. In particular, we show that one can learn the action of a unitary on entangled states having trained only product states. Since product states can be prepared using only single-qubit gates, this advances the prospects of learning quantum dynamics on near term quantum hardware, and further opens up new methods for both the classical and quantum compilation of quantum circuits.
Black-box machine learning models are recognized as useful tools for prediction applications, but the algorithmic complexity of some models causes interpretation challenges. Explainability methods have been proposed to provide insight into these models, but there is little research focused on supervised modeling with functional data inputs. We argue that, especially in applications of high consequence, it is important to explicitly model the functional dependence in a black-box analysis to not obscure or misrepresent patterns in explanations. As such, we propose the V ariable importance E xplainable E lastic S hape A nalysis (VEESA) pipeline for training supervised machine learning models with functional inputs. The pipeline is an analysis process that includes the data preprocessing, modeling, and post-hoc explanations. The preprocessing is done using elastic functional principal components analysis, which accounts for vertical and horizontal variability in functional data and, ultimately, allows for explanations in the original data space that identify the important functional variability without bias due to correlated variables. Here, we demonstrate the pipeline on two high-consequence applications: explosives classification for national security and inkjet printer identification in forensic science. The applications exhibit the VEESA pipeline’s ability to provide an understanding of the characteristics of the functional data useful for prediction. Code for implementing the pipeline is available in the veesa R package (and supplemental python code).
Energy-efficient scientific applications require insight into how high performance computing system features impact the applications' power and performance. This insight can result from the development of performance and power models. Here, in this article, we use the modeling and prediction tool MuMMI (Multiple Metrics Modeling Infrastructure) and 10 machine learning methods to model and predict performance and power consumption and compare their prediction error rates. We use an algorithm-based fault-tolerant linear algebra code and a multilevel checkpointing fault-tolerant heat distribution code to conduct our modeling and prediction study on the Cray XC40 Theta and IBM BG/Q Mira at Argonne National Laboratory and the Intel Haswell cluster Shepard at Sandia National Laboratories. Our experimental results show that the prediction error rates in performance and power using MuMMI are less than 10% for most cases. By utilizing the models for runtime, node power, CPU power, and memory power, we identify the most significant performance counters for potential application optimizations, and we predict theoretical outcomes of the optimizations. Based on two collected datasets, we analyze and compare the prediction accuracy in performance and power consumption using MuMMI and 10 machine learning methods.
First-principle simulations are at the heart of the high-energy physics research program. They link the vast data output of multi-purpose detectors with fundamental theory predictions and interpretation. This review illustrates a wide range of applications of modern machine learning to event generation and simulation-based inference, including conceptional developments driven by the specific requirements of particle physics. New ideas and tools developed at the interface of particle physics and machine learning will improve the speed and precision of forward simulations, handle the complexity of collision data, and enhance inference as an inverse simulation problem.
Efficient quantum control is necessary for practical quantum computing implementations with current technologies. Conventional algorithms for determining optimal control parameters are computationally expensive, largely excluding them from use outside of the simulation. Existing hardware solutions structured as lookup tables are imprecise and costly. By designing a machine learning model to approximate the results of traditional tools, a more efficient method can be produced. Such a model can then be synthesized into a hardware accelerator for use in quantum systems. In this study, we demonstrate a machine learning algorithm for predicting optimal pulse parameters. This algorithm is lightweight enough to fit on a low-resource FPGA and perform inference with a latency of 175 ns and pipeline interval of 5 ns with > 0.99 gate fidelity. In the long term, such an accelerator could be used near quantum computing hardware where traditional computers cannot operate, enabling quantum control at a reasonable cost at low latencies without incurring large data bandwidths outside of the cryogenic environment.
Abstract Generalization bounds are a critical tool to assess the training data requirements of Quantum Machine Learning (QML). Recent work has established guarantees for in-distribution generalization of quantum neural networks (QNNs), where training and testing data are drawn from the same data distribution. However, there are currently no results on out-of-distribution generalization in QML, where we require a trained model to perform well even on data drawn from a different distribution to the training distribution. Here, we prove out-of-distribution generalization for the task of learning an unknown unitary. In particular, we show that one can learn the action of a unitary on entangled states having trained only product states. Since product states can be prepared using only single-qubit gates, this advances the prospects of learning quantum dynamics on near term quantum hardware, and further opens up new methods for both the classical and quantum compilation of quantum circuits.
Ab initio molecular dynamics (AIMD) simulations have become an important tool used in the construction of equations of state (EOS) tables for warm dense matter. Due to computational costs, only a limited number of system state conditions can be simulated, and the remaining EOS surface must be interpolated for use in radiation-hydrodynamic simulations of experiments. In this work, we develop a thermodynamically consistent EOS model that utilizes a physics-informed machine learning approach to implicitly learn the underlying Helmholtz free-energy from AIMD generated energies and pressures. The model, referred to as PIML-EOS, was trained and tested on warm dense polystyrene producing a fit within a 1% relative error for both energy and pressure and is shown to satisfy both the Maxwell and Gibbs–Duhem relations. In addition, we provide a path toward obtaining thermodynamic quantities, such as the total entropy and chemical potential (containing both ionic and electronic contributions), which are not available from current AIMD simulations.