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.
Novel uncertainty quantification approach for machine learning based physics informed graph neural networks capable of quickly computing bulk energy system contingencies from a large search space with tolerance to changing grid topologies.
Informational presentation on the use of AI for autonomous design at INL's Scientific Computing and AI division (C500). Presented to visitors on 05/15/2025.
Atmospheric corrosion of metallic parts is a widespread materials degradation phenomena that is challenging to predict given its dependence on many factors (e.g. environmental, physiochemical, and part geometry). For materials with long expected service lives, accurately predicting the degree to which corrosion will degrade part performance is especially difficult due to the stochastic nature of corrosion damage spread across years or decades of service. The Finite Element Method (FEM) is a computational technique capable of providing accurate estimates of corrosion rate by numerically solving complex differential Eqs. characterizing this phenomena. Nevertheless, given the iterative nature of FEM and the computational expense required to solve these complex equations, FEM is ill-equipped for an efficient exploration of the design space to identify factors that accelerate or deter corrosion, despite its accuracy. In this work, a machine learning based surrogate model capable of providing accurate predictions of corrosion with significant computational savings is introduced. Specifically, this work leverages AdaBoosted Decision trees to provide an accurate estimate of corrosion current per width given different values of temperature, water layer thickness, molarity of the solution, and the length of the cathode for a galvanic couple of aluminum and stainless steel.
We describe the ongoing efforts to apply Machine Learning techniques to improve the performance of our accelerator and target. Specially, we are looking to minimize halo beam losses in the absence of a proper physics model, automatically detect and log anomalies in the target support systems such as cooling, and detect and prevent errant beam pulses in the linac. We also describe the infrastructure we use to acquire and stream data to the GPU cluster for training, our code development cycle, and edge computing for model inference. To minimize halo beam losses, we use a Reinforcement Learning technique tested on a virtual accelerator. The target anomaly detection is trained on archived data using incomplete physics models and is made part of the existing target reporting system. The errant beam prevention analyzes beam current and beam phase waveforms as well as accelerator configuration data to predict errant pulses. We also develop continual learning to adapt to changes in the accelerator.
Electron ptychography is a powerful and versatile tool for high-resolution and dose-efficient imaging. Iterative reconstruction algorithms are powerful but also computationally expensive due to their relative complexity and the many hyperparameters that must be optimised. Gradient descent-based iterative ptychography is a popular method, but it may converge slowly when reconstructing low spatial frequencies. Here, in this work, we present a method for accelerating a gradient descent-based iterative reconstruction algorithm by training a neural network (NN) that is applied in the reconstruction loop. The NN works in Fourier space and selectively boosts low spatial frequencies, thus enabling faster convergence in a manner similar to accelerated gradient descent algorithms. We discuss the difficulties that arise when incorporating a NN into an iterative reconstruction algorithm and show how they can be overcome with iterative training. We apply our method to simulated and experimental data of gold nanoparticles on amorphous carbon and show that we can significantly speed up ptychographic reconstruction of the nanoparticles.
Thermal distortion and residual stresses are two important factors that affect the quality and reliability of steel parts manufactured by laser powder bed fusion (LPBF) processes. A cost-effective model for evaluation of those heat effects is needed to refine the manufacturing process and provides insights into the product design and heat treatment. In this study, the layer heating method and sophisticated track-layer scanning method were applied to simulate the thermo-mechanical response of IN625 tube parts built by LPBF. Based on the similarity of temperature field in each layer deposit, a swept mesh was constructed to perform the thermal analysis for top layer, with the rest of layers referring to the temperature by node number offsetting. A novel explicit finite element analysis code accelerated by graphics processing unit was used for the massive-element numerical analysis. The computational accuracy and efficiency of the layer heating and track-layer scanning methods were compared in detail. It is shown that layer heating method can efficiently capture the pattern of stress distribution with reasonable accuracy in stress magnitude. The grouped track-layer scanning method can predict the residual stress and strain more accurately at a higher cost (5 ~ 10×). The elastic strain distribution was compared with the measurement by X-ray diffraction, confirming the accuracy of residual stress prediction.
In the search for stable and active catalysts, density functional theory and machine learning (ML) based models can accelerate the screening of materials. While stability is conveniently addressed on the bulk level of computation, the modelling of catalytic activity requires expensive surface simulations. Here, in this work, we develop models for the surface adsorption energy of O and OH intermediates across a consistent and extensive dataset of pure transition metal oxide surfaces. We show that adsorption energies across metal oxidation states of +2 to +6 are well captured from the metal-oxygen bond strength extracted from the bulk level calculation. Specifically, we calculate the integrated crystal orbital Hamiltonian population (ICOHP) of the metal-oxygen bond in the bulk oxide and employ a simple normalization scheme to obtain a strong correlation with adsorption energetics. By combining our ICOHP descriptor with non DFT features in a Gaussian Process regression (GPR) model, we achieve high model accuracy with mean absolute errors of 0.166 and 0.219 eV for OH and O adsorption, respectively. By targeting the O-OH adsorption energy difference with our GPR model, we predict the the oxygen evolution reaction (OER) activity from bulk descriptors only. Furthermore, we utilize the strong correlation between the COHP and metal oxygen bond lengths to rapidly predict adsorption energetics and catalytic activity from the optimized bulk geometry. Our approach can enable an efficient search for active catalysts by eliminating the need for surface calculations in the initial screening phase.
The use of machine learning (ML)-based surrogate models is a promising technique to significantly accelerate simulation-driven design optimization of internal combustion (IC) engines, due to the high computational cost of running computational fluid dynamics (CFD) simulations. However, training the ML models requires hyperparameter selection, which is often done using trial-and-error and domain expertise. Another challenge is that the data required to train these models are often unknown a priori. In this work, we present an automated hyperparameter selection technique coupled with an active learning approach to address these challenges. The technique presented in this study involves the use of a Bayesian approach to optimize the hyperparameters of the base learners that make up a super learner model. In addition to performing hyperparameter optimization (HPO), an active learning approach is employed, where the process of data generation using simulations, ML training, and surrogate optimization is performed repeatedly to refine the solution in the vicinity of the predicted optimum. The proposed approach is applied to the optimization of a compression ignition engine with control parameters relating to fuel injection, in-cylinder flow, and thermodynamic conditions. It is demonstrated that by automatically selecting the best values of the hyperparameters, a 1.6% improvement in merit value is obtained, compared to an improvement of 1.0% with default hyperparameters. Overall, the framework introduced in this study reduces the need for technical expertise in training ML models for optimization while also reducing the number of simulations needed for performing surrogate-based design optimization.
Abstract Background Sequence alignment lies at the heart of genome sequence annotation. While the BLAST suite of alignment tools has long held an important role in alignment-based sequence database search, greater sensitivity is achieved through the use of profile hidden Markov models (pHMMs). Here, we describe an FPGA hardware accelerator, called HAVAC, that targets a key bottleneck step (SSV) in the analysis pipeline of the popular pHMM alignment tool, HMMER. Results The HAVAC kernel calculates the SSV matrix at 1739 GCUPS on a $$\sim$$ ∼ $3000 Xilinx Alveo U50 FPGA accelerator card, $$\sim$$ ∼ 227× faster than the optimized SSV implementation in nhmmer . Accounting for PCI-e data transfer data processing, HAVAC is 65× faster than nhmmer’s SSV with one thread and 35× faster than nhmmer with four threads, and uses $$\sim$$ ∼ 31% the energy of a traditional high end Intel CPU. Conclusions HAVAC demonstrates the potential offered by FPGA hardware accelerators to produce dramatic speed gains in sequence annotation and related bioinformatics applications. Because these computations are performed on a co-processor, the host CPU remains free to simultaneously compute other aspects of the analysis pipeline.
Low-rank methods have shown success in accelerating simulations of a collisionless plasma described by the Vlasov equation, but still rely on computationally costly linear algebra every time step. We propose a data-driven factorization method using artificial neural networks, specifically with convolutional layer architecture, that trains on existing simulation data. At inference time, the model outputs a low-rank decomposition of the distribution field of the charged particles, and we demonstrate that this step is faster than the standard linear algebra technique. Numerical experiments show that the method achieves comparable reconstruction accuracy for interpolation tasks, generalizing to unseen test data in a manner beyond just memorizing training data; patterns in factorization also inherently followed the same numerical trend as those within algebraic methods (e.g., truncated singular-value decomposition). However, when training on the first 70% of a time-series data and testing on the remaining 30%, the method fails to meaningfully extrapolate. Despite this limiting result, the technique may have benefits for simulations in a statistical steady-state or otherwise showing temporal stability. These results suggest that while the model offers a computationally efficient alternative for datasets with temporal stability, its current formulation is best suited for interpolation rather than for predicting future states in time-evolving systems. This study thus lays the groundwork for further refinement of neural network-based approaches to low-rank matrix factorization in high-dimensional plasma simulations.
This year, the Nuclear Energy Advanced Modeling Simulation program (NEAMS) thermal-hydraulics verification and validation (V&V) work has focused in three areas of Nek5000 V&V-driven development. First, in a close collaborative effort with the U. S. Nuclear Regulatory Commission (NRC) staff, we have continued V&V efforts for the HYMERES-2 project using the OECD/NEA sponsored testing in the PSI PANDA facility. This year’s focus of ANL-NRC collaboration involves Nek5000 setups and validation for a range of problems relevant to and including the HYMERES-2 benchmark from PSI. The primary outcome of this year efforts is a more efficient geometry and inlet modeling simplification after a careful sensitivity study of the inlet profiles and pipe geometries. The resulting modeling choice of a short recycling/fully-developed turbulent inlet is within the experimental uncertainty estimate. This finding simplifies the next step of the cross-V&V HYMERES-2 project. In addition, the ANL team continue to provide assistance to the NRC staff in the form of Nek5000 application support in general and on the use of the HPC platforms of ALCF and INL in particular. This supports the NRC’s assessment of Nek5000 for use with the NRC Blue CRAB code suite. Second, we have implemented and tested more robust model of URANS, namely the k – τ model, a variant of the k-ω model, along with other improvements to RANS Nek5000 modeling in general. Because of its demonstrated robustness and stability, the k – τ model is the only RANS model that has been implemented in the new GPU version of the Nek5000 code, nekRS. Lastly, we report the initial implementation of Jacobian-free Newton Krylov approach to the direct Newton method for steady fluid solvers aimed at acceleration of RANS modeling and at IC improvement for LES campaigns. Also leveraging the Exascale Computing Project (ECP) ANL/CEED & SMR team’s software development effort to support NEAMS problems at large scale of the advanced computing architectures, NekRS, a GPU variant of Nek5000, built on top of kernels from libParanumal using OCCA for portability, has been successfully run on the full system of Summit (4608 nodes, 27648 GPUs).
The Department of Energy’s (DOE’s) vision for an Integrated Research Infrastructure (IRI) is to empower researchers to smoothly and securely meld the DOE’s world-class user facilities and research infrastructure in novel ways in order to radically accelerate discovery and innovation. Performant IRI arises through the continuous interoperability of research workflows with compute, storage, and networking infrastructure, fulfilling researchers’ quests to gain insight from observational and experimental data. Decades of successful research, pilot projects, and demonstrations point to the extraordinary promise of IRI but also indicate the intertwined technological, policy, and sociological hurdles it presents. Creating, developing, and stewarding the conditions for seamless interoperability of DOE research infrastructure, with clear value propositions to stakeholders to opt into an IRI ecosystem, will be the next big step. Governance, funding, and resource allocation are beyond the scope of this document: it seeks to provide a high-level view of potential benefits, focus areas, and the working groups whose formation would further define the testbed’s design, activities, and goals.
The filamentary network of intergalactic medium (IGM) gas that gives origin to the Lyα forest in the spectra of distant quasars encodes information on the physics of structure formation and the early thermodynamics of diffuse baryonic material. Here we use a massive suite of more than 400 high-resolution cosmological hydrodynamical simulations run with the Graphics Processing Unit–accelerated code Cholla to study the IGM at high spatial resolution maintained over the entire computational volume. The simulations capture a wide range of possible IGM thermal histories by varying the photoheating and photoionizing background produced by star-forming galaxies and active galactic nuclei. A statistical comparison of synthetic spectra with the observed 1D flux power spectra of hydrogen at redshifts 2.2 ≤ z ≤ 5.0 and with the helium Lyα opacity at redshifts 2.4 < z < 2.9 tightly constrains the photoionization and photoheating history of the IGM. By leveraging the constraining power of the available Lyα forest data to break model degeneracies, we find that the IGM experienced two main reheating events over 1.2 Gyr of cosmic time. For our best-fit model, hydrogen reionization completes by z R ≈ 6.0 with a first IGM temperature peak of T 0 ≃ 1.3 × 10 4 K and is followed by the reionization of He ii that completes by z R ≈ 3.0 and yields a second temperature peak of T 0 ≃ 1.4 × 10 4 K. We discuss how our results can be used to obtain information on the timing and the sources of hydrogen and helium reionization.
The Elastic Analysis Facility (EAF) hosted at Fermi National Accelerator Laboratory (Fermilab) is a platform being developed with the goal of providing a fast and efficient facility for physics analysis. As high-energy physics moves towards collecting larger datasets, such as those from the High-Luminosity LHC, the EAF strives to provide a powerful and adaptable framework for future colliders and multi-experiment institutions. Currently, the EAF supports several experiments including CMS, NOvA, and DUNE as well as serving accelerator physicists and beam line operations through integrated software and secure connections to Fermilab's computing resources. In addition, the EAF was designed with a user-friendly interface, intended to be more intuitive for emerging generations of physicists, that is still accessible for established styles of analysis. The EAF can also achieve better analysis efficiency due to the modernization of software and tools that can better utilize Fermilab's computing power. Furthermore, its design incorporates industry standards whenever possible, enhancing its sustainability and making it a possible template for other national or international laboratories and research facilities. Overall, the EAF is a forward-looking solution that will meet the evolving needs of particle physics, ensuring readiness for future colliders and multi-experiment research institutions.
HAMR is a library defining an accelerator technology agnostic memory model that bridges between accelerator technologies (CUDA, HIP, ROCm, OpenMP, Sycl, OpenCL, Kokos, etc) and traditional CPUs in heterogeneous computing environments. HAMR is light weight and implemented in modern C++. HAMR can be used to manage memory with in a single code or as a data model for coupling codes in a technologically agnostic way. HAMR provides a Python module for coupling C++ and Python codes which implements zero-copy data transfers to and from Python using the Numpy array interface and Numba CUDA array interface protocols.
Scientists, engineers and programmers at Fermilab are tackling today’s most challenging computational problems. Their solutions, motivated by the needs of worldwide research in particle physics and accelerators, help America stay at the forefront of innovation.
We present a novel machine learning (ML)-based method to accelerate conservative-to-primitive inversion, focusing on hybrid piecewise polytropic and tabulated equations of state. Traditional root-finding techniques are computationally expensive, particularly for large-scale relativistic hydrodynamics simulations. To address this, we employ feedforward neural networks (NNC2PS and NNC2PL), trained in PyTorch (2.0+) and optimized for GPU inference using NVIDIA TensorRT (8.4.1), achieving significant speedups with minimal accuracy loss. The NNC2PS model achieves 𝐿 1 and 𝐿 ∞ errors of 4.54 × 10 −7 and 3.44 × 10−6, respectively, while the NNC2PL model exhibits even lower error values. TensorRT optimization with mixed-precision deployment substantially accelerates performance compared to traditional root-finding methods. Specifically, the mixed-precision TensorRT engine for NNC2PS achieves inference speeds approximately 400 times faster than a traditional single-threaded CPU implementation for a dataset size of 1,000,000 points. Ideal parallelization across an entire compute node in the Delta supercomputer (dual AMD 64-core 2.45 GHz Milan processors and 8 NVIDIA A100 GPUs with 40 GB HBM2 RAM and NVLink) predicts a 25-fold speedup for TensorRT over an optimally parallelized numerical method when processing 8 million data points. Moreover, the ML method exhibits sub-linear scaling with increasing dataset sizes. We release the scientific software developed, enabling further validation and extension of our findings. By exploiting the underlying symmetries within the equation of state, these findings highlight the potential of ML, combined with GPU optimization and model quantization, to accelerate conservative-to-primitive inversion in relativistic hydrodynamics simulations.
Here, in this study, we present a technique for using machine learning (ML) for single-qubit gate synthesis on field-programmable logic for a superconducting transmon-based quantum computer based on simulated studies. Our approach is multi-stage. We first “bootstrap” a model based on simulation with access to the full state vector for measuring gate fidelity. We next present an algorithm, named adapted randomized benchmarking (ARB), for fine-tuning the gate on hardware based on measurements of the devices. We also present techniques for deploying the model on programmable devices with care to reduce the required resources. While the techniques here are applied to a transmon-based computer, many of them are portable to other architectures.