Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Reduced order representation”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

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

At least 127 records · Page 7
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Deep material network via a quilting strategy: visualization for explainability and recursive training for improved accuracy

Recent developments integrating micromechanics and neural networks offer promising paths for rapid predictions of the response of heterogeneous materials with similar accuracy as direct numerical simulations. The deep material network is one such approaches, featuring a multi-layer network and micromechanics building blocks trained on anisotropic linear elastic properties. Once trained, the network acts as a reduced-order model, which can extrapolate the material’s behavior to more general constitutive laws, including nonlinear behaviors, without the need to be retrained. However, current training methods initialize network parameters randomly, incurring inevitable training and calibration errors. Here, we introduce a way to visualize the network parameters as an analogous unit cell and use this visualization to “quilt” patches of shallower networks to initialize deeper networks for a recursive training strategy. The result is an improvement in the accuracy and calibration performance of the network and an intuitive visual representation of the network for better explainability.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Carbon–concentration and carbon–climate feedbacks in CMIP6 models and their comparison to CMIP5 models

Results from the fully and biogeochemically coupled simulations in which CO 2 increases at a rate of 1 % yr –1 (1pctCO2) from its preindustrial value are analyzed to quantify the magnitude of carbon–concentration and carbon–climate feedback parameters which measure the response of ocean and terrestrial carbon pools to changes in atmospheric CO 2 concentration and the resulting change in global climate, respectively. The results are based on 11 comprehensive Earth system models from the most recent (sixth) Coupled Model Intercomparison Project (CMIP6) and compared with eight models from the fifth CMIP (CMIP5). The strength of the carbon–concentration feedback is of comparable magnitudes over land (mean ± standard deviation = 0.97 ± 0.40 PgC ppm –1 ) and ocean (0.79 ± 0.07 PgC ppm –1 ), while the carbon–climate feedback over land (–45.1 ± 50.6 PgC °C –1 ) is about 3 times larger than over ocean (–17.2 ± 5.0 PgC °C –1 ). The strength of both feedbacks is an order of magnitude more uncertain over land than over ocean as has been seen in existing studies. These values and their spread from 11 CMIP6 models have not changed significantly compared to CMIP5 models. The absolute values of feedback parameters are lower for land with models that include a representation of nitrogen cycle. The transient climate response to cumulative emissions (TCRE) from the 11 CMIP6 models considered here is 1.77 ± 0.37 °C EgC –1 and is similar to that found in CMIP5 models (1.63 ± 0.48 °C EgC –1 ) but with somewhat reduced model spread. The expressions for feedback parameters based on the fully and biogeochemically coupled configurations of the 1pctCO2 simulation are simplified when the small temperature change in the biogeochemically coupled simulation is ignored. Decomposition of the terms of these simplified expressions for the feedback parameters is used to gain insight into the reasons for differing responses among ocean and land carbon cycle models.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Efficient and Flexible Hierarchical Data Layouts for a Unified Encoding of Scalar Field Precision and Resolution

To address the problem of ever-growing scientific data sizes making data movement a major hindrance to analysis, we introduce a novel encoding for scalar fields: a unified tree of resolution and precision, specifically constructed so that valid cuts correspond to sensible approximations of the original field in the precision-resolution space. Furthermore, we introduce a highly flexible encoding of such trees that forms a parameterized family of data hierarchies. We discuss how different parameter choices lead to different trade-offs in practice, and show how specific choices result in known data representation schemes such as zfp[52], idx[58], and jpeg2000 [76]. Lastly, we provide system-level details and empirical evidence on how such hierarchies facilitate common approximate queries with minimal data movement and time, using real-world data sets ranging from a few gigabytes to nearly a terabyte in size. Experiments suggest that our new strategy of combining reductions in resolution and precision is competitive with state-of-the-art compression techniques with respect to data quality, while being significantly more flexible and orders of magnitude faster, and requiring significantly reduced resources.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Reduced‐Order Modeling of Energetic Materials Using Physics‐Aware Recurrent Convolutional Neural Networks in a Latent Space (LatentPARC)

Physics-aware deep learning (PADL) has gained popularity for use in spatiotemporal dynamics simulations, such as those in computational modeling of energetic materials (EM). We show that the challenge PADL methods face while learning complex field evolution problems can be simplified and accelerated by decoupling it into two tasks: learning complex geometric features in evolving fields and modeling dynamics over these features in a lower-dimensional feature space. We build upon our previous work on physics-aware recurrent convolutional neural networks (PARC). PARC embeds knowledge of underlying physics into its neural network architecture for more robust and accurate prediction of evolving physical fields. PARC was shown to effectively learn complex nonlinear features such as the formation of hotspots and coupled shock fronts in various initiation scenarios of EMs, as a function of microstructures, serving effectively as a microstructure-aware burn model. Here, we further accelerate PARC and reduce its computational cost by projecting the original dynamics onto a lower-dimensional invariant manifold, or “latent space.” The projected latent representation encodes the complex geometry of evolving fields (e.g., temperature and pressure) in a set of data-driven features. The reduced dimension of this latent space allows us to learn the dynamics during the initiation of EM with a lighter and more efficient model. We observe a significant decrease in training and inference time while maintaining results comparable to PARC at inference. This work takes steps towards enabling rapid prediction of EM thermomechanics at larger scales and characterization of EM structure–property–performance linkages at a full application scale.

Mathematics and Computing↗
DOE OSTI2025

Even Higher-Level Synthesis: An Exploration of AI Hardware Accelerators using HLS4ML

With the rise of artificial intelligence, the popularization of deep learning, and a constantly evolving industry, the demand for flexible and efficient tools has never been greater. As algorithms grow more complex, their runtime and energy consumption increase exponentially. Customized hardware accelerators, long used for specific mathematical operations, remain essential for managing modern applications' computational and power demands. Hardware accelerators can speed up complex computations by orders of magnitude, but their manual design and verification processes are often challenging and time-consuming. High-Level Synthesis (HLS) provides a solution by transforming high-level algorithm descriptions, typically written in C++ or SystemC, into synthesizable RTL suitable for hardware implementation. This approach reduces development time for RTL engineers while offering flexibility beyond what traditional handwritten RTL can provide. We extended this capability to the machine-learning domain with the open-source framework hls4ml, which allows neural networks trained in Python frameworks like Tensorflow or PyTorch to be synthesized into efficient hardware representations for the traditional FPGA and ASIC flows. This breakthrough addresses the growing need for reduced design turnaround and easy verification of ML hardware accelerators with low latency and power efficiency constraints. During this tutorial, we will demonstrate how Python complements HLS by simplifying the ML design process, bridging the gap between software and hardware development. Attendees will explore how we translate neural networks modeled in Python into fixed-point C++ models suitable for HLS workflows. We will dive into strategies like Value-Range Analysis and Quantization-Aware Training, which optimize these designs for deployment and evaluate their accuracy, power consumption, and energy efficiency. To exemplify these concepts, experts from Fermilab will share their experiences applying this technology to high-energy physics experiments, where real-time, low-latency processing is critical. Over the years, Fermilab engineers have demonstrated how deep neural networks, optimized for hardware using hls4ml, can meet the stringent requirements of trigger systems at the CERN Large Hadron Collider. These systems rely on rapid decision-making to process immense data volumes while retaining only the most relevant events for further analysis. The application of hls4ml has also been extended to innovative technologies like smart pixel arrays. These smart pixels integrate ML inference capabilities directly into sensor devices, enabling localized data processing at the pixel level. This approach drastically reduces the need to transmit raw data to external processing units, significantly decreasing power consumption and latency. By embedding neural networks within the pixel architecture, the smart pixels can identify and prioritize relevant data in real time, providing a highly efficient solution for edge computing in scenarios such as particle detectors and imaging systems. Fermilab's work highlights the potential of hardware-accelerated ML in scenarios where both speed and power efficiency are mission-critical. Through this tutorial, attendees will gain valuable insights into the challenges and solutions of deploying ML in hardware. Understanding how HLS and hls4ml streamline the development of neural network-based hardware accelerators is fundamental for the industry's future. Participants will learn how these technologies are shaping the future of AI and scientific computing.

Di Guglielmo, Giuseppe [Fermilab]↗
DOE OSTI2023

Facility-Level Industry Representation for Decarbonization Modeling [Slides]

The largest facilities of energy-intensive materials processing industries are disproportionate contributors to U.S. greenhouse gas (GHG) emissions. In general, industrial energy system modeling for the United States currently represents industrial demands at a much lower resolution than other end-use sector modeling. Therefore, characterizing even a subset of energy-intensive materials processing facilities will capture a significant portion of industrial GHG emissions. In order to further the development of publicly-available data to support modeling of industrial decarbonization, we summarize a set of approaches and results for characterizing the location, energy intensity and mix, process emissions intensity, and general production technology of existing clinker, ammonia, and iron and steel facilities in the United States. We also characterize facilities that represent options for reducing GHG emissions from each industry.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2020

Cluster dynamics modeling of irradiation growth in single crystal Zr

Irradiation growth of single-crystal Zr is modeled with a reduced set of cluster dynamics equations. Nucleation and growth of basal and prismatic dislocation loops are both accounted for in the model. Equations are developed for the time evolution of single point defects, a limited number of point defect clusters, interstitial and vacancy loop nucleation rates, as well as for the growth of vacancy loops on basal planes and interstitial loops on prismatic planes. Reduction of the usual infinite hierarchy of cluster dynamics equations to the simple set studied here is justified on the physical basis of the stability of small loops once nucleated. This simplified cluster dynamics model with a small number of adjustable parameters avoids the complexity of explicit representation of higher order point defect clusters. The model shows consistency with experimental observations of the following aspects: (1) the growth rates of Zr crystals along the a- and c-axes; (2) the onset dose for breakaway irradiation growth; (3) the saturation dislocation loop densities of vacancy and interstitial loops; (4) the effects of cold work and temperature on irradiation growth.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Unifying Combinatorial and Graphical Methods in Artificial Intelligence

Recently, a new graph Laplacian, called the inner product Laplacian, was introduced which generalizes many existing Laplacians, including the normalized and combinatorial Laplacian and their weighted variants. The key observation behind the inner product Laplacian is that by defining appropriate inner product spaces on the vertices and edges, the standard Laplacians can be recovered as Hodge Laplacians over the simplicial complex formed by the edges and vertices. These inner product spaces form a natural way to incorporate non-combinatorial information into the definition of a domain-specific Laplacian. In particular, in contrast to current domain-specific weighting schemes which rely solely on edge weights, information regarding the similarity of non-adjacent vertices and arbitrary pairs of edges can be effectively incorporated into the Laplacian. In order to illustrate this approach we consider the problem of calculating the potential energy of an atomistic configuration using Graph Neural Networks. In comparison with start-of-the-art approaches, such as SchNet, our approach replaces a learned (via auto-encoder) representation of the atom types with an inner product space on atoms based on scientific knowledge (e.g., electronegativity). We will illustrate how this approach captures key chemical properties of the molecules and compare the energy calculations with state-of-the-art neural network approaches. However, to compute the resulting Laplacian involves a mixture of sparse and dense matrix computation and yields a dense matrix as the basis for the graph convolution. This dense convolutional kernel necessitates moving away from the standard message passing framework for graph neural networks and increases the computational cost of applying the kernel. In order to mitigate these costs we investigate means of leveraging the mixed sparse and dense computations to reduce the overall computational cost and how these approaches can be automatically transferred to energy efficient hardware (e.g., field programmable gate arrays (FPGAs)).

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Electron heating in kinetic-Alfvén-wave turbulence

We report analytical and numerical investigations of subion-scale turbulence in low-beta plasmas using a rigorous reduced kinetic model. We show that efficient electron heating occurs and is primarily due to Landau damping of kinetic Alfvén waves, as opposed to Ohmic dissipation. This collisionless damping is facilitated by the local weakening of advective nonlinearities and the ensuing unimpeded phase mixing near intermittent current sheets, where free energy concentrates. The linearly damped energy of electromagnetic fluctuations at each scale explains the steepening of their energy spectrum with respect to a fluid model where such damping is excluded (i.e., a model that imposes an isothermal electron closure). In conclusion, the use of a Hermite polynomial representation to express the velocity-space dependence of the electron distribution function enables us to obtain an analytical, lowest-order solution for the Hermite moments of the distribution, which is borne out by numerical simulations.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

FY25 Theory and Simulation Performance Target: Development of an integrated modeling framework for fusion reactor design and assessment (Final Report)

This report documents the FY25 Theory and Simulation Performance Target (TSPT) of developing an integrated modeling framework for fusion reactor design and assessment (FREDA). Over Q1-Q4, new capabilities were developed across both plasma and engineering domains and demonstrated on an example representation of a Compact Advanced Tokamak with a Dual Cooled Lead Lithium blanket. This represents a first-of-a-kind demonstration of coupled core-to-wall-to-engineering for a reactor. Self-consistent CESOL workflows were applied to provide core, pedestal, and SOL prediction; new modules were developed for energetic particle stability (FAR3D) and transport (TGLF-EP) analysis; and boundary plasma modeling (SOLPS-ITER, BOUT++/Hermes-3) was expanded to evaluate wall and divertor heat fluxes and interface with engineering thermal analysis. A parameterized CAD tool, TRACER, was expanded to generate medium-fidelity divertor, blanket, and coil geometries; OpenFOAM and Diablo workflows were applied for first-wall and divertor thermal analyses with helium cooling; and reduced-order models were created for high-mass-flux divertor cooling. Magnet multiphysics capabilities were verified between Elmer, Diablo, and a new MFEM-based solver, and workflows enable stress, thermal, and neutron-fluence analysis of TF coils with neutronics-driven heating. Nuclear and blanket analysis workflows were demonstrated, including tritium breeding, transport, and CFD-informed thermo-mechanical assessment. Preliminary multi-fidelity uncertainty quantification workflows were applied to boundary modeling codes and shown to achieve variance reductions with fewer high-fidelity boundary simulations. Key findings highlight the challenges of resolving the ITEP gap to find suitable balance between wall and divertor loads, neutron heating, and practical limits of PFC cooling. Next step priorities are to develop automated workflows to check boundary code convergence and detachment, implement tighter physics-engineering CAD provenance tracking, and inclusion of plasma-material interface models for SLAG and tungsten cracking behavior. Collectively, these developments establish sophisticated capabilities for predictive, multi-fidelity, whole-device modeling that integrates plasma physics, materials, magnets, and nuclear engineering to guide pathways to viable Fusion Pilot Plant design points.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

Size-dependent stochastic tensile properties in additively manufactured 316L stainless steel

Recent work in metal additive manufacturing (AM) suggests that mechanical properties may vary with feature size; however, these studies do not provide a statistically robust description of this phenomenon, nor do they provide a clear causal mechanism. Because of the huge design freedom afforded by 3D printing, AM parts typically contain a range of feature sizes, with particular interest in smaller features, so the size effect must be well understood in order to make informed design decisions. This work investigates the effect of feature size on the stochastic mechanical performance of laser powder bed fusion tensile specimens. A high-throughput tensile testing method was used to characterize the effect of specimen size on strength, elastic modulus and elongation in a statistically meaningful way. The effective yield strength, ultimate tensile strength and modulus decreased strongly with decreasing specimen size: all three properties were reduced by nearly a factor of two as feature dimensions were scaled down from 6.25 mm to 0.4 mm. Hardness and microstructural observations indicate that this size dependence was not due to an intrinsic change in material properties, but instead the effects of surface roughness on the geometry of the specimens. Finite element analysis using explicit representations of surface topography shows the critical role surface features play in creating stress concentrations that trigger deformation and subsequent fracture. The experimental and finite element results provide the tools needed to make corrections in the design process to more accurately predict the performance of AM components.

316L stainless steel↗
DOE OSTI2022

Neural-network learning of SPOD latent dynamics

Here, we aim to reconstruct the latent space dynamics of high dimensional, quasi-stationary systems using model order reduction via the spectral proper orthogonal decomposition (SPOD). The proposed method is based on three fundamental steps: in the first, once that the mean flow field has been subtracted from the realizations (also referred to as snapshots), we compress the data from a high-dimensional representation to a lower dimensional one by constructing the SPOD latent space; in the second, we build the time-dependent coefficients by projecting the snapshots containing the fluctuations onto the SPOD basis and we learn their evolution in time with the aid of recurrent neural networks; in the third, we reconstruct the high-dimensional data from the learnt lower -dimensional representation. The proposed method is demonstrated on two different test cases, namely, a compressible jet flow, and a geophysical problem known as the Madden-Julian Oscillation. An extensive comparison between SPOD and the equivalent POD-based counterpart is provided and differences between the two approaches are highlighted. The numerical results suggest that the proposed model is able to provide low rank predictions of complex statistically stationary data and to provide insights into the evolution of phenomena characterized by specific range of frequencies. The comparison between POD and SPOD surrogate strategies highlights the need for further work on the characterization of the interplay of error between data reduction techniques and neural network forecasts.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Enabling probabilistic learning on manifolds through double diffusion maps

Here, we present a generative learning framework for probabilistic sampling that extends Probabilistic Learning on Manifolds (PLoM), which is designed to generate statistically consistent realizations of a random vector in a finite-dimensional Euclidean space, informed by a (representative) set of observations. In its original form, PLoM constructs a reduced-order probabilistic model by combining three main components: (a) kernel density estimation to approximate the underlying probability measure, (b) Diffusion Maps to characterize the manifold of the data, and (c) a reduced-order Itô Stochastic Differential Equation (ISDE) to sample from the learned distribution. However, its sampling dynamics are posed in the ambient space and the retained number of reduced coordinates is chosen by projection-reconstruction error. In practice, this often (i) requires more coordinates than the data’s intrinsic dimension to achieve stable sampling and (ii) lacks a smooth, basis-independent lifting back to the data domain; moreover, standard Diffusion Maps emphasize harmonic eigenfunctions and can miss non-harmonic latent structure. We address these limitations by decoupling geometry learning from sampling: a first Diffusion Maps pass identifies non-harmonic coordinates on which we formulate a full-order ISDE directly in the latent space, while Double Diffusion Maps captures multiscale geometric features and Geometric Harmonics (GH) learns a smooth lifting map to the ambient variables that is independent of the particular diffusion basis. This hybrid design preserves the system’s dynamical richness with a compact geometric representation and enables principled out-of-sample inference. The effectiveness and robustness of the proposed method are illustrated through two numerical studies: one based on data generated from two-dimensional Hermite polynomial functions and another based on high-fidelity simulations of a detonation wave in a reactive flow.

Double diffusion maps↗
DOE OSTI2021

Development of a Unified Taxonomy for HVAC System Faults

Detecting and diagnosing HVAC faults is critical for maintaining building operation performance, reducing energy waste, and ensuring indoor comfort. An increasing deployment of commercial fault detection and diagnostics (FDD) software tools in commercial buildings in the past decade has significantly increased buildings’ operational reliability and reduced energy consumption. A massive amount of data has been generated by the FDD software tools. However, efficiently utilizing FDD data for ‘big data’ analytics, algorithm improvement, and other data-driven applications is challenging because the format and naming conventions of those data are very customized, unstructured, and hard to interpret. This paper presents the development of a unified taxonomy for HVAC faults. A taxonomy is an orderly classification of HVAC faults according to their characteristics and causal relations. The taxonomy includes fault categorization, physical hierarchy, fault library, relation model, and naming/tagging scheme. The taxonomy employs both a physical hierarchy of HVAC equipment and a cause-effect relationship model to reveal the root causes of faults in HVAC systems. A structured and standardized vocabulary library is developed to increase data representability and interpretability. The developed fault taxonomy can be used for HVAC system ‘big data’ analytics such as HVAC system fault prevalence analysis or the development of an HVAC FDD software standard. A common type of HVAC equipment-packaged rooftop unit (RTU) is used as an example to demonstrate the application of the developed fault taxonomy. Two RTU FDD software tools are used to show that after mapping FDD data according to the taxonomy, the meta-analysis of the multiple FDD reports is possible and efficient.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗
DOE OSTI2022

Final Technical Report for "Cloud-based Low-Scaling Quantum Chemistry Simulations for Materials"

The objective of phase I was to develop the basic infrastructure for performing efficient hybrid DFT calculations on solid materials with a turnaround time amenable to the use in large-scale data-driven approaches. To fulfill the goal we have developed a novel algorithm that significantly reduces the cost of exchange matrix formation. The algorithm does so by making use of three ingredients (a) interpolative decomposition of the electron integrals (b) robust pseudospectral method and (c) occ-RI approach. Our published work demonstrates that the algorithm is orders of magnitude faster than any other hybrid-DFT method and is on the order of only 3-4 times slower than pure DFT. All steps of the algorithm developed during phase I can be accelerated to the point that, for large enough systems, the diagonalization of the Fock matrix will become the most expensive step. To treat such large systems we have created an interface to the ASCR-funded PEXSI library. The PEXSI method is used to compute the density matrix directly from the Fock matrix in a manner that preserves the sparsity of the local representation.

Shiozaki, Toru↗
DOE OSTI2024

ZFP: A compressed array representation for numerical computations

HPC trends favor algorithms and implementations that reduce data motion relative to FLOPS. We investigate the use of lossy compressed data arrays in place of traditional IEEE floating point arrays to store the primary data of calculations. Simulation is fundamentally an exercise in controlled approximation, and error introduced by finite-precision arithmetic (or lossy compression) is just one of several sources of error that need to be managed to ensure sufficient accuracy in a computed result. We describe ZFP, a compressed numerical format designed for in-memory storage of multidimensional arrays, and summarize theoretical results that demonstrate that the error of repeated lossy compression can be bounded and controlled. Furthermore, we establish a relationship between grid resolution and compression-induced errors and show that, contrary to conventional floating point, ZFP reduces finite-difference errors with finer grids. We present example calculations that demonstrate data reduction by 4x or more with negligible impact on solution accuracy. Our results further demonstrate several orders-of-magnitude increase in accuracy using ZFP over IEEE floating point and Posits for the same storage budget.

Lindstrom, Peter↗
DOE OSTI2026

Latent Twins

Over the past decade, scientific machine learning has transformed the development of mathematical and computational frameworks for analyzing, modeling, and predicting complex systems. From inverse problems to numerical partial differential equations (PDEs), dynamical systems, and model reduction, these advances have pushed the boundaries of what can be simulated. Yet they have often progressed in parallel, with representation learning and algorithmic solution methods evolving largely as separate pipelines. With Latent Twins, we propose a unifying mathematical framework that creates a hidden surrogate in latent space for the underlying equations. Whereas digital twins mirror physical systems in the digital world, Latent Twins mirror mathematical systems in a learned latent space governed by operators. Through this lens, classical modeling, inversion, model reduction, and operator approximation all emerge as special cases of a single principle. We establish the fundamental approximation properties of Latent Twins for both ordinary differential equations (ODEs) and PDEs and demonstrate the framework across three representative settings: (i) canonical ODEs, capturing diverse dynamical regimes; (ii) a PDE benchmark using the shallow-water equations, contrasting Latent Twin simulations with deep operator network and forecasts with a four-dimensional variational method baseline; and (iii) a challenging real-data geopotential reanalysis dataset, reconstructing and forecasting from sparse, noisy observations. Latent Twins provide a compact, interpretable surrogate for solution operators that evaluate across arbitrary time gaps in a single-shot, while remaining compatible with scientific pipelines such as assimilation, control, and uncertainty quantification. Looking forward, this framework offers scalable, theory-grounded surrogates that bridge data-driven representation learning and classical scientific modeling across disciplines.

Latent Twins↗