Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “physics”

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 1,369 records · Page 76
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Programmable intrusion detection for distributed energy resources in cyber–physical networked microgrids

We present a programmable intrusion detection method is presented to identify the malicious attacks to distributed energy resources (DERs) in the cyber-physical networked microgrids. The proposed method injects small programmable signals into the system and uses the response to identify abnormal conditions. Because of the low or even zero inertia induced by integrations of DER power-electronic-interfaces, microgrids have very limited resilience capability; and thus, being sensitive to attacks. One microgrid's malfunction caused by attacks can easily propagate to its neighboring systems when several microgrids are connected, leading to catastrophic electricity supply failures. Through the presented method, malicious intrusions can be effectively detected, located, and defended for securing microgrids. Theoretical derivations are provided to define the programmable detection rules. The detection rule is easy and flexible to update, making it difficult for attack actors to gain the knowledge of the detection rules, in order to avoid being detected. Numerical results on a cyber-physical networked microgrids system show that the proposed method is effective and efficient in precisely locating intrusion attacks to the microgrids system.

42 ENGINEERING↗
DOE OSTI2024

A methodology to develop multi-physics dynamic fuel cell system models validated with vehicle realistic drive cycle data

Fuel cell (FC) technology has been identified as a technically attractive solution to decarbonize the transportation sector, especially for heavy-duty vehicles. In this context, the industry and the scientific community are in need of advanced fuel cell systems (FCS) models that are able to replicate real -world operating conditions. Due to the scarcity of said models in the open literature, this study aimed to develop a comprehensive methodology to calibrate and validate multi -physics dynamic FCS models. Therefore, the key contribution of this paper is the detailed description of the calibration process for each component and the calibration order. The specific focus here was to accurately describe the behavior of the FC stack as well as the cathode, anode, and cooling circuits of the balance of plant. The model was calibrated with the aid of experimental data from a Toyota Mirai FC electric vehicle, which was predominantly retrieved from the vehicle's Controller Area Network (CAN) bus system thereby negating the need for major intrusion into the powertrain system. The validation process was deemed successful with the model being able to truthfully replicate the characteristics of the FC vehicle operated on the World-wide harmonized Light duty Test Cycle (WLTC) 3b and US06 driving cycle. The time -resolved physical parameters such as the cathode pressure, mass flow, or the FC stack temperature were captured with high fidelity, while the overall performance parameters such as the H2 consumption in the stack and the system, and the compressor energy consumption were predicted accurately with a deviation lower than 0.47%, 1.75% and 1.89% with respect to the experimental data, respectively.

Lopez-Juarez, Marcos↗
DOE OSTI2025

Physics-based hybrid machine learning for critical heat flux prediction with uncertainty quantification

Critical heat flux (CHF) is a key quantity in nuclear system modeling due to its impact on heat transfer, safety margins, and reactor performance. This study develops and validates an uncertainty-aware hybrid modeling approach that combines machine learning with physics-based models to predict CHF in cases of dryout. The Biasi and Bowring empirical correlations were paired with three ML uncertainty quantification (UQ) techniques: deep neural network (DNN) ensembles, Bayesian neural networks (BNNs), and deep Gaussian processes (DGPs). A pure ML model without a base model was evaluated for comparison. Model performance was assessed under plentiful (7,350 points) and limited (9 points) training data scenarios using parity, uncertainty distributions, and calibration curves. Results show that the Biasi hybrid DNN ensemble achieved the best overall performance, with a mean absolute relative error of 1.846%, and well-calibrated uncertainty estimates. The BNN-based hybrids showed slightly higher error (2.14%) but superior uncertainty calibration. DGP models underperformed, with over 6% error and poor uncertainty calibration. All hybrid models outperformed pure machine learning configurations, demonstrating resistance against data scarcity. These findings indicate that hybrid modeling significantly improves predictive accuracy, interpretability, and resilience to data scarcity. The integration of uncertainty awareness provides actionable confidence in CHF predictions, which is vital for safety-critical decisions in nuclear applications. This hybrid approach offers a viable pathway for deploying ML models in reactor analysis tools while preserving domain knowledge and physical consistency.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2020

Molecular Transducers of Physical Activity Consortium (MoTrPAC): Mapping the Dynamic Responses to Exercise

Exercise provides a robust physiological stimulus that evokes cross-talk between a wide range of tissues via hormonal, cellular and molecular signals that when repeated regularly over time (i.e., training) improves physiological capacity, benefits numerous organ systems and decreases the risk for premature mortality. In spite of its fundamental importance, a gap remains in identifying the detailed molecular signals and responses to exercise and its benefits for health and disease prevention. To further knowledge in this emerging area, the Molecular Transducers of Physical Activity Consortium (MoTrPAC) was established to generate a molecular map of the effects of acute and chronic exercise. MoTrPAC investigators are conducting preclinical and clinical studies to better understand the systemic effects of exercise across multiple tissues. The preclinical studies were performed on more than 700 rats with the analysis of twenty or more organs per rat. The clinical study involves a large cohort of ~2600 individuals with a range of ages and fitness levels (untrained, trained, highly trained) who will be evaluated physiologically and by molecular probing of blood, muscle and adipose tissues before and after acute and chronic endurance or resistance exercise. Multi-omic analyses (genomics, epigenomics, transcriptomics, proteomics and metabolomics/lipidomics) will be performed followed by state-of-the-art bioinformatics to create a molecular map of exercise. This mapping project will provide a public database that is expected to enhance our understanding of the health benefits of physical activity in many organ systems and potentially provide insight into mitigating diseases through activity.

Sanford, James A.↗
DOE OSTI2021

On the eigenvector bias of Fourier feature networks: From regression to solving multi-scale PDEs with physics-informed neural networks

Physics-informed neural networks (PINNs) are demonstrating remarkable promise in integrating physical models with gappy and noisy observational data, but they still struggle in cases where the target functions to be approximated exhibit high-frequency or multi-scale features. Here in this work we investigate this limitation through the lens of Neural Tangent Kernel (NTK) theory and elucidate how PINNs are biased towards learning functions along the dominant eigen-directions of their limiting NTK. Using this observation, we construct novel architectures that employ spatio-temporal and multi-scale random Fourier features, and justify how such coordinate embedding layers can lead to robust and accurate PINN models. Numerical examples are presented for several challenging cases where conventional PINN models fail, including wave propagation and reaction–diffusion dynamics, illustrating how the proposed methods can be used to effectively tackle both forward and inverse problems involving partial differential equations with multi-scale behavior.

42 ENGINEERING↗
DOE OSTI2020

Coupling laser physics to radiation-hydrodynamics

In order to accurately model implosion hydrodynamics in a radiation-hydrodynamics code, it is essential to include accurate accounting for energy deposition physics. In inertial confinement fusion (ICF), where capsules are driven by lasers or laser-driven x-rays, energy deposition profiles and energy transport have a strong impact on the development and evolution of capsule dynamics and hydrodynamic instabilities. Nevertheless, accurately modeling laser beam propagation in radiation-hydrodynamics codes presents unique challenges associated with disparate resolution requirements, the potential to seed spurious noise in highly unstable systems, and computational expense. We discuss a new method for coupling laser ray-tracing physics to a radiation hydrodynamics code, developed in the process of implementing the Mazinisin laser ray-trace into the xRAGE radiation hydrodynamics code. In contrast to previous approaches, in which laser ray-tracing is performed on the radiation-hydrodynamics mesh, our method involves a mesh generation and evolution strategy that addresses the unique requirements of the laser ray-trace in a separate mesh, enabling performance enhancements and strategies to reduce noise seeded by the discretization of beams into computational rays. In addition, we have employed several methods to ensure that spurious mesh imprinting is minimized. These involved optimizing the laser and radiation-hydrodynamics meshes as well as interpolation between them and requires the use of an exact initialization method for the radiation-hydrodynamics mesh. Furthermore, these techniques have enabled efficient computation of laser-driven implosions and other experiments with minimal introduction of spurious noise.

47 OTHER INSTRUMENTATION↗
DOE OSTI2026

Validating corrosion models: Influence of physical properties

In chloride containing environments, two metals in physical contact can undergo galvanic corrosion limiting the lifetime of components. Being able to accurately predict galvanic corrosion damage distributions over time has not been widely presented in literature. Therefore, Finite Element Method (FEM) corrosion models were experimentally validated for two galvanic couples as a function of governing equations, environment, anode material, anode:cathode ratio, and time. Carbon steel/stainless steel (CS/SS) and zinc/stainless steel (Zn/SS) galvanic couples were exposed to NaCl solutions at room temperatures for up to 14 days. For the galvanic couples and environments, the Laplace equation with variable conductivity and reactions was sufficient to model the experimental corrosion damage. Increasing the chloride concentration for the CS/SS galvanic couple, regardless of the anode:cathode ratio, decreased the observed and modeled corrosion damage. Decreasing the anode:cathode ratio (i.e., increasing the cathode length), increased the experimentally observed and modeled corrosion damage. For small anode:cathode ratios, the governing equations deviate over long time periods with the Nernst-Plank equation being conservative. For the Zn/SS galvanic couple, the cathode length controls the dominant cathodic reduction reaction. For small cathode lengths, the hydrogen evolution reaction is dominant. For larger cathode lengths, the oxygen reduction reaction is dominant. The results are discussed with regard to the influence of solution chemistry, ohmic drop, and governing reactions. Overall, validated FEM models were presented, and the resultant models, physics, and mechanisms can be applied to other corrosion scenarios with confidence.

Carbon steel↗
DOE OSTI2021

Jas4pp — A data-analysis framework for physics and detector studies

This paper describes the Jas4pp framework for exploring physics cases and for detector-performance studies of future particle collision experiments. Jas4pp is a multi-platform Java program for numeric calculations, scientific visualization in 2D and 3D, storing data in various file formats and displaying collision events and detector geometries. It also includes complex data-analysis algorithms for function minimization, regression analysis, event reconstruction (such as jet reconstruction), limit settings and other libraries widely used in particle physics. The framework can be used with several scripting languages, such as Python/Jython, Groovy and JShell. Several benchmark tests discussed in the paper illustrate significant improvements in the performance of the Groovy and JShell scripting languages compared to the standard Python implementation in C. Furthermore, the improvements for numeric computations in Java are attributed to recent enhancements in the Java Virtual Machine.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Striving to translate shale physics across ten orders of magnitude: What have we learned?

Shales will play an important role in the successful transition of energy from fossil-based resources to renewables in the coming decades. Aside from being a significant source of low-carbon intensity fuels, like natural gas, they also serve as geologic seals of subsurface formations that may be used to isolate nuclear waste, sequester CO 2 , or store intermittent energy (e.g., solar hydrogen). Despite their importance, shales pose significant engineering and environmental challenges due to their nanoporous structure and extreme heterogeneity that spans at least ~10 orders of magnitude in spatial scale. Two challenges inhibit a system-level understanding: (1) the physics of fluid flow and phase behavior in shales are poorly understood due to the dominant molecular interactions between minerals and fluids under confinement, and (2) the apparent lack of scale separation that prevents a reliable (closed) description of the physics at any single scale of observation. In this review, we focus on the latter issue and discuss scale translation, which in its broadest sense is transforming data or simulations from one spatiotemporal scale to another. While effective scale translation is not exclusive to shales, but all geologic porous media, the need for it is especially acute in shales given their high degree of heterogeneity. Classical theories like homogenization, while indispensable, fail when scales are not separated. Other methods, like numerical upscaling, scale-translate in only one direction: small to large, but not the reverse, called downscaling. However, the confluence of advances in three areas are bringing challenging problems such as shales within reach: increased computational power and scalable algorithms; high-resolution imaging and multi-modal data acquisition; and machine learning to process massive amounts of data. While these advances equip geoscientists with a wide array of experimental and computational tools, no individual tool can probe the entire gamut of heterogeneity in shales. Their effective use, therefore, requires an ability to bridge between various data types obtained at different scales. The aim of this review is to present a coherent account of computational and experimental methods that may be used to achieve just that, i.e., to perform scale translation. We provide a broader definition of scale translation, one that transcends classical homogenization and upscaling methods, but is consistent with them and accommodates notions like downscaling and data translation. After a brief introduction to homogenization, we review hybrid methods, numerical upscaling and its recent extensions, multiscale computing, high-resolution imaging, and machine learning. We place particular emphasis on multiscale computing and propose an algorithmic framework to bridge between the pore (micro) and Darcy (macro) scales. Throughout the paper, we draw comparisons between the various methods and highlight their (often hidden) similarities, differences, benefits, and pitfalls. We finally conclude with two case studies on shales that exemplify some of the methods presented.

58 GEOSCIENCES↗
DOE OSTI2026

Development of physics-consistent conditional diffusion model to overcome data scarcity in critical heat flux

Deep generative modeling provides a powerful pathway to overcome data scarcity in energy-related applications where experimental data are often limited. By learning the underlying probability distribution of the training dataset, deep generative models, such as the diffusion model, can generate high-fidelity synthetic samples that statistically resemble the training data. Such synthetic data generation can significantly enrich the size and diversity of the available training data, and more importantly, improve the robustness of downstream machine learning models in predictive tasks. The objective of this paper is to investigate the effectiveness of diffusion models for overcoming data scarcity in nuclear energy applications. By leveraging a public dataset on critical heat flux which covers a wide range of commercial nuclear reactor operational conditions, we developed a diffusion model that can generate an arbitrary amount of synthetic samples. Since a vanilla diffusion model can only generate samples randomly, we also developed a conditional diffusion model capable of generating targeted critical heat flux data under user-specified thermal-hydraulic conditions. The performance of the diffusion model was evaluated based on its ability to capture empirical feature distributions and pair-wise correlations, as well as to maintain physical consistency. The results showed that both the diffusion model and conditional diffusion model can successfully generate realistic and physics-consistent critical heat flux data. Furthermore, uncertainty quantification results demonstrate that the conditional diffusion model is highly effective in augmenting critical heat flux data while maintaining acceptable levels of uncertainty.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2023

Machine learning surrogate of physics-based building-stock simulator for end-use load forecasting

Building energy models are used to simulate heat and mass transfer and estimate end-use load in buildings. With the proliferation of solar photovoltaics on residential and commercial buildings, increasingly, buildings are expected to provide grid services, for which accurate and computationally efficient building energy simulations and end-use load prediction are imperative. Existing building energy simulation tools, however, have significant computational overhead that make them less practical in real-time deployment for optimization, design, uncertainty quantification and control in building energy management systems. Here this article presents a data-driven machine learning model based on light gradient boosting method (LightGBM) as a surrogate for a physics-based simulator for residential buildings to predict end-use load. The machine learning based surrogate model accounts for time-series related variables, seasonality and trend component of end-use load, and history of end-use load. The accuracy of the surrogate model is assessed on the prediction of the load profiles of 100 different houses in Cook County, Illinois, USA. The LightGBM surrogate model is shown to reduce the root-mean-squared error by 53% relative to a reference decision tree (DT) based model reported previously in the literature. Moreover, the model predicts the load spikes and high-ramp rate events throughout the year which are often the Achilles heel of other models in the literature. The machine learning based surrogate model is demonstrated to be computationally efficient, with a ten-fold reduction in the computational time compared to a physics-based building energy simulation, and suitable for uncertainty analysis and real-time control of building characteristics in response to uncertainty.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2026

Disruption modelling for engineering and physics design of ST-E1 fusion power plant

Plasma disruptions represent a critical challenge for high-performance tokamak operations, as they can compromise machine integrity and reduce operational availability. Although future fusion devices essentially need to incorporate strategies to minimise disruption occurrence, complete avoidance remains unattainable. Consequently, assessing and characterising unmitigated disruption consequences is fundamental for the design and qualification of next-generation fusion power plants. This work supports the pre-conceptual design of ST-E1, a low aspect-ratio Tokamak Fusion Power Plant developed by Tokamak Energy Ltd., by presenting a comprehensive disruption modelling approach applied across different design stages. The methodology integrates both physics and engineering considerations to evaluate the impact of disruptions on machine performance and structural integrity. From an engineering perspective, several ST-E1 layout options were analysed to investigate the electromagnetic response of key components under disruption-induced loads, enabling comparison between alternative design solutions. On the physics side, a broad set of disruption scenarios was explored, scanning operational space parameters, plasma-material interactions, and associated thermal loads. Furthermore, the study examined variations in disruption behaviour arising from different reference equilibria, focusing on a range starting from Double Null to Single Null configurations, reflecting the increasing up-down asymmetry consequences. The results reveal significant contrasts in plasma dynamics and structures electromagnetic behaviour between configurations, highlighting the importance of disruption modelling in guiding design choices. These analyses have proven instrumental in shaping ST-E1 development, offering critical insights for mitigating risks and optimising future fusion power plant designs.

Borowiec, Katarzyna [ORNL] (ORCID:0000000335911739↗
DOE OSTI2023

A physics-driven Σ-Y atomization model for heavy-duty engine simulations

The atomization of a liquid jet is a multi-scale and multi-physics problem of interest for many engineering applications. Particularly, it drives the fuel-oxidizer mixing and dictates the efficiency of combustion engines. High-fidelity multi-phase simulations remain challenging due to the excessive computational cost required to capture all the atomization scales. Thus, atomization models are necessary to represent sub-grid liquid structures. Here, in this work, a modified Σ-Y model in the context of Eulerian-Lagrangian Spray Atomization (ELSA) is used to transport the surface area density of the spray. The model's predictive performance is assessed under various operating conditions relevant to heavy-duty engines using the Engine Combustion Network (ECN) Spray C and Spray D research-grade injectors. The classic droplet collision formulation of the Σ-Y model alone does not replicate the response of the measured spray surface area to changes in injector, ambient pressure, and injection pressure, requiring individual tuning of the model's parameters. Instead, a transition between dense and dilute spray breakup mechanisms is proposed in terms of the average droplet spacing. The collision breakup mechanism represents the dilute spray, whereas the droplet size in the dense spray is driven by a competition between the integral scale of turbulence and the balance between the turbulent kinetic energy and the surface energy of the droplets. Such an approach minimizes the requirements for model tuning. Moreover, the role of the constants of a compressible standard k-ε RANS model is assessed in the injector's internal flow and external spray simulation framework, and an updated set is proposed. The results are validated against x-ray radiography and Ultra-Small Angle X-ray Scattering (USAXS) data and highlight the predictive capabilities of the proposed physics-driven Σ-Y model, which is compatible with engine simulation turnaround times.

42 ENGINEERING↗
DOE OSTI2022

Dose Summation Strategies for External Beam Radiation Therapy and Brachytherapy in Gynecologic Malignancy: A Review from the NRG Oncology and NCTN Medical Physics Subcommittees

Definitive, nonsurgical management of gynecologic malignancies involves external beam radiation therapy (EBRT) and/or brachytherapy (BT). Summation of the cumulative dose is critical to assess the total biologic effective dose to targets and organs at risk. Cumulative dose calculation from EBRT and BT can be performed with or without image registration (IR) and biologic dose summation. Among these dose summation strategies, linear addition of dose-volume histogram (DVH) parameters without IR is the global standard for composite dose reporting. This approach stems from an era without image guidance and simple external beam and brachytherapy treatment approaches. With technological advances, EBRT and high-dose-rate BT have evolved to allow for volume-based treatment planning and delivery. Modern conformal therapeutic radiation involves volumetric or intensity modulated EBRT, capable of simultaneously treating multiple targets at different specified dose levels. Therefore, given the complexity of modern radiation treatment, the linear addition of DVH parameters from EBRT and high-dose-rate BT is challenging to represent the combined dose distribution. Deformable image registration (DIR) between EBRT and image guided brachytherapy (IGBT) data sets may provide a more nuanced calculation of multimodal dose accumulation. However, DIR is still nascent in this regard, and needs further development for accuracy and efficiency for clinical use. Biologic dose summation can combine physical dose maps from EBRT and each IGBT fraction, thereby generating a composite DVH from the biologic effective dose. However, accurate radiobiologic parameters are tissue-dependent and not well characterized. A combination of voxel-based DIR and biologic weighted dose maps may be the best approximation of dose accumulation but remains invalidated. The purpose of this report is to review dose summation strategies for EBRT and BT, including conventional equivalent dose in 2-Gy fractions dose summation without image registration, physical dose summation using 3-dimensional rigid IR and DIR, and biologic dose summation. We also provide general clinical workflows for IGBT with a focus on cervical cancer.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2020

NSFnets (Navier-Stokes flow nets): Physics-informed neural networks for the incompressible Navier-Stokes equations

In the last 50 years there has been a tremendous progress in solving numerically the Navier-Stokes equations using finite differences, finite elements, spectral, and even meshless methods. Yet, in many real cases, we still cannot incorporate seamlessly (multi-fidelity) data into existing algorithms, and for industrial-complexity applications the mesh generation is time consuming and still an art. Moreover, solving ill-posed problems (e.g., lacking boundary conditions) or inverse problems is often prohibitively expensive and requires different formulations and new computer codes. Here, we employ physics-informed neural networks (PINNs), encoding the governing equations directly into the deep neural network via automatic differentiation, to overcome some of the aforementioned limitations for simulating incompressible laminar and turbulent flows. We develop the Navier-Stokes flow nets (NSFnets) by considering two different mathematical formulations of the Navier-Stokes equations: the velocity-pressure (VP) formulation and the vorticity-velocity (VV) formulation. Since this is a new approach, we first select some standard benchmark problems to assess the accuracy, convergence rate, computational cost and flexibility of NSFnets; analytical solutions and direct numerical simulation (DNS) databases provide proper initial and boundary conditions for the NSFnet simulations. The spatial and temporal coordinates are the inputs of the NSFnets, while the instantaneous velocity and pressure fields are the outputs for the VP-NSFnet, and the instantaneous velocity and vorticity fields are the outputs for the VV-NSFnet. This is unsupervised learning and, hence, no labeled data are required beyond boundary and initial conditions and the fluid properties. The residuals of the VP or VV governing equations, together with the initial and boundary conditions, are embedded into the loss function of the NSFnets. No data is provided for the pressure to the VP-NSFnet, which is a hidden state and is obtained via the incompressibility constraint without extra computational cost. Unlike the traditional numerical methods, NSFnets inherit the properties of neural networks (NNs), hence the total error is composed of the approximation, the optimization, and the generalization errors. Here, we empirically attempt to quantify these errors by varying the sampling (“residual”) points, the iterative solvers, and the size of the NN architecture. For the laminar flow solutions, we show that both the VP and the VV formulations are comparable in accuracy but their best performance corresponds to different NN architectures. The initial convergence rate is fast but the error eventually saturates to a plateau due to the dominance of the optimization error. For the turbulent channel flow, we show that NSFnets can sustain turbulence at , but due to expensive training we only consider part of the channel domain and enforce velocity boundary conditions on the subdomain boundaries provided by the DNS data base. We also perform a systematic study on the weights used in the loss function for balancing the data and physics components, and investigate a new way of computing the weights dynamically to accelerate training and enhance accuracy. In the last part, we demonstrate how NSFnets should be used in practice, namely for ill-posed problems with incomplete or noisy boundary conditions as well as for inverse problems. We obtain reasonably accurate solutions for such cases as well without the need to change the NSFnets and at the same computational cost as in the forward well-posed problems. As a result, we also present a simple example of transfer learning that will aid in accelerating the training of NSFnets for different parameter settings.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Physics-based adaptivity of a spectral method for the Vlasov–Poisson equations based on the asymmetrically-weighted Hermite expansion in velocity space

We propose a spectral method for the 1D-1V Vlasov–Poisson system where the discretization in velocity space is based on asymmetrically-weighted Hermite functions, dynamically adapted via a scaling α and shifting u of the velocity variable. Specifically, at each time instant an adaptivity criterion selects new values of α and u based on the numerical solution of the discrete Vlasov–Poisson system obtained at that time step. Once the new values of the Hermite parameters α and u are fixed, the Hermite expansion is updated and the discrete system is further evolved for the next time step. The procedure is applied iteratively over the desired temporal interval. The key aspects of the adaptive algorithm are: the map between approximation spaces associated with different values of the Hermite parameters that preserves total mass, momentum and energy; and the adaptivity criterion to update α and u based on physics considerations relating the Hermite parameters to the average velocity and temperature of each plasma species. For the discretization of the spatial coordinate, we rely on Fourier functions and use the implicit midpoint rule for time stepping. The resulting numerical method possesses intrinsically the property of fluid-kinetic coupling, where the low-order terms of the expansion are akin to the fluid moments of a macroscopic description of the plasma, while kinetic physics is retained by adding more spectral terms. Moreover, the scheme features conservation of total mass, momentum and energy associated in the discrete, for periodic boundary conditions. A set of numerical experiments confirms that the adaptive method outperforms the non-adaptive one in terms of accuracy and stability of the numerical solution.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Multi-head physics-informed neural networks for learning functional priors and uncertainty quantification

In numerous applications, the integration of prior knowledge and historical information is essential, particularly for tasks requiring the solution of ordinary or partial differential equations (ODEs/PDEs) in data-sparse or noisy environments. For instance, achieving accurate solutions to time-dependent PDEs with limited initial condition measurements necessitates an effective strategy for embedding prior knowledge. Hard-parameter sharing architectures in neural networks (NNs) have demonstrated success in both traditional and scientific machine learning domains, facilitating the learning of informative representations. Here, in this study, we introduce a novel, yet efficient, method to enhance physics-informed neural networks (PINNs) by incorporating a multi-head structure that enables the learning of functional priors from both empirical data and governing physical laws. This prior information can then be used to address data sparsity and high-level noise in solving ODE/PDE problems with uncertainty quantification (UQ). The approach, termed Multi-Head PINN (MH-PINN), consists of a shared body NN and multiple head NNs, each corresponding to an individual PINN instance. Our framework for functional prior learning is carried out in two stages: (1) training the MH-PINNs to develop a shared body NN alongside multiple head NNs, and (2) employing these trained head NNs to estimate a prior distribution through a normalizing flow-based density estimator. The learned functional prior can then be applied as a regularization mechanism in deterministic contexts or as an informative prior within a Bayesian inference framework, aiding in the resolution of subsequent ODE/PDE tasks. We evaluate the efficacy of MH-PINNs across five benchmark problems, including a high-dimensional parametric PDE, all characterized by data sparsity or substantial noise levels. Our findings reveal that MH-PINNs deliver accurate solutions and robust UQ, demonstrating adaptability across a range of complex and challenging scenarios.

Bayesian inference↗