Multi-Agent Reinforcement Learning for Adaptive Mesh Refinement
Explore the source record for details and available documents.
Mathematics and Computing, Computer science↗
SEARCH · Search NASA
Results for “adaptive refinement”
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.
Developing Performance Portable Adaptive Mesh Refinement for Astrophysics
Explore the source record for details and available documents.
Computer Science↗
Parthenon: Adaptive Mesh Refinement for Exascale Astrophysics
Explore the source record for details and available documents.
Astronomy and Astrophysics↗
Gradient-enhanced physics-informed neural networks for forward and inverse PDE problems
Deep learning has been shown to be an effective tool in solving partial differential equations (PDEs) through physics-informed neural networks (PINNs). PINNs embed the PDE residual into the loss function of the neural network, and have been successfully employed to solve diverse forward and inverse PDE problems. However, one disadvantage of the first generation of PINNs is that they usually have limited accuracy even with many training points. Here, we propose a new method, gradient-enhanced physics-informed neural networks (gPINNs), for improving the accuracy of PINNs. gPINNs leverage gradient information of the PDE residual and embed the gradient into the loss function. We tested gPINNs extensively and demonstrated the effectiveness of gPINNs in both forward and inverse PDE problems. Our numerical results show that gPINN performs better than PINN with fewer training points. Additionally, we combined gPINN with the method of residual-based adaptive refinement (RAR), a method for improving the distribution of training points adaptively during training, to further improve the performance of gPINN, especially in PDEs with solutions that have steep gradients.
42 ENGINEERING↗
Multilevel Monte Carlo methods for the Grad-Shafranov free boundary problem
The equilibrium configuration of a plasma in an axially symmetric reactor is described mathematically by a free boundary problem associated with the celebrated Grad-Shafranov equation. The presence of uncertainty in the model parameters introduces the need to quantify the variability in the predictions. This is often done by computing a large number of model solutions on a computational grid for an ensemble of parameter values and then obtaining estimates for the statistical properties of solutions. In this study, we explore the savings that can be obtained using multilevel Monte Carlo methods, which reduce costs by performing the bulk of the computations on a sequence of spatial grids that are coarser than the one that would typically be used for a simple Monte Carlo simulation. We examine this approach using both a set of uniformly refined grids and a set of adaptively refined grids guided by a discrete error estimator. Numerical experiments show that multilevel methods dramatically reduce the cost of simulation, with cost reductions typically on the order of 60 or more and possibly as large as 200. Furthermore, adaptive griding results in more accurate computation of geometric quantities such as x-points associated with the model.
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
A block-spectral adaptive H-/$p$-refinement strategy for shock-dominated problems
An adaptive H-/p-refinement strategy using a novel sensor is devised and tested in a block-spectral compressible Euler code equipped with adaptive-mesh refinement (AMR) and high-order flux-reconstruction numerics. At each Gauss quadrature point (or solution point) within each spectral block (or mesh element) the discrete velocity jump ΔU = ∂U/∂y 1 Δy 1 + ∂V/∂y 2 Δy 2 + ∂W/∂y 3 Δy 3 is calculated and normalized by the local speed of sound, a. Here, the grid spacing, Δx i , is calculated in each direction as the distance between auxiliary Gauss-Lobatto points, staggered relative to the solution points. The polynomial order is increased from p = 0 to p = p max in regions of weak compression, (ΔU/a) crit < ΔU/a < 0 and kept at p = p max in regions of flow expansion ΔU/a ≥ 0, while staying at the H = 0 base mesh level. Regions experiencing strong compressions, i.e. ΔU/a < (ΔU/a) crit , are H-refined up to H = H max where H max is applied at the location of maximum compression, ΔU/a = min(ΔU/a) in the domain, while keeping p = 0 to guarantee robustness and monotonicity of the solution in the H refined region. The critical value of (ΔU/a) crit = -0.06 is found to effectively separate smooth and non-smooth solution regions, supported by a 1D detonation initiation test case in ideal gas and a shock-to-detonation transition in high explosives. Using this value, the Sod shock tube, Shu-Osher problem, double Mach reflection and a 2D detonation in a high-explosive are simulated with the proposed adaptive H-/p-refinement. In the Sod shock tube case, p-refinement resolves the (weak) contact discontinuity while H-refinement enhances the grid resolution in the shock exploiting the monotonicity of the p = 0 reconstruction. For the Shu-Osher problem, p-refinement captures the small-scale oscillations trailing the shock that would be otherwise attenuated, while H-refinement triggered by the ΔU-sensor appropriately tracks the shock. In the double Mach reflection problem, H-refinement confines the numerical diffusion around the reflected shock while p-refinement recaptures many physical features trailing the shock. Finally, in the 2D high-explosive detonation case, H-refinement follows the leading shock and resolves the curvature of the detonation wave, while p-refinement adds resolution to the trailing reaction zone. Finally, the proposed methodology is tested in a detonation-wave propagation test case in high-explosives with numerical predictions comparing favorably against experiments.
97 MATHEMATICS AND COMPUTING↗
From PINNs to PIKANs: recent advances in physics-informed machine learning
Physics-Informed Neural Networks (PINNs) have emerged as a key tool in Scientific Machine Learning since their introduction in 2017, enabling the efficient solution of ordinary and partial differential equations using sparse measurements. Over the past few years, significant advancements have been made in the training and optimization of PINNs, covering aspects such as network architectures, adaptive refinement, domain decomposition, and the use of adaptive weights and activation functions. A notable recent development is the Physics-Informed Kolmogorov-Arnold Networks (PIKANS), which leverage a representation model originally proposed by Kolmogorov in 1957, offering a promising alternative to traditional PINNs. In this review, we provide a comprehensive overview of the latest advancements in PINNs, focusing on improvements in network design, feature expansion, optimization techniques, uncertainty quantification, and theoretical insights. We also survey key applications across a range of fields, including biomedicine, fluid and solid mechanics, geophysics, dynamical systems, heat transfer, chemical engineering, and beyond. Lastly, we review computational frameworks and software tools developed by both academia and industry to support PINN research and applications.
Kolmogorov-Arnold networks↗
An adaptive space-time phase field formulation for dynamic fracture of brittle shells based on LR NURBS
Here, we present an adaptive space-time phase field formulation for dynamic fracture of brittle shells. Their deformation is characterized by the Kirchhoff–Love thin shell theory using a curvilinear surface description. All kinematical objects are defined on the shell’s mid-plane. The evolution equation for the phase field is determined by the minimization of an energy functional based on Griffith’s theory of brittle fracture. Membrane and bending contributions to the fracture process are modeled separately and a thickness integration is established for the latter. The coupled system consists of two nonlinear fourth-order PDEs and all quantities are defined on an evolving two-dimensional manifold. Since the weak form requires C 1 -continuity, isogeometric shape functions are used. The mesh is adaptively refined based on the phase field using Locally Refinable (LR) NURBS. Time is discretized based on a generalized-α method using adaptive time-stepping, and the discretized coupled system is solved with a monolithic Newton–Raphson scheme. Finally, the interaction between surface deformation and crack evolution is demonstrated by several numerical examples showing dynamic crack propagation and branching.
42 ENGINEERING↗
CFD simulations of electric motor end ring cooling for improved thermal management
Proper thermal management of an electric motor for vehicle applications extends its operating range. One cooling approach is to impinge Automatic Transmission Fluid (ATF) onto the rotor end ring. Increased ATF coverage correlates to enhanced heat transfer. Computational Fluid Dynamics (CFD) analytical tools provide a mechanism to assess motor thermal management prior to hardware fabrication. The complexity of the fluid flow (e.g., jet atomization, interface tracking, wall impingement) and heat transfer makes these simulations challenging. Computational costs are high when solving these flows on high-speed rotating meshes. Typically, a Volume-of Fluid (VOF) technique (i.e., two-fluid system) is used to resolve ATF dynamics within this rotating framework. Suitable numerical resolution of the relevant physics for thin films under strong inertial forces at high rotor speeds is computationally expensive, further increasing the run times. In this work, a numerical study of rotor-ring cooling by ATF is presented using a patent automated Cartesian cut-cell based method coupled with Automatic Mesh Refinement (AMR). This approach automatically creates the Cartesian mesh on-the-fly and can effectively handle complex rotating geometries by adaptively refining the mesh based on local gradients in the flow field which results in better resolution of the air-ATF interface. A Single non-inertial Reference Frame (SRF) approach is used to account for the rotating geometry and to further improve the overall computational efficiency. Quasi-steady state conditions are targeted in the analysis of the results. Important physics such as ATF jet structure, velocity detail near the air-jet interface, ATF coverage/accumulation on the ring surface, and cooling capacity are presented for a low-resolution Reynolds averaged Navier-Stokes (RANS), high-resolution RANS, and high-resolution Large-Eddy Simulation (LES) models. Computations are scaled over hundreds of cores on a supercomputer to maximize turnaround time. Each numerical approach is shown to capture the general trajectory of the oil jet prior to surface impingement. The high-resolution LES simulation, however, is superior in capturing small scale details and heat transfer between the free jet and surrounding air.
42 ENGINEERING↗
A comprehensive study of non-adaptive and residual-based adaptive sampling for physics-informed neural networks
Physics-informed neural networks (PINNs) have shown to be effective tools for solving both forward and inverse problems of partial differential equations (PDEs). PINNs embed the PDEs into the loss of the neural network using automatic differentiation, and this PDE loss is evaluated at a set of scattered spatio-temporal points (called residual points). The location and distribution of these residual points are highly important to the performance of PINNs. However, in the existing studies on PINNs, only a few simple residual point sampling methods have mainly been used. Here, we present a comprehensive study of two categories of sampling for PINNs: non-adaptive uniform sampling and adaptive nonuniform sampling. We consider six uniform sampling methods, including (1) equispaced uniform grid, (2) uniformly random sampling, (3) Latin hypercube sampling, (4) Halton sequence, (5) Hammersley sequence, and (6) Sobol sequence. We also consider a resampling strategy for uniform sampling. To improve the sampling efficiency and the accuracy of PINNs, we propose two new residual-based adaptive sampling methods: residual-based adaptive distribution (RAD) and residual-based adaptive refinement with distribution (RAR-D), which dynamically improve the distribution of residual points based on the PDE residuals during training. Hence, we have considered a total of 10 different sampling methods, including six non-adaptive uniform sampling, uniform sampling with resampling, two proposed adaptive sampling, and an existing adaptive sampling. We extensively tested the performance of these sampling methods for four forward problems and two inverse problems in many setups. Our numerical results presented in this study are summarized from more than 6000 simulations of PINNs. Here, we show that the proposed adaptive sampling methods of RAD and RAR-D significantly improve the accuracy of PINNs with fewer residual points for both forward and inverse problems. Furthermore, the results obtained in this study can also be used as a practical guideline in choosing sampling methods.
97 MATHEMATICS AND COMPUTING↗
Adaptive Space-Time Methods for Large Scale Optimal Design
When modeling complex physical systems with advanced dynamics, such as shocks and singularities, many classic methods for solving partial differential equations can return inaccurate or unusable results. One way to resolve these complex dynamics is through r-adaptive refinement methods, in which a fixed number of mesh points are shifted to areas of high interest. The mesh refinement map can be found through the solution of the Monge-Ampére equation, a highly nonlinear partial differential equation. Due to its nonlinearity, the numerical solution of the Monge-Ampére equation is nontrivial and has previously required computationally expensive methods. In this report, we detail our novel optimization-based, multigrid-enabled solver for a low-order finite element approximation of the Monge-Ampére equation. This fast and scalable solver makes r-adaptive meshing more readily available for problems related to large-scale optimal design. Beyond mesh adaptivity, our report discusses additional applications where our fast solver for the Monge-Ampére equation could be easily applied.
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
MOCMC: Method of Characteristics Moment Closure, a Numerical Method for Covariant Radiation Magnetohydrodynamics
In this work, we present a conservative numerical method for radiation magnetohydrodynamics with frequency-dependent full transport in stationary spacetimes. This method is stable and accurate for both large and small optical depths and radiation pressures. The radiation stress–energy tensor is evolved in flux-conservative form, and closed with a swarm of samples that each transport a multigroup representation of the invariant specific intensity along a null geodesic. In each zone, the enclosed samples are used to efficiently construct a Delaunay triangulation of the unit sphere in the comoving frame, which in turn is used to calculate the Eddington tensor, average source terms, and adaptively refine the sample swarm. Furthermore, radiation four-forces are evaluated in the moment sector in a semi-implicit fashion. The radiative transfer equation is solved in invariant form deterministically for each sample. Since each sample carries a discrete representation of the full spectrum, the cost of evaluating the transport operator is independent of the number of frequency groups, representing a significant reduction of algorithmic complexity for transport in frequency-dependent problems. The major approximation we make in this work is performing scattering in an angle-averaged way. Local adaptivity in samples also makes this scheme more amenable to nonuniform meshes than a traditional Monte Carlo method. We describe the method and present results on a suite of test problems. We find that Method of Characteristics Moment Closure converges at least as ~N -1 , rather than the canonical Monte Carlo N -1/2 , where N is the number of samples per zone.
79 ASTRONOMY AND ASTROPHYSICS↗
End-to-end GPU acceleration of low-order-refined preconditioning for high-order finite element discretizations
In this article, we present algorithms and implementations for the end-to-end GPU acceleration of matrix-free low-order-refined preconditioning of high-order finite element problems. The methods described here allow for the construction of effective preconditioners for high-order problems with optimal memory usage and computational complexity. The preconditioners are based on the construction of a spectrally equivalent low-order discretization on a refined mesh, which is then amenable to, for example, algebraic multigrid preconditioning. The constants of equivalence are independent of mesh size and polynomial degree. For vector finite element problems in H(curl) and H(div) (e.g., for electromagnetic or radiation diffusion problems), a specially constructed interpolation–histopolation basis is used to ensure fast convergence. Detailed performance studies are carried out to analyze the efficiency of the GPU algorithms. The kernel throughput of each of the main algorithmic components is measured, and the strong and weak parallel scalability of the methods is demonstrated. The different relative weighting and significance of the algorithmic components on GPUs and CPUs is discussed. Results on problems involving adaptively refined nonconforming meshes are shown, and the use of the preconditioners on a large-scale magnetic diffusion problem using all spaces of the finite element de Rham complex is illustrated.
97 MATHEMATICS AND COMPUTING↗
Efficient Computation of Doppler-Broadened Elastic Scattering Kernel Moments Using Ladder-Operator Formulation
Anefficient routine for computing Legendre moments of the Doppler-broadened elastic scattering kernel, including resonance scattering effects, has been implemented in the ISOXML module of Griffin. Isotropic scattering in the center-of-mass system and the ideal gas model for target motion are assumed. A ladder-operator formulation is introduced to compute all Legendre moments from order 0 to N simultaneously, enabling near-linear scaling of computational cost with respect to the maximum Legendre order. A physics-based strategy for constructing outgoing energy grids has also been developed, in which a tailored base grid is combined with adaptive refinement to maintain accuracy while limiting the number of outgoing energy points. For energies between resonances, a constant cross-section model is employed to further reduce computational cost. In addition, a quantitative criterion is derived to determine isotope-wise cut-off incident energies based on a prescribed up-scattering probability coverage. For 238U, up to incident energies of approximately 75, 230, and 661 eV at 294, 900, and 2500 K (corresponding to a 2% up-scattering probability threshold), computation of P0 kernels requires 1–8 s and computation of P0–P5 kernels requires 0.4–4 min using a single thread, while maintaining 1–3% relative error in up-scattering probability. These results demonstrate that the proposed formulation enables accurate and computationally practical Doppler-broadened kernel generation for online multigroup cross-section production in Griffin.
Doppler-broadening↗
Magnetic mesh generation and field line reconstruction for scrape-off layer and divertor modeling in stellarators
The design of divertor targets and baffles for optimal heat and particle exhaust from magnetically confined fusion plasmas requires a combination of fast, low-fidelity models (such as EMC3-Lite (Feng 2022 Plasma Phys. Control. Fusion 64 125012)) for scoping studies and high-fidelity ones (such as EMC3-EIRENE (Feng et al 2014 Contrib. Plasma Phys. 54 426–31)) for verification. Both of those approaches benefit from a magnetic flux tube mesh for fast interpolation and mapping of field line segments (Feng et al 2005 Phys. Plasmas 12 052505). A new automated mesh generator for unstructured quadrilateral flux tubes with adaptive refinement is presented and integrated into FLARE (Frerichs 2024 Nucl. Fusion 64 106034). For HSX with an extended first wall, it is found that several layers of flux tubes can span the entire half field period before splitting is required. This is an advantage over the traditional setup of the EMC3-EIRENE mesh where careful construction of several sub- domains is required already for the much tighter present first wall. In particular, there is no longer the need to manually construct a suitable outer boundary for the mesh. The divide and conquer paradigm with unstructured mesh layout offers a powerful alternative for fast head load approximation that is suitable for integration into optimization workflows. Further examples for W7-X and CTH demonstrate the versatile application range.
70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
A deformation-dependent coupled Lagrangian/semi-Lagrangian meshfree hydromechanical formulation for landslide modeling
Abstract The numerical modelling of natural disasters such as landslides presents several challenges for conventional mesh-based methods such as the finite element method (FEM) due to the presence of numerically challenging phenomena such as severe material deformation and fragmentation. In contrast, meshfree methods such as the reproducing kernel particle method (RKPM) possess unique features conducive to modelling extreme events such as the absence of a structured mesh and the ease of adaptive refinement, among others. While the semi-Lagrangian reproducing kernel (SL-RK) shape functions of RKPM defined in the current configuration have proven to be effective in extreme event modelling, the computational cost for the re-evaluation of the shape functions at every time step is costly. In this work, a deformation-dependent coupling of the Lagrangian reproducing kernel (L-RK) and SL-RK approximations is proposed for the solution of a hydro-mechanical formulation for effective simulations of landslides. The ramp function is constructed based on an equivalent plastic strain as a deformation-dependent transition from L-RK shape functions to SL-RK ones as the deformation progresses. The particular focus of the paper will be on modelling seepage-induced landslides with a mixed $$u$$ u – $$p$$ p formulation to couple the solid and fluid phases. Examples are presented to examine the effectiveness of this coupled Lagrangian/semi-Lagrangian reproducing kernel (L–SL RK) formulation and to highlight its performance in landslide modelling.
Baek, Jonghyuk↗
NucMesh: nuclear reactor geometry creation and mesh generation module in NEMoSys
NucMesh is a parameterized geometry and mesh generator for nuclear reactors developed within the Nuclear Energy Modeling System NEMoSys at Illinois Rocstar. NEMoSys is a platform developed for mesh generation, adaptive refinement, and solution verification. NucMesh is implemented to be generalized and extensible with a robust computer aided design engine and multiple mesh generation algorithms for unstructured triangular, quad-dominant, and structured quadrilateral meshing. In this paper, we present the geometric and meshing features of NucMesh. Geometrically objects are constructed bottom-to-top and overlaps are addressed automatically. A sophisticated object tracking algorithm prevents data from being lost for segmented objects. We discuss the primitive objects of circle and polygons that constitute the module and show how they are used with example inputs. Arrays of primitives and arrays of arrays are utilized to build large assemblies of objects. The concept of saved objects is discussed to demonstrate how repetitive objects can be reused easily and augmented in place. Three dimensional meshes can be obtained through mesh extrusion where all materials and side sets are extended to three dimensions. We show that side sets can be defined nearly anywhere within the geometry and can then be applied to the mesh. Finally, example reactor meshes are demonstrated for the Idaho National Laboratory Advanced Test Reactor and Los Alamos National Lab Empire reactor, both of which use control drums that NucMesh handles easily. (authors)
22 GENERAL STUDIES OF NUCLEAR REACTORS↗
A Virtual Element Method for the Full MHD system
In this manuscript we present a novel discretization for the incompressible MHD system. Our approach follows the framework of the Virtual Element Method and offers two main advantages. The method can be implemented in unstructured meshes making it highly versatile and capable of handling a wide array of problems involving interfaces, free-boundaries or adaptive refinements on the mesh. The second advantage involves the divergence of the magnetic field, our approach guarantees that it remains solenoidal. We include a theoretical proof of the condition on the magnetic field as well as energy estimates and a well-posedness study. The latter sheds light as to the stability properties of the method.
75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗