Rethinking materials simulations: Blending direct numerical simulations with neural operators (2023-12-08T00:00:00.000000Z)

TL;DR

This methodology is based on the integration of a community numerical solver with a U-Net neural operator, enhanced by a temporal-conditioning mechanism to enable accurate extrapolation and efficient time-to-solution predictions of the dynamics.

Abstract

Materials simulations based on direct numerical solvers are accurate but computationally expensive for predicting materials evolution across length- and time-scales, due to the complexity of the underlying evolution equations, the nature of multiscale spatiotemporal interactions, and the need to reach long-time integration. We develop a method that blends direct numerical solvers with neural operators to accelerate such simulations. This methodology is based on the integration of a community numerical solver with a U-Net neural operator, enhanced by a temporal-conditioning mechanism to enable accurate extrapolation and efficient time-to-solution predictions of the dynamics. We demonstrate the effectiveness of this hybrid framework on simulations of microstructure evolution via the phase-field method. Such simulations exhibit high spatial gradients and the co-evolution of different material phases with simultaneous slow and fast materials dynamics. We establish accurate extrapolation of the coupled solver with large speed-up compared to DNS depending on the hybrid strategy utilized. This methodology is generalizable to a broad range of materials simulations, from solid mechanics to fluid dynamics, geophysics, climate, and more.

Authors

G. Karniadakis

6 papers

Vivek Oommen

1 papers

K. Shukla

1 papers

TL;DR

Abstract

Authors

References89 items

Laplace neural operator for solving differential equations

Residual-based attention in physics-informed neural networks

Deep neural operators as accurate surrogates for shape optimization

Learning the intrinsic dynamics of spatio-temporal processes through Latent Dynamics Networks

A deep material network approach for predicting the thermomechanical response of composites

Trade-offs in the latent representation of microstructure evolution

An AI framework integrating physics-informed neural network with predictive control for energy-efficient food production in the built environment

On the training and generalization of deep operator networks

Artificial Intelligence in Predicting Mechanical Properties of Composite Materials

Prediction of Diblock Copolymer Morphology via Machine Learning

MultiSOM: Multi-layer Self Organizing Maps for local structure identification in crystalline structures

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

Real-time Inference and Extrapolation via a Diffusion-inspired Temporal Transformer Operator (DiTTO)

Machine Learning Surrogate Model for Acceleration of Ferroelectric Phase-Field Modeling

Novel DeepONet architecture to predict stresses in elastoplastic structures with variable complex geometries and loads

Understanding and design of metallic alloys guided by phase-field simulations

A Framework Based on Symbolic Regression Coupled with eXtended Physics-Informed Neural Networks for Gray-Box Learning of Equations of Motion from Data

Emulating microstructural evolution during spinodal decomposition using a tensor decomposed convolutional and recurrent neural network

Latent Dynamics Networks (LDNets): learning the intrinsic dynamics of spatio-temporal processes

Morphology-Controlled Vapor-Phase Nanowire Growth with Ruddlesden–Popper Lead Bromide Perovskite

Spatiotemporal prediction of microstructure evolution with predictive recurrent neural network

LNO: Laplace Neural Operator for Solving Differential Equations

Electrochemically Induced Fracture in Llzo: How the Interplay between Flaw Density and Electrostatic Potential Affects Operability

Learning bias corrections for climate models using deep neural operators

Deep neural operators can serve as accurate surrogates for shape optimization: A case study for airfoils

Wavelet Neural Operator for solving parametric partial differential equations in computational mechanics problems

Predicting traction return current in electric railway systems through physics-informed neural networks

Adaptive physics-informed neural operator for coarse-grained non-equilibrium flows

An unsupervised latent/output physics-informed convolutional-LSTM network for solving partial differential equations using peridynamic differential operator

Mechanical neural networks: Architected materials that learn behaviors

Towards Multi-spatiotemporal-scale Generalized PDE Modeling

FourCastNet: Accelerating Global High-Resolution Weather Forecasting Using Adaptive Fourier Neural Operators

A comprehensive study of non-adaptive and residual-based adaptive sampling for physics-informed neural networks

Physics-Informed Deep Neural Operator Networks

Accelerating phase-field predictions via recurrent neural networks learning the microstructure evolution in latent space

Learning time-dependent deposition protocols to design thin films via genetic algorithms

Approximation rates of DeepONets for learning operators arising from advection-diffusion equations

Generic bounds on the approximation error for physics-informed (and) operator learning

Graph neural network modeling of grain-scale anisotropic elastic behavior using simulated and measured microscale data

U-NO: U-shaped Neural Operators

Learning two-phase microstructure evolution using neural operators and autoencoder architectures

Learning Deep Implicit Fourier Neural Operators (IFNOs) with Applications to Heterogeneous Material Modeling

Stability of immiscible nanocrystalline alloys in compositional and thermal fields

U-FNO - an enhanced Fourier neural operator based-deep learning model for multiphase flow

Simulating progressive intramural damage leading to aortic dissection using DeepONet: an operator–regression neural network

Finite-Volume Methods

Cluster-based network modeling—From snapshots to complex dynamical systems

Physics-informed machine learning

A Physics-Informed Neural Network for Quantifying the Microstructural Properties of Polycrystalline Nickel Using Ultrasound Data: A promising approach for solving inverse problems

Learning the solution operator of parametric partial differential equations with physics-informed DeepONets

Convergence rate of DeepONets for learning operators arising from advection-diffusion equations

Accelerating phase-field-based microstructure evolution predictions via surrogate models trained by machine learning methods

Operator learning for predicting multiscale bubble growth dynamics.

Multi-fidelity Bayesian Neural Networks: Algorithms and Applications

A data-driven surrogate model to rapidly predict microstructure morphology during physical vapor deposition

DeepM&Mnet for hypersonics: Predicting the coupled flow and finite-rate chemistry behind a normal shock using neural-network approximation of operators

Benchmark problems for the Mesoscale Multiphysics Phase Field Simulator (MEMPHIS).

Fourier Neural Operator for Parametric Partial Differential Equations

DeepM&Mnet: Inferring the electroconvection multiphysics fields based on operator approximation by neural networks

Microstructure morphology and concentration modulation of nanocomposite thin-films during simulated physical vapor deposition

Benchmark problems for the Mesoscale Multiphysics Phase Field Simulator (MEMPHIS)

Transformer Networks for Trajectory Forecasting

Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators

Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations

Group Normalization

Attention is All you Need

Gaussian Error Linear Units (GELUs)

A guide to convolution arithmetic for deep learning

U-Net: Convolutional Networks for Biomedical Image Segmentation

Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation

ImageNet classification with deep convolutional neural networks

Finding Structure with Randomness: Probabilistic Algorithms for Constructing Approximate Matrix Decompositions

Ultrathin Films of Single‐Walled Carbon Nanotubes for Electronics and Sensors: A Review of Fundamental and Applied Aspects

Spectral/hp Element Methods for Computational Fluid Dynamics

Phase-Field Models for Microstructure Evolution

Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its application to dynamical systems