Learning Mesh-Based Simulation with Graph Networks

Mesh-based simulations are central to modeling complex physical systems in many disciplines across science and engineering. Mesh representations support powerful numerical integration methods and their resolution can be adapted to strike favorable trade-offs between accuracy and efficiency. However, high-dimensional scientific simulations are very expensive to run, and solvers and parameters must often be tuned individually to each system studied. Here we introduce MeshGraphNets, a framework for learning mesh-based simulations using graph neural networks. Our model can be trained to pass messages on a mesh graph and to adapt the mesh discretization during forward simulation. Our results show it can accurately predict the dynamics of a wide range of physical systems, including aerodynamics, structural mechanics, and cloth. The model's adaptivity supports learning resolution-independent dynamics and can scale to more complex state spaces at test time. Our method is also highly efficient, running 1-2 orders of magnitude faster than the simulation on which it is trained. Our approach broadens the range of problems on which neural network simulators can operate and promises to improve the efficiency of complex, scientific modeling tasks.

[1]  Comsol , 2021, IEEE Microwave Magazine.

[2]  J. Zico Kolter,et al.  Combining Differentiable PDE Solvers and Graph Neural Networks for Fluid Flow Prediction , 2020, ICML.

[3]  Sara Hooker,et al.  The hardware lottery , 2020, Commun. ACM.

[4]  Vladlen Koltun,et al.  Lagrangian Fluid Simulation with Continuous Convolutions , 2020, ICLR.

[5]  Huamin Wang,et al.  NNWarp: Neural Network-Based Nonlinear Deformation , 2020, IEEE Transactions on Visualization and Computer Graphics.

[6]  Danilo Jimenez Rezende,et al.  Equivariant flow-based sampling for lattice gauge theory , 2020, Physical review letters.

[7]  Pushmeet Kohli,et al.  Unveiling the predictive power of static structure in glassy systems , 2020 .

[8]  S. M. Ali Eslami,et al.  PolyGen: An Autoregressive Generative Model of 3D Meshes , 2020, ICML.

[9]  Jure Leskovec,et al.  Learning to Simulate Complex Physics with Graph Networks , 2020, ICML.

[10]  Rui Wang,et al.  Towards Physics-informed Deep Learning for Turbulent Flow Prediction , 2019, KDD.

[11]  Derek Nowrouzezahrai,et al.  Subspace neural physics: fast data-driven interactive simulation , 2019, Symposium on Computer Animation.

[12]  Karthik Duraisamy,et al.  Prediction of aerodynamic flow fields using convolutional neural networks , 2019, Computational Mechanics.

[13]  M. S. Albergo,et al.  Flow-based generative models for Markov chain Monte Carlo in lattice field theory , 2019, Physical Review D.

[14]  Leslie Pack Kaelbling,et al.  Graph Element Networks: adaptive, structured computation and memory , 2019, ICML.

[15]  German Capuano,et al.  Smart finite elements: A novel machine learning application , 2019, Computer Methods in Applied Mechanics and Engineering.

[16]  Nils Thuerey,et al.  Deep Learning Methods for Reynolds-Averaged Navier–Stokes Simulations of Airfoil Flows , 2018, AIAA Journal.

[17]  Jiajun Wu,et al.  Learning Particle Dynamics for Manipulating Rigid Bodies, Deformable Objects, and Fluids , 2018, ICLR.

[18]  Raia Hadsell,et al.  Graph networks as learnable physics engines for inference and control , 2018, ICML.

[19]  Jessica B. Hamrick,et al.  Relational inductive biases, deep learning, and graph networks , 2018, ArXiv.

[20]  Donghyun You,et al.  Data-driven prediction of unsteady flow over a circular cylinder using deep learning , 2018, Journal of Fluid Mechanics.

[21]  Nils Thürey,et al.  Latent Space Physics: Towards Learning the Temporal Evolution of Fluid Flow , 2018, Comput. Graph. Forum.

[22]  Mathieu Aubry,et al.  A Papier-Mache Approach to Learning 3D Surface Generation , 2018, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[23]  Yao Zhang,et al.  Application of Convolutional Neural Network to Predict Airfoil Lift Coefficient , 2017, ArXiv.

[24]  Patrick Gallinari,et al.  Deep learning for physical processes: incorporating prior scientific knowledge , 2017, ICLR.

[25]  Xiangyu Hu,et al.  Liquid Splash Modeling with Neural Networks , 2017, Comput. Graph. Forum.

[26]  Pierre Vandergheynst,et al.  Geometric Deep Learning: Going beyond Euclidean data , 2016, IEEE Signal Process. Mag..

[27]  Max Welling,et al.  Semi-Supervised Classification with Graph Convolutional Networks , 2016, ICLR.

[28]  Wei Li,et al.  Convolutional Neural Networks for Steady Flow Approximation , 2016, KDD.

[29]  J. Alonso,et al.  SU2: An Open-Source Suite for Multiphysics Simulation and Design , 2016 .

[30]  Barbara Solenthaler,et al.  Data-driven fluid simulations using regression forests , 2015, ACM Trans. Graph..

[31]  F. V. Antunes,et al.  A review on 3D-FE adaptive remeshing techniques for crack growth modelling , 2015 .

[32]  James F. O'Brien,et al.  Adaptive tearing and cracking of thin sheets , 2014, ACM Trans. Graph..

[33]  James F. O'Brien,et al.  Folding and crumpling adaptive sheets , 2013, ACM Trans. Graph..

[34]  James F. O'Brien,et al.  Adaptive anisotropic remeshing for cloth simulation , 2012, ACM Trans. Graph..

[35]  L. Parker The conference paper , 2011 .

[36]  Christoph Schwarzbach,et al.  Three-dimensional adaptive higher order finite element simulation for geo-electromagnetics—a marine CSEM example , 2011 .

[37]  James F. O'Brien,et al.  Dynamic local remeshing for elastoplastic simulation , 2010, ACM Trans. Graph..

[38]  Jingtian Tang,et al.  3D direct current resistivity modeling with unstructured mesh by adaptive finite-element method , 2010 .

[39]  Hamouine Abdelmadjid,et al.  A state-of-the-art review of the X-FEM for computational fracture mechanics , 2009 .

[40]  David Pardo,et al.  A self-adaptive goal-oriented hp-finite element method with electromagnetic applications. Part II: Electrodynamics , 2007 .

[41]  N. Ramakrishnan,et al.  An analysis of springback in sheet metal bending using finite element method (FEM) , 2007 .

[42]  Hermann A. Maurer,et al.  New Results and New Trends in Computer Science , 1991, Lecture Notes in Computer Science.

[43]  Antoine Vacavant,et al.  Reconstructions of Noisy Digital Contours with Maximal Primitives Based on Multi-scale/Irregular Geometric Representation and Generalized Linear Programming , 2017, DGCI.

[44]  Ah Chung Tsoi,et al.  The Graph Neural Network Model , 2009, IEEE Transactions on Neural Networks.

[45]  S. Marburg,et al.  Computational acoustics of noise propagation in fluids : finite and boudary element methods , 2008 .

[46]  R. Ramamurti,et al.  Simulation of Flow About Flapping Airfoils Using Finite Element Incompressible Flow Solver , 2001 .

[47]  Paul S. Heckbert,et al.  A Pliant Method for Anisotropic Mesh Generation , 1996 .

[48]  Emo Welzl,et al.  Smallest enclosing disks (balls and ellipsoids) , 1991, New Results and New Trends in Computer Science.