Physics-Driven Learning of the Steady Navier-Stokes Equations using Deep Convolutional Neural Networks

DOI10.4208/cicp.OA-2021-0146zbMath1500.76071arXiv2106.09301OpenAlexW3167808783WikidataQ114911802 ScholiaQ114911802MaRDI QIDQ5042008

Nils Thürey, Xiangyu Hu, Hao Ma, Oskar J. Haidn, Yuxuan Zhang

Publication date: 18 October 2022

Published in: Communications in Computational Physics (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/2106.09301

zbMATH Keywords

incompressible Navier-Stokes equationslow Reynolds number flowfirst-order accuracyiterative stochastic gradient descent methodU-net architecture

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05) Navier-Stokes equations for incompressible viscous fluids (76D05) Basic methods in fluid mechanics (76M99)

Related Items (3)

Learning Invariant Representation of Multiscale Hyperelastic Constitutive Law from Sparse Experimental Data ⋮ Bounded nonlinear forecasts of partially observed geophysical systems with physics-constrained deep learning ⋮ Less Emphasis on Hard Regions: Curriculum Learning of PINNs for Singularly Perturbed Convection-Diffusion-Reaction Problems

Uses Software

• Gmsh
• PyTorch
• ImageNet
• Adam
• DeepXDE
• U-Net
• PDE-Net
• AlexNet

Cites Work

• Unnamed Item
• Unnamed Item
• Numerical simulation of laminar flow past a circular cylinder
• Non-intrusive reduced order modeling of nonlinear problems using neural networks
• Hidden physics models: machine learning of nonlinear partial differential equations
• Model reduction for compressible flows using POD and Galerkin projection
• Surrogate modeling for fluid flows based on physics-constrained deep learning without simulation data
• Non-intrusive reduced order modeling of unsteady flows using artificial neural networks with application to a combustion problem
• Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data
• PDE-Net 2.0: learning PDEs from data with a numeric-symbolic hybrid deep network
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• A paradigm for data-driven predictive modeling using field inversion and machine learning
• Gmsh: A 3-D finite element mesh generator with built-in pre- and post-processing facilities
• Reynolds averaged turbulence modelling using deep neural networks with embedded invariance
• Turbulence Modeling in the Age of Data

This page was built for publication: Physics-Driven Learning of the Steady Navier-Stokes Equations using Deep Convolutional Neural Networks