DiscretizationNet: A Machine-Learning based solver for Navier-Stokes Equations using Finite Volume Discretization

Abstract

Over the last few decades, existing Partial Differential Equation (PDE) solvers have demonstrated a tremendous success in solving complex, non-linear PDEs. Although accurate, these PDE solvers are computationally costly. With the advances in Machine Learning (ML) technologies, there has been a significant increase in the research of using ML to solve PDEs. The goal of this work is to develop an ML-based PDE solver, that couples important characteristics of existing PDE solvers with ML technologies. The two solver characteristics that have been adopted in this work are: 1) the use of discretization-based schemes to approximate spatio-temporal partial derivatives and 2) the use of iterative algorithms to solve linearized PDEs in their discrete form. In the presence of highly non-linear, coupled PDE solutions, these strategies can be very important in achieving good accuracy, better stability and faster convergence. Our ML-solver, DiscretizationNet, employs a generative CNN-based encoderdecoder model with PDE variables as both input and output features. During training, the discretization schemes are implemented inside the computational graph to enable faster GPU computation of PDE residuals, which are used to update network weights that result into converged solutions. A novel iterative capability is implemented during the network training to improve the stability and convergence of the ML-solver. The ML-Solver is demonstrated to solve the steady, incompressible Navier-Stokes equations in 3-D for several cases such as, lid-driven cavity, flow past a cylinder and conjugate heat transfer.

An unsupervised learning approach to solving heat equations on chip based on Auto Encoder and Image Gradient

“…A "Deep Galerkin Method (DGM)" based deep neural network is trained to solve high-dimensional PDEs in [13] and is able to solve PDEs up to 200 dimensions. Besides, there has been researches using Convolutional Neural Network (CNN) based structure as well, for example, a U-Net architecture is used in [14] and a CNN-based encoder-decoder model is used in [15] for constructing numerical solvers for PDEs. With recent developments in one of the most useful but perhaps underused techniques in scientific computing: automatic differentiation (AD) [16], the idea of defining partial differential equations utilizing AD, imposing physics through neural network losses together with easy access to backpropagation revolutionized PDE solving by using Physics-Informed Neural Networks (PINN).…”

Physics-consistent deep learning for structural topology optimization

“…Although deep learning has many diverse applications and has demonstrated extraordinary results in several real-world scenarios, our focus in this paper is the recent application of deep learning to learn a system's underlying physics. There has been an increased interest in learning physical phenomena with neural networks in order to reduce the data requirement and achieve better performance with very little or no data 4,9,10,12,15,[18][19][20][21]23,27,31,32 . One method by which this can be achieved is by modifying the loss function Using the DLTO framework, we predict the optimal density of the geometry without any requirement of iterative finite element evaluations.…”

DiffusionNet: Accelerating the solution of Time-Dependent partial differential equations using deep learning

“…These reasons, among others, are a prime motivation for utilizing machine learning techniques. The research for solving partial differential equations with machine learning can be divided into data-driven and data-free; in data-driven methods, the data is generated experimentally or numerically and then fed to the model to learn the underlying relations [11].In datafree methods, the neural network yields solution by utilizing loss formulation to constrain the partial governing differential. In this paper, we focus on the data-driven methods of solutions to solve three partial differential equations: (1) Transient heat conduction, (2) Inviscid Burgers' equation, (3) Steady-state heat conduction.…”