.png){kind=icon}
返回
DeLISA: Deep learning based iteration scheme approximation for solving PDEs
.png){kind=icon}
PRE
AI
DOI：10.1016/j.jcp.2021.110884.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Solving the high dimensional partial differential equations (PDEs) with the classical numerical methods is a challenge task. As possessing the power of progressing high dimensional data, deep learning is naturally considered to solve PDEs. This paper proposes a deep learning framework based iteration scheme approximation, called DeLISA. First, we adopt the implicit multistep method and Runge-Kutta method for time iteration scheme. Then, such iteration scheme is approximated by a neural network. Due to integrating the physical information of governing equation into time iteration schemes and introducing time-dependent input, our method achieves the continuous time prediction without a mass of interior points. Here, the activation function with adaptive variable adjusts itself during the iteration process. Finally, we present numerical experiments results for some benchmark PDEs, including Burgers, Allen-Cahn, Schrodinger, carburizing and Black-Scholes equations, and verify that the proposed approach is superior to the state-of-the-art techniques on accuracy and flexibility. Moreover, the Frequency Principle is also illustrated by the changes of prediction at different iterations in this paper. (c) 2021 Elsevier Inc. All rights reserved.
Keyword:
Deep learning
PDE
Adaptive activation
Iteration scheme approximation
期刊
IF:
3.8
论文数:
1.6W
被引数:
7.4W
机构
引用论文
分享收藏Neural network algorithm based on Legendre improved extreme learning machine for solving elliptic partial differential equations.png)
SOFT COMPUTING
SOFT COMPUTINGIF2.5
PREAI
分享收藏Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations物理信息神经网络: 一种用于解决涉及非线性偏微分方程的正反问题的深度学习框架
OAAI
OAAI分享收藏Ab initio solution of the many-electron Schrodinger equation with deep neural networks多电子薛定谔方程的深度神经网络从头算解OAAI
OAAI分享收藏Cross-linkable Polymer Matrix for Enhanced Thermal Stability of Succinonitrile-based Polymer Electrolyte in Lithium Rechargeable Batteries可交联聚合物基质，用于增强锂可充电电池中基于丁二腈的聚合物电解质的热稳定性
OAAI
OAAI分享收藏
查看更多内容