Year: 2022
Author: Chenguang Duan, Yuling Jiao, Yanming Lai, Dingwei Li, Xiliang Lu, Jerry Zhijian Yang
Communications in Computational Physics, Vol. 31 (2022), Iss. 4 : pp. 1020–1048
Abstract
Using deep neural networks to solve PDEs has attracted a lot of attentions recently. However, why the deep learning method works is falling far behind its empirical success. In this paper, we provide a rigorous numerical analysis on deep Ritz method (DRM) [47] for second order elliptic equations with Neumann boundary conditions. We establish the first nonasymptotic convergence rate in $H^1$ norm for DRM using deep networks with ${\rm ReLU}^2$ activation functions. In addition to providing a theoretical justification of DRM, our study also shed light on how to set the hyperparameter of depth and width to achieve the desired convergence rate in terms of number of training samples. Technically, we derive bound on the approximation error of deep ${\rm ReLU}^2$ network in $C^1$ norm and bound on the Rademacher complexity of the non-Lipschitz composition of gradient norm and ${\rm ReLU}^2$ network, both of which are of independent interest.
You do not have full access to this article.
Already a Subscriber? Sign in as an individual or via your institution
Journal Article Details
Publisher Name: Global Science Press
Language: English
DOI: https://doi.org/10.4208/cicp.OA-2021-0195
Communications in Computational Physics, Vol. 31 (2022), Iss. 4 : pp. 1020–1048
Published online: 2022-01
AMS Subject Headings: Global Science Press
Copyright: COPYRIGHT: © Global Science Press
Pages: 29
Keywords: Deep Ritz method deep neural networks convergence rate analysis.
Author Details
-
Solving Parametric Partial Differential Equations with Deep Rectified Quadratic Unit Neural Networks
Lei, Zhen | Shi, Lei | Zeng, ChenyuJournal of Scientific Computing, Vol. 93 (2022), Iss. 3
https://doi.org/10.1007/s10915-022-02015-2 [Citations: 5] -
Solving a Class of High-Order Elliptic PDEs Using Deep Neural Networks Based on Its Coupled Scheme
Li, Xi’an | Wu, Jinran | Zhang, Lei | Tai, XinMathematics, Vol. 10 (2022), Iss. 22 P.4186
https://doi.org/10.3390/math10224186 [Citations: 0] -
Deep Ritz Method for Elliptical Multiple Eigenvalue Problems
Ji, Xia | Jiao, Yuling | Lu, Xiliang | Song, Pengcheng | Wang, FengruJournal of Scientific Computing, Vol. 98 (2024), Iss. 2
https://doi.org/10.1007/s10915-023-02443-8 [Citations: 1] -
Deep convolutional Ritz method: parametric PDE surrogates without labeled data
Fuhg, J. N. | Karmarkar, A. | Kadeethum, T. | Yoon, H. | Bouklas, N.Applied Mathematics and Mechanics, Vol. 44 (2023), Iss. 7 P.1151
https://doi.org/10.1007/s10483-023-2992-6 [Citations: 12] -
Generalization of PINNs for elliptic interface problems
Jiang, Xuelian | Wang, Ziming | Bao, Wei | Xu, YingxiangApplied Mathematics Letters, Vol. 157 (2024), Iss. P.109175
https://doi.org/10.1016/j.aml.2024.109175 [Citations: 0] -
Solving PDEs on unknown manifolds with machine learning
Liang, Senwei | Jiang, Shixiao W. | Harlim, John | Yang, HaizhaoApplied and Computational Harmonic Analysis, Vol. 71 (2024), Iss. P.101652
https://doi.org/10.1016/j.acha.2024.101652 [Citations: 1] -
WAN Discretization of PDEs: Best Approximation, Stabilization, and Essential Boundary Conditions
Bertoluzza, Silvia | Burman, Erik | He, CuiyuSIAM Journal on Scientific Computing, Vol. 46 (2024), Iss. 6 P.C688
https://doi.org/10.1137/23M1588196 [Citations: 0] -
Imaging conductivity from current density magnitude using neural networks*
Jin, Bangti | Li, Xiyao | Lu, XiliangInverse Problems, Vol. 38 (2022), Iss. 7 P.075003
https://doi.org/10.1088/1361-6420/ac6d03 [Citations: 8] -
Solving Elliptic Problems with Singular Sources Using Singularity Splitting Deep Ritz Method
Hu, Tianhao | Jin, Bangti | Zhou, ZhiSIAM Journal on Scientific Computing, Vol. 45 (2023), Iss. 4 P.A2043
https://doi.org/10.1137/22M1520840 [Citations: 2] -
Current density impedance imaging with PINNs
Duan, Chenguang | Huang, Junjun | Jiao, Yuling | Lu, Xiliang | Yang, Jerry ZhijianJournal of Computational and Applied Mathematics, Vol. 452 (2024), Iss. P.116120
https://doi.org/10.1016/j.cam.2024.116120 [Citations: 0] -
Deep learning algorithms for solving differential equations: a survey
Kumar, Harender | Yadav, NehaJournal of Experimental & Theoretical Artificial Intelligence, Vol. (2023), Iss. P.1
https://doi.org/10.1080/0952813X.2023.2242356 [Citations: 0] -
Convergence analysis of deep Ritz method with over-parameterization
Ding, Zhao | Jiao, Yuling | Lu, Xiliang | Wu, Peiying | Yang, Jerry ZhijianNeural Networks, Vol. 184 (2025), Iss. P.107110
https://doi.org/10.1016/j.neunet.2024.107110 [Citations: 0] -
Stationary Density Estimation of Itô Diffusions Using Deep Learning
Gu, Yiqi | Harlim, John | Liang, Senwei | Yang, HaizhaoSIAM Journal on Numerical Analysis, Vol. 61 (2023), Iss. 1 P.45
https://doi.org/10.1137/21M1445363 [Citations: 2] -
О решении дифференциальных уравнений с помощью нейронных сетей: целевые функционалы и верификация результатов
Muzalevskii, Aleksey Vitalyevich | Repin, Sergey IgorevichМатематическое моделирование, Vol. 36 (2024), Iss. 6 P.179
https://doi.org/10.20948/mm-2024-06-12 [Citations: 0] -
ERROR ESTIMATES OF RESIDUAL MINIMIZATION USING NEURAL NETWORKS FOR LINEAR PDES
Shin, Yeonjong | Zhang, Zhongqiang | Karniadakis, George EmJournal of Machine Learning for Modeling and Computing, Vol. 4 (2023), Iss. 4 P.73
https://doi.org/10.1615/JMachLearnModelComput.2023050411 [Citations: 16]