BI-GreenNet: Learning Green’s Functions by Boundary Integral Network

Abstract

Green’s function plays a significant role in both theoretical analysis and numerical computing of partial differential equations (PDEs). However, in most cases, Green’s function is difficult to compute. The troubles arise in the following threefold. Firstly, compared with the original PDE, the dimension of Green’s function is doubled, making it impossible to be handled by traditional mesh-based methods. Secondly, Green’s function usually contains singularities which increase in the difficulty to get a good approximation. Lastly, the computational domain may be very complex or even unbounded. To override these problems, we develop a new framework for computing Green’s function leveraging the fundamental solution, boundary integral method and neural networks with reasonably high accuracy in this paper. We focus on Green’s function of Poisson and Helmholtz equations in bounded domains, unbounded domains. We also consider Poisson equation and Helmholtz equations in domains with interfaces. Extensive numerical experiments illustrate the efficiency and accuracy of our method for solving Green’s function. In addition, Green’s function provides the operator from the source term and boundary condition to the PDE solution. We apply Green’s function to solve PDEs with different sources, and obtain reasonably high-precision solutions, which shows the good generalization ability of our method. However, the requirements for explicit fundamental solutions to remove the singularity of Green’s function hinder the application of our method in more complex PDEs, such as variable coefficient equations, which will be investigated in our future work.

Keywords

Green’s function / Partial differential equation / Boundary integral / Neural network

References

[1] Alpert BK. Hybrid gauss-trapezoidal quadrature rules. SIAM J. Sci. Comput.. 1999, 20 5 1551-1584

[2] Amini S, Kirkup SM. Solution of Helmholtz equation in the exterior domain by elementary boundary integral methods. J. Comput. Phys.. 1995, 118 2 208-221

[3] Baydin, A.G., Pearlmutter, B.A., Radul, A.A., Siskind, J.M.: Automatic differentiation in machine learning: a survey. J. Mach. Learn. Res. 18 (2018)

[4] Bayliss A, Goldstein CI, Turkel E. The numerical solution of the Helmholtz equation for wave propagation problems in underwater acoustics. Comput. Math. Appl.. 1985, 11 7–8 655-665

[5] Berg J, Nyström K. A unified deep artificial neural network approach to partial differential equations in complex geometries. Neurocomputing. 2018, 317 28-41

[6] Cai, W., Xu, Z.-Q.J.: Multi-scale deep neural networks for solving high dimensional PDEs. Preprint at https://arxiv.org/abs/1910.11710 (2019)

[7] Chai Y, Gong Z, Li W, Li T, Zhang Q. A smoothed finite element method for exterior Helmholtz equation in two dimensions. Eng. Anal. Bound. Elem.. 2017, 84 237-252

[8] Chen Y, Lu L, Karniadakis GE, Dal Negro L. Physics-informed neural networks for inverse problems in nano-optics and metamaterials. Opt. Express. 2020, 28 8 11618-11633

[9] de La Bourdonnaye A. Some formulations coupling finite element and integral equation methods for Helmholtz equation and electromagnetism. Numer. Math.. 1995, 69 3 257-268

[10] Devlin, J., Chang, M.-W., Lee, K., Toutanova, K.: Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805 (2018)

[11] Dissanayake M, Phan-Thien N. Neural-network-based approximations for solving partial differential equations. Commun. Num. Methods Eng.. 1994, 10 3 195-201

[12] Duffy DG. Green’s Functions with Applications. 2015 New York: Chapman and Hall/CRC

[13] Economou EN. Green’s Functions in Quantum Physics. 2006 Berlin, Heidelberg: Springer

[14] Evans LC. Partial Differential Equations. 2010 Providence, R.I.: American Mathematical Society

[15] Friedman A, Kinderlehrer D. A one phase Stefan problem. Indiana Univ. Math. J.. 1975, 24 11 1005-1035

[16] Gin CR, Shea DE, Brunton SL, Kutz JN. DeepGreen: deep learning of Green’s functions for nonlinear boundary value problems. Sci. Rep.. 2021, 11 1 1-14

[17] Goodfellow I, Bengio Y, Courville A, Bengio Y. Deep Learning. 2016 Cambridge, MA: MIT Press

[18] Greenberg MD. Applications of Green’s Functions in Science and Engineering. 2015 Mineola, NY: Courier Dover Publications

[19] Hancock, M.J.: Method of Green’s Functions. Lecture notes (2006)

[20] Hansbo A, Hansbo P. An unfitted finite element method, based on Nitsche’s method, for elliptic interface problems. Comput. Methods Appl. Mech. Eng.. 2002, 191 47–48 5537-5552

[21] Hornik K, Stinchcombe M, White H. Multilayer feedforward networks are universal approximators. Neural Netw.. 1989, 2 5 359-366

[22] Hsiao GC, Wendland WL. Boundary Integral Equations. 2008 Berlin, Heidelberg: Springer

[23] Jackson, J.D.: Classical Electrodynamics. American Association of Physics Teachers (1999)

[24] Kapur S, Rokhlin V. High-order corrected trapezoidal quadrature rules for singular functions. SIAM J. Numer. Anal.. 1997, 34 4 1331-1356

[25] Kellogg OD. Foundations of Potential Theory. 1967 Berlin, Heidelberg: Springer

[26] Krizhevsky, A., Sutskever, I., Hinton, G.E.: Imagenet classification with deep convolutional neural networks. Adv. Neural Inf. Process. Syst. 25 (2012)

[27] Kukla S, Siedlecka U, Zamorska I. Green’s functions for interior and exterior Helmholtz problems. Sci. Res. Inst. Math. Comput. Sci.. 2012, 11 1 53-62

[28] Lagaris IE, Likas A, Fotiadis DI. Artificial neural networks for solving ordinary and partial differential equations. IEEE Trans. Neural Netw.. 1998, 9 5 987-1000

[29] Lee L, LeVeque RJ. An immersed interface method for incompressible Navier-Stokes equations. SIAM J. Sci. Comput.. 2003, 25 3 832-856

[30] Li, Z., Kovachki, N., Azizzadenesheli, K., Liu, B., Bhattacharya, K., Stuart, A., Anandkumar, A.: Fourier neural operator for parametric partial differential equations. Preprint at https://arxiv.org/abs/2010.08895 (2020)

[31] Lin, G., Hu, P., Chen, F., Chen, X., Chen, J., Wang, J., Shi, Z.: BINet: learning to solve partial differential equations with boundary integral networks. Preprint at https://arxiv.org/abs/2110.00352 (2021)

[32] Lu L, Jin P, Pang G, Zhang Z, Karniadakis GE. Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators. Nat. Mach. Intell.. 2021, 3 3 218-229

[33] Lyu, L., Zhang, Z., Chen, M., Chen, J.: MIM: a deep mixed residual method for solving high-order partial differential equations. J. Comput. Phys. 110930 (2022)

[34] Marques AN, Nave J-C, Rosales RR. A correction function method for Poisson problems with interface jump conditions. J. Comput. Phys.. 2011, 230 20 7567-7597

[35] Melnikov YA. Some applications of the Greens’ function method in mechanics. Int. J. Solids Struct.. 1977, 13 11 1045-1058

[36] Raissi M, Perdikaris P, Karniadakis GE. Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. J. Comput. Phys.. 2019, 378 686-707

[37] Sirignano J, Spiliopoulos K. DGM: a deep learning algorithm for solving partial differential equations. J. Comput. Phys.. 2018, 375 1339-1364

[38] Wapenaar K, Fokkema J. Green’s function representations for seismic interferometry. Geophysics. 2006, 71 4 33-46

[39] Weinan E, Han J, Jentzen A. Deep learning-based numerical methods for high-dimensional parabolic partial differential equations and backward stochastic differential equations. Commun. Math. Stat.. 2017, 5 4 349-380

[40] Weinan E, Yu B. The deep Ritz method: a deep learning-based numerical algorithm for solving variational problems. Comm. Math. Stat.. 2018, 6 1 1-12

[41] Zhang, L., Luo, T., Zhang, Y., Xu, Z.-Q.J., Ma, Z.: MOD-Net: a machine learning approach via model-operator-data network for solving PDEs. Preprint at https://arxiv.org/abs/2107.03673 (2021)

[42] Zhao S. High order matched interface and boundary methods for the Helmholtz equation in media with arbitrarily curved interfaces. J. Comput. Phys.. 2010, 229 9 3155-3170

Funding

National Natural Science Foundation of China(12071244)