Subregion–identified physics-informed neural networks for the Navier–Stokes/Darcy interface model

i

This article has been updated. A Correction to this article was published on .

i

The Original Article was published on

i

The Original Article was published on

i

This article has been updated

i

This article was retracted on

i

This article has been updated

Abstract

Physics-informed neural networks (PINNs), built upon general neural networks, integrate physical information into the loss function, achieving remarkable results in solving partial differential equations (PDEs). Meanwhile, PINNs face challenges in solving complex interface problems, such as the Navier–Stokes/Darcy interface model considered in this paper. To address these challenges, we extend the general PINNs to propose subregion–identified physics-informed neural networks (SI-PINNs), which consist of three different neural networks for different subregions (Navier–Stokes, Darcy, and interface) with separated datasets. By identifying and classifying the regional features of the neural networks and datasets, the complexity of the physical information that SI-PINNs need to process is effectively split, while the networks also take all physical information into account during the training. Additionally, this splitting feature substantially reduces the training costs while enhancing or maintaining the accuracy of the general PINNs for the target model. Numerical experiments are performed to validate the effectiveness of the SI-PINN method.

Keywords

SI-PINNs / subregion–identification / physics-informed neural networks / Navier–Stokes/Darcy interface model / Beavers–Joseph interface condition

CHANGE HISTORY

A Correction to this paper has been published:
https://doi.org

The Retraction Note was updated to reflect reception of statements of disagreement from all authors.

This article has been retracted.Please see the Retraction Notice for more detail:
https://doi.org

References

[1] T. Arbogast, D. S. Brunson, A computational method for approximating a Darcy-Stokes system governing a vuggy porous medium, Comput. Geosci., 11 (2007), 207-218. https://doi.org/10.1007/s10596-007-9043-0

[2] T. Arbogast, H. L. Lehr, Homogenization of a Darcy-Stokes system modeling vuggy porous media, Comput. Geosci., 10 (2006), 291-302. https://doi.org/10.1007/s10596-006-9024-8

[3] J. Hou, D. Hu, X. M. He, C. Qiu, Modeling and a Robin-type decoupled finite element method for dual-porosity-Navier-Stokes system with application to flows around multistage fractured horizontal wellbore, Comput. Methods Appl. Mech. Eng., 388 (2022), 114248. https://doi.org/10.1016/j.cma.2021.114248

[4] J. Hou, M. Qiu, X. M. He, C. Guo, M. Wei, B. Bai, A dual-porosity-Stokes model and finite element method for coupling dual-porosity flow and free flow, SIAM J. Sci. Comput., 38 (2016), B710-B739. https://doi.org/10.1137/15M1044072

[5] Y. Zhang, C. Zhou, C. Qu, M. Wei, X. M. He, B. Bai, Fabrication and verification of a glass-silicon-glass micro-nanofluidic model for investigating multi-phase flow in unconventional dual-porosity porous media, Lab Chip, 19 (2019), 4071-4082. https://doi.org/10.1039/C9LC00847K

[6] Y. Gao, D. Han, X. M. He, U. Rüde, Unconditionally stable numerical methods for Cahn-Hilliard-Navier-Stokes-Darcy system with different densities and viscosities, J. Comput. Phys., 454 (2022), 110968. https://doi.org/10.1016/j.jcp.2022.110968

[7] Y. Gao, X. M. He, L. Mei, X. Yang, Decoupled, linear, and energy stable finite element method for the Cahn-Hilliard-Navier-Stokes-Darcy phase field model, SIAM J. Sci. Comput., 40 (2018), B110-B137. https://doi.org/10.1137/16M1100885

[8] M. Discacciati, R. Oyarzúa, A conforming mixed finite element method for the Navier-Stokes/Darcy coupled problem, Numer. Math., 135 (2017), 571-606. https://doi.org/10.1007/s00211-016-0811-4

[9] W. J. Layton, F. Schieweck, I. Yotov, Coupling fluid flow with porous media flow, SIAM J. Numer. Anal., 40 (2002), 2195-2218. https://doi.org/10.1137/S0036142901392766

[10] M. Discacciati, E. Miglio, A. Quarteroni, Mathematical and numerical models for coupling surface and groundwater flows, Appl. Numer. Math., 43 (2002), 57-74. https://doi.org/10.1016/S0168-9274(02)00125-3

[11] D. Vassilev, C. Wang, I. Yotov, Domain decomposition for coupled Stokes and Darcy flows, Comput. Methods Appl. Mech. Eng., 268 (2014), 264-283. https://doi.org/10.1016/j.cma.2013.09.009

[12] M. Discacciati, A. Quarteroni, A. Valli, Robin-Robin domain decomposition methods for the Stokes-Darcy coupling, SIAM J. Numer. Anal., 45 (2007), 1246-1268. https://doi.org/10.1137/06065091X

[13] M. Discacciati, P. Gervasio, A. Giacomini, A. Quarteroni, The interface control domain decomposition method for Stokes-Darcy coupling, SIAM J. Numer. Anal., 54 (2016), 1039-1068. https://doi.org/10.1137/15M101854X

[14] M. Gunzburger, X. M. He, B. Li, On Stokes-Ritz projection and multistep backward differentiation schemes in decoupling the Stokes-Darcy model, SIAM J. Numer. Anal., 56 (2018), 397-427. https://doi.org/10.1137/16M1099601

[15] Y. Cao, M. Gunzburger, X. M. He, X. Wang, Parallel, non-iterative, multi-physics domain decomposition methods for time-dependent Stokes-Darcy systems, Math. Comput., 83 (2014), 1617-1644. https://doi.org/10.1090/S0025-5718-2014-02779-8

[16] Y. Liu, Y. Boubendir, X. M. He, Y. He, New optimized Robin-Robin domain decomposition methods using Krylov solvers for the Stokes-Darcy system, SIAM J. Sci. Comput., 44 (2022), B1068-B1095. https://doi.org/10.1137/21M1417223

[17] P. Chidyagwai, B. Riviére, On the solution of the coupled Navier-Stokes and Darcy equations, Comput. Methods Appl. Mech. Eng., 198 (2009), 3806-3820. https://doi.org/10.1016/j.cma.2009.08.012

[18] B. Riviére, I. Yotov, Locally conservative coupling of Stokes and Darcy flows, SIAM J. Numer. Anal., 42 (2005), 1959-1977. https://doi.org/10.1137/S0036142903427640

[19] G. Kanschat, B. Riviere, A strongly conservative finite element method for the coupling of Stokes and Darcy flow, J. Comput. Phys., 229 (2010), 5933-5943. https://doi.org/10.1016/j.jcp.2010.04.021

[20] J. Camaño, G. N. Gatica, R. Oyarzúa, R. Ruiz-Baier, P. Venegas, New fully-mixed finite element methods for the Stokes-Darcy coupling, Comput. Methods Appl. Mech. Eng., 295 (2015), 362-395. https://doi.org/10.1016/j.cma.2015.07.007

[21] K. Lipnikov, D. Vassilev, I. Yotov, Discontinuous Galerkin and mimetic finite difference methods for coupled Stokes-Darcy flows on polygonal and polyhedral grids, Numer. Math., 126 (2014), 321-360. https://doi.org/10.1007/s00211-013-0563-3

[22] N. Jiang, H. Yang, Highly efficient ensemble algorithms for computing the Stokes-Darcy equations, Comput. Methods Appl. Mech. Eng., 418 (2024), 116562. https://doi.org/10.1016/j.cma.2023.116562

[23] Y. Boubendir, S. Tlupova, Domain decomposition methods for solving Stokes-Darcy problems with boundary integrals, SIAM J. Sci. Comput., 35 (2013), B82-B106. https://doi.org/10.1137/110838376

[24] Y. Boubendir, S. Tlupova, Stokes-Darcy boundary integral solutions using preconditioners, J. Comput. Phys., 228 (2009), 8627-8641. https://doi.org/10.1016/j.jcp.2009.08.014

[25] M. Mu, J. Xu, A two-grid method of a mixed Stokes-Darcy model for coupling fluid flow with porous media flow, SIAM J. Numer. Anal., 45 (2007), 1801-1813. https://doi.org/10.1137/050637820

[26] S. Münzenmaier, G. Starke, First-order system least squares for coupled Stokes-Darcy flow, SIAM J. Numer. Anal., 49 (2011), 387-404. https://doi.org/10.1137/100805108

[27] S. Zeng, Z. Xie, X. Yang, J. Wang, Fully discrete, decoupled and energy-stable Fourier-spectral numerical scheme for the nonlocal Cahn-Hilliard equation coupled with Navier-Stokes/Darcy flow regime of two-phase incompressible flows, Comput. Methods Appl. Mech. Eng., 415 (2023), 116289. https://doi.org/10.1016/j.cma.2023.116289

[28] W. Chen, M. Gunzburger, D. Sun, X. Wang, An efficient and long-time accurate third-order algorithm for the Stokes-Darcy system, Numer. Math., 134 (2016), 857-879. https://doi.org/10.1007/s00211-015-0789-3

[29] V. Girault, D. Vassilev, I. Yotov, Mortar multiscale finite element methods for Stokes-Darcy flows, Numer. Math., 127 (2014), 93-165. https://doi.org/10.1007/s00211-013-0583-z

[30] X. Li, W. Gong, X. M. He, T. Lin, Variational data assimilation and its decoupled iterative numerical algorithms for Stokes-Darcy model, SIAM J. Sci. Comput., 46 (2024), S142-S175. https://doi.org/10.1137/22M1492994

[31] C. Mingard, H. Rees, G. V. Pérez, A. A. Louis, Deep neural networks have an inbuilt Occam’s razor, Nat. Commun., 16 (2025), 220. https://doi.org/10.1038/s41467-024-54813-x

[32] Q. Guan, How can deep neural networks fail even with global optima?, Int. J. Numer. Anal. Model., 21 (2024), 674-696. https://doi.org/10.4208/ijnam2024-1027

[33] L. Storm, H. Linander, J. Bec, K. Gustavsson, B. Mehlig, Finite-time Lyapunov exponents of deep neural networks, Phys. Rev. Lett., 132 (2024), 057301. https://doi.org/10.1103/PhysRevLett.132.057301

[34] J. Cheng, Q. Li, T. Lin, Z. Shen, Interpolation, approximation, and controllability of deep neural networks, SIAM J. Control Optim., 63 (2025), 625-649. https://doi.org/10.1137/23M1599744

[35] Z. Zhang, F. Bao, G. Zhang, Improving the expressive power of deep neural networks through integral activation transform, Int. J. Numer. Anal. Model., 21 (2024), 739-763. https://doi.org/10.4208/ijnam2024-1030

[36] D. Frerichs-Mihov, L. Henning, V. John, Using deep neural networks for detecting spurious oscillations in discontinuous Galerkin solutions of convection-dominated convection-diffusion equations, J. Sci. Comput., 97 (2023), 36. https://doi.org/10.1007/s10915-023-02335-x

[37] S. Ji, S. Peng, Y. Peng, X. Zhang, A novel control method for solving high-dimensional Hamiltonian systems through deep neural networks, SIAM J. Sci. Comput., 47 (2025), C873-C898. https://doi.org/10.1137/24M1647084

[38] B. Chudomelka, Y. Hong, J. Morgan, H. Kim, J. Park, Deep neural network for solving differential equations motivated by Legendre-Galerkin approximation, Int. J. Numer. Anal. Model., 21 (2024), 652-673. https://doi.org/10.4208/ijnam2024-1026

[39] J. Y. Lee, J. W. Jang, H. J. Wang, The model reduction of the Vlasov-Poisson-Fokker-Planck system to the Poisson-Nernst-Planck system via the deep neural network approach, ESAIM: M2AN, 55 (2021), 1803-1846. https://doi.org/10.1051/m2an/2021038

[40] C. Fan, M. A. Ali, Z. Zhang, Decoupling numerical method based on deep neural network for nonlinear degenerate interface problems, Comput. Phys. Commun., 303 (2024), 109275. https://doi.org/10.1016/j.cpc.2024.109275

[41] M. Raissi, P. Perdikaris, G. E. Karniadakis, Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, J. Comput. Phys., 378 (2019), 686-707. https://doi.org/10.1016/j.jcp.2018.10.045

[42] S. Wang, Y. Teng, P. Perdikaris, Understanding and mitigating gradient flow pathologies in physics-informed neural networks, SIAM J. Sci. Comput., 43 (2021), A3055-A3081. https://doi.org/10.1137/20M1318043

[43] W. Cao, W. Zhang, An analysis and solution of ill-conditioning in physics-informed neural networks, J. Comput. Phys., 520 (2025), 113494. https://doi.org/10.1016/j.jcp.2024.113494

[44] X. Jin, S. Cai, H. Li, G. E. Karniadakis, NSFnets (Navier-Stokes flow nets): Physics-informed neural networks for the incompressible Navier-Stokes equations, J. Comput. Phys., 426 (2021), 109951. https://doi.org/10.1016/j.jcp.2020.109951

[45] A. F. Psaros, K. Kawaguchi, G. E. Karniadakis, Meta-learning PINN loss functions, J. Comput. Phys., 458 (2022), 111121. https://doi.org/10.1016/j.jcp.2022.111121

[46] Q. Zhang, C. Qiu, J. Hou, W. Yan, Advanced physics-informed neural networks for numerical approximation of the coupled Schrödinger-KdV equation, Commun. Nonlinear Sci. Numer. Simul., 138 (2024), 108229. https://doi.org/10.1016/j.cnsns.2024.108229

[47] Z. Gao, Z. Fu, M. Wen, Y. Guo, Y. Zhang, Physical informed neural network for thermo-hydral analysis of fire-loaded concrete, Eng. Anal. Boundary Elem., 158 (2024), 252-261. https://doi.org/10.1016/j.enganabound.2023.10.027

[48] S. Xu, Y. Dai, C. Yan, Z. Sun, R. Huang, D. Guo, et al., On the preprocessing of physics-informed neural networks: How to better utilize data in fluid mechanics, J. Comput. Phys., 528 (2025), 113837. https://doi.org/10.1016/j.jcp.2025.113837

[49] Z. Fu, W. Xu, S. Liu, Physics-informed kernel function neural networks for solving partial differential equations, Neural Networks, 172 (2024), 106098. https://doi.org/10.1016/j.neunet.2024.106098

[50] A. D. Jagtap, E. Kharazmi, G. E. Karniadakis, Conservative physics-informed neural networks on discrete domains for conservation laws: Applications to forward and inverse problems, Comput. Methods Appl. Mech. Eng., 365 (2020), 113028. https://doi.org/10.1016/j.cma.2020.113028

[51] A. D. Jagtap, G. E. Karniadakis, Extended physics-informed neural networks (XPINNs): A generalized space-time domain decomposition based deep learning framework for nonlinear partial differential equations, Commun. Comput. Phys., 28 (2020), 2002-2041. https://doi.org/10.4208/cicp.OA-2020-0164

[52] A. Kopaničáková, H. Kothari, G. E. Karniadakis, R. Krause, Enhancing training of physics-informed neural networks using domain decomposition-based preconditioning strategies, SIAM J. Sci. Comput., 46 (2024), S46-S67. https://doi.org/10.1137/23M1583375

[53] V. Dolean, A. Heinlein, S. Mishra, B. Moseley, Multilevel domain decomposition-based architectures for physics-informed neural networks, Comput. Methods Appl. Mech. Eng., 429 (2024), 117116. https://doi.org/10.1016/j.cma.2024.117116

[54] C. Si, M. Yan, Initialization-enhanced physics-informed neural network with domain decomposition (IDPINN), J. Comput. Phys., 530 (2025), 113914. https://doi.org/10.1016/j.jcp.2025.113914

[55] G. S. Beavers, D. D. Joseph, Boundary conditions at a naturally permeable wall, J. Fluid Mech., 30 (1967), 197-207. https://doi.org/10.1017/S0022112067001375

[56] Y. Cao, M. Gunzburger, F. Hua, X. Wang, Coupled Stokes-Darcy model with Beavers-Joseph interface boundary condition, Commun. Math. Sci., 8 (2010), 1-25. https://doi.org/10.4310/CMS.2010.v8.n1.a2

[57] Y. Cao, M. Gunzburger, X. Hu, F. Hua, X. Wang, W. Zhao, Finite element approximations for Stokes-Darcy flow with Beavers-Joseph interface conditions, SIAM J. Numer. Anal., 47 (2010), 4239-4256. https://doi.org/10.1137/080731542

[58] Y. Cao, M. Gunzburger, X. M. He, X. Wang, Robin-Robin domain decomposition methods for the steady-state Stokes-Darcy system with the Beaver-Joseph interface condition, Numer. Math., 117 (2011), 601-629. https://doi.org/10.1007/s00211-011-0361-8

[59] X. M. He, J. Li, Y. Lin, J. Ming, A domain decomposition method for the steady-state Navier-Stokes-Darcy model with Beavers-Joseph interface condition, SIAM J. Sci. Comput., 37 (2015), S264-S290. https://doi.org/10.1137/140965776

[60] K. Hornik, Approximation capabilities of multilayer feedforward networks, Neural Networks, 4 (1991), 251-257. https://doi.org/10.1016/0893-6080(91)90009-T

[61] D. Kingma, J. Ba, Adam: A method for stochastic optimization, preprint, arXiv:1412.6980.

[62] L. Lu, X. Meng, Z. Mao, G. E. Karniadakis, DeepXDE: A deep learning library for solving differential equations, SIAM Rev., 63 (2021), 208-228. https://doi.org/10.1137/19M1274067

[63] V. Nair, G. E. Hinton, Rectified linear units improve restricted Boltzmann machines, in Proceedings of the 27th International Conference on Machine Learning (ICML-10), Omnipress, (2010), 807-814.

[64] K. He, X. Zhang, S. Ren, J. Sun, Delving deep into rectifiers: surpassing human-level performance on imagenet classification, in Proceedings of the IEEE International Conference on Computer Vision, IEEE, (2015), 1026-1034.

[65] P. Ramachandran, B. Zoph, Q. V. Le, Searching for activation functions, preprint, arXiv:1710.05941.

[66] X. Glorot, Y. Bengio, Understanding the difficulty of training deep feedforward neural networks, in Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, PMLR, (2010), 249-256.

[67] L. Lu, Y. Shin, Y. Su, G. E. Karniadakis, Dying ReLU and initialization: Theory and numerical examples, Commun. Comput. Phys., 28 (2020), 1671-1706. https://doi.org/10.4208/cicp.OA-2020-0165

[68] M. Stein, Large sample properties of simulations using Latin hypercube sampling, Technometrics, 29 (1987), 143-151. https://doi.org/10.1080/00401706.1987.10488205