×

Solving multiscale steady radiative transfer equation using neural networks with uniform stability. (English) Zbl 1492.65286

Summary: This paper concerns solving the steady radiative transfer equation with diffusive scaling, using the physics informed neural networks (PINNs). The idea of PINNs is to minimize a least-square loss function, that consists of the residual from the governing equation, the mismatch from the boundary conditions, and other physical constraints such as conservation. It is advantageous of being flexible and easy to execute, and brings the potential for high dimensional problems. Nevertheless, due the presence of small scales, the vanilla PINNs can be extremely unstable for solving multiscale steady transfer equations. In this paper, we propose a new formulation of the loss based on the macro-micro decomposition. We prove that, the new loss function is uniformly stable with respect to the small Knudsen number in the sense that the \(L^2\)-error of the neural network solution is uniformly controlled by the loss. When the boundary condition is an-isotropic, a boundary layer emerges in the diffusion limit and therefore brings an additional difficulty in training the neural network. To resolve this issue, we include a boundary layer corrector that carries over the sharp transition part of the solution and leaves the rest easy to be approximated. The effectiveness of the new methodology is demonstrated in extensive numerical examples.

MSC:

65M99 Numerical methods for partial differential equations, initial value and time-dependent initial-boundary value problems
45K05 Integro-partial differential equations
65R20 Numerical methods for integral equations
68T07 Artificial neural networks and deep learning
80A21 Radiative heat transfer
82C70 Transport processes in time-dependent statistical mechanics
PDFBibTeX XMLCite
Full Text: DOI arXiv

References:

[1] Bardos, C.; Santos, R.; Sentis, R., Diffusion approximation and computation of the critical size, Trans. Am. Math. Soc., 284, 617-649 (1984) · Zbl 0508.60067
[2] Bensoussan, A.; Lions, P-L; Papanicolaou, GC, Boundary layers and homogenizatlon of transport processes, Publ. Res. Inst. Math. Sci., 15, 53-157 (1979) · Zbl 0408.60100
[3] Boscarino, S.; Pareschi, L.; Russo, G., Implicit-explicit Runge-Kutta scheme for hyperbolic systems and kinetic equations in the diffusion limit, SIAM J. Sci. Comput., 35, 22-51 (2013) · Zbl 1264.65150
[4] Chen, Z., Liu, L., Mu, L.: Solving the linear transport equation by a deep neural network approach, arXiv preprint arXiv:2102.09157 (2021) · Zbl 1484.65213
[5] Dimarco, G.; Pareschi, L., Numerical methods for kinetic equations, Acta Numer., 23, 369-520 (2014) · Zbl 1398.65260
[6] Egger, H.; Schlottbom, M., An lp theory for stationary radiative transfer, Appl. Anal., 93, 1283-1296 (2014) · Zbl 1292.35091
[7] Golse, F.; Jin, S.; Levermore, CD, A domain decomposition analysis for a two-scale linear transport problem, ESAIM: Math. Model. Numer. Anal., 37, 869-892 (2003) · Zbl 1078.65125
[8] Han, H.; Tang, M.; Ying, W., Two uniform tailored finite point schemes for the two dimensional discrete ordinates transport equations with boundary and interface layers, Commun. Comput. Phys., 15, 797-826 (2014) · Zbl 1373.82076
[9] Han, J., Jentzen, A., et al.: Algorithms for solving high dimensional pdes: from nonlinear monte carlo to machine learning. arXiv preprint arXiv:2008.13333 (2020) · Zbl 1490.60202
[10] Han, J.; Jentzen, A.; Weinan, E., Solving high-dimensional partial differential equations using deep learning, Proc. Natl. Acad. Sci., 115, 8505-8510 (2018) · Zbl 1416.35137
[11] Hwang, HJ; Jang, JW; Jo, H.; Lee, JY, Trend to equilibrium for the kinetic Fokker-Planck equation via the neural network approach, J. Comput. Phys., 419 (2020) · Zbl 07507228
[12] Jin, S., Asymptotic preserving (AP) schemes for multiscale kinetic and hyperbolic equations: a review, Riv. Mat. Univ. Parma, 2, 177-216 (2012) · Zbl 1259.82079
[13] Jin, S., Ma, Z., Wu, K.: Asymptotic-preserving neural networks for multiscale time-dependent linear transport equations, arXiv preprint arXiv:2111.02541 (2021)
[14] Jin, S.; Pareschi, L.; Toscani, G., Uniformly accurate diffusive relaxation schemes for multiscale transport equations, SIAM J. Numer. Anal., 38, 913-936 (2000) · Zbl 0976.65091
[15] Jin, S.; Tang, M.; Han, H., A uniformly second order numerical method for the one-dimensional discrete-ordinate transport equation and its diffusion limit with interface, Netw. Heterogeneous Media, 4, 35-65 (2009) · Zbl 1191.65181
[16] Klar, A., Asymptotic-induced domain decomposition methods for kinetic and drift diffusion semiconductor equations, SIAM J. Sci. Comput., 19, 2032-2050 (1998) · Zbl 0918.65090
[17] Klar, A., An asymptotic-induced scheme for nonstationary transport equations in the diffusive limit, SIAM J. Numer. Anal., 35, 1073-1094 (1998) · Zbl 0918.65091
[18] Lagaris, IE; Likas, A.; Fotiadis, DI, Artificial neural networks for solving ordinary and partial differential equations, IEEE Trans. Neural Netw., 9, 987-1000 (1998)
[19] Lee, J. Y., Jang, J. W., Hwang, H. J.: The model reduction of the Vlasov-Poisson-Fokker-Planck system to the Poisson-Nernst-Planck system via the deep neural network approach, arXiv preprint arXiv:2009.13280 (2020) · Zbl 07420189
[20] Lemou, M.; Méhats, F., Micro-macro schemes for kinetic equations including boundary layers, SIAM J. Sci. Comput., 34, B734-B760 (2012) · Zbl 1266.82060
[21] Lemou, M.; Mieussens, L., New asymptotic preserving scheme based on micro-macro formulation for linear kinetic equations in the diffusion limit, SIAM J. Sci. Comput., 31, 334-368 (2008) · Zbl 1187.82110
[22] Lewis, E., Miller, W., Jr.: Computational Methods of Neutron Transport. Wiley, London (1983)
[23] Li, Q.; Lu, J.; Sun, W., Diffusion approximations and domain decomposition method of linear transport equations: asymptotics and numerics, J. Comput. Phys., 292, 141-167 (2015) · Zbl 1349.82096
[24] Li, Q.; Wang, L., Implicit asymptotic preserving method for linear transport equations, Commun. Comput. Phys., 22, 157-181 (2017) · Zbl 1488.65262
[25] Li, W., Song, P., Wang, Y.: An asymptotic-preserving imex method for nonlinear radiative transfer equation, arXiv preprint arXiv:2008.06730 (2020) · Zbl 07549615
[26] Liu, L., Zeng, T., Zhang, Z.: A deep neural network approach on solving the linear transport model under diffusive scaling, arXiv preprint arXiv:2102.12408 (2021)
[27] Lu, J., Lu, Y.: A priori generalization error analysis of two-layer neural networks for solving high dimensional schrödinger eigenvalue problems, arXiv preprint arXiv:2105.01228 (2021)
[28] Lu, Y., Lu, J., Wang, M.: A priori generalization analysis of the deep ritz method for solving high dimensional elliptic partial differential equations. In: Conference on Learning Theory, PMLR, pp. 3196-3241 (2021)
[29] Manteuffel, TA; Ressel, KJ; Starke, G., A boundary functional for the least-squares finite-element solution of neutron transport problems, SIAM J. Numer. Anal., 37, 556-586 (1999) · Zbl 0949.65142
[30] Mishra, S., Molinaro, R.: Estimates on the generalization error of physics informed neural networks (PINNs) for approximating PDEs (2020). arXiv preprint arXiv:2006.16144 · Zbl 07524707
[31] Peng, Z.; Cheng, Y.; Qiu, J-M; Li, F., Stability-enhanced ap imex-ldg schemes for linear kinetic transport equations under a diffusive scaling, J. Comput. Phys., 415 (2020) · Zbl 1440.65147
[32] 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., 378, 686-707 (2019) · Zbl 1415.68175
[33] Shalev-Shwartz, S.; Ben-David, S., Understanding Machine Learning: From Theory to Algorithms (2014), Cambridge: Cambridge University Press, Cambridge · Zbl 1305.68005
[34] Sirignano, J.; Spiliopoulos, K., Dgm: a deep learning algorithm for solving partial differential equations, J. Comput. Phys., 375, 1339-1364 (2018) · Zbl 1416.65394
[35] Sun, W.; Jiang, S.; Xu, K., An implicit unified gas kinetic scheme for radiative transfer with equilibrium and non-equilibrium diffusive limits, Commun. Comput. Phys., 22, 889-912 (2017) · Zbl 1488.80008
[36] Tang, M.; Wang, L.; Zhang, X., Accurate front capturing asymptotic preserving scheme for nonlinear gray radiative transfer equation, SIAM J. Sci. Comput., 43, B759-B783 (2021) · Zbl 1469.65141
[37] Wang, S., Teng, Y., Perdikaris, P.: Understanding and mitigating gradient pathologies in physics-informed neural networks, arXiv preprint arXiv:2001.04536 (2020) · Zbl 07395804
[38] Wang, S., Yu, X., Perdikaris, P.: When and why pinns fail to train: a neural tangent kernel perspective, arXiv preprint arXiv:2007.14527 (2020) · Zbl 07524768
[39] Weinan, E.; Yu, B., The deep ritz method: a deep learning-based numerical algorithm for solving variational problems, Commun. Math. Stat., 6, 1-12 (2018) · Zbl 1392.35306
[40] Yang, X.; Golse, F.; Huang, Z.; Jin, S., Numerical study of a domain decomposition method for a two-scale linear transport equation, Netw. Heterogeneous Media, 1, 143 (2006) · Zbl 1115.65396
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.