We propose a physics-informed neural network (PINN)-based method to solve the heat transfer equation under imperfect contact conditions. A major challenge arises from the discontinuity of the solution across the interface, where the exact jump is unknown and implicitly determined by the Kapitza thermal resistance condition. Since the neural network function is smooth on the entire domain, conventional PINN could be inefficient to capture such discontinuities without certain modifications. One remedy is to extend a piecewise continuous function on $ \mathbb{R}^d $ to a continuous function on $ \mathbb{R}^{d+1} $. This is achieved by applying a Sobolev extension for the solution within each subdomain and introducing, additional coordinate variable that labels the subdomains. This formulation enables the design of neural network functions in the augmented variable, which retains the universal approximation property. We define the PINN in an augmented variable by the minimizer of the loss functional, which includes the implicit interface conditions. Once the loss functional is minimized, the solution obtained by the axis-augmented PINN satisfies the implicit jump conditions. In this way, our method offers a user-friendly way to solve heat transfer equations with imperfect contact conditions. Another advantage of using a continuous representation of solutions in augmented variables is that it allows error analysis in the space of smooth functions. We provide an error analysis of the proposed method, demonstrating that the difference between the exact solution and the predicted solution is bounded by the physics-informed loss functional. Furthermore, the loss functional can be made small by increasing the parameters in the neural network such as the number of nodes in the hidden layers.
Citation: Hansaem Oh, Gwanghyun Jo. Physics-informed neural network for the heat equation under imperfect contact conditions and its error analysis[J]. AIMS Mathematics, 2025, 10(4): 7920-7940. doi: 10.3934/math.2025364
We propose a physics-informed neural network (PINN)-based method to solve the heat transfer equation under imperfect contact conditions. A major challenge arises from the discontinuity of the solution across the interface, where the exact jump is unknown and implicitly determined by the Kapitza thermal resistance condition. Since the neural network function is smooth on the entire domain, conventional PINN could be inefficient to capture such discontinuities without certain modifications. One remedy is to extend a piecewise continuous function on $ \mathbb{R}^d $ to a continuous function on $ \mathbb{R}^{d+1} $. This is achieved by applying a Sobolev extension for the solution within each subdomain and introducing, additional coordinate variable that labels the subdomains. This formulation enables the design of neural network functions in the augmented variable, which retains the universal approximation property. We define the PINN in an augmented variable by the minimizer of the loss functional, which includes the implicit interface conditions. Once the loss functional is minimized, the solution obtained by the axis-augmented PINN satisfies the implicit jump conditions. In this way, our method offers a user-friendly way to solve heat transfer equations with imperfect contact conditions. Another advantage of using a continuous representation of solutions in augmented variables is that it allows error analysis in the space of smooth functions. We provide an error analysis of the proposed method, demonstrating that the difference between the exact solution and the predicted solution is bounded by the physics-informed loss functional. Furthermore, the loss functional can be made small by increasing the parameters in the neural network such as the number of nodes in the hidden layers.
| [1] |
L. J. Challis, Kapitza resistance and acoustic transmission across boundaries at high frequencies, J. Phys. C: Solid State Phys., 7 (1974), 481. https://doi.org/10.1088/0022-3719/7/3/005 doi: 10.1088/0022-3719/7/3/005
|
| [2] |
E. T. Swartz, R. O. Pohl, Thermal resistance at interfaces, Appl. Phys. Lett., 51 (1987), 2200–2202. https://doi.org/10.1063/1.98939 doi: 10.1063/1.98939
|
| [3] |
R. J. Stoner, H. J. Maris, Kapitza conductance and heat flow between solids at temperatures from 50 to 300 K, Phys. Rev. B, 48 (1993), 16373–16387. https://doi.org/10.1103/PhysRevB.48.16373 doi: 10.1103/PhysRevB.48.16373
|
| [4] |
D. G. Cahill, W. K. Ford, K. E. Goodson, G. D. Mahan, A. Majumdar, H. J. Maris, et al., Nanoscale thermal transport, J. Appl. Phys., 93 (2003), 793–818. https://doi.org/10.1063/1.1524305 doi: 10.1063/1.1524305
|
| [5] |
P. Kapitza, Heat transfer and superfluidity of helium II, Phys. Rev., 60 (1941), 354–355. https://doi.org/10.1103/PhysRev.60.354 doi: 10.1103/PhysRev.60.354
|
| [6] |
C. W. Nan, R. Birringer, Determining the kapitza resistance and the thermal, Phys. Rev. B, 57 (1998), 8264–8268. https://doi.org/10.1103/PhysRevB.57.8264 doi: 10.1103/PhysRevB.57.8264
|
| [7] |
J. Yvonnet, Q. C. He, Q. Z. Zhu, J. F. Shao, A general and efficient computational procedure for modelling the Kapitza thermal resistance based on XFEM, Comput. Mater. Sci., 50 (2011), 1220–1224. https://doi.org/10.1016/j.commatsci.2010.02.040 doi: 10.1016/j.commatsci.2010.02.040
|
| [8] |
J. Weisz, On an iterative method for the solution of discretized elliptic problems with imperfect contact condition, J. Comput. Appl. Math., 72 (1996), 319–333. https://doi.org/10.1016/0377-0427(96)00003-9 doi: 10.1016/0377-0427(96)00003-9
|
| [9] |
S. Hüeber, B. I. Wohlmuth, Thermo-mechanical contact problems on non-matching meshes, Comput. Method. Appl. M., 198 (2009), 1338–1350. https://doi.org/10.1016/j.cma.2008.11.022 doi: 10.1016/j.cma.2008.11.022
|
| [10] |
G. H. Jo, D. Y. Kwak, Enriched P1-Conforming Methods for Elliptic Interface Problems with Implicit Jump Conditions, Adv. Math. Phys., 2018 (2018), 9891281. https://doi.org/10.1155/2018/9891281 doi: 10.1155/2018/9891281
|
| [11] | D. Y. Kwak, S. Lee, Y. Hyon, A new finite element for interface problems having robin type jump, Int. J. Numer. Anal. Model., 14 (2017), 532–549. |
| [12] |
J. Yvonnet, Q. C. He, C. Toulemonde, Numerical modelling of the effective conductivities of composites with arbitrarily shaped inclusions and highly conducting interface, Compos. Sci. Technol., 68 (2008), 2818–2825. https://doi.org/10.1016/j.compscitech.2008.06.008 doi: 10.1016/j.compscitech.2008.06.008
|
| [13] |
J. T. Liu, S. T. Gu, E. Monteiro, Q. C. He, A versatile interface model for thermal conduction phenomena and its numerical implementation by XFEM, Comput. Mech., 53 (2014), 825–843. https://doi.org/10.1007/s00466-013-0933-9 doi: 10.1007/s00466-013-0933-9
|
| [14] |
T. Xue, X. Zhang, K. K. Tamma, Generalized heat conduction model involving imperfect thermal contact surface: Application of the GSSSS-1 differential-algebraic equation time integration, Int. J. Heat Mass Tran., 116 (2018), 889–896. https://doi.org/10.1016/j.ijheatmasstransfer.2017.09.081 doi: 10.1016/j.ijheatmasstransfer.2017.09.081
|
| [15] |
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 doi: 10.1016/j.jcp.2018.10.045
|
| [16] | Automatic Differentiation in PyTorch, Adv. Neural Inf. Process. Syst., 2017. Available from: https://openreview.net/forum?id = BJJsrmfCZ |
| [17] |
E. Haghighat, A. C. Bekar, E. Madenci, R. Juanes, A nonlocal physics-informed deep learning framework using the peridynamic differential operator, Comput. Method. Appl. Mech. Eng., 385 (2021), 114012. https://doi.org/10.1016/j.cma.2021.114012 doi: 10.1016/j.cma.2021.114012
|
| [18] |
H. Bararnia, M. Esmaeilpour, On the application of physics informed neural networks (PINN) to solve boundary layer thermal-fluid problems, Int. Commun. Heat Mass Tran., 132 (2022), 105890. https://doi.org/10.1016/j.icheatmasstransfer.2022.105890 doi: 10.1016/j.icheatmasstransfer.2022.105890
|
| [19] |
L. Ning, Z. Cai, H. Dong, Y. Liu, W. Wang, Physics-informed neural network frameworks for crack simulation based on minimized peridynamic potential energy, Comput. Method. Appl. Mech. Eng., 417 (2023), 116430. https://doi.org/10.1016/j.cma.2023.116430 doi: 10.1016/j.cma.2023.116430
|
| [20] |
M. Baldan, P. D. Barba, D. A. Lowther, Physics-Informed Neural Networks for Inverse Electromagnetic Problems, IEEE T. Magn., 59 (2023), 1–5. https://doi.org/10.1109/TMAG.2023.3247023 doi: 10.1109/TMAG.2023.3247023
|
| [21] |
W. F. Hu, T. S. Lin, M. C. Lai, A discontinuity capturing shallow neural network for elliptic interface problems, J. Comput. Phys., 469 (2022), 111576. https://doi.org/10.1016/j.jcp.2022.111576 doi: 10.1016/j.jcp.2022.111576
|
| [22] |
H. J. Park, G. H. Jo, A physics-informed neural network based method for the nonlinear Poisson-Boltzmann equation and its error analysis, J. Comput. Phys., 522 (2025), 113579. https://doi.org/10.1016/j.jcp.2024.113579 doi: 10.1016/j.jcp.2024.113579
|
| [23] |
P. Hansbo, M. G. Larson, Discontinuous Galerkin methods for incompressible and nearly incompressible elasticity by Nitsche's method, Comput. Method. Appl. Mech. Eng., 191 (2002), 1895–1908. https://doi.org/10.1016/S0045-7825(01)00358-9 doi: 10.1016/S0045-7825(01)00358-9
|
| [24] | D. Leguillon, E. Sanchez-Palencia, Computation of singular solutions in elliptic problems and elasticity, New York: John Wiley & Sons, Inc., 1987. http://doi.org/10.5555/39281 |
| [25] | Evans, C. Lawrence, Partial differential equations, 2 Eds., Providence, RI: Am.Math.Soc., 2010. http://doi.org/10.1090/gsm/019 |
| [26] | J. Wloka, Partial differential equations, Cambridge: Cambridge Univ.Press, 1987. http://doi.org/10.1017/CBO9781139171755 |
| [27] |
K. Hornik, M. Stinchcombe, H. White, Multilayer feedforward networks are universal approximators, Neural Netw., 2 (1989), 359–366. https://doi.org/10.1016/0893-6080(89)90020-8 doi: 10.1016/0893-6080(89)90020-8
|
| [28] |
R. E. Caflisch, Monte Carlo and quasi-Monte Carlo methods, Acta Numer., 7 (1998), 1–49. https://doi.org/10.1017/S0962492900002804 doi: 10.1017/S0962492900002804
|
| [29] |
S. Mishra, R. Molinaro, Estimates on the generalization error of physics-informed neural networks for approximating PDEs, IMA J. Numer. Anal., 43 (2022), 1–43. https://doi.org/10.1093/imanum/drab093 doi: 10.1093/imanum/drab093
|
| [30] |
S. Mishra, R. Molinaro, Estimates on the generalization error of physics-informed neural networks for approximating a class of inverse problems for PDEs, IMA J. Numer. Anal., 42 (2021), 981–1022. https://doi.org/10.1093/imanum/drab032 doi: 10.1093/imanum/drab032
|
| [31] |
T. D. Ryck, A. D. Jagtap, S. Mishra, Error estimates for physics-informed neural networks approximating the Navier-Stokes equations, IMA J. Numer. Anal., 44 (2023), 83–119. https://doi.org/10.1093/imanum/drac085 doi: 10.1093/imanum/drac085
|
| [32] |
T. D. Ryck, S. Lanthaler, S. Mishra, On the approximation of functions by tanh neural networks, Neural Netw., 143 (2021), 732–750. https://doi.org/10.1016/j.neunet.2021.08.015 doi: 10.1016/j.neunet.2021.08.015
|
| [33] |
K. He, X. Zhang, S. Ren, J. Sun, Delving deep into rectifiers: Surpassing human-level performance on ImageNet classification, Proc. IEEE Int. Conf. Comput. Vis., (2015), 1026–1034. https://doi.org/10.1109/ICCV.2015.123 doi: 10.1109/ICCV.2015.123
|
| [34] |
D. W. Marquardt, An Algorithm for Least-Squares Estimation of Nonlinear Parameters, J. Soc. Ind. Appl. Math., 11 (1963), 431–441. https://doi.org/10.1137/0111030 doi: 10.1137/0111030
|
| [35] |
D. C. Liu, J. Nocedal, On the limited memory BFGS method for large scale optimization, Math. Program., 45 (1989), 503–528. https://doi.org/10.1007/BF01589116 doi: 10.1007/BF01589116
|
| [36] |
C. He, X. Hu, L. Mu, A mesh-free method using piecewise deep neural network for elliptic interface problems, J. Comput. Appl. Math., 412 (2022), 114358. https://doi.org/10.1016/j.cam.2022.114358 doi: 10.1016/j.cam.2022.114358
|
| [37] |
J. F. Urbán, P. Stefanou, J. A. Pons, Unveiling the optimization process of Physics Informed Neural Networks: How accurate and competitive can PINNs be?, J. Comput. Phys., 523 (2025), 113656. https://doi.org/10.1016/j.jcp.2024.113656 doi: 10.1016/j.jcp.2024.113656
|
| [38] |
C. Liu, L. Zhu, M. Belkin, Loss landscapes and optimization in over-parameterized non-linear systems and neural networks, Appl. Comput. Harmon. Anal., 59 (2022), 85–116. https://doi.org/10.1016/j.acha.2021.12.009 doi: 10.1016/j.acha.2021.12.009
|
Figures(8) / Tables(3)
Hansaem Oh, Gwanghyun Jo. Physics-informed neural network for the heat equation under imperfect contact conditions and its error analysis[J]. AIMS Mathematics, 2025, 10(4): 7920-7940. doi: 10.3934/math.2025364
DownLoad: