fPINNs: Fractional Physics-Informed Neural Networks
Open accessThe DOI matches its Crossref record; 43 of the 43 references it cites were found in a registry or an index.
Abstract
Physics-informed neural networks (PINNs), introduced in [M. Raissi, P. Perdikaris, and G. Karniadakis, J. Comput. Phys., 378 (2019), pp. 686--707], are effective in solving integer-order partial differential equations (PDEs) based on scattered and noisy data. PINNs employ standard feedforward neural networks (NNs) with the PDEs explicitly encoded into the NN using automatic differentiation, while the sum of the mean-squared PDE residuals and the mean-squared error in initial-boundary conditions is minimized with respect to the NN parameters. Here we extend PINNs to fractional PINNs (fPINNs) to solve space-time fractional advection-diffusion equations (fractional ADEs), and we study systematically their convergence, hence explaining both fPINNs and PINNs for the first time. Specifically, we demonstrate their accuracy and effectiveness in solving multidimensional forward and inverse problems with forcing terms whose values are only known at randomly scattered spatio-temporal coordinates ( black-box (BB) forcing terms). A novel element of the fPINNs is the hybrid approach that we introduce for constructing the residual in the loss function using both automatic differentiation for the integer-order operators and numerical discretization for the fractional operators. This approach bypasses the difficulties stemming from the fact that automatic differentiation is not applicable to fractional operators because the standard chain rule in integer calculus is not valid in fractional calculus. To discretize the fractional operators, we employ the Grünwald--Letnikov (GL) formula in one-dimensional fractional ADEs and the vector GL formula in conjunction with the directional fractional Laplacian in two- and three-dimensional fractional ADEs. We first consider the one-dimensional fractional Poisson equation and compare the convergence of the fPINNs against the finite difference method (FDM). We present the solution convergence using both the mean $L^2$ error as well as the standard deviation due to sensitivity to NN parameter initializations. Using different GL formulas we observe first-, second-, and third-order convergence rates for small size training sets but the error saturates for larger training sets. We explain these results by analyzing the four sources of numerical errors due to discretization, sampling, NN approximation, and optimization. The total error decays monotonically (below $10^{-5}$ for a third-order GL formula) but it saturates beyond that point due to the optimization error. We also analyze the relative balance between discretization and sampling errors and observe that the sampling size and the number of discretization points ( auxiliary points) should be comparable to achieve the highest accuracy. As we increase the depth of the NN up to certain value, the mean error decreases and the standard deviation increases, whereas the width has essentially no effect unless its value is either too small or too large. We next consider time-dependent fractional ADEs and compare white-box (WB) and BB forcing. We observe that for the WB forcing, our results are similar to the aforementioned cases; however, for the BB forcing fPINNs outperform FDM. Subsequently, we consider multidimensional time-, space-, and space-time-fractional ADEs using the directional fractional Laplacian and we observe relative errors of $10^{-3}\sim10^{-4}$. Finally, we solve several inverse problems in one, two, and three dimensions to identify the fractional orders, diffusion coefficients, and transport velocities and obtain accurate results given proper initializations even in the presence of significant noise.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · cited 1×
“We also used the extended stochastic gradient descent Adam algorithm [18] to optimize the loss function.”
MethodRaissi, M., Perdikaris, P., & Karniadakis, G. E. (2019) · Journal of Computational Physics · doi:10.1016/j.jcp.2018.10.045 · Cited by 20,384 · cited 4×
“We just need to add the parameters to be identified in the inverse problem to the list of parameters to be optimized in the forward problem without changing anything else [35].”
BackgroundMethodInfluentialSelf-citationModel-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
Finn, C., Abbeel, P., & Levine, S. (2017) · arXiv · doi:10.48550/arxiv.1703.03400 · Cited by 8,512 · cited 1×
“We can tune these attributes of NN architecture automatically by leveraging metalearning techniques [65, 66].”
MethodImproving neural networks by preventing co-adaptation of feature detectors
Hinton, G. E., Srivastava, N., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. R. (2012) · arXiv · doi:10.48550/arxiv.1207.0580 · Cited by 6,382 · cited 1×
“Additionally, the noise in the data can be naturally taken into account by employing regularization techniques, such as L(1), L(2) and L∞ regularization [32], early stopping [33], as well as dropout [34, 35].”
MethodDiscovering governing equations from data by sparse identification of nonlinear dynamical systems
Brunton, S. L., Proctor, J. L., & Kutz, J. N. (2016) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1517384113 · Cited by 4,796 · cited 10×
“In this paper, we focus on the identification of the parameters of fractional PDEs whose overall form is known but some coefficients and most importantly the operators themselves are not known.”
BackgroundInfluentialNeural Architecture Search with Reinforcement Learning
Zoph, B., & Le, Q. V. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1611.01578 · Cited by 3,974 · cited 1×
“We can tune these attributes of NN architecture automatically by leveraging meta-learning techniques [54].”
MethodArtificial neural networks for solving ordinary and partial differential equations
Lagaris, I. E., Likas, A., & Fotiadis, D. I. (1998) · IEEE Transactions on Neural Networks · doi:10.1109/72.712178 · Cited by 2,412 · cited 1×
“In particular, we concentrate on physics-informed neural networks (PINNs) [7, 46, 19, 35], which belong to the second aforementioned category.”
BackgroundHidden physics models: Machine learning of nonlinear partial differential equations
Raissi, M., & Karniadakis, G. E. (2018) · Journal of Computational Physics · doi:10.1016/j.jcp.2017.11.039 · Cited by 1,349 · cited 1×
“Other examples include, but not limited to, [21, 22, 23].”
BackgroundSelf-citationApplication of a fractional advection‐dispersion equation
Benson, D. A., Wheatcraft, S. W., & Meerschaert, M. M. (2000) · Water Resources Research · doi:10.1029/2000wr900031 · Cited by 1,301 · cited 1×
“Among these are solute transport in fractured and porous media [2, 3], acoustic wave propagation with frequency-dependent dissipation [4, 5, 6, 7], laminar and turbulent flows [8, 9, 10], viscoelastic constitutive laws [11], just to mention a few.”
BackgroundA fully discrete difference scheme for a diffusion-wave system
Sun, Z.-Z., & Wu, X. (2006) · Applied Numerical Mathematics · doi:10.1016/j.apnum.2005.03.003 · Cited by 1,235 · cited 2×
“The truncating error for the L1 scheme is (\Delta t) 2 - \gamma [44].”
Method
Research works printed but not linked: 15
- Blum A., 1988, Advances in Neural Information Processing Systems, 1, 494
- 10.1007/978-3-642-14574-2 · doi:10.1007/978-3-642-14574-2
- Glorot X., 2010, Proceedings of the 13th International Conference on Artificial Intelligence and Statistics, 249
- 10.1007/978-3-319-40663-3_36 · doi:10.1007/978-3-319-40663-3_36
- 10.1007/978-3-662-43930-2 · doi:10.1007/978-3-662-43930-2
- Gulian M., 2018, Machine Learning of Space-Fractional Differential Equations, preprint, arXiv:1808.00931 · arXiv
- Jin B., 2015, IMA J. Numer. Anal., 36, 197
- Kilbas A. A. A., 2006, North-Holland Math. Stud. 204
- Lee J. D., 2016, Proceedings of the Conference on Learning Theory, 1246
- Lei L., 2017, Advances in Neural Information Processing Systems, 2348
- 10.1142/p614 · doi:10.1142/p614
- Meerschaert M. M., 2004, Fract. Calc. Appl. Anal., 7, 61
- Podlubny I., 1999, Fractional Differential Equations, to Methods of Their Solution and Some of Their Applications, Math. Sci. Eng. 198
- Samko S. G., 1993, Gordon and Breach
- Srivastava N., 2014, J. Mach. Learn. Res., 15, 1929
Cited by
Physics-informed machine learning
Karniadakis, G. E., Kevrekidis, I. G., Lu, L., Perdikaris, P., Wang, S., & Yang, L. (2021) · Nature Reviews Physics · doi:10.1038/s42254-021-00314-5 · Cited by 8,718
Self-citationLearning nonlinear operators via DeepONet based on the universal approximation theorem of operators
Lu, L., Jin, P., Pang, G., Zhang, Z., & Karniadakis, G. E. (2021) · Nature Machine Intelligence · doi:10.1038/s42256-021-00302-5 · Cited by 3,418
Self-citationScientific Machine Learning Through Physics–Informed Neural Networks: Where we are and What’s Next
Cuomo, S., Di Cola, V. S., Giampaolo, F., Rozza, G., Raissi, M., & Piccialli, F. (2022) · Journal of Scientific Computing · doi:10.1007/s10915-022-01939-z · Cited by 2,784
“The solution proposed in Pang et al [128], is then extended in Kharazmi et al [77] where they address also time-dependent fractional orders.”
BackgroundMethodInfluentialDeepXDE: A Deep Learning Library for Solving Differential Equations
Lu, L., Meng, X., Mao, Z., & Karniadakis, G. E. (2021) · SIAM Review · doi:10.1137/19m1274067 · Cited by 2,512
“Finally we note that with AD we differentiate the NN and therefore we can deal with noisy data [42].”
BackgroundMethodInfluentialSelf-citationPhysics-informed neural networks (PINNs) for fluid mechanics: a review
Cai, S., Mao, Z., Wang, Z., Yin, M., & Karniadakis, G. E. (2021) · Acta Mechanica Sinica · doi:10.1007/s10409-021-01148-1 · Cited by 2,084
“Beneficial due to the high flexibility and the expressive ability in function approximation, PINNs have been extended to solve various classes of PDEs, e.g., integro-differential equations [21], fractional equations [21], surfaces PDEs [22] and stochastic differential equations [23].”
MethodSelf-citationWhen and why PINNs fail to train: A neural tangent kernel perspective
Wang, S., Yu, X., & Perdikaris, P. (2022) · Journal of Computational Physics · doi:10.1016/j.jcp.2021.110768 · Cited by 1,496
Physics-informed neural networks for high-speed flows
Mao, Z., Jagtap, A. D., & Karniadakis, G. E. (2020) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2019.112789 · Cited by 1,282
Self-citationD. Jagtap, A., & Em Karniadakis, G. (2020) · Communications in Computational Physics · doi:10.4208/cicp.oa-2020-0164 · Cited by 1,130
“Thanks to their universal approximation properties, they can be exploited to construct alternative approaches for solving PDEs.”
BackgroundInfluentialSelf-citationInterpretable machine learning: Fundamental principles and 10 grand challenges
Rudin, C., Chen, C., Chen, Z., Huang, H., Semenova, L., & Zhong, C. (2022) · Statistics Surveys · doi:10.1214/21-ss133 · Cited by 958
Physics-Informed Neural Networks with Hard Constraints for Inverse Design
Lu, L., Pestourie, R., Yao, W., Wang, Z., Verdugo, F., & Johnson, S. G. (2021) · SIAM Journal on Scientific Computing · doi:10.1137/21m1397908 · Cited by 771
Self-citationhp-VPINNs: Variational physics-informed neural networks with domain decomposition
Kharazmi, E., Zhang, Z., & Karniadakis, G. E. M. (2021) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2020.113547 · Cited by 818
Self-citationWu, C., Zhu, M., Tan, Q., Kartha, Y., & Lu, L. (2023) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2022.115671 · Cited by 723
“In addition to standard PDEs, PINNs have also been extended to solve other types of PDEs, including integro-differential equations [3], fractional PDEs [9], and stochastic PDEs [10].”
BackgroundMethodSelf-citationGradient-enhanced physics-informed neural networks for forward and inverse PDE problems
Yu, J., Lu, L., Meng, X., & Karniadakis, G. E. (2022) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2022.114823 · Cited by 722
“In some PDEs, it is possible to enforce the boundary conditions exactly and automatically by modifying the network architecture [20, 4, 25, 8], which eliminates the loss term of boundary conditions.”
BackgroundMethodSelf-citationPPINN: Parareal physics-informed neural network for time-dependent PDEs
Meng, X., Li, Z., Zhang, D., & Karniadakis, G. E. (2020) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2020.113250 · Cited by 654
Self-citationLu, L., Meng, X., Cai, S., Mao, Z., Goswami, S., Zhang, Z., & Karniadakis, G. E. (2022) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2022.114778 · Cited by 514
“For example, using the DeepONet as a building block in [43] for modeling hypersonic flows, we were able to obtain very accurate results outside the Mach number training range [8 , 10], e.g., at Mach number 7 and 11, by assimilating only up to five new measurements of velocity and temperature.”
MethodResultSelf-citationSelf-adaptive loss balanced Physics-informed neural networks
Xiang, Z., Peng, W., Liu, X., & Yao, W. (2022) · Neurocomputing · doi:10.1016/j.neucom.2022.05.015 · Cited by 507
“These discretization-based methods approximate the solution of PDEs through their values at a set of grid points distributed over the spatial and temporal domain.”
MethodInfluentialShin, Y., Darbon, J., & Em Karniadakis, G. (2020) · Communications in Computational Physics · doi:10.4208/cicp.oa-2020-0193 · Cited by 284
“A series of works have shown the effectiveness of PINNs in one, two or three dimensional problems: fractional PDEs [30, 36], stochastic differential equations [18, 39], biomedical problems [33], and fluid mechanics [27].”
BackgroundSelf-citationVariational Physics-Informed Neural Networks For Solving Partial Differential Equations
Kharazmi, E., Zhang, Z., & Karniadakis, G. E. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1912.00873 · Cited by 183
“The recent development of PINNs has been well established for forward and inverse problem of solving di ff erential equations [30, 32, 39, 29] and since then a number of extensions has been made to tackle several physical and biomedical problems [28, 18, 19].”
BackgroundSelf-citationUncertainty quantification in scientific machine learning: Methods, metrics, and comparisons
Psaros, A. F., Meng, X., Zou, Z., Guo, L., & Karniadakis, G. E. (2023) · Journal of Computational Physics · doi:10.1016/j.jcp.2022.111902 · Cited by 455
Self-citationMoseley, B., Markham, A., & Nissen-Meyer, T. (2023) · Advances in Computational Mathematics · doi:10.1007/s10444-023-10065-9 · Cited by 360
“PINNs have been extended to fractional differential equations [Pang et al., 2019], stochastic differential equations [Yang et al., 2020, Zhang et al., 2019], and have even been used to discover the underlying differential equations themselves [Chen et al., 2020a].”
MethodYuan, L., Ni, Y.-Q., Deng, X.-Y., & Hao, S. (2022) · Journal of Computational Physics · doi:10.1016/j.jcp.2022.111260 · Cited by 422
Zhuang, X., Guo, H., Alajlan, N., Zhu, H., & Rabczuk, T. (2021) · European Journal of Mechanics - A/Solids · doi:10.1016/j.euromechsol.2021.104225 · Cited by 430
Analyses of internal structures and defects in materials using physics-informed neural networks
Zhang, E., Dao, M., Karniadakis, G. E., & Suresh, S. (2022) · Science Advances · doi:10.1126/sciadv.abk0644 · Cited by 393
“Different residual terms (PDEs, BCs, and data) require different residual points.”
BackgroundInfluentialSelf-citationMishra, S., & Molinaro, R. (2022) · IMA Journal of Numerical Analysis · doi:10.1093/imanum/drab032 · Cited by 336
“There has been an explosive growth of papers that present PINNs for different applications and a very incomplete list of references includes [15,27,29,39,44,51,54,61].”
BackgroundEstimates on the generalization error of physics-informed neural networks for approximating PDEs
Mishra, S., & Molinaro, R. (2023) · IMA Journal of Numerical Analysis · doi:10.1093/imanum/drab093 · Cited by 222
Rai, R., & Sahu, C. K. (2020) · IEEE Access · doi:10.1109/access.2020.2987324 · Cited by 315
From PINNs to PIKANs: recent advances in physics-informed machine learning
Toscano, J. D., Oommen, V., Varghese, A. J., Zou, Z., Ahmadi Daryakenari, N., Wu, C., & Karniadakis, G. E. (2025) · Machine Learning for Computational Science and Engineering · doi:10.1007/s44379-025-00015-1 · Cited by 239
“To handle fractional derivatives, several studies have employed Monte Carlo methods [23, 102–106].”
InfluentialSelf-citationMIONet: Learning Multiple-Input Operators via Tensor Product
Jin, P., Meng, S., & Lu, L. (2022) · SIAM Journal on Scientific Computing · doi:10.1137/22m1477751 · Cited by 226
Self-citationThe Old and the New: Can Physics-Informed Deep-Learning Replace Traditional Linear Solvers?
Markidis, S. (2021) · Frontiers in Big Data · doi:10.3389/fdata.2021.669097 · Cited by 247
“For instance, fPINN (fractional PINN) is a PINN with a residual network capable of calculating residuals of governing equations including fractional calculus operators [28].”
BackgroundThree ways to solve partial differential equations with neural networks — A review
Blechschmidt, J., & Ernst, O. G. (2021) · GAMM-Mitteilungen · doi:10.1002/gamm.202100006 · Cited by 246
“In addition, PINNs have been applied successfully in a wide range of applications, including fluid dynamics [113,115,117,160,177], continuum mechanics and elastodynamics [66,132,162], inverse problems [91,121], fractional advection–diffusion equations [135], stochastic advection–diffusion–reaction equations [34], stoc…”
MethodKim, Y., Choi, Y., Widemann, D., & Zohdi, T. (2022) · Journal of Computational Physics · doi:10.1016/j.jcp.2021.110841 · Cited by 221
“Such models include, but are not limited to, attempts to mimic temporal evolution by incorporating a time integrator in a loss function [59,15,36,47,4] and to represent the solution with a trained NN [59,77,6], the deep Galerkin method [65], approximating spatial gradient functions with a multilayer feedforward NN [30…”
MethodZhao, C., Zhang, F., Lou, W., Wang, X., & Yang, J. (2024) · Physics of Fluids · doi:10.1063/5.0226562 · Cited by 197
Parametric deep energy approach for elasticity accounting for strain gradient effects
Nguyen-Thanh, V. M., Anitescu, C., Alajlan, N., Rabczuk, T., & Zhuang, X. (2021) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2021.114096 · Cited by 229
Cai, S., Wang, Z., Lu, L., Zaki, T. A., & Karniadakis, G. E. (2021) · Journal of Computational Physics · doi:10.1016/j.jcp.2021.110296 · Cited by 214
“In addition to solving forward problems, neural networks are also used to identify the unknown coefficient values of PDEs from data [12, 13, 14, 15, 16].”
MethodSelf-citationPerspectives on the integration between first-principles and data-driven modeling
Bradley, W., Kim, J., Kilwein, Z., Blakely, L., Eydenberg, M., Jalvin, J., Laird, C., & Boukouvala, F. (2022) · Computers & Chemical Engineering · doi:10.1016/j.compchemeng.2022.107898 · Cited by 231
“The authors extend the PINN approach, to solve fractional advection-diffusion equations (Pang et al., 2019).”
Tang, K., Wan, X., & Yang, C. (2023) · Journal of Computational Physics · doi:10.1016/j.jcp.2022.111868 · Cited by 190
“Two major types of deep leaning methods have been proposed for solving PDEs, including the variational form subject to deep learning techniques [4, 5, 6, 7] and the physics-informed neural networks (PINNs) [8, 9, 10, 3], both of which reformulate a PDE problem as an optimization problem and train a deep neural network…”
MethodSchiassi, E., Furfaro, R., Leake, C., De Florio, M., Johnston, H., & Mortari, D. (2021) · Neurocomputing · doi:10.1016/j.neucom.2021.06.015 · Cited by 205
Numerical methods for nonlocal and fractional models
D’Elia, M., Du, Q., Glusa, C., Gunzburger, M., Tian, X., & Zhou, Z. (2020) · Acta Numerica · doi:10.1017/s096249292000001x · Cited by 200
Non-invasive inference of thrombus material properties with physics-informed neural networks
Yin, M., Zheng, X., Humphrey, J. D., & Karniadakis, G. E. (2021) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2020.113603 · Cited by 186
“Successful applications of PINNs range from flow visualization [30], to high-speed flows [20], to stochastic PDEs [47], to fractional PDEs [26], and cardiac flows [13], to name a few.”
BackgroundSelf-citationLin, S., & Chen, Y. (2022) · Journal of Computational Physics · doi:10.1016/j.jcp.2022.111053 · Cited by 183
“NSFnets [2], VPINNs [3], fPINNs [4], B-PINNs [5] and hp-VPINNs [6], were devised and targeted at different application situations.”
MethodError estimates for physics-informed neural networks approximating the Navier–Stokes equations
Ryck, T. D., Jagtap, A. D., & Mishra, S. (2022) · IMA Journal of Numerical Analysis · doi:10.1093/imanum/drac085 · Cited by 242
Lu, L., Pestourie, R., Johnson, S. G., & Romano, G. (2022) · Physical Review Research · doi:10.1103/physrevresearch.4.023210 · Cited by 232
Self-citationPhysics informed neural networks for simulating radiative transfer
Mishra, S., & Molinaro, R. (2021) · Journal of Quantitative Spectroscopy and Radiative Transfer · doi:10.1016/j.jqsrt.2021.107705 · Cited by 231
“PINNs have been successfully applied to simulate a variety of forward and inverse problems for PDEs, see [5, 22, 23, 26, 33, 36, 37, 38] and references therein.”
BackgroundPang, G., D’Elia, M., Parks, M., & Karniadakis, G. E. (2020) · Journal of Computational Physics · doi:10.1016/j.jcp.2020.109760 · Cited by 211
Self-citationKashefi, A., & Mukerji, T. (2022) · Journal of Computational Physics · doi:10.1016/j.jcp.2022.111510 · Cited by 162
“[52, 45, 23, 47, 75]) fall in the category of unsupervised or weakly-supervised deep learning.”
A review of machine learning methods applied to structural dynamics and vibroacoustic
Cunha, B. Z., Droz, C., Zine, A., Foulard, S., & Ichchou, M. N. (2023) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2023.110535 · Cited by 201
Survey of Optimization Algorithms in Modern Neural Networks
Abdulkadirov, R. I., Lyakhov, P. A., & Nagornov, N. N. (2023) · Mathematics · doi:10.3390/math11112466 · Cited by 216
“There is a model for solving fractional differential equations in [207].”
Kharazmi, E., Cai, M., Zheng, X., Zhang, Z., Lin, G., & Karniadakis, G. E. (2021) · Nature Computational Science · doi:10.1038/s43588-021-00158-0 · Cited by 171
Self-citationMao, Z., & Meng, X. (2023) · Applied Mathematics and Mechanics · doi:10.1007/s10483-023-2994-7 · Cited by 167
“And it has become a powerful tool in various applications, such as fluid mechanics, hydrogeophysics([14]), transport phenomena in porous media([13]), and compressible Euler equations, since the PINN algorithm has several advantages compared with the classical numerical methods: (i) the algorithm of PINN is easy coding…”
MethodApplying machine learning to study fluid mechanics
Brunton, S. L. (2021) · Acta Mechanica Sinica · doi:10.1007/s10409-021-01143-6 · Cited by 177
Optimal Control of Pdes Using Physics-Informed Neural Networks
Mowlavi, S., & Nabi, S. (2022) · Journal of Computational Physics · doi:10.1016/j.jcp.2022.111731 · Cited by 163
Physics‐Informed Neural Network Method for Forward and Backward Advection‐Dispersion Equations
He, Q., & Tartakovsky, A. M. (2021) · Water Resources Research · doi:10.1029/2020wr029479 · Cited by 154
nn-PINNs: Non-Newtonian physics-informed neural networks for complex fluid modeling
Mahmoudabadbozchelou, M., Karniadakis, G. E., & Jamali, S. (2021) · Soft Matter · doi:10.1039/d1sm01298c · Cited by 153
Self-citationGuo, L., Wu, H., Yu, X., & Zhou, T. (2022) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2022.115523 · Cited by 149
“To approximate the time-fractional derivative of the neural network output u NN ( x , t ) , a finite di ff erence L 1 scheme is employed in [1].”
BackgroundMethodInfluentialNiu, S., Zhang, E., Bazilevs, Y., & Srivastava, V. (2022) · Journal of the Mechanics and Physics of Solids · doi:10.1016/j.jmps.2022.105177 · Cited by 148
Hanna, J. M., Aguado, J. V., Comas-Cardona, S., Askri, R., & Borzacchiello, D. (2022) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2022.115100 · Cited by 146
“PINN was first introduced in [28] and has been used to solve many forward and inverse problems [29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39].”
Error analysis for physics-informed neural networks (PINNs) approximating Kolmogorov PDEs
Ryck, T. D., & Mishra, S. (2022) · Advances in Computational Mathematics · doi:10.1007/s10444-022-09985-9 · Cited by 196
“A very incomplete list of references includes [35, 23, 26, 32, 40, 13, 14, 27, 28, 29, 1] and references therein.”
BackgroundMachine learning for polymer composites process simulation – a review
Cassola, S., Duhovic, M., Schmidt, T., & May, D. (2022) · Composites Part B Engineering · doi:10.1016/j.compositesb.2022.110208 · Cited by 135
Chen, X., Yang, L., Duan, J., & Karniadakis, G. E. (2021) · SIAM Journal on Scientific Computing · doi:10.1137/20m1360153 · Cited by 133
Self-citationDeep learning of free boundary and Stefan problems
Wang, S., & Perdikaris, P. G. (2020) · Journal of Computational Physics · doi:10.1016/j.jcp.2020.109914 · Cited by 131
Physics-Informed Neural Networks for Nonhomogeneous Material Identification in Elasticity ImagingPreprint
Zhang, E., Yin, M., & Karniadakis, G. E. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2009.04525 · Cited by 121
“We place 41×41 training points on uniform grid nodes in the square domain in the undeformed configuration, which are used for evaluating the loss terms for PDEs and the incompressibility condition.”
MethodInfluentialSelf-citationEnhancing PINNs for solving PDEs via adaptive collocation point movement and adaptive loss weighting
Hou, J., Li, Y., & Ying, S. (2023) · Nonlinear Dynamics · doi:10.1007/s11071-023-08654-w · Cited by 119
“In this section, we give a quick introduction to PINNs, which are used to solve nonlinear PDEs.”
BackgroundInfluentialWu, Z., Wang, H., He, C., Zhang, B., Xu, T., & Chen, Q. (2023) · Industrial & Engineering Chemistry Research · doi:10.1021/acs.iecr.3c02383 · Cited by 127
AT-PINN: Advanced time-marching physics-informed neural network for structural vibration analysis
Chen, Z., Lai, S.-K., & Yang, Z. (2023) · Thin-Walled Structures · doi:10.1016/j.tws.2023.111423 · Cited by 120
A gradient-enhanced physics-informed neural networks method for the wave equation
Xie, G., Fu, B., Li, H., Du, W., Zhong, Y., Wang, L., Geng, H., Zhang, J., & Si, L. (2024) · Engineering Analysis with Boundary Elements · doi:10.1016/j.enganabound.2024.105802 · Cited by 118
ERROR ESTIMATES OF RESIDUAL MINIMIZATION USING NEURAL NETWORKS FOR LINEAR PDES
Shin, Y., Zhang, Z., & Karniadakis, G. E. (2023) · Journal of Machine Learning for Modeling and Computing · doi:10.1615/jmachlearnmodelcomput.2023050411 · Cited by 117
Self-citationDwivedi, V., Parashar, N., & Srinivasan, B. (2020) · Neurocomputing · doi:10.1016/j.neucom.2020.09.006 · Cited by 115
Navidi, S., Thelen, A., Li, T., & Hu, C. K. (2024) · Energy storage materials · doi:10.1016/j.ensm.2024.103343 · Cited by 123
“Table 5: A qualitative comparison of state-of-the-art PIML approaches covered in this study problems [72, 73].”
BackgroundDarbon, J., Langlois, G. P., & Meng, T. (2020) · Research in the Mathematical Sciences · doi:10.1007/s40687-020-00215-6 · Cited by 135
Inverse Dirichlet weighting enables reliable training of physics informed neural networks
Maddu, S., Sturm, D., Müller, C. L., & Sbalzarini, I. F. (2021) · Machine Learning Science and Technology · doi:10.1088/2632-2153/ac3712 · Cited by 108
“de tion [22], as well as stochastic [23], high-dimensional [24], and fractional differential equations [25].”
Jiang, Z., Zhu, M., & Lu, L. (2024) · Reliability Engineering & System Safety · doi:10.1016/j.ress.2024.110392 · Cited by 125
Self-citationMahmoudabadbozchelou, M., Caggioni, M., Shahsavari, S., Hartt, W. H., Karniadakis, G. E., & Jamali, S. (2021) · Journal of Rheology · doi:10.1122/8.0000138 · Cited by 107
Self-citationMiao, Z., & Chen, Y. (2023) · Physica D Nonlinear Phenomena · doi:10.1016/j.physd.2023.133945 · Cited by 107
Data-driven methods for flow and transport in porous media: A review
Yang, G., Xu, R., Tian, Y., Guo, S., Wu, J., & Chu, X. (2024) · International Journal of Heat and Mass Transfer · doi:10.1016/j.ijheatmasstransfer.2024.126149 · Cited by 112
Temperature field inversion of heat-source systems via physics-informed neural networks
Liu, X., Peng, W., Gong, Z., Zhou, W., & Yao, W. (2022) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2022.104902 · Cited by 103
“…et al., 2020], hidden fluid mechanics [Raissi et al., 2020, Tartakovsky et al., 2020], biomedicine [Sahli Costabal et al., 2020, Kissas et al., 2020] and the parameter identification of wave, stochastic and fractional PDEs [Song and Alkhalifah, 2020, Tartakovsky et al., 2020, Pang et al., 2019].”
BackgroundMethodGW-PINN: A deep learning algorithm for solving groundwater flow equations
Zhang, X., Zhu, Y. P., Wang, J., Ju, L., Qian, Y., Ye, M., & Yang, J. (2022) · Advances in Water Resources · doi:10.1016/j.advwatres.2022.104243 · Cited by 103
Ryck, T. D., Mishra, S., & Molinaro, R. (2024) · SIAM Journal on Numerical Analysis · doi:10.1137/22m1522504 · Cited by 139
Correcting model misspecification in physics-informed neural networks (PINNs)
Zou, Z., Meng, X., & Karniadakis, G. E. (2024) · Journal of Computational Physics · doi:10.1016/j.jcp.2024.112918 · Cited by 114
Self-citationPan, I., Mason, L. R., & Matar, O. K. (2021) · Chemical Engineering Science · doi:10.1016/j.ces.2021.117271 · Cited by 118
A physics-informed feature weighting method for bearing fault diagnostics
Lü, H., Nemani, V. P., Barzegar, V., Allen, C., Hu, C. K., Laflamme, S., Sarkar, S., & Zimmerman, A. T. (2023) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2023.110171 · Cited by 111
“Notable applications of physics-informed deep learning have been attempted by the scientific computing community, focusing on solving partial differential equations [36–39], fractional equations [40,41], integral–differential equations [42,43], etc.”
BackgroundPhysics-Informed Neural Networks (PINNs) for Parameterized PDEs: A Metalearning ApproachPreprint
Penwarden, M., Zhe, S., Narayan, A., & Kirby, R. M. (2021) · SSRN Electronic Journal · doi:10.1016/j.jcp.2023.111912 · Cited by 96
Finite element method-enhanced neural network for forward and inverse problems
Meethal, R. E., Kodakkal, A., Khalil, M., Ghantasala, A., Obst, B., Bletzinger, K., & Wüchner, R. (2023) · Advanced Modeling and Simulation in Engineering Sciences · doi:10.1186/s40323-023-00243-1 · Cited by 93
“The surrogate canbeused topredict simulation results for parametric PDEs of the form L(u, λ) = 0.”
BackgroundInterpretability of Machine Learning: Recent Advances and Future Prospects
Gao, L., & Guan, L. (2023) · IEEE Multimedia · doi:10.1109/mmul.2023.3272513 · Cited by 104
“In [85], a fractional physics-informed NN model was proposed to address multidimensional forward and inverse problems with forcing terms whose values are only known at randomly scattered spatio-temporal coordinates (black-box forcing terms).”
BackgroundFallah, A., & Aghdam, M. M. (2023) · Engineering With Computers · doi:10.1007/s00366-023-01799-7 · Cited by 89
Zhang, Z., Zou, Z., Ellen Kuhl, A., & Karniadakis, G. E. (2023) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2023.116647 · Cited by 106
Self-citationFractional SEIR model and data-driven predictions of COVID-19 dynamics of Omicron variant
Cai, M., Karniadakis, G. E., & Li, C. (2022) · Chaos An Interdisciplinary Journal of Nonlinear Science · doi:10.1063/5.0099450 · Cited by 87
Self-citationde la Mata, F. F., Gijón, A., Molina-Solana, M., & Gómez‐Romero, J. (2022) · Physica A Statistical Mechanics and its Applications · doi:10.1016/j.physa.2022.128415 · Cited by 110
Suzuki, J. L., Gulian, M., Zayernouri, M., & D’Elia, M. (2022) · Journal of Peridynamics and Nonlocal Modeling · doi:10.1007/s42102-022-00085-2 · Cited by 96
“Other recent works addressing a similar learning problem for fractional operators include optimization-based approaches such as the one used in [84], fractional/nonlocal physics-informed neural network approaches such as [74, 170] and operator-regression techniques such as the one developed in [171].”
BackgroundMo, Y.-F., Ling, L., & Zeng, D. (2021) · Physics Letters A · doi:10.1016/j.physleta.2021.127739 · Cited by 84
Chen, Y., & Negro, L. D. (2022) · APL Photonics · doi:10.1063/5.0072969 · Cited by 82
Multifidelity modeling for Physics-Informed Neural Networks (PINNs)
Penwarden, M., Zhe, S., Narayan, A., & Kirby, R. M. (2021) · Journal of Computational Physics · doi:10.1016/j.jcp.2021.110844 · Cited by 95
“…collaborators have extended these methods to conservative PINNs (cPINNs) [27], variational PINNs (vPINNS) [28], parareal PINNs (pPINNs) [29], stochastic PINNs (sPINNs) [30], fractional PINNs (fPINNs) [31], LesPINNs (LES PINNs) [32], non-local PINNs (nPINNs) [33] and eXtended PINNs (xPINNs) [34].”
MethodData-driven learning of nonlocal physics from high-fidelity synthetic data
You, H., Yu, Y., Trask, N., Gulian, M., & D’Elia, M. (2020) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2020.113553 · Cited by 79
Guo, J., Zhu, H., Yang, Y., & Guo, C. (2025) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2025.112044 · Cited by 79
Qi, X., de Almeida, G. A. M., & Maldonado, S. (2024) · Journal of Hydrology · doi:10.1016/j.jhydrol.2024.131263 · Cited by 78
Lin, S., & Chen, Y. (2022) · Physica D Nonlinear Phenomena · doi:10.1016/j.physd.2022.133629 · Cited by 78
“This deep learning framework has breathed new life into the area of scientific computing and many efficient and significant variants on the basic structure were devised, such as VPINNs [5], fPINNs [6], XPINNs [7], g-PINNs [8], B-PINNs [9], hp-VPINNs [10], NSFnets [11] and so on.”
MethodMosaic flows: A transferable deep learning framework for solving PDEs on unseen domains
Wang, H., Planas, R., Chandramowlishwaran, A. M., & Bostanabad, R. (2021) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2021.114424 · Cited by 101
“…variational learning (to improve predictions on the boundaries) [30, 57], surrogating fractional PDEs (where AD cannot calculate residual errors) [58], emulating long time-dependent PDEs (where long temporal features render the training very difficult) [32], and learning operators that map…”
BackgroundPhysics-informed deep learning for digital materials
Zhang, Z., & Gu, G. X. (2021) · Theoretical and Applied Mechanics Letters · doi:10.1016/j.taml.2021.100220 · Cited by 76
VB-DeepONet: A Bayesian operator learning framework for uncertainty quantification
Garg, S., & Chakraborty, S. (2022) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2022.105685 · Cited by 90
Zhou, Z., & Yan, Z. (2020) · Physics Letters A · doi:10.1016/j.physleta.2020.127010 · Cited by 75
“[6–19] and references therein), in which the physics-information neural networks (PINNs) deep learning method [11] is a simple and effective method to solving the partial differential equations (PDEs) via deep learning, and was extended to other types of differential equations such as integro-differential equations, f…”
MethodPhysical laws meet machine intelligence: current developments and future directions
Muther, T., Dahaghi, A. K., Syed, F. I., & Pham, V. V. (2022) · Artificial Intelligence Review · doi:10.1007/s10462-022-10329-8 · Cited by 82
“Specifically, in a physics-informed machine learning setup, different PDEs (such as stochastic, differential, integer-order, integro-differential) along with their respective initial conditions (ICs) and boundary conditions (BCs) are integrated into the loss function of a neural network as (Raissi et al. 2019; Lu et a…”
Background
Metadata
- Authors
- Pang, G., Lu, L., & Karniadakis, G. E.
- Venue
- SIAM Journal on Scientific Computing
- Published
- 2019-01
- Volume
- 41
- Issue
- 4
- Pages
- A2603-A2626
- ISSN
- 1064-8275 (ISSN-L)
- 1095-7197
- OpenAlex
- W2900501930
- Cited by
- 1,105
- References
- 38
- Citations by year
- 2026: 253 · 2025: 260 · 2024: 195 · 2023: 155 · 2022: 119 · 2021: 82 · 2020: 35 · 2019: 6
- FWCI
- 28.01
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- SIAM Journal on Scientific Computing (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Army Research Office (W911NF-18-1-0301, W911NF-15-1-0562)
- U.S. Department of Energy (DE-SC0019434, DE-SC0019453)
- Topics
- Physical Sciences › Physics and Astronomy › Statistical and Nonlinear Physics › Model Reduction and Neural Networks
- Physical Sciences › Mathematics › Modeling and Simulation › Fractional Differential Equations Solutions
- Physical Sciences › Computer Science › Artificial Intelligence › Neural Networks and Applications
- Keywords
- physics-informed neural networks, fractional advection-diffusion equation, fractional Poisson equation, convergence analysis, inverse problems, sampling error, optimization error
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.