REVIEW 66 references
Learnable-Differentiable Finite Volume Solver for Accelerated Simulation of Flows
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Simulation of fluid flows is crucial for modeling physical phenomena like meteorology, aerodynamics, and biomedicine. Classical numerical solvers often require fine spatiotemporal grids to satisfy stability, consistency, and convergence conditions, leading to substantial computational costs. Although machine learning has demonstrated better efficiency, they typically suffer from issues of interpretability, generalizability, and data dependency. Hence, we propose a learnable and differentiable finite volume solver, called LDSolver, designed for efficient and accurate simulation of fluid flows on spatiotemporal coarse grids. LDSolver comprises two key components: (1) a differentiable finite volume solver, and (2) an learnable module providing equivalent approximation for fluxes (derivatives and interpolations), and temporal error correction on coarse grids. Even with limited training data (e.g., only a few trajectories), our model could accelerate the simulation while maintaining a high accuracy with superior generalizability. Experiments on different flow systems (e.g., Burgers, decaying, forced and shear flows) show that LDSolver achieves state-of-the-art performance, surpassing baseline models with notable margins.
Reference graph
Works this paper leans on
-
[1]
1995.Computational Fluid Dynamics
John David Anderson and John Wendt. 1995.Computational Fluid Dynamics. Vol. 206. Springer
1995
-
[2]
Yohai Bar-Sinai, Stephan Hoyer, Jason Hickey, and Michael P Brenner. 2019. Learning data-driven discretizations for partial differential equations.Proceedings of the National Academy of Sciences116, 31 (2019), 201814058
2019
-
[3]
Ernnie Illyani Basri, Adi Azriff Basri, Vizy Nazira Riazuddin, S F Shahwir, Z Mohammad, and K A Ahmad. 2016. Computational fluid dynamics study in biomedical applications: a review.International Journal of Fluids and Heat Transfer 1, 2 (2016), 2–14
2016
-
[4]
Peter Bauer, Alan Thorpe, and Gilbert Brunet. 2015. The quiet revolution of numerical weather prediction.Nature525, 7567 (2015), 47–55
2015
-
[5]
Federico Berto, Stefano Massaroli, Michael Poli, and Jinkyoo Park. 2022. Neural solvers for fast and accurate numerical optimal control.Advances in Neural Information Processing Systems(2022)
2022
-
[6]
Steven L Brunton, Bernd R Noack, and Petros Koumoutsakos. 2020. Machine learning for fluid mechanics.Annual review of fluid mechanics52, 1 (2020), 477–508
2020
-
[7]
Keaton J Burns, Geoffrey M Vasil, Jeffrey S Oishi, Daniel Lecoanet, and Benjamin P Brown. 2020. Dedalus: A flexible framework for numerical simulations with spectral methods.Physical Review Research2, 2 (2020), 023068
2020
-
[8]
Zhao Chen, Yang Liu, and Hao Sun. 2021. Physics-Informed Learning of Govern- ing Equations from Scarce Data.Nature Communications12, 1 (2021), 6136
2021
Show all 66 references
-
[9]
Gideon Dresdner, Dmitrii Kochkov, Peter Christian Norgaard, Leonardo Zepeda- Nunez, Jamie Smith, Michael Brenner, and Stephan Hoyer. 2023. Learning to correct spectral methods for simulating turbulent flows.Transactions on Machine Learning Research(2023)
2023
-
[11]
Han Gao, Luning Sun, and Jian-Xun Wang. 2021. PhyGeoNet: Physics-informed geometry-adaptive convolutional neural networks for solving parameterized steady-state PDEs on irregular domain.J. Comput. Phys.428 (2021), 110079
2021
-
[12]
Somdatta Goswami, Katiana Kontolati, Michael D Shields, and George Em Karni- adakis. 2022. Deep transfer operator learning for partial differential equations under conditional shift.Nature Machine Intelligence4, 12 (2022), 1155–1164
2022
-
[13]
Gaurav Gupta, Xiongye Xiao, and Paul Bogdan. 2021. Multiwavelet-based opera- tor learning for differential equations.Advances in Neural Information Processing Systems34 (2021), 24048–24062
2021
-
[14]
Jayesh K Gupta and Johannes Brandstetter. 2023. Towards Multi-spatiotemporal- scale Generalized PDE Modeling.Transactions on Machine Learning Research (2023)
2023
-
[15]
Quercus Hernandez, Alberto Badias, Francisco Chinesta, and Elías Cueto. 2023. Thermodynamics-informed neural networks for physically realistic mixed reality. Computer Methods in Applied Mechanics and Engineering407 (2023), 115912
2023
-
[16]
Philipp Holl and Nils Thuerey. 2024. PhiFlow: Differentiable Simulations for PyTorch, TensorFlow and Jax. InInternational Conference on Machine Learning
2024
-
[17]
Masanobu Horie and Naoto Mitsume. 2024. Graph Neural PDE Solvers with Conservation and Similarity-Equivariance.International Conference on Machine Learning(2024)
2024
-
[18]
Steeven Janny, Aurélien Beneteau, Madiha Nadri, Julie Digne, Nicolas Thome, and Christian Wolf. 2023. Eagle: Large-scale learning of turbulent fluid dynamics with mesh Transformers. InInternational Conference on Learning Representations
2023
-
[19]
Hrvoje Jasak, Aleksandar Jemcov, Zeljko Tukovic, et al. 2007. OpenFOAM: A C++ library for complex physics simulations. InInternational workshop on coupled methods in numerical dynamics, Vol. 1000. Dubrovnik, Croatia, 1–20
2007
-
[20]
2005.Spectral/hp Element Methods for Computational Fluid Dynamics
George Karniadakis and Spencer J Sherwin. 2005.Spectral/hp Element Methods for Computational Fluid Dynamics. Oxford University Press, USA
2005
-
[21]
T Kobayashi and M Tsubokura. 2009. CFD application in automotive industry. In 100 Volumes of ‘Notes on Numerical Fluid Mechanics’ 40 Years of Numerical Fluid Mechanics and Aerodynamics in Retrospect. Springer, 285–295
2009
-
[22]
Dmitrii Kochkov, Jamie A Smith, Ayya Alieva, Qing Wang, Michael P Bren- ner, and Stephan Hoyer. 2021. Machine learning–accelerated computational fluid dynamics.Proceedings of the National Academy of Sciences118, 21 (2021), e2101784118
2021
-
[23]
Zongyi Li, Daniel Zhengyu Huang, Burigede Liu, and Anima Anandkumar. 2023. Fourier neural operator with learned deformations for pdes on general geometries. Journal of Machine Learning Research24, 388 (2023), 1–26
2023
-
[24]
Zongyi Li, Nikola Kovachki, Kamyar Azizzadenesheli, Burigede Liu, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. 2021. Fourier Neural Operator for Parametric Partial Differential Equations. InInternational Conference on Learning Representations
2021
-
[25]
Zongyi Li, Nikola Kovachki, Chris Choy, Boyi Li, Jean Kossaifi, Shourya Otta, Mohammad Amin Nabian, Maximilian Stadler, Christian Hundt, Kamyar Aziz- zadenesheli, et al. 2024. Geometry-informed neural operator for large-scale 3D PDEs.Advances in Neural Information Processing S...
2024
-
[26]
Zongyi Li, Nikola Borislavov Kovachki, Kamyar Azizzadenesheli, Burigede Liu, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. 2021. Fourier Neural Operator for Parametric Partial Differential Equations. InInternational Conference on Learning Representations
2021
-
[27]
Zijie Li, Dule Shu, and Amir Barati Farimani. 2024. Scalable transformer for PDE surrogate modeling.Advances in Neural Information Processing Systems36 (2024)
2024
-
[28]
Xin-Yang Liu, Min Zhu, Lu Lu, Hao Sun, and Jian-Xun Wang. 2024. Multi- resolution partial differential equations preserved learning framework for spa- tiotemporal dynamics.Communications Physics7, 1 (2024), 31
2024
-
[29]
Zichao Long, Yiping Lu, and Bin Dong. 2019. PDE-net 2.0: Learning PDEs from data with a numericsymbolic hybrid deep network.J. Comput. Phys.399 (2019), KDD ’25, August 3–7, 2025, Toronto, ON, Canada Mengtao Yan et al. 108925
2019
-
[30]
Zichao Long, Yiping Lu, Xianzhong Ma, and Bin Dong. 2018. PDE-net: Learning PDEs from data. InInternational Conference on Machine Learning. 3208–3216
2018
-
[31]
Lu Lu, Pengzhan Jin, Guofei Pang, Zhongqiang Zhang, and George Em Karni- adakis. 2021. Learning Nonlinear Operators via DeepONet Based on the Universal Approximation Theorem of Operators.Nature Machine Intelligence3, 3 (2021), 218–229
2021
-
[32]
Yiping Lu, Aoxiao Zhong, Quanzheng Li, and Bin Dong. 2018. Beyond finite layer neural networks: Bridging deep architectures and numerical differential equations. InInternational Conference on Machine Learning. 3276–3285
2018
-
[33]
2023.An introduction to ansys fluent 2023
John E Matsson. 2023.An introduction to ansys fluent 2023. Sdc Publications
2023
-
[34]
2016.The Finite Volume Method
Fadl Moukalled, Luca Mangani, and Marwan Darwish. 2016.The Finite Volume Method. Springer
2016
-
[35]
Tobias Pfaff, Meire Fortunato, Alvaro Sanchez-Gonzalez, and Peter Battaglia
-
[36]
Michael Poli, Stefano Massaroli, Atsushi Yamashita, Hajime Asama, and Jinkyoo Park. 2020. Hypersolvers: Toward fast continuous-depth models.Advances in Neural Information Processing Systems33 (2020), 21105–21117
2020
-
[37]
Maziar Raissi, Paris Perdikaris, and George E Karniadakis. 2019. 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
2019
-
[38]
Maziar Raissi, Alireza Yazdani, and George Em Karniadakis. 2020. Hidden Fluid Mechanics: Learning Velocity and Pressure Fields from Flow Visualizations. Science367, 6481 (2020), 1026–1030
2020
-
[39]
Chengping Rao, Pu Ren, Yang Liu, and Hao Sun. 2022. Discovering Nonlin- ear PDEs from Scarce Data with Physics-Encoded Learning. InInternational Conference on Learning Representations
2022
-
[40]
Chengping Rao, Pu Ren, Qi Wang, Oral Buyukozturk, Hao Sun, and Yang Liu
-
[41]
Chengping Rao, Hao Sun, and Yang Liu. 2021. Physics-Informed Deep Learning for Computational Elastodynamics Without Labeled Data.Journal of Engineering Mechanics147, 8 (2021), 04021043
2021
-
[42]
Pu Ren, Chengping Rao, Yang Liu, Zihan Ma, Qi Wang, Jian-Xun Wang, and Hao Sun. 2023. Physr: Physics-informed deep super-resolution for spatiotemporal data.J. Comput. Phys.492 (2023), 112438
2023
-
[43]
Pu Ren, Chengping Rao, Yang Liu, Jian-Xun Wang, and Hao Sun. 2022. PhyCRNet: Physics-informed convolutional-recurrent network for solving spatiotemporal PDEs.Computer Methods in Applied Mechanics and Engineering389 (2022), 114399
2022
-
[44]
Alvaro Sanchez-Gonzalez, Jonathan Godwin, Tobias Pfaff, Rex Ying, Jure Leskovec, and Peter Battaglia. 2020. Learning to simulate complex physics with graph networks. InInternational Conference on Machine Learning. 8459–8468
2020
-
[45]
Tapio Schneider, João Teixeira, Christopher S Bretherton, Florent Brient, Kyle G Pressel, Christoph Schär, and A Pier Siebesma. 2017. Climate goals and computing the future of clouds.Nature Climate Change7, 1 (2017), 3–5
2017
-
[46]
J S Shang. 2009. Computational fluid dynamics application to aerospace science. The Aeronautical Journal113, 1148 (2009), 619–632
2009
-
[47]
2014.CFD vision 2030 study: a path to revolutionary computational aerosciences
Jeffrey P Slotnick, Abdollah Khodadoust, Juan Alonso, David Darmofal, William Gropp, Elizabeth Lurie, and Dimitri J Mavriplis. 2014.CFD vision 2030 study: a path to revolutionary computational aerosciences. Technical Report
2014
-
[48]
Zhiqing Sun, Yiming Yang, and Shinjae Yoo. 2023. A neural pde solver with temporal stencil modeling. InInternational Conference on Machine Learning. PMLR, 33135–33155
2023
-
[49]
Zhiqing Sun, Yiming Yang, and Shinjae Yoo. 2023. A neural PDE solver with temporal stencil modeling. InInternational Conference on Machine Learning. 33135–33155
2023
-
[50]
Kejun Tang, Jiayu Zhai, Xiaoliang Wan, and Chao Yang. 2024. Adversarial Adaptive Sampling: Unify PINN and Optimal Transport for the Approximation of PDEs. InInternational Conference on Learning Representations
2024
-
[51]
Alasdair Tran, Alexander Mathews, Lexing Xie, and Cheng Soon Ong. 2022. Factorized Fourier neural operators. InInternational Conference on Learning Representations
2022
-
[52]
Kiwon Um, Robert Brand, Yun Raymond Fei, Philipp Holl, and Nils Thuerey
-
[53]
Qi Wang, Pu Ren, Hao Zhou, Xin-Yang Liu, Zhiwen Deng, Yi Zhang, Ruizhi Chengze, Hongsheng Liu, Zidong Wang, Jian-Xun Wang, et al. 2024. P 2C2 Net: PDE-Preserved Coarse Correction Network for Efficient Prediction of Spatiotem- poral Dynamics.Advances in Neural Information Proce...
2024
-
[54]
Rui Wang, Karthik Kashinath, Mustafa Mustafa, Adrian Albert, and Rose Yu
-
[55]
Rui Wang, Robin Walters, and Rose Yu. 2021. Incorporating symmetry into deep dynamics models for improved generalization. InInternational Conference on Learning Representations
2021
-
[56]
Haixu Wu, Huakun Luo, Haowen Wang, Jianmin Wang, and Mingsheng Long
-
[57]
Mengtao Yan, Zhiming Zhang, Shangce Gao, and Shuyang Cao. 2022. Predicting aerodynamic pressure on a square cylinder from wake velocity field by masked gated recurrent unit model.Physics of Fluids34, 11 (2022)
2022
-
[58]
InProceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining
Towards Physics-Informed Deep Learning for Turbulent Flow Prediction. InProceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 1457–1466
-
[59]
Ruiyang Zhang, Yang Liu, and Hao Sun. 2020. Physics-Informed Multi-LSTM Networks for Metamodeling of Nonlinear Structures.Computer Methods in Applied Mechanics and Engineering369 (2020), 113226
2020
-
[60]
Jiawei Zhuang, Dmitrii Kochkov, Yohai Bar-Sinai, Michael P Brenner, and Stephan Hoyer. 2021. Learned discretizations for passive scalar advection in a two- dimensional turbulent flow.Physical Review Fluids6, 6 (2021), 064605
2021
-
[61]
2005.The Finite Element Method: Its Basis and Fundamentals
Olek C Zienkiewicz, Robert L Taylor, and Jian Z Zhu. 2005.The Finite Element Method: Its Basis and Fundamentals. Elsevier. Learnable-Differentiable Finite Volume Solver for Accelerated Simulation of Flows KDD ’25, August 3–7, 2025, Toronto, ON, Canada Appendix This supplementa...
2005
-
[63]
Bocheng Zeng, Qi Wang, Mengtao Yan, Yang Liu, Ruizhi Chengze, Yi Zhang, Hongsheng Liu, Zidong Wang, and Hao Sun. 2025. PhyMPGN: Physics-encoded Message Passing Graph Network for spatiotemporal PDE systems.International Conference on Learning Representations(2025)
2025
-
[67]
The learning rate is set to5× 10−4, with a decay factor of 0.9 applied every 5000 steps
and the Adam optimizer. The learning rate is set to5× 10−4, with a decay factor of 0.9 applied every 5000 steps. The model is trained for 5000 epochs with a batch size of 20. PeRCNN: The PeRCNN model [40] is used with the standard settings. The optimization is performed using ...
2025
-
[2020]
Solver-in-the-loop: Learning from differentiable physics to interact with iterative pde-solvers.Advances in Neural Information Processing Systems33 (2020), 6111–6122
2020
-
[2021]
InInternational Conference on Learning Representations
Learning meshbased simulation with graph networks. InInternational Conference on Learning Representations
-
[2023]
Encoding physics to learn reaction–diffusion processes.Nature Machine Intelligence5, 7 (2023), 765–779
2023
-
[2024]
International Conference on Machine Learning(2024)
Transolver: A fast transformer solver for PDEs on general geometries. International Conference on Machine Learning(2024)
2024
Discussion (0). Sign in to comment.