Surrogate Modeling of 3D Rayleigh-Benard Convection with Equivariant Autoencoders
Not on the map. Why?
Abstract (arXiv)
The use of machine learning for modeling, understanding, and controlling large-scale physics systems is quickly gaining in popularity, with examples ranging from electromagnetism over nuclear fusion reactors and magneto-hydrodynamics to fluid mechanics and climate modeling. These systems - governed by partial differential equations - present unique challenges regarding the large number of degrees of freedom and the complex dynamics over many scales both in space and time, and additional measures to improve accuracy and sample efficiency are highly desirable. We present an end-to-end equivariant surrogate model consisting of an equivariant convolutional autoencoder and an equivariant convolutional LSTM using G-steerable kernels. As a case study, we consider the three-dimensional Rayleigh-B'enard convection, which describes the buoyancy-driven fluid flow between a heated bottom and a cooled top plate. While the system is E(2)-equivariant in the horizontal plane, the boundary conditions break the translational equivariance in the vertical direction. Our architecture leverages vertically stacked layers of D4-steerable kernels, with additional partial kernel sharing in the vertical direction for further efficiency improvement. We demonstrate significant gains in sample and parameter efficiency, as well as a better scaling to more complex dynamics. The accompanying code is available under https://github.com/FynnFromme/equivariant-rb-forecasting.
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 46 of its 82 references that are in the map. First posted on arXiv 19 May 2025.
Builds on
46 of 82 references.
References · 46
- 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
- 2019A Comprehensive Survey on Graph Neural Networkscited by 2,919
- 2020Fourier Neural Operator for Parametric Partial Differential Equationscited by 2,908
- 2019DeepONet: Learning nonlinear operators for identifying differential equations based on the universal approximation theorem of operatorscited by 2,018
- 20163D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotationcited by 1,476
- 2015Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcastingcited by 1,436
- 2021Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gaugescited by 1,167
- 2016Group Equivariant Convolutional Networkscited by 1,165
- 2015Scheduled Sampling for Sequence Prediction with Recurrent Neural Networkscited by 803
- 2019General E(2)-Equivariant Steerable CNNscited by 446
- 2019Machine Learning for Fluid Mechanicscited by 385
- 2016Steerable CNNscited by 305
34 more
- 2017Learning SO(3) Equivariant Representations with Spherical CNNscited by 248
- 2022Towards Multi-spatiotemporal-scale Generalized PDE Modelingcited by 197
- 2022FourCastNet: Accelerating Global High-Resolution Weather Forecasting using Adaptive Fourier Neural Operatorscited by 188
- 2016A guide to convolution arithmetic for deep learningcited by 185
- 2018Data-Driven Forecasting of High-Dimensional Chaotic Systems with Long Short-Term Memory Networkscited by 155
- 2017Linearly-Recurrent Autoencoder Networks for Learning Dynamicscited by 145
- 2024The Well: a Large-Scale Collection of Diverse Physics Simulations for Machine Learningcited by 145
- 2020MDP Homomorphic Networks: Group Symmetries in Reinforcement Learningcited by 142
- 2020Incorporating Symmetry into Deep Dynamics Models for Improved Generalizationcited by 126
- 2020Forecasting Sequential Data using Consistent Koopman Autoencoderscited by 115
- 2024Physics-Informed Diffusion Modelscited by 90
- 2019Data-driven approximation of the Koopman generator: Model reduction, system identification, and controlcited by 85
- 2017Polar Transformer Networkscited by 81
- 2021Coordinate Independent Convolutional Networks -- Isometry and Gauge Equivariant Convolutions on Riemannian Manifoldscited by 73
- 20223DLinker: An E(3) Equivariant Variational Autoencoder for Molecular Linker Designcited by 33
- 2025Generative Latent Neural PDE Solver using Flow Matchingcited by 28
- 2023Effective control of two-dimensional Rayleigh--B'enard convection: invariant multi-agent reinforcement learning is all you needcited by 27
- 2024Clifford-Steerable Convolutional Neural Networkscited by 25
- 2023Distributed Control of Partial Differential Equations Using Convolutional Reinforcement Learningcited by 24
- 2021Steerable Partial Differential Operators for Equivariant Neural Networkscited by 22
- 2025Sparse identification of nonlinear dynamics and Koopman operators with Shallow Recurrent Decoder Networkscited by 20
- 2019Representing Closed Transformation Paths in Encoded Network Latent Spacecited by 17
- 2022Lorentz group equivariant autoencoderscited by 15
- 2024Temporally Consistent Koopman Autoencoders for Forecasting Dynamical Systemscited by 15
- 2022Direct data-driven forecast of local turbulent heat flux in Rayleigh-B'enard convectioncited by 10
- 2024Multi-agent reinforcement learning for the control of three-dimensional Rayleigh-B'enard convectioncited by 8
- 2023Equivariance and partial observations in Koopman operator theory for partial differential equationscited by 6
- 2024Advanced deep-reinforcement-learning methods for flow control: group-invariant and positional-encoding networks improve learning speed and qualitycited by 5
- 2024Towards aerodynamic surrogate modeling based on β-variational autoencoderscited by 5
- 2025Solving Turbulent Rayleigh-B'enard Convection using Fourier Neural Operatorscited by 5
- 2022Rotationally Equivariant Super-Resolution of Velocity Fields in Two-Dimensional Fluids Using Convolutional Neural Networkscited by 3
- 2024Koopman-Based Surrogate Modelling of Turbulent Rayleigh-B'enard Convectioncited by 3
- 2023Reconstructing Rayleigh-Benard flows out of temperature-only measurements using Physics-Informed Neural Networkscited by 2
- 2024Solving Partial Differential Equations with Equivariant Extreme Learning Machinescited by 2
Built on by
Citing papers per year: 2025: 2; 2026: 1.
3 papers.