mlsanityData to 7 Oct 2026

Learning Modular Simulations for Homogeneous Systems

arXiv 2210.16294 ↗First posted 28 Oct 2022cs.LG cs.MAcited by 4

Not on the map. Why?

Abstract (arXiv)

Complex systems are often decomposed into modular subsystems for engineering tractability. Although various equation based white-box modeling techniques make use of such structure, learning based methods have yet to incorporate these ideas broadly. We present a modular simulation framework for modeling homogeneous multibody dynamical systems, which combines ideas from graph neural networks and neural differential equations. We learn to model the individual dynamical subsystem as a neural ODE module. Full simulation of the composite system is orchestrated via spatio-temporal message passing between these modules. An arbitrary number of modules can be combined to simulate systems of a wide variety of coupling topologies. We evaluate our framework on a variety of systems and show that message passing allows coordination between multiple modules over time for accurate predictions and in certain cases, enables zero-shot generalization to new system configurations. Furthermore, we show that our models can be transferred to new system configurations with lower data requirement and training effort, compared to those trained from scratch.

Key facts. Cited by 4 papers in the map (data through 7 Oct 2026). Builds on 26 of its 67 references that are in the map. First posted on arXiv 28 Oct 2022.

Builds on

26 of 67 references.

References · 26

  1. 2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,282
  2. 2018Neural Ordinary Differential Equationscited by 3,733
  3. 2021Efficiently Modeling Long Sequences with Structured State Spacescited by 2,424
  4. 2020Learning to Simulate Complex Physics with Graph Networkscited by 867
  5. 2020Learning Mesh-Based Simulation with Graph Networkscited by 840
  6. 2016Interaction Networks for Learning about Objects, Relations and Physicscited by 503
  7. 2021Machine learning accelerated computational fluid dynamicscited by 461
  8. 2019Augmented Neural ODEscited by 396
  9. 2020Lagrangian Neural Networkscited by 388
  10. 2020Scalable Gradients for Stochastic Differential Equationscited by 264
  11. 2019Deep Lagrangian Networks: Using Physics as Model Prior for Deep Learningcited by 205
  12. 2019Neural Stochastic Differential Equations: Deep Latent Gaussian Models in the Diffusion Limitcited by 168
14 more
  1. 2019Graph Neural Ordinary Differential Equationscited by 153
  2. 2019Neural Jump Stochastic Differential Equationscited by 147
  3. 2020Learning to Control PDEs with Differentiable Physicscited by 131
  4. 2020Simplifying Hamiltonian and Lagrangian Neural Networks via Explicit Constraintscited by 102
  5. 2019Neural SDE: Stabilizing Neural ODE Networks with Stochastic Noisecited by 98
  6. 2017Deep Learning for Physical Processes: Incorporating Prior Scientific Knowledgecited by 95
  7. 2019Physics-Informed Neural Networks for Power Systemscited by 80
  8. 2018Fashionable Modelling with Fluxcited by 53
  9. 2018Learning to Predict the Cosmological Structure Formationcited by 38
  10. 2019Neural Dynamics on Complex Networkscited by 38
  11. 2019Combining Generative and Discriminative Models for Hybrid Inferencecited by 34
  12. 2020Structured Mechanical Models for Robot Learning and Controlcited by 21
  13. 2019A General Framework for Structured Learning of Mechanical Systemscited by 13
  14. 2021Composing Modeling and Simulation with Machine Learning in Juliacited by 3

Built on by

Cited by year

Citing papers per year: 2023: 1; 2024: 0; 2025: 3.

4 papers.

  1. 2023PGODE: Towards High-quality System Dynamics Modelingcited by 7
  2. 2025STFlow: Data-Coupled Flow Matching for Geometric Trajectory Simulationcited by 7
  3. 2025Sample-Efficient Diffusion-based Control of Complex Physics Systemscited by 0
  4. 2025Spatiotemporal graph neural process for reconstruction, extrapolation, and classification of cardiac trajectoriescited by 0