Learning Modular Simulations for Homogeneous Systems
This addresses the problem of scalable and efficient simulation for complex systems in engineering, though it is incremental as it builds on existing neural methods.
The paper tackles modeling homogeneous multibody dynamical systems by introducing a modular simulation framework that combines graph neural networks and neural differential equations, enabling accurate predictions and zero-shot generalization to new configurations with reduced data and training effort.
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.