Efficient Parallelization of Message Passing Neural Network Potentials for Large-scale Molecular Dynamics
This work solves a scalability bottleneck for researchers using MPNN potentials in large-scale molecular dynamics simulations, enabling unprecedented scale applications.
The paper tackles the challenge of efficiently parallelizing message passing neural network (MPNN) potentials for large-scale molecular dynamics, achieving linear scaling and enabling simulations with over 100 million atoms as fast as local models.
Machine learning potentials have achieved great success in accelerating atomistic simulations. Many of them relying on atom-centered local descriptors are natural for parallelization. More recent message passing neural network (MPNN) models have demonstrated their superior accuracy and become increasingly popular. However, efficiently parallelizing MPNN models across multiple nodes remains challenging, limiting their practical applications in large-scale simulations. Here, we propose an efficient parallel algorithm for MPNN models, in which additional data communication is minimized among local atoms only in each MP layer without redundant computation, thus scaling linearly with the layer number. Integrated with our recursively embedded atom neural network model, this algorithm demonstrates excellent strong scaling and weak scaling behaviors in several benchmark systems. This approach enables massive molecular dynamics simulations on MPNN models as fast as on strictly local models for over 100 million atoms, vastly extending the applicability of the MPNN potential to an unprecedented scale. This general parallelization framework can empower various MPNN models to efficiently simulate very large and complex systems.