FluidFormer: Transformer with Continuous Convolution for Particle-based Fluid Simulation

TL;DR

FluidFormer introduces a novel Transformer-based architecture with continuous convolutions and self-attention for particle-based fluid simulation, effectively capturing local details and global dependencies to improve stability and accuracy.

Contribution

It pioneers a Transformer architecture tailored for fluid simulation, combining local continuous convolutions with global self-attention in a dual-pipeline framework.

Findings

Achieves state-of-the-art simulation stability and accuracy.

Effectively models long-range physical phenomena.

Outperforms existing neural fluid simulation methods.

Abstract

Learning-based fluid simulation networks have been proven as viable alternatives to traditional numerical solvers for the Navier-Stokes equations. Existing neural methods follow Smoothed Particle Hydrodynamics (SPH) frameworks, which inherently rely only on local inter-particle interactions. However, we emphasize that global context integration is also essential for learning-based methods to stabilize complex fluid simulations. We propose the first Fluid Attention Block (FAB) with a local-global hierarchy, where continuous convolutions extract local features while self-attention captures global dependencies. This fusion suppresses the error accumulation and models long-range physical phenomena. Furthermore, we pioneer the first Transformer architecture specifically designed for continuous fluid simulation, seamlessly integrated within a dual-pipeline architecture. Our method establishes a new paradigm for neural fluid simulation by unifying convolution-based local features with attention-based global context modeling. FluidFormer demonstrates state-of-the-art performance, with stronger stability in complex fluid scenarios.