OmniFluids: Physics Pre-trained Modeling of Fluid Dynamics

TL;DR

OmniFluids is a physics pre-trained model for fluid dynamics that efficiently adapts to various tasks, providing fast, accurate predictions and turbulence statistics with minimal data, outperforming existing AI methods.

Contribution

The paper introduces OmniFluids, a physics pre-trained model that captures fluid laws and adapts to diverse tasks with minimal data, combining physics-based pre-training and novel architecture.

Findings

Outperforms state-of-the-art AI methods in flow prediction and turbulence statistics.

Achieves 10-100x speedups over traditional CFD solvers.

Accurately identifies unknown physical parameters from sparse data.

Abstract

Computational fluid dynamics (CFD) drives progress in numerous scientific and engineering fields, yet high-fidelity simulations remain computationally prohibitive. While machine learning approaches offer computing acceleration, they typically specialize in single physical systems or require extensive training data, hindering their applicability in highly nonlinear and 3D flow scenarios. To overcome these limitations, we propose OmniFluids, a pure physics pre-trained model that captures fundamental fluid dynamics laws and adapts efficiently to diverse downstream tasks with minimal data. We develop a training framework combining physics-only pre-training, coarse-grid operator distillation, and few-shot fine-tuning. This enables OmniFluids to retain broad physics knowledge while delivering fast and accurate predictions. Architecturally, OmniFluids integrates a mixture of operators, a multi-frame decoder, and factorized Fourier layers, seamlessly incorporating physics-based supervision while allowing efficient and scalable modeling of diverse tasks. Extensive tests on a broad range of 2D and 3D benchmarks show that OmniFluids outperforms state-of-the-art AI-driven methods in terms of flow field prediction and turbulence statistics. It delivers 10--100$\times$ speedups over traditional solvers while maintaining a comparable accuracy and accurately identifies unknown physical parameters from sparse, noisy data. This work demonstrates the potential of training a unified CFD solver exclusively from physics knowledge, offering a new approach for efficient and generalizable modeling across complex fluid systems.