FLUID-LLM: Learning Computational Fluid Dynamics with Spatiotemporal-aware Large Language Models

Max Zhu,Adrián Bazaga,Pietro Liò

2024-06-07

Abstract:Learning computational fluid dynamics (CFD) traditionally relies on computationally intensive simulations of the Navier-Stokes equations. Recently, large language models (LLMs) have shown remarkable pattern recognition and reasoning abilities in natural language processing (NLP) and computer vision (CV). However, these models struggle with the complex geometries inherent in fluid dynamics. We introduce FLUID-LLM, a novel framework combining pre-trained LLMs with spatiotemporal-aware encoding to predict unsteady fluid dynamics. Our approach leverages the temporal autoregressive abilities of LLMs alongside spatial-aware layers, bridging the gap between previous CFD prediction methods. Evaluations on standard benchmarks reveal significant performance improvements across various fluid datasets. Our results demonstrate that FLUID-LLM effectively integrates spatiotemporal information into pre-trained LLMs, enhancing CFD task performance.

Machine Learning,Artificial Intelligence,Computation and Language

What problem does this paper attempt to address?

The paper attempts to address the issues of computational complexity and resource consumption associated with traditional numerical simulation methods in Computational Fluid Dynamics (CFD). Specifically, by introducing a new framework called FLUID-LLM, the study aims to combine pre-trained large language models (LLMs) with spatiotemporal-aware encoders to predict unsteady fluid dynamics. FLUID-LLM leverages the temporal autoregressive capabilities of LLMs and spatial-aware layers, thereby achieving significant performance improvements in fluid dynamics tasks.

The main contributions of the paper are:
1. **Proposing the FLUID-LLM framework**: Combining pre-trained LLMs with spatiotemporal-aware encoders to predict fluid dynamics.
2. **Performance improvement**: In standard benchmarks, FLUID-LLM demonstrated significant performance enhancements across various fluid datasets.
3. **Integration of spatiotemporal information**: Effectively integrating spatiotemporal information into pre-trained LLMs, enhancing their performance in fluid dynamics tasks.
4. **Contextual learning capability**: Enhancing prediction accuracy through historical state information, achieving better generalization and few-shot learning.

Through these methods, FLUID-LLM can significantly reduce computation time and resource requirements when handling complex fluid dynamics simulations.