NeurIPS 2024 ML4CFD Competition: Results and Retrospective Analysis

TL;DR

The paper reports on the results of the NeurIPS 2024 ML4CFD competition, demonstrating how machine learning models can surpass traditional CFD solvers in aerodynamic simulations through a comprehensive benchmarking and analysis.

Contribution

It introduces a systematic competition for ML-based surrogate modeling in CFD, providing insights into effective approaches and evaluation strategies for scientific machine learning.

Findings

Top ML models outperformed traditional solvers on aggregate metrics.

The competition attracted over 240 teams, fostering diverse innovative solutions.

Analysis identified key design principles for robust ML surrogates in CFD.

Abstract

The integration of machine learning (ML) into the physical sciences is reshaping computational paradigms, offering the potential to accelerate demanding simulations such as computational fluid dynamics (CFD). Yet, persistent challenges in accuracy, generalization, and physical consistency hinder the practical deployment of ML models in scientific domains. To address these limitations and systematically benchmark progress, we organized the ML4CFD competition, centered on surrogate modeling for aerodynamic simulations over two-dimensional airfoils. The competition attracted over 240 teams, who were provided with a curated dataset generated via OpenFOAM and evaluated through a multi-criteria framework encompassing predictive accuracy, physical fidelity, computational efficiency, and out-of-distribution generalization. This retrospective analysis reviews the competition outcomes, highlighting several approaches that outperformed baselines under our global evaluation score. Notably, the top entry exceeded the performance of the original OpenFOAM solver on aggregate metrics, illustrating the promise of ML-based surrogates to outperform traditional solvers under tailored criteria. Drawing from these results, we analyze the key design principles of top submissions, assess the robustness of our evaluation framework, and offer guidance for future scientific ML challenges.