CFDBench (Fluid Dynamics)

← Back to all benchmarks

Date: 2024-10-01

Name: CFDBench Fluid Dynamics

Domain: Mathematics

Focus: Neural operator surrogate modeling

Task Types: Surrogate modeling

Metrics: L2 error, MAE

Models: FNO, DeepONet, U-Net

AI/ML Motif: Regression

Resources

Benchmark: Visit

Keywords

neural operators CFD FNO DeepONet

Citation

• Yining Luo, Yingfa Chen, and Zhen Zhang. Cfdbench: a large-scale benchmark for machine learning methods in fluid dynamics. 2024. URL: https://arxiv.org/abs/2310.05963.

@misc{luo2024cfdbenchlargescalebenchmarkmachine,
  title={CFDBench: A Large-Scale Benchmark for Machine Learning Methods in Fluid Dynamics},
  author={Luo, Yining and Chen, Yingfa and Zhang, Zhen},
  year={2024},
  url={https://arxiv.org/abs/2310.05963}
}

Ratings

CategoryRating

Software

5.00

The benchmark provides Python scripts for data loading, preprocessing, and model training/evaluation

Specification

0.00

Not listed

Dataset

0.00

Not given

Metrics

5.00

Quantitative metrics (L2 error, MAE, relative error) are clearly defined and align with regression task objectives.

Reference Solution

5.00

Baseline models like FNO and DeepONet are implemented, hardware specified.

Documentation

5.00

Associated paper gives all necessary information.

Average rating: 3.33/5

Radar plot

Edit: edit this entry