CFD Benchmark Dataset

CFD Benchmark is a comprehensive benchmark dataset designed for training and evaluating neural PDE solvers. It comprises six standard tasks covering regular grids, structured grids, and irregular geometries. The dataset was utilized for unified evaluation in the ICML 2024 paper Transolver. Some data within this benchmark originates from the FNO and Geo-FNO works. The dataset contains six subsets: airfoil, darcy, elasticity, ns (Navier-Stokes), pipe, and plas.