NGC Catalog
CLASSIC
Welcome Guest
Resources
Modulus Datasets: Cylinder Flow

Modulus Datasets: Cylinder Flow

For downloads and more information, please view on a desktop device.
Logo for Modulus Datasets: Cylinder Flow
Description
Cylinder flow dataset in npy format used in the mesh-reduced GNN model: https://arxiv.org/pdf/2201.09113.pdf
Publisher
NVIDIA
Latest Version
v1
Modified
April 11, 2024
Compressed Size
1.58 GB

Dataset overview

This dataset includes the simulations used in training and evaluation of the model proposed in Predicting physics in mesh-reduced space with temporal attention for the cylinder flow use case. The dataset is generated by solving the incompressible equations based on the Finite Volume Method (FVM). The open-source library OpenFOAM is used to conduct all CFD simulations. All flow domains are discretized by triangle (tri) mesh cells and the mesh resolutions are adaptive to the richness of flow features. The dataset consists of 51 training and 50 test trajectories.

License

Attribution-ShareAlike 4.0 International

Reference

Predicting physics in mesh-reduced space with temporal attention