The Cost-Accuracy Trade-Off In Operator Learning With Neural Networks

The repo contains data for the paper @article{de2022cost, title={The Cost-Accuracy Trade-Off In Operator Learning With Neural Networks}, author={De Hoop, Maarten and Huang, Daniel Zhengyu and Qian, Elizabeth and Stuart, Andrew M}, journal={arXiv preprint arXiv:2203.13181}, year={2022} }. The data set is generated to study different neural networks for AI + engineering. The data set contains simulation data for the following 4 benchmark problems 1) Navier stokes equation : NavierStokes_inputs.npy & NavierStokes_outputs.npy. 2) Helmholtz equation : Helmholtz_inputs.npy & Helmholtz_outputs.npy. 3) Structural mechanics equation : StructuralMechanics_inputs.npy & StructuralMechanics_outputs.npy. 4) Advection equation : Advection_inputs.npy & Advection_outputs.npy. The data are stored as nx by ny by ndata arrays (2d problems) or nx by ndata arrays (1d problems).