NurshatMenglik/

NeuralConvection_2D_TrainTest

📄 Paper: A Neural Surrogate Approach for Simulating Natural Convection Problems (arXiv:2606.25259) — Nurshat Menglik, Alex Shao, David Hyde. Paired 2D Boussinesq -> compressible natural-convection fields used to train and evaluate the Fourier Neural Operator (FNO) surrogate in A Neural Surrogate Approach for Simulating Natural Convection Problems. Each split provides two tensors:

X.pt — Boussinesq input, shape (N, 4, 64, 64), float32… See the full description on the dataset page: https://huggingface.co/datasets/NurshatMenglik/NeuralConvection_2D_TrainTest.