poisson_CNN is a convolutional neural network model which estimates the solution of the Poisson equation with four Dirichlet or Neumann boundary conditions on rectangular grids of variable sizes.
Installation requires CUDA set up to work with tensorflow-gpu version 2.3 or newer. To install, please use the Dockerfile appropriate for your CPU architecture (in most cases, docker/Dockerfile-amd64
)
An article describing the performance of our model is available: journal | arXiv
If you use this repo in your work, please cite our paper:
Özbay AG, Hamzehloo A, Laizet S, Tzirakis P, Rizos G, Schuller B. Poisson CNN: Convolutional neural networks for the solution of the Poisson equation on a Cartesian mesh. Data-Centric Engineering. [Online] Cambridge University Press; 2021;2: e6. Available from: doi:10.1017/dce.2021.7