Code and examples for "Structure-Preserving Digital Twins via Conditional Neural Whitney Forms" by Kinch et al.
Standard machine learning and scientific computing dependencies.
To generate basis functions and some matrices, we use scikit-fem (https://github.com/kinnala/scikit-fem).
This repo provides Jupyter notebooks for all the pedagogical examples in the paper. The 1D advection-diffusion example will run on a CPU relatively quickly, and is therefore the best place to start.