In this project, we investigate physics-informed neural networks (PINNs) that are constructed in such way that they separate potential (curl-free) and solenoidal (divergence-free) vector fields, what is known as Helmholtz--Hodge decomposition.
-
Clone the project with
git
via SSH:git clone [email protected]:dmitry-kabanov/hhpinn.git
or HTTPS:
git clone https://github.com/dmitry-kabanov/hhpinn.git
-
Install conda environment for the project:
conda env create -f environment.yml [-n env-name]
where optional argument env-name
has the default value hhpinn
.