/FYS5429

Primary LanguagePython

Project repo for FYS5429-Advanced Machine Learning for Physical Sciences

Requirements

All requirements can b installed by creating an environment using:

conda create --name <env> --file heartpinn_env.txt

Training

1D cable

pinn1D.py can be run as is, the output will be the RMSE for the test data:

python pinn1D.py

2D square domain

´pinn2D.py' takes command line arguments train and predict. The code uses Pytorch as backlend, which has to be specified. For training run:

DDE_BACKEND=pytorch python pinn2D.py train

MRI-based geometry

Similar to the 2D square run:

DDE_BACKEND=pytorch python heartpinn.py train