train.py

This file contains scripts to train the ICNN using the physics-guided loss function.

train_weak(model, datasets, fem_material, noise_level):

Input arguments:

• model - class instance of the ICNN model that is to be trained.
• datasets - dataset to be used to train the ICNN model.
• fem_material - name of the material to be learned (for file naming convention).
• noise_level - level of noise used to condition the dataset (for file naming convention).

Output arguments:

• model - returns trained model class instance
• loss_history - returns loss history