Recent advances in deep learning for physics have focused on discovering shared
representations of target systems by incorporating physics priors or inductive biases into …