More of a minor detail, but I think in the total_force_loss method of the BurgersPDE class, self.scalars['Total Force'] should be set to l1 / self.dt instead of just l1. This way, the values are independent of the time step size and comparable to the values of the FieldEffect used to generate the ground truth.
More of a minor detail, but I think in the total_force_loss method of the BurgersPDE class, self.scalars['Total Force'] should be set to l1 / self.dt instead of just l1. This way, the values are independent of the time step size and comparable to the values of the FieldEffect used to generate the ground truth.