Moving Time Window Example

Physics Simulation PhysicsNeMo-ML Model Framework Technical Support (PhysicsNeMo Only)
1

Hi,
I am developing a Time Window PINNs solver (similar to the Taylor Green example) but with data loss. However, I am unsure how to best implement data loss since data will change with time. Here is what I have so far.

for i, specific_time in enumerate(np.linspace(0,time_window_size,nr_time_windows)):
        velData_invar, velData_outvar=GetVelocityData(velocity_data_normalized[i],VelocityArrayName,DistanceThresholdPercentile)
        data = PointwiseConstraint.from_numpy(
            nodes=nodes,
            invar=velData_invar,
            outvar=velData_outvar,
            batch_size=cfg.batch_size.data,
        )
        domain.add_constraint(data, "data_constraints")

Is this the correct approach? Will data be called each time-window call from sequential solver? Thanks a bunch in advance!