Error Calculation (Difference Between True Solution and Prediction)

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

Hello and thank you for the support,

I have a question about the prediction results: In the validation function, how is the u_diff calculated? Is it using the L2 norm between the prediction and the true solution? Where can I look at the implementation of the validation function?

fully_connected_VAL_004_u
fully_connected_VAL_004_u1477×424 69.6 KB

2

Hi @cpe.sk

Good question. The built in validators calculate the the relative L2 error for the output loss metric, but this isn’t what is plotted. The absolute error is what is shown in the contours by the default validation plotter.