Operator Learning on Earthquake problem Eq. (3.16) (train)
1Relative L2 ErrorVanilla DeepONet
Evaluation results
| Method | Links | |
|---|---|---|
| Vanilla DeepONetTrunk network=Vanilla DNN, Trainable parameters=1,909,1482026.06 | 1 | |
| Improved Causality-DeepONetTrunk network=RepNet, Causality approach=Causal windowing [66], Trainable parameters=1,909,1482026.06 | 1.27 | |
| Fourier Neural Operator (SOAP)Fourier layers=4, Channels=32, Fourier modes=300, Integration scheme=SOAP [67], Trainable parameters=2,466,2732026.06 | 4.26 | |
| Fourier Neural OperatorFourier layers=4, Channels=32, Fourier modes=300, Trainable parameters=2,466,2732026.06 | 4.32 | |
| Improved Causality-DeepONet (SOAP)Trunk network=RepNet, Causality approach=Causal windowing [66], Integration scheme=SOAP [67], Trainable parameters=1,909,1482026.06 | 7.05 | |
| Improved DeepONetTrunk network=RepNet, Trainable parameters=1,909,1482026.06 | 9.51 |