Autoregressive Forecasting on Kuramoto-Sivashinsky (KS) equation (LPSDA protocol, test)
0.03NMSEPACE-FNO (AR)
Evaluation results
| Method | Links | |
|---|---|---|
| PACE-FNO (AR)Symmetry (Sym.)=windowed g1 + PACE g2–g3, Architecture=m=20, w=64, 4 layers, 20 frames, Number of training samples=5122026.05 | 0.03 | |
| PACE-FNO (AR)Symmetry (Sym.)=windowed g1 + structured g2–g3, Architecture=m=20, w=64, 4 layers, 1 frame, Number of training samples=5122026.05 | 0.08 | |
| FNO (AR)Symmetry (Sym.)=g1g2g3, Architecture=m=32, w=256, 5 layers, 20 frames, Number of training samples=5122026.05 | 0.09 | |
| FNO (AR)Symmetry (Sym.)=-, Architecture=m=32, w=256, 5 layers, 20 frames, Number of training samples=5122026.05 | 0.1 |