2D Darcy flow on 2D Darcy flow s = 141 x 141
0.0039Relative L2 LossU-ABLE
Evaluation Results
| Method | Links | |
|---|---|---|
| U-ABLETraining velocity(s/epoch)=4.49 ± 0.02, Flops=6.19G, Best learning rate=2e-32026.05 | 0.0039 | |
| U-FNOTraining velocity(s/epoch)=2.61 ± 0.03, Flops=6.11G, Best learning rate=3e-32026.05 | 0.0045 | |
| ABLETraining velocity(s/epoch)=2.22 ± 0.05, Flops=0.25G, Best learning rate=5e-32026.05 | 0.0055 | |
| FreqMoETraining velocity(s/epoch)=5.48 ± 0.07, Flops=0.18G, Best learning rate=5e-3 (2stage)2026.05 | 0.0062 | |
| FNOTraining velocity(s/epoch)=1.17 ± 0.03, Flops=0.18G, Best learning rate=6e-32026.05 | 0.0063 | |
| LSMTraining velocity(s/epoch)=4.56 ± 0.02, Flops=3.76G, Best learning rate=1e-32026.05 | 0.0068 | |
| GF-NOTraining velocity(s/epoch)=1.53 ± 0.43, Flops=0.18G, Best learning rate=3.6e-2(1000)2026.05 | 0.007 | |
| SAOTTraining velocity(s/epoch)=6.51 ± 0.47, Flops=0.85G, Best learning rate=3e-32026.05 | 0.008 | |
| UNetTraining velocity(s/epoch)=1.00 ± 0.14, Flops=2.99G, Best learning rate=2e-32026.05 | 0.0085 | |
| WNOTraining velocity(s/epoch)=4.20 ± 0.17, Flops=0.58G, Best learning rate=3e-32026.05 | 0.0285 | |
| AFNOTraining velocity(s/epoch)=6.85 ± 0.02, Flops=1.48G, Best learning rate=4e-32026.05 | 0.0336 |