Chaotic system forecasting on Lorenz-63
1.2VPT (0% noise)MP-NODE
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| MP-NODEValidation threshold (epsilon)=0.3, Evaluation trials=302025.11 | 1.2 | 1.1 | 0.86 | 0.45 | 0.2 | 0.02 | 0.04 | 0.18 | 6.67 | 11.14 | |
| RAFDAValidation threshold (epsilon)=0.3, Evaluation trials=302025.11 | 1.51 | 3.13 | 2.62 | 1.91 | 1.16 | 0.23 | 0.01 | 0.002 | 0.02 | 0.06 | |
| Strong NODEValidation threshold (epsilon)=0.3, Evaluation trials=302025.11 | 2.55 | 1.88 | 1.39 | 0.8 | 0.36 | 0.001 | 0.006 | 0.02 | 0.23 | 6.82 | |
| WP-NODEValidation threshold (epsilon)=0.3, Evaluation trials=302025.11 | 2.7 | 2.77 | 2.9 | 2.23 | 0.96 | 0.006 | 0.002 | 0.002 | 0.01 | 0.03 | |
| Weak NODEValidation threshold (epsilon)=0.3, Evaluation trials=302025.11 | 3.19 | 3.29 | 2.77 | 1.6 | 0.76 | 0.019 | 0.002 | 0.001 | 0.003 | 0.41 | |
| DeepSkipValidation threshold (epsilon)=0.3, Evaluation trials=302025.11 | 4.47 | 1.54 | 0.56 | 0.34 | 0.11 | 0.02 | 0.05 | 10.8 | 12.19 | 18.82 |