One-step check of training-compatible alternatives
4/4 numerical diagnostics complete.
These generated probes preserve the original five noisy training observations and physical bounds. One linear repair is proposed per preselected seed. Only the subsequent nonlinear training check establishes feasibility.
| Processing | Feasible new points | Sampled N spread | Sampled E spread | Sampled height spread |
|---|---|---|---|---|
| Zero removal | 0/2 | 0.000 m | 0.000 m | 0.000 m |
| Profiled removal | 0/2 | 0.000 m | 0.000 m | 0.000 m |
Spreads include the original truth. These are sampled separations, not outer bounds. The original 20/20/50 m width targets are unchanged.
The proposal deliberately constrains its computed future north prediction to remain near the seed. It uses no future observation. This is a counterexample search, not an unbiased held-out prediction test.
All failures are preserved. There are no iterated repairs, new aircraft fits or changes to existing studies. No statistical confidence level or aircraft error law is established.