Complete training-conditioned direct-source prediction
36/48 sector calculations passed.
Every combined range meets the unchanged twofold target: no.
| Candidate / removal | Complete | Fresh widths N / E / height (m) | Combined widths (m) | Combined target |
|---|---|---|---|---|
| Globe with specified field / zero | No | unavailable | unavailable | No |
| Globe with specified field / profiled | No | unavailable | unavailable | No |
| Globe with zero field / zero | Yes | 0.7666 / 0.7998 / 0.5658 | 0.7666 / 0.7998 / 0.5658 | Yes |
| Globe with zero field / profiled | No | unavailable | unavailable | No |
| Flat disc / zero | Yes | 0.6459 / 0.2002 / 0.3924 | 0.6459 / 0.2002 / 0.3924 | Yes |
| Flat disc / profiled | Yes | 0.7440 / 0.2370 / 0.4865 | 0.7440 / 0.2370 / 0.4865 | Yes |
Elapsed 5091.8 seconds. Charged 336/336 bounds and 4838400/4,838,400 logical bound steps. The fresh range unions every surviving sector. The combined range intersects that union with the independently qualified 192-box union; both remain visible. Missing or failed sectors prevent qualification.
No new native predictions or original-aircraft reads ran. The same six generated 7.2-second histories, original physical bounds, observations, calibration references and 78 coordinates remain. All old constraints are preserved. This is a method-usefulness test, not calibrated observational coverage or a full-recording aircraft validation.