SURVIVES_LOCAL_GEOMETRY_CONTROL
Real mean D□ = 0.0266275
| Control | Null mean D□ | Fav z | p | Quality |
|---|---|---|---|---|
| N0 | 0.996786 | 1022.66 | 0.00990099 | FROZEN ROUTE-I CONTROL |
| N1 | 0.329091 | 106.404 | 0.00990099 | PASS |
| N2 | 0.223649 | 44.6945 | 0.00990099 | PASS |
Eligible source objects: 4957
Geometry-conditioned partial rank ρ(D□, enstrophy) = 0.179019
Geometry-conditioned partial rank ρ(D□, dissipation proxy) = 0.120047
Partial-rank values condition on scale, time, route-overlap geometry, distance/feature compatibility, and branch/merge flags. They test whether physical intensity retains association with D_square after those controls.
The null hierarchy tests whether low D_square survives increasingly local route controls available from the frozen object/edge tables. N1/N2 preserve degree and route layers while constraining recomputed centroid geometry; N2 also constrains enstrophy-based feature similarity. They do NOT preserve exact voxel overlap/Iou because voxel masks are not part of the frozen ROUTE-I run. Survival therefore rejects these graph-local geometric explanations, not every possible field-level geometric surrogate.