Manuscript En06-B@1 · En06-B.tex · section · ee886ae8

What has been repaired

Debugging audit and exact mathematical release › What has been repaired · lines 2045–2098

Potential gap	Repair proved here
Endpoint-only routing	Rounded affine trajectories satisfy the every-prefix estimate (eq:generatorprefix).
Boosted idle-wire pacing	The entire transformed causal segment is sampled; all route, wait, and access support is estimated.
False global lane separation	Disjoint residue classes give disjoint actual site sets, including in one dimension.
Unproved junction capacity	An explicit finite swap–payload–swap frame is charged $B τ_{n}$ physical time.
Directional service delay	Each long wire has its own site residue and moves in parallel; no factor proportional to the number of wires slows all routes.
Complexity-dependent coarse error	The coarse coefficient is $2 L_{g}$ ; graph-dependent constants occur only in a vanishing fixed-source fine error.
Declared versus active support	Both physical Hausdorff convergence and the separate declared support construction are stated.
Report-law versus instrument equality	The complete logical channel is equal at every sufficiently fine scale, with arbitrary references.
Channel adjoints versus algebra maps	Observable comparison is a faithful $*$ -isomorphism on an explicit dilation subalgebra, not arbitrary compression.
Shared backward ancestors	Microcausality is proved at a common tensor-factor cut.
Input-independent sampling	The Bell instrument acts on arbitrary inputs and retains an exact quantum-memory channel.