The four missing computational cases pass

All four new fits, both start comparisons and the remaining original-input replay pass the independent audit. The separate run finishes in 4.6 minutes after 364 source-bound software checks. It keeps the original inputs, starting points, physical bounds, all 26 or 27 nuisance columns and numerical limits.

The preceding matrix retains its 44 passing fits and four interrupted starts. This follow-up does not change those labels or charges. Its four starts have a separate source/environment freeze, prepayments and deadline; there are no retries. All 241 source files shared with the preceding matrix remain unchanged.

The new run precharges 239,816 native calls and 1,506,051,012 packet visits. Its actual work is 238,288 native calls, 151,343,676 visits and 44 linear operations. Every result and charge is retained. Publication verifies saved inputs, forecasts, timing checks and repeat agreement without fitting again.

The next gate is a separately tested, saved-only audit of the combined evidence: 44 earlier completed cases plus these four, covering all 48 cases, all 24 start pairs and all four original bindings exactly once under identical shared sources and environment. Until that audit qualifies, full matrix qualification remains open. No calibration threshold, aircraft confidence claim or independent confirmation follows from these software controls.

Verified report · Four fits · Start comparisons · Frozen plan · Previous interrupted matrix · Next analyses.