Qualify the conditional remainder in native propagation
The separately frozen comparison completed with all twelve controls passing after 63 focused software checks passed. Its fixed source, inputs, environment and allowance are recorded in its study plan. It does not use an existing study's allowance.
Use twelve fixed controls: three physical candidates × two processing assumptions × two force-constraint scopes. Each uses the first 800 packets of the existing generated history, retaining the original broad calibration box and ±1-second clock range. Predict at 2.5 seconds, whose complete clock support lies inside that four-second prefix.
The first scope uses the physical-domain force bound. The second adds a fixed synthetic force-change constraint of ±0.2 m/s² per axis and independently verifies its affine support certificate. This exercises a training-conditioned bound without choosing a favorable calibration point. Every predeclared probe must satisfy its stated domain; a failing probe is reported, not skipped.
Compare the unchanged native bound with a separate candidate applying only the proved remainder inequality. Preserve the original shared caps, coefficients, physical parameter directions, reference and native center. Require all map identities and source bindings to match, every bound component to remain no wider, and every known probe to remain contained. Require an actual remainder reduction so that an unused code path cannot qualify the comparison.
Proposed finite maximum: 12 identities, four native queries and two bounds per identity; 48 native queries / 38,400 conservatively charged packet visits; 24 bounds / 172,800 logical bound steps. Reserve complete source, derivative, point and direct streams before each bound. Conditional support work has its own explicit per-case proposal/verification counts.
Use one numerical worker/thread while reserving the running study's two threads. Estimate roughly 15–30 minutes, with a proposed 80-minute cooperative study ceiling and three minutes per bound. Final limits must be frozen before launch; no retries, process signals or later enlargement.
Passing these short controls would qualify only this component's native consumption. Recording-length usefulness, full-period prediction coverage and the observation-error law remain separate V01 requirements. All earlier results and the current 20 / 20 / 50 m targets remain unchanged.