Test prediction ranges with the coupled calibration map

Reuse all 48 saved orientation sectors and their six generated cases from the completed joint-product prediction comparison. Preserve the original native expression, every coefficient, physical bound, error radius and outcome. Add only necessary relationships from the qualified complete static map.

Each sector constructs its original source reference and verifies the complete saved report. Bound the map on the existing contracted physical box, then at its fixed midpoint. Form 48 inequalities involving the original 51 coordinates, padding the 27 arithmetic-product columns with zeros. No generated truth or future outcome determines the box, midpoint or error assumptions.

An independent exact calculation must reconstruct every exported coupling row from the stored gradient intervals before any contraction or LP. Keep all original training constraints. Apply one interval contraction with at most eight passes, then compute the six signed future supports. Reuse old certificates by padding their multipliers; audit the six new endpoint certificates independently. Do not repeat old baseline audits or hide additional support work.

Finite ceiling: 48 identities, 48 source setups, 96 interval-map evaluations, 48 coupling audits, 48 contraction runs, 288 one-second LP proposals and 1,152 exact support checks. Contractions retain 78 columns and at most 1,024 rows. Four workers use one numerical thread each; 480 cooperative seconds. No retries, new native predictions/bounds, original-aircraft reads or observed fits.

Freeze sources, inputs, environment, per-operation reservations and source qualification receipt before launch. Preserve empty and unsupported sectors. After ranges are fixed, verify retention of every sealed generated truth in its declared sector. Require complete correctness and the unchanged twofold width target on every axis of all six groups relative to the original projected 7.2-second baseline. Report the immediate previous widths too.

Passing this comparison would establish usefulness only for these generated histories and declared numerical assumptions. Flight-length uncertainty, observational covariance and withheld aircraft prediction remain separate gates.