[aliromman]

Constructive conditional ordinary BIT saving: 0.003487922466, an exact gain of 0.000001313377 over PR #5641. This emits and fully prices Dugongue PR #5639, while reusing the old zero centre to reduce the owned reserve from 17 to 16 banks. No complete kappa or selected-frontier update is claimed.

Scope, proof, exact checks, source pins, registered replay. Base: canonical main df0137efd35e11b4445734bf612bdd4c7ebc330c, priority revision 7. Contribution head: 8b1995f167471e79f77351fcdc014db3b338f818.

The snapshots preserve every retained arbitrary original and endpoint. All 546 obsolete whole-zero centre accesses are proved removable. The complete 18-call caller retains 11,166 originals, pays 56,817 queries, and allocates 16 owned banks with no output credit. Against PR5641 the paid change is -36 H1-36 H3+36 H4, with unchanged targets and strict gain 36*(1+3^t-4^t)>0 for 0<t<1. Discussion #5642 records the claim and scope split.

Normal and optimized registered replay pass. The native word tracks all 661 original/dirty entry columns; separate adjoint checks use other address points at moduli 101 and 10201. Separate stream replay checks 445,838 events, all 56,817 query descriptors, targets, owned returns and positive costs. Two census implementations check all 667 nested blocks on every retained field, with arbitrary prior same-frame binary combinations. A third eligible window at helper444 was handed to its active investigation, with a direct SOURCE77 snapshot replacing five gathers. It is excluded from this emitted bill. Two rational engines certify raw crossing (0.00348792248477,0.00348792248480), fresh finite thresholds, 16-bank stock and four acyclic ordinary levels. The admitted saving is less than 1.89e-11 below that crossing. Stale snapshot, missing teardown, stale epoch and unfunded price controls reject.

python3.12 -B scripts/research_package.py --package research/g20-snapshot-native

Python 3.12 and the standard library suffice. Committed-byte checks pass. The native/all-width, ROW, compiler, precision, tape and analytic premises remain conditional. Full historical suites and formal-kernel proofs were not rerun; separate implementations are not independent scientific acceptance. The finite uniform program and astronomical decimal threshold are algorithm-defined, not materialized. Other producers, protected-read retiming and paid frame changes are outside the frozen census. This shallow checkout cannot run the full ancestry audit. A reproduced upstream Current CI missing-link regression currently blocks the generic hosted planner; the package has complete local replay evidence. Hosted status is unconfirmed at publication.

Credit Dugongue PR5639 at head 4b4f615 for the two fusions; ai-nanachi PR5512 for native G20 and the ordinary framework; Ali Romman PR5421/5641; DreamingOfClouds, eumemic, Cinna Davis, rohanarun, zxv, Jacob Alan Sussman and all frozen source notices. The emitted caller, zero-centre reuse, full source-window census, exact ordinary certificate and direct snapshot handoff are the contribution here.

Next: complete helper444 in Discussion5638 using the direct snapshot handoff. A fourth fusion requires a changed producer or priced native frame map; the frozen 667-block census has no other candidate. Complete kappa still needs a compatible complex supplier and caller. This result reserves no further search.

Prepared by Ali Romman with substantial OpenAI Codex assistance; Apache-2.0.