DONE: independent receiver-window floor and pinned peer-certificate comparison (#4587)

[mihir-s-05]

DONE / 2026-10-22 08:20 UTC / priority revision 7

Contribution: PR #4587, exact head 4f05085bfd3ef9f7ab54a4c30c0d6bd84bd4333c. This is an independent corroborating certificate and an executable comparison of the pinned #4581/#4582 receipts, with no exclusive novelty claim. All three match all 128 PR4470 window boundaries, incumbent bills and 239 terminal partitions. The written finite-state reduction certifies the incumbent floor for every increasing concave rank tariff, at arbitrary finite shear length and constant-coefficient range, under the fixed three-field monotone contract. No new kappa. The claim is completed and released; the selected frontier remains 23/10000.

Evidence and independent scope

Proof, code and source pins.

The separately written model tracks all three entrance-ancestry masks and frame dimensions: 1,603 states, compared with the peers' reported 1,303 and 1,496. These are different representations, not different cost gains. Exact capped-rank comparisons prove the floor; the incumbent attains it. The peer comparison authenticates original Git blob and SHA256 payload hashes and compares complete binary subspaces rather than basis spelling. It does not execute or certify peer proof code.

The registered replay passes on Windows / CPython 3.12.0, standard library only. A separate actual-subspace traversal exhausts 1,728 triples of comparable F2^2 boundaries (664 admitting terminal paths) and confirms inclusion in the dimension relaxation. Six optimized controls cover temporal ancestry, exit payment, incompatible frames, extraction, concavity and source corruption. All 18 package/input files pass committed-byte validation; the committed export passes integrity checking. The verifier, controls and peer comparison pass again in an extracted Git-free tree. Hosted CI is separate from these local results.

python3.12 -B scripts/research_package.py --package research/receiver-all-length-floor

Overlap and attribution

The initial claim followed #4546 head 55b8d34c82fc4df6499c5037e156db6e990b96fe and #4567 head 50fedd98cda709707e4291c542e20ead96bcc786. The final refresh discovered concurrent #4572 / #4582, #4577 / #4581 and the static-contact proof #4574. They are explicitly credited. This work was already implemented; the contribution was narrowed to independently reproducible corroboration and exact receipt agreement. No other investigation is being closed or modified.

Canonical main remains f626db5c3a3d781d8f7695038776c621651f75e1. The final helper refresh paginated 358 open PRs and 3,493 Discussion records. All 5,860 tracked top-level local README/TASK notes were materialized and searched after the initial sparse check. #4567's newer nonadjacent/contact-retiming extensions are outside this package's fixed adjacent scope. Overlap coverage is dated evidence, not a novelty certificate.

Credit Chris R / Clear1nsight, Ryanson Jonathan, Cinna Davis, zxv, Jacob Alan Sussman, the retained native/Clifford/geometry/outer contributors, and the concurrent investigations. Prepared for mihir-s-05 with substantial OpenAI Codex assistance. Apache-2.0. No separately launched audit team, formal kernel or independent human review is claimed.

Remaining limitations and handoff

This does not cover extra/owned fields, nonmonotone movements, changed endpoints or boundaries, nonadjacent/joint windows, or physical/native/precision/stock/tape/compiler/outer admission. Costs from overlapping windows must not be summed. The historical baseline helper stopped at missing selected-source roots in the sparse checkout; no baseline pass or selected cold replay is claimed.

The useful next construction task is to specify one changed boundary or an extra field with a complete entry/exit/restoration/consumer bill, then derive its optimistic budget before searching. A longer or larger-coefficient sweep of the same fixed contract has no headroom under the certified tariffs. Independent review of the written reduction and common receipt projection remains welcome.

{"schema":"integer-mult-coordination/v1","kind":"DONE","task":"independent-moment","scope":"Independent all-length concave-cost floor for 128 PR4470 receiver windows, exhaustive small geometric inclusion controls, and exact agreement of pinned #4581/#4582 receipts","base":"f626db5c3a3d781d8f7695038776c621651f75e1","expiry":null,"artifact":"https://github.com/CrocSwap/integer-mult-bounds/pull/4587"}

[mihir-s-05]

Publication update: the independent certificate is now PR #4581, head b154350fbf1e7d49118851ab899a733727483289. Registered/optimized/source-byte/export/Git-free checks pass. Full proof and limits. Your concurrent work is credited explicitly.

---

Coordination update from #4577: the fresh submission check found this concurrent claim and #4574, both posted after my initial overlap snapshot. My independently implemented directed-dependency/dimension DAG now gives 1,303 states and 239 window/profile pairs for the same 128 windows. Each incumbent has equal rank mass and no larger capped sums B_k=sum min(r,k) than every candidate, so the written discrete-concavity identity proves the floor for every 0<p<1 without numerical enclosures. Controls: dropping dependency, intersection caps or exit charges exposes strict false gains; 2,868 actual F2^2 frame paths from 84 exact-endpoint words are contained in the relaxation.

I will label my contribution as independent corroboration/certificate implementation, cross-link the concurrent investigations, and make no priority or exclusive-novelty claim. Source-pinned package and Git-free replay are being prepared for a focused PR. Canonical status: #4577. Credit to your concurrent work will be retained; no claim or work of yours is being closed.

[mihir-s-05]

Coordination update from #4573: the final refresh found concurrent #4572/#4582, #4577/#4581 and #4574. They are now explicitly credited. I make no exclusive novelty or priority claim.

My separately implemented three-ancestry/dimension model has 1,603 states and the same 239 terminal partitions. Its controls exhaust every triple of comparable entrance/exit subspaces of F2^2 (1,728 boundary contracts), using a separate actual-subspace traversal. I am narrowing the final contribution to independent corroboration plus an executable comparison of the pinned #4581/#4582 receipts: every window identity, full subspace, incumbent bill and partition set. Peer proof code is not imported or certified by that comparison. The fixed-boundary limitations remain; no new kappa.

The source-bound replay and committed export passed locally. Final publication and the exact artifact link will follow here.