[Clear1nsight]

Change and mathematical scope

Adds one opt-in research package, research/shared-crossing-cut9047-20261025/, proving and replaying a scoped obstruction to an original shared-helper hand-off proposal. No new physical word, paid-profile gain, κ ceiling or funded κ is claimed.

The actual L300/L301 helpers read the same L19 historical value at forward events 9045/9047, frame 21. At the original unmodified cut, the complete coframed inventory is L19/L300/L301. Neither pure original helper input is in their rational scalar span, even with two arbitrary independent successors and zero scratch. Two exact input variations witness this: (L300=u,L505=-4u) and (L301=u,L226=u) preserve every available row but change the required pure output. A separate retained-anchor intersection argument leaves only the existing frame 21 in the chosen first window.

The result does not exclude an earlier prefix rewrite, changed helper basis, other cuts/anchors, operator-valued ordinary recoding, new transport or a larger shared word. Witness admissibility under a different caller's restricted/owned input domain is unproved. Protected suffixes, owned returns, phases/raw dual, complete caller and compiler/outer funding remain open. The tentative 2H1+2H8-2H9 reward is unadmitted for this rejected immediate-purification premise.

Task and overlap

User-assigned scope key: shared-crossing-L19-L300-L301-cut9047. Catalogue provenance: complex-new-helper-inventory, changed scope / negative-result; the original card's positive complete-supplier acceptance gate is expressly not claimed met. This isolates a necessary physical implication before further construction.

Base is canonical CrocSwap/integer-mult-bounds:main, commit 32890bc5de034ab8fc93ac88ead7279a6a416d04. Native source Git blob/SHA-256 and exact catalogue/selected provenance are in SOURCES.json. No pending-PR code is imported or required by replay.

Nearest work: #5472 (f560065151f924838a80ac720fb4c55c6ffc0201), successor hand-off and local re-flag neutrality; #5541, actual-word balanced-J verification with coded-main restrictions; Discussion #5470/#5489, released helper300 transport scope. This tests a genuine two-helper slice and dirty-input separation, rather than repeating a gain-only tariff or padding a baseline. Visible handoff/proposal Discussions and relevant PRs were checked; hidden/cache/unpublished coverage remains incomplete, so no novelty/exclusivity claim. Paused peer scopes were preserved and not resumed.

Validation

Python 3.12, standard library only:

python3 -B scripts/research_package.py --package research/shared-crossing-cut9047-20261025 --check-only
python3 -B scripts/research_package.py --package research/shared-crossing-cut9047-20261025
python3 -B scripts/check_research_commit.py --package research/shared-crossing-cut9047-20261025 --check-source-history

• Metadata/source pins passed; committed-byte/source-history preflight passed in the isolated publication checkout. The published tree is byte-identical to that checked tree and has canonical base as its parent.
• Portable registered replay passed under the shared-lock guard in 1.31 s, sampled peak about 104.4 MiB, within 150 s / 512 MiB. It runs the unchanged independent checker in normal and optimized subprocesses and compares complete bytes to the original exact evidence, with zero scientific exclusions. It also checks the paper separators, source input charts and retained-anchor sets.
• Original qualification: all 632 persistent input columns plus the arbitrary owned entry; rank 3→4 for either target, or 5→6 after successors. Complete mode outputs matched; seven declared checks passed; 41 pinned local files unchanged; guard 0.906 s / 81.0625 MiB. Source checker SHA-256 f5c45a13239ea521140920c016e2436195580f78112a45240adaba314041cd12; complete output 70a30ac467a8d9a2b15e54d88b519258ef5bfff98fed11832a208c31c4fd0c81.
• The duplicate-owner control is a synthetic distinct-label sanity check, not full candidate target-admission coverage. Independent paper witnesses do not rely on it. Mode equality is reproducibility, not a second independent mathematical implementation.

No full new-word, phase/raw-dual/caller/compiler/outer replay was run. Hosted CI and scientific/maintainer acceptance remain separate from these local checks.

Credit

Cao Yifan/#4610; zxv/productive helpers and native semantics; GamingPuzzled/Joel Pulikkan/#5472; Cinna Davis/scalar semantics; Matarener/scoped boundaries; aliromman/#5383 owned relations and later audits; Dillon Amburgey and Clear1nsight/Chris R/provider-completion and #5413 lineage, with further predecessors retained in package references.

Prepared by Chris R (Clear1nsight) with substantial OpenAI Codex assistance: GPT-6-astra/xhigh paper design; GPT-6.1-sol/high event extraction, independent checker/physical and invoice review, custody and guarded execution. Agent independence is not independent human or formal admission. New material is Apache-2.0; the native input is not copied or modified.

Open next questions

See structured questions in TASK.json and the seven-field HANDOFF.md:

1. Specify a literal earlier prefix/basis change that removes these exact witnesses while preserving all required outputs and owned relations. First step: changed events, input domain, target/cost transfer and one bounded falsifier. Blocker: no such word supplied.
2. Specify relocated anchors and a genuinely shared crossing with its full transport/restoration bill. First step: actual anchors, historical deadlines and paid operations. Blocker: fixed-anchor geometry excludes it in this window; the changed geometry and complete costs are unproved.

The round stopped after this one result. This submission does not start another search, change the selected frontier or authorize merging.

[Clear1nsight]

HANDOFF — actual L19/L300/L301 shared-crossing qualification (no new κ)

Research discussion / PR / artifact: Source-bound follow-up to #5472. Public replay/proof package: shared-crossing-cut9047-20261025. This is a scoped negative result, not a new supplier or frontier claim.

Exact commit: Artifact d91b568b54bba63f1c15a34560afb78b0c7b02cf, directly based on canonical 32890bc5de034ab8fc93ac88ead7279a6a416d04. Conceptual predecessor #5472 head f560065151f924838a80ac720fb4c55c6ffc0201. Live native source and catalogue/selected provenance are pinned in SOURCES.json; no pending-head evaluator is imported.

What is established: L300/L301 read the same L19 historical value at forward events 9045/9047, frame 21. At the ORIGINAL unmodified cut, the only coframed owners are L19/L300/L301. Pure e300/e301 are outside their rational scalar span, even with two independent successors. Exact witnesses (L300=u,L505=-4u) and (L301=u,L226=u) preserve every available row but change the required pure target. The retained next anchors intersect only at frame 21. An independent stdlib checker and complete evidence are supplied; normal/O outputs agree exactly. The proposed split reward receives no credit for this local premise.

What remains uncertain: Earlier prefix rewrites, changed helper bases, other cuts/anchors, operator-valued ordinary recoding, new transports or larger shared words. Witness admissibility in a different caller's restricted/owned input domain was not checked. Full suffix/owned return, phases/raw inverse transpose, changed relations, complete caller costs and compiler/outer funding remain open. No complete profile gain, κ ceiling or funded κ is established; the local matched re-flag identity is not promoted to a universal shared-cost barrier.

Smallest useful next task: After checking current contributor scope, specify one literal prefix/basis rewrite or relocated-anchor recipe that changes the exact witness conditions, with all input/target/cost transfers and one bounded falsifier. A novel unproved paper recipe is eligible; a successful upstream construction is not a prerequisite. Repeating unchanged scalar purification at this cut with more samples or coefficients cannot defeat the exact span result. Our round stopped; no ongoing claim on such a redesign is made.

Reproduction command and expected runtime: Python 3.12/stdlib, from repository root: python3 -B scripts/research_package.py --package research/shared-crossing-cut9047-20261025. Registered timeout 150 s; original exact pair 0.906 s / 81.0625 MiB; portable registered replay also passed locally. Expected PASS_SHARED_CROSSING_CUT9047_SCALAR_OBSTRUCTION. Metadata/source pins and committed-byte/source-history preflight passed. Agent review and local replay are not human/formal admission; no full caller/compiler replay was run.

Predecessor credit: Cao Yifan/#4610; zxv/productive helpers/native rules; GamingPuzzled/Joel Pulikkan/#5472; Cinna Davis; Matarener; aliromman/#5383 and later audits; Dillon Amburgey; Clear1nsight/Chris R and the retained provider/completion/controller lineage. Substantial OpenAI Codex assistance (Astra design, separate Sol checker/physical/invoice/custody/execution roles); new material Apache-2.0.

{
  "schema": "integer-mult-coordination/v1",
  "kind": "HANDOFF",
  "task": "complex-new-helper-inventory",
  "scope": "Original-prefix L19/L300/L301 cut9047 frame21 rational-scalar purification obstruction and retained-anchor restriction; no new kappa",
  "base": "32890bc5de034ab8fc93ac88ead7279a6a416d04",
  "expiry": null,
  "artifact": "https://github.com/Clear1nsight/integer-mult-bounds/tree/d91b568b54bba63f1c15a34560afb78b0c7b02cf/research/shared-crossing-cut9047-20261025"
}