[Clear1nsight]

Change and mathematical scope

Independently qualifies 182 previously uncovered ordered native frame maps from PR #5545 head 18968423711fd902aaee1099979497d55603e1dc, including all 11 private returns. Combined with the retained receipts in PR #5548/PR #5550, this covers 224/2981 pairs and 254/3624 local paid occurrences. The residual is explicitly 2757 public rank 1–3 pairs and 3370 paid occurrences. This is an exact finite native audit, with a conditional all-width interpretation under the retained framework. The paid profile and κ are unchanged; no full parent admission is claimed.

Task and overlap

Base 32890bc5de034ab8fc93ac88ead7279a6a416d04; publication 998d312fc3af584d39a3cf4c10a45e02a8395f48. Scope is refined from the scalar-to-moment research task to a missing native verification layer. This complements PR5548 and PR5550, now merged. PR #5551's later four-new-map audit concerns its changed map universe and does not duplicate these 182 predecessor maps. This package does not certify automatic transfer to PR #5551, PR #5561 or successors. The current upstream review prioritizes those stronger constructions before assigning further old-head residual work.

Validation

Python 3.12/stdlib. Run python3 -B scripts/research_package.py --package research/p104-native-subset-independent-audit.

The original exact normal/-O pair passed in 24.449 s, sampled 86.71 MiB, with complete result SHA256 fa2fe7ce93e72569f611ca810f440e6b77d92221350c51b7b6ab19ca34d6c64b. The portable registered pair passed in 26.494 s, sampled 115.93 MiB, under the 150 s / 512 MiB guard. A separate Git-free read-only export with spaces in its path passed in 8.350 s, sampled 89.54 MiB, with empty PATH. All frozen package/export bytes remained unchanged. Metadata, source-history and exact remote-commit byte preflight passed for 41 package/input files.

Checks compare the actual native recipe against all 18 signed Pauli images and an independent exact dense column. Thirteen scoped controls comprise 10 operator/authentication faults and 3 structural sanity checks; they are not 13 complete source mutations. Static wrapper corruption/symlink/extra-file tests reject before scientific launch. Hosted CI is pending at submission. Independent human/formal acceptance, the residual low-rank maps, uniform precision/tape/stock and complete compiler/outer funding were not newly established.

Credit

Ali Romman for the P104 construction and PR #5544/PR #5545; Gareth Davidson, David Arias, Cinna Davis, Jacob Alan Sussman, zxv, Cao Yifan and retained native/controller/geometry/tensor/compiler contributors. Credit thu-nmrc's separate successor audit. New independent qualification/publication for Chris R/Clear1nsight with substantial OpenAI Codex assistance. Apache-2.0 and source provenance retained.

Open next questions

Rebind any reused map evidence to literal frame bytes, actual recipe and occurrences on the strongest current construction before further native work. Assess PR #5561/PR #5569 and the overlapping PR #5567 endpoint audit first. A separate frozen budget and nomination are required for any remaining old-head rank band. Native coverage does not improve economics or fund κ.

[Clear1nsight]

Seven-field research handoff

1. Research discussion / PR / artifact: PR5573, package research/p104-native-subset-independent-audit; independent high-rank/private native verification supporting PR #5545 and the retained PR #5548/PR #5550 receipts.
2. Exact commit: publication 998d312fc3af584d39a3cf4c10a45e02a8395f48, tree cc4d85bb8ea9c5ee0fa857a078e6bbf5cec63c73; original parent 32890bc5de034ab8fc93ac88ead7279a6a416d04, current PR base 5e0052ca0aa3036a3afa8b0438927377c8d97d53. Reviewed construction is PR #5545 head 18968423711fd902aaee1099979497d55603e1dc; original checker SHA256 c5b62fee065d6d7029e613e5be45e9e429cf2274a3f557b0d403676c557143c2.
3. Established: 182 newly checked ordered maps covering 246 local paid occurrences, including all 11 private returns. Combined exact-source coverage is 224/2981 maps and 254/3624 paid occurrences. Complete normal/-O scientific output agrees, SHA256 fa2fe7ce93e72569f611ca810f440e6b77d92221350c51b7b6ab19ca34d6c64b. Ten operator/authentication faults and three structural sanity controls reject. No changed paid profile, economic gain or new κ.
4. Uncertain: 2757 public rank 1–3 maps / 3370 paid occurrences remain. All-width framework/owned erasure, uniform precision/tape/stock, complete compiler/outer funding and independent human/formal acceptance remain separate. Coverage does not automatically transfer to PR #5551/PR #5561/PR #5569. Hosted CI is pending at this handoff.
5. Smallest useful next task: bind actual source frame bytes, recipes and occurrences on the strongest successor before assigning residual native checks. PR #5551 already audits its four added maps; PR #5567 covers a separate endpoint-contract question. Assess PR #5569 first and avoid repeating those checks. A new old-head rank-band audit needs its own frozen budget and nomination.
6. Reproduction / runtime: Python 3.12/stdlib: python3 -B scripts/research_package.py --package research/p104-native-subset-independent-audit. Original pair 24.449 s / sampled 86.71 MiB; portable registered pair 26.494 s / sampled 115.93 MiB under 150 s / 512 MiB; separate Git-free read-only export with spaces and empty PATH passed in 8.350 s / sampled 89.54 MiB. Metadata and exact published-byte/source-history preflight passed for 41 package/input files. Portable and original expected bytes match; no source suite was silently repaired.
7. Credit: Ali Romman, Gareth Davidson, David Arias, Cinna Davis, Jacob Alan Sussman, zxv, Cao Yifan and retained P104/native/controller/geometry/compiler contributors; thu-nmrc for the distinct successor audit. New independent audit/publication for Chris R/Clear1nsight with substantial OpenAI Codex assistance; Apache-2.0 and predecessor provenance retained.

{"schema":"integer-mult-coordination/v1","kind":"FINDING","task":"p104-native-subset-independent-audit","scope":"Exact PR5545 head 18968423711fd902aaee1099979497d55603e1dc: 182 new native pairs/246 local paid occurrences, all 11 private returns; combined 224/2981, unchanged economics; no automatic successor transfer","base":"32890bc5de034ab8fc93ac88ead7279a6a416d04","expiry":null,"artifact":"https://github.com/CrocSwap/integer-mult-bounds/pull/5573"}

[Clear1nsight]

Hosted validation has now passed at exact publication head 998d312fc3af584d39a3cf4c10a45e02a8395f48: run 40228913317, planning and the registered package replay. This is scoped package validation, not complete parent admission.

Overlap update from the fresh upstream checkpoint: Matarener's PR5572, head dba3cf67c800f32f04c80437138a51310dc08c1b, supplies a much broader PR #5551 native/caller/funding audit. Its committed full-run receipt contains 3188 operator task records; the default registered mode freshly rechecks 344 operators against that receipt. We have inspected the source/receipt boundaries, not independently replayed its full audit. Credit this work before assigning further native coverage.

This PR retains its exact PR #5545 source domain and 182-map independent-checker evidence. It does not certify transfer to PR #5551 or later endpoint changes. The useful next step is source/recipe/occurrence-bound reuse of existing receipts on the strongest successor, rather than automatically extending the old low-rank band. No new κ is claimed.