Skip to content

Audit both factor-seven input columns in the chart-595 native graph - #7701

Open
aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-039
Open

aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-039

Conversation

@aliromman

@aliromman aliromman commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

Independent scalar audit: both split owners are needed to test the factor-seven divisions in PR #7697. All 630 literal basis products with a denominator divisible by 49 are confined to input columns 281 and 1501, with 60 and 570 products respectively, across 40 rows. The first old-owner witness is (-2/7)*(-40/21) = 80/147; the fresh-owner witness is (1/7)*(-40/21) = -40/147. A basis test that omits both owners misses this obligation.

Read the proof, exact receipt, checker and scope record. The complete 8738317-product replay independently confirms the source's reported minimum literal-product grid 45158400 and complete-snapshot grid 161280. The support census and separate causal witnesses make the new division obligation easier to review. The reported minima remain predecessor results.

No new kappa or supplier admission is claimed. This covers only normalized scalar coefficients on the pinned graph, with arbitrary independent entrances and one common scale. Physical Clifford maps, full root induction, early public installation, dirty-state movement, compiler funding and outer inequalities are outside scope. All five selected assumptions remain. This completes the advance claim under Clear1nsight's independent-review handoff; its active successor search is separate.

Normal and optimized CPython 3.12.16 checks pass on macOS arm64. Four maximal-prime witnesses and two split-owner witnesses are replayed by separate one-column Fraction arithmetic. Six under-grid controls stop at a real nonzero division remainder. Metadata, source history, committed bytes, modes, export and archive integrity pass. The read-only optimized archive replay has no Git or usable PATH and preserves all input hashes and modes.

With CPython 3.12 and its standard library, from the repository root:

python3.12 -I -W error -B research/chart595-factor-seven-audit/verify.py
python3.12 -B scripts/research_package.py --package research/chart595-factor-seven-audit
python3.12 -B scripts/check_research_commit.py --package research/chart595-factor-seven-audit --check-source-history --check-modes

Expected markers:

PASS_CHART595_FACTOR_SEVEN_AUDIT
PASS_REGISTERED_RESEARCH_PACKAGE
PASS_RESEARCH_COMMIT_BYTES

Hosted CI did not certify this package. The plan job failed in the shared test_current_ci_section_matches_selected_record: the base guide names frontier6030 while the selected record requires make frontier7000-verify. The log reports 760 tests, one failure and 24 skipped; package replay was skipped. This package changes no shared CI file. Local replay remains the scientific evidence. Full candidate and selected replays, physical/root checks, Lean and Docker were not run. Independent human review and formal-kernel acceptance are unclaimed.

Prepared by @aliromman with substantial OpenAI Codex assistance. @Clear1nsight supplied PR #7697, the exact graph and the grid claims in its pinned physical proof and grid checker. The graph and its notice are copied unchanged with exact origins. No source verifier is imported.

Ali Romman's PR #7680, pinned grid proof, supplied the gcd and product-versus-snapshot method and rational controls. This adapts that method; it is not blind review. Credits and the retained notice preserve the helper, caller and native lineage. New audit work is Apache-2.0.

Base: canonical main e54523517d0f03e58352e647c403a5af3e525a57, priority revision 8. Card uniform-transfer-audit, scope changed to this numerical sub-obligation. Current handoffs and refreshed open PR/Discussion overlap were checked. The complete closed snapshot and updated-order delta have no later changes. Sparse local and unpublished work remain coverage limits. A possible next task is a row-atomic integer accumulator with a proved numerator-width and ordinary-work budget; this audit does not establish that changed provider.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant