Repository navigation
Exclude a historical C5 caller with a common-cut23 bound - #3931
Clear1nsight wants to merge 3 commits into
Conversation
Carry the focused change from 89031d2 (PR #8) onto this active branch. Initialise TMPDIR after runner setup, preserving workflow triggers, repository guard, permissions, pinned actions, timeouts, replay commands and all research inputs. Validated YAML, exact semantic delta, shell syntax and TMPDIR assignment including spaces. Substantial OpenAI assistance.
|
Publication verification at 2026-10-19 04:47 UTC: current head ce148e5 has the exact tested research bytes; the only delta from scientific commit af26d5b is the existing P112 workflow repair. Both 17-file publication/source-history checks pass. All three mathematical/control scripts passed from the committed read-only Git-free export, with all 18 exported hashes unchanged. Upstream run39200317292 passed planning and selected exactly this package; on the last live API/browser inspection its replay job was still fetching repository history, before the mathematical replay. Hosted scientific success remains unclaimed. The canonical C5 finding is published and the claim released. Independent review and the separate 4,5^3,6^6 caller remain the next tasks; no further census or frontier change is proposed. |
…lusion Add a code-disjoint table/Fourier check and detailed written proof audit. Reconcile the literal 4,5^3,6^6 row with Nick Petalas's existing PR3295, giving conditional69/69 coverage only in the named historical inventory. Preserve all original scientific and predecessor bytes. No full C5, supplier, kappa or selected-frontier change. Fresh registered audit/controls and original PR3931/3295 registered replays pass. Substantial OpenAI Codex assistance; same-author audit, no independent human/model or proof-kernel acceptance.
1 participant
Change and mathematical scope
Exclude the literal historical P112 independent-piece caller
2,3,4^2,7^6at code-intersection dimension at least six. For ONE common set of at most23 live sites, at most THREE projective directions have residual binary support rank at most7, over every characteristic-zero field using the standard E8 Weyl-chamber theorem. The caller needs six independent directions under a common22-site cut. Live root normals each need at least21 deletions, while four distinct normals share at most14 off-sites: four qualifying directions would require at least421-323=15, contradiction. A two-direction fixture works at22; attainment of three and sharpness of23 are unclaimed.Follow-up audit and correction: Nick Petalas's PR3295 already conditionally excludes
4,5^3,6^6through a six-piece concentration theorem. The earlier suggestion to investigate it as the next gap overlooked that public predecessor. The source-bound composition is now 69/69 in PR3289's fixed named historical inventory, conditional on its inherited contracts. The last caller exclusion belongs to PR3295. This is not a global unresolved count or full C5.Detailed fresh proof audit finds no objection to the characteristic-zero/common-set/quotient argument and explains PR3295's coefficient-column and concentration proof. A new table/Fourier implementation imports no predecessor executable and reproduces the complete flat and fourfold histograms. This is same-author AI-assisted checking, not independent human/model or proof-kernel acceptance.
Exact changes and sources
Three commits,20 changed files:
Canonical main20ecf2891d80b5acf18d980b2164fb44ea7faa90 remains unchanged at the publication checkpoint; priority revision6. The audit's pinned base is ce148e5. Relevant predecessor heads: #3163 8941910; #3289 e8570c5; #3322 131605e; #3405 4c5e080; #3295 4cff812. All16 PR3295 file Git blobs match current canonical main. Its ledger status is partially reviewed; archival intake is not scientific approval. Runtime needs no foreign pending Git objects.
Validation and CI
CPython3.12.3/Linux x86-64, standard library only:
All three registered package replays passed this follow-up session. PR3295's primary, separate generating-basis check and ten behavioral groups passed. Its inherited global12/20 and all-A2 joint22/32 premises were authenticated and inspected, not independently reproved or replayed through their complete source chain.
The fresh audit checks43,435 binary seven-spaces,5,212,200 root/space pairs, all8,214,570 fourfold intersections and both full histograms. Character orthogonality replaces the old bit-mask flat count. Four literal primal supports check all43,435 dual spaces (173,740 comparisons); four semantic faults and three actual optimized CLI corruptions are rejected. The direct fourfold census is corroboration, not a fundamentally new method for that bound. The original package retains its separate lattice/RREF/pair-join checks, common22-site positive fixture and separate/common-cut controls.
The exact audit commit passes the18-file published-byte/source-history check and official export integrity check. Both actual audit commands pass from a committed read-only export without Git metadata under -I -O -Werror -B:4.259s and10.343s, all19 exported hashes unchanged. These are single-run timings, not benchmarks. Syntax/JSON/whitespace checks pass. Ruff crashed locally before diagnostics in the original session; no lint pass is claimed.
Original head ce148e5 passed hosted CI: upstream run39200317292, completed2026-10-19 at04:48:12UTC. Both plan and registered replay succeeded; logs contain all three package PASS markers and outcome=completed. This supersedes the earlier pending status. The earlier fork scientific push failed before jobs on the inherited workflow parse issue; no mathematical failure is inferred from it.
Current audit head33b6be4 also passed hosted CI: upstream run39206283413, completed2026-10-19 at05:50:03UTC. Plan and both registered replay jobs succeeded. Logs include the new Fourier audit/control PASS markers and the original theorem package's three PASS markers, each with outcome=completed. The committed-input checks passed. This confirms execution of both registered packages, not independent human/kernel or full-C5 acceptance.
Task, overlap and next obligations
User-assigned C5 application/proof audit, explicitly changed scope from complex-rank-shape-decoder; task cards record the decision. Latest entry scan covered643 open PR bodies and final scan647, with seven full REST pages plus an empty eighth, including drafts. Repository overlap helper also searched2,487 local records and surfaced PR3295 plus the later mixed566-cut32 package that explicitly credits it. Browser-loaded latest C5 replies and handoffs were inspected. Discussion coverage remains partial: all-board refresh is unavailable without gh; arbitrary diffs, all closed bodies and unpublished work are nonexhaustive. No novelty certificate.
Canonical investigation. No continuing reservation. The broader uniform normal-plane question remains unsettled, but this conditionally complete inventory supplies no gap requiring it. Before another census, bind a different literal TWOSIDE caller, quotient, common-set budget and needed span bound. Independent review of the Weyl/scalar-extension argument and inherited global/A2 premises remains useful.
Full C5/FINITE, all inherited successor scientific suites, selected cold replay, physical/native/restoration/consumer/supplier/all-size admission, upstream manual P112 execution, Lean, Docker and independent human/kernel acceptance remain unrun or unestablished. No new decoder, paid supplier, kappa or selected-frontier change.
Credit and merge danger
Credit DreamingOfClouds/#3154 for C5/TWOSIDE/FINITE; Nick Petalas/#3160/#3161/#3163 for geometry and quantifier controls, #3282/#3289 for accounting, #3253/#3322 and #3397/#3405 for distinct exclusions, and #3295 with #3278/#3267/#3260/#3262/#3269 for the already existing last caller. Retain #3283's related concentration credit, #3286's counterexample, joaqcoo and the E8/P112 lineage. Chris R / Clear1nsight with substantial OpenAI Codex assistance; Apache-2.0 and predecessor notices preserved.
Door: two-way. Blast radius: two isolated research packages and the separate existing one-file P112 runner-context repair. No selected record or frozen input is changed; maintainer edits remain disabled. No merge or closure is requested by the audit.