Skip to content

Prove exact kind A last-call finishability (no κ) - #7649

Open
petalas wants to merge 2 commits into
CrocSwap:mainfrom
petalas:research/w5-1117a-kind-a-last-call
Open

petalas wants to merge 2 commits into
CrocSwap:mainfrom
petalas:research/w5-1117a-kind-a-last-call

Conversation

@petalas

@petalas petalas commented Nov 17, 2026 •

Copy link
Copy Markdown
Contributor

Summary

Scoped written proof, produced by Nick Petalas. In the formal frame game at any address width m ≥ 3, take a kind A shared-middle part with nonzero rational coefficient c and a final call that raises its plain term to level three. Assume the extra term is outside the raw frame. With free preparation and addition of earlier outputs, the call finishes the part exactly when both terms have the same frame and (-c)^ord(γ) ≠ 1, where γ is the free element defining the extra term.

The proof projects the output span onto one coset of the raw-frame stabilizer. Distinct frames leave a forced pair of coefficients. On a common frame the surviving columns give a cyclic binomial inversion test. This supplies the all-width proof left open by jacobalansussman's finite result. Source question and finite evidence in #7629 · Advance claim.

Evidence and reproduction

CPython 3.12.15 on arm64 macOS. The direct verifier and the registered replay passed with PASS_KIND_A_LAST_CALL_COSET cases=8 post_call_control=1. The eight exact cases use five and six address bits. A separate free post-call control changes one excluded result to an attainable one. Committed bytes and source history passed.

Run from the repository root. Any CPython 3.12 executable can replace the local path.

/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -I -W error -B research/kind-a-last-call-coset/verify.py
/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -B scripts/research_package.py --package research/kind-a-last-call-coset
/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -B scripts/check_research_commit.py --package research/kind-a-last-call-coset --check-source-history

Limits

The result concerns one part and one last-call output. Joint output decoding, a free operation on the last-call output, removability, and Pauli phases lie outside the theorem. The package does not run a native caller or a formal proof kernel. It supplies no supplier admission or κ claim. The mathematical contribution is confined to one research package.

Open next questions

  • For a distinct-frame part that finishes after a free operation on the last output, must the two terms acquire the same frame after the call? Start by projecting the newly available free translates.
  • Can joint decoding of the two last-call outputs evade the separate obstruction? Start with both paid column families for one three-direction instance with a spare bit.

Related open work

  • #7629 posed the question and gives finite frame-game finishability and removability tables. This package proves the single-output kind A finishing criterion.
  • #7551 proves a cut-ratio ceiling on the same frame graph. The present result uses formal group coefficients in one last-call span.

CI notes

This PR carries the shared CI-doc and replay-gap sync used by other open PRs.

All three hosted checks passed at f642d2740265a61f8ac64b3a087f0118382c56e3: plan and the registered replays of research/historical-selected-replay-gap and research/kind-a-last-call-coset. GitHub Actions run. All six current-status tests and the historical replay-gap package passed locally. The kind A package and committed-byte/source-history checks also passed at this head.

On the original base, test_current_ci_section_matches_selected_record expected make frontier7000-verify while the current CI guide named #6030. The historical verifier now replays against its recorded source instead of the live Makefile. Earlier planning log with the #6030 mismatch.

Project the formal one-output span onto a stabilizer coset to isolate
distinct-frame contamination and derive the same-frame binomial test
at every address width.
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