Skip to content

Prove that a common free post-map cannot rescue kind B finishing - #7684

Open
aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-027
Open

aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-027

Conversation

@aliromman

@aliromman aliromman commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

Result and scope

One common free post-map adds no finishing power when a shared part has a unique level-two leading term and only lower-level companions. If one prepared last call followed by that map reaches the level-three target, projection onto the target frame forces the map to be a unit of Q[P intersection TPT]. It can be moved before the call.

This extends the nonplain kind B obstruction in PR #7683 to arbitrary common free postprocessing, at every address width and every nonzero rational coefficient. It resolves the kind B post-call boundary left untested in PR #7629. The post-map is not assumed invertible; the proof derives that fact.

Read the proof, exact receipt and matrices, verifier and scope record. The proof also gives an explicit false positive for a relaxed column-span test: TM=T y-c R T(SD) with R,S free. This uses two independently prepared and postprocessed call outputs. The theorem proves that one common-map call cannot realize the target. A kind A example with a second level-two term supplies an actual one-call positive control outside the theorem's hypotheses.

The general model is Q[Sp(2m,2)], m>=3. Let T be a paid involution, level(M)=2, level(TM)=3, and y=M+z, where every term of z has level at most one. Let U be the free left module generated by y and any earlier terms of level at most one. Then TM=u+F T v, with u,v in U and F free, holds exactly when TM is in U+TU. For the kind B corollary use M=WD, z=cW gamma, gamma free, c!=0, and W gamma outside PD. The exact call-alone proof is a mathematical dependency.

No new kappa, general two-in-five exclusion, joint decoder, phase lift or physical implementation is claimed. The same F must act on the entire output of one call. Multiple high-level companions and independently postprocessed multiple outputs are outside this theorem. The selected frontier and all five assumptions remain unchanged. This completes the advance claim.

Evidence and reproduction

Normal and optimized CPython 3.12.16 replay passed on macOS arm64. The complete fixed first/shared-call projection table covers 344064 preparation/call pairs: all 10752 three-bit free elements and all 32 calls that reach level three. Exactly 768 columns survive each target-frame projection, for 24576 in total. All are in the required intersection subgroup; the lower-level companion families contribute no target-frame coefficient.

Exact group-algebra unit, nonunit-kernel and outside-subgroup inverse checks cover all 32 calls and 96 wider conjugated contracts. The relaxed false positive is checked at five nonzero coefficients, with its call-alone separator checked on all 21504 old/paid binomials and all earlier singleton families. The kind A control verifies a full one-call post-map identity. These finite checks support the written all-width proof; they do not enumerate arbitrary rational maps or every wider group.

Metadata, committed bytes, modes, source ancestry, the seven-file official export and its integrity check passed. A read-only optimized export replay passed without Git or a usable PATH; all eight payload/manifest hashes and modes stayed unchanged. The proof and README were checked with GitHub Markdown rendering. No external runtime source is used.

From the repository root, with CPython 3.12 and its standard library:

python3.12 -I -W error -B research/last-call-common-postmap/verify.py
python3.12 -I -W error -B -O research/last-call-common-postmap/verify.py
python3.12 -B scripts/research_package.py --package research/last-call-common-postmap
python3.12 -B scripts/check_research_commit.py --package research/last-call-common-postmap --check-source-history --check-modes

Expected status markers:

PASS_LAST_CALL_COMMON_POSTMAP
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. The predecessor's full call-alone census, native operators, selected cold replay, Lean and Docker were not rerun. Independent human and formal-kernel review are unclaimed.

Sources and credit

Prepared by @aliromman with substantial OpenAI Codex assistance. Jacob Sussman (@jacobalansussman) supplied the definitions and untested kind B post-call question in PR #7629, exact source. Nick Petalas (@petalas) supplied the related coset method and post-call boundary in PR #7649, exact proof. Ali Romman's PR #7683 supplies the required call-alone theorem and reused binary matrix routines, originally from PR #7679. The common-factor proof and separation from the relaxed span are new. Kind A post-call gains are prior work and are used only as a boundary control. New work is Apache-2.0.

Base: canonical main e54523517d0f03e58352e647c403a5af3e525a57, priority revision 8; card complex-rank-shape-decoder refined to this common-post-map boundary. Refreshed open/closed PR and Discussion overlap contains the earlier finite, kind A and call-alone results. Sparse local notes and unpublished work remain coverage limits. No source PR or selected artifact is edited. Joint decoding must retain one preparation/post-map factorization per actual call output.

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