Skip to content

Exclude rank saving in extremal binary cores at k relations (no κ) - #7694

Open
aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-034
Open

aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-034

Conversation

@aliromman

@aliromman aliromman commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

Result and scope

An extremal irreducible binary scalar core cannot save paid rank with at most k exponent relations. Precisely, the word consists of noncoordinate rank-one-idempotent calls I+([A_j]-1)P_j over the paid-generated torsion-free exponent group. Its pattern dimension is k>=2, its irreducible joint-image dimension is 2k-1, its saturated exponent relation rank is at most k, and its full ambient endpoint is diagonal monomial.

Read the proof, exact receipt, verifier and scope record. All bridge calls must form one consecutive block with exponent sum zero. Deleting that block preserves the full ambient linear map, including every spectator-input column. Every target exponent is then a sum of a disjoint subset of paid exponents; total rational target rank is at most paid rank. Word length, ambient width and address ranks are unrestricted within these hypotheses.

The new result extends the prior five-dimensional restricted word theorem to every k. Full ambient deletion uses the unique ambient link from the recent classification, not just equality of restricted actions. The proof preserves the sheared endpoint's allowed edge entries and uses the entire degree-one kernel in the paid-generated group. The normal-form shear remains a proof device.

No new kappa, native supplier, rounding contract or complete multiplier invoice is supplied. Coordinate calls, reducible or smaller images, and larger relation budgets are outside the 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. Exact star-quotient checks cover one through twenty edge pairs. Matrix-product degree-two checks cover 88748 type orders in eleven stated domains. Separate full Laurent products cover all 20532 type orders with two bridge calls at core widths three, five and seven: 5166 have the expected diagonal endpoint, and every other order has a forbidden off-edge entry. Bridge exponents are assigned in chronological order; these are type orders, not separately labelled bridge permutations. The width-five case is retained as a prior control.

Ten ambient fixtures use nonsymmetric fresh factors, dirty-spectator columns and unused coordinates. The whole bridge block cancels in the full matrices. Omitting one bridge changes an explicit dirty column. A wrong-sign bridge has the same linear kernel and passes the degree-two quotient but fails the full endpoint. One extra relation permits an exactly diagonal word with nonconsecutive bridge calls; this is a chronology counterexample beyond the budget, not a saving. An even-generator control shows why an arbitrary ambient group gives the wrong parity kernel.

Metadata, committed bytes, modes, source ancestry, eight-file export and archive integrity pass. A read-only optimized export replay passes without Git or a usable PATH; all nine payload/manifest hashes and modes stay unchanged. Finite checks corroborate the written all-length argument.

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

python3.12 -I -W error -B research/extremal-core-word-obstruction/verify.py
python3.12 -I -W error -B -O research/extremal-core-word-obstruction/verify.py
python3.12 -B scripts/research_package.py --package research/extremal-core-word-obstruction
python3.12 -B scripts/check_research_commit.py --package research/extremal-core-word-obstruction --check-source-history --check-modes

Expected markers:

PASS_EXTREMAL_CORE_WORD_OBSTRUCTION
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 predecessor package replays, selected cold replay, Lean and Docker were not run. Independent human and formal-kernel review are unclaimed. The full ambient identity is an exact formal scalar statement; it is not a native compiler or finite-precision implementation.

Sources and credit

Prepared by @aliromman with substantial OpenAI Codex assistance. @CubeJerry supplied the five-core bridge theorem, associated-graded quotient and chronology argument in PR #6782, exact word proof. Its Laurent checker supplies the adapted recurrence. This adaptation is not a separately authored review. The prior five-core result and wrong-sign control keep their credit.

Ali Romman's PR #7693, exact proof, supplies the all-width classification and unique ambient link. It uses the matching theorem in PR #7692, exact proof. These are explicit mathematical dependencies. Credit details retain the source lineage. New work is Apache-2.0.

Base: canonical main e54523517d0f03e58352e647c403a5af3e525a57, priority revision 8. Card bit-new-register-model, scope changed to an exact negative word theorem. No native construction is proposed. Live open PR and Discussion overlap and handoffs show no competing all-k bridge claim. The closed-PR refresh timed out; the prior complete 6175-record snapshot and a fresh updated-order delta show that no closed PR changed since that snapshot. Sparse local and unpublished work remain coverage limits. A new construction must state a different hypothesis, exact alphabet, complete paid/target invoice and native interface.

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