Skip to content

Exclude rank saving for orthogonal fresh-bridge families - #7710

Open
aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-045
Open

aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-045

Conversation

@aliromman

@aliromman aliromman commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

This scoped negative proof excludes paid-rank saving for the prescribed family of orthogonal fresh-coordinate bridges. It covers any old basis dimension N and any number b of these bridges with N*b even, subject to the exact word and relation hypotheses below. This settles the paid-word question for the named two-link alphabet in PR #7707 under that relation budget. No new kappa follows.

Read the proof, exact receipt, checker and credits.

The old projectors are an arbitrary binary image basis and its dual, including coordinate projectors. Each bridge has old factors equal to their respective sums, its own literal fresh image axis, and a fresh covector given by a row of Z=I_b+(N modulo 2)J_b. Require Z invertible, equivalently N*b even. Every old and bridge type occurs, with no additional types. Calls are I+([A_j]-1)P over the paid-generated torsion-free abelian exponent group. Its saturated relation rank is at most the actual off-pattern dimension k=N-d+b, where d is the old diagonal-intersection dimension. The endpoint must be diagonal monomial on the full original ambient space.

No old call lies between the first and last occurrence of any bridge type. The bridge types commute, so they can interleave and then be collected without crossing old calls. Every collected bridge total is zero. Deletion preserves all ambient inputs, and each physical target exponent is a sum of a disjoint subset of old paid exponents. Rank subadditivity gives total target rank at most total paid rank.

The new step handles the star dependences: their rank is Nb for odd N and Nb-b+1 for even N. Independent bridge-variable sets force the chronology even when those dependences exist. This extends the prior one-bridge method. It does not assume literal adjacency of one bridge type, or infer exact exponent cancellation from parity alone.

The exceptional sixth projector in the (k,nu,n)=(4,0,5) complete pool is excluded. Its pattern couples two bridge totals and changes the first-order equations. Other even maximizers, larger relation budgets, singular fresh pairings and arbitrary fresh embeddings remain outside this theorem. No native provider, physical implementation or complete multiplication invoice is supplied. The selected kappa stays 0.00363005817, with all five assumptions retained.

With CPython 3.12 and the standard library, run:

python3 -B scripts/research_package.py --package research/orthogonal-bridge-family-obstruction

Both normal and optimized modes print:

PASS_ORTHOGONAL_BRIDGE_FAMILY_OBSTRUCTION
20335 bases; 81002 basis/bridge checks; 55250 full Laurent words; 10148 passing orders
15 larger fixtures; all bridge totals vanish in the stated class.

The finite census checks all 20,335 bases through old dimension four and each permitted b through four. Every one of the 55,250 word orders has a degree-two and full original-coordinate Laurent replay; a separate gap count confirms the 10,148 passing orders. Fifteen larger fixtures reach old dimension ten and six bridges, with repeated calls, literal interleavings and unused spectators. Omitted-call, wrong-pairing, added-relation, wrong-ambient-group, nonzero-even-total and sixth-projector controls pass. Twelve singular pairings have explicit common-kernel vectors.

Normal/optimized replay, committed bytes and modes, source history, export and sealed Git-free replay pass. 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 unbounded conclusion rests on the written proof. Full predecessor suites, selected cold replay, native/physical replay, Lean, Docker, independent human and formal-kernel review were not run. The adapted checker is not an independent source review.

Base: current main e54523517d0f03e58352e647c403a5af3e525a57. Ali Romman's PR #7695, head 996c002d82420141c19797e4e3b699ed88aee330, supplies the Boolean partition, augmentation argument and target matching, and the adapted primitives. The two-link geometry and pool exception come from PR #7707, head 57e24923b609d2588b83321f530fa2a1e75ed364, and remain prior results.

Credit @CubeJerry for the original chronology and Laurent methods in PR #6782, head 685457252b51b9c7932ece363365bdf0b12c9c31: word proof and original recurrence. Ali Romman's PR #7694, head 96b8b22f5925c4cf4514ed27575f465ebc95e51a, supplies the edge-family extension. The linked credits retain the earlier geometric contributors. Prepared by Ali Romman with substantial OpenAI Codex assistance, Apache-2.0.

The claim records the two-link starting question and its generalization within this single result. Live source/phrase overlap and the closed delta were checked. Sparse local and unpublished work remain coverage limits. The open questions retain the exceptional six-projector pool and other specified alphabets or relation budgets.

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