Skip to content

Exclude single cocharted BIT helper relations with complete causal-bank and target accounting - #7720

Open
aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-053
Open

aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-053

Conversation

@aliromman

@aliromman aliromman commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

A single cocharted binary helper relation cannot improve the complete signed moment of the fixed PR7001 BIT word for 0<=kappa<=1/100, in either orientation. This exclusion grants free storage of every past receiver bank and free scalar XOR work, but charges the lost productive helper. No new kappa is claimed.

Read the proof, summary receipt, full certificate, checker, input pins and credit. The canonical claim records the changed helper contract and overlap check.

The source has 659 public fields, 437 helpers in nine boundary groups, 237 owned labels and 2871 paid moves. At each chart, all entrance and retained paid rows are independent. A single helper relation can therefore remove at most one paid bank at that chart: the last token in its unique receiver witness. The exact receiver-space certificate reduces all possible relations to the following finite cases and terminal bounds.

Orientation Exceptional directions Maximum nonterminal clean dimension
Forward 664 4
Transpose 125 2

Every exceptional direction saves strictly less integer paid rank than the productive rank it loses. The exact inequality d^99>(d-1)^100 converts that deficit into a strict signed-moment loss throughout the stated interval. Every other direction can remove only one terminal bank of rank at most the lost target. The largest group has 2^410-1 nonzero relations; a complete terminal prefix-rank certificate covers the unlisted directions.

Both implementations recount the complete 36-call invoice: 107352 paid calls, 19950 productive targets and rank surplus 1506. Fixed query, idle and root terms cancel; the native difference is multiplied by 36. No paid-count-only or frozen-caller ceiling claim is used. This result blocks a specific proposed improvement route; it does not improve the selected saving .00363005817 or change the source BIT admission.

Base: e54523517d0f03e58352e647c403a5af3e525a57. Inputs are the exact native and caller from PR #7001, head d777f3a0aa2878bea0478918923dabcf86b9878b. The native SHA-256 is 51203e7fc75ba10bd3ac3f8e675759f47e13d262bea66b6600d1e515095cad78.

This differs from the productive splits in PR #7304 and PR #7573: it surrenders one old productive helper. The older helper deletion and fixed-anchor exclusion in PR #1043 and PR #1250 use another word. The helper splits in PR #5679 and PR #5708 and rational quotients in PR #7713 supply related methods, not this binary certificate. The premise lift in PR #7618 is separate.

Use CPython 3.12 and its standard library:

python3.12 -B scripts/research_package.py --package research/bit-single-relation-bank-obstruction

Both registered commands print:

PASS_SINGLE_BIT_RELATION_OBSTRUCTION 18 group-orientations; 789 exceptional directions; 6 rejected mutations; exact frame and full invoice checks

Normal and optimized replay pass. Packed-row/highest-pivot construction agrees with a set-based/lowest-pivot replay on every paid row and all public endpoints. The second checker proves each receiver dimension by projection rank and checks terminal coverage by prefix ranks. All 636 rational frames and 2374 strict nested paid pairs pass. Six mutations are rejected: missing debit, receiver, noncoordinate direction, witness token, causal order and terminal pivot. Exact input/notice bytes, committed bytes/modes, source history, export, sealed read-only Git-free replay, public links and rendered publication readback 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 exclusion keeps paid forms and their order fixed. Multiple relations, new routes, changed capture forms, mixed boundary groups, matrix-valued receiver coefficients and compositions across orientations are outside scope. The physical caller identity, native phases, owned storage, compiler, tape and analytic interfaces remain inherited, with all five framework assumptions. Full predecessor suites, selected cold replay, Lean and Docker were not run. This is author checking after reading the source, not blind, independent human or formal acceptance. No selected-frontier change or edit to another workstream is included.

Credit Jerry Chen / @CubeJerry for the paired source and related alternate-carrier analysis; zxv and retained authors for the native; DreamingOfClouds for the chain-root caller; ToikY for the fixed-bank exclusion; Ali Romman for the related helper splits and quotient method; Cinna Davis and retained contributors for the framework. The source notice supports the inherited credit, and all notices remain unchanged. Prepared by Ali Romman with substantial OpenAI Codex assistance, Apache-2.0.

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