Skip to content

Conditional κ ≥ 0.0037194530062 (+0.06374% over #7660) with full-extent private banks - #7663

Open
aliromman wants to merge 2 commits into
CrocSwap:mainfrom
aliromman:codex/p104-joint-chart-owned-banks
Open

aliromman wants to merge 2 commits into
CrocSwap:mainfrom
aliromman:codex/p104-joint-chart-owned-banks

Conversation

@aliromman

@aliromman aliromman commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

This construction gives conditional κ ≥ 0.0037194530062. It adds 0.0000023694495 (0.06374%) over the strongest open bound I found. It also supplies an exact-arithmetic, full-private-extent implementation for the changed word using fresh finite physical costs under the five selected framework assumptions. The resulting bound is 2.46263% above the selected 0.00363005817. Most of that difference was already present numerically in the open parent. Proof · Full signed receipt · Parent construction, PR #7660 · Selected application

What changed

Fifteen scalar contacts use a common hyperplane restriction and eighteen use a common line extension. Six helper endpoints change. Four paid positions disappear per native invocation. Across the actual 17-call signed caller, the paid count falls from 108,581 to 108,513, with 22,461 independent productive fields and rank-mass surplus 1,424. Both paid and target rank distributions change. Two directed evaluators place the raw crossing between 0.00371945300635 and 0.00371945300636. The interior probe passes the 54 strict and four functional outer conditions. Literal edit · Source pins

Each of the 3,978 native lifetimes across the signed caller receives its own full physical home. An extent guide generated from the geometry drives a complete overwrite in 2L+3 local transitions for L cells, including dirty or blank interior cells. A common-grid induction establishes exact signed division after every completed child in this implementation. The five remaining assumptions cover native all-column adapters, owned allocation, compiler closure, fixed-tape movement and storage, and analytic recovery. The historical supplier for arbitrary rounded residuals at one fixed tariff remains open. Extent and scalar implementation · Prior typed initialization, PR #7586 · Rounded parent-clock alternative, PR #7662

Verification

Run from the repository root with CPython 3.12 and its standard library.

python3.12 -B scripts/research_package.py --package research/p104-joint-chart-banks

Expected markers include

PASS_JOINT_NATIVE 1501 234 [100502, 187, True]
PASS_JOINT_APPLICATION 0.0037194530062 108513 22461
PASS_REGISTERED_RESEARCH_PACKAGE research/p104-joint-chart-banks outcome=completed

The registered replay passed locally and on Ubuntu 24.04 with CPython 3.12. The hosted scientific job reproduced the signed κ and all six catalogue shards. It checked every public row, all 234 ordered erasure identities, 3,023 native operators, and 104 main boundary pairs. It explicitly compared 512 forward and 512 adjoint columns on each of the 18 new chart pairs. Controls rejected a phase-only fault, an equal-price scalar sign error, an aliased stage, a self-referential erasure witness and an odd-tick exact division. A literal full multiplier execution and independent human review remain open. Registered checks

Source and credit

The base is e54523517d0f03e58352e647c403a5af3e525a57. @Clear1nsight supplied the parent word at d12b854d4825e9edfa3a3466909e1abf3525e929. @DreamingOfClouds supplied its large-neighbourhood and swap-composite predecessors. @SYLforge supplied their 233-erasure base. @rohanarun supplied the signed caller at 07a76c5e2738882c0cb322278e56f6cae3c9885d. @CubeJerry supplied the selected complex and BIT constructions. Original source notices are preserved. This contribution received substantial OpenAI Codex assistance. PR #7658 · PR #7641 · PR #7637 · PR #7113 · PR #7000 · PR #7001 · Independent caller residuals, PR #7661 · Coordination claim

The scientific commit is b81f40997de646de30749d367484d89e59fc3013. A separate prerequisite commit, 75cdd54ffcceedcc664e09b9c22b32e620fa6fa0, carries @ToikY’s copy of @Clear1nsight’s published CI repair. All five file blobs match its original PR. This corrects the base guide’s obsolete replay command and restores the historical verifier’s pinned inputs. The hosted planning and historical replay jobs also passed. Shared repair, PR #7257 · Source of the reused commit, PR #7643

Reuse Clear1nsight's exact CrocSwap#7257 guide and historical-source snapshot repair, already validated by the hosted checks on CrocSwap#7644. The research package bytes are unchanged.

(cherry picked from commit 694667c)
(cherry picked from commit c58206c)
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.

2 participants