Skip to content

Conditional κ ≥ 0.0036521182092 (+8.315456e-7 over #7644) with 74 erasures - #7645

Open
Clear1nsight wants to merge 3 commits into
CrocSwap:mainfrom
Clear1nsight:kappa-original74-main46-on7644
Open

Clear1nsight wants to merge 3 commits into
CrocSwap:mainfrom
Clear1nsight:kappa-original74-main46-on7644

Conversation

@Clear1nsight

@Clear1nsight Clear1nsight commented Nov 17, 2026 •

Copy link
Copy Markdown
Contributor

Change and mathematical scope

This construction raises the conservative conditional bound from 0.0036512866636 to 0.0036521182092 with the original 74 ordered erasures and the same signed 17-call caller as #7644. The difference between the published bounds is +0.0000008315456 (about 0.022774%).

Role 46 travels directly from chart 513 to 127. The original path visited 914 and 245. An entrance sequence of 22 ADDs and two SCALEs, plus three later ADDs, preserves the intermediate private observations. #7644’s edits to roles 82 and 185 are retained. Across all 17 invocations the paid counts change by −34 at rank 1, −17 at rank 4 and +17 at rank 6. Total paid positions fall from 110,332 to 110,298 with the same 22,461 productive fields. For each invocation, strict subadditivity gives 6^t < 4^t + 1^t + 1^t for every 0 < t < 1. The fixed-invoice crossing is approximately 0.003652118209307584843, about 1.076 × 10⁻¹³ above the conservative claim.

Task and overlap

Refined revision 8 card p104-side-correction from main e54523517d0f03e58352e647c403a5af3e525a57. Source #7644 at 32cfa1d20d3e4587dc601e69ccac1cd24a239e5c, word 86bc8a1d1276c5b8d80a19704f90279e85b24b9490a02b9afa67c1a5b119c114. Candidate word 6c52b2fba7c20343a50cb92ed4b997aa9921c78f212ac6428d13fff2bcd014cf. Scoped Discussion claim.

#7611 supplied the joint repair on a changed CLEAR-content interface. #7616 ported a wider family and qualified a 113-lifetime successor. This package isolates the role-46 change on the exact #7644 word and qualifies its original 74-erasure application. #7642 covered the final hop, and #7643 addresses roles 134/329. The larger figure in #7641 uses 234 lifetimes and an additional common-U premise. Bounded findings on that route.

Validation

python3.12 -I -B scripts/research_package.py --package research/p104-original74-main46-on7644
  • Two exact scalar checks matched 1,501 arbitrary-input public rows, all 630 retained private/COPY observations and all 74 ordered erasures. The independently written interpreter rejected a wrong-sign correction at output 145, input 36.
  • The native replay covered 2,958 chart pairs at 104 ports. Its detailed checks covered 58 affected pairs and four route compositions with 512 forward and 512 adjoint columns, including absolute phase. All 21 native rejection controls passed.
  • The signed application checked 3,043 fresh catalogue maps and 104 main boundary pairs with signed Pauli action and exact global phase, plus 54 strict and four functional outer conditions. The ordinary bound was 782,645,968. Precision factor 2,430 and finite tape counts are in the packaged result. Directed decimal and dyadic invoice calculations agree.
  • The registered offline command ran to completion on Linux with Python 3.12.3 in 939.20 seconds, with 663,928 KiB peak RSS. Its fresh native and signed stages passed. A relocated run of the optional independent checker matched all 24 stable fields of its original result in 47.53 seconds.

Review the literal edit and proof, complete signed result and portable independent checker.

Credit

DreamingOfClouds supplied the joint correction (#7611, bbdbc8aea2f4fd94a927f6d8c438372c0c69ae6c), SYLforge the wider W1/W2 port (#7616, 22b310ba50a554b6111804824782793beb41073f), ToikY the #7644 baseline, Ali Romman the original protected #7605 application, and Rohan Arun the #7113 signed caller. Clear1nsight prepared this source-bound port, verification and packaging with substantial OpenAI Codex assistance. ChatGPT Pro assisted the separate #7641 search. Original licenses and notices are retained.

CI prerequisite

The first hosted plan stopped in the shared CI navigation test before the package replay. A separate commit carries ToikY’s published CI repair from #7643. It points the CI guide to the already-selected #7000 and runs the historical replay against its original snapshot and rejection controls. All five file blobs match ToikY’s repair. All six navigation tests passed locally. The construction commit and the CI prerequisite have the same 41-file package tree, c4e25871096fe621edf72fc9f75459e1bbe734bc.

The next hosted job exposed an empty PID file in the package’s source-free process-cleanup test. Commit 8ee7e6869b57ce445dbebd1d720366eab1f8ca40 fixes the writer with a same-directory atomic rename and keeps the group leader unreaped until cleanup. Delaying the writer reproduced the failure on the previous head. The revised control passed the delayed-write case, an early parent exit with child termination, and 24 ordinary repetitions. This commit changes only the control and its README. The constructor, both word digests, pricing code and all mathematical verifiers retain their reviewed bytes. Final-head workflow.

Open next questions

The bound inherits five framework assumptions covering all-column operators and spectators, owned erasure, the uniform compiler, precision and fixed-tape tariffs, and analytic integer recovery. A maintainer can review the literal edits, emulator controls, resource bills and source-bound caller before considering admission. The package is confined to its research directory.

Clear1nsight and others added 3 commits November 18, 2026 08:58
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)
@Clear1nsight

Copy link
Copy Markdown
Contributor Author

Final verification at 8ee7e6869b57ce445dbebd1d720366eab1f8ca40 passed. The plan, historical replay and candidate replay all succeeded. The registered candidate job regenerated word 6c52b2fba7c20343a50cb92ed4b997aa9921c78f212ac6428d13fff2bcd014cf and completed the native and signed applications. Its full result matched the final local replay, including κ ≥ 0.0036521182092, 110,298 paid positions and 22,461 productive fields. The canonical Discussion handoff records the five inherited assumptions, verification receipts and next maintainer review. The contributor hub links back here.

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