Skip to content

Extend owned-wire floor to free native/helper boundaries and nonnested C moves - #5140

Merged
Mister-Meeseeks merged 2 commits into
CrocSwap:mainfrom
1337isnot1337:research/owned-wire-boundary-floor-20261023
Oct 23, 2026
Merged

Mister-Meeseeks merged 2 commits into
CrocSwap:mainfrom
1337isnot1337:research/owned-wire-boundary-floor-20261023

Conversation

@1337isnot1337

Copy link
Copy Markdown
Contributor

Extend #5136 from the eight owned high contacts to the complete owned C wire. In the actual round-three word, moving either center side, both helper boundaries or every native input/output boundary cannot lower its owned-wire rank mass below 64 under the fixed scalar/main-boundary and monotone persistent-L contract. Each block pays moment at least 8^t for 0<t<=1. The source attains the bound. No higher κ or whole-invoice optimum is claimed.

This extends Matarener's #5060 fixed-native/fixed-helper nested-wire floor. The complete new argument combines the existing 128 main-entrance paths with 144 outgoing paths to fixed main exits. Nine independent terminal annihilators per block force the common low-contact cap to zero. The Lagrangian projection inequality lower-bounds actual relative Clifford B-block rank by subspace symmetric-difference distance, so nonnested canonical owned moves and arbitrary paid intermediate Clifford charts are included. All 240 native events per orientation are cut; helper endpoints supply no seed or cap.

The package retains all source attribution: Matarener's propagation/path floor, Dugongue's owned-retiming and chart grammar, Cao Yifan's round-three word, Cinna Davis's native/controller/general-Gaussian work, zxv, Jacob Alan Sussman and the retained lineage. Frozen predecessors are unchanged. This extends the existing package, with written proof, deterministic source-bound witnesses and updated open questions.

Validation:

  • Registered normal and optimized replays pass: 272 paths, 4720 links, 300 controls, the complete original paid/helper baseline and 18459 small Lagrangian-pair calibrations. The universal rank argument is written explicitly, not inferred from finite calibration.
  • Metadata/pins, deterministic regeneration, committed-byte comparison and whitespace checks pass.
  • Official committed export/integrity check passes. The exported verifier also passes under isolated optimized Python in a Git-free temporary directory.
  • The optional full source-history preflight remains unavailable in this shallow checkout. All live scientific inputs are retained on canonical main at the original package base; no foreign-branch executable is imported.

The earlier #5136 hosted run failed on eight shared onboarding fixtures before package replay; that skipped replay is not green. This PR starts from refreshed canonical main a5df48c. Guidance, selected κ=.00302015 and the exact scientific input are unchanged. Hosted checks at this new head are separate.

The next changed premise must address nonmonotone persistent L paths, a changed scalar word/operand interface or changed main boundaries. The old common-kernel scan also fixed native maps, so it does not exclude a joint nonmonotone/shared-map experiment. Native phase/endpoint/ownership replay and complete compiler/precision/outer funding remain obligations for any positive candidate.

Canonical investigation: #4611 (comment) . Prepared by Cinna Davis / 1337isnot1337 with substantial OpenAI Codex assistance; no subagents; Apache-2.0. Written scoped proof and author-checked exact finite evidence, not independent human or proof-kernel acceptance.

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