Skip to content

Final-coordinate carrier pricing: conditional κ = 5.2791461623702e-5 - #98

Closed
rohanarun wants to merge 14 commits into
CrocSwap:mainfrom
rohanarun:research/final-coordinate-pricing
Closed

rohanarun wants to merge 14 commits into
CrocSwap:mainfrom
rohanarun:research/final-coordinate-pricing

Conversation

@rohanarun

@rohanarun rohanarun commented Oct 8, 2026 •

Copy link
Copy Markdown
Contributor

Under the inherited analytic and fixed finite-alphabet multitape hypotheses,

$$T(n)=O\bigl(n(\log n)^{1-\kappa}\bigr),\qquad\kappa=\frac{26395730811851}{500000000000000000}=5.2791461623702\times10^{-5}.$$

This is a small 0.0006927669% improvement over #95, with bit saving 6599281088677/125000000000000000.

Current comparison: #97, opened while this was being published, reports the higher conditional κ = 6.3965813×10⁻⁵ using deferred bit endpoints and a signed complex schedule. This PR is an incremental improvement to the PR95 construction, not a record claim.

At h25, price legal carrier choices in the final coordinate order, including #93 and #95's later relabelings. Previously the producer made these choices using an earlier coordinate order. The unchanged compiler, with this existing parameter updated, reproduces the selected new physical word byte-for-byte. h23 remains #95's word.

Roles remain 26,387 / 34,772, width 130,417,912, and total recursive rank 74,988,452,500. The complete internal child distribution changes. Every exterior, center-copy charge, side-growth contribution and early/late restoration gate remains paid; the bridge and complex layer are unchanged.

Full verification passed: make -j1 verify (92 isolated test modules, 20 historical patch checks), both parent control suites, full inherited producer-chain regeneration and six own controls completed in 6,633.390 seconds on f07c7db7d1fd378b0deae68e33b53421456319ae. All 57 research-head GitHub checks passed; no source drift. The committed validation receipt records the commands, timestamps and evidence hashes. Subsequent changes only add that receipt and update package documentation and its fingerprint.

Checks:

  • Fresh h25 compilation with the pristine inherited engine, byte-identical selected word, and dense scalar support audit.
  • Complete physical replay for both axes, including arbitrary-dirty restoration, and fresh full CRT profiles with zero disagreements.
  • Exact and independent moment enclosures, 47 constraints, seven margins, both next-grid exclusions and exact rejection of Coordinate descent beyond PR94: conditional κ = 5.279109590448e-5 #95's complete profile at the new saving.
  • Six adversarial/roundtrip controls and 2,764 pinned source files.
python3 research/final-coordinate-pricing/verify.py
python3 research/final-coordinate-pricing/test_controls.py
python3 research/final-coordinate-pricing/verify.py --regenerate
make verify

The package contains the general parameter-transfer argument, full paid accounting, source pins and a Python 3.11/3.13/3.14 workflow. This candidate emerged from the completed 16-experiment exchange batch, but no long cycles were accepted: the gain is from coordinate-aware pricing alone. No search optimality, new Lean proof or practical speedup is claimed. All inherited all-size and analytic hypotheses remain assumptions.

Credit Chafik Boukhalfa (#91), Maxime Fleury with Codebuff assistance (#94), Rohan Arun (#93/#95 and geometric pricing), Alejandro Zarzuelo Urdiales, Thomas DiFiore, Dominik Scholz, eumemic, Avi Eisenberg, Rohan Garg, icekylinx, James Chang, Zhihao Chen, Aurel Prosz/Paureel, Swapnil Jain, RaD/hipotures, Douglas Colkitt, OpenAI and all predecessors. Original code and notices are preserved.

Prepared by Rohan Arun with OpenAI Codex assistance.

@rohanarun
rohanarun marked this pull request as ready for review October 9, 2026 00:37
@Mister-Meeseeks

Copy link
Copy Markdown
Contributor

Thank you @rohanarun. This contribution is now recorded on main in the scoped reconciliation review and contributor credits.

Contribution: Coordinate pricing with the baseline legal engine.

Review and limits: Exact-head CI evidence. The long-cycle search accepted zero candidates; retained as a scoped negative experiment.

The complete original source, proof/research notes, notices and relevant replay/CI evidence are retained through the pinned source archive. The review distinguishes fresh runs, exact-head historical CI and prior integration evidence. Closing this historical PR as reconciled; its obsolete headline does not replace the current selected bound, and draft/retracted claims retain their explicit qualifications.

Reviewed head: 6862c47f5fcb69dfb424e1bb7f32f415d8659063. Publication: b84ba67156a01f9c2cb34fe1e3cb1dcdeab00d9c.

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.

3 participants