Skip to content

Prove exact four-stage pure-shear swap tariffs (no κ) - #6407

Merged
Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
petalas:research/w5-1102b-first-improvement
Nov 10, 2026
Merged

Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
petalas:research/w5-1102b-first-improvement

Conversation

@petalas

@petalas petalas commented Nov 2, 2026 •

Copy link
Copy Markdown
Contributor

Summary

Prove the exact swap tariff for at most four pure shear stages on two arbitrary K-coordinate banks, over every field, with arbitrary invertible within-bank bridges. Claim type: scoped proof with finite exact corroboration. No new kappa or supplier admission.

minimum rank charge = 3K
optimal alternating profiles = (r, K, K, K-r)
minimum directional tariff = f_U(K)+f_L(K)+min_r[f_U(r)+f_L(K-r)]

The tariff formula holds for nondecreasing subadditive directional tariffs. A common concave power gives exactly 3 K^theta. The proof classifies the complete noncommuting block word, characterizes equality by an idempotent and gives every attaining rank split. The new result settles the pure-stage subclass of themilkman216's PR2052 even-stage question. Simultaneous mixed-direction stages remain outside it.

Credit themilkman216 / PR2052, Johnny Franks / cross-shear-stage-bound, Jacob Alan Sussman and the archived borrowing-chain/controller lineage. Unpaidmoon's finite F2 mixed-stage result is cited separately below. The new proof and verifier are produced by Nick Petalas. Independent human and formal proof-kernel review are unclaimed.

Evidence

Before: PR2052's general stage inequality leaves an even-stage scheduling gap. After: any four-stage improvement over three stages must use a genuinely simultaneous mixed-direction stage. At K=4 the pure-stage minimum is exactly 12, with witness profile (1,4,4,3).

At head 7442594d23a351624b20417684afdb31590a2ce9, CPython 3.12.15, macOS 27.0.1 arm64:

  • Exhaust all 65,536 alternating binary words on two-dimensional banks: 96 complete swap endpoints, 48 rank-optimal words, minimum rank six and minimum square tariff ten. Full dense matrix execution agrees with the separate block criterion in every case.
  • Exhaust scalar words over F2,F3,F5,F7. Check 176 attaining complete-input constructions, 176 within-bank transports, 325 monotone directional tariff pairs, 800 integer price fixtures and seven named semantic controls.
  • Normal and optimized warning-strict direct checks pass in 0.712 s and 0.623 s. Both registered VERIFY commands pass locally. These are check durations, not a speedup claim.
  • All six committed package bytes, executable modes and source-history ancestry pass. Official export integrity and declared-input checks pass. Optimized replay from a read-only Git-free export with an empty executable PATH emits the identical complete receipt.

Exact reproduction from the repository root:

/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -I -W error -B research/four-stage-pure-shear-tariff/verify.py
/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -I -O -W error -B research/four-stage-pure-shear-tariff/verify.py
/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -B scripts/research_package.py --package research/four-stage-pure-shear-tariff
/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -B scripts/check_research_commit.py --package research/four-stage-pure-shear-tariff --check-source-history --check-modes

Expected markers: PASS_FOUR_STAGE_PURE_SHEAR_TARIFF, PASS_REGISTERED_RESEARCH_PACKAGE and PASS_RESEARCH_COMMIT_BYTES. Another CPython 3.12 path can replace the local interpreter. The standalone verifier writes no files and generates a receipt bound to all package hashes.

Limitations

The result requires one direction per stage. Splitting a simultaneous mixed stage changes the depth and cannot apply this theorem to the full PR2052 model. Additional banks, incomplete returns, physical bridge/native/helper/owner costs, precision, tape, compiler and outer transfer remain outside scope. No selected cold replay, Lean, Docker, physical supplier or complete multiplication theorem was run or claimed. All five selected framework assumptions remain; the selected frontier is unchanged. No predecessor scientific verifier was replayed.

Open next questions

Can four simultaneous mixed-direction stages over Q swap four bank pairs in eleven rank units? Enforce HE=EH=0, all bridges and the complete arbitrary-input endpoint. The present pure-stage identity cannot decide that altered depth contract. The full K=4 instance over F2 is already covered by the cited predecessor below.

Related open work

  • unpaidmoon fork PR125, head d9f9103be548e5b1b4a4ee1bf216ac5084ca7824, reports the stronger mixed-direction exclusion at K=4 over F2. Our already-computed pure F2 K=4 witnesses also cost 12. This is agreement on that shared restricted number, not an independent verification of the predecessor's finite mixed-stage search. Our new theorem is all-field and all-K, with a pure-stage hypothesis and a complete tariff/equality classification.

Advance claim and scope update. Source main was refreshed and live PR/claim overlap rechecked before commit; no other open PR on this specific swap tariff was identified. Unpublished work remains outside coverage.

Completion refresh read this account's 12 other open PRs created since the advance CLAIM. None computes this two-bank swap tariff at depth at most four, so no additional same-quantity comparison or disagreement was found. The related fork PR125 remains open at its cited head; its finite mixed-stage result was not replayed.

CI notes

Both hosted checks, plan and replay (research/four-stage-pure-shear-tariff), passed at exact head 7442594d23a351624b20417684afdb31590a2ce9 in GitHub Actions run 41528019817. The completed checks were confirmed once with gh pr checks 6407 --repo CrocSwap/integer-mult-bounds; run metadata confirms the successful conclusion and matching head. The planning job took 4m41s and the package replay took 1m24s.

Hosted replay certifies execution of the registered normal and optimized checks. Local direct/registered scientific replay, committed-byte/source-history/mode checks and read-only Git-free exported replay also pass. Independent scientific and formal proof-kernel acceptance, and the out-of-scope physical/supplier obligations, remain unclaimed.

Merge Danger

Door: two-way.

Blast Radius: scoped.

Six new files in one self-contained research package. No shared code, selected record or predecessor evidence changes.

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