Skip to content

Record that a second missing edge does not save a dirty one-move G20 word - #277

Draft
unpaidmoon wants to merge 1 commit into
cursor/crocswap-main-e545235-5df7from
cursor/g20-dirty-second-edge-5df7
Draft

unpaidmoon wants to merge 1 commit into
cursor/crocswap-main-e545235-5df7from
cursor/g20-dirty-second-edge-5df7

Conversation

@unpaidmoon

Copy link
Copy Markdown
Owner

Change and mathematical scope

Exhaustive finite census. No new κ. The selected conditional exponent stays 363005817/100000000000 (0.00363005817).

On the emitted local word of CrocSwap#5708, CrocSwap#6923 classifies 686,779 length-at-least-5 one-move assignments as dirty failures. Open CrocSwap#7074 inserts one route at the first missing edge and leaves 686,386 words still rejected. This package recomputes that domain (the first-edge multiplicity table matches CrocSwap#7074's SHA-256 8f38eb2f6e526ede87a62c1fc88c35fd5d3ba87a16f755df690a76cea5a68494) and inserts the second missing edge as well. Both inserted edges have rank 1.

None of the 686,386 words is a strict saving on (0, 1). The same absence holds for every rank pair in 1..7: a missing third edge is missing at every such rank, and raising an inserted edge's rank adds a nonnegative term u(r^t − 1) to the rank-1 moment.

Class Assignments Outcome
A third edge is still missing 673,425 no paid histogram
Emitted, strict loss by rank split 12,953 not a saving
Emitted, strict loss by the elementary rewrite 5 not a saving
Emitted, positive at t = 1/2 and negative at t = 4/5 3 not a saving
Strict saving 0 —

The lexicographically first still-rejected word, in order (interval start, interval end, correction index, moved frame), is scalar positions 1,003–8,891, length 12. Correction 0, ADD(583, 87), moves from frame 1 to frame 326. The outer gate is ADD(583, 60). After the first route (326, 1) is added, the rejecting edge is (591, 553). With both edges present, the walk still misses (467, 396).

The three sign-changing words have shape H1 + 3 H2 − H4 − H5. The first of them is positions 5,661–7,047, correction 8 moved from frame 593 to frame 360, edges (360, 358) and (360, 131). This is not the CrocSwap#6923 repair on that interval, which uses frame 358. These three words are not strict losses either.

This is not a proof that no other routing model can save a word, and it does not insert a third edge. It does not change the selected frontier.

Task and overlap

Base e54523517d0f03e58352e647c403a5af3e525a57. Card bit-helper-contract, scope refined. This is open question 1 of CrocSwap#7074. CrocSwap#6923 is the merged one-move census. CrocSwap#6963 settled two moved corrections on length 5. CrocSwap#7079 is the two-move length-6 census. Those packages hold the published route table fixed. This package adds a second route on a dirty one-move failure. It does not rerun those screens and does not move a second correction.

The fork PR is based on a pin of upstream main at e54523517d0f03e58352e647c403a5af3e525a57, because the fork's main is a different history. The intended upstream base is CrocSwap main at that commit.

Validation

Python 3.12, standard library, from the repository root. Both registered commands passed, and scripts/research_package.py printed PASS_REGISTERED_RESEARCH_PACKAGE with outcome=negative-result.

python3 -I -B research/g20-dirty-second-edge/verify.py
python3 -I -O -B research/g20-dirty-second-edge/verify.py
python3 -B scripts/research_package.py --package research/g20-dirty-second-edge
PASS_G20_DIRTY_SECOND_EDGE still=686386 third_missing=673425 split_loss=12953 elementary_loss=5 sign_change=3 savings=0 witness=1003-8891/0/326/326-1/591-553/third=467-396 no_new_kappa=true

Checks raise exceptions, so -O does not disable them. Every emission was replayed from the start of the word by paid_histogram. The lexicographically first word, one still-rejected word of each of the 28 lengths, and the sign-changing witness were replayed independently. The sign-changing witness was also checked with its first added edge moved from rank 1 to rank 7.

Not run: the CrocSwap#5708 integration replay, the 18-call caller, adjoint columns, moduli 101 and 10201, ordinary admission, Lean, Docker, and make frontier7000-verify.

Credit

The emitted word, route table, A2 relation, and the local-only conjugation count are Ali Romman's research/g20-conjugation-admission, following Dugongue, ai-nanachi, and the retained G20 lineage named there. The length-4-and-above contact-frame obstruction is Nick Petalas's CrocSwap#6305. The two-move classification at lengths 2..4 is Nick Petalas's CrocSwap#6664. The length-1..3 one-move obstruction is unpaidmoon's CrocSwap#6337. The length-4 one-move obstruction is unpaidmoon's CrocSwap#6633. The length-at-least-5 one-move obstruction is unpaidmoon's CrocSwap#6923. The length-5 two-move census is unpaidmoon's CrocSwap#6963. The one-added-route census, including the 686,386 still-rejected words and the first-edge table checked here, is unpaidmoon's CrocSwap#7074.

Prepared by unpaidmoon with substantial Cursor Grok assistance. Apache-2.0.

Open next questions

  1. Does adding the third missing edge as well turn one of the 673,425 still-rejected words into a strict saving? The first step is the same lexicographically first word: its third missing edge is (467, 396). Replay that word with all three edges present at rank 1. This package inserts only the first two missing edges.

Workflow feedback (optional)

Recording the second missing edge while the one-edge walk is still running makes this follow-up a direct replay. CrocSwap#7074 already noted that the one-move census does not store the missing pair. Feedback thread: CrocSwap#1200

Open in Web Open in Cursor 

…word

Recompute the 686,386 still-rejected dirty failures, insert the edge that
rejects each one-edge walk, and show that both edges at rank 1 produce no
strict saving on (0, 1).

Co-authored-by: unpaidmoon <unpaidmoon@users.noreply.github.com>
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