Skip to content

org(review): PR58, an independent proof of 11 and a literature check - #61

Draft
abu-aska wants to merge 5 commits into
review/F7-threadsfrom
ccr-d120afc1-r8sv8m
Draft

abu-aska wants to merge 5 commits into
review/F7-threadsfrom
ccr-d120afc1-r8sv8m

Conversation

@abu-aska

@abu-aska abu-aska commented Nov 15, 2026 •

Copy link
Copy Markdown

Follow-up to #58 and its review record in #60. Draft. A person merges.

The base is the branch of #58, so the diff shows only this work. It can be merged into that branch, or its commits cherry-picked there.

Update. #58's author answered every blocking and major review item in twelve commits of their own. This branch merged them and took their version of every file in threads/. What remains in threads/ is a literature correction to two notes and the manifest. Everything else is the new folder pr58-trust/.

Check the headline yourself in one minute

cd org/review/2026-11/science/pr58-trust/four
python3 -O check.py short_table.json certs/short_table.cert.json 11

Python 3.8+ and nothing else. No floating point, no third-party library. It ends PROVED exactly 11 balance points.

What this adds

1. A separate proof that four charges have exactly 11 balance points (pr58-trust/four/NOTE.md)

  • check.py was written without reading indep_count.py, verify_arb.py or lower_bound_check.py. It uses only the standard library, and every real number is enclosed in an integer interval.
  • An untrusted search writes a certificate. The checker replays every leaf from a root box it computes itself. It uses no hand-derived derivative bound.
  • All four certificates give exactly 11 (7 of index 2, 4 of index 1). Re-run after the merge: output byte-identical.
  • Controls:
    • 8 configurations with other counts all match the threads or indep_count.py, including one 1.2e-17 from a fold.
    • 26 corrupted inputs are all rejected under -O.
  • An adversarial review of check.py found one unsound path, now fixed.
    • An unquoted JSON number passed through a float. That could move a configuration across a fold: 11 proved where the exact count is 9.
    • The headline inputs were always quoted, so they don't change.
    • Both attack inputs are now controls.

2. A literature check (pr58-trust/LITERATURE.md, full texts read)

  • No report of more than 9 for four Coulomb charges was found.
  • But 11 at the regular tetrahedron is already in print, and org(review): threads after the science pass, with four charges at 11 balance points #58's notes, including its repeated search, still missed or misread it:
    • Gabrielov–Novikov–Shapiro (2007) note 11 effective cells, above 9, for their potential with a large exponent.
    • arXiv:2501.05315 reads the degenerate centre as 11 co-located critical points (with a minimum) and does not realise them. The note said it "says only that it splits into three saddles".
  • Theorem 1.1 of arXiv:2609.03965 bounds four positive charges by 570 balance points, with no genericity assumption.
  • I checked all three quotes against the PDFs myself.

3. Why the routes stop at 11 (pr58-trust/four/T444.md)

  • The tetrahedron centre is Arnold's T(4,4,4), so μ = 4+4+4−1 = 11 follows from the classification. A Gröbner basis check agrees: 27 − 16 simple = 11.
  • The thread's multiplicities 8 and 10 are the types P8 and T(3,4,4) on the same ladder.
  • The next rung, U12 (μ = 12), would open a route to 13. Searches find no flat four-charge point of that type.
  • A prediction from the ladder was written down first, then tested on the thread's own 23 samples: log-log slope 1.01 against the predicted 1.

4. Two independent write-ups agree. #58's author and an agent working on this PR each wrote out the three-charge derivations separately. Both found the same two over-claims in three-charges.md. The author's version is kept.

Changes to threads/

File Change
four-charges-eleven.md Cites the 2007 remark and describes arXiv:2501.05315 as it reads. Adds "for the Coulomb potential" and the 570 bound. Names the source of the rounding sentence: rounding the printed table again gives 7, not 9. Lists the later checker and the T(4,4,4) identification.
README.md One sentence on the two prior statements.
MANIFEST.sha256 Two checksums. checks/verify_folder.py passes: 1,481 files, 193 of 193 ledger claims, 22 checkers.

For the owners

  • An outside human check remains the step that matters most. check.py is about 600 lines, meant to be read against the four lemmas in NOTE.md.
  • Whom to tell is yours to decide. The literature check adds two groups beyond the July authors. No one was contacted.
  • Nothing here changes a claim label on a merged record.

Limits

All work here, like the threads and the review, was done by AI models of one family. A shared misunderstanding would pass all of it.

Automated posting by agentic team with human oversight.

🤖 Generated with Claude Code

https://claude.ai/code/session_012tWK9X51ZybvPt7yCm7Zbt

… balance points

A checker written without reading the earlier ones proves exactly 11
balance points (7 of index 2, 4 of index 1) for all four certificates in
the threads folder, including the five-decimal table typed in from
four-charges-eleven.md.

check.py is the whole trusted base: standard library only, no floating
point, integer intervals on a 2^-128 grid with outward rounding. It
re-derives every leaf of an untrusted bisection tree written by find.py,
from a root box it computes itself. It uses no hand-derived derivative
bound. Controls: seven configurations with other counts all match, and
nineteen corrupted inputs are all rejected under python3 -O.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012tWK9X51ZybvPt7yCm7Zbt
…major wording items

Blocking items, as listed in the review record of PR 60:
1. unit-point.md gives the counts certray_868.json holds at each eps.
2. Recurrence labels match README.md; PROVEN (defined only in an
   unmerged PR) is dropped; the four entries with no separate verifier
   run are named.
3. The "every checker raises and ran under -O" sentence names the nine
   the record supports; pr58-trust/CHECKER-AUDIT.md lists all 57.
4. checks/three/DERIVATIONS.md writes out every lemma of the
   three-charge proof; derivations_check.py checks 62 items under -O
   and fails on flipped signs. Two over-claims in three-charges.md are
   narrowed to what the certificate proves.
5. gravity.md and neutrino-like.md say plainly they cannot be re-run
   from this checkout; stale paths fixed where the target exists.

Also: the major wording items (l2288, central configurations, fermion
window, fabric tag, growth constant), and literature corrections in
four-charges-eleven.md: the 2007 remark of 11 at the tetrahedron for a
large exponent is cited, arXiv:2501.05315 is described as it reads, and
"for the Coulomb potential" is added. MANIFEST.sha256 regenerated:
1,445 of 1,445 verify.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012tWK9X51ZybvPt7yCm7Zbt
… by adversarial review

- LITERATURE.md: what is published, with quotes and every query. No
  report of more than 9 for four Coulomb charges was found; the count
  11 at the regular tetrahedron is in print for a related potential
  (2007) and as a reading of the degenerate centre (2026).
- four/T444.md: the tetrahedron centre is Arnold's T(4,4,4), so its
  multiplicity 11 follows from the classification; a Groebner check
  agrees. The thread's multiplicities 8 and 10 are P8 and T(3,4,4) on
  the same ladder. No flat four-charge point of type U12 is found. A
  prediction from the ladder passes a first test on the thread's own
  samples (log-log slope 1.01).
- four/check.py: an adversarial review found that an unquoted JSON
  number passed through a float, which could move a configuration
  across a fold. Unquoted decimals and repeated keys are now refused
  and the certificate shape is checked first. Controls: 8 other
  counts, 26 corrupted inputs. Headline results unchanged.
- README.md: index, findings in order, and what remains for owners.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012tWK9X51ZybvPt7yCm7Zbt
@abu-aska abu-aska changed the title org(review): PR58 answered — an independent proof of 11, a literature check, and fixes for the five blocking items org(review): PR58 answered, with an independent proof of 11, a literature check, and fixes for the five blocking items Nov 15, 2026
…the literature findings

The author of #58 answered the review's blocking and major items in
twelve commits while this branch did the same. Their version of every
file in threads/ is taken as it stands, and this branch's own fixes and
its second three-charge write-up are dropped. Both write-ups found the
same two over-claims in three-charges.md.

What remains in threads/ from this branch is what their repeated
literature search did not find, read from the full texts:
- the 2007 remark of 11 effective cells at the regular tetrahedron for
  a large exponent (Gabrielov, Novikov, Shapiro), now cited;
- arXiv:2501.05315 described as it reads (11 co-located, with a
  minimum, not realised), not as "three saddles" only;
- "for the Coulomb potential" in the novelty sentences;
- Theorem 1.1 of arXiv:2609.03965: at most 570 for four charges;
- the source of the rounding sentence, the later checker, and the
  T(4,4,4) identification.
checks/verify_folder.py passes: 1,481 files, 193 of 193 ledger claims.

pr58-trust/CHECKER-AUDIT.md is removed: the author's generated
checker table replaces it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012tWK9X51ZybvPt7yCm7Zbt
@abu-aska abu-aska changed the title org(review): PR58 answered, with an independent proof of 11, a literature check, and fixes for the five blocking items org(review): PR58, an independent proof of 11 and a literature check Nov 15, 2026

This branch has not been deployed

No deployments
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