s(k² − 4) = k for every k ≥ 8 in the Lean kernel: ValidTilt9 (Hosono, after Daniel): tracking an unreported result

Open
#643 2 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
5/5
Estimated time
Over a week
Newbie friendliness
8/100
Issue type
Feature
Clarity
Needs clarification
Activity status
Active
Domain
testing

Research direction

Start by reading the three theorems in SquarePacking.LemmaELeaf under problems/square-lower-bounds/certificates/k2m4_n77_L9/lean/ and checking them against library.sha256 and the retained Basic.lean, Bentz4.lean and Bentz4Data.lean. The open items are retention, registration, comparison with evand/square-packing and statement review. The full kernel replay is on hold because it needs about 7,400 CPU-hours and 230 GB.

Written by the indexing model from the issue text.

Description

intake

This issue tracks a significant result that reached the project through a watched repository rather than an issue. Its author is Hiroaki Hosono (@wand125), building on Evan Daniel's reduction and certificate, and the source is wand125/square-packing at 319a779 (the proof is added in 2c75edc and its receipts in e99f5e3, both of 16 October 2026). We opened it so that the result's intake, review and validation have one public place.

The claim. A Lean 4 kernel proof, problems/square-lower-bounds/certificates/k2m4_n77_L9/lean/, of three theorems in SquarePacking.LemmaELeaf:

  • validTilt9 : Bentz4.ValidTilt9, the finite premise of Evan Daniel's k² − 4 theorem;
  • valid9 : Bentz4.Valid9;
  • bentz4 : ∀ k : ℕ, 8 ≤ k → minSide (k ^ 2 - 4) = k, that is, s(k² − 4) = k for every k ≥ 8.

The source's axiom receipts list only propext, Classical.choice and Quot.sound, with no sorry and no native_decide: each leaf and row group is checked by decide +kernel.

What has been checked here (a read-only triage, 16 October):

  • The proof's statement files match what the project already holds. The import closure's library.sha256 lists 59 files: 56 Lean modules and 3 build files. The project retains 19 of them from Evan Daniel's 3 October release, 16 Lean modules (among them Basic.lean, S32.lean, ValidSplit9.lean, Bentz4.lean and Bentz4Data.lean) and the 3 build files, and each is byte-identical to the project's copy. So the three theorems use the same definitions as the project's own 3 October build of the reduction.
  • Nothing has been built or replayed yet.

What is open:

  • Retaining the lean/ directory at that commit, with the release validtilt9-lean-v1 assets pinned by digest.
  • Registering the result as reported, as a formalization beside Evan Daniel's theorem.
  • Comparing the other 40 files of the import closure, the proof machinery, with evand/square-packing.
  • A review of the statement binding.
  • The full kernel replay: about 7,400 CPU-hours, and about 230 GB of memory for the final assembly, by the source's receipts. That is far above the project's compute cap and this project's machines, so it is on hold. A sampled rebuild of a few roots may run later. It would check components only and would not move a rung.

We will post the result's id and rungs here once it is on main, and update the open items as they close.


Generated by Claude Code

Dominant language
Python
Stars
54
Forks
6
Avg merge
9h 11m
Merged PRs (30d)
239

Getting set up

  • No Dockerfile or Docker Compose file
  • Has a pull request template
  • No contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from jlevy/squares

All issues in jlevy/squares

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.