T-250: Hosono reports a Lean kernel proof of Daniel's Valid7, and so of s(k2−3)=k for every k≥6

V0 C0 Formalization recorded

2026-10-17 published · Hosono after Daniel, Burns, Massaccesi; Hosono after Daniel, Levy, Burns, Massaccesi · 13 cases, n=33 to 321

Hosono (wand125) reports a Lean 4 kernel proof of Daniel's finite premise Valid7, and with Daniel's reduction a proof that s(k^2 - 3) = k for every integer k >= 6: a formalization of T-064, which this record holds at V3/C3 on two exact checkers of Valid7 and a build here of the reduction. Its theorems, in namespace SquarePacking.LemmaELeaf, are valid7 : Bentz.Valid7, bentz : ∀ k : ℕ, 6 ≤ k → minSide (k ^ 2 - 3) = k, and validTilt7 : Bentz.ValidTilt7.

This entry's scope is that theorem's projection onto the cases this record holds, k = 6 to 18, as T-064's is: s(33)=6, s(46)=7, s(61)=8, s(78)=9, s(97)=10, s(118)=11, s(141)=12, s(166)=13, s(193)=14, s(222)=15, s(253)=16, s(286)=17 and s(321)=18.

The proof checks the pose tree of the third k2m3 run on the box cover of L4_k02_box7.txt, one generated module per root cell, 9,800 of them, glued over the D4 root [0, 7/2]^2 x (0, 1/2] in a generated Main.lean, with Bridge.lean identifying the cover with Bentz.box7Cover.

Version 1, merged into Daniel's repository on 11 October, generates the root modules from the per-root certificates of the release data/valid7-lean-certs-v1 and reports a build of roughly 10^4 CPU-hours, later put at about 6,650. Version 2, in wand125/square-packing on 17 October, re-checks the same theorem from the same certificates with re-certified leaves on Daniel's library with a faster row-minimum check, 62,626 generated files, and a recorded run of about 798 CPU-hours. Both report the axioms propext, Classical.choice and Quot.sound, no sorry and no native_decide. Nothing here has built either.

Here the names were audited: the 9,800 pickles of version 1's release asset and version 2's 9,800 root modules are R00000 to R09799, one per root of the run's leaves file, which fills the 35 x 35×8 grid gen_main.py asserts.

Hosono after Daniel, square-packing and square-packing, 11 and 17 October 2026. Both commits name Claude as co-author.

Significance, composition and next rung
Significance
A machine-checked proof, by another author, of the finite premise of a bound family the record holds on two exact checkers and a Lean reduction: with it the whole theorem s(k^2 - 3) = k for every k >= 6 is in Lean, unconditional. A machine audit, S3, below T-064's S4 because it proves no new bound, as T-236 is scored.
Composition
Two cited entries, the source's reports of version 1 (E-k2m3-wand125-valid7-lean-v1-report) and version 2 (E-k2m3-wand125-valid7-lean-v2-report), each claiming the exact values in Lean, the grid inside the reduction. T-064's checkers and its reduction build are not cited: they are T-064's evidence. No build is retained here, so V0; no review has read it, so C0.
Next rung
C1 needs the stage 4 mathematical review of valid7, Bridge.lean, the glue in gen_main.py and the generators, read from the retained packets. V3 and C3 need a complete build here or by a third party retained here, with the axiom receipt: version 2's is about 798 CPU-hours with 37 GB per leaf module and 51 GB for Main.lean, about 20 times the owner's 40 CPU-hour item cap, and a sample of roots is a component check that moves no rung; which to run, if either, is the owner's decision (think-qu37, #537). Rung 5 needs a human expert's review of the formalization beside it.
Unfinished confirmations
C1 (think-0ayk): the stage 4 mathematical review of valid7, bentz and validTilt7, Bridge.lean, the glue of gen_main.py and the generators emit_v7.py and run_v7.py, read from the retained packets. Not priced.

C3 (think-qu37): a complete build of version 2 from its release's generated sources on its library commit, verify.py's checks of every module, Main.lean and Axioms7.lean, with the axiom receipt, here or by a third party whose build is retained here. About 798 CPU-hours (priced 2026-10-26).
Novelty
previously-published Present in an identified source

The cases

This result concerns 13 cases, too many to draw one by one. Each is listed with the film’s bounds, the proved lower bound and the best known side, and links to its case record, where its packing and number line are drawn.

Results on these cases

16 results in the register on these cases, oldest first
  1. 2005 published T-007 · 13 of these cases

    s(n)≥min(⌈n⌉,n−2⌊n⌋+1+1) for 4≤n≤324

    Lower bound incomplete on these cases, superseded by T-008, T-063, T-064 and T-067

    Nagamochi · Nagamochi 2005 · source · register

  2. 2010-09-13 published T-004 · case 46

    Bentz 2010, Theorem 8 (s(46)≥7) is correct as printed, machine-audited in full

    Audit confirmed

    Bentz · Bentz 2010 · source · register

  3. 2010-09-13 published T-008 · case 46

    s(46)=7

    Optimality confirmed

    Bentz · Bentz 2010 · source · register

  4. 2018 published T-087 · case 61

    s(37)≥53/2+22−1 and s(61)≥73/2+22−1, by optimal piercing

    Lower bound reviewed superseded by T-063

    Bašić, Slivková · Basic-Slivkova 2018 · source · register

  5. 2026-09-04 published T-085 · 13 of these cases

    Nagamochi 2005, Lemma 1 is false for every container with a>3 and b>2

    Correction confirmed

    Karakuş; Vlasenko · Karakuş 2026 · chelokot Nagamochi counterexample 2026 · packet · register

  6. 2026-09-27 published T-045 · cases 61, 78

    Rectangle-density lower bounds replayed at 15 counts in n=18…78

    Lower bound confirmed superseded by T-063, T-064 and T-067

    Hosono after Tokoharu, Levy, Stromquist, Nagamochi, Burns, Massaccesi · wand125 rectangle bounds 2026 · packet · packet · source · review · register

  7. 2026-09-27 published T-046 · cases 61, 78

    Rectangle-density lower bounds verified at 48 counts in n=18…95

    Lower bound confirmed superseded by T-063, T-064 and T-067

    Hosono after Tokoharu, Levy, Stromquist, Nagamochi, Burns, Massaccesi · wand125 rectangle bounds 2026 · wand125 rectangle bounds 2026-09-28 · packet · packet · source · review · register

  8. 2026-09-28 published T-063 · case 61

    s(61)=8, by monotonicity from T-062

    Optimality confirmed

    Daniel after Burns, Massaccesi · evand square-packing 2026-10-01 · packet · packet · packet · source 1 · source 2 · review 1 · review 2 · register

  9. 2026-09-29 published T-058 · 5 of these cases

    Rectangle-certificate ceiling α·UB(n) proved for n=1..100; B·UB(n) on 64 grid rows

    Method limit confirmed

    Hosono after Tokoharu, Daniel · wand125 tools 2026 · packet · register

  10. 2026-09-29 published T-083 · 13 of these cases

    s(n)≥1/2+n−⌊n⌋+1/4 for every nonsquare 8≤n≤324

    Lower bound confirmed on these cases, superseded by T-008, T-063, T-064 and T-067

    Karakuş · Karakuş 2026 · source · register

  11. 2026-09-30 published T-064 · 13 of these cases

    s(k2−3)=k for every integer k from 6 up; k=6…18 are the cases held here

    Optimality confirmed

    Daniel after Burns, Massaccesi · evand square-packing 2026-10-01 · packet · packet · packet · packet · source 1 · source 2 · review 1 · review 2 · register

  12. 2026-10-01 published T-067 · case 78

    s(77)=9, by a mixed cover extended from Daniel's s(60) cover

    Optimality confirmed

    Hosono after Daniel, Levy, Stromquist, Nagamochi, Burns, Massaccesi · wand125 exact covers 2026-10-01 · packet · source · review · register

  13. 2026-10-07 published T-124 · 13 of these cases

    Reported non-strict local minima for 178 source configurations

    Restricted optimality recorded

    Daniel after Couzo · Daniel exact and local reports 2026 · packet · register

  14. 2026-10-12 published T-261 · 13 of these cases

    Reported grade-C local minima for 233 source configurations of field degree at most 4

    Restricted optimality incomplete

    Daniel after Schadt, Ellsworth, Levy · Daniel exact and local reports 2026-10-12 · packet · packet · register

  15. 2026-10-13 published T-180 · 11 of these cases

    Closed-form lower bound s(N)≥a(N) for every non-square N≥49, reported

    Lower bound reviewed on these cases, superseded by T-063, T-064 and T-067

    Hosono · wand125 closed form 2026-10-14 · packet · packet · register

  16. 2026-10-17 published T-250 this result · 13 of these cases

    Hosono reports a Lean kernel proof of Daniel's Valid7, and so of s(k2−3)=k for every k≥6

    Formalization recorded

    Hosono after Daniel, Burns, Massaccesi; Hosono after Daniel, Levy, Burns, Massaccesi · wand125 Valid7 Lean 2026-10-11 · wand125 Valid7 Lean v2 2026-10-17 · packet · packet · source 1 · source 2 · register