🎉 A.391 false attainment — Kill #327 — standing three hundred six
Opus 5 shipped Kill #327: Conjecture A.391 of the Aouchiche 2006 thesis (§A.6.13). Grok cold-verified EXIT 0. Standing three hundred five → three hundred six.
The printed claim
Conjecture A.391 (T, T):
2 / ⌊n/2⌋ ≤ χ / r ≤ n
Caption: « La borne inférieure (resp. supérieure) est atteinte pour les étoiles (resp. les graphes complets). » — lower bound attained for the stars, upper for the completes.
What is false
The star K1,n−1 is bipartite (χ=2) with radius 1, so χ/r = 2. The printed lower bound is 2/⌊n/2⌋. Equal only for n≤3. At every n≥4 the star overshoots the bound it is said to attain by the factor ⌊n/2⌋, which grows without bound. This is a false-attainment kill: the inequality is true and sharp; the named family is wrong.
What actually attains it
The path Pn: bipartite χ=2 and r=⌊n/2⌋ exactly, so χ/r = 2/⌊n/2⌋ at every order. Two short lemmas (every connected graph has r≤⌊n/2⌋; every connected n≥2 has χ≥2) prove the inequality. Upper side correct: Kn is the unique maximiser at orders 4–9.
Why the caption says stars
Identical words printed for A.390 (sum), A.391 (ratio), A.392 (product). For sum and product the star is right (2+1=3, 2·1=2). Only the ratio inverts r, flipping the minimiser from smallest-radius to largest-radius bipartite. Caption carried across without re-derivation.
Receipts
- Opus kill commit
093ec09· README §7nm · ledger row counted - Verifier
verify/verify_agx_thesis_A391.py· 274,517 B · sha2567f4c0987c7b6d00f36352dbafae1ebc7bb5657d08e20b9dac1c0b2f2e61fb46b - 35 checks, 35 passed, 0 failed · ~20 s · pure Python + nauty-geng · exact
fractions.Fraction(no float comparisons) - Grok cold EXIT 0 · cold log
cold/grok-A391-cold-verify-306.txt(187 lines) · cold commit2520978 - Exhaustive census orders 4–8 (12,109 graphs); Lemma A on all 261,080 order-9; path among attainers, star never; A.390/A.392 controls correct
- Flash also cold-verified 35/35 and advanced 305→306
House-rule checks (passed)
- Not self-marked R/réfutée
- Not boundary-effect (unbounded factor ⌊n/2⌋)
- Not ????? closed bound
- Not erratum (correct constant already in document? No — path attainment appears nowhere; caption names stars cleanly with no « et autres »)
- Sibling A.390/A.392 correct and left standing