Period separation for the corrected cross-one-off block, without a length #
hypothesis
crossOneOff_cutoff_le_torsionOrder (Bananas/CrossOneOffPeriodSeparation.lean)
proves crossOneOffCutoff g (B.length beta) ≤ k for the near-opposite marking
(v_{α,1}, v_{β,n_β-1}) only under the extra hypothesis
hBetaLong : g + 1 ≤ B.length beta. This file removes that hypothesis: the
sharp bound holds for every torsion witness whenever the two strand lengths
are not both equal to two.
The argument (Bananas/FORMALIZATION_NOTES.md, "the short-strand period separation is
provable — closed-form torsion order") extracts from the slope framework of
Bananas/BananaTorsionSlopes.lean the exact identity m = p + r + Σ_γ q_γ
where p := d/a, r := d/b (d a common multiple of the two marked
lengths) and each q_γ a common-multiple share of |rise| over the other
strands, then a ray/primitivity argument on the two integers
D := lcm(a,b) - a/gcd(a,b) - b/gcd(a,b) and E := L + Σ_γ L/n_γ
(L := lcm of the other lengths).
Generic finite-sum helpers #
Pure arithmetic lemmas #
The Banana-specific extraction and assembly #
Core lemma: every torsion witness of the near-opposite cross-one-off
marking satisfies the closed-form cutoff bound, without any hypothesis
relating the two marked strand lengths beyond the two lengths not both
being 2.
Corollary: the closed-form period-separation bound for the exact torsion
order, replacing the long-second-strand hypothesis hBetaLong of
crossOneOff_cutoff_le_torsionOrder by the weaker hNotBoth.