Answers are grounded in this paper and cite their sources.
We determine the largest number V(m, n) of turns of a cycle in the grid graph of m × n lattice points for all 2 ≤ m ≤ n. For m ≡ 2 (mod 4), m ≥ 6, it is mn − 2⌈(2n + m − 2)/6⌉ − 2δ, where δ ∈ {0, 1} is 1 exactly when n lies in an explicit set 𝒳m, which is infinite for m = 6 and has at most five elements for m ≥ 10. For odd m it is (m − 1)n, except that V(m, m) = m(m − 1) − 2 for m ≥ 7; in particular, the maximum on the 19 × 19 Go board is 340. For odd m and even n, and for 4 | m, where V(m, n) = m(n − 1), these are the maxima for Hamiltonian cycles found by Tan and Zhang. For Hamiltonian cycles with m ≡ 2 (mod 4), which they determined up to an additive 2, we obtain the exact maximum for n ∈ 𝒳m and for m ∈ {6, 10, 14}. For m = 10 it is less than V(10, n) exactly when n ≡ 2 (mod 3) and n ∉ 𝒳10, so on these infinitely many boards every cycle with the most turns skips lattice points; for m = 14 this happens only on [20] × [14]. We also disprove their conjecture that the error depends eventually only on n − m modulo 6. All proofs are by hand, and the finite verifications they use are printed in full. 2020 Mathematics Subject Classification: Primary 05B50; Secondary 05C38, 05C45, 52C05. Files: the paper (PDF) and a reproduction archive with programs and their stored outputs that independently confirm the exact values, the constructions, the printed certificates and the figures of the paper, including a checker that re-verifies from the PDF every construction datum printed in it; no proof depends on these programs. See README.md in the archive. Assisted by AI.
No takes yet. Share an insight, caveat, or question.
Sungsoo Na (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: