Super-polynomial growth of 3-uniform complete Ramsey

Conjecture 4.6 · arXiv:2404.02021

arXiv Conjecture high confidence— first stated 2024-04-02

Status open high confidence

No post-2024 paper resolving Conjecture 4.6 was found. The conjecture asks whether every polynomial growth rate $2^{n^c}$ is achievable as a lower bound for $r(K_s^{(3)}, K_n^{(3)})$ for some fixed $s$ depending on $c$. The closest follow-up work (arXiv:2411.13812) studies the complementary polynomial-growth regime for general $H$, and arXiv:2603.16069 achieves quasipolynomial rates for specific non-complete hypergraphs; neither directly addresses the complete-hypergraph setting of Conjecture 4.6. The sole internal reference (arXiv:2312.15572) concerns a different conjecture about VC-dimension and is unrelated.

Reviewer notes. Two related follow-up papers were found but do not resolve Conjecture 4.6: arXiv:2411.13812 ('When are off-diagonal hypergraph Ramsey numbers polynomial?', 2024) characterizes polynomial growth for general H, the complementary direction; arXiv:2603.16069 ('Hypergraph Ramsey numbers with quasipolynomial growth rate', Xiaoyu He, 2026) constructs specific 3-graphs achieving n^{Theta(log n)} Ramsey numbers, demonstrating intermediate growth rates but for non-complete hypergraphs. Neither paper addresses the complete hypergraph K_s^{(3)} setting of Conjecture 4.6. The conjecture is closely related to the stepping-up paradigm (Erdős-Hajnal-Rado) applied to complete hypergraphs, but no explicit resolution via that approach was located in the indexed literature.

Auto-reviewed 2026-05-15 with claude-sonnet-4-6 (web search enabled).

Conjecture. For every $c>0$, there exists $s$ such that $r(K_{s}^{(3)},K_{n}^{(3)})\geq 2^{\Omega_{s}(n^{c})}$.

Context

The authors ask what growth rates are possible for off-diagonal hypergraph Ramsey numbers, noting that for the family $\mathcal{H}$ of links of nonbipartite graphs the rate $2^{\Theta_{\mathcal{H}}(n)}$ is achievable. They contrast this with the tripartite setting, where $r(H,K_{n,n,n}^{(3)})\leq 2^{O_H(n^2)}$ for all $H$, and note the conjecture fails with $K_{n,n,n}^{(3)}$ in place of $K_{n}^{(3)}$.

Source paper

On off-diagonal hypergraph Ramsey numbers
David Conlon, Jacob Fox, Benjamin Gunby, Xiaoyu He, Dhruv Mubayi, Andrew Suk, Jacques Verstraete · 2024-04-02
https://arxiv.org/abs/2404.02021