Pure pair in triangle-free 3-blockade

Question 1.3 · arXiv:2111.00532

arXiv Question high confidence— first stated 2024-02-06

Authors: Alex Scott, Paul Seymour, Sophie Spirkl

Status open high confidence

Question 1.3 from Pure pairs IX asks whether, in any 3-blockade with no transversal triangle, there exist distinct blocks $B_i, B_j$ containing a pure pair of polynomial size in the blockade width. The authors note it is a variant of a Conlon–Fox–Sudakov conjecture, equivalent to the $k=3$ case of Theorem 1.4, and reduce it to a sparse case, but leave it open. No follow-up paper resolving the question was found in a broad web search through May 2026.

Reviewer notes. No follow-up found. The conjecture is the k=3 triangle case of a family of results about transversal cycles in blockades; the k>=4 cases are handled by Theorem 1.4 of the same paper. The authors reduce Question 1.3 to the sparse case but do not resolve it.

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

Question. Do there exist $c, \varepsilon > 0$ with the following property? Let $\mathcal{B} = (B_1, B_2, B_3)$ be a blockade in a graph $G$, and let $W$ be its width. If there is no $\mathcal{B}$-transversal triangle, then there exist distinct $i, j \in \{1, 2, 3\}$ and $X \subseteq B_i$ and $Y \subseteq B_j$, such that $X, Y$ is a pure pair, and $|X|, |Y| \geq \varepsilon W^c$.

Context

This is a variant of a conjecture of Conlon, Fox and Sudakov. Settling Question 1.3 was the authors' initial goal, though they were unable to do so. It is also equivalent to the $k=3$ case of Theorem 1.4 (which handles cycles of length $k \geq 4$), and can be reduced to the sparse case where every vertex in $B_i$ has at most $|B_j|/100$ neighbours in $B_j$ for all distinct $i, j \in \{1,2,3\}$.

Source paper

Pure pairs. IX. Transversal trees
Alex Scott, Paul Seymour, Sophie Spirkl · 2024-02-06
https://arxiv.org/abs/2111.00532 PDF source