Erdős–Pósa for anti-complete (X,Y)-paths

Status open high confidence

Conjecture 6 from arXiv:2512.17232 posits an optimal induced Menger-type theorem: for any $k$, $G$, and $X,Y\subseteq V(G)$, either $G$ contains $k$ pairwise anti-complete $(X,Y)$-paths or there is a set $Z$ of size at most $k-1$ whose closed neighbourhood hits all $(X,Y)$-paths. The source paper itself proves a weaker form with separator bound $78(k-1)$ (improvable to $4(k-1)$ using radius-4 balls), leaving the optimal $k-1$ bound as an open conjecture. No follow-up paper resolving or partially resolving Conjecture 6 was found in the five months since the preprint appeared.

Reviewer notes. The conjecture, if proved, would imply an Induced Grid Minor Theorem for $K_{1,t}$-induced-subgraph-free graphs conjectured by Dallard et al. The paper is only ~5 months old; no follow-up was found in indexed literature. Conjecture 6 is closely related to but distinct from Conjecture 2 in the same paper (the $A$-path version), for which Albrechtsen, Knappe, and Wollan independently announced a partial solution ($d=2$ case). No such concurrent progress on Conjecture 6 (the $(X,Y)$-path variant) was identified.

Auto-reviewed 2026-05-14 with claude-sonnet-4-6 (web search enabled).