Multicolour Erdős--Hajnal Conjecture
Status partial high confidence
The multicolour Erdős–Hajnal conjecture remains open in general, even for two colours. Since the problem was posted (2019), Axenovich, Snyder, and Weber confirmed the conjecture for three-colour edge-colorings of $K_n$ that avoid any family of at most three coloured triangle patterns; separately, Axenovich and Weber reduced the full conjecture to the case where the host clique uses at most one more colour than the forbidden pattern, and exhibited surprising non-monotone behaviour in homogeneous-set sizes.
Cited literature (3)
-
Confirms the multicolour Erdős–Hajnal conjecture for three-colour edge-colorings of $K_n$ that avoid any family of at most three coloured triangle patterns, providing asymptotically tight bounds.
-
Reduces the multicolour EH-conjecture to the case where the host-clique colour count is equal to or exactly one more than that of the forbidden pattern, and demonstrates that allowing an extra colour can non-monotonically decrease the size of the largest homogeneous set.
-
Formulates and partially resolves a geometric analogue of the multicolour EH-conjecture for finite projective geometries, proving a Gallai-type decomposition for rainbow-triangle-free three-colourings of PG(n-1,2).
Reviewer notes. The two ScienceDirect URLs returned HTTP 403; they appear to be the journal-published versions of arXiv:2005.09269 (Discrete Math. 2021) and arXiv:2311.03249 (Discrete Math. 2025) respectively, but could not be confirmed — both papers are cited via their verified arXiv pages only. The 2025 projective-geometry paper (2505.13781) addresses a geometric analogue rather than the original graph-theoretic statement. The original conjecture remains open for all k≥2 and all colourings χ, including the two-colour case.
Discussion
See [FGP].
Bibliography
-
[FGP]Jacob Fox, Andrey Grinshpun and János Pach: The Erdős–Hajnal conjecture for rainbow triangles, J. Combin. Theory, Series B. 111 (2016), 75--125.