On a question of Erdős and Gimbel on the cochromatic number
Annika Heckel
TL;DR
The paper addresses Erdős and Gimbel's question on whether the gap between the chromatic number $\chi(G)$ and the cochromatic number $\zeta(G)$ in the random graph $G_{n,1/2}$ tends to infinity. It proves that this difference is not whp bounded by $n^{1/2-o(1)}$, tying any such bound to the concentration interval length of $\chi(G_{n,1/2})$ through a comparison with the complement graph and Harris's lemma, and employing non-concentration results for $\chi(G_{n,1/2})$. A key proposition shows that any proposed bound $g(n)$ with $\mathbb{P}(\chi(G)-\zeta(G)\le g(n))>0.999$ must eventually satisfy $g(n) > c \frac{\sqrt{n}\log\log n}{\log^3 n}$ along a subsequence, highlighting the link to chromatic concentration. The author conjectures whp $\chi(G)-\zeta(G)=\Theta(n/\log^3 n)$, which aligns with a first-moment heuristic for the thresholds of colorings vs cocolorings in $G_{n,1/2}$.
Abstract
In this note, we show that the difference between the chromatic and the cochromatic number of the random graph $G_{n,1/2}$ is not whp bounded by $n^{1/2-o(1)}$, addressing a question of Erdős and Gimbel.
