Oriented trees in $O(k \sqrt{k})$-chromatic digraphs, a subquadratic bound for Burr's conjecture
Stéphane Bessy, Daniel Gonçalves, Amadeus Reinald
TL;DR
The paper advances Burr's conjecture by establishing a subquadratic universal bound for oriented trees in digraphs with large chromatic number. It introduces gluing lemmas for leaves and directed/oriented paths, forming a win-win induction that either attaches many leaves or decomposes into few directed paths to grow the target tree while controlling the bound. The main results include a subquadratic universal bound of $8 \sqrt{\frac{2}{15}} k\sqrt{k} + \frac{11}{3}k + \sqrt{\frac{5}{6}} \sqrt{k} + 1$ for general oriented trees, an improved arborescence bound of $\sqrt{\frac{4}{3}} k \sqrt{k} + k/2$, and a linear bound $(b-1)(k-3)+3$ for $b$-block paths ($b\ge 2$). These methods provide a framework that combines structural decompositions with constructive path-gluing to push towards tighter universality results and illuminate directions for further improvements.
Abstract
In 1980, Burr conjectured that every directed graph with chromatic number $2k-2$ contains any oriented tree of order $k$ as a subdigraph. Burr showed that chromatic number $(k-1)^2$ suffices, which was improved in 2013 to $\frac{k^2}{2} - \frac{k}{2} + 1$ by Addario-Berry et al. We give the first subquadratic bound for Burr's conjecture, by showing that every directed graph with chromatic number $8\sqrt{\frac{2}{15}} k \sqrt{k} + O(k)$ contains any oriented tree of order $k$. Moreover, we provide improved bounds of $\sqrt{\frac{4}{3}} k \sqrt{k}+O(k)$ for arborescences, and $(b-1)(k-3)+3$ for paths on $b$ blocks, with $b\ge 2$.
