Flip colouring of graphs II
Xandru Mifsud
TL;DR
This paper advances the theory of flip colourings by (i) producing small $(b,r)$-flip graphs for $4\le b<r<b+2\left\lfloor\frac{b+2}{6}\right\rfloor^2$ via Cayley-graph constructions rooted in sum-free subsets, yielding a new upper bound on $h(b,r)$, and (ii) establishing both upper and lower bounds on $q(k)$, the extent of fixed early terms in $k$-flip sequences, including a lower bound showing $q(k) \gtrsim k/4$ and a general upper bound $q(k) < k/3$ (in specific residue classes) or $< \lceil k/2\rceil$ otherwise. The methods synthesize graph products, packing, and Abelian-group Cayley graphs with sum-free sets to connect local neighbourhood constraints to global realizability, enabling controlled constructions and unbounded behaviour in large final colours. The results illuminate the growth of $h$ and $q(k)$, reveal non-polynomial growth in some regimes, and pose open problems regarding polynomial bounds tied to the index $p(k)$ and the feasible parameter ranges for $h(b,r)$. The work integrates combinatorial design with algebraic graph theory to push the boundary of what is constructible under flip-colouring constraints.
Abstract
We give results concerning two problems on the recently introduced \textit{flip colourings of graphs}. For positive integers $b, r$ with $b < r$, we say that a $b + r$ regular graph is a $(b,r)$-\textit{flip graph} if there exists a red/blue edge colouring such that the red degree of every vertex is $r$, the blue degree of every vertex is $b$, yet in the closed neighbourhood of every vertex there are more blue edges than red edges. We prove that for integers $b, r$ with $4 \leq b < r < b + 2 \left\lfloor\frac{b+2}{6}\right\rfloor^2$, small constructions of $(b,r)$-flip graphs on $Θ(b+r)$ vertices are possible. Furthermore, we prove that there exist $k$-flip sequences $(a_1, \dots, a_k)$ where $k > 4$, such that $a_k$ can be arbitrarily large whilst $a_i$ is constant for $1 \leq i < \frac{k}{4}$.
