Cops and Attacking Robbers with Cycle Constraints
Alexander Clow, Melissa A. Huggan, M. E. Messinger
TL;DR
This work advances the study of the Cops and Attacking Robbers variant by delivering a structural characterization of triangle-free graphs with attacking cop number at most 2, leveraging domination-elimination concepts. It proves a tight bound for bipartite planar graphs, showing $cc(G)\le 4$ with a constructive tight example, and develops guardability lemmas for geodesic paths to support planar strategies. The authors also demonstrate a concrete separation between the attacking cop number and the classical cop number by constructing 17 graphs of order 58 with $cc(H)=6$ and $c(H)=3$, establishing $cc(H)-c(H)\ge 3$, and provide a lower-bound framework via $G^2-E$ that informs these separations. The paper closes with conjectures and open problems that chart a rich research agenda on triangle-free, planar, and general graphs in pursuit–evasion settings. The results offer new tools and directions for understanding how attack power interacts with traditional guarding and domination parameters in graph pursuit games.
Abstract
This paper considers the Cops and Attacking Robbers game, a variant of Cops and Robbers, where the robber is empowered to attack a cop in the same way a cop can capture the robber. In a graph $G$, the number of cops required to capture a robber in the Cops and Attacking Robbers game is denoted by $\attCop(G)$. We characterise the triangle-free graphs $G$ with $\attCop(G) \leq 2$ via a natural generalisation of the cop-win characterisation by Nowakowski and Winkler \cite{nowakowski1983vertex}. We also prove that all bipartite planar graphs $G$ have $\attCop(G) \leq 4$ and show this is tight by constructing a bipartite planar graph $G$ with $\attCop(G) = 4$. Finally we construct $17$ non-isomorphic graphs $H$ of order $58$ with $\attCop(H) = 6$ and $\cop(H)=3$. This provides the first example of a graph $H$ with $\attCop(H) - \cop(H) \geq 3$ extending work by Bonato, Finbow, Gordinowicz, Haidar, Kinnersley, Mitsche, Prałat, and Stacho \cite{bonato2014robber}. We conclude with a list of conjectures and open problems.
