On the generalized Tur\'an problem for odd cycles

TL;DR

This paper extends the understanding of the maximum number of odd cycles in graphs avoiding shorter odd cycles, proving that balanced blow-ups of cycles are optimal for large graphs and disproving a related conjecture.

Contribution

It generalizes the Turán problem for odd cycles, establishing conditions under which balanced blow-ups maximize cycle counts and disproving a previous conjecture.

Findings

Balanced blow-ups of cycles maximize the number of cycles for large graphs.

The maximization property fails for smaller graphs without size constraints.

Disproves a conjecture relating cycle maximization to cycle length ratios.

Abstract

In 1984, Erd\H{o}s conjectured that the number of pentagons in any triangle-free graph on $n$ vertices is at most $(n/5)^5$, which is sharp by the balanced blow-up of a pentagon. This was proved by Grzesik, and independently by Hatami, Hladk\'y, Kr\'al', Norine and Razborov. As an extension of this result for longer cycles, we prove that for each odd $k\geq 7$, the balanced blow-up of $C_k$ (uniquely) maximises the number of $k$-cycles among $C_{k-2}$-free graphs on $n$ vertices, as long as $n$ is sufficiently large. We also show that this is no longer true if $n$ is not assumed to be sufficiently large. Our result strengthens results of Grzesik and Kielak who proved that for each odd $k\geq 7$, the balanced blow-up of $C_k$ maximises the number of $k$-cycles among graphs with a given number of vertices and no odd cycles of length less than $k$. We further show that if $k$ and $\ell$ are odd and $k$ is sufficiently large compared to $\ell$, then the balanced blow-up of $C_{\ell+2}$ does not asymptotically maximise the number of $k$-cycles among $C_{\ell}$-free graphs on $n$ vertices. This disproves a conjecture of Grzesik and Kielak.