Long-cycle conjecture for graphs of given average degree

Prove that for every cepsilon>0 there exist d_0,cgamma>0 such that every graph G of average degree dcgeq d_0 satisfies that, when pcgeq(1+cepsilon)/d, the percolated graph G_p contains a cycle of length at least cgamma d asymptotically almost surely as |G|ctocinfinity.

Background

For d-regular graphs, prior work cited in the paper gives cycles of length cOmega(d) in the supercritical percolated graph. The paper proves the analogous component-size result for arbitrary graphs of average degree d, namely the existence of a component of order cOmega(d) when p is at least (1+cepsilon)/d.

The conjecture asks whether the stronger conclusion about a long cycle extends from regular graphs to arbitrary graphs with sufficiently large average degree. The authors mention a partial result showing that percolation at pcgeq C/d yields a cycle of length at least (1-cgamma)d, with cgamma tending to zero as C tends to infinity, and suggest that combining those techniques with the methods of the paper might resolve the conjecture.

References

We reiterate a conjecture of Krivelevich and Samotij (see also ), stating that such a result should still be true if we only assume that the average degree is $d$. \begin{conjecture}[]\label{conj:long cycle} For every $\varepsilon>0$ there exist $d_0,\gamma>0$ such that the following holds. If $G$ is a graph of average degree $d\geq d_0$ and $p\geq (1+\varepsilon)/d$, then $G_p$ contains a cycle of length at least $\gamma d$ a.a.s.\ as $|G| \to \infty$. \end{conjecture}

The critical probability for percolation on finite graphs  (2608.19145 - Christoph et al., 19 Aug 2026) in Conjecture 6.1, Section 6, Concluding remarks (conj:long cycle)