The refined graph-container bound at activity of order inverse degree

Let dXd_X and dYd_Y be sufficiently large integers, let δ1\delta\geq 1 and δ,δ>0\delta',\delta”>0, and let Σ=XY\Sigma=X\sqcup Y be a δ\delta-approximately (dX,dY)(d_X,d_Y)-biregular graph satisfying the assumptions of Lemma ML. Write G(a,g)\mathcal{G}(a,g) for the family of sets occurring there and let λ\lambda be the hard-core activity. The refined graph-container conjecture. A bound of the type in Lemma ML should hold for λ=Ω~(1/dX)\lambda=\widetilde{\Omega}(1/d_X); equivalently, there should be a constant κ\kappa and a function λ:NR\lambda^\ast:\mathbb{N}\to\mathbb{R} with λ(d)=Ω~(1/d)\lambda^\ast(d)=\widetilde{\Omega}(1/d) as dd\to\infty such that, under the assumptions of Lemma ML, if λλ(dX)\lambda\geq\lambda^\ast(d_X), then

AG(a,g)λAY(1+λ)gexp{(ga)/dXκ}.\sum_{A\in\mathcal{G}(a,g)}\lambda^{|A|}\leq |Y|(1+\lambda)^g\exp\left\{-(g-a)/d_X^{\kappa}\right\}.

The proved lemma only establishes the corresponding estimate for λ\lambda above a constant multiple of log2dX/dX1/2\log^2 d_X/d_X^{1/2}, and the conjecture would extend the useful range to essentially inverse degree; the authors state that their current methods encounter a barrier at this point.

Sources & referencesView supporting material

Primary source

Matthew Jenssen, Alexandru Malekshahian and Jinyoung Park, “A refined graph container lemma and applications to the hard-core model on bipartite expanders”, arXiv:2411.03393 (2026).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.