ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let (Yk)kotinV(Y_k)_{k otin V} be a family of NN random variables admitting a dependency graph of degree D∈Z≥0D\in\mathbb{Z}_{\geq 0}. Assume that there exists δ∈(2,3]\delta\in(2,3] such that Aδ<∞\mathcal{A}_\delta<\infty, where Aδ=∑k∈VE[∣Yk−ck∣δ]\mathcal{A}_\delta=\sum_{k\in V}\mathbb{E}[|Y_k-c_k|^\delta], and let WW be the standardized sum and ξδ\xi_\delta the renormalized standard deviation defined by

ξδ=(NAδ)1/δVar⁡(S)N(D+1).\xi_\delta=\left(\frac{N}{\mathcal{A}_\delta}\right)^{1/\delta}\sqrt{\frac{\operatorname{Var}(S)}{N(D+1)}}.

Berry–Esseen dependency-graph conjecture. The bound

dKol(W,N(0,1))≲ξδ−δ(D+1N)(δ−2)/2d_{\mathrm{Kol}}(W,\mathcal{N}(0,1))\lesssim \xi_\delta^{-\delta}\left(\frac{D+1}{N}\right)^{(\delta-2)/2}

should hold.

This would extend the finite-moment Berry–Esseen estimate from independent variables to variables with a sparse dependency graph, with the effective sample-size factor N/(D+1)N/(D+1). The paper presents this as a conjecture; no resolution is supplied in the given text.

Record

Comments

Nothing has been published against this problem yet, and nothing has been said about it.

Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.

Sign in

Solve with an agent

Open the statement in a chat, with the problem and the ground rules already written into the prompt.

This opens a third-party site. Nothing is posted back to ProbXiv and nothing you write there is recorded here — what a model gives you is an attempt, which a person still has to check.