ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let GNK\mathcal G^K_{\mathcal N} denote the class of Gaussian-dependent p-value vectors indexed by the hypotheses, and define

fK0,K(α):=sup⁡P∈GNKE[FDαRDα],f_{K_0,K}(\alpha):=\sup_{\mathbf P\in\mathcal G^K_{\mathcal N}}\mathbb E\left[\frac{F_{\mathcal D_\alpha}}{R_{\mathcal D_\alpha}}\right],

where K0K_0 is the number of true null hypotheses. Let

fK(α):=max⁡K0∈KfK0,K(α).f_K(\alpha):=\max_{K_0\in\mathcal K}f_{K_0,K}(\alpha).

Gaussian BH-FDR bound. There exists c⩾1c\geqslant 1 such that

fK(α)⩽cαf_K(\alpha)\leqslant c\alpha

for all α∈(0,1)\alpha\in(0,1) and K∈NK\in\mathbb N. This would give a dimension-free linear bound on the worst-case false discovery rate under Gaussian dependence. The paper states that the value of cc, particularly at small significance levels, is important, but whether such a universal constant exists remains open.

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.