ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let Gn\mathbb{G}_n be the set of graphs with nn vertices, and let G(n,1/2)G(n,1/2) and G(n,1/2,k,q)G(n,1/2,k,q) denote the null and alternative graph distributions, respectively. For a sequence k=knk=k_n satisfying k≤nβk\le n^\beta for some 0<β<1/20<\beta<1/2, let q∈(1/2,1]q\in(1/2,1], and consider tests ψn:Gn→{0,1}\psi_n:\mathbb{G}_n\to\{0,1\}, where P0\mathbf{P}_0 and P1\mathbf{P}_1 denote probabilities under the null and alternative hypotheses.

The dense subgraph detection conjecture. For any sequence of randomized polynomial-time tests (ψn)n(\psi_n)_n,

lim inf⁡n→∞{P0(ψn(A)=1)+P1(ψn(A)=0)}≥13.\liminf_{n\to\infty}\left\{\mathbf{P}_0(\psi_n(A)=1)+\mathbf{P}_1(\psi_n(A)=0)\right\}\ge \frac{1}{3}.

Equivalently, no randomized polynomial-time algorithm can correctly identify the dense subgraph with probability tending to 11. This conjecture is used to establish a computational lower bound for matrix logistic regression; its resolution is not given in the source.

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.