ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let G(n,a/n,b/n)\mathcal{G}(n,a/n,b/n) be the planted partition model with kk communities, where a>ba>b. Let c(k)c(k) be a threshold depending only on kk. Decelle–Krzakala–Moore–Zdeborová's Potts-model conjecture. For any kk, there exists c(k)c(k) such that

(a−b)2a+(k−1)b<c(k)\frac{(a-b)^2}{a+(k-1)b}<c(k)

implies that the clustering problem cannot be solved;

c(k)<(a−b)2a+(k−1)b<kc(k)<\frac{(a-b)^2}{a+(k-1)b}<k

implies that it is solvable but not in polynomial time; and

(a−b)2a+(k−1)b>k\frac{(a-b)^2}{a+(k-1)b}>k

implies that it can be solved in polynomial time. When k≤4k\leq4, c(k)=kc(k)=k, so the intermediate case does not occur; when k≥5k\geq5, c(k)<kc(k)<k. This conjecture predicts distinct information-theoretic and computational transitions for the multi-community model, motivated by phase transitions in the Potts model on trees. The source notes that non-reconstruction results are less precise for k>2k>2 and gives no resolution of the conjecture.

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.