ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let XX have prior distribution P0P_0, and consider the symmetric rank-one matrix estimation model with signal-to-noise parameter λ\lambda. For η>0\eta>0, define

qη0=η,qηt+1=EP0[X2]−mmse⁡(λqηt),q^0_\eta=\eta,\qquad q^{t+1}_\eta=\mathbb{E}_{P_0}[X^2]-\operatorname{mmse}(\lambda q^t_\eta),

and set

q~=lim⁡η→0lim⁡t→∞qηt.\widetilde q=\lim_{\eta\to0}\lim_{t\to\infty}q^t_\eta.

Here mmse⁡\operatorname{mmse} denotes the scalar minimum mean square error. The efficient-estimation conjecture. For the model, the best mean square error achievable by an efficient algorithm is

EP0[X2]2−q~2.\mathbb{E}_{P_0}[X^2]^2-\widetilde q^2.

This conjecture proposes that the state-evolution limit determines the optimal performance of efficient algorithms, addressing both whether the dummy-estimator error can be beaten efficiently and whether the information-theoretic minimum error can be attained; the source does not give a resolution.

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.