ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let X1∼p1(X1;θ1)X_1 \sim p_1(X_1; \theta_1) defined for all θ1∈R\theta_1 \in \mathbb{R} and X2∼p2(X2;θ2)X_2 \sim p_2(X_2; \theta_2) defined for all θ2∈R\theta_2 \in \mathbb{R} be random elements of a general nature. Define the Fisher Informations in X1X_1 and X2X_2 as I1I_1 and I2I_2, respectively, i.e.\ for i=1,2i=1,2,

Ii=E[(δδθilog⁡pi(Xi;θi))2].I_i = \mathbb{E}\left[\left(\frac{\delta}{\delta \theta_i} \log p_i(X_i; \theta_i)\right)^2\right].

Consider a statistics U=f(X1,X2)U = f(X_1, X_2) with a property

U∼p(u;θ1,θ2)=p(u,θ1+θ2).U \sim p(u; \theta_1, \theta_2) = p(u, \theta_1 + \theta_2).

For such UU that satisfies the above, a version of (a generalization of) the Stam Inequality holds:

1IU(θ)≥1I1(θ)+1I2(θ).\frac{1}{I_U(\theta)} \geq \frac{1}{I_1(\theta)} + \frac{1}{I_2(\theta)}.

It is of interest to see if for UU with g(θ1,θ2):Θ1×Θ2→Θg(\theta_1, \theta_2): \Theta_1 \times \Theta_2 \rightarrow \Theta,

U∼p(u;θ1,θ2)=p(u,g(θ1,θ2)),U \sim p(u; \theta_1, \theta_2) = p(u, g(\theta_1, \theta_2)),

is there an analogy of the Stam Inequality?

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.