ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let K≥2K\geq 2, let R1,…,RKR_1,\ldots,R_K be the link rates, and let ϵ1,…,ϵK\epsilon_1,\ldots,\epsilon_K be the type-I error thresholds at the KK decision centers. Let E∗(R1,…,RK,ϵ1,…,ϵK)\mathcal{E}^*(R_1,\ldots,R_K,\epsilon_1,\ldots,\epsilon_K) denote the fundamental type-II error-exponents region, and choose a permutation π ⁣:{1,…,K}→{1,…,K}\pi\colon\{1,\ldots,K\}\to\{1,\ldots,K\} such that

ϵπ(1)≥ϵπ(2)≥⋯≥ϵπ(K),\epsilon_{\pi(1)}\geq\epsilon_{\pi(2)}\geq\cdots\geq\epsilon_{\pi(K)},

with ϵπ(0):=1\epsilon_{\pi(0)}:=1. For i∈{1,…,K}i\in\{1,\ldots,K\}, define

ℓi∗:=max⁡ℓ{ℓ ⁣:ℓ∈{π(i),…,π(K)}}.\ell_i^*:=\max_{\ell}\{\ell\colon\ell\in\{\pi(i),\ldots,\pi(K)\}\}.

Simplified exponents-region conjecture. The region E∗(R1,…,RK,ϵ1,…,ϵK)\mathcal{E}^*(R_1,\ldots,R_K,\epsilon_1,\ldots,\epsilon_K) is the set of all exponent tuples (θ1,…,θK)(\theta_1,\ldots,\theta_K) for which there exist nonnegative rates Ri,ℓR_{i,\ell} satisfying

θk≤min⁡i∈{1,…,π(k)}[∑ℓ=1kηℓ(Ri,ℓ)],k∈{1,…,K},\theta_k\leq\min_{i\in\{1,\ldots,\pi(k)\}}\left[\sum_{\ell=1}^{k}\eta_\ell(R_{i,\ell})\right],\qquad k\in\{1,\ldots,K\},

and

Rℓ≥∑i∈{1,…,K}:ℓi∗≥ℓ(ϵπ(i−1)−ϵπ(i))Ri,ℓ,ℓ∈{1,…,K}.R_\ell\geq\sum_{\substack{i\in\{1,\ldots,K\}:\\ \ell_i^*\geq\ell}}\left(\epsilon_{\pi(i-1)}-\epsilon_{\pi(i)}\right)R_{i,\ell},\qquad \ell\in\{1,\ldots,K\}.

The formula is obtained by retaining only the rate-sharing components indexed by the ordered decision centers; it is proved in the paper for K=2K=2 and K=3K=3, whereas its validity for arbitrary K≥2K\geq2 is conjectured.

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.