ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let aˉ:=1−a\bar{a}:=1-a and a∗b:=a(1−b)+(1−a)ba*b:=a(1-b)+(1-a)b. Let α,ϵ,f,g∈(0,1)\alpha,\epsilon,f,g\in(0,1), and define ss by

s:=(ϵˉ−ϵ)[log⁡(α∗ϵ)−log⁡(αˉ∗ϵ)]log⁡α−log⁡αˉ.s:=\frac{(\bar{\epsilon}-\epsilon)[\log(\alpha*\epsilon)-\log(\bar{\alpha}*\epsilon)]}{\log\alpha-\log\bar{\alpha}}.

Define cc by the two equivalent expressions in the source, with the quantities kk, f∗gˉ∗ϵf*\bar{g}*\epsilon, hh, PXYP_{XY}, and (X^,Y^)(\hat X,\hat Y) interpreted as in the paper; when α=12\alpha=\frac12, these expressions are defined by continuity.

Supporting-functional inequality.

sH(X^,Y^)−I(X^;Y^)≤s[h(ϵ)+h(α)]−[h(α∗ϵ)−h(ϵ)]+c(f−12)(g−12)f∗gˉ∗ϵ.sH(\hat X,\hat Y)-I(\hat X;\hat Y)\le s[h(\epsilon)+h(\alpha)]-[h(\alpha*\epsilon)-h(\epsilon)]+\frac{c(f-\frac12)(g-\frac12)}{f*\bar g*\epsilon}.

Equality holds at the four points specified in the source.

This is presented as a conjectured inequality that would imply the one-way tradeoff conjecture for binary symmetric sources. Because the supplied context does not define all notation needed to state it self-containedly, its precise formulation should be checked against the paper.

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.