The supporting-functional inequality for the binary symmetric key-interaction conjecture

From papers

Let aˉ:=1a\bar{a}:=1-a and ab:=a(1b)+(1a)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:=(ϵˉϵ)[log(αϵ)log(αˉϵ)]logαlogαˉ.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, fgˉϵf*\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(αϵ)h(ϵ)]+c(f12)(g12)fgˉϵ.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.

Progress summary

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Sources & referencesView supporting material

Primary source

Jingbo Liu, Paul Cuff and Sergio Verdú, “Secret Key Generation with Limited Interaction”, arXiv:1601.00899 (2017).

Solutions 0

No solutions have been posted yet.