The BEC bound for sandwiched Rényi conditional entropy

Let X1,X2X_1,X_2 be binary random variables with quantum side information B1,B2B_1,B_2, and let τ\tau denote the product construction in which the output is X1+X2X_1+X_2. Write K~α(XiBi)\widetilde K_\alpha^\downarrow(X_i|B_i) for the corresponding sandwiched Rényi conditional quantity, and let δαH\delta_\alpha^H denote the parameter appearing in the binary erasure-channel bound. The BEC bound. For α(0,2][3,)\alpha\in(0,2]\cup[3,\infty),

H~α(X1+X2B1B2)τ11αlog((δαHK~α(X1B1))(δαHK~α(X2B2))1δαH)+δαH.\widetilde H_\alpha^\downarrow(X_1+X_2|B_1B_2)_\tau\leq\frac{1}{1-\alpha}\log\left(\frac{(\delta_\alpha^H-\widetilde K_\alpha^\downarrow(X_1|B_1))(\delta_\alpha^H-\widetilde K_\alpha^\downarrow(X_2|B_2))}{1-\delta_\alpha^H}\right)+\delta_\alpha^H.

For α[2,3]\alpha\in[2,3], the same statement holds with the inequality direction reversed; for α{2,3}\alpha\in\{2,3\}, equality holds. The conjecture is intended as the Rényi-information analogue of the binary erasure-channel bound, but the supplied text does not provide a resolution for general α\alpha.

Sources & referencesView supporting material

Primary source

Christoph Hirche, Xinyue Guan and Marco Tomamichel, “Chain Rules for Renyi Information Combining”, arXiv:2305.02589 (2023).

Progress summary

Never refreshed

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

Solutions 0

No solutions have been posted yet.