Bound Intrinsic Information. Examples 2 and 3 suggest that, in analogy to bound entanglement of a quantum state, bound classical information exists, i.e., conditional intrinsic information which cannot be used to generate a secret key in the classical scenario. We give a formal definition of bound intrinsic information. || ↓ || Definition 2 Let PXY Z be a distribution with I(X; Y Z) > 0. If S(X; Y Z) > 0 holds for this distribution, the intrinsic information between X and Y , given Z, is called free. Otherwise, if S(X; Y Z) = 0, the intrinsic information is called bound. ❢ Note that the existence of bound intrinsic information could not be proven so far. However, all known examples of bound entanglement, combined with all known advantage-distillation protocols, do not lead to a contradiction to Con- jecture 1! Clearly, it would be very interesting to rigorously prove this conjecture because then, all pessimistic results known for the quantum scenario would im- mediately carry over to the classical setting (where such results appear to be much harder to prove).
Bound Intrinsic Information. Examples 2 and 3 suggest that, in analogy to bound entanglement of a quantum state, bound classical information exists, i.e., conditional intrinsic information which cannot be used to generate a secret key in the classical scenario. We give a formal definition of bound intrinsic information. ↓ Definition 2 Let PXY Z be a distribution with I(X; Y Z) > 0. Then if ||
