Simplifying information-theoretic arguments by post-selection

Abstract

Devices and protocols for information processing are often required to work for arbitrary inputs. For example, in channel coding theory, one demands that a coding scheme transmits any possible input state reliably over a given noisy channel. Similarly, in quantum cryptography, security of a protocol should hold independently of the inputs, even if they are chosen maliciously. In this short paper, we review the Post-Selection Technique introduced in (Phys. Rev. Lett. 102:020504, 2009). Its main purpose is to simplify the analysis of information processing schemes so that only one single input needs to be considered. If a scheme satisfies a desired criterion when acting on this particular input, then—under certain symmetry conditions—the same criterion is automatically met for arbitrary inputs. We illustrate the Post-Selection Technique at the example of quantum cryptography. Here, it can be used to show that security of a Quantum Key Distribution scheme against general attacks—somewhat surprisingly—follows from its security against one specific attack. This not only simplifies security proofs, but also has other remarkable implications, e.g., that no randomness is needed for privacy amplification.