I have an interesting idea for a proof approach that someone might find useful. Here it is.

Suppose we are given a quantum qubit channel $N$ (for example the amplitude damping channel) whose Holevo information we are trying to prove is additive for two uses, i.e, we are trying to show that $\chi_{N \otimes N} = 2 \chi_N$.

Given the channel $N$ and another channel $M$ that we know has strongly additive Holevo information, suppose we construct a channel $N^\prime$ that simulates the channel $N$ with probability $(\chi_{N \otimes N} - 2 \chi_N)$, and simulates the channel $M$ with probability $1 - (\chi_{N \otimes N} - 2 \chi_N)$, i.e,

$$N^\prime = (\chi_{N \otimes N} - 2 \chi_N) N + (1 - (\chi_{N \otimes N} - 2 \chi_N))M.$$

Note that for qubit channels $\chi_{N \otimes N} \leq 2$ and $\chi_N \leq 1$, so the assumption that $(\chi_{N \otimes N} - 2 \chi_N) \leq 1$ -- and so can be used as a probability -- is reasonable. From the construction of the channel, we can see that if $N$ is additive, then $(\chi_{N \otimes N} - 2 \chi_N) = 0$, and $N^\prime$ is strongly additive.

That was the idea; to use $(\chi_{N \otimes N} - 2 \chi_N)$ in the construction of a new channel. May be this idea, or a variant of it, can be used to prove something.

I would be interested to know about similar ideas that have been used in proofs before.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.