A Holevo quantity inequality

For some reason, I wanted to know the following fact at some point.

Let (p_i,\rho_i) be an ensemble of states of a bipartite system AB. For \chi the Holevo information, we have \chi_{AB}\leq\chi_A+\chi_B+\bar{I} where \bar{I}=\sum_ip_iI_i is the expected value of the quantum mutual information I_i=S(\rho_i^A)+S(\rho_i^B)-S(\rho_i^{AB}).

\begin{aligned} \chi_{AB}&=S\left(\sum_{i}p_i\rho_i\right)-\sum_{i}p_iS(\rho_i)\\ &\leq S\left(\sum_{i}p_i\rho^A_i\right)+S\left(\sum_{i}p_i\rho^B_i\right)-\sum_{i}p_iS(\rho_i)\\ &= S\left(\sum_{i}p_i\rho^A_i\right)+\sum_{i}p_iS(\rho^A_i)-\sum_{i}p_iS(\rho^A_i)+S\left(\sum_{i}p_i\rho^B_i\right)-\sum_{i}p_iS(\rho_i)\\ &= \chi_A+\chi_B+\sum_{i}p_iS(\rho^A_i)+\sum_{i}p_iS(\rho^B_i)-\sum_{i}p_iS(\rho_i)\\ &= \chi_A+\chi_B+\sum_{i}p_i\left(S(\rho^A_i)+S(\rho^B_i)-S(\rho_i)\right)\\ &= \chi_A+\chi_B+\sum_{i}p_iI_i  \end{aligned}

Since the Holevo information gives an upper bound for the mutual information between the random variable X\sim (p_i) and the outcome of any measurement that can be made on the received state, setting \chi_A=\chi_B=0 we see that \bar{I} may be meaningfully taken as an upper bound for the amount of hidden information.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s