1: \begin{definition}[Chernoff information]

2: The Chernoff information between probability measures $P_1$ and $P_2$ is expressed as follows in

3: terms of the R\'{e}nyi divergence:

4: \begin{align}

5: C(P_1, P_2) = \max_{\alpha \in [0,1]} \Bigl\{(1-\alpha) D_{\alpha}(P_1\|P_2)\Bigr\}

6: \end{align}

7: and it is the best achievable exponent in the Bayesian probability of error for binary hypothesis testing

8: (see, e.g., \cite[Theorem~11.9.1]{Cover_Thomas}).

9: \end{definition}