HDS

Exercise 15.4: More Properties of Shannon Entropy

chapter 15

Recall the definition \begin{equation} H(X \cond Y) = \E \log \frac{\sd \P_{X \cond Y}}{\sd \mu}. \end{equation}

(a)

Use non-negativity of the mutual information: \begin{equation} H(X \cond Y) - H(X) = \E \log \frac{\sd \P_{X \cond Y} / \sd \mu}{\sd \P_{X} / \sd \mu} = \E_Y \KL(\P_{X \cond Y}, \P_{X}) \ge 0. \end{equation}

(b)

Follows from a direct computation: \begin{equation} H(X) + H(Y \cond X) = \E \log \frac{\sd \P_{X}}{\sd \mu}\frac{\sd \P_{Y \cond X}}{\sd \mu} = \E \log \frac{\sd \P_{X, Y}}{\sd \mu} = H(X, Y). \end{equation}

(c)

Follows from (a) applied to the r.h.s. of (b).

Published on 29 January 2022.