Paper 1, Section I, 4I4 \mathrm{I}

Coding and Cryptography
Part II, 2014

State and prove Gibbs' inequality.

Show that, for a pair of discrete random variables XX and YY, each taking finitely many values, the joint entropy H(X,Y)H(X, Y) satisfies

H(X,Y)H(X)+H(Y)H(X, Y) \leqslant H(X)+H(Y)

with equality precisely when XX and YY are independent.