1: \begin{definition} If $\sigma\in\M(\X,\T,\basemeasure)$, the \emph{relative entropy} of $\T$ over $\theta$ with respect to $\sigma$ is defined by the equations

2: \begin{align*}

3: h_\sigma(\T\on\theta) &:= \sup_\Par h_\sigma(\T\on\theta;\Par)\\

4: h_\sigma(\T\on\theta;\Par) &:= \lim_{n\rightarrow\infty}\frac{1}{n}H_\sigma\left(\bigvee_{j = 0}^{n - 1}\T^{-j}\Par\on\pi^{-1}\epsilon_\Omega\right)

5: \end{align*}

6: (The supremum is taken over all partitions $\Par$ of $\X$ such that $H_\sigma(\Par\on\pi^{-1}\epsilon_\Omega) < \infty$. $\epsilon_\Omega$ and $\epsilon_\X$ are the partitions into points of $\Omega$ and $\X$, respectively.) For proof of the existence of the limit see [\cite{Bo} Theorem 2.2, p.102].

7: \end{definition}

8: