1: \begin{definition}

2: i) A $[0,\infty]$-valued stochastic process $A=(A_t)_{t\geq 0}$

3: on $(\Omega',\mathcal{G})$

4: is called a \emph{positive continuous additive functional (PCAF)} of $X$

5: (in the strict sense), if $A_t$ is $\mathcal{G}_t$-measurable for every

6: $t\geq 0$ and if there exists a set $\Lambda \in \mathcal{G}$,

7: called a \emph{defining set} for $A$, such that

8: \begin{enumerate}

9: \item [a)] for all $x\in E$, $P_x[\Lambda]=1$,

10: \item [b)] for all $t\geq 0$, $\theta_t(\Lambda)\subset \Lambda$,

11: \item [c)] for all $\om \in \Lambda$,

12: $[0,\infty)\ni t\mapsto A_t(\om)$ is a $[0,\infty)$-valued continuous function

13: with $A_0(\om)=0$ and

14: \begin{align}

15:  A_{t+s}(\om)=A_t(\om)+ A_s \circ \theta_t (\om), \qquad \forall s,t\geq 0.

16: \end{align}

17: \end{enumerate}

18: 

19: 

20: ii) Two such functionals $A^1$ and $A^2$ are called \emph{equivalent}

21: if $P_x[A^1_t=A^2_t]=1$ for all $t>0$, $x\in E$, or equivalently,

22: there exists a defining set $\Lambda \in \mathcal{G}_\infty$ for

23: both $A^1$ and $A^2$ such that $A^1_t(\om)=A^2_t(\om)$ for all $t\geq 0$,

24: $\om\in \Lambda$.

25: 

26: iii) For any such $A$, a Borel measure $\mu_A$ on $E$ satisfying 

27: \begin{align} \label{eq:rev_corr_gen}

28:  \int_{E} f(y) \, \mu_A(dy)

29: 	=\lim_{t\downarrow 0} \frac 1 t \int_{E} E_x \Bigl[ \int_0^t f(B_s) \, dA_s \Bigr] \, m(dx)

30: \end{align}

31: for any non-negative Borel function $f:E\to[0,\infty]$ is called the

32: \emph{Revuz measure} of $A$, which exists uniquely by general theory

33: (see e.g.\ \cite[Theorem~A.3.5]{CF12}). 

34: %Equivalent PCAFs give rise to the same Revuz measure.

35: \end{definition}

36: