Let $X$ a stochastic process and $\mathcal F_s=\sigma (X_u\mid u\leq t)$. In the book "Continuous Martingale and Brownian motion" (third edition) of Yor and Revuz, page 80 : Let $$P_{s,t}(X_s,A)=\mathbb P\{X_t\in A\mid \mathcal F_s\}.$$

Q1) They say that $\mathbb E[f(X_t)\mid \mathcal F_s]=P_{s,t}f(X_s)$ a.s. where $f$ is a bounded and continuous function. Is it a definition or a proposition ? Because, they say that it can be proved using classical argument on conditional expectation, but I don't see how I can prove it.

Q2) After, they say that for $s<t<v$, $$\mathbb P\{X_v\in A\mid \mathcal F_s\}=\mathbb P\{X_v\in A\mid \mathcal F_t\mid \mathcal F_s\}=\mathbb E[P_{t,v}(X_t,A)\mid \mathcal F_s]=\int P_{s,t}(X_s,dy)P_{t,v}(y,A).$$

1) I'm not sure what mean $\mathbb P\{X_v\in A\mid \mathcal F_t|\mathcal F_s\}$. Is it $$ \mathbb P \{\mathbb P\{X_v\in A\mid \mathcal F_t\}|\mathcal F_s\}= \mathbb E[\mathbb E[\boldsymbol 1_{X_v\in A}\mid \mathcal F_t]\mid \mathcal F_s] \ \ ?$$

2) How can I prove that $$\mathbb E[P_{t,v}(X_t,A)\mid \mathcal F_s]=\int P_{s,t}(X_s,dy)P_{t,v}(y,A),$$ I by the way, didn't even know that we have a formula for $\mathbb E[P_{t,v}(X_t,A)\mid \mathcal F_s]$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.