OK, this is kind of re-posting, but I think I can clarify the question more, so it's worth a shot.

Consider a real valued process $(X_t)_{t \leq T}$, cadlag on a probability space $(\Omega, (\mathcal{F}^\circ_t)_{t \leq T}, \mathbb{P}). \mathcal{F}^\circ_t=\sigma(X_s;s\leq t)$ is the uncompleted, natural filtration generated by $X_t$. Unfortunately $X_t$ neither has independent increments, nor is it markov. Since $\Omega$ is a Polish space, $\mathcal{F}^\circ_T$ and also $\mathcal{F}^\circ_t$ are countably generated, so we know, there exists a regular version of the conditional probability of $\mathbb{P}$ for any fixed $t$ for $\mathbb{P}$-a.a. $\omega$, i.e. for fixed $t$, $\mathbb{P}(\cdot|\mathcal{F}_t)(\omega)$ is a prob. measure f.a.a. $\omega$.

Hence we know, that for all $t\in [0,T]\cup \mathbb{Q}$, we find a regular conditional probability f.a.a. $\omega$, depending on $t$. In words: Given almost any path of the process up to time $t$, we can deduce the probablity of events, taking that information into account.

On the remaining $\omega$'s, define some meaningless measure, so we have a measure $\forall \omega$. How can I extend this to all $t$ in a reasonable way? Reasonable means: There is one Null set $N$, so that $\forall t$ $\mathbb{P}(\cdot|\mathcal{F}_t)(\omega)$, $\omega\in N^c$, is a measure Anybody seen anything like this?

I read something like this only for Markov and Feller processes using infinitesimal generators, but this cannot be carried over one to one, because we do not have a transition semigroup.

Maybe I have a deep misunderstanding here. Grateful for any objections, hints and comments.


1 Answer 1


Let's assume that we are working with the canonical probability space $\Omega = D(\mathbb R)$ of càdlàg functions, and $\mathbb P$ is the law of the process. I would doubt that there is a satisfactory answer at the level of maximal generality you've stated. At the very least, the measure $\mathbb P$ should be Radon. There are extremely general results on the existences of RCPs for Radon measures (cf. Leão, Fragoso and Ruffino, Regular conditional probability, disintegration of probability and Radon spaces).

The RCP is a measure-valued function $P : [0,T] \times \Omega \to \mathcal M(\Omega)$ such that for $\mathbb P$-almost every $\omega$, the measure $P(t,\omega, \cdot)$ is a version of $\mathbb P(\cdot|\mathcal F_t)$. Do you want the function $(t, \omega) \mapsto P(t,\omega,\cdot)$ to simply exist and be measurable? If so, this can be done in the wide generality stated above; see Leão et al.

Recently, I have needed more regularity properties for RCPs, namely, continuity. Consider the space $\mathcal M(\Omega)$ of Radon measures on $\Omega$ equipped with the topology of weak convergence of measures. We say that the RCP is a continuous disintegration (or continuous RCP) when it satisfies the following property: $$\mbox{if $\omega_n \to \omega$, then the measures $P(t,\omega_n,\cdot)$ converge weakly to $P(t,\omega,\cdot)$.}$$

If the law is Gaussian, then my preprint Continuous Disintegrations of Gaussian Processes gives a necessary and sufficient condition for the law $\mathbb P$ to have a continuous disintegration. I haven't thought about this in the case of càdlàg functions, but I'm pretty sure that this will extend easily. Note that this is just for fixed $t$.

To show that the map $(t, \omega) \mapsto P(t,\omega,\cdot)$ jointly continuous, a little more work is needed. As part of a larger project, Janek Wehr and I have general results in this direction for stationary, Gaussian processes. If this is what you need, I'm happy to discuss this with you further.

Open Question: If the law $\mathbb P$ is not Gaussian but at least is log-Sobolev, then all the same results should hold. This is because log-Sobolev measures satisfy very strong concentration-of-measure properties. I have some ideas how to do this, but I haven't worked out the details because I've been busy with other projects. If anybody is interested in collaborating on extending this work to the log-Sobolev case, please contact me.

  • 1
    $\begingroup$ I was not as precise, as I hoped I was: The situation is indeed much simpler. $\mathbb{P}$ is just the law of $X_T$. The question is, what can I say about the outcome of $X$ in T, given the path up to t. I want to compute $\mathbb{P}(\{f(X_T)\in \cdot \}|F_t)$ using $(\mathbb{P}|F_t)(\cdot,\omega)$ as a RCP. One may not need the path space $D$ here, though it might be the way to prove it. What I worry about is, that you always have to declare some nonsense measure on a Null set of $\Omega$ for each $t$ and you have no control which $\omega$ that concerns for some $t$. $\endgroup$
    – Pierre
    Jul 24, 2011 at 21:20
  • 1
    $\begingroup$ You have control one a dense subset of [0,T], because then it is only a Nullset of $\Omega$, where you have to declare the nonsense measure. So I thought some kind of continuity (from the right?) might help, because we have cadlag paths. $\endgroup$
    – Pierre
    Jul 24, 2011 at 21:30
  • 1
    $\begingroup$ @TomLaGatta Have you managed to extend this to the log-Sobolev case? (Since I see this post is a bit old :) ) $\endgroup$ Mar 19, 2021 at 8:50
  • $\begingroup$ Hi @Bernard_Karkanidis, alas I lost this thread many years ago and never was able to move it along. Janek & I did publish our Geodesics of Random Riemannian Metrics which sufficed for stationary, Gaussian-distributed random tensors, which we then transformed pointwise into a space of non-negative metric tensors. So the resulting space still satisfied nice conditional properties, because conditioning could directly pull back into the Gaussian situation. Hope this helps! I'd certainly be interested to learn about other cases as other working mathematicians develop those solutions. $\endgroup$ Jun 14 at 3:07

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.