Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have a question as follow:

"Let $X$ be a general Markov process, $M$ is a running maximum process of $X$ and $T$ be an exponential distribution, independent of $X$.

I learned that there is the following result:

Probability: $P_x(X_T\in dz \mid M_T=y)$ is independent of starting point $x$ of the process $X$. Where $y, z \in R$"

Is there anyone who knows some references which mentioned the result above? I heard that this result was found around the seventies but I haven't found any good reference yet.

Thanks a lot!

share|cite|improve this question
What do you call a running process of a Markov process? – Did Dec 27 '11 at 17:44
Hello Didier Piau, $M_t:= \sup \limits_{s \leq t}X_s $ is called a running process of $X$. – ltt Dec 27 '11 at 19:26
Clarify your central expression please. P(X) should be P(X = x) or P(X <=x) or something similar (the first term in parenthesis), depending on whether it's discrete or continuous. Next, this certainly doesn't hold in all generality. Imagine a constant process. Any such process is Markovian, and will have X and M constant as well, but at different values, so their probability measures will be very different. – gnometorule Dec 27 '11 at 19:58
@gnometorule: I have edited the post. I hope it is clearer. – ltt Dec 28 '11 at 0:06
up vote 3 down vote accepted

For real-valued diffusion processes, this is essentially a local form of David Williams' path decomposition, and can be deduced from Theorem A in a paper "On the joint distribution of the maximum and its location for a linear diffusion" by Csaki, Foldes and Salminen [Ann. Inst. H. Poincare Probab. Statist., vol. 23 (1987) pp. 179--194].

For more general Markov processes, you will need to look into the theory of "last-exit times". Although these are not stopping times, many Markov processes possess a sort of strong Markov property at such times. This theory can be applied to the last time before $T$ that the process is at level $y$. One place to start might be the paper of Meyer, Smythe and Walsh "Birth and death of Markov processes" in vol. III (pp. 295-305) of the Proceedings of the Sixth Berkeley Symposium on Mathematical Statistics and Probability (1972). See also the work of P.W. Millar from roughly the same time period.

share|cite|improve this answer
Thanks a lot @John Dawkins. I have also found the paper from Millar:… – ltt Dec 28 '11 at 10:41
+1. Nice bibliographic search. I wonder if, once the research papers you indicate were digested by the community, this result got explained in some textbooks as well... – Did Dec 28 '11 at 15:44

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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