Let $X$ be a stochastic process with a.s. continuous sample paths on $[0, 1]$ such that $\mathbb E [X_t]$ is finite for all $t \in [0, 1]$. Given any non null subset $Y$ of the probability space, define $\mathbb Q_Y$ to be the restricted probability measure $\mathbb Q_Y [E] = P(E \cap Y)/P(Y)$.

Does it follow there exists some non null $Y$ such that that the function $f: [0, 1] \to R$ defined $f(t)$ $=$ $\mathbb E_{Q_Y} [X_t]$ is continuous a.e.?


Assuming you don't mean continuous sample paths, the answer is no. For example, $X_t = $ the indicator of the rationals with probability 1, or on a space that is an atom of mass 1 if you prefer. Then there is only 1 set of positive probability and it doesn't work.

Since you do mean continuous sample path, the answer is 'yes', take $Y = \lbrace max |X_t| < A \rbrace $ which can be made to be of positive probability by the continuity of $X_t$. Then, by dominated convergence (convergence because of continuity, and dominated by the bound A) $f$ is continuous.

  • $\begingroup$ Sorry, I meant continuous sample paths. $\endgroup$ – James Baxter Aug 2 at 12:10
  • $\begingroup$ Can you elaborate on how the conclusion follows from dominated convergence? $\endgroup$ – James Baxter Aug 2 at 13:27

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.