I am currently reading Strook's $\textit{Probability Theory: An Analytic View}$, and I am confused by the following statement on page 156:

"I take for $D(\mathbb{R}^N)$ the measurable structure given by the $\sigma$-algebra $\mathcal{F}:=\sigma(\psi(t): t\in[0,\infty))$ generated by the maps $\psi\mapsto\psi(t)$ as $t$ runs over $[0,\infty)$. The reason for insisting on this choice is that I want two $D(\mathbb{R}^N)$-valued stochastic processes $X_t$ and $Y_t$ to induce the same measure on $D(\mathbb{R}^N)$ if they same distribution. Seeing as $\mathcal{F}\subset\mathcal{B}(D(\mathbb{R}^N)$, this would not be true if I were to choose the Borel structure."

Any help clarifying this is greatly appreciated.


Suppose that $X$ and $Y$ are equal in distribution in some space $(\Omega,\mathcal{F},\mathbb{P}$). Stroock would like to choose $\mathcal{F}$ such that $\mathbb{P} (X \in A) = \mathbb{P} (Y \in A)$ for any $A \in \mathcal{F}$. This does not hold when you equip the set $\Omega$ with its natural Borel structure $\mathcal{F} := \sigma$ say. Instead, he defines $\mathcal{F} := \mathfrak{B}(D(\mathbf{R}^N))$ to allow this.

There are many problems with this choice of $\mathcal{F}$. In the same page he discusses how $\mathcal{F}$ measurability no longer follows from continuity. Determining which (infinite dimensional) stochastic processes are now measurable is.. considerably more difficult.

On page 160 he asks us to show that $\mathfrak{B}(D(\mathbf{R}^N))$ is not $(\mathbb{R}^N)^{[0,\infty[}$ measurable nor is a special subset of it (he calls it $C$). The set $\mathfrak{B}(D(\mathbf{R}^N))$ is rather bizarre.

  • $\begingroup$ Thanks for your answer Arbias. But why specifically does there exist an $A\in\mathcal{F}$ such that $\mathbb{P}(X\in A)\neq\mathbb{P}(Y\in A)$ even when $X\overset{\mathcal{D}}{=}Y$ when $\mathcal{F}$ is the Borel $\sigma$-algebra? $\endgroup$ – Mike Oct 6 '14 at 13:02
  • $\begingroup$ Apologies for the long wait - and you may no longer be interested in this question but I have looked at Stroock's book again and exercises 4.1.9, 4.1.10 and 4.1.11 explain why and give such an example. I am happy (although it will be torturous) to construct such an example and to explain Stroock's choice. $\endgroup$ – AXH Feb 8 '15 at 23:15

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.