Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

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

A basic question really. Consider the infinite sequence of random variable $\left(X_t\right)_{t \in \mathbb{N}}$ where $X_t:\Omega \mapsto \mathbb{R},\, \forall t$ and the spaces are $\left(\Omega, \mathscr{F}, \mathbb{P}\right)$ and $\left(\mathbb{R}, \mathscr{B}, \mathbb{P}_X\right)$. The random sequence induces the product topology $\left(\mathbb{R}^{\infty}, \mathscr{B}^{\infty}\right)$ generated by the finite dimensional cylinders. Also, we have one-one and onto transformation $\mathsf{T}:\Omega \mapsto \Omega$.

I want to know what the appropriate definition of a measure-preserving transformation is. Is it

$\begin{equation} \mathbb{P}\left[\omega \in \Omega\vert \left(X_t(\omega)\right)_{t \in \mathbb{N}} \in E\right] = \mathbb{P}\left[\omega \in \Omega\vert \left(X_t(\mathsf{T}\omega)\right)_{t \in \mathbb{N}} \in E\right],\; \forall E\in \mathscr{B}^{\infty} \end{equation}$

or is it

$\begin{equation} \mathbb{P}\left[\omega \in \Omega\vert \left(X_t(\omega)\right)_{t \in \mathbb{N}} \in E\right] = \mathbb{P}\left[\mathsf{T}\omega \in \Omega\vert \left(X_t(\mathsf{T}\omega)\right)_{t \in \mathbb{N}} \in E\right],\; \forall E\in \mathscr{B}^{\infty} \end{equation}$

I am failing to make the connection here.

share|cite|improve this question
up vote 0 down vote accepted

The second formula is a tautology since $\{\omega\in\Omega\mid Y(\omega)\in E\}=\{T\omega\in\Omega\mid Y(T\omega)\in E\}$ for every $Y$ defined on $\Omega$ and every onto $T:\Omega\to\Omega$. To wit, the second set can be rewritten as $$ \{\alpha\in\Omega\mid\exists\omega\in\Omega,\,\alpha=T\omega,\, Y(\alpha)\in E\}. $$ If the condition that there exists $\omega\in\Omega$ such that $\alpha=T\omega$ is fulfilled for every $\alpha\in\Omega$, to cancel it does not change the definition.

The first definition (the correct one) says that the distributions of $Y$ and $Y\circ T$ coincide, where $T:(\Omega,\mathscr F)\to(\Omega,\mathscr F)$ should be measurable and where in your case it happens that $Y=(X_t)_t$ hence $Y$ is a random variable $Y:(\Omega,\mathscr F,\mathbb P)\to(\mathbb R^{\infty},\mathscr B^{\infty})$, but in fact, $Y$ could be any random variable $Y:(\Omega,\mathscr F,\mathbb P)\to(E,\mathscr E)$ for any measurable space $(E,\mathscr E)$.

This is also equivalent to the fact that the distribution of $Y$ is the same with respect to $\mathbb P$ and $\mathbb Q$, where $\mathbb Q$ is the probability measure defined on $\mathscr F$ by $\mathbb Q(A)=\mathbb P(T^{-1}(A))$ (a definition which makes sense as soon as $T:(\Omega,\mathscr F)\to(\Omega,\mathscr F)$ is measurable).

share|cite|improve this answer

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.