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

Why is a random walk on $\mathbb{R}^d$ (see below) a time-homogeneous Markov process? Specifically, why does it satisfy requirement #2 of definition 17.3 that the map $\left(x,B\right)\mapsto\mathrm{P}_x\left[X\in B\right]$ be a stochastic kernel?

Relevant Definitions

Let $Y_1,Y_2,\dots$ be i.i.d.$\mathbb{R}^d$-valued random variables and let $$S_n^x=x+\sum_{i=1}^n Y_i\space\space \mathrm{for}\space x\in\mathbb{R}^d\space\mathrm{and}\space n\in\mathbb{N}_0$$

Define probability measures $\mathrm{P}_x$ on $\left(\left(\mathbb{R}^d\right)^{\mathbb{N}_0},\left(\mathcal{B}\left(\mathbb{R}^d\right)^{\otimes\mathbb{N}_0}\right)\right)$ by $\mathrm{P}_x=\mathrm{P}\circ\left(S^x\right)^{-1}$. Then the canonical process $X_n:\left(\mathbb{R}^d\right)^{\mathbb{N}_0}\rightarrow\mathbb{R}^d$ is a Markov chain with distributions $\left(\mathrm{P}_x\right)_{x\in\mathbb{R}^d}$. The process $X$ is called a random walk on $\mathbb{R}^d$ with initial value $x$. [Klenke, Example 17.5, p. 347]

share|cite|improve this question
up vote 2 down vote accepted

You have to show that, for each $A$ fixed in the countable product $\sigma$-algebra, the map $x \to \mathbb{P}_x(A) = \mathbb{P}(S^x \in A)$ is measurable.

Since $S^x = S^0 + \underline{x}$, where $\underline{x} = (x, x, ..., x, x, ...)$, we have $\mathbb{P}_x(A) = \mathbb{P}_0(A - \underline{x})$. Setting $A_x = A - \underline{x}$, we have $\mathbb{P}_x(A) = \mathbb{P}_0(A_x) = \mathbb{E}_{\mathbb{P}_0}(\chi_{A_x})$.

Fact : If $(X, \mathcal{A})$ is a measurable space, $(\Omega,\mathcal{T}, \mathbb{P})$ is a probability space, and $f : X \times \Omega \to [0,+\infty]$ is a measurable map, then the map $x \to \mathbb{E}_{\mathbb{P}}(f(x, .))$ is measurable.

(if you want to prove it, it's an application of a Dynkin-class argument : prove it first for characteristic functions of measurable rectangles, and then extend the result in the usual way)

So, we are reduced to prove that the map $(x, \omega) \to \chi_{A_x}(\omega)$ is measurable. But this function is just the characteristic function of the set of all $(x, \omega)$ such that $\underline{x} + \omega \in A$, and since the map $(x, \omega) \to \underline{x} + \omega$ is measurable (exercise !), we are done.

share|cite|improve this answer
Thanks, Ahriman! That's a terrific answer. I've got two suggestions, though. 1. It would be better to show that $\left(x,\omega\mapsto\underline{x}+\omega\right)$ before using the expression $\mathbb{P}_0\left(A-\underline{x}\right)$ in paragraph #2 rather than in the end, since this expression makes sense only if $A-\underline{x}$ is measurable. – Evan Aad Aug 20 '12 at 13:07
2. Dynkin's class argument can be bypassed by invoking Fubini's theorem. See, for instance, part (a) of Ash & Doléans-Dade's statement of Fubini's Theorem (Theorem 2.6.4, p. 105) – Evan Aad Aug 20 '12 at 13:08
Yes, you're right, but to prove this part of Fubini's theorem, you use exactly the same kind of Dynkin-class argument. – Ahriman Aug 20 '12 at 13:19

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.