Let triple $(\Omega, \Sigma, P)$ be a probability space. A (temporal) stochastic process $X(t, \omega): (0,\infty]\times\Omega\rightarrow R$ can be viewed as:

  1. for each $t$, $X(t, \cdot)$ is a random variable
  2. for each fixed $\omega\in\Omega$, $X(\cdot, \omega)$ is a sample path (measurable function).

(1) is easy to understand. I'm confused with (2). What does it mean by fixing $\omega$?

For example, let's say ($\mathbb{R}$, $\mathcal{B}(\mathbb{R})$, $P$), by fixing $\omega=a\in\mathbb{R}$ with a number, then the sample path is

$$ X(1, \omega=a), X(2, \omega=a), X(3, \omega=a)\,\cdots, X(t, \omega=a) $$

Is this correct by choosing an $\omega\in\Omega$?

Could someone give an explicit example of $(\Omega, \Sigma, P)$, and one fixed $\omega\in\Omega$ of Brownian motion? What are the $\Omega, \Sigma, \omega$ here

Similar question:

  1. Stochastic Processes and Trajectories He says $\omega$ should be a sequence. But doesn't this violate $\omega\in\Omega$? Does it mean for every time step $t$, we have to "enlarge" the sample space?

  2. An example of stochastic process

  • $\begingroup$ For $\omega\in\Omega$, you can simply define a mapping $X':(0,\infty]\rightarrow R$ by $X'(t):=X(t,\omega)$. $\endgroup$
    – peer
    Jan 10 '20 at 15:40

Let us start with a simple example. Assume that $( \Omega, \mathcal{F}, P)$ is some probability space and let $X : \Omega \rightarrow [0, 10]$ be a random variable which is uniformly distributed on $[0,10]$.

Let us define a stochastic process $X( t, \omega)$ in the following way: $$ X ( t, \omega ) := t X(\omega), \quad ( t, \omega) \in [0, \infty) \times \Omega. $$ Now let us first fix $t=100$. Then $X ( 100 , \omega) = 100X(\omega)$. In other words, we get $100X$, which is again a random variable, since a random variable multiplied by a constant is again a random variable. This argument can be used to conclude that $X(t, \omega) = t X(\omega)$ is a random variable for every fixed $t \geq 0$.

Let us now fix some $\omega_0 \in \Omega$ and assume that $X(\omega_0) = 3$. This then means that for all $t \geq 0$ $$ X( t, \omega_0 ) =tX(\omega_0) = 3t. $$ And if we let $t$ vary over $[0, \infty)$ we will simply get a linear function of the form $$ y(t)=t X( \omega_0) = 3t, \quad t \in [ 0, \infty) $$ which will be the sample path of the stochastic process $X(t, \omega)$ for the fixed $\omega_0 \in \Omega$. Extending this argument, we see that all the sample paths of the stochastic process $X(t,\omega) = t X( \omega )$ are from the following class of linear functions: $$ \{ y : [0, \infty ) \rightarrow \mathbb{R} : y(t) = bt, \ b \in [ 0, 10 ]\}. $$ Let us now turn to a Brownian $B(t, \omega)$ defined on (some other) probability space $( \Omega, \mathcal{F}, P )$. The Brownian motions starts almost surely at $0$, which means that $B(0, \omega)$, which is just a random variable for the fixed $t = 0$, is equal to $0$ with a probability of $1$, i.e. $P( B(0, \omega ) = 0 ) = 1$. Moreover, we know that a Brownian motion has continuous sample paths, which means that for every fixed $\omega_0 \in \Omega$, $y(t) = B(t, \omega_0)$ is a continuous function of $t$. Therefore, with probability $1$, the sample paths of the Brownian motion will be from the following class of functions $$ \{ y : [0, \infty ) \rightarrow \mathbb{R} : y(t) \ \text{is continuous}, \ y(0) = 0 \}. $$


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.