0

Im having difficulty regarding the properties of standard Brownian motion. Can someone briefly describe the 4 properties?

And also, it say $\Bbb E[Z(2)\,Z(4)]$ :-

  1. $\Bbb E[Z(2)\,Z(4)\mid Z(2)]$
  2. $Z(2)\,\Bbb E[Z(4)\mid Z(2)]$
  3. $Z(2) \times Z(2)$
  4. $Z^2(2)$

And then he goes calculating the value by double expectation. Can someone please explain the steps 1 and 2?

Thank you!

Graham Kemp
  • 133,231
Nehal
  • 155

1 Answers1

2

The properties that model a Wiener process are:

  • $Z(0)=0$
  • $Z(t)$ is almost everywhere contrinuous.
  • $Z(t)$ occurs in independent increments.
  • $Z(t)-Z(s)\sim \mathcal N(0, t-s)$ for all $0\leqslant s < t)$

So you are looking at:

$\begin{align}\Bbb E[Z(2)\,Z(4)] \tag 1 & = \Bbb E[~\Bbb E[Z(2)\,Z(4)\mid Z(2)]~] \\[1ex] \tag 2 & =\Bbb E[~ Z(2)\,\Bbb E[Z(4)\mid Z(2)]~] \\[1ex] \tag 3 & = \Bbb E[~ Z(2) \times Z(2)~] \\[1ex] \tag 4 & = \Bbb E[~Z^2(2)~]\end{align}$

Step $1$ is the Law of Total Epectation. What you have listed is the interior part of the nested expectation: $$\Bbb E[Z(2)\,Z(4)\mid Z(2)]$$

Step $2$ is that $\mathsf E(XY\mid X)=X~\mathsf E(Y\mid X)$, since when measured against $X$, a factor of $X$ itself can be distributed out of the operator.   So: $$Z(2)\,\Bbb E[Z(4)\mid Z(2)]$$

Step $3$ is just evaluating the conditional expectation; $\Bbb E[Z(4)\mid Z(2)]= Z(2)$.

$$\begin{align} &\qquad \Bbb E[Z(4)\mid Z(2)] \\[1ex] &=~ \Bbb E[Z(2)\mid Z(2)]+\Bbb E[Z(4){-}Z(2)\mid Z(2)] &&\text{Linearity of Expectation} \\[1ex] &=~ Z(2)+\Bbb E[Z(4){-}Z(2)\mid Z(2)] && \text{Since }\mathsf E(X\mid X)=X \\[1ex] &=~ Z(2)+ \Bbb E[Z(4){-}Z(2)] && \text{Wiener process occurs in independent increments}\\[1ex] &= ~ Z(2)+0 &&\text{Because } Z(4){-}Z(2)\sim \mathcal N(0, 2) \end{align}$$

Step $4$ is just multiplying the factors.$$Z^2(2)$$

Then take the expectation of that to conclude that $$\Bbb E[Z(2)\;Z(4)] = \mathbb E[Z^2(2)]$$

From here recall again that $Z(2)\sim\mathcal N(0;2)$ so therefore: $\mathbb E[Z^2(2)]= 2$

Graham Kemp
  • 133,231
  • Thank you. I can get some idea from it now. Just to be sure, in step 3: E[Z(4)|Z(2)] = E[Z(t+s)|{Z(u): 0<u<t}] ~ N(Z(t),s) = E[Z(2+2)|{Z(u):0<u<2}] ~N(Z(2),2) Is this correct? – Nehal May 16 '17 at 19:57
  • I've expanded step 3, and included the rules for a Wiender Process to clarrify. – Graham Kemp May 17 '17 at 00:08
  • Can you please have a look at the question

    https://math.stackexchange.com/q/2288421/446754

    Thanks a lot!

    – Nehal May 20 '17 at 00:25