How would one find the expectation and thus the covariance of a process, say,

$$X(t)=\alpha +B(t)-tB(1)$$

is the process and $t\in[0,1]$ and $\alpha\in\mathbb{R}$?

When considering this process would you take a discrete time method at both end points or a continuous time over the entirety of $[0,1]$?

I have assumed that $(B(t))_{t\in [0,1]}$ is a Brownian motion in my mathematics.

Any help explaining the lines are arguments would be greatly appreciated thanks!


Basically, we calculate the expected value, the variance and the autocovariance of a process by calculating the expected value and the variance of the random variable $X(t)$ and the covariance between the variables $X(s)$ and $X(t)$.

Hence, the expectation of the process is given by $$ \operatorname EX(t)=\alpha+\operatorname EB(t)-t\operatorname EB(1)=\alpha $$ since $\operatorname EB(t)=0$ for each $t\in[0,1]$. The variance of the process is given by \begin{align*} \operatorname{Var}X(t) &=\operatorname{Var}[\alpha+B(t)-tB(1)]\\ &=\operatorname{Var}[B(t)-tB(1)]\\ &=\operatorname E[B(t)-tB(1)]^2\\ &=\operatorname E[B^2(t)-2tB(t)B(1)+t^2B^2(1)]\\ &=t(1-t) \end{align*} using the fact that $\operatorname EB^2(t)=t$ and $\operatorname E[B(s)B(t)]=\min\{s,t\}$. Finally, \begin{align*} \operatorname{Cov}[X(s),X(t)] &=\operatorname E[(B(s)-sB(1))(B(t)-tB(1))]\\ &=\operatorname E[B(s)B(t)]-t\operatorname E[B(s)B(1)]-s\operatorname E[B(1)B(t)]+st\operatorname EB^2(1)\\ &=\min\{s,t\}-st \end{align*} for $s,t\in[0,1]$.


If $\{X(t), t\geq 0\}$ be an independent increment process, then it has the following properties:

  1. $Cov(X(s),X(t)) = min\{s,t\}$.
  2. $Var(X(t)-X(s)) = Var(X(t))-Var(X(s))$.

By your supposition, $\{B(t), t\geq 0\}$ is a standard Brownian motion process, for $t \in [0,t]$ and a process is defined by $X(t)=\alpha + B(t) - tB(1)$. By the definition of standard Brownian motion process, we know that, it has stationary and independent increments and

$B(t)\sim N(0,t)$, meaning that, $E(B(t))=0$ and $Var(B(t))=t$.


\begin{equation} E(X(t))=E(\alpha + B(t) - tB(1)) = \alpha. \end{equation} since $E(B(t))=0$ and $E(tB(1))=tE(B(1))=0$.


\begin{eqnarray*} Var(X(t)) &=& Var(\alpha + B(t) - t\;B(1)) = Var( B(t) - t\;B(1))\\ &=& Var(B(t)) - t^2 Var(B(1)) = t - t^2 = t(1-t).\quad \mbox{by prop-2 above} \end{eqnarray*}

Covariance: \begin{eqnarray*} Cov\{X(s),X(t)\} &=& Cov\{B(s)-s\;B(1), B(t)-t\;B(1)\} \\ &=& Cov\{B(s),B(t)\} - Cov\{B(s),t B(1)\} - Cov\{sB(1),B(t)\} + Cov\{sB(s),tB(t)\}\\ &=& Cov\{B(s),B(t)\} - t\; Cov\{B(s), B(1)\} - s\; Cov\{B(1),B(t)\} + st\; Cov\{B(s),B(t)\}\\ &=& min\{s,t\} - t\; min\{s,1\} - s\; min\{1,t\} + st\; min\{s,t\}\\ &=& min\{s,t\} - ts - st + st\\ &=& min\{s,t\} - st. \end{eqnarray*}


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.