0
$\begingroup$

I am studying stochastic processes and found the next problem:

Let $A$ and $\Phi $ be two independent random variables such that $E(A) = 0$, $E(A^2) < \infty$, and $\Phi$ is uniformly distributed on $[0, 2\pi]$. Consider the stochastic process $$X_t =A \cos{(\Phi+λt)}, t≥0$$

where $\lambda$ is a fixed real parameter.

Show that the correlation between two different variables of the process is given by $$E(X_tX_s)= \frac{1}{2}E(A^2)\cos {\lambda}(t-s)$$

what i have done:

Using the fact that $Cov (X_s, X_t) = E\{ (X_s - EX_s)(X_t - EX_t) \}$, I compute this and found that $Cov (X_s, X_t) = E(A^2)E\{ \cos(\Phi + \lambda s)\cos (\Phi + \lambda t)- A \cos (\Phi +\lambda s)E(A \cos (\Phi + \lambda t )) - E(A \cos (\Phi+\lambda s))A\cos (\Phi+\lambda s) + E(A)E(A)E(\cos (\Phi+\lambda s) \cos (\Phi + \lambda t)) \}$

Using the hipothesis $E(A) = 0$,

I have that

$Cov (X_s, X_t) = E(A^2)E(\cos (\Phi + \lambda s) \cos (\Phi + \lambda t))$

but I don't know how to calculate $E(\cos (\Phi + \lambda s) \cos (\Phi + \lambda t))$, I know that I need to use the fact that $\Phi \sim U_{[0, 2\pi]}$ but I don't know how to do this

(I think that if I can make this, is possible to use the theorem that says $Cov(X,Y)= E(XY)- EXEY$ to solve the problem)

$\endgroup$
  • $\begingroup$ Use the fact that $\cos(\Phi+\lambda t)\cos(\Phi+\lambda t) = \frac{1}{2}(\cos (\lambda(t-s))+\cos(2 \Phi+\lambda(t+s)))$. $\endgroup$ – copper.hat Mar 10 '13 at 17:33
1
$\begingroup$

Note that $$\cos(x) \cdot \cos(y) = \frac{1}{2}(\cos(x-y) + \cos(x+y))$$

i.e. $$\cos(\Phi+\lambda \cdot s) \cdot \cos(\Phi+\lambda \cdot t) = \frac{1}{2} \cdot( \cos(\lambda \cdot (t-s)) + \cos(2\Phi+\lambda \cdot (s+t))$$

Thus $$\mathbb{E}\big(\cos(\Phi+\lambda \cdot s) \cdot \cos(\Phi+\lambda \cdot t) \big) = \frac{1}{2} \left( \cos(\lambda \cdot (t-s)) + \mathbb{E}\cos(2\Phi+\lambda \cdot (s+t)) \right)$$

Moreover,

$$\mathbb{E}\cos(2\Phi+\lambda \cdot (s+t)) = \frac{1}{2\pi} \int_0^{2\pi} \cos(x+ \lambda \cdot (t+s)) \, dx$$ since $\Phi \sim U(0,2\pi)$. By the periodicity of $\cos$ we conclude $$\mathbb{E}\cos(2\Phi+\lambda \cdot (s+t))=0$$

So we arrive at $$\text{cov}(X_s,X_t) = \frac{1}{2} \mathbb{E}(A^2) \cdot \cos(\lambda \cdot (t-s))$$

By the way, there is a typo in the second line of your computations: Instead of $$\mathbb{E}(A) \cdot \mathbb{E}(A) \cdot \mathbb{E}(\cos(\Phi+\lambda \cdot s) \cdot \cos(\Phi+\lambda \cdot t))$$ it should read $$\mathbb{E}(A) \cdot \mathbb{E}(A) \cdot \mathbb{E}\cos(\Phi+\lambda \cdot s) \cdot \mathbb{E}\cos(\Phi+\lambda \cdot t)$$ (Of course, that doesn't change anything in this case since $\mathbb{E}(A)=0$.)

$\endgroup$
  • $\begingroup$ Momentary lack of caffeine. Have deleted the comment. $\endgroup$ – copper.hat Mar 10 '13 at 17:42
  • $\begingroup$ thank u, I really appreciate your help $\endgroup$ – LFRC Mar 10 '13 at 17:58

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.