1
$\begingroup$

I want to compute the spectral measure of the 1-dimensional Ornstein-Uhlenbeck process with covariance function \begin{align} \rho(t) = e^{-|\beta|t}, \qquad t \in \mathbb{R}. \end{align} Since $\rho$ is a proper covariance function, Bochner's theorem says that $\rho$ is the covariance function of some stationary Gaussian random field if and only if \begin{align} \rho(t) = \int_\mathbb{R} e^{i\omega t}\ d\mu(\omega), \end{align} for some finite non-negative symmetric Borel measure on $\mathbb{R}$. Now, I know as well that if $\mu \ll \lambda$, when $\mu$ is absolutely continuous w.r.t. the Lebesgue-measure, $\rho$ could be written as \begin{align} \rho(t) = \int_\mathbb{R} e^{i\omega t}f(\omega)\ d \omega, \end{align} where $f$ is the spectral density, i.e. the Fourier transform of the covariance function $\rho$. Therefore, \begin{align} f(\omega)= \frac{1}{2\pi} \int_\mathbb{R} e^{-i\omega t} \rho(t)\ dt. \end{align} However, I do not know see how to compute the spectral measure explicitly. Any help is appreciated!

$\endgroup$
1
$\begingroup$

At first, the covariance function of OU process should be $\rho(t)=e^{-\beta|t|}$, $\beta>0$. If this is true, then the spectral density is the following: \begin{align} f(\omega)&=\frac1{2\pi}\int_{\mathbb{R}}e^{-i\omega t}\rho(t)\,dt =\frac1{2\pi}\biggl[\int_{-\infty}^0e^{-i\omega t}e^{-\beta|t|}\,dt+ \int_{0}^{\infty}e^{-i\omega t}e^{-\beta|t|}\,dt\biggr]\\ &=\frac1{2\pi}\Bigl[\frac{1}{\beta-i\omega}+\frac{1}{\beta+i\omega}\Bigr] =\frac{\beta}{\pi(\beta^2+\omega^2)}. \end{align}

$\endgroup$
  • $\begingroup$ Okido, thanks for your reaction. How to find the spectral measure from the spectral density? $\endgroup$ – iJup Apr 8 '17 at 8:31

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.