1
$\begingroup$

I have to find the characteristic function of a random Gaussian variable of $$ \sigma_z (w) = E e^{i w z } $$. This is the variable and I know , from the theory that the characteristic function of this variable is the Fourier transform of the probability density. From theory I read that the probability density is $$\frac{1}{\sigma \sqrt 2 \pi } e^{- \frac{(z - \mu)^2 }{2 \sigma^{2}}} $$. Now I’m blocked because I don’t know how to calculate this :/ Thank you so much

$\endgroup$
3
$\begingroup$

In general, the characteristic function of a random variable is related to the fourier transform of the distribution as follows:

$$\varphi_Z(-\omega) = \mathscr F \big\{ f_Z(z) \big\}$$

Why?

Because, fourier transform of a PDF $f_Z(z)$ is:

$$\mathscr F \big\{ f_Z(z) \big\} = \int^{\infty}_{-\infty} f_Z(z)\cdot e^{-j\omega z} \ \mathrm dz$$

And Characteristic function of a PDF is given by:

$$\varphi_Z(\omega) = \mathbf E_Z \big\{e^{j\omega z} \big\} = \int^{\infty}_{-\infty} f_Z(z) \cdot e^{j\omega z} \ \mathrm dz$$

As you can see the difference is just in the argument $\omega$. I think now you can get the characteristic function of a Gaussian Random Variable. Just calculate the fourier transform of the PDF and replace $\omega$ by $-\omega$.

Fourier Transform of Gaussian PDF :

$$\mathscr F \big\{ f_Z(z) \big\} = \int^{\infty}_{-\infty} \frac{1}{\sqrt{2\pi \sigma^2}}e^{-\frac{1}{2\sigma^2}(z-\mu)^2}\cdot e^{-i\omega z} \ \mathrm dz$$

Now, just consider the power of exponential and simplify as follows: $$(\frac{-(z-\mu)^2}{2\sigma^2} - i\omega z )= -\frac{1}{2\sigma^2}(z^2+\mu^2-2\mu z + i2\sigma^2 \omega z)$$ $$= -\frac{1}{2\sigma^2}(z^2 -2(\mu - i\sigma^2 \omega)z + \mu^2 + (\mu - i\sigma^2 \omega)^2 - (\mu - i\sigma^2 \omega)^2)$$ $$ = -\frac{1}{2\sigma^2}((z - (\mu - i\sigma^2 \omega))^2 - (\mu - i\sigma^2 \omega)^2 + \mu^2)$$

Now, basically, the integral becomes the following: $$\mathscr F \big\{ f_Z(z) \big\} = \int^{\infty}_{-\infty} \frac{1}{\sqrt{2\pi \sigma^2}}e^{-\frac{1}{2\sigma^2}(z-(\mu - i\sigma^2 \omega))^2}\cdot e^{\frac{1}{2\sigma^2}((\mu - i\sigma^2 \omega)^2 - \mu^2)} \ \mathrm dz$$ The second exponential can come out of the integral since it does not contain $z$, then what you are left with is just area under a Gaussian PDF with mean $(\mu - i\sigma^2 \omega)$ and variance $\sigma^2$. Area under any PDF is equal to $1$. So, basically the fourier transform of the Gaussian PDF becomes: $$\mathscr F \big\{ f_Z(z) \big\} = e^{\frac{1}{2\sigma^2}((\mu - i\sigma^2 \omega)^2 - \mu^2)} = e^{-\frac{\sigma^2 \omega^2}{2} - i\mu \omega}$$ Once you have fourier transform, you can get the characteristic function by substituting $\omega = -\omega$ as follows: $$ \varphi_Z(\omega) = e^{-\frac{\sigma^2 \omega^2}{2} + i\mu \omega}$$

$\endgroup$
4
  • $\begingroup$ Assuming $$ f_z z $$ as $$ \frac{1}{\sigma \sqrt 2 \pi } e^{\frac{-(z-\mu)^2}{2\sigma ^{2}}} $$ now I don’t know how to solve $$ \frac{1}{\sigma \sqrt 2 \pi } \int_{\infty}^{-\infty} e^{\frac{-(z-\mu)^2}{2\sigma ^{2}}+iwz} $$ $\endgroup$ – Elena Martini Apr 20 '20 at 14:26
  • 2
    $\begingroup$ @ElenaMartini Try to write $e^{\frac{-(z-\mu)^2}{2\sigma^2} + \mathbb i \omega z} $ as sum of a complete square and a residual term without $z$. $\endgroup$ – DSP Rookie Apr 20 '20 at 14:36
  • $\begingroup$ @ElenaMartini I have added the derivation in case you need further help. $\endgroup$ – DSP Rookie Apr 20 '20 at 15:17
  • $\begingroup$ Thank you so much , now I’ll do all the step from the beginning for better understand $\endgroup$ – Elena Martini Apr 20 '20 at 15:40

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.