X is a real-valued random variable with characteristic function:

$\hspace{20mm}$$\phi(t) = \frac{1}{2}[\cos(t) + \cos(t\pi)]$, $\hspace{10mm}-\infty<t<\infty$.

  1. Is the distribution of X completely continuous?

  2. Is there an $r$ such that $E(X^r)$ does not exist or is infinite?

    1. The distribution is not absolutely continuous as the limit of $\sin(x)$ as $x$ approaches infinity does not exist.

    2. ?


1 Answer 1


By definition of the Fourier transform $\mathcal{F}$ (up to some convention-dependent constants),

$$ \delta_x \stackrel{\mathcal{F}}{\mapsto} e^{itx}. $$


$$ \frac{1}{2} (\delta_1 + \delta_{-1}) \stackrel{\mathcal{F}}{\mapsto} \cos t, $$


$$ \frac{1}{2} (\delta_{\pi} + \delta_{-\pi}) \stackrel{\mathcal{F}}{\mapsto} \cos \pi t, $$

So your $X$ induces the uniform measure supported on $\{1, -1, \pi, -\pi\}$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.