I was looking at a table of Fourier transform pairs, and one entry is really confusing me. There's one on the second page that states $$ \mathcal{F}(\cos(\omega_0t))(\omega) = \pi(\delta(\omega - \omega_0) + \delta(\omega + \omega_0)) $$ I know that a Fourier transform is supposed to map a function in the time domain to an equivalent function in the frequency domain, so lets see what that means in this case. The period of $\cos(\omega_0t)$ is $2\pi/\omega_0$, meaning the frequency in rad/s is $\omega_0$. Thus at $\omega = \pm\omega_0$, the frequency domain should be $1$, because the amplitude of $\cos$ is $1$. When we plug it in, however, we get $\pi$: $$ \pi(\delta(0) + \delta(2\omega_0)) = \pi(1 + 0) = \pi $$ Why is this true? Isn't the amplitude of the cosine $1$, not $\pi$?

  • $\begingroup$ When you plug $\pm\omega_0$, you get "$\pi\infty$", if that makes sense. And with the usual convention, the coefficient is $1/2$, not $\pi$. $\endgroup$ – Yves Daoust Jul 10 '15 at 13:42

The Dirac distribution needs an integration to lead to finite values.

And that might also explain the factor $\pi$, because that integration is the inverse Fourier transformation and that might have a compensating factor in your case.

Your text lists $$ f(t) = \frac{1}{2\pi} \int\limits_{-\infty}^{\infty} \!\! F(\omega)\,e^{j\omega t}\, d\omega $$

so this gives $$ f(t) = \frac{1}{2\pi} \int\limits_{-\infty}^{\infty} \!\! \pi \left[ \delta(\omega - \omega_0) + \delta(\omega + \omega_0) \right] \, e^{j\omega t}\, d\omega = \frac{1}{2}\left[ e^{j\omega_0 t} + e^{-j\omega_0 t}\right] = \cos(\omega_0 t) $$ so we could read the transform as $$ F(\omega) = 2\pi \, \frac{1}{2} \left[ \delta(\omega - \omega_0) + \delta(\omega + \omega_0) \right] $$ where the $2\pi$ is due to the specific constants choosen for the Fourier transformation pair.


Evaluating the Dirac delta function at zero would give you $\infty$. Remember that the Dirac delta is not, strictly speaking, a function, but rather a distribution. The Fourier transform gives the distribution of the function in the frequency domain, so the cosine function is made of of only two frequencies, with no spreading.

  • $\begingroup$ Yes that's weird, it only takes a value if you integrate it together with someone else. Like a function chameleon, a funchameleon. $\endgroup$ – mathreadler Jul 10 '15 at 13:47
  • 2
    $\begingroup$ I remember I asked my analysis professor if the set of integrable functions could be viewed as a commutative ring with identity under addition and convolution. He asked me what the identity was, I said the delta function. Then he told in no uncertain terms that the Dirac delta was NOT a function, and not to treat it like one. $\endgroup$ – Alex S Jul 10 '15 at 13:50

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.