Suppose $f(t)$ and $F(\omega)$ are a Fourier transform pair. I want to show that $$\mathcal{F}^{-1} \left\{\frac{F(\omega)}{i\omega}\right\} = \int_{-\infty}^t f(\eta)\ \text{d}\eta$$ I start with the Fourier transform of the RHS and use integration by parts: $$\begin{align*} \mathcal{F}\left\{\int_{-\infty}^tf(\eta)\ \text{d}\eta \right\}\ &= \int_{-\infty}^{\infty} \int_{-\infty}^t f(\eta)\ \text{d}\eta\ e^{-i\omega t}\ \text{d}t \\&=\underbrace{\frac{-1}{i\omega}\left[\int_{-\infty}^t f(\eta)\ \text{d}\eta\ e^{-i\omega t}\right]_{t\ =-\infty}^{t\ =\ \infty}}_{=\ 0 ?}\ +\ \frac{1}{i\omega}\int_{-\infty}^{\infty}f(t)\ e^{-i\omega t}\ \text{d}t\\ &= \frac{F(\omega)}{i\omega} \end{align*}$$ Hence $$\int_{-\infty}^t f(\eta)\ \text{d}\eta\ = \mathcal{F}^{-1} \left\{\frac{F(\omega)}{i\omega}\right\}$$

If the result is true I can see the first term on the RHS after performing the integration by parts must vanish but I'm not sure how to justify it. Any help on justifying it (or a cleaner approach to show the result) would be appreciated, thanks!


One may recall that

$$ \mathcal{F}(g')(\omega)=i\omega\mathcal{F}(g)(\omega) \tag1 $$

applying it with $$ g(t)=\int_{-\infty}^t f(\eta)\ \text{d}\eta, \quad g'(t)=f(t),\quad $$ the notation $F(\omega):=\mathcal{F}(f)(\omega)$, it gives $$ \frac{F(\omega)}{i\omega} =\mathcal{F}\left( \int_{-\infty}^t f(\eta)\ \text{d}\eta\right)(\omega) \tag2 $$ that is

$$ \mathcal{F}^{-1} \left\{\frac{F(\omega)}{i\omega}\right\} = \int_{-\infty}^t f(\eta)\ \text{d}\eta. \tag3 $$

Are you Ok with a proof of $(1)$ ?

  • $\begingroup$ Très bon. :D No that's fine, I've already proved (1), thank you. $\endgroup$ – AtticusFinch95 Apr 27 '16 at 23:29
  • $\begingroup$ You are welcome. $\endgroup$ – Olivier Oloa Apr 27 '16 at 23:30

Let $F(\omega)$ be the Fourier transform of the square integrable, continous function $f(t)$ as given by

$$F(\omega)=\int_{-\infty}^\infty f(t)e^{-i\omega t}\,dt$$

Let $I_L(\omega,)$ be defined by the integral

$$I_L(\omega)=\int_{-L}^L \int_{-\infty}^tf(t')\,dt'\,e^{-i\omega t}\,dt$$

Integrating by parts with $u=\int_{-\infty}^tf(t')\,dt'$ and $v=\frac{e^{-i\omega t}}{-i\omega}$ yields

$$I_L(\omega)=\frac{1}{i\omega}\int_{-L}^L f(t)e^{-i\omega t}\,dt+\frac{1}{i\omega}\left(e^{i\omega L}\int_{-\infty}^{-L}f(t')\,dt'-e^{-i\omega L}\int_{-\infty}^{L}f(t')\,dt'\right) \tag 1$$

Assuming that $\int_{-\infty}^t f(t')\,dt'$ is also a square integrable function, then we must have $$\lim_{L\to \infty}\int_{-\infty}^L f(t')\,dt'=0$$

Then, the boundary term on the right-hand side of $(1)$ vanishes in the limit as $L\to \infty$ and the limit of $I_L(\omega)$ becomes

$$\begin{align} \bbox[5px,border:2px solid #C0A000]{\int_{-\infty}^\infty \left(\int_{-\infty}^t f(t')\,dt'\right)\,e^{-i\omega t}\,dt=\frac{1}{i\omega}F(\omega)}\end{align}$$

as was to be shown!

  • $\begingroup$ Hello, thanks for your answer! Just a small typo on the last line, it should be $e^{-i\omega t}$. ;) Sorry, I'm kinda new to this so bear with me here...why must $\lim_{L\to\infty} \int_{-\infty}^{L} f(t')\ \text{d}t$ vanish if $\int_{-\infty}^{t} f(t')\ \text{d}t$ is square integrable? Reading the wiki page on it isn't seeming to help... $\endgroup$ – AtticusFinch95 Apr 29 '16 at 11:53
  • $\begingroup$ You're welcome. My pleasure. -Mark $\endgroup$ – Mark Viola Apr 29 '16 at 13:28
  • 1
    $\begingroup$ Thank you for the catch. I've edited accordingly. I also added the previously tacitly assume assumption that $f$ is continuous. Then, $\int_{-\infty}^t f(t')\,dt'$ is continuous and differentiable (this permits the integration by parts). Moreover, if $\int_{-\infty}^t f(t')\,dt'$ is square integrable and continuous, then it must vanish at $\infty$. -Mark $\endgroup$ – Mark Viola Apr 29 '16 at 13:31
  • $\begingroup$ I've got it now, thank you. Ah, I was just a lowly newbie beforehand and couldn't upvote any answers, but I can now. Done. ;) $\endgroup$ – AtticusFinch95 Apr 29 '16 at 13:51
  • $\begingroup$ Thank you; much appreciative! I must ask ... is Atticus Finch your real name or are you a fan of "To Kill a Mockingbird?" $\endgroup$ – Mark Viola Apr 29 '16 at 14:18

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.