1. Not finding help here? Sign up for a free 30min tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Property fourier transform

  1. Dec 26, 2008 #1
    1. The problem statement, all variables and given/known data
    If [tex]F(k)=TF\{f(x)\},k\neq 0[/tex] where TF is the Fourier transform ,and

    [tex]F(0)=\frac{1}{\sqrt{2\pi}}\int_{-\infty}^{\infty}f(u)du\neq 0[/tex] ,

    show that

    [tex]TF\{\int_{-\infty}^{x}f(u)du\}=-i \frac{F(k)}{k} +\pi F(0)\delta(k)[/tex]

    2. Relevant equations

    3. The attempt at a solution
    I attempt the following:

    [tex]TF\{\int_{-\infty}^{x}f(u)du\}=\frac{1}{\sqrt{2\pi}}\int_{-\infty}^{\infty} \{\int_{-\infty}^{x}f(u)du\}e^{-ikx}dx[/tex]

    and , integration by parts, with




    [tex]\frac{1}{\sqrt{2\pi}}\{\int_{-\infty}^{x}f(u)du \frac{i}{k} e^{-ikx}\}|_{-\infty}^{\infty} -\frac{1}{\sqrt{2\pi}} \frac{i}{k}\int_{-\infty}^{\infty}f(x)e^{-ikx}dx[/tex]

    The last term is

    [tex] -\frac{i}{k}F(k)[/tex]

    What should i do with the first?. I do this, but it probably be bad:

    [tex]\frac{i}{k} e^{-ikx}\}|_{-\infty}^{\infty}=\int_{-\infty}^{\infty}e^{-ikx}dx = 2\pi\delta(k)[/tex]


    [tex]\frac{1}{\sqrt{2\pi}}\left[\lim_{x\rightarrow\infty}\int_{-\infty}^xf(z)dz-\lim_{x\rightarrow -\infty}\int_{-\infty}^xf(z)dz\right] = F(0) [/tex]


    \int_{-\infty}^{\infty}\frac{dx}{\sqrt{2\pi}}\int_{-\infty}^xf(z)dze^{-ikx}=-i\frac{F(k)}{k}+2\pi F(0)\delta(k)

    where is an extra 2.

    I tried to do this problem using convolution, but again i cant achieve the desired result:


    If [tex]g(x-u)[/tex] is the Heaviside function, and with [tex]x-u>0[/tex] , the convolution is




    But, the fourier transform of the Heaviside function is:



    [tex]TF\{\int_{-\infty}^{x}f(u)du\}=\frac{-iF(k)}{k}+\pi F(k)\delta(k)[/tex]

    But, in the last term appears [tex]F(k)[/tex] , and i want to get [tex]F(0)[/tex]. What is wrong??

  2. jcsd
  3. Dec 26, 2008 #2


    User Avatar
    Science Advisor
    Homework Helper

    There's nothing wrong with your convolution argument. Think what happens if you integrate a test function against F(k)*delta(k). You get the same thing as if you integrate it against F(0)*delta(k) if F is continuous.F(k)*delta(k) and F(0)*delta(k) are the same distribution.
  4. Dec 27, 2008 #3
    Ok,i understand it now, thanks for the help :smile:
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?