Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Delta function

  1. Mar 5, 2009 #1
    1. The problem statement, all variables and given/known data

    How does one prove that [tex]\int^\infty_{-\infty}\lim_{\epsilon \rightarrow 0}(1/\pi)\frac{\epsilon g(x)}{(x-a)^{2}+\epsilon^{2}}dx=g(a)[/tex]?
  2. jcsd
  3. Mar 5, 2009 #2

    George Jones

    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    One doesn't, since the result isn't true. The limit has to be outside the integral sign.

    Mathematical Physics by Butkov has a nice proof on pages 238-239. The idea is, for positive [itex]\epsilon[/itex] to write

    [tex]\lim_{\epsilon \rightarrow 0} \int_{-\infty}^\infty = \lim_{\epsilon \rightarrow 0} \int_{-\infty}^{a-\epsilon} + lim_{\epsilon \rightarrow 0} \int_{a-\epsilon}^{a+\epsilon} + lim_{\epsilon \rightarrow 0} \int_{a+\epsilon}^{\infty},[/tex]

    and then to assume [itex]g[/itex] is bounded to show that the first and last terms go to zero.

    For [itex]\epsilon[/itex] small and [itex]g[/itex] continuous, [itex]g(x)[/itex] is approximately equal to the constant value [itex]g(a)[/itex] over the middle interval, so pull this outside of the middle integral, or, more rigorously, use the mean value theorem for integrals.
  4. Mar 6, 2009 #3
    Thanks, got it right now. The limit was indeed before the integral sign.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook