Results 1 to 4 of 4

Math Help - Convolutions

  1. #1
    Junior Member
    Joined
    Nov 2008
    Posts
    53

    Convolutions

    f(x) = e^-(x^2) and g(x) = e^-2(x^2).
    Compute f*g (their convolution). Use it to show that if
    h_t(x) = 1/(4 \pi t) e^{(-x^2)/4t}
    we have h_t * h_s = h_(t+s)

    My first thought was to take the Fourier transform of f and g - these are standard for example f^(k) = sqrt(\pi) e^{-(\pi ^2) k^2} - then use the property for the convolution of Fourier transform, namely (f*g)^(k) = f^(k) g^(k), but to no avail. Alternatively, I can't evaluate the integral when using the definition of convolution.
    Sorry, LaTex doesn't seem to be working....
    Follow Math Help Forum on Facebook and Google+

  2. #2
    MHF Contributor
    Opalg's Avatar
    Joined
    Aug 2007
    From
    Leeds, UK
    Posts
    4,041
    Thanks
    7
    Quote Originally Posted by davidmccormick View Post
    f(x) = e^-(x^2) and g(x) = e^-2(x^2).
    Compute f*g (their convolution). Use it to show that if
    h_t(x) = 1/(4 \pi t) e^{(-x^2)/4t}
    we have h_t * h_s = h_(t+s)

    My first thought was to take the Fourier transform of f and g - these are standard for example f^(k) = sqrt(\pi) e^{-(\pi ^2) k^2} - then use the property for the convolution of Fourier transform, namely (f*g)^(k) = f^(k) g^(k), but to no avail. Alternatively, I can't evaluate the integral when using the definition of convolution.
    The key to calculating the convolution is to complete the square. Take f(x) = e^{–ax^2} and g(x) = e^{–bx^2}.To find f*g, you want to integrate the exponential of –at^2 – b(x–t)^2 (with respect to t). Complete the square to get –at^2 – b(x–t)^2 = –(a+b)[t – (bx)/(a+b)]^2 – (abx^2)/(a+b). Now make the substitution u = t – (bx)/(a+b) in the integral. The answer should be \sqrt{\pi/(a+b)} times the exponential of –(abx^2)/(a+b).

    For the second part, there is a mistake in the question. The function h_t(x) should be given by h_t(x) = 1/(\sqrt{4 \pi t}) e^{(-x^2)/4t}. If you substitute a = 1/(4t) and b = 1/(4s) in the previous calculation, then everything works out nicely.

    Quote Originally Posted by davidmccormick View Post
    Sorry, LaTex doesn't seem to be working....
    I hope it gets restored soon. It's a nightmare trying to read or write complicated maths without it.
    Follow Math Help Forum on Facebook and Google+

  3. #3
    Junior Member
    Joined
    Nov 2008
    Posts
    53
    Thanks for your great post. It solved the problem completely.
    Follow Math Help Forum on Facebook and Google+

  4. #4
    Forum Admin topsquark's Avatar
    Joined
    Jan 2006
    From
    Wellsville, NY
    Posts
    9,897
    Thanks
    326
    Awards
    1
    Quote Originally Posted by Opalg View Post
    I hope it gets restored soon. It's a nightmare trying to read or write complicated maths without it.
    See post #8 on this thread. It helps.

    -Dan
    Follow Math Help Forum on Facebook and Google+

Similar Math Help Forum Discussions

  1. Well Definedness and C^infinity Closure of Convolutions
    Posted in the Differential Geometry Forum
    Replies: 0
    Last Post: March 5th 2012, 04:58 PM
  2. How to evaluate these convolutions:
    Posted in the Calculus Forum
    Replies: 7
    Last Post: October 20th 2011, 12:34 PM
  3. solve by method of convolutions....
    Posted in the Differential Equations Forum
    Replies: 4
    Last Post: April 25th 2011, 09:16 PM
  4. Inverse Fourier Transform using convolutions
    Posted in the Calculus Forum
    Replies: 2
    Last Post: May 17th 2010, 02:18 PM
  5. Convergence of Convolutions (Approximate Convolution Identity)
    Posted in the Differential Geometry Forum
    Replies: 12
    Last Post: June 26th 2009, 05:22 AM

Search Tags


/mathhelpforum @mathhelpforum