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

Convolution theorem

  1. Sep 29, 2014 #1
    I am having a hard time to understand why convolution integral gives the area overlaps of the two signal functions.
    if we use

    http://en.wikipedia.org/wiki/Convolution#mediaviewer/File:Comparison_convolution_correlation.svg

    for convolution, it is pretty obvious that one of the functions gives always 1 and therefore we just need to find another function's integral (area) and that area represents the area overlaps.

    What I feel is though, this seems one special coincidence that convolution of the two function gives the overlapped area of the two functions.


    if we use some arbitrary two functions, lets say


    x(t)=e^2t

    and

    h(t)=e^(-7t+2)


    then convolution is


    integral from negative infity to infinity x(τ)*(t-τ)
    = integral from negative infinity to infinity e^2τ*e^(-7t-7τ+2)

    we are finding integral of product of the two functions, and it will likely be greater than the overlapped are of the two functions.

    Is the convolution 'always' the overlapped area of two signal function
     
  2. jcsd
  3. Sep 30, 2014 #2

    olivermsun

    User Avatar
    Science Advisor

    No, it is not the overlapped area of two signal functions.

    The picture is just depicting the unit "top hat" function and the ramp-shaped function overlapping with different offsets. The convolution at each offset is the integral over of the point-wise product of the two functions with that offset.
     
  4. Oct 2, 2014 #3

    analogdesign

    User Avatar
    Science Advisor

    Do you understand the principle of Superposition? One way to visualize convolution is to imagine taking your single and splitting it up into however many impulses. Then the convolution is the sum of the response of all the different impulses with the other signal or system. If you're doing a convolution in discrete math it is done in just this way. I found it more intuitive.
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook




Similar Discussions: Convolution theorem
  1. Convolution of signals (Replies: 2)

  2. Graphical convolution (Replies: 4)

  3. Convolution in Quartus (Replies: 1)

  4. More convolution. (Replies: 2)

Loading...