As stated in the title, I have two questions

  1. Why convolution is defined as $(f*g)(x) = \int_{-\infty}^\infty f(t) g(x - t) dt$ instead of just $\int_{-\infty}^\infty f(t) g(x + t) dt$ ? Why we need to flip $g$ ?

  2. Is there any intuitive explanation of convolution theorem (Fourier transform) ?

  • $\begingroup$ I had voted to close this question as being too broad, not as a duplicate of a previous question. In fact, I do not agree that "Filpping...." is at all an answer to this question which is asking why convolutions are defined the way that they are instead of the other way (without flipping) which is commonly called correlation instead of convolution. So, I am voting to re-open this question. $\endgroup$ – Dilip Sarwate Dec 24 '18 at 18:28

Because the front part of the input hits the filter first.

The graphical diagrams shown in most text books with an exponentially decaying signal multiplied by a step function show this directly.

While Causal response is not required for valid convolution, it is essential that the filter output responds and evolves from the head of the input.

Convolution also commutes, so the head of the impulse function also hits the signal first.

Correlation overlays the pair of functions which why it is typically used in measurements associated with similarly measures.

  • $\begingroup$ Can you explain you answer in more details ? $\endgroup$ – HOANG GIANG Dec 23 '18 at 15:47
  • $\begingroup$ i just did, can you look at a book. pictures convey more information than words $\endgroup$ – Stanley Pawlukiewicz Dec 23 '18 at 15:49

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.