Suppose that I have two signals $x[n] = \left\{2,4,1\right\}$ and $p[n] = \left\{5,1,8\right\}$ and I want to multiply them.

  • How do you do that?
  • How different is it from convolving two signals?

I understand that multiplication in one domain is equal to convolution in other domain. How do you choose as to what to use : multiplication or convolution?

  • 2
    $\begingroup$ How different are apples and oranges? This part of the question is too broad, please specify what kind of difference interests you. $\endgroup$ – chirlu Aug 23 '13 at 7:02
  • 2
    $\begingroup$ This is like asking "What's better, an umbrella or a hammer?" Depends what you're trying to do. $\endgroup$ – endolith Aug 23 '13 at 14:19
  • $\begingroup$ Will Normal Time domain convolution result in frequency domain multiplication ? or it should be CIRCULAR CONVOLUTION ! Somewhere I read Property of Discrete Fourier Transform [DFT] if x(n) and h(n) are two time-domain signals, circular convolution between them x(n)©h(n) is equivalent to multiplication in the frequency domain. Because this property used in cyclic prefix for OFDMA systems. $\endgroup$ – rajez79 Aug 26 '13 at 16:45

Yes, you are correct. Multiplication in time domain means convolution in frequency domain and vice versa. Multiplying your signals $x[n]$ and $y[n]$ will give an output:

\begin{align} z[n]&=\{2\cdot 5, 4\cdot 1, 1\cdot 8\}\\ &= \{10, 4, 8\}\end{align}

Remember that this output is in time domain. When you convolve $x[n]$ and $y[n]$, you will get $z[n]$ in time domain as:

\begin{align} z[n]&=\{5\cdot 2, 5\cdot 4+1\cdot 2, 5\cdot 1+1\cdot 4+8\cdot 2, 1\cdot 1+8\cdot 4, 8\cdot 1\}\\ &= \{10,22,25,33,8\} \end{align}

Just flip one of the signals around zero and start moving right one place at a time. Multiply the corresponding points as you go along. The output has a larger sequence because convolution output has ${\rm length}(x)+{\rm length}(y)-1$ points.

To answer your query about where to use multiplication & convolution, assume you want to pass signal $x(n)$ through filter $y(n)$. The output of the filter $z(n)$ will be convolution of $x(n)$ and $y(n)$.
Now assume that you first converted from time to frequency domain, i.e. $X(e^{i\omega})$ and $Y(e^{i\omega})$ are frequency domain representation of $x(n)$ and $y(n)$. Now, to find $Z(e^{i\omega})$(output in frequency domain), you have to multiply $X(e^{i\omega})$ and $Y(e^{i\omega})$. To get the output in time domain i.e. $z(n)$ you have to apply inverse transform.

So that's how you use convolution and multiplication.

| improve this answer | |
  • 1
    $\begingroup$ So when to use multiplication and when to use convolution? From what you explained convolution is equal to polynomial multiplication. $\endgroup$ – user2381442 Aug 23 '13 at 7:24
  • $\begingroup$ @user2381442 edited the answer to include your query. And yes, convolution is a polynomial multiplication operation $\endgroup$ – KharoBangdo Aug 23 '13 at 8:17
  • 1
    $\begingroup$ Convolution is filtering in the spatial domain, multiplication is filtering in the Fourier domain. What to use depends on what you are trying to do. $\endgroup$ – geometrikal Aug 23 '13 at 8:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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