Reply by December 20, 20052005-12-20
abariska@student.ethz.ch wrote:
> Almost. The convolution of s(n) with y(t) is the inverse FT of the > (point-wise) product of S(f) with Y(f)!
With appropriate zeropadding (otherise, this describes circular convolution).
Reply by December 20, 20052005-12-20
sulphox wrote:
> Hi, > > Do you mean that convolution is the multiplication of 2 complex > number? > let's say if i have S(f) which is FFT of s(n), > also i have Y(f) which is the FFT of y(t). > > Do you mean the convolution of S(f) and Y(f) is the Product of the > Multiplication of S(f) and Y(f)?
Almost. The convolution of s(n) with y(t) is the inverse FT of the (point-wise) product of S(f) with Y(f)!
Reply by sulphox December 20, 20052005-12-20
Hi,

Do you mean that convolution is the multiplication of 2 complex
number?
let's say if i have S(f) which is FFT of s(n), 
also i have Y(f) which is the FFT of y(t).

Do you mean the convolution of S(f) and Y(f) is the Product of the
Multiplication of S(f) and Y(f)?

Sulphox

Reply by Fred Marshall December 20, 20052005-12-20
<abariska@student.ethz.ch> wrote in message 
news:1134990975.314279.326830@f14g2000cwb.googlegroups.com...
> sulphox wrote: >> Hi all, >> >> As the subject suggest, I would like to know if anyone know how to >> perform convolution in excel? >> >> In excel only have the Fourier analysis, but no convolution Function. > > You can use the convolution theorem: > > conv(x,y) = ifft(fft(zeropad(x,length(x)+length(y)-1)) * > fft(zeropad(y,length(x)+length(y)-1))) > > where * denotes the point-wise multiplication of complex vectors.
I didn't know that Excel had "zeropad" or "length" ..... ?? In Excel: "Signal1" in column A "Filter" or reversed "Signal2" in column B Column C filled with SUMPRODUCT($nonzero range of cells in Column A ($ for unchanging), running range of cells in Column B of same length as range in Column A) Example: The fixed-length function is in Column D The longer function is in Column C. The convolution is in Column E. E51=SUMPRODUCT(D$4:D$54,C1:C51) where the shortest array "D" is of length 51 and "C" can be of any length. C starts with 51 zeros so the leading transient can be computed. Not fancy but it works. Fred
Reply by December 19, 20052005-12-19
sulphox wrote:
> Hi all, > > As the subject suggest, I would like to know if anyone know how to > perform convolution in excel? > > In excel only have the Fourier analysis, but no convolution Function.
You can use the convolution theorem: conv(x,y) = ifft(fft(zeropad(x,length(x)+length(y)-1)) * fft(zeropad(y,length(x)+length(y)-1))) where * denotes the point-wise multiplication of complex vectors.
Reply by sulphox December 19, 20052005-12-19
Hi all,

As the subject suggest, I would like to know if anyone know how to
perform convolution in excel?

In excel only have the Fourier analysis, but no convolution Function.

Thanks in advance.

Sulphox