Fast Fourier Transform Question in Matlab

I have a signal with 8192 samples with lots of intermittent 0s. I want to compute the FFT of this signal however I don't want to use all of these 0s.

How can I shorten the original signal to get rid of these 0s and compute a FFT that is equivalent to the original signal?

Answer 1

[Azzi Abdelmalek]

You can use sparse function

y=rand(1,8192);
yout=sparse(y)

But to compute the fft you will need to use y, unless you use your own code