When ‘x’ is 0, ‘G’ becomes 0/0, and the IEEE standard defines that as NaN. Any NaN in a vector will propagate through all calculations involving it to result in the entire vector being NaN.
You can get around that by creating a version of L’Hospital’s rule by addint eps to ‘x’:
G=sin(pi*(x+eps)/5)./(pi*(x+eps))
Thar results in the entire vector — and its fft — being defined and not NaN.
