Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.



Re: Trignometric interpolation using FFT
Posted:
May 20, 2014 4:41 PM


On Wednesday, May 21, 2014 8:20:09 AM UTC+12, jan wrote: > Given N sampled points, using the FFT we can get the Fourier transform of those N points Xk. With N/2 the Nyquist frequency and X0 the DC value. Using the inverse we can then get back the original function we just measured. However if we would like more points then just the N we have measured but instead we would like M, how can u use the inverse FFT to find the trigonometric interpolation? We can assume the N is even and that M>N. And wat if we would drop values out of , how would you find a trigonometric interpolation of the original signal.
There's an inbuilt function that does it: help interpft
But the principle is quite simple. To get closer intervals in the time domain, you must increase the Nyquist frequency in the frequency domain. You do this simply by adding zeros to the ends of the Fourier transform. The details are complicated, but interpft does it all for you.



