What is the computational complexity of the n-dimensional FFT with m points along each dimension?
Asked
Active
Viewed 4,850 times
4
-
possible duplicate of [Computational complexity of an n-dimensional Fast Fourier Transform?](http://stackoverflow.com/questions/12249275/computational-complexity-of-an-n-dimensional-fast-fourier-transform) – Apr 26 '15 at 20:39
1 Answers
7
For a 1D FFT it's O(m log m)
.
For a 2D FFT you have to do m x 1D FFTs in each axis so that's O(2 m^2 log m)
= O(m^2 log m)
.
It's too early in the morning here to get my head round n >= 3
but I'm guessing it's probably:
O(m^n log m)

Paul R
- 208,748
- 37
- 389
- 560
-
4This is right, but it is really O(2^n m^n log m) - just pointing out that the coefficient grows exponentially with n as well. – GeorgeWilson Jun 04 '14 at 23:12