Fft In Matlab And Numpy / Scipy Give Different Results

I am trying to re-implement one of the matlab toolboxes. they use fft over there. when i perform same operation on the same data i get different results to those from matlab. Ju

Solution 1:

Matlab applies the fft over the columns of the matrix, numpy applies the fft over the last axis (the rows) by default. You want:

>>> np.fft.fft(Msig.T, axis=0)
array([[ 0.+0.j,  0.+0.j,  0.+0.j,  0.+0.j,  1.+0.j,  0.+0.j],
       [ 0.+0.j,  0.+0.j,  0.+0.j,  0.+0.j,  0.-1.j,  0.+0.j],
       [ 0.+0.j,  0.+0.j,  0.+0.j,  0.+0.j, -1.+0.j,  0.+0.j],
       [ 0.+0.j,  0.+0.j,  0.+0.j,  0.+0.j,  0.+1.j,  0.+0.j]])

or

>>> np.fft.fft(Msig).T
array([[ 0.+0.j,  0.+0.j,  0.+0.j,  0.+0.j,  1.+0.j,  0.+0.j],
       [ 0.+0.j,  0.+0.j,  0.+0.j,  0.+0.j,  0.-1.j,  0.+0.j],
       [ 0.+0.j,  0.+0.j,  0.+0.j,  0.+0.j, -1.+0.j,  0.+0.j],
       [ 0.+0.j,  0.+0.j,  0.+0.j,  0.+0.j,  0.+1.j,  0.+0.j]])