Numpy vectorization of 2D array differences

Question

I'd like to vectorise the difference of two M x N arrays across different slices in NumPy. Something like this:

dA = A[1:,:] - A[:-1,:]
dB = B[:,1:] - B[:,:-1]
C = dA * dB

But since dA is (M-1) x N and dB is M x (N-1), it's not a valid operation.

In other words, is there a way to vectorise this loop in NumPy?

for i in range(M-1):
    for j in range(N-1):
        C[i,j] = (A[i+1,j] - A[i,j])*(B[i,j+1] - B[i,j])

Bi Rico · Accepted Answer · 2012-03-20 00:19:34Z

3

It looks like you want:

dA = A[1:, :-1] - A[:-1, :-1]
dB = B[:-1, 1:] - B[:-1, :-1]
C = dA * dB

answered Mar 20, 2012 at 0:19

Bi Rico

25.9k3 gold badges57 silver badges75 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

segasai · Accepted Answer · 2012-06-07 23:57:10Z

2

You could also use numpy.diff function

    np.diff(A, axis=0)[:, :-1] * np.diff(B, axis=1)[:-1, :]

answered Mar 20, 2012 at 0:28

segasai

8,5581 gold badge31 silver badges40 bronze badges