Original array:
array([[ 8, 9, 10, 11],
[12, 13, 14, 15]])
With offset = 2 the result would be:
array([[ 10, 11, 12, 13],
[14, 15, 0, 0]])
With offset = -2:
array([[ 0, 0, 8, 9],
[10, 11, 12, 13]])
Is there a way to do this other than creating multiple slices and concatenating them with np.r_?
Something like this would work:
def shift1d(a, offset):
if offset == 0:
return a.copy()
b = np.zeros_like(a)
if offset<0:
b.ravel()[-offset:] = a.ravel()[:offset]
else:
b.ravel()[:-offset] = a.ravel()[offset:]
return b
shift1d(a, 2)
# array([[10, 11, 12, 13],
# [14, 15, 0, 0]])
shift1d(a,-2)
# array([[ 0, 0, 8, 9],
# [10, 11, 12, 13]])