How to pad with zeros a tensor along some axis (Python)

Question

I would like to pad a numpy tensor with 0 along the chosen axis. For instance, I have tensor r with shape (4,3,2) but I am only interested in padd

Answer 1

You can use np.pad():

a = np.ones((4, 3, 2))

# npad is a tuple of (n_before, n_after) for each dimension
npad = ((0, 0), (1, 2), (2, 1))
b = np.pad(a, pad_width=npad, mode='constant', constant_values=0)

print(b.shape)
# (4, 6, 5)

print(b)
# [[[ 0.  0.  0.  0.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  0.  0.  0.]
#   [ 0.  0.  0.  0.  0.]]

#  [[ 0.  0.  0.  0.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  0.  0.  0.]
#   [ 0.  0.  0.  0.  0.]]

#  [[ 0.  0.  0.  0.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  0.  0.  0.]
#   [ 0.  0.  0.  0.  0.]]

#  [[ 0.  0.  0.  0.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  1.  1.  0.]
#   [ 0.  0.  0.  0.  0.]
#   [ 0.  0.  0.  0.  0.]]]

Answer 2

This function would pad at the end of certain axis.
If you wish to pad both side, just modify it.

def pad_along_axis(array: np.ndarray, target_length: int, axis: int = 0):

    pad_size = target_length - array.shape[axis]

    if pad_size <= 0:
        return array

    npad = [(0, 0)] * array.ndim
    npad[axis] = (0, pad_size)

    return np.pad(array, pad_width=npad, mode='constant', constant_values=0)

example:

>>> a = np.identity(5)
>>> b = pad_along_axis(a, 7, axis=1)
>>> print(a, a.shape)
[[1. 0. 0. 0. 0.]
 [0. 1. 0. 0. 0.]
 [0. 0. 1. 0. 0.]
 [0. 0. 0. 1. 0.]
 [0. 0. 0. 0. 1.]] (5, 5)

>>> print(b, b.shape)
[[1. 0. 0. 0. 0. 0. 0.]
 [0. 1. 0. 0. 0. 0. 0.]
 [0. 0. 1. 0. 0. 0. 0.]
 [0. 0. 0. 1. 0. 0. 0.]
 [0. 0. 0. 0. 1. 0. 0.]] (5, 7)