Reshape an array in NumPy

Consider an array of the following form (just an example):

[[ 0  1]
 [ 2  3]
 [ 4  5]
 [ 6  7]
 [ 8  9]
 [10 11]
 [12 13]
 [14 15]
 [16 17]]

It's shape is [9,2]. Now I want to transform the array so that each column becomes a shape [3,3], like this:

[[ 0  6 12]
 [ 2  8 14]
 [ 4 10 16]]
[[ 1  7 13]
 [ 3  9 15]
 [ 5 11 17]]

The most obvious (and surely "non-pythonic") solution is to initialise an array of zeroes with the proper dimension and run two for-loops where it will be filled with data. I'm interested in a solution that is language-conform...

a = np.arange(18).reshape(9,2)
b = a.reshape(3,3,2).swapaxes(0,2)

# a: 
array([[ 0,  1],
       [ 2,  3],
       [ 4,  5],
       [ 6,  7],
       [ 8,  9],
       [10, 11],
       [12, 13],
       [14, 15],
       [16, 17]])


# b:
array([[[ 0,  6, 12],
        [ 2,  8, 14],
        [ 4, 10, 16]],

       [[ 1,  7, 13],
        [ 3,  9, 15],
        [ 5, 11, 17]]])

numpy has a great tool for this task ("numpy.reshape") link to reshape documentation

a = [[ 0  1]
 [ 2  3]
 [ 4  5]
 [ 6  7]
 [ 8  9]
 [10 11]
 [12 13]
 [14 15]
 [16 17]]

`numpy.reshape(a,(3,3))`

you can also use the "-1" trick

`a = a.reshape(-1,3)`

the "-1" is a wild card that will let the numpy algorithm decide on the number to input when the second dimension is 3

so yes.. this would also work: a = a.reshape(3,-1)

and this: a = a.reshape(-1,2) would do nothing

and this: a = a.reshape(-1,9) would change the shape to (2,9)