python - Slicing 3d numpy array returns strange shape -

if slice 2d array set of coordinates

>>> test = np.reshape(np.arange(40),(5,8)) >>> coords = np.array((1,3,4)) >>> slice = test[:, coords] 

then slice has shape expect

>>> slice.shape (5, 3) 

but if repeat 3d array

>>> test = np.reshape(np.arange(80),(2,5,8)) >>> slice = test[0, :, coords] 

then shape now

>>> slice.shape (3, 5) 

is there reason these different? separating indices returns shape expect

>>> slice = test[0][:][coords] >>> slice.shape (5, 3) 

why these views have different shapes?

slice = test[0, :, coords] 

is simple indexing, in effect saying "take 0th element of first coordinate, of second coordinate, , [1,3,4] of third coordinate". or more precisely, take coordinates (0,whatever,1) , make our first row, (0,whatever,2) , make our second row, , (0,whatever,3) , make our third row. there 5 whatevers, end (3,5).

the second example gave this:

slice = test[0][:][coords] 

in case you're looking @ (5,8) array, , taking 1st, 3rd , 4th elements, 1st, 3rd , 4th rows, end (5,3) array.

edit discuss 2d case:

in 2d case, where:

>>> test = np.reshape(np.arange(40),(5,8)) >>> test array([[ 0,  1,  2,  3,  4,  5,  6,  7],        [ 8,  9, 10, 11, 12, 13, 14, 15],        [16, 17, 18, 19, 20, 21, 22, 23],        [24, 25, 26, 27, 28, 29, 30, 31],        [32, 33, 34, 35, 36, 37, 38, 39]]) 

the behaviour similar.

case 1:

>>> test[:,[1,3,4]] array([[ 1,  3,  4],        [ 9, 11, 12],        [17, 19, 20],        [25, 27, 28],        [33, 35, 36]]) 

is selecting columns 1,3, , 4.

case 2:

>>> test[:][[1,3,4]] array([[ 8,  9, 10, 11, 12, 13, 14, 15],        [24, 25, 26, 27, 28, 29, 30, 31],        [32, 33, 34, 35, 36, 37, 38, 39]]) 

is taking 1st, 3rd , 4th element of array, rows.