I've Googled and asked ChatGPT and looked through NumPy docs and can't find any way to do this, so thought I'd ask here.
Suppose I have a 4-dimensional array -- in this case, of shape (3, 2, 2, 2):
a = np.array([
[[[0, 0], [0, 0]],
[[0, 0], [0, 0]]],
[[[0, 0], [0, 0]],
[[0, 0], [0, 0]]],
[[[0, 0], [0, 0]],
[[0, 0], [0, 0]]],
])
and I want to set the last (second) element of the last dimension to a different value according to each row of the first dimension. In my example I have 3 rows in the first dimension, so let's suppose I wanted to apply the values [1, 2, 3] to result in:
[
[[[0, 1], [0, 1]],
[[0, 1], [0, 1]]],
[[[0, 2], [0, 2]],
[[0, 2], [0, 2]]],
[[[0, 3], [0, 3]],
[[0, 3], [0, 3]]],
]
The closest syntax I've been able to think of would be:
a[:, ..., 1] = [1, 2, 3]
But it produces an error (ValueError: could not broadcast input array from shape (3,) into shape (3,2,2)). No error is produced if I try:
a[:, ..., 1] = [1, 2]
but it produces a different result which isn't what I want:
a = np.array([
[[[0, 1], [0, 2]],
[[0, 1], [0, 2]]],
[[[0, 1], [0, 2]],
[[0, 1], [0, 2]]],
[[[0, 1], [0, 2]],
[[0, 1], [0, 2]]],
])
Is there any way to elegantly and compactly do what I want?
For now I've written a loop to cycle over every row of the first dimension and then set the values per-row, but I wonder if there's a more powerful way to do this in a single line.
You need to provide a 3-1-1 array on the right hand side of the assignment for numpy to be able to broadcast it to your slice's 3-2-2 shape. If I got your request right, following expression would work
a[:, ..., 1] = np.array([1, 2, 3])[:, None, None]