如何使用未知长度的列表切片np.array？

Question

I am probably using the wrong names/notation (and an answer probably exists on SO, but I can't find it). Please help me clarify so that I can update the post, and help people like me in the future.

I have an array A of unknown dimension n , and a list of indexes of unknown length l , where l<=n .

I want to be able to select the slice of A that correspond to the indexes in l . Ie I want:

A = np.zeros([3,4,5])
idx = [1,3]
B = # general command I am looking for

B_bad = A[idx] # shape = (2,4,5), not what I want!
B_manual = A[idx[0], idx[1]] # shape = (5), what I want, but as a general expression.
# it is okay that the indexes are in order i.e. 0, 1, 2, ...

Answer 1

You need a tuple:

>>> A[tuple(idx)]
array([0., 0., 0., 0., 0.])
>>> A[tuple(idx)].shape
(5,)

Indexing with a list doesn't have the same meaning. See numpy 's documentation on indexing for more information.