如何将2D数组连接到每个3D数组中？

Question

I am trying to concatenate the same 2D array, A with shape (n, m) , into every 2D array of a 3D array, B with shape (N, n, k) .

I tried with stacks and concatenate but it didn't work due to only one dimension match. I tried the following example to test the idea:

a = np.array([[1],[2],[3]])
b = np.ones((2,3,4))

np.hstack((a,b))

ValueError: all the input arrays must have same number of dimensions

What I was expecting is the following result:

array([[[1., 1., 1., 1., 1.],
        [2., 1., 1., 1., 1.],
        [3., 1., 1., 1., 1.]],

       [[1., 1., 1., 1., 1.],
        [2., 1., 1., 1., 1.],
        [3., 1., 1., 1., 1.]])

I am aware that it is possible to do it with a for loop but I am looking for a more compact and optimised solution.

Answer 1

I know this is a bit messy but its getting the job done

a = np.array([[1],[2],[3]])
b = np.ones((2,3,4))
a=np.expand_dims(a,axis=0)
a=np.concatenate((a,a),axis=0)
np.dstack((a,b))

Answer 2

You currently have a (3, 1) array that you want to prepend to a (2, 3, 4) array as did if it were broadcast to (2, 3, 1) . This is one of those cases where you would have to do the broadcasting yourself. If you use broadcast_to , you will get an object that does not copy the original data, and is suitable for copying into a new array, as a minimum:

c = np.broadcast_to(a, b.shape[0:1] + a.shape)
result = np.concatenate((c, b), axis=2)

https://ideone.com/ypDpyT