如何将 3D 阵列的内积转换为二维阵列？

Question

I have two Numpy array

b=np.array([[1, 2, 3], [4, 5, 6]])
a=np.array([[[1, 2,1], [3, 4,1],[4,5,6],[6,7,8]], [[5, 6,1], [7, 8,1],[4,5,6],[6,7,8]]])

a.shape,b.shape
((2, 4, 3), (2, 3))

I want to calculate dot product of these array. I tried below code:

s=np.flip(np.dot(a,b).transpose((0,2,1)),1)

but it throws below error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-140-6ada0906cbfc> in <module>
----> 1 s=np.flip(np.dot(a,b).transpose((0,2,1)),1)

<__array_function__ internals> in dot(*args, **kwargs)

ValueError: shapes (2,4,3) and (2,3) not aligned: 3 (dim 2) != 2 (dim 0)

I want my resultant array should be of shape -(2,4)

How will I calculate this? Plz suggest some other method to do that.

Answer 1

a 's shape should be (*, 3, 2) because b 's shape is (2, 3) .

a_ · b = │ a11 a12 | * | b11 b12 b13 |
         │ a21 a22 |   | b21 b22 b23 |
         │ a31 a32 | 

where a_ is an element of a .

With the following a :

a=np.array([[
    [1, 2], [1, 3], [4, 1]
], [
    [4, 5] ,[6, 6], [7, 8]
], [
    [5, 6], [1, 7], [8, 1]
], [
    [4, 5], [6, 6] ,[7, 8]
]])

I got no error. So try:

np.dot(a.reshape(4, 3, 2), b)