Search code examples
numpymultidimensional-arraymatrix-multiplicationarray-broadcastingvector-multiplication

Numpy array muliplication

I have two input arrays and one output array like this:

M=np.array([[1,2,3],[3,4,5],[6,7,8]])
u=np.array([[1,2,3],[4,5,7],[2,4,9]])
res=np.zeros((3,))

I want to do the following calculation:

for i in range(3):
      res[i]=np.matmul(np.matmul(u[0:,i].T,M),u[0:,i])
#res=array([ 231.,  594., 1957.])

Can I do it without doing for loop since for loop will take alot of time in larger size matrix

so the goal is to acheive quicker method

Solution

  • The most intuitive way to do it is probably via np.einsum:

    res = np.einsum('ki,kl,li->i', u, M, u)