如何在 numpy 中将矩阵变为对角矩阵？

Question

given matrix:

x = matrix([[ 0.9,  0.14], [ 0.15,  0.8]])

how can you make the first column, x[:,0] , into a diagonal matrix in numpy? to get:

matrix([[0.9, 0],
        [0, 0.15]])

Answer 1

There is a diagflat that 'Create a two-dimensional array with the flattened input as a diagonal.'. It both ravels the input, and wraps the result in np.matrix (matching the input array type):

In [122]: np.diagflat(x[:,0])
Out[122]: 
matrix([[ 0.9 ,  0.  ],
        [ 0.  ,  0.15]])

So it's doing all the work of jez answer, just wrapping it in a generalized function:

np.matrix(np.diag(np.asarray(x[:,0]).ravel()))

Answer 2

numpy.diag( x.A[ :, 0 ] )

should do it.

The difference between a matrix and an array is crucial here. You won't get the same result from just numpy.diag( x[ :, 0 ] ) . xA is a shorthand for numpy.asarray( x ) when x is a matrix .

So by the same token, to answer your question precisely I guess I shouldn't forget convert the answer from an array back to a matrix :

numpy.matrix( numpy.diag( x.A[ :, 0 ] ) )

Answer 3

You can use np.diag(np.diag(your numpy array)) . Eg

>>> import numpy as np
>>> b = np.arange(1,10).reshape(3,3)
>>> b
array([[1, 2, 3],
       [4, 5, 6],
       [7, 8, 9]])
>>> np.diag(np.diag(b))
array([[1, 0, 0],
       [0, 5, 0],
       [0, 0, 9]])