Python 3.8 Raise matrix to power of -1 results in inf-CodePudding

I have a simple matrix in Python:

[[1134.01    0.  ]
 [   0.   1134.01]]

And I need to raise it to the power of -1. I have tried

mat**-1 and mat = pow(mat, -1)

Both methods give me an infinity in the 0 place (I get it you can't divide by zero). The result looks like:

[[0.00088183        inf]
 [       inf 0.00088183]]

And it gives me this warning because it is trying to divide by zero:

RuntimeWarning: divide by zero encountered in reciprocal

So other than manually replacing the inf with 0 or adding a small value to my original matrix, is there a way to tell python not to produce the inf and put a 0 instead?

CodePudding user response：

You can pass a where attribute to np.power to avoid calculating over zero:

a = np.array([
    [1134.01, 0.],
    [0., 1134.01]
])
np.power(a, -1, where=a != 0)

Which gives:

array([[0.00088183, 0.        ],
       [0.        , 0.00088183]])

CodePudding user response：

You can use numpy.linalg.inv instead.

>>> np.linalg.inv(np.array([[1134.01, 0], [0, 1134.01]]))
array([[0.00088183, 0.        ],
       [0.        , 0.00088183]])