Description
Hi,
Here is an example of inv that doesn't throw an error when dealing with singular matrix and return wrong results.
import numpy as np
import scipy as si
A=np.array([[1,2,3],[4,5,6],[7,8,9]])
b=np.array([15,15,15])
print np.dot(np.linalg.inv(A),b)
array([-32., 48., -24.] ## wrong result
np.linalg.solve(A,b) # the result of np.dot(np.linalg.inv(A),b) should be same as this.
array([-39., 63., -24.]) # correct result
It is necessary to say that scipy throws an error though.
np.dot(si.linalg.inv(A),b)
My numpy version is '1.9.2' and python 2.7.6.