Use unravel_index:
unravel_index
numpy.unravel_index(A.argmin(), A.shape)
[Corrected typo]
Another simple solution is
ri, ci = A.argmin()//A.shape[1], A.argmin()%A.shape[1]
As numpy.argmin returns the index reading in row-major order
Yes, you are right, it was a typo, which worked for square matrix