Eigenvectors of matrix satisfy , where is the eigenvalue associated with eigenvalue . I.e. is parallel to .
Repeated multiplication of by will cause it to shrink exponentially. , , .
can be written as linear sum of its components () parallel to the eigenvectors. I.e.
Repeated multiplication of by will shrink components corresponding to small . I.e. except , are transients.
Eventually only left with component parallel to eigenvector with largest eigen value.