Rank normal form
= Rank normal form
Every <matrix> of rank $r$ is equivalent to a block <matrix> with $I_r$ in its upper-left corner and zeros elsewhere. This follows by choosing bases adapted to the kernel and image of the represented <linear map>.