Given a matrix, A, that can be square, but, is most likely rectangular; and given that an identity matrix can be formed on the Rank(A) left most columns, is there a nice way