If all eigenvalues are < 1, fixed point iterations converges to the only solution

291 Views Asked by At

Theorem states that for every initial value fixed point iteration x = Bx+b converges to the only solution of the system if all $|\lambda| $ < 1. Prove it using Jordans normal form. Initial form is Ax=b and to get B from A us whatever you want(Althoug nothing is known about A). Only thing is that if all eigenvalues of B are less than 1 then there is only 1 solution and fixed point iteration will find it.

  1. How can I represent the matrix B using Jordans normal form?
  2. What does Jordans normal form say about eigenvalues?
  3. Is there anything else i need to know to prove this?

I understand that Jordan normal form consists of Jordan boxes where diagonal has eigenvalue and diagonal on top has 1-s. But what determines the Jordans box size?