12
$\begingroup$

I am stuck for weeks with the following problem:

Let $A$ and $x$ be $n \times n$ and $n \times 1$ matrices, respectively, with all entries real and strictly positive. Assume that $A^2 x = x$. Show that $A x = x$.

This was on the first problem set on a course of linear algebra based on the book written by Hoffman & Kunze. We haven't seen eigenvalues and eigenvectors yet. So, while any solution that uses anything more advanced than the first 3 chapters of that book is welcome (it may incentivize me to study something!), it does not solve the problem.

Can anyone help? Thanks!

EDIT: My question was marked as a exact duplicate of a question by Igor Caetano Diniz. While that is the exact same question, that post has one wrong answer and one answer that has a theorem that I haven't studied yet. So it doesn't solve my problem.

$\endgroup$
  • $\begingroup$ Which topics are covered in the first 3 chapters of the book? $\endgroup$ – user84413 Jul 2 '16 at 22:33
  • $\begingroup$ Linear equations, vector spaces and linear transformations $\endgroup$ – Gabriel Jul 2 '16 at 22:43
  • $\begingroup$ Edited stating why I think that my question is not a duplicate. $\endgroup$ – Gabriel Jul 2 '16 at 23:03
  • $\begingroup$ The answers there don't seem suitable, I agree. $\endgroup$ – Hoot Jul 3 '16 at 0:51
  • 1
    $\begingroup$ I have written an outline of an elementary proof that involves only the concept of linear independence. You may see if it's useful. $\endgroup$ – user1551 Jul 3 '16 at 1:50
1
$\begingroup$

The Perron-Frobenius Theorem says, among others, the following: A square matrix $A$ with strictly positive entries has a positive eigenvalue $\rho$ and corresponding eigenvector $y$ with strictly positive entries. All other eigenvalues of $A$ are strictly smaller in absolute value, and the corresponding eigenvectors do not have all positive entries.

In the situation at hand we are given a matrix $A$ with positive entries. Let $\rho$ and $y$ be as in the theorem. The matrix $A^2$ has positive entries as well, and $y$ is an eigenvector of $A^2$ with corresponding eigenvalue $\rho^2$. Since we are told that the all-positive vector $x$ is an eigenvector of $A^2$ with corresponding eigenvalue $1$, it follows from the uniqueness part of the theorem that in fact $\rho^2=1$ and $x=\lambda y$ for a $\lambda>0$. This allows to conclude that $Ax=x$.

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.