Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It only takes a minute to sign up.
Sign up to join this community
Consider a square matrix $A$ over the real numbers that is not orthogonal. Let $v \neq 0$ be a non-zero vector such that $A^T A v = A A^T v = v$. Is it true that $v$ must be an eigenvector of $A$?
It's not hard to see that if $v$ is an eigenvector, then its eigenvalue must be on the unit circle. But I can't determine whether or not $v$ is necessarily an eigenvector.
Somehow this looks like it would be a consequence of the spectral theorem, but I can't see a crisp proof. Thanks!
No. Let $Q\ne\pm I$ be any $n\times n$ orthogonal matrix, $u\in\mathbb R^n$ be any vector that is not an eigenvector of $Q$, and $a\ne\pm1$. Let $$ A=\pmatrix{Q&0\\ 0&a},\ v=\pmatrix{u\\ 0}. $$ Then $A$ is not orthogonal and $AA^Tv=A^TAv=v$, but $v$ is not an eigenvector of $A$.
In general $v$ will not be an eigenvector. For instance let $A$ be a Jordan block matrix with respect to the eigenvalue $0$. (That is $A$ has 1 over the main diagonal and all other entries are zero). Let $e_i$ denote the column vector with the i'th entry equal to 1 and the other entries zero. Then $Ae_1 = 0$ and $Ae_i = e_{i-1}$ for $i = 2, \ldots, n$. And $A^Te_n = 0$ and $A^Te_i = e_{i+1}$ for $i = 1, \ldots, n-1$. Hence $AA^Te_i = A^TAe_i = e_i$ for $i = 2,\ldots,n-1$.
