Properties of annihilators and eigenspaces

412 Views Asked by At

I'm stuck on the following practice question, and I'm uncertain of how to proceed with the proof. We've only very briefly touched on annihilators in class, and I really don't understand how to combine them with eigenvectors and eigenvalues. Any help would be appreciated.

Q. Consider a vector space $V$ with $\dim(V)=n$ and $T \in \operatorname{End}_K(V)$

(a) Show that $\operatorname{Ann}(V)=\operatorname{Ann}(u)$ for some eigenvector $u$ of $T$ if and only if $T$ is diagonalizable with exactly one eigenvalue.

(b) Show that $\operatorname{Ann}(V)=\operatorname{Ann}(u)$ for some eigenvector $u$ of $T$ if and only if $T=\lambda I$

EDIT For a vector space V over a field K, we denote $\operatorname{Ann}(u)=\{p(x) \in K[x] ; p(T)(u)=0 \}$.

We denote $\operatorname{Ann}(V)=\{p(x) \in K[x] ; p(T)=0 \}$.

1

There are 1 best solutions below

0
On BEST ANSWER

Note that $p(T) = 0$ if and only if $p$ is divisible by the minimal polynomial of $T$, which we can call $m_T(x)$. So, we can write $\operatorname{Ann}(V) = \{m_T(x)q(x) : q \in K[x]\}$, the set of all polynomials divisible by $m_T(x)$.

Now, if $u$ is an eigenvector with eigenvalue $\lambda$, then for any $p$, $p(T)u = p(\lambda) u$. It follows that $p(T)u = 0$ iff $p(\lambda) = 0$, which is to say that $x - \lambda$ divides $p$. So, we can write $\operatorname{Ann}(u) = \{(x - \lambda)q(x) : q \in K[x]\}$, the set of all polynomials divisible by $x - \lambda$ (alternatively, we can think of this as the reapplication of the same argument over an invariant subspace).

Now, if these two sets are the same, what can we deduce about $m_T$?