Everyone, I get stuck in an exercise of Functional Analysis.

Let $T \in B(H)$ (H a complex Hilbert space) and $T^*$ adjoint of $T$. Supose $T$ is a normal operator.

1) Prove that $Ker(T)= Ker(T^*) = R(T)^\perp$ - I've finished this.

2) Using previous proof. If $\alpha$ is an eigenvalue of $T$ then the conjugate $\bar\alpha$ is an eigenvalue of $T^*$.

3) If $\alpha \neq \beta $ are eigenvectors of T, then the associated eigenspaces are orthogonal between them.

My try:

2) Let $x \in H$ such that $Tx = \alpha x $.

$$<x,Tx> = <T^*x,x> = \alpha<x,x> = <\bar\alpha x,x>$$

Then we have that

$$<T^*x - \bar \alpha x,x>=0$$

Here I don't know if above implies that $T^*x - \bar \alpha x =0$.

3) I didn't make a great progress here.

  • 1
    $\begingroup$ Use T^* to get $T^*$. Your notation without the * in the exponent can be confusing. $\endgroup$ – Disintegrating By Parts Nov 29 '18 at 19:05

For part 2, using the proof of part 1 (that $\langle Tx, Tx\rangle = \langle T^*x, T^*x \rangle$), we see$$||T^*x-\overline{\alpha}x||^2 = \langle T^*x-\overline{\alpha}x, T^*x-\overline{\alpha}x\rangle = \langle T^*x, T^*x\rangle - \langle \overline{\alpha}x, T^*x\rangle - \langle T^*x, \overline{\alpha}x\rangle+\langle \overline{\alpha} x, \overline{\alpha} x\rangle$$ $$ = \langle Tx, Tx \rangle-\langle Tx, \alpha x\rangle-\langle \alpha x, Tx\rangle + \langle \alpha x, \alpha x\rangle = 0.$$

For part 3, if $Tx = \alpha x, Ty = \beta y$, then $$\alpha\langle x,y \rangle = \langle Tx,y \rangle = \langle x,T^*y \rangle = \langle x, \overline{\beta} y \rangle = \beta \langle x,y \rangle$$ implies $\langle x,y \rangle = 0$. This means that the eigenspaces are orthogonal.

  • $\begingroup$ Thank you! for this part. $\endgroup$ – Kutz Nov 29 '18 at 19:10
  • $\begingroup$ @Kutz I added part 3 $\endgroup$ – mathworker21 Nov 29 '18 at 19:13
  • $\begingroup$ Thank you for answer. $\endgroup$ – Kutz Nov 29 '18 at 19:22

$T$ is normal means $T^*T=TT^*$, which is equivalent to $$ \|Tx\| = \|T^*x\|,\;\;\; x\in H. $$ So $\mathcal{N}(T)=\mathcal{N}(T^*)$ follows. The sum of normal operators is normal, and any scalar times a normal operator is normal. And the identity $I$ is normal. So, if $T$ is normal, then so is $\alpha I-T$ for any scalar $\alpha$. Therefore $\mathcal{N}(T-\alpha I)=\mathcal{N}(T^*-\overline{\alpha}I)$, and any eigenvector of a normal $T$ with eigenvalue $\alpha$ is an eigenvector of $T^*$ with eigenvalue $\overline{\alpha}$. If $Tx=\alpha x$ and $Ty=\beta y$, then

\begin{align} (\alpha-\beta)\langle x,y\rangle & = \langle \alpha x,y\rangle-\langle x,\overline{\beta}y\rangle \\ & = \langle Tx,y\rangle-\langle x,T^*y\rangle \\ & = \langle Tx,y\rangle-\langle Tx,y\rangle =0. \end{align} Therefore, if $\alpha\ne \beta$, it follows that $x\perp y$.


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.